Grainger & Allison's Diagnostic Radiology, 5th Edition

Contributors

†

: deceased

Andy Adam MBBS (Hons), FRCP, FRCS, FRCR, FFRRCSI (Hon)

Professor of Interventional Radiology Department of Radiology St Thomas’ Hospital King’s College London London, UK E. Jane Adam MBBS, MRCP, FRCR Consultant Radiologist Department of Radiology St George’s Hospital London, UK Judith E. Adams MBBS, FRCR, FRCP Chair, Diagnostic Radiology Imaging Science and Biomedical Engineering University of Manchester Honorary Consultant Radiologist Royal Infirmary Manchester, UK David J. Allison BSc, MD, MRCS, LRCP, MBBS, DMRD, FRCR, FRCP

Emeritus Professor of Imaging Imperial College London, UK Sandra Allison MD Assistant Professor of Radiology Director, Radiology Residency Program Director, Ultrasound Georgetown University Hospital Washington DC, USA Philip Anslow FRCR Consultant Neuroradiologist Department of Radiology Radcliffe Infirmary Oxford, UK Susan M. Ascher MD Georgetown University Medical Center Washington DC, USA Zelena A. Aziz MD, MRCP, FRCR Consultant Radiologist Department of Radiology London Chest Hospital London, UK

Clive I. Bartram FRCS, FRCP, FRCR Emeritus Consultant, St Mark’s Hospital and Honorary Professor of Gastrointestinal Radiology Faculty of Medicine Imperial College London, UK Philip P. W. Bearcroft FRCP, FRCR Consultant Radiologist Department of Radiology Cambridge University Hospitals NHS Foundation Trust Addenbrooke’s Hospital Cambridge, UK Anna-Maria Belli DMRD, FRCR Consultant Vascular Radiologist and Reader in Radiology Department of Radiology St George’s Hospital London, UK Anthony R. Berendt BM, BCh, FRCP Consultant Physician-in-Charge Bone Infection Unit Nuffield Orthopaedic Centre Oxford, UK Lol Berman FRCP, FRCR University Department of Radiology Addenbrooke’s Hospital Cambridge, UK Martin J. K. Blomley†

Gisele Brasil Caseiras PhD Research Fellow Insitute of Neurology University College London London, UK Jackie E. Brown BDS, MSc, FDSRCP, DDRRCR Consultant Oral and Maxillofacial Radiologist Kings College London Dental Institute Guy’s Dental Hospital London, UK Dina F. Caroline MD, PhD Professor Emerita Department of Radiology Temple University Hospital Philadelphia Pennsylvania, USA Silvia D. Chang MD, FRCP(C) Assistant Professor University of British Columbia Department of Radiology Vancouver General Hospital Vancouver, Canada W. K. ‘Kling’ Chong BMedSci, MD, MRCP, FRCR Consultant Neuroradiologist Department of Radiology Great Ormond Street Hospital for Children NHS Trust London, UK Bairbre Connolly MBBCh, BAO, FRCSI, MCH, FFRRCSI, FRCP(C)

Carol A. Boles MD Associate Professor of Radiology Associate, Surgical Sciences Orthopedic Surgery Wake Forest University North Carolina, USA Jamshed B. Bomanji MBBS, MSc, PhD Consultant in Nuclear Medicine UCLH Trust Middlesex Hospital London, UK

Medical Director and Division Head of Image Guided Therapy Pediatric Interventional Radiologist Assistant Professor, University of Toronto Department of Diagnostic Imaging The Hospital for Sick Children Toronto, Canada Susan J. Copley MBBS, MD, MRCP, FRCR Consultant Radiologist and Honorary Senior Lecturer Radiology Department Hammersmith Hospital London, UK

x

CONTRIBUTORS

David O. Cosgrove MA, MSc, FRCP, FRCR Emeritus Professor Imaging Sciences Department Faculty of Medicine Imperial College Hammersmith Hospital London, UK

Robert J. Eckersley PhD Research Associate Imaging Sciences Department Faculty of Medicine Imperial College Hammersmith Hospital London, UK

Philip C. Goodman MD Professor of Radiology Chief, Thoracic Imaging Division Department of Radiology Duke University Medical Center Durham North Carolina, USA

Nigel Cowan PGDipLATHE, FRCP Oxford, UK

Andrew J. Evans MRCP, FRCR Consultant Radiologist Nottingham Breast Institute Nottingham City Hospital Nottingham, UK

Isky Gordon FRCR, FRCP Professor of Paediatric Imaging Institute of Child Health London, UK

Justin J. Cross MRCP, FRCR Consultant Neuroradiologist Department of Radiology Addenbrooke’s Hospital Cambridge, UK Paras Dalal BSc, MRCP, FRCR Research Fellow in Thoracic Imaging Department of Radiology Royal Brompton Hospital London, UK Maria Daskalogiannaki MD Registrar in Radiology Department of Radiology University Hospital of Heraklion Crete, Greece A. Mark Davies FRCR Consultant Radiologist Royal Orthopaedic Hospital Birmingham, UK Adrian K. Dixon MD, FRCR, FRCP, FRCS, FMedSci, FFRRCSI (Hon), FRANZCR (Hon)

Professor of Radiology Department of Radiology Addenbrooke’s Hospital University of Cambridge Cambridge, UK Rose de Bruyn DMRD, FRCR Consultant Radiologist Department of Radiology Great Ormond Street Hospital for Sick Children NHS Trust London, UK Claudio Defilippi MD Consultant Radiologist Department of Radiology OIRM - S. Anna Hospital Turin, Italy Sujal R. Desai MD, MRCP, FRCR Consultant Radiologist Department of Radiology King’s College Hospital London, UK

Jane Evanson BSc, MBBS, MRCP, FRCR Consultant Neuroradiologist, Barts & The London Hospital NHS Trust The Royal London Hospital London, UK Laura Fender BMedSci, BMBS, MRCP, FRCR Consultant Radiologist Nottingham University Hospital Nottingham, UK Alan H. Freeman MBBS, FRCR Consultant Radiologist Department of Radiology Addenbrooke’s Hospital Cambridge, UK Julia Gates MD Assistant Professor of Radiology Department of Radiology Tufts University School of Medicine Springfield Massachusetts, USA Robert N. Gibson MBBS, MD, FRANZCR, DDU Professor of Radiology Department of Radiology University of Melbourne Royal Melbourne Hospital Victoria, Australia Raymond J. Godwin MA, MB, Bchir, FRCP, FRCR Consultant Radiologist Department of Radiology West Suffolk Hospital Suffolk, UK Karen Goldstone BSc, MSc, Csci, FIPEM Radiation Protection Advisor Acting Head of Department of Medical Physics and Clinical Engineering East Anglian Regional Radiation Protection Service (EARRPS) Addenbrooke’s Hospital Cambridge, UK

Nicholas Gourtsoyiannis FRCR (Hon) Professor of Radiology University of Crete Faculty of Medicine Heraklion, Crete Greece Andrew J. Grainger BM, BS, MRCP, FRCR Consultant Radiologist Chapel Allerton Orthopaedic Centre Leeds Teaching Hospitals Leeds, UK Ronald G. Grainger MB ChB(Hons), MD, FRCP, DMRD, FRCR, FACR(Hon), FRACR(Hon)

Professor of Diagnostic Radiology (Emeritus) University of Sheffield Honorary Consultant Radiologist Royal Hallamshire Hospital and Northern General Hospital Sheffield, UK Philippe Grenier FRCR (Hon) Professor of Radiology Service de Radiologie Polyvalente Diagnostique et Interventionnelle Hôpital Pitié-Salpêtrière Paris, France Roxana S. Gunny BS, BSc, MRCP, FRCR Consulant Neuroradiologist Department of Radiology Great Ormond Street Hospital for Children NHS Trust London, UK Christine M. Hall MBBS, DMRD, FRCR MD Professor of Paediatric Radiology Great Ormond Street Hospital for Children NHS Trust London, UK David M. Hansell MD, FRCP, FRCR, LRSM Professor of Thoracic Imaging Department of Radiology Royal Brompton Hospital London, UK

CONTRIBUTORS

George G. Hartnell FRCR, FRCP Director of Cardiovascular and Interventional Radiology Department of Radiology Baystate Medical Center Springfield Professor of Radiology Tufts University Medical School Boston Massachusetts, USA Hedvig Hricak MD, Dr. Med, SC, Dr. h.c, FRCR (Hon) Chairman, Department of Radiology Carroll and Milton Petrie Chair Professor of Radiology, Weill Medical College of Cornell University Memorial Sloan-Kettering Cancer Center New York, USA James E. Jackson MRCP, FRCR Consultant Radiologist Department of Imaging Hammersmith Hospital London, UK H. Rolf Jäger FRCR, MD Reader in Neuroradiology Institute of Neurology University College London Honorary Consultant Neuroradiologist The National Hospital for Neurology and Neurosurgery and University College Hospital London, UK Jonathan J. James BMBS, FRCR Consultant Radiologist Nottingham Breast Institute Nottingham City Hospital Nottingham, UK Renee M. Kendzierski DO Assistant Professor of Radiology Department of Radiology Temple University Hospital Philadelphia Pennsylvania, USA Dow-Mu Koh MRCP, FRCP Senior Lecturer and Honorary Consultant Department of Radiology Royal Marsden Hospital Sutton, UK Isla Lang MBChB, MRCP, FRCR Consultant Paediatric Radiologist Sheffield Children’s Hospital Sheffield, UK

Adrian K. P. Lim MD, FRCR Consultant Radiologist and Senior Lecturer Imaging Sciences Department Faculty of Medicine Imperial College Hammersmith Hospital London, UK

Stuart E. Mirvis MD, FACR Professor of Radiology Department of Radiology University of Maryland School of Medicine Baltimore Maryland, USA

David J. Lomas MA, MB, BChir, FRCR, FRCP Professor of Clinical Magnetic Resonance Imaging Department of Radiology Addenbrooke’s Hospital Cambridge, UK

Sameh K. Morcos FRCS, FFRRCSI, FRCR Professor of Diagnostic Imaging University of Sheffield Consultant Radiologist Department of Diagnostic Imaging Northern General Hospital Sheffield, UK

Sharyn L. S. MacDonald MBChB, FRANZCR Consultant Radiologist Department of Radiology Christchurch Hospital Christchurch, New Zealand David MacVicar MA, MRCP, FRCP, FRCR Consultant Radiologist Department of Diagnostic Radiology Royal Marsden Hospital Sutton, UK Adrian Manhire BSc, MBBS, FRCP, FRCR Consultant Radiologist Nottingham City Hospital Nottingham, UK Tarik F. Massoud MA, MD, PhD, FRCR University Lecturer and Honorary Consultant in Neuroradiology University Department of Radiology University of Cambridge School of Clinical Medicine Addenbrooke’s Hospital Cambridge, UK Kieran McHugh FRCPI, DCH, FRCR Department of Radiology Great Ormond Street Hospital for Sick Children NHS Trust London, UK James Meaney FRCR, FFRRCSI Director of MRI St James’s Hospital Dublin, Ireland Hylton B. Meire FRCR, DRCOG, DMRD Consultant Radiologist (Retired) King’s College Hospital London, UK Kenneth A. Miles MD, FRCR, MSc, FRCP Clinical Imaging Sciences Centre Brighton and Sussex Medical School University of Sussex Falmar, Brighton, UK

Robert A. Morgan MBChB, MRCP, FRCR Consultant Radiologist Department of Radiology St George’s Hospital London, UK Iain Morrison MBBS, MRCP, FRCR Consultant Radiologist Radiology Department Kent and Canterbury Hospital Canterbury, UK Nestor L. Müller MD, PhD, FRCPC Professor and Chairman Department of Radiology University of British Columbia Head and Medical Director Department of Radiology Vancouver General Hospital Vancouver, Canada Graham Munneke MRCP, FRCR Consultant in Interventional Radiology Department of Radiology St. George’s Hospital London, UK Alison D. Murray MB ChB (Hons), FRCR, FRCP Senior Lecturer in Radiology Department of Radiology University of Aberdeen Aberdeen, UK Richard A. Nakielny FRCR Honorary Clinical Lecturer Directorate of Medical Imaging & Medical Physics Royal Hallamshire Hospital Sheffield, UK Hrudaya Nath MD Professor of Radiology Department of Radiology University of Alabama Hospitals Birmingham Alabama, USA

xi

xii

CONTRIBUTORS

Tony Nicholson MSc, FRCR Consultant Vascular Radiologist Department of Clinical Radiology Leeds Teaching Hospitals Leeds, UK Amaka C. Offiah BSc, MBBS, MRCP, FRCR, PhD Consultant Radiologist (Academic) Great Ormond Street Hospital for Children NHS Trust London, UK Simon Padley BSc, MBBS, FRCP, FRCR Consultant Radiologist Department of Radiology Chelsea and Westminster Hospital London, UK Martyn N. J. Paley PhD, FInstP Professor of MR Physics Academic Radiology University of Sheffield Sheffield, UK Nickolas Papanikolaou PhD Biomedical Engineer Department of Radiology University Hospital of Heraklion Crete, Greece Jai Patel MBChB, MRCP, FRCR Consultant Vascular Radiologist Department of Clinical Radiology St James’s University Hospital Leeds, UK Anne Paterson MBBS, MRCP, FRCR, FFR RCSI Consultant Paediatric Radiologist Radiology Department Royal Belfast Hospital for Sick Children Belfast, UK Praveen Peddu MRCS, FRCR Specialist Registrar in Radiology Department of Radiology King’s College Hospital London, UK A. Michael Peters BSc, MD, MSc, MRCP, MRCPath, FRCR

Professor of Nuclear Medicine Brighton and Sussex Medical School University of Sussex Brighton, UK William H. Ramsden BM, FRCR Consultant Paediatric Radiologist Department of Clinical Radiology St James’s University Hospital Leeds, UK

Sheila Rankin FRCR Consultant Radiologist Department of Radiology Guy’s and St. Thomas’ Foundation Trust London, UK

John Rout BDS, FDSRCS, MDentSc, DDRRCR, FRCR Consultant Oral and Maxillofacial Radiologist Birmingham Dental Hospital Birmingham, UK

Padma Rao MBBS, BSc, MRCP, FRCR,

Michael B. Rubens MB, DMRD, FRCR Consultant Radiologist Department of Radiology Royal Brompton Hospital London, UK

FRANZCR

Consultant Paediatric Radiologist Royal Children’s Hospital Parkville Melbourne Victoria, Australia Christine Reek BSc, FRCR Consultant Radiologist Department of Radiology Greenfield Hospital Leicester, UK John H. Reynolds DMRD, FRCR, MMedSci Consultant Radiologist Birmingham Heartlands Hospital Birmingham, UK Rodney H. Reznek FRANZCR (Hon), FRCP, FRCR Professor of Diagnostic Imaging The Centre for Cancer Imaging St Bartholomew’s Hospital and The London Queen Mary’s School of Medicine and Dentistry London, UK Philip M. Rich BSc, FRCS, FRCR Consultant Neuroradiologist Department of Neuroradiology Atkinson Morley Wing St George’s Hospital London, UK Andrea Rockall MD, BS, BSc, MRCP, FRCP Department of Radiology St Bartholomew’s Hospital London, UK Giles Roditi FRCP, FRCR Consultant Radiologist Department of Radiology Glasgow Royal Infirmary Glasgow, UK

Asif Saifuddin BSc (Hons), MBChB, MRCP, FRCR Consultant Radiologist Department of Radiology Royal National Orthopaedic Hospital NHS Trust Stanmore, UK Evis Sala MD, PhD, FRCR University Lecturer in Oncology Imaging University Department of Radiology Addenbrooke’s Hospital Cambridge, UK Caron Sandhu FRCR Consultant Radiologist Department of Radiology Guy’s and St. Thomas’ Hospital London, UK Dawn Saunders MD, MRCP, FRCR Consultant Neuroradiologist Department of Radiology Great Ormond Street Hospital for Children NHS Trust London, UK Daniel J. Scoffings MRCP, FRCR Specialist Registrar in Neuroradiology Addenbrooke’s Hospital Cambridge, UK Djilda Segerman MA, MSc, MIPEM Head of Nuclear Medicine Physics Department of Medical Physics Brighton and Sussex University Hospitals NHS Trust Brighton, UK

Lee F. Rogers MD Clinical Professor of Radiology Department of Radiology University of Arizona Health Services Tucson Arizona, USA

Kathirkama Shanmuganathan MD Associate Professor of Radiology Department of Radiology University of Maryland School of Medicine Baltimore Maryland, USA

Giles Rottenberg FRCR Consultant Radiologist Department of Radiology Guy’s and St. Thomas’ Foundation Trust London, UK

Ashley S. Shaw MRCP, FRCR Consultant Radiologist Department of Radiology Addenbrooke’s Hospital Cambridge, UK

CONTRIBUTORS

Mihra S. Taljanovic MD, MA Associate Professor of Clinical Radiology and Clinical Orthopedic Surgery Head - Musculoskeletal Imaging Section Department of Radiology Tucson Arizona, USA

Satinder P. Singh MD, FCCP Associate Professor of Radiology Director Cardiac CT Director, Combined Cardiopulmonary and Abdominal Fellowship Chief of Cardiopulmonary Radiology Department of Radiology University of Alabama Hospitals Birmingham Alabama, USA

FRCR

S. Aslam A. Sohaib MRCP, FRCR Radiology Department Royal Marsden Hospital London, UK

Consultant Radiologist Department of Clinical Radiology Great Ormond Street Hospital for Children NHS Trust London, UK

Alan Sprigg MBChB, DCH, DRCOG, DMRD, FRCR, FRCPCH

Consultant Paediatric Radiologist Sheffield Children’s Hospital Sheffield, UK John M. Stevens MBBS, DRACR, FRCR Lyshom Department of Neuroradiology Radiology Department The National Hospital for Neurology and Neurosurgery London, UK Dennis J. Stoker MB, FRCP, FRCS, FRCR Emeritus Consultant Radiologist Henley-on-Thames, UK Nicola H. Strickland BM BCh, MA (Hons),

Andrew M. Taylor BA (Hons), BM BCh, MRCP

Stuart Taylor BSc, MD, MRCP, FRCR Consultant Radiologist Department of Intestinal Imaging St Marks Hospital Northwick Park Harrow, UK Henrik S. Thomsen MD Professor and Chairman Department of Diagnostic Radiology Copenhagen University Hospital Herlev, Denmark Paolo Toma MD Radiologist-in-Chief Radiology Department G. Gaslini Institute Genoa, Italy

(Oxon), FRCP, FRCR

Consultant Radiologist Imaging Department Hammersmith Hospitals NHS Trust Honorary Senior Lecturer Imperial College London, UK Louise E. Sweeney MBBCH, BAO, DCH, DMRD, FRCR, FFR, RCSI

Consultant Paediatric Radiologist Radiology Department Royal Belfast Hospital for Sick Children Belfast, UK

Peter Twining FRCR, BSc, BS MB Consultant Radiologist Nottingham University Hospital Nottingham, UK John A. Verschakelen MD, PhD Professor of Chest Radiology Department of Radiology University Hospitals Gasthuisberg Leuven, Belgium

Sarah J. Vinnicombe BSc, MRCP, FRCR Consultant Radiologist Department of Radiology St Bartholomew’s Hospital London, UK Gustav K. von Schulthess MD, PhD Professor and Director Department of Radiology University Hospital Zurich, Switzerland Iain D. Wilkinson BSc, MSc, PhD, CSci, ARCP, FIPEM

Reader in Magnetic Resonance & Consultant Clinical Scientist Academic Radiology University of Sheffield and Sheffield Teaching Hospitals NHS Foundation Trust Sheffield, UK A. Robin M. Wilson FRCR, FRCP(E) Consultant Radiologist King’s College Hospital and Guy’s and St Thomas’ Foundation Trusts London, UK David J. Wilson MBBS, BSc, FRCP, FRCR Consultant Musculoskeletal Radiologist Nuffield Orthopaedic Centre and University of Oxford Oxford, UK Stuart J. Yates MSci, MSc, CSci, MIPEM Principal Physicist Department of Medical Physics & Clinical Engineering Cambridge University Hospitals NHS Foundation Trust Cambridge, UK

xiii

Preface

We hope that this 5th edition of Diagnostic Radiology will continue to build on the original vision of Professors Grainger and Allison who, back in the early 1980s, saw the need for a ‘bible’ for doctors studying for postgraduate examinations in radiology, and to provide a bench book for reporting and reference. The success of the first four editions, which were extremely well received by an increasingly international readership, speaks for the realisation of their dream. Few could have predicted at that stage the extraordinary growth of radiology, or its increasing importance within all aspects of modern medicine. The unprecedented expansion in the imaging repertoire, together with the trend for increasing subspecialisation, have led to changes in training and in the methods used for teaching and learning. This book has had to evolve to reflect these changes, adapting to the perceived needs of those facing postgraduate examinations and also to all radiologists who wish to have an up-to-date basic general textbook for ready reference and illustration. An attempt to cover every subject in detail would have resulted in a huge book that would have been very difficult to use.We have chosen to concentrate on those subjects that most radiologists need to know well, and to pay special attention to the needs of trainee radiologists preparing for examinations. Because training throughout Europe is moving towards a three

year basic course followed by two years of training in selected subspecialties, the factual examination in the UK has moved to an earlier stage in training with less emphasis on some of the diagnostic rarities so beloved by examiners of old. The curriculum is now somewhat less comprehensive and the reduction in size of this 5th Edition reflects that – down from three volumes to two. In this electronic age there are many databases of images available on the internet, with accompanying text. Nevertheless, we believe that well structured textbooks remain an essential part of medical education and practice as they present information in a format that facilitates learning, guiding the reader through an unfamiliar field. We are convinced that Diagnostic Radiology will remain a valuable resource for many years to come. We are again extremely grateful to the distinguished international cast of authors who have all worked hard to deliver fresh and up-to-date material. We are also most grateful to Michael Houston, Gavin Smith and Nora Naughton for their professional skills and publishing expertise and to Jeremy Rabouhans for invaluable help with proof reading. We could never have done it without them! Andy Adam Adrian Dixon

Acknowledgements

This edition could not have occured without the large amount of work done by all the contributing authors and their colleagues. However, the vision and overall planning of Michael Houston at Elsevier have been fundamental in bringing the book to fruition. So, too, has the meticulous gathering and editing of material by Gavin Smith. Finally, the skilful copyediting, and other tasks provided by Nora Naughton, and her remarkable team, must not be forgotten; without them the Editors simply could not have managed. All of these col-

leagues remained remarkably cheerful throughout and kept strong heads even when chapters were late, images missing, and all the other hiccups that can hinder progress in a project of this kind. At a local level, all the Editors would like to thank their various secretaries, technicians and colleagues who have helped proofread, collect material and made various other invaluable contributions. The Editors

CHAPTER

Picture Archiving and Communication Systems (PACS) and Digital Radiology

1

Nicola H. Strickland

The role of PACS Advantages and disadvantages of PACS • Advantages of PACS • Disadvantages of PACS Planning for PACS • The PACS project team • Tendering for a PACS • The PACS contract • Economic considerations • Purchasing versus leasing a PACS • PRE-PACS workflow and preparation • Acronyms: DICOM, HL7 and IHE PACS project implementation • Implementation of digital image acquisition prior to PACS • Integration of PACS ‘Value-added’ PACS Modern PACS architecture • PACS networks • Storage requirements and solutions for PACS PACS workstations • Monitor quality on PACS workstations Conference room design Graphic user interface • Soft-copy tools Software concepts • Prefetching

Compression • Lossless • Lossy • Need for compression Digitization policy Teleradiology • Review of images from home • Teleradiology linkage between two or more hospitals for joint MDTMs • Outsourcing of imaging examinations for reporting • Tele-education Security • User-specific log-in and password • Monitor screen savers • Workstation time-outs • Encryption • Audit trails • Firewalls • Teleradiology security PACS training Quality assurance • Plain radiography • PACS workstations PACS housekeeping • Storage commitment • Modality performed procedure step Current and future directions of PACS

THE ROLE OF PACS A picture archiving and communication system (PACS) aims to replace conventional analogue film and paper clinical request forms and reports with a completely computerized electronic network whereby digital images are viewed on monitors in

conjunction with the clinical details of the patient and the associated radiological report displayed in electronic format. Clearly PACS must replace the functions of traditional X-ray film; i.e. image acquisition, storage, transportation and display.

4

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Were these the only roles of a PACS, it would be an extremely complex and expensive means of replacing traditional film. A PACS must improve upon a film-based system, preferably in a cost-neutral manner. The major added value of a PACS is efficiency of data management. True efficiency benefits can only be realized once a PACS is at least hospital-wide, since any more limited installation means running two systems in parallel, i.e. it entails continuing to produce conventional film and moving it around the hospital, as well as the cost of installing and maintaining a PACS. Thus, even if funds are limited initially, it is advisable at least to aim and plan for growing the PACS installation into a hospital-wide system ultimately. This means decid-

ing upon a potential time scale in which the hospital-wide PACS is to be achieved, and deciding upon a scaleable PACS architecture. Ideally the whole hospital infrastructure should be adapted at the outset so that a hospital-wide PACS can be accommodated at a later stage. This includes providing an uninterruptable power supply (UPS), allowing sufficient cabling space in floors and ceilings, and adapting the air conditioning system for PACS. The hospital information technology (IT) network is likely to need upgrading to enable large amounts of image data to be transported, and it is advisable to install multiple PACS ‘drops’ (workstation outlets) so that more PACS workstations can easily be added at a later date.

ADVANTAGES AND DISADVANTAGES OF PACS Although the concept of PACS has now been in existence for over 20 years, advances in computer hardware technology only enabled it to become a realistic clinical entity in the 1990s. PACS installations are now rising exponentially worldwide, and although most institutions with PACS have achieved a completely filmless working environment, few function in a paperless mode. PACS has proved itself over the last 10 years in the clinical environment, however, and it would now be unthinkable not to implement PACS when installing a new imaging department, or a new hospital.

ADVANTAGES OF PACS There are a number of powerful advantages accruing from a hospital-wide PACS: • Once correctly acquired onto the PACS, no image can ever be lost or misfiled and is always available when needed.This is a major benefit considering that in many hospitals up to 20 per cent of conventional films are missing at the time they are clinically required. In addition to the convenience of always having the appropriate image available when it is wanted for viewing, no patient is re-irradiated simply because a previous key study has been lost. • PACS facilitates the easy comparison of a patient’s current and historical examinations, and of examinations performed on the same body part using different imaging techniques, which is desirable in the interests of optimum clinical practice, and always possible since none of the relevant comparative images is missing1. • All images remain accessible from the PACS archives day and night, every day of the year. • Simultaneous multilocation viewing of the same image is possible on any workstation connected to the PACS network, whereas hard-copy film can manifestly only be in one place at any one time. • Image retrieval is infinitely quicker from PACS than it is using conventional film where someone physically has to go and fetch the film packet2. • The benefits of a computerized system mean that all images correctly and permanently reside under the appropriate

•

•

•

•

imaging study, remain in their correct orientation, and are automatically chronologically ordered. Database searches for a particular patient or study are rapidly effected. Computerized data can easily be duplicated, i.e. ‘backed up’ as a precaution against loss, and cheaply stored in a distant location if desired for disaster recovery purposes. Viewing of images on monitors allows numerous postprocessing soft-copy manipulations: a range of different window width and level settings can be applied to CT images, for example, within a fraction of a second, whereas previously further sheets of film had to be printed with the appropriate settings for soft tissue, lung or bone as desired3,4. There are some direct cost savings due to PACS: there is no longer a film budget, film packet cost or chemical processing. (The ability to print film need not be retained provided a CD burner is linked to the PACS to burn imaging studies for transfer of patients outside the institution, or in the event of a PACS failure or planned downtime). Ancillary staff – filing clerks and darkroom technicians – are no longer needed. The most frequently cited benefit of PACS by nonradiological clinicians is the very substantial time saving incurred by their never having to search for or retrieve films. This time saving represents a considerable indirect financial benefit of a hospital-wide PACS, and should certainly be costed when a business case for PACS is being made. After a hospital-wide PACS has been installed, old films can progressively be removed from the film filing room (film store) starting with the oldest first.The timing of this manoeuvre will depend upon local and national policy, but it has been shown5 that most radiological comparisons are made with studies obtained during the preceding 6 months, although this may be longer in institutions with a large oncological or paediatric practice. Maintenance of a film store, with its associated lighting, heating and cleaning, and the value of the physical space itself, is particularly costly in hospitals located in cities where the space is at a premium. Even off-site film storage at a cheaper location is worth introducing as a cost-saving measure in such institutions in the early years after PACS

CHAPTER 1

•

PICTURE ARCHIVING AND COMMUNICATION SYSTEMS (PACS) AND DIGITAL RADIOLOGY

has been installed, before the film store can be dispensed with altogether. (In the UK it is the radiological report, not the images that are deemed to be the legal document, with certain restrictions for paediatric and educationally challenged patients.) • The installation of a PACS infrastructure in an institution (a local area network [LAN]) sets the stage for the introduction of teleradiology over a wide area network (WAN) if desired. Teleradiology offers the potential for improvements in efficiency, for example in geographically remote areas by centralizing a reporting service, or increasing the referrals to a particular institution. A number of perceived potential benefits of PACS have not been substantiated, or not consistently demonstrated, by audit studies. These include the possibility of a reduced hospital inpatient stay or a greater throughput of outpatients6. It is hardly surprising that such benefits cannot be attributed to PACS since there are so many other variable factors that influence these issues. Also the type of study required to prove any given putative benefit of PACS is fraught with practical difficulties. The classical ‘before’ and ‘after’ study comparing the pre-PACS with the post-PACS era is inevitably complicated by the numerous other concurrent environmental, technological (and often political) changes that have taken place in the interim. These include, for example, changes in the medical personnel and in the type of clinical practice pursued in the hospital under study. Studies comparing a PACS institution with a nonPACS institution may be similarly flawed by the difficulty in adequately correcting for other inherent differences between the two institutions, which may or may not be related to the presence of a hospital-wide PACS7.

DISADVANTAGES OF PACS The advantages of PACS described earlier must be set against its potential disadvantages: • Even though the costs of hardware and storage media continue to reduce in price, PACS remains an expensive technology. Most estimates suggest that a PACS installation should aim at becoming cost-neutral in less than 5 years. Some would argue that PACS should be viewed in the same way as any new imaging technique, and as such it represents an advance in health care management and should not be assessed merely in terms of cost–benefit. • The technological complexity of a PACS and the absolute dependency of a hospital on the PACS once it becomes filmless require a dedicated maintenance programme for the PACS and a carefully devised plan detailing how to supply a minimal clinical service should the whole PACS fail for a significant period of time. This inevitably means that there will be a requirement for new or retrained hospital personnel specializing in computer engineering/information technology, as well as a vendor-provided maintenance service, and these costs must be set against the savings made in respect of less highly paid clerical staff and darkroom technicians. • Once a hospital-wide PACS is in operation and film has been withdrawn, there is no ‘fall-back position’. The hospital is no longer equipped to run a film-based service. This is a daunting prospect that may act as an initial psychological deterrent to embarking upon a large-scale PACS project. • Changing from a hard-copy to a soft-copy imaging environment will raise many issues necessitating a change of work patterns involving: the training of the users, maintenance of the system, action to be taken in the event of a PACS failure and the institution of specific quality assurance protocols (see later).

PLANNING FOR PACS THE PACS PROJECT TEAM The planning and installation of a PACS requires the coordinated input of a multidisciplinary team that might beneficially include representatives from the IT department, computer scientists, physicists, nonradiologist clinicians, radiologists, radiographers, nurses, hospital management, the hospital financial manager and, ultimately, a representative from the chosen PACS vendor. This emphasizes that PACS must have ‘buy in’ from the whole hospital/health care enterprise, and is not a ‘radiological toy’. The project team needs a leader with sufficient time to commit to the project8. This leader need not be a radiologist, or indeed a clinician, but must have a comprehensive practical grasp of the clinical workings of a hospital environment as well as an understanding of basic IT issues and the requirements of imaging.

TENDERING FOR A PACS When tendering for a PACS, a detailed request for tender/ proposal (RFT/RFP) will need to be drawn up9,10. This

document should define the clinical and logistical requirements of the PACS, rather than merely stipulating technical specifications. It is up to the radiologists and other clinicians to specify the current and projected future performance requirements that the PACS must fulfill (including the issues of PACS maintenance, uptime and data migration) and it is the role of the vendors to specify how these clinical requirements are to be met technically.

THE PACS CONTRACT Every PACS project should be based on a firm contract between the institution and the vendor, defining the responsibilities of each in detail. The PACS contract serves as a legal document and requires careful wording. Any subsequent changes or additions to the contract should be made as formal addenda ‘change control notices’ (CCNs) to the original contract, to preserve the legal integrity of the document. This also serves as an audit trail, keeping the contract up to date, and simplifying the situation for both purchaser and vendor.

5

6

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

It is important to define what is meant by ‘the life of the system’. This is often taken as being 8–10 years after the completion date.The majority of PACS installations are now based on some form of leasing agreement, rather than an outright capital purchase of the hardware and software. The date at which the maintenance contract will commence, its cost and its terms, all need careful definition. Responsibility must be defined for migration of the PACS archive data at the end of the PACS contract, or in the event of premature termination of the contract. It is advisable to define exactly what is meant by an ‘update’ to the system (generally a minor software release that merely corrects bugs in the system), and what is meant by an ‘upgrade’ (which constitutes a new software installation comprising major new features giving enhanced functionality), whether these are to be included in the purchase price/maintenance agreement, and at what frequency these will occur. An important consideration in negotiation with a vendor is the terms under which the delivered PACS hardware will be replaced if upgrades are released that it cannot support. It is vital to define which party is ultimately responsible for the functioning of the interfaces to the various pieces of imaging equipment, and to other hospital computer systems including the hospital and radiological information systems (HIS and RIS), electronic patient record (EPR), speech recognition dictation system, and electronic remote requesting (order communications) system, if these are already extant in the institution.

ECONOMIC CONSIDERATIONS The main reason for putting in a PACS (ultimately a hospital-wide or larger PACS) is to improve the efficiency of data handling throughout the whole of that health care environment. Much of the benefit will be experienced outside the imaging department itself, which is why PACS must be hospital driven (and funded), not radiology driven. PACS should be part of the hospital-wide information management and technology strategy, with the aim of achieving cost neutrality over about 3 years.

PURCHASING VERSUS LEASING A PACS The options for PACS procurement are broadly 2-fold11: 1 Capital purchase, using the hospital’s capital allowance, or 2 Some form of leasing arrangement. The fundamental difference between the two is that with capital purchase the hospital owns the assets (the PACS), whereas with any leased arrangement the hospital procures a service that contractually provides agreed functional outcomes (the output-based specification [OBS]). The vast majority of PACS installations nowadays are on a leasing scheme with a managed service provided by the PACS vendor. A leasing arrangement means that all the PACS equipment (hardware and software) is provided as a service from

a private company (with a maintenance contract). Nothing is owned by the hospital, i.e. the private company retains all the assets. The advantage of a leasing arrangement is that it allows hospitals with no hope of ever having the large capital sum necessary to purchase a hospital-wide PACS to make a quantum leap in technology to move to a PACS solution, and it transfers the risk to the company. It has to be appreciated, as with renting a house, that the hospital would be left with nothing if it were to stop paying the lease. At the end of the contractual term, the hospital does not own anything. However, in these circumstances the hospital is usually given an option to buy the installed assets (the PACS hardware and software) at ‘a fair market value’, as negotiated with the provider. In a leased service the PACS company provides the PACS hardware and software necessary to meet the workload and performance requirements stipulated by the hospital, both at the time of leasing and in the future. For example, such requirements might include the need to perform 400 000 imaging examinations per year, with a short-term storage of a year, and a display time for these examinations (first image to screen) of 3 s or less.The PACS company is responsible for maintaining a PACS with an agreed level of technology throughout the contractual term. The technological risk rests with the provider. The risk covers the following three areas: • Utilization of the system • Availability of the system to users • Future planning and implementation of new technology into the hospital’s system. The risk is transferred to the service provider by linking their revenues to agreed performance targets for the above. In addition the hospital should be provided with a guaranteed programme of equipment replacement and a guarantee to keep pace with technology. Leasing arrangements vary from about 8–30 years, with some planned ‘built-in’ equipment replacement ‘hardware swap-out’ at 5-yearly intervals, or 3–4-yearly technological reviews with ‘technology refreshments’ of software and/or hardware as may be necessary to provide greater productivity or capability. This technology refreshment may be a chargeable issue depending upon the contract and/or negotiations. Companies generally expect to recoup their capital outlay in approximately 3–5 years in a leased arrangement, but this will of course depend upon the financial model being used.

PRE-PACS WORKFLOW AND PREPARATION A number of careful and detailed preliminary pre-PACS studies of the workload and workflow pattern within the hospital/ health care facility will need to be conducted before the clinical requirements to be met by the PACS can be specified12.This includes documenting how many imaging examinations in each modality are performed annually, the average (and maximum) number of images per examination for each modality, and making a prediction of the rate of growth of the imaging workload.This prediction will be influenced by the expectation,

CHAPTER 1

•

PICTURE ARCHIVING AND COMMUNICATION SYSTEMS (PACS) AND DIGITAL RADIOLOGY

for example, of acquiring a new 64-slice (or greater) multidetector CT machine, the intention to start performing high image acquisition studies in MR such as cardiac or breast imaging and so forth. The predicted image storage requirement should always be an overestimate to allow for unexpected demand. It is also advisable to know which outpatient clinics are performed when, and how many film packets are pulled for each, to give some estimate of the network traffic to be expected. It is also important to define the role of the imaging department in other imaging-related activities, such as the radiological steps involved in conducting multidisciplinary team meetings (MDTMs) and radiological presentations at staff rounds, and undergraduate and postgraduate teaching sessions. PACS must be able to fulfill all these functions. The Imaging Directorate would be wise to know before PACS is installed, exactly how many imaging studies are never reported (for various reasons such as: the film packets are never returned to the department for reporting), the time between image examination acquisition and dictation of a report by a radiologist for each investigation and the time delay (if any) between dictation of the report and the availability of the verified report to other clinicians. These workflow deficiencies need to be addressed and corrected prior to the installation of a PACS, since the PACS will not ameliorate the situation but instead will expose these issues by making such information available in computerized form throughout the PACS institution. Seamless integration with other IT systems, and with imaging acquisition devices, is absolutely vital for a PACS to function successfully, and every modern PACS depends upon adherence to DICOM and HL7 standards and IHE (see next paragraph for an explanation of each) workflow processes to achieve this integration13. Before installing PACS, old equipment and old data information systems will need to be upgraded to a minimum level of DICOM and HL7 compliance, respectively. Often it is cheaper to replace such products with modern versions, rather than to pay to upgrade them. A full inventory must be made of the equipment and IT systems to be connected to the PACS with a precise description of the level (if any) of DICOM or HL7 compliance supported, before a PACS project can be properly planned and costed. The full DICOM conformance statement of every piece of DICOM-compliant equipment (e.g. a computed tomography [CT] scanner, an ultrasound [US] machine, a workstation etc.) is available on the Internet. It is generally the users’

responsibility to list the make and model of all the equipment possessed by the hospital/health care facility, and the PACS vendor’s responsibility to look up and interpret the DICOM conformance statements and to make it clear which pieces of equipment will need upgrading/replacing, and to explain the connectivity limitations if such DICOM upgrades are not undertaken. One of the major causes of interoperability problems when linking equipment from different vendors to a PACS is the use of ‘private DICOM attributes’ by many vendors, which, in simple terms, means that information stored in these particular private DICOM fields is not available to be shared with other apparatus, manufactured by a different vendor, which may be linked to it on a network. This leads to a loss of functionality on the recipient apparatus, for example not being able to post-process scanner images received on a workstation from a different vendor.

ACRONYMS: DICOM, HL7 AND IHE DICOM stands for digital communication in medicine and refers to a worldwide multipart standard to which all modern imaging equipment and PACS must adhere, and has now been extended to other disciplines, including cardiology, endoscopy, and ophthalmology. The DICOM conformance statement of every piece of modern imaging equipment is obligatorily available on the Internet, and the description of the various DICOM attributes possessed by each appliance will predict its connectivity with another piece of equipment. All apparatus is described as being a ‘user’ or a ‘provider’ of services such as storage.The DICOM standard is continuously under development, but each new part of the DICOM standard has to be backwardly compatible with the current DICOM standard. HL7 stands for health level 7 and refers to a worldwide standard for data information systems such as HIS and RIS. It is a less rigorous standard than DICOM. HL7 messages from data information systems are conveyed to DICOM apparatus (including PACS) by a ‘broker’, which acts as an integrating and translation device. IHE stands for integrating the health care enterprise and is not a standard, but a comprehensive workflow descriptor of how processes, such as reporting, for example, are achieved in imaging. Its use of integration profiles eliminates the need to reconcile the details of HL7 messages and DICOM conformance statements among multiple vendors. It is now being expanded outside the discipline of imaging.

PACS PROJECT IMPLEMENTATION The PACS project should be treated as a major IT project and divided into stages (milestones) with specific dates set for the completion of each stage. It may be useful in the UK to base the project management on the PRINCE guidelines published by the HMSO14 for major IT equipment contracts. Only after each consecutive milestone is satisfactorily completed and assessed is the next milestone embarked upon.

Each milestone requires conformance testing and clinical acceptance15. These are distinct entities: conformance testing should be carried out independently by hospital employees as well as by the vendor company itself. Payment is best deferred until after clinical acceptance at each stage. Final payment is only made after agreed satisfactory functioning of the system under loaded conditions, in a clinical setting.

7

8

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

A realistic definition of the expected dates of completion of each milestone (including installation dates), with penalty clauses written into the contract to come into effect if these dates are overrun, protects the institution from major delays since penalties can be extracted from the vendor if such delays occur.

IMPLEMENTATION OF DIGITAL IMAGE ACQUISITION PRIOR TO PACS Plain radiography (chest, abdominal and skeletal plain images) still constitute the majority (usually 60 per cent) of imaging examinations in most general radiology departments. These examinations therefore need to be acquired in a digital format in order to be transferred to PACS. This represents a considerable challenge since conventional plain film work is still such a significant part of the total departmental workload. The other imaging investigations (computed tomography [CT], magnetic resonance [MR], nuclear medicine [NM], positron emission tomography [PET], US, digital subtraction angiography, and fluoroscopy) are either already digital in nature at acquisition or can be rendered so by screen capture/frame grabbing from the acquisition device, and can thus be easily transferred to PACS. There are three basic means of rendering plain radiographic images digital: 1 Digitizing conventional analogue film 2 Photostimulable phosphor plate technology, commonly known as computed radiography (CR). 3 Direct digital radiography. There is considerable new technology associated with acquiring plain radiographic images digitally, whichever method is chosen. In situations other than the opening of a new hospital/health care facility, it is circumspect not to introduce this new technology concurrently with a PACS to avoid the risk of ‘technology overload’ for the users. Once the bulk of the imaging studies (i.e. the plain radiography) is being acquired and stored digitally, the introduction of PACS and the complete withdrawal of film is a less daunting task. The overlap period, in which a digital archive is being built up whilst continuing to distribute film outside the imaging department, should be kept to a minimum for financial reasons: it is obviously expensive to run hard- and soft-copy systems concurrently. A period of 95 per cent of the injected dose is excreted in urine with normal renal function. A very small amount (1 month

35 s

3h

N/A

24 h

N/A

**Amount of excess chelate (mg/ml)

0.40

None

12.00

0.23

None

N/A

28.40

*The higher the figure, the higher is the stability of the contrast agent. **The presence of excess chelate is an indirect marker of the instability of the contrast agent. The higher the figure, the higher the instability of the agent. N/A, not applicable.

Unlike all other Gd chelates, gadobenate dimeglumine has a capacity for weak and transient protein binding and is eliminated through both the renal and hepatobiliary pathways. The hepatic uptake represents 2–4 per cent of the injected dose. It behaves as a conventional extracellular contrast agent in the first minutes following IV administration and as a liver-specific agent in a later delayed phase (40–120 min after administration) when it is taken up specifically by normal functioning hepatocytes.13

CLINICAL USE OF EXTRACELLULAR MRI CONTRAST AGENTS These agents accumulate in tissues with abnormal vascularity (malignant and inflammatory lesions) and in regions where the blood–brain barrier is disrupted. Owing to their rapid equilibration in the interstitial space of both normal and tumour/ inflammatory tissues, the use of dynamic MR imaging after bolus injection makes the best use of the narrow imaging window with a transiently increased tumour/inflammatory to normal tissue contrast. Generally the recommended dose for clinical use is 0.1 mmol kg-1 body weight but up to 0.3 mmol kg-1 may be used, particularly in MRA.13–15

anaphylactoid reaction, which are observed with radiographic iodinated contrast media, may also be seen with extracellular MRI contrast agents but the incidence is much lower. The incidence of mild adverse effects is less than 5 per cent. Life-threatening reactions are very rare, with an incidence around 1:100 000. Fatal reactions may however occur but are extremely rare.13

CONTRAST MEDIUM-INDUCED NEPHROPATHY Extracellular MRI agents are more nephrotoxic than iodinated contrast media in equimolar doses. However, nephrotoxicity after contrast-enhanced MRI examinations is not common even in patients with renal disease as the dose of Gd-based contrast medium required for a diagnostic MRI examination is small in comparison to the doses of iodinated contrast media that are routinely used for CT or other radiographic examinations. However, in patients with marked reduction in renal function (particularly due to diabetes mellitus), contrast medium-induced nephropathy may occur even at the standard doses of extracellular MRI contrast agents (see later section on NSF).16

SAFETY OF EXTRACELLULAR MRI CONTRAST AGENTS

NEPHROGENIC SYSTEMIC FIBROSIS (NSF)

Extracellular MRI contrast agents are well tolerated with a low incidence of adverse effects. There are no differences in the safety of the various agents except when it comes to extravasation in soft tissue; then the osmolality is important and high osmolar agents are likely to induce more local damage. In blood, the osmotic load of all Gd-based contrast media is very low, compared to iodinated contrast media, because only a small amount of the contrast agent is required to produce a diagnostic MRI examination. Adverse reactions such as nausea/vomiting, urticaria, bronchospasm, laryngeal oedema, hypotension and generalized

Recently (2006 and 2007) several reports have suggested that the administration of extracellular gadolinium based contrast agents (Gd-CA) may trigger the development of NSF in patients with advanced renal impairment (GFR < 30 ml/ min) or on dialysis. The incidence of this condition in this group of patients is around 3–5% and the onset varies from a few days to 3 months after administration of Gd-CA. The disease is characterised by scleroderma- like skin changes that mainly affect the limbs and trunk. The induration of the skin can progress to cause flexion contracture of joints. The fibrotic changes may also affect other organs such as mus-

CHAPTER 2

cles, heart, liver and lungs. The disease can be aggressive, in some patients leading to serious physical disability or even death. The considerable majority of reported NSF complications followed the administration of non-ionic gadodiamide, but in a few patients NSF followed injection of gadopentate dimeglamine, gadovestomide and other gadolinium contrast agents.18–21 The stability of the binding of the gadolinium ion (Gd+++) within the chelate could be an important factor in the pathogenesis of NSF and may explain the strong association between gadodiamide and this condition. The stability of the Gd-chelates is influenced by the configuration of the molecule whether linear or macrocyclic and ionicity. Macrocyclic chelates offer better protection and binding to Gd+++ in comparison to the linear molecules. Replacement of carboxyl group by a less strongly coordinating group to produce a non ionic Gd-CA also weaken the binding of the chelate to Gd+++ particularly in the non ionic linear molecule. Stability measurements (Table 2.5) indicate that the least stable molecules are the non ionic linear chelates, and most stable is the ionic cyclic chelate. In vivo, transmetallation with endogenous ions leading to the release of the highly toxic gadolinium ions (Gd+++) is likely to occur with unstable Gd-CA particularly if they remain in the body for a long period, as is the case in patients with advanced renal impairment or on dialysis. Recent reports demonstrated gadolinium in the skin lesions of patients with NSF strengthening the hypothesis that free gadolinium play an important role in triggering this condition. It is advisable at this stage of our understanding to *avoid the administration of non-ionic linear chelates in patients with advanced renal impairment (GFR 10 mm24 to 93% (outperforming colonoscopy) in the detection of polyps > 8 mm25. CTC outperforms double contrast barium enema (DCBE) in all studies. The clinical significance of ‘missed’ and diminutive polyps has also been questioned, given the slow progression of polyps over time26. Moreover, there is wide variation in the reporting of polyps even among ‘expert’ readers27, illustrating that this truly is a technique in evolution from a technical perspective, with a steep learning curve for the radiologist. Multicentre trials are currently underway in the USA (ACRIN) and the UK (SIGGAR 1) to evaluate the use of CTC in the context of a screening programme, but at present the majority view is that colonoscopy should be viewed as the gold standard, with CTC likely to replace the DCBE in the future. Again CAD techniques are under evaluation.

identifying the arterial and venous anatomy of the donor for surgical planning. Following transplantation, we have found MDCT to be particularly useful in evaluating the vascular supply to the graft in cases of suspected ischaemia where US is nondiagnostic (Fig. 4.7).

Genitourinary The evaluation of a patient presenting with haematuria may involve intravenous urography, ultrasound, CT or MRI in combination with cystoscopy. Increasingly, MDCT is being used as the primary imaging investigation as it has a high sensitivity for detecting malignant lesions, calculi and traumatic injuries. The other major causes (infection, coagulopathy, instrumentation) may be diagnosed on clinical history and blood and urine testing.The technique employed should encompass an unenhanced study (to detect calculi), a nephrographic phase (approximately 100 s, to assess the renal parenchyma) and an excretory phase (8–10 min, to assess the ureters). The latter two stages may be combined by giving the intravenous contrast medium in two parts, 8–10 min apart, and then imaging 100 s after the second dose. This will significantly reduce the radiation dose to the patient. As MDCT may not depict flat tumours of the bladder wall, cystoscopy must not be omitted. The increased anatomical resolution and multiplanar reconstructions possible with MDCT have enabled its use for surgical planning. In particular, urologists are able to accurately assess whether resection of a malignant lesion requires partial or complete nephrectomy, whilst potential organ donors may be noninvasively evaluated with an accurate depiction of the renal vasculature.

Paediatric Imaging children with CT poses a number of problems. It requires cooperation with instructions and the need to remain still, requires a high degree of spatial resolution to depict smaller organs and their vessels and exposes patients to a relatively high dose of radiation. Some hold that around 1 per

Hepatobiliary The use of multiple phases of contrast-enhanced imaging has probably had the greatest impact in liver imaging, which is reflected in the extensive US, CT and MR literature on the subject. The detection and characterization of focal liver lesions is largely based on patterns of vascular enhancement, with the hypervascular lesions—hepatocellular carcinoma (HCC), regenerative nodules, focal nodular hyperplasia and adenoma—in particular providing diagnostic dilemmas. In order to optimize contrast enhancement, our practice is to trigger the CT study from the arrival of contrast in the abdominal aorta. CT-based surveillance programmes for early detection of HCC with cirrhosis consist of an arterial and portal venous phase study at the very least, with many centres also obtaining unenhanced images. Both early and late arterial phase imaging can be performed, but no additional benefit for this has yet been shown. In living related liver transplantation, MDCT may be used to determine liver volumes and is increasingly utilized for

Figure 4.7 Coronal MIP image demonstrating hepatic artery thrombosis with collateral arteries in a patient following liver transplantation.

CHAPTER 4

1200–2000 patients undergoing CT might develop cancer because of the effects of the CT radiation, and these risks are greater in children28. Advances in MDCT have led to a significant reduction in acquisition time, which has been shown to reduce the need for sedation of paediatric patients29. Simultaneously, it has become possible to reduce the collimation below 1 mm, dramatically improving the resolution, and thereby not only aiding the diagnosis but also enabling accurate assessment of congenital abnormalities and surgical planning, particularly in patients with malignancy. Unlike in adult patients, imaging in different vascular phases is discouraged because of radiation issues. Therefore it is important to select a single optimal imaging sequence whenever possible (e.g. in a Wilms’ tumour, the late arterial phase will also opacify the renal veins and upper IVC). Optimizing an MDCT study for a paediatric patient should also include reducing the radiation dose as much as is possible while maintaining diagnostic quality. Reducing the kV and/or mAs will significantly reduce the dose to the patient30, and this can now be modulated automatically during imaging on newer machines, giving a constant image signal-to-noise ratio throughout the study. Additionally, increasing the pitch reduces the radiation dose significantly without loss of diagnostic quality30.

WHOLE BODY MDCT IN ASYMPTOMATIC ADULTS Whole body CT imaging of asymptomatic patients is controversial, yet is becoming commonplace in several countries, particularly the United States. Clearly, the potential benefits of identifying early stage malignant lesions or coronary heart disease with one noninvasive test could be enormous, but as yet the case for whole body CT screening is far from proven. When considering these studies, it should be borne in mind that a single CT examination cannot be optimized for looking at all organs at once. Consequently, the sensitivity and specificity of a whole-body study is likely to be significantly lower than with the dedicated organ-specific studies currently being evaluated. Inevitably, further follow-up investigations will be required for incidental findings, resulting in anxiety and distress for the patient and occasionally lead to morbidity and mortality. Moreover, the radiation dose is nontrivial. A recent publication exploring the hypothetical situation of a 45-year-old man undergoing annual screening CT until the age of 75 calculated the overall estimated lifetime attributable risk of cancer mortality to be approximately 1.9% (1 in 50)31. This combination of high cost, low specificity and high risk indicates that it is unlikely that whole body CT screening would be of benefit to the population. In the UK, such CT examinations are not recommended by the National Screening Committee32.

RADIATION DOSE CONSIDERATIONS In the late 1980s, CT represented approximately 2% of radiological investigations and 20% of the collective dose to the

•

MULTIDETECTOR COMPUTED TOMOGRAPHY

population. In 2003–2004, it is estimated that this has increased to 9%, with the dose from CT contributing around half of all radiation from medical exposures33. Since 1997, CT has been designated a high-dose procedure by the European Union, along with interventional radiology and radiotherapy. The reasons for this expansion in practice have been outlined above, with MDCT able to perform increasingly complex diagnostic procedures noninvasively and, in some instances, requiring multiple phases of imaging. Furthermore, unlike conventional plain radiography, where the radiation dose from each exposure is to some extent regulated by automatic detectors (increasing the dose will result in an overexposed, dark radiograph of little diagnostic value), increasing the exposure factors (and patient dose) for a CT study will provide the user with higher quality images. More difficult to assess is the effect that an increasingly litigious society has had on the practice of medicine and in particular the number of requests for studies ‘to rule out’ underlying tumour, pulmonary embolism, etc., even when these are clinically unlikely. The responsibility for reducing patient dose should be shouldered by all parties. The referring clinician should ensure that the radiologist is given full clinical information to ensure that CT is indeed the most appropriate test. The radiologist should ensure that each study is justified, that the imaging protocols are optimized to answer the clinical question and that the dose to the patient follows the ALARA (as low as reasonably achievable) principle. CT manufacturers also play a key role in this area through the continued development of dose modulation and the installation of low-dose preset protocols. In order to maximize patient safety, it is essential that all these issues are addressed in each case.

RADIOTHERAPY Over the past decade, significant advances have been made in the planning and delivery of radiotherapy. The introduction of MDCT enabled the oncologist to map the extent of tumour in three planes, allowing accurate dose planning of irregular shapes while minimizing the dose to the surrounding normal tissues. This 3D conformal radiotherapy (CRT) results in a significant reduction in side-effects. However, there were limitations in corrections that could be made to the delivered dose. More recently, intensity modulated radiotherapy (IMRT) has been developed, whereby each radiation beam is divided into 1-cm2 ‘beamlets’, each delivering a different prescribed dose. This has reduced even further the dose delivered to the surrounding normal tissues through increasing accuracy of delivery. Current research in radiotherapy includes the development of techniques to account for movement during radiotherapy, particularly respiration, with 4D CRT. Moreover, techniques are being developed which take account of the changes to the patient and tumour (weight loss and tumour shrinkage) that occur during a course of radiotherapy, so-called adaptive radiotherapy. It is anticipated that both of these will again further the accuracy of radiotherapy with more precise dose delivery and reduced side-effects. The increasing use of functional imaging of tumours with 18FDG-PET is also beginning to impact on

85

86

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

radiotherapy planning. The co-registration of anatomical and physiological images using CT-PET has resulted in significant advances in lung tumour radiotherapy in particular, as it is now possible to differentiate between the central obstructing tumour and distal collapse.

10.

11.

FUTURE DIRECTIONS The gantry speeds and coverage currently achieved in MDCT are more than adequate for almost all clinical applications. The most notable exception, cardiac imaging, will undoubtedly drive manufacturers to produce yet faster machines with greater coverage until high-quality,‘real-time’ images are achieved, which could potentially directly replace all diagnostic coronary angiography. A significant improvement in resolution will be more difficult to achieve using the current technology with up to 64 banks of detectors.This would require significant improvements in detector efficiency and size reduction. However, the next generation of CT machines may instead employ flat panel detectors. Early reports of prototype machines indicate that spatial resolution is greatly improved, with isotropic voxels of the order of 0.25 mm, with coverage of approximately 20 cm in the z-axis per revolution. However, contrast resolution is relatively poor and data acquisition is still slow on these machines at present. One of the most important potential applications for such high-resolution studies, in addition to those described above, would be imaging of the breast. Early reports indicate that the technique is feasible at a dose comparable to that from 2-view mammography, with the added advantages of better tumour localization and potentially better detection of tumours. The tumour detection and dose characteristics will clearly be of greater importance than the speed of data acquisition for breast work, but increased speed will be crucial for cardiac applications.

REFERENCES 1. Ambrose J, Hounsfield G N 1973 Computerized transverse axial tomography. Br J Radiol 46: 148–149 2. Kalender W A, Seissler W, Klotz E, Vock P 1990 Spiral volumetric CT with single-breath-hold technique, continuous transport, and continuous scanner rotation. Radiology 176: 181–183 3. Tipper G, U-King-Im J M, Price S J et al 2005 Detection and evaluation of intracranial aneurysms with 16-row multislice CT angiography. Clin Radiol 60: 565–572 4. Matsumoto M, Kodama N, Sakuma J et al 2005 3D-CT arteriography and 3D-CT venography: the separate demonstration of arterial-phase and venous-phase on 3D-CT angiography in a single procedure. Am J Neuroradiol 26: 635–641 5. Miles K A, Hayball M, Dixon A K 1991 Colour perfusion imaging: a new application of computed tomography. Lancet 337: 643–645 6. Gillard J H, Minhas P, Hayball M P et al 2000 Evaluation of quantitative computed tomographic cerebral perfusion imaging with H215O positron emission tomography. Neurol Res 22: 457–464 7. Wintermark M, Fischbein N J, Smith W S, Ko N U, Quist M, Dillon W P 2005 Accuracy of dynamic perfusion CT with deconvolution in detecting acute hemispheric stroke. Am J Neuroradiol 26: 104–112 8. Wintermark M, van Melle G, Schnyder P et al 2004 Admission perfusion CT: prognostic value in patients with severe head trauma. Radiology 232: 211–220 9. O’Rourke R A, Brundage B H, Froelicher V F et al 2000 American College of Cardiology/American Heart Association Expert Consensus document

12.

13.

14.

15.

16.

17.

18.

19. 20.

21.

22.

23.

24.

25.

26. 27.

28.

29. 30.

31. 32. 33.

on electron-beam computed tomography for the diagnosis and prognosis of coronary artery disease. Circulation 102: 126–140 Nieman K, Cademartiri F, Lemos P A, Raaijmakers R, Pattynama P M, de Feyter P J 2002 Reliable noninvasive coronary angiography with fast submillimeter multislice spiral computed tomography. Circulation 106: 2051–2054 Ropers D, Baum U, Pohle K et al 2003 Detection of coronary artery stenoses with thin-slice multi-detector row spiral computed tomography and multiplanar reconstruction. Circulation 107: 664–666 Schuijf J D, Bax J J, Salm L P et al 2005 Noninvasive coronary imaging and assessment of left ventricular function using 16-slice computed tomography. Am J Cardiol 95: 571–574 Cademartiri F, Mollet N R, Runza G et al 2006 Diagnostic accuracy of multislice computed tomography coronary angiography is improved at low heart rates. Int J Cardiovasc Imaging 22: 101–105; discussion 107–109 Cross J J L, Kemp P M, Walsh C G, Flower C D R, Dixon A K 1998 A randomized trial of spiral CT and ventilation perfusion scintigraphy for the diagnosis of pulmonary embolism. Clin Radiol 53: 177–182 O’Neill J, Murchison J T, Wright L, Williams J 2005 Effect of the introduction of helical CT on radiation dose in the investigation of pulmonary embolism. Br J Radiol 78: 46–50 Quiroz R, Kucher N, Zou K H et al 2005 Clinical validity of a negative computed tomography scan in patients with suspected pulmonary embolism: a systematic review. JAMA 293: 2012–2017 Prologo J D, Gilkeson R C, Diaz M, Cummings M 2005 The effect of single-detector CT versus MDCT on clinical outcomes in patients with suspected acute pulmonary embolism and negative results on CT pulmonary angiography. Am J Roentgenol 184: 1231–1235 Cham M D, Yankelevitz D F, Henschke C I 2005 Thromboembolic disease detection at indirect CT venography versus CT pulmonary angiography. Radiology 234: 591–594 Tins B J, Cassar-Pullicino V N 2004 Imaging of acute cervical spine injuries: review and outlook. Clin Radiol 59: 865–880 Brohi K, Healy M, Fotheringham T et al 2005 Helical computed tomographic scanning for the evaluation of the cervical spine in the unconscious, intubated trauma patient. J Trauma 58: 897–901 Griffen M M, Frykberg E R, Kerwin A J et al 2003 Radiographic clearance of blunt cervical spine injury: plain radiograph or computed tomography scan? J Trauma 55: 222–226 Elton C, Riaz A A, Young N, Schamschula R, Papadopoulos B, Malka V 2005 Accuracy of computed tomography in the detection of blunt bowel and mesenteric injuries. Br J Surg 92: 1024–1028 Taylor S A, Halligan S, Bartram C I et al 2003 Multi-detector row CT colonography: effect of collimation, pitch, and orientation on polyp detection in a human colectomy specimen. Radiology 229: 1–2 Rockey D C, Paulson E, Niedzwiecki D et al 2005 Analysis of air contrast barium enema, computed tomographic colonography, and colonoscopy: prospective comparison. Lancet 365: 305–311 Pickhardt P J, Choi J R, Hwang I et al 2003 Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 349: 2191–2200 Macari M, Bini E J, Jacobs S L et al 2004 Significance of missed polyps at CT colonography. Am J Roentgenol 183: 127–134 Johnson C D, Harmsen W S, Wilson L A et al 2003 Prospective blinded evaluation of computed tomographic colonography for screen detection of colorectal polyps. Gastroenterology 125: 311–319 Brenner D J, Elliston C D, Hall E J, Berdon W E 2001 Estimated risks of radiation-induced fatal cancer from pediatric CT. Am J Roentgenol 176: 289–296 Pappas J N, Donnelly L F, Frush D P 2000 Reduced frequency of sedation of young children with multisection helical CT. Radiology 215: 897–899 Donnelly L F, Emery K H, Brody A S et al 2001 Minimizing radiation dose for pediatric body applications of single-detector helical CT: strategies at a large children’s hospital. Am J Roentgenol 176: 303–306 Brenner D J, Elliston C D 2004 Estimated radiation risks potentially associated with full-body CT screening. Radiology 232: 735–738 Dixon A K 2004 Whole-body CT health screening. Br J Radiol 77: 370–371 Shrimpton P C, Hillier M C, Lewis M A, Dunn M 2005 Doses from computed tomography (CT) examinations in the UK—2003 review. National Radiation Protection Board, Oxon, UK

CHAPTER

Magnetic Resonance Imaging: Basic Principles

5

Iain D. Wilkinson and Martyn N. J. Paley

• • • • • • • • •

Historical perspective Spin physics Excitation and relaxation: free induction decay and echoes Signal localization: techniques for building images Instrumentation: magnets, coils and computers Physical parameters that provide contrast Pulse sequences Safety considerations Conclusion

This chapter considers the basic principles of the MRI technique. Magnetic resonance imaging (MRI) is a noninvasive method of mapping the internal structure and certain aspects of function within the body. It uses nonionizing electromagnetic radiation and appears to be without exposure-related hazard. It employs radiofrequency (rf) radiation in the presence of carefully controlled magnetic fields in order to produce high quality cross-sectional images of the body in any plane. It portrays the distribution of hydrogen nuclei and parameters relating to their physical surroundings in water and lipids. MRI has progressed over 30 years from being a technique with great potential to one that has become the primary diagnostic investigation for many clinical problems. Its application, initially limited to the neuro-axis, now covers all regions of the body and an increased knowledge base has provided a better understanding of how it can best be utilized, either alone or in conjunction with other techniques, in order to maximize diagnostic certainty. Technical advances have included improvements in spatial resolution, types of contrast and, in particular, speed of imaging. The range of information that can be obtained from a multitude of different types of image contrast is one of its biggest selling points: modern MR is truly multimodal, capable of depicting function as well as anatomy with a high sensitivity to the presence of disease.

HISTORICAL PERSPECTIVE The phenomenon of ‘nuclear induction’, later to be termed nuclear magnetic resonance (NMR), was described independently but almost simultaneously by Bloch1 and Purcell2 and their colleagues in 1946 and for this they were jointly awarded the Nobel Prize for Physics in 1952. Much later, for some reason the term ‘nuclear’ was dropped. In applications to medicine, it is now commonly referred to as magnetic resonance (MR). Since its discovery, NMR has been used extensively as a laboratory method for studying the properties of matter at the molecular level (NMR spectroscopy). In 1971, Damadian noted that in vitro, animal tumours had elevated MR relaxation times when compared to normal control tissue, and in the following year he filed a US patent: ‘apparatus and method for detecting cancer in tissue’. The application of MR to imaging required a method of spatial localization. Published in 19733, Lauterbur’s paper showed how this could be done by applying a linearly varying magnetic field across a liquid. In the same year, Mansfield and Grannell reported revealing structure within a solid by the use of a linear magnetic field gradient4. Human in vivo images were first published in 1977 by Mansfield and Maudsley5, Damadian et al6 and Hinshaw et al7. Multiplanar imaging ability was first demonstrated by Hawkes et al in 1980, who also reported the first demonstration of intracranial lesions8, and 1980 also saw the introduction of the basis of the most common spatial encoding method – spin-warp imaging by Edelstein et al9.The introduction and subsequent improvements in diffusion, perfusion, flow and spectroscopy from a large number of groups have been providing fascinating new insights into cerebral and cardiac pathology10–16. Functional MRI (fMRI) of the brain, which uses endogenous blood oxygen level dependent (BOLD) contrast, was introduced by Ogawa et al in 199217 and developed by Rosen et al18. Methods have been developed that have enabled areas that have previously been difficult to assess, such as pulmonary ventilation, to be imaged with the aid

88

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

of hyperpolarized helium gas19. New parallel encoding methods used with phased-array rf coils, such as SMASH introduced by Sodickson et al in 199720 and SENSE introduced by Pruessmann et al in 199921, are changing the rules of imaging speed, particularly when coupled with the introduction of 3T whole-body systems.

SPIN PHYSICS MR describes the phenomenon whereby the nuclei of certain atoms, when placed in a magnetic field, absorb and emit energy of a specific or resonant frequency. Nuclei suitable for MR are those which have an odd number of protons or neutrons and therefore possess a net charge distribution. They also exhibit the property of nuclear spin, which gives them angular momentum. The combination of charge and angular momentum causes these nuclei to behave as magnetic dipoles. Almost all medical images produced to date have used the simplest of all nuclei, that of hydrogen (a single proton), which is present in virtually all biological material and exhibits relatively high MR sensitivity. Other relevant naturally occurring nuclei include phosphorus (31P), sodium (23Na), carbon (13C) and potassium (39K). Noble gases such as helium (3He) and xenon (129Xe) can also be made sufficiently sensitive by laser pre-polarization techniques. The proton can be regarded as a small, freely suspended bar magnet spinning rapidly about its magnetic axis. Place a group of protons in a uniform magnetic field and their magnetic moments experience a torque, tending to line them up with the applied field. Due to thermal energy not all of the spins line up and at body temperatures the difference in numbers between those that do and those that do not (the net magnetization) is small. The stronger the applied magnetic field, the larger the net magnetization and the higher the available MR signal.The direction of the applied magnetic field conventionally defines the z-axis, which is generally in the craniocaudal direction in the common cylindrical imaging magnet configuration. Because the nuclei are spinning, they respond to the magnetic torque like a gyroscope and their axes are tilted so they rotate about the magnetic field’s axis in a movement termed ‘precession’. The frequency of precession is directly proportional to the applied magnetic field. For protons in a field of 3T, it is almost 128 MHz. This relationship is expressed by the Larmor equation: f0 = γB0 where f0 is the resonant frequency, γ is the gyromagnetic ratio (constant for a nuclear species) and B0 is the applied magnetic field.

EXCITATION AND RELAXATION: FREE INDUCTION DECAY AND ECHOES If rf energy is then applied, there is a strong interaction or resonant effect, providing that the frequency of the rf is equal

to the proton precession frequency. This is called magnetic resonance (MR) and manifests itself in the following way: rf energy is absorbed by the nuclei causing the motion of the nuclear dipoles to be disturbed (they become ‘excited’). Their net magnetization along the z-axis deviates through an angle that depends upon the amount of rf energy absorbed. Once the applied rf is turned off, the magnetization gradually returns to its equilibrium position. As it does so, the changing magnetization induces a small voltage in a receiver coil. In its simplest form, the electrical signal detected following an rf pulse is known as the free induction decay (FID), the size and length of which is determined by the sample’s proton density and relaxation times. The time it takes for the net magnetization to revert back to its equilibrium position along the longitudinal z-axis is governed by the longitudinal or T1 relaxation time. In this way, energy is given up by the spins to the overall structure or lattice (T1 can be called the spin–lattice relaxation time) and once the energy associated with a given excitation pulse has decayed away by this mechanism, it cannot be retrieved: you have to start again by applying another rf excitation pulse, imparting more energy to the system. In addition to the transfer of energy to the overall structure, the spins interact with each other. As this occurs, their net magnetization in the x–y plane (perpendicular to the z-axis) becomes less coherent or dephases. The rate at which it does so is governed by T2, the transverse or spin–spin relaxation time. As will be described below, for a given rf excitation pulse, part of the T2 decay or dephasing can be reversed, leading to the production of echoes. Due to the different nature of T1 and T2 decay, T2 can never be longer than T1. In liquids, T2 starts to approach the value of T1 and in human tissue, T2 is approximately a tenth of T1.The spin–spin interaction (T2 decay) can be hastened by the presence of local changes in magnetic field homogeneity. When this occurs, the transverse magnetization decays with time constant T2*, which is shorter than its intrinsic T2. So far, the production of a FID has been described, occurring after termination of an rf excitation pulse.To facilitate the spatial encoding process and to enable ‘weighting’ of the signal by various contrast mechanisms in differing proportions, it is necessary to delay the production of the signal. Within the overall decay envelope given by T2 relaxation, a signal echo (or echoes) can be formed in different ways. The two most common types of echo are the spin echo and the gradient echo (or field echo) (Fig. 5.1). Both cause rephasing or refocusing of the transverse magnetization (in the x–y plane) at an echo time (TE) following the excitation pulse. Spin echoes are produced by an 180-degree, second rf pulse which flips the spins over in the x–y plane: instead of spreading further apart, they refocus, moving into phase with each other again, pass through a maximum and then dephase once more. Gradient echoes are, as the name suggests, produced by the introduction of a magnetic field gradient: the gradient causes the spins to realign in the x–y plane while they are ‘pointing’ in the same direction (i.e. without flipping them over). At sufficiently long TE, spin echoes and gradient echoes produce signals that are

CHAPTER 5

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

90
rf

A

FID

Echoes

180
rf

B

SR

G

C

GE TE

Figure 5.1 Schematic depiction of signal formation: (A) application of a 90-degree rf pulse leads to the production of the free induction decay (FID); (B) a spin echo (SE) is produced at the echo time (TE) by the application of a 180-degree rf pulse at TE/2; (C) a gradient echo (GE) is produced by the application of a magnetic field gradient (G).

T2-weighted and T2*-weighted, respectively; that is, different tissues having different T2 and T2* relaxation times yield different signals that depend on the values of those relaxation times. There are other echo types: Hahn (or partial) echoes and stimulated echoes. The former will be briefly mentioned in the section outlining gradient-echo sequences.

SIGNAL LOCALIZATION: TECHNIQUES FOR BUILDING IMAGES The physical relationship that makes building up an image possible is the proportional relationship of the resonant frequency to the strength of the magnetic field (Larmor equation again). In understanding the MR spatial-encoding process, two concepts need to be separated: that of resonant frequency (as given by the Larmor equation: spins precessing at a high resonant frequency when in a high magnetic field) and spatial frequency (the high spatial frequency components of an image corresponding to the fine detail in that image). Imaging is performed using the properties of the Larmor relationship but, in addition to this, the data are encoded in the spatial frequency domain! The latter refers to the sampled signal being in the wonderful world of k-space. Different gradient schemes have been devised to traject through k-space in different ways but most images are acquired using a two-dimentional (2D) encoding technique which samples kspace as a set of rectangular Cartesian coordinates (like the regular spacing in a tight climbing net). The beauty of this method is that the image is then just the 2D Fourier transform of the sampled signal (Fig. 5.2). The basic components of this strategy can be split into three parts: 1 Slice selection is performed during the rf excitation process. The excitation pulse consists of a narrow bandwidth of frequencies. A linearly increasing field (field gradient) is applied across the object in the slice-select direction; the combination of the limited excitation bandwidth in the

Figure 5.2 (A) Two-dimensional (2D) ‘k-space’ magnitude data array and (B) corresponding sagittal T1-weighted proton density image formed after 2D Fourier transformation.

presence of the gradient means that only spins within a finite distance in the slice-encode direction (i.e. a 2D slice) will resonate and be excited. The slice position and thickness can be altered by adjusting the rf centre frequency, pulse bandwidth and/or gradient strength (rate of change of magnetic field across the object). A negative gradient lobe is added after excitation so that once slice selection has been accomplished, the spins across the chosen slice precess in-phase with one another. 2 Phase encoding is performed following excitation and prior to the formation of the signal echo. A second gradient is applied along one direction within the plane of the excited slice. While it is on, the presence of the gradient causes a linear change in spin precession frequency along its direction. As with slice selection, once it has been switched off, all spins within the slice revert to precessing at the static field’s resonance frequency. However, the phase-encoding gradient will have introduced an incremental phase shift along its direction, the size of which depends upon its magnitude. This corresponds to one sampled spatial frequency (one value of ky) along the phase-encode direction. In order

89

90

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

to sample other spatial frequencies in that direction, the excitation process is repeated to fill the desired matrix size. During each repeat, which occurs every repeat time (TR), the strength of the phase-encode gradient is altered, so that a different phase shift is introduced, thereby encoding a different spatial frequency in the phase-encode direction. 3 Frequency encoding is performed during data sampling of the signal echo. A third linear field gradient is applied in the imaging plane, perpendicular to the phase-encode and slice-selection directions. At different positions along the frequency-encoding direction, spins will be resonating at different resonant frequencies as the signal is sampled. As the echo evolves, successive spatial frequencies of the object are encoded along this third dimension. Thus, in a basic spin-echo or gradient-echo technique (Fig. 5.3), during any one TR, the slice is selected, the total number of spatial frequencies (say 256 or 512) is sampled along the frequency-encode direction at one spatial frequency along the phase-encode direction. This process is repeated for the total number of spatial frequencies required in the phase-encode direction (say 256 or 512). This can lead to inappropriate, excessive imaging times if high spatial resolution is required (> 25 min for 512 phase encodes with a TR of 3 s). However, developments in sequence design have enabled multiple phase-encoding steps to be performed within each TR, speeding up the process using, for example, fast spin-echo (FSE) techniques (Fig. 5.4). It is important to realize that individual elements in k-space do not correspond on a one-to-one basis with individual pixels in the final image, but rather that each k-space element contains information about that spatial frequency component throughout the image. An alternative method to slice selection is to use an additional (secondary) phase-encoding gradient along the slice selection axis, yielding true three-dimensional (3D)-encoded datasets that typically have high spatial resolution in all three directions (often submillimeter). Putting together phase and frequency encoding plus slice selection has introduced the concept of a pulse sequence that consists of a series of applied rf and gradient pulses that lead to

TR 90


90


RF Slice Phase Frequency TE Signal

Figure 5.3 Basis of the gradient-echo (spin-warp) pulse sequence. Three orthogonal gradients are used for: slice selection (while the rf excitation pulse is applied); phase encoding (the amplitude of which is iterated every repeat time [TR]) and frequency encoding (leads to echo formation at the echo time [TE]). The process is repeated after time TR for the total number of required phase-encoding steps.

RF

90


180


180


TR

180


90
etc

Phase train

Figure 5.4 Train of 180-degree refocusing rf pulses that form the basis of the fast spin-echo (FSE) sequence. Each 180-degree rf pulse has a different magnitude phase-encoding gradient. A whole train of 2, 4, 8, etc. echoes are produced per TR, reducing the number of TRs required for a given total number of phase-encoding steps.

the production of the signal in the form of an echo which has been encoded in k-space.Variations in the timing of the component parts of these pulse sequences can be used dramatically to alter image contrast; this being one of the major advantages of MR, helping to identify anatomy and characterize normal and pathological tissue types. Adjustments to the building blocks and parameters that define a pulse sequence can be used to manipulate image contrast. Most sequences in use today are based on variants of the above. There are several other methods of moving or trajecting through k-space that have potential advantages. Of these, the most commonly used is echo-planar imaging (EPI). EPI was one of the first imaging schemes to be proposed. It requires very rapid gradient switching and is thus technically demanding and, in its early forms, prone to various artefactual complications (such as Eddy current effects). Technical developments have enabled its introduction on commercial clinical systems over the last 10 years. Its speed is a great asset, being capable of yielding an image in under 100 ms, and has led to its use in areas such as imaging of water diffusion and brain function. It is important to note that MRI is a true digital imaging technique as discrete data points are sampled which, when transformed, form an image made up of 2D picture elements (pixels) or 3D volume elements (voxels).

INSTRUMENTATION: MAGNETS, COILS AND COMPUTERS The major parts of an imaging system comprise (A) the magnet subsystem to produce a spatially and temporally constant B0; (B) the gradient subsystem to produce time-varying magnetic field gradients primarily for spatial encoding; (C) the rf subsystem to transmit and receive the rf energy; and (D) the associated microprocessors/computers to specify and control the pulse sequence, calculate, process, display, store and transfer the resultant images. The magnet, gradient and rf coils are all situated within an rf-shielded examination or magnet room. This shielding (a Faraday cage) consists of a conductive metal lining (copper or aluminium) through which rf electromagnetic radiation will not pass. It serves to keep external electrical noise out and the generated rf energy within the examination room. All electrical connections into the room must be filtered appropriately as they pass through this shield. Most screened rooms possess wave guides or tubes (with a high length-to-diameter ratio; > 5:1) fixed into the wall of the

CHAPTER 5

screened room, through which nonconductive tubing may be passed (plastic gas tubing or fibre-optic cabling).The examination room door should be shut while acquiring data to enable the shield to work effectively.

Static magnetic field (Fig. 5.5) Although it is possible to perform MR experiments within the earth’s magnetic field (being approximately 0.00004T!), most clinical MRI systems are constructed around magnets which produce fields in the range 0.2–3T. Research systems capable of human in vivo imaging at 8T and above are available. There has been much debate over the optimum B0 for clinical imaging and this debate continues.The vast majority of installed systems operate at 1.5T, although there is presently a surge in systems operating at the new high-field strength of 3T. Low-field, reduced cost systems (0.2–0.5T) are often marketed for various ‘niche’ applications (e.g. musculoskeletal extremity imaging). At this low end of the range, permanent ferromagnetic composites or room temperature electromagnets tend to be utilized. Above 0.5T, superconducting windings (a niobium–tin alloy embedded in a copper matrix) are supercooled in a bath of liquid helium (at 4.2 K or −269°C) such that they offer zero resistance to the high currents required to produce high magnetic flux densities. Superconducting magnets, which are constantly ‘on’ or ‘at field’, are mostly cylindrical in geometry, especially at 1.5T and above, although more ‘open’ designs are becoming available at 1T and below. Developments in magnet technology have meant that, provided the helium refrigeration pump remains in operation (usually a rhythmical thud can be heard while the system is not acquiring data), one fill of liquid helium should last several years. The overall size of these superconducting magnet systems has decreased substantially in recent years, yielding friendlier designs with reduced levels of patient anxiety/claustrophobia. Why buy a high B0 system? The protons’ net magnetization increases with B0, leading to an overall increase in imaging signalto-noise ratio (approximately a linear relationship). Various other parameters also change with increasing B0, leading to

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

potential benefits such as increased magnetic susceptibility contrast (useful for depicting nuclei within the brainstem or fMRI responses), greater resolution between resonance peaks in spectroscopy (potentially helpful for water/fat imaging) and an increase in tissue T1s (prolonged blood signal and better background suppression for time-of-flight angiography). However, higher is not automatically better as the aforementioned can also produce increased metallic foreign-body artefact, increased chemical shift artifact and changed T1 contrast. At high resonant frequencies, rf homogeneity also provides greater technical challenges. A high degree of homogeneity of the main magnetic field together with gradient linearity are essential for correct geometrical representation, this being especially pertinent when MR is used directly for therapy (e.g. in stereotactic radiosurgery to the trigeminal nerve or the placement of deep-brain stimulators). To maximize B0 homogeneity, a combination of passive and active shimming is employed. The former can take the form of a series of metallic coin-like discs or nougats, the configuration of which is usually site specific, determined and placed during installation. In high-end systems, active shimming can be computer optimized on each imaging volume for each patient episode.

Magnetic field gradients In cylindrical systems, the gradient coil sets required to produce the three orthogonal, linear gradient fields (gradients), are located adjacent to the inner surface of the magnet cryostat. The maximum gradient amplitude that can be generated (given in milliTesla per metre) has a significant effect on the minimum slice thickness and field of view and, thereby, the highest spatial resolution that can be achieved. The rate at which the gradients can achieve the desired amplitude is known as the gradient slew rate (given in milliTesla per metre per second). This sets a limit to the minimum TR and TE values that can be achieved, and thereby the minimum scan acquisition time. Applications such as fast cardiac, MR-digital subtraction angiography (MR-DSA), or high sensitivity diffusion tensor imaging require high amplitude gradients with a

Figure 5.5 MRI systems: (A) Modern system based on a cylindrical cryogenic magnet; (B) open magnet for interventional MRI; (C) low-field system for niche applications.

91

92

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

fast slew rate. Fast gradient switching can lead to the induction of artefact-producing eddy currents within the conductive structures of the system. Improvements in gradient coil design and compensation methodology have significantly reduced this problem, although correct set-up during installation is essential for artefact-free use of gradient-intensive sequences such as echo-planar diffusion or sensitive techniques such as gradient localized spectroscopy. Powerful gradient amplifiers are an essential component of the gradient subsystem and are sited in a room with restricted access, adjacent to the magnet room.

Radiofrequency field Many of the rf transmit and receive subsystem components (digital-to-analogue converters, frequency mixers, power amplifiers, analogue-to-digital converters, etc.) are also sited within the electronics/cabinet room. These comprise all but the rf coils and associated receiver pre-amplifiers, which are placed close to the anatomical area under investigation in order to maximize their effectiveness. A typical system will have been purchased with a number of signal-detecting rf ‘receive’ coils, each dedicated to investigating a particular anatomical area (Fig. 5.6); often used in conjunction with the main rf transmitter body coil located adjacent to the gradient coil sets. The body coil is designed to provide a homogeneous rf excitation field throughout the required imaging volume. Many of the receiver coils will comprise several coil ‘elements’. Coil sensitivity needs to be maximized and the best signal-to-noise ratio can be obtained from a small volume adjacent to a close

fitting coil, but small coils only detect from a limited volume. To achieve high sensitivity over a large volume or area, several coil elements can be used in ‘phased-array’ configurations (e.g. for imaging the entire length of the spinal cord) or signals from multiple coils themselves (e.g. a coil ‘matrix’ covering the whole body) can be used in tandem.To facilitate this multicoil technology, many rf channels are used to pipe the detected signals away from the examination room. Thirty-two channel systems are available as standard product while those under research development can contain over 100. The number of coils/elements used is likely to grow substantially over the next few years. In addition to signal-to-noise ratio considerations, the inherent spatial information provided by arrays of coils/elements can be put to good use in parallel imaging techniques, where SENSE and SMASH technology can reduce the number of required phase-encoding steps and hence imaging time to yield a prescribed resolution. This is particularly pertinent to high-field strength systems where there is sufficient signal-to-noise ratio.

Pulse sequence control, data manipulation and image handling MR depends on highly sophisticated computing. The relentless increase in computing power has had a direct beneficial effect upon performance. Real-time image reconstruction is common without noticeable delay following data acquisition, which is important when hundreds (e.g. tissue perfusion studies) or thousands of images (e.g. BOLD fMRI studies) are acquired during a single examination. It is also now possible to

Figure 5.6 Examples of dedicated rf receive coils: (A) ‘CTL’ phased-array coil for imaging the whole spine plus typical anatomical coverage (B).

Continued

CHAPTER 5

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

Figure 5.6 cont’d (C) a neurovascular coil that allows MR angiography from the circle of Willis to the aortic arch (D); (E) a dedicated wrist coil designed to produce high-resolution imaging of the wrist joint (F).

register two or more datasets accurately with each other using complex computationally intensive algorithms. Slice positions can be prescribed automatically using such image registration techniques. Highly accurate comparisons between images from the same examination can be obtained ‘on the fly’ (e.g. statistical analysis of BOLD fMRI data) or from different examinations (e.g. subtraction of images to determine parenchymal volume changes in dementia or slowly evolving lesions such as meningioma). The real-time manipulation of 3D datasets (maximum intensity projections, surface rendering or general multiplanar reformatting) enables complex interactive postacquisition viewing strategies to be employed by the radiologist, which is particularly important when assessing complex vascular structures such as arteriovenous malformations or visualizing intracranial cortical developmental malformations. Digital data processing speeds will become even more critical if

ultra-high resolution imaging at very high field strengths (≥ 3T) with acquisition matrices of at least 2048 × 2048 becomes widespread. The common PC–user interface can be found on the majority of new MR systems. The introduction of picture archiving and communications systems (PACS) and the data standard, DICOM, has been evident. Local archive is still possible (commonly to DVD in new systems), but the provision of a central PACS archive is often assumed. The introduction of soft copy reviewing and reporting has meant that the radiologist is no longer confronted with fixed viewing parameters on sheets of film, but has the advantage of being able to interact with window widths and levels during the review process, enabling the interrogation of features that may be outside the acuity of the human visual system when observed on hard copy format. The incorporation of a plethora of functional investigative techniques into clinical practice (perfusion,

93

94

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

exogenous contrast medium uptake, water diffusion, etc., and not just BOLD fMRI) often benefits from representation in colour overplayed on high resolution anatomical images. The rapid cinematic presentation of multiple contiguous sections can facilitate integration of information across several sections. Ciné presentations are frequently used in the analysis of dynamic studies, particularly of the vasculature.

PHYSICAL PARAMETERS THAT PROVIDE CONTRAST MR offers a vast range of possible image contrasts (Fig. 5.7), which, in turn, provides high sensitivity about the presence of many abnormalities. However, pixel values do not represent an absolute measure (e.g. absolute T2, in milliseconds). They are only absolute on the rare occasions when specialist quantitative techniques are invoked (e.g. calculation of absolute T2 of the hippocampal structures for the definition and lateralization of sclerosis in temporal lobe epilepsy). Rather, they reflect various physical parameters, including proton density, T1, and T2, and their scaling is influenced by factors such as body size (which affects rf coil loading and hence the induced voltage in the receiver coil). This is unlike CT, where pixel values in Hounsfield units correspond directly to X-ray attenuation (electron cloud density). When a particular physical parameter, e.g. the spin–spin relaxation time T2, dominates the relative pixel intensities in the image (i.e. the image contrast mostly reflects differences in T2), that image is said to be T2 weighted. The diversity of available weightings can lead to difficulties assessing what is the optimum amount of information to gather. As well as in-depth anatomical knowledge, interpretation requires knowledge regarding the type of weighting present in an image, how that relates to a particular physical contrast mechanism and how that contrast varies in the normal population and with disease. This section outlines the most common contrast mechanisms in common clinical use and emphasizes how they can manifest as artefacts or false image information if not properly analysed. Some examples are given.

Proton density It seems intuitive that the proton density (PD or ρ) of an object will fundamentally be linked to the size of the signal returned by that object. Indeed, this is the case; the higher the number of ‘MR visible’ protons, the greater the potential signal becomes. To provide image weighting so that the PD dominates pixel intensity, the effects of other potential influences should be minimized. This approach to providing the majority weighting by one factor is common practice in sequence design and protocol implementation (see ‘sequence’ section below). It is useful to rank body components according to their relative PDs. Highest are cerebrospinal fluid (CSF) and other fluids. These are followed in order by various soft tissues such as liver, kidney, spleen, grey matter and white matter, then articular cartilage, fibrocartilage, membranes, cortical bone and, lowest of all, air. It is important to recognize that the ‘MR visible’ proton pool consists of mobile protons. Hydrogen nuclei that form part of large molecular structures or are immobilized in solids have tight chemical bonds that ensure that they interact with each other so rapidly (i.e. have very short T2 values) that they provide no detectable signal over the timescales involved in standard clinical MR technique, and appear to have a PD of zero. PD is increased in some lesions.These include oedema, infection, inflammation, acute demyelination, acute haemorrhage, various tumours and cysts. Since the formation of oedema is a common response to a wide variety of insults, an increase in PD is a frequent occurrence. A decrease in PD may be seen in scar formation, fibrosis, some tumours, capsule and membrane formation, as well as with calcification. The changes in PD are often relatively small in comparison to the changes seen on T2-weighted images.

T1 (longitudinal relaxation time or spin–lattice relaxation time) This time constant characterizes the rate of the protons’ recovery from their excited state to their resting or equilibrium state, in the longitudinal or z-direction (parallel to the main field), for each component of the object. It relates to the transfer of energy from the protons to the tissue’s overall structure. Body

Figure 5.7 Patient with multiple sclerosis with plaques of demyelination shown on (A) fast spin-echo (FSE) proton density; (B) FSE T2; and (C) FSE FLAIR. There is no discernible abnormality on T1-weighted images without contrast medium (D).

CHAPTER 5

fluid T1 is long (typically 1500–2500 ms) compared to soft tissues (typically 300-1200 ms), while those of fatty structures are shorter still (100–150 ms). T1 increases with B0, requiring an alteration in sequence parameters to provide equivalent T1 image contrast at different field strengths. For any tissue, T1 is always longer than T2. Although the relaxation processes characterized by T1 and T2 are physically different, both tend to increase with most disease processes, primarily due to the accumulation of fluid associated with oedema, inflammation and infection. Despite an overall increase in T1 with pathology, T1-weighted images are often favoured for the depiction of anatomical structure. Standard T1-weighted spin-echo images can be acquired in less time than their corresponding T2-weighted counterparts, due to the reduction in TR used to introduce the degree of T1 weighting. Images with T1 weighting are heavily influenced by the presence of the most common exogenous contrast agent, chelated gadolinium (Gd) (see section below on ‘exogenous contrast media’), which can cause a dramatic increase in signal on T1-weighted imaging (Fig. 5.8).

T2 (transverse relaxation time or spin–spin relaxation time) This time constant characterizes the rate of phase dispersion of the spins’ magnetization in the transverse plane (perpendicular to the main field). This fanning out of the x–y magnetization occurs as the protons exchange energy (or quantum, mechanical chatter) with each other. For a given tissue, the dependence of T2 on B0 is small when compared to that of T1. Fluids have long T2 values (typically 600–1200 ms) those of soft tissues being shorter (typically 50–200 ms) and fat shorter still (10–100 ms). All of these are shorter than their corresponding T1 values: for a given tissue, absolute T2 can never exceed T1. The elevation of T2 in disease processes leads to a very high

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

sensitivity using a standard T2-weighted spin-echo imaging technique, which is often the ‘bread and butter’ sequence for detecting lesions. When exogenous contrast media are administered, T2 is shortened. This mechanism is commonly used in abdominal imaging. For the use of Gd-chelates in the central nervous system (CNS), this effect is generally less and rarely used by comparison to its corresponding T1 effect. However, monitoring the passage of rare earth metals such as Gd is routinely performed using its modulatory effect on a closely related parameter, T2*.

T2* and endogenous susceptibility The concept of T2* was introduced above. When an image is weighted by T2*, not only is it weighted by tissue T2, but also by another component that can be ascribed to the hastening of the intrinsic spin–spin interaction by perturbations in the local magnetic field. Such perturbations are caused by the presence or introduction of objects that have different magnetic susceptibilities (χ). Magnetic susceptibility defines how magnetized an object becomes when placed in a magnetic field. When adjacent structures inside the homogeneous MR magnet have different χ, a small magnetic field gradient will be present. This gradient is the perturbing agent. Most of the constituents of the body are diamagnetic, having weak susceptibility, and they lead to a subtle reduction in the field that passes through them. Exogenous contrast agents (e.g. Gd-chelates) and some endogenous agents (such as deoxyhaemoglobin) exhibit paramagnetic properties which enhance local magnetic flux. Ferromagnetic objects exhibit an extreme form of paramagnetism, experiencing strong forces when placed in a magnetic field. Increasing B0 leads to shortening in T2* and an increase in T2* contrast. Artefacts can result from local decreases in T2*. Signal dropout adjacent to nonferrous metallic implants can be dramatic, even on low- and mid-field strength systems

Figure 5.8 Typical appearance of a brain tumour, specifically a cerebellar medulloblastoma. (A) The tumour has a low signal on spinecho T1-weighted images owing to a prolonged T1. (B) After the injection of Gd-DTPA the tumour enhances avidly, depicting breakdown of the blood–brain barrier.

95

96

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

(0.2–1T). At high field, signal dropout occurs at the edges of various intracranial structures, such as the base of the frontal lobes, due to differences between the χ of cortical parenchyma and air within the sinuses. Dropout can be minimized by the use of spatially localized high-order shimming and increasing sequence bandwidth (number of Hertz per pixel). The presence of endogenous susceptibility gradients can be put to good use, providing a valuable contrast mechanism. One such example is BOLD f MRI, used not just within clinical neuroradiology, but also within associated neuroscientific specialities such as psychiatry and psychology. It is able to detect regional haemodynamic responses to simple or complex stimulation tasks. Although still not totally understood, neuronal/synaptic activity results in a change in local energy consumption which gives rise to an alteration in the ratio of intravascular oxy- and deoxy-haemoglobin and blood flow: the result being a localized net increase in the amount of fresh oxyhaemoglobin and hence T2*-weighted signal.This contrast is very subtle: activity within the primary visual or motor areas leads to signal changes of the order of 2–3% at 1.5T which approximately doubles to 4–6% at 3T. Much data have to be collected, processed and compared using computer-intensive algorithms to obtain statistically significant contrast. Clinical applications include aiding pre-operative neurosurgical planning and intra-operative neurosurgical guidance (Fig. 5.9). There are numerous other important uses of endogenous susceptibility-related, T2*weighted contrast, such as the depiction and timing of blood breakdown products following haemorrhage (Fig. 5.10).

Exogenous contrast agents Several agents have been tested, approved and are commercially available. Two groups will be considered: Gd compounds (the most common) and super-paramagentic iron oxide particles (SPIOs). Being a rare earth metal, Gd is toxic in isolation. Its toxicity is negated by appropriate chelation: the initial product to be introduced uses diethylenetriamene penta-acetic acid for this purpose (Gd-DTPA, Magnevist, Schering AG). Predominantly administered by intravenous injection, Gd-DTPA is rapidly distributed

Figure 5.9 Blood oxygen level dependent functional MRI is being investigated for its use in aiding pre-operative neurosurgical planning and intra-operative neurosurgical guidance. These data were obtained from a patient with a glioma in the left parietal lobe. Areas of activation due to finger tapping are overlaid in orange/yellow. Two anatomical slices are shown for right-hand motion and left-hand motion. Activation that correlates with right-hand movement can be seen close to the top of the lesion (bottom right).

throughout the vasculature. It is excreted by the kidneys, having a biological half-life of approximately 90 min. The agent lowers localized T1,T2 and T2*, as outlined above. Decreased T1 leads to bright pixels on T1-weighted spin-echo images (the most commonly used contrast mechanism [see Fig 5.8]), and decreased T2 can lead to less intense pixels on heavily T2-weighted images, particularly at high Gd concentrations. The effect on T2* is dramatic: the decrease causes low signal on T2*-weighted images which, on a suitable dynamic dataset, depicts parenchymal perfusion in real-time (Fig. 5.11; see section on ‘perfusion’ below).

Figure 5.10 Haematoma. A patient presents with a history suggestive of subarachnoid haemorrhage but MRI shows an extensive left frontal haematoma. The haemoglobin in the haematoma is in different stages of breakdown as shown on (A) the spin-echo T1- and (B) fast spin-echo T2-weighted images. (C) Note the high signal rim on FLAIR imaging indicative of oedema. (D) Gradient-echo sequences are very sensitive for acute haemorrhage and show prominent ‘blooming’ of reduced signal due to susceptibility effects.

CHAPTER 5

A

INTENSITY v, TIME

B

INV. INTENSITY v, TIME

0,00 13,2 28,6 39,7 52,9 66,2 79,4 92 1

2

C

ROI AREA 1. 426 4 2. 363 9

PEAK 96.2 87.6

D

0,00 13,2 28,6 39,7 52,9 66,2 79,4 92,6sec 1

2

Figure 5.11 Exogenous perfusion data obtained from a time series of T2*-weighted echo planar images in a 70-year-old woman who presented with amaurosis fugax and was found to have a 95% stenosis of the right internal carotid artery. (A) A base image shows two regions of interest within the middle cerebral artery territories. (B) A drop in signal intensity can be seen due to the first pass of the Gd-chelate from which (C) a concentration–time curve is calculated and a gamma-variate fit is performed (solid lines). (D) The resultant time-to-peak (TTP) map shows prolonged TTP in the affected hemisphere shown as high signal.

Gd-chelates do not cross the intact blood–brain barrier, but accumulate where it has been compromised (e.g. intracranial tumours, abscesses and acute demyelination).This has been one of their primary clinical uses. They are particularly effective in the delineation of small tumours, such as acoustic neuromas. High-grade malignancies generally show more enhancement than tumours of lower grade. Areas of cystic necrosis may also show enhancement but the greatest clinical value of chelated Gd has been in distinguishing tumour from oedema. Care should be exercised as the integrity of the blood–brain barrier may be altered by the modulatory influences of corticosteroids. Applications outside the CNS benefit, for example, dynamic contrast uptake characteristics may aid breast lesion differentiation. Having Gd-chelates ‘on-board’ can increase overall vessel conspicuity when using standard TOF or phase-contrast angiography techniques, although the most common angiographic application is in contrast-enhanced MRA (CE-MRA)

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

and MR digital subtraction angiography (MR-DSA), where blood T1-shortening effects are used to visualize the passage of a bolus of contrast medium. If a particularly ‘tight’ bolus is needed, higher concentrations are available (e.g. 1 M Gadovist, Schering AG). Oral preparations can be used to visualize the gastrointestinal tract. The other class of agents considered here, SPIOs, is primarily for use in liver (Fig. 5.12) or spleen imaging (Endorem, Guerbet; Resovist, Schering AG).These particles have also been used in oral form for labelling the gut. In clinical applications to date, relatively large particles have been used which, after intravenous injection, lodge in the reticulo-endothelial system (Kupfer cells). Once lodged, they cause strong perturbations to the main magnetic field, producing a loss of transverse coherence, shortening T2. Tumours or other lesions which do not have reticulo-endothelial cells do not take up the SPIO and retain their high MR signal on T2- or T2*-weighted imaging. Lesions down to a few millimetres in diameter can be delineated in the liver.These iron-based agents are metabolized into the iron pools of the body and are excreted over a period of weeks. A selection of other exogenous contrast agents is available or under development, including coated iron-oxide and Gdcomplex blood-pool agents/strong albumin binding agents, hepatocyte-specific Gd-chelates (MultiHance, Guerbet), manganese-containing agents (Teslascan, GE-Healthcare) for the liver, pancreas and cortex of the kidneys; Mn-DPDP may be useful for assessment of acute myocardial ischaemia. Natural compounds, such as blueberry juice, may act as negative contrast agent in upper abdominal MR investigations, such as MR cholangiopancreatography. Other targeted agents, which may be necrosis specific (bis-Gd-mesoporphyrin), provide lymphographic contrast, or are specific for inflammation detection. The exact role of these agents is currently unclear, but they could considerably alter the practice of radiology.

Figure 5.12 Liver contrast agents—super-paramagnetic iron oxide particles (SPIOs). (A) Pre and (B) post iron oxide single-shot fast spinecho images. Both areas of FNH take up the iron oxide. Notice the central scar which has high signal on T2 and low signal on T1. (Courtesy of Dr A Blakeborough, Royal Hallamshire Hospital, Sheffield, UK.)

97

98

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Chemical shift, spectroscopy and water/fat imaging The phenomenon of chemical shift (δ) has important consequences for imaging as well as MR spectroscopy (MRS). In addition to the provision of clinical information, an overview of MRS can aid the understanding of many aspects and underlying principles of MR. Analytical chemists have used bench-top spectrometers since the 1950s in their quest to identify, understand and model molecular structure. The technique relies on information provided by chemical shift. In proton MRS (H-MRS), a hydrogen nucleus attached to a particular molecule does not ‘see’ the exact field produced by the magnet. This is because it is not a free proton; rather it is a proton in a chemical bond attached to a molecule. Neighbouring atoms that form that molecule have associated electron clouds which shield the hydrogen nucleus from the main magnetic field. The shielding effects within different molecules will differ. Back to the Larmor equation, since the resonant frequency is proportional to the field, the hydrogen nucleus will resonate at a particular frequency that is characteristic of that molecular environment. A measurement of the frequency components that emanate from a sample (in the absence of a frequencyencoding imaging gradient) will yield a frequency spectrum. This usually contains a number of peaks, the position of which (or shift along the frequency axis) is given in units of parts per million (ppm) of B0. Each peak can be attributed to a chemical group. Thus, MRS provides direct biochemical information. This can be spatially localized using a number of methods: by the spatial sensitivity of a surface coil, by the intersection of three slice-select gradients (single-voxel selection [SVS]), or a 2D or 3D metabolite map can be created by phase encoding the spectroscopic signal (chemical shift imaging [CSI]). It has become feasible to assess clinically important chemi-

cals and metabolities which are present in tissues in fairly low concentrations (millimoles in proton spectroscopy of the CNS). Spectroscopy is demanding in terms of the signal-to-noise ratio of the system and B0 homogeneity. The homogeneity within the spectroscopic region of interest must be such that any changes in resonant frequency are dominated by the chemical shift and not any heterogeneity in the applied magnetic field itself. Spectroscopy packages (particularly proton) have developed over the last 25 years to a stage where complex high-order shimming (B0 homogeneity optimization), signal localization, and data sampling and post-processing can be performed by ‘one-press’ button automation. Since its implementation in whole-body medical imaging systems, MRS has sometimes struggled to find a true clinical role. In part, that may follow from being seen as a disparate, complicated technique compared to MRI. On the contrary, it is often simpler. The technique has become more robust, forming part of a clinical investigative brain examination in many centres. It is often sensitive to the presence of disease, but lacks specificity; its clinical use (as for many imaging techniques) relies on other contextual information. Just as with standard imaging, spectra can be metabolite concentration-, T1- or T2-weighted, or a mixture of the three, depending upon acquisition parameters such as TE and TR. The brain has been the most extensively studied organ to date by proton MRS (Figs 5.13 and 5.14). If left alone, the hydrogen nuclei from water will dominate the MR ‘visible’ cerebral metabolites. This large peak from water (at 4.77 ppm) needs to be suppressed, so that an appropriate dynamic range can be obtained for metabolite detection. The three main cerebral metabolites are attributed to choline-containing

Figure 5.13 Two examples of the applications of proton spectroscopy in neuroimaging. (A) The patient is a child with known mitochondrial abnormality who presents with seizures. MRI shows extensive areas of hypoxic damage and proton spectroscopy shows the doublet characteristic of lactate. Lactate is not normally seen. The other three major peaks (choline, creatine and N-acetyl aspartate [NAA]) are normal except for a mild reduction in the Na/Cr ratio. (B) In comparison, a child with a cerebellar tumour shows a massive increase of choline and complete loss of NAA relative to creatine. This is highly suggestive of tumour and characteristic of medulloblastoma, which was confirmed on histology.

CHAPTER 5

Figure 5.14 A ‘metabolite map’ or chemical shift image (CSI) where phase encoding is used in two dimensions to map the spectral metabolites. This example shows the distribution of N-acetyl aspartate in a patient with a periventricular space-occupying lesion. A colour overlay is used. (Courtesy of Dr P. Pattany, University of Miami, Florida, USA.)

compounds (Cho) centred at 3.2 ppm, the total creatine pool (creatine plus phosphocreatine) at 3.0 ppm, and N-acetyl (NA) groups at 2.0 ppm (such as N-acetylaspartate [NAA] and N-acetyl-aspartyl-glutamate [NAG]). The biochemical role of NA is not completely understood, but the peak serves as a putative marker of neuronal integrity/function, being predominantly confined within neuronal cell bodies and axons. In a variety of disorders the NA signal may be decreased when there is little or no change on MRI. These include diseases in

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

which there is acute or chronic neuronal loss, such as some of the dementias, chronic head injury, temporal lobe epilepsy and infective encephalopathies. In the latter, in particular HIV-associated encephalopathy, spectroscopy has been shown to follow clinical neurological status more closely than imaging following the administration of antiretroviral medication. Proton MRS appears complementary to MRI, which is generally of most value in acute and subacute disease. At short TEs (where T2 and modulation effects are reduced) more detail on other biochemicals within the brain, such as glutamate, glutamine and myo-inositol can be obtained. Phase-encoded CSI yields metabolite images whose spatial resolution is sufficient to allow differentiation between normal tissues and pathology. Potential applications of proton MRS to other organ systems have been investigated. Most notably it has been investigated for the diagnosis and staging of prostate cancer where the level of citrate decreases, while the choline peak is high. This is the reverse of findings in the normal peripheral zone of the prostate, where citrate is high and choline low, and benign prostatic hypertrophy with intermediate citrate and choline levels. From an imaging point of view, the slightly different resonant frequencies of fat and water can lead to band artefacts at tissue boundaries (the component water and fat images are shifted with respect to each other).These can be minimized by increasing the range of resonance frequencies covered along the frequency-encode direction (acquisition bandwidth); as ever, there is a trade-off as more noise is sampled. Chemical shift can be used to provide images of just water or just fat, or their relative contributions can be altered. Fat- or water-only images can be produced by selective excitation, where the centre frequency of an rf excitation pulse (ChemSat) of narrow bandwidth is placed over either the water or fat peak (Fig. 5.15). Another method utilizes interference effects between fat and water signals. It is useful to consider the phase of the signal. Because of the difference in resonant frequency, their signals will alternatively be in and out of phase. The detected signals cancel out when they are 180 degrees out of phase and add together when they come

Figure 5.15 Saturation of orbital fat using a chemical shift selective pulse. (A) Axial fast spin-echo T2 images in a 2-year-old child showing proptosis of the right eye. (B) Axial images through the orbits show a mass intermingled with the orbital fat and conal musculation. (C) The anatomy is clarified using a post Gd-DTPA fat-saturation sequence.

99

100

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

back into phase. The phenomenon is like the beating effect from two musical notes which are slightly out of tune. At 1.5T, the first water–fat cancellation point occurs at TE = 2.2 ms (Fig. 5.16). Such imaging can provide invaluable information in certain areas (e.g. adrenal imaging).

Flowing spins The given description of the spatial encoding process in the earlier section makes a bold assumption for the production of artefact-free images – that protons stay still while they are being excited and their position is encoded. This is often a tall order for the patient in a magnet with an itchy nose and, of course, impossible for humans to have totally still spins while they have a functioning circulatory system! The different MRI behaviour of fluids can be harnessed to yield information regarding fluid flow. There are various forms of in vivo flow or flow-like movement that that can be imaged to provide spin-movement related contrast.

Macrovascular flow (angiography) Flow within large vessels (arterial and venous) can lead to various appearances using standard imaging techniques: vessels can be void of signal (Fig. 5.17) or often they return hyperintense signal. An understanding of the principles underlying the appearance of flowing blood on MRI has led to the development of magnetic resonance angiography (MRA), where the protons in flowing blood produce a high signal against a background of little or no signal from stationary tissues. Such high contrast datasets can be manipulated and viewed, often using maximum intensity projection (MIP) algorithms, to reveal projected vascular anatomy. The common contrast mechanisms can be split into three groups: (A) TOF; (B) phase contrast; and (C) contrast enhanced. Time of flight effects utilize differences in magnetization when very short TR is used. Contrast is provided between stationary background spins in the imaging volume and inflowing, fresh

Figure 5.16 Colorectal metastasis in a fatty liver. (A) In-phase and (B) out-of-phase T1 gradient-echo axial images at same level. There is marked loss of signal from the liver parenchyma on the out-of-phase image indicating fatty infiltration. Notice the metastasis in the right lobe, and an artefact from aortic pulsation in the midline. (Courtesy of Dr A Blakeborough, Royal Hallamshire Hospital, Sheffield. UK.)

Figure 5.17 Flow effects. This patient presented with right-sided seizures. (A) Fast spin-echo T2-weighted images show a large area of signal void within the area of the left sensorimotor cortex, which was present on all other sequences. This is characteristic of a high-flow vascular malformation. (B,C) Time of flight angiograms before (B) and after Gd-DTPA (C) confirm the presence of a large pial arteriovenous malformation.

CHAPTER 5

blood.The stationary spins are repeatedly exposed to the excitation pulse and, due to the short TR, there is not enough time for their longitudinal magnetization to return to equilibrium: it becomes saturated and there is a lack of magnetization with which to form a signal. The inflowing blood, on the other hand, experiences its first excitation pulse and it returns a high signal. This phenomenon provides endogenous contrast. The imaging volume can be encoded as a set of 2D slices or a 3D dataset. High field strength systems (3T in particular) can provide excellent vascular depiction due to high signal-to-noise ratio and prolongation of both the T1 of blood and stationary background protons (Fig. 5.18). Gradient-echo sequences with high T1-weighting are often used as they can be acquired at very short TR, maximizing the TOF contrast. Phase contrast techniques utilize the basic phenomenon that phase changes are introduced to the transverse magnetization when spins are exposed to a magnetic field gradient.This is put to use in phase encoding where the phase difference in a single TR is used to encode the signal at one spatial frequency in the phase-encode direction. If, after a time delay, the phase-encode gradient were to be reversed, then the phase differences in the spins across the sample introduced by the initial gradient will be completely reversed or rewound. This is only true for stationary spins. If a spin has moved in the direction of the gradient between its initial application and its reversal, then the spin will either not have been rewound enough or it will have been rewound too much. It will have developed an overall phase change. Thus, phase-contrast encoding is performed by a pair of additional ‘bipolar gradients’, the magnitude, duration and time interval between which will determine the phase change experienced by a spin moving at a particular velocity. This information can be used to quantitate the velocity of the moving spins. In qualitative PC angiography, velocity encoding has to be applied in more than one direction. For a 3D implementation, separate flow encoding is applied in each direction. As with the TOF technique, PC is also an endogenous contrast mechanism.

Figure 5.18 3D Time of flight MR angiography projection of an intracranial, middle-cerebral artery berry aneurysm.

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

Contrast-enhanced depiction of flow uses fast T1-weighted acquisitions to capture signal from a bolus injection of exogenous Gd-based contrast media. The addition of Gd into the vasculature causes a decrease in the T1 of blood which will appear bright on T1-weighted images. In order to delineate the arterial phase, the acquisition is required to be performed within one cardiac cycle. In practice, the low spatial frequency components are collected during the arterial phase as these form the major components of the image. Care is necessary in recognizing flow effects, for it is relatively easy to diagnose a solid lesion when all that is being seen is an unusual flow effect within a fluid-filled space.

Microvascular flow (perfusion) One of the key indicators of normal tissue function is that of normal arterial flow and tissue perfusion. Perfusion-weighted MRI has also been developed and can be performed with the aid of endogenous or exogenous contrast agents.The latter are by far the most common implementation. Techniques using endogenous contrast media rely on the rf labelling of proximal arterial blood, which can then be detected as it reaches an image plane selected through the region of interest. The noninvasive nature of this measurement makes it ideal in certain situations, e.g. in the assessment of placental perfusion in utero or when several repeat studies are needed. However, the contrast-to-noise ratios are less than those that can be obtained following the administration of exogenous contrast media. Gd-chelates are suitable exogenous agents for use in MR perfusion studies (see Fig. 5.11). The presence of the Gd–ion complex shortens both T1 and T2*. If a time series of T2*weighted images is acquired with high temporal resolution (approximately 1 frame s−1), the presence of Gd can be detected as a drop in signal as blood transporting the agent arrives at the image plane. A concentration–time curve can be obtained from which parameters such as bolus arrival time (BAT), mean transit time (MTT), time-to-peak (TTP) and relative blood volume (rBV) can be derived, and the application of the central volume theorem can yield information regarding relative cerebral blood flow (rCBF). Early MR perfusion studies performed using clinical systems were limited by image acquisition times. Fast gradientecho techniques (such as FLASH) could be used but usually on a single-slice basis. The introduction of EPI into the clinical setting has revolutionized MR perfusion imaging: enabling subsecond, multislice datasets to be acquired. As with diffusion-weighted imaging (DWI), most of the experience of perfusion-weighted imaging is in the brain, and particularly in stroke and other vascular abnormalities. Ischaemic stroke is usually due to occlusion of the cervicocranial vessels. The resulting reduced blood flow will have effects on the parameters measured by perfusion-weighted MR. In the acute setting the most commonly encountered changes are: increased time-to-peak; increased mean transit time; reduced cerebral blood volume and flow. It is hoped that a combination of DWI and perfusion-weighted imaging will permit delineation of the ischaemic penumbra in stroke, defining potentially

101

102

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

salvageable tissue. Another area of interest is myocardial perfusion—used to provide information regarding the viability of the myocardium.

Very slow flow (molecular diffusion) The diffusion of water molecules is very slow random movement which can be thought of as slow directional flow when its randomness is restricted by a surrounding barrier (e.g. an axonal tract in brain parenchyma). Just as with arterial and venous intravascular blood flow and CSF flow, molecular DWI contrast can be introduced into MRI. The technique utilizes the same principles as outlined in the phase-contrast MRA section above: depending on the alteration in the phase of protons attached to water molecules as they travel along a magnetic field gradient. If a bi-lobed diffusion-encoding gradient pulse scheme is applied along one physical direction between the initial rf excitation pulse and data collection, any protons which are displaced along the direction of diffusion encoding will accumulate a relative phase change.Those that are stationary will not. The bipolar diffusion gradients are most commonly added to a single-shot EPI sequence, which enables a snap shot of the molecular diffusion to be obtained in one direction. This can be repeated along the three mutually orthogonal directions within the image plane. The degree of diffusion weighting can be changed by altering what is referred to as the b-value. The amplitude and duration of the bipolar gradients determine this b-value and images that have strong diffusion weighting require the application of very strong field gradients. Due to the inclusion of the time-consuming bipolar gradient scheme, data collection occurs at a relatively long effective TE and thus the diffusion-weighted images are also T2 weighted. This can lead to ‘T2 shinethrough’: high pixel values in areas of increased T2 as well as in areas of abnormal diffusion, which can lead to confusion! It is common practice to acquire a set of images with the bipolar gradients turned off, yielding a set of images with identical T2 but no diffusion weighting. Post-processing can then be performed to eliminate

the T2-dependent image contrast, yielding maps of apparent diffusion coefficients (ADCs). DWI is being assessed in many types of brain disease. Its major contribution appears to be in the distinction between cytotoxic oedema (which produces restricted diffusion) and vasogenic oedema (which does not restrict diffusion of water). The pathology that primarily produces cytotoxic oedema is ischaemia/infarction and DWI appears to be a sensitive and reasonably specific means of detecting ischaemic stroke. DWI is often used in conjunction with dynamic, Gd-enhanced perfusion imaging in the investigation of acute ischaemic stroke (Fig. 5.19).

PULSE SEQUENCES The most commonly used groups of pulse sequences are outlined below. Two main groupings are given: sequences where the echo results from the application of (A) an rf pulse and (B) a gradient.

Variations on a spin echo Basic two-dimensional spin-echo sequence In this pulse sequence; the digitized signal echo is produced by flipping the spins over in the x–y plane by a 180-degree rf pulse. This is applied at a time TE/2 after an initial 90-degree excitation pulse, producing a spin echo at time TE. Multiple slices are encoded by staggering each slice-selecting excitation pulse within each TR period and the Larmor frequency of each slice. Each of these slices experiences the application of one phase-encode step. Slice timing within each TR can be sequential (adjacent slices excited one after another: 1,2,3,4, 5…) or interleaved (1,3,5….2,4,6…): the latter is performed to reduce the effects of any overlap between slice edges that can occur at low TR (cross-talk). Parameters that alter image contrast are TE and TR. This is a very versatile sequence that can be PD, T2 or T1 weighted. PD-weighted images result if

Figure 5.19 Imaging of ischaemic stroke by multisequence MR. (A) An axial fast spin-echo T2-weighted image from a patient with an acute onset of left hemiparesis which shows a large area of T2 prolongation in the confines of the middle cerebral artery. The image is somewhat degraded by movement artefact as is often the case when using standard sequences in acutely unwell patients. (B) The single-shot fast spin-echo (imaging time < 1 s for one slice), which gives the same information. (C) A diffusion-weighted image acquired in 15 s shows high signal and (D) the apparent diffusion coefficient map shows a low signal, indicating an acute stroke.

CHAPTER 5

(A) TE is short (≤ 20 ms), thereby not allowing any differences in signal to evolve between tissues that may have different T2s, at the same time as (B) TR is long (≈ 3000 ms), enabling the spins to return to equilibrium between each rf excitation pulse. Differences in signal will be due to differences in the number of spins: a high PD will be represented by a bright pixel while a low PD will be represented by a dark pixel. If the TR is kept long, then as TE is increased (80–100 ms), different tissues with different T2 values will return different signals, and the T2 weighting of the resultant images will increase. In this case, tissues with long T2 values (e.g. CSF) will still have high signal at long TE and be represented by a bright pixel and vice versa. If on the other hand TE is short (to minimize T2 influences as in the PD-weighted case) but TR is decreased (≈500 ms), the T1 weighting of the sequence increases. The latter occurs due to different amounts of saturation experienced by tissues that have different T1 values. A tissue with a short T1 (e.g. fat) with respect to the specified TR, will relax back to equilibrium before the next excitation pulse: maximum magnetization will be available and so a bright pixel will ensue. A tissue with a long T1 (e.g. CSF) will not have time to relax back to its equilibrium state before the next excitation: only a small component of its longitudinal magnetization will and so the signal returned will be low and hence the corresponding pixels will be dark. Thus, pixel intensity resulting from a T1weighted spin-echo sequence (a partial saturation sequence) is inversely related to T1. A dataset can be obtained from the spin-echo technique in the order of several minutes.

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

slices can produce motion artefact-free images, but care should be exercised as unwanted movement can occur between slices, altering the relative slice positions. Blurring can occur in the phase-encode direction. Every phase-encode step is acquired at a different TE, leading to a variation of T2-weighting across spatial frequency components. All resultant images appear heavily T2 weighted. Use of the single-shot FSE (SS-FSE) technique includes MR cholangiopancreatography, imaging of the diaphragm, bowel imaging and imaging of the fetus in utero (Fig. 5.20). It is also useful as a fast T2-weighted screening sequence in a moving patient!

Inversion recovery In this sequence type, the magnetization is ‘prepared’ before the initial 90-degree excitation pulse by the addition of an 180-degree ‘inversion’ pulse.The signal obtained will be influenced by the relative degree of recovery experienced by the spins along the z-axis.The inversion time (TI) is the time allotted for this recovery process to evolve between the 180-degree inversion and 90-degree excitation pulses. The inversion pulse introduces heavy T1 weighting into this sequence. Specific forms of the technique are fluid attenuated inversion recovery (FLAIR) where the TI is chosen so that the magnetization from fluids is nulled (≈ 2200 ms) in the detection (x–y) plane and TE is long (80–100 ms), introducing a lot of T2 as well as T1 weighting. The combination of fluid nulling with T2 weighting is particularly useful in assessing lesions with prolonged T2 adjacent to fluid structures (e.g. periventricular demyelination; see Fig. 5.7). In this same acquisition, T1 shortening brought

Multi-spin echo More than one echo can be recalled per 90-degree excitation using multiple 180-degree rf refocusing pulses which yield image datasets with different contrasts (as they are obtained at different TEs).The maximum number of slices per TR will reduce as the number of recalled echoes increases. The most common type used in clinical practice has been the dual spin-echo with short and long TE datasets (≈20 and 80 ms, respectively) at long TR, yielding PD- and T2-weighted datasets, respectively.

Fast-spin echo or turbo-spin echo or rapid acquisition with relaxation enhancement (RARE) As mentioned previously, the introduction of more than one phase-encode, refocusing pulse and frequency-encoded echo per excitation pulse (increasing the echo train length [ETL]), can speed imaging up by factors of between 2 and 32. During any one TR period, different spatial frequency components in the phase-encode direction will be acquired at different TEs, leading to a complex mixture of image contrasts. It is impossible to produce images that are as heavily PD weighted as can be achieved with the standard spin-echo technique above. Both T2- and T1-weighted images are commonly produced.

Single-shot fast-spin echo or half-Fourier acquired single-shot turbo-spin echo (HASTE) This is an extension of the FSE technique that enables all phase-encoding steps to be obtained following one initial excitation. One slice is acquired at a time, in approximately 1 s. Covering a volume prone to movement with multiple single

Figure 5.20 Imaging of the fetus in utero using single-shot fast spin-echo. Each set of 20 slices takes 20 s to acquire. This fetus has agenesis of the corpus callosum.

103

104

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

about by the presence of exogenous Gd-based contrast agents will return high signal due to the sequence’s T1 weighting. In short-tau inversion recovery (STIR), T1 is kept short, and is typically used to null signal from fat. An example of its use in this context is in the delineation of breast lesions. There are several other methods of fat suppression; such sequences often demonstrate subtle lesions to best effect.

Spin-echo echo planar imaging In this variant of EPI, several or all lines of k-space are swept through at once during the evolution and decay of the main echo. The main echo is produced by an 180-degree rf pulse, i.e. it is a spin echo which is less influenced by field heterogeneities than the gradient-echo variety (below). This spin-echo EPI variant tends to be used for DWI as it is less prone to susceptibility-related artefacts than the gradient-echo EPI variant (see below).The sequence is actually a mix of gradient echoes, refocused by the EPI technique, that occur during the evolution of the spin echo that results from the refocusing rf pulse. Readers are referred elsewhere for further explanation and discussion of mixed gradient and spin echo (GRASE) sequences which have mixed spin-echo/gradient-echo contrast features.

Variations on a gradient echo/field echo Basic two-dimensional gradient-echo sequence This is as above for the 2D spin-echo sequences, except that the echo is produced by a gradient (rather than an rf pulse), causing the components of the spin magnetization to refocus in the x–y plane, producing an echo, without ‘flipping’ them over (i.e. without using an 180-degree pulse) (see Fig. 5.3). Unlike 2D spin-echo sequences, reduction in signal amplitude resulting from faster dephasing due to the presence of magnetic susceptibility differences/localized heterogeneity is not cancelled in this technique. Images can have mixed contrast consisting of PD, T1 and T2* weighting. When a different type of echo formation is present (the Hahn echo), gradient-echo sequences can also be T2 weighted. An understanding of the Hahn echo is needed to understand how different combinations of these four parameters can be made to dominate the image contrast. The TR used in a gradient-echo sequence is often shorter than the component tissue T2 values. In other words, the magnetization in the x–y plane will not have dephased completely before the next excitation pulse is applied (unlike the transverse magnetization in the spin-echo sequence which completely dephases, even when short TR is used to introduce saturation of the longitudinal component, i.e. to introduce T1 weighting). This net transverse coherence plus the addition of a further rf pulse produces the Hahn echo. So, two echoes can be present: the gradient echo and the Hahn echo. Each or both of these echoes can be used as follows. 1 Spoiled gradient-echo or spoiled gradient-recalled echo (SPGR) or RF spoiled Fourier acquired in the steady state (RF-FAST) or T1 fast field echo (T1 FFE). Any magnetization coherence in the x–y plane can be destroyed by the addition of a gradient pulse or rf ‘spoiler’, applied towards the end of the TR period. This sequence is

often termed a ‘spoiled gradient echo’. There is no Hahn echo, just the gradient echo. At short TE (< 8 ms) and short TR (≈ 50 ms), T1-weighting predominates (as with spin echo at short TR, due to differential saturation effects), but the weighting is not as heavy as the T1-weighted spin-echo variants. FLASH is a variant of the spoiled gradient echo. Instead of an initial 90-degree rf pulse, a low flip-angled pulse (α) is used. Very short repetition times can be used as saturation along the z-axis is minimized; enabling fast acquisition times and refresh rates for dynamic and angiographic studies. A relatively long TE (≈ 18 ms) increases the T2* weighting, with low α (≈ 25 degrees), decreasing the T1 weighting, allowing the T2* weighting to dominate. 2 Contrast-enhanced Fourier acquired in the steady state (CE-FAST) or PSIF (mirrored FISP) or T2 fast field echo (T2 FFE). T2-weighted images can be produced by imaging the steady-state coherence or Hahn echoes (that build up over a few repetitions) and not the gradient echo. The Hahn echoes are sampled towards the end of the TR period and the effective TE is approximately twice the TR. 3 Gradient recalled acquisition in the steady state (GRASS) or fast imaging with steady-state precession (FISP) or Fourier acquired in the steady state (FAST) or fast field echo (FFE). Residual phase coherence at the end of each TR is not destroyed and both the gradient and Hahn echoes contribute to the image. Complex, combined T1/T2 weighting results.

Three-dimensional gradient echo Secondary phase encoding is performed instead of slice selection. The penalty associated with this is an increase in acquisition time. However, as gradient-echo sequences can run with very short TR, this is manageable with typical standard 3D gradient-echo acquisition times in the order of 5–15 min. Acquisition times can be reduced substantially (with lower coverage and resolution), enabling breath-hold techniques (10–20 s). The spatial resolution in the second phase-encode direction is generally less than 2 mm (above this, 2D techniques tend to be used) and the signal-to-noise ratio at high resolution is superior to 2D techniques. Isotropic, submillimeter voxels facilitate true 3D visualization strategies.

Magnetization prepared rapid acquisition gradient echo (MP-RAGE) or turbo-FLASH or inversion recoveryprepped fast spoiled gradient-recalled echo As with the inversion recovery spin-echo sequence, an inversion pulse is applied before the excitation pulse, producing a T1-weighted sequence that is often used with rapid 3D encoding for high-resolution imaging of, for example, brain malformations. Coupled with its ability to depict exogenous contrast medium which has traversed the damaged blood–brain barrier, it is also useful for intra-operative neurosurgical guidance.

Gradient-echo echo planar imaging This is similar to spin-echo EPI but the main echo is produced by a gradient lobe. This sequence is heavily weighted to reflect differences in magnetic susceptibility. It is often used

CHAPTER 5

for BOLD fMRI studies and EPI-based exogenous perfusion assessment.

SAFETY CONSIDERATIONS Undergoing an MR investigation or working in a MR unit does not involve exposure to ionizing radiation and no clear long-term biological effects have been reported. However, the environment is quite unusual, presenting several potential risks to human safety that need to be fully addressed by those who work in it and for those who have occasion to visit it. Each unit should have a set of local rules based on advice outlined in documents from the relevant regulatory bodies (UK’s Medicines and Healthcare products Regulatory Agency (MHRA), European Union, etc.) and all MR unit staff should be familiar with these. Inner and outer ‘controlled areas’ that have restricted access should be detailed. Accountability to the hospital’s radiation safety committee is to be commended, ensuring that incidents are reported appropriately and that necessary safeguards/recommendations are acted upon. The invisible MR environment includes potential effects resulting from the following.

Magnet Ferromagnetic objects (mainly containing iron, nickel or cobalt), when near to a magnet, experience a force of attraction towards the magnet bore (magnetic isocentre). The force depends upon factors including magnetic field magnitude, proximity, object mass and composition.The danger associated with this hazard is real: fatalities have resulted. All patients, visitors, non-MR personnel and pieces of equipment should be screened appropriately before entry into the controlled area. Cardiac pacemakers contraindicate entry as do intracranial aneurysm clips of ferromagnetic/unknown composition. In very high fields (>> 1.5T) mild sensory activations (such as visual magnetophosphenes and balance alteration) have been reported. There is no evidence that these are harmful. If the magnet’s liquid helium/nitrogen Dewar is being replenished, the magnet room should be kept clear due to risk of cryogenic burns and suffocation. There may be occasion when the magnet needs to be shut down in an emergency, e.g. if a person becomes trapped between the magnet and a ferromagnetic object. Once the emergency run down unit (ERDU) button is depressed, liquid helium will vaporize and there is a valve/ pipe system to vent this gas to the outside. However, as a matter of caution, all personnel should exit the magnet room and the door should be kept closed.

Time-varying magnetic field gradients At high amplitudes and slew rates, these can cause unpleasant peripheral nerve stimulation. This is an acute effect and not harmful. Modern system software will alert the operator when there is risk of attaining such levels. The acoustic noise generated due to the fast switching of currents within the gradient coils while restrained in a high static magnetic field (an expensive loudspeaker!) may be considerable20, levels being up to

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

108 dBA. Protective ear plugs/headphones should be worn by patients (and accompanying personnel), particularly in systems operating at 1.5T and above. Recent advances in gradient coil design and construction have led to a decrease in gradientassociated noise on some systems for given performance levels.

Radiofrequency field The transmitted rf field deposits energy within the body. Energy deposition is expressed as the specific absorption rate (SAR) in watts per kilogram. The levels given in MHRA guidelines (for SARs applied over a 15-min period) have been set to limit tissue heating to less than 1°C. Modern system software indicates the SAR to the operator for each acquisition. Care should be exercised when examining anaesthetized patients who require ECG monitoring. Burns can result from rf Eddy currents being induced under the skin electrodes (e.g. where clinical indications suggest SAR-intensive sequences such as FSE imaging of the lumbar spine). In these circumstances, pain sensation cannot be communicated while a burn is developing. Caution should also be made when the patient’s circulatory system is compromised. Clinical need should be noted in cases such as these. MR is not advised during the first trimester of pregnancy. This is precautionary as no deleterious effects to mother or child are known. Patients with various implants (from deep brain stimulators to stents) should be assessed for possible interactions with static field and rf heating: evidence from an up-to-date reference should be sought. At the time of writing, an EU Directive covering work exposure to electromagnetic radiation is due to become part of UK law in 2008. It will affect working practices in clinical MR units. The relevant exposure is likely to be related to the time-varying fields produced by the gradient subsystem. Staff will not be able to stand at the entrance to the magnet bore to comfort the patient, initiate examinations from the magnet fascia, perform close monitoring of anaesthetized patients or take part in MR-guided interventional procedures. Various professional bodies are presently attempting to address these issues and it is hoped that these restrictions will not be necessary.

CONCLUSION In the first 25 years of clinical practice, MRI has moved from being a curiosity to being the technique of first choice in a variety of diseases. The physics necessary for image interpretation is more complex than that for any other technique in radiology, yet the great diversity of the technique provides a wide range of possibilities, many of which are being developed or are yet to be explored. Clinical MR is truly 3D, and with the advent of modern workstations there is an opportunity to obtain radiographic views that have not been available previously.The recent development of fast imaging ciné studies, perfusion and diffusion techniques, hyperpolarized gas imaging (Fig. 5.21) and solid

105

106

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Figure 5.21 Image of a healthy lung obtained with a two-dimensional, rapid, steady-state free precession sequence with 300 ml. Hyperpolarized 3 He at 30% polarization. (Courtesy of Dr J.M. Wild, University of Sheffield, UK.)

state methods provide new possibilities. The development of new contrast agents, targeted at specific organs, diseases, cell or gene types is likely to result in a considerable expansion of the options now available. Parallel acquisition methods promise radically to increase the throughput of MR techniques and allow contrast mechanisms, which previously could only be acquired statically, to be acquired in dynamic mode. Specialized MR systems targeted at specific body regions or diseases may increase access to MR and provide options for monitoring and even performing minimally invasive therapy. The full potential of MR is yet to be realized.

REFERENCES 1. Bloch F, Hansen W W, Packard M E 1946 Nuclear induction. Phys Rev 69: 127. 2. Purcell E M, Torrey H C, Pound C V 1946 Resonance absorption by nuclear magnetic movements in a solid. Phys Rev 64: 37–38. 3. Lauterbur P C 1973 Image formation by induced local interactions: examples employing NMR. Nature 242: 190–191. 4. Mansfield P, Grannell P K 1973 NMR ‘diffraction’ in solids? J Phys C 6: 422–426. 5. Mansfield P, Maudsley A A 1977 Medical imaging by NMR. Br J Radiol 50: 188–194. 6. Damadian R, Goldsmith M, Minkoff L 1977 NMR in cancer: Fonar image of the live human body. Physiol Chem Phys 8: 97–108. 7. Hinshaw W S, Bottomley P A, Holland G N 1977 Radiographic thin section of the human wrist by nuclear magnetic resonance. Nature 270: 722–723. 8. Hawkes R C, Holland G N, Moore W S, Worthington B S 1980 NMR tomography of the brain: a preliminary clinical assessment with demonstration of pathology. J Comput Assist Tomogr 4: 577–586. 9. Edelstein W A, Hutchinson J M S, Johnson G, Redpath T 1980 Spin warp NMR imaging and application to human whole body imaging. Phys Med Biol 25: 751–756. 10. Thomas D L, Lythgoe M F, Pell G S, Calamante F, Ordidge R J 2000 The measurement of diffusion and perfusion in biological systems using magnetic resonance imaging. Phys Med Biol 45: R97–138.

11. Wilkinson I D, Griffiths P D, Hoggard N, et al 2003 Short-term changes in cerebral micro-hemodynamics following carotid stenting assessed by MR perfusion imaging. Am J Neuroradiol 24:1501–1507. 12. Neumann-Haefelin T, Moseley M E, Albers G W 2000 New magnetic resonance imaging methods for cerebrovascular disease: emerging clinical applications. Ann Neurol 47: 559–570. 13. Taylor A M, Keegan J, Jhooti P, Gatehouse P D, Firmin D N, Pennell D J 2000 A comparison between segmented k-space FLASH and interleaved spiral MR coronary angiography sequences. J Magn Reson Imaging 11: 394–400. 14. Wilkinson I D 2005 Perfusion and diffusion imaging in chronic carotid disease. In: Gillard J, Waldman A, Barker P (eds) Clinical MR neuroimaging: diffusion, perfusion and spectroscopy. Cambridge: Cambridge University Press. 15. Paley M 1996 Human brain proton spectroscopy. In: Bydder G M, Bradley W (eds) Advanced MR imaging techniques. London: Martin Dunitz. 16. Schaefer S, Balaban R S (eds). 1992. Cardiovascular magnetic resonance spectroscopy. New York: Springer. 17. Ogawa S, Tank D W, Menon R, et al 1992 Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 89: 5951–5955. 18. Rosen B R, Aronen H J, Kwong K K, Belliveau J W, Hamberg L M, Fordham J A 1993 Related articles advances in clinical neuroimaging: functional MR imaging techniques. RadioGraphics 13: 889–896. 19. Middleton H, Black R D, Saam B, et al 1995 MR imaging with hyperpolarized 3He gas. Magn Reson Med 33: 271–275. 20. Sodickson D K, Manning W J 1997 Simultaneous acquisition of spatial harmonics (SMASH): fast imaging with radiofrequency coil arrays. Magn Reson Med 38: 591–603. 21. Pruessmann K P, Weiger M, Scheidegger M B, Boesiger P 1999 SENSE: sensitivity encoding for fast MRI. Magn Reson Med 42: 952–962.

FURTHER READING Overviews Edelman R R, Hesselink R R 2005 Clinical magnetic resonance imaging. Philadelphia: WB Saunders. McRobbie D W, Moore E A, Graves M J, Prince M R 2003 MRI from picture to proton. Cambridge: Cambridge University Press. Reimer P, Parizel PM (rds) 2006 Clinical MR imaging: A practical approach. Berlin: Springer-Verlag.

Sequences design/spatial encoding/hardware Haacke E M, Brown R W, Thompson M R, Venkatesan R 1999 Magnetic resonance imaging: Physical principles and sequence design. Chichester: J Wiley & Sons. Holland G N, MacFall J R 1992 An overview of digital spectrometers for MR imaging. J Magn Reson Imaging 2: 241–246. Landini L, Positano V 2005 Advanced image processing in magnetic resonance imaging. CRC Press. Nassaivier M 1996 All you really need to know about MRI physics. Simply physics. www.simplyphysics.com. Robitaille P-M, Berliner L (eds) 2006 Ultra high field magnetic resonance imaging. Berlin: Springer-Verlag. Schmitt F, Stehling M K, Turner R 1998 Echo-planar imaging: theory, techniques and application. Berlin: Springer-Verlag. Schoenberg S O 2007 Parallel imaging in clinical MR applications. Berlin: Springer-Verlag. Tweig D B 1983 The k-trajectory formulation of the NMR imaging process with applications in analysis and synthesis of imaging methods. Med Phys 10: 610–623.

Diffusion, perfusion and spectroscopy Danielson E B, Ross B 1998 Magnetic resonance spectroscopy diagnosis of neurological diseases. New York: Marcel Dekker.

CHAPTER 5

Gillard J, Waldman A, Barker P (eds) 2005 Clinical MR neuroimaging: diffusion, perfusion and spectroscopy. Cambridge: Cambridge University Press. Warach S, Davis S 2003 Magnetic resonance imaging in stroke. Cambridge: Cambridge University Press. Young I R (ed) 2000 Methods in biomedical magnetic resonance imaging and spectroscopy. Chichester: J Wiley & Sons.

BOLD fMRI Buxton R B 2001 Introduction to functional magnetic resonance imaging: Principles and techniques. Cambridge: Cambridge University Press. Frackowiak R S J, Friston K J, Frith C D, Dolan R J, Mazziotta J C 1997 Human brain function. New York: Academic Press.

Exogenous contrast Jackson A 2005 Dynamic contrast-enhanced magnetic resonance imaging in oncology. Berlin: Springer-Verlag. Kauczor H-U (ed) 2000 Special issue: hyperpolarized gases in MRI. NMR Biomed 13: 173–264. Stark D D, Weissleder R, Elizando G, et al 1989 Superparamagnetic iron oxide: clinical application as a contrast agent for magnetic resonance imaging. Radiology 168: 297–301. Weinmann H-J, Brasch R C, Press W R, Wesbey G E 1984 Characteristics

•

MAGNETIC RESONANCE IMAGING: BASIC PRINCIPLES

of gadolinium-DTPA complex: a potential NMR contrast agent. Am J Roentgenol 142: 619–624.

Angiography/cardiology/interventional Bogaert J, Dymarkowski S, Taylor A M (eds) 2005 Clinical cardiac MRI. Berlin: Springer-Verlag. Graves M J 1997 Magnetic resonance angiography. BJR 70: 6–28. Lee V 2006 Cardiovascular MRI: Physical principles to practical protocols. Philadelphia: Lippincott, Williams & Wilkins. Lufkin R B 1999 Interventional MRI. St Louis: Mosby. Morris E, Liberman L 2005 Breast MRI: Diagnosis and intervention. Berlin: Springer-Verlag.

Safety Medical Devices Agency 2002 Guidelines for magnetic equipment in clinical use with particular reference to safety. London: Medical Devices Agency. Moseley I F 1994 Safety of magnetic resonance imaging. Br Med J 308: 1181–1182. National Radiation Protection Board 1991 Principles for the protection of patients and volunteers during clinical MRI. NRPB 2 (no 1). Shellock S G 2006 Reference manual for magnetic resonance safety, implants, and devices: 2007 edition. BRPG. Shellock R, D Services Inc. www.mrisafety.com.

107

CHAPTER

Angiography: Principles, Techniques (Including CTA and MRA) and Complications

6

James E. Jackson, David J. Allison and James Meaney

Multidetector CT angiography • Clinical applications Magnetic resonance angiography • Background • Contrast mechanisms • Post-processing • MRA in clinical practice • ‘Supplemental’ imaging: when is imaging of the lumen not enough? • Future directions Catheter arteriography • Technique • Preparation of the patient

• Contraindications • Anaesthesia • Arterial puncture • Digital subtraction angiography • Intravenous digital subtraction angiography (IVDSA) • Aftercare • Complications Catheter venography • Techniques • Complications Embolization techniques • Embolic materials and techniques • Indications for therapeutic arterial embolization

The imaging of blood vessels has changed considerably since the first edition of this textbook and, indeed, there have been significant new developments in cross-sectional imaging techniques even since the 4th edition. These have made many of the diagnostic catheter angiographic techniques described in previous editions almost obsolete. On the whole this is clearly a welcome advance; the newer multidetector CT angiographic techniques are obviously less invasive and, therefore, safer. Furthermore, in many instances these techniques will give more diagnostic information than could be obtained by conventional catheter arteriography because of the concurrent visualization of surrounding tissues and the ability to reconstruct the data in any plane. One of the disadvantages of the decline in the number of diagnostic catheter

angiograms performed, however, is that it has become more difficult for radiologists to acquire suitable expertise in the catheter techniques that are still required for more complex therapeutic interventional procedures. As a good understanding of the basic principles and techniques of catheter angiography remains essential for those intending to become interventional radiologists (and it becomes less likely that they will be able to obtain sufficient practical experience during their training for the reasons given earlier) it perhaps becomes more important that this information is available in this textbook. The newer cross-sectional techniques for imaging blood vessels will, however, be discussed first as these will, quite rightly, be requested before (and often instead of ) conventional catheter angiography.

110

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

MULTIDETECTOR CT ANGIOGRAPHY (MDCTA) The development of CT machines combining a fan-shaped x-ray source and multiple detector rows has led to the ability to acquire image data from a large tissue volume in a single breath hold. With IV contrast medium and appropriate timing, exquisite images can be obtained of blood vessels during any particular vascular phase. Optimal imaging of the vessels requires the relatively rapid IV injection of iodinated contrast medium (usually 3–5 mls-1) and the acquisition of data at the appropriate time of vascular enhancement. The latter can be estimated based upon the ‘normal’ time of arrival of the contrast medium within the organ being imaged or, more commonly nowadays, by the more accurate use of contrast bolus detection technology. ‘Tight’ boluses of contrast medium using a chaser of normal saline may be useful not only to improve vascular opacification but also to reduce the total volume of contrast medium required. Depending upon the region and volume of the body being imaged hundreds, if not thousands, of axial images will be acquired; whilst all the diagnostic information is available in this data-set, evaluation of the axial images alone can be extremely time-consuming and is helped considerably by reconstruction of the data in axial, coronal, and oblique planes without loss of resolution, so-called multiplanar reconstruction (MPR). Tortuous vessels can be ‘straightened’ by curved MPR to aid in the assessment of luminal narrowing due to, for example, atheromatous disease or encasement by tumour. Maximum intensity projection (MIP) and volume rendering (VR) techniques are additional tools that help greatly in the assessment of blood vessels. Each of these reconstruction techniques has its advantages and disadvantages: 1 MPR is very useful for the rapid review of blood vessels in any plane including the surrounding bone and soft tissues, and will allow the assessment of vessel walls that might be obscured in MIP and VR techniques by the presence of, for example, calcification or an endoluminal stent. Each image, however, gives

A

B

only one ‘slice’ of information and multiple separate images are required to see the vessel in its entirety (Fig. 6.1). 2 MIP techniques produce a planar image from a volume of data within which the pixel values are determined by the highest voxel value in a ray projected along the data set in a specified direction.The images obtained are those most similar to a conventional arteriogram but one of the clear disadvantages of this technique is that any tissue of high density (such as bone or vascular calcification) lying within the ray through a vessel will determine the pixel intensity instead of the contrast medium itself within the blood vessel (Fig. 6.2).This is a common cause of overestimation of vascular stenoses. 3 Volume-rendering techniques assess the entire volume of data with an attenuation threshold for display and produce a three-dimensional image. Typically, tissues are assigned a

Figure 6.2 An MIP image from an MDCT of a renal transplant artery clearly demonstrating the vascular anatomy.

C

Figure 6.1 MDCT images of a gastroduodenal artery pseudoaneurysm. (A) An axial image demonstrates an enhancing pseudoaneurysm cavity surrounded by a large haematoma in the region of the pancreatic head. (B) A sagittal MPR image demonstrates a pseudoaneurysm arising from the gastroduodenal artery. (C) A further sagittal MPR image demonstrates that the right hepatic artery, from which the gastroduodenal artery arises, originates from the superior mesenteric artery.

CHAPTER 6

• ANGIOGRAPHY: PRINCIPLES, TECHNIQUES (INCLUDING CTA AND MRA) AND COMPLICATIONS

colour that is dependent upon their attenuation values, facilitating the differentiation of structures of differing density (Figs 6.3, 6.5). The final images can be rotated in real time to find the best projection to display anatomy and pathology and this is the most important feature of this technique. Vascular stenoses can be overestimated, however, and small vessels may not be clearly visualized. It should be remembered that, whilst these post-processing techniques are very helpful for diagnostic assessment and for display in multidisciplinary team meetings, the axial source images are essential and often allow the operator to distinguish between artefact and disease when an abnormality is suggested on reformatted views.

CLINICAL APPLICATIONS MDCT angiography (MDCTA) is replacing conventional angiography in many, if not all, body areas and is indicated, therefore, in any disease process that requires the visualization

A

B

of blood vessels to improve diagnosis and outcome. Within the thorax, for example, this would include the assessment of: pulmonary embolic disease1,2 (Fig. 6.4); thoracic aortic disease3–5; coronary artery graft patency6–10; and bronchial artery anatomy and pathology in massive haemoptysis11. In the abdomen common indications include the pre-operative planning and posttreatment assessment of abdominal aortic aneurysmal disease12,13 (Fig. 6.5); the assessment of native and transplant renal arteries14–16; the staging of hepato-pancreaticobiliary neoplasms17–20; and the assessment of vascular complications in patients suffering severe trauma21. It is also used increasingly in the assessment of peripheral arterial disease in the lower (and upper) limbs where it is less invasive, less expensive and exposes the individual to less radiation than conventional catheter angiography22–26. The scanning technique (positioning of the patient, rate of contrast medium administration, time of image acquisition), and the post-processing techniques most suited to the different indications listed earlier will clearly vary and lie outside the scope of this chapter; interested readers are referred to other chapters within this book and to other texts cited in the reference list.

C

D

Figure 6.3 The value of MDCT MPR and VR images in the assessment of pulmonary sequestration. (A) The CXR demonstrates a mass projected through the left side of the cardiac silhouette. (B) An axial image from an MDCT examination demonstrates a left paraspinal soft-tissue mass. A feeding vessel arising from the thoracic aorta is visible. (C) MPR and (D) VR images show the full length of the oblique course of feeding artery.

Figure 6.4 MDCT demonstration of bilateral pulmonary emboli. (A) Axial and (B) coronal MPR images demonstrating extensive bilateral pulmonary emboli in central pulmonary arteries.

111

112

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Figure 6.5 MDCT images of abdominal aortic aneurysmal disease before and after stent insertion. A large infrarenal abdominal aortic aneurysm is seen on axial (A and D), coronal MPR (B and E) and VR (C and F) images before and after the insertion of an endoluminal bifurcated stent

MAGNETIC RESONANCE ANGIOGRAPHY BACKGROUND Magnetic resonance angiography (MRA) is a method for generating images of blood vessels with magnetic resonance imaging (MRI)27,28. With improved understanding of the nature of the signals emanating from blood vessels on MR images, it became evident that rather than simply contribut-

ing a source of artefacts on MR images, flow phenomena could be harnessed to generate diagnostic ‘angiograms’27–29. MRA has undergone a revolution over the last decade, replacing catheter angiography as the primary diagnostic tool for the evaluation of most vascular territories (apart from the coronary arteries); this change is due mainly to the success of contrast-enhanced techniques30.

CHAPTER 6

• ANGIOGRAPHY: PRINCIPLES, TECHNIQUES (INCLUDING CTA AND MRA) AND COMPLICATIONS

CONTRAST MECHANISMS Unenhanced (time-of-flight [TOF] and phase contrast [PC]) MRA With these techniques, the intravascular signal depends on inherent properties of flow, and the MR parameters must be carefully tailored to ensure a high intravascular signal27,28. In the case of TOF MRA, for example, data must be acquired perpendicular (and ideally orthogonal) to the direction of flow, and the time between successive radio frequency (RF) pulses (the repetition time [TR]), must be sufficiently long to allow an adequate ‘inflow’ of fully relaxed protons into the imaging slice27.The TR, therefore, is dictated by expected flow rates within the regionof-interest and, typically, should be 35 ms or greater. As the data acquisition time is directly proportional to the TR (acquisition time = TR × number of phase-encoding steps × number of slices × number of excitations), a TR substantially greater than the shortest possible for gradient-echo imaging (3–5 ms) must be used, with resultant prolongation of scan time. Additionally, owing the predominant head–foot orientation of the major arteries (e.g. the aorta, ilio-femoral and infra-popliteal vessels), the axial plane must be used, which also prolongs the data acquisition owing to the large number of slices required. Despite ‘faster’ techniques, therefore, acquisition times for TOF MRA are artificially prolonged, firstly as a result of physiological bloodflow rates that mandate the use of a relatively long TR and secondly by the requirement for axial imaging, which affords poor spatial coverage in comparison with sagittal or coronal imaging30. Selective arteriograms or venograms can be acquired by employing a (travelling) saturation pulse placed downstream of the imaging slice for MRA (to eliminate venous return from the opposite direction) or upstream of the imaging slice to generate MRV’s (venograms). If no saturation pulses are employed, both arteries and veins are identified on the same image.

Time-of-flight MRA methodology and limitations TOF angiography relies on the fact that the blood enters the volume under consideration with relatively high velocity and traverses it quickly, so that it receives very few RF pulses27,28. In order to maintain a highest possible inflow effect, all protons within the imaging volume must be replenished between successive TRs, though maximal inflow may not be necessary in clinical practice and some trade-offs can be accepted.An oblique course of the blood vessel being imaged in relation to the slice orientation and short TRs both adversely affect signal-to-noise ratios (SNR) as a result of protons under these circumstances experiencing more RF pulses whilst in the imaging slice. The severity and length of stenoses also tend to be overestimated on TOF MRA images because of intra-voxel dephasing secondary to turbulent, slow or pulsatile flow. As a result of these limitations,TOF MRA has failed to offer a viable non-invasive screening alternative to conventional arteriography, and has not had a major impact on clinical practice.

Phase-contrast MRA Phase-contrast angiography (PCA) is now seldom used in clinical MRA. The methodology underpinning the technique

is somewhat complex and, like TOF MRA, images are prone to artefacts and data acquisition is lengthy29. The term ‘phase’ refers to the angle that the transverse magnetization makes relative to a reference axis. When protons are moving, for example, in flowing blood, a change in phase proportional to the distance the proton moves (and, therefore to blood velocity) is induced by the imaging gradients, in particular the slice-selection gradient and the frequencyencoding gradient. As these gradients actually consist of a pair of gradient pulses applied in opposite directions (so-called bipolar gradients), the effect of each gradient on the phase of protons for stationary (i.e. background, motionless) tissues is equal and opposite, and the effect cancels out. For moving (flowing blood) protons, the position of each proton will change between consecutive pulses resulting in a phase change relative to stationary protons that is proportional to velocity. This observation of gradient-induced phase change provides the basis for phase-contrast angiography in which the gradient pulses are designed to produce phase changes for a given velocity range. In this way, the signals do not cancel, phase information is preserved during the image reconstruction. Pixel brightness is directly proportional to the phase-shift acquired by a moving proton in the magnetic field and, therefore, to velocity. In practice, as the method is only sensitive for the velocity component applied along the flow-encoding gradient, the acquisition must be repeated a total of four times in order to generate an ‘angiogram’: an initial flow-compensated sequence is followed by three flow-encoded acquisitions one for each direction of flow (head–foot, left–right, anteroposterior), followed by a complex subtraction to generate the final angiographic image. As the phase is unchanged for static protons, the subtraction completely suppresses the signal from the background tissue thus facilitating the generation of high-quality images29. One of the major challenges of PCA relates to the requirement of the operator to appropriately select the velocity-encoding gradient.As the signal intensity is proportional to velocity, the range of velocities present within the vessels of interest must be inferred or measured to allow the operator to set the velocity-encoding gradient (Venc) correctly.This assumes an a priori knowledge of the blood flow velocities within the relevant artery and, though this can be rapidly measured directly by acquiring a series of 2D PC images with different phase-encoding values, it is time-consuming and the presence of different flow velocities within the arteries enclosed within a single field of view may introduce artefacts29.

Limitations of unenhanced MRA and requirement for contrast agents At each stage of the evolution of MRA, high accuracy was reported for almost all techniques when compared with catheter angiography27–29 but TOF and PC MRA did not gain widespread acceptance in clinical practice because of long examination times, suboptimal resolution and frequent artefacts. TOF MRA was used to evaluate disease involving the carotid bulb and the femoro-popliteal and pedal arteries as these vascular territories were ideally suited to this technique due to the relatively straight course of their vessels, which

113

114

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

meant that adequate ‘inflow’ could be ensured. MRA was also established as an accurate modality for portrayal of the proximal intra-cranial arteries (despite the fact that the arteries demonstrate marked tortuosity) because of the relatively small volume of tissue that needed to be covered, coupled with the fact that constant blood flow within the cerebral arteries over the cardiac cycle optimizes intravascular signal29,30.

Contrast-enhanced MRA (CEMRA) Because of their unmatched high contrast-to-noise ratios, high spatial resolution, rapid speed of acquisition and relative freedom from artefacts, contrast-enhanced techniques have almost universally replaced non-contrast techniques in clinical practice30. Unlike TOF and PCA techniques, where the intravascular signal is dependent on inherent properties of flow and is, therefore, at the mercy of alterations in flow rate secondary to vascular disease, intravascular signal for contrast-enhanced MRA (CEMRA) depends on a T1 shortening effect induced by the injection of a paramagnetic contrast agent (usually gadolinium based). Images can, therefore, be acquired in any plane including coronal, which affords the best anatomical coverage for virtually all vascular territories outside the brain (Fig. 6.6). In addition, the ability to exploit ultrafast 3D acquisitions (by using the shortest TRs possible), allows rapid image acquisition that can easily be accommodated within a single breath-hold, an important factor when imaging in the chest and abdomen. In order to generate ‘selective’ arteriograms, images are acquired during the first arterial passage of the contrast agent before its arrival within the veins.The synchronization of data acquisition with the peak arterial bolus is one of the major challenges of CEMRA as the rate of transit of contrast medium from the peripheral vein injection site to the vessel of interest is affected by a number of factors including heart rate, stroke volume and the presence or

Figure 6.6 A surgically created dialysis (arteriovenous) fistula in the left arm of patient with chronic renal failure.

absence of proximal steno-occlusive lesions. Although the circulation time can be measured using a test bolus, or can be inferred by making some assumptions about the patient’s cardiovascular status, the process is now automated by employing an MR fluoroscopic approach – a technique that demonstrates contrast medium arrival in real time on the display monitor, thus signalling the appropriate time for data acquisition31. The unique nature of k-space (the array of data from which the final image is generated) whereby the central lines determine image contrast and the peripheral lines determine image resolution, can be uniquely exploited to generate CEMRA images with unrivalled signal-to-noise ratios32. In situations where breath-holding is not required (e.g. peripheral MRA and carotid artery imaging) as long as collection of the contrast-defining central lines of k-space is completed during the arterial peak before contrast medium reaches the veins, the continued collection of resolution-defining peripheral lines of k-space during venous enhancement does not result in venous contamination of the images32. CEMRA is now the standard of reference for MRA against which all new techniques must be measured.

POST-PROCESSING Regardless of which method is employed to generate MR angiograms, the aim of all techniques is to make the arteries the brightest structures on the images, and to extract the vascular data by means of a maximum intensity projection (MIP) computer algorithm28 (Figs 6.7, 6.8). Other methods of post-processing include multi-planar reformatting, volume rendering and surface-shaded displays. For phase-contrast MRA, there is inherent complete background suppression because of the absence of bulk motion in background tissues.

Figure 6.7 Normal MRA. Note clear depiction of the abdominal aorta, iliac arteries and renal arteries on the frontal MIP.

CHAPTER 6

• ANGIOGRAPHY: PRINCIPLES, TECHNIQUES (INCLUDING CTA AND MRA) AND COMPLICATIONS

monary angiography; and the fact that catheter pulmonary angiography, the reference standard against which new and improved MRA techniques should be compared, has largely disappeared from clinical practice due to the success of CTA, thus depriving MRA of a valid arbiter for comparative studies. Improvements in spatial resolution bring the subsegmental arteries within the realm of MRA (Fig. 6.9) and further refinements including MR perfusion and ventilation (mirroring the ventilation and perfusion components of nuclear medicine studies albeit at much higher resolution) offer additional functionality to determine the location and distribution of small emboli36.

Abdominal aorta, renal and mesenteric arteries

Figure 6.8 Severe left renal artery stenosis and right common iliac artery occlusion.

For TOF MRA, the background is suppressed by virtue of the short TR in relation to the longish T1s of background tissues. Although fat remains bright, it can be eliminated by use of fat-saturation techniques, albeit with an additional time penalty. For CEMRA, there is the additional benefit that the background tissues can be completely eliminated by the subtraction of a mask acquired before the injection of contrast material.

CEMRA is the MR technique of choice for imaging the aorta in its entirety and for evaluating its large and medium-sized branches including the renal (Fig. 6.8) and proximal mesenteric arteries30,37–40. Because of the need for breath-holding the technique’s spatial resolution remains inferior to that of catheter angiography because of the inability to collect a highresolution data-set that matches that of DSA during a breathhold. Nonetheless, numerous studies and meta-analyses attest to the accuracy of MRA in clinical practice38. An additional benefit of MRA lies in its ability directly to measure the flow rate to each kidney using a two-dimensional (2D) cardiactriggered, phase-contrast approach, which facilitates both the assessment of both end-organ damage and the likelihood of success of transluminal angioplasty37. MRA is also highly accurate for depicting the mesenteric arteries in patients with suspected chronic mesenteric ischaemia39. In patients with abdominal aortic aneurysms, the external dimensions of the aneurysm can easily be delineated, should this be necessary, with targeted pre-contrast or post-contrast

MRA IN CLINICAL PRACTICE MRA is an excellent technique for imaging most vascular territories but is generally avoided in unstable and/or ventilated patients and those with severe trauma because of the hazards of the MR environment and the difficulties in monitoring patients within the MR room. Standard contra-indications to MRI (pacemakers, intracranial uneurysm clips) also preclude use of MRA.

Thoracic aorta and great arteries Because of the relatively large size of these vessels, they can be demonstrated with a wide variety of techniques but CEMRA is favoured in most instances owing to its rapid speed of acquisition and the quality of the images generated33,34.

Pulmonary arteries Although several studies have established high accuracy for MRA compared with pulmonary angiography for the evaluation of suspected pulmonary embolism, it is not widely used clinically35,36. Reasons for this include a reluctance to refer potentially unstable patients to MRI; the availability of CT pul-

Figure 6.9

Normal pulmonary MRA.

115

116

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

images as only the lumen is demonstrated with CEMRA40. Calcium within the wall is not demonstrated, however, and aneurysm assessment for planning endovascular stenting is usually performed with CTA.

Carotid arteries Because of the requirement accurately to differentiate stenosis at a 70 per cent cut-off within a relatively small (internal carotid) artery, there are stringent spatial resolution requirements for carotid MRA41. As the carotid bifurcation does not move with respiration a relatively long data acquisition that generates images with sufficiently high (isotropic) spatial resolution is recommended. Despite the fact that the resultant data acquisition is substantially longer than the arterio-venous transit

A

D

G

B

E

H

time of 8–12 s (the blood–brain barrier prevents the parenchymal extraction of gadolinium and therefore facilitates rapid transit from artery to vein), this does not lead to unacceptable venous contamination of the images as there is sufficient time within this arterio-venous window for acquisition of the contrast-defining central lines of k-space. Clearly, acquisition of these central lines must be synchronized with the peak arterial bolus by combining some form of bolus detection as described previously with a ‘centric’ order of k-space filling. In comparison with cathether angiography CEMRA has demonstrated high accuracy in evaluating the carotid artery, e.g. for differentiating between significant and insignificant stenoses, differentiating between critical stenoses and occlusions, and for depicting carotid and vertebral dissections (Fig. 6.10).

C

F Figure 6.10 T2, FLAIR and diffusion-weighted images in a patient with right-sided weakness and aphasia. All the images were acquired in the axial plane at the same level. (A–D) Note the high signal on T2 (A), FLAIR (B), and Diffusion-weighted images (C) (arrowed) with a corresponding low-signal intensity on ADC map (D), ringed indicating an acute cerebral infarct. (E, F) Whole-volume MIP images from a TOF MRA in coronal (E) and axial (F) orientation demonstrate reduced signal intensity within the left internal carotid (short arrows) and left middle cerebral arteries. Targeted MIP images of the left-sided carotid and vertebral arteries demonstrate a severe stenosis of the internal carotid artery just beyond the bulb (not shown). Axial T1 (G) and T2w (H) images with fat-saturation reveal crescentic high-signal intensity within the left ICA (long arrows) at the site of the stenosis demonstrated on CEMRA, diagnostic of acute dissection.

CHAPTER 6

• ANGIOGRAPHY: PRINCIPLES, TECHNIQUES (INCLUDING CTA AND MRA) AND COMPLICATIONS

Peripheral arteries In the peripheral arteries, as in most other areas, TOF MRA has been superseded by CEMRA. Although the spatial coverage offered by single-field-of-view imaging is insufficient to address all of the relevant anatomy, the introduction of moving-table MRA has opened the way for routine noninvasive MRA of the entire run-off arteries in a short timeframe ( 50

Eye/forehead

Nuclear medicine

2.04

5–10

injection and positioning of the patient on the couch, is difficult to shield. Typical doses received by staff in diagnostic imaging departments are shown in Tables 9.10 and 9.11.

Pregnant staff The dose to the fetus during the declared term of pregnancy (i.e. after the employer has been informed in writing) should be less than 1 mSv. For those receiving exposure to diagnostic X-rays this is equivalent to about 2 mSv to the surface of the abdomen. An individual risk assessment must be carried out for the pregnant employee but, as can be seen from Table 9.10, it is unlikely that a member of staff will exceed this dose.The evidence available indicates that there should be no need for a change in work patterns except perhaps for staff who are involved in a heavy interventional radiology workload or working with high activity unsealed sources (e.g. in the radiopharmacy). In these cases it may be advisable to change work schedules or to limit the number of procedures performed and to offer additional monitoring. However, it should be stressed that such measures are often mainly to give peace of mind to the staff concerned.

REFERENCES 1. ICRP 2000 Avoidance of radiation injuries from medical interventional procedures. ICRP Publication 85. Ann ICRP 30. Pergamon, Oxford. ICRP website: www.icrp.org/docs/ ICRP_85_Interventional_s.pps 2. ICRP 2001 Supporting Guidance 2. Radiation and your patient: a guide for medical practitioners. Ann ICRP 31. Pergamon, Oxford. ICRP website: www.icrp.org/docs/ ICRP_85_Interventional_s.pps 3. Dendy P P 2005 Low dose radiation risk: UKRC 2004 debate. Br J Radiol 78: 1–2 4. Watson S J, Jones A L, Oatway W B et al 2005 Ionising radiation exposure of the UK population: 2005 review. Health Protection Agency, Chilton 5. ICRP 1991 The 1990 recommendations of the International Commission on Radiological Protection. ICRP Publication 60. Ann ICRP 21. Pergamon, Oxford 6. ICRP 2000 Pregnancy and medical radiation. ICRP Publication 84. Ann ICRP 30. Pergamon, Oxford. ICRP website: www.icrp.org/docs/ICRP_84_ Pregnancy_s.pps 7. ICRP 2003 Biological effects after prenatal irradiation (embryo and fetus). ICRP Publication 90. Ann ICRP 33. Pergamon, Oxford. 8. The Justification of Practices Involving Ionising Radiation Regulations 2004 Statutory Instrument 2004 No. 1769. The Stationery Office, London 9. The Ionising Radiation (Medical Exposure) Regulations 2000 Statutory Instrument 2000 No. 1059. The Stationery Office, London

CHAPTER 9

• RADIATION ISSUES: RADIATION PROTECTION AND PATIENT DOSES IN DIAGNOSTIC IMAGING

10. Royal College of Radiologists 2003 Making the best use of a department of clinical radiology: guidelines for doctors, 5th edn. Royal College of Radiologists, London 11. The Medicines (Administration of Radioactive Substances) Regulations 1978 SI 1006/1978. HMSO, London. And Amendment Regulations 1995 (SI 2147/1995) 12. Health and Safety Commission 2000 Work with ionising radiation. The Ionising Radiation Regulations 1999: approved code of practice and guidance. The Stationery Office, Norwich 13. Radioactive Substances Act 1993 (Chapter 12). HMSO, London 14. Hart D, Hillier M C, Wall B F 2002 Doses to patients from medical X-ray examinations in the UK: 2000 Review. NRPB W14. NRPB, Chilton 15. Institute of Physics and Engineering in Medicine 2004 Guidance on the establishment and use of diagnostic reference levels for medical X-ray examinations. Report 88. IPEM, York 16. Hart D, Jones D G, Wall B F 1994 Estimation of effective dose in diagnostic radiology from entrance surface dose and dose–area product measurements. NRPB-R262. NRPB, Chilton 17. Hart D, Jones D G, Wall B F 1996 Coefficients for estimating effective doses from paediatric X-ray examinations. NRPB-R279. NRPB, Chilton 18. Keat N 2002 ImPACT CT patient dosimetry calculator (version 0.99m). ImPACT, London. ImPACT website: www.impactscan.org/ctdosimetry.htm 19. Institute of Physics and Engineering in Medicine 2005 The commissioning and routine testing of mammographic X-ray systems. Report 89: 1987–1997. IPEM, York 20. ICRP 1998 Radiation doses to patients from radiopharmaceuticals. ICRP Publication 80. Ann ICRP 28. Pergamon, Oxford 21. Administration of Radioactive Substances Advisory Committee 1998 Notes for guidance on the clinical administration of radiopharmaceuticals and use of sealed radioactive sources. The Stationery Office, London 22. RADAR—The internal dose assessment system. RADAR website: www. doseinfo-radar.com/RADARIntSys.html 23. Institute of Physics and Engineering in Medicine 1998 Cost-effective methods of patient dose reduction in diagnostic radiology. Report 82. IPEM, York 24. Rawlings D 2005 Options for radiation protection of the patient. Br J Radiol 78: 877–879 25. Honey I D, MacKenzie A, Evens D S 2005 Investigation of optimum energies for chest imaging using film-screen and computed radiography. Br J Radiol 78: 422–427

26. ICRP 2004 Managing patient dose in digital radiology. ICRP Publication 93. Ann ICRP 34. Elsevier, Oxford. ICRP website: www.icrp.org/docs/ ICRP_93_digital_educational_version_20April04.pdf 27. Lewis M A, Edyvean S 2005 Patient dose reduction in CT. Br J Radiol 78: 880–883 28. ICRP 2000 Managing patient dose in computed tomography (CT). ICRP Publication 87. Ann ICRP 30. Pergamon, Oxford. ICRP website: www.icrp. org/docs/ICRP_87_CT_s.pps 29. Shrimpton P C, Hillier M C, Lewis M A et al 2005 Doses from computed tomography (CT) examinations in the UK: 2003 Review. NRPB-W67. NRPB, Chilton 30. National Radiation Protection Board 1998 Diagnostic medical exposures. Advice on exposure to ionizing radiation during pregnancy. NRPB, Chilton 31. ICRP 2002 Doses to the embryo and fetus from intakes of radionuclides by the mother. ICRP Publication 88. Ann ICRP 31, corrected version. Elsevier, Oxford 32. Winer-Muram H T, Boone J M, Brown H L et al 2002 Pulmonary embolism in pregnant patients: fetal radiation dose with helical CT. Radiology 224: 487–492 33. Cook J V, Shah K, Pablot S et al 1998 Guidelines on best practice in the X-ray imaging of children. St George’s Hospital, London 34. Paediatric Task Group of the European Association of Nuclear Medicine 1990 A radiopharmaceuticals schedule for imaging in paediatrics. Eur J Med 17: 127–129 35. ICRP 2004 Doses to infants from ingestion of radionuclides in mothers’ milk. ICRP Publication 95. Ann ICRP 34. Elsevier, Oxford 36. Becket J R, Kotre C J, Michaelson J S 2003 Analysis of benefit:risk ratio and mortality reduction for the UK breast screening programme. Br J Radiol 76: 309–320 37. Whitby M, Martin C J 2003 Radiation doses to the legs of radiologists performing interventional procedures: are they a cause for concern? Br J Radiol 76: 321–327

SUGGESTED FURTHER READING Institute of Physics and Engineering in Medicine 2002 Medical and dental guidance notes. A good practice guide on all aspects of ionising radiation protection in the clinical environment. IPEM, York

171

CHAPTER

Clinical Governance and Audit in Radiology

10

Richard A. Nakielny, Adrian Manhire and Raymond J. Godwin

Clinical governance in radiology • Definition of clinical governance • Setting standards for clinical audit • Risk management • Revalidation

Clinical audit—‘reality rather than belief’ • Clinical audit’s history and the wider perspective • The audit cycle/spiral • Achieving successful organization of audit • The re-launch of audit

CLINICAL GOVERNANCE IN RADIOLOGY Richard A. Nakielny In the mid/late 1990s a series of perceived major failures of the UK Medical Health System received intense media publicity. It was felt that the professional accountability of doctors to the public needed reinforcing.The government response culminated in two White Papers, The New NHS: Modern, Dependable1 and A First Class Service: Quality in the New NHS2, in which the concept of clinical governance was introduced and defined. These two White Papers signalled a culture change from an emphasis on numbers treated to an emphasis on quality of care, cooperation and patient involvement3. The new system is a multiprofessional approach which includes all the healthcare professionals within a clinical team. One of the objectives of this change was to reduce the substantial variations in medical practice and outcome across the country. It was recognized that this change would not happen overnight and the stated target was a 10-year programme of modernization of the NHS.

responsibility for the maintenance and development of service quality as well as the financial affairs of the Trust. Most NHS Trusts have set up committees in the following areas to satisfy external reviews of compliance with clinical governance: • Clinical audit (critical analysis of clinical care set against known standards) • Risk management • Clinical effectiveness (extent to which processes do what they are intended to do, i.e. greatest health gain from available resources) • Quality assurance • Staff development • Research and development.

DEFINITION OF CLINICAL GOVERNANCE

Local self-regulation, particularly through personal audit, is a cornerstone of clinical governance. All hospital doctors will be required to participate in a national audit programme appropriate to their specialty or subspecialty2. Within radiology, written standards related to equipment, process and appropriate clinical outcome need to be in place to comply with this aspect of clinical governance. Most radiology departments already have a substantial number of standards in place, particularly with regard to equipment and process. In addition, there are Royal College of

Clinical governance was defined as ‘a framework through which NHS organisations are accountable for continuously improving quality of their services and safeguarding high standards of care by creating an environment in which excellence in clinical care will flourish’. The essence of clinical governance is local accountability for the continuous monitoring and improvement of clinical quality.The Trust Chief Executive has statutory

SETTING STANDARDS FOR CLINICAL AUDIT

174

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Radiologists’ publications containing suggested standards4. These can be categorized under the following headings: 1 Equipment standards 2 Process standards: • Referral guidelines—there is national advice on referral guidelines5 • Requests for examination—legible, sufficient clinical detail, signed, dated, etc. • Procedure—target times for waiting lists, satisfactory patient identification, systems, informed consent, radiation, protection protocols, etc. • Patient care—written information about examination, acceptable environment, user satisfaction questionnaire, defined system for dealing with patient complaints, etc. • Reporting times • Critical incident reporting • Continuing professional development (CPD) for staff— induction programmes, fire regulations, radiation protection, cardiopulmonary resuscitation training, etc. • Staffing levels—national advice is available6. 3 Outcome standards These are much more difficult to set than process standards. There are few nationally accepted figures for minimum diagnostic accuracy in an everyday work situation.The accuracies of various imaging methods are published in the literature but these levels are often obtained by specialists who are working under optimal conditions. It is essential that any standards set under the auspices of clinical governance are practicable and achievable in the working situation. • Mammography screening—this was set up with process and outcome standards in position from the outset. However, mammography screening has the advantage of dealing with a single anatomical area and, in effect, a single pathological entity. This enables outcome standards to be more readily set and monitored than in the more complex everyday work situation where a wide number of investigations and pathologies are encountered. • Interventional radiology—The Society of Cardiovascular Interventional Radiology (SCVIR) in the USA has developed and published extensive standards relating to the success and complication rates of interventional procedures.The British Society of Interventional Radiology (BSIR) is developing a comparable set of practical and achievable standards. • Radiology—setting outcomes standards in other subspecialty areas of radiology is difficult and complex. National advice has been issued on possible audit projects7 but the majority are process rather than outcome audits. This illustrates the practical difficulties of identifying viable outcome audit projects. How, for example, does one assess the accuracy of a chest X-ray report other than on the most simplistic criteria? As the issue of standards in radiology is so problematical, it may be necessary to await central directives on standards in radiology from the National Institute for Health and Clinical Excellence (NICE) (NHS, UK) after appropriate consultation with national bodies and subsequently to set up audits

in these particular areas to ensure conformance with the standards set by NICE. In the interim it will be essential to develop an involvement with audit, not only to demonstrate departmental conformance with clinical governance, but also because individual revalidation by the General Medical Council (GMC) will require proof of audit activity.

RISK MANAGEMENT Clinical governance and risk management are inextricably linked. There are two main components of risk management: • the identification of potential problems which could compromise the safety of patients, visitors and staff • the installation of procedures and protocols to minimize these. There are already many areas of well-established risk management regulations including health and safety regulations, equipment maintenance and safety, radiation safety, infection control, cardiopulmonary resuscitation, IT security, etc. The introduction of clinical governance has raised the profile of several other important aspects in the practice of radiology.

Staffing levels/workload Clinical governance emphasizes quality of care, and adequate staffing levels are a prerequisite for this. Stress caused by understaffing will affect performance.There is at present a significant undersupply of radiologists in the UK to accommodate the increasing number of referrals and the complexity of modern radiology. The Royal College of Radiologists has attempted to define a reasonable workload for a radiologist, and also to assess the impact of new clinical appointments in other clinical specialties on radiology staffing6.

Skill mix Team-working is emphasized by clinical governance. There is national support from the Royal College of Radiologists for a responsible introduction of skill mix8 so that appropriately trained, supervised and audited (usually nonmedical graduate) staff can help to offset the radiological workload. An important prerequisite for this is that both the delegator and the person to whom the task has been delegated agree to the delegation which must conform to GMC guidelines9. In particular, it must be established that the person to whom the work is delegated is competent to carry out the task.This person then assumes a clinical and medico-legal responsibility for their actions but overall medical responsibility can only be transferred, by referral, to another medically qualified practitioner. It is essential that improved quality of patient care, rather than a reduction in costs, is the main aim of skill mix. Improved quality of care may be achieved with skill mix by releasing highly trained medical practitioners from time-consuming yet relatively straightforward tasks to allow them to concentrate on tasks that require a level of expertise commensurate with their ability and training.

CHAPTER 10

Continuing professional development The concept of lifelong learning is a firmly established component of clinical governance. It is not sufficient to view a qualification examination certificate, no matter how advanced, as being the final stage of medical education. Most radiologists have been voluntarily pursuing further learning throughout their careers and the formalization of this process does not pose any major conceptual difficulty. The public require reassurance that doctors, and indeed all staff, are keeping up to date with advances in medical knowledge. Documentation of this is a vital component of risk management and is also required for revalidation. Continuing medical education (CME) forms the backbone of continuing professional development (CPD) but CPD also includes the development of managerial, appraisal, teaching and other skills where appropriate.

Quality of reporting The quality and timeliness of reporting are central to the input into patient care by a radiology department. However, the setting of standards for quality of reports is problematical. In diagnosis, errors in perception are much more common than errors of interpretation. Overload of work, fatigue, repeated distractions and environmental conditions all have an important bearing on the incidence of these errors and it is important to minimize these adverse factors. Attendance at clinicoradiological and multidisciplinary team meetings is important to enable feedback on the accuracy of reporting to occur. Good communication between team members involved in clinical care is emphasized repeatedly in clinical governance. Formal records of attendance at these meetings would be useful documentary evidence for revalidation. Formal medical discrepancy meetings (see later) to discuss (anonymously) cases where possible errors have occurred are an essential development of clinical governance.

Working beyond competence All clinicians are becoming more specialized and radiology is no exception. Subspecialization is, however, a two-edged sword. It can produce a high-quality service during normal working hours but it also leads to a situation in which oncall work may be in an area outside the expertise of the consultant on-call. Consultant staff should together define who within the department has adequate expertise to perform and/or interpret the specialized procedures that may occur in an on-call setting. Arrangements should then be put in place, possibly involving other NHS Trusts, for on-call cover for these procedures. If such cover cannot be made available on an on-call basis owing to the local situation, risk managers must be made aware of this. National advice for on-call radiological practice10 includes the following: • Only those examinations that will affect immediate patient management during the out-of-hours period should be

• CLINICAL GOVERNANCE AND AUDIT IN RADIOLOGY

performed, and each department should have a portfolio of examinations that it believes can be offered safely and reliably out of hours. This list should be agreed with the Trust. • Out of hours, a radiologist should only carry out those procedures that they are competent to perform in normal working hours. • Appropriate staff and equipment must be available for outof-hours work that would normally be available for in-hours work. Another aspect of working beyond competence is the introduction of new radiological procedures. New procedures need to be evidence-based. Careful consideration should be given to organizing adequate study leave to acquire the skills for new procedures. The cost of complications incurred while performing a new procedure for which the radiologist is not adequately trained may be considerably more than the cost of obtaining training in that procedure.

Informed consent Clinical governance stresses greater patient involvement in decision making. The attitude of the general public to the amount and quality of the information they require before consenting to medical procedures has changed radically. The ease of access to information through the media and internet, together with an irreversible move away from the presumed infallibility of doctors, has resulted in a climate in which it is no longer acceptable to give inadequate information about medical procedures. In an excellent review of informed consent11, the differences in attitude in the UK and America were highlighted. In America the law requires that the patient is given all the relevant information. This contrasts with the UK where the law allows, to some extent, clinical judgement to determine what information is given to the patient. There must still be ‘sufficient disclosure’ to allow the patient to make an informed choice. The legal meaning of ‘sufficient disclosure’ is that patients must be informed of any serious risk, even if it is of low frequency. They must also be told of less serious risks which occur more commonly. Details/risks of a procedure may only be withheld if it is felt that they are likely to cause ‘serious harm to the mental or physical health of a patient’. If a patient asks a direct question about risks, this must be answered truthfully and as fully as the patient demands, i.e. information cannot be withheld if a direct question is asked. The GMC has issued advice on informed consent12. Aspects of this advice relevant to radiological practice in the UK have been incorporated into a guidance document issued by the Royal College of Radiologists13. Careful explanation and oral consent will be sufficient for the majority of radiological investigations. High-risk procedures require a full and careful explanation, and adequate time must be allowed for the patient to assimilate this information. This should occur prior to any pre-medication. Written aids and

175

176

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

patient-focused literature have a positive role in this process. The person explaining the procedure must have sufficient knowledge and experience to answer any relevant questions fully and truthfully. Written consent is then required, but it is not a legal safeguard if complications arise that were not explained to the patient. If any information is withheld on the grounds that it may cause serious harm to the patient, this must be recorded before the procedure in the clinical notes together with the reason for doing so, as this may need to be justified in law. Examinations involving high radiation doses (e.g. CT, extended fluoroscopy) should have the risks/benefits of the procedure explained in terms that can be clearly understood. Again, pre-prepared information sheets may well be helpful. Research procedures must not be contrary to the interests of the patient, and a full explanation and written consent are mandatory.

Professional registration Policies must be in place for checking the professional registration of staff within the department. If the employment of certain groups of staff is subcontracted then the responsibility for confirming appropriate staff registration must be clearly defined.

Patient record security The confidentiality of patient records, both written and held on computer, is an important part of risk management. Clear policies must be in place for the maintenance of this security. Safeguards governing the access to, and storage of, confidential patient information must be in place.

Major accident response Clear policies defining the departmental response in the face of a major accident must be in place.

Critical incident reporting Clear policies for recording, openly discussing and disseminating any lessons learned from critical incidents must be in place. Any lessons learned must be applied promptly. The Royal College of Radiologists has produced an overview of the impact of risk management on clinical radiology14.

REVALIDATION Clinical governance incorporates both departmental and individual performance. The GMC have stated their view of the components of good medical practice for individual doctors9 and in future will require all UK doctors to undergo a revalidation process to maintain their licence to practise15. There is national guidance about how good medical practice impacts on radiologists16. However, the Shipman Inquiry findings17 (where a general practitioner was found guilty of the murder of more than 200 of his patients)

have criticized the proposed GMC revalidation procedure for failing to incorporate a robust assessment of a doctor’s fitness to practise. A leading article in the British Medical Journal18 succinctly discusses the rationale that underpins this criticism. The implementation of the GMC version of revalidation had been planned for April 2005, but this has now been postponed to allow for the incorporation of any future recommendations resulting from the Shipman Inquiry. The following is a summary of the principles of the GMC revalidation so far; however, it should be borne in mind that there will be additions to this when further recommendations are published, taking into account the findings of the Shipman Inquiry. • Individual clinical performance meets any national professional standards. • The basis will be appraisal and assessment of local performance in the workplace in relation to any appropriate national standards (i.e. an examination system is not thought to be appropriate). • It is seen to be fair to doctors and open and clear to the public and employers. • It is capable of appraising and assessing all doctors whatever the circumstances of their practice. • It is simple, unobtrusive, economical in time and effort, and is as inexpensive as is consistent with effectiveness (i.e. detailed performance assessment will only be invoked in cases in which there is local evidence of serious dysfunction in performance). It has been proposed that revalidation should be primarily based on the outcome of the local annual appraisal process for those employed within a managed setting, and will occur on a 5-yearly cycle. It will be essential for doctors to collect and maintain appropriate documentary evidence for revalidation. The GMC has stated that there will need to be evidence available in the following categories in order to obtain revalidation.

Suggested documentary evidence for radiologists 1 Good medical practice: • Audit results and record of attendance at audit meetings • Medical discrepancy personal records • Record of attendance at medical discrepancy meetings. 2 Maintaining good medical practice: • CME/CPD records • Personal development plan • Record of attendance at clinicoradiological and/or multidisciplinary team meetings. 3 Working with colleagues: • 360-degree appraisal documentation (see later) • Record of attendance at any radiology team meetings. 4 Relationships with patients: • 360-degree appraisal documentation (see later) • Record of complaints/plaudits.

CHAPTER 10

5 Research 6 Teaching/training: • Feedback documentation (anonymous). 7 Health/probity:

• Sickness record • Self-signed statement that ‘health (of the doctor) has never endangered patients or colleagues’ • Self-signed statement that ‘conduct work to highest ethical and moral standards’ • 360-degree appraisal documentation (see later). Most of the above documentary evidence can be collected relatively easily. However, documentary evidence for two of the cornerstones of clinical governance—namely medical discrepancy meetings (quality improvement by learning from errors) and 360-degree appraisals (team working)—requires more active organization. The following is a précis of suggested methods for setting up these two processes with the emphasis on simplicity.

Medical discrepancy meetings • Empathic lead person: involves everyone in the process, encouraging a constructive, nonconfrontational atmosphere. • Case collection: lockable collection boxes placed in easyto-reach sites with standardized forms adjacent (and replenished regularly!). • Meetings at regular intervals (e.g. monthly). • Cases presented by a lead person: the radiologist involved should remain anonymous. It is impossible to re-create the original reporting conditions but it is important to present the same clinical information that the reporting radiologist had available. • Discussion focused on learning, not blame. • Consensus vote on whether error actually has occurred. • Simple consensus scoring system for degree of error (grade and significance). • Lead person gives confidential feedback if an error has occurred. • Annual analysis by lead person of cases discussed to see if there are any patterns that may require a more structured solution. • Attendance recorded formally (documentary evidence for revalidation).

360-degree appraisal Satisfactory team working and a willingness to listen and act on constructive comments about performance from patients and colleagues (medical and nonmedical) are essential in clinical governance.Three hundred and sixty degree appraisal is a process where the views of patients and colleagues are

• CLINICAL GOVERNANCE AND AUDIT IN RADIOLOGY

gathered together and fed back to the individual at predetermined time intervals (approximately 3 years if there are no significant problems). Although 360-degree appraisal is viewed as an essential part of the overall appraisal process (and hence for revalidation), it must be stressed that 360degree appraisal is only one part of this overall appraisal, and is not a pass/fail process. The tasks and interactions of radiologists are different from those of physicians and surgeons. Consequently, the following suggested method for 360-degree appraisal has been adapted to conform to the requirements of radiologists, and has been issued as guidance by the Royal College of Radiologists on their website, www.rcr.ac.uk. The essence of the 360-degree appraisal questionnaire is as follows: • Five sections, the first three to be completed only by the relevant professional group (medical colleagues, radiographers/nurses and clerical/secretarial staff) and the final two sections to be completed by all staff groups. • Each section has about 10 simple questions relevant to that professional group in appraising the radiologist. There is a simple numerical scoring system ranging from 1 (poor) to 10 (excellent). • Within the radiographer/nurse section there are questions relating to patient interactions. This element allows radiologists with infrequent patient contact to obtain some documentary evidence for the ‘relationships with patients’ section for revalidation. • The penultimate section contains a simple question on the health and probity of the radiologist undergoing the 360degree appraisal. • The final section is for free text comments. • A pilot study has shown that the average time taken to complete the questionnaire is 7 min. • A minimum of 10–12 questionnaires (e.g. four from each of the three professional groups chosen) must be completed to give a reasonable overview and also to maintain anonymity of the staff completing the questionnaires. • A system for collecting the questionnaires anonymously and analysing them for feedback at the overall appraisal process needs to be in place. • A reasonable time interval between 360-degree appraisals, assuming there are no particular problems, would be 3 years to allow documentary evidence to be available for the 5yearly revalidation process. In an age where form filling is in danger of proliferating out of control, it is important that the 360-degree appraisal questionnaire is kept as simple as possible. If any significant problem areas are identified, these can then have in-depth assessment at the annual appraisal.

177

178

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

CLINICAL AUDIT—‘REALITY RATHER THAN BELIEF’ Adrian Manhire and Raymond J. Godwin You will almost certainly have been referring to other chapters in this publication, using it as a reliable source of current opinion in diagnostic radiology, searching for best practice and the latest knowledge, and for research evidence in support of it. Having gathered such knowledge, are you able to show that you are practising to these new standards of care? Clinical audit is the tool that should enable you to produce evidence to show that you have achieved these standards in your own practice and you need to be able to do this. Audit is an integral part of clinical governance. In the foreword to Principles for Best Practice in Clinical Audit, Hine (Chair of Commission for Health Improvement) and Rawlins (Chairman, National Institute for Clinical Excellence) state19: Public and professional belief in the essential quality of clinical care has been hit hard in recent years, not least by a number of highly public failures. Clinical governance is the organizational approach for quality that integrates the perspectives of staff, patients and their carers and those charged with managing our health service. Clinical audit is at the heart of clinical governance.

One of the most prominent of the public failures was investigated in the Shipman report20. How well doctors carry out their professional activities has also been brought to the centre of public debate, along with the process of audit as a method of enquiry21. In this section, we review the origins of clinical audit, consider what audit is, how it can be carried out, how it can help underpin clinical governance, and how it might be built into departmental practices, creating the environment in which clinical audit can flourish.

The definition of audit The systematic, critical analysis of the quality of medical or clinical care, including the procedures used for diagnosis and treatment, the use of resources, and the resulting outcome and quality of life for the patient. Secretaries of State for Health, Wales, Northern Ireland and Scotland (1989)22

CLINICAL AUDIT’S HISTORY AND THE WIDER PERSPECTIVE Audit of medical care is not new. As early as the Crimean War (1854–1856), Florence Nightingale used a form of audit as an aid to her management of the injured and sick in her care, using standardized methods for the collection of information on death and infection23. She used this audit evidence to assist her argument for resources and changes in practice. In the United States, in 1917, the American College of Surgeons introduced a process of reviewing clinical notes against a set of minimum explicit standards, questioning their quality and adequacy of facilities (including radiology)24. These processes developed into criterion-based patient outcome audit

by 1975 under the overview of a national body, the Joint Commission on Accreditation of Healthcare Organisations25. At the same time, mainly as an aid to keeping public healthcare spending under control, the US Congress created Professional Standards Review Organisations. These were instituted to review the appropriateness of medical services and their quality through medical audit. There is a marked similarity to the development of the Commission for Health Improvement (CHI) and the UK National Institute for Health and Clinical Excellence (NICE).

Audit in the UK In the United Kingdom, prior to the late 1980s, there was no requirement within the NHS for clinicians to demonstrate any evidence of the quality of their clinical practice. In 1989, the UK government introduced medical audit as a requirement within all doctors’ job plans22, extended this in 1997 to include all healthcare professionals as clinical audit1,2,26, and integrated it into clinical governance. Together with the logical requirement to practise evidence-based medicine, the infrastructure for clinical governance was now in place, awaiting its formal introduction2. Not only is there now a mandatory requirement for all doctors to participate in clinical audit, the GMC9 advises all doctors that they: …must take part in regular and systematic medical and clinical audit, recording data honestly. Where necessary, you must respond to the results to improve your practice, for example by undertaking further training.

Although most doctors routinely practise audit informally by comparing their work to published data, there is now a clear obligation to record this and have it available to validate their personal practice and that of their clinical team, department and hospital.

Making it possible There are two factors that are important for successful audit: • creating a local environment that is supportive of audit (including providing adequate resources in terms of time and assistance, and ensuring that the resulting change occurs) • using audit methods that are most likely to lead to audit projects that result in real improvement.

Difficulties in audit Many of those involved in audit have unfortunately lost enthusiasm because of the difficulties that they have encountered. • Poor project design has led to data of poor quality. Information has been collected because it is available rather than being a relevant measure of clinical quality. • Many projects are poorly managed. Demonstrating inadequate care is not sufficient unless it can be carried through into changes that improve practice. Change is

CHAPTER 10

often the most difficult part of audit but it is often left to inexperienced junior staff without appropriate support, influence and resources. As much attention needs to be devoted to change for improvement as to the collection and analysis of data. It is the perception of improvements in care that drives the individual on to further audit.

What does audit really mean? Many clinicians find the term ‘audit’, with its financial undertones and rather formal definition (above), confusing and difficult to remember. Fowkes27 has given a more pragmatic definition: ‘Comparison of actual practice to a standard of practice and as a result of the comparison, any deficiencies in actual practice may be identified and change undertaken to rectify the deficiencies.’

The words ‘actual practice’, ‘standard of practice’, ‘comparison’, ‘deficiencies’ and ‘change undertaken to rectify’ clearly describe the audit cycle and emphasize the essential need for change for the better if standards are not achieved. Donabedian28 has subdivided audit into three types: structure, process and patient health outcome. Structure—What you need The availability and organization of resources (material as well as human) required for the delivery of a service. An example of this is the availability of adequate resuscitation equipment within a department of radiology. Process—What you do How well has a required procedure been followed? Have all radiology reports been checked and validated by the reporting radiologist prior to circulation? Patient health outcome—What you expect This describes the alteration in healthcare status of an individual which is directly attributable to clinical intervention. Outcome audits in radiology are often related to interventional radiology procedures, where the clinical improvement achievable with such techniques is more readily measured; however, it is not unreasonable to include accuracy of imaging diagnosis as an outcome audit, as it can also aid in the patient’s clinical improvement or cure29.

Audit is not research Research and audit are often confused.There are clear differences, the awareness of which enables differentiation. Donabedian30 distinguished between the assessment of medical technology Table 10.1

• CLINICAL GOVERNANCE AND AUDIT IN RADIOLOGY

(research) and the assessment of the quality of medical care (audit) (Table 10.1). The most important illustration of these differences is that an audit can be carried out on a relatively small number of cases and does not always require the time and expense of a research programme. Research sets the standards and audit determines whether clinical practice meets them.

What not to audit There is a tendency for new users of audit methodology to use it as a tool to demonstrate the inadequacies of other clinicians’ practice. It is tempting to audit how others use imaging services and the use of the published guidance on how best to use imaging encourages this31,32. Audit of third parties is unlikely to achieve the essential elements of change and improvement, which are the key features of a successful audit33. All involved should be committed to the audit or it will generate ill-feeling and the results are likely to be dismissed or ignored. Using such standards as part of an inclusive multidisciplinary audit process alongside nonradiological colleagues is much more likely to achieve change and reinforce, not strain, local professional relationships. Audit of personal practice should take the prime place. A standard is the keystone to the process and unless a clear standard is identified at the beginning of an audit, it is unlikely to succeed. It helps to ensure that a project is audit and not research. Audit activity must be relevant to current local activity and needs, and must reflect the problems encountered in everyday work.The likely required change should be achievable, otherwise effort used in trying to implement it is likely to be wasted. It is also better not to attempt audits with a high level of complexity, as these are more likely to fail to complete their second cycle34.

THE AUDIT CYCLE/SPIRAL The main aims of audit are to demonstrate either: • compliance with an agreed standard of care, or • to use the results of the initial audit to identify possible change(s) which, following implementation of those changes, may enable the standard to be achieved. The original concept was the audit cycle (or loop) (Fig. 10.1). Firstly, a topic is chosen for audit and a standard is identified. By identifying a suitable indicator (see later) and collection of related data, the reality of practice is identified and compared with a previously agreed target. • If the set target is achieved, the audit, on this occasion, is completed, reassurance has been achieved, and the audit result is available as governance evidence.

DIFFERENCES BETWEEN RESEARCH AND AUDIT

Research

Audit

Identifies what is best practice

Determines if this has been put into practice

Is concerned with techniques, instruments or materials

Is concerned with the performance of individuals or teams

Uses statistical models and usually requires statistical compliance

Does not have to reach statistical significance

Usually requires a long time scale for completion

May be carried out in a very short time (sometimes a matter of a few hours)

179

180

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

Select a topic

Accept a standard of practice (e.g. from Research)

Observe your practice (i.e. Audit)

Implement change

Compare your practice with the standard

Figure 10.1 The audit cycle.

• If the target is not achieved, the need for some form of change is indicated to enable the required improvement in performance. After the introduction of the agreed change(s), the process of data collection is repeated. The second cycle will show if the changes have improved practice and whether the target has been achieved. The concept of the audit spiral (Fig. 10.2) adds a third dimension of continued improvement, recognizing that standards and targets can change with time and new developments. The more important audits are likely to be continuous processes, with multiple cycles year on year, rather than closed loops.

a successful audit such as the success and complication rates of angioplasty. Audit is a tool for showing how a practitioner measures up to local and national standards which are now becoming more readily available. In the UK, nationwide audits are now organized by the Royal College of Radiologists (RCR) or one of the affiliated clinical interest bodies, such as the audit of nephrostomy practice (RCR and British Society of Interventional Radiology).This enables comparison of local practice and its outcomes with similar institutions across the country. At the individual level, the collected evidence may assist in revalidation, and take its place in a personal folder for discussion at annual appraisal. However, with clinical governance, there is a need to demonstrate corporate accountability for clinical performance2 through a process of ‘regular and systematic formal Clinical Review’ (or clinical audit). Quality monitoring within the specific areas described earlier will be essential to provide this evidence. Some of these areas may require the creation of running audits as a monitoring process (e.g. waiting times for radiological examinations). Suggestions for what these areas of audit could be are available in a publication from the RCR35 which contains illustrative recipes. Topics and standards for governance issues are frequently incorporated into NICE guidance and Department of Health publications. An illustration of likely audit activity related to governance is shown in Figure 10.3. It is also possible to prioritize the choice of subjects for audit using the list below36, recognizing that not all areas can be audited at once. As an aid to prioritization, consider audit topics of activities which involve: • high risk • high volume • high cost • wide variation in clinical practice • local clinical anxiety (e.g. untoward events or questionable clinical performance).

Choosing topics to audit

What standard should be used?

Areas of concern in clinical practice that arouse the interest of an individual are more likely to produce enthusiasm for

When designing an audit, it is essential to identify an appropriate standard at the beginning of the process.

Figure 10.2 The audit spiral. (From Godwin R J, DeLacey G, Manhire A (eds) 1996 Clinical audit in radiology: 100+ recipes. Royal College of Radiologists, London, with permission.)

5 Re-audit 1 Select a standard 4 Implement change 2 Assess local practice

3 Compare with standard

Improvement or reassurance

CHAPTER 10

• CLINICAL GOVERNANCE AND AUDIT IN RADIOLOGY

Adverse events detected. Investigated. Lessons learnt and translated into change in practice.

Poor clinical performance identified early. Then dealt with skilfully, speedily and sensitively, in order to avoid harm to patients.

Quality of data necessary for monitoring clinical care to be of a consistently high standard.

Leadership skills developed at the clinical team level.

Systematic learning from clinical complaints. Translated into change in practice.

Where is your evidence

?

Continuing professional development programmes in place.

Quality improvement processes (clinical audit) to be integrated into an organizational quality programme.

Clinical risk reduction programme in place and of high quality.

Evidence-based practice and infrastructure in place and utilized.

Clinical audit will provide: • the evidence • indication of where changes need to be made • the help needed in order to meet the Trust’s statutory obligation1,2

Figure 10.3 Acquiring the evidence on effective governance. (Courtesy of Dr G DeLacey.)

Standards may be based upon research evidence, but this is frequently not available. A guideline may also be used as a standard, derived either nationally (e.g. from a specialty group or the National Institute for Health and Clinical Excellence) or locally agreed, based upon the best available information, respected opinion and local circumstances. Each standard has three components: A recommendation + an indicator + a target A Recommendation: a statement about the structure, process or outcome against which the quality of performance is to be judged. B Indicator: the variable (or item of information) that needs to be measured in order to determine whether the recommendation is being met.This is also known as the criterion. It may be represented as a percentage of compliance with a standard. C Target: the expected level of achievement or the minimum score that is considered locally to be acceptable in

good practice. It may be that when auditing for the second or third time, the target can be gradually raised. In this way, early failure, disappointment and disillusion might be avoided33.

The measurable indicator There may be more than one indicator within a single audit, each representing a step along a multilevel standard. This situation arises within audits of care pathways, where a number of criteria for completeness are required for the standard to be achieved. As an example of an indicator within an audit of double reading of breast screening mammograms, the recommendation might be that ‘all screening mammograms will be read by two radiologists’. The indicator here would be the actual percentage of mammograms reported by two radiologists.The target would be the minimum percentage conformity such as 90%. Similarly, the recommendation may be that pneumothoraces after lung biopsy should not exceed 15% at 1 h post-procedure.

181

182

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

The indicator would be the number of patients with a pneumothorax on a chest film taken at 1 h post-procedure, and the initial target might be 20%. The standard will indicate ideal expectation, the indicator signifies reality, and the target gives the required minimum achievable result locally, perhaps during the development period of the biopsy service.

Numbers for audit As mentioned in the section on audit and research, large numbers are not necessary for a successful audit and may indeed lead to failure of completion of the audit cycle. The choice of sample size can be difficult and should be decided for each individual circumstance.The number of cases or episodes audited should reflect the number seen locally in practice. There is no need for controls. In audits where a high percentage compliance is required (e.g. 100%) a relatively small sample might show failure of compliance early on (even after review of the first case). It may be that in an audit of a high risk, low frequency activity (e.g. percutaneous nephrolithotomy), the preference may be to include all cases for audit prospectively, running the audit as a continuous process, reviewing results on a month-by-month, or year-by-year basis, in order to show trends. For a higher frequency clinical activity, it may be decided to review only a randomly chosen 5% sample of cases or reports within the audit (e.g. double reporting of CT staging scans in cases of lung cancer). The error rate in reporting can be a useful audit process to improve the quality of report content and structure37. Audit of even a low number of cases may indicate that there is a cause for concern and that further work is required. It is important not to draw premature conclusions without considering all the factors influencing the initial audit outcome. An excellent analysis of numbers in audit is available in a RCR governance publication (see Suggested Further Reading).

Implementing change and re-audit The most difficult, but also the most satisfying aspect of audit, is the successful implementation of any required or recommended change. The change required will depend upon local circumstances, available resources, and a willingness to adopt change in practice. The most important aid to introducing change is the acceptance by those involved in the audit of the need for change, should the audit show a failure to reach target. It is also essential to involve and inform those with the authority to introduce (and fund) change, and identify and empower those who will implement the changes. It has been shown that when the real cost of carrying out an audit is known to managers and clinicians, the recommended changes are more likely to be implemented38. Clinical audit can be a powerful force in the process of introducing new techniques, managing staff development and creating improved services and circumstances for staff and patients. As a business tool, it can be a two-edged sword, not only identifying poor performance but also supplying the evidence of need where resource inadequa-

cies exist39,40. The date by which any changes should have been introduced must be made clear and also the date by which any second audit will be carried out. Re-audit is an essential part of the process in order to demonstrate that the changes have really produced the expected improvements. To these ends, it is essential to create written reports of any audits carried out, with the associated recommendations for change clearly stated. Circulate these effectively and use them in the clinical governance report to the directorate and Trust. These actions will make the results and the process for change explicit and available as part of the governance process, and available for review. Here we come to the important matter of confidentiality.

Confidentiality In the collection of data during the audit process, details about patients and clinicians will be identified. If this information is available within the public domain, clinicians will inevitably become less willing to give further information and to cooperate. Although the results of audit need to be available to those with a legitimate interest in generating high clinical standards (Trusts, Royal Colleges, managers, purchasers and patients), such results should be of a general nature rather than person specific. Such person-specific audit information needs to be protected. There is, of course, a requirement that a responsible individual within each Trust (usually the Medical Director) has access to information relating to any one individual or group. This is particularly important where matters of clinical performance are brought into question. The GMC recommends that patient data should be kept anonymous for the protection of individuals41.

ACHIEVING SUCCESSFUL ORGANIZATION OF AUDIT The essential requirements for successful departmental audit are time, facilities, clarity of organization and responsibility, multidisciplinary involvement and a readiness to accept change. An absence of any of these, particularly time42, makes the achievement of successful audit much more difficult. Time allocation for audit work and meetings to receive and discuss results is essential. It is part of the agreed job plan for doctors under the new NHS contract introduced in 2003 and should be achievable and acceptable to the directorate. Advice and help with data collection and IT support are present within all Trusts, funding having been allocated for audit staff. Each department requires a clearly identifiable leader for audit with the responsibility to organize meetings, coordinate appropriate audits and create an annual audit report to the Trust43. The audit leader is also supported by the RCR audit subcommittee. Opportunities should be sought for cross-specialty audit and the creation of multidisciplinary care pathways with agreed standards.

CHAPTER 10

All staff groups should ideally be included within the audit processes and consideration must also be given to inclusion of patients in the audit process44.

THE RE-LAUNCH OF AUDIT It is well recognized that since its launch in the UK, the expected improvements in clinical practice from clinical audit have not materialized, despite major investment and work by many to facilitate its acceptance. One reason for this is the tendency for medical staff to see audit as a time-consuming process separate from the rest of their clinical activities, rather than as an integrating tool to show what they are achieving and what resources they need. The bulk of the work is often left to junior medical staff or radiographers who do have a role to play as team members. Furthermore, there may well be different agendas and priorities held by clinicians and managers. With the requirement to develop real clinical governance, and the responsibility for quality of care now clearly lying with chief executives, we may now see the tool of audit being used more coherently with less divergence of opinion and priority45. The development of multidisciplinary working and clear standards of care and process in recent years should make it easier to identify effective and reproducible audit processes that can be shared nationally and allow comparison of outcomes and methods over the whole National Health Service.

REFERENCES 1. Department of Health 1997 The new NHS: modern, dependable. The Stationery Office, London 2. Dobson F 1998 A first class service: quality in the new NHS. Government White Paper 35. The Stationery Office, London 3. Scally G, Donaldson L J 1998 Clinical governance and the drive for quality improvement in the new NHS in England. Br Med J 317: 61–65 4. Royal College of Radiologists 1999 Good practice guide for clinical radiologists. RCR, London 5. Royal College of Radiologists 2003 Making the best use of a department of clinical radiology: guidelines for doctors, 5th edn. RCR, London 6. Royal College of Radiologists 1999 Workload and manpower in clinical radiology. RCR, London 7. Royal College of Radiologists 2000 Clinical governance and revalidation: a practical guide for radiologists. RCR, London 8. Royal College of Radiologists 1999 Skills mix in clinical radiology. RCR, London 9. General Medical Council 2001 Good medical practice. GMC, London 10. Royal College of Radiologists 1996 Advice to clinical radiology members and fellows with regard to out of hours working. RCR, London 11. Panting G 1999 Where do we stand on informed consent? Continuing Medical Education Journal: Radiology Update 1: 29–31 12. General Medical Council 1999 Seeking patients’ consent: the ethical considerations. GMC, London 13. Royal College of Radiologists 1999 Guidance on consent by patients to examination or treatment in a Department of Clinical Radiology. RCR, London 14. Royal College of Radiologists 2002 Risk management in clinical radiology. RCR, London

• CLINICAL GOVERNANCE AND AUDIT IN RADIOLOGY

15. General Medical Council 2003 A licence to practise and revalidation. GMC, London 16. Royal College of Radiologists 2004 Individual responsibilities—a guide to good medical practice for clinical radiologists. RCR, London 17. Shipman Inquiry 2004 Fifth report—Safeguarding patients: lessons from the past, proposals for the future. HMSO website: www.theshipman- inquiry.org.uk/fifthreport.asp 18. Smith R 2005 The GMC: expediency before principle. Br Med J 330: 1–2 19. National Institute for Clinical Excellence (NICE) 2002 Principles for best practice in clinical audit. Radcliffe Medical Press, Oxford 20. Department of Health 2000 Harold Shipman’s clinical practice 1974–1998: a review commissioned by the Chief Medical Officer. The Stationery Office, London 21. Lanier D C, Roland M, Burstin H, Knottnerus J A 2003 Doctor performance and public accountability. Lancet 362: 1404–1408 22. Secretaries of State for Health, Wales, Northern Ireland and Scotland 1989 Working for patients. Working Paper 6: Medical audit. HMSO, London 23. Nightingale F 1863 Notes on hospitals, 3rd edn. Longman Green/ Longman, Roberts and Green, London, p 63 24. American College of Surgeons 1924 The minimum standard of the American College of Surgeons’ hospital standardisation program. Bull Am Coll Surgeons 8: 4 25. Joint Commission on Accreditation of Hospitals 1975 Supplement to the accreditation manual for hospitals. JCAH, Chicago 26. Department of Health 2000 The NHS plan: a plan for investment, a plan for reform. The Stationery Office, London 27. Fowkes F R G 1982 Medical audit cycle. A review of methods and research in clinical practice. Med Educ 16: 228–238 28. Donabedian A 1966 Evaluating the quality of medical care. Milbank Mem Fund Q 44: 166–206 29. Godwin R J, DeLacey G, Manhire A (eds) 1996 Clinical audit in radiology: 100+ recipes. RCR, London 30. Donabedian A 1988 The assessment of technology and quality. A comparative study of certainties and ambiguities. Int J Technol Assess Health Care 4: 487–496 31. Royal College of Radiologists Working Party 2003 Making the best use of a Department of Clinical Radiology: guidelines for doctors, 5th edn. RCR, London 32. McCreath G T, O’Neill K F, Kincaid W C, Hay L A 1999 Audit of chest X-rays in general practice: a case for local guidelines? Health Bull 57: 180–185 33. Godwin R 1995 Nothing succeeds like success—some do’s and don’ts in clinical audit. Clin Radiol 50: 818–820 34. Jackson G 1997 Clinical audit—KISS is better. Int J Clin Pract 51: 83 35. DeLacey G, Godwin R J, Manhire A R (eds) 2000 Clinical governance and revalidation. RCR, London 36. Shaw C 1989 Medical audit: a hospital handbook. King’s Fund Centre, London 37. Peters M A, Bomanji J, Costa D C et al 2004 Clinical audit in nuclear medicine. Nucl Med Commun 25: 97–103 38. Tomalin D, Renshaw M 1999 Clinical audit. Count the cost. Health Serv J 109: 28–29 39. DeLacey G 1995 Don’t look a gift horse in the mouth. Clin Radiol 87: 815–817 40. Jackson S 1999 Achieving a culture of continuous improvement by adopting the principles of self assessment and business excellence. Int J Health Care Qual Assur Inc Leadersh Health Serv 12: 59–64 41. General Medical Council 2004 Confidentiality: protecting and providing information. GMC, London 42. Manhire A, Cook A, Adam J et al 1998 Audit in radiology: a survey of hospital departments in the UK. Health Trends 30: 72–77 43. Renshaw M, Ireland A 2003 Specialty audit leads: has this concept been effective in implementing clinical audit in an acute hospital? Int J Health Care Qual Assur Inc Leadersh Health Serv 16: 136–142 44. Avis M 1997 Incorporating patients’ voices in the audit process. Qual Health Care 6: 86–91 45. Berger A 1998 Why doesn’t audit work? Br Med J 316: 875–876

183

184

SECTION 1

• IMAGING TECHNIQUES AND GENERAL ISSUES

SUGGESTED FURTHER READING: CLINICAL AUDIT DeLacey G, Godwin R J, Manhire A R (eds) 2000 Clinical governance and revalidation. RCR, London Dixon N 1996 Good practice in clinical audit—a summary of selected literature to support criteria for clinical audit. National Centre for Clinical Audit, London

Godwin R J, DeLacey G, Manhire A (eds) 1996 Clinical audit in radiology: 100+ recipes. RCR, London National Institute for Clinical Excellence 2002 Principles for best practice in clinical audit. Radcliffe Medical Press, Oxford

CHAPTER

Techniques in Thoracic Imaging

11

Zelena A. Aziz and David M. Hansell

• • • • • • • •

Chest radiography Computed tomography High-resolution computed tomography Ultrasound Magnetic resonance imaging Ventilation–perfusion scintigraphy Positron emission tomography Positron emission tomography–computed tomography

Chest radiography and computed tomography (CT) are the two most important imaging tests used to evaluate respiratory disease. The basic technique of chest radiography has changed little over the past 100 years but recent developments in image receptor technology have resulted in the more efficient acquisition of chest radiographs with the benefit of a lower radiation dose. These images are produced in digital format, so facilitating their incorporation into picture archiving and communications systems (PACS). Advancing CT technology has meant that multidetector row CT (MDCT) systems have largely replaced single-detector CT. The resultant increase in data acquisition speed and z-axis spatial resolution has meant that volumetric high-resolution acquisitions are increasingly becoming the norm. Protocols for MDCT continue to be developed and refined, and currently, particular attention is being directed at dose-reducing strategies. The recent development of fused CT–positron emission tomography (PET) images has revolutionized the investigation of patients with suspected neoplastic disease, enabling the simultaneous assessment of both metabolic function and anatomical location.The role of other imaging techniques such as magnetic resonance imaging (MRI) and ultrasound (US) is limited to specific clinical situations and this is likely to remain the case given the obvious capabilities of MDCT.

CHEST RADIOGRAPHY Many chest radiographs are still acquired with conventional film–screen radiography systems that provide, at low cost,

good image quality and high spatial resolution1,2. However, the disadvantages of film–screen radiography are a limited exposure range, a relatively high retake rate and inflexibility of image display and manipulation1. As computer technology and storage capacities have developed over recent years, the considerable advantages of digital imaging systems and of PACS have become increasingly evident. As a result, digital imaging systems are now commonplace in radiology departments. Early digital imaging systems used a photostimulable phosphor image receptor plate (generally termed ‘computed radiography’ [CR]); these CR systems continue to be widely used because of their compatibility with existing radiography equipment. The phosphor plate stores some of the energy of the incident X-ray as a latent image. On scanning the plate with a laser beam, the stored energy is emitted as light that is detected by a photomultiplier and converted to a digital signal. The digital information is then manipulated, displayed and stored in whatever format is desired. More recently, full-field digital amorphous silicon flat-panel X-ray detector radiography systems based on caesium iodide and amorphous silicon have become commercially available. These thin film transistor (TFT) flat-panel detector systems (also referred to as DR [direct radiography]) are now widely available.The advantages of this technology include high detection efficiency and rapid image display.These systems have excellent image quality3,4 and allow a significant reduction in effective dose compared with either conventional film–screen or storage phosphor based CR systems5. Digital radiography systems have many advantages over conventional radiography: the photostimulable phosphor plate is reusable, user controlled post-processing is automatically performed to generate the display features desired for the anatomical part selected, and there is more efficient image archiving, retrieving and transmission. One of the most important advantages over conventional radiography is the wide dynamic range or latitude of the image plate—consequently, exposure errors are reduced and the need for repeat examinations is lessened.

Additional radiographic views Frontal and lateral projections are adequate for most purposes. Other radiographic views are less frequently performed

188

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

because of the ready availability of cross-sectional imaging. One projection that is occasionally used is the lateral decubitus view, taken as a frontal projection with a horizontal beam and the patient lying on his/her side. Its main purpose is to identify an effusion that is not visible on an erect chest radiograph, or to demonstrate the movement of fluid in the pleural space.The lateral decubitus view is recommended by American Thoracic Society (ATS) guidelines for all patients presenting with community-acquired pneumonia; the rationale being that if the thickness of the effusion on the lateral decubitus view is < 1 cm, the effusion is small enough not to require further intervention6 (in practice this guideline is studiously ignored).The contrary view is that a parapneumonic effusion large enough to be potentially clinically significant can usually be defined by postero-anterior (PA) and lateral radiographs, and that if loculations are suspected, then CT will define the pleural space more accurately than a lateral decubitus radiograph7. Occasionally the lateral decubitus radiograph can confirm the presence of small amounts of fluid if there is a subpulmonic effusion or if the costophrenic angles are obscured by a pulmonary infiltrate. Radiographs exposed in expiration are valuable in the investigation of air trapping, particularly in paediatric practice in the context of a suspected inhaled foreign body. An expiratory radiograph may also enhance the demonstration of a small pneumothorax.

Portable chest radiography The imaging problems associated with portable chest radiography are (A) scattered radiation; (B) the inability of the radiograph to capture all relevant information; and (C) the lack of control over the overall optical density of the resulting image when there is slight over- or under-exposure. Additionally, the shorter focus–film distance results in undesirable, and sometines misleading, magnification of structures. High kilovoltage techniques cannot be used because portable machines are unable to deliver a sufficiently high kilovoltage, and as the maximum current is limited, long exposure times are needed, increasing movement artefact.The development of CR systems has provided solutions to some of the limitations of portable chest radiography by controlling optical density and contrast, but it does not eliminate the problem of scatter. Though their instant readout capabilities would be a great advantage at the bedside, DR detectors are not applicable to portable work, largely due to cost considerations. Therefore, it is expected that storage phosphors will continue to be the media of choice for portable radiography for some time to come.

COMPUTED TOMOGRAPHY The introduction of spiral (helical) CT in the early 1990s constituted a fundamental evolutionary step in the ongoing refinement of CT imaging, replacing the discontinuous acquisition of data in conventional CT with volumetric data acquisition. In 1998 several CT manufacturers introduced MDCT systems, which provided considerable improvement in data acquisition speed and longitudinal resolution, and more efficient use of X-rays8–10. These systems typically offered simultaneous

acquisition of four sections with a gantry rotation time of 0.5 s. Since then, there has been further rapid improvement in scanner performance with increased numbers of detector rows and faster tube rotation; currently, systems with 16-, 32-, 40- and 64-active detector rows are available. Rotation times of the X-ray tubes have decreased from 0.5 s to 0.33 s per rotation. The faster data acquisition enables not only better coverage in a single breath-hold, but also results in a significant reduction in patient movement artefacts. In paediatric practice this has meant less frequent need for sedation11. The introduction of MDCT has expanded the clinical indications for CT; these are summarized in Table 11.1. With MDCT systems, different section widths are achieved by collimating and adding together the signals of neighbouring detector rows. The Somatom Sensation 4 system (Siemens, Forchheim), for example, uses the adaptive array detector design and has eight detector rows. Their widths in the longitudinal direction range from 1 to 5 mm at the isocentre and this arrangement allows the following collimated section widths: two sections at 0.5 mm, four at 1 mm, four at 2.5 mm, four at 5 mm, two at 8 mm and two at 10 mm. Currently, there is a trend amongst thoracic radiologists towards acquiring high-resolution (1–1.25-mm thickness) volumetric images which can then be reconstructed at 1.25–5-mm intervals for interpretation depending on the clinical question. Hence, from the same dataset, both narrow sections for high spatial resolution detail or three-dimensional (3D) post-processing, and wide sections for better contrast resolution or quick review, can be derived.The convenience of a single protocol is particularly useful for patients with suspected focal and interstitial lung disease. Thin section reconstructions are recommended for volumetric assessment12 and characterization13 of pulmonary nodules, the evalution of interstitial lung disease and the evaluation of pulmonary embolism14, whereas 3–5-mm reconstructions are usually adequate for the initial assessment of mediastinal masses and for lung cancer staging studies. In younger patients, however, a more critical approach

Table 11.1

INDICATIONS FOR CT OF THE CHEST

In the acute setting Chest trauma Evaluation of acute aortic syndromes (dissection, transection) Demonstration of pulmonary embolism Identification of complications post thoracic surgery (mediastinal haematomas, complex pleural collections) In the nonacute setting Further evaluation of nodules, hilar or mediastinal masses identified on a chest radiograph Lung cancer diagnosis and staging Assessment of congenital anomalies of the thoracic great vessels Characterization of interstitial lung disease Identification of bronchiectasis/small airways disease Detection of pulmonary metastases from known extrathoracic malignancy

CHAPTER 11

should be adopted with the CT examination being tailored to the specific clinical question being asked, to avoid unnecessary radiation dose. The introduction of 16- and 64-detector MDCT systems has allowed the goal of truly isotropic imaging to be realized. Here, each image data element (voxel) is of equal dimensions in all three spatial axes, and forms the basis for image display in any arbitrarily chosen imaging plane. The acquisition of volumetric high-resolution data has particularly revolutionized the noninvasive assessment of vascular disease in the chest, and also paved the way for the further development of more sophisticated 3D image processing techniques. Many anatomical features of the chest do not conform to a single two-dimensional (2D) axial plane and full exploitation of isotropic MDCT data requires 2D and 3D post-processing techniques to harness the added advantage of improved z-axis resolution and coverage. Table 11.2 summarizes the main post-processing techniques used in evaluating chest disease.

Definition of spiral pitch An important parameter for characterizing helical CT is the pitch, which according to International Electrotechnical Commission specifications is defined as p = TF/W, where TF is the table feed per rotation and W is the total width of the

Table 11.2

• TECHNIQUES IN THORACIC IMAGING

collimated beam.15 With four sections at 1-mm collimation and a table feed of 6 mm per rotation, the pitch is p = 6/(4 × 1) = 1.5. This definition holds true for both single- and multidetector row CT systems. In the early days of four-detector CT, the term detector pitch was introduced, which accounted for the width of a single section in the denominator. For the sake of uniformity, the term detector pitch should no longer be used16.

Dose considerations Despite the undisputed clinical benefits of MDCT, there is the issue of increased radiation compared to single-detector CT to consider. In a CT X-ray tube, a small area on the anode plate emits X-rays that penetrate the patient and are registered by the detector. A collimator between the X-ray tube and the patient, the pre-patient collimator, is used to shape the beam and establish the dose profile. In general, the collimated dose profile is a trapezoid in the longitudinal direction. In the umbral region, X-rays emitted from the entire area of the focal spot fall on the detector; however, in the penumbral regions, only a part of the focal spot illuminates the detector—the pre-patient collimator blocking off other parts. With single-detector CT, the entire trapezoidal dose profile can contribute to the detector

POST-PROCESSING TECHNIQUES THAT MAY BE APPLIED IN EVALUATING CHEST PATHOLOGY

Post-processing technique

Technical considerations

Clinical applications

Multiplanar and curved multiplanar reconstructions (MPR and CMPR)

2D techniques that provide alternate viewing perspectives, usually with conventional window settings. Images are obtained by a reordering of the voxels into 1 voxel-thick tomographic sections, excluding those voxels outside the imaging plane

Evaluation of the large airways and pulmonary emboli, particularly for interpretative difficulties on axial sections either due to partial volume averaging or the inability to differentiate periarterial from endoluminal abnormalities

Maximum intensity projection (MIP)

A ray is cast through the CT data and only data that are above an assigned value are displayed, thus reducing all data in the line of the ray to a single plane. Sliding slabs of 5–10 mm are commonly used

Main use is in vascular imaging (Fig. 11.1) and in the evaluation of micronodular disease (more accurate identification of nodules versus vessels, and more precise characterization of nodule distribution)

Mininum intensity projection (MinIP)

Similar to MIP, but only data below an assigned value are displayed and thus it is best suited for showing areas of low density

May improve conspicuity of subtle density differences of lung parenchyma and therefore highlight regions of emphysema or air trapping

Shaded surface display (SSD)

This technique reformats data around a threshold that defines the interface of tissues. SSD does not reveal any internal detail

Evaluation of airway abnormalities

Volume rendering

Volume rendering is a unique form of 3D visualization. In this process a ray is projected through the dataset and a weighted representation of all the Hounsfield units encountered is displayed depending on their representation within the tissues. Voxels that are only partially filled with a density of interest are also included. The resultant images contain depth information whilst maintaining 3D spatial relationships

Used in angiographic examinations and also to evaluate large airway abnormalities (Fig. 11.2)

‘Virtual bronchoscopy’

Surface rendering and volume rendering are used to produce endoscopic simulations of the airway (Fig. 11.3)

Virtual endoscopic or perspective volume rendering images are not widely applied as they seldom give information that cannot be obtained by MPR. However, virtual CT bronchoscopy used in association with 3D techniques providing extraluminal information can provide additional information such as safe routes for tracheobronchial biopsy. Monitoring the position of airway stents is another potential application of this technique

189

190

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 11.1 Maximum intensity projection (MIP) in vascular imaging. MIP reconstructed using 5-mm slabs demonstrates a markedly tortuous thoracic aorta. The descending thoracic aorta is aneurysmal with extensive mural thrombus.

signal, and thus the relative dose utilization of a single-detector CT system can be close to 100%. With MDCT, only the plateau region of the dose profile is used to ensure an equal signal level for all detector elements. The penumbral region is then discarded, either by a post-patient collimator or by the intrinsic self-collimation of the MDCT, and represents ‘wasted’ dose. The relative contribution of the penumbral region decreases with increasing section width and with an increasing number of simultaneously acquired images (Fig. 11.4). Thus, the relative

dose utilization with four-section 1-mm collimation CT is 70% or less depending on the scanner type, whereas with 16-section CT systems, dose efficiency can be improved to 84%. The CT parameters that affect radiation dose include gantry geometry, tube current and voltage, acquisition modes, collimation, pitch and gantry rotation time. Reduction in tube current is the most practical means of reducing CT radiation dose. A 50% reduction in tube current can halve effective radiation dose17. Authors of several studies using MDCT have suggested that it is possible to reduce tube current markedly (to between 40 and 70 mAs) in chest examinations without affecting image quality18,19. On a 64-detector MDCT, the dose for a volumetric high-resolution (1-mm sections) acquisition of the thorax in a 70 kg adult can be as low as 3.6 mSv if parameters of 120 kVp and 90 mAs (pitch of 1) are used (E. Castellano-Smith, personal communication). In lung cancer screening examinations, tube current can be remarkably low and yet yield images of diagnostic quality. Itoh et al have shown that images obtained at an effective tube current of 20 mAs are of equal diagnostic utility to those obtained at 50 mAs for the detection of 6-mm simulated nodules20. Another recommendation comprises acquisition of the entire chest using a 1-mm collimation (MDCT) at 120 kVp and 10–40 mAs depending on the body habitus of the individual21. At a tube current of 10 mAs, the effective radiation dose is 0.27 mSv; equivalent to just five conventional PA chest radiographs. In the paediatric population, some institutions favour the use of 1 mAs kg−1 for imaging the thorax; an approach that significantly reduces radiation dose. Tube potential (peak voltage) determines the incident X-ray mean energy, and variation in tube potential causes a substantial change in CT radiation dose. The effect of tube voltage on image quality is complex, since it affects both image noise and

Figure 11.2 Volume rendering to evaluate large airway abnormalities. (A) The axial image of a patient with recurrent adenoid cystic carcinoma shows narrowing of the right main bronchus with abnormal soft tissue surrounding the right upper lobe bronchus. (B) The volume rendered 3D image enables the entire extent of the stenosis to be visualized on a single image.

CHAPTER 11

• TECHNIQUES IN THORACIC IMAGING

Figure 11.3 ‘Virtual bronchoscopy’ of the airway. (A) Axial CT and (B) virtual bronchoscopic rendition demonstrating a carcinoid tumour protruding into the lumen of the distal trachea just above the level of the carina.

tissue contrast.Thus, the image quality ramifications of a decrease in tube voltage to reduce radiation exposure must be carefully examined before being implemented. For chest examinations, 120 kVp is commonly used. In thin patients (< 50 kg) and in the paediatric population, 100 kVp is recommended; the use of 80 kVp has been found to be associated with unacceptable beam hardening even in the smallest of patients (Fig. 11.5)22. With helical CT systems, beam collimation, table speed and pitch are interlinked parameters that affect diagnostic image quality. Faster table speed for a given collimation, resulting in a higher pitch, is associated with a reduced radiation dose (if other data acquisition parameters, including tube current, are held constant) because of a shorter exposure time. However, this is not true for some multidetector systems that use an effective milliampere–second setting (defined as milliampere seconds divided by pitch). Here, the effective milliampere–second level is held constant (by automatic tube current adjustment) irrespective of pitch value, so that radiation dose does not vary as pitch is changed23. Caution should be exercised when extrapolating dose reduction strategies from single- to multi-detector CT systems.

4–slice collimator

CT

detector

16 – slice collimator

CT

Automatic tube current modulation is a technical innovation that can substantially reduce patient dose.There are two methods used currently with CT systems: z-axis modulation and angular (x- and y-axis) modulation. In z-axis modulation, tube current is adjusted to maintain a user-selected quantum noise level in the image data. z-axis modulation attempts to render all images with similar noise, independent of patient size and anatomy. In angular modulation, the tube current is adjusted to minimize X-rays in projections (angles) that have less importance for the reduction of overall image noise content. With this technique, the tube output is adapted to the patient geometry during each rotation to compensate for strongly varying X-ray attenuation in asymmetric body regions such as the shoulders. A recent investigation of CT imaging studies in children in whom angular modulation was used demonstrated a mean reduction of 22% in dose without loss of image quality24. Ultimately, the complexity of the interrelationships between the different CT parameters and dose requires a close collaboration between radiologists and medical physicists to ensure that the radiation burden to patients is as low as possible without diagnostic accuracy being compromised.

Figure 11.4 Dose profiles for 4- and 16detector MDCT. The relative contribution of the penumbral region (P), representing wasted dose, decreases with increasing number of simultaneously acquired sections.

detector

Umbra

P

Detectors

P

191

192

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

sibly a reduced volume. Typical injection parameters for fourdetector MDCT are 100–150 ml of 240–320 mg ml−1 of iodine injected at a rate of 3–4 ml s−1. For 64-detector MDCT, some institutions have experimented with 90–120 ml of 320–370 mg ml−1 of iodine injected at 3.5–5 ml s−1. A recent study using a four-detector system evaluated the influence of iodine flow concentration on vessel attenuation29. There was significantly better visualization of fourth-, fifth- and sixth-order pulmonary arteries using a protocol based on 90 ml of 400 mg ml−1 of iodine when compared with 120 ml of 300 mg ml−1. An injection rate of 4 ml s−1 was used in both groups. A specific application of contrast enhancement is in the differentiation between benign and malignant pulmonary nodules and the reader is referred to two papers for details of this protocol30,31.

Window settings

Figure 11.5 Image obtained at 80 kVp on a 64-detector MDCT in a 1-year old with cough. Right lower lobe collapse is identified but the image is unacceptably noisy, making evaluation of the lung parenchyma difficult.

Intravenous contrast medium enhancement The following section will cover the basic principles of intravenous enhancement as applied to the evaluation of pulmonary disease. Intravenous enhancement is used routinely for CT angiography of the thorax; the most common indications being the evaluation of the pulmonary arterial tree in suspected pulmonary embolism, the aorta and lung cancer staging studies. Intravenous enhancement is influenced by several factors: body size and cardiac output of the patient, the concentration and volume of contrast material, the rate and duration of the injection, the delay between the injection and the initiation of data acquisition, the duration of data acquisition and whether bolus tracking or a set delay is used. With single-detector CT, protocols are relatively straightforward. A volume of 100 ml of 150 mg ml−1 of iodine injected at a rate of 2.5 ml s−1 after a 25-s delay is recommended for general thoracic work25. Suggested protocols for evaluating the pulmonary arterial tree using single-detector CT use between 120 and 140 ml of 240–300 mg ml−1 of iodine injected at a rate of 3–4 ml s−1 26,27 with either a fixed delay of 20 s28 or the use of automated triggering mechanisms. With advances in CT technology, however, the way contrast medium is delivered has had to be rethought. A CT study acquired using a 16-detector system in < 10 s leaves little room for error, and imaging at peak enhancement requires not only precise timing but careful tailoring of the volume and rate of delivery of contrast medium. One dilemma for fast CT is the chance of contrast medium still being injected when data acquisition is complete. Protocols for CT angiography of the chest using MDCT are still being refined, but it is generally accepted that the faster acquisition times of MDCT require a faster rate of injection, a higher concentration of contrast medium and pos-

The density within each voxel is represented by a Hounsfield unit (HU) value. In the thorax these units encompass a wide range, from aerated lung (approximately −800 HU) to ribs (+700 HU). No single-window setting can depict this wide range of densities on a single image. For this reason, a thoracic CT examination requires viewing in at least two settings in order to demonstrate the lung parenchyma and the soft tissues of the mediastinum. Furthermore, it may be necessary to adjust the window settings to improve the demonstration of a particular structure or abnormality. Preferred window settings for thoracic CT vary between institutions, but some generalizations can be made. For the soft tissues of the mediastinum and chest wall a window width of 300– 500 HU and a centre of +40 HU are appropriate. For the lungs a wide window of approximately 1500 HU or more at a centre of approximately −600 HU is usually satisfactory. The window settings have a profound influence on the visibility and apparent size of normal and abnormal structures. The most accurate representation of an object appears to be achieved if the value of the window level is halfway between the density of the structure to be measured and the density of the surrounding tissue. For example, the diameter of a pulmonary nodule, measured on soft tissue settings appropriate for the mediastinum, will be grossly underestimated31. It is also important to remember that when inappropriate window settings are used, smaller structures (e.g. peripheral pulmonary vessels) are proportionately much more affected than larger structures.

HIGH-RESOLUTION COMPUTED TOMOGRAPHY For the majority of patients being investigated exclusively for suspected interstitial lung disease, interspaced (as opposed to volumetric) high-resolution CT (HRCT) remains an adequate examination and should be used for younger patients. This is because the dose of interspaced HRCT is considerably lower than a volumetric high-resolution acquisition. Even when techniques are optimized for dose, volumetric HRCT of the chest incurs a dose that is at least three times higher than interspaced HRCT, and in certain cases the dose

CHAPTER 11

increase can be up to 10-fold, particularly if a relatively high milliampere–second setting (170 mAs) is used32.The fundamental components of an HRCT technique include thin collimation, usually 1–2 mm, coupled with a high spatial frequency algorithm reconstruction. Thin collimation improves spatial resolution and consequently enhances the detection of key morphologic features in HRCT interpretation: thickened interlobular septa, ground-glass opacification, small nodules and abnormally thickened or dilated airways. Reducing the section thickness below 1 mm will not yield any significant further improvement in spatial resolution and at the same time will reduce the signal-to-noise ratio of the image. A sharp reconstruction algorithm reduces image smoothing and makes structures visibly sharper, although image noise becomes more obvious33. Images are usually obtained in the supine position from the apices to the lung bases at full inspiration and at 10- or 20-mm intervals. When early interstitial fibrosis is suspected, HRCT is often performed in the prone position to prevent confusion with the increased opacification often seen in the dependent posterobasal segments in the usual supine position (Fig. 11.6). However, there is no advantage in prone CT if there is obvious diffuse lung disease on a contemporary chest radiograph34. Prone CT is mandatory in the evaluation of asbestosis in which subtle parenchymal abnormalities are most frequently seen in the caudal parts of the posterobasal segments (Fig. 11.7)35. The necessity of expiratory CT sections is somewhat controversial. In most patients with clinically significant small airways disease, the mosaic attenuation pattern attributable to small airways disease is apparent on inspiratory images. However, images obtained at end-expiration can certainly accentuate the regional heterogeneity of lung density, thus revealing small or subtle areas of air trapping (Fig. 11.8)36,37.

• TECHNIQUES IN THORACIC IMAGING

ULTRASOUND The main advantages of chest ultrasonography are its bedside availability, absence of radiation and the ease of guided aspiration of pleural fluid and some solid tumours. Visualization of the chest wall requires a high frequency linear probe (5– 7.5 MHz), whereas pleural and pulmonary pathology is better detected with a sector or phased-array probe with a lower frequency (3.5 MHz). Most pleural fluid collections of clinical significance are readily identified on standard chest radiographs, but in the intensive care setting even small effusions may cause respiratory compromise and ultrasound (US) is an effective way of detecting and subsequently guiding aspiration. US is also valuable for identifying septations within loculated collections which may influence the choice of treatment38. With real-time US, the movement of the diaphragm may be observed and the reduced motion of paralysis may be of diagnostic value. US is also a quick and effective way of guiding percutaneous needle biopsy of peripheral lung, pleural or chest wall lesions39, but cannot be used if there is any aerated lung between the ultrasound probe and the lesion. An early study reported a potential use of US in diagnosing chest wall invasion in lung cancer staging (disruption of the pleural line which is normally seen as an echogenic interface), but this technique has not become widely used40.

Endoscopic ultrasonography Endoscopic ultrasonography (EUS) is a unique investigation in which a high-frequency US transducer is incorporated into the tip of an endoscope to provide high-resolution images of the gastrointestinal wall and structures in close proximity to the gastrointestinal tract. Linear echoendoscopes that can image parallel to the long axis of the instrument allow visualization

Figure 11.6 HRCT for suspected interstitial fibrosis. (A) Supine HRCT image of a patient being investigated for suspected interstitial lung disease reveals increased opacification in the posterior aspects of both lower lobes, which completely resolve with the patient prone (B).

193

194

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 11.7 HRCT for suspected asbestosis. (A) HRCT image in the supine position demonstrates fine reticulation and increased subpleural density (arrows). (B) These changes (arrows) persist on the prone image and may represent early asbestosis in this patient who had an appropriate asbestos exposure.

Figure 11.8 (A) Inspiratory and (B) end-expiratory HRCT images in a patient with known hypersensitivity pneumonitis. There is evidence of a subtle mosaic attenuation on the inspiratory image but this is made much more conspicuous on the end-expiratory image revealing spared secondary pulmonary lobules.

of a projecting needle, relative to adjacent tissue, making EUSguided aspiration or intervention possible. Transoesophageal EUS-guided real-time fine-needle aspiration (FNA) of mediastinal lymph nodes has become a useful, minimally invasive and safe method for staging the mediastinum41–43. In patients with non-small cell lung cancer (NSCLC) who have enlarged mediastinal nodes on CT, the accuracy of EUS-guided FNA is approximately 96%43. EUS-guided FNA is particularly suited for posterior mediastinal staging, with enlarged lymph nodes in the subcarina, aortopulmonary window, para-oesophageal area and para-aortic area. EUS can also play a significant role in identifying patients with unresectable (N3) NSCLC when adenopathy was not present on CT44. The importance of histological confirmation by EUS–FNA is emphasized as echo characteristics alone are not adequately sensitive to predict malignancy.

MAGNETIC RESONANCE IMAGING General technical considerations of magnetic resonance imaging (MRI) are outlined elsewhere and its specific applications in the chest are described where appropriate in the succeeding chapters. Accumulated evidence has demonstrated that MRI is no substitute for CT in the investigation of most thoracic conditions that require cross-sectional imaging; this is largely due to the relatively poor spatial resolution of MRI, the extremely low proton density of normal lung, the further decrease of signal by strong susceptibility artefacts induced by the multiple air–soft tissue interfaces within the lung, and the consequences of cardiac and respiratory movement. The advent of MDCT has allowed the acquisition of high-resolution thin sections and with this, the ability to produce multiplanar reconstructions. Consequently, MDCT is used for most aspects of thoracic

CHAPTER 11

imaging, including areas thought previously to be the domain of ‘problem-solving’ MRI. The main indications for MRI in the chest include the evaluation of the heart and great vessels, characterization of mediastinal lesions that are equivocal on CT, evaluation of superior sulcus tumours, particularly if brachial plexus involvement is suspected, and in demonstrating pulmonary embolism if radiation and intravenous contrast medium need to be avoided. This section will summarize the spectrum of MRI techniques that are used in thoracic imaging. Traditional imaging sequences have included T1-weighted spin echo (SE) with or without the use of gadolinium chelates (for the initial detection of abnormalities or the demonstration of anatomy) and T2-weighted fast spin echo (FSE) (for further characterization of abnormalities). Occasionally, a fatsaturation MRI technique (phase-shift gradient-echo imaging or proton-selective fat-saturation imaging) can be useful to detect fat and distinguish it from haemorrhage in the evaluation of mediastinal masses45. MRI is useful in confirming the cystic nature of mediastinal lesions that appear solid on CT (cysts containing nonserous fluid can have high attenuation on CT) as these cysts will have characteristically high signal intensity when imaged with T2-weighted sequences regardless of the nature of the cyst contents. To overcome the problem of respiratory motion, other sequences (fast low-angle shot [FLASH] and half Fourier turbo-spin echo [HASTE]) have been developed that can be acquired in one breath-hold with acquisition times well below 30 s. Additional techniques to compensate for respiratory motion in non-breath-hold MRI have also been evaluated. These include navigator techniques with registration of the diaphragm, reordering of phase encoding (ROPE) and phase encoding and reordering (PEAR). Recently developed 3D gradient-recalled echo (GRE) sequences for volumetric interpolated breath-hold imaging of the lung may introduce new capabilities for MRI of lung morphology with high spatial resolution46. Early studies indicate that images obtained with this technique provide good visualization of lung anatomy with a low prevalence of artefacts47. The introduction of new lymph node-specific contrast agents used in conjunction with MRI is an interesting development. Ultrasmall superparamagnetic iron oxide nanoparticles traverse the vascular endothelium and are phagocytosed by macrophages in normally functioning lymph nodes. This results in a uniform signal loss in T2- and T2*-weighted images, a feature that was first demonstrated in animal models48. Preliminary data in patients with bronchogenic carcinoma have shown a good sensitivity, but a relatively low specificity for the diagnosis of metastatic normal sized lymph nodes49. Currently, the most important clinical application of MRI is in the imaging of the heart and great vessels, and the specific techniques required for this are dealt with in detail in Chapter 22. In summary, pulse sequences used for cardiac and great vessel imaging can generally be divided into dark or black blood, and bright or white blood imaging techniques. In black blood imaging techniques, rapidly flowing blood is black or of low signal intensity. Examples of this technique include conven-

• TECHNIQUES IN THORACIC IMAGING

tional spin-echo, breath-hold FSE or turbo spin-echo (TSE) variants with double inversion recovery pulses to suppress blood signal (HASTE, double IR TSE–FSE).These techniques are typically used for anatomical delineation of the heart, pericardium, mediastinum and great vessels. In bright blood techniques flowing blood is white or of high signal intensity.These are usually GRE sequences (Fig. 11.9). Cine GRE sequences that produce a motion picture loop throughout the cardiac cycle are used for assessment of cardiac function (and can be used to assess relative motion of adjacent structures in the context of masses abutting vessels or the heart). A single slice multiphase or multislice single-phase acquisition can be performed in a short breath-hold period. Examples of this technique include fast low angle shot (TurboFLASH), fast spoiled

Figure 11.9 (A) Black blood T1 spin-echo image and (B) white blood GRE cine image of the heart. Spin-echo imaging is particularly useful for high-definition anatomical imaging of the heart and vessels, whereas the acquisition of cine loops during a GRE sequence is useful for the assessment of valves and regional contractile function of the myocardium. (Courtesy of Dr Sanjay K. Prasad.)

195

196

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

GRE, turbo field echo and fast field echo. More recently, faster short TE–GRE sequences with completely refocused gradients have been used to provide excellent contrast between myocardium and blood pool (true FISP, balanced fast field echo, FIESTA). For imaging the intrathoracic vessels, contrast-enhanced MR angiography (MRA) is performed using short, breathhold, 3D T1-weighted GRE sequences (Fig. 11.10). Using the T1-shortening effects of gadolinium-based contrast agents, blood appears bright regardless of flow patterns or velocity. Synchronization of image acquisition and arrival of the contrast bolus is crucial to obtain high image quality. Newer sequences that allow near real-time assessment of a gadolinium contrast bolus are now available, and may be useful for assessment of shunts and fistulas. There is also continued interest in the role of MRA for the diagnosis of pulmonary embolism, either by direct demonstration of the intravascular thrombus50,51 or by decreased signal areas representing underperfused lung on gadolinium-enhanced MRI52. Although still largely a research tool, the introduction of hyperpolarized noble gas imaging using either 3He or 129Xe has been used to demonstrate ventilated parts of the lung53 and

Figure 11.10 3D Gadolinium-enhanced MRA of a patient with aortic dissection. The intimal flap (arrows) is clearly delineated. (Courtesy of Dr Sanjay K. Prasad.)

thus enabled the evaluation of structure–function relationships in lung disease54. Diffusion-sensitive MRI techniques allow mapping of the ‘apparent diffusion coefficient’ (ADC) of 3He within lung spaces, where ADC is physically related to local bronchoalveolar dimensions55. ADC values are increased in fibrosis and emphysema, and show good agreement with predicted lung function56; however, these sophisticated techniques are not widely available.

VENTILATION–PERFUSION SCINTIGRAPHY Ventilation–perfusion (V/Q) scintigraphy is a noninvasive technique for the assessment of the distribution of pulmonary blood flow and alveolar ventilation and has primarily been used for the diagnosis of pulmonary embolism. Perfusion scintigraphy is performed following the intravenous injection of 99m Tc-labelled protein microparticles which, because of their size, undergo micro-embolization in the pulmonary vascular bed. The number of particles injected may range from a minimum of 60 000 to a maximum of 700 000, with the recommended average being 200 000. Because of the theoretical possibility of embolization of a medium-sized systemic vessel, microspheres should be used rather than larger macroaggregates in the presence of a known right-to-left shunt. Krypton-81m is in many ways the ideal agent of choice for ventilation imaging but it has a very short half-life and is expensive to produce. Unlike longer-lived radioactive gases, such as 133Xe, 81mKr does not accumulate progressively in regions of lung with a low ventilatory turnover. The lung can be imaged in multiple projections and in each projection, perfusion and ventilation images can be acquired sequentially, or, with the newer digital cameras, simultaneously. Another commonly used agent for ventilation imaging is 133Xe (half-life of 5.3 d). 133Xe has three advantages over 81mKr: it is cheaper, is readily available in comparison to 81mKr which is available from its generator for 1 d only, and it can be used to detect air trapping. Since the energy of 133Xe is less than that of 99mTc, the ventilation images must be acquired before the 99mTc injection. Furthermore, only one projection is available and, with the lower energy of xenon, the images are of poorer resolution. Other agents include 99mTc-diethylenetriaminepentaacetic or 99mTc-technegas. Technegas is an ultrafine aerosol that is considered to behave truly like a gas due to the mean aerodynamic diameter of the particles being between 30 and 90 nm. The small particle size results in efficiency values of up to 20% (efficiency being defined as the ratio between the amount of applied activity and its actual pulmonary deposition) compared to conventional aerosols which show a degree of efficiency between 1 and 3%. Lung scintigraphy remains part of the diagnostic algorithm in the investigation of patients with pulmonary embolism, and guidelines suggest that it may be considered, subject to its availability, as the initial imaging investigation provided the chest radiograph is normal and there is no significant symptomatic concurrent cardiopulmonary disease57. A recent study evaluating ventilation–perfusion lung scintigraphy performed using SPECT technique (as opposed to

CHAPTER 11

merely using planar acquisitions) has shown that diagnostic accuracy is comparable with MDCT (four-detector system with 1.25-mm effective slice thickness)58. Analogous to developments in the fields of neurology and cardiology, it can be expected that SPECT will replace planar lung scintigraphy in the near future, but lack of availability will mean that this technique is unlikely to be frequently used for patients with suspected pulmonary embolism.

POSITRON EMISSION TOMOGRAPHY Positron emission tomography (PET) was initially used as a research tool for functional studies of the brain and the assessment of cardiac metabolism, but in the past 8 years most of its indications have been in the ambit of assessment of patients with suspected malignancy, particularly those being considered for surgery59. The different applications relevant to lung cancer are listed in Table 11.3, but a discussion of the advantages of PET in the staging of lung cancer is given in Chapter 18. The use of [F-18]fluoro-2-deoxy-glucose (FDG)-PET in oncology is based on its ability to identify the differences between glucose metabolism in various tissues. Neoplastic cells have a much higher rate of glycolysis that non-neoplastic cells. [F 18]FDG, a glucose analogue in which the oxygen molecule in position 2 is replaced by the positron-emitting 18 F, undergoes the same uptake as glucose, but is metabolically trapped and accumulates in neoplastic cells after phosphorylation by hexokinase. To gain a quantitative measure of FDG uptake, attenuationcorrected images are required; this is because the intensity of the photon emission of a lesion is position dependent, and therefore the intensity seen on the nonattenuation corrected whole-body images does not truly reflect the actual FDG uptake. If the images are corrected for photo-attenuation by a ‘transmission’ study which estimates the attenuating characteristics of the patient, quantification of FDG metabolism becomes possible. The use of a transmission study allows the

Table 11.3 INDICATIONS FOR [F-18]FLUORO-2DEOXYGLUCOSE PET IN RESPIRATORY ONCOLOGY Common clinical indications Evaluation of nodules and masses

• TECHNIQUES IN THORACIC IMAGING

Standardized Uptake Value (SUV) to be reported.The SUV of a lesion is a semiquantitative index of glucose utilization that is obtained by normalizing the accumulation of FDG in the lesion to the injected dose and patient body weight60.The criterion for a positive result is either greater uptake in the lesion than in the background mediastinum or a SUV of > 2.561. The most important role of FDG–PET is in the clarification of the nature of an incidental nodule or mass identified on an ‘anatomical’ study62. Based on several prospective studies, FDG–PET has proven to be accurate in differentiating benign from malignant lesions as small as 1 cm with an overall sensitivity of 96% (range 83–100%) and specificity of 79% (range 52–100%). Potential pitfalls in sensitivity are due to the fact that a critical mass of metabolically active malignant cells is required for PET diagnosis; false-negative findings can occur in lesions < 1 cm63 and in lesions with low metabolic activity, e.g. carcinoid tumours64 and bronchioloalveolar cell carcinomas65. Errors in specificity are due to FDG uptake in inflammatory conditions, such as bacterial pneumonia66 and particularly granulomatous diseases such as sarcoidosis67, tuberculosis, or Wegener’s granulomatosis. A further emerging use of PET is in the staging of nodal metastases with studies consistently showing significantly greater accuracy of PET compared with CT for the detection or exclusion of mediastinal nodal disease68. The use of PET will undoubtedly increase as the technique becomes more available and a routine part of the investigative algorithm of patients with suspected lung cancer. Table 11.4 summarizes the indications for PET in the diagnosis and staging of patients with lung cancer. PET may well have a role in the evaluation of other intrathoracic malignancies, particularly in the assessment of response to chemotherapy that may occur before there is any discernible morphological change in the tumour on conventional imaging. Finally, a whole new field—applying PET to molecular biology using new radiopharmaceutical probes—is under investigation. These techniques may allow the evaluation of molecular-targeted lung cancer therapies or even gene therapy.

POSITRON EMISSION TOMOGRAPHY– COMPUTED TOMOGRAPHY PET/CT imaging was introduced in 1998.The concept of combined PET/CT imaging is to supplement metabolic information

Locoregional staging Extrathoracic staging Applications under investigation Radiotherapy planning Evaluation of response post radiotherapy Evaluation of response post (induction) chemotherapy Follow-up and diagnosis of recurrence Molecular applications Early assessment of chemotherapy Assessment of molecular targeted therapy

Table 11.4 INDICATIONS FOR PET IN THE STAGING OF LUNG CANCER All patients who are staged on CT as candidates for surgery: to identify involved intrathoracic lymph nodes and distant metastases Patients who are otherwise surgical candidates but have, on CT, limited (one to two stations) N2/N3 disease of uncertain pathological significance All patients who are candidates for radical radiotherapy To investigate solitary pulmonary nodules in cases in which biopsy is not possible or has failed, depending on nodule size

197

198

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

from a whole-body PET study with more detailed anatomical information, thereby improving diagnostic accuracy.

Image acquisition Conventional PET employs transmission images for photon attenuation correction using an external radiation source and consequently, a conventional whole-body PET study covering six to eight bed positions requires about 1 h for completion. PET/CT imaging differs in that it utilizes whole-body CT data for attenuation correction. Depending on the number of CT detectors used, attenuation correction is achieved within seconds to slightly over 1 min and thus whole-body imaging times are reduced by 50%. The incremental value of PET/CT over PET alone for staging and restaging of cancer has not been fully established, but preliminary data suggest significant increments in diagnostic and staging accuracy of NSCLC69,70. The latter study demonstrated that tumour staging and nodal staging were significantly more accurate with integrated PET/CT than with PET alone. Moreover, PET/CT provided additional information in 41% of patients, including localization of lymph nodes (n = 9), precise identification of chest wall infiltration (n = 3), correct differentiation between tumour and inflammation (n = 7) and localization of distant metastases (n = 2)70. However, it is important to note that this study did not examine prospectively whether the ‘additional information’ led to significant changes in patient management. More clinical trials with greater patient numbers will be required to firmly establish possible advantages of PET/CT over PET or CT alone for each type of cancer.

REFERENCES 1. Rong X J, Shaw C C, Liu X et al 2000 Comparison of an amorphous silicon/cesium iodide flat-panel digital chest radiography system with screen/film and computed radiography systems—a contrast-detail phantom study. Med Phys 28: 2328–2335 2. Garmer M, Hennigs S P, Jager H J et al 2000 Digital radiography versus conventional radiography in chest imaging: diagnostic performance of a large-area silicon flat-panel detector in a clinical CT-controlled study. Am J Roentgenol 174: 75–80 3. Floyd C E J, Warp R J, Dobbins J T et al 2001 Imaging characteristics of an amorphous silicon flat-panel detector for digital chest radiography. Radiology 218: 683–688 4. Fink C, Hallscheidt P J, Noeldge G et al 2002 Clinical comparative study with a large-area amorphous silicon flat-panel detector: image quality and visibility of anatomic structures on chest radiography. Am J Roentgenol 178: 481–486 5. Bacher K, Smeets P, Bonnarens K et al 2003 Dose reduction in patients undergoing chest imaging: digital amorphous silicon flat-panel detector radiography versus conventional film-screen radiography and phosphorbased computed radiography. Am J Roentgenol 181: 923–929 6. American Thoracic Society 2001 Guidelines for the management of adults with community-acquired pneumonia. Am J Respir Crit Care Med 163: 1730–1754 7. Metersky M L 2005 Is the lateral decubitus radiograph necessary for the management of a parapneumonic pleural effusion? Chest 124: 1129–1132 8. Klingenbeck-Regn K, Schaller S, Flohr T et al 1999 Subsecond multislice computed tomography: basics and applications. Eur J Radiol 31: 110–124

9. Hu H, He H D, Foley W D, et al 2000 Four multidetector-row helical CT: image quality and volume coverage speed. Radiology 215: 55–62 10. McCollough C H, Zink F E 1999 Performance evaluation of a multi-slice CT system. Med Phys 26: 2223–2230 11. Pappas J N, Donnelly L F, Frush DP 2000 Reduced frequency of sedation of young children with multisection helical CT. Radiology 215: 897–899 12. Wormanns D, Kohl G, Klotz E et al 2004 Volumetric measurements of pulmonary nodules at multi-row detector CT: in vivo reproducibility. Eur Radiol 14: 86–92 13. Yi C A, Lee K S, Kim E A et al 2004 Solitary pulmonary nodules: dynamic enhanced multi-detector row CT study and comparison with vascular endothelial growth factor and microvessel density. Radiology 233: 191–199 14. Revel M P, Petrover D, Hernigou A et al 2005 Diagnosing pulmonary embolism with four-detector row helical CT: prospective evaluation of 216 outpatients and inpatients. Radiology 234: 265–273 15. International Electrotechnical Commission 2002 60601-2-44 Amendment 1. Medical Electrical Equipment, part 2-44: Particular requirements for the safety of x-ray equipment for computed tomography. International Electrotechnical Commission, Geneva 16. Flohr T G, Schaller S, Stierstorfer K et al 2005 Multi-detector row CT systems and image-reconstruction techniques. Radiology 235: 756–773 17. Kalra K K, Maher M M, Toth T L et al 2004 Strategies for CT radiation dose optimization. Radiology 230: 619–628 18. Jung K J, Lee K S, Kim S Y et al 2000 Low-dose, volumetric helical CT: image quality, radiation dose, and usefulness for evaluation of bronchiectasis. Invest Radiol 35: 557–563 19. Yi C A, Lee K S, Kim T S et al 2003 Multidetector CT of bronchiectasis: effect of radiation dose on image quality. Am J Roentgenol 181: 501–505 20. Itoh S, Ikeda M, Arahata S et al 2000 Lung cancer screening: minimum tube current required for helical CT. Radiology 215: 175–183 21. Schoepf U J, Becker C R, Obuchowski N A et al 2001 Multi-slice computed tomography as a screening tool for colon cancer, lung cancer and coronary artery disease. Eur Radiol 11: 1975–1985 22. Cody D D, Moxley D M, Krugh K T et al 2004 Strategies for formulating appropriate MDCT techniques when imaging the chest, abdomen, and pelvis in pediatric patients. Am J Roentgenol 182: 849–859 23. Mahesh M, Scatarige J C, Cooper J et al 2001 Dose and pitch relationship for a particular multislice CT scanner. Am J Roentgenol 177: 1273–1275 24. Greess H, Nomayr A, Wolf H et al 2002 Dose reduction in CT examination of children by an attenuation-based on-line modulation of tube current (CARE Dose). Eur Radiol 12: 1571–1576 25. Leung A N 1997 Spiral CT of the thorax in daily practice: optimization of technique. J Thorac Imag 12: 2–10 26. Remy-Jardin M, Remy J, Artaud D et al 1997 Peripheral pulmonary arteries: optimization of the spiral CT acquisition protocol. Radiology 204: 157–163 27. Kuzo R S, Goodman L R 1997 CT evaluation of pulmonary embolism: technique and interpretation. Am J Roentgenol 169: 959–965 28. Hartmann I J, Lo R T, Bakker J et al 2002 Optimal scan delay in spiral CT for the diagnosis of acute pulmonary embolism. J Comput Assist Tomogr 26: 21–25 29. Schoellnast H, Deutschmann H A, Fritz G A et al 2005 MDCT angiography of the pulmonary arteries: Influence of iodine flow concentration on vessel attenuation and visualization. Am J Roentgenol 184: 1935–1939 30. Swensen S J, Viggiano R W, Midthun D E et al 2000 Lung nodule enhancement at CT: multicenter study. Radiology 214: 73–80 31. Harris K M, Adams H, Lloyd D C F et al 1993 The effect on apparent size of simulated pulmonary nodules of using three standard CT window settings. Clin Radiol 47: 241–244 32. Kelly D M, Hasegawa I, Borders R et al 2004 High-resolution CT using MDCT: Comparison of degree of motion artefact between volumetric and axial methods. Am J Roentgenol 182: 757–759 33. Murata K, Khan A, Rojas K A et al 1988 Optimization of computed tomography technique to demonstrate the fine structure of the lung. Invest Radiol 23: 170–175

CHAPTER 11

34. Volpe J, Storto M L, Lee K et al 1997 High-resolution CT of the lung: determination of the usefulness of CT scans obtained with the patient prone based on plain radiographic findings. Am J Roentgenol 169: 369–374 35. Aberle D R, Gamsu G, Ray C S 1988 High-resolution CT of benign asbestos-related diseases: Clinical and radiographic correlation. Am J Roentgenol 151: 883–891 36. Arakawa H, Webb W R, McCowin M et al 1998 Inhomogeneous lung attenuation at thin-section CT: diagnostic value of expiratory scans. Radiology 206: 89–94 37. Lucidarme O, Coche E, Cluzel P et al 1998 Expiratory CT scans for chronic airway disease: correlation with pulmonary function test results. Am J Roentgenol 170: 301–307 38. Lomas D J, Padley S P G, Flower C D R 1993 The sonographic appearances of pleural fluid. Br J Radiol 66: 619–624 39. Yang P C 1997 Ultrasound-guided transthoracic biopsy of peripheral lung, pleural, and chest-wall lesions. J Thorac Imag 12: 272–284 40. Suzuki N, Saitoh T, Kitamura S 1993 Tumour invasion of the chest wall in lung cancer: diagnosis with US. Radiology 187: 39–42 41. Fritscher-Ravens A, Soehendra N, Schirrow L et al 2000 Role of transesophageal endosonography-guided fine-needle aspiration in the diagnosis of lung cancer. Chest 117: 339–345 42. Bhutani M S, Hawes R H, Hoffman B J 1997 A comparison of the accuracy of echo features during endoscopic ultrasound (EUS) and EUSguided fine-needle aspiration for diagnosis of malignant lymph node invasion. Gastrointest Endosc 45: 474–479 43. Gress F G, Savides T J, Sandler A et al 1997 Endoscopic ultrasonography, fine-needle aspiration biopsy guided by endoscopic ultrasonography, and computed tomography in the preoperative staging of non-smallcell lung cancer: a comparison study. Ann Intern Med 127: 604–612 44. LeBlanc J, Devereaux B M, Imperiale T F et al 2005 Endoscopic ultrasound in non-small cell lung cancer and negative mediastinum on computed tomography. Am J Respir Crit Care Med 171: 177–182 45. Outwater E K, Siegelman E S, Hunt J L 2001 Ovarian teratomas: tumor types and imaging characteristics. RadioGraphics 21: 475–490 46. Semelka R C, Cem B N, Wilber K P et al 2000 Breath-hold 3D gradientecho MR imaging of the lung parenchyma: evaluation of reproducibility of image quality in normals and preliminary observations in patients with disease. J Magn Reson Imaging 11: 195–200 47. Biederer J, Both M, Graessner J et al 2003 Lung morphology: fast MR imaging assessment with a volumetric interpolated breath-hold technique: initial experience with patients. Radiology 226: 242–249 48. Vassallo P, Matei C, Heston W D et al 1994 AMI-227-enhanced MR lymphography: usefulness for differentiating reactive from tumorbearing lymph nodes. Radiology 193: 501–506 49. Pannu H K, Wang K P, Borman T L et al 2000 MR imaging of mediastinal lymph nodes: evaluation using a superparamagnetic contrast agent. J Magn Reson Imag 12: 899–904 50. Meaney J F M, Weg J G, Chenevert T L et al 1997 Diagnosis of pulmonary embolism with magnetic resonance angiography. N Engl J Med 336: 1422–1427 51. Gupta A, Frazer C K, Ferguson J M et al 1999 Acute pulmonary embolism: diagnosis with MR angiography. Radiology 210: 353–359 52. Amundsen T, Kvaerness J, Jones R A et al 1997 Pulmonary embolism: detection with MR perfusion imaging of lung—a feasibility study. Radiology 203: 181–185 53. Kauczor H U, Hofmann D, Kreitner K F et al 1996 Normal and abnormal pulmonary ventilation: visualization at hyperpolarized He-3 MR imaging. Radiology 201: 564–568 54. Eberle B, Markstaller K, Schreiber W G et al 2001 Hyperpolarised gases in magnetic resonance: a new tool for functional imaging of the lung. Swiss Med Wkly 131: 503–599 55. Hanisch G, Schreiber W, Diergarten T et al 2000 Investigation of intrapulmonary diffusion by 3He MRI. Eur Radiol 10 (suppl 1): S345

• TECHNIQUES IN THORACIC IMAGING

56. Kauczor H 2003 Hyperpolarized helium-3 gas magnetic resonance imaging of the lung. Top Magn Reson Imaging 14: 223–230 57. British Thoracic Society Standards of Care Committee Pulmonary Embolism Guideline Development Group. British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 58: 470–483 58. Reinartz P, Wildberger J E, Schaefer W et al 2004 Tomographic imaging in the diagnosis of pulmonary embolism: a comparison between V/Q lung scintigraphy in SPECT technique and multislice spiral CT. J Nucl Med 45: 1501–1508 59. Vansteenkiste J F 2002 Imaging in lung cancer: positron emission tomography scan. Eur Respir J 19 (suppl 35): 49s–60s 60. Meikle S R, Bailey D L, Hooper P K et al 1995 Simultaneous emission and transmission measurements for attenuation correction in wholebody PET. J Nucl Med 36: 1680–1688 61. Patz E F, Lowe V J, Hoffman J M et al 1993 Focal pulmonary abnormalities: evaluation with F-18 fluorodeoxyglucose PET scanning. Radiology 188: 487–490 62. Vansteenkiste J F, Stroobants S G, De Leyn P R et al 1997 Mediastinal lymph node staging with FDG-PET scan in patients with potentially operable non-small cell lung cancer: a prospective analysis of 50 cases. Leuven Lung Cancer Group. Chest 112: 1480–1486 63. Dewan N A, Gupta N C, Redepenning L S et al 1993 Diagnositc efficacy of PET-FDG imaging in solitary pulmonary nodules. Potential role in evaluation and management. Chest 104: 997–1002 64. Erasmus J J, McAdams H P, Patz E F Jr et al 1998 Evaluation of primary pulmonary carcinoid tumors using FDG PET. Am J Roentgenol 170: 1369–1373 65. Kim B T, Kim Y, Lee K S et al 1998 Localized form of bronchioloalveolar carcinoma. FDG PET findings. Am J Roentgenol 170: 935–939 66. Kapucu L O, Meltzer C C, Townsend D W et al 1998 Fluorine-18fluorodeoxyglucose uptake in pneumonia. J Nucl Med 39: 1267–1269 67. Brudin L H, Valind S O, Rhodes C G et al 1994 Fluorine-18 deoxyglucose uptake in sarcoidosis measured with positron emission tomography. Eur J Nucl Med 21: 297–305 68. Steinert H C, Hauser M, Allemann F et al 1997 Non-small cell lung cancer: nodal staging with FDG PET versus CT with correlative lymph node mapping and sampling. Radiology 202: 441–446 69. Cerfolio R J, Ojha B, Bryant A S et al 2004 The accuracy of integrated PET-CT compared with dedicated PET alone for the staging of patients with nonsmall cell lung cancer. Ann Thorac Surg 78: 1017–1023 70. Lardinois D, Weder W, Hany T F et al 2003 Staging of non-smallcell lung cancer with integrated positron-emission tomography and computed tomography. N Engl J Med 348: 2500–2507

SUGGESTED FURTHER READING Flohr T G, Schaller S, Stierstorfer K et al 2005 Multi-detector row CT systems and image-reconstruction techniques. Radiology 235: 756–773 Kalra M K, Maher M M, Toth T L et al 2004 Strategies for CT radiation dose optimization. Radiology 230: 619–628 Kauczor H U, Chen X J, Beek E R J et al 2001 Pulmonary ventilation imaged by magnetic resonance: at the doorstep of clinical application. Eur Respir J 17: 1008–1023 Müller N L 2002 Computed tomography and magnetic resonance imaging: past, present and future. Eur Respir J 19 (suppl 35): 3s–12s Remy-Jardin M, Remy J, Mayo J R et al 2001 CT angiography of the chest. Lippincott, Williams and Wilkins, Philadelphia Schoepf U J (ed) 2004 Multidetector-row CT of the thorax. Springer Verlag, Berlin Schrevens L, Lorent N, Dooms C et al 2004 The role of PET scan in diagnosis, staging and management of non-small cell lung cancer. Oncologist 9: 633–643

199

CHAPTER

The Normal Chest

12

Simon Padley and Sharyn L. S. MacDonald

• • • • • •

The lungs The central airways The lungs beyond the hila The hila The mediastinum The diaphragm

THE LUNGS Each lung is divided into lobes surrounded by pleura. There are two lobes on the left: the upper and lower, separated by the major (oblique) fissure; and three on the right: the upper, middle and lower lobes separated by the major (oblique) and minor (horizontal) fissures. The fissures are frequently incomplete, particularly medially, containing localized defects which form an alveolar pathway for collateral air drift and the spread of disease. For a fissure to be visualized on conventional radiographs, the X-ray beam has to be tangential to the fissure. In most people, some or all of the minor fissure is seen in the frontal projection, but neither major fissure can be identified. In the lateral view, both the major and minor fissures are often identified, but usually only part of any fissure is seen; in fact, it is very unusual to see both left and right major fissures in their entirety. The major fissures have similar anatomy on the two sides. They run obliquely anteriorly and inferiorly from approximately the fifth thoracic vertebra to pass through the hilum and contact the diaphragm 0–3 cm behind the anterior costophrenic angle. Each major fissure follows a gently curving plane somewhat similar to a propeller blade (Fig. 12.1), with the upper portion facing anteriorly and laterally, and the lower portion facing forward and medially. Owing to the undulating course of the major fissure, either fissure may be seen as two lines on the lateral view. Consequently it may appear to the unwary that a fissure is displaced when it is in fact in its normal position, or both fissures may appear to be in their normal positions when in reality one of them is so displaced that it is no longer visible.

The inferior few centimetres of either or both major fissures may be widened owing to the presence of fat or pleural thickening between the leaves of the pleura. In these circumstances the contact with the diaphragm will often be broadened and lead to a localized loss of silhouette, an appearance referred to as the juxtaphrenic peak. With modern multislice computed tomography (CT), the normal major fissures are frequently visible, but if not clearly defined the position can be inferred from the presence of a relatively avascular zone that forms the outer cortex of the lobe. With high-resolution CT (HRCT), a normal major fissure is seen as a thin line traversing the avascular zone1, although it may be represented as two parallel lines on at least one level in approximately one-third of the population because of an artefact related to cardiac and respiratory motion2. The minor fissure fans out anteriorly and laterally from the right hilum in a horizontal direction to reach the chest wall. On a standard chest radiograph, the minor fissure contacts the chest wall at the axillary portion of the right sixth rib. The fissure curves gently, with its anterior and lateral portion usually curving downwards. Because of the curvature of the major fissure described above, part of the minor fissure may be projected posterior to the right major fissure on the lateral view. On CT the minor fissure position is represented by an oval area of reduced vascularity at the level of the bronchus intermedius (Fig. 12.2). The normal minor fissure is not seen as a line on axial CT imaging but is apparent on multiplanar reformats. In 1% of the population an accessory fissure3 called the ‘azygos lobe fissure’ (Fig. 12.3) is seen. This fissure contains the azygos vein at its lower end and results from failure of normal migration of the azygos vein from the chest wall to its usual position in the tracheobronchial angle and persistence of the invaginated visceral and parietal pleurae.There is no corresponding alteration in the segmental architecture of the lung, so the term ‘lobe’ is a misnomer. The ‘azygos lobe’ may, however, be smaller and therefore less transradiant than corresponding normal lung4. On CT the altered course of the azygos vein can be seen traversing the lung (Fig.12.3B).

202

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 12.1 The different position and shape of the major fissures (arrows) in the lower and the upper zones is well shown by CT. Note that (A) below the hila, the major fissures bow forward, whereas (B) above the hila, the major fissures bow backward. (The images are high-resolution thin-section [1.5 mm] CT scans.) Sagittal reformats from the (C) right and (D) left lungs demonstrate the detail of fissure anatomy available on this 16-channel CT study.

Figure 12.2 Minor fissure on CT. (A) The minor fissure is apparent as an area of avascularity anterior to the major fissure. In this example the slightly bowed horizontal fissure undulates through the plane of the slice. (B) The position of the minor fissure, in another patient, is indicated by the oval deficiency of vessels in the right mid zone (arrows).

CHAPTER 12

• THE NORMAL CHEST

Figure 12.3 (A) Azygos lobe fissure (curved arrows) and aortic nipple (horizontal straight arrow). The azygos vein in the lower end of the fissure is well seen (lower curved arrow). Note its absence from its usual location in the right tracheobronchial angle (open arrow). The aortic nipple, due to the left superior intercostal vein, is particularly large in this example. (B) CT of the azygos lobe fissure.

Other accessory fissures are occasionally identified3. A minor fissure may separate the lingular segments from the remainder of the upper lobe, similar to the right minor fissure. A horizontally orientated fissure, a superior accessory fissure, may separate the apical segment from the basal segments of either lower lobe. An inferior accessory fissure is sometimes seen in one or other lower lobe, usually the right, separating the medial and anterior basal segments. This fissure runs obliquely upward and medially towards the hilum from the diaphragm. The inferior pulmonary ligaments5 are pleural reflections from the mediastinum which hang down from the hila and are analogous in shape to the peritoneal reflections forming the broad ligament of the uterus. These two layers of pleura may extend down to the diaphragm or may have a free inferior edge. The intersegmental septum of the lower lobe, a septum within the lung immediately beneath the inferior pulmonary ligament, is often visible on CT (Fig. 12.4)6.When the inferior pulmonary ligament reaches the diaphragm it may contain a small amount of fat. This may efface the diaphragm, resulting in a juxtaphrenic peak. Otherwise neither the intersegmental septum nor the inferior pulmonary ligament is visible on plain radiographs.

THE CENTRAL AIRWAYS The trachea is a straight tube that, in children and young adults, passes inferiorly and posteriorly in the midline. In subjects

Figure 12.4 Intersegmental septum deep to the inferior pulmonary ligament shown by CT (arrows).

with unfolding and ectasia of the aorta the trachea may deviate to the right and may also bow forward. In cross-section the trachea is usually round, oval, or oval with a flattened posterior margin. Maximum coronal and sagittal diameters in adults on plain chest radiography are 21 and 23 mm, respectively, for women, and 25 and 27 mm for men7. On CT, which allows precise assessment of diameters and cross-sectional areas without magnification, the mean transverse diameter is 15.2 mm (sd 1.4) for women and 18.2 mm (sd 1.2) for men,

203

204

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

the lower limit of normal being 12.3 mm for women, and 15.9 mm for men8. The diameters in growing children and young adults have been documented9. Calcification of the cartilage rings of the trachea is a common normal finding after the age of 40 years, increasing in frequency with age. The trachea divides into the two mainstem bronchi at the carina. In children the angles are symmetrical, but in adults the right mainstem bronchus has a steeper angle than the left. The range of angles is wide, and alterations in angle can be diagnosed only by right–left comparisons, not by absolute measurement. The left main bronchus extends up to twice as far as the right main bronchus before giving off its upper lobe division. The lobar and segmental branching pattern is shown in Figure 12.5. There are many variations of the segmental and subsegmental branches10,11. Airways to subsegmental level can be routinely identified on volumetric thin-collimation CT.

THE LUNGS BEYOND THE HILA The usual method of deciding normal lung density in the frontal view is by comparison with equivalent areas on the opposite side. Since this is not possible on the lateral chest radiograph, the detection of subtle densities is more difficult, but the density over the spine should decrease gradually as the eye travels down the spine until the diaphragm is reached. Certain other comparisons can be made, but are less reliable: the density of the high retrosternal areas is approximately equal to that of the area immediately posterior to the left ventricle; the density over the heart is usually similar to that over the shoulders; and, apart from the cardiac fat pads and overlying ribs, there should be no abrupt change in density over the heart shadow. The segmental bronchi divide into smaller and smaller divisions until after 6–20 divisions they become bronchioles and

Figure 12.5 Diagram illustrating the anatomy of the main bronchi and segmental divisions. The nomenclature is that approved by the Thoracic Society. (Courtesy of the Editors of Thorax.)

CHAPTER 12

no longer contain cartilage in their walls. The bronchioles divide and the last of the purely conducting airways are known as the terminal bronchioles, beyond which lie the alveoli. The walls of the segmental bronchi are invisible on the chest radiograph unless seen end-on, when they may cause ring shadows (Fig.12.6). The acinus, which is 5–6 mm in diameter, comprises respiratory bronchioles, alveolar ducts and alveoli. The acini are grouped together in lobules of three to five acini which, in the lung periphery, are separated by septa and together comprise

the secondary pulmonary lobule.These peripheral interlobular septa, when thickened by disease, are the so-called septal or Kerley B lines. The bronchopulmonary segments are based on the divisions of the bronchi.The boundaries between segments are complex in shape and have been likened to the pieces of a three-dimensional (3D) jigsaw puzzle; there is no septation between them (except in the rare instance of a patient with accessory fissures). Atelectasis or pneumonia may predominate in one or other segment, but rarely conforms precisely to the whole of just one segment, since collateral air drift occurs across the segmental boundary. The position of the segments as seen on standard radiographs is illustrated in Figure 12.7. The pulmonary blood vessels (Fig. 12.8) are responsible for branching linear markings within the lungs both on conventional radiographs and CT. It is not possible to distinguish arteries from veins in the outer two-thirds of the lungs on plain radiographs. Centrally, the orientations of the arteries and veins differ: the lower lobe veins run more horizontally and the lower lobe arteries more vertically. In the upper lobes, the arteries and veins show a similar gently curving vertical orientation, but the upper lobe veins (when not superimposed on the arteries) lie lateral to the arteries and can sometimes be traced to the main venous trunk, the superior pulmonary vein. The diameter of the blood vessels beyond the hilum varies with the position of the patient and with various haemodynamic factors. On plain chest radiographs taken in the upright position, there is a gradual increase in the relative diameter of vessels equivalent in distance from the hilum as the eye travels from apex to base.The differences are abolished when the patient lies supine. These observations correlate

Figure 12.6 Ring shadows (arrows) due to end-on bronchi as a normal finding on chest radiography. The patient has a dual chamber pacing system.

1

6 10

2 3

7

H

H

8

4 9

5

A

H

H

15

11 12

14

16

18

13

17

B

6 1 2 7

3

15

10 H

H

v

4 11

5 12

C

9

14 18 13

D

17

• THE NORMAL CHEST

16

Figure 12.7 Diagrams of position of segments seen on plain frontal and lateral chest radiographs. There is substantial overlap of the projected images of the segments in both views; this overlap is worse in the frontal than the lateral projection. (A) Shows only the segments in the upper lobes and the middle lobe; (B) shows only the segments in the lower lobes; (C,D) show all the segments in the right and left lung, respectively, in the lateral view. H = hila, 1 = apical segment of right upper lobe (RUL), 2 = posterior segment of RUL, 3 = anterior segment of RUL, 4 = lateral segment of right middle lobe (RML), 5 = medial segment of RML, 6 = apical posterior segment of left upper lobe (LUL), 7 = anterior segment of LUL, 8 = superior segment of lingula, 9 = inferior segment of lingula, 10 = apical (superior) segment of right lower lobe (RLL), 11 = medial basal segment of RLL, 12 = anterior basal segment of RLL, 13 = lateral basal segment of RLL, 14 = posterior basal segment of RLL, 15 = apical (superior) segment of left lower lobe (LLL), 16 = anterior basal segment of LLL, 17 = lateral basal segment of LLL, 18 = posterior basal segment of LLL.

205

206

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 12.8 Pulmonary angiogram shows appearances in (A) arterial phase and (B) venous phase. Note the difference in arrangement of the central arteries and veins. The peripheral arteries show similar anatomy to the peripheral veins. (C) Coronal maximum intensity projection slab image from a CT pulmonary angiogram demonstrating the combined arterial and venous phase image in a patient with normal circulation.

with physiological studies of perfusion which show that in the erect position there is a gradation of blood flow (the lower zones showing greater blood flow than the upper zones) from apex to base, a difference that is less obvious in the supine patient. While a general statement regarding these differences in zonal blood vessel size can be made, it is difficult to draw conclusions from the size of any particular peripheral pulmonary vessel. Certain measurements have, however, been suggested for upright chest radiographs: 1 The artery and bronchus of the anterior segment of either or both upper lobes are frequently seen end-on.The diameter of the artery is usually much the same as the diameter of the bronchus (4–5 mm). In the authors’ experience, an endon vessel with a diameter of over 1.5 times the diameter of the adjacent bronchus indicates that the vessel is increased in size. 2 Vessels in the first anterior interspace should not exceed 3 mm in diameter. A rich network of lymphatic vessels drains the lung and pleura to the hilar lymph nodes. The sub-pleural lymphatics are found beneath the pleura at the junction of the interlobular septa with the pleura. These vessels connect with each other and with the lymphatic vessels accompanying the veins in the interlobular septa. Lymph then flows to the hilum via deep lymphatic channels that run peribronchially and in the deep septa of the lungs. Under normal circumstances the lymphatic network is invisible radiographically but when thickened the septa are seen as line shadows known as septal or Kerley lines. Thickened interlobular septa correspond to Kerley B lines and thickened deep septa correspond to Kerley A lines. There are a few intrapulmonary lymph nodes, but they are small and cannot be identified on a chest radiograph but may be seen as small, peripherally located ellipsoid nodules on CT12,13.

THE HILA Understanding the normal hilum on plain radiography, CT and magnetic resonance imaging (MRI) requires an appreciation of the anatomy of the major blood vessels (Figs 12.9–12.13). On plain radiograph and CT the densities of the normal hilum are due mainly to blood vessels (Figs 12.10–12.12). Normal lymph nodes cannot be recognized as discrete structures, and the bronchial walls contribute little to the bulk of the hila, being thin and easily recognized for what they are. On MRI (Fig. 12.13), the lack of signal from fast-flowing blood within the vessels or from air in the bronchi means that there is relatively little signal generated from normal hilar structures on standard spin-echo sequences. The only signal will be from slow-flowing blood in the vessels, from the bronchial walls, and from the fat and hilar nodes. Normal lymph nodes of just a few millimeters in size are often evident as discrete structures on submillimeter collimation multislice CT. The major points to remember when viewing the hila are: 1 The transverse diameter of the lower lobe arteries before their segmental divisions can be determined with reasonable accuracy: they measure 9–16 mm on the normal postero-anterior (PA) chest radiograph (Fig. 12.11). 2 The posterior walls of the right main bronchus and its division into the right upper lobe bronchus and bronchus intermedius are outlined by air and appear as a thin stripe on lateral plain radiographs (Fig. 12.14) and on CT (Fig. 12.12).The posterior walls of the equivalent bronchi on the left are rarely visible on the plain radiograph because the left lower lobe artery intervenes between the lung and the bronchial tree. The lung does, in fact, frequently invaginate between the left lower lobe artery and the descending aorta to contact the posterior wall of the left lower lobe bronchus, but this is usually only visible on CT or MRI. 3 The right pulmonary artery passes anterior to the major bronchi, whereas the left pulmonary artery arches superior to the left main bronchus. The central portion of the right

CHAPTER 12

• THE NORMAL CHEST

Figure 12.9 Diagrams of the relationships between the hilar blood vessels and bronchi. (A) Frontal view. (B) Right posterior oblique view of right hilum. (C) Left posterior oblique view of left hilum. (D) Lateral chest radiograph with major blood vessels drawn in. IPV = inferior pulmonary vein—only one has been drawn in since they are superimposed, LPA = left pulmonary artery, LSPV = left superior pulmonary vein, RPA = right pulmonary artery; RSPV = right superior pulmonary vein. (Diagrams drawn by Ron Ervin and reproduced with permission from Armstrong P (ed) 1983 Critical problems in diagnostic radiology. Lippincott, Philadelphia.)

207

Figure 12.10 Normal digital PA chest radiograph demonstrating position and density of the hilar structures. Arrows indicate the hilar points, where the superior pulmonary vein crosses the descending lower lobe artery, the left normally being level with or slightly higher than the right.

Figure 12.11 Frontal view of hila in plain chest radiograph. The measurement points for the diameter of the right lower lobe artery are indicated.

Figure 12.12 CT of normal hila. Two-millimetre collimation images have been obtained through the hilar structures during contrast medium enhancement and displayed on lung windows (L-500, W 1500). (A) Section just below the tracheal carina at the origin of the right upper lobe bronchus, immediately posterior to the upper lobe vein (v). (B) Section through level of right main pulmonary artery (RPA) and bronchus intermedius (curved arrow). Note the tongue of lung that contacts the left main bronchus between the aorta (A) and the left lower lobe artery (straight arrow). Note also that the right lung contacts the posterior wall of the bronchus intermedius as it extends into the azygo-oesophageal recess. (C) Section through the level of the middle lobe bronchus (long arrow) at the point of origin of the bronchus to the superior segment of the right lower lobe (arrow). Note that the middle lobe bronchus separates the right lower lobe artery from the right superior pulmonary vein as it enters the left atrium (LA). The lung contacts the posterior wall of the right lower lobe bronchus as it extends into the azygo-oesophageal recess. (D) Section through the level of the inferior pulmonary veins (arrows). At this level the lower lobe arteries have bilaterally divided into basal segmental divisions and are therefore narrower than 1 cm in diameter.

CHAPTER 12

• THE NORMAL CHEST

5 The pulmonary veins are similar on the two sides.The superior pulmonary vein is the anterior structure in the upper and mid hilum on both sides. Since, however, the central portions of the pulmonary arteries are so differently organized on the two sides, the relationships of the major veins to the arteries differ. On the right the superior pulmonary vein is separated from the central bronchi by the lower division of the right pulmonary artery, whereas on the left the superior pulmonary vein is separated from the pulmonary artery by the bronchial tree.

Figure 12.13 MRI of normal hilum. Note that there is relatively little signal from the hila because there is no signal either from air in the bronchial tree or from fast-flowing blood in the pulmonary artery or vein branches.

Both inferior pulmonary veins travel obliquely anteriorly and superiorly, inferior to the branches of the left lower lobe artery, to enter the left atrium. They are slightly posterior to the plane of the left lower lobe bronchi. They may be seen either end-on or in oblique cross-section in PA, lateral and oblique projections and may, therefore, simulate a mass.

THE MEDIASTINUM The radiographic anatomy of the mediastinum can be described from many points of view, depending on the technique that is under discussion. In this chapter only plain radiographs, CT and MRI will be considered in any detail, CT and MRI being illustrated first because an appreciation of the cross-sectional anatomy of the mediastinum helps in understanding the appearances on plain chest radiographs. The mediastinum is conventionally divided into superior, anterior, middle and posterior compartments. The exact anatomical boundaries of these divisions are unimportant to the radiologist (indeed they vary according to different authors), since they do not provide a clear-cut guide to disease and nor do their boundaries form any barriers to the spread of the disease.

Computed tomography and magnetic resonance imaging

Figure 12.14 Lateral view of the hila showing normal thickness of the posterior wall of the bronchus intermedius (arrow).

hilum consists of a combination of the right pulmonary artery and the superior pulmonary vein. Since these two vessels are immediately adjacent to one another (on the left, the left main bronchus lies between them), they may be responsible for a density that is sufficiently great to be confused with a mass on lateral plain radiographs and even, on occasion, on CT. 4 On lateral chest radiographs the angles between the middle and right lower lobe bronchi on the right, and the upper and lower lobe bronchi on the left, do not contain any large end-on vessels; a rounded shadow of greater than 1 cm in these angles is therefore unlikely to be a normal vessel14.

The blood vessels, trachea and main bronchi make up the bulk of the mediastinum, and the CT/MRI anatomy of these structures is illustrated in Figures 12.12–12.16. The thymus is situated anterior to the aorta and right ventricular outflow tract or pulmonary artery; it is often best appreciated on a section through the aortic arch or great vessels (Fig. 12.17). Before puberty15 the thymus fills in most of the mediastinum in front of the great vessels. During this period of life the gland varies so greatly in size that measurement is of little value in deciding normality. Approximate symmetry is the rule. Also, the thymus fills in the spaces between the great vessels and the anterior chest wall as if moulded by these structures. In adults the thymus is bilobed or triangular in shape. The maximum width and thickness of each lobe decreases with advancing age. Between the ages of 20 and 50, the average thickness as measured by CT decreases from 8–9 mm to 5–6 mm, the maximum thickness of each lobe being up to 15 mm. These diameters are greater on MRI, presumably because MRI demonstrates the thymic tissue even when it is partially replaced by fat. On MRI, sagittal images demonstrate the gland to be 5–7 cm long in its craniocaudad dimension.

209

Figure 12.15 CT of normal mediastinum. (A–E) Five 1-cm thick sections have been selected to show the important anatomical features. The level of each section is illustrated in the diagram. A. Ao = ascending aorta, Ao arch = aortic arch, AV = azygos vein, D. Ao = descending aorta, IA = innominate artery, LA = left atrium, LCA = left carotid artery, LIV = left innominate vein, LPA = left pulmonary artery, LSA = left subclavian artery, MPA = main pulmonary artery, N = normal lymph node, OES = oesophagus, RA = right atrium, RIV = right innominate vein, RPA = right pulmonary artery, RVO = right ventricular outflow tract, SPV = superior pulmonary vein, SVC = superior vena cava, T = trachea.

CHAPTER 12

• THE NORMAL CHEST

Figure 12.16 MRI of normal mediastinum and hila. Four sections have been chosen to show the important anatomical features: (A) is just below the tracheal carina; (B) is 1 cm below A; (C) is at the level of the right main pulmonary artery; (D) is at the level of the mid left atrium. A.Ao = ascending aorta, AV = azygos vein, BI = bronchus intermedius, D.Ao = descending aorta, LA = left atrium, LMBr = left main bronchus, LPA = left pulmonary artery; LV = left ventricle, MPA = main pulmonary artery, Oes = oesophagus, RA = right atrium, RMBr = right main bronchus, RSPV = right superior pulmonary vein, SPR = superior pericardial recess, SVC = superior vena cava, Th = thymus.

Figure 12.17 CT of normal thymus (arrows) in a young adult man.

In younger patients, the CT density of the thymus is homogeneous and close to that of other soft tissues, but after puberty the density gradually decreases owing to fatty replacement, so that above 40 years of age the thymus usually has an attenuation value identical to that of fat and is often indistinguishable from the adjacent mediastinal fat, apart from some residual thymic parenchyma which may be visible as streaky or nodular densities within the fat (Fig. 12.18)16,17. On MRI the intensity of the thymus in T1weighted images is similar to that of muscle and appreciably lower than that of mediastinal fat, although, as would be expected, this difference decreases with age. On T2weighted images, the intensity differences are slight and do not vary with age.

211

212

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 12.18 Thymic residues (arrows) shown by CT.

Figure 12.19 Diagram showing AJCC–UICC classification of regional lymph nodes.

Lymph nodes are widely distributed in the mediastinum. Ninety-five per cent of normal mediastinal lymph nodes are less than 10 mm in diameter, and the remainder, with few exceptions, are less than 15 mm in diameter18–22. Lymph nodes in the paraspinal areas, in the region of the brachiocephalic veins and in the space behind the diaphragmatic crura are generally smaller, 6 mm or less, whereas nodes in the aortopulmonary window, pretracheal and lower paratracheal spaces, and subcarinal compartment are often 6–10 mm in diameter. Lymph nodes encircle the trachea and main bronchi except where the aorta, pulmonary artery, or oesophagus are in direct contact with the airway.There is no clear division between the various nodes, but they can be categorized according to site.

The nomenclature of mediastinal lymph nodes should accord with the terms agreed by the American Joint Committee on Cancer and the Union International Contre le Cancer (AJCC– UICC) designed for staging carcinoma of the bronchus (Fig. 12.19; Table 12.1)23,24. These terms replaced the previous American Thoracic Society (ATS) classification. The AJCC–UICC classification is based on cross-sectional imaging in that it is directly referable to axial cross-sectional anatomy. The plane tangential to the upper margin of the aortic arch is an important dividing plane with nodes above this level being designated as: ‘highest mediastinal nodes’ (station 1 if they are above the upper rim of the left brachiocephalic vein); ‘right, left and posterior upper paratracheal’ (stations 2R,

Table 12.1 AJCC–UICC CLASSIFICATIONS OF REGIONAL LYMPH NODES 1. Highest mediastinal nodes lie above a horizontal line at the upper rim of the bracheocephalic (left innominate) vein 2. Upper paratracheal nodes lie above a horizontal line drawn tangential to the upper margin of the aortic arch and below the inferior boundary of No1 nodes 3. Prevascular and retrotracheal nodes may be designated 3A and 3P: midline nodes are considered to be ipsilateral 4. Lower paratracheal nodes lie to the right or left of the midline of the trachea between a horizontal line drawn tangential to the upper margin of the aortic arch and a line extending across the right or left main bronchus at the upper margin of the ipsilateral upper lobe bronchus. They are contained within the mediastinal pleural envelope. NB: The left lower paratracheal nodes lie medial to the ligamentum arteriosum 5. Subaortic (aortopulmonary window) nodes lie lateral to the ligamentum arteriosum or the aorta or left pulmonary artery and proximal to the first branch of the left pulmonary artery and lie within the mediastinal pleural envelope 6. Para-aortic nodes (ascending aorta or phrenic) lie anterior and lateral to the ascending aorta and the aortic arch or the innominate artery, beneath a line tangential to the upper margin of the aortic arch 7. Subcarinal nodes lie caudal to the carina of the trachea, but not associated with the lower lobe bronchi or arteries within the lung 8. Para-oesophageal nodes (below carina) lie adjacent to the right or left of the midline, excluding subcarinal nodes 9. Pulmonary ligament nodes lie within the pulmonary ligament, including those against the posterior wall and lower part of the inferior pulmonary vein 10. Hilar nodes lie distal to the mediastinal pleura reflection and the nodes adjacent to the bronchus intermedius on the right 11. Interlobar nodes lie between the lobar bronchi 12. Lobar nodes lie adjacent to the distal lobar bronchi 13. Segmental nodes lie adjacent to the segmental bronchi 14. Subsegmental nodes lie around the subsegmental bronchi NB. Station 1 through 9 nodes lie within the mediastinal pleural envelope, whereas station 10 through 14 nodes lie outside the mediastinal pleura within the visceral pleura.

CHAPTER 12

2L and 3P, respectively); and ‘prevascular’ if they lie anterior to the arteries to the head and neck (station 3A). The nodes below the plane tangential to the upper margin of the aortic arch include the following: right and left lower paratracheal (stations 4R and 4L); subaortic (aortopulmonary window) nodes (station 5); para-aortic nodes which lie anterior and lateral to the ascending aorta, the aortic arch, or the proximal brachiocephalic artery (station 6); and subcarinal nodes, which lie beneath the main bronchi within the mediastinal pleura (station 7). Low down in the mediastinum are the para-oesophageal (station 8) and pulmonary ligament nodes (station 9). Nodes are also present in the retrocrural areas and cardiophrenic angles. The nodes outside the mediastinal pleura are hilar (station 10), interlobar (station 11), lobar, segmental and subsegmental (stations 12–14); all these nodes are removed at pneumonectomy. The oesophagus is visible on all axial CT and MRI sections from the root of the neck down to the diaphragm. It may contain a small amount of air in approximately 80% of normal people. If there is sufficient mediastinal fat, the entire circumference of the oesophagus can be identified, and if air is present in the lumen, the uniform thickness of the wall can be appreciated. Without air, the collapsed oesophagus appears circular or oval in shape and measures approximately 1 cm in its narrowest diameter. On MRI the signal intensity on T1-weighted images is similar to muscle but on T2-weighted images the oesophagus often shows much higher signal intensity than muscle.

• THE NORMAL CHEST

Ant. junction line IA LCA Paratracheal stripe

SVC

LSA T

Posterior tracheal stripe

Oesophagus Post. junction line

Pleuro-oesophageal line

A

Anterior junction line

Paratracheal stripe

Ao Arch

SVC T

Post. tracheal stripe

B

Oesophagus

Pleuro-oesophageal line

Radiographic appearances Plain chest radiographs provide limited information regarding mediastinal anatomy, since only the interfaces between the lung and the mediastinum are visualized (Fig. 12.20).

Junction lines25,26 When there is only a small amount of fat anterior to the ascending aorta and its major branches, the two lungs may be separated anteriorly by little more than the four intervening layers of pleura. In such patients an anterior junction line is visible on frontal chest radiographs (Fig. 12.21).The line diverges and fades out superiorly and cannot be identified above the level of the clavicles. It descends for a variable distance, usually deviating to the left, but never extending lower than the point where the two lungs separate to envelop the right ventricular outflow tract. The lungs may also come close together behind the oesophagus, forming the posterior junction line (Fig. 12.21). This line, unlike the anterior junction line, separates to envelop the aortic arch. It may reform below the aortic arch where the two lungs occasionally abut behind the oesophagus. Superiorly, the posterior junction line extends to the level of the lung apices where it diverges and disappears, a level appreciably higher than the medial ends of the clavicles. The differences in the superior extent of the anterior and posterior junction lines are related to the sloping boundary between the root of the neck and the thorax.

Oesophagus Azygo-oesophageal line Right paraspinal line

C

Ao Azygos vein Left paraspinal line

Figure 12.20 Diagrams illustrating the mediastinal boundaries and junction lines. The visualization of the junction lines on a plain chest radiograph is variable, depending on how much fat is present in the mediastinum and on how closely the two lungs approximate to one another. (A) Section just above the level of the aortic arch; (B) section through the aortic arch; (C) section through the heart.

The major value of being able to identify the anterior and posterior junction lines is that a mass, or other space-occupying process, in the junctional areas can be excluded if these lines are visible. Since both junction lines are inconsistently seen, however, the lack of visualization of one or both is not a reliable sign of disease.

213

214

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 12.22 Right tracheal stripe (straight arrows) and pleurooesophageal line (curved arrows) demonstrated on (A) plain radiograph and (B) unenhanced CT.

Figure 12.21 Anterior junction line (curved arrows) and the supraaortic posterior junction line (straight arrows). Note that the supraaortic posterior junction line goes well above the level of the clavicles and extends down to the top of the aortic arch but then stops, whereas the anterior junction line starts below the clavicles and continues well below the aortic arch.

Right mediastinum above the azygos vein The right superior mediastinal border is formed by the right brachiocephalic (innominate) vein and the superior vena cava. With aortic or brachiocephalic (innominate) artery ectasia or unfolding, either of these veins may be pushed laterally or the mediastinal border may be formed by the aorta or the right brachiocephalic artery. The right paratracheal region can be seen through the right brachiocephalic vein and superior vena cava because the lung contacts the right tracheal wall from the level of the clavicles down to the azygos vein, producing a visible stripe of uniform thickness known as the right paratracheal stripe (Fig.12.22), between the tracheal air column and the lung.This stripe, which should be no more than 5 mm wide, is visible in approximately two-thirds of normal people. It consists of the wall of the trachea and the adjacent mediastinal fat, but no focal bulges due to individual paratracheal lymph nodes can be seen. As with the junction lines, the diagnostic value of this stripe is that its presence excludes a spaceoccupying process in the area where the stripe is visible. The azygos vein is outlined by air in the lung at the lower end of the right paratracheal stripe. The diameter of the azygos vein in the tracheobronchial angle is variable: it may be considered normal when its diameter is 10 mm or less. The nodes immediately beneath the azygos vein are known as azygos nodes and are not recognizable on the normal chest radiograph. The lung posterior to the trachea contacts the right wall of the oesophagus so that a recognizable border may be seen in

the frontal projection. If the oesophagus at this level contains air, then the right wall of the oesophagus is seen as a stripe, the so-called oesophageal–pleural stripe27, curving superiorly and laterally behind the tracheal air column (Fig. 12.22). In summary, on a frontal radiograph three interfaces are potentially recognizable in the right mediastinum above the azygos vein: the superior vena cava border, the right wall of the trachea, and the right wall of the oesophagus.

Left mediastinum above the aortic arch The mediastinal shadow to the left of the trachea above the aortic arch is of low density and is caused by the left carotid and left subclavian arteries together with the left brachiocephalic (innominate) and jugular veins. The usual appearance on the frontal projection is a gently curving border formed by the left subclavian artery, which fades out where the artery enters the neck. A separate interface may occasionally be discernible for the left carotid artery or left brachiocephalic vein. The outer margin of the left tracheal wall is virtually never outlined, because the lung is separated from the trachea by the aorta and other vessels listed above.

Trachea and retrotracheal area in the lateral view The air column in the trachea can be seen throughout its length as it descends obliquely inferiorly and posteriorly. The course of the trachea on a normal lateral view is straight, or bowed anteriorly in patients with aortic unfolding, with no visible indentation from adjacent vessels. Small indentations into the air column of the trachea from tracheal cartilage rings may be apparent on the lateral view. The carina cannot be identified on the lateral view (though the right main bronchus is often mistaken for it). Its anterior wall is visible in a minority of patients, but the posterior wall is usually seen because lung often passes behind the trachea, thereby permitting visualization of the posterior tracheal (stripe) band28.The thickness of this stripe is 2–3 mm, provided it is formed solely by the tracheal wall and pleura (Fig. 12.23). If a large amount

CHAPTER 12

• THE NORMAL CHEST

Figure 12.23 Lateral view of trachea and major bronchi. (A) In this example, the posterior wall of the trachea is outlined by lung posterior to it (arrow). (B) In this example, the collapsed oesophagus is between the lung and the trachea (arrow).

of air is present in the oesophagus, the posterior tracheal band may be much thicker, since it then comprises the combined thicknesses of the posterior tracheal wall and the anterior oesophageal wall. Alternatively, the lung may be separated from the trachea by the full width of a collapsed oesophagus, leading to a band of density measuring 10 mm or more (Fig. 12.23).

Supra-aortic mediastinum on the lateral view A variable proportion of the aortic arch and its major branches is visible on the lateral view, depending largely on the degree of aortic unfolding. The brachiocephalic (innominate) artery is the only branch vessel that is recognizable with any frequency. It arises anterior to the tracheal air column; usually the origin is unclear but, after a variable length, the posterior wall can be seen as a gently S-shaped interface as it crosses the tracheal air column. The left and right brachiocephalic (innominate) veins are also sometimes visible on the lateral view. The left brachiocephalic vein is seen as an extrapleural bulge behind the manubrium in a small proportion of normal people (Fig. 12.24).

Figure 12.24 Bulge behind manubrium representing normal left innominate (brachiocephalic) vein (arrow).

Right middle mediastinal border below the azygos arch Below the azygos arch, the right lower lobe makes contact with the right wall of the oesophagus and the azygos vein as it ascends next to the oesophagus. This portion of the lung is known as the azygo-oesophageal recess, and the interface is known as the azygo-oesophageal line (Fig. 12.25). The shape of the azygos arch varies considerably in different subjects and therefore the shape of the upper portion of the azygo-oesophageal line varies accordingly. The upper few centimetres of the azygo-oesophageal line are, however, always straight or concave toward the lung, so that a convex shape suggests the

presence of a subcarinal mass or left atrial enlargement. The azygo-oesophageal line can be traced down to the posterior costophrenic angle in normal subjects.

Left cardiac border below the aortic arch This left cardiac border is formed by the main pulmonary artery and heart. The pleura smoothing the angle between the mid-portion of the aortic arch and the main and left pulmonary artery, the so-called aortic–pulmonary mediastinal stripe29, is the lateral extent of the aortopulmonary

215

216

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 12.25 Azygo-oesophageal line (arrows).

window. Because the aortopulmonary window is a sensitive place to look for lymph node enlargement, Blank and Castellino30 investigated the variable shape of this pleural reflection, as illustrated in Figure 12.26. A small ‘nipple’ may occasionally be seen projecting laterally from the aortic knuckle owing to the presence of the left superior intercostal vein31,32 (see Fig. 12.3A). The vein, which is formed by the junction of the left first to fourth intercostal veins, arches forward around the aorta just below the origin of the left subclavian artery to enter the left brachiocephalic vein. This normal nipple should not be misinterpreted as adenopathy projecting from the aortopulmonary window. The interface between the lung and the left wall of the aorta can almost invariably be followed down to the level of the diaphragm, though contact with the proximal portion of the left pulmonary artery may silhouette a small portion of the interface.The shape varies with the degree of aortic unfolding. Though the lung invaginates between the heart and aorta to contact the left wall of the oesophagus, the interface with the oesophagus may be seen as a line if air is present within the lumen of the oesophagus.

Paraspinal lines Although lymph nodes and intercostal veins occupy the space between the spine and the lung, they cannot normally be recognized individually. In individuals with little fat, the interfaces, known as the paraspinal lines, may closely reflect the

Figure 12.26 Patterns of pleural reflection along the left border of the great vessels and heart. The heavy line indicates the visible pleural interface. (Adapted from Blank N, Castellino R A 1972 Patterns of pleural reflections of the left superior mediastinum: normal anatomy and distortions produced by adenopathy. Radiology 102: 585–589, with permission from the Radiological Society of North America).

undulations of the lateral spinal ligaments, but the more fat there is, the more these undulations are smoothed out. The thickness of the left paravertebral space is usually greater than that of the right and can be more than 10 mm in obese subjects. Aortic unfolding contributes to the thickness of the left paraspinal line; as the aorta moves posteriorly and laterally, it strips the pleura from its otherwise close contact with the profiled portions of the spine.

Retrosternal line The band-like opacity simulating pleural or extrapleural disease is often seen along the lower third of the anterior chest wall on a lateral chest radiograph (Fig. 12.27)33. This density is due to mediastinal fat and to the differing anterior extent of the left and right lungs. The left lung does not contact the most anterior portion of the left thoracic cavity at these levels because the heart occupies the space. The band-like opacity is therefore accounted for by the normal heart and mediastinum, rather than by disease.

THE DIAPHRAGM The diaphragm consists of a large dome-shaped central tendon surrounded by a sheet of striated muscle which is attached to ribs 7 to 12 and to the xiphisternum. The two diaphragmatic crura, which arise from the upper three lumbar vertebrae, arch superiorly and anteriorly to form the margins of the aortic

CHAPTER 12

• THE NORMAL CHEST

Figure 12.28 Right phrenic nerve as it passes over the surface of the right hemidiaphragm (arrows).

Figure 12.27 Retrosternal stripe (arrowheads) and inferior vena cava in lateral projection (arrow).

and oesophageal hiatuses. The median arcuate ligament connecting the two crura forms the anterior margin of the aortic hiatus, and the crura themselves form its lateral boundary. The oesophageal hiatus lies anterior to the aortic hiatus, and anterior to that lies the hiatus for the inferior vena cava, which is situated within the central tendon immediately beneath the right atrium. In most individuals, the diaphragm has a smooth domed shape, but a scalloped outline is also common. The angle of contact with the chest wall is acute and sharp, but blunting of this angle can be normal in athletes, because they can depress their diaphragm to a remarkable degree on deep inspiration. The normal right hemidiaphragm is found at about the level of the anterior portion of the sixth rib, with a range of approximately one interspace above or below this level34. In most people, the right hemidiaphragm is 1.5–2.5 cm higher than the left, but the two hemidiaphragms are at the same level in some 9% of the population. In a few normal individuals the left hemidiaphragm is up to 1 cm higher than the right. The normal excursion of the diaphragm is usually between 1.5 and 2.5 cm, though greater degrees of movement are not uncommon. Transabdominal ultrasound, which is capable of providing accurate real-time measurement of movement, shows a considerable normal range of between 2.0 and 8.6 cm, the mean excursion of the right hemidiaphragm on deep inspiration being 53 mm (sd 16.4) and that of the left being 46 mm (sd 12.4)35. Incomplete muscularization, known as eventration, is also common. An eventration is composed of a thin membranous sheet replacing what should be muscle. Usually it is partial, involving one-half to one-third of the hemidiaphragm. The lack of muscle manifests itself radiographically as elevation of the affected portion of the diaphragm, and the usual appear-

ance is one of a smooth hump on the contour of the diaphragm.Total eventration of a hemidiaphragm, which is much more common on the left than the right, results in elevation of the whole hemidiaphragm; on fluoroscopy hemidiaphragm movement is poor, absent, or paradoxical, and severe cases of congenital eventration cannot be distinguished from acquired paralysis of the phrenic nerve. A linear density arising from the lateral wall of the inferior vena cava (Fig. 12.28) is often seen coursing over the surface of the right hemidiaphragm.This line represents pleura and an envelope of fat investing the phrenic nerve, according to Berkman et al36, or the inferior phrenic artery and vein, according to Ujita et al37.

REFERENCES 1. Glazer H S, Anderson D J, DiCroce J J et al 1991 Anatomy of the major fissure: evaluation with standard and thin-section CT. Radiology 180: 839–844 2. Mayo J R, Muller N L, Henkelman R M 1987 The double-fissure sign: a motion artifact on thin-section CT scans. Radiology 165: 580–581 3. Ariyurek O M, Gulsun M, Demirkazik F B 2001 Accessory fissures of the lung: evaluation by high-resolution computed tomography. Eur Radiol 11: 2449–2453. Epub 2001 May 19 4. Caceres J, Mata J H, Alegnet X 1993 Increased density of the azygos lobe on frontal radiographs simulating disease: CT findings in seven patients. Am J Roentgenol 160: 245–248 5. Rabinowitz J G, Cohen B A, Mendleson D S 1984 The pulmonary ligament. Radiol Clin North Am 22: 659–672 6. Berkman Y M, Drossman S R, Marboe C C 1993 Intersegmental (intersublobar) septum of the lower lobe in relation to the pulmonary ligament: anatomic, histologic, and CT correlations. Radiology 185: 389–393 7. Breatnach E, Abbott G C, Fraser R E 1984 Dimensions of the normal human trachea. Am J Roentgenol 142: 903–906 8. Vock P, Spiegel T, Fram E K et al 1984 CT assessment of the adult intra-thoracic cross section of the trachea. J Comput Assist Tomogr 8: 1076–1082 9. Griscom N T, Wohl M E 1986 Dimensions of the growing trachea related to age and gender. Am J Roentgenol 146: 233–237

217

218

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

10. Jardin M, Remy J 1986 Segmental bronchovascular anatomy of the lower lobes: CT analysis. Am J Roentgenol 147: 457–468 11. Lee K S, Bae W K, Lee B H et al 1991 Bronchovascular anatomy of the upper lobes: evaluation with thin section CT. Radiology 181: 765–772 12. Oshiro Y, Kusumoto M, Moriyama N et al 2002 Intrapulmonary lymph nodes: thin-section CT features of 19 nodules. J Comput Assist Tomogr 26: 553–557 13. Hyodo T, Kanazawa S, Dendo S et al 2004 Intrapulmonary lymph nodes: thin-section CT findings, pathological findings, and CT differential diagnosis from pulmonary metastatic nodules. Acta Med Okayama 58: 235–240 14. Park C-K, Webb W R, Klein J S 1991 Inferior hilar window. Radiology 178: 163–168 15. Mendelson D S 2001 Imaging of the thymus. Chest Surg Clin North Am 11: 269–293 16. Dixon A K, Hilton C J, Williams C T 1981 Computed tomography and histological correlation of the thymic remnant. Clin Radiol 32: 255–257 17. Moore A V, Korobkin M, Olanow W et al 1983 Age-related changes in the thymus gland: CT-pathologic correlation. Am J Roentgenol 141: 241–246 18. Genereux G P, Howie J L 1984 Normal mediastinal lymph node size and number: CT and anatomic study. Am J Roentgenol 142: 1095–1100 19. Glazer G M, Gross B H, Quint L E et al 1985 Normal mediastinal lymph nodes: number and size according to American Thoracic Society mapping. Am J Roentgenol 144: 261–265 20. Schynder P A, Gamsu G 1982 CT of the pretracheal retrocaval space. Am J Roentgenol 136: 303–308 21. Ingram C E, Belli A M, Lewards M D et al 1989 Normal lymph node size in the mediastinum: a retrospective study in two patient groups. Clin Radiol 40: 355–339 22. Murray J G, O’Driscoll M, Curtin J J 1995 Mediastinal lymph node size in an Asian population. Br J Radiol 68: 348–350 23. Mountain C F, Dresler C M 1997 Regional lymph node classification for lung cancer staging. Chest 111: 1718–1723

24. Cymbalista M, Waysberg A, Zacharias C et al 1999 CT demonstration of the 1996 AJCC–UICC regional lymph node classification of lung cancer staging. Radiographics 19: 900 25. Proto A V, Simmons J D, Zylak C J 1983 The anterior junction anatomy. CRC Crit Rev Diagn Imaging 19: 111–173 26. Proto A V, Simmons J D, Zylak C J 1983 The posterior junction anatomy. CRC Crit Rev Diagn Imaging 20: 121–173 27. Cimmino C V 1981 The oesophageal–pleural stripe: an update. Radiology 140: 607–613 28. Bachman A L, Teixidor H S 1975 The posterior tracheal band: a reflection of local mediastinal abnormality. Br J Radiol 48: 352–359 29. Keats T E 1975 The aortic–pulmonary mediastinal stripe. Am J Roentgenol 116: 107–109 30. Blank N, Castellino R A 1972 Patterns of pleural reflections of the left superior mediastinum: normal anatomy and distortions produced by adenopathy. Radiology 102: 585–589 31. Lane E J, Heitzman E R, Dinn W M 1976 The radiology of the superior intercostal veins. Radiology 120: 263–267 32. Abiru H, Ashizawa K, Hashmi R et al 2005 Normal radiographic anatomy of thoracic structures: analysis of 1000 chest radiographs in Japanese population. Br J Radiol 78: 398–404 33. Whalen J P, Meyers M A, Oliphant M et al 1973 The retrosternal line. Am J Roentgenol 117: 861–872 34. Lennon E A, Simon G 1965 The height of the diaphragm in the chest radiograph of normal adults. Br J Radiol 38: 937–943 35. Houston J G, Morris A D, Howie C A et al 1992 Quantitative assessment of diaphragmatic movement: a reproducible method using ultrasound. Clin Radiol 46: 405–407 36. Berkman Y M, Davis S D, Kazam E 1989 Right phrenic nerve: anatomy, CT appearance, and differentiation from the pulmonary ligament. Radiology 173: 43–46 37. Ujita M, Ojiri H, Arizumi M et al 1993 Appearance of the inferior phrenic artery and vein on CT scans of the chest: a CT and cadaveric study. Am J Roentgenol 160: 745–747

CHAPTER

The Chest Wall, Pleura, Diaphragm and Intervention

13

John A. Verschakelen

The chest wall • Soft tissues • Bony structures The pleura • Pleural effusion

• Pneumothorax • Pleural thickening and fibrothorax • Pleural calcification • Pleural tumours The diaphragm

THE CHEST WALL Although there is a wide variety of tissues and structures that make up the chest wall, based on their radiographic presentation, its components can be grouped into two major parts: the soft tissues and the bony structures.

SOFT TISSUES On the chest radiograph the soft tissues present as areas of increased density that in part project next to the bony chest wall and in part overlay the different components of the chest. Abnormalities of the soft tissues will present as an abnormal increase or decrease in density often combined with the appearance of an abnormal contour or the disappearance of a normal contour. Because of better density resolution and multiplanar reformatting, computed tomography (CT) can better demonstrate the different tissues of the chest wall. CT has the advantage over magnetic resonance imaging (MRI) of higher spatial resolution and the ability to better identify bony structures. MRI, however, yields greater soft tissue contrast which can be important1,2. Multiplanar imaging and three-dimensional (3D) reformation can be performed with both techniques.

Ultrasound may also be used to examine the chest wall. In general it provides less detailed and comprehensive information but it usually enables the lesion to be localized, allows a distinction to be made between cystic and solid lesions, and enables guided aspiration/biopsy to be performed under imaging control.

Breasts On the female chest radiograph it is mandatory to check that both breasts are present. Unilateral radical mastectomy is usually easy to detect because it generates a unilateral mid/lower zone transradiancy and an abnormally straight anterior axillary fold that passes upwards and inwards towards the mid clavicle (Fig. 13.1). Bilateral radical mastectomy is more difficult to identify, but an overall increase in basal transradiancy and axillary fold abnormalities should provide adequate clues. Surgical interventions short of radical mastectomy may be impossible to detect, but close attention to the relative transradiancy of the breast regions and to the breast contours may provide suggestive findings. Nipple shadows can mimic intrapulmonary nodules. A putative nipple should be checked for compatible size (5– 15 mm), shape and location—its relation to the breast outline in a woman or the pectoralis opacity in a man.

220

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.1 Left mastectomy. The left hemithorax is more transradiant than the right.

Muscles On the chest radiograph the pectoralis major produces a broad, band-like opacity extending downwards and medially from the axilla. Unilateral absence or hypoplasia of the pectoralis major results in a unilateral transradiancy and an abnormal anterior axillary fold as seen with mastectomy. In Poland’s syndrome these changes are accompanied by ipsilateral hand and arm anomalies (particularly syndactyly) with or without absence of pectoralis minor, rib anomalies, and hypoplasia of breast and nipple.

Soft tissue calcification Soft tissue calcification may occur in the chest wall, and clues to its site and nature are provided by its morphology distribution and the clinical history. Possible causes to consider include granulomatous lymph nodes, parasites (Taenia solium and Dracunculus medinensis), calcinosis universalis, childhood dermatomyositis, tuberculosis (spine, ribs, or soft tissues) and bone neoplasms. Ossification is rare and most commonly seen in fibrodysplasia ossificans progressiva.

Subcutaneous emphysema Subcutaneous emphysema of the chest wall is not uncommon following surgery, pleural drain placement, trauma, or in cases of spontaneous or acquired pneumomediastinum. Air dissects along tissue planes and between muscle bundles, giving an overall pattern of linear transradiancies which can significantly interfere with the interpretation of the underlying structures. In this way diagnosis of pneumothorax can become very difficult. In case of doubt CT can be performed.

Soft tissue tumours A soft tissue tumour of the chest wall gives rise to an opacity. Malignant and inflammatory lesions cause bony destruction and

benign ones result in rib separation and notch-like remodelling from pressure erosion. The most common benign chest wall tumour is a lipoma, but a variety of other mesenchymal tumours occur, including neurofibromas (focal or plexiform), neurilemmomas, haemangiomas and lymphangiomas (cystic hygromas). On CT, lipomas are well-demarcated homogeneous masses of low density (−90 to −150 HU).They contain few, if any, other soft tissue components; the presence of the latter in a fatty tumour suggests a liposarcoma. MRI features are also characteristic, with high signal on T1-weighted images, intermediate signal on T2-weighted images, and low signal with fat suppression1,2. Neurofibromas on CT characteristically have a lower density than muscle both before and after intravenous contrast medium. On MRI, neurofibromas give low to intermediate signal on T1-weighted images but high signal on T2-weighted images and marked contrast enhancement after gadolinium, which allows clear delineation of their extent. Haemangiomas are uncommon lesions that occasionally show phlebolithic calcification on plain radiography. Findings on CT include phleboliths, bone remodeling, and an enhancing mass. MRI is the best investigation for delineating their extent. Lesions give an intermediate signal on T1-weighted images and a high signal on T2-weighted images, accompanied by inhomogeneities generated by vessels, soft tissue, and elements derived from haemorrhage1,2. Lymphangiomas on CT have the features of a fluid-filled cyst with or without septation. On MRI they have the features of a cyst with low protein content. Malignant primary tumours arising in the soft tissues of the chest wall are unusual, the most common being lipo- or fibrosarcomas. Secondary tumours of the chest wall are common, particularly when due to local spread (carcinoma of the breast and lung, lymphoma) (Fig. 13.2); see bronchial carcinoma and Pancoast’s tumour, below.

BONY STRUCTURES Although depicting bone abnormalities is not the primary goal of a chest radiograph, the bony structures that are, as a result of the technique, often only partially visible should be carefully examined. CT, when indicated, is better at demonstrating congenital or acquired remodelling or complicated fractures; multidetector 2D or 3D reformations can be helpful.

Ribs There are normally 12 pairs of ribs. Cervical ribs occur in 1–2% of the population and are commonly bilateral, though often asymmetrical. Congenital abnormalities of modelling may be confined to one or two ribs or be generalized. One or a few upper ribs are commonly bifid, splayed, fused, or hypoplastic. Usually occurring in isolation, these anomalies are occasionally part of a syndrome (e.g. basal cell naevus syndrome) or associated with other anomalies (e.g. Sprengel’s deformity). With acquired remodelling, abnormalities tend to be focal, affecting one or many ribs. Such acquired changes may follow

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.2 Invasive malignant T-cell lymphoma. (A) High-resolution chest wall ultrasound image through the intercostal space showing echoic cortical rib and consecutive echo void behind. Within the intercostal space hypo-echoic tumour tissue (arrowheads) is seen invading the posterior chest wall. (B) Same patient, transverse contrast-enhanced CT. Enhancing peripheral tumour tissue is widely invading the posterior chest wall. Exact definition of invasive tissue is impaired owing to low soft tissue contrast resolution in CT. (C) Same patient, sagittal T1-weighted MRI pre- (left) and post- (right) contrast medium nicely display the widespread invasion of the posterior chest wall by enhancing tumour tissue. There is already invasion of two ribs, including cortical rib destruction (arrowheads). The central nonenhancement of the tumour is due to necrosis. (Courtesy of Dr R. Bittner.)

fracture, surgery, osteomyelitis and empyema drainage, or result from external pressure (rib notching). The two main causes of rib notching are coarctation of the aorta and neurofibromatosis Type I. Ribs may fracture and the callus formed can sometimes mimic an intrapulmonary opacity. Destructive rib lesions occur most commonly in osteomyelitis or neoplastic disease. The former is uncommon and may be haematogenous (e.g. staphylococcal or tuberculous) or caused by direct spread from lung and pleural space (e.g. in actinomycosis). Bronchial carcinoma, including Pancoast’s tumour, commonly spreads from lung to rib. In this latter condition MRI can be performed to study the extent of the disease, especially the relationship between the tumour and the plexus brachialis in case of a Pancoast’s tumour3 (Fig. 13.3). Multidetector CT (MDCT) can also play an important role especially because it can better evaluate the invasion in the bony cortex of the ribs4 (Fig. 13.3D). Also, 3D image reconstruction methods can be used in selected cases to clarify a complex relationship between the tumour invading the chest wall and vascular structures of the thoracic inlet. Various primary and secondary tumours can affect ribs, causing localized lesions. Benign primary tumours are infrequent, and of these, the cartilaginous tumours (chondromas, osteochondromas) are the most common. They are predominantly anterior and may show characteristic cartilaginous calcification. Other lesions that broadly fall into this category include fibrous dysplasia, histiocytosis X, haemangioma and aneurysmal bone cyst1 (Fig. 13.4). The most common malignant rib tumours are metastatic deposits and myeloma. Primary malignant tumours are rare, chondrosarcomas being the least uncommon. Other malignancies that occur occasionally include lymphoma, osteosarcoma and round-cell tumours.

Sternum This is well displayed in a lateral chest radiograph but is inconspicuous in the frontal projection, in which only the manubrial margins are sometimes visible, giving rise to confusing shadows that may mimic mediastinal widening. Various sternal deformities are described, and the most important radiologically is the depressed sternum (funnel chest, pectus excavatum) in which there is approximation of the lower half of the sternum and the spine (Fig. 13.5). This may be an isolated abnormality or it may be associated with other disorders such as Marfan’s syndrome or congenital heart disease (particularly atrial septal defect [ASD]). The radiological signs on a postero-anterior (PA) chest radiograph consist of a shift of the heart to the left, straightening of the left heart border with prominence of the main pulmonary artery segment, loss of the descending aortic interface, and an increased opacity in the right cardiophrenic angle, often accompanied by a loss of clarity of the right heart border which simulates right middle lobe disease. The diagnosis can be suspected on a PA radiograph from the steep inferior slope of the anterior ribs and undue clarity of the lower dorsal spine seen through the heart. Pigeon chest (pectus carinatum) represents the reverse deformity and may be congenital or acquired. Neoplasms of the sternum are usually malignant (myeloma, chondrosarcoma, lymphoma or metastatic carcinoma), the most common benign tumour being a chondroma. Non-neoplastic processes that may affect the sternum include osteomyelitis, histiocytosis X, Paget’s disease, fibrous dysplasia, osteitis fibrosa cystica and intersternocostoclavicular hyperostosis. CT is the best investigation for imaging the sternum because it eliminates overlapping structures, detects bony destruction, allows imaging of adjacent soft tissues (the parasternal–internal mammary zone), and has good contrast resolution superior to that of conventional radiography or tomography.

221

222

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.3 Pancoast’s tumour. (A,B) MRI. (A) Coronal and (B) sagittal image. Large tumour in the left upper lobe invading the soft tissues and displacing the vascular structures anteriorly (arrows). The brachial plexus has also been invaded (arrowheads). (C,D) CT (different patient). (C) Coronal and (D) sagittal image bone window setting. Large tumour in the left upper lobe invading the soft tissues, displacing and invading the left subclavian artery and invading a rib (arrow).

Clavicles The medial clavicular ends are important landmarks used together with the spine in assessing rotation on a radiograph. The joints at both ends are synovial but only the acromioclavicular joint can be assessed with confidence on a chest radiograph. It may be eroded in any synovitis, particularly rheumatoid arthritis, and is also commonly fuzzy and illdefined in hyperparathyroidism and rickets. Neoplasms of the clavicle are usually malignant (myeloma or metastatic). Other

primary tumours and tumour-like lesions include osteosarcoma, Ewing’s sarcoma, post-radiation sarcoma, aneurysmal bone cyst, histiocytosis X and intersternocostoclavicular hyperostosis. Either CT or MRI is required to provide a full evaluation of the medial clavicular ends.

Spine Kyphoscoliosis makes assessment of the chest radiograph difficult and CT is often necessary to evaluate possible thoracic disease.

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.4 Fibrous dysplasia in a rib; chest radiograph detail of the left lung. Compared with the other ribs the ninth rib shows an increase in density and is slightly broadened.

Figure 13.5 Depressed sternum. (A) PA chest radiograph. The depressed sternum displaces the heart to the left and rotates it so that the left heart border adopts a straight configuration. The right heart border becomes ill-defined and is bounded by a hazy opacity, simulating collapse of the right middle lobe. The ribs show their characteristic configuration—horizontal posteriorly and steeply oblique anteriorly. The posterior displacement of the sternum is better demonstrated on (B) the lateral chest radiograph and (C) the axial CT.

223

224

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

THE PLEURA The chest radiograph is still the most important and widely used means of demonstrating and following the progress of pleural disease, though ultrasound, CT and MRI can play a significant role in a number of specific situations. Pleural disease is manifest by the accumulation of fluid or air in the pleural space, by pleural thickening (with or without calcification), or by the presence of a pleural mass.

and left effusions with pancreatitis, pericarditis, oesophageal rupture and aortic dissection. Massive effusions are most commonly due to malignant disease, particularly metastases (lung or breast), but may also occur in heart failure, cirrhosis, tuberculosis, empyema and trauma.

PLEURAL EFFUSION

Free pleural fluid A small amount of free fluid may be undetectable on an erect PA chest radiograph as it tends initially to collect under the lower lobes. Such small subpulmonary effusions can be demonstrated by ultrasound or CT. An alternative technique, the lateral decubitus chest radiograph has largely been replaced by these newer techniques6. As the amount of effusion increases, the posterior and then the lateral costophrenic angles become blunted, by which time a 200–500 ml effusion is present. Following this the classical signs develop, viz homogeneous opacification of the lower chest with obliteration of the costophrenic angle and the hemidiaphragm. The superior margin of the opacity is concave to the lung and is higher laterally than medially. Above and medial to the meniscus there is a hazy increase in opacity owing to the presence of fluid posterior and anterior to the lungs (Fig. 13.6). Massive effusions cause dense opacification of the hemithorax with contralateral mediastinal shift (Fig. 13.7). Absence of mediastinal shift with a large effusion raises the strong possibility of obstructive collapse of the ipsilateral lung or extensive pleural

A number of different types of fluid may accumulate in the pleural space, the most common being transudate, exudate (thin or thick), blood and chyle. Occasionally effusions are highly specific, not falling into any of the above categories and containing, for example, bile, cerebrospinal fluid, or iatrogenic fluids. All types of pleural effusion are radiographically identical, though historical, clinical and other radiological features may help limit the diagnostic possibilities. Sometimes, also CT and MRI can help to specify the diagnosis. Bilateral pleural effusions tend to be transudates because they develop secondary to generalized changes that affect both pleural cavities equally—a rise in capillary pressure or a fall in blood proteins, etc. Some bilateral effusions are exudates, however, and this is seen with metastatic disease, lymphoma, pulmonary embolism, rheumatoid disease, systemic lupus erythematosus (SLE), post-cardiac injury syndrome, myxoedema and some ascites-related effusions. Right-sided effusions are typically associated with ascites, heart failure and liver abscess,

Imaging pleural effusion5 Chest radiograph

Figure 13.6 Bilateral pleural effusion. (A) Erect and (B) supine chest radiograph. The pleural effusion obscures the diaphragm and both costophrenic angles. It has a curvilinear upper margin concave to lung and is higher laterally than medially. This is opposite to the findings on the supine chest radiograph where the pleural effusion is hardly visible as a hazy opacity affecting the lower part of the thorax. Note also that the costophrenic angles are not obscured and that the vascular opacities are preserved in the overlying lung.

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.7 Massive pleural effusion with mediastinal shift to the left. (A) Chest radiograph and (B) CT coronal reconstruction. A massive effusion displaces the mediastinum to the left. CT shows the important pleural effusion together with the enhanced atelectatic left lung. Note also the depression of the right hemidiaphragm (arrows).

malignancy, such as may be seen with mesothelioma or metastatic carcinoma (Table 13.1). Large effusions sometimes cause diaphragmatic inversion, particularly on the left where the diaphragm lacks the support of the liver7. Although pleural fluid collects initially under the lung, it is unusual for it to remain localized in this site once its volume exceeds 200–300 ml.This does happen occasionally, however, and may be suspected from an erect PA and lateral radiograph. On a PA radiograph this subpulmonary effusion7 presents as a ‘high hemidiaphragm’ with an unusual contour that peaks more laterally than usual, has a straight medial segment and falls away rapidly to the costophrenic angle laterally, which may or may not be blunted. Ultrasound or CT will confirm the diagnosis (see Fig. 13.20). Loculated (encysted, encapsulated) pleural fluid Fluid can loculate between visceral pleural layers in fissures or between visceral and parietal layers, usually against the chest wall. It is unusual for this to happen without some additional radiographic clue to the presence of pleural disease (Fig. 13.8). Both ultrasound and CT can be used to distinguish loculated fluid from solid lesions.

Table 13.1

CAUSES OF OPACIFICATION OF A HEMITHORAX

Pleural effusion Consolidation Collapse Massive tumour Fibrothorax Combination of above lesions Pneumonectomy Lung agenesis

Pleural effusion in the supine patient In the supine patient, pleural fluid layers out posteriorly and the meniscus effect, present from front to back, is not appreciated because of the projection. The main radiographic finding is a hazy opacity like a veil affecting the whole or the lower part of the hemithorax, with preserved vascular opacities in the overlying lung (see Fig. 13.6B). Additional signs include haziness of the diaphragmatic margin, blunting of the costophrenic angle, a pleural cap to the lung apex, thickening of the minor fissure and widening of the paraspinal interface.

Ultrasound6,8 Pleural fluid, especially when it is a transudate, is commonly echo-free and marginated on its deep aspect by a highly echogenic line at the fluid–lung interface. Exudative and haemorrhagic effusions may be echogenic and are often accompanied by pleural thickening. The pattern of echoes may be homogeneous, complex or septated. Features that help distinguish a fluid from a solid echogenic lesion include changes in shape with breathing, the presence of septa and fibrous strands, and movement of components induced by breathing (Fig. 13.9). Occasionally, in the absence of such features, some echogenic fluid pleural effusions are indistinguishable from solid ones.6 Ultrasound has a number of important roles in the evaluation and management of pleural fluid. It can be used to distinguish between pleural fluid, solid pleural (or extrapleural) lesions, and peripheral lung lesions. In peripheral lung lesions, the presence of fluid bronchograms and vessels on Doppler examination will positively identify consolidation. In addition, pleural lesions characteristically make an obtuse angle with the chest wall, whereas with intrapulmonary lesions the angle is often acute.This ability of ultrasound to distinguish pulmonary lesions (collapse, consolidation, abscess) from pleural effusion

225

226

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.8 Encapsulated fluid on (A) PA and (B) lateral chest radiographs. Pleural fluid is encapsulated in the major fissure and against the anterior chest wall. These encysted fluid collections can mimic a lung tumour.

extent and location of the latter. Accurate localization of such loculated effusions is useful before drainage. CT distinguishes between parenchymal lung disease and pleural disease, a distinction that is often facilitated by a bolus of intravenous contrast medium. CT can characterize the morphology of pleural thickening that often accompanies a pleural effusion, distinguishing between malignant thickening (nodular, with focal masses) and benign thickening, which is typically uniform. CT can also identify any underlying lung disease that might have provoked an effusion and it facilitates percutaneous aspiration and biopsy (Fig. 13.10).

Figure 13.9 Ultrasound of an empyema. The pleural fluid is separated by septa (arrows). Although the pleural fluid is echo-free in part, some areas return echoes owing to the turbid nature of the empyema fluid.

is particularly useful when it comes to the evaluation of the opaque hemithorax. Ultrasound can also be used to identify small amounts of pleural fluid, or pleural fluid in unusual locations, as with a subpulmonary effusion (see Fig. 13.20). Ultrasound is widely used to localize pleural fluid for aspiration and identify any solid components to allow guided biopsy. Furthermore, ultrasound may identify the cause of an effusion when it lies inside or even outside the chest (e.g. subphrenic abscess, metastasis).

Computed tomography6,9 CT is very sensitive in detecting pleural fluid and can distinguish between free and loculated fluid, identifying the

Figure 13.10 CT of malignant pleural disease. In this right pleural effusion CT identifies the extensive and irregular pleural thickening characteristic of a malignant process (pleural metastases). Note also the primary tumour in the right breast.

CHAPTER 13

A pleural effusion appears on CT as a dependent sickleshaped opacity with a lower CT number than that of any adjacent pleural thickening or mass. CT numbers do not allow a distinction between transudate and exudate. However, parietal pleural thickening at contrast-enhanced CT almost always indicates the presence of pleural exudates. The higher density of clotted blood in a haemothorax is sometimes apparent.The fat-containing chylothorax does not have a lower CT number than normal, because of its protein content. Loculated effusions have a lenticular configuration with smooth margins and they displace the adjacent parenchyma.

Magnetic resonance imaging6 MRI has a limited role in the evaluation of pleural effusion. Pleural fluid has a low signal on T1-weighted sequences and a high signal on T2-weighted images, with a tendency for exudates to give a higher signal than transudates on T2-weighted sequences. In addition, complex exudates have greater signal intensity than simple exudates. It may also be possible to differentiate transudates from exudates using triple echo-pulse sequence, and benign from malignant changes using high-resolution MRI.10 Chylous effusion can cause high signal intensity on T1-weighted images similar to subcutaneous fat. In the subacute and chronic stage, haematomas show bright signal intensity on T1-weighted images, surrounded by a dark rim caused by haemosiderin.

Some specific pleural effusions Exudates and transudates Pleural effusion is common in heart failure and tends to be more frequent and larger on the right. All types of pericardial disease may be associated with pleural effusion which is predominantly left sided. Pleural effusion is a characteristic finding in the post-cardiac injury syndrome, seen in about 80% of patients. It may be bilateral or unilateral and is commonly accompanied by consolidation and pericardial effusion. Pulmonary embolism is commonly associated with pleural effusion which is seen in 25–50% of cases. A number of drugs have been described as causing pleural effusions. The most common agents are cytotoxics (methotrexate, procarbazine, mitomycin, busulphan, bleomycin and interleukin-2), nitrofurantoin, antimigraine drugs (ergotamine, methysergide), amiodarone, propylthiouracil, bromocriptine and gonadotropins. With a number of these agents pleural thickening is more common than a pleural effusion. Pleural effusion is also a recognized complication of hepatic cirrhosis. The principal mechanism of its production is the transdiaphragmatic passage of ascites, though other factors such as hypoalbuminaemia may contribute in a small number of cases. Both acute and chronic pancreatitis are associated with pleural effusions which have high amylase levels. In acute pancreatitis, exudative and often blood-stained effusions form in 15% of patients, particularly on the left side where the diaphragm is closely related to the pancreatic tail. Associated elevation of the hemidiaphragm and basal lung consolidation are common. In chronic pancreatitis, effusions tend to be large and recurrent and patients present with dyspnoea, unlike effusions in acute

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

pancreatitis in which abdominal symptoms predominate. The pathogenesis of pleural effusion in chronic pancreatitis is fistula formation following ductal rupture. Pleural effusion is common with subphrenic abscess and occurs in about 80% of patients. The effusion is often accompanied by basal lung collapse and consolidation, an elevated hemidiaphragm and a subdiaphragmatic air–fluid level. Pleural effusion may occur in a number of renal conditions. Exudative effusions may be seen in uraemia and are often accompanied by pericarditis. Effusions can be large or small and are often unilateral, behaving in a rather indolent fashion. In common with other hypoproteinaemic states, bilateral effusions develop in about 20% of patients with nephrotic syndrome. Peritoneal dialysis can produce pleural effusions by the direct transdiaphragmatic passage of fluid, as occurs with cirrhotic ascites. In common with other ascites-related effusions they are predominantly right sided, but these effusions have a diagnostically high level of glucose. Patients with acquired immune deficiency syndrome are at risk for a variety of pleural infections and neoplasms that can be associated with pleural effusion. These effusions are most frequently caused by pneumonic infections but can also be the result of non-Hodgkin’s lymphoma. Empyema is a suppurative exudate usually parapneumonic. Less commonly it is caused by transdiaphragmatic extension of a liver abscess or by bronchopleural fistula (Fig. 13.11).

Bronchopleural fistula Bronchopleural fistula differs from a pneumothorax in that the communication with the pleural space is via airways rather than distal air spaces. It occurs in two main settings, following partial or complete lung resection and in association with necrotizing infections.

Figure 13.11 Empyema. An enhanced CT shows a fluid collection in the right pleural space. The pleura is thickened but smooth and enhancing. The empyema followed pneumonia. Soft tissue medially is collapsed and consolidated lung. Note the oedema of the extrapleural fat.

227

228

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Chylothorax Chylous effusions are commonly milky because they contain triglycerides in the form of chylomicrons. Chylous and non-chylous pleural effusions are indistinguishable on the chest radiograph. In addition, despite its high fat content, the increased protein level of a chylothorax gives it an attenuation on CT similar to that of other pleural effusions. Chylous effusion can cause high signal intensity on T1-weighted images similar to subcutaneous fat. Chyle collects in the pleural space following rupture of the thoracic duct or seepage from collaterals. Rarely, it crosses the diaphragm from the abdomen in the presence of chylous ascites.

Haemothorax On the plain chest radiograph an acute haemothorax is indistinguishable from other pleural fluid collections. Once the blood clots there is a tendency for loculation and occasionally a fibrin body will form. Pleural thickening and calcification are recognized sequelae. On CT a haemothorax may show areas of hyperdensity, and in the subacute or chronic stage it will appear on MRI as a high signal on T1- and T2-weighted images, possibly with a low signal rim caused by haemosiderin. The most common cause of haemothorax is trauma, but it is seen in a number of other conditions, including ruptured aortic aneurysm, pneumothorax, extramedullary haemopoiesis and coagulopathies.

Table 13.2

CAUSES OF ADULT PNEUMOTHORAX

Spontaneous, primary Spontaneous, secondary Airflow obstruction

Asthma Chronic obstructive pulmonary disease Cystic fibrosis

Pulmonary infection

Cavitary pneumonia Tuberculosis Fungal disease AIDS Pneumatocoele

Pulmonary infarction Neoplasm Diffuse lung disease

Metastatic sarcoma Histiocytosis X Lymphangioleiomyomatosis Fibrosing alveolitis Other diffuse fibroses

Hereditable disorders of fibrous connective tissue

Marfan’s syndrome

Endometriosis (catamenial pneumothorax)

Traumatic, noniatrogenic Ruptured oesophagus/trachea Closed chest trauma (± rib fracture) Penetrating chest trauma

PNEUMOTHORAX

Traumatic, iatrogenic Thoracotomy/thoracocentesis

Air in the pleural space is a pneumothorax. When air and liquid are present the nomenclature depends on their relative volumes and the type of liquid. Small amounts of liquid are disregarded and the condition is still called a pneumothorax; otherwise the prefix hydro-, haemo-, pyo- or chylo- is added, depending on the nature of the liquid.

Primary spontaneous pneumothorax Iatrogenic causes apart, the most common type of pneumothorax in the adult is the so-called primary spontaneous pneumothorax (PSP). A pneumothorax occurring without an obvious precipitating event is spontaneous, and if the patient has essentially normal lungs it is in addition primary. PSP occurs predominantly in young adults (65% are between 20 and 40 years of age) and it is five times more common in men than women. Untreated, at least one-third of patients will have a recurrence, most commonly within a few years and on the ipsilateral side. PSP is nearly always caused by the rupture of an apical pleural bleb. Although not detectable on interval chest radiographs, one taken at the time of the pneumothorax will show one or more blebs projecting from the apical lung margin in 20% of patients; such abnormal apical airspaces are much more commonly shown by interval CT.11

Secondary spontaneous pneumothorax A large number of conditions predispose to pneumothorax (Table 13.2). In a number of these disorders pneumothorax occurs frequently.

Percutaneous biopsy Tracheostomy Central venous catheterization

Diagnosis The diagnosis of pneumothorax is made with the chest radiograph, which also detects complications and predisposing conditions and helps in management5 (Fig. 13.12).

Typical signs These are seen on erect radiographs in which the pleural air rises to the lung apex. Under these conditions the visceral pleural line at the apex becomes separated from the chest wall by a transradiant zone devoid of vessels. Though this sounds a straightforward sign to assess, difficulties of interpretation can arise with avascular lung apices, as in bullous disease and when linear shadows are created by clothing or dressing artefacts, tubes and skin folds. Skin folds cause problems particularly in neonates and in old people radiographed slumped against a cassette in the AP projection (Fig. 13.13) Features that help identify artefacts and skin folds include extension of the ‘pneumothorax’ line beyond the margin of the chest cavity, laterally located vessels, and an orientation of a line that is inconsistent with the edge of a slightly collapsed lung. In addition, the margin of skin folds tends to be much wider than the normally thin visceral pleural line. In indeterminate circumstances a repeat

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.13 Skin folds mimicking a right pneumothorax (arrows). The laterally located blood vessels, the wide margin of the lines, and the orientation of the lines that is inconsistent with the edge of a slightly collapsed lung help to differentiate them from a real pneumothorax.

chest wall at the apex or laterally. Signs that suggest a pneumothorax under these conditions are12,13 (Fig. 13.14): • ipsilateral transradiancy, either generalized or hypochondrial • a deep, finger-like costophrenic sulcus laterally • a visible anterior costophrenic recess seen as an oblique line or interface in the hypochondrium; when the recess is manifest as an interface it mimics the adjacent diaphragm (‘double diaphragm sign’) • a transradiant band parallel to the diaphragm and/or mediastinum with undue clarity of the mediastinal border • visualization of the undersurface of the heart, and of the cardiac fat pads as rounded opacities suggesting masses • diaphragm depression.

Figure 13.12 Left primary spontaneous pneumothorax. Chest radiograph (A) at deep inspiration and (B) at deep expiration. The left lung has partially collapsed and an area of extreme low density without vascular markings becomes visible. The pneumothorax is accentuated on the chest radiograph at suspended deep expiration (B).

chest radiograph, an expiratory radiograph (see Fig. 13.12B) or one taken with the patient decubitus may clarify the situation. Should doubt still remain, then CT is particularly helpful in distinguishing between bullae and a pneumothorax.

In a patient who cannot stand, the presence of a pneumothorax can be confirmed with a lateral decubitus view or a supine decubitus projection with the cassette placed dorsolaterally at 45 degrees and the X-ray tube angled perpendicular to the cassette. When the pleural space is partly obliterated a pneumothorax may be loculated, and must be differentiated from other localized transradiancies. These include cysts, bullae, pneumatocoeles, pneumomediastinum and local emphysema. These cannot always be differentiated by plain radiographs, but can be by CT.

Atypical signs

Complications Haemopneumothorax

These arise when the patient is supine or the pleural space partly obliterated. In the supine position, pleural air rises and collects anteriorly, particularly medially and basally, and may not extend far enough posteriorly to separate lung from the

This is a common complication of traumatic pneumothorax. Small amounts of serous or bloody fluid may also occur with a spontaneous pneumothorax but only 2% of individuals develop a clinically significant haemothorax in these circumstances.

229

230

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.14 Supine pneumothorax. Portable chest radiograph (A) before and (B) immediately after development of a pneumothorax. There is an increase of transradiancy at the left lung base, the costophrenic sulcus laterally is more pronounced, and the diaphragm is slightly depressed.

Blood may clot in the pleural space, producing a mass which can mimic a pleural tumour.

Tension pneumothorax This life-threatening complication is present when intrapleural pressure becomes positive relative to atmospheric pressure for a significant part of the respiratory cycle. Tension has an adverse effect on gas exchange and cardiovascular performance, causing a rapid deterioration in the patient’s clinical condition. The diagnosis is usually made clinically and treatment instituted without a radiograph. Should a chest radiograph be taken, it will show contralateral mediastinal shift and ipsilateral diaphragm depression. Mild degrees of contralateral mediastinal shift are not unusual with a nontension pneumothorax because of the negative pressure in the normal pleural space. Moderate or gross mediastinal shift, however, should be taken as indicating tension, particularly if the ipsilateral hemidiaphragm is depressed.This latter sign is the more reliable and is almost invariably present with significant tension.

Pyopneumothorax This unusual complication is seen most commonly following necrotizing pneumonia or oesophageal perforation.

Adhesions These generate straight band shadows extending from the lung margin to the chest wall. They limit collapse but at the same time may account for continued air leakage from the lung surface, and if they tear they may bleed. They can be identified with CT.

Re-expansion oedema This unusual complication is sometimes seen following the rapid therapeutic re-expansion of a lung that has been markedly collapsed for several days or more. Oedema comes on within hours of drainage, may progress for a day or two and clears within a week. It usually causes only mild morbidity.

PLEURAL THICKENING AND FIBROTHORAX5 Pleural thickening is common and usually represents the organized end stage of various active processes such as infective and noninfective inflammation (including asbestos exposure and pneumothorax) and haemothorax. When generalized and gross, it is termed a fibrothorax and may cause significant ventilatory impairment. Radiologically, pleural thickening gives fixed shadowing of water density, most commonly located in the dependent parts of the pleural cavity. Viewed en profile, it appears as a band of soft tissue density up to approximately 10-mm thick, more or less parallel to the chest wall and with a sharp lung interface. En face, it causes ill-defined, veil-like shadowing. Blunting of the costophrenic angle, often with tenting of the diaphragm, is a common finding. On ultrasound, benign pleural thickening produces an homogeneous echogenic layer just inside the chest wall. It is not reliably detected unless it is 1 cm or more thick. CT on the other hand is very sensitive at detecting pleural thickening, which is most easily assessed on the inside of the ribs, where there should normally be no

CHAPTER 13

soft tissue opacity. In chronic conditions pleural thickening is commonly accompanied by thickening of the normally inconspicuous fatty layer that lies immediately outside the parietal pleura, a feature that can be appreciated on CT. Fibrous pleural thickening is common in the apical pleural cupola. This may be secondary to tuberculosis or represent apical cap. Caps are age-related changes of unknown aetiology. Sometimes they have a scalloped contour or are associated with a tenting towards the lung. They are as commonly unilateral as bilateral. Caps should be distinguished from the companion shadows of the upper ribs, from extrapleural linear fat deposition and most importantly from a Pancoast’s tumour. Companion shadows of the ribs are usually smoothly bordered towards the lung apex while extrapleural fat is usually bilateral, symmetrical, and also located along the lateral chest wall. Caps may be indistinguishable from Pancoast’s tumour on the chest radiograph. In case of doubt CT or MR should be performed. Fibrous pleural thickening can be induced by asbestos exposure14 (Fig. 13.15). This thickening can be diffuse or is more often multifocal. These pleural plaques can undergo

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

hyaline transformation, calcify, or ossify. They are most commonly found along the lower thorax and on the diaphragmatic pleura. On CT, they appear as circumscribed areas of pleural thickening separated from the underlying rib and extrapleural soft tissues by a thin layer of fat. Because of their higher density they can easily be differentiated from circumscribed increase of extrapleural fat, as sometimes seen in obese patients. Diffuse pleural thickening is also a manifestation of asbestos exposure. The radiographic definition of diffuse pleural thickening or fibrothorax is somewhat arbitrary. It has been suggested to consider as fibrothorax a smooth uninterrupted pleural density that extends over at least one-quarter of the chest wall. On CT fibrothorax has been defined as a pleural thickening which extends more than 8 cm in the craniocaudal direction, 5 cm laterally and with a thickness of more than 3 mm. Common causes of fibrothorax are empyema, tuberculosis and haemorrhagic effusion. Asbestos exposure-related fibrothorax is less common than pleural plaques and is usually the sequel of a benign exudative effusion. CT may be helpful

Figure 13.15 (A–D) Pleural plaques caused by asbestos exposure. Pleural plaques are most commonly found along the lower thorax, on the diaphragmatic pleura and, when involvement is extensive, also along the lateral and anterior thorax (arrows). They can partially or completely calcify or ossify.

231

232

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

to find the aetiology of the fibrothorax. Extensive calcification favours previous tuberculosis or empyema5 (Fig. 13.16). Asbestos exposure-related fibrothorax is usually bilateral and rarely calcified. Generalized, postinflammatory pleural thickening must be distinguished from diffuse pleural malignancy caused by mesothelioma, metastatic disease (particularly adenocarcinoma), lymphoma and leukaemia. Mesothelioma and adenocarcinoma cause diffuse pleural thickening which is often lobulated, and may surround the whole lung and extend into and along fissures. These features are frequently obscured by an effusion. The most useful signs on CT that indicate malignant as opposed to benign pleural thickening are circumferential thickening, nodularity, parietal thickening of more than 1 cm, and involvement of the mediastinal pleura5 (see Figs 13.18, 13.19). MRI signal intensity seems to be a valuable additional feature for differentiating benign from malignant disease, especially since MRI is often able to demonstrate the tumour extension into the chest wall. Signal hypointensity with long TR sequences is a reliable predictive sign of benign pleural disease15.

PLEURAL CALCIFICATION Pleural calcification is most commonly seen following asbestos exposure, empyema (usually tuberculous) and haemothorax (Fig. 13.16). In the last two conditions, calcification is irregular, resembles a plaque or sheet, and is contained within thickened pleura. It may occur anywhere but is most common in the lower posterior half of the chest and is usually unilateral, unlike that found in silicosis, particularly of the asbestos-related type, where calcification occurs as more discrete collections within plaques and is usually bilateral.

PLEURAL TUMOURS Localized pleural tumours16 These are relatively uncommon, the most common being a localized fibrous tumour (localized mesothelioma) (Fig.

13.17). These lesions most commonly present in middle age, about half the patients being asymptomatic. Hypertropic osteoarthropathy is a well-recognized complication (10–30% of patients) and uncommonly the tumour produces hypoglycaemia. Microscopically two-thirds are benign and one-third is malignant.The plain radiographic findings are of a pleurallybased, well-demarcated, rounded and often slightly lobulated mass (2–20-cm diameter) which may, because of pedunculation, show marked positional variation with changes in posture and respiration. Pleural fibromas usually make an obtuse angle with the chest wall and may reach enormous sizes. Occasionally they may arise in a fissure. CT findings are similar to those observed on plain radiography: a mobile mass, often heterogeneous because of necrosis, haemorrhage, frequently enhancing after contrast medium administration, and rarely calcified. Malignant types are usually larger than 10 cm and may invade the chest wall. Typically these tumours show low signal intensity on both T1- and T2-weighted images, although tumours with intermediate to high signal intensity have been described.17 Lipomas are asymptomatic benign tumours that are usually discovered incidentally on chest radiographs as sharply defined pleural masses. Diagnosis is easy with CT because this examination can delineate the pleural origin and the fatty composition. This fatty density is homogeneous. When heterogeneous and when also soft tissue attenuation components are found, a liposarcoma should be suspected. Pleural lipomas have high signal intensity on T1-weighted images. On T2-weighted images signal is moderately bright. Diagnosis of pleural extension of bronchogenic carcinoma on a chest radiograph is very difficult.The only reliable indicator is rib destruction.With CT and MRI also diagnosis can be difficult. Features such as a large contact (> 3 cm) between the mass and the pleura, an obtuse angle between the tumour and the chest wall, an associated pleural thickening and the presence of pleural tags usually considered as signs of chest wall invasion also occur in benign lesions. The accuracy of CT can be increased by performing 2D and 3D reconstructions.

Figure 13.16 Pleural calcification. (A,B) On the chest radiograph an extensive sheet-like calcification of the right pleura with additional pleural thickening (old tuberculous empyema) is seen. (C) CT demonstrates the extent and thickness of the pleural calcification (arrow).

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.17 Benign pleural fibroma. (A) Frontal and (B) lateral radiographs show a small well-demarcated, homogeneous, slightly lobulated mass (arrows). (C) CT shows that the mass is pleural based, sharply defined, and slightly enhancing.

In cases where tumour invasion is obvious, 2D sagittal or coronal reconstructions can be helpful in ascertaining the extent of the mass. MRI has a slight advantage over CT in the evaluation of chest wall and pleura invasion. Before spiral CT, MRI was considered better for studying superior sulcus tumours and their extension to the chest wall (see Fig. 13.3). However, studies have shown that spiral CT and MRI showed comparable sensitivity but that spiral CT had higher specificity. CT is superior in the detection of pleural calcifications and osseous destruction (see Fig. 13.3D). MDCT can also be used in selected cases to clarify a complex relationship between tumour invading the chest wall and vascular structures of the thoracic inlet. Pleural metastases are the most common pleural neoplasms. They are usually an adenocarcinoma with sites of origin including the ovary, stomach, breast and lung. Pleural metastatic disease can present as a solitary mass but more often multiple pleural locations are seen (Fig. 13.18). Pleural metastases are very often accompanied by pleural effusion, which can be the only finding on a chest radiograph. CT, MRI and ultrasound are more sensitive to demonstrate pleural metastasis as the cause of the pleural effusion.9

On CT malignant mesothelioma presents as a nodular soft tissue mass sometimes with hypodense areas corresponding with necrosis. Metastatic enlargement of hilar and mediastinal nodes is seen in up to 50% of patients. Malignant mesothelioma has a minimally increased signal on T1 and a moderately increased signal on T2. MRI may be superior to CT in determining extent of disease because it allows better evaluation of the relationship of the tumour to the structures of the chest wall, mediastinum and diaphragm. However,

Diffuse pleural tumours Diffuse tumoural thickening of the pleura can be caused by malignant mesothelioma or by pleural metastasis. Both entities are usually indistinguishable with imaging. Diffuse malignant mesothelioma is a rare primary neoplasm and its development is strongly related to asbestos exposure. It presents on a chest radiograph as an irregular and nodular pleural thickening with or without associated pleural effusion. Tumour extension into the interlobular fissures, accompanying pleural effusion, and invasion into the chest wall are better appreciated with CT (Fig. 13.19).

Figure 13.18 Malignant pleural thickening caused by metastatic disease. Malignant pleural thickening was caused by pleural metastases. Note the compression on the right hemidiaphragm and the extension of the tumour into the liver (arrows).

233

234

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

lesion is adjacent to major cardiovascular structures, such as the aorta.21

Ultrasound Ultrasound is well suited for interventional procedures in the pleura. Because of the development of high-resolution, highfrequency probes with special biopsy ports, ultrasound guided biopsy of small pleural lesions has become possible22. Ultrasound is particularly indicated to guide percutaneous aspiration and catheter drainage of a pleural fluid or air collection, even in small amounts. Advantages of this technique include real-time visualization during needle placement, absence of ionizing radiation, and in the case of biopsy of a mass, the ability to target non-necrotic portions for sampling20. In addition, ultrasound is a safe and convenient method of guiding interventional procedures at the bedside of the patient and obviates the need to transport patients on life support devices to the radiology department. A disadvantage is that ultrasound is limited by attenuation of the beam as it transverses air-filled lung or pleura.

Computed tomography

Figure 13.19 Malignant mesothelioma. (A, B: axial and coronal CT) Diffuse lobulated and nodular thickening of the pleura with tumour extension into the lobar fissure (arrows). Note the metastatic enlargement of some hilar and mediastinal lymph nodes.

in most cases CT and MRI provide similar information. Ultrasound may be a supplementary method for biopsy and surgery planning.18,19

Pleural interventions Fluoroscopy Uni- or bi-planar fluoroscopy was the first imaging technique used to guide percutaneous pleural interventions. The technique is widely available, allows real-time control of the procedure, and gives an overview of the thorax. In addition, the technique is familiar to most investigators.20 However, fluoroscopy is not suitable for every lesion. Small lesions may be difficult or impossible to identify. Some lesions may be superimposed on or not separable from normal thoracic structures. Another important limitation is that biopsy or drainage using fluoroscopic guidance may not be advisable if the

Major advantages of CT over fluoroscopy are its multiplanar capabilities and its exquisite anatomical detail. CT is particularly useful for sampling lesions visible in only a single radiographic projection or when great imaging detail is required for the interventional procedure. The administration of intravenous contrast medium is mandatory for the identification of tissue necrosis, fluid content, and identification of normal and abnormal vascular structures. CT allows for determination of an optimal cutaneous entry point for the biopsy needle or for tube placement. Disadvantages of CT guided interventional procedures include greater patient discomfort lying on the CT table and greater expense than with fluoroscopically guided biopsies. A disadvantage compared to ultrasound is the fact that this technique also requires ionizing radiation. However, the introduction of CT continuous imaging, also called CT fluoroscopy, has improved the ease of performing interventional thoracic procedures because it allows real-time visualization of the lesion and of the progression of the needle or tube23. Compared with conventional spiral CT, there is also a markedly decreased patient radiation dose because the procedure can be shortened.

Magnetic resonance imaging Although MRI is often used for guidance of interventional procedures, little experience has been gained in thoracic or pleural interventions24. This technique combines the absence of ionizing radiation with good anatomical detail and has become possible with the introduction of non-ferromagnetic MRI compatible biopsy needles. Major disadvantages, however, include high cost, limited availability, the length of the procedure and the lack of real-time control.

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

THE DIAPHRAGM The diaphragm is only seen because there is air-containing lung adjacent to it superiorly. It is 2–3-mm thick, but this will only be appreciated if there is air immediately beneath it, as with a pneumoperitoneum. Localized loss of clarity occurs when the diaphragm is not tangential to the X-ray beam, but usually indicates adjacent pulmonary or pleural disease, e.g. the costophrenic or costovertebral angles are obliterated by pleural fluid, and much of the diaphragmatic outline may be obliterated by basal pneumonia. Each hemidiaphragm is normally represented on the PA radiograph by a smooth, curved line which is convex upwards. The lateral attachment of the diaphragm to the ribs is represented by the lateral costophrenic recess, a sharply-defined acute angle. When the diaphragm is flat, as in emphysema, the most lateral muscle slips extend slightly upwards and may be seen as digitations. The costophrenic angle then becomes less acute, or even obtuse, and the appearance may simulate a small pleural effusion. Medially, the diaphragm meets the heart at the cardiophrenic angle. This is higher than the costophrenic angle and unlike the latter is often ill-defined owing to the presence of fat. On the right, this may simulate disease in the middle lobe, and on the left, disease in the lower lobe or lingula. Prominent fat pads at the cardiophrenic angles are an occasional cause of overestimation of the transverse cardiac diameter, particularly if the film is underexposed. On correctly exposed radiographs, the relatively low radio-opacity of the fat pad enables it to be distinguished from the cardiac apex. On the lateral radiograph each dome makes an acute angle with the ribs posteriorly to form the posterior costophrenic recesses. The latter lie considerably lower than the highest part of each leaf—a point of great importance, as localized pulmonary or pleural disease adjacent to the posterior aspect of the diaphragm will often not be recognized on the PA radiograph, on which only the highest anterior portion of the diaphragmatic dome is represented. The right hemidiaphragm makes an upward curve as it extends anteriorly to the sternum. This part of its attachment is often poorly defined because of adjacent fat. Localization of disease requires the correct identification of each leaf on the lateral radiographs. The left diaphragm is obscured anteriorly by the heart and usually has an air-distended gastric fundus beneath it; whichever leaf is nearer the film is related to the ribs least magnified by the diverging beam.

Height In most people the diaphragm in the mid lung field lies at the level of the fifth or sixth anterior rib interspace. It may lie at a lower level in normal young individuals, particularly those of an asthenic build and at a slightly higher level in the obese, the elderly and young infants. In over 90% of normal people the right hemidiaphragm is higher than the left. This difference in height on the PA film is usually about 15 mm, but may be as much as 30 mm. Depression of the diaphragm occurs in emphysema and in acute severe asthma, but flattening only occurs in emphysema.

Inversion of the diaphragm is sometimes seen with a tension pneumothorax and with large basal bullae. It is also a common accompaniment of pleural effusions.Table 13.3 shows the most common causes of bilateral symmetrical elevation of the diaphragm. Elevation of a single hemidiaphragm is usually secondary to adjacent pleural, pulmonary or subphrenic disease, or to phrenic nerve palsy (Table 13.4). A minor degree of diaphragmatic elevation is a common accompaniment of pleurisy, lower lobe pneumonia and pulmonary thromboembolism. In the latter there may be no visible change in the affected lung. Upper abdominal inflammatory processes and rib fractures may also cause a high diaphragm. A high hemidiaphragm may be mimicked by a subpulmonary pleural effusion (Fig. 13.20), a large well-defined tumour adjacent to the dome, or by combined middle and lower lobe collapse.

Eventration In eventration a part of the normal diaphragmatic muscle is replaced by a thin layer of connective tissue and a few scattered muscle fibres25. The unbroken continuity differentiates it

Table 13.3 CAUSES OF BILATERAL SYMMETRICAL ELEVATION OF THE DIAPHRAGM Supine position Poor inspiration Obesity Pregnancy Abdominal distension (ascites, intestinal obstruction, abdominal mass) Diffuse pulmonary fibrosis Lymphangitis carcinomatosa Disseminated lupus erythematosus Bilateral basal pulmonary emboli Painful conditions (after abdominal surgery) Bilateral diaphragmatic paralysis

Table 13.4 CAUSES OF UNILATERAL ELEVATION OF THE DIAPHRAGM Posture—lateral decubitus position (dependent side) Gaseous distension of stomach or colon Dorsal scoliosis Pulmonary hypoplasia Pulmonary collapse Phrenic nerve palsy Eventration Pneumonia or pleurisy Pulmonary thromboembolism Rib fracture and other painful conditions Subphrenic infection Subphrenic mass

235

236

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.20 Subpulmonary pleural effusion. On the (A) erect PA and (B) lateral radiograph the effusion simulates a high hemidiaphragm. (C) Ultrasound and (D) CT clearly show that the effusion is located above the diaphragm. Arrows = diaphragmatic area.

from diaphragmatic hernia. Some authors consider eventration to be a congenital anomaly resulting from failure of muscularization of part or all of the diaphragmatic leaf. Most authors, however, also include within the definition elevation occurring as a result of acquired paralysis with atrophy of the diaphragmatic muscle; an inclusion justified by the fact that many adults with surgically proven eventration have previously had normal chest radiographs. Total eventration shows a marked left-sided predominance, for which there is no acceptable explanation. Although eventration is a recognized cause of respiratory distress in the newborn, it is not usually associated with symptoms in the adult. Localized forms of the condition are relatively common, particularly in the elderly, and predominantly affect the right hemidiaphragm at its anteromedial aspect (Fig. 13.21). The distinction between a localized eventration and a small diaphragmatic hernia or a mass arising from the lung,

pleura, or diaphragm is best made using CT or MRI. The various causes of focal elevation or bulging of a diaphragm are given in Table 13.5.

Movement and paralysis Unequal excursion of the two hemidiaphragms occurs in approximately 80% of normal people. However, this inequality of diaphragmatic excursion is less than 10 mm in most people. While normal young adults can move the diaphragm over at least 30 mm, this range is greatly reduced in the elderly. As the chest radiograph is exposed at the end of a full inspiration, any severe unilateral limitation of diaphragmatic movement will be apparent on this static examination. Diaphragmatic movement is, however, better assessed by fluoroscopy, which should, ideally, be performed in both the AP and lateral projections with the patient erect and supine. The

CHAPTER 13

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

Figure 13.21 Focal eventration. (A) PA chest radiograph reveals a soft tissue opacity arising from the diaphragm. (B) CT shows the presence of fat and liver under the elevated part of the diaphragm.

latter position is useful as the range of movement is usually greater than it is in the erect position. With the patient in the lateral position, any inequality of movement of the two leaves is readily assessed and localized restriction of movement identified better26. Restriction of diaphragmatic movement occurs secondary to disease of the phrenic nerve and secondary to inflammatory and painful conditions adjacent to the diaphragm, such as lower lobe pneumonia and subphrenic infection. Phrenic palsy is most commonly secondary to involvement of the phrenic nerve by tumour—usually a bronchial carcinoma. Phrenic nerve paresis may be caused by trauma (road accidents, birth injury, brachial plexus block and phrenic crush), irradiation and a variety of neurological conditions such as poliomyelitis, herpes zoster, and cervical disc degeneration. The recognition of phrenic paresis depends upon finding a high hemidiaphragm which exhibits absent, restricted, or paradoxical movement.The latter is particularly well demonstrated by sniffing. Diaphragmatic motion can also be examined with ultrasound26. Especially in patients who cannot come to the fluoroscopy room, bedside ultrasound is very useful. This technique has also a high accuracy to discover absent and paradoxical diaphragmatic

Table 13.5 CAUSES OF FOCAL ELEVATION (BULGE) OF THE DIAPHRAGM Partial eventration Diaphragmatic hernia Diaphragmatic tumour Pleural tumour Pulmonary tumour Focal diaphragmatic dysfunction Focal diaphragmatic adhesions

motion. In addition measurement of diaphragmatic thickness can be helpful to confirm diaphragmatic paralysis, since a paralysed diaphragm does not thicken during inspiration. An important mimic of phrenic paresis is eventration (usually left sided, see above). In a significant small number of patients in whom there is little doubt that a phrenic paresis exists, no cause can be discovered. In this ‘idiopathic’ group the right leaf is more commonly affected than the left and it has been suggested that the palsy may be a legacy of previous viral neuritis. Weakness or paralysis of both hemidiaphragms is most commonly seen in association with chronic neuromuscular disease and causes severe clinical disability. Bilateral paralysis may not be recognized by fluoroscopic examination, for passive descent of the diaphragm may occur with inspiration.

Diaphragmatic hernias Intrathoracic herniation of abdominal contents occurs through congenital defects in the muscle, through traumatic tears or, most commonly, through acquired areas of weakness at the central oesophageal hiatus. Congenital hernias presenting in childhood are discussed elsewhere. When the defect is small it may not come to attention until adulthood, when it usually presents as an incidental abnormality on the chest radiograph. Bochdalek defects through the pleuroperitoneal canal occur along the posterior aspect of the diaphragm and the hernia usually contains retroperitoneal fat or a portion of kidney or spleen27 (Fig. 13.22). The majority occur on the left. A well-defined, dome-shaped, soft tissue opacity is seen midway between the spine and lateral chest wall on the frontal view and above the posterior costophrenic recess on the lateral view. It may appear to ‘come and go’ on serial PA radiographs because of varying degrees of inspiration and differences in transdiaphragmatic pressure. It has been shown that asymptomatic small Bochdalek hernias are present in 6% of otherwise

237

238

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.22 Bochdalek hernia. (A) Lateral chest radiograph shows a focal bulge on the diaphragmatic contour just above the posterior costophrenic recess (arrows). (B) CT shows a fatty mass abutting the defect in the posteromedial aspect of the left hemidiaphragm (arrowheads).

normal adults.These hernias appear on a lateral radiograph as a focal bulge centred approximately 4–5 cm anterior to the posterior diaphragmatic insertion. On CT and MRI the diagnosis can be made when a soft tissue or fatty mass is seen protruding through a small defect in the posteromedial aspect of either hemidiaphragm. A Morgagni hernia presents in adulthood as an anterior opacity at the right cardiophrenic angle. It frequently contains omentum and may contain gut. Its smooth, welldefined margin and soft tissue radiodensity usually allow its differentiation from the much more common fat pad collection at this site. It is more difficult to differentiate from a low-lying pericardial cyst. Morgagni hernias containing gut can be diagnosed using barium but the diagnosis is more simply established by means of CT or MRI. Hernias through the oesophageal hiatus are extremely common, particularly in the elderly in whom they may be an incidental finding on CT.

Diaphragmatic trauma Because diaphragmatic rupture is often associated with thoracic or abdominal injuries that require surgical treatment, many cases are diagnosed during surgery28. If surgery is not indicated, diaphragmatic tear can be missed, especially when it is small and when there is no herniation of abdominal structures to the chest. That is why suspicion is needed in all cases of trauma to the lower chest, but also in patients with severe pelvic trauma. The chest X-ray should be evaluated carefully. Special attention has to be given to small changes in the diaphragm or to basal lung atelectasis or consolidation. If possible, the post-traumatic thorax should always be compared with previous chest X-rays. The diagnostic tools are different in the acute and latent phase. In the acute phase surgical procedures are often necessary and if the patient has severe injuries bedside examinations, such as chest X-ray and ultrasound, should be relied upon (Fig. 13.23). In the latent phase barium studies, spiral CT and MRI can give additional diagnostic information.

Figure 13.23 Traumatic diaphragmatic rupture. (A) Ultrasound and (B) CT show traumatic laceration of the liver and herniation of liver tissue into the fluid-filled pleural space.

CHAPTER 13

In the acute phase the chest X-ray is normal in about onequarter of cases. In some cases gas and fluid shadows are seen in the thorax. Sometimes there is only a localized density in close relationship to the diaphragm, or an alteration in the diaphragmatic shape. The position of a nasogastric tube can help to localize the gastric fundus, but does not tell anything about the position of the diaphragm, which is essential in the diagnosis of a diaphragmatic tear. A follow-up X-ray of an acutely injured patient showing progressive opacification of one thorax side by a gas-filled structure is strongly suggestive for diaphragmatic rupture (Fig. 13.24). Barium studies can be very helpful in making the correct diagnosis, when an extrinsic narrowing occurs on the border of the stomach or bowel at the point where they pass the

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

diaphragmatic tear. However, since barium studies cannot be used in emergency situations, they are predominantly indicated in the latent phase and eventually in the obstructive phase. Pneumoperitoneum can be established by bringing a small amount of air into the abdominal cavity. If air shifts through the diaphragmatic tear and a pneumothorax occurs, the test is diagnostic for diaphragmatic rupture. However, no shift of air will occur when the tear is closed by adhesions or by the herniated organs themselves. In this case the exact position of the diaphragm can be visualized since it is delineated by the subdiaphragmatic air. Ultrasound can be diagnostic if both the diaphragm and the herniated organs can be visualized. Examination of the right hemidiaphragm is facilitated by the presence of the liver,

Figure 13.24 Traumatic rupture of the diaphragm diagnosed 2 months after the trauma. (A) Detail of the left hemithorax. The supine chest radiograph immediately after the trauma shows multiple rib fractures, a pleural effusion and a poorly-defined opacity at the left lung base. (B) One month after the trauma the chest radiograph is normal but (C) 2 months later a large gas-filled structure corresponding with the air-containing stomach is seen in the left hemithorax suggesting rupture and herniation. (D) CT confirmed the diagnosis of diaphragmatic rupture and shows the herniated stomach (S).

239

240

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 13.25 Primary malignant tumour of the diaphragm. (A) PA chest radiograph shows a small focal bulge of the diaphragm in combination with a small pleural effusion. (B) CT and (C) MRI show an irregular mass with central necrosis in continuity with the right hemidiaphragm (arrows).

acting as an acoustic window. However, this technique is limited by the often minimal visualization of the diaphragm itself, the tenderness over the upper abdomen and the presence of gas in herniated bowel. The MDCT diagnosis of diaphragmatic rupture is largely based on the fact that abdominal organs are seen in the pleural space outside the diaphragm. However, the identification of the diaphragm on standard CT images can be very difficult; multiplanar CT reconstructions can help to show the defect directly. The more usual CT signs of diaphragmatic rupture include29,30: discontinuity of the diaphragm with direct visualization of the diaphragmatic injury; herniation of abdominal organs with liver, bowel or stomach in contact with the posterior ribs (‘dependent viscera sign’); thickening of the crus (‘thick crus sign’); constriction of the stomach or bowel (‘collar sign’); active arterial extravasation of contrast material near the diaphragm; and, in the case of a penetrating diaphragmatic injury, depiction of a missile or puncturing instrument trajectory. Because it is in most cases difficult to perform an MRI examination during the acute phase, this technique is more valuable in the latent phase. It allows both a static and a dynamic view of the diaphragm. However, as in CT, the parts of the diaphragm in contact with the liver and spleen are not visible.

Neoplasms of the diaphragm Primary tumours of the diaphragm are rare (Fig. 13.25). Both benign and malignant varieties are mostly derived from muscle, fibrous tissue, blood vessels, or fat. They are usually well defined and on the right may mimic an elevated diaphragm or local eventration. Calcification has been described in lipomas. Malignant tumours may present as a pleural effusion. Secondary invasion of the diaphragm by malignant tumours of the lung, pleura, stomach, or pancreas may occur. Imaging with CT or MRI is particularly helpful in such patients.

REFERENCES 1. Kuhlman J E, Bouchardy L, Fishman E K et al 1994 CT and MR imaging evaluation of chest wall disorders. RadioGraphics 14: 571–595 2. Fortier M, Mayo J R, Swensen S J et al 1994 MR imaging of the chest wall lesions. Radiographics 14: 597–606

3. Takasugi J E, Rapoport S, Shaw C 1989 Superior sulcus tumors: the role of imaging. J Thorac Imaging 4: 41–48 4. Deschildre F, Petyt L, Remy-Jardin M et al 1994 Evaluation de la TDM par balayage spirale volumique (BSV) vs IRM dans le bilan d’extension pariétal des masses thoraciques. Rev Im Med 6(S): 188 5. Müller N L 1993 Imaging the pleura. Radiology 186: 297 6. McLoud T C, Flower C D R 1991 Imaging the pleura: sonography CT, and MR imaging. Am J Roentgenol 156: 1145–1153. 7. Fleischner F G 1963 Atypical arrangement of free pleural effusion. Radiol Clin North Am 1: 347–362 8. Lomas D J, Padley S G, Flower C D R 1993 The sonographic appearances of pleural fluid. Br J Radiol 66: 619–624 9. McLoud T C 1998 CT and MR in pleural disease. Clin Chest Med 19: 261–276 10. Bittner R C, Schnoy N, Schonfeld N et al 1995 High-resolution magnetic resonance tomography (HR-MRT) of the pleura and thoracic wall: normal findings and pathological changes. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 162: 296–303 11. Lesur O, Delorme N, Fromaget J M et al 1990 Computed tomography in the etiologic assessment of idiopathic spontaneous pneumothorax. Chest 98: 341–347 12. Gordon R 1980 The deep sulcus sign. Radiology 136: 25–27 13. Rhea J T, vanSonnenberg E, McLoud T C 1979 Basilar pneumothorax in the supine adult. Radiology 133: 593–595 14. Aberle D R, Gamsu G, Ray C S 1988 High resolution CT of benign asbestos-related diseases: clinical and radiographic correlation. Am J Roentgenol 151: 883 15. Falaschi F, Battolla L, Maschalchi M et al 1996 Usefulness of MR signal intensity in distinguishing benign from malignant pleural disease. Am J Roentgenol 166: 963–968 16. England D M, Hochholzer L, McCarthy M J 1989 Localized benign and malignant fibrous tumors of the pleura. A clinicopathologic review of 223 cases. Am J Surg Pathol 13: 640–658 17. Kinoshita T, Ishii K, Miyasato S 1997 Localized pleural mesothelioma: CT and MR findings. Magn Reson Imaging 15: 377–379 18. Layer G, Schmitteckert H, Steudel A et al 1999 MRT, CT and sonography in the preoperative assessment of the primary tumor spread in malignant pleural mesothelioma. RoFo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 170: 365–370 19. Rusch V W, Godwin J D, Shuman W P 1988 The role of computed tomography scanning in the initial assessment and the follow-up of malignant pleural mesothelioma. J Thorac Cardiovasc Surg 96: 171 20. Klein J S, Zarka M 1997 Transthoracic needle biopsy: an overview. J Thorac Imaging 12: 232–249 21. vanSonnenberg E, Casola G, Ho M et al 1988 Difficult thoracic lesions: CT-guided biopsy experience in 150 cases. Radiology 167: 457–461 22. Ikezoe J, Morimoto S, Kozuka T 1991 Sonographically guided needle biopsy of thoracic lesions. Semin Intervent Radiol 8: 15–22

CHAPTER 13

23. White C S, Meyer C A, Templeton P A 2000 CT fluoroscopy for thoracic interventional procedures. Radiol Clin North Am 38: 303–322 24. Buecker A, Adam G, Neuerburg J M et al 1998 MR-guided biopsy using a T2-weighted single-shot zoom imaging sequence (local look technique). J Magn Reson Imaging 8: 955–959 25. Deslauriers J 1988 Eventration of the diaphragm. Chest Surg Clin North Am 8: 315–330 26. Houston J G, Fleet M, Cowan M D et al 1995 Comparison of ultrasound with fluoroscopy in the assessment of suspected hemidiaphragmatic movement abnormality. Clin Radiol 50: 95–98

• THE CHEST WALL, PLEURA, DIAPHRAGM AND INTERVENTION

27. Demartini W J, House A J S 1980 Partial Bochdalek herniation: computerized tomographic evaluation. Chest 77: 702–704 28. Shah R, Sabanathan S, Mearns J A et al 1995 Traumatic rupture of diaphragm. Ann Thorac Surg 60: 1444–1449 29. Bergin D, Ennis R, Keogh C et al 2001 The “dependent viscera” sign in CT diagnosis of blunt traumatic diaphragmatic rupture. Am J Roentgenol 177: 1137–1140 30. Leung J C, Nance M L, Schwab C W et al 1999 Thickening of the diaphragm: a new computed tomography sign of diaphragm injury. J Thorac Imaging 14: 126–129

241

CHAPTER

The Mediastinum, Including the Pericardium*

14

Sharyn L. S. MacDonald and Simon Padley

Mediastinal diseases • Mediastinal masses • Other mediastinal lesions The pericardium • Normal anatomy

• Imaging pericardial disease • Developmental anomalies • Acquired pericardial disease

MEDIASTINAL DISEASES MEDIASTINAL MASSES Incidence The true incidence of mediastinal masses is difficult to ascertain because most surgical series are biased towards patients requiring surgery and therefore do not include all aneurysms, intrathoracic goitres, or lymph node masses in patients with established diagnoses such as lymphoma or sarcoidosis. In adult surgical series1–3, the most frequent tumours are of neurogenic (17–23%) or thymic (20–25%) origin, or are neoplastic disorders of lymph nodes (10–20%). Developmental cysts, thyroid masses, and germ-cell tumours constitute the next most frequent group (approximately 10% each). In children, neuroblastoma/ganglioneuroma, foregut cysts and germ-cell tumours account for over three-quarters of cases, whereas thymoma and thyroid masses are rare. Mediastinal masses are conventionally divided by location into anterior, middle, or posterior mediastinal compartments. This division into compartments is for descriptive convenience only, since it is not based on anatomical boundaries that limit spread; nor do radiologists in general use these terms in the way they are defined by anatomists. Localization of a mass to one of these compartments is a useful step towards reaching the most appropriate diagnosis or differential diagnosis; the age of the patient and such characteristics as the presence of calcification, fat, fluid, or soft tissue within the mass, invasion of the mediastinal fat (indicating malignant rather that

benign disease) and contrast enhancement characteristics on computed tomography (CT) and magnetic resonance imaging (MRI) are important in narrowing the differential diagnosis, which is considered later in this chapter.

Imaging techniques Mediastinal masses are often incidentally detected on chest radiograph. Despite diagnostic limitations the chest radiograph is also important for detecting and localizing mediastinal masses when suspected clinically.

Computed tomography Computed tomography is the most useful investigation for localizing, characterizing and demonstrating the extent of a mediastinal mass and its relationship to adjacent structures. Multidetector CT (MDCT) following intravenous contrast medium with multiplanar reformats provides an excellent assessment of mediastinal structures, including vessels, and has largely obviated the need to proceed to MRI for imaging in the coronal and sagittal planes. CT may also be used to guide biopsy, plan resection and follow response to therapy.

Magnetic resonance imaging Magnetic resonance imaging remains useful for imaging suspected neurogenic tumours, for demonstrating intraspinal extension of a mediastinal mass and for further evaluating the relationship of a mass to the heart, pericardium and larger intrathoracic vessels. MRI may have advantages over

*The section on the pericardium is adapted from Chapter 42 by Anna Rozenshtein, Lawrence M. Boxt, Kathleen Reagan and Robert M. Steiner from the fourth edition of this textbook.

244

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

contrast-enhanced CT for distinguishing between solid tissue and adjacent vessels (fast flowing blood in vessels results in a signal void on spin-echo sequences) and may be useful for confirming that a mass is cystic. Unlike CT, MRI is not sensitive to the presence of calcification.

Ultrasound Ultrasound of the mediastinum, including echocardiography and endoscopic ultrasound, may be of use in selected patients, in particular for distinguishing cystic from solid mediastinal masses and for distinguishing cardiac from paracardiac masses. Ultrasound may also be used to guide mediastinal biopsy.

Radionuclide examinations Radionuclide examinations have a limited role in the assessment of mediastinal masses. Positron emission tomography (PET) and PET–CT using [F-18]2-deoxy-d-glucose (18F-FDG) has proven useful in the evaluation of mediastinal lymph node involvement in lung cancer and lymphoma4,5. Radionuclide examinations may also be useful in the imaging of thyroid masses, neuroendocrine tumours and phaechromocytomas.

Thyroid masses Most thyroid masses (Fig. 14.1) in the mediastinum represent downward extensions of either a multinodular colloid goitre or, occasionally, an adenoma or carcinoma. Intrathoracic thyroid masses usually have a well-defined outline, which may be spherical or lobular. Rounded or irregular, well-defined areas of calcification may be seen in benign areas, whereas amorphous cloud-like calcification is occasionally seen within carcinomas. Almost all intrathoracic thyroid masses displace the trachea, which may also be substantially narrowed. The direction of displacement depends on the location of the mass. Thyroid

masses are most commonly anterior and lateral to the trachea. Posteriorly placed masses often separate the trachea and the oesophagus, and such separation by a localized mass rising into the neck is virtually diagnostic of a thyroid mass. Radionuclide imaging with 123I or 131I demonstrates the presence of thyroid tissue within the mediastinum in almost all intrathoracic goitres. Although radionuclide imaging is a sensitive and specific method of determining the thyroid nature of an intrathoracic mass, CT is more useful as the initial investigation because it provides more information should the mass prove to be something other than a thyroid lesion and is almost as specific as nuclear medicine in diagnosing a thyroid origin. CT optimally demonstrates the shape, size and position of the mass (Fig. 14.2)6,7. It is usually possible to diagnose a thyroid origin by noting a well-defined mass in the paratracheal or retrotracheal region, almost invariably being continuous with the thyroid gland in the neck. Another useful sign is that normal thyroid tissue within the mass shows a higher attenuation value than muscle on images obtained both before and after contrast medium. Focal areas of calcification are frequently identified and rounded, focal, nonenhancing low density areas are common. It is not possible to distinguish between a benign and malignant mass on CT unless the tumour has clearly spread beyond the thyroid gland. It should, however, be noted that multiple masses are a feature of benign multinodular goitre, though carcinoma can develop in multinodular goitre. MRI of intrathoracic goitre, like CT, can identify cystic and solid components, and can show haemorrhage to advantage, but will not reliably demonstrate calcification.

Parathyroid masses The parathyroid glands may migrate into the chest during fetal development. Mediastinal parathyroid tumours causing hyperparathyroidism are most commonly located in or around

Figure 14.1 Intrathoracic thyroid mass on (A) AP and (B) lateral radiographs. This benign multinodular goitre is predominantly posterior to the trachea with components to either side, resulting in forward displacement and narrowing of the trachea.

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

otherwise normal gland, may lie within a thymoma, or may follow thymic irradiation for Hodgkin’s disease10.

Thymomas

Figure 14.2 Intrathoracic extension of thyroid on coronal image from multidetector CT. Contrast medium has been administered. The thyroid demonstrates heterogeneous contrast enhancement, and is mainly of greater attenuation than skeletal muscle. There are flecks of calcification in the gland.

the thymus. They are small and almost never visible on plain radiographs. They are probably best detected using ultrasound (Fig. 14.3) and if not readily apparent, by subsequent 99m Tc-sestamibi imaging with CT or MRI only in selected cases8,9.

Thymic tumours Tumours that arise within the thymus include thymoma, lymphoma, thymic carcinoid, germ-cell tumour/teratoma and thymolipoma. Thymic cysts are another cause of a thymic mass (Fig. 14.4). They may be simple cysts and occur in an

Figure 14.3 High-resolution ultrasound of the left lobe of the thyroid (LLT) anterior to a parathyroid adenoma (white arrows). There is a small area of cystic degeneration within the posterior aspect of the adenoma. Black arrows indicate the common carotid artery.

Thymomas are the most common tumour of the thymus in adults, and the most common primary tumour of the anterior mediastinum in adults. The average age at diagnosis is approximately 50 years, earlier in those who present with myasthenia gravis. Thymomas are rare under 20 years of age and extremely unusual below the age of 15. Up to 50% of patients with thymoma have myasthenia gravis, and approximately 10–20% of patients with myasthenia gravis have a thymoma. A variety of other syndromes are seen in patients with thymoma, including hypogammaglobulinaemia and red cell aplasia11,12. Most thymomas (90%) arise in the upper anterior mediastinum. The mass is usually anterior to the ascending aorta, lying above the right ventricular outflow tract and pulmonary artery (Fig. 14.5). A few are situated more inferiorly, projecting from the left or right heart border, or lying close to the cardiophrenic angles. They are usually spherical or oval in shape and may show lobulated borders. They may contain one or more cysts and a few are predominantly cystic (Fig. 14.6). Calcification, punctate or curvilinear, may be seen (Fig. 14.7). All of these features are best demonstrated using CT13,14, which is the most sensitive technique for the detection of thymoma in patients with myasthenia gravis15. The diagnosis depends on identifying a focal swelling rather than applying a specific measurement. Thymomas as small as 1.5– 2.0 cm in diameter are readily identified over the age of 40, largely because the rest of the thymus is atropic. Before age 40, and particularly before 30, diagnosing a small thymoma can be difficult, because the normal gland is variable in size and in myasthenia gravis the associated hyperplasia may cause a bulky gland. In these circumstances a useful rule is that thymoma usually gives rise to an asymmetrical focal swelling. Fortunately, thymoma is so infrequent in children that the potentially difficult problem of finding a thymoma in a child with myasthenia gravis rarely arises. Thymomas usually show homogeneous density and uniform enhancement after contrast medium and may occasionally be cystic. Invasion of the mediastinal fat and adjacent pleura may be identified with invasive thymomas (Fig. 14.8), and while CT shows such invasion to advantage it cannot reliably diagnose invasive thymoma if the tumour is still confined to the thymus16. Remote pleural metastases resulting from transpleural spread are a feature of invasive thymomas, and therefore the whole of the pleural cavity should be carefully examined16 (Fig. 14.8). MRI, in general, provides similar information to CT, though it can be useful to show mediastinal spread when there is doubt on the latter. On T1-weighted images, thymomas have a signal intensity similar to that of muscle and the adjacent normal thymic tissue. On T2-weighted images, the signal intensity increases and may make it difficult to distinguish a thymoma from adjacent mediastinal fat. Heterogeneity of signal intensity caused by septation, cystic change and haemorrhage is common16.

245

246

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 14.4 Thymic cyst producing an anterior mediastinal mass on (A) AP and (B) lateral chest radiographs, filling in the normal retrosternal window and widening the mediastinum. (C) The cystic nature is best demonstrated by CT.

Figure 14.5 Thymic enlargement on (A) PA and (B) lateral radiographs of a patient with a malignant thymic mass. The lateral view demonstrates pleural metastases posteriorly (arrows). (C) CT confirms the anterior position of the primary tumour suspected from the filling of the retrosternal window apparent on the lateral radiograph (B).

Thymic carcinomas Thymic carcinomas are aggressive locally invasive malignancies that have frequently metastasized to regional lymph nodes and distant sites at presentation. They are typically large, heterogeneous masses, containing areas of necrosis and calcification and demonstrating evidence of invasion of adjacent structures, in particular the mediastinum, pericardium and pleura17,18.

Thymic lymphoma Thymic lymphoma is usually part of generalized disease, but isolated involvement is very occasionally encountered. It is most commonly associated with Hodgkin’s disease. The imaging features are the same as those of thymoma. Figure 14.6 Cyst formation in a thymoma demonstrated on CT in a patient with myasthenia gravis. The wall is irregular and enhances following administration of intravenous contrast medium.

Thymic carcinoid Thymic carcinoid is histologically distinct from thymoma. A noteworthy feature is that these tumours may secrete

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

Figure 14.7 Thymoma presenting on a chest radiograph obtained before orthopaedic surgery in an otherwise asymptomatic elderly female patient. There is a large anterior mediastinal mass (A) with coarse calcification visible on (B) the lateral view and (C) contrast-enhanced CT.

Figure 14.8 Invasive thymoma in a young man. (A) Shows a lobular anterior mediastinal mass associated with a pleural effusion. (B) Image obtained through the lower chest demonstrates mixed soft tissue (arrows) and fluid attenuation owing to transpleural spread of tumour.

adrenocorticotropic hormones in sufficient quantities for the patient to present with Cushing’s syndrome. The plain radiograph and CT features of thymic carcinoid are indistinguishable from those of thymoma.

Thymolipomas Thymolipomas are rare tumours composed of a mixture of mature fat and normal-looking or involuted thymic tissue.The age range is 3–60 years. Individual cases have been reported in association with a variety of conditions, including myasthenia gravis, aplastic anaemia, Graves’ disease and hypogammaglobulinaemia. Thymolipomas can grow to a very large size before discovery and, being soft, mould themselves to the adjacent mediastinum and diaphragm, and may mimic cardiomegaly or lobar collapse19. CT/MRI show the fatty nature of the mass, with islands of thymus and fibrous septa running through the lesion19,20.

Thymic hyperplasia The most common association of thymic hyperplasia is myasthenia gravis, but thymic hyperplasia is also seen in other conditions, notably thyrotoxicosis. Thymic hyperplasia is rarely severe enough to cause visible enlargement of the thymus, but when it does, both lobes are enlarged, usually uniformly, though on occasion thymic hyperplasia may mimic a thymic mass. The thymus may atrophy due to stress or as a consequence of steroid or antineoplastic drug therapy21,22.The gland usually returns to its original size on recovery or cessation of treatment, but it may become larger than its previous normal size in the phenomenon known as rebound thymic hyperplasia. It may then be difficult to distinguish between thymic rebound and thymic involvement by neoplasm. The diagnosis depends on a known reason for thymic rebound, the absence of clinical

247

248

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

features to indicate tumour recurrence and the presence of an enlarged, normally-shaped thymus22,23.

Germ-cell tumours of the mediastinum Germ-cell tumours of the mediastinum are believed to be derived from primitive germ-cell elements left behind after embryonal cell migration. The mediastinum is the most common extragonadal site for these tumours, almost all of which arise in the anterior mediastinum, within, or in intimate contact with, the thymus. Mediastinal germ-cell tumours include mature teratoma (benign) and a number of malignant forms, chiefly seminoma, malignant teratoma, embryonal carcinoma, choriocarcinoma, endodermal sinus tumour and tumours with mixtures of these cell types24. Malignant germ-cell tumours secrete human chorionic gonadotrophin and α-fetoprotein, which can be used as markers to diagnose and monitor the tumour.

Mature teratomas Mature teratomas are the most common mediastinal germcell tumour2; most are cystic. Mature teratomas are found at all ages, particularly in adolescents and young adults, with women slightly outnumbering men25,26. They are usually asymptomatic and diagnosed incidentally on chest radiography or CT, but may give rise to cough, dyspnoea, or chest pain if they compress the bronchial tree or superior vena cava, or if they rupture into the mediastinum or lung. They are frequently large and may be huge, occupying much of one hemithorax. The lesions are usually stable, but haemorrhage or infection may lead to a rapid increase in size. On chest radiograph or CT most mature teratomas present as a well-defined, rounded or lobulated mass, localized to the anterior mediastinum. Fat and calcification may occasionally be identified on chest radiograph. CT appearances are variable, combinations of fat, fluid, soft tissue components and calcification may be seen25 (Fig. 14.9). The presence of fat, either as focal collections or fluid fat, is a very helpful diagnostic feature

favouring mature (benign) cystic teratoma over the other causes of anterior mediastinal mass. MRI provides similar information to CT, although it may not detect calcification.

Malignant germ-cell tumours Malignant germ-cell tumours are usually seen in young adults and are much more common in men (>90%) than women. Seminoma is the most common form27. They are more commonly more symptomatic than mature teratoma, usually due to mass effect or invasion of adjacent structures. The plain radiographic findings are similar except that the malignant tumours are more often lobular in outline, fat density is not seen and visible calcification is rare. Because they are malignant tumours, they grow rapidly and metastasize readily to the lungs, bones, or pleura. CT shows a lobular, asymmetrical mass.The adjacent mediastinal fat planes may be obliterated, and the tumours are either of homogeneous soft tissue density or show multiple areas of contrast enhancement interspersed with rounded areas of decreased attenuation due to necrosis and haemorrhage28,29. MRI provides similar information to CT (Fig. 14.10).

Mediastinal lymphadenopathy Lymph node calcification Extensive lymph node calcification is common following tuberculosis and fungal infection, and is occasionally seen with other infections. It may also be encountered in a variety of other conditions, notably sarcoidosis, silicosis and amyloidosis. Although it may be seen in lymph node metastases from calcifying primary malignancies, such as osteosarcoma, chondrosarcoma and mucinous colorectal and ovarian tumours, lymph node calcification is rare in metastatic neoplasm. It is virtually unknown in untreated lymphoma though it is occasionally seen in nodes involved by Hodgkin’s disease following therapy. CT demonstrates more calcification than plain radiographic techniques. Calcification is not usually visible on MRI. Two

Figure 14.9 Teratoma in a young man undergoing an immigration chest radiograph. (A) There are no specific features on the plain radiograph to indicate the nature of the mass. (B) CT demonstrates that the opacity visible on the chest radiograph is well defined and contains soft tissue and fat densities.

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

with inflammatory disorders, particularly tuberculosis, fungal disease32, sarcoidosis and neoplasm. When striking, it points particularly to the diagnosis of metastatic neoplasm from a highly vascular primary tumour, such as melanoma, renal and thyroid carcinoma, carcinoid tumour, or leiomyoma/sarcoma. A rare cause of strikingly uniform contrast enhancement is Castleman’s disease. A low-density centre with rim enhancement of the enlarged node is a useful pointer towards the diagnosis of tuberculous infection31 (Fig. 14.12).

Lymph node enlargement Figure 14.10 Malignant germ-cell tumour in a 25 year old man presenting with chest pain, dyspnoea, malaise and features of pericardial tamponade. The CT shows a lobular asymmetrical mass with low attenuation areas corresponding to necrotic tumour intersected by neoplastic septation.

common patterns of calcification are coarse, irregularly-distributed clumps within the node and homogeneous calcification of the whole node. A strikingly foamy appearance is seen with Pneumocystis jiroveci (previously P. carinii) infection in acquired deficiency syndrome (AIDS) patients30 and in some cases of metastatic mucinous neoplasms. Sometimes there is a ring of calcification at the periphery of the node—so-called ‘eggshell calcification’, which is a particular feature of sarcoidosis and of prolonged dust exposure in coal and metal mines.

Low attenuation nodes On CT, areas of low attenuation within enlarged nodes, corresponding to necrosis (Fig. 14.11), may be seen in a variety of conditions, particularly tuberculosis31 and occasionally in fungal disease32, infections in immunocompromised patients, metastatic neoplasm (notably from testicular tumours33) and lymphoma34. Attenuation values below that of water are seen in fatty replacement of inflammatory nodes and have also been described in Whipple’s disease35.

When mediastinal or hilar nodes are greater than 2 cm in their short axis diameter, the enlargement is likely to be due to metastatic carcinoma, malignant lymphoma, sarcoidosis, tuberculosis, or fungal infection. With lesser degrees of enlargement the differential diagnosis broadens to include lymph node hyperplasia and pneumoconiosis. Widespread moderate mediastinal lymph node enlargement is a frequent accompaniment of chronic diffuse lung disease and bronchiectasis. Normal sized nodes are demonstrable at CT/MRI, but are not visible on plain chest radiographs or on conventional tomography. The ease with which enlarged nodes can be recognized using plain radiography varies according to their location. Nodes in the right paratracheal group are readily identified: they show uniform or lobular widening of the right paratracheal stripe. Enlarged azygos nodes displace the azygos vein laterally and enlarge the shadow that normally represents just the azygos vein to over 10 mm in its short axis diameter. If the lymph nodes beneath the aortic arch become large enough to project beyond the aortopulmonary window they cause a local bulge in the angle between the aortic arch and the main pulmonary artery. Hilar lymph node enlargement causes enlargement and/or lobulation of the outline of the hilar shadows (Fig. 14.13).The diagnosis of lymph node enlargement on

Contrast-enhanced CT Contrast enhancement of enlarged nodes, when moderate in degree, is non-specific, being seen

Figure 14.11 Low attenuation lymph node enlargement. There is necrosis within malignant right hilar and subcarinal nodes which have arisen from the primary tumour in the right lung.

Figure 14.12 Tuberculous lymphadenopathy. Following contrast enhancement there is rim enhancement and central low attenuation due to caseation (arrows).

249

250

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

normally concave toward the lung, flattens or becomes convex towards the lung, an appearance that may be confused with left atrial enlargement. Posterior mediastinal lymph node enlargement causes localized displacement of the paraspinal and paraoesophageal lines. Lymph nodes elsewhere in the mediastinum are only recognizable on plain radiographs when substantially enlarged (Fig.14.14). CT is an excellent method for detecting mediastinal lymph node enlargement (Figs 14.15–14.17). It is usually easy to distinguish between the normal vascular structures and enlarged lymph nodes using contrast-enhanced CT, although without excellent opacification of the left atrium it may, on occasion, be difficult to distinguish enlarged subcarinal lymph nodes from a normal or enlarged left atrium. The short axis measurement provides the most representative guide to true size, since long axis measurements vary to a significant degree according to the orientation of the lymph node within the CT section. In the assessment of lymph node enlargement, MRI provides essentially the same information as CT, although its use is limited to selected cases due to longer acquisition times and relatively limited spatial resolution (which may make measurement of individual nodes difficult). Figure 14.13 Sarcoidosis producing symmetrical bilateral hilar lymph node enlargement.

plain radiography depends on the recognition of the edge of a round or oval hilar mass, an analysis that requires a detailed understanding of the normal anatomy of the hilar blood vessels. Subcarinal lymph node enlargement widens the carinal angle and displaces the azygo-oesophageal line, so that the subcarinal portion of the azygo-oesophageal line, which is

Sarcoidosis Sarcoidosis is the most common cause of intrathoracic lymph node enlargement, the hilar nodes being enlarged in almost all cases36. Additionally, tracheobronchial, aortopulmonary and subcarinal nodes are enlarged in over half the patients37. Anterior mediastinal nodes occasionally increase in size, but posterior mediastinal node enlargement is very unusual and it is seldom that either is seen in isolation. The important diagnostic feature of lymphadenopathy in sarcoidosis is its symmetry (Fig. 14.13).

Figure 14.14 Massively enlarged lymph nodes. (A,B) Massive anterior mediastinal lymphadenopathy due to malignant lymphoma. There are huge lobulated swellings in the upper anterior mediastinum which are easily seen in both projections. (C) Massive anterior mediastinal nodal enlargement secondary to Hodgkin’s disease demonstrated by CT. There is marked compression and distortion of the mediastinal structures and bilateral small pleural fluid reactions.

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

Malignant lymphoma and leukaemia

Figure 14.15 Right paratracheal lymph node enlargement (arrows) due to sarcoidosis.

Malignant lymphoma often involves mediastinal and hilar lymph nodes, multiple nodal groups usually being involved, particularly in Hodgkin’s disease. Lymph node enlargement is seen in a higher proportion of patients with Hodgkin’s than non-Hodgkin’s lymphoma. Any intrathoracic nodal group may be enlarged and the possible combinations are legion, but the following generalizations regarding plain radiograph, CT and MRI findings can be made38,39. 1 The anterior mediastinal and paratracheal nodes are the groups most frequently involved, the tracheobronchial and subcarinal nodes also being enlarged in many cases. In most cases, the lymphadenopathy is bilateral but asymmetrical. Hodgkin’s disease, particularly the nodular sclerosing form, has a propensity to involve the anterior mediastinal and paratracheal nodes (Fig. 14.14). 2 Hilar node enlargement is rare without accompanying mediastinal node enlargement, particularly in Hodgkin’s disease. 3 The posterior mediastinal nodes are infrequently involved— the enlarged nodes are often low down in the mediastinum and contiguous retroperitoneal disease is likely. 4 The paracardiac nodes are rarely involved but become important as sites of recurrent disease because they may not be included in the initial radiation therapy fields40.

Lymph node enlargement is also seen occasionally with leukaemia, the pattern being the same as with lymphoma. The lymph node enlargement in both lymphoma and leukaemia may resolve remarkably rapidly with therapy (Fig. 14.18).

Figure 14.16 Malignant mediastinal lymph node enlargement due to metastases from a small cell carcinoma of the lung.

Tuberculosis and histoplasmosis Lymph node enlargement due to tuberculous or fungal infection may affect any of the nodal groups in the hila or mediastinum. One or more lymph nodes may be visibly enlarged and an associated area of pulmonary consolidation may or may not be present. Occasionally, widespread massive mediastinal and hilar node enlargement is seen. With healing, the nodes usually become smaller, often returning to normal size. Dense calcification is frequent both in nodes that stay enlarged and in those that shrink. The enlarged nodes, together with surrounding fibrosis, may compress the superior vena cava or pulmonary veins and cause obstruction. Rim enhancement with a low density centre may be seen with tuberculosis on contrast-enhanced CT examination31 (see Fig. 14.12). Metastatic carcinoma Mediastinal lymph node metastases from bronchial carcinoma are discussed in Chapter 18. Metastases may also occur from extrathoracic primary carcinomas. In one large series, half the cases of mediastinal lymph node enlargement from extrathoracic primary carcinomas arose from tumours of the genitourinary tract, particularly the kidney and testis41. Other major sources are head and neck tumours and breast carcinomas.

Figure 14.17 Metastatic malignant teratoma involving mediastinal nodes and directly invading the lumen of the superior vena cava (arrow), where it is outlined by intravenous contrast medium.

Reactive hyperplasia in nodes draining infection or neoplasm may cause mild nodal enlargement that is recognizable on CT but rarely so with plain radiograph techniques. Castleman’s disease is a specific type of lymph node hyperplasia of uncertain aetiology which can cause substantial lymph node enlargement

251

252

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 14.18 (A,B) Acute lymphocytic leukaemia in a 9-year-old boy showing rapid resolution of massive mediastinal adenopathy following chemotherapy. The two radiographs were taken 7 d apart.

in many sites in the body. When seen within the thorax, the enlarged nodes are usually situated in the middle or posterior mediastinum. The lymph node mass is often localized to one area, can be huge, and may be very vascular. The nodes may calcify and may show striking contrast enhancement on both CT and MRI42,43.

Foregut duplication cysts ‘Foregut duplication cyst’ is a useful term that covers various congenital cysts derived from the embryological foregut, including bronchogenic, enteric and neurenteric cysts.

Bronchogenic cysts Bronchogenic cysts are usually solitary asymptomatic mediastinal masses which may present at any age. Typically they have

a thin fibrous capsule, are lined with respiratory epithelium and contain cartilage.The cyst contents usually consist of thick mucoid material. Most are located adjacent to the trachea or main bronchi44. The cysts can grow very large without causing symptoms, but they may compress surrounding structures, particularly the airways and give rise to symptoms. In rare cases they become infected or haemorrhage occurs into the cyst; these complications may be life-threatening, particularly in infants and young children. On chest radiography nonmalignant mediastinal cysts present as spherical or oval masses with smooth outlines projecting from either side of the mediastinum (Fig. 14.19). Most are unilocular and do not have a lobulated outline, though lobulation may be seen. They usually contact the carina or

Figure 14.19 Oesophageal duplication cyst on (A) chest radiography and (B) CT. This case shows the typical features of a well-defined spherical mass projecting from the mediastinum.

CHAPTER 14

main bronchi, but may be seen anywhere along the course of the trachea and larger airways, and frequently project into the middle and/or posterior mediastinum. Calcification of the wall or of the cyst contents is rare. When located in the subcarinal area, these cysts may closely resemble an enlarged left atrium. Foregut duplication cysts frequently push the carina forward and the oesophagus backward—displacements that are almost never seen with other masses (the exceptions being thyroid masses and an aberrant left pulmonary artery). CT is an excellent method of demonstrating the size, shape and position of a bronchogenic cyst (Fig. 14.20). In some cases, it may demonstrate a thin-walled mass, with contents of uniform CT attenuation close to that of water (0 HU), thereby effectively making the diagnosis of a fluid-filled cyst45. In other cases, the CT attenuation is similar to soft tissue and therefore to tumour, in which case the differential diagnosis becomes wider (Fig. 14.20). Rarely, the cyst may show uniformly high density, probably due to a high protein content, or very high density indicating a very high calcium content (milk of calcium) within the fluid46. MRI shows the expected features of a fluid-filled cyst.

Oesophageal duplication cysts Oesophageal duplication cysts are uncommon44. They usually present first in childhood, but may not present until adulthood: initial presentation up to the age of 61 has been reported. They are distinguished from bronchogenic cysts pathologically by the presence of smooth muscle in the walls and contain mucosa resembling that of the oesophagus, stomach, or small intestine. Many are clinically silent and are first discovered as an asymptomatic mass on an imaging examination of the chest, but they may cause dysphagia, pain, or other symptoms due to the compression of adjacent structures. A duplication cyst may become infected or ectopic gastric mucosa within the cyst may cause haemorrhage or perforation. The imaging features of oesophageal duplication cysts

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

(see Fig. 14.19) on CT and MRI are identical to those of bronchogenic cysts (see Fig. 14.20) except that in the former the wall of the lesion may be thicker, the mass may assume a more tubular shape, and it may be in more intimate contact with the oesophagus44. Due to their close proximity to the oesophageal wall, barium swallow will show the features of extrinsic or intramural compression.

Neurenteric cysts Neurenteric cysts result from incomplete separation of the foregut from the notochord in early embryonic life. The cyst wall contains both gastrointestinal and neural elements with an enteric epithelial lining. There is usually a fibrous connection to the spine or an intraspinal component. Communication with the subarachnoid space or the gastrointestinal tract may be present, but communication with the oesophageal lumen is rare. There are typically associated vertebral body anomalies such as butterfly or hemivertebra. These cysts frequently produce pain and are often found early in life. Radiologically44, a neurenteric cyst is a well-defined, round, oval or lobulated mass in the posterior mediastinum between the oesophagus (which is usually displaced) and the spine. Appearances on CT and MRI are similar to those of other foregut duplication cysts, with MRI being the investigation of choice for demonstrating the extent of intraspinal involvement44.

Mediastinal pancreatic pseudocyst On rare occasions, a pancreatic pseudocyst extends into the mediastinum. Most patients are adults and have the clinical features of chronic pancreatitis; in children, the usual cause of the pseudocyst is trauma. On imaging examinations, most patients have left-sided or bilateral pleural effusions. The mediastinal component of the pseudocyst is almost always in the posterior mediastinum adjacent to the oesophagus, having gained access to the chest via the oesophageal or aortic hiatus47. CT is the optimal method of demonstrating these

Figure 14.20 Bronchogenic cyst. (A) Bronchogenic cyst in right paratracheal area in a young asymptomatic man. (B) In this instance the CT attenuation was almost the same as that of the other soft tissue structures and it was not possible to predict the cystic nature of the mass. The cyst was surgically removed.

253

254

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

thin-walled cysts, which show continuity with the pancreas and any peripancreatic fluid collections.

Neurogenic tumours Neurogenic tumours are the most common tumours to arise in the posterior mediastinum, and most neurogenic tumours occur in this location44. Most neurogenic tumours in adults are benign and are discovered as asymptomatic masses on chest radiography, though some, particularly the malignant lesions, cause chest pain.They can be classified as tumours arising from peripheral nerves, including neurofibroma, neurilemmoma (schwannoma) and malignant tumours of nerve sheath origin (neurogenic sarcomas), or as tumours arising from sympathetic ganglia. MRI is the best investigation for these tumours48.

Peripheral nerve tumours Peripheral nerve tumours typically originate in an intercostal nerve in the paravertebral region. Radiologically, the benign tumours (neurofibromas and schwannomas) present as welldefined round or oval posterior mediastinal masses. Pressure deformity causing a smooth, scalloped indentation on the adjacent ribs, vertebral bodies, pedicles, or transverse processes is common, particularly with the larger lesions44,49 (Fig. 14.21). The scalloped cortex is usually preserved and is often thickened.These bone changes are diagnostic of a neurogenic lesion, the only differential diagnosis being that of a lateral thoracic meningocele. The rib spaces and the intervertebral foramina may be widened by the tumour44,49. On CT the tumours may be homogeneous or heterogeneous, usually enhancing heterogeneously after intravenous contrast medium50. Punctate foci of calcification may be seen. On MRI neurofibroma and neurilemmoma have variable T1-weighted signal intensity that may be similar to spinal cord (Fig. 14.22). They may have

Figure 14.22 Neurofibroma in left paravertebral region. This coronal T1-weighted, spin-echo image demonstrates the tumour well and shows that it does not enter the spinal canal or encroach significantly on the adjacent foramina.

characteristic high signal intensity peripherally and low signal intensity centrally (target sign) on T2-weighted images51, and enhance uniformly after gadolinium. Ten per cent of paravertebral neurofibroma and neurilemmoma extend into the spinal canal and appear as dumb-bell-shaped masses with widening of the affected neural foramen52. Malignant tumours of nerve sheath origin are rare neoplasms, typically occurring in the third to fifth decades, although they may occur earlier in patients with neurofibromatosis Type 1. Radiologically the masses are usually larger than 5 cm in diameter44. Although MRI cannot reliably differentiate benign from malignant neurogenic tumours, sudden change in size of a pre-existing mass, the development of heterogeneous signal intensity (caused by haemorrhage and necrosis), or infiltration of adjacent mediastinum or chest wall are cause for concern48.

Sympathetic ganglion tumours

Figure 14.21 Neurofibrosarcoma showing widening and pressure deformity of adjacent ribs. It was not possible to predict the malignant nature of this tumour from the plain radiographs. A benign neurofibroma would have had identical features.

Sympathetic ganglion tumours are rare neoplasms forming a continuum ranging from benign ganglioneuroma to malignant neuroblastoma, with ganglioneuroblastoma being an intermediate form44. Ganglioneuromas are benign neoplasms usually occurring in children and young adults. Ganglioneuroblastomas exhibit variable degrees of malignancy and usually occur in children53. Neuroblastomas are highly malignant tumours that typically occur in children younger than 5 years of age53. The posterior mediastinum is the most common extra-abdominal location of a neuroblastoma. Ganglioneuromas and ganglioneuroblastomas usually arise from the sympathetic ganglia in the posterior mediastinum and therefore usually present radiologically as well-defined elliptical masses, with a vertical orientation, extending over the anterolateral aspect of three to five vertebral bodies44,54. Calcification occurs in approximately 25%. CT appearance is variable44. On MRI ganglioneuromas and ganglioneuroblastomas

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

are usually of homogeneous intermediate signal intensity on T1- and T2-weighted images. Neuroblastomas are typically more heterogeneous due to areas of haemorrhage, necrosis, cystic degeneration and calcium. They may be locally invasive and have a tendency to cross the midline48.

Mediastinal paragangliomas Intrathoracic paragangliomas are of two types: chemodectomas or phaeochromocytomas (functioning paragangliomas), either of which may be benign or malignant. Almost all intrathoracic chemodectomas are in a location close to the aortic arch and are classified as aortic body tumours. Other mediastinal chemodectomas are very rare55.They are usually single, but multicentric cases are reported. Fewer than 2% of phaeochromocytomas occur in the chest. Most intrathoracic phaeochromocytomas are found in the posterior mediastinum or closely related to the heart, particularly in the wall of the left atrium or the interatrial septum. Approximately one-third of mediastinal phaeochromocytomas are nonfunctioning and asymptomatic, the remainder presenting with the symptoms, signs and laboratory findings of overproduction of catecholamines. The various paragangliomas have similar appearances on chest radiography, CT and MRI. They form rounded, soft tissue masses, which are usually very vascular and therefore enhance intensely on CT56. On MRI, phaeochromocytomas usually show a signal intensity similar to muscle on T1-weighted images and very high signal intensity on T2weighted images57. MRI is particularly useful for demonstrating intracardiac phaeochromocytomas. Radio-iodine MIBG (meta-iodobenzylguanidine) and somatostatin receptor scintigraphy both show increased activity in paragangliomas, and are useful techniques for identifying extra-adrenal phaeochromocytomas57,58.

Lateral thoracic meningocele Lateral thoracic meningoceles are protrusions of the spinal meninges through an intervertebral foramen. Like neurofibromas, they are commonly associated with neurofibromatosis59. They are rare lesions that present as an asymptomatic mass, often with pressure deformity of the adjacent bone, indistinguishable on plain radiographs from neurofibromas. CT and MRI can both indicate the correct diagnosis by showing the mass to be fluid filled rather than solid44. If necessary, the diagnosis can be established by CT with intrathecal contrast medium demonstrating flow into the lesion.

Extramedullary haematopoiesis Extramedullary haematopoiesis is a rare phenomenon caused by compensatory expansion of bone marrow in various anaemias, particularly congenital haemolytic anaemias. The mass itself almost never causes symptoms. Radiographically, extramedullary haematopoietic tissue typically produces one or more smooth, lobular or spherical masses in the paravertebral gutter, usually in the lower thorax (Fig. 14.23).The bones may be normal or may show an altered lacelike trabecular pattern due to marrow expansion. The masses are usually of homogeneous soft tissue attenuation on CT, although occasionally,

Figure 14.23 Extramedullary haematopoiesis showing smooth pleurally-based masses and altered bone texture in this patient with thalassaemia. There is also a small right pleural effusion.

a fatty component may be visible60. Usually the masses are bilateral and reasonably symmetrical.

Mesenchymal tumours and tumour-like conditions Lymphangiomas (cystic hygromas) Lymphangiomas (cystic hygromas) are focal mass-like congenital malformations of the lymphatic system comprising complex lymph channels or cystic spaces containing clear or strawcoloured fluid. Lymphangiomas can occur in any part of the mediastinum, but are most common in the anterior or superior mediastinum. Mediastinal lymphangiomas may on occasion be wholly confined to the mediastinum but they are more frequently an extension from a lymphangioma in the neck. Most cervicomediastinal lymphangiomas present in early life as a neck mass, whereas the purely mediastinal lymphangiomas usually present in older children and adults as an asymptomatic mediastinal mass. Typically they appear as cystic masses, with the attenuation of the contents close to that of water on CT61.

Fatty tumours of the mediastinum Fatty tumours of the mediastinum are rare. On chest radiography, regardless of whether they are benign or malignant, fatty tumours are seen as well-defined round or oval mediastinal masses. Benign lipomas are soft and do not, therefore, compress surrounding structures unless they are very large. On CT they show uniform fat attenuation apart from a few strands of soft tissue62. Mediastinal liposarcomas are malignant fat-containing tumours.They often occur in the anterior mediastinum where the fat appears heterogeneous on CT. In contradistinction to benign lipomas, they usually contain large areas of soft tissue density material. Lipoblastoma, a benign tumour of childhood, contains fat and soft tissue63,64. Occasionally, the amount of fat attenuation is relatively small. Angiolipoma and myelolipoma are both benign tumours which may show a combination of soft tissue and fat attenuation on CT and therefore can be indistinguishable from liposarcoma65,66. Other mesenchymal tumours, such as benign and malignant fibrous tumours and haemangiomas, may occur anywhere in the mediastinum.

255

256

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Herniation of abdominal fat Herniation of omental and perigastric fat is a common cause of a localized fatty mass in the mediastinum.The fat may herniate through the oesophageal hiatus, the foramen of Morgagni, or the foramen of Bochdalek. Such herniations are usually readily diagnosed because of their characteristic locations. On CT or MRI, appearances consistent with fat eliminate confusion with other mediastinal masses64.

Mediastinal lipomatosis Relatively large collections of fat are often present in the cardiophrenic angles, particularly in obese subjects. These cardiophrenic fat pads may resemble a mass. Massive collections of fat throughout the mediastinum may be seen in so-called ‘mediastinal lipomatosis’, a phenomenon seen particularly in Cushing’s disease, in patients on steroid therapy and in obese subjects. When the fat deposits are extensive and symmetrical, the diagnosis is usually obvious. Localized masses may also be seen and CT is helpful in these cases since it can clearly demonstrate fat attenuation throughout the mass.

Sternal and spinal disease Disorders of the stern um and spine may give rise to anterior and posterior mediastinal masses, respectively. The conditions that are particularly likely to do so are paravertebral abscess, myeloma, metastasis, traumatic haematoma, lymphoma, and primary tumours (Figs 14.24–14.26).

Aortic aneurysms and aortic arch anomalies

Figure 14.25 Sternal destruction due to a radiation-induced sarcoma following treatment for breast carcinoma many years previously. Note the thickening of the overlying skin (arrows), the radical mastectomy and pleural effusion.

Prevascular masses Almost all masses anterior to the ascending aorta and the head and neck vessels are: 1 thyroid masses 2 thymic masses 3 germ-cell tumours/cystic teratomas 4 lymphadenopathy.

The differential diagnosis of a mediastinal mass depends on the age of the patient, the location, shape, size and characteristics of the lesion, and on the number of masses present. Location in particular is important in the differential diagnosis. This is best assessed with cross-sectional imaging, in particular CT.

Thyroid masses can usually be specifically diagnosed or excluded because of their contiguity with the thyroid gland in the neck and their high CT attenuation. In addition, many show cystic areas of attenuation close to water, as well as one or more areas of discrete calcification. Almost all masses located superiorly in the anterior mediastinum which cause focal deviation of the trachea are likely to be thyroid in origin. Thymic masses and germ-cell tumours/cystic teratomas can be thought of together, as most arise within the thymus. Clinical and laboratory features may help distinguish between

Figure 14.24 Sternal metastatic deposit. CT demonstration of bone destruction by a soft tissue mass in a patient with an adenocarcinoma of unknown primary.

Figure 14.26 Sternal destruction due to direct extension from mediastinal lymphoma. Note the soft tissue swelling and obliteration of fat planes in the right-sided pectoral muscles owing to soft tissue involvement.

These important causes of mediastinal masses are discussed in Chapter 27.

Differential diagnosis of mediastinal masses

CHAPTER 14

the two, e.g. myasthenia gravis is associated with thymoma, whereas elevated human chorionic gonadotrophin levels are seen with malignant germ-cell tumours. Fat, fluid, or teeth within an anterior mediastinal mass are pathognomonic of cystic teratoma. Rarer causes of prevascular masses are parathyroid adenoma, lymphangioma (cystic hygroma), pericardial cyst, aortic body chemodectoma, lipoma/liposarcoma or other mesenchymal tumours, or aneurysms. Many of these masses have features that permit a specific diagnosis to be made: parathyroid adenomas are usually associated with hyperparathyroidism; lymphangiomas almost always have broad contact with the root of the neck and show numerous areas of nonenhancing, water or near-water attenuation on CT; lipomas show uniform fat attenuation, apart from a few soft tissue strands; liposarcomas show an unusual mixture of fat interspersed by irregular strands or masses of soft tissue attenuation; aneurysms should be recognized by luminal enhancement; pericardial cysts are, in general, of uniform water attenuation with a thin wall of uniform thickness, and need only be considered when the mass in question is in contact with the pericardium. Mesenchymal tumours such as fibrosarcomas or haemangiomas have no distinguishing features.

Paracardiac masses The likely diagnoses for paracardiac masses in direct contact with the diaphragm are pericardial cyst, diaphragmatic hernia, fat pad or lymphadenopathy. If the mass is separated from the diaphragm, the likely differential diagnosis widens to include germ-cell, mesenchymal and pericardial tumours, cystic teratomas and thymic masses. Approximately 20% of thymomas are found in a paracardiac location, though contact with the diaphragm is very unusual. Lack of connection with the diaphragm eliminates the possibility of a diaphragmatic hernia.

Masses in the paratracheal, subcarinal and paraoesophageal regions These three regions are contained within a common fascial sheath which continues into the neck. The likely possibilities for a mass in these locations are lymphadenopathy, intrathoracic thyroid mass, developmental foregut cyst, oesophageal tumour, hiatus hernia, paraspinal mass encroaching on the middle mediastinum and aortic aneurysm. Aneurysms are readily diagnosed on CT by observing contrast enhancement of the lumen.The nature of other masses can often be predicted with reasonable certainty: nonvascular masses in the aortopulmonary window or deep to the azygos vein are almost invariably enlarged lymph nodes; bronchogenic cysts can be diagnosed with confidence if the criteria of a thin wall and contents exhibiting uniform water attenuation are met; and splitting of the trachea from the oesophagus is a characteristic shared only by thyroid masses, bronchogenic cysts, oesophageal tumours and an aberrant origin of the left pulmonary artery. Patients with oesophageal carcinoma (the most common oesophageal tumour) nearly always present with dysphagia when the tumour mass is still relatively small, whereas a leiomyoma or other mesenchymal tumour of the oesophagus may occasionally present first as an asymptomatic mediastinal mass. The intimate

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

relationship with the oesophagus usually leads to a barium swallow examination. Hiatus hernia is an exceedingly common cause of enlargement of the mediastinum in the region of the lower oesophagus and plain radiography is so characteristic that a barium swallow is rarely required for diagnosis.

Paravertebral masses Neurogenic lesions and neoplastic lymphadenopathy dominate the differential diagnosis for paravertebral masses. The less common causes of paravertebral masses include: extramedullary haematopoiesis; pancreatic pseudocyst; mesenchymal tumours such as lipoma, fibroma and haemangioma; and lesions arising from the oesophagus, pharynx, spine, or aorta. The oesophageal or pharyngeal lesions that may project posteriorly include leiomyoma, foregut duplication cyst, and congenital or acquired diverticula of the oesophagus. The spinal origin of masses such as paraspinal abscess, primary or metastatic tumours of the vertebral body, or haematoma from trauma to the spine, are usually readily diagnosed by observing corresponding changes in the spine.Aneurysms of the descending aorta that truly mimic a mediastinal mass are uncommon, as the majority of large aneurysms in this location are obvious dilatations of the descending aorta and have curvilinear calcification in their wall. An aneurysm is readily diagnosed on CT when opacification of its lumen can be demonstrated.

OTHER MEDIASTINAL LESIONS Acute mediastinitis Acute infection of the mediastinum is rare. Oesophageal perforation, either iatrogenic or from swallowed objects, is the most frequent cause. Forceful vomiting may tear the oesophageal wall (Boerhaave’s syndrome) and if the tear is deep enough, air, alimentary juices and food may leak into the mediastinum causing acute mediastinitis. Such tears are almost invariably just above the gastro-oesophageal junction. Other causes of acute mediastinal infection are leakage from the oesophagus into the mediastinum through a necrotic neoplasm, and extension of infection from the neck, retroperitoneum, or adjacent intrathoracic or chest wall structures into the mediastinum. Clinically, the patients are often very ill with an abrupt onset of high fever, tachycardia and chest pain. The chest radiograph may show widening and lack of clarity of the mediastinal outline adjacent to the oesophagus. Streaks or round collections of air may be seen within the mediastinum, and there may even be one or more mediastinal air–fluid levels. Pleural effusions are frequent and are usually confined to, or greatest on, the left. Lower lobe pneumonia or atelectasis often complicate the radiographic picture. A swallow using nonionic contrast medium may show the site of perforation, with extravasation into the mediastinum. CT shows obliteration of the normal mediastinal fat planes and gas bubbles may be identified within the mediastinum (Fig. 14.27). In advanced cases there may be walledoff discrete fluid or air–fluid collections indicating abscess formation. There may be an associated empyema, subphrenic or pericardial collection. When acute mediastinitis

257

258

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 14.27 Abscess formation. (A) Abscess in anterior mediastinum demonstrated on CT. (B) This coronal reformat of an axial CT dataset demonstrates a tuberculous mediastinal abscess and associated lung changes in a different patient.

is suspected following sternotomy, CT shows the extent of inflammation and any drainable mediastinal or pericardial fluid collections67. Distinguishing a retrosternal haematoma from reactive granulation tissue or cellulitis is difficult, as is distinguishing osteomyelitis from the direct effects of the surgical incision68. It should be remembered that substernal fluid collections and dots of air are normal in the first 20 d following sternotomy. Therefore, before gas-forming infections can be diagnosed, the air collections must appear de novo or must progressively increase in the absence of any other explanation69.

Fibrosing mediastinitis Fibrosing mediastinitis (sclerosing mediastinitis or mediastinal fibrosis) is a disorder that results in proliferation of fibrous tissue and collagen within the mediastinum. It is usually due to

previous infection from histoplasmosis or tuberculosis70. The fibrosis is usually maximal in the upper mediastinum but may extend to the lung roots. The most common clinical consequences are obstruction to the superior vena cava and, occasionally, obstruction to the central pulmonary arteries or veins. Other causes of fibrosing mediastinitis include idiopathic (similar to retoperitoneal fibrosis/peri-aortitis), autoimmune disease, radiation therapy and drugs (in particular methysergide). The chest radiograph is non-specific and often underestimates the extent of mediastinal disease. In fibrosing mediastinitis due to previous tuberculous or fungal infection, the chest radiograph may show calcification of mediastinal or hilar lymph nodes. CT typically shows an infiltrative, often extensively calcified, hilar or mediastinal process (Fig. 14.28), which may be relatively focal when disease is due to previous histoplasmosis or tuberculosis, and more diffuse in the idiopathic form71. Airway

Figure 14.28 Mediastinitis. (A) Fibrosing mediastinitis. There is confluent soft tissue infiltration throughout the mediastinum without evidence of a discrete mass. Note the marked narrowing of the superior vena cava (SVC). The patient had clinical evidence of SVC compression and a history of previous radiotherapy for lymphoma, diagnosed by surgical biopsy through a median sternotomy. This original biopsy, 30 years previously, had been complicated by post-operative infection. (B) Tracheal narrowing from mediastinal fibrosis of unknown cause in a different patient. The trachea (arrow) is markedly narrowed and distorted and lies within the fibrotic scarring. The more posterior oesophagus is relatively dilated and gas filled.

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

narrowing (Fig. 14.28), vascular encasement and obstruction may also be seen. MRI provides similar information to CT, with the fibrosis appearing as heterogeneous signal intensity on T1and T2-weighted imaging72, but lacks sensitivity for detection of calcification, which is an important feature for differentiating fibrosing mediastinitis from other infiltrative disorders of the mediastinum, such as lymphoma and metastatic carcinoma.

Mediastinal haemorrhage Mediastinal haemorrhage is most commonly due to trauma to the arteries and veins within the mediastinum, with other causes including rupture of an aneurysm, aortic dissection and complications of central venous catheterization. Radiologically haemorrhage produces an increase in the mediastinal diameter which is maximal at the point of bleeding73. Blood may track through the mediastinum, frequently running over the apex of the left lung to produce a smooth and well-defined apical cap. When haemorrhage is severe blood may rupture into the pleural cavity or dissect into lung along peribronchovascular sheaths, resulting in a radiographic pattern resembling interstitial oedema. On unenhanced CT, acute haemorrhage may appear of relative high attenuation. The appearance of mediastinal haematoma on MRI varies with the age of the haemorrhage.

Mediastinal emphysema Air may enter the mediastinum from a perforation of the pharynx, oesophagus, or major airways. In many instances, however, a pneumomediastinum is the result of an air leak from a tear in a small intrapulmonary airway, the air dissecting through the lung via the hilum into the mediastinum. Asthma is the most common precipitating cause. In other cases the leak is probably related to abrupt changes in intrathoracic pressure such as those associated with vomiting. Occasionally, air tracks into the mediastinum from retroperitoneal air collections. The presence of a pneumomediastinum is, in itself, of little significance (though it may be responsible for substernal chest pain), but the condition causing the air leak (particularly bronchial, oesophageal, or pharyngeal perforation) may be of great significance to the patient. On imaging the condition is recognized as streaky translucencies within the mediastinum that are usually most obvious adjacent to the left heart border, aortic knuckle, main pulmonary artery and adjacent left hilum (Fig. 14.29). The air dissects through

Figure 14.29 Mediastinal emphysema in a patient with Pneumocystis jiroveci (formerly carinii) pneumonia. The air has tracked through the mediastinum into the neck and chest wall.

the perivascular areolar tissues and may track up into the neck, supraclavicular areas and axillae, as well as down into the retroperitoneum. It may also track extraserosally on either side of the diaphragm, which will occasionally be seen as a continuous line of transradiancy known as the ‘continuous diaphragm sign’74. The differential diagnosis of a pneumomediastinum on chest radiograph includes a medially placed pneumothorax and a ‘Mach effect’ due to the abrupt change in density between the lung and the adjacent heart and mediastinum. It is easy to appreciate why a medial pneumothorax can be mistaken for a pneumomediastinum, since in both instances there is a linear collection of air bounded on its lateral side by a thin line of pleura. Deciding whether the line is mediastinal parietal pleura or visceral pleura can be difficult; the distinction often depends on recognizing the full extent of the air and looking carefully for a pneumothorax lying against the chest wall, or looking for evidence of air elsewhere in the mediastinum. Pneumomediastinum is easy to diagnose on CT as streaks or rounded collections of air surrounding the vessels and other structures within the mediastinum.

THE PERICARDIUM NORMAL ANATOMY The pericardium is a fibrous bag consisting of the (inner) visceral pericardium, the (outer) parietal pericardium, and the 20–60-ml cavity between them (Fig. 14.30). Beneath the visceral pericardium is either myocardium or epicardial fat. The visceral pericardium extends for short distances along the pulmonary veins, the superior vena cava below the azygos vein, the inferior vena cava, and the ascending aorta to

a point 20–30 mm above the aortic root and the main pulmonary artery as far as its bifurcation. It then reflects upon itself to become the parietal pericardium. The reflection of the two pericardial layers around the great arteries and veins forms the two ‘appendages’ or diverticula of the pericardium. The arterial mesocardium is an anterior extension of all layers of the pericardium around the ascending aorta and pulmonary trunk. It lies obliquely in the coronal plane and is higher on the right, over the ascending aorta, than on the

259

260

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Superior vena cava

Transverse pericardial sinus

Ascending aorta Pulmonary trunk

Left pulmonary veins Right pulmonary veins Oblique pericardial sinus Inferior vena cava

Diaphragm

Pericardium (cut edge)

Diaphragmatic part of pericardium

A Figure 14.30 The normal pericardium. (A) Drawing of the reflections of the normal pericardial sac. (B) The normal pericardium on narrow-section CT is demonstrated as a thin (1–2 mm) soft tissue density line separated from the cardiac muscle by the epicardial fat (arrows).

left, over the pulmonary trunk. The arterial mesocardium is divided into an anterior compartment (the pre-aortic recess), lying between the ascending aorta and pulmonary trunk, and another compartment behind the ascending aorta immediately above the right pulmonary artery (the retroaortic recess) (Fig. 14.31). Posterior and lateral to the heart, the extraparenchymal pulmonary veins and the superior and inferior venae cavae are enveloped by the venous mesocardium, which has the shape of an inverted U. Beneath the apex of the U is the space between the pulmonary veins and the left atrium—the oblique sinus. The arterial and venous

mesocardia are connected by the transverse sinus. This space is limited anteriorly by the aorta and pulmonary arteries, and posteriorly by the superior vena cava and left atrium. It forms a communication, at the base of the heart, between the right and left sides of the pericardial cavity. The pericardium is considered to support the heart and cardiac function in three major ways: it contains the heart and limits its motion within the middle mediastinum; it acts as a protective membrane to shield the heart from local inflammatory disease; and it limits excessive acute dilatation of the cardiac chambers in response to increased preload75.

Figure 14.31 Contrast-enhanced CT demonstrating normal pericardial recesses. (A) The retro-aortic recess appears as a crescentic structure closely apposed to the posterior wall of the aorta (arrows). It is characteristically of slightly lower attenuation than the unenhanced blood pool. In this example of a patient in mild cardiac failure there is slightly more fluid than is usually seen. (B) Pericardial fluid in the anterior compartment (pre-aortic recess) (arrows).

CHAPTER 14

On the lateral chest radiograph the normal pericardium may be seen as a 1–2-mm thick curved stripe anterior to the heart, set between more radiolucent mediastinal fat anteriorly and epicardial fat posteriorly. The visceral pericardium is normally very thin and is therefore not visualized separately by any imaging modality. The combination of the visceral pericardium and the small layer of physiological pericardial fluid constitutes the normal pericardium routinely visualized on CT and MRI as a 1–2-mm thick layer76,77 (Fig. 14.30B), which can appear focally thicker at the sites of its major attachments. Although the pericardium is readily visualized overlying the right atrium and right ventricle in most individuals, it is often not visible over the lateral and inferior walls of the left ventricle. It is essential to appreciate the anatomical extent and location of the pericardial recesses78 since they are frequently seen on both CT79,80 and MRI and may be confused with aortic dissection, adenopathy, a mediastinal mass, or thymus.

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

collections and masses. Limitations of MRI include its inability to reliably depict calcification, relatively longer data acquisition times, and greater demands on the patient with regard to cooperation with breath-holding. Additionally, arrhythmias, which commonly occur in association with pericardial disease, may detrimentally affect image acquisition and quality.

DEVELOPMENTAL ANOMALIES Congenital absence of the pericardium

Computed tomography is the best investigation for localizing, characterizing and demonstrating the extent of a collection or mass in the acute setting. Multidetector CT (MDCT) with multiplanar reformats, particularly if ECG gated, provides excellent motion-free assessment of the pericardium. MDCT has the advantage of speed and generally greater availability and accessibility. It is highly sensitive for the detection of pericardial calcification, an important finding in constrictive pericarditis. Pericardial assessment with CT may be limited by motion artefact if not ECG gated; additionally, differentiating pericardial fluid from thickening may occasionally be difficult on CT.

Compromise of the vascular supply to the pleuropericardial membrane during embryological development is associated with congenital defects in the pericardium. Pericardial deficiency is associated with congenital anomalies of the heart and lungs, including atrial septal defect, tetralogy of Fallot, patent ductus arteriosus, bronchogenic cysts and pulmonary sequestration81. The defects vary in size from small communications between the pleural and pericardial cavities to complete (bilateral) absence of the pericardium. The most common form is complete absence of the left pericardium, with preservation of the pericardium on the right. Bilateral and isolated right-sided lesions are very rare. Pericardial defects are frequently associated with large defects in the parietal pleura, through which the left lung can herniate and surround the intrapericardial vascular structures82,83. Complete absence of the pericardium is usually asymptomatic, whereas partial or localized absence of the pericardium may be complicated by herniation and entrapment of a cardiac chamber, in particular the left atrial appendage in left-sided defects. Chest radiograph findings are frequently subtle and nonspecific82,83. When present, chest radiograph findings in complete absence of the left pericardium include displacement of the heart into the left chest and interposition of lung between the aorta and pulmonary artery, as well as between the left hemidiaphragm and cardiac silhouette. Both the medial and lateral borders of the main pulmonary artery may be visualized more clearly due to absence of the anterior pericardial reflection between the aorta and the pulmonary artery. Due to leftward displacement and rotation, the right cardiac border may not be seen. In partial pericardial defects, varying degrees of prominence of the pulmonary artery and/or left atrial appendage may be seen, while the heart retains its normal position in the thorax82 (Figs 14.32, 14.33). A definitive diagnosis of absence of the pericardium can be obtained with either CT or MRI.

Magnetic resonance imaging

Pericardial cysts and diverticula

Magnetic resonance imaging can provide a comprehensive assessment of the pericardium. When T1- and T2-weighted imaging sequences (a number of which can be performed using ECG-gated breath-hold techniques), are combined with gradient-echo cine-based functional cardiac imaging, both pericardial disease and its impact on cardiac function can be assessed. MRI has some advantages over echocardiography and CT in the detection and characterization of pericardial

Pericardial cysts and diverticula are thought to be the result of persistence of blind-ending ventral parietal pericardial recesses. Those cysts that communicate with the pericardial space are termed pericardial diverticula. They almost invariably appear as a well-defined, oval or occasionally lobulated mass attached to the pericardium84 (Fig.14.34). Most occur in the right cardiophrenic angle, with a proportion being situated in the left cardiophrenic angle and some higher in the

IMAGING PERICARDIAL DISEASE Chest radiography Chest radiography is of limited use in the detailed assessment of pericardial disease although pericardial effusions, calcification and secondary signs and complications of pericardial disease may be evident.

Ultrasound Ultrasound is the most commonly used investigation in the initial evaluation of pericardial disease. Restricted acoustic windows limit its evaluation of the entire pericardium. Loculated collections, intrapericardial blood clot and pericardial thickening in particular may be difficult to assess or may be overlooked.

Computed tomography

261

262

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 14.32 Complete absence of the pericardium. (A) PA radiograph. The heart is displaced into the left chest, obscuring the right heart border by the spine. The cardiac apex (arrow) is elevated, and air-filled lung is seen beneath it. (B) On lateral barium swallow, the increased density of the left ventricle surrounded by the air-filled lung (arrows) is apparent.

Figure 14.33 Partial absence of the pericardium. Axial spin-echo MRI. (A) Image through the aortic valve and proximal ascending aorta (Ao). The heart is displaced into the left chest and rotated in a clockwise manner. (B) Image 1-cm cephalad through the pulmonary valve (PV). A sliver of lung (arrow) invaginates to come into contact with the ascending aorta. (C) Image 1-cm cephalad to the main (MP) and transverse right (RP) pulmonary arteries. The MP protrudes to the left and is in contact with the lung.

Figure 14.34 Pericardial cyst. Axial spin-echo MRI at the base of the heart. (A) An intermediate signal intensity smooth mass extrinsic to the heart is identified (arrow). Ao = ascending aorta, LA = left atrium, PA = main pulmonary artery, S = superior vena cava. (B) Chest radiograph, lateral view. The smooth lobulated density superimposed on the cardiac silhouette (arrows) is a pericardial cyst.

PA Ao S LA

A

CHAPTER 14

mediastinum. They contain clear fluid and can be recognized as fluid-filled cysts surrounded by normal pericardium on echocardiography, CT, or MRI85, the cyst contents showing characteristics similar to water.

Intrapericardial bronchogenic cysts Intrapericardial bronchogenic cysts are rare. They may cause symptoms secondary to mass effect on adjacent cardiac structures. Appearances are similar to bronchogenic cysts occurring elsewhere in the mediastinum.

ACQUIRED PERICARDIAL DISEASE Pericarditis including pericardial effusion Inflammation of the pericardium (pericarditis) may occur in response to a variety of insults. It typically results in cellular proliferation, or the production of fluid (pericardial effusion) or fibrin, either alone or in combination. Causes include myocardial infarction (acute or post-myocardial infarction [Dressler] syndrome), pericardiotomy, mediastinal irradiation, infection (viral or bacterial), connective tissue disease (rheumatoid arthritis, systemic lupus erythematosus), metabolic disorders (uraemia, hypothyroidism), neoplasia and AIDS. The most common imaging manifestation of acute pericarditis is a pericardial effusion, the nature of the fluid varying with the underlying cause (Fig. 14.35). Transudative pericardial effusions may develop after cardiac surgery or in congestive heart failure, uraemia, post-pericardiectomy syndrome, myxoedema and collagen–vascular diseases. Haemopericardium may be due to trauma, aortic dissection, aortic rupture, or neoplasm (especially primary pericardial mesothelioma) (Fig. 14.35). Chylopericardium resulting from injury or obstruction of the thoracic duct is rare. On chest radiograph sudden increase in the size of the cardiac silhouette without specific chamber enlargement suggests the diagnosis of pericardial effusion. Filling in of the retrosternal space, effacement of the normal cardiac borders, development of a ‘flask’ or ‘water bottle’ cardiac conFiguration, and bilateral hilar overlay are features of pericardial

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

effusion (Fig. 14.36). The epicardial fat pad ‘sign’ is positive when, visualized in the lateral projection, an anterior pericardial stripe (bordered by epicardial fat posteriorly and mediastinal fat anteriorly) is thicker than 2 mm. This sign is diagnostic of pericardial thickening or fluid86,87. Echocardiography is the most commonly used method for diagnosing pericardial effusion. It is highly sensitive and specific although visualization may be limited in patients with marked obesity or emphysema, and loculated collections and intrapericardial clot in postoperative patients may be difficult to detect88. CT and MRI are indicated when a loculated or haemorrhagic effusion or pericardial thickening is suspected. Increased attenuation in a pericardial effusion on CT suggests haemorrhage89 (see Fig. 14.35), although occasionally the attenuation values of a haemorrhagic effusion may overlap with the pericardial fluid found in patients with hypothyroidism. On spin-echo MRI, the signal characteristics of pericardial collections vary depending on the composition of the fluid. In the absence of haemorrhage, effusions are typically of predominantly low signal intensity, although intermediate signal intensity may be seen in inflammatory conditions such as uraemia, tuberculosis, or trauma, possibly reflecting high protein content and when more focal, the presence of adhesions limiting normal flow of pericardial fluid in the pericardial space77,85. In haemorrhagic effusions, signal intensity varies depending on the age of blood products. Thickened inflamed pericardium can appear of moderate to high signal intensity on spin-echo MRI, and pericardial enhancement may be seen on both MRI or CT performed after intravenous contrast medium administration.

Constrictive pericarditis Any insult to the pericardium can progress from an acute pericarditis with pericardial effusion to a subacute stage of resorption of the effusion with organization, and then to a chronic phase of fibrous scarring, pericardial thickening and obliteration of the pericardial cavity. Constrictive pericarditis is the condition in which a thickened, fibrotic and often calcified pericardium restricts diastolic filling of the heart (Fig. 14.37).

Figure 14.35 Pericardial effusion. (A) Small pericardial effusion is present in the anterior pericardial sac in a patient with advanced pulmonary artery hypertension. Note the dilated right-sided cardiac chambers. (B) Large haemopericardium complicating a type A aortic dissection. This is an unenhanced image and the haemopericardium is the same density as soft tissue structures (compare to A). (C) The dissection flap can be seen on this enhanced CT within the transverse arch.

263

264

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 14.36 Large pericardial effusion. (A) The heart had become rapidly enlarged in this patient who had previously undergone aortic valve replacement. (B) Lateral view demonstrates the pleural fluid lying posteriorly. (C) Unenhanced CT through the level of the valve replacement demonstrates the large pericardial effusion.

Figure 14.37 Dense pericardial calcification demonstrated on (A,B) chest radiograph (arrows) and (C) CT. There are bilateral pleural effusions in this patient with constrictive calcific pericarditis due to previous tuberculosis.

Constrictive pericarditis is usually the result of a chronic pericardial insult. The aetiology is unknown in many cases, presumed to be secondary to an occult viral pericarditis and other causes of pericarditis90. Outside the USA, the most common cause is probably tuberculosis or fungal aetiology. Constriction due to neoplastic infiltration of the pericardium is most commonly secondary to carcinoma of the lung or breast, lymphoproliferative malignancies and melanoma. Pericardial constriction after mediastinal irradiation, usually performed to treat breast carcinoma or Hodgkin’s disease, may occur months to years after treatment91. Pericardial thickening is seen in up to 88% of confirmed cases of constrictive pericarditis92. In the majority of cases, constrictive pericarditis involves the entire pericardium, restricting filling of all cardiac chambers. Occasionally, in particular anterior to the right ventricle in postoperative patients, the pericardial thickening is more localized78. Patients with constrictive pericarditis frequently present with symptoms of heart failure such as dyspnoea, orthopnoea and fatigability; they may occasionally present with hepatomegaly and ascites. The clinical findings of constriction overlap with those of restrictive cardiomyopathy, a primary disorder of the myocardium. The differential diagnosis is important since the patients with pericardial constriction may benefit from

pericardiectomy, while restrictive cardiomyopathy is managed medically or by cardiac transplantation. The hallmarks of pericardial constriction are pericardial thickening, pericardial calcification and abnormal diastolic ventricular function. On chest radiograph the heart size may be normal or appear increased due to the presence of pericardial fluid. The superior vena cava and azygos vein may be of increased size due to raised right heart pressures. Although echocardiography is routinely performed and provides an excellent assessment of haemodynamic function, it is not highly accurate at depicting pericardial thickening. CT and MRI are significantly more sensitive, with CT having the added advantage over MRI of being able to demonstrate the presence of calcification, which is associated with pericardial constriction (see Fig. 14.37C). Pericardial thickening of 4 mm or more is abnormal and, when accompanied by clinical features of constriction, is highly suggestive of constrictive pericarditis85,92. Both CT and MRI may show the secondary effects of constriction on the central cardiovascular structures. The right ventricle tends to be of reduced volume and has a narrow tubular configuration. A sigmoid-shaped interventricular septum or prominent leftward convexity of the septum may be seen78. The right atrium,

CHAPTER 14

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

superior and in particular inferior venae cavae, and hepatic veins may be dilated. Hepatomegaly and ascites may be seen. Cardiac MRI can also be used to provide a more detailed assessment of cardiac function and myocardial wall thickness, which has prognostic implications for outcome after pericardiectomy.

Cardiac tamponade Gradual accumulation of pericardial fluid may fail to produce clinical signs or symptoms for an extended period of time. However, rapid accumulation of as little as 100–200 ml of fluid can cause a haemodynamically significant compression of the heart, which severely impedes diastolic filling, the condition known as pericardial tamponade. Even in the face of preserved ejection fraction, diminished ventricular end-diastolic volume leads to reduced stroke volume. In addition to other clinical signs of pericardial effusion, those of cardiac tamponade include pulsus paradoxus, which is exaggeration of the normal inspiratory drop in systolic blood pressure. Since acute tamponade may occur with small effusions, clinically important pericardial enlargement may be difficult to detect on plain radiographs. Subtle changes in cardiac contour may only be detectable by comparison with previous studies. If there is decreased pulmonary vascularity in spite of the cardiac enlargement or if the superior vena cava and azygos veins are dilated, tamponade should be suspected. Echocardiographic demonstration of pericardial effusion and the clinical findings are usually sufficient to make the diagnosis of tamponade. CT and MRI are frequently instrumental in suggesting the cause of the effusion (i.e. haemorrhage, neoplastic involvement, inflammation due to tuberculosis, or other infectious processes, etc).

Pericardial neoplasms Primary pericardial neoplasms are rare, with approximately equal incidence of benign versus malignant pericardial neoplasms. Benign tumours include teratomas, fibromas, neurofibromas, lipomas, haemangiomas, lymphangiomas and hamartomas. Although these patients are usually symptom free, pericardial effusion or constriction, particularly in the case of childhood teratomas, may occur.

Figure 14.38 Phaeochromocytoma of the heart. There is abnormal soft tissue in the right atrioventricular groove (arrows) that proved to be a primary cardiac phaeochromocytoma on histological examination following resection.

in approximately 10% of all patients with malignancy94. The most common malignancies encountered are lung, lymphoma, breast, leukaemia, stomach, melanoma, liver and colon94 (Fig. 14.39). A pericardial effusion is the most common finding in pericardial malignancy, whether primary pericardial or metastatic. Intrapericardial neoplasms tend to compress and deform normal intrapericardial structures, whereas extrapericardial masses tend to displace the intrapericardial structures without compression or distortion. Chest radiographs are often abnormal, but are non-specific. Alteration of fat-pad contours, cardiac enlargement, mediastinal widening, hilar adenopathy, or a hilar mass may be seen. Echocardiography is usually the initial technique for evaluation of a suspected pericardial neoplasm, with MRI and CT being useful for further evaluation. Both MRI and CT are excellent at providing information regarding the size, location and extent of pericardial neoplasms, but are not tissue specific. Fatty tumours (lipomas, fat-containing teratomas) are the exception, due to their typically low attenuation on CT and increased signal intensity on spin-echo T1-weighted MRI. Fatty tumours must be differentiated from the focal

Malignant mesothelioma Malignant mesothelioma is the most common primary pericardial malignancy. A causal relationship between it and asbestosis is uncertain because of the low prevalence of this neoplasm. Mesothelioma may present as a well-defined single mass, multiple nodules, or diffuse plaques involving the visceral and parietal pericardium and wrapping around the cardiac chambers and great vessels. Clinically it presents with haemorrhagic effusion and tamponade, congestive heart failure, arrhythmia and occasionally pericardial constriction93. Other malignant primary tumours include lymphoma, sarcoma, phaeochromocytoma and liposarcoma (Fig. 14.38).Teratomas of the pericardium may also be malignant and are most commonly seen in children.

Pericardial metastases Pericardial metastases are much more common than primary pericardial neoplasms. They are identified at autopsy

Figure 14.39 Metastatic melanoma. There is a mass within the right atrial cavity that demonstrates high signal intensity on this T1-weighted image in keeping with the known diagnosis of metastatic malignant melanoma (arrows). Note the enlarged left axillary lymph nodes.

265

266

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

deposits of subepicardial fat and non-neoplastic lesions that can simulate fatty tumours, such as mesenteric fat in a hiatal hernia. Metastatic melanoma may have high signal intensity on T1- and T2-weighted images (see Fig. 14.40)95, a feature that may be useful in differentiating it from other metastatic neoplasms, which are frequently of low signal intensity on T1and high signal intensity on T2-weighted images96. In addition to discrete masses and effusions, metastatic involvement of the pericardium may cause focal or diffuse pericardial thickening, which may be irregular and usually enhances. Primary lipoma, liposarcoma and lymphoma of the pericardium typically appear as large heterogeneous masses frequently associated with a serosanguinous pericardial effusion93.

REFERENCES 1. Benjamin S P, McCormack L J, Effler D B, Groves L K 1972 Primary tumors of the mediastinum. Chest 62: 297–303 2. Wychulis A R, Payne W S, Clagett O T, Woolner L B 1971 Surgical treatment of mediastinal tumors: a 40 year experience. J Thorac Cardiovasc Surg 62: 379–391 3. Cohen A J, Thompson L N, Edwards F H et al 1991 Primary cysts and tumors of the mediastinum. Ann Thorac Surg 51: 378–386 4. Friedberg J W, Chengazi V 2003 PET scans in the staging of lymphoma: current status. Oncologist 8: 438–447 5. Verboom P, van Tinteren H, Hoekstra O S et al 2003 Cost-effectiveness of FDG-PET in staging non-small cell lung cancer: the PLUS study. Eur J Nucl Med Mol Imaging 30: 1444–1449 6. Glazer G M, Axel L, Moss A 1982 CT diagnosis of mediastinal thyroid. Am J Roentgenol 138: 495–498 7. Bashist B, Ellis K, Gold R P 1983 Computed tomography of intrathoracic goiter. Am J Roentgenol 140: 455–460 8. Lee V S, Spritzer C E, Coleman R E, Wilkinson R H Jr, Coogan A C, Leight G S Jr 1996 The complementary roles of fast spin-echo MR imaging and double-phase 99mTc-sestamibi scintigraphy for localization of hyperfunctioning parathyroid glands. Am J Roentgenol 167: 1555–1562 9. Lee V S, Spritzer C E 1998 MR imaging of abnormal parathyroid glands. Am J Roentgenol 170: 1097–1103 10. Baron R L, Sagel S S, Baglan R J 1981 Thymic cysts following radiation therapy for Hodgkin disease. Radiology 141: 593–597 11. Souadjian J V, Enriquez P, Silverstein M N, Pepin J M 1974 The spectrum of diseases associated with thymoma. Coincidence or syndrome? Arch Intern Med 134: 374–379 12. Strollo D C, Rosado-de-Christenson M L 1999 Tumors of the thymus. J Thorac Imaging 14: 152–171 13. Brown L R, Muhm J R, Sheedy P F, Unni K K, Bernatz P E, Hermann R C Jr 1983 The value of computed tomography in myasthenia gravis. Am J Roentgenol 140: 31–35 14. Ellis K, Austin J H, Jaretzki A III 1988 Radiologic detection of thymoma in patients with myasthenia gravis. Am J Roentgenol 151: 873–881 15. Chen J L, Weisbrod G L, Herman S J 1988 Computed tomography and pathologic correlations of thymic lesions. J Thorac Imaging 3: 61–65 16. Rosado-de-Christenson M L, Galobardes J, Moran C A 1992 Thymoma: radiologic–pathologic correlation. RadioGraphics 12: 151–168 17. Lee J D, Choe K O, Kim S J, Kim G E, Im J G, Lee J T 1991 CT findings in primary thymic carcinoma. J Comput Assist Tomogr 15: 429–433 18. Do Y S, Im J G, Lee B H et al 1995 CT findings in malignant tumors of thymic epithelium. J Comput Assist Tomogr 19: 192–197 19. Chew F S, Weissleder R 1991 Mediastinal thymolipoma. Am J Roentgenol 157: 468 20. Rosado-de-Christenson M L, Pugatch R D, Moran C A, Galobardes J 1994 Thymolipoma: analysis of 27 cases. Radiology 193: 121–126

21. Choyke P L, Zeman R K, Gootenberg J E, Greenberg J N, Hoffer F, Frank J A 1987 Thymic atrophy and regrowth in response to chemotherapy: CT evaluation. Am J Roentgenol 149: 269–272 22. Kissin C M, Husband J E, Nicholas D, Eversman W 1987 Benign thymic enlargement in adults after chemotherapy: CT demonstration. Radiology 163: 67–70 23. Cohen M, Hill C A, Cangir A, Sullivan M P 1980 Thymic rebound after treatment of childhood tumors. Am J Roentgenol 135: 151–156 24. Moran C A, Suster S 1997 Primary germ cell tumors of the mediastinum: I. Analysis of 322 cases with special emphasis on teratomatous lesions and a proposal for histopathologic classification and clinical staging. Cancer 80: 681–690 25. Moeller K H, Rosado-de-Christenson M L, Templeton P A 1997 Mediastinal mature teratoma: imaging features. Am J Roentgenol 169: 985–990 26. Strollo D C, Rosado de Christenson M L, Jett J R 1997 Primary mediastinal tumors. Part 1: tumors of the anterior mediastinum. Chest 112: 511–522 27. Knapp R H, Hurt R D, Payne W S et al 1985 Malignant germ cell tumors of the mediastinum. J Thorac Cardiovasc Surg 89: 82–89 28. Lee K, Im J, Han M, Kim C, Kim W 1989 Malignant primary germ cell tumors of the mediastinum: CT features. Am J Roentgenol 153: 947–951 29. Rosado-de-Christenson M L, Templeton P A, Moran C A 1992 From the archives of the AFIP. Mediastinal germ cell tumors: radiologic and pathologic correlation. RadioGraphics 12: 1013–1030 30. Radin D R, Baker E L, Klatt E C et al 1990 Visceral and nodal calcification in patients with AIDS-related Pneumocystis carinii infection. Am J Roentgenol 154: 27–31 31. Pombo F, Rodriguez E, Mato J, Perez-Fontan J, Rivera E, Valvuena L 1992 Patterns of contrast enhancement of tuberculous lymph nodes demonstrated by computed tomography. Clin Radiol 46: 13–17 32. Landay M J, Rollins N K 1989 Mediastinal histoplasmosis granuloma: evaluation with CT. Radiology 172: 657–659 33. Yousem D M, Scatarige J C, Fishman E K, Siegelman S S 1986 Lowattenuation thoracic metastases in testicular malignancy. Am J Roentgenol 146: 291–293 34. Hopper K D, Diehl L F, Cole B A, Lynch J C, Meilstrup J W, McCauslin M A 1990 The significance of necrotic mediastinal lymph nodes on CT in patients with newly diagnosed Hodgkin disease. Am J Roentgenol 155: 267–270 35. Samuels T, Hamilton P, Shaw P 1990 Whipple disease of the mediastinum Am J Roentgenol 154: 1187–1188 36. Miller B H, Rosado-de-Christenson M L, McAdams H P, Fishback N F 1995 Thoracic sarcoidosis: radiologic–pathologic correlation. RadioGraphics 15: 421–437 37. Bein M E, Putman C E, McLoud T C, Mink J H 1978 A reevaluation of intrathoracic lymphadenopathy in sarcoidosis. Am J Roentgenol 131: 409–415 38. Castellino R A, Hilton S, O’Brien J P, Portlock C S 1996 Non-Hodgkin lymphoma: contribution of chest CT in the initial staging evaluation. Radiology 199: 129–132 39. Castellino R A, Blank N, Hoppe R T, Cho C 1986 Hodgkin disease: contributions of chest CT in the initial staging evaluation. Radiology 160: 603–605 40. Jochelson M S, Balikian J P, Mauch P, Liebman H 1983 Peri- and paracardial involvement in lymphoma: a radiographic study of 11 cases. Am J Roentgenol 140: 483–488 41. McLoud T C, Kalisher L, Stark P, Greene R 1978 Intrathoracic lymph node metastases from extrathoracic neoplasms. Am J Roentgenol 131: 403–407 42. McAdams H P, Rosado-de-Christenson M, Fishback N F, Templeton P A 1998 Castleman disease of the thorax: radiologic features with clinical and histopathologic correlation. Radiology 209: 221–228 43. Yamashita Y, Hirai T, Matsukawa T, Ogata I, Takahashi M 1993 Radiological presentations of Castleman’s disease. Comput Med Imaging Graph 17: 107–117

CHAPTER 14

44. Strollo D C, Rosado-de-Christenson M L, Jett J R 1997 Primary mediastinal tumors: part II. Tumors of the middle and posterior mediastinum. Chest 112: 1344–1357 45. Nakata H, Nakayama C, Kimoto T et al 1982 Computed tomography of mediastinal bronchogenic cysts. J Comput Assist Tomogr 6: 733–738 46. Mendelson D S, Rose J S, Efremidis S C, Kirschner P A, Cohen B A 1983 Bronchogenic cysts with high CT numbers. Am J Roentgenol 140: 463–465 47. Kawashima A, Fishman E K, Kuhlman J E, Nixon M S 1991 CT of posterior mediastinal masses. RadioGraphics 11: 1045–1067 48. Erasmus J J, McAdams H P, Donnelly L F, Spritzer C E 2000 MR imaging of mediastinal masses. Magn Reson Imaging Clin North Am 8: 59–89 49. Reed J C, Hallet K K, Feigin D S 1978 Neural tumors of the thorax: subject review from the AFIP. Radiology 126: 9–17 50. Kumar A J, Kuhajda F P, Martinez C R, Fishman E K, Jezic D V, Siegelman S S 1983 Computed tomography of extracranial nerve sheath tumors with pathological correlation. J Comput Assist Tomogr 7: 857–865 51. Bhargava R, Parham D M, Lasater O E, Chari R S, Chen G, Fletcher B D 1997 MR imaging differentiation of benign and malignant peripheral nerve sheath tumors: use of the target sign. Pediatr Radiol 27: 124–129 52. Aughenbaugh GL 1984 Thoracic manifestations of neurocutaneous diseases. Radiol Clin North Am 22: 741–756 53. Adam A, Hochholzer L 1981 Ganglioneuroblastoma of the posterior mediastinum: a clinicopathologic review of 80 cases. Cancer 47: 373–381 54. Bar-Ziv J, Nogrady M B 1975 Mediastinal neuroblastoma and ganglioneuroma. The differentiation between primary and secondary involvement on the chest roentgenogram. Am J Roentgenol Radium Ther Nucl Med 125: 380–390 55. Olson J L, Salyer W R 1978 Mediastinal paragangliomas (aortic body tumor): a report of four cases and a review of the literature. Cancer 41: 2405–2412 56. Spizarny D L, Rebner M, Gross B H 1987 CT evaluation of enhancing mediastinal masses. J Comput Assist Tomogr 11: 990–993 57. van Gils A P, Falke T H, van Erkel A R et al 1991 MR imaging and MIBG scintigraphy of pheochromocytomas and extraadrenal functioning paragangliomas. RadioGraphics 11: 37–57 58. Krenning E P, Kwekkeboom D J, Bakker W H et al 1993 Somatostatin receptor scintigraphy with [111In-DTPA-D-Phe1]- and [123I-Tyr3]octreotide: the Rotterdam experience with more than 1000 patients. Eur J Nucl Med 20: 716–731 59. Miles J, Pennybacker J, Sheldon P 1969 Intrathoracic meningocele. Its development and association with neurofibromatosis. J Neurol Neurosurg Psychiatry 32: 99–110 60. Long J A Jr, Doppman J L, Nienhuis A W 1980 Computed tomographic studies of thoracic extramedullary hematopoiesis. J Comput Assist Tomogr 4: 67–70 61. Shaffer K, Rosado-de-Christenson M L, Patz E F Jr, Young S, Farver C F 1994 Thoracic lymphangioma in adults: CT and MR imaging features. Am J Roentgenol 162: 283–289 62. Glazer H S, Wick M R, Anderson D J et al 1992 CT of fatty thoracic masses. Am J Roentgenol 159: 1181–1187 63. Whyte A M, Powell N 1990 Mediastinal lipoblastoma of infancy. Clin Radiol 42: 205–206 64. Gaerte S C, Meyer C A, Winer-Muram H T, Tarver R D, Conces D J Jr 2002 Fat-containing lesions of the chest. RadioGraphics 22: 615–678 65. Kline M E, Patel B U, Agosti S J 1990 Noninfiltrating angiolipoma of the mediastinum. Radiology 175: 737–738 66. Kim K, Koo B C, Davis J T, Franco-Saenz R 1984 Primary myelolipoma of mediastinum. J Comput Tomogr 8: 119–123 67. Jolles H, Henry D A, Roberson J P, Cole T J, Spratt J A 1996 Mediastinitis following median sternotomy: CT findings. Radiology 201: 463–466 68. Bitkover C Y, Cederlund K, Aberg B, Vaage J 1999 Computed tomography of the sternum and mediastinum after median sternotomy. Ann Thorac Surg 68: 858–863 69. Carter A R, Sostman H D, Curtis A M, Swett H A 1983 Thoracic alterations after cardiac surgery. Am J Roentgenol 140: 475–481

• THE MEDIASTINUM, INCLUDING THE PERICARDIUM

70. Goodwin R A, Nickell J A, Des Prez R M 1972 Mediastinal fibrosis complicating healed primary histoplasmosis and tuberculosis. Medicine Balt 51: 227–246 71. Sherrick A D, Brown L R, Harms G F, Myers J L 1994 The radiographic findings of fibrosing mediastinitis. Chest 106: 484–489 72. Rossi S E, McAdams H P, Rosado-de-Christenson M L, Franks T J, Galvin J R 2001 Fibrosing mediastinitis. RadioGraphics 21: 737–757 73. Woodring J H, Loh F K, Kryscio R J 1984 Mediastinal hemorrhage: an evaluation of radiographic manifestations. Radiology 151: 15–21 74. Bejvan S M, Godwin J D 1996 Pneumomediastinum: old signs and new signs. Am J Roentgenol 166: 1041–1048 75. Holt JP 1970 The normal pericardium. Am J Cardiol 26: 455–465 76. Bull R K, Edwards P D, Dixon A K 1998 CT dimensions of the normal pericardium. Br J Radiol 71: 923–925 77. Sechtem U, Tscholakoff D, Higgins C B 1986 MRI of the normal pericardium. Am J Roentgenol 147: 239–244 78. Rienmuller R, Groll R, Lipton M J 2004 CT and MR imaging of pericardial disease. Radiol Clin North Am 42: 587–601, vi 79. Groell R, Schaffler G J, Rienmueller R 1999 Pericardial sinuses and recesses: findings at electrocardiographically triggered electron-beam CT. Radiology 212: 69–73 80. Budoff M J, Lu B, Mao S et al 2000 Evaluation of fluid collection in the pericardial sinuses and recesses: noncontrast-enhanced electron beam tomography. Invest Radiol 35: 359–365 81. Spodick D H 2001 Pericardial disease. In: Braunwald E (ed) Heart disease: a textbook of cardiovascular medicine. WB Saunders, Philadelphia, pp 1823–1876 82. Nasser W K, Helmen C, Tavel M E, Feigenbaum H, Fisch C 1970 Congenital absence of the left pericardium. Clinical, electrocardiographic, radiographic, hemodynamic, and angiographic findings in six cases. Circulation 41: 469–478 83. Van Son J A, Danielson G K, Schaff H V, Mullany C J, Julsrud P R, Breen J F 1993 Congenital partial and complete absence of the pericardium. Mayo Clin Proc 68: 743–747 84. Feigin D S, Fenoglio J J, McAllister H A, Madewell J E 1977 Pericardial cysts. A radiologic-pathologic correlation and review. Radiology 125: 15–20 85. Sechtem U, Tscholakoff D, Higgins C B 1986 MRI of the abnormal pericardium. Am J Roentgenol 147: 245–252 86. Lane E J Jr, Carsky E W 1968 Epicardial fat: Lateral plain film analysis in normals and in pericardial effusion. Radiology 91: 1–5 87. Carsky E W, Mauceri R A, Azimi F 1980 The epicardial fat pad sign: analysis of frontal and lateral chest radiographs in patients with pericardial effusion. Radiology 137: 303–308 88. Yousem D, Traill T T, Wheeler P S, Fishman E K 1987 Illustrative cases in pericardial effusion misdetection: correlation of echocardiography and CT. Cardiovasc Intervent Radiol 10: 162–167 89. Tomoda H, Hoshiai M, Furuya H et al 1980 Evaluation of pericardial effusion with computed tomography. Am Heart J 99: 701–706 90. Cameron J, Oesterle S N, Baldwin J C, Hancock E W 1987 The etiologic spectrum of constrictive pericarditis. Am Heart J 113: 354–360 91. Ling L H, Oh J K, Schaff H V et al 1999 Constrictive pericarditis in the modern era: evolving clinical spectrum and impact on outcome after pericardiectomy. Circulation 100: 1380–1386 92. Masui T, Finck S, Higgins C B 1992 Constrictive pericarditis and restrictive cardiomyopathy: evaluation with MR imaging. Radiology 182: 369–373 93. Grebenc M L, Rosado de Christenson M L, Burke A P, Green C E, Galvin J R 2000 Primary cardiac and pericardial neoplasms: Radiologic– pathologic correlation. RadioGraphics 20: 1073–1103 94. Abraham K P, Reddy V, Gattuso P 1990 Neoplasms metastatic to the heart: review of 3314 consecutive autopsies. Am J Cardiovasc Pathol 3: 195–198 95. Mousseaux E, Meunier P, Azancott S, Dubayle P, Gaux J C 1998 Cardiac metastatic melanoma investigated by magnetic resonance imaging. Magn Reson Imaging 16: 91–95 96. Fujita N, Caputo G R, Higgins C B 1994 Diagnosis and characterization of intracardiac masses by magnetic resonance imaging. Am J Card Imaging 8: 69–80

267

268

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

SUGGESTIONS FOR FURTHER READING Choe Y H, Im J G, Park J H, Kim C W 1987 The anatomy of the pericardial space: a study in cadavers and patients. Am J Roentgenol 149: 693–697 Schoepf U J (ed) 2004 CT of the heart: Principles and applications. Humana Press, New Jersey Sharma A, Fidias P, Hayman L A, Loomis S L, Taber K H, Aquino S L 2004 Patterns of lymphadenopathy in thoracic malignancies. RadioGraphics 24: 419–434

Spindola-Franco H, Fish B G 1985 Radiology of the heart. Springer Verlag, New York Strollo D C, Rosado de Christenson M L, Jett J R 1997 Primary mediastinal tumors. Part 1: tumors of the anterior mediastinum. Chest 112: 511–522 Strollo D C, Rosado-de-Christenson M L, Jett J R 1997 Primary mediastinal tumors: part II. Tumors of the middle and posterior mediastinum. Chest 112: 1344–1357

CHAPTER

Pulmonary Infection in Adults

15

Philip C. Goodman

Specific pneumonias • Lobar pneumonias • Bronchopneumonia • Anaerobic pneumonias • Atypical pneumonia • Pulmonary tuberculosis • Nontuberculous mycobacterial disease • Fungal infections • Protozoal and metazoal diseases Pulmonary complications of HIV infection and AIDS Infections • Malignancies

The variety of infectious agents that produce pneumonia in humans is vast, encompassing bacteria, viruses, fungi, protozoa and parasites.The chest radiograph is invariably abnormal with pneumonia and when on occasion it is normal, CT of the chest may reveal underlying lung involvement. From a diagnostic standpoint radiographic findings should be combined with clinical and laboratory information to suggest the cause of infection. Differentiation of aetiologies based solely on the radiograph is not reliable, yet the pattern of abnormalities should be very useful in formulating a differential diagnosis of the nature of disease1.

SPECIFIC PNEUMONIAS Specific pneumonias are caused by organisms drawn from virtually all forms of microbial life. In the following sections the type of infections are classified in a variety of ways: their predominant appearance, by the situations in which they are acquired, and taxonomically. Some organisms cross over pattern and situational boundaries and will only be addressed briefly when that occurs. First an explanation of some terms is presented. Lobar pneumonia is usually unifocal, develops in the distal airspaces adjacent to the visceral pleura, and then spreads via collateral air drift routes to produce uniform homogeneous opacification of partial or complete segments of lung and occasionally an entire lobe. As the airways are not primarily involved and remain patent, there is little to no volume loss and air bronchograms are common. On the other hand, bronchopneumonia, frequently caused by aspiration of secretions from a colonized trachea, is usually multifocal and centred in distal airways. The process is initially heterogeneous and distributed along the course of the airways. Thus, radiologically a bronchopneumonia is characterized by large heterogeneous, scattered opacities which only later, with worsening of disease, become more homogeneous. An air bronchogram is usually absent.

The term atypical pneumonia was initially applied to the clinical and radiographic appearance of lung infection not behaving or looking like that caused by Streptococcus pneumoniae. Radiographically focal or diffuse small heterogeneous opacities are seen uniformly distributed in the involved lung. Frequently these opacities are described as reticular or reticulonodular. Community-acquired pneumonias are caused by a variety of typical and atypical organisms with a myriad of radiographic abnormalities. However, many community-acquired pneumonias are still commonly caused by Strep. pneumoniae and are lobar in appearance. Nosocomial pneumonias occur during hospitalization. They frequently appear as bronchopneumonia caused by Staphylococcus aureus and Gram-negative organisms. They may look like atypical infections, similar to those caused by Mycoplasma, viruses and Chlamydia. In practice, however, remember that radiological patterns can be indeterminate and that one infectious agent can produce many patterns2.

LOBAR PNEUMONIAS Streptococcus pneumoniae Strep. pneumoniae pneumonia (pneumococcal pneumonia) is the most common community-acquired bacterial pneumonia

270

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

in adults. Predisposing factors include chronic illness, alcoholism, sickle-cell disease and splenectomy. Chest radiographs generally reveal a peripheral, homogeneous opacity with or without air bronchograms (Fig. 15.1).Though commonly basal and solitary, this pneumonia can occur in any lobe and may be multifocal. Round-shaped pneumonias may simulate lung masses (Fig. 15.2). Lobar volume usually remains unchanged but rarely it increases. Cavitation is very unlikely. Parapneumonic effusion is fairly common; empyema is less frequent. Atypical patterns of heterogeneous opacification have been described but are unusual. Radiographic resolution is fairly rapid with some improvement commonly seen within 1 week and total resolution within 2–6 weeks (Fig. 15.3). Failure to improve may be due to drug resistance, incorrect or inadequate antimicrobial treatment, an obstructing lesion, development

Figure 15.2 Pneumococcal pneumonia—simulates mass. A 56 year old man with fever and productive cough. (A) PA and (B) lateral chest radiographs demonstrate a discrete fairly well marginated opacity in the right lower lobe. The abnormality resolved following appropriate antibiotic therapy and Gram stains of sputum demonstrated Strep. pneumoniae. Figure 15.1 Pneumococcal pneumonia. A 38 year old man with Strep. pneumoniae pneumonia. A close-up of the PA chest radiograph demonstrates a homogeneous lingular opacity with central air bronchograms.

Figure 15.3 Pneumococcal pneumonia—resolution time. A 48 year old man with productive cough and fever. (A) AP chest radiograph demonstrates peripheral homogeneous poorly marginated opacity. (B) AP image demonstrates significant improvement 5 d after institution of therapy.

CHAPTER 15

of lung abscess or empyema, or noninfectious cause of the opacity, e.g. broncho-alveolar carcinoma3.

Klebsiella Klebsiella pneumoniae pneumonia may arise in the community or in hospital. The chest radiograph frequently demonstrates homogeneous opacity similar to Strep. pneumoniae or may reveal scattered focal heterogeneous opacities, a bronchopneumonia, as produced by other Gram-negative bacteria. Early series reported rapid cavitation of lobar consolidation often accompanied by bulging fissures signifying a very exudative response.

Legionella Legionella pneumophila causes a pneumonia (Legionnaires’ disease) which when severe has a 10–30% mortality rate. Legionnaires’ disease may be acquired in a community, nosocomial, or epidemic fashion associated with contaminated water sources. Predisposing factors include post-transplantation, chronic obstructive pulmonary disease (COPD) and heart failure. The typical radiographic patterns include solitary or multifocal, lobar pneumonia-like, homogeneous opacities, simulating Strep. pneumoniae infection4, but with a tendency to round and mass-like appearance (Fig. 15.4). Rapid progression is common, with confluence and/or spread of the initial frequently unilateral consolidation to other lobes. Cavitation has been described in immunocompromised and post renaltransplant patients. An effusion occurs in 10–35% of cases. Resolution may be quick when treated appropriately but in some instances may take weeks.

Actinomycosis Actinomycosis is caused by Actinomyces israeli, an anaerobic, Gram-positive bacterium.The organisms reside as commensals in the mouth and oropharynx and cause infection when access occurs to devitalized or previously infected tissues, particularly

•

PULMONARY INFECTION IN ADULTS

in the cervicofacial region and abdomen. A chronic inflammatory reaction spreads across fascial planes and causes abscesses and fistulas containing tiny sulphur granules. In fewer than one-quarter of cases the lungs are involved from aspiration or spread from abdominal or cervicofacial foci. The chest radiograph usually reveals homogeneous opacification as a lobar type pneumonia or mass5. Cavitation is common and the appearance mimics bronchogenic carcinoma. Focal fibrosis and contraction may be severe.Widespread small nodules have been reported. Pleural effusion, pleural thickening or empyema formation, and extension of disease to the contiguous soft tissue or bones (as a periostitis) set this pneumonia apart from the usual bacterial types. (Tuberculosis and nocardiosis may have a similar appearance.) At CT, scattered peripheral areas of homogeneous consolidation with central low attenuation and adjacent pleural thickening are suggestive of actinomycosis6.

Nocardiosis Nocardia asteroides is an aerobic, Gram-positive, weakly acid-fast bacillus, related to, and sometimes confused with, A. israeli and the Mycobacteria. Infections by these organisms occur worldwide, but most reported cases are from North America. The majority of affected patients are immunocompromised, including patients with acquired immune deficiency syndrome (AIDS). Nocardiosis usually begins with a focus of pulmonary infection and may disseminate to other organs, notably the brain. The chest radiographic findings vary7. Pulmonary consolidation, either unifocal or multifocal, is the usual feature and cavitation is frequent. However, many chest radiographs reveal single or multiple pulmonary nodules resembling primary lung cancer or metastatic disease (Fig. 15.5). Cavitation is common as is pleural effusion. Lymphadenopathy or chest wall involvement may be evident, especially on CT 8.

Moraxella (Branhamella) catarrhalis Moraxella catarrhalis is a Gram-negative coccus that commonly causes otitis media and sinusitis in children, and acute tracheobronchitis or a relatively mild pneumonia in older patients with COPD. Many patients have additional serious underlying disease, including bronchial carcinoma, or are taking steroids. Presentation is most common during the winter and early spring, and the most frequent radiological pattern is lobar or segmental homogeneous opacification9. Bibasal and heterogeneous or mixed heterogeneous/homogeneous opacities are also reported. Small effusions occur in one-third of patients.

Chlamydial pneumonia

Figure 15.4 Legionnaires’ disease. A 35 year old man with high fevers. The AP radiograph demonstrates homogeneous opacities in the right upper lobe. The medial one resembles a mass. Subsequent images demonstrated bilateral spread of disease determined to be caused by Legionella pneumophila.

Chlamydiae are bacteria implicated in community-acquired pneumonias. C. psittaci and C. pneumoniae (strain TWAR) cause infections in adults, and C. trachomatis largely, but not entirely, in neonates. C. psittaci causes psittacosis (ornithosis), which is usually seen following direct bird contact but occasionally after outdoor activity, e.g. mowing lawns. The clinical manifestations may be indistinguishable from an acute bacterial pneumonia. Chest radiography reveals small to large homogeneous opacities and/ or perihilar or basal reticular opacities10. Hilar nodes are occasionally enlarged and small effusions are sometimes present.

271

272

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

acquired by inhalation from farm livestock or their products, and occasionally from domestic animals. Radiographic abnormalities usually consist of unilateral, but occasionally bilateral, segmental, or lobar opacities14. In 20% of patients reticular opacities are observed15. Pleural effusion occurs but is not common. Radiographic resolution occurs within 6 weeks of starting treatment. Other rickettsial infections such as Rocky Mountain spotted fever are usually tick-borne and occasionally demonstrate diffuse heterogeneous or homogeneous opacities on chest radiographs, perhaps representing vasculitis or cardiogenic pulmonary oedema16.

Francisella tularensis Francisella tularensis is a Gram-negative coccobacillus that causes tularaemia. It is endemic in parts of Europe, Asia and North America. The disease is acquired from handling infected animals, bites of insect vectors and inhalation. In patients with pleuropulmonary tularaemia the most common radiographic findings in the chest are scattered airspace consolidation, hilar adenopathy and pleural effusion17.Various other, less common, changes are seen, including cavitation.

Yersinia (Pasturella) pestis

Figure 15.5 Nocardia infection. A young immune suppressed patient presented with fever and cough. (A) PA chest radiograph demonstrates large mass with possible cavitation in the right upper lobe. (B) Single CT section demonstrates well defined large mass with central necrosis.

Yersinia pestis is a Gram-negative coccobacillus that causes plague, a disease that is endemic in parts of South America Africa and Asia. It is also found in the southwestern USA, about 20 cases being reported annually, of which 10% have respiratory involvement18. Plague exists in three forms—primary septicaemic, bubonic and pneumonic. The latter two forms are the most common and both are associated with chest radiographic changes. The principal radiographic findings are basally predominant, bilateral, multifocal areas of consolidation that rapidly progress to become confluent. Pleural effusion is common and there may be mediastinal adenopathy, the latter sometimes being the sole manifestation in the chest of the bubonic form of the disease19.

Leptospirosis The radiographic opacity characteristically clears slowly and persistent changes at 3 months are not uncommon11. C. pneumoniae (strain TWAR) causes respiratory infections in adults that are often asymptomatic or mild. It is one of the most common causes of community-acquired pneumonia. The illness may be biphasic, with initial sore throat and hoarseness followed a few weeks later by lower respiratory tract symptoms. The most common radiographic pattern in primary infection is usually unifocal and occasionally multifocal homogeneous opacity. With recurrent infection, changes are more commonly bilateral and equally homogeneous and heterogeneous. Up to one-half of patients have pleural fluid, either small or moderate in size. In one series it was not possible reliably to differentiate between pneumonia caused by C. pneumoniae and S. pneumoniae12,13.

The Leptospira interrogans complex is a group of spirochaetes pathogenic for almost all mammals, including man.The disease in humans is usually acquired from contact with contaminated water (sewage workers, water-sport participants) or directly from infected animals. Leptospirosis is a biphasic illness characterized initially by fever, headache and myalgia, followed by skin rash and renal, neurological and hepatic disorder. Pulmonary involvement, reported in 11–67% of cases, is characterized by a haemorrhagic pneumonitis. The initial radiographic pattern is often nodular (1–7-mm diameter) evolving into confluent airspace or ground-glass opacities. Changes are bilateral, commonly peripherally predominant, and resolve in about 2 weeks20. Small pleural effusions and discoid atelectasis are also described.

Rickettsial pneumonia

Other pneumonias that need to be included in a differential list of lobar type infections include primary tuberculosis (see section on mycobacterial disease), coccidioidomycosis and blastomycosis (see section on fungal disease).

The most common rickettsial lung infection is sporadic or epidemic Q-fever pneumonia caused by Coxiella burnetii, an intracellular, Gram-negative bacterium. Infection is mainly

Miscellaneous infections

CHAPTER 15

•

PULMONARY INFECTION IN ADULTS

BRONCHOPNEUMONIA Staphylococcus aureus Staph. aureus pneumonia occurs in debilitated hospitalized or institutionalized patients, and less frequently as a community-acquired infection. It is usually acquired by aspiration from the upper respiratory tract. Chest radiographs typically demonstrate heterogeneous opacities in a scattered multifocal and bilateral distribution21 (Fig. 15.6). Pleural effusion or empyema, and cavitation are common; air bronchograms are unusual. Pneumatoceles may form, particularly in children. Septicaemic infections, as opposed to those acquired by aspiration, cause disseminated, poorly marginated, peripheral, multifocal, nodules which can cavitate.These septic emboli are seen in drug addicts, immunocompromised patients and patients with infective endocarditis or indwelling catheters.

Gram-negative pneumonias Gram-negative pneumonias are chiefly caused by enterobacteria (Enterobacter sp., Serratia marcescens, Proteus sp, Escherichia coli, Pseudomonas aeruginosa and Haemophilus influenzae) in a hospital setting. Patients affected are invariably debilitated by a chronic medical or pulmonary disease. These bacteria are generally aspirated from a colonized upper respiratory tract or may be inhaled or spread hematogenously. The lower lobes predominantly tend to be affected and the radiographic pattern is similar to that seen with Staph. aureus infections in adults (Fig. 15.7). A CT study of Ps. aeruginosa pneumonias revealed multifocal, predominantly upper lobe, airspace consolidation, random large nodules, tree-in-bud opacities, ground-glass opacity, necrosis and pleural effusion22.

Figure 15.6 Bronchopneumonia—staphylococcus. A 24 year old man with fever and cough developed within the hospital. The AP chest radiograph demonstrates bilateral scattered homogeneous and heterogeneous opacities, some almost nodular.

Figure 15.7 Gram-negative pneumonia—septic emboli. A 52 year old man with high fevers that developed during hospitalization. AP chest radiograph demonstrates scattered peripheral discrete and rounded opacities. Right upper lobe nodule is cavitated with thin walls. Left lower lobe nodule has thick-wall cavitation. Escherichia coli was cultured from blood.

ANAEROBIC PNEUMONIAS Most anaerobic pneumonias result from aspiration of bacteria, including Bacteroides, Peptostreptococcus, microaerophilic streptococcus, and Fusobacterium following a bout of altered consciousness, or mechanical ventilation23. Immediately following aspiration the chest radiograph is frequently normal although scattered opacities representing partial atelectasis caused by aspirated food may be seen. The appearance of pneumonia is usually delayed 24–72 h, at which time heterogeneous opacities are seen in dependent lung segments (posterior upper lobe, superior, or posterobasal lower lobe.) Involvement may be uni- or bi-lateral. Empyema is a common complication, may be large, and may occur with or without radiographic evidence of pneumonia. Multiple cavities reflecting severe lung necrosis may be seen 1–3 weeks following aspiration (Fig. 15.8). Patients who delay seeking medical help for 3–4 weeks may present with a discrete lung abscess24. Nearly two-thirds of these lesions occur in the apicoposterior segments of upper lobes and the superior segments of lower lobes. These may be large, ranging in size from 2 to 12 cm. Characteristically the wall is thick and irregular. Mediastinal nodal enlargement, though not common, may be observed. The above radiological features, coupled with an indolent course and systemic symptoms, closely resemble the findings of post primary tuberculosis or bronchogenic carcinoma.

273

274

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 15.8 Anaerobic pneumonia. A 51 year old man with chronic fevers and productive cough. (A) PA and (B) lateral chest radiographs demonstrate left lower lobe opacity with multiple air–fluid levels. The patient was treated for an anaerobic pneumonia and improved rapidly. Incidentally noted are right paratracheal calcified lymph nodes from prior tuberculosis or histoplasmosis.

ATYPICAL PNEUMONIA Mycoplasma pneumonia Mycoplasma pneumoniae is a major nonbacterial cause of community-acquired pneumonia in patients between the ages of 20 and 40 years. Spread is from person to person by droplets during close contact, particularly in some communities (e.g. military barracks). Symptoms resemble a viral infection with progression from the upper to the lower respiratory tract. Pneumonia occurs in less than one-tenth of those infected, is accompanied by a minimally productive cough, and usually runs a self-limiting course. Extrapulmonary complications are well described. The radiological findings are variable and may be striking in the face of minor clinical signs. The most common pattern is unilateral lower lobe involvement beginning as heterogeneous, reticular, segmental, peribronchial opacifications that may become lobar and homogeneous. Bilateral or multilobar involvement is a frequently observed variation (Fig. 15.9). Pleural effusions are uncommon. Nodal enlargement is an unusual finding in adults, occurring in about one-fifth of patients in some series11. Radiological clearing is variable and may occasionally take as long as 6 weeks. CT demonstrates ground-glass and homogeneous opacities, a bronchiolitis with centrilobular nodules, and bronchovascular thickening in approximately 80% of patients25.

Influenza A and B Influenza A and B are common causes of pneumonia in adults, particularly the elderly. Radiographic findings within a few days after symptoms begin include scattered homogeneous opacities that rapidly become bilateral, extensive and confluent. Pleural effusion is uncommon or rare. Clinical

Viral pneumonias Viral pneumonia is common in infants and children but unusual in adults. Studies of community-acquired pneumonia suggest that about 8% are viral. The pneumonia may be solely a manifestation of respiratory tract involvement (e.g. influenza) or part of a more generalized viral illness (e.g. varicella). Viral infections predispose to secondary bacterial pneumonia.

Figure 15.9 Mycoplasma pneumonia. A 35 year old man presents with nonproductive cough and fever. The PA chest radiograph demonstrates bilateral perihilar and lower lobe heterogeneous reticular opacities as well as a more focal left upper lobe homogeneous opacity. Findings are characteristic of atypical community-acquired infections.

CHAPTER 15

•

PULMONARY INFECTION IN ADULTS

relapse about 2 weeks after the onset of influenza may be due to secondary bacterial pneumonia (often Strep. pneumoniae or Staph. aureus). In patients with underlying haematological malignancies, chest radiographs demonstrate scattered bilateral consolidation and ill-defined nodules, and CT reveals groundglass opacities nodules and a tree-in-bud appearance26.

Herpes simplex virus (HSV)

Adenovirus

Hantavirus

Adenovirus, a common upper respiratory tract pathogen, rarely results in respiratory failure and a radiographic appearance of acute respiratory distress syndrome (ARDS)27,28. Initial radiographs show homogeneous lobar pneumonia that progresses to diffuse bilateral heterogeneous and homogeneous opacities. A similar appearance has been described in paediatric patients29.

Hantavirus is transmitted by human contact with infected deer mice and produces a pulmonary syndrome most frequently reported in the southwestern USA. Bilateral homogeneous opacities resembling ARDS are seen33 and nearly all patients have pleural fluid. The mortality rate of treated patients can approach 35%.

Infectious mononucleosis

Cytomegalovirus (CMV) pneumonia has been seen in increasing numbers with the proliferation of bone marrow and solid organ transplantation. Chest radiographs demonstrate focal and diffuse hazy opacification and multiple small (less than 5 mm) nodules, and less commonly focal consolidation34. CT reveals small centrilobar nodules and ground-glass or homogeneous consolidation generally in a symmetric and bilateral distribution35. Solitary or multiple large nodules are frequently identified at CT.

This is caused by the Ebstein–Barr virus. Radiological manifestations in the chest are uncommon and include hilar lymphadenopathy, heterogeneous opacities, small pleural effusions and lobar homogeneous consolidation30.

Varicella Varicella is unlike other viruses in that it causes pneumonia more frequently in adults than in children. Typically, young adults are affected, some predisposed to the infection by lymphoma, pregnancy, or steroid therapy. Pulmonary involvement follows the skin rash by 1–6 d. The radiological findings are characteristic, consisting of widespread 5–10 mm in diameter poorly marginated nodules or acinar opacities, which may become confluent (Fig. 15.10). The nodules usually resolve in a week or two but can persist for months (simulating metastases). CT findings are similar, including nodules and coalescing opacities; ground-glass opacities are also observed31. In resolution, numerous small irregular calcified nodules can develop; these can be evident on plain radiographs.

Figure 15.10 Varicella pneumonia. A 30 year old man with lymphoma and new development of fever and skin rash. The PA chest radiograph demonstrates bilateral poorly marginated 5–10 mm in diameter nodular opacities.

Herpes simplex virus (HSV) may also cause pneumonia characterized by hazy and homogeneous opacities seen in a segmental or subsegmental distribution, and pleural effusion in approximately 50% of patients. At CT similar findings are noted32.

Cytomegalovirus (CMV)

PULMONARY TUBERCULOSIS Mycobacterium tuberculosis accounts for more than 95% of pulmonary mycobacterial infections. Other mycobacterial species, mainly M. kansasii and the M. avium–intracellulare complex (MAC) account for the remainder. Infections are acquired via droplet inhalation from other infected individuals. In people who are previously unexposed, hypersensitivity to tuberculoprotein is absent and primary tuberculosis develops. This form is commonly seen in infants and children. With improved control of tuberculosis in western societies, however, more people reach adulthood without exposure, and primary patterns of disease are being seen with increasing frequency in adulthood. If a patient already possesses hypersensitivity to tuberculoprotein by virtue of a previous infection or BCG vaccination, then post-primary tuberculosis is seen. Should the primary disease pass into the post-primary form without a break, the term progressive primary tuberculosis is used. Factors that contribute to the large number of cases seen worldwide are human immunodeficiency virus (HIV) infection, inner city poverty, homelessness and immigration from areas with high rates of infection. Other predisposing conditions are diabetes mellitus, alcoholism, silicosis, malignancy, immune compromise from a variety of causes and living in closed institutions. Symptoms include loss of weight and appetite, malaise, fever, night sweats and cough, which may or may not be productive and accompanied by haemoptysis. Treatment is by chemotherapy and, given cooperation by the patient and a sensitive organism, it is very successful. Recourse to surgery is rare.The radiographs of patients with tuberculosis take many forms36,37, and are best discussed as primary and post-primary disease.

275

276

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Primary tuberculosis Pathologically, primary tuberculosis is characterized by macrophages, other monocytes and inflammatory fluid in an area of peripheral pneumonia. Spread of the bacilli to regional nodes and throughout the body follows within 2–6 weeks. Immunological changes in the host at this stage lead to the ability to kill the organism and healing by fibrosis, with or without subsequent calcification. Repeated episodes of arrest and progression may lead to a growing nodule, a tuberculoma. Radiographically, primary tuberculosis causes a pneumonia that is homogeneous and mimics community-acquired pneumonias, such as Strep. pneumoniae (Fig. 15.11). Any lobe may be involved; size varies from subsegmental to an entire lobe. Multifocal involvement is unusual and cavitation rare, the occurrence of the latter suggesting progressive primary disease. Compared with community-acquired pneumonia, primary tuberculosis may exhibit nodal enlargement, usually ipsilateral, hilar and/or mediastinal. In fact, lymphadenopathy is the most common manifestation of primary tuberculosis in children and occurs with or without pneumonia. In the former case, the pneumonia may sometimes obscure hilar enlargement. In adults hilar or mediastinal lymphadenopathy is less common declining to about 50% of cases in the older population. Asymmetrical bilateral hilar involvement is less commonly described38. Nodal pressure and bronchial erosion may cause segmental or lobar collapse; commonly in the anterior segment of the right upper lobe and the middle lobe. Bronchial perforation may permit endobronchial spread of disease giving rise to scattered, heterogeneous opacities mimicking a bronchopneumonia. Pleural effusion as a manifestation of primary tuberculosis occurs in children, who usually have parenchymal or nodal disease, or in teenagers and young adults, when it is frequently isolated. The effusions are often large and unilateral. Residual pleural change is unusual, pleural thickening and calcification

Figure 15.11 Primary tuberculosis. A 39 year old man with cough and fever. Homogeneous lingular opacity is noted. No definite lymphadenopathy is identified but this would not necessarily be expected in an adult.

being much more commonly due to a tuberculous empyema seen with post-primary disease. Although classically a manifestation of primary disease, miliary tuberculosis is now more commonly seen as a post-primary process in older patients. Multiple small (1–2 mm) discrete nodules are scattered evenly throughout both lungs (Fig. 15.12). Other features of primary tuberculosis may or may not be present. With therapy, the nodules clear, often rather slowly over months, leaving no residual changes. Calcification within miliary nodules is rare or nonexistent. Usually the primary pneumonia resolves completely. In one-third of patients a residual well defined rounded or irregular (linear) opacity, with or without calcification remains. This is a Ghon lesion or focus. Nodal calcification may occur in the ipsilateral hilum or mediastinum and is heterogeneous and irregular. When a Ghon lesion or focus and ipsilateral lymph node calcification are seen together the combination is termed a Ranke complex. This appearance reflects prior primary tuberculosis (Fig. 15.13). (Remember that this picture is evidence of an old infection but does not imply activity of disease. For that the clinical findings must be incorporated.)

Post-primary tuberculosis This term is used to describe tuberculosis in patients who by reason of previous infection or BCG vaccination have acquired tuberculoprotein hypersensitivity. Most cases are due to reactivation of quiescent lesions, but occasionally a new infection from an exogenous source occurs. Pathologically, the ability of the host to respond immunologically results in a greater inflammatory reaction and caseous necrosis. Radiographically, in 95% of patients the initial lesions are poorly marginated, nodular and linear opacities approximately 5–10 mm in diameter, which arise in the apicoposterior segments of an upper lobe and/or the superior segment of a lower lobe (Fig. 15.14). Isolated involvement of the anterior segment of an upper lobe with few exceptions virtually excludes the diagnosis of tuberculosis, although the anterior segment may become involved from contiguous segmental disease. Changes may be unilateral or bilateral. With progression the opacities clump together and coalesce.

Figure 15.12 Miliary tuberculosis. A 26 year old man with fevers, shortness of breath. The close-up of a PA chest radiograph demonstrates multiple, discrete 1–2 mm in diameter nodular opacities.

CHAPTER 15

Figure 15.13 Primary tuberculosis—Ghon focus, Ranke complex. A 70 year old man with a prior history of tuberculosis. The close-up of the PA chest radiograph demonstrates a right mid lung calcified nodule with ipsilateral right hilar lymph node calcification. The solitary calcified nodule is termed a Ghon focus. The combination of this with ipsilateral calcified lymph nodes is termed a Ranke complex.

Cavitation is seen in the region of abnormality in 40–80% of cases. Cavities may be single or multiple, large or small. Wall thickness varies from thin to thick. Air-fluid levels are unusual but have been recorded in up to 20% of cases (Fig. 15.15). A Rasmussen aneurysm is a rare life-threatening complication of cavitary tuberculosis caused by granulomatous weakening of a pulmonary arterial wall.

•

PULMONARY INFECTION IN ADULTS

Figure 15.15 Post-primary tuberculosis. A 53 year old man presents with night sweats and fever. A close-up of a PA chest radiograph demonstrates coalescence of poorly marginated reticular and nodular opacities in the right upper lobe with an irregularly margined moderately thick-walled cavity and small air–fluid level.

Some of the opacities calcify, though this occurs less commonly than with primary tuberculosis (Fig. 15.16). Bronchiectasis and the formation of cysts and bullae may be created by the lung distortion coupled with secondary bacterial infections. Endobronchial spread can occur with or without cavitary disease and is similar to that seen with primary tuberculosis.

Healing results in scar formation. Cavities are usually obliterated but rarely a sterile cavity remains. The fibrosis produces well defined, upper lobe nodular and linear opacities, often with evidence of severe volume loss and pleural thickening.

Pleural effusion accompanying post-primary tuberculosis is more likely an empyema which may lead to pleural thickening or calcification.

Figure 15.14 Post-primary tuberculosis. A 40 year old man presents with night sweats, nonproductive cough and fever. The chest radiograph demonstrates clumped nodular opacities, particularly in the superior segment of left lower lobe, but also in the peripheral right lung, probably in the posterior segment of the right upper lobe.

Figure 15.16 Post-primary tuberculosis. A 62 year old man with a history of tuberculosis. The chest radiograph demonstrates right upper lobe well marginated and perhaps calcified nodular opacities, as well as significant volume loss in the left upper lobe with upward retraction of the left hilum and left apical pleural thickening.

277

278

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Miliary tuberculosis now occurs more commonly as a manifestation of post-primary than of primary disease and has the same radiographic appearance as described above. If the patient goes untreated the miliary nodules get bigger but rarely more than 5 mm in diameter before death occurs. A tuberculoma may occur in the setting of primary or postprimary tuberculosis and probably represents localized parenchymal disease that alternately activates and heals.These 10–15 mm in diameter nodules are commonly single but may be multiple. The margins of the nodule are usually well defined and there may be satellite lesions nearby, though this is not a specific finding. Calcification is common. Tuberculomas frequently remain stable for years, but always carry the potential risk of activation and dissemination. Chest wall involvement may be due to haematogenous seeding or direct spread from the lung and may affect soft tissue, rib, or costal cartilage. Surgery for tuberculosis is rarely performed now but patients may still be seen and exhibit the results of phrenic nerve ablation, plombage (lucite balls or oil, inserted extrapleurally), and thoracoplasty. Several studies have looked at the value of CT in diagnosing both primary and post-primary tuberculosis39. CT can identify lymphadenopathy and parenchymal lesions not appreciated on plain radiography. Over 80% of patients have lymph nodes with low attenuation centres when contrast medium is administered. Complications or spread of tuberculosis may be better revealed by CT. However, routine use of CT would probably not be cost-effective; it is recommended when normal or equivocal radiographs are seen in association with the clinical suspicion of tuberculosis, or when complications are suspected. There is no simple answer to the frequently posed question as to whether a radiographically detected tuberculous lesion is active or not. Ill-defined coalesced nodules, poorly marginated linear opacities, and especially cavitary disease in the appropriate segments are suspicious for active disease, whereas well defined opacities are not. Exceptions occur, however, and clinical findings are necessary for the diagnosis. Stability of the radiographic appearance is comforting but it should be remembered that all tuberculous lesions, even those that strongly suggest a healed lesion, even calcified, are capable of reactivation. Resolution of abnormal opacities, decrease in cavity size, and volume loss from fibrosis all suggest a satisfactory response to treatment, with a time course in the order of weeks and months rather than days. Occasionally, during the first month of treatment, opacities will extend despite appropriate therapy. Extension of opacities or incomplete resolution on treatment should also raise the question of drug-resistant tuberculosis, failure to take the medication, or an associated bronchial neoplasm. Chest radiography may be used to screen populations for pulmonary tuberculosis and to aid in the management of known cases. Nonselective radiographic screening is still employed in some countries but has been abandoned in others

because the falling prevalence of tuberculosis has resulted in low detection rates, particularly of smear-positive cases, which, from an epidemiological point of view, are the important ones to detect. Screening is still used in some selected groups such as the military, prisons and in socio-economically disadvantaged communities. Follow-up chest radiography for known cases of tuberculosis is usually sufficient after 1, 6 and 9 months, or at the end of therapy. Long-term surveillance is no longer considered necessary in straightforward cases. This regimen may be modified if complications arise, and follow-up may be extended if there is doubt about patient compliance, if risk factors are present, or if there is severe lung damage.

NONTUBERCULOUS MYCOBACTERIAL DISEASE As mentioned above, 1−3% of pulmonary mycobacterial infections are caused by agents other than M. tuberculosis: usually M. avium–intracellulare complex (MAC ) and less commonly M. kansasii. These are free-living saprophytes, and infections are not acquired from human contacts but from the environment by inhalation or ingestion. Patients are often predisposed by reason of underlying debilitating disease, immune compromise, chronic airflow obstruction, previous pulmonary tuberculosis, or silicosis and following lung transplantation. MAC, in particular, is also seen in otherwise healthy, older women. Clinically, MAC may be an indolent process with symptoms of cough, with or without sputum production. The radiological pattern of M. kansasii is generally indistinguishable from post-primary tuberculosis40,41 and changes equivalent to primary tuberculosis are rarely described. More commonly, MAC presents with a radiological pattern that does not resemble that of post-primary tuberculosis. It consists of multiple nodules, with or without small ring opacities, showing no specific lobar predilection and bronchiectasis particularly in the lingula and right middle lobe. CT has identified a similar pattern with easier detection of bronchiectasis42,43. High-resolution CT (HRCT) demonstrates small centrilobular nodules, small airway or bronchiolar ectasia and tree-in-bud opacities (Fig. 15.17). Pleural effusion is uncommon and nodal enlargement and haematogenous spread are rare except in patients with AIDS.

FUNGAL INFECTIONS Cryptococcosis (torulosis) Cryptococcosis, also known as torulosis and European blastomycosis, is caused by inhaling spores of Cryptococcus neoformans, a fungus of worldwide distribution which is found in soil and in bird droppings. Most reports of the disease are from North America. Many patients have no symptoms and the pulmonary lesions heal spontaneously but in some the disease may spread to

CHAPTER 15

•

PULMONARY INFECTION IN ADULTS

Figure 15.17 Mycobacterium avium complex (MAC). An elderly woman with a long history of emphysema and chronic cough. (A) PA chest radiograph demonstrates several poorly marginated nodular opacities in the right perihilar region as well as overlying the right costophrenic angle. More coarse linear opacities emanating from the right hilum suggest the possibility of scarring or bronchiectasis. (B) Lateral radiograph demonstrates bronchiectasis in the lateral segment of the right middle lobe. (C) CT section demonstrates right middle lobe tree-in-bud opacities as well as right lower lobe mucus-filled dilated bronchi. On the lateral view a pectus excavatum is also demonstrated on lateral projection. This finding has been associated with MAC.

many organs, meningoencephalitis being the most serious consequence. Approximately half the cases of symptomatic infection are associated with immunodeficiency. The chest radiograph44,45 shows three patterns: pulmonary masses, which are usually single, but may be multiple; homogeneous segmental or lobar opacifications with or without air bronchograms, cavitation and lymphadenopathy; and diffuse nodular, occasionally military, or reticulonodular opacities. The masses usually have an ill-defined edge and may cavitate and range from approximately 5 mm to very large (Fig. 15.18). CT reveals similar findings: some ‘acinar’ nodules but no tree-in-bud abnormalities46.

Histoplasmosis Histoplasma capsulatum is a fungus found in moist soil and in bird or bat excreta in many parts of the world, but human infection is only seen with any frequency in North America, particularly in the major river valley regions of the mid and eastern USA and Canada47. Occasional epidemics of symptomatic infection (similar to the flu) are reported in areas where construction is occurring or following exposure from entering bat caves or cleaning out chicken pens; however, most cases are asymptomatic. If detected while symptomatic the chest radiograph may reveal multiple poorly-defined nodules approximately 5–10 mm in diameter (Fig. 15.19).

Figure 15.18 Cryptococcus—asymptomatic. This young man presented for routine physical examination. (A) PA chest radiograph demonstrates poorly marginated left mid lung nodular opacity. (B) Single CT section demonstrates solid left lower lobe nodule with minimal surrounding halo. Surgical resection revealed Cryptococcus infection.

279

280

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 15.19 Primary histoplasmosis. A 27 year old woman with flu-like illness 2 weeks after cave exploration in Mexico. A PA chest radiograph demonstrates three or four poorly marginated nodular opacities overlying the left lower lobe. Her husband, who was suffering from similar symptoms, had identical nodules.

Less commonly a segmental or lobar pneumonia is seen. Chronic pulmonary histoplasmosis radiologically resembles post-primary tuberculosis, with upper lobe contraction, calcification and cavitation. There may be substantial adjacent pleural thickening. Hilar and mediastinal lymph nodes are frequently enlarged48. If unsuspected the disease may be diagnosed in retrospect years later by the appearance on chest radiograph of multiple calcified 3–4 mm in diameter sharplymarginated, round nodules and calcified lymph nodes in the hila and mediastinum (Fig. 15.20). Occasionally a solitary, well defined nodule may form and is then termed a histoplasmoma. When the centre of this lesion calcifies it forms a ‘target’ lesion which is very specific for this entity (Fig. 15.21). In some cases of histoplasmosis fibrosing mediastinitis may develop and can lead to constriction of mediastinal structures, including the airways, superior vena cava, pulmonary arteries and pulmonary veins. CT is helpful in demonstrating this complication.

Figure 15.20 Chronic histoplasmosis. A 55 year old man undergoing routine pre-employment physical examination. A PA chest radiograph demonstrates several well defined uniform-sized calcified nodules in both lungs with bilateral hilar and mediastinal lymph node calcification. The patient was a long-time resident of the central USA, an area endemic for histoplasmosis.

The chest radiographic manifestations are variable49. In primary coccidioidomycosis unifocal or multifocal homogeneous opacities resembling community-acquired bacterial pneumonia may be seen. Cavitation and hilar/mediastinal adenopathy may be seen with approximately 20% of these lesions (Fig. 15.22). Often the pneumonias are round and produce thin-walled cavities that can lead to a pneumothorax. Primary disease almost invariably resolves spontaneously or reveals only small residual linear or nodular scars. Chronic fibronodular cavitary disease resembling post-primary tuberculosis may also be encountered. CT demonstrates expected findings of soft tissue solitary nodule, sometimes with a low attenuation centre, cavitation, or calcification. Ground-glass halos may be seen around these nodules50. Disseminated coccidioidomycosis may cause miliary nodules.

Coccidioidomycosis Coccidioidomycosis is caused by Coccidioides immitis, a fungus which is found in soil in arid regions of the southwestern USA and northern Mexico47. Infection is acquired by inhaling dust containing the fungus. Almost half the patients develop a febrile illness; the rest are asymptomatic. The illness usually resembles a non-specific viral infection but may present with erythema nodosum, erythema multiforme and arthritis. In most instances, the disease is self-limiting, but it may progress and can even result in fatal disseminated disease, particularly in the immunocompromised patient.

Figure 15.21 Histoplasmoma. The chest radiograph obtained for left lower lobe pneumonia in a 68 year old man. The right upper lobe demonstrates central target calcification with surrounding soft tissue opacity very suggestive of histoplasmoma.

CHAPTER 15

Figure 15.22 Coccidioidomycosis. An 18 year old man with flu-like symptoms. The PA chest radiograph demonstrates homogeneous left upper lobe opacity with cavitation as well as ipsilateral hilar adenopathy.

North American blastomycosis North American blastomycosis is due to Blastomyces dermatiditis47. Pulmonary infection may be accompanied by infection of the skin, bones and genitourinary tract. Pulmonary blastomycosis is often asymptomatic.When symptoms do occur, they may be non-specific or may be those of an acute pneumonia. The chest radiograph reveals homogeneous unifocal or multifocal segmental or lobar opacification indistinguishable from acute pneumonia. Cavitation occurs in approximately 15% of cases. Sometimes, the pneumonia is spherical in shape, closely resembling bronchial carcinoma.51 Pleural thickening or pleural effusion may accompany the pneumonia in 10–15% of cases. Lymph node enlargement is infrequent. Blastomycosis may cause miliary nodules particularly in immunocompromised patients. A chronic fibrocavitary form of the disease, which resembles post-primary tuberculosis, is also seen. CT reveals pulmonary masses, consolidation, effusions and cavitation as seen on chest radiographs.

•

PULMONARY INFECTION IN ADULTS

of bronchial carcinoma. When considering this differential diagnosis, it should always be borne in mind that mycetomas may be difficult to see in portions of the lung that have been severely distorted by previous cavitary disease. On occasion, bleeding may be severe enough to warrant surgical resection of the lung containing the mycetoma, and bronchial artery embolization may sometimes be helpful in management. The radiological diagnosis depends on recognizing a mass within a cavity and formation of an air crescent (Fig. 15.23). Calcification within the mass is extremely rare and fluid levels are infrequent. The cavity wall and the adjacent pleura may thicken. A freely-moving fungus ball can be demonstrated on decubitus images (Fig. 15.24). The differential diagnosis of mycetoma includes blood clot or lung debris in a cavity, echinococcal disease and cavitary neoplasm. Allergic bronchopulmonary aspergillosis (ABPA) describes a hypersensitivity reaction which occurs in the major airways. It is associated with elevated serum IgE, positive serum precipitins and skin reactivity to aspergillus, and it is the most common cause of pulmonary eosinophilia in the UK. The radiographic appearances consist of: nonsegmental areas of opacity most common in the upper lobes, lobar collapse, branching thick tubular opacities due to bronchi distended with mucus and fungus, and occasionally pulmonary cavitation. Bronchial wall thickening with tramlines and ring formation indicates bronchiectasis. The lungs are often overinflated, while late in the disease there may be volume loss due to fibrosis. In chronic necrotizing (formerly semi-invasive) aspergillosis, local invasion of lung parenchyma takes place, usually in the upper lobes.This form is seen in debilitated patients and in patients with pre-existing lung damage or chronic lung disease. Thus there is some decrease in host response to the fungus. Radiographically heterogeneous opacities resembling tuberculosis are followed by an enlarging, thick-walled cavity which develops over a period of weeks. Adjacent pleural thickening

Aspergillus infection Aspergillus infections of the lung are usually caused by Aspergillus fumigatus and can take different forms depending on an individual’s immune response to the organism52. Aspergillus mycetomas are saprophytic growths which colonize a pre-existing cavity in the lung. There is relatively little invasion of the cavity wall or surrounding lung. Most cavities (e.g. from sarcoidosis or tuberculosis) and thus mycetomas are in the upper lobes or superior segments of the lower lobes. The great majority of aspergillomas are asymptomatic. Haemoptysis is the important complication. When it occurs, there may be difficulty in distinguishing between mycetoma formation, reactivation of tuberculosis and the development

Figure 15.23 Aspergilloma. A 30 year old man with a right upper lobe cavity of uncertain etiology. Within the cavity there is a rounded soft tissue opacity and superior air crescent indicative of mycetoma.

281

282

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 15.24 Aspergilloma. A 60 year old man with chronic lung disease of indeterminate aetiology. Patient presents with mild haemoptysis. (A) Supine and (B) prone CT of the upper lobe demonstrate a fungus ball moving within the left upper lobe cavity.

and involvement of the chest wall may occur. Occasionally a mycetoma is observed. Bilateral involvement may occur and multiple nodules have been reported. Invasive aspergillosis is virtually confined to immunocompromised hosts. Radiologically, the appearances are variable, but a common pattern is of one or more rounded poorly marginated areas of homogeneous opacification with or without air bronchograms (Fig. 15.25). With time the margins may become more discreet and the lesions resemble masses. They may cavitate with the formation of an air crescent (Fig. 15.26). Wedge-shaped peripheral opacities, believed to be pulmonary infarcts may also be seen. Rarely, miliary nodules are observed.

Figure 15.25 Invasive aspergillosis. A 65 year old man with immunosuppression and neutropenia presents with fever. A PA chest radiograph demonstrates multiple bilateral poorly marginated nodular opacities with some coalescence of nodules in the right lower lobe.

Figure 15.26 Invasive aspergillosis. A 46 year old man on steroids and methotrexate for asthma. The patient developed cough and fever. (A) Initial chest radiograph demonstrates peripheral homogeneous opacities in both upper lobes and heterogeneous opacities in the left lower lobe. (B) A subsequent PA chest radiograph demonstrates cavitation within the peripheral left opacity following appropriate antifungal therapy and reconstitution of neutrophils 6 d into course.

CHAPTER 15

CT reveals findings as expected from chest radiography but is more sensitive and may be abnormal when radiography is normal. CT may reveal ground-glass haloes around the nodules, as may be seen in other diseases (Fig. 15.27). Optimized HRCT may permit more specific diagnoses of angio-invasive infection by actually demonstrating occlusion of the vessel supplying the focal pulmonary lesion53.

PROTOZOAL AND METAZOAL DISEASES Protozoal infections Pleuropulmonary amoebiasis caused by Entamoeba histolytica is usually secondary to liver involvement and develops in about one-fifth of such patients. It is characteristically a disease of young adults and shows a distinct male preference. Lung involvement usually occurs in the right lung base and consists of hemidiaphragmatic elevation, pleural effusion and/or thickening and plate-like atelectasis. With erosion of a liver abscess through the diaphragm, basal homogeneous opacification develops and frequently cavitates. Occasionally haematogenous spread gives rise to similar disease in other lung locations. Amoebiasis should always be considered in the appropriate clinical setting with isolated right basal radiological changes that abut the hemidiaphragm54.

Metazoal infestations With the exception of Armillifer armillatus, metazoal pulmonary disease is due to either roundworms or flatworms (tapeworms and flukes). Roundworms generally cause pulmonary consolidation with eosinophilia (acute eosinophilic pneumonia) and very occasionally an isolated pulmonary nodule (Dirofilaria immitis)55.The manifestations of flatworm infestations are more varied, and paragonimiasis and echinococcosis are considered here in detail. Two others, Taenia solium and Schistosoma sp.,

Figure 15.27 Invasive aspergillosis. A 30 year old man with neutropenia presents with fever. A chest radiograph demonstrated multiple lung nodules. This thin CT slice demonstrates heterogeneous and ground-glass opacity in the azygo-oesophageal lung recess as well as a lingular nodule consisting of an opaque centre and ground-glass halo very suggestive of invasive aspergillosis.

•

PULMONARY INFECTION IN ADULTS

can also cause radiological changes on the chest radiograph. T. solium gives rise to multiple calcified cysticerci in the chest wall muscles which appear as calcified oval opacities of 3–10 mm in diameter. Schistosoma infestation manifests as pulmonary consolidation with eosinophilia, pulmonary arterial hypertension with or without miliary nodulation or reticulonodular opacities, and slightly larger individual nodules.

Paragonimiasis Paragonimiasis is caused by a fluke (Paragonimus westermani) that develops from a larval form in the lung, where it lives, often for years, producing ova. Water snails and crustaceans are intermediate hosts and infestations are acquired from eating raw or incompletely cooked fresh water crabs and crayfish. The disease mainly occurs in the Far East, southeast Asia and Africa, and the usual presentation is with chronic cough and sputum production with haemoptysis. Radiological changes tend to be bilateral and affect any lobe (particularly the mid lung). Various abnormalities are observed including a mixture of consolidation, nodules and band, tubular and ring opacities56. Ring opacities range in size from 5 to 30 mm and may or may not be accompanied by adjacent consolidation. In some series about half the patients have had a pleural effusion. In the lower lobes parenchymal changes mimic bronchiectasis, and in the upper lobes, tuberculosis. CT may reveal peripheral linear opacities thought to be worm migration tracks56. The diagnosis is established by detecting ova in the sputum or by identifying anti-Paragonimus antibody in the blood.

Hydatid disease Hydatid disease (echinococcosis) is caused by a tapeworm, usually Echinococcus granulosus. Humans are accidental hosts and acquire infection by ingesting ova from fomites or contaminated water and by direct contact with dogs. Cysts develop in the lung, or less commonly in the mediastinum and rarely in the heart and pulmonary arteries. They are usually solitary but in one series approximately 10% of cases were multiple and/or bilateral57. CT series have demonstrated a higher percentage of multiple cysts. They may be ruptured (two-thirds) or unruptured (one-third) at the time of presentation. The radiological findings with unruptured pulmonary cysts are one or more homogeneous, roughly spherical or oval, sharply demarcated mass lesions. They range in size from 1 to 10 cm and occur particularly in the mid and lower lobes. The intrapulmonary masses are of soft tissue density and almost never calcify, unlike mediastinal lesions. Cysts are easily deformed and this leads to: lobulation or eccentricity of contour, where they come up against major bronchovascular structures; flattening of peripheral aspects in contact with the chest wall or mediastinum, in such a way that in the latter case a mediastinal mass is simulated; and changes in shape with breathing. Cyst rupture is usually associated with secondary infection and may occur into the airways or pleural space. Acute symptoms often develop and frequently precipitate presentation. There are three layers to the wall, two in the cyst itself (endo- and ecto-cyst) and a third derived from the surrounding lung (pericyst). If the two inner layers remain intact, airway

283

284

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

communication results in a ring opacity containing a rounded, homogeneous density. The radiographic appearance resembles the air crescent of a mycetoma. Should there be disruption of the inner layers, a complex cavitary lesion results with one or more of the following radiographic features: an air–fluid level, a floating membrane (water lily sign, camalote sign), a double wall, an essentially dry cyst with crumpled membranes lying at its bottom (rising sun sign, serpent sign), a cyst with all its contents expectorated (empty cyst sign). Secondary infection of a

hydatid cyst may produce a lung abscess with or without surrounding lung opacity. CT can demonstrate a number of these characteristic features to better advantage than the chest radiograph58. High specificity of CT for the diagnosis of perforated pulmonary hydatid cyst (‘air bubble’ sign) has been reported59. Rupture into the pleural space causes an effusion or, if there is additional airway communication, a hydropneumothorax.The diagnosis may be established by serological testing, or examination of the sputum if there is rupture into airways.

PULMONARY COMPLICATIONS OF HIV INFECTION AND AIDS INFECTIONS The great majority of the pulmonary complications of HIV infection are infectious in origin. Of these, the three most important are Pneumocystis jiroveci (carinii) pneumonia, which is still a common disease leading to the diagnosis of AIDS; tuberculosis, which is extremely common in developing countries and is thus by far the most important (worldwide) pulmonary complication of HIV infection; and communityacquired pneumonias and airways disease. The prevalence of HIV-associated pulmonary infections varies greatly between different countries, so that it is difficult to identify them in order of importance; the clinical and radiographic features of each of the most common causative agents are considered below.

Pneumatoceles are seen in 10% of patients with Pneumocystis pneumonia and AIDS. They generally appear within a few days of the initial pneumonia, are thin walled, may rapidly increase or decrease in size, and over the course of 2–3 months

Pneumocystis jiroveci (formerly carinii) P. jiroveci, formerly carinii, was for a long time considered a parasite but has been shown to be a fungus. The widespread use of the abbreviation PCP to refer to pneumonia caused by this organism continues and will be used here. P. jiroveci is ubiquitous and is believed to infect the majority of humans early in life. It only manifests as pneumonia when immunosuppressive disorders, including AIDS, cause a profound depression of cellular immunity. The majority of HlV-infected patients who develop PCP present with fever, dyspnoea, non-productive cough, weakness and weight loss. Elevation of the serum lactase dehydrogenase (LDH) has been used as a sensitive but non-specific indicator of PCP and may have some prognostic value. The definitive diagnosis of PCP is made by demonstrating typical organisms in secretions obtained from the lungs or in lung tissue itself. The chest radiograph in most patients with Pneumocystis pneumonia demonstrates bilateral, diffuse, symmetrical, fine to medium reticular opacities. The same pattern confined to one or two lobes or segments of lung may be seen (Fig. 15.28). In some cases upper lobe predominance of infiltrates has simulated reactivation tuberculosis. Unusual radiographic presentations include diffuse or focal miliary nodules, homogeneous opacities, solitary or multiple well formed nodules and moderate to thick-walled cavitary nodules60. Approximately 5–10% of patients will have normal chest radiographs at presentation, perhaps even more frequently in less severe disease. Pleural fluid and lymphadenopathy are rare or do not occur unless extrapulmonary involvement has been observed, usually in patients who have received prophylactic aerosolized pentamidine.

Figure 15.28 Pneumocystis jiroveci pneumonia (PCP). (A) PA chest radiograph demonstrates the typical bilateral distribution of fine to medium reticular opacities. (B) A close-up of the right upper lobe of the same patient reveals the nature of this typical pattern which occasionally is confined to one lobe.

CHAPTER 15

gradually resolve61. In some cases they persist as chronic, thinwalled, air-filled cavities (Fig. 15.29). Spontaneous pneumothorax, has been observed in approximately 5% of patients with PCP (Fig. 15.30). Management is notoriously difficult, because bronchopleural fistulas are common.

•

PULMONARY INFECTION IN ADULTS

Significant radiographic improvement is usually seen within 10 d of beginning treatment.The radiographic appearance may get worse during the first 3 d of therapy, especially with intravenous trimethoprim–sulphamethoxazole, possibly related to overhydration pulmonary oedema and possibly to an inflammatory reaction to dead and dying parasites (Fig. 15.31). Eventually, in most patients, the pneumonia resolves completely and the chest radiograph becomes normal, though in some, pulmonary fibrosis develops. The value of CT in PCP diagnosis is questionable given the cost of imaging but it has been used to differentiate between pneumonias and to exclude the diagnosis62. CT demonstrates unilateral or bilateral ground-glass or homogeneous opacities in geographical distribution. The possibility of recurrence of PCP due to immune reconstitution following retroviral therapy has been reported63.

Mycobacterium tuberculosis In the USA tuberculosis has occurred in approximately 4% of all patients with AIDS, but in sub-Saharan Africa, where millions of people are co-infected with HIV and tubercle bacilli,

Figure 15.29 Pneumocystis jiroveci pneumonia (PCP). (A) A close-up of the right upper lobe demonstrates medium reticular opacities. (B) This close-up of the right upper lobe was obtained 3 weeks after (A). The lung disease has resolved but a thin-walled pneumatocele 30 mm in diameter is now demonstrated (arrows).

Figure 15.30 Pneumocystis jiroveci pneumonia (PCP). An AP chest radiograph demonstrates a large right tension pneumothorax. Severe underlying bilateral pneumocystis pneumonia was diagnosed. What probably represents a large left upper lobe pneumatocele is also noted.

Figure 15.31 Pneumocystis jiroveci pneumonia (PCP). (A) AP chest radiograph demonstrates a mild, diffuse reticular pattern typical of P. jiroveci infection. (B) On the fourth day of trimethoprim– sulphamethoxazole therapy the chest radiograph demonstrates a worse, coarse, bilateral reticular pattern. Following diuretic therapy the radiographic appearance quickly returned to baseline.

285

286

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

tuberculosis has become the most common and most important manifestation of AIDS. Worldwide, tuberculosis is the most common cause of death among AIDS patients. Clinical features depend on the stage of HIV-induced immunosuppression at the time that tuberculosis is encountered. Tuberculosis may be indistinguishable from ‘ordinary’ disease in patients whose HIV infection is at an early stage and whose cellular immunity is better preserved. Cutaneous reactivity to tuberculin is present, the chest radiograph is typical, and extrapulmonary disease occurs with the same frequency (15–20%) as it does in patients not infected with HIV. However, when tuberculosis develops in the late stages of HIV disease, patients often have a negative tuberculin skin test reaction; more than half will have extrapulmonary (especially lymph node) involvement; and the chest radiographs are usually atypical. The diagnosis of tuberculosis rests on isolation and identification of M. tuberculosis. Antituberculous chemotherapy is usually started with three or four drugs, including isoniazid, rifampicin and pyrazinamide, with ethambutol and or other drugs added when there is a possibility of drug resistance. Longer courses of treatment than usual have been recommended. In patients with early HIV disease and limited immunosuppression the chest radiograph is similar to that seen in the general population. Homogeneous segmental or lobar opacification mimicking Strep. pneumoniae is observed with or without ipsilateral hilar and/or mediastinal adenopathy. Alternatively the post-primary type of disease with upper lobe clumped nodular or coalesced coarse reticulonodular opacities with or without cavitation may be noted. However, with advanced HIV-induced immunosuppression, diffuse bilateral coarse reticulonodular opacities are typically demonstrated. The pattern is commonly distinctive enough to suggest that an infection other than Pneumocystis pneumonia is present. Hilar and/or mediastinal adenopathy is well documented, as is a mid or lower lobe predominance of lesions (Fig. 15.32). Cavitation is not expected in this setting64. A high prevalence of pleural effusion was observed in some HIV-infected patients with tuberculosis, but this has not been universally observed.With effective antituberculosis therapy, the majority of patients will demonstrate both clinical and radiographic improvement within 1–2 weeks. In comparison, if the radiographic appearances deteriorate, a second disease process or possibly infection with a drug-resistant strain of M. tuberculosis warrants consideration65. The use of highly active antiretroviral therapy (HAART) may lead to increasing mediastinal adenopathy and worsening or new lung opacities due to a systemic inflammatory response resulting from immune reconstitution66.

Mycobacterium avium complex Mycobacterium avium complex (MAC), a ubiquitous microorganism that is found in house dust, soil and water, may cause complications for patients with end-stage HIV infection but has decreased in incidence as antiretroviral regimens have been administered. The role of MAC in causing pulmonary abnormalities should be taken seriously since disseminated MAC infection decreases life expectancy. There are no distinctive chest radiographic abnormalities in patients with MAC infection and AIDS;67 moreover, other

Figure 15.32 Tuberculosis. A PA chest radiograph demonstrates a diffuse, bilateral coarse nodular pattern associated with right hilar adenopathy. This combination of findings should suggest the presence of fungal or mycobacterial disease.

associated opportunistic diseases are common. Among patients with MAC, diffuse bilateral opacities, focal consolidation, pleural fluid, adenopathy and normal radiographs have been reported. Cavitation is rare. Considerable overlap has been found in the CT findings of nontuberculous mycobacterial infection and tuberculosis. Both demonstrate centrilobular nodules, some ground-glass attenuation and lymphadenopathy68.

Other nontuberculous mycobacteria Virtually all nontuberculous mycobacterial disease in AIDS patients is caused by MAC, but M. kansasii, M. gordonae, M. fortuitum and M. chelonei have also been implicated. The radiographic features of these agents in AIDS patients are variable and include diffuse infiltrates, focal infiltrates, cavitation, heterogeneous and homogeneous opacities predominantly in the upper lobes, lymphadenopathy and pleural effusions69,70.

Pyogenic organisms Community-acquired pneumonias, especially Strep. pneumoniae and Haemophilus influenzae, are common in HIV-infected patients. Patients with HIV infection also develop pneumonias related to their cell-mediated immune deficiencies from organisms such as Nocardia asteroides, Salmonella sp. and Legionella sp. Infections with Staph. aureus, Moraxella, Branhamella catarrhalis, Rhodococcus equi and M. pneumoniae are also reported. The clinical presentation of HIV-related Strep. pneumoniae or H. influenzae pneumonia is indistinguishable from that seen in the normal host, and is characterized by the sudden onset of high fever and a productive cough71. The radiographic features of pyogenic bacterial pneumonia in patients with AIDS are similar to those seen in non-immunosuppressed individuals. Chest radiographs demonstrate focal or multiple areas of lobar homogeneous opacification72 and, occasionally, pleural effusion (Fig. 15.33). Bronchitis and bron-

CHAPTER 15

•

PULMONARY INFECTION IN ADULTS

empyema are also seen. This aetiology should be considered when cavitary pneumonias are observed in HIV patients with low CD4 lymphocyte counts. CT reveals mediastinal lymph node enlargement simulating lymphoma74.

Cytomegalovirus

Figure 15.33 Streptococcal pneumoniae pneumonia. An AP radiograph demonstrates homogeneous opacification of the right lower lobe. This pattern of pyogenic bacterial pneumonia in AIDS patients is no different from that seen in nonimmunosuppressed individuals.

chiectasis may develop acutely or may be a sequel of recurrent pyogenic pneumonias in patients with AIDS. The abnormalities on plain radiographs include peribronchial thickening and ‘tram tracking’. On CT, dilated bronchi and peribronchial thickening with or without mucous plugs are observed73. Airways disease may also result from other infectious or noninfectious aetiologies including lymphocytic interstitial pneumonia (LIP), PCP and tuberculosis. Pulmonary nocardiosis presents with cavitation, abscesses, mixed heterogeneous and homogeneous patterns and pleural fluid. Rhodococcus equi may cause necrotizing pneumonia. Patients present with chest pain, fever, productive cough and haemoptysis. Chest radiographs demonstrate homogeneous opacity frequently with cavitation (Fig. 15.34). Pleural effusions and

Cytomegalovirus pneumonia is well documented in immunocompromised patients, especially after bone marrow, kidney, lung, or heart transplantation. However, its role in respect to pneumonia in patients with AIDS is problematical due to superimposed infections and difficulty in establishing the diagnosis. Cytomegalovirus can be isolated from respiratory secretions or lung tissue in one-quarter to one-third of patients with AIDS-related lung diseases, especially PCP. In this setting it is impossible to identify what contribution, if any, the virus is making to the clinical and radiographic features of the patient’s pulmonary disorder. The principal chest radiographic abnormality of suspected cytomegalovirus pneumonia in patients with AIDS is a bilateral fine reticular pattern, similar to that seen with PCP. An analysis of 21 patients with AIDS and cytopathological evidence of cytomegalovirus infection revealed chest CT abnormalities, including ground-glass opacities and dense consolidation, bronchial wall thickening, heterogeneous densities and discrete nodules75.

Cryptococcus neoformans The usual disease caused by Cryptococcus neoformans in patients with HIV infection is meningitis, but about one-third of these patients have simultaneous pulmonary involvement and present with respiratory symptoms. Patients with cryptococcal pneumonia typically present with weight loss, fever, productive cough and breathlessness. The definitive diagnosis is usually made by cytology and culture of induced sputum or broncho-alveolar lavage fluid. When cryptococcal lung disease is found, the patient should be evaluated for meningeal involvement. The most common chest radiographic manifestation is diffuse reticular opacification. However involvement may be limited to one lung or lobe and focal homogeneous opacity, pleural effusion, hilar adenopathy, and cavitation have all been described (Fig. 15.35). CT may provide additional information76.

Histoplasma capsulatum

Figure 15.34 Rhodococcus equi. A 22 year old man with fever and cough. A PA chest radiograph demonstrates a thick-walled cavity adjacent to heterogeneous opacities in the right upper lobe. Necrotizing pneumonias are typical with this organism.

Cases of AIDS-related progressive disseminated histoplasmosis have been recognized in endemic regions of the USA. The same phenomenon may occur in endemic areas of Central and South America and elsewhere. Progressive systemic disease produces prominent weight loss and prolonged fever. Cough and dyspnoea are common in patients with chest radiographic abnormalities.The diagnosis is usually made by the biopsy and culture of bone marrow or blood. The radiographic features of histoplasmosis in patients with AIDS include normal chest radiographs in 50% of AIDS patients with extrapulmonary histoplasmosis, nodular or linear opacities in 50%, nearly 20% with pleural effusions, and approximately 10% with lymphadenopathy77 (Fig. 15.36).The lung abnormalities may be coarse and nodular, thus distinguishable from the

287

288

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 15.35 Cryptococcus. An AP chest radiograph demonstrates a fine reticular pattern, right paratracheal and hilar adenopathy, and left pleural fluid. Although the pattern may simulate that seen with PCP, the presence of adenopathy and pleural fluid should direct attention towards another diagnosis.

fine reticular pattern seen with PCP. The presence of lymphadenopathy would also distinguish histoplasmosis from PCP.

Figure 15.37 Coccidioidomycosis. A close-up of a PA chest radiograph demonstrates coarse reticulonodular opacities typical of disseminated fungal or mycobacterial disease in AIDS patients. The possibility of left hilar adenopathy is raised.

Coccidioides immitis

Aspergillosis

Coccidioidomycosis, in AIDS patients may be seen in the endemic areas of the southwestern USA. Fever, weight loss, cough and fatigue are usually present in patients with AIDS-related disseminated coccidioidomycosis. Spherules can usually be identified in sputum and fungal cultures from broncho-alveolar lavage, or transbronchial biopsy are likely to be positive. Diffuse, medium to coarse nodular opacities similar to those seen with histoplasmosis and with disseminated tuberculosis were seen in 55% of patients; 36% had focal abnormalities, including single or multiple nodules, cavities and hilar and mediastinal adenopathy78 (Fig. 15.37).

Despite its importance as a complication of other immunosuppressive disorders, especially leukaemia and organ transplantation, aspergillosis is relatively infrequent in patients with HIV-induced immune deficiency. Two types of HIV-associated aspergillosis have been observed: invasive pulmonary aspergillosis, which is characterized by prolonged cough and fever; and obstructing bronchial aspergillosis, which is characterized by breathlessness, cough and chest pain. The radiographic features of aspergillosis in patients with AIDS include focal and occasionally persistent homogeneous opacities. These may remain stable for several months but are indicative of invasive aspergillosis. Approximately one-third of patients demonstrate cavitation, chiefly in upper lobe homogeneous opacities. Disseminated heterogeneous and homogeneous patterns are observed with widespread disease. CT may reveal nodules with surrounding ground-glass halos, or airway disease manifest by centrilobular nodules and peribronchial opacities79.

Protozoal diseases Toxoplasma gondii

Figure 15.36 Histoplasmosis. A PA chest radiograph demonstrates the typical findings of histoplasmosis in AIDS patients. A pattern composed of small nodules 2–4 mm in diameter is seen in both lungs.

Judging from serological studies, Toxoplasma gondii is a prevalent infection (40–60%) in the general population. Reactivation of central nervous system (CNS) toxoplasmosis is a common cause of seizures, focal neurological deficits and/or encephalopathy in patients with AIDS. In view of the frequency of CNS toxoplasmosis, pulmonary involvement is surprisingly unusual, though it may occur and progress to respiratory failure. Chest radiographs in the few reported patients demonstrate bilateral lower lobe homogeneous opacities, a solitary nodule and bilateral diffuse heterogeneous opacities. In our series, bilateral coarse nodular opacities were the most common

CHAPTER 15

•

PULMONARY INFECTION IN ADULTS

Kaposi’s sarcoma

Kerley B lines are occasionally present. The lung abnormalities tend to coalesce together unlike the nodules seen with lymphoma82. Unilateral or bilateral pleural fluid is reported in 33–50% of patients and lymphadenopathy has been observed in 10–30% (Fig. 15.39). Rapid progression from a poorlydefined nodular pattern to one of airspace consolidation has usually been seen in patients with haemoptysis, and probably represents haemorrhage into the lung (Fig. 15.40). Unusual radiographic features of KS include pericardial fluid accumulation due to pericardial KS and the plain chest

At the beginning of the AIDS epidemic, Kaposi’s sarcoma (KS) was the indicator disease in 20–25% of all cases but its incidence has fallen considerably81. KS is caused by the human herpes virus 8 (HHV8). Most patients with KS present with one or more typical violaceous plaques on their skin or mucous membranes. Primary involvement of visceral organs may also occur, particularly in the gastrointestinal tract; other sites include lymph nodes, liver, spleen, heart and pericardium. Primary pulmonary KS is unusual. Most cases of pulmonary KS occur in the presence of obvious disease in the skin or elsewhere. The manifestations of pulmonary KS are fever, cough and breathlessness, and thus are indistinguishable from those of many pneumonias. Haemoptysis and upper airway obstruction rarely occur. Dyspnoea from effusions may require treatment. Compared with the non-specific radiographic abnormalities observed with the opportunistic pulmonary infections discussed above, the findings of KS are more specific and the diagnosis may sometimes be suggested on the basis of the chest radiograph and CT abnormalities.The usual pattern is a poorly defined peribronchovascular nodular opacity which typically measures 10–20 mm in diameter. Although solitary nodular KS may occur, bilateral multiple lesions are typically present. Coarse linear opacities are also commonly scattered throughout the lungs, particularly in the perihilar and lower lungs and

Figure 15.39 Kaposi’s sarcoma. A PA chest radiograph demonstrates coarse linear opacities in the perihilar regions. Some nodular opacities are noted in the right upper lobe. Left pleural fluid is present. This constellation of findings is highly suggestive of Kaposi’s sarcoma.

Figure 15.38 Toxoplasmosis. A close-up of a portable AP chest radiograph demonstrates a coarse nodular pattern which was present in the perihilar and lower lobe regions of the lung.

Figure 15.40 Kaposi’s sarcoma. Fairly large, coarse nodules are noted in the right upper lobe and left lung. An area of homogeneous opacification in the right middle lobe appeared at the same time as this patient developed haemoptysis. This radiographic abnormality probably represents haemorrhage in the lung.

pattern; lymphadenopathy was not observed and pleural effusion was uncommon80 (Fig. 15.38).

MALIGNANCIES Some malignancies have close association with infectious agents, and as they may mimic infection they are included here.

289

290

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

diffuse parenchymal opacities, pleural fluid and rapidlygrowing well defined parenchymal nodules (Fig. 15.41). These nodules range in size from 10 to 60 mm, may be solitary or multiple and only rarely cavitate. CT may reveal similar findings and would be expected to be more sensitive than chest radiography.

REFERENCES

Figure 15.41 Non-Hodgkin’s lymphoma. (A) A close-up of a PA chest radiograph demonstrates a fairly well defined nodule, 15 mm in diameter, in the left upper lobe. (B) A close-up of the chest radiograph obtained 10 d later demonstrates considerable enlargement of the nodule which, on open lung biopsy, was shown to represent non-Hodgkin’s lymphoma.

radiographic demonstration of a large intratracheal Kaposi’s lesion. Endobronchial lesions may cause complete occlusion of airways. Occasionally septic emboli in intravenous drug abusers with AIDS may simulate KS. In this situation cavitation would be expected, whereas cavitation in KS is rare or not observed.

Non-Hodgkin’s lymphoma Although intrathoracic involvement by non-Hodgkin’s lymphoma in patients with AIDS is not common, when lesions occur they include: hilar and/or mediastinal adenopathy,

1. Franquet T 2001 Imaging of pneumonia: trends and algorithms. Eur Respir J 18: 196–208 2. Tarver R D, Teague S D, Heitkamp D E, Conces D J Jr 2005 Radiology of community-acquired pneumonia. Radiol Clin North Am 43: 497–512 3. Rome L, Murali G, Lippmann M 2001 Nonresolving pneumonia and mimics of pneumonia. Med Clin North Am 85: 1511–1530 4. Coletta F S, Fein A M 1998 Radiological manifestations of Legionella/ Legionella-like organisms. Semin Respir Infect 213: 109–115 5. Mabeza G F, Macfarlane J 2003. Pulmonary actinomycosis. Eur Respir J 21: 545–551. 6. Cheon J E, Im J G, Kim M H et al 1998 Thoracic actinomycosis: CT findings. Radiology 209: 229–233 7. Feigin D S 1986 Nocardiosis of the lung: chest radiographic findings in 21 cases. Radiology 159: 9–14 8. Yoon H K, Im J G, Ahn J M et al 1995 Pulmonary nocardiosis: CT findings. J Comput Assist Tomogr 19: 52–55 9. Wright P W, Wallace R J, Shepherd J R 1990 A descriptive study of 42 cases of Branhamella catarrhalis pneumonia. Am J Med 88: 5A, 2S–8S 10. Itoh I, Ishida T, Hashimoto T et al 2000 Chest radiograph of atypical pneumonia: comparison among Chlamydia pneumoniae, Pneumonia ornithosis, and Mycoplasma pneumoniae pneumonia. Kanesenshogaku Zasshi 74: 954–960 11. MacFarlane J T, Miller A C, Roderick Smith W H et al 1984 Comparative radiographic features of community acquired Legionnaires’ disease, pneumococcal pneumonia, mycoplasma pneumonia, and psittacosis. Thorax 39: 28–33 12. McConnell C T, Plouffe J F, File T M et al, and CBPIS Study Group 1994 Radiographic appearances of Chlamydia pneumoniae (TWAR strain) respiratory infections. Radiology 192: 819–824 13. Kauppinen M T, Lahde S, Syrjala H 1996 Roentgenographic findings of pneumonia caused by Chlamydia pneumonia: a comparison with Streptococcus pneumonia. Arch Intern Med 156: 1851–1856 14. Gikas A, Kofteridis D, Bouros D et al 1999 Q fever pneumonia: appearance on chest radiographs. Radiology 210: 339–343 15. Caron F, Meurice J C, Ingrand P et al 1998 Acute Q fever pneumonia. A review of 80 hospitalized patients. Chest 114: 808–813 16. Faul J L, Doyle R L, Kao P N et al 1999 Tick-borne pulmonary disease. Update on diagnosis and management. Chest 116: 222–230 17. Rubin S A 1978 Radiographic spectrum of pleuropulmonary tularemia. Am J Roentgenol 131: 277–281 18. Weinberg A N 1991 Respiratory infections transmitted from animals. Infect Dis Clin North Am 5: 649–661 19. Alsofrom D J, Mettler F A, Mann J M 1981 Radiographic manifestations of plague in New Mexico, 1975–1980. Radiology 139: 561–565 20. Im J-G, Yeon K M, Han M C, et al 1989 Leptospirosis of the lung: radiographic findings in 58 patients. Am J Roentgenol 152: 955–959 21. Wiita R M, Cartwright R R, Davis J G 1961 Staphylococcal pneumonia in adults. A review of 102 cases. Am J Roentgenol 86: 1083–1091 22. Shah R M, Wechsler R, Salazar A M, Spirn P W 2002 Spectrum of CT findings in nosocomial Pseudomonas aeruginosa pneumonia. J Thorac Imaging 17: 53–57 23. Robert R, Grollier G, Frat J P et al 2003 Colonization of lower respiratory tract with anaerobic bacteria in mechanically ventilated patients. Intensive Care Med 29: 1062–1068 24. Groskin S A, Panicek D M, Ewing D K et al 1991 Bacterial lung abscess: a review of the radiographic and clinical features of 50 cases. J Thorac Imaging 6: 62–67

CHAPTER 15

25. Reittner P, Muller N L, Heyneman L et al 2000 Mycoplasma pneumoniae pneumonia: Radiographic and high-resolution CT features in 28 patients. Am J Roentgenol 174: 37–41 26. Okonomou A, Muller N L, Nantel S 2003 Radiographic and highresolution CT findings of influenza virus pneumonia in patients with hematologic malignancies. Am J Roentgenol 181: 507–511 27. Klinger J R, Sanchez M, Curtin L A et al 1998 Multiple cases of lifethreatening adenovirus pneumonia in a mental health care center. Am J Respir Crit Care Med 157: 645–649 28. Matar L D, McAdams H P, Palmer S M et al 1999 Respiratory viral infections in lung transplant recipients: radiologic findings with clinical correlation. Radiology 213: 735–742 29. Han B K, Son J A, Yoon H K, Lee S 1998 Epidemic adenoviral lower respiratory tract infection in pediatric patients: radiographic and clinical characteristics. Am J Roentgenol 170: 1077–1080 30. Miyake H, Matsumoto A, Komatsu E 1998 Infectious mononucleosis with pulmonary consolidation. J Thorac Imaging 11: 158–160 31. Kim J S, Ryu C W, Lee S I 1999 High-resolution CT findings of varicellaZoster pneumonia. Am J Roentgenol 172: 113–116 32. Aquino S L, Dunagan D P, Chiles C et al 1998 Herpes simplex virus 1 pneumonia: patterns on CT scans and conventional chest radiographs. J Comput Assit Tomogr 22: 795–800 33. Ketai L H, Williamson M R, Telepak R J et al 1994 Hantavirus pulmonary syndrome: radiographic findings in 16 patients. Radiology 19: 66–68 34. Leung A N, Gosselin M V, Napper C H 1999 Pulmonary infections after bone marrow transplantation: clinical and radiographic findings. Radiology 210: 699–710 35. Franquet T, Lee K S, Muller N L 2003 Thin-section CT findings in 32 immunocompromised patients with cytomegalovirus pneumonia who do not have AIDS. Am J Roentgenol 181: 1059–1063 36. McAdams H P, Erasmus J, Winter J A 1995 Radiologic manifestations of pulmonary tuberculosis. Radiol Clin North Am 33: 655–678 37. Goodman P C 2005 Pulmonary tuberculosis. In: McAdams H P, Reddy G (eds) Cardiopulmonary imaging: categorical course syllabus. Leesberg, VA: American Roentgen Ray Society, 2005: 205–218. 38. Leung A N, Muller N L, Pineda P R, FitzGerald J M 1992 Primary tuberculosis in childhood: Radiographic manifestations. Radiology 182: 87–91 39. Lee K S, Im J G 1995 CT in adults with tuberculosis of the chest: characteristic findings and role in management. Am J Roentgenol 164: 1361–1367 40. Griffith D E 2002 Management of disease due to Mycobacterium kansasii. Clin Chest Med 23: 613–621 41. Erasmus J J, McAdams H P, Farrell M A et al 1999 Pulmonary nontuberculous mycobacterial infection: Radiologic manifestations. RadioGraphics 19: 1487–1503 42. Hartman T E, Swensen S J, Williams D E 1993 Mycobacterium avium– intracellulare complex: evaluation with CT. Radiology 187: 23–26 43. Hollings N P, Wells A U, Wilson R, Hansell D M 2002 Comparative appearances of non-tuberculous mycobacteria species: a CT study. Eur Radiol 12: 2211–2217 44. Patz E F Jr, Goodman P C 1992 Pulmonary cryptococcosis. J Thorac Imaging 7: 51–55 45. Himmel J E 2003 Cryptococcal pneumonia in an immunocompetent host: Radiographic findings. Semin Respir Infect 18: 129–131 46. Murayama S, Sakai S, Soeda H et al 2004 Pulmonary cryptococcosis in immunocompetent patients: HRCT characteristics. Clin Imaging 28: 191–195 47. Goldman M, Johnson P C, Sarosi G A 1999 Fungal pneumonias. The endemic mycoses. Clin Chest Med 20: 507–519 48. Gurney J W, Conces D J 1996 Pulmonary histoplasmosis. Radiology 199: 297–306 49. Batra P, Batra R S 1996 Thoracic coccidioidomycosis. Semin Roentgenol 31: 28–44 50. Kim K I, Leung A N, Flint J D et al 1998 Chronic pulmonary coccidioidomycosis: computed tomographic and pathologic findings in 18 patients. Can Assoc Radiol J 49: 401–407 51. Patel R G, Patel B, Petrini M F et al 1999 Clinical presentation, radiographic findings, and diagnostic methods of pulmonary

52. 53.

54.

55. 56.

57. 58. 59. 60.

61.

62.

63.

64. 65.

66.

67. 68.

69.

70.

71.

72. 73. 74. 75.

76.

•

PULMONARY INFECTION IN ADULTS

blastomycosis: a review of 100 consecutive cases. South Med J 92: 289–295 Gefter W B 1992 The spectrum of pulmonary aspergillosis. J Thorac Imaging 7: 56–74 Sonnet S, Buitrago-Tellez C H, Tamm M, Christen S, Steinbrich W 2005 Direct detection of angioinvasive pulmonary aspergillosis in immunosuppressed patients: preliminary results with high-resolution 16-MDCT angiography. Am J Roentgenol 184: 745–751 Kimura K, Stoopen M, Reeder M M et al 1997 Amebiasis: modern diagnostic imaging with pathological and clinical correlation. Semin Roentgenol 32: 250–275 Echeverri A, Long R F, Check W et al 1999 Pulmonary dirofilariasis. Ann Thorac Surg 67: 201–202 Im J G, Whang H Y, Kim W S et al 1992 Pleuropulmonary paragonimiasis: radiologic findings in 71 patients. Am J Roentgenol 159: 39–43 Jerray M, Benzarti M, Garrouche A et al 1992 Hydatid disease of the lungs. Study of 386 cases. Am Rev Respir Dis 146: 185–189 Kervancioglu R, Bayram M, Elbeyli L 1999 CT findings in pulmonary hydatid disease. Acta Radiologica 40: 510–514 Kokturk O, Ozturk C, Diren B et al 1999 “Air bubble”: a new diagnostic CT sign of perforated pulmonary hydatid cyst. Eur Radiol 9: 1321–1323 Boiselle P M, Crans C A Jr, Kaplan M A 1999 The changing face of Pneumocystis carinii pneumonia in AIDS patients. Am J Roentgenol 172: 1301–1309 Sandhu J, Goodman P C 1989 Pulmonary cysts associated with Pneumocystis carinii pneumonia in patients with the acquired immunodeficiency syndrome. Radiology 173: 33–35 Hidaldo A, Falco V, Mauleon S et al 2003 Accuracy of high-resolution CT in distinguishing between Pneumocystis carinii pneumonia and non-Pneumocystis carinii pneumonia in AIDS patients. Eur Radiol 13: 1179–1184 Crothers K, Huang L 2003 Recurrence of Pneumocystis carinii pneumonia in an HIV-infected patient: apparent selective immune reconstitution after initiation of antiretroviral therapy. HIV Med 4: 346–349 Goodman P C 1995 Tuberculosis and AIDS. Radiol Clin North Am 33: 707–717 Small P M, Hopewell P C, Schecter G F et al 1994 Evolution of chest radiographs in treated patients with pulmonary tuberculosis and HIV infection. J Thorac Imaging 9: 74–77 Fishman J E, Saraf-Lavi E, Narita M et al 2000 Pulmonary tuberculosis in AIDS patients: transient chest radiographic worsening after initiation of antiretroviral therapy. Am J Roentgenol 174: 43–49 Washington L, Miller W T Jr 1998 Mycobacterial infection in immunocompromised patients. J Thorac Imaging 13: 271–281 Laissy J P, Cadi M, Cinqualbre A et al 1997 Mycobacterium tuberculosis versus nontuberculous mycobacterial infection of the lung in AIDS patients: CT and HRCT patterns. J Comput Assist Tomogr 21: 312–317 Tenholder M F, Moser R J, Tellis C J 1988 Mycobacteria other than tuberculosis. Pulmonary involvement in patients with acquired immunodeficiency syndrome. Arch Intern Med 148: 953–955 Fishman J E, Schwartz D S, Sais G J 1997 Mycobacterium kansasii pulmonary infection in patients with AIDS: spectrum of chest radiographic findings. Radiology 204: 171–175 Feldman C, Glatthaar M, Morar R et al 1999 Bacteremic pneumococcal pneumonia in HIV-seropositive and HIV-seronegative adults. Chest 116: 107–114 Aviram G, Boiselle P M 2004 Imaging features of bacterial respiratory infections in AIDS. Curr Opin Pulm Med 10: 183–188 McGuinness G, Gruden J F, Bhalla M et al 1997 AIDS-related airway disease. Am J Roentgenol 168: 67–77 Wicky S, Cartei F, Mayor B et al. Radiological findings in nine AIDS patients with Rhodococcus equi pneumonia. Eur Radiol 6: 826–830 McGuinness G, Scholes J V, Garay S M et al 1994 Cytomegalovirus pneumonitis. Spectrum of parenchymal CT findings with pathologic correlation in 21 AIDS patients. Radiology 192: 451–459 Sider L, Westcott M A 1994 Pulmonary manifestations of cryptococcosis in patients with AIDS. CT features. J Thorac Imaging 9: 78–84

291

292

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

77. Conces D J Jr, Stockberger S M, Tarver R D et al 1993 Disseminated histoplasmosis in AIDS: Findings on chest radiographs. Am J Roentgenol 160: 15–19 78. Fish D, Ampel N, Galgiani J et al 1990 Coccidioidomycosis during human immunodeficiency virus infection: a review of 77 patients. Medicine 69: 384–391 79. Pompili G G, Alineri S, Arbori G et al 1998 Invasive aspergillosis in AIDS: findings with high-resolution computerized tomography. Radiol Medica 96: 325–330

80. Goodman P C, Schnapp L M 1992 Pulmonary toxoplasmosis in AIDS. Radiology 814: 791–793 81. Engels E A, Goedert J 2005 Human immunodeficiency virus/acquired immunodeficiency syndrome and cancer: past, present, and future. J Natl Cancer Inst 97: 407–409 82. Goodman P 1992 Kaposi’s sarcoma. J Thorac Imaging 6: 43–48

CHAPTER

Large Airway Disease and Chronic Airway Obstruction

16

Philippe Grenier

• • • • • • •

Tracheal disorders Bronchiectasis Broncholithiasis Emphysema Chronic bronchitis Asthma Obliterative (constrictive) bronchiolitis

This chapter reviews lesions involving the trachea and proximal bronchi, describes the radiological signs of bronchiectasis, and discusses the role of imaging in obstructive lung disease, a group of diffuse lung diseases associated with chronic airflow obstruction that includes chronic obstructive pulmonary disease (COPD), asthma and obliterative bronchiolitis. In obstructive lung disease, decreased expiratory flow may be related to loss of lung recoil or small airway obstruction or a combination of both. The pathological lesion that best correlates with loss of recoil is emphysema. The process that causes the small airway obstruction is inflammatory in nature and is characterized by thickening of all the layers of the bronchiolar walls, as well as an accumulation of mucus in the airway lumen (COPD and asthma) and/or an irreversible fibrosis (obliterative bronchiolitis).

TRACHEAL DISORDERS1–4 The trachea may be affected by a variety of extrinsic or intrinsic processes. Extrinsic processes, particularly masses, displace and distort the trachea, while intrinsic ones cause narrowing, widening, or a mass effect.Tracheal narrowing may affect a short or a long segment, and may extend to the mainstem bronchi. Tracheal disease initially missed on the chest radiograph is usually evident on careful evaluation of the frontal and lateral radiograph5. Computed tomography (CT) allows precise delineation of the intratracheal and extratracheal extent of the abnormality. Multidetector CT (MDCT), by combining helical volumetric CT acquisition and thin collimation during a single breath-

hold, provides an accurate assessment of the proximal airways, allowing multiplanar reformations and three-dimensional (3D) rendering of very high quality. Complementary CT acquisition at suspended or continuous expiration allows tracheal collapsibility to be assessed.

Post-traumatic strictures Strictures of the trachea are usually secondary to damage from a cuffed endotracheal or tracheostomy tube, or to external neck trauma.The lesions consist of granulation tissue followed by the development of dense mucosal and submucosal fibrosis associated with distortion of cartilage plates. The two principal sites of stenosis following intubation or the insertion of a tracheostomy tube are at the stoma or at the level of the endotracheal or tracheostomy tube balloon. On radiographs, the stenosis may be seen as a focus of circumferential or eccentric narrowing associated with a segment of increased soft tissue. The size of narrowing is usually clearly seen on CT.The narrowing is often concentric. Postintubation stenosis extends for several centimetres and typically involves the trachea above the level of the thoracic inlet. Post-tracheostomy stenosis typically begins 1–1.5 cm distal of the inferior margin of the tracheostomy stoma and involves 1.5–2.5 cm of tracheal wall. Multiplanar reformations are particularly helpful in defining accurately the site, length and degree of the stenosis. In selected cases, the degree of stenosis may also be defined by use of virtual bronchoscopy.

Infectious tracheobronchitis A number of infections, both acute and chronic, may affect the trachea and proximal bronchi, resulting in both focal and diffuse airway disease. Subsequent fibrosis may result in localized airway narrowing. The most common causes of infectious tracheobronchitis are bacterial tracheitis in immunocompromised patients, tuberculosis, rhinoscleroma (Klebsiella rhinoscleromatis) and necrotizing invasive aspergillosis. On CT, the extent of irregular and circumferential tracheobronchial narrowing is clearly demonstrated, and in some patients an accompanying mediastinitis (opacification of the mediastinal fat) is evident.

294

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

In active disease, the narrowed trachea and frequently one or other main bronchus have an irregularly thickened wall. In the fibrotic or healed phase, the trachea is narrowed but has a smooth wall of normal thickness.

Primary malignant neoplasms These are uncommon, accounting for less than 1% of all thoracic malignancies. The vast majority are squamous cell carcinoma and adenoid cystic carcinoma. Other neoplasms, such as mucoepidermoid carcinoma, carcinoid tumour, lymphoma, plasmacytoma and adenocarcinoma are rare. On CT, they appear as a soft tissue mass, most often in the posterior and lateral wall (Fig. 16.1A). Often sessile and eccentric, resulting in asymmetrical luminal narrowing, rarely they may appear circumferential. They can be polypoid and are mostly intraluminal with mediastinal extension in 30–40%. The surface of the tumour is often irregular in squamous cell carcinoma, whereas it is smooth in adenoid cystic carcinoma. Multiplanar reformation and volumetric rendering images are recommended for a precise pre-therapeutic assessment of tumour extent (Fig. 16.1B). The tumours are best treated surgically, especially with primary resection and re-anastomosis followed by radiation.

Secondary malignant neoplasms The large airways may be involved secondarily by malignant neoplasms as a result of either haematogeneous metastasis or direct invasion from the oesophagus, thyroid, mediastinum, or lung. Neoplasms that have a propensity to metastasize to the trachea and major bronchi include renal cell carcinoma and melanoma. On CT the abnormalities are usually focal and include intraluminal soft tissue nodules and wall thickening.

Benign neoplasms The most common benign neoplasms are hamartoma, leiomyoma, neurogenic tumour and lipoma. They are usually well demarcated, round and less than 2 cm in diameter. The radiological appearance typically consists of a smoothly marginated intraluminal polyp. Hamartomas and lipomas may demonstrate fat attenuation on CT. Tracheobronchial papillomatosis is a particular entity caused by human papillomavirus infection usually acquired at birth from an infected mother. The larynx is affected most commonly; extension into the trachea and proximal bronchi occurs occasionally. Exceptionally the infection spreads into the lung parenchyma. The typical radiological findings consist of multiple small nodules projecting into the airway lumen or diffuse nodular thickening of the airway wall. Although benign, papilloma may undergo transformation to squamous cell carcinoma.

Wegener’s granulomatosis Involvement of the large airways is a common manifestation of Wegener’s granulomatosis. Inflammatory lesions may be present with or without subglottic or bronchial stenosis, ulcerations and pseudotumours. Radiological manifestations include thickening of the subglottic region and proximal trachea with a smooth symmetrical or asymmetrical narrowing over a vari-

Figure 16.1 Adenoid cystic carcinoma of the trachea. (A) Axial CT at the level of the supra-aortic part of the mediastinum. Irregular stenosis of the tracheal lumen due to a soft tissue mass developing from the posterior and left lateral wall of the trachea. (B) 3D external volume rendering of the airways in a coronal view. The level of the tracheal lumen involvement (arrow) is accurately assessed with respect to the larynx and carina. The length and the degree of the stenosis are also clearly seen.

able length. Stenosis may also be seen on any main lobar or segmental bronchus. Nodular or polypoid lesions may also be seen on the inner contour of the airway lumen.

Relapsing polychondritis This is a rare systemic disease of autoimmune pathogenesis that affects cartilage at various sites, including the ears, nose, joints and tracheobronchial tree. Histologically, the acute inflammatory infiltrate present in the cartilage and perichondrial tissue induces progressive dissolution and fragmentation of the cartilage followed by fibrosis. Symmetrical subglottic stenosis is the most frequent manifestation in the chest (Fig. 16.2A). As the disease progresses, the distal trachea and bronchi may be involved. CT shows smooth thickening of the airway wall associated with more or less diffuse narrowing (Fig. 16.2B). In the early stage, the posterior wall of the trachea is spared

CHAPTER 16

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

Figure 16.2 Relapsing polychondritis. (A) PA chest radiograph targeted on the upper and mid parts of the mediastinum. The upper part of the tracheal lumen is narrowed (black arrows). The right paratracheal band is abnormally thickened (white arrows). (B) Axial CT at the level of the aortic arch showing abnormal thickening of the anterior and lateral walls of the trachea associated with calcium deposits (arrow). The posterior membranous wall of the trachea is unaffected.

but in advanced disease circumferential wall thickening occurs. The trachea may become flaccid with considerable collapse at expiration. Gross destruction of the cartilaginous rings with fibrosis may cause stenosis.

Amyloidosis Deposition of amyloid in the trachea and bronchi may be seen in association with systemic amyloidosis or as an isolated manifestation. As a result, the amyloid forms either multifocal or diffuse submucosal plaques or masses.The overlying mucosa is usually intact. Dystrophic calcification or ossification is frequently present. CT shows focal or, more commonly, diffuse thickening of the airway wall and narrowing of the lumen. Calcification may be seen. Narrowing of the proximal bronchi can lead to distal atelectasis, bronchiectasis, or obstructive pneumonia.

Tracheobronchopathia osteochondroplastica This rare disorder is characterized by the presence of multiple cartilaginous nodules and bony submucosal nodules on the inner surface of the trachea and proximal airways. Men are more frequently affected than women and most patients are older than 50. Histologically, the nodules contain heterotopic bone, cartilage and calcified acellular protein matrix. The overlying bronchial mucosa is normal and because it contains no cartilage, the posterior wall of the trachea is spared. The chest radiograph may be normal or may demonstrate lobar collapse or infective consolidation. If the tracheal air column is clearly seen, multiple sessile nodules that project into the tracheal lumen extending over a long segment of the trachea can be appreciated. On CT tracheal cartilage rings are thickened and shows irregular calcifications. The

nodules may protrude from the anterior and lateral walls into the lumen; they usually show foci of calcification.

Sabre-sheath trachea6 Characterized by a diffuse narrowing involving the intrathoracic trachea, this entity is almost always associated with COPD.The pathogenesis of the lesion is obscure, but probably it is an acquired deformity related to the abnormal pattern and magnitude of intrathoracic pressure changes in COPD. On radiographs and CT, the condition is easily recognized by noting that the internal side-to-side diameter of the trachea is halved or less than the corresponding sagittal diameter (Fig. 16.3A). On the postero-anterior (PA) radiograph and CT multiplanar reformations (Fig. 16.3B), the narrowing usually affects the whole intrathoracic trachea, with an abrupt return to normal calibre at the thoracic inlet. The trachea usually shows a smooth inner margin but occasionally has a nodular contour. Calcification of the tracheal cartilage rings is frequently evident.

Tracheobronchomegaly (Mounier–Kuhn disease) Tracheobronchomegaly refers to patients who have marked dilatation of the trachea and mainstem bronchi. It is often associated with tracheal diverticulosis, recurrent lower respiratory tract infection and bronchiectasis. Atrophy affects the elastic and muscular elements of both the cartilaginous and membranous parts of the trachea. The diagnosis is based on radiological findings. The immediately subglottic trachea has a normal diameter, but it expands as it passes to the carina and this dilatation often continues into the major bronchi. Atrophic mucosa prolapses between cartilage rings and gives the trachea a characteristically corrugated outline on a plain

295

296

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 16.3 Sabre-sheath trachea in a patient with chronic obstructive pulmonary disease. (A) Axial CT at the level of the upper lobes shows a significant reduction of the coronal diameter of the trachea. Bilateral centrilobular and paraseptal emphysematous areas are also present in the upper lobes. (B) Coronal oblique reformation along the long axis of the trachea. Reduction of the coronal diameter of the tracheal lumen (arrows). The upper part of the trachea above the thoracic inlet has a normal appearance.

radiograph. Corrugations may become exaggerated to form sacculations or diverticula. On CT a tracheal diameter of greater than 3 cm (measured 2 cm above the aortic arch) and diameters of 2.4 and 2.3 cm for the right and left bronchi, respectively, are diagnostic criteria (Fig. 16.4A). Additional findings include tracheal scalloping (Fig. 16.4B) and diverticula (especially along the posterior membranous tracheal wall).

Tracheobronchomalacia6 This abnormality results from weakened tracheal cartilage rings and is seen in association with a number of disorders, including tracheobronchomegaly, COPD, diffuse tracheal inflammation such as relapsing polychondritis, as well as following trauma. On radiographs, an almost 300% reduction in the sagittal diameter at expiration is an excellent

Figure 16.4 Tracheobronchomegaly in a patient with bilateral bronchiectasis and recurrent pulmonary infections. (A) Axial CT at the level of the upper part of the chest shows the dilated irregular lumen of the trachea. There is bilateral cylindrical and varicose bronchiectasis in the upper lobes. (B) 3D external volume rendering of the airways showing irregular dilatation of both trachea and main bronchi. Dilatation is also present in the lobar bronchi. In addition, the trachea is irregular in contour with multiple corrugations. (Reprinted from “Imagerie Thoracique de l’ Adulte” 3rd. ed., 2006, Flammarion, with permission).

diagnostic indicator. On CT, diagnosis is based on a narrowing of the diameter of the lumen by more than 50% on expiration compared with that on inspiration. The increase in compliance is due to the loss of integrity of the wall’s structural components and is particularly associated with damaged or destroyed cartilage. The coronal diameter of the trachea becomes significantly larger than the sagittal one, producing a lunate configuration. The flaccidity of the trachea or bronchi is usually most apparent during coughing or forced expiration. In patients with COPD with high downstream resistance particularly high dynamic pressure gradients can be generated across the tracheal wall, and it is likely that calibre changes of more than 50% can occur at expiration with normal tracheal compliance. As a result only a decrease in the cross-sectional area of the tracheal

CHAPTER 16

lumen of greater than 70% at expiration indicates tracheomalacia. Dynamic expiratory MDCT may offer a feasible alternative to bronchoscopy in patients with suspected tracheobronchomalacia7. Dynamic expiratory CT may show collapse of the more than 75% of the airway lumen; indeed, complete collapse may be seen (Fig. 16.5). Involvement of the central tracheobronchial tree may be diffuse or focal. The reduction of the airway may have an oval or crescentic shape.The crescentic form is due to the bowing of posterior membranous trachea.

Tracheobronchial fistula and dehiscence8 MDCT with thin collimation is the most accurate technique to identify peripheral bronchopleural fistulas, which are most commonly caused by necrotizing pneumonia or secondary to traumatic lesions. Nodobronchial and nodobronchoesophageal fistulas, which are most commonly caused by Mycobacterium tuberculosis infection, are characterized by the presence of gas in cavitated hila or mediastinal lymphadenopathy adjacent to the airways. Tracheal diverticula and tracheobronchoesophageal fistulas may also be diagnosed, even in adults. Malignant neoplasia, particularly oesophageal, are the most common cause of tracheo-oesophageal fistulas in adults. Occasionally congenital fistulas first manifest in adults. Infection and trauma are the most frequent nonmalignant causes. MDCT has a high degree of sensitivity and specificity for depicting bronchial dehiscence occurring after lung transplantation. Bronchial dehiscence is seen as a bronchial wall defect associated with extraluminal air collections.

BRONCHIECTASIS1,3,4 Bronchiectasis is a chronic condition characterized by local, irreversible dilatation of bronchi, usually associated with inflammation. In spite of its decreased prevalence in developed countries, bronchiectasis remains an important cause of haemoptysis and chronic sputum production. Although the causes of bronchiectasis are numerous, there are three mechanisms by which the dilatation can develop: bronchial obstruction, bronchial wall

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

damage, and parenchymal fibrosis (Table 16.1). In the first two mechanisms, the common factor is the combination of mucus plugging and bacterial colonization. Cytokines and enzymes released by inflammatory cells plus toxins from the bacteria result in a vicious cycle of increasing airway wall damage, mucous retention and bacterial proliferation. In parenchymal fibrosis, the dilatation of bronchi is caused by maturation and retraction of fibrous tissue located in the parenchyma adjacent to an airway (traction bronchiectasis). Pathologically, bronchiectasis has been classified into three subtypes, reflecting increasing severity of disease: cylindrical, characterized by relatively uniform airway dilatation; varicose, characterized by nonuniform and somewhat serpiginous dilatation; and cystic. As the extent and degree of airway dilatation increase, the lung parenchyma distal to the affected airway shows increasing collapse from fibrosis.

Radiographic findings Chest radiography reveals abnormalities in the majority of cases (Figs 16.6 and 16.7). Thickened bronchial walls are visible either as single thin lines or as parallel line opacities (tramlines). When seen end-on, bronchiectatic airways appear as poorly defined ring or curvilinear opacities (Fig. 16.6B). Dilated bronchi filled with mucous or pus result in tubular or ovoid opacities of variable size. Cystic bronchiectasis manifests as multiple thin-walled ring shadows often containing air–fluid levels (Fig. 16.7). Pulmonary vessels may appear increased in size and may be indistinct because of adjacent peribronchial inflammation fibrosis. In generalized bronchiectasis, such as that associated with cystic fibrosis, overinflation is often present (Fig. 16.7). Localized forms Table 16.1

MECHANISMS AND CAUSES OF BRONCHIECTASIS

• Bronchial obstruction Carcinoma Fibrous stricture (e.g. tuberculosis) Broncholithiasis Extensive compression (lymphadenopathy, neoplasm) • Parenchymal fibrosis (traction bronchiectasis) Tuberculosis Sarcoidosis Idiopathic pulmonary fibrosis • Bronchial wall injury Cystic fibrosis Childhood viral and bacterial infection Immunodeficiency disorders Dyskinetic cilia syndrome Allergic bronchopulmonary aspergillosis Lung and bone marrow transplantation Panbronchiolitis Systemic disorders (rheumatoid arthritis, Sjögren’s syndrome, inflammatory bowel disease, yellow nail syndrome)

Figure 16.5 Tracheobronchomalacia. Axial CT acquired during dynamic expiratory manœuvre. The collapse of the tracheal lumen is almost complete. The tracheal lumen is crescentic in shape because of the bowing of the posterior membranous trachea.

α1-Antitrypsin syndrome Congenital Williams Campbell syndrome

297

298

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

cardinal sign of bronchiectasis) (Fig. 16.8), internal bronchial diameter greater than that of the adjacent pulmonary artery (signet ring sign) (Fig. 16.9), visualization of bronchi within 1 cm of the costal pleura or abutting the mediastinal pleura, and mucus-filled dilated bronchi (see Fig. 16.4A). In varicose bronchiectasis, the bronchial lumen assumes a beaded configuration (Fig. 16.8). Cystic bronchiectasis is seen as a string of cysts caused by sectioning irregular dilated bronchi along their

Figure 16.7 Cystic fibrosis. The PA radiograph shows slight overinflation and the presence of multiple thin-walled ring shadows in the right lung and the upper part of the left lung, reflecting cystic bronchiectasis. Some ring shadows contain air–fluid levels.

Figure 16.6 Bronchiectasis and obliterative bronchiolitis. (A) PA chest radiograph shows oligaemia in the lung bases with pulmonary blood flow redistribution in the upper parts of the lungs and slight overinflation of the lungs, more marked on the right side. (B) Targeted image of the right lung base in the same patient shows tramlines and ring opacities reflecting the presence of dilated and thick-walled bronchi.

are frequently accompanied by atelectasis which may be mild and detected only because of vascular crowding, fissure displacement, or obscuration of part of the diaphragm.

Computed tomography findings The major sign of bronchiectasis on thin-collimation CT (highresolution CT [HRCT]) is dilatation of the bronchi, with or without bronchial wall thickening. Bronchial dilatation on CT is often manifested by lack of tapering of bronchial lumina (the

Figure 16.8 Allergic bronchopulmonary aspergillosis. HRCT shows cylindrical and varicose bronchiectasis in the right upper lobe and the apicoposterior segment of the left upper lobe. Small centrilobular nodules representing infectious or inflammatory bronchiolitis are seen in the anterior segment of the right upper lobe. An area of parenchymal consolidation surrounded by a halo of ground-glass opacity is present within the superior segment of the left upper lobe (arrow). (Reprinted from “Imagerie Thoracique de l’ Adulte, 3rd. ed., 2006, Flammarion, with permission).

CHAPTER 16

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

Figure 16.9 Cystic fibrosis. HRCT image in the upper lobes shows bilateral bronchiectasis and thickening of the bronchial walls.

lengths, or a cluster of cysts, caused by multiple dilated bronchi lying adjacent to each other (Fig. 16.10). Clusters of cysts are most frequently seen in atelectatic lobes. Air–fluid levels, caused by retained secretions, may be present in the dependent portion of the dilated bronchi. Secretion accumulation within bronchiectatic airways is generally easily recognizable as lobulated gloved finger, V- or Y-shaped densities (Figs 16.11 and 16.12). When oriented perpendicular to the data acquisition plane the filled dilated bronchi are visualized as nodular opacities and recognized by observation of the homologous pulmonary arteries, whose diameters are smaller than those of the dilated filled bronchi. CT may show a completely collapsed lobe containing bronchiectatic airways. Subtle degrees of volume loss may be seen in lobes in relatively early disease.This is most evident in the lower lobes on the basis of crowding of the mildly dilated bronchi and posterior displacement of the oblique fissure. Associated CT findings of bronchiolitis are seen in about 70% of patients with bronchiectasis. Small centrilobular nodu-

Figure 16.10 Post-infectious cystic bronchiectasis. Presence of the clusters of cysts abutting the mediastinum and located within the left lower lobe and the inferior segment of the lingula. There is volume loss of the left lower lobe.

Figure 16.11 Allergic bronchopulmonary aspergillosis. HRCT of the upper lobes. Mucoid impactions are present within segmental and subsegmental dilated bronchi in the upper lobes. Small centrilobular linear branching opacities are seen in the periphery of the right upper lobe.

lar and linear branching opacities (tree-in-bud sign) express inflammatory and infectious bronchiolitis (Fig. 16.12). Areas of decreased attenuation and vascularity, mosaic perfusion pattern and expiratory air trapping reflect the extent of obliterative bronchiolitis (Fig. 16.13). These abnormalities are very common in patients with severe bronchiectasis and can even precede the development of bronchiectasis. The obstructive defect found at pulmonary tests in patients with bronchiectasis seems not to be related to the degree of collapse of large airways on expiratory CT or the extent of mucous plugging

Figure 16.12 Bronchiectasis with mucoid impactions and infectious bronchiolitis. Oblique reformat (4-mm thick slab) of the right lung with maximum intensity projection. Presence of mucoid impactions (arrows) in the dilated bronchi located within the anterior segment of the right upper lobe. Presence of multiple small centrilobular nodular and linear branching opacities (tree-in-bud sign).

299

300

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Cystic fibrosis1,4

Figure 16.13 Bronchiectasis and obliterative bronchiolitis. HRCT performed at full expiration shows air trapping in the right middle lobe, the superior segment of the left lower lobe, and some lobules of the left upper and right lower lobes. There is a cluster of cysts (cystic bronchiectasis) in the right middle lobe.

of the airway, but the consequence of an obstructive involvement of the peripheral airways (obliterative bronchiolitis)9. The extent of small airway disease commonly evident on CT (decreased lung attenuation, expiratory air trapping) in patients with bronchiectasis has proven to be the major determinant of airflow obstruction.

Accuracy of CT HRCT has replaced bronchography in the diagnosis and assessment of the extent of bronchiectasis10. By combining helical volumetric CT acquisition and thin collimation, CT has gained greater advantages by circumventing the limitations of HRCT, particularly the risk of missing bronchiectasis strictly localized within the intervals between slices11. Currently, MDCT with thin collimation is the technique of choice for the detection and the assessment of the extent of bronchiectasis. Multiplanar reformations increase the detection rate and the reader’s confidence as to the distribution of bronchiectasis, and improves agreement between observers as to the diagnosis12. In addition, maximum intensity projections improve the detection and display of both mucoid impactions and small centrilobular and linear branching opacities (tree-in-bud sign), characteristic of infectious bronchiolitis. The reliability of CT for determining the causes of bronchiectasis is somewhat controversial. An underlying cause for bronchiectasis is found in fewer than half of patients and CT features alone do not usually allow a confident distinction between idiopathic bronchiectasis and bronchiectasis with a specific cause4. Bilateral upper lobe distribution is most common in patients with cystic fibrosis and allergic bronchopulmonary aspergillosis; unilateral upper lobe distribution is most common in patients with tuberculosis; and a lower lobe distribution is most often seen after childhood viral infections. However, CT remains of little value in diagnosing specific aetiologies of bronchiectasis.

Cystic fibrosis results from an autosomal recessive genetic defect in the structure of the cystic fibrosis transmembrane regulation protein which leads to abnormal chloride transport across epithelial membranes. Although the mechanisms by which this defect leads to lung disease are not entirely understood, an abnormally low water content of airway mucus is at least partially responsible for decreased mucus clearance, mucous plugging of airways and an increased incidence of bacterial airway infection. Bronchial wall inflammation progressing to secondary bronchiectasis is always present in patients with long-standing disease. In patients with early or mild disease, the findings on chest radiography may be subtle. Hyperinflation reflects the presence of obstruction of the small airways (see Fig. 16.7). Thickening of the wall of the upper lobar bronchi can also be seen on the lateral radiograph. In more advanced disease, the radiographs can be diagnostic, showing increased lung volume, accentuated linear opacities in the upper lung areas, resulting from bronchial wall thickening or bronchiectasis, proximal bronchiectasis and mucoid impaction. Additional findings include cystic regions of the upper lobes (see Fig. 16.7), representing cystic bronchiectasis, healed abscess, cavities, or bullae; and atelectasis, findings of pulmonary hypertension or cor pulmonale, pneumothorax or pleural effusion. The chest radiographs are sufficient for clinical management, but it is important to know that usually there is a little visible radiographic change associated with clinical exacerbation. Several studies have shown that CT can offer an alternative to routine radiographic and clinical methods for monitoring disease status and progression as well as for assessing response to treatment. These studies consistently document close correlation between HRCT findings and both clinical and pulmonary functional evaluation of these patients. On CT, peripheral and/or central bronchiectasis is present in all patients with advanced cystic fibrosis (see Fig. 16.9). All lobes are typically involved, although early in the disease abnormalities are often predominantly distributed in the upper lobes, and sometimes with a right upper lobe predominance. Bronchial wall and/or peribronchial interstitial thickening is also commonly present. It is generally more evident than bronchial dilatation in patients with early disease. Mucous plugging is present in 25–50% of patients, and may be seen in all lobes. Collapse or consolidation is visible in up to 80% of patients. Lobar volume loss is often present in patients with advanced disease. Bullae may be difficult to distinguish from cystic bronchiectasis, particularly in fibrotic upper lobes. Abscesses may be difficult to distinguish from cystic bronchiectasis, particularly as both may contain air–fluid levels. Pleural thickening, which is often apparent on chest radiographs, is demonstrated to advantage by CT. Small centrilobular nodular and branching linear opacities (tree-in-bud sign) can be an early sign of disease.They reflect the presence of mucous impactions in dilated bronchioles associated with peribronchiolar inflammation. Focal areas of decreased lung attenuation are frequently present, representing air trapping and mosaic perfusion due to obstruction of the small airways. These areas often correspond to lobules and subsegments where mucous plugging in dilated airways is present.

CHAPTER 16

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

At an early stage of the disease, HRCT can demonstrate airway abnormalities in patients who are asymptomatic and have normal pulmonary function and a normal chest radiograph. In patients with more advanced disease, HRCT is superior to chest radiography in detecting bronchiectasis and mucous plugging.

dilection. Despite this upper lobar shrinkage, the lung volume is frequently increased, reflecting overinflation in the lower lobes due to obstruction of the small airways and the presence of bullae in cavitation in the upper lobes.

Allergic bronchopulmonary aspergillosis1, 4

The dyskinetic cilia syndrome results from an autosomal recessive genetic abnormality and is characterized by abnormal ciliary structure and function, leading to a reduced mucociliary clearance and chronic airway infection. Bronchiectasis and sinusitis are common manifestations. About half of patients have also situs inversus. The combination of bronchiectasis, sinusitis and situs inversus is termed Kartagener’s syndrome. Men and women are equally affected, but in men the syndrome may be associated with immotile spermatozoa and infertility. Respiratory symptoms can generally be traced back to childhood. Bronchiectasis develops in childhood and adolescence and is associated with recurrent pneumonia. Both plain radiographs and CT typically show bilateral bronchiectasis with a basal (lower or middle lobe) predominance, similar to that seen in patients with other causes of postinfectious bronchiolitis. Cylindrical bronchiectasis is the most common type and a diffuse bronchiolitis may be present.

Allergic bronchopulmonary aspergillosis (ABPA) is a hypersensitivity reaction to aspergillus and is characterized by asthma, blood, eosinophilia, radiographic pulmonary opacities and evidence of allergy to antigens of aspergillus species. It may also occur in patients with cystic fibrosis. Recurrent acute episodes cause progressive lung damage that can be controlled with steroids. The radiological features can be classified as acute and transient, or chronic and permanent. The most common acute changes are transient consolidation, mucoid impaction and atelectasis. Consolidation ranges from massive and homogeneous to lobar or segmental in configuration, or to subsegmental or smaller. When consolidation clears, it often leaves residual bronchiectasis, which creates a favourable environment for fungal recolonization, a finding that accounts for the fact consolidation often recurs in the same area. Mucoid impaction obstructs the airway lumen which becomes distended by retained secretions. At the same time, lung parenchyma remains aerated by collateral drift, permitting the visualization of the impacted airway. Bronchoceles appear as opacities of a variety of shapes (linear, branching or non-branching, band-like opacities that point to the hilum, toothpaste opacities, V- and Y-shaped opacities, gloved finger opacities). These opacities disappear once their airway contents have been coughed up, leaving ring or parallel linear opacities. Atelectasis is subsegmental, segmental or lobar and has a tendency to recur in the same area. Permanent changes indicate irreversible lung damage and are the clue that an asthmatic patient has ABPA when he/ she is in remission. Bronchiectasis is responsible for most of the permanent radiological changes. It affects lobar bronchi and the first- and second-order segmental bronchi. Beyond the proximal bronchi, more distal airways remain normal and patent, though small airway abnormalities are present on HRCT. These abnormalities include the tree-in-bud appearance reflecting mucoid impaction in dilated bronchioles, focal areas of decreased lung attenuation, and air trapping reflecting obstruction of the small airways. Compared with other bronchiectatic diseases, bronchiectasis in ABPA is more commonly widespread in central location, and more likely to contain cystic or varicose components. Mucus plugs within the ectatic airways are frequently seen on HRCT (see Fig. 16.11). High attenuation within the plugs is also relatively frequent, reflecting the presence of calcium concentration by the fungus. Hyperattenuated mucous plugs may be depicted within the areas of consolidation. Parenchymal scarring represents the fibrotic stage of the disease. It commonly follows bronchiectasis, and manifests as linear opacities and lobar shrinkage. Mirroring the distribution of bronchiectasis, these features have a strong upper zone pre-

Dyskinetic cilia syndrome1,4

BRONCHOLITHIASIS3,8 Broncholithiasis is a condition in which peribronchial calcified nodal disease erodes into or distorts an adjacent bronchus. The underlying abnormality is usually granulomatous lymphadenitis caused by Mycobacterium tuberculosis or fungi such as Histoplasma capsulatum. A few cases have been reported with silicosis. Calcified material in a bronchial lumen or bronchial distortion by peribronchial disease results in airway obstruction. This leads to collapse, obstructive pneumonitis, mucoid impaction, or bronchiectasis. Symptoms include cough, haemoptysis, recurrent episodes of fever and purulent sputum. Broncholithiasis is more common on the right, and obstructive changes particularly affect the right middle lobe. On chest radiographs, three major types of changes may be seen: • disappearance of a previously identified calcified nidus • change in position of a calcified nidus • evidence of airway obstruction, including segmental or lobar atelectasis, mucoid impaction, obstructive pneumonitis, obstructive oligaemia with air trapping. Calcified hilar or mediastinal nodes are a key feature. CT and fibre-optic bronchoscopy complement each other in this condition. Broncholithiasis is recognized on CT by the presence of a calcified endobronchial or peribronchial lymph node, associated with a bronchopulmonary complication caused by obstruction (including atelectasis, pneumonia, bronchiectasis and air trapping), in the absence of an associated soft tissue mass.

301

302

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

EMPHYSEMA1,3,6,13 Emphysema is defined as a condition of the lung characterized by permanent, abnormal enlargement of airspaces distal to the terminal bronchioles, accompanied by the destruction of their walls without obvious fibrosis. The most important aetiological factor by far is cigarette smoking. Other inhaled pollutants have also been implicated, including gases such as nitrogen oxides and phosphogenes, as well as particulate smoke. There is also a causal relationship between HIV infection and the development of early emphysema. Various genetic disorders associated with emphysema include α1-antitrypsin deficiency, heritable diseases of connective tissue such as cutix laxa, Marfan syndrome and familial emphysema. Emphysema is thought to result from the destruction of elastic fibres caused by an imbalance between proteases and protease inhibitors in the lung and from the mechanical stresses of ventilation and coughing. Proteases are normally released in low concentration by phagocytes in the lung. Protease inhibitors, mainly α1-protease inhibitor (α1-antitrypsin), prevent them from causing structural damage to the lung. Imbalance in the protease–antiprotease activity may result from antiprotease deficiency (α1-antitrypsin deficiency) from excess release of protease stimulated by environmental agents, or from the defective repair of protease-induced damage. Tobacco smoke increases the number of pulmonary macrophages and neutrophils, reduces antiprotease activity, and may impair the synthesis of elastin. As emphysema develops lung destruction progresses, airspaces enlarge and elastic recoil declines, reducing radial traction on bronchial walls and on blood vessels, allowing airways and vessels to collapse.

Pathological classification Classification of emphysema is traditionally based on the microscopic localization of disease within the secondary pulmonary lobule. The principal types are centrilobular, panlobular, paraseptal and irregular emphysema. Centrilobular (centriacinar) emphysema affects mainly the proximal respiratory bronchioles and alveoli in the central part of the acinus. The process tends to be most developed in upper parts of the lungs. It is strongly associated with cigarette smoking. Inflammatory changes in the small airways are common with plugging, mural infiltration and fibrosis, leading to stenosis, distortion and destruction. Paraseptal emphysema selectively involves the alveoli adjacent to the connective tissues of septa and bronchovascular bundles, particularly at the margins of the acinus and lobule but also subpleurally and adjacent to the bronchovascular bundles. Airspaces in paraseptal emphysema may become confluent and develop into bullae, which may be large. Airway obstruction and physiological disturbance may be minor. Panlobular (panacinar) emphysema is characterized by a dilatation of the airspaces of the entire acinus and lobule. With progressive destruction, all that eventually remains are thin strands of deranged tissue surrounding blood vessels. It is the most widespread and severe type of emphysema. Pathological changes are distributed throughout the lungs, but they

are often basely predominant. Panlobular emphysema is the type of emphysema that occurs in α1-antitrypsin deficiency and in familial cases. Irregular (or paracicatricial) emphysema refers to irregular airspace enlargement, and occurs in patients with pulmonary fibrosis. It is commonly seen adjacent to localized parenchymal scars, in diffuse pulmonary fibrosis, and in pneumoconiosis, particularly progressive massive fibrosis.

Radiographic findings The main radiographic manifestations of emphysema are overinflation and alterations in the lung vessels. Signs of overinflation are the best predictors of the presence and severity of emphysema. Signs of overinflation include the height of the right lung being greater than 29.9 cm, location of the right hemidiaphragm at or below the anterior aspect of the seventh rib, flattening of the hemidiaphragm, enlargement of the retrosternal space, widening of the sternodiaphragmatic angle and narrowing of the transverse cardiac diameter (Fig. 16.14). Alterations in lung vessels include arterial depletion, whereas vessels of normal, or occasionally increased, calibre are present in unaffected areas of the lung, absence or displacement of vessels caused by bullae, widened branching angles with loss of side branches and vascular redistribution. With the development of cor pulmonale, or left heart failure, the radiographic appearance will alter and may become less obviously abnormal. The heart may then appear to be normal in size, or sometimes enlarged, the diaphragm becomes less flat and the pulmonary vessels less attenuated. Bullae may be as small 1 cm in diameter or may occupy the whole hemithorax causing marked relaxation collapse of the adjacent lung. Bullae caused by paraseptal emphysema are much more common in the upper zones, but when they are associated with widespread panlobular emphysema, the distribution is much more even. Occasionally the wall is completely absent and in such cases bullae can be difficult to detect. Plain radiographs markedly underestimate the number of bullae. The presence of emphysema associated with large bullae is referred to as bullous emphysema. An entity mainly seen in young men, characterized by the presence of large progressive upper lobe bullae which occupy a significant volume of a hemithorax and are often asymmetrical, is referred to as giant bullous emphysema, vanishing lung syndrome or primary bullous disease of the lung. Large bullae may be seen as avascular transradiant areas usually separated from the remaining lung parenchyma by a thin curvilinear wall (Fig. 16.15). They can cause marked relaxation collapse of the adjacent lung and can even extend into the opposite hemithorax, particularly by way of the anterior junctional area. Spontaneous pneumothorax commonly occurs in association with localized areas of emphysema or bullae affecting the lung apices. Bullae may enlarge progressively over months or years; a period of stability may be followed by a sudden expansion. Bullae may also disappear, either spontaneously or following infection or haemorrhage. The main complications of bullae include pneumothorax, infection and haemorrhage. In case of

CHAPTER 16

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

Figure 16.15 Giant bullous emphysema. The PA chest radiograph shows large avascular transradiant areas in the upper and lower parts of the right lung. The bullae are marginated with thin curvilinear opacities.

Computed tomography findings CT, particularly HRCT, is the most accurate means of detecting emphysema and determining its type and extent in vivo. On HRCT, emphysema is characterized by the presence of areas of abnormally low attenuation which can be easily contrasted with surrounding normal lung parenchyma if sufficiently low window values (−800 to −1000 HU) are used. Focal areas of emphysema usually lack distinct walls as opposed to lung cysts. In many patients, it is possible to classify the type of emphysema on the basis of its HRCT appearance, although the different types, as well as bullae, may be present in association in the same patient.

Centrilobular emphysema Figure 16.14 Severe diffuse emphysema. (A) PA and (B) lateral chest radiographs. The diaphragm is displaced downwards and appears flattened. On the PA radiograph (A) the transverse cardiac diameter is reduced. The diaphragm appears irregular in contour due to an abnormal visibility of diaphragmatic insertions on the ribs. Note the depression of vessels in the periphery of the lungs. On the lateral radiograph (B) there is a widening of the sternodiaphragm angle and an increase of dimensions of the retrosternal transradiant area.

infection or haemorrhage, bullae contain fluid and develop an air–fluid level.When a bulla becomes infected the hairline wall becomes thickened and may mimic a lung abscess. Carcinoma arising in or adjacent to bullae should be suspected in case of mural nodule, mural thickening, a change in diameter of the bulla, pneumothorax and the accumulation of fluid within the bulla.

Centrilobular emphysema predominantly affects the central portion of the lobule. On HRCT it is characterized by the presence of multiple, small round areas of abnormally low attenuation, distributed throughout the lungs but commonly having an upper lobe predominance (Fig. 16.16)14.The emphysematous spaces often appear to be grouped near the centre of secondary lobules surrounding the centrilobular arteries. Even when the centrilobular location of low attenuation areas is not recognized on HRCT, the presence of multiple, small areas of emphysema scattered throughout the lung is diagnostic of centrilobular emphysema. As the emphysema becomes more severe the areas of low attenuation become confluent and the centrilobular distribution becomes less apparent. In most cases, the areas of low attenuation have no visible walls. However, sometimes very thin walls may be seen when the areas of emphysema are extensive. These apparent walls probably represent atelectasis or interlobular septa adjacent to the emphysematous spaces.

303

304

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Occasionally the remaining interlobular septa may appear particularly prominent on the chest radiograph and in this way mimic lymphangitic carcinomatosis. HRCT of course will readily identify the true cause.

Panlobular emphysema Panlobular emphysema is characterized by widespread areas

of abnormally low attenuation, representing the uniform destruction of the pulmonary lobule. Pulmonary vessels in the affected lung appear fewer and smaller than normal (Fig. 16.18). Panlobular emphysema is almost always most severe in the lower lobes, where long lines may be present reflecting the presence of fibrosis within the remaining interlobular septa (Fig. 16.18). The characteristic appearances of extensive lung destruction and the associated paucity of vascular markings are easily recognized. On the other hand, mild and even moderately severe panlobular emphysema can be very subtle and difficult to detect radiologically15. Panlobular emphysema, secondary to α1-antitrypsin deficiency, is frequently associated with bronchiectasis. Figure 16.16 Centrilobular emphysema. HRCT of the right lung shows multiple small round areas of low attenuation that are distributed through the lungs, mainly around the centrilobular arteries (arrows).

Irregular emphysema

Paraseptal emphysema

mally low attenuation associated with features of fibrosis. It may be associated with diffuse pulmonary fibrosis or progressive massive fibrosis.

Paraseptal emphysema is characterized on HRCT by areas of low attenuation visible in the subpleural areas, along the peripheral or mediastinal pleura, mainly in the upper lobe and along the fissures (Fig. 16.17). The emphysematous spaces often have very thin but visible walls, mostly corresponding to interlobular septa thickened by associated fibrosis. Subpleural bullae are a frequently associated finding. They are commonly found in the azygoesophageal recess, adjacent to the superior mediastinal border and along the anterior junctional region. Because of its location adjacent to structures with soft-tissue attenuation, even mild paraseptal emphysema is easily detected on HRCT.

Figure 16.17 Paraseptal emphysema. HRCT of the right upper lobe shows multiple small areas of low attenuation distributed along the peripheral and mediastinal pleura (arrows).

Irregular emphysema is recognized on HRCT as areas of abnor-

Bullae Bullae are seen as avascular, low-attenuation areas that are larger than 1 cm in diameter and that can have a thin but perceptible wall. In most patients the parenchymal abnormalities are not visible on the chest radiograph. CT is more sensitive than the chest radiograph in demonstrating bullae and allows accurate assessment of their number, size, and position. CT is particularly useful when there is difficulty in distinguishing bullous disease from pneumothorax (Fig. 16.19A)16. Inspiratory and expiratory CT images indicate the extent to which a bulla is ventilated and the appearance of the rest of the lung helps in assessing the extent and degree of diffuse lung disease. This

Figure 16.18 Panlobular emphysema in a patient with α1-antitrypsin deficiency. HRCT of the lung bases showing the presence of large areas of decreased lung attenuation with a paucity of pulmonary vessels, more marked in the lower lobes. Long lines are visible within the remaining parenchyma of lung bases.

CHAPTER 16

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

can be highlighted on a CT image (density mask technique) and expressed as a percentage of the total pixels included in the lung section (Fig. 16.19B)18. Gevenois et al have shown that a threshold of −950 HU provides an accurate estimate of both macroscopic and, to a slightly lesser degree, microscopic emphysema19,20. However, this technique is very sensitive to technical factors. In particular, reducing the milliamperage may affect the measured extent of low attenuated lung. With MDCT, it has become feasible to apply density thresholding technique to volumetric data. Lung volume and volume of emphysema are quantified and displayed at a regional level, for a better preoperative assessment of lung disease in patients with severe emphysema who are candidates for bullectomy, lung volume reduction surgery, or lung transplantation. The use of the density mask method on both expiratory and inspiratory images has been used as a means of distinguishing areas of simple hyperinflation without tissue destruction from areas of emphysema. Expiratory CT does not correlate as well as inspiratory HRCT with the morphological extent of emphysema, but the expiratory HRCT is superior to inspiratory HRCT in reflecting functional air trapping. This technique has shown good correlations with indices of airflow obstruction and air trapping, particularly when a threshold of −900 HU is used.

CHRONIC BRONCHITIS1,6,13

Figure 16.19 Giant bullous emphysema. (A) Coronal reformation after MDCT thin collimation acquisition. Presence of large confluent bullae within the right lung associated with destruction of the right upper lobe. Presence of paraseptal emphysematous bullae within the left upper lobe along the mediastinum. (B) The same coronal reformation as in (A) after applying the density mask technique (−950 HU), making it possible to segment automatically the areas of emphysema before quantitative assessment.

ability makes CT useful for identifying patients suitable for treatment with bullectomy.

Assessment of extent of emphysema with computed tomography The distribution and severity of emphysema may be quantitated by CT. CT can be assessed by subjective visual methods, density measurements or by post-processing and texture analysis. Studies using various CT section thicknesses have shown good correlation between macroscopic pathological scores and visually assessed CT scores. However, intraand inter-observer variation in such subjective approaches is low to moderate. Objective methods using CT densitometry provides better correlation with a morphological reference17. The density of emphysematous areas is abnormally low. If a histogram plot is made of frequency against pixel density (HU), the emphysematous curve is shifted to the left compared with normal. Pixels with values below a certain number

Chronic bronchitis is defined as a clinical disorder characterized by excessive mucus secretion by the bronchial tree, manifested by chronic or recurring productive cough on most days in more than 3 months of each of 2 successive years. The pathogenesis of chronic bronchitis is related to cigarette smoking, air pollution and infection. The histological abnormalities present in chronic bronchitis include bronchial submucosal hyperplasia, smooth muscle hypertrophy, chronic inflammation and the obstruction of small airways. On pulmonary function tests, a patient with pure chronic bronchitis has normal total lung capacity and normal elastic recoil, but reduced expiratory flow and elevated residual volume. Airflow obstruction, which is concentrated in the small bronchioles, has both reversible (mucous plugging, inflammation, smooth muscle hypertrophy) and irreversible components (fibrosis and stenosis).

Radiographic findings The majority of patients with symptoms of chronic bronchitis have a normal chest radiograph. When radiographic abnormalities are present, they can include hyperinflation, oligaemia, bronchial wall thickening and accentuation of linear lung markings. Hyperinflation and oligaemia (sparse and attenuated lung vessels) can occur in patients with chronic bronchitis in the absence of emphysema, as a result of obstruction of the small airways (Fig. 16.20). Thickening of the bronchial walls leads to tubular and ring shadows. Increased lung markings cause the appearance of a ‘dirty chest’, a term widely used for describing a loss in clarity of the lung vessels (Fig. 16.20). In spite of these findings, it is

305

306

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 16.20 Chronic bronchitis and obstructive lung disease. PA chest radiograph shows mild overinflation. A ring shadow is visible above the left hilum (arrow) reflecting bronchial wall thickening. There is also accentuation of linear markings in the right lung base.

widely admitted that the chest radiograph has little to offer in the detection or exclusion of chronic bronchitis. Sabre-sheath trachea may be present. Cor pulmonale is a recognized complication which is seen almost exclusively in hypoxic patients. With the onset of heart failure the heart and hilar and intermediate lung vessels become enlarged. Enlargement of vessels is present in all zones and affects particularly segmental vessels and a few divisions beyond.

Computed tomography findings On CT, bronchial wall thickening is present. Using thin-collimation MDCT acquisition and multiplanar reformations associated with minimum intensity projections, air-filled outpouchings or diverticula are seen in addition to the lumen of the main lobar or segmental bronchi. These abnormalities reflect the enlargement of mucous glands and are related to low or subepithelial connective tissue and herniation of airway mucosa between small muscle bundles. Because of the deficit of bronchial cartilage in patients with COPD, prominent collapse of airway lumen may occur with maximum force expiratory manœuvre. On expiratory CT, the lumen of segmental and subsegmental bronchi (mainly in the lower lobes) may collapse excessively, particularly in the lower lobes where the cartilage deficiency is most apparent. Airway remodelling occurs in COPD patients, but this abnormality involves essentially the small airways. As there is a significant association between the dimension of the small and large airways in COPD patients, measuring airway dimension in the larger bronchi can provide an estimate of small airway remodelling. It is likely that the same pathophysiological process that causes small airway obstruction also

takes place in larger airways where its functional effects are smaller. By using CT to assess the extent of emphysema and measure airway wall area in a cohort of COPD patients and asymptomatic smokers, some investigators have shown that individual COPD patients may have emphysema or airway wall remodelling as their predominant phenotypes21.The ability to separate airway-predominant from parenchymal-predominant pathology in COPD may prove useful in applying specific therapies designed to prevent or ameliorate the airway remodelling or parenchymal destruction. In the lung parenchyma, thin-section CT has proven its ability in demonstrating the presence of small airway abnormalities and centrilobular emphysema in asymptomatic smokers, before the development of an obstructive lung disease (Fig. 16.21). In a study of healthy adult volunteers, 20–25% of the smokers showed multiple areas of ground-glass attenuation and small nodules (Fig. 16.21)22. In another CT–pathology correlation study in heavy smokers, the areas of ground-glass attenuation corresponded to histological findings of respiratory bronchiolitis and the small centrilobular nodules corresponded to bronchiolectasis with peribronchiolar fibrosis23. The introduction of expiratory thin-section CT has demonstrated that air trapping observed in healthy volunteers is related to smoking and can be observed before pulmonary function deteriorates. In patients with COPD, the extent of lung hypoattenuation at expiration probably reflects air trapping more than reduction of the alveolar wall surface. However, expiratory air trapping in these patients may be the result of either airway obstruction caused by a loss of alveolar attachment to the airways, directly related to emphysema, or of intrinsic bronchial or bronchial abnormalities associated with cigarette smoking.

ASTHMA1,8 Asthma is a chronic inflammatory condition involving the airways. This inflammation causes a generalized increase in existing bronchial hypersensitivity to a variety of stimuli. This is commonly used in practice to confirm the clinical diagnosis of asthma. In susceptible individuals, this inflammation induces recurrent episodes of wheezing, chest tightness, breathlessness

Figure 16.21 Respiratory bronchiolitis and centrilobular emphysema in a heavy smoker. HRCT of the upper lobes shows patches of groundglass opacity associated with areas of centrilobular emphysema.

CHAPTER 16

and coughing, usually associated with widespread but variable airflow obstruction that is often reversible either spontaneously or with treatment. The chronic inflammation process leads to structure changes, such as new vessel formation, airway smooth muscle thickening and fibrosis, which may result in irreversible airway narrowing.

Radiographic findings Chest radiography is usually recommended in all asthmatic patients who are ill enough to justify admission to a hospital. Hyperinflation may be seen in both relapse and remission.The prevalence of hyperinflation is generally higher in children and in patients needing hospital admission. While hyperinflation is often transient, it may be a permanent change. Bronchial wall thickening is more frequent in children, but in adults when it becomes visible it is usually an irreversible phenomenon. The walls of end-on segmental airways become thickened and the normally invasive airways parallel to the radiographs appear as parallel or single line opacities. It may be present in up twothirds of patients. Chest radiography may depict complications including consolidation, atelectasis, mucoid impaction, pneumothorax and pneumomediastinum. Consolidation is commonly infective but in some cases it is due to eosinophilic consolidation probably associated with allergic aspergillosis. Collapse ranges from subsegmental to lobar and occasionally involves the whole lung. Collapse is due to mucoid impaction in large airways or more commonly mucous plugging in many small airways.

Computed tomography findings The clinical indications for CT in patients with asthma include the detection of bronchiectasis in patients with suspected ABPA, the documentation of the presence and extent of emphysema in smokers with asthma, and the identification of conditions that may be confused with asthma, such as hypersensitivity pneumonitis. In uncomplicated asthma, HRCT may show bronchial dilatation, bronchial wall thickening, mucoid impaction, decreased lung attenuation, air trapping and small centrilobular opacities24,25. These abnormalities may or may not be reversible with steroid treatment. The prevalence of these thin-section CT abnormalities increases with increasing severity of symptoms. Considerable variation exists, however, in the reported frequency of abnormalities. This variation is related to differences in diagnostic criteria and patient selection. Bronchial wall thickness measured on CT is prominent in patients with more severe asthma26. It correlates with the duration and severity of disease and the degree of airflow obstruction. Peribronchial inflammation may be partly responsible for the bronchial wall thickening, but if this feature is not reversible with steroid treatment, it reflects the development of hyperplasia and hypertrophy of smooth muscle on the bronchial wall reflecting airway wall remodelling. This observation supports the concept that quantitative assessment of bronchial wall area on CT could be used to assess airway wall remodelling in asthmatic patients in longitudinal studies to evaluate the effects of new therapies.

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

Focal and diffuse areas of decreased lung attenuation seen in 20–30% of asthmatic patients are likely the results of a combination of air trapping and pulmonary oligaemia owing to alveolar hypoventilation. The areas of decreased attenuation in acute asthma almost always reflect hypoxic vasoconstriction in parts of the lung that are underventilated as a result of bronchospasm, and such areas of air trapping are more conspicuous and extensive on expiratory CT. In chronic asthma, morphological features of emphysema on CT are almost invariably related to cigarette smoking, rather than the asthma per se, in which the decreased attenuation areas represent small airway obstruction. Expiratory CT can show abnormal air trapping even in patients who have normal inspiratory images. The extent of such air trapping correlates with the severity of the asthma. Abnormal expiratory air trapping has been observed in 50% of asthmatic patients.This reflects the luminal obstruction of the airways and is potentially, but not always, reversible. CT may depict air trapping before lung function deteriorates. The mosaic perfusion pattern is frequent in patients with moderate persistent asthma27. In severe persistent asthma, diffuse decreased lung attenuation and expiratory air trapping make the pattern difficult to distinguish from that of obliterative bronchiolitis. In patients with persistent asthma, no change in air trapping scores after inhalation of a bronchodilator suggest that the air trapping may reflect permanent changes resulting from small airway remodelling27. Airway remodelling caused by smooth muscle hypertrophy and hyperplasia accounts for the faster and greater decrease in forced expiratory volume per second with age in asthmatics compared with controls.

OBLITERATIVE (CONSTRICTIVE) BRONCHIOLITIS1,3,4,6,13 Inflammation of the bronchioles (bronchiolitis) is a very common lesion in the lungs. However, the extent of such lesions is rarely extensive enough to cause clinical symptoms. Pathological studies have repeatedly emphasized the frequent involvement of the bronchioles in diverse diffuse disease. Inflammation of the bronchioles may be reversible under specific or antiinflammatory treatment or lead to subsequent scaring and obliteration. Obliterative bronchiolitis is a condition characterized by bronchiolar and peribronchiolar inflammation and fibrosis that ultimately leads to luminal obliteration affecting the membranous and respiratory bronchioles. Obliterative bronchiolitis is the result of a variety of causes but in rare cases it is idiopathic (Table 16.2). When a large proportion of the airways is affected, patients usually present with progressive shortness of breath and functional evidence of airflow obstruction.

Pathological features The pattern of obliterative bronchiolitis is characterized by the development of an irreversible circumferential submucosal fibrosis, resulting in bronchiolar narrowing or obliteration of bronchioles in the absence of intraluminal granulation tissue polyps or surrounding parenchymal inflammation. Proliferation of fibrosis extends predominantly between the epithelium

307

308

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 16.2 CAUSES OF AND ASSOCIATION WITH OBLITERATIVE (CONSTRICTIVE) BRONCHIOLITIS • Postinfection Childhood viral infection (adenovirus, respiratory syncytial virus, influenza, parainfluenza) Adulthood and childhood (Mycoplasma pneumoniae, Pneumocystis carinii in AIDS patients, endobronchial spread of tuberculosis, bacterial bronchiolar infection) • Postinhalation (toxic fume and gases) Nitrogen dioxide (silo filler’s disease), sulphur dioxide, ammonia, chlorine, phosgene Hot gases • Gastric aspiration Diffuse aspiration bronchiolitis (chronic occult aspiration in the elderly, patients with dysphagia) • Connective tissue disorders Rheumatoid arthritis Sjögren’s syndrome • Allograft recipients Bone marrow transplant Heart–lung or lung transplant • Drugs Penicillamine Lomustine • Ulcerative colitis • Other conditions Bronchiectasis Chronic bronchitis Cystic fibrosis Hypersensitivity pneumonitis Sarcoidosis Microcarcinoid tumorlets (neuroendocrine cell hyperplasia) Sauropus androgynus ingestion • Idiopathic

and the muscular mucosa and along the long axis of the airway, impairing collateral ventilation and leading to airflow obstruction. The epithelium overlying the abnormal fibrosis tissue may be flattened or metaplastic and is usually intact without any ulceration. In some instances, the accompanying artery is also obliterated by the same fibrotic process.

Radiographic findings The chest radiograph is often normal. In a small number of patients, mild hyperinflation, subtle peripheral attenuation of the vascular markings, widespread and conspicuous abnormalities in lung attenuation, and central bronchiectasis may be seen (see Fig. 16.6). Thin-section CT is superior to radiography in demonstrating the presence and extent of abnormalities. The main thin-section CT findings usually consist of areas of decreased lung attenuation associated with vessels of decreased calibre on inspiratory images and air trapping on expiratory images. Because the lesions of bronchiolar narrowing or obstruc-

tion are heterogeneously distributed throughout the lungs, redistribution of blood flow to areas of normal lung or less diseased areas give a pattern of mosaic perfusion. Bronchial wall thickening and bronchiectasis, both central and peripheral, are also commonly present. Although the vessels within areas of decreased attenuation on thin-section CT may be of markedly reduced calibre, they are not distorted as in emphysema. The lung areas of decreased attenuation related to decreased perfusion can be patchy or widespread. They are poorly defined or sharply demarcated, giving a geographical outline, and represent a collection of affected secondary pulmonary lobules. Redistribution of blood flow to the normally ventilated areas causes increased attenuation of lung parenchyma in these areas. The patchwork of abnormal areas of low attenuation and normal lung or less diseased areas, appearing normal in attenuation or hyperattenuated, gives the appearance of mosaic attenuation. The vessels in the abnormal hypoattenuated areas are reduced in calibre, whereas the vessels in normal areas are increased in size; the resulting pattern is called mosaic perfusion. The difference in vessel size between low and high attenuation areas allows the mosaic perfusion pattern to be distinguished from mosaic attenuation due to an infiltrative lung disease with patchy distribution, in which the vessels have the same calibre in both high and normal attenuation areas.The areas of decreased lung attenuation and perfusion may be confined to or predominant in one lung, particularly in Swyer–James or MacLeod syndrome, which is a variant form of postinfectious obliterative bronchiolitis in which the obliterative bronchiolar lesions affect predominantly one lung. Usually the regional inhomogeneity of the lung density seen at end-inspiration on thin-section CT is accentuated on sections obtained at end, or during, expiration because the high attenuation areas increase in density and the low attenuation areas remain unchanged. In the case of more global involvement of the small airways, the lack of regional homogeneity of the lung attenuation is difficult to perceive on inspiratory CT, and as a result mosaic perfusion becomes visible only on expiratory images (Fig. 16.22). In patients with particularly severe and widespread involvement of the small airways, the patchy distribution of hypoattenuation and mosaic pattern is lost. On inspiratory CT there is uniformly decreased attenuation in the lungs, and images taken at end-expiration may appear unremarkable. In these patients, the most striking features are a paucity of pulmonary vessels and no difference in the cross-sectional areas of the lung at comparable levels on inspiratory and expiratory images. In such a situation, there is a risk of misdiagnosis between obliterative bronchiolitis and panlobular emphysema. Both conditions are characterized by bronchial wall thickening and generalized decreased attenuation of the lung parenchyma and bronchial dilatation. However, patients with panlobular emphysema demonstrate parenchymal destruction with higher frequency and greater extent than those with obliterative bronchiolitis. Long lines reflecting limited thickened interlobular septa were significantly more frequent in patients with panlobular emphysema28.

CHAPTER 16

Figure 16.22 Obliterative bronchiolitis. HRCT acquired at (A) full inspiration and (B) full expiration. The mosaic perfusion appearance is very difficult to perceive on the inspiration image (A). The contrast in attenuation between normal and abnormal areas is accentuated at expiration. The areas that did not change in attenuation between inspiration and expiration represent areas of lung parenchyma containing obliterative lesions on the bronchioles.

Assessment of air trapping with computed tomography The most commonly used CT technique for the assessment of air trapping is based on postexpiratory thin-section images obtained during suspended respiration following a forced exhalation. Each of the postexpiratory images is compared with the inspiratory image that most closely duplicates its level to detect air trapping. More recently, dynamic expiratory manœuvre performed during helical CT acquisition has been described29. Motion artefacts, which increase as temporal resolution decreases, represent the major limitation of continuous expiratory CT. However, motion artefacts are at a maximum during the early phase of expiration and at a minimum during its late phase, which allows good visualization of lobular air trapping with helical CT. The extent of air trapping and the relative contrast scores are significantly higher with continuous expiratory CT than those obtained with suspended end-

•

LARGE AIRWAY DISEASE AND CHRONIC AIRWAY OBSTRUCTION

expiratory CT. This improvement can be explained by a small increase in the degree of expiration, which leads to a better detection of air trapping29. This technique is recommended when patients have difficulty performing the suspended endexpiration manœuvre adequately. MDCT with thin collimation over the lungs and low dose has become routine in many institutions to improve the conspicuity and apparent extent of air trapping. The technique of multiplanar volume rendering slab associated with the technique of minimum intensity projection increases the contrast between areas of normal lung attenuation and areas of lung hypoattenuation. This helps depict the mosaic perfusion pattern. Its application in expiratory CT can also facilitate the detection of air trapping and the assessment of its extent. The texture analysis technique has been developed to discriminate between patterns of obstructive lung disease on the basis of parenchymal texture alone on HRCT. The extent of air trapping present on expiratory CT can be measured using a semiquantitative scoring system that estimates the percentage of lung that appears abnormal in each study. In the scoring system proposed by Stern et al, estimates of air trapping were made at each level and for each lung on a four-point scale: 0: no air trapping; 1: 1–25%; 2: 26–50%; 3: 51–75%; and 4: 76–100% of the cross-sectional area of the lung affected. The air trapping score is the summation of these numbers for the level studied.This scoring system allows good inter- and intra-observer agreement. The extent of expiratory air trapping at CT has proved to correlate with the degree of airflow obstruction at pulmonary function tests in patients with obliterative bronchiolitis30. Objective measurement of air trapping can be done using CT densitometry. The assessment of extent can be expressed as a mean density of the voxels included in a chosen region of interest; as a histogram that shows the distribution of attenuation values within the lung; and as a density mask that highlights, or as a calculation that summarizes, the pixels with a density below a certain critical value. In the density mask technique, all the pixels included in areas of air trapping are segmented by thresholding at −910 HU and are highlighted and automatically counted. This permits calculation of the pixel index, which is defined as the percentage of pixels in both lungs on a single study that show an attenuation lower than a predetermined threshold value. Expressing lung density on a histogram has the advantage that changes in the distribution of attenuation values are detectable when mean attenuation is unchanged. Density changes between full inspiration and full expiration can be compared, and expiratory:inspiratory ratios can be calculated. The density mask has the advantage that it combines density measurement with the visual assessment of pathology. Using MDCT with thin collimation over the lungs performed at full expiration, an exhaustive assessment of the volume of air trapping may be provided, as well as a 3D visualization of the distribution of air trapping. Sophisticated image processing techniques can be used to compensate for the nondependent–dependent lung attenuation gradient.

309

310

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

REFERENCES 1. Hansell D M, Armstrong P, Lynch D A et al 2005 Imaging of diseases of the chest, 4th edn. Elsevier Mosby, Philadelphia 2. Kwong J S, Muller N L, Miller R R 1992 Diseases of the trachea and main-stem bronchi: correlation of CT with pathologic findings. RadioGraphics 12: 645–657 3. Muller N L, Fraser R G, Lee K S et al 2003 Diseases of the lung. Lippincott Williams & Wilkins, Philadelphia 4. Naidich D P, Webb W R, Grenier P A et al 2005 Imaging of the airways. Lippincott Williams & Wilkins, Philadelphia 5. Berkmen Y M 1984 The trachea: the blind spot in the chest. Radiol Clin North Am 22, 539–562 6. Takasugi J E, Godwin J D 1998 Radiology of chronic obstructive pulmonary disease. Radiol Clin North Am 36: 29–55 7. Baroni R H, Feller-Kopman D, Nishino M et al 2005 Tracheobronchomalacia: comparison between end-expiratory and dynamic expiratory CT for evaluation of central airway collapse. Radiology 235: 635–641 8. Grenier P A, Beigelman-Aubry C, Fetita C et al 2002 New frontiers in CT imaging of airway disease. Eur Radiol 12: 1022–1044 9. Roberts H R, Wells A U, Milne D G et al 2000 Airflow obstruction in bronchiectasis: correlation between computed tomography features and pulmonary function tests. Thorax 55: 198–204 10. Grenier P, Maurice F, Musset D et al 1986 Bronchiectasis: assessment by thin-section CT. Radiology 161: 95–99 11. Lucidarme O, Grenier P, Coche E et al 1996 Bronchiectasis: comparative assessment with thin-section CT and helical CT. Radiology 200: 673–679 12. Remy-Jardin M, Amara A, Campistron P et al 2003 Diagnosis of bronchiectasis with multislice spiral CT: accuracy of 3-mm-thick structured sections. Eur Radiol 13: 1165–1171 13. Webb W R 1994 High-resolution computed tomography of obstructive lung disease. Radiol Clin North Am 32: 745–757 14. Foster W L Jr, Pratt P C, Roggli V L et al 1986 Centrilobular emphysema: CT–pathologic correlation. Radiology 159: 27–32 15. Spouge D, Mayo J R, Cardoso W et al 1993 Panacinar emphysema: CT and pathologic findings. J Comput Assist Tomogr 17: 710–713 16. Stern E J, Webb W R, Weinacker A et al 1994 Idiopathic giant bullous emphysema (vanishing lung syndrome): imaging findings in nine patients. Am J Roentgenol 162: 279-282 17. Bankier A A, De Maertelaer V, Keyzer C et al 1999 Pulmonary emphysema: subjective visual grading versus objective quantification with macroscopic morphometry and thin-section CT densitometry. Radiology 211: 851–858

18. Muller N L, Staples C A, Miller R R et al 1988 “Density mask.” An objective method to quantitate emphysema using computed tomography. Chest 94: 782–787 19. Gevenois P A, de Maertelaer V, De Vuyst P et al 1995 Comparison of computed density and macroscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 152: 653–657 20. Gevenois P A, De Vuyst P, de Maertelaer V et al 1996 Comparison of computed density and microscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 154: 187–192 21. Nakano Y, Muro S, Sakai H et al 2000 Computed tomographic measurements of airway dimensions and emphysema in smokers. Correlation with lung function. Am J Respir Crit Care Med 162: 1102–1108 22. Remy-Jardin M, Remy J, Boulenguez C et al 1993 Morphologic effects of cigarette smoking on airways and pulmonary parenchyma in healthy adult volunteers: CT evaluation and correlation with pulmonary function tests. Radiology 186: 107–115 23. Remy-Jardin M, Remy J, Gosselin B et al 1993 Lung parenchymal changes secondary to cigarette smoking: pathologic–CT correlations. Radiology 186: 643–651 24. Grenier P, Mourey-Gerosa I, Benali K et al 1996 Abnormalities of the airways and lung parenchyma in asthmatics: CT observations in 50 patients and inter- and intraobserver variability. Eur Radiol 6: 199–206 25. Park C S, Muller N L, Worthy S A et al 1997 Airway obstruction in asthmatic and healthy individuals: inspiratory and expiratory thinsection CT findings. Radiology 203: 361–367 26. Niimi A, Matsumoto H, Amitani R et al 2000 Airway wall thickness in asthma assessed by computed tomography. Relation to clinical indices. Am J Respir Crit Care Med 162: 1518–1523 27. Laurent F, Latrabe V, Raherison C et al 2000 Functional significance of air trapping detected in moderate asthma. Eur Radiol 10: 1404–1410 28. Copley S J, Wells A U, Muller N L et al 2002 Thin-section CT in obstructive pulmonary disease: discriminatory value. Radiology 223: 812–819 29. Lucidarme O, Grenier P A, Cadi M et al 2000 Evaluation of air trapping at CT: comparison of continuous versus suspended-expiration CT techniques. Radiology 216: 768–772 30. Hansell D M, Rubens M B, Padley S P et al 1997 Obliterative bronchiolitis: individual CT signs of small airways disease and functional correlation. Radiology 203: 721–726

CHAPTER

Pulmonary Lobar Collapse: Essential Considerations

17

Susan J. Copley

• • • • •

Mechanisms and causes of lobar collapse Radiographic considerations Computed tomography of lobar collapse Other imaging techniques in lobar collapse Patterns of lobar collapse

Collapse and atelectasis are terms which are often used synonymously and refer to loss of volume within the lung. In North America, the term collapse is often reserved to denote complete loss of volume within an entire lobe or lung1.

MECHANISMS AND CAUSES OF LOBAR COLLAPSE Broadly, lobar collapse can be divided into those due to endobronchial obstruction (either intrinsic or extrinsic) and those without obstruction2,3.The causes of lobar collapse are summarized in Table 17.1.The common causes differ slightly between adults and children. In adults the frequent causes of intrinsic obstruction are tumours and mucus plugs. In the clinical context of a middleaged or elderly smoker, lobar collapse should always be suspected to be due to a bronchogenic carcinoma until proved otherwise. All cell types of bronchogenic carcinoma can potentially cause intrinsic large airway obstruction and produce segmental, lobar or whole lung collapse (Fig. 17.1)2. More rarely, foreign bodies, broncholiths and focal bronchostenosis due to inflammation or trauma may be encountered. In children, causes such as inhaled foreign bodies or mucus plugs are common (Fig. 17.2), with tumours being very rare.

RADIOGRAPHIC CONSIDERATIONS The cardinal radiographic features of lobar collapse are increased opacity of the affected lobe and volume loss. The latter can be inferred by direct and indirect signs. Direct signs of volume loss refer to displacement of interlobar fissures, pulmonary vessels and bronchi, whereas indirect signs include

compensatory shifts of adjacent structures such as hyperinflation of other lobes. The effects of a lobar collapse are often maximal on immediately adjacent structures, e.g. an upper lobe collapse often results in a shift of the superior mediastinum, whereas a lower lobe collapse often demonstrates elevation of the posterior part of the diaphragm in particular. However, the general principles and fundamental radiographic signs are similar for all lobes. A collapsed lobe appears radiographically dense due to a combination of retained secretions or fluid within the lobe and reduction in aeration of the lobe4. However, retained fluid is the dominant process resulting in increased opacity of a partially collapsed lobe, as virtually complete collapse is required to displace sufficient air for the normally radiographically hyperlucent lung to appear dense.

Direct signs of volume loss Displacement of fissures is a reliable feature of lobar collapse, and is generally characteristic depending on the affected lobe5. The pulmonary vessels and bronchi become crowded together in the affected lobe as the lung loses volume. The sign may be one of the earliest seen in lobar collapse and can often be readily appreciated by comparison with previous radiographs. Hilar elevation on the PA chest radiograph is a wellknown sign of upper lobe collapse: the ipsilateral interlobar and lower lobe arteries remain visible as these structures are still outlined by aerated lung. It would seem logical to consider ‘hilar depression’ to be a sign of lower lobe collapse, but some authorities believe the small hilum to be a more accurate description5. This is due to the fact that when a lower lobe collapses, the opaque, collapsed lobe obscures the lower lobe artery that lies within it, and the interlobar artery is usually rotated so the margin is no longer in profile to the frontal X-ray beam. Consequently, it is difficult to recognize the hilum as being depressed and instead, smaller vascular structures are noted at the expected position of the hilum. Occasionally, confusion with a central hilar mass/ adenopathy can arise if the convex margin of an interlobar artery remains visible due to minimal rotation6.

312

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 17.1 CAUSES OF LOBAR COLLAPSE 2,3 Lobar collapse due to endobronchial obstruction • Intrinsic Bronchogenic carcinoma Bronchial carcinoid Adenoid cystic carcinoma Metastases (e.g. breast, renal cell and colonic carcinoma, melanoma, sarcoma) Lymphoma Benign tumours (e.g. lipoma, hamartoma, papillomas, endometriomas) Granulomatous diseases (e.g. sarcoidosis and tuberculosis) Miscellaneous conditions (e.g. aspirated foreign bodies, mucus plugs, gastric contents, malpositioned endotracheal tubes, bronchial torsion or rupture, amyloidosis, Wegener’s granulomatosis) • Extrinsic Hilar or mediastinal lymphadenopathy (commonly due to bronchogenic or breast carcinoma) Mediastinal masses Fibrosing mediastinitis Aortic aneurysms and congenital vascular anomalies Cardiac enlargement

Lobar collapse without endobronchial obstruction • Miscellaneous conditions (e.g. passive collapse due to pleural fluid or pneumothorax, radiation-induced collapse, tumour replacement [bronchiolo-alveolar cell carcinoma])

Figure 17.1 Total left lung collapse. (A) Frontal and (B) lateral chest radiographs. The cause of the collapse is a bronchogenic carcinoma; the endobronchial component is visible as an abrupt cut-off of the left main bronchus. Note the marked displacement of the right lung anteriorly and posteriorly across the midline (arrows). Note the marked anterior hyperlucency of the thorax on the lateral view (B).

As well as vascular reorientation, hilar bronchial alterations also occur. The central large bronchi undergo characteristic changes in position with collapse of either the upper or lower lobe. When either upper lobe collapses significantly, the ipsilateral main bronchus becomes more horizontally orientated than usual, hence the bronchus intermedius and the left lower lobe bronchus swing laterally. Conversely, when either lower lobe collapses, each main bronchus is more vertically orien-

tated than usual, with a medial swing of the bronchus intermedius on the right and the lower lobe bronchus on the left.

Indirect signs of volume loss Compensatory hyperinflation of adjacent lobes occurs with lobar collapse, resulting in fewer vessels per unit volume of lung. It is often easier to detect a paucity of vessels, which are more widely spaced than on the unaffected side, than

CHAPTER 17

Figure 17.2 Total right lung collapse in a neonate. The patient was ventilated for respiratory distress syndrome and the cause of the total lung collapse was a mucus plug.

subtle increased radiolucency. In isolation, the sign may be due to causes other than lobar collapse and other confirmatory features should be sought before making the diagnosis. The normal lung parenchyma should expand proportionally to compensate for the degree of collapse and often the greater the degree of lobar collapse, the greater the compensatory overinflation. Therefore when small lung volumes are involved, the hyperinflation usually only involves the remainder of the ipsilateral lung, whereas with larger volumes, the contralateral lung may expand across the midline. On a frontal radiograph the lung may expand across the midline superiorly, thus displacing the anterior junctional line to the contralateral side (see Fig. 17.1A). On a lateral view, the anterior medias-

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

tinum appears hyperlucent (see Fig. 17.1B). Displacement of the azygo-oesophageal line and posterior junctional line on the PA radiograph, which denote protrusion of contralateral lung through other weak areas between the oesophagus and vertebral column and the retrocardiac space respectively, may be more difficult to recognize. Although the term ‘mediastinal herniation’ is sometimes used, some authorities emphasize that there is no actual mediastinal defect or hiatus and the sign more accurately denotes displacement of mediastinal structures7. A divergent or parallel pattern of vascular reorientation seen near the hilum has been described in marked upper lobe collapse8.The pattern is seen more commonly on the left than on the right, as a result of the different degree of compensatory overinflation in the superior segment of the ipsilateral lower lobe on each side8. The right middle lobe can also overinflate in compensation, which further explains the lesser degree of overinflation of the superior segment of the right lower lobe. The sign of vascular reorientation can be helpful when unusual patterns of upper lobe collapse are present. Hyperexpansion may also result in a change in position of lung lesions, such as granulomas resulting in the so-called shifting granuloma sign (Fig.17.3). Of particular note, the Luftsichel sign (from German, meaning air crescent) is due to the overinflated superior segment of the ipsilateral lower lobe occupying the space between the mediastinum and the medial aspect of the collapsed upper lobe, resulting in a paramediastinal translucency (Fig. 17.4)9. The sign is more common on the left than the right and is regarded as a typical appearance of left upper lobe collapse9. CT demonstrates the increased paramediastinal lucency to be due to a wedge shape of the collapsed upper lobe, with the apex of the V resulting from tethering of the major fissure by hilar structures (Fig. 17.4B).

Figure 17.3 Shifting granuloma sign. (A) Pre and (B) post right lower lobe collapse.

313

314

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 17.4 Luftsichel sign. (A) A left upper lobe collapse demonstrating paramediastinal lucency (arrow). (B) CT shows interposition of aerated lung between the collapse and the mediastinum (arrow). There is also a large right paratracheal node causing some distortion of the SVC.

Mediastinal shift is another indirect sign of volume loss and the degree varies according to the position of the affected lobe. Usually the least mediastinal shift occurs in right middle lobe collapse, whilst the greatest shift, particularly of the inferior mediastinum, is seen with lower lobe collapse. The amount of mediastinal shift due to upper lobe collapse is often dependent on the chronicity: in acute upper lobe collapse there is often little shift, whereas in chronic upper lobe volume loss with fibrosis, the shift may be greater. The position of the trachea may be a useful indicator of superior mediastinal shift as it should be central in the superior mediastinum between the anterior ends of the clavicles or slightly deviated to the right by the aortic arch. Inferiorly within the mediastinum, anywhere between one-half and one-fifth of the cardiac outline normally lies to the right of the midline and greater or lesser variations indicate mediastinal shift. However, because of the wide variation in normal subjects, displacement of the cardiac outline may be more difficult to assess than changes in position of the trachea. The hemidiaphragms may be elevated in lobar collapse, particularly involving the left upper lobe and to a lesser extent the right upper and both lower lobes. However, the sign is of limited value because the position of the right hemidiaphragm is highly variable (0–3 cm higher than the left on the frontal chest radiograph). A useful ancillary sign of upper lobe collapse (or a combination of right upper and middle lobe collapse) is a juxtaphrenic peak of the diaphragm (Fig. 17.5)10. The sign refers to a small triangular density at the highest point of the dome of the hemidiaphragm, due to the anterior volume loss of the affected upper lobe resulting in traction and reorientation of an inferior accessory fissure11,12. Reduction in the volume of a hemithorax may result in relative reduction of the spaces between the ribs by comparison to the unaffected side. Rib crowding or approximation may be recognizable on the frontal radiograph in cases of chronic lobar collapse, but in acute collapse it may be more difficult to appreciate. Furthermore, the sign is considered to be unreli-

Figure 17.5 Juxtaphrenic peak sign. A small triangular density (arrow) is seen in a left upper lobe collapse. The sign is due to reorientation of an inferior accessory fissure.

able as patient rotation and minor degrees of scoliosis may result in apparent rib crowding.

Ancillary features of lobar collapse Occasionally the cause of a lobar collapse may be apparent and an endobronchial lesion may be clearly demonstrated radiographically (Fig. 17.1A). However, although the actual endobronchial component is often not directly visualized, lobar collapse due to a central obstructing bronchogenic carcinoma is most likely when Golden’s S sign is seen (Fig. 17.6). The sign refers to the S shape (or more accurately, reverse S on the right) of the fissure due to the combination of collapse and mass centrally resulting in a focal convexity with a concave outline peripherally. Although the sign was originally described in the right upper lobe, it can be seen in any lobe5,13. The CT equivalent is discussed later. Generally, absence of air

CHAPTER 17

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

Figure 17.6 Golden’s S sign. A right upper lobe collapse demonstrating peripheral concavity and central convexity (arrows) due to an underlying bronchogenic carcinoma resulting in a reverse S shape.

bronchograms within the affected lobe should also raise the suspicion of a central obstructing lesion as there is absorption of air from both the lung parenchyma and airways. The sign may be useful to distinguish a central obstructing mass from a consolidative process such as bacterial pneumonia (Fig. 17.7). The rare important caveats are when a mass results in only partial obstruction of the airways or in cases of acute bronchopneumonia where the airways are filled with an inflammatory exudate. However, the sign is not as reliable on CT, and often distal air bronchograms are visible in part of a collapsed lobe due to a central neoplasm (Fig. 17.8).

Figure 17.8 CT of a collapsed right upper lobe due to a squamous cell carcinoma. Note the peripheral air bronchograms (arrow) in (A) despite a central obstructing mass with amorphous calcification (B). There is a convex border of the collapsed lobe (arrows) (B) which is the CT equivalent of Golden’s S sign.

COMPUTED TOMOGRAPHY OF LOBAR COLLAPSE CT has become an invaluable method for the investigation of patients with lobar collapse. The obvious benefits are a lack of superimposition of overlying structures with the added advantage of demonstration of anatomical structures in the axial and, with computer reformatting, coronal and sagittal planes. Not only does CT aid the understanding of the radiographic appearances of lobar collapse, it also provides invaluable information about the cause which may not be apparent on chest radiography. The most common indication for CT in adults with lobar collapse is to identify an endobronchial or compressing lesion.

Technique

Figure 17.7 Air bronchograms in a collapsed and consolidated right lower lobe. The sign can be helpful in excluding a central obstructing mass and in this case the cause was a bacterial pneumonia.

Careful attention to CT technique is sometimes required to accurately demonstrate an obstructing lesion resulting in lobar collapse. The old recommendations using single slice CT14 have been superseded by techniques using helical and now multidetector CT (MDCT). Such systems can often provide high resolution images from the same data acquisition as used for other purposes with a saving in radiation dose. The facility to reconstruct and reformat volumetric data has resulted

315

316

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

in advancements in the display of tracheobronchial anatomy. Three-dimensional (3D) and multiplanar (2D) images provide an extremely useful adjunct to axial images15.

Utility In some cases, the aetiology of lobar collapse can be determined from the patient’s clinical history, examination, and chest radiographic features. Using fibre-optic bronchoscopy as the reference standard, CT is clearly more sensitive than chest radiography for the detection of an obstructing carcinoma16. Reported sensitivities for detection by CT range from 83 to 100%17–20 but generally, when an endobronchial lesion is sufficiently large to cause lobar collapse, CT is a reliable method for detection2,16. False-positive diagnoses may be due to bronchial strictures, plugs of mucus or secretions and compression by large pleural effusions2,3,16. CT is not histologically specific, however, as bronchogenic carcinoma, endobronchial metastases, bronchial adenomas and lymphoma may all have similar appearances2. The accuracy of CT is, to some extent, dependent on technique and it may be more difficult to demonstrate endobronchial lesions in the right middle lobe and lingular bronchi owing to their oblique orientation relative to the axial plane. This is much less of a problem with MDCT. Accurate delineation of a tumour mass from a surrounding collapsed lobe may be problematical, but collapsed lung usually enhances to a greater degree than tumour with dynamically contrast-enhanced CT (Fig. 17.9)21. The difference in attenuation value is maximal between 40 s and 2 min after a bolus injection of contrast medium21. In practice, the distinction between tumour and collapsed lung may not be important and the technique is reserved for cases where more accurate delineation is necessary for treatment purposes, e.g. radiotherapy planning. Golden’s S sign on chest radiography has a CT equivalent that may be helpful in identifying an obstructing tumour22,23. Usually, a collapsed lobe is associated with concavity of the

Figure 17.9 CT of right upper lobe collapse due to bronchogenic carcinoma. Note how the attenuation of the necrotic tumour is lower than the adjacent collapsed lung which enhances with intravenous contrast medium.

adjacent fissure and a localized convexity is highly suggestive of an underlying mass (see Fig. 17.8). The sign is not entirely specific, but it is strongly indicative of a bronchogenic carcinoma. Unlike the frontal chest radiograph, in which the S sign is only helpful in the right upper lobe and to a lesser extent the right and left lower lobes, the S sign can be applied to all lobes on CT. Another CT sign that is highly suggestive of an obstructing lesion causing lobar collapse is the CT mucous bronchogram sign24. Histopathologically, the lobar and segmental bronchi are filled with inspissated secretions and are usually dilated. The airways are optimally demonstrated as tubular, low attenuation branching structures within the enhancing collapsed lobe following intravenous contrast enhancement (Fig. 17.10).

Figure 17.10 Left lung collapse. (A, B) Contrast-enhanced CT sections of whole lung collapse due to a squamous cell carcinoma in the left main bronchus (arrow in A). There is also a left pleural effusion and a small pericardial effusion. Note the low-attenuation areas relative to the densely enhancing left lower lobe parenchyma (B) which represent mucus-filled airways—the CT mucous bronchogram sign.

CHAPTER 17

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

Obstructing lesions such as bronchogenic carcinoma or benign causes, including tuberculous bronchostenosis, should be considered. The sign may also result from excessive mucus production combined with decreased mucociliary function in conditions such as allergic bronchopulmonary aspergillosis, asthma and cystic fibrosis25. CT also has a role in complicated or atypical lobar collapse as their appearances may be confusing on chest radiography. In particular, combined right middle lobe and right upper lobe collapse may be difficult to diagnose on chest radiography when the collapse is nearly complete26. CT is useful for demonstrating mediastinal anatomy and provides information about mediastinal lymph nodes and the staging of a tumour causing lobar collapse.Additional signs of lobar collapse, such as compensatory overinflation and the Luftsichel sign (described above), are also well demonstrated, providing explanations for the radiographic appearances of lobar collapse27,28.

Potential pitfalls The increased sensitivity of CT by comparison with radiography means that the presence of an air bronchogram within a lobar collapse does not necessarily exclude a central obstructing lesion (see Fig. 17.8.)24. In this context, an air bronchogram may be seen in the peripheral part of a collapsed lobe due to collateral air drift or tumour necrosis16. Similarly, a proximal obstructing lesion may not cause complete lobar collapse when a fissure is incomplete allowing ventilation by collateral air drift29. Occasionally the parenchyma and airways become filled with fluid owing to the presence of a central obstructing lesion with little or no associated volume loss, and the lobe may even be expanded giving rise to the appearance termed ‘drowned lobe’. The CT equivalent of the Golden’s S sign is particularly well demonstrated with right-sided lobar collapse and, on the left, care should be taken in interpretation owing to the fact that normal mediastinal structures may mimic a mass (e.g. thoracic aorta)22 (Fig. 17.11).

Figure 17.11 Left lower lobe collapse. Contrast-enhanced CT showing a tight left lower lobe collapse. Normal mediastinal structures (particularly left-sided) may cause a focal bulge in the contour of a lobar collapse (in this case by the well opacified descending thoracic aorta) and should not be confused with a Golden’s S sign due to tumour.

The accurate determination of the reversibility and chronicity of a lobar collapse may be problematical. Relatively acute collapses may show apparent bronchiectatic dilatation of the airways and may mimic a long-standing irreversible event (Fig. 17.12); a meaningful evaluation of the airways in the context of a lobar collapse is therefore often difficult.

OTHER IMAGING TECHNIQUES IN LOBAR COLLAPSE Magnetic resonance imaging (MRI) has been largely surpassed by CT in the investigation of lobar collapse owing to

Figure 17.12 Resolution of left lower lobe collapse. (A) An initial high-resolution CT of a young female patient with symptoms of recurrent respiratory tract infections shows a collapsed left lower lobe with possible bronchiectatic airways, raising the possibility of chronicity. (B) Follow-up conventional CT at the same level several months later shows complete resolution of the left lower lobe collapse and normal airways. This case illustrates the difficulty in making an accurate assessment of the airways in patients with lobar collapse.

317

318

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

the superior spatial resolution of the latter. In particular, endobronchial tumours and smaller bronchi are less well demonstrated by MRI than CT30. Studies have investigated the ability of MRI to differentiate a tumour mass from postobstructive collapse by utilizing differences in signal characteristics31–33. Sometimes the distinction can be made on T1-weighted images, but it is generally accepted that T2-weighted images are superior as the tumour is of lower signal intensity than the obstructed lung which has a higher water content. On ultrasound of a large pleural effusion, the underlying collapsed lung is often visible as a hyperechoic wedge-shaped area within hypo-echoic or anechoic fluid. In practice, the main utility of ultrasound is to readily distinguish pleural effusion from a collapsed and consolidated lung when radiographic appearances are equivocal. On positron emission tomography (PET), a collapsed lobe often demonstrates less uptake of [18F]fluorodeoxyglucose than tumour. By comparison with CT, PET may therefore provide more accurate delineation of tumour from postobstructive collapsed lung, which may be useful in treatment with radiotherapy34.

Figure 17.13 Right upper lobe collapse. Typical example of a collapsed right upper lobe demonstrating the slightly concave inferior border of the opacified lung due to the horizontal fissure.

PATTERNS OF LOBAR COLLAPSE Right upper lobe collapse On the frontal radiographic view of a right upper lobe collapse, the collapsed lobe forms increased density at the apex of the hemithorax adjacent to the right side of the mediastinum, with the elevated horizontal fissure resulting in a concave inferior outline depending on the degree of collapse (Fig. 17.13). Even in cases where there is no obstructing lesion, there is often a small convexity at the hilum due to the pulmonary veins and artery where the apex of the lobe is attached to the hilum. On the lateral view, the horizontal and oblique fissure approximate and are both displaced superiorly and medially with the collapsed lobe forming a superior ill-defined wedgeshaped density. In cases where the collapse is very severe, the horizontal fissure parallels the mediastinum and appearances may simulate an apical cap of pleural fluid (Fig. 17.14) or mediastinal widening on the frontal radiograph (Fig. 17.15). There is also usually compensatory hyperinflation of the right middle and lower lobes, resulting in elevation and a more horizontal course of the lower lobe pulmonary artery and right main bronchus. The vascular reorientation can be recognized on the frontal view, but the right main and lower lobe bronchial displacement can be difficult to appreciate on both the frontal and lateral view. On CT the right upper lobe forms a triangular density with the base anteriorly against the chest wall and the apex at the hilum (Fig. 17.16). A focal bulge of the lateral border usually indicates an underlying mass. Compensatory hyperinflation of not only the right middle and right lower lobes but also the left upper lobe is often more easily appreciated on CT.

Left upper lobe collapse The cardinal features of left upper lobe collapse are fundamentally different from right upper lobe collapse as there is very rarely a horizontal fissure on the left. Consequently, the main

Figure 17.14 Right upper lobe collapse. An example of right upper lobe collapse mimicking an apical cap of fluid (arrow).

direction of volume loss is anteriorly and medially rather than superiorly, and the entire oblique fissure is displaced in that direction parallel to the chest wall on the lateral view. On the frontal view the signs may be variable depending on the degree of collapse, but there is a ‘veil-like’ increased density of the whole of the affected hemithorax in most cases. The increased density is often greatest at the hilum and it gradually fades out laterally, superiorly, and inferiorly without the clear inferior demarcation of the horizontal fissure as seen in right upper lobe collapse.The difference in transradiancy may be relatively subtle and therefore overlooked by the unwary. The other features that aid diagnosis on the frontal view are loss of the normal silhouette of structures

CHAPTER 17

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

Figure 17.15 Tight right upper lobe collapse. Note how the collapsed lobe (due to a central bronchogenic carcinoma) results in increased right paramediastinal density.

Figure 17.16 CT of right upper lobe collapse. The collapsed lobe forms a triangular wedge of soft tissue anteriorly in the right hemithorax.

adjacent to the collapse, such as the left heart border, mediastinum, and aortic arch, as these structures are no longer adjacent to aerated lung. There is some variability in which outlines are obscured depending on the degree of collapse. In cases of relatively less severe collapse, the left heart border, left mediastinal outline and aortic knuckle are obscured (see Fig. 17.5), whereas in more severe cases the apical segment of the left lower lobe is hyperexpanded superiorly adjacent to the aortic arch and somewhat paradoxically the aortic knuckle outline is therefore visible in more severe cases as it is adjacent to aerated lung (Fig. 17.17A). The Luftsichel sign (described above; see Fig. 17.4) is a particular manifestation of the hyperexpansion, and literally describes an ‘air crescent’ which may be seen between the aortic arch and the medial border of the collapse. On the lateral view the anterior outline of the ascending thoracic aorta can be seen with unusual clarity and this is due to compensatory hyperinflation of the right upper lobe across the midline and rotation of the medi-

Figure 17.17 Left upper lobe collapse. (A) A typical example of left upper lobe collapse demonstrating increased angulation between the left main bronchus and the lower lobe bronchus (arrow) on the frontal view. The aortic knuckle is visible in this example due to compensatory hyperinflation of the left lower lobe. (B) The lateral view demonstrates anterior displacement of the oblique fissure. Note the increased retrosternal lucency (see Fig. 17.18).

astinum so the anterior aspect of the aorta is outlined by aerated lung tangential to the X-ray beam (Fig. 17.17B). This feature is often readily appreciated on CT (Fig. 17.18). On the frontal radiograph the left main bronchus is reorientated and has a more horizontal course than usual. The superior displacement of this structure results in angulation between the left main bronchus and the left lower lobe bronchus (Fig. 17.17A). The CT appearances of left upper lobe collapse are similar to that of the right upper lobe with a triangular soft tissue density, the apex at the origin of the upper lobe bronchus and the base against the anterior chest wall, adjacent to the left border of the mediastinum. However, in the left upper lobe the lingular segment is seen as a density closely opposed to the left heart border. Rarely, left upper lobe collapse may mimic right upper lobe collapse (Fig. 17.19). The appearance is due to collapse of the apicoposterior and anterior segments of the left upper lobe with sparing of the lingular portion resulting in a concavity to the inferior border of the collapse, even in the absence of a left

319

320

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 17.18 Left upper lobe collapse. Intravenous contrastenhanced CT of left upper lobe collapse shows increased wedge-shaped density of the left upper lobe adjacent to the mediastinum. Note the displacement of the right lung across the midline anteriorly, resulting in retrosternal hyperlucency and increased clarity of the anterior ascending thoracic aorta on the lateral view (see Fig. 17.17B).

collapsed lobe lies obliquely in the chest, more parallel with the major fissure, the only sign on the frontal radiograph may be indistinctness of a portion of the right atrial border (Fig. 17.21). By comparison, the triangular density of the collapsed right middle lobe is relatively easy to identify on the lateral view, with approximation of the minor and inferior portion of the major fissure, the apex of the triangle being at the hilum (Fig. 17.21B). In increasingly severe collapse the triangular shape is less marked as the fissures become almost parallel with only a thin wedge of density separating them. The CT appearances are characteristically of a triangularshaped density of varying size adjacent to the heart border. Depending on the orientation of the collapse, only a small portion may be identified on each section as the collapse represents a relatively flat sheet of tissue. The so-called middle lobe syndrome refers to a collapsed right middle lobe with bronchiectasis due to a focal bronchostenosis secondary to pulmonary tuberculosis. Although in theory any lobe may be affected, the middle lobe is the most common, resulting in characteristic CT features (Fig. 17.23).

Right and left lower lobe collapse minor fissure5. Apart from being on the left, isolated collapse of the lingula has a very similar appearance to right middle lobe collapse (Fig. 17.20).

Right middle lobe collapse The features of right middle lobe collapse may be extremely subtle on the frontal view and consequently easy to overlook. The collapsed lobe lies adjacent to the right heart border and there is loss of the silhouette of this structure to a variable degree (Fig. 17.21). There may or may not be a recognizable increase in density depending on the orientation of the collapse relative to the X-ray beam. When the collapse is orientated roughly parallel to the beam or if the patient is in a lordotic position, a triangular, sail-shaped density may be seen adjacent to the heart border (Fig. 17.22). However, if the

The features of right and left lower lobe collapse are very similar and will be considered together. In collapse of the lower lobes, the oblique fissure is displaced posteriorly and medially, and the collapsed lobe lies in the posteromedial portion of the chest, a feature readily appreciated on CT (see Fig. 17.11). On the frontal radiograph, the collapsed lower lobes usually form a triangular density behind the heart (Fig. 17.24). The medial portion of the hemidiaphragm may be obscured as it is no longer outlined by aerated lung (Fig. 17.25), but if the inferior pulmonary ligament is incomplete and does not attach to the diaphragm, the medial contour of the diaphragm may still be visualized. On the lateral radiograph, a posterior portion of the hemidiaphragm may not be seen (Fig. 17.24B), but in more severe collapse the contour may reappear as it becomes outlined by aerated lung from the

Figure. 17.19 Atypical left upper lobe collapse. (A) The frontal radiograph demonstrates the inferior concave border of the collapsed lobe and resembles a right upper lobe collapse. (B, C) CT images show increased triangular density to the left of the mediastinum (B), which does not extend along the left heart border (C), a feature usually seen in left upper lobe collapse. The appearance is due to sparing of the lingular segments.

CHAPTER 17

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

Figure 17.20 Lingular collapse. (A) Frontal view of isolated collapse of the lingular segments of the left upper lobe showing loss of clarity of the left heart border and a raised hemidiaphragm. (B) The similarity to a right middle lobe collapse can be appreciated on the lateral view.

Figure 17.21 Right middle lobe collapse. (A) Frontal view of a typical example showing loss of clarity of the right heart border. (B) The lateral view shows the wedge-shaped density extending anteriorly from the hilum.

Figure 17.22 Right middle lobe collapse. An example showing a triangular-shaped density adjacent to the right heart border.

Figure 17.23 Middle lobe syndrome. High-resolution CT showing right middle lobe collapse and bronchiectasis due to previous tuberculous infection.

321

322

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 17.25 Left lower lobe collapse. A typical appearance of left lower lobe collapse resulting in a triangular density behind the heart (arrowheads). The contour of the medial left hemidiaphragm is lost.

Figure 17.24 Right lower lobe collapse. (A) Frontal view of an example of right lower lobe collapse demonstrating a triangular density which does not obscure the right hemidiaphragm silhouette. (B) The lateral radiograph shows the typical features of increased density of the posterior costophrenic angle and loss of the silhouette of the right diaphragm posteriorly.

hyperexpanded upper lobe. In addition, the vertebral column appears progressively denser inferiorly in lower lobe collapse (Fig. 17.24B), whereas normally the converse is true. On the frontal radiograph the lower lobe pulmonary artery is usually not seen in lower lobe collapse as it is no longer outlined by aerated lung (Fig. 17.26). The major airways, including the right and left main bronchi, are also displaced more vertically in lower lobe collapse and often the relevant aircontaining bronchus can be identified as leading directly into the triangular density of the collapsed lobe. There are several features involving the upper mediastinum which are sometimes helpful in diagnosing lower lobe collapse35. The first of these is the ‘superior triangle sign’ and refers to a triangular density to the right of the mediastinum seen in right lower lobe collapse due to displacement of anterior junctional structures36 (Fig. 17.26). The appearance should not be confused with right upper lobe collapse. The ‘flat waist sign’ is seen in extensive

collapse of the left lower lobe and describes flattening of the contours of the aortic knuckle and main pulmonary artery due to cardiac rotation and displacement to the left37. Third, the outline of the superior aortic knuckle may be lost in severe left lower lobe collapse35. On CT, the collapsed lower lobes form a triangle of soft tissue density posteriomedially in the thorax, adjacent to the spine. On the left, the collapsed lower lobe is seen to drape over the descending aorta, giving a focal convexity to the lateral border, a feature which potentially can cause confusion with an underlying mass on an unenhanced CT (Fig. 17.11).

Whole lung collapse Collapse of an entire lung results in complete opacification or ‘white-out’ of the affected hemithorax. In adults, the cause is often an obstructing neoplasm in the right or left main bronchi (see Fig. 17.1). There is marked volume loss with compensatory hyperinflation of the contralateral lung across the midline. The cardinal feature of volume loss can help discriminate between collapsed lung and a large pleural effusion, the latter usually resulting in mediastinal shift to the contralateral side. The lateral radiograph shows accentuation of the retrosternal space as the displacement of the contralateral lung is greatest anteriorly (see Fig. 17.1). By comparison, the opacity of the hemithorax is more uniform on the lateral view in large pleural effusion and may be a useful discriminating feature in equivocal cases.

Combinations of lobar collapse Occasionally various combinations of lobar collapse occur. Collapse of the right middle and right lower lobes is often due to an obstructing lesion in the bronchus intermedius (Fig. 17.27). The features are similar to right lower lobe collapse with the exception that the opacity extends laterally to the costophrenic

CHAPTER 17

•

PULMONARY LOBAR COLLAPSE: ESSENTIAL CONSIDERATIONS

Figure 17.26 Superior triangle sign. (A) An initial image shows the normal appearances (note the lower lobe artery is clearly visible). (B) The subsequent image shows a right lower lobe collapse demonstrating the superior triangle sign (arrow) (which should not be confused with a right upper lobe collapse). The lower lobe artery can no longer be seen.

Figure 17.27 Combined right middle and right lower lobe collapse. (A) On the frontal view the increased density extends to the right costophrenic angle. (B) On the lateral view the increased density also extends from the anterior to the posterior chest wall. The cause in this case was a bronchogenic carcinoma obstructing the bronchus intermedius.

angle on the frontal view and from the front to the back of the hemithorax on the lateral view5 (Fig. 17.27B). Collapse of the right upper and right middle lobes is more unusual as these lobes do not have a common bronchial origin which spares the lower lobe. In adults the cause is often a carcinoma which obstructs one bronchus and causes extrinsic compression of the other due to mass

effect. Combined collapse of the right upper and right middle lobes results in an appearance very similar to left upper lobe collapse on both frontal and lateral radiographs and CT38. Both bilateral lower lobe and upper lobe collapse are exceedingly rare and may occur as a result of metachronous bronchial neoplasms or mucous plugging (Fig. 17.28).

323

324

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 17.28 Bilateral lower lobe collapse. Bilateral triangular densities are seen with obscuration of the medial portions of the hemidiaphragms. The cause was mucous plugging.

REFERENCES 1. Tuddenham W J 1984 Glossary of terms for thoracic radiology: recommendations of the Nomenclature Committee of the Fleischner Society. Am J Roentgenol 143: 509–517 2. Naidich D P, McCauley D I, Khouri N F, Leitman B S, Hulnick D H, Siegelman S S 1983 Computed tomography of lobar collapse: 1. Endobronchial obstruction. J Comput Assist Tomogr 7: 745–757 3. Naidich D P, McCauley D I, Khouri N F, Leitman B S, Hulnick D H, Siegelman S S 1983 Computed tomography of lobar collapse: 2. Collapse in the absence of endobronchial obstruction. J Comput Assist Tomogr 7: 758–767 4. Stein L A, Vidal J J, Hogg J C, Fraser R G 1976 Acute lobar collapse in canine lungs. Invest Radiol 11: 518–527 5. Proto A V, Tocino I 1980 Radiographic manifestations of lobar collapse. Semin Roentgenol 15: 117–173 6. Proto AV 1984 The chest radiograph: anatomic considerations. Clin Chest Med 5: 213–246 7. Lodin H 1957 Mediastinal herniation and displacement studied by transversal radiography. Acta Radiol 48: 337–350 8. Proto A V, Moser E S Jr 1987 Upper lobe volume loss: Divergent and parallel patterns of vascular reorientation. RadioGraphics 7: 875–887 9. Webber M, Davies P 1981 The Luftsichel: an old sign in upper lobe collapse. Clin Radiol 32: 271–275 10. Kattan K R, Eyler W R, Felson B 1980 The juxtaphrenic peak in upper lobe collapse. Semin Roentgenol 15: 187–193 11. Cameron D C 1993 Juxtaphrenic peak (Katten’s sign) is produced by rotation of an inferior accessory fissure. Australas Radiol 37: 332–335 12. Davis S D, Yankelevitz D F, Wand A, Chiarella D A 1996 Juxtaphrenic peak in upper and middle lobe volume loss: assessment with CT. Radiology 198: 143–149 13. Golden R 1925 The affect of bronchostenosis upon the roentgen-ray shadows in carcinoma of the bronchus. Am J Roentgenol Radiat Ther 13: 21–30 14. Naidich D P, Webb W R, Müller N L, Krinsky G A, Zerhouni E A, Siegelman S S 1999 Computed tomography and magnetic resonance of the thorax, 3rd edn. Lippincott–Raven, Philadelphia

15. LoCicero J, Costello P, Campos C T et al 1996 Spiral CT with multiplanar and three-dimensional reconstructions accurately predicts tracheobronchial pathology. Ann Thorac Surg 62: 818–822 16. Woodring J H 1988 Determining the cause of pulmonary atelectasis: a comparison of plain radiography and CT. Am J Roentgenol 150: 757–763 17. Henschke C I, Davis S D, Auh Y et al 1987 Detection of bronchial abnormalities: comparison of CT and bronchoscopy. J Comput Assist Tomogr 11: 432–435 18. Webb W R, Gamsu G, Speckman J M 1983 Computed tomography of the pulmonary hilum in patients with bronchogenic carcinoma. J Comput Assist Tomogr 1983;7:219–225. 19. Naidich D P, Lee J-J, Garay S M, McCauley D I, Aranda C P, Boyd A D 1987 Comparison of CT and fibreoptic bronchoscopy in the evaluation of bronchial disease. Am J Roentgenol 148: 1–7. 20. Mayr B, Ingrisch H, Häussinger K, Huber R M, Sunder-Plassmann L 1989 Tumours of the bronchi: role of evaluation with CT. Radiology 172: 647–652 21. Onitsuka H, Tsukuda M, Araki A, Murakami J, Torii Y, Masuda K 1991 Differentiation of central lung tumor from postobstructive lobar collapse by rapid sequence computed tomography. J Thorac Imaging 6: 28–31 22. Reinig J W, Ross P 1984 Computed tomography appearance of Golden’s “S” sign. J Comput Assist Tomogr 8: 219–223 23. Khoury M B, Godwin J D, Halvorsen R A, Putman C E 1985 CT of obstructive lobar collapse. Invest Radiol 20: 708–716 24. Woodring J H 1988 The computed tomography mucous bronchogram sign. J Comput Assist Tomogr 12: 165–168 25. Glazer H S, Anderson D J, Sagel S S 1989 Bronchial impaction in lobar collapse: CT demonstration and pathologic correlation. Am J Roentgenol 153: 485–488 26. Saida Y, Itai Y, Kujiraoka Y, Tohno E, Shimizu H T 1997 Bronchoarterial inversion: radiographic–CT correlation in combined right middle and lower lobe collapse. J Thorac Imaging 12: 59–63 27. Flanagan J J, Flower C D, Dixon A K 1982 Compensatory emphysema shown by computed tomography. Clin Radiol 33: 553–554 28. Blankenbaker D G 1998 The Luftsichel sign. Radiology 208: 319–320 29. Woodring J H, Reed J C 1996 Radiographic manifestations of lobar atelectasis. J Thorac Imaging 11: 109–144 30. Mayr B, Heywang S H, Ingrisch H, Huber R M, Häussinger K, Lissner J 1987 Comparison of CT with MR imaging of endobronchial tumors. J Comput Assist Tomogr 11: 43–48 31. Shioya S, Haida M, Ono Y, Fukuzaki M, Yamabayashi H 1988 Lung cancer: differentiation of tumor, necrosis, and atelectasis by means of T1 and T2 values measured in vitro. Radiology 167: 105–109 32. Herold C J, Kuhlman J E, Zerhouni E A 1991 Pulmonary atelectasis: signal patterns with MR imaging. Radiology 178: 715–720 33. Bourgouin P M, McLoud T C, Fitzgibbon J F et al 1991 Differentiation of bronchogenic carcinoma from postobstructive pneumonitis by magnetic resonance imaging: histopathologic correlation. J Thorac Imaging 6: 22–27 34. Nestle U, Walter K, Schmidt S et al 1999 18F-deoxyglucose positron emission tomography (FDG-PET) for the planning of radiotherapy in lung cancer: high impact in patients with atelectasis. Int J Radiat Oncol Biol Phys 44: 593–597 35. Kattan K R 1980 Upper mediastinal changes in lower lobe collapse. Semin Roentgenol 15: 183–186 36. Kattan K R, Felson B, Holder L E, Eyler W R 1975 Superior mediastinal shift in right lower lobe collapse: the “upper triangle sign.” Radiology 116: 305–309 37. Kattan K R, Wiot J F 1976 Cardiac rotation in left lower lobe collapse: “the flat waist sign.” Radiology 118: 275–279 38. Saterfiel J L, Virapongse C, Clore F C 1988 Computed tomography of combined right upper and middle lobe collapse. J Comput Assist Tomogr 12: 383–387

CHAPTER

Pulmonary Neoplasms

18

Simon Padley and Sharyn L. S. MacDonald

• • • • • • •

Bronchial carcinoma Pulmonary sarcoma and other primary malignant neoplasms Benign pulmonary tumours Benign lymphoproliferative disorders Malignant lymphoproliferative disorders Metastases Evaluation of the solitary pulmonary nodule

such as the inappropriate secretion of antidiuretic hormone or a peripheral neuropathy, are the cardinal symptoms at a stage when lobectomy or pneumonectomy may be curative; whereas hoarseness, chest pain, brachial plexus neuropathy and Horner’s syndrome (Pancoast’s tumour), superior vena caval obstruction, dysphagia and the problems of pericardial tamponade indicate invasion of the mediastinum or chest wall, and a poorer prognosis.

Early diagnosis

BRONCHIAL CARCINOMA Lung cancer is the most common cause of cancer-related death in both men and women1,2.Tobacco smoke is the most important causative agent imparting a 20–30-fold increased risk in smokers compared to non-smokers. Other risk factors include passive smoking3, exposure to inorganic substances such as asbestos, nickel and arsenic, interstitial pulmonary fibrosis and radiotherapy.

Pathology The World Health Organization (WHO) classification1 divides bronchial carcinoma into several histological subtypes. Four major cell types: adenocarcinoma, squamous cell carcinoma, large cell carcinoma and small cell carcinoma account for 95% of cases. Of the non-small cell carcinomas, adenocarcinoma accounts for 30–35% of cases. Its relative incidence is rising; it is now the predominant histological subtype in many countries4. The relative incidence of bronchiolo-alveolar carcinoma, which is a subtype of adenocarcinoma, is also increasing as smoking declines. In comparison, the relative incidence of squamous cell carcinoma (30–35% of cases) is decreasing. Large cell carcinoma accounts for 10–15% of cases. Small cell carcinoma accounts for 20–30% of cases.

Clinical presentation Clinical features vary with cell type5 and extent of disease. Approximately 25% of patients are asymptomatic at the time of diagnosis, following the discovery of an abnormality on chest radiograph or computed tomography (CT). Pneumonia is the other common presentation. Cough, wheeze, haemoptysis, symptoms of pneumonia and paraneoplastic syndromes,

Late presentation is one of the factors that have contributed to the lack of significant improvement in survival rates over the past 30 years, despite advances in detection methods and treatments. Median survival from diagnosis has remained at 6–12 months and overall 5-year survival is poor, approximately 5–15%6,7. No study has described definite benefit from screening for lung cancer. Recent large CT studies have demonstrated increased lung cancer detection in at-risk populations but many questions remain unanswered. Most importantly, there is no current evidence that screening confers a disease-specific survival benefit. In an effort to provide definitive evidence, a large-scale trial (National Lung Cancer Screening Trial [USA]) began in 2002 and is due to continue until 2009, after which the findings will be reported. This is a study of overall and comparative utility of chest radiography and spiral CT for lung cancer screening and has recruited 50 000 patients.

Imaging techniques Although lung cancers are frequently detected on chest radiograph, the chest radiograph is of limited use in the evaluation and staging of lung cancer. CT is a more sensitive test for the detection of lung cancer, and is the imaging investigation most widely used to evaluate the primary tumour and the extent of intrathoracic and regional extrathoracic disease. Magnetic resonance imaging (MRI) may provide additional information in selected cases when mediastinal or chest wall invasion is suspected8. Positron-emission tomography (PET) with 18F-fluorodeoxyglucose (18F-FDG) has become a very useful tool in lung cancer staging. It has been found to increase the accuracy of preoperative staging, particularly when used in conjunction with CT (PET–CT)9.

326

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Imaging features The thoracic imaging features of bronchial carcinoma are discussed under three headings: peripheral tumours; central tumours (arising in a large bronchus at or close to the hilum); and staging intrathoracic spread of bronchial carcinoma.

Peripheral tumours Approximately 40% of bronchial carcinomas arise beyond the segmental bronchi, and in 30% a peripheral mass is the sole radiographic finding10 (Fig. 18.1). Tumour shape and margins Tumours at the lung apex (Pancoast’s tumours, superior sulcus tumours) may resemble apical pleural thickening; however, the majority of peripheral lung cancers are approximately spherical or oval in shape. Lobulation, a sign that indicates uneven growth rates in different parts of the tumour, is common. Occasionally, a dumb-bell shape is encountered or two nodules are seen next to one another. The term corona radiata is used to describe numerous fine strands radiating into the lung from a central mass, sometimes with transradiant lung parenchyma between these strands11. While not specific, this sign is highly suggestive of bronchial carcinoma (Fig. 18.2). Absolutely spherical, sharply defined, smooth-edged nodules due to carcinoma of the lung are rare. A peripheral line shadow or ‘tail’ may be seen between a peripherally located mass lesion and the pleura12, a phenomenon that occurs in both benign and malignant lesions. When associated with carcinoma of the lung, the ‘tail’ probably represents either plate-like atelectasis secondary to bronchial obstruction beyond the mass, or septal oedema due to lymphatic obstruction. Although the edges of a tumour are frequently well defined, some peripheral cancers, notably adenocarcinoma and bronchiolo-alveolar carcinoma, have ill-defined edges similar to pneumonia (Fig. 18.3). Cavitation may be identified in tumours of any size (Fig. 18.4) and is best demonstrated by CT (Fig. 18.5). Squamous cell

Figure 18.1 Bronchial carcinoma in the left lower lobe showing typical rounded, slightly lobular configuration. The mass shows a notch posteriorly.

carcinoma is the most likely cell type to show cavitation. The walls of the cavity are of irregular thickness and may contain tumour nodules, but sometimes the wall has smooth inner and outer margins. The cavity wall is usually 8-mm thick or greater. Fluid levels are common. Calcification within bronchogenic carcinomas is rarely seen on chest radiograph but is identified on CT in 6–10% of cases13,14. Some foci of calcification represent pre-existing calcified granulomatous disease engulfed by tumour (Fig. 18.6). However, amorphous or cloudlike calcification consistent with dystrophic tumour calcification is seen in a significant proportion (Fig. 18.7). Most calcified tumours are large with a diameter of 5 cm or more, but calcification can also be seen in small peripheral tumours. Other findings Air bronchograms and bubble-like lucencies or pseudocavitation may be seen within lung cancers, in particular with bronchiolo-alveolar carcinoma and adenocarcinoma15. Occasionally, dilated mucus-filled bronchi (bronchocele, mucocele, mucoid impaction) are seen distal to a carcinoma obstructing a segmental or subsegmental bronchus16. Ground-glass attenuation may be seen as a component of nodules and is associated with a greater risk of malignancy than that of purely solid nodules. It is more commonly associated with bronchiolo-alveolar carcinoma17, which may present as a purely ground-glass opacity.

Central tumours The cardinal imaging signs of a central tumour are collapse/ consolidation of the lung beyond the tumour and the presence of hilar enlargement, signs that may be seen in isolation or in conjunction with one another. Collapse/consolidation in association with central tumours Obstruction of a major bronchus often leads to a combination of atelectasis and retention of secretions with consequent

CHAPTER 18

Figure 18.2 CT demonstrating a second primary bronchogenic carcinoma in the right lung in a patient who had undergone a previous left pneumonectomy 7 years earlier. The new tumour has spiculated edges infiltrating into the adjacent lung (corona radiata).

•

PULMONARY NEOPLASMS

pulmonary opacity18, but collateral air drift may partially or completely prevent these postobstructive changes. Secondary infection may occur beyond the obstruction. The following features suggest that pneumonia is secondary to an obstructing neoplasm: 1 The shape of the collapsed or consolidated lobe may be altered because of the bulk of the underlying tumour. In cases with lobar collapse due to a central tumour mass, the fissure in the region of the mass is unable to move in the usual manner and, therefore, the fissure may show a bulge (the Golden S sign) (Fig. 18.8). 2 The presence of pneumonia in an at-risk patient, confined to one lobe (or more lobes if there is a common bronchus supplying these lobes) that persists unchanged for longer than 2–3 weeks, or a pneumonia that recurs in the same lobe, particularly if the lobe shows loss of volume and no air bronchograms. Simple pneumonia often clears or spreads to other segments within a few weeks. In practice,

Figure 18.3 (A) Squamous cell carcinoma resembling pneumonia. The entire opacity seen on this radiograph is due to the carcinoma itself. (B) Apical bronchiolo-alveolar cell carcinoma of the left upper lobe with ground-glass attenuation margins.

Figure 18.4 Examples of neoplastic cavitation on chest radiography. (A) The cavity is eccentric (large cell undifferentiated carcinoma). (B) The inner wall of the cavity is irregular and an air–fluid level is present (squamous cell carcinoma). (C) The cavity wall is very thin (squamous cell carcinoma).

327

328

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 18.5 CT showing cavitating squamous cell carcinoma. The wall of the cavity is variable in thickness.

complete resolution of pneumonia virtually excludes an obstructing neoplasm as a cause of infection. Although consolidation may improve partially on appropriate antibiotic therapy, it almost never resolves completely if secondary to an underlying carcinoma. Occasionally, the opacified lobe will appear larger than normal because of the build-up of infected secretions beyond the obstructing carcinoma, an appearance that has been labeled the ‘drowned lobe’. 3 The presence of a visible mass or irregular stenosis in a mainstem or lobar bronchus. Careful analysis of CT images will demonstrate the presence of an obstructing tumour in virtually every case of postobstructive atelectasis due to a lung carcinoma19. 4 Simple pneumonia rarely causes radiographically visible hilar adenopathy, though enlarged central nodes may be seen on

Figure 18.6 Calcified infectious granuloma engulfed by lung cancer. CT shows a cluster of densely calcified small nodules almost at the centre of a small carcinoma.

Figure 18.7 Tumour calcification. Large bronchial carcinoma in left lower lobe showing extensive amorphous and cloud-like calcification. Initial examination; no treatment had been given.

CT or MRI. Lung abscess can occasionally be confused with bronchial carcinoma because it may result in hilar or mediastinal adenopathy20. 5 Mucus-filled dilated bronchi may be visible within collapsed lobes on a CT examination as branching, tubular low-density structures, and when seen should prompt a search for a centrally obstructing tumour (Fig. 18.9). Hilar enlargement is a common presenting feature in patients with bronchial carcinoma. It may reflect a proximal tumour, lymphadenopathy, consolidated lung, or a combination of these phenomena21–24. In general, the more lobular the shape, the more likely that metastatic lymphadenopathy is present. A mass superimposed on the hilum may lead to increased density of the hilum owing to summation of the opacity of the mass and that of the normal hilar shadows (Fig. 18.10).This sign may be the only indication of lung cancer on a frontal chest radiograph; when suspected, it is essential to inspect a lateral radiograph with care. Radiographic patterns based on cell type The radiographic pattern of bronchial carcinoma varies to a degree with the cell type, which may aid in the differential diagnosis prior to obtaining histological confirmation25. Early, often massive, hilar or mediastinal lymphadenopathy (Fig. 18.11) and direct mediastinal invasion are well-recognized phenomena in both small cell carcinoma and large cell carcinoma. Adenocarcinoma frequently shows hilar and mediastinal adenopathy26, though the nodal enlargement is not as massive as it is with small cell and large cell undifferentiated tumours. A mass in, or adjacent to, the hilum is a particular characteristic of small cell carcinoma, seen in 78% of cases. A peripheral nodule is very common in adenocarcinoma (72% of cases) and large cell tumours (63% of cases); it occurs approximately twice as often as with squamous or small cell carcinomas. The largest peripheral masses are seen with squamous and large cell tumours, whereas most adenocarcinomas and small cell carcinomas are less than 4 cm in diameter.

CHAPTER 18

•

PULMONARY NEOPLASMS

Figure 18.8 Lobe collapse. (A) Collapse of a lobe around a central mass. (B) The middle lobe has undergone collapse, but there is a central mass causing the central portion of both the oblique and horizontal fissures to bulge outwards (arrows).

Squamous cell cancers may attain great size and they cavitate more frequently than the other cell types; in one series cavitation was seen in 12% of squamous cell carcinomas presenting as a peripheral mass, compared with only 4–6% of peripheral large cell and peripheral adenocarcinomas27. Collapse/consolidation of the lung beyond the tumour is the most frequent feature seen with squamous cell carcinoma, in keeping with the predominantly central origin of this form of neoplasm. Pleural effusion (with dyspnoea) is a feature of adenocarcinoma.

Figure 18.9 Fluid-filled dilated bronchi beyond a central obstructing carcinoma are visible in this collapsed and consolidated left lower lobe.

Bronchiolo-alveolar carcinomas arise from the alveoli and the immediately adjacent small airways. They therefore present as peripheral pulmonary opacities rather than with the effects of large airway obstruction. The most common radiographic finding28 is a solitary lobulated or spiculated pulmonary mass indistinguishable from other types of carcinoma. Bubble-like lucencies corresponding to patent small bronchi, air-containing cystic lucencies, or air bronchograms and cavitation may be seen29,30. Bronchiolo-alveolar carcinomas

Figure 18.10 Dense hilum. The right hilum is dense owing to a mass superimposed directly over it. The mass proved to be a squamous cell carcinoma.

329

330

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 18.11 Massive mediastinal adenopathy in a patient with small (oat) cell carcinoma of the bronchus. The primary carcinoma is not visible because it lies centrally in the bronchial tree.

may also appear as an ill-defined opacity resembling a patch of pneumonia; homogeneous consolidation of one lobe (which may be expansile); patchy consolidation; atelectasis; or multiple ill-defined nodules spread widely through multiple lobes in one or both lungs (Figs 18.12–18.14). Less commonly, a lepidic (scale-like) growth pattern or focal ground-glass opacity is seen31.

Figure 18.13 Bronchiolo-alveolar carcinoma. (A) Bronchiolo-alveolar carcinoma with widespread lung involvement. The appearance closely resembles bronchopneumonia or pulmonary oedema. (B) CT of a similar case showing the typical airspace filling with an obvious air bronchogram.

Staging intrathoracic spread of tumour

Figure 18.12 Bronchiolo-alveolar carcinoma occupying the right lower lobe. The appearance is identical to consolidation with partial collapse. Air bronchograms are present.

Lung cancer staging provides information about the anatomical extent and histological nature of disease, which allows the most appropriate treatment to be planned and gives an indication of prognosis. In those in whom surgery is not deemed appropriate, assessment of disease burden aids the oncologist to plan radiotherapy and chemotherapeutic regimens, and assists in the assessment of response to therapy. Small cell lung cancer is usually disseminated at the time of diagnosis. It is almost always treated medically, with the role

CHAPTER 18

•

PULMONARY NEOPLASMS

Figure 18.14 Bronchiolo-alveolar carcinoma. (A) Bronchiolo-alveolar carcinoma demonstrating lepidote growth in the left upper lobe. CT obtained during fine needle aspiration biopsy. (B,C) Biopsy-proven bronchiolo-alveolar cell carcinoma presenting as diffuse consolidation and ground-glass shadowing on chest radiography (B) and CT (C).

of imaging largely being to determine the extent of intrathoracic (limited versus extensive) and extrathoracic disease for the purposes of treatment planning. The International Staging System for Lung Cancer7 uses the TNM system to describe the findings (Table 18.1), and the stage is derived from the TNM description (Table 18.2) (T

signifies the primary tumour, N the regional lymph nodes and M distant metastases.) The system is used for non-small cell lung cancers. In essence: • Following the 1997 revision of the International Staging System for Lung Cancer7 Stage I has been divided into IA and IB based on demonstrable survival differences between

Table 18.1 TNM DEFINITIONS FOR THE INTERNATIONAL STAGING SYSTEM FOR LUNG CANCER7 Primary tumour (T) TX

Tumour proved by the presence of malignant cells in bronchopulmonary secretions but not visualized radiographically or bronchoscopically, or any tumour that cannot be assessed as in a retreatment staging

T0

No evidence of primary tumour

T1S

Carcinoma in situ

T1

A tumour that is 3 cm or less in its greatest dimension, surrounded by lung or visceral pleura and without evidence of invasion proximal to a lobar bronchus at bronchoscopy* (i.e. not in the main bronchus)

T2

A tumour more than 3 cm in its greatest dimension, or a tumour of any size that either invades the visceral pleura or has associated atelectasis or obstructive pneumonitis extending to the hilar region. At bronchoscopy the proximal extent of demonstrable tumour must be within a lobar bronchus or at least 2 cm distal to the carina. Any associated atelectasis or obstructive pneumonitis must involve less than an entire lung

T3

A tumour of any size with direct extension into the chest wall (including superior sulcus tumours), diaphragm, or the mediastinal pleura or pericardium without involving the heart, great vessels, trachea, oesophagus, or vertebral body; or a tumour in the main bronchus within 2 cm of the carina without involving the carina

T4

A tumour of any size with invasion of the mediastinum or involving the heart, great vessels, trachea, oesophagus, vertebral body, or carina, or presence of malignant pleural effusion**, or with satellite tumour nodule(s) within the ipsilateral primary tumour lobe of the lung

Nodal involvement (N) NX

Regional lymph nodes cannot be assessed

N0

No demonstrable metastasis to regional lymph nodes

N1

Metastasis to lymph nodes in the peribronchial or the ipsilateral hilar region, or both, including direct extension

N2

Metastasis to ipsilateral mediastinal lymph nodes and subcarinal lymph nodes

N3

Metastasis to contralateral mediastinal lymph nodes, contralateral hilar lymph nodes, ipsilateral or contralateral scalene, or supraclavicular lymph nodes

Distant metastasis (M) MX

Presence of distant metastases cannot be assessed

M0

No (known) distant metastasis

M1

Distant metastasis present—specify sites

*The uncommon superficial tumour of any size with its invasive component limited to the bronchial wall, which may extend proximal to the main bronchus, is also classified T1. **Most pleural effusions associated with lung cancer are due to tumour. However, there are a few patients in whom multiple cytopathological examinations of the fluid reveal no tumour. In these cases the fluid is non-bloody and is not an exudate. When these elements and clinical judgement dictate that the effusion is not related to the tumour, the effusion should be excluded as a staging element and the patient’s disease should be staged T1, T2, or T3. Pericardial effusion is classified according to the same rules.

331

332

SECTION 2

Table 18.2

• THE CHEST AND CARDIOVASCULAR SYSTEM

STAGING GROUPING. TNM SUBSETS7

Stage

TNM subset

0

Carcinoma in situ

IA

T1N0M0

IB

T2N0M0

IIA

T1N1M0

IIB

T2N1M0

of collapsed lung relative to tumour on contrast-enhanced CT may assist; collapsed lung tends to enhance to a greater extent than the adjacent tumour32, although this is not always the case32–34.T1 and T2 tumours are the most amenable to surgical resection using standard techniques. T3 tumours that involve the chest wall or mediastinum to a limited extent may also be candidates for surgical resection, albeit with more complex techniques. T4 tumours are irresectable.

T3N0M0 IIIA

T3N1M0 T1N2M0 T2N2M0 T3N2M0

IIIB

T4N0M0 T4N1M0 T4N2M0 T1N3M0 T2N3M0 T3N3M0 T4N3M0

IV

Any T Any N M1

*Staging is not relevant for occult carcinoma, designated TXM0N0. For each stage, the prognoses, or estimated 5-year survival rates, in Europe are as follows:

• Stage IA—60% • Stage IB—38% • Stage IIA—34% • Stage IIB—24% • Stage IIIA—13% • Stage IIIB—5% (Stage IIIB and IV lesions are non-resectable.) • Stage IV—< 1%

patients with T1N0M0 lesions at presentation compared with T2N0M0 lesions. Both these lesions are resectable with a reasonable hope of cure. • For similar reasons, Stage II has also been subdivided into IIA and IIB. These lesions are the same T lesions as stage I but with hilar node involvement or resectable mediastinal/ chest wall invasion. They are resectable for potential cure but with less good prognosis than stage I lesions. • Stage III is divided into IIIA, in which there is locally extensive intrathoracic disease and/or hilar and ipsilateral lymph node involvement which may be surgically resectable, and IIIB where the intrathoracic disease is beyond the limits of conventional surgical resection. IIIB tumours may be considered localized in terms of planning radiotherapy. • Stage IV includes all patients with distant metastatic disease. Staging the primary tumour CT is the most commonly used tool in evaluation of the primary tumour. Defining the primary tumour in terms of T staging enables prediction of resectability.Tumour size, location, margins and relationship to adjacent structures should be described. Assessment of tumour size may not be straightforward as distinction from adjacent collapsed lung may be impossible. Differential enhancement

Mediastinal invasion Plain radiograph evidence of mediastinal invasion relies on demonstrating phrenic nerve paralysis. Caution is needed, however, before deciding that a high hemidiaphragm is caused by phrenic nerve invasion, because lobar collapse can also lead to elevation of a hemidiaphragm, a subpulmonary effusion may mimic it, and diaphragmatic eventration is common. The major CT and MRI signs of mediastinal invasion include the demonstration of visible tumour deep within the mediastinal fat, particularly if tumour surrounds the mediastinal vessels, oesophagus, or proximal mainstem bronchi (Figs 18.15, 18.16). Associated pneumonia or atelectasis may make it very difficult to determine whether or not mediastinal contact is present. Even clear-cut contact with the mediastinum is not enough for the diagnosis of invasion, and the apparent interdigitation of tumour with mediastinal fat can be a misleading sign on both CT and MRI. Glazer et al35 showed that the presence of (A) less than 3 cm of contact with the mediastinum; (B) less than 90 degrees of circumferential contact with the aorta; or (C) a visible mediastinal fat plane between the mass and any vital mediastinal structures indicated a very high likelihood of technical resectability, even if the tumour had crossed into the mediastinum, and that most tumours in their series conforming to this description had no mediastinal invasion at surgery. When the question is turned round to enquire as to the criteria for irresectability, however, the answer is less certain36–39. Tumours that obliterate fat planes or show greater contact than that described above are not necessarily irresectable, though the greater the degree of invasion and the extent of contact, the more likely it is that there is significant mediastinal involvement37. MRI does not appear to offer any advantages over CT for the routine diagnosis of mediastinal invasion, its role being limited to problem solving in specific cases (Fig. 18.16)8. Before the advent of multidetector CT (MDCT), the multiplanar capabilities of MRI (Fig.18.17) could be used to advantage to identify involvement of major mediastinal blood vessels8,40 and the tracheal carina. MDCT has largely obviated the need to proceed to MRI to take advantage of multiplanar imaging alone; however, MRI sequences optimized for evaluation of the heart and vessels may still offer advantages where there is concern about invasion of hilar or mediastinal vessels41, the heart or pericardium. Chest wall invasion The presence of chest wall invasion alone does not preclude surgical resection, though it does adversely affect prognosis42,43. The necessarily more extensive surgery is associated with increased morbidity and mortality and it therefore helps the surgeon to know the extent of any chest wall invasion pre-operatively.

CHAPTER 18

Figure 18.15

•

PULMONARY NEOPLASMS

(A) Extensive deep mediastinal invasion by primary bronchial carcinoma. (B) On lung windows there are pulmonary metastases.

Figure 18.16 MRI of involved mediastinal nodes in a patient with a right lower lobe non-small cell lung cancer.

Figure 18.17 MRI of a left lower lobe tumour that has directly invaded the aortic wall which has altered signal adjacent to the tumour.

The diagnosis of chest wall involvement adjacent to a tumour is unreliable on CT, unless there is clear-cut bone destruction or a large soft tissue mass38–40,42–45 (Fig. 18.18). Local chest wall pain remains the single most specific indicator of whether or not the tumour has spread to the parietal pleura or chest wall44. Contact with the pleura on CT examination, even if the pleura is thickened (see Fig. 18.21), does not necessarily indicate invasion, though the greater the degree of contact and the greater the pleural thickening, the more likely it is that the parietal pleura has been invaded, particularly if the extrapleural fat plane is obliterated45. A definite extrapleural mass that is not explicable by previous chest trauma is likely to be the result of invasion by tumour45, but even this sign may be misleading since soft tissue swelling may be due to inflammation and fibrosis rather than neoplasm46. Conversely, a clear extrapleural fat plane adjacent to the mass may be helpful, but again not definitive, in excluding chest wall invasion47 (Fig. 18.19). In selected cases MRI has proved to be better than CT in demonstrating chest wall48,49 and diaphragmatic invasion. MRI is regarded as the optimal modality for demonstrating the extent of superior sulcus tumours (Pancoast’s tumour) (Fig. 18.18), reliably diagnosing mediastinal invasion, extension into the root of the neck and involvement of vascular and neural structures. Transthoracic ultrasound can identify chest wall invasion with a high degree of accuracy; however, the technique is rarely used for this purpose in Europe or the USA50. 99m Tc radionuclide skeletal scintigraphy is a sensitive technique with which to assess bone invasion and it may be positive when the plain radiograph still shows no bony abnormality. Intrathoracic lymph node metastases The American Thoracic Society’s description of nodal stations is the most widely used system to define the anatomical position of lymph nodes. Lung cancers normally spread to ipsilateral hilar nodes, then

333

334

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 18.18 Chest wall invasion by a Pancoast’s tumour. Involvement of the soft tissues of the chest wall is appreciated on the (A) coronal T1- and (B) T2-weighted MRI images. (C) This example from a different patient shows the better demonstration of bone involvement (arrows) on CT.

ipsilateral mediastinal, contralateral mediastinal and supraclavicular nodes. Though nodal spread is most often sequential, skip metastases to mediastinal nodes in the absence of hilar nodes is seen in 33% of cases51. Generally chest radiography is insensitive for nodal staging. However, the presence of enlarged hilar or paratracheal nodes has been shown to be specific (92%) for N2–N3 disease8. Lymph node assessment on CT and MRI is limited to size, shape and location, with size being the major criterion used to predict metastatic involvement (Fig. 18.20). Normal mediastinal lymph node size on CT or MRI varies according to the location of the nodes within the mediastinum, but a simple and reasonably accurate rule is that nodes with a short axis diameter of less than 10 mm fall within the 95th percentile and nodes above this size should, therefore, be considered enlarged. The problem with using size as the only criterion for malignant involvement is that intrathoracic lymph node enlargement

has many nonmalignant causes, including previous tuberculosis, histoplasmosis, pneumoconiosis, sarcoidosis and, most importantly, reactive hyperplasia to the tumour (Fig. 18.21) or associated pneumonia/atelectasis: it has repeatedly been shown that one-half to two-thirds of enlarged nodes draining postobstructive pneumonia/atelectasis are free of tumour52. Conversely, microscopic involvement by tumour can be present in normal sized nodes. It will, therefore, be clear that there is no measurement above which all nodes can be assumed to be malignant and below which all can be considered to be benign. The sensitivity and specificity of CT for diagnosing metastatic involvement of mediastinal lymph nodes varies greatly in different published series, reflecting different size criteria and the methods used to confirm or exclude lymph node metastases. A reasonable generalization in the USA (where fungal infection is endemic) is that both sensitivity and specificity are in the 50 to low 60% range when the cut-off point for normal is a short axis diameter of 1 cm8,53,54. Better specificity figures

Figure 18.19 Cavitating bronchogenic carcinoma. There is preservation of the extrapleural fat plane at the point of contact with the chest wall. Although the pleura may be involved the chest wall is likely to be otherwise spared.

Figure 18.20 True-positive CT for metastatic lymphadenopathy. There are several enlarged nodes in the right paratracheal area. The largest measured 16 mm in its short axis diameter (arrow). The primary tumour was a bronchial carcinoma in the right lung.

CHAPTER 18

Figure 18.21 False-positive CT for metastatic mediastinal lymphadenopathy. The largest of the right paratracheal nodes (arrow) is 17 mm in its short axis diameter. This node proved to be free of malignant tumour at thoracotomy. The enlargement was due to reactive hyperplasia. None of the hilar or mediastinal nodes in this patient was involved by tumour. The primary tumour can be seen in the right lung. It shows extensive contact with the right chest wall, but no definite evidence of invasion of the chest wall on CT. At surgery there was invasion of the soft tissues of the chest wall but no spread to the ribs.

have been obtained in Europe55 and Japan56, probably because the prevalence of coincidental histoplasmosis is much lower than in the USA.The positive predictive value for nodal metastatic disease may be improved (to up to 95%) by ensuring that nodes draining the tumour are larger than nodes elsewhere in the mediastinum55. The accuracy of MRI, despite its improved contrast resolution, is limited by the same constraint as for CT of overlap of features of benign and malignant causes of node enlargement. Although it is generally considered that the MRI signal within nodes is not a useful predictor of involvement, a recent study has reported that STIR imaging produces sufficient signal

•

PULMONARY NEOPLASMS

difference between normal and pathological nodal tissue to detect metastases with 93% sensitivity and 87% specificity57. The previously cited advantage of MRI over CT in nodal detection due to its ability to distinguish small nodes from vessels without intravenous enhancement has been effectively negated by the advantages of MDCT. Endoscopic ultrasound (EUS) can be used to assess the size and morphology of, and to guide fine needle aspiration (FNA) of, aortopulmonary, subcarinal and posterior mediastinal nodes58, achieving greater sensitivity and specificities for nodal involvement than CT and PET in some series59. Ultrasound assessment (± FNA) of supraclavicular lymph nodes improves sensitivity for detection of supraclavicular lymph node involvement; its routine use has been suggested as a method to improve the accuracy of pre-operative staging60. PET imaging with fluorodeoxyglucose (FDG) is increasingly used for staging lung carcinoma, with published studies consistently demonstrating greater accuracy compared to CT and MRI in the detection of nodal disease (Fig. 18.22). Falsepositive results still occur, most commonly due to inflammation and reactive hyperplasia. In one meta-analysis of nodal staging the sensitivity of PET was 79% and specificity was 91%, compared with 60% and 70% respectively for CT61. Fused PET–CT imaging provides registration of FDG metabolic activity with the anatomical detail of CT (Fig. 18.22); it has been reported to be more accurate than PET or CT alone in staging patients with non-small cell lung cancer62–64. Decision analysis studies have shown that PET can be incorporated into the work-up of lung cancer in a cost-effective manner, with savings derived from identifying inoperable patients before thoracotomy65,66. Mediastinoscopy and mediastinotomy remain the most widely employed techniques for mediastinal lymph node sampling. They have high sensitivity and specificity for detecting malignant disease and, although invasive, are indicated prior to thoracotomy when other forms of imaging suggest nodal involvement.

Figure 18.22 Recurrent malignant right hilar lymph nodes from a small peripheral non-small cell lung cancer. (A) CT demonstrates nodes at the right hilum. (B) The PET–CT image confirms high FDG uptake in keeping with malignant involvement.

335

336

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pleural involvement may occur as a result of direct spread, lymphatic involvement, or tumour emboli. On occasion, adenocarcinoma takes the form of a sheet of lobular pleural thickening indistinguishable from malignant mesothelioma. A pleural effusion in association with a primary lung cancer designates the tumour as being T4. The exception is the few patients who have clinical evidence of another cause for the effusion (e.g. heart failure) and in whom cytology examinations of multiple pleural fluid samples are negative for tumour cells, in which case the effusion can be disregarded as a staging criterion. Attempts to characterize the nature of the pleural fluid based on density measurements at CT or signal intensities at MRI have not so far proved useful. Several studies suggest PET may have a role in the evaluation of pleural effusion in patients with lung cancer, although this requires further evaluation67–69. Summary Staging the intrathoracic extent of lung cancer is a multidisciplinary process utilizing imaging, bronchoscopy and biopsy. Chest radiography, CT and PET (where available) are currently the routine imaging procedures for assessing intrathoracic spread and determining resectability, with MRI and ultrasound reserved for specific indications. The essential points to establish when staging the intrathoracic extent of non-small cell cancers are: (A) whether or not the tumour has spread to hilar or mediastinal nodes; (B) if it has, which nodal groups are involved; (C) whether or not the tumour has invaded the chest wall or mediastinum; and (D) if it has, whether it is still potentially curable surgically. If chest radiography and CT ± PET show no evidence of spread beyond the lung (other than to ipsilateral hilar nodes) in a patient who is suitable for surgery, and in whom bronchoscopy shows the tumour to be resectable, then that patient should be offered surgical resection without further pre-operative invasive procedures. Spread to ipsilateral nodes, whilst not necessarily precluding surgical resection, has a significantly adverse effect on prognosis and if surgery is undertaken, it is performed with the understanding that 5-year survival rates are poor. The poor specificity of CT in determining nodal involvement must be appreciated. Nodal enlargement, whilst probably due to metastatic carcinoma, may also be due to coincidental benign disease, reactive hyperplasia to the presence of the tumour, or to any associated obstructive consolidation/atelectasis. Thus biopsy confirmation of neoplastic nodal involvement by mediastinoscopy, mediastinotomy, or needle aspiration is usually essential before a patient is denied surgery. Positive PET findings for nodal involvement do not obviate the need for histological confirmation of nodal involvement. However, in patients with no enlarged lymph nodes on CT and normal findings on PET, the likelihood of nodal involvement is so low that mediastinoscopy can be omitted. For lung cancers that have invaded the mediastinum or chest wall, it is important to decide whether the tumour is nevertheless resectable for possible cure, again recognizing that the prognosis will be poorer than for tumours confined to the lung. CT may show definitively that the tumour is too

extensive for resective surgery (i.e. that it is a T4 lesion)54. Alternatively CT may leave the issue in doubt and MRI may then help to solve the problem.

Extrathoracic staging of lung cancer Lung cancer is commonly associated with widespread haematogenous dissemination at the time of presentation. Sites of spread include the adrenal glands, bones, brain, liver and more distant lymph nodes. Detection of metastatic disease precludes surgical resection of the primary tumour. There is evidence to support the approach of extending the staging chest CT to include the liver and adrenals, with no further imaging being undertaken in the absence of clinical features suggesting metastatic disease. Currently, there is a lack of consensus regarding whether or not to perform more extensive extrathoracic screening in patients who otherwise have potentially operable disease. The increased availability of PET and PET–CT, with its greater sensitivity for detecting occult extrathoracic metastatic disease, may well result in a change in practice to include routine assessment for extrathoracic disease. In patients initially selected for curative resection using standard tumour staging, PET–CT has been reported to detect occult metastatic disease in 11–14% of patients, and to alter management in up to 40%61,70,71. A more detailed discussion of this topic is beyond the scope of this chapter.

Missed lung cancer Whether the failure to diagnose a lung cancer on an initial chest radiograph, seen only in retrospect, constitutes malpractice or whether such ‘misses’ are inevitable can be a difficult decision72,73. Forty-five of the 50 potentially visible primary peripheral lesions in one part of the National Cancer Institutes (NCI) screening programme had been overlooked on at least one previous chest radiograph, though it should be realized that the opacities in question were often very subtle74. Similarly, Heelan et al found that 65% of cancers had been overlooked in a yearly screening programme75. Lung cancers may also be overlooked at CT; the majority of missed cancers on CT are endobronchial or perihilar in location. If failure to diagnose a lung cancer at the first opportunity, however subtle the abnormality, were automatically to be regarded as malpractice, then radiologists would be found guilty of malpractice even though their radiographic reporting conformed to the same standards as those of a group of experienced chest radiologists. Clearly the dividing line between negligence and acceptable practice is difficult to define72. It is worth noting, however, that a legal case will sometimes hinge on the technical adequacy of the radiographs and appropriate communication of the findings, as well as on the issue of interpretation.

PULMONARY SARCOMA AND OTHER PRIMARY MALIGNANT NEOPLASMS The majority of pulmonary sarcomas in the lungs are metastases from extrathoracic primary tumours. Primary pulmonary sarcomas are rare, the most common primary forms

CHAPTER 18

being fibrosarcoma and leiomyosarcoma. Chondrosarcoma, fibroleiomyosarcoma, rhabdomyosarcoma, malignant fibrous histiocytoma, carcinosarcoma, liposarcoma and osteosarcoma are among the other sarcomas that may occasionally arise as primary airway or pulmonary tumours. All the above neoplasms present as a solitary pulmonary nodule or as a tracheal or endobronchial mass indistinguishable radiologically from bronchial carcinoma. Angiosarcomas of the pulmonary artery extend or arise intravascularly. The autoimmune deficiency syndrome (AIDS) epidemic has led to an increased number of cases of Kaposi’s sarcoma involving the lung. Kaposi’s sarcoma in the respiratory tract appears to be rare in the absence of cutaneous involvement76,77. Coincidental involvement of the tracheobronchial tree is relatively frequent but parenchymal involvement may occur in the absence of endobronchial disease. Imaging may show the disease to be focal or widespread78–80. Focal segmental or lobar opacities are usually due to the tumour itself, but endobronchial Kaposi’s sarcoma may result in atelectasis or postobstructive pneumonia81. Radiographically widespread disease is the more frequent pattern, with a tendency to perihilar predominance of linear, rounded, or reticulonodular shadowing, reflecting a bronchocentric distribution of the lesions79 (Fig. 18.23). The pulmonary opacities of Kaposi’s sarcoma do not fluctuate in severity, whereas the major differential diagnoses—pulmonary oedema and opportunistic infections—may do so. Intrathoracic hilar/mediastinal lymphadenopathy has been detected in 25–60% of cases in some series78,79,82. Pleural involvement is frequent78,83; pleural effusions are most commonly bilateral and may on occasion be large.

•

PULMONARY NEOPLASMS

Other rare malignant pulmonary neoplasms include haemangiopericytoma, pulmonary blastoma, plasmacytoma, choriocarcinoma, teratoma and Askin tumours. The most common malignant tumour of the trachea is invasion from an adjacent neoplasm, notably bronchial carcinoma. Primary malignant tracheal tumours are rare and are virtually confined to adults.The least infrequent is squamous cell carcinoma, followed by adenoid cystic carcinoma and mucoepidermoid carcinoma84,85 (Fig. 18.24). These three tumours make up over 90% of primary malignant tracheal tumours, the remaining 10% encompassing a wide variety of neoplasms, including sarcoma, lymphoma, adenocarcinoma, chondrosarcoma, plasmacytoma, small cell carcinoma and metastases. Some malignant tracheal tumours may present on imaging studies as a mural nodule with lobular or irregular contours, whereas some grow circumferentially as a stenosing lesion of variable length. All these tumours grow through the tracheal wall to produce a paratracheal mass, a feature most frequently seen with adenoid cystic carcinoma. Like bronchial carcinoid and various benign tumours, adenoid cystic carcinomas may calcify.

BENIGN PULMONARY TUMOURS Bronchial carcinoids Bronchial carcinoids are uncommon, constituting less than 5% of pulmonary tumours.The peak age at diagnosis is in the fifth decade, but the age range is wide and includes children. Two forms of bronchial carcinoid are described: typical (85– 90%) and atypical (10–15%).Typical carcinoids most commonly arise in central airways. Atypical carcinoids usually arise in the

Figure 18.23 Kaposi’s sarcoma in two patients with AIDS. (A) Plain chest radiograph showing extensive pulmonary shadowing consisting of a mixture of ill-defined rounded and bandlike shadows maximal in the perihilar regions and lower zones. (B) CT showing the peribronchial distribution of the illdefined pulmonary nodules. There is interlobular septal thickening, a feature that is also frequently identified on the chest radiograph.

337

338

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pulmonary hamartoma

Figure 18.24 Adenoid cystic carcinoma of the trachea. There is irregular polypoid tumour within the tracheal lumen.

lung periphery. Despite their classification as benign neoplasms, bronchial carcinoids can invade locally and may metastasize to hilar and mediastinal lymph nodes as well as to the brain, liver and bone. The atypical carcinoids have histological and clinical features intermediate between typical bronchial carcinoid and small cell carcinoma of the lung86, and have a poorer prognosis. Bronchial carcinoid may present with wheeze, pneumonia, or haemoptysis. Even when small, tumours may secrete adrenocorticotropic hormone (ACTH) in sufficient quantities to cause Cushing’s syndrome. Carcinoid syndrome is very rare if the tumour is still confined to the lung. Radiographic appearances vary with location of the tumour87–89. There is no lobar predilection and on rare occasions carcinoids may arise in the trachea. Bronchial carcinoids, particularly those located centrally, may calcify and occasionally ossify. Calcification is seen on CT in up to one-third of cases, but is only occasionally visible on chest radiograph90. Marked contrast enhancement may be seen on CT. Carcinoids arising in central bronchi (80–90% of cases) often show a larger mass external to the bronchus than within the lumen (‘iceberg’ lesions), and the extrabronchial component may be visible as a hilar mass (Fig. 18.25). Central lesions usually produce partial or complete bronchial obstruction, resulting in atelectasis with or without pneumonia. Central bronchial obstruction may be complicated by development of distal bronchiectasis or lung abscess. Occasionally, a bronchial carcinoid in a segmental or subsegmental bronchus may obstruct bronchial secretions, thereby causing a mucocele. Peripheral lesions (10–20% of carcinoids) present as solitary spherical or lobular nodules, 2–4 cm in diameter, with a welldefined smooth edge. Noncalcified peripheral bronchial carcinoid tumours closely resemble bronchial carcinomas, both radiologically and cytologically and are therefore frequently removed surgically in the belief that they are carcinomas.

Hamartomas are tumour-like malformations composed of an abnormal mixture of mature tissues normally found in the organ in which the tumour occurs. Pulmonary hamartomas consist predominantly of masses of cartilage with clefts lined by bronchial epithelium and may contain large collections of fat. Malignant transformation is either nonexistent or extremely rare. Pulmonary hamartomas are very occasionally multiple. A triad of pulmonary chondroma(s) (often multiple), gastric epithelioid leiomyosarcoma (leiomyoblastoma) and functioning extra-adrenal paragangliomas, known as Carney’s triad, has been reported, as has a form with just pulmonary chondromas and gastric smooth muscle tumours91,92. The age range for hamartoma is from young adulthood to old age, with presentation peaking in the seventh decade; they are only occasionally seen in children. The distribution of pulmonary hamartomas is opposite to that seen with bronchial carcinoid: 90% are peripheral and present as a solitary pulmonary nodule, while the remaining 10% arise within a major bronchus. Central lesions may lead to major airway obstruction and the features are then identical to those seen with bronchial carcinoids. On plain chest radiography93,94 the tumour is seen as a spherical or slightly lobulated, well-defined nodule, usually less than 4 cm in size, with normal surrounding lung (Fig. 18.26). Some hamartomas show calcification, which may be spotty or linear or show the characteristic ‘popcorn’ configuration associated with calcification in cartilage (Fig. 18.26). The frequency of calcification increases significantly with the size of the lesion. Popcorn calcification, if present, is virtually diagnostic of a hamartoma (the only differential diagnosis is a chondrosarcoma). Central fat density on CT is another important finding, which, if present, establishes the diagnosis95. The lesions grow slowly, usually much more slowly than carcinoma of the bronchus, and cavity formation is almost unknown.

Other benign pulmonary neoplasms Fibroma, chondroma, lipoma, haemangioma, benign clear cell tumours, neurogenic tumours, chemodectoma and granular cell myoblastoma are benign neoplasms that are occasionally encountered in the trachea, bronchi, or lungs. The plain radiograph and CT findings vary with the size and location of the tumour mass, but no features distinguish any one of these lesions from any other, and therefore the specific diagnosis has to be made histologically. They are indistinguishable radiologically from carcinoid tumour and solitary metastasis. Leiomyoma of the lung may be a solitary lesion, radiographically indistinguishable from the other benign connective tissue neoplasms. Multiple leiomyomas present as multiple discrete nodules in the lungs96. They are given a wide variety of names, including benign metastasizing leiomyoma. In women these tumours may be very slow growing metastases from a uterine leiomyoma; women with multiple pulmonary leiomyomas often have a history of previous hysterectomy for uterine fibroids.

CHAPTER 18

•

PULMONARY NEOPLASMS

Figure 18.25 Carcinoid tumour. (A) A small tumour is completely occluding the right main bronchus and causing extensive collapse in the right lung. The endoluminal component is well seen (arrows), but there is poor differentiation of the tumour from adjacent collapsed lung. (B) A well-defined perihilar carcinoid tumour (arrows) is demonstrated anterior to the artery to the right lower lobe. (C) On lung windows there is only a small band of atelectasis in the middle lobe. (D) A small peripheral carcinoid tumour indistinguishable from a number of other causes of a solitary pulmonary nodule.

Intrapulmonary teratomas are very unusual. Most are benign, though malignant lesions are occasionally encountered. Radiographically and on CT, intrapulmonary teratomas appear as lobulated masses that may show calcification or cavitation97.

to be viral in origin. Rarely, these papillomas are also present in the lung and are seen on plain chest radiography or CT as multiple, small, widely scattered and well-defined, round pulmonary nodules, frequently showing cavitation101,102.

Plasma cell granuloma of the lung (inflammatory pseudotumour) is the name given to a lesion that is presumed to be reactive inflammatory granulomatous tissue98. The age range is wide and includes children. Most patients present with an asymptomatic solitary pulmonary nodule. Cavitation and calcification have both been described.

Benign tumours of the trachea are rare. They are most frequent in children, in whom squamous papillomas are the most common type, the next most common being haemangiomas. Haemangiomas are often associated with cutaneous haemangiomas and frequently present in the first year of life with stridor; imaging examinations show them to be eccentrically located nodular masses, most often in the subglottic region.

Sclerosing haemangioma is a benign neoplasm99,100, which almost always presents as an asymptomatic solitary pulmonary mass. Calcification may be seen. Squamous papillomas of the trachea, bronchi and lungs are most commonly associated with laryngeal papillomatosis, a disease that usually commences in childhood and is believed

BENIGN LYMPHOPROLIFERATIVE DISORDERS Lymphocytic interstitial pneumonia Lymphocytic interstitial pneumonia (LIP) is an uncommon non-neoplastic lymphoproliferative disorder characterized by

339

340

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 18.26 Hamartoma of the lung. (A,B) Round, completely smooth, hamartoma in a 57 year old asymptomatic man. There is typical coarse popcorn calcification in this lesion which is unusually large.

diffuse infiltration of the pulmonary parenchymal interstitium by lymphocytes and plasma cells98. Histological differentiation between benign proliferation and low grade lymphoma can be difficult. LIP may occur as an isolated entity (it is included in the classification of idiopathic interstitial pneumonias); however, this is rare. It is more commonly seen in association with an underlying immunological abnormality such as Sjögren’s syndrome and AIDS. The main imaging findings are of bilateral areas of ground-glass opacification and cysts103.

Follicular bronchiolitis Follicular bronchiolitis, also known as diffuse lymphoid hyperplasia, is characterized by hyperplasia of bronchial mucosa associated lymphoid tissue (MALT) in relation to airways. Reticular or reticular nodular shadowing with centrilobular nodules and ground-glass opacity and occasionally bronchial wall thickening, bronchial dilatation, interlobular septal thickening and peribronchovascular airspace consolidation, is seen104.

visible intrathoracic adenopathy, whereas in the non-Hodgkin’s lymphomas, isolated pulmonary involvement is not uncommon107. If the mediastinal and hilar nodes have been previously irradiated, then recurrence confined to the lungs may be seen in both Hodgkin’s and non-Hodgkin’s lymphoma. The radiographic appearances of lung involvement in malignant lymphoma vary108,109. The usual patterns are: (A) one or more areas of pulmonary consolidation resembling pneumonia (Fig. 18.27); (B) multiple pulmonary nodules (Fig. 18.28); and, occasionally (C) miliary nodulation or reticulonodular shadowing resembling lymphangitis carcinomatosa (Fig. 18.29). The areas of pulmonary consolidation, which may contain air bronchograms, may be segmental or lobar in shape, but often they radiate from the hila or mediastinum without conforming to segmental anatomy, in keeping with the concept that extension into the lungs is by direct invasion from involved hilar or mediastinal nodes. Peripheral subpleural masses or areas of consolidation without any visible connection to enlarged nodes in the mediastinum and hila are, however, common in both Hodgkin’s disease and non-Hodgkin’s lymphoma. Very rapid increase in the size of lymphomatous deposits in the lung, so rapid that the disease may be confused with pneumonia, has been reported with high grade non-Hodgkin’s lymphoma110. Primary lymphoma of the lung (see Fig 18.27) (i.e. lymphoma isolated to the lung at initial presentation) is very uncommon, non-Hodgkin’s lymphoma of MALT type being the most frequently encountered form. These are low grade B-cell lymphomas of MALT (also called bronchus associated lymphoid tissue or BALT), which consists of mucosal lymphoid follicles located in distal bronchi and bronchioles, particularly at airway bifurcations98. The second most common primary tumour, known as angiocentric immunoproliferative lesion or lymphoid granulomatosis, is high grade and may have B- or T-cell phenotype98. Primary pulmonary Hodgkin’s disease is notably rare.

MALIGNANT LYMPHOPROLIFERATIVE DISORDERS Lymphoma Only pulmonary parenchymal involvement by lymphoma is considered in this chapter. Pulmonary parenchymal involvement can be broadly divided into that occurring in association with existing or previously treated nodal disease, and that due to primary lymphoma of the lung (Hodgkin’s or non-Hodgkin’s). Parenchymal involvement is comparatively rare at initial presentation (10–15% of cases105), but it becomes considerably more common as the disease progresses. It is particularly frequent in patients who relapse after treatment106. Involvement of the lung appears to be three times as frequent in Hodgkin’s lymphoma as it is in non-Hodgkin’s lymphoma105. In Hodgkin’s lymphoma the lung disease is almost invariably accompanied by

Figure 18.27 Primary pulmonary lymphoma. This appearance had been very slowly progressive over several years.

CHAPTER 18

•

PULMONARY NEOPLASMS

be a striking feature. A few of the lesions show cavitation, but calcification does not occur. MALT lymphomas are relatively rarely associated with pleural effusions despite contact with the pleura.

Other findings in pulmonary lymphoma

Figure 18.28 Pulmonary involvement by lymphocytic lymphoma showing multiple pulmonary masses.

The imaging features of MALT111–113 lymphomas are solitary or multifocal, round or segmental areas of pulmonary consolidation. There is no lobar predilection and the consolidations may be placed centrally or peripherally in the lung parenchyma. Air bronchograms are frequently visible and may

Lobar atelectasis caused by endobronchial lymphoma is occasionally encountered, but, somewhat surprisingly, atelectasis as a result of extrinsic compression by enlarged lymph nodes is rare, with encasement rather than obstruction being the usual pattern of disease. Pleural effusions are common except in MALT lymphoma. They are usually unilateral and accompanied by visible intrathoracic adenopathy. They frequently disappear once the mediastinal nodes have been irradiated; in such cases they are probably due to venous or lymphatic obstruction rather than neoplastic involvement of the pleura. The usual radiographic problem is in deciding whether the pulmonary abnormality is due to involvement by lymphomatous tissue, infection or a complication of therapy. It should be remembered that the pattern of pulmonary infection in patients with lymphoma is modified because they are immunocompromised hosts, owing either to their disease or, more often, to the drugs used for treating the disorder. In many instances, a biopsy is the only way to establish the precise diagnosis. Since Hodgkin’s disease is believed to spread from nodal sites, a useful guideline is that if a patient presents with Hodgkin’s lymphoma and a pulmonary opacity, but no evidence of hilar or mediastinal disease, it is more likely that the opacity represents something other than Hodgkin’s lymphoma114. A caveat here is that the patient should not previously have received radiation therapy to the mediastinum.

Leukaemia

Figure 18.29 Pulmonary involvement by non-Hodgkin’s lymphoma showing an appearance closely resembling lymphangitis carcinomatosa with widespread nodules and thickened septal lines.

The incidence of leukaemic infiltration of the lungs, mediastinal lymph nodes and pleura varies with the course of the disease. Pulmonary infiltration by leukaemic cells is found at autopsy in nearly two-thirds of patients who have leukaemia. However, provided those patients with leukostasis (see below) are considered separately, leukaemic infiltration of the lungs, though very common pathologically, is usually asymptomatic and is rarely a cause of significant pulmonary opacity on chest radiograph. When respiratory impairment is present, pulmonary infection, oedema or haemorrhage, are more likely causes of the patient’s symptoms115. Imaging features include diffuse bilateral reticulation and patterns resembling interstitial oedema; lymphangitic carcinomatosis, small nodules, ground-glass opacification and consolidation have also been described.116 Radiographically visible hilar and/or mediastinal lymph node enlargement may be present and pleural effusions are common, though it is not possible to state the cause of the effusion with any confidence. The distribution of nodal enlargement closely resembles that of the lymphomas. T-cell leukaemias may show massive mediastinal adenopathy that responds rapidly to chemotherapy or radiation treatment. Huge mediastinal masses of T-cell leukaemia may disappear within a few days following appropriate treatment.

341

342

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pleural thickening due to a mass of leukaemic cells in patients with myeloid leukaemia, so-called granulocytic sarcoma or chloroma formation (because of its green appearance), may be encountered on rare occasions117–119. Leukostasis is seen in patients with acute myeloid leukaemia with very high white blood cell counts in the order of 100 000–300 000 cells mm−3. The patients may be dyspnoeic because of the obliteration of their small pulmonary blood vessels by the leukaemic cells117. The chest radiograph may be normal or show airspace shadowing, which is probably due to pulmonary oedema rather than directly to the accumulation of leukaemic cells in the lungs120,121.

METASTASES Pulmonary metastases122 in adults are usually from breast, gastrointestinal tract, kidney, testes, head and neck tumours or from a variety of bone and soft tissue sarcomas. The basic sign of haematogenous pulmonary metastasis is one or more discrete pulmonary nodules (Fig. 18.30), usually in the outer portions of the lungs, a distribution that is most evident on CT (Fig. 18.31). The nodules are usually spherical and well defined, but they may be almost any shape and can occasionally have a very irregular edge. Such irregular edges are seen particularly with metastases from adenocarcinomas (Fig. 18.32). Cavitation is occasionally seen in pulmonary metastases; it is a particular feature of squamous cell carcinoma123. Calcification is very unusual except in osteosarcoma and chondrosarcoma. Even if the primary tumour shows calci-

Figure 18.30 Typical pulmonary metastases showing multiple, welldefined spherical nodules in the lungs. Rib metastases with associated soft tissue swelling are also present (arrows). In this case the primary tumour was a synovial cell carcinoma.

Figure 18.31 Pulmonary metastases (arrow). CT demonstrating a single peripheral metastasis (arrow). There were multiple lesions at other levels. The volume loss and scarring in the left lung is secondary to previous resection of the primary bronchogenic carcinoma.

fication, e.g. in breast and colon, visible calcification in the pulmonary metastases is rare.The rate of growth of metastases is highly variable; in some choriocarcinomas and osteosarcomas, for example, it may be explosive and double the volume of the lesions in less than 30 d124. Alternatively, metastases can remain unchanged in size for a long time, as in some cases of thyroid carcinoma125. A solitary pulmonary metastasis may be the presenting feature in a patient without a known primary tumour. However, a metastasis is a rare cause of the asymptomatic pulmonary

Figure 18.32 Irregular pulmonary metastases due to metastatic adenocarcinoma from an unknown primary. The nodules are irregular in outline. A large left pleural effusion is also present.

CHAPTER 18

nodule in patients who do not have a known extrathoracic primary neoplasm, comprising no more than 2–3% of most series. The simplest technique for diagnosing pulmonary metastases is the plain postero-anterior (PA) and lateral chest radiograph. High-kV techniques are often used routinely, since substantial portions of the lungs are obscured on low-kV radiographs by overlying structures such as the diaphragm, heart, mediastinum, hila and ribs. Such radiographs will detect most lung metastases above 1 cm in diameter. Increasing sensitivity can be obtained with CT, in particular MDCT. The increase in sensitivity for small nodules is, however, at the cost of decreasing specificity. On CT, lesions smaller than 1 cm are regularly demonstrated, together with most lesions above 3 mm in diameter. Below 1 cm and particularly below 6 mm, the differential diagnosis from granulomas due to tuberculosis, histoplasmosis, or other fungi becomes difficult. Where calcification can be identified, metastases (except from osteogenic sarcoma or chondrosarcoma) can effectively be dismissed from consideration. If the nodules are not calcified, the best that can be done in most instances is to give a statistical probability of the nodules being metastases. With a plain chest radiograph showing multiple noncalcified nodules, the probability is high, well over 90%, even in areas endemic for fungus granulomas, and approaches 100% in areas where fungus granulomas are rare or nonexistent. With the smaller lesions detectable on CT, this probability diminishes. Depending on the prevalence of infectious granulomas in the community and the likelihood of a particular tumour metastasizing to the lung, the probability that a pulmonary nodule seen solely on CT is indeed a metastasis may drop to as low as 50%. The use of chest CT to detect pulmonary metastases should be limited to selected patients. There are no universally agreed guidelines but general indications include: 1 Investigation of patients with a normal chest radiograph in whom the likelihood of metastasis is high and in whom demonstration of the presence of pulmonary metastases would significantly alter management, e.g. patients with osteosarcoma, choriocarcinoma and testicular germ-cell tumours. 2 Investigation of patients who are being considered for surgical resection of a known pulmonary metastasis to look for further occult lesions. 3 Distinction of solitary from multiple pulmonary nodules in a patient with an extrathoracic primary tumour in whom the diagnostic question is metastasis versus new primary bronchial carcinoma. A truly solitary pulmonary nodule may represent a primary bronchogenic carcinoma, whereas multiple nodules make metastases more likely. Currently CT is the most cost-effective and widely used method of screening for pulmonary metastases. PET can be useful for detection of thoracic metastases for tumours such as melanoma, colon and breast. The major weakness of PET alone is the limited sensitivity for nodules less than 10 mm in diameter.

•

PULMONARY NEOPLASMS

Lymphangitic carcinomatosis Lymphangitic carcinomatosis is the name given to permeation of pulmonary lymphatics and/or their adjacent interstitial tissue by neoplastic cells.The most common tumours that spread in this manner are carcinomas of the bronchus, breast, stomach and prostate126. Lymphangitic carcinomatosis may develop secondary to blood-borne emboli lodging in smaller pulmonary arteries and subsequently spreading through the vessel walls into the perivascular interstitium and lymphatic vessels. Such spread tends to give rise to bilateral symmetric pulmonary abnormality. Alternatively, lymphangitic carcinomatosis may result from direct extension of tumour from hilar lymph nodes into peribronchovascular interstitium, from the pleura into adjacent interlobular septa, or from a primary carcinoma of the lung into the adjacent peribronchovascular interstitium. Tumour spreading by these mechanisms tends to be more localized98. The radiological findings are fine reticulonodular shadowing and/or thickened septal lines (Figs 18.33, 18.34). These signs occur because of a combination of dilated lymphatics and interstitial oedema, together with shadows due to the tumour cells themselves along with any desmoplastic response which may have been induced by the tumour98,126. Another useful sign of lymphangitis carcinomatosa is subpleural oedema resulting from lymphatic obstruction by tumour cells, a feature that is most readily visible as thickening of the fissures. Pleural effusion is common, seen in about 30%. As would be expected, CT is more sensitive than plain radiography in the detection of lymphangitic spread and may show changes in patients whose chest radiograph is normal. CT, particularly high-resolution CT127–130, shows nonuniform, often nodular, thickening of the interlobular septa and irregular thickening of the bronchovascular bundles in the

Figure 18.33 Unilateral lymphangitic carcinomatosis due to carcinoma of the bronchus, showing thickened septal lines and nodules confined to the right lung.

343

344

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 18.34 Bilateral lymphangitic carcinomatosis showing bilateral thickened septal lines together with widespread nodulation of the lungs. The primary tumour in this 71 year old woman was presumed to be a bronchial carcinoma (a diagnosis based on sputum cytology).

Figure 18.35 High-resolution CT of lymphangitic carcinomatosis. Note the variable thickening of the interlobular septa and the enlargement of the bronchovascular bundle in the centre of the secondary pulmonary lobules. The polygonal shape of the walls (septa) of the secondary pulmonary lobules is particularly well shown anteriorly. The pulmonary nodule is due to a discrete metastasis, a relatively frequent finding in this condition.

Tumour emboli central portions of the lungs (Fig. 18.35). Small, peripherally located, wedge-shaped densities are sometimes seen as well; these may represent volume averaging of the thickened septa. There is often patchy airspace shadowing, but an important differential diagnostic feature from pulmonary oedema is that many of the acini subtended by thickened interlobular septa are normally aerated. Nodular shadows may be seen scattered through the parenchyma. The abnormalities may involve all zones of both lungs or they may be centrally or peripherally predominant; sometimes, particularly when lymphangitis is due to bronchial carcinoma, they are confined to a lobe or one lung. Hilar lymph node enlargement is seen in only some of the patients.

Unusual patterns of metastatic cancer Endobronchial metastases Endobronchial metastases are most unusual. Melanoma and renal, colorectal and breast carcinomas are the primary tumours that most frequently give endobronchial submucosal metastases131. In such cases the effect of airway obstruction is the dominant feature.

Miliary metastases Occasionally, innumerable tiny nodules closely resembling miliary tuberculosis are seen throughout both lungs, with no large masses and no evidence of lymphatic obstruction, such as is seen in lymphangitis carcinomatosa. Metastases are, however, one of the rarest causes of this pattern. The primary tumours that are most likely to produce miliary nodulation of the lungs are thyroid and renal carcinomas, bone sarcomas and choriocarcinoma.

Radiologically recognizable pulmonary arterial hypertension may occur on rare occasions as a result of tumour emboli blocking small pulmonary arteries132. Many tumours can embolize in this fashion, particularly hepatoma, carcinoma of the breast, kidney, stomach, and prostate and choriocarcinoma132.

EVALUATION OF THE SOLITARY PULMONARY NODULE A solitary pulmonary mass or nodule is defined as a solitary circumscribed pulmonary opacity with no associated pulmonary, pleural, or mediastinal abnormality, measuring less than 3 cm in diameter133. Many are discovered incidentally at chest radiography or CT. Although most are benign, up to 40% of these nodules may be malignant, with the relative incidence varying amongst different patient populations, e.g. there is a relatively higher incidence of granulomas in countries where fungal disease is endemic. Table 18.3 lists some of the causes of the solitary nodule, and Table 18.4 some of the mimics. The primary role of radiological investigation is to differentiate benign from malignant disease. It is as important to avoid a false-positive diagnosis of cancer with an unnecessary operation as to avoid a false-negative diagnosis leaving a potentially resectable cancer untreated. Chest radiography and CT are the primary imaging investigations used, with CT optimal for characterization. The following discussion reviews factors used to differentiate between the benign and malignant lesion. Of those listed below, the two primary criteria are rate of growth/stability over time and the attenuation of the nodule.

CHAPTER 18

Table 18.3

CAUSES OF A SOLITARY PULMONARY MASS

Bronchial carcinoma Bronchial carcinoid Granuloma Hamartoma Metastasis Chronic pneumonia or abscess Hydatid cyst Pulmonary haematoma Bronchocele Fungus ball Massive fibrosis in coal workers Bronchogenic cyst Sequestration Atriovenous malformation Pulmonary infarct Round atelectasis

Table 18.4

SIMULANTS OF A SOLITARY PULMONARY MASS

Extrathoracic artefacts Cutaneous masses Bony lesions Pleural tumours or plaques

•

PULMONARY NEOPLASMS

Attenuation and enhancement The attenuation of a pulmonary nodule on CT can be classified as soft tissue, calcification, fat and ground glass. A dense central nidus or laminated calcification are good indications of a granulomatous process (tuberculosis, histoplasmosis). Irregular ‘popcorn’ calcification is very suggestive of a hamartoma, but is comparatively uncommon in such tumours when they are less than 30 mm in diameter. Granular calcification on CT is seen in up to 7% of carcinomas, usually in large rather than small tumours. Sometimes the calcification represents tumour calcification, sometimes it is due to benign granulomatous calcification engulfed by a carcinoma. CT is of most value for the detection of fat, its presence being virtually diagnostic of hamartoma (present in 20–30% of cases). A lack of enhancement (< 15 HU) following an intravenous bolus of contrast medium is also indicative of benignity138 (Fig. 18.36). Recent studies of screening CT have found that mixed soft tissue and ground-glass attenuation nodules have a greater likelihood of malignancy compared to soft tissue nodules, and that groundglass attenuation nodules when malignant are more likely to be of bronchiolo-alveolar carcinoma cell type17.

Size The size of the mass is of little diagnostic value. Although only a small percentage of nodules under 1 cm in diameter are malignant, over 40% of malignant nodules are less than 2 cm and 15% are less than 1 cm in diameter15,139.

Encysted pleural fluid Pulmonary vessels

The age of the patient is also a significant distinguishing feature. Only 1% of solitary masses in patients under the age of 35 years will be carcinomas.

Margins The features of the margins of the mass are of some predictive value and are best demonstrated by high-resolution CT15. Carcinomas typically have ill-defined margins which are irregular, speculated, or lobulated and may exhibit umbilication or a notch. Unfortunately all these features can be seen with benign disease. On the other hand, a well-defined mass

Rate of growth/stability over time Considerable amounts of data have been accumulated on the growth rates of benign and malignant masses. The measure of growth is the volume doubling time. Benign lesions almost invariably have a doubling time of less than 1 month or more than 18 months, with the volume doubling time for most peripheral pulmonary carcinomas being between 1 and 18 months (median 3 months)134–136. Bronchiolo-alveolar carcinomas are an exception; they may grow slowly, with volume doubling times much longer than those usually quoted for bronchial carcinoma137. Monitoring of growth by serial radiography is now not recommended, but these data serve to underline the fact that a mass which has demonstrably not changed in size over a period of 2 years can be regarded as almost certainly benign, and emphasize the importance of obtaining old images for comparison wherever possible. Volumetric analysis tools on CT provide a more accurate determination of interval growth. Growth in and of itself does not indicate malignancy, but does increase the likelihood that a nodule is malignant.

Figure 18.36 Contrast-enhanced CT for the evaluation of a solitary pulmonary nodule. There is differential enhancement in this lesion that was due to a primary adenocarcinoma.

345

346

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

with a perfectly smooth, pencil-sharp margin is unlikely to be bronchial carcinoma, although a metastasis may occasionally exhibit these features. PET or PET–CT has been shown to have sensitivity, specificity and accuracy of 90% or greater in the diagnosis of benign nodules133. False-positive results may occur in patients with infectious or inflammatory processes, and false-negative results may occur in slow growing malignancies such as bronchiolo-alveolar carcinoma. PET also has difficulty evaluating lesions less than 10 mm in diameter. In a patient with a known extrathoracic primary malignancy the lesion should be considered as a probable metastasis. There is no place for searching for an occult extrapulmonary primary malignancy which, as the source of an isolated metastasis, accounts for only 2% of all solitary pulmonary masses. Such a search is far more fruitfully conducted after surgical excision or needle biopsy of the mass, when there will be some indication of the primary site from the histology available. Sputum cytology provides a low yield in solid, as opposed to cavitating, masses, but is easy to perform and may provide a definitive diagnosis of malignancy. Whilst percutaneous biopsy has an extremely low falsepositive rate for malignancy (well under 1%), the false-negative rate is around 10%. Other limitations of the technique are the low yield of a specific diagnosis of many benign lesions and the imperfect correlation of cell type between the cytological sample and the eventual histology. Most of these series, however, relied on FNA cytology. Improved diagnostic rates of benign pathology may be achieved with cutting needle biopsy techniques140.

Summary When a solitary pulmonary mass is detected every attempt should be made to obtain old images to check for interval change in size. The margins of the lesion and the presence or absence of calcification should be assessed. This frequently involves the use of CT which may also demonstrate specific signs of a benign lesion (atriovenous malformation, bronchocele and fungus ball). Lesions that are unchanged in size over a 2-year period may be presumed to be benign and followed up at 6-monthly intervals for a further 2 years. The presence of central or ringlike calcification also places the lesion in the benign category. Solitary pulmonary masses in patients under 40 years of age which are truly spherical with a clear-cut margin are unlikely to be malignant. Whilst it may be difficult to firmly establish the benign nature of the lesion in this way, the absence of evidence of malignancy in a satisfactory specimen allows a ‘wait and see’ policy in some patients. The other indications for biopsy are a presumed solitary metastasis; a presumed bronchial carcinoma when thoracotomy is inadvisable owing to the general health of the patient or is contraindicated because the lesion is unresectable or there are known metastases; and for the investigation of the solitary pulmonary mass occurring in the immunosuppressed patient.

REFERENCES 1. Travis W D, Colby T V, Corrin B et al 1999 World Health Organization Pathology Panel: World Health Organization. Histological typing of lung and pleural tumors: international histological classification of tumours. Springer Verlag, New York 2. Jemal A, Tiwari R C, Murray T et al 2004 Cancer statistics, 2004. CA Cancer J Clin 54: 8–29 3. Hackshaw A K, Law M R, Wald N J 1997 The accumulated evidence on lung cancer and environmental tobacco smoke. BMJ 315: 980–988 4. Travis W D, Lubin J, Ries L, Devesa S 1996 United States lung carcinoma incidence trends: declining for most histologic types among males, increasing among females. Cancer 77: 2464–2470 5. Filderman A E, Shaw C, Matthay R A 1986 Lung cancer. Part I: Etiology, pathology, natural history, manifestations, and diagnostic techniques. Invest Radiol 21: 80–90 6. Janssen-Heijnen M L, Gatta G, Forman D, Capocaccia R, Coebergh J W 1998 Variation in survival of patients with lung cancer in Europe, 1985–1989. EUROCARE Working Group. Eur J Cancer 34: 2191–2196 7. Mountain C F 1997 Revisions in the International System for Staging Lung Cancer. Chest 111: 1710–1717 8. Webb W R, Gatsonis C, Zerhouni E A et al 1991 CT and MR imaging in staging non-small cell bronchogenic carcinoma: Report of the radiologic diagnostic oncology group. Radiology 178: 705–713 9. Gilman M D, Aquino S L 2005 State-of-the-art FDG-PET imaging of lung cancer. Semin Roentgenol 40: 143–153 10. Auerbach O, Garfinkel L 1991 The changing pattern of lung carcinoma. Cancer 68: 1973–1977 11. Heitzman E R, Markarian B, Raasch B N, Carsky E W, Lane E J, Berlow M E 1982 Pathways of tumor spread through the lung: radiologic correlations with anatomy and pathology. Radiology 144: 3–14 12. Hill CA 1982 “Tail” signs associated with pulmonary lesions: critical reappraisal. Am J Roentgenol 139: 311–316 13. Mahoney M C, Shipley R T, Corcoran H L, Dickson B A 1990 CT demonstration of calcification in carcinoma of the lung. Am J Roentgenol 154: 255–258 14. Grewal R G, Austin J H 1994 CT demonstration of calcification in carcinoma of the lung. J Comput Assist Tomogr 18: 867–871 15. Zwirewich C V, Vedal S, Miller R R, Muller N L 1991 Solitary pulmonary nodule: high-resolution CT and radiologic–pathologic correlation. Radiology 179: 469–476 16. Felson B 1979 Mucoid impaction (inspissated secretions) in segmental bronchial obstruction. Radiology 133: 9–16 17. Henschke C I, Yankelevitz D F, Mirtcheva R, McGuinness G, McCauley D, Miettinen O S 2002 CT screening for lung cancer: frequency and significance of part-solid and nonsolid nodules. Am J Roentgenol 178: 1053–1057 18. Burke M, Fraser R 1988 Obstructive pneumonitis: a pathologic and pathogenetic reappraisal. Radiology 166: 699–704 19. Woodring J H 1988 Determining the cause of pulmonary atelectasis: a comparison of plain radiography and CT. Am J Roentgenol 150: 757–763 20. Rohlfing B M, White E A, Webb W R, Goodman P C 1978 Hilar and mediastinal adenopathy caused by bacterial abscess of the lung. Radiology 128: 289–293 21. Lehar T J, Carr D T, Miller W E, Payne W S, Woolner L B 1967 Roentgenographic appearance of bronchogenic adenocarcinoma. Am Rev Respir Dis 96: 245–248 22. Byrd R B, Miller W E, Carr D T, Payne W S, Woolner L B 1968 The roentgenographic appearance of small cell carcinoma of the bronchus. Mayo Clin Proc 43: 337–341 23. Byrd R B, Miller W E, Carr D T, Payne W S, Woolner L B 1968 The roentgenographic appearance of large cell carcinoma of the bronchus. Mayo Clin Proc 43: 333–336 24. Byrd R B, Carr D T, Miller W E, Payne W S, Woolner L B 1969 Radiographic abnormalities in carcinoma of the lung as related to histological cell type. Thorax 24: 573–575

CHAPTER 18

25. Byrd R B, Carr D, Miller W E et al 1964 Radiographic abnormalities in carcinoma of the lung as related to histological cell type. Thorax 166: 699–704 26. Woodring J H, Stelling C B 1983 Adenocarcinoma of the lung: a tumor with a changing pleomorphic character. Am J Roentgenol 140: 657–664 27. Theros E G 1977 1976 Caldwell Lecture: varying manifestation of peripheral pulmonary neoplasms: a radiologic–pathologic correlative study. Am J Roentgenol 128: 893–914 28. Epstein D M 1990 Bronchioloalveolar carcinoma. Semin Roentgenol 25: 105–111 29. Adler B, Padley S, Miller R R, Muller N L 1992 High-resolution CT of bronchioloalveolar carcinoma. Am J Roentgenol 159: 275–277 30. Kuhlman J E, Fishman E K, Kuhajda F P et al 1988 Solitary bronchioloalveolar carcinoma: CT criteria. Radiology 167: 379–382 31. Huang D, Weisbrod G L, Chamberlain D W 1986 Unusual radiologic presentations of bronchioloalveolar carcinoma. Can Assoc Radiol J 37: 94–99 32. Onitsuka H, Tsukada M, Araki A et al 1991 Differentiation of central lung tumour from post-obstructive lobar collapse by rapid sequence computed tomography. J Thorac Imaging 6: 28–31 33. Kono M, Adachi S, Kusumoto M, Sakai E 1993 Clinical utility of GdDTPA-enhanced magnetic resonance imaging in lung cancer. J Thorac Imaging 8: 18–26 34. Tobler J, Levitt R G, Glazer H S, Moran J, Crouch E, Evens R G 1987 Differentiation of proximal bronchogenic carcinoma from postobstructive lobar collapse by magnetic resonance imaging. Comparison with computed tomography. Invest Radiol 22: 538–543 35. Glazer H S, Kaiser L R, Anderson D J et al 1989 Indeterminate mediastinal invasion in bronchogenic carcinoma: CT evaluation. Radiology 173: 37–42 36. McLoud T C 1989 CT of bronchogenic carcinoma: indeterminate mediastinal invasion. Radiology 173: 15–16 37. Herman S J, Winton T L, Weisbrod G L, Towers M J, Mentzer S J 1994 Mediastinal invasion by bronchogenic carcinoma: CT signs. Radiology 190: 841–846 38. Armstrong P 1994 Preoperative computed tomographic scanning for staging lung cancer [editorial; comment]. Thorax 49: 941–943 39. White P G, Adams H, Crane M D et al 1994 Preoperative staging of carcinoma of the bronchus: can computed tomographic scanning reliably identify stage III tumours. Thorax 49: 951–957 40. Mayr B, Lenhard M, Fink U et al 1992 Preoperative evaluation of bronchogenic carcinoma: Value of MR in T- and N-staging. Eur J Radiol 14: 245–251 41. Ohno Y, Adachi S, Motoyama A et al 2001 Multiphase ECG-triggered 3D contrast-enhanced MR angiography: utility for evaluation of hilar and mediastinal invasion of bronchogenic carcinoma. J Magn Reson Imaging 13: 215–224 42. McCaughan B C, Martini N, Bains M S, McCormack P M 1985 Chest wall invasion in carcinoma of the lung. Therapeutic and prognostic implications. J Thorac Cardiovasc Surg 89: 836–841 43. Allen M S, Mathisen D J, Grillo H C, Wain J C, Moncure A C, Hilgenberg A D 1991 Bronchogenic carcinoma with chest wall invasion. Ann Thorac Surg 51: 948–951 44. Glazer H S, Duncan-Meyer J, Aronberg D J, Moran J F, Levitt R G, Sagel S S 1985 Pleural and chest wall invasion in bronchogenic carcinoma: CT evaluation. Radiology 157: 191–194 45. Ratto G B, Piacenza G, Frola C et al 1991 Chest wall involvement by lung cancer: computed tomographic detection and results of operation. Ann Thorac Surg 51: 182–188 46. Pearlberg J L, Sandler M A, Beute G H, Lewis J W Jr, Madrazo B L 1987 Limitations of CT in evaluation of neoplasms involving chest wall. J Comput Assist Tomogr 11: 290–293 47. Scott I R, Muller N L, Miller R R et al 1988 ResecTable stage III lung cancer: CT, surgical and pathological correlation. Radiology 166: 75–79 48. Musset D, Grenier P, Carette M F et al 1986 Primary lung cancer staging: prospective comparative study of MR imaging with CT. Radiology 160: 607–611

•

PULMONARY NEOPLASMS

49. Padovani B, Mouroux J, Seksik L et al 1993 Chest wall invasion by bronchogenic carcinoma: evaluation with MR imaging. Radiology 187: 33–38 50. Suzuki N, Saitoh T, Kitamura S 1993 Tumor invasion of the chest wall in lung cancer: diagnosis with US. Radiology 187: 39–42 51. Tateishi M, Fukuyama Y, Hamatake M, Kohdono S, Ishida T, Sugimachi K 1994 Skip mediastinal lymph node metastasis in non-small cell lung cancer. J Surg Oncol 57: 139–142 52. Webb W R, Gamsu G, Stark D D, Moore E H 1984 Magnetic resonance imaging of the normal and abnormal pulmonary hila. Radiology 152: 89–94 53. McLoud T C, Bourgouin P M, Greenberg R W et al 1992 Bronchogenic carcinoma: Analysis of staging in the mediastinum with CT by correlative lymph node mapping and sampling. Radiology 182: 319–323 54. Primack S L, Lee K S, Logan P M, Miller R R, Muller N L 1994 Bronchogenic carcinoma: utility of CT in the evaluation of patients with suspected lesions. Radiology 193: 795–800 55. Buy J N, Ghossain M A, Poirson F et al 1988 Computed tomography of mediastinal lymph nodes in nonsmall cell lung cancer. A new approach based on the lymphatic pathway of tumor spread. J Comput Assist Tomogr 12: 545–552 56. Ikezoe J, Kadowaki K, Morimoto S et al 1990 Mediastinal lymph node metastases from nonsmall cell bronchogenic carcinoma: reevaluation with CT. J Comput Assist Tomogr 14: 340–344 57. Ohno Y, Hatabu H, Takenaka D et al 2004 Metastases in mediastinal and hilar lymph nodes in patients with non-small cell lung cancer: quantitative and qualitative assessment with STIR turbo spin-echo MR imaging. Radiology 231: 872–879 58. Potepan P, Meroni E, Spagnoli I et al 1996 Non-small-cell lung cancer: detection of mediastinal lymph node metastases by endoscopic ultrasound and CT. Eur Radiol 6: 19–24 59. Fritscher-Ravens A, Davidson B L, Hauber H P et al 2003 Endoscopic ultrasound, positron emission tomography, and computerized tomography for lung cancer. Am J Respir Crit Care Med 168: 1293–1297 60. van Overhagen H, Brakel K, Heijenbrok M W et al 2004 Metastases in supraclavicular lymph nodes in lung cancer: assessment with palpation, US, and CT. Radiology 232: 75–80 61. Dwamena B A, Sonnad S S, Angobaldo J O, Wahl R L 1999 Metastases from non-small cell lung cancer: mediastinal staging in the 1990s—meta-analytic comparison of PET and CT. Radiology 213: 530–536 62. Antoch G, Stattaus J, Nemat A T et al 2003 Non-small cell lung cancer: dual-modality PET/CT in preoperative staging. Radiology 229: 526–533 63. Lardinois D, Weder W, Hany T F et al 2003 Staging of non-smallcell lung cancer with integrated positron-emission tomography and computed tomography. N Engl J Med 348: 2500–2507 64. Shim S S, Lee K S, Kim B T et al 2005 Non-small cell lung cancer: Prospective comparison of integrated FDG PET/CT and CT alone for preoperative staging. Radiology 236: 1011–1019 65. Gambhir S S, Hoh C K, Phelps M E, Madar I, Maddahi J 1996 Decision tree sensitivity analysis for cost-effectiveness of FDG-PET in the staging and management of non-small-cell lung carcinoma. J Nucl Med 37: 1428–1436 66. Scott W J, Shepherd J, Gambhir S S 1998 Cost-effectiveness of FDGPET for staging non-small cell lung cancer: a decision analysis. Ann Thorac Surg 66: 1876–1883 67. Bury T, Paulus P, Dowlati A, Corhay JL, Rigo P, Radermecker MF 1997 Evaluation of pleural diseases with FDG-PET imaging: preliminary report. Thorax 52: 187–189 68. Erasmus J J, McAdams H P, Rossi S E, Goodman P C, Coleman R E, Patz E F 2000 FDG PET of pleural effusions in patients with non-small cell lung cancer. Am J Roentgenol 175: 245–249 69. Schaffler G J, Wolf G, Schoellnast H et al 2004 Non-small cell lung cancer: evaluation of pleural abnormalities on CT scans with 18F FDG PET. Radiology 231: 858–865

347

348

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

70. Valk P E, Pounds T R, Hopkins D M 1995 Staging non-small cell lung cancer by whole-body positron emission tomographic imaging. Ann Thorac Surg 60: 1573–1582 71. Dietlein M, Weber K, Gandjour A et al 2000 Cost-effectiveness of FDGPET for the management of solitary pulmonary nodules: a decision analysis based on cost reimbursement in Germany. Eur J Nucl Med 27: 1441–1456 72. Potchen E J, Bisesi M A 1990 When is it malpractice to miss lung cancer on chest radiographs? Radiology 175: 29–32 73. Woodring J H 1990 Pitfalls in the radiologic diagnosis of lung cancer. Am J Roentgenol 154: 1165–1175 74. Muhm J R, Miller W E, Fontana R S, Sanderson D R, Uhlenhopp M A 1983 Lung cancer detected during a screening program using fourmonth chest radiographs. Radiology 148: 609–615 75. Heelan R T, Flehinger B J, Melamed M R et al 1984 Non-small cell lung cancer: results of the New York screening program. Radiology 151: 289–293 76. Lemlich G, Schwam L, Lebwohl M 1987 Kaposi’s sarcoma and acquired immunodeficiency syndrome. Postmortem findings in twenty-four cases. J Am Acad Dermatol 16: 319–325 77. Nash G, Fligiel S 1984 Kaposi’s sarcoma presenting as pulmonary disease in the acquired immunodeficiency syndrome: diagnosis by lung biopsy. Hum Pathol 15: 999–1001 78. Davis S D, Henschke C I, Chamides B K, Westcott J L 1987 Intrathoracic Kaposi sarcoma in AIDS patients: Radiographic pathologic correlation. Radiology 163: 495–500 79. Sivit C J, Schwartz A M, Rockoff S D 1987 Kaposi’s sarcoma of the lung in AIDS: radiologic–pathologic analysis. Am J Roentgenol 148: 25–28 80. Naidich D P, Tarras M, Garay S M, Birnbaum B, Rybak B J, Schinella R 1989 Kaposi’s sarcoma. CT–radiographic correlation. Chest 96: 723–728 81. Huang R M, Naidich D P, Lubat E, Schinella R, Garay S M, McCauley D I 1989 Septic pulmonary emboli: CT–radiographic correlation. Am J Roentgenol 153: 41–45 82. Zibrak J D, Silvestri R C, Costello P et al 1986 Bronchoscopic and radiologic features of Kaposi’s sarcoma involving the respiratory system. Chest 90: 476–479 83. Hannon F B, Easterbrook P J, Padley S, Boag F, Goodall R, Phillips R H 1998 Bronchopulmonary Kaposi’s sarcoma in 106 HIV-1 infected patients. Int J STD AIDS 9: 518–525 84. Gelder C M, Hetzel M R 1993 Primary tracheal tumours: a national survey. Thorax 48: 688–692 85. Allen M S 1993 Malignant tracheal tumors. Mayo Clin Proc 68: 680–684 86. Choplin R H, Kawamoto E H, Dyer R B, Geisinger K R, Mills S E, Pope T L 1986 Atypical carcinoid of the lung: radiographic features. Am J Roentgenol 146: 665–668 87. Altman R L, Miller W E, Carr D T, Payne W S, Woolner L B 1973 Radiographic appearance of bronchial carcinoid. Thorax 28: 433–434 88. Forster B B, Muller N L, Miller R R, Nelems B, Evans K G 1989 Neuroendocrine carcinomas of the lung: clinical, radiologic, and pathologic correlation. Radiology 170: 441–445 89. Nessi R, Basso R P, Basso R S, Bosco M, Blanc M, Uslenghi C 1991 Bronchial carcinoid tumors: radiologic observations in 49 cases. J Thorac Imaging 6: 47–53 90. Zwiebel B R, Austin J H, Grimes M M 1991 Bronchial carcinoid tumors: assessment with CT of location and intratumoral calcification in 31 patients. Radiology 179: 483–486 91. Carney J A 1983 The triad of gastric epithelioid leiomyosarcoma, pulmonary chondroma, and functioning extra-adrenal paraganglioma: a five-year review. Med Balt 62: 159–169 92. Mazas-Artasona L, Romeo M, Felices R et al 1988 Gastro-oesophageal leiomyoblastomas and multiple pulmonary chondromas: an incomplete variant of Carney’s triad. Br J Radiol 61: 1181–1184 93. Bateson E M, Abbott E K 1960 Mixed tumors of the lung, or hamartochondromas. A review of the radiological appearances of cases published in the literature and a report of fifteen new cases. Clin Radiol 11: 232–247 94. Poirier T J, van Ordstrand H S 1971 Pulmonary chondromatous hamartomas. Report of seventeen cases and review of the literature. Chest 59: 50–55

95. Siegelman S S, Khouri N F, Scott W W Jr et al 1986 Pulmonary hamartoma: CT findings. Radiology 160: 313–317 96. Martin E 1983 Leiomyomatous lung lesions: a proposed classification. Am J Roentgenol 141: 269–272 97. Morgan D E, Sanders C, McElvein R B, Nath H, Alexander C B 1992 Intrapulmonary teratoma: a case report and review of the literature. J Thorac Imaging 7: 70–77 98. Muller N L, Fraser R S, Lee K S, Johkoh T 2003 Diseases of the Lungs, 1st edn. Lippincott Williams & Wilkins, Philadelphia 99. Katzenstein A L, Gmelich J T, Carrington C B 1980 Sclerosing hemangioma of the lung: a clinicopathologic study of 51 cases. Am J Surg Pathol 4: 343–356 100. Sugio K, Yokoyama H, Kaneko S, Ishida T, Sugimachi K 1992 Sclerosing hemangioma of the lung: radiographic and pathological study. Ann Thorac Surg 53: 295–300 101. Kwong J S, Muller N L, Miller R R 1992 Diseases of the trachea and main-stem bronchi: correlation of CT with pathologic findings. RadioGraphics 12: 645–657 102. McCarthy M J, Rosado-de-Christenson M L 1995 Tumors of the trachea. J Thorac Imaging 10: 180–198 103. Johkoh T, Muller N L, Pickford H A et al 1999 Lymphocytic interstitial pneumonia: thin-section CT findings in 22 patients. Radiology 212: 567–572 104. Howling S J, Hansell D M, Wells A U, Nicholson A G, Flint J D, Muller N L 1999 Follicular bronchiolitis: thin-section CT and histologic findings. Radiology 212: 637–642 105. Filly R, Bland N, Castellino R A 1976 Radiographic distribution of intrathoracic disease in previously untreated patients with Hodgkin’s disease and non-Hodgkin’s lymphoma. Radiology 120: 277–281 106. Cobby M, Whipp E, Bullimore J, Goodman S, Davies E R, Goddard P 1990 CT appearances of relapse of lymphoma in the lung. Clin Radiol 41: 232–238 107. Jenkins P F, Ward M J, Davies P, Fletcher J 1981 Non-Hodgkin’s lymphoma, chronic lymphatic leukaemia and the lung. Br J Dis Chest 75: 22–30 108. Au V, Leung A N 1997 Radiologic manifestations of lymphoma in the thorax. Am J Roentgenol 168: 93–98 109. Lee K S, Kim Y, Primack S L 1997 Imaging of pulmonary lymphomas. Am J Roentgenol 168: 339–345 110. Dunnick N R, Parker B R, Castellino R A 1976 Rapid onset of pulmonary infiltration due to histiocytic lymphoma. Radiology 118: 281–285 111. King L J, Padley S P, Wotherspoon A C, Nicholson A G 2000 Pulmonary MALT lymphoma: imaging findings in 24 cases. Eur Radiol 10: 1932–1938 112. Lee D K, Im J G, Lee K S et al 2000 B-cell lymphoma of bronchusassociated lymphoid tissue (BALT) : CT features in 10 patients. J Comput Assist Tomogr 24: 30–34 113. Wislez M, Cadranel J, Antoine M et al 1999 Lymphoma of pulmonary mucosa-associated lymphoid tissue: CT scan findings and pathological correlations. Eur Respir J 14: 423–429 114. Bragg D G 1987 Radiology of the lymphomas. Curr Probl Diagn Radiol 16: 177–206 115. Maile C W, Moore A V, Ulreich S, Putman C E 1983 Chest radiographic– pathologic correlation in adult leukemia patients. Invest Radiol 18: 495–499 116. Heyneman L E, Johkoh T, Ward S, Honda O, Yoshida S, Muller N L 2000 Pulmonary leukemic infiltrates: high-resolution CT findings in 10 patients. Am J Roentgenol 174: 517–521 117. Siegel M J, Shackelford G D, McAlister W H 1981 Pleural thickening. An unusual feature of childhood leukemia. Radiology 138: 367–369 118. Neiman R S, Barcos M, Berard C et al 1981 Granulocytic sarcoma: a clinicopathologic study of 61 biopsied cases. Cancer 48: 1426–1437 119. Lee M J, Grogan L, Meehan S, Breatnach E 1991 Pleural granulocytic sarcoma: CT characteristics. Clin Radiol 43: 57–59 120. van Buchem M A, Wondergem J H, Kool L J et al 1987 Pulmonary leukostasis: radiologic-pathologic study. Radiology 165: 739–741

CHAPTER 18

121. Vernant J P, Brun B, Mannoni P, Dreyfus B 1979 Respiratory distress of hyperleukocytic granulocytic leukemias. Cancer 44: 264–268 122. Coppage L, Shaw C, Curtis A M 1987 Metastatic disease to the chest in patients with extrathoracic malignancy. J Thorac Imaging 2: 24–37 123. Chaudhuri M R 1970 Cavitary pulmonary metastases. Thorax 25: 375–381 124. Ishihara T, Kikuchi K, Ikeda T, Yamazaki S 1973 Metastatic pulmonary diseases: biologic factors and modes of treatment. Chest 63: 227–232 125. Schaner E G, Chang A E, Doppman J L, Conkle D M, Flye M W, Rosenberg S A 1978 Comparison of computed and conventional whole lung tomography in detecting pulmonary nodules: a prospective radiologic-pathologic study. Am J Roentgenol 131: 51–54 126. Janower M L, Blennerhassett J B 1971 Lymphangitic spread of metastatic cancer to the lung. A radiologic–pathologic classification. Radiology 101: 267–273 127. Johkoh T, Ikezoe J, Tomiyama N et al 1992 CT findings in lymphangitic carcinomatosis of the lung: correlation with histologic findings and pulmonary function tests. Am J Roentgenol 158: 1217–1222 128. Munk P L, Muller N L, Miller R R, Ostrow D N 1988 Pulmonary lymphangitic carcinomatosis: CT and pathologic findings. Radiology 166: 705–709 129. Ren H, Hruban R H, Kuhlman J E et al 1989 Computed tomography of inflation-fixed lungs: the beaded septum sign of pulmonary metastases. J Comput Assist Tomogr 13: 411–416 130. Stein M G, Mayo J, Muller N, Aberle D R, Webb W R, Gamsu G 1987 Pulmonary lymphangitic spread of carcinoma: Appearance on CT scans. Radiology 162: 371–375 131. Baumgartner W A, Mark J B 1980 Metastatic malignancies from distant sites to the tracheobronchial tree. J Thorac Cardiovasc Surg 79: 499–503 132. Chan C K, Hutcheon M A, Hyland R H, Smith G J, Patterson B J, Matthay R A 1987 Pulmonary tumor embolism: a critical review of clinical, imaging, and hemodynamic features. J Thorac Imaging 2: 4–14

•

PULMONARY NEOPLASMS

133. Hartman T E 2005 Radiologic evaluation of the solitary pulmonary nodule. Radiol Clin North Am 43: 459–465, vii 134. Hayabuchi N, Russell W J, Murakami J 1983 Slow-growing lung cancer in a fixed population sample. Radiologic assessments. Cancer 52: 1098–1104 135. Nathan M H, Collins V P, Adams R A 1962 Differentiation of benign and malignant pulmonary nodules by growth rate. Radiology 79: 221–232 136. Steele J D, Buell P 1973 Asymptomatic solitary pulmonary nodules. Host survival, tumor size, and growth rate. J Thorac Cardiovasc Surg 65: 140–151 137. Hill C A 1984 Bronchioloalveolar carcinoma: a review. Radiology 150: 15–20 138. Swensen S J, Morin R L, Schueler B A et al 1992 Solitary pulmonary nodule: CT evaluation of enhancement with iodinated contrast material—a preliminary report. Radiology 182: 343–347 139. Gurney J W 1993 Determining the likelihood of malignancy in solitary pulmonary nodules with Bayesian analysis. Part I. Theory. Radiology 186: 405–413 140. McLoud T C 1998 Should cutting needles replace needle aspiration of lung lesions? Radiology 207: 569–570

SUGGESTIONS FOR FURTHER READING Hansell D M, Armstrong P, Lynch D A, McAdams H P 2005 Imaging of diseases of the chest, 4th edn. Elsevier, Philadelphia Muller N L, Fraser R S, Lee K S, Johkoh T 2003 Diseases of the lung, 1st edn. Lippincott Williams & Wilkins, Philadelphia

349

CHAPTER

High-Resolution Computed Tomography of Interstitial and Occupational Lung Disease

19

David M. Hansell, Zelena A. Aziz and Nestor L. Müller

• High-resolution computed tomography patterns of interstitial lung disease • Idiopathic interstitial pneumonias • Sarcoidosis • Hypersensitivity pneumonitis • Langerhans cell histiocytosis • Lymphangioleiomyomatosis • Connective tissue diseases • Systemic vasculitides • Drug-induced lung disease • Occupational lung disease

The pulmonary interstitium is the network of connective tissue fibres that supports the lung. It includes the alveolar walls, interlobular septa and the peribronchovascular interstitium. The term interstitial lung disease (ILD) is used to refer to a group of disorders that mainly affects these supporting structures. Although the majority of these disorders also involve the airspaces, the predominant abnormality is thickening of the interstitium which may be due to the accumulation of fluid, cells, or fibrous tissue. The chest radiograph remains part of the initial assessment of ILD, but the radiographic pattern is often non-specific, observer variation is considerable1 and it is relatively insensitive to early ILD2–4. High-resolution computed tomography (HRCT) has revolutionized the imaging of ILD as it enables early detection of disease, allows a histospecific diagnosis to be made in certain cases, and provides insight into disease reversibility and prognosis.

HIGH-RESOLUTION COMPUTED TOMOGRAPHY PATTERNS OF DIFFUSE LUNG DISEASE Diffuse abnormalities of the lung on HRCT may be broadly classified into one of the following four patterns: (A) reticular

or linear; (B) nodular; (C) ground-glass opacity through to consolidation; and (D) areas of decreased lung attenuation.

Reticular pattern A reticular pattern on CT almost always represents significant ILD. Morphologically, a reticular pattern may be caused by thickened interlobular or intralobular septa or honeycomb (fibrotic) destruction. Numerous thickened interlobular septa indicate an extensive interstitial abnormality and causes include infiltration by fibrosis (interstitial fibrosis), abnormal cells (lymphangitis carcinomatosa), or fluid (pulmonary oedema). Although thickened interlobular septa can be a consequence of infiltration by fibrosis, this feature is not a frequent finding in idiopathic pulmonary fibrosis (IPF). Interlobular septal thickening is usually described as smooth (seen in pulmonary oedema and alveolar proteinosis) or irregular (lymphangitic spread of tumour), but the distinction is not always easily made. Sarcoidosis causes nodular septal thickening although this pattern is not usually the dominant feature5. Intralobular septal thickening manifests as a fine reticular pattern on HRCT and is seen in all ILDs but most commonly in IPF. Often, the intralobular septal thickening may be so fine that HRCT does not demonstrate discrete intralobular opacities but a generalized increase in lung density (ground-glass opacification). Severe pulmonary fibrosis usually results in a coarse reticular pattern made up of interlacing irregular linear opacities. The reticular pattern of end-stage fibrotic (honeycomb) lung is characterized by cystic airspaces surrounded by irregular walls. The distortion of normal lung morphology by extensive fibrosis results in irregular dilatation of segmental and subsegmental airways (traction bronchiectasis/bronchiolectasis); in the periphery of the lung, it can be difficult to distinguish dilated airways from true honeycomb change.

352

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Nodular pattern A nodular pattern is a feature of both interstitial and airspace disease.The distribution and density of nodules can help narrow what can be a lengthy differential diagnosis. Nodules within the lung interstitium, especially those related to the lymphatic vessels, are seen in the interlobular septa, subpleural and peribronchovascular regions; a distribution seen most commonly in sarcoidosis but also in lymphangitis carcinomatosa. Centrilobular nodules are seen in several conditions (Table 19.1). In particular, distinguishing between subacute hypersensitivity pneumonitis and respiratory bronchiolitis–interstitial lung disease (RB–ILD) can be difficult, because both cause relatively low density, poorly defined centrilobular nodules which may look identical on HRCT. A random distribution of very small well-defined nodules is seen in patients with haematogenous spread of tuberculosis, pulmonary metastases, pneumoconiosis and rarely in pulmonary sarcoidosis.

lung disease. In the first two processes, the decreased attenuation (‘black’) lung is abnormal; in infiltrative lung disease it is the ‘grey’ lung that is abnormal. In a study of 70 patients in whom a mosaic attenuation pattern was the dominant abnormality, Worthy et al showed that small airways disease and infiltrative lung disease were correctly identified but the mosaic attenuation pattern caused by occlusive vascular disease was frequently misinterpreted6. Bronchial abnormalities and the presence of air trapping on expiratory CT are the most useful discriminatory features in identifying small airways disease as the cause of mosaic attenuation. However, the phenomenon of hypoxic bronchodilatation in chronic occlusive vascular disease7, and the demonstration that air trapping is seen on expiration in acute pulmonary embolism8 complicates the interpretation. Nevertheless, the differentiation between the three basic causes of a mosaic attenuation pattern is usually easily made when clinical and physiological information is taken into account.

Ground-glass pattern A ground-glass pattern on HRCT is defined as a generalized increase in opacity that does not obscure pulmonary vessels. At a microscopic level, the changes responsible for ground-glass opacity are complex and include partial filling of the airspaces, considerable thickening of the interstitium, or a combination of the two. Ultimately though, the pattern of groundglass opacity on HRCT results from displacement of air from the lungs. Many conditions result in the non-specific pattern of ground-glass opacity but the most common causes include subacute hypersensitivity pneumonitis, acute respiratory distress syndrome (ARDS), acute interstitial pneumonia (AIP), non-specific interstitial pneumonia (NSIP) and diffuse pneumonias, particularly Pneumocystis jirovecii (carinii) pneumonia in patients with acquired immune deficiency syndrome (AIDS). The definite identification of dilated airways within areas of ground glass is usually an indication of fine fibrosis and thus usually indicates irreversible disease. The caveat is that, in certain entities (e.g. organizing pneumonia), dilated airways that are present within areas of ground glass in the acute setting may completely resolve following successful treatment.

Mosaic attenuation pattern The term mosaic attenuation pattern refers to regional attenuation differences demonstrated on HRCT.The attenuation of a given area of lung depends on the amount of blood, parenchymal tissue and air in that area, and thus the sign of a mosaic attenuation pattern is non-specific. It is the dominant abnormality in three completely different types of diffuse pulmonary disease: small airways disease, occlusive vascular disease and infiltrative

Table 19.1 CONDITIONS CHARACTERIZED BY PROFUSE CENTRILOBULAR NODULES IN HRCT Subacute hypersensitivity pneumonitis Respiratory bronchiolitis–interstitial lung disease Diffuse panbronchiolitis Endobronchial spread of tuberculosis or bacterial pneumonia Cryptogenic organizing pneumonia (unusual pattern)

IDIOPATHIC INTERSTITIAL PNEUMONIAS The term idiopathic interstitial pneumonia (IIP) is applied to a group of disorders with no known cause, and with more or less distinct histological and radiological appearances. Over the years, there have been additions and subtractions to the classification, but in 2001 an American Thoracic Society (ATS)/ European Respiratory Society (ERS) consensus panel consisting of clinicians, radiologists and pathologists sought to clarify the nomenclature by combining the histopathological pattern seen on lung biopsy with clinical and radiological features9. The current classification of IIPs is outlined in Table 19.2. Current guidelines recommend a multidisciplinary approach to the diagnosis of the IIPs and a recent study has demonstrated that dynamic interaction between clinicians, radiologists and pathologists improves interobserver agreement and diagnostic confidence10.

Usual interstitial pneumonia/idiopathic pulmonary fibrosis Usual interstitial pneumonia (UIP) is the most common histopathological pattern in patients with the clinical presentation of cryptogenic fibrosing alveolitis (CFA)/IPF. Under the new classification, the term IPF is exclusively reserved for patients with the idiopathic clinical syndrome associated with the morphological pattern of UIP. Other causes of a UIPtype pattern on histology include chronic hypersensitivity pneumonitis, asbestosis, connective tissue disease and rarely drugs. The pathological features of UIP are the presence of fibroblastic foci, normal areas, dense fibrosis and honeycombing; the crucial finding is of areas of fibrosis at different stages of maturity. The number of fibroblastic foci on lung biopsy is an important predictor of survival11. Classic chest radiographic features include bilateral asymmetric peripheral reticular opacities most profuse at the lung bases associated with lung volume loss, although in the presence of coexisting emphysema, lung volumes may be preserved or increased. The characteristic and virtually pathognomonic appearance of IPF on HRCT is of a predominantly subpleural bibasal

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

Table 19.2 HISTOLOGICAL AND CLINICAL CLASSIFICATION OF THE IDIOPATHIC INTERSTITIAL PNEUMONIAS Clinico–radiological–pathological criteria

Histological pattern

HRCT features

Idiopathic pulmonary fibrosis

Usual interstitial pneumonia

Reticular opacities Honeycombing Areas of ground-glass opacity associated with traction bronchiectasis

Non-specific interstitial pneumonia

Non-specific interstitial pneumonia

Areas of ground-glass opacity ± traction bronchiectasis Honeycombing minimal

Cryptogenic organizing pneumonia

Organizing pneumonia

Peripheral or peribronchial consolidation Areas of ground-glass opacity Perilobular pattern (increasingly recognized)

Acute interstitial pneumonia

Diffuse alveolar damage

Consolidation (dependent lung) Areas of ground-glass opacity Traction bronchiectasis (organizing phase)

Respiratory bronchiolitis–interstitial lung

RB–ILD

disease (RB–ILD)

Poorly defined centrilobular nodules Areas of ground-glass opacity Bronchial wall thickening Limited emphysema

Desquamative interstitial pneumonia (DIP)

DIP

Areas of ground-glass opacity Features of interstitial fibrosis

Lymphoid interstitial pneumonia (LIP)

LIP

Areas of ground-glass opacity Centrilobular nodules Thickened interlobular septa Thin-walled discrete cysts

reticular pattern within which there are areas of honeycomb destruction (Fig. 19.1)12. As the disease progresses, it often appears to ‘creep’ around the periphery of the lung to involve the anterior aspects of the upper lobes (Fig. 19.2); the finding of upper lobe irregularities (reticulation) is an important discriminator between UIP and other conditions with similar

Figure 19.1 Usual interstitial pneumonia. HRCT abnormalities predominate in the posterior, subpleural regions of the lower lobes and comprise honeycombing and traction bronchiectasis within the abnormal lung.

clinical presentations13. In the study by Hunninghake et al, the presence of both honeycombing and upper lobe irregularities increased the specificity for UIP from 69% (honeycombing alone) to 81%13. The presence of ground-glass opacification is not a dominant feature and when present, there is usually obvious traction bronchiectasis and bronchiolectasis. Mediastinal lymphadenopathy (up to 2 cm) unrelated to infection or malignancy is a frequent accompaniment14.

Figure 19.2 Usual interstitial pneumonia. In the upper lobes anteriorly there are peripheral irregular lines with areas of honeycombing.

353

354

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Studies have demonstrated that when a confident diagnosis of IPF is made on HRCT, the diagnosis is invariably correct15,16, and it has been suggested that a confident diagnosis of IPF made by experienced observers should obviate biopsy16. HRCT also has a role in predicting survival. A study by Flaherty et al suggested that patients with histological UIP who had definite or probable UIP by HRCT criteria had a worse prognosis than those who had interdeterminate HRCT findings17. The rapid development of a diffuse increase in the attenuation of lung parenchyma in patients with IPF should raise the possibility of an opportunistic infection (such as PCP), an accelerated phase of the disease (Fig. 19.3)18, or concurrent pulmonary oedema. Other complications include lung cancer19 and pulmonary tuberculosis (Fig. 19.4); the latter usually has atypical appearances on CT due to the presence of underlying lung fibrosis20.

Figure 19.4 Tuberculosis on a background of usual interstitial pneumonia. Biopsy of the area of consolidation in the right lower lobe confirmed tuberculosis.

Non-specific interstitial pneumonia NSIP is characterized by varying degrees of interstitial inflammation and fibrosis without the specific features that allow a diagnosis of UIP or desquamative interstitial pneumonia (DIP)21. While NSIP may have significant fibrosis, it is usually of uniform temporality (in comparison to UIP), and fibroblastic foci and honeycombing, if present, are scanty. Although the clinical features of NSIP resemble those of UIP, prognosis is considerably better22,23. Non-idiopathic NSIP is most often found on lung biopsy in patients with connective tissue disease and may be the pattern identified in some cases of drug-induced lung disease and hypersensitivity pneumonitis. On HRCT, NSIP is characterized by a predominant pattern of ground-glass opacification in a predominantly basal and subpleural distribution with or without associated distortion of airways (Fig. 19.5). A reticular pattern is common but honeycombing is sparse or absent24. In general, NSIP may be distinguished from UIP on CT by a more prominent component of ground-glass attenuation and a finer reticular pattern in the absence of honeycombing25. However, the variability of CT appearances reflects the heterogeneity of the pathological processes encompassed by NSIP and a confident diagnosis of NSIP based on CT alone is less readily made than in cases of UIP. Consolidation is reportedly a highly variable feature (0–98%24,25) and this discrepancy probably reflects the fact that some patients with non-idiopathic NSIP have significant amounts of histological organizing pneumonia, making classification of individual cases difficult.

Cryptogenic organizing pneumonia Figure 19.3 Usual interstitial pneumonia. HRCT performed (A) before and (B) after clinical deterioration in a patient with biopsy proven usual interstitial pneumonia. HRCT obtained during the accelerated phase of the disease demonstrates a generalized increase in lung attenuation and progression of both the reticular and honeycomb patterns.

A component of organizing pneumonia is identifiable in a variety of different contexts, including infection26, malignancy27, drug-related lung injury28 and in association with connective tissue disease29. However, in 1983, Davison et al30 described a clinicopathological entity of isolated organizing pneumonia in patients without an identifiable associ-

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

distribution is typically seen in patients with polymyositis or dermatomyositis. Ground-glass opacification, subpleural linear opacities and a distinctive perilobular pattern (Fig. 19.7)34 are also commonly encountered. The histopathological appearance of organizing pneumonia is a uniform temporal appearance of mild interstitial chronic inflammation associated with an intraluminal organizing fibrosis in distal airspaces. The lung architecture is generally well preserved. A complete response to a long (2–3 months) course of high-dose steroid treatment is the general rule, although in a minority of patients the process progresses with the incorporation of the organizing pneumonia into the alveolar walls as mature fibrosis35.

Respiratory bronchiolitis–interstitial lung disease and desquamative interstitial pneumonia

Figure 19.5 Non-specific interstitial pneumonia. The predominant abnormality is patchy, bilateral ground-glass opacification, mild reticulation and traction bronchiectasis. There is no frank honeycombing destruction.

ated disease. In 1985, Epler et al31 described the same entity and used the term bronchiolitis obliterans organizing pneumonia (BOOP). The ATS/ERS Consensus statement9 recommends that the term cryptogenic organizing pneumonia (COP) be used because it avoids confusion with airway diseases such as constrictive obliterative bronchiolitis. On a chest radiograph the most frequent feature of COP is patchy, often subpleural and basal, areas of consolidation with preservation of lung volumes. The areas of airspace consolidation have a propensity to progress and change location over time. On HRCT, consolidation corresponding to areas of organizing pneumonia is the cardinal feature found more frequently in the lower zones, with either a subpleural or a peribronchial distribution (Fig. 19.6)32,33; the peribronchial

Figure 19.6 Cryptogenic organizing pneumonia. HRCT through the upper lobes demonstrates areas of consolidation in a subpleural and peribronchial distribution in association with areas of ground-glass opacification (left upper lobe).

These two entities are considered together because of their strong association with cigarette smoking. All cigarette smokers have, to some degree, inflammation around their small airways (‘respiratory bronchiolitis’) but this is clinically unimportant and not considered further here. Patients generally present with an insiduous onset of dyspnoea and cough. The chest radiograph is relatively insensitive for the detection of RB–ILD and DIP and a normal chest radiograph has been reported in up to 20% of patients with RB–ILD36 and 25% in DIP37. On HRCT, the features of RB–ILD include areas of patchy ground-glass opacification (resulting from macrophage accumulation within alveolar spaces and alveolar ducts) and poorly defined low attenuation centrilobular nodules (Fig. 19.8). In addition, upper lobe centrilobular emphysema, usually of very limited extent and areas of air trapping, reflecting that the bronchiolitic element of this entity may be present38. Some patients show thickening of the interlobular septa and features of interstitial fibrosis, but this is unusual36. Ground-glass opacification is also the dominant feature seen in DIP (Fig. 19.9). The distribution is typically

Figure 19.7 Biopsy proven organizing pneumonia. There are poorly defined arcade-like and polygonal opacities (the perilobular pattern) in the subpleural and posterior regions of both lungs. The opacities resemble ill-defined thickened interlobular septa.

355

356

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 19.8 Respiratory bronchiolitis–interstitial lung disease. HRCT shows (A) subtle areas of ground-glass opacification and (B) ill-defined centrilobular nodules.

and Langerhans cell histiocytosis (LCH) and interstitial fibrosis, the global term smoking related-interstitial lung disease (SRILD) has been proposed to encompass DIP, RB–ILD, LCH and interstitial fibrosis (Fig. 19.10)36,40,41.

Acute interstitial pneumonia

Figure 19.9 Several areas of non-specific ground-glass opacification in the right middle lobe and both lower lobes.

lower zone, peripheral and may be patchy39. In some patients there are HRCT features of established fibrosis (in the form of architectural distortion with dilatation of some bronchi), usually of limited extent. The majority of patients with DIP or RB–ILD have a relatively stable clinical course. Smoking cessation is an important part of the management of patients but the influence of smoking on the clinical course of these patients has not been fully delineated; some patients have persistent abnormalities on HRCT even with smoking cessation and corticosteroid therapy. Because of the significant overlap between the clinical, imaging and histological features of DIP and RB–ILD and to a lesser extent between these two patterns

Acute interstitial pneumonia (AIP) can be regarded as an idiopathic form of the ARDS and is histologically (and clinically) distinct from the other interstitial pneumonias. The histological pattern seen in AIP is that of diffuse alveolar damage (DAD), which is also found in infection, connective tissue disease, drug toxicity, toxic inhalation, uraemia and sepsis. DAD has an acute exudative phase and a subsequent organizing and fibrotic phase. Lung biopsy shows diffuse involvement with temporal homogeneity, which may imply lung injury due to a single event42. The chest radiograph shows bilateral patchy airspace opacification43. HRCT demonstrates a combination of ground-glass opacification, consolidation, bronchial dilatation and architectural distortion44. Ground-glass opacification on HRCT is found in all three phases of AIP, but coexistent traction bronchiectasis probably reflects the incorporation of established fibrosis in the proliferative and fibrotic phases45. Follow-up CT studies in survivors demonstrate clearing of the ground-glass attenuation and consolidation, leaving reticular opacities consistent with residual fibrosis. Anterior nondependent fibrotic damage in survivors secondary to barotrauma has also been reported46.

Lymphoid interstitial pneumonia The term lymphoid interstitial pneumonia (LIP) was proposed by Liebow and Carrington47 to describe a disease entity characterized by a widespread interstitial lymphoid infiltrate

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

Figure 19.10 Smoking related-interstitial lung disease. Images of the (A) upper and (B) lower lobes of a 42 year old man with a 25-pack year smoking history and dyspnoea. The combination of a fine reticular pattern representing fibrosis and ground-glass opacification on a background of emphysema suggests a diagnosis of smoking related-interstitial lung disease.

of the lung, resembling lymphoma but with a clinical course more akin to a chronic interstitial pneumonia. Although in the past LIP has been considered by some to be a pulmonary lymphoproliferative disorder, evolution to frank lymphoproliferative disease is rare and thus LIP remains within the group of interstitial pneumonias9. Classically, LIP occurs in association with autoimmune diseases, most often Sjögren’s syndrome. Other diseases associated with LIP include dysproteinaemias, autologous bone marrow transplantation and viral, mycobacterial and human immunodeficiency virus (HIV) infections. Intrathoracic Castleman’s disease is frequently associated with LIP48. The incidence of LIP is approximately two-fold greater in women and symptoms of progressive cough and dyspnoea usually predominate. Common HRCT findings are nodules of varying sizes (which may be ill-defined), areas of groundglass opacification, thickened bronchovascular bundles, interlobular septal thickening and thin-walled cysts (1–30 mm) (Fig. 19.11)49,50. Airspace disease, large nodules and pleural effusions are rare in these patients. The cysts in LIP are usually discrete, are not found in clusters and are found deep within the lung parenchyma49. The cysts have been postulated to result from the lymphocytic infiltrate compressing bronchioles, causing stenosis or obstruction and subsequent postobstructive bronchiolar ectasia.

SARCOIDOSIS Sarcoidosis is a multisystem granulomatous disorder of unknown aetiology. As a consequence, the diagnosis of this syndrome is defined by the presence of characteristic clinical and radiological data along with histological evidence of noncaseating granuloma. Granulomas in the lung have a characteristic distribution along the lymphatics in the bronchovascular sheath and, to a lesser extent, in the interlobular septa and subpleural lung regions. Sarcoidosis is a disease of young

Figure 19.11 Lymphocytic interstitial pneumonitis. There is a background of ground-glass opacification and a few thin-walled cystic airspaces (the pathogenesis of these cysts is unclear).

adults, with a peak incidence in the second to fourth decades. The hilar and mediastinal nodes and the lungs are affected clinically much more commonly than any other organ or system. They are followed in decreasing order of frequency by the skin (26%), peripheral lymph nodes (22%), eyes (15%), spleen (6%), central nervous system (4%), parotid glands (4%) and bones (3%)51. Pulmonary involvement accounts for most of the morbidity and mortality associated with sarcoidosis. Sarcoidosis is traditionally staged according to its appearance on the chest radiograph: stage I, lymphadenopathy; stage II, lymphadenopathy with parenchymal opacity; stage III, parenchymal opacity alone52. Low stages at presentation are reported to have a

357

358

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

better prognosis than high stages, although the precision and clinical usefulness of such ‘staging’ is questionable.

Lymphadenopathy Sarcoidosis is characterized by bilateral, symmetrical hilar and paratracheal lymphadenopathy. Some degree of lymphadenopathy is evident on a chest radiograph in about 70–80% of patients at some time during the course of the condition. Hilar lymph node enlargement ranges from the barely detectable to the massive and gives the hila a lobulated and usually well-demarcated outline. Occasionally hilar lymphadenopathy appears to be asymmetrical or, in 1–5% of cases, may even be strictly unilateral although this is distinctly unusual53,54. Marked asymmetry that is confirmed by CT is sufficiently unusual to bring the diagnosis into question. Clinically significant compression of adjacent airways, arteries and veins is extremely unusual, even though lymph node enlargement is often massive. Paratracheal lymphadenopathy may be bilateral or unilateral and in the latter instance is usually right sided. The most common manifestation of left-sided lymphadenopathy is enlargement of the aortopulmonary window nodes—a common and characteristic feature on the chest radiograph. Other mediastinal nodes (anterior prevascular, posterior and subcarinal) are often not identified as being enlarged on the chest radiograph but on CT are seen to be affected in about half of patients55. In 90% of patients with lymphadenopathy, nodal enlargement is maximal on the first radiograph and usually disappears within 6–12 months. In about 5%, however, large nodes persist more or less indefinitely and these can be a source of confusion. Recurrence of lymphadenopathy is exceedingly rare. The lymph nodes may calcify, sometimes in a characteristic eggshell fashion. This latter feature is shared by only a few conditions (Table 19.3). Although lymph node calcification is seen on the radiograph in 5% or less of patients with sarcoidosis, it may be evident on CT in up to 40% of patients with long-standing disease56. The calcification is of variable intensity but may be relatively light, and the affected lymph nodes are usually small in volume and evenly distributed throughout the mediastinum and hila (very different from calcified nodes due to tuberculous infection which usually follow a drainage path)57. About 40% of patients presenting with nodal enlargement will develop parenchymal opacities, usually within a year, and of these about one-third will go on to have persistent (fibrotic) shadowing. Nodal enlargement does not develop after parenchymal opacities have appeared.

Table 19.3 CAUSES OF EGGSHELL NODAL CALCIFICATION Sarcoidosis Silicosis Histoplasmosis Lymphoma (postirradiation) Blastomycosis Amyloidosis

Parenchymal changes Parenchymal changes probably occur histologically in all patients but are only detected on the chest radiograph in 50–70% of cases. Characteristically, parenchymal abnormalities appear as the nodal enlargement is subsiding (in lymphoma such abnormalities tend to progress in unison).The most common radiographic pattern, seen in 75–90% of patients with parenchymal opacities, is of rounded or irregular nodules 2– 4 mm in diameter, which are usually moderately well defined. Smaller or larger opacities are not uncommon, though they rarely exceed 5 mm.Very small aggregated opacities sometimes give a ground-glass appearance. All zones tend to be affected but there is usually a mid and upper zonal predominance. The second most common pattern, seen in 10–20% of patients with parenchymal opacity, is patchy airspace consolidation. Opacities sometimes contain air bronchograms and have ill-defined margins that commonly break up into a nodular pattern. They tend to involve predominantly the peribronchovascular regions of the middle and upper lungs zones, although they may be diffuse or, occasionally, have a subpleural predominance. The parenchymal opacities described above will clear completely in about two-thirds of cases and progress to fibrosis in one-third. Permanent fibrotic shadowing is unusually coarse with a mid and upper zone predominance. The radiographic pattern consists of coarse linear opacities with evidence of volume loss and ring shadowing caused by bullae or traction bronchiectasis. Occasionally a conglomerate opacity develops resembling progressive massive fibrosis. Cor pulmonale, bullous disease with or without mycetoma formation, and pneumothorax are all recognized complications of this fibrotic stage.

High-resolution computed tomography features Parenchymal opacities are well demonstrated on HRCT (Fig. 19.12)5,58 and HRCT appearances have a high sensitivity and specificity for the diagnosis. The most consistent pulmonary parenchymal abnormality is the presence of nodular opacities (1–5 mm) distributed in a perilymphatic fashion, predominantly along the bronchovascular bundles and subpleurally and, to a lesser extent, along interlobular septa. Other findings include irregular and beaded interfaces, larger ill-defined nodules with/without an air bronchogram, patchy groundglass opacities and occasional interlobular septal thickening. In advanced disease there is evidence of fibrosis, predominantly in the perihilar regions of the middle and upper lung zones (Fig. 19.13). Air trapping is a common HRCT feature of sarcoidosis and its presence shows a good correlation with indices of small airways disease on pulmonary function tests59.The combination of a peribronchovascular, subpleural distribution, small well-defined nodules, fibrosis and a mid and upper zone distribution has been highlighted as the features most helpful in making a diagnosis of sarcoidosis. In a very small number of cases, sarcoidosis has been shown to mimic IPF with intralobular septal thickening and ground-glass opacity seen predominantly in the basal subpleural regions of the lung60. Despite the better delineation of parenchymal disease on HRCT, it is not recommended as part of the initial diagnostic work-up in patients with suspected sarcoidosis; its greatest use being

CHAPTER 19

Figure 19.12

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

Sarcoidosis. Typical HRCT features are (A) nodular opacities which (B) may become confluent, and (C) interlobular septal thickening.

They are seen most often in the context of established disease and are subacute, lasting weeks or months. The prevalence of effusions is about 2%.

Bronchial stenosis and airflow obstruction Mild large airway narrowing may be due to nodal compression, but significant lesions are usually due to intrinsic mural sarcoidosis. Stenoses may be single or multiple and particularly affect larger airways to segmental level65. Such stenoses are very rare but can cause significant airflow obstruction or atelectasis, particularly in the middle lobe. However, the functional severity of airflow obstruction seems to be largely determined by the extent of a reticular pattern, representing established fibrosis, on HRCT66.

Figure 19.13 Fibrotic sarcoidosis. There are areas of conglomerate fibrosis in a perihilar distribution with associated bronchial distortion and volume loss. The appearances superficially mimic progressive massive fibrosis seen in the pneumoconioses.

in patients who present with an atypical chest radiograph61. Previous HRCT studies have shown that areas of parenchymal consolidation and ground-glass opacity are usually reversible, whereas little resolution is identified following treatment in patients with reticulation and architectural distortion56,62. Despite this distinction between reversible and irreversible disease on HRCT, studies comparing HRCT assessment of disease activity to clinical, scintigraphic and bronchoscopic findings have yielded contradictory results63,64. Hence, HRCT is not generally used to guide prognosis in patients with sarcoidosis.

Other thoracic findings Pleural thickering and effusions Pleural thickening and effusions are unusual manifestations of sarcoidosis and do not occur in isolation. Effusions, though commonly unilateral, may be bilateral and are usually small.

HYPERSENSITIVITY PNEUMONITIS Hypersensitivity pneumonitis, also known as extrinsic allergic alveolitis, is an immunologically mediated lung disease characterized by an inflammatory reaction to specific antigens contained in a variety of organic dusts67. Common causes include avian proteins (e.g. bird breeder’s lung) and thermophilic bacteria present in mouldy hay (farmer’s lung), mouldy grain (grain handler’s lung), or heated water reservoirs (humidifier or air conditioner lung). A more comprehensive list is given in Table 19.4. These antigens reach the alveoli where they provoke an immunological reaction that includes both type III (immune complex response) and type IV (cell-mediated) mechanisms. The cell-mediated response results in a delayed hypersensitivity reaction and the presence of granulomatous inflammation within the pulmonary interstitium. Interestingly, several studies have shown that cigarette smoking has a suppressive effect that interferes with the immunopathological process that ultimately leads to hypersensitivity pneumonitis68,69. The clinical features of hypersensitivity pneumonitis are characteristic. Approximately 6 h after exposure the patient develops fever, chills, dyspnoea and cough.There is no eosinophilia, and

359

360

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Table 19.4 MAJOR CAUSES OF HYPERSENSITIVITY PNEUMONITIS Agent

Source

Disease

Aspergillus

Mouldy hay

Farmer's lung

Thermophilic actinomyces

Compost

Mushroom worker's lung

Trichosporum asahii

Tatami mats

Japanese summer-type hypersensitivity pneumonitis

Isocyanates

Paint sprays, plastics Isocyanate hypersensitivity pneumonitis

Mycobacterium avium complex

Hot tubs

Unknown

Metal working fluids Metal worker's lung

Bird proteins

Bird feathers, excrement

Hot tub lung

Bird fancier's lung

wheeze is not a prominent feature. The radiological findings are influenced by the stage of the disease. A chest radiograph taken during the acute episode can be normal70, but typical radiographic findings include diffuse ground-glass opacification and a fine nodular or reticulonodular pattern; these two features become more prominent in the subacute phase71. Between acute attacks the radiograph may return to normal and the fluctuating nature of changes on serial radiographs is highly suggestive of the diagnosis. In chronic hypersensitivity pneumonitis, fibrosis with upper lobe retraction, reticular opacity, volume loss and honeycombing may be seen. On HRCT, the nodules of hypersensitivity pneumonitis are typically poorly defined, < 5 mm in diameter72, centrilobular70 and seen throughout the lung, although a mid to lower lung zone predominance has been variably reported (Fig. 19.14)73. Ground-glass opacity is most common in the acute phase but may also be a feature of subacute and chronic hypersensitivity pneumonitis, especially if there is ongoing exposure74. A mosaic attenuation pattern is common in hypersensitivity pneumonitis; the presence of lobular areas of decreased vascularity that show air trapping on expiratory HRCT, reflecting the coexisting bronchiolitis caused by antigen deposition in the small airways (Fig. 19.15)75. The combination on HRCT

Figure 19.14 Subacute extrinsic allergic alveolitis. HRCT shows numerous poorly defined, relatively low attenuation nodules.

Figure 19.15 Hypersensitivity pneumonitis. (A) Inspiratory image shows patchy density differences reflecting both the interstitial infiltrate of subacute hypersensitivity pneumonitis and coexisting small airways disease. (B) End-expiratory image enhances the density differences revealing several secondary pulmonary lobules of decreased attenuation.

of features of infiltrative (ill-defined nodules and ground-glass opacity) and small airways disease may be remarkably similar to that seen in patients with RB–ILD; however, the distinction can usually be made with knowledge of the smoking history. Lymph node enlargement (smaller than 20 mm) has been described in both acute and subacute hypersensitivity pneumonitis14, and the presence of thin-walled lung cysts is also an occasional feature in subacute hypersensitivity pneumonitis. Cysts range in size from 3 to 25 mm and resemble those seen in lymphocytic interstitial pneumonia76, although their pathogenesis remains uncertain. Emphysema is a reported sequela of farmer’s lung and a study has demonstrated that in hypersensi-

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

tivity due to farmer’s lung, emphysema was a more prominent feature than honeycombing/fibrosis (even in never smokers) and was seen in approximately one-third of patients74. This is in comparison to pigeon breeder’s disease, where lung fibrosis is the major complication. The chronic stage of hypersensitivity pneumonitis is characterized by fibrosis, although evidence of active disease is often present. Radiological findings include intralobular and interlobular interstitial thickening, traction bronchiectasis and honeycomb destruction (Fig. 19.16)77. In some cases, there is a mid zone predominance, but the fibrotic appearance may be seen in the upper or lower lobes73. Patients with hypersensitivity pneumonitis may exhibit histological and imaging features of NSIP78,79 or UIP80,81, and thus should be considered as a differential diagnosis when either IPF or NSIP is being considered on HRCT appearances. Imaging

Figure 19.16 Chronic hypersensitivity pneumonitis. The reticular pattern with distortion of the lung parenchyma indicates established fibrosis in this case of chronic hypersensitivity pneumonitis.

features that favour hypersensitivity pneumonitis over IPF include an upper or mid zone predominance, the presence of ground-glass opacity and air trapping81.

LANGERHANS CELL HISTIOCYTOSIS Langerhans cell histiocytosis (LCH), formerly known as pulmonary histiocytosis X or eosinophilic granuloma of the lung, is a granulomatous disorder characterized histologically by the presence of large histiocytes containing rod- or racket-shaped organelles (Langerhans cells)67. The male-to-female ratio is about 4:1, and the vast majority of adult patients are cigarette smokers. In the earliest stages, patients are often asymptomatic. Others present with dyspnoea, cough, constitutional symptoms or a spontaneous pneumothorax. Pulmonary involvement is widespread, bilateral and usually symmetrical. At presentation, usually because of dyspnoea or a pneumothorax, the chest radiograph is abnormal. Typical appearances are of reticulonodular shadowing in the mid and upper zones of the lungs that are of normal or increased volume82. The nodules vary in size from micronodular to approximately 1 cm in diameter and, although histopathological examination will often demonstrate cavitation83, this feature is often difficult to appreciate on chest radiography. The classical appearances of LCH on HRCT are nodules (ranging in size from a few millimetres to 2 cm), several of which show cavitation (this feature often clinches the diagnosis) and have bizarre shapes (Fig. 19.17). At this stage of the disease, there are no obvious features of fibrosis. The distribution of disease is a useful diagnostic pointer and the typical sparing of the extreme lung bases and anterior tips of the right middle lobe and lingula is preserved even in end-stage disease84. The typical nodules of LCH85 tend to show a predictable progression through the following stages: cavitation of the nodules, thin-walled cystic lesions, and finally emphysematous and fibrobullous destruction86.

Figure 19.17 Langerhans cell histiocytosis. (A) Shows the characteristic combination of thin-walled cysts and poorly defined nodules, some of which are just beginning to cavitate. (B) Image from a patient with more advanced disease. There are numerous irregularly-shaped cysts bilaterally and a pneumothorax on the right.

361

362

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

LYMPHANGIOLEIOMYOMATOSIS Lymphangioleiomyomatosis (LAM) is a disease characterized histologically by two key features: cysts and proliferation of atypical smooth muscle cells (LAM cells) of the pulmonary interstitium, particularly in the bronchioles, pulmonary vessels and lymphatics67. LAM is a rare disease seen almost exclusively in women, the vast majority of cases being diagnosed during childbearing age. Similar pulmonary abnormalities can be seen in approximately 1% of patients with tuberous sclerosis. The most commonly described radiographic manifestation of LAM is a pattern of generalized, symmetrical, reticular, or reticulonodular opacities with normal or increased lung volumes87,88. Pleural effusions occur in 10–40% of patients89–91 (these may be unilateral or bilateral) and pneumothoraces in approximately 50% of cases.The effusions are chylous and result from involvement of the thoracic duct by the leiomyomatous tissue. The CT manifestations of LAM are distinctive, characterized by numerous thin-walled cysts randomly distributed throughout the lungs with no zonal predilection92 (Fig.19.18). Imaging features that help distinguish LAM from LCH include a more diffuse distribution of cysts typically with no sparing of the bases, more regularly shaped cysts and normal intervening lung parenchyma. Occasionally HRCT may demonstrate interlobular septal thickening88 (attributed to dilatation of lymphatic channels secondary to obstruction of pleuropulmonary lymphatics) or patchy areas of ground-glass attenuation (presumably the result of pulmonary haemorrhage)89.

in each disease separately, depending upon whether imaging, physiological, or histological criteria are used to judge involvement. Although the radiographic and HRCT appearances are not specific for any of the collagen vascular disorders, they frequently provide good corroborative evidence in substantiating what is often a difficult clinical diagnosis.

Rheumatoid disease Rheumatoid arthritis (RA) is a connective tissue disease characterized by a symmetrical inflammatory arthritis. The majority of patients have extra-articular disorders, thus the term rheumatoid disease is commonly used to emphasize the systemic nature of the disorder. RA is associated with a broad spectrum of pleural and pulmonary manifestations. Most, but not all, patients with pleuropulmonary disease have other clinical evidence of RA. In a significant minority of patients with rheumatoid disease, pleuropulmonary disease antedates the development of arthritis and in general, pleuropulmonary involvement is not related to the severity of the arthritis. The most frequently encountered manifestations of rheumatoid disease in the chest are listed in Table 19.5. Pleural involvement, either manifesting as effusions or thickening, is common. Pleural effusions can be unilateral or bilateral, are usually small or moderate in size, and the majority resolve spontaneously93.

Table 19.5 INTRATHORACIC MANIFESTATIONS OF RHEUMATOID DISEASE Pleural effusion or thickening

CONNECTIVE TISSUE DISEASES

Interstitial fibrosis (most frequently usual interstitial pneumonia type) Constrictive obliterative bronchiolitis

The connective tissue diseases form a heterogeneous group of chronic inflammatory and immunologically mediated disorders, all of which affect the lung and pleura to a variable extent and in various ways. Although the lung is a particularly vulnerable target organ, the frequency of pleuropulmonary involvement varies widely within the spectrum of disease and also

Bronchiectasis Organizing pneumonia Follicular bronchiolitis Drug-induced lung disease (methotrexate) Necrobiotic nodules/Caplan's syndrome

Figure 19.18 Lymphangioleiomyomatosis. (A) There is a profusion of thin-walled cystic airspaces scattered evenly throughout the lungs. The cysts are relatively uniform in size. (B) In a more advanced case of LAM, note the small left-sided pleural effusion.

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

ILD in RA is more common in men with seropositive disease. The most common histopathological pattern in RA-associated ILD is UIP with HRCT features that are indistinguishable from idiopathic cases; namely reticular opacities with honeycombing predominantly in the subpleural regions of the lung (Fig. 19.19)94. It is thought that the prognosis for RA–ILD is better than for idiopathic cases; Flaherty et al have demonstrated that patients with a connective tissue disease-associated UIP pattern had fewer fibroblastic foci and better survival when compared with patients with the idiopathic type95. NSIP is also seen but is less prevalent than in the other connective tissue diseases such as systemic sclerosis. Other pulmonary abnormalities seen in RA include follicular bronchiolitis, bronchiectasis (in up to 30% of cases) (Fig. 19.20)96,97, obliterative bronchiolitis (this

Figure 19.19 Rheumatoid arthritis with a usual interstitial pneumonia (UIP)-type pattern. In this case the HRCT appearances of peripheral reticular abnormality and honeycombing are indistinguishable from that of UIP.

can occur in patients who are on penicillamine, gold or no treatment)98, methotrexate-induced pneumonitis and organizing pneumonia94. Rheumatoid (necrobiotic) pulmonary nodules are an uncommon feature of the disease.They are usually associated with the presence of subcutaneous nodules, and like them may wax and wane. They may be single or multiple, vary in size from a few millimetres to several centimetres, are well circumscribed and may cavitate99.They are usually asymptomatic and may occur in association with pulmonary fibrosis and pleural changes. Radiologically identical nodules, characteristically appearing rapidly and in crops, may occur in patients with rheumatoid arthritis who have been exposed to silica. Radiographic findings of pneumoconiosis may be present but usually are not a prominent feature100. This phenomenon was originally described in Welsh coalminers (Caplan’s syndrome). These nodules contain dust particles and are quite different from necrobiotic nodules on histological examination. Follicular bronchiolitis (discussed here because of its frequent association with rheumatoid disease) is part of the spectrum of lymphoproliferative disease and is characterized histologically by a diffuse peribronchiolar proliferation of hyperplastic lymphoid follicles and mild, if any, alveolar interstitial inflammation101. Clinically the patients usually present during young adulthood or middle age with insidious dyspnoea102. Most cases of follicular bronchiolitis are associated with connective tissue disease, especially RA, Sjögren’s syndrome and scleroderma, but it is also seen in association with immunodeficiency syndromes including AIDS, pulmonary infections, or ill-defined hypersensitivity reactions. The cardinal features of follicular bronchiolitis on HRCT consist of centrilobular nodules measuring 1–12 mm in diameter, variably associated with peribronchial nodules and patchy areas of ground-glass opacity103. Nodules and ground-glass opacities are generally bilateral and diffuse in distribution. Mild bronchial dilatation with wall thickening and a tree-inbud pattern are less frequent findings104.

Sjögren’s syndrome

Figure 19.20 Rheumatoid arthritis. HRCT demonstrates both mild cylindrical bronchiectasis and constrictive obliterative bronchiolitis (reflected by areas of low attenuation in which there is a reduction in the number of vessels present) in this patient with rheumatoid arthritis.

Sjögren’s syndrome (SjS) is a chronic autoimmune inflammatory disease characterized by a triad of clinical features: dry mouth (xerostomia), dry eyes (keratoconjunctivitis sicca) and arthritis105. SjS can occur alone as primary SjS or in association with other autoimmune diseases—secondary SjS. A recent study evaluating the radiological and pathological manifestations of lung diseases associated with primary SjS found that NSIP was the most common entity; other pathologies included bronchiolitis, lymphoma, amyloid and atelectasis106. HRCT studies have demonstrated LIP in patients with Sjögren’s syndrome107,108; the imaging findings of which are described under the section on the IIPs. The association of LIP and amyloidosis (manifesting on HRCT as multiple irregular nodules) in patients with SjS is recognized (Fig. 19.21)107, but as these patients are also at increased risk of pulmonary lymphoma108, the finding of LIP on HRCT in conjunction with multiple nodules in a patient with SjS should at least prompt the consideration of a neoplastic process.

363

364

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 19.21 Sjögren’s syndrome—lymphoid interstitial pneumonia and amyloid. There are numerous thin-walled cysts in association with multiple irregular solid nodules, some of which are heavily calcified. Histopathological examination showed marked thickening of the interstitium with an infiltrate of small, mature lymphocytes and plasma cells. Multiple deposits of amyloid were seen throughout the specimen and there was no evidence of malignancy.

Progressive systemic sclerosis (scleroderma) Progressive systemic sclerosis (SSc) is a collagen vascular disease characterized by the deposition of excessive extracellular matrix with vascular occlusion involving several organs. It commonly affects the skin (scleroderma), peripheral vasculature, kidneys, oesophagus and lungs. As with systemic lupus erythematosus, SSc occurs more frequently in women. Cutaneous features dominate the clinical picture, at least in the early stages, although the prognosis is usually determined by involvement of the heart, lungs and kidneys. ILD is common in patients with SSc and causes considerable morbidity and

mortality. The interstitium and pulmonary vasculature are the predominant sites that are affected109,110. The HRCT findings of interstitial fibrosis in SSc include peripheral reticular opacities, ground-glass attenuation associated with traction bronchiectasis and occasionally honeycomb destruction111,100. At microscopy, NSIP is increasingly regarded as the more prevalent histological pattern in patients with SSc112,113, and indeed CT studies have confirmed that patients with SSc typically have HRCT features more akin to idiopathic NSIP with a less coarse fibrosis when compared with IPF and a greater proportion of ground-glass opacification (Fig. 19.22)114. A UIP pattern is thought to occur in 5–10% of cases113. Pleural disease is much less common in SSc than in other connective tissue diseases; pleural thickening being seen on HRCT in approximately 10% of patients115. As in other diffuse fibrosing lung diseases, enlarged mediastinal lymph nodes (which histologically show reactive hyperplasia) are a frequent finding on CT116.

Polymyositis/dermatomyositis Polymyositis (PM) is an idiopathic autoimmune inflammatory myopathy that results in proximal muscle weakness117. Dermatomyositis (DM) is similar except that it is accompanied by a skin rash. Pulmonary complications of PM/DM are important determinants of the clinical course with aspiration pneumonia being the most important pulmonary disease due to its prevalence as well as its associated morbidity and mortality118. Respiratory muscle weakness and a poor cough reflex are responsible for the high prevalence of recurrent aspiration. ILD in PM/DM was first described in 1956119 and occurs in an estimated 5–47% of patients. Initial clinical presentation is with cough, dyspnoea and fever, prior to musculoskeletal manifestations of arthralgia, myalgia and weakness in 30%, with simultaneous occurrence in only 20%120. NSIP is thought to

Figure 19.22 Scleroderma. (A,B) Two patients with scleroderma showing ground-glass opacification in association with traction bronchiectasis and a fine reticular pattern. The pattern of fibrosis is closest to that of non-specific interstitial pneumonia. Note the dilated oesophagus in both examples.

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

be the most common histological pattern seen in PM/DM120. The ILD can be acute and aggressive, similar to AIP, with some series reporting up to 10.5% mortality29,121, or more slowly progressive. In some, the lung disease is responsive to steroids and immunosuppression121. At presentation, the most common HRCT features of PM/DM are linear opacities with a lower lung predominance, ground-glass opacities, irregular interfaces and areas of consolidation (Fig. 19.23). Parenchymal micronodules and honeycombing are less frequently observed29. Histologically, organizing pneumonia is the correlate of consolidation and ground-glass opacification seen on HRCT. DAD is demonstrated in some cases and is associated with widespread involvement, dense dependent consolidation and extensive diffuse ground-glass opacification. Organizing pneumonia in PM/DM can also be admixed with interstitial fibrosis with a predominance of reticular elements and architectural distortion, traction bronchiectasis and honeycombing, and this overlap entity is associated with a poor prognosis122. When comparing patients with PM/DM ILD as a whole, the 3-year survival is 74.7% and 5-year survival 60.4%120, which is better than in IPF, but not significantly different from patients with idiopathic NSIP.

Systemic lupus erythematosus Systemic lupus erythematosus (SLE) is a chronic multisystem disease of unknown origin characterized by the presence of autoantobodies against various cell nuclear antigens123. SLE is associated with widespread inflammatory changes in the connective tissues, vessels and serosal surfaces. Pleuropulmonary disease will occur in more than half of patients with SLE at some point during the course of their illness124. Although pleuritis is the most common manifestation of SLE, diverse thoracic manifestations which range from diaphragmatic dysfunction (shrinking lung syndrome) to life-threatening pneumonitis or

pulmonary haemorrhage are encountered. A list of the other pulmonary complications may be found in Table 19.6. Pleural effusions are the most common radiographic abnormality. They are frequently bilateral, usually only small in volume and, unlike those in rheumatoid disease, are often associated with pleuritic pain. Thick horizontal band shadows at the lung bases due to linear atelectasis may be secondary to the pleurisy or, more likely, restricted diaphragmatic movement. Pulmonary consolidation in patients with SLE may cause diagnostic difficulty as it may be a consequence of infection (the incidence of respiratory tract infection in patients with SLE is high due to the immunological abnormalities, immunosuppression from steroids and respiratory muscle weakness), pulmonary oedema, lupus pneumonitis, or pulmonary haemorrhage. Pulmonary oedema may be secondary to renal

Table 19.6 INTRATHORACIC MANIFESTATIONS OF SYSTEMIC LUPUS ERYTHEMATOSUS Pleural effusion Segmental or subsegmental collapse Lupus pneumonitis Pulmonary infection Pulmonary oedema Diaphragmatic dysfunction Interstitial fibrosis (rare) Pericardial effusion Pulmonary vascular disease Pulmonary arterial hypertension Vasculitis/capillaritis Pulmonary embolism Pulmonary veno-occlusive disease

Figure 19.23 Polymyositis/dermatomyositis. HRCT features include (A) reticular opacities and (B) areas of ground-glass opacification. The appearances of (B) are compatible with organizing pneumonia being incorporated as fibrosis.

365

366

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

disease or cardiac failure. Acute lupus pneumonitis is a wellrecognized but rare manifestation of the disease that is characterized by fever, severe hypoxaemia and diffuse pulmonary infiltrates125. Radiological features are typically patchy consolidation and focal atelectasis seen predominantly in the lower lung zones with concomitant pleural effusions. Histological findings are not diagnostic but include alveolar wall damage, inflammatory cell infiltration and haemorrhage125. Compared with many other collagen vascular diseases, SLE is not commonly associated with chronic diffuse ILD. When present, reported HRCT findings include irregular linear and bandlike opacities (in part atelectasis), ground-glass opacities and interlobular septal thickening. Honeycombing which can resemble IPF is extremely rare100. Loss of lung volume is sometimes a prominent feature and is secondary to a myopathy of the diaphragmatic muscle. Diffuse alveolar haemorrhage is a rare but dramatic complication of SLE126, which manifests radiologically as widespread ground-glass opacity and consolidation. SLE is associated with increased risk of malignancy, with lymphoma being the most common124.

Ankylosing spondylitis Ankylosing spondylitis is a chronic inflammatory disease that affects mainly the axial skeleton, particularly the costovertebral, apophyseal and sacroiliac joints127. The majority of patients with ankylosing spondylitis have airway and interstitial abnormalities evident on HRCT, but these are usually mild and seldom evident on the chest radiograph128. Apical fibrosis, evident on chest radiography, is seen in approximately 1% of patients100. The upper lobe fibrosis causes upward retraction of the hila, and is often associated with bullous formation and apical pleural thickening. The changes are usually bilateral but may be unilateral, especially initially, and are indistinguishable from tuberculosis. Occasionally pulmonary changes may antedate the spondylitis. Ankylosing spondylitis is one of a number of causes (albeit a very rare one) of upper lobe fibrosis (Table 19.7). As with tuberculosis, mycetomas may form within the upper lobe cavities. The HRCT findings of ankylosing spondylitis include apical fibrosis, mild peripheral interstitial fibrosis, parenchymal bands, bronchiectasis and bullae128,129.

SYSTEMIC VASCULITIDES A number of disorders are characterized histologically by a systemic vasculitis in which the primary pathogenetic mechanism is the deposition of immune complexes in the walls of Table 19.7 CAUSES OF BILATERAL UPPER LOBE FIBROSIS Tuberculosis (including atypical mycobacterial infections) Sarcoidosis Histoplasmosis Allergic bronchopulmonary aspergillosis Chronic extrinsic allergic alveolitis Ankylosing spondylitis Progressive massive fibrosis (distinctive mass-like opacities)

blood vessels. The systemic vasculitides that most commonly affect the lung are Wegener’s granulomatosis, Churg–Strauss syndrome and Behçet’s disease. Only Churg–Strauss syndrome and Wegener’s granulomatosis will be covered in this section.

Wegener’s granulomatosis Wegener’s granulomatosis is a multisystem disease with variable clinical expression. It is characterized histologically by necrotizing granulomatous inflammation of the upper and lower respiratory tracts, the lungs being involved in approximately 90% of cases; focal necrotizing glomerulonephritis; and a small vessel vasculitis affecting arteries, capillaries and veins130. The majority of patients present with symptoms referable to the nose, paranasal sinuses, or chest; in some patients, the disease manifests solely in the respiratory tract and is known as limited (nonrenal) Wegener’s granulomatosis131. Chest symptoms include cough, dyspnoea, pleuritic chest pain and haemoptysis. Multiple nodules or masses are the most common imaging finding in Wegener’s granulomatosis, being seen in approximately 70% of cases132. Nodules range in size from a few millimetres to 10 cm, are frequently multiple, and increase in size and number as the disease progresses. Nodules are bilateral in 75% of cases, have no predilection for any lung zone, and usually show cavitation at about 2 cm in size133. HRCT may demonstrate nodules not apparent on radiography and is superior in demonstrating the presence of cavitation. Airspace consolidation and ground-glass opacities also may occur with or without the presence of nodules. Several patterns of consolidation have been described: (A) peripheral wedge-shaped lesions abutting the pleura mimicking pulmonary infarcts134; (B) a peribronchial distribution of consolidation135; and (C) a region of focal consolidation with or without cavitation (Fig. 19.24). Diffuse bilateral areas of ground-glass opacification are often a consequence of pulmonary haemorrhage132, but may also be related to necrotizing granulomatous inflammation similar to that associated with nodules. Rarely, Wegener’s granulomatosis may present as a fibrosing lung disease (closest to a UIP-type pattern) on HRCT136. Histopathologists have also described bronchocentric granulomatosis137 and organizing pneumonia138 in patients with Wegener’s granulomatosis, all of whom also demonstrated typical features of necrotizing vasculitis. Unilateral or bilateral pleural effusions are present in about 10% of patients and hilar or mediastinal lymphadenopathy has also been reported. Airway involvement is common in Wegener’s granulomatosis which may lead to subglottic, tracheal or bronchial narrowing (the latter resulting in segmental or lobar atelectasis). Mild bronchiectasis is an additional feature in Wegener’s granulomatosis occurring in up to 40% of cases134.

Churg–Strauss syndrome Churg–Strauss syndrome is an antineutrophil cytoplasmic antibody (ANCA)-associated systemic vasculitis affecting small arteries and veins. It is characterized histologically by the presence of necrotizing vasculitis and extravascular granulomatous inflammation rich in eosinophils, and clinically by the presence of asthma, fever and blood eosinophilia130. While the vasculitis affects both arteries and veins, predominant small vessel involvement is rarely encountered, and it is therefore not sur-

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

Figure 19.24 Wegener’s granulomatosis. Images through the (A) mid and (B) upper zones in a patient with Wegener’s granulomatosis. Thick-walled cavitating mass in the left upper lobe (A). Note also the focal narrowing of the mid intrathoracic trachea (B) reflecting focal involvement by Wegener’s granulomatosis.

prising that diffuse pulmonary haemorrhage is an uncommon manifestation of Churg–Strauss syndrome. HRCT appearances largely reflect the eosinophilic infiltrate and are largely nonspecific. HRCT features include ground-glass opacities, areas of airspace consolidation, centrilobular nodules (some of which may display cavitation) and airways abnormalities attributable to asthma (Fig. 19.25)139. Histologically the airspace disease is due to eosinophilic infiltrate or foci of organizing pneumonia139. Interlobular septal thickening may be seen as a result of interstitial pulmonary oedema secondary to cardiac involvement. However, a significant proportion (up to 25%) of patients with Churg–Strauss syndrome have few or no imaging abnormalities and imaging is often of little help in making this somewhat elusive diagnosis. Even when HRCT abnormalities exist they

are not specific and the diagnostic accuracy for Churg–Strauss syndrome was less than 50% in one study140.

DRUG-INDUCED LUNG DISEASE The lung is less commonly the site of drug-induced disease than other organs such as the skin and gastrointestinal tract. Nevertheless, approximately 350 drugs can cause injury to the lungs, and the list of drugs and patterns of involvement continues to increase. Respiratory disease secondary to drugs may be the result of the pharmacological action of the drug in normal or excessive dosage, or caused by an allergic or idiosyncratic reaction. The radiological manifestations of drug-induced ILD, although heterogeneous and non-specific,

Figure 19.25 Churg–Strauss syndrome. Spectrum of HRCT features: (A) areas of ground-glass opacification, (B) small cavitating nodules,

Continued

367

368

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 19.25, Cont’d (C) thickened interlobular septa, and (D) an area of airspace opacification, likely to be a peripheral infarct.

enable many alternative diagnoses to be excluded. There is no specific radiological pattern of parenchymal change associated with drug-induced lung disease and in the early stages of disease, patients with symptoms secondary to drug reaction may have a normal chest radiograph. Furthermore, data based on a small number of cases suggest that the different histological patterns of drug reaction are not always reflected by distinctive HRCT findings141. Despite these limitations, it is reasonable to understand the radiological manifestations of drug-induced lung disease via an appreciation of the underlying histological patterns of drug-induced disease142. The most common histological manifestations can be classified into DAD, chronic interstitial pneumonia (a vague term used by histopathologists which incorporates drug-induced lung disease with histologiTable 19.8

cal features that resemble either NSIP or less commonly UIP), hypersensitivity pneumonitis, organizing pneumonia and eosinophilic pneumonia142. Most drugs typically cause more than one type of histological pattern. Table 19.8 lists the drugs associated with the different histological patterns.

Diffuse alveolar damage Chemotherapeutic drugs such as busulphan, cyclophosphamide, carmustine (BCNU) and bleomycin constitute the largest group of drugs associated with this pattern of lung toxicity141. DAD usually develops a few weeks or months after initiating therapy and disease onset is heralded by progressive dyspnoea.The corresponding radiological features, not surprisingly, are similar to those found in ARDS with bilateral patchy or homogeneous

HISTOLOGICAL PATTERN OF DRUG-INDUCED LUNG DISEASE

Diffuse alveolar damage

Diffuse alveolar haemorrhage

Non-specific interstitial pneumonia

Organizing pneumonia

Eosinophilic pneumonia

Amiodarone

Amphotericin B

Amiodarone

Amiodarone

Ampicillin

Bleomycin

Cyclophosphamide

Carmustine

Carbamazepine

Captopril

Methotrexate

Nitrofurantoin

Busulphan

Interferon

Chlorpropamide

Nitrofurantoin

Amiodarone

Methotrexate

Methotrexate

Ethambutol

Carmustine

Anticoagulants

Nitrofurantoin

Sotalol

Indomethacin

Busulphan

Penicillamine

Melphalan

Nitrofurantion

Mesalazone

Sulphasalazine

Streptokinase

Phenytoin

Minocycline

Naproxen

Vinblastine

Haloperidol

Simvastatin

Phenytoin

Tetracycline Ranitidine Propranolol

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

airspace consolidation involving mainly the middle and lower lung zones. HRCT demonstrates extensive bilateral groundglass opacities and dependent areas of airspace consolidation (Fig. 19.26)143. In most circumstances there are no histological features that allow separation of drug toxicity from other potential causes of DAD and the diagnosis of drug-induced lung disease requires vigorous exclusion of other potential aetiologies, most importantly opportunistic infection.

Non-specific interstitial pneumonia Drugs reported to cause an NSIP-type pattern include amiodarone, busulphan, carmustine, methotrexate, phenytoin and simvastatin142. Descriptions of HRCT findings are available for a limited number of agents, but demonstrate the same range of abnormalities described in patients with the idiopathic form of NSIP (Fig. 19.27)143,144. With disease progression, there may be evidence of fibrosis with development of a reticular pattern and traction bronchiectasis. The fibrosis is patchy in distribution and predominantly peribronchovascular,

a pattern most commonly seen in patients receiving nitrofurantoin. In some cases, however, HRCT features suggestive of irreversible fibrosis may show complete resolution on cessation of nitrofurantoin145. NSIP is the most common manifestation of amiodarone-induced lung disease146. HRCT features that have been described with amiodarone-induced lung disease include ground-glass opacities in association with fine intralobular reticulation seen predominantly in a peripheral distribution. Foci of consolidation have also been described147, and are likely to represent areas of organizing pneumonia.

Hypersensitivity pneumonitis Several drugs have been associated with a pattern of lung toxicity with radiological and histopathological features indistinguishable from hypersensitivity pneumonitis144, although in general, this pattern is an uncommon manifestation of drug-induced lung disease. Methotrexate is the best known offender; similar changes have been attributed to cyclophosphamide, fluoxetine, nitrofurantoin and amitriptyline. The radiological and HRCT findings are similar to those seen in hypersensitivity pneumonitis secondary to the inhalation of organic dust and consist of bilateral ground-glass opacities (Fig. 19.28) and/or small, poorly defined centrilobular nodular opacities141,144. HRCT and lung biopsies in methotrexate toxicity show features more characteristic of NSIP in the majority of patients with a pattern resembling hypersensitivity pneumonitis seen in only a few patients142.

Organizing pneumonia

Figure 19.26 Diffuse alveolar damage secondary to amiodarone. There is extensive bilateral ground-glass opacification and airspace consolidation. Note also the bilateral pleural effusions.

Figure 19.27 Non-specific interstitial pneumonia secondary to bleomycin. The dominant abnormality is ground-glass opacification in association with a fine reticular pattern. The pattern of fibrosis most closely resembles non-specific interstitial pneumonia.

An organizing pneumonia-like reaction has been reported most frequently in association with methotrexate, cyclophosphamide, gold, nitrofurantoin, amiodarone, bleomycin and busulphan141. The chest radiograph shows patchy bilateral areas of consolidation, masses or nodules, which may be asymmetric or symmetric. HRCT may demonstrate patchy asymmetrical ground-glass opacity and areas of consolidation which often have a predominantly peripheral or peribronchiolar distribution (Fig. 19.29)144.

Figure 19.28 Hypersensitivity pneumonitis secondary to sertraline. HRCT shows extensive bilateral ground-glass opacification and lobular areas of air trapping (arrows).

369

370

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 19.29 Organizing pneumonia secondary to (A,B) nitrofurantoin and (C) amiodarone. The HRCT features of ground-glass opacification and consolidation (A,C) and a perilobular pattern (B) are in keeping with organizing pneumonia. The areas of consolidation in (C) are both peribronchial and perilobular in distribution.

Eosinophilic pneumonia Eosinophilic pneumonia is characterized histologically by the accumulation of eosinophils in the alveolar airspaces and infiltration of the adjacent interstitial space by eosinophils and variable numbers of lymphocytes and plasma cells. Peripheral blood eosinophilia is present in around 40% of patients. Eosinophilic pneumonia secondary to drug reaction is seen most commonly in association with methotrexate, sulphasalazine, para-aminosalicylic acid, nitrofurantoin and non-steroidal anti-inflammatory drugs. Chest radiography and HRCT show bilateral airspace consolidation, which tends to involve mainly the peripheral lung regions and the upper lobes141,143. The diagnosis of drug-induced disease will be missed unless specifically sought as a cause of unexplained diffuse pulmonary shadowing in patients at known risk with clinical symptoms of lung disease. It is particularly important, though often difficult, to differentiate between drug-induced disease, infections (particularly of the opportunistic variety) and metastatic malignancy in patients who are susceptible to a combination of these processes.

OCCUPATIONAL LUNG DISEASE Diseases of the lung caused by workplace and environmental exposures are common throughout both developed and developing worlds, and as industrial techniques continue to evolve, new occupational diseases will be recognized. The following section highlights the imaging features of the main pneumoconioses—silicosis, coal worker’s pneumoconiosis and asbestos-related pulmonary disease. Table 19.9 summarizes some of the other main occupational lung diseases. Hypersensitivity pneumonitis is covered in the preceding section on ILD. Work-related asthma is one of the most frequently reported occupational lung diseases in a number of industrialized countries148 but as these patients are not frequently imaged (and the contribution of imaging is negligible), this topic is not further discussed.

The International Labour Office Classification The International Labour Office (ILO) International Classification of Radiographs for the Pneumoconioses is a system

Table 19.9 EXAMPLES OF OCCUPATIONAL EXPOSURES THAT CAUSE LUNG PATHOLOGY Occupational lung disease Flock worker's lung

Pathology

Radiology

Lymphocytic bronchiolitis

Ground-glass opacities with centrilobular nodules181

Flavour worker's lung Obliterative (flavouring agents used bronchiolitis in microwave popcorn)

Mosaic attenuation pattern, air trapping, bronchial wall thickening182

Berylliosis

Noncaseating granulomas (indistinguishable from sarcoidosis) accompanied by mononuclear cell infiltrates and interstitial fibrosis

Nodules with a similar distribution to sarcoidosis, ground-glass opacities, thickened interlobular septa, reticular opacities and honeycombing (rare)183,184. Mediastinal adenopathy is less common than in sarcoidosis. Conglomerate masses are seen in advanced disease

Hard metal pneumoconiosis (alloys of tungsten carbide and cobalt, titanium and tantalum)

Giant cell interstitial pneumonia

Ground-glass attenuation and consolidation. Cysts and reticular abnormality may also occur185

used for the recording of chest radiographic abnormalities related to the inhalation of dusts. Its intent was to improve health workers’ health surveillance by facilitating international comparisons of pneumoconiosis statistics and research reports149. Thus, it was designed primarily for population epidemiology, rather than for individual diagnosis. In the ILO system, the size, shape and profusion of opacities on radiographs are classified in a detailed manner by trained observers using a set of standard radiographs. Rounded or nodular opacities are graded as p (< 1.5 mm diameter), q (1.5–3 mm), or r (3–10 mm). Irregular opacities are classified as s, t, or u, using the same size criteria. Large opacities (> 10 mm) are graded as A, B and C based on the combined dimensions of all large opacities present. The classification also scores the extent and thickness of plaques, pleural thickening, fissural thickening and

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

calcified nodules. Profusion of the opacities is classified into four categories (0–3); category 0 indicating that there is no excess of small opacities above normal. The use of two profusion categories is useful when appearances lie between those of the standard radiographs. Thus, 1/0 indicates that appearances most closely resemble category 1, but that the reader has also considered category 0. Despite acknowledged limitations and problems with the ILO classification (interobserver variability, the presence of background opacities that are unrelated to dust exposure, the relative insensitivity of the chest radiograph to early disease, and the misuse of the classification in legal settlements for compensation), it remains a useful shorthand whose meaning is widely understood for population studies. HRCT classification systems for the pneumoconioses have been developed150, but it is too early to gauge whether such a classification will be widely accepted and adopted.

Silicosis/coal worker’s pneumoconiosis Silicon dioxide or silica is the most abundant mineral on earth and is formed from the elements silicon and oxygen under conditions of increased heat and pressure. Any occupation that disturbs the earth’s crust or exposes the worker to the use or processing of silica-containing rock or sand has potential risks. Mining, tunnelling through rock, quarrying, stone cutting and foundry work, amongst others, are potentially hazardous occupations. Coal worker’s pneumoconiosis (CWP) is a consequence of the inhalation of coal dust. Both coalmine dust and silica predispose workers to chronic bronchitis, simple pneumoconiosis, emphysema, complicated pneumoconiosis (progressive massive fibrosis [PMF]), lung cancer (in excess of that expected from smoking alone) and mycobacterial pulmonary infection—the risk for tuberculosis is increased three-fold in patients with chronic silicosis151. As coal also contains a variable proportion of quartz, it has often been difficult to separate the pulmonary effects of coal dust from that of silica; in general, coal of high rank (high carbon content), such as anthracite, is associated with a higher incidence of CWP. Silica causes three distinct clinical patterns of lung disease (Table 19.10) which are related to both level and duration of exposure. The earliest radiographic changes of silicosis and CWP are nearly identical. Typical appearances are a profusion

Table 19.10 EXPOSURE

of small (1–3 mm) round nodules distributed in the posterior aspects of the upper two-thirds of the lung152. Radiologically, the only difference between simple CWP and simple silicosis is that the nodules in CWP are often smaller (typically p, rather than q opacities, according to the ILO classification). With advancing disease, the nodules increase in size and number to involve all lung zones. The nodules are sometimes calcified. Hilar and mediastinal lymph node enlargement with calcification of the eggshell type is not uncommon and may be seen on the chest radiograph or CT. On CT, the micronodules are sharply defined and distributed throughout the lungs but are frequently most numerous in the upper lung zones. The nodules may be centrilobular or subpleural in location; the subpleural micronodules may become confluent, forming a ‘pseudo-plaque’153. PMF refers to the coalescence of large nodules and is much more common in silicosis than in CWP. On the chest radiograph, PMF is seen as mass-like opacities, typically in the posterior upper lobes and associated with contraction of the upper lobes and hilar elevation. Sequential evaluation of these masses often demonstrates migration towards the hila, leaving a peripheral rim of cicatricial emphysema. The outer margins of PMF often parallel the contour of the adjacent chest wall. CT confirms the architectural distortion associated with PMF (Fig. 19.30). Large lesions (> 5 cm) often show irregular low attenuation regions on CT indicative of necrosis. Frank cavitation is a less frequent finding and when present should always raise the suspicion of tuberculosis (conventional or atypical). Unilateral or asymmetric PMF may be distinguished from lung cancer by the presence of lobar volume loss and peripheral emphysema. Acute silicoproteinosis develops after exposure to high concentrations of crystalline silica. The dominant feature is the presence of an alveolar proteinaceous exudate, similar to that found in pulmonary alveolar proteinosis, hence the term acute silicoproteinosis.The chest radiograph demonstrates widespread alveolar opacity with an upper and mid zone dominance. Air

PATTERNS OF DISEASE CAUSED BY SILICA

Clinical pattern

Duration and level of exposure

Acute silcoproteinosis

Occurs in response to a massive inhalation of silica (e.g. in sandblasting) usually within a few weeks to 4–5 years after exposure

Accelerated silicosis

Develops less than 10 years after first inhalation of high concentrations of silica. Its more rapid development than in simple silicosis indicates that the worker is at great risk for the development of progressive massive fibrosis

Chronic simple silicosis

The most common manifestation usually developing after 10–50 years of low level silica exposure

Figure 19.30 Progressive massive fibrosis in coalworker’s pneumoconiosis. Mass-like opacities are seen bilaterally in the upper lobes in association with multiple small nodules and calcified mediastinal lymphadenopathy.

371

372

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

bronchograms may be seen initially and hilar and mediastinal adenopathy also occur. Studies have shown that silica workers have an increased risk of IPF154,155, although epidemiological data are currently insufficient firmly to establish an aetiological link between exposure to silica and IPF-like diseases156. In addition, there has been a long-recognized association of silicosis with connective tissue disease (CTD)157. Among the CTDs, the association of silicosis and rheumatoid arthritis (Caplan’s syndrome) is more common than systemic sclerosis (Erasmus syndrome). In the development of CTD, it appears that exposure to very fine silica dust (silica flour) is necessary; this exposure may be experienced by dental technicians and workers exposed to fine scouring powders158.

Asbestos-related disease Asbestos is the generic term for a group of fibrous silicates that share the property of heat resistance. They are classified into two groups: the serpentines and the amphiboles.The only serpentine asbestos used commercially is chrysolite, which accounts for more than 90% of the asbestos used in the USA. The pathological hallmark of asbestos exposure is the asbestos body consisting of an asbestos fibre usually 2–5 µm in width. These bodies can be identified in tissue sections in interstitial fibrous tissue and intra-alveolar macrophages in broncho-alveolar lavage (BAL) fluid. The effects of asbestos on the lung are diverse and clinical manifestations of these abnormalities typically do not appear until 20 years or more after initial exposure, apart from asbestos-related pleural effusions which may be present as early as 5 years post exposure.

Benign pleural effusions The exact prevalence of benign pleural effusions is unknown, as many are subclinical. The effusions are typically haemorrhagic exudates of mixed cellularity and usually do not contain asbestos bodies. Their diagnosis is therefore reliant largely on the exclusion of other causes of effusions in an asbestos-exposed patient.The development of effusions is thought to be exposure dependent159.The effusions are often small, may be persistent or recurrent and may be simultaneously or sequentially bilateral160. Diffuse pleural thickening is the usual consequence.

Pleural plaques The most common manifestation of asbestos exposure is pleural plaques which macroscopically are discrete foci of pearly white fibrous tissue, usually 2–5-mm thick. They involve the parietal pleural almost exclusively and on the chest radiograph are classically distributed along the posterolateral chest wall between the 7th and 10th ribs, lateral chest wall between the 6th and 9th ribs, the dome of the diaphragm and the mediastinal pleura161. CT also demonstrates anterior and paravertebral plaques that are not well demonstrated on chest radiography. Calcification is reported in 10–15% of cases161. At histological examination, the plaques are relatively acellular, with a ‘basket-weave’ appearance of collagen bundles. Asbestos fibres (usually chrysolite) are often seen, but asbestos bodies are usually absent. CT is undoubtedly more sensitive for the detection of pleural plaques. Only

50–80% of cases of documented pleural thickening are detected by chest radiography162,163; on chest radiography pleural plaques were most commonly missed in the paravertebral and posterior regions of the costal pleural164. Studies have suggested that pleural plaques are not associated with significantly impaired lung function165,166.

Diffuse pleural thickening The frequency of diffuse pleural thickening increases with time from first exposure and is thought to be dose related. It results from thickening and fibrosis of the visceral pleura, which leads to fusion with the parietal pleura and may be caused by extension of interstitial fibrosis to the visceral pleura, consistent with the pleural migration of asbestos fibres. Diffuse pleural thickening superimposed on circumscribed plaques has been observed, often after a pleural effusion. Histologically, there is a similarity between pleural thickening and plaques, except that fusion of the pleural layers is suggestive of more intense inflammation. It has been shown that workers with diffuse pleural thickening have a significant reduction in forced vital capacity (FVC) and gas transfer (DLCO)167. CT is more sensitive and specific than chest radiography in the detection of diffuse pleural thickening168 and is better at the distinction between mild pleural disease and extrapleural fat169. Although oblique views can enhance detection of pleural abnormalities in cases in which HRCT is unavailable, they may also fail to distinguish plaques from extrapleural fat170.

Round atelectasis Round atelectasis, also known as folded lung, is a form of parenchymal collapse that occurs most commonly in the peripheral lung in the dorsal regions of the lower lobes. Pathological examination shows pleural fibrosis overlying the abnormal parenchyma as well as invaginations of fibrotic pleura into the region of collapse. The appearance suggests that retraction of collagen in the pleura as it matures is the cause of the collapse. Because of the pathogenetic association with fibrosis, the areas of atelectasis are always seen adjacent to the visceral pleura. A characteristic finding is the presence of crowding of bronchi and blood vessels that extend from the border of the mass to the hilum (‘comet tail’ sign)171. In most cases, the collapsed lung has a rounded or oval shape; however, wedgeand irregularly-shaped masses can also occur (Fig. 19.31).Volume loss of the affected lobe is invariably present and often associated with hyperlucency of the adjacent lung172. Serial examinations show a relatively stable appearance, and the differentiation from a lung neoplasm is usually straightforward on CT.

Asbestosis Asbestosis is defined as pulmonary parenchymal fibrosis secondary to inhalation of asbestos fibres. The lag between exposure and onset of symptoms is usually 20 years or longer. Histologically, fibrosis is first seen in the interstitium of respiratory bronchioles, particularly in the lower lobes adjacent to the visceral pleura. With advancing disease, the fibrous tissue extends into the adjacent alveolar septa, eventually involving the entire lobule173. In the

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

most severe cases there is diffuse interstitial fibrosis associated with parenchymal remodelling and honeycombing. Asbestos bodies are almost always identifiable microscopically in the fibrous tissue or macrophages in residual airspaces. Early CT changes indicative of asbestosis are the presence of subpleural curvilinear lines and dots, pleural-based nodular irregularities, parenchymal bands and septal lines164.The fine reticulation eventually progresses to a coarse linear pattern with honeycombing (Fig. 19.32). These abnormalities are usually most severe in the subpleural regions of the lower lobes. HRCT–pathological correlation studies have shown that subpleu-

ral dots and branching structures correspond to peribronchiolar fibrosis174. The sensitivity of HRCT over the chest radiograph for the identification of early fibrosis in asbestos-exposed individuals is well established175,176; however, sensitivity is not 100% and a histopathological diagnosis of asbestosis can be present in patients with normal or near-normal HRCTs177.The diagnosis of asbestosis has significant implications for the patient in terms of prognosis, work ability and the possibility of receiving legal compensation.Although both the chest radiograph and HRCT can confirm previous exposure, the diagnosis of asbestosis is largely inferential and based on

Figure 19.31 Atelectasis. Two examples of rounded atelectasis in association with (A) pleural thickening and (B) a pleural effusion. In both cases, there is evidence of lobar volume loss as evidenced by displacement of fissures. The most common location of rounded atelectasis is in the lower lobes.

Figure 19.32 Asbestosis. (A) HRCT features of early asbestosis include subpleural lines (arrowheads) and fine reticulation (arrows). These subtle abnormalities persisted on prone sections. (B) In more advanced disease, a coarse reticular pattern with honeycombing, often indistinguishable from usual interstitial pneumonia on HRCT, is seen. Note the calcified pleural plaques in both examples.

373

374

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

demonstrating a compatible structural lesion, an appropriate exposure history with a suitable latency, and the exclusion of other plausible conditions. One of the problems in interpreting the presence of interstitial fibrosis, whether on chest radiography or HRCT, is the fact that asbestos-exposed individuals are as likely as the rest of the population to develop other causes of fibrosis such as IPF178. Distinguishing asbestosis from IPF is also desirable, as asbestosis is associated with a much slower rate of progression and hence better prognosis. Discrimination between the two by HRCT appearances is by no means straightforward and is usually impossible. It has been suggested that subpleural dot-like or branching opacities are significantly more common in patients with asbestosis, whereas honeycombing, traction bronchiectasis with areas of confluent fibrosis and a mosaic perfusion pattern resulting from air trapping are more common in patients with IPF179. Additionally, pleural disease may be a discriminator: in Akira et al’s study, pleural disease was found in 83% (66/80) of patients with asbestosis but only in 4% (3/80) of patients with IPF179. Copley et al found no statistically significant differences in the coarseness of fibrosis between individuals with asbestosis and a cohort of individuals with biopsy-proven UIP, although the CT findings of asbestosis were strikingly different from NSIP; the quality of fibrosis was coarser, there was a lower proportion of ground-glass opacification, and a higher likelihood of a basal and subpleural distribution180.

REFERENCES 1. Collins C D, Wells A U, Hansell D M et al 1994 Observer variation in pattern type and extent of disease in fibrosing alveolitis on thin section computed tomography and chest radiography. Clin Radiol 49: 236–240 2. Orens J B, Kazerooni E A, Fernando J M et al 1995 The sensitivity of high-resolution CT in detecting idiopathic pulmonary fibrosis proved by open lung biopsy: a prospective study. Chest 108: 109–115 3. Hodgson M J, Parkinson D K, Karpf M 1989 Chest X-rays in hypersensitivity pneumonitis: a metaanalysis of secular trend. Am J Ind Med 16: 45–53 4. Gaensler E A, Carrington C B 1980 Open lung biopsy for chronic diffuse infiltrative lung disease: clinical, roentgenographic and physiological correlations in 502 patients. Ann Thorac Surg 30: 411–426 5. Müller N L, Kullnig P, Miller R R 1989 The CT findings of pulmonary sarcoidosis: analysis of 25 patients. Am J Roentgenol 152: 1179–1182 6. Worthy S A, Müller N L, Hartman T E et al 1997 Mosaic attenuation pattern on thin-section CT scans of the lung: differentiation among infiltrative lung, airway, and vascular diseases as a cause. Radiology 205: 465–470 7. Remy-Jardin M, Remy J, Louvegny S et al 1997 Airway changes in chronic pulmonary embolism: CT findings in 33 patients. Radiology 203: 355–360 8. Arakawa H, Kurihara Y, Sasaka K et al 2002 Air trapping on CT of patients with pulmonary embolism. Am J Roentgenol 178: 1201–1207 9. Travis W D, King T E, Bateman E D et al 2002 ATS/ERS International Consensus Classification of Idiopathic Interstitial Pneumonias. Am J Respir Crit Care Med 165: 277–304 10. Flaherty K R, King T E Jr, Raghu G et al 2004 Idiopathic interstitial pneumonia: What is the effect of a multi-disciplinary approach to diagnosis? Am J Respir Crit Care Med 170: 904–910 11. King T E Jr, Schwarz M I, Brown K et al 2001 Idiopathic pulmonary fibrosis: relationship between histopathologic features and mortality. Am J Respir Crit Care Med 164: 1025–1032 12. Johkoh T, Müller N L, Cartier Y et al 1999 Idiopathic interstitial pneumonias: diagnostic accuracy of thin-section CT in 129 patients. Radiology 211: 555–560

13. Hunninghake G W, Lynch D A, Galvin J R et al Radiologic findings are strongly associated with a pathologic diagnosis of usual interstitial pneumonia. Chest 124: 1215–1223 14. Niimi H, Kang E Y, Kwong J S et al 1996 CT of chronic infiltrative lung disease: prevalence of mediastinal lymphadenopathy. J Comput Assist Tomogr 20: 305–308 15. Raghu G, Mageto Y N, Lockhart D et al 1999 The accuracy of the clinical diagnosis of new-onset idiopathic pulmonary fibrosis and other interstitial lung disease: A prospective study. Chest 116: 1168–1174 16. Hunninghake G W, Zimmerman M B, Schwartz D A et al 2001 Utility of a lung biopsy for the diagnosis of idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 164: 193–196 17. Flaherty K R, Thwaite E L, Kazerooni E A et al 2003 Radiological versus histological diagnosis in UIP and NSIP: survival implications. Thorax 58: 143–148 18. Kondoh Y, Taniguchi H, Kawabata Y et al 1993 Acute exacerbation in idiopathic pulmonary fibrosis: analysis of clinical and pathologic findings in three cases. Chest 103: 1808–1812 19. Lee H J, Im J G, Ahn J M et al 1996 Lung cancer in patients with idiopathic pulmonary fibrosis: CT findings. J Comput Assist Tomogr 20: 979–982 20. Chung M J, Goo J M, Im J G 2004 Pulmonary tuberculosis in patients with idiopathic pulmonary fibrosis. Eur J Radiol 52: 175–179 21. Katzenstein A L A, Fiorelli R F 1994 Nonspecific interstitial pneumonia/ fibrosis: histologic features and clinical significance. Am J Surg Pathol 18: 136–147 22. Travis W D, Matsui K, Moss J et al 2000 Idiopathic nonspecific interstitial pneumonia: prognostic significance of cellular and fibrosing patterns: survival comparison with usual interstitial pneumonia and desquamative interstitial pneumonia. Am J Surg Pathol 24: 19–33 23. Nicholson A G, Colby T V, Du Bois R M et al 2000 The prognostic significance of the histologic pattern of interstitial pneumonia in patients presenting with the clinical entity of cryptogenic fibrosing alveolitis. Am J Respir Crit Care Med 162: 2213–2217 24. Johkoh T, Müller N L, Colby T V et al 2002 Nonspecific interstitial pneumonia: correlation between thin-section CT findings and pathologic subgroups in 55 patients. Radiology 225: 199–204 25. MacDonald S L, Rubens M B, Hansell D M et al 2001 Nonspecific interstitial pneumonia and usual interstitial pneumonia: comparative appearances at and diagnostic accuracy of thin-section CT. Radiology 221: 600–605 26. Prabhu M B, Barber D, Cockcroft D W 1991 Bronchiolitis obliterans and Mycoplasma pneumonia. Respir Med 85: 535–537 27. Vaiman E, Odeh M, Attias D et al 1999 T-cell chronic lymphocytic leukaemia with pulmonary involvement and relapsing BOOP. Eur Respir J 14: 471–474 28. Cameron R J, Kolbe J, Wisher M L et al 2000 Bronchiolitis obliterans organising pneumonia associated with the use of nitrofurantoin. Thorax 55: 249–251 29. Ikezoe J, Johkoh T, Kohno N et al 1996 High resolution CT findings of lung disease in patients with polymyositis and dermatomyositis. J Thorac Imaging 11: 250–259 30. Davison A G, Heard B E, McAllister W A C et al 1983 Cryptogenic organizing pneumonitis. Q J Med 207: 382–394 31. Epler G R, Colby T V, McLoud T C et al 1985 Bronchiolitis obliterans organizing pneumonia. N Engl J Med 312: 152–158 32. Müller N L, Staples C A, Miller R R 1990 Bronchiolitis obliterans organizing pneumonia: CT features in 14 patients. Am J Roentgenol 5: 983–987 33. Lee K S, Kullnig P, Hartman T E et al 1994 Cryptogenic organizing pneumonia: CT findings in 43 patients. Am J Roentgenol 162: 543–546 34. Ujita M, Renzoni E A, Veeraraghavan S et al 2004 Organizing pneumonia: perilobular pattern at thin-section CT. Radiology 232: 757–761 35. Cohen A J, King T E Jr, Downey G P 1994 Rapidly progressive bronchiolitis obliterans with organizing pneumonia. Am J Respir Crit Care Med 149: 1670–1675

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

36. Moon J, Du Bois R M, Colby T V et al 1999 Clinical significance of respiratory bronchiolitis on open lung biopsy and its relationship to smoking related interstitial lung disease. Thorax 54: 1009–1014 37. Carrington C B, Gaensler E A, Coutu R E et al 1978 Natural history and treated course of usual and desquamative interstitial pneumonia. N Engl J Med 298: 801–809 38. Park J S, Brown K K, Tuder R M et al 2002 Respiratory bronchiolitisassociated interstitial lung disease: radiologic features with clinical and pathologic correlation. J Comput Assist Tomogr 26: 13–20 39. Hartman T E, Primack S L, Swensen S J et al 1993 Desquamative interstitial pneumonia: thin-section CT findings in 22 patients. Radiology 187: 787–790 40. Hansell D M, Nicholson A G 2003 Smoking related diffuse parenchymal lung disease: HRCT–pathologic correlation. Semin Respir Crit Care Med 24: 377–392 41. Nagai S, Hoshino Y, Hayashi M et al 2000 Smoking-related interstitial lung diseases. Curr Opin Pulm Med 6: 415–419 42. Bouros D, Nicholson A G, Polychronopoulos V et al 2000 Acute interstitial pneumonia. Eur Respir J 15: 412–418 43. Primack S L, Hartman T E, Ikezoe J et al 1993 Acute interstitial pneumonia: radiographic and CT findings in nine patients. Radiology 188: 817–820 44. Johkoh T, Müller N L, Taniguchi H et al 1999 Acute interstitial pneumonia: thin-section CT findings in 36 patients. Radiology 211: 859–863 45. Howling S J, Evans T W, Hansell D M 1998 The significance of bronchial dilatation on CT in patients with adult respiratory distress syndrome. Clin Radiol 53: 105–109 46. Desai S R, Wells A U, Rubens M B et al 1999 Acute respiratory distress syndrome: CT abnormalities at long-term follow-up. Radiology 210: 29–35 47. Liebow A A 1975 Definition and classification of interstitial pneumonias in human pathology. Prog Respir Res 8: 1–33 48. Johkoh T, Müller N L, Ichikado K et al 1998 Intrathoracic multicentric Castleman disease: CT findings in 12 patients. Radiology 209: 477–481 49. Johkoh T, Müller N L, Pickford H A et al 1999 Lymphocytic interstitial pneumonia: thin-section CT findings in 22 patients. Radiology 212: 567–572 50. Ichikawa Y, Kinoshita M, Koga T et al 1994 Lung cyst formation in lymphocytic interstitial pneumonia: CT features. J Comput Assist Tomogr 18: 745–748 51. James D G, Neville E, Siltzbach L E et al 1976 A worldwide review of sarcoidosis. Ann N Y Acad Sci 278: 321–333 52. DeRemee R A 1983 The roentgenographic staging of sarcoidosis. Chest 83: 128–133 53. Kirks D R, McCormick V D, Greenspan R H 1973 Pulmonary sarcoidosis. Roentgenologic analysis of 150 patients. Am J Roentgenol 117: 777–785 54. Conant E F, Glickestein M F, Mahar P et al 1988 Pulmonary sarcoidosis in the older patient: Conventional radiographic features. Radiology 169: 315–319 55. Hamper U M, Fishman E K, Khouri N F et al 1986 Typical and atypical CT manifestations of pulmonary sarcoidosis. J Comput Assist Tomogr 10: 928–936 56. Murdoch J, Müller N L 1992 Pulmonary sarcoidosis: changes on followup CT examination. Am J Roentgenol 159: 473–477 57. Gawne-Cain M L, Hansell D M 1996 The pattern and distribution of calcified mediastinal lymph nodes in sarcoidosis and tuberculosis: a CT study. Clin Radiol 51: 263–267 58. Brauner M W, Grenier P, Mompoint D et al 1989 Pulmonary sarcoidosis: evaluation with high-resolution CT. Radiology 172: 467–471 59. Davies C W H, Tasker A D, Padley S P G et al 2000 Air trapping in sarcoidosis on computed tomography: correlation with lung function. Clin Radiol 55: 217–221 60. Padley S P G, Padhani A R, Nicholson A et al 1996 Pulmonary sarcoidosis mimicking cryptogenic fibrosing alveolitis on CT. Clin Radiol 51: 807–810 61. Hunninghake G W, Costabel U, Ando M et al 1999 ATS/ERS/WASOG statement on sarcoidosis. American Thoracic Society/European

62. 63.

64.

65. 66.

67. 68.

69.

70.

71. 72. 73.

74.

75.

76. 77.

78. 79.

80.

81.

82.

83.

84.

Respiratory Society/World Association of Sarcoidosis and other Granulomatous Disorders. Sarcoidosis Vasc Diffuse Lung Dis 16: 149–173 Brauner M W, Lenoir S, Grenier P et al 1992 Pulmonary sarcoidosis: CT assessment of lesion reversibility. Radiology 182: 349–354 Remy-Jardin M, Giraud F, Remy J et al 1994 Pulmonary sarcoidosis: role of CT in the evaluation of disease activity and functional impairment and in prognosis assessment. Radiology 191: 675–680 Leung A N, Brauner M W, Caillat-Vigneron N et al 1998 Sarcoidosis activity: correlation of HRCT findings with those of 67Gallium scanning, bronchoalveolar lavage, and serum angiotensin-converting enzyme assay. J Comput Assist Tomogr 22: 229–234 Olsson T, Bjornstad-Pettersen H, Stjernberg N L 1979 Bronchostenosis due to sarcoidosis. Chest 75: 663–666 Hansell D M, Milne D G, Wilsher M L et al 1998 Pulmonary sarcoidosis: morphologic associations of airflow obstruction at thin-section CT. Radiology 209: 697–704 Colby T V, Carrington C B 1995 Interstitial lung disease. In: Pathology of the lung, 2nd edn. Thieme Medical Publishers Inc, New York, pp 589–737 Dalphin J C, Debieuvre D, Pernet D et al 1993 Prevalence and risk factors for chronic bronchitis and farmer’s lung in French dairy farmers. Br J Ind Med 50: 941–944 Baldwin C I, Todd A, Bourke S et al 1998 Pigeon fanciers’ lung: effects of smoking on serum and salivary antibody responses to pigeon antigens. Clin Exp Immunol 113: 166–172 Remy-Jardin M, Remy J, Wallaert B et al 1993 Subacute and chronic bird breeder hypersensitivity pneumonitis: sequential evaluation with CT and correlation with lung function tests and bronchoalveolar lavage. Radiology 189: 111–118 Matar L D, McAdams H P, Sporn T A 2000 Hypersensitivity pneumonitis. Am J Roentgenol 174: 1061–1066 Silver S F, Müller N L, Miller R R et al 1989 Hypersensitivity pneumonitis: Evaluation with CT. Radiology 173: 441–445 Adler B D, Padley S P, Müller N L et al 1992 Chronic hypersensitivity pneumonitis: high-resolution CT and radiographic features in 16 patients. Radiology 185: 91–95 Cormier Y, Brown M, Worthy S et al 2000 High-resolution computed tomographic characteristics in acute farmer’s lung and in its follow-up. Eur Respir J 16: 56–60 Hansell D M, Wells A U, Padley S P G et al 1996 Hypersensitivity pneumonitis: correlation of individual CT patterns with functional abnormalities. Radiology 199: 123–128 Franquet T, Hansell D M, Senbanjo T et al 2003 Lung cysts in subacute hypersensitivity pneumonits. J Comput Assist Tomogr 27: 475–478 Patel R A, Sellami D, Gotway M B et al 2000 Hypersensitivity pneumonitis: patterns on high-resolution CT. J Comput Assist Tomogr 24: 965–970 Vogelmeier C 2003 IPF or NSIP? That is the question. Eur Respir J 22: 191–192 Jacobs R L, Andrews C P 2003 Hypersensitivity pneumonia-nonspecific interstitial pneumonia/fibrosis histopathologic presentation: a study in diagnosis and long-term management. Ann Allergy Asthma Immunol 90: 265–270 Zompatori M, Calabro E, Chetta A et al 2003 Chronic hypersensitivity pneumonitis or idiopathic pulmonary fibrosis? Diagnostic role of high resolution computed tomography (HRCT). Radiol Med (Torino) 106: 135–146 Lynch D A, Newell J D, Logan P M et al 1995 Can CT distinguish hypersensitivity pneumonitis from idiopathic pulmonary fibrosis? Am J Roentgenol 4: 807–811 Lacronique J, Roth C, Battesti J P et al 1982 Chest radiological features of pulmonary histiocytosis X: a report based on 50 adult cases. Thorax 37: 104–109 Travis W D, Borok Z, Roum J H et al 1993 Pulmonary Langerhans cell granulomatosis (histiocytosis X). A clinicopathologic study of 48 cases. Am J Surg Pathol 17: 971–986 Primack S L, Hartman T E, Hansell D M et al 1993 End-stage lung disease: CT findings in 61 patients. Radiology 189: 681–686

375

376

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

85. Brauner M W, Grenier P, Mouelhi M M et al 1989 Pulmonary histiocytosis X: evaluation with high-resolution CT. Radiology 172: 255–258 86. Brauner M W, Grenier P, Tijani K et al 1997 Pulmonary Langerhans cell histiocytosis: evolution of lesions on CT scans. Radiology 204: 497–502 87. Sherrier R H, Chiles C, Roggli V Pulmonary lymphangioleiomyomatosis: CT findings. Am J Roentgenol 153: 937–940 88. Lenoir S, Grenier P, Brauner M W et al 1990 Pulmonary lymphangiomyomatosis and tuberous sclerosis: Comparison of radiographic and thin-section CT findings. Radiology 175: 329–334 89. Müller N L, Chiles C, Kullnig P 1990 Pulmonary lymphangiomyomatosis: Correlation of CT with radiographic and functional findings. Radiology 175: 335–339 90. Kitaichi M, Nishimura K, Itoh H et al 1995 Pulmonary lymphangioleiomyomatosis: a report of 46 patients including a clinicopathologic study of prognostic factors. Am J Respir Crit Care Med 151: 527–533 91. Abbott G F, Rosado-de-Christenson M L, Frazier A A et al 2005 From the archives of the AFIP: lymphangioleiomyomatosis: radiologic– pathologic correlation. RadioGraphics 25: 803–828 92. Aberle D R, Hansell D M, Brown K et al 1990 Lymphangiomyomatosis: CT, chest radiographic and functional correlations. Radiology 176: 381–387 93. Tanoue L T 1998 Pulmonary manifestations of rheumatoid arthritis. Clin Chest Med 19: 667–685 94. Lee H K, Kim D S, Yoo B et al 2005 Histopathologic pattern and clinical features of rheumatoid arthritis-associated interstitial lung disease. Chest 127: 2019–2027 95. Flaherty K R, Colby T V, Travis W D et al 2003 Fibroblastic foci in usual interstitial pneumonia: idiopathic versus collagen vascular disease. Am J Respir Crit Care Med 167: 1410–1415 96. Remy-Jardin M, Remy J, Cortet B et al 1994 Lung changes in rheumatoid arthritis: CT findings. Radiology 193: 375–382 97. Akira M, Sakatani M, Hara H 1999 Thin-section CT findings in rheumatoid arthritis-associated lung disease: CT patterns and their courses. J Comput Assist Tomogr 23: 941–948 98. Epler G R, Snider G L, Gaensler E A et al 1979 Bronchiolitis and bronchitis in connective tissue disease: a possible relationship to the use of penicillamine. JAMA 242: 528–532 99. Gamsu G 1992 Radiographic manifestations of thoracic involvement by collagen vascular diseases. J Thorac Imag 7: 1–12 100. Primack S L, Müller N L 1998 Radiologic manifestations of the systemic autoimmune diseases. Clin Chest Med 19: 573–586 101. Yousem S A, Colby T V, Carrington C B 1985 Follicular bronchitis/ bronchiolitis. Hum Pathol 16: 700–706 102. Gibson M, Hansell D M 1998 Lymphocytic disorders of the chest: pathology and imaging. Clin Radiol 53: 469–480 103. Howling S J, Hansell D M, Wells A U et al 1999 Follicular bronchiolitis: thin-section CT and histologic findings. Radiology 212: 637–642 104. Franquet T 2001 High-resolution CT of lung disease related to collagen vascular disease. Radiol Clin North Am 39: 1171–1187 105. Cain H C, Noble P W, Matthay R A 1998 Pulmonary manifestations of Sjögren’s syndrome. Clin Chest Med 19: 687–699 106. Ito I, Nagai S, Kitaichi M et al 2005 Pulmonary manifestations of primary Sjögren’s syndrome. A clinical, radiologic and pathologic study. Am J Respir Crit Care Med 171: 632–638 107. Desai S R, Nicholson A G, Stewart S et al 1997 Benign pulmonary lymphocytic infiltration and amyloidosis: computed tomographic and pathologic features in three cases. J Thorac Imag 12: 215–220 108. Franquet T, Gimenez A, Monill J M et al 1997 Primary Sjögren’s syndrome and associated lung disease: CT findings in 50 patients. Am J Roentgenol 169: 655–658 109. Young R H, Mark G J 1978 Pulmonary vascular changes in scleroderma. Am J Med 84: 998–1004 110. Silver R M, Miller K S 1990 Lung involvement in systemic sclerosis. Rheum Dis Clin North Am 16: 199–216 111. Minai O A, Dweik R A, Arroliga A C 1998 Manifestations of scleroderma pulmonary disease. Clin Chest Med 19: 713–731 112. Fujita J, Yoshinouchi T, Ohtsuki Y et al 2001 Non-specific interstitial pneumonia as pulmonary involvement of systemic sclerosis. Ann Rheum Dis 60: 281–283

113. Bouros D, Wells A U, Nicholson A G et al 2002 Histopathologic subsets of fibrosing alveolitis in patients with systemic sclerosis and their relationship to outcome. Am J Respir Crit Care Med 165: 1581–1586 114. Desai S R, Veeraraghavan S, Hansell D M et al 2004 CT features of lung disease in patients with systemic sclerosis: Comparison with idiopathic pulmonary fibrosis and nonspecific interstitial pneumonia. Radiology 232: 560–567 115. Primack S L, Müller N L 1998 Radiologic manifestations of the systemic autoimmune diseases. Clin Chest Med 19: 573–586 116. Garber S J, Wells A U, Du Bois R M et al 1992 Enlarged mediastinal lymph nodes in the fibrosing alveolitis of systemic sclerosis. Br J Radiol 66: 983–986 117. Schwarz M I 1998 The lung in polymyositis. Clin Chest Med 19: 701–712 118. Marie I, Hachulla E, Hatron P Y et al 2001 Polymyositis and dermatomyositis: short term and longterm outcome, and predictive factors of prognosis. J Rheumatol 28: 2230–2237 119. Mills E S, Matthews W H 1956 Interstitial pneumonitis in dermatomyositis. JAMA 160: 1467–1470 120. Douglas W W, Tazelaar H D, Hartman T E et al 2001 Polymyositisdermatomyositis-associated interstitial lung disease. Am J Respir Crit Care Med 164: 1182–1185 121. Mino M, Noma S, Taguchi Y et al 1997 Pulmonary involvement in polymyositis and dermatomyositis: sequential evaluation with CT. Am J Roentgenol 169: 83–87 122. Akira M, Hara H, Sakatani M 1999 Interstitial lung disease in association with polymyositis-dermatomyositis: long-term follow-up CT evaluation in seven patients. Radiology 210: 333–338 123. Kim J S, Lee K S, Koh E M et al 2000 Thoracic involvement of systemic lupus erythematosus: clinical, pathologic, and radiologic findings. J Comput Assist Tomogr 24: 9–18 124. Murin S, Wiedemann H P, Matthay R A 1998 Pulmonary manifestations of systemic lupus erythematosus. Clin Chest Med 19: 641–665 125. Orens J B, Martinez F J, Lynch J P III 1994 Pleuropulmonary manifestations of systemic lupus erythematosus. Rheum Dis Clin North Am 20: 159–193 126. Zamora M R, Warner M L, Tuder R et al 1997 Diffuse alveolar hemorrhage and systemic lupus erythematosus (SLE): clinical presentation, histology, survival and outcome. Medicine 76: 192–202 127. Lee-Chiong T L 1998 Pulmonary manifestations of ankylosing spondylitis and relapsing polychondritis. Clin Chest Med 19: 747–757 128. Souza A S J, Müller N L, Marchiori E et al 2004 Pulmonary abnormalities in ankylosing spondylitis: inspiratory and expiratory high-resolution CT findings in 17 patients. J Thorac Imaging 19: 259–263 129. Fenlon H M, Casserly I, Sant S M et al 1997 Plain radiographic and thoracic high-resolution CT in patients with ankylosing spondylitis. Am J Roentgenol 168: 1067–1072 130. Fraser R S, Müller N L, Colman N et al 1999 Systemic vasculitides. In: Diagnosis of diseases of the chest. W B Saunders, Philadelphia, 1999 131. Luqmani R A, Bacon P A, Beaman M et al 1994 Classical versus nonrenal Wegener’s granulomatosis. Q J Med 87: 161–167 132. Cordier J F, Valeyre D, Guillevin L et al 1990 Pulmonary Wegener’s granulomatosis. A clinical and imaging study of 77 cases. Chest 97: 906–912 133. Weir I H, Müller N L, Chiles C et al 1992 Wegener’s granulomatosis: findings from computed tomography of the chest in 10 patients. Can Assoc Radiol J 43: 31–34 134. Maskell G F, Lockwood C M, Flower C D 1993 Computed tomography of the lung in Wegener’s granulomatosis. Clin Radiol 48: 377–380 135. Foo S S, Weisbrod G L, Herman S J et al 1990 Wegener granulomatosis presenting on CT with atypical bronchovasocentric distribution. J Comput Assist Tomogr 14: 1004–1006 136. Bicknell S G, Mason A C 2000 Wegener’s granulomatosis presenting as cryptogenic fibrosing alveolitis on CT. Clin Radiol 55: 890–891 137. Yousem S A 1991 Bronchocentric injury in Wegener’s granulomatosis: a report of five cases. Hum Pathol 22: 535–540

CHAPTER 19

•

HIGH-RESOLUTION COMPUTED TOMOGRAPHY OF INTERSTITIAL AND OCCUPATIONAL LUNG DISEASE

138. Uner A H, Rozum-Slota B, Katzenstein A L 1996 Bronchiolitis obliterans-organizing pneumonia (BOOP)-like variant of Wegener’s granulomatosis. A clinicopathologic study of 16 cases. Am J Surg Pathol 20: 794–801 139. Silva C I, Müller N L, Fujimoto K et al 2005 Churg–Strauss syndrome: high resolution CT and pathologic findings. J Thorac Imaging 20: 74–80 140. Johkoh T, Müller N L, Akira M et al 2000 Eosinophilic lung diseases: diagnostic accuracy of thin-section CT in 111 patients. Radiology 216: 773–780 141. Cleverley J R, Screaton N J, Hiorns M P et al 2002 Drug-induced lung disease: high-resolution CT and histological findings. Clin Radiol 57: 292–299 142. Myers J L, Limper A H, Swensen S J 2003 Drug-induced lung disease: A pragmatic classification incorporating HRCT appearances. Semin Respir Crit Care Med 24: 445–453 143. Rossi S E, Erasmus J J, McAdams H P et al 2000 Pulmonary drug toxicity: radiologic and pathologic manifestations. RadioGraphics 20: 1245–1259 144. Ellis S J, Cleverley J R, Müller N L 2000 Drug-induced lung disease: high-resolution CT findings. Am J Roentgenol 175: 1019–1024 145. Sheehan R E, Wells A U, Milne D G et al 2000 Nitrofurantoin-induced lung disease: two cases demonstrating resolution of apparently irreversible CT abnormalities. J Thorac Imag 24: 259–261 146. Kennedy J I, Myers J L, Plumb V J et al 1987 Amiodarone pulmonary toxicity. Clinical, radiologic, and pathologic correlations. Arch Intern Med 147: 50–55 147. Vernhet H, Bousquet C, Durand G et al 2001 Reversible amiodaroneinduced lung disease: HRCT findings. Eur Radiol 11: 1697–1703 148. Goe S K, Henneberger P K, Reilly M J et al 2004 A descriptive study of work aggravated asthma. Occup Environ Med 61: 512–517 149. International Labour Office 1980 Guidelines for the use of the ILO international classification of radiographs of the pneumoconioses, revised edition. International Labour Office, Geneva 150. Hering K G, Tuengerthal S, Kraus T 2004 Standardized CT/HRCTclassification of the German Federal Republic for work and environmental related thoracic diseases. Radiologe 44: 500–511 151. American Thoracic Society Committee of the Scientific Assembly on Environmental and Occupational Health 1997 Adverse effects of crystalline silica exposure. Am J Respir Crit Care Med 155: 761–768 152. Bergin C J, Müller N L, Vedal S et al 1986 CT in silicosis: correlation with plain films and pulmonary function tests. Am J Roentgenol 146: 477–483 153. Remy-Jardin M, Beuscart R, Sault M C et al 1990 Subpleural micronodules in diffuse infiltrative lung diseases: evaluation with thinsection CT scans. Radiology 177: 133–139 154. Brichet A, Wallaert B, Gosselin B et al 1997 “Primary” diffuse interstitial fibrosis in coal miners: a new entity? Study Group on Interstitial Pathology of the Society of Thoracic Pathology of the North. Rev Mal Respir 14: 277–285 155. Honma K, Chiyotani K 1993 Diffuse interstitial fibrosis in nonasbestos pneumoconiosis—a pathological study. Respiration 60: 120–126 156. De Vuyst P, Camus P 2000 The past and present of pneumoconioses. Curr Opin Pulm Med 6: 151–156 157. Rosenman K D, Moore-Fuller M, Reilly M J 1999 Connective tissue disease and silicosis. Am J Ind Med 35: 375–381 158. Koeger A C, Lang T, Alcaix D et al 1995 Silica-associated connective tissue disease. A study of 24 cases. Medicine (Baltimore) 74: 221–237 159. Epler G R, McLoud T C, Gaensler E A 1982 Prevalence and incidence of benign asbestos pleural effusion in a working population. JAMA 247: 617–622 160. Hillerdal G 2005 Non-malignant asbestos pleural disease. Thorax 36: 669–675 161. Peacock C, Copley S J, Hansell D M 2000 Asbestos-related benign pleural disease. Clin Radiol 55: 422–432 162. Friedman A C, Fiel S B, Fisher M S et al 1988 Asbestos-related pleural disease and asbestosis: A comparison of CT and chest radiography. Am J Roentgenol 150: 269–275

163. Schwartz D A, Galvin J R, Yagla S J et al 1993 Restrictive lung function and asbestos-induced pleural fibrosis. A quantitative approach. J Clin Invest 91: 2685–2692 164. Oksa P, Suoranta H, Koskinen H et al 1994 High-resolution computed tomography in the early detection of asbestosis. Int Arch Occup Environ Health 65: 299–304 165. Van Cleemput J, De Raeve H, Verschakelen J A et al 2001 Surface of localized pleural plaques quantitated by computed tomography scanning: no relation with cumulative asbestos exposure and no effect on lung function. Am J Respir Crit Care Med 163: 705–710 166. Sette A, Neder J A, Nery L E et al 2004 Thin-section CT abnormalities and pulmonary gas exchange impairment in workers exposed to asbestos. Radiology 232: 66–74 167. Kee S T, Gamsu G, Blanc P 1996 Causes of pulmonary impairment in asbestos-exposed individuals with diffuse pleural thickening. Am J Respir Crit Care Med 154: 789–793 168. Jarad N A, Poulakis N, Pearson M C et al 1991 Assessment of asbestos induced pleural disease by computed tomography—correlation with chest radiograph and lung function. Respir Med 85: 203–208 169. Lee Y C, Runnion C K, Pang S C et al 2001 Increased body mass index is related to apparent circumscribed pleural thickening on plain chest radiographs. Am J Ind Med 39: 112–116 170. Ameille J, Brochard P, Brechot J M et al 1993 Pleural thickening: a comparison of oblique chest radiographs and high-resolution computed tomography in subjects exposed to low levels of asbestos pollution. Int Arch Occup Environ Health 64: 545–548 171. Schneider H J, Felson B, Gonzalez L L 1980 Rounded atelectasis. Am J Roentgenol 134: 225–232 172. Lynch D A, Gamsu G, Ray C S et al 1988 Asbestos-related focal lung masses: Manifestations on conventional and high-resolution CT scans. Radiology 169: 603–607 173. Craighead J E, Mossman B T 1982 The pathogenesis of asbestosassociated diseases. N Engl J Med 306: 1446–1455 174. Akira M, Yamamoto S, Yokoyama K et al 1990 Asbestosis: Highresolution CT–pathologic correlation. Radiology 176: 389–394 175. Akira M, Yokoyama K, Yamamoto S et al 1991 Early asbestosis: evaluation with high-resolution CT. Radiology 178: 409–416 176. Gamsu G 1989 High-resolution CT in the diagnosis of asbestosrelated pleuroparenchymal disease. Am J Ind Med 16: 115–117 177. Gamsu G, Salmon C J, Warnock M L et al 1995 CT quantification of interstitial fibrosis in patients with asbestosis: a comparison of two methods. Am J Roentgenol 164: 63–68 178. Gaensler E A, Jederlinic P J, Churg A 1991 Idiopathic pulmonary fibrosis in asbestos-exposed workers. Am Rev Respir Dis 144: 689–696 179. Akira M, Yamamoto Y, Inoue Y et al 2003 High-resolution CT of asbestosis and idiopathic pulmonary fibrosis. Am J Roentgenol 181: 163–169 180. Copley S, Wells A, Sivakumaran P et al 2003 Asbestosis and idiopathic pulmonary fibrosis: comparison of thin-section CT features. Radiology 229: 731–736 181. Weiland D A, Lynch D A, Jensen S P et al 2003 Thin-section CT findings in flock worker’s lung, a work-related interstitial lung disease. Radiology 227: 222–231 182. Akpinar-Elci M, Travis W D, Lynch D A et al 2004 Bronchiolitis obliterans syndrome in popcorn production plant workers. Eur Respir J 24: 298–302 183. Newman L S, Buschman D L, Newell J D Jr, Lynch D A 1994 Beryllium disease: assessment with CT. Radiology 190: 835–840 184. Harris K M, McConnochie K, Adams H 1993 The computed tomographic appearances in chronic berylliosis. Clin Radiol 47: 26–31 185. Akira M 1995 Uncommon pneumoconioses: CT and pathologic findings. Radiology 197: 403–409

SUGGESTED FURTHER READING

Books/book chapters Hansell D M, Armstrong P, Lynch D A, McAdams H P (eds) 2005 Basic HRCT patterns of lung disease. Drug- and radiation-induced

377

378

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Reviews/papers diseases of the lung. Idiopathic interstitial pneumonias and immunologic diseases of the lung. Miscellaneous diffuse lung diseases. In: Hansell DM (ed) Imaging of Diseases of the Chest, 4th edn. Philadelphia, Elsevier Mosby, pp 143–181, 485–533, 535–629, 631–709 Franquet T 2005 High resolution computed tomography of the lungs. In: Wells A U, Denton C P (eds) Handbook of systemic autoimmune diseases, vol 2. Elsevier BV Lynch D A, Newell J D, Lee J S (eds) 1999 Imaging of diffuse lung disease. Elsevier, Ontario Schwarz M I, King Jr T E (eds) 2003 Interstitial lung disease, 4th edn. BC Decker Inc, Ontario Webb W R, Müller N L, Naidich D P (eds) 2001 High-resolution CT of the lung, 3rd edn. Lippincott-Raven, Philadelphia

Akira M 2002 High-resolution CT in the imaging of occupational and environmental disease. Radiol Clin North Am 40: 43–59 American Thoracic Society/European Respiratory Society 2002 International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias. Am J Respir Crit Care Med 165: 277–304 Cleverley JR, Screaton NJ, Hiorns MP et al 2002 Drug-induced lung disease: high-resolution CT and histological findings. Clin Radiol 57: 292–299 Kim J S, Lynch D A 2002 Imaging of non-malignant occupational lung disease. J Thoracic Imaging 17: 238–260 Lynch D A, Travis W D, Müller N L et al 2005 Idiopathic interstitial pneumonias: CT features. Radiology 236: 10–21 Myers J L, Limper A H, Swensen S J 2003 Drug-induced lung disease: A pragmatic classification incorporating HRCT appearances. Semin Respir Crit Care Med 24: 445–453

CHAPTER

Thoracic Trauma and Related Topics

20

John H. Reynolds

• • • • • •

Chest wall, lungs and pleura The diaphragm The mediastinum Thoracic imaging in the intensive care patient Lung transplantation Surgical treatment of emphysema

The major organs within the chest, namely the heart, lungs and major blood vessels, play a crucial role in providing oxygen to the tissues of the body and removing carbon dioxide. This occurs by means of pulmonary ventilation, gaseous exchange at the alveoli and the transport of oxygen and carbon dioxide by the cardiovascular system. Major trauma to the chest can disrupt one or more of these processes and consequently there is a range of serious and potentially life-threatening thoracic injuries, including aortic rupture, tracheal transection, haemothorax, haemopericardium and tension pneumothorax. In developed countries, trauma is the main cause of death in children and adults under 40 years of age. Approximately 20% of trauma deaths are directly related to chest trauma and of these around two-thirds occur as a result of motor vehicle accidents. Other causes of significant chest trauma include falls from a height, occupational injuries including falls and crush injuries, knife and gunshot injuries and domestic accidents. Trauma can be broadly divided into two categories, blunt and penetrating. Blunt trauma imparts kinetic energy to the body, which leads to tissue damage either by direct impact or by inducing shearing forces within body tissues. Penetrating injuries include knife and bullet wounds which, as well as damaging structures within the body, can also introduce infection1. Major chest injuries seldom occur in isolation. Commonly associated injuries include those to the head, extremities, spine, abdomen and pelvis2. Major trauma patients are managed on arrival at hospital in accordance with the Advanced Trauma Life Support (ATLS) protocol devised by the Committee on Trauma of the American College of Surgeons3. Patients are initially assessed with a primary survey that includes evaluation of the airway, breathing and circulation, followed by a

more detailed secondary survey once the immediate threats to life have been dealt with. Radiographs (antero-posterior [AP] supine) of the chest and pelvis are performed as adjuncts to the primary survey and the initial clinical and radiographic assessment may lead to further imaging. Protocols for the imaging of acutely injured patients have been undergoing a period of evolution in recent years. Although the chest radiograph remains a pivotal investigation in the rapid triage of trauma patients, computed tomography (CT) is a more accurate technique for the characterization of virtually all thoracic injuries, particularly those of the heart, pericardium, thoracic spine, mediastinum, aorta and lungs. Many trauma patients require imaging of several body areas and current multidetector CT (MDCT) systems are able to perform a whole body study in a matter of seconds. As well as allowing a detailed cross-sectional study of the trauma patient, these advanced CT machines minimize the time the patient has to spend away from the emergency department where he/ she can be cared for more safely. The quantity of image data obtained can be problematical—one solution may be to have a brief overview of the CT study at the time the patient is in the CT unit for triage purposes followed by a more detailed interrogation of the images by the radiologist once the patient has returned to the emergency unit. Concerns about the radiation dose relating to MDCT need to be kept in mind but the low threshold for the use of CT can save the patient from other radiological procedures such as radiography of the spine or catheter angiography4.

CHEST WALL, LUNGS AND PLEURA Rib fractures Rib fractures can occur in up to 50% of patients with blunt chest trauma. More than 50% of acute fractures are missed on initial radiographs owing to the superimposition of structures or because the fracture line is not tangential to the X-ray beam5. Additional lateral or oblique views to assess the ribs are inappropriate in the acute trauma patient. They are time consuming and do not alter clinical management as rib fractures

380

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

are almost invariably treated conservatively. The main purpose of the chest radiograph is to detect complications such as pneumothorax, haemothorax, or pulmonary contusion. Rib fractures lead to local bleeding and may result in an extrapleural haematoma. This is visible on the chest radiograph or CT study as a convex soft tissue bulge projecting towards the lung (Fig. 20.1). Fractures of the 1st to 3rd ribs imply severe trauma and may be associated with vascular, brachial plexus, spinal, or tracheobronchial injury. In one series of 730 patients, vascular injuries were seen in 24% of multiple trauma patients with 1st rib fractures6. Fractures of the 10th to 12th ribs, often better seen on an abdominal radiograph, are associated with injuries to the liver, spleen, or kidneys. Further imaging of these organs is mandatory when such fractures are detected2,7. In children, rib fractures are uncommon and are usually of the greenstick variety. Rib fractures are rare in accidental injury in children and their presence should raise the possibility of nonaccidental injury, particularly if they involve the posterior aspects of the ribs7. In children, teenagers and young adults the ribs have great elasticity. Major blunt trauma in this age group may well lead to significant intrathoracic injury, such as tracheobronchial, diaphragmatic, or aortic rupture without any associated rib fracture2. Double fractures of three or more adjacent ribs, or adjacent combined rib and sternal or costochondral fractures, result in a flail segment which moves paradoxically during the respiratory cycle (Figs 20.2, 20.3). This injury, referred to as ‘flail chest’, can lead to impaired ventilation and pulmonary atelectasis. There is usually severe associated pulmonary contusion which, together with associated injuries, contributes to a high mortality (in the region of 40%)2.

Figure 20.2 Flail chest. Chest radiograph with multiple left-sided lateral and posterior rib fractures resulting in a flail chest. There is associated left lung contusion.

Figure 20.3 Flail chest. CT image of a left-sided flail chest with a segment of the chest wall pushed inwards. This is known as a ‘stove-in-chest’.

Other chest wall injuries

Figure 20.1 Blunt trauma. CT image following blunt trauma to the right side of the chest demonstrating a rib fracture with associated extrapleural haematoma (arrowhead).

Sternoclavicular joint dislocations account for 3% of all dislocations5. Posterior dislocation of the clavicle is potentially dangerous as compression of the trachea or brachiocephalic vessels may result. This injury should be evident on clinical examination but will be better appreciated on CT. Sternal fractures occur in 8–10% of patients after major blunt trauma2. Mortality is high at 25% owing to associated injuries, which can include cardiac contusion, pulmonary contusion and haemothorax. The diagnosis can be made on a lateral radiograph. However, CT will allow assessment of associated injuries and coronal reformatted images may reveal subtle undisplaced fractures8. Fractures of the thoracic spine occur in around 9% of multiple trauma cases and around one-third of these will have

CHAPTER 20

cord injuries with neurological deficit. Fractures typically occur in the T9–T11 region and result from hyperflexion and/or axial loading, usually in motor vehicle accident victims or following a fall from a height9. Thoracic spine fractures are often missed initially as the clinical and radiographic signs can be overshadowed by other injuries (Fig. 20.4)10. An initial evaluation of the thoracic spine should be made on the chest radiograph performed on arrival in the emergency department.The radiograph should be sufficiently well exposed to allow this assessment. Any concern about overall alignment, vertebral body height, the pedicles, or widening of the paraspinal lines should raise suspicion of a spinal injury and be evaluated further. It has been traditional practice to perform spinal radiography in cases where there has been clinical suspicion of spinal fracture or where the patient’s conscious level does not allow a reliable clinical assessment. Plain radiographs are still used in clinical practice but MDCT is increasingly used in cases of non-trivial thoracic trauma. A number of studies have shown improved sensitivity of MDCT compared with plain radiographs. Wintermark et al studied 67 spinal fractures in 26 patients and found that sensitivity and interobserver agreement were much better with MDCT than with radiographs11. Sensitivity for fracture detection was 97.2% with CT compared with 33.3% with radiographs and the authors concluded that MDCT could replace plain radiography in the assessment of traumatic spinal injury.

Pneumothorax Pneumothorax is a common complication of blunt or penetrating trauma, occurring in 20–30% of major trauma victims5. In blunt trauma, the pneumothorax can result from pulmonary laceration from a fractured rib fragment or from a sudden rise in intra-alveolar pressure. On the chest radiograph a pneumothorax is diagnosed by visualizing the visceral pleura as a sharp thin line with absent lung markings beyond. Difficulty may occur with supine radiographs where air collects anterior to the lung and there is no visible lung edge. In

• THORACIC TRAUMA AND RELATED TOPICS

this situation a pneumothorax can produce an unusually sharp mediastinal border and hemidiaphragm and an abnormally deep costophrenic sulcus (Fig. 20.5)8. CT is much more sensitive than the chest radiograph and can detect the smallest of pneumothoraces (Fig. 20.6)4. Detection is important as even a small pneumothorax can increase in size rapidly if the patient receives positive-pressure mechanical ventilation. A tension pneumothorax develops if air can enter but not leave the pleural space. This may be rapidly fatal as the mediastinal shift to the contralateral side impedes venous return to the chest. Tension pneumothorax should be evident clinically. The radiograph may reveal mediastinal displacement, absent lung markings on the affected side and eversion of the diaphragm. A tension pneumothorax can develop even in the presence of a chest tube if the tube is occluded or malpositioned (Fig. 20.7)8.

Haemothorax Post-traumatic pleural haemothorax is found in 50% of major trauma victims5. The effusion is small if blood originates from the low-pressure pulmonary circulation as a result of lung contusion or laceration, but can be large and life-threatening if bleeding arises from large pulmonary vessels or from systemic vessels, such as the internal mammary arteries. On the chest radiograph an effusion may produce a meniscus appearance but, with the patient supine, it is likely that blood will accumulate posterior to the lung and will result only in a diffuse increase in density of the affected hemithorax (Fig. 20.8). Both ultrasound and CT can identify a haemothorax if the chest radiograph is equivocal12.

Pulmonary contusion Pulmonary contusion occurs in up to 75% of cases of blunt chest trauma5. In simple contusion the basic lung structure remains intact. Shock waves from blunt trauma lead to intra-alveolar and interstitial haemorrhage from the rupture of microvessels, associated with alveolar and interstitial oedema. Radiographically, contusion presents as nonsegmental consolidation typically

Figure 20.4 Thoracic spine fracture. (A) Chest radiograph, (B) localized view of the central chest and (C) thoracic spine MRI of a young man injured in a motorcycle road accident. A variety of injuries including a left-sided pneumothorax lead to the thoracic spine fracture being overlooked, but loss of height of one of the upper thoracic vertebral bodies can be seen on the localized view. MRI confirmed the fracture and demonstrated cord contusion.

381

382

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 20.7 Pneumothorax. CT image demonstrating a post-traumatic pneumothorax. Despite the presence of an intercostal drain, a tension pneumothorax was developing as the drain was blocked with congealed blood. The right-sided pneumothorax is situated anteriorly and the mediastinum is displaced to the left due to the tension.

Figure 20.5 Left-sided pneumothorax seen on a supine chest radiograph demonstrating the deep sulcus sign and an unusually sharp left heart border.

adjacent to the ribs, spine, or heart—kinetic energy from the impact tends to be absorbed by the lung tissue at interfaces between tissues of different densities (Fig. 20.9)13.The extent of the consolidation depends on the severity of the injury. Radiographic opacities appear within 6 h of impact and typically clear within 3–10 d5. Shadowing that increases in the days following admission is unlikely to be due to simple contusion and should alert those responsible for the management of the patient to other possibilities, such as infection, aspiration, fat embolism, or acute respiratory distress syndrome (ARDS)13. CT will detect areas of contusion not visible on the chest radiograph4.

Figure 20.6 Pneumothorax. CT image demonstrating a very small right anterior post-traumatic pneumothorax.

Pulmonary laceration Severe blunt trauma may induce shearing forces that lead to parenchymal disruption13. Owing to the elastic recoil of the lung, the resultant tear quickly becomes a space that can fill with blood, creating a haematoma, or with air, resulting in a pneumatocele, or both. Haematomas and pneumatoceles are usually small (2–5 cm) but can be larger (Figs 20.10, 20.11). They initially may be obscured by surrounding contusion and typically resolve over a period of months. Lung herniation is a rare complication of blunt chest trauma, the lung herniating through a defect caused by rib

Figure 20.8 Haemothorax. Chest radiograph of a 22 year old female pedestrian hit by a bus. The hazy opacification within the right hemithorax is due to the presence of a large haemothorax caused by a ruptured intercostal artery.

CHAPTER 20

• THORACIC TRAUMA AND RELATED TOPICS

Figure 20.10 Pneumatocoele. CT image following blunt trauma to the left side of the chest demonstrating left-sided contusion and a traumatic pneumatocoele in the periphery of the right lung, representing a contrecoup injury.

fractures or shoulder girdle dislocation. Most cases are treated conservatively13. Lung torsion is extremely rare and tends to occur in children or in adults who have undergone a lobectomy. Initially the chest radiograph shows an abnormal configuration of the pulmonary vessels but if the lung infarcts then complete opacification of the hemithorax will occur13.

THE DIAPHRAGM

Figure 20.9 Pulmonary contusion. (A) CT image in the axial plane and (B) coronal reformatted image illustrating bilateral post-traumatic pulmonary contusion. Note the subpleural predominance of the contusion.

The diaphragm can be ruptured by both blunt and penetrating trauma. In the UK blunt trauma is the more common cause, whilst in the USA the reverse is true14,15. Up to 70% of diaphragmatic tears are missed initially5. An increasing tendency for surgeons to manage injuries such as splenic rupture conservatively places a greater onus on radiologists to detect

Figure 20.11 Haematoma. (A) Axial and (B) sagittal reformatted CT images showing a large post-traumatic haematopneumatocoele involving the left lower lobe and lingular segments of the left upper lobe. Despite its large size, this lesion resolved spontaneously over a period of 6 months.

383

384

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

diaphragmatic injury16. Prompt surgical repair of diaphragmatic tears resulting from blunt trauma is essential in order to reduce the risk of subsequent complications such as bowel herniation or strangulation. Herniated bowel may itself rupture leading to contamination of the pleural space and empyema. Abdominal contents adjacent to the lung lead to impaired ventilation and atelectasis. Penetrating injuries of the diaphragm, such as may occur from a knife or bullet injury, are usually small, measuring 2 cm in length or less. They are normally diagnosed by surgical exploration in the region of the wound, including laparoscopic or thoracoscopic techniques. Imaging plays little part in their diagnosis15. Diaphragmatic rupture occurs in 0.8–5% of patients with major blunt thoraco-abdominal trauma17. A frontal impact to the abdomen leads to a massive, sudden rise in intra-abdominal pressure, which results in the rupture. In this situation, tears are usually 10 cm or more in length, are radially orientated and occur at the weakest part of the diaphragm, namely the musculotendinous junction in a posterolateral location15,17. Lateral impacts in motor vehicle accidents create a particularly high risk in that they cause compression of the thorax towards the vertebral body, which tends to stretch and tear the diaphragm16. Most clinical series show an increased incidence of left-sided tears (72–88%) in blunt trauma compared with right-sided tears17,18. Bilateral tears occur infrequently. This left-sided preponderance has been ascribed to the protective effect of the liver and the relative weakness of the left hemidiaphragm. Autopsy series, however, tend to show a more equal incidence of leftand right-sided tears19 and it is likely that many right-sided tears are missed.The smooth dome of the right lobe of the liver herniating through a right-sided tear can easily be mistaken for a normal diaphragmatic contour on the chest radiograph. If sufficient kinetic energy has been applied to the torso to rupture the diaphragm, there are usually significant associated injuries. With right-sided tears these are likely to be intra-abdominal injuries, such as hepatic, splenic, or bowel laceration. Left-sided tears are associated particularly with splenic injuries and with thoracic injury such as haemothorax and traumatic aortic injury17. The chest radiograph is relatively insensitive in diagnosing diaphragmatic rupture. In one series chest radiographs were diagnostic in only 20 of 44 patients with left-sided rupture and in 1 of 6 patients with right-sided rupture18. Radiographic findings in right-sided rupture include apparent elevation of the hemidiaphragm, loss of the diaphragmatic contour and shift of the mediastinum to the left side. Signs of left-sided tears include apparent diaphragmatic elevation, the presence of hollow viscera within the thorax, an obscured or discontinuous diaphragmatic contour and contralateral mediastinal shift. A nasogastric tube coiled within the left hemithorax is characteristic of a rupture (Fig. 20.12)17,18. Associated atelectasis, pleural effusions, lung contusion, or phrenic nerve paralysis may mask or mimic diaphragmatic tears8. Delayed imaging performed up to 6 h later may be of value in clarifying an initial equivocal chest radiograph18.This is particularly true in patients undergoing positive-pressure mechanical ventilation in whom the herniation of abdominal contents into the chest may be delayed20.

Figure 20.12 Diaphragmatic rupture. Chest radiograph showing a left-sided diaphragmatic rupture. Bowel can be seen herniating into the left hemithorax, the mediastinum is displaced to the right and there is a nasogastric tube seen coiled within an intrathoracic stomach. (Courtesy of Dr L. C. Morus, Birmingham, UK.)

Ultrasound has not gained widespread acceptance as a means of diagnosing diaphragmatic injury. The operatordependent nature of the technique together with practical difficulties such as dressings getting in the way and difficulties in interpretation in the presence of pneumothoraces or pneumoperitoneum have limited its usefulness. In practice, CT should be used whenever there is clinical or radiological suspicion of diaphragmatic injury. Before helical CT, the restriction of CT images to the axial plane created difficulties in assessing the diaphragm, which is best assessed by means of coronal or sagittal images. Despite this, certain CT signs of diaphragmatic rupture were recognized: in one series, abnormalities were found in 9 of 11 patients with surgically proven diaphragmatic rupture21 and in another CT was found to have a sensitivity of 61% and a specificity of 87% in detecting traumatic tears22. The key findings of diaphragm rupture on CT21,23 are: • discontinuity of the diaphragm • herniation of the abdominal organs into the chest • constriction of the stomach or colon as it passes through the tear (the ‘collar sign’) • dependent viscera sign whereby organs such as the spleen or liver have an abnormally posterior location due to the lack of the normal support from the diaphragm. The use of helical CT, with good quality coronal and sagittal reformatted images, should increase accuracy. Killeen et al detected 78% of left-sided and 50% of right-sided ruptures in a series of 41 patients, though cases of rupture still went undetected, particularly if no abdominal contents were herniating into the thorax at the time of data acquisition24. The advent of MDCT in the late 1990s should further enhance the role of CT with regard to diaphragmatic injury due to improved z-axis resolution and reduction in acquisition times with consequent reductions in motion artefact (Fig. 20.13).

CHAPTER 20

• THORACIC TRAUMA AND RELATED TOPICS

Figure 20.13 Rupture of the left hemidiaphragm following blunt trauma due to a road accident. (A) Chest radiograph reveals left mid zone contusion. (B) Axial and (C) sagittal reformatted CT images reveal a ruptured diaphragm on the left side with the stomach herniating through into the thorax. The stomach is constricted as it passes through the diaphragmatic tear—the so-called ‘collar sign’.

Magnetic resonance imaging (MRI) is well suited to visualizing the diaphragm, particularly on the left side. Even small tears in the region of the dome of the diaphragm can be shown by direct coronal and sagittal images (Fig. 20.14). In a study from Baltimore, MRI accurately diagnosed a diaphragmatic tear in seven patients and confirmed that the diaphragm was intact in nine25. T1-weighted images are useful as these sequences clearly show the diaphragm as a low signal intensity line with high intensity mediastinal and abdominal fat on either side. Tears are clearly depicted as defects in the low intensity line with herniation of omental fat or upper abdominal organs. Faster sequences can also be employed (e.g. fast gradient echo with fat suppression); the diaphragm is again seen as a hypodense line. Intravenous gadolinium can be helpful as it enhances contused lung and adjacent atelectasis16. Cardiac and respiratory gating is required to minimize motion artefact. MRI has the drawback that the life support devices required by the traumatized patient may not be MRI compatible and difficulties will occur in respect of patient monitoring. In practice, MRI is only performed in haemodynamically stable patients. A strategy for suspected diaphragmatic rupture has been suggested by Iochum et al16. A supine chest radiograph on admission is likely to remain the initial investigation for the foreseeable future. In the acute setting, MDCT of the chest, abdomen and pelvis should be the next investigation. If the state of the diaphragm remains in question then MRI in inspiration and expiration should be performed once the patient’s clinical condition allows.

THE MEDIASTINUM Pneumomediastinum Pneumomediastinum occurs in up to 10% of cases of blunt chest trauma. In a minority of cases, pneumomediastinum is caused

Figure 20.14 Diaphragmatic tear. Sagittal MRI showing a post-traumatic diaphragmatic tear. (Courtesy of Dr M. Bull, Sheffield, UK.)

by tracheobronchial or oesophageal rupture. In more than 95% of cases, it results from alveolar rupture due to lung trauma or positive-pressure ventilation or both. Following alveolar rupture, air tracks through the pulmonary interstitium along peribronchovascular sheaths and into the mediastinum, a process known as the Macklin effect26.Air may also enter the mediastinum from the neck or retroperitoneum. On the chest radiograph, a pneumomediastinum is manifested by lucent streaks that outline

385

386

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

mediastinal structures, elevate the mediastinal pleura and often extend into the neck or chest wall. Although the signs may be better seen on a lateral radiograph, CT is more sensitive (Fig. 20.15)27.

Tracheobronchial rupture Tracheobronchial rupture is relatively uncommon and occurs in up to 2.0% of cases of major blunt trauma28. In 90% of cases the tear is located within a mainstem bronchus and in 10% it is located in the trachea, within 2 cm of the carina5. The mechanism is that of sudden chest compression against a closed glottis. Associated injuries are common and include upper rib, sternal and thoracic spine fractures13. Bronchoscopy remains the definitive investigation8. Radiographic findings include pneumomediastinum and pneumothorax, the latter typically failing to respond to a chest drain. With complete rupture of a mainstem bronchus, the lung may sag to the floor of the pleural cavity—the ‘fallen lung sign’—as the intact vessels are unable to support the lung (Fig. 20.16)13. Chen et al reviewed findings with helical CT in a series of patients of whom 14 had tracheal rupture28. CT showed direct signs of rupture in 10 cases (71%) when either a defect or wall deformity was seen. All patients with tracheal rupture had pneumomediastinum and extraluminal air within the deep cervical soft tissues. Overall, CT had a sensitivity of 85% for detecting tracheal injury and the authors concluded that CT could detect injury directly or reveal pneumomediastinum, which in either case should lead to an early bronchoscopic assessment.

Traumatic aortic rupture Traumatic aortic rupture is the cause of 16% of motor vehicle accident deaths29. Rapid deceleration at impact leads to shearing forces at the aortic isthmus, the junction between the relatively mobile arch and the descending thoracic aorta. Contributory factors include tethering by the ligamentum arteriosum and the ‘osseous pinch’, which occurs between the anterior chest wall and the thoracic spine at impact.

Figure 20.15 Extensive pneumomediastinum seen on CT following a high speed road accident. Air can be seen around the trachea though no tracheal injury was seen at bronchoscopy.

Fig. 20.16 Fallen lung sign. Chest radiograph in a patient injured in a farming accident. The right lung is seen sagging to the floor of the right hemithorax (the ‘fallen lung sign’) and a completely ruptured right main bronchus was found at surgery.

In clinical series, 90% of aortic ruptures occur at the isthmus, just distal to the origin of the left subclavian artery. In autopsy series, 20–25% of aortic injuries occur in the ascending aorta30; ruptures at this site are usually rapidly fatal owing to exsanguination or haemopericardium and cardiac tamponade. Consequently these injuries account for only 5% of cases in clinical series. Approximately 1% of aortic ruptures due to blunt trauma occur at the level of the diaphragm5. In the majority of survivors the tear involves the intima and media with the adventitia remaining intact and maintaining aortic integrity for a time. The saccular outpouching that develops in this situation is known as a pseudoaneurysm and is normally lined by a thin layer of adventitia and surrounding tissues31. This is referred to as an incomplete aortic rupture. With complete rupture the adventitia is also disrupted and is normally associated with mediastinal haemorrhage which (if the patient survives) may progress to apical pleural capping and haemothorax5. Seventy per cent of all patients with aortic rupture die at the scene of trauma and 80–90% die before they can be treated in hospital. About 10–20% of patients with aortic injury will reach hospital and of these, 60–70% will survive, compared with only 2% who survive if the diagnosis is missed or delayed29,30. Untreated survivors invariably develop a chronic pseudoaneurysm at the site of the tear5. Early and accurate diagnosis is essential for maximum patient survival. The initial study in suspected traumatic aortic rupture is likely to be the supine chest radiograph. Portable radiographic equipment, monitoring devices in the field of view, the AP projection, gravitational effects on mediastinal vessels and adjacent pulmonary contusion all contrive to make interpretation difficult. The chest radiograph must be sufficiently well exposed to allow visualization of the aorta, the lateral aspect of which should be visible as a smooth contour

CHAPTER 20

from the aortic knuckle to the diaphragm. The radiograph must be studied for signs of mediastinal haematoma13 and therefore possible aortic rupture (Fig. 20.17). These signs include: • Widening of the mediastinum. As a guide, a wide mediastinum can be regarded as one with a width above the level of the carina of 8 cm or more or, alternatively, the mediastinum forms more than 25% of the width of the chest at this level. Too much weight, however, should not be placed on these absolute values if the subjective impression is of a wide mediastinum. Body habitus, radiographic projection, magnification and diminished inspiration all cause further diagnostic difficulties. • Blurring of the contours of the aortic arch and filling in of the aortopulmonary window. • A left apical pleural cap due to extrapleural haematoma and possibly a left pleural effusion. • Deviation of the trachea or a nasogastric tube to the right and depression of the left mainstem bronchus. • Widening of the right paratracheal stripe and the paraspinal lines. Although not practical in many trauma patients owing to haemodynamic instability and concerns over spinal injury, an erect PA chest radiograph, if it can be performed, will remove some of the interpretational difficulties seen with the supine radiograph. Despite the potential problems in its interpretation, the chest radiograph is of importance in deciding on the need for further imaging. Although occasional cases have been reported, it is very rare for a patient with traumatic aortic rupture to have a normal chest radiograph—the normal chest radiograph has a negative

Figure 20.17 Aortic rupture and mediastinal haematoma. Chest radiograph in a patient with a post-traumatic aortic rupture and mediastinal haematoma. Features present include a widened mediastinum, filling in of the aortopulmonary bay and the development of a left apical pleural cap. (Courtesy of Dr L. C. Morus, Birmingham, UK.)

• THORACIC TRAUMA AND RELATED TOPICS

predictive value of 96–98% for aortic rupture32.The chest radiograph does however lack specificity as, in the majority of cases, mediastinal haematoma is secondary to bleeding from small mediastinal veins. Only 12–25% of cases of mediastinal haematoma correspond to aortic rupture at angiography5. The difficulties in interpreting the chest radiograph in the acute trauma setting mean that further investigation is often necessary in cases of suspected traumatic aortic rupture, either to confirm or exclude aortic injury. Historically, catheter angiography has been the investigation of choice for this role. However, the increasing availability and improved image quality of MDCT, coupled with the need for CT of the head or spine, all make CT (rather than angiography) the second-line investigation after the chest radiograph. Furthermore, catheter angiography is time consuming and, although relatively safe, is not entirely without risk. Arterial puncture, catheter manipulation and the injection of contrast medium all pose a small risk to the patient33. Initially there was scepticism about the validity of using CT for this purpose but a number of large studies have now been performed which confirm the accuracy of CT in assessing the aorta34–36. The studies that have been performed have consistently demonstrated that good quality helical CT has a very high negative predictive value for aortic injury—usually at or approaching 100%34,35. This is of great practical importance as it means that when the aortic contour is smooth and clearly defined with uniform density of surrounding mediastinal fat, the aorta can be regarded as intact. Haematoma not adjacent to the aorta and without direct signs of aortic injury can be ascribed to bleeding from mediastinal veins. Peri-aortic haematoma with no direct signs of injury is more problematical and in some centres may lead to further assessment with angiography. The CT signs of aortic injury can be subdivided into direct and indirect. The direct signs consist of pseudoaneurysms and intimal flaps, whilst peri-aortic haematoma is regarded as an indirect sign (Figs 20.18, 20.19)37. In a review of 34 cases, pseudoaneurysm was seen in 97% of cases and intimal flaps in 91% and one or other of these signs was seen in all cases. Peri-aortic haematoma was identified in 91% of cases37. With advances in image quality more subtle aortic injuries are now being recognized. The term minimal aortic injury is used to describe lesions such as small intramural haematoma or intimal thrombus. Such lesions may be treated conservatively and serial CT studies may show resolution of the abnormality over time4. Surgical repair has been the traditional treatment for traumatic aortic rupture but recently endovascular stenting has been used with increasing frequency. MDCT allows effective follow-up of patients treated in this way31. Angiographic signs of traumatic aortic rupture include irregularity of the aortic isthmus, pseudoaneurysm formation and the presence of a linear radiolucency across the aorta5. The advances with regard to the assessment of trauma patients with CT mean that angiography is now seldom performed as a diagnostic study but may be performed as an adjunct to endovascular stent insertion.

387

388

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 20.18 Aortic dissection. (A) Axial and (B) sagittal reformatted CT images following a high speed road accident. An acute post-traumatic aortic dissection can be seen along with a small amount of mediastinal haematoma.

In many centres, transoesophageal echocardiography (TOE) is used effectively to assess aortic integrity as it gives good visualization of the intima in the region of the ligamentum arteriosum38. TOE has drawbacks in that it is operator dependent, does not assess the ascending aorta and is contraindicated in patients with maxillofacial injuries and in those in whom cervical spine or oesophageal injury is suspected.

swallow examination will reveal the site of a tear in 90% of cases. In most centres a water-soluble contrast agent is used initially, followed by a barium study if this is negative. Prompt diagnosis and treatment is required if mediastinitis is to be avoided39.

The oesophagus

Pneumopericardium can result from penetrating and, less frequently, from blunt trauma8. On the chest radiograph, air within the pericardium outlines the heart and only extends as far superiorly as the pericardial reflection at the root of the great vessels. The CT findings of blunt cardiac trauma include pneumopericardium, haemopericardium and intrapericardial contrast medium extravasation from a ruptured cardiac chamber8. Myocardial contusion, which is associated with sternal fracture, is relatively common but rarely clinically significant13.

The oesophagus may be ruptured by blunt or penetrating trauma though most perforations are iatrogenic, being due to endoscopy with or without therapeutic dilatation. Chest radiography in oesophageal perforation may show a pneumomediastinum and a left-sided pleural effusion. A contrast medium

The heart Pneumopericardium

Thoracic duct Thoracic duct disruption is almost always the result of penetrating trauma, with surgery being the most common cause13. The result is a chylous pleural effusion. Initial management is with chest tube drainage in the hope that this will encourage pleural adhesion and seal the leak. If this is unsuccessful, surgical ligation will be necessary39. An assessment of the site of rupture may be made with lymphoscintigraphy.

THORACIC IMAGING IN THE INTENSIVE CARE PATIENT Figure 20.19 Pseudoaneurysm. CT image following blunt trauma with aortic injury. A pseudoaneurysm is visible at the aortic isthmus (arrowhead) but in addition some extravasation of contrast has occurred indicating that there has been a complete disruption of the aortic wall. (Courtesy of Dr J. Poels, Birmingham, UK.)

The chest radiograph in the postoperative or critically ill patient on an intensive care unit (ICU) can present interpretational difficulties.The clinical problems are often complex and rapidly changing, yet the chest radiograph remains central to the diagnostic assessment. CT and ultrasound are useful adjuncts for

CHAPTER 20

when the chest radiograph alone cannot answer a clinical question. The AP projection, short tube–radiograph distance and less than full inspiration lead to lack of sharpness and particular difficulty in assessing the cardiac shadow and lung bases. The use of computed radiography and of grids can improve image quality but interpretation remains challenging.

Cardiopulmonary disease Atelectasis Atelectasis is a common finding in the critically ill patient and represents areas of nonaerated lung. Retained secretions is the most common cause. The extent can vary from linear bands of subsegmental atelectasis through to more extensive opacification to lobar collapse40. Air bronchograms may be visible. Atelectasis is usually basal with a particular predominance in the left lower lobe following cardiac surgery40.

Aspiration Aspiration factors that predispose to aspiration include a reduced conscious level and the presence of a nasogastric tube which disrupts the function of the oesophagogastric sphincter. Radiographic infiltrates normally appear within a few hours of aspiration of gastric contents and often progress for 24–48 h. There is usually a more pronounced radiological abnormality when acidic gastric contents are aspirated—pH neutral fluids such as blood may produce little or no abnormality. When present, the infiltrates are usually patchy and diffuse. They are usually bilateral or mainly right sided and are most commonly seen in the bases or superior segments of the lower lobes. Most cases show evidence of regression after 72 h. Persistent or increasing radiographic shadowing after this time raises the possibility of complicating infection or retained secretions. Any CT on an ICU patient showing dependent consolidation should raise the possibility of aspiration or infection (Fig. 20.20)41.

Pulmonary oedema Pulmonary oedema in the ICU patient may be due to a number of aetiologies.The most common causes are cardiac failure and overhydration41.

• THORACIC TRAUMA AND RELATED TOPICS

With cardiac failure the heart is typically enlarged and pleural effusions are common. Upper lobe blood diversion is a normal finding on a supine radiograph so this sign, useful in the erect situation, cannot be used. Cardiogenic pulmonary oedema results in diffuse airspace opacity in association with interstitial lines (Kerley A and B lines) and peribronchial cuffing. An enlarged vascular pedicle (the width of the mediastinum just above the aortic arch) may also be seen41. Overhydration oedema may be radiologically indistinguishable from cardiogenic oedema. Overhydration features a more central distribution of oedema and a wider vascular pedicle compared with cardiogenic oedema41. A comparison of the key radiographic features of cardiogenic versus noncardiogenic oedema is provided in Table 20.1.

Pneumonia Nosocomial, or hospital acquired, pneumonia is estimated to occur in about 10% of ICU patients and the most common infecting organisms are Gram-negative bacteria, Staphylococcus aureus and fungi42. The infection may be difficult to detect as the clinical features of outpatient pneumonia, such as fever, leukocytosis, or sputum production may not be present. The radiological appearances of pneumonia in the intensive care patient are non-specific41. There may be lobar or segmental consolidation containing air bronchograms. Consolidation without loss of lung volume is particularly suggestive of infection. Some patients may have more diffuse consolidation with air bronchograms that may be symmetrical or asymmetrical and may be indistinguishable from pulmonary oedema. The development of a cavity within an area of consolidation increases the likelihood of the presence of infection with necrosis or abscess formation. An associated pleural effusion may be parapneumonic. Loculation of pleural fluid would be suggestive of an empyema. Infection in the ICU patient may result from haematogenous spread with the development of septic emboli. Radiographically, these manifest as multiple rounded areas of consolidation which typically have a peripheral and basal predominance. The areas of consolidation typically cavitate—this is generally easier to appreciate on CT than on plain radiographs41. Pulmonary haemorrhage can produce consolidation that may mimic infection. This may occur following trauma or following surgical or other interventional procedures.

Table 20.1 COMPARISON OF RADIOGRAPHIC APPEARANCES IN CARDIAC VERSUS NONCARDIAC OEDEMA40

Figure 20.20 Recurrent aspiration. Axial CT image in a patient with oesophageal dysfunction due to systemic sclerosis. Multiple areas of airspace opacity in the lower zones of the lungs are due to recurrent aspiration.

Signs

Cardiac

Renal

ARDS

Cardiomegaly

Present

Present

Absent

Vascular redistribution

Present

Present

Absent

Widened vascular pedicle

Present

Present

Absent

Interstitial lines

Present

Present

Absent

Peribronchial cuffing

Present

Present

Absent

Airspace opacification

Diffuse perihilar

Central perihilar

Patchy, peripheral

Pleural effusions

Present

Present

Absent

389

390

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pulmonary embolism Pulmonary embolism is a common cause of morbidity and mortality in the ICU patient. Trauma patients are particularly susceptible to this complication. Predisposing factors include prolonged immobilization and the frequency of surgical procedures. The clinical signs are non-specific. The chest radiograph is of limited value. It may be normal or reveal non-specific atelectasis. A peripheral area of more or less wedge-shaped consolidation may indicate associated infarction (the so-called ‘Hampton’s hump’). Regional oligaema with sharp cut-off of a pulmonary embolism may be seen (the Westermark sign). Traditional investigations have included ventilation perfusion scintigraphy and pulmonary angiography but CT pulmonary angiography (CTPA) has become the preferred technique for confirming or excluding the presence of pulmonary embolism in the ICU patient. Studies have shown that CTPA has a very high sensitivity and specificity with a high negative predictive value for the detection of pulmonary embolism. CTPA may also identify other causes for the patient’s symptoms such as an undetected pneumothorax43.

Haemorrhage Bleeding can occur following thoracic surgery or other thoracic interventional procedures.The coagulation disturbances that are part of cardiopulmonary bypass predispose to a certain amount of postoperative bleeding, which normally results in a small quantity of blood passing out from mediastinal drains. More extensive bleeding may produce radiographic abnormality, depending on its location. Haemorrhage into the mediastinum may produce a widened mediastinum on the chest radiograph with displacement of drains or tubes. Haemorrhage into a lung will produces consolidation that can mimic a pneumonia44. Diffuse alveolar haemorrhage can occur as a complication of bone marrow transplantation. This typically produces bilateral airspace opacity on the chest radiograph, similar to that seen in pulmonary oedema45.

Table 20.2 CAUSES OF ACUTE RESPIRATORY DISTRESS SYNDROME48 Pulmonary causes

Extrapulmonary causes

Pulmonary contusion

Nonpulmonary injury (accidental and following surgery)

Aspiration of gastric acid contents

Burns

Smoke inhalation

Hypovolaemia

Near drowning

Hypoperfusion

Pneumonia

Massive blood transfusion

Fat embolism

Systemic sepsis

oedema, capillary congestion and airspace filling with oedema and red blood cells. Pulmonary vascular abnormalities including microvascular thromboses are also common. This phase is followed by the proliferative phase which occurs 7–14 d after initial injury and is characterized by organization of the airspace exudates by macrophages and fibroblasts. Cellular proliferation is accompanied by synthesis and deposition of collagen. If sufficient collagen is deposited, the patient may enter a fibrotic phase with parenchymal fibrosis, though in many patients much of the lung abnormality resolves with little or no residual histopathological or functional abnormality46. The earliest radiographic findings in the exudative phase of ARDS are those of patchy, ill-defined airspace opacities in both lungs (Fig. 20.21). Interstitial oedema is variably present46. The patchy opacities may progress to more diffuse consolidation. The airspace opacities tend to have a more peripheral distribution than those seen in relation to cardiogenic pulmonary oedema and pleural effusions are seldom seen on the supine radiographs obtained on such patients46. After a week or so, reticular opacities can be seen, corresponding to the fibrosis observed pathologically.

Acute respiratory distress syndrome A variety of direct and indirect insults to the lung can result in increased permeability of the pulmonary microvasculature, allowing protein rich fluid to pass into the alveolar spaces of the lung at normal hydrostatic pressures. Such patients may go on to develop the clinical syndrome of ARDS which is characterized by respiratory failure refractory to oxygen administration, diminished pulmonary compliance, normal pulmonary capillary wedge pressure and diffuse parenchymal infiltrates on the chest radiograph46,47. The terms ARDS and acute lung injury (ALI) describe essentially the same clinicopathological process.The difference is merely one of severity: ALI is defined as a ratio of arterial to inspired fraction of oxygen of less than 300 mmHg while ARDS is the more severe form of the disease with a ratio less than 200 mmHg48. The common causes of ARDS are summarized in Table 20.2. The pathological changes within the lung in patients within ARDS are described by pathologists as diffuse alveolar damage. Broadly speaking this can be divided into three phases. Initially there is an exudative phase, characterized by interstitial

Figure 20.21 Acute respiratory distress syndrome (ARDS). Chest radiograph in a patient with ARDS due to extrathoracic trauma. Bilateral airspace opacity is seen.

CHAPTER 20

CT findings in ARDS are characterized by diffuse groundglass opacity and gravity-dependent atelectasis. It has emerged that there are tendencies towards different patterns of CT abnormality depending on the underlying cause for the lung injury. Injuries that may lead to ARDS can be subdivided into direct causes, such as pneumonia, aspiration and near drowning and indirect (or extrapulmonary) causes such as sepsis, hypovolaemic shock, acute pancreatitis and non-thoracic trauma. Goodman et al studied 33 patients with ARDS with CT and found that ground-glass opacity was the dominant abnormality in patients with ARDS due to extrathoracic causes whilst with direct pulmonary injury, ground-glass opacity and consolidation were equally prevalent. Direct injury tended to cause asymmetrical consolidation whereas extrapulmonary causes tended to result in symmetrical ground-glass opacity. Air bronchograms were almost universal in both groups and small pleural effusions were seen in about a half of the patients49. Desai et al reviewed the CT appearances of 41 patients with ARDS. They found that what they referred to as a typical appearance of ARDS occurred more frequently in cases with an extrapulmonary cause50.The so-called typical feature of ARDS more strongly associated with an extrapulmonary cause is consolidation in the dependent, posterior parts of the lung with the density reducing more anteriorly. The reduction in density is usually gradual with consolidation merging with areas of ground-glass opacity with normal lung in the most anterior portion of the chest. With ARDS from pulmonary causes, the overall extent of consolidated lung and ground-glass opacity is about the same but the opacities tend to be patchily distributed throughout the lungs, without the gradation from dependent to nondependent areas (Fig. 20.22). Cystic spaces were a feature of the atypical appearance of ARDS. It is thought that the areas of dependent consolidation or opacification in ARDS represent areas of atelectatic lung which is compressed by overlying oedematous parenchyma50. Nondependent pulmonary opacities (seen primarily from pulmonary causes of ARDS) are likely to represent simple consolidation.The patterns of airspace opacity noted in the studies comparing pulmonary with extrapulmonary causes of ARDS can serve as a useful guide to the clinician but may not allow a specific cause to be identified. In many patients the causes of ARDS are multifactorial with a combination of pulmonary and extrapulmonary causes being present. Desai et al analysed the CT findings in long-term survivors of ARDS51. The most common abnormality in the survivors was a reticular pattern (indicating fibrosis) and this had a striking anterior distribution (Fig. 20.23). A suggested explanation for this anterior distribution is that the more dependent collapsed and consolidated lung is protected from injury due to barotrauma related to mechanical ventilation.

Extrapulmonary air Air may escape from the normal airways in the lung due to blunt or penetrating trauma. In the ICU setting alveolar disruption from barotrauma may lead to an air leak. Surgery and other medical procedures are another cause of extrapulmonary air. Possible locations for extrapulmonary air are:

• THORACIC TRAUMA AND RELATED TOPICS

Figure 20.22 Acute respiratory distress syndrome (ARDS). CT images in two patients. (A) This patient’s ARDS was due to an extrapulmonary cause and the CT shows increased opacification in the posterior, dependent portions of the lungs and ground-glass opacity more anteriorly. A right-sided intercostal tube is also present and part of a Swan–Ganz catheter can be seen in the left main pulmonary artery. (B) This patient’s ARDS was related to pulmonary infection and there is patchy airspace opacity present with no gradation from dependent to nondependent lung being seen.

• • • • •

interstitial pulmonary spaces the mediastinum the pleural space the pericardium subcutaneous tissues.

Interstitial pulmonary air in adults is difficult to recognize radiographically. Escaped air dissects along bronchovascular structures towards the mediastinum. The appearance may superficially appear as air bronchograms, but, unlike air bronchograms, the lucencies do not branch or taper towards the lung periphery52. Pneumomediastinum results in linear streaking of air density within the mediastinum. Depending on the amount of air present, normal anatomical structures may become visible.The thymus may be visible; air may be seen anterior to the pericardium (best appreciated on a lateral radiograph); air surrounding the pulmonary arteries can produce a ringlike lucency; air on either side of a bronchial wall results in unusually sharp

391

392

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

ing radiographically occult effusions. When performing ultrasound on the ICU it is important to evaluate the posterior, dependent aspect of the hemithorax in question in order that effusions are not overlooked—this will usually involve some manœuvring of the patient.

Support and monitoring apparatus Airway

Figure 20.23 Post acute respiratory distress syndrome (ARDS) fibrosis. CT image following recovery from ARDS. Reticular opacities and traction bronchiectasis can be seen anteriorly indicating fibrosis.

delineation of the wall—the double bronchial wall sign; and air over the diaphragmatic surface leads to the continuous diaphragm sign53. As has been discussed in the prior section on thoracic trauma, diagnosing a pneumothorax on a supine AP chest radiograph can be challenging. The classical appearance of a lung edge with absent lung markings beyond may well not be present. In the supine position, air preferentially accumulates anterior to the lungs and abuts the mediastinal structures. Unusually sharp demarcation of a heart border or mediastinal vascular structure such as the superior vena cava may be the only indication on the radiograph of a pneumothorax. An unusually deep costophrenic sulcus (the ‘deep sulcus sign’) is another indicator of pneumothorax on an anterior radiograph (see Fig. 20.5)52. With a subpulmonary pneumothorax there may be a hyperlucent upper quarter of the abdomen and a sharply demarcated diaphragmatic surface52. Pneumopericardium may result as a consequence of barotrauma in children but in adults is more likely to be a consequence of a cardiothoracic surgical procedure. Features indicating pneumopericardium rather than pneumomediastinum include outlining of the superior pericardial reflection around the great vessels with air and visualization of the main pulmonary artery52.

Pleural effusions Pleural effusions are common in the ICU setting and may be related to trauma, congestive cardiac failure, fluid overload, pneumonia, or surgery. In the erect patient an effusion manifests as blunting of the costophrenic angle with increased basal radiographic density. In the supine patient, pleural fluid, when free rather than loculated, tends to collect in a posterobasal location. This results in a diffuse, hazy increase in density over the lower lungs through which bronchovascular markings may still be seen. Fluid in a subpulmonary location may cause apparent elevation of the hemidiaphragm41. In cases of uncertainty both CT and ultrasound can be of value in detect-

Monitoring the presence and position of the various tubes and catheters used in the critically ill patient is an important aspect of reading an ICU chest radiograph. Endotracheal tubes are inserted to maintain an airway and administer oxygen. The ideal position for the tip is in the mid trachea about 5 cm cranial to the carina.This allows for a degree of upward or downward movement which can occur when the head is moved. A tube positioned too inferiorly will tend to enter the right main bronchus leading to impaired ventilation and, ultimately, collapse of the left lung54. For longer periods of intubation a tracheostomy tube is likely to be employed. These have the advantage of not moving with movement of the head. The tip of the tube should lie between one-half and two-thirds of the distance between the stoma and the carina.The cuff should fill but not distend the trachea wall54.

Intravascular devices Catheters are commonly inserted to monitor central venous pressure (CVP). On the chest radiograph the tip of the catheter should be projected between the medial end of the 1st rib, at the junction of the brachiocephalic vein and superior vena cava, or within the superior vena cava itself. Peripherally inserted central catheters (PICC) have a small calibre and can be left in place for longer durations to allow completion of a course of intravenous therapy. Ideally, PICCs should terminate within the superior vena cava54. Monitoring of left-sided cardiac pressures using pulmonary capillary wedge pressure (PCWP) is critical for maintaining accurate blood volume. The Swan–Ganz catheter normally used for this purpose is introduced into the pulmonary artery about 5 cm distal to the main pulmonary artery bifurcation. A balloon is then inflated to allow the catheter to float into the wedge position.The catheter, when in use, should not extend beyond the proximal interlobar arteries on the chest radiograph—more distal positioning increases the risk of associated pulmonary infarction54. Other devices that may be encountered include intra-aortic balloon counterpulsation devices, transvenous and epicardial pacing wires, thoracostomy tubes and nasogastric tubes. The main tubes and lines encountered on ICU images are summarized in Table 20.3.

LUNG TRANSPLANTATION Lung transplantation is widely accepted as a form of therapy for a range of end-stage lung and pulmonary vascular diseases. Surgical options include heart–lung, single lung or bilateral lung transplantations. Single lung transplantation has the advantage of increasing the overall number of patients who can potentially receive a

CHAPTER 20

Table 20.3

• THORACIC TRAUMA AND RELATED TOPICS

LINES AND TUBES ENCOUNTERED ON AN ICU CHEST RADIOGRAPH40,44

Appliance

Function

Optimum location of tip

Endotracheal tube

Ventilatory support

3–8 cm above carina

Swan–Ganz catheter

Wedge and right heart pressures

Right or left pulmonary artery

Central venous pressure catheter

Central venous pressure

Superior vena cava

Left atrial catheter

Left atrial pressure

Left atrium

Peripherally inserted central catheter line

Intravenous therapy

Superior vena cava

Mediastinal drains

Mediastinal fluid evacuation

Anterior mediastinum or posterior pericardium

Pleural tubes

Pleural space evacuation

In pleural space via mid axillary line, 6th to 8th rib spaces. Directed anteriorly for pneumothorax and posteriorly for effusion

Temporary pacing wires

Cardiac pacing

Over right heart

Nasogastric tube

Gastric evacuation

Left upper quadrant of abdomen, with side-holes in stomach

lung and is the preferred method in cases of nonsuppurative lung disease, such as emphysema, idiopathic pulmonary fibrosis, sarcoidosis, or lymphangioleiomyomatosis. Bilateral sequential lung transplantation is performed for suppurative lung diseases, such as cystic fibrosis and bronchiectasis and also for severe pulmonary hypertension. Heart–lung transplantation is performed in cases of combined heart and lung disease55. Aspects of imaging in relation to lung transplantation can be subdivided into pre-operative, peri-operative and postoperative.

Pre-operative imaging Typical imaging procedures performed before lung transplantation include the PA and lateral chest radiograph, chest CT and quantitative ventilation–perfusion scintigraphy. Pre-operative imaging can help determine the optimum side for a single lung transplant procedure and screen for potential lung cancer. The chest radiograph can also be used for donor–recipient size matching55.

Peri-operative imaging Most patients are extubated within 24–48 h of transplantation. In some cases, complications such as infection or early graft dysfunction may necessitate a longer period of ventilation, in which case the patient usually undergoes a tracheostomy.

Reperfusion oedema Reperfusion oedema (also known as the re-implantation syndrome) is caused by increased capillary permeability. Causes include interruption of lymphatic drainage in the donor lung, underlying donor lung injury, surfactant deficiency and ischaemic damage to pulmonary capillaries55. In the series reported by Kundu et al56, reperfusion oedema was found to be nearly universal, occurring in 44 of 45 patients. Radiographic signs were non-specific but appeared most commonly as airspace opacity in the mid and lower zones. Linear or reticular radiographic shadowing was also commonly seen. Anderson et al57 found radiographic signs of reperfusion oedema in 97% of a series of 105 lung transplant patients.There is poor correlation between the degree of radiographic abnormality and physiological measurements56. Peak radiographic shadowing usually occurs around day 456 and in most patients the shadowing will have cleared by the 10th postoperative day55.

Early graft dysfunction Early graft dysfunction is a general term that describes a range of early injuries, including reperfusion oedema, ARDS and graft failure. Although potentially related to a range of underlying problems, patients show a common clinical pattern of presentation which includes radiographic abnormalities, poor oxygenation and biopsies, if performed, demonstrate diffuse alveolar damage or organizing pneumonia55. Radiographic abnormalities range from mild changes of airspace opacity associated with reperfusion oedema through to complete lung opacification55.

Postoperative imaging Infection The lung transplant patient is particularly vulnerable to infection due to a variety of causes such as immunosuppressive therapy, lost cough reflex and impaired mucociliary function in the denervated transplanted lung. Organisms that may infect the post-transplant lung include bacterial, viral, fungal, mycobacterial and mycoplasma species. In a review of patients with post-transplant infection, the most common infecting organisms were cytomegalovirus, pseudomonas and aspergillus. The most common CT findings were consolidation, ground-glass opacity, septal thickening, multiple or single nodules and pleural effusion. No significant difference in the prevalence of the CT signs was seen between the different groups of infecting organism—imaging may identify signs of infection but does not allow the specific organism to be recognized58.

Acute rejection Acute rejection occurs in virtually all transplanted lungs and usually within the first 3 months. The diagnosis is confirmed histopathologically with a transbronchial biopsy which typically demonstrates perivascular and interstitial mononuclear infiltrates. The rejection is graded from 0 to 4 depending on the severity of the biopsy changes and the majority of cases respond to intravenous methylprednisolone therapy55. Radiographic findings are non-specific and include new or persisting airspace opacities 5–10 d following transplantation, pleural effusions and interstitial lines without other signs of heart failure. High-resolution CT (HRCT) signs are similarly non-specific though ground-glass opacity and septal lines may

393

394

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

be the predominant findings when acute rejection occurs after the first post-operative month55.

Bronchial anastomotic complications The bronchial anastamoses following transplantation may be complicated by dehiscence or stenosis (Fig. 20.24). Dehiscence (or separation of the two airway ends) tends to occur in the first few months following transplantation and may be associated with infection55. Factors that contribute to airway problems include ischaemia, acute allograft rejection, low cardiac output and prolonged postoperative ventilation55.

Obliterative bronchiolitis Long-term survival following lung transplantation is limited primarily by the development of obliterative bronchiolitis (OB). OB is characterized histopathologically by the presence of fibrosis in small airways associated with intimal thickening and sclerosis of vessels. The process is thought to represent chronic allograft rejection55. Episodes of acute rejection increase the likelihood of developing OB. Although it can occur as early as 2 months following transplantation, most cases are diagnosed 6–12 months following surgery. The chest radiograph in OB may be normal in the early stages though as the disease progresses signs of lung overinflation and subtle attenuation of peripheral airways may become apparent59. HRCT findings include areas of decreased lung attenuation associated with vessels of decreased calibre. Images acquired at end-expiration can reveal air trapping (Fig. 20.25)60. Air trapping on expiratory CT is the most sensitive and may be the only radiological sign of OB61. Redistribution of pulmonary blood flow to more normal areas of lung may lead to a mosaic attenuation pattern60. Bronchiectasis is commonly present60,62.

Post-transplantation malignant disease Post-transplantation lymphoproliferative disease (PTLD) occurs in 5–20% of lung transplant recipients, usually in the first year after transplantation55. Histological findings may range from benign hyperplasia of lymphocytes through to malignant lymphoma. PTLD is thought to be caused by

Figure 20.24 Post-transplant stenosis. (A) Axial plane and (B) three-dimensional reconstruction showing a post-transplant stenosis of the distal right main bronchus.

proliferation of donor B lymphocytes infected with the Epstein–Barr virus55. The most common CT manifestations are of multiple nodules, frequently in a predominantly peribronchovascular or subpleural distribution63.

Recurrence of the primary disease This has been described with a number of conditions, with sarcoidosis being the disease that recurs most commonly64.

SURGICAL TREATMENT OF EMPHYSEMA As described in Chapter 16, emphysema is defined as a condition of the lung characterized by permanent, abnormal enlargement of the airspaces distal to the terminal bronchioles, accompanied by destruction of their walls and without obvious fibrosis. The caveat relating to the absence of fibrosis was inserted to make emphysema distinct from honeycomb fibrosis but this has been controversial as some degree of inflammation and fibrosis can occur in association with emphysema65,66. Put simply, emphysema is thought to be caused by an imbalance between elastolytic enzymes, which tend to destroy lung tissue and anti-elastolytic proteins. Cigarette smoking can lead to recruitment of cells such as neutrophils and macrophages around bronchioles and these can initiate a local increase in elastolytic enzymes. Genetic deficiency of α1-antitrypsin can also predispose to emphysema. Typical features of emphysema on the chest radiograph include hyperinflation (flattened, depressed hemidiaphragms, increased retrosternal space), hyper-transradiancy and attenuated vessels. CT has a high sensitivity for the detection of emphysema and HRCT is more sensitive than conventional CT67. The four recognized subtypes of emphysema68—centrilobular emphysema, panlobular emphysema, paraseptal emphysema and irregular or ciccatricial emphysema—are described in Chapter 16.

Surgical options in emphysema Over the last century a number of surgical procedures have been used to attempt to relieve the distressing symptoms of advanced emphysema. Many of these have not stood the test of time but

CHAPTER 20

Figure 20.25 Post-lung transplantation obliterative bronchiolitis. The lungs are overinflated with mild cylindrical bronchiectasis and attenuation of pulmonary vessels. Areas of patchy ground-glass opacity in the periphery of the lung were thought to be due to cytomegalovirus pneumonitis. The patient died within a month of this examination.

three procedures are in current use.These are bullectomy, transplantation and lung volume reduction surgery (LVRS).

Bullectomy This consists of the excision of a large bulla or bullae by thoracostomy or by video-assisted thoracoscopic surgery (VATS). Bullectomy is considered in patients with a single or several large bullae identified on the chest radiograph and CT, usually with compression of adjacent lung parenchyma. The technique is usually reserved for patients with dyspnoea or recurrent pneumothoraces69.

Lung transplantation Lung transplantation can offer a marked improvement in the quality of life of the patient with advanced emphysema but the limited availability of donor organs is a major disadvantage. Transplantation is usually reserved for patients younger than 60–65 years of age and such patients will normally have resting hypoxia, hypercapnoea and a dependence on supplementary oxygen. Patient selection is based primarily on clinical and physiological criteria.The main roles of imaging in the preoperative patient are to assess whether alternative treatment such as LVRS might be more appropriate and to exclude incidental lung cancer.

Lung volume reduction surgery The surgical removal of the most emphysematous area of lung was first attempted in the 1950s by Professor Otto Brantigan from the University of Maryland, USA. Initially the technique did not gain widespread acceptance due to a relatively high operative mortality of 16%. During the 1990s, Joel Cooper of the Division of Cardiothoracic Surgery, Washington University School of Medicine, St Louis, USA revisited the technique and achieved an acceptable postoperative mortality of 4%70.

• THORACIC TRAUMA AND RELATED TOPICS

LVRS involves the removal of the most severely emphysematous portions of lung in the upper lobes. Typically about 30% of each lung is removed using either a median sternotomy or a video-assisted thoracoscopic technique. Air leaks from the lung can be a problem following LVRS. The use of bovine pericardial strips to buttress the staple line and the use of a ‘pleural tent’ to cover the staple line have kept this problem at an acceptable level70. LVRS is thought to have a beneficial effect by two main mechanisms: improved respiratory mechanics brought about by repositioning of the chest wall and diaphragm; and redistribution of ventilation which improves gas exchange.Advantages of LVRS include the relief of dyspnoea for patients without the problems of organ transplantation. Work so far has shown that a successful outcome depends on specific anatomical and physiological characteristics in the patient’s lungs and so careful patient selection is essential70. Patient selection Cooper’s results were encouraging but other centres which adopted the procedure did not have such successful outcomes and acceptable mortality rates. The use of LVRS became controversial towards the late 1990s and a number of trials were established to confirm the efficacy and safety of the technique. The most influential of these was the National Emphysema Treatment Trial (NETT) carried out in the USA71. The NETT recruited 1218 patients and of these 610 were randomized to medical treatment and 608 to surgical treatment, of whom 508 underwent surgery. The NETT overall showed improved exercise capacity in patients undergoing LVRS but no difference in survival between patients given medical or surgical therapy. With subgroup analysis, a survival advantage was seen for patients with predominantly upper lobe emphysema (Figs 20.26, 20.27) and a low baseline exercise capacity. Imaging plays a pivotal role in patient selection for LVRS. CT is the preferred method of assessing patients for LVRS but the chest radiograph remains a useful first-line investigation. The majority of patients will have a radiograph demonstrating hyperinflation with flattening of the hemidiaphragms. CT allows for more sensitive detection of emphysema and a more accurate assessment of disease distribution. Incidental lung cancers will also be more readily detected— such cancers do not preclude LVRS as, depending on their location, they may be removed by wedge excision at the same time as the volume reduction surgery (Fig. 20.28)72. An interim report from the NETT announced that patients with a forced expiratory volume in 1 s (FEV1) of less than 20% predicted value and either a homogeneous distribution of emphysema or a carbon monoxide diffusing capacity less than 20% predicted value were at higher risk of death following LVRS (16% 30-d mortality), thus reaffirming the importance of carefully assessing emphysema distribution with CT73. The severity and distribution of emphysema can be assessed either with a visual scoring system or by means of a density mask technique which identifies pixels below a predetermined value (usually around −910 HU)74.

395

396

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 20.26 Upper zone emphysema. Axial (A) upper zone and (B) lower zone CT images of a patient with upper zone predominant emphysema that would be potentially suitable for treatment with lung volume reduction surgery.

Figure 20.27 Panlobular emphysema. Axial (A) upper zone, (B) lower zone and (C) coronal reformatted CT images of a patient with panlobular emphysema due to α1-antitrypsin deficiency. Emphysema of this type would not be suitable for lung volume reduction surgery though lung transplantation would be a possible option.

Perfusion scintigraphy can provide useful information about regional perfusion differences in the lung reflecting the distribution of emphysema. Although there is generally good correlation between CT and scintigraphy, Cederlund et al found that a combination of the two provided the best pre-operative assessment of disease heterogeneity75. Postoperative imaging In clinical practice, chest radiographs are routinely performed following LVRS but CT is only used to deal with specific clinical problems. Complications following LVRS are generally the same as those following any thoracic surgical procedure but particular attention needs to be given to the possibility of prolonged air leak. The staple line with pericardial buttress is always seen on CT and sometimes on the postoperative chest radiograph and this may cause some confusion in interpretation (Fig. 20.29). If a pleural tent has been performed then this may be mistaken for a pneumothorax in the early postoperative period. Aside

from air leaks, other complications reported in the NETT included pneumonia and cardiac arrhythmias71,76. A number of studies have performed quantitative analyses of lungs following LVRS using CT. Lung volumes measured with CT typically reduce by around 25% following LVRS and the average lung parenchymal density increases by around 25 HU77. Summary of the role of imaging The key roles of imaging in possible LVRS cases can be summarized as follows: • Confirming that the patient does have emphysema and identifying the subtype. • Assessing the distribution of emphysema. Means of assessment and criteria for surgery have varied from study to study but in practice the emphysema needs to be primarily of the centrilobular type with an upper zone predominance. There should be complete or relative sparing of the lower zones. As a useful rule of thumb, the emphysema should be

CHAPTER 20

• THORACIC TRAUMA AND RELATED TOPICS

predominant emphysema related to α1-antitrypsin deficiency. Only a minority of patients referred for consideration of LVRS will ultimately prove suitable for the technique71.

REFERENCE

Figure 20.28 Upper zone emphysema. CT image from a patient with upper zone predominant emphysema. The mass in the left upper lobe is a probable lung cancer but this could potentially be resected at the same procedure as the lung volume reduction surgery.

Figure 20.29 Postoperative CT image after lung volume reduction surgery. The staple line with bovine pericardium buttresses can be seen (arrowhead).

twice as severe in the upper zones as in the lower zones before a patient can be considered for possible LVRS. • Identifying incidental cancers. Depending on their location, these may not necessarily preclude LVRS. • Excluding other lung diseases which could lead to intra-operative or postoperative problems and thus may represent absolute or relative contraindications to surgery. These include severe bronchiectasis and diffuse parenchymal lung diseases, such as idiopathic pulmonary fibrosis. • Postoperative assessment—routine assessment with chest radiography and CT when there are specific clinical questions to be addressed. Emphysema is a common problem worldwide and LVRS can provide improved physiology, lung function and survival in certain patients. The emphysema distribution that is most favorable is that of upper lobe predominance and in practice this means centrilobular emphysema related to cigarette smoking—there is currently no role for LVRS in lower zone

1. Boot D A 1999 Epidemiology of accidents. In: Alpar EK, Gosling P (eds) Trauma: a scientific basis for care. Edward Arnold, London, pp 1–19 2. Besson A, Saegessor F 1982 A colour atlas of chest trauma and associated injuries. Wolfe Medical, Weert, the Netherlands 3. Committee on Trauma of the American College of Surgeons 1997 Advanced trauma life support for doctors: student course manual, 6th edn. American College of Surgeons, Chicago, 1997 4. Rivas L A, Fishman J E, Munera F, Bajayo D E 2003 Multislice CT in thoracic trauma. Rad Clin North Am 41: 599–616 5. Stark P 1993 The radiology of thoracic trauma. Andover Medical, Boston 6. Gupta A, Jamshidi M, Rubin J R 1997 Traumatic first rib fractures: is angiography necessary? Cardiovasc Surg 5: 48–53 7. Rao P, Carty H 1999 Non-accidental injury: review of the radiology. Clin Radiol 54: 11–24 8. Shanmuganathan K, Mirvis S E 1999 Imaging diagnosis of non-aortic thoracic injury. Rad Clin North Am 37: 533–551 9. Reid D C, Henderson R, Soboe L, Miller J D R 1987 Etiology and clinical course of missed spinal fractures. J Trauma 27: 980–986 10. Meek S 1988 Fractures of the thoraco-lumbar spine in major trauma patients. BMJ 317: 1442–1443 11. Wintermark M, Mouhsine E, Theumann N et al 2003 Thoracolumbar spine fractures in patients who have sustained severe trauma: Depiction with multi-detector row CT. Radiology 227: 681–689 12. Ma O J, Mateer J R 1997 Trauma ultrasound examination versus chest radiography in the detection of haemothorax. Ann Emerg Med 29: 312–316 13. Dee P 1992 The radiology of chest trauma. Rad Clin North Am 30: 291–306 14. Bekassy S M, Dave K S, Wooler G H, Ionescu M I 1972 Spontaneous and traumatic rupture of the diaphragm. Ann Surg 177: 320–323 15. Shackleton K L, Stewart E T, Taylor A J 1998 Traumatic diaphragmatic injuries: spectrum of radiographic findings. Radiographics 18: 49–59 16. Iochum S, Ludig T, Walter F, Sebbag H, Grosdidier G, Blum A G 2002 Imaging of diaphragmatic injury: A diagnostic challenge? RadioGraphics 22: S103–S116 17. Boulanger B R, Milzman D P, Rosati C, Rogriguez A 1993 Comparison of right and left blunt traumatic diaphragmatic injury. J Trauma 35: 255–260 18. Gelman R, Mirvis S E, Gens D 1990 Diaphragmatic rupture due to blunt trauma: sensitivity of plain chest radiographs. Am J Roentgenol 156: 51–57 19. Estrera A S, Platt M R, Mills L J 1979 Traumatic injuries of the diaphragm. Chest 75: 306–313 20. Shapiro M J, Heiberg E, Durham R M et al 1996 The unreliability of CT scans and initial chest radiographs in evaluating blunt trauma induced diaphragmatic rupture. Clin Radiol 51: 27–30 21. Worthy S A, Kang E Y, Hartman T E et al 1995 Diaphragmatic rupture: CT findings in 11 patients. Radiology 194: 885–888 22. Murray J G, Caoili E, Grunden J F et al 1996 Acute rupture of the diaphragm due to blunt trauma: diagnostic sensitivity and specificity of CT. Am J Roentgenol 166: 1035–1039 23. Bergin D, Ennis R, Keogh C et al 2001 The “dependent viscera sign” in CT diagnosis of blunt traumatic diaphragmatic rupture. Am J Roentgenol 177: 1137–1140 24. Killeen K L, Mirvis S E, Shanmuganathan K 1999 Helical CT of traumatic diaphragmatic rupture. Am J Roentgenol 173: 1611–1616 25. Shanmuganathan K, Mirvis S E, White C S, Pomerantz S M 1996 MR imaging evaluation of hemidiaphragms in acute blunt trauma: experience with 16 patients. Am J Roentgenol 167: 397–402

397

398

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

26. Wintermark M, Wicky S, Schnyder P, Capasso P 1999 Blunt traumatic pneumomediastinum: using CT to reveal the Mackilin effect. Am J Roentgenol 172: 129–130 27. Bevjan S M, Godwin J D 1996 Pneumomediastinum: old and new signs. Am J Roentgenol 166: 1041–1048 28. Chen J D, Shanmuganathan K, Mirvis S E, Killeen K L, Dutton R P 2001 Using CT to diagnose tracheal rupture. Am J Roentgenol 176: 1273–1280 29. Trerotola S O 1995 Can CT replace aortography in thoracic trauma? Radiology 197: 13–15 30. Ledbetter S, Stuk J L, Kaufman J A 1999 Helical (spiral) CT in the evaluation of emergent thoracic aorta syndromes. Rad Clin North Am 37: 575–589 31. Alkadhi H, Wildermuth S, Desbiolles L et al 2004 Vascular emergencies of the thorax after blunt and iatrogenic trauma: multi-detector row CT and three-dimensional imaging. RadioGraphics 24: 1239–1255 32. Mirvis S E, Bidwell J, Buddmeyer E 1987 Value of chest radiography in excluding traumatic aortic rupture. Radiology 163: 487–493 33. Groskin S A 1992 Selected topics in chest trauma. Radiology 183: 605–617 34. Mirvis S E, Shanmuganathan K, Miller B H, White C S, Turney S Z 1996 Traumatic aortic injury: diagnosis with contrast-enhanced thoracic CT—five-year experience at a major trauma center. Radiology 200: 413–422 35. Gavant M L, Menke P G, Fabian T, Flick P A, Graney M J, Gold R E 1995 Blunt traumatic aortic rupture: detection with helical CT of the chest. Radiology 197: 125–133 36. Wicky S, Capasso P, Meuli R, Fischer A, von Segesser L, Schnyder P 1998 Spiral CT aortography: an efficient technique for the diagnosis of traumatic aortic injury. Eur Radiol 8: 828–833 37. Cleverley J R, Barrie J R, Raymond G S, Primack S L, Mayo J R 2002 Direct signs of aortic injury on contrast enhanced CT in surgically proven traumatic aortic injury: a multi-centre review. Clin Radiol 57: 281–286 38. Driscoll P A, Hyde J A J, Curzon I et al 1996 Traumamatic disruption of the thoracic aorta: a rational approach to imaging. Injury 27: 679–685 39. Hyde J A J, Shetty A, Graham T R 1998 Mediastinal trauma. In: Driscoll P, Skinner D (eds) Trauma care: beyond the resuscitation room. BMJ Publishing, London, pp 53–66 40. Trotman-Dickenson B 1998 Radiography in the critical care patient. In: McLeod T C (ed) Thoracic radiology: The requisites. Mosby, St Louis, pp 151–172 41. White C S, Pugatach R D 2003 Thoracic imaging in the intensive care unit. In: Mirvis S E, Shanmuganathan K (eds) Imaging in trauma and critical care, 2nd edn. W B Saunders, Philadelphia, pp 725–740 42. Lipchik R J, Kuzo R S 1996 Nosocomial pneumonia. Rad Clin North Am 34: 47–58 43. Remy-Jardin M, Mastora I, Remy J 2003 Pulmonary embolus imaging with multislice CT. Rad Clin North Am 41: 507–519 44. Henry D A 1996 Radiologic evaluation of the patient after cardiac surgery. Rad Clin North Am 34: 119–135 45. Winer-Muram H T, Gurney J W, Bozeman P M, Krance R A 1996 Pulmonary complications after bone marrow transplantation. Rad Clin North Am 34: 97–118 46. Muller N L, Fraser R S, Kyung S L, Johkoh T 2003 Pulmonary edema. In: Muller N L, Fraser R S, Kyung S L, Johkoh T (eds) Diseases of the lung: radiologic and pathologic correlations. Lippincott, Williams and Wilkins, Philadelphia, pp 255–265 47. Goodman L R 1996 Congestive heart failure and adult respiratory distress syndrome: new insights using computed tomography. Rad Clin North Am 34: 33–46 48. Sutcliffe A J 1999 ARDS: pathophysiology related to cardiorespiratory management. In: Alpar E K, Gosling P (eds) Trauma: a scientific basis for care. London: Edward Arnold, London, 142–152 49. Goodman L R, Fumagalli R, Tagliabue P et al 1999 Adult respiratory distress syndrome due to pulmonary and extrapulmonary causes: CT, clinical and functional correlations. Radiology 213: 545–552

50. Desai S R, Wells A U, Suntharalingam G, Rubens MB, Evans TW, Hansell D M 2001 Acute respiratory distress syndrome caused by pulmonary and extrapulmonary injury: A comparative CT study. Radiology 218: 689–693 51. Desai S R, Wells A U, Rubens M B, Evans T W, Hansell D M 1999 Acute respiratory distress syndrome: CT abnormalities at long-term follow-up. Radiology 210: 29–35 52. Tocino I, Westcott J L 1996 Barotrauma. Rad Clin North Am 34: 59–81 53. Zylak C M, Standen J R, Barnes G R, Zylak C J 2000 Pneumomediastinum revisited. RadioGraphics 20: 1043–1057 54. Henschke C I, Yankelevitz D F, Wand A, Davis S D, Shiau M 1996 Accuracy and efficacy of chest radiography in the intensive care unit. Rad Clin North Am 34: 21–31 55. Collins J 2002 Imaging of the chest after lung transplantation. J Thorac Imaging 17: 102–112 56. Kundu S, Herman S J, Winton T L 1998 Reperfusion edema after lung transplantation: radiographic manifestations. Radiology 206: 75–80 57. Anderson D C, Glazer H S, Semenkovich J W et al 1995 Lung transplant edema: chest radiography after lung transplantation—the first 10 days. Radiology 195: 275–281 58. Collins J, Muller N L, Kazerooni E A, Paciocco G 2000 CT findings of pneumonia after lung transplantation. Am J Roentgenol 175: 811–818 59. Skeens J L, Fuhrman C R, Yousem S A 1989 Bronchilitis obliterans in heart–lung transplantation patients: radiologic findings in 11 patients. Am J Roentgenol 153: 253–256 60. Leung A N, Fisher K, Valentine V et al 1998 Bronchiolitis obliterans after lung tranplantation: detection using expiratory HRCT. Chest 113: 365–370 61. Arakawa H, Webb W R 1998 Air trapping on expiratory high-resolution CT scans in the absence of inspiratory scan abnormalities: correlation with pulmonary function tests and differential diagnosis. Am J Roentgenol 170: 1349–1353 62. Worthy S A, Muller N L, Kim J S et al 1997 Bronchiolitis obliterans after lung transplantation: high resolution CT findings in 15 patients. Am J Roentgenol 169: 673–677 63. Collins J, Muller N L, Leung A N et al 1998 Epstein–Barr-virus-associated lymphoproliferative disease of the lung: CT and histologic findings. Radiology 208: 749–759 64. Collins J, Hartman M J, Warner T F et al 2001 Frequency and CT findings of recurrent disease after lung transplantation. Radiology 219: 503–509 65. Snider G L, Kleinerman J, Thurlbeck W M 1985 The definition of emphysema: report of a National Heart, Lung and Blood Institute, Division of Lung Disease Workshop. Am Rev Respir Dis132: 182–185 66. Muller N L, Fraser R S, Lee K S, Johkoh T 2003 Emphysema. In: Muller N L, Fraser R S, Lee K S, Johkoh T (eds) Diseases of the lung: radiologic and pathologic correlations. Lippincott, Wiliams and Wilkins, Philadelphia, pp 239–254 67. Webb W R 1997 Radiology of obstructive pulmonary disease. Am J Roentgenol 169: 637–647 68. Foster W L Jr, Giminez E I, Roubidoux M A et al 1993 The emphysemas: radiologic-pathologic correlation. RadioGraphics 13: 311–328 69. Snider G L 1996 Reduction pneumoplasty for giant bullous emphysema. Implications for surgical treatment of non-bullous emphysema. Chest 109: 540–548 70. Cooper J D 1997 The history of surgical procedures for emphysema. Ann Thorac Surg 63: 312–319 71. National Emphysema Treatment Trial Research Group 2003 A randomized trial comparing lung-volume-reduction surgery with medical therapy for severe emphysema. N Engl J Med 348: 2059–2073 72. McKenna R J Jr, Fischel R J, Brenner M, Gelb A F 1996 Combined operations for lung volume reduction surgery and lung cancer. Chest 110: 885–890 73. National Emphysema Treatment Trial Research Group. Patients at High Risk of Death after Lung-Volume-Reduction Surgery. N Engl J Med 2001; 345:1075-1083. 74. Muller NL, Staples CA, Miller RR, Abboud RT 1988 “Density mask”: an objective method to quantitate emphysema using computed tomography. Chest 94: 782–787

CHAPTER 20

75. Cederlund K, Hogberg S, Jorfeldt L et al 2003 Lung perfusion scintigraphy prior to lung volume reduction surgery. Acta Radiol 44: 246–251 76. Slone R M, Gierada D, Yusen R M 1998 Preoperative and postoperative imaging in the surgical management of pulmonary emphysema. Rad Clin North Am 36: 57–89 77. Screaton N J, Reynolds J H 2006 Lung volume reduction surgery for emphysema: what the radiologist needs to know. Clin Radiol 61: 237−249

SUGGESTIONS FOR FUTHER READING Books Mirvis S E, Shanmuganathan K (eds) 2003 Imaging in trauma and critical and critical care. W B Saunders, Philadelphia

• THORACIC TRAUMA AND RELATED TOPICS

Schnyder P, Wintermark M 2000 Radiology of blunt trauma of the chest. Springer-Verlag, Berlin Journal articles Mirvis S E, Shanmuganathan K 1999 Imaging diagnosis of nonaortic thoracic injury. Rad Clin North Am 37: 533–551 Rivas L A, Fishman J E, Munera F, Bajayo D E 2003 Multislice CT in thoracic trauma. Rad Clin North Am 41: 599–616

399

CHAPTER

Airspace Diseases

21

Praveen Peddu and Sujal R. Desai

• • • • • • • •

Radiological approach to diagnosis of airspace disease Pulmonary oedema Diffuse pulmonary haemorrhage Wegener’s granulomatosis Cryptogenic organizing pneumonia Eosinophilic lung disease Alveolar proteinosis Miscellaneous causes of airspace opacification

RADIOLOGICAL APPROACH TO DIAGNOSIS OF AIRSPACE DISEASE Disease processes that principally involve the airspaces are reasonably common. Not surprisingly, airspace opacification is a frequent radiological finding. The approach to diagnosis is potentially daunting since opacification of the airspaces is a non-specific radiological sign (Table 21.1). In essence, any pathological process that displaces air from the alveoli will lead to radiographic airspace opacification; this pattern is most commonly seen when either fluid accumulates (as in pulmonary oedema) or there are inflammatory cells (as with infection) in the airspaces. The plain chest radiograph is usually the first investigation requested in patients with suspected lung disease.With experience, the radiologist can usually offer a sensible diagnosis or, at worst, a limited list of differential diagnoses. In this regard, an appreciation of the clinical features, the distribution of radiographic abnormalities and the changes on serial examination are often invaluable. A review of the radiology together with the clinical features may be diagnostic in some instances: e.g. lobar opacification in a patient with pyrexia and a productive cough is most likely to be due to infection (Fig. 21.1). Alternatively, in a patient with known left ventricular dysfunction, it would be entirely reasonable to suggest that the presence of bilateral airspace opacification is likely to represent pulmonary oedema. Recognition of the distribution of airspace opacities on the plain radiograph can be another helpful differentiating feature.

In cryptogenic organizing pneumonia, areas of consolidation may be most pronounced in the periphery and lower zones of the lungs1 (Fig. 21.2), whereas in patients with chronic eosinophilic pneumonia, the changes tend to be in the upper zones and characteristically parallel to the chest wall2. A review of serial radiographs, to ascertain disease progression or regression, should also be part of the radiologist’s routine. Relatively rapid changes (occurring over a period of hours or, at most, a few days rather than weeks) are more in keeping with pulmonary oedema (Fig. 21.3) or intra-alveolar haemorrhage than with pneumonic consolidation. On the other hand, opacities that are transient and migratory, unaccompanied by significant constitutional disturbance, should prompt the radiologist to propose a diagnosis of an eosinophilic pneumonia. Computed tomography (CT) is frequently requested in patients with airspace disease and, in the appropriate clinical context, the CT features are occasionally characteristic enough to permit a reasonably confident diagnosis. An example which comes readily to mind (with a few important caveats3) is the so-called ‘crazy-paving’ pattern of alveolar proteinosis4. However, in other instances, the radiologist may only be able to limit the list of diagnostic possibilities based on the additional information from CT, e.g. CT may reveal cavitation that may not have been evident on plain radiographs. In such a case it would be reasonable to propose Wegener’s granulomatosis or septic emboli as possible diagnoses. Except in the select circumstances highlighted above, the advantages of CT over plain radiography in the diagnosis of airspace diseases are not clearly defined5. This chapter considers some of the common and a few of the more unusual causes of airspace opacification in clinical practice. Airspace diseases due to infection and malignancy are considered in detail elsewhere (see Chapters 15 and 18).

Anatomical considerations Before considering the causes of airspace opacification it is pertinent to revise relevant aspects of pulmonary anatomy.The terminal bronchioles are the last purely conducting airways of the bronchial tree6 and the region of lung subtended by a terminal bronchiole is termed the acinus (comprising the

402

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 21.1 CAUSES OF AIRSPACE OPACIFICATION • Oedema Cardiogenic Noncardiogenic • Inflammation/infection Wegener’s granulomatosis Cryptogenic organizing pneumonia • Blood Idiopathic pulmonary haemosiderosis Antibasement membrane antibody disease Systemic lupus erythematosus • Miscellaneous causes Eosinophilic pneumonia Alveolar proteinosis Alveolar cell carcinoma Alveolar microlithiasis Lymphoma (MALToma) Sarcoidosis

Figure 21.2 (Noninfective) organizing pneumonia in a patient following allogeneic bone marrow transplantation. (A) Chest radiograph and (B) CT through the lower zones demonstrating characteristic bilateral mid and lower zone airspace opacification.

Figure 21.1 Pneumonia. Chest radiograph in a young male patient. The presence of focal (cavitating) consolidation in a patient with pyrexia and a productive cough should prompt a radiological diagnosis of pneumonia.

respiratory bronchioles, alveolar ducts, alveolar sacs and alveoli). Important pathways of collateral ventilation (the pores of Kohn) link different alveolar units7 and help to maintain lung inflation in the presence of proximal airway obstruction. Thus certain disease processes (most notably infections) can spread readily into adjacent alveolar units. Another important unit of lung structure, the secondary pulmonary lobule, is the smallest unit of lung bounded by

connective tissue septa8. Individual lobules are irregular polyhedral units, best seen sub-pleurally, measuring between 5 and 30 mm in diameter, and incorporating 3–24 acini9; the centrilobular bronchiole and adjacent artery form the core structures. The centrilobular arteries (with a diameter of 0.2 mm) can be resolved on high-resolution computed tomography (HRCT) in the normal lung, whereas normal bronchioles with a diameter below 2 mm are generally not seen10. Infiltration of the interlobular interstitium by oedema fluid or malignant cells, or thickening by fibrosis, will render individual secondary pulmonary lobules visible on HRCT11.

Radiological signs of airspace disease One of the limitations of imaging studies such as chest radiography and CT is that a multitude of pathological processes

CHAPTER 21

• AIRSPACE DISEASES

Figure 21.3 (A–C) Rapid changes in radiographic appearances in pulmonary oedema. Serial chest radiographs over roughly a 48-h period show striking changes in the extent/severity of airspace opacification reflecting relatively rapid shifts of fluid between the intravascular compartment and the airspaces/interstitium.

in the airspaces manifest in only a limited number of ways: thus, for most airspace diseases, a nodular pattern, ground-glass opacification and consolidation represent the range of radiological abnormalities. A nodular pattern (defined as rounded opacities that are at least moderately well differentiated and no greater than 3 cm in maximum diameter12) as a sole manifestation of airspace disease is relatively uncommon. Historically, the term ‘acinar nodules’ or ‘acinar rosettes’ has been used to describe the appearance of a patchy nodular infiltrate on chest radiography13 (Fig. 21.4). However, the diagnostic value of localizing disease to the acinus has been questioned; in pathological studies, the acinar pattern on plain radiographs does not correspond to filling of acini as defined

anatomically14. This notwithstanding, the so-called acinar pattern is most frequently encountered in the context of bacterial infection15. Ground-glass opacification is a common manifestation of airspace disease on chest radiography and CT. On plain radiography ground-glass opacification is seen as an increase in lung density which obscures vessel markings12. By comparison, because of the greater contrast medium resolution of CT, this pattern appears as a hazy increase in lung attenuation but with preservation of the bronchial and vascular markings12 (Fig. 21.5); there may or may not be an associated air bronchogram. However ground-glass opacification is a non-specific radiological sign which may indicate disease within the airspaces and/or the interstitium16 (Fig. 21.6). Occasionally, ground-glass opacification on CT is extremely subtle. In cases of uncertainty, comparison of the (air) density within airways with that of lung parenchyma (the ‘black bronchus’ sign) may be useful (Fig. 21.7); normally the two densities are roughly comparable. Consolidation refers to an increase in parenchymal density in which the margins of vessels and airways are obscured both on chest radiography and CT (Fig. 21.8). As with ground-glass opacification, an air bronchogram may be seen. This radiological pattern occurs when air in the airspaces is replaced with either transudate or exudate.

PULMONARY OEDEMA

Figure 21.4 Nodular airspace opacities on chest radiography. Targetted image of the right lung showing numerous ill-defined nodules in a patient with disseminated pulmonary tuberculosis.

Pulmonary oedema, defined as an excess of extravascular lung water, may be due to an increase in hydrostatic pressure (sometimes called cardiogenic oedema) or vascular permeability (termed noncardiogenic oedema) (Table 21.2). Whilst being practical, the utility of this apparently simple and dichotomous classification of pulmonary oedema has been questioned17,18. Cardiogenic or hydrostatic pulmonary oedema occurs when there is shift of fluid out of the vascular compartment secondary to an increase in pulmonary venous and capillary pressure. A common cause of increased hydrostatic pressure is left heart failure but rarely, a reduction in plasma osmotic pressure (as in hypoalbuminaemic states) will lead to pulmonary oedema.

403

404

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 21.5 Ground-glass opacification on (A) chest radiography and (B) CT in a patient with pulmonary cytomegalovirus reactivation following bone marrow transplantation. On plain chest radiography there is veil-like opacification in the mid and lower zones of both lungs, which obscures vessel markings. In comparison, on CT, bronchovascular markings are clearly visible within regions of increased opacification.

Figure 21.6 Variable causes of ground-glass opacification on CT due to airspace and/or interstitial processes. (A) Diffuse pulmonary haemorrhage in a patient with disseminated malignancy. (B) Ground-glass opacification and thickened inter- and intra-lobular septa in pulmonary oedema. (C) Ground-glass opacification due to a predominant interstitial lung disease; there are grossly dilated segmental and subsegmental airways (‘traction bronchiectasis’) within ground-glass indicating fine fibrosis.

Noncardiogenic pulmonary oedema occurs when there is an increase in the permeability of the alveolar–capillary barrier. A good example of this type of pulmonary oedema is the acute respiratory distress syndrome (ARDS).

Chest radiographic signs

Figure 21.7 ‘Black bronchus’ sign on CT: there is an almost imperceptible increase in lung density. However, air within the segmental bronchi is clearly of lower attenuation than the surrounding parenchyma.

Plain chest radiography is more sensitive than clinical examination in the detection of early pulmonary oedema19. Indeed, chest radiography is an important investigation in patients with suspected pulmonary oedema. Because fluid passes from the intravascular compartment into the interstitium and only then into the alveoli, the radiographic changes of interstitial oedema generally precede frank airspace opacification20. Whilst the division into interstitial and alveolar oedema is perhaps too simplistic, for clarity these will be discussed separately in the following sections.

CHAPTER 21

• AIRSPACE DISEASES

Interstitial oedema. One of the classical radiographic manifestations of interstitial oedema is thickening of the interlobular septa. The characteristic Kerley B lines, which represent fluid in the interlobular septa (typically 1–2-mm wide and 30–60-mm long), are mainly seen in the sub-pleural lung, perpendicular to the pleural surface (Fig. 21.9). In comparison, Kerley A lines are longer (up to 80–100 mm), occasionally angulated and cross the inner two-thirds of the lung in varying directions but tend to point medially towards the hilum. In left heart failure, septal lines become visible as they distend with increasing extravascular fluid. Note that the visualization of oedematous septal lines may be hampered when neighbouring alveoli are also opacified. However, the demonstration of thickened interlobular septa is not diagnostic of pulmonary oedema; fibrosis and malignant infiltration (as in lymphangitis carcinomatosa) may also increase the conspicuity of interlobular septa. Another sign of interstitial oedema on chest radiographs is peribronchial cuffing in which, on a frontal radiograph, the normally thin and well defined wall of the airway appears thickened and somewhat indistinct (Fig. 21.10). A loss of conspicuity of the central pulmonary vessels (termed a perihilar haze) also occurs and, as with peribronchial cuffing, is assumed to be due to oedema of the perivascular interstitium. Because of the anatomical arrangement of septa in relation to the visceral pleura, oedema fluid may collect in the sub-pleural space. On chest radiography this is manifest as thickening of the interlobar fissures or as a lamellar ‘effusion’ in the costophrenic recesses.

Figure 21.8 (A) Consolidation on plain chest radiography in a patient with proven tuberculosis; there is evidence of cavitation within the area of consolidation. (B) Consolidation on CT in a patient with adenovirus infection following allogeneic bone marrow transplantation. In both cases, there is obscuration of bronchovascular markings by areas of dense parenchymal opacification.

Table 21.2

CAUSES OF PULMONARY OEDEMA

• Cardiogenic (hydrostatic) oedema • Noncardiogenic (increased permeability) oedema (Iatrogenic) fluid overload Drowning Drug-induced (e.g. narcotics, non-steroidal anti-inflammatory drugs, intravenous contrast) Acute respiratory distress syndrome High altitude Rapid re-expansion of collapsed lung Intracranial disease (cerebrovascular accident, raised intracranial pressure)

Figure 21.9 Magnified view of the left costophrenic region demonstrating multiple interstitial (Kerley B) lines. Each line is roughly perpendicular to the chest wall and extends to the pleural surface.

405

406

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 21.11 Asymmetrical distribution of airspace opacification in a patient with pulmonary oedema. There is patchy opacification in the right lung with relative sparing of the left.

Figure 21.10 (A) Normal appearance of segmental bronchus seen ‘end-on’ (arrow) adjacent to its accompanying pulmonary artery branch. The wall of the bronchus is well defined and thin. (B) Peribronchial cuffing. The wall of the anterior segmental bronchus appears thickened and ill-defined (arrows) in early interstitial oedema due to (iatrogenic) fluid overload.

Alveolar oedema Airspace opacification becomes evident on the chest radiograph as oedema fluid passes from the interstitium into the alveoli. The distribution of changes is variable and frequently random (Fig. 21.11) but, in general, there is sparing of the apices and extreme lung bases. Typically, there is bilateral opacification, although pulmonary oedema apparently confined to one lung on the chest radiograph also occurs21. On occasion, the central lungs are more affected, producing the characteristic ‘butterfly’ or ‘bat’s wing’ distribution (Fig. 21.12)22. As oedema progresses, opacities may coalesce to produce a general ‘whiteout’. An air bronchogram or alveologram may be seen when

Figure 21.12 Pulmonary oedema on chest radiography demonstrating the characteristic ‘bat’s wing’ distribution, with airspace opacification principally within the central lung.

there is intra-alveolar oedema. Resolution of airspace opacification on serial radiographs may be rapid (hours compared to days or weeks) and is a useful indicator that airspace disease is due to pulmonary oedema rather than any other process (Fig. 21.13). Redistribution of blood to the upper zones is seen in some patients with elevated pulmonary venous pressure23; when this occurs, vessels in the upper zones appear larger than comparable vessels in the lower zones (this appearance is called upper lobe blood diversion) (Fig. 21.14). Although not an invariable

CHAPTER 21

• AIRSPACE DISEASES

Figure 21.13 (A–C) Serial chest radiographic changes in pulmonary oedema. There is progressive clearing of airspace opacification over a 48-h period.

feature, the redistribution of blood is a recognized feature in both acute and chronic cardiac dysfunction24.The mechanisms underlying blood flow redistribution are interesting but not entirely clear; however, in the upright patient, oedema accumulates in the more dependent lung and it has therefore been suggested that interstitial oedema at the bases may compress the vessels, thereby increasing the resistance to flow and preventing the transmission of hydrostatic distending forces25. In some patients the radiographic appearances of intraalveolar oedema will be modified. The distribution of pulmonary oedema may vary with posture; in patients lying on one side for a prolonged period, for example, the dependent lung becomes more oedematous and pulmonary oedema may then be unilateral. Coexisting disease can also affect the distribution of extravascular lung water. In smokers, the distribution of oedema may be patchy owing to the destruction of lung parenchyma by emphysema. Similarly, in patients with pulmonary fibrosis, there may be more rapid fluid accumulation in the alveolar spaces in comparison with the interstitium26.

Radiographic differentiation of cardiogenic and noncardiogenic pulmonary oedema Plain chest radiography is probably more accurate than clinical examination in distinguishing between hydrostatic and increased permeability oedema27. However, whether chest radiography can consistently differentiate between cardiogenic and noncardiogenic oedema is more debatable28,29. In one early study, the pattern of blood flow (i.e. upper zone versus lower zone), the distribution of oedema (i.e. peripheral versus central) and the width of the vascular pedicle were considered to be discriminatory features28. In 50% of patients with cardiogenic oedema there was upper lobe blood diversion, whereas in patients with increased permeability oedema due to ARDS, only 10% showed this inverted pattern; normal or ‘balanced’ flow were more commonly seen in ARDS28. A peripheral distribution of oedema was strikingly absent in patients with cardiogenic oedema but was the most common pattern seen in patients with ARDS. Based on these findings and some ancillary features, the authors claimed an overall accuracy for chest radiography of 86–89%28. However, this conclusion has been refuted by the findings of subsequent studies29,30. Although there is high specificity for the finding of patchy or peripherally-distributed oedema in increased permeability oedema, Aberle et al showed that the sensitivity was below 50%29. Similarly, the discriminatory value of signs of interstitial fluid accumulation and pleural effusions have been questioned29,31. In summary, analysis of the radiographic pattern will sometimes allow a distinction to be made but the inconsistency of the radiographic signs suggests that differentiation between the various forms of pulmonary oedema, based on radiographic features alone, is unreliable.

Computed tomography

Figure 21.14 Upper lobe blood diversion. Vessels in the upper zones (arrows) are prominent in comparison to those in the lower lung zones.

The appearances of pulmonary oedema on CT are variable. As with chest radiography, the changes on CT may be bilateral or indeed confined to one lung and may be modified by co-existent disease. Because of its superior contrast medium resolution, CT may detect abnormalities before the transudation of fluid into the interstitium and airspaces: in an animal model of fluid overload there was an increase in background lung attenuation, attributed to an expansion of intracapillary

407

408

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

volume32. With more florid transudation, peribronchovascular cuffing, prominent interlobular septa, ground-glass opacification and consolidation become evident (Fig. 21.15).A perihilar distribution may be seen in some patients but is by no means invariable.

Table 21.3 CLASSIFICATION OF DIFFUSE PULMONARY HAEMORRHAGE SYNDROMES (MODIFIED FROM33) Nonimmunocompromised patients • Antibasement membrane antibody disease/Goodpasture’s syndrome • Diseases of presumed immune aetiology, with or without nephropathy Systemic lupus erythematosus

DIFFUSE PULMONARY HAEMORRHAGE

Rheumatoid arthritis Systemic sclerosis

Bleeding into the airspaces is a surprisingly frequent event in many common pulmonary diseases. It is well known that patients with lung cancer, bronchiectasis, or pneumonia frequently aspirate blood into the airspaces. However, because the bleeding tends to be localized in these cases and there is often an established underlying cause, the diagnosis is generally straightforward. In addition to these common clinical scenarios there are a number of pulmonary haemorrhage syndromes characterized by diffuse intra-alveolar bleeding. The severity of haemorrhage is variable, ranging from small symptomless bleeds to life-threatening episodes. A practical scheme for classifying diffuse pulmonary haemorrhage (DPH) categorizes syndromes according to whether or not the patient is immunocompromised33 (Table 21.3). In immunocompetent patients, DPH may be immunologically mediated (e.g. antiglomerular basement membrane disease), have a presumed immunological basis (e.g. systemic lupus erythematosus,Wegener’s granulomatosis), or have a nonimmunological mechanism (e.g. idiopathic pulmonary haemosiderosis, drug reactions). In immunocompromised patients, infection, tumours and blood dyscrasias account for the majority of cases of DPH. The clinical presentation of the DPH syndromes varies34. Recurrent haemoptyses, dyspnoea and chronic cough are the typical, but not invariable, symptoms. Non-specific clinical features, including intermittent pyrexia, headache, lethargy, basal crackles on auscultation and clubbing, may also be present. On

Systemic necrotizing vasculitis Wegener’s granulomatosis Microscopic polyarteritis • Diseases with no known immune aetiology Idiopathic pulmonary haemosiderosis Rapidly progressive glomerulonephritis without immune complexes Fibrillary glomerulonephritis Drug-induced (anticoagulants, trimellitic anhydride, cocaine) Valvular heart disease Disseminated intravascular coagulation Acute lung injury Tumours Immunocompromised patients • Blood dyscrasias • Infection • Tumours

histopathological examination blood and haemosiderin-laden macrophages are the key findings within the alveoli35. With repeated episodes there is thickening of alveolar septa due to fibrosis36, which can occasionally become florid37. Of the many DPH syndromes, idiopathic pulmonary haemosiderosis and antibasement membrane antibody disease (Goodpasture’s syndrome) have received greatest attention. These two entities are considered briefly below.

Idiopathic pulmonary haemosiderosis Idiopathic pulmonary haemosiderosis (IPH) is a rare disorder of unknown aetiology.The clinical picture is that of episodic intraalveolar haemorrhage, haemoptyses, iron-deficiency anaemia and airspace opacification on chest radiographs. The majority of patients are children (typically in the first decade) or young adults in the second to third decades36,38; sporadic cases in older subjects have been recorded39.The prognosis of patients with IPH varies: survival may range from a few days (following a severe episode of bleeding) to years.The pathogenesis of IPH is not clear although a number of hypotheses have been proposed38.

Antibasement membrane antibody disease

Figure 21.15 Pulmonary oedema on CT. There is diffuse groundglass opacification, smooth thickening of multiple interlobular septa and peribronchovascular cuffing. Bilateral pleural effusions are also seen.

The link between renal disease and diffuse intra-alveolar bleeding has long been established40.Whilst the historical term Goodpasture’s syndrome (referring to the entity in which circulating antibodies are directed against components of basement membrane in the lungs and kidneys) remains in common parlance, the term antibasement membrane antibody (ABMA) disease is now preferred to the eponymous title. ABMA disease

CHAPTER 21

typically affects young men with a male:female ratio of around 3:141. The pulmonary manifestations of ABMA disease often dominate the clinical presentation, though evidence of renal disease is present in the majority of cases.

Radiology The plain radiographic features of DPH syndromes are similar regardless of the underlying aetiology. Therefore, differentiation between different causes of DPH is generally not possible on the basis of radiological findings alone. Following an acute episode the chest radiograph is usually, but not invariably, abnormal. However, in individual patients, the radiographic appearances will vary depending on the extent of haemorrhage and the timing of the examination relative to individual bleeding episodes: acutely there is widespread ground-glass opacification or consolidation42. The changes tend to be more pronounced in the perihilar region of the mid and lower zones. A useful diagnostic clue is that, compared to other causes of airspace opacification (with the notable exception of pulmonary oedema), the changes of diffuse intra-alveolar haemorrhage typically clear over a few days. This is a consequence of the relatively efficient removal of blood by pulmonary alveolar macrophages (Fig. 21.16). With repeated episodes ill-defined nodular or reticulonodular opacities are seen and there may be enlargement of hilar lymph nodes. The CT abnormalities in the ‘subacute’ phase have been described in one study of six patients43. A nodular pattern with no zonal predilection was the dominant CT finding. Nodules measured 1–3 mm in diameter and corresponded on histopathological examination to conglomerates of haemosiderin-laden macrophages within alveoli. There was patchy or uniform ground-glass opacification and abnormal thickening of interlobular septa in the majority of cases.

WEGENER’S GRANULOMATOSIS Wegener’s granulomatosis is a multisystem disease of unknown aetiology characterized by varying proportions of (A) a necrotizing vasculitis involving medium- and small-sized vessels; (B) granulomatous necrosis; and (C) elements of both acute and chronic inflammation44. The disease affects men and women about equally and can present at virtually any age; indeed, there are reports of Wegener’s granulomatosis presenting both in childhood and in patients in the eighth decade45–47. The clinical manifestations of Wegener’s granulomatosis are both varied and non-specific48. However, the majority of patients complain of respiratory symptoms. The upper and lower respiratory tract are commonly involved and the typical patient with Wegener’s granulomatosis will also has evidence of renal disease. A ‘limited’ form of Wegener’s granulomatosis, in which disease is confined to the respiratory tract, is also recognized49. Without treatment, the disease is almost invariably fatal50. However, the outlook for patients with Wegener’s granulomatosis has improved dramatically with cyclophosphamide and corticosteroid therapy.

• AIRSPACE DISEASES

Radiology The spectrum of radiographic abnormalities in Wegener’s granulomatosis is wide. Nodules are seen in most patient and vary in size from a few millimeters to several centimeters in diameter51,52. There is no particular zonal predilection (Fig. 21.17). In most patients there are multiple nodules but in some, there may be a solitary pulmonary lesion. On CT a feeding vessel leading to the nodule may be identified and there may be air bronchograms within lung nodules. Linear bands, spiculation and pleural tags may also be seen in relation to pulmonary nodules. With treatment, nodules generally regress, but it is important to remember that the chest radiograph may not return to normal for up to a month following the initiation of therapy53. Furthermore, there may be residual parenchymal scarring on CT despite resolution of nodules and pulmonary consolidation54. Cavitation is generally regarded as the classical radiological finding in pulmonary Wegener’s granulomatosis and the demonstration of this sign is a useful pointer to the diagnosis (Fig 21.18). However, cavitation is by no means an invariable feature and thus, the absence of this sign should not preclude a diagnosis of Wegener’s granulomatosis. Consolidation and ground-glass opacities are recognized features on CT but are less common than nodules. Interestingly, the converse is apparently true in children, in whom nodules are seen less frequently55. As with nodules, foci of consolidation may cavitate. With CT it has been possible to highlight the more unusual abnormalities in Wegener’s granulomatosis such as bronchovascular thickening and frank bronchiectasis (Fig. 21.19). The other reported ancillary features of Wegener’s granulomatosis include areas of lobar or segmental atelectasis, pleural effusions or thickening, and rarely, hilar and mediastinal lymph node enlargement.

CRYPTOGENIC ORGANIZING PNEUMONIA The entity of cryptogenic organizing pneumonia (COP), first reported by Davison et al, refers to the histological pattern of an organizing pneumonia of unknown aetiology56. The characteristic finding on biopsy is of buds of granulation tissue in the alveoli and alveolar ducts with cellular infiltrates (mainly mononuclear cells and macrophages) in the surrounding airspaces56. It is important to realize, at the outset, that organizing pneumonia is simply a histological pattern and not a diagnosis in itself; the label ‘cryptogenic’ is only applied when other potential causes of an organizing pneumonia pattern (including a host of infections, certain drugs and connective tissue disorders) have been excluded57. In the past, the more confusing term idiopathic brochiolitis obliterans organizing pneumonia (BOOP) has also been used to describe this histological entity58. However, because the airway changes are thought to be secondary to the dominant process in the airspaces and there is no evidence of a ‘bronchiolitis obliterans’, the term COP is now preferred59. Patients with COP usually present with a short history of cough, dyspnoea,

409

410

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 21.16 (A–C) Chest radiographs and (D) HRCT in a patient with idiopathic pulmonary haemosiderosis. (A) There is diffuse groundglass opacification with no zonal predilection during an acute hospital admission with haemoptysis. (B) Radiograph taken 4 d later shows striking (but incomplete) resolution of airspace opacities. There is residual opacification around the right hilum. (C) Chest radiograph obtained 1 month following admission demonstrates ground-glass opacification in the right mid zone. (D) HRCT through the lower zones (concurrent with the radiograph in [A]) shows patchy ground-glass opacification with a somewhat unusual geographical distribution; there are thickened interlobular septa in areas of groundglass opacification.

CHAPTER 21

• AIRSPACE DISEASES

Figure 21.17 Wegener’s granulomatosis. (A) Multiple intrapulmonary nodules within both lungs on chest radiography in a patient with elevated c-ANCA titres. (B) CT through the lower zones also demonstrates multiple pulmonary nodules. However, there is clear evidence of cavitation (not seen on the chest radiograph) in one of the lesions in the right lower lobe.

Figure 21.18 HRCT through the lower lobes in a patient with biopsy-proven Wegener’s granulomatosis. There is multifocal consolidation and a thick-walled cavity in the left lower lobe.

Figure 21.19 Bronchocentric disease in Wegener’s granulomatosis; many of the segmental and subsegmental bronchi are thick walled (arrows). (Courtesy of Dr Kate Pointon, Nottingham City Hospital, UK.)

fever, weight loss, chills and myalgia. Whilst spontaneous resolution can occur, most patients treated with corticosteroids undergo relatively rapid and generally complete response occurs60. Disease relapse is relatively common but does not appear adversely to influence the long-term prognosis.

vascular regions in around two-thirds of patients (Fig. 21.21) but cavitation is only rarely seen. More recently, multifocal areas of ground-glass opacification with a surrounding rim of consolidation (known as the ‘reverse halo’ sign) have also been described63. Nodules, sometimes measuring up to 1 cm in diameter and representing focal areas of organizing pneumonia, are seen in some patients and, in occasional subjects, these may be the sole radiographic manifestation of COP61,62; as with areas of consolidation, there is no definite zonal predilection. In some patents nodules are large enough to simulate a neoplasm64. Murphy et al have described linear opacities as the dominant HRCT abnormality in some patients with COP65. Two types of opacity (termed Type I and II) were identified. Type I opacities are intimately related to bronchi and extend radially towards the pleura. Type I opacities measure 2–4 cm in

Radiology The radiological signs of COP may be predicted from a knowledge of the histopathological changes. Bilateral patchy areas of consolidation, which tend to be peripheral, are the cardinal features on chest radiography61 (Fig. 21.20). Although earlier radiographic series suggested a predilection for the mid and lower zones, data from CT indicate that all lung zones may be equally affected62. The changes of COP may be confined to one lung but this is an infrequent finding61. Consolidation in COP has a propensity for the sub-pleural and/or peribroncho-

411

412

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 21.20 Cryptogenic organizing pneumonia. (A) Patchy multifocal consolidation in the mid and upper zones in a patient presenting with a cough, breathlessness and weight loss. (B) One month after commencing corticosteroids there is complete clearing of the airspace opacities. (C) Five months after stopping treatment there is recurrence with consolidation in the right mid zone.

length and 1–2 mm in thickness. Areas of consolidation may coexist with Type I linear opacities. Type II linear opacities are sub-pleural and, unlike the Type I pattern, are not related to airways.Type II linear opacities tend to be parallel to the pleural surface but like Type I changes are frequently associated with multifocal airspace consolidation. Importantly, regardless of the type of opacity, there was radiological resolution on follow-up in the majority of patients. Another unusual manifestation of COP on CT, recently reported by Ujita et al, is a perilobular pattern in which curvilinear opacities (of greater thickness but less well defined than interlobular septa) give rise to an arcadelike appearance66 (Fig. 21.22).

EOSINOPHILIC LUNG DISEASE The eosinophilic lung diseases are a diverse group which, like the DPH syndromes, have proved difficult to classify. In one early attempt the pulmonary eosinophilias were defined as any condition in which pulmonary infiltration on the radiograph is accompanied by blood eosinophilia in a patient in whom pneumonia, hydatid disease of the lung, Hodgkin’s disease and sarcoidosis had been excluded67. A simplified classification of the pulmonary eosinophilias is given in Table 21.468, and some of the entities encompassed within the broad category of ‘eosinophilic pneumonia’ are discussed below.

Simple pulmonary eosinophilia (Löffler’s syndrome) In 1932, Löffler reported on a series of patients who presented with evidence of infiltrates on chest radiography, two of whom had a peripheral blood eosinophilia69. The term Löffler’s syndrome (synonymous with simple pulmonary eosinophilia) describes patients with transient radiographic infiltrates, minimal constitutional upset and an elevated eosinophil count in peripheral blood. The airspace opacification in Löffler’s syndrome is fleeting and may be either unior bi-lateral. Resolution of opacities within a period of days and, by definition, within a month is the rule. In many cases, no underlying aetiological factor is uncovered but there is an association with parasitic infection, in particular infestation with Ascaris lumbricoides.

Acute eosinophilic pneumonia Some patients with pulmonary eosinophilia have a more fulminant clinical course. In these subjects, there may be a brief history of a febrile illness, followed by respiratory distress with marked hypoxia, diffuse airspace opacification on chest radiography and elevated eosinophil levels in broncho-alveolar fluid; the term acute eosinophilic pneumonia has been applied to this entity70, which can occur at any age including in childhood. The clinical improvement with corticosteroids is often dramatic with fever and radiographic changes resolving within days and with very little risk of relapse on withdrawal of ther-

Figure 21.21 Cryptogenic organizing pneumonia (COP). CT through the (A) apex, (B) carina and (C) lower lobes in COP. Multifocal consolidation and ground-glass opacification are seen, predominantly in the peripheral lung. Note that all lung zones are affected.

CHAPTER 21

• AIRSPACE DISEASES

Table 21.4 MODIFIED CLASSIFICATION OF EOSINOPHILIC LUNG DISEASE • Idiopathic Simple pulmonary eosinophilia (Löffler’s syndrome) Acute eosinophilic pneumonia Chronic eosinophilic pneumonia Hypereosinophilic syndrome • Drug-induced Aminosalicylic acid Para-aminosalicylic acid Non-steroidal anti-inflammatory drugs Captopril Cocaine Minocycline Nitrofurantoin Phenytoin • Infection Parasitic (ascariasis, paragonomiasis, tropical eosinophilia) Fungal (aspergillus) Bacterial (tuberculosis, atypical mycobacterial infection, brucella) Viral (respiratory syncytial virus) • Immunological diseases Wegener’s granulomatosis Churg–Strauss syndrome Rheumatoid disease Sarcoidosis • Neoplasms

Figure 21.22 Cryptogenic organzing pneumonia in a young female patient. CT at (A) the level of the carina and (B) the lung bases demonstrating an arcade-like appearance due to perilobular curvilinear opacities.

apy. Spontaneous resolution of acute eosinophilic pneumonia, without therapy, also occurs71. On plain chest radiographs, there is bilateral airspace opacification and/or reticular infiltrates72. Pleural effusions are common. Areas of ground-glass opacification and consolidation are seen on CT and there may be smooth thickening of interlobular septa.

Chronic eosinophilic pneumonia The clinical and radiological features of chronic eosinophilic pneumonia are strikingly different from the entities described above. As the term suggests, the clinical course of chronic eosinophilic pneumonia is generally more protracted and the symptoms are often more marked than in patients with simple pulmonary eosinophilia. Pyrexia, cough, breathlessness, weight loss and night sweats are the common clinical features of chronic eosinophilic pneumonia; occasional patients also complain of haemoptysis and chest pain68. There is frequently eosinophilia in the peripheral blood and lung function tests reveal a restrictive defect with impaired gas transfer. Fortunately the prognosis is good and most patients respond to corticosteroid therapy with rapid clearing of their radiographic

Brochogenic carcinoma Bronchial carcinoid Lymphoma (Hodgkin’s, non-Hodgkin’s)

infiltrates, though the withdrawal of, or a reduction in, such therapy is not uncommonly associated with relapse. The plain radiographic abnormalities in chronic eosinophilic pneumonia may be characteristic: patchy, nonsegmental areas of consolidation are typical in the mid and upper zones2. A distinctive feature is that the opacities are peripheral and seem to parallel the chest wall, a finding that has been called the ‘photographic negative of pulmonary oedema’ (Fig. 21.23)2. Not surprisingly, the peripheral location of the airspace opacities is more readily appreciated on CT.

ALVEOLAR PROTEINOSIS Alveolar proteinosis (synonymous with alveolar lipoproteinosis and alveolar phospholipoproteinosis) is a rare disease characterized by the accumulation of a periodic acid-Schiffpositive lipoproteinaceous material within the alveoli73. The aetiology of alveolar proteinosis is unknown, although a defect of surfactant metabolism has long been suspected. In some cases, alveolar proteinosis is consequent on a pulmonary insult: exposure to inorganic dusts and some infections has been associated with so-called secondary alveolar proteinosis. Interestingly,

413

414

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 21.23 Chronic eosinophilic pneumonia. (A) Chest radiograph in a male patient with a dry cough, weight loss and marked blood eosinophilia. There is ill-defined airspace opacification in both upper zones and the changes appear to parallel the chest wall. (B) HRCT through the upper lobes with bilateral (but asymmetrical) areas of ground-glass opacification and a fine reticular component in a peripheral distribution.

there is also a recognized link between alveolar proteinosis and some adult haematological malignancies (i.e. lymphoma and leukaemia), and with immunodeficiency states in children74. The disease usually affects adults aged 20–50 years, but has been described in children, in whom the prognosis is generally worse. Men are affected more frequently than women. The clinical manifestations of alveolar proteinosis are variable but exertional dyspnoea and a non-productive cough are the most common symptoms; pyrexia, chest pain and haemoptysis are occasionally reported. Patients with alveolar proteinosis may have digital clubbing and inspiratory crackles can be heard on auscultation. Paradoxically, in some patients, there may be minimal clinical signs despite extensive radiographic changes. The diagnosis of alveolar proteinosis may be supported by the electron microscopic features of lung tissue, lavage samples or sputum, and histopathological examination is then required only in cases of uncertainty. Although spontaneous resolution may occur, the majority of patients require therapeutic (whole lung) broncho-alveolar lavage, a technique that has improved the outlook of patients with alveolar proteinosis75. Symptomatic improvement following lavage may be rapid and sustained but repeated treatments are frequently required.

Radiology The chest radiographic changes of alveolar proteinosis are nonspecific. In general, both lungs are involved and airspace opacification is most pronounced in the central lung (Fig. 21.24).The CT features are much more suggestive (although not entirely pathognomonic) of alveolar proteinosis: on thin-section images a ‘crazy-paving’ pattern (made up of a striking geographical distribution of ground-glass opacification and thickened interlobular septa) is the characteristic feature4 (Fig. 21.25). This pattern, which has traditionally been associated with alveolar proteinosis, is not specific: an indistinguishable CT appearance can be seen in some patients with (mucinous) bronchiolo-alveolar carcinoma, exogenous lipoid pneumonia and a variety of other diffuse lung diseases3.

Figure 21.24 Alveolar proteinosis. Chest radiograph in a 28 year old woman with alveolar proteinosis. There is bilateral symmetrical airspace opacification in the mid and lower zones with a predilection for the central lung, an appearance which can simulate the ‘bat’s wing’ appearance of pulmonary oedema.

MISCELLANEOUS CAUSES OF AIRSPACE OPACIFICATION Pulmonary alveolar microlithiasis is characterized by the deposition of tiny stones (calcipherites) within alveoli. The calcipherites are composed mainly of calcium and phosphorus and are of unknown origin. There is a characteristic radiographic appearance in which innumerable discrete high-density opacities (resembling grains of sand) are seen in both lungs; when profuse there may be a ‘white-out’ and the tiny stones are then best demonstrated on an overexposed radiograph. The disease exhibits a strong familial tendency, and whilst it has been recorded over a wide age range (the peak incidence is between 30 and 50 years), it seems likely that the process begins in early life. Most patients

CHAPTER 21

• AIRSPACE DISEASES

are asymptomatic at the time of diagnosis and disease progression is variable. However, there is a tendency for pulmonary fibrosis and the development of cor pulmonale. The fibrosis of alveolar microlithiasis is associated with the formation of bullae, particularly at the apices. Airspace opacification on chest radiography and CT is a feature of some malignant pulmonary diseases. In patients with bronchiolo-alveolar cell carcinoma (depending to some extent on the histological subtype), there may be consolidation and/ or ground-glass opacities. Similarly, multifocal consolidation is seen on CT in primary pulmonary lymphoma (Fig. 21.26), a rare malignancy (previously called ‘pseudolymphoma’) arising from mucosa-associated lymphoid tissue and considered to be a low grade B-cell lymphoma. Figure 21.25 HRCT of the ‘crazy paving’ pattern in alveolar proteinosis: patchy but geographical ground-glass opacification is seen and there are numerous thickened interlobular septa in areas of groundglass opacification.

Figure 21.26 Primary pulmonary lymphoma. (A) Chest radiograph and (B) CT through the lower zones in a patient with pulmonary non-Hodgkin’s lymphoma. There are bilateral foci of consolidation; the bronchocentric nature of the disease can be easily appreciated on CT.

REFERENCES 1. Haddock J A A, Hansell D M 1992 The radiology and terminology of cryptogenic organizing pneumonia. Br J Radiol 65: 674–680 2. Gaensler E A, Carrington C B 1977 Peripheral opacities in chronic eosinophilic pneumonia: the photographic negative of pulmonary oedema. Am J Roentgenol 128: 1–13. 3. Johkoh T, Itoh H, Müller N L et al 1999 Crazy-paving appearance at thin-section CT: spectrum of disease and pathologic findings. Radiology 211: 155–160 4. Murch C R, Carr D H 1989 Computed tomography appearances of pulmonary alveolar proteinosis. Clin Radiol 40: 240–243 5. Johkoh T, Ikezoe J, Kohno N et al 1996 Usefulness of high-resolution CT for differential diagnosis of multi-focal pulmonary consolidation. Radiat Med 14: 139–146 6. Lui Y M, Taylor J R, Zylak C J 1973 Roentgen–anatomical correlation of the individual human pulmonary acinus. Radiology 109: 1–5 7. Van Allen C M, Lindskog G E 1931 Collateral respiration in the lung: rôle in bronchial obstruction to prevent atelectasis and to restore patency. Surg Gynecol Obstet 53: 16–21 8. Miller W S 1947 Historical sketch. In: Miller W S (ed) The lung, 2nd edn. Springfield: Charles C Thomas, Springfield, pp 162–202 9. Itoh H, Murata K, Konishi J, Nishimura K, Kitaichi M, Izumi T 1993 Diffuse lung disease: pathologic basis for the high-resolution computed tomography findings. J Thorac Imaging 8: 176–188 10. Murata K, Itoh H, Todo G et al 1986 Centrilobular lesions of the lung: demonstration by high-resolution CT and pathologic correlation. Radiology 161: 641–645 11. Webb W R, Müller N L, Naidich D P 2001 High-resolution computed tomography findings of lung disease. In: Webb W R, Müller N L, Naidich D P (eds) High-resolution CT of the lung, 3rd edn. Lippincott Williams & Wilkins, Philadelphia, pp 71–192 12. Austin J H M, Müller N L, Friedman P J et al 1996 Glossary of terms for computed tomography of the lungs: recommendations of the nomenclature committee of the Fleischner Society. Radiology 200: 327–231 13. Gamsu G, Thurlbeck W M, Macklem P T, Fraser R G 1971 Roentgenographic appearance of the human pulmonary acinus. Invest Radiol 6: 171–175 14. Recavarren S, Benton C, Gall E A 1967 The pathology of acute alveolar diseases of the lung. Semin Roentgenol 2: 22–32 15. Itoh H, Tokunaga S, Asamoto H et al 1978 Radiologic–pathologic correlations of small lung nodules with special reference to peribronchiolar nodules. Am J Roentgenol 130: 223–231 16. Collins J, Stern E J 1997 Ground-glass opacity at CT: the ABCs. Am J Roentgenol 169: 355–367 17. Crapo J D 1993 New concepts in the formation of pulmonary oedema. Am Rev Respir Dis 147: 790–792 18. Ketai L H, Godwin J D 1998 A new view of pulmonary oedema and acute respiratory distress syndrome. J Thorac Imaging 13: 147–171

415

416

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

19. Harrison M O, Conte P J, Heitzman E R 1971 Radiological detection of clinically occult cardiac failure following myocardial infarction. Br J Radiol 44: 265–272 20. Staub N C, Nagano H, Pearce M L 1967 Pulmonary edema in dogs, especially the sequence of fluid accumulation in lungs. J Appl Physiol 22: 227–240 21. Youngberg A S 1977 Unilateral diffuse lung opacity. Radiology 123: 277–281 22. Fleischner F G 1967 The butterfly pattern of acute pulmonary oedema. Am J Cardiol 20: 39–46 23. Friedman W F, Braunwald E 1966 Alterations in regional pulmonary blood flow in mitral valve disease studied by radioisotope scanning: a simple nontraumatic technique for estimation of left atrial pressure. Circulation 34: 363–376 24. Baumstark A, Swensson R G, Hessel S J et al 1983 Evaluating the radiographic assessment of pulmonary venous hypertension in chronic heart disease. Am J Roentgenol 141: 877–884 25. West J B, Dollery C T, Heard B E 1964 Increased vascular resistance in the lower zone of the lung caused by perivascular oedema. Lancet 2: 181–183 26. Zwickler M P, Peters T M, Michel R P 1994 Effects of pulmonary fibrosis on the distribution of edema: computed tomographic scanning and morphology. Am J Respir Crit Care Med 149: 1266–1275 27. Pistolesi M, Miniati M, Milne E N C, Giuntini C 1985 The chest roentgenogram in pulmonary edema. Clin Chest Med 6: 315–344 28. Milne E N C, Pistolesi M, Miniati M, Giuntini C 1985 The radiologic distinction of cardiogenic and noncardiogenic edema. Am J Roentgenol 144: 879–894 29. Aberle D R, Wiener-Kronish J P, Webb W R, Matthay M A 1988 Hydrostatic versus increased permeability pulmonary edema: diagnosis based on radiographic criteria in critically ill patients. Radiology 168: 73–79 30. Smith R C, Mann H, Greenspan R H, Pope C F, Sostman H D 1987 Radiographic differentiation between different etiologies of pulmonary edema. Invest Radiol 22: 859–863 31. Wiener-Kronish J P, Matthay M A 1988 Pleural effusions associated with hydrostatic and increased permeability pulmonary edema. Chest 93: 852–858 32. Herold C J, Wetzel R C, Robotham J L, Herold S M, Zerhouni E A 1992 Acute effects of increased intravascular volume and hypoxia on the pulmonary circulation: assessment with high-resolution CT. Radiology 183: 655–662 33. Miller R R 1995 Diffuse pulmonary hemorrhage. In: Thurlbeck W M, Churg A M (eds) Pathology of the lung, 2nd edn. Thieme Medical Publishers, New York, 365–373 34. Collard H R, Schwarz M I 2004 Diffuse alveolar hemorrhage. Clin Chest Med 25: 583–592 35. Travis W D, Colby T V, Koss M N, Rosado-de-Christenson M L, Müller N L, King T E Jr 2002 Idiopathic interstitial pneumonia and other diffuse parenchymal lung diseases. In: Travis W D, Colby T V, Koss M N, Rosado-de-Christenson M L, Müller N L, King T E Jr (eds) Atlas of nontumor pathology: non-neoplastic disorders of the lower respiratory tract. 1 (Fascicle 2) ed. American Registry of Pathology and the Armed Forces Institute of Pathology, Washington, DC, pp 49–231 36. Morgan P G M, Turner-Warwick M 1981 Pulmonary haemosiderosis and pulmonary haemorrhage. Br J Dis Chest 75: 225–242 37. Buschman D L, Ballard R 1993 Progressive massive fibrosis associated with idiopathic pulmonary hemosiderosis. Chest 104: 293–295 38. Ioachimescu O C, Sieber S, Kotch A 2004 Idiopathic pulmonary haemosiderosis revisited. Eur Respir J 24: 162–169 39. Scadding J G 1956 Pulmonary fibrosis and collagen diseases of the lungs: a symposium. I: clinical problems of diffuse pulmonary fibrosis. Br J Radiol 29: 633–641 40. Goodpasture E W 1919 The significance of certain pulmonary lesions in relation to the etiology of influenza. Am J Med Sci 158: 863–870 41. Briggs W A, Johnson J P, Teichman S, Yeager H C, Wilson C B 1979 Antiglomerular basement membrane antibody-mediated glomerulonephritis and Goodpasture’s syndrome. Medicine 58: 348–361

42. Bruwer A J, Kennedy R L J, Edwards J E 1956 Recurrent pulmonary hemorrhage with hemosiderosis: so-called idiopathic pulmonary hemosiderosis. Am J Roentgenol 76: 98–107 43. Cheah F K, Sheppard M N, Hansell D M 1993 Computed tomography of diffuse pulmonary haemorrhage with pathological correlation. Clin Radiol 48: 89–93 44. DeRemee R A, Colby T V 1995 Wegener’s granulomatosis. In: Thurlbeck W M, Churg A M (eds) Pathology of the lung, 2nd edn. Thieme Medical Publishers, New York, pp 401–402 45. Hoffman G S, Kerr G S, Leavitt R Y et al 1992 Wegener granulomatosis: an analysis of 158 patients. Ann Intern Med 116: 488–498 46. Singer J, Suchet I, Horwitz T 1990 Paediatric Wegener’s granulomatosis: two case histories and a review of the literature. Clin Radiol 42: 50–51 47. Cordier J F, Valeyre D, Guillevin L, Loire R, Brechot J-M 1990 Pulmonary Wegener’s granulomatosis: a clinical and imaging study of 77 cases. Chest 97: 906–912 48. Langford C A, Hoffman G S 1999 Rare diseases 3: Wegener’s granulomatosis. Thorax 54: 629–637 49. Carrington C B, Liebow A A 1966 Limited forms of angiitis and granulomatosis of Wegener’s type. Am J Med 41: 497–527 50. Langford C A 2003 Treatment of ANCA-associated vasculitis. N Engl J Med 349: 3–4 51. Kuhlman J E, Hruban R H, Fishman E K 1991 Wegener granulomatosis: CT features of parenchymal lung disease. J Comput Assist Tomogr 15: 948–952 52. Lohrmann C, Uhl M, Kotter E, Burger D, Ghanem N, Langer M 2005 Pulmonary manifestations of Wegener granulomatosis: CT findings in 57 patients and a review of the literature. Eur J Radiol 53: 471–477 53. Grotz W, Mundinger A, Würtemberger G, Peter HH, Schollmeyer P 1994 Radiographic course of pulmonary manifestations in Wegener’s granulomatosis under immunosuppressive therapy. Chest 105: 509–513 54. Attali P, Begum R, Ben Rhomdhane H, Valeyre D, Guillevin L, Brauner M W 1998 Pulmonary Wegener’s granulomatosis: changes at follow-up CT. Eur Radiol 8: 1009–1113 55. Wadsworth D T, Siegel M J, Day D L 1994 Wegener’s granulomatosis in children: chest radiographic manifestations. Am J Roentgenol 163: 901–904 56. Davison A G, Heard B E, McAllister W A C, Turner-Warwick M E H 1983 Cryptogenic organizing pneumonitis. Q J Med 52: 382–394 57. Cordier J F 2000 Organising pneumonia. Thorax 55: 318–328 58. Epler G R, Colby T V, McLoud T C, Carrington C B, Gaensler E A 1985 Bronchiolitis obliterans organizing pneumonia. N Engl J Med 312: 152–158 59. Travis W D, King T E Jr, and the Multidisciplinary Core Panel 2002 American Thoracic Society/European Respiratory Society international multidiscplinary consensus classification of idiopathic interstitial pneumonias. Am J Respir Crit Care Med 165: 277–304 60. Cordier J F 2004 Cryptogenic organizing pneumonia. Clin Chest Med 25: 727–738 61. Flowers J R, Clunie G, Burke M, Constant O 1992 Bronchiolitis obliterans organizing pneumonia: the clinical and radiological features of seven cases and a review of the literature. Clin Radiol 45: 371–377 62. Lee K S, Kullnig P, Hartman T E, Müller N L 1994 Cryptogenic organizing pneumonia: CT findings in 43 patients. Am J Roentgenol 162: 543–546 63. Kim S J, Lee K S, Ryu Y H et al 2003 Reversed halo sign on highresolution CT of cryptogenic organizing pneumonia: diagnostic implications. Am J Roentgenol 180: 1251–1254 64. Akira M, Yamamoto S, Sakatani M 1998 Bronchiolitis obliterans organizing pneumonia manifestating as multiple large nodules or masses. Am J Roentgenol 170: 291–295 65. Murphy J M, Schnyder P, Verschakelen J, Leuenberger P, Flower C D R 1999 Linear opacities on HRCT in bronchiolitis obliterans organising pneumonia. Eur Radiol 9: 1813–1817 66. Ujita M, Renzoni E A, Veeraraghavan S, Wells A U, Hansell D M 2004 Organizing pneumonia: perilobular pattern at thin-section CT. Radiology 232: 757–761 67. Crofton J W, Livingstone J L, Oswald N C, Roberts A T M 1952 Pulmonary eosinophilia. Thorax 7: 1–35

CHAPTER 21

68. Fraser R S, Müller N L, Colman N, Paré P D 1999 Eosinophilic lung disease. In: Fraser R S, Müller N L, Colman N, Paré P D (eds) Diagnosis of diseases of the chest, 4th edn. W B Saunders, Philadelphia, pp 1743–1756 69. Löffler W 1932 Zur differential-diagnose der lungen infiltreierunger: III uber fluchtige succendan—infiltrate (mit eosinophilia). Beitr Klin Tuberk 79: 368–392 70. Allen J N, Pacht E R, Gadek J E, Davis W B 1989 Acute eosinophilic pneumonia as a reversible cause of noninfectious respiratory failure. N Engl J Med 321: 569–574 71. Hayakawa H, Sato A, Toyoshima M, Imokawa S, Taniguchi M 1994 A clinical study of idiopathic eosinophilic pneumonia. Chest 105: 1462–1466

• AIRSPACE DISEASES

72. Cheon J E, Lee K S, Jung G S, Chung M H, Cho Y D 1996 Acute eosinophilic pneumonia: radiographic and CT findings in six patients. Am J Roentgenol 167: 1195–1199 73. Fraser R S, Müller N L, Colman N, Paré P D 1999 Metabolic pulmonary disease. In: Fraser R S, Müller N L, Colman N, Paré P D (eds) Diagnosis of diseases of the chest, 4th edn. W B Saunders, Philadelphia, pp 2699–2735 74. Prakash U B S, Barham S S, Carpenter H A, Dines D E, Marsh H M 1987 Pulmonary alveolar phospholipoproteinosis: experience with 34 cases and a review. Mayo Clin Proc 62: 499–518 75. Ramirez R J 1967 Pulmonary alveolar proteinosis: treatment by massive bronchopulmonary lavage. Arch Int Med 119: 147–156

417

CHAPTER

Cardiac Anatomy and Imaging Techniques

22

George G. Hartnell and Julia Gates

• Normal cardiac anatomy • Cross-sectional assessment of cardiac enlargement • Cardiac imaging techniques

Cardiac imaging is challenging for many reasons. No other target organ is in constant motion; essential to life; or as commonly affected by potentially lethal disease. For these reasons cardiac imaging is very common, often costly, and may put patients at risk. There are major financial implications for health care professionals, hospitals, vendors and payers. There are disputes over appropriate utilization of, and responsibility for performing and receiving payment for, studies1. Imaging capabilities change rapidly and the relative value of each technique requires constant re-evaluation2. For this reason it is important to understand the anatomy of the heart, the current capabilities of each available imaging technique and the best methods for imaging each clinical problem. The use of chest radiography in cardiac diagnosis has been largely replaced by echocardiography. Many regard echocardiography as an integral part of the cardiac examination, providing essential information of cardiac function and anatomy. Angiography is mainly used to evaluate coronary artery anatomy, while nuclear imaging is used mainly in the evaluation of ischaemic heart disease (see Ch. 25). Interest in cardiac imaging by magnetic resonance imaging (MRI) and computed tomography (CT) has waxed and waned. There have been periods of enthusiasm associated with the introduction of cine-magnetic resonance angiography (MRA), coronary MRA and electron beam or ultrafast CT (EBCT or UFCT). These developments, although academically significant, had relatively little impact on routine cardiac care. More recently both MRI and CT have developed to the stage where reliable, economic, and comprehensive assessment of cardiac function, perfusion, viability and coronary artery anatomy is possible as part of routine care3. The heart may be studied by the conventional projection imaging techniques of chest radiography, angiography and radionuclide imaging, or by the two-dimensional (2D) imaging techniques of echocardiography, cardiac CT and cardiac MRI and MRA (CMR). When used appropriately in suit-

able subjects, echocardiography, multidetector CT (MDCT), UFCT and CMR provide similar information concerning anatomy and function, and can often be used interchangeably, although echocardiography is used most widely for cardiac diagnosis. In this chapter, MDCT, CMR and angiographic anatomy are used to illustrate the orientation of the heart and its internal structure. Although much cardiac anatomy is conventionally shown by angiography, this is being replaced in many cases by echocardiography MDCT and CMR.

NORMAL CARDIAC ANATOMY Orientation of the cardiac chambers The relationship of the cardiac chambers to each other, the positions and inclinations of the cardiac septae and valves (Table 22.1), and the great vessel relationships are shown by casts of chambers of the heart (Figs 22.1, 22.2). The right atrium forms the right border of the heart. The superior vena cava enters the upper posterior aspect of the right atrium. The atrial septum forms the posteromedial wall of the right atrium.

Table 22.1 ORIENTATIONS OF MAJOR CARDIAC STRUCTURES WHEN ASSESSED BY MRI4 Mean angle

SD

Optimum projection

Aortic arch

−28

8

LAO 62 (P)

Interventricular septum

+56

12

LAO 56 (E)

Interatrial septum

+61

10

LAO 61 (E)

Mitral valve ring

−38

10

RAO 38 (E)

Tricuspid valve ring

−42

11

RAO 43 (E)

Origin right coronary

−17

13

LAO 73 (P)

Origin left coronary

−66

15

LAO 14 (P)

Long axis left ventricle

+48

12

RAO 42 (P)

Long axis right ventricle

+65

9

RAO 25 (P)

Right pulmonary artery

−64

8

RAO 26 (P)

Left pulmonary artery

+37

8

LAO 53 (P)

LAO = left anterior oblique, RAO = right anterior oblique. E = structure best viewed from end-on perspective; P = structure best viewed from perpendicular perspective.

420

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

The aortic root, lying in the middle of the heart, is anteromedial to the right atrium. To the left and anterior to the right atrium lies the right ventricle, separated from the atrium by the tricuspid valve.The right ventricle forms the bulk of the front of the heart. The pulmonary valve, at the top of the right ventricular infundibulum, lies cephalad, anterior and to the left of the aortic root. The curved ventricular septum bulges into the right

ventricle, narrowing it from behind and the left.The curvature of the septum varies from anterior to posterior and from above downwards. The posterior aspect of the heart is formed by the left atrium: the left atrial appendage projects anteriorly and to the left from its upper left border. The pulmonary veins enter the posterior left atrium. The left atrium is flattened antero-posteriorly. The left atrium and ventricle are separated by the mitral valve,

Figure 22.1 Anatomy of the right heart cavities. Cast of the right heart cavities compared with maximum intensity projection (MIP) and volume rendered (VRT) CTA. (A) Frontal and (B) left lateral. The black area on the cast represents the intraventricular component of the membranous septum (m) under the septal leaflet of the tricuspid valve. The white area (s) represents that component of the membranous septum that lies above the tricuspid valve and separates the left ventricle from the right atrium. The entry of the main draining vein of the heart, the coronary sinus (cs), into the right atrium between the inferior vena cava and the tricuspid valve is seen. (C,D) Coronal 3D reconstruction of the right heart chambers using MDCTA. Right anterior oblique views of the right heart shown by (C) CTA (maximum intensity projection [MIP]) and (D) volume rendered (VRT) shows the right coronary artery (RCA, arrows in C) running in the low attenuation fat of the anterior atrioventricular groove. Continued

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.1 Cont’d (E,F) Sagittal 3D reconstruction of the right heart chambers using MDCTA by (E) MIP and (F) VRT showing the position of the pulmonary valve (white arrows). Mitral valve leaflets indicated by black arrows on (E). Ao = aortic root, CSV = crista supraventricularis, DA = descending aorta, fo = foramen ovale, Inf = infundibulum separating the right ventricular inflow and outflow, IVC = inferior vena cava, MPA = main pulmonary artery, pb = parietal band, PV = pulmonary valve (between white arrows), RA = right atrium, RAA = right atrial appendage, RV = right ventricle, sb = septal band, SVC = superior vena cava, TV = tricuspid valve.

Figure 22.2 Anatomy of the left heart cavities. Cast of the left heart chambers. (A) Frontal. The left atrium (LA) can be seen at the back with its appendage (LAA) protruding to the left. The carrot-shaped left ventricle points anteriorly, inferiorly and to the left, and hence is foreshortened. Three components of the ventricular septum can be seen. The membranous septum (ms) between the right and noncoronary sinuses of Valsalva provides the right attachment to the anterior leaflet of the mitral valve. The smooth muscular septum (sms) curves round the upper part of the left ventricle, almost forming part of the left border of the left ventricle. Below lies the trabeculated muscular septum (ts). The right coronary sinus, seen en face, is indicated by the origin of the right coronary artery (rca). (B) Left lateral. The left ventricle is foreshortened as it points to the left as well as downwards and forwards. The lack of a muscular infundibulum in the left ventricle allows the upper left ventricle to reach up between the right coronary and left coronary sinuses of the aortic root. The origin of the anterior papillary muscle (apm) of the mitral valve from the left ventricular free wall may be identified (in this cast, it lies superimposed on the posterior papillary muscle). The right coronary sinus of Valsalva is indicated by the origin of the right coronary artery. (C–G) 3D reconstructions of the left heart chambers using MDCTA. (C) Coronal view of the left heart (maximum intensity projection [MIP]) shows the position (white line) for setting up a fourchamber view. (D) The four-chamber view shows the right coronary artery (RCA) running in the low attenuation fat of the anterior atrioventricular groove and the circumflex coronary artery (CX) running in the low attenuation fat of the posterior atrioventricular groove. Note the prominent crista terminalis (CT) in the wall of the right atrium, a normal structure commonly mistaken for a mass. The white line indicates the position for setting up a two-chamber view of the left heart. Continued

421

422

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.2 Cont’d (E) Oblique coronal MIP reconstruction of the left heart chambers (E) showing the relationship of the LAD running in the anterior interventricular groove and the position of the bifurcation of the LMCA. (F) Volume rendered (VRT) image (left anterior cranial) showing the relationship of the ventricles, the origins of the ascending aorta and pulmonary artery and positions of the left anterior descending coronary artery (LAD) and right coronay artery (RCA). (G) VRT image (left lateral) showing the relationship of the left ventricle (LV) to the LAD, coronary sinus (CS) and CX. AA = ascending aorta, AV = aortic valve, CT = crista terminalis, PA = main pulmonary artery, RA = right atrium, RV = right ventricle.

which lies in the right anterior oblique plane. The left ventricle, which forms the bulk of the left heart border, lies anterior to the left atrium. The atrial and ventricular septae lie in approximately the same left anterior oblique plane, forming the septal plane of the heart, and separate the right heart chambers anteriorly and on the right from the left heart chambers posteriorly and on the left (Fig. 22.2D,F). The plane of the ventricular septum is marked on the surface of the heart by the interventricular groove encircling the ventricular mass. The aortic valve lies in the centre of the cardiac mass with the left main (LMCA) and right (RCA) coronary arteries arising from the aortic root about 1 cm cephalad to the valve ring (Fig. 22.3).The left anterior descending (LAD) coronary artery (a branch of the left coronary artery) runs in the anterior interventricular groove to supply the anterior aspect of the ventricular septum (Figs 22.2F, 22.4). Septal branches pass perpendicularly from the LAD into the septal myocardium (Fig. 22.4B). Diagonal branches pass from the LAD over the posterior surface of the heart (Fig.

22.4C). In about 10% a large diagonal branch arises from the bifurcation of the LMCA to form the ramus intermedius artery (Fig. 22.5C). The posterior descending coronary artery, which may be a branch of the right or left coronary arteries, lies in the postero-inferior interventricular groove and supplies the posterior base of the ventricular septum.The relative contributions of these arteries to septal blood supply can vary considerably. The atrioventricular plane is marked on the surface of the heart by the atrioventricular groove.The right coronary artery passes anteriorly in the right (anterior) atrioventricular groove (Figs 22.1C,D, 22.6); the circumflex branch of the left coronary artery lies in the left (posterior) atrioventricular groove. The atrioventricular and posterior interventricular grooves intersect on the postero-inferior surface of the heart at the crus (crux) of the heart. The coronary artery that reaches the crux and supplies the posterior interventricular branch is termed the dominant coronary artery. This is usually the right coronary artery (85%) but there may be circumflex dominance (8%) or mixed dominance (7%).

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.3 Evaluation of aortic valve and coronary artery origins using ECG-gated MDCTA. (A) Coronal MDCTA image shows the positions for imaging the coronary artery origins and the aortic valve. (B) Short axis view through the aortic valve clearly shows the position of the three raphe of the aortic valve (arrows). There is a small and clinically insignificant amount of calcification on the noncoronary cusp. (C) Image just superior to (B) shows the origins of the main coronary arteries with a calcified moderate proximal stenosis of the left main coronary artery (LMCA). Note the ability to image even small vessels such as the sinus node artery and proximal diagonal artery. AA = Ascending aorta, Ao = aortic root, D = diagonal coronary artery, LA = left atrium, LAD = left anterior descending coronary artery, LMCA = left main coronary artery, LV = left ventricle, RA = right atrium, RCA = right coronary artery, RV = right ventricle, RVOT = right ventricular outflow tract, SNA = sinus node artery.

Individual heart chambers Right atrium The right atrium (Fig. 22.1) has a smooth globular body and a triangular, trabeculated appendage pointing upwards, forwards, and to the left. A muscular ridge, the crista terminalis, separates the body of the atrium from the appendage (Fig. 22.2D). The broad-based, squat right atrial appendage characterizes the morphological right atrium, and in complex congenital heart disease distinguishes it from the left atrium (Figs 22.1C,D, 22.2C). The superior and inferior venae cavae enter the posterior wall of the right atrium some way from its centre (but not at the upper and lower ends). The fossa ovalis, surrounded by the slight elevation of its limbus, lies in the middle of the interatrial septum, which forms the left posterior wall of the right atrium. The tricuspid valve opens to the left and anteriorly from the body of the right atrium. The coronary sinus (Figs 22.2G, 22.4C), the major draining vein of the heart, enters the posterior wall of the right atrium between the tricuspid valve and the inferior vena cava. On a frontal image, a catheter or pacing wire in the coronary sinus may appear to lie in the right ventricular outflow tract, but a lateral view will show it to be posterior. The right atrium forms the right margin of the heart (Fig. 22.1). Three features characterize right atrial internal morphology: 1 the limbus of the fossa ovalis on the septal aspect 2 the crista terminalis 3 a squat, broad-based appendage Right atrial size is estimated by echocardiography, CT, or CMR in the same way as that of the left atrium, particularly using an apical four-chamber view or its equivalent. Causes of right atrial enlargement are listed in Table 22.2.

Right ventricle The right ventricle has a complex shape (Fig. 22.1). From the front it resembles a triangle (although with four distinct edges) based on the diaphragm. Superiorly a muscular conus or infundibulum separates the pulmonary valve from the tricuspid valve (Figs 22.1D, 22.7B). On the left and inferiorly is a coarsely trabeculated apex. The lower end of the septal aspect of the conus has a discrete muscular elevation, the crista supraventricularis, extending to the right and forwards over the tricuspid valve to the right ventricular free wall to form the parietal band. The normal right ventricle makes virtually no contribution to the cardiac silhouette in the frontal projection (Fig. 22.1D). In the lateral view the right ventricle forms the anterior aspect of the heart (Fig. 22.1E,F), with the lower half normally in contact with the sternum. Three features characterize right ventricular morphology: 1 a muscular conus or infundibulum separating the entry atrioventricular valve from the exit semilunar valve by a ring of muscle 2 a tricuspid atrioventricular valve which is in part connected to the septum of the chamber, either by direct chordal attachment or through a small papillary muscle of the conus 3 coarse trabeculation of the septal aspect of the ventricle (Fig. 22.7) Restricted acoustic access often makes it difficult to image the right ventricle by echocardiography, and it is more completely evaluated by CT or CMR. Causes of right ventricular enlargement are listed in Table 22.3.

Left atrium The left atrium (Figs 22.2, 22.4) usually receives the four pulmonary veins at the ‘corners’ of its posterior surface (Fig. 22.8) and empties through the mitral valve in its left lower

423

424

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.4 Evaluation of left coronary artery anatomy by ECG-gated MDCTA. (A) Axial MDCTA source image showing positions for imaging the left anterior descending coronary artery. (B) Oblique thin maximum intensity projection (MIP) reconstruction along the line of the left anterior descending coronary artery (LAD) with calcified plaque at several levels as well as the origin of the circumflex coronary artery and several small septal arteries (S). (C) Oblique thin MIP reconstruction posterior to (B) showing the distal LAD and the origin of a large diagonal branch (arrows). Note the visibility of the coronary sinus (CS) and its branches, which should not be confused with arteries. (D) Curved thin MIP reconstruction along line shown (inset) to demonstrate length of the LAD. Arrows indicate plaque causing variable degrees of stenosis in both the right coronary artery and the LAD. Ao = aortic root, LA = left atrium, LAA = left atrial appendage, LV = left ventricle, MCV = middle coronary vein, MV = mitral valve, PA = pulmonary artery, PM = papillary muscle, RAA = right atrial appendage, S = septal branches from the LAD.

anterior aspect. The long finger-like forward-projecting left atrial appendage is its only characteristic morphological feature (Figs 22.4C, 22.5C). The left atrium makes no definite contribution to the normal cardiac silhouette in the frontal view as the normal left atrial appendage is buried in the epicardial fat and does not produce a discrete shadow (Fig. 22.9). In the lateral view (Fig. 22.2G), the left atrium contributes to the upper posterior border of the heart, but normally the left atrium is not in contact with the air-containing lung and there is no properly defined outline. The features that characterize left atrial morphology are an absence of those suggesting right atrial morphology, and more positively the long finger-like atrial appendage.

The left atrium is posterior to the aortic root (Fig. 22.3). Left atrial enlargement is usually measured using echocardiography but parallel images, from CMR or MDCT, can also be used to measure atrial volumes. Causes of selective left atrial enlargement are listed in Table 22.4.

Left ventricle The left ventricle (Figs 22.2, 22.4) is a cone-shaped structure with a long axis pointing inferiorly, anteriorly and to the left. Its base is the fibrous skeleton of the mitral and aortic valves and its apex usually forms the apex of the cardiac shadow as seen on the frontal chest radiograph. The left ventricle has no conus or infundibulum: the mitral and

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.5 Vascular relationships to the left heart using ECG-gated MDCTA. (A) Axial MDCTA source image showing the positions (from anterior to posterior) of coronal reconstructions illustrated in B–D. This image shows the origin of the left main coronary artery from the aortic root. (B) Thin maximum intensity projection (MIP) coronal reconstruction through the centre of the aortic valve shows part of the left anterior descending coronary artery (LAD) passing obliquely through the image. (C) A more posterior image shows a complex high attenuation (calcified) and low attenuation (fatty) plaque at the origin of the left main coronary artery (left main coronary artery [LMCA]). Note the trifurcation of the LMCA due to the presence of a ramus intermedius artery. The same image shows the origin of the circumflex coronary artery and the left atrial appendage superior to the left ventricular myocardium. The characteristic ridged appearance of the left atrial appendage must be differentiated from thrombus. High attenuation at the base of the posterior mitral leaflet may be an early sign of mitral annulus calcification. (D) The most posterior image shows the upper part of the coronary sinus running in the posterior atrioventricular groove and inflow of the right upper pulmonary vein. AA = ascending aorta, Ao = aortic root, AV = aortic valve leaflets, CS = coronary sinus, CX = circumflex coronary artery, DA = descending aorta, LA = left atrium, LAA = left atrial appendage, LPA = left pulmonary artery, LV = left ventricle, MV = mitral valve, PA = pulmonary artery, RA = right atrium, RAA = right atrial appendage, RI = ramus intermedius coronary artery, RPA = right pulmonary artery, RUPV = right upper pulmonary vein.

425

426

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.6 Evaluation of right coronary artery anatomy by ECG-gated MDCTA. (A) Axial MDCTA source image showing positions for imaging the proximal RCA. Usually the same orientation will image the proximal circumflex coronary artery (CX). (B) Oblique reconstruction showing the origin of the right coronary artery (RCA) in the anterior atrioventricular groove. (C) Adjacent more posterior image showing a long length of RCA running to the inferior margin of the heart. In all these images it is important to avoid confusing the coronary arteries and cardiac veins, structures which are not usually seen on coronary arteriography. (D) Curved thin maximum intensity projection (MIP) reconstruction along line shown (inset) to demonstrate length of the RCA. Areas of coronary stenosis are indicated by black and white arrows. Ao = aortic root, AV = aortic valve leaflets, CS = coronary sinus, LA = left atrium, LAA = left atrial appendage, LV = left ventricle, MCV = middle coronary vein, PMV = posterior mitral valve leaflet, RA = right atrium, RCA = right coronary artery, RV = right ventricle, SNA = sinus node artery.

CHAPTER 22

Table 22.2

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

CAUSES OF A LARGE RIGHT ATRIUM

Volume overload Tricuspid regurgitation. Most commonly due to congestive cardiac failure with right ventricular dilatation* Atrial septal defect (due to increased flow from left atrium) Atrioventricular canal (due to shunt into right atrium and atrioventricular valve regurgitation) Pulmonary hypertension (usually with a degree of tricuspid regurgitation) Anomalous pulmonary venous drainage to right atrium Sinus of Valsalva fistula Pressure overload Right ventricular failure Tricuspid stenosis (with mitral stenosis when rheumatic aetiology; con genital tricuspid stenosis is very rare) Restrictive cardiomyopathy (especially in diabetes mellitus) Right atrial myxoma or other tumours causing tricuspid valve obstruction (very rare) Acquired tricuspid stenosis (carcinoid syndrome; more commonly causes tricuspid regurgitation) *Tricuspid regurgitation is also seen in Ebstein's anomaly, arrhythmogenic right ventricular dysplasia, carcinoid syndrome, endomyocardial fibrosis and fibro-elastosis.

aortic valves are in fibrous continuity, the anterior leaflet of the mitral valve arising from the noncoronary and part of the left coronary sinuses of Valsalva of the aortic valve. The mitral valve is bicuspid.The inflow valves are an integral part of their respective ventricles. The ventricular septum lies obliquely from right posterior to left anterior. It is a curved structure arching over the ventricle so that the membranous part of the ventricular septum faces to the right under the noncoronary and right coronary sinuses of Valsalva. The smooth muscular septum derived from the conus cushions of the cardiac tube faces to the left under the right and left coronary sinuses of Valsalva. The distal ventricular septum is finely trabeculated. The feature that best characterizes left ventricular morphology is fibrous continuity of the entry atrioventricular valve and the exit semilunar valve, due to the lack of any conus or infundibulum. The conical shape and relatively fine trabeculation (absent on the upper septum) are not helpful. In diastole, the mitral valve is usually open wide and the aortic valve is closed. Contraction should be concentric, and filling defects about the middle of the anterior and inferior walls are usually due to the papillary muscles (Fig. 22.4C). Left ventricular volume can reliably be estimated by echocardiography, which is the most frequently used method for assessing left ventricular dimensions and function. A number of methods have been described, using combinations of apical four-chamber, apical two-chamber, or multiple short axis views. Left ventricular myocardial thickness can readily be measured by echocardiography, MDCT, or CMR. The thickness of the posterior wall is the distance from the anterior border of the endocardium to that of the epicardium, excluding the pericardial space. When measuring septal thickness, it is essential to exclude the trabeculum septomarginalis, which runs alongside the interventricular septum within the right ventricular cavity. The distribution of

Figure 22.7 Axial MRI in mitral stenosis. (A) Axial ECG-gated spin-echo image of patient with mitral stenosis leading to right heart failure. Slow flow in the left atrium has produced high signal in that chamber. (B) Oblique coronal breath-hold cine-MRA image of the heart. There is dilatation of the right atrium (RA) and right ventricle (RV). There is mild tricuspid regurgitation (jet of flow dephasing = open arrow). aa = Ascending aorta, la = left atrium, sb = septal band, t = tricuspid valve, inf = infundibulum of RV. The pulmonary valve is shown between the black arrows.

Table 22.3

CAUSES OF A LARGE RIGHT VENTRICLE

Left heart failure (most common cause, particularly associated with mitral valve disease) Pulmonary artery hypertension due to lung disease (cor pulmonale) Chronic thromboembolic disease Left-to-right shunt Vasculitis Idiopathic pulmonary hypertension Tricuspid regurgitation Pulmonary regurgitation Pulmonary arteriovenous malformations Severe pulmonary stenosis

427

428

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.8 Left atrial and pulmonary vein anatomy. (A) High signal or attenuation allows better 3D image reconstruction with a wide range of reconstruction techniques. Also, the larger the structure being imaged the more accurate the reconstruction. Contrast-enhanced MRA for pulmonary venography provides very high signal compared with background. (B) Using thin-slice maximum intensity projection (MIP) reconstruction the inflow of the pulmonary veins (RPVS, LPVS) into the left atrium is accurately represented. (C) When this type of high contrast image with a uniform signal level surface shaded (SSD) reconstruction can be used to guide ablation procedures. This posterior view shows late confluence of the right lower lobe apical pulmonary vein (arrow) within 1 cm of the pulmonary vein ostium (over incorporation). AA = Aortic arch, AO = upper descending aorta, DA = lower descending aorta, LA = left atrium, LPV = left pulmonary vein, RPV = right pulmonary vein.

left ventricular hypertrophy is well demonstrated by echocardiography and is particularly important in patients with hypertropic cardiomyopathy. Hypertrophy is often asymmetrical with severe secondary hypertrophy, and localized basal septal hypertrophy is a common finding particularly in the elderly. Causes of selective left ventricular enlargement are listed in Table 22.5.

CROSS-SECTIONAL ASSESSMENT OF CARDIAC ENLARGEMENT Echocardiography is very useful in establishing the cause of cardiomegaly and is the usual method for determining inter-

nal cardiac dimensions. Both MDCT and CMR can serve the same functions, with CMR being probably the more accurate technique. Normal dimensions should be corrected for patient size. It is not uncommon for several abnormalities to be present in one patient. Thus aortic (stenosis or regurgitation) or mitral (regurgitation) disease cause left ventricular dilatation and increased wall thickness. In these patients, left ventricular enddiastolic pressure rises, so that left atrial size is increased while associated fluid retention may produce a pericardial effusion. Chronic left-sided heart disease, eventually causing right heart failure, may lead to right atrial and ventricular dilatation with secondary valvar regurgitation (Fig. 22.7).

Table 22.4

CAUSES OF A LARGE LEFT ATRIUM

Due to volume overload Mitral regurgitation (often with left ventricular failure) Ventricular septal defect Patent ductus arteriosus Atrial septal defect with shunt reversal (i.e. pulmonary hypertension) Atrial septal defect with tricuspid atresia (obligatory shunt reversal) Aortopulmonary window Due to pressure overload Mitral stenosis Noncompliant left ventricle: hypertension, hypertrophic cardiomyopathy, aortic stenosis

Figure 22.9 Coronal MRA. Oblique breath-hold cine-MRA in a patient with mild aortic regurgitation (black arrow) shows the left atrial appendage (LAA) embedded in epicardial fat. There is mild dilatation of the ascending aorta (aa) as a result of the aortic regurgitation. Between curved arrows = aortic valve. lv = Left ventricle, pa = pulmonary artery, RA = right atrium.

Left ventricular failure (often with secondary mitral regurgitation) Left atrial myxoma Other causes (both rare) Atrial fibrillation Isolated/idiopathic

CHAPTER 22

Table 22.5

CAUSES OF A LARGE LEFT VENTRICLE

Any cause of myocardial damage Myocardial ischaemia (most common cause, with or without myocardial infarction) Cardiomyopathy (including ischaemic and hypertensive disease) Myocarditis Chronic volume overload Mitral regurgitation Ventricular septal defect (congenital or acquired) Patent ductus arteriosus Aortopulmonary window Atrioventricular canal High cardiac output Chronic anaemia Thyrotoxicosis Large arteriovenous fistula Extensive Paget's disease of bone* Chronic pressure overload (ventricular dilatation is a late feature which indicates a worsening of prognosis in aortic stenosis, systemic hypertension, and coarctation of the aorta) *

Paget's disease is frequently listed as a cause of left ventricle dilatation resulting from a high-output state and is said to be due to shunting through involved bones. In reality, by far the most common cause of an enlarged heart in patients with Paget's disease is unrelated hypertensive or ischaemic heart disease.

Ventricular volumes can be rapidly estimated, with reasonable accuracy, using 2D images from echocardiography, CT and CMR, and applying one of several formulae (e.g. Simpson’s rule, see below), which make assumptions about ventricular geometry (the normal left ventricle is assumed to be a prolate ellipse). These methods become less accurate in the dilated ventricle.

Biventricular enlargement In most adults with acquired heart disease, the disease process, although beginning on the left side of the heart, eventually involves the pulmonary circulation and the right heart chambers, leading to gross cardiac enlargement (Table 22.6). When this occurs, all the cardiac chambers are likely to be involved, and it is not possible to determine the contribution of each chamber with plain radiography.The heart takes on a globular shape when all chambers are enlarged.

Table 22.6 CAUSES OF GROSS CARDIAC ENLARGEMENT Multiple valve disease (usually initially due to aortic or mitral valve disease, particularly regurgitation, which progresses to left and then right heart failure) Pericardial effusions (no single chamber enlargement identifiable) Dilated cardiomyopathy Ebstein's anomaly (frequently mimics pericardial effusion) Arrhythmogenic right ventricular dysplasia Heterotropic cardiac transplantation

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

CARDIAC IMAGING TECHNIQUES Planar imaging orientation and cross-sectional imaging planes There are many ways to obtain imaging information about the heart. Each technique has advantages, disadvantages, costs and risks. The choice depends on the nature of the data required, availability of imaging equipment, interpretative and operator skills; all work well and are accurate enough in most routine cases. A detailed discussion of the capabilities of each technique is less important than understanding the anatomy and pathology, and then using the appropriate technique well. For example, left ventricular systolic function can be assessed with similar accuracy by echocardiography, CT angiography (CTA), MRA, nuclear imaging and angiocardiography. The decision on which to use depends on the need for reproducibility (favouring MRA or nuclear imaging); patient mobility (echocardiography is best for unstable patients where imaging must go to the patient); need for other imaging (selective coronary angiography with angiocardiography); viability (MRI or nuclear imaging); and assessment of valve function (echocardiography). Nuclear imaging (planar) and angiography are planar techniques, which require image acquisition in the correct orientation and depend on the configuration and orientation of the heart. For ventriculography (nuclear or angiographic) this will usually be in a right anterior oblique (RAO; Fig 22.10) orientation (and occasionally left anterior oblique [LAO]) profiling the long axis of the left ventricle. Sectional (2D) imaging by echocardiography or CMR requires acquisition to be positioned to best identify the abnormality using a variety of imaging planes (short axis, two- and four-chamber long axis approaches).

Quantitative assessment of cardiac performance Assessment of cardiac performance is important since it reflects the heart’s ability to adapt to a pathological state. Poor ventricular function causes symptoms and also predicts those patients with a reduced life expectancy, whether due to ischaemic heart disease, valve disease, or other pathologies. Regional abnormalities may identify areas of viable but ischaemic or nonviable myocardium or other focal disease (such as right ventricular dysplasia). Systolic ventricular function can be assessed by echocardiography, CTA (to some extent), cine-MRA, nuclear imaging and angiocardiography. Quantitative analysis can be applied with varying degrees of accuracy and ease to all these techniques. The accuracy of assessment depends on temporal resolution (the faster the frame rate the better), spatial resolution and ability to define the entire endocardial surface accurately (best for echocardiography and MRA), freedom from artefact (angiocardiography may be affected by arrhythmia during contrast medium injection), and ability to apply computer analysis techniques. Quantitative estimation of pump performance is usually limited to the left ventricle and to the events occurring during the ejection of blood. The ejection fraction is the most

429

430

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.10 (A,B) Normal left ventriculogram, right anterior oblique (RAO). (A) Diastole. The aortic valve (arrows) is closed. The mitral valve is not seen in profile, due to inadequate obliquity, but its site could be identified by nonopaque blood passing through it (asterisk). (B) Systole. There has been good emptying with concentric contraction so that anterior wall, apex and inferior wall of the left ventricle (LV) have all moved equally. The aortic valve is open and the mitral valve closed. Filling defects in the contrastfilled left ventricle indicate the papillary muscles (white arrowhead = anterior, black arrowhead = posterior, open black arrow = contribution of the septum to the left ventricular outline). (C,D) Normal left ventricle on breath-hold cine-MRA (RAO equivalent). (C) Enddiastole. The closed mitral valve is well seen (between white arrows), as is the posterior papillary muscle (black arrow). (D) End-systole. There has been good emptying with concentric contraction of the LV so that anterior wall, apex and inferior wall have all moved equally. The mitral valve (between white arrows) is again closed. Signal defects in the left ventricle indicate thickening of the papillary muscle (black arrow).

common index of pump performance and is calculated from the expression: EF =

(LVEDV–LVESV) LVEDV

where LVEDV is the left ventricular end-diastolic volume and LVESV the left ventricular end-systolic volume calculated from 2D images. The most accurate measurements are obtained using multiple contiguous short axis views and calculating volumes using Simpson's rule (which can also be used with CT and CMR). Simpson's rule is a method for volume or mass calculation based on dividing the ventricular volume into a series of slices equidistant along the long axis of the ventricle. It makes no assumption about the shape of the ventricle. Modifications, based on the assumption that the left ventricle is a prolate ellipse, can be applied to single 2D long axis sections.

Echocardiography General principles Ultrasound examination of the heart, or echocardiography, offers a wealth of anatomical and physiological information without ionizing radiation, patient discomfort, or significant risk. It is therefore well suited to the initial assessment of cardiac disease and to serial studies. Echocardiography is so widely available and provides such useful information on nearly all aspects of cardiac disease that many regard it as an integral part of the clinical examination of the heart. The experience of the examiner or the centre in which it is used will determine its relationship to other techniques of diagnosis and management. For example, one centre may consider echocardiography information sufficient to proceed with surgical therapy while another centre may demand confirmatory data be obtained by cardiac catheterization, CMR, or other imaging techniques. The evolving capabilities of Doppler, power Doppler and 3D echocardiography continue to extend its uses2.

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Two-dimensional echocardiography This is the imaging tool used most frequently to evaluate the anatomy and function of the heart. The ultrasound beam is directed mechanically or more usually electronically through a plane intersecting the chosen cardiac or great vessel structures. This allows imaging of a large segment of the heart and great vessels from a small, often limited, acoustic window on the chest wall. Commercially available ultrasound systems may generate real-time, 2D images at frame rates up to or greater than 100 Hz depending on the total sector angle and distance of the structures from the transducer. When a transthoracic approach is inadequate, transoesophageal echocardiography (TOE) often provides the best images of the heart. In addition, the portability of echocardiography often means that it is the most appropriate method for very sick patients (Fig. 22.11). Through computer reconstruction of 2D images, 3D echocardiograms have become possible. Similar to other forms of tomographic imaging with 3D reconstruction (e.g. CMR, CT), the approach involves obtaining multiple 2D images to produce a 3D image. New, faster techniques are making realtime viewing of 3D images possible5.With improving technology, speed of acquisition and reconstruction, these techniques continue to find increasing clinical use. Current uses have been to determine ventricular size, mass and function, qualitative and quantitative assessment of anatomy, and the placement of medical devices.

Doppler echocardiography The display of blood or tissue velocity can be used qualitatively as an intracardiac stethoscope to validate the source of heart sounds and murmurs. Measurement of blood velocity can lead to quantitative estimation of blood flow and pressure gradients. Assessment of tissue velocity can yield information on the motion of cardiac muscle.

Figure 22.11 Transoesophageal echocardiography (TOE) of aortic dissection. TOE view of the aortic root through the left atrium showing the dissection flap (arrows) in acute Type A aortic dissection. This provides the highest spatial and temporal resolution of any cross-sectional imaging technique, and is the best way to demonstrate the precise anatomy of the aortic root in acute aortic dissection. (Courtesy of Hani Aziz, MD.)

Doppler colour flow mapping presents spatially oriented velocity data in real-time superimposed on an M-mode or 2D image for anatomical and temporal reference (Fig. 22.12). Components of velocities towards or away from the transducer are represented as shades of red or blue, respectively. Since the velocity data are displayed in real-time, the video image correctly gives the impression of blood moving through the cardiac chambers and vessels. With faster computers, 3D colour Doppler imaging has also become possible.

Figure 22.12 Transoesophageal echocardiography (TOE). (A) TOE is usually the most accurate method for imaging valve disease, as in this patient with endocarditis of the aortic valve complicated by perforation of a cusp (arrow). It is doubtful that any other imaging technique could approach the clarity of this image. (B) Echocardiography also has the advantage that accurate flow information is easily obtained along with anatomical information. Here colour flow Doppler shows severe aortic regurgitation (between arrows). Ao = aortic root, LV = left ventricle, LA = left atrium, RV = right ventricle. (Courtesy of Hani Aziz, MD.)

431

432

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pulsed Doppler echocardiography displays velocity information recorded from a well-localized area or sample volume identified on the 2D image. Most machines display this sample volume along a cursor identical to that used to obtain an M-mode echocardiogram. Pulsed Doppler echocardiography is used to assess localized valve stenosis or regurgitation, to quantify flow at a specific point in a chamber or vessel, and to detect abnormal flow patterns. Continuous wave Doppler echocardiography simultaneously displays the velocities of all volumes of blood intersected by the 1D cursor. In return for not being able to provide spatial separation of velocities, it is able to record blood velocities far in excess of the pulsed Doppler method.

Contrast medium echocardiography Contrast medium echocardiography is an exceptionally useful adjunct to ultrasound imaging, especially with the development of newer contrast agents6. The intravascular contrast effect can be produced when microbubbles of gas, present in solution, are injected into a peripheral vein or the injection of stable solutions of microbubbles prepared in advance. Some agents contain very small microbubbles which traverse the pulmonary vascular bed. Other agents filtered by the lungs opacify the left heart only when a right-to-left shunt is present. Contrast medium echocardiography may clarify anatomy incompletely identified by 2D imaging, or improve the definition of endocardial boundaries, improving the detection of wall motion abnormalities (see Fig. 25.10B).Valvular regurgitation and intracardiac or pulmonary shunts may be assessed by peripheral or catheter injection. Finally, myocardial perfusion can be evaluated by contrast medium enhancement after intracoronary or venous contrast injection.

There are two approaches to overcoming this problem, one of which now allows much more widespread use of CT for cardiac diagnosis, including coronary artery imaging8–10. The first but less commonly used approach is EBCT (Fig. 22.13, see Fig. 25.3). This approach uses an electronically steered beam of electrons aimed at a series of tungsten rings to generate X-rays. This allows very rapid pulsing and rotation of the X-ray beam producing typical exposure times of 33, 50, or 100 ms at up to 30 frames s−1, a level of performance not matched by MDCT. Pulsing the X-ray beam can be ECG triggered, minimizing blurring due to cardiac motion, and reducing radiation dose as exposure occurs only during useful image acquisition. Although there are many potential advantages to using this approach for cardiac imaging, for numerous technical and historical reasons there are relatively few EBCT machines in operation. The major current interest in cardiac CT focuses on the development of MDCT11. The development of MDCT (typically 16 or more rows of static detectors) with fast tube rotation, ECG triggering of image acquisition and ECG controlled modulation of tube current (and therefore of X-ray output), along with sophisticated post-processing, has opened up a wide range of options for cardiac imaging. Slice thickness can be collimated to as little as 0.4 mm. Fast tube rotation (which may be as little as 330 ms) and ECG triggering mean that image data can be acquired during only a short portion (typically 100– 200 ms) of the tube rotation (partial rotation reconstruction). When data are acquired toward the end of diastole, when there is usually the least cardiac motion, blurring and other artefacts are minimized (Fig. 22.14). Even the fastest current MDCT

Clinical applications Echocardiography can be helpful in clarifying the presence or absence of heart disease but is not recommended for routine screening of patients with ‘innocent’ murmurs, and should not replace thorough history taking and physical examination7.The value of echocardiography is seen in the diagnosis, differential diagnosis, or follow-up of virtually all types of cardiac disease, as well as risk stratification and assessment of cardiac function. Stress echocardiography incorporates the same techniques used to assess function at rest (e.g. shortening fraction, ejection fraction, cardiac output, segmental wall motion analysis) with a stressor such as exercise or a pharmacological agent such as dobutamine infusion (see Fig. 25.4). Doppler echocardiography can estimate changes in flow during exercise. Sensitivity and specificity depend on the protocol used, choice of correlative imaging, patient population (overall accuracy may be better in men), and nature of the disease (more sensitive for LAD disease when imaging single vessel disease, more sensitive for multivessel disease), and are quoted to be between 80% and 100%.

Computed tomography For many years CT had limited use in the evaluation of heart disease. The slow image acquisition led to image blurring.

Figure 22.13 Axial electron beam CT (EBCT) showing heavy calcification (straight arrows) anteriorly in the pericardium following localized radiotherapy and mitral annulus (curved arrow). Note the clear definition of the normal thickness pericardium (open arrow), which is well shown using the very short exposure (100 ms) possible with EBCT.

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.14 Pulsatility artefact on CTA. (A) Axial CTA image through the ascending aorta (AA) reconstructed without ECG gating. Note the double shadows (arrows) in the ascending aorta. The double densities in the aorta prevent confident exclusion of aortic dissection, and obscure the separation of the aorta from the right ventricular outflow (RV). There is also marked blurring of calcification in the distal left main coronary artery (arrowheads). (B) Axial CTA image through the ascending aorta (AA) now reconstructed with ECG gating at end-diastole using the same acquisition as in (A). The wall of the aorta is now clearly defined, as is calcification in the coronary arteries. In addition, areas of low attenuation plaque (arrows) are now visible. (C) Pulsatility effects are evident when viewing 3D reconstructions. Here an oblique maximum intensity projection reconstruction (MIP) shows a step artefact in the ascending aorta (AA, arrows) due to pulsatility. (D) The same 3D reconstruction using ECG-gated source images reconstructed at enddiastole produces a smooth wall of the ascending aorta (AA). There is also clearer definition of the coronary sinus (white arrows) running below the left atrium, calcification in the distal right coronary artery (white arrowhead) and calcification in the aortic valve ring (black arrow).

acquisition is longer than for EBCT and especially for coronary angiography (typically acquisition times of only 10 ms). Reliable imaging by MDCT requires a regular, relatively slow heart rate (< 75 beats min−1 is recommended), which may require medication, such as beta-blockers. Images are reconstructed in three dimensions to represent the complex anatomy of the coronary arteries12,13. Comprehensive evaluation may require extensive

post-processing using maximum intensity projection (MIP; Figs 22.1–22.6), curved multiplanar reconstruction (MPR; Figs 22.4D, 22.6D, see Fig. 25.7), surface shaded (SSD; Fig. 22.8C), and volume rendering techniques (VRT; Figs 22.1D,F, 22.2F,G, see Fig. 25.8). When done manually this is time consuming but semi-automatic methods are being developed which reduce processing time14.

433

434

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Cardiac evaluation Cardiac CT can be performed with or without contrast medium. Useful information has long been obtainable with conventional CT but image quality is markedly improved by ECG gating, especially when evaluating small structures. CT without contrast medium can show abnormalities characterized by differences in attenuation due to fat (e.g. lipomatous hypertrophy of the atrial septum, fatty degeneration of the myocardium; Fig 22.15), or calcification (coronary artery, aneurysm, valves, etc; Fig 22.13, see Fig. 25.3). Pericardial effusion or thickening is outlined between lung and epicardial fat. The position of implants and postoperative complications (such as focal fluid collections) can be seen (Fig. 22.16). Some anomalous coronary artery anatomy can be revealed even without the use of contrast medium, although accuracy is greatly increased with CTA (see Fig. 25.19). The use of intravenous contrast medium allows better evaluation of the cardiac chambers and adjacent vascular anatomy8–11. MDCT and EBCT are used to evaluate great vessel anatomy (arterial and venous), cardiac dimensions, ventricular function, viability, perfusion and coronary artery anatomy, including obstructive disease. With the exception of great vessel anatomy, where CTA is of proven accuracy and value, cardiac CT in these applications is in evolution. Artefacts due to metal are less of a problem than with CMR and some patients with implantable devices cannot be imaged by CMR (Fig. 22.16).

Three-dimensional image reconstruction High quality CTA and MRA produce a huge amount of image data. Although 2D source images can be used for diagnosis, this is time consuming and cumbersome. Most 3D datasets are best viewed using one of several 3D reconstruction

techniques. The best technique depends on the nature of the source data and the anatomy being viewed. For small diameter vessels, such as coronary arteries, submillimetre voxel dimensions should be used. Several techniques are widely used to present these data10,14.

Figure 22.15 CT diagnosis based on detection of low attenuation structures. The ability of CT to identify low attenuation areas allows some specific diagnoses even without contrast medium. This ECG-gated axial image shows low attenuation (indicating fat) in the interatrial septum (white arrows), right ventricular myocardium (white arrowheads) and left ventricular myocardium (indicating fatty degeneration of the myocardium; black arrows). Arrhythmogenic right ventricular dysplasia is also characterized by low attenuation but this is usually limited to the right ventricular free wall, and occasionally the septum.

Figure 22.16 Imaging by CTA in the presence of implanted devices. (A) CTA is less frequently used for evaluation of congenital heart disease than CMR, but in patients with contraindications to CMR (such as pacemakers and implantable defibrillators) CTA is useful, although still affected by metal artefacts. In this patient with anatomically corrected transposition of the great arteries, wires (white arrows) from a pacing system in the superior vena cava and right atrial appendage (RAA) produce some artefact. Note the abnormal anterior and left-sided position of the ascending aorta (AA) relative to the pulmonary artery (PA) as well as the abnormal origins of the coronary arteries (black arrows). (B) Artefact from denser material in the right atrial components of the pacing system (white arrows) obscures atrial anatomy and the transposed connection of the right atrium to the left ventricle (LV). A metal tricuspid valve replacement (black arrow) separates the left atrium (LA) from the transposed right ventricle (RV), identified by the presence of a prominent moderator band (M). (C) An oblique multiplanar reconstruction (MPR) reconstruction shows the abnormal relationship of the PA to the more anterior and left-sided AA as well as the infundibulum (I) which identifies the ventricle connecting to the AA as being a transposed morphological RV. Artefact (arrows) from pacing wires in the transposed anterior LV is included in this image although the wires are not in fact in this position.

CHAPTER 22

Multiplanar reconstruction This technique is used to view any desired 2D slice or section from a 3D dataset. When nearisometric voxels are acquired, the spatial resolution and image quality is the same for any orientation of the image. The use of MPR facilitates a view of relevant anatomy from an appropriate perspective; e.g. the coronary arteries can be viewed in a similar perspective as seen on conventional coronary arteriography. Thin sections should be used (typically only a few millimetres) as thicker sections may superimpose image data obscuring detail. Curved MPR sections can be generated manually or automatically to follow tortuous anatomy, which is especially useful when evaluating the coronary arteries (Figs 22.4D, 22.6D). Maximum intensity projection In this method the highest attenuation voxels (CTA) or highest signal voxel (MRA) along any ray (line) through the image are identified and projected into a 3D image; images of any required thickness can be rotated and viewed from any appropriate direction (Figs 22.1–22.5). This method is widely used in contrast-enhanced 3D MRA as well as CTA (Fig. 22.8B). Surface shaded display All voxels with an attenuation or signal above a certain level are represented as a 3D image which can be rotated in any direction. An imaginary source of illumination is used to provide a 3D perspective of the surface. This technique has been largely replaced by VRT for CTA. It can be useful for demonstrating large structures with uniform high attenuation or signal, such as the pulmonary veins using CTA or contrast-enhanced MRA (Fig. 22.8C). Volume rendered techniques Volumes with a certain range of attenuations or signal intensity are assigned a colour or grey scale density (Figs 22.1–22.5). Colours and threshold levels are varied depending on the application. This technique can be used for CTA or contrast-enhanced MRA, as well as imaging the airways, bowel and bone. Three-dimensional reconstruction benefits from editing of the 3D dataset to remove areas of high attenuation (e.g. bone for CTA) or signal (e.g. fat for MRA) from the image, which might obscure the area of interest. Such image reconstruction facilitates understanding of spatial relationships and allows evaluation of very large quantities of imaging data, but there is a significant potential for errors of interpretation if undue reliance is placed on just one type of image reconstruction15, or if the operator is not fully aware of the range of anatomical variations15,16. Typically, CTA covering 20 cm of the chest for cardiac imaging, using a 0.65-mm slice thickness, can generate over 300 source images. Although these images may need to be reviewed to identify other abnormalities, the vascular anatomy is usually best understood by reviewing the 3D images.

Arterial and venous great vessel anatomy Even before the widespread availability of ECG-gated CT, aortic and major venous anatomy was well demonstrated by CT, with or without contrast medium. Dimensions and branch vessel relationships are well shown along with major congenital anomalies of arterial connection (Fig. 22.16) and

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

venous drainage (Fig. 22.8). With ECG gating and contrast enhancement, CT is a definitive method for evaluating aortic dissection, planning endovascular aortic repair and assessing pulmonary venous anatomy, an increasingly important application in patients with atrial arrhythmias treated with radiofrequency ablation of the pulmonary veins.

Cardiac dimensions The assessment of cardiac dimensions and mass has long been an accepted application for EBCT. Although EBCT assessment of ventricular size and configuration has been shown to be accurate in many conditions, and especially superior to echocardiography when evaluating the right ventricle, the need to give contrast medium and to use radiation, as well as the limited availability of EBCT, has prevented its widespread use. Multidetector CTA (MDCTA) with ECG gating is also accurate but the limitations imposed by the need for iodinated contrast medium and radiation remain. Usually echocardiography is more than adequate. In situations where echocardiography is limited due to poor acoustic access or reproducibility, MRI is usually the preferred method, with a high level of accuracy established over nearly two decades. MDCTA is useful in patients who cannot have MRI.

Ventricular function EBCT has long been used to assess systolic cardiac function, but has not been accepted into widespread routine practice. MDCT can assess left ventricular function but there are concerns about the accuracy of endomyocardial definition due to the long period of image acquisition. For a heart rate of 70 beats min−1, systole lasts about 300 ms, allowing acquisition of only three 100-ms systolic images using conventional MDCTA.Although the images presented to support this application appear well defined, the precision with which systolic events in particular are demonstrated is in doubt. By comparison echocardiography, EBCT and cine-MRA typically provide 30-ms images for frame rates of 30 frames s−1 or more. At present MDCTA can be used to assess systolic function as part of a comprehensive cardiac evaluation, but with less precision than other techniques.

Assessment of myocardial viability and perfusion Passage of contrast medium through the myocardium can be analysed in a similar way to perfusion echocardiography and MRI, as well as first-pass studies on nuclear scintigraphy. Delayed enhancement may also indicate poor viability17. The usefulness of this approach is as yet unclear. There is much more experience and validation with nuclear techniques, while echocardiography and MRI techniques do not have the limitations of contrast medium-related toxicity inherent in CT perfusion techniques. There are reports of CT perfusion studies providing useful information as part of a comprehensive cardiac study in patients where contrast medium administration is not a major concern. To do this accurately requires careful attention to correction for contrast medium dilution as it passes through the heart, beam hardening effects as contrast enhancement progresses, and potential nonlinear attenuation due to the use of partial rotation reconstruction with data

435

436

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

acquired through different volumes of lung and bone. At present other techniques using scintigraphy, MRI, or ultrasound contrast agents are usually preferred.

Coronary computed tomography angiography Coronary angiography using noninvasive imaging by CTA (Figs 22.3–22.6) or MRA has long been the focus of substantial research. Although there have been promising reports of techniques using EBCT and a variety of MRA techniques, none has generated as much enthusiasm for real world clinical use as MDCTA3,14. Technical developments allowing thinner (< 0.5 mm) slice acquisition, rapid imaging and retrospective data reconstruction during a short period at end-diastole provide worthwhile coronary artery imaging18.19. There are limitations related to heavy calcification, which may prevent visualization of the artery lumen (see Fig. 25.12) and imaging of vessels smaller than 2 mm. In spite of this, MDCTA is already accepted as a definitive technique for demonstrating congenital coronary anomalies (see Fig. 25.19), bypass graft patency (Fig. 22.17, see Fig. 25.15), and ruling out significant coronary stenosis in low risk patients (see Ch. 25).

Magnetic resonance imaging The diagnostic power of MRI lies in its ability to provide a wide range of information, including anatomy, function, flow and chemistry, with minimal risk. Cardiovascular MRI (CMR) is well established, particularly for the assessment of congenital heart disease, cardiac function, tumours and diseases of the aorta and pericardium20,21.Technological improvements (e.g. 3T imaging, faster gradient switching, more powerful gradient amplifiers and new pulse sequences) have continued to expand the utility of CMR. Images, including cine-MRA,

can be acquired in one breath-hold period (which eliminates respiratory artefact). Different sequences can be used to assess anatomy (spin-echo, fast spin-echo, half-Fourier single-shot turbo spin-echo), function (time-of-flight or phase-contrast cine-MRA), perfusion or viability (with contrast enhancement) and vascular anatomy (time-of-flight, phase-contrast or contrast-enhanced MRA). The key to achieving accurate CMR is to establish a good ECG trace and to choose the appropriate imaging sequences to assess the clinical problem (Figs 22.18, 22.19). In general black blood imaging (spin-echo, turbo spin-echo) is used for anatomical imaging (Figs 22.7, 22.19A), while cine-MRA (white blood imaging) is used to assess dynamic function (Figs 22.10, 22.18). Phase-contrast imaging is used to measure blood flow velocity. Contrast enhancement can be used for extracardiac angiography using 3D gradient-echo sequences (Figs 22.8, 22.20), while inversion recovery type sequences (with or without contrast medium injection) can be used to identify areas of myocardial injury22. A variety of well validated analysis and 3D reconstruction techniques are available to provide accurate assessment of cardiac dimensions, wall motion, perfusion, flow and 3D anatomy. Used correctly, CMR is a definitive, stand-alone cardiac imaging technique with no need for alternative corroborative imaging. It is only in the search for coronary artery stenosis that CMR has as yet failed to make a significant impact.

Cardiac gating CMR studies are degraded by motion artefact unless cardiac gated or the more difficult ‘one-shot’ technique is used. Quantitative measurement of flow requires cardiac synchronization to avoid errors. Even with synchronization, some errors occur

Figure 22.17 CTA of coronary artery bypass grafts. (A) Thin maximum intensity projection (MIP) showing patent saphenous vein bypass graft (arrows) to the distal right coronary artery. Also note thrombus in the left pulmonary arteries (arrowheads). (B) VRT 3D image shows three patent grafts passing towards the right coronary artery and two left-sided branches (arrows). AA = Ascending aorta, LA = left atrium, PA = pulmonary artery, RA = right atrium, RV = right ventricle, SVC = superior vena cava.

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.18 Correct choice of MRA imaging sequence. (A) Four-chamber long axis view of the left ventricle (LV) using cine-FLASH sequence shows poor definition of the endocardial surface of the LV and left atrium (LA). This is a reflection of low signal from slow, in-plane flow. There is higher signal from the right atrium (RA) and right ventricle (RV) where the flow is passing obliquely through the imaging plane. (B) Four-chamber long axis view of the LV using cine-FISP in the same slice position in the same patient. This sequence produces higher signal and much better definition of the endocardial surfaces (arrows).

Figure 22.19 Tissue characterization using multiple CMR sequences. (A) Axial, turbo spin-echo MRI image shows a high signal in a smooth, possibly cystic, well-defined mass (C) posterior to the left atrium (LA). High signal suggests fluid or viscous contents. (B) Sagittal, cine-FISP image shows that the lesion behind the LA has high signal similar to that of flowing blood in the cardiac chambers, suggesting the contents are fluid. (C) Coronal, T1-weighted HASTE image shows high signal in the mass and confirms that it is well defined. (D) Coronal, inversion recovery HASTE image shows intermediate signal. This combination of signal characteristics in a well-defined, rounded mass in this position indicates that this is a benign cyst, in this position most probably a bronchial cyst. This was confirmed at surgery.

437

438

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 22.20 Multi-sequence comprehensive evaluation of postoperative changes. (A) Axial ECG-gated spin-echo image showing bifurcation of the pulmonary artery (PA) directly anterior to the ascending aorta (AA) in an 8 year old patient following surgical correction of transposition of the great arteries. The vessel lumens are black (signal void due to flow). 3D image reconstruction with this type of image is possible but difficult. (B). Coronal thin maximum intensity projection (MIP) reconstruction from coronal acquisition. Contrast-enhanced MRA shows dilatation of the ascending aorta (AA) separating the left (LPA) and right (RPA) pulmonary arteries. (C) Lateral view MIP from the same CE-MRA shows the relationship of the aorta to the main pulmonary artery (MPA) as well as minor narrowing at the site of repair of coarctation (arrow). DA = descending aorta, LV = left ventricle, RA = right atrium, RV = right ventricle.

due to beat-to-beat variations resulting from a number of physiological variations including respiration. There are two forms of cardiac gating. Prospective gating detects the QRS complex of the ECG, which then triggers the application of the sequence, such that the image is formed from a specific part of the cardiac cycle. For cine-MRA, the sequence is repeated rapidly in order to acquire images at a number of stages of the cardiac cycle; the repeat time and number of pulse repetitions are adjusted such that, when added to the initial delay, they require slightly less time than the shortest expected R–R interval. The other form of cardiac synchronization is retrospective gating. For this approach, the sequence is repeated continuously with a constant repeat time while the ECG is monitored. The data are sorted at the completion of the sequence and adjusted to give images at specific parts of the cardiac cycle. In this way variations in the R–R interval can be corrected. Prevention of respiratory motion is especially important in coronary artery imaging. Coronary MRA using breathhold acquisition is difficult because patients are required to breath-hold at the same diaphragm position and therefore slice misregistration is common. In addition, the signal-tonoise ratio and spatial resolution are inherently low because of the limited acquisition time of a single breath-hold. Navigator techniques monitor respiratory motion by tracking the position of the diaphragm in real-time, which is used to select and reconstruct only data acquired in a selected part of the respiratory cycle23.

The heart under stress Cardiac function in patients with coronary artery disease may be within normal limits at rest and unless it is possible to study the effects of stress on such patients, the value of CMR is lim-

ited. Stressing patients in the MR machine using an exercise apparatus is possible but difficult. Pharmacological stress, using dobutamine or adenosine, can be applied and is well established for assessing myocardial ischaemia24,25.This technique is used in a similar way to stress echocardiography, but has no limitations from restricted acoustic access. The level of stress is increased while monitoring wall motion, with the development of a new wall motion abnormality (which precedes ECG changes) indicating ischaemia. This technique can be combined with perfusion and viability imaging using a contrast agent.

Contrast agents The most widely clinically used MR contrast agents are paramagnetic compounds that influence the relaxation times of adjacent tissues26. Gadolinium is a powerful paramagnetic substance which when complexed with ligands such as diethylenetriaminepenta-acetic acid (DTPA) is no longer toxic. The main indications for the use of contrast agents in CMR are: assessment of myocardial perfusion, contrast-enhanced MRA, improved demarcation of normal and infarcted myocardium and delineation of viability after coronary occlusion (Fig. 22.21, see Fig. 25.15)22,24.

Pulse sequences There are innumerable pulse sequences, often known by acronyms for ease of recall. In cardiac imaging the most important are the spin-echo, gradient-echo and velocity mapping sequences. Spin-echo imaging In spin-echo images the myocardium and vascular wall appear bright and the blood is dark (black blood imaging). Spin-echo CMR is useful for high-definition anatomical imaging of the heart and vessels but it is rather slow (Figs 22.7A, 22.20A).Variable signal from slow or in-plane

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.21 Evaluation of cardiac perfusion and viability by CMR. Multiplanar cine-MRA is used to evaluate systolic cardiac function in multiple orientations including usually four-chamber long axis and short axis. (A) Four-chamber long axis view from cine-FISP MRA at end-diastole shows apical myocardial thinning. The position of the perfusion image (C) and scout images (D–H) used to assess optimum inversion times for viability imaging is shown (white line). (B) Four-chamber long axis view from cine-FISP MRA at end-systole shows persisting apical myocardial thinning (arrows) while the basal myocardium thickens normally. (C) Short axis view obtained during first pass of contrast (gadopentetate) through the myocardium shows reduced enhancement in the septum (arrows). (D–H) Five images from a series of 10 acquired at varying inversion times (TT) to identify the best parameters to null signal in the normal myocardium (F, asterisk) compared with delayed enhancement of nonviable myocardium best shown at 182 ms (F, arrows). (I) Delayed enhancement four-chamber long axis image shows extensive apical and septal enhancement (arrows) indicating nonviable myocardium.

blood flow produces artefacts. Fast spin-echo (FSE; Fig. 22.19A), half-Fourier single-shot turbo spin-echo (HASTE; Fig. 22.19C,D) and multi-echo, spin-echo sequences that change the phase-encode gradient for each echo allow acquisition of multiple lines of k-space in a given repetition time (TR). This significantly reduces imaging time. Using different

pulse sequences, different signals can be extracted from lesions to indicate the nature of the tissue (Fig. 22.19D). Gradient-echo imaging In this sequence, blood produces a higher signal (white blood imaging) as only one radiofrequency (RF) pulse is used and there is less time for blood to

439

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

move out of the imaging plane between slice selection and data acquisition. The gradient-echo sequence can be repeated more rapidly with a reduced RF flip angle, allowing the acquisition of cine loops (Figs 22.10, 22.18). Cine acquisitions are useful for the assessment of valves and regional contractile function of the myocardium. Turbulent and other irregular flow types, which contain higher order motion, produce signal loss on gradient-echo images (e.g. valve stenosis or regurgitation). Shortening echo times (TE) reduces phase incoherence and loss of signal. The shorter the TE, the higher the threshold of turbulence intensity at which signal is lost. Phase-shift velocity mapping The MR signal has three components: frequency, amplitude and phase, which are used in image reconstruction. It is possible to construct an image to display phase changes caused by special sequences. If a gradient is applied in the direction of blood flow for a finite time and then turned off, the relationship of phase will change in relation to the two ends of the gradient. The protons at the stronger end of the gradient precess at a faster rate during its application than those at the weaker end. When the gradient is turned off, the rate of precession becomes constant again in the slice, but the phase relationship has changed and the phase signature remains. When the gradient is reversed and applied for the same period of time as the original gradient, the phase signature of still material is cancelled, but flowing blood moving during the gradients to a different phase territory retains a phase change proportional to velocity (Fig. 22.22).

Quantification of myocardial contraction Myocardial tagging has been used to study regional wall motion and myocardial strain (see Fig. 25.6). In this technique modulation of the MRI is achieved using alternating RF pulses and magnetic gradients such that narrow planes of very low signal appear in a latticework. Computer analysis can derive quantitative measurements of intrinsic myocardial mechanics by tracking the motion of the points of intersection of the tagging lines. The results are observer independent and objective. This technique should become more widely used once the computer analysis issues have been ironed out and made more widely available. Myocardial motion can also be quantified using MR velocity mapping.

Magnetic resonance angiography (see also Chs 6, 27 and 28) Tomographic imaging is not ideal for demonstrating tortuous vessels and 3D imaging methods are often required. Although 3D imaging can be performed using time-of-flight or phasecontrast MRA, the bulk of extracardiac studies are performed using contrast-enhanced MRA. Contrast-enhanced MRA has become increasingly popular as it can be acquired during a single breath-hold (> 20 s). Contrast-enhanced MRA relies on fast acquisition of images after the administration of a gadolinium contrast agent (Figs 22.18, 22.21), which produces signal enhancement due to the shortening of the T1-relaxation time of blood26. The result is optimal when data collection (in particular the centre of the k-space) occurs when the contrast agent arrives at the targeted vessel. Therefore timing of the data acquisition with respect to the intravenous injection of contrast agent is important. Contrast-enhanced MRA is particularly useful for the comprehensive evaluation of aortic disease and major branch vessel abnormalities (renal, mesenteric, peripheral, carotid), as well as venous disease (Fig. 22.23). The use of contrast-enhanced MRA for evaluating many aspects of congenital heart disease is well defined (Figs 22.20, 22.22).

Clinical applications Myocardial function at rest CMR images are dimensionally accurate.The volume of a chamber can be measured in systole and diastole by summing the areas in contiguous images.Ventricular stroke volumes are now commonly measured using multiple contiguous fast cine gradient-echo imaging acquired during a breath-hold to cover the ventricles.The end-diastolic and end-systolic volumes are computed from the endocardial outlines. Volume measurements by these methods are independent of cavity shape, unlike other methods where geometric assumptions usually have to be made, and are ideal for the right ventricle. In theory, it does not matter which plane is used for the measurements, but the short axis plane appears to be the best because of the good contrast between blood and myocardium, although endocardial measurements at the base

60 AA DA

50 40 30

Flow (l min−1)

440

20 10 0 −10 −20 −30

B

C

0

200

400

600

Time (ms)

Figure 22.22 Gradient-echo images in a coronal plane acquired during ventricular (A) systole and (B) diastole in a patient with Marfan's syndrome and aortic valve regurgitation (arrow). aa = ascending aorta, lv = left ventricle, pa = pulmonary artery, ra = right atrium. (C) Flow volume curves in the ascending aorta (AA) and descending thoracic aorta (DA) measured from the complete line acquisition. Note the large reverse flow during diastole.

CHAPTER 22

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

Figure 22.23 Comprehensive cardiovascular evaluation in complex congenital heart disease using contrast-enhanced MRA. (A) An advantage of contrast-enhanced MRA is that multiple vascular territories can be evaluated safely with the same study. In particular the major veins and arteries which may be affected by complications of surgical treatment of congenital heart disease are well shown. In this 18 month old patient with central line-related endocarditis following closure of a patent foramen ovale and repair of a ventricular septal defect, a coronal, thin (6-mm thick) maximum intensity projection (MIP) image shows thrombus (arrow) in the superior vena cava (SVC). (B) A more posterior thin MIP generated from the same 3D acquisition shows a shelf (arrow) at the level of a coarctation and a large false aneurysm (FA) of the abdominal aorta. There are areas of reduced pulmonary enhancement (arrowheads) indicating pulmonary infarction from emboli. (C) As the imaging acquisition generates a 3D dataset with isometric voxels it is possible to reconstruct images in the best orientations to visualize the abnormalities detected. A sagittal, thin MIP through the descending aorta shows the eccentric shelf of the coarctation (arrow). (D) A sagittal, thin MIP through the SVC shows the thrombus (arrow) attached to the anterior wall of and partly obstructing the SVC. AA = ascending aorta, LA = left atrium, PA = pulmonary artery, RA = right atrium, RV = right ventricle.

can be difficult in the short axis. Muscle volume and hence mass can also be measured.

assessed. Myocardial tagging has been used to study regional wall motion, myocardial strain and myocardial viability.

Myocardial function during stress Cardiac function at rest may be normal in the presence of severe coronary artery disease. Abnormal function may represent neither coronary artery disease nor irreversible damage (hibernation). Therefore, a full assessment of cardiac function should include stress assessment for inducible ischaemia and reversible dysfunction21,23. This may also be achieved using similar ultrasound and nuclear imaging techniques. Reversible myocardial ischaemia can be shown on cineMRA, following pharmacological stress, by the development of abnormal wall motion. Using low dose dobutamine infusion, residual myocardial mass (viability) in areas of asynergy can be

Myocardial perfusion and viability Myocardial perfusion imaging can be combined with wall motion analysis, stress testing and evaluation of viability (Fig. 22.21, see Fig. 25.15). MR contrast agents equilibrate quickly with the extracellular compartment and are washed out rapidly. They are therefore neither blood pool tracers nor fixed in the myocardium. Nevertheless, by treating them mainly as extracted tracers, it is possible to measure myocardial perfusion from the peak myocardial enhancement. The model assumes a linear relationship between tracer concentration and signal enhancement. This is only true for gadolinium–DTPA at low concentrations, which appears to be the case in animal experiments and in clinical

441

442

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

practice. Ultrafast gradient-echo imaging can provide at least one image of each cardiac cycle during the passage of the tracer; measurement of myocardial perfusion is feasible as part of the routine evaluation of ischaemic heart disease (Fig. 22.21C). Delayed contrast enhancement shown by inversion recovery techniques which null the signal from normally perfused viable myocardium (Fig. 22.21D–I, see Fig. 25.15C) shows promise in assessing the size of myocardial infarction and detecting nonviable myocardium with accuracy similar to nuclear imaging, stress echocardiography and positron emission tomography27. Delayed hyperenhancement on contrast-enhanced MRI can identify recent myocardial infarction or nonviable myocardium in those with chronic ischaemia (Fig. 22.21, see Fig. 25.15). Absence of hyperenhancement correlates well with nuclear imaging and echocardiographic determinations of viability, regardless of resting contractile function. Coronary arteries and bypass grafts Coronary MRA can be performed using a variety of techniques23,28,29. Although patency can be determined by 2D breath-hold techniques the best 3D images of the coronary arteries and bypass grafts are obtained using 3D techniques combined with navigator techniques and using contrast enhancement23,29,30. Coronary MRA is useful for the assessment of anomalous coronary arteries, patency of the infarct-related coronary artery and saphenous vein graft patency28. Although good images of coronary arteries are obtained fairly easily in normal volunteers, current techniques are less reliable for defining coronary artery stenosis30. Valvular heart disease In valvular heart disease, CMR information complements and can go beyond that obtainable by echocardiography31. Although CMR is less suitable than echocardiography for visualizing valve leaflets and suspensory apparatus or infective vegetations, it can, in most cases, measure regurgitation. CMR velocity mapping allows measurement of peak velocities across stenoses. Cine-MRA provides a wide, unrestricted view of dimensions, movements and relations of chambers and visualization of jets (Figs 22.7, 22.9). CineMRA in multiple contiguous slices allows accurate, reproducible measurements of ventricular volume and myocardial mass, which can be important in relation to valve disease. CMR provides quantification of valvular regurgitation and stenosis, assesses atrial and ventricular size and function, visualizes associated intracardiac thrombus, detects paravalvar abscesses and fistulas, and assesses prosthetic valves. Gradient-echo sequences show turbulent blood flow as areas of signal loss within the high blood signal. The degree of signal loss is dependent on the echo time, field strength and other acquisition parameters.The severity of stenosis can be evaluated using phase-contrast MRA and the Modified Bernoulli Equation, which in its simplest form is: ∆P = 4V 2 where ∆P = pressure drop across the stenosis (mmHg) and V = velocity (m s−1). This assumes that the blood velocity upstream of the stenosis is negligible. This calculation can be easily applied to determine the peak instantaneous gradient and the mean gradient across the stenosis. It can be used to determine the pressure difference between two chambers with a regurgitant valve.

Other established applications of CMR20,21 These include: • Pericardium: tumour invasion, pericardial thickness, pericardial effusion, congenital absence of the pericardium. • Cardiac thrombus and tumours: detection of thrombus and tumours, identification of tumour types from signal characteristics; this is mainly limited to lipomas (high signal) and fibromas (low signal)32. • Cardiomyopathy: identification and assessment of the three main types of cardiomyopathies: hypertropic, dilated and restrictive. • Arrhythmogenic right ventricular dysplasia: focal thinning and abnormal motion of the right ventricle with or without signal changes in the involved myocardium. • Acute myocarditis: focal wall motion with abnormal high signal on T2-weighted images and contrast enhancement on T1-weighted images33. • Cardiac infiltration: sarcoidosis (high signal on T2-weighted images; contrast enhancement on T1-weighted or delayed contrast-enhanced MRI); amyloidosis (thickening of the myocardium, interatrial septum, valves, leaflets and papillary muscles, atrial dilatation, pleural and pericardial effusions; high signal on T2-weighted images; and contrast enhancement on T1-weighted or DE-MRI); myocardial iron overload (diffuse reduction in contraction, reduced myocardial signal). • Congenital heart disease: definition of great vessels and attachments, atrioventricular connection, extracardiac shunts, postoperative changes.

Limitations CMR has long been touted as a potential means of providing comprehensive cardiac imaging. Although CMR provides some of the most accurate information on cardiac anatomy, function, perfusion and viability, there are limitations. Coronary MRA is still not robust enough to replace routine coronary angiocardiography. Up to 4% of the normal population is claustrophobic and some unstable patients may not tolerate an examination. Controversy surrounds the safe MR of patients with pacemakers or implanted defibrillators. Some metal implants are a contraindication, although many extracerebral metallic objects such as sternal wires, metallic clips and prosthetic valves are not ferromagnetic and are safe in current clinical scanners34. However, prosthetic material gives little MR signal and metal causes a localized image defect due to distortion of the magnetic field. The defect is small for spinecho images but larger for some gradient-echo images, making it difficult to assess velocities and blood flow around the prosthesis. Other limitations include long acquisition times (which are getting shorter), poor image quality in patients with cardiac arrhythmias and cost.

Nuclear cardiac imaging Nuclear cardiology is an established subspecialty that has grown significantly in recent years because of developments in imaging hardware, software and tracers. Alongside these technical developments, there has been an increasing appreciation of the role that the functional information provided by nuclear techniques can play in clinical cardiology. This chapter covers

CHAPTER 22

only basic principles of nuclear cardiology and the most commonly used techniques, and it reviews some more common clinical applications, emphasizing the role of myocardial perfusion imaging.

Techniques (see also Ch. 7) Single photon emission computed tomography (SPECT) Tomography is an extension of the multiprojection method of acquiring planar images. Multiple planar projections are acquired as the camera rotates around the patient. The images are reconstructed into tomograms and presented as images in orthogonal planes parallel and perpendicular to the long axis of the left ventricle. SPECT is used routinely for myocardial perfusion studies, with accuracy approaching that of PET35. Its advantages include a true 3D display of the distribution of radionuclide, improved image contrast because of the elimination of overlying structures, and the potential for quantification of tracer uptake. Positron emission tomography (PET) Positrons travel a very short distance in matter before an annihilation reaction with an electron causes the emission of two gamma photons in opposite directions. Photons sensed simultaneously by opposing detectors are presumed to come from annihilation along the line between the detectors. Coincidence detection allows tissue attenuation to be determined accurately, which has important implications for quantifying tracer uptake. PET has several advantages over SPECT for cardiac imaging, including better spatial resolution, higher sensitivity and the ability to measure tracer distribution in absolute terms as a function of time30. PET is relatively expensive and is not widely available, but a conventional gamma camera can be used, either by using a high-energy collimator designed for 511 keV photons, or by using opposing detectors without collimation but with coincidence imaging. Radiopharmaceuticals Thallium-201 is still commonly used for myocardial perfusion studies. It is administered intravenously as thallous chloride and the usual dose is 80–110 MBq (maximum 80 MBq in the UK). Approximately 88% is cleared from the blood after the first circulation, with 4% of the injected dose localizing in the myocardium. Distribution in the myocardium is proportional to perfusion over a wide range of values, although at high rates of flow, extraction may become rate limiting. Thallium is an excellent tracer of myocardial perfusion and it has been used clinically for more than two decades, but has significant limitations. The long physical half-life results in a radiation dose of the same order as coronary angiography. The relatively low injected dose results in a low signal-tonoise ratio and images can be suboptimal, especially in obese patients. Third, the relatively low-energy emission leads to low-resolution images and significant attenuation by soft tissue. Technetium-99m (99mTc) compounds do not have these limitations and this has encouraged the development of such tracers.

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

99m

Tc-labelled tracers include 99mTc-2-methoxy-isobutyl-isonitrile (MIBI), 99mTc-1, 2-bis[bis(2-ethoxyethyl) phosphino] ethane (tetrofosmin) and 99mTc-teboroxime. Each of these agents has specific attributes but all have advantages over thallium due to the superior imaging characteristics of 99m Tc. Positron-emitting radiopharmaceuticals A number of positron-emitting radiopharmaceuticals can be used as tracers of myocardial perfusion. [15O]-Labelled water can be administered either intravenously or by inhalation of [15O]CO2 with rapid transformation to water by carbonic anhydrase in the lungs. Tracers such as nitrogen-13 ammonia (as [13N]NH4+), rubidium-82 and potassium-38 have also been used. Kinetic models of these tracers have also been used to calculate myocardial perfusion. Tomography is now almost universally used for myocardial perfusion imaging, although planar imaging is sometimes used with older equipment or if the patient is unable to lie still and flat for the duration of tomographic imaging. The best imaging protocol depends upon the tracer used, the dose given, the type of collimator and the number of heads of the camera. Thallium is usually given at peak exercise with exercise continued for 2 min in order to maintain stable conditions over the period of extraction by the myocardium. Imaging starts within 5–10 min of injection and should be completed within 30 min. During this period the distribution of thallium within the myocardium is relatively fixed and the images reflect myocardial distribution at the time of injection. Thallium washes out of the myocardium at a slower rate in underperfused than normally perfused myocardium. Areas with decreased initial uptake appear to have a relative increase in uptake when imaged 2–4 h later, a phenomenon known as redistribution. Comparison between the stress and redistribution images distinguishes between the reversible defect of inducible hypoperfusion and the fixed defect of myocardial necrosis, although in some cases redistribution may be incomplete at 4 h. A second injection of thallium can then be given and re-injection images acquired for a more accurate assessment of myocardial viability. Technetium-labelled tracers Unlike thallium, MIBI and tetrofosmin are essentially fixed in the myocardium with no redistribution, and separate injections are used to assess stress and resting perfusion (see Fig. 25.5) with a high degree of accuracy37. The 6-h half-life of 99mTc means that the two studies should be performed on separate days to allow for the decay of activity from the first injection. The 2-d protocol is performed with an initial stress study, followed by a resting study on a different day using a similar dose. Imaging starts 30–60 min after injection.The two studies can be performed on the same day if a larger dose is given on the second occasion in order to swamp residual activity from the first injection, and some time is allowed also for the first injection to decay.When the diagnosis of ischaemia is important, the stress study should be performed first in order to avoid reducing the contrast of a stress-induced defect by a previous normal resting study. Conversely, if the objective is to detect viable myocardium and reversibility of a

443

444

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

defect, perhaps in a patient with previous infarction, then it is better to perform the rest study first. Image interpretation Normal myocardial perfusion and hence tracer distribution is uniform. Regional defects indicate either reduced perfusion in viable myocardium, or a reduced amount of viable myocardium, or a combination of both. If a stress defect returns to normal in the resting images, this indicates an inducible perfusion abnormality, which is sometimes referred to as inducible ischaemia (see Fig. 25.26). Areas of infarction show defects on both stress and rest images and the depth of the defect indicates the myocardial loss differentiating large from small infarcts (see Figs 25.16, 25.17). Ischaemia can be superimposed upon partial thickness infarction and in this case a stress defect may be only partially reversible at rest. Fixed perfusion defects can also occur due to overlying soft tissue, something to be especially considered in women, even with SPECT (Fig. 22.24). Movement can produce uneven signal distribution (Fig. 22.25), potentially obscuring significant perfusion defects. Reverse distribution is the term used to describe a defect in redistribution thallium images that is less apparent in the stress images. When the phenomenon is seen, it is often the result of artefact or other technical differences between images and, in this case, it is not strictly speaking reverse redistribution. When true reverse redistribution is seen, it is the result of rapid washout of tracer as might occur in an area of partial thickness infarction supplied by a patent artery after angio-

Figure 22.25 Motion artefact. Movement during image acquisition for cardiac scintigraphy blurs the image and by obscuring or even producing areas of reduced activity can produce nondiagnostic studies. Here movement during the stress phase (rows A and C) of a nuclear stress perfusion study shows distortion of activity with artefactual thickening (arrows) of the myocardium, not seen at rest (rows B and D). The imaging agent is 99mTc-Tetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). The images in rows A and B are short axis views; those in rows C and D are vertical long axis views. (Courtesy of Laurie Gianturco, MD.)

plasty or thrombolysis. It is not associated with future coronary events and it is not normally clinically significant. Perfusion defects are not always the result of coronary artery disease. Abnormalities can be seen with coronary spasm, anomalous arteries, muscle bridges, small vessel disease as may occur in diabetes or syndrome X, the dilated and hypertropic cardiomyopathy, hypertrophy caused by outflow obstruction or hypertension, infiltrative disorders such as sarcoidosis and amyloidosis, connective tissue disorders and conduction defects such as left bundle branch block. True perfusion defects must also be distinguished from artefact caused by motion and by attenuation. Attenuation can be seen as reduced counts in the inferior wall, often in slim men, or in the anterior wall, often from breast attenuation in women.

Figure 22.24 Attenuation artefact. There are several causes of artefactual reduction of activity on cardiac scintigraphy. Increased overlying tissue as with the female breast (shown here), breast augmentation, or a high diaphragm, can reduce activity and mimic a fixed perfusion defect (arrows; row A stress, row B rest). The imaging agent is 99mTc-Tetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). (Courtesy of Laurie Gianturco, MD.)

Stress techniques The response to stress is central to the assessment of most aspects of cardiovascular function. This is particularly so for coronary artery disease because resting coronary flow is normal until the luminal area of a coronary artery is reduced by approximately 85–90% (equivalent to a 60–75% reduction in diameter), and resting ischaemia only occurs when the artery is virtually occluded. The most commonly used technique is dynamic exercise. This has the advantage that it mimics the stress that may provoke symptoms in patients with coronary artery disease. However, many patients with coronary artery disease are unable to exercise adequately.

CHAPTER 22

The potent coronary arteriolar dilator dipyridamole increases myocardial perfusion by between 250% and 600%. Perfusion reserve is reduced by a coronary stenosis, and the differential flow between territories served by normal and by stenosed arteries may produce a defect of uptake, even though flow may be increased in the abnormal area. The radiotracer is given 4 min after the end of the dipyramidole infusion, at which time vasodilatation is maximal. Myocardial perfusion remains near peak levels for up to 10 min, returning to baseline over 30 min. Adenosine has a very short plasma half-life, which increases mean coronary flow by 4.4 times the resting value. During infusion there is a slight fall in blood pressure with an increase in heart rate, cardiac output and pulmonary capillary pressure. Sideeffects are more common, occurring in up to 80% of patients. Dobutamine may be used in patients with asthma or other contraindications to dipyridamole or adenosine. Dobutamine increases perfusion in normal myocardium approximately twofold (similar to exercise) by a combination of coronary vasodilatation and increased myocardial oxygen demand by increasing myocardial work, leading to a secondary increase in myocardial perfusion. Dobutamine is safe in patients with coronary artery disease, and serious complications are uncommon. Imaging ventricular function Ventricular function can be imaged either by following the first pass of a tracer after injection, or by labelling the blood pool for an equilibrium study. Imaging of the cardiac blood pool is normally performed in the left anterior oblique projection that best separates the left and right ventricles: the ‘best septal projection’. Fifteen to 30 degrees of caudal tilt also separates the left ventricle and atrium. ECG gating is used to acquire 16–32 frames through the cardiac cycle. Once the background is subtracted, the left ventricular ejection fraction can be calculated from maximum and minimum counts within the left ventricular region of interest as follows:

EF = ED − ES ED Where EF = Ejection fraction, ED = End-diastole counts and ES = End-systole counts. This method makes no assumptions about ventricular geometry, and it is therefore superior to techniques such as echocardiography and X-ray ventriculography, particularly for abnormal and irregular ventricles. First-pass studies are also performed to measure shunting in patients with congenital heart disease. The pulmonary activity–time curve from a first-pass study shows a single peak as the bolus passes through the lungs, and a later more diffuse one as it recirculates. Myocardial infarction imaging A number of compounds localize in areas of myocardial infarction, including derivatives of tetracycline and dicarboxylates, and the bone-seeking agents such as pyrophosphate. 99mTc-pyrophosphate and 111Inlabelled antimyosin have been used most extensively.

Clinical applications Diagnosis of coronary artery disease—chronic chest pain Published figures for sensitivity and specificity of myocardial

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

perfusion imaging vary widely depending upon the population studied (sex, presenting symptoms, medication, presence of previous infarction), the imaging technique used, and the experience of the diagnostic centre. Using SPECT imaging, good accuracy can be achieved with sensitivity of 91% and specificity of 89%. This is significantly better than exercise electrocardiography, for which a large meta-analysis reported sensitivity of 68% and specificity of 77%. Commercially available tracers have similar accuracy for the detection of coronary artery disease. Thallium has better uptake characteristics and, in theory, provides defects with greater contrast, but 99mTc-based images are superior in terms of resolution and susceptibility to attenuation artefact. Diagnosis of coronary artery disease—acute chest pain In general terms, nuclear techniques are less commonly used in patients presenting with acute chest pain, mainly because of the logistical problems of imaging in the emergency department or coronary care unit. A resting perfusion defect has a high positive predictive value for acute infarction in patients without previous infarction, particularly if associated with a wall motion abnormality38. A normal perfusion scintigraphy excludes acute infarction and suggests that exercise ECG or stress perfusion imaging should be the next diagnostic step. Prognosis in coronary artery disease Beyond diagnosis, the most valuable contribution that perfusion imaging can make to the management of known or suspected coronary artery disease is to assess the likelihood of a future major coronary event39. Prognosis is strongly influenced by the extent and severity of inducible perfusion abnormality and this can guide the need for invasive investigation and revascularization. Myocardial perfusion imaging is a better indicator of prognosis than clinical assessment, exercise ECG, or coronary angiography, and it provides incremental prognostic value even after the other tests have been performed. The addition of SPECT information on ventricular function provides further useful prognostic information. Preoperative risk assessment A common clinical problem is that of assessing cardiac risk in patients who require noncardiac surgery. In this as in other clinical settings, perfusion imaging provides useful prognostic information although these patients are generally at low cardiac risk and the predictive value of a normal perfusion study is greater than that of an abnormal study. The decision on whether investigations for risk assessment are required should be based upon the urgency for surgery and its own cardiac risk, the risk factors of the individual and his/her exercise tolerance. Patients with only minor clinical predictors (older than 70 years, abnormal resting ECG, history of stroke or hypertension) who are scheduled for low to moderate risk surgery are not at high risk and do not require further investigation. Patients with intermediate clinical predictors (mild angina, prior cardiac infarction, treated heart failure, or diabetes) or with minor predictors and impaired exercise tolerance need further assessment if they are to undergo moderate or high risk surgery. Patients at high clinical risk (recent infarction or

445

446

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

unstable angina, severe heart failure, or significant arrhythmias) may require investigation even for low risk surgery. Management after myocardial infarction An important aspect of management after infarction is to identify patients at high risk of further events and to intervene to prevent them. Indicators of high risk in the acute phase include hypotension, left ventricular failure and malignant arrhythmias. These patients are candidates for early coronary angiography. After the acute phase, however, prognosis is related to the degree of left ventricular dysfunction and the extent and severity of residual ischaemia. Left ventricular ejection fraction (LVEF) at the time of discharge or 10–14 d after infarction predicts mortality40. Patients with impaired LV function, in particular, can benefit from perfusion imaging to assess viability of the infarct zone and determine whether remote myocardium is also ischaemic. Patients without high risk clinical markers or severely impaired LV function are at lower risk, but some form of stress testing is required in order to assess exercise tolerance and the presence of residual ischaemia. Viability and positron emission tomography Positronemitting radionuclides can assess myocardial perfusion and metabolism36. This relies on the detection of reduced resting perfusion assessed by [13N]ammonia with normal or increased fluorodeoxyglucose (FDG) uptake (perfusion–metabolism mismatch), as opposed to a matched reduction of perfusion and metabolism in fully infarcted myocardium or scar. The reported positive and negative accuracies for predicting recovery of regional function range from 72% to 95% and from 75% to 100%, respectively. Mismatch also predicts improvement in global left ventricular function after revascularization. Thallium is also widely used to identify viability and hibernation. Because redistribution can be slow or incomplete in regions of reduced perfusion, the usual stress/redistribution protocol can underestimate myocardial viability, and additional steps to ensure complete assessment of viability are required. These include late redistribution imaging at 8–72 h after stress injection, re-injection of tracer at rest after redistribution imaging and a resting injection on a separate day with both early and delayed imaging. 99m Tc-labelled agents have also been used for the detection of viable and hibernating myocardium. In theory these tracers may underestimate viability in areas with reduced resting perfusion because they are combined tracers of viability and perfusion without the property of redistribution that allows viability to be assessed independently. Other myocardial disease Many agents, including alcohol, drugs, viruses and metabolic defects, can damage the myocardium. Heterogeneous uptake of perfusion and metabolic tracers is seen in most patients with dilated cardiomyopathy although the findings are not specific.The distinction between idiopathic dilated and ischaemic cardiomyopathy cannot be made with certainty. Radionuclide ventriculography is of value for monitoring ventricular function and is standard practice for patients

receiving cardiotoxic chemotherapy such as doxorubicin. Cessation of chemotherapy when the ejection fraction falls below normal leads to stabilization of the ventricular function and avoids the development of heart failure.

REFERENCES 1. Lipton M J, Boxt L M, Hijazi Z M 2000 Role of the radiologist in cardiac diagnostic imaging. Am J Roentgenol 175: 1495–1506 2. Gibbons R J, Araoz P A 2005 The year in cardiac imaging. J Am Coll Cardiol 46: 542–550 3. Fuster V, Fayad Z A, Moreno P R, Poon M, Corti R, Badimon J J 2005 Atherothrombosis and high-risk plaque part II: Approaches by noninvasive computed tomographic/magnetic resonance imaging. J Am Coll Cardiol 46: 1209–1218 4. Hartnell G G, Bradley F M 1993 Correct positioning for cardiac angiography—insights from MRI. Cathet Cardiovasc Diagn 30: 101–103 5. Nosir Y F M, Salustri A, Kasprzak J D, Breburda C S, Ten Cate F J, Roelandt J R T C 1998 Left ventricular ejection fraction in patients with normal and distorted left ventricular shape by three-dimensional echocardiographic methods: a comparison with radionuclide angiography. J Am Soc Echocardiogr 11: 620–630 6. De Maria A N 1997 A new generation of ultrasound contrast agents for echocardiography. Clin Cardiol 20 (suppl I): 1–48 7. Duvall W L, Croft L B, Goldman M E 2003 Can hand-carried ultrasound devices be extended for use by the noncardiology medical community? Echocardiography 20: 471–476 8. Araoz P A 2003 Computed tomography for functional evaluation of the heart. Semin Roentgenol 38: 303–308 9. Boxt L M, Lipton M J, Kwong R Y, Rybicki F, Clouse M E 2003 Computed tomography for assessment of cardiac chambers, valves, myocardium and pericardium. Cardiol Clin 21: 561–585 10. Hofmann L K, Becker C R, Flohr T, Schoepf U J 2003 Multidetector-row CT of the heart. Semin Roentgenol 38: 135–145 11. Chan F P 2003 Cardiac multidetector-row computed tomography: principles and applications. Semin Roentgenol 38: 294–302 12. Wintersperger B J, Nikolaou K 2005 Basics of cardiac MDCT: techniques and contrast application. Eur Radiol 15 (suppl 2): B2–9 13. Mahnken A H, Wildberger J E, Koos R, Gunther R W 2005 Multislice spiral computed tomography of the heart: technique, current applications, and perspective. Cardiovasc Interven Radiol 28: 388–399 14. Lawler L P, Pannu H K, Fishman E K 2005 MDCT evaluation of the coronary arteries, 2004: How we do it—data acquisition, postprocessing, display, and interpretation. Am J Roentgenol 184: 1402–1412 15. Nakanishi T, Kayashima Y, Inoue R, Sumii K, Gomyo Y 2005 Pitfalls in 16-detector row CT of the coronary arteries. RadioGraphics 25: 425–440 16. Broderick L S, Brooks G N, Kuhlman J E 2005 Anatomic pitfalls of the heart and pericardium. RadioGraphics 25: 441–453 17. Koyama Y, Matsuoka H, Mochizuki T et al 2005 Assessment of reperfused acute myocardial infarction with two-phase contrastenhanced helical CT: prediction of left ventricular function and wall thickness. Radiology 235: 804–811 18. Raff G L, Gallagher M J, O'Neill W W 2005 Diagnostic accuracy of noninvasive coronary angiography using 64-slice spiral computed tomography. J Am Coll Cardiol 46: 552–557 19. Heuschmid M, Kuettner A, Schroedr S et al 2005 ECG-gated 16-MDCT of the coronary arteries: Assessment of image quality and accuracy in detecting stenoses. Am J Roentgenol 184: 1413–1419 20. Constantine G, Shan K, Flamm S D, Sivananthan M U 2004 Role of MRI in clinical cardiology. Lancet 363: 2162–2171 21. Boxt L M 1999 Cardiac MR imaging: A guide for the beginner. RadioGraphics 19: 1009–1025 22. Schneider G, Ahlhelm F, Seidel R et al 2003 Contrast-enhanced cardiovascular magnetic resonance imaging. Top Magn Reson Imaging 14: 386–402

CHAPTER 22

23. Botnar R M, Stuber M, Kissinger K V, Manning W J 2000 Free-breathing 3D coronary MRA: the impact of ‘isotropic’ image resolution. J Magn Reson Imaging 11: 389–393 24. Hundley W G, Hamilton C A, Rerkpattanapipat P 2003 Magnetic resonance imaging assessment of cardiac function. Curr Cardiol Rep 5: 69–74 25. Paetsch I, Jahnke C, Wahl A et al 2004 Comparison of dobutamine stress magnetic resonance, adenosine stress magnetic resonance, and adenosine stress magnetic resonance perfusion. Circulation 110: 835–842 26. Edelman R R 2004 Contrast-enhanced MR imaging of the heart: overview of the literature. Radiology 232: 653–668 27. Ibrahim T, Nekolla S G, Hornke M et al 2005 Quantitative measurement of infarct size by contrast-enhanced magnetic resonance imaging early after acute myocardial infarction: comparison with single photin emission tomography using Tc99M-sestaMIBI. J Am Coll Cardiol 45: 544–552 28. Langerak S E, Vliegen H W, de Roos A et al 2002 Detection of vein graft disease using high-resolution magnetic resonance angiography. Circulation 105: 328–333 29. Larson A C, Simonetti O P, Li D 2002 Coronary MRA with 3D undersampled projection reconstruction TrueFISP. Magn Reson Med 48: 594–601 30. Fuster V, Kim R J 2005 Frontiers in cardiovascular magnetic resonance. Circulation 112: 135–144 31. Glockner J F, Johnston D L, McGee K P 2003 Evaluation of cardiac valvular disease with MR imaging: qualitative and quantitative techniques. RadioGraphics 23: e9 32. Sparrow P J, Kurian J B, Jones T R, Sivananthan M U 2005 MR imaging of cardiac tumors. RadioGraphics 25: 1255–1276

•

CARDIAC ANATOMY AND IMAGING TECHNIQUES

33. Roditi G H, Cohen M C, Hartnell G G 2000 MRI changes in myocarditis—evaluation with spin echo, cine MR angiography and contrast enhanced spin echo imaging. Clin Radiol 55: 752–758 34. Shellock F G, Crues J V 2004 MR procedures: biologic effects, safety, and patient care. RadioGraphics 232: 635–652 35. Chen E Q, MacIntyre W J, Go R T et al 1997 Myocardial viability studies using fluorine-18-FDG SPECT: a comparison with fluorine-18-FDG PET. J Nucl Med 38: 582–586 36. Shagam J Y 2001 Positron emission tomography. Radiol Technol 72: 551–569 37. Gunning M G, Anagnostopoulos C, Knight C J et al 1998 Identification of hibernating myocardium. A comparison of Tl-201, Tc-99m tetrofosmin and dobutamine cine magnetic resonance imaging. Circulation 98: 1869–1874 38. Kontos M C, Jesse R L, Schmidt K L, Ornato J P, Tatum J L 1997 Value of acute rest sestamibi perfusion imaging for evaluation of patients admitted to the emergency department with chest pain. J Am Coll Cardiol 30: 976–982 39. Marie P Y, Danchin N, Durand J F et al 1995 Long-term prediction of major ischemic events by exercise thallium-201 single-photon emission computed tomography. Incremental prognostic value compared with clinical, exercise testing, catheterization and radionuclide angiographic data. J Am Coll Cardiol 26: 879–886 40. Sharir T, Germano G, Kavanagh P B et al 1999 Incremental prognostic value of post-stress left ventricular ejection fraction and volume by gated myocardial perfusion single photon emission computed tomography. Circulation 100: 1035–1042

447

CHAPTER

Congenital Heart Disease: General Principles and Imaging

23

Ronald G. Grainger, Andrew M. Taylor and Christine Reek General principles • Embryology • Diagnosis of congenital heart disease • Conventional radiology Imaging of congenital heart disease • Magnetic resonance imaging

• • • •

Computed tomography Sequential segmental analysis Acyanotic heart disease Cyanotic heart disease

GENERAL PRINCIPLES Ronald G. Grainger

EMBRYOLOGY The fetal heart develops from a series of very complex and very rapid changes that occur between the second (total fetal length 2 mm) and eighth week of intrauterine life. It is during the third to fifth weeks of intrauterine life (when the forelimbs are developing) that the cardiac structures develop most actively and are therefore most susceptible to adverse external influence (e.g. rubella virus or drugs such as thalidomide), resulting in congenital heart disease (CHD). By 8 weeks, the heart has already assumed its definitive form. Table 23.1 lists the embryological precursors and the more frequent important abnormalities that may result from aberrations in their development during the first 3–5 weeks of intrauterine life. The right and left atria are separated from, but connected to, the corresponding right and left ventricle by the tricuspid (right) and mitral (left) atrioventricular valves which develop from the atrioventricular endocardial cushions arising from the margins of the primitive atrioventricular groove. Between the second and seventh week of intrauterine life, the primitive cardiac tube grows in length much more rapidly than does the fetal trunk. The caudal and cephalad ends of the cardiac tube are therefore relatively fixed, and the rapidly elongating cardiac tube is compelled to bend into a loop, to twist and to rotate.

The loop or curve so formed is almost always convex to the right of the fetus—the D or dextro loop as viewed from the front. The axial rotation or twist of the looped cardiac tube is almost always clockwise viewed from the caudal end of the fetus. This axial rotation results in the definitive orientation of cardiac chambers after birth with the right atrium anterior and on the right, left atrium posterior, right ventricle anterior and left ventricle posterior and on the left. The fetal right atrium receives both superior vena cava (SVC) (desaturated blood from the head and myocardium) and inferior vena cava (IVC) blood (partially oxygenated from the fetal placenta and the veins from the lower part of the body). There is almost no blood flow to the fetal airless lungs because of the very high pulmonary vascular resistance.The only blood supply to the fetal left heart and therefore the systemic arteries is via (a) an atrial right atrium-to-left atrium shunt through the patent foramen ovale; and (b) a fetal pulmonary artery shunt to the descending thoracic aorta through the large patent ductus arteriosus (PDA). Normal fetal development is only sustained by the continued patency of these two essential right-to-left shunts throughout fetal life (Fig. 23.1).

Fetal circulation and circulatory changes at birth The only oxygen supply of the fetus is obtained from the oxygenated maternal blood in the placenta supplied by branches from the maternal abdominal aorta. The oxygenated maternal

450

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 23.1 MORE COMMON ANOMALIES OF CARDIAC DEVELOPMENT Embryological precursor

Abnormality

Sinus venosus and tributaries

Anomalous systemic venous drainage*

Interartrial septum and atria

Persistent foramen ovale* Ostium secundum defect* Sinus venous defect Ostium primum defect

Endocardial cushions

Ostium primum defect Endocardial atrioventricular defect Tricuspid atresia/stenosis Ebstein's anomaly Cor triatrium

Interventricular septum and ventricles

Ventricular septal defect (membranous, muscular) Common ventricle Hypoplastic left ventricle† Uhl's dysplastic right ventricle

Bulbus cordis and ventricular

Pulmonary stenosis (valve/infundibular)*

outflow

Aortic stenosis (valve/subaortic)*

anoxic vasoconstriction of the fetal pulmonary arterioles. This major fall in fetal pulmonary arteriolar pressure permits a major increase in pulmonary arterial and venous flow which distends the left atrium and closes the flap valve of the foramen ovale. Thus within a few days of postnatal life, the rightto-left shunting is abolished through both the foramen ovale and PDA. The success of the fetal circulation (Fig. 23.1) in oxygenating and nourishing the rapidly developing fetus depends on: 1 The crossing of two streams of blood (partially oxygenated blood from the fetal IVC and desaturated blood from the fetal SVC and coronary veins) which remain reasonably distinct with a minimum of mixing. 2 The open fetal foramen ovale permitting a right atriumto-left atrium shunt of partially oxygenated blood returning to the fetal heart from the placenta via the fetal IVC. 3 The patency of the fetal arterial duct (PDA) permitting a major right-to-left shunt (pulmonary artery to descending aorta) of poorly oxygenated blood from the SVC via the right ventricle to perfuse the lower half of the fetus, including the placenta where the fetal blood becomes fully oxygenated.

Tetralogy of Fallot*† Ventriculo-arterial discordance (uncorrected and corrected transposition)*† Truncus arteriosus

Common arterial trunk, persistent truncus ateriosus*† Aortopulmonary window Uncorrected transposition (discordance ventricles and great arteries)*† Congenitally corrected transposition of great arteries Pulmonary arterial trunk atresia

Branchial (aortic) arches

Double aortic arch: aortic rings Right aortic arch either isolated or with congenital heart disease* Aberrant right subclavian artery* Interruption or absence of right or left pulmonary artery Patent arterial duct (ductus arteriosus)* Supravalvular aortic stenosis Coarctation of the aorta*

*

†

Most frequent anomalies; usually cyanotic.

blood transmits oxygen through the capillaries of the placenta to the fetal desaturated blood from the fetal placental arteries. At the very moment of birth, the placenta is shed and this source of oxygen supply to the fetus is permanently removed and is replaced by the atmospheric air inhaled into the infant's lungs. This momentous development necessitates major essential changes within the fetal heart, lungs, aorta and pulmonary artery and arterial duct (PDA). Immediately after birth, there is a major increase in the infant's pulmonary blood flow as the infant's lungs become fully expanded with inhaled air, which greatly reduces the

Figure 23.1 Normal fetal circulation. Oxygenated blood (black) is carried by the umbilical vein (UV) through the ductus venosus (DV) and inferior vena cava into the right atrium (RA) within which it is directed to pass through the flap valve foramen ovale (O) into the left atrium (LA) and thence into the left ventricle and aorta and systemic arteries. Mixed venous (superior and inferior venae cavae) blood (light grey) is passed by the venae cavae into the RA, thence the RV into pulmonary artery (PA). The peripheral pulmonary arteries (pp) are very small in the fetus due to the nonfunction of the fetal lungs and high resistance of the peripheral arterial circulation. Most of the output of the right ventricle therefore passes through the large valveless patent ductus arteriosus (D) into the descending aorta to supply the lower parts of the fetus and the fetal umbilical arteries which supply the fetal placental plexus with poorly oxygenated blood for oxygenation in the placenta. The dark grey colour indicates blood having an oxygen content intermediate between mixed venous blood (light grey) and fully oxygenated blood (black).

CHAPTER 23

DIAGNOSIS OF CONGENITAL HEART DISEASE Types of abnormality The great majority of anomalies of cardiac development can be categorized as: 1 An abnormal communication [e.g. atrial septal defect (ASD), ventricular septal defect (VSD), PDA] between the right and left sides of the heart or between their main vessels. 2 An obstruction usually involving or near a heart valve in the pathway of blood flow (e.g. tricuspid atresia, pulmonary, mitral and aortic valve stenosis, coarctation of the aorta). 3 Common (combined right and left) chambers, i.e. common atrium, ventricle, or great artery (common arterial trunk) receiving both oxygenated and desaturated blood. 4 Abnormal connections (discordance) between the cardiac chambers and great arteries (e.g. right ventricle to transposed aorta, left ventricle to transposed pulmonary artery)— uncorrected transposition of great arteries (UTGA). 5 The discordance may be congenitally uncorrected (UTGA) or congenitally corrected (CCTGA) by atrioventricular inversion whereby the right atrium connects to the left ventricle (which gives rise to the discordant pulmonary artery), and the left atrium connects to the right ventricle (which gives rise to the discordant aorta). 6 Abnormal situs (position) of the heart chambers (see below). These cardiac abnormalities may exist singly or in combination, enabling a useful but not invariable classification into (A) cyanotic; and (B) non-cyanotic congenital heart disease.

Cyanotic congenital heart disease For clinically detectable central cyanosis, there must be at least 5 g of reduced haemoglobin per 100 ml of aortic blood. The entry of desaturated blood into the aorta (to cause central cyanosis) may occur in three ways: 1 Direct right-to-left shunt of desaturated venous blood into an otherwise normal systemic circuit. This usually entails a septal defect plus a more distal right heart obstruction that increases the pressure of the proximal right-sided chamber sufficiently to make the shunt go from right to left (e.g. tetralogy of Fallot) (Table 23.2). 2 Transposition (discordance) of the great arteries (UTGA). The infant right ventricle pumps systemic desaturated venous Table 23.2

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

return directly into the aorta and thence to the systemic arteries of the body, causing intense central cyanosis. The infant's left ventricle pumps fully oxygenated pulmonary venous return straight back via the transposed pulmonary artery to the lungs (a useless exercise). For the infant to be viable, there must be mixing of the right and left circulations. 3 Common mixing situations account for most of the other causes of central cyanosis. At some point in the circulation, the systemic venous (desaturated) and pulmonary venous (oxygenated) bloods return to the heart and are obliged to mix. Examples are common atria, common ventricles and persistent truncus arteriosus. Central cyanosis may be present within a few hours after birth. This indicates a very severe abnormality, such as UTGA, in which the aorta arises from the right ventricle (ventriculoarterial discordance) and thus conveys desaturated blood to the systemic circulation. Such an anomaly is not compatible with life unless there is shunting (ASD, VSD and PDA) which enables mixing of oxygenated and desaturated blood. Alternatively, central cyanosis may develop within the next few months or years. Infants with tetralogy of Fallot usually develop this delayed cyanosis, due to an increasing degree of obstruction of the outflow of the right ventricle by muscle hypertrophy and fibro-elastosis.

CONVENTIONAL RADIOLOGY The single most important radiograph is the erect frontal chest radiograph taken in full inspiration (total lung capacity) with a radiographic exposure showing the peripheral pulmonary vessels, the right and left bronchi (indicating the atrial situs, see below), and the location of the aortic arch. A high kV filtered frontal chest exposure may be necessary to demonstrate the important mediastinal anatomy. The type of CHD abnormality and its natural history and haemodynamic consequences may be evident on a single frontal chest radiographic exposure, such as ASD, pulmonary valve stenosis (PVS), coarctation of the aorta, and Eisenmenger reaction (see Fig. 23.4). However, it is often impossible to make a specific diagnosis from the chest radiograph, particularly in the supine infant. It may only be possible to suggest the broad classification, such as a left-to-right shunt, without being able to localize the site

SHUNTS: ACYANOTIC LEFT-TO-RIGHT SHUNT AND CYANOTIC RIGHT-TO-LEFT SHUNT Acyanotic

Abnormal communication

Left-to-right shunt if no obstruction

Cyanotic Abnormal communication + distal right heart obstruction

Shunt reversal with distal obstruction. Right-to-left shunt

ASD

LA → RA

TS/PS/ER

RA → LA

VSD

LV → RV

PS/ER

RV → LV

PDA

Ao → PA

ER

PA → Ao

APW

Ao → PA

ER

PA → Ao

Ao = aorta, APW = aortopulmonary window, ASD = atrial septal defect, LA = left atrium, LV = left ventricle, PA = pulmonary artery, PDA = patent ductus arteriosus, PS = pulmonary stenosis or atresia, RA = right atrium, RV = right ventricle, TS = tricuspid stenosis or atresia.

451

452

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

of the abnormal communication. Conventional chest images become more reliable as the child grows. Clinical, ECG, echocardiographic, angiocardiographic, catheter pressure and oximetry data, and cross-sectional imaging [computed tomography (CT) and magnetic resonance imaging (MRI)] are of great and usually decisive importance in establishing the diagnosis. Major recent developments in the noninterventional techniques very often obviate the need for interventional catheter techniques (see below).

Diagnostic features The pulmonary vasculature (normal, plethoric, oligaemic, congested) The radiological assessment of the pulmonary circulation is probably the most important observation on the plain chest radiograph. Radiologically normal pulmonary vascularity is present in CHD if the patient is not in heart failure, if no large shunt is present, and if there is no extreme pulmonary stenosis. The pulmonary vasculature may, however, look normal on the conventional radiograph even in the presence of substantial CHD. Increased pulmonary perfusion (pulmonary plethora) is recognized by enlarged central and peripheral pulmonary arteries and veins in all zones (Fig. 23.2), as in ASD, VSD and PDA with large left-to-right shunts (Table 23.3).

Table 23.3

INCREASED PULMONARY PERFUSION (PLETHORA)

Level of shunt

Anomaly

Atrium

Ostium primum defect* Ostium secundum defect* Sinus venosus defect Anomalous pulmonary veins†

Atrioventricular valves

Endocardial cushion defects Ostium primum defect* Muscular ventricular septal defect (VSD)*

Ventricles

Membranous VSD* Bulvar VSD* Double outflow ventricle Single ventricle

Aorta

Patent arterial duct* Aortopulmonary window Common arterial trunk (persistent truncus arteriosus)† Coronary artery-to-right heart fistula Uncorrected transposition of the great arteries with atrial or venous septal defect

* Most common causes of plethora without cyanosis; †most common causes of plethora with cyanosis.

Decreased pulmonary perfusion (oligaemia) (Fig. 23.3) may be difficult to assess even on good quality radiographic images, and this becomes impossible if the image is overexposed. Isolated obstruction of the right heart, such as PVS, may not cause radiologically demonstrable pulmonary oligaemia unless the obstruction is very severe with a considerable proximal right-to-left shunt (Table 23.4). Pulmonary venous congestion and oedema is the result of the failure of the left heart to clear all the blood delivered to it. The underlying cause may be pump (left ventricular) failure, e.g. fibro-elastosis of the left ventricle, or a severe obstructive lesion anywhere from the pulmonary veins to the aorta, e.g. severe mitral or aortic obstruction (Table 23.5).

Figure 23.2 Pulmonary plethora in atrial septal defect. Large heart, large central and peripheral pulmonary arteries. Note the recruitment of the upper zone vessels into the active pulmonary circulation.1 There is no evidence of pulmonary congestion or oedema. (Reprinted from Clin Radiol, 21, Grainger R G, Transposition of the great arteries and of the pulmonary veins, including an account of cardiac embryology and chamber indentification, 335–354, 1970, with permission from The Royal College of Radiologists©.)

Figure 23.3 Pulmonary oligaemia. Very sparse small calibre vessels in all zones and small hila (tetralogy of Fallot).

CHAPTER 23

Table 23.4

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

DECREASED PULMONARY PERFUSION (ANOMALY) Anomaly*

Level of abnormality Tricuspid valve

Tricuspid atresia Tricupid stenosis Ebstein's anomaly

Right ventricular outflow

Pulmonary infundibular stenosis (severe) Pulmonary valvar stenosis (severe) Tetralogy of Fallot Uhl's disease (right ventricular hypoplasia)

Pulmonary arterial

Pulmonary artery or trunk atresia Right or left pulmonary artery interruption Peripheral pulmonary artery interruption or stenosis Common arterial trunk (persistent truncus) (Type IV) Transposition (ventriculo-arterial discordance) with pulmonary valve stenosis Eisenmenger reaction (lung periphery only)

*

Most of these abnormalities may cause central cyanosis if the foramen ovale is patent or if proximal septal defect is present.

The diagnostic radiological features of pulmonary venous congestion and oedema in CHD are essentially the same as in acquired heart disease (see Ch. 26) with both alveolar and interstitial oedema and thickened, oedematous pleural fissures. Severe obstructive pulmonary arterial hypertension (Eisenmenger reaction or syndrome) is due to a greatly increased peripheral pulmonary arteriolar resistance, which is a response to a left-to-right shunt. The central pulmonary arteries enlarge (Fig. 23.4) and the peripheral pulmonary arteries become smaller than normal. The increase in size of the central pulmonary arteries and possible calcification of the wall of the central pulmonary arteries (and patent arterial duct) are more marked when the Eisenmenger reaction

Table 23.5 PULMONARY OEDEMA AND VENOUS CONGESTION IN CONGENITAL HEART DISEASE Level of anomaly Pulmonary veins

Anomaly Obstruction by septa Total anomalous pulmonary venous drainage obstructed (especially infradiaphragmatic Type 3)

Left atrium

Cor triatrium Mitral valve stenosis or atresia Mitral valve incompetence (including atrioventricular cushion defect)

Left ventricle

Hypoplastic left ventricle Fibro-elastosis of left ventricle Cardiomyopathy

Aorta

Aberrant left coronary artery from pulmonary artery Aortic, supravalvar or valve stenosis or atresia Coarctation

Figure 23.4 Obstructive pulmonary arterial hypertension (Eisenmenger syndrome) complicating a large atrial septal defect with previous major left-to-right shunt. Very large central pulmonary arteries with small peripheral vessels. Linear calcification of the central pulmonary arteries. Moderate cardiac enlargement.

complicates a previous long-standing severe left-to-right shunt (as in ASD) compared to Eisenmenger VSD. Increased bronchial circulation1 An increased bronchial arterial circulation may be recognizable in cyanotic CHD by a nodular lung pattern in the central third of the lung parenchyma, with many small rounded opacities representing many enlarged bronchial arteries seen end-on. Increased bronchial artery circulation occurs in cyanotic CHD as an adaptive mechanism1 whereby the enlarged bronchial arteries containing partially desaturated blood can gain oxygen from the alveolar air and transmit it usefully to the pulmonary veins. The usual cause in CHD is severe tetralogy of Fallot.

Size of the heart Mechanical outflow obstruction before heart failure develops causes muscular hypertrophy of the obstructed ventricle and not dilatation of its cavity. The heart enlarges in two different haemodynamic situations: (a) diastolic volume overload (increased preload); and (b) pump failure, which may be intrinsic or secondary to increased afterload (distal obstruction). Diastolic volume overload indicates that the heart chamber(s) must accommodate an excessive volume of blood during its filling phase whenever there is a major left shunt in order to maintain increased cardiac output, or when there is severe valvular regurgitation. Pump failure may affect either the right or left ventricle, causing considerable enlargement of that chamber and a marked

453

454

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

reduction in its ejection fraction. In CHD this may result from an inherent pump weakness such as fibro-elastosis, or from failure of the ventricle to overcome a severe mechanical challenge such as very severe obstruction of its outflow.

Shape of the heart This is influenced by the size, shape and position of its different chambers and great vessels. Classic descriptions, e.g. coeur en sabot in tetralogy of Fallot, figure of 8 in total superior anomalous pulmonary venous drainage (Type 1), are now much less frequently seen as early diagnosis and corrective surgery prevent the development of major cardiac changes and the adoption of the characteristic shapes.

Position, size and shape of the ascending aorta and its arch The side of the aortic arch is almost always clearly visible on the frontal chest radiograph of an adolescent or adult, but may be much more difficult to detect in the infant and young child. The side of the aortic arch can usually be inferred to be on that side of the trachea where there is more mediastinal soft tissue at the level of the third and fourth dorsal vertebrae: if there is any doubt, confirmation will be obtained by cross-sectional imaging—MRI or CT, or a barium swallow which will demonstrate a concave indentation and displacement of the oesophagus away from the side of the aortic arch. In right-sided aortic arch with associated CHD, there is usually mirror image branching of the aortic arch, i.e. the first branch is the left innominate (which divides into the left common carotid and left subclavian arteries), followed by the right carotid and right subclavian arteries. No major arterial branch usually passes behind the oesophagus, which is therefore rarely indented from behind as it often is by the left subclavian artery in isolated right aortic arch without associated cardiac anomaly. The position of the ascending aorta is of considerable importance. In a normally positioned heart (laevocardia) in situs solitus, the ascending aorta usually forms a gently convex right border to the superior mediastinum. In uncorrected transposition (UCTA)2, the ascending aorta is usually only very slightly to the right of the midline in front of the midline pulmonary arterial trunk, so producing a vascular pedicle which is narrow in the frontal view and wide on the lateral radiograph. In congenitally corrected transposition (CCTGA), the ascending aorta may produce a long convexity on the left upper mediastinal contour—a very significant diagnostic feature. A dilated ascending ‘aorta’, rising high in the mediastinum to form a high prominent ‘aortic’ arch, is seen typically in persistent arterial truncus (up to 50% have a right arch) or tetralogy of Fallot (20–30%). An abnormally shaped aortic arch is seen in many cases of coarctation and high cervical arch (pseudocoarctation). The appearances of the aortic arch in these two conditions may be identical (deformed high aortic knuckle, sharp forward kink of the upper descending aorta, an indented aortic arch left border, figure of 3 indentation deformity of the left border of the oesophagus). The distinguishing radiological feature between these two conditions is rib notching in a haemodynamically significant coarctation, denoting a compensating collateral intercostal circulation

conveying blood to the descending aorta.There is no rib notching in persistent cervical arch (pseudocoarctation) as there is usually minimal or no stenosis of the aortic lumen.

Situs (position) of the viscera, bronchi (and by inference of the atria) The normal position of the viscera (heart and stomach on the left, liver on the right, short right bronchus, longer left bronchus) is known as situs solitus. Cardiac situs is discussed further below.

Presence of associated features, e.g. skeletal changes The most important skeletal feature seen on the conventional chest radiograph of a patient with CHD is rib notching of the lower margin of the third to eighth ribs in aortic coarctation due to the enlarged, tortuous intercostal arteries supplying blood to (instead of conducting blood from) the descending aorta. (The notching is not usually seen in children younger than 5 years.) It is usually bilaterally symmetrical but may be exclusively on the right (aortic coarctation involving the left subclavian artery), or only on the left (aberrant right subclavian artery arising below the coarctation). Several genetic, metabolic and skeletal disorders are associated with congenital malformations of the heart. CHD is more frequent when the upper limbs are deformed, probably because the forelimbs develop at the same time as the heart in the first 3–5 weeks of intrauterine life, i.e. before the lower limbs. Trisomy chromosome 21 (Down’s syndrome, mongolism) is complicated in about 25% of cases by heart malformation usually involving atrioventricular endocardial cushion defects. Turner's and Noonan’s syndromes are often associated with aortic coarctation and/or pulmonary artery stenosis. Holt– Oram syndrome (absent or hypoplastic forearms and thumbs) may be associated with ASD and other CHD. Ellis van Creveld syndrome (chondro-ectodermal dysplasia) may be complicated by a common atrial chamber. Thalidomide toxicity produces phocomelic changes (small undeveloped limbs), more extreme but similar to Holt–Oram syndrome, and 17% of affected patients are said to have a cardiac malformation. Post-surgical stigmata such as surgically deformed ribs, unilateral rib notching, surgical sutures and valve and conduit prostheses are most valuable evidence and must always be carefully looked for on the plain radiographs.

Chamber identification The geographical position of the four individual heart chambers and of the great vessels relative to each other, their spatial orientation within the thorax, and their interconnections may all be abnormal and these criteria cannot therefore be used for chamber identification. It is essential therefore positively to identify the different cardiac chambers by their intrinsic anatomical or morphological characteristics2, regardless of their intrinsic connections or their

CHAPTER 23

spatial orientation (see section on sequential segmental analysis below). For most of the 20th century, catheter angiocardiography was the undisputed gold standard for identifying the detailed

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

anatomy of the heart chambers and the detailed diagnosis of CHD2. That position has been recently very successfully displaced by noninterventional echocardiography and crosssectional imaging—CT and MRI (see below).

IMAGING OF CONGENITAL HEART DISEASE Andrew M. Taylor and Christine Reek CHD has an incidence of 6–8 per 1000 at birth3,4. The diagnosis of CHD has increased because of improvements in diagnosis and treatment, leading to more patients surviving into adulthood. There is an increasing number of children with acquired heart disease, in particular related to anthracylcine cardiotoxicity following treatment for oncological disease in early childhood. Management of patients with CHD relies very heavily on imaging. Echocardiography is the first-line investigation6. It can be used to assess accurately anatomy and function, is not very expensive, is portable, and is easy to use in experienced hands. However, echocardiography may be limited by poor acoustic windows, can be poor at imaging the vessels beyond the heart, and is very user dependent.These problems are exacerbated in patients with CHD, as it is often the right ventricle and pulmonary arteries that require assessing (both of which are difficult to assess at echocardiography), and patients may have undergone multiple operations, which can further reduce acoustic window access. Traditionally, more detailed anatomical and functional assessment has been acquired with X-ray cardiac catheterization. Cardiac catheterization gives good definition of vascular anatomy and enables assessment of haemodynamics (vascular stenosis, quantification of cardiac shunting and pulmonary vascular resistance), but is associated with risks due to the invasive nature of the procedure and the exposure to ionizing radiation. In addition, X-ray fluoroscopy provides a projection image and has only limited three-dimensional (3D) capabilities. Radionuclide imaging has been used to quantify left-toright shunts in order to avoid cardiac catheterization and to assess the relative pulmonary perfusion, but this also involves significant exposure to ionizing radiation7. More recently, cross-sectional cardiovascular imaging (MRI8 and CT9) has become a very important tool for diagnosis and follow-up of patients with CHD, both in adults [grown-up congenital heart (GUCH) disease] and in children. Not only does cross-sectional imaging allow accurate description of cardiac and vascular anatomy in relation to the other structures of the chest8,9, but MRI in particular can provide accurate quantification of cardiac function and vascular flow in vivo. This section describes the common types of CHD and stresses the role of the chest radiograph, MRI and CT. A detailed discussion of the echocardiography is beyond the scope of this chapter.

MAGNETIC RESONANCE IMAGING10 The majority of cardiovascular MR images are acquired using cardiac (ECG) gating during a single breath-hold to

reduce the artefacts associated with cardiac and respiratory motion. For a complex case, MRI is performed over approximately 1 h, though this time can be considerably reduced if a focused question is being addressed. Imaging sequences can be broadly divided into: • ‘Black blood’ spin-echo images, where signal from blood is nulled and thus not seen—for accurate anatomical imaging. • ‘White blood’ gradient-echo or steady-state free precession (SSFP) images, where a positive signal from blood is returned—for anatomical, cine imaging and quantification of ventricular volumes, mass and function. • Phase-contrast imaging, where velocity information is encoded—for quantification of vascular flow. • Contrast-enhanced MR angiography (MRA), where nonECG-gated 3D data are acquired after gadolinium contrast medium has been administered—for thoracic vasculature imaging. All these sequences can be acquired in a single breath-hold, reducing the overall time in the MR machine, and enabling the acquisition of accurate data in the majority of patients. Importantly, ‘white blood’ cine images can be acquired in a continuous short axis stack along the heart, enabling accurate quantification of right and left ventricular function. Imaging should be performed in the presence of a cardiovascular MRI expert in conjunction with an MRI technician to ensure that the appropriate clinical questions are answered. Currently, the authors' own practice is to perform all cardiovascular MRI in children younger than 8 years of age under general anaesthesia as this enables the safe acquisition of accurate data (reduced movement and respiratory artefact). With the development of even faster sequences, breath-holding may become less of a necessity, and MRI data may then be acquired more easily during sedation.

COMPUTED TOMOGRAPHY9 With the development of multidetector CT (MDCT), it is now possible to acquire volumes of CT data that can be reformatted in any imaging plane. MDCT images of the entire thorax can be acquired in 3–10 s, depending on the size of the patient. Using iodinated contrast agents, CT angiography (CTA) can now be rapidly performed, and 3D imaging algorithms applied to create 3D images which considerably aid in the 3D appreciation of complex vascular anatomy. At present, ECG gating for cardiac CT is limited to patients with slow heart rates (< 60 beats min−1), which excludes many subjects

455

456

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

with CHD. Cardiac CT images of intracardiac anatomy are often blurred and of limited value; however, with the advent of 64- and potentially 256-slice CT machines, this will improve in the near future. With the speed of MDCT acquisition, most young children can be imaged either with sedation or unsedated with feed and wrap. Contrast medium injection should optimally be through a large (20 G minimum) cannula with pump injection, but a hand injection may also produce adequate results. Contrast agent bolus timing is important and may be crucial for optimal image quality, vessel delineation and diagnostic confidence. This is best achieved by triggering the data acquisition after visually watching contrast medium arrive in the appropriate vessel. Data are acquired in an axial spiral, with a reconstruction slice width of 0.75–1.5 mm, during a single breath-hold or, if not possible, shallow respiration. Images are then transferred to a workstation for the creation of 3D images. MRI is currently the 3D imaging investigation of choice, as it allows accurate assessment of intracardiac anatomy, function, and flow without a radiation burden or potential complications of iodinated contrast media. However, cardiac CT may be preferred, particularly for vascular imaging in young patients, who would otherwise require a general anaesthetic for adequate MRI, and in those for whom MRI is contraindicated or when additional information may be acquired from CT (e.g. airways and lung).

SEQUENTIAL SEGMENTAL ANALYSIS

3

The nomenclature of complex CHD is based on sequential segmental analysis. This requires description of atrial position (situs), atrioventricular and ventriculo-arterial connections, and other associated lesions3.

Step 1—Atrial situs In the normal heart, the morphological right atrium is located to the right of the morphological left atrium (situs solitus). Abnormalities in the atrial situs are often associated with major cardiac malformations and abnormal abdominal and thoracic anatomy. In situs solitus, the IVC is to the right of the abdominal aorta, with a right-sided liver and a left-sided spleen.The rightsided bronchus is much shorter than the left-sided bronchus. In situs inversus, the mirror image of the normal anatomy is present and is usually easily demonstrated with echocardiography. In situs ambiguus, atrial situs is uncertain. The most common lesions are right or left isomerism in which both atrial chambers have features of the same-sided atrium. In isomeric lesions, there is often an endocardial cushion defect with varying degrees of atrioventricular septal defect (AVSD). The atrial appendages are the best method of determining atrial identification. The right atrial appendage is triangular with a wide base, whilst the left atrial appendage is a tubular structure. The presence of the terminal crest and pectinate muscle in the appendage are more specific internal characteristics of the right atrium and its appendage.

Right isomerism is usually associated with bilateral (short) right bronchi, trilobed lungs and bilateral right atrial appendages, asplenia and a midline liver. Left isomerism is usually associated with bilateral left (long) bronchi, bilobed lungs, left atrial appendages, polysplenia and IVC interruption. All of these abnormalities can be determined with MRI, particularly 3D balanced-SSFP techniques. Both right- and left-sided isomerism are associated with gut malrotation (Table 23.6).

Step 2—Atrioventricular connection Determination of ventricular morphology allows analysis of atrioventricular and ventriculo-arterial connection or concordance. Irrespective of atrioventricular concordance, the atrioventricular valve is concordant with the ventricle, i.e. the tricuspid valve always connects to the right ventricle and the mitral valve connects to the left ventricle.The septal insertion of the tricuspid valve is nearer the cardiac apex than that of the mitral valve, and allows determination of the ventricular morphology. The right ventricle is more coarsely trabeculated than the left ventricle, with a muscular infundibulum separating the inflow and outflow valves, and there is a mid ventricular ‘moderator band’. The size, thickness and shape of the ventricles, although different in normal subjects, are not good indicators of ventricular origin in CHD as they are dependent on load effects.

Step 3—Ventriculo-arterial connection Description of ventriculo-arterial connections represents the final element of sequential segmental analysis. The aorta and pulmonary arteries are identified by their typical branching and distribution patterns and not by the chambers to which they are connected. Cross-sectional imaging is very useful in

Table 23.6 ASSOCIATIONS WITH ATRIAL ISOMERISM Defect

Right isomerism

Left isomerism

Atrioventricular septal defect

Yes

Yes

Anomalous pulmonary veins

Yes

No

Bilateral superior vena cavae

Yes

Yes

Vena azygos continuation

No

Yes

Double outflow right ventricle

Yes

No

Pulmonary stenosis/atresia

Yes

Yes

Dextrocardia

Cardiac

Yes

Yes

Transposition ventriculo-arterial Yes discordance

No

Complete heart block

No

Yes

Bronchi

Bilateral short right

Bilateral long left

Lungs

Bilateral trilobed lungs

Bilateral bilobed lungs

Liver

Central large

Central

Spleen

Asplenia

Polysplenia

Gut

Malrotation

Malrotation

Noncardiac

CHAPTER 23

determining the arrangement of the great vessels and their ventricular connections.

Step 4—Identification of other abnormalities Other abnormalities to be considered include abnormal venous connections, ASD,VSD, AVSD and valve abnormalities. In CHD, most abnormalities are single lesions (e.g. PVS, coarctation of the aorta) that are easily described. Many patients with CHD have a combination of anomalies (e.g. tetralogy of Fallot and transposition of the great arteries). Almost any combination of abnormalities and connections can occur, and using the sequential segmental analysis method the description of all conceivable combinations and diagnoses is possible although not commonly required.

ACYANOTIC HEART DISEASE Congenital anomalies of the aorta Aortic coarctation Coarctation occurs in 6–8% of live CHD births with a predominance of boys4. There is an area of narrowing in the thoracic aorta in the region of insertion of the arterial duct (aortic isthmus).There is a risk of systemic hypertension, atherosclerosis and end-organ damage, even in patients who have undergone surgical repair. Treatment in infancy is usually surgical excision of the narrowing, though in older subjects balloon angioplasty may be undertaken, and following re-coarctation, aortic stenting can be used. In the neonatal population, echocardiography is used in the initial diagnosis. Echocardiography can also be used in followup, but as children grow this becomes more difficult and imaging with MRI (Fig. 23.5A) and CT is required in later life to establish if there is re-coarctation after repair (3–35% of patients), aneurysmal dilatation, or left ventricular hypertrophy secondary to hypertension. MRI is preferred if there are no contraindications, as this reduces population radiation.

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

Imaging is crucial for management decisions in order to establish the location and degree of stenosis, length of coarctation segment, associated aortic arch involvement (such as tubular hypoplasia), the collateral pathways (internal mammary and posterior mediastinal arteries), relationship to aberrant subclavian artery, post-stenotic dilatation and left ventricular hypertrophy. Three-dimensional contrast-enhanced MRA may display the severity and extent of involvement (Fig. 23.5B). MRI flow mapping can define the severity of the stenosis by measuring velocity jets at the level of coarctation. Assessment of collateral flow can also be performed by measuring flows in the proximal and descending aorta. Reassessment of collateral flow following treatment can be used to assess the success of the treatment. MRI can also be used to assess secondary pathology in patients with coarctation, e.g. aortic root for dilatation secondary to a bicuspid aortic valve—frequency in coarctation of 15%; aortic valve incompetence and stenosis; and ventricular function and left ventricular mass (an indirect indicator of increased left ventricular after-load). Discrete localized stenoses may be amenable to balloon dilatation. MRI is very useful for noninvasive follow-up and directing further treatment if necessary.

Anomalies of the aortic arch Embryologically, the left fourth branchial arch persists and the right fourth arch involutes, except for the proximal segment, which forms the proximal right subclavian artery. Understanding the various potential anomalies that may occur is based on Edward's embryonic hypothetical double aortic arch schema (Fig. 23.6). Atretic aortic segments may persist (including right and left arterial ducts), which may not be visualized on imaging, and which may have the potential to cause complete vascular compressive rings (Table 23.7). In double aortic arch, the right arch is usually larger than the left (which may be atretic or hypoplastic) and may result in the formation of a vascular ring, which may compress the trachea and oesophagus. The arterial duct is usually left sided, but can be right sided or even bilateral and increase the compression. Two further potential vascular rings can occur: (a) the left pulmonary artery sling (the left pulmonary artery arises from the proximal right pulmonary artery and then turns abruptly posteriorly and then to the left, indenting and compressing the trachea in anteriorly and the oesophagus posteriorly in order to reach the left hilum); and (b) a right aortic arch, aberrant left subclavian artery, connected to the fibrosed left arterial duct.

Congenital supravalvular aortic stenosis

Figure 23.5 Aortic coarctation. (A) Oblique sagittal, ‘black-blood’ turbo spin-echo image through a severe, discrete aortic coarctation (arrowhead). (B) 3D, volume-rendered MRA image of another severe aortic coarctation (arrow), with multiple collaterals (asterisk).

This is a rare congenital heart defect that is often associated with Williams' syndrome of infantile hypercalcaemia. The luminal stenosis is due to local thickening of the proximal ascending aortic wall. The main roles of cardiovascular MRI are pre-operative anatomical and functional assessment of the stenosis and postoperative assessment of possible restenosis. Contrast-enhanced MRA or CTA is an accurate and less traumatic method than catheter angiography of assessing the type

457

458

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 23.6 A schematic representation of Edward's hypothetical double aortic arch, showing right and left aortic arches and bilateral arterial ducts. (A) Division of the double arch along the bold line in (B) gives a normal left-sided arch, ringed trachea and striated oesophagus posteriorly. (C) A double arch with the ringed trachea and vertically striated oesophagus within the vascular ring. (Courtesy of Professor Robert H Anderson, Dr Andrew C Cook and Gemma Price, Cardiac Unit, Institute of Child Health.)

and severity of the defect as well as its relationship to the aortic valve.

Septal defects These consist of ‘holes’ or defects in the atrial septum, ventricular septum, or a combination of the two.They may be simple, isolated lesions, or may be associated with more complex CHD.

Atrial septal defects Atrial septal defects are the most common congenital heart defect detected in adults, with an incidence of 941 per million live births4. Ostium secundum defects form about 80% of all ASDs and are located in the fossa ovalis. The ostium primum defect, although previously regarded as an ASD, is actually a partial AVSD (see below). The sinus venosus defect is found at the junction of either of the caval veins and the right atrium. This type of ASD is rare and is often associated with partial anomalous pulmonary venous drainage.

Table 23.7 VASCULAR RINGS Symptoms Usually symptomatic

Anatomy Double aortic arch Right aortic arch with right descending aorta + aberrant left subclavian artery + persistent left arterial duct Left aortic arch with right descending aorta + right arterial ductus Left pulmonary sling

Occasionally symptomatic

Anomalous right innominate artery Anomalous left common carotid artery Right aortic arch with left descending aorta + left ductus

Usually asymptomatic

Left aortic arch + aberrant right subclavian artery Left aortic arch with right descending aorta Right aortic arch with right descending aorta + mirror image branching Right aortic arch with right descending aorta + left aberrant subclavian artery Right aortic arch with right descending aorta + isolation of left subclavian artery

Pathological consequences of ASDs are secondary to atrial left-to-right shunting. This volume overload leads to atrial dilatation, predisposing to tachyarrythmias and biventricular volume overload. The presence of an ASD or patent foramen ovale is also a risk factor for paradoxical thromboembolic stroke. Evaluation of ASDs requires definition of type and location of the defect, quantification of the shunt, detection of any intra-atrial thrombus, assessment of right ventricular function, and visualization of the pulmonary venous anatomy. Management of ASDs may be with transcatheter ASD mechanical closure devices or open surgical closure for larger defects. Trans-thoracic echocardiography has a limited ability to visualize small ostium secundum and sinus venosus defects: detection of pulmonary venous abnormalities is very difficult. Transoesophageal echocardiography has become the main imaging technique used to assess ASDs. However, it cannot be used to quantify the shunt (Qp:Qs) accurately and it may be difficult to delineate pulmonary venous anatomy. Cardiovascular MRI can be used if there remains doubt with regard to the anatomical rim of the defect or the functional significance of ASDs, imaging data that compare well with operative findings.

Atrioventricular septal defects Atrioventricular septal defects and, possibly, migraine are caused by abnormal development of the endocardial cushion, and can be categorized into partial with a defect of the atrial septum alone (previously described as primum ASD) or complete with defects of both the atrial and ventricular septa (Fig. 23.7). AVSDs can further be divided into balanced (relatively equal sized atrioventricular valves and ventricles) or unbalanced, when inflow through the atrioventricular valves is predominantly into one of the ventricles. AVSDs are relatively complex lesions, and although they are listed here as an acyanotic condition, they may be associated with other complex anomalies (e.g. atrial isomerism or double outlet right ventricle).There is also a high incidence of AVSDs in patients with Down's syndrome (trisomy 21).

Ventricular septal defects (VSDs) Ventricular septal defects (VSDs) are the most common congenital heart lesions with an incidence of 3570 per million live

CHAPTER 23

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

Patent arterial duct Although the arterial duct is essential for life in the fetus, if it fails to close after birth, a persistent communication is left between the aorta and pulmonary artery. The consequences will depend on the size of the duct with a large duct resulting in a large left-to-right shunt. A small duct may not be detected for many years but still has a risk of infective endocarditis. Closure may be surgical or by catheter device depending on the anatomy and associated conditions. A large duct left unclosed may result in Eisenmenger syndrome of pulmonary arterial hypertension (PAH) and therapeutic closure is then contraindicated.The duct and the central pulmonary arteries may become calcified and may be visible on the chest radiograph along with evidence of PAH.

Eisenmenger syndrome Figure 23.7 Complete atrioventricular septal defect. Balanced steady-state free precession four-chamber view MRI. Arrowhead = small ventricular septal defect, asterisk = complete absence of the atrial septum, white arrow = left atrioventricular valve, dashed arrow = position of right atrioventricular valve.

births4.VSDs are a heterogeneous group of lesions with shunting at the ventricular level. Assessment in the first instance is with echocardiography. The haemodynamic effects of VSDs are dependent on the shunt volume and may vary from insignificant up to left heart failure and pulmonary vascular disease, and the severity determines treatment. Quantification of leftto-right shunts using velocity-encoded phase-contrast MRI compares well to invasive catheterization. Invasive catheterization may still be used to quantify pulmonary vascular resistance to exclude pulmonary hypertension in some long-standing cases before defect closure. Perimembranous lesions account for 80% of all VSDs and are characterized by the relationship with ventricular inlet or outlet. The majority of perimembranous VSDs will close spontaneously. Large perimembranous VSDs are unlikely to close and may lead to the development of obstructive pulmonary vascular (Eisenmenger) disease. In these patients, the VSD must not be therapeutically closed. Between 5% and 20% of VSDs occur in the muscular interventricular septum. Muscular VSDs are an extremely heterogeneous group and can be difficult to close surgically. Optimum treatment (either operative or transcatheter) depends on accurate demonstration of theVSD anatomy. Muscular defects often have a complex 3D structure with multiple defects—‘Swiss cheese appearance’—which are more accurately represented by MRI than echocardiography. Approximately 50% of patients with a VSD have an associated additional cardiac abnormality. Coarctation of the aorta and aortic stenosis are particularly important and are associated with posterior malalignment of the VSD. Cardiovascular MRI is useful in the assessment of both.

The Eisenmenger syndrome of obstructive PAH may develop in any patient with a left-to-right shunt and transmitting near systemic pressures to the pulmonary arteries. This results in atheromatous hypertensive reaction of the pulmonary arteries, which is progressive and may eventually become irreversible. The syndrome involves a high pulmonary vascular resistance with reversed or bidirectional shunt. This may occur early in life, especially if the shunt is large and either aortopulmonary or interventricular. The Eisenmenger syndrome tends to occur earlier in life in VSD and patent ductus, but after 30–40 years in large shunt ASD. In the latter cases, the central pulmonary arteries become very large and may become calcified (see Fig. 23.4). The prevalence of Eisenmenger syndrome has declined as diagnosis and surgical closure of shunts has greatly improved. However, once PAH is established and irreversible, the balanced physiology should not be disturbed. The remaining surgical option is heart–lung transplantation when end-stage heart and lung disease has developed. Anatomy and flow rates can be assessed by MRI, but investigation of pulmonary vascular resistance and its reversibility still requires cardiac catheterization.

Pulmonary valvular stenosis Pulmonary valvular stenosis is frequently associated with ASD or patent foramen ovale, and a dysplastic pulmonary valve is often seen in patients with Noonan's syndrome. Obstruction to the right ventricular outflow tract is also considered in the section on tetralogy of Fallot and pulmonary atresia below. When severe, pulmonary valvular stenosis can present in the neonatal period. The right ventricle may be small in these patients, and presentation and management are similar to those of pulmonary atresia with intact ventricular septum (see below). In older children with normal cardiac output, pulmonary stenosis is graded according to the gradient across the valve: mild—25–50 mmHg; moderate—50–70 mmHg; and severe > 70 mmHg. Diagnosis is usually based on clinical suspicion and findings at echocardiography. Angiocardiography demonstrates convex doming of the pulmonary valve and post-stenotic dilatation of the main and left pulmonary arteries, but this invasive

459

460

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

procedure is now rarely indicated. Patients with moderate to severe pulmonary stenosis can be treated with balloon valvuloplasty in the first instance (cardiac catheterization or surgery). Cross-sectional imaging, in particular MRI, can be used for the follow-up of these patients.The sequelae of pulmonary incompetence are of particular importance in these patients, and may require surgical or transcatheter (percutaneous pulmonary valve implantation11) intervention at a later date (see tetralogy of Fallot below).

CYANOTIC HEART DISEASE Tetralogy of Fallot Tetralogy of Fallot is the most common cyanotic congenital heart defect with an incidence of approximately 420 per million live births4. It is caused by malalignment of the infundibular septum, which leads to right ventricular outflow (RVOT) obstruction, a subaortic VSD with aortic override and right ventricular hypertrophy (Fig. 23.8). Trans-thoracic echocardiography is the imaging modality of choice for initial diagnosis and assessment. Initial chest radiographic appearances (Fig. 23.9) may include right ventricular heart silhouette, a boot-shaped heart, small hila, pulmonary oligaemia or asymmetry and concave pulmonary artery segment.The aortic arch is right sided in about 30%. The anatomy can be confirmed by conventional angiography (Fig. 23.8) or by noninvasive cross-sectional imaging.

LSCA

RSCA

DAD C Ao

Current management consists of early single stage reconstructive surgery, with closure of the VSD, and relief of the RVOT obstruction, with possible placement of a transannular patch. Staged reconstruction is still required if there is significant hypoplasia of the central pulmonary arteries, with placement of a modified Blalock–Taussig (BT) shunt—a systemic-to-pulmonary anastomosis, usually between the innominate artery and the right pulmonary artery. This shunt is then taken down during subsequent definitive repair. Cross-sectional imaging has a role in delineating pulmonary artery anatomy in the younger child, defining whether reconstructive pulmonary artery surgery is required prior to definitive surgery. However, the main role of imaging in patients with tetralogy of Fallot is the assessment of postoperative complications (Fig. 23.10). Chest radiography at this stage may show evidence of previous BT shunt surgery, asymmetric pulmonary circulation, and calcification of the transannular patch. The most common late postoperative complication is pulmonary regurgitation secondary to transannular patch reconstruction of the RVOT/pulmonary annulus, and is often associated with aneurysmal dilatation of the RVOT. Surgical or transcatheter valve replacement is the current method of managing patients with severe pulmonary regurgitation11 (Fig. 23.10). Accurate quantification of regurgitation and assessment of RVOT anatomy and right ventricular function are particularly important in deciding the type, timing and multiplicity of procedures. Branch pulmonary stenosis may also be present in this patient group and is best imaged by MRA.This can contribute to right ventricular dysfunction, and significant obstructions should be repaired at the same time as valve replacement. The final role of MRI is in evaluating right ventricular function. This is important for timing of invasive therapeutic measures and evaluating the effect of any invasive procedure. It has been shown that, using a combination of MR ventricular volumetry and tricuspid and pulmonary flow maps, precise information about global and diastolic ventricular function can be assessed in patients with repaired tetralogy of Fallot.

LPA

RPA

RVOT

RV

Figure 23.8 Tetralogy of Fallot. Right ventricular angiogram showing long stenosis of the right ventricle outflow tract (RVOT). Immediate opacification of the right aortic arch. Immediate opacification of the proximal right pulmonary artery (RPA): interruption of the proximal left pulmonary artery. The distal left pulmonary artery opacifies later and fainter via collaterals from the aorta (Ao), left subclavian artery (LSCA), and from other collaterals (C).

Figure 23.9 Tetralogy of Fallot. Boot-shaped heart, small hila and pulmonary oligaemia.

CHAPTER 23

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

Figure 23.10 Repaired tetralogy of Fallot. (A) Balanced steady-state free precession four-chamber image showing right ventricular dilation (asterisk). (B) 3D, volume-rendered MRA image of the right ventricular outflow tract (RVOT) and proximal pulmonary arteries—lateral view. Note the large RVOT aneurysm is seen at the site of previous transannular patch repair (dashed arrow), the pulmonary trunk (arrowhead), and the narrowed proximal left pulmonary artery (long complete arrow). (C) Flow plotted against time for the aorta and pulmonary artery demonstrating pulmonary incompetence (regurgitant fraction = 30%); data acquired from phase-contrast velocity maps (not shown).

Pulmonary atresia Pulmonary atresia is defined as a lack of continuity between the RVOT and the central pulmonary arteries with a variable degree of hypoplasia of these structures. Pulmonary atresia can be separated into two groups depending on the presence of a VSD. As the diagnosis and subsequent management of these two groups is different, it is useful to consider them separately.

Pulmonary atresia with a ventricular septal defect This is the more common variant and is considered by some to be a severe form of tetralogy of Fallot, with a subaortic VSD, overriding aorta and often major aortopulmonary collaterals (MAPCAs). Surgical repair aims to establish RVOT to the pulmonary artery continuity with a homograft, repair the VSD and bring the aortopulmonary collaterals into the pulmonary circulation. As with tetralogy of Fallot, the main role of MRI in patients with pulmonary atresia and a VSD is assessment of postoperative complications.The most common long-term complication is homograft failure; usually mixed stenosis and regurgitation leading to right ventricular dysfunction. Conduit stenosis is often secondary to calcification of the nonviable homograft,

and although calcified tissue is difficult to visualize using MRI, it is clearly seen on CT. Other long-term complications are similar to those found in tetralogy of Fallot.

Pulmonary atresia with an intact ventricular septum This is the less common variant of pulmonary atresia, and is associated with a variable degree of right ventricular hypoplasia. The type of surgical repair depends on the size and shape of the right ventricular cavity. The presence of a right ventricular infundibulum allows a biventricular repair. If the right ventricular cavity is small then single ventricular physiology is established and angiography demonstrates large venous sinusoids in the right ventricular wall. MRI assessment is considered in the section below on the single ventricle.

Transposition of the great arteries2 Transposition of the great arteries (TGA) is the second most common cyanotic CHD in the first year of life with an incidence of 315 per million live births4. It is defined as ventriculo-arterial discordance with an anterior aorta arising from the anterior right ventricle, and the pulmonary artery arising from the posterior left ventricle (Fig. 23.11). Forty per cent of patients with

Figure 23.11 Transposition of the great arteries. (A) Balanced steady-state free precession (SSFP) sagittal image; the aorta (Ao) arises anteriorly from the hypertrophied right ventricle (arrowhead); the posterior pulmonary artery (PA) arises from the left ventricle (dashed arrow). LV = left ventricle, RV = right ventricle. (B) Balanced SSFP images of intra-atrial baffles in a patient who has undergone the atrial switch Senning operation—pulmonary venous blood into the right atrium (asterisk), systemic venous blood into the left atrium, superior pathway of systemic venous blood (dashed arrow), inferior pathway (arrow). (C) 3D volume-rendered MRA of the aorta (Ao) after the arterial switch operation. Anterior pulmonary arteries are marked: asterisk = pulmonary trunk, arrow = left pulmonary artery.

461

462

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

TGA have a VSD and 30% of these patients have subpulmonary stenosis. Surgical therapy for this condition was revolutionized in about the 1960s with the introduction of the Senning and Mustard procedure, in which blood is diverted by an intra-atrial baffle from the right atrium to the left ventricle, and from the left atrium to the right ventricle. Both procedures produce a physiologically normal, but an anatomically very abnormal circulation (systemic venous return to the left atrium, then to the left ventricle, and then to the transposed discordant pulmonary artery; pulmonary venous return diverted to the right atrium, then to the right ventricle and then to the transposed aorta). In 1975, the first arterial switch operation was performed. In this operation, the aorta and main pulmonary artery are transected just above the origin of the coronary arteries, switched and re-anastomosed to the correct ventricle. The coronary arteries and a small portion of aortic sinus are excised when the aorta is transected, and re-implanted once the aorta is connected to the left ventricular outflow track. This results in both a physiological and anatomically normal circulation, and for this reason, the arterial switch operation has become the procedure of choice for TGA. The arterial switch operation is performed in the first few days of life, and currently trans-thoracic echocardiography is the imaging modality of choice for pre-operative diagnosis and assessment. The role of cross-sectional imaging is mainly in the diagnosis of postoperative complications, particularly those that develop as the child grows older. The main complications of the arterial switch operation are RVOT or branch pulmonary artery obstruction. Due to the unusual position of the pulmonary arteries immediately behind the sternum, trans-thoracic echocardiography is poor at detecting these lesions. Cardiovascular MRI is ideal for imaging the RVOT and branch pulmonary arteries in this group of patients (Fig. 23.11). Contrast-enhanced MRA can be used to visualize the relationship between the pulmonary arteries and the aorta, while spin-echo sequences are used to assess accurately the degree of stenosis. A less common complication of the arterial switch operation is coronary stenosis secondary to the re-implantation procedure. Although the majority of coronary complications cause early postoperative mortality, a subset of patients suffers from late coronary events. In this group, coronary catheter angiography probably represents the modality of choice for investigation. However, coronary MRA and MDCT angiography are useful noninvasive methods for investigating the coronary arteries, particularly the proximal segments. Although intra-atrial baffle repair has been superseded by the arterial switch operation, there is a sizeable population that has undergone either a Senning or Mustard operation. The most common complications of intra-atrial repair are baffle obstruction or leak, arrhythmias and right ventricular dysfunction. A combination of contrast MRA, and spin-echo and phase-contrast MRI techniques allows comprehensive assessment of intra-atrial baffles (Fig. 23.11).

Congenitally corrected transposition Congenitally corrected transposition is a rare disorder characterized by both atrioventricular discordance and ventricle–arterial

discordance (right atrium to left ventricle to pulmonary artery to lung; and left atrium to right ventricle to aorta). Thus, although the heart is anatomically abnormal, it is physiologically normal in terms of the pulmonary and systemic circuits. This deformation does not usually cause cyanosis; however, many of the problems are similar to those experienced by patients with TGA. Congenitally corrected transposition (CCTGA) may be asymptomatic and in some patients is an incidental finding. However, the majority of patients with CCTGA have associated cardiac lesions, the most common being VSD. Pulmonary stenosis is present in approximately 50% of cases and tricuspid valve abnormalities (i.e. Ebstein's abnormality) are found in 20% of cases. Even without associated abnormalities, the majority of patients with CCTGA eventually develop systemic ventricular failure. The main role of MRI is in evaluation of associated lesions, quantification of ventricular function, and assessment of postoperative complications.

Double outlet right ventricle In double outlet right ventricle (DORV), both great vessels emerge from the right ventricle, with the left ventricle emptying through a VSD into the right ventricle.The clinical picture and type of surgical correction depend on the arrangement of the great vessels and the anatomy of the VSD. The most common variant is a normal arrangement of the great vessels and a subaortic VSD. This variant is often referred to as the ‘Fallot's type’, as it is often associated with pulmonary stenosis and has a similar presentation. DORV may also be associated with an anterior aorta and a subpulmonary VSD, known as the ‘Taussig–Bing’ anomaly. Surgical correction for the ‘Fallot type’ variant consists of patch closure of the VSD, which redirects blood to the aorta, and correction of any pulmonary stenosis. For the ‘Taussig– Bing’ anomaly the surgical approach depends on the presence of pulmonary obstruction. In the absence of pulmonary obstruction, correction consists of patch closure of the VSD and arterial switch. In the presence of obstruction, left ventricular flow is tunnelled through the VSD to the aorta, and a right ventricle–pulmonary artery pathway is established—the Rastelli procedure. MRI plays an important role in pre-operative assessment. The 3D anatomy of the VSD and the arrangement of the great vessels are well visualized by MRI and are particularly important when deciding the type of surgery. Spin-echo black blood MRI of the VSD has been shown to compare well with surgical findings and is able to indicate the optimal type of repair.

Common arterial trunk Common arterial trunk (truncus arteriosus) is defined as a single arterial trunk arising from both ventricles, which overrides a large misaligned VSD.The pulmonary, systemic and coronary arteries originate from the single common arterial trunk. The classification of common arterial trunk relies on the branching pattern of the pulmonary artery. In Type I, a short main pulmonary artery arises from the common trunk and subsequently divides. In Type II, the right and left pulmonary arteries originate from the posterior wall of the common trunk

CHAPTER 23

and in Type III, the right and left pulmonary arteries emerge from the lateral wall of the common trunk. The truncal valve is often abnormal with varying degrees of stenosis and insufficiency. About 40% of truncal arches are on the right side: most truncal arches rise higher in the mediastinum than the normal aortic arch. Surgical repair consists of reconstruction of the common trunk to produce a systemic vessel from the left ventricle, patch closure of the VSD and establishment of a right ventricle-to-pulmonary artery conduit. The main role of MRI is in assessment of postoperative complications (homograft failure, truncal valve regurgitation,VSD patch leak). MRA can be used to better delineate the vascular anatomy before surgery.

The single ventricle The single ventricle can be either an anatomical entity, e.g. in tricuspid atresia (single left ventricle) or hypoplastic left heart syndrome (single right ventricle), or a functional entity, where two ventricles are connected by a large VSD. Even when anatomically there is a single ventricle, there is usually a vestigial remnant of the other ventricle. Depending on the size of the ventricles, and the great vessel anatomy, it may be possible surgically to ‘septate’ the ventricles to create a biventricular circuit. If the ventricular sizes are not potentially equal, separation of the pulmonary and aortic circulations is required (the Fontan circulation) (Fig. 23.12), such that the single ventricle pumps blood into the systemic circulation, and systemic venous return is directed to the pulmonary circulation. The creation of this single ventricle circuit is performed in a stepwise surgical fashion (Fig. 23.12).

Bidirectional Glenn circulation The first stage in the creation of a single ventricular circulation is the bidirectional Glenn operation (or hemi-Fontan operation). In this procedure, a side-to-side anastomosis is created between the SVC and the pulmonary arteries, and a patch is inserted to divide the SVC from the right atrium. Any previous surgical systemic-to-pulmonary artery shunts are also

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

taken down at this time. MRI can be used to assess branch pulmonary artery narrowing and pulmonary venous obstruction prior to completion of the Fontan circulation, otherwise the circulation may fail.

Fontan circulation The Fontan circulation (Fig. 23.12) is completed by either anastomosis of the right atrium to the Glenn shunt (classical Fontan operation) or creation of a lateral or extracardiac tunnel between the IVC and the Glenn shunt [total caval pulmonary connection (TCPC)]. The latter procedure is now the preferred option; there remains, however, a sizeable population with a standard Fontan circulation who require diagnostic assessment. Right atrial dilatation is the major complication of the Fontan procedure and the 3D anatomy of the atrium can be delineated by MRI. Right atrial dilatation may cause pulmonary vein compression, which can lead to failure of the Fontan circulation. MRI also evaluates the branch pulmonary arteries and tunnels, and can be used accurately to assess ventricular function.

Hypoplastic left heart syndrome Hypoplastic left heart syndrome (HLHS) is the fourth most common cardiac malformation to present in the first year of life, with an incidence of approximately 266 per million live births. HLHS constitutes a spectrum of CHD with hypoplasia or atresia of the left heart components, and normal relation of the great vessels. At birth, the right ventricle supplies both the systemic and pulmonary circulations via the pulmonary artery and patent arterial duct. After birth, closure of the PDA and the presence of a restrictive patent foramen ovale lead to increasing cyanosis, heart failure and early death. Surgical treatment has been revolutionized by the introduction of the Norwood procedure in 1980 with subsequent conversion to a total cavopulmonary connection (TCPC). At all stages MRI is used to assess fully 3D anatomy, complications, function and flow.

Figure 23.12 Fontan circulation. 3D volume-rendered MRA. (A) Bidirectional Glenn shunt (blue), posterior view. Arrowhead = right pulmonary artery, arrow = left pulmonary artery, asterisk = superior vena cava, dashed arrow = descending aorta. (B) Classical Fontan circulation showing severe right atrial dilation (arrowhead). Arrow = right atrial appendage, asterisk = inferior vena cava drainage. (C) Total caval pulmonary connection (TCPC) lateral tunnel Fontan circulation. Arrowhead = lateral tunnel.

463

464

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Abnormal pulmonary venous drainage MRI can help evaluate the course of the pulmonary veins, differentiating forms of partial anomalous pulmonary venous return (e.g. right upper pulmonary vein to SVC in association with ASD–sinus venosus defect) and total anomalous pulmonary venous return. In total anomalous pulmonary venous drainage (TAPVD) the pulmonary veins coalesce posterior to the left atrium, but do not drain into it. Drainage from this venous confluence to the right atrium may be: (a) via either an ascending vein to the innominate vein, and then to the SVC (supracardiac, Type I) (Fig. 23.13A,B); (b) the coronary sinus directly into the right atrium (cardiac, Type II); or (c) via a descending vein, which passes through the diaphragm into either the IVC or portal venous system (infracardiac, Type III) (Fig. 23.13C). The infracardiac type (Fig. 23.13C) is the least common variant and makes up approximately 12% of TAPVD cases. There is a usually a degree of obstruction as the descending vein passes through the diaphragm. Thus, unlike the supracardiac and cardiac variants, which present with a left to right shunt and cardiac failure, the clinical picture for infracardiac TAPVD is one of pulmonary congestion with tachypnoea, tachycardia, liver enlargement, cyanosis, pulmonary oedema and respiratory distress. Symptoms usually appear within 24– 36 h of birth. Importantly, the diagnosis of infracardiac TAPVD can be missed at echocardiography, and must always be considered in the differential diagnosis of pulmonary oedema on the neonatal chest radiograph.

Associations of TAPVD are complex cardiac anomalies such as heterotaxy syndromes (in particular right isomerism), AVSD, pulmonary stenosis, DORV, HLHS, common arterial trunk, transposition of the great arteries and aortic coarctation. The anomalous pulmonary venous connection is easily visualized with ultrasound, but in patients with poor echocardiography acoustic windows, cross-sectional imaging with MRI and CT is very useful, often avoiding the need for a diagnostic X-ray catheterization.

Ebstein's anomaly Ebstein's anomaly is a congenital abnormality of the tricuspid valve. The septal and mural leaflets are more apically placed than normal, resulting in a malfunctioning, regurgitant tricuspid valve and atrialization of the proximal right ventricle5.The consequence is gross right atrial enlargement and raised right atrial pressure. The anomaly is usually associated with an ASD and there is thus right-to-left shunting at the atrial level and subsequent cyanosis. Ultimately, Ebstein's anomaly results in gross enlargement of the cardiac contour with a prominent curved right atrial border on the plain chest radiograph.Treatment is problematical, though expert surgical repair of the tricuspid valve is possible in some centres. MRI can be used to assess the valve morphology, quantify ventricular function and size the right atrial enlargement.

Cor triatrium This is a rare congenital anomaly in which the pulmonary veins and the immediately related left atrium are separated from the main body of the left atrium by a perforated

Figure 23.13 Supracardiac total anomalous pulmonary venous drainage (TAPVD). (A) Angiogram with contrast medium injected into the pulmonary artery. The pulmonary veins (single black arrows) join behind the heart to form a vertical vein (double black arrows) which passes upwards to join the left innominate vein which passes to the right to enter the superior vena cava which enters the right atrium. (From Grainger R G 1970 Transposition of the great arteries and of the pulmonary veins, including an account of cardiac embryology and chamber identification. Clin Radiol 21: 335–354.) (B) Dextrocardia with complex cardiac anomalies. 3D volume-rendered MDCT angiogram viewed from the left anterior oblique showing a supracardiac total anomalous pulmonary venous drainage. The pulmonary veins (white arrows) drain into an ascending vein (asterisk) and then into the brachiocephalic vein (white arrowhead and black crossed arrow) and the superior vena cava, which drains into the right atrium. Double arrow = descending aorta. (C) Total anomalous infradiaphragmatic pulmonary venous drainage (Type 3). Contrast medium injected into the main pulmonary artery illustrates all the pulmonary veins joining behind the heart to form a vertical vein which descends (arrow) through the oesophageal hiatus to enter the portal veins. The vertical vein is often compressed in its lower segment which can contribute to pulmonary oedema. (From Grainger R G 1970 Transposition of the great arteries and of the pulmonary veins, including an account of cardiac embryology and chamber identification. Clin Radiol 21: 335–354.) (With permission from the Royal College of Surgeons.)

CHAPTER 23

•

CONGENITAL HEART DISEASE: GENERAL PRINCIPLES AND IMAGING

Figure 23.14 Anomalous coronary arteries. Maximum intensity projection images reconstructed from the 3D coronary MRAs. (A) Oblique sagittal view demonstrating an anomalous origin of the left circumflex (asterisk) from the right coronary artery (arrow). (B) Coronal view demonstrating an anomalous accessory left anterior descending artery (arrowhead) arising from the same ostia as the right coronary artery (dashed arrow).

fibromuscular septum. Cor triatrium is often associated with septal defects, anomalous pulmonary venous drainage and tetralogy of Fallot. Symptoms depend on the size of the orifice in the septum and may present with a syndrome similar to mitral valve stenosis with pulmonary venous congestion and oedema. Diagnosis is best by echocardiography.Treatment is by surgical excision of the obstructing membranous septum.

Coronary arteries Anomalous coronary arteries Although the anomalous origin and epicardial course of the coronary arteries are uncommon entities, they may be the cause of chest pain and sudden cardiac death, often in a younger asymptomatic patient. Together with hypertropic cardiomyopathy and arrhythmogenic right ventricular dysplasia, anomalous origin of the left coronary artery from the right coronary or pulmonary artery is one of the most frequent causes of sudden cardiac death in competitive young athletes. The prevalance of anomalous coronary arteries has been reported to be between 0.3% and 1.0% of the normal population, although this is likely an underestimate, as many asymptomatic individuals may be unrecognized. From an anatomical point of view, the anomalies are classified according to the coronary artery involved, the origin of the anomalous coronary artery, and the anatomical course of the proximal segment. From a clinical point of view, the anomalies are divided into ‘benign’ and ‘malignant’ lesions (Fig. 23.14). The latter, especially those of left coronary artery origin from the pulmonary artery and in which the anomalous artery passes between the aortic root behind and right ventricular outflow tract or pulmonary artery in front, have an increased risk for developing myocardial ischaemia and sudden cardiac death. Even with multiple projections, the precise location of the proximal course of the vessel in a patient with an abnormal origin of a coronary artery can be difficult to depict with conventional angiography. However, MRA and CT angiography provide reliable visualization of the root of the arteries, and the coronary artery tree (Fig. 23.14).

CONCLUSION Cardiovascular imaging is important for the diagnosis and follow-up of children and adults with CHD. In young patients, echocardiography is the first-line imaging modality; however, when echocardiography cannot provide a complete diagnosis, cross-sectional imaging with MRI and CT is rapidly becoming the next line of investigation, with interventional catheter angiography reserved for problem solving, the assessment of the coronary arteries and pulmonary vascular resistance, and for catheter-guided therapeutic procedures. In older children and adults, where echocardiography is less easy (poor acoustic windows and multiple operations), crosssectional imaging is essential. MRI in particular is important in this setting, as there is no radiation burden, and both anatomy and function can be assessed in order to suggest optimal follow-up and timing of future interventions. Used in combination, echocardiography, MRI, CT and Xray catheter angiography can provide a complete assessment of patients with CHD.The further application of MRI and CT to CHD should enable radiologists to take a more proactive role in the assessment of CHD in a multidisciplinary team setting.

Acknowledgements We would like to acknowledge the following for their help and support: Professor Jan Bogaert, Professor Steven Dymarkowski, Professor Reza Razavi, Dr Vivek Muthurangu, Dr. Sanjeet Hegde, Professor John Deanfield, Professor Philipp Bonhoeffer, Dr. Catherine Owens and Springer-Verlag.

REFERENCES 1. Grainger R G 1985 The pulmonary circulation: The radiology of adaptation. Clin Radiol 36: 103–116 2. Grainger R G 1970 Transposition of the great arteries and of the pulmonary veins, including an account of cardiac embryology and chamber identification. Clin Radiol 21: 335–354 3. Anderson R H, Baker E J, Macartney F J, Rigby M L, Shinebourne E A, Tynan M (eds) 2002 Paediatric cardiology, 2nd edn. Churchill Livingstone, London, 2002

465

466

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

4. Hoffman J I, Kaplan S 2002 The incidence of congenital heart disease. J Am Coll Cardiol 39: 1890–1900 5. Anderson R H, Razavi R, Taylor A M 2004 Cardiac anatomy revisited. J Anat 205: 159–177 6. Snider A R, Serwer G A, Ritter S B 1997 Echocardiography in pediatric heart disease. Mosby, St Louis 7. Iskandrian A E, Verani M S (eds) 2002 Nuclear cardiac imaging: Principles and applications. Oxford University Press, New York 8. Bogaert J, Dymarkowski S, Taylor A M (eds) 2005 Clinical cardiac MRI. Springer, New York

9. Ohnesorge B M, Becker C R, Flohr T G, Reiser M F (eds) 2005 Multi-slice CT in cardiac imaging: Technical principles, imaging protocols, clinical indications and future perspective. New York, Springer 10. Gatzoulis M A, Webb G A, Piers Daudeney P (eds) 2003 Diagnosis and management of adult congenital heart disease. Churchill Livingstone, London 11. Khambadkone S, Coats L, Taylor A et al 2005 Percutaneous pulmonary valve implantation in humans – Results in 59 consecutive patients. Circulation 112: 1189–1197

CHAPTER

Nonischaemic Acquired Heart Disease

24

Hrudaya Nath and Satinder P. Singh

• Role of imaging • Valvular heart disease • Prosthetic cardiac valves • Diseases of the heart muscle (cardiomyopathies) • Tumours of the heart • Trauma to the heart

The incidence of cardiac disease is steadily rising all over the world, and at the beginning of the 21st century, cardiovascular disease caused nearly half the deaths in the developed world, and about 25% in the developing world1. Coronary arterial disease accounts for slightly greater than half of these cases2. The remainder form a heterogeneous group of diseases, including acquired valve diseases, diseases of the heart muscle (most resulting from hypertension), cardiac tumours, and trauma. These nonischaemic causes of acquired heart disease (NIHD) are the subject of this chapter.

ROLE OF IMAGING The diagnosis, assessment, and follow-up of NIHD have changed considerably over the years. While chest radiography is still useful in judging cardiac size and pulmonary haemodynamics, echocardiography and cardiac magnetic resonance imaging (MRI) have become increasingly useful. Echocardiography remains the most important imaging method in assessing ventricular function and status of the cardiac valves, and estimating transvalvar gradient as well as pulmonary artery pressures. Helical computed tomography (CT), particularly the emerging faster multidetector CT systems (MDCT) using ECG gating, is beginning to make inroads into the assessment of NIHD. This technology has the potential to become even more important in patients with contraindications for MRI.

Chest radiography While very non-specific, postero-anterior (PA) and lateral views of the chest are the most frequently performed radio-

graphic examinations in the evaluation of cardiac disease. In many conditions of NIHD, the chest radiographs are normal. However, when abnormalities are noted, they are quite useful in judging the functional state of the heart and for serial follow-up. Cardiomegaly (when the cardiothoracic ratio in a PA view of the chest exceeds 50%) is a very useful sign in detecting cardiac dysfunction, and is often due to ventricular enlargement. Right atrial enlargement and pericardial effusion, in the absence of ventricular enlargement, are far less common causes of enlargement of cardiac silhouette. Chest radiographs can also frequently detect moderate to significant left atrial enlargement (Fig. 24.1). It is not possible to confidently distinguish between right and left ventricular enlargement and detection of mild degrees of enlargement of any cardiac chamber (e.g. compared to chamber dimensions from echocardiography or MRI). Chest radiography may also detect abnormal cardiac calcifications, most notably that of the aortic valve (Fig. 24.2) and mitral annulus (Fig. 24.3). Chest radiographs are also useful in assessing pulmonary venous hypertension and pulmonary oedema (Fig. 24.4). However, in patients with chronic, compensated congestive failure, often the only radiographic abnormality is cardiac enlargement.

Echocardiography Echocardiography is a noninvasive, portable imaging technique that allows high-resolution, two- and three-dimensional (2D and 3D) views of the cardiac chambers, valves, and pericardium.These techniques, either with a transthoracic or semi-invasive transoesophageal approach, can assess cardiac anatomy and ventricular function. When combined with Doppler and colour Doppler techniques, valvar regurgitation and transvalvar pressure gradients can also be assessed. Echocardiography is the most commonly performed imaging examination in the assessment of NIHD.

Magnetic resonance imaging Cardiac magnetic resonance imaging (CMR) is rapidly becoming a very useful imaging method in the assessment of NIHD. Its role is most valuable in (A) serial measurement of

468

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.3 Lateral view of chest shows C-shaped dense calcification of the mitral annulus (arrows). Figure 24.1 Classical appearance of rheumatic mitral stenosis. PA view of the chest. The heart size is normal. The enlarged left atrium (A) displaces the left bronchus upwards (asterisk) and creates a right retrocardiac double density. The left atrial appendage is enlarged (arrowheads). There is severe pulmonary venous hypertension.

ventricular function in patients with cardiomyopathy (considered superior to echocardiography in reproducibility); (B) evaluation of valve function, including stenosis and regurgitation; (C) morphology and extent of involvement in cardiac tumours; and (D) value of post-contrast delayed enhancement in determining prognosis in hypertrophic cardiomyopathy, dilated cardiomyopathy (DCM), as well as many types of infiltrative myocardial diseases including sarcoidosis and myocarditis.

Figure 24.2 Aortic valve calcification. Lateral chest radiograph shows several calcifications in the aortic valve (arrow).

Computed tomography Until recently, conventional computed tomography (CT) had little role in the evaluation of NIHD. While electron beam CT is comparable to echocardiography and CMR in assessing cardiac function, its limited availability was a drawback to its widespread use. The increasing use of ECG-gated helical CT techniques coupled with greater than 16-detector MDCT has

Figure 24.4 PA view of the chest. Severe mitral re-stenosis following closed mitral commissurotomy. The heart is enlarged. Note the concavity in the region of the left atrial appendage resulting from closed commissurotomy (arrows). Compare to Fig. 24.1 (typical mitral stenosis).

CHAPTER 24

the potential to make cardiac CT a viable alternative in assessing cardiac function. More recently cardiac CT was shown to be useful in the assessment of valvar function3. However, at present, CT has no demonstrable advantage over echocardiography and CMR in the evaluation of NIHD.

VALVULAR HEART DISEASE In the past few decades the aetiology of acquired valve disease shifted from rheumatic and infective causes to degenerative diseases, with increasing prevalence of mitral regurgitation and aortic stenosis compared to the higher incidence of mitral stenosis and aortic insufficiency. Consequently, the age at which these diseases become clinically evident also has advanced. Many of these disease processes have few radiographic findings, and their evaluation primarily involves echocardiography4.

Mitral valve disease5 In the western world, non-rheumatic mitral valve disease is now the most common manifestation of mitral valve abnormality. Rheumatic heart disease, with the virtual disappearance of rheumatic fever, has become a relative rarity in the western world. Worldwide, rheumatic heart disease still remains a prevalent and dangerous condition.

Non-rheumatic mitral valve disease Non-rheumatic disease is usually manifest as mitral regurgitation: non-rheumatic mitral stenosis is extremely rare. Many conditions can result in significant mitral regurgitation (Table 24.1). Mitral valve prolapse6 Elongation of the chordae tendineae associated with myxomatous degeneration of the valve leaflets allows prolapse of the mitral valve leaflets back into the left atrium in ventricular systole. The condition most commonly

Table 24.1

CAUSES OF MITRAL REGURGITATION

Abnormalities of valve leaflets Mitral valve prolapse (idiopathic, Marfan’s syndrome) Bacterial endocarditis Prosthetic mitral valve leaks Mitral commissurotomy/balloon valvuloplasty Congenital (endocardial cushion defect, etc.) Mucopolysaccharidosis Systemic lupus erythematosis Rheumatoid arthritis

Abnormalities of supporting structures Dilatation of mitral valve annulus Ruptured chordae tendineae Ischaemic heat disease causing papillary muscle Dysfunction, rupture Hypertrophic cardiomyopathy Mitral annular calcification Atrial myxoma

•

NONISCHAEMIC ACQUIRED HEART DISEASE

occurs alone but may also occur in Marfan’s syndrome and is frequently seen in patients with atrial septal defect. The diagnosis is made by 2D ultrasound (US), mitral prolapse also being visible on CMR (Fig. 24.5) and conventional left ventriculography (Fig. 24.6). The severity of the pathological change varies from involvement of one of the three scallop-shaped segments of the posterior leaflet of the mitral valve, to the whole of both leaflets. Prolapse of the mitral valve may be associated with chest pain and ECG changes, which may suggest ischaemic heart disease. Chordal rupture Mitral regurgitation may also result from rupture of groups of chordae tendineae, which may complicate bacterial endocarditis or less frequently myocardial infarction or connective tissue diseases.This allows part of a leaflet to flail, preventing proper co-aptation in systole and producing severe mitral regurgitation.The murmur is usually mid to late systolic in onset. When this event occurs suddenly, acute pulmonary oedema may occur, with normal cardiac size (Fig. 24.7). The severity of mitral regurgitation can be estimated with CMR and echocardiography, where signal loss due to turbulent regurgitant flow allows semiquantitative assessment of regurgitation (Fig. 24.8). More accurate estimates of the severity of mitral regurgitation use phase-contrast CMR, comparing forward mitral flow with forward aortic flow7. In the absence of aortic regurgitation or a ventricular septal defect, the difference indicates the amount of mitral regurgitation. One of the most widely used applications of intra-operative transoesophageal US is in determining the functional status of mitral valve repair by detecting and quantifying mitral regurgitation during the repair, thus allowing further correction if necessary. Bacterial endocarditis may lead to mitral regurgitation by causing cusp perforation or chordal rupture. Vegetations and penetrating abscesses may also interfere with valve function. While these abnormalities can be well shown by transthoracic US or MRI, transoesophageal US may be required to best show paravalvar abscesses or cusp perforation. Subacute bacterial endocarditis usually, but not invariably, affects abnormal valves, while acute bacterial endocarditis can affect normal or abnormal valves. Papillary muscle dysfunction/rupture Mitral regurgitation may also result from papillary muscle dysfunction. This is the presumed origin of the systolic murmurs that are commonly heard in the course of inferior myocardial infarction, though the amount of regurgitation is rarely significant.When the papillary muscle ruptures as the result of myocardial infarction, the degree of regurgitation is massive and often rapidly fatal, unless it can be urgently corrected surgically. Functional mitral regurgitation Mitral regurgitation may also result from dilatation of the left ventricle (possibly by causing tethering of chordae tendinae), or mitral annulus dilatation, as

469

470

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.5 Mitral valve prolapse. (A) M-mode echocardiography shows late systolic prolapse of the posterior mitral leaflet (arrows). (B) 2D echocardiography (parasternal long axis) in the same patient shows posterior mitral leaflet prolapse (arrow) into the left atrium. (C) Colour Doppler (apical four-chamber) in the same patient shows an eccentric jet of mitral regurgitation (following direction of arrows). Gradient-echo cine-MRI (D) four-chamber projection and (E) left ventricular outflow tract (LVOT) view. There is bulging of the posterior cusp of the mitral valve into the left atrium (arrow). There is moderate mitral regurgitation (arrowheads), better seen on the LVOT view (E). Ao = Ascending aorta, LA = left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

occurs in dilated cardiomyopathy or ischaemic cardiac failure9. Mitral regurgitation in these situations is very common, but is rarely severe. The appearances on chest radiography depend on the duration and severity of the mitral regurgitation and any other associated heart disease. Acute, severe, non-rheumatic mitral regurgitation may present with pulmonary oedema, but with a virtually normal heart size and shape. After an interval, the heart usually compensates by developing left ventricular dilatation, which may be marked (Fig. 24.9). Selective left atrial enlargement may be absent, slight, or moderate, and the left atrial appendage is usually not enlarged. Pulmonary vascular changes reflect the haemodynamic derangement and the effects of treatment. Mitral annulus calcification10 Calcification occurs in the angle between the left ventricle wall and the mitral valve cusps, and is readily recognized on the chest radiograph as a C-shaped open ring (see Fig. 24.3); the gap in the ring occurs where the anterior mitral leaflet base is in contact with the posterior aortic valve ring. Calcification rarely occurs before the age of 70 years and is more common in women. The mitral annulus calcification is not always benign, since it may be associated with transient ischaemic attacks due to

emboli or carotid stenosis, and atrioventricular conduction disturbances11. Mitral valve function may be affected, most commonly leading to mild mitral regurgitation, but significant mitral stenosis from heavy annulus calcification has been described. The risk of infective endocarditis may be increased. There is suggestion that mitral annular calcification is another marker for atherosclerosis, and premature calcification may be seen in hypercalcaemic states, particularly in patients with end-stage renal disease.

Rheumatic mitral valve disease Acute rheumatic carditis results in a pancarditis. During this phase, mitral regurgitation is often present, resulting from myocarditis. Such mitral regurgitation is usually transient and reversible, as the myocarditis subsides. Chronic rheumatic mitral valve disease often results in stenosis. This is the result of fusion of the commissures, thickening and shortening of the chordae tendineae, with fibrosis of the papillary muscles. Often there is a combination of all these features, in varying degrees. Depending on the severity of the process, pure valve regurgitation results if there is extensive leaflet destruction. Pure stenosis results if commissural fusion occurs with virtually normal leaflets, but the most common result is a mixture of mitral stenosis and regurgitation, either of which may be

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.6 Non-rheumatic mitral regurgitation due to mitral valve prolapse. Left ventriculography (right anterior oblique [ROA] projection). (A) Diastole. Arrows indicate the recess under the mitral valve, which persists in diastole. (B) Early systole. Prolapse of the mitral valve (arrows) is just appearing. (C) Mid systole. Mitral valve prolapse (arrows) has reached its fullest extent. (D) Late systole. After maximum prolapse, reflux into the left atrium (arrowhead) begins.

worsened by chordal involvement. Now it is believed that the valve stenosis results from a combination of the smouldering rheumatic process as well as abnormal turbulence from the flow across the fused commissures, resulting in fibrosis of the valve structures and superimposed calcification12. Mitral stenosis, if present, limits the amount of regurgitation, so that both cannot be severe at the same time. Other acquired causes of mitral stenosis are extremely rare and congenital mitral stenosis is quite uncommon. Valve replacement or valvuloplasty is indicated for severe stenosis (valve area 0.8–1.2 cm2 depending on the exercise tolerance required). Mitral valve disease increases left atrial pressure: this is transmitted back to the pulmonary veins and, when severe, leads first to interstitial and then to alveolar pulmonary oedema. Parenchymal lung changes of haemosiderosis and intrapulmonary ossification may appear after several years of pulmonary venous congestion (Fig. 24.10B). Secondary pulmonary arterial hypertension may develop and lead to pulmonary valve regurgitation, right ventricular dilatation, and functional tricuspid regurgitation. There may also be organic involvement of the tricuspid valve by the rheumatic process, leading to either tricuspid stenosis and/

or regurgitation. Long-standing mitral stenosis can result in atrial fibrillation, complicated by thrombi in the left atrium. These thrombi can result in systemic emboli. Clinical features After the attack of rheumatic fever, mitral valve damage may remain asymptomatic until stenosis becomes critical or regurgitation becomes severe, when clinical cardiac failure develops. In western countries a history of rheumatic fever is often absent. Acute cardiac failure may be precipitated by the development of atrial fibrillation, an almost inevitable consequence of rheumatic mitral valve disease. Chest radiography The cardinal radiological feature of rheumatic mitral valve disease is left atrial enlargement. The left atrial appendage is particularly affected; when it is seen to be enlarged this suggests a rheumatic aetiology. The appearances vary from a simple straightening of the left heart border (see Fig. 24.1) to a large bulge in the characteristic site of the appendage, immediately below the left main bronchus (Fig. 24.10). The left atrium, when grossly dilated, can also enlarge to the right (Fig. 24.11) posteriorly causing dysphagia.

471

472

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.7 Acute mitral regurgitation due to chordal rupture in a patient with mitral valve prolapse. (A) Baseline PA view of the chest which appears normal apart from hiatus hernia. (B) At the time of acute mitral regurgitation, note the enlarged heart, left atrial and ventricular enlargement, and greater pulmonary oedema in the right lung.

The features of enlargement of the left atrium vary from being visible on the plain chest radiograph as a ‘double density’ through the heart (see Fig. 24.1), to ‘aneurysmal enlargement’ when the left atrium reaches to within an inch or so of the chest wall on one or both sides of the chest.Very large left atria are more commonly seen in long-standing mitral regurgitation with atrial fibrillation. Occasionally, a large left atrium may displace the oesophagus backwards and to the left instead

Figure 24.8 Mitral regurgitation. Axial gradient-echo systolic MRI shows a jet-like signal void in the left atrium due to moderate mitral regurgitation. LA = Left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

of the more frequent displacement to the right. Enlargement of the left ventricle is not a feature of pure mitral stenosis but is seen with significant mitral regurgitation. Calcification may occur, either in the wall of the left atrium, or in clot lining the wall (Fig. 24.12).The calcification is usually curvilinear, at least in part, and lies fairly high in the cardiac shadow, higher than is usually seen with pericardial calcification. The lateral radiograph shows calcification in the upper posterior aspect of the heart at the top of the left atrium. Fluoroscopy or CT can be used to detect dystropic calcification that takes place in the mitral valve in long-standing rheumatic valve disease. It occurs in the mitral valve, lying between the left atrium and left ventricle, near the postero-inferior aspect of the heart in the lateral view. Unlike the J- or C-shape of mitral annulus calcification (see Fig. 24.3), calcification in the mitral valve itself is strongly suggestive of a rheumatic aetiology. The right ventricle is enlarged if there is pulmonary arterial hypertension, pulmonary valve regurgitation, or tricuspid valve regurgitation. Tricuspid valve disease, stenosis, or regurgitation may be suspected if there is right atrial enlargement. The constantly present left atrial hypertension is reflected back into the pulmonary veins and produces changes in the pulmonary vascularity and the lungs that may be seen on the plain chest radiograph (see Figs 24.1, 24.4). Symptomatic mitral stenosis of such severity as to require surgical treatment may, however, be present with an entirely normal chest radiograph, particularly if diuretic treatment is vigorous. Closed mitral valvotomy, with amputation of the left atrial appendage to access the valve via the left atrium, produces

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.9 (A) Acute mitral regurgitation, ruptured chordae tendineae. Acute onset of dyspnoea. The heart size is normal with some straightening of the left heart border. The lungs show upper lobe blood diversion and pulmonary oedema. (B) Compensated phase (another patient). The physical signs of non-rheumatic mitral regurgitation were present. The heart shows compensatory left ventricular enlargement but no evidence of left atrial enlargement.

a concavity in the left heart border below the left bronchus (see Fig. 24.4).This may give the heart a left ventricular configuration, making left ventricular enlargement difficult to assess. The relative contribution of mitral stenosis and regurgitation cannot be recognized with certainty, and might be due to an associated aortic valve lesion. If the left atrium is very large, mitral regurgitation should be suspected. Echocardiography is the best method for routine assessment of mitral valve disease.

Other imaging techniques In mitral stenosis, CMR shows restricted opening of the mitral valve and thickened leaflets (Fig. 24.13). Phase-contrast CMR can be used as a substitute for Doppler in assessing valve area. In mitral regurgitation, CMR can be used as a semiquantitative method for assessing mitral regurgitation in the same way as colour Doppler13. Cardiac catheterization and angiocardiography is used in those rare situations where US has failed to elucidate the contribution of each

Figure 24.10 (A) Severe mitral valve disease. The left atrial appendage is large, producing a convex bulge (arrow). The heart is considerably enlarged. (B) Pulmonary haemosiderosis in mitral stenosis. Long-standing severe mitral stenosis. The heart and left atrium are enlarged. Bilateral nodular interstitial prominence is due to pulmonary haemosiderosis.

473

474

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.11 Severe mitral valve disease. (A) The left atrium (white arrow) has enlarged to the right, protruding well beyond the border of the right atrium (black arrowheads). (B) It is also enlarged posteriorly, displacing the barium-filled oesophagus (and the left bronchus) posteriorly. The left atrial appendage is not grossly enlarged.

valve lesion, or when coexistent coronary artery disease needs assessment.

Tricuspid valve disease The tricuspid valve may be directly involved by rheumatic disease, producing organic regurgitation (Fig. 24.14) or, less commonly, stenosis. The tricuspid valve may also become functionally regurgitant when the right ventricle becomes dilated in the presence of pulmonary hypertension, e.g. secondary to mitral valve disease. Severe tricuspid regurgitation is now being seen more often in patients who have undergone mitral valve replacement many years previously. In Ebstein’s anomaly (Fig. 24.15), the insertion of the septal cusp of the tricuspid valve is displaced towards the apex of the right ventricle. Severe tricuspid regurgitation can also occur in

endomyocardial fibrosis when the papillary muscles become scarred and retracted. In carcinoid syndrome, the valve cusps are mildly thickened and shortened; in long-standing cases they may not be identifiable. There is subendocardial fibrosis of the right ventricle, increasing its stiffness. Carcinoid syndrome usually causes tricuspid regurgitation but may also cause tricuspid stenosis.

Other causes of tricuspid valve abnormality The tricuspid valve may be made regurgitant by endocarditis, a frequent cause where intravenous drug abuse is widespread. Infective emboli to the lungs may cause cavitating pneumonia.The tricuspid valve may be disrupted by trauma. It may be abnormal in congenital heart disease, especially in congenital stenosis, Ebstein’s anomaly (Fig. 24.15), and corrected transposition of the great arteries.

Figure 24.12 Mitral stenosis and regurgitation following mitral commissurotomy. (A) PA and (B) lateral views of the chest. There is massive left atrial dilatation and partly calcified mural thrombus in the dilated left atrium (arrows in B). Note the left thoracotomy.

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.15 Ebstein’s anomaly. Gradient-echo axial MRI. Due to adherence of septal and posterior leaflets of tricuspid valve to the right ventricle, the free portion of the leaflets (arrows) is located at a variable distance below the atrioventricular annulus within the right ventricle (RV) (downward displacement). The right atrium (RA) and the atrialized portion of the RV (asterisk) are enlarged with reduced functional RV chamber size.

Figure 24.13 Mitral stenosis. (A) CMR, four-chamber projection (diastolic image) shows a large right heart due to chronic mitral stenosis with an atrial septal defect (ASD) (open arrow)—Lutembacher syndrome. There is restricted opening of the mitral valve (between curved arrows). The left atrium is not significantly enlarged due to the coexisting ASD. (B) CMR, four-chamber projection. There is fusion of commissures and doming of the mitral valve (arrows).

The clinical recognition of tricuspid valve disease can be difficult. High venous pressure with a big ‘v’-wave and a pulsating enlarged liver suggests tricuspid regurgitation. A slow ‘y’ descent in neck vein pulsation suggests tricuspid stenosis. The murmurs of tricuspid and mitral valve disease are similar.

Figure 24.14 Tricuspid valve regurgitation (arrows). CMR, fourchamber projection, systolic image. LA = Left atrium, RA = right atrium.

Tricuspid valve disease produces a large right atrium, bulging to the right and increasing the curvature of the right heart border, or by detecting enlargement of right atrial appendage in the lateral projection, seen as increased retrosternal density in the lateral projection superior to the expected location of the right ventricle. This is usually seen just below the level of the aortic arch. A prominent right atrium developing in rheumatic heart disease always suggests tricuspid valve involvement (Fig. 24.16). Sometimes regression of right atrial enlargement may be seen in functional tricuspid regurgitation when mitral valve disease is treated. In the authors’ experience, right atrial enlargement, unless gross, is difficult to assess when other cardiac chambers are enlarged. Rheumatic tricuspid valve calcification is almost unknown. As with mitral disease, US is usually the definitive investigation. When regurgitation is very severe, the pressure

Figure 24.16 Tricuspid valve disease. Gross right atrial enlargement (arrow), extending to the right, developing in a patient with severe mitral valve disease, leading to right heart failure.

475

476

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

drop across the valve is low or, in the absence of effective resistance at the level of the orifice, it may be absent altogether. Velocities are therefore low, with peak values occurring early in right ventricular ejection, and equilibration of right ventricular and right atrial pressures before the end of systole. Severe regurgitation can usually be recognized on colourflow Doppler or CMR, although it should be remembered that when the valve is effectively absent, regurgitant velocities may be so low that they do not cause aliasing, and so may not immediately be recognized.

Aortic valve disease Aortic stenosis This section considers acquired stenosis of the aortic valve in adults14. Congenital aortic stenosis, supravalvar and subvalvar aortic stenosis are considered in Chapter 23. Acquired aortic stenosis may be degenerative or inflammatory. Calcific aortic stenosis The changes in the natural history of valve disease not only refer to rheumatic heart disease but also to the aortic valve. Previously, aortic stenosis in adults was most commonly due to calcification on a congenitally deformed bicuspid valve. This condition occurs in patients over 30 years old, especially men.The predominant cause of aortic stenosis in western industrialized countries now appears to be degenerative calcific disease in middle-aged or elderly patients15. Unlike in mitral stenosis, where calcium is deposited on an already stenosed valve, calcific aortic stenosis results from the deposition of masses of calcium on the aortic cusps which obstruct outflow by their bulk as well as by stiffening the cusps. Thus

calcification causes the stenosis. Other common causes include rheumatic heart disease and degeneration of a normal trileaflet aortic valve (usually in a more elderly population). Clinical presentation may include breathlessness, chest pain, or syncope. The ejection click of the bicuspid aortic valve, detected in earlier life, disappears as the cusps become rigid and is replaced by an ejection systolic murmur. Usually left ventricular hypertrophy is seen on the ECG.The clinical diagnosis and assessment of severity are usually straightforward, though in the elderly with severe aortic stenosis and heart failure, recognition of the systolic murmur can be very difficult.The chief differential diagnostic problem is presented by the combination of a systolic murmur and left ventricular hypertrophy, and this may be due to aortic stenosis, non-rheumatic mitral regurgitation, or hypertrophic cardiomyopathy. Echocardiography distinguishes between these. Chest radiography Rounding of the cardiac apex may suggest left ventricular hypertrophy; however, there is usually some degree of cardiac dilatation, which may be marked if there is aortic regurgitation. Localized prominence of the ascending aorta may indicate post-stenotic dilatation, but in older patients the whole thoracic aorta may be widened from atherosclerosis (Figs 24.17, 24.18). The lateral view may show aortic valve calcification (see Fig. 24.2). This is best seen by fluoroscopy or CT (Fig. 24.19). The extent of calcification has only a rough relationship to the severity of the stenosis. Significant aortic stenosis may be present with a virtually normal frontal image, though there is usually evidence of left ventricular enlargement on the lateral view. There may be no visible aortic abnormality.

Figure 24.17 Aortic valve calcification. (A) PA and (B) lateral views of chest. A 45 year old male with calcific stenosis of bicuspid aortic valve. There is prominence of the ascending aorta in the PA view (arrows). In the lateral view, left ventricular hypertrophy is identified by the positive Rigler sign. Arrows = Calcified aortic valve, arrowheads = outline of inferior vena cava.

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.18 Complications of aortic valve replacement. (A) PA and (B) lateral views of the chest. There is aneurysm of the ascending aorta with a Bjork–Shiley aortic valve prosthesis inserted 10 years ago. CT (not shown) demonstrated chronic proximal aortic dissection with aneurysm of the false channel.

Other imaging techniques Calcification in the aortic valve leaflets and adjacent aortic root is well shown by CT17 (Fig. 24.19), which may lead to the diagnosis before the development of symptoms. Ultrafast CT can assess the severity of aortic stenosis by imaging the aortic valve opening but this is not used routinely. CMR can demonstrate impaired aortic valve opening, morphology of the valve (bicuspid or tricuspid, presence of calcification), and assess stenosis, although less accurately than Doppler (Fig. 24.20).The area of systolic flow dephasing seen in the aortic root on CMR has a loose relationship to the severity of aortic stenosis (Fig. 24.20). However, phase-contrast techniques and

planimetry of the aortic valve during systole in gradient-echo sequences obtained in a view optimized to demonstrate the aortic valve in its cross-section (Fig. 24.21) have fair accuracy and compare favorably with US18. During phase-contrast imaging, care should be taken to set theVENC value sufficiently high to detect the highest peak velocity. Left ventricular function and hypertrophy are accurately measured by CMR, which is probably the most accurate method for determining ventricular mass. Associated aortic regurgitation is shown by diastolic flow dephasing in the left ventricular outflow tract (Fig. 24.22). On CMR the cross-sectional area of the flow dephasing immediately below the aortic valve is probably the best guide to the severity of aortic regurgitation, although some advocate phase-contrast MRI19. In older adults with aortic stenosis, there is frequent association of coronary artery disease, and coronary angiography is necessary before surgery for valve replacement. Rheumatic aortic stenosis is caused by inflammatory fusion of the commissures of the aortic valve cusps. It is often associated with aortic regurgitation and usually involvement of the mitral valve.The clinical features are similar to those of calcific aortic stenosis, but dyspnoea is usually pronounced because of associated mitral disease. Additional valve lesions complicate clinical assessment, and US is often required to determine the relative severity of each lesion.

Figure 24.19 Aortic valve calcification. Axial CT at aortic valve level shows calcification of the aortic leaflets (arrows).

Chest radiography The appearances of rheumatic aortic stenosis are commonly dominated by the presence of associated mitral valve disease, producing left atrial enlargement and making recognition of any left ventricular disorder difficult. Post-stenotic dilatation of the aorta is rare in rheumatic aortic stenosis, though the aorta may be dilated if there is also aortic regurgitation. Gross aortic valve calcification is rare in rheumatic disease.

477

478

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.20 Tricuspid aortic stenosis. CMR of (A) aortic valve view, diastolic image, (B) left ventricular outflow tract (LVOT) view, systolic image, and (C) coronal, ascending aorta view. In the aortic valve view (A), fusion of the commissures of a markedly thickened valve is noted (arrowheads). Doming of the valve leaflets is noted in the LVOT projection (arrowheads) (B). The systolic image demonstrates the turbulent jet (arrows) from aortic stenosis (C). AO = ascending aorta, LA = left atrium, LV = left ventricle, PA = pulmonary artery, RA = right atrium, RV = right ventricle. (D) 2D US, short-axis view, showing distribution of calcification in the aortic valve cusps (arrows). The true valve area cannot be seen due to limited spatial resolution and acoustic shadowing by calcifications. (E) 2D US in same patient, parasternal long axis view, showing the stenosed aortic valve (open arrow). There is left ventricular hypertrophy due to aortic stenosis. The posterior wall of the left ventricle is thickened (between curved arrows) and there is a large posterior pericardial effusion (p).

Figure 24.21 Severe aortic stenosis. (A) Short-axis CMR of the aortic valve (diastole) in severe aortic stenosis shows a trileaflet appearance with areas of irregular low signal intensity indicating areas of calcification (arrows). (B) Short-axis CMR of the aortic valve in the same patient (systole) shows restricted opening of the aortic valve (arrow = valve orifice).

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

In rheumatic aortic regurgitation, there is a slow destruction of the free edges of the cusps, often with commissural fusion. Rheumatic aortic regurgitation may occur alone or in association with aortic stenosis, but almost always with involvement of the mitral valve. The slow onset of chronic aortic regurgitation allows the left ventricle to dilate to receive the regurgitant flow, compliance increases, and end-diastolic pressure remains low until cardiac failure develops. Aortic regurgitation remains asymptomatic until the patient develops heart failure. Once cardiac failure develops, the prognosis is markedly worsened but the natural history is difficult to predict; many patients may remain asymptomatic for many years. It is as important to avoid too early a valve replacement as it is to delay too long, when irreversible ventricular dysfunction may have developed. Deterioration is heralded by a changing ECG and an enlarging heart as detected by US or chest radiography. Figure 24.22 CMR in aortic regurgitation due to aortic dilatation. Left anterior oblique equivalent view shows the dilated aortic root and a jet of signal loss in the left ventricular outflow due to aortic regurgitation (arrow).

Aortic regurgitation Aortic regurgitation may result from disease of the cusps of the

aortic valve, or from disease of the aortic walls (Table 24.2). Cusp involvement may precipitate acute aortic regurgitation in bacterial endocarditis or, rarely, after trauma or as a result of acute avulsion of a leaflet into the left ventricle in aortic dissection. Chronic aortic regurgitation may be the result of congenital deformity (bicuspid aortic valve or subarterial ventricular septal defect) or rheumatic heart disease. Aneurysms of the ascending aorta may cause dilatation of the valve ring, leading to aortic regurgitation. Chronic aortic regurgitation The common causes of acquired aortic regurgitation are complications of a congenital abnormality of the valve or rheumatic heart disease.The congenital deformity is often a bicuspid valve in which the inadequately supported cusps allow progressive development of aortic regurgitation; there may be some degree of stenosis because of inadequate development of the valve commissures or dystropic calcification.

Table 24.2 CAUSES OF AORTIC REGURGITATION Degenerative Congenital (bicuspid or unicuspid valve) Rheumatic (with or without coexistent aortic stenosis) Subacute or acute bacterial endocarditis Aortic dissection Takayasu’s arteritis Marfan’s syndrome Ankylosing spondylitis Rheumatoid arthritis Ehlers–Danlos syndrome Syphilis Trauma Surgical or balloon valvoplasty

Chest radiography Chronic aortic regurgitation may produce dramatic appearances, with enlargement of the left ventricle which may be very severe. Enlargement parallels the severity of the condition and is one factor suggesting the need for surgical intervention. After surgery the reduction in heart size may be considerable (Fig. 24.23). The whole thoracic aorta may be moderately enlarged, unless there is aortitis or chronic aortic dissection, when aortic dilatation may be considerable. Valve calcification is not as marked but flecks of calcium may be seen. When mitral valve disease is present, left atrial enlargement may dominate the appearance, obscuring significant left ventricular enlargement. Other imaging techniques US is the main imaging technique for most patients with aortic valve disease. The complications of endocarditis such as paravalvar abscess may be better seen with CMR. Aortic root diseases such as annuloaortic ectasia or dissection are best investigated by CT angiography (CTA) or CMR20, and the severity of aortic regurgitation can be estimated from the extent of signal loss on CMR (see Fig. 24.22). In stable patients, CMR is the most accurate method for delineating almost all diseases of the thoracic aorta and simultaneous evaluation of the left ventricular function27,28. The evolving ECG-gated cardiac CT can provide similar information, and is a satisfactory alternative to CMR, particularly when the latter is contraindicated. There is a limited role for conventional aortography if good CMR or CT is available. As with aortic stenosis in adults, coronary angiography may be necessary in some patients with aortic regurgitation. In patients with large aneurysms of the aortic root, selective catheterization of the coronary arteries may be very difficult. In such patients, coronary CTA may provide all the necessary information in presurgical planning. Acute aortic regurgitation results from acute damage to the cusps, usually by bacterial endocarditis, very rarely by trauma, or even by spontaneous avulsion of the cusps. It may also result from acute dissecting aneurysms of the aorta. Severe aortic

479

480

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.23 Chest radiographs in pure aortic regurgitation. (A) Pre-operative radiograph. There is marked left ventricular enlargement. The aorta is slightly prominent but there is no post-stenotic dilatation. (B) Radiograph 8 years later after aortic valve replacement. There has been a marked reduction in the heart size.

regurgitation develops rapidly, with increasing left ventricular end-diastolic pressure and the acute onset of heart failure. In those patients who survive, left ventricular dilatation may occur and the condition may become compensated and chronic. The clinical diagnosis is usually obvious when the characteristic early diastolic murmur of aortic regurgitation is first detected in the course of an otherwise unexplained febrile illness, or in association with an acute attack of chest pain and clinical evidence of aortic dissection. In acute aortic regurgitation, ventricular compliance cannot compensate, and a relatively small regurgitant volume produces a very large rise in the ventricular end-diastolic pressure. This limits regurgitant flow as well as precipitating cardiac failure. The limited regurgitant flow means that the murmur may be soft or inaudible and the diagnosis missed. Chest radiography Depending on speed of onset, the heart may be normal in size or show only minor left ventricular enlargement. If there is left heart failure, there will be upper lobe blood diversion or overt pulmonary oedema. Acute aortic regurgitation is an important cause of pulmonary oedema in

a patient with a normal-sized heart. The aorta is usually unremarkable unless there is associated aortic disease causing dilatation as well as aortic regurgitation, as in Marfan’s syndrome. Other imaging techniques Acute aortic regurgitation is usually caused by infective endocarditis or aortic dissection. Both conditions can be well studied by 2D US. Transthoracic US can usually delineate aortic vegetations more than 2–3 mm in size. Transoesophageal US is very useful in both conditions, being more sensitive in detecting aortic vegetations than transthoracic US, particularly when the valve cusps are independently thickened by pre-existing disease. It is the technique of choice for displaying cusp perforation and aortic root abscess29. Abscess cavities may extend into the interventricular septum, the right side of the heart, the left atrium, or the wall of the ascending aorta. Abscesses may also be shown by CT or MRI. In aortic dissection, CTA and CMR are very valuable. Both can demonstrate the intimal flap, its relation to the coronary ostia, and any associated haemopericardium. CMR allows detection of the aortic regurgitation (see Fig. 24.22) as well as the morphology of the aortic valve (Fig. 24.24).

Figure 24.24 Calcified, stenotic bicuspid aortic valve. CMR, aortic valve view in (A) diastole and (B) systole. The calcified leaflets result in a signal void in diastolic frame (arrowheads). In systole the signal outlines the small opening of the stenotic valve orifice. (C) Left ventricular outflow tract view shows signal loss in the ascending aorta due to aortic stenosis. Note the doming of the aortic valve (arrowheads) and dilated ascending aorta (Ao). LA = Left atrium, RA = right atrium, RV = right ventricle. Compare to Fig. 24.21.

CHAPTER 24

PROSTHETIC CARDIAC VALVES Heart valve replacement with mechanical, donor, or bioprosthetic valves has resulted in an improved quality of life for patients with cardiac valvular disease. Since Harken et al38 in 1959 and Starr and Edwards22 in 1960 successfully implanted the first caged ball prosthesis in the mitral and aortic areas, continuous improvement in operative mortality, prosthetic design, and postoperative complication rates has been achieved23. Malfunctions, however, continue to occur. Five per cent of patients with aortic valve replacement die within 5 years, and 14% with mitral valve replacement with a St Jude prosthesis die due to intrinsic heart disease or to complications related to the prosthesis24. Two major types of artificial valves are currently available for both the aortic and atrioventricular positions. These are mechanical valves and bioprostheses. Either type should be of lasting structural durability, capable of successful implantation in the native valve annulus, and be chemically inert, free of thrombogenicity, and offer little resistance to blood flow25. Normal as well as abnormal function of the valve should be readily evaluated by minimally invasive or noninvasive diagnostic techniques25–27. The role of diagnostic imaging is complicated because there are over 40 types of cardiac valves, with different modes of action and complication rates, designed for use in the aortic, mitral, and tricuspid areas, as well as for vascular conduits. As each prosthesis has different inlet and outlet diameters and flow patterns, a knowledge of the inlet–outlet relationship is critical for proper identification of blood flow efficiency through the prosthesis. After the native valve has been removed and replaced with a tissue or mechanical prosthesis, complications are often found which are characteristic of specific models or classes of prosthesis.These include: paravalvar leak due to sewing ring dehiscence, occluder variance, or erosion; structural complications

•

NONISCHAEMIC ACQUIRED HEART DISEASE

such as strut fracture; pseudoaneurysm; thrombosis; and bland vegetations or tissue overgrowth. Infective endocarditis and primary valve failure with regurgitation, usually occurring in bioprostheses, are additional complications. Other complications to be considered are those of anticoagulation, embolism, and post-pericardiotomy syndrome. The primary method of evaluation of a prosthetic valve is with echocardiography. The role of the diagnostic radiologist in identifying the complications of prosthetic valves is based on the findings seen on plain chest radiography27–29, fluoroscopy, and CT16, and angiography30 can help to solve problems in individual cases. At present, the artefacts caused by almost all the prosthetic valves limit their evaluation with CMR. Valves have differing rates and types of early and late failure depending on their specific type. Knowledge of the anticipated mechanism of failure is essential for evaluation of a prosthetic valve. It is valuable to perform baseline 2D and Doppler echocardiography after the valve is implanted, for comparison with follow-up studies which may be undertaken when and if there is a suspicion of malfunction or altered cardiac status.The normal diagnostic parameters for many of the more widely used prostheses have been published and are available for guidance to the diagnostician26,31–33.

Cardiac valve prostheses in current use and of historical interest Central ball occluder valve This valve (Fig. 24.25) consists of a spherical occluder or poppet and a base ring located within the valve annulus. A metallic cage limits excursion of the occluder. This type of prosthesis is large and occupies enough space to reduce blood flow in those patients with a small left ventricular chamber or aortic root. Both thromboembolism and haemolysis are troublesome

Figure 24.25 Three Starr–Edwards prostheses in a 53 year old patient with rheumatic valvular disease. Chest radiograph in (A) frontal and (B) lateral views. Note the perpendicular orientation of the atrioventricular valves to the aortic valve. 1 = Aortic position, 2 = mitral position, 3 = tricuspid position.

481

482

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

complications of central ball occluder valves. Examples include the Starr–Edwards, Harken, Smelloff–Cutter, Braunwald–Cutter, and Magovern–Cromie valves. These valves have not been implanted for many years, and the central ball occluder valve will only rarely be seen today.

Central caged disc occluder valve Low profile valves have the advantage of minimal intrusion into a small valve orifice or cardiac chamber. However, over time the disc edge erodes and valvar regurgitation will develop. Examples include the Beall and Starr–Edwards 6500 series.

Eccentric monocuspid disc valve Until recently, the eccentric monocuspid disc was the most commonly used type of mechanical valve. Its advantages include a large cross-sectional area in the open position, so that flow is less inhibited than with other valves. As a result, haemolysis is trivial. Specific models of one monocuspid valve, the Bjork–Shiley concavoconvex occluder valve, developed an alarming number of outlet strut fractures and disc embolization (see page 487)34. This problem led the manufacturers to recall the unimplanted valves, including the Bjork–Shiley, Omniscience, and Medtronic–Hall valves.

Bileaflet disc valve This valve (Fig. 24.26) consists of two semilunar discs which pivot into position, eliminating supporting struts. There is good hydraulic function with a low incidence of thromboembolism. They are in common use today and include the popular St Jude33 and the Carbomedics and Sorin bicarbon valves (Figs. 24.26, 24.27).

Bioprosthesis In common use are the Carpentier–Edwards porcine prosthesis (Fig. 24.28), the Hancock porcine bioprosthesis, and the Ionescu–Shiley bovine pericardial bioprosthesis (Fig. 24.29). All have metallic components that permit recognition on chest radiographs. Recent introductions include the Stentless Porcine Xenografts for the aortic position. Since

Figure 24.27 Cine-fluoroscopy of St Jude prosthesis in aortic position. Tangential view in (A) open and (B) closed position. The leaflets form an angle of 35 degrees in (A) and 125 degrees in (B), indicating stenosis of the prosthesis in this patient, as the normal angle between the open leaflets is about 10 degrees. Thrombosis was demonstrated at surgery (see also Fig. 24.31C). In the open position, there are three streams of blood, one between the almost parallel discs and one on the outer side of each disc.

Figure 24.26 Bileaflet valve. Implanted Duramedic valve (replaced mitral valve) in the open position.

the stent decreases the effective valve area, and causes more stress on the leaflets, the stentless valves are considered to have more physiological flow, and are increasingly used in patients with small aortic roots. This group of prostheses includes the Toronta SPV stentless valve34, the Edwards stentless valve35, and the Medtronic Freestyle valve36. Their long-term prosthetic durability and survival are not yet established. The major advantage of the bioprosthetic valves is that anticoagulants can be discontinued after several months of therapy. These valves are used preferentially in women of child-bearing age who plan to have further children, in older patients, and in very young patients, and particularly those patients for whom anticoagulation is a serious risk.

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.28 Carpentier–Edwards prosthesis. (A) Photography from outlet aspect. The commissures of the heterograft aortic valve are supported by three struts. (B,C) Two frames from cine-fluoroscopy, showing the usual displacement of the base support during the cardiac cycle. The estimated displacement in this patient is less than 2 degrees.

Figure 24.29 Heterograft valves. (A) Hancock porcine valves—only the supporting ring is radio-opaque. (B) Carpentier–Edwards porcine bioprosthesis with metal frame. (C) Ionescu–Shiley calf pericardial tissue valve.

Cardiac valve reconstruction and annuloplasty Surgical valvuloplasty for treatment of mitral insufficiency sometimes requires the use of a metallic ring (Carpentier– Edwards or Duran ring37) to reduce the diameter of a patulous tricuspid or mitral valve annulus. A flexible annuloplasty ring has recently been introduced for this purpose. The radioopaque ring is inserted in the annulus of the mitral valve38 (Fig. 24.30).

Diagnostic imaging following valve implantation Chest radiography During the immediate postoperative period, interpretation of valve position and integrity will usually be based on the frontal bedside chest radiograph. In the frontal projection, the aortic valve prosthesis is oriented in partial profile along the direction of flow as determined by the occluder position in a mechanical aortic valve.The mitral valve is usually more vertically oriented and is more likely than the aortic valve to be visualized en face. The direction of flow in the case of the mitral valve is towards the cardiac apex28.

Perhaps the most useful role for plain chest radiography is the identification of the signs of increased pulmonary venous pressure due to new valvar regurgitation secondary to paravalvar leak29. Plain chest radiography is particularly valuable in the identification of strut fracture, occluder embolization39,40, or calcification of a degenerated bioprosthesis41,42. If malfunction occurs, echocardiography is the study of choice for characterization of the location and extent of regurgitant flow.

Fluoroscopic evaluation of selected prosthetic valves Fluoroscopic examination with videotape or digital recording is helpful in analysing the integrity of the radio-opaque components of prosthetic valves (Fig. 24.31). However, in practice fluoroscopy is reserved for analysis of older mechanical valves and in selected cases when echocardiography does not clearly answer the question addressed. Some important features to assess are: motion of the occluder and base ring, integrity of the opaque struts, and calcification of tissue valves and surrounding supporting structures.

483

484

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.30 Mitral valve annuloplasty in a patient with severe mitral insufficiency. (A) Carpentier–Edwards ring has been inserted in the mitral annulus. (B) Sculptor flexible annuloplastic ring in mitral annulus in another patient with mitral insufficiency. (C) Lateral view of chest shows Duran band for reconstruction of an incompetent mitral valve.

Calcifications of the aorta, coronary arteries, and head and neck vessels can be studied at the same time. Since normal motion of the base ring differs for each valve model and for each patient, a baseline examination is helpful for future reference. The normal and abnormal ranges of motion of many of the valves has been described, including the St Jude25,43, the Bjork–Shiley30,44,45, Starr–Edwards, Medtronic–Hall, and Beall valves. To visualize an aortic central occluding valve in profile, left anterior oblique views with cranial or caudal angulation are necessary.The right anterior oblique view with cranio/caudal angulation is appropriate for examination of the mitral valve. In the opposite oblique position, the base ring is positioned in an en face projection, so that the occluder or calcifications can be observed through the base ring. Carm fluoroscopy is helpful for obtaining a completely en face projection, by angling the tube into the sagittal as well as the axial plane46. The position of the mitral valve annulus in relation to the base ring can be appreciated by locating the fat plane of the atrioventricular groove. A partially separated mitral valve prosthesis may lie above the fat line of the atrioventricular groove within the left atrium47, or below the fat line within the left ventricle48,49. The range of motion of the base ring of a prosthetic aortic valve is generally smaller than that of a mitral valve, probably due to the greater variation in residual tissue in the mitral

annulus after surgery. The normal range of motion of the aortic sewing ring has been described as up to 9 degrees (6–9 degrees). The normal range of motion of the mitral prosthesis is 9–12 degrees31,50.The Bjork–Shiley prosthesis, when in the open position, forms an angle of 60 degrees to the axis of the base ring. An aperture of 45 degrees is considered appropriate for this valve. Bileaflet prosthesis (e.g. St Jude) The St Jude valve51 is a low-profile central-flow device consisting of two semicircular discs that pivot in position without supporting struts. Blood flow occurs in three streams: a central stream between the discs and two outer streams on the outer side of each disc.The orifice ring is impregnated with tungsten to provide faint radio-opacity. Fluoroscopy with craniocaudal angulation appears to be most useful to study the St Jude aortic valve (see Fig. 24.27). Placing the patient in a slight left anterior oblique position provides an en face view31,33. CT is useful in demonstrating this valve to advantage because of the high temporal resolution of this technique (Fig. 24.32). When a St Jude prosthesis is observed en face, the base ring appears as a circle and the leaflets in the open position as two almost parallel lines. In the tangential view, the leaflets in open position appear to form an angle of approximately 10 degrees. The leaflets make an angle of 120 degrees in the closed position for a prosthesis of 19–25 mm; the closing angle for valves of 27–33 mm is 130 degrees (see Figs 24.27, 24.32).

CHAPTER 24

Figure 24.31 Cine-fluoroscopic findings with commonly implanted prosthetic valves. (A) Starr–Edwards aortic valve in the closed position during diastole and a Beall mitral valve in the open position (upper panel). The lower panel shows the Starr–Edwards valve in the open position during systole and the mitral central occluding the disc Beall valve in the closed position. (B) Upper panel shows a Bjork–Shiley valve in the closed mitral position. Lower panel shows the disc to be 50 degrees from the horizontal baseline (ring) during systole in the open position. (C) In the upper panel, the St Jude aortic bileaflet valve is in the closed position during diastole. In the lower panel, it is in the open position during systole, permitting two blood streams at the outer sides of the leaflets and one stream between the leaflets.

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.32 Ultrafast CT in the axial plane at the level of the aortic root demonstrates a St Jude bileaflet valve (between arrows which define the closure line between the two leaflets) in the closed position. (Courtesy of William Stanford, MD.)

models. Since there are few Beall valves in place today, this problem is an historical one.

Porcine bioprosthesis The major problem with porcine bioprostheses is their poor durability. Cusp tears, degeneration, perforation, fibrosis, and

Tissue valves Fluoroscopy is of limited value in detecting abnormalities of tissue valve function, but abnormal tilting of the radio-opaque base ring will suggest dehiscence, and calcification of the tissue components will indicate degeneration25,26.

Complications of prosthetic valves Strut fracture and poppet embolization Strut fractures have been reported in some types of mechanical valves. During the past decade, fractures of the concavoconvex Bjork–Shiley valve affecting the outlet struts have been reported in a highly significant number of patients, particularly those with the larger size mitral values. Plain chest radiography using microfocus, fluoroscopy, and CT have been useful in identifying fractures of these radioopaque components (Figs 24.33, 24.34). The high incidence of disc embolization with the Beall valve is related to severe grooving and wear occurring in the Teflon® disc of the earlier

Figure 24.33 Broken struts. Cine-fluoroscopy demonstrates two broken struts in this DeBakey aortic valve. (Reproduced with permission from Zumbro et al39.)

485

486

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Infection typically develops in the coronary ostia, the ascending aorta, and the aortic annulus, related to high-pressure regurgitant jets of blood which damage valve attachments and cause development of local abscesses (Fig. 24.35). Pannus and vegetations obstruct flow through the valve orifice or limit occluder motion (Fig. 24.36). Sinus of Valsalva and perivalvar pseudoaneurysms may develop (Fig. 24.37)53. Paravalvar insufficiency occurs as tissue fragments and sutures loosen with infection. Two-dimensional echocardiography is useful in demonstrating the site, form, and size of vegetations or thrombi on a tissue valve prosthesis. However, transoesophageal echocardiography (TEE), CT, and CMR have also been useful in identifying vegetations, particularly pseudoaneurysms53. Contrast angiography is also useful for detecting pseudoaneurysm and evidence of infection (Fig. 24.37). Cine-fluoroscopy has been disappointing for detecting vegetations, but if the vegetation is large enough to interfere with valve motion, decreased excursion of the occluder may be recognized53.

Valve regurgitation

Figure 24.34 Bjork–Shiley disc embolized following strut fracture. The strut is in the proximal left common iliac artery (lower arrow). The disc is in the abdominal aorta at the T12–L1 interspace (upper arrow). (Reproduced with kind permission of Springer Science and Business Media, from Morse et al47.)

calcification appear on about the fifth postoperative year, and by the 10th year 20% have failed and require re-implantation. Fortunately, bioprostheses seldom require urgent replacement, so that replacement can be carried out on an elective basis. Fracture of the stent is rare in this type of prosthesis.

Minor regurgitation is found normally with many mechanical valves, but major regurgitation usually occurs as a result of sewing ring dehiscence, thrombus, disc wear, and sticking, cocking, or vegetations of the valve leaflets. Perivalvar leak occurs more frequently after mechanical valve implantation than with bioprostheses. Dehiscence with valve detachment may occur in a heavily calcified annulus or when there is underlying defective collagen, as in Marfan’s syndrome39,54. Perivalvar regurgitation or valve dysfunction should be suspected when sudden cardiac enlargement and pulmonary oedema occur: a regurgitant murmur will occur in 80% of patients with regurgitation. In the mitral position, normal motion of the prosthesis is anterior and posterior along the long axis of the left ventricle, and side to side along the vertical axis. Strain on the suture line is greatest at the medial and lateral aspects of the sewing ring. If a suture breaks in one of these locations, the adjacent sutures suffer additional strain, eventually causing complete dehiscence, leading to detachment and possible embolization. The incidence of leakage

Infective endocarditis Prosthetic valve endocarditis is an infrequent complication of cardiac valve replacement. The overall incidence of infective endocarditis ranges from 0.9 to 4.4%, and is most frequent within 6 months of valve implantation. Potential sources of infection include contaminated blood products, surgical field contamination, and pre-existing infection. Postoperative sources include indwelling catheters, endotracheal tubes, and pacemakers. Organisms most frequently involved are coagulase-negative staphylococci, Serratia marcescens, streptococci, Candida albicans, and Staphylococcus aureus. Diphtheroids are seen more frequently in early infective endocarditis, and there is a lower incidence of Gramnegative bacilli and fungal infections52. Pre-operative factors that predispose to the development of infection are previous endocarditis and post-surgical wound infection.

Figure 24.35 Transoesophageal echocardiography through the aortic valve in a patient with aortic prosthesis. An abscess (AB) is seen involving the aorta at the site of the noncoronary cusp. LA = Left atrium, LV = left ventricle, PAV = prosthetic aortic valve, RA = right atrium. (Courtesy of N Nanda, MD.)

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

sensitive than fluoroscopy in identifying dehiscence and other causes of prosthetic regurgitation.

Thromboembolism

Figure 24.36 Transoesophageal echocardiographic view in a patient with aortic valve endocarditis. There is a marked thickening of the aortic leaflets with a vegetation involving the aortic annulus at its arterial side. LA = Left atrium, LV = left ventricle, PAV = prosthetic aortic valve, RV = right ventricle, V = vegetation. (Courtesy of N. Nanda, MD.)

Thrombosis in the valve orifice will reduce the excursion of the occluding ball, disc, or leaflet. Prosthetic mitral valve thrombosis is more common than aortic valve thrombosis. The frequency of thrombosis has decreased markedly during the past few years. With the early prostheses, the incidence of thromboembolism was 24–37%. Since the introduction of cloth-covered prostheses, the incidence has decreased to 3–5%. The risk of embolism diminishes with time after implantation and is similar for single or multiple valve replacements55. The incidence of St Jude prosthetic valve thrombosis is as low as 0.2% per patient per year56. Fluoroscopy is helpful in demonstrating the abnormal motion of the occluder due to thrombus (see Fig. 24.27). M-mode and 2D imaging are useful in characterizing the thrombus, as are CT and MRI if the thrombus is large enough. Obstruction may occur because of spontaneous thrombosis, despite adequate anticoagulation. The patient will present with a low-output state, and the normal circulatory sounds of the valve are not audible. Since this phenomenon is a surgical emergency, immediate valve replacement or intervention by catheter remobilization and intracardiac thrombolysis, is required57.

Effect of magnetic resonance on the prosthetic valve

Figure 24.37 Mycotic aneurysm in the ascending aorta in a patient with infective endocarditis. There is a Bjork–Shiley valve in the aortic position. (Reproduced with kind permission of Springer Science and Business Media, from Morse et al47.)

produced by dehiscence is near to zero in noninfected valves. Factors which predispose to dehiscence are infective endocarditis and calcium in the annulus of the recipient valve. Rocking of the mitral valve radio-opaque base ring by more than 9–12 degrees and of the aortic valve by more than 10 degrees is highly suggestive of dehiscence. However, there must be at least a 40% continuous suture interruption around the circumference of the base ring before abnormal tilting motion will occur; thus, fluoroscopy without contrast medium is helpful only in advanced cases. Echocardiography is more

In an in vitro study, nine different mechanical bioprosthetic valves were studied with medium- and high-field MRI magnets. At 1.5T, six mechanical valves deflected 0.25–3.0 degrees. The largest deflections were observed in several Starr–Edwards models, which are made of dense cobalt–nickel alloy (Stellite)58. Other investigators found no significant temperature increase in a series of commonly used valves, or evidence of significant image distortion. A recent in vitro study using a 4.5-T system revealed highest deflection with the Duran Mitral annuloplasty ring59. CMR demonstration of the valves yields no significant information about valve condition. The diagnostic value of CMR in detecting regurgitation in prosthetic valves was compared with the transoesophageal colour Doppler echocardiography study by Deutsch et al60. They found 96% agreement between the methods in distinguishing physiological and pathological regurgitation across a prosthetic valve. The degree of regurgitation was detected by a signal void emanating from the prosthesis (Fig. 24.38). However, with many of the metallic prostheses, the artefacts from the metallic cage can obscure small paravalvar leaks. In addition, CMR is well suited to study the cardiac function that may be affected by prosthetic valve dysfunction61.

DISEASES OF THE HEART MUSCLE (CARDIOMYOPATHIES) Cardiomyopathies are a heterogeneous group of disorders in which the dominant feature is diseased cardiac muscle itself,

487

488

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.38 Prosthetic valve insufficiency. Gradient-echo CMR in transverse section, showing a prosthetic valve in the mitral position (curved arrow) with paravalvar leakage. The signal void from eccentric regurgitation (wide short arrow) originates from the septal insertion of the prosthesis. LA = Left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle. (Reprinted from Deutsch H J, Bachmann R, Sechtem U, Curtius J M, Jungehulsing M, Schicha H, Hilger H H 1992 Regurgitant flow in cardiac valve prosthesis: Diagnostic value of gradient echo nuclear magnetic resonance imaging in reference to transesophageal two-dimensional color Doppler echocardiography. J Am Coll Cardiol 19:1500–1507 with permission from The American College of Cardiology Foundation.)

often leading to heart failure. By definition, this group of diseases is not the result of congenital, valvular, pericardial, or hypertensive disease. Since the most common cause of congestive heart failure is coronary artery disease, these disorders must also be distinguished from the sequela of myocardial ischaemia. The most widely recognized classification of cardiomyopathy is by the World Health Organization (WHO) and the International Society and Federation of Cardiology (ISFC)62. In this scheme, the cardiomyopathies are classified on the basis of their pathophysiological features. When the muscle disease is part of a generalized or particular cardiac disorder, they are referred to as ‘specific cardiomyopathies’ (Table 24.3).

Dilated cardiomyopathy Also known as congestive cardiomyopathy (CCM), dilated cardiomyopathy (DCM) is the most common type of cardiomyopathy. There is global impairment of contractility with Table 24.3 SPECIFIC CARDIOMYOPATHIES Ischaemic cardiomyopathy Valvular cardiomyopathy Hypertensive cardiomyopathy Inflammatory cardiomyopathy Metabolic cardiomyopathy General systemic disease Muscular dystrophies Neuromuscular disorders Sensitivity and toxic reactions Peripartum cardiomyopathy

dilatation of all chambers, particularly the left ventricle (Fig. 24.39), presenting clinically as congestive cardiac failure. Sometimes symptoms may be noticed after an acute influenza-like illness, hinting that a myocarditis may have been the underlying cause. Chest pain is very unusual. There is a familial form63. Physical signs of heart failure may be accompanied by mitral regurgitation due to dilatation of the mitral ring.There may be a pericardial effusion but tamponade is rare. Mimics of DCM are the specific myopathies, e.g. alcoholic heart disease; diffuse, silent ischaemic disease; and critical aortic stenosis, when the murmur may be inaudible. The course of the illness is variable, often with a good response to initial drug therapy. Sooner or later, periods of relapse end in death from low cardiac output; DCM is the most common reason for cardiac transplantation. At necropsy the left ventricle is found to be dilated, its wall is thickened (to a variable degree), and its trabeculae are effaced.There may be clot within the ventricle. Other cardiac chambers are usually dilated to a variable degree. Histological examination reveals cardiac muscle fibre hypertrophy and interstitial fibrosis, appearances that are not specific and may be seen in any form of left ventricular failure.

Imaging The plain chest radiograph is almost always abnormal at clinical presentation and most often the only abnormality is cardiac enlargement. The configuration of the heart may be solely left ventricular, or all chambers may be enlarged, although there is usually left ventricular prominence. In the untreated patient, upper lobe blood diversion and a variable degree of pulmonary oedema are usually seen. A good response to treatment is indicated by clearing of the pulmonary oedema and a reduction of the heart size (Fig. 24.40). Radionuclide ventriculography, echocardiography, and MRI can all show the dilated left ventricle and its reduced systolic contractility. Mild regional variations in contractility may mimic the irregular distribution of ischaemic myocardial disease, or the ventricle may exhibit global hypokinesis. CMR may be a particularly useful noninvasive test to distinguish DCM from ischaemic heart disease. Regional wall thinning may be obvious in short axis views. Use of MR contrast agents is particularly useful in recognizing the presence of infarcted myocardium, from both first-pass perfusion images and delayed enhancement of the infarcted segments. It is suggested that contrast-enhanced CMR is currently the best imaging method to categorize various forms of cardiomyopathies and more specifically, to distinguish the far more common ischaemic myocardial disease (Fig 24.41).When the DCM is secondary to active myocarditis, there may be delayed enhancement due to active inflammation. Such enhancement may be patchy, and can mimic infarcted myocardium. However, the perfusion images will be normal, and often there will be no focal thinning of the myocardium. The right ventricle is usually similarly affected but to a less obvious degree due to its lower working pressure—a normal right ventricle should raise suspicions of global left ventricular impairment due to multiple infarctions from ischaemic heart disease. Severe right ventricular dilatation is a poor prognostic sign64.

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.39 Left ventriculography in congestive cardiomyopathy (same patient as in Fig. 24.40). (A) Systole and (B) diastole corresponding to Fig. 24.40A (before treatment). In (A) and (B) the ventricle is markedly dilated with little change in size from diastole to systole. The shape is slightly globular and the trabeculae are effaced. There was mitral incompetence, demonstrated by opacification of the left atrium (LA). (C) Systole and (D) diastole corresponding to Fig. 24.40B (after treatment). Although much smaller, the ventricle remains significantly dilated, even though the heart appears normal on plain radiography. Mitral incompetence is no longer present. At end systole (C), the left ventricle is much smaller than in (A).

Figure 24.40 Congestive cardiomyopathy (same patient as Fig. 24.39). (A) Before treatment. The heart is large with some prominence of the left ventricle, though all chambers are enlarged. There is a small right pleural effusion, redistribution of blood to the upper zones, and slight pulmonary oedema. (B) After treatment. The heart has reverted to normal size and the lungs are clear.

489

490

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.41 Differentiation of (A,B) dilated and (C–E) ischaemic cardiomyopathy using CMR. (A) Four-chamber projection shows dilatation of all four chambers with greater involvement of the left ventricle (LV). (B) Same patient, post gadolinium administration delayed image, two-chamber projection. There is no regional hyperenhancement of the LV myocardium. (C) Short axis, mid ventricular level precontrast gradient-echo image. (D,E) Post contrast medium, delayed inversion recovery sequence. Before injection of contrast medium, LV wall thinning involving the posterior wall is seen. Post contrast medium delayed images show extensive hyperenhancement involving anterior, apical, septal, and posterior segments (arrowheads). LA = Left atrium, RA = right atrium, RV = right ventricle.

In all patients suspected of having DCM, the aortic valve must be examined in detail in order to exclude end-stage aortic stenosis. The two conditions may have very similar clinical presentations. It is very important to identify this presentation of aortic stenosis, because the response to valve replacement is frequently excellent. Echocardiography and CMR are very useful in recognizing the aortic stenosis.

Hypertrophic cardiomyopathy Hypertrophic cardiomyopathy (HCM), which is often familial and much less common than DCM, is characterized by excessive hypertrophy of the left ventricular myocardium, the myocytes showing a very disorganized orientation when compared to the regular alignment of ordinary left ventricular hypertrophy. HCM must be distinguished from ventricular hypertrophy resulting from hypertension and left ventricular outflow obstructions, including aortic stenosis. In a small proportion of patients, perhaps 25%65, the upper interventricular septum is most affected, resulting in a subvalvular obstruction. This form is also referred to as asymmetric septal hypertrophy or idiopathic hypertrophic subaortic stenosis (IHSS). However, the ventricular hypertrophy is symmetrical in the majority of patients with HCM; the right ventricle may also be involved. The myocardium may outgrow its blood supply, cause anginal pain, and show diffuse fibrosis. There may rarely be associated myocardial infarction. Ventricular arrhythmia is common and death may be the mode of presentation. Similar pathological

changes may be seen in the hearts of patients with Noonan’s syndrome and Friedreich’s ataxia. A rare variant affects only the apex of the left ventricle66. In most cases, contractility is good or even hyperdynamic, and together with the hypertrophy leads to a very small endsystolic left ventricular volume. As a consequence, the chordae of the mitral valve may become slack, allowing the tip of the anterior leaflet to float towards the septum: once it strays into the left ventricular outflow tract it is picked up by the flow of blood and slammed against the interventricular septum, rather like a sail caught by the wind. In doing so it partially obstructs the subaortic region, causing an end-systolic gradient and giving further stimulus to left ventricular hypertrophy. This mechanism is called systolic anterior motion of the mitral valve (SAM). This abnormality can also contribute to the left ventricular outflow obstruction. The distortion of the mitral leaflet also causes mitral regurgitation, usually mild. Occasionally there is a preterminal phase when the ventricle dilates; SAM is then lost and the picture more closely resembles DCM.

Imaging The plain chest radiographic appearances67 may vary from a heart that is entirely normal in size and shape to varying degrees of left ventricular shape, sometimes with unusual bulges, and the left atrium is sometimes enlarged. The dilated phase will resemble DCM. The lungs may reflect left atrial hypertension. The cardiac silhouette may remain unchanged

CHAPTER 24

for years, but it grows in size with clinical deterioration, often associated with the onset of atrial fibrillation68. Echocardiography has played a major part in elucidating the pathophysiology and natural history of HCM. The condition, and particularly SAM, is well displayed by cross-sectional echocardiography (Fig. 24.42).

•

NONISCHAEMIC ACQUIRED HEART DISEASE

the left ventricle is irritable and catheterization may produce arrhythmias. Coronary angiography usually shows little or no arterial disease: the vessels are typically large and the septal perforators may show compression during left ventricular systole. Careful pressure measurement with an end-hole catheter shows the subaortic gradient together with any valvar aortic stenosis that may be coincidental or causative. On left ventriculography, the characteristic angiographic appearances depend on the classical distribution of hypertrophy. The left ventricular cavity (Fig. 24.44) is usually of normal size in diastole, but the shape varies from virtually normal to the characteristic curved shape, almost like a banana, when gross septal hypertrophy is present. Left ventricular trabeculation is usually coarser than normal and the free wall of the left ventricle may be obviously thickened.

CMR can demonstrate the pattern of hypertrophy, and assess left ventricle mass and mitral regurgitation (Fig. 24.43). Recent investigations have identified a subset of patients in whom echocardiography fails to demonstrate left ventricular hypertrophy or more severe thickening of the anterolateral wall (the latter is associated with a high risk for sudden cardiac death)69. Serial measurement of left ventricle mass by CMR is the best technique to follow the natural history of the disease. Postcontrast CMR may demonstrate patchy delayed enhancement, which is also associated with increased risk for sudden cardiac death (Fig. 24.43).

Restrictive cardiomyopathy70

Cardiac catheterization and angiography usually add little to the diagnosis or assessment and can be dangerous because

Heart failure due to impaired contractility of the left ventricle is easily understood. Less easy, but often as important, is failure of the ventricle to fill properly, initially hindering the

RV

RV Se

Se

LV

Ao

LV LA

LA

Figure 24.42 Hypertrophic cardiomyopathy. (A) M-mode echocardiogram, showing systolic anterior motion (SAM) of mitral valve apparatus. In addition, the diastole closure rate of the mitral valve is very low, so that it remains in apposition to the septum throughout diastole. IVS = interventricular septum, PW = posterior wall. (B ,C) Cross-sectional echocardiogram, long-axis parasternal view. (B) Diastole—the increase in septal thickness is apparent. (C) Systole—the anterior leaflet of the mitral valve has moved forward to touch the septum and obstruct the outflow tract.

Figure 24.43 Hypertrophic cardiomyopathy. CMR, left ventricular outflow tract view of (A) systolic and (B) diastolic phases show massive hypertrophy of the left ventricle, with near complete obliteration of the left ventricle (LV) cavity in systole (A). Ao = Ascending aorta, LA = left atrium.

491

492

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.44 Left ventriculography in hypertrophic cardiomyopathy. Right anterior oblique (RAO) projection. (A) Diastole. The left ventricular cavity is of normal size and has a curved axis and coarse trabeculations. Some left ventricular wall thickening is present. There is mitral regurgitation into the left atrium (la). (B) Systole. There is almost complete emptying of the left ventricle with some contrast medium remaining under the aortic valve. The mid cavity filling defect of the posterior papillary muscle is seen (arrows). The indentation by the hypertrophic upper septum can be seen (arrowheads). Contrast medium is trapped at the apex of the left ventricle. Mitral incompetence is present, as shown by opacification of the left atrium (la).

expected increase in stroke volume with exercise and eventually reducing the resting cardiac output. Elements of diastolic dysfunction often coexist with systolic dysfunction as mentioned above, notably in HCM, hypertension, and ischaemic heart disease. This section will discuss those pathologies in which diastolic dysfunction is the predominant mechanism. In general terms there are two situations. First, the ventricle(s) may have stiff walls which fail to relax and require high filling pressures in order to achieve a reasonable stroke volume; this is typical of the myocardial infiltrative conditions. Alternatively, there may be a fixed maximum volume for the ventricle, which can easily be achieved with normal filling pressure, but beyond which there is only an abrupt rise in pressure but no further expansion; this is very similar to constrictive pericarditis. The differentiation between the two mechanisms is often blurred and they may well coexist with systolic impairment. Classification is therefore not as easy as for DCM and its differential. Mitral inflow Doppler studies during echocardiography or phase-contrast CMR underline the difference, with volume restriction giving rapid, early diastolic filling and the stiffened (low compliance) ventricles showing low velocity, protracted filling with accentuation of the atrial contraction wave (A wave). However, a mixture of the two can yield a spuriously normal wave form. The great majority of patients with restrictive cardiomyopathy have a ‘stiff ’ (diminished compliance) left ventricle.This is frequently seen in patients with hypertension, myocardial ischaemia, and advanced stages of HCM. However, these groups of patients are not labelled as restrictive cardiomyopathy. The common cause of the latter is from myocardial infiltration by a systemic disease process.

Amyloid heart disease Cardiac amyloidosis is a classic example of low compliance restrictive cardiomyopathy. Different forms of amyloidosis are

classified by the composition of the amyloid fibril, which infiltrates the involved organ. AL amyloidosis (previously known as primary amyloidosis) is derived from light-chain immunoglobulin produced by the monoclonal plasma cells. This is the most severe form, and the most common type encountered in the USA, and commonly involves the kidneys, liver, heart, and peripheral nerves. Familial amyloidosis is an uncommon autosomal dominant disease resulting from the production of an unstable variant of the serum protein transerythretin (ATTR amyloidosis), causing cardiac and/or neurological dysfunction. Senile systemic amyloidosis is the result of the deposition of wild-type transerythretin and is almost exclusively limited to cardiac involvement. This disease is becoming more prevalent, due to the ageing of the general population. Cardiac involvement in all these forms not only indicates a poor prognosis, but also predicts poor tolerance to high dose chemotherapy and stem cell transplantation71. Therefore, recognizing cardiac involvement in patients with amyloidosis is very important. The condition presents clinically in an intractable form of lowoutput cardiac failure, often with low pulse and blood pressures. There are usually no murmurs. The ECG is often of low voltage.The process leads to a combination of systolic dysfunction in which the heart does not pump properly, and diastolic dysfunction in which the stiff ventricles do not allow proper filling. Echocardiography is considered the noninvasive test of choice in the diagnosis of amyloidosis in a selected subset of patients with low output heart failure, but lacks specificity when co-morbid conditions such as hypertension are present. Gadolinium-enhanced CMR may give more specific findings, including decreased T1 of the myocardium and qualitative global and subendocardial delayed enhancement72. Both echocardiography and CMR can demonstrate systolic and diastolic dysfunction as well as any related atrioventricular valve regurgitation. CMR can also be helpful in excluding the presence of constrictive pericarditis.

CHAPTER 24

Sarcoidosis Sarcoid granulomas in the myocardium can give rise to cardiac arrhythmia as well as restrictive cardiomyopathy. Involvement of the papillary muscles may give rise to mitral regurgitation. Pericardium is uncommonly involved. Only 5% of patients with systemic sarcoid have clinical evidence of cardiac involvement, but at autopsy 20–30% have histological evidence of granulomas in the heart. Myocardial biopsy, the most specific test, can be falsely negative, due to patchy involvement of the myocardium. CMR can identify sarcoid granulomas as high signal areas in T2-weighted images with fat suppression, or as areas of hyperenhancement after administration of gadolinium (Fig. 24.45). These findings are nonspecific, and can be observed in many other inflammatory states. However, in patients with known sarcoidosis, these findings may help in identifying the site for endomyocardial biopsy and/or monitoring results of therapy.

Haemochromatosis Haemochromatosis can be a primary, inherited autosomal recessive disease, in which iron deposition occurs in the liver, heart, and pancreas. Secondary haemochromatosis results commonly from repeated blood transfusions, and less commonly from long-term haemodialysis and alcohol abuse. This form involves the liver and spleen, while iron deposition in primary haemochromatosis spares the spleen. Cardiac involvement results in diastolic dysfunction and in advanced stages systolic dysfunction as well, resembling DCM. Since iron reduces tissue T1- and T2-relaxation rates, haemochromatosis can be diagnosed by CMR. The amount of signal decrease in T2-weighted images correlates with iron level in the tissues, and CMR can be used to monitor the results of treatment of haemochromatosis73.

Volume-restricted cardiomyopathy Idiopathic volume-restricted cardiomyopathy is mainly a disease of children74 and includes some contracted-type endocardial fibro-elastosis with no other associated pathology, but is

•

NONISCHAEMIC ACQUIRED HEART DISEASE

mostly a mixed group of dysplastic ventricles. Systolic contractility is often near normal and stiffness may be minimal. Chest radiographs may show a normal heart size but dilatation of the atria often causes an enlarged cardiac shadow. Pulmonary congestion is usual and there may be severe secondary pulmonary hypertension. US and MRI typically show large atria with normal sized ventricles, and the mitral inflow Doppler shows rapid early diastolic filling typical of volume restriction. In the occasional adult case, much of this is shared with constrictive pericarditis. Endocardial irregularity and bizarre ventricular modelling may be easily seen on US but is often best visualized by angiography. Volume-restricted cardiomyopathy in the adult is usually due to endomyocardial fibrosis (EMF)75. This condition predominantly affects the peoples of equatorial Africa, southern India, Sri Lanka, and Brazil, but it is also recognized in caucasians (Loeffler’s endocarditis) in Europe and North America. It is characterized by the deposition of a layer of fibrous tissue on the ventricular endocardium, beginning at the apex and spreading proximally to encroach on the atrioventricular valve. Either ventricle or both may be affected.Thrombus may deposit on the endocardium and become organized, resembling a ventricular tumour. The fibrosis does not initially severely affect left ventricular contractile function and the ventricles empty reasonably well, but the fibrosis prevents them from expanding fully in diastole. Valve involvement may lead to atrioventricular valve incompetence. In non-caucasians, the pathological process usually involves the right heart, either alone or more severely than the left, leading to right heart failure with the murmur of tricuspid incompetence. In caucasians, when the condition is often associated with present or past evidence of eosinophilia23, the left heart and mitral valve tend to be more severely involved. Imaging in endomyocardial fibrosis In right heart involvement, on the plain chest radiograph the heart is enlarged with a right heart or globular configuration. The right atrium may be very large. The lungs may be clear if there is no left heart involvement. Pleural effusions may be seen. Caucasian cases may have a normal heart silhouette or some left ventricular and left atrial enlargement and pulmonary congestion. Very rarely, curvilinear calcification due to endocardial calcification may be seen.

Left ventricular noncompaction

Figure 24.45 Cardiac sarcoid. CMR, short axis view. Post gadolinium delayed image demonstrating focal area of myocardial enhancement (arrow) in a patient with known pulmonary sarcoidosis. LV = Left ventricle, RV = right ventricle.

Left ventricular noncompaction (LVNC) is a rare form of congenital cardiomyopathy, resulting in arrest of normal intrauterine developmental progression of the loose myocardial fibre network. Two forms of LVNC are recognized: a nonisolated form associated with other congenital heart diseases, such as ventricular septal defect or hypoplastic left ventricle; and an isolated form that is often undetected and more common in adults. The latter type is becoming increasingly recognized in adults76–78. The involved myocardium (more often the left ventricle) has a ‘spongy’ appearance with more trabeculations and deep intertrabecular recesses. The latter tend to have sludging of the intracavitary blood and form

493

494

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

thrombi. The disorder is often familial. The clinical presentation may range from the absence of symptoms to cardiac arrhythmias, embolic events, and severe heart failure. Echocardiographic criteria for diagnosis of LVNC are based upon trabeculations in the ventricular apex and thickness of the left ventricular free wall in end diastole. CT and CMR can demonstrate similar findings (Fig. 24.46). Some reports suggest that CMR may be more sensitive in the diagnosis of

LVNC than echocardiography78. CMR appearances range from increased left ventricular trabeculations, to abnormal wall motion, to the presence of two myocardial layers with different degrees of tissue compaction77.

Less common ‘specific’ causes of cardiomyopathy Hurler’s syndrome79 Cardiac failure may be due to involvement of the myocardium or the mitral or aortic valves. Thyrotoxicosis in the elderly may present with atrial fibrillation unresponsive to the usual methods of treatment. In the younger patient, the characteristic cardiac abnormality is high output cardiac failure with enlargement of all chambers. Myxoedema usually causes a pericardial effusion that resolves when the myxoedema is brought under control, but cardiac dilatation may occur. Acromegaly Hypertension and coronary artery disease are the usual features of this endocrine condition, but it is generally believed that there is a specific heart muscle abnormality that is associated with cardiac failure. Beri-beri Deficiency of thiamine (vitamin B1) may lead to cardiac failure, commonly of high output but occasionally of low output type. Both types show a large heart and pulmonary congestion, and they respond dramatically to thiamine. Alcoholic heart disease80 Alcohol appears to have a direct effect on the myocardium, rather than an indirect effect from a poor diet. There is no specific test for alcoholic heart disease, and withdrawal of alcohol may not improve function. Beta-blocking drugs These have a negative inotropic action (i.e. they depress myocardial contraction). When administered in large doses, particularly to control angina, they may precipitate congestive cardiac failure in the already diseased heart.The cardiac failure is reversible when the drug is discontinued. Cytotoxic drugs81 Among many cytotoxic drugs used for the treatment of malignant disease, doxorubicin is the most likely to precipitate a syndrome of congestive cardiomyopathy. This may remit if the drug is stopped. Postpartum cardiomyopathy usually remits entirely but may recur in subsequent pregnancies.

Figure 24.46 Left ventricular noncompaction. (A) Contrast-enhanced CT at mitral valve level shows increased trabeculations (arrows) near the apex of the left ventricle (LV). The heart is enlarged. This young patient was being investigated for arrhythmia and a follow-up echocardiography confirmed the presence of noncompaction. (B) CMR of a different patient, two-chamber projection, shows markedly increased fine trabeculations in the LV wall more marked near the ventricular apex with a faint line of demarcation between ‘normal’ and ‘abnormal’ segments (arrowheads). LA = Left atrium, LV = left ventricle, RV = right ventricle.

Neuromuscular disorders82 In a number of familial muscular disorders, death ultimately occurs from congestive cardiac failure due to heart muscle involvement.The radiological appearances are of non-specific cardiac dilatation and failure. Endocardial fibroelastosis This condition is characterized by the deposition of elastic tissue on the inside of the left ventricle, commonly involving the mitral valve and producing mitral regurgitation. Two forms are recognized: the dilated variety that occurs in the absence of other cardiac abnormality and produces a clinical syndrome identical to that of DCM; and a

CHAPTER 24

contracted form that occurs in association with various forms of left ventricular outflow obstruction.The dilating form is the most common type of heart muscle disease in childhood. In the dilated form, echocardiography shows the left ventricular cavity to be dilated and the ejection fraction to be reduced (Fig. 24.47). The thickened endocardium may be visible as a bright line adjacent to the ventricular cavity and separating it from the underlying myocardium. The right ventricle may be involved in a similar way. Mural thrombus may be present in either ventricle. Doppler may demonstrate functional mitral and/or tricuspid regurgitation. Hypertensive heart failure This is a form of congestive cardiac failure, the cause of which is obvious when the high systemic blood pressure is known, but may be obscured if the blood pressure falls as a result of cardiac failure. Rarely, hypertension may lead to the development of aortic incompetence. Echocardiography is useful in determining the degree of left ventricular hypertrophy and contractile impairment. Severe left ventricular hypertrophy may cause cardiac enlargement on the plain chest radiograph, but usually left ventricular enlargement is due to dilatation, and the picture may resemble a dilated cardiomyopathy, although some septal hypertrophy characteristically persists in the latter. Hypertension is a risk factor in the development of coronary artery disease, which may have its own effects on the ventricle. Anaemia, when severe, may precipitate cardiac failure, particularly when pre-existing heart disease is present. Radiation damage83 to the heart may involve pericardium, myocardium, or coronary arteries. Glycogen storage disease (Pompe’s disease) Most patients with glycogen storage disease present with ‘floppy baby’ syndrome with macroglossia. Rarely, the condition may present as heart failure before the characteristic general muscle weakness becomes obvious.The plain chest radiograph reveals gross cardiomegaly with either a left ventricular or a non-specific

•

NONISCHAEMIC ACQUIRED HEART DISEASE

dilatation. Often the liver may be enlarged.The left ventricle is large and poorly contracting, but the left ventricle wall is very markedly increased in thickness and irregular in outline on its inner aspect, giving characteristic echocardiographic and ventriculographic appearances84. Fabry’s disease is similar to Pompe’s disease but the myocardial wall thickening is gross.

TUMOURS OF THE HEART Metastatic tumours to the heart and pericardium are 20–40 times more common than primary heart tumours. Among primary cardiac tumours, benign tumours are more common than malignancy. Cardiac tumours can clinically present as heart failure, arrhythmia, chest pain, or embolic disease85.The diagnosis is usually obvious when clinical evidence of cardiac involvement, arrhythmias, or (haemorrhagic) pericardial effusion develops in a patient with a known primary malignant neoplasm. Intrathoracic, extracardiac tumours, benign or malignant, may produce cardiac symptoms and signs by compressing the heart and great vessels, and mimic obstructive lesions of these vessels. Obstructive murmurs developing in the course of the malignant disease are well recognized. Cardiac tumours and ventricular and aortic aneurysms may resemble each other radiographically. Echocardiography has been the imaging technique of choice for the diagnosis of intracardiac tumours. Real-time imaging can show tumour mobility and distensibility features which are typically seen in atrial myxomas and less likely in sarcomas and metastases. CT and MRI can be used accurately to image the heart and the surrounding mediastinum. CT can depict calcifications and fat, and may allow tissue diagnosis of some masses such as lipomas. MRI allows better soft tissue characterization than CT and can provide functional information such as flow direction and velocity, as well as a more reliable diagnosis of true invasion of cardiac structures by intrathoracic tumours86,87.

Metastasis

Figure 24.47 Endocardial fibro-elastosis, cross-sectional echocardiogram, apical two-chamber view. Note the enlargement of the left ventricular (LV) cavity size. The cavity is abnormally round, and the amplitude of endocardial (En) and subendocardial echoes is increased.

There are three ways of cardiac involvement by metastases: (A) direct mediastinal infiltration by lung cancer, breast cancer, or lymphoma; (B) haematogenous metastasis from a systemic tumour such as malignant melanoma, lymphoma, leukaemia, or sarcoma; and (C) transvenous spread, from the inferior vena cava (renal or hepatic tumours) (Fig. 24.48A) and via the pulmonary vein (Fig. 24.48B) or superior vena cava in cases of lung cancer. CT is extremely useful in tumour localization and staging for surgical resection. CT features that suggest a malignant nature of a cardiac neoplasm are wide attachment to the wall of the heart, destruction of the cardiac chamber wall, involvement of more than one cardiac chamber, invasion of the pericardium, especially with haemorrhage, extension into the pulmonary artery, pulmonary vein, or cava, and involvement beyond the pericardium, lung, or mediastinum (Figs 24.49, 24.50).

495

496

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.48 Intravascular extension of tumours into the heart. (A) Contrast-enhanced CT in a patient with hepatocellular carcinoma extending directly into the right atrium via the inferior vena cava (arrows). Renal cell carcinoma and adrenal carcinoma also commonly spread this way. (B) Direct extension of lung cancer via the pulmonary vein into the left atrium. Contrast-enhanced CT in a patient with right lower lobe bronchogenic carcinoma (T) shows tumour growing into the left atrium via the right inferior pulmonary vein (arrows). C = Inferior vena cava; L = left atrium; LV = left ventricle; R = right atrium; RV = right ventricle.

CT is not an accurate technique for detecting direct invasion of the mediastinum and cardiovascular structures by lung cancer. The sensitivity for assessment of mediastinal invasion by CT ranges from 40 to 78% and specificity from 67 to 99%88. The contact area between the thoracic mass and adjacent mediastinal structure, and the obliteration of intervening fat are the commonly used CT features for local invasion, but neither is reliable. The sliding motion between two contacting areas can be the definitive evidence of absence of invasion and this feature can be detected by MRI86.

Primary cardiac tumours Primary tumours of the heart89–92 are rare and the great majority are benign, the malignant tumours being sarcomas. Myxomas are by far the most common tumours, followed by rhabdomyomas and fibromas. The clinical features of these tumours depend on their sites of origin. Intracavity tumours

are commonly pedunculated and may impact on and/or occlude the valves, or fill cardiac chambers, leading to obstruction, arrhythmias, and cardiac failure. Extension of tumours to the pericardium may produce haemorrhagic pericardial effusion that may lead to tamponade.

Cardiac myxoma This most common primary cardiac tumour commonly arises in the left (75%) or right atrium (20%), and rest in either ventricle. Myxoma tends to be solitary, pedunculated or polypoid, occurs more commonly in women than men, and in 85% of cases is characteristically attached to the interatrial septum near the fossa ovalis93. Patients may be asymptomatic or have the triad of peripheral embolic phenomena, symptoms and signs of mitral valve obstruction, and constitutional symptoms of fever, anaemia, raised ESR, and sometimes finger clubbing mimicking infective endocarditis. Blood cultures,

Figure 24.49 Undifferentiated sarcoma. (A–C) Contrast-enhanced CT of the chest shows a large lobulated mass beginning near the tricuspid valve (arrows in A,B), extending into the right ventricular outflow tract (C). A similar mass is also present in the left ventricle (M) in (A); both confirmed at echocardiography. LA = left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

CHAPTER 24

Figure 24.50 Metastasis to the right ventricular myocardium from carcinoma of the vulva. Axial, contrast-enhanced chest CT. There is irregular infiltration of the anterior free wall of the right ventricle (arrows). LA = left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

however, are sterile and splenomegaly does not occur. Familial myxomas constitute fewer than 10% of all myxomas, tend to present earlier (median age 20 years), are more likely to have multiple myxomas at atypical locations and to develop recurrent tumours, and have associated dermatological and endocrine abnormalities (Carney complex).94 On chest radiography, the heart is commonly enlarged and there is often evidence of selective left atrial enlargement (Fig. 24.51A), though rarely is there a large appendage (Fig. 24.51B), which would suggest rheumatic heart disease. Pulmonary venous hypertension, pulmonary oedema, or even pulmonary

•

NONISCHAEMIC ACQUIRED HEART DISEASE

arterial hypertension may be present. The appearances of the chest radiograph may, however, be normal. The rare calcified myxoma may be identified on chest radiograph or CT (Fig. 24.52), and at fluoroscopy it moves forward through the mitral valve in ventricular diastole. Both MRI and MDCT95 are capable of demonstrating cardiac tumours, MRI being particularly good at showing muscle invasion. Myxomas demonstrate an intermediate but variable signal intensity on spin-echo images similar to that of myocardium. On gradient-echo images, myxomas often have a low signal intensity caused by partial calcification, and higher signal intensity is seen on T2-weighted sequences. Intratumoural subacute or chronic haemorrhage appears as high signal intensity on both T1- and T2-weighted sequences. A moderate contrast enhancement after Gd-DTPA is caused by increased vascularization96 (Fig. 24.53). A cine-MRI display may show the mobility of the tumour. Echocardiography shows the tumour as a rounded, mobile mass often prolapsing through the mitral orifice in ventricular diastole (Fig. 24.54). Atrial myxoma must be distinguished from thrombus. Thrombus usually occurs in an enlarged chamber, atrial appendage is commonly involved, and atrial fibrillation is likely to be present (Figs 24.55A–C).Thrombi are usually sessile and do not demonstrate contrast enhancement which is seen to a variable degree in myxomas. Myxomas can be sessile or pedunculated and are commonly attached to the interatrial septum. Purely from the imaging perspective, other cardiac tumours and valvular vegetations may also present as intracardiac masses. Clinical features help in differentiating one from the other. Besides the presence of clinical symptoms and signs of endocarditis, vegetations are always related to a cardiac valve (Fig. 24.55D). Once diagnosed, myxoma should be removed without delay.

Figure 24.51 Left atrial myxoma. (A) PA view of the chest shows a large heart with all chambers involved. There is interstitial pulmonary oedema. (B) Chest radiograph in a patient with known left atrial myxoma demonstrating enlargement of the left atrial appendage (arrows). This finding is commonly seen in rheumatic mitral valve disease and is rare in atrial myxoma.

497

498

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Hypoxic spells similar to tetralogy of Fallot have been reported. These tumours are often inoperable because they are commonly deep seated, poorly demarcated, and multiple.

Fibroma

Figure 24.52 Left atrial myxoma. Unenhanced CT at the level of the aortic root shows a faintly calcified soft tissue mass (arrows) in the left atrium (LA) attached to the interatrial septum. Ao = Aortic root, LV = left ventricle, RA = right atrium.

Lipomas Lipomas can occur in the myocardial free wall or the septae. These lesions are easily identified on CT as low density masses. Lipomatous hypertrophy of the interatrial septum is now a well known entity and is often found incidentally on routine chest CT. Lipomas have high signal intensity on MRI due to their short T1 and long T2 relaxation, consistent with fat (Fig. 24.56).

Rhabdomyomas These are the most common paediatric primary cardiac tumours and occur in association with tuberous sclerosis in up to 50% of cases, where they may be congenital and manifest in the neonatal period.They are usually intramural and start in the interventricular septum, encroaching on the ventricular cavities as they enlarge. Arrhythmias, which may be fatal, are their main presenting feature.

A fibroma is usually a single tumour of the ventricular wall, generally on the left, and may be resectable. The presentation is with arrhythmia or congestive cardiac failure, and plain radiographs show a large heart. Though the tumour is rare, it is of radiological interest in that it may calcify and show characterizing whorls of calcium, which may suggest the specific diagnosis. On MRI, fibromas are iso- to hyper-intense on T1-weighted spin-echo images when compared to skeletal muscle. Due to the short T2-relaxation time of the fibrous tissue, fibromas show a decrease in signal intensity relative to the myocardium on T1- and T2-weighted spin-echo images.

Hydatid disease Hydatid disease may involve the heart, the cysts behaving as a benign myocardial tumour; they may calcify.

Primary malignant tumours of the heart Sarcomas96 Sarcomas account for nearly all primary malignant cardiac neoplasms (Fig. 24.57, see Fig. 24.49). Angiosarcoma is the most common malignant tumour of the heart. Most exhibit a strong male prevalence and arise in childhood. The right atrium and then the right ventricle are the chambers most commonly affected. The tumours are mainly intramural and infiltrating, though they may encroach on the cavity of the chamber. They usually present with intractable arrhythmias or relentless cardiac failure, features of obstruction to the vena cava, the tricuspid or pulmonary valves, or more rarely the mitral and aortic valves. Cardiac pain may occur. Involvement of the pericardium leads to pericardial effusion, often haemorrhagic, and this may lead to tamponade.The

Figure 24.53 Left atrial myxoma. CMR (A) gradient-echo and (B) post gadolinium contrast-enhanced axial images demonstrate lobulated soft tissue mass (M) in the left atrium attached to the interatrial septum, exhibiting significant enhancement after contrast medium administration (B). The typical location and post contrast enhancement helps in differentiating myxoma from thrombus. Compare to Fig. 24.55. LV = Left ventricle, RA = right atrium, RV = right ventricle.

CHAPTER 24

•

NONISCHAEMIC ACQUIRED HEART DISEASE

Figure 24.54 Left atrial myxoma. (A) Transoesophageal echocardiogram shows a large tumour (T) in the left atrium (LA) in ventricular systole. (B) In diastole it prolapses through and obstructs the mitral annulus, dragging down its septal attachment.

Figure 24.55 Other masses in the atria. (A,B) Left atrial thrombus in a patient with mitral stenosis. (A) Contrast-enhanced CT of the chest shows a large filling defect in the left atrium (LA) extending into the left atrial appendage (LAA). (B) More caudal image. The left atrium is dilated and dense calcification is seen in the stenotic mitral valve. (C) Right atrial thrombus. Contrast-enhanced chest CT. A large intraluminal nonenhancing mass is seen in the right atrium away from the interatrial septum. Compare to Fig. 24.53. (D) Valvular vegetation. Contrast-enhanced CT at the level of the inferior vena cava–right atrium (RA) junction shows an intraluminal filling defect (arrows) protruding into the right atrium. This patient has a history of intravenous drug abuse and there were multiple cavitating lesions in both lungs suggestive of septic emboli (not shown). This lesion may represent thrombus, RA myxoma, or valvular vegetation. At surgery a Eustachian valve vegetation was found. LA = Left atrium, LV = left ventricle, RV = right ventricle.

499

500

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.56 Fat-containing cardiac masses. Lipomatous hypertrophy of interatrial septum. (A) CT and (B) T1-weighted axial MRI shows extensive fatty infiltration (F) in the interatrial septum, extending into the posterior atrial walls. (C) Left ventricular lipoma. Spin-echo CMR. High signal intensity mass (M) is seen in the left ventricular cavity. Due to short T1- and long T2-relaxation properties of fat, lipomas are bright on T1- and T2-weighted images. LA = Left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

cardiac shadow is usually enlarged on the plain radiograph, and echocardiography will show a mass which is atypical of myxoma, with or without an effusion. MRI will show the extent of the lesion to greater advantage, and coronary angiography may demonstrate abnormal vascularity. However, this is not necessarily typical of sarcoma and may be seen in tumours such as myxomas, or even long-standing thrombi.

Lymphoma Lymphoma (Fig. 24.58) may affect the heart with little evidence of involvement elsewhere. Characteristically it takes the form of a diffuse infiltration of the myocardium, particularly of the right ventricular outflow tract. Both primary and metastatic (secondary) cardiac lymphoma occur more commonly in immunocompromised individuals. Patients who are positive for human immunodeficiency virus (HIV) disease are more prone to all forms of extranodal lymphoma, which are typically aggressive B-cell lymphomas. Primary cardiac lymphoma typically involves the right atrium, and involvement of other chambers is less common.

Figure 24.57 Angiosarcoma of the right atrium. Contrast-enhanced chest CT. There is a heterogeneous soft tissue abnormality lateral and anterior to the right atrium (RA). There are associated pericardial (P) and bilateral pleural effusions.

Most patients have pericardial extension, but valves are rarely involved97,98.

TRAUMA TO THE HEART99,100 Penetrating trauma Gunshot and knife wounds that penetrate the heart usually lead to haemorrhage and pericardial tamponade. This clinical situation is usually too critical to allow for more than supine radiography. When penetration is in the location of the heart, and resuscitation has not been successful with adequate blood transfusion, tamponade is assumed and the pericardium is surgically explored. After the acute phase, retained foreign bodies need to be carefully located by radiography (Fig. 24.59) and angiocardiography before attempts are made to remove them.

Blunt trauma101,102 (Fig 24.60) The clinical effects of blunt cardiac trauma may be delayed and may cause subsequent insidious deterioration of cardiac func-

Figure 24.58 Patient with human immunodeficiency virus (HIV) infection and aggressive B-cell lymphoma invading the mediastinum and heart. A large heterogeneous soft tissue mass is invading the mediastinum, pericardium, heart, and left pleural space displacing the cardiac structures to the right. Ao = Aortic root; LA = left atrium; R = right atrial appendage.

CHAPTER 24

tion.Traumatic pericarditis with haemorrhage may result from blunt trauma, as may traumatic haemorrhage with necrosis of the myocardium. In the latter, involvement of small areas may be clinically silent or of no great significance, but necrosis of large areas of the myocardium may be fatal. The lesion resembles myocardial infarction. The pericardium may be disrupted in blunt trauma, allowing herniation and strangulation of parts of the heart, or the abdominal or thoracic contents may enter the torn pericardium.

Injury common to both types of trauma Perforation of the ventricular septum Though more common with penetrating than with blunt trauma, perforation of the septum may occur with both types. The development of a murmur of ventricular septal defect following chest trauma points to the diagnosis. A shunt of haemodynamic significance should be surgically repaired. The clinical presentation often is subacute and usually allows full investigation by echocardiography, cardiac catheterization, and angiocardiography.The ventricular septal defect tends to be low or apical in the ventricular septum. Coronary arteriography is usually performed at the same time in order to recognize atheromatous coronary artery disease and the possibility of trauma to the coronary arteries, as damage to either may lead to septal infarction and perforation.

False aneurysms False aneurysms may develop when rupture or perforation of the free wall myocardium is held in check by pericardium. Such aneurysms increase in size (Fig. 24.61) and commonly rupture, either early or after many years. They may first present as an abnormal cardiac contour on chest radiography. If

Figure 24.59 Contrast-enhanced CT in a patient after gunshot injury shows bullet with streak artefacts lodged in the right side of the interventricular septum.

•

NONISCHAEMIC ACQUIRED HEART DISEASE

present for several years, they may calcify. False aneurysms may be seen at ventriculography and can be distinguished from postinfarction aneurysms, as they usually fill through a narrow neck from the free wall of the left ventricle. Usually the coronary arteries are not very abnormal but trauma may transect or thrombose a coronary artery and lead to myocardial infarction and possible false aneurysm formation.

Valve disruption Any of the cardiac valves may be damaged by trauma, but the aortic valve is the most commonly affected, leading to aortic incompetence. Tricuspid incompetence may be virtually without symptoms, but traumatic mitral incompetence usually requires urgent correction. The characteristic murmur of valvar incompetence occurring after chest trauma suggests the correct diagnosis.

Trauma to the coronary arteries Trauma to the coronary arteries may lead to laceration of the artery, with haemorrhage that may require surgical control. Traumatic thrombosis may lead to myocardial infarction, the extent of which depends on the size of the vessel that is occluded. Rarely, trauma may lead to a coronary artery to cardiac cavity (cameral) fistula. A continuous murmur developing for the first time after major chest trauma should suggest this complication. False aneurysms may develop after trauma to the coronary arteries.

Trauma to the pericardium Pericardial tamponade is one of the most common causes of death after trauma to the heart. Traumatic pericarditis may calcify but rarely leads to constriction.

Figure 24.60 Blunt chest trauma causing rupture of right atrial appendage. There is a large anterior mediastinal haematoma (H) and haematoma surrounding the right atrial appendage (arrows). At surgery, rupture of the right atrial appendage near its junction with the right atrium was noted. Ao = aortic root, RA = right atrium.

501

502

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 24.61 Traumatic false aneurysm. Plain radiographs (A) immediately after a car accident involving chest trauma, showing an insignificant bulge on the left heart border, and (B) 16 months later, a large bulge has developed in the mid left heart border. (C) Left ventriculogram shows that the bulge is a narrow-necked false aneurysm filling from the upper left ventricular cavity.

REFERENCES 1. Murray C J L, Lopez A D 1996 The global burden of disease. Harvard School of Public Health, Cambridge, MA 2. Gaziano J M 2005 Global burden of cardiovascular disease. In: Zipes D P, Libby P, Bonow R O, Braunwald E (eds) Braunwald’s heart disease: A textbook of cardiovascular medicine. W B Saunders, Philadelphia, pp 1–19 3. Alkadhi H, Bettex D, Wildermuth S et al 2005 Dynamic cine imaging of the mitral valve with 16-MDCT: A feasibility study. Am J Roentgenol 185: 636–646 4. Chen L, Otto C M 1998 Longitudinal assessment of valvular heart disease by echocardiography. Curr Opin Cardiol 13: 397–403 5. Roberts W C 1983 Morphologic features of the normal and abnormal mitral valve. Am J Cardiol 51: 1005–1027 6. Leathem A, Bridgen W 1980 Mild mitral regurgitation and the mitral valve prolapse fiasco. Am Heart J 99: 659–664 7. Fujita N, Chazoulleres A F, Hartiala J J et al 1994 Quantification of mitral regurgitation by velocity-encoded cine nuclear magnetic resonance imaging. J Am Coll Cardiol 23: 951–958 8. Hundley W G, Li H F, Willard J E et al 1995 Magnetic resonance imaging assessment of the severity of mitral regurgitation. Comparison with invasive techniques. Circulation 92: 1151–1158 9. Otsuji Y, Gilon D, Jiang L et al 1998 Restricted diastolic opening of the mitral leaflets in patients with left ventricular dysfunction: Evidence for increased valve tethering. J Am Coll Cardiol 32: 398–404 10. Fulkerson P K, Beaver B M, Auseon J C, Graber H L 1979 Calcification of the mitral valve annulus: Etiology, clinical associations, complications and therapy. Am J Med 66: 967–977 11. Adler Y, Koren A, Fink N et al 1998 Association between mitral annulus calcification and carotid atherosclerotic disease. Stroke 29: 1833–1837 12. Lachman A S, Roberts W C 1978 Calcific deposits in stenotic mitral valves. Circulation 57: 808–815 13. Higgins C B, Caputo G R 1993 Role of MR imaging in acquired and congenital cardiovascular disease. Am J Roentgenol 161: 13–22 14. Pomerance A 1972 Pathogenesis of aortic stenosis and its relation to age. Br Heart J 34: 569–574 15. Passik C S, Ackermann D M, Pluth J R, Edwards W D 1987 Temporal changes in the causes of aortic stenosis: A surgical pathological study of 646 cases. Mayo Clin Proc 52: 119–123 16. Boxt L M 2005 CT of valvular heart disease. Int J Cardiovasc Imaging 21: 105–113 17. Shavelle D M, Budoff M J, Buljubasic N et al 2003 Usefulness of aortic valve calcium scores by electron beam computed tomography as a marker of aortic stenosis. Am J Cardiol 92: 349–353 18. Friedrich M G, Schulz-Menger J, Poetsch T et al 2002 Quantification of valvular aortic stenosis by magnetic resonance imaging. Am Heart J 144: 329

19. Dulce M-C, Mostbeck G H, O’Sullivan M, Cheitlin M, Caputo R, Higgins C B 1992 Severity of aortic regurgitation: Interstudy reproducibility of measurement of velocity-encoded cine MR imaging. Radiology 185: 235–240 20. Takahashi K, Stanford W 2005 Multidetector CT of the thoracic aorta. Int J Cardiovasc Imaging 21: 141–153 21. Carr J C, Finn J P 2003 MR imaging of the thoracic aorta. Magn Reson Imaging Clin N Am 11: 135–148 22. Starr A, Edwards M L 1961 Mitral replacement: Clinical experience with a ball-valve prosthesis. Ann Surg 154: 726–740 23. Hammond G L, Geha A S, Kopf G S, Hashim S W 1987 Biological versus mechanical valves: Analysis of 1116 valves inserted in 1012 adult patients with 4818 patient-years and 5327 valve-years follow-up. J Thorac Cardiovasc Surg 93: 182–193 24. Smith J A, Westlake G W, Mullerworth M H, Sillington P D, Tatoulis J 1993 Excellent long-term results of cardiac valve replacement with the St. Jude medical valve prosthesis. Circulation 88: 49–54 25. Kotler M N, Goldman A, Parry W R 1986 Non-invasive evaluation of cardiac valve prosthesis. In: Kotler M N, Steiner R M (eds) Cardiac imaging: new technologies and clinical applications. Davis, Philadelphia, pp 201–241 26. Kotler M N, Mintz G S, Panidis I, Morgan R J, Segal B L, Ross J 1983 Noninvasive evaluation of normal and abnormal prosthetic valve function. J Am Coll Cardiol 2: 151–173 27. Steiner R M, Mintz G S, Morse D et al 1988 The radiology of cardiac valve prostheses. RadioGraphics 8: 277–298 28. Gross B H, Shirazi K K, Slater A D 1983 Differentiation of aortic and mitral valve prostheses based on postoperative frontal chest radiographs. Radiology 149: 389–391 29. Lewell D B, Ofole S, McCorkell S J 1988 Prosthetic heart valve malfunction: Plain film findings. J Can Assoc Radiol 39: 182–185 30. Bjork V O, Henze A, Jerelo M 1973 Aortographic follow-up in patients with the Bjork–Shiley disc valve prosthesis. Scand J Thorac Cardiovasc Surg 7: 1–6 31. White A F, Dismore R E, Buckley M J 1973 Cineradiographic evaluation of prosthetic cardiac valves. Circulation 48: 882–890 32. Mehlman D L, Resnikov L 1984 A guide to the radiographic identification of prosthetic valves. Circulation 69: 102–112 33. Castadena-Zuniga W, Nicoloff D, Jorgeensen C, Nath P H, Zollikofer E, Amplatz K 1980 In vivo radiographic appearance of the St. Jude valve prosthesis. Radiology 134: 775–776 34. Yun K L, Sintek C F, Fletcher A D et al 1999 Aortic valve replacement with the Freestyle stentless prosthesis: Five year experience. Circulation 100: 11–17 35. Dossche K, Vanerman H, Daernen W et al 1996 Edwards stentless aortic valve xenograft: Early results of a multicenter clinical trial. Thorac Cardiovasc Surgeon 44: 11

CHAPTER 24

36. Westaby S, Jin X Y, Katsumata T, Arifi A 1998 Valve replacement with a stentless bioprosthesis: Versatility of the porcine aortic root. J Thorac Cardiovasc Surg 116: 477 37. Landay M J, Estrera A S, Bordlee R P 1992 Cardiac valve reconstruction and replacement: A brief review. RadioGraphics 12: 659–671 38. Kirklin J W, Barratt-Boyes B G 1986 Cardiac surgery. Wiley, New York, 1986, pp 331–332 39. Zumbro G L, Cunder P E, Fishback M E, Galloway R F 1977 Strut fracture in DeBakey valve. J Thorac Cardiovasc Surg 74: 469–470 40. Guit G L, Van Voorthuisen A E, Steiner R M 1985 Outlet strut fracture of the Bjork–Shiley mitral prosthesis. Radiology 154: 298–299 41. Cipriano P R, Billingham M E, Oyer P E, Kutsche L M, Stinson E B 1982 Calcification of porcine prosthetic heart valves. Circulation 60: 1100–1104 42. Knight J P, Torell J A, Hunter R E 1984 Bacterial endocarditis associated with porcine heterograft heart valve calcification. Am J Cardiol 53: 370–372 43. Paquet D J, Randall P, Parker F B 1988 Radiographic appearance of the St. Jude medical valve. J Can Assoc Radiol 39: 186–189 44. Sands M J, Lachman A S, O’Reilly D J, Leach C W, Sappington J E, Katz A M 1982 Diagnostic value of cinefluoroscopy in the evaluation of prosthetic valve dysfunction. Am Heart J 4: 622–627 45. Mamourian A C, Pae W E 1986 The radiographic appearance of the Medtronic-Hall valve prosthesis. Am J Radiol 146: 485–486 46. Green C E, Glass-Royal M, Bream P R, Soto B, Elliot L P 1988 Cinefluoroscopic evaluation of periprosthetic cardiac valve regurgitation. Am J Radiol 151: 455–459 47. Steiner R M, Flicker S 1985 The radiology of prosthetic heart valves. In: Morse D, Steiner R M, Fernandez J (eds) A guide to prosthetic cardiac valves. Springer, New York, pp 53–77 48. Hipona F A, Lerona P T, Paredes S 1971 Radiologic diagnosis of late complications associated with cardiac valve surgery in acquired heart disease. Radiol Clin North Am 9: 265–283 49. Gabriel O 1969 Postoperative radiography of aortic valve prosthesis. J Thorac Cardiovasc Surg 58: 248–249 50. Green C E, Glass-Royal M, Bream P R, Soto B, Elliott L P 1988 Cinefluoroscopic evaluation of periprosthetic cardiac valve regurgitation. Am J Radiol 151: 455–459 51. Silber H, Khan S S, Matloff J M, Chaux A, DeRobertis M, Gray R 1993 The St. Jude valve thrombolysis as the first line of therapy for cardiac thrombosis. Circulation 87: 30–37 52. Heimberger T S, Duma R J 1989 Infections of prosthetic heart valves and cardiac pacemakers. Infect Dis Clin North Am 3: 221–245 53. Winkler M L, Higgins C B 1986 MRI of perivalvular infectious pseudoaneurysms. Am J Roentgenol 7: 253–256 54. Chun P K E, Rajfer S I, Donahue K J, Bowen T E, Davia J E 1980 BjorkShiley mitral valve dehiscence. Am Heart J 99: 230–234 55. Venkataraman K, Beer R, Mathews N P et al 1980 Thrombosis of Bjork–Shiley aortic valve prostheses. Radiology 37: 43–47 56. Edmunds L H 1981 Thromboembolic complications of current cardiac valvular prostheses. Ann Thorac Surg 34: 96–106 57. Hartzler G, Diehl A M, Reed W A 1984 Non-surgical correction of a frozen disc valve prosthesis using a catheter technique and intracardiac streptokinase infusion. J Am Coll Cardiol 4: 779–783 58. Soulen R L, Budinger T F, Higgins C B 1985 Magnetic resonance imaging of prosthetic heart valves. Radiology 154: 705–707 59. Edwards M B, Ordidge R J, Hand J W, Taylor K M, Young I R 2005 Assessment of magnetic field (4.7T) induced forces on prosthetic heart valves and annuloplasty rings. J Magn Reson Imaging 22: 311–317 60. Deutsch H J, Bachmann R, Sechtem U et al 1992 Regurgitant flow in cardiac valve prosthesis: Diagnostic value of gradient echo nuclear magnetic resonance imaging in reference to transesophageal twodimensional color Doppler echocardiography. J Am Coll Cardiol 19: 1500–1507 61. Globits S, Higgins Ch B 1995 Assessment of valvular heart disease by magnetic resonance imaging. Am Heart J 129: 361–381 62. Richardson P, McKenna W, Bristow M et al 1996 Report of the 1995 World Health Organization/International Society and Federation of Cardiology Task Force on the Definition and Classification of Cardiomyopathies. Circulation 93: 841

•

NONISCHAEMIC ACQUIRED HEART DISEASE

63. Canady M, Hagye M, Kallai A, Forster T, Szarazajtai T 1995 Familial dilated cardiomyopathy—a worse prognosis compared to sporadic forms. Br Heart J 74: 171–173 64. Juilliere Y, Barbier G, Feldmann L, Grentzinger A, Danchi N, Cherrier F 1997 Additional predictive value of both left and right ventricular ejection fractions on long term survival in dilated cardiomyopathy. Eur Heart J 18: 276–280 65. Braunwald E, Seidman CE, Sigwart U 2002 Contemporary evaluation and management of hypertrophic cardiomyopathy. Circulation 106: 1312 66. Nishiyama S, Yamaguchi H, Ishimura T et al 1979 Echocardiographic features of apical HCM. J Cardiogr 8: 177–183 67. Chapman A H, Raphael M J, Steiner R E, Oakley C M 1978 Unusual chest x-ray appearances in hypertrophic cardiomyopathy. Clin Radiol 29: 9–16 68. Cannan C, Reeder G, Bailey K et al 1995 Natural history of hypertrophic cardiomyopathy. Circulation 92: 2488–2495 69. Rickers C, Wilke N M, Jerosch-Herold M et al 2005 Utility of cardiac magnetic resonance imaging in the diagnosis of hypertrophic cardiomyopathy. Circulation 112: 855–861 70. Wilmshurst P T, Katritis D 1990 Restrictive cardiomyopathy (editorial). Br Heart J 63: 323–324 71. Kwong R Y, Falk R H 2005 Cardiovascular magnetic resonance in cardiac amyloidosis. Circulation 111: 122–124 72. Maceira A M, Joshi J, Prasad S K et al 2005 Cardiovascular magnetic resonance in cardiac amyloidosis. Circulation 111: 186–193 73. Blankenberg F, Eisenberg S, Scheinman M N et al 1994 Use of cine GRASS MR in the imaging of cardiac hemochromatosis. J Comput Assist Tomogr 18: 136 74. Gewilig M, Mertens L, Moerman P, Dumoulin M 1996 Idiopathic restrictive cardiomyopathy in children. Eur Heart J 17: 1413–1420 75. Cockshott W P 1969 Cardiomyopathy in the tropics: endomyocardial fibrosis. Semin Roentgenol 4: 367–373 76. Ivan D, Flamm S D, Abrams J, Kindo M, Heck K, Frazier O H 2005 Isolated ventricular non-compaction in adults with idiopathic cardiomyopathy: Cardiac magnetic resonance and pathologic characterization of the anomaly. J Heart Lung Transplant 74: 781–786 77. Petersen S E, Selvanayagam J B, Wisemann F et al 2005 Left ventricular non-compaction: Insights from cardiovascular magnetic resonance imaging. J Am Coll Cardiol 46: 101–105 78. Murphy R T, Thaman R, Blanes J G et al 2005 Natural history and familial characteristics of isolated left ventricular non-compaction. Eur Heart J 26: 187–192 79. Krovetz J, Lorincz A E, Schiebler G L 1965 Cardiovascular manifestations of the Hurler’s syndrome. Circulation 31: 132–141 80. Brigden W, Robinson J. Alcoholic heart disease. BMJ 1964;2: 1283–1289 81. Rhoden W, Hasleton P, Brooks N 1993 Anthracyclines and the heart. Br Heart J 70: 499–502 82. Perloff J K 1971 Cardiomyopathy associated with heredofamilial neuromyopathic diseases. Mod Concepts Cardiol Dis 40: 23–26 83. Sebag-Montefiore D, Hope-Stone H 1993 Radiation induced coronary heart disease. Br Heart J 69: 481–482 84. Hohn A R, Lowe C U, Sokal J E, Lambert E C 1965 Cardiac problems in the glycogenoses with special reference to Pompe’s disease. Pediatrics 35: 313–321 85. Sabatine M S, Colucci W, Schoen F J 2005 Primary tumors of the heart. In: Zipes D P, Libby P, Bonow R O, Braunwald E (eds) Braunwald’s heart disease: A textbook of cardiovascular medicine. W B Saunders, Philadelphia, pp 1741–1745 86. Seo J S, Kim Y J, Choi B W, Choe K O 2005 Usefulness of magnetic resonance imaging for evaluation of cardiovascular invasion: evaluation of sliding motion between thoracic mass and adjacent structures on cine MR images. J Magn Reson Imaging 22: 234–241 87. Sparrow P J, Kurian J B, Jones T R, Sivananthan M U 2005 MR imaging of cardiac tumors. RadioGraphics 25: 1255–1276 88. Herman S J, Winton T L, Weisbrod G L, Towers M J, Mentzer S J 1994 Mediastinalinvasion by bronchogenic carcinoma: CT signs. Radiology 190: 841–846

503

504

SECTION 2 • THE CHEST AND CARDIOVASCULAR SYSTEM

89. Sallee D, Spector M L, van Heeckeren D W, Patel C R 1999 Primary pediatric cardiac tumors: a 17 year experience. Cardiol Young 9: 155–162 90. O’Neill M B Jr, Grehi T M, Hurley E J 1979 Cardiac myxomas: a clinical diagnostic challenge. Am J Surg 138: 68–76 91. Arciniegas E, Hakimi M, Farooki Z Q, Truccone N T, Green E W 1980 Primary cardiac tumors in children. J Thorac Cardiovasc Surg 79: 582–591 92. St. John Sutton M G, Mercier L, Giuliani E R, Lie J T 1980 Atrial myxomas: a review of clinical experience in 40 patients. Mayo Clin Proc 55: 371–376 93. Schvartzman P R, White R D 2000 Imaging of cardiac and paracardiac masses. J Thoracic Imaging 15: 266–273 94. Carney J A, Gordon H, Carpenter P C et al 1985 The complex of myxomas, spotty pigmentation, and endocrine overactivity. Medicine (Baltimore) 64: 270–283 95. Mousseaux E, Hernigou A, Azencot M et al 1996 Evaluation by electron beam computed tomography of intracardiac masses suspected by transesophageal echocardiography. Heart 76: 256–263

96. Hoffmann U, Globits S, Frank H 1998 Cardiac and paracardiac masses: Current opinion on diagnostic evaluation by magnetic resonance imaging. Eur Heart J 19: 553–563 97. Duong M, Dubois C, Buisson M et al 1997 Non-Hodgkin’s lymphoma of the heart in patients infected with human immunodeficiency virus. Clin Cardiol 20: 497–502 98. Araoz P A, Eklund H E, Welch T J, Breen J F 1999 CT and MR imaging of primary cardiac malignancies. RadioGraphics 19: 1421–1434 99. Soulen R L, Freeman E 1971 Radiologic evaluation of traumatic heart disease. Radiol Clin North Am 9: 285–297 100. Jackson D H, Murphy G W 1976 Non-penetrating cardiac trauma. Mod Concepts Cardiovasc Dis 45: 123–128 101. Banning A P, Pillai R 1997 Non-penetrating cardiac and aortic trauma. Heart 78: 226–229 102. Chirillo F, Totis O, Cavarzerani A et al 1996 Usefulness of trans-thoracic and transesophageal echocardiography in recognition and management of cardiovascular injuries after blunt chest trauma. Heart 75: 301–306

CHAPTER

Ischaemic Heart Disease

25

George G. Hartnell and Julia Gates

• • • • • • • •

Selection of imaging tests appropriate to the task Imaging Coronary artery anatomy Manifestations of atheromatous coronary artery disease Myocardial infarction Percutaneous coronary intervention Nonpathological narrowing Coronary artery fistulas

Ischaemic heart disease (IHD) is the most common symptomatic manifestation of cardiovascular disease and by far the leading cause of mortality in western countries, accounting for over 1.5 million deaths each year in the European Union1. In the USA 13 000 000 individuals have symptomatic IHD (history of myocardial infarction in 7 100 000; angina pectoris in 6 400 000).

It is not surprising that the demand for invasive therapy has grown rapidly. For example, in the UK in 2001, 28 500 coronary artery bypass (CABG) procedures were performed, and in 2003, a total of 53 261 coronary angioplasty and stent (percutaneous coronary intervention [PCI]) procedures2. In the USA in 2002 an estimated 1 463 000 inpatient cardiac catheterizations were performed, with an estimated 657 000 PCI and 515 000 CABG procedures3. The number of procedures is still growing. It is therefore not surprising that IHD is by far the most common indication for all types of cardiac imaging. It is also one of the most challenging problems in diagnostic imaging due to the myriad manifestations of IHD, the constant movement of the heart, and the size and complex three dimensional (3D) anatomy of the coronary arteries (Figs 25.1, 25.2). The role of different imaging techniques in the evaluation of IHD is changing rapidly as technological advances increase the

Figure 25.1 Normal coronary anatomy. (A) Anterior view of the heart showing the left anterior descending (LAD) and right coronary arteries (black arrowheads). (B) Diagrammatic depiction of the three coronary arteries. The dotted lines represent the atrioventricular (AV) and interventricular grooves posteriorly. The continuous parallel lines represent the same grooves anteriorly. The interventricular groove for the left anterior descending artery is the more vertical of the two grooves.

506

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.2 Normal coronary arteriography. (A,B) Left coronary artery: The normal coronary artery has a smooth, gently tapering outline. (A) Left anterior oblique (LAO) projection; (B) right anterior oblique (RAO) projection. Open arrowhead = left anterior descending artery (LAD), solid arrowhead = diagonal, arrow = circumflex/obtuse marginal vessels. The LAD in (A) is framed between its diagonal branch posteriorly and its septal branches anteriorly. It reaches to the apex of the heart. The circumflex and its marginal branches overlap. In (B) the LAD and its diagonal branch overlap but the LAD may be identified by its septal branches and its extension to the apex of the heart. The small circumflex branch (arrow) has the characteristic straight appearance which suggests that it is running in the left atrioventricular (AV) groove. The larger obtuse marginal branch has left the AV groove and is running on the surface of the left ventricle. (C,D) Right coronary artery: (C) LAO projection; (D) RAO projection. Solid white arrowhead = right coronary artery, open arrowhead = posterior descending coronary artery, black arrowhead = sinus node artery, black arrow = inverted U at the crux, white arrow = posterolateral left ventricular branch. In (C) the entire right coronary artery is seen in profile as it passes around the heart in the right side of the AV groove. The posterior descending branch is the last branch before the crux. Beyond the crux (marked by the inverted U) are posterolateral left ventricular branches supplying the free wall of the left ventricle. In (D) both the origin and the terminal part of the right coronary artery are foreshortened but the sinus node artery (black arrowhead) can be distinguished from the conus and other right ventricular (RV) branches. As seen in (D) it runs posteriorly to the atria whereas the conal and RV branches run anteriorly to the right ventricle.

value of noninvasive imaging techniques (especially computed tomography [CT] and magnetic resonance imaging [MRI]) and the spectrum of invasive procedures (e.g. coronary stents). The development of new drugs that are effective for both preventing and treating IHD (e.g. statins, angiotensin-converting enzyme inhibitors) has added further impetus for the early diagnosis of IHD, including the diagnosis of asymptomatic disease. Consideration of imaging IHD therefore needs to be put in the context of the nature of the problem for the individual patient.

SELECTION OF IMAGING TESTS APPROPRIATE TO THE TASK Screening for asymptomatic ischaemic heart disease Atherosclerotic vascular disease is very common and most adults who lead a western lifestyle have some degree of atherosclerosis. The presence of risk factors such as hypertension, smoking, hypercholesterolaemia, diabetes and obesity identifies a large proportion of patients who are at increased risk

CHAPTER 25

for developing symptomatic IHD. Recognized risk factors can be treated but in intermediate risk cases the value of this may be unclear. For example, the cost-effectiveness of statin therapy for patients with mildly elevated serum cholesterol is unclear. The demonstration of an above average burden of coronary artery disease (CAD) may identify those in whom lipid lowering should be pursued more aggressively. Although this approach is still regarded as controversial, many advocate the use of CT for this purpose. Electron beam CT (EBCT) has been used for coronary calcification for over a decade and has high sensitivity for coronary artery calcification indicating the presence of atherosclerosis. The presence of high attenuation material in the coronary arteries can be quantified as the Agatston calcium score (Fig. 25.3). The problem of using this information is in determining what is normal, what is abnormal and whether a high calcium score indicates something that should be treated in the absence of symptoms and independent of other established risk factors, such as family history4.Various scoring systems and risk stratification protocols have been proposed5,6. Some studies have shown that a high calcium score can predict an increased risk of adverse coronary events. EBCT is not widely available and multidetector CT (MDCT) is now being used for coronary calcium screening. There is evidence that MDCT is more sensitive and shows less interstudy variability than EBCT7, but there is less experience with MDCT in predicting outcomes. As yet there is no good evidence that basing treatment on finding a high calcium score affects outcomes. Although both stress radionuclide imaging and stress echocardiograms have been used to screen for asymptomatic disease, in a low risk population there is very low sensitivity and specificity, which prevents widespread use. Both techniques are dependent on the presence of significant (flow limiting) coronary stenosis to cause wall motion or perfusion abnormalities, which are relatively uncommon in the absence of

•

ISCHAEMIC HEART DISEASE

symptoms. A large proportion (up to two-thirds) of myocardial infarctions are caused by plaque rupture on a nonobstructive (< 50%) lesion. Such lesions are unlikely to be detected by stress testing. Some aspects of cardiac MRI (CMR), such as plaque imaging, may eventually be applicable to assessment of asymptomatic IHD. Currently there is no generally agreed role for CMR in an asymptomatic population. Coronary angiography is also inappropriate in this context. Most professional society guidelines do not endorse screening using any form of imaging for IHD in asymptomatic patients.

Risk stratification in symptomatic ischaemic heart disease There are more options for imaging patients with symptoms of possible or proven IHD. Conventional approaches include stress ECG, stress radionuclide imaging and stress echocardiography. These are all dependent on producing stress-induced abnormalities in either the ECG tracing, wall motion, or perfusion. This requires a flow limiting coronary artery stenosis in at least one coronary artery. Although there are limitations with these techniques (ECG changes may be uninterpretable in bundle branch block; perfusion imaging may fail to show changes in diffuse three vessel disease), there is substantial experience and established value in identifying patients at higher risk. Further evaluation by coronary angiography may be indicated for a high risk Duke treadmill score, a large stress-induced perfusion defect, or an extensive stress echocardiography wall motion abnormality developing at a low heart rate (Table 25.1). Careful use of noninvasive testing should reduce the need for further invasive imaging by coronary angiography. In this respect there has been only partial success as up to 30% of patients undergoing coronary angiography have no significant coronary artery obstruction. Part of the problem in reducing the need for angiography in diagnosing IHD is the perception that angiography gives

Figure 25.3 EBCT screening. (A) EBCT image with conventional soft tissue windowing and with a threshold set at 130 HU. (B) Image shows an area of dense calcification (circled) in the proximal left anterior descending artery (LAD) in this apparently asymptomatic patient. The calcium score was high (total score 887; LAD score 227). On further questioning the patient remembered one episode of severe chest pain. (C) This prompted a contrastenhanced cine-EBCT study which showed poor left ventricular function with thinning of the cardiac apex, and apical calcification (arrow) indicating the presence of an apical aneurysm from an old anterior myocardial infarction and suggesting occlusion of the LAD at the site of the calcified plaque. (Reproduced with permission from Lee R T, Braunwald E (eds) 1998 Atlas of cardiac imaging. Current Medicine Inc/Churchill-Livingstone, Philadelphia.)

507

508

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 25.1 ACC/AHA GUIDELINES FOR CORONARY ANGIOGRAPHY32 Severe resting left ventricular dysfunction (LVEF < 35%)

higher contrast imaging at 3T, and blood pool contrast agents have substantially improved image quality, CMR coronary angiography is still not ready for routine use except in a few research programmes12.

High risk treadmill score (score ≥11) Severe exercise left ventricular dysfunction (exercise LVEF < 35%) Stress-induced large perfusion defect (particularly if anterior) Stress-induced moderate size multiple perfusion defects Large, fixed perfusion defect with left ventricular dilatation or increased lung uptake (201Tl) Stress-induced moderate size perfusion defect with left ventricular dilatation or increased lung uptake (201Tl) Echocardiographic wall motion abnormality (involving more than two segments) developing at a low dose of dobutamine or at a low heart rate Stress echocardiography evidence of extensive ischaemia

a more definitive diagnosis. Being able to see a widely patent artery is for many much more convincing than seeing normal perfusion or wall motion, even under stress. This has been one reason for developing techniques for noninvasive angiography with CTA or CMR. Although well established for noncoronary angiography, the accuracy and utility of coronary CT angiography (CTA) and CMR is still being evaluated. Performance depends on the hardware available and the protocols used. Current experience with coronary CTA using MDCT (with 16 or more rows of detectors) is very encouraging8,9. Although stenosis quantification remains a challenge, the ability reliably to demonstrate normal coronary arteries and detect stenosis can provide high positive (90%) and negative (95%) predictive values9. With adequate image quality, this allows reliable identification of patients who do not need coronary angiography, and may allow selection of those who need PCI at a specialist centre, or may only need a diagnostic coronary arteriogram as a prelude to CABG. In addition, in up to 10% of cases, MDCT can detect significant noncardiac diagnoses to explain symptoms (such as pneumonia and pneumothorax), which are not detectable by other cardiac imaging techniques. However, numerous pitfalls limit the reliability of MDCT in individual cases10. A slow heart rate (< 70 beats min−1) and freedom from artefact due to respiration or arrhythmia are essential. Beam hardening artefacts from calcification, stents or contrast medium in the superior vena cava can all limit assessment of important segments of coronary arteries10. Careful multiplanar reconstruction with a full awareness of the potential for artefact and good understanding of the 3D anatomy of the coronary arteries is essential11. With further improvements in performance expected as MDCT technology progresses, it will play an increasing role in cardiac diagnosis. Coronary angiography by CMR has been available for over 10 years but is generally still not sufficiently reliable for routine use. There are several important limitations. In addition to the usual contraindications to MRI (pacemakers, claustrophobia, etc), the need to establish reliable ECG gating can be difficult and the problems of compensating for respiratory motion remain a challenge. Although navigator sequences,

Stress testing: echocardiography, scintigraphy and cardiac magnetic resonance imaging Stress testing is widely used to determine the functional consequences of CAD.The stress used can be exercise, pacing, cold pressor testing, or pharmacological (e.g. dobutamine, dipyridamole or adenosine, alone or in combination with each other or with atropine)13,14. The use of stress testing is dependent on a cascade of events when there is a flow limiting stenosis. Perfusion defects are seen before the development of ischaemia causes a reduction in wall motion and myocardial thickening. These events precede ECG changes and are therefore more sensitive for detecting ischaemia than the conventional ECG-based exercise test. For imaging techniques, the end-point of the test is usually the appearance of new wall motion abnormalities (mainly stress echocardiography; Fig. 25.4) or a new perfusion defect (mainly scintigraphy; Fig. 25.5). Some state that a new wall motion abnormality is more significant as this is probably due to ischaemia, while a perfusion defect may reflect an attenuation artefact. Others state that sensitivity is better with perfusion imaging as abnormal blood flow occurs before the development of ischaemic dysfunction. Both precede the development of ECG changes. Wall motion can be reported as a general qualitative description of reduced, absent, or paradoxical motion or by calculating a wall motion score. Wall motion scores are produced by dividing the ventricular wall into segments (which requires that all segments are adequately visualized) and assigning a value to each segment. The best method of imaging wall motion depends on local expertise. Echocardiography is the most commonly used method15. In patients with poor acoustic access, both scintigraphic and CMR techniques can be used and combined with perfusion imaging16 and for CMR, coronary angiography16,17. This approach has significantly greater specificity and sensitivity in detecting CAD than simple exercise testing and is equivalent to myocardial perfusion imaging. Tagging sequences (Fig. 25.6) and strain imaging techniques when available can increase the accuracy and reproducibility of wall motion analysis18. Experience with echo contrast medium perfusion is limited but may provide a new option for perfusion imaging.

Characterization of known coronary artery disease Once the diagnosis of IHD has been made and a significant degree of risk established, either on the basis of symptoms or stress testing, further imaging may be required to define the most appropriate treatment. Conventionally this has been done by invasive catheter-based coronary angiography. This remains the most reliable and by far the most widely used technique for demonstrating coronary artery anatomy and determining the degree of coronary artery obstruction. Alternative methods using CTA and CMR remain investigational. The ability to characterize the content of plaque by CTA (Fig. 25.7) or

CHAPTER 25

•

ISCHAEMIC HEART DISEASE

Figure 25.4 Stress echocardiography. (A) End-diastolic apical two-chamber view at rest shows thinning of the apical myocardium (arrow). (B) Endsystolic apical two-chamber view at rest shows concentric contraction of left ventricle, except for the cardiac apex. (C) End-diastolic apical two-chamber view during stress (dobutamine infusion) shows dilatation of the left ventricle when compared with (A). (D) End-systolic apical two-chamber view during stress (dobutamine infusion) shows dilatation of the left ventricle due to akinesia of the anterior wall to contract (arrows). (E) Selective coronary arteriogram (right anterior oblique with cranial angulation) in the same patient shows a significant left anterior descending coronary artery stenosis (arrow). (Courtesy of Hani Aziz, MD.)

Figure 25.5 Stress nuclear perfusion imaging. Reversible inferolateral myocardial ischaemia. (A) Vertical long axis views. Upper row images obtained during stress (dipyridamole) show an area of reduced activity (arrows) not seen on resting images (lower row images). The imaging agent is 99m Tc-Tetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). (B) Upper row images show short axis images during stress (dipyridamole) with an area of reduced activity (arrows) not seen on resting images (lower row images). (C) Upper row images show horizontal long axis images during stress (dipyridamole) with an area of reduced activity (arrows) not seen on resting images (lower row images). Findings suggest ischaemia in the circumflex territory. (Courtesy of Laurie Gianturco, MD.)

509

510

SECTION 2

A

• THE CHEST AND CARDIOVASCULAR SYSTEM

B

C

D

Figure 25.6 Inferior myocardial infarction shown by short axis breath-hold cine-MRA and breath-hold tagging. (A) End-diastolic image shows minor narrowing of the inferior wall and the lower part of the interventricular septum (arrow). (B) End-systolic image shows normal thickening of the posterior and upper septal myocardium. There has been little thickening of the inferior wall and the lower part of the interventricular septum (arrow) representing the area affected by the inferior myocardial infarction. (C) End-diastolic, breath-hold tagging image shows no displacement of the orthogonal grid lines applied immediately before this image at the end of diastole. (D) End-systolic tagging image shows deformation of the tagging grid applied to the posterior and upper septal myocardium, indicating normal motion and thickening. There has been little distortion of the tagging grid, indicating poor contraction and thickening, in the region of the inferior myocardial infarction.

CMR may allow better risk stratification, but this remains to be established. Additional methods for assessing coronary stenosis using intravascular ultrasound (IVUS), intravascular Doppler and angioscopy may be combined with coronary angiography, especially when performing PCI. Several considerations need to be assessed during coronary angiography. These include the severity and length of stenosis, complete occlusions, collateral vessels, number of vessels affected, vessel diameter, stenosis configuration (smooth, ulcerated), presence of thrombus and variant anatomy. The relative roles of medical therapy, percutaneous intervention and surgery are constantly changing, are not universally agreed, and depend on locally available skills and facilities. For patients with chronic stable angina symptoms resistant to medical

therapy, suggestive findings on noninvasive testing, such as a high treadmill score, a large stress-induced perfusion defect, or wall motion abnormality at a low heart rate are indications for intervention. Many recommend early coronary angiography, taking the view that PCI with drug eluting stents is so effective that established guidelines for intervention are no longer applicable. In general, significant (> 50% diameter reduction) accessible stenoses in one or two vessels with a diameter of greater than 2 mm are treated by PCI. Left main coronary artery disease and significant disease affecting all three coronary arteries are usually best treated by CABG. Poor ventricular function increases the risk of both of these options but also increases the potential for benefit, as a poor prognosis is related to the degree of ventricular dysfunction. The greatest benefit from revascularization in terms of improving life expectancy is seen in those patients with significant left ventricular damage. The assessment of patients with unstable angina or nonST elevation myocardial infarction follows a different path. Risk stratification is required urgently and is currently based on the 12-lead ECG and cardiac biomarkers (e.g. troponin, creatine kinase-MB). This risk assessment may take a significant amount of time. Early experience with risk assessment using CTA suggests that this may be a cost-effective and rapid alternative for separating low from high risk patients. Having made the diagnosis, further risk stratification may be achieved by echocardiography or scintigraphy. Perfusion imaging by CT may have some value but this remains to be established19. Higher risk patients may benefit from early PCI, provided the appropriate skills are readily available.

Evaluation following coronary revascularization

Figure 25.7 Coronary CTA. Curved maximum intensity projection reconstruction from ECG-gated MDCTA of a complex left anterior descending artery stenosis showing both eccentric low attenuation plaque (arrowheads) and high attenuation calcified plaque (arrow).

Patients require re-study after PCI or CABG if symptoms have not been relieved or if they recur. Previously, graft patency was usually studied by selective catheterization, but experience with CTA and CMR indicates that graft patency can be determined with essentially equivalent accuracy (Fig. 25.8). However, with the exception of some research protocols, these techniques cannot give

CHAPTER 25

•

ISCHAEMIC HEART DISEASE

IMAGING Echocardiography

Figure 25.8 Coronary artery bypass graft CTA. Volume rendered reconstruction from ECG-gated MDCTA shows patent bypass grafts to the right coronary artery (saphenous vein, arrowheads) and left anterior descending coronary artery (internal mammary artery, arrows).

an effective indication of graft flow or reliably show the quality of the distal graft anastomosis and the distal native vessels. Complete obstruction of the graft will produce a small aortic protrusion at the graft origin and the grafted vessel may or may not fill by collateral flow. Selective angiography will show graft stenosis, the arterial status distal to the anastomosis and disease progression in the coronary arteries.

The main value of echocardiography in IHD is in demonstrating the effects of IHD on ventricular function and in detecting structural complications such as ventricular septal defect (VSD) (Fig. 25.9), papillary muscle dysfunction (causing mitral regurgitation) and ventricular thrombus (Fig. 25.10). Chronic ischaemia can cause diffuse dysfunction while acute myocardial infarction shows more localized changes. For patients who are poor echocardiography subjects, the use of echo contrast medium improves definition of the margins of the ventricular cavity and enhances the accuracy of echocardiography for assessing ventricular function (Fig. 25.10). In patients with reversible ischaemia, stress (exercise or pharmacologically induced) produces reversible wall motion abnormalities, which can indicate a need for revascularization (see Fig. 25.4). Using the development of a new wall motion abnormality as the criterion for ischaemia provides good specificity as this change is likely to reflect ischaemia, but sensitivity may be affected by deciding where to regard a change in wall motion as being a significant change. The accuracy of stress echocardiography (as with other stress imaging) is dependent on the threshold for defining significant stenosis, whether there is single or multivessel disease, whether adequate stress is achieved (especially for exercise-based protocols) and the presence of other disease processes which affect myocardial function (such as cardiomyopathy, microvascular disease, or hypertrophy). Sensitivity increases with a threshold set at 70% stenosis compared with 50%. Sensitivity for single vessel disease is less than for multivessel disease. For exercise echocardiography, with a threshold for significant disease of 50%, sensitivity is 71–97% with overall accuracy of 69–92%15. This compares with dobutamine echocardiography, with the same threshold for significant disease, where sensitivity is 70– 96% with overall accuracy of 76–92%15.

Figure 25.9 Infarct ventricular septal defect (VSD). (A) Long axis view from transthoracic echocardiography shows an 8-mm echo-free area (arrow) in the interventricular septum representing an infarct VSD. AoR = aortic root, LA = left atrium, LV = left ventricle, RV = right ventricle. (B) Equivalent view with colour Doppler shows variable colour (high velocity flow, arrow) in the infarct VSD. (Courtesy of Hani Aziz, MD.)

511

512

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.10 Left ventricular aneurysm and apical thrombus. (A) Apical four-chamber view from transthoracic echocardiography shows apical thrombus (arrow) complicating anterior myocardial infarction. (B) Enlarged view following intravenous injection of a contrast agent (Optison™, Perflutren Protein-Type A Microspheres for Injection, USP; GE Healthcare). The lumen of the left ventricle has become very echogenic and outlines the apical thrombus (between arrow tips) which is not enhanced by the contrast agent. (Courtesy of Hani Aziz, MD.)

Stress testing with echocardiography (and the other stress imaging techniques) is useful for risk stratification in patients with known or suspected IHD. A normal study indicates a very good prognosis with less than 1% incidence of cardiac events per year. Stress imaging is also used for risk stratification after myocardial infarction, pre-operative risk assessment and determining viability. Stress echocardiography for assessing viability (which can be complicated and is based on assessing changes at low and high levels of stress) has a specificity of 74–88% and specificity of 73–87%15. Performance is dependent on local skills and the ability to obtain a clear view during stress, which is not possible in 5–15% of patients. Replacement of myocardium by fibrosis is common in IHD. Fibrous tissue is stiffer than normal myocardium and it reflects ultrasound (US) better than normal myocardium. Regional loss of reflectivity has been detected in the septum after massive myocardial infarction in elderly patients. Its genesis is uncertain, but it may represent necrosis or myocardial haematoma. The origins of the main coronary arteries can usually be imaged by echocardiography in adults, and almost invariably in children, in the absence of thoracic deformity. Excellent images of proximal coronary arteries can be obtained by transoesophageal echocardiography (TOE). This is important in patients with anomalous origin of a coronary artery who present with severe left ventricular dysfunction, which must be distinguished from dilated cardiomyopathy or myocarditis. The distinction is important, particularly in infants and young children, since successful coronary artery surgery is frequently followed by substantial improvement in ventricular function. Echocardiography is also very useful in detecting coronary artery aneurysms in Kawasaki disease.

Cardiac computed tomography and computed tomographic angiography There have been substantial developments in noninvasive coronary artery imaging by CT and MRI. Cardiac imaging,

including coronary angiography and cine-ventriculography (see Fig. 25.3C) has been available for a long time with EBCT. Unfortunately several important problems prevented widespread adoption of this technology. More recently the development of MDCT systems with the ability to perform ECG gating and partial rotation image reconstruction has opened up a range of cardiac conditions to evaluation by CT.The ability to acquire imaging data in very thin slices allows the use of isometric voxels in 3D reconstruction and visualization of the coronary arteries from any relevant perspective (Fig 25.11). The current best MDCT machines allow submillimetre imaging with an ECG-gated acquisition time of 100–250 ms, still longer than the 50 ms or less obtainable with EBCT, but with better spatial resolution. Images can be obtained in a few heart beats, admittedly at the price of a significant dose of radiation when compared with other imaging procedures (Table 25.2).The issue of radiation dose is important, especially when applied to a potentially very large and possibly asymptomatic patient population. Comments in some publications and on the Internet that dose for cardiac CT is equivalent to only a few chest radiographs are misleading and poorly informed. Radiation doses are dependent on protocol, especially whether or not there is tube current modulation (which can nearly halve dose) and slice thickness. The value of CTA is well established in the evaluation of aortic and other vascular disease as well as some structural cardiac abnormalities, such as pericardial constriction, tumours and thrombus. With current MDCT performance, imaging of the more proximal coronary arteries allows reliable exclusion or detection of significant disease provided the heart rate is slow enough (< 80 and preferably < 70) and there is good breath-holding for the duration of the scan (typically < 10 s)20. Sensitivity and specificity for detecting significant (> 50%) coronary stenoses exceeds 90% with the best current MDCTA. Calcification can prevent assessment of stenosis (Fig. 25.12), although heavily calcified lesions are often stenosed, but a high score (> 400) strongly suggests a stenosis not detectable by MDCTA21.The ability to identify complex plaque

CHAPTER 25

•

ISCHAEMIC HEART DISEASE

Figure 25.11 Multiplanar visualization of complex coronary stenoses. (A). Oblique axial thin maximum intensity projection (MIP) reconstruction from coronary CTA and orientated along the left anterior descending artery (LAD) shows multiple high attenuation (calcified, arrows) and low attenuation (fatty, black arrowheads) lesions. Ao = aortic root, LA = left atrium, LV = left ventricle, PA = pulmonary artery. (B) Oblique coronal thin MIP reconstruction from coronary CTA shows more of the length of the LAD and multiple high attenuation (calcified, arrows) and low attenuation (fatty, arrowheads) lesions from a different perspective.

(alternating high and low attenuation areas) may be valuable in risk stratification if these are shown to represent the vulnerable plaques which cause acute thrombosis when plaque rupture occurs.When compared with IVUS there is good agreement for detection of coronary plaque. Table 25.2 RADIATION DOSE FROM CARDIAC IMAGING STUDIES COMPARED WITH OTHER IMAGING STUDIES20,39 Procedure

Effective dose (mSv)

Yearly background radiation

3.6

Chest radiograph

0.032–0.1

Skull examination

0.15

Lumbar spine series

3

Bone scintigram

4.4

Coronary arteriogram

2–6

EBCT calcium scoring

0.5–1

Coronary MDCTA

6–8 (constant tube current)

Although not often used for this purpose, MDCT multiphase reconstruction can provide cine images of ventricular and valve movement. Resting myocardial perfusion can be assessed and used to evaluate the size of acute myocardial infarction and recovery following reperfusion19. MDCT can also accurately determine CABG patency and may be able to determine the patency of the infarct-related artery after coronary thrombolysis.

Cardiac magnetic resonance imaging and magnetic resonance angiography Reliable coronary MR angiography (MRA) is still elusive but there are other important applications of CMR in patients with IHD. While echocardiography usually is the most appropriate method for assessing cardiac function and the complications of IHD, in those patients where echocardiography is difficult, breath-hold CMR is the next best alternative for demonstrating wall motion abnormalities (Fig. 25.13), thrombus, VSD, aneurysm, false aneurysm (Fig. 25.14), or valve dysfunction22. Figure 25.12 Heavy coronary calcification. (A) Axial thin maximum intensity projection (MIP) reconstruction from coronary CTA shows multiple areas of dense calcification which prevent visualization of most of the arterial lumen in the left anterior descending (LAD) (curved arrow), ramus intermedius (arrowhead), and circumflex (arrow) coronary arteries. An area of low attenuation in the ramus intermedius represents an area of clear occlusion by fatty plaque. (B) Oblique coronal thin MIP reconstruction shows more of the length of the LAD (arrow) but again due to heavy calcification, the lumen cannot be visualized. A = aortic arch, LA = left atrium, LV = left ventricle, PA = pulmonary artery.

513

514

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.13 True aneurysm. (A) Oblique two-chamber view of the left ventricle at end-diastole using cine-MRA (FISP) shows a dilated left ventricle. Note the relative diameters of the base (white arrow) and apical (black arrow) ventricle at end-diastole. (B) Oblique two-chamber view of the left ventricle at end-systole using cine-MRA (FISP) shows vigorous contraction of the basal part of the left ventricle (shortening of white arrow) but no contraction of the apical aneurysm (no change in black arrow).

Although this information can often be obtained by CTA, this requires significant radiation and iodinated intravascular contrast medium. CMR for these applications does not usually require contrast agent, and even when required, the risk is much less than that associated with iodinated contrast media. Tissue characterization to identify infarcted or nonviable tissue can be assessed in various ways. In the days after acute myocardial infarction there may be high signal on T2weighted imaging. As with MDCT, contrast-enhanced MRA can also accurately determine CABG patency23 and possibly that of the infarct-related artery after coronary thrombolysis24. In spite of many enthusiastic reports and individual cases, coronary MRA is not yet established as an accurate alternative to conventional coronary arteriography12. Direct comparison

of the best MDCTA and coronary MRA shows that these techniques have essentially equivalent accuracy (sensitivity and specificity about 80%, positive predictive value 40%, negative predictive value > 90%, overall accuracy about 80%), which may allow either to be used to rule out significant coronary obstruction25. MDCTA is a quicker study and is easier to use. Perfusion imaging by CMR has been developed over the past decade, has similar accuracy to nuclear techniques16, and can be combined with assessment of anatomy and wall motion at rest or with stress. Recent myocardial infarction also shows poor perfusion with first-pass contrast medium imaging, but delayed enhancement with gadolinium-based perfusion agents26,27. Delayed contrast enhancement (DE-MRI) or high T2 signal is not specific and can be seen in other conditions causing myocardial damage, such as sarcoidosis or myocarditis28. However, in the correct clinical context, MRI is a useful method for assessing the size of a myocardial infarction, as well as the consequences of infarction, such as left ventricular dysfunction or aneurysm which are well shown by cine-MRA and tagging sequences18 (see Fig. 25.6). Breath-hold DE-MRI using patient-specific inversion times (TI) to null signal in the normal myocardium (Fig. 25.15) is widely used for assessing viability when planning revascularization and may be at least as accurate as viability imaging using positron emission tomography (PET)28,29. In the acute phase following myocardial infarction, DE-MRI is probably the best method for determining the size of the infarction.

Nuclear cardiac imaging The principles of cardiac scintigraphy are described in Chapter 22. Nuclear cardiac imaging is a first-line technique for evaluating myocardial perfusion, viability and function. Planar nuclear imaging has been largely replaced by single-photon emission computed tomography (SPECT) which allows attenuation and scatter correction and 3D data acquisition, all of which help to overcome some of the most significant limitations of planar

Figure 25.14 False aneurysm. (A) Axial spin-echo MRI image shows a large false aneurysm (FA) arising from the posterior wall of the left ventricle (LV). Note the abrupt change in contour at the mouth of the FA (arrow). (B) Axial cine MRA image at the same level as (A) shows flow into the FA (arrows).

CHAPTER 25

•

ISCHAEMIC HEART DISEASE

Figure 25.15 Delayed contrast enhancement (DE)-MRI in nonviable myocardium. (A) End-diastolic image from short axis cine-MRA shows a dilated and thin walled left ventricle. (B) End-systolic image from short axis cine-MRA at the same level as (A) shows that the left ventricle has hardly contracted at all. In particular there has been little thickening of the superoseptal and, to a lesser extent, inferobasal myocardium (arrows). (C) Short axis DE-MRI image shows delayed full thickness (arrow) or > 50% thickness (arrowhead) contrast enhancement indicating nonviable myocardium. The TI has been chosen to null signal from the viable posterior wall myocardium.

scintigraphy. Combining SPECT with CT allows attenuation correction while providing simultaneous and anatomical information. New perfusion agents are being developed which may improve quantification of perfusion, while molecular imaging agents may identify vulnerable plaque. Thallium was long the dominant myocardial perfusion agent but has largely been replaced by 99mTc-labelled agents, such as SestaMIBI (99mTc-methoxyisobutyl-isonitrile), Tetrofosmin and Teboroxime. 99mTc has better imaging characteristics (higher energy, less scatter, and shorter half-life, allowing higher doses). The choice of agent depends to some extent

on local preference, instrumentation and experience. In spite of the theoretical advantages of 99mTc-labelled agents, there is little formal evidence of substantial superiority over thallium, and some imaging protocols use a combined 99mTc/thallium approach for stress testing. Using SPECT rather than planar imaging improves accuracy significantly. Myocardial perfusion imaging depends on differences in flow delivering different amounts of activity to normal and ischaemic myocardium. Resting perfusion abnormalities occur in areas of myocardial infarction (Figs 25.16, 25.17) or fibrosis, or in areas supplied by a very severe (> 85%) stenosis.

Figure 25.16 Extensive antero-apical myocardial infarction. (A) Fixed perfusion defect in a patient with large antero-apical myocardial infarction as shown by severely reduced activity (arrow) along the anterior surface and extending around the cardiac apex on the horizontal long axis view. The left ventricle is dilated. The imaging agent is 99mTc-Tetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). (B) Fixed perfusion defect (arrow) in the anterior and septal myocardium on short axis view. (C) Fixed perfusion defect (arrow) along the septal surface and extending to the cardiac apex on the vertical long axis view. The left ventricle is dilated. (Courtesy of Laurie Gianturco, MD.)

Figure 25.17 Limited inferobasal myocardial infarction. (A) Fixed perfusion defect in a patient with inferobasal myocardial infarction shows reduced activity in the inferobasal wall (arrow) on the horizontal long axis view. The imaging agent is 99mTcTetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). (B) Fixed perfusion defect in the inferior wall (arrow) on short axis view. (C) Fixed perfusion defect in the inferior wall (arrow) on the vertical long axis view. (Courtesy of Laurie Gianturco, MD.)

515

516

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Because many significant stenoses may allow normal or near normal blood flow at rest, stress is often used to increase sensitivity for rate-limiting stenoses. Stress increases the difference between flow in normal and abnormally perfused myocardium (Fig. 25.5). Stress can be induced by exercise (usually on a treadmill) or, in patients who cannot exercise sufficiently to produce a significant difference in blood flow detectable by nuclear imaging, pharmacologically using vasodilators such as adenosine (which has the advantage of being very short acting) or dipyridamole. Given the choice, most patients prefer exercise. Other stress agents such as dobutamine work better with techniques used to assess abnormal wall motion (echocardiography, cine-MRA). Current indications for pharmacological perfusion imaging include an inability to exercise, pre-operative risk assessment, postinfarction risk assessment, left bundle branch block and fixed rate pacemaker. The latter two prevent assessment of ECG changes with ischaemia. Although there is substantial experience with nuclear perfusion imaging in risk assessment, there are important potential sources of error, apart from misinterpretation of the perfusion images or the correlative angiograms.These include inadequate stress or a delay in imaging after stress, medications that reduce ischaemia, non-flow limiting stenosis, overlap or collateral circulation, balanced hypoperfusion, or hypoperfusion in areas with attenuation. PET has been available for many years but the original agent, 18F-fluorodeoxyglucose (FDG), requires a cyclotron for it’s generation. PET allows better attenuation correction and may produce improved spatial resolution. More widespread use of perfusion imaging has become possible with the availability of rubidium-82 from a generator. Rubidium82 may have better specificity than SPECT for detecting coronary artery disease. Many regard PET (using rubidium82 to assess flow and FDG to assess glucose metabolism) as the best method for determining myocardial viability. Others believe that DE-MRI is equivalent or even superior to PET for assessing myocardial viability, and has superior spatial resolution.

Conventional coronary arteriography Conventional coronary arteriography requires selective injection of contrast medium into the right and left coronary arteries and into the left ventricle, while recording the resultant moving images. Percutaneous femoral arterial catheterization is the usual approach and requires at least three shaped catheters, one for each coronary artery, and one for the left ventricle (usually a pigtail configuration). Multiple catheter configurations are available to address different anatomical configurations. An alternative percutaneous route in patients without peripheral arterial disease is the radial artery. Percutaneous techniques have a very high success rate and usually a lower complication rate than the cutdown, transbrachial Sones method, which is now seldom used. Low osmolality contrast media (LOCM) are safest in that their haemodynamic and ECG effects are low. Iso-osmolar agents are possibly the safest of all30. Early reports of thrombosis occurring more commonly with nonionic contrast

media were probably a reflection of poor angiographic technique. Typically 3–10 ml of LOCM is injected for selective coronary angiography, while for left ventriculography, a power injection of 24–40 ml at 8–12 ml s−1 is typically used. Rapid filming at 25 or more frames s−1 is needed to demonstrate cardiac motion and requires cine film or digital acquisition. To best demonstrate the complex 3D anatomy of the coronary arteries, images are acquired in multiple orientations with appropriate oblique and craniocaudal angulation.This requires at least three views for the right coronary artery and five views for the left coronary arteries. Powerful X-ray generators are required to allow very short exposures (5–10 ms) to freeze cardiac motion and deliver exposures optimized to give the best image contrast. The overall mortality of coronary angiography is about 0.2% for elective cases31,32 but there are wide variations in complications depending on the population being studied, the experience of the operator, and the performance of PCI. The mortality rate may rise to nearly 1% if there is major left main coronary artery disease or severe left ventricular dysfunction. The use of anticoagulants or platelet inhibitors during PCI increases the risk of haemorrhagic complications. Conversely, in patients with normal hearts undergoing diagnostic coronary angiography, the mortality rate is very low in experienced hands. Complications of coronary angiography are listed in Table 25.3.

Table 25.3 COMPLICATIONS OF CARDIAC ANGIOGRAPHY31,32 Vascular Haematoma False aneurysm Arteriovenous fistula Mycotic aneurysm Retroperitoneal haematoma Acute occlusion Arterial dissection

Cardiac Arrhythmias (catheter manipulation) Myocardial infarction Coronary dissection Systemic embolus (including stroke) Myocardial perforation

Contrast medium Heart failure (reduced with low osmolality contrast agent) Arrhythmias (sinus bradycardia, sinus arrest, ventricular fibrillation in 0.1–1.0%, mainly with wedged coronary injections) ECG changes (wide QRS, long QT, ST changes, change in QRS axis) Hypotension (reduced with low osmolality contrast agent) Allergic/idiosyncratic (reduced with nonionic contrast agent) Renal impairment (possibly reduced with low osmolality contrast agent)

CHAPTER 25

CORONARY ARTERY ANATOMY An understanding of the coronary artery anatomy is essential to interpretation of cardiac imaging. This applies to coronary imaging (CTA, CMR, coronary angiography), perfusion imaging (echocardiography, radionuclide imaging, CMR) and stress testing (echocardiography, radionuclide imaging, CMR). Usually two coronary arteries arise from the aorta (Figs 25.1, 25.2). The left coronary artery originates from the left coronary sinus of Valsalva, which lies on the left and slightly posteriorly (Fig. 25.2A,B). The right coronary artery originates from the right coronary sinus of Valsalva (Fig. 25.2C,D), which lies anteriorly and only slightly to the right.

Left coronary artery After a variable length and direction passing to the left, the left coronary artery divides into the left anterior descending (LAD) and circumflex (Cx) coronary arteries (Fig. 25.2A,B). The LAD passes forwards in the anterior interventricular groove on the front of the heart, giving off parallel straight descending branches to the interventricular septum. Branches passing to the left from the LAD to the free (lateral) wall of the left ventricle are called diagonal branches. The Cx passes backwards and leftwards, lying in the left atrioventricular groove. The atrioventricular groove, often filled with fat, encircles the surface of the heart and marks the plane of the atrioventricular valves. The Cx gives rise to one or more obtuse marginal branches to the lateral wall of the left ventricle and continues as a much reduced vessel to the inferior surface of the heart, terminating before the inferior interventricular groove. In about 10% the left main coronary has a trifurcation with an intermediate or ramus medianus artery, arising between the LAD and Cx coronary arteries (Fig. 25.12A).This vessel is effectively a diagonal branch serving the lateral aspect of the left ventricle.

Right coronary artery After originating from the right coronary sinus ofValsalva on the anterior aspect of the ascending aorta, the right coronary artery (RCA) passes to the right to lie in the anterior atrioventricular groove (Fig. 25.2C,D). The RCA encircles the right side of the heart to reach the crux (the junction of the atrioventricular and the interventricular septal planes) on the posterior undersurface. The RCA then rises vertically in a loop into the myocardium (the inverted U-loop of the crux; Fig. 25.2), descends again, and sometimes continues on as a small branch in the atrioventricular groove on the left. Its first branch is usually the conus branch to the front of the right ventricular conus, although this branch frequently has a separate origin from the right coronary sinus of Valsalva.The second branch is the sinus node artery which arises from the RCA in about two-thirds of cases and passes posteriorly to the interatrial septum to supply the sino-atrial node.The right ventricular (acute marginal) branches pass forward and to the left to supply the right ventricle. The last branch before the inverted U of the crux is usually the posterior descending coronary artery, which passes forward in the inferior interventricular groove on the inferior

•

ISCHAEMIC HEART DISEASE

surface of the heart to supply the lower part of the ventricular septum and the adjacent ventricular walls. The posterior descending branch can be recognized by the straight, vertical septal vessels arising from it. At the apex of the inverted ‘U’ of the crux, a straight short vessel, the atrioventricular nodal branch, rises into the septum. When the RCA continues in the left atrioventricular groove beyond the crux it gives off forward-running posterolateral left ventricular branches to the free wall of the left ventricle.

Coronary artery dominance The arrangement in which the RCA reaches the crux of the heart and supplies the posterior descending artery occurs in about 85% and is termed right dominance (Fig. 25.2). Dominance indicates the coronary artery that supplies the posterior descending branch to the inferior surface of the heart. Variations of balance in the coronary arteries are common. In about 8%, the Cx becomes a large trunk passing in the left atrioventricular groove to reach the crux, and gives off the posterior descending and usually the atrioventricular nodal branches; this arrangement is termed left dominance.

Congenital variations of coronary anatomy Congenital variations of coronary anatomy are common and may be best demonstrated by CTA or CMR22,33. In most cases their main importance is because they cause difficulty in performing coronary arteriography and PCI. Major congenital variants of coronary anatomy occur in about 1–2% of patients. The most common major variant (other than variations of dominance), which represents about 60% of all major congenital anomalies, is for the Cx to originate from the RCA or as a separate artery from the right coronary sinus (Fig. 25.18); occasionally this aberrant vessel may be diseased (Fig. 25.19). Congenital variations of anatomy are more important when they lead to ischaemia. This may occur if the anomalous vessel passes between, and may be compressed by, the pulmonary artery and the aortic root. Compression of the anomalous vessel may cause angina, arrhythmias, myocardial infarction, or sudden death (Fig 25.20). Demonstration of this potentially lethal course is often best shown by CTA (Fig 25.19) or CMR22,33.

Figure 25.18 Anomalous circumflex coronary artery. Selective coronary arteriogram (left anterior oblique) following injection into the ostium, from the right coronary sinus of Valsalva, of an anomalous circumflex coronary artery (arrow) passing inferior and to the right of the aortic root.

517

518

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.19 Anomalous origin of stenosed left coronary artery. (A) Axial maximum intensity projection (MIP) reconstruction from ECG-gated MDCTA shows an abnormally large origin to the right coronary artery (RCA) which arises directly anteriorly (arrow) and the course of an anomalous left main coronary artery (LMCA, arrowheads) running between the left atrium (LA) and the aortic root (Ao). There is no left coronary artery in the normally expected area (asterisk) adjacent to the left sinus of Valsalva (SV). (B) Axial MIP reconstruction from ECG-gated MDCTA immediately inferior to (A) shows an abnormal bifurcation (white arrow) of the RCA giving off the anomalous LMCA (black arrow) running between the left atrium (LA) and the aortic root (Ao). Note the effect of the beam hardening effect from dense contrast medium in the right atrium (asterisk) obscuring part of the proximal LMCA. In most cases this anomaly has a benign prognosis. In this case an atherosclerotic stenosis (black arrow) has developed causing symptoms. (C) Sagittal MIP reconstruction from ECG-gated MDCTA showing the curve of the anomalous LMCA (arrow) as it passes under the right sinus of Valsalva.

MANIFESTATIONS OF ATHEROMATOUS CORONARY ARTERY DISEASE Coronary artery stenosis The normal coronary artery is a smooth, gently tapering, branching structure.The normal lumen is preserved as atherosclerosis develops with increasing thickness in the vessel wall being accompanied by an increase in outer diameter. It is only in the late stages that a stenosis develops. As time passes areas of calcification develop. Calcification indicates the presence of atherosclerosis and the greater the calcification the greater the likelihood of there being a stenosis, although not necessarily at

Figure 25.20 Anomalous left main coronary artery causing myocardial infarction. Selective coronary arteriogram (left anterior oblique) following injection into the ostium, from the right coronary sinus of Valsalva, of an anomalous left main coronary artery (arrow) passing between the pulmonary artery and the aorta. This led to compression of the anomalous left main coronary artery and was complicated by myocardial infarction. The position of the mitral valve is shown by the sewing ring of a valve replacement.

the same place. The development of calcification in atheroma provides the rationale for using CT to detect calcification as a method for screening for asymptomatic CAD. Atheroma causes irregularity which progresses to stenoses as plaques spread circumferentially, narrowing the arterial lumen (Fig. 25.21). A reduction of lumen diameter of over 50–60% is thought to be sufficient to reduce flow on demand and to represent a haemodynamically significant stenosis. Because atheroma is

Figure 25.21 Progression of atheromatous coronary artery disease. (A) Right coronary artery, first arteriogram. The outline of the contrast-filled artery is irregular due to atheromatous plaques. The arrowhead indicates a large atheromatous plaque. (B) Right coronary arteriogram, same patient, 2 years later. The previously localized deposit of atheroma now encircles the lumen of the artery, producing a localized stricture (black arrowhead).

CHAPTER 25

usually diffuse in the arterial wall, although not necessarily causing stenosis, arteriography tends to underestimate the degree of atheroma and the severity of the stenosis. Acute occlusion, usually due to plaque rupture with subsequent thrombosis, prevents the passage of contrast medium but, unless an arterial stump is identified, the findings should be interpreted with caution as apparent absence of flow beyond a stenosis may be caused by collateral (or CABG) flow. In the presence of coronary artery obstruction, anastomoses around the right ventricular conus, between the RCA and the anterior descending artery, can form the ring of Vieussens (Fig. 25.22). Septal anastomoses develop in the ventricular septum between branches of the anterior descending and posterior descending arteries, or between atrial branches proximally and distally. An anastomosis between an atrial branch and the atrioventricular nodal artery is called Kugel's artery. Anastomoses may also occur by direct terminal communications of large arteries.

Coronary artery aneurysms Most aneurysms of the coronary arteries are atheromatous (Fig. 25.23). They may be localized but are more usually part of generalized ectasia or arteriomegaly. Men are more commonly affected than women. Aneurysms may lead to symptoms or sudden death by rupture, pressure on the parent artery, or distal embolization. False aneurysm may complicate enthusiastic PCI (especially with cutting balloons or atherectomy). Aneurysms and strictures of the coronary arteries may also be seen in children in the mucocutaneous lymph node syndrome (Kawasaki disease) (Fig. 25.24). These patients present with fever and swollen cervical lymph nodes: they may have ischaemic pain and ECG abnormalities. Coronary aneurysms can often be fully evaluated using CTA or CMR.

•

ISCHAEMIC HEART DISEASE

Figure 25.23 Coronary artery ectasia. Left coronary arteriogram (right anterior oblique) shows marked dilatation of the left anterior descending (LAD, solid arrow) and circumflex (open arrow) coronary arteries. Notice the poor opacification of the distal LAD due to the effect of gravity in preventing flow of the heavy contrast medium into the more anterior segment of the vessel.

of myocardial infarction where it forms on an area of rupture of an unstable plaque. Thrombus may occasionally form in a vessel of normal diameter. Dissection is common after percutaenous transluminal coronary angioplasty (PTCA)—in fact it is part of the mechanism of PTCA, but in this circumstance it is often of no consequence unless it causes obstruction to flow (Fig. 25.25). Spontaneous dissection can occur and cause chest pain or sudden death, either in association with aortic dissection or in isolation.

Ventricular function

Thrombus within the lumen of the coronary arteries is common, and commonly underestimated by angiography. Usually thrombus is associated with a stenosis and is the typical cause

The degree of emptying of the left ventricle is reflected by the ejection fraction, which is the stroke volume divided by the end-diastolic volume. The normal left ventricle ejects about two-thirds of its content with each beat, to produce an ejection fraction of 66%. Localized abnormalities of contraction of the wall of the left ventricle may be recognized in patients

Figure 25.22 Vieussen's ring. Anterior view following injection into an occluded right coronary artery (arrow) shows a collateral vessel, Vieussen's artery (open arrow), opacifying the left coronary artery (curved arrow) distal to a severe proximal stenosis. The black arrow indicates the aortic catheter.

Figure 25.24 Kawasaki disease. Ascending aortogram (left anterior oblique) showing a normal ascending aorta with aneurysms of the proximal right (solid arrow) and left (open arrow) coronary arteries.

Intraluminal lesions

519

520

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.25 Coronary artery dissection complicating stent deployment. Oblique axial maximum intensity projection (MIP) reconstruction from ECG-gated MDCTA in a patient with impaired flow following stenting of the coronary artery circumflex coronary artery shows a faint dissection flap (arrow) in the left main coronary artery proximal to the stent in the circumflex (between arrowheads). Ao = aortic root, LA = left atrium, PA = pulmonary artery.

with areas of ischaemia or scarring from IHD, although the changes are not specific. Reversible wall motion changes can occur with stress-induced ischaemia. Impaired wall motion is termed hypokinesia (reduced wall motion), akinesia (absent wall movement; see Fig. 25.13) and dyskinesia (paradoxical wall movement). In the resting state, localized contraction abnormalities usually represent areas of infarction. Slow recovery of previously ischaemic and nonfunctioning myocardium may occur, the ‘stunned myocardium’. Significant overall impairment of the left ventricle is present if the ejection fraction falls below 50%, severe impairment is represented by an ejection fraction below 30%, and very severe impairment by an ejection fraction below 10%. The severity of left ventricular dysfunction is one of the most important prognostic indicators of survival after myocardial infarction or in patients with cardiomyopathy. Ventricular function can be demonstrated by echocardiography, first-pass or gated blood-pool radionuclide imaging, EBCT, MDCT, cine-MRA, or conventional left ventriculography. Multilevel cine-MRA is probably the most accurate method for comprehensively assessing ventricular function (see Figs 25.6, 25.13, 25.15). Tagging sequences can be used to highlight abnormal regional wall motion (see Fig. 25.6) and reduced myocardial thickening in systole18.

Myocardial stunning and myocardial hibernation Ventricular dysfunction persists in spite of there being patent coronary arteries in two conditions in IHD. Myocardial stunning refers to prolonged but temporary ventricular dysfunction that follows a period of ischaemia. There is abnormal ventricular function but the myocardium is viable, has

contractile reserve, and can regain normal function with revascularization. Stunning can be seen after relief of ischaemia by thrombolysis, PCI, coronary bypass grafting, reversal of vasospasm, or after exercise. Stunning can be identified by recovery of left ventricular function during extended pharmacological stress testing with imaging by echocardiography, radionuclide imaging, or cine-MRA. Myocardial perfusion imaging using contrast echocardiography, radionuclide imaging, or first-pass MRI will identify areas of perfused but ischaemic myocardium that show impaired function but may benefit from revascularization16. Infarcted, nonviable myocardium can be identified as it fails to recover systolic function on extended pharmacological stress, retains contrast on DE-MRI (see Fig. 25.15), and does not recover function after reperfusion. Reperfusion of viable myocardium improves ventricular function and survival34. Myocardial hibernation is a more chronic condition resulting from months or years of ischaemia causing ventricular dysfunction that persists until normal blood flow is restored. The affected myocardium shows contractile reserve which, as with stunning, allows differentiation from ischaemia using pharmacological stress during assessment of left ventricular function by echocardiography, radionuclide imaging, or cine-MRA35. The dysfunction is reversible with revascularization. Perfusion imaging with echocardiography, radionuclides, CMR, or DE-MRI can also be used to differentiate hibernating myocardium from infarct, and to identify those patients who will benefit from revascularization of a dysfunctional area.

Angina pectoris Angina pectoris, caused by reversible myocardial ischaemia, is a clinical syndrome of periodic chest pain, usually of a crushing nature, often with characteristic radiation to the left arm or jaw. Typically it is triggered by exercise, cold, or emotion, and relieved by rest or nitrates. Pain can occur spontaneously. Pain with spontaneous onset associated with ST segment elevation is termed Prinzmetal's or variant angina. ECG changes may precede and outlast the pain, or even occur without pain, and the pain may occur without ECG changes. Angina pectoris with apparently completely normal coronary arteries is known as syndrome X and is usually thought to be due to spasm. Angina pectoris may be investigated using stress ventriculography (echocardiography, first-pass or equilibrium radionuclide, cine-MRA) or stress perfusion imaging (contrast agent, echocardiography, radionuclide, contrast CMR). The most widely used perfusion tests use radionuclides to identify perfusion defects at rest and those developing only with stress (see Fig. 25.5). In reversible ischaemia, thallium-201 radionuclide scintigraphy, first on exercise (or drug stimulation) and then at rest, shows perfusion defects on stress that fill in at rest (Fig. 25.26). Thallium radionuclide scintigraphy is less specific than hoped. SestaMIBI can be used in the acute phase of myocardial infarction to reveal perfusion deficits within the myocardium and their evolution over time, as well as in the assessment of angina, although stress and rest images need to be acquired on different days. A hybrid thallium/SestaMIBI protocol can be completed on the same day.

CHAPTER 25

Figure 25.26 Reversible ischaemia. Reversible distal anteroseptal and apical myocardial ischaemia on vertical long axis views. Upper row images are during stress (dipyridamole) with an area of reduced activity (arrows) not seen on resting images (lower row images). This suggests ischaemia in the distal left anterior descending artery (LAD) territory. The imaging agent is 99mTc-Tetrofosmin (Myoview, Amersham Health, Arlington Heights, IL). (Courtesy of Laurie Gianturco, MD.)

Angina may occur with other cardiac disorders, especially where there is an increased demand for myocardial blood flow, even when the coronary arteries are normal. It may occur in any valvular heart disease, though most commonly in severe aortic stenosis. Angina may be a feature of disease of heart muscle, especially in hypertropic cardiomyopathy where the coronary arteries are usually rather large and rarely stenosed but the myocardium demands a greater blood flow. In hypertrophied hearts, increased wall tension during diastole, when most myocardial blood flow occurs, may also limit perfusion and cause symptoms. Angina may also develop in patients with tachyarrhythmias, usually when there is underlying CAD.

Dressler's syndrome About 10–30 d after myocardial infarction or cardiac surgery, some patients develop chest pain associated with a high sedimentation rate. This may be due to Dressler's syndrome, characterized by the triad of pleuritis (small pleural effusion), pneumonitis (ill-defined basal lung shadows) and pericarditis. Pericardial involvement may progress to a substantial pericardial effusion and rarely to tamponade. The syndrome may remit and relapse over weeks or months. It should not be confused with extension of the myocardial infarction or with pulmonary embolism with pulmonary infarction. It usually responds dramatically to aspirin or steroids, a useful distinguishing feature.

MYOCARDIAL INFARCTION Coronary arteriography after myocardial infarction Most patients with myocardial infarction have some obstructive CAD, which may vary in extent from a lesion in only one artery to multiple lesions in all three main coronary arteries. The infarct-causing lesion is not necessarily the most severe

•

ISCHAEMIC HEART DISEASE

stenosis. The majority of myocardial infarctions are caused by plaque rupture involving less than 50% stenosis. Myocardial infarction can, however, occur in the presence of apparently normal coronary arteries, and then has a rather better prognosis. The natural history of coronary thrombosis is complex36,37. Plaque fissuring occurs in a cholesterol-rich atheromatous lesion, which is not necessarily very large or flow limiting. Thrombosis develops on the exposed intimal area. This leads to infarction, or less frequently to unstable angina. Thrombus may resolve spontaneously or after thrombolytic treatment. It can be visualized as an intraluminal filling defect seen on coronary arteriography. If the underlying plaque causes significant stenosis, angina or re-infarction may occur; this requires treatment with PCI (Fig. 25.27) or, with severe multivessel involvement, CABG.

Indications for coronary arteriography In many patients it is clear that the diagnosis is chest pain due to atheromatous CAD, and coronary arteriography is performed to guide management. Coronary arteriography is essential in the anatomical assessment of disease severity and deciding whether CABG or PCI is appropriate. CABG surgery or PCI is indicated for patients with: • severe CAD and angina which is not responding to medical treatment • life-threatening disease such as a severe left main coronary artery stenosis • angina persisting after myocardial infarction • positive stress test and severe disease of the corresponding coronary arteries • acute myocardial infarction when access to expert PCI is rapidly available.

Surgery The usual operation is a left internal mammary artery graft to the LAD, and reversed saphenous vein aortocoronary grafts (see Fig. 25.8) to the other arteries. There is increasing use of the right internal mammary artery, gastro-epiploic artery and other arterial conduits for bypass grafting as these have better long-term patency rates than conventional vein grafts. Coronary arteriography is an essential roadmap for surgery, indicating the number and sites of stenosed vessels to be bypassed and the state of the vessel at the potential level of graft insertion. Small and irregular vessels beyond the stricture may indicate diffuse disease and suggest a difficult graft procedure. Assessment of left ventricular function is required, as operative mortality is higher in patients with poor left ventricular function, but if untreated, prognosis is also poor. Coronary arteriography using EBCT, MDCT, or CMR may in some cases show coronary stenoses with sufficient clarity to allow treatment to be decided without conventional coronary arteriography8,9,24.

Mechanical complications Most patients who die from CAD develop a fatal arrhythmia, or have suffered so much myocardial infarction that insufficient

521

522

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.27 Acute myocardial infarction treated by mechanical thrombolysis and stenting. (A) Left anterior oblique coronary arteriogram shows thrombotic occlusion (asterisk) of the proximal right coronary artery. (B) Subsequent coronary arteriogram in the same orientation shows a patent vessel following mechanical thrombolysis (Angiojet, Boston Scientific, Natick, MA) and subsequent stent deployment (between arrows). (Courtesy of Gregory Giugliano, MD.)

muscle remains to support life. Some patients develop complications that interfere with effective ventricular contraction, although sufficient contractile muscle remains to maintain the circulation. Such patients may have their heart failure relieved by surgery.

Cardiac rupture Rupture is a not uncommon complication. The myocardium ruptures and pericardial tamponade leads rapidly to death.There is usually little opportunity for imaging. Very rarely, rupture is subacute, leading either to pericardial effusion with tamponade or to the rupture being walled off within the pericardium, producing a false or pseudoaneurysm (see Fig. 25.14).

Left ventricular aneurysm Left ventricular aneurysm is the most common mechanical complication encountered in routine practice. The definition of ‘aneurysm’ varies according to different experts; hence there are different values for the incidence of symptoms and prognosis. The pathological definition is a large thin-walled fibrous sac, bulging not only from the lumen of the left ventricular cavity but also from the external surface of the heart, and usually clearly demarcated from normal myocardium. The clinical findings include those of left ventricular failure, angina, persisting arrhythmias and systemic embolism. Persisting Q waves and ST segment elevation are seen on the ECG. Chest radiograph findings include a normal cardiac silhouette, diffuse left ventricular enlargement, or an obvious focal bulge from the left ventricle. Curvilinear calcification in the wall of the aneurysm develops after several years (see Fig. 25.3). The lungs may show evidence of cardiac failure.

False or pseudoaneurysm (see Fig. 25.14) results from a localized perforation of the ventricular myocardium, following myocardial infarction or trauma. It may originate from unusual parts of the left ventricle (e.g. high on the anterior wall), and may have a narrow neck; it fills and empties slowly with contrast medium. Cross-sectional imaging shows an abrupt change in wall thickness, often with an abrupt angulation at the mouth of the false aneurysm. There may be extensive thrombus. There is no myocardium in the wall of the false aneurysm, and unlike a true aneurysm there is no enhancement on DE-MRI. False aneurysms have a tendency to rupture and should be excised. Echocardiography, CT and MRI can distinguish left ventricular aneurysm from global enlargement (see Figs 25.10, 25.13). Aneurysms are fibrous and thus akinetic; the limited elasticity of collagen fibres is such that they cannot expand significantly with the pressures normally developed by the left ventricle. Apparent paradoxical motion results in part from optical illusion and in part from overall motion of the heart in space. In making the diagnosis of an aneurysm, therefore, reversal of curvature at the ‘neck’ must be shown (see Fig. 25.13). This is present during systole and may persist during diastole. There is marked thinning of the wall of the ventricle and there may be adherent thrombus in the aneurysm. Mitral regurgitation is frequently present. Nonviable myocardium in the wall of a true aneurysm shows enhancement on DE- MRI (see Fig. 25.15).

Postinfarction ventricular septal defect Large myocardial infarcts involving the ventricular septum may lead to rupture, establishing a large interventricular shunt.

CHAPTER 25

Defects may occur anywhere in the muscular septum but are most frequently seen towards the apex. Acute volume overload may precipitate severe cardiac failure. The usual clinical story is of recovery from recent anteroseptal infarction with sudden deterioration associated with the appearance of a systolic murmur.The prognosis is poor and many patients need immediate surgery to close the defect. Occasionally the diagnosis may be suggested fortuitously on early imaging post myocardial infarction where extreme septal thinning or even dissection can be seen on the echocardiogram or CMR (Fig. 25.28). The chest radiograph shows a large heart and pulmonary oedema. Patients who survive with the shunt intact, may have severe pulmonary plethora. The diagnosis is usually made by echocardiography. Acquired VSD following acute myocardial infarction can be demonstrated by transthoracic echocardiography (see Fig. 25.9) or TOE. Acquired VSDs can be identified on Doppler by the high velocity jet arising from the septum and visible in the right ventricle (Fig. 25.9B). Peak velocity on continuous wave (CW) Doppler gives an indication of the peak systolic pressure difference across the septum, and thus of peak right ventricular pressure, once systemic arterial pressure is known. An intracardiac shunt may be shown by first-pass nuclear ventriculography and will be seen as a defect on the ventricular septum associated with a flow void (due to turbulent flow) on cine-MRA.

Postinfarction mitral regurgitation Acute mitral regurgitation may be caused by papillary muscle dysfunction or chordal rupture. A systolic murmur following inferior myocardial infarction is common and is presumed to be due to mitral regurgitation from ischaemia of the posterior (also called inferior) papillary muscle. Occasionally, inferior infarction leads to necrosis and rupture of the posterior papillary muscle leading to severe acute mitral regurgitation, and death is usual unless urgent surgical repair is possible. Clinical presentation and timing is similar to that of septal perforation. Echocardiography shows a flail mitral leaflet with severe mitral regurgitation into a normal or slightly enlarged left atrium and inferior dyskinesia. It should be remembered that mitral regurgitation is also common secondary to left ventricular dilatation.

•

ISCHAEMIC HEART DISEASE

Ventricular thrombus Ventricular thrombus can be shown by echocardiography (see Fig. 25.10), contrast-enhanced CT and MRI.TOE is probably the most reliable technique, particularly after acute myocardial infarction.Thrombus may occur with cavity dilatation in congestive cardiomyopathy, ventricular aneurysm, or after acute myocardial infarction. Fresh thrombus is mobile and strands may be observed floating within the cavity. Later the thrombus may become flattened against the ventricular wall and become more echogenic as it becomes organized. Thrombus is most commonly apical, and is less common on the posterior wall or the proximal septum. The texture of the echoes in the thrombus usually differs from that of adjacent myocardium. Using a contrast agent may emphasize the extent of the thrombus (see Fig. 25.10). Motion of the underlying myocardium is either reduced or absent; normal motion makes thrombus unlikely.

PERCUTANEOUS CORONARY INTERVENTION Percutaneous intervention for CAD has developed enormously since the first PTCA was performed by Andreas Gruentzig in Zurich in 1977. In addition to balloon dilatation, other recanalization techniques have been developed, including atherectomy, US ablation and especially stents. Doppler flow probes and IVUS devices augment the information provided by coronary arteriography. These have shown that arteriography often fails to show the full extent of CAD. These devices are widely used during PCI with improvements in immediate success, better long-term outcomes and reduced complications. This is especially so for drug eluting stents (DES) which have had a huge impact on restenosis rates and other outcomes following PCI. A better understanding of how to use the numerous available devices and the development of better anticoagulation regimens, including liberal use of pharmacological agents such as IIb/IIIa inhibitors, has reduced complications and broadened the indications for intervention. IVUS is widely used to assess the results of PCI and to determine whether or not to use adjunctive techniques. It is also used for assessing the adequacy of stent deployment, which is often underestimated by arteriography. Unlike arteriography, which shows a shadow of the arterial lumen, IVUS shows a

Figure 25.28 Acute myocardial infarction complicated by septal dissection. (A) Axial ECG-gated spinecho MRI shows irregular thinning (arrow) of the interventricular septum. This is an example of septal dissection complicating myocardial infarction, a potential precursor of an infarct related ventricular septal defect. (B) Oblique four-chamber view of the left ventricle at end-systole (note closed mitral valve leaflets, arrowheads) using cine-MRA (FISP) shows flow signal extending into the septal dissection. LA = left atrium, LV = left ventricle, RA = right atrium, RV = right ventricle.

523

524

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

tomographic, cross-sectional view of the vessel. This orientation enables direct measurements of lumen dimensions, which are considered to be more accurate than angiographic dimensions. Pullback images can be reconstructed into longitudinal images to assess complex lesions. IVUS allows characterization of atheroma size, calcification (Fig. 25.29), plaque ulceration (Fig. 25.30) and lesion composition, as well as procedural dissections, and residual irregularities or stenosis. Adequacy of arterial flow can be assessed with intravascular Doppler.

As elsewhere in the body, the ideal lesion for PCI is a short, discrete, noncalcified, relatively concentric stenosis that does not involve the vessel origin or a branch (Fig. 25.31). Most procedures are performed on lesions that are not ideal, such as ostial or bifurcation stenoses, calcified, eccentric, long lesions (> 2 cm) and lesions on bends. Factors that have led to the improved success rate include the development of more flexible balloons, increasing expertise, assessment by IVUS or Doppler and, particularly, stents. Stents have several benefits compared with conventional PTCA.

Figure 25.29 Calcified plaque. IVUS image showing calcified plaque with transducer centred in the main lumen of the coronary artery and high level echoes with reverberation (asterisk) from eccentric calcification (arrows). (Courtesy of Gregory Giugliano, MD.)

Figure 25.30 Plaque rupture. IVUS image showing ruptured plaque with the transducer centred in the main lumen of the coronary artery and low echogenicity in the area of ulceration (asterisk). Note the thickness of the wall of the artery, something not demonstrated by conventional angiography. (Courtesy of Gregory Giugliano, MD.)

Figure 25.31 Stenting of tandem stenoses. (A) Selective coronary arteriogram (left anterior oblique view) shows severe stenosis in the middle part of the right coronary artery (RCA, arrow, mid lesion) beyond the first acute marginal artery and then a more severe distal stenosis at the origin of the posterior descending coronary artery (PDA, arrow, RCA distal lesion). (B) Selective coronary arteriogram (in the same orientation following stenting of the two stenoses) shows widely patent RCA and PDA. (Courtesy of Gregory Giugliano, MD.)

CHAPTER 25

The use of balloon expandable stents reduces restenosis rates by about 50% in ideal, short lesions in vessels wider than 3 mm. The use of DES has reduced restenosis rates still further (survival without vessel failure at 3 years with DES is 87.9% compared with 67.3% with bare metal stents38). The risk of thrombosis has been reduced without an increased risk of haemorrhage by using antiplatelet drugs during PCI, especially intravenous glycoprotein IIb/IIIa inhibitors. Most PCI for acute and chronic disease as well as treatment for stenosis developing in bypass grafts (Fig. 25.32) is now performed with stents. The routine use of arterial closure devices may reduce the risk of haemorrhage from puncture sites when anticoagulation is required.

NONPATHOLOGICAL NARROWING Narrowing of no clinical significance is not uncommon on coronary arteriography. Catheter-induced spasm causes a smooth, short, tapered narrowing at the catheter tip, which may deform the course of the artery. This is most common in the RCA. Muscle bridging causes narrowing of a segment of vessel, usually part of the LAD, which narrows in systole and relaxes in diastole. This appearance is due to the (normally superficial) coronary artery passing into the myocardium, which may be shown by CTA, and being bridged and constricted by myocardial action. It usually has no significance unless the appearances are very gross, or it occurs in hypertropic cardiomyopathy, when it may be associated with sudden death.

Figure 25.32 Stenting of stenosis developing in a bypass graft. (A) Right anterior oblique selective arteriogram shows thrombus (asterisk) almost occluding a saphenous vein bypass graft. (B) Image in the same orientation shows balloon inflation during deployment of a stent (length of stent between tips of long thin black arrows). (C) Following stent deployment there is no more residual narrowing. (Courtesy of Gregory Giugliano, MD.)

•

ISCHAEMIC HEART DISEASE

CORONARY ARTERY FISTULAS Coronary artery fistulas into any cardiac chamber or vein may occur and are usually congenital (see Ch. 23) but may follow trauma. The fistula opens into the right heart in 90% of cases, most often into the right ventricle. The RCA is much more commonly involved than the left, although both may be involved. Coronary artery fistulas are associated with other cardiac malformations (e.g. pulmonary atresia, patent ductus arteriosus) in about 20% of cases. The fistula may be simple with a single vessel opening into a chamber (Fig. 25.33), or complex with multiple tortuous vessels. The clinical presentation depends on the site of the shunt and the chamber that it enters. Large shunts may produce congestive cardiac failure. In older patients, atrial fibrillation and even bacterial endocarditis may occur. Ischaemic changes may occur with fistulas if these divert coronary blood from the peripheral parts of the coronary arterial tree. The chest radiograph may show pulmonary plethora. In Bland–Garland–White syndrome (left coronary artery arising from the pulmonary artery with collateral retrograde flow from the RCA) the chest radiograph may suggest a dilated cardiomyopathy (Fig. 25.34). The diagnosis is usually obvious on aortography and, in older patients, selective coronary arteriography. In children, fistulas are often visible on

Figure 25.33 Simple coronary artery fistula. (A) Right coronary arteriogram (left anterior oblique [LAO]) shows a large fistula between the right coronary artery and the right atrium. There is aneurysmal dilatation (asterisk) of the distal coronary artery immediately before the fistula. (B) Right coronary arteriogram (LAO) after balloon occlusion shows no further filling of the distal coronary artery. Contrast medium in the detached balloon is marked by the white arrow.

525

526

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 25.34 Bland–Garland–White syndrome (left coronary artery from the pulmonary artery). (A) Lateral aortogram at the beginning of the contrast medium injection. (B) Lateral aortogram towards the end of filming. Only the large right coronary artery fills directly from the aorta in (A). In (B) a small left coronary artery (white arrow) has filled by collateral flow from the right coronary artery which in turn fills the pulmonary artery (open arrow).

echocardiography. The coronary artery supplying the fistula is almost always enlarged and tortuous and may be aneurysmal. The site and number of entries of a fistula into the receiving chamber can usually be identified by selective coronary arteriography. Beyond the fistula, the coronary artery may be very small. Percutaneous treatment by embolization is usually preferred for fistulas causing a significant shunt.

REFERENCES 1. British Heart Foundation 2000 European cardiovascular disease statistics, 2000 edn. www.heartstats.org/datapage.asp?id=744 2. British Heart Foundation 2004 Coronary heart disease statistics, 2004 edn. www.heartstats.org/datapage.asp?id=836 3. American Heart Association 2005 Heart disease and stroke statistics— 2005 update. American Heart Association, Dallas 4. Nasir K, Michos E D, Rumberger J A et al 2004 Coronary artery calcification and family history of premature coronary heart disease. Sibling history is more strongly associated than parental history. Circulation 110: 2150–2156 5. Agatston A, Janowitz W, Hildner F et al 1990 Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 15: 827–832 6. Janowitz W R, Agatston A S, Viamonte M 1991 Comparison of serial quantitative evaluation of calcified coronary artery plaque by ultrafast computed tomography in persons with and without obstructive coronary artery disease. Am J Cardiol 68: 1–6 7. Horiguchi J, Yamamoto H, Akiyama Y et al 2005 Variability of repeated coronary artery calcium measurements by 16-MDCT with retrospective reconstruction. Am J Roentgenol 184: 1917–1923 8. Heuschmid M, Kuettner A, Schroedr S et al 2005 ECG-gated 16-MDCT of the coronary arteries: Assessment of image quality and accuracy in detecting stenoses. Am J Roentgenol 184: 1413–1419 9. Garcia M J 2005 Noninvasive coronary angiography. Hype or new paradigm? JAMA 293: 2531–2533 10. Nakanishi T, Kayashima Y, Inoue R 2005 Pitfalls in 16-detector row CT of the coronary arteries. RadioGraphics 25: 425–440

11. Lawler L P, Pannu H K, Fishman E K 2005 MDCT evaluation of the coronary arteries, 2004: How we do it—data acquisition, postprocessing, display, and interpretation. Am J Roentgenol 184: 1402–1412 12. Dianias P G, Roussakis A, Ioannidis J P A 2004 Diagnostic performance of coronary magnetic resonance angiography as compared against conventional X-ray angiography. A meta-analysis. J Am Coll Cardiol 44: 1867–1876 13. Katz W, Gulati, A Q, Mahler C M et al 1997 Quantitative evaluation of the segmental left ventricular response to dobutamine stress by tissue Doppler echocardiography. Am J Cardiol 79: 1036–1042 14. Dendale P A, Franken P R, Waldman G J et al 1995 Low-dosage dobutamine magnetic resonance imaging as an alternative to echocardiography in the detection of viable myocardium after acute infarction. Am Heart J 130: 134–140 15. Armstrong W F, Zoghbi W A 2005 Stress echocardiography: Current methodology and clinical applications. J Am Coll Cardiol 45: 1739–1747 16. Plein S, Radjenovic A, Ridgway J P et al 2005 Coronary artery disease: Myocardial perfusion MR imaging with sensitivity encoding versus conventional angiography. Radiology 235: 423–430 17. Nagel E, Klein C, Paetsch I et al 2003 Magnetic resonance perfusion measurements for the noninvasive detection of coronary artery disease. Circulation 108: 432–437 18. Castillo E, Lima J A C, Bluemke D A 2003 Regional myocardial function: Advances in MR imaging and analysis. RadioGraphics 23: S127–S140 19. Koyama Y, Matsuoka H, Mochizuki T et al 2005 Assessment of reperfused acute myocardial infarction with two-phase contrast enhanced helical CT: Prediction of left ventricular function and wall thickness. Radiology 235: 804–811 20. Hoffman M H K, Shi H, Schmitz B L et al 2005 Noninvasive coronary angiography with multislice computed tomography. JAMA 293: 2471–2478 21. Lau G T, Ridley L J, Schieb M C et al 2005 Coronary artery stenoses: Detection with calcium scoring, CT angiography, and both methods combined. Radiology 235: 415–422 22. Hartnell G G 1999 Breathhold cardiac MRI and MRA. Int J Card Imaging 15: 139–150 23. Salm L P, Bax J J, Vliegen H W et al 2004 Functional significance of stenoses in coronary artery bypass grafts. Evaluation by single-photon emission computed tomography perfusion imaging, cardiovascular magnetic resonance and angiography. J Am Coll Cardiol 44: 187–188 24. Hundley W G, Clarke G D, Landau C et al 1995 Noninvasive determination of infarct artery patency by cine magnetic resonance angiography. Circulation 91: 1347–1353 25. Kefer J, Coche E, Legros G et al 2005 Head-to-head comparison of three-dimensional navigator-gated magnetic resonance and 16-slice computed tomography to detect coronary artery stenosis in patients. J Am Coll Cardiol 46: 92–100 26. Selvanayagam J B, Kardos A, Francis J M et al 2004 Value of delayed– enhancement cardiovascular magnetic resonance imaging in predicting myocardial viability after surgical revascularization. Circulation 110: 1535–1541 27. Arai A, Kwong R Y 2004 Detecting acute coronary syndrome in the emergency department using cardiac magnetic resonance imaging. Cardiovasc Rev Rep 25: 149–154 28. Hunold P, Schlosser T, Vogt F M et al 2005 Myocardial late enhancement in contrast-enhanced cardiac MRI. Distinction between infraction scar and non-infarct related disease. Am J Roentgenol 184: 1420–1426 29. Klein C, Nekolla S G, Bengel F M et al 2002 Assessment of myocardial viability with contrast-enhanced magnetic resonance imaging. Comparison with positron emission tomography. Circulation 105: 162–167 30. Aspelin P, Lundkvist J, Aubry P et al 2003 Nephrotoxic effects in highrisk patients undergoing angiography. N Engl J Med 348: 491–499 31. Davis K, Kennedy J W, Kemp H G Jr et al 1979 Complications of coronary arteriography from the collaborative study of coronary artery surgery (CASS). Circulation 59: 1105–1111

CHAPTER 25

32. ACC/AHA 1999 Guidelines for coronary angiography: Executive summary and recommendations 1999. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Coronary Angiography) Developed in collaboration with the Society for Cardiac Angiography and Interventions. Circulation 99: 2345–2357 33. Datta J, White C S, Gilkeson R C et al 2005 Anomalous coronary arteries in adults: Depiction at multi-detector row CT angiography. Radiology 235: 812–818 34. Bax J J, Poldermans D, Elhendy A et al 1999 Improvement in left ventricular ejection fraction, heart failure symptoms and prognosis after revascularization in patients with chronic coronary artery disease and viable myocardium detected by dobutamine stress echocardiography. J Am Coll Cardiol 34: 163–169

•

ISCHAEMIC HEART DISEASE

35. Wahl A, Paetsch I, Gollesch A et al 2004 Safety and feasibility of highdose dobutamine–atropine stress cardiovascular magnetic resonance for diagnosis of myocardial ischemia: Experience in 1000 consecutive cases. Eur Heart J 25: 1230–1236 36. Hansson G K 2005 Inflammation, atherosclerosis, and coronary artery disease. N Engl J Med 352: 1685–1695 37. Davies M J, Thomas A C 1985 Plaque fissuring—the cause of acute myocardial infarction, sudden ischaemic death, and crescendo angina (review). Br Heart J 53: 363–373 38. Fajadet J, Morice M-C, Bode C et al 2005 Maintenance of long-term clinical benefit with sirolimus-eluting coronary stents: Three-year results of the RAVEL trial. Circulation 111: 1040–1044 39. Hall E 1994 Radiobiology for the radiologist, 4th edn. J B Lippincott, Philadelphia

527

CHAPTER

Pulmonary Circulation and Pulmonary Thromboembolism

26

Paras Dalal and Michael B. Rubens

Pulmonary circulation • Pulmonary anatomy • Pulmonary physiology • Pulmonary vascular patterns Pulmonary thromboembolic disease • Investigation

The pulmonary circulation is unique in many ways, e.g. its response to hypoxia is arterial constriction. In addition it

followsva dual flow model (both pulmonary and bronchial circulations supply the lung). Appreciation of pulmonary anatomy and normal physiology enables a better understanding of abnormal conditions (and their relevant radiographic features). A potentially devastating consequence of interruption to lung blood supply is that of pulmonary infarction. The most common cause of this is pulmonary thromboembolism. The second half of this chapter concerns the demography, pathophysiology and imaging features of pulmonary thromboembolism.

PULMONARY CIRCULATION PULMONARY ANATOMY In adulthood the main pulmonary artery measures approximately 5 cm in length and is entirely enveloped within the pericardium. From its origin at the right ventricle it passes superiorly and posteriorly, initially anterior and then to the left of the ascending aorta. At about the level of the fifth thoracic vertebrae, it divides to form the left and right pulmonary arteries. The central pulmonary arteries have a lamella of elastic tissue in their media.

Left pulmonary artery This is the shorter of the two branches. Its course runs cephalad and posteriorly from the main pulmonary artery arching superiorly over the left main bronchus to enter the left hilum. Within the hilum it bifurcates into ascending and descending trunks. The ascending trunk divides almost immediately into the apico-posterior and anterior segmental branches which supply the left upper lobe. The descending trunk gives a branch to the lingula which itself divides into two segmental arteries (superior and inferior lingular segmental arteries). The next branch from the descending trunk is the superior segmental artery, which sup-

plies the apical segment of the left lower lobe. Subsequent branches supply the anterior basal, lateral basal, posterior basal and medial basal segments of the left lower lobe.

Right pulmonary artery The right pulmonary artery arises at a right angle from the main pulmonary trunk, traverses the mediastinum posterior to the superior vena cava and ascending aorta and anterior to the right main bronchus, and just before entering the hilum it divides into the ascending and descending trunks. The ascending branch divides into apical, anterior and posterior segmental branches. The posterior segmental branch may however also originate at the bifurcation of the right main pulmonary artery or the right descending trunk. The descending trunk passes vertically inferiorly from the right hilum. It gives rise to the middle lobe artery (which divides into the lateral and medial segmental branches) and the artery to the apical segmental lobe of the right lower lobe. The distal branches of the descending trunk arise in the order: medial, anterior, posterior and lateral basal segmental arteries. In addition a suprabasal segmental branch artery may arise between the superior and basal segmental arteries and supply an additional segmental artery of the right lower lobe.

530

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pulmonary arterioles These are continuations of the segmental and subsegmental arteries. In adults they have discontinuous muscular walls with thin media and wide lumens.

Bronchial vessels Bronchial arterial anatomy is subject to wide variability. The right bronchial artery usually shares its origin from the dorsolateral aorta with the first intercostal artery. From this common trunk the bronchial artery then courses in a long sharp ‘hairpin’ loop. The left bronchial artery may have up to three separate origins from the descending aorta (just below the origin of the left subclavian artery). Compared to the pulmonary arteries, the bronchial arteries are generally of small calibre and enter the hilum closely applied to the bronchial wall. They are distributed mainly in the central third of each lung, where they supply oxygenated blood to the lung interstitium. Bronchial veins drain into the left atrium.

PULMONARY PHYSIOLOGY Normal pulmonary artery pressures are shown in Table 26.1. In the normal patient during ventricular systole, pulmonary arterial pressure rises equal to that in the right ventricle. At the end of systole as blood flows into the lung capillaries, pulmonary artery pressure falls slowly. Mean capillary pressure is normally significantly lower than that in the pulmonary artery (indirectly calculated mean capillary pressure is 7 mmHg). This gradient is important for the maintenance of forward flow of blood and has implications with respect to fluid exchange. The hydrostatic pressure within the pulmonary capillaries tends to force fluid into the interstitium of the lung. This is partly counteracted by the plasma oncotic pressure which attracts fluid back into the capillaries. Imbalance of these pressure ratios can lead to abnormal fluid shift and thus overflow of fluid into the lung parenchyma. The pulmonary interstitial space is usually kept dry by pulmonary lymphatic channels. These drain the ‘transudated’ fluid which enters the interstitium from the alveoli. However, if the rate of accumulation of fluid exceeds the ability of the lymphatic channels to clear it, accumulation of fluid within the interstitium begins. If this is continuous, it can lead to alveolar fluid accumulation at which point gas exchange may become compromised (gas exchange is not usually compromised by the presence of interstitial oedema alone). An important difference between the pulmonary and systemic vasculature is the response to hypoxia. In the pulmo-

Table 26.1

NORMAL PULMONARY ARTERIAL PRESSURES

nary system hypoxia results in local vasoconstriction, causing diversion of blood to regions of better ventilation. Although contrary to the effect in the rest of the body, this mechanism serves to protect the alveolar–arteriolar PO2 balance and thus minimizes ventilation–perfusion differences in cases of diffuse and regional disease, i.e. it supplies blood to regions of the lung that will most efficiently oxygenate it. This homeostatic mechanism is responsible for ‘matched defects’ seen in cases of pneumonic consolidation on ventilation–perfusion imaging (see below).

PULMONARY VASCULAR PATTERNS Pulmonary venous hypertension Pulmonary venous hypertension (PVH) is caused by increased resistance in the pulmonary veins. It may occur secondary to left-sided heart disease, or less commonly, mediastinal disease (Table 26.2).The most common causes are left ventricular failure, mitral valve disease, and aortic valve disease. The severity of mitral valve stenosis can be noninvasively gauged by assessing the degree of PVH1. In cases of aortic valve disease, however, the degree of PVH is more indicative of myocardial failure than severity of stenosis. The radiological findings can be thought of essentially as a progressive series of changes that occur in response to the underlying changes in physiology. As pulmonary venous pressure rises, the upper lobe veins distend.They initially reach the size of, and eventually become larger than, the lower lobe vessels (thus reversing the normal ‘gravity-dependent’ pattern). This is described as ‘upper lobe venous diversion’ and is often the first recognized radiological sign of pulmonary venous hypertension (Fig. 26.1). If the pulmonary venous pressure continues to rise and exceeds the plasma oncotic pressure, fluid will begin to accumulate in the lung interstitium. This is known as interstitial pulmonary oedema. Radiologically this is associated with the appearance of interstitial (Kerley B) lines (Fig. 26.2). These lines were first described in 19332 and represent thickening of interlobular septa (as a result of fluid accumulation) within the lung. They were originally classified into three groups: 1 Kerley A lines are approximately 4 cm in length and are most conspicuous in the upper and mid portions of the lung. They are deep septal lines (lymphatic channels) that radiate from the hila into the central portions of the lungs but do not reach the pleura. Their presence normally indicates a more acute or severe degree of oedema.

Table 26.2 CAUSES OF PULMONARY VENOUS HYPERTENSION AND PULMONARY OEDEMA Left ventricular outflow obstruction, e.g. aortic coarctation, aortic stenosis, hypoplastic left heart

Arterial pressure phase

Mean pressure (mmHg)

Left ventricular failure

Systolic

25

Mitral valve disease

Diastolic

8

Left atrial myxoma

Mean Mean wedge

15 5

Fibrosing mediastinitis Pulmonary veno-occlusive disease

CHAPTER 26

Figure 26.1 Upper lobe venous distension. The upper zone vessels are distended. Note also the double density over the left heart border due to there is also left atrial enlargement.

2 Kerley B lines are shorter (1 cm or less) interlobular septal lines, found predominantly in the lower zones peripherally, and parallel to each other but at right angles to the pleural surface. 3 The originally described Kerley C lines are now designated as due to overlapping Kerley B lines. The term is no longer used3. Septal lines can be differentiated from blood vessels as the latter are not visible in the outer 1 cm of the lung. In addition, deep septal lines do not branch uniformly (as is the case for blood vessels) and are seen with a greater clarity (as they represent a sheet of tissue) than a blood vessel of similar calibre. Under normal circumstances septal lines caused by interstitial fluid overload would be expected to disappear after suitable reduction in pulmonary venous pressure. Exceptionally, however, they may persist, e.g. in long-standing PVH, where haemosiderin deposition or fibrosis has occurred. Other causes of persistent septal lines include idiopathic interstitial fibrosis, lymphangitis carcinomatosa and pneumoconiosis. Differentiation between the causes on plain radiography may be helped by ancillary signs (e.g. cardiomegaly and calcification of the mitral valve which both favour PVH as the diagnosis). Other signs of interstitial fluid overload include perihilar haze (loss of visible clarity of the lower lobe and hilar vessels) and peribronchial cuffing (apparent thickening of proximal bronchial walls as a result of interstitial fluid accumulating around their walls). As the pulmonary venous pressure continues to increase fluid begins to accumulate in the alveolar spaces. This is termed alveolar oedema (Fig. 26.3). Kerley B lines, airspace nodules, bilateral symmetric consolidation in the mid and lower lung zones and pleural effusions may be seen.

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

Figure 26.2 Thickened interlobular septae (Kerley B lines) in a patient with mitral stenosis. Short, horizontal linear densities are present in the periphery of the right lower lobe. This is also a small lamellar pleural effusion.

Certain patterns of opacification may suggest particular diagnoses. The often cited ‘perihilar bat’s wing’ pattern of airspace consolidation is seen most commonly in left ventricular and renal failure, whereas alveolar oedema localized to the right upper zone is suggestive of severe mitral regurgitation.The latter

Figure 26.3 Alveolar pulmonary oedema. Diffuse shadowing is present throughout the lungs; it ranges from ground glass to consolidation. Careful scrutiny of the radiograph also reveals upper lobe venous distension.

531

532

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

is thought to be a result of predominant regurgitant blood flow in the right upper lobe pulmonary vein, from the superiorly and posteriorly positioned mitral valve. Computed tomography (CT) findings in pulmonary oedema include thickening of septal and bronchovascular structures. In addition, perihilar ground-glass opacity may be found in cases of mild parenchymal oedema. Alveolar oedema may initially be recognized as peribronchovascular airspace nodules progressing to dense airspace consolidation. In chronic pulmonary venous hypertension signs of pulmonary arterial hypertension may also develop. In addition, a fine nodular pattern may appear throughout both lungs.These nodules represent haemosiderin deposition. This pattern was previously most commonly seen in patients with long-standing severe mitral stenosis. In very severe chronic PVH pulmonary ossicles (up to 1 cm in size) may develop (Fig. 26.4). Attempts have been made to correlate the stage of radiographic pulmonary venous hypertension with pulmonary venous pressure (as measured by the pulmonary capillary wedge pressure [PCWP])4–6. When the PCWP is normal (8– 12 mmHg) the chest radiograph is not expected to demonstrate any specific abnormality related to pulmonary venous pressure. Mild PCWP elevation (12–18 mmHg) is associated with upper lobe venous distension. Further PCWP increase (19–25 mmHg) leads to interstitial oedema (peribronchial cuffing, Kerley lines). Above this value (25 mmHg) airspace opacities are seen. Although not accurate, these correlates serve to give an approximation of intravascular pressure. Although most cases of PVH are associated with valvular and/or myocardial dysfunction leading to cardiomegaly, this is not always the case. An important example of this is in the early post-myocardial infarction phase. Here, up to 50% of patients have been shown to exhibit radiographic signs of PVH in the first 24–48 h post infarction7,8. This is due to an acute decrease in myocardial compliance which essentially resolves in the first week post infarction. The other situation where signs of

pulmonary oedema may be seen with a normal heart size is noncardiogenic pulmonary oedema (Table 26.3).

Pulmonary arterial hypertension Pulmonary arterial hypertension (PAH) is defined as an elevation in mean pulmonary arterial pressure above 30 mmHg during exercise (and 25 mmHg at rest). Arterial pressure may be considered as a function of blood flow and vascular resistance. Vascular resistance depends upon the cross-sectional area of the vascular bed. The pulmonary vessels are more compliant than their systemic counterparts owing largely to their thin walls and also (and perhaps more importantly) their larger diameter. Furthermore, the pulmonary bed can also respond to increasing flow by opening up additional vascular channels. These features allow considerable latitude in the system; nonetheless, hypertension will occur when these mechanisms become insufficient to cope with increasing flow. Some causes of PAH are listed in Table 26.4. Radiographically, cardiac enlargement (right atrial and ventricular enlargement), enlargement of the central pulmonary arteries (main pulmonary artery and its branches down to the segmental level) (Fig. 26.5) and tapering of peripheral arterial branches (vessels beyond segmental level)—termed ‘peripheral pruning’—are seen. In long-standing cases the central pulmonary arteries may develop calcification due to atheroma (a feature not seen in non-hypertensive pulmonary arteries). Central arterial enlargement may mimic enlarged hilar lymph nodes. However, careful scrutiny usually allows differentiation as lymphadenopathy characteristically has a lobulated border whereas arterial enlargement has a smoother outline. A recognized method of assessing pulmonary arterial size is by measuring the size of the right descending pulmonary artery9. Enlargement may be diagnosed if the transverse diameter of the artery at its midpoint is greater than 17 mm. Overall, although relatively specific, the sensitivity of chest radiography for the diagnosis of PAH is low.

Table 26.3 OEDEMA

CAUSES OF NONCARDIOGENIC PULMONARY

Aspiration Adult respiratory distress syndrome (ARDS) Near drowning Hepatorenal failure Rapid lung re-expansion post thoracocentesis Drugs, e.g. aspirin, nitrofurantoin, morphine

Table 26.4

CAUSES OF PULMONARY ARTERY HYPERTENSION

Chronic lung disease, e.g. chronic obstructive pulmonary disease, interstitial pneumonia Pulmonary embolic disease Pulmonary venous hypertension Intracardiac shunts (left-to-right or bidirectional)

Figure 26.4 Mixed mitral valve disease. Multiple small calcified nodules are seen due to pulmonary ossification (a sign related to chronically raised venous pressure).

Pulmonary arteritides Idiopathic

CHAPTER 26

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

has been reported that if the diameter of the main pulmonary artery is greater than that of the adjacent ascending aorta, then PAH is present12. This finding has a positive predictive value of 0.93 but a negative predictive value of 0.44. Thus, a nondilated artery does not exclude the diagnosis. In cases of severe PAH pericardial thickening and effusion may also be found on CT13. With advances in magnetic resonance imaging (MRI) techniques, studies observing changes in physiological variables have also been reported, e.g. it has been demonstrated that vessel distensibility is reduced in patients with PAH14 (Fig. 26.6).

Pulmonary overcirculation (plethora)

Figure 26.5 Pulmonary arterial hypertension. Chest radiograph demonstrates gross dilatation of the main, left and right pulmonary arteries in a patient with Eisenmenger atrial septal defect.

Studies using CT to measure the main pulmonary artery10 have found good correlation between arterial cross-sectional area and mean pulmonary artery pressure (although not all dilated arteries relate to increased pressure11). In addition, it

Pulmonary overcirculation is caused by an increase in blood flow through the lungs, most commonly secondary to a left-to-right cardiac shunt. It may also be seen in individuals with bidirectional shunts or increased cardiac output states (e.g. in well-trained athletes and pregnant women). Radiologically, the central pulmonary arteries enlarge. In addition, peripheral pulmonary vessels become visible in the outer third of the lung—a sign known as ‘pulmonary plethora’ (unlike in PAH where there is enlargement of central arteries with peripheral arterial pruning) (Fig. 26.7). A left-toright shunt of at least 2:1 is required before it becomes apparent on plain radiography. When radiological pulmonary plethora is combined with clinical examination, the underlying cause can often be identified; in the absence of cyanosis, pulmonary

Figure 26.6 Pulmonary arterial hypertension. (A) Chest radiograph demonstrating cardiomegaly and enlargement of central pulmonary arteries. (B) CT section demonstrating an enlarged main pulmonary artery when compared to the ascending aorta. (C) MR angiogram showing the enlarged main pulmonary artery. This series of images shows the relative ease with which even minor degrees of pulmonary artery dilation (inferring raised arterial pressures in the pulmonary system) can be detected on cross-sectional imaging. This may not have been adequately recognized on chest radiography alone. (Image C courtesy of Dr R Mohiaddin, Royal Brompton Hospital, London.)

533

534

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 26.7 Pulmonary plethora. There is a marked increase in the size and number of visible vessels in both lungs (seen in a patient with an atrioventricular septal defect). The prominent right superior mediastinum is due to normal thymus.

Figure 26.8 Normal chest radiograph.

plethora is most commonly due to a left-to-right shunt, whereas in the presence of cyanosis a bidirectional shunt is inferred.

pulmonary embolism. Cardiovascular causes of this finding are rare but include previous shunt operations (e.g. Blalock–Taussig shunt) for congenital heart disease, pulmonary artery stenosis and pulmonary arteriovenous fistulas.

Systemic supply to the lungs

Pulmonary arteriovenous malformations

Systemic supply to the lungs develops in situations where there is severe right ventricular outlet obstruction (usually secondary to a severe Tetralogy of Fallot or pulmonary atresia). In these cases no main pulmonary artery is visible and the peripheral vascular pattern is often disorganized (and may resemble interstitial lung disease).

These occur in four main forms: 1 a single fistula 2 multiple discrete fistulas with one/a few predominant lesions 3 multiple discrete fistulas of similar size 4 diffuse telangectasia.

Pulmonary oligaemia This sign implies decreased blood flow in the lungs and is usually due to right ventricular outflow tract obstruction, often in association with a right-to-left shunt. Radiographically there appear to be fewer and smaller vascular markings in the lungs (compare Fig. 26.8 with Fig. 26.9). In addition, the pulmonary trunk appears small (or may not be visible at all).When seen in an otherwise healthy neonate it may be a physiological finding as the pulmonary arterioles may not have fully relaxed from their intrauterine high resistance state.

Uneven vascularity Uneven pulmonary vascularity may be ‘apparent’ or ‘real’. Apparent causes of this finding include patient rotation (which may be due to poor technique or patient scoliosis), unilateral mastectomy and atrophy or congenital absence of pectoral muscles. In cases of apparent uneven pulmonary vascularity close scrutiny of the pulmonary vessels reveals no difference in number or calibre between the two lungs. Real causes of uneven or unilateral reduced pulmonary vascular attenuation include: bullae, emphysema, Macleod’s syndrome (Fig. 26.10), previous pulmonary resection and

Figure 26.9 Pulmonary oligaemia. The peripheral vascular pattern is diminished in a patient with tetralogy of Fallot and a right aortic arch.

CHAPTER 26

Figure 26.10 Uneven vascularity. In a patient with Macleod’s syndrome a hypoplastic right main pulmonary artery leading to a reduction in vascular markings and a small right hemithorax are seen.

Fistulas may be congenital or acquired. When acquired they may be seen in conjunction with liver cirrhosis, schistosomiasis and metastatic thyroid carcinoma. In congenital cases, 50% have hereditary haemorrhagic telangectasia15–17. Clinically they may produce systemic arterial desaturation

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

and give rise to signs of dyspnoea, hypoxia, cyanosis and heart failure. Although 10% of cases present in the first decade16, most do not manifest clinically until the third or fourth decade. Multiple lesions are seen in up to 50% of cases15,16. Radiographically they may appear as round, oval, or lobulated opacities with an associated prominent vascular shadow (Fig. 26.11), but if small and discrete they may not be detected on plain chest radiography15. They occur most frequently in the lower lobes (Fig. 26.11). The feeding and draining vessels may only be seen (or certainly more clearly seen) on CT (Fig. 26.12). Furthermore, their vascular nature may be suggested by change in size with respiratory manœuvres (reduction in size during valsalva). Diagnosis may be confirmed with bilateral digital subtraction pulmonary angiography, which will demonstrate the feeding artery and vein draining into the left atrium. This not only aids pre-therapeutic assessment of the known lesion but will aid detection of other (perhaps smaller fistulas) not seen on chest radiography. A study comparing CT (with three-dimensional [3D] reconstructions) and pulmonary angiography concluded that selective angiography was still the ‘gold standard’ for pre-therapeutic pulmonary arteriovenous malformation investigation18.

Congenital abnormalities of the pulmonary artery Congenital pulmonary artery absence This congenital abnormality results from a failure of development of the right or left sixth branchial arterial arch. It is a rare finding which is characterized by short segment atresia of the proximal left or right pulmonary artery19–25. More distal

Figure 26.11 Pulmonary arteriovenous malformation. (A) Plain chest radiograph shows a band-like opacity leading to a nodular opacity in the left lower lobe. (B) close-up of (A).

535

536

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Pulmonary artery stenosis This occurs in three forms: 1 central form involving the bifurcation of the main pulmonary artery 2 peripheral form involving the origin of the lobar segmental or subsegmental pulmonary arteries 3 diffuse form where there is general hypoplasia of the entire pulmonary arterial system27. Associations of this condition include rubella syndrome, William’s syndrome and Ehlers–Danlos syndrome28. Approximately two-thirds of patients with pulmonary artery stenosis have additional cardiac lesions26. Radiographically there may be no discernible abnormality. However, sausage-shaped arteries produced by proximal stenosis and post-stenotic dilation29 may be seen. Optimally the stenosis is demonstrated using pulmonary arteriography. Figure 26.12 Pulmonary arteriovenous malformation. CT image showing arteriovenous malformation (arrow) and its associated feeding and draining vessels leading to the right hilum.

segments are usually present. It is associated with various congenital cardiac defects (e.g. tetralogy of Fallot) and normally occurs on the side opposite the aortic arch.The principal radiological features are a small volume ipsilateral lung without air trapping (compare with Macleod’s syndrome [see Fig. 26.10]) and a small ipsilateral hilum. Occasional opacities may be found in the affected lung owing to systemic–pulmonary collaterals. Confirmation of the absence of the pulmonary artery can be achieved using radionuclide imaging or angiography (conventional, CT, or MRI).

Congenital absence of the pulmonary valve This is associated with aneurysmal dilatation of the main pulmonary artery and hilar vessels (especially the left pulmonary artery26) and is almost always associated with cyanotic heart disease (typically tetralogy of Fallot). Pulmonary valve stenosis characteristically produces enlargement of the left hilar, lobar and segmental pulmonary arteries. Idiopathic pulmonary artery dilatation may give an identical appearance but it is not associated with a pressure gradient across the pulmonary valve.

Pulmonary artery aneurysms Aneurysms rarely occur in the pulmonary circulation30,31 but when they do, they usually involve main, lobar, or segmental arteries. They may be congenital or acquired (e.g. secondary to a patent ductus arteriosus). Most commonly, however, they are mycotic aneurysms32 which may be acquired secondary to septic embolization or direct extension of a parenchymal infection (Table 26.5). Mycotic aneurysms need to be monitored closely as they have a propensity to enlarge rapidly and rupture. On chest radiography aneurysms appear as fusiform or saccular dilatations. Confirmation of their presence (and number) may be performed with CT, MRI, or conventional pulmonary angiography.

Table 26.5

CAUSES OF MYCOTIC ANEURYSMS

Infected ventricular shunts Valvular bacterial endocarditis of right ventricle Septic emboli Tuberculosis (Rasmussen aneurysm) which may resemble a mycetoma on chest radiograph Behçet’s disease Trauma

PULMONARY THOMBOEMBOLIC DISEASE The annual incidence of pulmonary embolus (PE) is estimated to be 65 cases per 100 000 population33,34. One-third of these episodes cause (or contribute to) the death of the patient. This figure is reduced to 8% if anticoagulant therapy is commenced35. Risk factors for developing PE include increasing age, hypercoagulable state36,37, orthopaedic surgery, malignancy, medical illness, instrumentation (e.g. indwelling intravenous catheters38) and pregnancy. Post-mortem studies suggest that up to 65% of inpatients have emboli lodged in their arteries39. Furthermore, it has been stated that PE is the cause of 7% of deaths in hospital inpatients40 but the incidence

of clinically diagnosed embolic disease in hospital inpatients is less than 1%41. This gives evidence for the widely held view that clinical diagnosis for pulmonary embolic disease is very insensitive. These data suggest a justification for empiric treatment of all patients suspected of having PE. However, there is a high (up to 30%) incidence of haemorrhagic complications in patients on anticoagulant therapy which mandates an accurate diagnosis before therapy. That PE is a complication of deep venous thrombosis is often overlooked. Thus management of these patients must address not only treatment of the existing embolus but also the issue of reducing further embolism from

CHAPTER 26

the deep veins. The effects of PE, once lodged in the pulmonary tree, are principally dependent on its size and the presence of any pre-existing pulmonary or cardiac disease. The clinical presentation of PE ranges widely and as far as cardiac arrest. Clinical evaluation most consistently, however, reveals dyspnoea, chest pain (which may be pleuritic), haemoptysis, cough, hypotension and tachycardia. A large PE may lead to pulmonary oedema, presumably as a result of precipitation of left ventricular failure42. As these are relatively non-specific findings the diagnosis may be delayed unless specifically thought of. Physiologically multiple effects are caused by vascular obstruction43. These include: • Arterial hypoxia commonly occurs in pulmonary embolic disease; the degree of hypoxia being proportional to the size of the embolus44. • Cessation of blood flow to a part of the lung leads to the development of alveolar dead space40, although for this to occur effectively greater than 80% of the blood flow must be reduced. • Depletion in surfactant may lead to pulmonary oedema (if reserves of surfactant are depleted). • Lung volume reduction. • PAH—it is estimated that greater than 50% of the vasculature needs to be obstructed for there to be a significant rise in pulmonary arterial pressure. However, in addition to ‘simple’ vascular obstruction, humoral agents and reflex mechanisms causing vasoconstriction may also play a role, thus making hypertension more likely. Emboli normally lodge in branch pulmonary arteries (only a few are situated at the bifurcation of the main pulmonary artery, so-called ‘saddle emboli’). Once an embolus is lodged in the pulmonary tree, it is normally either lysed by the patient’s fibrinolytic system or becomes organized and canalized. The degree to which each of these processes occurs depends to some extent on the patient’s fibrinolytic system, the amount of thrombus deposited on the embolus and the degree of organization of the embolic material itself. In cases of repeated thromboembolism without lysis of the embolic material, arterial hypertension is more likely to develop. This is important to identify as, in this circumstance, it may be treatable surgically45. Pulmonary infarction is a relatively rare event due in large part to the presence of a second, ‘systemic’ supply to the lungs from the bronchial arteries46,47. Infarction tends only to occur in cases where both pulmonary and bronchial supplies are inadequate (e.g. where emboli occur in patients with heart failure)46,48. Overall infarction has been found to occur in 15% of cases of PE43 and is seen most commonly in the lower lobes.

INVESTIGATION Various techniques are used in the investigation of PE. Clinical evaluation alone is not sufficiently reliable for diagnosis. Other nonimaging tests include ECG and measurement of arterial pO2 and d-dimers. Although not diagnostic for a pulmonary

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

embolus, these tests are of particular use in suggesting other causes for the patient’s symptoms, e.g. myocardial infarction. The ECG and arterial pO2 assessment have a low specificity for diagnosis of a pulmonary embolus. On ECG the classic findings of ‘S1 Q3 T3’, right bundle branch block, right axis deviation and right ventricular hypertrophy are all signs of cor pulmonale of any cause. This is thus likely only to be seen in patients with severe embolism (i.e. not in the majority of patients). A similar problem arises with the use of arterial blood gas analysis as no specific levels of hypoxia have been set below which PE can reliably be diagnosed. Thus although hypoxia may commonly be found in these patients it is almost impossible to differentiate the myriad of causes solely by its presence. d-Dimers have been used extensively in screening patients for deep venous thrombosis. d-Dimers are a breakdown product of cross-linked fibrin, which is found in increasing amounts when thromboembolism is present, and are a measure of fibrinolytic activity. The use of d-dimers in PE testing has been revealed to have a high sensitivity but low specificity49.The use of this assay in confirming PE appears limited owing to a high false-positive rate, but it is useful in excluding PE (especially in patients with a low pretest probability) as it has a very high negative predictive value50,51. Thus it may be useful to screen patients before embarking on further investigation. Indeed in some centres it has been combined with clinical probability assessment to screen patients who may need further investigation52–55. However, this practice can lead to false reassurance as false-negative tests can occur, particularly in cases of subsegmental emboli56.

Radiological investigation Increasing reliance is placed on radiological techniques in the work-up of patients with possible PE. Tests include chest radiography, radionuclide ventilation–perfusion imaging, spiral CT and magnetic resonance and pulmonary angiography. In addition to these ‘direct’ investigations, ‘indirect’ investigations include Doppler ultrasound (US), conventional CT and magnetic resonance venography of the leg veins to look for deep vein thrombosis (DVT).

Chest radiography This is often the first imaging examination requested in patients suspected of having a PE. It is however not a reliable test for this diagnosis as patients with emboli may have a normal radiograph (seen in up to 40% of patients with PE). Indeed the accuracy of diagnosing a PE on chest radiography alone was shown to be only 33% in one study57. In addition, the PIOPED (prospective investigation of pulmonary embolism diagnosis) investigators found in patients with a suspected PE that abnormalities seen on the chest radiographs of patients with proven PE were seen with similar frequency in patients with no proven PE58. Thus the overall sensitivity and specificity of chest radiography for the diagnosis is low57,59–61. In light of these findings the purpose of chest radiography is probably best thought of as one of excluding other causes of the patient’s symptoms (e.g. pneumothorax) rather than diagnosing PE.

537

538

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

The most common radiographic signs of pulmonary embolus (not causing infarction) are: • Localized peripheral oligaemia (Westermark’s sign)—secondary to the embolus lodging in a peripheral artery. It may be associated with proximal arterial dilatation. Diagnosis of this often subtle sign may be made easier by comparison with previous radiographs (if available). • Peripheral airspace opacification, which represents pulmonary haemorrhage. • Linear atelectasis—caused by ischaemic injury leading to surfactant deficiency (as a result of damaged Type II pneumocytes). • Pleural effusion—often small (in the absence of infarction). • Enlargement of the central pulmonary arteries may be found also, but normally occurs secondary to chronic repeated embolic disease. Less commonly cardiac enlargement may be seen in patients with a large PE with associated cor pulmonale. As discussed, owing to good collateral bronchial and pulmonary blood flow, infarction is relatively rare after embolism (occurring in < 10% patients). It is most common in those patients with impaired cardiac function (causing impairment to collateral flow). The changes seen in those without infarction may be seen in patients with infarction also. To distinguish infarction from noninfarction causing pulmonary embolus radiographically is tricky. Radiographic signs more associated with infarction that may be of use include: • A pleurally based, wedge-shaped opacity (Hampton’s hump), which is normally seen in the lateral or posterior costophrenic sulcus of the lung. It may rarely contain air bronchograms. The apex of the triangle points toward the occluded feeding vessel and its base rests against the pleural surface. An infarct is rarely truly ‘wedge’- or ‘triangular’shaped owing to the plane of the radiograph and surrounding haemorrhage. Although often mentioned, it is neither specific nor commonly seen in PE. • Consolidation (which may be multifocal) is seen predominantly in the lower lobes46. This may be seen from 12 h up to several days post embolism. Its exact shape may vary depending on the location and state of the underlying lung62. • Distinction between parenchymal changes related to infarction and those where no infarction has occurred may be made by observing serial radiographs. Relatively rapid resolution of parenchymal changes (up to 7–10 d) is associated with noninfarction-causing emboli, whereas those associated with infarction cause changes that take considerably longer (weeks to months) to resolve, and in addition normally heal with scarring63 or localized pleural thickening. • Cavitation (which is reported to occur in < 5% of cases and is normally related to secondary infection at the site of infarction or due to a septic embolus)64. • Haemorrhagic pleural effusion (which may be slow to clear) and may be seen in up to 50% of patients with PE. Such an effusion may be the cause of ipsilateral chest pain65.

Ventilation–perfusion imaging This is a common ‘next step’ in the diagnostic algorithm of suspected PE. It is a widely available and cost-effective test66,67 as a (technically adequate) negative examination can allow confident exclusion of the diagnosis of PE. Its utility is greatest when accompanied by a normal chest radiograph implying that a ventilation–perfusion mismatch is not due to parenchymal disease. Perfusion (Q) scintigraphy (to assess the distribution of pulmonary blood flow) is performed using injection of microparticles (10–100 µm) of 99mTc micro-aggregate albumin (MAA) with the patient lying supine (to maximize flow to the lung apices). These particles micro-embolize in the lung and, providing the study is technically adequate, the radioactivity emitted by these particles (detected using a gamma camera) can provide a map of pulmonary blood flow. The effective dose to the patient using this technique is less than 1 mSv. Ventilation (V) scintigraphy is performed by inhalation of krypton-81m, xenon-133, 99mTc-diethylenetriamine penta-acetic acid (DTPA) or ‘technegas’. 81mKr has a short half-life and is expensive. It is the optimal imaging agent for this purpose as it emits high energy photons (190 keV) and owing to its short halflife can be continuously administered to the patient, including during perfusion imaging. 133Xe is less optimal owing to its longer half-life and low photon energy (80 keV), but it is cheaper than 81m Kr. 99mTc-DTPA and technegas aerosols cannot be administered during perfusion imaging as both aerosols are labelled with 99m Tc (as used to label the MAA). Furthermore, aerosol imaging provides a static image of lung ventilation whereas imaging with 81m Kr is more dynamic. Imaging with technegas aerosol has been shown to provide images comparable to those with 81mKr (both of which provide better images than diethylene triamine pentaacetate [DTPA]68). Eight images are conventionally acquired (anterior, posterior, oblique and lateral projection on both sides). Additionally, SPECT imaging has been shown also to be of use in assessment. The ventilation and perfusion images are then compared to observe for V/Q ‘mismatches’ (i.e. defects on perfusion imaging in regions that are normal on the ventilation study), which refer to an underperfused region of lung and suggest the presence of PE. The physiological basis of this in noninfarct-causing PE is that despite interruption of perfusion the ventilation remains intact; thus the ventilation image remains normal. If such an area develops into an infarct a defect may be seen on both studies, i.e. it will become matched (as the area will no longer be ventilated). However, where infarction does occur a ventilation defect is seen but is smaller than the perfusion defect as the peri-infarct lung remains ventilated. A normal V/Q study has been shown to have a negative predictive value of 100%69, which would thus justify its widespread use. Matched ventilation and perfusion defects are commonly seen in patient with obstructive airways disease, e.g. asthma and chronic obstructive pulmonary disease. These are caused by pulmonary hypoxic vasoconstriction, which (as previously mentioned) reduces the flow of blood to poorly ventilated parts of the lung. In situations where this process is not complete ‘a reversed mismatch’ may be found. In this circumstance more prominent ventilation than perfusion defect is seen. Causes of

CHAPTER 26

this phenomenon include lobar collapse, pneumonic consolidation and a large pleural effusion in addition to obstructive airway causes (Fig. 26.13). Although a commonly utilized test, the efficacy of V/Q examinations may be limited. In the PIOPED study only 27% of all V/Q examinations showed high probability or normal results70. Furthermore, there is wide interobserver variability

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

in interpretation and few studies have compared the accuracy of the various patterns of V/Q abnormality that are used to diagnose PE. The criteria for diagnosing PE with lung scintigraphy have been standardized by the PIOPED study. Table 26.6 shows a modified summary based on the data from the PIOPED study. Fundamentally radionuclide imaging provides a probability assessment of the risk of the presence of a PE. Figure 26.13 Pulmonary thromboembolism. (A) Normal V/Q study (ventilation [v] images on top row and perfusion [p] images beneath). No defects are seen in either series. (B) Matched defects—multiple foci of non-tracer uptake seen in ventilation and perfusion series. Scrutiny of the images reveals that the defects are well matched for position on both series. (C) Perfusion defect (wedgeshaped peripherally) seen on perfusion imaging which is not replicated on ventilation imaging. This suggests a high probability for the presence of pulmonary embolus. (Courtesy of Dr S. E. M. Clarke, Guy’s Hospital, London.)

539

540

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Table 26.6 PIOPED STUDY OF THE VALUE OF VENTILATION–PERFUSION IMAGING IN THE DIAGNOSIS OF PULMONARY EMBOLUS. SUMMARIZED FINDINGS OF THE FEATURES OF DIFFERENT PULMONARY EMBOLUS PROBABILITIES70 Normal Normal perfusion

Very low probability

Low probability (≤ 19%)

Intermediate probability (20–79%)

High probability (≥ 80%)

≤ 3 small perfusion defects with a normal chest radiograph

> 3 small perfusion defects with a normal chest radiograph

Lower zone matched ventilation– perfusion defect

≥ 2 large ventilation–perfusion mismatches with a normal chest radiograph

Solitary perfusion defect considerably smaller than the chest radiograph defect

Solitary moderate perfusion defect (matched/mismatched) with a normal chest radiograph

Ventilation–perfusion match with some normal perfusion and a normal chest radiograph

Perfusion defect in areas of severe obstructive lung disease, pulmonary oedema, or pleural effusion

Nonsegmental perfusion defects (e.g. cardiomegaly or pleural effusion)

Abnormality that does not ‘fit’ high or low probability status

Perfusion defect surrounded by normal lung perfusion Matched ventilation–perfusion defect with a corresponding defect on a chest radiograph in upper or mid lung zones Small = ≤ 25% area of a pulmonary segment; moderate = 25–75% area of a pulmonary segment; large = > 75% area of a pulmonary segment. JAMA 263 (20): 2753–2759. Copyright © 1990 American Medical Association.

The criteria used to make this assessment are generally widely accepted. A normal V/Q scintigram effectively excludes PE owing to the high sensitivity of the test. Conversely those patients with a high probability test (especially those with consistent clinical symptoms) can be confidently treated for PE. However, patients with intermediate probability studies or a low probability study with strong clinical suspicion should have additional tests to confirm the diagnosis (as up to 40% of these patients have embolus). Finally, care must be taken when interpreting V/Q scintigrams as an erroneous label of PE may be given in certain circumstances. These include: • pulmonary artery wall causes, e.g. vasculitis, infection, irradiation • vascular malformations, e.g. arterial agenesis, arteriovenous malformations, surgical shunts • extrinsic compression of vasculature, e.g. from hilar adenopathy or tumour • prior PE that has not completely resolved • pulmonary artery luminal block, e.g. from non-thrombotic material or tumour. Clues to these entities may be obtained from the chest radiograph and clinical history. However, in difficult cases other tests (e.g. angiography) may be required.

Computed tomography pulmonary angiography Since the mid 1990s helical CT has progressively replaced pulmonary angiography as the mainstay investigation for pulmonary thromboembolism. Although its ability to detect emboli has been extensively studied71–73 its exact role in PE investigation has not

been finalized74–76. However, owing to its widespread availability and noninvasive nature its use is increasing rapidly. As in conventional angiography, acute embolism is seen as an intravascular filling defect. Contrast medium may be seen to flow around or adjacent to the clot (giving a ‘tram track’ appearance only if the vessel is in the plane of the image section) (Fig. 26.14). Ancillary signs include small pleural effusions and focal infarcts in the costophrenic recesses (seen as consolidation)77. More chronic emboli may be seen as crescentic thrombi adherent to the arterial wall; these may be calcified or show evidence of recanalization78–80. Enlargement of the bronchial vessels and a prominent mosaic attenuation pattern (most obvious on thin-section images) may also be seen in chronic cases. These ancillary signs provide enough justification to mandate viewing all image sections on both mediastinal and ‘lung parenchymal’ windows as subtle changes may be missed if only one or the other is viewed77. Infarction may cause a peripheral wedge-shaped region of consolidation (analogous to the Hampton’s hump on chest radiography). This may be better defined on CT81 but is a no more specific sign, unless a vessel can be traced into the apex of the wedge82. CT has proven to be a sensitive method of detecting segmental, lobar and main pulmonary artery emboli and may be reliably used to detect emboli in up to fourth-order vessels (approximately 7 mm in diameter)83,84. Studies have shown the sensitivity of spiral CT for detection of segmental emboli to range between 75% and 91% and specificity between 78% and 96%85–89; estimates for catheter angiography are sensitivity 98% and specificity 97%90. Although pulmonary arteriography is capable of detecting emboli down to the small subsegmental level, the reported

CHAPTER 26

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

Figure 26.14 Pulmonary thromboembolism. (A) Chest radiograph demonstrating a moderate right pleural effusion on the day of symptom onset. (B) CT image taken 2 d after symptom onset reveals a filling defect (representing thrombus) in the right lower lobe pulmonary artery. (C) Further chest radiograph taken 10 d after (A) with the patient on anticoagulant therapy reveals resolution of the previous effusion. As a sequelae, a thin band of atelectasis is seen in the right lower zone.

interobserver variations of 45% and 66%91,92 limit its efficacy at this level. Technical considerations. Although CTPA is normally confined to the area encompassing the central and segmental pulmonary arteries (i.e. from the aortic arch to the inferior pulmonary veins), the remainder of the lung should ideally also be examined. This may reveal ancillary features of PE such as an altered perfusion pattern or pleural effusions, or

evidence of another cause of the patient’s symptoms (e.g. pneumothorax). Technical parameters should be optimized in order to maximize information with the shortest necessary patient breathhold and minimum radiation exposure. Single-slice helical acquisitions tend to be made with thin collimation (2–5 mm) and during the rapid infusion of intravenous contrast medium (3–5 ml s−1). Narrower collimations have been shown to image subsegmental emboli to better advantage93. A table feed of

541

542

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

5 mm s−1 may be used under normal circumstances; however, this rate may be increased for extremely dyspnoeic patients. Inadequate breath-holding can significantly impair the quality of the image because of arterial flow rate changes during the various phases of respiration81 and motion artefact associated with breathing.The advent of multidetector CT (MDCT) has allowed refinement of the technique, such that the whole lung may be examined during a single breath-hold in almost all patients. Indeed, in one recent prospective study the sensitivity, specificity and accuracy of MDCT were 100%, 93% and 95%, respectively, whereas for invasive pulmonary angiography these calculations were 86%, 100% and 97%, respectively94. Improvements in contrast medium administration and data acquisition timing have been further optimized such that adequate vascular opacification may more often be achieved. Furthermore MDCT (with its faster data acquisition and thinner sections) has improved the ability of the technique to resolve peripheral arteries95–99. In one study 40% more subsegmental emboli were detected on MDCT than with 3-mm collimation images. CT should take place during the maximal enhancement with intravenous contrast medium. Experience has shown that high concentration contrast media (350 mg ml−1 of iodine or greater) are associated with significant streak artefact (e.g. in the superior vena cava), which can obscure accurate assessment. Evaluation of various parameters has led to recommendation of an iodine concentration of 120–250 mg ml−1 injected at a rate of 4–5 ml s−1 and a volume of 120–140 ml100,101. Accurate timing of the data acquisition is essential to achieve images at optimal pulmonary arterial opacification. Several methods have been developed. The crudest is a ‘best guess’ method, based on experience that suggests that an 8–15s delay is adequate in most patients. More sophisticated methods include performing ‘time–density’ curve analysis after the injection of a small volume test bolus100. Newer CT systems can be programmed to start acquiring data when predefined levels of enhancement are achieved in the central pulmonary arteries. Whether these techniques offer significantly better results than the ‘best guess’ method has yet to be proven. Patient-related factors affecting opacification include superior vena cava thrombosis and persistent intracardiac shunts, both of which reduce adequacy of opacification. Assessment of the pulmonary arteries may be aided by reformatting the data to allow more suitable planes for visualizing individual arteries (e.g. middle lobe arteries which often run tangential to the plane of a conventional transverse image)102. In addition to modifying the plane of view, dynamic alteration of window setting (from the standard mediastinal setting) may help identify the presence of emboli103. Careful analysis will reduce the incidence of misdiagnosis; e.g. mistaking poorly opacified pulmonary veins and mediastinal lymph nodes for thrombus within an artery. To help minimize such error it is essential to view all pulmonary angiograms on a workstation where vessels can more easily be followed along their course.

Conventional pulmonary angiography Following significant advances in CT technology conventional angiography is now rarely used in the diagnosis of

pulmonary embolic disease. It does however remain the gold standard test104–107. Although invasive, its safety is reassuring (in the hands of an experienced operator) with a mortality ranging from 0% to less than 1%105,108–112 and nonfatal complication rate less than 5%104,113. The most common nonfatal complications are cardiac arrhythmias and cardiac perforation. Mortality is usually related to sudden right ventricular failure from elevated pulmonary artery pressure secondary to contrast medium injection. A contemporaneous perfusion V/Q or chest radiograph study can help refine the target area for investigation and thus reduce the length of study. Subsegmental PEs can also be detected (but with less reliability)114,115. Studies based on clinical follow-up have shown for pulmonary angiography a negative sensitivity of up to 99% with almost 100% specificity for positive tests. The indications for conventional angiography in the past were: • indeterminate V/Q result • high clinical suspicion for PE but low probability V/Q study • high clinical suspicion with indeterminate or negative CT • high probability V/Q findings but relative or absolute contraindication to anticoagulation • before considering fibrinolytic therapy or embolectomy • where in situ thrombolysis for a central PE is contemplated. In light of the vast improvement in noninvasive technologies, possibly the only remaining justification to perform angiography is before in situ thrombolysis. PE may be diagnosed on angiography when an intraluminal filling defect or complete occlusion of an artery is seen. The latter however is a less specific sign for acute embolism as it may also be seen in regions of previous embolism. A further but less specific sign is delayed opacification of the pulmonary tree106,116,117. This sign, however, may also be seen in cases of obstructive airways disease and bronchiectasis. Once an embolus is identified, unless surgical intervention or thrombolysis is considered, the study should be terminated as detection of other emboli will not affect the management of the patient but may significantly increase the overall risk of the procedure.

Noninvasive imaging for deep vein thrombosis As 90% of PEs result from DVT and because the treatment of proximal DVT and PE is identical, diagnosis of the former provides another (usually earlier) endpoint in many diagnostic algorithms. It should be remembered that a negative leg venous US does not firmly exclude PE and so these patients should be further investigated. To this end the use of US and more recently CT and MRI has been investigated. Once again contrast medium studies remain the ‘gold standard’ but advances in the other three techniques have all but left it redundant. Owing to accessibility, time and cost, US has become the first choice investigation for suspected DVT in patients with nondiagnostic PE investigations. In skilled hands US is 90–95% sensitive and 95–98% specific for the diagnosis

CHAPTER 26

of acute DVT. False-positive studies occur in patients with previous DVT and pelvic masses causing venous compression mimicking thrombosis. Recently the technique of CT venography has been investigated. The rationale for this is that both pulmonary emboli and venous thrombus are part of the same process and that detection of either is important. CT features of acute thrombus in the deep veins include: low-attenuation material in the vein, enhancement of the wall of the vein, and more rarely, perivenous oedema. CT should be performed with a suitable delay (in the order of 3 min)118–120 after CTPA, thus obviating the need for further contrast medium. Consensus on optimal CT parameters and whether the added radiation burden of the test is justifiable121 have yet to be reached. Nonetheless, interobserver variability is good. A small percentage of cases have positive CT venography with a negative CTPA. It is not yet known whether they represent a proportion of those patients who are found to have PE but have an initial negative CTPA. Magnetic resonance imaging is also capable of noninvasively imaging the larger pulmonary arteries for the presence of emboli (Fig. 26.15). Advantages over CT include: • contrast medium may not be required • multiplanar imaging • flow studies can aid estimation of potential emboli and coexisting pulmonary arterial hypertension.

Disadvantages however include: • poor spatial resolution relative to CT • long examination time leading to increasing effect of motion artefact • lack of MRI availability. The early studies with MRI suggested its sensitivity for PE in large arteries was 85–90% and specificity was 62–77%. More recently, however, it has been shown to be almost equivalent to that of CTPA122–128. The development of MR angiography (MRA) allows ECG-gated acquisitions and short echo times, thus reducing motion (respiratory) artefact and length of study

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

time. Post-acquisition processing can be used to construct 3D images which can be of use in embolus detection129.The efficacy of 2D angiography has been shown to be at least as good as that of longer sequences in detecting emboli greater than 1 cm in diameter. Contrast-enhanced multiplanar MRA is not yet sufficiently sensitive for the detection of small peripheral emboli130. MR venography (used to look for DVTs) has a sensitivity and specificity profile similar to that of conventional angiography as shown in studies using 2D time-of-flight (TOF) imaging. The added advantage of MRI over other modalities is the ability of T1/T2 relaxation times to age thrombus, thus allowing distinction between acute and chronic DVT.

CONCLUSION The investigation of changes in pulmonary vascular physiology has many facets. The initial clinical assessment may significantly influence further investigations (e.g. postoperative pleuritic chest pain after a hip replacement).Traditionally chest radiography has been the primary method used to investigate pulmonary vascular disorders. Even today it remains a valuable tool which is easily and cost-effectively (in financial and radiation terms) repeatable. Patterns seen in the various altered physiological states (as described above) are well established. Although not always pathognomonic, many findings may be labelled on the basis of the pattern which they adopt (e.g. cardiomegaly with upper lobe venous distension and septal lines referring to raised pulmonary venous pressure, whereas septal lines alone may represent a multitude of conditions). In situations with many seemingly conflicting signs or where no defining abnormality is seen on the chest radiograph, attention may be turned to other imaging modalities. In pulmonary thromboembolism, radionuclide V/Q scintigraphy still has a role in diagnosis. In certain centres cost and radiation burden are reduced by simply performing a perfusion study if the chest radiograph demonstrates no abnormality. The principal cross-sectional tool however in this regard is currently CT. With MDCT, the entire lung can normally be imaged during a single breath-hold.The added information gained from the test must however be weighed against the degree of prior clinical uncertainty as the radiation burden associated with CT is significant (with respect to chest radiography). However, in suspected PE the need to establish a diagnosis far outweighs the relative radiation burden in high risk patients. In cases that are not so clear-cut other tests (e.g. d-dimers) may first be used to stratify the need for further investigation. Furthermore, the use of radionuclide imaging should also be considered. Newer techniques such as MRI have yet to find their niche in the standardized diagnostic algorithm131 but may in specific circumstances have an important role.

REFERENCES Figure 26.15 Pulmonary thromboembolism. MR angiographic image reveals thrombus (arrow) in the right pulmonary artery. (Courtesy of Dr R. Mohiaddin, Royal Brompton Hospital, London.)

1. Newell J D, Higgins C B, Kelley M J 1980 Radiographic– echocardiographic approach to acquired heart disease: diagnosis and assessment of severity. Radiol Clin North Am 18: 387–409 2. Kerley P 1933 Radiology in heart disease. BMJ 2: 594–597

543

544

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

3. Heitzman E R 1993 The radiologic–pathologic correlations. Mosby, St Louis, 1993 4. Jefferson K, Rees S 1973 Clincial cardiac radiology. Butterworth, London, 1973 5. Kostuk W J, Kazamias T M, Gander M P, Simon A L, Ross J Jr 1973 Left ventricular size after acute myocardial infarction. Serial changes and their prognostic significance. Circulation 47: 1174–1179 6. McHugh T J, Forrester J S, Adler L, Zion D, Swan H J 1972 Pulmonary vascular congestion in acute myocardial infarction: hemodynamic and radiologic correlations. Ann Intern Med 76: 29–33 7. Higgins C B, Lipton M J 1980 Radiography of acute myocardial infarction. Radiol Clin North Am 18: 359–368 8. Battler A, Karliner J S, Higgins C B et al 1980 The initial chest x-ray in acute myocardial infarction. Prediction of early and late mortality and survival. Circulation 61: 1004–1009 9. Chang C H 1962 The normal roentgenographic measurement of the right descending pulmonary artery in 1085 cases. Am J Roentgenol 87: 929–935 10. Kuriyama K, Gamsu G, Stern R G, Cann C E, Herfkens R J, Brundage B H 1984 CT-determined pulmonary artery diameters in predicting pulmonary hypertension. Invest Radiol 19: 16–22 11. Ring N J, Marshall A J 2002 Idiopathic dilatation of the pulmonary artery. Br J Radiol 75: 532–535 12. Ng C S, Wells A U, Padley S P 1999 A CT sign of chronic pulmonary arterial hypertension: the ratio of main pulmonary artery to aortic diameter. J Thorac Imaging 14: 270–278 13. Baque-Juston M C, Wells A U, Hansell D M 1999 Pericardial thickening or effusion in patients with pulmonary artery hypertension: a CT study. Am J Roentgenol 172: 361–364 14. Bogren H G, Klipstein R H, Mohiaddin R H et al 1989 Pulmonary artery distensibility and blood flow patterns: a magnetic resonance study of normal subjects and of patients with pulmonary arterial hypertension. Am Heart J 118: 990–999 15. Higgins C B, Wexler L 1976 Clinical and angiographic features of pulmonary arteriovenous fistulas in children. Radiology 119: 171–175 16. Gomes M M, Bernatz P E 1970 Arteriovenous fistulas: a review and tenyear experience at the Mayo Clinic. Mayo Clin Proc 45: 81–102 17. Giampalmao A 1950 The arteriovenous angiomatosis of the lung with hypoxemia. Acta Med Scand 139 (Suppl 248): 1–67 18. Remy J, Remy-Jardin M, Giraud F, Wattinne L 1994 Angioarchitecture of pulmonary arteriovenous malformations: clinical utility of threedimensional helical CT. Radiology 191: 657–664 19. Carvalho P, Lavender J P 1989 The incidence and etiology of the ventilation/ perfusion reverse mismatch defect. Clin Nucl Med 14: 571–576 20. Alexander M S, Peters A M, Cleland J P, Lavender J P 1992 Impaired left lower lobe ventilation in patients with cardiomegaly. An isotope study of mechanisms. Chest 101: 1189–1193 21. Lavender J P, Irving H, Armstrong J D 1981 Krypton-81m ventilation scanning: acute respiratory disease. Am J Roentgenol 136: 309–316 22. Spies W G, Burstein S P, Dillehay G L, Vogelzang R L, Spies S M 1986 Ventilation–perfusion scintigraphy in suspected pulmonary embolism: correlation with pulmonary angiography and refinement of criteria for interpretation. Radiology 159: 383–390 23. Hull R D, Raskob G E, Coates G, Panju A A, Gill G J 1989 A new noninvasive management strategy for patients with suspected pulmonary embolism. Arch Intern Med 149: 2549–2555 24. Braun S D, Newman G E, Ford K, Miller G A, Coleman R E, Dunnick N R 1984 Ventilation–perfusion scanning and pulmonary angiography: correlation in clinical high-probability pulmonary embolism. Am J Roentgenol 143: 977–980 25. Robinson P J 1989 Lung scintigraphy—doubt and certainty in the diagnosis of pulmonary embolism. Clin Radiol 40: 557–560 26. Ia D C, Agustsson M H, Bicoff J P, Weinberg M Jr, Rcilla R A 1964 Stenotic lesions of the pulmonary arteries. Clinical and hemodynamic findings in 84 cases. Am J Cardiol 13: 441–450 27. Schlesinger F G, Meester G T 1967 Supravalvar stenosis of the pulmonary artery. Br Heart J 29: 829–837

28. Lees M H, Menashe V D, Sunderland C O, Mgan C L, Dawson P J 1969 Ehlers–Danlos syndrome associated with multiple pulmonary artery stenoses and tortuous systemic arteries. J Pediatr 75: 1031–1036 29. Baum D, Khoury G H, Ongley P A, Swan H J, Kincaid O W 1964 Congenital stenosis of the pulmonary artery branches. Circulation 29: 680–687 30. Jaffe R B, Condon V R 1975 Mycotic aneurysms of the pulmonary artery and aorta. Radiology 116: 291–298 31. Deterling R A, Clagett O T 1947 Aneurysms of the pulmonary artery. Review of literature and report of a case. Am Heart J 34: 479 32. Remy J, Lemaitre L, Lafitte J J, Vilain M O, Saint M J, Steenhouwer F 1984 Massive hemoptysis of pulmonary arterial origin: diagnosis and treatment. Am J Roentgenol 143: 963–969 33. Oger E 2000 Incidence of venous thromboembolism: a community-based study in Western France. EPI-GETBP Study Group. Groupe d’Etude de la Thrombose de Bretagne Occidentale. Thromb Haemost 83: 657–660 34. Heit J A, Melton L J III, Lohse C M et al 2001 Incidence of venous thromboembolism in hospitalized patients vs community residents. Mayo Clin Proc 76: 1102–1110 35. Dalen J E, Alpert J S 1975 Natural history of pulmonary embolism. Prog Cardiovasc Dis 17: 259–270 36. Provenzale J M, Ortel T L 1995 Anatomic distribution of venous thrombosis in patients with antiphospholipid antibody: imaging findings. Am J Roentgenol 165: 365–368 37. Gilkeson R C, Patz E F Jr, Culhane D, McAdams H P, Provenzale J M 1998 Thoracic imaging features of patients with antiphospholipid antibodies. J Comput Assist Tomogr 22: 241–244 38. Black M D, French G J, Rasuli P, Bouchard A C 1993 Upper extremity deep venous thrombosis. Underdiagnosed and potentially lethal. Chest 103: 1887–1890 39. Freiman D G, Suyemoto J, Wessler S 1965 Frequency of pulmonary thromboembolism in man. N Engl J Med 272: 1278–1280 40. Uhland H, Goldberg L M 1964 Pulmonary embolism: a commonly missed clinical entity. Dis Chest 45: 533–536 41. Morrell M T 1970 The incidence of pulmonary embolism in the elderly. Geriatrics 25: 138–143 42. Yuceoglu Y Z, Rubler S, Eshwar K P, Tchertkoff V, Grishman A 1971 Pulmonary edema associated with pulmonary embolism: a clinicopathological study. Angiology 22: 501–510 43. Moser K M 1977 Pulmonary embolism. Am Rev Respir Dis 115: 829–852 44. Dantzker D R, Bower J S 1982 Alterations in gas exchange following pulmonary thromboembolism. Chest 81: 495–501 45. Moser K M, Auger W R, Fedullo P F, Jamieson S W 1992 Chronic thromboembolic pulmonary hypertension: clinical picture and surgical treatment. Eur Respir J 5: 334–342 46. Dalen J E, Haffajee C I, Alpert J S III, Howe J P, Ockene I S, Paraskos J A 1977 Pulmonary embolism, pulmonary hemorrhage and pulmonary infarction. N Engl J Med 296: 1431–1435 47. Im J G, Choi Y W, Kim H D, Jeong Y K, Han M C 1996 Thin-section CT findings of the lungs: experimentally induced bronchial and pulmonary artery obstruction in pigs. Am J Roentgenol 167: 631–636 48. Tsao M S, Schraufnagel D, Wang N S 1982 Pathogenesis of pulmonary infarction. Am J Med 72: 599–606 49. Perrier A, Desmarais S, Goehring C et al 1997 d-dimer testing for suspected pulmonary embolism in outpatients. Am J Respir Crit Care Med 156: 492–496 50. Egermayer P, Town G I, Turner J G, Heaton D C, Mee A L, Beard M E 1998 Usefulness of d-dimer, blood gas, and respiratory rate measurements for excluding pulmonary embolism. Thorax 53: 830–834 51. Ginsberg J S, Wells P S, Kearon C et al 1998 Sensitivity and specificity of a rapid whole-blood assay for d-dimer in the diagnosis of pulmonary embolism. Ann Intern Med 129: 1006–1011 52. Wells P S, Anderson D R, Rodger M et al 2001 Excluding pulmonary embolism at the bedside without diagnostic imaging: management of patients with suspected pulmonary embolism presenting to the emergency department by using a simple clinical model and d-dimer. Ann Intern Med 135: 98–107

CHAPTER 26

53. Kruip M J, Slob M J, Schijen J H, van der H C, Buller H R 2002 Use of a clinical decision rule in combination with d-dimer concentration in diagnostic workup of patients with suspected pulmonary embolism: a prospective management study. Arch Intern Med 162: 1631–1635 54. MacGillavry M R, Lijmer J G, Sanson B J, Buller H R, Brandjes D P 2001 Diagnostic accuracy of triage tests to exclude pulmonary embolism. Thromb Haemost 85: 995–998 55. Burkill G J, Bell J R, Chinn R J et al 2002 The use of a d-dimer assay in patients undergoing CT pulmonary angiography for suspected pulmonary embolus. Clin Radiol 57: 41–46 56. De M W, Sanson B J, MacGillavry M R et al 2002 Embolus location affects the sensitivity of a rapid quantitative D-dimer assay in the diagnosis of pulmonary embolism. Am J Respir Crit Care Med 165: 345–348 57. Greenspan R H, Ravin C E, Polansky S M, McLoud T C 1982 Accuracy of the chest radiograph in diagnosis of pulmonary embolism. Invest Radiol 17: 539–543 58. Worsley D F, Alavi A, Aronchick J M, Chen J T, Greenspan R H, Ravin C E 1993 Chest radiographic findings in patients with acute pulmonary embolism: observations from the PIOPED Study. Radiology 189: 133–136 59. Buckner C B, Walker C W, Purnell G L 1989 Pulmonary embolism: chest radiographic abnormalities. J Thorac Imaging 4: 23–27 60. Talbot S, Worthington B S, Roebuck E J 1973 Radiographic signs of pulmonary embolism and pulmonary infarction. Thorax 28: 198–203 61. Szucs M M Jr, Brooks H L, Grossman W et al 1971 Diagnostic sensitivity of laboratory findings in acute pulmonary embolism. Ann Intern Med 74: 161–166 62. Heitzman E R, Markarian B, Dailey E T 1972 Pulmonary thromboembolic disease. A lobular concept. Radiology 103: 529–537 63. Simon G 1970 Further observations on the long line shadow across a lower zone of the lung. Br J Radiol 43: 327–332 64. Wilson A G, Joseph A E, Butland R J 1986 The radiology of aseptic cavitation in pulmonary infarction. Clin Radiol 37: 327–333 65. Bynum L J, Wilson J E III 1978 Radiographic features of pleural effusions in pulmonary embolism. Am Rev Respir Dis 117: 829–834 66. Burkill G J, Bell J R, Padley S P 1999 Survey on the use of pulmonary scintigraphy, spiral CT and conventional pulmonary angiography for suspected pulmonary embolism in the British Isles. Clin Radiol 54: 807–810 67. Hull R D, Pineo G F, Stein P D, Mah A F, Butcher M S 2001 Costeffectiveness of currently accepted strategies for pulmonary embolism diagnosis. Semin Thromb Hemost 27: 15–23 68. James J M, Lloyd J J, Leahy B C et al 1992 99Tcm-Technegas and krypton-81m ventilation scintigraphy: a comparison in known respiratory disease. Br J Radiol 65: 1075–1082 69. Macdonald W B, Patrikeos A P, Thompson R I, Adler B D, van der Schaaf A A 2005 Diagnosis of pulmonary embolism: ventilation perfusion scintigraphy versus helical computed tomography pulmonary angiography. Australas Radiol 49: 32–38 70. Anonymous 1990 Value of the ventilation/perfusion scan in acute pulmonary embolism. Results of the prospective investigation of pulmonary embolism diagnosis (PIOPED). The PIOPED Investigators. JAMA 263: 2753–2759 71. Remy-Jardin M, Remy J, Wattinne L, Giraud F 1992 Central pulmonary thromboembolism: diagnosis with spiral volumetric CT with the singlebreath-hold technique—comparison with pulmonary angiography. Radiology 185: 381–387 72. Goodman L R, Curtin J J, Mewissen M W et al 1995 Detection of pulmonary embolism in patients with unresolved clinical and scintigraphic diagnosis: helical CT versus angiography. Am J Roentgenol 164: 1369–1374 73. van Rossum A B, Treurniet F E, Kieft G J, Smith S J, Schepers-Bok R 1996 Role of spiral volumetric computed tomographic scanning in the assessment of patients with clinical suspicion of pulmonary embolism and an abnormal ventilation/perfusion lung scan. Thorax 51: 23–28 74. Hansell D M, Padley S P 1996 Continuous volume computed tomography in pulmonary embolism: the answer, or just another test? Thorax 51: 1–2

•

PULMONARY CIRCULATION AND PULMONARY THROMBOEMBOLISM

75. Henschke C I, Yankelevitz D F, Sicherman N 1997 Evaluation of algorithms for the diagnosis of pulmonary embolism. Semin Ultrasound CT MR 18: 376–382 76. ACCP Consensus Committee on Pulmonary Embolism, American College of Chest Physicians, 1998 Opinions regarding the diagnosis and management of venous thromboembolic disease. Chest 113: 499–504 77. Coche E E, Muller N L, Kim K I, Wiggs B R, Mayo J R 1998 Acute pulmonary embolism: ancillary findings at spiral CT. Radiology 207: 753–758 78. Schwickert H C, Schweden F, Schild H H et al 1994 Pulmonary arteries and lung parenchyma in chronic pulmonary embolism: preoperative and postoperative CT findings. Radiology 191: 351–357 79. Bergin C J, Sirlin C B, Hauschildt J P et al 1997 Chronic thromboembolism: diagnosis with helical CT and MR imaging with angiographic and surgical correlation. Radiology 204: 695–702 80. Roberts H C, Kauczor H U, Schweden F, Thelen M 1997 Spiral CT of pulmonary hypertension and chronic thromboembolism. J Thorac Imaging 12: 118–127 81. Hughes J M, Glazier J B, Maloney J E, West J B 1968 Effect of lung volume on the distribution of pulmonary blood flow in man. Respir Physiol 4: 58–72 82. Ren H, Kuhlman J E, Hruban R H, Fishman E K, Wheeler P S, Hutchins G M 1990 CT of inflation-fixed lungs: wedge-shaped density and vascular sign in the diagnosis of infarction. J Comput Assist Tomogr 14: 82–86 83. Remy-Jardin M, Remy J, Wattinne L, Giraud F 1992 Central pulmonary thromboembolism: diagnosis with spiral volumetric CT with the singlebreath-hold technique—comparison with pulmonary angiography. Radiology 185: 381–387 84. Geraghty J J, Stanford W, Landas S K, Galvin J R 1992 Ultrafast computed tomography in experimental pulmonary embolism. Invest Radiol 27: 60–63 85. van Rossum A B, Pattynama P M, Ton E R et al 1996 Pulmonary embolism: validation of spiral CT angiography in 149 patients. Radiology 201: 467–470 86. van Rossum A B, Pattynama P M, Mallens W M, Hermans J, Heijerman H G 1998 Can helical CT replace scintigraphy in the diagnostic process in suspected pulmonary embolism? A retrolective–prolective cohort study focusing on total diagnostic yield. Eur Radiol 8: 90–96 87. Mayo J R, Remy-Jardin M, Muller N L et al 1997 Pulmonary embolism: prospective comparison of spiral CT with ventilation-perfusion scintigraphy. Radiology 205: 447–452 88. Remy-Jardin M, Remy J, Deschildre F et al 1996 Diagnosis of pulmonary embolism with spiral CT: comparison with pulmonary angiography and scintigraphy. Radiology 200: 699–706 89. Remy-Jardin M, Remy J, Wattinne L, Giraud F 1992 Central pulmonary thromboembolism: diagnosis with spiral volumetric CT with the singlebreath-hold technique—comparison with pulmonary angiography. Radiology 185: 381–387 90. van Erkel A R, van Rossum A B, Bloem J L, Kievit J, Pattynama P M 1996 Spiral CT angiography for suspected pulmonary embolism: a costeffectiveness analysis. Radiology 201: 29–36 91. Diffin D C, Leyendecker J R, Johnson S P, Zucker R J, Grebe P J 1998 Effect of anatomic distribution of pulmonary emboli on interobserver agreement in the interpretation of pulmonary angiography. Am J Roentgenol 171: 1085–1089 92. Stein P D, Athanasoulis C, Alavi A et al 1992 Complications and validity of pulmonary angiography in acute pulmonary embolism. Circulation 85: 462–468 93. Remy-Jardin M, Remy J, Artaud D, Deschildre F, Duhamel A 1997 Peripheral pulmonary arteries: optimization of the spiral CT acquisition protocol. Radiology 204: 157–163 94. Winer-Muram H T, Rydberg J, Johnson M S et al 2004 Suspected acute pulmonary embolism: evaluation with multi-detector row CT versus digital subtraction pulmonary arteriography. Radiology 233: 806–815 95. Patel S, Kazerooni E A, Cascade P N 2003 Pulmonary embolism: optimization of small pulmonary artery visualization at multi-detector row CT. Radiology 227: 455–460

545

546

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

96. Ghaye B, Szapiro D, Mastora I et al 2001 Peripheral pulmonary arteries: how far in the lung does multi-detector row spiral CT allow analysis? Radiology 219: 629–636 97. Schoepf U J, Holzknecht N, Helmberger T K et al 2002 Subsegmental pulmonary emboli: improved detection with thin-collimation multidetector row spiral CT. Radiology 222: 483–490 98. Remy-Jardin M, Tillie-Leblond I, Szapiro D et al 2002 CT angiography of pulmonary embolism in patients with underlying respiratory disease: impact of multislice CT on image quality and negative predictive value. Eur Radiol 12: 1971–1978 99. Coche E, Pawlak S, Dechambre S, Maldague B 2003 Peripheral pulmonary arteries: identification at multi-slice spiral CT with 3D reconstruction. Eur Radiol 13: 815–822 100. Kuzo R S, Goodman L R 1997 CT evaluation of pulmonary embolism: technique and interpretation. Am J Roentgenol 169: 959–965 101. Remy-Jardin M, Tillie-Leblond I, Szapiro D et al 2002 CT angiography of pulmonary embolism in patients with underlying respiratory disease: impact of multislice CT on image quality and negative predictive value. Eur Radiol 12: 1971–1978 102. Remy-Jardin M, Remy J, Cauvain O, Petyt L, Wannebroucq J, Beregi J P 1995 Diagnosis of central pulmonary embolism with helical CT: role of two-dimensional multiplanar reformations. Am J Roentgenol 165: 1131–1138 103. Brink J A, Woodard P K, Horesh L et al 1997 Depiction of pulmonary emboli with spiral CT: optimization of display window settings in a porcine model. Radiology 204: 703–708 104. Stein P D, Athanasoulis C, Alavi A et al 1992 Complications and validity of pulmonary angiography in acute pulmonary embolism. Circulation 85: 462–468 105. Bell W R, Simon T L 1982 Current status of pulmonary thromboembolic disease: pathophysiology, diagnosis, prevention, and treatment. Am Heart J 103: 239–262 106. Dalen J E, Brooks H L, Johnson L W, Meister S G, Szucs M M Jr, Dexter L 1971 Pulmonary angiography in acute pulmonary embolism: indications, techniques, and results in 367 patients. Am Heart J 81: 175–185 107. Robin E D 1977 Overdiagnosis and overtreatment of pulmonary embolism: the emperor may have no clothes. Ann Intern Med 87: 775–781 108. Moses D C, Silver T M, Bookstein J J 1974 The complementary roles of chest radiography, lung scanning, and selective pulmonary angiography in the diagnosis of pulmonary embolism. Circulation 49: 179–188 109. Cheely R, McCartney W H, Perry J R et al 1981 The role of noninvasive tests versus pulmonary angiography in the diagnosis of pulmonary embolism. Am J Med 70: 17–22 110. Novelline R A, Baltarowich O H, Athanasoulis C A, Waltman A C, Greenfield A J, McKusick K A 1978 The clinical course of patients with suspected pulmonary embolism and a negative pulmonary arteriogram. Radiology 126: 561–567 111. Marsh J D, Glynn M, Torman H A 1983 Pulmonary angiography. Application in a new spectrum of patients. Am J Med 75: 763–770 112. Mills S R, Jackson D C, Older R A, Heaston D K, Moore A V 1980 The incidence, etiologies, and avoidance of complications of pulmonary angiography in a large series. Radiology 136: 295–299 113. Zuckerman D A, Sterling K M, Oser R F 1996 Safety of pulmonary angiography in the 1990s. J Vasc Interv Radiol 7: 199–205 114. Stein P D, Henry J W, Gottschalk A 1999 Reassessment of pulmonary angiography for the diagnosis of pulmonary embolism:

115.

116. 117. 118.

119.

120.

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

relation of interpreter agreement to the order of the involved pulmonary arterial branch. Radiology 210: 689–691 van Beek E J, Bakker A J, Reekers J A 1996 Pulmonary embolism: interobserver agreement in the interpretation of conventional angiographic and DSA images in patients with nondiagnostic lung scan results. Radiology 198: 721–724 Bookstein J J, Silver T M 1974 The angiographic differential diagnosis of acute pulmonary embolism. Radiology 110: 25–33 Sagel S S, Greenspan R H 1971 Nonuniform pulmonary arterial perfusion. Pulmonary embolism? Radiology 99: 541–548 Yankelevitz D F, Gamsu G, Shah A et al 2000 Optimization of combined CT pulmonary angiography with lower extremity CT venography. Am J Roentgenol 174: 67–69 Bruce D, Loud P A, Klippenstein D L, Grossman Z D, Katz D S 2001 Combined CT venography and pulmonary angiography: how much venous enhancement is routinely obtained? Am J Roentgenol 176: 1281–1285 Loud P A, Grossman Z D, Klippenstein D L, Ray C E 1998 Combined CT venography and pulmonary angiography: a new diagnostic technique for suspected thromboembolic disease. Am J Roentgenol 170: 951–954 Rademaker J, Griesshaber V, Hidajat N, Oestmann J W, Felix R 2001 Combined CT pulmonary angiography and venography for diagnosis of pulmonary embolism and deep vein thrombosis: radiation dose. J Thorac Imaging 16: 297–299 Erdman W A, Peshock R M, Redman H C et al 1994 Pulmonary embolism: comparison of MR images with radionuclide and angiographic studies. Radiology 190: 499–508 Schiebler M L, Holland G A, Hatabu H et al 1993 Suspected pulmonary embolism: prospective evaluation with pulmonary MR angiography. Radiology 189: 125–131 Grist T M, Sostman H D, MacFall J R et al 1993 Pulmonary angiography with MR imaging: preliminary clinical experience. Radiology 189: 523–530 Haage P, Piroth W, Krombach G et al 2003 Pulmonary embolism: comparison of angiography with spiral computed tomography, magnetic resonance angiography, and real-time magnetic resonance imaging. Am J Respir Crit Care Med 167: 729–734 Oudkerk M, van Beek E J, Wielopolski P et al 2002 Comparison of contrast-enhanced magnetic resonance angiography and conventional pulmonary angiography for the diagnosis of pulmonary embolism: a prospective study. Lancet 359: 1643–1647 Hurst D R, Kazerooni E A, Stafford-Johnson D et al 1999 Diagnosis of pulmonary embolism: comparison of CT angiography and MR angiography in canines. J Vasc Interv Radiol 10: 309–318 Ley S, Kauczor H U, Heussel C P et al 2003 Value of contrastenhanced MR angiography and helical CT angiography in chronic thromboembolic pulmonary hypertension. Eur Radiol 13: 2365–2371 Wielopolski P A, Haacke E M, Adler L P 1992 Three-dimensional MR imaging of the pulmonary vasculature: preliminary experience. Radiology 183: 465–472 Loubeyre P, Revel D, Douek P et al 1994 Dynamic contrastenhanced MR angiography of pulmonary embolism: comparison with pulmonary angiography. Am J Roentgenol 162: 1035–1039 British Thoracic Society 2003 British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 58: 470–483

CHAPTER

The Aorta, including Intervention

27

Tony Nicholson and Jai Patel

• The normal aorta • Congenital aortic abnormalities • Acquired aortic abnormalities

THE NORMAL AORTA The aorta is the main artery of the body delivering oxygenated blood from the left ventricle to all parts. In common with other arteries it has three histologically distinct layers; an intima consisting of a thin endothelial layer; a media containing an elastic lamella, smooth muscle and connective tissue; and a thin outer adventitia made of connective and elastic tissues and which contains nerves, lymphatics and the vasa vasorum1. The aortic root begins at the upper part of the left ventricle and is approximately 3 cm in diameter. A normal aorta passes superiorly and to the right for approximately 5 cm, then arches posteriorly and to the left over the root of the left lung, descending within the thorax on the left side of the vertebral column, gradually achieving the median plane, and becoming the abdominal aorta, when it enters the abdominal cavity through the aortic hiatus in the diaphragm.The abdominal aorta is approximately 2 cm in diameter; it ends slightly to the left of the median plane at the lower border of the fourth lumbar vertebra by dividing into the right and left common iliac arteries2. The aortic root and most of the ascending aorta is contained within the pericardium. The root consists of three sinuses: the right coronary artery arising from the right coronary sinus, the left coronary artery from the left coronary sinus and a noncoronary sinus which is usually to the right and posterior. The ascending aorta forms the right mediastinal border on a postero-anterior (PA) chest radiograph. It becomes the aortic arch at the origin of the innominate artery and also gives rise to the left common carotid and left subclavian arteries. Around three-quarters of aortas show this ‘normal’ branch pattern, but in 20% the innomi-

nate and left common carotid arteries have a common origin, and in 6% the left vertebral artery arises directly from the aortic arch. The aortic arch ends and the descending thoracic aorta begins immediately beyond the origin of the left subclavian artery. At this site the ligamentum venosum (the embryological ductus arteriosus which closes within a few days of birth) joins the inferior concavity of the aortic arch to the main pulmonary artery. The aorta is fixed at this point. Occasionally the duct may persist as a short diverticulum3.

CONGENITAL AORTIC ABNORMALITIES Vascular rings A vascular ring is a condition in which an anomalous configuration of the arch and/or its associated vessels completely or incompletely surrounds the trachea and oesophagus causing compression of these structures. Neonates present with respiratory distress, older children present with stridor or dysphagia. The two most common types of complete vascular rings, which account for 85–95% of cases, are double aortic arch and right aortic arch with left ligamentum arteriosum. Tracheomalacia results from compression of the trachea by vascular rings4. The key to understanding vascular rings is to appreciate how the arch develops. From about 4 weeks gestation there are paired ventral aortas joined to paired dorsal aortas by six pairs of arterial arches, though these are never all present simultaneously (Fig. 27.1). The fourth arch is the most important when considering vascular rings.

Double aortic arch In double aortic arch both the right and left fourth arches persist (Fig. 27.2). The right arch is the larger in 75%. In about 30% of cases the smaller, or less dominant, of the arches is atretic but remains in continuity with the descending aorta, maintaining the complete ring.This may be difficult to identify radiologically. The normally positioned left, or anterior, arch exits the pericardium and joins the left-sided descending thoracic aorta after

548

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

presence is suggestive of the existence of an associated anomaly as persistence of the right arch with involution of the left creates a situation in which the origins of the left subclavian artery and ductus arteriosus can vary.These configurations can produce a vascular ring. Ventral Aorta 4th

4th

Dorsal Aorta

Figure 27.1 Embryology of the aortic arch. Six pairs of arterial arches develop between paired ventral and dorsal aortas from approximately day 26. They are not all present at the same time and normally the left fourth arch develops into the aortic arch.

giving off the left subclavian artery. It is anterior to the trachea. The ligamentum arteriosum is positioned normally. The posterior, or right, arch joins the descending thoracic aorta at the same level as the anterior arch but reaches that point from an extreme posterior course behind the oesophagus. Hence the descending aorta is more commonly on the left but can be on either side5. Other vascular rings associated with aortic arch abnormalities

Where the left fourth branchial arch involutes and the right remains, a right aortic arch is present. It has a frequency of 1 in 100 000 and can occur in the absence of any other anomalies. Its

Right aortic arch with aberrant left subclavian artery and left ligamentum arteriosum (Fig. 27.3). In these cases, the right arch first gives off the left carotid artery, which travels anterior to the trachea. It then gives off the right carotid, followed by the right subclavian artery, and, lastly, the left subclavian artery, which courses in a retro-oesophageal position and gives rise to the ligamentum arteriosum from its base, completing the ring as it attaches to the pulmonary artery. Right aortic arch with mirror-image branching and retrooesophageal ligamentum arteriosum (Fig. 27.4). In these cases, only partial resorption of the distal left fourth arch occurs. The first vessel originating from the right arch is a left innominate artery which, in turn, branches into the left carotid and a left subclavian artery. These course anterior to the trachea. The right carotid artery and a right subclavian artery then arise. The ligamentum arteriosum is the final structure arising from the arch in this sequence. It originates from the Kommerell diverticulum, which represents the nonresorbed remnant of the left fourth arch and is situated at the point of merger between the right arch and the proximal descending thoracic aorta. The ligamentum passes leftward and behind the oesophagus and then travels anteriorly to join with the left pulmonary artery, completing the ring.

Vascular rings associated with left aortic arch Two extremely rare complete rings occur in the presence of a left aortic arch, and both are associated with a right-sided descending thoracic aorta.

Figure 27.2 Double aortic arch. (A) 3D MRA reconstruction. (Courtesy of Dr M. Shivanathan, The Leeds Teaching Hospitals NHS Trust, UK.) (B) Diagram. LSCA = left subclavian artery, RCCA = right common carotid artery.

LSCA RCCA

B

CHAPTER 27

RSCA

LSCA

RCCA LCCA

Figure 27.3 Diagram of right aortic arch with aberrant left subclavian artery and left ligamentum arteriosum. LCCA = left common carotid artery, LSCA = left subclavian artery, RCCA = right common carotid artery, RSCA = right subclavian artery.

Left aortic arch with right descending aorta and right ligamentum arteriosum The first arch vessel to exit the left aortic arch is the right common carotid, which passes anterior to the trachea. The left carotid is next, followed by the left subclavian artery. The right subclavian artery arises more distally as a branch of the proximal right-sided descending aorta. The ligamentum arteriosum arises from the base of the right subclavian artery or a nearby diverticulum and travels to the right pulmonary artery. Left aortic arch, right descending aorta and atretic right aortic arch The brachiocephalic vessels arise from the left-sided arch in a normal arrangement. The left arch passes behind the oesophagus to join a right-sided descending aorta. An atretic right arch is present and completes the ring.

Aortic arch abnormalities without an anatomic ring

• THE AORTA, INCLUDING INTERVENTION

Anomalous innominate artery The innominate artery originates from a more distal and leftward position on the arch than normal. As it takes its course from left to right, it crosses the trachea anteriorly and in doing so compresses the trachea. Retro-oesophageal right subclavian artery with an otherwise normal left arch This is the most common of the arch vessel anomalies, occurring in about 0.5% of the population. In these cases, the right subclavian artery does not arise from an innominate trunk with the right carotid artery but originates as the last brachiocephalic branch from the descending aorta and takes a retro-oesophageal route to its destination.

Imaging Chest radiography is the first and most commonly performed imaging investigation. The identification of a right aortic arch on a chest radiograph in a child with airway difficulties, respiratory distress, or dysphagia should alert the radiologist to the likelihood of a vascular ring (Fig. 27.5). In double aortic arch, the arch location is often ill defined (Fig. 27.6). However, with this and other types of vascular ring, compression of the trachea on antero-posterior (AP) and lateral radiographs, hyperinflation and/or atelectasis of some of the lobes of either lung may be a feature. Many authorities consider barium oesophagography to be the most important study in patients with a suspected vascular ring, and it is diagnostic in the vast majority of cases (Fig. 27.7). A double aortic arch produces bilateral and posterior compressions of the oesophagus, which remain constant regardless of peristalsis (Fig. 27.8). The right indentation is usually slightly higher than the left and the posterior one is usually rather wide and courses in a downward direction from right to left. Where the right subclavian artery takes a retro-oesophageal course there is a posterior defect slanting upward from left to

It should be noted that there are abnormalities of the aortic arch which produce compression symptoms without an anatomic ring.

RSCA

RCCA

LlnA

Figure 27.4 Diagram of right aortic arch with mirror image branching and retro-oesophageal ligamentum arteriosum. LInA = left innominate artery, RCCA = right common carotid artery, RSCA = right subclavian artery

Figure 27.5 Chest radiograph of child with right aortic arch. Stridor in this situation suggests the presence of a vascular ring.

549

550

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.6 Chest radiograph of a child presenting with stridor and swallowing difficulties. The superior mediastinum is slightly widened and the aortic knuckle is not seen. The possibility of a double aortic arch should be considered.

right.The posterior defect in these cases is usually not as broad as that found in a double aortic arch. Echocardiography has been increasingly used for the diagnosis of a vascular ring using the suprasternal window but there are limitations. Structures without a lumen, such as a ligamen-

Figure 27.8 Barium swallow in a child with a double aortic arch. Note the constant impressions on either side of the oesophagus.

tum arteriosum or an atretic arch, are difficult to identify with colour-flow echocardiography. Identification of compressed midline structures and their relationship to encircling vascular anomalies may be difficult to detect. However, associated congenital cardiac defects can be examined. Computed tomography (CT), magnetic resonance imaging (MRI) and digital subtraction angiography (DSA) can be useful diagnostic tools because they reveal the positions of vascular, tracheobronchial and oesophageal structures, and their relationships to one another. Magnetic resonance angiography (MRA) is an excellent substitute for DSA but young patients may require general anaesthesia and where there is already airway compromise this should be avoided5.

Coarctation of the aorta

Figure 27.7 (A,B) Barium swallow in a child with a vascular ring. Note the constant posterior impressions on the oesophagus.

Aortic coarctation is rare, most commonly affecting the aortic isthmus (95% of cases) and rarely the more distal thoracic and abdominal aorta, where it is a cause of mid aortic syndrome (see below). Presentation is usually with hypertension and, in infants, failure to thrive.The femoral pulses are usually delayed and weakened compared with the carotid and arm pulses and there is a characteristic murmur, though this can disappear in older patients if the site of coarctation becomes occluded6. Eighty per cent of patients are male. In females, coarctation is associated with Turner's syndrome. Patients can present in infancy or adulthood.The infantile type of coarctation is usually proximal to the ductus arteriosus and 50% are associated with

CHAPTER 27

• THE AORTA, INCLUDING INTERVENTION

other congenital heart defects such as bicuspid aortic valve, ventricular septal defect (VSD), or hypoplastic left heart syndrome. Cystic medial necrosis of the aorta is also associated. At birth the ductus arteriosus closes, resulting in reduced blood supply to the distal aorta, which is perfused up to this time via the duct. The consequent increased strain on the heart leads to heart failure, as no collateral pathways are required in utero. The adult type of coarctation is usually distal to the ductus arteriosus and the left subclavian artery and therefore collaterals develop in utero. Consequently, whilst infants present with hypertension and failure to thrive, adults present with hypertension and the classical signs of collateral vessels on the chest radiograph (Fig. 27.9). Rib notching usually takes several years to develop. It is caused by pressure erosion of the inferior aspects of the upper adjacent ribs by enlarged and tortuous intercostal arteries. It is usually bilateral but asymmetric, and most often spares the first two ribs where intercostal arteries arise from the costocervical trunk proximal to the usual site of coarctation and do not form part of the collateral circulation. The rib notches may be shallow or deep and usually have a corticated margin. Unilateral absence of rib notching is seen—on the left in the presence of a stenosed or occluded left subclavian artery, and on the right in association with anomalous origin of the right subclavian artery from below the level of the coarctation. Other radiographic features include cardiomegally, particularly in older adults, a prominent ascending aorta (especially with a bicuspid aortic valve) and various aortic knuckle abnormalities. There may be a ‘3’ sign due to enlargement of the left subclavian artery above the coarctation (Fig. 27.9), the

narrowed segment and then localized post-stenotic dilatation of the aorta below. Occasionally this post-stenotic dilatation gives the impression of a low aortic knuckle. The whole area of the aortic knuckle may appear small and flat. On a lateral radiograph the enlarged internal mammary artery may be seen behind the sternum. The above signs together with a chest radiograph are usually enough to make the diagnosis. Echocardiography may be difficult in adults and older children but in infants it may be useful in identifying the area of stenosis and the associated congenital cardiac defects. Using continuous wave Doppler measurements of flow velocities above and below the coarctation and a modified Bernoulli equation it is possible to assess the degree of stenosis. MRI is the imaging technique of choice in both infantile and adult coarctation (Fig. 27.10). It has considerable advantages as it is noninvasive and also is useful for post-treatment follow-up.T1weighted spin-echo sequences will show the whole of the aorta, the major branches and the larger collaterals. Cine phase-contrast imaging can be used to estimate the gradient across the coarctation. Gadolinium-enhanced three-dimensional (3D) MRA gives the best anatomical images. Multidetector CT (MDCT) can also provide exquisite images but is rarely used because of the radiation implications. Angiography, previously the imaging procedure of choice, is now rarely required unless cardiac catheterization is necessary for the investigation of associated cardiac abnormalities. The coarctation can usually be crossed from the femoral arterial route but this may be impossible and require brachial artery catheterization. Asymmetry of the lesion may require the acquisition of multiple views.

Figure 27.9 Chest radiograph in a patient with coarctation. There is rib notching and enlargement of the left subclavian artery causing a ‘3’ sign.

Figure 27.10 Spin-echo T1-weighted oblique MRI of a patient presenting with hypertension and radiofemoral delay. There is a coarctation of the aorta just beyond the left subclavian artery.

551

552

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Management If undetected or untreated, death from cardiac failure, aortic rupture, infective endocarditis, and intracerebral haemorrhage from associated cerebral aneurysms is inevitable, with only 50% of patients surviving into their 30s6. Surgery used to be the most common treatment for significant coarctation and is still often required. If the lesion is short and the aorta can be adequately mobilized, resection and end-to-end anastomosis may be possible; this will give the best long-term result. Failing this, the most usual procedure is repair by subclavian patch. This involves resecting the lesion, transecting the left subclavian artery before the origin of the vertebral artery, incising longitudinally and turning it down as a patch repair.This repair will grow with the patient. Collateral supply to the left subclavian artery, which will be mainly by the vertebral artery, causes few problems. Synthetic graft material is unsuitable in children because it does not stretch. There is a mortality of 2.6–3.1%, a 5.4% incidence of aneurysm formation and a paraplegia rate of 1%7. Percutaneous transluminal angioplasty (PTA) was first used in 1982 to treat a recoarctation in a critically ill patient who had undergone previous surgical repair. PTA continues to be very successful at treating post-surgical recoarctation with early success rates of 90% and restenosis rates of 16–30%. The aorta is carefully measured proximal to the coarctation site and a balloon chosen that is 2-mm narrower than this. With native coarctation, it is generally accepted that PTA success, defined as a reduction in gradient across a coarctation to less than 20 mmHg, can be obtained in 80% of patients (Figs 27.11, 27.12). There is a 7.4% incidence of aneurysm formation in neonates and adults, and recoarctation has occurred in every series, though this is less common in adults and older children (7.3–8%) than in neonates. Long-term normalization of blood pressure can be achieved in 75% of patients. In view of the acceptable success

Figure 27.12 Arch aortogram of the patient described in Fig. 27.11 following balloon angioplasty of the coarctation site. There is still a mild stenosis but the gradient was reduced to less that 5 mmHg.

rate and low complication rate compared to surgery, PTA is the primary method of treatment in adults, adolescents and children outside of infancy7. Some consider that inserting stents at the coarctation site post balloon dilatation gives better long-term results. However, there is no evidence for this currently. Stents should be reserved for initial failure of PTA due to recoil. They should not be used in infants other than as a short-term measure in the critically ill who are unsuitable candidates for immediate surgery. Stent grafts should be available at all times in order to treat the very rare but invariably fatal complication of aortic rupture at PTA8.

Pseudocoarctation An elongated aortic arch will bulge posteriorly above the point at which it is fixed by the ligament. This can give an appearance on a chest radiograph of a ‘3’ sign similar to true coarctation. There is usually no significant haemodynamic obstruction. MRI will demonstrate the true anatomy.

Aortic atresia

Figure 27.11 Arch aortography in a patient with coarctation. The aortogram demonstrates the coarctation just beyond the left subclavian artery. There was a gradient of 65 mmHg across the coarctation.

Aortic atresia is associated with hypoplastic left heart syndrome. The ascending aorta is variable in size but is usually very small and no larger than one of the brachiocephalic arteries. Blood flow from the heart to the aorta is through the pulmonary trunk and the arterial duct, with the aortic arch filling in a retrograde direction. The brachiocephalic branches arise normally from the arch, and the coronary arteries are supplied via the diminutive ascending aorta. Survival depends on maintaining patency of the duct by giving prostaglandin E1. The Norwood operation converts the morphological right ventricle into the systemic ventricle, by anastomosing the pulmonary trunk to the ascending aorta.The atrial septum is excised. Blood flow to the pulmonary arteries is maintained through a modified Blalock–Taussig shunt. The duct is then closed.

CHAPTER 27

Interrupted aortic arch is rare and thought to be a result of faulty development of the aortic arch system during the fifth to seventh weeks of fetal development. It is almost always associated with a large VSD. It can occur distal to the left subclavian artery (type A, 30–40%), distal to the left common carotid artery (type B, 53%), or distal to the innominate artery (type C, 4%). Patients with type B often have a chromosomal abnormality called the DiGeorge syndrome.

ACQUIRED AORTIC ABNORMALITIES Traumatic aortic injury Trauma to the thoracic aorta can result from both penetrating and blunt chest injuries. In the majority of patients (80–90%) there is complete rupture of the aorta, with death occurring at the scene of the accident. In half of the 10–20% of patients who reach hospital alive, the aorta ruptures within the ensuing 24 h. These patients have injuries in which the adventitia has remained intact or a false aneurysm has formed, thus maintaining the integrity of the aorta. Only 2% of patients survive beyond 4 months9. The clinical diagnosis of traumatic aortic injury (TAI) can be difficult due to the lack of specific symptoms or signs in many patients and poor diagnostic accuracy of initial chest radiographs performed as part of a trauma series. Given this, the mechanism of trauma plays a crucial role in alerting the clinician to the possibility of TAI; examples include road traffic accidents (RTAs) at speeds greater than 30 mph; unrestrained occupants of vehicles and pedestrians involved in RTAs; falls from greater than 10 feet; and severe crush injuries to the chest. A high index of suspicion should be exercised in these patients to the possibility of TAI. A number of likely mechanisms of injury to the aorta

• THE AORTA, INCLUDING INTERVENTION

have been proposed.The aorta has a natural ‘weakness’ between the junction of the fairly mobile aortic arch and relatively fixed descending aorta, the site of the ligamentum arteriosum. Sudden deceleration can lead to shearing forces acting on the aorta at the level of the aortic isthmus, resulting in injury and rupture. Other mechanisms of injury include AP compression displacing the heart to the left (torsion stress). This mechanism is responsible for injuries involving the ascending aorta and is seen with vertical deceleration in falls from a height. ‘Osseous pinch’ is suggested to occur from compression of the heart and aorta between the sternum and vertebral column. The aortic isthmus is the most common site of injury (approximately 90% of cases). The ascending aorta is involved in 7–8% of cases; injury at this site is invariably fatal. Injuries to the descending aorta at the diaphragmatic hiatus occur in 2–3% of cases10. Aortic injuries can range from intimal tears to complete transection. Lesions can be classified as follows: (A) intimal haemorrhage; (B) intimal haemorrhage with laceration; (C) medial laceration; (D) complete laceration; (E) false aneurysm; and (E) peri-aortic haemorrhage9. Due to the nature of the initiating trauma, injuries to other organs are commonly associated. These may detract attention from a possible TAI, or may in themselves be a significant factor endangering the patient's life.

Imaging Chest radiography has been the traditional screening investigation in the initial evaluation of TAI, allowing triage of patients for subsequent thoracic aortography. A number of radiographic signs may be seen in TAI, largely based on the presence of mediastinal haematoma (Table 27.1, Fig. 27.13)10.The most commonly quoted sign is widening or abnormality of the mediastinum,

Table 27.1 CHEST RADIOGRAPH FINDINGS ASSOCIATED WITH TRAUMATIC AORTIC INJURY9

Features directly related to the aortic injury Features related to the presence of mediastinal haematoma

Other features

*

Most commonly seen signs.

Sensitivity (%)

Specificity (%)

Irregularity or blurring of the aortic knuckle contour*

72

47

Enlargement of the aortic knuckle*

35

60

Upper mediastinal widening

90

19

Aortopulmonary window opacification*

42

83

Displacement of a nasogastric tube to the right of the midline

9

96

Displacement of the trachea or an endotracheal tube to the right

20

92 N/A

*

Anterior displacement of trachea on lateral radiograph

N/A

Downward displacement of left main stem bronchus

3

99

Enlarged cardiac silhouette secondary to haemopericardium

N/A

N/A

Right lateral displacement of superior vena cava

7

96

Obscuration of the azygos vein

N/A

N/A 95

Left apical cap

5

Opacification of the medial border of the left lung

12

95

Left haemothorax

5

97

Widened right paratracheal stripe

30

99

Widened right or left paraspinal stripe

2

97

553

554

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.13 Supine chest radiograph of a patient involved in a road traffic accident demonstrates signs of traumatic aortic injury. There is widening of the superior mediastinum (M/C ratio approximately 3). The right paratracheal stripe is widened and there is deviation of the trachea and the nasogastric tube to the right of the midline. The contour of the aortic knuckle is enlarged and partially obscured by mediastinal haematoma.

which is seen in 92.7% of patients. It is the only sign where good interobserver agreement is consistently achieved. However, the specificity of mediastinal widening is low, as it is frequently the result of bleeding from small mediastinal arteries or veins. Given that the chest radiograph in patients with potential TAI is taken in the supine position, the interpretation of mediastinal widening can be problematical. In an attempt to overcome this, absolute values for the upper limit of normal for mediastinal width and ratios of the mediastinal width to chest width at the level of the aortic arch (M/C ratio) have been proposed (8 cm and 0.25, respectively). However, these values result in a wide range of reported sensitivities and specificities. It must be remembered that a normal mediastinum does not exclude a significant aortic injury and is seen in 7.3% of patients with subsequent proven Figure 27.14 (A) CT of a patient who was a pedestrian in a road traffic accident. There is disruption of the aortic wall with contrast pooling in an extra-arterial location in the distal arch. (B) Thoracic aortogram confirms the presence of a pseudoaneurysm at the aortic concavity just distal to the arch. The aortic tear can be seen as a linear lucency traversing the aortic lumen at this site (arrowheads).

TAI. However, an overall 90% sensitivity, 25% specificity and 95% negative predictive value makes chest radiography a useful screening tool for mediastinal haemorrhage. Thoracic aortography is accurate but it is also expensive, invasive and not without complications. Aortography in two planes has a sensitivity and specificity of 96% and 98%, respectively10. False-positive diagnoses often result from the presence of a prominent ductus diverticulum, severe aortic atheroma, or double densities from overlapping adjacent vessels; false-negative diagnoses are made when there is poor opacification of the aorta with inadequate views and are mostly due to small intimal defects. Angiographic findings of TAI include the presence of an intimal flap, dissection, pseudoaneurysm or, less commonly, pseudocoarctation (Fig. 27.14). Angiographic appearances are those of an irregularity of the aortic contour due to an intimal tear. A luminal filling defect may be seen if there is thrombus formation associated with the intimal injury. Dissection results in a linear filling defect dividing the aortic lumen into two. There may be differential opacification of the true and false lumens. If there is a pseudoaneurysm, discontinuity or asymmetry/irregularity of the aortic contour may be seen. There is usually an acute margin at the junction of the abnormal and normal aortic wall, differentiating a pseudoaneurysm from a ductus diverticulum, which classically has a smooth symmetrical contour and obtuse margins with the ‘normal’ aorta (Fig. 27.15). Transoesophageal echocardiography (TOE) has a sensitivity of 91% and specificity of 98% for demonstration of isthmic aortic injuries10. Direct signs of aortic injury include an intimal flap, intramural haematoma, pseudoaneurysm, pseudocoarctation and complete transaction. TOE can be performed at the patient's bedside in 15–20 min. However, it is operator dependent and may be contraindicated in the presence of severe facial injuries or unstable cervical spine fractures. The entire aortic circumference may not be adequately visualized in approximately 30% of patients.TOE is also extremely useful for guiding endovascular intervention, in particular providing excellent visualization of the proximal aorta for accurate placement of stent grafts adjacent to aortic branch vessel origins.

CHAPTER 27

• THE AORTA, INCLUDING INTERVENTION

Figure 27.15 Aortograms of two patients demonstrating the difference between (A) a normal ductus diverticulum and (B) a traumatic pseudoaneurysm. The ductus diverticulum has a smooth symmetrical contour with obtuse margins at its junction with the ‘normal’ aorta (arrows). In comparison the pseudoaneurysm is asymmetric and has an acute proximal margin with the normal aorta (arrowhead).

A

The development of faster machines has led to spiral CT angiography (spCTA) being used not just for screening but also for preoperative imaging, in many cases bypassing thoracic aortography altogether. SpCTA can demonstrate direct signs of aortic trauma such as intramural haematoma and contrast extravasation in addition to those seen with conventional aortography (Fig. 27.16), as well as indirect signs such as mediastinal haematoma. Most trauma units will now have a low threshold for the use of SpCTA in assessing patients with an appropriate mechanism of injury. Given the potential risk of missing a TAI, the aorta should be examined in any trauma patient undergoing CT for evaluation of head and/or abdominal injury. In most cases, an initial unenhanced CT of the head will be followed by a post-contrast arterial phase CT of the thorax and subsequent examination of the abdomen. The finding of peri-aortic haematoma on abdominal CT is also said to be a predictor of TAI and should instigate a search to exclude TAI11. SpCTA has a sensitivity of 96.2%, a specificity of 99.8%, an accuracy of 99.7% and a negative predictive value of 99.9% (in the presence of a normal mediastinum)10. Falsenegative examinations result from poor contrast enhancement, partial volume effect and axial images failing to demonstrate transverse injuries and lesions on the convexity of the aortic arch. The rate of false-negative examinations should decrease with the wider availability of MDCT, allowing thinner slices and better multiplanar reconstructions. As with aortography, false positives can arise from the presence of severe atheroma or a ductus diverticulum. These can be differentiated from a small pseudoaneurysm by the absence of surrounding mediastinal haematoma. Isolated anterior mediastinal haematoma is unlikely to be the result of TAI. The supreme intercostal vein, bronchial artery infundibulum and movement artefact are also sources of false-positive diagnoses. Images can also be degraded by streak artefact from the shoulders, contrast injection into a left arm vein and monitoring lines.

Management The traditional practice of immediate repair of TAI is based on the high mortality in the first 24 h in patients who survive the initial traumatic event. However, the outcome in these patients is related as much to associated injuries as it is to the TAI itself. This has led some investigators to suggest delayed repair or conservative management of TAI, allowing time to manage other serious or potentially life-threatening injuries12. Controlled hypotension is instigated using beta-blockers and vasodilators to keep the mean arterial pressure below 70 mmHg. Beta-blockers reduce the rate rise of systolic ejection of the left ventricle, decreasing the shearing force on the aortic wall. The selective use of this approach has successfully reduced the overall morbidity and mortality in patients with other significant injuries12,13. As surgery has a high morbidity and mortality, thoracic stent grafting (TSG) has been introduced for the management of TAI14 (see Fig. 27.16). Small cohort series and case reports have shown TSGs to be successful in the emergent management of TAI with substantially reduced morbidity and mortality compared to surgical repair. In common with the principles of stent grafting of aortic aneurysms (see below), endovascular repair of TAI ideally requires at least 15 mm of aorta proximal to the injury to achieve an adequate seal. Given that the isthmus is the most common site of TAI, there may be an insufficient length of aorta between the origin of the left subclavian artery and the site of the injury. An increase in the length of the proximal ‘landing zone’ can be achieved by intentionally covering the left subclavian artery origin and extending the stent graft to the origin of the left common carotid artery. Initial concerns over acute upper limb ischaemia have not been unfounded; most patients have adequate collateral supply via the circle of Willis and left vertebral artery to make surgical carotid-to-subclavian bypass unnecessary. The calibre of most currently available devices is 22–24 F, requiring approximately 8-mm vessels for their passage. Since many patients with TAI are young, they often have iliac arteries that are of too small a calibre, necessitating

555

556

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.16 (A) Axial and (B) sagittal oblique reconstruction CT of the patient in Fig. 27.13 confirm the presence of a superolateral intimal flap and an inferomedial pseudoaneurysm producing a pseudocoarctation. CT also demonstrates significant mediastinal haemorrhage (displaced nasogastric tube). Intra-operative aortograms (C) pre and (D) post thoracic stent graft insertion.

a common iliac or aortic conduit to provide adequate access. The incidence of paraplegia is low at 2% and is reduced by limiting the length of the stent graft to the minimum required to treat the lesion. Cerebrospinal fluid (CSF) drainage is mandatory to reduce CSF pressure and improve spinal cord perfusion, should neurological symptoms develop in the lower limbs. Until long-term data become available, all patients should have regular follow-up to look for problems such as endoleaks (see ‘Aortic aneurysms’ below) and fatigue fractures of the metal components of the stent graft with annual CT and plain radiographs, respectively.

Aortic dissection Aortic dissection is the most common non-traumatic acute aortic emergency with an overall inhospital mortality of 15–20%. Mortality is increased markedly in patients with complicated dissection.

The aetiology is frequently unknown, but is related to advancing age and hypertension. Cystic medial degeneration in connective tissue disorders such as Marfan's syndrome and Ehlers–Danlos syndrome is a predisposing factor, as are coarctation, aortitis, pregnancy and blunt chest trauma. The classic dissection is initiated by an intimal tear, which allows blood to penetrate into and split the medial layer. A cleavage plane is produced between the inner two-thirds and outer one-third of media. Arterial pressure results in extension of the dissection for a variable distance distal and, sometimes, proximal to the entry tear producing a false channel.This ‘false’ lumen is separated from the ‘true’ lumen by an intimomedial flap. Additional communications between the two lumens can be caused either by shear forces producing re-entry tears in the flap, or avulsion of the flap attachment at branch vessel origins producing natural fenestrations within the flap. Reduction in the amount of elastic tissue within the wall of the false lumen leads to subsequent aneurysmal dilatation.

CHAPTER 27

With modern high-resolution imaging, two other pathological entities have also been postulated as potential precursors of dissection: intramural haematoma and penetrating atherosclerotic ulcer15,16 (Fig. 27.17). Intramural haematoma results from hypertensive rupture of the vasa vasorum within the aortic media. The haematoma may remain localized or propagate and rupture through the intima. Intramural haematoma is distinguished from mural thrombus with cross-sectional imaging. The latter lies on top of the intima, which is frequently calcified, whereas the former is subintimal. In addition, intramural haematoma is hyperdense on unenhanced CT. In penetrating atherosclerotic ulcer, ulceration of an atheromatous plaque disrupts the internal elastic lamina, exposing the media to pulsatile arterial flow and subsequent haematoma formation in the media. This is distinguished from an atheromatous plaque by the presence of a focal, contrast medium-filled outpouching surrounded by an intramural haematoma. An atheromatous plaque does not extend beyond the intima, is frequently calcified and lacks an intramural haematoma.

• THE AORTA, INCLUDING INTERVENTION

Classification There are a number of classification systems for aortic dissection, depending on the extent of the thoracic aorta involved (Table 27.2).The Stanford system is based on whether or not surgery is required and has largely superseded the other systems. In addition to the above, all dissections are classified as acute if the duration of symptoms is shorter than 14 d and chronic if longer than this.

Mechanisms of branch vessel ischaemia Restriction of flow into the aortic branch vessels can occur from one of two mechanisms leading to end-organ ischaemia (Fig. 27.18). 1 Dynamic obstruction affects vessels arising from the true lumen. Collapse of the true lumen results in bowing of the dissection flap across the true lumen, either proximal to or at the level of the ostium of a branch vessel, restricting or occluding flow. 2 Static obstruction results from extension of the dissection into the branch vessel without a re-entry point. Increased pressure or thrombus formation in the false lumen within the branch vessel produces a focal stenosis and end-organ ischaemia. Both dynamic and static obstruction can coexist, and identification of the mechanism of ischaemia is vital as the endovascular management of each differs (see below).

Imaging

Figure 27.17 Sagittal reconstruction CT shows contrast medium tracking into the aortic wall outside the confines of the normal lumen in keeping with a penetrating atherosclerotic ulcer. The aortic wall proximal to this is also ‘thickened’ from the presence of intramural haematoma.

Table 27.2

The purpose of imaging is to establish the diagnosis, define the extent of dissection, identify the true and false lumens, identify the entry and re-entry tears and assess for the presence of complications (involvement of aortic branch vessels, rupture and aneurysm formation). SpCT is the primary imaging investigation used for the diagnosis of aortic dissection owing to its wide availability, and sensitivity and specificity approaching 100%. Image acquisition is commenced a few centimetres cephalad to the arch extending inferiorly down to the aortic bifurcation or femoral heads. Unenhanced CT may demonstrate intramural haematoma, which will appear as an area of increased attenuation. The dissection flap may be visible as a linear track of high attenuation (from intimal calcification) within the aortic lumen. Contrast-enhanced CT allows the detection of the dissection flap in 70% of cases16, seen as a band of low attenuation separating the contrast-enhanced true and false lumen.

CLASSIFICATION SYSTEMS FOR AORTIC DISSECTION Classification system

Site of dissection

Crawford

DeBakey

Stanford

Both ascending and descending aorta

Proximal dissections

Type I

Type A

Ascending aorta and arch only

Proximal dissections

Type II

Type A

Descending aorta only (distal to left subclavian artery)

Distal dissections

Type III IIIa—limited to thoracic aorta IIIb—extends to abdominal aorta

Type B

557

558

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.18 Diagram of branch vessel ischaemia. (A) Dynamic obstruction. The intimomedial flap bows across the true lumen (arrow) and obstructs flow into the branch vessel. (B) Static obstruction. The dissection extends into the branch vessel and may thrombose (arrowheads), causing stenosis of the vessel origin. F = false lumen, T = true lumen.

Injection of contrast medium via the left upper limb should be avoided as the very high attenuation from contrast medium within the left brachiocephalic vein can produce streak artefact across the aortic arch, potentially causing diagnostic difficulty (Fig. 27.19). In the era of endovascular treatment the differentiation of the false lumen from the true lumen is essential; a number of imaging findings can be helpful (Fig. 27.20). A search is also made for potential ischaemic complications. The dissection is followed caudally to determine if it intersects a branch vessel (static obstruction) and whether adequate true lumen calibre is maintained from the heart to the branch vessels (dynamic obstruction) (Fig. 27.21). A dissection with thrombosed false lumen may be confused with an aneurysm with calcified mural thrombus. High attenuation within the false lumen on unenhanced images can help to identify the former. In addition, a dissection tends to spiral as it passes along the aorta whereas thrombus maintains a constant relationship with the aortic wall. Mural thrombus arising in an aneurysm also tends to have an irregular internal border whilst a dissection has a smooth internal border. Finally,

calcification of the intima may be identified at the periphery of the thrombus in an aneurysm. The use of MRI in the evaluation of aortic dissection has, until recently, been largely confined to patients undergoing long-term follow-up. High-speed pulse sequences are mandatory for vascular imaging, e.g. true fast imaging with steadystate precession (FISP) has allowed more realistic imaging of acute aortic dissection. A complete study of the thoracic aorta can be performed within a few minutes and achieves high sensitivity and specificity in the diagnosis of aortic dissection and other aortic diseases17. TOE has the advantage of providing real-time imaging. It is able to localize the site of intimal tears and provide haemodynamic information about flow in the true and false lumen. In the presence of a type A dissection, it provides valuable information about the structural and functional status of the aortic valve and can assess the degree of involvement of the coronary arteries (Fig. 27.22). It permits unobstructed views of the descending aorta in type B dissection. TOE is, however, limited by poor visualization of the distal ascending aorta and parts of the aortic arch due to the interposition of the trachea and right main bronchus between the aorta and oesophagus. In addition, the abdominal aorta and its branch vessels cannot be assessed. Intravascular ultrasound (IVUS) at 12.5 MHz provides intraluminal cross-sectional images of vessels and has been useful in both the diagnosis and intervention of aortic dissection. IVUS is able to demonstrate the entry tear and extent of dissection, but is particularly useful in differentiating the true and false lumen, and demonstration of dynamic obstruction18,19. However, the transducers are single use only and expensive, and most departments have limited experience of IVUS as a diagnostic tool.

Management The initial goals in all patients presenting with aortic dissection are to eliminate pain, reduce systolic blood pressure (below 100–200 mmHg) and reduce the rate of rise of left ventricular

Figure 27.19 (A) CT demonstrates a dissection flap in the proximal descending thoracic aorta. Contrast has been given via the left subclavian artery causing streak artefact across the aorta. (B) Sagittal reconstruction shows the dissection arising immediately distal to the left subclavian artery origin (type B), though the streak artefact gives the impression that there is retrograde extension proximal to the left subclavian artery (type A).

CHAPTER 27

T

*

*

F

• THE AORTA, INCLUDING INTERVENTION

F

* F T

T

* T F *

S

C

R

T F

T F

T F

* T F

Figure 27.20 Differentiating the true and false lumen on CT. The false lumen usually tracks around the convexity of the aortic arch (A) and is more often than not the larger of the two lumens (B–E). The outer wall of the false lumen produces an acute angle at its junction with the dissection flap (asterix) (A–C,H). Occasionally, linear strands of low attenuation may be seen within the false lumen (cobweb sign). These represent residual strands of media incompletely sheared away at the time of dissection (arrowheads) (B,D). Finally, the true lumen can be correctly identified by its continuity with the nondissected aorta. Intimal calcification can be seen along the dissection flap (arrows) (C,F). C = coeliac axis, F = false lumen, R = right renal artery, S = superior mesenteric artery, T = true lumen.

systolic pressure. These are achieved with a combination of intravenous beta-blockers and peripheral vasodilators. Subsequent management is based on the Stanford type and presence of complications.

C

Type B dissection The current treatment of acute type B dissections is based on a complication-specific approach. In uncomplicated type B dissection (no evidence of rupture or branch vessel ischaemia) medical treatment is initially implemented, as both medical and emergent surgical management are associated with similar mortality rates15. Patients who fail medical management (persistent pain and/or progression of dissection) or develop complications are referred for either surgical or endovascular intervention. Early surgery is recommended for patients with Marfan's syndrome.

S R

Type A dissection These dissections account for 75% of cases. Emergent surgical repair is indicated in all patients owing to the high mortality (> 50% within 48 h) if untreated20. Fatal complications include aortic rupture, cardiac tamponade, acute aortic regurgitation and acute myocardial infarction (Fig. 27.23). Involvement of the arch vessels results in morbidity from neurological complications.

L

Figure 27.21 Acute type B dissection in a patient presenting with mesenteric ischaemia. There is true lumen collapse in the descending thoracic aorta (A); the true lumen is seen as an opacified crescent anteriorly (arrows). The coeliac axis (B) and superior mesenteric/right renal arteries (C) arise from the true lumen and are thus compromised. The left renal artery arising from the false lumen (D) allows normal perfusion to the left kidney. C = coeliac axis, L = left renal artery, R = right renal artery, S = superior mesenteric artery.

Endovascular treatment of type B dissection Surgery for type B dissection is associated with mortality in excess of 50% in cases with end-organ ischaemia19. Endovascular techniques have been successfully employed in the treatment of surgical type B dissection with reduced morbidity and mortality21,22. The three techniques employed are stent insertion, stent graft insertion and/or fenestration of the intimal flap. The indications for stent or stent graft placement in type B dissection are two-fold: 1 Contained rupture. Persistent flow in the false lumen is associated with aneurysmal dilatation and increased risk of

559

560

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.22 Transoesophageal echocardiogram of a type A dissection shows (A) the dissection distal to the aortic valve (arrow) with (B) prolapse of the dissection flap (arrow) through the aortic valve into the left ventricular outflow tract (LVOT). AO = aortic valve, LA = left atrium, RVOT = right ventricular outflow tract. (Courtesy of Dr K. E. Berkin, The Leeds Teaching Hospitals NHS Trust, UK)

rupture of the false lumen (20–50% of patients within 1–5 years19). Placement of a stent graft across the site of the entry tear can promote thrombosis of the false lumen and stabilization of the dissection, reducing the risk of rupture.Acute type B dissections are the most appropriate group to treat as the dissection flap is thin and mobile, and will readily re-appose to the aortic wall on exclusion of the false lumen. In chronic dissections, the flap becomes thickened and rigid and, thus, is less likely to result in complete false lumen exclusion. 2 Branch vessel ischaemia. The treatment of branch vessel ischaemia is dependent on the cause. Dynamic obstruction results from true lumen collapse. Sealing the entry tear with stent graft placement directs blood flow back into the true lumen, increasing the size of the true lumen and moving the dissection flap away from the branch vessel, thus

relieving branch vessel ischaemia. Static obstruction can be successfully treated by direct stent insertion in the compromised vessel via the true lumen. In cases where the dissection extends distally to cause lower limb ischaemia, direct access to the true lumen is gained by puncturing the side with the weak or absent femoral pulse. Currently, stent grafts are used largely in patients who have indications for surgical intervention. Given that the majority of patients undergo thrombosis of the false lumen following stent graft placement, it is not inconceivable that, in the future, this may become the treatment of choice for all patients with acute type B dissection. However, more follow-up data are required on the long-term durability of these devices before this becomes a reality.

Figure 27.23 CT of a type A dissection. (A) The dissection flap is seen in the ascending aorta with partial thrombosis of the false lumen. (B) The left ventricular outflow tract is compressed by the acutely thrombosed false lumen which is seen as increased attenuation of the false lumen. There has also been rupture into the pericardium.

CHAPTER 27

Since the introduction of stent grafts, percutaneous fenestration has been less frequently employed in the management of branch vessel ischaemia due to true lumen compression. It may be used in conjunction with, or as an alternative to, entry tear stent graft placement when a pressure gradient exists between the true and false lumen. Access is gained into both lumen and a stiff guidewire passed across the flap, from the true lumen into the false lumen, using either a balloon or IVUS catheter as the target. A large balloon is then inflated across the flap to produce the fenestration. If true lumen compression persists a stent can be placed across the fenestration to maintain flow through it. Patients who are managed conservatively require long-term follow-up with either CT or MRI to monitor the false lumen for aneurysmal dilatation and distal extension of the dissection over time. Any patient who undergoes endovascular treatment similarly requires long-term follow-up to assess for integrity of the device.

Inflammatory diseases of the aorta and mid aortic syndrome There are many causes of acquired inflammatory aortitis, including ergotism, radiation fibrosis, syphilis, tuberculosis, giant cell arteritis, Buerger’s, Behçet’s, Cogan’s and Kawasaki diseases. There are also congenital inflammatory abnormalities that affect the aorta, such as Ehlers–Danlos and Marfan's syndromes, as well as neurofibromatosis. Inflammatory aneurysmal aortic disease is described below under aortic aneurysms.

Mid aortic syndrome Mid aortic syndrome is characterized by segmental narrowing of the proximal abdominal aorta and ostial stenosis of its major branches (Fig. 27.24). It is usually diagnosed in young adults, but may present in childhood. There are a number of causes and the clinical presentation and radiology depend on the cause, but hypertension is a feature in all cases.

Figure 27.24 Longitudinal abdominal ultrasound examination in a 12 year old girl presenting with hypertension. There is clearly a stenosis at the origin of the superior mesenteric artery and narrowing and irregularity of the aorta. This is characteristic of mid aortic syndrome. The patient had café au lait spots consistent with the diagnosis of neurofibromatosis.

• THE AORTA, INCLUDING INTERVENTION

Congenital aortic coarctation is a very uncommon cause of mid aortic syndrome, where the aortic narrowing occurs in the thoracic or abdominal aorta. It may be seen in fetal alcohol syndrome when it is associated with intellectual disability. Granulomatous vasculitis (Takayasu's disease) is a chronic inflammatory disease that involves the aorta, its branches and the pulmonary arteries, causing varying degree of stenosis, occlusion, or dilatation of the involved vessels.The aetiology and precise pathogenesis are unknown. It is more common in parts of the world where there is a high incidence of tuberculosis, but exceptions like Japan need to be explained. It is predominantly a disease of young adults, but it can affect children. It is very rare in infancy. The female-to-male ratio has varied from 9:1 in reports from Japan to 1.3:1 in India. The pattern of vessel involvement also varies in different parts of the world. The involvement of the aortic arch and its branches is common in Japan, whereas the thoraco-abdominal aorta is mainly involved in patients from Korea and India. It is not known whether this variation reflects differing causes of the disease or differing HLA associated genetic subtypes. Racial variation also occurs, the disease being uncommon in caucasians and affecting Sephardic Jews but not Ashkenazi Jews23,24. The initial site of inflammation is around the vasa vasorum in the media and adventitia but later nodular fibrosis in all layers of the artery is seen and the intima can obliterate the lumen (Fig. 27.25). The diagnosis depends on typical angiographic morphology, a history or presence of constitutional symptoms suggestive of a systemic illness and the differential diagnosis of other, similar conditions as listed above. Atherosclerosis of the aorta is distinguished on clinical and morphological grounds, but secondary atherosclerotic changes may occur in older patients with Takayasu’s arteritis. The angiographic features occur late in the course of the disease and include luminal irregularity, vessel stenosis, occlusion, dilatation, or aneurysms in the aorta or its primary branches (Fig. 27.26). Neurofibromatosis of the abdominal aorta and some other causes of mid aortic syndrome may produce an identical angiographic picture in children. Based

Figure 27.25 MRI of a 20 year old woman presenting with left arm ischaemia and hypertension. Thickening of all layers of the descending thoracic aortic wall was also seen in the left subclavian origin. The patient had an elevated ESR and a diagnosis of Takayasu's disease was made.

561

562

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.26 A 47 year old man presenting with acute pulmonary oedema and uncontrolled hypertension. A contrast-enhanced MRA confirms mid aortic syndrome with focal occlusion of the juxtarenal aorta. The distal aorta is perfused by the mesenteric arteries. There is a grossly hypertrophied inferior mesenteric artery together with hypertrophied marginal artery and arc of Riolan. The source images (not shown) demonstrated a high grade right renal artery stenosis but normal left renal and superior mesenteric arteries. Axial FISP images (not shown) confirmed a thickened aortic wall at the level of the occlusion consistent with Takayasu's disease.

on angiographic morphology, Takayasu’s arteritis is divided into type I (involving the aortic arch and its branches), type II (thoraco-abdominal aorta and its branches) and type III (involving lesions of both type I and II). Involvement of pulmonary arteries, in addition to any of the above types, is grouped as type IV. The infrarenal aorta and the iliac vessels are not usually involved in Takayasu's arteritis. Similarly, the inferior mesenteric artery is rarely involved. Unlike coarctation of the aorta, intercostal collaterals rarely occur as the diffuse intimal disease in the aorta also involves the ostia of these intercostal vessels. Aortic intimal calcification may be seen. Saccular or fusiform aneurysms of the aorta occur in 2–26% of cases and usually coexist with stenotic lesions. Aneurysms without stenosis occur in 1–2% of cases. Pseudoaneurysm or dissection of the aorta are extremely rare. CT, ultrasound and particularly contrast-enhanced MRI and MRA have been used to provide information on mural changes of the vessels. These noninvasive modalities may replace angiography for the diagnosis and for monitoring therapy and progression of the disease. Takayasu’s arteritis in children has a mortality of 10–30% at follow-up. The prognosis has significantly improved due to interventional procedures for the treatment of renal and aortic stenosis. Long-term follow-up data on children are not available. Five and 10 year survival in adults is 91% and 84%, respectively. The presence of severe hypertension, aortic regurgitation, retinopathy, aneurysms, or cardiac involvement predicts a poorer outcome. A relatively stable course is anticipated in the absence of these severe complications. Eighty

per cent of chronic, burnt out lesions of Takayasu’s arteritis remain stable for years but 20% can progress. In the acute phase, treatment with corticosteroids leads to clinical remission in 60% of cases. Cytotoxic drugs can also be used in resistant cases. The major morbidity and mortality of Takayasu’s arteritis results from stenosis and occlusion of the aorta, and renal and carotid arteries. PTA of stenosed segments has revolutionized the treatment of Takayasu’s arteritis. Renal artery angioplasty is successful in up to 90% of cases and blood pressure control is achieved in 60%. Restenosis may occur in 20–25% of cases. Renal artery stents are not usually required. PTA is preferably performed in the chronic phase of disease, but successful dilatation may be done during the acute phase of Takayasu’s arteritis. Similarly, PTA of aortic stenosis is highly effective, even in diffuse, long segment stenoses. Success rates approximating 90% have been reported for aortic angioplasty in children. Restenosis may occur in 14–20% at follow-up. Stents should not be used in children. Surgical treatment is not preferred for Takayasu’s arteritis because of the diffuse, inflammatory, and possibly progressive nature of the disease, except for undilatable, symptomatic, stenotic lesions and for large aneurysms23,24. von Recklinghausen's disease (type 1 neurofibromatosis) is a genetic disorder associated with chromosome 17, distinguished from other causes of mid aortic syndrome by the presence of café au lait skin lesions and neurofibromas. Presentation is from birth to 15 years (median age 5 years). Approximately 2% of patients develop vascular abnormalities including renal, aortic and mesenteric stenoses (Fig. 27.27). Vessels are surrounded by neurofibromatous or ganglioneuromatous tissue in the adventitia. Alagille syndrome is a multisystem autosomal dominant disorder caused by mutations in the JAG1 gene on chromosome 20p122,4. It often presents with clinical symptoms involving the liver during infancy and early childhood. Alagille syndrome is the most common form of the inherited disorders that cause cholestasis. Patients have distinctive facial features with deep set eyes, frontal bossing, bulbous tip of the nose, a down-turned mouth and a small mandible with pointed chin. Mid aortic coarctation is one of several heart, bone and ocular defects that can also occur. Williams' syndrome is a rare genetic condition estimated to occur in 1 in 20 000 births. Children have characteristic facial features and metabolic problems, particularly hypercalcaemia. It is a multisystem syndrome and hypertension is common, usually caused by mid abdominal or thoracic coarctation.

Aortic occlusive disease Atherosclerosis is the predominant cause of chronic aortic occlusive disease (over 90% of cases), with Takayasu's disease (see above) accounting for the rest. Acute occlusion is the result of aortic bifurcation ‘saddle’ embolus or in situ aortic thrombosis.

Chronic aortic occlusive disease Atherosclerotic aortic occlusive disease affects a younger population than those generally presenting with lower limb

CHAPTER 27

• THE AORTA, INCLUDING INTERVENTION

A number of key points need to be addressed, irrespective of the method of investigation, as these will have a bearing on the surgical and endovascular options considered: 1 What is the upper limit of the lesion? Is it infrarenal or juxtarenal? 2 What is the lower limit of the lesion? Is there involvement of the aortic bifurcation? 3 Are the coeliac axis (CA) and superior mesenteric artery (SMA) normal? This is important if intervention has the potential to compromise the inferior mesenteric artery (IMA).

Figure 27.27 Right oblique abdominal aortogram in a hypertensive patient with the cutaneous stigmata of neurofibromatosis. Note the mid aortic stenosis and the coeliac axis and superior mesenteric artery stenoses with collateral supply from the inferior mesenteric artery.

arterial disease. Patients are typically female, heavy smokers with hyperlipidaemia and have a small infrarenal aorta and hypoplastic iliofemoral arteries (hypoplastic aorto-iliac syndrome). The infra-inguinal arteries are ‘protected’ by the aortic lesion and are characteristically disease free. Patients present with chronic lower limb ischaemia and are graded according to the severity of their disease. Aortic occlusive disease is largely associated with Fontaine grade I or II symptoms (see Ch. 28). Symptoms of critical limb ischaemia (grade III or IV) are unusual at initial presentation as these are associated with both supra- and infra-inguinal disease. As a general rule, the management of grade I and IIa patients is based solely on risk-factor modification. Those with grades IIb–IV are investigated further and, in addition to risk factor modification, are offered revascularization, if appropriate. Investigation and management In the presence of significant aortic disease, the femoral pulses are diminished or absent and there is a reduction in the ABPI (grade IIb 0.5–0.8, grades III–IV < 0.5). Duplex data acquisition plays very little role in the investigation of these patients as the aorta is often difficult to visualize and assess. Angiography is currently the investigation of choice, though MRA offers major advantages. It is noninvasive, avoiding a brachial puncture and its attendant risks. It provides excellent images of not only the aortic lesion but also the run-off and negates the need for large volumes of iodinated contrast medium. It therefore allows planning of an appropriate management strategy with almost no risk to the patient.

In the past, aortic occlusive disease was treated by aortic bypass surgery or aortic endarterectomy. Though surgery is associated with excellent primary patency (75–90% and 90–95%, respectively), there is also an appreciable morbidity (9–27%) and mortality (1–7%)25. Endovascular techniques have been used as an alternative since the early 1990s and are now the treatment of choice. The options lie between angioplasty alone, angioplasty with selective stenting, or primary stenting. Angioplasty alone is used in short focal stenoses (< 2 cm) and is associated with a primary patency of 85% and low incidence of major complications (3.6%). Stenting has, by and large, been reserved for flow-limiting dissection or residual stenosis following angioplasty. Primary stenting has been advocated for the treatment of occlusions and complex lesions (eccentric, ulcerated, or calcified plaques) where there is significant concern of distal embolization (Fig. 27.28).

Acute aortic occlusive disease Acute aortic occlusive disease is a vascular emergency resulting from either saddle embolus to the aortic bifurcation or in situ thrombosis of an aortic stenosis, aortic aneurysm or traumatic aortic dissection. Given the proximal nature of the occlusion, patients often present with neurological deficit (including paralysis) in the lower limbs. This can lead to investigation for spinal cord compression and a consequent delay in diagnosis. If the vascular nature of the presenting symptoms is unrecognized, mortality is high (75%)26.The key to the diagnosis is the absence of femoral pulses. Imaging and management The role of imaging is dependent on the severity of ischaemia. Where there is major tissue loss, absent capillary return with marbling, profound paralysis and sensory loss, or absent Doppler signals, the ischaemia is irreversible and amputation inevitable. When there is a degree of any of the above emergency surgery with no imaging may be appropriate. With less severe degrees of ischaemia, expedient imaging can provide useful information. As with chronic occlusion, the choice lies between catheter angiography and MRA, with similar risks and benefits. The purpose of investigation is to confirm the proximal extent of occlusion and the state of the run-off, and to differentiate embolic from thrombotic occlusion. If an embolic aetiology or thrombosis of an aneurysm/ traumatic dissection is confirmed, surgical intervention with embolectomy or bypass grafting, as appropriate, is the treatment

563

564

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.28 Chronic aortic occlusion in a 56 year old woman presenting with short distance claudication. (A) There is a short segment occlusion of the infrarenal aorta. The aorta proximal to this is also diseased and narrowed. The lumbar arteries are markedly hypertrophied. (B) A plaque is seen at the aortic bifurcation, but the iliac arteries are relatively disease free. (C) Following primary stenting of the occlusion the patient's symptoms subsided, despite underdilatation of the lesion. (D) The extreme calcific nature of the lesion is seen on the native image.

CHAPTER 27

of choice. In situ thrombosis of a pre-existing stenosis is suggested by the presence of collaterals. Though surgical bypass is an option, endovascular treatment with thrombolysis can also be considered if the severity of ischaemia allows time for this. Following successful thrombolysis, aortic angioplasty or stenting can be performed to treat the underlying lesion (Fig. 27.29). Even with appropriate intervention, overall inhospital mortality is high (21%), being higher in those with an embolic aetiology26. If the occlusion is embolic, following successful revascularization a search must be made for a source. Echocardiography will exclude a cardiac source, following which the thoracic aorta should be carefully examined. Emboli often originate from a thrombus forming on an aortic plaque or intimal flap.

Aortic aneurysms Aortic aneurysm is a relatively common condition, characterized by degeneration and remodelling of the aortic wall. Aneurysm development is multifactorial in nature, with both a genetic predisposition and environmental factors acting together to initiate a cascade of arterial degeneration. They are commonly atherosclerotic, but can also result from trauma, infection including tuberculosis and syphilis and syndromes such as Marfan's and Ehlers–Danlos when they most commonly affect the aortic root, ascending aorta and arch27. The relevance of dividing aneurysms of the aorta into true or false is simply that false aneurysms result from a contained rupture and the term stable should not be applied.

Atherosclerotic aortic aneurysms Ninety-five per cent of atherosclerotic aneurysms affect the abdominal rather than the thoracic aorta.

• THE AORTA, INCLUDING INTERVENTION

The natural history of such aneurysmal disease is progressive remodelling, expansion and eventual rupture. As only 14% of patients have symptoms, ruptured aneurysm is a major cause of death in western populations. Patients always have major co-morbidities such as coronary disease, peripheral vascular disease, obstructive pulmonary disease, diabetes and renal failure. In patients who initially survive a ruptured aneurysm, more than 50% die at or following surgery. Histologically, an inflammatory cell infiltrate has been demonstrated in all atherosclerotic aortic aneurysms. Matrix metalloproteinases (MMPs) play an important role in the remodelling process by degrading extracellular matrix proteins such as elastin and collagen, both of which are needed to maintain the structural integrity and mechanical properties of the aortic wall. In healthy tissue, MMP activity is tightly regulated by tissue inhibitors. Several MMPs have been identified, MMP-2 being most strongly associated with small aneurysms, while MMP-9 has been linked to medium-sized, large, or ruptured aneurysms. Furthermore, plasma concentrations of MMP-9 appear not only to be associated with aneurysmal disease, but also with size and expansion rate of abdominal aortic aneurysms (AAAs), plasma levels falling after successful exclusion of the AAA28. The asymptomatic thoracic aortic aneurysm (TAA) is often detected as a soft tissue mediastinal mass which may or may not be calcified on a chest radiograph taken for some other reason, though it can be confused with tumour (Fig. 27.30). Likewise, AAAs can be detected incidentally on abdominal or lumbar spine radiographs because of their mass effect and calcification (Fig. 27.31). However, in the abdomen, ultrasound is the imaging method of choice and most centres now routinely

Figure 27.29 Acute aortic occlusion. (A) Infrarenal aortic occlusion with meniscus sign suggests that this is an embolic occlusion. (B) However, following successful thrombolysis, an underlying stenosis is revealed which appears secondary to extrinsic compression of the aorta. (C) A good angiographic result is achieved post stent placement.

565

566

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 27.30 Control image from intravenous urogram requested because of left renal colic. Curvilinear calcification (arrowheads) is consistent with a significant size calcified abdominal aortic aneurysm.

Figure 27.31 Lateral aortogram in a patient with severe mid back pain and lumbar spine images which demonstrated anterior erosion of the lower thoracic vertebral bodies. The angiogram demonstrates that this has been caused by a pulsatile thoraco-abdominal aortic aneurysm.

check for abdominal aneurysmal disease in patients presenting for other reasons. In the past, clinical examination and an ultrasound were all that were required before elective surgery. The introduction of endovascular stent grafting has made contrast-enhanced CT mandatory now, though spin-echo MRI can provide the same information with no radiation dose or contrast. The relatively recent introduction of MDCT has made the diagnosis and evaluation of the extent of an aneurysm even easier. Individual slices can be around 1-mm thick. Multiplanar reconstructions of AAAs allow full appreciation of the lumen, the wall, the extent of any inflammatory material, and the position and degree of distension of the all important adjacent ureters. It is generally, though not universally, accepted that an AAA below 5.5 cm which is not enlarging rapidly and is asymptomatic requires nothing more than regular follow-up ultrasound29. However, an aneurysm that does not conform to this description requires treatment if rupture is to be prevented and the benefit of surgery is to be maximized. This size criterion does not apply in the thorax, though rapid expansion or symptoms do. The purpose of CT for elective aneurysms is to evaluate the potential of endovascular treatment30. For TAAs the main features that should be assessed are: • Size and morphology of the aneurysm. • Its position superiorly in relation to the great vessels of the arch that might contraindicate primary stent grafting, though it should be noted that the left subclavian artery can be covered provided that both vertebral arteries are patent. • Its diameter at any potential stent graft ‘drop zone’ superiorly. • Amount of atheromatous material at this site which might affect the proximal stent seal. • Distal extent of the aneurysm and its relationship to the visceral arteries which might require bypass or stenting if they have to be covered. • Any evidence that there might be a large radicular artery supplying the spinal cord that could be covered by a stent graft. The incidence of paraplegia is much lower with thoracic stent grafts than with surgery but is still 2% and can be devastating. Anticipation of such a complication can prevent it by CSF drainage after the procedure. • Size and condition of the abdominal aorta and iliac arteries which might prevent passage of the stent graft delivery system. • Any other incidental findings in the chest, abdomen, or pelvis. Endovascular treatment of AAAs is more complex technically as most AAA stent grafts are bifurcated, though aorto uni-iliac stent grafts with surgical femorofemoral cross-over grafting are also common, especially where the aneurysm has ruptured. Tube grafts are now the exception rather than the rule31. CT evaluation should aim to determine the following (Fig. 27.32): • AP and transverse size of the AAA. • Diameter of the aorta at the levels of and just below the visceral arteries.

CHAPTER 27

• THE AORTA, INCLUDING INTERVENTION

Figure 27.32 (A,C) Sagittal and (B,D) coronal reconstructions of an axially acquired contrast-enhanced CT with measurements. Such reconstructions and measurements are vital to the successful use of stent grafts to treat significant sized abdominal and thoracic aortic aneurysms.

• Length of the aneurysm neck from the lowest renal artery to the origin of the aneurysm; this needs to be at least 15 mm. • Shape of the neck; conical necks may lead to poor proximal seals or late endoleaks. • Angulation of the neck in the AP and lateral planes; greater than 60 degrees and the seal may be difficult or the stent graft may dislodge. • Presence of excessive atheroma in the neck which may prevent a perfect seal.

• Any accessory renal arteries that may have to be covered. If this is the case an assessment of split renal function may be necessary. • Distance from the lowest renal artery to the aortic bifurcation. This will affect the length of the stent graft body. • Tortuosity and degree of calcification of the iliac arteries. • Diameter of the common iliac arties. If these are aneurysmal it may be necessary to counsel the patient regarding possible buttock claudication following internal iliac artery embolization and extension of the stent graft into the external iliac artery.

567

568

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

• Size of the common femoral and external iliac arteries and the presence of any stenoses that might prevent the stent graft being delivered. • Any incidental findings, especially concerning the size and function of the kidneys and the presence of a double inferior vena cava, which can be disastrous if not recognized at open surgery. Endovascular stent grafting is still a new technique and complications can occur, particularly endoleaks (Fig. 27.33), but the evidence from the two major randomized trials suggests that it will remain a valuable tool in the treatment of AAA and TAA in the future31,32.

Aortic sinus aneurysms These aneurysms can be congenital, particularly in Asian populations, but can also be seen secondary to infective endocarditis and in Marfan's syndrome and ankylosing spondylitis. The most common site for these is from the right aortic sinus into the right ventricle or right atrium, but also from the noncoronary sinus into the left atrium. They may rupture into the heart and present subacutely with a left-to-right shunt and a continuous murmur. When dilatation is confined to the aortic root, it will not be seen on the PA chest radiograph, though it may be revealed on the lateral radiograph. When the ascending aorta is involved, the right mediastinal border will be prominent and, on the lateral radiograph, the aorta will obliterate the retrosternal space above the heart. Left ventricular dilatation results from aortic regurgitation. The aortic root and ascending aorta are very well shown by MRI, particularly by T1-weighted spinecho imaging, and cine MRI will show aortic regurgitation as well as fistulas. TOE is also very useful and can be carried out at the bedside. There may be very good reasons for avoiding angiography in these patients, particularly in the presence of infection.

Inflammatory aneurysms Inflammatory abdominal aortic aneurysms (IAAAs) are defined as dilation of the aorta with a thickened aneurysm wall, marked peri-aneurysmal and retroperitoneal fibrosis, and dense adhesions to adjacent abdominal organs. IAAAs represent 3–10% of all abdominal aortic aneurysms and are seen more commonly in men. Mean age of occurrence ranges from 62 to 68 years (5–10 years younger than patients with other atherosclerotic aneurysms). More patients with IAAAs have a positive family history of aneurysms (17%) when compared with patients with noninflammatory aneurysms (1.7%). IAAAs are more often symptomatic compared to the more common noninflammatory aneurysm. In addition to abdominal or back pain these patients also present with weight loss and elevated erythrocyte sedimentation rate (ESR)33. The aetiology of IAAAs is the same as for other atherosclerotic AAAs, but the inflammatory component is more pronounced. Current thinking favours one pathological process with varying degrees of inflammation rather than a distinct clinical entity. The importance of the IAAA lies in its potential treatment. Its true natural history is unknown but the risk of rupture remains. Steroid therapy has been used to control the inflammatory process but no controlled studies exist. Surgery is technically difficult as the ureters can be involved in the inflammatory process and may need to be stented to protect them.The duodenum and left renal vein are often adherent to the aneurysm sac. IAAA repair halts the progression of retroperitoneal fibrosis but does not cure it. Complete regression of the retroperitoneal fibrosis is seen in 23–53% of cases after surgery. Partial regression or no change occurs in the remainder. In view of the technical difficulty of open IAAA repair and the increased morbidity and mortality, endovascular aneurysm repair is an attractive option. However, the longer-term regression of the peri-aneurysmal fibrosis seen in up to 53% of open repairs is said to occur infrequently after endovascular repair, though the number of cases reported is small. Though ultrasound is very important, CT has become the mainstay of assessing IAAAs. In peri-aortitis, CT shows a thick

Figure 27.33 Four types of endoleak are identified which can complicate the exclusion of an aneurysm by stent graft. Type I is the most dangerous and is due to failure to obtain a seal either proximally or distally. This should not be left and is usually solved with a moulding balloon and/or a cuff. Type II is caused by retrograde flow into the sac from lumbar or inferior mesenteric arteries. Mostly these resolve themselves but if they do not and the sac continues to enlarge they have to be embolized retrogradely. Type III is due to a failure of the stent graft material. This can be due to a manufacturing problem or long-term wear and tear and changes in aneurysm morphology causing holes in the graft or dislocations of modular components. They can be fixed either by repeat stent grafting or open surgery. Type IV is unimportant and rarely occurs now. It was due to graft porosity. The term endotension (Type V) is applied where there is sac expansion in the absence of a recognizable Type I to IV endoleak.

CHAPTER 27

cuff of enhancing soft tissue around the aorta (Fig. 27.34). It can usually be differentiated from lymphoma around an aneurysm, though this can be mimicked by tumour infiltration. Liposarcoma and bladder cancer can cause a strong inflammatory fibrous reaction, which can cause problems of interpretation. Haemorrhage is another potential diagnostic pitfall. When an aneurysm has ruptured, the tissue planes within the retroperitoneum become poorly defined, which can make the identification of an inflammatory component difficult. However, fresh blood has a higher CT attenuation than muscle and usually tracks into the parare-

• THE AORTA, INCLUDING INTERVENTION

nal fat well away from the aneurysm, except at the very point of rupture. CT and ultrasound have nearly replaced intravenous urography, which, for the clinical indication of inflammatory aneurysm, is of historical interest only. MRI is best reserved for specific unanswered questions or when the patient is in renal failure. Spin-echo T1- and T2-weighted sequences provide a good overall assessment of an inflammatory aneurysm and are ideal for follow-up when trying to minimize radiation dose.

Mycotic aneurysms Infection can cause thrombosis of the vasa vasorum with consequent destruction of the aortic intima and media. Commonly, such infection is due to emboli from infective endocarditis, septicaemia, or local spread (Fig. 27.35). Imaging of mycotic aneurysms is no different from that of other aneurysms. The results of surgical treatment can be poor and often the aorta has to be tied off and axillo-bifemoral grafting performed. More recently stent grafting has been tried with mixed results.

REFERENCES

Figure 27.34 Contrast-enhanced CT demonstrates a significant sized abdominal aortic aneurysm with a thick well-defined outer wall. There is hydroureter on the left and a hydroureter with a nonfunctioning kidney on the right. These features are typical of inflammatory abdominal aortic aneurysm.

Figure 27.35 Aortic angiogram in a patient who presented with emboli to both feet 6 weeks after an episode of acalculous cholecystitis. Aortography demonstrated an eccentric lower abdominal aortic aneurysm. This was later proved to be mycotic secondary to salmonella infection. The aorta was surgically tied off and the patient underwent axillo-bifemoral grafting.

1. Standring S 2005 The anatomical basis of clinical practice. In: Gray's anatomy, 39th edn. Elsevier/Churchill Livingstone, Edinburgh, p141 2. Aronberg D J, Glazer H S, Madsen K et al 1984 Normal thoracic aortic diameters by computed tomography. J Comput Assist Tomogr 8: 247–250 3. Johnson D 2005 The anatomical basis of clinical practice. In: Gray's anatomy, 39th edn. Elsevier/Churchill Livingstone, Edinburgh, p1021–1023 4. Stewart J R, Kincaid O W, Edwards J E 1964 An atlas of vascular rings and related malformations of the aortic arch system. Charles C. Thomas, Springfield, Illinois 5. Jaffe R B 1991 Radiographic manifestations of congenital anomalies of the aortic arch. Radiol Clin North Am 29: 319–334 6. Campbell M 1970 Natural history of coarctation of the aorta. Br Heart J 32: 633–640 7. Paddon A J 2000 Endovascular repair of thoracic and dissecting aneurysms and aortic coarctation. In: Dyet J F, Ettles D F, Nicholson A A, Wilson S E (eds) Textbook of endovascular procedures. Churchill Livingstone, Philadelphia 8. Paddon A J, Nicholson A A, Ettles D F et al 2000 Long term follow up of percutaneous balloon angioplasty in adult aortic coarctation. Cardiovasc Intervent Radiol 23: 364–367 9. Parmley L F, Mattingly T W, Manion W C et al 1958 Non-penetrating traumatic injury of the aorta. Circulation 17: 1086–1101 10. Wintermark M, Wicky S, Schnyder P 2002 Imaging of acute traumatic injuries of the thoracic aorta. Eur Radiol 12: 431–442 11. Curry J D, Recine C A, Snavely E et al 2002 Periaortic haematoma on abdominal computed tomography scanning as an indicator of thoracic aortic rupture in blunt trauma. J Trauma 52: 699–702 12. Degiannis E, Boffard K 2002 Critical decisions in trauma of the thoracic aorta. Injury 33: 317–322 13. Iannelli G, Piscione F, Di Tommaso L et al 2004 Thoracic aortic emergencies: impact of endovascular surgery. Ann Thorac Surg 77: 591–596 14. Dake M D, Miller D C, Semba C P et al 1994 Transluminal placement of endovascular stent-grafts for the treatment of descending thoracic aneurysms. N Engl J Med 331: 1729–1734 15. Prendergast B D, Boon N A, Buckenham T 2002 Aortic dissection: advances in imaging and endoluminal repair. Cardiovasc Intervent Radiol 25: 85–97 16. Castañer E, Andreu M, Gallardo X et al 2003 CT in nontraumatic acute thoracic aortic disease: typical and atypical features and complications. RadioGraphics 23 (suppl): S93–S110

569

570

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

17. Pereles F S, McCarthy R M, Baskaran V et al 2002 Thoracic aortic dissection and aneurysm: evaluation with nonenhanced true FISP MR angiography in less than 4 minutes. Radiology 223: 270–274 18. Lee D Y, Williams D M, Abrams G D 1997 The dissected aorta. Part II. Differentiation of the true from the false lumen with intravascular US. Radiology 203: 32–36 19. Chavan A, Lotz J, Oelert F et al 2003 Endoluminal treatment of aortic dissection. Eur Radiol 13: 2521–2534 20. Karmy-Jones R, Aldea G, Boyle E M 2000 The continuing evolution in the management of thoracic aortic dissection. Chest 117: 1221–1223 21. Dake M D, Kato N, Mitchell R S et al 1999 Endovascular stent-graft placement for the treatment of acute aortic dissection. N Engl J Med 340: 1546–1552 22. Nienaber C A, Fattori R, Lund G et al 1999 Nonsurgical reconstruction of thoracic aortic dissection by stent-graft placement. N Engl J Med 340: 1539–1545 23. Matsunaga N, Hayashi K, Sakamoto I et al 1997 Takayasu arteritis: protean radiologic manifestations and diagnosis. RadioGraphics 17: 579–594 24. Yamato M, Lecky J W, Hiramatsu K et al 1986 Takayasu arteritis: radiographic and angiographic findings in 59 patients. Radiology 161: 329–334 25. Sandu C, Belli A-M 2001 Abdominal aortic stenting: current practice. Abdom Imaging 26: 452–460

26. Surowiec S M, Isiklar H, Sreeram S et al 1998 Acute occlusion of the abdominal aorta. Am J Surg 176: 193–197 27. Gillum R F 1995 Epidemiology of aortic aneurysms in the United States. J Clin Epidemiol 48: 1289–1298 28. Koch A E, Haines G K, Rizzo R J 1990 Human abdominal aortic aneurysms. Immunophenotypic analysis suggesting an immune mediated response. Am J Pathol 137: 1199–1213 29. The UK Small Aneurysm Trial Participants 1998 Mortality results for RCT of early elective surgery or ultrasound surveillance for small abdominal aortic aneurysms. Lancet 352: 1649–1655 30. Mitchel R S, Dake M D, Semba C P 1996 Endovascular stent graft repair of thoracic aortic aneurysms. J Thorac Cardiovasc Surg 111: 1054–1062 31. The Dutch Randomized Endovascular Aneurysm Management (DREAM) Trial Group 2004 A randomized trial comparing conventional and endovascular repair of abdominal aortic aneurysms. N Engl J Med 351: 1607–1618 32. The EVAR trial participants 2004 Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. Lancet 364: 843–848 33. Tang T, Boyle J R, Dixon AK et al 2005 Inflammatory abdominal aortic aneurysms. Eur J Vasc Endovasc Surgery 29: 353–362

CHAPTER

Peripheral Vascular Disease

28

Robert A. Morgan, Anna-Maria Belli and Graham Munneke

• Interventional radiology techniques Arterial system • Pelvic and lower extremity arteries • Upper extremity arteries • Gastrointestinal system • The carotid arteries Venous system • Lower extremity venous system • Upper extremity venous obstruction • Inferior vena cava filters • Complications of endovascular procedures

Since the first edition of this textbook, vascular radiology has expanded beyond recognition. It was only 20 years ago that the role of radiology in the vascular system was to provide diagnostic images using invasive angiography. Since the development of interventional techniques in the late 1980s, interventional radiologists have assumed a major role not only in the diagnosis of vascular disorders, but also in their treatment. The other main advance in vascular radiology has been the development of noninvasive imaging such as duplex ultrasound, computed tomographic angiography (CTA) and magnetic resonance angiography (MRA).The current range of diagnostic modalities and interventional techniques is too extensive for it to be described fully in a general textbook of radiology. The aim of this chapter is to provide a brief overview of salient features of diagnostic angiography and to describe the role of the main interventional techniques in the vascular system. A description of noninvasive angiographic imaging is described briefly elsewhere and a discussion of vascular disease affecting the aorta and renal arteries is covered in Chapters 27 and 39.

the blood vessel and advanced to the site of the lesion. The balloon is inflated for a short period of time. After deflating the balloon, a check angiogram is performed to assess for the success of the procedure (Fig. 28.1).

Stenting This refers to the placement of a metallic mesh tube across a vascular stenosis or occlusion. There are two main types of stent: balloon expandable endoprostheses are mounted on a balloon catheter and deployed by inflating the balloon; and self-expanding stents are compressed on a delivery catheter and released by withdrawing an outer sheath, allowing them to expand by their own radial force (Fig. 28.2)

Plaque

Angioplasty balloon pre-inflation

Vessel wall

Guide wire

Inflation of angioplasty balloon

INTERVENTIONAL RADIOLOGY TECHNIQUES Plaque

The following are brief descriptions of the main interventional methods used by vascular radiologists.

Angioplasty Angioplasty refers to treatment of a vascular stenosis or occlusion with a balloon catheter, which is introduced into

Figure 28.1

Diagram of angioplasty.

Post angioplasty

572

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Plaque

Stents may be used as the primary method of treatment or may be reserved for use if angioplasty is unsuccessful, depending on the location and type of the lesion.

Embolization Thrombotic occlusion

Guidewire

Balloon expandable stent Plaque

Stent

Balloon

Plaque

Post stenting

Figure 28.2

Diagram of stenting.

Embolization refers to the occlusion of a blood vessel by injection of embolic material through a catheter, which has been passed into that vessel percutaneously from a distant puncture site. Embolization may be a definitive treatment in nonmalignant lesions or may be used pre-operatively to reduce blood loss or to alleviate symptoms. There is a large variety of embolic agents, including metallic springs (coils), particulate matter, gelatin sponge, glue and absolute alcohol. The choice depends on the anatomical site, the nature of the lesion and the personal preference of the operator. Some embolic agents, such as gelatin sponge, are temporary, and can be used to control haemorrhage when recanalization of the parent vessel may be desirable once the ‘acute’ lesion has healed, e.g. traumatic injury to the internal iliac artery following pelvic trauma. Permanent particulate emboli are made from various agents, but polyvinyl alcohol is the best known example (Fig. 28.3A–E). These agents are suspended in contrast medium and carried to the lesion where they silt up the blood supply and cause occlusion. The level of occlusion depends on the size and type of particles chosen, but if

Figure 28.3 Embolization. (A) Lytic metastasis in the left humeral head in a patient with renal cell carcinoma. (B) Subclavian artery angiogram demonstrating tumour vascularity over the humeral head (arrow). (C,D) Selective catheterization of the circumflex humeral arteries (arrows) before embolization showing the tumour circulation to advantage. (E) Angiogram after embolization with 355–500 µm polyvinyl alcohol particles demonstrating devascularization of the tumour (arrow). Continued

CHAPTER 28

•

PERIPHERAL VASCULAR DISEASE

Figure 28.3 Cont’d (F) Femoral angiogram in a patient who had an iatrogenic arterial injury during a dynamic hip screw procedure. There is active extravasation of contrast medium from a branch of the profunda femoris artery (arrow). (G) The bleeding has ceased after selective embolization with 3-mm coils (arrow).

there are large arteriovenous communications, they may pass through these into the venous system. Permanent particulate emboli are chosen for the treatment of benign or malignant tumours, e.g. uterine leiomyomata. Coils are used in situations analogous to tying of a vessel surgically but knowledge of vascular anatomy is important to avoid retrograde filling of a lesion from collateral vessels. Coils are best used in end arteries where back filling is unlikely to occur (Fig. 28.3F,G). They are useful for packing the lumen of pseudoaneurysms or can be placed across the neck of a pseudoaneurysm to prevent ‘front and back door’ entry of blood. Liquid embolic agents include sclerosants such as absolute alcohol, polidocanol, hypertonic dextrose and glue-like materials such as isobutyl-2-cyanoacylate. Sclerosant materials are useful in venous embolization, e.g. in varicoceles and low flow vascular malformations. Glue-like materials are useful in arteriovenous malformations and some aneurysms. Common indications for embolization are visceral haemorrhage from the gastrointestinal tract or kidneys, or as an adjunct

to another interventional procedure, e.g. embolization of the internal iliac arteries before insertion of aortic stent grafts.

Thrombolysis This refers to the dissolution of blood clots within an artery or vein by the injection or infusion of a thrombolytic (clot dissolving) drug directly into the thrombus through a percutaneous catheter which has been advanced directly into the thrombus. Although successful thrombolysis may be achieved within a short time, it is usual for the lytic agent to be infused over 24−48 h. Patients usually undergo periodic check angiography to assess the progress of the treatment. In most cases, successful clearance of the thrombus reveals an underlying causative lesion, which should be treated by angioplasty or stenting during the same procedure. Thrombolysis was a very popular technique for the treatment of acute lower limb ischaemia about 10 years ago. It is less often used now, but it has defined indications in several conditions.

ARTERIAL SYSTEM PELVIC AND LOWER EXTREMITY ARTERIES Angiographic anatomy (Fig. 28.4A) At the level of L4 the aorta divides into the common iliac arteries, which pass in front of the iliac veins and give off no major branches. At the level of the mid sacrum they divide into the external and internal iliac arteries. The internal iliac arteries supply the pelvis and surrounding musculature. They divide into anterior divisions which supply the viscera, and posterior divisions which mainly supply the musculature.

The external iliac artery has no major branches although it gives rise to the inferior epigastric artery at the junction with the common femoral artery. At the level of the inguinal ligament, the external iliac artery becomes the common femoral artery—a short vessel that gives rise to the profunda femoris (or deep femoral artery) which supplies the muscles of the thigh; and the superficial femoral artery (SFA), which has no major branches and passes distally. At the level of the adductor canal, the SFA becomes the popliteal artery which gives rise to the vessels of the calf, which are the anterior and

573

574

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Aorta Common iliac artery External iliac artery Circumflex iliac artery Inguinal ligament Common femoral artery

Internal iliac artery Inferior epigastric artery Subclavian artery

Profunda femoris artery and branches

Axillary artery

Superficial femoral artery Circumflex humeral arteries

Brachial artery

Popliteal artery Tibio-peroneal trunk Anterior tibial artery

Posterior tibial artery Common interosseous artery Ulnar artery Peroneal artery

Radial artery Posterior interosseous artery Anterior interosseous artery

Superficial palmar arch

B

A Figure 28.4

Diagram of (A) lower and (B) upper limb anatomy.

posterior tibial arteries and the peroneal artery. At the level of the ankle, the anterior tibial artery becomes the dorsalis pedis artery and the posterior tibial artery becomes the medial and lateral plantar arteries. The anterior tibial artery is the most lateral calf vessel whereas the posterior tibial artery is the most medial. On the forefoot, the plantar arch is formed by the lateral plantar branch of the posterior tibial artery and the dorsalis pedis artery. There are anatomical variations of the lower extremity arteries which are outside the scope of this chapter.

The clinical effect of occlusive disease varies depending on the type, location and number of arterial lesions present. Patients may be asymptomatic, suffer from pain on walking (intermittent claudication), pain while at rest, or tissue loss in the form of either ulceration or gangrene. In general, patients with intermittent claudication are not treated with invasive procedures unless their claudication distance is very short or their symptoms substantially limit their lifestyle. Patients with rest pain and tissue loss are at risk of limb loss and must be treated by angioplasty, stenting, or surgery.

Arterial disease affecting the lower extremity

Angiographic diagnosis

The most common condition affecting the arteries of the lower extremity is ischaemia due to occlusive disease. Occlusive disease may be acute, acute-on-chronic (where acute occlusion occurs in the presence of a previous chronic stenosis or occlusion), or chronic occlusion. Most patients present with symptoms of chronic occlusive disease. The most common cause of arterial occlusive disease in the lower extremity is atherosclerosis. Less common causes include thromboembolism, acute thrombotic occlusion, micro-embolism, trauma and vasculitis, including vasospastic disorders and Buerger’s disease.

Most pathological processes affecting the lower extremity arteries cause stenosis, occlusion or dilatation, i.e. aneurysm formation. Atherosclerosis may affect the arteries at any level, from the iliac arteries to the small vessels of the foot. While it is true that a stenosis or occlusion is almost always due to atherosclerosis, it is important to consider other aetiologies in the differential diagnosis. The clinical history may often be of help in this respect. For example, a patient who develops acute severe pain in the lower leg with no previous history or symptoms has probably sustained an acute embolus in the femoral or popliteal arteries, rather than a long-standing atherosclerotic

CHAPTER 28

occlusion. Patients with diabetes develop arterial occlusive disease, which involves mainly the distal vessels of the calf and feet. Patients with a history of radiotherapy to the pelvis for the treatment of carcinoma of the cervix may develop occlusive lesions of the common and external iliac arteries due to ischaemic vasculitis induced by the radiation.

•

PERIPHERAL VASCULAR DISEASE

Treatment of chronic limb ischaemia Iliac artery disease

no evidence that primary stenting is better than a policy of angioplasty with selective stenting for percutaneous transluminal angioplasty (PTA) failure1. Four-year patency rates are around 60–70%2 which, although lower than those of surgical aortobifemoral bypass, are acceptable considering the minimally invasive nature of this treatment. Patients with diffusely stenosed iliac arteries respond less well to angioplasty. These patients are often treated with stents although there are no data to support this policy.

Stenosis. In the treatment of stenotic lesions balloon angioplasty has a technical success rate approximating 100% (Fig. 28.5A,B). Stents are used when angioplasty is unsuccessful or when lesions recur soon after a previous angioplasty. There is

Occlusions of the iliac artery are usually amenable to endovascular treatment with a technical success rate of recanalization of around 80%. Most operators favour primary stent

Figure 28.5 Iliac angioplasty and stenting. (A) Left anterior oblique iliac angiogram performed with a 4-F flush catheter (black arrow) inserted via the right common femoral artery. There is a 90% stenosis of the mid external iliac artery (white arrow). (B) Improved lumen dimensions following 8-mm balloon angioplasty. Note the fissuring, which is a normal post-angioplasty appearance (arrow). (C) Flush angiogram in another patient. There is occlusion of the right external iliac artery just after the iliac bifurcation (arrow). The external iliac artery reconstitutes distally via collateral vessels (white arrow). (D) A guidewire has been passed retrograde through the occlusion from the right and a 9-mm self-expanding stent deployed. This has resulted in successful recanalization of the occluded segment (arrow).

575

576

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

insertion because distal angioplasty carries a 10–40% risk of embolization of the calf vessels (Fig. 28.5C,D). The durability of endovascular treatment of iliac artery occlusion is similar to that for iliac artery stenosis. Total iliac occlusions are also amenable to endovascular treatment.

Common femoral artery and profunda femoris Stenoses of the common femoral artery are amenable to angioplasty. Access to these lesions must be gained from the contralateral groin involving catheter and guidewire manipulation across the aortic bifurcation, which usually does not present a problem. However, common femoral endarterectomy is a straightforward procedure and may be performed under local anaesthesia. Therefore, surgery may be a better option than angioplasty for lesions involving the common femoral artery, particularly as they are often calcified and eccentric. If the SFA is patent or salvageable by intervention, angioplasty of a stenosis in the profunda artery is usually not carried out. However, if the SFA is occluded, the profunda becomes the main route for blood to reach the lower leg and any stenosis of the main profunda trunk should be treated.The success and durability rates of angioplasty are similar to those for the SFA (see below). Occlusions of the common femoral artery and profunda femoris are generally treated surgically.

Superficial femoral artery Stenosis Angioplasty is the mainstay of treatment for stenosis of the superficial femoral artery (Fig. 28.6A,B). Patency is around 50% at 4 years, which is lower than for surgical bypass. However, most patients are treated by angioplasty in view of the lower rate of complications compared with surgery. Angioplasty can be repeated if lesions recur. An additional advantage of angioplasty is that it spares the long saphenous vein, which is commonly used for femoropopliteal bypass, but which is also commonly needed for coronary artery bypass. The results of angioplasty are less satisfactory if the vessel is diffusely stenosed or if the number of calf run-off vessels is reduced. The technical success of stenting is probably slightly higher than for angioplasty, but its durability is very similar. Therefore, in view of the cost of stents, angioplasty is the preferred option3. Occlusions SFA occlusions are usually treated by angioplasty (Fig. 28.6C,D). Many radiologists use a technique called subintimal angioplasty in which a catheter and guidewire are manipulated outside the lumen of the vessel underneath the intima and into the subintimal space. Unless the vessel is heavily calcified, it is easy to advance the catheter and guidewire down the occluded vessel via the subintimal space. When the guidewire reaches the level of the patent vessel below the occlusion, it usually re-enters the lumen. After replacing the catheter with a balloon catheter, the subintimal space is dilated in the usual manner. Data on the results of subintimal angioplasty for long SFA occlusions are limited. Overall, the results for endovascular management of SFA occlusions are lower than for stenoses. However, angioplasty is still favoured over surgery in view of its lower morbidity and repeatability.

Popliteal artery The principles of treatment, results and durability are similar to those in the SFA. In general, the more distal the lesion, the more likely it is to produce symptoms of critical limb ischaemia. If treatment of these lesions fails, the limb may be lost. In general, lesions in the popliteal artery are only treated if the patient has critical limb ischaemia or very short distance claudication.

Calf vessels Angioplasty has become the main method of treatment for focal or diffuse lesions (stenosis or occlusions) of the tibial and peroneal arteries (Fig. 28.7A,B). In view of the small calibre of these vessels, it is necessary to use small calibre catheters and guidewires. Interventions in the calf vessels are only performed in the setting of critical limb ischaemia, i.e. rest pain or tissue loss. The limb salvage rates are 50–70% at 2 years. Some companies are now manufacturing stents for use in the tibial arteries. However, the data regarding their efficacy are limited (Fig. 28.7C–E). Endovascular intervention in the pedal circulation is seldom performed4,5.

Treatment of acute lower limb ischaemia Patients with acute limb ischaemia usually present with severe rest pain. In many cases, the limb is threatened and patients may develop paraesthesia or motor dysfunction. In such circumstances treatment is required to prevent limb loss. In the 1990s, this condition was often treated with catheter directed intra-arterial thrombolysis. However, this form of treatment has fallen from favour due to the lack of data favouring thrombolysis over surgical management and the relatively high incidence of complications6. Some patients presenting with acute limb ischaemia have small emboli lodged in the popliteal artery. They can be treated by percutaneous aspiration of the thrombus using wide bore catheters placed directly into the thrombus via femoral artery access. A variety of dedicated thrombectomy devices have been tried over the years but none has proved to be better than simple aspiration catheters. Most patients presenting with acute limb ischaemia have a large thrombus load, which cannot be treated in this way.

UPPER EXTREMITY ARTERIES Anatomy (Fig. 28.4B) The subclavian artery extends to the lateral border of the first rib and continues as the axillary artery. The axillary artery extends to the lower border of the teres major, where it becomes the brachial artery. At the elbow, the brachial artery gives rise to the radial artery and ulnar arteries. At the wrist, the radial artery gives rise to the deep carpal arch which anastomoses with branches of the ulnar artery. The ulnar artery gives rise to the superficial carpal arch. The digital arteries originate from both arches.

Pathology Most lesions involving the arteries of the upper limb are caused by atherosclerosis. However, other processes form a

CHAPTER 28

•

PERIPHERAL VASCULAR DISEASE

Figure 28.6 Superficial femoral artery angioplasty. (A) Tight distal superficial femoral artery (SFA) stenosis (arrow) in a patient with claudication. (B) Appearance after angioplasty with a 5-mm balloon (arrow). (C) Occlusion of the SFA in another patient (black arrow). There is reconstitution of the popliteal artery via collaterals (white arrow). (D) The occlusion has been crossed subintimally with hydrophilic guidewire and a 5-mm PTA performed, restoring flow.

577

578

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 28.7 Tibial arteries. (A) Angiogram of the tibial vessels in a patient with rest pain. There is a tight stenosis of the tibioperoneal trunk (arrow). (B) A 0.018-inch guidewire has been placed across the stenosis and a 3-mm PTA performed (arrow). (C) Angiogram of the tibial area in another patient with critical limb ischaemia. There is stenosis of the only run-off vessel, the posterior tibial (arrow). (D) Appearance after angioplasty with a 3-mm balloon. There has been little improvement in lumen dimensions. (E) A 3-mm balloon expandable stent has been deployed across the stenosis to good effect. The proximal and distal stent markers are indicated (arrows).

greater proportion of lesions compared with the legs, including Takayusu’s arteritis, thoracic outlet syndrome, thromboembolism and other vasculitides.

Endovascular treatment Endovascular techniques have a major role in the treatment stenoses of subclavian artery (Fig. 28.8).These lesions usually occur at the origins of the subclavian arteries and are amenable to angioplasty and/or stent insertion with technical success rates of around 95%. There are no convincing data on the superiority of stents versus angioplasty (although most interventionalists would now treat these lesions with stents). Subclavian artery occlusions can be recanalized but the technical success rates are substantially lower than for stenotic disease.

Acute thromboembolism is a common cause of upper extremity arterial problems.Thrombolysis has a limited role to play in the treatment of these patients and surgery remains the treatment of choice. Angioplasty or stenting can be used in the treatment of occlusive lesions distal to the subclavian arteries. However, these lesions are not usually treated unless the patient has limb-threatening ischaemia.

GASTROINTESTINAL SYSTEM The coeliac arteries and the superior mesenteric artery (SMA) usually arise at the level of T12 and L1, respectively. The

CHAPTER 28

•

PERIPHERAL VASCULAR DISEASE

Figure 28.8 Subclavian artery occlusion. (A) Left anterior oblique projection flush aortogram performed via a pigtail catheter in the ascending aorta. There is left subclavian artery occlusion in this patient who presented with arm claudication. Note that the left vertebral artery (the third vessel from the right) arises directly from the aortic arch, rather than off the left subclavian. The stump of the left subclavian is marked (white arrow). There is reconstitution of the distal left subclavian artery (black arrow). (B) A guidewire (black arrow) has been placed across the occlusion from the left brachial artery. A balloon expandable stent (white arrows) is seen in position ready for deployment. (C) After deployment of a 7-mm stent (arrow). Continuous flow has been restored.

inferior mesenteric artery (IMA) arises at the level of L3. The coeliac artery and SMA anastomose with each other via the pancreaticoduodenal arcades while the superior and inferior mesenteric arteries anastomose via the middle colic branch of the SMA and left colic branch of the IMA, just proximal to the splenic flexure.

Angiography (Fig. 28.9) The main pathological processes affecting the gastrointestinal circulation are haemorrhage, occlusive disease and aneurysms.

Mesenteric haemorrhage Upper gastro-intestinal (GI) haemorrhage is defined as bleeding proximal to the duodenal–jejunal flexure. It is commonly caused by peptic ulceration, inflammatory disease such as pancreatitis, or as a complication of endoscopic, surgical, or percutaneous biliary procedures. Lower GI haemorrhage is less common and is usually due to angiodysplasia, diverticular disease, neoplasms, or haemorrhoids. Angiography for mesenteric haemorrhage is performed if endoscopy is negative or if it is not possible to control the site of haemorrhage or identify the precise bleeding point (Fig. 28.10). Angiography can detect active bleeding at a rate of 0.5 ml min−1. If there is no active haemorrhage but the site of bleeding is known or clinically suspected because of associated pathology or intervention, prophylactic embolization may be successful. This is easier in the upper GI than the lower GI tract, where collateral supply is less good and precise identification of the bleeding site is required to avoid ischaemia or infarction. Red cell scintigraphy is more sensitive than angiography and can detect bleeding rates as low as 0.1 ml min−1, but is not able to localize precisely the site of haemorrhage. If all other tests are negative, ultrasound, computed tomography (CT) and magnetic resonance imaging (MRI) may help to detect masses and aneurysms and may demonstrate active haemorrhage. Selective catheterization of the coeliac axis, and superior mesenteric and inferior mesenteric arteries is required. The only direct sign of haemorrhage is contrast medium extravasation into the bowel lumen, but this may not be visible if the

bleeding is intermittent or if its rate is too low. Carbon dioxide angiography may detect slower bleeding but this is not widely available. Indirect signs to identify the source of haemorrhage can be helpful: these include the presence of a pseudoaneurysm, early venous return, vascular lakes or tumour circulation and irregularity of the vessel wall. Manœuvres to provoke haemorrhage include injecting spasmolytic drugs or repeating the contrast medium injection. Injection of thrombolytic drugs can be performed but is not generally recommended as haemorrhage may be massive and full anaesthetic and surgical support is required. Once the source of haemorrhage is identified, embolization can be used to treat the patient, or to enable stabilization of the patient’s condition before surgery. Consideration of the anatomical site will dictate the type of agent used and the extent of the embolization required to avoid ischaemia or infarction of normal vascular territories. For example, consideration should be given to whether the vessel is an end artery (when the risk of infarction is greater) or whether there is significant collateral supply. The embolization agent is selected depending on the anatomy and presence of collateral supply. As a very general rule, particulate emboli are used in the upper GI tract where there is good collateral supply, and coils are used elsewhere where the collateral supply is poor7–9.

Visceral artery aneurysms These are uncommon. The splenic artery is the vessel most frequently involved, followed by the hepatic artery and SMA. The aneurysms may be found incidentally on imaging but otherwise present when they rupture. They may occur as a result of trauma, atherosclerosis, arteritis, collagen vascular disorders, or infection. Interventional radiological procedures used in the management of such aneurysms have evolved considerably (Fig. 28.11). Depending on the vascular anatomy, endovascular treatment may involve embolization (with coils, glue, or particles), insertion of stent grafts, or combinations of these methods. For example, a wide necked aneurysm can be occluded with coils, followed by insertion of an uncovered stent to hold them in place. Alternatively, a covered stent alone may be used. If the aneurysm occurs at a site where the

579

580

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 28.9 Mesenteric anatomy. (A) Coeliac artery: 1 = coeliac axis, 2 = left gastric artery, 3 = splenic artery, 4 = common hepatic artery, 5 = proper hepatic artery, 6 = right hepatic artery, 7 = left hepatic artery, 8 = gastroduodenal artery, 9 = superior pancreaticoduodenal arteries, 10 = right gastroepiploic arteries. (B) Superior mesenteric artery: 1 = sidewinder catheter in the superior mesenteric artery, 2 = jejunal arteries, 3 = ileal arteries, 4 = ileocolic artery, 5 = right colic artery, 6 = middle colic artery. (C) Inferior mesenteric artery: 1 = catheter in the inferior mesenteric artery, 2 = left colic artery, 3 = sigmoid artery, 4 = superior rectal artery.

Figure 28.10 Embolization for gastro-intestinal bleeding. (A) Selective angiogram of the superior mesenteric artery performed with a sidewinder catheter (white arrow). There is active extravasation of contrast medium into the bowel lumen (black arrow) in this patient who presented with melaena. (B) Superselective angiogram performed via a co-axial microcatheter (black arrow). Contrast medium extravasation is seen from a jejunal branch (white arrow). (C) After embolization with microcoils (white arrow) there is no further bleeding.

CHAPTER 28

Renal artery

PERIPHERAL VASCULAR DISEASE

Kidney

Covered stent graft

Uncovered stent

Aneurysm

A

Embolisation coils

B

Micro catheter placed into aneurysm sac past balloon

C Post embolisation and balloon deflation

Aneurysm embolised with onyx

Occlusion balloon placed across the neck of the aneurysm

D

•

E

Figure 28.11 Embolization of aneurysms. (A) Aneurysm of the main renal artery. (B) Exclusion of the aneurysm by covered stent. (C) Coil embolization of the aneurysm with an uncovered stent to hold the coils in place. (D) Aneurysm close to the bifurcation of the renal artery. Occlusion balloon placed in the renal artery across the neck of the aneurysm. Embolization of the aneurysm with Onyx (Micro Therapeutics. Inc., Irvine, CA, USA) via a microcatheter placed in its sac. The balloon prevents the embolization material entering the renal artery. (E) After the aneurysm has been filled with Onyx, the balloon is removed and the Onyx remains in place excluding the aneurysm from the circulation. (F) Angiogram demonstrating a saccular aneurysm (arrow) of the anterior division of the left renal artery (G) Appearance after embolization with onyx (arrow). Both anterior and posterior divisions remain patent.

581

582

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

main artery divides or if there are important side branches which should remain patent, filling the aneurysm with the glue-like material, whilst avoiding extrusion into the parent vessel with a conformable balloon catheter, may be the best method of treatment. These techniques are of value when the parent vessel must remain patent. If the parent vessel can be sacrificed, coils may be placed across the neck, e.g. hepatic artery aneurysm when the portal vein is patent10,11.

Mesenteric occlusive vascular disease Acute mesenteric ischaemia (AMI) is often diagnosed at laparotomy. If suspected, and the bowel is viable, thrombolysis is a potential treatment option. The evidence for this treatment is limited to small case series and reports12,13. As removal of necrotic bowel is essential, endovascular therapy is usually followed by exploratory laparoscopy. Chronic mesenteric ischaemia (CMI) presents with post-prandial abdominal pain and weight loss. Autopsy series quote mesenteric vessel atherosclerosis in 35–70% of unselected patients14. However, clinical symptoms are rare because of the excellent collateral vessels between mesenteric arteries. At least two of the three mesenteric arteries must be significantly stenosed for symptoms to occur. The main therapeutic options for CMI are surgery and angioplasty, with or without stenting (Fig. 28.12). The diagnosis may be made noninvasively by colour Doppler ultrasound, CT, or MRI. Catheter angiography is reserved for confirmation before or at the time of intervention. Lateral aortography is the best way to assess the anatomy before intervention. It is important to exclude extrinsic compression of the coeliac axis by the median arcuate ligament of the diaphragm (MALC) as this requires surgical release of the ligament. MALC causes a nonostial asymmetric narrowing on the superior aspect of the coeliac axis, accentuated on expiratory angiography. Angioplasty of the mesenteric vessels may be performed using a femoral or brachial approach. The early series report the results of angioplasty alone, whilst later series report on PTA and stenting. As most stenoses are found at the origin and are caused by aortic disease, stenting is generally preferred. Although PTA and stenting of one vessel will relieve symptoms, it may be worthwhile treating more than one vessel, if technically feasible, to improve long-term outcome. The relevant literature, which is not extensive, suggests that the initial technical and clinical success rates of endovascular therapy are high (80–100%). Re-stenosis may require repeat angioplasty. Stenting may reduce this problem15,16.

Bronchial artery embolization Massive haemoptysis is defined as more than 300 ml of blood loss over 24 h. Moderate haemoptysis is more than three episodes of 100 ml d−1 within 1 week. The aetiology is variable and includes tuberculosis, cystic fibrosis, malignancy, bronchiectasis, aspergilloma and lobar pneumonia. Bronchial artery embolization is an effective method of stopping haemorrhage and preserving lung parenchyma. However, it does not tackle the underlying problem so re-bleeding is likely with time.

The bronchial arteries arise anterolaterally from the descending thoracic aorta at the level of the fifth or sixth thoracic vertebra. Their anatomy is highly variable but the most common configurations are of an intercostobronchial trunk (ICBT) on the right and two bronchial arteries on the left; one ICBT on the right and one bronchial artery on the left; or one ICBT, one right bronchial artery and two left bronchial arteries.They are catheterized selectively with cobra, sidewinder, or multipurpose catheters to assess whether they are abnormal. Signs of abnormality include hypertrophy, bronchial–pulmonary artery or –venous shunting, peribronchial vascularity, aneurysm and contrast medium extravasation. Bronchoscopy can localize the side of haemorrhage when there is bilateral pulmonary disease but care must be taken in interpreting its findings as they may be misleading. Chest radiography and CT should be reviewed to establish the sites of the disease process as this may help the search for non-bronchial collaterals which can arise from below the diaphragm in lower lobe disease and from the branches of the subclavian artery in upper lobe disease. A descending thoracic aortogram can rapidly identify any hypertrophied bronchial arteries. These arteries are catheterized selectively by using pre-shaped catheters, such as cobra or sidewinder catheters (Fig. 28.13). 3-F coaxial catheters may be passed through these catheters to obtain a super-selective position to avoid any reflux of embolic material into the aorta or into important side branches, such as the anterior spinal artery, which is supplied by the radiculomedullary branch from the right intercostobronchial trunk in 5–10%. Embolization is usually performed using polyvinyl alcohol particles of 300–500 µm. It is prudent to check for non-bronchial collaterals which may arise from the subclavian, axillary, internal thoracic, intercostal, or thyrocervical arteries in upper lobe disease or from branches of the coeliac axis or inferior phrenic artery in lower lobe disease. Bronchial embolization provides immediate relief of symptoms in 75% of cases. If bleeding does not stop immediately, repeat angiography and embolization should be performed. The source of haemorrhage lies in the bronchial arteries in 85–90% and in non-bronchial arteries in 10–15% of cases. The pulmonary arteries are rarely the cause of massive haemoptysis. Recurrent haemoptysis may occur if the underlying pathology cannot be treated effectively and varies from approximately 30% in tuberculosis to 100% in aspergillomas, but the procedure is repeatable. Complications of bronchial embolization include broncho-oesophageal fistula, oesophageal embolization causing dysphagia or necrosis, spinal cord ischaemia causing myelitis or paraplegia, and retrograde aortic embolization.

THE CAROTID ARTERIES Angiography With the development of duplex ultrasound, and MR and CT angiography (MRA and CTA), catheter-based angiography

CHAPTER 28

•

PERIPHERAL VASCULAR DISEASE

Figure 28.12 Mesenteric revascularization. (A) Lateral flush aortogram in a patient who had significant weight loss and post-prandial pain. There are critical stenoses of the coeliac axis (white arrow) and superior mesenteric artery (SMA) (black arrow). The inferior mesenteric artery (IMA) was occluded. (B) Image demonstrating a guidewire in the SMA and a 6-mm balloon expandable stent ready for deployment. The proximal and distal markers of the stent are indicated (arrows). (C) Appearance of the SMA after stent insertion (arrow). Angioplasty of the coeliac axis had been attempted. The vessel exhibits considerable elastic recoil and continued stenosis at its origin (white arrow). (D) Final angiogram after insertion of stents in the coeliac axis (white arrow) and SMA (black arrow)

has little place in the diagnosis of carotid disease except as a problem-solving tool when the noninvasive methods are discordant. Arch aortography should be performed first, as this has less chance of producing distal embolic complications than selective carotid angiography. Selective carotid angiography is associated with a risk of stroke in 1% of procedures. Selective angiography is performed using one of a

variety of catheters, including the sidewinder, the Berenstein, the Headhunter, or the Mani (Cordis, Johnson and Johnson, NJ)17.

Endovascular treatment of carotid artery stenosis Internal carotid stenosis is an important cause of ischaemic stroke and transient ischaemic attack (TIA) (Fig. 28.14).

583

584

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

Figure 28.13 Bronchial embolization. (A) Bronchial angiogram in a patient with lower lobe bronchiectasis who presented with haemoptysis. A large abnormal bronchial artery has been selected with a cobra catheter (arrow). It is providing supply to both lower lobes. (B) Angiogram after embolization with 355–500-µm polyvinyl alcohol particles.

Patients with symptomatic carotid stenosis are at a higher risk of developing further ischaemic cerebral events than patients with asymptomatic stenosis. The risk of developing cerebral infarction after an ischaemic neurological event is highest in the first 6 months. Standard nonmedical treatment is surgical carotid endarterectomy, but carotid angioplasty and stenting is emerging as a promising alternative treatment. Patients who have had a TIA or ischaemic stroke have a 4% increased risk of developing a further event as a result of selective intra-arterial angiography, with a 1% risk of permanent neurological damage17. Therefore, the diagnosis of ICA stenosis is generally made noninvasively by duplex

Figure 28.14 Carotid stenting. (A) Lateral projection angiogram following selective injection into the common carotid artery. There is a tight stenosis at the origin of the internal carotid artery (arrow). The external carotid artery is marked (white arrow). (B) Angiogram following self-expanding stent deployment.

ultrasound, CTA, or MRA. Angiography is usually performed immediately before radiological intervention or to assess suitability. Two trials (European Carotid Surgery Trial [ECST] and the North American Symptomatic Carotid Endarterectomy Trial [NASCET]) showed benefit in surgically treating patients with 70–99% stenosis. The criteria for treating patients with angioplasty and stenting are the same as for surgery. The angioplasty and stenting technique varies from angioplasty at other sites in that predilatation is required for very tight stenoses to avoid trauma to the arterial wall and embolization of material on advancing the stent, and cerebral protection

CHAPTER 28

devices are usually used to catch any embolic debris released during the stenting procedure. One published randomized trial (the Carotid and Vertebral Artery Transluminal Angioplasty Study [CAVATAS]) using predominantly balloon angioplasty before cerebral protection devices became available, showed no difference in the rates of major outcome events between endovascular treatment and surgery (10.9% versus 9.9% for any stroke lasting longer than 7 d or resulting in death). Longer-term follow-up to 3

•

PERIPHERAL VASCULAR DISEASE

years showed no difference in stroke rate between the two groups even though re-stenosis was more frequent 1 year after endovascular treatment than after endarterectomy (14% versus 4%)18. The technique used in the CAVATAS trial has been superseded by stenting and cerebral protection. The results of ongoing randomized trials between best medical therapy, endarterectomy and stenting in symptomatic and asymptomatic cases are awaited19,20.

VENOUS SYSTEM LOWER EXTREMITY VENOUS SYSTEM The main pathology affecting the venous system of the lower extremity is thrombosis. This may be due to factors causing procoagulation such as patient immobility, dehydration and thrombocythaemia; or due to an underlying stenosis/occlusion in the iliac veins. Although some interventionalists advocate catheter directed thrombolysis as first-line treatment for patients with acute lower limb venous thrombosis, this technique has not been widely accepted because it is time consuming and has significant complications.Thrombolysis is occasionally performed for patients with limb-threatening ischaemia as a result of acute venous occlusion in the condition of phlegmasia caerulea dolens. After successful thrombolysis of lower limb deep vein thrombosis (DVT), an underlying stenosis or occlusion of the common iliac vein may be found. Lower limb vein thrombosis or oedema due to a common iliac occlusive lesion is referred to as May–Thurner syndrome. Patency of the iliac vein can be restored by endovascular stent placement.

or IVC, iliac or femoropopliteal DVT who cannot be treated with anticoagulants, or in whom anticoagulants have failed to prevent further PE or progression of thrombus. Other possible indications for IVC filters include protection against PE in pregnant women with proven DVT during caesarean section or childbirth; and pre-operatively in patients with iliofemoral DVT when anticoagulation is contraindicated or pelvic manipulation is expected Many filters are designed for permanent placement. However, retrievable filters are available which can either be left in situ permanently or removed within a certain timeframe. These are advantageous when uncertainty exists regarding the need for permanent filters. Before placing a filter, the IVC should be assessed by vena cavography, the diameter of the IVC measured and the position of the renal veins documented (Fig. 28.15).This is because

UPPER EXTREMITY VENOUS OBSTRUCTION The main causes of upper limb venous occlusion are thoracic outlet syndrome (Paget–Schroetter syndome) and occlusive disease related to the presence of dialysis fistulas. Paget–Schroetter syndrome refers to acute subclavian or axillary vein thrombosis caused by underlying venous obstruction due to the muscles or bony structures of the thoracic outlet. Patients presenting with subclavian or axillary vein thrombosis due to thoracic outlet syndrome are usually treated first by thrombolysis. If the thrombus can be cleared by this method, the first rib should be resected to create space for the vein, followed by angioplasty of any residual stenosis. Patients with high pressure in the upper extremity veins resulting from the presence of dialysis fistulas are prone to develop venous stenoses and occlusions. These can be treated by angioplasty or stenting with high technical success rates. However, recurrence is frequent and long-term durability is very uncommon.

INFERIOR VENA CAVA FILTERS Inferior vena cava (IVC) filters are placed to prevent fatal pulmonary embolism (PE) in patients with a documented PE,

Figure 28.15 Inferior vena cava filter. Cavogram performed with a pigtail catheter (white arrow) placed from the right internal jugular vein. Unopacified blood from the renal veins creates a void (black arrows) in the column of contrast medium and thus delineates their position. Thrombus is seen as a filling defect distally (white arrowhead). (B) Post filter (black arrow) deployment. The delivery catheter is marked (white arrow).

585

586

SECTION 2

• THE CHEST AND CARDIOVASCULAR SYSTEM

the ideal position for the IVC filter is in the infrarenal IVC with the apex of the filter at or just below the level of the renal veins. The manufacturer’s recommendation should be followed with regard to the diameter of the IVC for which a particular filter is recommended. Filters can be inserted via the femoral or jugular venous route depending on the site and extent of the thrombus. Retrieval is via the jugular route, the right jugular vein being the ideal choice. Suprarenal positioning of the filter may be necessary when IVC thrombosis extends above the renal veins or there is renal vein thrombosis. Other indications include thrombus above a previously placed filter, pregnant women in whom there will be compression of the infrarenal vena cava, PE following gonadal vein thrombosis and anatomical variants (double IVC). Other sites where filters have been implanted include the iliac and subclavian veins, and the superior vena cava (SVC).

COMPLICATIONS OF ENDOVASCULAR PROCEDURES Complications occurring after endovascular procedures are divided into major and minor. Major complications include death and those complications where intervention is required, while minor complications are usually selflimiting and do not require treatment. Death occurring after angioplasty or stenting procedures occurs in fewer than 1% of cases. It is generally related to comorbidity rather than to the procedure itself. Other major complications of angioplasty or stenting include vessel rupture, access vessel pseudoaneurysm formation, retroperitoneal haematoma, distal embolization and vessel dissection, stent migration and severe reactions to intravascular contrast medium. Major complications occur in around 1% of procedures2. Minor complications of these procedures are self-limiting haematoma, mild vessel dissection not requiring treatment, minor contrast medium reactions, self-limiting fever and nausea. Thrombolysis is associated with haemorrhage, mainly at the access site, in up to 30% of cases6.

REFERENCES 1. Tetteroo E, van der Graaf Y, Bosch J L et al 1998 Randomised comparison of primary stent placement versus primary angioplasty followed by selective stent placement in patients with iliac artery occlusive disease. Dutch Iliac Stent Trial Study Group. Lancet 351: 1153–1159

2. Dormandy J A, Rutherford R B 2000 Management of peripheral arterial disease (PAD). TASC working group. Trans-Atlantic Inter-Society Consensus (TASC). J Vasc Surg 31: S1–S296 3. Tsetis D, Belli A-M 2004 Guidelines for stenting in infrainguinal arterial disease. Cardiovasc Intervent Radiol 27: 198–203 4. Soder H K, Manninen H I, Jaakkola P et al 2000 Prospective treatment of infra popliteal artery balloon angioplasty for critical limb ischaemia: angiographic and clinical results. J Vasc Intervent Radiol 11: 1021–1031 5. Kudo T, Chandra S A, Ahn S S 2005 Effectiveness of percutaneous transluminal angioplasty for the treatment of critical limb ischaemia: a 10-year experience. J Vasc Surg 41: 423–435 6. Ouriel K, Shortell C K, De Weese J A et al 1994 A comparison of thrombolytic therapy with operative revascularization in the initial treatment of acute peripheral arterial ischaemia. J Vasc Surg 19: 1021–1030 7. Silver A, Bendrick P, Wasvary H 2005 Safety and efficacy of superselective angio embolisation in control of lower gastro intestinal haemorrhage. Am J Surg 189: 361–363 8. Lefkovitz Z, Cappell M S , Lookstein R, Mitty H A, Gerard P S 2002 Radiologic diagnosis and treatment of intestinal hemorrhage and ischemia. Med Clin North Am 86: 1357–1399 9. Schenker M P, Duszak R Jr, Soulen M C, Smith K P, Baum R A , Cope C 2001 Upper gastrointestinal haemorrhage and trans catheter embolotherapy: clinical and technical factors impact in success and survival. J Vasc Interv Radiol 12: 1263–1271 10. Deterling R A 1971 Aneurysm of the visceral arteries. J Cardiovasc Surg 12: 309–322 11. Saltzberg S S, Maldonado T S, Lamparello P J et al 2005 Is endovascular therapy the preferred treatment for all visceral artery aneurysms? Am Vasc Surg 19: 507–515 12. Brountzos E N, Critselis A, Magoulas D, Kagianni E, Kelekis D A 2001 Emergency endovascular treatment of a superior mesenteric artery occlusion. Cardiovasc Intervent Radiol 24: 57–71 13. Simo G, Echenagusia A J, Camunez F et al 1997 Superior mesenteric arterial embolism: local fibrinolytic treatment with urokinase. Radiology 204: 775–779 14. Sheeran S R, Murphy T P, Khwaja A et al 1999 Stent placement for treatment of mesenteric artery stenosis or occlusions. J Vasc Intervent Radiol 10: 861–867 15. Sharafuddin M J, Olson C H, Sun S et al 2003 Endovascular treatment of celiac and mesenteric arteries stenoses: application and results. J Vasc Surg 38: 692–698 16. Matsumoto A H, Angle J F, Spinosa D J et al 2002 Percutaneous transluminal angioplasty and stenting in the treatment of chronic mesenteric ischaemia: results and long-term follow-up. J Am Coll Surg 194 (suppl 1): S22–31 17. Hankey G J, Warlow C P, Sellar R I 1990 Cerebral angiographic risk in mild cerebovascular disease. Stroke 21: 209–222 18. CAVATAS Investigators 2001 Endovascular versus surgical treatment in patients with carotid stenosis in the Carotid and Vertebral Artery. Transluminal Angioplasty Study. (CAVATAS): A randomised trial. Lancet 357: 1729–1737. 19. Kasirajar K, Schneider P A, Kent K C 2003 Filter devices for cerebral protection during carotid angioplasty and stenting. J Endovasc Ther 10: 1039–1045 20. MacDonald S, McKevitt F, Venables G S, Cleveland T T, Gaines P A 2003 Neurological outcomes after carotid stenting protected with the Neuroshield filter compared to unprotected stenting. J Endovasc Ther 9: 777–785

CHAPTER

The Plain Abdominal Radiograph and Associated Anatomy and Techniques

29

Iain Morrison Techniques The plain radiograph The plain abdominal radiograph • Normal appearances • The acute abdome

Abnormal plain radiograph findings • Dilatation of bowel • Abnormal gas distribution • Abnormal bowel wall pattern • Inflammatory conditions • Abdominal calcification

TECHNIQUES The various radiological techniques that are applied to the abdomen are discussed elsewhere in various chapters according to anatomical areas. In this chapter the techniques will be discussed mainly with reference to the acute abdomen. Plain radiographs are frequently still the first

investigation in the acute abdomen. Ultrasound (US) and computed tomography (CT) are increasingly available and important in the early assessment of the patient with acute abdominal pain. Their relative merits and indications will be discussed.

THE PLAIN RADIOGRAPH Plain abdominal radiographs are still commonly performed for the clinically acute abdomen or renal colic. For non-acute situations, the relatively small yield of information from a plain radiograph does not justify the radiation dose that it entails. There are exceptions, such as in the follow-up of renal calculi and in assessment of bowel transit times in constipation. A supine abdomen and an erect chest can be regarded as the basic standard radiographs in the acute abdomen.The clinical condition of the patient will determine whether he or she can stand or sit for the erect radiograph. In a patient who is too ill to be moved it may only be possible to obtain a lateral decubitus with a horizontal ray (Fig. 29.1). The patient should ideally remain in a given position for 10 minutes before the horizontal-ray radiograph to allow time for any free gas to rise to the highest point, although this is rarely achieved in practice.The supine radiograph should ideally be taken with an empty bladder, and should include the area from the diaphragm to the hernial orifices. The erect chest radiograph is superior to the erect abdominal view for the demonstration of free intra-abdominal gas since, in the erect abdominal view, the X-ray beam is passing through any free gas under the diaphragm at an oblique angle.The exposure is also unfavourable for detecting small amounts of gas since

this part of the film is usually overexposed. On the erect chest radiograph, the beam is passing almost tangentially to the free gas, with a better exposure (Fig. 29.2). The erect chest is there-

Figure 29.1 Lateral decubitus X-ray (right side up) showing free gas in the peritoneum over the liver.

590

SECTION 3

•

GASTROINTESTINAL IMAGING

fore invaluable in diagnosing visceral perforation. A chest radiograph is also useful because chest diseases such as pneumonia, pulmonary infarction, aortic dissection or myocardial infarction can present with abdominal symptoms. Historically, an erect abdominal radiograph was performed for the acute abdomen in order to assess the number and length of any fluid levels within bowel.This was thought to distinguish between obstruction and ileus. However, this distinction is highly unreliable, and evidence shows that the erect abdominal radiograph can be misleading. It is now established practice to offer the supine abdominal and erect chest radiographs in the first instance1,2. The erect chest X-ray appearances can be confusing when looking for free peritoneal gas, as discussed later in this chapter. Occasionally a left lateral decubitus radiograph can clarify, since small amounts of gas can be seen over the liver if there has been a perforation.

Ultrasound US has become the first abdominal investigation in numerous clinical circumstances. It is particularly useful in the assessment of the solid organs, namely the liver, biliary tree, pancreas, spleen and kidneys. It is the first investigation of choice for the uterus, ovaries and adnexae, and can be useful for diagnosing conditions of the viscera, such as appendicitis and diverticulitis (see later). The reader is referred to Chapter 3. US should be used wherever possible in children. A more recent development in US is that of US contrast media; microbubbles of gas measuring around 1 µ are injected intravenously and these, sequentially, enhance the arteries, capillary beds and veins. A special low-energy US programme is required3. This technique is particularly useful in increasing the sensitivity of US for detecting small liver lesions such as metastases, and also for helping to characterize liver lesions.

Figure 29.2 Pneumoperitoneum resulting from perforation of a duodenal ulcer. Erect chest radiograph. Typical free gas between the liver and the right hemidiaphragm. Note also the small triangular collection between the loops of the splenic flexure of the colon, beneath the left hemidiaphragm.

Barium and contrast studies The reader is referred to Chapters relevant to the gastrointestinal tract.

Computed tomography The reader is referred to Chapter 4. The indications for CT in the acute abdomen are discussed later.

Magnetic resonance imaging (MRI) Although widely used for certain abdominal applications (e.g. MRCP), MRI is still not widely used as a general abdominal investigation. As its niche roles grow and as the technical aspects become easier, it is likely that its usage will increase.

THE PLAIN ABDOMINAL RADIOGRAPH NORMAL APPEARANCES Relatively large amounts of gas are normally present in the stomach and colon but only a small amount is usually seen in the small bowel. The presence of bowel gas is useful in assessing the diameter and position of the bowel. It is usual to be able to identify the gastric rugae on a supine radiograph.There is rarely sufficient gas present in the small bowel to outline more than a short length, and although the mucosal pattern may be seen, the thin bands of the valvulae conniventes are seldom identified in a normal patient. Air and fluid are normal contents of the small bowel, and short fluid levels are not abnormal on an erect radiograph, should one be obtained.The

following statements about fluid levels on an erect radiograph also apply to CT performed in the supine position, Fluid levels are common in normal people, and they usually lie in the colon. Three to five fluid levels less than 2.5 cm in length may be seen, particularly in the right lower quadrant, without any evidence of intestinal obstruction or paralytic ileus. However, more than two fluid levels in dilated small bowel (calibre greater than 2.5 cm) are said to be abnormal, and usually indicate paralytic ileus or intestinal obstruction4. Fluid levels at different heights in the same loop of small bowel do not help differentiate obstruction from paralytic ileus and may occur in normal people. The significance of air–fluid levels is often overstated. Small-bowel fluid levels are by no means

CHAPTER 29

• THE PLAIN ABDOMINAL RADIOGRAPH AND ASSOCIATED ANATOMY AND TECHNIQUES

specific for obstruction, and reference to Table 29.1 illustrates the number of alternative causes of such an abnormality. The value of the erect abdominal radiograph in diagnosing intestinal obstruction is therefore highly questionable1,2,5. Most of the gas within the bowel has been swallowed; it normally reaches the colon within 30 min6. When severe pain is present anywhere in the body, or when respiration is laboured as in pneumonia or asthma, the amount of air swallowed is increased, producing a dramatic plain abdominal radiograph; such gas-filled, slightly dilated loops of bowel contain relatively little fluid, and the term meteorism is applied to this appearance6. It is sometimes difficult to distinguish meteorism, produced for example by renal colic, from intestinal obstruction but the clinical history and examination frequently enable the radiological findings to be interpreted correctly. The amount of gas present in a normal colon is extremely variable, from almost none to what may appear to be abnormal gaseous distension. Sufficient gas is usually present for the colonic haustra to be identified readily. Large-bowel calibre is very variable, and there is considerable overlap between the normal and the abnormal. The transverse colon normally measures less than 5.5 cm across, but the caecum is easily distensible. A diameter of 9 cm is a critical dimension beyond which danger of perforation exists7. The borders of the kidneys, psoas muscles, bladder and peritoneum, and the posterior borders of the liver and spleen, can often be identified by the fat that surrounds them. The fat lines can be displaced by enlargement of these organs or effaced by inflammation or fluid. It is important to stress, however, that in a number of normal people these fat lines may be blurred or not identified at all. The outline of the spleen cannot be identified in 42 per cent of normal individuals8 and the right psoas is blurred in 19 per cent of adults9. In around half of normal children, the psoas outlines are indistinct. These facts must be considered carefully when assessing the significance of fat line changes. On plain radiography a leaking abdominal aortic aneurysm can cause the

Table 29.1

SOME CAUSES OF SMALL BOWEL FLUID LEVELS

Small-bowel obstruction Large-bowel obstruction Paralytic ileus Gastroenteritis Hypokalaemia Uraemia Jejunal diverticulosis Mesenteric thrombosis Saline cathartics Peritoneal metastases Cleansing enemas Normal (3 cm) are almost always benign but have a higher rate of complications of bleeding or perforation. The en face radiographic signs of gastric ulcers are best seen on double-contrast barium studies and to a lesser extent on compression views in the single-contrast examination. The primary sign of ulcer is a collection of barium on the dependent wall (Fig. 31.8). Most benign ulcers are round or oval; some may have tear-drop or linear contour. If an ulcer is present on a nondependent surface or is not filled with barium, it may be demonstrated as a ‘ring’ shadow, with barium coating the edge of the ulcer crater. If a ring shadow is seen with the patient in the supine position, it may be filled when the patient is turned prone if it is on the anterior wall of the stomach. If the ulcer is on the posterior wall it may be filled using the ‘flow’ technique – turning the patient with fluoroscopic guidance and watching while the barium bolus washes over the posterior wall. En face, a smooth mound of oedema is often seen surrounding the ulcer crater causing a circular filling defect. Radiating folds seen in healing ulcers should be smooth and symmetric and continue to the edge of the crater.The presence of normal areae gastricae extending to the ulcer crater is a good sign of benignancy. The classic description of a benign gastric ulcer refers to lesser curvature ulcers seen in profile. The ulcer, often referred to as the ‘ulcer niche’, projects beyond the lumen of the stomach26 (Fig. 31.9). Sometimes a pencil-thin line of lucency, ‘Hampton’s line’, is present crossing the base of the ulcer27. This is believed to represent preserved gastric mucosa with undermining of the more vulnerable submucosa. This sign

Figure 31.8 En face appearance of benign gastric ulcer. (A) Posterior wall ulcer is nearly filled with barium in this RPO projection. Thin regular radiating folds (best seen around the inferior border of the ulcer) are seen converging to the ulcer. (B) Unfilled benign ulcer crater is outlined by a ‘ring’ shadow. This ulcer is surrounded by a prominent ring of oedema – the lucent area around the crater.

633

634

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 31.9 Profile views of benign gastric ulcer. (A) Hampton's line, a thin line of radiolucency crossing the opening of an ulcer, a virtually infallible sign of a benign ulcer. (B) Lesser curvature ulcer with a clearly visible ulcer collar. (C) Large lesser curvature ulcer niche, its projection from the lumen of the stomach strongly suggesting a benign lesion. (D) Smooth, straight radiating folds converge at the ulcer crater.

is not common, but is virtually diagnostic of a benign ulcer. More often there is a thicker (2–4 mm) smooth rim of lucency at the base of the ulcer termed the ulcer collar. This is also a sign of benignancy.When there is more visible oedema associated with an ulcer it forms an ulcer mound. The ulcer mound should be a symmetrical gently sloping mass. For full characterization of ulcers, barium studies should be performed to include double-contrast views using high-density barium, followed by a low-density barium suspension with prone and/or upright compression views. Almost all (>95 per cent) benign gastric ulcers heal in about 8 weeks when treated medically28. Benign ulcers may heal completely without any radiographic residua. As benign ulcers heal they may change shape from round or oval to linear crevices.There may be subtle retraction or stiffening of the affected wall (Fig. 31.10A). An easily recognizable radiographic sign of healed gastric ulcer is the presence of folds converging to the site of the healed ulcer. There may be a residual central pit or depression (Fig. 31.10B). The radiating folds should be uniform. Incomplete

healing, irregularity of the folds, residual mass or loss of mucosal pattern all suggest the possibility of an underlying malignancy. Occasionally benign ulcers may heal with significant scarring. Severe retraction of the greater curvature from healed ulceration of the lesser curvature may cause narrowing of the mid-body of the stomach. Healing of antral ulcers may form prominent transverse folds or significant antral narrowing and deformity. Such scarring may lead to significant obstruction, an important complication of peptic ulcer disease (PUD). Penetration and perforation are other potentially life-threatening complications of gastric ulcers.

Gastric erosions Gastric erosions or aphthous ulcers are shallow ulcerations that do not penetrate the muscularis mucosa. They usually appear as small, shallow collections of barium 1–2 mm in diameter surrounded by a radiolucent rim of oedema (Fig. 31.11).These are called ‘complete’ or ‘varioloform’ erosions. When the halo of oedema is lacking the erosions are called ‘incomplete’.

Figure 31.10 Healing gastric ulcer. (A) Focal retraction along the incisura angularis with small residual out-pouching is present. Converging smooth folds no longer fill an ulcer crater. (B) Radiating folds converging to a linear scar. (C) Scarred antrum with constriction at site of previous ulcer causing narrowing and deformity.

CHAPTER 31

• THE STOMACH

Figure 31.11 Gastric erosions. (A) Complete or ‘varioliform’ erosions in antrum seen on double-contrast views. (B) Incomplete erosions in antrum on compression view. The erosions do not have a rim of oedema.

These appear as short linear or serpentine lines or dots of barium. Gastric erosions are most easily detected radiographically when they are multiple and complete. They are most often detected on double-contrast barium studies but may also be seen with compression technique29. Single or incomplete erosions are commonly detected endoscopically but are infrequently identified radiographically30. Erosions heal without scarring. Gastric erosions are most often causally related to H. pylori infection31. Other causes include alcohol, NSAID ingestion and Crohn’s disease. They may also be seen as a response to stress, for example in patients with severe trauma.

Gastritis Gastritis is a descriptive term with sometimes conflicting pathological, endoscopic and radiographic definitions. It is now better understood that many causes of gastritis including H. pylori, alcohol and NSAID gastritis lead to similar morphological changes. Since H. pylori has been shown to be the most common cause of gastritis, its radiological manifestations have been described31. The most common findings are thick (>5 mm) folds with or without nodularity (Fig. 31.12). Erosions, while less commonly seen, are a frequent sign of H. pylori gastritis. Other signs of gastritis include antral narrowing, inflammatory polyps and prominent areae gastricae. The radiological findings are similar to endoscopic findings. A study by Sohn et al10 demonstrated that the finding of thickened gastric folds, although nonspecific, is still the single most useful radiographic sign for the diagnosis of H. pylori gastritis and that the combination of thick folds and enlarged areae gastricae may be the most specific findings. The limitation of the radiological signs is their lack of specificity.There is considerable overlap of findings between patients who are biopsy positive or negative for H. pylori. Therefore it is often not possible at this time to distinguish between ulcers and gastritis caused by H. pylori or those caused by chemical irritants (alcohol, NSAIDs and other aetiologies of gastritis)32.

Owing to the prevalence of H. pylori, its association with many gastric diseases and effective treatment options, it is important for the radiologist to recognize findings that suggest the presence of the infection.

Atrophic gastritis Atrophic gastritis is a combination of atrophy of the gastric glands with histological inflammatory changes. It is associated with pernicious anaemia and is more common with advancing age.This disease is increasing and is caused by a decreased production of intrinsic factor with malabsorption of vitamin B12. Atrophic gastritis is found in more than 90 per cent of patients with pernicious anaemia and is characterized by loss of parietal and chief cells leading to achlorhydria, and atrophy of the mucosa and mucosal glands33. There is an association

Figure 31.12 Diffuse erosive gastritis with thick nodular folds. Erosions are scattered along the folds.

635

636

SECTION 3

•

GASTROINTESTINAL IMAGING

with gastric polyps and carcinoma34 and ulcers, both benign and malignant, may occur. Intestinal metaplasia, which may be seen histologically in atrophic gastritis, is considered a premalignant condition. This diagnosis may be suggested by focal enlargement of the areae gastrica35. Radiographic findings of atrophic gastritis include loss of rugal folds and a tubular, featureless narrowed stomach (Fig. 31.13). Areae gastricae may be absent. The radiographic features are nonspecific, but because of the important prognostic implications of atrophic gastritis an appearance suggesting this diagnosis should trigger an appropriate clinical workup.

Infectious gastritis H. pylori gastritis is by far the most common infection affecting the stomach. Tuberculosis, histoplasmosis and syphilis are usually lumped together with other granulomatous processes causing gastritis. Ulceration, thick folds and mucosal nodularity are common features, with antral narrowing being a late feature of these diseases. Monilia may involve the stomach.This almost always occurs in the presence of severe oesophageal disease. Prominent aphthous ulceration may be seen in such cases36. In immunocompromised patients, cytomegalovirus, crytosporidiosis and toxoplasmosis may occur. Radiographic findings are nonspecific but there are some suggestive signs. Deep ulceration, and even fistulization to adjacent structures may be seen with cytomegalovirus37. Cryptosporidium primarily affects the small bowel causing severe diarrhoea and thick, small bowel folds. It rarely involves the stomach but has been shown to cause deep ulcers, antral narrowing and rigidity38,39. Strongyloides is a parasitic infection with a worldwide distribution that affects the upper gastrointestinal tract, duodenum and proximal small bowel with thickened, effaced folds and narrowing. In advanced cases, the stomach may be narrowed and have thickened folds.

Crohn’s and other granulomatous diseases Granulomatous conditions of the stomach include Crohn’s disease, sarcoidosis, tuberculosis, syphilis and fungal diseases. Crohn’s disease is a chronic inflammatory bowel disease pri-

Figure 31.13 Atrophic gastritis. Featureless narrowed stomach. Note pyloric channel seen ‘end on’.

marily affecting the ileum and colon. Gastroduodenal involvement occurs in up to 20 per cent of cases most often in the presence of ileocolitis40. When the upper gastrointestinal tract is affected by Crohn’s disease both the stomach and duodenum are involved. However, involvement of the duodenum alone is more common than the stomach alone41. A wide range of symptoms may be seen in patients with gastroduodenal Crohn’s disease; some are asymptomatic, others have symptoms more typical of PUD or with symptoms related to antral narrowing or gastric outlet obstruction. Diarrhoea due to associated ileocolic disease is common. Gastrocolic fistula is an unusual complication. When it occurs, disease of the transverse colon extending to the stomach is most often the cause42. Radiographic findings of gastric Crohn’s disease almost always involve the antrum or antrum and body of the stomach43.With early disease, findings include aphthous ulcers, larger discrete ulcers, thickened and distorted folds and sometimes a nodular (‘cobble-stoned’) mucosa (Fig. 31.14). These are indistinguishable from aphthous ulcers or erosions due to other causes. These findings represent the nonstenotic phase of Crohn’s disease. Stenotic disease is caused by scarring and fibrosis narrowing of the gastric antrum and pylorus into a funnel or ‘rams-horn’ shape, sometimes foreshortening the stomach enough that it simulates a partial gastrectomy. This scarred, funnel-shaped antroduodenal region may also be seen in other granulomatous diseases, including tuberculosis, syphilis, sarcoid and eosinophilic gastroenteritis. Antral narrowing may also mimic scirrhous gastric carcinoma44.

Hypertrophic gastritis Hypertrophic gastritis is characterized radiographically by thickened folds, often greater than 10 mm in width, predominantly in

Figure 31.14 Crohn's disease. Multiple aphthous erosions are present on the antrum. Duodenal folds are thick and nodular.

CHAPTER 31

• THE STOMACH

the fundus and body, which are the acid-producing regions of the stomach.While the term is often used descriptively, the entity of hypertrophic gastritis is associated with glandular hyperplasia and increased acid secretion45,46. Histologically, inflammation is not a prominent feature thus ‘gastritis’ is somewhat a misnomer. The areae gastricae pattern may be prominent, up to 4–5 mm in size and more angular and polygonal than the usual round or oval configuration47 (Fig. 31.15). There is a high prevalence of duodenal and gastric ulcers in these patients. Many cases that had been classified as hypertrophic gastritis may in fact have been due to H. pylori infection. The differential diagnosis is primarily Ménétrier’s disease and lymphoma.

Ménétrier’s disease Ménétrier’s disease is a rare entity well known in the radiology literature because of its dramatic and characteristic appearance. This condition is characterized by hypertrophy of gastric glands, achlorhydria and hypoproteinaemia. Loss of protein from the hyperplastic mucosa into the gastric lumen results in a protein-losing enteropathy, and may produce disabling symptoms. This manifests as increased fluid in the small bowel. The disease is characterized by markedly enlarged, often bizarre gastric folds most prominent in the proximal stomach and along the greater curvature48. Radiographically, upper gastrointestinal or CT show massively thickened often lobular folds (Fig. 31.16). Increased fluid may prevent optimal mucosal coating with barium. The folds remain pliable, which helps to differentiate it from carcinoma, where the stomach becomes rigid and aperistaltic. While in the classic description of Ménétrier’s disease the antrum is spared, it has been found to be involved in up to 50 per cent of cases49 causing diffuse involvement of the stomach.

Figure 31.16 Ménétrier's disease. Classic appearance with massively distended folds in the body without abnormality in the antrum.

Zollinger-Ellison syndrome Patients with Zollinger-Ellison syndrome may have thickened gastric folds and increased gastric secretions. This syndrome is caused by gastrinomas, which are nonbeta islet cell tumours that secrete gastrin, stimulating acid secretion in the stomach. Seventy-five per cent of the tumours are found in the pancreas and 15 per cent in the duodenum. A small number are extraintestinal. The tumours may be malignant and metastases, primarily to the liver, are present in up to half of the cases. Zollinger-Ellison is one of the manifestations of multiple endocrine neoplasia (MEN) Type I, which includes parathyroid, pituitary and adrenal tumours.

Eosinophilic gastritis Eosinophilic gastroenteritis is characterized by focal or diffuse infiltration of the gastrointestinal tract by eosinophils.The clinical presentation includes crampy abdominal pain, diarrhoea, distension and vomiting, often in an atopic or asthmatic patient50. Peripheral eosinophilia is a frequent accompaniment. Any segment of the gastrointestinal tract may be affected but it most often involves the stomach, especially the antrum and the proximal small bowel. The clinical and imaging features depend on which layers of the GI tract wall are involved51. Involvement may be predominantly mucosal, muscular or subserosal52. Many cases are panmural and eosinophilic ascites is often seen in such cases53–55. Radiographically eosinophilic gastritis is characterized by fold thickening of the stomach and small bowel. Antral narrowing and rigidity with mucosal nodularity are frequently seen56 (Fig. 31.17).

Corrosive ingestion Figure 31.15 Hypertrophic gastritis in a patient with a recently healed lesser curvature gastric ulcer. This characteristic enlargement and prominence of the areae gastricae can be correlated with an increased incidence of gastric hypersecretion and PUD.

Ingestion of caustic chemicals may cause severe injury to the stomach, sometimes leading to gastric necrosis, perforation and death. Acids are more injurious to the stomach and duodenum. Since the stomach secretes hydrochloric acid, it has the ability

637

638

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 31.17 CT shows diffuse thickening of the gastric wall in a patient proven to have eosinophilic gastroenteritis. No ascites was present. Symptoms resolved with steroid therapy.

Figure 31.18 Corrosive gastritis following the ingestion of household bleach. The distal stomach has undergone considerable scarring and contraction in a manner similar to syphilitic gastritis or linitis plastica.

to neutralize alkaline agents. However, the already acidic gastric contents have no ability to neutralize strong ingested acids such as sodium hypochlorite (household bleach). The consequences of ingesting a corrosive agent follow a distinct course. First, there is necrosis with sloughing of the mucosal and submucosal layers. In severe cases, full-thickness necrosis of the gastric wall may lead to perforation. In less severe cases, the denuded gastric wall develops a granulating surface and the formation of collagen then leads to fibrosis and stricture. The final result is a deformed, contracted and occasionally obstructed stomach.The outcome is often total gastrectomy. The radiographic findings in corrosive gastritis depend on the severity of the chemical insult and the time that has elapsed since injury. Initially, swelling and irregularity of the gastric mucosa are seen, occasionally with visible blebs. As the mucosa sloughs, barium flows beneath it and the mucosa may then be seen as a thin radiolucent line paralleling the outline of the stomach. After a week or two, fibrotic contraction of the stomach becomes evident (Fig. 31.18). In severe cases, the lumen of the stomach may be no larger than that of the duodenal bulb.

pneumoniae. Infectious gastritis with any organism is unusual but is always a fulminant, highly morbid process. The radiological findings in both emphysematous gastritis and gastric emphysema include thin, curvilinear lines of radiolucent gas paralleling the gastric wall (Fig. 31.19).

Air in the gastric wall Disruption of the gastric mucosa permits air to enter the gastric wall. When this occurs without an underlying infection it is called gastric emphysema57. Causes include corrosive ingestion, gastric ulcer, gastric outlet obstruction, chronic obstructive pulmonary disease (COPD), ischaemia and trauma. Air in the gastric wall caused by an acute infection with a gas-forming organism is called emphysematous gastritis. Escherichia coli and Clostridium welchii are the usual causative agents. Acute panmural infectious gastritis by nongas-forming organisms may also occur and is referred to as phlegmonous gastritis. Causative infectious agents include alpha-haemolytic streptococcus, Staphylococcus aureus E. coli, C. welchii and Streptococcus

Amyloidosis Amyloidosis is a rare condition that may cause gastric fold and/or wall thickening and rigidity. Luminal narrowing may mimic infiltrative tumour (linitis plastica). The condition is caused by the deposition of amyloid, a protein-saccharide complex in the stomach58.

Positional abnormalities The stomach is attached to several peritoneal reflections, permitting it relative mobility. These include the gastrohepatic ligament (lesser omentum), the gastrosplenic ligament and the gastrocolic ligament, which is part of the greater omentum.The oesophagogastric junction passing through the oesophageal hiatus of the diaphragm normally fixes the proximal stomach while the distal stomach is anchored at the pyloroduodenal junction.

Hiatus hernia Hiatal hernias are the most common positional abnormality in which the stomach herniates into the chest through the oesophageal hiatus when there is widening of the opening between the diaphragmatic crura. The prevalence of hiatal hernia increases with age and is present in over 50 per cent of the aged population. Most hiatal hernias are small, involving a protrusion of a part of the gastric fundus at least 1.5–2 cm above the diaphragm. At the opposite extreme the entire stomach may be intrathoracic. Sliding hiatal hernias are the most common type of hiatal hernia. In this type of hernia the gastro-oesophageal junction slides proximally through

CHAPTER 31

• THE STOMACH

Figure 31.19 (A) Gastric emphysema on abdominal radiograph in a patient with ischaemic gastritis after extensive abdominal surgery. (B) CT of patient with infectious, emphysematous gastritis.

the diaphragmatic hiatus to assume an intrathoracic location. Small or moderate-sized hiatal hernias are often reducible, changing in size and configuration during barium evaluation. They are best demonstrated with the patient recumbent in the right anterior oblique position. Sliding hiatal hernias are often accompanied by gastro-oesophageal reflux and reflux oesophagitis (Fig. 31.20). In comparison to sliding hiatal hernias, in paraoesophageal hernias the gastro-oesophageal junction is in its normal position below the diaphragm. The proximal stomach herniates through the oesophageal hiatus usually to the left of the distal oesophagus in the posterior mediastinum. This type of hernia is important because it is more prone to incarceration and obstruction than a sliding hernia. Traumatic diaphragmatic hernias result from a tear in the diaphragm either from a direct penetrating injury or from a sudden increase in intra-abdominal pressure during blunt trauma. These hernias are almost always on the left side.

Herniation may occur immediately after trauma or may be delayed by many years. Diagnosis is often difficult both due to lack of specificity of symptoms and because it is often confused with simple elevation of the hemidiaphragm. On barium studies the recognition of the gastric hernia lateral to the normal oesophageal hiatus is crucial. CT is also often helpful in diagnosis.

Gastric volvulus Gastric volvulus occurs when the stomach twists on itself between the points of its normal anatomical fixation. It is clinically important as it may cause gastric outlet obstruction or vascular compromise resulting in a surgical emergency. Classically, it presents clinically with violent retching with little produced vomitus, severe epigastric pain and difficulty passing a nasogastric tube. Gastric volvulus is most common in the elderly but may occur at any age. It is usually associated with large sliding or para-oesophageal hiatal hernias59. Most

Figure 31.20 (A) Sliding hiatal hernia. Chest X-ray shows a large air collection overlying the heart shadow. (B) CT shows a large part of the bariumfilled stomach is in the chest. (C) In this patient, an upper GI exam shows that the gastric fundus and most of the body of the stomach is intrathoracic

639

640

SECTION 3

•

GASTROINTESTINAL IMAGING

of the time gastric volvulus involves a stomach that is partially or totally in the thoracic cavity and that rotates between the normally positioned gastric ligaments. Other predisposing factors include phrenic nerve palsy, eventration of the diaphragm, traumatic diaphragmatic hernia, gastric distension and abnormalities of the spleen60. Gastric volvulus is often divided into two types depending on the plane of torsion. In organo-axial volvulus the stomach rotates along its long axis, which is a line drawn between the cardia and the pylorus. Rotation may be to the right or left. The configuration of the torsed stomach depends on the original shape and position of the stomach (horizontal or vertical). If the normal stomach was in a horizontal position, volvulus flips the stomach upward so that the greater curvature is superior to the lesser curvature. If the stomach was originally vertically orientated, volvulus causes a right–left twist. Mesenteroaxial volvulus is less common but more likely to have significant clinical consequences. In this type of volvulus the stomach rotates on an axis perpendicular to the long axis of the stomach along a line joining the middle of the lesser curvature to the greater curvature. This corresponds to the axis of the mesenteric attachments of the greater and lesser omentum61. The characteristic appearance is an ‘upside-down stomach’ with the distal antrum and pylorus assuming a position cranial to the fundus and proximal stomach. This type of volvulus is often associated with traumatic diaphragmatic ruptures. Radiographic signs of gastric volvulus include a double air–fluid level of the stomach in the mediastinum and upper

abdomen on upright plain radiograph. On barium studies, the stomach may be inverted with the greater curvature above the lesser curvature, or the pylorus above the cardia and the torsed area identified as the source of the obstruction62 (Fig. 31.21).

Figure 31.21 Gastric volvulus. The stomach is inverted with the greater curvature above the lesser curvature.

EXTRINSIC IMPRESSIONS Organomegaly or masses in upper abdominal organs may displace or cause extrinsic impressions on the stomach. These most commonly involve the spleen, left lobe of the liver and pancreas. These processes may be deduced from the position and configuration of the compression on upper gastrointestinal examination63, but are primarily evaluated by computed tomography (CT) or ultrasound (US).

MISCELLANEOUS BENIGN CONDITIONS Prepyloric web (antral mucosal diaphragm) This occurs as a diaphragm-like, exaggerated fold of gastric mucosa64 oriented perpendicular to the long axis of the stomach. The thin web demarcates the distal antrum into a third small chamber between the proximal antrum and duodenal bulb. This is believed to be a congenital lesion. It appears as a thin persistent circumferential smooth band within 3–4 cm of the pylorus. While usually asymptomatic it may present with symptoms of gastric outlet obstruction. The antral chamber produced by the web may mimic a second duodenal bulb.

Diverticula Gastric diverticula are most common in the posterior aspect of the fundus59, below the oesophagogastric junction and near the

lesser curvature65, or rarely in the antrum (Fig. 31.22). These are true diverticula, i.e. containing muscularis propria, and thus are capable of peristalsis. They are usually several centimetres in size and readily fill with barium. They rarely present a diagnostic dilemma but may mimic a submucosal mass if they fail to fill with barium. They may be mistaken for a gastric ulcer. Intramural or partial gastric diverticula describes a rare anomaly in which there is invagination of gastric mucosa into the gastric wall. These diverticula are usually smaller than 1 cm in size and have a lenticular shape in profile with a small opening into the gastric lumen66,67. They are typically located on the greater curvature of the distal antrum and are generally considered to be asymptomatic. Radiologically they may present a problem as they can be mistaken for ulcers, or for aberrant (ectopic) pancreatic rests, which typically occur in the same region.

Hypertrophic pyloric stenosis Hypertrophic pyloric stenosis is a relatively frequent congenital disorder diagnosed in infancy. Presentation in adults occasionally occurs. The morphological features are due to hypertrophy and hyperplasia of the circular muscle with some contribution by the longitudinal muscle. The hypertrophied muscle lengthens and narrows the pyloric channel. Radiographically, there is lengthening of the pyloric

CHAPTER 31

• THE STOMACH

Figure 31.22 Large fundal diverticulum. The patient is upright and a barium/air level is present in the diverticulum.

channel (2–4 cm long) with smooth symmetrical narrowing. The hypertrophied muscle bulges retrograde into the antrum, creating a ‘shoulder’. In infants US generally provides the definitive diagnosis. A similar appearance may be seen with acquired hypertrophy of the distal antrum and pylorus. This is usually a sequelum of peptic or other inflammatory disease68. This form of pyloric stenosis typically lacks the retrograde bulge of muscle.

Varices Gastric varices are seen in most patients who have portal hypertension and oesophageal varices. Gastric veins provide one of the collateral pathways when there is obstruction of the portal vein. The presence of gastric varices in the absence of oesophageal varices is a sign of splenic vein thrombosis most often associated with pancreatitis or pancreatic carcinoma69. Varices are most often seen in the fundus around the oesophagogastric junction sometimes involving the proximal body. They appear as widened, effaceable polypoid folds. They may be nodular-appearing, ‘grape-like’ or appear masslike, in which case they may mimic gastric cancer70–72. Rarely they occur in the antrum without fundal involvement73. Transabdominal and endoscopic US are important techniques for definitive diagnosis of gastric varices. Differential diagnosis for thick polypoid gastric folds includes hypertrophic gastritis, Ménétrier’s disease and lymphoma (Fig. 31.23).

Gastric distension Gastric distension may be obstructive or nonobstructive. PUD and gastric cancer account for more than 90 per cent of cases of gastric outlet obstruction. PUD is the most common cause of gastric outlet obstruction in adults. Duodenal and pyloric channel and distal antral ulcers are the usual culprits. Obstruction is caused by one or more factors including spasm, scarring, acute inflammation and muscle hypertrophy. Carcinoma of the distal antrum or pylorus is the second most common cause of gastric outlet obstruction. Infrequent causes include

Figure 31.23 Multiple nodular and curved submucosal masses in the fundus are a common appearance of gastric varices.

Crohn’s disease, sarcoidosis, tuberculosis, syphilis (granulomatous diseases) and pancreatitis or pancreatic cancer. Abdominal radiographs show the outline of the dilated airfilled stomach or downward displacement of the transverse colon by a fluid- or air-filled stomach (Fig. 31.24). A mottled pattern of air and residual retained gastric contents may also be seen. If barium is given orally or through a nasogastric tube, an attempt should be made to delineate the margins and the contours of the obstruction to determine if an ulcer or mass is present. If some barium enters the duodenum, it can be assessed for scarring and/or ulceration. Barium studies are less than ideal for the delineation of the obstruction. CT is superior in depicting a mass or abnormal gastric wall when gastric malignancy is the cause of obstruction (Fig. 31.25). Nonobstructive gastric dilatation may be acute or chronic. Rapid sudden gastric distension, or gastric ileus occurs most often as a complication of abdominal surgery or acute trauma. Chronic gastric dilatation is common in diabetics, where gastroparesis, i.e. delayed gastric emptying, due to decreased or absent gastric peristalsis occurs in 20–30 per cent of patients. These are most often poorly controlled diabetics74.

Ectopic pancreas Ectopic pancreas (pancreatic rest, aberrant pancreas) is the presence of pancreatic tissue in the submucosa of the luminal gastrointestinal tract. It is most commonly visible along the greater curvature of the antrum. The ectopic pancreatic tissues are usually solitary deposits that appear as sharply defined submucosal nodules usually 2 cm in size88. Surprisingly, gastric carcinoma is much more common than gastric adenomas and it is therefore hypothesized that unlike the situation in the colon, many gastric cancers arise de novo and not from degeneration of adenomas85. Adenomas are polypoid, usually solitary lesions >1 cm in size, most often seen in the antrum. They may be sessile or pedunculated and are usually lobulated in contour. Villous tumours are adenomas characterized by numerous frond-like projections that radiographically may appear bubbly. These tumours carry a very high risk of malignancy. Adenomatous polyps are often found in patients with familial adenomatous polyposis coexisting with hypertrophic polyps.

• THE STOMACH

incidentally during evaluation for unrelated problems. Ulceration becomes more common as the lesions grow to >2 cm in size and symptoms, including epigastric pain and gastrointestinal bleeding, become more common. Radiographically the tumours appear as discrete submucosal masses typified by a smooth mucosal surface with borders forming right or slightly obtuse angles to the adjacent mucosa (Fig. 31.33). En face, the preservation of a normal areae gastricae pattern over the mass confirms the presence of normal mucosa and the extramucosal location of tumour. When there is ulceration it is usually seen as a central collection of barium in a smooth or slightly lobulated mass. This is sometimes called a ‘target’ or ‘bulls-eye’ lesion. Leiomyomas occasionally contain coarse mottled calcifications. In about 15 per cent of cases the tumours grow predominantly outside the stomach (exogastric) and may in less than 5 per cent of cases have an intra- and extraluminal growth pattern (‘dumbbell’). Occasionally they are pedunculated and may obstruct the pylorus or duodenum93 or act as the lead point of an intussusception. While the vast majority of this class of tumours are benign, up to 10 per cent are malignant. Unfortunately, the prediction of malignancy is difficult even by histological criteria. GISTs are classified by their estimated risk of recurrence and metastasis into low- or high-risk categories94. Imaging findings contribute importantly to the assessment of risk of malignancy. A favourable prognosis is associated with tumour size 5 cm, malignancy should be strongly considered. As a result of the large size of these tumours, they frequently outgrow their blood supply and ulcerations are therefore common. Tumours may arise from any of the mesenchymal elements of the gastric wall, including the neural elements.These are all radiographically indistinguishable from one another.

Table 31.4

LYMPHOMA VERSUS GASTRIC CARCINOMA Lymphoma

Gastric carcinoma

Wall thickening

Very thick, mean = 4 cm, 1.8 cm, focally thick

Less thick, mean = homogeneously thick

Perigastric fat planes

Usually preserved

May be obliterated

Regional adenopathy

Common

Common

Extent of adenopathy

May extend below level

Does not extend below renal vein

Extent

Large bulky nodes

Less bulky

May involve duodenum

Does not commonly involve duodenum

Metastatic disease The most common primary tumours that metastasize to the stomach are breast, malignant melanoma and lung106. Bloodborne metastases to the stomach appear radiographically as gastric wall lesions. Initially these can be seen as small intramural masses, usually multiple, that are indistinguishable from benign disease. These may contain central ulcerations, having a bulls-eye appearance. This is most frequently seen in metastatic melanoma, lymphoma and Kaposi’s sarcoma107. As these lesions grow, the aetiology becomes apparent. Breast carcinoma may produce a linitis plastica-type appearance, indistinguishable from primary gastric carcinoma.

ADVANCES IN GASTRIC IMAGING With the development of the multidetector CT systems, new computer applications have become available such as virtual colonography. It offers a noninvasive way to diagnose abnormalities in the colon such as colon cancer. This technique can likewise be applied to the stomach for virtual evaluation. Following air distension of the stomach (Fig. 31.37) we can apply the ‘fly-through technique’ to

the stomach to visualize the gastric folds, pylorus etc. (Fig. 31.38). While these advances are still investigational it certainly offers an alternative means of visualizing pathology in a patient who is not a candidate for more invasive procedures. Several similar articles have been published on virtual imaging using magnetic resonance.

CHAPTER 31

Positron emission tomography (PET) has proven itself to be valuable in the evaluation of patients with gastric cancer. The study utilizes [F-18] Fluoro-2-deoxy-glucose as a marker of tumor activity. It is a functional technique that images tissue metabolic activity.

• THE STOMACH

Fusion of PET with CT is rapidly becoming the standard of care when diagnosing malignancy. PET/CT is currently used for staging and restaging of gastric lymphoma (Fig. 31.39) as well as following response to therapy.

Figure 31.37 Virtual endoscopy. (A) Digital CT Radiograph following air distension of the stomach. (B) Axial view of the stomach using a threedimensional technique in diagnosing abnormalities.

Figure 31.38 Virtual endoscopy. (A) Normal appearance of gastric folds as seen on the virtual endoscopy. Image shows a view from the fundus looking into the body of the stomach. (B) Normal appearance of the pylorus as seen on the virtual endoscopy.

649

650

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 31.39 FDG PET/CT. The images depicted show first normal uptake of FDG by the stomach and second increased uptake in an area of proven gastric lymphoma. (Courtesy of Dr Alan H. Maurer, Nuclear Medicine Department, Temple University Hospital.)

REFERENCES 1. Gore R M, Lichtenstein J E 1998 The gastrointestinal tract: anatomicpathologic basis of radiologic findings. In: Traveras J M, Ferrucci J T (eds) Radiology, 4th edn, Lippincott-Raven, Philadelphia, chapter 4 2. Kawai K, Tanaka H 1974 Fundamentals of x-ray diagnosis. In: Differential Diagnosis of Gastric Diseases. Year Book, Chicago, pp 11–53 3. Gelfand D W 1984 Gastrointestinal Radiology. Churchill Livingstone, New York, pp 29–46, 55–66 4. Levine M S, Rubesin S E, Herlinger H et al 1988 Double-contrast upper gastrointestinal examination – technique and interpretation. Radiology 168:593–602 5. Evers K, Kressel H Y 1982 Principles of performance and interpretation of double-contrast gastrointestinal studies. Radiol Clin North Am 20:667–685 6. Fishman E K, Urban B A, Hruban R H 1996 CT of the stomach: spectrum of disease. Radiographics 16:1035–1054 7. Brant W E 1999 Gastrointestinal tract. In: Webb W R, Brant W E, Helms C A (eds) Fundamentals of Body CT, 2nd edn. Williams and Wilkins, Baltimore, MD, pp 723–730 8. Megibow A J 1994 Computed tomography of the gastrointestinal tract: techniques and principles of interpretation. In: Gore R M, Levine M S, Laufer I (eds) Textbook of Gastrointestinal Radiology, vol 1. WB Saunders, Philadelphia, p 103 9. Mackintosh C E, Kreel L 1977 Anatomy and radiology of the areae gastricae. Gut 18:855–864 10. Sohn J, Levine M S, Furth E E et al 1995 Helicobacter pylori gastritis: radiographic findings. Radiology 195:763–767 11. Watambe H, Magota S, Shiiba S et al 1983 Coarse areae gastricae in the proximal body and fundus: a sign of gastric hypersecretion. Radiology 146:303–306 12. Cho K C, Gold B M, Printz D A 1987 Multiple transverse folds in the gastric antrum. Radiology 164:339–341 13. Meyers M A 1994 Dynamic radiology of the abdomen, 4th edn. Springer-Verlag, New York 14. Pattison P C, Combs M J, Marshall B J 1997 Helicobacter pylori and peptic ulcer disease: evolution to revolution to resolution. Am J Roentgenol 168:1415–1420 15. Marshall B J, Warren J R 1984 Unidentified curved bacilli in the stomach of patients with gastritis and peptic ulceration. Lancet i:1311–1315

16. NIH Consensus Conference 1994 Helicobacter pylori in peptic ulcer disease: NIH Consensus Development Panel on Helicobacter pylori in peptic ulcer disease. JAMA 272:65–69 17. IARC monographs on the evaluation of carcinogenic risks to humans: schistosomes, liver flukes and Helicobacter pylori 1994 International Agency for Research on Cancer, Lyon, pp 177–240 18. Wotherspoon A C, Ortiz-Hidalgo C, Falzon M R et al 1991 Helicobacter pylori-associated gastritis and primary B-cell gastric lymphoma. Lancet 338:1175–1176 19. Marshall B J, Barrett L, Prakash C et al 1990 Urea protects Helicobacter (Campylobacter) pylori from the bactericidal effect of acid. Gastroenterology 99:697–702 20. Cullen D J, Collins B J, Christiansen K J et al 1993 When is Helicobacter pylori infection acquired? Gut 34:1681–1682 21. Graham D Y, Lew G H, Klein P D et al 1992 Effect of treatment of Helicobacter pylori infection on long term recurrence of gastric or duodenal ulcer. Ann Intern Med 116:707–708 22. Gelfand D W, Dale W J, Ott D J 1994 The location and size of gastric ulcer: radiologic and endoscopic evaluation. Am J Roentgenol 143: 755–758 23. Thompson G, Stevenson G W, Somers S 1983 Distribution of gastric ulcers by double-contrast barium meal with endoscopic correlation. J Can Assoc Radiol 34:296–297 24. Kottler R E, Tuft R J 1981 Benign greater curve gastric ulcer: the ‘sumpulcer’. Br J Radiol 54:651–654 25. Amberg J R, Zboralske F F 1966 Gastric ulcers after 70. Am J Roentgenol 96:393–399 26. Haudek M 1910 Zur rontgenologischen Diagnose der Ulcerationen in der pars media des magnes. Munchen Med Wschr 57:1587–1591 27. Schumacher E C, Hampton A O 1956 Radiographic differentiation of benign and malignant ulcers. Ciba Clin Symp 8:161–171 28. Levine M S, Creteur V, Kressel H Y et al 1987 Benign gastric ulcers: diagnosis and follow-up with double-contrast radiography. Radiology 164:9–13 29. Laufer I, Hamilton J, Mullens J E 1975 Demonstration of superficial gastric erosions by double-contrast radiography. Gastroenterology 68:387–391 30. Ott D J, Gelfand E W, Wu S C, Kerr R M 1982 Sensitivity of singlevs double-contrast radiology in erosive gastritis. Am J Roentgenol 138:263–266 31. Graham D Y, Go M F 1993 Helicobacter pylori: current status. Gastroenterology 105:279–282 32. Gelfand D W, Ott D J 1997 Helicobacter pylori and gastroduodenal diseases: a minor revolution for radiologists. Am J Roentgenol 168:1421–1422 33. Joske R A, Finckh E S, Wood I J 1955 Gastric biopsy: a study of 1000 consecutive successful gastric biopsies. Q J Med 24:269–294 34. Elsborg L, Mosbech J 1979 Pernicious anaemia as a risk factor in gastric cancer. Acta Med Scand 206:315–318 35. Levine M S 2000 Inflammatory conditions of the stomach and duodenum. In: Gore R M, Levine M S (eds) Textbook of gastrointestinal radiology, 2nd edn. WB Saunders, Philadelphia, chapter 33 36. Cronan J, Burrell M, Trepeta R 1980 Aphthoid ulcerations in gastric candidiasis. Radiology 134:607–611 37. Agel N M, Tanner P, Drury A et al 1991 Cytomegalovirus gastritis with perforation and gastrocolic fistula formation. Histopathology 18:165–168 38. Megibow A J, Balthazar E J, Hulnick D H 1987 Radiology of nonneoplastic gastrointestinal disorders in acquired immune deficiency syndrome. Semin Roentgenol 22:31–41 39. Falcone S, Murphy B J, Weinfeld A 1991 Gastric manifestations of AIDS: radiographic findings on upper gastrointestinal examination. Gastrointest Radiol 16:95–98 40. Levine M S 1987 Crohn’s disease of the upper gastrointestinal tract. Radiol Clin North Am 225:79–91 41. Legge D A, Carlson H C, Judd E S 1970 Roentgenologic features of regional enteritis of the upper gastrointestinal tract. Am J Roentgenol 110:355–360

CHAPTER 31

42. Laufer I, Joffe N, Stolberg H 1977 Unusual causes of gastrocolic fistula. Gastrointest Radiol 2:21–25 43. Cohen W N 1967 Gastric involvement in Crohn’s disease. Am J Roentgenol 101:425–430 44. Marshak R H, Maklansky D, Kurzban J D et al 1982 Crohn’s disease of the stomach and duodenum. Am J Gastroenterol 77:340–343 45. Stempien S J, Dagradi A E, Reingold I M et al 1964 Hypertrophic hypersecretory gastropathy. Am J Dig Dis 9:471–493 46. Tan D T D, Stempien S J, Dagradi A E 1971 The clinical spectrum of hypertrophic hypersecretory gastropathy. Gastrointest Endosc 18:69–73 47. Rose C, Stevenson G W 1981 Correlation between visualization and size of the areae gastricae and duodenal ulcer. Radiology 139:371–374 48. Reese D F, Hodgson J R, Dockerty M B 1962 Giant hypertrophy of the gastric mucosa (Ménétrier’s disease): a correlation of the roentgenographic, pathologic, and clinical findings. Am J Roentgenol 88:619–626 49. Olmsted W W, Cooper P H, Madewell J E 1976 Involvement of the gastric antrum in Ménétrier’s disease. Am J Roentgenol 126:524–529 50. Schulman A, Morton P C G, Dietrich B E 1980 Eosinophilic gastroenteritis. Clin Radiol 31:101–104 51. Lee C M, Changchien C S, Chen P C et al 1993 Eosinophilic gastroenteritis: 10 years experience. Am J Gastroenterol 88:70–74 52. Klein N C, Hargrove R L, Sleisenger M H et al 1970 Eosinophilic gastroenteritis. Medicine 49:299–319 53. Goldberg H I, O’Kieffe D, Jenis E H et al 1973 Diffuse eosinophilic gastroenteritis. Am J Roentgenol 119:342–351 54. Balfe D M 1989 Eosinophilic gastritis. Am J Roentgenol 152:1322 55. Wehunt W D, Olmsted W W, Neiman H L et al 1976 Eosinophilic gastritis. Radiology 120:85–89 56. MacCarthy R L, Talley N J 1990 Barium studies in diffuse eosinophilic gastroenteritis. Gastrointest Radiol 15:183–187 57. Kussin S Z, Henry C, Navarro C et al 1982 Gas within the wall of the stomach: report of a case and review of the literature. Dig Dis Sci 27:949 58. Carlson H C, Breen J F 1986 Amyloidosis and plasma cell dyscrasias: gastrointestinal involvement. Semin Roentgenol 21:128–138 59. Eisenberg R L 1994 Miscellaneous abnormalities. In: Gore R N, Levine M S, Laufer I (eds) Textbook of Gastrointestinal Radiology, vol 1. W B Saunders, Philadelphia, p. 717 60. Balthazar E J 1998 Positional abnormalities of the stomach. In: Taveras J M, Ferrucci J T (eds) Radiology diagnosis-imaging-intervention, 4th edn, Lippincott-Raven, Philadelphia, chapter 20 61. Gerson D E, Lewicki A M 1976 Intrathoracic stomach: when does it obstruct? Radiology 119:257–264 62. Menuck L 1976 Plain film findings of gastric volvulus herniating into the chest. Am J Roentgenol 126:1169–1174 63. Whalen J P, Evans J A, Shanser J 1971 Vector principle in the differential diagnosis of abdominal masses: the left upper quadrant. Am J Roentgenol 113:104–118 64. Felson B, Berkman Y M, Hoyumpa A M 1969 Gastric mucosal diaphragm. Radiology 92:513–517 65. Palmer E D 1951 Collective review: gastric diverticula. Int Abstr Surg 92:417–428 66. Rabushka S E, Melamed M, Melamed J L 1968 Unusual gastric diverticula. Radiology 90:1006–1008 67. Treichel J, Gerstenberg E, Palme G et al 1976 Diagnosis of partial gastric diverticula. Radiology 119:13–18 68. Eisenberg R L 1994 Miscellaneous abnormalities. In: Gore R M, Levine M S, Laufer I (eds) Textbook of Gastrointestinal Radiology, vol 1. W B Saunders, Philadelphia, pp 717–741 69. Muhletaler C, Gerlock J, Goncharenko V et al 1979 Gastric varices secondary to splenic vein occlusion: radiographic diagnosis and clinical significance. Radiology 132:593–598 70. Swischuk L E 1967 Gastric varices presenting as ‘pseudotumors’ of the cardia. Am J Dig Dis 12:839 71. Belgrard R, Carlson H C, Payne W S, Cain J C 1964 Pseudotumoral gastric varices. Am J Roentgenol 91:751–756

• THE STOMACH

72. Carucci L R, Levine M S, Rubesin S E 1999 Tumorous gastric varices: radiographic findings in 10 patients. Radiology 212:861–865 73. Csos T, Meyers M A, Baltaxe H A 1972 Nonfundic gastric varices. Radiology 105:579 74. Gramm H F, Reuter K, Costello P 1978 The radiologic manifestations of diabetic gastric neuropathy and its differential diagnosis. Gastrointest Radiol 3:151–155 75. Burhenne H J 1971 Postoperative defects of the stomach. Semin Roentgenol 6:182–192 76. Burrell M, Curtis A M 1977 Sequelae of stomach surgery. CRC Crit Rev Diagn Imag 10:17–97 77. Hardy J H 1964 Problems associated with gastric surgery. Am J Surg 108:699–716 78. Jay B S, Burrell M 1977 Iatrogenic problems following gastric surgery. Gastrointest Radiol 2:239–257 79. Burhenne H J 1967 The retained gastric antrum: preoperative roentgenologic diagnosis of an iatrogenic syndrome. Am J Roentgenol 101:549–597 80. Poppel M H 1962 Gastric intussusceptions. Radiology 78:602–608 81. Szemes G C, Amberg J R 1968 Gastric bezoars after partial gastrectomy: report of five cases. Radiology 90:765–768 82. Moskowitz H 1974 Phytobezoars of the small bowel following gastric surgery. Radiology 13:23–26 83. Feldman F, Seaman W B 1972 Primary gastric stump cancer. Am J Roentgenol 115:257–267 84. Kobayashi S, Prolla J C, Kirsner J B 1970 Late gastric carcinoma developing after surgery for benign conditions. Am J Dig Dis 15:905– 912 85. Ming S C 1976 The adenoma-carcinoma sequence in the stomach and colon. II. Malignant potential of gastric polyps. Gastrointest Radiol 1:121–125 86. Gordon R, Laufer I, Kressel H Y 1980 Gastric polyps found on routine double-contrast examination of the stomach. Radiology 134:27–30 87. Iida M, Yao T, Waatanabe H et al 1984 Fundic gland polyposis in patients without familial adenomatosis coli: its incidence and clinical features. Gastroenterology 86:1437–1442 88. Tomosulo J 1971 Gastric polyps. Histologic types and their relationship to gastric carcinoma. Cancer 27:1346–1355 89. Lewin K J, Appelman H D 1996 Tumours of the esophagus and stomach. In: Atlas of Tumor Pathology, AFIP, Washington, DC 90. Rosai J 1995 Ackerman’s Surgical Pathology, 8th edn. Mosby, St Louis, MO, pp 645–647 91. Miettinen M, Sarlomo-Rikala M, Lasota J 1999 Gastrointestinal stromal tumors: Recent advances in understanding of their biology. Human Pathol 30:1213–1220 92. Salmela H 1968 Smooth muscle tumors of the stomach. Acta Chir Scand 134:384–391 93. Short W F, Young B R 1968 Roentgen demonstration of prolapse of benign polypoid gastric tumors into the duodenum, including dumbbell-shaped leiomyoma. Am J Roentgenol 103:317–320 94. Franquemont D W 1995 Differentiation and risk assessment of gastrointestinal stromal tumors. Am J Clin Pathol 103:41–47 95. Powell J, McConkey C C 1992 The rising trend in oesophageal adenocarcinoma and gastric cardia. Eur J Cancer Prev 1:265–269 96. Shirakabe H, Nishzawa M, Maruyama M, Kobayashi S 1982 Atlas of X-ray diagnosis of early gastric cancer. Igaku-Shoin, New York 97. Maruyama M, Baha Y 1994 Gastric carcinoma. Radiol Clin North Am 32:1233–1252 98. Montesi A, Graziani L, Pesaresi A et al 1982 Radiologic diagnosis of early gastric carcinoma by routine double-contrast examination. Gastrointest Radiol 7:205–215 99. Miller F H, Kochan M L, Talamonti M S, Ghahremani G G, Gore R M 1997 Gastric cancer. Radiologic staging. Radiol Clin North Am 35: 331–348 100. Hori S, Tsuda K, Marayama S et al 1992 CT of gastric carcinoma: preliminary results with a new scanning technique. Radiographics 12:257–268

651

652

SECTION 3

•

GASTROINTESTINAL IMAGING

101. Isaacson P G 2005 Gastrointestinal lymphoma: where morphology meets molecular biology. J Pathol 2:255–274 102. Menuck L S 1976 Gastric lymphoma: a radiologic diagnosis. Gastrointest Radiol 1:157–161 103. Sherrick D W, Hidgson J R, Dockerty M B 1965 The roentgenologic diagnosis of primary gastric lymphoma. Radiology 84:925–932 104. Gore R M, Ghahremani G G 1994 CT evaluation of the stomach. In: Body CT: categorical course syllabus. American Roentgen Ray Society, Reston, VA, pp 131–139

105. Davies P M 1978 Smooth muscle tumors of the upper gastrointestinal tract. Clin Radiol 29:407–414 106. Ming S C 1973 Tumors of the oesophagus and stomach. Armed Forces Institute of Pathology, Washington, pp 253–255 107. Dutta S K, Costa B S 1979 Umbilicated gastric polyposis: an indicator of metastatic gastric cancer. Am J Gastroenterol 791:598–600

CHAPTER

The Duodenum and Small Intestine

32

Nicholas Gourtsoyiannis, Nickolas Papanikolaou and Maria Daskalogiannaki

The duodenum • Anatomy and normal appearances • Radiological investigation • Peptic ulceration • Gastric heterotopia • Diverticula • Neoplasms • Other conditions The small intestine • Anatomy and normal appearances • Radiological investigation • Plain radiographs • Barium studies • Angiography • Nuclear medicine studies • Computed tomography • Ultrasound • Magnetic resonance imaging

The abnormal small intestine • Crohn’s disease • Coeliac disease • Neoplasms • Infections and infestations • Chronic radiation enteritis • Mechanical small intestinal obstruction • Acute mesenteric ischaemia • Diverticula and blind loops • Neuromuscular disorders • Nodular lymphoid hyperplasia and immunoglobulin deficiency • Whipple’s disease • Intestinal lymphangiectasia • Eosinophilic gastroenteritis • Mastocytosis • Amyloidosis • Acquired immune deficiency syndrome • Behçet’s disease • Graft-versus-host disease • Nonsteroid anti-inflammatory drug enteritis

THE DUODENUM ANATOMY AND NORMAL APPEARANCES The duodenum measures 20–30 cm in length1. It forms an incomplete circle surrounding the head of the pancreas and is described as having first, second, third and fourth parts. The first (superior) part contains the duodenal cap or bulb and passes superiorly, posteriorly and to the right before turning down to become the second part. Posteriorly it is devoid of peritoneum. The second (descending) portion passes down anterior to the right kidney and posterior to the transverse colon. Above and below the transverse colon it is covered with peritoneum. The duodenum turns to the left and passes horizontally in front of the spine as the third (horizontal) part before it ascends in front and to the left of the aorta as the fourth (ascending) part to end at the duodenojejunal flexure (ligament of Treitz).

At barium examination parallel, or nearly parallel, folds are seen passing upwards from the base of the duodenal cap. They are effaced when the hypotonic duodenum is distended by gas at double-contrast barium examination. The mucosa of the remainder of the duodenum is thrown into numerous folds that disappear on distension. The circular folds, termed valvulae conniventes, are permanent; they begin in the second part of the duodenum and extend throughout the small intestine. The ampulla of Vater can be seen in 65 per cent of patients during routine double-contrast barium examination and the accessory papilla of Santorini’s duct in 25 per cent2. The ampulla of Vater is recognized by its fold pattern: a hooded fold and a distal longitudinal fold are usual, and oblique folds are frequently present. The accessory papilla is sited about 10 mm proximal to the ampulla. On a prone view the ampulla lies on the medial wall and the accessory papilla on the anterior wall.

654

SECTION 3

•

GASTROINTESTINAL IMAGING

RADIOLOGICAL INVESTIGATION Barium studies The barium examination is the principal radiological technique for examining the duodenal lumen although the technique has largely been replaced by endoscopy. It is easier to detect abnormalities if the duodenum is relaxed. Hypotonia is produced by IV injection of 20 mg of hyoscine butylbromide (Buscopan) or 0.2 mg of glucagon. With the table horizontal, the patient is first turned onto their right side so that barium fills the duodenum. The patient is next rotated onto their left side and gas passes rapidly from the stomach into the duodenum.Then, with the patient in the right anterior oblique position, a number of double-contrast views of the duodenal cap and duodenal loop (Fig. 32.1) are taken. Excellent views can be obtained by elevating the head of the table slightly so that barium drains from the superior duodenal flexure into the second part of the duodenum. Double-contrast views of the duodenal cap and the duodenal loop are also taken with the patient prone or prone oblique, with the left side slightly raised.

Hypotonic duodenography Hypotonic duodenography is performed if the duodenal loop is the main area of interest. The tip of a duodenal catheter is positioned in the lower part of the descending duodenum and about 40 ml of barium suspension is injected. The smooth muscle relaxant is then injected intravenously. Air is injected through the catheter to distend the duodenum and radiographs of the duodenal loop are taken.

Water-soluble contrast studies Water-soluble contrast medium is used in patients with suspected perforation of the stomach or duodenum. The

Figure 32.1 Double-contrast view of the duodenal loop. The normal duodenal cap is also seen.

examination is best performed under fluoroscopic control; alternatively, a right decubitus radiograph is taken after a short interval.

Other imaging techniques Angiography can be invaluable in the diagnosis of acute duodenal haemorrhage, when endoscopy has failed to locate the bleeding site. Ultrasound (US) and CT are used to evaluate secondary involvement of the duodenum by malignant disease and CT may also be helpful in assessing the extent of duodenal neoplasms3,4.

PEPTIC ULCERATION On double-contrast barium examination, duodenal ulcer craters are shown as sharply defined, constant collections of barium (Fig. 32.2), sometimes with a surrounding zone of oedema or radiating folds. Anterior wall ulcers are normally shown best on the prone view.

Giant duodenal ulcers The term ‘giant duodenal ulcer’ is given to a benign ulcer crater with a radiographic diameter greater than 2 cm.The ulcer, because of its large size, may be mistaken for the duodenal bulb, a pseudodiverticulum, or a true diverticulum. Barium studies show giant duodenal ulcers as being constant in size and shape – often round or oval with a sharp outline (Fig. 32.3).The floor may be irregular, particularly when the ulcer is penetrating an adjacent organ.

Postbulbar ulceration Postbulbar ulcers are occasionally seen, mostly on the concave border of the second part or in the immediate postbulbar area. The ulcer is shown as a typical crater, frequently with spasm of the opposite wall. There may be narrowing of the lumen and thickening of the mucosal folds. In some cases scar formation may obscure the ulcer crater. Postbulbar ulcers usually fail to heal on medical treatment.

Figure 32.2 Duodenal ulceration. The duodenal cap is deformed and a moderate-sized ulcer crater is outlined with barium.

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.3 Giant duodenal ulcer. A large meniscus-shaped ulcer (arrows) is seen resembling the duodenal bulb. There was no change in the size or shape of the barium collection on screening or on other radiographic views. The ulcer was also seen at endoscopy.

Figure 32.4 Gastric heterotopia. Multiple small irregular filling defects of varying size are seen in the duodenal cap. (Courtesy of Dr J. Virjee.)

Complications of peptic ulceration The principal complications of duodenal ulceration are perforation, bleeding, stenosis and penetration of adjacent organs. Free perforation is usually diagnosed on the clinical and plain radiographic findings. Occasionally, a water-soluble contrast examination may be necessary to confirm the diagnosis. The perforation is sometimes localized or ‘walled-off ’ with marked deformity of the duodenum due to the adjacent inflammatory reaction. Bleeding caused by duodenal ulceration is diagnosed by endoscopy and/or angiography. Duodenal stenosis may become quite marked and result in obstruction. Barium examination in this situation will show an excessive amount of fluid in a dilated stomach with considerable delay in emptying.

GASTRIC HETEROTOPIA Gastric heterotopia is present in a small percentage of normal people5. Irregular filling defects, varying in size from 1 to 6 mm, are seen in the duodenal cap extending from the pylorus distally (Fig. 32.4). Gastric heterotopia should be differentiated from lymphoid hyperplasia of the duodenal bulb.

NEOPLASMS Benign neoplasms Benign neoplasms of the duodenum are uncommon and often symptomless. Brunner’s gland hyperplasia is seen as single or multiple polypoid lesions in the first part of the duodenum (Fig. 32.5), often with a characteristic cobblestone appearance. Patients usually present with typical symptoms of peptic ulceration. A single Brunner’s gland adenoma is occasionally seen. Adenomatous polyps feature as solitary, mostly sessile, polypoid, intraluminal filling defects on barium studies, or as soft tissue masses on CT. Villous adenomas exhibit a characteristic ‘cauliflower’ or ‘soap bubble’ appearance on barium studies, caused by the trapping of barium in the crevices between the multiple frond-like projections of the tumour. Benign lymphoid hyperplasia6 is an occasional finding shown as multiple small rounded filling defects of uniform size (Fig. 32.6). Other benign neoplasms of the duodenum include periampullary adenomas (Fig. 32.7), gastrointestinal stromal tumours (GIST), lipomas7 (Fig. 32.8), neurogenic tumours or hamartomas exhibiting the same features as in the small intestine.

DIVERTICULA

Malignant neoplasms Primary carcinoma

Duodenal diverticula are present in 2–5 per cent of barium studies; most are an incidental finding. They are usually in the descending part of the duodenum, with 85 per cent arising from the medial surface. Frequently they are in contact with the pancreas and may be embedded in its surface. Occasionally a diverticulum contains aberrant pancreatic, gastric, or other functioning tissue and is the site of ulceration, perforation, or gangrene. Symptoms may also develop due to the retention of food or a foreign body. Cholangitis or pancreatitis may result from the aberrant insertion of the common bile duct or pancreatic duct into an intraluminal diverticulum.

Carcinoma of the papilla of Vater is the type most frequently encountered, usually presenting with jaundice. Barium studies show an enlarged papilla of Vater with irregular borders, sometimes with spiculation and ulceration8. Non-papillary carcinomas of the duodenum are adenocarcinomas and usually present clinically as duodenal obstruction. Barium examination shows the neoplasm as an ulcerative, polypoid or annular lesion. On CT, primary carcinomas are seen mostly as focal masses with asymmetric mural thickening with varying degrees of luminal narrowing 7 (Fig. 32.9). Coincident adenopathy or hepatic metastases may be present. Other malignant primary neoplasms

655

656

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 32.5 Brunner’s gland hyperplasia. Multiple filling defects are seen in the duodenal cap. Biopsies obtained at endoscopy showed Brunner’s gland hyperplasia.

Figure 32.7 Periampullary adenoma. A (axial image) and B (coronal reconstruction). Enhanced CT with water used as oral contrast shows a sharply defined mass (arrows) within the lumen of the second duodenal part extending to the infrapapillary area. A biliary stent is also seen.

Figure 32.6 Lymphoid hyperplasia. Multiple small filling defects characteristic of lymphoid hyperplasia are shown on a double-contrast view of the duodenal cap. (Reproduced from Darrah E R, Nolan D J 1999 Hosp Med 60:10–18 with permission.)

occasionally encountered in the duodenum include GIST and lymphomas4.

Secondary involvement The duodenum may be invaded by malignant neoplasms from adjacent organs or be the site of metastases. Carcinoma and lymphoma of the stomach can spread directly across the pylorus to involve the duodenum.This is reported to occur in up to 40 per cent of lymphomas and 25 per cent of adenocarcinomas of the gastric antrum9.

Carcinoma of the head of the pancreas frequently causes changes in the duodenal loop. There may be widening of the duodenal loop, a double contour, irregularity of the inner border and stricturing or distortion of the valvulae conniventes.The reversed ‘3’ sign of Frostberg10 is often quoted as being characteristic, but is an infrequent finding (Figs 32.10, 32.11). Carcinoma of the tail of the pancreas may compress or invade the duodenum, resulting in mucosal destruction, the patients presenting with bleeding or obstruction. The duodenum may be invaded by carcinomas in other adjacent organs. Carcinoma of the colon, particularly of the hepatic flexure, may distort and invade the duodenal loop.The infiltrating lesion may show on barium studies as destruction of the mucosal pattern, stricturing (Fig. 32.12), a postbulbar ulcer with associated deformity, or a duodenocolic fistula. The distal half of the first part of the duodenum may be displaced, compressed, or infiltrated by carcinoma of the gallbladder. Carcinoma of the bile duct occasionally spreads to the duodenum and, rarely, enlarged neoplastic retroperitoneal lymph glands may also invade the duodenum. CT is particularly helpful for determining the origin and extent of malignant neoplasms that secondarily invade the duodenum4.

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.8 Lipoma. (A) A large intraluminal filling defect is seen occupying and distending the second part of the duodenum on a barium examination. (B) CT shows the lesion to be a well-defined, round mass with low attenuation values, characteristic of fat.

Figure 32.9 Primary duodenal adenocarcinoma. Enhanced CT with water used as oral contrast shows asymmetric mural thickening and encroachment of the lumen at the second portion of the duodenum (arrow). Dilatation of the common bile duct is also seen. The head of the pancreas (P) appears normal.

OTHER CONDITIONS Pancreatitis Duodenal ileus may be seen on plain radiographs in acute pancreatitis. Mucosal oedema, enlargement of the duodenal loop and enlargement of the papilla of Vater are characteristic findings on barium studies.

Crohn’s disease The duodenum is affected in about 4 per cent of patients with Crohn’s disease. The radiological appearances are similar to

Figure 32.10 Carcinoma of the pancreas. Barium examination of the duodenum shows the characteristic reversed ‘3’ sign of Frostberg with effacement and distortion of the mucosal pattern on the medial wall of the second portion of the duodenum in a patient with carcinoma of the head of the pancreas.

657

658

SECTION 3

•

GASTROINTESTINAL IMAGING

those in the more distal parts of the small intestine. The valvulae conniventes are frequently thickened. At a more advanced stage of the disease there may be strictures with eccentric or concentric narrowing7. Cobblestoning, asymmetry and skip lesions may be seen but fissure ulcers, sinuses and fistulae are uncommon in the duodenum. The disease may cause tubular narrowing of the antrum and proximal duodenum in continuity (Fig. 32.13), resulting in the ‘pseudo post-Billroth I’ appearance.

Tuberculosis Tuberculosis of the duodenum is rare. Barium studies show narrowing of the lumen, sometimes with destruction of the mucosa and ulceration, mostly involving the descending duodenum. Tuberculous mesenteric lymphadenitis, in the absence of intrinsic duodenal tuberculosis, may produce extrinsic pressure on the duodenum and cause obstruction.

Radiation damage Figure 32.11 Secondary invasion from pancreatic adenocarcinoma. Enhanced CT shows irregular mural thickening of the second part of duodenum with spiculations. The reverse ‘3’ Frostberg configuration is also evident.

Damage to the duodenum from radiotherapy is very uncommon. The radiological changes of radiation damage are seen mostly in the second part of the duodenum as ulceration, thickening of the mucosal folds and stricture formation.

Progressive systemic sclerosis In patients with neuromuscular disorders, particularly progressive systemic sclerosis and visceral myopathy, the duodenum is frequently involved. There may be dilatation, which is often more pronounced in the second, third and fourth parts. The dilated duodenum may be slow to empty and the grossly dilated, atonic organ may produce a sump effect.

Intramural haematoma The most common cause of intramural haematoma is blunt abdominal trauma. Intramural haematoma is usually seen on barium studies as a concentric obstructive lesion in the duodenum. Infiltration of blood and oedema may result in

Figure 32.12 Carcinoma of the colon invading the duodenum. There is marked symmetrical narrowing with mucosal destruction of the second portion of the duodenum in a patient who presented with abdominal pain and vomiting. At operation, carcinoma in an inverted caecum was found invading the duodenum.

Figure 32.13 Crohn’s disease. Marked irregular narrowing of the antrum and first portion of the duodenum, giving the ‘pseudo post-Billroth I’ appearance.

CHAPTER 32

thickening of the valvulae conniventes. CT shows the extent of the haematoma, seen as a large mixed attenuation mass, surrounding the duodenum (Fig. 32.14).

Traumatic rupture The most frequent site of rupture is at the junction of the second and third parts of the duodenum. CT is the primary imaging modality for assessment. Imaging findings include retroperitoneal air adjacent to the duodenum, extravasation of oral contrast in the retroperitoneum, edema in the duodenal wall and stranding in the peripancreatic fat7.

Superior mesenteric artery compression syndrome The superior mesenteric artery compression syndrome is an unusual form of high intestinal obstruction thought to be caused by narrowing of the normal angle between the aorta and superior mesenteric artery. During a symptomatic episode,

• THE DUODENUM AND SMALL INTESTINE

barium meal examination shows strong to-and-fro peristalsis and duodenal dilatation due to compression of the third part of the duodenum. Superior mesenteric artery compression is seen as a sharp cut-off in the right anterior oblique position, with the compression and proximal dilatation persisting in the prone position. CT may be useful by allowing the distance between aorta and superior mesenteric artery to be measured11.

Compression from aortic aneurysm Abdominal aortic aneurysms may compress the third part of the duodenum and they occasionally cause obstruction. The duodenum, when faintly opacified with oral contrast medium and stretched around an aneurysm, can be misinterpreted as a contained leak or as a patch of perianeurysmal inflammation.

Aortoduodenal fistula An aortoenteric fistula should always be suspected in patients who have undergone aortic graft surgery and present with gastrointestinal haemorrhage. Aortoenteric fistulas more often involve the duodenum, particularly the third part. A combination of endoscopy and CT may offer the best chance of detecting a fistula12.

Bouveret’s syndrome On rare occasions a gallstone may become impacted in the duodenal cap, a condition known as Bouveret’s syndrome. Barium examination shows the calculus as a radiolucent mass filling the duodenal cap with a thin coat of barium suspension between the periphery of the calculus and the wall of the duodenum.

Duodenal varices Figure 32.14 Intramural haematoma. CT shows a mass of mixed attenuation, characteristic of haematoma, surrounding the third portion of the duodenum (arrowheads).

Varices are encountered occasionally in the duodenal cap and loop. They occur mainly in patients with extrahepatic portal hypertension, but may occur in portal hypertension without evidence of extrahepatic obstruction.

THE SMALL INTESTINE ANATOMY AND NORMAL APPEARANCES The small intestine measures approximately 5 m in length and extends from the duodenojejunal flexure to the ileocaecal valve. It is attached by its mesentery to the posterior abdominal wall and this allows it to be mobile. The proximal two-fifths constitute the jejunum and the distal three-fifths the ileum. The jejunum lies mainly in the left upper and lower quadrants and the ileum in the lower abdomen and the right iliac fossa.The jejunal and ileal branches of the superior mesenteric artery provide the blood supply. Normally the small intestine is in a collapsed or partially collapsed state. Its calibre diminishes as it passes distally. During enteroclysis the maximum diameter of the jejunal loops is 4 cm and of the ileal loops 3 cm. The valvulae conniventes

have a circular configuration and are about 2 mm thick in the distended jejunum, becoming more spiral shaped and about 1 mm thick in the ileum13. They may be absent in the distended terminal ileum, resulting in a rather featureless outline.

RADIOLOGICAL INVESTIGATION Radiological evaluation of small bowel remains of particular importance in clinical practice, despite the advances in endoscopic techniques. Barium examination and computed tomography (CT) are the main radiological techniques used to show the various manifestations of small intestinal pathology14. In many patients both procedures are used in a complementary fashion. The choice of initial examination will depend on the

659

660

SECTION 3

•

GASTROINTESTINAL IMAGING

clinical issue in question15. Plain abdominal radiography and US have a limited role. Angiography and nuclear medicine studies can be of value in selected cases. Magnetic resonance imaging (MRI) is an evolving technique that holds great promise for the evaluation of the small intestine16.

The most widely used barium techniques are the followthrough examination and enteroclysis14. In selected cases, the peroral pneumocolon technique or a reflux examination are used to depict the terminal ileum.

The technique the current authors use is similar to Sellink’s method13,17,21. A 10 French gauge radio-opaque Nolan tube (William Cook Europe A/S) is passed via the nasogastric route to the duodenum so that its tip lies at or just distal to the duodenojejunal flexure, or preferably 5–10 cm into the jejunum. A total of 800–1200 ml of barium suspension diluted to 20 per cent weight/volume is infused, using a pump, at about 75 ml min-1. Radiographs of the barium-filled jejunum and ileum are then taken (Fig. 32.15). Compression is an important part of the examination technique17. It is necessary for the patient to have a clean colon and take a low-residue diet, increased fluid intake and cathartics on the day before the examination. Cathartics are not given to patients with known Crohn’s disease, severe diarrhoea, an ileostomy, or intestinal obstruction. Enteroclysis is not widely used because of perceived patient discomfort. The mean effective patient dose during enteroclysis has been calculated to be 1.5 mSv22, which compares favourably with the radiation dose received during routine abdominal CT or other barium examinations.

Barium follow-through examination

Modifications of enteroclysis technique

The barium follow-through is performed following a barium meal examination of the oesophagus, stomach and duodenum. As the barium column progresses through the small intestine, radiographs of the abdomen are taken at intervals. When the barium column reaches the caecum, views of the terminal ileum are obtained17. It takes from 2–6 h for the head of the barium column to reach the caecum.

A double-contrast method using barium suspension followed by an aqueous suspension of methylcellulose is used in a number of centres23. An air-contrast method is also recommended by Ekberg24. Detailed mucosal changes, particularly small ulcers, are well shown by the air-contrast method. However, as air does not distend the intestine as well as dilute barium; sinuses and fistulae are often not demonstrated and stenoses may be overlooked25.

PLAIN RADIOGRAPHS Patients who present acutely with suspected perforation or obstruction of the small intestine are investigated initially with plain abdominal radiographs.

BARIUM STUDIES

Small bowel meal The small bowel meal is a modification of the follow-through18. It is performed as a specific examination of the small intestine, separate from the barium examination of the upper gastrointestinal tract. Five hundred ml of a 30–40 per cent weight/ volume suspension of barium sulphate is given to the patient, who is encouraged to drink it as fast as possible. Cold water is used in the preparation of the suspension to stimulate gastric emptying and reduce transit time. Full-length prone radiographs are exposed at 10, 30 and 60 min. All segments of the small intestine are examined fluoroscopically and compression is applied.

Peroral pneumocolon examination Excellent views of the terminal ileum and caecum can be obtained by giving barium orally and, when the head of the barium column has reached the ascending colon, introducing air per rectum19.

Enteroclysis Enteroclysis has established its role as an accurate technique to examine the small intestine with a reported sensitivity of 93.1 per cent and a specificity of 96.9 per cent for detecting disorders accounting for the patients’ presenting symptoms20. Barium introduced directly into the small intestine gives excellent visualization by challenging the distensibility of intestinal loops, thereby making it easier to identify the presence of morphological abnormalities.

Figure 32.15

Normal enteroclysis examination.

CHAPTER 32

ANGIOGRAPHY Selective visceral angiography of the small intestine can detect the site of obscure bleeding when extensive barium studies and endoscopy are negative. It is also used to determine the bleeding site in patients who present with massive acute bleeding from the small intestine. In acute active bleeding angiography is successful when the rate of blood loss exceeds 0.5 ml min−1. Acute lower gastrointestinal bleeding is, however, frequently intermittent rather than continuous, resulting in a high rate of negative angiographic examinations.

NUCLEAR MEDICINE STUDIES Nuclear medicine studies are useful alternative and adjunctive methods in the investigation of small bowel pathology. Leukocyte scintigraphy and FDG–PET are helpful in the diagnosis of suspected inflammatory bowel disease and/or the assessment of current disease activity. Older techniques, like pertechnetate scintigraphy for the detection of a Meckel’s diverticulum have yet to be surpassed by more modern imaging methods. Scintigraphy with labelled red cells remains a useful examination for localization of intestinal bleeding. Carcinoid tumours of the small bowel may be elegantly demonstrated by scintigraphy and PET techniques are already proving useful in small bowel oncology26.

Radionuclide imaging of Meckel’s diverticulum Radionuclide scintigraphy, using 99mTc pertechnetate, is a wellestablished non-invasive technique for identifying a Meckel’s diverticulum that contains gastric mucosa, as this agent is concentrated in the mucus-secreting cells and the parietal cells of the gastric mucosa in both the stomach and the diverticulum. The technique is more accurate in the paediatric age group than in adults15.

COMPUTED TOMOGRAPHY CT is becoming increasingly important in evaluating mural and extramural lesions and in assessing mesenteric involvement and ancillary intra-abdominal findings associated with inflammatory or neoplastic small intestinal diseases28. Multidetector CT and 3D imaging have improved CT evaluation of the intestine and mesenteric vasculature. For routine imaging, opacification of the intestinal lumen is achieved using orally administered positive contrast material such as iodinated or dilute barium solutions, starting approximately 1 h prior to the examination.Water is gaining widespread acceptance as an oral agent and may be used when a detailed CT study of the small bowel or mesenteric vessels is required, since it allows for better visualization of the bowel wall and does not interfere with 3D angiography. However, it has the disadvantage of rapid emptying, resulting in suboptimal distension of distal loops. IV contrast administration is essential for a comprehensive CT examination. Dual-phase CT may be helpful in selected cases. CT enteroclysis and CT enterography have been proposed for a detailed assessment of the small bowel. CT enteroclysis

• THE DUODENUM AND SMALL INTESTINE

combines the advantages of enteral volume challenge and crosssectional imaging with reformatting to depict mural and extramural manifestations of disease, providing complete visualization of the gastrointestinal tract without superimposition of intestinal loops29. Contrast medium is administered into the small intestine through a nasojejunal tube under fluoroscopic guidance, and CT images are obtained following distension of bowel loops. Methylcellulose solution as a neutral luminal contrast agent, watersoluble contrast medium or a dilute barium solution as positive luminal contrast can be used. CT enterography allows non-invasive global assessment of small bowel30. A high-density agent such as iodinated, dilute barium suspensions30 or a low-density agent such as Mucofalk31 (Dr F. Pharma, Freiburg, Germany) is given perorally, resulting in adequate small bowel distension. For both CT techniques, IV administration of a spasmolytic drug is essential, before image acquisition. Neither CT enteroclysis nor CT enterography are useful when mucosal detail is required. They are of value in patients with suspected complications of Crohn’s disease, intermittent small bowel obstruction, neoplasms and unexplained anaemia or gastrointestinal bleeding.

ULTRASOUND In clinical practice, US plays a rather limited role in the diagnosis and management of small bowel disorders. High-resolution ultrasonography is a quick, noninvasive method for assessing inflammatory bowel disease32. Small intestinal obstruction can also be recognized and primary intestinal neoplasms may be identified using a dedicated US technique33.

MAGNETIC RESONANCE IMAGING MR imaging (MRI) provides excellent soft-tissue contrast and three-dimensional imaging capabilities, which are of importance when studying the small intestine.A state-of-the-art MRI examination of the small intestine should comprise: adequate bowel distension, homogeneous lumen opacification, increased conspicuity of the bowel wall, demonstration of the mesenteries, information on bowel motility, ability to obtain dynamic post contrast images, high-contrast resolution, sufficient spatial resolution to evaluate subtle mucosal lesions, images free from artifacts – especially motion artifacts – and rapid acquisition times. The above can be provided by a comprehensive MR enteroclysis (MRE) examination protocol, which includes small bowel intubation, administration of a biphasic contrast agent, heavily T2W (T2W) single-shot turbo spin-echo (SSTSE) images for MR fluoroscopy and for monitoring the infusion process, T2W imaging employing half-Fourier acquisition single-shot turbo spin-echo (HASTE) and true FISP sequences (Fig. 32.16), and dynamic T1-weighted (T1W) imaging using a post-gadolinium FLASH sequence with fat suppression (Fig. 32.17). This protocol can provide anatomic demonstration of the normal intestinal wall, identification of wall thickening and neoplasms, lesion characterization and/or evaluation of disease activity, assessment of the extent of exoenteric/mesenteric disease and information concerning intestinal motility34.

661

662

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 32.16 Coronal true FISP image demonstrating small bowel at its entire length. The use of an iso-osmotic water solution as an intraluminal contrast agent results in homogeneous opacification of the bowel lumen. Note the increased conspicuity of the normal bowel wall due to the high-resolution capability and total absence of motion.

Figure 32.17 Two-dimensional FLASH coronal image with fat saturation, after IV administration of gadolinium and glucagon. Normal small bowel wall and valvulae conniventes are demonstrated with excellent conspicuity.

THE ABNORMAL SMALL INTESTINE CROHN’S DISEASE This chronic, progressive, transmural disease may affect any part of the gastrointestinal tract, but mostly involves the small intestine. The extent of involvement varies considerably but the terminal ileum is almost always affected. In patients with previous right hemicolectomy for Crohn’s disease, the site of anastomosis in the small bowel represents the most common site of recurrence. Abdominal pain and diarrhoea, often accompanied by weight loss, are the most frequent presenting symptoms. Patients may present with an ‘acute abdomen’, indistinguishable from acute appendicitis. Other presenting symptoms are anaemia, retardation of growth, anorexia and weight loss. Acute intestinal obstruction due to stenosis is occasionally the presenting symptom, as are fistulae, particularly fistula in ano.

Radiological appearances The most characteristic feature of Crohn’s disease of the small intestine is the variety of its radiological appearances and the multiplicity of imaging features often seen in the majority of patients35 (Table 32.1). Aphthoid ulcers, which are a characteristic feature, are usually visualized as small collections of barium with surrounding radiolucent margins due to oedema. Fissure ulcers are seen in profile, may penetrate deep into the thickened intestinal wall and may lead to abscess formation at their base and/or to the development of sinuses and fistulae. Fistulae pass to adjacent loops of ileum, the caecum, the sigmoid colon or the urinary bladder and occasionally to the skin or the vagina. Longitudinal ulcers (Fig. 32.18) running along the mesenteric border of the ileum are a characteristic but infrequent sign of the

Table 32.1 RADIOLOGICAL SIGNS OF CROHN’S DISEASE Ulceration Discrete ulcers Fissure ulcers Longitudinal ulcers Cobblestoning Thickening of the valvulae conniventes Stenosis Dilatation proximal to stenosis Asymmetrical involvement Skip lesions Inflammatory polyps (pseudopolyps) Featureless outline Thickened wall Enlarged ileocaecal valve Gross distortion Adhesions

disease process. Cobblestoning (Fig. 32.19) is fairly common and is caused by a combination of longitudinal and transverse ulceration, separating intact portions of mucosa. Narrowing of the intestinal lumen is frequent, as are strictures, which may be short, long, single or multiple, the latter being virtually diagnostic of Crohn’s disease. Solitary strictures (Fig. 32.20) are a common finding; they may be accompanied by proximal dilation and in the absence of additional evidence of Crohn’s disease have to be differentiated from other causes of stricture formation (Table 32.2). Discontinuous involve-

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.19 Cobblestoning of the terminal ileum, thickening of the wall of the terminal ileum, and an enlarged ileocaecal valve in Crohn’s disease. Figure 32.18 Longitudinal ulcer in Crohn’s disease. There is a long, longitudinal ulcer involving the mesenteric border of the terminal ileum.

ment of the intestinal wall is shown either as skip lesions or asymmetry. Asymmetrical involvement of the intestinal wall produces characteristic ‘pseudodiverticula’, which represent small patches of normal intestine in an otherwise severely involved segment. Inflammatory polyps (pseudopolyps) are seen in Crohn’s disease as small, discrete round filling defects, but are not a frequent finding. A smooth, featureless outline replacing the normal mucosal pattern, without any significant changes in the calibre of the lumen may occasionally be seen. Thickening of the wall of the diseased bowel is shown radiologically as displacement of the adjacent barium-filled loops. The valvulae conniventes are blunted, flattened, thickened and distorted or straightened in approximately 50 per cent of patients with Crohn’s disease35.These early changes are due to hyperplasia of the lymphoid tissue causing obstructive lymphoedema. Cross-sectional imaging modalities offer an important complementary diagnostic perspective in patients with Crohn’s disease, because of their ability to directly image the intestinal wall and surrounding mesentery and therefore to determine the extramucosal extent and spread of the disease. Computed tomography has an expanding role in evaluating mural, serosal and mesenteric abnormalities, as well as intestinal wall thickening, abscess formation and mesenteric lymphadenopathy. On CT, the thickened bowel wall may have a homogeneous or stratified appearance (Fig. 32.21A). Mural stratification is often seen in active lesions, particularly after IV

Figure 32.20 Long stricture in Crohn’s disease. A long segment of narrowing is seen in the ileum just proximal to the site of an ileocolic anastomosis in a patient who had undergone a previous resection for Crohn’s disease.

663

664

SECTION 3

•

Table 32.2

GASTROINTESTINAL IMAGING

STRICTURES OF THE SMALL INTESTINE

Crohn’s disease Tumors Primary carcinoma Carcinoid Lymphoma (including malignant histiocytosis) Invasion from other organs Metastases Tuberculosis Actinomycosis South American blastomycosis Strongyloidiasis Anisakiasis Radiation damage Ischaemia Intramural haemorrhage Diverticular mass Appendix mass Endometriosis Eosinophilic gastroenteritis Idiopathic ulcerative enteritis Behçet’s disease Nonsteroidal anti-inflammatory drugs

contrast administration. Transmural inflammation of the small bowel in Crohn’s disease usually involves the adjacent mesentery. Fibrofatty proliferation of the mesentery, resulting in increased CT attenuation, is the most common cause of separation of bowel loops in Crohn’s disease. Associated findings include mild reactive lymphadenopathy and mesenteric hypervascularity (Fig. 32.21B). With more advanced disease, perforation of intestine leads to a mesenteric phlegmon or interloop abscess. MRE is an emerging technique for the evaluation of small bowel in patients with Crohn’s disease. Characteristic lesions such as bowel wall thickening (Fig. 32.22), linear and fissure ulcers (Fig. 32.23) and cobblestoning are accurately depicted, especially when using the true FISP sequence. MRE is comparable with conventional enteroclysis in assessing the number and extent of involved small bowel segments and in demonstrating luminal narrowing and/or prestenotic dilatation36. MRE has a clear advantage over conventional enteroclysis in demonstrat-ing extramural manifestations and/or complications of Crohn’s disease, including fibrofatty proliferation (Fig. 32.24), mesenteric lymphadenopathy (Figs 32.25, 32.26), vascular engorgement, sinus tracts, fistulae (Fig. 32.27) and abscesses. It may be helpful in distinguishing inflammatory from fibrotic strictures. Disease activity can be also accurately assessed37.

Figure 32.21 Crohn’s disease. (A) Coronal reconstruction image of CT enterography shows thickened distal ileal loops and mural stratification resulting in a ‘target’ appearance (arrows). Prestenotic dilatation is also seen. (B) A coronal, three-dimensional projection of the same patient showing the vascular engorgement (arrows) of an involved ileal loop (comb sign).

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.22 True FISP coronal image of a patient with Crohn’s disease of the terminal ileum. Moderate luminal narrowing and mural thickening is shown. Black boundary artifact (arrow) can be easily differentiated from the thickened bowel wall, exhibiting moderate signal intensity.

Figure 32.25 Coronal true FISP spot view in a patient with Crohn’s disease showing multiple mesenteric lymph nodes (arrows) of variable size. Bowel wall thickening of ileal loops is also present.

Figure 32.23 34-year-old male patient with active Crohn’s disease (Crohn’s Disease Activity Index (CDAI) = 244). Coronal true FISP spot view demonstrates luminal narrowing and wall thickening in a segment of distal ileum. A fissure ulcer (arrow) penetrating the thickened wall and increased mesenteric vascularity are also disclosed (With permission from Gourtsoyiannis N C, Papanikolaou N, Karantanas A 2006 Magnetic resonance imaging evaluation of small intestinal Crohn’s disease. Best Pract Res Clin Gastroenterol 20:137–156.)

Figure 32.26 27-year-old patient with active Crohn’s disease (CDAI = 312). Coronal 2D FLASH image after IV gadolinium administration, showing multilayered mural enhancement of distal ileal loops (small arrow) and multiple enhancing mesenteric lymph nodes (arrowheads). Vascular engorgement is also present.

COELIAC DISEASE Figure 32.24 Extensive fibrofatty proliferation of the mesentery, accompanying involved ileal segments is demonstrated on a coronal true FISP image (With permission from Gourtsoyiannis N, Papanikolaou N, Rieber A, Brambs H J, Prassopoulos P 2002 Evaluation of the small intestine by magnetic resonance imaging. In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin pp 157–170.) (With kind permission of Springer Science and Business Media.)

Coeliac disease is a disorder of small intestinal mucosa caused by gluten toxicity, in genetically susceptible individuals, mostly children or young adults. Patients usually present with clinical symptoms of malabsorption, such as diarrhoea, weight loss, steatorrhoea, malnutrition, anaemia and abdominal pain. Patients with coeliac disease are at greater risk of developing malignant neoplasms, particularly lymphomas.

665

666

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 32.27 Four consecutive coronal true FISP images in a patient with fistulizing/perforating subtype of Crohn’s disease. Multiple entero-enteric fistulae forming a ‘star’-like sign are nicely demonstrated (arrow).

The diagnosis is established by demonstrating the abnormal villous pattern on specimens obtained at peroral jejunal biopsy. Radiological examination should be reserved for patients with normal jejunal biopsy, those with a suspected complication such as lymphoma, or those whose symptoms fail to respond to a gluten-free diet. Radiological appearances on barium studies include dilatation of bowel loops, increased intestinal fluid, straightened and thickened valvulae conniventes in the jejunum and presence of numerous mucosal folds in the ileum (jejunization). Transient nonobstructive intussusception may be also seen. CT is especially useful in clinically unsuspected adult patients, in whom it may demonstrate bowel dilatation and fluid excess, bowel wall thickening, a jejunoileal fold pattern reversal, transient small bowel intussusception, extraintestinal findings like benign mesenteric lymphadenopathy of low attenuation, cavitation of mesenteric lymph nodes and complications such as lymphoma38. MRI may be useful, especially in paediatric patients in whom avoidance of ionizing radiation is particularly important. Like CT, it can identify jejunization of the ileum, jejunoileal fold pattern reversal, mural thickening due to submucosal edema and varying degrees of inflammation, transient intussusception,

benign mesenteric lymphadenopathy, small-volume ascites and vascular engorgement39.

NEOPLASMS Primary neoplasms of the small intestine account for only 3–6 per cent of gastrointestinal neoplasms. The clinical presentation is often nonspecific, making their detection particularly challenging. The reported mean time period between the onset of symptoms and diagnosis is approximately 3 years for benign tumours and almost 2 years for such malignant neoplasms40. However, the radiological features of these tumors as shown by enteroclysis and CT correlate very well with the morphological changes seen in the gross pathology specimens41. The ability to accurately image small intestinal neoplasms represents a major improvement in their diagnosis and management.

Benign neoplasms Benign small intestinal neoplasms account for approximately 0.5–2 per cent of all gastrointestinal tract neoplasms. There are more than 12 histopathological types, amongst which adenoma and leiomyoma are the most common and the only two

CHAPTER 32

with definite malignant predisposition42. Lipomas, vascular and neurogenic tumours, hamartomas and heterotopias are less frequently encountered. Gastrointestinal stromal tumours (GISTs) are a unique type of mesenchymal tumours that may occur anywhere in the gastrointestinal tract. They exhibit a wide spectrum of clinical behaviour from benign, small, incidentally depicted nodules to frank malignant lesions43. The best defining feature of GISTs is their expression of KIT (CD 117), a tyrosine kinase growth factor receptor. Immunoreactivity for KIT distinguishes GISTs from other mesenchymal tumours, like leiomyomas, leiomyosarcomas, schwannomas and neurofibromas and determines the appropriateness of KIT inhibitor therapy44. However, radiological criteria to separate GISTs from other mesenchymal, nonepithelial, tumours have not yet been fully developed45. Leiomyoma is the most common symptomatic benign neoplasm. Patients usually present with acute bleeding. In cases of intraluminal tumours a broad-based, smooth, round or semilunar filling defect is usually seen, whereas with extralu-minal masses there is displacement and distortion of neighbouring bowel loops. Bidirectional or dumb-bell tumours combine features of both. Tenting deformity of the intestinal wall, ulceration and signs of intussusception may also encountered46. Ancillary CT findings include a round or semilunar, homogeneous soft-tissue mass associated with the intestinal wall, showing marked homogeneous or rim contrast enhancement and absence of mesenteric changes or metastasis. A peripheral crescent-shaped necrosis of the mass, shown on contrastenhanced CT, US or MRI is additionally suggestive of the diagnosis47. On angiography leiomyoma appears as a welldefined, round or lobulated hypervascular mass. Adenomatous polyps and villous adenomas are two terms widely used to designate the growth pattern and gross morphology of adenomas. Adenomatous polyps are often symptomless. On barium studies they appear as small, smooth, round or oval intraluminal filling defects, and are often solitary and sessile.When multiple, they usually affect a single segment, are of different sizes and may be pedunculated.Villous adenomas are usually > 3 cm in size, are invariably broad based and present radiographically as lobulated cauliflower-like filling defects, exhibiting multiple radiolucent striations, interspersed with frond-like projections. Hamartomatous polyps are a developmental anomaly and may be present in large numbers in the small intestine of patients with the Peutz-Jeghers’ syndrome. Recurrent abdominal pain caused by intussusception is the most common clinical presentation. The polyps are shown on barium studies as multiple round or lobulated filling defects and are often pedunculated; intussusception is frequently demonstrated (Fig. 32.28). Lipoma is the third most common benign neoplasm. On barium studies, lipomas appear as sharply marginated, solitary, sessile, intraluminal filling defects, averaging 3–4 cm in size. They are easily deformed by peristalsis or compression during fluoroscopy. On CT, they feature as smooth ovoid masses, exhibiting attenuation values of fat48.

• THE DUODENUM AND SMALL INTESTINE

Figure 32.28 Peutz-Jeghers syndrome. Jejunal intussusception is seen resulting from the presence of multiple hamartomatous polyps in a patient with Peutz-Jeghers syndrome.

Malignant neoplasms The most common primary malignant neoplasms of the small intestine are adenocarcinoma, carcinoid tumour, lymphoma and malignant GIST.

Carcinoid tumour Carcinoid tumours of the small intestine are regarded as lowgrade malignant neoplasms. Almost 90 per cent of them are located distally and they may be multiple in approximately one-third of cases. Carcinoid arises within the basal portion of the mucosa and often extends into the submucosa, infiltrating the intestinal wall and serosa. Less frequently intraluminal growth results in a polypoid lesion. Invasion of the muscular layers of the intestinal wall may lead to fibrosis of the surrounding submucosa, the mesentery and mesenteric vessels48. The primary tumour rarely produces symptoms, largely because of its small size and its deep mucosal location. Abdominal pain, diarrhoea or an abdominal mass may be encountered, but gastrointestinal haemorrhage is extremely rare. The carcinoid syndrome is seen in one-third of jejunoileal carcinoids that have metastasized to the liver. The radiological features of carcinoid tumour are nonspecific and reflect the stage of evolution of the pathological process at

667

668

SECTION 3

•

GASTROINTESTINAL IMAGING

the time of examination. They may be those of: (A) the primary lesion, appearing as solitary or multiple, round, smoothly outlined intraluminal filling defects (Fig. 32.29); (B) those of a secondary mesenteric mass, causing stretching, rigidity and fixation of ileal loops; (C) those due to interference with the ileal blood supply, resulting in thickening of valvulae conniventes and chronic ischaemic intestinal changes; or (D) those of fibrosis associated with tumour spread, presenting as sharp angulation of a loop or a stellate, spoke-like arrangement of adjacent loops49. Carcinoid tumours are best recognized on CT on the basis of mesenteric findings. Secondary mesenteric changes include a discrete, unifocal, soft tissue mass, usually associated with linear strands radiating into surrounding fat (Fig. 32.30),

while displacing adjacent intestinal loops. Hypervascular liver metastases, usually hypodense on unenhanced images, mesenteric lymphadenopathy and dystrophic calcification in metastatic nodes, liver metastases or in the mesenteric mass may be also encountered48.

Lymphoma Lymphoma represents 20 per cent of primary small intestinal malignancies. The clinical presentation depends on whether bowel involvement is primary or secondary, and on the features of any associated disorders, such as adult coeliac disease, immunoproliferative disease or immunodeficiency syndromes. The distal or terminal ileum is most commonly involved. Mediterranean-type lymphoma, complicating immunoproliferative small intestinal (alpha-chain) disease is an exception, as it affects the duodenum and proximal jejunum. Characteristic radiological signs include luminal narrowing with mucosal destruction, occasional shouldering of the margins and stricture formation, broad-based ulceration, cavitation (Fig. 32.31), thickening of the valvulae conniventes, discrete intraluminal filling defects and an extraluminal mass. Small nodules or polyps may be seen throughout the small intestine. Focal ‘aneurysmal’ dilatation, caused by extensive lymphomatous invasion of the muscle layers and neural plexuses is characteristic but only occasionally seen50. Ileo-ileal fistula formation may result from a cavitating mass invading adjacent ileal loops. CT or MR appearances of intestinal lymphoma are also variable and may be categorized as aneurysmal, nodular, ulcerative and constrictive, while mesenteric involvement will usually feature as ill-defined confluent masses encasing intestinal loops, mesenteric adenopathy (Fig. 32.32), a conglomerate

Figure 32.29 Carcinoid tumour. A round, well-defined, intraluminal filling defect (arrow) is seen in the distal ileum of a patient who presented with symptoms of intermittent obstruction but without any manifestations of the carcinoid syndrome.

Figure 32.30 Carcinoid tumour. CT shows a carcinoid mass (arrow) with a characteristic stellate radiating pattern and thickening of the adjacent intestinal wall.

Figure 32.31 Lymphoma. A large mass is seen infiltrating and compressing the pelvic loops of ileum. There is also a large cavitating ulcer that has eroded a number of loops of intestine producing an ileoileal fistula. The patient presented with abdominal pain, weight loss, anaemia and a palpable mass.

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.32 Patient with small intestinal lymphoma. The coronal true FISP image (A) shows significant mural thickening of the terminal ileum (arrow) and multiple, small and large, in size, mesenteric lymph nodes (dotted arrows). Diffusion-weighted image with a b-value of 1000 s/mm2 renders the lesions with high signal intensity, compatible with the presence of restricted water diffusion pattern due to hypercellularity.

mass of mesenteric/retroperitoneal tissue, or a ‘sandwich- like’ complex, due to encasement of vessels from enlarged, confluent mesenteric lymph nodes.

Adenocarcinoma Adenocarcinoma is the most common malignant neoplasm of the small intestine. It is a solitary lesion mostly located in the proximal small intestine. It is almost always symptomatic, with a nonspecific clinical presentation and a dismal prognosis, mainly due to late diagnosis. Its appearance on enteroclysis includes an annular constricting lesion, a filling defect, a polypoid and/ or ulcerated mass, or a combination of the above48. Infiltrative adenocarcinoma, featuring as circumferential narrowing with mucosal destruction and shouldering of the margins (Fig. 32.33), is the most common type. Adenocarcinoma appears on CT as mural thickening, not exceeding 1.5 cm, compromising the lumen either concentrically or asymmetrically.The tumour mass may be homogeneous or heterogeneous and shows moderate contrast enhancement. Infiltration of the mesentery is seen with advanced disease and associated lymphadenopathy is found in almost 50 per cent of patients.

Figure 32.33 Carcinoma. A spot compression view of a segment of jejunum shows a tight constricting lesion with mucosal destruction and shouldered margins.

Malignant GIST (leiomyosarcoma) Leiomyosarcoma, now termed malignant gastrointestinal stromal tumour (GIST), grows mainly extraluminally and eccentrically and often shows degenerative changes, such as necrosis, haemorrhage, calcification, fistula or secondary infection. It often presents with abdominal pain, bleeding or a palpable abdominal mass. The main radiological feature is a large, inhomogeneous extrinsic mass, displacing or distorting adjacent loops of intestine. It may be associated with ulcer-

ation, cavitation or fistula formation48. Although malignant, GISTs are usually larger than other malignant small intestinal neoplasms; they tend not to cause obstruction and are rarely accompanied by regional lymph node enlargement (Fig. 32.34). CT or MR examination may demonstrate hepatic or peritoneal metastases.

669

670

SECTION 3

•

GASTROINTESTINAL IMAGING

tumours with ulcerations, or mesenteric masses encasing or obstructing bowel loops may develop.

INFECTIONS AND INFESTATIONS Tuberculosis

Figure 32.34 Malignant GIST. CT shows a large, heterogeneous, soft tissue mass, merely hanging from an ileal segment. Absence of lymph node enlargement.

Secondary neoplasms Secondary neoplasms involve the small intestine by direct invasion from adjacent organs, lymphatic extension, peritoneal seeding and embolic metastases. More than one mechanism of spread may be encountered in the same patient. Direct invasion of the small intestine from primary neoplasms of the ovary, colon, prostate, uterus, or kidney indicates that an aggressive neoplasm has broken through fascial planes. The characteristic radiological appearance is of a mass invading the adjacent intestine, often over a considerable length, with mucosal destruction, narrowing of the lumen and obstruction. Tethering of the mucosal folds may be a conspicuous feature. Lymphatic spread plays a minor role in the spread of neoplasms to the small intestine; spread of caecal carcinoma to the terminal ileum is a classic example. Intraperitoneal seeding of abdominal neoplasms frequently localizes in the right lower quadrant at the distal mesentery. Stasis in the lower recess of the small intestinal mesentery results in the deposition and growth of secondary deposits. As a result the ileal loops in the right lower quadrant become separated with angled tethering of the mucosal folds, and the narrowed loops may align in a parallel configuration described as ‘palisading’. Blood-borne metastases to the small intestine from extraabdominal sites are uncommon. The most common primary tumours that metastasize to the small bowel are bronchogenic carcinoma, melanoma, breast carcinoma and renal cell carcinoma. Metastatic carcinomas are featuring as small, focal, nodular or infiltrating obstructive lesions, whereas metastatic melanomas appear as multiple submucosal polypoid lesions with central ulceration or cavitation, giving a ‘bull’s eye’ or ‘target’ appearance, on barium studies51. CT features are protean and often similar to primary intestinal tumours. Local spread and intraperitoneal seeding usually manifest with detectable CT mesenteric and intraperitoneal changes, including mesenteric infiltrates, peritoneal implants and omental ‘caking’. Blood-borne metastases feature as one or several intramural or intraluminal soft-tissue masses, which may lead to intussusception. At later stages, large infiltrating

Intestinal tuberculosis is rare in Europe. Patients usually present with abdominal pain, fever, weight loss, diarrhoea, intestinal obstruction and, rarely, with bowel perforation. Lesions are often multiple and usually involve more than one site. The ileum is most commonly affected, particularly the terminal ileum and the ileocaecal junction (Fig. 32.35). Discrete ulcers, usually transverse and circumferential, mucosal fold thickening and stricture formation are the main radiological features. In ileocaecal tuberculosis the terminal ileum is narrowed and thickened, the ileocaecal valve becomes rigid, irregular, gaping and incompetent and the caecum is usually involved. CT shows bowel wall thickening with homogeneous attenuation and lack of mural stratification. Enlarged, mesenteric nodes of low attenuation, with enhancing peripheral rings, representing caseous liquefaction are seen. CT findings of tuberculous peritonitis may be demonstrated, including diffuse omental and mesenteric infiltration, nodules, peritoneal thickening and ascites52.

Figure 32.35 Tuberculosis. A short, irregular stricture, about 4 cm long, is shown to involve the terminal ileum and ileocaecal valve. The narrowing caused considerable delay in the passage of barium, and dilatation of the ileum proximal to the stricture can be seen. The patient presented with intestinal obstruction.

CHAPTER 32

Yersiniosis The term yersiniosis refers to infections caused by the Gramnegative bacilli Yersinia enterocolitica and Yersinia pseudotuberculosis. Acute inflammation of the terminal ileum occurs and the clinical presentation is often indistinguishable from acute appendicitis, with right iliac fossa pain and tenderness, fever, diarrhoea and, occasionally, vomiting. The radiological changes of yersiniosis are limited to the distal 20 cm of ileum. The mucosal folds are tortuous, increased in number and thickened with the typical small, discrete nodular filling defects of lymphoid hyperplasia. Some thickening of the wall of the terminal ileum may be present.

Actinomycosis Abdominal actinomycosis is a rare condition caused by Actinomyces israelii, a common saprophyte in the mouth, throat and gastrointestinal tract. Predisposing factors include overt gastrointestinal perforation, previous surgery, neoplasms, diabetes mellitus, steroid therapy and poor dental hygiene. The appendix is the site most commonly affected in the abdominal cavity and the clinical presentation often suggests an appendiceal abscess. Barium studies and/or CT may show a mass causing ileocaecal compression. Sinus tracts and enteroenteric, enterocolic, enterocutaneous and enterovesical fistulae may also be demonstrated.

Giardiasis Giardia lamblia, a flagellate protozoon parasite of the upper small intestine, is commonly associated with diarrhoea in the Tropics. The disease is usually contracted by drinking contaminated water. The presence of G. lamblia in the stool or duodenal aspirate confirms the diagnosis. Irregular thickening of the valvulae conniventes in the duodenum and proximal jejunum is the principal change observed on barium studies. Small, well-defined nodular filling defects due to nodular lymphoid hyperplasia may be seen in patients who also have dysgammaglobulinaemia.

Strongyloidiasis Strongyloides stercoralis is widespread in the Tropics and is the most significant roundworm to involve the small intestine.The clinical symptoms include abdominal pain, vomiting, diarrhoea and weight loss. Microscopic examination of stool, sputum, or duodenal aspirate confirms the diagnosis. Barium studies may show delay in the passage of barium and thickening or absence of the valvulae conniventes in the duodenum and proximal jejunum. In severe cases a rigid ‘pipestem’ stenosis with irregular narrowing may be seen.

Anisakiasis Eosinophilic granulomas have been found in the gastrointestinal tract of patients who habitually eat raw, pickled, or salted fish containing Anisakis larvae. The stomach, terminal ileum and caecum are frequent sites of involvement. The condition is encountered mostly in Holland and Japan. Patients present with symptoms of an acute abdomen, similar to appendicitis or intestinal obstruction. Barium studies show concentric narrowing of

• THE DUODENUM AND SMALL INTESTINE

the involved segment of ileum, sometimes with proximal dilatation. The appearances may be indistinguishable from Crohn’s disease, although in anisakiasis the mucosa is usually intact.

Ascariasis Infestation of the gastrointestinal tract with Ascaris lumbricoides is very common in the Tropics. Intestinal obstruction may develop if a large number of worms are present in children. Large collections of coiled worms may be identified on plain abdominal radiographs.The worms can be seen in barium studies in adults, often symptomless, as single or multiple smooth longitudinal or coiled filling defects, sometimes with a thin central track of barium outlining the worm’s intestinal tract.

CHRONIC RADIATION ENTERITIS Radiation enteritis is a form of intestinal ischaemia resulting from damage to vascular endothelial cells that leads to endarteritis obliterans. High radiation doses or radiation treatment over a short time and a large treatment volume result in a higher incidence of chronic radiation enteritis. The distal ileum, particularly the pelvic loops, is the most frequent site of intestinal damage. The time interval between the radiation therapy and the development of symptoms varies considerably and may be as long as 25 years. The typical clinical presentation of patients with chronic radiation enteritis is colicky abdominal pain, diarrhoea, malabsorption and intermittent small intestinal obstruction. Radiological features include thickening of the valvulae conniventes, mural thickening, effacement of the mucosal pattern, ulceration and fixation and angulation of small intestinal loops. ‘Mucosal tacking’ seen as spiking and distortion of the mucosal folds on the antimesenteric border of the intestine caused by adhesions to the inflamed and thickened mesentery, is rather characteristic. Luminal narrowing and stenosis, single or multiple, are frequent findings (Fig. 32.36), sometimes leading to obstruction. Sinuses and fistulae are uncommon findings. Mural thickening is best assessed on CT. It is demonstrated as a target configuration representing oedema and inflammation, seen in the acute phase, and as homogeneous mural thickening in the chronic, healing fibrotic phase.

MECHANICAL SMALL INTESTINAL OBSTRUCTION The most common causes of small intestinal obstruction are adhesions, external or internal hernias, primary or secondary neoplasms and Crohn’s disease. Abdominal plain radiographs remain a preferred initial method of radiological evaluation in patients suspected of small bowel obstruction but are often inconclusive. Enteroclysis is recommended in patients with recurrent episodes of partial or low-grade obstruction. It can demonstrate the presence and determine the severity of obstruction. It can depict the presence of partially obstructing and nonobstructing symptomatic adhesions and it can accurately distinguish adhesions from metastases, tumours and radiation injury53.

671

672

SECTION 3

•

GASTROINTESTINAL IMAGING

reliable CT criterion for diagnosing small bowel obstruction (Figs 32.38, 32.39). The differentiation of simple obstruction from closed-loop and strangulated obstruction is a crucial issue. Characteristic CT appearances in closed-loop obstruction include U- or C-shaped configuration of dilated loops

Figure 32.36 Radiation stricture. This spot view of a barium infusion examination shows a short, tight stricture of the terminal ileum in a patient who 4 years earlier had undergone radiotherapy for carcinoma of the uterus.

The speed and ability of CT to demonstrate the level, the cause of obstruction and signs of threatened bowel viability and to exclude other causes of an acute abdomen make it particularly valuable in the acute clinical setting. It is especially beneficial in high-grade obstruction and in patients with a history of a previous abdominal malignancy, hernias (Fig. 32.37), inflammatory bowel disease, a palpable abdominal mass or sepsis. The identification of a definite point of obstruction, the ‘transition zone’, with dilated small bowel loops proximal to the site of obstruction and collapsed loops distally, is the most

Figure 32.37 CT – small intestinal obstruction. A section through the mid-abdomen shows dilated, mainly fluid-filled, small intestinal loops. The stretched valvulae conniventes can clearly be identified in some segments. The obstruction was due to an incarcerated paraumbilical hernia, the edge of which can be identified (arrow).

Figure 32.38 Small bowel obstruction secondary to an adhesive band. Enhanced CT, coronal reformation image of a patient with splenic lymphoma shows the transition zone (arrow) between the dilated, fluidfilled jejunal loops proximal to the site of obstruction and the collapsed loops distally. No mass or mural thickening is seen at the transition zone. Note the micronodular lymphomatous splenic involvement.

Figure 32.39 Adhesive small bowel obstruction. Enhanced CT, coronal reformation image shows the transition zone (white arrow), distended, fluid-filled proximal jejunal loops and collapsed distal loops (black arrow). No mass is seen.

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

and a fixed radial distribution of several dilated loops with prominent stretched mesenteric vessels converging towards the point of torsion. CT findings suggestive of strangulation include wall thickening with increased attenuation, poor or no contrast enhancement of the bowel wall, engorgement of mesenteric vasculature, mesenteric haziness and intestinal pneumatosis. Detection of ischaemic changes in the bowel wall and/or attached mesentery implies strangulation necessitating emergency laparotomy. CT enterography and CT enteroclysis can combine the advantages of volume challenge in detecting and grading partial obstruction with the ability of CT to demonstrate the cause and any pertinent extraintestinal manifestations, including vascular impairment.

ACUTE MESENTERIC ISCHAEMIA Acute mesenteric ischaemia can be the result of arterial embolism or thrombosis, venous occlusion or low flow states. Patients usually present with severe abdominal pain. The early diagnosis of bowel ischaemia is difficult both clinically and radiologically. Plain radiographic findings are seen in a minority of patients and are often inconclusive. In the past, angiography was the only radiological test for an early identification of ischaemia. Catheter angiography, whilst considered the procedure of choice due to its diagnostic and therapeutic potential, is often not available and time consuming. The introduction of multi-detector CT has renewed interest in utilizing this imaging modality for evaluating patients with suspected mesenteric ischaemia. Careful attention to CT technique is required.The use of water as oral contrast medium allows better assessment of the bowel wall. A rapidly administered IV bolus of contrast medium and dual-phase imaging are required for accurate mesenteric vessel evaluation. Thrombus typically appears as an intraluminal filling defect within a mesenteric artery or vein. Changes in the affected bowel loops include bowel distension and bowel wall thickening often associated with engorgement of mesenteric veins and ascites (Fig. 32.40). The bowel wall may appear hypoattenuated due to submucosal oedema. Abnormal wall enhancement patterns including diminished enhancement, delayed enhancement or lack of enhancement, may also be seen. Pneumatosis is usually a late finding and indicates irreversible disease (Fig. 32.41).

Figure 32.40 Mesenteric venous thrombosis. Enhanced CT, coronal reformation image shows circumferential thickening of distended jejunal loops (white arrows) and haziness of the adjacent mesentery. Nonopacification of superior mesenteric and jejunal branches is noted (black arrows). Ascites is also present.

or without perforation or mesenteric abscesses, bleeding and small bowel obstruction. On barium examination jejunal diverticula are usually multiple and are seen as fairly large outpouchings with a relatively narrow neck. In cases with jejunal diverticulitis, CT can assess the extent of the inflammatory process and suggest the correct diagnosis.

DIVERTICULA AND BLIND LOOPS Jejunal diverticula Jejunal diverticula are uncommon. They result from mucosal herniation along the mesenteric border. Symptomatic patients present with abdominal pain, distension, weight loss and megaloblastic anemia – the ‘blind loop’ syndrome. Symptoms develop as a result of bacterial overgrowth and can be effectively treated with antibiotics and replacement therapy. Less common complications include acute diverticulitis with

Figure 32.41 Small bowel infarction. Enhanced CT shows pneumatosis of small bowel loops (white arrows) and hepatic portal vein gas (black arrow).

673

674

SECTION 3

•

GASTROINTESTINAL IMAGING

Meckel’s diverticulum Meckel’s diverticulum results from failure of the yolk sac to close during fetal life and is present in 0.5–3 per cent of the population. It is located on the antimesenteric border of the ileum, 30–90 cm from the ileocaecal valve and ranges in size from 0.5 cm to 13 cm. It is estimated that approximately 20– 40 per cent of Meckel’s diverticula cause symptoms. Complications associated with the abnormality include ulceration, bleeding, perforation, inflammation, intussusception, internal hernia, volvulus and adhesions. Ectopic gastric mucosa is present in the diverticulum in 20 per cent of adults and 95 per cent of children who present with bleeding. The preoperative diagnosis of Meckel’s diverticulum may be difficult. Radionuclide imaging with 99mtechnetium pertechnetate (99mTc-pertechnetate) is more accurate in the paediatric age group than in adults. On enteroclysis, it is seen as a blindending sac arising from the antimesenteric border of the ileum (Fig. 32.42). A triradiate pattern of mucosal folds or a triangular plateau is sometimes present at the base of the diverticulum.The diverticulum may become inverted and give the appearance of a polypoid filling defect, often presenting with intussusception. The demonstration of a persistent vitelline artery is a hallmark for the angiographic diagnosis of Meckel’s diverticulum, in patients who present with chronic gastrointestinal bleeding54.

Blind loops Other causes of ‘blind loop’ syndrome include strictures of the small intestine and surgically produced blind loops. Blind pouches are a complication of side-to-side intestinal anastomosis. A blind pouch is invariably a short length of proximal bowel that lies beyond the stoma. It can become quite large, and patients may develop abdominal pain, distension, steatorrhoea, weight loss and megaloblastic anaemia.

Ileal diverticula Acquired diverticula of the ileum are rare. They are usually small, few in number and are located on the mesenteric border

of the terminal ileum. Complications are extremely rare but perforation, diverticulitis, fistula formation and bleeding have been reported.

NEUROMUSCULAR DISORDERS Progressive systemic sclerosis, visceral myopathies and visceral neuropathies can produce diffuse disorders of gastrointestinal motility and patients may present with recurring symptoms and signs of intestinal obstruction in the absence of true mechanical obstruction. In progressive systemic sclerosis barium examination shows dilatation of the duodenum and jejunum, diminished peristalsis, decreased motility and delayed transit. Sacculations, also known as pseudodiverticula, are seen frequently as large, broad-based outpouchings with a somewhat squared contour (Fig. 32.43A). Similar appearances, with the exception of pseudodiverticula, may be seen in visceral myopathy. A characteristic sign of progressive systemic sclerosis is an increased number of mucosal folds – the ‘wire spring or ‘hidebound’ appearance (Fig. 32.43B).

NODULAR LYMPHOID HYPERPLASIA AND IMMUNOGLOBULIN DEFICIENCY Nodular lymphoid hyperplasia is frequently seen in the terminal ileum and colon of children and young adults and is considered to be a normal finding. In older people nodular lymphoid hyperplasia is commonly associated with immunoglobulin deficiency and in particular with the late-onset type of variable hypogammaglobulinaemia. The lymphoid nodules are multiple, small, discrete, round in shape and measure 1–3 mm. The nodules are seen throughout the small intestine in most patients with immunoglobulin deficiency, increasing in number from the proximal to the distal end. The colon is frequently involved throughout its length.

WHIPPLE’S DISEASE Whipple’s disease is a rare condition related to a Grampositive bacillus, Tropheryma whippeli. Presenting symptoms include abdominal pain, diarrhoea, malabsorption, lymphadenopathy and polyarthritis. The characteristic radiological change is thickening of the valvulae conniventes, often with a nodular appearance in the proximal small intestine. CT findings include nonspecific bowel wall thickening and lowdensity retroperitoneal and mesenteric lymphadenopathy, due to increased amount of fat and fatty acids.

INTESTINAL LYMPHANGIECTASIA Figure 32.42 Meckel’s diverticulum. A blind-ending sac is shown arising from the antimesenteric border of the distal ileum (arrow).

Intestinal lymphangiectasia may be primary, part of a generalized hypoplasia of the lymphatic channels, seen in children or young adults, or secondary to lymph flow obstruction by

CHAPTER 32

• THE DUODENUM AND SMALL INTESTINE

Figure 32.43 Systemic sclerosis. (A) Sacculation shown as broad-based outpouchings. (B) Dilatation of a segment of intestine with the ‘hide-bound’ appearance of the valvulae conniventes seen on a spot compression view.

retroperitoneal fibrosis or malignant infiltration of the mesenteric or retroperitoneal lymph nodes. The diagnosis of the primary form can be made by demonstrating dilated submucosal lymphatics on histology, following peroral jejunal biopsy. Enteroclysis shows non-specific uniformly thickened, closely set and parallel valvulae conniventes and/or a micronodular mucosal surface pattern. CT will demonstrate mural thickening or alternative causes of secondary lymphangiectasia.

EOSINOPHILIC GASTROENTERITIS Eosinophilic gastroenteritis is characterized by eosinophilic infiltration of the walls of the stomach and small intestine. Radiologically there is thickening of the valvulae conniventes and mural thickening. Severe luminal narrowing leading to intestinal obstruction may be encountered when the mascularis propria is predominantly involved. The distribution of changes may be regional or generalized. The majority of patients also have gastric involvement, seen radiologically as nodularity and narrowing of the pyloric antrum.

MASTOCYTOSIS Mast cell disease or mastocytosis is a condition in which mast cell infiltration of the skin and other tissues has occurred. It is nearly always characterized by a typical skin rash, known as urticaria pigmentosa. Involvement of the small intestine is frequent and is shown radiologically as thickening of the valvulae conniventes with associated nodular mucosal defects, 2–5 mm

in diameter. The mucosal nodules are usually seen in short segments of jejunum but can occur in the ileum.

AMYLOIDOSIS Infiltration of the gastrointestinal tract with amyloid occurs in the majority of patients with primary amyloidosis. The radiological appearances vary from symmetrical thickening to effacement of the valvulae conniventes. The degree of thickening of the valvulae is related to the amount of vascular amyloid infiltration. Large deposits of amyloid may form intraluminal masses in the barium-filled small intestine. Diminished motility and atrophy of the valvulae may result from amyloid deposited throughout the layers of the intestinal wall and the lumen of the intestine may become dilated. CT findings are nonspecific but typically include symmetric wall thickening of the affected small bowel.

ACQUIRED IMMUNE DEFICIENCY SYNDROME The gastrointestinal tract is a major target organ in the acquired immune deficiency syndrome (AIDS) and there is evidence of small intestinal involvement in about half the affected patients. Opportunist infections, Kaposi’s sarcoma and AIDS-related lymphomas are the principal forms of intestinal involvement in AIDS patients. Mycobacterium avium-intracellulare complex and cryptosporidiosis are the most frequent opportunist infections of the small intestine. Clinical symptoms include a profuse watery diarrhoea and malabsorption. On barium examination there is thickening, sometimes with nodularity, of the valvulae conniventes in the proximal small intestine. On CT, bulky retroperitoneal and mesenteric nodal masses may be seen that

675

676

SECTION 3

•

GASTROINTESTINAL IMAGING

are often indistinguishable from AIDS-related Kaposi’s sarcoma or lymphoma. Thickening of the valvulae conniventes is also a feature of cryptosporidiosis.

BEHÇET’S DISEASE Occasionally the small intestine, particularly the ileocaecal region, is involved in Behçet’s disease. The appearances usually resemble ileocaecal tuberculosis or Crohn’s disease.

GRAFT-VERSUS-HOST DISEASE Graft-versus-host disease develops following allogenic bone marrow transplantation when the foreign donor lymphoid graft mounts an immunological reaction against the skin, liver and gastrointestinal tract of the host. Radiological examination of the gastrointestinal tract reveals three distinct phases. The acute phase, 4–15 days after the onset of gastrointestinal symptoms, consists of uniform thickening or flattening of the mucosal folds, thickening of the intestinal wall, luminal narrowing and ulceration. Ribbon-like narrowing of the lumen and mural thickening may be seen extending throughout most of the jejunum and ileum55. Examination during the second, subacute phase, 13–96 days after the onset of gastrointestinal symptoms, shows similar changes to the acute phase, often with a striking segmental distribution. The third or resolution phase shows improvement with no abnormality or effacement of mucosal folds, but with mural thickening confined to the terminal ileum.

NONSTEROIDAL ANTI-INFLAMMATORY DRUG ENTERITIS Nonspecific ulceration of the small intestine, with blood and protein loss, may develop in patients on long-term treatment with nonsteroidal anti-inflammatory drugs (NSAIDs). Characteristic pathological findings include concentric, circumferential diaphragm-like narrowings, resulting from submucosal fibrosis secondary to focal ulceration. These diaphragm-like narrowings, which can progress to strictures, can be depicted clearly on enteroclysis56.

REFERENCES 1. Warwick R, Williams P L 1980 Gray’s Anatomy, 36th edn. Churchill Livingstone, Edinburgh 2. Stevenson G W, Somers S, Virjee J 1980 Routine double-contrast barium meal: appearance of the normal duodenal papillae. Diagn Imaging 49:6–14 3. Megibow A J 1986 Duodenum. In: Megibow A J, Balthazar E J (eds) Computed tomography of the gastrointestinal tract. CV Mosby, St Louis pp 175–216 4. Horton K M, Fishman E K 2004 Multidetector-row computed tomography of small bowel neoplasms. Current concept in diagnosis. J Comput Assist Tomogr 28:106–116

5. Langkemper R, Hoek A C, Dekker W, Op den Orth J 1980 Elevated lesions in the duodenal bulb caused by heterotopic gastric mucosa. Radiology 137:621–624 6. Nahon J R 1955 The roentgen appearance of localized hyperplasia of the lymphoid follicles of the duodenum. AJR 73:211–214 7. Jayaraman M V, Mayo-Smith W W, Movson J S, Dupuy D E, Wallach M T 2001 CT of the duodenum: An overlooked segment gets its due. RadioGraphics 21:S147–S160 8. Nix G A J J 1980 Early carcinoma of the ampulla and papilla of Vater. Clin Radiol 31:95–100 9. Cho K C, Baker S R, Alterman D D, Fusco J M, Cho S 1996 Transpyloric spread of gastric tumours: comparison of adenocarcinoma and lymphoma. AJR 167:467–469 10. Frostberg N 1938 Characteristic duodenal deformity in cases of different kinds of perivaterial enlargement of the pancreas. Acta Radiol 19:164–173 11. Payawal J H, Cohen A J, Stamos M J 2004 Superior mesenteric artery syndrome involving the duodenum and jejunum. Emergency Radiology 10: 273–275 12. Limani K, Place B, Philippart P, Dubail D 2005 Aortoduodenal fistula following aortobifemoral bypass. Acta Chir Belg 105:207–209 13. Sellink J L, Miller R E 1982 Radiology of the small bowel: technique and atlas. The Hague: Martinus Nijhoff 14. Ha A, Levine M, Rubesin S, Laufer I, Herlinger H 2004 Radiographic examination of the small bowel: Survey of practice patterns in the United States. Radiology 231:407–412 15. Maglinte D D, Gourtsoyiannis N, Rex D, Howard T J, Kelvin F M 2003 Classification of small bowel Crohn’s subtypes based on multimodality imaging. Radiol Clin N Am 41:285–303 16. Gourtsoyiannis N C, Papanikolaou N, Karantanas A 2006 Magnetic resonance imaging evaluation of small intestinal Crohn’s disease. Best Pract Res Clin Gastroenterol 20:137–156 17. Nolan D J 1997 Radiological examination of the small intestine. In: Gourtsoyiannis N, Nolan DJ (eds) Imaging of Small Intestinal Tumors. Amsterdam: Elsevier pp 15–27 18. Lintott D J. The small bowel meal. In: Simpkins K C (ed.) A Textbook of Radiological Diagnosis, vol 4. The Alimentary Tract, the Hollow Organs and the Salivary Glands, 5th edn. H K Lewis, London, pp 322–327 19. Kellett M J, Zboralske F F, Margulis A R 1977 Peroral pneumocolon examination of ileocaecal region. Gastrointest Radiol 1:361–365 20. Dixon P M, Roulston M E, Nolan D J 1993 The small bowel enema: a ten year review. Clin Radiol 47:46–48 21. Traill Z C, Nolan D J 1995 Technical report: intubation fluoroscopy times using a new enteroclysis tube. Clin Radiol 50:339–340 22. Hart D, Hagget P J, Boardman P, Nolan D J, Wall B F 1994 Patient radiation doses from enteroclysis examinations. Br J Radiol 67:997–1000 23. Herlinger H 1978 A modified technique for the double-contrast small bowel enema. Gastrointest Radiol 3:201–207 24. Ekberg O 1977 Double contrast examination of the small bowel. Gastrointest Radiol 1:349–353 25. Miller R E, Sellink J L 1979 Enteroclysis: The small bowel enema. How to succeed and how to fail. Gastrointest Radiol 4:269–283 26. Skehan S J, Mernagh J R, Nahmias C 2002 Evaluation of the small intestine by nuclear medicine studies. In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin, pp 131–155 27. Sfakianakis G, Conway J J 1981 Detection of ectopic gastric mucosa in Meckel’s diverticulum and in other aberrations by scintigraphy: I. Pathophysiology and 10-year clinical experience. J Nucl Med 22:647–654 28. Balthazar E J 2002 Evaluation of the small bowel by computed tomography. In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin, pp 87–130 29. Maglinte D T, Bender G N, Heitkamp D E, Lappas J C, Kelvin F M 2003 Multidetector-row helical CT enteroclysis. Radiol Clin N Am 41: 249–262 30. Raptopoulos V, Schwartz R K, McNicholas M M, Movson J, Pearlman J, Joffe N 1997 Multiplanar helical CT enterography in patients with Crohn’s disease. AJR Am J Roentgenol 169:1545–1550

CHAPTER 32

31. Doerfler O C, Kohlmayr A J K, Reittner P et al 2003 Helical CT of the small bowel as an alternative oral contrast material in patients with Crohn’s disease. Abdom Imaging 28:313–318 32. Valette P J, Rioux M, Pilleul F et al 2001 Ultrasonography of chronic inflammatory bowel diseases. Eur Radiol 11:1859–1866 33. Bin W 1997 Ultrasonography. In: Imaging of Small Intestinal Tumors. Elsevier, Amsterdam 1997, pp 325–343 34. Gourtsoyiannis N, Papanikolaou N, Rieber A, Brambs H J, Prassopoulos P 2002 Evaluation of the small intestine by magnetic resonance imaging. In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin pp 157–170 35. Nolan D J, Gourtsoyiannis N 1980 Crohn’s disease of the small intestine: a review of the radiological appearances in 100 consecutive patients examined by a barium infusion technique. Clin Radiol 31:597–603 36. Gourtsoyiannis N, Grammatikakis J, Papamastorakis G, Koutroumbakis J, Prassopoulos P, Roussomoustakaki M, Papanikolaou N 2006 Imaging of small intestinal Crohn disease: Comparison between MR enteroclysis and conventional enteroclysis. Eur Radiol 16:1915–1925 37. Gourtsoyiannis N, Papanikolaou N, Grammatikakis J et al 2004 Assessment of Crohn’s disease activity in the small bowel with MR and conventional enteroclysis: preliminary results. Eur Radiol 14:1017–1024 38. Tomei E, Diacinti D, Marini M et al. 2005 Abdominal CT findings may suggest coeliac disease. Dig Liv Dis 37:402–406 39. Laghi A, Paolantonio P, Catalano C et al 2003 MR Imaging of the small bowel using polyethylene glycol solution as an oral contrast agent in adults and children with celiac disease: Preliminary observations. AJR 180:191–194 40. Maglinte D D T, O’Connor K, Bassette J, Gernish S M, Kelvin F M 1991 The role of physician in the late diagnosis of primary malignant tumors of the small intestine. Am J Gastroenterology 86:304–308 41. Gourtsoyiannis N, Mako E 1997 Imaging of primary small intestinal tumors by enteroclysis and CT with pathologic correlation. Eur Radiol 7:625–642 42. Olmsted W W, Ros P R, Hjermstad B M, McCarthy M S, Dachman A H 1987 Tumors of the small intestine with little or no malignant predisposition: a review of the literature and report of 56 cases. Gastrointest Radiol 12:231–239 43. Gourtsoyiannis N, Odze R D, Ros P 2002 Malignant small intestinal neoplasms In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin, pp 157–170

• THE DUODENUM AND SMALL INTESTINE

44. Miettinen M, Lasota J 2001 Gastrointestinal stromal tumors: definitions, clinical, histological, immunohistochemical and molecular genetic features and differential diagnosis. Virchows Arch 438:1–12 45. Levy A D, Remotti H E, Thompson W M, Sobin L H, Miettinen M 2003 Gastrointestinal stromal tumors: radiologic features with pathologic correlation. Radiographics 23:283–304 46. Gourtsoyiannis N, Bays D, Malamas M, Barouxis G, Liasis N 1992 Radiological appearances of small intestinal leiomyomas. Clin Radiol 45:94–103 47. Rioux M, Mailloux C 1997 Crescent-shaped necrosis: a new imaging sign suggestive of stromal tumor of the small bowel. Abdom Imaging 22:376–380 48. Gourtsoyiannis N, Bays D 2005 Primary tumors of the small intestine In: Gourtsoyiannis N, Ros P (eds) Radiologic-Pathologic Correlations: from head to toe. Springer, Berlin, Heidelberg, New York, pp 273 49. Jeffree M A, Barter S J, Hemingway A P, Nolan D J 1984 Primary carcinoid tumors of the ileum: the radiological appearances. Clin Radiol 35:451–455 50. Gourtsoyiannis N, Nolan D J 1988 Lymphoma of the small intestine: radiological features. Clin Radiol 39:639–645 51. Nolan D J 1997 Secondary neoplasms. In: Gourtsoyiannis N, Nolan D J (eds) Imaging of small intestinal tumors. Amsterdam, Elsevier, pp 193–211 52. Ha H K 2002 Parasitic and infections diseases of the small intestine. In: N. Gourtsoyiannis (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin, pp 429–446 53. Lappas J C, Maglinte D D T 2002 Radiologic approach to investigation of the small intestine. In: Gourtsoyiannis N (ed.) Radiological Imaging of the Small Intestine. Springer, Berlin, pp 447–463 54. Rossi P, Gourtsoyiannis N, Bezzi M et al 1996 Meckel’s diverticulum: imaging diagnosis. AJR 166:567–573 55. Fisk J D, Schulman H M, Greening R R, McDonald G B, Sale G E, Thomas E D 1981 Gastrointestinal radiographic features of human graft-vs-host disease. Am J Roentgenol 136:329–336 56. Scholz F J, Heiss F W, Roberts P L, Thomas C 1994 Diaphragm-like strictures of the small bowel associated with use of nonsteroidal antiinflammatory drugs. AJR 162:49–50

677

CHAPTER

The Large Bowel

33

Clive I. Bartram and Stuart Taylor

Anatomy Radiological investigation Tumours • Familial adenomatous polyposis • Peutz-Jeghers syndrome • Juvenile polyposis • Hereditary non-polyposis colorectal cancer • Diffuse polypoid malignancy • Radiographic features of polyps • Colorectal carcinoma • Lymphoma • Diverticulitis • Epiploic appendagitis Colitis • Strictures • Pseudodiverticula • Extent of colitis • Carcinoma in colitis • Acute colitis • Ischaemic colitis • Radiation colitis

• Behçet's syndrome • Infectious colitis • Pseudomembranous colitis • Neutropenic colitis • Parasitic colitis • Tuberculosis • Amoebiasis • Acquired immune-deficiency syndrome Miscellaneous conditions • Large-bowel strictures • Lipomatous disorders of the large bowel • Pneumatosis coli • Solitary rectal ulcer syndrome • Volvulus • Intussusception • Endometriosis • Retrorectal lesions • The postoperative colon • Functional disorders of the ano-rectum Anal fistula

ANATOMY The large bowel comprises the colon, rectum and anus. The ascending and descending colon and rectum are retroperitoneal. The transverse and sigmoid colon have a mesentery formed from a double layer of visceral peritoneum sandwiching connective and adipose tissue with vessels, nerves and lymphatics. The caecum, hepatic and splenic flexures may also have short mesenteries. The colonic mucosa is columnar in type with goblet and some enterochromaffin cells arranged in crypts (of Lieberkühn). The surface pattern consists of fine parallel grooves running transversely with short intercommunicating branches, and is called the innominate groove pattern. The lamina propria contains lymphoid follicles; the submucosa, adipose tissue with

neural elements (Meissner plexus), blood vessels and lymphatics.The muscularis propria has two layers, an inner circular and an outer longitudinal with the myenteric (Auerbach's) nerve plexus in between. The outer layer is thin, except where it is condensed into three narrow bands called the taeniae coli that contain more collagen and elastic tissue than muscle. The intraperitoneal colon is covered by mesenteric serosa. Subserosal fat in the caecum and sigmoid accumulates in small peritoneal pouches to form the epiploic appendages that may encase diverticula in the sigmoid. The retroperitoneal colon has an adventitial layer, separating muscle from peritoneal fat. The superior mesenteric artery supplies the colon proximal to the splenic flexure via the ileocolic, right and mid-colic

680

SECTION 3

•

GASTROINTESTINAL IMAGING

branches.The colon distally is supplied by the left colic, sigmoid and superior rectal artery branches of the inferior mesenteric artery.The marginal artery is a vascular arcade in the mesenteric border giving off short branches, the vasa recta, which penetrate the muscle layer close to the taenia mesocolica, and long branches entering between the taenia omentalis and libra. The veins and lymphatics follow the course of the arteries, draining up into the portal vein and celiac nodes, respectively. The midrectal veins drain into the internal iliac vein and so into the systemic circulation via the inferior vena cava. In the rectum lymph drains superiorly via superior rectal artery nodes to the inferior mesenteric chain, posteriorly by nodes around the median sacral artery, and laterally around the middle rectal artery to the internal iliac chain. The taeniae coli converge proximally into the appendix, and fuse distally into the longitudinal muscle of the rectum. The taeniae are thicker in the proximal colon where they interconnect with the circular muscle. Three rows of haustral sacculations arise between the taeniae, with haustral clefts between the sacculations (Fig. 33.1). In the distal colon haustra form only when the taeniae contract. There is often no haustration distal to the mid-transverse colon during a double-contrast barium enema (DCBE), but the absence of haustration in the proximal colon is always abnormal.

Lesser sac

The rectum is defined as beginning at the third sacral level, although the sacral promontory is often taken surgically as the reference point. There is no haustration in the rectum, but the valves of Houston create folds in the rectum. The main middle fold corresponds to the level of the anterior reflection of peritoneum in the pouch of Douglas. The presacral space is normally less than 1 cm at the fourth sacral segment, as measured from a lateral pelvic view during double-contrast examination, but may be up to 2 cm in the elderly and obese. The mesorectal fascia (Fig. 33.2) encloses the mesorectum1, which is derived from the hindgut and composed of loose adipose connective tissue containing the small perirectal lymph nodes and the superior rectal vessels. The lateral rectal ligaments comprise the mid-rectal vessels and nerves. The mesorectal fascia is separated posteriorly from the presacral fascia by the thin retrorectal space; anteriorly it blends with Denonvillier’s fascia, superiorly is contiguous with the sigmoid mesentery, and inferiorly it terminates close to the anus in the parietal fascia covering the levator ani. The anus has a complex sphincteric arrangement with an internal sphincter of smooth muscle and an external sphincter of striated muscle. Between these the longitudinal layer, comprising striated and smooth muscle with extensive fibroelastic tissue, anchors the anus in position. The vascular subepithelial tissues seal the canal to maintain continence.

Gastrocolic lig.

Transverse mesocolon TM

TO

Greater omentum TL Transverse colon Extraperitoneal

TO

TM

TM

Paracolic gutter

TO

Peritoneal reflection TL

TL

Ascending colon

Descending colon

3 taeniae

form 3 rows of haustral sacculations

TO = taenia omentalis TM = taenia mesocolica TL = taenia libera

TO — TM TO — TL TL — TM

Figure 33.1 Relationship of the three taeniae coli to the haustra and mesenteries.

Figure 33.2 Pelvic T2-weighted axial MRI showing the mesorectal fascia (arrows).

RADIOLOGICAL INVESTIGATION Cross-sectional imaging is the first-line investigation in many clinical situations and with the development of computed tomography (CT) colonography the role of the barium enema is diminishing, though it remains unsurpassed for mucosal definition. A water-soluble contrast enema (Gastrografin diluted 3:1 with water or Urografin 150; Schering AG, Berlin) is indicated

if there is any risk of perforation, to demonstrate a fistula, or to show the length of a stricture. These agents are hypertonic and produce diarrhoea. Proximal to colonic strictures water absorption will distend the colon and perforation resulting from this has been reported; care must be taken not to overfill the bowel proximal to a stricture. Isotonic contrast agents have

CHAPTER 33

obvious advantages but, for cost considerations, their use is not justified except in children. A DCBE involves full bowel preparation, an IV smooth muscle relaxant, partial filling of the colon with a barium suspension and insufflation of air or carbon dioxide to distend the colon. The objective is to acquire a series of images so that the entire colon is seen in double contrast, with no segment obscured by a barium pool or coated poorly. Evacuation proctography (EP) is a study of the dynamics of rectal evacuation. Thick barium paste is injected into the rectum and a video recording made of the voluntary evacuation of the paste.The small bowel should be opacified with a dilute barium suspension to show any enterocoele. The examination may be viewed in three stages2: rest, evacuation and recovery. At rest the anorectal junction is normally just above the plane of the ischial tuberosities. Evacuation is initiated by 3 cm descent of the pelvic floor, widening of the anorectal angle, and relaxation of the anal sphincters. The voluntary expulsion of contrast medium does not involve rectal contraction. The rectum distal to the main fold is squeezed by raised intraabdominal pressure against the levator ani to form a ‘zone of evacuation’ that empties in less than 30 s. At the end of evacuation the anal canal closes, and the pelvic floor ascends to its resting position as tone in the puborectalis returns. Standard abdominal CT protocols include bowel opacification either orally or rectally (2 per cent barium or Gastrografin suspension) and enhancement with IV contrast medium in the portal phase. Faecal residue is recognized by its air content. The colonic wall should be not more than 3 mm thick. The pericolic fat should be homogeneous with only a few vascular channels. Mucosal enhancement is seen following IV contrast medium (Fig. 33.3). Colonic and pericolonic abnormalities on CT are summarized in Table 33.1. CT colonography (CTC) with full bowel preparation is standard, although reduced preparation regimens with faecal tagging are being developed. Distension with air, or preferably carbon dioxide, is improved with IV Buscopan (Boehringer,

Table 33.1

Figure 33.3 CT of normal descending colon showing a thin wall with mucosal enhancement. The pericolic fat is homogeneous with only a few small vessels. Faecal residue containing gas is present intraluminally. (Reproduced with the kind permission of Dr A. McLean, St Bartholomew's Hospital.)

Ingelheim)3. Supine and prone sequences reduce the chances of any collapsed or fluid-filled segment hiding a lesion. Review in two-dimensional (2D) cine loop with three-dimensional (3D) interrogation of suspicious areas is standard. Improvements in viewing software speed and capability however make primary analysis using 3D endoluminal projections increasingly viable. The ileo-caecal valve (Fig. 33.4A, B), haustral folds and faecal residue must be distinguished from pathological entities4. Collimation with multidetector CT is typically 1–2.5 mm with

SUMMARY COLONIC AND PERICOLONIC CHANGES ON CT

Wall thickening

>3 mm

Neoplasia, colitis, diverticulitis, ischaemia

Wall attenuation

Target sign without enhancement

Fat in ulcerative colitis Gas in pneumatosis coli

Target sign with enhancement

Acute Crohn's disease or ulcerative colitis Ischaemic colitis Pseudomembranous colitis

Homogeneous with enhancement

Neoplasia Chronic Crohn's disease Radiation enteritis

Enlarged

Reactive or malignant infiltration

Increased attenuation

Kaposi's sarcoma

Decreased central attenuation

TB Mucinous Carcinoma

Mesenteric stranding

Inflammatory

Diverticulitis, colitis, necrosis

Fibro-fatty proliferation

Colon Rectum

Crohn's disease Ulcerative colitis

Pericolonic lymph nodes

• THE LARGE BOWEL

681

682

SECTION 3

•

GASTROINTESTINAL IMAGING

ultra low-dose studies in the 10 mAs range, bringing the dose down to 0.7 and 1.2 mSv for men and women, respectively5. Abdominal ultrasonography requires a graded compression technique6 for good views of the colon. Gas distension proximally often prevents visualization of both walls, but the haustral pattern of the anterior wall should still be seen. Usually only the low reflective muscularis propria and reflective submucosa are identi-

fied. The bowel wall thickness may be measured, Doppler flow assessed and the pericolic tissues interrogated. Endosonography allows higher frequency probes to be used (10–20 MHz range) to show the wall layers in detail. The sonographic pattern is created by a mixture of interface reflections between, and reflections from, the thin layers. A four-layer pattern is seen in the anal canal (Fig. 33.5) with a five-layer one in the rectum (Fig. 33.6).

Figure 33.4 (A) 2D CTC of the ileocaecal valve (arrow) with gas in the terminal ileum just under the valve. (B) 3D view of the biscuspid ileocaecal valve.

Figure 33.5 Endoanal US axial view in the mid-canal. Two bright reflections are seen from the cone, the subepithelial tissues (SE) are moderately reflective, the internal anal sphincter (IAS) is a well-defined low reflective ring surrounded by the mixed reflective longitudinal layer (LL) and the external anal sphincter (EAS) defined by inner and outer interface reflections.

Figure 33.6 Endorectal US with an inner reflective layer from the superficial mucosa, and the deep mucosa (DM) of low reflectivity as is the muscularis propria (MP), with the bright submucosa in between.

CHAPTER 33

• THE LARGE BOWEL

TUMOURS A polyp is simply a description of any elevated mucosal lesion (Table 33.2). Hyperplastic polyps are usually small and are very common in the rectum. The characteristic histological feature is of elongated tubules with a ‘saw-tooth’ edge. Traditionally these have not been considered to have any malignant potential, but recent work suggests that hyperplastic polyps with a ‘serrated’ appearance may also have molecular instability and a propensity to malignant conversion. Adenomas are sharply circumscribed areas of dysplastic epithelium. Dysplasia refers to the presence of intra-epithelial neoplasia. The overall pattern of the epithelium and stroma may be classified as tubular, villous, or mixed. Pedunculation results from the extrusion of a stalk of mucosa and muscularis mucosae as the head is pulled in the faecal stream. Sixty per cent of adenomas are found in the rectosigmoid, 18 per cent in the descending colon, 14 per cent in the transverse colon and 8 per cent in the ascending colon and caecum. Adenomas tend to be larger in the left colon, with two-thirds of polyps >2 cm in size in the rectosigmoid. Adenomas are rare in patients under the age of 30, but 25 per cent of the population over 50 years of age have an adenomatous polyp. Colorectal cancer is defined by extension of the intraepithelial neoplasia into the submucosa. In the general population most colorectal cancer develops from polypoid adenomas. Long-term follow-up of large (>1 cm) adenomas suggests a cumulative risk of cancer at 5, 10 and 20 years of 2.5 per cent, 8 per cent and 24 per cent, respectively7. Size is the most important single indicator for malignancy, approximately 3 mm). The en face configuration may be linear, transverse, serpiginous or rounded. Tangentially there is usually some undercutting of the mucosal edge, giving a ‘T’ or ‘collar stud’ shape. Fissuring ulceration with thorn-like cuts into the bowel wall is a classic feature of CD. Aphthoid ulcers are very superficial so that there is no disruption of the mucosal line. Barium precipitates in the sloughed ulcer base, creating a dense amorphous pool en face. The edge of the ulcer is oedematous, slightly elevated and does not coat with barium, creating a surrounding black halo (Fig. 33.30). Aphthoid ulceration is typical of CD, does not occur in UC but may be seen in amoebiasis, tuberculosis, Behçet's disease, and human immunodeficiency virus (HIV)-related infections.

Figure 33.28 The granular mucosa typical of UC. Note the intact mucosal line.

• THE LARGE BOWEL

Figure 33.29 DCBE in an acute attack of UC with collar stud ulcers (arrow) protruding through the mucosal line (arrowhead).

Reflux ileitis with a patulous ileo-caecal valve and granular distal 10–15 cm of ileum is a typical feature of UC, and disappears rapidly following colectomy. Extensive ulceration in acute colitis may virtually denude the mucosa, leaving oedematous remnants as pseudopolypoid elevations. When an acute attack remits, granulation tissue forms in the base of the healing ulcer. As the ulcers are

Figure 33.30 Aphthoid ulcers (arrows) in CD.

693

694

SECTION 3

•

GASTROINTESTINAL IMAGING

typically undermining in configuration, a flap of mucosa at the edge of the ulcer is left, which is prevented from being sealed down. These mucosal tags form sessile, filiform, adherent or bridging polyps, and are best called postinflammatory polyps (Fig. 33.31).They may remain for many years, even if the colitis goes into complete remission, and are really just markers of the sites of ulceration during the acute attack. They are found in about 20 per cent of patients with UC, less commonly in CD and rarely in other forms of colitis.

STRICTURES In chronic UC there is considerable hypertrophy of the muscularis mucosae and submucosal thickening with fat. The smooth muscle changes are probably responsible for the generalized shortening of the colon, and may produce localized strictures in the left colon in 10–20 per cent with extensive long-standing UC strictures. The strictures are smooth, tapering and symmetrical, with mucosal uniformity. Strictures in CD (Fig. 33.32) are usually asymmetrical with sacculation and secondary to ulceration on the antimesenteric border.

PSEUDODIVERTICULA Sacculation of the bowel wall is frequent in CD, secondary to fibrosis in healing eccentric ulceration. Pseudodiverticula may be seen in ischaemic strictures, but are rare in other forms of colitis and never occur in UC.Wide ‘square'-shaped diverticula in focal areas of bowel wall weakness are seen in scleroderma.

Figure 33.31 Filiform postinflammatory polyposis (arrow) following an acute attack of UC. The mucosal surface and haustration are normal as the colitis was inactive.

Figure 33.32 Extensive colonic CD with a short mid-transverse colon stricture (arrow) and generalized thickening of the bowel wall, also seen in the descending colon (arrowhead). There is some pericolic mesenteric stranding.

EXTENT OF COLITIS In proctitis the inflammatory changes are limited to the rectum, with distal colitis to the level of the left iliac crest, and with extensive colitis to the hepatic flexure when there is always total involvement histologically. US shows thickening (>4 mm) of the wall. The wall is typically stratified in UC with differentiation between the submucosal and muscularis propria (Fig. 33.33), although in chronic CD this may be lost. The surrounding fat is more reflective with acute inflammation. Ulceration is evident from focal disruption of the bowel wall layers and may be outlined by intracolonic gas. The findings on CT are similar. Loss of stratification during IV enhancement suggests CD. The pericolic tissues are normal in UC (unless perforation has occurred), but in CD there are serosal and mural irregularities (Fig. 33.34). In

Figure 33.33 Stratified wall thickening in UC on US. The outer low reflective muscle layer is well defined, but the thickened mucosa/ submucosa are poorly distinguished. The mucosal surface is indicated by the bright central reflective line (arrow).

CHAPTER 33

• THE LARGE BOWEL

ill-defined inflammatory mass without overt abscess formation. These present as poorly defined focal masses of increased attenuation in adjacent omentum or mesentery. Abscesses are of low density (10–30 HU) and often contain gas bubbles either from gas-forming bacteria or a direct communication to bowel.

Differential features

Figure 33.34 CD with wall thickening, serosal irregularity, increased vascularity and a small pericolonic fluid collection (reproduced with kind permission of Dr A. McLean, St Bartholomew's Hospital.)

chronic UC there is thickening of the muscularis mucosae and fatty infiltration in the submucosa creating a target sign even on unenhanced images24. Fatty proliferation is responsible for widening the presacral space in rectal disease (Fig. 33.35). Extramural changes are prominent in CD, with fibrofatty proliferation, increased fat attenuation, small lymph nodes 3 mm, are 92 per cent and 95 per cent, respectively (Fig. 36.10). The negative predictive value of the absence of gallbladder stones and a negative sonographic Murphy’s sign is 95 per cent17. US can be definitive in about 80 per cent of cases18. Gallstone(s) may be impacted in the neck of the gallbladder and this region must be carefully examined. Other US signs are gallbladder distension (diameter >5 cm), pericholecystic fluid, gallbladder wall striations and, occasionally, obvious wall hyperaemia on Doppler examination. Fine echoes seen within the gallbladder may be seen due to sludge or pus (gallbladder empyema). If liver function tests suggest duct obstruction a careful evaluation of the common bile duct should be made for choledocholithiasis. CT is less accurate than US for acute cholecystitis, but is widely used for the evaluation of patients with acute abdominal pain. The CT findings in acute cholecystitis include gallbladder wall thickening, subserosal oedema, gallbladder distension, high-density bile, pericholecystic fluid and inflammatory stranding in the pericholecystic fat (Fig. 36.11). Gallstones are identifiable in the minority and may be seen as high-density or, less often, low-density entities. Gallbladder wall enhancement is variable and not a reliable predictor of cholecystitis since normal gallbladders can show wall enhancement. Transient pericholecystic liver rim enhancement may be seen19.

Figure 36.11 Acute cholecystitis on CT. The gallbladder wall is thickened with oedema in the adjacent fat. There is no abnormal contrast enhancement in this case.

Hepatobiliary scintigraphy has a high diagnostic accuracy for acute cholecystitis. A positive result is nonvisualization of the gallbladder, which results from cystic duct obstruction. Imaging is performed up to 2–4 h after the administration of isotope. Whilst its accuracy is similar to US it is more time consuming and does not allow assessment of related organs. It can be helpful, however, when diagnostic uncertainty remains after US. Gallbladder wall thickening may result from many causes other than cholecystitis. These include nonfasting, generalized oedematous states, hepatitis, pancreatitis, gallbladder wall varices, adenomyomatosis and carcinoma, though the latter two usually cause focal rather than diffuse thickening.

Gangrenous cholecystitis

Figure 36.10 Acute cholecystitis. The gallbladder contains small stones in the neck (Nos.1–4) and its wall shows oedematous thickening (5 mm thickness).

This condition is suggested on US by pronounced irregularity or asymmetrical thickening of the gallbladder wall, internal membranous echoes resulting from sloughed mucosa and pericholecystic fluid. The clinical findings, paradoxically, may diminish with progression to gangrenous change. CT signs that suggest gangrenous cholecystitis are gas in the wall or lumen, discontinuous and/or irregular mucosal enhancement, internal membranes representing sloughed mucosa and pericholecystic abscess20,21. Gallbladder perforation, which is more often localized than generalized, occurs in 5–10 per cent of patients with acute cholecystitis19. It is suggested on US or CT by pericholecystic fluid and the features of gangrenous cholecystitis. Localized disruption of the gallbladder wall is seen on US in 40 per cent of cases and on CT in 80 per cent (Fig. 36.12). Less often, generalized peritoneal fluid may be present22.

769

770

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 36.12 Acute cholecystitis with localized perforation on (A) US and (B) CT. The thickened gallbladder wall shows a local defect (arrow) and on CT there is small amount of intraperitoneal fluid and oedema of adjacent fat.

Emphysematous cholecystitis This condition accounts for only 1 per cent of cases of acute cholecystitis but has a relatively high mortality rate. It is more common in men (the reverse of the usual female predominance in cholecystitis), about one-half of whom are diabetics, and stones are present in less than 50 per cent19. The diagnosis may be evident on plain radiographs (Fig. 36.13) and is readily made on CT, which shows intramural and/or intraluminal gas caused by gas-forming organisms. On US intramural gas appears as focal or diffuse bright echogenic lines. Intraluminal gas, in the nondependent portion of the gallbladder, causes a curvilinear, brightly echogenic band with shadowing, which can make recognition of the gallbladder difficult and lead to a false-negative US result (Fig. 36.14). Small foci of intramural gas may cause ring-down artefact and mimic the appearance seen with adenomyomatosis.

Chronic acalculous cholecystitis is a controversial entity as there are no clear clinical, pathological or imaging criteria for its diagnosis. The clinical setting is usually unexplained biliary-type pain, and patients have often previously undergone numerous other negative investigations. US may show

Acalculous cholecystitis Acute acalculous cholecystitis is most often seen in patients in intensive care units, or those who are otherwise critically ill, and the clinical presentation is usually one of sepsis. The US signs are gallbladder distension, gallbladder wall thickening, echogenic contents and, occasionally, sloughed membranes/mucosa and pericholecystic fluid. A positive diagnosis is often difficult as sludge and gallbladder distension may occur without cholecystitis in this group of patients. All investigation – US, CT and biliary scintigraphy – are substantially less diagnostically accurate than they are in acute calculous cholecystitis. Biliary scintigraphy is possibly the most accurate modality. Gallbladder aspiration has been used to aid diagnosis but is often unhelpful23. Localized gallbladder tenderness is a good predictive sign when present but is frequently difficult to assess in this group of patients.

Figure 36.13 Emphysematous cholecystitis. Image showing intramural (arrow) as well as intraluminal gallbladder gas.

CHAPTER 36 • THE BILIARY SYSTEM

Figure 36.14 Emphysematous cholecystitis. (A) CT – Intraluminal gas; (B) US – intraluminal gas appears as a bright curvilinear echogenic band (arrow) with ‘dirty’ shadowing.

gallbladder wall thickening and, by definition, no stones. Cholescintigraphy followed by the IV infusion of cholecystokinin (CCK), or one of its analogues, can be used to assess gallbladder contractibility. An ejection fraction greater than 35 per cent on CCK-cholescintigraphy is generally taken to be an indicator of gallbladder dysfunction and helps select patients who may benefit from cholecystectomy24. Xanthogranulomatous cholecystitis is an unusual form of chronic cholecystitis that may simulate malignancy radiologically and pathologically. It usually presents with the clinical features of cholecystitis or biliary obstruction (a variant of Mirizzi syndrome). It is characterized on imaging by gallbladder wall thickening, which may be focal or diffuse and can be quite marked. Gallbladder stones are present in the great majority of cases, and in a small percentage there is associated gallbladder carcinoma25.

GALLBLADDER FISTULAE Gallbladder fistulae are rare. A minority are due to neoplastic disease with the great majority due to chronic stone disease. Most are to the duodenum and most of the remainder to the colon. Chole-cystoduodenal fistulae may result in bowel obstruction due to the impaction of larger stones in the distal small bowel, so-called gallstone ileus, a condition associated in a minority of patients with a visible gallstone on plain radiographs or CT, and gas in the biliary tract.

PORCELAIN GALLBLADDER Porcelain gallbladder is an uncommon condition of mural calcification associated with chronic cholecystitis. It may be asymptomatic but cholecystectomy is often advocated as the reported incidence of a complicating carcinoma in up to 33 per cent19. The calcification follows the contour of the gallbladder wall, may be generalized or localized, and may be visible on CT or plain radiography. On US it can mimic emphysematous cholecystitis or gallstones but the ‘double-arc shadow’ sign of stones is absent.

ADENOMYOMATOUS HYPERPLASIA This condition is known by several names including adenomyomatosis and cholecystitis glandularis proliferans. It occurs in up to 9 per cent of cholecystectomy specimens and in 90 per cent of cases there are associated gallstones. It is characterized by thickening of the gallbladder wall due to epithelial and smooth muscle hyperplasia, with cystic epithelial invaginations into the wall (Rokitansky-Aschoff sinuses) and these spaces may contain small stones. Its distribution is fundal (most common), segmental (usually in mid-body) or diffuse. On US it appears as gallbladder wall thickening with secondary luminal narrowing (Fig. 36.15). The affected segment often contains bright echoes arising from the cystic spaces, or from small stones or crystals within them, often associated with

771

772

SECTION 3

•

GASTROINTESTINAL IMAGING

technique can show the subtle, thin, acoustic shadowing that makes the diagnosis of a stone. Lack of mobility favours a polyp rather than a stone though, occasionally, polyps have a pedicle and may show limited mobility. Polyps per se are usually of no significance, though a diameter of >10 mm or local disruption of the adjacent gallbladder wall suggests malignancy25. Doppler flow within an echogenic lesion differentiates it from tumefactive sludge but does not distinguish reliably between benign and malignant polyps.

GALLBLADDER CARCINOMA

Figure 36.15 Adenomyomatous hyperplasia. Gallbladder wall thickening in the fundus is associated with small stones (arrow).

‘comet-tail’ ring-down artefacts. CT shows wall thickening. Contrast gallbladder studies (e.g. OCG) and T2-weighted MR may show the intramural cystic spaces25.

GALLBLADDER POLYPS The great majority of polyps are cholesterol and less often adenomatous. Cholesterol polyps are usually 2–10 mm in size whereas adenomas can be up to 2 cm. Cholesterol polyps are multiple and not often associated with stones whereas adenomas tend to be solitary and associated with stones. They both appear as small echogenic nonshadowing foci adherent to the gallbladder wall, often in a nondependent portion (Fig. 36.16). The main differential diagnosis is small stones and careful US

Figure 36.16

Gallbladder carcinoma is an uncommon malignancy that has a very poor prognosis unless detected incidentally at cholecystectomy. A minority will be detected early as a polypoid intraluminal mass. Usually, however, it presents at a late stage in the sixth and seventh decades with right upper-quadrant pain, often presenting as hilar biliary obstruction. Apart from the findings associated with biliary obstruction, it may be seen on imaging as focal or diffuse irregular thickening of the gallbladder wall or as a larger mass in the gallbladder fossa with little or no gallbladder lumen identifiable. Gallbladder stones are present in the majority and may appear to be ‘buried’ in the mass. The disease tends to spread to lymph nodes around the por-tal vein relatively early in its course and at presentation there may be nodal masses extending down to the head of the pancreas. It also spreads to the adjacent liver (segments 4 and 5). The differential diagnosis includes Mirizzi syndrome and metastases to the gallbladder, which are uncommon but include melanoma. Doppler blood flow can frequently be shown within the mass, and contrast enhancement may occur on CT and MR which, in the larger masses, is in the periphery around areas of central necrosis (Fig. 36.17)19.

Gallbladder polyps. (A) Solitary, nondependent and nonshadowing polyp (arrow). (B) Multiple, nonshadowing cholesterol polyps.

CHAPTER 36 • THE BILIARY SYSTEM

Figure 36.17 Gallbladder carcinoma. (A) Polyp with breach of continuity of the underlying wall (arrow). (B) Advanced carcinoma extending outside the fundus, with a nodal metastasis posterior to the pancreatic head (arrow). An associated stone can be seen in the gallbladder neck.

ROLE OF RADIOLOGY IN INVESTIGATION OF JAUNDICE The principal role of imaging in the jaundiced patient is the iden-tification and detailed assessment of major bile duct obstruction. The clinical suspicion is based on a variable combination of dark urine, pale stools, pruritus, cholangitis and cholestatic liver function tests. US is the preferred initial imaging investigation, but will usually be supplemented with a combination of CT, MRCP, direct cholangiography and, in some centres, endoscopic and/or intraoperative US. The questions that need to be addressed are: 1 Is bile duct obstruction present? 2 What is the anatomical level of obstruction? 3 What is the cause of the obstruction? 4 If the obstruction appears to be malignant: a There evidence of nonresectability? b In those patients with malignant hilar obstruction who are unsuitable for surgical resection, what approach should be taken to palliative stenting? Attention to these questions at the initial US allows targeted use of the supplementary investigation to address the unanswered questions26. The first task is to determine if there is intrahepatic and/or extrahepatic duct dilatation as a marker of duct obstruction. The intrahepatic ducts should measure no more than 2–3 mm centrally; more peripherally they are usually only just visible on US and should be clearly smaller than the adjacent portal vein branches. Mild dilatation of the intrahepatic ducts may occur without duct obstruction in the elderly.

Selection of a single common duct diameter to predict distal bile duct obstruction is problematic. The maximum diameter of the normal common duct (includes the common hepatic and common bile duct) is influenced by age and where the duct is measured. A diameter of >7 mm is commonly used as a predictor of bile duct obstruction in the jaundiced patient but this is only a guide27. Lower values should be used in younger adults and, conversely, in the normal older population values of 8 mm or more are not unusual. If there has been a cholecystectomy the upper limit of ‘normal’ is less well defined and the duct tends to be larger, commonly up to 10 mm. Further investigation is guided by the level of clinical and laboratory evidence of duct obstruction. The diameter of the duct at the superior end of the portal vein tends to be less than it is in the more distal duct and this can sometimes be quite misleading. The author's practice is to attempt to visualize the whole length of duct and to measure the largest internal diameter, which tends to be in the suprapancreatic portion. If only the very upper end of the common duct is seen and is not dilated this does exclude pathological dilatation of the more inferior portion. Conversely, if there is mild dilatation of the suprapancreatic portion but the duct tapers to a normal size in its pancreatic portion, further imaging is not mandatory and should be guided by the clinical likelihood of duct obstruction27. Hilar biliary obstruction will produce only intrahepatic duct dilatation, whilst more distal obstruction will result in extrahepatic dilatation followed by intrahepatic dilatation.

773

774

SECTION 3

•

GASTROINTESTINAL IMAGING

Approximately 95 per cent of patients with bile duct obstruction have biliary dilatation, the degree of which is related to the duration and completeness of the obstruction. In the remaining 5 per cent there are usually sufficient clinical/biochemical indicators of duct obstruction to suggest that cholangiography of some form (MRCP or ERCP) is warranted. Most cases of biliary obstruction without duct dilatation are due to choledocholithiasis, primary sclerosing cholangitis or postoperative stricturing. If there is evidence of duct obstruction (i.e. duct dilatation) the next question is to determine the anatomical level, namely whether it is hilar (at or close to the confluence of the right and left hepatic ducts), or low/mid common duct (Fig. 36.18). This helps with the differential diagnosis as well as in the selection of further imaging tests. The main differential diagnoses are summarized in Table 36.1. The choice of further imaging tests is determined by what has been shown on US. If US shows choledocholithiasis, patients can proceed to endoscopic sphincterotomy or cholecystectomy. If US does not show stones but stones are highly likely on clinical grounds (e.g. pain and fever associated with jaundice) patients should proceed to ERCP in most cases, especially in the presence of sepsis. In patients with significant comorbidities that might contraindicate ERCP or surgery, MRCP is helpful in providing confirmatory evidence of stones or suggesting another cause of obstruction. US detects the level of obstruction in up to 95 per cent and cause in up to 88 per cent26. If the cause is not evident on US, and stones are not considered the most likely diagnosis, CT is usually the next most useful test, although MRCP, MRI and endoscopic ultrasound (EUS) may be substituted depending on local access and expertise. A meta-analysis has shown that MRCP identifies the presence of obstruction in 99 per cent, the level of obstruction in 96 per cent and detects tumour in 88 per cent of patients with a malignant cause28. Multislice CT is highly accurate for identifying the level and cause of obstruction, having similar accuracy to MRI and MRCP, especially with use of multiplanar and cholangiography-type CT reformats (Fig. 36.19)4,29,

Figure 36.18 Low biliary obstruction. Longitudinal US shows a very dilated bile duct (13 mm) and a large pancreatic head carcinoma.

Table 36.1

CAUSES OF MAJOR BILE DUCT OBSTRUCTION

Anatomical location

Malignant

Benign ***

Hilar

Gallbladder carcinoma Hepatocellular carcinoma**

Low/mid duct

Pancreatic carcinoma*** Ampullary carcinoma**

Pancreatitis (acute or chronic)**

Either

Cholangiocarcinoma*** Metastases*** Lymphoma* Benign biliary tumors*

Stones**** Mirizzi syndrome** Postoperative strictures*** Primary sclerosing cholangitis*** Other cholangiopathy* Hemobilia* Parasites*

Astersiks indicate approximate relative incidence (**** = most common). Low/mid duct obstruction is more common than hilar obstruction.

Figure 36.19 Low biliary obstruction. Multislice CT with curved coronal reformat displaying a pancreatic head tumour (arrows) obstructing the common bile duct and pancreatic duct.

the exception being that MRCP has a higher accuracy for detection of choledocholithiasis. The next questions relate to the detailed evaluation of malignant obstruction in regard to tumour resectability and biliary decompression options. In malignant hilar obstruction any evaluation should assess the proximal extent of stricturing into the right and left hepatic ducts, the presence of lobar atrophy, the patency of the portal veins (main, right and left branches) and the presence of any intrahepatic or local extrahepatic metastases. The proximal extent of stricturing is classified according to the modified Bismuth classification (Fig. 36.20)30. In malignant low obstruction, usually due to pancreatic carcinoma, the main factors to assess are tumour size, vascular involvement (portal vein, superior mesenteric vein and superior mesenteric artery), lymph node metastases and hepatic metastases Further information concerning pancreatic lesions giving rise to biliary obstruction may be found in Chapter 37, The Pancreas.

CHAPTER 36 • THE BILIARY SYSTEM

US (including Doppler), CT and MRI (including MRCP and MRA) can all provide information about tumour resectability. Angiography and direct cholangiography (PTC and ERCP) have been replaced in most centres by the other modalities for purposes of resectability assessment. Positron emission tomography (PET) scanning is more helpful in identifying metastases than it is in identifying primary biliary tumours31. Resectability assessment should, ideally, identify signs of nonresectability without excluding appropriate patients from the chance of surgical cure. The choice of imaging technique depends on local facilities, expertise and surgical practice. Multislice CT has a good overall accuracy for resectability assessment. For a hilar tumour MRCP is especially helpful in assessing the proximal extent of the lesion and determining its Bismuth classification. EUS is useful for the assessment of any involvement of the superior mesenteric and portal veins by pancreatic head or periampullary tumours. It also allows fineneedle aspiration cytology of the tumour or suspicious lymph nodes. Core biopsy or fine-needle aspiration of suspected malig-nant obstructing lesions can be guided by US or CT, the choice depending on US access and user preference. If surgical resection is being attempted then a preoperative biopsy is not usually appropriate. If palliative stenting is being performed it is preferable to perform a biopsy after decompression to reduce the risk of a bile leak.

Figure 36.20 Modified Bismuth classification of malignant hilar biliary obstruction based on proximal extent of tumour.

BENIGN BILE DUCT PATHOLOGY CHOLEDOCHOLITHIASIS At least 90 per cent of bile duct stones are stones that have passed from the gallbladder, so-called secondary stones. Primary stones are those that arise in the bile duct and these are pigment stones. For patients younger than 60 years undergoing cholecystectomy 8–15 per cent have duct stones, the figure increasing substantially in older patients32.

Ultrasound This is the most commonly used initial imaging modality. Reports of its sensitivity vary greatly with the upper range being 50–80 per cent. The sensitivity in jaundiced patients

tends to be better. The specificity is, however, about 95 per cent. Duct dilatation and acoustic shadowing are each absent in about 30 per cent of cases33. Positive stone diagnosis depends on the demonstration of an intraductal echogenic focus in both the longitudinal and transverse planes (Fig. 36.21). Conditions that may mimic stones on US are: 1 Intraductal gas - usually recognizable by its linear nature and its movement. 2 Haemobilia and sludge - they produce more diffuse echoes than stones. 3 Surgical clips, hepatic artery calcification and duodenal diverticula - these do not lie within the lumen of the duct. 4 Parasites.

775

776

SECTION 3

•

GASTROINTESTINAL IMAGING

Figure 36.22 Choledocholithiasis. A distal common bile duct stone (arrow) is slightly dense compared with the surrounding low-density bile. Figure 36.21 Choledocholithiasis. Small shadowing stone (arrow) in dilated bile duct.

US measurement of bile duct diameter has value as a predictor of the presence of bile duct stones, apart from the direct detection of stones. If the bile duct diameter is 25% visible glandular tissue.

BREAST PATHOLOGY Benign mass lesions Cysts Cysts are the most common cause of a discrete breast mass, although they are often multiple and bilateral. They are common between the ages of 20 and 50 years, with a peak incidence between 40 and 50 years. Simple cysts are not associated with an increased risk of malignancy and have no malignant potential. On mammography they are seen as well-defined, round or oval masses. Sometimes a characteristic halo is visible on mammography (Fig. 52.8A). Cysts can be readily diagnosed with ultrasound. They have well-defined margins, are oval or round in shape, and show an absence of internal echoes indicating the presence of fluid. The area of breast tissue behind a cyst appears bright on ultrasound (posterior enhancement) due to improved transmission on the ultrasound beam through the cyst fluid (Fig. 52.8B). When these features are present, a cyst can be diagnosed with certainty. Aspiration is easily performed under ultrasound guidance to alleviate symptoms or when there is diagnostic uncertainty. Cytology on cyst fluid is not routinely performed unless there are atypical imaging features or the aspirate is bloodstained. Figure 52.8 Cyst. (A) A well-defined rounded mass, with an associated lucent halo characteristic of a cyst. (B) The absence of internal echoes and the posterior enhancement of the ultrasound beam are diagnostic of a cyst.

1179

1180

SECTION 6

• WOMEN’S IMAGING

Fibroadenomas and related conditions Fibroadenomas are the most common cause of a benign solid mass in the breast. They present clinically as smooth, welldemarcated, mobile lumps. They are most frequently encountered in younger women with a peak incidence in the third decade.With the advent of screening, many previously asymptomatic lesions are detected. On mammography, fibroadenomas are seen as well-defined, rounded or oval masses. In the majority of cases they are solitary, but in 10–20% they are multiple (Fig. 52.9A)18. Coarse calcifications may develop within fibroadenomas, particularly in older women (Fig. 52.10). Ultrasound features have been described that are characteristic of benign masses17. These include hyperechogenicity compared to fat, an oval or well-circumscribed, lobulated, gently curving shape and the presence of a thin, echogenic pseudocapsule. If several of these features are present and there are no features suggestive of malignancy, then a mass can be confidently classified as benign. Many fibroadenomas have a smooth, well-defined margin, are gently lobulated and are oval in shape (Fig. 52.9B). Most fibroadenomas are isoechoic or mildly hypoechoic relative to fat. A thin echogenic pseudocapsule may be seen. In most cases, even though the mass has benign features, percutaneous biopsy is necessary to confirm the diagnosis. However, in patients with no suspicious features in whom several benign characteristics are present, a percutaneous biopsy may be avoided17. One such group may be women under the age of 25, where the risks of any mass being malignant are very small. Fibroadenomas must be distinguished from wellcircumscribed carcinomas; this is done by percutaneous biopsy. Phyllodes tumour can have a similar appearance to fibro-

Figure 52.10 Fibroadenomas may develop coarse ‘pop-corn’ type calcifications.

adenoma, leading to diagnostic difficulties (Fig. 52.11). The pathological characteristics can also be similar to those of large fibroadenomas. The majority of phyllodes tumours are benign, but some (less than 25%) are locally aggressive and may even metastasize18. When a diagnosis of phyllodes tumour is made, surgical excision must be complete with clear margins to prevent the possibility of recurrence. Many larger fibroadenomas (over 3 cm), and those that show a rapid increase in size, tend to be excised, in order to avoid missing a phyllodes tumour.

Figure 52.9 Fibroadenoma. (A) Two well-defined masses on mammography. (B) Ultrasound of the lesion nearer the nipple showed a well-defined oval mass. Both lesions were confirmed as fibroadenomas on ultrasound-guided core biopsy.

CHAPTER 52

• THE BREAST

the mass is associated with microcalcifications. On ultrasound, they typically appear as a filling defect within a dilated duct or cyst (Fig. 52.12B). On aspiration, any cyst fluid may be bloodstained. As it is impossible to differentiate papillomas from papillary carcinomas on imaging criteria, percutaneous biopsy is required. Papillomas may be associated with an increased risk of malignancy, particularly if they are multiple or occur in a more peripheral location within the breast. Consequently, excision of papillary lesions is desirable and may be therapeutic in cases of nipple discharge. In situations where percutaneous biopsy shows no evidence of cellular atypia, then an alternative to surgical excision is piecemeal percutaneous excision using a vacuum-assisted biopsy device.

Lipoma

Figure 52.11 Phyllodes tumour. The presence of several cystic spaces within this large, well-defined mass suggested the possibility of a phyllodes tumour. This was confirmed on core biopsy and surgical excision.

Papilloma Papillomas are benign neoplasms, arising in a duct, either centrally or peripherally within the breast. Many papillomas secrete watery material leading to a nipple discharge. As they are often friable and bleed easily, the discharge may be bloodstained. On mammography, they may be seen as a well-defined mass, commonly in a retroareolar location (Fig. 52.12A). Sometimes

Lipomas are benign tumours composed of fat. They present clinically as soft, lobulated masses. Large lipomas may be visible on mammography as a radiolucent mass (Fig. 52.13A). On ultrasound their characteristic appearance is that of a welldefined lesion, hyperechoic compared to the adjacent fat (Fig. 52.13B).

Hamartoma Hamartomas are benign breast masses composed of lobular structures, stroma and adipose tissue, the components that make up normal breast tissue.They occur at any age. On imaging they may be indistinguishable from other benign masses, such as fibroadenomas. Sometimes large hamartomas, detected on screening mammograms, are impalpable (Fig. 52.14). On mammography they classically appear as large, well-circumscribed masses containing a mixture of dense and lucent areas, reflecting the different tissue components present. Diagnostic

Figure 52.12 (A) Multiple small papillomas. Papillomas are frequently well defined on mammography, although part of the mass may have an irregular or ill-defined contour. (B) On ultrasound, the presence of a filling defect within a cystic structure suggests the diagnosis. Colour Doppler can be useful to distinguish debris within a cyst from a soft tissue mass.

1181

1182

SECTION 6

• WOMEN’S IMAGING

Figure 52.13 Lipoma. (A) On mammography, a lipoma may be seen as a well-defined mass of fat density, contained within a thin capsule. (B) On ultrasound, a well-defined hyperechoic lesion characteristic of a lipoma is seen.

extended across the basement membrane of the TDLU into the surrounding normal breast tissue, the carcinoma is invasive. Malignant cells contained by the basement membrane are termed noninvasive or in situ.

Classification of invasive breast cancer There is much confusion regarding the classification of breast cancer. Some tumours show distinct patterns of growth, allowing certain subtypes of breast cancer to be identified. Those with specific features are called invasive carcinoma of special type, while the remainder are considered to be of no special type (NST or ductal NST). Special type tumours include lobular, medullary, tubular, tubular mixed, mucinous, cribriform and papillary. Different types of tumour have different clinical patterns of behaviour and prognosis. It should be understood that when a tumour is classified as of a special type this does not imply a specific cell of origin, but rather a recognizable morphological pattern18,21. Histological grade has implications for tumour behaviour, imaging appearances and prognosis. The morphological features on which histological grade is based are tubule formation, nuclear pleomorphism and frequency of mitoses21. Low-grade tumours that are well differentiated are less likely to metastasize. Figure 52.14 Hamartoma. Hamartomas are frequently encountered on screening mammograms as large, lobulated masses with areas of varying density reflecting the presence of elements which are of fat and soft tissue density.

difficulty may be encountered because percutaneous biopsy specimens may be reported as normal breast tissue.

Invasive carcinoma Breast carcinomas originate in the epithelial cells that line the terminal duct lobular unit (TDLU).When malignant cells have

Imaging appearance of invasive breast cancer Mammography Carcinomas typically appear as ill-defined or spiculated masses on mammography (Fig. 52.15A,B). Lowergrade cancers tend to be seen as spiculated masses, due to the presence of an associated desmoplastic reaction in the adjacent stroma. Higher-grade tumours are usually seen as an illdefined mass, but sometimes a rapidly growing tumour may appear relatively well defined, with similar appearances to a benign lesion such as a fibroadenoma (Fig. 52.15C). Many breast cancers arise from areas of ductal carcinoma in situ (DCIS) and are associated with microcalcifications

CHAPTER 52

• THE BREAST

Figure 52.15 Mammographic appearances of invasive carcinoma. Spiculated and ill-defined masses are typical features of malignancy. The spiculated mass (A) and the ill-defined mass (B) were found to be ductal NST carcinomas of intermediate grade on core biopsy. (C) Sometimes highgrade tumours that exhibit rapid growth may appear more well defined. (D) Calcifications typical of high-grade DCIS may be found associated with invasive carcinomas.

1183

1184

SECTION 6

• WOMEN’S IMAGING

on mammography (Fig. 52.15D). This is particularly true for high-grade invasive ductal carcinomas that are often associated with high-grade DCIS22. Special type tumours can have particular mammographic characteristics: • Lobular carcinomas can be difficult to perceive on a mammogram due to their tendency to diffusely infiltrate fatty tissue. Compared with ductal NST tumours, lobular cancers are more likely to be seen on only one mammographic view, are less likely to be associated with microcalcifications, and are more often seen as an ill-defined mass or an area of asymmetrically dense breast tissue23. • Tubular and cribriform cancers often present as architectural distortions or small spiculated masses24. • Papillary, mucinous and medullary neoplasms may appear as new or enlarging multilobulated masses and may be well defined, simulating an apparently benign lesion25,26. • A small spiculated mass will be easily visible in a fatty breast, whereas even large lesions can be obscured by dense breast parenchyma. Sometimes the only clue to the presence of an invasive tumour may be abnormal trabecular markings, known as an architectural distortion, or the presence of microcalcifications, which tend to be visible even when the breast parenchyma is dense. The ability to perceive small or subtle cancers on a mammogram is improved by having the two standard mammographic views available and seeking out previous studies for comparison. An increase in the size of a mass or the presence of a new mass is suspicious of malignancy, whereas a lesion that remains unchanged over many years is invariably benign. Multiple masses in both breasts would favour a benign disease such as cysts or fibroadenomas. Ultrasound There are characteristic malignant features on ultrasound17: • Carcinomas are seen as ill-defined masses and are markedly hypoechoic compared to the surrounding fat (Fig. 52.16A). • Carcinomas tend to be taller than they are wide (the anterior to posterior dimension is greater than the transverse diameter). • There may be an ill-defined echogenic halo around the lesion, particularly around the lateral margins, and distor-

Figure 52.16 Ultrasound appearances of invasive carcinoma. (A) This irregular hypoechoic mass with acoustic shadowing and an echogenic halo is typical of a carcinoma. (B) Occasionally high-grade tumours may appear well defined, mimicking benign lesions. This shows the importance of performing a core biopsy even on apparently benign-appearing mass lesions. (C) Small echogenic foci of microcalcification associated with malignant lesions may be identified.

tion of the adjacent breast tissue may be apparent, analogous to spiculation on the mammogram. • Posterior acoustic shadowing is frequently observed, due to a reduction in the through transmission of the ultrasound beam via dense tumour tissue. Poorly differentiated, high-grade tumours are more likely to be well defined, without acoustic shadowing (Fig. 52.16B), hence the importance of carrying out a biopsy of solid masses even when the ultrasound appearances are benign. Microcalcifications are sometimes observed, associated with high-grade tumours arising in areas of DCIS, although this is less frequently encountered than with mammography (Fig. 52.16C). Lobular carcinomas can be difficult to demonstrate on ultrasound. They may produce vague abnormalities, such as subtle alterations in echotexture, or the ultrasound findings may even be normal. Doppler examination of malignant masses may show abnormal vessels that are irregular and centrally penetrating. In contrast, benign lesions such as fibroadenomas tend to show displacement of normal vessels around the edge of a lesion. Ultrasound is a useful tool in the local staging of breast cancer preoperatively. It tends to be a better predictor of tumour size than mammography and may detect intraductal tumour extension. Ultrasound may also detect small satellite tumour foci not visible on mammography (Fig. 52.17). It has long been recognized that involvement of axillary lymph nodes is one of the most important prognostic factors for women with breast cancer. Traditionally, the axilla has been staged at the time of surgery by lymph node sampling procedures, sentinel node biopsy, or clearance of the axillary lymph nodes. Surgical clearance of axillary lymph nodes is probably the ‘gold standard’; however, it carries the risk of significant postoperative morbidity, with some women developing disabling lymphoedema in the arm. Ultrasound can identify abnormal nodes preoperatively that can then be biopsied percutaneously under ultrasound guidance (Fig. 52.18), allowing a preoperative diagnosis of lymph node involvement to be made in just over 40% of patients who are lymph node positive27.This enables the more radical axillary clearance to be targeted to those patients with a preoperative diagnosis of axillary disease, with the sampling or sentinel node procedures reserved for those patients with a much lower risk of axillary involvement.

CHAPTER 52

Figure 52.17 A small satellite tumour focus (on the left) is visible adjacent to the main tumour mass. A duct can be appreciated extending between the two lesions.

The differential diagnosis of malignancy Many apparently suspicious findings seen on mammography or ultrasound can be caused by benign disease or even normal breast tissue. The presence of a surgical scar may cause a spiculated mass or an architectural distortion (Fig. 52.19). Radiographers should be encouraged to record the presence and position of any scars when performing a mammogram to aid image interpretation by the film reader.

• THE BREAST

Infection and inflammatory processes in the breast can be mistaken for malignancy on mammography and ultrasound. Breast abscesses are typically encountered in young lactating women.Treatment is with antibiotics and aspiration of the pus, frequently under ultrasound guidance. Inflammation in a nonlactating breast is a more worrying feature, although infections and more unusual inflammatory conditions such as granulomatous mastitis can occur. Skin erythema and oedema may be caused by an underlying carcinoma, termed ‘inflammatory carcinoma’. In this situation, skin thickening and oedema may be the only signs of malignancy recognized on the mammogram. In any case of unexplained inflammation, or when infection fails to resolve, percutaneous biopsy is required to make the diagnosis or exclude malignancy. Radial scars, also called complex sclerosing lesions, can produce a spiculated lesion indistinguishable from malignancy on both mammography and ultrasound (Fig. 52.20). Many of these lesions are asymptomatic and are encountered on screening mammography. Epithelial atypia, DCIS and invasive carcinoma are found in association with radial scars. Superimposition of normal breast tissue may produce apparent masses, distortions, or worrying asymmetric densities on mammography. These summation shadows are usually evaluated with additional mammographic views. Localized compression or paddle views are particularly helpful in deciding whether a lesion is real or just a summation shadow. Ultrasound of the area of mammographic concern can help to determine whether a lesion is truly present.

Microcalcifications Microcalcifications are frequently encountered on routine screening mammograms. In many cases these microcalcifications turn out to be benign, but occasionally are an important feature of DCIS. Some calcifications have a characteristic

Figure 52.18 Axillary lymph nodes can be assessed on the basis of shape and the morphology of the cortex. (A) Nodes are likely to contain tumour if their longitudinal to transverse diameter is less than 2 (the node appears round rather than oval). Nodes are more likely to contain tumour if the cortex is thickened to more than 2 mm. (B) The node has a normal shape, but part of the cortex had a thickness of 3 mm. Both these axillary lymph nodes were biopsied under ultrasound guidance and found to contain tumour.

1185

1186

SECTION 6

• WOMEN’S IMAGING

Benign microcalcifications

Figure 52.19 Postoperative scar. This patient had undergone a previous wide excision for a screen-detected cancer. There is associated deformity with indrawing of the skin. There is a focus of benign calcification, probably the result of fat necrosis.

benign appearance and require no further action. There is a considerable overlap between the appearance of benign and malignant microcalcifications, necessitating a percutaneous biopsy in many cases.

Many benign processes in the breast can cause microcalcifications, including fibrocystic change, duct ectasia, fat necrosis and fibroadenomatoid hyperplasia. Fibroadenomas and papillomas can also become calcified. Sometimes normal structures, such as the skin or small blood vessels, calcify. Calcifications can also develop in atrophic breast lobules or normal stroma. Vascular calcifications have a characteristic ‘tramline’ appearance caused by calcification in both walls of the vessel (Fig. 52.21). Similarly, duct ectasia has a classical appearance that rarely causes diagnostic difficulty. In this condition, coarse rod and branching calcifications are recognized due to calcification of debris within dilated ducts. These calcifications have been described as having a ‘broken needle’ appearance and are usually bilateral (Fig. 52.22A). Sometimes the debris may extrude from the ducts into the adjacent parenchyma, leading to an inflammatory type reaction. Fat necrosis may then occur and the calcifications take on a characteristic ‘lead-pipe’ appearance (Fig. 52.22B). In many cases the diagnosis is obvious, but sometimes biopsy may be required, particularly if the calcifications are unilateral or focal. Fibrocystic change is a common cause of microcalcifications (Fig. 52.23). On a lateral magnification view, layering of calcific fluid contained within microcysts can be appreciated, producing a characteristic ‘teacup’ appearance. However, in many cases, percutaneous biopsy is required to exclude DCIS. Fat necrosis is a frequently encountered cause of benign calcifications, particularly when there is a history of trauma or previous surgery (Fig. 52.24). It may present as ‘egg shell’

Figure 52.20 Radial scar. (A) Radial scar with a stellate appearance but no central mass. It is not possible to differentiate benign and malignant causes of parenchymal distortion on the basis of imaging alone. (B) Sometimes a radial scar can mimic a malignant lesion on ultrasound.

CHAPTER 52

• THE BREAST

calcifications within the wall of an oil cyst or as coarse dystrophic calcifications associated with areas of scarring. Fibroadenomas may become calcified, particularly after the menopause. Classically, the calcifications have a coarse, ‘popcorn’ appearance (Fig. 52.11). However, they can be small and punctuate, necessitating a biopsy to establish the diagnosis. Fibroadenomatoid hyperplasia is an increasingly common cause of microcalcifications detected during screening. Histologically, there are features of a fibroadenoma and fibrocystic change. There is usually no associated mass lesion and in many cases biopsy is required to exclude DCIS (Fig. 52.25). Skin calcifications are characteristically round, well defined, have a lucent centre and are very often bilateral and symmetrical.Talcum powder or deodorants on the skin, as well as tattoo pigments, can mimic microcalcifications.

Malignant microcalcifications

Figure 52.21 Vascular calcifications.

Microcalcifications are found associated with invasive breast cancer and DCIS. Calcifications are more likely to be malignant if they are clustered rather than scattered throughout the breast, if they vary in size and shape (pleomorphic), and if they are found in a ductal or linear distribution. Malignant microcalcifications associated with high histological grade DCIS are classically rod shape and branch. These calcifications are known as casting or comedo microcalcifications and represent necrotic debris within the ducts, hence their linear, branching structure (Fig. 52.26). Approximately one third of malignant microcalcification clusters have an invasive focus within them at surgical excision28. The greater the number of flecks of

Figure 52.22 Duct ectasia. (A) Broken needle appearance, typical of duct ectasia. (B) Sometimes thicker, more localized calcifications can be seen, giving a ‘lead-pipe’ appearance.

1187

1188

SECTION 6

• WOMEN’S IMAGING

Figure 52.23 Fibrocystic change. (A) ‘Teacups’ representing the layering out of calcific material in the dependent portion of microcysts on a lateral magnification view. (B) As calcifications associated with areas of fibrocystic change may not exhibit this characteristic appearance, stereotactic core biopsy is required.

microcalcification associated with an area of DCIS, the greater the risk of invasive disease28. In the screening setting, it is often the presence of mammographically visible calcifications associated with high-grade DCIS that leads to the diagnosis of small, high-grade cancers29. Calcifications are much less frequently found in low-grade DCIS, as there is usually no intraductal necrosis. When they

do occur, they are clustered, but otherwise have a nonspecific appearance. The sensitivity of ultrasound for detecting DCIS is significantly lower than mammography, which is one of the reasons why ultrasound is not a useful screening test for breast cancer. However, ultrasound may be able to identify areas of microcalcifications seen on a mammogram, aiding percutaneous biopsy.

Figure 52.24 ‘Egg shell’ calcifications of fat necrosis.

Figure 52.25 Small cluster of indeterminate microcalcifications. Stereotactic biopsy revealed fibroadenomatoid change.

CHAPTER 52

Figure 52.26 Ductal carcinoma in-situ (DCIS). Mammography shows the segmental distribution of pleomorphic microcalcifications. Granular, rod-shaped and branching calcifications can be identified. The appearances are typical of high-grade DCIS.

OTHER IMAGING TECHNIQUES Magnetic resonance imaging Although mammography and ultrasound remain the most frequently used techniques for imaging the breast, contrastenhanced MRI is becoming increasingly important, largely because of its high sensitivity for detecting invasive breast cancer, which approaches 100% in many studies. There are problems with specificity, with some benign lesions and even normal breast tissue having a worrying appearance on MR imaging. The problems with specificity are compounded by limited availability of MR-guided localization and biopsy systems. MRI is the technique of choice for assessing the integrity of breast implants. It is more accurate than mammography, ultrasound, or clinical examination in identifying implant failure. Unenhanced MRI is used to assess breast implants for complications.

Technique Successful breast MR studies require high field strength magnets (1.5 Tesla) and the use of a dedicated breast coil. Some breast coils have inbuilt compression devices to stabilize the breast and reduce the number of slices required to cover the whole of the breast. Patients are examined in the prone position with the breast hanging down into the coil. The intravenous injection of gadolinium-based contrast agent is required. It is the presence of abnormal vasculature within the lesion that enables detection. Some method of eliminating the signal from fat is required as an enhancing lesion and fat display similar high signal on a T1-weighted image. Fat suppression may be active or passive: active fat suppression is typically achieved by the use of spectrally selective pulse sequences to suppress the signal from fat; passive fat suppression involves subtraction of the unenhanced images from the enhanced images. Subtraction

• THE BREAST

allows faster imaging, with good spatial and temporal resolution, but it requires no patient movement between the two sets of images. New methods of active fat suppression, such as parallel imaging, including the sensitivity encoding technique (SENSE), allow fat suppression to be achieved with shorter examination times while maintaining good spatial and temporal resolution30. Fast, 3D gradient echo pulse sequences provide the optimum method for imaging small lesions.Temporal resolution is important because the optimum contrast between malignancy and normal breast tissue is achieved in the first 2 min following the injection of gadolinium contrast agent. Later, normal breast tissue may start to show nonspecific enhancement, masking the presence of malignancy. Other signs of malignancy, such as a rapid uptake of contrast agent followed by a ‘washout’ phase, may only be apparent if images are acquired dynamically every minute over a period of 6–7 min after the injection of gadolinium. Unfortunately, increased spatial resolution (thin slices) can only be achieved at the expense of an increased examination time. Improving temporal resolution allows rapid dynamic scanning but at the expense of spatial resolution or the volume of the breast imaged. With modern equipment it should be possible to achieve a slice thickness of 3–4 mm while maintaining a temporal resolution of 60–90 s, covering the whole of both breasts.

Lesion characterization There are two main approaches to image interpretation: the first relates to lesion morphology and the second to assessment of enhancement kinetics. The architectural features that indicate benign and malignant disease are similar to those already described for mammography and ultrasound. Benign lesions tend to be well defined with smooth margins whereas malignant lesions are poorly defined and may show spiculation or parenchymal deformity. Malignant lesions tend to enhance rapidly following the injection of contrast agent and may show characteristic ring enhancement. Dynamic MRI enables more detailed enhancement curves to be calculated to aid characterization. Malignant lesions usually show a rapid uptake of contrast agent in the initial phase of the examination, followed by a washout or plateau in the intermediate and late periods after injection, whereas benign lesions exhibit a steady increase in signal intensity throughout the time course of the examination31. One of the strengths of breast MR imaging is that invasive cancer can be effectively excluded with a high degree of certainty if no enhancement is seen. Investigators use a combination of architectural features and enhancement kinetics to differentiate benign from malignant lesions. However, there is some overlap in the enhancement characteristics of benign and malignant lesions. Sinister patterns of contrast enhancement have been observed in benign conditions, including fibroadenomas, fat necrosis and fibrocystic change. Even normal breast tissue may enhance and this enhancement is in part dependent on the phase of the menstrual cycle. Normal breast tissue is more likely to enhance

1189

1190

SECTION 6

• WOMEN’S IMAGING

during the middle of the cycle, between the sixth and seventeenth days32. Recent surgery or radiotherapy can interfere with image interpretation. It used to be recommended that at least 12 months should elapse between breast radiotherapy and an MRI study, but more recent work has suggested that enhancement patterns return to normal between 3 and 6 months after radiotherapy33. Percutaneous breast biopsy (FNA, core, or vacuumassisted biopsy) rarely interferes with MRI interpretation.

Indications for breast MRI Contrast-enhanced breast MRI is most frequently used for local staging of primary breast cancer. MRI is the most accurate technique for sizing invasive breast carcinomas and will sometimes show unsuspected multifocal disease in the same breast or even additional tumour foci in the contralateral breast34. Additional information on tumour extent can be expected in 15–30% of patients, which can lead to a change in the therapeutic approach, avoiding inappropriate breastconserving surgery. There is an argument for breast MRI in all patients considering breast-conserving surgery. However, due to scarcity of resources and specificity issues, MRI is usually reserved for patients where estimating tumour size is proving difficult by conventional methods, including mammographically occult lesions, patients with mammographically dense breasts, and where there is significant discrepancy between size estimations at mammography, ultrasound and clinical examination. Another group of patients who benefit from preoperative staging with MRI are those whose carcinomas have lobular features. Lobular carcinomas are more likely to be multifocal compared to ductal NST tumours. They are more difficult to detect and their size is more difficult to measure by conventional methods because of their infiltrating growth pattern. In approximately 50% of such patients MRI will show more extensive tumour (Fig. 52.27)35. Another important role of MRI is identifying an occult primary tumour in women presenting with malignant axillary lymphadenopathy with a normal mammogram and breast ultrasound. In this situation, MRI is highly sensitive for identifying an occult primary. MRI is also useful in the postsurgical breast, differentiating surgical scarring from tumour recurrence. MRI can help to assess the response to treatment in women receiving neoadjuvant chemotherapy for locally advanced primary breast cancers. It can recognize nonresponders to treatment earlier than other imaging methods by demonstrating a reduction in lesion size, or a change in the enhancement pattern, with the level of enhancement reducing or taking on a more benign appearance. MRI can be used to screen younger women with a high familial risk of breast cancer. Some of these women (e.g. known gene mutation carriers) have a lifetime risk of developing breast cancer of around 85%. In these younger women the sensitivity of mammography for detecting malignancy is low, largely due to the presence of mammographically dense breast parenchyma. Screening with MRI is superior to mammography in detecting

invasive breast cancer in such women, although mammography remains more sensitive for detecting DCIS36,37.

Pitfalls of MRI Although the sensitivity of MRI for invasive breast cancers approaches 100%, its sensitivity for detecting DCIS is more variable. Specificity is also a problem. When MRI is used for staging breast cancer, problems arise when additional enhancing lesions are detected away from the primary tumour site, as these have to be differentiated from additional tumour foci, incidental benign lesions and areas of enhancing normal glandular breast tissue. In such cases the architectural features of the lesion and enhancement kinetics should be assessed, and the MRI findings correlated with those of mammography. Probably most useful is a targeted, second-look ultrasound of the area. In most cases, ultrasound will identify any additional lesions and facilitate image-guided biopsy. For younger women, repeating the examination at a different phase of the menstrual cycle may exclude spontaneous hormone-induced enhancement as a cause of an additional enhancing focus. For lesions that are considered low risk, follow-up MRI after a suitable period of time is acceptable. There is concern that the increasing use of breast MRI may result in unnecessary percutaneous biopsies or diagnostic surgery.The problems of specificity are compounded by the difficulties associated with MRI-guided biopsy and localization, as these techniques are time consuming and problematic. However, when there is no mammographic or ultrasound correlate, MRI-guided biopsy may be necessary. Dedicated biopsy coils and MRI-compatible needles are now becoming commercially available.

MRI for imaging breast implants MRI is the technique of choice for assessing the integrity of breast implants, with a sensitivity and specificity of over 90%. When imaging breast implants, no contrast agent is required unless malignancy is suspected. Imaging should be performed in the prone position using a dedicated breast coil. The main goal is to determine whether the implant has ruptured and, if so, to establish the location of the leaked filler (usually silicon). When implants fail, the rupture may be either intracapsular or extracapsular: intracapsular rupture occurs when silicon has escaped from the plastic shell of the implant, but is contained within the fibrous implant capsule (Fig. 52.28); signs of intracapsular rupture include the ‘wavy line’, ‘linguini’, ‘key-hole’ and ‘salad oil’ sign38. False-positive interpretations can be made when normal implant folds are mistaken for signs of rupture. Extracapsular rupture is diagnosed when silicon is demonstrated outside the fibrous capsule. In this situation, free silicon, silicon granulomas, or silicon in axillary lymph nodes may be demonstrated (Fig. 52.29).

Scintimammography Scintimammography is a nuclear medicine technique. It was developed following the observation that many breast cancers

CHAPTER 52

• THE BREAST

280 260 240 220 Signal intensity(SI)

200 180 160 140 120 100 80 60 40 20 0

C

0

1

2

3 Time(min)

4

5

6

Figure 52.27 MRI for local tumour staging. This patient presented with a mass in the left breast. Mammography showed a spiculated lesion lying centrally within the breast, best appreciated on the CC view (A). Biopsy indicated a carcinoma with lobular features. MRI confirmed the presence of a malignant spiculated lesion (B) with a typically malignant enhancement curve (rapid uptake of contrast agent followed by a washout phase) (C). An additional tumour focus was identified away from the primary tumour site (D). This was confirmed at biopsy.

show uptake of isotopes used in cardiac imaging—for example, 99mTc-MIBI (Sestamibi). Indications for use overlap with those for MRI, including local staging, searching for a mammographically occult primary and detecting recurrence in the postsurgical breast. Scintimammography has failed to establish a place in routine practice due to problems with spatial resolution and its inability to detect DCIS.

BREAST CANCER SCREENING Introduction Breast cancer mortality in the UK is amongst the highest in the world.The causes of breast cancer are not well understood and, in the absence of any effective preventative measures, much effort and health care resources have been focused on

1191

1192

SECTION 6

• WOMEN’S IMAGING

Figure 52.28 Intracapsular implant rupture. On these T2-weighted fast spin-echo images, the plastic shell of the left breast implant can be seen floating within the silicon, producing a ‘wavy line’ or ‘linguini’ sign. Note the presence of a couple of bright dots of water-like material, the ‘salad oil’ sign.

the quest to reduce breast cancer mortality by early detection through screening. A number of randomized controlled trials (RCTs) and case control studies carried out since the mid 1960s have shown that screening by mammography can reduce breast cancer mortality. The UK National Health Service Breast Screening Programme was set up following the publication of the Forrest Report in 198639. This document, commissioned by the UK Department of Health under the chairmanship of Professor Sir Patrick Forrest, reviewed the scientific evidence for population breast cancer screening. It recommended the immediate introduction of screening by mammography in the UK. Within a year of publication, population breast cancer screening, free at the point of delivery, was introduced into the UK National Health Service (NHS). This was the first population-based breast screening programme in the world. Currently in the UK, breast cancer screening by mammography is provided for all women over the age of 50. Women between the ages of 50 and 70 are invited every 3 years. Women over 70 are not invited but are encouraged to attend by self referral. Two-view mammography is now used for all screens. The mammograms are read by at least one film reader, and in the majority of units the mammograms are double read.

The evidence for screening Data from RCTs provide the strongest evidence of the efficacy of screening in reducing breast cancer mortality. The design of RCTs enables the elimination of lead-time bias. The majority of the RCTs of screening were carried out in Sweden. The latest overview of these trials was published in 2002 and included data from Malmo, Gothenburg, Stockholm and the Ostergotland arm of the Two Counties study. Data from the Kopparberg arm of the Two Counties study was not made available. Almost a quarter of a million women were included in these studies, with approximately half being invited for screening and the other half making up the control group. The median trial time was 6.5 years and the median followup 15.8 years. The overall results indicated a 21% reduction in breast cancer mortality. The mortality reduction was largest in women aged 60–69 (33%). Due to recent criticisms of RCTs of breast screening, which suggested that the overall mortality may be higher in those screened due to adverse effects of treatment, this study also looked at total cause mortality. This showed a relative risk of dying of any cause in the study arm of 0.98, which was of borderline statistical significance40. The precise mortality reduction attributable to screening is controversial as RCTs may underestimate the benefit of screening due to nonattendance and contamination (mammography occurring within the control group). A recent paper has suggested that regular attendance for mammographic screening may result in a 63% reduction in breast cancer deaths41. Figure 52.29 Extracapsular implant rupture. (A) A collection of free silicon is seen anterior to this ruptured left breast implant. (B) Ultrasound may be useful in the diagnosis of extracapsular rupture, with free silicon or silicon granulomas having a typical ‘snow storm’ appearance.

Which age groups should be screened? There is definite evidence from RCTs of screening of a reduction in mortality in women aged 55–69. Previous meta-analyses have supported the introduction of screening at age 50 but

CHAPTER 52

• THE BREAST

these data are based on 10-year age bands. Data analysis based on 5-year age bands of screening women aged 50–55 has never shown a mortality benefit in this age group.The reasons for this are unclear but it has been postulated that this may be due to the unusual behaviour of breast cancer in perimenopausal women. There is no evidence from RCTs to support the screening of women over the age of 70. However, the number of women over the age of 70 in these studies is low. Although the mammograms of older women are easy to read and the incidence of cancer is high, there would be a significant risk of overdiagnosis in this age group. Overdiagnosis is the detection and treatment of cancers that would not become clinically apparent or threaten life. Meta-analysis of RCTs screening women aged 40–49 at randomization have shown a statistically significant mortality reduction of 29%42. Failure of the most recent meta-analysis in 2002 to show a mortality benefit in this age group may be because of the noninclusion of the Kopparberg arm of the Two Counties study. The Malmo43 and Gothenburg44 studies have both shown statistically significant mortality reductions in this age group. As breast cancer is only half as common in women in their forties compared with women in their fifties, some authors have suggested that presenting data in terms of percentage reduction in population mortality may be misleading. On the other hand, preventing breast cancer deaths in younger women will result in a larger number of life years gained and it has been shown that breast cancers arising in women in their forties account for 34% of life years lost to breast cancer. The RCTs of screening were not designed to look at particular age groups and such subanalysis has been criticised. In particular, a proportion of the screening episodes occurring in women aged 40–49 at randomization actually occurred when women were over the age of 50. In addition, women in the control groups of these studies were not always screened at 50. Therefore, it is possible that part of the mortality benefit demonstrated in these women may be due to screening episodes over the age of 50. The low cancer incidence in women under the age of 50 results in the specificity of both recall and surgical biopsy being lower than that in older women. It has been said that there is reduced mammographic sensitivity in women in their forties; however, recent data suggest that the use of two views and the high film density substantially improve sensitivity in younger women. The most important hindrance to screening younger women is the short lead-time of screening. The lead-time of screening is that time between mammographic detection and clinical presentation.The presence of this short lead-time indicates the need to screen more frequently in women under the age of 50.The ideal screening interval would be 12 months. The high frequency of screening required in younger women and the low incidence of breast cancer has led some to question the cost-effectiveness of screening in this age group. However, these disadvantages may be partly negated by the large number of life years gained per life saved.

cancer detection of 24% and a 15% lower recall rate using two views45. The cost per cancer detected was similar using one view or two views. Two-view screening resulted in a 54% increase in invasive cancer detection less than 10 mm in size46. Initial data suggest that introduction of two views at every screen has resulted in an increase in small invasive cancer detection. There is no evidence that adding physical examination at screening episodes is associated with an improvement in breast cancer mortality reduction.

Number of views

Screening intervals

Mammographic screening is best performed using two views: MLO and CC. The RCT of screening with one view versus two views at prevalent round (first round) showed an increased

The interval at which a screening mammogram needs to be repeated is related to the lead-time of screening. In breast cancer the lead-time of mammographic screening is age related.

Who should interpret mammographic images? In the UK, interpretation of screening mammograms is limited to practitioners who read a high volume of images (greater than 5000 examinations per year) and it is recommended that film readers also participate in screening assessment.These recommendations are based on a number of studies that suggest the total number of mammograms read and access to regular feedback are the most important factors relating to radiologists’ performance47. A recent study using a difficult test set (PERFORMS 2) indicated that high volume readers had a significantly increased sensitivity for detection of breast cancer compared to medium and low volume readers48. In the UK, there has been a national shortage of radiologists, and in particular breast screening radiologists. This had led to the introduction of radiographer film readers. Radiographers have been shown to have identical sensitivity and specificity when compared with screening radiologists once they have been trained.

How should the images be interpreted? In independent double reading, if one of the readers requests recall, the patient is recalled without discussion. In consensus double reading, cases with disparate opinions are discussed by the two film readers and a consensus achieved. When arbitration is used, a third reader decides on whether a film recalled by one reader but not by another, necessitates patient recall. Recent data from the UK screening programme have shown that double reading with arbitration results in the best small invasive cancer detection rate and at an acceptable recall rate49.

The screening process and assessment Screening mammograms are carried out by female radiographers, either at static sites or using mobile vans. Attendance is best when using timed appointments, which can be changed by telephone. Approximately 5% of women are called back for assessment and such assessments are carried out by a multidisciplinary team at a static site. Breast screening assessment involves a combination of extra mammographic views, ultrasound and physical examination. Figure 52.30 is a flow diagram of the assessment process Approximately one in seven of those recalled have breast cancer. Four out of five operations provoked by screening are treatment operations for breast cancer.

1193

1194

SECTION 6

• WOMEN’S IMAGING

Screening

Normal/ benign

Routine recall >93%

Abnormal (10 cm) that may contain several hypoechoic areas. The hypoechoic areas may correspond to atretic follicles or to regions of haemorrhage within the ovary. The symptoms associated with OHSS usually begin 5–8 d after hCG is given, but they can be most severe in patients who actually achieve pregnancy. Recent studies have shown that hyperstimulation is unlikely in women whose ovaries contain several large (>15 mm) follicles, and tends to occur when there are several small or intermediate-sized follicles6.With supportive therapy, this syndrome usually regresses spontaneously.

ASSISTED REPRODUCTION TECHNOLOGY The success of in vitro fertilization (IVF) has been due in part to the US identification of follicular maturity and subsequent transvaginal US-guided oocyte recovery techniques. The principles of IVF involve controlled ovarian hyperstimulation using gonadotrophins. This allows several follicles to develop, as opposed to just one dominant follicle in a natural cycle, allowing for the recruitment of several oocytes. The first stage of IVF involves the administration of a gonadotrophin-releasing hormone (GnRH) analogue to downregulate the pituitary. Follicular development is then induced through the administration of follicle-stimulating hormone (FSH). The follicles are monitored using TV US until at least three large follicles are present. Ovulation can be triggered once follicular maturity has been demonstrated by human chorionic gonadotrophin. This allows for suitable timing of ovulation so that the oocytes can be aspirated prior to the occurrence of in vivo ovulation. Follicular aspiration is possible with a needle-guide attachment to the vaginal probe for subsequent IVF or intracytoplasmic sperm injection (ICSI). For these aspiration techniques, a long (30-cm) 18-gauge needle is used, which is scored at the tip, making it easier to visualize under US. Transvaginal sonography (TVS)-guided follicular aspiration

CHAPTER 53

•

IMAGING IN OBSTETRICS AND INFERTILITY

Figure 53.2 Congenital uterine abnormalities HSG images. (A) Bicornis bicollis (single vagina, two cervices, two separate uterine horns). (B) Bicornuate (single vagina, single cervix, two separate uterine horns). (C) Septate – preoperative (i) and initial post resection (ii) images. (D) Subseptate – deep fundal indentation. (E) Arcuate uterine fundus (F) Unicornuate (HSG images reproduced from Chapter 94B in Grainger & Allison, 4th Edition, written by Anne Hemingway).

1203

1204

SECTION 6

• WOMEN’S IMAGING

Figure 53.5 Ovarian hyperstimulation. The ovary is massively enlarged and contains several follicles, follicular cysts and stromal oedema.

Figure 53.3 Adenomyosis. Abnormal uterine margins with contrast penetrating the myometrium.

has now become the preferred procedure of choice for oocyte retrieval over the previously used laparoscopic technique7. On the second or third day following fertilization in vitro, the selected embryos are transferred to the recipient uterus. The favourability of the endometrium for implantation is difficult to predict with US. Studies have looked at endometrial thickness8 and endometrium volume using 3D US and have found no correlation on pregnancy outcomes. Pregnancy rates have been found to be lower in patients in whom the endometrium volume was below 2 ml and no pregnancies were observed in cases with an endometrial volume less than 1 ml9. Doppler studies of the uterine artery and endometrial vascularity may prove helpful in predicting favourable uterine receptivity. Increased pulsatility indices in the uterine arteries have been found to be associated with a reduced implantation rate10. Increased endometrial vascularity is also associated with successful conception cycles11. Figure 53.4 Uterine polyps. Well-defined filling defects are present in the lower uterine segment.

OBSTETRIC IMAGING NORMAL PREGNANCY Transvaginal ultrasound (TVS) plays an important role in confirming the site and viability of an early pregnancy, particularly in the context of bleeding in the first trimester. US is more accurate in establishing gestational age than dates from the menstrual history and can determine chorionicity in multiple pregnancies. An endometrial decidual reaction can be visualized by TVS following implantation 3–5 days prior to the first missed

period. It, however, is not a reliable sign and does not confirm an intrauterine pregnancy. With implantation trophoblastic hCG gains access to the maternal circulation and a sensitive urinary pregnancy test will be positive at this stage. The first reliable US indicator of pregnancy is the presence of a chorionic sac, or a gestational sac, and is usually evident on TVS at 4.5 weeks post last menstrual period (LMP). A normal gestational sac is round or oval and consists of a central sonolucent area surrounded by an echogenic ring, which will later form

CHAPTER 53

the placenta (Fig. 53.6). It lies within the decidua and therefore lies in an eccentric position within the uterus. It should not be confused with a pseudogestational sac in cases of ectopic pregnancy or fluid within the endometrial cavity, which lie centrally within the endometrial cavity. The gestational sac

•

IMAGING IN OBSTETRICS AND INFERTILITY

contains the embryonic disk within a small amniotic sac and the yolk sac. The yolk sac is the first structure to be visualized on TVS within the chorionic sac at 5.5 weeks’ menstrual age. The yolk sac is a rounded structure with a central lucency. It should be evident once the gestational sac measures 10 mm. Visualization of the yolk sac precedes that of the embryo and amniotic sac by 3–7 d. Initially the yolk sac occupies much of the chorionic sac, though as the embryo grows it pushes the yolk sac to the side. The yolk sac degenerates at 12 weeks’ LMP in normal pregnancies. The embryo is first seen on high-resolution images as a thickening on the margin of the yolk sac.With high resolution, the heartbeat is seen as a regular flutter in the embryo, first evident at 5 mm crown to rump length (CRL) (6.2 weeks). Thus it is possible to see viable embryos without heartbeats. In such cases, a follow-up study in 5–7 d will almost always demonstrate the heartbeat in healthy embryos. On TVS embryos should be seen at mean sac diameters (MSDs) of 18 mm. With lower resolution abdominal US, embryos should be seen with MSDs of 25 mm.The gestational sac size is useful in early pregnancy in the diagnosis of miscarriage and anembryonic pregnancies. It should be measured in three orthogonal planes and the MSD calculated by averaging these measurements.

NONVIABLE PREGNANCY Miscarriages occur in up to 25 per cent of pregnancies. They may present with vaginal bleeding or be silent. It is important to adhere to strict guidelines when determining the viability of an early pregnancy12 and to be familiar with the normal first trimester US landmarks. Hypoechoic areas behind the choriodecidua may be related to retrochorionic haemorrhage. Small haematomas may resolve but larger bleeds are associated with a high rate of pregnancy loss. The gestational sac may appear irregular in outline or appear to be collapsing in a failing pregnancy. A fetal pole should be evident if the mean sac diameter is greater than 20 mm. An anembryonic pregnancy occurs when the embryo fails to form and the gestational sac remains empty with a mean sac diameter greater than 20 mm. Embryonic demise can be diagnosed if no cardiac activity is evident in a fetal pole measuring more than 6 mm. If uncertainty remains and the gestational sac measures less than 20 mm or the fetal pole length less than 6 mm, then a repeat examination should be performed after an interval of 7 d.

ECTOPIC PREGNANCY

Figure 53.6 TVS of early pregnancy. (A) TVS at 6 weeks’ gestation demonstrates a gestational sac and an echogenic ring within the deciduas. (B) A yolk sac is identified by 5.5 weeks’ gestation as a circular echogenic ring with a central lucency. (C) Gestation sac containing a yolk sac – Y and embryo – E.

The incidence of ectopic pregnancy has increased significantly in the past few decades. This is in part due to an increased incidence of pelvic inflammatory disease and assisted reproduction techniques. Based on hospital discharge data, the incidence of ectopic pregnancy has risen from 3.45 cases per 1000 pregnancies between 1966 and 1970 to 15.5 cases per 1000 pregnancies from 1994 to 199613.

1205

1206

SECTION 6

• WOMEN’S IMAGING

Figure 53.7 TVS of ectopic pregnancy. (A) Uterus containing an intraluminal fluid collection surrounded by a mild decidual reaction. (B) Left adnexa contains an echogenic ring containing a live embryo consistent with an unruptured ectopic pregnancy. (C) Unruptured ectopic pregnancy surrounded by blood within a distended tube.

TVS is the first-line investigation in haemodynamically stable women presenting with suspected ectopic pregnancy and is accurate in up to 90 per cent of cases14. Unruptured ectopic pregnancies appear as a complex adenexal mass separate from both the uterus and the ovary in 80 per cent of cases (Fig. 53.7). The mass may contain a yolk sac or evidence of an embryo, which is the only true sign of an ectopic pregnancy. A hyperechoic ring may surround a gestational sac, the doughnut or bagel sign. Following Fallopian tube rupture or tubal abortion, intraperitoneal fluid is present and is visible with the pouch of Douglas on TVS.The uterus should be examined for evidence of an intrauterine pregnancy. In a low-risk patient, with no history of pelvic inflammatory disease (PID) or infertility, the risk of a heterotopic pregnancy is one in 30 00015, though this can increase up to 7-fold in women with predisposing risk factors16. The endometrial cavity may contain fluid and areas of hyperechogenicity, the so-called pseudo sac. This lies centrally within the uterine cavity and not in the usual eccentric position of a normal gestational sac. An ectopic pregnancy can be suspected if the transvaginal US examination does not detect an intrauterine gestational sac when the β-hCG level is higher than 1000 mIU per mL. This has a positive predictive value of 86 per cent and a sensitivity of 93 per cent17.

ESTABLISHING GESTATIONAL AGE It is important to accurately establish the gestational age during pregnancy in order to estimate the date of delivery and to correctly time biochemical screening in the second and

third trimesters. The most accurate assessment of gestational age is the crown rump length18.This is more accurate than the menstrual history alone or the gestational sac size. During the second and third trimesters, fetal biometry can provide some assessment of gestational age though there is a greater scatter in distribution. The fetal head circumference, biparietal diameter and abdominal circumference can be correlated with appropriate charts19,20.

MULTIPLE PREGNANCIES Of twin pregnancies, 80 per cent are dizygotic. By implication two separate amniotic sacs are present in dizygotic twin pregnancies and the placentas are always dichorionic but may be fused or unfused (Fig. 53.8). The remaining 20 per cent of twin pregnancies are monozygotic twins resulting from the fertilization of one ovum followed by division of the zygote. The chorionicity and the amnionicity of a monozygotic pregnancy depends on the time of division of the zygote (Table 53.2). The chorionicity can be assessed by US most effectively between 9 and 10 weeks’ gestation. A thick membrane is seen to separate diamniotic dichorionic twins and the so-called lambda sign can be seen (Fig. 53.9). A thin membrane is evident between monochorionic diamniotic twins and no membrane at all is present in cases of rarer monoamniotic twinning (Fig. 53.10). It is important to determine the chorionicity of a twin pregnancy as monochorionic twins are associated with an increased risk of fetal malformations, prematurity and twinto-twin transfusion syndrome.

CHAPTER 53

•

IMAGING IN OBSTETRICS AND INFERTILITY

Table 53.2 MONOZYGOTIC TWIN TYPES Time of division

Type of twinning

13 d

Conjoined twins

Monoamniotic Monochorionic

Monochorionic Diamniotic

Dichorionic diamniotic [fused placenta]

Dichorionic diamniotic [separate placenta]

Figure 53.8 Different types of fetal membrane patterns in twinning. (From Twining P, McHugo J M, Pilling D W (eds.) 2006 Textbook of Fetal Abnormalities, 2nd edition, Churchill Livingstone.)

Figure 53.9 Dichorionic diamniotic twin pregnancy. (A) A thick membrane separates the two gestation sacs, known as the lambda sign. (B) 3D US demonstration of the lambda sign.

1207

1208

SECTION 6

• WOMEN’S IMAGING

The nuchal translucency can be measured between 11 weeks and 13 weeks 6 d (Fig. 53.11). An increase in nuchal translucency is associated not only with chromosomal abnormalities but also structural abnormalities and some rare genetic syndromes (Table 53.3). Therefore a detailed assessment of fetal anatomy should be made in the presence of a raised nuchal translucency and normal karyotype.

FIRST TRIMESTER DETECTION OF FETAL ABNORMALITIES

Figure 53.10 Monochorionic diamniotic twin pregnancy. A thin membrane between the two gestation sacs is shown.

NUCHAL TRANSLUCENCY SCREENING It has been observed that there is an increase in lymphatic fluid accumulation under the skin of the fetal neck in cases of Down’s syndrome21. The measurement of this, the nuchal translucency, can be used as a screening tool for Down’s syndrome. When combined with the maternal age-related risk for trisomy 21, nuchal translucency screening detects up to 80 per cent of cases of Down’s syndrome with a false-positive rate of 5 per cent22. The detection rate can be increased to almost 90 per cent when combined with first or second trimester biochemical screening23.

Fetal anomaly US is routinely performed in the UK at around 20 weeks’ gestation. It has been proposed that the first trimester is the optimum time to assess the fetus along with nuchal translucency measurements. However, comparable detection rates are only obtained with TVS at this gestation period, which is not routinely available and is limited by reduced US probe manoeuverability. In addition some conditions have variable onset and may only present later in gestation. A number of conditions are readily detected at a first trimester dating and these include anencephaly, gastroschisis and omphalocoele (Fig. 53.12).

SECOND TRIMESTER ANOMALY SCREENING Second trimester US is widely used to detect fetal anomalies. The published sensitivity of antenatal anomaly screening varies from 34 per cent24 to 74 per cent25. These studies were however conducted at a time when the standard of US equipment and expertise are not as they are today. For a full description of the range of antenatal abnormalities and their detection the reader is referred to specialist literature26,27.

Figure 53.11 Nuchal translucency measurement. (A) The nuchal translucency should be measured on a magnified image of the fetal head and neck in the neutral position. The amniotic membrane is seen separate to the nuchal membrane. (B) An increased nuchal translucency associated with Down’s syndrome.

CHAPTER 53

Table 53.3 CAUSES OF AN INCREASED NUCHAL TRANSLUCENCY Chromosomal abnormalities Cardiac abnormalities Diaphragmatic hernia Exomphalos Skeletal dysplasias Noonan’s syndrome Myotonic dystrophy Spinal muscle atrophy Smith-Lemli-Opitz syndrome Congenital adrenal hyperplasia Fetal akinesia deformation sequence

•

IMAGING IN OBSTETRICS AND INFERTILITY

Abnormalities of the central nervous system account for up to one-third of fetal malformations. The intracranial structures can be clearly visualized in the majority of patients if the correct sectional planes are obtained through the thalami, ventricles and cerebellum. The lateral ventricles should be routinely measured and should measure less than 10 mm. Associated abnormalities are present in up to 67 per cent of cases. These include neural tube defects, congenital infection and chromosomal abnormalities (Fig. 53.13). Congenital heart defects are the most common neonatal abnormalities, with an incidence of eight per 1000 (Fig. 53.14)28. The sensitivity of antenatal US in detecting congenital cardiac defects is variable, with reported detection rates ranging from 4 per cent to 78 per cent29. The four-chamber view is the

Figure 53.12 First trimester fetal anomalies. (A) Omphalocoele at 14 weeks’ gestation. (B) Gastroschisis containing free loops of bowel. (C) Large septated cystic hygroma in a fetus with Turner’s syndrome. (D) Anencephaly at 13 weeks’ gestation. The cranium is absent and a thin membrane is present.

1209

1210

SECTION 6

• WOMEN’S IMAGING

Figure 53.13 Neural tube defects. (A) Lemon-shaped skull associated with a neural tube defect. The cisterna magna is effaced due to an Arnold–Chiari malformation. (B) In the same fetus the coronal view of the lumbar shows splaying of the posterior ossification centres in a fetus with an open neural tube defect.

Figure 53.14 Fetal cardiac imaging. (A) Large atrioventricular septal defect in diastole. (B and C) Ventricular septal defect confirmed with myocardial motion imaging. (From Twining26 2000 with the permission of Churchill Livingstone, Edinburgh.)

CHAPTER 53

most easily obtained view of the fetal heart and allows examination of the atrioventricular connection, ventricular septum and pulmonary veins. A number of critical cardiac anomalies such as transposition of the great arteries will have a normal four-chamber view and the detection of fetal cardiac abnormalities can be further increased by imaging the cardiac outflow tracts30. Space-occupying lung lesions may cause a deviation in the cardiac axis. Congenital cystic adenomatoid malformations and pulmonary sequestrations are well demonstrated by ante-

•

IMAGING IN OBSTETRICS AND INFERTILITY

natal US (Fig. 53.15). They often become less apparent in the third trimester as the echogenecity of the normal, fluid-filled lung increases. Renal tract anomalies are relatively common, accounting for 20 per cent of fetal abnormalities. The significance of antenatally detected renal abnormalities remains contentious in particular mild renal pelvis dilatation. However, severe bilateral renal abnormalities account for 10 per cent of all terminations for lethal abnormalities (Fig. 53.16).

Figure 53.15 Congential cystic adenomatoid malformation. (A) Axial image through the fetal thorax shows significant deviation of the heart to the left by the hyperechoic right-sided cystic adenomatoid malformation. (B) The involvement of the whole right lower lobe is confirmed on the sagittal image.

Figure 53.16 Renal abnormalities. (A) Unilateral pelvi-ureteric junction obstruction. Coronal imaging demonstrating hydronephrosis and cortical thinning. (B) Bilateral multicystic dysplastic kidney. Multiple, peripheral cortical cysts are present bilaterally. Severe oligohydramnios is present. (From Twining P, McHugo J M, Pilling D W 2000 Textbook of Fetal Abnormalities. Churchill Livingstone, Edinburgh, with the permission of Churchill Livingstone, Edinburgh.)

1211

1212

SECTION 6

• WOMEN’S IMAGING

HYDROPS

THREE- AND FOUR-DIMENSIONAL ULTRASOUND

Fetal hydrops is defined as the presence of excess fluid in more than one body cavity, such as ascites, pleural effusions and pericardial effusions (Fig. 53.17). With the introduction of prophylactic anti-D immunoglobulin, immune hydrops is no longer the most common aetiology. Chromosomal causes predominate in cases presenting before 24 weeks’ gestation (Table 53.4). After this gestation period cardiac causes, including structural cardiac disease and cardiac arrthymias, account for one-quarter of cases. US remains essential in the initial assessment of the hydropic fetus, to exclude structural abnormalities, and to assess fetal well-being using biophysical assessment and umbilical artery Doppler. Middle cerebral artery Doppler may also be useful in the assessment of the hydropic fetus. Raised peak systolic velocities in the middle cerebral artery are seen in cases of moderate to severe fetal anaemia with a sensitivity of 100 per cent31. Doppler of the middle cerebral artery is also a useful noninvasive tool for the assessment for fetal anaemia in monitoring patients with known red-cell alloimmunization. This removes the need for regular cordocentesis with its associated morbidity and mortality.

The development of three-dimensional (3D) and fourdimensional (4D) US of the fetus has received much attention for its social benefits. The clinical applications however are still evolving and the huge potential for the diagnosis of fetal abnormalities is being explored32,33. 3D US has already proven to be beneficial in the diagnosis of facial clefts, skeletal abnormalities and neural tube defects (Fig. 53.18). Attention is now turning to fetal cardiac applications. Recent technological advances using 4D echocardiography allow an automatic volume acquisition of the fetal heart.The 3D data can then be analyzed in different planes to examine the short- and long-axis views of the fetal heart and outflow tracts.

DISORDERS OF PLACENTATION US has a pivotal role in the detection of placental abnormalities and determining placental site for invasive procedures such as chorionic villus sampling. In a patient with bleeding in the second or third trimester, sonography is essential to delineate placenta praevia. Placenta praevia is defined as a placenta that is partially or wholly positioned in the lower uterine segment. It should not be diagnosed before 32 weeks’ gestation as the lower segment has not formed at this time. For this diagnosis, the presence of placenta covering the area of the internal cervical os should be documented by US. An overly distended urinary bladder may compress the lower uterine segment inwards, simulating the appearance of a lowlying placenta and TVS is the gold standard in the diagnosis of placenta praevia. Areas of retroplacental haemorrhage can also be evaluated sonographically. In these patients, hypoechoic areas that indent the placenta, indicating space-occupying lesions, can be identified.

Figure 53.17 Fetal hydrops. Longitudinal view of a fetus with skin oedema, ascites and a hydrothorax.

Table 53.4 CAUSES OF NONIMMUNE HYDROPS AT LESS THAN 20-WEEKS’ GESTATION Chromosomal cause

65.6%

Infection

7.3%

Cardiovascular cause

2.1%

Fetal akinesia

8.3%

Multiple malformation

2.1%

Other anatomical defect

3.1%

Haematological cause

1.1%

Unknown

10.4%

(From Twining P, McHugo J M, Pilling D W (eds.) 2006 Textbook of Fetal Abnormalities, 2nd edition, Churchill Livingstone.)

Figure 53.18

Facial cleft demonstrated on 3D US.

CHAPTER 53

CERVICAL LENGTH The incidence of preterm birth has not altered significantly in the past few decades and remains the most significant cause of neonatal morbidity and mortality with long-term sequelae for survivors. One cause of premature delivery is cervical incompetence, which is defined as the painless dilatation of the cervix and delivery without uterine contractions during the midtrimester. Cervical length is a useful tool to assess the cervix during pregnancy. Although it can be visualized by the transabdominal and transperineal route,TVS is more accurate (Fig. 53.19). TVS is more accurate in determining cervical shortening and funnelling than digital examination and may also demonstrate prolapsing membranes34,35. The sensitivity and positive predictive value of screening for cervical length improve in the high-risk population. Surveillance should be offered to women with the following risk factors: • previous mid-trimester fetal loss • previous cervical surgery or cervical manipulations • polyhydramnios. Cervical length measurements are also useful in predicting premature delivery in women presenting with signs of prema-

•

IMAGING IN OBSTETRICS AND INFERTILITY

ture labour. Of women presenting with a cervical length less than 16 mm, 49 per cent deliver within 7 d36. Cervical cerclage is currently the treatment of choice for patients with documented cervical incompetence. The literature however is not conclusive for the benefits of cerclage when compared to bed rest alone.

FETAL MAGNETIC RESONANCE IMAGING Fetal MRI is now a useful adjunct to US in the detection and delineation of fetal anomalies.T2-weighted single-shot turbo spin-echo sequences, such as HASTE, have rapid acquisition times, which lends itself to imaging the fetus without the need for transplacental sedation. Each slice is acquired in less than 1 s so the effects of fetal movement are reduced. Fetal MRI will not replace US as a screening tool but it is useful to clarify abnormalities detected on US and in cases such as large maternal body habitus and oligohydramnios, which limit visualization on US (Fig. 53.20). Fetal MRI has proven diagnostic benefits in cranial abnormalities37,38 (Figs 53.21, 53.22) and also in abnormalities of the renal tract39 (Fig. 53.23).

Figure 53.19 Cervix. (A) Normal cervix (between arrows) with hypoechoic endocervical canal. (B) Dilatation of the endocervical canal in a patient with cervical incompetence.

1213

1214

SECTION 6

• WOMEN’S IMAGING

Figure 53.21 Agenesis of the corpus callosum. (A) Axial MRI through the fetal brain shows parallel, separated ventricles. (B) Coronal imaging confirms agenesis of the corpus callosum.

Figure 53.20 Congenital CMV. (A) Sagittal fetal MRI in a case with anhydramnios demonstrates fetal ascites. (B) Coronal view of the fetal brain shows high signal in the brain parenchyma consistent with cerebritis and haemorrhage in the cerebellum.

Figure 53.22 Occipital meningocoele. (A) Axial MRI through the cranio-cervical junction shows a large open neural tube defect. Continued

CHAPTER 53

Figure 53.22 Cont’d agenesis.

•

IMAGING IN OBSTETRICS AND INFERTILITY

Occipital meningocoele. (B) Sagittal image showing the extent of the defect. (C) This was associated with cerebellar vermian

Figure 53.23 Polycystic renal disease. (A) Longitudinal fetal MRI confirms the large polycystic kidney detected on US. (B) Axial MRI also shows that the contralateral kidney is dysplastic with small cortical cysts (arrow).

1215

1216

SECTION 6

• WOMEN’S IMAGING

REFERENCES 1. te Velde E R, Eijkemans R, Habbema H D 2000 Variation in couple fecundity and time to pregnancy, an essential concept in human reproduction. Lancet 355:1928–1929 2. Hull M G, Glazener C M, Kelly N J et al 1985 Population study of causes, treatment, and outcome of infertility. Br Med J 291:1693–1697 3. National Collaborating Centre for Women’s and Children’s Health 2004 Fertility: assessment and treatment for people with fertility problems. Commissioned by the National Institute for Clinical Excellence. Feb 4. Swart P, Mol B W, van der Veen F, van Beurden M, Redekop W K, Bossuyt P M 1995 The accuracy of hysterosalpingography in the diagnosis of tubal pathology: a meta-analysis. Fertil Steril 64:486–491 5. Johnson N, Vandekerckhove P, Watson A, Lilford R, Harada T, Hughes E 2005 Tubal flushing for subfertility. Cochrane Database Syst Rev 18 6. Blankstein J, Shaley J, Saadon T et al 1987 Ovarian hyperstimulation syndrome: prediction by number and size of preovulatory ovarian follicles. Fertil Steril 47:597–602 7. Dellenbach P, Nisand I, Moreau L et al 1985 Transvaginal sonographically controlled follicle puncture for oocyte retrieval. Fertil Steril 44:656–662 8. Gonen Y, Casper R F, Jacobson W, Blankier J 1989 Endometrial thickness and growth during ovarian stimulation: a possible predictor of implantation in in vitro fertilization. Fertil Steril 52:446–450 9. Raga F, Bonilla-Musoles F, Casan E M, Klein O, Bonilla F 1999 Assessment of endometrial volume by three-dimensional ultrasound prior to embryo transfer: clues to endometrial receptivity. Hum Reprod 14:2851–2854 10. Steer C, Tan S, Dillon D, Mason B A, Campbell S 1995 Vaginal color Doppler assessment of uterine artery impedance correlates with immunohistochemical markers of endometrial receptivity required for the implantation of an embryo. Fertil Steril 63:101–108 11. Kupesic S, Bekavac I, Bjelos D, Kurjak A 2001 Assessment of endometrial receptivity by transvaginal color Doppler and three-dimensional power Doppler ultrasonography in patients undergoing in vitro fertilization procedures. J Ultrasound Med 20:125–134 12. Royal College of Radiologists and Royal College of Obstetricians and Gynaecologists 1995 Guidance on Ultrasound Procedures in Early Pregnancy 13. Rajkhowa M, Glass M R, Rutherford A J, Balen A H, Sharma V, Cuckle H S 2000 Trends in the incidence of ectopic pregnancy in England and Wales from 1966 to 1996. BJOG 107:369–374 14. Condous G, Okaro E, Khalid A et al 2005 The accuracy of transvaginal ultrasound for the diagnosis of ectopic pregnancy prior to surgery. Hum Reprod 20:1404–1409 15. DeVoe R W, Pratt J H 1948 Simultaneous intrauterine and extrauterine pregnancy. Am J Obst Gynecol 56:1119 16. Richards S R, Stempel L E, Carlton B D 1982 Heterotopic pregnancy: Reappraisal of incidence. Am J Obst Gynecol 142:928–930 17. Gabrielli S, Romero R, Pilu G et al 1992 Accuracy of transvaginal ultrasound and serum hCG in the diagnosis of ectopic pregnancy. Ultrasound Obstet Gynecol 2:110–115 18. Robinson H P 1973 Sonar measurement of fetal crown-rump length as means of assessing maturity in first trimester of pregnancy. Br Med J 4:28–31 19. Chitty L S, Altman D G, Henderson A, Campbell S 1994 Charts of fetal size: 2. Head measurements. Br J Obstet Gynaecol 101:35–43

20. Chitty L S, Altman D G, Henderson A, Campbell S 1994 Charts of fetal size: 3. Abdominal measurements. Br J Obstet Gynaecol 101:125–131 21. Nicolaides K H, Azar G, Byrne D, Mansur C, Marks K 1992 Fetal nuchal translucency: ultrasound screening for chromosomal defects in first trimester of pregnancy. Br Med J 304:867–869 22. Snijders R J M, Noble P, Sebire N, Souka A, Nicolaides K H 1998 UK multicentre project on assessment of risk of trisomy 21 by maternal age and fetal nuchal-translucency thickness at 10–14 weeks of gestation. Lancet 352:343–346 23. Nicolaides K H, Spencer K, Avgidou K, Faiola S, Falcon O 2005 Multicenter study of first-trimester screening for trisomy 21 in 75 821 pregnancies: results and estimation of the potential impact of individual risk-orientated two-stage first-trimester screening. Ultrasound Obstet Gynecol 25:221–226 24. Ewigman B G, Crane J P, Frigoletto F D, LeFevre M L, Bain R P, McNellis D and the RADIUS Study Group 1993 Effect of prenatal ultrasound screening on perinatal outcome. N Engl J Med 329:821–827 25. Chitty L S, Hunt G H, Moore J, Lobb M O 1991 Effectiveness of routine ultrasonography in detecting fetal structural abnormalities in a low risk population. Br Med J 303:1165–1169 26. Twining P, McHugo J M, Pilling D W 2000 Textbook of Fetal Abnormalities. Churchill Livingstone, Edinburgh 27. Nyberg D A, McGahan J P, Pretorius D H, Pilu G 2006 Diagnostic imaging of fetal anomalies. Lippincott Williams & Wilkins, New York 28. Mitchell S C, Korones S B 1971 Congenital heart disease in 56 109 births. Incidence and natural history. Circulation 43:323–332 29. Pitkin R M 1991 Screening and detection of congenital malformation. Am J Obstet Gynecol 164:1045–1048 30. Carvalho J S, Mavrides E, Shinebourne E A, Campbell S, Thilaganathan G 2002 Improving the effectiveness of routine prenatal screening for major congenital heart defects. Heart 88:487–491 31. Mari G, Deter R L, Carpenter R L et al 2000 Noninvasive diagnosis by Doppler ultrasonography of fetal anemia due to maternal red cell alloimmunization. Collaborative Group for Doppler Assessment of the Blood Velocity in Anemic Fetuses. N Engl J Med 342:9–14 32. Merz E 1998 3-D ultrasound in obstetrics and gynecology. Lippincott Williams & Wilkins, New York 33. Bega G, Lev-Toaff A, Kuhlman K, Kurtz A, Goldberg B, Wapner R 2001 Three-dimensional ultrasonographic imaging in obstetrics. J Ultrasound Med 20:391–408 34. Berghella V, Tolosa J E, Kuhlman K, Weiner S, Bolognese R J, Wapner R J 1997 Cervical ultrasonography compared with manual examination as a predictor of preterm delivery. Am J Obstet Gynecol 177:723–730 35. Iams J D, Johnson F F, Sonek J, Sachs L, Gebauer C, Samuels P 1995 Cervical competence as a continuum: a study of ultrasonographic cervical length and obstetric performance. Am J Obstet Gynecol 175 (4 Pt 1):1097–1103 36. Tsoi E, Fuchs I B, Rane S, Geerts L, Nicolaides K H 2005 Sonographic measurement of cervical length in threatened preterm labor in singleton pregnancies with intact membranes. Ultrasound Obstet Gynecol 25:353–356 37. Garel C 2003 MRI of the Fetal Brain: Normal Development and Cerebral Pathologies. Springer-Verlag, Berlin and Heidelberg GmbH & Co. 38. Griffiths P D, Paley M N, Widjaja E, Taylor C, Whitby E H 2005 In utero magnetic resonance imaging for brain and spinal abnormalities in fetuses. Br Med J 331:562–565 39. Hormann M, Brugger P C, Balassy C, Witzani L, Prayer D 2006 Fetal MRI of the urinary system. Eur J Radiol 57:303–311

CHAPTER

Imaging in Gynaecology

54

Evis Sala, Sandra Allison, Susan M. Ascher and Hedvig Hricak

• • • • • • • •

Imaging techniques Congenital anomalies of the female genital tract Infertility Pelvic pain Uterus and cervix The adnexa Vulva and vagina Contraception

The results of diagnostic imaging tests frequently change treatment strategies and impact our understanding of disease processes. This chapter gives a brief review of common gynaecological entities and presents indications for each imaging technique used while focusing on advances in ultrasound (US), computed tomography (CT), magnetic resonance imaging (MRI) and positron emission tomography (PET). As with any changing technological arena, imaging strategies are not static and require on-going updates and re-evaluation.

IMAGING TECHNIQUES Ultrasound Ultrasound (transabdominal or transvaginal) is accepted as the primary imaging technique for examining the female pelvis. Currently, the main role of ultrasound (US) in gynaecology includes evaluation of a suspected pelvic mass, evaluation of the causes of uterine enlargement, identification of endometrial abnormalities in a patient with postmenopausal bleeding and characterization of ovarian masses. It is also the primary imaging technique of choice in the evaluation of women with acute pelvic pain. In addition, ultrasound has become invaluable in guiding a wide selection of invasive procedures. For example, it is used for transabdominal and transvaginal guidance of fluid or tissue sampling, transvaginal-guided drain placement, guidance for placement of brachytherapy for cervical and endometrial malignancy and intraoperative assessment for completion of evacuation and instrument placement, especially when the anatomy is difficult to assess preoperatively.

A full bladder provides a sonic window and is mandatory for transabdominal US. It is usual to employ 3.5–5.0 MHz transducers. Transvaginal ultrasound, which is optimally performed with an empty bladder, provides greater detail of the anatomy and pathology due to the closer apposition to the pelvic organs as well as the higher frequencies of insonation (5–8 MHz). Colour, power and spectral Doppler provide additional information regarding associated vascularity. Ultrasound has many advantages in routine pelvic imaging: it is relatively inexpensive, provides multiplanar views, is widely available and lacks ionizing radiation or contrast media. Its portability allows use in virtually any setting, including the ultrasound suite, operating room, patient bedside, or radiation therapy suite. However, US also has a number of limitations: it is operator dependent and image quality varies with patient body habitus. Although transvaginal, sonohysterography and endorectal US provide improved spatial resolution, they are not as useful as either CT or MRI in the staging of pelvic malignancies, including the evaluation of regional extent or metastatic spread.

Normal ultrasound anatomy The normal uterus measures between 5 and 9 cm in length and is usually visualized in an anteverted position (in relation to the urinary bladder). The appearance of the endometrium changes in response to the menstrual cycle (Fig. 54.1). In the proliferative phase of the cycle, the endometrium is well defined and may measure up to 8 mm. In midcycle, the endometrium assumes a trilaminar appearance and may measure up to 12–16 mm. During the secretory phase, the layers become hyperechoic due to the increasing complexity of glandular structure and secretions. The Fallopian tubes are usually not identified unless distended with fluid. Following menopause, the uterus may decrease in size and an endometrial thickness of 5 mm serves as a threshold for endometrial biopsy. An endometrial thickness of up to 8 mm is considered acceptable for those on hormonal therapy. Normal ovaries are visualized in the majority of premenopausal patients. They are more readily identified on transvaginal US. The iliac vessels lie immediately posterior to the ovary

1218

SECTION 6

• WOMEN’S IMAGING

myometrium, endometrium, cervix and parametrium) and the major pelvic vessels. With helical and multidetector CT, reformatting of images is also possible, including the creation of 3D volume renderings.

Normal CT anatomy CT examination displays the uterus as a triangular or ovoid soft tissue structure behind the urinary bladder (Fig. 54.2). On unenhanced images, secretions within the endometrial canal demonstrate centrally located decreased attenuation. Following intravenous (IV) contrast medium administration, the myometrium enhances, helping to delineate the endometrium. The vagina, cervix and uterine corpus can be distinguished by morphological and enhancement pattern characteristics. The uterine corpus is typically triangular, whereas the cervix is more rounded. At the level of the fornix, the vagina is seen as a flat rectangle.The broad and round ligaments can be seen coursing laterally and anteriorly, respectively. Occasionally, the uterosacral ligaments are depicted as arc-like structures extending from the cervix to the sacrum. In the premenopausal patient

Figure 54.1 (A) Transvaginal US shows normal endometrium (arrows) in proliferative phase and (B) in follicular phase (arrows). (C) Sagittal transvaginal US shows a normal ovary (O) with follicles. Note the location of the ovary anterior and medial to the internal iliac vessels (I) within the ovarian fossa.

and assist in its identification (Fig. 54.1). A normal ovary typically measures 30 mm in any two dimensions but may measure 50 mm or more in one plane. The volume of the ovary can be estimated from the formula for an ellipsoid (0.5 × length × width × breadth) and is usually less than 10 cm3. Follicular and luteal cysts are seen within the ovary.

Computed tomography CT is the most commonly used primary imaging study for evaluating the extent of gynaecological malignancy and for detecting persistent and recurrent pelvic tumours. CT-guided biopsy can be used to confirm pelvic recurrence. Advantages of CT include oral and rectal contrast opacification of the gastrointestinal tract, intravenous contrast enhancement of blood vessels and viscera, fast data acquisition and high spatial resolution. Disadvantages of CT include the use of ionizing radiation (this is why it is essential to exclude the possibility of pregnancy before performing CT), degradation of image quality by body habitus or metallic hip prosthesis and the risk of morbidity and mortality associated with iodinated contrast agents. Although CT is useful in the later stages of pelvic malignancy, it often has limited utility in characterizing early-stage disease. The advent of helical (spiral) and multidetector CT (MDCT) has made it possible for images to be acquired during the arterial, capillary and venous phases of enhancement following contrast medium administration. The new generation of CT equipment provides good delineation of uterine anatomy (i.e.

Figure 54.2 (A) Normal uterus. Helical CT of the normal uterus and adnexa shows the low attenuation endometrial canal (*) flanked by enhancing myometrium (arrowheads). Enhancing endocervical mucosa (short solid white arrows) surrounds the endocervical canal. The fibrous cervical stroma (open black arrows) enhances less than the uterine corpus myometrium. A physiological cyst is seen in the right ovary. (B) Normal ovary. Helical CT shows bilateral physiological ovarian cysts (*). The ovaries are in their expected location, anterior to the internal iliac vessels and posterior to the external iliac vessels. The zonal anatomy of the uterus is faintly defined.

CHAPTER 54

• IMAGING IN GYNAECOLOGY

the normal ovaries are routinely seen, usually posterolateral to the uterine corpus (Fig. 54.2).Their uniform soft tissue density is punctuated by small cystic regions representing follicles. In the postmenopausal patient the ovaries are small and may not be discernible.

Magnetic resonance imaging The role of MRI in gynaecology has evolved during the last two decades. There is now a substantial body of evidence that MRI is useful in evaluating Müllerian duct anomalies and both benign and malignant conditions of the pelvis. MRI has been shown to be superior to CT in the work-up of uterine and cervical cancer and may be a useful problem-solving tool in the evaluation of ovarian cancer. In addition, there is evidence that MRI may aid the differentiation of radiation fibrosis from recurrent tumour. The accuracy of MRI assessment of lymph node invasion is similar to that of CT; both rely on size criteria to detect lymphadenopathy. In addition, MR-guided biopsies are also gaining wider clinical acceptance. Although MRI is still relatively expensive, it has been shown to minimize costs in some clinical settings by limiting or eliminating the need for further expensive and/or more invasive diagnostic or surgical procedures. Advantages of MRI include superb spatial and tissue contrast resolution, no use of ionizing radiation, multiplanar capability and fast (i.e. breath-hold and breathing-independent) techniques. MRI is the technique of choice for patients with previous reactions to iodinated IV contrast media or impaired renal function. However, MRI is contraindicated in patients with implants such as pacemakers, neural stimulators or cochlear implants, certain vascular clips and metallic objects. Some patients may experience claustrophobia, causing difficulty in completing the examination or requiring sedative pre-medication in subsequent MR examinations.

Normal MRI anatomy Pelvic anatomy is exquisitely demonstrated by MRI (Fig. 54.3). On T1-weighted sequences, the normal pelvic musculature and viscera demonstrate homogeneous low to intermediate signal intensity. However, it is the contrast resolution of T2weighting that is the basis for the superb tissue characterization of MRI. On T2-weighted sequences, the uterus, cervix and vagina exhibit distinct layers of different signal intensity— the so-called zonal architecture.The endometrium yields high signal intensity on T2-weighted images. The peripheral myometrium, in comparison, is intermediate in signal intensity, higher than striated muscle. Interposed between these two layers is a narrow band of decreased signal intensity, the junctional zone (JZ), which corresponds to the innermost myometrium. Its signal properties reflect its lower water content, compared with the remainder of the myometrium, which may be a function of its decrease in extracellular matrix/unit volume. The three zones seen on MR images, however, are not identical to the different zones seen on US. The width of the endometrium (both leaflets) varies with the menstrual cycle and, on the sagittal plane of section, measures up to 3 mm in the proliferative phase and up to 7 mm in the secretory phase. In postmenopausal women not receiving exogenous

Figure 54.3 Zonal anatomy of the uterus. Sagittal T2-weighted MRI. The central, high-signal intensity stripe represents the endometrium (small arrows); the band of low signal intensity subjacent to the endometrial stripe represents the inner myometrium or junctional zone (arrows). The outer layer of the myometrium is of intermediate signal intensity (open arrow). bl = bladder.

hormones, uterine zonal anatomy is often indistinct and the endometrium measures less than 3 mm. The cervix also demonstrates zonal architecture on T2weighted images, with the normal cervix demonstrating a central area of high signal intensity (endocervical glands and mucus) surrounded by low signal intensity stroma (elastic fibrous tissue). Around the periphery of the cervix, smooth muscle predominates, resulting in a rim of intermediate signal intensity similar to that of myometrium. Occasionally, intermediate signal intensity cervical mucosal folds (plicae palmatae) can be seen interposed between the low signal intensity cervical stroma and the high signal intensity endocervical canal. T2-weighted images of the vagina reveal two zones: the bright vaginal mucosa and the intermediate intensity vaginal wall. The ligamentous structures are identified with low signal intensity by their anatomical location. Following IV administration of paramagnetic gadolinium chelates, the zonal anatomy of the uterus is demonstrated on T1-weighted images. The endometrium and outer myometrium enhance to a greater extent than the JZ. Similarly, the inner cervical mucosa and outer smooth muscle enhance more than the fibrocervical stroma. The parametrial tissues, vaginal walls and submucosa also enhance after IV contrast medium administration. The normal MRI appearance of the ovaries varies depending on the pulse sequence used. On T1-weighted images, the ovaries display homogeneous low to intermediate signal intensity, whereas on T2-weighted images the follicles become brighter than the surrounding stroma. The normal Fallopian tubes are not routinely seen because of their small size and tortuous course.

Positron emission tomography PET (and PET-CT) is maturing as an imaging technique. PET takes advantage of the biochemical changes associated

1219

1220

SECTION 6

• WOMEN’S IMAGING

with malignancy that often precede and are more specific than the structural changes visualized by conventional means. Specifically, with the most commonly used glucose analogue, 2-[18F]-fluoro-2-deoxy-D-glucose (FDG), PET exploits the accelerated rate of glycolysis common to neoplastic cells to image tumours. Because PET requires specialized instrumentation and a local source of positron emitters, it is not widely available and its routine use is often limited to tertiary care settings. Most work to date with PET and gynaecological malignancy has focused on cervical and ovarian cancer.

Hysterosalpingography

Figure 54.4 Normal hysterosalpingogram. Cervix and uterine body are delineated by contrast media. Both Fallopian tubes are shown (arrows), with early peritoneal spill.

Hysterosalpingography (HSG) is an imaging test whereby radio-opaque contrast medium is instilled into the uterus and Fallopian tubes. It is used mainly to evaluate infertility (e.g. tubal obstruction). In the past, HSG was used for the evaluation of congenital abnormalities of the uterus but has been replaced by US and MRI. The optimal time to perform HSG is towards the end of the first week after the menstrual period, when the isthmus is most distensible and the Fallopian tubes most readily filled with contrast medium and the chance of pregnancy is small. Nonionic contrast media have little advantage compared to ionic agents but may cause less peritoneal irritation. HSG is contraindicated in pregnancy and active pelvic infection and in the pre- and postmenopausal phases. Possible complications are pain, pelvic infection, haemorrhage and vasovagal attacks.

Normal hysterosalpingography anatomy The cervical canal is usually 30–40 mm long and tends to shorten after childbirth. It measures about one third of the entire length of the uterus and is often spindle shaped. Glandular filling often occurs in the normal cervix. The isthmus is seen as a distinct segment, narrower than the uterine body and cervical canal, in only half of all normal hysterograms.The internal os may appear as a short constriction of the lumen. The cavity of the uterine body is triangular in shape (Fig. 54.4), its walls normally regular and straight or concave. Both the average length and the intercornual diameter measure approximately 35 mm. The cornual sphincters are pear or spindle shaped and are often separated from the uterine body by a short, dark line due to a mucosal fold. The apex of each cornu opens continually to the tubal lumen. The Fallopian tubes are approximately 5 or 6 cm long, with a variable degree of tortuosity. The isthmic portion is of uniform diameter and opens laterally into the wide ampulla.

Sonohysterography This technique involves placement of a 5 F catheter through the cervix and distension of the uterine cavity with sterile saline under direct ultrasound visualization (Fig. 54.5). The procedure is well tolerated with few contraindications and virtually no complications. It can be performed rapidly on an outpatient basis without the use of ionizing radiation or contrast agent. Several studies have shown that the accuracy of US HSG exceeds that of endovaginal US alone. US HSG can make a more precise diagnosis in cases where endovaginal US only shows abnormal thickening of the endometrium

Figure 54.5 Sonohysterography. Sagittal transvaginal US (A) demonstrates the inflated balloon of the sonohysterographic catheter (*) within the endometrial canal. Following the instillation of 40 cc of sterile saline (B), fluid distends the endometrial canal.

and can differentiate intracavitary, endometrial and subendometrial pathology. It can therefore select and effectively triage patients to appropriate methods of endometrial sampling and direct hysteroscopic removal of pathology when appro-

CHAPTER 54

• IMAGING IN GYNAECOLOGY

priate. Because the single layer endometrium is visualized, focal pathology can be differentiated from diffuse endometrial conditions with increased accuracy. In the evaluation of tubal patency, the identification of free fluid in the pouch of Douglas indicates that at least one tube is patent.

CONGENITAL ANOMALIES OF THE FEMALE GENITAL TRACT General considerations Müllerian duct anomalies result from nondevelopment or varying degrees of nonfusion or nonresorption of the Müllerian ducts. These congenital anomalies occur in 1–15% of women. Müllerian duct anomalies are associated with menstrual disorders, infertility and obstetric complications. Renal anomalies (especially agenesis or ectopia) may be present in up to 50% of these patients. Embryologically, the uterus, the upper two thirds of the vagina and the Fallopian tubes are derived from paired Müllerian ducts. At approximately 10 weeks following conception, the ducts migrate caudally and undergo fusion and subsequent canalization.This process may be interrupted at any time. Failure of the ducts to develop leads to various types of uterine, cervical and vaginal agenesis, while absent or incomplete fusion results in didelphys or bicornuate uteri, respectively. In those instances where fusion does occur, but there is absent or incomplete resorption of the septum between the Müllerian duct components, a septate uterus will result. Evaluation of Müllerian duct anomalies with physical examination and more traditional imaging studies (HSG and US) is often inconclusive. MRI is the most accurate imaging technique for evaluating patients with congenital anomalies, allowing both precise classification and demonstration of associated complications.

Müllerian duct anomalies The clinical classification of Müllerian duct anomalies follows the guidelines proposed by Buttram and Gibbons/American Fertility Society. The role of imaging is to provide a detailed map of the pelvic anatomy, including the presence and extent of any anomalies. Surgical management in female patients with anatomical anomalies is aimed at preventing endometriosis and preserving fertility.

Class I: Müllerian agenesis or hypoplasia Uterine agenesis or hypoplasia results from nondevelopment or rudimentary development of the Müllerian ducts. A subtype of uterine agenesis is the Mayer–Rokitansky– Kuster–Hauser (MRKH) syndrome. In this syndrome, the presence of vaginal agenesis or hypoplasia with intact ovaries and Fallopian tubes is accompanied by variable anomalies of the uterus, urinary tract and skeletal system. Uterine remnants may be present. Patients with this malformation benefit from US as well as MRI. On MR images the absence or anomalies of the uterus and upper vagina, with varying degrees of development of the lower vagina, is reliably detected on a combination of sagittal and axial images (Fig. 54.6). Normal

Figure 54.6 Absence of the uterus. Sagittal T2-weighted MRI shows no uterine tissue.

ovaries are usually present. Uterine hypoplasia is diagnosed when the uterus is small, the endometrium is atrophic and on T2-weighted MR images the myometrium is of lower than normal signal intensity.

Class II: Unicornuate uterus The unicornuate uterus results from nondevelopment or rudimentary development of one Müllerian duct. The other Müllerian duct is fully developed and demonstrates a ‘banana-like’ configuration. The role of imaging is to demonstrate: 1 the presence or absence of a rudimentary horn 2 whether or not the rudimentary horn contains endometrium 3 whether or not the rudimentary horn communicates with the main uterine cavity. Patients with a rudimentary horn communicating with the uterine cavity might benefit from surgical removal of this rudimentary horn. On T2-weighted MR images, unicornuate uterus demonstrates a curved, elongated uterus with tapering of the fundal segment off midline (the ‘banana-like’ configuration). Normal uterine zonal anatomy is maintained. The rudimentary horn, when present, usually demonstrates lower signal intensity.

Class III: Uterus didelphys Uterus didelphys results from nonfusion of the two Müllerian ducts. Two separate normal-sized uterine horns and cervices are demonstrated on T2-weighted MR images. A longitudinal vaginal septum is present in 75% of cases, occasionally complicated by transverse septa causing obstruction. The two uterine horns are usually widely separated, with preservation of the endometrial and myometrial widths. An oblique plane of imaging parallel to the long axis of the uterus is useful to delineate the uterus, whereas the axial oblique plane is useful to delineate the vaginal septum. Haemorrhage in the obstructed segment is best seen on T1-weighted images.

1221

1222

SECTION 6

• WOMEN’S IMAGING

Class IV: Bicornuate uterus Partial fusion of the Müllerian ducts results in the bicornuate uterus (Figs 54.7, 54.8). There is incomplete fusion of the cephalad extent of the uterovaginal horns with resorption of the uterovaginal septum. MRI shows uterine horns separated by an intervening cleft longer than 1 cm, in the external fundal myometrium. Normal zonal anatomy is seen in each horn and there is a dividing septum composed of myometrium.

Class V: Septate uterus A septate uterus is seen when there is incomplete resorption of the final fibrous septum between the two uterine horns.

The septum may be partial, or it may be complete and extend to the external cervical os. The differentiation between the septate and the bicornuate uterus is clinically important. The septate uterus is associated with a higher rate of reproductive complications, with only 3% of pregnancies delivered at term. Lacking the necessary blood supply, the collagenous septum in the septate uterus cannot support a pregnancy as well as the myometrial septum in the bicornuate uterus.The abortion rate in patients with a septate uterus is twice that of patients with a bicornuate uterus. When clinically warranted, the dividing septum can be removed hysteroscopically. T2-weighted MR images taken parallel to the long axis of the uterus demonstrate a convex, flat, or concave (< 1 cm) external uterine contour and the presence of the fibrous septa1.

Vaginal anomalies A number of anomalies have been found in vaginas that have not developed normally.Vaginal anomalies may be categorized as follows.

Congenital absence of the Müllerian ducts (vaginal aplasia, MRKH syndrome) The pathophysiology of vaginal absence may be either a result of failure of the vaginal plate to form, or failure of cavitation. Absence of the uterus and Fallopian tubes indicates total failure of Müllerian duct development and is known as a Mayer– Rokitansky–Kuster–Hauser (MRKH) syndrome. The presence of vaginal agenesis or hypoplasia with intact ovaries and Fallopian tubes is accompanied by variable anomalies of the uterus, urinary tract and skeletal system (Fig. 54.9). (See above section on Müllerian duct anomalies for further discussion of MRKH syndrome.)

Disorder of vertical fusion Figure 54.7 Uterus bicornuate. Coronal T2-weighted MRI demonstrating two endometrial canals (*).

The transverse vaginal septum prevents loss of menstrual blood and results in haematocolpos. Most patients present as teenagers with cyclical abdominal pain and a haematocolpos might be palpable within the pelvis (Fig. 54.10). In these patients a careful pelvic examination and US are helpful for diagnosis. On MRI, T2-weighted images show dilatation of the vagina with intraluminal fluid of intermediate or high signal intensity and the occasional presence of fluid/debris levels. The lower third of the vagina is replaced by low signal intensity fibrous tissue with loss of normal zonal anatomy. T1-weighted images with fat suppression confirm blood products in haematometrocolpos and associated endometriosis if present.

Disorder of lateral fusion These patients often present with the incidental finding of a vaginal septum that is usually asymptomatic. It may first be diagnosed during pregnancy and excision will be necessary to ensure a vaginal delivery. This malformation may be missed if careful examination is not performed.

INFERTILITY Figure 54.8 Bicornuate uterus with pregnancy in one horn. Transverse transvaginal US shows two endometrial cavities (arrows) in a patient with a bicornuate uterus. A living fetus (*) was noted in the left horn.

Imaging plays an important role in the evaluation of infertility. In this section we describe the use of imaging in patients

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Figure 54.9 Uterine hypoplasia, horseshoe kidney. The sagittal T2weighted MRI shows a hypoplastic uterus (A), without zonal architecture. On the axial image a horseshoe kidney is shown (B).

Figure 54.10 Vaginal septum. Sagittal T2-weighed MRI shows the presence of hematometra(*) caused by a transverse vaginal septum.

undergoing modern infertility treatments. The success of in vitro fertilization (IVF) has been due in part to the correct timing of ovulation and subsequent oocyte recovery that US can provide. Using transabdominal US, ovarian follicles of 3–5 mm diameter can be visualized.They appear as echo-free structures amidst the more echogenic ovarian tissue. Their rate of growth is linear and the mean diameter before ovulation is 20.2 mm (range 8–42 mm). Structures within the follicle such as the cumulus oophorus can also be visualized. Following ovulation, internal echoes appear because of bleeding. Free fluid may also be observed in the pouch of Douglas. Transvaginal US has largely replaced the transab-

dominal approach in infertility practice. Follicular aspiration is possible with a needle-guide attachment to the vaginal probe, as is fine needle aspiration of fluid in the pouch of Douglas. A more precise measurement of follicular size is possible and the corpus luteum is easily recognized. US is also useful for accurate timing of artificial insemination, while a postcoital test can help differentiate between inadequate sperm penetration and poor mucus production in the presence of immature follicles. Ultrasound is also used to monitor patients on clomiphene therapy and there is good correlation between follicular diameters and plasma oestradiol concentrations. The hyperstimulation syndrome is uncommon if gonadotrophin therapy is monitored by US in conjunction with measurements of plasma oestradiol levels. The waveforms of blood flow in vessels supplying the ovaries of women undergoing IVF have also been studied using transvaginal pulsed Doppler US2. Certain blood flow patterns may help predict implantation failure. MRI is valuable in the investigation of infertility where uterine or adnexal pathology is suspected. It provides a particularly high diagnostic yield in patients with dysmenorrhoea and menorrhagia3, confidently diagnosing leiomyomas, adenomyosis and/or endometriosis. It should be part of the investigation of patients with persistent unexplained infertility awaiting costly procedures such as gamete intra-Fallopian transfer (GIFT) and IVF.

PELVIC PAIN Chronic pelvic pain (CPP) has been described as noncyclic pelvic pain of greater than 6 months’ duration that is not relieved by strong analgesics. Common causes of CPP are endometriosis, adenomyosis, leiomyomas, pelvic varices and pelvic inflammatory disease (PID). A detailed discussion of each entity is included under the anatomical area affected (e.g. uterus, adnexa). Radiological evaluation of women with CPP often includes US and MRI. US (transvaginal and transabdominal) is considered the primary imaging technique in the evaluation of CPP, while MRI is reserved as a problem-solving tool4.

1223

1224

SECTION 6

• WOMEN’S IMAGING

UTERUS AND CERVIX Benign uterine conditions Leiomyoma Leiomyomas are the most common uterine tumours. These benign tumours are found in up to 40% of women in their reproductive years. They are usually multiple and may be subserosal, intramural, or submucosal in location. Symptoms may be caused by the location of the leiomyoma (e.g. menorrhagia with submucosal leiomyomas) and/or their mass effect (e.g. urinary frequency caused by a large subserosal leiomyoma). Hysterectomy has been the traditional primary treatment for debilitating leiomyomas. While hysterectomy is curative, alternative uterine-sparing procedures may be appropriate for some patients. Specifically, myomectomy has been successfully performed for many years and, more recently, transcatheter uterine arterial embolization (UAE) is being used as an alternative, less invasive therapy for symptomatic leiomyomas. Another recent, minimally invasive treatment is MR-guided high intensity focused ultrasound ablation. Diagnostic ultrasound is often the initial radiological evaluation in these patients, while MRI is usually reserved for patients with inconclusive US results or patients undergoing myomectomy, uterine embolization, or MR-guided focused ultrasound ablation. In all of these situations, MR is used to assist in appropriate selection of patients for UAE5.

Ultrasound In most cases, US can accurately detect leiomyomas and distinguish them from extrauterine disease. Twenty per cent of small leiomyomas may be occult by ultrasound. The myomatous uterus is typically enlarged and its outline may be irregular or lobular. The most common appearance of a leiomyoma is that of a well-marginated, hypoechoic, rounded and/or oval mass within the uterine body. Depending on the proportion of smooth muscle, fibrosis and degeneration present within the myoma, the appearance may range from hypoechoic to echogenic, homogenous to heterogeneous, with or without acoustic shadowing (Fig. 54.11). Calcification when present appears as shadowing echogenic foci.

Distortion of the endometrial complex is helpful in identifying a submucosal component and associated menorrhagia may cause a prominent endometrial echo. Submucosal leiomyomas may mimic endometrial lesions and US HSG may aid in making the final diagnosis, particularly in the case of a pedunculated or intracavitary leiomyoma. Special care should be taken when evaluating a pregnancy within a leiomyomatous uterus, as multiple leiomyomas increase the frequency of malposition, retained placenta and premature uterine contractions. In addition, because of hormonal influence, leiomyomas can grow during pregnancy, further compromising the fetus.

Magnetic resonance imaging is indicated when the US examination is indeterminate or limited. It allows precise determination of the size, location and number of leiomyomas. It is also very useful in differentiating a pedunculated subserosal leiomyoma from an adnexal mass5. MRI may help select patients for invasive treatment (myomectomy versus UAE versus hysterectomy). Additionally, MRI may be useful to monitor the success of UAE and assess its durability. The effects of hormonal therapy on leiomyomas can also be monitored. MRI is the most accurate noninvasive diagnostic imaging investigation available so far for differentiation of a leiomyoma from adenomyosis6. This distinction impacts clinical management, as an accepted surgical treatment of a leiomyoma is myomectomy, whereas standard treatment of debilitating adenomyosis is hysterectomy. On T1-weighted images, leiomyomas most commonly present as well-circumscribed, rounded lesions with intermediate signal intensity, often indistinguishable from adjacent myometrium. Optimum contrast is achieved on T2-weighted images, where the tumour is of lower signal intensity relative to the myometrium or endometrium (Fig. 54.12).The presence of calcification usually results in areas of signal void on both T1- and T2-weighted images; however, on MRI, signal void can also be produced by fast flowing blood and therefore is not as specific for calcification as it is on CT. A variety of degenerative processes can alter the characteristic appearance of a leiomyoma, making differential diagnosis more difficult. Following administration of IV contrast medium, most leiomyomas enhance less than adjacent myometrium, whereas degenerated areas may not enhance. Fat saturation T1-weighted images may be helpful in cases of haemorrhagic degeneration5,6. Figure 54.12 Leiomyoma, MRI. T2-weighted sagittal MRI. A subserosal leiomyoma (arrows) distends the posterior aspect of the uterus, displacing the endometrium.

Figure 54.11 Leiomyoma, transvaginal US. Transvaginal US demonstrating hypo- to isoechoic well-defined intramural heterogeneous masses (T), the typical ultrasound appearance of leiomyomas.

CHAPTER 54

Computed tomography Leiomyomas are usually an incidental finding during CT performed for other reasons. CT is not recommended for the evaluation of leiomyomas. Leiomyomas usually have a soft tissue density similar to that of normal myometrium, although necrosis or degeneration may result in low attenuation. On CT, while a contour deformity is the most common sign of a uterine leiomyoma, calcification is the most specific finding for a leiomyoma.

Hysterosalpingography Submucosal

leiomyomas are particularly likely to cause distortion of the uterine cavity, whereas subserosal and small intramural leiomyomas are often associated with normal hysterographic findings. A single leiomyoma will cause a smooth and rounded filling defect on the uterine contour (Fig. 54.13). Multiple submucosal leiomyomas are associated with separate filling defects and sometimes gross distortion of the uterine cavity. HSG is no longer recommended for the evaluation of submucosal leiomyomas.

Adenomyosis Adenomyosis is the presence of endometrial tissue within the myometrium and secondary smooth muscle hypertrophy/ hyperplasia. It can be diffuse or focal. The most frequent symptoms are dysmenorrhoea and dysfunctional uterine bleeding. It is found in 15–27% of hysterectomy specimens, with an increased incidence in multiparous women. Transvaginal US is the initial imaging investigation, with MRI being reserved for indeterminate cases or those undergoing uterus-sparing surgery6,7. Pitfalls in the diagnosis of uterine adenomyosis include leiomyoma, endometrial carcinoma and myometrial contractions8.

Ultrasound Transvaginal US has an accuracy of 68–86% in the diagnosis of diffuse adenomyosis. Typically, the uterus assumes an enlarged but globular configuration, often with antero-posterior asymmetry. Myometrial heterogeneity is related to the presence of endometrial implants and intervening smooth muscle hypertrophy.The implants present as diffuse echogenic nodules, subendometrial echogenic linear striations and nodules and 2–6 mm subendometrial cysts (present in 50% of cases) which represent haemorrhage within the implants. Other ultrasound features of adenomyosis include endometrial pseudowidening, poor definition of the endomyometrial junction and multiple fine areas of attenuation throughout the lesion—the ‘rain shower’ appearance. Colour Doppler examination demonstrates a speckled pattern of increased vascularity

Figure 54.13 Leiomyoma, hysterosalpingogram. A submucosal leiomyoma presents as filling defect (arrow).

• IMAGING IN GYNAECOLOGY

within the heterogeneous area8. While the diagnosis of diffuse adenomyosis can be suggested on transvaginal US, the findings of focal adenomyosis are hard to distinguish from leiomyoma. Because the findings may be subtle, real time or video evaluation (as opposed to static images) of women suspected of adenomyosis may be the key to making the diagnosis.

Magnetic resonance imaging On T2-weighted MRI, adenomyosis appears as areas of low myometrial signal intensity, which presents as focal or diffuse thickening of the JZ (Fig 54.14). When diffuse, a widened low intensity JZ > 12 mm diagnoses disease with high accuracy, while a JZ < 8 mm excludes disease with high accuracy. For indeterminate cases (JZ 8–12 mm), ancillary criteria are used. These include the presence of high signal intensity linear striations (finger-like projections) extending out from endometrium into myometrium on T2-weighted images and high signal intensity foci on T1-weighted images. These foci are believed to represent endometrial rests and/or small punctate haemorrhages6,7.

Endometrial polyps Benign endometrial polyps are common in the endometrial cavity at all ages, with their greatest prevalence after age 50. They consist of stromal cores with mucosal surfaces projecting above the level of the adjacent endometrium. Endometrial polyps must be differentiated from submucosal leiomyomas and malignant neoplasms. Pelvic US often does not reveal endometrial polyps. While MRI is diagnostic, it is rarely employed because of high cost. US HSG is highly accurate in detecting endometrial polyps, even in the setting of endometrial hyperplasia, but differentiating a broad-based polyp from a submucosal fibroid may be problematic. Polyps are typically isoechoic to and continuous with the endometrium and preserve the endomyometrial interface.They are usually homogeneous but may be centrally cystic. Colour Doppler may reveal a characteristic feeding vessel. Outpatient hysteroscopy is the technique of choice for their diagnosis.

Endometrial hyperplasia Endometrial hyperplasia may be subdivided into cystic hyperplasia (simple), adenomatous hyperplasia (complex) and atypical hyperplasia. The main presentations are abnormal uterine bleeding, infertility and postmenopausal bleeding. The important practical considerations in the natural history of endometrial hyperplasia are coexisting endometrial carcinomas, coexisting ovarian carcinomas and the risk of progression to endometrial carcinoma. The main objective of investigating a woman found to have endometrial hyperplasia is to exclude invasive endometrial cancer or ovarian cancer. If endometrial hyperplasia has been diagnosed using an outpatient biopsy instrument, a formal examination under anaesthesia, hysteroscopy and curettage are required to palpate the adnexa and explore the endometrial and endocervical cavities. US evaluation might reveal endometrial thickening, which may be due to endometrial hyperplasia, polyps, or carcinoma. In postmenopausal women, the threshold value for serious endometrial abnormalities on transvaginal US is 5 mm. Although there is no reliable threshold value for premenopausal

1225

1226

SECTION 6

• WOMEN’S IMAGING

Figure 54.14 MRI, adenomyosis. Sagittal (A) and axial (B) T2-weighted MR images through the pelvis demonstrate focal junctional zone widening and multiple punctate high signal intensity foci with the areas of thickening (*, A, B) characteristic of focal adenomyosis. Sagittal (C) and axial (D) T2-weighted MR images in a different patient demonstrate widening of the entire junctional zone (*, C, D) which contains multiple foci of high signal intensity that represent endometrial rests. Appearances are typical of diffuse adenomyosis.

women, limited data in the literature suggest that endometrial thickness greater than 8 mm during the proliferative phase, or greater than 16 mm during the secretory phase, is abnormal. On US HSG endometrial hyperplasia presents as focal or, more commonly, diffuse endometrial thickening without a localized mass or abnormality. Because secretory phase endometrium may mimic hyperplasia, the timing of sonohysterog-

raphy is essential to making or excluding the diagnosis. The focal form of hyperplasia may be difficult to distinguish from a broad-based polyp. Endometrial hyperplasia usually presents as diffuse thickening of the endometrial stripe on T2-weighted images.The signal intensity of the endometrial stripe is isointense or slightly hypointense relative to normal endometrium. However, these

CHAPTER 54

MRI characteristics are nonspecific and are also seen with endometrial carcinoma.

Uterine infections Uterine infections most often occur in the puerperium. Endometritis might also be caused by septic abortion or postoperatively. Pyometra may occur in patients with cervical stenosis due to carcinoma of the cervix or following radiotherapy, or as a complication of endometritis (Fig. 54.15). The distended uterus may be identified on US or cross-sectional studies. The uniform thick-walled appearance of the distended uterus and failure to identify the uterus as a separate structure should differentiate the appearances from a pelvic abscess.

Cervical incompetence Cervical incompetence is responsible for approximately 15% of second- and third-trimester abortions. Primary incompetence may be congenital, associated with diethylstilbestrol exposure, or caused by reduced collagen within the cervix. Secondary incompetence usually results from multiple gestations, gynaecologic/obstetric trauma, or increased prostaglandin production. US is currently the investigation of choice for diagnosing cervical incompetence during pregnancy. A number of US parameters have been described to indicate cervical incompetence. Shortening of the cervix in the absence of any other changes may not alter the prognosis. The cervical length in normal pregnancy should measure more than 3 cm. The width of the cervical canal is by far the most reliable parameter to predict cervical incompetence and should measure less than 2 cm in the second trimester. If bulging of membranes into the cervical canal is seen, the prognosis is unfavourable9. US criteria for diagnosing cervical incompetence have not been established in the nongravid patient.

• IMAGING IN GYNAECOLOGY

MRI offers the potential to diagnose cervical incompetence in the nonpregnant as well as the pregnant patient. Four MRI findings have been described as suggestive of cervical incompetence10: • shortening of the endocervical canal (< 3 cm) • widening of the internal cervical os (> 4 mm) • asymmetric widening of the endocervical canal • thinning or absence of the low signal intensity cervical stroma. When one or more of these findings are present, cervical incompetence should be suspected.

Malignant uterine conditions Endometrial carcinoma Endometrial carcinoma is the fourth most common female cancer and the most common malignancy of the female reproductive tract. Due to earlier diagnosis and treatment advances, its overall mortality has decreased by 28% over the last two decades. Presenting as postmenopausal bleeding, the disease occurs most frequently in white women, with a peak incidence between the ages of 55 and 65 years. Risk factors include unopposed oestrogen intake, nulliparity, obesity, diabetes and polycystic ovarian syndrome. The prognosis of endometrial carcinoma depends on a number of factors, including stage, depth of myometrial invasion, nodal status and grade. Preoperative evaluation of prognostic factors helps in subspecialist treatment planning.

Detection, diagnosis and staging Endometrial carcinomas are typically diagnosed at endometrial biopsy or dilatation and curettage, with imaging being reserved to evaluate extent of disease. Imaging criteria for staging of endometrial cancer are based on the TNM/FIGO classification (Table 54.1).

Ultrasound Transvaginal US is superior to transabdominal US

Figure 54.15 Pyometra. Helical CT shows a distended fluid-filled endometrial canal (*) in a postpartum patient with endometritis and pyometra. Without the presence of gas, it may be impossible to differentiate a pyometra from a hydrometra.

for imaging endometrial abnormalities11. The most common appearance of endometrial cancer is nonspecific thickening of the endometrium. While this thickening is indistinguishable from that found with hyperplasia or polyp, the diagnosis of endometrial cancer should be considered when the endometrial/myometrial junction is disrupted or the endometrial surface is irregular (Fig. 54.16). In postmenopausal women, the threshold value for serious endometrial abnormalities (i.e. endometrial carcinoma, atypical hyperplasia, or complex hyperplasia) is 5 mm. In sonohysterosalpingography, endometrial cancer may appear as an intracavitary polyp or as asymmetric thickening of the endometrial lining. Doppler and colour US have been advocated to improve endometrial carcinoma detection. The use of US is limited to the evaluation of stage I disease with emphasis on the evaluation of the depth of myometrial invasion. For evaluation of myometrial invasion, the presence and continuity of the hypoechoic halo that surrounds the outer layer of the endometrium is assessed (i.e. intact, focally disrupted, or totally disrupted). The extent of myometrial invasion is then estimated by measuring the distance from the central lumen of the uterus to the distal junction between tumour and normal myometrium.

1227

1228

SECTION 6

• WOMEN’S IMAGING

Table 54.1 TNM AND FIGO CLASSIFICATION FOR ENDOMETRIAL CARCINOMA TNM

FIGO

Uterus

Tis

0

Carcinoma in situ

T1

I

Limited to the uterus

T1a

Ia

Tumour limited to endometrium

T1b

Ib

Invasion less than or equal to half of the myometrium

T1c

Ic

Invasion greater than half of the myometrium

T2

II

Invasion of cervix, but not beyond the uterus

T2a

IIA

Endocervical glandular involvement

T2b

IIB

Cervical stromal invasion

T3 and/or N1

III

Local and/or regional spread

T3a

IIIA

Tumour involves the serosa and/or adnexa and/or positive peritoneal cytology

T3b

IIIB

Vaginal involvement

T3c

IIIC

Metastatic to pelvic and/or para-aortic nodes

T4

IV

Tumour extends outside pelvis or invades bladder or rectal mucosa

T4a

IVA

Tumour invades bladder and/or bowel mucosa

M1

IVB

Distant metastasis

Computed tomography High quality pelvic CT in patients

Magnetic resonance imaging Dynamic contrast-enhanced

with endometrial carcinoma requires good bowel opacification. Prolonged oral and IV contrast medium are mandatory. The use of rectal contrast is optimal. The use of IV contrast serves to enhance normal myometrium and to delineate endometrial and myometrial tumours. On contrast-enhanced CT, endometrial tumour is seen as a hypodense mass relative to normal myometrium. CT, especially MDCT, has been used for the staging of endometrial cancer. Its greatest clinical impact is in confirming parametrial and sidewall extension in stage III tumours and in detecting pelvic lymphadenopathy. Limitations of CT include a tendency to understage endometrial carcinoma because of failure to detect bowel or bladder invasion and cervical extension of the tumour.

MR (CEMR) imaging offers a ‘one-stop’ examination with the highest efficacy for pre-treatment evaluation in patients with endometrial cancer12.The MRI protocol includes T1-weighted axial images with a large field of view (FOV) to evaluate the entire pelvis for lymphadenopathy, T2-weighted images (small FOV) in the axial and sagittal planes and dynamic contrastenhanced T1-weighted images (small FOV) in the sagittal and axial planes to evaluate local disease. On unenhanced T1weighted images, endometrial carcinoma is isointense with the normal endometrium. Although endometrial cancer may demonstrate high signal intensity on T2-weighted sequences, it is more typically heterogeneous and may even be of low signal intensity. Routine use of dynamic IV contrast enhancement is necessary for state-of-the-art MR evaluation of endometrial carcinoma12. Following IV contrast medium administration, there is early avid enhancement of normal myometrium. Endometrial cancer enhances slower than the adjacent myometrium allowing identification of small tumours, even those contained by the endometrium. In the later phases of enhancement, tumour appears hypointense relative to the myometrium (Fig. 54.17). MRI is significantly superior to US and CT in the evaluation of both tumour extension into the cervix and myometrial invasion. The overall staging accuracy of MRI has been reported to be between 85 and 93%. Stage I endometrial cancers include tumours confined to the corpus. Stage IA tumours (limited to endometrium) appear as a normal or widened (focal or diffuse) endometrium. An intact JZ and a band of early subendometrial enhancement (SEE) exclude deep myometrial invasion. Regardless of sequence, the tumour–myometrium interface is smooth and sharp. In stage IB disease, tumour extends less than 50% into the myometrium with associated disruption or irregularity of the JZ and SEE. If these landmarks are not present, stage IB tumour is suggested by an irregular tumour–myometrium interface. The presence of low

Figure 54.16 Endometrial carcinoma, transvaginal US. The endometrium is thickened and irregular in this postmenopausal patient. Near the fundus, the endometrial–myometrial junction is indistinct, indicating myometrial invasion (arrow).

CHAPTER 54

• IMAGING IN GYNAECOLOGY

medical contraindication to surgical staging. CT is very useful in screening for lymphatic or peritoneal metastases in patients with a poorly differentiated carcinoma or sarcoma and for confirmation of stage III or stage IV disease. PET imaging is promising in the post-treatment surveillance of endometrial cancer patients13.

Figure 54.17 Endometrial carcinoma, MRI. Sagittal gadoliniumenhanced T1-weighted fat-suppressed MR image shows an endometrial cancer (T) with deep myometrial invasion. Note the thin rim of normal myometrium (black arrows). The disease extends to the upper third of the vagina (white arrow).

signal intensity tumour (later phases of enhancement) within the outer myometrium or beyond indicates deep myometrial invasion—stage IC disease. Stage II includes tumour extension beyond the uterine corpus into the cervix. In stage IIA, invasion of the endocervix appears as widening of the internal os and endocervical canal with preservation of the normal low signal intensity fibrocervical stroma. Disruption of the fibrocervical stroma by high signal intensity tumour on T2-weighted images, together with disruption of normal enhancement of the cervical mucosa by low signal intensity tumour on early dynamic CEMR, indicate cervical invasion—stage IIB disease. In stage III disease, tumour extends outside the uterus but not the true pelvis. Parametrial involvement—stage IIIA— appears as disruption of the serosa with direct extension into the surrounding parametrial fat. In stage IIIB disease, tumour extends into the upper vagina and there is segmental loss of the low signal intensity vaginal wall. In stage IIIC disease, lymphadenopathy is present. Stage IV disease is tumour that extends beyond the true pelvis or invades the bladder or rectum.

Recommended imaging approach US, especially transvaginal US, is often considered to be the primary imaging approach. The role of MRI in the work-up of a patient with endometrial malignancy includes the assessment of the depth of myometrial invasion and the accurate determination of stage in patients with an equivocal pelvic examination or a

Impact of imaging on treatment Endometrial cancer primarily presents at stage I (80% of cases), for which the standard treatment is total abdominal hysterectomy and bilateral salpingo-oophorectomy. The major diagnostic factors necessary for the preoperative evaluation of endometrial cancer are as follows: 1 Differentiation between stages IA and IB; this is becoming critical with increased use of medical (hormonal) treatment for stage IA disease. 2 Determination of the risk of lymph node metastasis in order to have skilled surgical consultation available (i.e. retroperitoneal lymph node sampling may be indicated). Differentiation of stage IB from stage IC has prognostic as well as morbidity implications, as stage IB patients would undergo lymph node sampling whereas stage IC patients would undergo radical lymph node surgical resection. 3 Diagnosing gross cervical invasion, which requires preoperative radiation therapy or a different treatment plan, i.e. radical hysterectomy instead of total abdominal hysterectomy. In summary, the role of imaging is to depict noninvasively the depth of myometrial invasion and the presence of lymphadenopathy and to stage the tumour extent before treatment planning.

Uterine sarcomas (leiomyosarcomas, endometrial sarcomas, malignant mixed Müllerian tumours) Sarcomas of the uterus are often highly malignant. They are rare, with an incidence of approximately 2 per 100 000 women over the age of 20 and account for 3–5% of all uterine cancers. Leiomyosarcomas show an extremely bizarre US configuration. The tumours are frequently very large at the time of the examination and it is difficult to determine the primary origin of the mass. MRI can provide an accurate preoperative assessment of uterine size and degree of involvement. Because uterine sarcomas commonly metastasize to the lung, chest CT should be considered for staging purposes.

Gestational trophoblastic disease Gestational trophoblastic neoplasms include the tumour spectrum of hydatidiform mole, invasive mole (choriocarcinoma destruens) and choriocarcinoma. They arise from fetal tissue within the maternal host and are composed of both syncytiotrophoblastic and cytotrophoblastic cells. In addition to being the first and only disseminated solid tumours that have proved to be highly curable by chemotherapy, they elaborate a unique and characteristic tumour marker—human chorionic gonadotropin (hCG). The role of imaging in gestational trophoblastic disease (GTD) has been primarily to document metastatic disease at initial diagnosis or to evaluate persistent disease. As yet, there

1229

1230

SECTION 6

• WOMEN’S IMAGING

are no specific imaging findings that allow differentiation of complete mole from invasive mole or choriocarcinoma.

Ultrasound The US appearance of GTD is most commonly a large-for-gestational-age uterus with an echogenic soft tissue mass distending the endometrial canal, punctuated by multiple small cystic spaces. In cases of complete hydatidiform mole, the small spaces correspond to the hydropic villi (Fig. 54.18)14. Although US has been used for the initial diagnosis of GTD and to exclude a normal intrauterine pregnancy, the findings are nonspecific and the diagnosis relies heavily on history and serology.

Computed tomography Uterine enlargement is the most common CT feature of GTD. Following administration of IV contrast medium, uterine enhancement is typically heterogeneous and focal enlargement or irregular hypodense regions may be seen within the myometrium14. The extrauterine manifestations of GTD are well seen on CT. Adnexal findings include bilateral ovarian enlargement by multilocular, theca lutein cysts. Locoregional spread is characterized by enhancing soft tissue density in the parametria and/or obliteration of the pelvic fat or muscle planes.

Magnetic resonance imaging On T2-weighted images, GTD is seen as a heterogeneous, predominantly high signal intensity mass that obliterates normal uterine zonal anatomy14,15. On T1-weighted images, it may be iso- or hyperintense to adjacent myometrium.The tumours are hypervascular and enlarged vessels in the broad ligament and the uterus are depicted as signal voids on both T1- and T2-weighted images. The tumours avidly enhance following injection of IV contrast medium. Carcinoma of the cervix Cervical cancer is the third most common gynaecological malignancy. Between 1960 and 1990 there was a 63% decrease in cervical cancer mortality. This improvement in mortality has been attributed to the development of the Papanicolaou smear and only minor improvement has been achieved in the survival of invasive cervical cancer. Established risk factors for

cervical cancer include early sexual activity (especially with multiple partners), cigarette smoking, immunosuppression and infection with human papilloma viruses 16 and 18. Abnormal uterine bleeding (especially after intercourse) and vaginal discharge may be symptoms leading to the diagnosis. In colposcopy, a lesion may be detected and vaginal cytology is usually positive.

Staging Recommendations for diagnostic evaluation of tumour staging derive from the TNM/FIGO clinical staging system (Table 54.2). Accurate tumour staging is important not only for prognosis, but also in determining appropriate therapy. Among the various prognostic indicators, the most critical include tumour grade, tumour size, depth of stromal invasion, parametrial extension and lymph node involvement. Because of a relatively high likelihood of parametrial invasion and/or lymph node metastases, cross-sectional imaging is recommended in evaluating cervical carcinoma patients with clinical stage IB disease or greater when the primary lesion is larger than 2 cm16,17. Imaging techniques for detection, diagnosis and staging Ultrasound Transabdominal US can show the presence of hydronephrosis but has a limited role in the evaluation of the local extent of cervical cancer. Transrectal and transvaginal US have been used in the assessment of local disease but are limited in the detection of parametrial disease and pelvic sidewall involvement due to poor soft tissue contrast, small FOV and operator dependence. More advanced cervical cancer can be visualized with transvaginal US. The cervix appears as an enlarged, irregular, hypoechoic mass that may mimic a cervical myoma. If tumour obstructs the endocervical canal, hydroand/or haematometra result.

Computed tomography Typically, CT findings in cervical cancer include enlargement of the cervix. After administration of IV contrast medium, low-attenuation areas may be seen within the tumour. These regions of decreased attenuation are a function of tumour necrosis/ulceration and/or inherent differences in the attenuation between tumour and normal

Figure 54.18 Gestational trophoblastic disease. Transverse transvaginal US (A) shows an echogenic mass with multiple cystic spaces within the endometrial cavity in a woman with a hydatidiform mole. The small cystic spaces (*, A) are felt to represent hydropic villi. Sagittal transvaginal US with colour flow (arrows, B) documents flow to the mole. Hydatidiform mole is a subtype of gestational trophoblastic disease.

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Table 54.2 TNM AND FIGO CLASSIFICATION FOR CERVICAL NEOPLASM TNM

FIGO

Cervix

Tis

0

Carcinoma in situ

T1

I

Carcinoma confined to the cervix

T1a

Ia

Invasive carcinoma identified only microscopically All gross lesions, even with superficial invasion, are stage Ib cancers Measured stromal depth should be no greater than 5 mm and no wider than 7 mm

T1a1

Ia1

Measured invasion no greater than 3 mm in depth and no wider than 7 mm

T1a2

Ia2

Measured invasion greater than 3 mm but no greater than 5 mm in depth and no wider than 7 mm

T1b1

Ib1

Clinical lesions no greater than 4 cm in size

T1b2

Ib2

Clinical lesions confined greater than 4 cm in size

T2

II

Carcinoma extending beyond cervix and involving the vagina (but not to pelvic sidewall or lower third of vagina)

T2a

IIa

Carcinoma has involved the vagina

T2b

IIb

Carcinoma has infiltrated the parametrium

T3

III

Carcinoma involving the lower third of the vagina and/or extending to the pelvic sidewall (there is no free space between the tumour and the pelvic sidewall)

T3a

IIIa

Carcinoma involving the lower third of the vagina

T3b

IIIb

Carcinoma extending to the pelvic wall and/or hydronephrosis or nonfunctioning kidney due to ureterostenosis caused by tumour

T4

IVa

Carcinoma involving the mucosa of the bladder or rectum and/or extending beyond the true pelvis

M1

IVb

Spread to distant organs

cervical tissue. Distinguishing tumour from normal cervix and parametrium may be problematic. Obstruction of the endocervical canal can result in uterine enlargement with a fluidfilled endometrial cavity. There is a consensus in the literature that the value of CT increases with higher stages of disease (Fig. 54.19) and that CT has limited value (a positive predictive value of 58%) in the evaluation of early parametrial invasion. However, CT has an accuracy of 92% in the depiction of advanced disease (Fig. 54.19).

Magnetic resonance imaging can accurately determine tumour location (exophytic or endocervical), tumour size, depth of stromal invasion and extension into the lower uterine segment17. MRI is the best single imaging investigation for this purpose, since it is better than either CT or physical examination in demonstrating parametrial invasion and as good as CT in detecting nodal metastases18. The staging accuracy of MRI ranges from 75 to 96%. The reported sensitivity of MRI in the evaluation of parametrial invasion is 69% and the specificity is 93%17,18. The MR imaging protocol should include T1-weighted axial images with a large FOV to evaluate the entire pelvis for lymphadenopathy and T2-weighted images in the axial and sagittal planes with a small FOV. Use of dynamic contrastenhanced MRI is optional16,17. On T1-weighted images, tumours are usually isointense with the normal cervix and may not be visible. On T2-weighted images, cervical cancer appears as a relatively hyperintense mass and is easily distinguishable from low signal intensity cervical stroma. Stage I tumours are confined to the uterus. Stage IA is defined as a microinvasive tumour that cannot be demonstrated at MRI. Stage IB carcinoma appears as a high signal intensity

mass in contrast to the low signal intensity fibrocervical stroma on T2-weighted images. In stage IIA tumours, segmental disruption of the upper two thirds of the vaginal wall without parametrial invasion is demonstrated on T2-weighted images. The lack of preservation of low signal intensity cervical stroma is highly indicative of parametrial invasion—stage IIB disease (Fig. 54.20). In stage IIIA, vaginal involvement reaches the lower third of the vaginal canal without extending to the pelvic sidewall. When the tumour extends to the pelvic sidewall (pelvic musculature or iliac vessels) or causes hydronephrosis, it is defined as stage IIIB. Once tumour invades the adjacent organs such as the bladder and rectal mucosa, or distant metastasis occurs, the stage is defined as IV. Although pelvic node metastases do not change the FIGO stage, para-aortic or inguinal node metastases are classified as stage IVB.

PET Although the use of PET in the initial evaluation of cervical cancer is still under investigation, PET can be used to assess nodal disease and tumour recurrence. In the detection of metastatic lymph nodes in patients with cervical cancer, PET has been reported to have a sensitivity of 91% and a specificity of 100%, which are higher than the sensitivity (73%) and specificity (83%) of MRI19. PET has added value in patients with recurrent cervical cancer who undergo salvage therapy, as it can provide precise re-staging information.

Recommended imaging approach Transvaginal US has poor soft tissue contrast and thus limited value in cervical cancer detection and in the determination of stromal invasion or parametrial extension. Multidetector contrastenhanced CT may be helpful in local staging; however, it is mainly used in advanced disease and in the assessment of lymph nodes. CT is also performed to detect distant metastases

1231

1232

SECTION 6

• WOMEN’S IMAGING

Figure 54.19 Cervical cancer, CT. Axial CT images (A–C) show a cervical cancer (T, A) which is contiguous with the adjacent parametrial fat, indicating parametrial invasion (white arrows, A). Note the presence of a filling defect within the right external femoral vein suggesting deep venous thrombosis (black arrow, A). There is bilateral para-aortic (black arrows, B) and retrocrural lymphadenopathy (black arrows, C). Also, note the presence of right hydronephrosis (H, B). Axial CT image (D) of extensive cervical cancer (T, D) in a different patient. The tumour extends to both parametria (black arrows, D) and invades the posterior aspect of the bladder (white arrows, D) and anterior rectal wall (white arrow, D).

for radiotherapy planning and for guiding interventional procedures. For locoregional staging, MRI is the method of choice because it is accurate in the determination of tumour size, location (exophytic or endocervical), depth of stromal invasion and the local extent of the tumour. As noted earlier, although the role of PET in the initial evaluation of cervical cancer is still under investigation, it can be used to assess nodal disease and tumour recurrence.

Impact of imaging on treatment The most important issue in the staging of cervical cancer is to distinguish early disease (stages I and IIA) that can be treated with surgery from advanced disease that must be treated with radiation alone or combined with chemotherapy. Imaging techniques must be directed to solve this clinically important question. Conventional radiological studies such as excretory urography, barium enema and lymphangiography have become obsolete and cross-sectional imaging, particularly CT and MRI, is now used.

THE ADNEXA Benign ovarian conditions Physiological cysts Benign ovarian cysts are common, frequently asymptomatic and often resolve spontaneously. One of the most frequently seen ovarian masses is the follicular cyst. Follicular cysts are a common asymptomatic finding, usually larger than the typical preovulatory follicle and varying in size from 3 to 8 cm in diameter. These cysts represent the failure of the fluid in an incompletely developed follicle to be reabsorbed. Follow-up US is recommended in 6 weeks, preferentially immediately after menstruation. No treatment is required in the majority of cases as the cysts usually regress spontaneously in 2 months. Two types of lutein cyst are recognized: corpus luteum and theca lutein cysts. Corpus luteum cysts are functional, nonneoplastic enlargements of the ovary, following ovulation. A persistent corpus luteum cyst may cause local pain and tenderness and either amenorrhoea or delayed menstruation, thus

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Figure 54.20 Cervical cancer, MRI. Sagittal (A) and axial (B) T2-weighted images show a large cervix cancer (T) involving the anterior fornix of the vagina (arrow, A). The tumour invades the fibrocervical stroma on the right (arrows, B). Note the tumour extends into the lower endometrial canal (*, A). Incidentally, presence of a large uterine leiomyoma (L, A) is noted.

simulating the clinical picture associated with ectopic pregnancy. Theca lutein cysts are of small to medium size, usually bilateral and filled with clear, straw-coloured fluid. These cysts are found in association with polycystic ovarian disease, hydatidiform mole, choriocarcinoma and chorionic gonadotrophin or clomiphene therapy.

Endometriosis Endometriosis is the presence of endometrial epithelium and stroma outside of endometrium and myometrium. Common locations of endometrial tissue include ovary, uterine ligaments, Fallopian tubes, rectovaginal septum, pouch of Douglas, bladder wall and umbilicus. Endometriosis is perhaps the most prevalent cause of CPP, occurring in up to 65% of women with pelvic pain. It usually affects women of reproductive age. The clinical manifestations are variable: some patients are asymptomatic while others present clinically with dysmenorrhoea, dyspareunia, abdominal pain, dysfunctional uterine bleeding and infertility. Endometriosis is a syndrome that can present with endometriomas, adhesions, or endometrial implants. Staging of endometriosis requires a laparoscopic procedure to look for any of these elements. Although imaging is most commonly used for the diagnosis and follow-up of endometriomas, laparoscopy is the gold standard as it can provide a complete evaluation for endometrial implants in both the abdomen and pelvis. Transvaginal US is the primary imaging investigation, with MRI reserved for masses atypical on US.

Ultrasound On ultrasound, endometriomas appear as cystic masses with diffuse uniform low-level echoes (Fig. 54.21). After repeated episodes of bleeding and re-bleeding, they may develop irregular walls and echogenic mural nodules. Fluid– fluid levels or fluid–debris levels represent blood products.

When the cysts have thin or thick septations, it may sometimes be difficult to differentiate them from malignant ovarian masses. US plays no role in the evaluation of endometrial implants or adhesions.

Magnetic resonance imaging Endometriomas appear hyperintense onT1-weighted images and heterogeneously hyperintense on T2-weighted images. The use of fat suppression increases the conspicuity of endometriosis (Fig. 54.21)20. Endometriomas larger than 1 cm are routinely seen, but imaging small implants remains problematic. Endometrial implants can be detected on MRI, but evaluation on MRI is inferior to laparoscopic staging. Polycystic ovarian disease (Stein–Leventhal syndrome) Polycystic ovarian disease is characterized by bilaterally enlarged polycystic ovaries, secondary amenorrhoea or oligomenorrhoea and infertility. About 50% of patients are hirsute and many are obese. Many cases of female infertility secondary to failure of ovulation are due to polycystic ovarian disease.The classic appearance on ultrasound is enlarged ovaries with echogenic central stroma and greater than 10 peripherally placed cysts less than 9 mm in diameter (Fig. 54.22).

Pelvic inflammatory disease Pelvic inflammatory disease (PID) is defined as the acute clinical syndrome associated with ascending spread of micro-organisms from the vagina or cervix to the endometrium, Fallopian tubes and/or contiguous structures. Chlamydia trachomatis and Neisseria gonorrhoeae are probably the most prevalent sexually transmitted bacteria in the western world. Early diagnosis of PID is of paramount importance in the management of women with acute upper genital tract infection, as delay will

1233

1234

SECTION 6

• WOMEN’S IMAGING

Figure 54.21 Endometrioma. Transvaginal US in sagittal (A) and coronal (B) planes demonstrates a complex cystic mass in the left ovary consistent with endometrioma (E in A, B). Although endometriomas can appear similar to haemorrhagic cysts, the irregular contour, homogeneity of the internal echoes and persistence over an extended period favours the diagnosis of endometrioma. Axial T1-weighted (C) and T1-weighted fat-suppressed MR (D) images in a different patient show multiple high signal intensity lesions within the left ovary (arrow, C), suggesting either endometriosis or haemorrhagic cysts. Note how fat suppression increases the conspicuity of haemorrhagic lesions and helps differentiate them from dermoids. Diagnosis of endometriosis was confirmed at surgery.

result in late treatment, with an attendant increased risk of complications (e.g. infertility). Classically, women with PID present with subacute lower abdominal pain that is dull in nature and usually bilateral.They also have a fever, purulent vaginal discharge, bilateral adnexal tenderness with cervical excitation and an elevated erythrocyte sedimentation rate. A number of other gynaecological and surgical conditions where lower abdominal pain is the prin-

cipal presenting complaint also mimic the clinical picture of acute PID, causing diagnostic uncertainty. Furthermore, there is evidence that in many women with acute upper genital tract infection, the condition goes unnoticed because it is atypical or asymptomatic and yet the extent of tubal damage in these women is the same as in symptomatic disease. At US, the uterus appears slightly enlarged, more hypoechoic than normal and the uterine margins are not clearly defined.

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Figure 54.22 Polycystic ovaries. Sagittal (A) and transverse (B) transvaginal ultrasound of the left ovary depicting multiple subcentimetre peripherally placed follicles in enlarged ovaries with echogenic central stroma.

These early changes are often subtle and difficult to assess. The endometrial echo may be prominent due to associated endometritis, with sonolucent margins. The adnexa appear prominent and have a complex but usually symmetrical ultrasound pattern. Small amounts of fluid may be present in the peritoneal cul-de-sac. Pyosalpinx or hydrosalpinx (Fig. 54.23) as well as tubo-ovarian abscesses (Fig. 54.24) may be present due to PID-associated adhesions. Septic abortion, intrauterine manipulation and pelvic operations may also provoke a pyogenic adnexal abscess. On US, a tubo-ovarian abscess presents as an adnexal mass with a thickened echogenic wall. There may also be hypoechoic areas with associated thick and irregular septations. On CT, an adnexal abscess is seen as a soft tissue mass with central areas of low attenuation.Thick, irregular walls are commonly present and it may be difficult to differentiate an ovarian

Figure 54.23 Hydrosalpinx. Sagittal transvaginal ultrasound image demonstrates a cystic lesion of tubular appearance within the pelvis. Small internally projecting nodules are compatible with the fimbriae and give the cyst (*) a ‘cogwheel’ appearance, typical of hydrosalpinx.

abscess from a necrotic tumour or endometrioma based solely on the CT finding. MRI features of adnexal abscess are similarly nonspecific. Pyosalpinx and hydrosalpinx may present as a flask-shaped cystic adnexal mass that may contain faintly echogenic material due to debris. A large hydrosalpinx may be indistinguishable from an ovarian cyst. Both CT and MRI may demonstrate a serpentine cluster of cysts.

Pelvic varices Pelvic varices are dilated veins in the broad ligament and ovarian plexus. When symptomatic, the condition is called ‘pelvic congestion syndrome’. Symptoms have been described as dull, aching pain, often occurring during walking, sexual intercourse, or other activities that create increased intra-abdominal pressure. Varices are most often found in multiparous women of reproductive age. The criteria for diagnosis include venous structures greater than 4 mm in diameter, slower than 3 cm s−1 velocity flow and connecting arcuate veins within the myometrium.

Figure 54.24 Pelvic tubo-ovarian abscess. Transverse transvaginal US shows a complex left ovarian mass (*) in a woman with pelvic pain and fever. This constellation of findings is consistent with pelvic inflammatory disease complicated by tubo-ovarian abscess. U = uterus.

1235

1236

SECTION 6

• WOMEN’S IMAGING

Varices show on CT as serpiginous structures that enhance following administration of contrast medium. 3D T1 gradient-echo sequences performed after the IV administration of gadolinium are the most effective MR imaging sequence for demonstrating pelvic varices. Blood flow in pelvic varices appears with high signal intensity4.

of virilization can be seen in patients presenting with this tumour type.

Benign tumours of the ovary

Granulosa cell tumours are all malignant tumours but are mentioned here because they are generally confined to the ovary when they present and so have a good prognosis. However, they grow very slowly and recurrences are frequently diagnosed.

Ninety per cent of all ovarian tumours are benign, although this varies with age.

Ultrasound The ovaries are well imaged with transvaginal

Germ cell tumours are among the most common ovarian tumours seen in women less than 30 years of age. Overall, only 2–3% of germ cell tumours are malignant.

Dermoid cysts (mature cystic teratoma) are the only benign germ cell tumours and are quite common. Dermoid cysts stem from cells that differentiate into embryonic tissues and account for around 40% of all ovarian neoplasms. They are most common in young women with a median age at presentation of 30 years.

Mature solid teratomas are rare tumours containing mature tissues just like the dermoid cysts. They must be differentiated from immature teratomas, which are malignant. Benign epithelial tumours The majority of ovarian neoplasms, both benign and malignant, arise from the ovarian surface epithelium.

Serous cystadenoma is the most common benign epithelial tumour and is usually a unilocular cyst with papilliferous processes on the inner surface and occasionally on the outer surface.

Mucinous cystadenoma contributes to 15–25% of all ovarian tumours and is the second most common epithelial tumour.

Endometrioid cystadenomas are often malignant and difficult to distinguish from endometriosis.

Brenner tumours account for only 1–2% of all ovarian tumours and probably arise from Wolffian metaplasia of the surface epithelium. Sex cord stromal tumours Theca cell tumours are almost always benign, solid and unilateral. Many produce oestrogens in sufficient quantity to provoke systemic effects, such as precocious puberty, postmenopausal bleeding, endometrial hyperplasia, or endometrial cancer.

Fibromas are uncommon tumours and present most frequently around 50 years of age. Most are derived from stromal cells and are similar to thecomas. Sertoli–Leydig cell tumours are rare and usually of low grade malignancy. Many produce androgens and clinical signs

US. Changes in their appearance can be correlated with their functional status. Follicular cysts may be up to 10 cm in diameter and difficult to distinguish from other ovarian tumours unless serial ultrasound examinations are obtained, in which regression of the follicular cyst is demonstrated.The ultrasound appearance of dermoid cysts varies. The three most common appearances are: (A) a cystic mass with an echogenic nodule projecting into the lumen; (B) a predominantly echogenic mass with posterior sound attenuation owing to the presence of sebaceous material and hair; and (C) a cystic mass with fine internal echogenic lines also representing hair. A fluid–fluid level may be seen which represents sebaceous material floating on fluid. The imaging features of other benign ovarian tumours are nonspecific. Different imaging techniques are combined for ovarian lesion characterization, as discussed more extensively in the next section.

Magnetic resonance imaging The MRI appearance of simple cysts is diagnostic: low signal intensity on T1-weighted images, high signal intensity on T2-weighted images and no enhancement following the administration of IV contrast medium. The strength of MRI is its ability to characterize adnexal lesions, particularly in the case of masses that are indeterminate on US (see next section for a full discussion). Among benign lesions, MRI can also be used to diagnose dermoid cysts with confidence as the fat or sebum within the cyst parallels the signal intensity of fat on all pulse sequences. Specifically, by exploiting the processional frequency differences between fat and water protons, fat saturation techniques cause the fatty elements to lose signal and are 100% specific and 96% accurate in identifying dermoid cysts (Fig. 54.25). Additional features include fat–water chemical shift artefact, fat–fluid and/or fluid–fluid levels, layering debris, low signal intensity calcifications (e.g. teeth) and soft tissue protuberances (Rokitansky nodules or dermoid plugs) attached to the cyst wall.

Malignant ovarian conditions Ovarian carcinoma Ovarian neoplasms account for more cancer-related deaths than all other primary cancers of the reproductive system. In general, ovarian cancer is a disease of postmenopausal women and, occasionally, prepubescent girls. The cause of ovarian cancer is unknown, although a number of risk factors have been identified. Chronic anovulation, multiparity and a history of breast feeding seem to be protective, whereas genetic factors appear to play an important role in the development of progression of ovarian cancer. Ten per cent of ovarian

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Figure 54.25 Ovarian dermoid, intrauterine pregnancy. T2-weighted sagittal (A) and T1-weighted axial (B) MRI. The sagittal image demonstrates a fetus (F) in vertex position. Situated posterior to the uterus in the cul-de-sac is a hyperintense cystic and solid, rounded mass (M). A portion of the mass is high signal on T1-weighted image and falls in signal intensity following a fat saturation pulse (arrow B). This is consistent with fat, allowing the diagnosis of dermoids to be made with confidence.

cancers are due to hereditary syndromes such as BRCA-1 and BRCA-2 mutations (risk of breast cancer) and Lynch syndrome II (risk of colon cancer). Approximately 90% of ovarian cancers are of epithelial origin. Epithelial tumours are subtyped as serous (50%), mucinous (20%), endometroid (20%), clear cell (10%), or undifferentiated (1%). Nonepithelial cancers include malignant granulosa cell tumour, dysgerminoma, immature teratoma, endodermal sinus tumour and metastases to the ovary.

Staging Cross-sectional imaging is better accepted and more commonly used in the evaluation and staging of ovarian car-

cinoma than for other gynaecological malignancies21,22. The TNM and FIGO staging systems are outlined in Table 54.3. Imaging is an adjunct to surgical staging and is becoming a valuable tool in the detection of nonresectable disease. Findings that indicate nonresectable disease include: • in the pelvis: invasion of pelvic sidewall or urinary bladder, or urinary obstruction • in the upper abdomen: tumour deposits greater than 1 cm in size in the gastrosplenic ligament, lesser sac, fissure for the ligamentum teres, porta hepatis, subphrenic space, small bowel mesentery, or retroperitoneal nodes above the renal hila • distant metastases (i.e. liver, spleen, lung).

Table 54.3 TNM AND FIGO STAGING FOR PRIMARY OVARIAN CANCER TNM

FIGO

Ovary

T1

I

Growth limited to ovaries

T1

Ia

Growth limited to one ovary, no ascites; no tumour on external surface; capsule intact

T1b

Ib

Growth limited to both ovaries, no ascites; no tumour on external surface; capsule intact

T1c

Ic

Tumour either stage Ia or Ib, but tumour on surface of one or both ovaries; or with capsule ruptured; or with ascites present containing malignant cells; or with positive peritoneal washings

T2

II

Growth involving one or both ovaries with pelvic extension

T2a

IIa

Extension and/or metastasis to the uterus or tubes

T2b

IIb

Extension to other pelvic tissue

T2c

IIc

Tumour either stage IIa or IIb but tumour on surface of one or both ovaries; or with capsule ruptured; or with ascites present containing malignant cells; or with positive peritoneal washings

T3 and/or N1

III

Growth involving one or both ovaries with peritoneal implants outside the pelvis or positive retroperitoneal or inguinal nodes

T3a

IIIa

Tumour grossly limited to the true pelvis with negative nodes but with histologically confirmed microscopic seeding of abdominal peritoneal surfaces

T3b

IIIb

Tumour with histologically confirmed implants on abdominal peritoneal surfaces, none exceeding 2 cm in diameter

Superficial liver metastasis equals stage III

Nodes are negative T3c and/or N1

IIIc

M1

IV

Abdominal implants greater than 2 cm in diameter or positive retroperitoneal or inguinal nodes Growth involving one or both ovaries with distant metastases If pleural effusion is present there must be positive cytology to allot a case to stage IV Parenchymal liver metastasis equals stage IV

1237

1238

SECTION 6

• WOMEN’S IMAGING

Imaging techniques for detection, diagnosis and staging Ultrasound Combined transabdominal and transvaginal US has been advocated for the detection of ovarian carcinoma. These studies provide superb morphological detail of the adnexa, allowing detection of masses before they are clinically apparent. Ultrasound features that suggest benignancy and malignancy have been well described. Although several scoring systems have been devised to predict the nature of an adnexal mass, most researchers agree that wall irregularity, thick septations (> 3 mm), papillary projections, solid components and size (> 9 cm) are suspicious for malignancy23. As colour and duplex US become more widely available, their potential value for differentiating benign from malignant ovarian masses is being explored (Fig. 54.26)23. Malignant tumours often have neovascularity that consists of blood vessels with walls that have little or no smooth muscle support. As a result, these vessels frequently have a characteristic waveform with a low resistive index (RI < 0.4) (peak systolic–end diastolic Doppler shift/peak systolic Doppler shift). A meta-analysis of the literature demonstrated that the best adnexal lesion characterization is achieved by the combined use of grey scale and colour Doppler US24. Assessment of adnexal masses on imaging for malignancy or benignancy is only possible if the ovaries can be visualized. This may be problematic in postmenopausal women.

Computed tomography is the most commonly performed study for the preoperative staging of a suspected ovarian carcinoma. It is particularly useful in determining the extent of cytoreductive surgery required to optimize subsequent chemotherapeutic response. Ovarian cancer is frequently bilateral (Fig. 54.27D). On CT, ovarian cancer demonstrates varied morphological patterns, including a multilocular cyst with thick internal septations and solid mural or septal components, a partially cystic and solid mass (Fig 54.27D) and a lobulated papillary heterogeneous solid mass. The outer border of the mass may be irregular and poorly defined and amorphous

coarse calcifications and contrast enhancement may be seen in the cyst wall or soft tissue components25. It is possible to suggest the histological subtype of the epithelial cancer based on the imaging findings. Calcification suggests a serous tumour, whereas high density within the locules of a multilocular tumour is suggestive of proteinaceous fluid in a mucinous tumour. Endometroid carcinomas are associated with endometrial hyperplasia or carcinoma in 20–30% of cases. Intraperitoneal dissemination is the most common route of spread of ovarian cancer. Peritoneal implants appear as nodular or plaque-like enhancing soft tissue masses of varying size (Fig. 54.27E). Ascitic fluid may outline small implants, facilitating detection. Although peritoneal implants may occur anywhere in the peritoneal cavity, the most common sites include the pouch of Douglas, paracolic gutters, surface of the small and large bowel, greater omentum, surface of the liver (perihepatic implants) and subphrenic space. CT is useful in differentiating between subcapsular and parenchymal liver metastasis, which alters staging and therapy26. Ascites is a nonspecific finding but, in a patient with ovarian cancer, usually indicates peritoneal metastases27. In addition to peritoneal implantation, ovarian cancer also spreads by local continuity to the ovary, uterus and other pelvic organs. Surgically important features of local spread that may be detected by imaging are invasion of the pelvic sidewall, rectum, sigmoid colon, or urinary bladder. Pelvic sidewall invasion should be suspected when the primary tumour lies within 3 mm of the pelvic sidewall or when the iliac vessels are surrounded or distorted by tumour.The ovarian lymphatic vessels are another important route of metastatic spread.While enlarged nodes are likely to be involved, CT is unable to exclude disease in normal-sized nodes. The CT appearance of ovarian metastasis is indistinguishable from that of a primary ovarian neoplasm and the stomach and colon should be carefully examined as potential primary tumour sites. Occasionally, it is not possible to determine whether the origin of a pelvic mass is ovarian or uterine on CT.

Magnetic resonance imaging The role of MRI in patients

Figure 54.26 Ovarian carcinoma. Sagittal transvaginal US image demonstrates a large complex cystic mass arising from the left adnexa. The presence of flow within the solid nodule suggests malignant aetiology.

with suspected or known ovarian carcinoma is still evolving. To optimize MR detection and characterization of an adnexal mass, contrast-enhanced protocols and attention to eliminating, or at least limiting, bowel motion are needed. Both transvaginal US and contrast-enhanced MRI have high sensitivity (97% and 100%, respectively) in the identification of solid components within an adnexal mass. MRI, however, shows higher accuracy (93%)27. Primary and ancillary criteria for characterizing an adnexal mass as malignant have been established. Primary criteria for malignancy include: (A) size larger than 4 cm; (B) a cystic lesion with a solid component; (C) irregular wall thickness greater than 3 mm; (D) septa greater than 3 mm and/ or the presence of vegetations or nodularity; and (E) a solid mass with the presence of necrosis. Ancillary criteria for malignancy include: (A) involvement of pelvic organs or sidewall; (B) peritoneal, mesenteric, or omental disease; (C) ascites; or (D) adenopathy. Utilizing unenhanced T1-, T2- and contrastenhanced T1-weighted sequences, the presence of at least one of the primary criteria coupled with a single criterion from

CHAPTER 54

• IMAGING IN GYNAECOLOGY

Figure 54.27 Bilateral ovarian carcinoma. Transverse (A) transvaginal ultrasound image of the pelvis shows bilateral cystic adnexal masses (T). Sagittal images of right (B) and left (C) ovaries demonstrate cystic mass (T) with mural nodularity (B) and multiple septations (C). MDCT of a different patient shows bilateral complex solid and cystic adnexal masses (T, D), highly suggestive of ovarian carcinoma, and demonstrates the presence of omental tumour implants (white arrows, E). Note also the presence of left para-aortic, interaortocaval (black arrows, E) and superior diaphragmatic (arrow, F) lymphadenopathy.

1239

1240

SECTION 6

• WOMEN’S IMAGING

the ancillary group correctly characterizes 95% of malignant lesions28. MRI is very sensitive (95%) for detection of peritoneal metastases, which show delayed enhancement on contrastenhanced MRI29.

PET PET and PET-CT have a potential role in evaluating patients for recurrent ovarian cancer, particularly those with negative CT or MRI findings and rising tumour marker levels30. PET is also useful in the detection of implants difficult to assess by conventional imaging studies, such as serosal bowel implants or tumour deposits within the small bowel mesentery.

Recommended imaging approach US is the primary technique for detection and characterization of adnexal masses. MRI is a problem-solving investigation in cases of indeterminate adnexal masses and the best technique to assess pelvic sidewall invasion. Multidetector contrast-enhanced CT of the abdomen and pelvis is the imaging technique of choice for preoperative staging and for follow-up. CT is the primary technique for prediction of tumour resectability, tumour extent into the upper abdomen and related complications such as hydronephrosis and bowel obstruction. PETCT is very valuable in the setting of recurrent disease and particularly useful for detecting tumour deposits in mesentery and bowel serosa.

Impact of imaging on treatment Early ovarian cancer is treated with comprehensive staging laparotomy, which includes transabdominal hysterectomy and bilateral salpingooophorectomy (TAH/BSO), omentectomy, retroperitoneal lymph node sampling, peritoneal and diaphragmatic biopsies and cytology of peritoneal washings. Advanced but operable disease is treated with primary cytoreductive surgery (debulking) followed by adjuvant chemotherapy. Patients with nonresectable disease may benefit from neoadjuvant (preoperative) chemotherapy before debulking. Cross-sectional imaging plays a crucial role: • in treatment planning by identifying patients with inoperable disease who may be more appropriately managed by neoadjuvant chemotherapy followed by cytoreductive surgery after tumour shrinkage • following treatment response • in detection of recurrent disease. It is important to realize that second-look surgery is no longer routine and imaging diagnosis of recurrence may obviate a second-look laparotomy since secondary cytoreduction is only justified if resection is possible with no residual tumour22,25.

VULVA AND VAGINA Carcinoma of the vulva Metastases from carcinoma of the vulva spread to the superficial inguinal nodes and subsequently to the deep inguinal and

external iliac groups. CT and MRI are usually performed for staging of vulvar carcinoma.

Vulvar varices These usually occur in multiparous women, often with varicose veins of the legs. Contrast medium may be injected directly into vulvar varicose veins to show their connections.

Congenital abnormalities Abnormalities of the vagina have been described in previous sections.

Vaginal malignancies Primary vaginal malignancies are uncommon. Metastatic tumours to the vagina are more frequent and originate most commonly from carcinoma of the endometrium and cervix, followed by melanoma and carcinoma of the colon and kidney.

CONTRACEPTION Intrauterine contraceptive devices The intrauterine contraceptive device (IUCD) is an effective, long-acting method of contraception, which is perhaps underrated in the developed world. It exerts a local inflammatory reaction within the cavity of the uterus, which probably interferes with the viability of both sperm and eggs in addition to inhibiting implantation. Modern copper-containing devices are effective for long periods and it is now recommended that they need to be changed only every 5 years. Using modern IUCDs, the rates for perforation and expulsion are 4–9 and 0.2 per 100 users, respectively. The incidence of both complications is increased with postpartum insertions, which should be delayed until 6–8 weeks after delivery. Transvaginal ultrasound is the initial technique of choice in evaluating possible IUCD malposition. Most devices are echogenic on ultrasound and visualized as a linear structure within the endometrial cavity. During the examination, the presence of an intrauterine gestation is excluded and the position of the device established. The device may be identified within the myometrium. If it has been completely extruded from the uterus, a plain radiograph will establish whether it has either been expelled from the patient or lies within the pelvis. CT may accurately depict the presence of the device within the pelvic cavity. IUCDs can be safely imaged with MRI and their presence does not create artefacts that impede image interpretation31.

REFERENCES 1. Carrington B M, Hricak H, Nuruddin R N, Secaf E, Laros R K Jr, Hill E C 1990 Mullerian duct anomalies: MR imaging evaluation. Radiology 176: 715–720 2. Baber R J, McSweeney M B, Gill R W et al 1988 Transvaginal pulsed Doppler ultrasound assessment of blood flow to the corpus luteum in IVF patients following embryo transfer. Br J Obstet Gynaecol 95: 1226–1230

CHAPTER 54

3. De Souza N M, Brosens J J, Schwieso J E, Paraschos T, Winston R M 1995 The potential value of magnetic resonance imaging in infertility. Clin Radiol 50: 75–79 4. Kuligowska E, Deeds L 3rd, Lu K 3rd 2005 Pelvic pain: overlooked and underdiagnosed gynecologic conditions. Radiographics 25: 3–20 5. Murase E, Siegelman E S, Outwater E K, Perez-Jaffe L A, Tureck R W 1999 Uterine leiomyomas: histopathologic features, MR imaging findings, differential diagnosis and treatment. Radiographics 19: 1179–1197 6. Ascher S M, Jha R C, Reinhold C 2003 Benign myometrial conditions: leiomyomas and adenomyosis. Top Magn Reson Imaging 14: 281–304 7. Tamai K, Togashi K, Ito T, Morisawa N, Fujiwara T, Koyama T 2005 MR imaging findings of adenomyosis: correlation with histopathologic features and diagnostic pitfalls. Radiographics 25: 21–40 8. Reinhold C, Tafazoli F, Mehio A et al 1999 Uterine adenomyosis: endovaginal US and MR imaging features with histopathologic correlation. Radiographics 19: S147–160 9. Leitich H, Brunbauer M, Kaider A, Egarter C, Husslein P 1999 Cervical length and dilatation of the internal cervical os detected by vaginal ultrasonography as markers for preterm delivery: a systematic review. Am J Obstet Gynecol 181: 1465–1472 10. Hricak H, Chang Y C, Cann C E, Parer J T 1990 Cervical incompetence: preliminary evaluation with MR imaging. Radiology 174: 821–826 11. Arko D, Takac I 2000 High frequency transvaginal ultrasonography in preoperative assessment of myometrial invasion in endometrial cancer. J Ultrasound Med 19: 639–643 12. Manfredi R, Mirk P, Maresca G et al 2004 Local–regional staging of endometrial carcinoma: role of MR imaging in surgical planning. Radiology 231: 372–378 13. Saga T, Higashi T, Ishimori T et al 2003 Clinical value of FDG-PET in the follow up of post-operative patients with endometrial cancer. Ann Nucl Med 17: 197–203 14. Green C L, Angtuaco T L, Shah H R, Parmley T H 1996 Gestational trophoblastic disease: a spectrum of radiologic diagnosis. Radiographics 16: 1371–1384 15. Hricak H, Demas B E, Braga C A, Fisher M R, Winkler M L 1986 Gestational trophoblastic neoplasm of the uterus: MR assessment. Radiology 161: 11–16 16. Scheidler J, Heuck A F 2002 Imaging of cancer of the cervix. Radiol Clin North Am 40: 577–590, vii

• IMAGING IN GYNAECOLOGY

17. Okamoto Y, Tanaka Y O, Nishida M, Tsunoda H, Yoshikawa H, Itai Y 2003 MR imaging of the uterine cervix: imaging–pathologic correlation. Radiographics 23: 425–445, quiz 534–535 18. Hricak H, Powell C B, Yu K K et al 1996 Invasive cervical carcinoma: role of MR imaging in pretreatment work-up: cost minimization and diagnostic efficacy analysis. Radiology 198: 403–409 19. Reinhardt M J, Ehritt-Braun C, Vogelgesang D et al 2001 Metastatic lymph nodes in patients with cervical cancer: detection with MR imaging and FDG PET. Radiology 218: 776–782 20. Woodward P J, Sohaey R, Mezzetti T P Jr 2001 Endometriosis: radiologic–pathologic correlation. Radiographics 21: 193–216; questionnaire 288–294 21. Tempany C M, Zou K H, Silverman S G, Brown D L, Kurtz A B, McNeil B J 2000 Staging of advanced ovarian cancer: comparison of imaging modalities—report from the Radiological Diagnostic Oncology Group. Radiology 215: 761–767 22. Woodward P J, Hosseinzadeh K, Saenger J S 2004 From the archives of the AFIP: radiologic staging of ovarian carcinoma with pathologic correlation. Radiographics 24: 225–246 23. Brown D L, Doubilet P M, Miller F H et al 1998 Benign and malignant ovarian masses: selection of the most discriminating gray-scale and Doppler sonographic features. Radiology 208: 103–110 24. Kinkel K, Hricak H, Lu Y, Tsuda K, Filly R A 2000 US characterization of ovarian masses: a meta-analysis. Radiology 217: 803–811 25. Coakley F V 2002 Staging ovarian cancer: role of imaging. Radiol Clin North Am 40: 609–636 26. Coakley F V, Choi P H, Gougoutas C A et al 2002 Peritoneal metastases: detection with spiral CT in patients with ovarian cancer. Radiology 223: 495–499 27. Hricak H, Chen M, Coakley F Vet al 2000 Complex adnexal masses: detection and characterization with MR imaging—multivariate analysis. Radiology 214: 39–46 28. Stevens S K, Hricak H, Stern J L 1991 Ovarian lesions: detection and characterization with gadolinium-enhanced MR imaging at 1.5 T. Radiology 181: 481–488 29. Ricke J, Sehouli J, Hach C, Hanninen E L, Lichtenegger W, Felix R 2003 Prospective evaluation of contrast-enhanced MRI in the depiction of peritoneal spread in primary or recurrent ovarian cancer. Eur Radiol 13: 943–949

1241

CHAPTER

Skull and Brain: Methods of Examination and Anatomy

55

Dawn Saunders, H. Rolf Jäger, Alison D. Murray and John M. Stevens

Methods of examination • Plain radiography • Cross-sectional imaging techniques • Special techniques • Magnetic resonance imaging Anatomy Advanced magnetic resonance imaging • Magnetic resonance diffusion imaging • Functional imaging techniques • Functional magnetic resonance imaging

Vascular imaging • Techniques Computed tomography angiography Magnetic resonance angiography Anatomy of the cerebral arteries and veins • Intracranial arteries • Posterior cerebral arteries • Intracranial veins

METHODS OF EXAMINATION Almost all neuroradiological examinations consist of cross-sectional imaging with computed tomography (CT) and magnetic resonance imaging (MRI). Plain radiography is assuming a historical role, but general radiologists and neuroradiologists still need to be familiar with the appearances of plain radiographs of the skull.Vascular grooves and other bony landmarks are shown well by skull radiographs and remain part of the core knowledge required in professional examinations. Knowledge of these structures is also useful for the interpretation of more advanced imaging techniques, such as CT and MRI of the skull base and pituitary region. For these reasons much of the original section on plain radiography is retained in this edition. The technical principles of CT and MRI and radionuclide studies are covered elsewhere.This chapter discusses only specific issues concerning their application to the imaging of the brain, covering advanced imaging methods such as MR diffusion and MR perfusion imaging and functional imaging with MRI, positron emission tomography (PET), and singlephoton emission computed tomography (SPECT). Noninvasive vascular imaging techniques, such as MR angiography (MRA) and CT angiography (CTA) compete now with intra-arterial cerebral angiography, replacing it for many indications. The technical principles of invasive and noninva-

sive imaging of the extra- and intracranial vessels are discussed and followed by an overview of the vascular anatomy.

PLAIN RADIOGRAPHY General considerations Skull radiography has been replaced by axial imaging methods such as CT and MRI but may still be used on occasions. Scrupulous patient positioning is essential and high-definition films in grid cassettes (24–40 lines cm−1) are preferred. Tube voltages of 50–90 kVp are employed, with a focal spot no larger than 0.6 mm and a film-focus distance of 90 cm. An isocentric skull unit is desirable. Recommendations in this chapter are based on those of the 1961 Commission on Neuroradiology of the World Federation of Neurology, with certain modifications suggested by du Boulay1.

Lines2 1 The anthropological base line is drawn from the lower margin of the orbit to the superior border of the external auditory meatus (EAM) known as Reid’s or Frankfurt line; or from the outer canthus to the centre of the meatus: orbitomeatal (OM) line.

1246

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

2 The auricular line: perpendicular to the above, drawn vertically through the EAM. 3 The interpupillary line: through both pupils, perpendicular to the median sagittal plane (see below).

Planes 1 The medial sagittal plane is the anatomical midline. 2 The horizontal (Frankfurt) plane contains both anthropological base lines; it is perpendicular to the above. A corresponding orbitomeatal plane includes both orbitomeatal lines. 3 The frontal biauricular (coronal) plane: perpendicular to both the preceding planes, passing through the EAM. The full ‘skull series’ includes the four views described below.

Lateral projection (Fig. 55.1) With the midsagittal plane parallel to the detector plate, a horizontal beam is centred 25 mm anterior to the EAM and 10 mm above the orbitomeatal line with the patient supine or sitting, thus placing the sella turcica in the centre of the beam. The anterior clinoid processes and the orbital roofs on the two sides should be superimposed.

Posteroanterior (occipitofrontal, OF) projection (Fig. 55.2) The midsagittal and orbitomeatal planes are perpendicular to the plate: this is achieved by resting the nose and forehead on the cassette. The tube is angled 20 degrees caudally, with the beam centred on the nasion. A fronto-occipital (anteroposterior [AP]) projection should not be used as it causes magnification and blurring of the more important anterior structures.The petrous ridges should be projected at or near the inferior orbital margins.

Half-axial anteroposterior (Towne’s) projection (Fig. 55.3) The median sagittal plane is again perpendicular to the plate. Placing the occiput on the cassette, with the orbitomeatal or anthropological line perpendicular to it, and angling the tube 30 degrees caudally gives an effective caudal angulation of

Figure 55.1

25–40 degrees. The beam is centred on the foramen magnum. Lateral rotation is assessed as described earlier.

Submentovertical (base) projection (Fig. 55.4) With the patient supine, the neck is fully hyperextended by placing a thick pillow or bolster under the shoulders so that the anthropological line is parallel with the plate; the median sagittal plane is again perpendicular to it. The beam is centred on the bi-auricular line, halfway between the angles of the mandible. A satisfactory radiograph shows no rotation and the angles of the mandibles lie just anterior to the middle ear cavities.

Anatomy The standard projections are shown in Figures 55.1–55.4 and the detailed radioanatomy of the pituitary fossa in Figure 55.5. For anatomical purposes, the basal portion of the skull is divided into three fossae. The anterior cranial fossa lies above the orbital roofs, anterior to the ridge formed by the greater and lesser wings of the sphenoid. It contains the frontal lobes and the olfactory bulbs and tracts. The middle cranial fossa lies posteroinferior to the sphenoid ridge, on either side of the basisphenoid, and is bounded laterally by the squamous temporal bone and posteroinferiorly by the petrous ridge. It contains the temporal lobe and should not be confused with the temporal fossa, which is the extracranial space deep to the zygomatic arch. The posterior cranial fossa comprises all the space below the tentorium or the centrally placed tentorial hiatus and above the foramen magnum. It is bounded anteriorly by the clivus (basisphenoid and basiocciput) in the midline, anterolaterally by the posterior surface of the petrous bone, and elsewhere by the occipital bone. Superolaterally its extent is indicated on skull radiographs by the groove for the transverse venous sinus, but in the midline the apex of the tentorium lies almost at the level of the pineal. Marked variations in the shape of the tentorium (e.g. the straight sinus can be

Lateral radiograph and diagram of the skull. (For key, see Figure 55.4)

CHAPTER 55

Figure 55.2

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Occipitofrontal radiograph and diagram of the skull. (For key, see Figure 55.4)

Figure 55.3 Half-axial (Towne’s) radiograph and diagram of the skull. (For key, see Figure 55.4)

1247

1248

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 55.4 Submentovertical radiograph and diagram of the skull. Key for Figures 55.1–55.4: a = alveolus, ac = air cells in petrous bone, at = atlas, c = clivus, cc = carotid canal, co = cochlea, cs = coronal suture, csp = cervical spine, ds = dorsum sellae, eam = external auditory meatus (superimposed on lateral projection), eop = external occipital protuberance, es = ethmoid sinus, eu = Eustachian tube, fm = foramen magnum, fo = foramen ovale, fs = frontal sinus, fsp = frontal spinosum, fz = frontozygomatic synostosis, gw = greater wing of sphenoid bone, h = hyoid bone, hp = hard palate, iam = internal auditory meatus (superimposed on lateral projection), il = innominate line, iof = inferior orbital fissure, iop = internal occipital protuberance, it = inferior turbinate, lo = lateral wall of orbit, ls = lambdoid suture, lw = lateral wall of maxillary antrum, m = mastoid process, ma = maxillary antrum, mm = groove for middle meningeal artery, mn = mandible, mw = medial walls of orbit and maxillary antrum (superimposed), np = nasopharynx, ns = nasal septum, o = odontoid, or = roof of orbit, os = occipital squame, oss = ossicles (auditory), p = petrous bone, pc = posterior clinoid process, pr = petrous ridge, ps = planum sphenoidale, pt = pterygoid plates, pte = pterion, rp = retropharyngeal soft tissue, sg = groove for superior sagittal sinus, sof = superior orbital fissure, sps = sphenoid sinus, sr = sphenoid ridge, ss = sagittal suture, tm = temporomandibular joint, tr = tympanic ring, ts = groove for transverse sinus, tt = temporal tubercle, v = venous markings, z = zygomatic arch.

Figure 55.5 Diagram of the sellar region. (A) Lateral projection; (B) frontal projection; (C) from above. acp = anterior clinoid process, c = cortical bone lining sphenoid sinus, cl = clivus, ds = dorsum sellae, es = ethmoid sinus, f = floor of sella turcica, gw = greater wing of sphenoid, l = lamina papyracea, ld = lamina dura (cortical bone lining sella turcica), ls = limbus sphenoidale, mcp = middle clinoid process (inconstant), ns = nasal septum, oc = optic canal, pcp = posterior clinoid process, ps = planum sphenoidale, s = carotid sulcus, sc = sulcus chiasmaticus, sof = superior orbital fissure, ss = sphenoid suture, ts = tuberculum sellae.

CHAPTER 55

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Pituitary region Here thinner sections are used, usually 2–3 mm, and smaller fields of view. Imaging can be in the axial plane, with multi-planar reformatting to generate views in the coronal plane if deemed appropriate. Many centres prefer imaging in the direct coronal plane, with the patient prone and head maximally extended, which is not always possible. The preliminary digital CT radiograph is used to select the most favourable plane to avoid metallic dental work projecting artefact into the field of interest. IV contrast medium is given (see later).

Figure 55.6 The cranial fossae and dural reflections. The right side of the cranial vault has been removed, as have all the cranial contents apart from the dura mater. acf = floor of the anterior cranial fossa (= orbital roof), acp = anterior clinoid process, cg = crista galli, cr = cribriform plate region, ds = dorsum sellae, f = falx cerebri, fe = free edge of the tentorium, h = tentorial hiatus, iss = inferior sagittal sinus, ls = lateral (transverse) sinus, mcf = middle cranial fossa, pr = petrous ridge, pt = pterion, s = cavernous sinus, sr = sphenoid ridge, ss = straight sinus, sss = superior sagittal sinus, t = tentorium, th = torcular Herophili (confluence of the dural venous sinuses).

almost vertical or nearly horizontal) make plain radiographic assessment of the size of the posterior fossa unreliable. It contains the brainstem, cerebellum, fourth ventricle, lower cranial nerves and vertebro-basilar arterial tree. The major components of the cranial cavity defined by the dural structures are shown schematically in Figure 55.6.The base of the skull is perforated by a number of foramina and canals (the latter being longer than the former) (Table 55.1). Causes of physiological intracranial calcification are listed in Table 55.2.

CROSS-SECTIONAL IMAGING TECHNIQUES Computed tomography Routine CT examinations of the brain and specific areas Brain computed tomography The routine study of the head is made in the axial plane. Some centres prefer to angle the plane so that it is parallel to the orbital roof, and therefore the routine examination does not include the eye. Others image parallel to the cantho-meatal line, which then includes the orbital contents. Slice thickness was originally 5–10 mm for the supratentorial compartment and 5 mm or less for the posterior fossa, mainly in an attempt to reduce beam-hardening artefact. But multidetector CT can provide thinner cuts and many more options, including multiplanar reconstructions. Window widths and levels are set to maximize contrast between grey and white matter, and are kept constant from patient to patient. Head injury examinations should be reviewed using a bone reconstruction algorithm.

Craniocervical junction This region is best examined by MRI. Therefore this investigation is not routine, but can be invaluable in unravelling complex skeletal anomalies in the region, and checking the integrity of bone in erosive or destructive processes prior to excisions or stabilization procedures. Thin sections are necessary and high-resolution bone review algorithms. The axial plane is appropriate, using 2 mm or less slice thickness in either spiral or multislice modes. Multi-planar reformatting and occasionally shaded surface rendering can be extremely helpful in interpretation. Orbits and petrous bones These special areas are covered in dedicated chapters.

Intravenous contrast medium A plain unenhanced study always should be performed first. IV contrast enhancement shows areas of blood–brain barrier breakdown within the brain, which is a very nonspecific phenomenon; it can make small lesions much more conspicuous. Some centres still insist on IV contrast medium for all CT examinations of the head but most centres are more selective. Guidelines for contrast medium use include: (A) when plain CT is abnormal and there is a reasonable expectation that enhancement may improve diagnostic accuracy; (B) when lesions are suspected close to the skull base or in the posterior fossa (this includes pituitary and imaging for visual failure); (C) when staging for carcinomas known frequently to metastasize to the brain; (D) when suspecting focal intracranial infections; and (E) when meningeal disease is suspected such as may be caused by sarcoidosis or metastases (e.g. cranial nerve palsies, especially if multiple). An IV injection of iodinated contrast medium licensed for intravascular use, at dose equivalents of 15–30 g of iodine generally are given; some clinics use two or even three times that dosage. Most units now prefer to inject through indwelling cannulae rather than ordinary needles, so that IV access is assured should an adverse reaction occur. Life-support equipment and medical staff trained in its use should be nearby.

Anatomy on CT The electron density of grey matter is slightly greater than white, and adequate images should allow clear differentiation of the larger grey and white matter areas of cerebral hemispheres and cerebellum. Usually only the decussation of the superior cerebellar peduncles can be differentiated in the brainstem, which is located in the lower midbrain. The boundaries of brain to cerebrospinal fluid (CSF) are clearly defined, and those of bone to soft tissue are even better. General anatomy is

1249

1250

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 55.1 THE CRANIAL FORAMINA AND CANALS Site

From

To

Contents

Optic canal

Basisphenoid

Orbital apex

Middle cranial fossa

Optic nerve and 6 mm sheath; ophthalmic diameter artery 8 mm long

Optic canal view

1 mm difference in size suspicious; keyhold and figure of eight variants

Superior orbital fissure

Between greater & lesser wing of sphenoid

Orbital apex

Middle cranial fossa

Very variable III, IV, V1, VI; superior ophthalmic vein; middle meningeal artery branch

Occipitofrontal

Thin greater wing may stimulate erosion of lower border

Foramen rotundum

Greater wing of sphenoid

Middle cranial fossa

Pterygopalatine fossa

V2, artery of the foramen rotundum

3–4 mm diameter

Occipitofrontal

May be surrounded by extensive sphenoid sinus

Pterygoid (vidian) canal

Body of sphenoid, lateral to f. rotundum

Foramen lacerum

Pterygopalatine fossa

Vidian nerve and artery

Smaller than f. rotundum

Occipitofrontal

Foramen ovale

Greater wing of sphenoid, posteriorly

Middle cranial fossa

Infratemporal fossa

V3, accessory meningeal artery; veins

5 ⫻ 9.5 mm

Submentovertical

Frequently poorly seen on one or both sides. May be confluent with f. spinosum

Foramen spinosum

Greater wing of sphenoid, posterolateral to f. ovale

Middle cranial fossa

Infratemporal fossa

Middle meningeal artery

2.5–3 mm, rarely 5 mm

Submentovertical

May be double

Carotid canal

Petrous temporal

Skull base

Middle cranial fossa, above f. lacerum

Internal carotid artery and sympathetic plexus

6–9 mm diameter; 1.5 cm+ in length

Submentovertical

Runs posteromedial to Eustachian tube; rarely passes through middle ear; absent in aplasia of internal carotid artery

Internal auditory meatus

Petrous temporal

Posterior cranial Inner ear fossa

VII, VIII and dural 5–6 mm in sheath; internal height auditory artery

Perorbital

Height difference of 2 mm + is suspicious

Jugular foramen

Between petrous temporal and basiocciput

Posterior cranial Extracranial fossa jugular fossa

Pars nervosa; IX, 11 × 17 mm; inferior petrosal right often sinus; pars vascularis; larger X, XI, internal jugular vein and ascending pharyngeal and occipital artery branches

Under-tilted submentovertical

Pars nervosa and vascularis may be separate

Foramen magnum

Basiocciput

Posterior cranial Cervical spinal fossa canal

Medulla oblongata, 30 × 35 mm meninges and ligaments; XI (spinal root); vertebral and spinal arteries and veins

Lateral; submentovertical

Shape very variable

Hypoglossal (anterior condylar) canal

Occipital condyle

Foramen magnum

XII; branch of ascending pharyngeal artery

Reversed Stenvers: Stockholm ‘C’

Medial to jugular fossa

demonstrated in Figure 55.7. Anatomy of the skull base on CT with bone window settings is shown in Figure 55.8.

SPECIAL TECHNIQUES Computed tomography cisternography This test is rarely performed these days, as its only indication is to identify the site of CSF leaks before operative closure. An

Size

Best projection

Foramen/canal

5 mm diameter

Notes

important requirement is that the patient is actually leaking at the time of the examination, and many workers apply pressure to the neck by hand to temporarily occlude both jugular veins for 4 or 5 min before the study, to encourage active leakage. Water-soluble contrast medium (licensed for intrathecal use, e.g. iohexol) is instilled by lumbar puncture; a concentration of 240 mg ml−1 and of about 10 ml are usually more than adequate. Head-down tilting with the patient on their side ensures cranial penetration. Generally it is best to start

CHAPTER 55

Table 55.2

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

marks is shown in Figure 55.9. An axial series with the patient supine often is performed as well, especially if CSF leakage is profuse; if leakage is very profuse, rapid axial imaging in the supine plane may be required initially.

PHYSIOLOGICAL INTRACRANIAL CALCIFICATION

Pineal gland (60 per cent of adults) Habenular commissure (30 per cent) Choroid plexus (10 per cent) Dura mater falx cerebri (7 per cent) and superior sagittal sinus

Xenon computed tomography

Tentorium dural plaques (frequently parasagittal)

There are two fundamentally different methods on how CT can be used to assess cerebral blood perfusion. One uses the inhalation of xenon, the other a bolus injection of an iodinated contrast medium3,4. Xenon is a stable gas that has an atomic number close to iodine and therefore attenuates the X-ray beam in a similar fashion. Unlike iodine, xenon is freely diffusable and penetrates the blood–brain barrier. Current set-ups for xenon CT consist of the inhalation of a gas containing 28 per cent xenon during sequential acquisition of CT images over a period of approximately 6 min3. The distribution of xenon in the brain depends on the regional blood flow and is slightly quicker in grey matter than in white matter. The change of the Hounsfield numbers (CT numbers) over time

Petroclinoid (12 per cent) and interclinoid ligaments Diaphragm sellae Pituitary gland (rare)2 Carotid arteries (in elderly patients)

imaging in the direct coronal plane with the patient prone, as leaking is likely to be maximal in this position. Jugular vein compression is applied at this stage just before imaging if there is doubt about leaking.Thin sections (1–2 mm) on highresolution modes and smaller fields of view are made through the general area of suspected leakage (which often is already known). A normal CT cisternogram with its anatomical land-

mc

tp

sf

i tca ba

p

ac

cpd ba

pc

fm c l lc

lgb

aw

cq

csp 3

th

cs

v

ac

fh mc

pg

s

cp st t

ss

A

B

C

f

if f

c

s cv

bv

cs

f

D

E

Figure 55.7 (A–E) Normal contrast-enhanced CT anatomy. 3 = third ventricle, ac = anterior cerebral artery, ba = basilar artery, bv = body of lateral ventricle, c = caudate nucleus, cc = corpus callosum (genu), cp = choroids plexus, cpd = cerebral peduncle, cq = colliqulus, cs = centrum semiorale, csp = cave of septum pellucidum, cv = internal cerebral vein, f = falx, fh = frontal horn of lateral ventricle, fm = foramen of Monro, i = infundibulum of pituitary, ic = internal capsule, if = intracerebral fissure, l = lateral sulcus, lgb = lateral geniculate body, mc = middle cerebral artery, o = white matter tracts, p = pons, pc = posterior cerebral artery, pg = pineal gland, s = sulcus, sf = sylvian fissure, sp = septum between lateral ventricles, st = straight sinus, ss = sagital sinus, tca = terminal carotid artery, th = thalamus, tp = temporal horn, tr = trigone of lateral ventricle.

1251

1252

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Perfusion computed tomography The second technique of CT perfusion imaging tracks transient attenuation changes in the blood vessels and brain parenchyma during the first-pass passage of an intravenously injected contrast medium, similar to MR perfusion imaging (see later). A series of images is acquired at a predetermined level with a temporal resolution of one image every 1 or 2 s. The passage of the contrast-medium bolus causes a transient increase in Hounsfield units, which is proportional to the iodine concentration in the perfused tissue. Maps of cerebral blood volume (CBV), mean transit time (MTT) and cerebral blood flow (CBF) can be obtained from a pixel-by-pixel analysis of the density changes over time. Although absolute quantification of CBF is theoretically possible with this method, because of the linear relationship between iodine concentration and Hounsfield numbers, there remains some doubt about its accuracy in practice. Blood flow measurements of cortical grey matter using the bolus perfusion technique were systematically lower compared to data of xenon CT studies4.

vc

spf

fo fs

fl cc

jf

Figure 55.8 CT. Base of skull. Thin-section (1.5 mm) slice showing the important foramina at the skull base. cc = carotid canal, fo = foramen ovale, fl = foramen lacerum, fr = foramen rotundum, fs = foramen spinosum, fv = vidian canal, jf = jugular foramen, spf = sphenopalatine foramen.

MAGNETIC RESONANCE IMAGING Routine sequences and imaging protocols Routine brain magnetic resonance imaging A dedicated head coil is essential for most indications. This is performed in the axial plane, baseline as for CT. Section thicknesses of 4–5 mm are adequate throughout the head.A standard study usually consists of a sagittal scout multislice acquisition using a fast gradient echo sequence, with T1-weighted contrast, followed by an axial dual-echo multislice series providing balanced (proton density) contrast on the first echo and T2weighted contrast on the second (Fig. 55.10). Optimal timings,

during inhalation of xenon forms the basis of blood flow calculations, which are usually displayed as colour maps. The washout of xenon occurs relatively rapidly, allowing a repeat examination after 15–20 min. A disadvantage of this method is that any patient movement during the 6-minute period of imaging causes misregistration of data. Xenon uptake may also be impaired in patients with severe pulmonary disease.

oc olb

ca

ss vc

on fo

A

B

Figure 55.9 CT cisternography. Thin-section (1.5 mm) slices at the (A) level of the olfactory grooves and (B) foramen ovale. The intrathecal contrast outlines the subarachnoid space and extends into the optic nerve sheaths, outlining the optic nerves. ca = carotid artery, fo = foramen ovale, oc = optic chiasm, ob = olfactory bulb, on = optic nerve, ss = sphenoid sinus, vc = vidian canal.

CHAPTER 55

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Figure 55.10 Conventional MRI. (A) Fast spin-echo proton density (TR/TE = 3000/20). (B) Fast spin-echo T2-weighted (3000/90). (C) Spin-echo T1weighted images (400/14).

matrix sizes and fields of view vary between machines. The coronal plane is preferred for the dual-echo acquisition in patients with epilepsy and may be added if judged useful in other situations. A sagittal T2-weighted acquisition may aid detection of corpus callosum involvement in multiple sclerosis, which frequently is in the clinical differential diagnosis. Most units substitute a FLAIR (fluid-attenuated inversion recovery) sequence for the proton density-weighted acquisition in their routine examination5. Gradient-recalled susceptibilityweighted acquisitions can be useful on occasions to emphasize the presence of blood products in potentially haemorrhagic lesions. Strategies to reduce acquisition time at the expense of image quality may be necessary in restless and claustrophobic patients. Most modern equipment has the capability of fast and very fast acquisitions over a few seconds. Multislice modes utilize gradient echos, but single-slice fast spin-echo techniques can be a satisfactory alternative in some cases. Contraindications to MRI are discussed in Chapter 5.

Special areas The pituitary gland is studied at higher resolution using smaller fields of view, usually with T1-weighted contrast, with contiguous acquisitions in both sagittal and coronal planes. High-resolution modes and thin sections (2–3 mm) are desirable for examining the posterior fossa cisterns, middle and inner ears and the craniocervical junction using acquisitions emphasizing T2-dependent contrast.

Intravenous magnetic resonance contrast medium As with CT, many units use this routinely, but most are more selective. An unenhanced T1-weighted acquisition is essential first, as with CT. An IV injection of around 10 ml of one of the few MR contrast agents (e.g. gadolinium DTPA) currently licensed for intravascular use is made, imposing a far lower solute load than needed for CT. Similar precautions in case of adverse reactions should be taken, however. Indications for selective use are entirely similar to CT.

ANATOMY (Fig. 55.11) Grey white matter contrast, and contrast within and between different white matter regions, is far greater on MRI than on CT. A major contribution to this contrast is myelin density, increases in which generally lower signal intensity, though the effect is more complex with T1-weighted contrast where signal is increased in all but the most densely myelinated and

compacted tracts. Contrast discrimination within the brain and brainstem generally is greatest on mainly proton density-weighted images: T2-dependent contrast is present as well and signal-to-noise is significantly higher in the first than second echo of a dual-echo acquisition.

1253

1254

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

cs pof

cc

fl th

cf

ec

tha t

aca oc

sf

4

ba

fm

cc

pm th

3

tl

gp

pvs

a tl

p pg

A

cn

B

h

gf

C

Figure 55.11 Normal MRI. T2-weighted sagittal images through the midline (A). Coronal T2-weighted images through the hippocampi (B). Coronal T1-weighted images through the level of the third ventricle (C). 3,4 = third and fourth ventricle, a = amygdala, aca = anterior cerebral artery, ba = basilar artery, cc = corpus callosum, cf = calcarine fissure, ch = cerebellar hemisphere, cn = caudate nucleus, cs = central sulcus, ec = external capsule, fh = frontal horn, fl = frontal lobe, fm = foramen of Munro, gf = gyrus fusiformis, gp = globus pallidus, h = hippocampus, mca = middle cerebral artery, oc = optic chiasm, oh = occipital horn, p = pons, pg = parahippocampal gyrus, pm = putamen, pof = parieto-occipital fissure, pvs = perivascular spaces, sf = sylvian fissure, t = tectal plate, th = temporal horn, tha = thalamus, tl = temporal lobe.

ADVANCED MAGNETIC RESONANCE IMAGING MAGNETIC RESONANCE DIFFUSION IMAGING Diffusion-weighted MRI exploits the presence of random motion (Brownian motion) of water molecules to produce image contrast, thereby providing information not available on standard T1- or T2-weighted images. This is achieved by applying a pair of diffusion sensitizing gradients symmetrically around a 180 refocusing RF pulse of a T2-weighted MR sequence. Mobile molecules acquire phase shifts, which prevent their complete rephasing and result in signal loss. The loss of signal is proportional to the degree of microscopic motion that occurs during the pulse sequence6. On diffusion-weighted images, regions of relatively stationary water molecules appear much brighter than areas with a higher molecular diffusion. The degree of phase shift and signal loss depends also on the strength and duration of the diffusion sensitizing gradient, which is expressed by the ‘b-value’. B-values used for imaging of acute stroke lie typically around 1000 s mm−2. Quantitative analysis of the apparent diffusion coefficient (ADC) requires sequences with at least two different b-values and additional postprocessing. ADC maps are solely based on differences of tissue diffusion, independent of any T2 effects7. The ADC in the normal brain ranges from 2.94 × 10−3 mm2 s−1 for CSF to 0.22 × 10−3 mm2 s−1 for white matter; grey matter lies in between with a ADC of 0.76 × 10−3 mm2 s−1 8. Areas with a decreased ADC appear dark on ADC maps, which is the converse to diffusion-weighted images where areas of decreased diffusion appear bright7. A further feature of diffusion in the brain is its directional dependence, or anisotropy. This is particularly prominent in compacted white matter tracts, and least evident in grey matter. Diffusion tension imaging (DTI) explores anisotropy from six to nine directions and is capable of generating notionally invariant values, which characterize anisotropy on a pixel-by-pixel basis.

Magnetic resonance perfusion imaging MR perfusion imaging exploits magnetic susceptibility effects within the brain tissue during the first pass of an intravenously injected gadolinium-based contrast agent. During its first pass through the brain, the contrast medium causes a transient signal drop on T2*-weighted (susceptibility-weighted) MRI8 (see Chapter 8 for images). Images are typically acquired with a temporal resolution of one image every 1–2 s, similar to CT perfusion. The use of singleshot echoplanar imaging (EPI) however, allows multislice imaging with full-brain coverage. MR perfusion imaging is, however, at present only semiquantitative and cannot provide absolute values9. A newer MR perfusion technique, arterial spin labelling, that does not require exogenous contrast medium, is currently being developed particularly at higher field strengths, but is not yet robust enough for clinical use10. The sequential changes in signal intensity can be plotted as a time–signal intensity curve of a chosen region of interest or reproduced as pixel-based colour maps. Summary parameters are then generated by the manufacturer’s software. The relative cerebral blood volume (rCBV) is proportional to the area under the curve on the time–signal intensity graph. Other measurements that can be derived are arrival time (T0), time to peak (Tp) and mean transit time (MTT) of the gadolinium bolus. Using tracer kinetics, the rCBF can be estimated by dividing the relative blood volume by the mean transit time (rCBF = rCBV divided by MTT). In the absence of absolute quantification of the CBF, comparison with the contralateral hemisphere provides the easiest way to analyse MR perfusion images. This becomes, however, problematic if the perfusion of the contralateral hemisphere is not normal, as in the presence of bilateral carotid artery disease11.

CHAPTER 55

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Magnetic resonance spectroscopy Magnetic resonance spectroscopy (MRS) is a noninvasive in vivo method that allows the investigation of biochemical changes in both animals and man. Histochemical and cell culture studies have shown that specific cell types or structures have metabolites that give rise to 1H-MRS peaks. A change in the resonance intensity of these marker compounds may reflect loss or damage to a specific cell type. The acquisition of long echo time data (TE = 270 ms, TR = 3 ms) allows the detection of N-acetylaspartate (NAA), creatine (Cr/PCr) and choline (Cho) in normal brain, and lactate in areas of abnormality. The methyl resonance of NAA produces a large sharp peak at 2.01 p.p.m. and acts as a neuronal marker as it is almost exclusively found in neurons in the human brain, where it is found predominantly in the axons and nerve processes12. The creatine peak (3.03 p. p.m.) arises from both phosphocreatine- and creatine-containing substances in the cell and choline (3.22 p.p.m) is thought to arise from choline-containing substances in the cell membrane. The acquisition of short echo time data (TE = 30 ms, TR = 2 s) has become the standard spectroscopy sequence and has the advantage of reduced effects from T2 losses and therefore provides spectra with better signal to noise. In addition, it detects additional resonances from metabolites with complex MR spectra such as myo-inositol, glutamate and glutamine (Chapter 8). Whilst providing more information, the broad background signal consisting of low concentration metabolites, and macromolecules and lipids, increases the difficulty of peak area estimation13.

FUNCTIONAL IMAGING TECHNIQUES A variety of different techniques are available for functional brain imaging. Those most widely used are SPECT, PET and functional magnetic resonance imaging (fMRI).

Single-photon emission computed tomography SPECT images are formed from detection of gamma rays emitted during radionuclide decay as part of a nuclear medicine examination. Gamma rays or photons are detected by a gamma camera which, if rotated about the patient’s head, allows reconstruction of tomographic slices of distribution of activity in that part of the patient. Radionuclide imaging of the brain requires radiopharmaceuticals that cross the blood–brain barrier. SPECT may be used to produce images of a rCBF using the radiopharmaceuticals 133Xe, 123 I isopropyl iodoamphetamine (IMP), 99mTc ethyl cysteinate dimer (ECD) or 99mTc hexamethylpropylene amine oxide (HMPAO) (Fig. 55.12). Clinical applications include dementia, cerebrovascular disease, epilepsy, encephalitis and head injury. SPECT can also be used to image uptake at neurotransmitter receptors using various radiopharmaceuticals usually labelled with 123I. Many different SPECT radiopharmaceuticals are taken up into intracranial tumours,

Figure 55.12 SPECT. Normal 99mTc HMPAO SPECT of the brain, axial (L) and sagittal (R) images.

including 201Tl chloride, 99mTc MIBI, 123I α-methyl tyrosine and 111In octreotide. Because of the requirement for lead collimation, SPECT has inherently poorer resolution than PET and absolute quantitation is not possible. However, in its favour, SPECT is available in most nuclear medicine departments, is relatively inexpensive and has good patient acceptability.

Positron emission tomography Positron emission tomography (PET), like SPECT, produces tomographic images. Positron-emitting radioisotopes decay by emission of positrons, or positively charged electrons. These quickly combine with an adjacent electron in an annihilation reaction with the emission of two high-energy gamma rays in opposing directions. Detection of these simultaneously emitted photons allows calculation of their site of origin and, therefore, a map of radiopharmaceutical distribution in the patient. PET can be used to study different physiological processes in the brain. A cyclotron is required to generate positron-emitting isotopes that can be made from a variety of biologically interesting compounds, such as 18F, 11C, 13N and 15O. Physiological parameters can be derived, for example, cerebral glucose uptake, using 1,8fluorodeoxyglucose (FDG), oxygen metabolism, using 15O2 or 11CO and rCBF using H215O. PET rCBF studies can be used to study cortical activation during ‘brain tasks’ such as finger tapping or visual stimulation. The area of cerebral cortex responsible for a particular function demonstrates slightly increased rCBF during activation. A number of radiopharmaceuticals are available for PET receptor imaging. FDG, 11C methionine and 1,8Fα-methyl tyrosine are used for tumour imaging. Disadvantages of PET compared with SPECT are its limited availability and high cost due to the necessity of a cyclotron close to the PET unit. PET has the advantage over SPECT and fMRI of enabling absolute quantitation of, for example, CBF, provided arterial blood sampling is performed. Compared with fMRI the main disadvantages of PET are limited availability and ionizing radiation, limiting the number of repeat studies in any one patient.

1255

1256

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

FUNCTIONAL MAGNETIC RESONANCE IMAGING Functional MRI techniques can also be used to study cortical activation. The most commonly applied technique is measurement of a tiny increase in signal intensity on T2*weighted acquisitions in the relevant cortex during neuronal activation. This occurs as a result of the magnetic susceptibility effects of oxyhaemoglobin. Oxyhaemoglobin is diamagnetic while deoxyhaemoglobin is paramagnetic. During cortical activation there is an increase in rCBF and thus an increase in oxygen delivery to the activated brain, which exceeds the local oxygen metabolic requirement. There is, therefore, a net increase in oxyhaemoglobin concentration in the venules and veins in the vicinity of the activated brain, which results in a tiny increase in MR signal, the so called blood oxygenation level dependent or, BOLD effect. The magnitude of this MR signal change is field dependent, being greater at higher field strengths. Nevertheless, it

is a tiny signal and quantitative comparison must be made between the MR signal during the resting state and the activation state during multiple repetition in order to detect activation. Major advantages of fMRI over PET are the lack of ionizing radiation and higher temporal resolution. However, due to the haemodynamic response time, fMRI will never approach the time resolution of electrophysiological methods such as EEG. Another significant limitation of fMRI is that the magnitude of the MR signal change is not directly proportional to rCBF change and therefore absolute quantification is not possible. For activation purposes this is not an important limitation as it is relative changes during activation tasks that are important. Although fMRI is being increasingly used for brain mapping, the technique has limited clinical applications and is used primarily for the identification of eloquent cortex, particularly the motor strip, prior to surgery in patients with structural lesions and arteriovenous malformations14.

VASCULAR IMAGING TECHNIQUES Intra-arterial catheter angiography has been the mainstay for the investigation of neurovascular diseases in the past. Its role is changing with continuous advances in noninvasive vascular imaging, which include Doppler sonography, magnetic resonance angiography and CT angiography. These noninvasive techniques have replaced intra-arterial angiography for a number of diagnostic indications. This section discusses the techniques of vascular imaging (invasive and noninvasive) before giving a brief overview of vascular neuroanatomy, mostly illustrated with intra-arterial angiograms but equally applicable to the noninvasive techniques.

Conventional intra-arterial angiography techniques of catheterization15,16 General principles and basic arteriographic techniques are described elsewhere. Most diagnostic cerebral angiography can be performed under local anaesthesia. General anaesthesia is indicated in very anxious or restless patients and if interventional endovascular procedures are planned to follow the diagnostic study. The transfemoral route is now almost exclusively used for catheterization of the cerebral vessels and puncture of the axillary artery or direct puncture of the carotid artery are only rarely performed. Insertion of a femoral sheath is not necessary for straightforward cases, but may be useful in more complex cases where a change of catheter during the procedure is anticipated, and is mandatory for interventional procedures. A variety of catheters are available for catheterization of the carotid and vertebral arteries and the choice of catheter is largely a personal one. The most frequently used catheters are 4F or

5F with a tapered J-shaped tip; some neuroradiologists prefer more complex shapes such as a Mani catheter. In elderly or hypertensive patients, the aortic arch and great vessels are often ectatic and tortuous. In such cases it is best to use a reverse curve catheter, such as a Sidewinder catheter. The use of hydrophilic guidewires greatly facilitates catheterization of the cerebral vessels. It is important to choose a guidewire of the appropriate size, occupying the lumen of the catheter.The choice of a guide that is too small facilitates reflux of blood into the catheter, which can clot and be the source of embolic complications. For catheterization of the carotid and vertebral arteries the surgeon should always lead with a soft-tipped guidewire and advance the catheter over the wire, in order to avoid trauma to the intima. Most intracranial abnormalities require selective internal carotid and/or vertebral artery injections, depending on the clinical problem. Common carotid artery injection can, however, be performed in elderly patients and those with significant atheromatous disease at the carotid bifurcation. If the latter is suspected, the carotid bifurcation should be visualized under fluoroscopy or with an angiographic run, before advancing the guidewire into the internal carotid artery. The left vertebral artery is larger or of equal size than the right vertebral artery in approximately 75 per cent and therefore represents the first-line approach to angiography of the posterior circulation. Should the left vertebral artery appear to be absent, it probably arises from the arch of the aorta between the left common carotid and subclavian arteries. Very rarely neither vertebral artery can be catheterized, and under these circumstances the subclavian artery can be injected during inflation of a blood pressure cuff, which reduces flow of contrast medium and blood down the arm.

CHAPTER 55

External carotid artery catheterization is necessary for head and neck lesions in intracranial lesions with a potential meningeal supply (such as meningiomas or cerebral arteriovenous malformations [AVMs] and dural fistulae). The origin of the external carotid artery lies anterior and medial to that of the internal carotid artery in the majority of cases. Once the catheter has been positioned in the appropriate vessel, a double-flush technique, withdrawing blood into one syringe and saline flushing from another, is used to minimize the risks of embolism. Non-ionic low-osmolality contrast media is recommended for cerebral angiography. For a modern digital angiography system, using a concentration of 150 mg I ml−1 is sufficient in the case of internal carotid or vertebral artery injections. A higher concentration of contrast medium (up to 300 mg I ml−1) may be necessary for common carotid artery injections and highflow lesions, such as large AVMs. Injection of contrast medium into the external carotid artery is uncomfortable in a number of patients. If the procedure is performed under a local anaesthetic, it is best to warn the patient of a hot feeling in the face and ‘funny taste’ in the mouth. Similarly, patients should be warned prior to a vertebral artery injection that they might experience flashing lights in the eyes. Contrast medium can be injected manually or with an automatic pump (6–8 ml of contrast medium at a rate of 3–4 ml s−1 for internal carotid and vertebral artery injections when using digital subtraction angiography [DSA]); less forceful and lowervolume injections are needed in the external carotid artery and its branches.

Technique of image acquisition Nowadays, cerebral angiography is carried out on a DSA system. Modern digital systems provide an image resolution of 1024 × 1024 pixels and good-quality digital fluoroscopy with additional features such as ‘road mapping’, which is useful for superselective catheterization with microcatheters during interventional procedures. Standard radiographic projections for carotid angiography include a lateral projection, centred on the pituitary fossa, and an AP view with the petrous ridge projected approximately over the roof of the orbit. Ipsilateral anterior oblique views are routinely performed for the investigation of aneurysms in subarachnoid haemorrhage and a number of additional views such as an occipitomental or periorbital view may be necessary. Three standard projections are employed for the vertebral angiogram: lateral, half-axial (Towne’s), and AP, with the petrous ridge superimposed on the lower border of the orbit. A biplane angiography unit is of major advantage in neuroangiography. It allows simultaneous acquisition of two projections (such as AP and lateral or two oblique views) during a single injection. This reduces the number of contrast medium injections and thereby the risk of catheter-related complications. Two major technical advances in digital angiography are rapid-frame-rate acquisition (up to 30 images s−1) and threedimensional (3D) rotational angiography. Routine cerebral angiography is carried out with a frame rate of 2 or 3 images s−1 for the arterial phase and 1–2 images s−1 for the venous phase.

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

The investigation of AVMs benefits from a higher frame rate in the arterial phase, in the order of 6 images s−1. Occasionally a higher frame rate (15 frames s−1) can be used to analyze the haemodynamics in high flow lesions or certain types of aneurysms17. 3D rotational angiography involves image acquisition with a precisely calibrated rotating X-ray tube (describing an 180-degree arc) before and during pump injection of contrast medium. This allows the acquisition of volumetric datasets, which are post-processed on a computer workstation. Following removal of the bony structures, high-resolution images of the cerebral vessels can be manipulated in different ways and viewed from any angle. This obviates the need for multiple angiographic runs and has proved very useful in the planning of endovascular treatment of aneurysms, providing a 3D view of the aneurysm morphology and its neighbouring vessels18.

Indications Owing to the advent of new noninvasive imaging techniques, particularly computed tomographic angiography (CTA), the indications for intra-arterial cerebral angiography are changing. In departments where CTA is routinely used, cerebral angiography is used to resolve discrepancies between two noninvasive methods and as an integral part of endovascular interventional procedures. It may be used for the investigation of aneurysms in subarachnoid haemorrhage, cerebral AVMs and carotid artery disease to confirm a significant stenosis suspected on noninvasive imaging. Preoperative angiography is sometimes performed in glomus jugulare tumours and meningiomas to assess tumour vascularity, and is frequently combined with preoperative embolization in very vascular tumours19.

Contraindications There are very few absolute contraindications to cerebral angiography, but since it is not without risk, it should never be carried out if it is clear that the results will not influence management. A well-documented history of untoward reactions to contrast media is a relative contraindication. Intra-arterial angiography is now increasingly used in the context of thrombolytic treatment of acute stroke20. Treatment with anticoagulant drugs does not contraindicate arteriography, provided the prothrombin level is within the normal therapeutic range.

Complications Local and general complications of arteriography are discussed in Chapter 6. Specific risks of catheterization of the aortic arch or cervical arteries include cerebral embolism and damage to the arteries by the catheter or guidewire, which include spasm, thrombosis and dissection. A previous study showed that the stroke risk of arch aortography and DSA is similar to that of selective injections, which suggests that the risk is mainly due to embolism from catheter or guidewires21. Reported risks of cerebral angiography in studies published over the last 15 years vary from 0.5–2.3 per cent22–24. The North American Asymptomatic Carotid Stenosis trial (ACAS) identified the risk of significant disabling stroke as being 1.2 per cent,

1257

1258

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

which was similar to the stroke risk of carotid endarterectomy in asymptomatic patients25. A recent, retrospective study of 454 patients mainly investigated for suspected aneurysms or AVMs (with one-third

of the studies performed in the acute stage of intracranial haemorrhage), showed an overall neurological complication rate of 2.3 per cent with persistent neurological deficits in 0.4 per cent24.

COMPUTED TOMOGRAPHY ANGIOGRAPHY (Fig. 55.13) Selective imaging of blood vessels with CT has become possible with the introduction of spiral CT systems26. The technical principles of spiral CT data acquisition are explained in Chapter 4. For CTA, a volumetric dataset is acquired during the vascular phase of an iodinated contrast, which is injected intravenously. The data acquisition time and amount of contrast medium used depend on the area to be covered and choice of slice collimation and table speed. Earlier spiral CT systems offered only a limited area coverage (such as the carotid bifurcation or circle of Willis), due to X-ray tube heating limits. The quality of CTA has dramatically improved with the introduction of multidetector CT27 with improved area coverage and spatial resolution. Modern multidetector-CT machines are able to cover an area from the carotid bifurcation up to the circle of

Figure 55.13 CTA 3D volume-rendered CTA. The arterial anatomy is well visualized. Filling of the internal cerebral venous and venous sinuses is also seen.

Willis or the entire intracranial circulation from the skull base to the vertex with a single data acquisition. Timing of data acquisition in relation to the administration of contrast medium is critical for maximum arterial opacification. The operator can either set a standard delay or use an automating bolus detection system, which is available on most machines. Recommended standard delays between the start of the injection and image acquisition are 12 s for examination of the extracranial and 15 s for examination of the intracranial vessels28. Automated bolus detection is, however, more satisfactory because it adjusts for individual variations in circulation time. Injection rates and volumes vary with the cannula size, number of slices and cardiac output. Typical volumes of 100– 120 ml contrast medium are given at a rate of 3–5 ml s−1. The quality of CT angiograms depends heavily on postprocessing of the image data. Enhanced blood vessels are extracted from a 3D dataset by applying specific density thresholds.The vessels can then be displayed as 2D projectional images, which resemble conventional angiograms, or as 3D surface-rendered structures. Separation of vessels running close to bone (near the skull base and cranial vault may be difficult.These difficulties can be at least partially resolved by using thick-section multiplanar reformats (which can be angled in such a way to exclude bone) and by interactive viewing of the source data29. CTA has been used successfully for the evaluation of carotid artery stenosis30,31, carotid artery dissection32, intracranial vascular occlusion33 and intracranial aneurysms34,35. CTA for the assessment of cerebral AVM is subject to ongoing research; their complex structure needs more sophisticated post-processing and interactive viewing36. CTA can also be used to examine the cerebral venous system (CT venography), its main application being suspected dural sinus thrombosis37. The advantages of CTA over MRA are that it can be used in claustrophobic patients and patients with cardiac pacemakers, or other implants that preclude MR data acquisition. Its disadvantages are the use of ionizing radiation and iodinated contrast media.

MAGNETIC RESONANCE ANGIOGRAPHY MRA has progressed rapidly in the last decade. Its basic principles are outlined in Chapter 6.Traditionally, MRA has been performed without administration of an exogenous contrast medium, relying on inflow of unsaturated spin (time-of-flight [TOF] MRA) or accumulation of phase shifts proportional to the flow velocity (phase contrast MRA)38. More recently,

MRA has been used in conjunction with the injection of gadolinium-based contrast media (contrast-enhanced MRA)39. Both non-enhanced and enhanced MRA techniques benefited from the development of high-performance gradients allowing shorter repetition (TR) and echo (TE) times as well as software developments such as zero-filled interpolation

CHAPTER 55

processing. Both TOF and phase contrast techniques can be performed with a 2D or 3D data acquisition40. 2D-TOF or 3D-TOF MRA of the neck vessels have limitations in areas of turbulent or slow flow, which may remain undetected due to intravoxel dephasing41,42.This can simulate the presence or exaggerate the degree of carotid artery stenoses. For imaging of the intracerebral vessels, 3D-TOF MRA is the technique of choice43. A single slab 3D-TOF aquisition is adequate for imaging the circle of Willis, but coverage of a larger part of the intracranial circulation requires three or four multiple overlapping slabs (MOTSA technique) (Fig. 55.14). Data are usually displayed as maximum intensity projections, but inspection of the source data should always be performed to resolve difficult cases or to confirm the suspicion of an artefact43. Intracranial 3D-TOF MRA has been successfully used for the detection of aneurysms (with a high sensitivity for aneurysms larger than 5 mm)44–46, assessment of intracranial stenosis and, to a limited extent, AVMs47. Phase contrast MRA is based on the detection of phase shifts generated by a flow-encoding gradient.The phase shift is proportional to the velocity of blood, and care must be taken to choose an appropriate ‘velocity window’ depending on the area studied. Typical velocity parameters are 10–20 cm s-1 for dural sinuses and 50–60 cm s−1 for major cerebral arteries. Although generally inferior to 3D MRA, it is more sensitive for detection of slow flow (with the appropriate velocity encoding) and can therefore be used for imaging cerebral veins48. It does not suffer from T1contamination artefact and it can provide information about the direction of blood flow. 3D phase contrast MRA can also be used to show the direction of collateral blood flow across the circle of Willis (Fig. 55.15) with significant differences between normal volunteers and patients with carotid artery occlusion49. The administration of a gadolinium-based contrast medium has been shown to have some benefit in conjunction with intracranial 3D-TOF MRA for conditions such as AVMs and intracranial stenoses50,51. The acquisition of 3D gradient-echo

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Figure 55.15 Phase contrast MRA of circle of Willis. Phase contrast MRA allows encoding of the flow direction. Here the flow-encoding direction is right (R) to left (L): flow in this direction appears white, whereas flow in the opposite direction (towards the right) appears black.

images, during the first pass of a contrast medium bolus, represents a relatively recent development. Demonstration of vascular structures with this technique depends on the T1 shortening of blood and not on inflow or phase-shift effects39 (Fig. 55.16). Contrast-enhanced MRA of the carotid arteries does not suffer from artificial signal loss due to turbulent and slow flow (like the TOF techniques) and initial results show that it compares favourably with intra-arterial angiography52. Applications of this method for imaging intracranial vessels are emerging53.

Doppler ultrasound Technical principles and Doppler US and its use in the assessment of the carotid artery stenosis are discussed in Chapter 3.

ACCM A1

M2

CS M1

PCOM

BA P2 P1

petr CA

Figure 55.14 3D TOF MRA of the intracranial circulation: axially collapsed maximum intensity projection. A1 = precommunicating segment of anterior cerebral artery, ACOM = anterior communicating artery, BA = basilar artery, CS = carotid siphon, M1 = first (horizontal) segment of middle cerebral artery, M2 = M2 segments of middle cerebral artery, P1 = precommunicating segment of posterior cerebral artery, P2 = P2 segment of posterior cerebral artery, PCOM = posterior communicating artery, petr CA = petrous segment of ICA.

Figure 55.16 Contrast-enhanced MRA of aortic arch. A 3D gradientecho sequence has been acquired during the first pass of an intravenously injected gadolinium bolus. It shows the origins of the great vessels. Note also that there is background opacification of the pulmonary vessels.

1259

1260

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

ANATOMY OF CEREBRAL ARTERIES AND VEINS16 INTRACRANIAL ARTERIES The internal carotid arteries supply the anterior cerebral circulation and the vertebral and basilar arteries supply the posterior circulation. The external carotid arteries supply most extracranial head and neck structures (except the orbits) and make an important contribution to the supply of the meninges. There are numerous anastomoses between the external carotid arteries and the anterior and posterior circulation.

Anterior circulation The right common carotid artery is the first main branch of the innominate or brachiocephalic artery and the left common carotid artery is the second main branch of the aortic arch (Fig. 55.17). Each common carotid artery runs within a fascial plane, the carotid sheath, lateral to the vertebral column, and bifurcates between the level of the third and fifth cervical vertebrae into external and internal carotid arteries. At the bifurcation, the internal carotid lies usually posterior and lateral to the external carotid artery.

Internal carotid artery16,54,55 The internal carotid artery can be divided into a number of segments between the carotid bulb and its bifurcation into

Figure 55.17 Arch aortogram, left anterior oblique projection, arterial phase, 1 = arch of aorta, 2 = innominate (brachiocephalic) artery, 3 = right subclavian artery, 4 = right vertebral artery, 5 = right common carotid artery, 6 = right internal carotid artery, 7 = right external carotid artery, 8 = left common carotid artery, 9 = left external carotid artery, 10 = left subclavian artery, 11 = left vertebral artery.

middle and anterior cerebral arteries. Unfortunately there are several classification systems, some numbering the segments with the direction of blood flow and others against it54. Until a consensus is reached, it is best to use anatomical names for the internal carotid artery segments. A simplified anatomical division distinguishes between cervical, petrous, cavernous and supraclinoid segments of the internal carotid artery56. No named branches arise from the cervical segment of the internal carotid artery. The petrous segment is intraosseous. It begins at the carotid canal where the artery enters the skull base. The internal carotid artery then runs forward and medially in the foramen lacerum and lies extradurally until it reaches the petrolingual ligament; after this it becomes the cavernous segment, which gives off several important branches. After leaving the cavernous sinus, it pierces the dura and enters the subarachnoid space at the level of the anterior clinoid process, after which it becomes the supraclinoid segment. The carotid siphon is formed by the cavernous and supraclinoid segments. The supraclinoid segment terminates, dividing into the anterior and middle cerebral arteries (Fig. 55.18). The principal branches of the cavernous segment are: 1 The posterior trunk (meningohypophyseal artery) arises posteriorly from the superior aspect of the first bend of the cavernous segment and gives off branches to the pituitary gland and tentorium cerebelli (marginal tentorial artery). 2 The inferolateral trunk arises more anteriorly and laterally from the horizontal portion of the cavernous segment. It supplies the third, fourth and fifth cranial nerves and has important anastomoses with the external carotid system: with the maxillary artery through the foramen rotundum and ovale and with the middle meningeal artery through the foramen spinosum.

Figure 55.18 Internal carotid arteriogram; lateral projection, arterial phase. 1 = cervical portion of internal carotid artery, 2 = petrous portion, 3 = cavernous portion (siphon), 4 = ophthalmic artery, 5 = choroidal (ophthalmic) crescent, 6 = anterior choroidal artery, 7 = anterior cerebral artery, 8 = pericallosal artery, 9 = callosomarginal artery, 10 = middle cerebral artery branches.

CHAPTER 55

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

The principal branches of the supraclinoid segment are: 1 The ophthalmic artery is usually given off just after the carotid artery leaves the cavernous sinus, but its origin is variable and it can sometimes arise from the middle meningeal artery. 2 The posterior communicating artery arises posteriorly from the distal loop of the siphon, to link the internal carotid artery with the posterior cerebral artery, which arises normally from the posterior (vertebral) circulation. It is extremely variable in size and, when small, inconstantly opacified by carotid injection. The posterior cerebral artery can arise directly from the internal carotid artery, which is called a ‘fetal arrangement’. 3 The anterior choroidal artery arises just distal to the posterior communicating artery and runs posterosuperiorly and laterally. It is an important artery that supplies the posterior limb of internal capsule, cerebral peduncle and optic tract, medial temporal lobe and choroid plexus.

Anterior cerebral artery16,57–60 The anterior cerebral artery is divided into three anatomical segments: the horizontal or precommunicating segment (A1), vertical or postcommunicating segment (A2), and distal ACA including cortical branches (A3). The A1 segment runs beneath the frontal lobe of the brain and courses over the optic nerves and chiasm to the midline, where it is joined to the contralateral A1 segment by the anterior communicating artery. The A1 segment gives rise to a variable number of perforating branches, the medial lenticulostriate arteries. The recurrent artery of Huebner is the largest of the perforating branches and may arise from the A1 or A2 segment. It derives its name from the fact that it doubles back on its parent artery at an acute angle to join the lenticulostriate vessels. A common anatomical variation is hypoplasia or aplasia of the A1 segment, in which case the distal segments fill preferentially from the other side via the anterior communicating artery. Other variations include a fusion of the A2 segment in the midline to give a single ‘azygos’ anterior cerebral artery, which then supplies both hemispheres. The A2 segment turns upwards, gives off a frontopolar branch and divides, at the level of the genu of the corpus callosum, into the callosomarginal and pericallosal arteries, which are the A3 segments. The former lies in the cingulate sulcus and the latter course along the posterior aspect of the corpus callosum, which it supplies. Cortical branches of the callosomarginal artery supply the medial frontal lobe (frequently as far back as the Rolandic fissure), whereas cortical branches of the pericallosal artery supply the medial parietal lobe.

Middle cerebral artery16,59,60 The middle cerebral artery is divided into four anatomical segments: the horizontal segment (M1), insular segment (M2), opercular segment (M3) and cortical branches (M4 segments). Medial and lateral lenticulostriate arteries are perforating branches that arise from the M1 segment (Fig. 55.19); they supply the basal ganglia and capsular region.

Figure 55.19 Internal carotid arteriogram: AP projection, arterial phase. 1 = petrous segment of internal carotid artery, 2 = cavernous portion, 3 = supraclinoid (subarachnoid) portion, 4 = anterior cerebral artery precommunicating segment, lying above the pituitary fossa, 5 = pericallosal and callosomarginal arteries, superimposed, lying in the midline, 6 = anterior choroidal artery, 7 = lenticulostriate artery, 8 = major divisions of the middle cerebral artery, 9 = cortical branches, which extend to the cranial vault.

The M1 segment runs in the Sylvian fissure and gives off an anterior temporal artery of variable size before dividing into two or three main trunks (M2 segments). Its branches run over the frontoparietal and temporal opercula (M3 segments). The characteristic loops formed by the upward and downward course of the insular and opercular segments form a straight line on the lateral projection, which represents the upper border of the ‘Sylvian triangle’, its inferior border being formed by main middle cerebral artery trunk. The ‘Sylvian point’ is the highest and most medial point where the angular artery turns inferolaterally to exit the Sylvian fissure. Displacement of these landmarks has been used in the past to locate cerebral mass lesions; they are now largely of historical interest. The cortical branches (M4 segments) of the middle cerebral artery are variable and complex, but temporal, ascending frontoparietal, parietal, angular and posterior temporal branches can usually be identified (Figs 55.18, 55.19) and supply most of the lateral surface of the cerebral hemisphere, excluding a narrow superomedial strip supplied by the anterior and posterior cerebral arteries.

1261

1262

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Posterior circulation15,16,59 (Figs 55.20, 55.21) Vertebro-basilar system The right and left vertebral arteries usually arise as the first branches of the corresponding subclavian arteries. Each then enters the foramen transversarium of the sixth cervical vertebra, and runs directly upwards in the bony vertebral canal formed by these foramina before arching laterally then medially around the anterior arch of the atlas behind the lateral mass of the atlas to pierce the dura mater and enter the subarachnoid space at the level of the foramen magnum, subsequently fusing with its fellow behind the clivus and in front of the lower pons, to give rise to the midline basilar artery. The vertebral arteries give muscular branches, which frequently anastomose with those of the ascending pharyngeal and occipital arteries, and they commonly furnish important feeding vessels to the cervical spinal cord. One of the vertebral arteries often gives off the posterior meningeal artery, which passes upwards through the foramen magnum to run posteriorly in the midline on the dura mater of the occipital bone. Soon after entering the cranial cavity, each vertebral artery gives off a posterior inferior cerebellar artery, which runs around the medulla oblongata, looping under the olive, to lie near its fellow in the midline behind the medulla,

Figure 55.21 Vertebral angiogram. Half-axial projection, arterial phase. (For Key, see Fig. 55.20 caption.)

Figure 55.20 Vertebral angiogram. Lateral projection, arterial phase. Key for Figures 55.20 and 55.21: 1 = vertebral artery, 2 = posterior inferior cerebellar artery, 3 = inferior vermian branch, 4 = basilar artery, 5 = anterior inferior cerebellar artery, 6 = superior cerebellar artery (duplicated on left), 7 = posterior cerebral arteries, 8 = posterior temporal branches, 9 = internal occipital and calcarine branches, 10 = posterior choroidal arteries, 11 = thalamoperforating arteries, 12 = filling of middle cerebral arterial branches via posterior communicating artery.

before running posteriorly above the cerebellar tonsil, where it lies close to the roof of the fourth ventricle, and continuing on the undersurface of the cerebellum as the inferior vermian artery. The posterior inferior cerebellar artery also gives off tonsillar and hemispheric branches. The vertebral arteries are commonly unequal in size; when this is the case, the left is usually the larger, but the right is of greater calibre in about one-fifth of cases. When one of the arteries is very small, it frequently supplies only the ipsilateral posterior inferior cerebellar artery territory, which is called a PICA termination of the vertebral artery. The basilar artery runs superiorly on the anterior surface of the pons and gives off anterior inferior cerebellar, superior cerebellar and posterior cerebral arteries on both sides. Terminating just above the tip of the dorsum sellae, it generally shows a slight anterior convexity and deviates from the midline following the curve of the dominant vertebral artery; its form is sufficiently variable, especially in the elderly, to render assessment of lateral or posterior displacement difficult. The anterior inferior cerebellar arteries arise close to the origin of the basilar artery and run laterally on the pons and anteroinferior surface of the cerebellum. They loop in the cerebellopontine angle, and supply the surrounding structures, their branches including the internal auditory arteries, to the nerves in the internal auditory meatus. The cerebellar branches anastomose with those of the posterior inferior cerebellar artery and there is frequently a reciprocal relationship:

CHAPTER 55

if the posterior artery is small on one side, the corresponding anterior artery is larger, branching more extensively, and vice versa. The superior cerebellar arteries arise several millimetres below the posterior cerebral arteries that are the terminal branches of the basilar artery, from which they are separated by the tentorium cerebelli. The superior cerebellar arteries are frequently duplicated, in which case the individual vessels are smaller. They pass around the brainstem to fan out over the superior surface of the cerebellar hemispheres, while their main trunks run back over the superior vermis, giving a precentral branch that passes down between the roof of the fourth ventricle and the central lobule of the cerebellum.

POSTERIOR CEREBRAL ARTERIES The bifurcation of the basilar arteries can appear either Vshaped (caudal fusion of the posterior cerebral arteries) or Tshaped (cranial fusion of the cerebral arteries). It can also be asymmetrical with a caudal fusion on one side and a cranial fusion on the other. Basilar tip aneurysms are much more frequently associated with a caudal fusion than with a cranial fusion of the posterior cerebral arteries61. After bifurcating, the basilar artery gives rise to the two posterior cerebral arteries, each of which has four segments. P1 is the precommunicating segment before which it joins with the posterior communicating arteries to become the P2 or ‘ambient’ segment and P3 or ‘quadrigeminal’ segment, named after the basal cistern in which it runs. The P4 segment is the terminal segment of the posterior cerebral artery, which includes the occipital and inferior temporal branches. There is reciprocity in calibre of the precommunicating (P1) segments of the posterior cerebral arteries and the posterior communicating arteries: if the

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

latter are large and the main source and give rise to a so-called fetal origin of the posterior cerebral artery, the P1 segments may be so small that little or no distal filling is seen on vertebral arteriography. The appearances are commonly asymmetrical. The posterior communicating arteries give off the anterior thalamoperforating arteries, and the P1 segment the posterior thalamoperforating and thalamogeniculate arteries, which pass posterosuperiorly into the interpeduncular fossa to enter the posterior perforated substance. The medial posterior choroidal artery arises from the P2 segment and passes around the midbrain, then superiorly, over the pulvinar of the thalamus to reach the third ventricle. Two or more lateral posterior choroidal arteries arise also from the P2 segment and follow a similar course, but lie more posteriorly on the lateral view. Cortical branches arise from the P2 segment (anterior and posterior temporal arteries) and form the P4 segment, which divides into a group of the inferior temporal arteries supplying a considerable portion of the inferior surface of the temporal lobe and the parieto-occipital and calcarine branches, supplying the medial surface of the occipital lobe, including the visual cortex.

External carotid artery15,16,62,63 The major branches of the external carotid artery are shown in Figure 55.22; in general they are named simply for their territory of supply. They are best examined using the lateral projection. The first, anterior branch, the superior thyroid artery, may arise from the terminal common carotid artery. The lingual and facial arteries also arise anteriorly, sometimes from a common trunk, and run forwards, the former deep to and the latter lateral to the mandible. In addition to the structures from which they take their names, they also supply the salivary glands. The ascending pharyngeal artery (Fig. 55.23) runs vertically upwards (often obscured on common carotid

Figure 55.22 External carotid artery. Proximal injection. Radiograph (A) and diagram (B) of principal branches. (For key, see Fig. 55.24 caption.)

1263

1264

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

18

19 17

Figure 55.23 (Left) Injection of the ascending pharyngeal artery. (For key, see Fig. 55.24 caption.) 9

injections by the much larger internal carotid artery), giving fine branches to the pharynx, the dura mater of the posterior cranial fossa and, in many individuals, to the posterior lobe of the pituitary gland. Posterior branches include the occipital artery, through which the carotid system frequently communicates with the vertebral arteries. The artery supplies muscles, scalp and, via a petromastoid branch, the dura mater. The posterior auricular artery is often very small. The terminal branches of the external carotid artery are the internal maxillary and superficial temporal arteries. The former turns forwards, deep to the mandible, giving inferior dental, middle meningeal, deep temporal, accessory meningeal, sphenopalatine, infraorbital and descending palatine branches. Of these, the middle meningeal artery (Fig. 55.24) is of particular radiological interest; it runs superiorly, often appearing to cross the superficial temporal artery on the lateral projection, through the foramen spinosum, where it makes an angular forward bend to run in a smooth curve around the greater wing of the sphenoid and up over the convexity to the midline at the vertex. It gives a posterior branch that runs backwards across the squamous temporal bone towards the lambda. Supplying the dura mater and the inner table of the skull, the middle meningeal artery may also give off the ophthalmic artery; conversely, it may arise as a recurrent branch of the latter. The superficial temporal artery is the main feeder to the scalp. It gives off a very proximal major branch, the transverse facial artery, which runs forwards parallel with the zygomatic arch, the branches over the cranium, with a more tortuous course than that of the middle meningeal artery.

Anastomotic pathways16,43,64,65 There are three main categories of collateral supply to the brain: extracranial–intracranial anastomoses, the circle of Willis and leptomeningeal collaterals. The extracranial– intracranial collaterals are actual or potential anastomotic

Figure 55.24 (Right) Injection of the middle meningeal artery. Key for Figures 55.22–55.24: 1 = internal carotid artery, 2 = superior thyroid artery, 3 = lingual artery, 4 = facial artery, 5 = occipital artery, 6 = posterior auricular artery, 7 = internal maxillary artery, 8 = inferior dental artery, 9 = middle meningeal artery MMA, 10 = middle deep temporal artery, 11 = anterior deep temporal artery, 12 = infraorbital artery, 13 = descending (greater) palatine artery, 14 = superficial temporal artery, 15 = transverse facial artery, 16 = common carotid artery, 17 = ascending pharyngeal artery, 18 = anterior branch of the MMA, 19 = posterior branch of the MMA.

connections between branches of the external carotid artery and the internal carotid or vertebral arteries. These play a role in chronic cerebrovascular occlusive disease and their knowledge is of importance for interventional endovascular procedures. Principal anastomoses between the external and internal carotid artery are: • facial artery • middle meningeal artery • ophthalmic artery • superficial temporal artery • artery of foramen rotundum • vidian artery • carotid siphon Principal anastomoses between the external and posterior circulation are: • occipital artery • ascending pharyngeal artery • vertebral artery The circle of Willis65 plays a critical role as a collateral supply in acute and chronic cerebrovascular occlusive disease and during balloon occlusion of one of the internal carotid arteries. Its anterior part is formed by the distal

CHAPTER 55

internal carotid arteries, precommunicating segments of the anterior cerebral arteries (A1 segments), and anterior communicating artery; its posterior part is formed by the distal basilar artery, precommunicating segments of the posterior cerebral arteries (P1 segments), and posterior communicating arteries. The A1 segments course above the optic nerves and the posterior communicating arteries course below the optic tracts. The circle of Willis is well demonstrated with axial projections of MR angiograms and phase contrast MRA can provide information about the flow direction in its various components (see Figs 55.14, 55.25). A complete circle of Willis is only found in about 40 per cent of people and various segments of the circle may be sufficiently small or absent to be ineffective as a collateral channel. Common variations include absence or hypoplasia of one of the A1 segments and of one or both posterior communicating arteries. Another common variation is a fetal origin of the posterior cerebral artery from the internal carotid artery, which occurs in 20–30 per cent and is often associated with hypoplasia of the P1 segment on that side. There are other developmental connections between the anterior (carotid) and posterior (vertebrobasilar) circulation that may persist into adult life (Fig. 55.26, Table 55.3); of these only the trigeminal artery is encountered with any frequency, but is found in less than 1 per cent of normal people. Leptomeningeal (pial) collaterals are end-to-end anastomoses between distal branches of the intracerebral arteries that can provide collateral flow across vascular watershed zones. These are highly variable and are of great importance in acute occlusion of intracerebral vessels.

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Figure 55.26 Persistent caroticovertebral connections. B = basilar artery, EC = external carotid artery, H = hypoglossal artery, IC = internal carotid artery, O = otic artery, PAI = proatlantal intersegmental artery, PC = posterior cerebral artery, PCo = posterior communicating artery, T = trigeminal artery, V = vertebral artery.

Table 55.3 PERSISTENT CAROTICOVERTEBRAL ANASTOMOSES Artery

Origin

Termination

Route

Pro-atlantal

Cervical internal

Vertebral artery

Via foramen intersegmental carotid artery magnum

Hypoglossal

Internal carotid

Vertebral artery

Via hypoglossal artery canal

Otic

Petrous internal

Basilar artery

Via internal (exceptionally rare) carotid artery auditory meatus

Trigeminal

Precavernous

Basilar artery

Transdural internal carotid artery

INTRACRANIAL VEINS16,66,67 Dural sinuses

Figure 55.25 Fetal origin of the posterior cerebral artery. A 3D TOF MRA of the circle of Willis shows a fetal origin of the left posterior cerebral artery (arrow), which arises from the left internal carotid artery and is associated with hypoplasia of the left P1 segment.

The dural sinuses run within the major dural septa: the superior sagittal sinus between the layers of the upper part of the falx cerebri and the inferior sagittal sinus in the lower border of the falx, running backwards to join the great vein of Galen. The straight sinus is formed by the confluence of the vein of Galen and inferior sagittal sinus and runs downwards in the junction of the falx cerebri and tentorium cerebelli towards the torcular Herophili. The transverse (or lateral) sinuses run in the outer border of the tentorium itself, where it attaches to the vault. They appear frequently asymmetrical in size and the right is usually the dominant one. They become the sigmoid sinuses as they turn downward behind the lateral portions of the petrous bones to discharge into the internal jugular veins, which run in the lateral portion (the pars vascularis) of the jugular foramina. The superior petrosal sinuses extend from the cavernous sinus to the sigmoid sinuses and run along the attachment

1265

1266

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

of the tentorium cerebelli to the petrous ridge. The inferior petrosal sinuses connect the cavernous sinus to the jugular bulb and run in a groove between the petrous apices and clivus. The disposition of the inferior petrosal sinuses is highly variable, which has practical implications for inferior petrosal sampling, a procedure sometimes performed to lateralize pituitary microadenomas.

bv vg

Cerebral veins (Figs 55.27, 55.28)

sgs

Angiographically, these consist of two groups: the deep, sub-ependymal veins and the cortical veins. The former are rather constant, while the latter are extremely variable. In the angiographic series, the cortical veins fill before the deep ones, and usually from frontal to occipital, but deviations from this pattern are not necessarily abnormal. The deep and superficial groups are in fact joined by fine medullary veins, which run a straight course, perpendicular to the surface of the brain.

sts ts

vc

A ss

Deep veins

icv

The septal veins course directly posteriorly on either side of the midline, on the septum pellucidum, to join the thalamostriate veins as the latter run anteromedially across the floor of the lateral ventricle. They meet at the posterior lips of the foramina of Monro forming the venous angle, from which the internal cerebral veins run posteriorly on the roof of the third ventricle, near the midline. All the aforementioned veins are paired bilateral structures, as are the basal veins (of Rosenthal), which arise in the region of the choroidal fissures and course posterosuperiorly around the midbrain.

aca sts ica

vg

bv ba

va

B Figure 55.28 B3D TOF MR venogram in axial (A) and sagittal (B) plane. Note that arteries can sometimes also be seen on MR venograms. aca = anterior cerebral artery, ba = basilar artery, bv = basal vein of Rosenthal, ica = internal cerebral artery, icv = internal cerebral vein, sgs = sigmoid sinus, smcv = superficial middle cerebral vein, sts = straight sinus, ss = sagittal sinus, ts = transverse sinus, va = vertebral artery, vc = venous confluence, vg = vein of Galen. 8 3

4

7

2 6

9

5

10 11 12

Figure 55.27 Venous phase of internal carotid arteriogram. Lateral projection. 1 = septal vein, 2 = venous angle indicating the foramen of Monro, formed by junction of 3 = internal cerebral vein and 4 = thalamostriate vein, 5 = basal vein (of Rosenthal), 6 = great vein of Galen, 7 = straight sinus, 8 = superior sagittal sinus, 9 = superficial middle cerebral vein, 10 = temporoparietal cortical vein (inferior anastomotic vein of Labbé), 11 = lateral sinus, 12 = internal jugular vein.

The confluence of both internal cerebral and both basal veins gives rise to the unpaired great vein of Galen, which lies in the quadrigeminal cistern and shows a characteristic upward concavity as it delineates the posterior end of the corpus callosum before discharging into the straight sinus. Because of their constant relationships to the ventricular system, and the fact that they generally become visible at the point at which they reach the ependyma, the deep cerebral veins are an indicator of the size and shape of the lateral ventricles. The spread of the thalamostriate veins on the AP projection indicates the size of the central part of the lateral ventricles. Displacement of the deep cerebral veins from the midline is seen with posteriorly placed masses, whereas anterior masses cause displacement of the anterior cerebral arteries.

CHAPTER 55

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

Superficial veins Cortical veins can be divided into three main groups. The largest numbers drain upwards and medially to the superior sagittal sinus. Veins in the inferior frontoparietal and temporal regions drain to the superficial middle cerebral vein, thence to the sphenoparietal sinus. Inferior parietal, posterior temporal and occipital veins drain directly to the transverse sinuses. Two large cortical veins running posterosuperiorly across the parietal lobe to the superior sagittal sinus and posteroinferiorly over the temporal lobe to the transverse sinus are the superior and inferior anastomotic veins (of Trolard and Labbé, respectively); it is uncommon for both to be well developed.

Posterior fossa veins (Figs 55.29, 55.30) The anatomy of the posterior fossa veins is very variable.There are three principal drainage pathways: the vein of Galen, the superior petrosal sinus and direct tributaries into the transverse and straight sinuses. The pontomesencephalic veins outline the midbrain and brainstem. The anterior and lateral pontomesencephalic veins are connecting channels of a plexus of small veins, closely applied to the anterior surface of the pons. The precentral cerebellar vein and superior cerebellar vein outline the anterior and posterior aspects of the superior cerebellar vermis, respectively.They enter the vein of Galen either jointly or separately. The inferior vermian veins are paired paramedian vessels that enter the straight sinus or anastomose with the superior vermian vein. The cerebellar hemispheres are drained by hemispheric veins that usually enter the transverse sinuses. The petrosal veins course anterolaterally below the trigeminal nerves to enter the superior petrosal sinuses just above the internal auditory meatus. They receive numerous tributaries from the cerebellum, pons and medulla and inner ear.

Figure 55.29 Vertebral angiogram. Lateral projection, venous phase. (For key, see Fig. 55.30 caption.)

Figure 55.30 Vertebral angiogram. Half-axial projection, venous phase. Key for Figures 55.29–55.30: 1 = inferior vermian vein, 2 = anterior pontomesencephalic vein, 3 = posterior mesencephalic vein, 4 = lateral mesencephalic vein, 5 = precentral cerebellar vein, 6 = superior vermian vein, 7 = great vein of Galen, 8 = straight sinus, 9 = petrosal vein, 10 = cerebellar hemispheric veins, 11 = transverse sinus, 12 = sigmoid sinus, 13 = internal jugular vein. Note the normal asymmetry of the posterior fossa.

REFERENCES 1. Du Boulay E 1980 Principles of X-ray diagnosis of the skull. Butterworth, London, pp 1–28 2. Newton T, Potts G 1971 Radiology of the skull and brain: The Skull, vol. 1, Mosby, St Louis 3. Jungreis C A, Yonas H, Firlik A D, Wechsler L R 1999 Advanced CT imaging (functional CT). Neuroimaging Clin North Am 9: 455–464 4. Koening M, Klotz E, Luka B, Vanderink D, Spittler J, Heuser L 1998 Perfusion CT of the brain: diagnostic approach for early detection of ischaemic stroke. Radiology 209: 85–93 5. Okuda T, Korogi Y, Shigematsu Y et al 1999 Brain lesions: when should fluid-attenuated inversion-recovery sequences be used in MR evaluation. Radiology 212: 793–798 6. Beauchamp N, Ulug A, Passe T, Zijl P V 1998 MR diffusion imaging in stroke: review and controversies. Radiographics 18: 1269–1283 7. Provenzale J, Sorensen G 1999 Diffusion-weighted MR imaging in acute stroke: theoretic considerations and clinical applications. Am J Roentgenol 173: 1459–1467 8. Yoshiura T, Wu O, Sorensen A 1999 Advanced MR techniques. Diffusion MR imaging, perfusion MR imaging, and spectroscopy. Neuroimaging Clin North Am 9: 439–453 9. LeBihan D 1995 Diffusion and Perfusion Magnetic Resonance Imaging: Applications to Functional MRI. Raven Press, New York, pp 365 10. Duyn J H, van Gelderen P, Talagala L, Koretsky A, de Zwart J A 2005 Technological advances in MRI measurement of brain perfusion. J Magn Reson Imaging 22: 51–53 11. Keston P, Murray A D, Jackson A 2003 Cerebral perfusion imaging using contrast-enhanced MRI. Clin Radiol 58: 505–513 12. Tallan H, Moore S, Stein W 1956 N-acetyl-L-aspartic acid in brain. J Biochem 219:257–264

1267

1268

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

13. Saunders D, Howe F, van den Boogart A, Griffiths J, Brown M 1997 Discrimination of metabolites from lipid and macromolecule resonances in cerebral infarction in humans using short echo proton spectroscopy. J Magn Resonance Imag 7: 1116–1121 14. Dillon W, Robert T 1999 The limitations of functional MR imaging: a caveat. Am J Neuroradiol 20: 536 15. Morris P (ed) 1997 Practical Neuroangiography, vol. 1. Williams & Wilkins, Baltimore, pp 1–408 16. Osborn A G 1999 Diagnostic cerebral angiography, 2nd edn. Lippincott Williams & Wilkins, Washington, pp 462 17. Tenjin H, Asakura F, Nakahara Y et al 1998 Evaluation of intraneurysmal blood velocity by time-density curve analysis and digital subtraction angiography. Am J Neuroradiol 19: 1303–1307 18. Anxionnat R, Bracard S, Macho J et al 1998 3D angiography. Clinical interest. First applications in interventional neuroradiology. J Neuroradiol 25: 251–262 19. Probst E N, Grzyska U, Westphal M, Zeumer H 1999 Preoperative embolization of intracranial meningiomas with a fibrin glue preparation. Am J Neuroradiol 20: 1695–1702 20. Ng P P, Higashida R T, Cullen S P, Malek R, Dowd C F, Halbach V V 2004 Intraarterial thrombolysis trials in acute ischemic stroke. J Vasc Interv Radiol 15 (1 Pt 2): S77–S85 21. Stevens J M, Barter S, Kerslake R, Schneidau A, Barber C, Thomas D J 1989 Relative safety of intravenous digital subtraction angiography over other methods of carotid angiography and impact on clinical management of cerebrovascular disease. Br J Radiol 62: 813–816 22. Willinsky R A, Taylor S M, TerBrugge K, Farb R I, Tomlinson G, Montanera W 2003 Neurologic complications of cerebral angiography: prospective analysis of 2,899 procedures and review of the literature. Radiology 227:522–528. Epub 2003 Mar 13 23. Heiserman J E, Dean B L, Hodak J A, Flom R A, Bird C R, Drayer B P 1994 Neurologic complications of cerebral angiography. Am J Neuroradiol 15: 1401–1407 24. Leffers A, Wagner A 2000 Neurologic complications of cerebral angiography. Acta Radiol 41: 204–210 25. National Institute of Neurological Disorders and Stroke 1995 Carotid endarterectomy for patients with asymptomatic internal carotid artery stenosis. J Neurol Sci 129: 76–77 26. Napel S 1995 Principles and techniques of 3D spiral CT angiography. In: Fishman E, Jeffrey B (eds) Spiral CT: Principles, Techniques and Clinical Applications. Raven Press, New York, pp 167–182 27. Rubin G D, Shiau M C, Schmidt A J et al 1999 Computed tomographic angiography: historical perspective and new state-of-the-art using multi detector-row helical computed tomography. J Comput Assist Tomogr 23 (suppl 1): S83–S90 28. Provenzale J, Beauchamp N 1999 Recent advances in imaging of cerebrovascular disease. Radiol Clin North Am 37: 467–488 29. Velthuis B K, Leeuwen M S, Witkamp T D, Boomstra S, Ramos L, Rinkel G J E 1997 CT angiography: source images and post-processing techniques in the detection of cerebral aneurysms. Am J Roentgenol 169: 1411–1417 30. Link J, Brossmann J, Grabener M et al 1996 Spiral CT angiography and selective digital subtraction angiography of internal carotid artery stenosis. Am J Neuroradiol 17: 89–94 31. Leclerc X, Godefroy O, Lucas C et al 1999 Internal carotid arterial stenosis: CT angiography with volume rendering. Radiology 210: 673–682 32. Leclerc X, Godefroy O, Salhi A, Lucas C, Leys D, Pruvo J P 1996 Helical CT for the diagnosis of extracranial internal carotid artery dissection. Stroke 27: 461–466 33. Wong K S, Liang E Y, Lam W W, Huang Y N, Kay R 1995 Spiral computed tomography angiography in the assessment of middle cerebral artery occlusive disease. J Neurol Neurosurg Psychiatry 59: 537–539 34. Tipper G, U-King-Im JM, Price S J et al 2005 Detection and evaluation of intracranial aneurysms with 16-row multislice CT angiography. Clin Rad 60: 565–572 35. Karamessini M T, Kagadis G C, Petsas T et al 2004 CT angiography with three-dimensional techniques for the early diagnosis of intracranial

36.

37. 38. 39.

40.

41. 42.

43. 44.

45.

46.

47. 48. 49. 50. 51.

52.

53.

54. 55.

56.

57.

58. 59.

aneurysms. Comparison with intra-arterial DSA and the surgical findings. Eur J Radiol 49: 212–223 Jäger H R, Grieve J, Moore E, Kitchen N, Taylor W 1999 Comparison of CTA and intra-arterial DSA in patients with cerebral AVMs. Neuroradiology 41: 227–228 Casey S O, Alberico R A, Patel M et al 1996 Cerebral CT venography. Radiology 198: 163–170 Graves M J 1997 Magnetic resonance angiography. Br J Radiol 70: 6–28 Leclerc X, Gauvrit J Y, Nicol L, Pruvo J P 1999 Contrast-enhanced MR angiography of the craniocervical vessels: a review. Neuroradiology 41: 867–874 Isoda H, Takehara Y, Isogai S et al 1998 Technique for arterial-phase contrast-enhanced three-dimensional MR angiography of the carotid and vertebral arteries. Am J Neuroradiol 19: 1241–1244 Anderson C, Saloner D, Tsuruda J 1990 Artifacts in maximum-intensityprojection display of MR angiograms. Am J Roentgenol 154: 623–629 Ozsarlak O, Van Goethem J W, Maes M, Parizel P M 2004 Angiography of the intracranial vessels: technical aspects and clinical applications. Neuroradiology 46: 955–972. Epub 2004 Dec 4 Barboriak D, Provenzale J 1998 MR arteriography of intracranial circulation. Am J Roentgenol 171: 1469–1478 Ida M, Kurisu Y, Yamashita M 1997 MR angiography of ruptured aneurysms in acute subarachnoid hemorrhage. Am J Neuroradiol 18: 1025–1032 Jäger H R, Mansmann U, Hausmann O, Partzsch U, Moseley I F, Taylor W J 2000 MRA versus DSA in acute subarachnoid haemorrhage: a blinded multireader study of prospectively recruited patients. Neuroradiology 42: 313–326 Oelerich M, Lentschig M G, Zunker P, Reimer P, Rummeny E J, Schuierer G 1998 Intracranial vascular stenosis and occlusion: comparison of 3D time-of-flight and 3D phase-contrast MR angiography. Neuroradiology 40: 567–573 Jäger H, Grieve J 2000 Advances in non-invasive imaging of intracranial vascular disease. Ann R Coll Surg Engl 82: 1–5 Liauw L, Buchem M, Spilt A et al 2000 MR angiography of the intracranial venous system. Radiology 214: 678–682 Ross M R, Pelc N J, Enzmann D R 1993 Qualitative phase contrast MRA in the normal and abnormal circle of Willis. Am J Neuroradiol 14: 19–25 Yano T, Kodama T, Suzuki Y, Watanabe K 1997 Gadolinium-enhanced 3D time-of-flight MR angiography. Acta Radiol 38: 47–54 Jung H W, Chang K H, Choi D S, Han M H, Han M C 1995 Contrastenhanced MR angiography for the diagnosis of intracranial vascular disease: Optimum dose of gadopentate dimeglumine. Am J Roentgenol 165: 1251–1255 Remonda L, Heid O, Schroth G 1998 Carotid artery stenosis, occlusion and pseudo-occlusion: first-pass gadolinium-enhanced, threedimensional MR angiography-preliminary study. Radiology 209: 95–102 Isoda H, Takehara Y, Osogai S et al 2000 Software-triggered contrastenhanced three-dimensional MR angiography of the intracranial arteries. Am J Roentgenol 174: 371–375 Morris P 1997 The internal carotid artery. In: Morris P (ed.) Practical neuroangiography. Williams and Wilkins, Baltimore, pp 117–163 Willinsky R, Lasjaunias P, Berenstein A 1987 Intracavernous branches of the internal carotid artery [ICA]: comprehensive review of their variations. Surg Radiol Anat 9: 201–215 Osborn A 1994 Diagnostic neuroradiology. In: Patterson A (ed.) Diagnostic Neuroradiology, vol. 1. Mosby-Year Book, St Louis, pp 1–936 Perlmutter D, Rhoton A L 1976 Microsurgical anatomy of the anterior cerebral-anterior communicating-recurrent artery complex. J Neurosurg 45: 259–271 Perlmutter D, Rhoton A L 1978 Microsurgical anatomy of the distal anterior cerebral artery. J Neurosurg 49: 204 Lasjaunias P, Berenstein A 1990 Surgical Neuroangiography, Functional Vascular Anatomy of Brain, Spinal Cord and Spine, 1st edn, vol 3. Springer-Verlag, Berlin, pp 1–337

CHAPTER 55

60. Gibo H, Carver C C, Rhoton A L, Lenkey C, Mitchell R J 1981 Microsurgical anatomy of the middle cerebral artery. J Neurosurg 54: 151–169 61. Campos C, Churojana A, Rodesch G, Alvarez H, Lasjaunias P 1998 Basilar tip aneurysms and basilar tip anatomy. Intervent Neuroradiol 4: 121–125 62. Lasjaunias P L, Choi I S 1991 The external carotid artery: functional anatomy. Rev Neuroradiol 4 (suppl 1): 39–45 63. Williams P L 1995 Carotid system of arteries. In: Gray’s Anatomy, 38th edn. Churchill Livingstone, New York, pp 1513–1523 64. Helgertner L, Szostek M, Malek A K, Staszkiewicz W 1994 Collateral role of the external carotid artery and its branches in occlusion of the internal carotid artery. Int Angio 13: 5–9 65. Wolpert S M 1997 The circle of Willis. Am J Neuroradiol 18: 1033–1034 66. Curé J K, Van Tussel P, Smith M T 1994 Normal variant anatomy of the dural venous sinuses. Semin Ultrasound CT MR 15: 499–519 67. Carpenter A 1996 Cerebral veins and venous sinuses. In: Carpenter’s Human Neuroanatomy, 9th edn. Williams and Wilkins, Media, PA, pp120–128

•

SKULL AND BRAIN: METHODS OF EXAMINATION AND ANATOMY

FURTHER READING Atlas S 1996 Magnetic resonance imaging of the brain and spine. In: Atlas S (ed.) 2nd edn, Lippincott-Raven, New York, pp 1–1675 Beauchamp N, Ulug A, Passe T, Van Zijl P 1998 MR diffusion imaging in stroke: review and controversies. Radiographics 18: 1269–1283 De Deyn P P, Dierckx R A, Alave A, Pickut B A (eds) 1997 A Textbook of SPECT in Neurology and Psychiatry. John Libbey, London Frackowiak R S J, Friston K J, Frith C D, Dolan R J, Mazziotta J C 1997 Human Brain Function. Academic Press, London Keats T 1996 Atlas of Normal Roentgen Variants That May Stimulate Disease. Mosby 6th edn. Lasjaunias P, Berenstein A 1990 Surgical Neuroangiography, Functional Vascular Anatomy of Brain, Spinal Cord and Spine, 1st edn, vol 3. Springer-Verlag, Berlin, pp 1–337 Ockner J, Nesbit G 1999 Angiographic evaluation and correlative anatomy. Neuroimaging Clin North Am 9: 475–490 Osborn A G 1999 Diagnostic cerebral angiography, 2nd edn. Lippincott Williams & Wilkins, Washington

1269

CHAPTER

Benign and Malignant Intracranial Tumours in Adults

56

H. Rolf Jäger, Gisele Brasil Caseiras and Philip M. Rich

Radiological investigations in intracranial tumours Computed tomography Magnetic resonance imaging • Structural MRI • Physiology-based MR imaging • MR perfusion imaging • Permeability imaging (KTRANS) • MR diffusion imaging • MR spectroscopy • FMRI

Classification of intracranial tumours Intra-axial tumours • Neuroepithelial tumours • Lymphomas • Metastases • Intraventricular tumours • Extra-axial tumours • Skull base tumours • Pituitary region tumours

RADIOLOGICAL INVESTIGATIONS IN INTRACRANIAL TUMOURS Plain radiographic findings in brain tumours are of historical interest, but radiologists should still be familiar with signs of raised intracranial pressure (ICP) such as erosion of the lamina dura of the dorsum sellae, or a ‘J-shaped’ sella. Skull radiography may also demonstrate tumour calcification and enlargement of middle meningeal artery grooves in meningiomas. The use of diagnostic catheter angiography for brain tumours has dramatically decreased with the advances in crosssectional imaging. It is occasionally performed to assess the vascular supply of meningiomas pre-operatively. Otherwise it

is now mostly performed in conjunction with pre-operative or palliative tumour embolizations or intra-arterial chemotherapy for treatment of high-grade gliomas. Magnetic resonance imaging (MRI) is the preferred investigation of patients with suspected intracranial tumours. It provides a better soft-tissue differentiation and tumour delineation than CT and advanced MR imaging techniques, such as diffusion-weighted (DWI) and perfusion-weighted (PWI) imaging and MR spectroscopy (MRS), allow the assessment of physiological and metabolic processes.

COMPUTED TOMOGRAPHY Most clinically symptomatic brain tumours are detectable on CT, by virtue of mass effect and/or altered attenuation. Intraaxial tumours are usually of low attenuation on non-enhanced CT images. High attenuation areas within a tumour indicate tumour calcification or recent intratumoural haemorrhage. Tumours frequently exhibiting these two features are listed in Table 56.1.

Bone-window settings can reveal bone erosion or hyperostosis, associated with extra-axial tumours. Contrast-enhancement improves the visualization of strongly enhancing mass lesions such as menigiomas, neuromas, metastases and certain types of glial tumours. More recently CT perfusion has emerged as a technique to assess the relative blood volume (rCBV) and permeability

1272

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 56.1 COMMONLY CALCIFIED AND HAEMORRHAGIC LESIONS Commonly calcified lesions

Commonly haemorrhagic lesions

Oligodendrogliomas (90%)

GBM (grade 4 glioma)

Choroid plexus tumours

Oligodendroglioma

changes in brain tumours1. Compared with MR perfusion techniques it has still limited area coverage despite progress in multi-detector technology. It has, however, the advantage of a direct relationship between the CT attenuation coefficient and contrast material concentration in tissue.

Ependymoma Central neurocytoma

Metastases

Meningioma

– Melanoma

Craniopharyngioma

– Lung

Teratoma

– Breast

Chordoma

MAGNETIC RESONANCE IMAGING STRUCTURAL MRI The MRI protocol for structural tumour imaging should include T2-weighted (T2W), fluid-attenuated recovery (FLAIR) sequence and T1-weighted (T1W) images before and after injection of a gadolinium-based contrast medium. Most tumours appear hypointense on T1W and hyperintense on T2W and FLAIR images. The latter provide a particularly good contrast between normal brain tissue and glial tumours and show signal loss in cystic tumour components2. Highly cellular tumours such as lymphomas and primitive neuroectodermal tumours have decreased water content and therefore appear relatively hypointense on T2W images. Intratumoural haemorrhage or calcification are also hypointense on T2W images and become more conspicuous on T2*W images, where magnetic susceptibility effects are stronger. Hyperintensities on T1W images can be due to haemorrhage, calcification, melanin (in metastatic melanomas) or fat. Enhancement with gadolinium is seen in vascular extraaxial tumours such as meningiomas and in intra-axial tumours, which disrupt the blood–brain barrier. This is generally a feature of high-grade intra-axial tumours, but can also be present in certain low-grade tumours, such as pilocytic astrocytomas and WHO grade II oligodendrogliomas. The visibility of contrast enhancement on MR can be improved by magnetization transfer imaging, by doubling or tripling the gadolinium dose3, or by using high relaxivity gadolinium compounds4.

PHYSIOLOGY-BASED MR IMAGING DWI, PWI, MRS and functional MRI (fMRI) with blood oxygen level dependent (BOLD) imaging provide information about physiological and metabolic processes not available on standard MRI. Much of the recent progress in tumour imaging is based on the use of these methods, which are now increasingly implemented in clinical practice5–8.

MR PERFUSION IMAGING Dynamic susceptibility-weighted contrast-enhanced (DSC) MR imaging is the most widely used technique of PWI in

brain tumours. It offers the possibility to assess the density of blood vessels within a tumour and provides an indirect measure of tumour neovascularity, an important feature of malignant tumours5,8,9. It differs from contrast enhancement, which is an indicator of vascular endothelial (blood–brain barrier) integrity. rCBV measurements derived from PWI correlate closely with angiographic and histological markers of tumour vascularity10, and also with the expression of vascular endothelial growth factor (VGEF) in tumours11, an important determinant of angiogenesis. High-grade glial tumours tend to have higher rCBV values than low-grade tumours and PWI significantly increases the specificity and sensitivity of conventional MRI in the classification of gliomas. MRP can be preformed using a spin-echo or gradient-echo technique. The latter is influenced by vessels of a larger size and has been shown to be superior in discriminating low-grade from highgrade gliomas12. Maps of rCBV may also be a useful adjunct for stereotactic tumour biopsies, and can help to direct tissue sampling towards areas with maximal angiogenesis.

PERMEABILITY IMAGING (KTRANS) Microvascular permeability of brain tumours can be quantified by measuring the transfer coefficient KTRANS, which is influenced by endothelial permeability, vascular surface area and flow. In can be measured using a T1W steady-state or a firstpass T2*W gradient-echo technique. The former has a higher spatial resolution and is more accurate but requires longer acquisition times and more complicated post-processing; the latter can be combined with DSC perfusion imaging. KTRANS correlates with tumour grade and might be even more sensitive than rCBV measurements for glioma grading13.

MR DIFFUSION IMAGING Diffusion-weighted imaging (DWI) measures Brownian motion of water molecules within the tissue. Isotropic- trace-weighted DW images are obtained by measuring the signal loss on typically T2W images following the application of diffusion gradients. The signal loss depends on several factors including the gradient strength and apparent diffusion coefficient (ADC),

CHAPTER 56

which describes water diffusibility in tissue. The more mobile the water molecules are, the higher the ADC and the greater the signal loss on DW images8. Trace-weighted DW images are to some degree still dependent on T2 effects, which can lead to T2 shine-through artefacts, whereas ADC maps provide a quantitative voxel-by-voxel representation of water movement. Visual inspection of trace-weighted DW images plays a limited role in the diagnosis of brain tumours. It may be useful to identify lesions with severely restricted diffusion, such as acute infarcts or abscesses, which can occasionally mimic brain tumours on standard MRI. More detailed analysis of brain tumours requires measurement of the ADC, which is an indicator of disruption of tissue microstructure, cellular density and matrix composition in tumours. ADC measurements correlate inversely the histological cell count of gliomas14 and positively with the presence of hydrophilic substances in the tumour matrix15. DWI and ADC measurements assess the overall freedom of water-movement whereas diffusion tensor imaging (DTI) provides additional information about the direction of water diffusion16. The tendency of water to move in some directions more than others is called anisotropy and can be quantified using parameters such as fractional anisotropy (FA). Compact white matter tracts show normally a high degree of anisotropy, which can be lost if they are infiltrated by tumour cells, which destroy the ultrastructural boundaries formed by myelin sheaths. Post-processing of DTI allows the depiction of important white matter tracts and their connections (tractography), which can be displayed in direction-encoded colour images. Tractography is useful in the pre-operative assessment of brain tumours and can differentiate between displacement and infiltration of white matter tracts17.

MR SPECTROSCOPY Proton MR spectroscopy (MRS) analyses the biochemistry of a brain tumour and provides semiquantitative information about major metabolites7,18. A common pattern in brain tumours is a decrease in N-acetylaspartate (NAA), a neuronspecific marker, and creatine (Cr) and an increase in choline (Cho), lactate (Lac), lipids (L) (Fig. 56.1).The concentration of Cho is a reflection of the turnover of cell membranes (due to accelerated synthesis and destruction) and is more elevated in regions with a high neoplastic activity. Lactate (Lac) is the end product of nonoxidative glycolysis and a marker of hypoxia in tumour tissue. This is of increasing interest as tumour hypoxia is now recognized as a major promoter of tumour angiogenesis and invasion. Lac is probably associated with viable but hypoxic tissue, whereas mobile lipids are thought to reflect tissue necrosis with breakdown of cell membranes. The choice of echo time (TE) is an important technical consideration for performing MRS. It can be short (20–40 ms), intermediate (135–144 ms) or long (270–288 ms). MRS with

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.1 Proton magnetic resonance spectroscopy. The choline peak (3.22 p.p.m.) is elevated, the creatine peak (3.03 p.p.m.) is low and the N-acetyl aspartate peak (2.01 p.p.m.) is nearly undetectable; characteristic spectroscopic appearance of gliomas (CHO = choline, PCr/Cr = creatine, NAA = N-acetylaspartate).

a short TE has the advantage of demonstrating additional metabolites, which may improve tumour characterization, such as myo-inositol, glutamate/glutamine (Glx) and lipids but is hampered by baseline distortion and artefactual NAA peaks. Intermediate echo times have a better defined baseline and quantification of NAA and Cho is more accurate and reproducible. Long echo times lead to a decrease of signal to noise. MRS is presently a sensitive but not very specific technique. Single voxel acquisition provides good-quality spectra but is prone to sampling errors. Chemical shift imaging is technically more demanding but covers a larger volume of tissue.

FMRI Blood oxygen level-dependent (BOLD) imaging detects changes in regional cerebral blood flow during various forms of brain activity. Paradigms using motor tasks, language and speech productions and memory are able to show activation of relevant cortical areas (Fig. 56.2). The main use of fMRI in tumour imaging is the pre-operative localization of eloquent cortical regions that may have been displaced, distorted or compressed by the tumour18. This can improve the safety of surgery and allow for a more radical resection. If possible fMRI should be combined with MR tractography in order to minimize intra-operative injury to white matter tracts connected to eloquent cortical areas.

1273

1274

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 56.2 fMRI in a patient with a right peri-insular and temporal lobe tumour. Images acquired during a picture-naming task show activation in the visual cortex and in the right Broca’s area, which lies outside the tumour.

CLASSIFICATION OF INTRACRANIAL TUMOURS There are several ways of classifying brain tumours: primary versus secondary, intra-axial (arising from the brain parenchyma) versus extra-axial (arising from tissues covering the brain, such as the dura) and various regional classifications (supratentorial, infratentorial, intraventricular, pineal region and sellar region tumours). The World Health Organization (WHO) classification of intracranial tumours is a universally accepted histological classification of brain tumours. It was extensively revised

in 1993 and updated in 2000, stratifying neoplasms by their overall biological potential19,20. The WHO classification no longer relies on standard pathological features alone but includes information from immunochemistry and molecular tumour profiling. An outline of this classification is given in Table 56.2. Intra-axial tumours, which conform largely to the tissue types 1, 4 and 5 and 9, will be discussed first, followed by extra-axial tumours corresponding to the tissue types 2, 3, 6, 7 and 8.

Table 56.2 ABBREVIATED WHO CLASSIFICATION OF BRAIN TUMOURS 1.

Tumours of neuroepithelial tissue

4

Lymphoma and haemopoetic tumours

1.1

Astrocytic tumours

5

Germ cell tumours

1.2

Oligodendroglial tumours

5.1

Germinoma

1.3

Ependymal tumours

5.2

Teratoma

1.4

Mixed gliomas

5.3

Choriocarcinoma

1.5

Choroid plexus tumours

6

Cysts and tumour-like conditions

1.6

Uncertain origin

6.1

Rathke’s cleft cyst

1.7

Neuronal and mixed neuronal-glial tumours

6.2

Epidermoid cyst

1.8

Pineal tumours

6.3

Dermoid cyst

1.9

Embryonal tumours

6.4

Colloid cyst

2

Tumours of cranial and spinal nerves

7

Tumours of the sellar region

2.1

Schwannoma

7.1

Pituitary adenoma

2.2

Neurofibroma

7.2

Craniopharyngioma

3.

Tumours of the meninges

8

Local extension from regional tumours

3.1

Meningioma

9

Metastases

3.2

Mesenchymal tumours (incl. haemangiopericytoma)

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

INTRA-AXIAL TUMOURS NEUROEPITHELIAL TUMOURS Neuroepithelial tumours account for 50–60 per cent of all primary brain tumours and represent a broad spectrum of neoplasms arising from or sharing morphological properties of neuroepithelial cells. They include glial neoplasms; choroid plexus tumours; tumours with predominant neuronal phenotype (ganglioglioma, dysembryoplastic neuroepithelial tumour and neurocytoma); pineal tumours and embryonal tumours (neuroectodermal tumours, medulloblastoma).

Gliomas Gliomas are the commonest neuroepithelial tumours. There are three types, distinguished depending on the cell type they originate from: atrocytomas, oligodendrogliomas and oligoastrocytomas21.

Astrocytic tumours Astrocytomas account for approximately 75 per cent of glial tumours.The WHO classification distinguishes four histological grades, ranging from the benign pilocytic astrocytomas (grade I) to glioblastoma multiforme (GBM) (grade IV), which is the most malignant astrocytic tumour.The incidence of the various types of astrocytic tumours varies with age. In children, most of these are relatively benign tumours (pilocytic or low-grade astrocytomas); in young adults low-grade astrocytomas predominate, whereas anaplastic astrocytomas have a peak incidence around 40 years and GBM usually occurs after 40 years21. WHO grade I is now reserved for pilocytic astrocytomas. These are well-circumscribed, potentially resectable lesions with a low proliferative potential and a predilection for the posterior fossa (Fig. 56.3). They are primarily seen in children

Figure 56.3 Cerebellar pilocytic astrocytoma. Axial T1W postgadolinium MRI. There is a cystic lesion in the cerebellum with a small, enhancing mural nodule but otherwise nonenhancing cyst wall. The fourth ventricle is compressed causing hydrocephalus (note enlargement of the temporal horns). The differential diagnosis of this lesion is a cerebellar haemangioblastoma.

and are rare in adults. Pilocytic atrocytomas have usually a significant cystic component and show enhancement that can be nodular or ring-like. Infratentorial pilocytic astrocytomas in adults are frequently mistaken for haemangioblastomas, which have a similar appearance and represent the commonest primary intra-axial tumour below the tentorium cerebelli in adults. About 20 per cent of haemangioblastomas occur in association with von Hippel–Lindau disease. Typical CT and MR appearance is of a cystic mass with an enhancing mural nodule. Some haemangioblastomas may, however consist of only solid components, which enhance strongly, whereas others may appear mainly cystic without significant enhancement. Diffuse astrocytomas WHO grade II are infiltrating lowgrade tumours that occur typically in the hemispheres of young adults, involving cortex and white matter. They have less well-defined borders than pilocytic astrocytomas and contrast enhancement is usually absent. WHO grade II astrocytomas show a low mitotic activity but have a propensity to progress to a higher histological grade.They have variable mass effect, appear iso- or hypodense on CT and show areas of calcification in up to 20 per cent. MRI (and specifically a FLAIR sequence) is better in defining the extent of the low-grade gliomas (Fig. 56.4), which are hyperintense on T2W images and FLAIR images and hypo/isointense on T1W images. Anaplastic astrocytomas (WHO grade III) are high-grade gliomas with an increased mitotic activity and raised immunohistochemical proliferation indices. According to the revised WHO classification of 2000, presence of a single mitosis is, however, no longer a distinguishing criterion between grade II and grade III astrocytomas20.Anaplastic astrocytomas show usually contrast enhancement, and infiltration of the peritumoural tissues is more extensive than in grade II lesions22,23. They may also be accompanied by vasogenic oedema (Fig. 56.5). Pleomorphic xanthoastrocytoma (PXA) arises near the surface of the cerebral hemispheres and is frequently cystic. The tumour may enhance strongly and is usually associated with little or no oedema. Despite its fat content, it is T1 hypointense and T2 hyperintense on MRI31. PXA may be WHO grade II or III, depending on its proliferation rate24. GBM (WHO grade IV) shows poorly differentiated, often highly pleomorphic glial tumour cells with vascular proliferation and necrosis. It has the worst prognosis but is unfortunately also the commonest primary intracranial neoplasm in adults25. These rapidly growing, highly mitotic tumours may arise from pre-existing lower grade astrocytomas or occur de novo (particularly in older patients)21. Vasogenic oedema and contrast enhancement are usually much more extensive than in anaplastic astrocytomas. Tumour necrosis is a hallmark of GBM and appears on MRI as areas of non-enhancing T1 hypointensity, frequently surrounded by irregularly enhancing regions of active mitosis. Intratumoural haemorrhage contributes to the heterogeneous MR appearance of GBM with areas of high signal on T1W images and low signal on T2W images. A small number of GBMs may show evidence of subarachnoid seeding.

1275

1276

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 56.4 WHO grade II astrocytoma. Axial T2W (A), FLAIR (B) images showing a left frontal hyperintense mass lesion with well-defined borders and small cystic areas. On the trace-weighted DW image (C) the tumour is not very conspicuous as T2 effects and diffusion effects cancel each other out. On the ADC map (D) the glioma is easily identified as an area of increased diffusivity compared to normal brain parenchyma.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.5 WHO grade III astrocytoma. Coronal T1W post-contrast image (A) of a frontal irregularly enhancing mass with some cystic areas, which appears heterogeneous on T2W images (B). The latter also shows associated vasogenic oedema at the posterior margin of the tumour. There is marked mass effect with midline shift. The trace-weighted DW images (C) and ADC map (D) appear heterogeneous with cystic areas and more restricted diffusion peripherally.

1277

1278

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Oligodendrogliomas Oligodendrogliomas account for 10–15 per cent of all gliomas, and occur predominantly in adults. They are diffusely infiltrating neoplasms that are found almost exclusively in the cerebral hemispheres, most commonly in the frontal lobes, and typically involve subcortical white matter and cortex (Fig. 56.6). The WHO classification distinguishes between WHO grade II (well-differentiated low-grade) and WHO grade III (anaplastic high-grade) oligodendrogliomas. The former are slowly growing tumours with rounded homogenous nuclei; the latter have increased tumour cell density, mitotic activity, microvascular proliferation and necrosis. Both low- and high-grade oligodendral tumours express proangiogenic mitogens and may contain regions of increased vascular density with finely branching capillaries that have a ‘chicken wire’ appearance. This contributes to their appearance on contrast-enhanced MRI and MRP. Up to 90 per cent of oligodendrogliomas contain visible calcification on CT, which can be central, peripheral or ribbon-like26. On MRI, areas of calcification may be more difficult to appreciate due to the variable appearance of calcification. Intratumoural calcification appears typically T2 hypo- and T1 hyperintense but intratumoural haemorrhage, which occurs uncommonly in oligodendrogliomas, may have a similar appearance. Contrast enhancement is variable and often heterogeneous. Unlike in astrocytomas, contrast enhancement is not a reliable indicator of tumour grade in oligodendrogliomas. WHO grade II tumours do not infrequently exhibit some contrast enhancement whereas WHO grade III oligodendrogliomas may not enhance27. Lowgrade oligodendrogliomas may also have an elevated rCBV on PWI28 (Fig. 56.7).The identification of oligodendroglial tumour with loss of chromosomes 1p and 19q has become important as these subtypes show a better response to chemotherapy. Oligodendrogliomas with 1p/19q loss appear to have a significantly higher rCBV than those with intact 1p and 19q29. Conventional MR images may provide a further clue to the oligodendroglioma genotype.Tumours with intact1p/19q show more homogeneous signal on T1W and T2W images and have sharper borders than the tumours with chromosome deletions30.

MRS of low-grade tumours with oligodendral elements shows increased levels of myo-inositol/glycine as well as glutamine and glutamate7.

Physiological MR imaging in the differential and grading diagnosis of glial tumours Recent studies demonstrated that PWI and DWI can help to differentiate low-grade astrocytic from low-grade oligodendroglial tumours. WHO grade II oligodendrogliomas have significantly higher rCBV than WHO grade II astrocytomas with median values of 3.68 versus 0.92, respectively28. This concurs with the histological findings of regions with increased vascular density seen in oligodendrogliomas. Using a method of histogram analysis of all intra-tumoural ADC measurements, oligodendrogliomas were shown to have significantly lower ADC values than atrocytomas, which is probably a reflection of their higher cellular density and different tumour matrix composition31. Several studies have investigated the potential of advanced MR imaging to distinguish between low- and high-grade gliomas. Formation of new blood vessels (angiogenesis) represents an important aspect of tumour progression and growth and the microvascular density in glial tumours correlates with histological tumour grade. MR perfusion imaging is a noninvasive method of assessing the tumour microvasculature. Sugahara et al10 found mean maximum rCBV values of 7.32, 5.84 and 1.26 for glioblastomas, anaplastic astrocytomas and low-grade gliomas, respectively, which correlate closely with the findings of other investigators32. A recent study of 160 primary cerebral gliomas showed that rCBV measurements significantly increased the sensitivity and positive predictive value of conventional MR imaging in glioma grading33. PWI had a sensitivity of 95 per cent and positive predictive value of 87 per cent for distinguishing low-grade from high-grade gliomas when an rCBV threshold of 1.75 was used33. A separate study of low-grade gliomas demonstrated rCBV values above 1.75 to be associated with a more rapid tumour progression34. The role of DWI in differentiating high-grade from lowgrade gliomas is more controversial. Initial reports were

Figure 56.6 Oligodendroglioma. CT after IV contrast medium (A) shows a large left frontal tumour that involves the cortex. It is predominantly solid with irregular enhancement, but there are also cysts and coarse calcification. Follow-up after 2 years with CT (B), T2W MRI (C) and T1W post-contrast MRI (D) shows more extensive cyst formation and calcification than on the first scan. The calcification is much less apparent on MRI and appears as nonspecific low signal areas. Posterior infiltration of the tumour is, however, best seen on MRI (C). Note that the patient had undergone a left frontal craniotomy after the first CT.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.7 WHO grade II oligodendroglioma. Axial T2W image (A) and colour rCBV map (B) showing areas of increased rCBV (yellow and red areas) within the inhomogenous left frontal tumour.

encouraging by showing lower ADC measurements in highgrade lesions14,32.This could be accounted for by the increased cellularity and lesser content of hydrophylic components in the extracellular matrix of high-grade tumours. Subsequent studies failed, however, to demonstrate significant ADC differences between glioma grades35. The role of ADC measurements in predicting the glioma grade remains at present unclear and appears less promising than the rCBV measurements. The role of MRS in glioma grading is also subject to debate and remains currently investigational. Low-grade tumours have generally lower Cho levels than high-grade tumours, and lipid and lactate levels correlate with necrosis present in highgrade neoplasms. A threshold value of 1.56 for the Cho/Cr ratio produced a 75.8 per cent sensitivity and 47.5 per cent specificity for differentiating high- from low-grade gliomas7. Radiation necrosis is a late complication of radiotherapy or gammaknife surgery, and can present as an enhancing mass lesion, difficult to distinguish from recurrent tumour on conventional imaging. PWI and DWI may help to distinguish between radiation necrosis and tumour recurrence. In radiation necrosis the enhancing lesion has a low rCBV, which tends to be high in tumour recurrence as a consequence of increased new vessel formation8. ADC measurements of the enhancing components in recurrent tumour are significantly lower than in radiation necrosis36, mirroring the higher cellular density in recurrent neoplasm. Investigation of peritumoural regions with physiology-based MR techniques may be just as important as the assessment of the tumour itself. Differences in the peritumoural tissues of low-grade and high-grade gliomas have been demonstrated with DWI, PWI or MRS8,23,37–39. The peritumoural regions

of high-grade gliomas show a more marked decrease in ADC, fractional anisotropy and NAA and increase in rCBV compared to low-grade tumours. This is a reflection of the more invasive nature of these tumours, which infiltrate the adjacent brain tissue along vascular channels, leading to an rCBV increase; destroy ultrastructural boundaries with a consequent decrease in ADC and FA; and replace normal brain tissue, resulting in a drop of NAA. Metastases on the other hand are surrounded by ‘pure’ vasogenic oedema, which contains no infiltrating tumour cells. These peritumoural regions in metastases therefore show no increase in rCBV or decrease in FA.

Tumours of predominantly neuronal cell origin These include gangliocytomas, gangliogliomas, dysembryoplastic neuro-epithelial tumours (DNETs) and central neurocytomas. The latter is discussed under the section ‘Intraventricular tumours’ below. Gangliogliomas and gangliocytomas are slow-growing tumours with a low malignant potential, which occur preferentially in young adults and in the temporal lobe presenting with epilepsy. Gangliogliomas contain a mixture of neural and glial elements with neoplastic large ganglion cells; gangliocytomas have only neuronal elements. CT and MRI show peripherally located mixed solid/cystic lesions that commonly calcify. Enhancement can be variable and is often peripheral. Dysembryoplastic neuroepithelial tumours (DNETs) are highly polymorphic tumours that arise during embryogenesis. They are preferentially located in the supratentorial cortex and frequently manifest through intractable complex partial seizures. DNETs are usually hypodense on CT and T1-hypointense and T2-hyperintense on MRI. Small intratumoural cysts may be

1279

1280

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

present and cause a ‘bubbly’ appearance (Fig. 56.8) Calcification is seen in about 25 per cent but enhancement is uncommon. Thinning of the overlying bone is present in approximately half of the cases, reflecting the extremely slow growth of these tumours, which allows bone remodelling to occur40.

Pineal region tumours Pineal region tumours account for approximately 1 per cent of intracranial tumours in adults, whereas they represent 10 per cent of all paediatric brain tumours. They present either with obstructive hydrocephalus secondary to aqueduct compression or with problems with eye movements and accommodation, due to compression of the underlying tectal plate. More than half of all pineal region tumours are of germ cell origin (germinoma, teratoma, yolk sac tumours and choriocarcinoma)41. The diagnosis of a germ cell tumour is often made by the presence of marker proteins (such as α-fetoprotein or chorionic gonadotropin) in the cerebrospinal fluid. Germinomas are usually rounded, and often grey matter isointense on standard MRI. They show marked, homogeneous contrast enhancement and virtually never calcify. Germinomas may be multifocal (the second commonest site being the hypothalamic region) or show diffuse subependymal and subarachnoid spread, best appreciated on contrast-enhanced T1W images42. Teratomas appear more lobulated and inhomogeneous on CT and MRI, reflecting fat content and calcification. The margins of these tumours are often irregular.

The remaining pineal region tumours are of pineal cell (pineoblastomas, pineocytomas) or glial origin (astrocytomas). Pineocytomas are histologically benign tumours that lack specific imaging features and are therefore difficult to differentiate from other pineal region neoplasms. Pineoblastomas belong to the group of primitive neuroectodermal tumours (PNETs), and behave like cerebellar medulloblastomas, with frequent seeding via the CSF. They tend to be of low signal intensity on T2W images and can appear bright on DWI. Benign pineal cysts are common and must be differentiated from pineal tumours. They are smooth and well defined and can exhibit rim enhancement. Their signal on T1W, proton density-weighted and FLAIR images may be higher than CSF due to their protein content. They do not, however, cause hydrocephalus or a midbrain syndrome.

Embryonal tumours These are also called PNETs (primary neuro-ectodermal tumours) and represent high-grade (WHO grade IV) tumours of neuro-ectodermal origin, which include medulloblastomas and the aforementioned pineoblastomas. Medulloblastomas are the commonest posterior fossa tumour in children and arise classically from the super medullary velum at the roof of the 4th ventricle (Fig. 56.9). PNETs have a high cellular density and appear therefore hyperdense on CT, hyperintense on DWI and of intermediate to low density on T2W images. A strong (3–4-fold) elevation of Cho and Lip on MRS is also an expression of the high cellularity and cell turnover of these tumours7. PNETs enhance with IV contrast medium and have a propensity for dissemination in the subarachnoid space with leptomeningeal deposits. Staging of these tumours therefore requires a contrast-enhanced MRI of the entire neuroaxis23.

LYMPHOMAS

Figure 56.8 Coronal FLAIR of a dysembryoplastic neuro-epithelial tumour (DNET) showing a right parietal, pyramidal-shaped, predominantly cortically based, tumour. It has peripheral cystic areas and a linear area of hyperintensity extends towards the right lateral ventricle.

Primary cerebral lymphoma (PCL) has tripled in incidence over the past 2 decades23.This is partly due to a rise in patients with AIDS but PCL has also increased in immunocompetent patients. PCL appears as a single (less frequently multiple) lobulated enhancing mass, often abutting an ependymal or meningeal surface and involving basal nuclei. Enhancement is uniform in immunocompetent patients and ring-like in immunocompromised patients, in whom PCL frequently contains areas of central necrosis.The high cellular density and nucleus-to-cytoplasm ratio make PCL appear hyperdense on CT and hypointense on T2W images (Fig. 56.10). The ADC of PCL is lower than in gliomas43 or toxoplasmosis44, which is an important differential diagnosis in immunocompromised patients. PCL grows in an angiocentric fashion around existing blood vessels without extensive new vessel formation. Perfusion-weighted MRI therefore shows only a modest increase in rCBV, much less marked than in high-grade gliomas, where angiogenesis is a prominent feature23. A characteristic finding is rapid resolution of the tumour following administration of steroids and/or radiotherapy. Diffuse meningeal involvement is more common in secondary lymphoma and relatively rare in PCL.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.9 Medulloblastoma. Sagittal gadolinium-enhanced T1W (A), axial T2W images (B) and ADC map (C) demonstrate a heterogeneously enhancing mass posterior to the 4th ventricle, which is obliterated. The increased cellularity of this tumour is reflected by its relative hypointensity on the ADC map. (Courtesy of Dr R. Gunny.)

METASTASES The primary neoplasms that most commonly metastasize to the brain are carcinoma of the lung, breast and malignant melanoma (Fig. 56.11). Generally, metastases appear as multiple rounded lesions with a tendency to seed peripherally in the cerebral substance, at the grey/white matter junction.They can, however, occur anywhere in the cerebrum, brainstem or cerebellum and can also spread to the meninges. Metastases are characterized by oedema in the surrounding white matter, which appears dark on trace-weighted DWI and is often

Figure 56.10 Primary cerebral lymphoma. CT before (A) and after IV contrast medium (B). An irregular mass that is hyperdense to grey matter expands the splenium of the corpus callosum and extends into the left hemisphere. It is surrounded by extensive white matter oedema and enhances avidly with contrast.

1281

1282

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

disproportionate to the size of the tumour itself. On T2W images, the neoplastic nodule may blend with the surrounding oedema, giving a picture of widespread vasogenic oedema and obscuring the diagnosis. Most metastases enhance strongly with IV contrast medium, either uniformly, or ring-like if the metastasis has outgrown its blood supply. Most metastases from lung and breast are similar in density to normal brain parenchyma on CT, but some types are spontaneously dense, particularly deposits from malignant melanoma45. Haemorrhage occurs in about 10 per cent of metastases, resulting in high signal on T1W images and high or low signal on T2W images. Similar signal characteristics can also occur in nonhaemorrhagic metastases from melanoma, due to the paramagnetic properties of melanin. Small metastases and those that are not made conspicuous by surrounding oedema are

often only detected on contrast-enhanced studies. Increasing the contrast dose or relaxivity of gadolinium compounds can improve the sensitivity for detection of metastases on MRI3. Advanced MRI methods can also contribute towards diagnosis and differential diagnosis of metastases. DWI may help to predict the histology of metastases. Well-differentiated adenocarcinoma metastases are hypointense on trace-weighted DWI, whereas small cell and neuroendocrine metastases are hyperintense, due to their higher cellularity46,47. On standard MRI it may occasionally be difficult to distinguish a single metastasis from a glioma. PWI and MRS of the peritumoural rather than intratumoural region were shown to be useful in differentiating the two, as mentioned earlier. DWI is helpful to differentiate cystic metastasis (Fig. 56.12) from cerebral abscesses (Fig. 56.13)47. The latter contain more

Figure 56.11 Metastases: (A) melanoma (MRI): axial T2W (2000/80). There are at least three foci of signal hypointensity in the right hemisphere, the largest in the right posterior frontal cortex and the others deeper in the subcortical parietal region. This T2 shortening is attributable to melanin. (B) Axial post-contrast T1W with magnetization transfer (650/16). At a slightly different level, this post-contrast study discloses at least four rounded hyperintense metastatic deposits, all in the cortex or subcortical regions.

Figure 56.12 Cystic metastasis from CA breast. Axial T1W post-contrast image (A) demonstrates a peripherally enhancing, centrally necrotic lesion in the right thalamus. The lesion appears dark on the trace-weighted DW image (B) and bright on the ADC map (C), which is consistent with a relatively unrestricted diffusion in the center of the mass.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.13 Brain abscess. T1W post-contrast image (A) shows a cystic lesion with an enhancing rim centred on the region of the right basal ganglia. The trace-weighted DW image (B) shows a hyperintense lesion that appears markedly hypointense on the ADC map (C), and is distinct from the surrounding hyperintense vasogenic oedema. These features indicate severely restricted diffusion in the abscess cavity.

viscous fluid and pus and show a more marked restriction of water diffusion than necrotic tumours. Abscesses appear therefore bright on trace-weighted DWI and dark on ADC maps.

INTRAVENTRICULAR TUMOURS Approximately one-tenth of all primary intra-axial brain tumours involve the ventricular system. The precise anatomical location of the tumour within the ventricles often provides an important clue to the nature of the lesion. Common intraventricular tumours and cysts and their sites of predilection are summarized in Table 56.3. Intraventricular tumours arising from neuro-epithelial tissue (ependymomas, central neurocytoma and choroid plexus tumours) are discussed first.

Ependymomas Ependymomas arise from neoplastic transformation of the ependyma and account for about 5 per cent of adult primary brain tumours, being twice as frequent in children. Ependymomas are usually intraventricular, although extraventricular rests of ependymal cells may give rise to hemisphere tumours. Supratentorial tumours occur in young adults and fourth ventricular ependymomas (Fig. 56.14), which frequently extend through the foramina of Magendie and Luschka and have two

Table 56.3 INTRAVENTRICULAR LESIONS

Figure 56.14 Ependymoma of the fourth ventricle. Sagittal gadolinium-enhanced T1W (A) and axial T2W (B) MRI. A heterogeneously enhancing mass (arrow) fills the lower half of the fourth ventricle and extends through the foramina of Lushka (arrowhead) and Magendie to lie posterior to the medulla oblongata and upper cervical spinal cord, which are compressed from behind. There is obstructive hydrocephalus.

age peaks: at 5 and 35 years of age. They are well-demarcated, lobulated mass lesions that show calcification on CT in over 50 per cent and are of mixed signal intensity on MRI (predominantly hyperintense on T2W and iso- to hypointense on T1W images). MRI may demonstrate small cysts but calcification is less conspicuous than on CT. Enhancement is mild to moderate and often heterogeneous.

Tumour

Typical site

Colloid cyst

Foramen of Monro/third ventricle

Meningioma

Trigone of lateral ventricle

Central neurocytomas

Choroid

Fourth ventricle

Ependymoma

Lateral ventricle (more common in children) and fourth ventricle

Neurocytoma

Lateral ventricles (involving septum pellucidum)

Metastases

Lateral ventricles, ependyma and choroid plexus

Central neurocytomas are slow-growing intraventricular tumours of purely neuronal origin48. Before the advent of immunohistological methods they were frequently misdiagnosed as subependymal oligodendrogliomas. These relatively benign tumours occur predominantly in the second

1283

1284

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

and third decades of life, and represent probably the commonest lateral ventricular masses in this age group. They typically arise from the septum pellucidum and occupy the frontal horns and bodies of the lateral ventricles, and sometimes extend through the foramen of Monro. CT frequently shows calcification and small cysts. MRI shows a heterogeneously enhancing mixed-signal intensity mass containing septated cysts, susceptibility artefact from calcification and grey-matter-isointense nodules (Fig. 56.15). Obstructive hydrocephalus is common.

Choroid plexus tumours Choroid plexus papillomas are much more common than choroid plexus carcinomas (Fig. 56.16).The location and incidence of choroid plexus papillomas varies with age. They are relatively more common in childhood (3 per cent of primary brain tumours), presenting as a ‘cauliflower-like’ mass in the trigone of the lateral ventricle. In adults, papillomas are less common, and occur predominantly in the fourth ventricle. CT shows an iso- to hyperdense mass with punctate calcification and homogeneous enhancement. On MRI the papillomas appear as lobulated, intraventricular masses of heterogeneous, predominantly intermediate signal intensity on both T1W and T2W images, with intense contrast enhancement. Angiography, which is now rarely indicated, shows a highly vascular mass supplied predominantly by the anterior and posterior choroidal arteries. Choroid plexus carcinomas are rare, highly malignant tumours that invade the adjacent brain parenchyma to a greater degree than papillomas.

Colloid cysts

Figure 56.16 Choroid plexus papilloma. Coronal T1W post-gadolinium MRI. There is a lobulated, strongly enhancing tumour in the trigone of the left lateral ventricle. Both lateral ventricles are dilated due to hydrocephalus associated with this tumour.

cerebrospinal fluid from the lateral ventricles and are smooth, spherical lesions that are characteristically hyperdense on unenhanced CTs (Fig. 56.17). Their MR appearance varies depending on the cyst content (calcium, cholesterol, haemosiderin); some can have similar signal to CSF, but most are of high signal on T1W and on T2W images.

Meningiomas

These benign lesions occur exclusively at the paraphysis, which lies in the posterior lip of the foramen of Monro, between the third and lateral ventricles. They tend to cause hydrocephalus, by intermittently or continuously obstructing the outflow of

This is the commonest cause of a mass in the trigone of the lateral ventricle after the first decade of life. The CT and MRI appearances are similar to those of extraventricular meningiomas (see later): a well-defined, globular lesion that is usually hyperdense on CT and may give similar signal to cerebral

Figure 56.15 Central neurocytoma. Axial proton density (A) and coronal T1W post-gadolinium (B) MRI. A partly cystic, multi-septated, enhancing mass, which is related to the septum pellucidum, fills the bodies of both lateral ventricles and causes hydrocephalus with dilatation of the left temporal horn.

Figure 56.17 Colloid cyst. Unenhanced CT. There is a dense, rounded mass in the region of the foramen of Monro causing enlargement of the lateral ventricles, and indenting the anterior aspect of the third ventricle.

CHAPTER 56

cortex on T1W and T2W images. They usually show marked contrast enhancement on CT and MRI.

EXTRA-AXIAL TUMOURS Primary extra-axial neoplasms arise from the meningothelial arachnoidal cells (meningiomas), mesenchymal pericytes (haemangiopericytoma) or cranial nerves (schwannomas, neurofibromas) and include developmental cysts or tumourlike lesions (epidermoid and dermoid cysts). Metastatic involvement of the meninges and tumours in specific regions (around the sella turcica and skull base) are discussed separately. Overall, meningiomas represent the commonest nonglial intracranial neoplasm, accounting for approximately 20 per cent of all primary intracranial tumours. Multiple meningiomas and cranial nerve tumours are found in neurofibromatosis type 2. Extra-axial lesions occur much more frequently in adults than in children and account for the majority of primary infratentorial tumours in adults, with three lesions sharing a predilection for the cerebellopontine region: vestibular schwannoma, meningioma and epidermoid cysts. When analysing an extra-axial lesion, it is important to pay attention to associated bone changes: meningiomas tend to induce a hyperostotic bone reaction, whereas dermoid cysts and schwannomas tend to cause bone thinning resulting in enlargement of, for example, the middle cranial fossa or internal auditory meatus. Other features distinguishing extra- from intra-axial mass lesions are ‘buckling’ and medial displacement of the grey–white matter interface, a CSF cleft separating the base of the mass from adjacent brain and a broad base along a dural or calvarial surface37.

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Meningiomas Meningiomas originate from arachnoid cell rests, related to arachnoid granulations of the dura mater and may assume a spherical, well-circumscribed shape or be flat, infiltrating (‘en plaque’) lesions49. There are several histological types of meningioma including meningothelial, fibrous/fibroblastic, transitional and psammomatous tumours. Most of them correspond to WHO grade I. Meningiomas with atypical features such as increased mitotic activity, patternless growth and necrosis correspond to grade II (atypical) or grade III (anaplastic meningioma), depending on the extent of these features20. Of meningiomas, 90 per cent are supratentorial arising, in decreasing order of frequency, from the parasagittal region, cerebral convexities, sphenoid ridge and olfactory grove. Infratentorial meningiomas are most frequently located on the posterior surface of the petrous bones and clivus and can mimic acoustic neuromas. Bone sclerosis is in favour of meningioma and enlargement of the internal auditory meatus is much more common in neuroma. On CT, 60 per cent of meningiomas are spontaneously hyperdense and up to 20 per cent contain calcification. Enhancement on CT is usually intense and uniform (Fig. 56.18). Hyperostosis, best seen on bone window settings, indicates the site of the tumour attachment to the meninges. On MRI, meningiomas appear frequently isointense to cerebral cortex on both T1W and T2W images and may be difficult to detect without IV contrast medium. Meningiomas can have ‘capping cysts’ of similar MRI signal intensity to CSF. As on CT, meningiomas enhance vividly and homogeneously, except for the uncommon cystic and very densely calcified tumours. There may also be a linear, contrast-enhancing ‘dural tail’ extending from the tumour along the dura mater.

Figure 56.18 Subfrontal meningioma. CT before (A) and after (B) IV contrast medium, and lateral projection of common carotid arteriogram (C). There is a large circumscribed mass in the anterior cranial fossa that is isodense to normal grey matter, contains foci of calcification centrally and enhances homogeneously. There is oedema in the white matter of both frontal lobes and posterior displacement and splaying of the frontal horns of the lateral ventricles. On the arteriogram (C) the mass is delineated by a tumour blush and there is posterior displacement of the anterior cerebral arteries (arrowhead), mirroring the mass effect seen on CT. The ophthalmic artery is enlarged as its ethmoidal branches supply the tumour (arrow).

1285

1286

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

The ‘dural tail sign’, once thought to be pathognomonic for meningioma (Fig. 56.19) can also be seen with other tumours such as schwannoma or metastasis. Vasogenic oedema is not infrequently associated with meningiomas. The extent of the vasogenic oedema does not correlate with the size of the meningioma and, as with metastases, even small lesions can cause quite extensive oedema50,51. Meningiomas abutting the superior sagittal or transverse sinuses can compress or invade these venous structures.The distinction between compression and occlusion is important for preoperative planning, and can be made by magnetic resonance or computed tomography angiography (MRA and CTA, respectively). Physiological MR imaging methods are also of help in the diagnosis of meningiomas. DWI shows differences between typical and atypical meningiomas: the latter have lower ADC values than the former52. MRS may show an alanin peak, which is characteristic for a meningioma but this is seen in less than 50 per cent of cases7. Meningiomas have usually a markedly elevated rCBV on PWI (Fig. 56.19C), which can be used to differentiate from dural metastases, which tend to have a lower rCBV53.Angiography is occasionally performed for pre-operative assessment of the blood supply, and can be combined with pre-operative embolization, to reduce the blood loss during operation. The cardinal angiographic findings are supplied from meningeal vessels and a dense, homogeneous, persistent blush. Parasitization of cortical vessels by tumours over the cerebral convexity or of branches of the ophthalmic artery by subfrontal masses is not rare (Fig. 56.18.C). Haemangiopericytomas enter into the differential diagnosis for meningiomas. Features suggestive of a haemangiopericytoma rather than meningioma, are a lobulated (rather than spherical) dural-based mass, absence of calcification and hyperostosis and multiple areas of flow void on MRI, reflecting the high vascularity of these tumours54.

Cranial nerve sheath tumours Cranial nerve sheath tumours, originating from cranial nerves, account for 6–8 per cent of primary intracranial tumours. Most are benign neoplasms arising from Schwann cells (‘schwannomas’) of the nerve sheaths. Schwannomas arise eccentrically from the sheath and compress the parent nerve rather than invading it. All cranial nerves except I (olfactory) and II (optic), which are white matter tracts of the cerebrum, have nerve sheaths; however schwannomas usually grow on the sensory nerves, most frequently from the superior vestibular division of the vestibulocochlear nerve (‘acoustic neuroma’) and, with decreasing frequency, from the trigeminal, glossopharyngeal and lower cranial nerves. Pure motor cranial nerves rarely form schwannomas. Multiple cranial nerve schwannomas are found in neurofibromatosis type 2 and bilateral vestibular schwannomas are pathognomonic of this condition. Neurofibromas are benign tumours, composed of fibroblasts, reticulin and a mucoid matrix in addition to Schwann cells. Cranial nerve tumours are T1 iso/ hypointense and T2 hyperintense, and larger lesions often contain areas of cystic degeneration (Fig. 56.20). Cranial nerve tumours almost invariably show marked enhancement with IV contrast medium, which is solid in two-thirds and ring-like or heterogeneous in one-third of cases. Acoustic or, more accurately, vestibular schwannomas account for over 80 per cent of cerebellopontine lesions and widen the internal auditory meatus (IAM) if large enough. MRI is much more sensitive than CT for the detection of small vestibular schwannomas. High-resolution, thin-section, T2W fast spin-echo images of the posterior fossa shows the 7th and 8th nerves in detail (Fig. 56.21) and can detect small tumours causing focal nerve thickening. If the findings are equivocal, gadolinium-enhanced images refute or confirm the suspicion of a tumour.

Figure 56.19 Meningioma with perfusion-weighted imaging. Axial T2W (A), gadolinium-enhanced T1W (B) and perfusion-weighted (C) MRI. A grey-matter isointense mass deeply indents the left cerebral convexity (A). Its broad dural base, the surrounding displaced cerebral sulci and the small pial vessel between the tumour and the brain surface (arrowhead) are all features of an extra-axial lesion. The tumour enhances and there is a ‘dural tail’ (arrow) (B), which is a frequent radiological finding in meningioma, but is not pathognomonic (see Fig 56.24). Perfusion-weighted MRI (C): a colour map of the relative cerebral blood volume (rCBV) shows increased blood volume of the tumour compared to normal cortex and white matter, confirming its highly vascular nature.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.22 Suprasellar dermoid tumours. CT (A). There is a midline, fat density tumour (arrowheads) occupying the suprsasellar region. (B) Coronal T1W MRI of a different patient with a ruptured dermoid tumour. There is a lobulated high signal mass in the chiasmatic cistern compressing and displacing the optic chiasm to the left (arrow). Fat globules, which have spilled into the subarachnoid space, are seen as high signal foci in the left Sylvian fissure. The patient has had previous surgery via a right temporal approach, causing right temporal atrophy and enlargement of the right temporal horn.

Figure 56.20 Cystic vestibular schawannoma. T2W image reveals a large right cerebellopontine angle tumour with a medial cystic component. The mass extends into and expands the internal auditory meatus and distorts the right middle cerebellar peduncle.

Epidermoid and dermoid tumours Epidermoid and dermoid cysts, or ‘pearly tumours’, result from inclusion of ectodermal elements during the closure of the neural tube. Intracranial dermoid cysts lie usually near the midline and contain all skin elements, including fat, which is why they appear to be of very low density on CT and high signal intensity on T1W images (Fig. 56.22).They may rupture and release fat globules, which float in the CSF space. Epidermoid cysts can be central (chiasmatic and quadrigeminal plate cisterns) or eccentric (cerebellopontine angle,

middle cranial fossa, Sylvian fissure). Although present at birth, these lesions grow slowly by accumulating desquamated epithelium and conform to the shape of the portion of the subarachnoid space they occupy, sometimes invaginating into the brain parenchyma. On CT and standard T1W and T2W images, epidermoid cysts are non-enhancing lesions of similar density or signal intensity to CSF. They have to be differentiated from arachnoid cysts, which have better defined margins and cause bone thinning. DWI is very helpful to distinguish epidermoid tumours from arachnoid cysts (Fig. 56.23): water diffusion is markedly restricted in epidermoid tumours (which appear bright on DWI) but not in arachnoid cysts, which have similar signal characteristics to CSF.

Meningeal metastases

Figure 56.21 Vestibular schwannoma (Acoustic neuroma). Axial, high-resolution T2W MRI. There is a small soft-tissue mass (arrowhead) in the right internal auditory meatus, which is only minimally expanded. The normal 7th and 8th nerves on the left are clearly demonstrated.

Meningeal metastases may involve the pachymeninges (dura mater), leptomeninges (arachnoid and pia mater) or both. Contrast-enhanced MRI is much more sensitive than contrast-enhanced CT for detection of metastatic meningeal involvement. Carcinomatosis of the dura mater, common in carcinoma of the breast, manifests itself as focal curvilinear or diffuse contrast enhancement closely applied to the inner table of the skull, which does not follow the convolutions of the gyri. Focal segmental lesions may be difficult to distinguish from en plaque meningioma (Fig. 56.24). Leptomeningeal carcinomatosis produces linear or finely nodular contrast enhancement of the surface of the brain, extending into the sulci and following the convolutions of the brain. It may be indistinguishable from infective meningitis or sarcoidosis. Leptomeningeal disease is commonly seen in leukaemia, lymphoma and breast or lung cancer.

1287

1288

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 56.23 Epidermoid tumour. Axial T2W image (A) and sagittal gadolinium-enhanced T1W image (B) shows a large nonenhancing lesion of similar signal intensity to CSF, which occupies the chiasmatic and ambient cisterns, and distorts the medial aspect of the left temporal lobe. On the trace-weighted DW image (C) the lesion appears markedly hyperintense, indicating restricted diffusion.

Chordomas

Figure 56.24 Dural metastasis from breast carcinoma. Coronal T1W post-contrast MRI. There is a heterogeneously enhancing mass with an irregular surface that arises from the dura over the right cerebral convexity. It displaces the underlying brain and causes considerable low signal oedema within it. There is a dural ‘tail’ extending away from the tumour (arrowhead).

SKULL BASE TUMOURS Tumours of the skull base include a large pathological spectrum, such as metastases, myeloma/plasmacytoma, meningioma, caudally extending pituitary adenomas, direct extension of nasopharyngeal malignancies as well as tumours and inflammatory lesions arising in the paranasal sinuses. Two specific lesions of the central and posterior skull base are discussed later: chordoma and glomus jugulare tumours.

Chordomas originate from malignant transformation of notochordal cells and their most frequent location in the skull base is the spheno-occipital synchrondrosis of the clivus, followed by the basiocciput and petrous apex: tumours away from the midline are considerably less common. These tumours present usually with pain and lower cranial nerve palsies. They cause bone destruction and contain calcification, both of which are well demonstrated on CT with bone windows. On MRI the tumours exhibit mixed, heterogeneous signal55, and may have a ‘soap-bubble’ appearance (Fig. 56.25). The solid components show variable, but often marked contrast enhancement. Fatsuppressed T1W spin-echo sequences are particularly helpful for demonstrating the extent of the tumour and distinguishing pathological enhancement from the high signal of adjacent clival fat. The differential diagnosis includes chondrosarcoma, metastasis and nasopharyngeal carcinoma.

Glomus jugulare tumours Glomus jugulare tumours (chemodectomas) arise from paraganglion cells, the precursors of the chemo- and baroreceptors of the great vessels. The most common site is the jugular bulb and their presentation is with pulsatile tinnitus, deafness, vertigo and lower cranial nerve palsies. The tumour causes bone destruction with enlargement of the pars vasorum of the jugular foramen, well demonstrated on CT. The tumour, which fills the expanded foramen, gives high signal on T2W images. Glomus jugulare tumours enhance intensely with IV contrast medium and, because of their extreme vascularity, they tend to show areas of flow void that correspond to dilated vessels (Fig. 56.26). These tumours frequently obstruct the internal jugular vein, which may show signal changes indicative of thrombosis.

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

Figure 56.25 Clivus chordoma. Axial T2W (A), coronal T1W (B) and sagittal T1W post-gadolinium (C) MRI. A large mass has destroyed the clivus and extends superiorly compressing the midbrain and hypothalamus. It invades the right cavernous sinus, encasing the internal carotid artery, and extends into the nasopharyngeal soft tissues, posterior ethmoid air cells and optic canals. Tumour also projects into the pontine cistern but appears restrained by dura. The mass returns mixed signal and there is irregular enhancement following contrast injection.

Figure 56.26 Glomus jugulare tumour. An axial CT (A) demonstrates expansion of the right jugular foramen and bone destruction in the adjacent petrous bone by a mass that is markedly enhancing on axial T1W post-contrast images (B). The mass contains areas of flow voids, corresponding to the dilated tumour vessels seen on the right external carotid artery angiogram (C). (Courtesy of Dr M. Adams.)

1289

1290

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

PITUITARY REGION TUMOURS The region of the sella turcica contains many different tissues that have a propensity to undergo neoplasia55 and the differential diagnosis for sellar and parasellar masses is indeed large (Table 56.4). In addition to tumours, non-neoplastic lesions such as arachnoid cysts or giant aneurysms have to be considered.

Pituitary adenomas Pituitary adenomas are the most common neoplasms in the sellar region. They are classified as microadenomas (diameter < 1 cm) and macroadenomas (> 1 cm).They become symptomatic either because of their endocrine activity (microadenomas and functioning macroadenomas) or by the mass effect they exert (nonfunctioning macroadenomas), usually manifest by visual symptoms secondary to compression of the optic nerves and chiasm. Nonfunctioning microadenomas are not uncommon and can be found incidentally on MRI studies performed for other reasons. MRI is the investigation of choice for the detection of microadenomas. The standard protocol should include 3-mm thick coronal and sagittal T1W images through the pituitary gland before and after IV gadolinium. It is desirable to perform a fat-saturated T1W sequence after administration of IV contrast medium to eliminate high signal from fat in the clivus and clinoid processes, which could be mistaken for enhancement. The conspicuity of microadenomas can be increased by performing dynamic pituitary MRI (Fig. 56.27), which involves the acquisition of a series of rapid images with a time interval of approximately 10–15 s and exploits the time differences in enhancement between the adenoma and normal gland56. Functioning microadenomas can produce prolactin (prolactinomas), ACTH (in Cushing’s disease) or growth hormone (eosinophilic microadenomas). Prolactinomas are the most common functioning microadenomas and tend to arise later-

Table 56.4

ally within the anterior lobe of the pituitary gland. They may depress the floor of the sella turcica or expand one side of the gland, causing a subtle upwardly convex bulge and contralateral displacement of the infundibulum. Microadenomas are best shown on contrast-enhanced images and usually enhance later and/or to a lesser degree than normal pituitary tissue. The primary treatment of prolactin-secreting microadenomas is medical and the role of imaging in cases of hyperprolactinaemia is therefore mainly to exclude a macroadenoma. Precise localization of the microadenoma is, however, important in ACTH and thyroid-stimulating hormone (TSH)producing adenomas, which are treated surgically. If MRI is not conclusive, petrosal venous sampling may be necessary to lateralize an adenoma in pituitary-driven Cushing’s disease. Eosinophilic microadenomas can cause enlargement of the sella along with other features of acromegaly. Macroadenomas balloon the pituitary fossa and can have a suprasellar component or extend inferiorly into the sphenoid sinus and clivus. Suprasellar extension leads first to elevation, then to compression of the optic chiasm and intracranial optic nerves and a large tumour may compress brain parenchyma, often in the region of the hypothalamus. Most macroadenomas are isointense with brain parenchyma on unenhanced T1W images and hyperdense on CT. Administration of IV contrast may show uniform or heterogenous enhancement and facilitates the detection of cavernous sinus invasion57. Macroademomas may contain cystic or haemorrhagic components. Acute haemorrhage into a pituitary macroademona can lead to rapid expansion of the gland resulting in acute compression of the optic chiasm (pituitary apoplexy). Haemorrhage appears hyperintense on non-enhanced T1W images (Fig. 56.28) and, in the acute stage, hyperdense on CT.

Craniopharyngiomas Craniopharyngiomas are suprasellar tumours that occur most frequently in childhood, but may arise in adult life, and have a second peak of incidence at about the 6th decade. Symptoms

PRIMARY TUMOURS IN THE SELLAR AND PARASELLAR REGION

Tumour

Typical features

Pituitary macroadenoma

Enlarged sella turcica, strong enhancement, sometimes haemorrhage

Meningioma

Broad dural base, enhancement along planum sphenoidale Hyperostosis, ‘blistering’ of sphenoid sinus

Schwannoma

T1-hypo- and T2-hyperintense, strong (e.g. of fifth nerve) enhancement

Chordoma

Bone destruction on CT, heterogeneous signal and enhancement on MRI

Chondrosarcoma

Bone destruction and calcification on CT, T2 hyperintense on MRI

Crangiopharyngioma

Calcification, cysts, nodular enhancement

Rathke’s cleft cyst

T1-hyperintense on MRI, smooth peripheral enhancement

Dermoid

Hypodense on CT and T1 hyperintense on MRI

Epidermoid

Isodense to CSF on CT and isointense to CSF on T1 and T2 weighting, brighter than CSF on FLAIR and DWI

Tuber cinerum

Grey matter isointense on T1 weighting and T2 hamartoma hyperintense

Optic glioma

Thickening of chiasm, spread along optic pathways

Germ cell tumours

Located in midline, intense enhancement; can be synchronous with pineal germinomas

CHAPTER 56

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

solid and associated cysts return low signal on T1W images, similar to CSF. The solid components of craniopharyngiomas show intense contrast enhancement and may be partially calcified.

Rathke’s cleft cysts

Figure 56.27 Dynamic contrast-enhanced MRI of pituitary adenoma. Dynamically enhanced thin-section coronal T1W images of the pituitary gland. Images were acquired at 15-s intervals following injection of gadolinium. The scan at 90 s (A) reveals a microadenoma (arrowhead) to the right of the midline, which has enhanced to a lesser degree than the surrounding normal pituitary tissue at this stage. On the conventional enhanced MRI, acquired over 4 min (B), the enhancement of the lesion is similar to the rest of the gland.

are due to compression of the optic chiasm or to raised intra-cerebral pressure (ICP) secondary to obstruction of the foramen of Monro. They arise from epithelial remnants of Rathke’s pouch, from which the anterior pituitary develops, and can be cystic, solid or mixed cystic/solid (Fig. 56.29). Craniopharyngiomas tend not to expand the pituitary fossa unless they become very large (a differentiating feature from pituitary macroadenomas). The MRI appearances differ according to the histological type58. The adamantinous type, more commonly occurring in children, is more likely to have large T1 hyperintense cystic components, whereas the squamous-papillary type, which is more frequent in adults, is more likely to be predominantly

Symptomatic Rathke’s cleft cysts are much less common than craniopharyngiomas but asymptomatic lesions are relatively commonly found at postmortem. The cysts usually lie within the pituitary gland, although others are found adjacent to the infundibulum, above the sella. On MRI they appear frequently as T1 hyperintense cysts but may also exhibit similar signal characteristics to CSF. Rathke’s cysts do not usually enhance following IV contrast medium, although enhancement of the cyst wall is possible.

Other sellar region tumours Parasellar meningiomas can arise from the dura mater of the cavernous sinus or the tuberculum, dorsum or diaphragma sellae (Fig. 56.30). Clinical presentation is with cranial nerve palsies or visual symptoms. Parasellar meningiomas are strongly enhancing masses that expand the cavernous sinus and frequently encase and narrow of the cavernous portion of the internal carotid arteries. Suprasellar meningiomas often show a forward extension along the dura mater of the anterior cranial fossa and are associated with dilatation (‘blistering’) of the sphenoid sinus. Intracranial extension of optic nerve sheath meningiomas characteristically involves the planum shenoidale. Optic nerve gliomas are astrocytic tumours, which occur in childhood and may involve the optic nerves, optic chiasm and optic tracts (Fig. 56.31). These tumours can be associated with NF 1 but chiasmic tumours are more frequently seen in patients who do not have NF 159.

Figure 56.28 Pituitary apoplexy due to haemorrhage into a pituitary macroadenoma. Coronal (A) and sagittal (B) T1W images demonstrate a hyperintense area at the superior aspect of the tumour that contains a fluid level and it is consistent with a recent intratumoural haemorrhage. The optic chiasm is stretched across the apex of the mass.

1291

1292

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 56.29 Craniopharyngioma. CT following IV contrast medium. There is a partly calcified, partly cystic lesion in the suprasellar region. There is inhomogenous enhancement of the solid tumour components.

Figure 56.30 Suprasellar meningioma. Sagittal T1W post-gadolinium MRI. A lobulated, enhancing suprasellar mass arises from the region of the tuberculum sellae and extends down into the pituitary fossa displacing the pituitary stalk posteriorly. Enhancing dural ‘tails’ (arrowheads) can be seen extending over the planum sphenoidale and clivus.

Figure 56.31 Intra-cranial optic nerve glioma in a patient without NF1. T2W (A) and T1W post-contrast images (B) show an inhomogeneous enhancing mass occupying the chiasmatic cistern, which is inseparable from the optic chiasm. The mass invaginates the left temporal lobe medially and the T2W image (A) shows extension of abnormal signal into the right optic tract.

CHAPTER 56

Metastases, particularly from breast carcinoma, can cause thickening of the pituitary stalk and can present with diabetes insipidus. Similar appearances may be seen in histiocytosis and sarcoidosis.

REFERENCES 1. Hoeffner E G, Case I, Jain R et al 2004 Cerebral perfusion CT: technique and clinical applications. Radiology 231:632–644 2. Bynevelt M, Britton J, Seymour H, MacSweeney E, Thomas N, Sandhu K 2001 FLAIR imaging in the follow-up of low-grade gliomas: time to dispense with the dual-echo? Neuroradiology 43:129–133 3. Yuh W, Maley J 1999 Contrast dosage in the neuroimaging of brain tumors. Principles and indications. MRI Clin N Am 6:113–124 4. Essig M 2006 MR imaging of CNS tumors: are all contrast agents created the same? Neuroradiology 48(Suppl 1):3–8 5. Cha S 2004 Perfusion MR imaging of brain tumors. Top Magn Reson Imaging 15:279–289 6. Cha S 2006 Update on brain tumor imaging: from anatomy to physiology. AJNR Am J Neuroradiol 27:475–487 7. Law M 2004 MR spectroscopy of brain tumors. Top Magn Reson Imaging 15:291–313 8. Provenzale J M, Mukundan S, Barboriak D P 2006 Diffusion-weighted and perfusion MR imaging for brain tumor characterization and assessment of treatment response. Radiology 239:632–649 9. Folkman J 1997 Angiogenesis and angiogenesis inhibition: an overview. Department of Surgery, Children’s Hospital, Boston. E X S 79:1–8 10. Sugahara T, Koroghi Y, Kochi M et al 1998 Correlation of MR imagingdetermined cerebral blood maps with histologic and angiographic determination of vascularity of gliomas. AJR Am J Roentgenol 171:1479–1486 11. Maia A C Jr, Malheiros S M, da Rocha A J et al 2005 MR cerebral blood volume maps correlated with vascular endothelial growth factor expression and tumor grade in nonenhancing gliomas. AJNR Am J Neuroradiol 26:777–783 12. Sugahara T, Korogi Y, Kochi M, Ushio Y, Takahashi M 2001 Perfusionsensitive MR imaging of gliomas: comparison between gradient-echo and spin-echo echo-planar imaging techniques. AJNR Am J Neuroradiol 22:1306–1315 13. Cha S, Yang L, Johnson G et al 2006 Comparison of microvascular permeability measurements, K(trans), determined with conventional steady-state T1-weighted and first-pass T2*-weighted MR imaging methods in gliomas and meningiomas. AJNR Am J Neuroradiol 27:409–417 14. Sugahara T, Korogi Y, Kochi M et al 1999 Usefulness of diffusionweighted MRI with echo-planar technique in the evaluation of cellularity in gliomas. J Magn Reson Imaging 9:53–60 15. Sadeghi N, Camby I, Goldman S et al 2003 Effect of hydrophilic components of the extracellular matrix on quantifiable diffusionweighted imaging of human gliomas: preliminary results of correlating apparent diffusion coefficient values and hyaluronan expression level. AJR Am J Roentgenol 181:235–241 16. Le Bihan D, Mangin J F, Poupon C et al 2001 Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging 13:534–546 17. Field A S, Alexander A L 2004 Diffusion tensor imaging in cerebral tumor diagnosis and therapy. Top Magn Reson Imaging 15:315–324 18. Vlieger E J, Majoie C B, Leenstra S, Den Heeten G J 2004 Functional magnetic resonance imaging for neurosurgical planning in neurooncology. Eur Radiol 14:1143–1153 19. Kleihues P, Burger P, Scheithaur B 1993 The new WHO classification of brain tumours. Brain Pathol 3:255–268 20. Kleihues P, Louis D N, Scheithauer B W et al 2002 The WHO classification of tumors of the nervous system. J Neuropathol Exp Neurol 61:215–225; discussion 226–229 21. Behin A, Hoang-Xuan K, Carpetier A F, Delattre J Y 2003 Primary brain tumours in adults. Lancet 361:323–331

•

BENIGN AND MALIGNANT INTRACRANIAL TUMOURS IN ADULTS

22. Wilms G, Demaerel P, Sunaert S 2005 Intra-axial brain tumours. Eur Radiol 15:468–484 23. Young R J, Knopp E A 2006 Brain MRI: tumor evaluation. J Magn Reson Imaging 24:709–724 24. Patel M R, Tse V 2004 Diagnosis and staging of brain tumors. Semin Roentgenol 39:347–360 25. Nelson S J, Cha S 2003 Imaging glioblastoma multiforme. Cancer J 9:134–145 26. Ricci P 1999 Imaging of adult brain tumors. Neuroimaging Clin North Am 9:651–669 27. White M L, Zhang Y, Kirby P, Ryken T C 2005 Can tumor contrast enhancement be used as a criterion for differentiating tumor grades of oligodendrogliomas? AJNR Am J Neuroradiol 26:784–790 28. Cha S, Tihan T, Crawford F et al 2005 Differentiation of low-grade oligodendrogliomas from low-grade astrocytomas by using quantitative blood-volume measurements derived from dynamic susceptibility contrast-enhanced MR imaging. AJNR Am J Neuroradiol 26:266–273 29. Jenkinson M D, Smith T S, Joyce K A et al 2006 Cerebral blood volume, genotype and chemosensitivity in oligodendroglial tumours. Neuroradiology 48:703–713 30. Jenkinson M D, du Plessis D G, Smith T S, Joyce K A, Warnke P C, Walker C 2006 Histological growth patterns and genotype in oligodendroglial tumours: correlation with MRI features. Brain 129:1884–1891 31. Tozer D J, Jager H R, Danchaivijitr N et al 2006 Apparent diffusion coefficient histograms may predict low-grade glioma subtype. NMR Biomed 20: 49–57 32. Yang D, Korogi T, Sugahara T et al 2002 Cerebral gliomas: prospective comparison of multivoxel 2D chemical-shift imaging proton MR spectroscopy, echoplanar perfusion and diffusion-weighted MRI. Neuroradiol 44:656–666 33. Law M, Yang S, Wang H et al 2003 Glioma grading: sensitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. AJNR Am J Neuroradiol 24:1989–1998 34. Law M, Oh S, Babb J S et al 2006 Low-grade gliomas: dynamic susceptibility-weighted contrast-enhanced perfusion MR imaging – prediction of patient clinical response. Radiology 238:658–667 35. Catalaa I, Henry R, Dillon W P et al 2006 Perfusion, diffusion and spectroscopy values in newly diagnosed cerebral gliomas. NMR Biomed 19:463–475 36. Hein P A, Eskey C J, Dunn J F, Hug E B 2004 Diffusion-weighted imaging in the follow-up of treated high-grade gliomas: tumor recurrence versus radiation injury. AJNR Am J Neuroradiol 25:201–209 37. Chiang I C, Kuo Y T, Lu C Y et al 2004 Distinction between highgrade gliomas and solitary metastases using peritumoral 3-T magnetic resonance spectroscopy, diffusion, and perfusion imagings. Neuroradiology 46:619–627 38. Provenzale J M, McGraw P, Mhatre P, Guo A C, Delong D 2004 Peritumoral brain regions in gliomas and meningiomas: investigation with isotropic diffusion-weighted MR imaging and diffusion-tensor MR imaging. Radiology 232:451–460 39. Stadlbauer A, Ganslandt O, Buslei R et al 2006 Gliomas: histopathologic evaluation of changes in directionality and magnitude of water diffusion at diffusion-tensor MR imaging. Radiology 240:803–810 40. Stanescu Cosson R, Varlet P, Beuvon F et al 2001 Dysembryoplastic neuroepithelial tumors: CT, MR findings and imaging follow-up: a study of 53 cases. J Neuroradiol 28:230–240 41. Tien R D, Barkovich A J, Edwards M S 1990 MR imaging of pineal tumors. AJR Am J Roentgenol 155:143–151 42. Sumida M, Uozumi T, Kiya K et al 1995 MRI of intracranial germ cell tumours. Neuroradiology 37:32–37 43. Guo A C, Cummings T J, Dash R C, Provenzale J M 2002 Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 224:177–183 44. Camacho D L, Smith J K, Castillo M 2003 Differentiation of toxoplasmosis and lymphoma in AIDS patients by using apparent diffusion coefficients. AJNR Am J Neuroradiol 24:633–637

1293

1294

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

45. Young R J, Sills A K, Brem S, Knopp E A 2005 Neuroimaging of metastatic brain disease. Neurosurgery 57:S10–S23; discusssion S11–S14 46. Hayashida Y, Hirai T, Morishita S et al 2006 Diffusion-weighted imaging of metastatic brain tumors: comparison with histologic type and tumor cellularity. AJNR Am J Neuroradiol 27:1419–1425 47. Guzman R, Barth A, Lovblad K O et al 2002 Use of diffusion-weighted magnetic resonance imaging in differentiating purulent brain processes from cystic brain tumors. J Neurosurg 97:1101–1107 48. Zhang D, Wen L, Henning T D et al 2006 Central neurocytoma: clinical, pathological and neuroradiological findings. Clin Radiol 61:348–357 49. Drevelegas A 2005 Extra-axial brain tumors. Eur Radiol 15:453–467 50. Vaz R, Borges N, Cruz C, Azevedo I 1998 Cerebral edema associated with meningiomas: the role of peritumoral brain tissue. J Neuro-oncol 36:285–291 51. Zee C S, Chin T, Segall H D, Destian S, Ahmadi J 1992 Magnetic resonance imaging of meningiomas. Semin Ultrasound CT MR 13:154–169 52. Filippi C G, Edgar M A, Ulug A M, Prowda J C, Heier L A, Zimmerman R D 2001 Appearance of meningiomas on diffusion-weighted images: correlating diffusion constants with histopathologic findings. AJNR Am J Neuroradiol 22:65–72

53. Kremer S, Grand S, Remy C et al 2004 Contribution of dynamic contrast MR imaging to the differentiation between dural metastasis and meningioma. Neuroradiology 46:642–648 54. Chiechi M V, Smirniotopoulos J G, Mena H 1996 Intracranial hemangiopericytomas: MR and CT features. AJNR Am J Neuroradiol 17:1365–1371 55. Simonetta A 1999 Imaging of suprasellar and parasellar tumors. In: Ricci P (ed.) Neuroimaging Clinics of North America. Philadelphia: W B Saunders Company, pp 717–732 56. Bartynski W S, Lin L 1997 Dynamic and conventional spin-echo MR of pituitary microlesions. AJNR Am J Neuroradiol 18:965–972 57. Cottier J, Destrieux C, Brunereau L et al 2000 Cavernous sinus invasion by pituitary adenoma: MR imaging. Radiology 215:463–469 58. Sartoretti-Schefer S, Wichmann W, Aguzzi A, Valavanis A 1997 MR differentiation of adamantinous and squamous-papillary craniopharyngiomas. AJNR Am J Neuroradiol 18:77–87 59. Kornreich L, Blaser S, Schwarz M et al 2001 Optic pathway glioma: correlation of imaging findings with the presence of neurofibromatosis. AJNR Am J Neuroradiol 22:1963–1969

CHAPTER

Cerebrovascular Disease and Nontraumatic Intracranial Haemorrhage

57

Philip M. Rich, H. Rolf Jäger

Cerebral ischaemia • Imaging of cerebral ischaemia Nontraumatic intracranial haemorrhage • Subarachnoid haemorrhage • Cerebral aneurysms • Intracerebral haemorrhage • Arteriovenous malformations Subdural and extradural haemorrhage

There have been exciting recent advances in cerebrovascular imaging, particularly for acute ischaemic stroke and

noninvasive angiography. This has been driven by the desire to rapidly identify candidates for thrombolysis for whom ‘time is brain’, and a generally more interventional approach to stroke management, which seems to be associated with better outcomes. This chapter begins with a discussion of ischaemic stroke, including imaging strategies in the hyperacute stage, when thrombolysis may be appropriate. Subsequent sections cover spontaneous intracranial haemorrhage, aneurysms and vascular malformations. Traumatic intracranial haemorrhage is discussed elsewhere.

CEREBRAL ISCHAEMIA Stroke is the clinical syndrome of a sudden neurological deficit of vascular origin. Ischaemic and haemorrhagic stroke cannot be distinguished clinically and around 30 per cent of patients presenting with a stroke-like episode have a nonvascular cause1, hence the importance of diagnostic imaging. After exclusion of subarachnoid haemorrhage (SAH), ischaemia accounts for 85 per cent of strokes and spontaneous intracranial haemorrhage for 15 per cent2. Ischaemic stroke may be caused by atherothrombotic arterial occlusion, or embolism from carotid or vertebral artery atheroma or the heart. Cervical arterial dissection, vasculitis, venous thrombosis and substance abuse also cause stroke and rarely in adults stroke-like episodes are due to metabolic disorders such as mitochondrial cytopathies. A transient ischaemic attack (TIA) by definition resolves within 24 h. This includes amaurosis fugax, a transient loss of vision in one eye. The risk of stroke following a TIA is higher than previously thought, maybe up to 8 per cent in the first week and 12 per cent within a month3, and even more in those awaiting endarterectomy for a symptomatic carotid stenosis4. These figures do not allow for mild TIAs that go unrecognized

but even so a recent TIA should now prompt urgent investigation for treatable risk factors including cervical arterial stenoses. Implementation of such a policy has major resource implications but this may be offset if strokes are prevented. The role of stroke imaging has shifted away from merely excluding haemorrhage and nonvascular causes and towards the identification of early ischaemia and the differentiation of viable from irreversibly damaged tissue: the concept of the ischaemic penumbra, which will be discussed later. Advanced magnetic resonance imaging (MRI) techniques have been the cornerstone of this approach, particularly perfusion and diffusion-weighted imaging (DWI) but the advent of multi detector computed tomography (MDCT) provides a reasonable alternative5. The rapidly enlarging literature on advanced techniques should not distract us from the importance of properly performed and interpreted basic imaging. Unenhanced brain CT remains the workhorse of stroke imaging in most institutions and its enduring value should not be diminished, particularly because it is readily available and easy to use in critically ill patients. The additional value of some newer techniques

1296

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

has yet to be proven in randomized trials. This may change as more is learnt about how imaging can be used to predict outcome from strokes and treatment. It seems likely this will come via MRI, which has a greater inherent capacity than CT to measure different tissue parameters.

IMAGING OF CEREBRAL ISCHAEMIA Pathophysiological considerations6–8 Angiography and perfusion studies show flow abnormalities immediately after onset. Cerebral autoregulation responds to a fall in cerebral perfusion pressure (CPP) with vasodilatation and recruitment of collateral vessels, thus increasing cerebral blood volume (CBV) and reducing resistance, in order to maintain cerebral blood flow (CBF). After the vessels are fully dilated the autoregulatory system cannot properly respond to any further reduction in CPP and therefore both CBF and CBV start to fall. Oxygen extraction goes up to compensate, but once this is maximal any further fall in CBF causes cellular dysfunction. A CBF of around 23 ml 100 g–1 min–1 causes a reversible neurological deficit. Electrical activity ceases below about 18–20 ml 100 g–1 min–1. Such tissue remains viable but infarction is likely if perfusion is not restored. At 10–15 ml 100 g–1 min–1, failure of energy-dependent membrane pumps leads to a shift of sodium and water into cells causing swelling of neurons and cytotoxic oedema, shown earliest on MR diffusion-weighted imaging (DWI) as restriction of water diffusion, but it can also produce subtle signs on CT. This degree of perfusion compromise results in an infarct much more rapidly. In time, structural breakdown of the blood–brain barrier occurs due to ischaemic damage to capillary endothelium. Leakage of intravascular fluid and protein into the extracellular space and later net influx of water to the infarcted area cause vasogenic oedema. Clinical outcome depends on the arterial territory involved and the adequacy of collateral blood supply9. Proximal occlusions cause infarcts of an entire arterial territory unless sufficient collateral circulation exists, in which case there may be deep infarcts with sparing of overlying cortex. Emboli or

Figure 57.1 Large MCA territory infarct on CT. (A) Acute left middle cerebral artery thrombus appears dense (arrows). (B). Grey-white differentiation is lost around the left putamen and insular cortex (white arrow) and there is mild effacement of the Sylvian fissure (black arrow), early signs of an MCA territory infarct. The ‘insula ribbon’ and lentiform nucleus appear normal on the right. (c) Three days later there is malignant oedema and haemorrhagic transformation (higher density areas).

occlusion of terminal branches cause cortical infarcts. Perforator vessel occlusions cause lacunar infarcts (see small vessel disease later). Most carotid territory infarcts involve the middle cerebral artery. Anterior cerebral artery (ACA) collateral flow is generally excellent and emboli relatively rare.The commonest cause of ACA infarcts is vasospasm following subarachnoid haemorrhage.The basilar artery supplies the posterior cerebral arteries unless the posterior communicating artery is large, in which case emboli from the carotid circulation may enter their territory. Brainstem infarcts are commonly due to occlusion of short perforating vessels. A combination of infratentorial, thalamic and occipital infarcts suggests an occlusion of distal basilar artery, or ‘top of the basilar’ syndrome10. Multiple infarcts in different arterial territories suggest a cardiac rather than a carotid source of emboli, or haemodynamic strokes due to hypotension if the distribution conforms to the arterial border zones.

Imaging This account is intentionally directed towards very early stroke imaging and thrombolysis, reflecting the many recent publications on this subject. It is acknowledged that the great majority of stroke patients are not managed in this way, and many never have any imaging other than unenhanced brain CT. However the fundamental concepts are applicable to all stroke patients.

Structural imaging A dense artery is the earliest detectable change on CT. As it is caused by fresh thrombus occluding the vessel it can be seen at the onset of the ictus (Fig. 57.1). Thrombus may rapidly disperse so this sign is not always present. When found in the proximal MCA it correlates with large infarcts, although it has a better prognosis if limited to an MCA branch within the Sylvian fissure11 (Fig. 57.2). MCA calcification can mimic this sign but is often bilateral. The basilar artery may also appear dense in the case of posterior circulation infarcts particularly the ‘top of basilar’ syndrome (Fig. 57.3). Thrombus may cause loss of a normal arterial flow void on MRI. However arterial high signal may be seen in a patent vessel on FLAIR MRI due to altered flow, a useful qualitative sign of reduced perfusion when the parenchyma usually still

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.2 Sylvian ‘dot’ sign. (A) CT shows dense MCA branch due to occlusive acute thrombus (short arrow). There is very subtle loss of greywhite differentiation between the insular cortex and lateral border of putamen posteriorly (arrowhead), whereas more anteriorly it is preserved (long arrow). (B) DWI confirms a small acute cortical infarct adjacent to the thrombosed vessel.

appears normal12(Fig. 57.4A). Intravascular enhancement due to sluggish flow in affected vessels may be seen on contrastenhanced CT and MRI in the first few days after an infarct, becoming less obvious towards the end of the first week13 (Fig. 57.4B). The early parenchymal signs on CT are reduced grey matter density and brain swelling, manifest as effacement of sulci. These changes are traditionally thought to reflect cytotoxic oedema, which reduces the Hounsfield number of grey matter so it is indistinguishable from adjacent white matter. In early MCA infarcts this causes a reduction in clarity of the margins of the lentiform nucleus and cortex, particularly in the insula (Figs 57.1, 57.2).

The equivalent MRI signs are cortical swelling and T1/T2 prolongation, more obvious on T2-weighted sequences, particularly FLAIR. T2 hyperintensity is often absent or very subtle in infarcts only a few hours old14, suggesting that early parenchymal low density on CT may be due to abnormal perfusion (reduced CBV) rather than oedema5. Furthermore brain swelling on CT without accompanying low density does not always progress to infarction. Such cases may also be due to abnormal perfusion, but a compensatory increase in CBV rather than a reduction15. Thus whilst it is generally accepted that swelling with obvious low density on CT is an indication of infarction, perhaps subtle low density or swelling without low density are sometimes signs of compromised perfusion that may be reversible, particularly the latter (see earlier). Hypodensity on early CT examinations affecting more than 50 per cent of the MCA territory is associated with a high mortality rate16. Greater than one-third involvement is commonly a contraindication to thrombolysis (although the increased mortality in this group does not reach statistical significance17). The subtle early CT signs of ischaemia are easily overlooked and it can be difficult to accurately estimate percentage involvement of a vascular territory, particularly for less experienced observers. The sensitivity of CT for infarcts has been reported to be only 30 per cent at 3 h18 and 60 per cent at 24 h. A post hoc expert review of CT studies in the European Cooperative Acute Stroke Study (ECASS) showed that the initial ‘on site’ interpretation had overlooked an early infarct in 11 per cent of the patients19. These difficulties led to the development of the Alberta Stroke Program Early CT Score (ASPECTS)20. In this simple rating scale the affected middle cerebral artery territory is divided into ten segments, namely: internal capsule, caudate nucleus, lentiform nucleus, insula and six segments for cortical areas (Fig. 57.5). One point is lost for each area that shows early ischaemic changes of swelling or reduced attenuation, a lower total score carrying a worse prognosis. ASPECTS

Figure 57.3 ‘Top of the basilar’ syndrome. T2-weighted MRI shows multiple infarcts in the basilar and posterior cerebral artery territories including the left thalamus (A), both occipital lobes (B), and cerebellar hemispheres (C). Note the absence of flow void in the distal basilar artery in B (arrow).

1297

1298

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.4 Vascular signs in acute infarcts. (A) FLAIR axial image of an acute right striatocapsular infarct. Note asymmetrical high signal returned from patent right MCA cortical branches due to compromised flow (arrows). (B) Contrast-enhanced CT shows early infarct changes of mild low density and local swelling in posterior part of right MCA territory. There is asymmetrical enhancement of MCA cortical branches because of sluggish flow (arrows). Figure 57.5 ASPECTS for early ischaemic change on CT (from Barber P A, Demchuk A M, Zhang J et al 2000 Validity and reliability of a quantitative computed tomography score in predicting outcome of hyperacute stroke before thrombolytic therapy. Lancet 355:1670– 1674 with permission from Elsevier). A = anterior circulation; P = posterior circulation; C = caudate; L = lentiform; IC =internal capsule; I = insular ribbon; M1 = anterior middle cerebral artery (MCA) cortex; M2 = MCA cortex lateral to insular ribbon; M3 = posterior MCA cortex; M4, M5 and M6 are anterior, lateral and posterior MCA territories immediately superior to M1, M2 and M3, rostral to basal ganglia. Subcortical structures are alotted 3 points (C, L and IC), MCA cortex is alotted 7 points (insular cortex, M1, M2, M3, M4, M5 and M6). A point is lost for each area that shows early ischaemic change (low density or swelling). The range of scores is 0–10, representing an infarct of the entire territory and normal findings, respectively.

can be used to predict outcome and risk of post-thrombolysis haemorrhage20. It correlates well with DWI findings at presentation21 and facilitates more accurate interpretation of emergency CT by nonexperts22. Even in patients not suitable for thrombolysis it seems intuitive that a methodical approach such as ASPECTS is likely to increase accuracy of CT interpretation, at least for supratentorial events. Brainstem infarcts are notoriously difficult to detect by CT at any stage after onset, even using the most modern CT equipment. In the subacute phase there is structural breakdown and blood–brain barrier disruption. Fluid leaks into the extracellular space causing low attenuation on CT and T2 hyperinten-

A

A

M4

M1 C L IC

I M5

M2 M3

M6 P

P

sity on MRI that involves both grey and white matter in large infarcts. The severity and duration of brain swelling depends on infarct size. It usually increases during the first week, persists during the second week and then regresses. Other diagnoses such as tumour or infection should be considered if there is extensive white matter oedema without cortical involvement or prolonged brain swelling. Contrast enhancement on CT and MR due to blood–brain barrier disruption is common in the subacute stage; indeed on MRI it occurs in almost all cases by the end of the first week13 and persists for several months. The pattern is variable and therefore not always specific, however gyriform enhancement,

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

if present, is most characteristic of a cortical infarct. Lack of enhancement of large cortical lesions on MRI suggests alternative diagnoses such as low-grade glioma. CT has a much higher sensitivity than spin-echo MRI for the detection of acute haemorrhage, the main differential diagnosis of ischaemic stroke. However the T2*-weighted gradient echo sequence has equivalent sensitivity or better than CT. Gradient echo imaging also detects the haemosiderin that can persist for years following a haemorrhage23–25. Establishing the diagnosis of an old haemorrhage is not possible with CT, since blood clots become hypodense within a few weeks and old haemorrhagic cavities can resemble infarcts. Haemorrhagic transformation due to secondary bleeding into reperfused ischaemic tissue occurs during the first 2 weeks. It is shown in up to 80 per cent of infarcts on MRI26, appearing hyperintense on T1-weighted and hypointense on T2-weighted images. It is often seen in the basal ganglia and cortex, where it can assume a gyriform pattern (Fig. 57.6 and see Fig. 57.1C). The occurrence and severity of haemorrhagic transformation correlates with the size of the infarct and degree of contrast enhancement in the early stage26. The final stage of an infarct is encephalomalacia and atrophy, causing enlargement of adjacent sulci and ventricles. The density on CT and signal intensity on MRI approaches that of CSF. Wallerian degeneration is sometimes visible as faint T2 hyperintensity in the ipsilateral corticospinal tract and asymmetrical brainstem atrophy. Proton magnetic resonance spectroscopy (MRS) of cerebral infarcts shows the appearance of lactate, while Nacetylaspartate (NAA), a neuronal marker, and total creatine are reduced compared to the contralateral hemisphere (Fig. 57.7). Longitudinal studies have shown progressive reduction of NAA suggesting that ischaemic injury continues for more than a week following infarction27. However since the advent of DWI, MRS is not required to diagnose an acute infarct. Furthermore it requires the subject to keep still for several minutes, which is not always possible in acute stroke patients who may be confused and restless.

Vascular imaging Catheter angiography has been the traditional way of investigating intracranial vessel occlusions and collateral pathways and is still used for intra-arterial thrombolysis. Noninvasive angiography by US, CTA or MRA is preferable for other patients. CTA can be performed as part of the same examination as perfusion imaging with separate IV infusions of contrast medium. No delay is necessary between the two acquisitions. A three-dimensional (3D) time-of-flight technique is preferred for MR angiography.

Advanced techniques—perfusion and diffusion imaging Currently perfusion imaging for acute stroke patients is most conveniently achieved using dynamic bolus tracking techniques, either perfusion CT (CTP) or perfusion-weighted MRI (PW-MRI)28,29. Unlike other techniques the perfusion data is acquired in under a minute on the same machine (CT or MR) used to acquire diagnostic brain images. Other modalities used for perfusion imaging clinically or in research include xenon CT, arterial spin labelling MRI, transcranial Doppler US, single-photon emission CT (SPECT) and positron emission tomography (PET). Their various strengths and weaknesses are explored in a recent review28. PW-MRI produces maps of time-to-peak contrast (TTP), mean transit time (MTT), CBV and CBF (CBF=CBV/MTT). TTP provides a qualitative overview of brain perfusion. A threshold of 4 s delay seems to indicate tissue at risk and correlates with a CBF of under 20 ml 100 g–1 min–1 30. However proximal vessel stenosis can delay TTP even if CBF via collaterals is normal and tissue viability not threatened. As outlined earlier, within an area of prolonged MTT (or TTP), moderate ischaemia may cause increased CBV, however reduced CBV indicates inadequate collateral supply and high risk of infarction. On perfusion imaging a CBV deficit seems to be the best predictor of initial infarct size (and final size if successfully reperfused).The MTT and CBF indicate tissue at risk5, in other words the final infarct volume unless reperfusion occurs,

Figure 57.6 Haemorrhagic transformation. (A) Unenhanced brain CT 2 weeks after a large right MCA territory infarct shows a gyriform pattern of haemorrhagic transformation in right cerebral cortex. There is also haemorrhage in the basal ganglia (arrows). (B) Coronal T1-weighted image shows swelling and signal alteration of the caudate and lentiform nuclei. Sparing of the cortex is due to adequate leptomeningeal collateral circulation. The central T1-hyperintense area within the infarct indicates haemorrhagic transformation.

1299

1300

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Lac Cr/Pcr

NAA

mI Chol Glx Glx

4

B

Figure 57.7 MR spectroscopy in acute stroke. (A) Axial T2-fast spin-echo image (TR/TE = 3500/95) of a 55-year-old male stroke patient at 14 h from stroke onset, to show the positions of voxels within the infarct and contralateral hemisphere. (B) Proton spectrum (TR 2000/TE 30 ms) acquired at time of presentation on a 1.5 T MR system. The total creatine (3.03 p.p.m.) and choline (3.22 p.p.m.) are reduced, the NAA peak (2.01 p.p.m.) is almost absent and there is a large lactate doublet at 1.33 p.p.m. compared to the contralateral hemisphere. (C) Normal spectrum acquired from the contralateral hemisphere. Resonance peaks are lactate (Lac), N-acetyl aspartate (NAA), creatine (Cr/Pcr), myoinositol (ml), and glutamate and glutamine (Glx).

Cr/Pcr

Chol

Glx

4

C

1

ppm

NAA

mI

2

3

Glx

3

2

1

ppm

either naturally or following medical intervention. CBV in the infarcted region increases in the subacute stage31, often associated with vessel recanalization, followed by a decrease later (Fig. 57.8). Absolute CBF levels may allow more accurate prognostication and treatment targeting, however quantitative data is not readily obtained from PW-MRI.This does not seem to be a major problem in acute stroke, MRI usually adequately defining hypoperfused brain relative to a normal contralateral hemisphere. However it may have implications if there are bilateral or multifocal deficits, for example in bilateral severe carotid stenoses or vasospasm following subarachnoid haemorrhage. Calculated arterial input functions may make MR data more quantitative in due course. Quantitative perfusion data can be obtained from CTP or arterial spin-labelled MRI techniques, although at the moment the latter takes longer to acquire than contrast-enhanced perfusion MRI. The whole brain is covered on PW-MRI whereas CTP only allows partial coverage, although this will increase with more modern MDCT systems. However this technical limitation does not seem to significantly affect the practical application of CTP in most cases.

Perfusion imaging, including PET, is also used for elective assessment of haemodynamic reserve and stroke risk (Fig. 57.9). For example perfusion may be normal at rest despite a significant carotid stenosis but show reduced blood flow following acetazolamide challenge, which is the reverse of normal32,33. SPECT with 99mTc HMPAO or ECD will show a perfusion defect as soon as vascular occlusion occurs, although care must be taken in interpretation of HMPAO SPECT studies 10 days or more after the onset of stroke due to hyperfixation of the radiopharmaceutical in infarcted tissue34. Quantitation of the degree of ischaemia using HMPAO SPECT will predict risk of intracranial haemorrhage following intra-arterial thrombolysis35. DWI has a pre-eminent role in acute stroke imaging due to its high sensitivity in the first few hours after infarction, when T2-weighted images are usually normal. DWI should be interpreted in conjunction with an apparent diffusion coefficient (ADC) map, which is derived from the DWI data and in the clinical environment is displayed as a grey scale ‘image’ for ease of use (Fig. 57.10). Restricted diffusion in acute infarcts (low ADC) returns high signal on DWI and appears dark on the ADC map. After 5–14 d, loss of structural integrity results in

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.8 Ischaemic penumbra on CT perfusion imaging and thrombolysis. (A) CT 5 hours after onset of right hemiplegia and aphasia shows subtle low density in left lentiform nucleus (arrow). (B) MTT map shows large deficit corresponding to entire left MCA territory. (C). CBV map shows almost symmetrical CBV apart from a dark-appearing deficit in left basal ganglia region (arrow). The basal ganglia deficit represents the infarcted core. The surrounding area of MTT deficit is the potentially salvageable ischaemic penumbra, known as the area of mismatch. (D) CTA shows proximal occlusion of left MCA (arrow). This did not appear dense on the admission CT. (E) 24 hours post intra-arterial thrombolysis the patient had made an almost complete functional recovery. An axial T2 weighted image shows a basal ganglia infarct but preserved cortex. (F) Repeated CBV map shows local increase in CBV (arrow). (G) MRA confirms patency of left MCA (short arrow). Note asymmetrical prominence of left lenticulostriate perforators (long arrow) supplying area of increased perfusion following ischaemia.

increased water mobility and the imaging appearance reverses to low signal on DWI and bright on the ADC map36,37. Between the two states is a transition period of DWI ‘pseudonormalization’, during which small infarcts can be masked. Larger lesions will still be obvious on structural images. Prolonged restriction of diffusion in small white matter infarcts

lasting several weeks has been reported, the explanation for which is not entirely clear38. Chronic lesions with very long T2 relaxation times may appear high signal on DWI due to ‘T2 shine through’, but in comparison to acute infarcts they will also appear bright on the ADC map. Another potential pitfall of DWI is acute haemorrhage,

1301

1302

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.9 Autoregulatory response to chronic ischaemia on CT perfusion imaging. (A) CEMRA shows right internal carotid artery (ICA) occluded from its origin (long arrow). There is intracranial reconstitution via the circle of Willis (short arrow). Note the entire arterial tree from the aortic arch to the circle of Willis is shown on a single examination. (B) MTT is prolonged throughout right ICA territory. (C) CBV appears symmetrical. Measured regions of interest showed slight relative increase in the right frontal lobe due to opening of collateral pathways and vasodilatation responding to a reduction in perfusion pressure. (D) CT shows small mature watershed infarcts (arrows) but the majority of the right ICA territory is preserved.

Figure 57.10 New and old infarcts on DWI. (A) FLAIR shows high signal ischaemic change of indeterminate age in both frontal lobes. There is a small mature lacunar infarct on the right with a low signal central cavity (arrow). (B) DWI shows an acute high signal lesion on the left. (C) The ADC map confirms the left-sided lesion is acute (dark indicates low ADC; long arrow). The infarcts on the right are bright, indicating increased ADC and therefore older lesions (short arrows).

which can return high signal resembling an infarct. However there is often also a low signal margin produced by susceptibility effects39. Analysis of other sequences should indicate the correct diagnosis. The CBV deficit on perfusion CT and extent of low density on CT angiogram source images (CTA-SI) both correlate sufficiently well with DWI lesion size to make them practical CT surrogates for DWI40 (Fig. 57.8). The explanation for low density on CTA-SI may lie in a reduced CBV rather than cytotoxic oedema (see earlier)17.

Imaging strategies in acute stroke Only neuroimaging can accurately identify patients with haemorrhage or a nonvascular diagnosis. Exclusion of haemorrhage allows aspirin to be prescribed, which is beneficial if commenced within 48 h41. In principle it seems preferable that either CT or MR (usually CT) is performed at the time of admission or as soon as possible afterwards so the correct diagnosis is established. Economic as well as medical arguments have been advanced to justify this approach42,43 and it is accepted practice in many countries, although implementing

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

such a policy where it is not already in place has significant implications for local resources and work patterns. Patients on anticoagulation should have urgent imaging so it can be reversed if they have had a haemorrhagic stroke. Other reasons for urgent neuroimaging include: a history of immunosuppression or malignancy as such patients may have a cerebral infection or tumour; depressed conscious level; and a suspected posterior cranial fossa haematoma that may require surgical evacuation. CT is usually the first-line investigation, being more widely available outside routine working hours and generally regarded as a ‘simpler’ technique, although in fact early infarcts are much easier to detect on DWI, particularly for the inexperienced observer. Other advantages of DWI over CT include the demonstration of a new infarct on a background

of chronic ischaemic damage (Figs 57.10, 57.11) and importantly the correct diagnosis of a transient ischaemic attack, which indicates a high short-term risk of stroke. Abnormalities may be shown on DWI following a TIA, particularly if there is a motor deficit, aphasia or an event lasting more than 1 h, although lesions may be transient44. The presence of a lesion confirms a cerebrovascular substrate and the location indicates the vascular territory involved. In general MRI more clearly shows the extent of ischaemic change, the anatomical distribution and to some degree allows a better estimation of type of stroke and age of different lesions (large or small vessel, embolic, hypoperfusion). Cervical and sometimes intracranial angiography is often important after stroke and in most circumstances conventional catheter angiography has been replaced by noninvasive

Figure 57.11 Small-vessel disease. (A) On CT there is diffuse cerebral white matter low density (leukoaraiosis) indicating small vessel disease. There is an acute haematoma in the left lateral ventricle with a small blood level in the dependent part of the right occipital horn. (B) The acute intraventricular blood returns low signal on an axial T2-weighted image. There are high signal foci due to small infarcts and widened perivascular spaces in the thalami and basal ganglia. (C) Axial gradient echo image shows the acute intraventricular blood as low signal. There are also numerous low signal foci in the deep nuclei and cerebral hemispheres. These are old areas of haemosiderin staining known as microbleeds. Note they are not visible on CT. (D). Axial FLAIR of a different patient shows diffuse high signal indicating small vessel ischaemic change in deep and periventricular cerebral white matter. There is a mature lacunar infarct in the right parietal lobe (arrow). (E). Axial DWI shows an acute infarct as high signal (short arrow) and the mature lacune as low signal (long arrow). Only DWI can differentiate the acute infarct from the surrounding signal abnormality.

1303

1304

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

methods. Doppler ultrasonography is simple to perform at the bedside, although it is operator dependent and does not clearly show arterial segments that are anatomically inaccessible or heavily calcified. CTA or MRA allows a full survey including the aortic arch and cervical vessels, circle of Willis and major intracranial arteries. CTA is quicker to perform than MRA but requires more time at the workstation to properly interpret. These techniques will be discussed further later. The concept of the ischaemic penumbra is central to under-standing current thinking on the interrelationship between early stroke pathophysiology, treatment and imaging (Fig. 57.8). In the hyperacute stage there may be a discrepancy in the size of lesions shown separately on DWI and perfusion imaging. In this model the ischaemic zone consists of two areas: the core of restricted diffusion (or CT CBV deficit) already infarcted surrounded by an underperfused penumbra representing salvageable ischaemic tissue liable to infarct without early spontaneous or therapeutic recanalization. If the penumbra is larger than the core the difference between the two is the area of mismatch. A matched defect (i.e. the two areas coincide) is assumed to indicate a completed infarct without areas of reversibility. This view has been complicated by a realization that restricted diffusion may be reversible in the earliest period after onset36,45,46 and the penumbral perfusion defect may include areas of benign oligaemia as well as tissue at risk47. It appears likely that diffusion can already be impaired if the blood flow lies between 10–20 ml 100 g–1 min–1, above the threshold for irreversible ischaemia. Interestingly this implies low density on CT may sometimes be more specific for infarction than DWI signal change assessed visually, although quantifying the ADC increases specificity for infarcts in patients with strokes or TIAs48. Relatively few stroke patients are currently offered thrombolysis. Acute stroke imaging protocols are designed to rapidly obtain the necessary anatomical, vascular and functional information, using CT (unenhanced CT brain, perfusion imaging and CTA) or MRI (usually with scout, FLAIR, gradient-echo, DWI, intracranial MRA and perfusion imaging). Treatment options include IV or intra-arterial thrombolysis and mechanical clot disruption. Management strategies vary but frequently IV thrombolysis is given within 3 h of symptom onset as soon as haemorrhage or an obvious large infarct has been excluded. Because early DWI lesions may reverse, some thrombolysis centres treat patients presenting within 3 h, even if perfusion imaging shows a matched defect. Intra-arterial thrombolysis or mechanical disruption may be attempted if the patient does not rapidly improve and a proximal vessel occlusion is shown angiographically or as a dense artery on CT. Beyond 3 h restricted diffusion is less likely to be reversible so thrombolysis is usually only appropriate if there is mismatch and many centres opt for primary intra-arterial treatment if there is also a proximal vessel occlusion. There is little randomized evidence to support some of these therapeutic manoeuvres and they are performed at the risk of provoking catastrophic intracranial haemorrhage. The

published literature includes arguments for and against the use of CT or MRI and various treatment strategies, although in practice much still depends on local scanner access and available expertise5,8,17,29,49. It is noteworthy that the US Food and Drug Administration’s approval for the use of recombinant tissue plasminogen activator in acute ischaemic stroke was based on a trial that selected patients with unenhanced CT alone50. A recent UK Cochrane Collaboration review acknowledged the net benefit of thrombolysis in acute stroke but in somewhat guarded terms with emphasis on the risk of haemorrhage and suggested that further randomized trials were required51. Hopefully further research and experience with advanced imaging will refine patient selection for this important but hazardous treatment.

Other patterns of cerebrovascular disease Small vessel ischaemic disease (Fig. 57.11) Cerebral white matter small vessel disease is the norm in older people. More than 95 per cent of those over 65 years of age have white matter lesions on MRI, usually limited in extent52. More severe ischaemic damage is associated with cognitive impairment and gait disturbance. Risk factors include age, hypertension and elevated glycated haemoglobin levels52,53. The basal ganglia are often also involved. Arterioles of the long penetrating arteries become occluded, the outcome probably depending on vessel size. Occlusion of a large vessel causes a lacunar infarct; blockage of smaller arterioles results in ischaemic demyelination and gliosis. Although highly variable in extent, it affects predominantly the periventricular and deep cerebral white matter, basal ganglia and ventral pons. CT shows white matter hypodensities (‘leukoaraiosis’) but MRI, particularly FLAIR, is much more sensitive. There may be cavities indicating lacunar infarcts. DWI will confirm the site of an acute subcortical infarct even if there is widespread pre-existing ischaemic change. New infarcts develop every few months in small vessel disease and are clinically silent unless they arise in eloquent areas, although they will be shown on fortuitously timed DWI54. It is important to realize that these white matter changes are nonspecific and similar appearances may be encountered in vasculitis55. Foci of old haemorrhage known as microbleeds occur with small vessel disease. They are common in ischaemic stroke patients, occurring in up to 26 per cent56, but they are also present in around 6 per cent of asymptomatic older people, associated with age, hypertension and radiological extent of small vessel disease57. They are a marker of vascular fragility in hypertensive small vessel disease, their distribution mirroring symptomatic haemorrhages58,59. At present it is not clear if they should be a contraindication for anticoagulant or antiplatelet therapy, or whether they increase the risk of post thrombolysis haemorrhage60,61. The susceptibility effect of haemosiderin reveals their presence on T2-weighted gradient-echo MRI, spin-echo sequences being much less sensitive. Microbleeds are not visible on CT.

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Global cerebral hypoperfusion and anoxia (Fig. 57.12) Inadequate oxygen supply to the entire brain can be the consequence of severe hypotension or impaired blood oxygenation. Global hypoperfusion can result in watershed infarcts, shown as wedge-shaped lesions in the frontal and/or parietal lobes at the margins of anterior and middle and middle and posterior cerebral artery territories, respectively. DWI can be especially useful, as the early signs on CT or MRI are rendered even subtler by their frequently symmetrical distribution. Anoxia due to defective blood oxygenation such as in carbon monoxide poisoning tends to cause infarcts in sensitive regions, for example the basal ganglia.

Venous infarcts Causes of cerebral venous thrombosis include trauma, infection (particularly subdural empyema) and hypercoagulability disorders including those due to oral contraceptives. Venous infarcts do not conform to arterial territories and are often haemorrhagic and multifocal. The superior sagittal sinus is most commonly involved, which can lead to bilateral parasagittal infarcts. Isolated occlusion of the transverse sinus or the deep cerebral veins can occur (Fig. 57.13). On unenhanced CT acute thrombosis will cause a venous sinus to appear expanded and hyperdense. IV contrast medium causes more intense enhancement of the walls of the sinuses than of their contents, the so-called ‘delta sign’ (Fig. 57.14). MRI shows lack of flow void in the affected sinuses but normal slow flowing blood may appear bright, simulating a thrombus and acute thrombus can appear hypointense on spin-echo images mimicking a flow void (Fig. 57.13). Therefore structural MRI is not always reliable for the exclusion of venous occlusion, particularly if chronic, and a phase-contrast MR venogram (MRV) is usually acquired, which depicts only flow and not thrombus. A lowvelocity encoding (VENC) is chosen to demonstrate venous flow. CT venography is a satisfactory alternative if MRV is unavailable or equivocal62. Using a combination of structural

images, MR and CT venography, there is very little need to resort to DSA to confirm the diagnosis of venous thrombosis.

Other pathology A variety of conditions can mimic infarcts, including tumours, inflammatory conditions, in particular cerebritis or encephalitis and sometimes an early presentation of multifocal leukoencephalopathy. Knowledge of the distribution of vascular territories and attention to clinical details and radiological signs are helpful in distinguishing these conditions from an infarct. In many cases modern imaging and expert neuroradiological review will provide the answer. Follow-up imaging is sometimes useful and variance from the usual pattern of infarct evolution should prompt a review of the diagnosis.

Angiography in ischaemic stroke Most vascular studies in patients with ischaemic stroke or TIAs are carried out for secondary prevention of strokes and primarily concern the extracranial vessels. Intracranial arterial imaging may be obtained in the acute phase if thrombolysis is an option (see earlier). Intracranial studies may also be appropriate in black and oriental stroke patients, who have a higher incidence of intracranial arterial stenosis than white patients, the reverse being the case for extracranial disease63.

Imaging of extracranial vessels (Fig. 57.16) Atheroma can occur at vessel origins (including vertebral arteries), at the carotid bifurcation, and in the distal course of internal carotid or vertebral arteries. The carotid bifurcation is the commonest site. The North America Symptomatic Carotid Endarterectomy Trial64 and European Carotid Surgery Trial65 found patients with symptomatic 70–99 per cent stenosis of the internal carotid artery benefited from surgery. Stenosis measurements in these trials were performed on conventional catheter angiograms, using slightly different methods,

Figure 57.12 Global hypoperfusion. (A) Acute watershed infarcts and diffuse brain swelling on FLAIR axial image after a cardiac arrest (arrows). (B) FLAIR axial image in a different patient with a generalized hypoxic ischaemic brain insult after self-hanging shows diffuse high signal that could be mistaken for normal because of the perfectly symmetrical appearance. (C) DWI shows diffuse grey matter restricted diffusion markedly different from white matter signal intensity, alerting the observer that the scan is abnormal despite its symmetry. Obvious grey-white differentiation is not a normal feature of DWI (see Figs 57.2B, 57.11E).

1305

1306

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.13 Acute deep venous system thrombosis. (A). CT in a young woman presenting obtunded postpartum. The internal cerebral veins are dense and expanded (short arrow). The straight sinus is also hyperdense (long arrow) whereas the superior sagittal sinus is normal in appearance (black arrow). The surrounding thalami and basal ganglia appear as low density. There is mild hydrocephalus due to a combination of thalamic swelling and impaired CSF absorption. (B) Axial T2 confirms oedema in deep venous territory. Note thrombosed internal cerebral veins return low signal mimicking flow void, due to deoxyhaemoglobin in acute thrombus (arrows). (C). A phase contrast MR venogram shows a normal superior sagittal sinus (short arrow) but no flow in the deep venous system (position indicated by long arrow, see also Fig. 57.13E). (D) 1 week later the patient recovered and a repeat T2 axial image shows almost complete resolution of oedema. The ventricles are also reduced in size. (E). Sagittal T1-weighted image shows high signal thrombus in the internal cerebral vein, vein of Galen and straight sinus (arrows).

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.14 Delta sign in superior sagittal sinus thrombosis. (A) Young man with a seizure after several days of headaches. An unenhanced CT shows a small right parietal acute haematoma. The superior sagittal sinus is occluded by acute thrombus. It appears expanded and the same density as the intraparenchymal clot (arrow). (B) Contrast-enhanced CT shows the dura around the sinus (long arrow) is higher density than the lumen (short arrow). This is the ‘delta sign’, although in acute sinus thrombosis the unenhanced CT alone may indicate the correct diagnosis, as in this case. (C) T2weighted MRI shows the recent haematoma. Note the superior sagittal sinus returns intermediate signal rather than flow void (arrow).

which are illustrated in Fig. 57.15. Surgery also reduces the risk of stroke in asymptomatic carotid stenosis of 70 per cent or more as measured by US66. Symptomatic 50–69 per cent stenoses may be a suitable target for intervention but in both cases the benefits are smaller and the advantages of surgery could more easily be outweighed by poor patient selection or excess morbidity from surgery (or angiography) in comparison with trial centres67. Carotid intervention should not be considered for a stenosis of less than 50 per cent regardless of symptoms. Since these trial results were published there has been tremendous progress in noninvasive vascular imaging, a great attraction of which, apart from ease of use, is avoidance of the small but definite risk of stroke associated with catheter angiography, which for steno-occlusive cerebrovascular disease approaches 1 per cent in most published series68.

b ICA

a

ECA c

a= b= c= d=

minimum residual lumen distal internal carotid lumen original internal carotid lume common carotid lumen

CCA d

Figure 57.15 Different ways of measuring percentage of carotid artery stenosis (adapted from86): a) NASCET method = [1–(a/b)] × 100; b) ESCT method = [1–(a/c)] × 100; common carotid method = [1–(a/d)] × 100.

A review of Doppler US, spiral CTA and time-of-flight (TOF) MRA showed similar accuracy of all three techniques, with US showing the best correlation with DSA. Combining two or three noninvasive techniques improved accuracy69. However TOF MRA is a flow-sensitive technique prone to artifactual signal loss due to changes in vessel orientation and high-grade stenoses. It has been superseded by contrastenhanced MR angiography (CEMRA). Multidetector CT also represents a technical advance, with each new generation scanner offering higher resolution than the last at shorter examination times. CEMRA and CTA both allow a full assessment of the arterial tree from the aortic arch to the circle of Willis and beyond, which is not possible on US or usually on TOF MRA. There is good correlation between CEMRA, four-detector row CTA and Doppler US for the assessment of symptomatic carotid bifurcation stenosis70. CEMRA also compares well with DSA for assessment of carotid and vertebral artery disease71,72. A tendency for carotid stenosis to be exaggerated on CEMRA has been reported.Vessel narrowing can be underestimated on DSA if plaque is partially obscured on frames acquired when contrast density is maximal71. Using a fast frame rate in different projections or inspecting the source images from rotational angiography may avoid this.The length of an occlusion can be exaggerated on CEMRA if the vessel partially fills by retrograde flow on late frames of a DSA.This may be avoided in the future with time-resolved MRA techniques72. The nonvertical segments of tortuous vertebral artery origins can appear narrowed on CEMRA, although the true lumen diameter can often be appreciated on source images. In practice even CEMRA and CTA combined sometimes fail to adequately resolve vertebral artery origins; in such cases catheter angiography is definitive. Review of CTA should include assessment of any stenosis in the axial plane in addition to reconstructed images and this is particularly helpful in differentiating between string flow and occlusion, often a difficult judgement on other

1307

1308

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

currently remains the investigation against which others are judged. It shows sequential opacification of the arterial tree, including collateral pathways and steal phenomena. However excessive use of DSA reduces the benefit of carotid revascularization by exposing the patient population to the additional stroke risk of the investigation. Decision making is even more finely balanced for patients with an asymptomatic stenosis70. Irregularity due to atheroma must be differentiated from catheter-induced spasm (Fig. 57.17), fibromuscular dysplasia and spontaneous or iatrogenic dissection. Fibromuscular dysplasia causes extensive, concentric corrugation of the artery, frequently bilateral and rarely extending above the skull base (Fig. 57.18). There may be associated renal artery stenosis. Dissection of the cervical arteries is an important cause of stroke, particularly in younger patients. It can be diagnosed acutely with MRI, which shows an expanded artery with high signal intramural haematoma around a narrow, often eccentric signal void indicating the true lumen. Internal carotid artery dissection often involves the vessel just below the skull base and may be identified on the lowermost routine axial T2-weighted images through the brain, without the need for additional special sequences. If not it is

Figure 57.16 Carotid bifurcation atheroma. Examples of internal carotid artery stenoses (arrows) from different patients on (A) CEMRA MIP, (B) CTA volume rendering and (C) DSA. Note the right vertebral artery origin is clearly demonstrated in A (black arrow).

modalities. Calcified plaque exaggerates stenoses on MIPs; carefully windowed minimum thickness multiplanar reformat (MPR) images are likely to be a better guide in such circumstances. Inevitably there are local variations but many centres opt to combine more than one technique to improve accuracy, for example US plus CEMRA, with CTA used as a problem-solving tool if there is discrepancy. DSA is still occasionally necessary if other techniques are inconclusive and it

Figure 57.17 Catheter-induced arterial spasm. (A) Elective arteriogram following aneurysm coiling 6 months earlier. AP view of left internal carotid artery shows marked spasm in response to selective catheterization (short arrows). The catheter was withdrawn into the common carotid artery before this run (long arrow). (B) Four minutes later the spasm has almost resolved (arrows).

CHAPTER 57

•

Figure 57.18 Fibromuscular dysplasia; internal carotid arteriogram, lateral projection. The cervical portion of the artery shows regular concentric corrugations: appearances were similar on the other side.

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

usual to employ fat-suppressed T1- or T2-weighted axial sequences through the neck. In practice vertebral artery dissection is more difficult to diagnose in this way, perhaps because these signs are harder to detect in a smaller vessel (Fig. 57.19). Signs of dissection on angiography include a tapered stenosis or occlusion of the true lumen (‘rat’s tail’), which in the internal carotid arteries often starts at or just above the carotid bifurcation and sometimes extends intracranially. An intraluminal flap may be visible. Imaging at fast frame rates may show differential opacification and contrast washout in the true and false lumens if they are both patent. Many of these features can be shown on MRA or CTA and the diagnosis can often be made without recourse to DSA. CTA has shown promise in the noninvasive diagnosis of vertebral artery dissection73. The source images and MPRs may show an eccentric lumen within an enlarged vessel analogous to the appearance on axial MRI (Fig. 57.19). Internal carotid artery dissection may also be diagnosed using Doppler US (see Ch. 3). Finally within this section, cross-sectional imaging (MRI, CTA and US) is being used to image atheromatous plaque in the hope that characteristics such as lipid content might help to predict stroke risk and guide therapy74.These are largely still

Figure 57.19 Diagnosis of cervical arterial dissection on MRA and CTA. (A) CEMRA in a 50-year-old man with left-sided neck pain and a right Horner’s syndrome after cervical manipulation (vertebral arteries removed for clarity). There are bilateral ICA dissections with irregular narrowing (long arrows), extending into the proximal petrous canal on the left (arrowhead). There are also pseudoaneurysms on both sides (short arrows). (B) Axial T2-weighted image through the neck at C2 level in a different patient. On both sides there is high signal mural thrombus (short arrows) with eccentric flow void indicating the position of the true lumen (long arrows). The appearance is diagnostic of ICA dissections. Continued

1309

1310

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.19, Cont’d (C) DWI in a different patient shows a small left superior cerebellar infarct (arrow), which would be easily overlooked on D, the corresponding T2-weighted image (arrow). (E) Coronal reformat from CTA shows expanded left vertebral artery at C1 with mural thrombus (short arrow) and eccentric enhancing lumen (long arrow), analogous to the MRI appearance in B. (F) Volume rendering from the same examination shows the left vertebral artery (short arrow) as it passes around the left lateral mass of C1, viewed from the left and slightly in front. A small pseudoaneurysm is shown (long arrow).

research techniques at present. DSA can show plaque ulceration but no information can be derived regarding composition of plaque or arterial wall.

Imaging of intracranial vessels In addition to stroke (see earlier) and haemorrhage (see later), intracranial vessels are sometimes imaged for suspected cerebral vasculitis. This heterogeneous group of inflammatory diseases mainly affects smaller parenchymal and leptomeningeal vessels. Conditions that cause cerebral vasculitis, other than primary (isolated) angiitis of the central nervous system, include infection, malignancy, radiotherapy, cocaine ingestion and autoimmune diseases such as systemic lupus erythematosus, polyarteritis nodosa, giant cell arteritis and Sjögren’s syndrome.

High-resolution intra-arterial angiography is superior to MRA and CTA, particularly for smaller, more peripheral vessels. Angiographic signs suggesting a vasculitis include stenoses, occlusion, thromboses or arterial beading, although they are not specific (Fig. 57.20). Angiography is frequently negative and it should not be regarded as the gold standard investigation; a brain or meningeal biopsy is often necessary to make a firm diagnosis.

Moya moya Moya moya is Japanese for ‘puff of smoke’, describing the angiographic appearance of dilated collateral vessels that develop secondary to progressive occlusion of the supraclinoid internal carotid arteries. Moya moya represents an idiopathic arteriopathy

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.20 Arteritis caused by septic emboli of cardiac origin. (A) Common carotid arteriogram, anteroposterior projection. Filling defects are seen (arrows), many vessels are irregular and under filled, and the middle cerebral artery bears a mycotic aneurysm (double arrows). (B) Beaded appearance of cortical arteries (arrows), lateral projection.

mainly seen in Japan and the Pacific Rim. In advanced cases there may be extensive dural, leptomeningeal and pial collateral circulation. The characteristic vascular changes can also be shown on MRA or CTA and MRI may show associated

infarcts. A moya moya-like pattern may be found in other conditions such as sickle cell disease, Down’s syndrome, previous radiotherapy or tuberculous meningitis and Type 1 neurofibromatosis75.

NONTRAUMATIC INTRACRANIAL HAEMORRHAGE Intracranial haemorrhage may be traumatic or spontaneous (nontraumatic) and is described anatomically as intraparenchymal, intraventricular, subarachnoid, subdural, or extradural. Common causes of spontaneous haemorrhage include hypertension, amyloid angiopathy, aneurysms and various vascular malformations.

to the perimesencephalic area. Patients with a nonaneurysmal perimesencephalic SAH (who by definition have a negative angiogram) have a very good long-term prognosis76. The risk of further bleeding is thought to be no higher than that in the general population. Variations in venous anatomy found with this pattern of SAH suggest a venous origin77.

Appearance on CT and MRI

SUBARACHNOID HAEMORRHAGE Spontaneous SAH is due to a ruptured arterial aneurysm in 70–80 per cent of patients and an arteriovenous malformation in about 10 per cent. In the remaining approximately 15 per cent, no underlying cause is found on angiography, which is more likely when the subarachnoid blood is confined

CT is positive for SAH in 98 per cent within 12 h of onset78 but this falls to less than 75 per cent by the third day79. Recent SAH causes increased density of the cerebrospinal fluid (CSF) spaces on CT, assuming sufficient elevation of the CSF haematocrit (Fig. 57.21). Most aneurysms are located on or close to the circle of Willis and blood is therefore seen in the basal cisterns, although the entire intracranial subarachnoid space

Figure 57.21 Anterior communicating artery aneurysm rupture. (A) CT shows diffuse subarachnoid haemorrhage (short arrows) and a small haematoma in the septum pellucidum indicating the likely source is an aneurysm of the anterior communicating artery (long arrow). The temporal horns of the lateral ventricles and anterior recesses of the third ventricle are enlarged (arrowheads) due to secondary communicating hydrocephalus, an extremely common finding in acute SAH. (B) A 3D angiogram shows a lobulated aneurysm in the predicted location (long arrow). The left proximal anterior cerebral artery (short arrow), middle cerebral artery (arrowhead) and internal carotid artery (open arrow) are clearly shown.

1311

1312

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

may be opacified and intraventricular blood is common. In some cases surrounding clot indicates the aneurysm location. A lumbar puncture should be performed when there is a strong clinical suspicion of SAH but negative CT. Spin-echo MRI sequences are unreliable in SAH but using a T2* gradient-echo or FLAIR sequence (Fig. 57.22), sensitivities of 94–100 per cent and 81–87 per cent can be achieved in the acute (less than 4 d) and subacute (more than 4 d) periods, respectively80. However FLAIR is less sensitive at low CSF red blood cell concentrations following normal CT81. The susceptibility effects of para-magnetic iron cause low signal on gradient echo sequences; on FLAIR the CSF appears high signal due to the presence of increased protein. FLAIR may remain positive for at least 45 d after a haemorrhage82, at a time when the blood has long since become invisible on CT. Therefore MRI may be used to diagnose SAH following normal CT, but a lumbar puncture should still be performed following a negative CT and MRI in someone with a high clinical suspicion of SAH. Supplemental inspired oxygen83 and contrast medium leakage into the subarachnoid space after an acute infarct84 can both cause the CSF to return high signal on FLAIR and CSF flow is a common cause of artifactual high signal in daily practice. CT and MRI may also show a number of complications in patients with SAH, of which communicating hydrocephalus and ischaemia secondary to vasospasm are the most important. It is very common to see mild dilatation of the ventricles, particularly the temporal horns, at diagnosis. Indeed it may be a useful clue to the diagnosis if the presence of blood is not obvious. It usually resolves over several days, but may progress and necessitate a ventricular drain or shunt.Vasospasm usually occurs between 4 and 11 d after the

haemorrhage and is a significant cause of morbidity during this period85. It is more likely if the initial CT shows a large amount of subarachnoid blood. MRI in chronic repeated SAH may show evidence of superficial siderosis with haemosiderin staining of the leptomeninges, particularly around the midbrain and in the posterior fossa. Such patients often present with symptoms related to the lower cranial nerves, ataxia or gradual cognitive decline (Fig. 57.23).

Cerebral aneurysms may be saccular, fusiform, or dissecting86. Fusiform aneurysms can be regarded as an extreme form of focal ectasia in arteriosclerotic disease. Intracranial aneurysms can also develop following an arterial dissection. However the majority are saccular aneurysms, which are usually round or lobulated and arise from arterial bifurcations. Giant aneurysms by definition measure over 25 mm in diameter and account for approximately 5 per cent of all cerebral aneurysms. They often contain layers of organized thrombus. Aneurysms tend to present with SAH or mass effect on adjacent structures, most commonly a posterior communicating artery aneurysm causing a third nerve palsy. Increasingly saccular aneurysms are discovered incidentally on scans for other indications and this represents a management problem. A recent large-scale study suggests that the annual risk of haemorrhage from small incidental aneurysms is substantially lower than previously thought and the risks of elective intervention higher87,88. Current data indicates that there is no benefit from treating aneurysms of

Figure 57.22 Subarachnoid haemorrhage on FLAIR. Blood is shown as high signal in occipital sulci (long arrows) and layered in the occipital horns of both lateral ventricles (short arrows).

Figure 57.23 Superficial siderosis. Axial T2-weighted image shows the pons, mesial temporal lobes and cerebellar folia are outlined by a low signal intensity haemosiderin rim indicating repeated SAH.

CEREBRAL ANEURYSMS

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

the anterior circulation 7 mm or less in diameter regardless of the patient’s age, if there is no prior history of SAH. The difference between the risks of haemorrhage and treatment may favour intervention for larger or posterior circulation aneurysms, depending on aneurysm size and remaining life expectancy88. Around 90 per cent of intracranial aneurysms arise from the carotid circulation, the remaining 10 per cent from vertebral or basilar arteries86. The anterior and posterior communicating arteries give rise to approximately one-third each of all intracranial aneurysms, with another 20 per cent from middle cerebral arteries and 5 per cent from the basilar termination. The remainder arises from other vessel origins and bifurcations. A clot in the septum pellucidum, possibly extending into one or other frontal lobe, is virtually diagnostic of an aneurysm of the anterior communicating artery (Fig. 57.21). Aneurysms of the distal anterior cerebral artery related to pericallosal branches are less common. Aneurysms of the MCA bleed into the Sylvian fissure, sometimes with a clot in the temporal lobe. Aneurysms of the posterior communicating artery (which arise from the internal carotid artery at the origin of this vessel) are a frequent cause of SAH but can also present with isolated third nerve palsy due to pulsatile pressure on the nerve. Other relatively common sites of aneurysms of the internal carotid artery are the origin of the ophthalmic and anterior choroidal arteries and its terminal bifurcation. Aneurysms of the posterior circulation are commonly located at the basilar artery bifurcation and if they rupture blood may be seen in the interpeduncular fossa, brainstem or thalamus; prognosis is frequently poor. The second commonest site in the posterior circulation is at the origin of one of the posterior inferior cerebellar arteries. They often haemorrhage into the ventricular system via the fourth ventricle and downwards into the spinal subarachnoid space. Larger aneurysms are shown on CT and MRI. On CT they appear as rounded enhancing lesions. Giant aneurysms have an enhancing lumen and a wall of variable thickness that often contains laminated thrombus and may be calcified. On spinecho MRI sequences a patent aneurysm appears as an area of flow void. Areas of increased signal intensity within the aneurysm may represent mural thrombus or turbulent, slow flow. Surrounding white matter oedema suggests a mycotic aneurysm, particularly if very extensive. Mycotic aneurysms are caused by septic emboli and tend to occur peripherally, typically on the branches of the MCA.They commonly present with haemorrhage, usually with a peripheral intraparenchymal clot, which, while not specific, is highly suggestive of such a lesion in a patient with known septicaemia or bacterial endocarditis (Figs 57.20, 57.24).

Angiography in subarachnoid haemorrhage and intracranial aneurysms Until recently SAH was an incontestable indication for intraarterial angiography but this has changed with improved CT technology. Early CTA studies produced inconsistent results but in expert hands a very high sensitivity for aneurysm detec-

tion can now be achieved. A comparative study of spiral CTA, DSA and surgical findings for small aneurysms less than 5 mm in diameter found that CTA outperformed DSA, with sensitivities of 98–100 per cent and 95 per cent, respectively89. In neurovascular centres CTA is now at least the equal of two-dimensional (2D) DSA for the diagnosis and anatomical assessment of aneurysms90 (Fig. 57.25). It is essential to methodically review source images on a workstation, in addition to MPRs and MIPs in multiple planes and 3D renderings. Particular care should be taken close to the skull base, where adjacent bone may reduce the conspicuity of small aneurysms. CTA images can be degraded by vasospasm or inadequate opacification if the images are not acquired during the arterial phase of contrast enhancement. It is apparent on visual inspection when this is the case. The images can be rotated in multiple planes (like in 3D rotational DSA, see Fig. 57.21) allowing better demonstration of an aneurysm and its neck than sometimes achieved on DSA, especially since overlying vessels can be ‘removed’. In patients subsequently undergoing catheter angiography CTA image rotation allows prior selection of optimal projections. There are occasions when aneurysms are masked by superimposition on DSA but visible on a CTA. However the greatest benefit of CTA is its ease of use. It takes only a few minutes to prepare the patient and plan the examination and on a 16-detector CT system the whole head is imaged in less than 10 s. This is ideal in patients with SAH, who are often restless and unwell. Unenhanced cranial CT and perfusion data can be acquired during the same examination if necessary. It avoids the need for an arterial puncture and the risk of ischaemic stroke from catheter angiography, admittedly extremely small if SAH is the indication for the procedure. Patient preference has been shown for CTA over MRA or DSA in the investigation of carotid stenosis69 and it is reasonable to assume that the same applies to aneurysm patients. CTA can now be viewed as a complementary investigation to DSA, the latter being reserved for problem solving or in some centres dispensed with altogether as a diagnostic investigation90. It seems superficially attractive to consider devolving CTA to the general hospital environment. However accurate interpretation requires experience in neurovascular radiology and CTA is part of the overall care of SAH patients, which is currently delivered in a neuroscience environment. The sensitivity of MRA for prospective detection of aneurysms larger than 5 mm is 77–94 per cent, varying with observer experience, but it is much less sensitive for aneurysms smaller than 3 mm91–93. Like CTA, MRA images can be rotated and may show aneurysms missed on DSA94. Recent SAH may cause image degradation on TOF MRA due to T1 shortening from haemorrhage. Giant aneurysms are rarely visualized in their full extent on 3D TOF MRA because of slow and turbulent flow in their fundus.The lumen is properly opacified on CTA, which also shows mural thrombus and the aneurysm wall. MRA is a reasonable first-line option for elective imaging of aneurysms, although its modest sensitivity for smaller aneurysms should be borne in mind. It is also used for following coiled aneurysms. CTA is now preferable in acute SAH.

1313

1314

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.24 Mycotic aneurysm. (A) CT of a patient with infective endocarditis shows an acute right frontal intraparenchymal haemorrhage. (B) Coronal MIP from a CTA shows pooling of contrast at the periphery of the haematoma, suggesting an aneurysm (arrow). (C) Lateral right ICA arteriogram confirms a peripheral mycotic aneurysm (arrow).

Practice varies in patients with a negative CTA following SAH. There is a body of opinion that a single technically adequate CTA is sufficient following a classical perimesencephalic SAH95. Some institutions now rely solely on CTA for all diagnostic imaging in SAH. However in many centres it remains routine practice to confirm a negative CTA result with DSA, which is also used if for any reason aneurysm anatomy is not adequately displayed on CTA. Traditionally a ‘four vessel’ catheter angiogram was performed because up to 20 per cent of aneurysms are multiple. This involves bilateral common or internal carotid injections and one or both vertebral arteries depending on whether reflux of contrast medium displays the contralateral posterior inferior cerebellar artery origin. If a dural fistula is a possibility the external carotid arteries should also be injected. A standard examination comprises sufficient projections to resolve all vessels without superimposition and demonstrates the anatomy of any aneurysms and adjacent arteries. It can be difficult to

decide which aneurysm has ruptured if more than one is found and cross-sectional imaging shows a symmetrical distribution of blood (or if SAH was diagnosed by lumbar puncture after a negative scan). Ruptured aneurysms often have an irregular shape and may show a ‘nipple’ indicating the site of rupture. Alternatively the larger of multiple aneurysms is frequently incriminated, but this is not a very reliable rule. It is justifiable to perform a limited angiogram when the territory of the bleed is indicated from cross-sectional imaging, particularly if remote vessels are normal on CTA or the patient is elderly or very sick. Angiography should be performed as soon as possible following SAH since the aneurysm re-bleed rate is greatest during the first 48 h and vasospasm can adversely affect the quality of angiograms performed several days after the haemorrhage. If a negative angiogram is marred by vasospasm, a repeat study is indicated. Three-dimensional angiography reduces the need for multiple angiographic runs and provides high-resolution 3D

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.25 Aneurysms on CTA. (A) CT shows acute subarachnoid blood in interhemispheric and right Sylvian fissures (short arrows) and a clot in the right side of the chiasmatic cistern extending into the temporal lobe (open arrow). Within this is a rounded area of lower density, suggesting an aneurysm (long arrow). There is also an acute right subdural haematoma (arrowheads). (B) CTA viewed from above, behind and the left shows a large right distal ICA aneurysm arising at the level of the posterior communicating artery (not visible as anatomically hypoplastic). The neck is clearly shown (arrowhead) and there is a small lobule (short arrow) on the fundus possibly indicating the site of rupture. Note the left posterior communicating (long arrow) and anterior communicating arteries (open arrow). The decision to treat by endovascular coiling was based on this examination. (C, D) Lateral right ICA arteriograms from coiling procedure, immediately before and after occlusion of the aneurysm. Note confirmation of aneurysm anatomy as shown on CTA, including terminal lobule (arrow). (E) Coronal MIP from CTA in a different patient with a complex right middle cerebral artery aneurysm. Note the anatomical detail of separate lobules and artery arising from the neck of the superior aneurysm (arrow).

1315

1316

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

images of the proximal cerebral vessels that show the relationship of an aneurysm to adjacent vessels, although CTA often provides similar information (Figs 57.21, 57.25). Manipulation at the workstation permits the image to be rotated and viewed from any angle. Aneurysms can be treated by surgical clipping or endovascular coiling (Fig. 57.25). The latter is performed via a microcatheter placed in the aneurysm sac through which a number of electrically detachable platinum coils are deployed. A multicentre randomized comparison of surgical clipping and endovascular coiling showed superior outcomes at 1 year for coiling over clipping (death or dependency 23.5 per cent vs. 30.9 per cent; absolute risk reduction 7.4 per cent). This difference was maintained at 7 years with a lower risk of epilepsy but more episodes of re-bleeding in the coiled group96. There is still a role for surgical clipping if the anatomy of the aneurysm is unfavourable for endovascular treatment.

INTRACEREBRAL HAEMORRHAGE Nontraumatic intracerebral haemorrhage in older people is frequently from rupture of a small perforating vessel due to hypertension. The preferential sites of hypertensive haemorrhage are the basal ganglia, thalamus and pons. Larger hypertensive bleeds in the basal ganglia often extend into the ventricles or Sylvian fissure. Peripheral or lobar haemorrhages in the elderly are suggestive of amyloid angiopathy, particularly if they are multifocal97 and accompanied by microbleeds elsewhere, best shown on T2 gradient-echo imaging. Primary or secondary brain neoplasms may also cause intracerebral haemorrhage. Haemorrhage may also be due to vascular malformations and ‘recreational’ drugs such as cocaine and ecstasy98. Intracranial haemorrhage due to aneurysms is usually associated with SAH, but very occasionally a ruptured aneurysm can cause an apparently isolated intracerebral clot (particularly if the surrounding subarachnoid space has been ‘sealed off ’ by preceding SAH). Other causes include coagulopathies, anticoagulation, vasculitis, venous infarcts and haemorrhagic transformation of arterial infarcts.

Appearance on CT and MRI Intracerebral haemorrhage is reliably detected on CT, appearing as increased density. Calcified or highly proteinaceous material and contrast enhancement of tumours can reach a similar density to fresh blood clot, but the clinical context or correlation with unenhanced images should prevent confusion. An acute intracerebral clot is usually of fairly homogeneous density. Hyperacute unclotted blood will appear less dense, which may cause a blood–fluid level to be visible.This appearance is most commonly due to haemorrhage from coagulopathies (usually anticoagulant medication).Very rarely in severely anaemic patients with a haematocrit level below 20 per cent haematomas can be isodense to the surrounding brain99. Deep or extensive haemorrhage may leak into the ventricles, forming a haematoma, or a blood–fluid level in the occipital horns, which are dependent with the patient supine.

There is typically only a fine rim of low density around a fresh clot and extensive oedema at presentation suggests an underlying neoplasm. Other features favouring a neoplastic haemorrhage are a more complex structure, extensive surrounding vasogenic oedema in the acute phase and enhancing areas not immediately adjacent to the blood clot. In some cases the diagnosis can only be made after follow-up studies. Over the course of several days, an untreated haematoma becomes less dense, from the periphery towards the centre and therefore appears smaller. The timing depends on the size of the clot. Small haemorrhages can look identical to infarcts on CT by 8–9 d, which clearly has important treatment ramifications42. MRI will distinguish between the two.Vasogenic oedema may develop in the surrounding white matter and should contrast medium be given at this stage, it usually produces a halo of enhancement. After several weeks, the blood products become hypodense and are eventually absorbed to leave a focal cavity or area of atrophy. The MRI appearance of intracerebral haemorrhage changes over time as red cells lyse and haemoglobin degrades, ultimately taken up by macrophages as haemosiderin100 (Fig. 57.26 and Table 57.1). Factors such as protein and water content, fibrin formation and clot retraction can alter the sequence and timing of changes in appearance on MRI. Gradient-echo imaging is much more sensitive to the magnetic field inhomogeneities induced by paramagnetic blood products than spin-echo sequences and this applies to both acute and old haemorrhage (deoxyhaemoglobin and haemosiderin, respectively) (Fig. 57.11).

Angiography in intracerebral haemorrhage The indications for angiography in intracerebral haemorrhage are determined by clinical factors as much as imaging appearance. It is unlikely a treatable vascular abnormality will be found in a basal ganglia bleed in an elderly hypertensive patient whereas a haemorrhage in the same location in a young, normotensive patient warrants further investigation with angiography to exclude an AVM. It is also noteworthy that some ‘recreational’ drugs are associated with aneurysm formation and rupture so angiography is often appropriate in such patients. The timing of angiography depends on the size and mass effect of the haematoma. Occasionally urgent angiography is required before surgical evacuation of a haematoma. Otherwise it is usually preferable to defer angiography until the haematoma has resolved because smaller vascular lesions can be compressed by an acute haematoma and not be apparent angiographically. CTA seems to be a reasonable alternative to DSA in the emergency setting for the detection of ruptured AVMs and aneurysms, and it is certainly much easier and quicker to perform in very sick patients. However it is not yet established that a small arteriovenous fistula or malformation can be reliably excluded using CTA so DSA is still preferred for elective investigation of such patients. The same reservation applies to MRA although the dynamic gadolinium bolus technique of MR DSA, which is time resolved, has produced some interesting early results in patients with large dural AVFs101. It remains to be seen if spatial resolution improves sufficiently to increase the sensitivity of MRA to clinically useful levels.

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.26 Cerebral haemorrhage: a simplified explanation of serial MRI signal changes. (A) Red cells containing oxyhaemoglobin (O), which gives signal similar to brain on both T1 and T2 weighting, are extravasated. (B) The oxyhaemoglobin in some of the cells at the centre of the clot is converted to deoxyhaemoglobin (DE), giving lower signal on T1-weighted images. The clot is now surrounded by a variable amount of cerebral oedema (small circles), which gives non-specific increased signal on T2-weighted images. (C) Most of the haemoglobin is now converted to methaemoglobin (M), which gives high signal on both T1- and T2weighted images: however, lower signal due to the continued presence of deoxyhaemoglobin (DE) may still be evident centrally. Progressive lysis (interrupted outline) of the red cells is occurring. Surrounding oedema may become more extensive. (D) The red cells have now broken down, leaving a post-haemorrhagic cyst. This still contains methaemoglobin (M), and returns high signal on all the commonly used sequences. The oedema around the clot has resolved, but the macrophages (circles with three engulfed particles) which surround the cavity contain haemosiderin granules (H) which, because of susceptibility effects, give markedly reduced signal with T2 weighting and, to a lesser extent, on T1-weighted images. Persistence of high signal from methaemoglobin and low signal from haemosiderin is variable, but in some patients these effects can be seen for many months after documented bleeding.

ARTERIOVENOUS MALFORMATIONS Intracranial vascular malformations can be classified according to the presence or otherwise of arteriovenous shunting102. The former comprises cerebral (or subpial) arteriovenous malformations (AVMs) and dural fistulae; the latter includes developmental venous anomalies (DVAs), cavernous angiomas (‘cavernomas’) Table 57.1

and capillary telangiectasias. Intra-arterial angiography is still the method of choice for the investigation of cerebral AVMs and dural fistulae. Only occasionally is it necessary to confirm a DVA angiographically as the MRI appearance is usually characteristic. Cavernous angiomas and telangiectasias are angiographically occult or ‘cryptic’ vascular malformations. Cerebral (subpial) AVMs are probably congenital anomalies consisting of direct arteriovenous shunts without a normal intervening capillary bed. Some are essentially fistulous; others have a plexiform nidus or a combination of the two. They lie within the brain substance or cerebral sulci and are supplied by branches of the internal carotid artery or vertebrobasilar system, sometimes recruiting additional supply from meningeal arteries. Cerebral haemorrhage is the commonest clinical presentation, others being epilepsy, headache or focal neurological deficit. They are usually detectable on CT or MRI as serpiginous areas of high density (with marked contrast enhancement) or mixed signal, respectively. CT may show calcification and the MR signal comprises areas of flow void and high signal, which may represent thrombosis or flow-related enhancement.There may be haemorrhage at different stages of evolution. AVMs may be surrounded by areas of ischaemic damage that are low attenuation on CT and hyperintense on T2-weighted MRI. Dilated feeding arteries and early opacification of draining veins are the angiographic hallmarks of these lesions. Dural arteriovenous fistulae are direct shunts between branches of the external carotid artery or meningeal branches of the cerebral vessels and dural sinuses. They are thought to be acquired and may be due to prior venous thrombosis. The clinical presentation depends on their location and venous drainage pattern103. Lesions shunting into the cavernous sinus commonly present with proptosis. Shunting into the transverse or sigmoid sinus may cause pulsatile tinnitus. Intracranial haemorrhage, which may be intracerebral, subarachnoid or subdural, usually occurs in lesions that reflux into cortical veins. They may go undetected on MRI or CT unless there are enlarged dural sinuses or cortical veins. MRA or CTA may show abnormal vessels more clearly but intra-arterial angiography is still required to make a definitive diagnosis. Angiography for an AVM or dural fistula should include injections of all possible feeding vessels using a high frame rate to improve delineation of the nidus, which otherwise can be obscured by overlying veins in rapidly shunting lesions. There may be associated aneurysms, either on the feeding arteries or within the nidus and venous drainage may be via deep and/or superficial systems. There may be venous varices or stenoses (Figs 57.27, 57.28).There is an increased risk

MR SIGNAL CHARACTERISTICS OF INTRACEREBRAL HAEMORRHAGE (ACCORDING TO BRADLEY100)

Stage

Form of haem iron

T1-weighted MRI

T2-weighted MRI

Hyperacute (first few hours)

Oxyhaemoglobin

Iso- or hypointense

Slightly hyperintense

Acute (1–3 d)

Deoxyhaemoglobin

Slightly hypointense

Hypointense

Early subacute (3–7 d)

Intracellular methaemoglobin

Hyperintense

Hypointense

Late subacute (1–4 weeks)

Extracellular methaemoglobin

Hyperintense

Hyperintense

Chronic

Haemosiderin

Iso- or hypointense

Markedly hyperintense

1317

1318

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 57.27 Cerebral AVM on MRI. (A) Sagittal T2-weighted MRI shows a parietal AVM. There are varices (short arrow), dilated arteries (long arrow) and draining veins (notched arrow). (B) 3D TOF MRA shows hugely dilated left middle cerebral artery feeders (long arrow), nidus (short arrow), varices (arrowhead) and superficial draining vein (open arrow). This shows the basic components but no information about direction of flow. (C–E) Series of three MR DSA frames in a lateral projection acquired at 1-second intervals during an IV infusion of gadolinium contrast medium. The left internal carotid artery dominates as blood shunts via it to the AVM. There are feeding middle cerebral artery branches (long arrows), the nidus (short arrow), varices (arrowhead) and a large superficial draining vein (open arrow) all apparent on the first frame, indicating the speed of shunting. Subsequent frames show opacification of transverse sinus (open arrow) and later superior sagittal sinus (white arrow in E). Note the detail of a small venous pouch on the main draining vein identical to the structural image (notched arrows on Figs 57.27A and 57.27E). This technique may in the future provide temporal and spatial resolution closer to that of DSA.

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

Figure 57.28 Cerebral AVM on DSA. (A) Arterial and (B) venous phase of DSA in a patient with a cerebral AVM. The AVM is fed by branches of the anterior and middle cerebral arteries and the venous drainage is predominantly superficial into the superior sagittal and transverse sinuses.

of haemorrhage in the presence of intranidal aneurysms, a single draining vein, deep venous drainage and venous stenoses104,105. CTA and MRA show the components of an AVM94 but they currently lack the spatial resolution of intraarterial angiography and do not produce sequential images necessary to judge transit time and direction of flow through the constituent vessels. In due course MR DSA may go some way to providing this information101 (Fig. 57.27). The treatment options for cerebral AVMs include surgery, radiosurgery and endovascular embolization. More than one technique may be used in combination. Cavernous angiomas are mulberry-like lesions consisting of vascular spaces with little intervening tissue and haemorrhage of different ages. The incidence of clinically symptomatic haemorrhage remains uncertain, but is less frequent than with cerebral AVMs or dural fistulae. A previous bleed and infratentorial location are the main prognostic factors for recurrent haemorrhage. Lesions in or close to the cerebral cortex may cause epilepsy. They are occasionally intraventricular or arise on a cranial nerve. They appear as relatively well-defined, dense or calcified lesions on CT, which may show patchy contrast enhancement. On MRI they appear multilobular with mixed signal intensity centrally surrounded by a dark haemosiderin rim106 (Fig. 57.29). Not surprisingly gradient-echo sequences are the most sensitive. They may be multiple, particularly in familial cases107. In many clinical situations the discovery of a cavernoma represents an incidental finding. Developmental venous anomalies are not malformations but represent a benign variation in venous drainage. They may be found with cavernomas. They consist of radially arranged, dilated transmedullary veins that have a typical ‘caput medusa’ appearance on the venous phase of conventional angiograms (Fig. 57.30).They may drain into the superficial or deep venous system. They are readily diagnosed by contrast-enhanced CT or MRI106. Capillary telangiectasias are benign nests of dilated capillaries with normal brain tissue in between.They are usually found on postmortem examinations and are occasionally visible on MRI as areas of very subtle T2 hyperintensity or ill-defined enhancement. They do not cause haemorrhage.

Figure 57.29 Cavernous haemangioma. (A) T2-weighted axial image showing typical mixed signal intensity lesions. High signal is due to methaemoglobin and the low signal intensity rim of haemosiderin indicates an old haemorrhage. The ‘popcorn’ appearance of the larger lesion is typical of a ‘cavernoma’. Note the blood–fluid level in the smaller lesion (arrow). (B) Unenhanced CT of the same patients shows the lesions to be predominantly high density with tiny foci of calcification (arrows). Figure 57.30 Developmental venous anomaly (DVA) (arrow) draining into a large thalamostriate vein. Internal carotid arteriogram, lateral projection, venous phase.

1319

1320

SECTION 7

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

SUBDURAL AND EXTRADURAL HAEMORRHAGE Acute subdural and extradural haematomas are almost always post-traumatic and are considered in Chapter 59. Occasionally rupture of a cerebral aneurysm may cause an acute subdural haematoma, most frequently a posterior communicating artery aneurysm lying next to the free edge of the tentorium cerebelli. A dural arteriovenous fistula may also bleed into the subdural space. Angiography is therefore indicated following a spontaneous acute subdural haematoma. Chronic subdural haematomas represent a different entity. These are frequently bilateral and occur in elderly patients or alcoholics with underlying brain atrophy, patients on anticoagulants or following shunting for hydrocephalus. The underlying mechanism is thought to be leakage from bridging cortical veins following minor trauma. They may present with increasing confusion and a reduction in conscious level. Burr holes for drainage of a chronic subdural collection, sometimes under a local anaesthetic, is one of the few neurosurgical operations performed on the very elderly. On CT they appear to be of lower density than the brain but may contain areas of high density, or even fluid levels, due to more recent haemorrhage. The MRI appearance evolves in a similar pattern to intraparenchymal haemorrhage. Chronic subdural haematomas continue to give high signal on T2-

weighted images, while returning low signal on T1-weighted images, without becoming isointense to CSF, because of their higher protein content. Repeated episodes of bleeding can produce variable changes of signal intensity (Fig. 57.31) analogous to the variable density changes on CT. A pseudomembrane, which forms around chronic subdural haematomas, may show marked contrast enhancement or haemosiderin staining. Shallow subdural fluid collections and occasionally overt haemorrhage may also develop around the cerebral hemispheres and cerebellum secondary to mild brain descent in the low CSF volume syndrome108. In this condition patients usually present with postural headache that is worse on standing and relieved by lying down. There is sometimes a history of vigorous Valsalva, lumbar puncture or other spinal intervention. The MRI features, other than subdural collections, are diffuse dural thickening shown best on FLAIR or contrastenhanced T1-weighted images and mild cerebellar ectopia. These changes resolve after successful treatment.

ACKNOWLEDGEMENT Dr Andrew Carne (Department of Radiology, St George’s Hospital, London UK) for his technical expertise during the preparation of images for this chapter.

Figure 57.31 Subdural haematomas. (A) Axial T2-weighted and (B) coronal T1-weighted MRI of bilateral spontaneous subdural haematomas of different ages. The leftsided collection is a few days old (early subacute stage) and is of low signal on T2 weighting and high signal on T1 weighting. The right-sided collection is a few weeks old (late subacute stage) and appears of high signal on both sequences, but it appears less bright on the T1-weighted sequence than the more recent contralateral collection (see Table 57.1).

REFERENCES 1. Blight A, Pereira A C, Brown M M 2000 A single consultation cerebrovascular disease clinic is cost effective in the management of transient ischaemic attack and minor stroke. J R Coll Physicians Lond 34: 452–455 2. Bath P, Lees K 2000 Acute stroke. Br Med J 320: 920–992 3. Lovett J K, Dennis M S, Sandercock P A G et al 2003 Very early risk of stroke after a first transient ischemic attack. Stroke 34: e138–e140

4. Fairhead J F, Mehta Z, Rothwell P M 2005 Population-based study of delays in carotid imaging and surgery and the risk of recurrent stroke. Neurology 65: 371–375 5. Zimmerman R D 2004 Stroke wars: Episode IV. CT strikes back. AJNR Am J Neuroradiol 25: 1304–1309 6. Ueda T, Yuh W, Taoka T 1999 Clinical application of perfusion and diffusion MR imaging in acute ischemic stroke. J Magn Reson Imaging 10: 305–309 7. Markus H S 2004 Cerebral perfusion and stroke. J Neurol Neurosurg Psychiatry 75: 353–361

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

8. Halpin S F S 2004 Brain imaging using multislice CT: a personal perspective. Br J Radiol 77: 20–26 9. Liebeskind D S 2003 Collateral circulation. Stroke 34: 2279–2284 10. Caplan L R 1980 ‘Top of the basilar’ syndrome. Neurology 30: 72–79 11. Barber P A, Demchuk A M, Hudon M E et al 2001 Hyperdense Sylvian fissure MCA ‘dot’ sign: a CT marker of acute ischemia. Stroke 32: 84–88 12. Toyoda K, Masahiro I, Fukuda K 2001 Fluid-attenuated inversion recovery intra-arterial signal: an early sign of hyperacute cerebral ischemia. AJNR Am J Neuroradiol 22:1021–1029 13. Karonen J O, Partanen P L K, Vanninen R L et al 2001 Evolution of MR contrast enhancement patterns during the first week after acute ischemic stroke. AJNR Am J Neuroradiol 22: 103–111 14. Mullins M E, Schaefer P W, Sorensen A G et al 2002 CT and conventional and diffusion-weighted MR imaging in acute stroke: study in 691 patients at presentation to the emergency department. Radiology 224: 353–360 15. Na D G, Kim E Y, Ryoo J W et al 2005 CT sign of brain swelling without concomitant parenchymal hypoattenuation: comparison with diffusionand perfusion-weighted MR imaging. Radiology 235: 992–998 16. Kummer R, Meyding-Lamade U, Forsting M et al 1994 Sensitivity and prognostic value of early CT in occlusion of the middle cerebral artery trunk. AJNR Am J Neuroradiol 15: 9–15 17. Demchuk A M, Coutts S B 2005 Alberta stroke program early CT score in acute stroke triage. Neuroimag Clin N Am 15: 409–419 18. Bryan R N, Levy L M, Whitlow W D et al 1991 Diagnosis of acute cerebral infarction: comparison of CT and MR imaging. AJNR Am J Neuroradiol 12: 611–620 19. Hacke W, Kaste M, Fieschi C et al 1995 Intravenous thrombolysis with recombinant tissue plasminogen activator for acute hemispheric stroke. The European Cooperative Acute Stroke Study (ECASS). JAMA 274:1017–1025 20. Barber P A, Demchuk A M, Zhang J et al for the ASPECTS study group 2000 Hyperacute stroke: the validity and reliability of a novel quantitative CT score in predicting outcome prior to thrombolytic therapy. Lancet 355: 1670–1674 21. Barber P A, Hill M D, Eliasziw M et al 2005 Imaging of the brain in acute ischaemic stroke: comparison of computed tomography and magnetic resonance diffusion-weighted imaging. J Neurol Neurosurg Psychiatry 76:1528–1533 22. Coutts S B, Demchuk A M, Barber P A, et al 2004 Interobserver variation of ASPECTS in real time. Stroke 35: e103–e105 23. Liang L, Korogi Y, Sugahara T et al 1999 Detection of intracranial hemorrhage with susceptibility-weighted MR sequences. AJNR Am J Neuroradiol 20: 1527–1534 24. Fiebach J B, Schellinger P D, Gass A et al 2004 Stroke magnetic resonance imaging is accurate in hyperacute intracerebral haemorrhage: a multicenter study on the validity of stroke imaging. Stroke 35: 502–506 25. Kidwell C S, Chalela J A, Sayer J L et al 2004 Comparison of MRI and CT for detection of acute intracerebral haemorrhage. JAMA 292: 1823–1830 26. Mayer T E, Schuffe-Altedorneburg G, Droste D W et al 2000 Serial CT and MRI of ischaemic cerebral infarcts: frequency and clinical impact of haemorrhagic transformation. Neuroradiology 42: 233–239 27. Saunders D E, Howe F A, van den Boogart A et al 1995 Continuing ischaemic damage following acute middle cerebral artery infarction in man demonstrated by short echo proton spectroscopy. Stroke 26: 2272–2276 28. Wintermark M, Sesay M, Barbier E et al 2005 Comparative overview of brain perfusion imaging techniques. Stroke 36: e83–e99 29. Schellinger P D 2005 The evolving role of advanced MR imaging as a management tool for adult ischemic stroke: A Western-European perspective. Neuroimag Clin N Am 15: 245–258 30. Sobesky J, Weber O Z, Lehnhardt F-G et al 2004 Which time-topeak threshold best identifies penumbral flow? A comparison of perfusion-weighted magnetic resonance imaging and positron emission tomography in acute ischemic stroke. Stroke 35: 2843–2847

31. Kim J H, Shin T, Chung J D et al 1998 Temporal pattern of blood volume change in cerebral infarction: evaluation with dynamic contrastenhanced T2-weighted MR imaging. AJR Am J Roentgenol 170: 765–770 32. Guckel F J, Brix G, Schmiedek P et al 1996 Cerebrovascular reserve capacity in patients with occlusive cerebrovascular disease: assessment with dynamic susceptibility contrast-enhanced MR imaging and the acetazolamide stimulation test. Radiology 201: 405–412 33. Kuwabara Y, Ichiya Y, Sasaki M 1998 PET evaluation of cerebral hemodynamics in occlusive cerebrovascular disease pre- and postsurgery. J Nucl Med 39: 760–765 34. Sperling B, Lassen N 1999 Cerebral blood flow by SPECT in ischaemic stroke. In: Deyn P D, Diercks R, Alave A, Pickut B (eds) A Textbook of SPECT in Neurology and Psychiatry. John Libbey, London 35. Alexandrov A, Masdeu J, Devous S 1997 Brain single-photon emission CT with HMPAO and safety of thrombolytic therapy in acute ischemic stroke. Stroke 28: 1830–1834 36. Provenzale J, Sorensen G 1999 Diffusion-weighted MR imaging in acute stroke: theoretic considerations and clinical applications. AJR Am J Roentgenol 173: 1459–1467 37. Burdette J 1998 Cerebral infarction: time course of signal intensity changes on diffusion-weighted images. AJR Am J Roentgenol 171: 791–795 38. Geijer B, Lindgren A, Brockstedt S et al 2001 Persistent high signal on diffusion weighted MRI in the late stages of small cortical and lacunar ischaemic lesions. Neuroradiology 43: 115–122 39. Morita N, Harada M, Yoneda K et al 2002 A characteristic feature of acute haematomas in the brain on echo-planar diffusion-weighted imaging. Neuroradiology 44: 907–911 40. Schramm P, Schellinger P D, Klotz E et al 2004 Comparison of perfusion computed tomography and computed tomography angiography source images with perfusion-weighted imaging and diffusion-weighted imaging in patients with acute stroke of less than 6 hours’ duration. Stroke 35: 1652–1657 41. Chen Z M, Sandercock P A G, Pan H C, et al, on behalf of the CAST and IST collaborative groups 2000 Indications for early aspirin use in acute ischaemic stroke. A combined analysis of 40,000 randomised patients from the Chinese Acute Stroke Trial and the International Stroke Trial. Stroke 31:1240–1249 42. Wardlaw J M 2002 Recent developments in imaging of stroke. Imaging 14: 409–419 43. Wardlaw J M, Farrell A J 2004 Editorial: Diagnosis of Stroke on Neuroimaging. Br Med J 328: 655–656 44. Cristomo R A, Garcia M M, Tong D C 2003 Detection of diffusionweighted MRI abnormalities in patients with transient ischaemic attack. Stroke 34: 932–937 45. Yuh W, Ueda T, White M et al 1999 The need for objective assessment of the new imaging techniques and understanding the expanding roles of stroke imaging. AJNR Am J Neuroradiol 20: 1779–1784 46. Latchaw R 1999 The roles of diffusion and perfusion imaging in acute stroke management. AJNR Am J Neuroradiol 20: 957–959 47. Kidwell C S, Alger J R, Saver J L 2003 Beyond mismatch. Evolving paradigms in imaging the ischemic penumbra with multimodal magnetic resonance imaging. Stroke 34: 2729–2735 48. Winbeck K, Bruckmaier K, Etge T et al 2004 Transient ischemic attack and stroke can be differentiated by analyzing early diffusion-weighted imaging signal intensity changes. Stroke 35: 1095–1099 49. Liebeskind D S, Kidwell C S on behalf of UCLA thrombolysis investigators 2005 Advanced MR imaging of acute stroke: the University of California at Los Angeles endovascular therapy experience. Neuroimag Clin N Am 15: 455–466 50. National Institute of Neurological Disease and Stroke rt-PA Study Group 1995 Tissue plasminogen activator for acute ischaemic stroke. N Engl J Med 355; 1581–1587 51. Wardlaw J M, del Zoppo G, Yamaguchi T et al 2005 Thrombolysis for acute ischaemic stroke. The Cochrane Database of Systematic Reviews Issue 4 52. Longstreth W T, Manolio T A, Arnold A et al for the Cardiovascular Health Study Collaborative research group 1996 Clinical correlates of

1321

1322

SECTION 7

53.

54.

55.

56.

57.

58.

59. 60.

61.

62.

63. 64.

65.

66.

67. 68.

69.

70.

71.

72.

73.

•

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

white matter findings on cranial magnetic resonance imaging of 3301 elderly people. Stroke 27: 1274–1282 Murray A D, Staff R T, Shenkin S D et al 2005 Brain white matter hyperintensities: Relative importance of vascular risk factors in nondemented elderly people. Radiology 237: 251–257 O’Sullivan M, Rich P M, Barrick T R et al 2003 Frequency of subclinical lacunar infarcts in ischemic leukoaraiosis and CADASIL. AJNR Am J Neuroradiol 24: 1348–1354 Pomper M, Miller T, Stone H et al 1999 CNS vasculitis in autoimmune disease: MR imaging findings and correlation with angiography. AJNR Am J Neuroradiol 20: 75–85 Kwa V I, Franke C L, Verbeeten B Jr 1998 Silent intracerebral microhemorrhages in patients with ischemic stroke. Amsterdam Vascular Medicine Group. Ann Neurol 44: 372–377 Roob G, Schmidt R, Kapeller P et al 1999 MRI evidence of past cerebral microbleeds in a healthy elderly population. Neurology 52: 991–994 Lee S H, Kwon S J, Kim K S et al 2004 Cerebral microbleeds in patients with hypertensive stroke. Topographical distribution in the supratentorial area. J Neurol 251: 1183–1189 Lee S H, Kwon S J, Kim K S et al 2004 Topographical distribution of pontocerebellar microbleeds. AJNR Am J Neuroradiol 25: 1337–1341 Kakuda W, Thijs V N, Lansberg M G et al 2005 Clinical importance of microbleeds in patients receiving IV thrombolysis. Neurology 65: 1175–1178 Kidwell C S, Saver J L, Villablana J P et al 2002 Magnetic resonance imaging detection of microbleeds before thrombolysis. Stroke 33: 95–98 Ozsvath R R, Casey S O, Lustrin E S et al 1997 Cerebral venography: comparison of CT and MR projection venography. AJR Am J Roentgenol 169: 1699–1707 Caplan L R, Gorelick P B, Hier D B 1986 Race, sex and occlusive cerebrovascular disease: a review. Stroke 17: 648–655 North American Symptomatic Carotid Endarectomy trial collaborators 1991 Beneficial effect of carotid endarterectomy in symptomatic patients with high grade carotid stenosis. N Engl J Med 325: 445–453 European Carotid Surgery Trialists’ collaborative group 1991 MRC European carotid surgery trial: interim results for symptomatic patients with severe (70–99%) or with mild (0–29%) carotid stenosis. Lancet 337: 1235–1243 Halliday A, Mansfield A, Marro J et al 2004 Prevention of disabling and fatal strokes by successful carotid endarterectomy in patients without recent neurological symptoms: randomised controlled trial. Lancet 363: 1491–1502 Chaturvedi S, Bruno A, Feasby T et al Carotid endarterectomy – An evidence based review. Neurology 65: 794–801 Executive Committee for the Asymptomatic Carotid Atherosclerosis (ACAS) Study 1995 Endarterectomy for asymptomatic carotid artery stenosis. JAMA 273: 1421–1428 Patel S G, Collie D A, Wardlaw J M et al 2002 Outcome, observer reliability, and patient preferences if CTA, MRA, or Doppler ultrasound were used, individually or together, instead of digital subtraction angiography before carotid endarterectomy. J Neurol Neurosurg Psychiatry 73: 21–28 Nonent M, Serfaty J-M, Nighoghossian N et al 2004 Concordance rate differences of 3 noninvasive imaging techniques to measure carotid stenosis in clinical routine practice: results of the CARMEDAS multicenter study. Stroke 35: 682–686 Cosottini M, Pingitore A, Puglioli M et al 2003 Contrast-enhanced three-dimensional magnetic resonance angiography of atherosclerotic internal carotid stenosis as the noninvasive imaging modality in revascularization decision making. Stroke 34: 660–664 Yang C W, Carr J C, Futterer S F et al 2005 Contrast-enhanced MR angiography of the carotid and vertebrobasilar circulations. AJNR Am J Neuroradiol 26: 2095–2101 Chen C-J, Tseng Y-C, Lee T-H et al 2004 Multisection CT angiography compared with catheter angiography in diagnosing vertebral artery dissection. AJNR Am J Neuroradiol 25: 769–774

74. Fayad Z A, Guest editor 2002 Imaging of atherosclerosis. Neuroimaging Clinics of North America 12(3) 75. Barkovich A J 2005 Pediatric Neuroimaging, 4th edn. Lippincott Williams & Wilkins, USA p. 906 76. Rinkel G J, Wijdicks E F, Hasan D et al 1991 Outcome in patients with subarachnoid haemorrhage and negative angiography according to pattern of haemorrhage on computed tomography. Lancet 338: 964–968 77. van der Schaaf I C, Velthius B K, Gouw A et al 2004 Venous drainage in perimesencephalic haemorrhage. Stroke 35:1614–1618 78. van der Wee N, Rinkel G J, Hasan D et al 1995 Detection of subarachnoid haemorrhage on early CT: is lumbar puncture still needed after a negative scan? J Neurol Neurosurg Psychiatry 58: 357–359 79. Adams Jr H P, Kassell N F, Torner J C et al 1983 CT and clinical correlations in recent aneurysmal subarachnoid hemorrhage: a preliminary report of the Cooperative Aneurysm Study. Neurology 33: 981–988 80. Mitchell P, Wilkinson I D, Hoggard N et al 2001 Detection of subarachnoid haemorrhage with magnetic resonance imaging. J Neurol Neurosurg Psychiatry 70: 205–211 81. Mohamed M, Heasely D C, Yagmurlu B et al 2004 Fluid-attenuated inversion recovery MR imaging and subarachnoid hemorrhage: Not a panacea. AJNR Am J Neuroradiol 25: 545–550 82. Noguchi K, Ogawa T, Seto H, et al 1997 Subacute and chronic subarachnoid hemorrhage: diagnosis with fluid-attenuated inversionrecovery MR imaging. Radiology 203: 257–262 83. Deliganis A V, Fisher D J, Lam A M et al 2001 Cerebrospinal fluid signal intensity increase on FLAIR MR images in patients under general anesthesia: the role of supplemental O2. Radiology 218: 152–156 84. Dechambre S D, Duprez T, Grandin C B et al 2000 High signal in cerebrospinal fluid mimicking subarachnoid haemorrhage on FLAIR following acute stroke and intravenous contrast medium. Neuroradiology 42: 608–611 85. Kassell N F, Sasaki T, Colohan A R T et al 1985 Cerebral vasospasm following aneurysmal subarachnoid hemorrhage. Stroke 16: 562–572 86. Osborn A G 1999 Diagnostic cerebral angiography, 2nd edn. Lippincott Williams & Wilkins, Washington, p. 462 87. Wiebers D O, Whisnant J P, Huston J 3rd et al 2003 International Study of Unruptured Intracranial Aneurysms Investigators. Unruptured intracranial aneurysms: natural history, clinical outcome and risks of surgical and endovascular treatment. Lancet 362:103–110 88. Vindlacheruvu R R, Mendelow A D, Mitchell P 2005 Risk-benefit analysis of the treatment of unruptured intracranial aneurysms. J Neurol Neurosurg Psychiatry 76: 234–239 89. Villablanca J P, Jahan R, Hooshi P et al 2002 Detection and characterisation of very small cerebral aneurysms by using 2D and 3D helical CT angiography. AJNR Am J Neuroradiol 23: 1187–1198 90. Hoh B L, Cheung A C, Rabinov J D et al 2004 Results of a prospective protocol of computed tomographic angiography in place of catheter angiography as the only diagnostic and pretreatment planning study for cerebral aneurysms by a combined neurovascular team. Neurosurgery 54: 1329–1340 91. Huston J, Nichols D A, Luetmer P H et al 1994 Blinded prospective evaluation of sensitivity of MR angiography to known intracranial aneurysms: importance of aneurysm size. AJNR Am J Neuroradiol 15: 1607–1614 92. Atlas S, Sheppard L, Goldberg H I et al 1997 Intracranial aneurysms: detection and characterization with MR angiography with use of an advanced postprocessing technique in a blinded-reader study. Radiology 203: 807–814 93. Ida M, Kurisu Y, Yamashita M 1997 MR angiography of ruptured aneurysms in acute subarachnoid hemorrhage. AJNR Am J Neuroradiol 18: 1025–1032 94. Jager H R, Grieve J P 2000 Advances in non-invasive imaging of intracranial vascular disease. Ann R Coll Surg Engl 82:1–5 95. Ruigrok Y M, Rinkel G J, Buskens E et al 2000 Perimesencephalic hemorrhage and CT angiography: A decision analysis. Stroke 31: 2976–2983

CHAPTER 57

•

CEREBROVASCULAR DISEASE AND NONTRAUMATIC INTRACRANIAL HAEMORRHAGE

96. Molyneux A J, Kerr R S C, Yu L-M et al 2005 International subarachnoid aneurysm trial (ISAT) of neurosurgical clipping versus endovascular coiling in 2143 patients with ruptured intracranial aneurysms: a randomised comparison of effects on survival, dependency, seizures, rebleeding, subgroups, and aneurysm occlusion. Lancet 366: 809–817 97. Greenberg S M 1998 Cerebral amyloid angiopathy: prospects for clinical diagnosis and treatment. Neurology 51: 690–694 98. McEvoy A, Kitchen N, Thomas D 2000 Intracerebral haemorrhage in young adults: the emerging importance of drug abuse. Br Med J 320: 1322–1324 99. Gaskill-Shipley M 1999 Routine CT evaluation of acute stroke. Neuroimaging Clin North Am 9: 411–422 100. Bradley W G Jr 1994 Hemorrhage and hemorrhagic infections in the brain. Neuroimaging Clin North Am 4: 707–732 101. Coley S C, Romanowski C A J, Hodgson T J et al 2002 Dural arteriovenous fistulae: non-invasive diagnosis with dynamic MR digital subtraction angiography. AJNR Am J Neuroradiol 23: 404–407

102. Valavanis A 1996 The role of angiography in the evaluation of cerebral vascular malformations. Neuroimaging Clin North Am 6: 679–704 103. Rodesch G, Lasjaunias P 1992 Physiopathology and semeiology of dural arteriovenous shunts. Rivista di Neuroradiologia 5: 11–21 104. Meisel H J, Mansmann U, Alvarez H et al 2000 Cerebral arteriovenous malformations and associated aneurysms: analysis of 305 cases from a series of 662 patients. Neurosurgery 46: 793–800 105. Mast H, Young W L, Koennecke H C et al 1997 Risk of spontaneous hemorrhage after diagnosis of cerebral arteriovenous malformation. Lancet 350: 1065–1068 106. Wilms G, Demaerel P, Bosmans H et al 1999 MRI of non-ischemic vascular disease: aneurysms and vascular malformations. Eur Radiol 9: 1055–1060 107. Brunereau L, Labauge P, Tournier-Lasserve E et al 2000 Familial form of intracranial cavernous angioma: MR imaging findings in 51 families. Radiology 214: 209–216 108. Goadsby P J, Boes C 2002 New daily persistent headache. J Neurol Neurosurg Psychiatry 72 (suppl 2): ii6–ii9

1323

CHAPTER

Cranial and Intracranial Disease: Infections, AIDS, Inflammatory, Demyelinating and Metabolic Diseases

58

Daniel J. Scoffings and John M. Stevens

Intracranial infections • Brain abscess • Intracranial empyema • Ventriculitis • Tuberculosis • Encephalitis • Meningitis • Parasitic infestations HIV infection and AIDS • HIV encephalopathy • Cerebral toxoplasmosis • Primary cerebral lymphoma • Cryptococcosis • Progressive multifocal leukoencephalopathy • Tuberculosis

• Candidiasis • Incidental white-matter hyperintensities • Herpes viruses • Cerebrovascular disease • Histoplasmosis • Neurosyphilis • Spinal cord disorders Demyelinating, metabolic and nonspecific inflammatory disorders • Wilson’s disease • Intestinal encephalopathies • Multiple sclerosis • Osmotic myelinolysis • Sarcoidosis • Behçet’s disease

INTRACRANIAL INFECTIONS Intracranial infections take the following forms: • Cerebritis: focal, usually pyogenic infection, without a capsule or pus formation • Abscess: a focal, encapsulated, pus-containing cavity • Empyema: an abscess that forms in an enclosed space (or potential space), e.g. sub- or extradural • Granuloma: a focal, more or less encapsulated, inflammatory lesion, usually chronic, without pus formation • Encephalitis: direct infection of the brain, usually viral and often diffuse

• Meningitis: infection of the meninges, which may be suppurative or granulomatous • Osteomyelitis of the skull may also occur; infections of the paranasal sinuses are discussed elsewhere.

BRAIN ABSCESS In immunocompetent patients most brain abscesses are bacterial, streptococci accounting for the majority. In 20–40 per cent no causative organism is identified. Brain abscesses arise by

1326

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

haematogenous dissemination, penetrating trauma or direct spread from contiguous infection. The site of an abscess depends on its cause; frontal sinusitis will result in an abscess in or beneath the adjacent frontal lobe, whilst mastoiditis will give rise to a temporal lobe or cerebellar lesion. Blood-borne infection can occur anywhere in the brain, but has a predilection for the territory of the middle cerebral arteries, particularly the frontoparietal region. A thorough search for a predisposing factor should be made; a cardiac cause is frequently overlooked (occult endocarditis and septal defects). Abscesses are frequently subcortical or periventricular. Four stages of development are described: early and late cerebritis and early and late capsule formation. Patients present with fever (in 50 per cent), headache and focal neurological deficits. Brain abscesses are multiple in 10–50 per cent1. On computed tomography (CT), cerebritis appears as illdefined low attenuation and shows thick ring enhancement that may progress centrally on delayed images. With capsule formation the abscess shows central low attenuation, because of pus or necrotic debris and a rim of slightly higher attenuation surrounded by low attenuation vasogenic oedema. After contrast medium, a ring of enhancement corresponds to the capsule. The enhancing rim typically has a smooth inner margin and shows thinning of its medial aspect2 (Fig. 58.1). In comparison to cerebritis, the centre of the abscess never enhances on delayed images. The degree of enhancement is diminished in patients who are immunocompromised or

Figure 58.1 Cerebral abscess. CT: Low attenuation central abscess cavity surrounded by an enhancing rim and white matter vasogenic oedema. The medial aspect of the enhancing rim is subtly thinned.

are on corticosteroid therapy3. Abscesses rarely contain gas, most often caused by surgical intervention or communication with a cranial air-space, rarely because of a gas-forming organism. On magnetic resonance imaging (MRI), the signal of the abscess centre is between that of cerebrospinal fluid (CSF) and white matter on T1-weighted (T1W) images, and iso- or slightly hyperintense to CSF on T2-weighted (T2W) images. On T2W images the abscess rim is relatively hypointense; it may be slightly hyperintense to white matter on T1W images4. The pattern of rim enhancement is similar to that shown by CT. Surrounding vasogenic oedema is of low signal on T1W and high signal on T2W images. The abscess centre is high signal on diffusion-weighted imaging (DWI) and low signal on maps of apparent diffusion coefficient (ADC), because of restricted diffusion in the viscous pus5 (Fig. 58.2). Though typical, the appearance of a brain abscess as a rim-enhancing mass is nonspecific and may be mimicked by

Figure 58.2 Streptococcal abscess due to penetrating trauma. MRI. (A) Axial T2W fast spin-echo image. Note low signal of the abscess capsule and extensive high signal perilesional oedema. (B) Diffusion-weighted image shows high signal in the abscess centre, indicating restricted diffusion.

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

metastasis, glioblastoma, resolving haematoma or subacute infarct. A thick, irregular rind of enhancement is more suggestive of tumour. Abscesses are more likely to show small satellite lesions. Despite initial hopes that restricted diffusion would reliably distinguish abscess from tumour, reduced ADC has been subsequently reported in metastases and glioblastomas. Dynamic contrast-enhanced perfusion MRI may help distinguish between brain abscess and tumour; abscesses have a lower relative cerebral blood volume in their enhancing rim than gliomas6. The management of brain abscesses can require both medical and neurosurgical therapy. CT diagnosis has been responsible for a marked reduction in the mortality of brain abscesses. Follow-up imaging is recommended at biweekly intervals or when new symptoms arise. Sufficient treatment is indicated by resolution of rim enhancement or disappearance of the low signal rim on T2W images. Treatment response may be better assessed with DWI than conventional MRI; low signal on DWI correlates with a good clinical response whilst increasing signal implies reaccumulation of pus5. Fungal cerebral abscesses are seen most typically in immunocompromised patients. They are radiologically similar to pyogenic abscesses (Fig. 58.3) but more likely to show areas of haemorrhage.

may coexist7. An empyema may accompany a cerebral abscess and, if unrecognized, delay response to treatment. CT typically shows a thin fluid collection, slightly denser than CSF, overlying a convexity or in the interhemispheric fissure. Enhancement of the margin of the empyema is characteristic; IV contrast medium is therefore recommended when this diagnosis is suspected. MRI is more sensitive, showing an extracerebral lesion with prolonged T1 and T2. Contrast enhancement may be considerable. Empyemas tend to be loculated, multiple or complex (Fig. 58.4).

INTRACRANIAL EMPYEMA

Involvement of the central nervous system occurs in 5 per cent of cases of tuberculosis, and is especially common in patients younger than 20 years. The chest radiograph is abnormal in 45–60 per cent1. Tuberculous meningitis is the most frequent manifestation and tends to involve the basal leptomeninges. CT shows obliteration of the basal cisterns by isodense or slightly hyperdense exudate, which shows diffuse enhancement with IV contrast medium.The most useful CT criteria of abnormal basal meningeal enhancement are: (A) linear enhancement of the middle cerebral artery cisterns; (B) obliteration by contrast of the CSF spaces around normal vascular enhancement; (C) Y-shaped enhancement at the junction of the suprasellar and middle cerebral artery cisterns and (D) asymmetry of enhancement9. The meningeal exudate obstructs CSF resorption and causes communicating hydrocephalus; this is seen in 50 per cent of adults and 85 per cent of children1. Infarctions of the basal ganglia and internal capsules can occur, caused by an arteritis of the penetrating arteries at the base of the brain. With healing, meningeal calcification may rarely be seen. MRI (Fig. 58.6) depicts the basal meningeal enhancement, hydrocephalus and basal ganglia infarcts with greater sensitivity than CT. The differential diagnosis includes fungal meningitis, sarcoid and carcinomatous meningitis. Tuberculomas (parenchymal granulomas) occur most often at the corticomedullary junction. On CT they appear as small, rounded lesions isodense or hypodense to brain, with variable amounts of surrounding oedema. Enhancement is homogeneous when lesions are solid and shows rim enhancement when central caseation or liquefaction occurs. The ‘target sign’ of central high attenuation with rim enhancement is not pathognomonic for tuberculoma. On MRI small tuberculomas show prolonged T1 and T2; caseation results in low signal on

Intracranial empyema is usuaully caused by spread of infection from the paranasal sinuses or ears. Trauma, meningitis and complications of intracranial shunts are additional causes. Empyemas are more often subdural than extradural, but both

Figure 58.3 Fungal abscess in a 48-year-old diabetic patient. MRI. (A) Axial T2W image. Central high-signal abscess cavity with surrounding vasogenic oedema. (B) Coronal post-gadolinium T1W image. Large multiloculated abscess cavity with enhancement of the capsule and abscess wall. Note relative thinness of the medial wall compared with the thicker, more irregular, lateral component. Mild mass effect is evident.

VENTRICULITIS Ventriculitis is uncommon. Causes include trauma, intraventricular rupture of an abscess, shunt infection and haematogenous spread of infection to the ependyma or choroid plexus. On MRI (Fig. 58.5) the most frequent finding is intraventricular debris, seen as increased signal on fluidattenuated inversion recovery (FLAIR) and DWI sequences. Periventricular and subependymal high signal and enhancement of the ventricular margins are less common8.

TUBERCULOSIS

1327

1328

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 58.4 Subdural empyema. (A) Axial contrast-enhanced CT shows a low attenuation subdural fluid collection abutting the posterior falx with peripheral enhancement. (B) Coronal T1W post-gadolinium MRI shows the empyema is loculated and also extends along the right tentorial leaflet.

Figure 58.5 Ventriculitis: MRI. (A) Subependymal enhancement, most marked posteriorly, extends along the margins of the dilated ventricles. (B) Diffusion-weighted image shows restricted diffusion as high signal.

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

Figure 58.6 Tuberculous meningitis. Contrast-enhanced T1W MRI shows basilar meningeal enhancement, and multiple ring-enhancing tuberculomas in the suprasellar and ambient cisterns and the medial Sylvian fissures. Marked dilatation of the temporal horns indicates hydrocephalus.

T2W images. Tuberculomas may calcify when healed, but, as with meningeal disease, this is uncommon. Tuberculous abscesses are uncommon; they may resemble tuberculomas but are usually larger with a thinner enhancing rim10.

Figure 58.7 Herpes simplex encephalitis. Axial T2W MRI shows swelling and high signal in the anteromedial right temporal lobe with normal appearance on the left.

patchy white matter oedema and cortical areas of increased density on CT and low signal on T2W MRI. Lesions progress to areas of multicystic encephalomalacia.

Acute disseminated encephalomyelitis

Most encephalitides are caused by diffuse inflammation of the brain parenchyma by viral infection. Other causes include tick-borne bacterial infection, post-infectious autoimmune disorders and paraneoplastic limbic encephalitis.

Acute disseminated encephalomyelitis (ADEM) is a monophasic demyelinating disorder that occurs after vaccination or a viral illness. It has a fulminant course, results in encephalopathy and focal neurological deficits, and usually resolves without longterm sequelae. MRI typically shows multiple, large, irregular T2hyperintense lesions in the subcortical white matter, cerebellum or brainstem14. Lesion enhancement is variable. Lesions of the thalamus, basal ganglia and spinal cord occur infrequently.

Herpes encephalitis

Acute haemorrhagic leukoencephalopathy

Herpes simplex type 1 is the most frequent cause of viral encephalitis (Fig. 58.7) in adults and is often fatal without treatment. It results from reactivation of latent infection in the trigeminal ganglion, or re-infection by the olfactory route. CT appears normal in the first 3–5 days after onset before showing low attenuation in the antero-medial temporal lobe, with or without involvement of the insula and orbital surface of the frontal lobe. Haemorrhage is seen as a late feature and not usually a prominent finding. Enhancement may be patchy or gyriform. MRI is more sensitive;T2W and FLAIR sequences show high signal within 2 d of onset11.The abnormal signal is mainly cortical, with secondary involvement of the subjacent white matter. MRI is also more sensitive than CT to haemorrhagic foci. DWI shows cortical hyperintensity with greater sensitivity than conventional MRI12. Cerebral blood flow, measured by perfusion CT or SPECT, is increased in the acute phase13. Neonatal herpes simplex encephalitis is caused by intrapartum infection with herpes simplex virus type 2. Imaging shows

This aggressive variant of ADEM is frequently fatal within 1 week of onset. Appearances are similar to ADEM but with more oedema and mass effect, and small haemorrhages15.

ENCEPHALITIS

Subacute sclerosing panencephalitis Usually caused by the measles virus, this process characteristically produces cerebral atrophy in addition to low CT attenuation and increased MRI relaxation times in the white matter16. However, the diagnosis is made from clinical and electroencephalogram (EEG) features rather than radiologically.

MENINGITIS CT is usually normal in uncomplicated pyogenic meningitis, but it is useful for detecting complications such as hydrocephalus, subdural effusion, abscess or cerebral infarction. MRI may show sulcal high signal on FLAIR images but this is nonspecific and is

1329

1330

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

also seen with subarachnoid haemorrhage and leptomeningeal metastases17. Meningeal enhancement is more sensitive; contrastenhanced FLAIR may be better than contrast-enhanced T1W imaging in its depiction18.

PARASITIC INFESTATIONS Neurocysticercosis has varied imaging appearances depending on the developmental stage of the larva. Four stages are recognized and lesions in different stages may be seen concurrently. Cysticerci are ovoid cysts that contain an invaginated larval head, or scolex. Intracranial cysticerci are most frequent at the corticomedullary junction; intraventricular and racemose subarachnoid lesions also occur. In the vesicular stage the parasite is alive and incites little or no perilesional oedema. The cyst is isointense to CSF, the scolex isointense to white matter (Fig. 58.8). Enhancement is minimal or absent. In the colloidal vesicular stage the larva dies and lesions show ring enhancement with surrounding oedema. Cyst retraction in the granular nodular stage results in small enhancing nodules with mild oedema. Finally, in the nodular calcified stage, the lesions calcify. Subarachnoid cysticerci, which usually lack a scolex, may cause obstructive hydrocephalus19. Hydatid cysts are most often solitary and found in the middle cerebral artery territory. CT and MRI show a welldefined spherical lesion with the attenuation and signal characteristics of CSF. The cyst wall is hypointense on T2W images. Enhancement and perilesional oedema are seen only if the cyst is superinfected. Fewer than 1 per cent of cysts calcify20.

Figure 58.8 Cysticercosis: Axial FLAIR image shows a left frontal lobe cysticercus in the vesicular stage. Note the scolex surrounded by cyst fluid isointense to CSF. There is no perilesional oedema.

HIV INFECTION AND AIDS After the lung, the central nervous system (CNS) is the organ most frequently affected by the human immunodeficiency virus (HIV). Postmortem studies show CNS abnormalities in up to 70 per cent of acquired immune deficiency syndrome (AIDS) patients. Neurological symptoms in HIV infection and AIDS occur because of opportunistic infections, the effects of HIV itself, and adverse effects of therapy. Highly active antiretroviral therapy (HAART) has decreased both mortality rates and the incidence of opportunistic infections in HIV-infected patients. Initial symptoms may be nonspecific and neurological signs difficult to elicit. Serology is often unhelpful because many opportunistic infections are reactivations of previous infections and the patient may not mount an immune response. Neuroimaging is thus a fundamental part of the assessment of HIV-infected patients with suspected CNS disease. Asymptomatic HIV-infected patients have MRI abnormalities in only 20 per cent21. Abnormalities are increasingly common with significant immune suppression, in the later stages of disease and as patients live longer. Diseases affecting the CNS that have been described in HIV infection and AIDS are listed in Table 58.1, roughly grouped according to the likelihood of encountering such cases in radiological practice. This may be modified according to the HIV risk group of the patient: for example, CNS

tuberculosis is most often seen in those with a history of IV drug misuse. MRI is more sensitive than CT in evaluating HIV-infected patients with suspected CNS disease, and as such is the imaging investigation of choice. Small lesions, and cortical or subcortical lesions, are shown with greater sensitivity by FLAIR than T2W images22. Whilst IV contrast medium improves the characterization of lesions, it infrequently reveals lesions that are not visible on unenhanced images and the clinical value of its routine use is unproven23. The commonest AIDS-related diseases seen on MRI may be classified according to simple radiological patterns (Table 58.2). The following text will discuss in more detail the typical and atypical manifestations of the more common CNS diseases.

HIV ENCEPHALOPATHY HIV-associated cognitive-motor complex (also termed HIV encephalopathy, HIV-associated dementia or AIDS dementia complex) presents with cognitive impairment, behavioural change and motor symptoms. The clinical picture is of a subcortical dementia with slowness, forgetfulness and apathy. The prevalence of HIV encephalopathy is 10–20 per cent of AIDS

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

Table 58.1 CNS DISEASE IN HIV INFECTION/AIDS Commonest

Less common

Rarest reported

Table 58.2 COMMON TYPICAL MRI PATTERNS IN HIV-RELATED CNS DISEASE Focal parenchymal

HIV encephalopathy

Mycobacterial (TB/MAC)

Syphilis

Toxoplasmosis

Candida

Nocardia*

+mass +enhancement *

Primary lymphoma

Cytomegalovirus

Histoplasmosis

Toxoplasmosis

PML

Herpes simplex

Aspergillosis

Lymphoma

Cryptococcosis

Varicella zoster

Mucormycosis

Tuberculosis

Infarcts

Coccidioidomycosis

Candida

Metastatic lymphoma

Amoebiasis

Others—histoplasmosis, aspergillosis, coccidioidomycosis

Trypanosomiasis

+mass –enhancement

Protothecosis

Cryptococcomas

Measles

–mass +enhancement

Adenovirus

Diffuse toxoplasmosis

Kaposi’s sarcoma

Viral encephalitides—HSV, CMV

Metastatic carcinoma

Infarcts—± VZV arteritis

Metabolic encephalopathy

–mass –enhancement

*

Higher incidences in North America

PML

Diffuse parenchymal

cases and is rising as patients live longer. The incidence has halved since the introduction of HAART. Multinucleate giant cells and microglial nodules characterize HIV encephalitis, the pathological correlate of HIV encephalopathy. Myelin loss, macrophage infiltration and gliosis are also seen. HIV encephalitis occurs because of cerebral infection with HIV. The commonest imaging finding in HIV encephalopathy is cerebral atrophy, the extent of volume loss correlates with cognitive impairment24. White matter lesions in the centrum semiovale and periventricular regions are the next most frequent abnormality. These appear as areas of low attenuation on CT, and T2-prolongation on MRI, which lack mass effect and do not enhance25. The white matter changes tend to progress with time, becoming diffuse and confluent.Atrophy and white matter lesions can coexist or occur independently of one another (Fig. 58.9). MR spectroscopy (MRS) shows decreased N-acetyl aspartate (NAA), reflecting neuronal loss, increased choline, a marker of membrane turnover, and increased myoinositol, a glial cell marker. These abnormalities can be detected in patients with normal MRI. Cognitively normal HIV-infected patients may also show increased choline and myoinositol, but little or no change in NAA. The spectroscopic abnormalities can reverse with HAART26. PET and SPECT may show hypermetabolism in the basal ganglia and thalami in patients with a normal MRI; these abnormalities correlate with neuropsychiatric measures of dementia. In advanced disease cortical hypometabolism is seen. Although the sensitivity of these techniques is high, the specificity is undetermined and the role in clinical practice is not established. Diffusion tensor MR imaging (DTI) shows reduced whole-brain fractional anisotropy (FA) in cognitively impaired HIV-infected patients.The reduction in FA correlates with the severity of cognitive impairment27. Patients receiving HAART may show stabilization or even regression of MRI abnormalities. Early follow-up imaging may show lesion progression but this is not indicative of treatment failure28.

HIV encephalopathy

Meningitis/meningeal disease HIV meningo-encephalitis Cryptococcosis Metastatic lymphoma Viral meningitis

Ventriculitis CMV necrotizing ventriculitis

Basal ganglia +mass Toxoplasmosis Cryptococcosis –mass Metabolic encephalopathy

White-matter disease Small nonspecific focal white-matter hyperintensities on T2W HIV encephalopathy PML Viral encephalitis—CMV

Myelopathy CMV VZV Vacuolar myelopathy CMV = cytomegalovirus; VZV = Varicella zoster virus; HSV = Herpes simplex virus.

CEREBRAL TOXOPLASMOSIS Toxoplasmosis is the commonest cause of a cerebral mass lesion in AIDS and also the most treatable. It results from reactivation of latent infection by Toxoplasma gondii, for which the definitive host is the cat. Patients present subacutely with headache, fever, confusion, personality change and focal

1331

1332

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 58.9 Advanced HIV encephalopathy. Axial T2W image. There is diffuse confluent and symmetrical abnormal high signal returned from the white matter of the cerebral hemispheres (A), which is also extending into the brainstem to involve the cerebral peduncles (B). In this patient there is also generalized atrophy. Features that help to differentiate HIV from PML are the symmetry of the changes and the lack of signal abnormalities on T1W images (cf. Fig. 58.17).

neurological deficits. Histopathology shows a multifocal haemorrhagic necrotizing encephalitis with the development of organizing abscesses. Imaging typically shows multiple lesions, 1–4 cm across, at the corticomedullary junction and in the basal ganglia (Fig. 58.10). Single lesions and lesions in the brainstem or cerebellum are uncommon.The lesions show ring or nodular enhancement with associated oedema and mass effect29 (Fig. 58.11). Enhancement may be diminished or absent in severely immunocompromised patients. MRI is more sensitive than CT to foci of haemorrhage within the lesions. The principal differential diagnosis is from primary CNS lymphoma, which can appear identical and can coexist in the same patient (Fig. 58.12). In comparison to pyogenic abscesses cerebral toxoplasmosis is hypointense to white matter on DWI, indicating no restriction of diffusion30. A diffuse form of toxoplasmosis is not infrequent at postmortem examination. It appears as ill-defined foci of T2 prolongation at the corticomedullary junction, with mild and patchy enhancement. Lifelong treatment is necessary, usually with pyrimethamine and sulfadiazine. Most lesions show reduced enhancement, oedema and mass effect within 2–4 weeks. Small lesions may heal completely whilst larger lesions can show persistent enhancement for more than 2 years on maintenance therapy.

PRIMARY CEREBRAL LYMPHOMA

Figure 58.10 Typical toxoplasma abscesses and response to treatment. Transverse T2W images (A, C, D) and coronal T1W image (B). Multiple masses of varying sizes with a propensity to involve the basal ganglia and corticomedullary junction and associated with perilesional oedema may occur (A). High signal seen on the T1W images is due to haemorrhage (B). In response to therapy for toxoplasma (C, D), the size of the lesions and the surrounding oedema are reduced. Responding lesions may show increased intensity on T2W images and some show a surrounding low signal rim due to haemosiderin (arrow).

Primary cerebral lymphoma is the AIDS-defining diagnosis in 5 per cent of patients. The incidence has reduced in the era of HAART. Most patients present with rapid progression of confusion, lethargy, memory loss and focal neurology. Cerebral lymphoma is often multifocal in AIDS; lesions are commonest in the cerebral white matter but also occur in the basal ganglia, corpus callosum and ventricular margins (Figs 58.13, 58.14). Lesions abut the ependyma, leptomeninges or both in 75 per cent. Imaging shows well-defined round or oval lesions of high attenuation on unenhanced CT, and lower signal intensity than grey matter on T2W MRI. This reflects the dense cellularity of lymphoma. Lesions have relatively little mass effect and oedema for their size31. Haemorrhage is unusual and calcification seen only after treatment. Enhancement is typical, often in a smooth or nodular ring surrounding a zone of central necrosis, in contrast to the solid enhancement seen in immune-competent patients. Whilst they can be indistinguishable, there are some features that favour a diagnosis of lymphoma over toxoplasmosis. A single enhancing mass lesion in AIDS is more likely to be lymphoma, as are larger lesions and those with central low intensity on T2W images. Subependymal spread is a feature of lymphoma but is not seen in toxoplasmosis32. Thallium-201 SPECT and FDG-PET show greater uptake in lymphoma than toxoplasmosis though this is unreliable in lesions smaller than 2 cm33. Although lymphoma sometimes shows restricted diffusion, ADC values often overlap with those of toxoplasmosis, limiting the value of DWI in distinguishing between the two34.

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

Figure 58.11 Enhancement in toxoplasmosis. (A) Axial T2W and (B) gadolinium-enhanced T1W images. Toxoplasma abscess in the right thalamus shows extensive surrounding vasogenic oedema and irregular peripheral enhancement.

Figure 58.13 Multifocal primary cerebral lymphoma. Transverse T2W spin-echo image. Multiple masses, most of which show mixed signal intensity on T2W images, are present; like multiple toxoplasmosis, they involve the basal ganglia. However, subependymal tumour spread is clearly seen around the lateral and the 4th ventricles (arrows), which favours the diagnosis of lymphoma.

Figure 58.12 Coexistent lymphoma and toxoplasmosis. Transverse T2W spin-echo images (A, C) and enhanced coronal T1W spin-echo image. Dual or triple disease processes frequently occur in AIDS-related neurological disease. In this patient cerebral toxoplasmosis (A, B – arrow) and lymphoma involving the pineal (C, D – curved arrows) were confirmed at postmortem.

Metastases from systemic lymphoma typically involve the meninges (Fig. 58.15); parenchymal disease without leptomeningeal involvement is rare. Lymphoma may respond dramatically to radiotherapy and/ or corticosteroids but usually the prognosis is poor, HAART has prolonged median survival from 2–8 months.

CRYPTOCOCCOSIS This is the second commonest opportunistic CNS infection in AIDS. Patients most often present with headache, fever and altered mental state. The earliest imaging manifestation is dilatation of perivascular spaces, most often in the basal ganglia but also in the brainstem and cerebral white matter (Fig. 58.16). These spaces are distended by mucoid material, organisms and inflammatory cells, and appear as multiple small foci of high signal on T2W images35. With disease progression cryptococcomas develop at these sites, forming lesions 3 mm to several cm in size. Cryptococcomas are of low to intermediate signal

1333

1334

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 58.14 Primary cerebral lymphoma involvement of the corpus callosum. Transverse T2W spin-echo image (A) and coronal enhanced T1W spin-echo image (B). Lymphomatous masses may involve the corpus callosum, as in this patient who had multifocal primary cerebral lymphoma. Rim enhancement of the mass is seen.

defects, speech abnormalities, ataxia and dementia. Pathologically, there is demyelination, astrocytosis and oligodendrocytes with intranuclear inclusions. Lesions can occur in any part of the brain but are commonest in the parieto-occipital regions. MRI shows multifocal, asymmetric bilateral white matter lesions that are of high signal on T2W and low signal on T1W images (Fig. 58.17). Extension to the subcortical U-fibres gives the lesions a characteristic ‘scalloped’ appearance. Apparent involvement of the basal ganglia can occur when lesions affect the white matter tracts that course through this site. Rarely, lesions may show mild mass effect and peripheral enhancement; restricted diffusion can be observed in areas of active disease progression37.

Figure 58.15 Metastatic lymphoma. Transverse T2W spin-echo image. The lymphomatous deposit is based on, and is lifting, the dura (arrow). There is oedema in the underlying brain substance, which is displaced.

on T1W and high signal on T2W images, lack surrounding oedema and do not show restricted diffusion36. Enhancement of cryptococcomas or the leptomeninges is rare because these patients are profoundly immunocompromised.

PROGRESSIVE MULTIFOCAL LEUKOENCEPHALOPATHY Progressive multifocal leukoencephalopathy (PML) is a central demyelinating disease resulting from the reactivation of a latent infection of oligodendrocytes by JC polyomavirus. Eighty per cent of adults show serological evidence of prior exposure to the JC virus. The incidence of PML is about 4–5 per cent of AIDS cases; it is the AIDS-defining illness in about 1 per cent. Clinically, limb weakness is the commonest presentation, with an insidious onset and a progressive course involving visual field

TUBERCULOSIS (See earlier in this Chapter) Intracranial tuberculosis is mostly seen amongst IV drug misusers. The radiological manifestations are similar to those in immunocompetent patients, hydrocephalus and meningeal enhancement being the commonest. Parenchymal lesions, in the form of tuberculomas and abscesses, are more frequent in HIV infection10.

CANDIDIASIS Although mucocutaneous candidiasis is common in HIVinfected patients, CNS involvement is rare. Haematogenous dissemination results in meningitis and/or cerebral abscesses. Imaging appearances are nonspecific; clinical confirmation is dependent on CSF analysis or brain biopsy.

INCIDENTAL WHITE-MATTER HYPERINTENSITIES Focal white-matter hyperintensities, often multiple, are seen in up to 26 per cent of HIV-positive patients and up to 24

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

Figure 58.16 Cryptococcomas. Transverse T2W spin-echo images (A, B) and coronal T1W spin-echo image (C). Expanded Virchow-Robin spaces (arrow) of high signal on T2 and low on T1 are seen in the brainstem and the basal ganglia, so that the ganglia look like ‘Swiss cheese’.

Figure 58.17 Progressive multifocal leukoencephalopathy. Axial T2W fast spin-echo image (A), FLAIR (B) and T1W spin-echo image (C). Asymmetrical signal abnormalities in the parieto-occipital white matter of both hemispheres extend to the subcortical U-fibres. There is no mass effect associated with the lesions.

per cent of seronegative men of matched ages. No associations with neurological abnormalities, CD4 count, or vascular risk factors have been identified. These lesions are probably incidental and of no clinical significance38.

HERPES VIRUSES Cytomegalovirus, herpes simplex and varicella zoster viruses can cause encephalitis, necrotizing ventriculitis (Fig. 58.18), and myelitis in AIDS. In encephalitis imaging may be normal, show nonspecific white matter lesions or focal enhancing lesions. Ependymal enhancement occurs with ventriculitis; myelitis manifests as nonspecific swelling and signal change in the spinal cord39.

CEREBROVASCULAR DISEASE Cerebral infarcts occur in fewer than 5 per cent of AIDS patients. Causes include infective vasculitis (CMV, varicella zoster or tuberculosis) and embolism from HIV cardiomyopathy. HIV also causes a dilating vasculopathy that results in fusiform aneurysms of the intracranial vessels40.

HISTOPLASMOSIS Histoplasmosis occurs in up to 5 per cent of AIDS patients in areas where Histoplasma capsulatum is endemic. CNS manifestations include meningitis with involvement of adjacent vessels,

1335

1336

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

from the meninges, and appear as mass lesions with variable MR signal characteristics and enhancement41.

SPINAL CORD DISORDERS42

Figure 58.18 Herpes simplex ventriculitis. Transverse T2W spinecho image (A) and coronal T1W spin-echo image (B) after IV contrast medium. The necrotic right-sided periventricular lesion shows central low signal on T1 and high on T2 with peripheral enhancement. Enhancement of the periventricular tissues (note the involvement of the corpus callosum) is also present in this case.

AIDS-associated vacuolar myelopathy presents insidiously and progresses to severe paraparesis; the thoracic cord is most often affected. MRI is usually normal or shows nonspecific changes such as diffuse symmetrical signal abnormalities in the cord. Primary HIV myelitis is rare and presents acutely with paraparesis and a sensory level; MRI shows multifocal asymmetrical signal change in the cord. Other diseases affecting the spinal cord in AIDS include herpes virus infections (Fig. 58.19), toxoplasmosis and tuberculosis.

and single or multiple abscesses. Imaging may show meningeal enhancement, cerebral infarcts, or focal enhancing lesions with mass effect and oedema39.

NEUROSYPHILIS CNS involvement can occur at any stage of syphilis, in HIVinfection its course may be more aggressive. Meningovascular syphilis causes a small-vessel endarteritis that appears as arterial segmental ‘beading’ on angiography, with associated infarcts in the basal ganglia. Cerebral gummas are rare, typically arise

Figure 58.19 Herpes zoster cord myelitis. Axial T2W gradient echo at level of T8. Following an attack of shingles in a T8 distribution, this patient developed cord symptoms. Focal high signal is seen involving the dorsal columns (arrows).

DEMYELINATING, METABOLIC AND NONSPECIFIC INFLAMMATORY DISORDERS There are innumerable diseases in which myelin is formed abnormally or is destroyed. Most such familial and metabolic disorders occur in infancy and childhood with nonspecific imaging findings. The reader is referred elsewhere for an account beyond the scope of this chapter43. Congenital leukodystrophies usually manifest in childhood, though adrenoleukodystrophy (ALD) and Krabbe’s disease may not present until adulthood. White matter lesions in the leukodystrophies are typically symmetrical, manifest on CT as low attenuation, and on MRI as increased signal on T2W images and often less extensive decreased signal on T1W images. ALD usually begins posteriorly, and Krabbe’s disease often involves the pyramidal tracts. The aminoacidurias and mitochondrial cytopathies affect grey as well as white matter or instead of white matter. Some mitochondrial disorders may not present until late

adulthood. A striking feature is infarct-like lesions in the basal ganglia and cortex that are not confined to vascular territories and may be transient.

WILSON’S DISEASE Mutations in the ATP7B gene, which encodes a copper transporter, cause Wilson’s disease. Copper first accumulates in the liver and then the brain, producing multifocal necrosis. The commonest MRI finding is high signal in the putamen on T2W images; lesions also occur in the pons, midbrain, cerebellum and subcortical white matter. Lesions initially show restricted diffusion44 and diminish in size with copper chelation. Some patients with hepatic failure develop symmetrical lesions in basal ganglia and cerebral white matter, a notable

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

feature being increased signal in basal ganglia on T1W images.

INTESTINAL ENCEPHALOPATHIES Ataxia is the commonest neurological complication of coeliac disease; MRI shows cerebellar atrophy in 79 per cent and nonspecific white matter lesions in 19 per cent45. Bilateral calcification in the parieto-occipital cortex, associated with seizures, has also been reported in coeliac disease46. Intracranial venous and dural sinus thrombosis may occur, especially in ulcerative colitis.

MULTIPLE SCLEROSIS Multiple sclerosis (MS) is an inflammatory disease of the CNS characterized pathologically by demyelination and axonal injury. Relapsing-remitting MS is the commonest form: patients experience episodic neurological deficits with intervening partial or complete recovery. The diagnosis requires evidence of dissemination of lesions in space and time, and is primarily clinical. Paraclinical studies including CSF analysis, visual evoked responses and MRI can be used to support the diagnosis47 (Table 58.3). Lesions can occur anywhere in the CNS but are commonest in the periventricular white matter. Typical lesions are ovoid, with the long axis orientated perpendicular to the ventricle wall (Fig. 58.20). Such lesions are better shown by proton density and FLAIR than T2W images because of increased lesionto-CSF contrast. Lesions of the corpus callosum, both at the callososeptal interface and subcallosal striations, are characteristic, and best shown on sagittal FLAIR images. FLAIR is less sensitive than T2W imaging for infratentorial lesions, which tend to occur in the brainstem and middle cerebellar peduncles (Fig. 58.21). Spinal cord involvement is common; lesions

are generally less than two vertebral segments long and aligned with the long axis of the cord. Acute lesions can show solid or ring enhancement with IV gadolinium, which can persist for up to 3 months but generally resolves within weeks. Occasionally, large acute lesions with associated oedema and mass effect can mimic glioma (Fig. 58.22). The enhancement in such cases of tumefactive MS often forms an incomplete ring48. Lesions show reduced magnetization transfer ratio (MTR), reflecting decreased amounts of myelin. MTR is also reduced in normal-appearing white matter of MS patients compared with controls. This ‘occult’ tissue damage can also be detected by diffusion tensor imaging, which shows reduced fractional anisotropy49. Most MS lesions seem asymptomatic; disability correlates poorly with T2 lesion load but rather better with the number of low signal lesions on T1W MRI (‘black holes’). Better correlation with disability has also been found with brain and spinal cord atrophy, which develops later in the disease. The differential diagnosis for the imaging appearance of MS includes ADEM, vasculitis (e.g. lupus, anti-phospholipid syndrome, Behçet’s disease), sarcoidosis and white matter lesions associated with small vessel ischaemia. These latter processes tend to spare the subcortical U-fibres, which are affected by MS lesions.

OSMOTIC MYELINOLYSIS50 Precipitous correction of severe hyponatraemia can cause acute demyelination in the central pons, and also in extrapontine sites such as the cerebellum, subinsular regions, basal ganglia and thalami. MRI is often normal initially, before showing swelling and high signal in the basal pons on T2W images. DWI is more sensitive in the acute stage, showing restricted diffusion as early as 24 h.

Table 58.3 MRI CRITERIA TO DEMONSTRATE LESION DISSEMINATION IN SPACE AND TIME 1.1.1

Dissemination in space

Three of the following: 1. At least one gadolinium-enhancing lesion or nine T2 hyperintense lesions if there is no gadolinium-enhancing lesion 2. At least one infratentorial lesion 3. At least one juxtacortical lesion 4. At least one periventricular lesion Note: A spinal cord lesion can be considered equivalent to a brain infratentorial lesion: an enhancing spinal cord lesion is considered to be equivalent to an enhancing brain lesion, and individual spinal cord lesions can contribute together with individual brain lesions to reach the required number of T2 lesions.

1.1.2

Dissemination in time

There are two ways to show dissemination in time using imaging: 1. Detection of gadolinium enhancement at least 3 months after the onset of the initial clinical event, if not at the site corresponding to the initial event. 2. Detection of a new T2 lesion if it appears at any time compared with a reference scan done at least 30 days after the onset of the initial clinical event.

1337

1338

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 58.20 Multiple sclerosis. (A) Axial proton density and (B) gadolinium-enhanced T1W images show multiple lesions in the periventricular white matter, two of which are in the acute phase and enhance.

SARCOIDOSIS51

Figure 58.21 Infratentorial MS lesion. Axial T2W fast spin-echo image shows a typical MS lesion in the left middle cerebellar peduncle.

Symptomatic CNS involvement occurs in 5 per cent of patients with this multi-system granulomatous disorder; MRI is the investigation of choice for assessment. Meningeal disease is the most frequent finding; dural thickening and masses that can mimic meningioma, and enhancement of the basal and suprasellar meninges are well-recognized. Small enhancing granulomas are usually located superficially in brain parenchyma bordering the basal cisterns. Nonenhancing lesions in the periventricular white matter and brainstem are common and can mimic MS. Less often, subependymal granulomatous infiltration causes hydrocephalus. In over 80 per cent of established cases the chest radiograph is abnormal. About 70 per cent of cases have skin, or visceral involvement, especially the liver.

BEHÇET’S DISEASE52 CNS involvement most often occurs as a chronic meningoencephalitis. Lesions tend to occur in the brainstem, diencephalon, basal ganglia and deep hemispheric white matter, and may resemble those of MS. Brainstem atrophy is seen in chronic cases.

CHAPTER 58

•

CRANIAL AND INTRACRANIAL DISEASE: INFECTIONS, AIDS, INFLAMMATORY, DEMYELINATING AND METABOLIC DISEASES

Figure 58.22 Tumefactive MS. (A) Axial T2W fast spin-echo and (B) gadolinium-enhanced T1W spin-echo images show a large MS plaque with considerable associated oedema in the right frontal lobe. Note the smaller periventricular lesions on the T2W image, and the ‘open ring’ of contrast enhancement.

REFERENCES 1. Roos K L 2005 Principles of neurologic infectious disease. McGraw Hill, New York 2. Stevens E A, Norman D, Kramer R A et al 1978 Computed tomographic brain scanning in intraparenchymal pyogenic abscesses. Am J Roentgenol 130:111–114 3. Enzmann D R, Britt R H, Placone R 1983 Staging of human brain abscesses by computed tomography. Radiology 146:703–708 4. Haimes A, Zimmerman R D, Morgello S et al 1989 MR imaging of brain abscesses. Am J Roentgenol 152:1073–1085 5. Cartes-Zumelzu F W, Stavrou I, Castillo M et al 2004 Diffusionweighted imaging in the assessment of brain abscess therapy. Am J Neuroradiol 25:1310–1317 6. Holmes T M, Petrella J R, Provenzale J M 2004 Distinction between cerebral abscesses and high-grade neoplasms by dynamic susceptibility contrast perfusion MRI. Am J Roentgenol 183:1247–1252 7. Rich P M, Deasy N P, Jarosz J M 2000 Intracranial dural empyema. Br J Radiol 73:1329–1336 8. Fujikawa A, Tsuchiya K, Honya K et al 2006 Comparison of MRI sequences to detect ventriculitis. Am J Roentgenol 187:1048–1053 9. Pryzbojewksi S, Andronikou S, Wilmhurst J 2006 Objective CT criteria to determine the presence of abnormal basal enhancement in children with suspected tuberculous meningitis. Paediar Radiol 36:687–696 10. Bernaerts A, Vanhoenacker F M, Parizel P M et al 2003 Tuberculosis of the central nervous system: overview of neuroradiological findings. Eur Radiol 12:1876–1890 11. Tien R D, Feisberg G J, Osumi A K 1993 Herpesvirus infections of the CNS: MR findings. Am J Roentgenol 161:167–176

12. Küker W, Nägele T, Schmidt et al 2004 Diffusion-weighted MRI in herpes simplex encephalitis: a report of three cases. Neuroradiology 46:122–125 13. Marco de Lucas E, Gonzalez Mandly A, Gutierrez A et al 2006 Computed tomography perfusion usefulness in early imaging diagnosis of herpes simplex encephalitis. Acta Radiol 47:878–871 14. Singh S, Alexander M, Korah I P 1999 Acute disseminated encephalopmyelitis: MR imaging features. Am J Roentgenol 173:1101–1107 15. Gibbs W N, Kreidie M A, Kim R C et al 2005 Acute haemorrhagic leukoencephalitis: neuroimaging features and neuropathologic diagnosis. J Comp Assist Tomogr 29:689–693 16. Ozturk A, Gurses C, Baykan B et al 2002 Subacute sclerosing panencephalitis: clinical and magnetic resonance imaging evaluation of 36 patients. J Child Neurol 17:25–29 17. Kamran S, Bari Bener A, Alper D et al 2004 Role of fluid-attenuated inversion recovery in the diagnosis of meningitis: comparison with contrast-enhanced magnetic resonance imaging. J Comp Assist Tomogr 28:68–72 18. Splendani A, Puglielli E, De Amicis R et al 2005 Contrast-enhanced FLAIR in the early diagnosis of infectious meningitis. Neuroradiology 47:591–598 19. Noujaim S E, Rossi M D, Rao S K et al 1999 CT and MR imaging of neurocysticercosis. Am J Roentgenol 173:1485–1490 20. Tüzün M, Hekimoglu B 1998 Hydatid disease of the CNS: imaging features. Am J Roentgenol 171:1497–1500 21. Post M J D, Berger J R, Duncan R, Quencer R M, Pall L, Winfield D 1993 Asymptomatic and neurologically symptomatic HIV-seropositive subjects: results of long term MR imaging and clinical follow-up. Radiology 188:727–733

1339

1340

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

22. Thurnher M M, Thurnher S A, Fleischmann D et al 1997. Comparison of T2-weighted and fluid-attenuated inversion recovery fast spin-echo MR sequences in intracerebral AIDS-associated disease. Am J Roentgenol 18:1601–1609 23. Malcolm P N, Howlett D C, Saks A et al 1999 MRI of the brain in HIVpositive patients: what is the value of routine intravenous contrast medium? Neuroradiology 41:687–695 24. Patel S H, Kolson D L, Glosser G et al 2002 Correlation between percentage of brain parenchymal volume and neurocognitive performance in HIV-infected patients. Am J Neuroradiol 23:543–549 25. Post M J D, Tate L G, Quencer R M et al 1988 CT, MR and pathology in HIV encephalitis and meningitis. Am J Roentgenol 151:373–380 26. Chang L, Ernst T 2005 Physiological MR to evaluate HIV-associated brain disorders. In: Gillard J, Waldman A, Barker P Clinical MR Neuroimaging: Diffusion, Perfusion and Spectroscopy. Cambridge University Press, Cambridge pp 460–478 27. Ragin A B, Storey P, Cohen B A et al 2004 Whole brain diffusion tensor imaging in HIV-associated cognitive impairment. Am J Neuroradiol 25:195–200 28. Thurnher M M, Schindler E G, Thurnher S A et al 2000 Highly active antiretroviral therapy for patients with AIDS dementia complex: effect on MR imaging findings and clinical course. Am J Neuroradiol 21:670–678 29. Post M J D, Kursunoglu S J, Hensely G T et al 1985 Cranial CT in acquired immunodeficiency syndrome: spectrum of diseases and optimal contrast enhancement technique. Am J Roentgenol 145:929–940 30. Camacho D L A, Smith J K, Castillo M 2003 Differentiation of toxoplasmosis and lymphoma in AIDS patients by using apparent diffusion coefficients. Am J Neuroradiol 24:633–637 31. Koeller K K, Smirniotopoulos J G, Jones R V 1997 Primary central nervous system lymphoma: radiologic–pathologic correlation. Radiographics 1497–1526 32. Dina T S 1991 Primary central nervous system lymphoma versus toxoplasmosis in AIDS. Radiology 179:823–828 33. Young R J, Ghesani M V, Kagetsu N J et al 2005 Lesion size determines accuracy of thallium-201 brain single-photon emission tomography in differentiating between intracranial malignancy and infection in AIDS patients. Am J Neuroradiol 26:1973–1979 34. Schroeder P C, Post M J, Oschatz E et al 2006 Analysis of the utility of diffusion-weighted MRI and apparent diffusion coefficient values in distinguishing central nervous system toxoplasmosis from lymphoma. Neuroradiology 48:715–720 35. Miszkiel K A, Hall-Craggs M A, Miller R F et al 1996 The spectrum of MRI findings in CNS cryptococcosis in AIDS. Clin Radiol 51:842–850 36. Ho T L, Lee H J, Lee K W et al 2005 Diffusion-weighted and conventional magnetic resonance imaging in cerebral cryptococcoma. Acta Radiol 46:411–414

37. Küker W, Mader I, Nägele T et al 2006 Progressive multifocal leukoencephalopathy: value of diffusion-weighted and contrast-enhanced magnetic resonance imaging for diagnosis and treatment control. Eur J Neurol 13:819–826 38. McArthur J C, Kumar A J, Johnson D W et al 1990 Incidental white matter hyperintensities on magnetic resonance imaging in HIV infection. Multicenter AIDS cohort study. J Acq Immun Def Synd 3:252–259 39. Berger J R, Cohen B A 2005 Opportunistic infections of the central nervous system in AIDS In: Gendelman H E, Grant I, Everall I P et al The Neurology of AIDS, 2nd edn. Oxford University Press, Oxford, pp 485–491 40. Corr P D 2006 Imaging of cerebrovascular and cardiovascular disease in AIDS patients. Am J Roentgenol 187:236–241 41. Brightbill T C, Ihmedian I H, Post M J D et al 1995 Neurosyphilis in HIV-positive and HIV-negative patients: neuroimaging findings. Am J Neuroradiol 16:703–711 42. Thurnher M M, Post M J, Jinkins J R 2000 MRI of infections and neoplasms of the spine and spinal cord in 55 patients with AIDS. Neuroradiology 42:551–563 43. Barkovich A J 2005 Pediatric neuroimaging, 4th edn. Lippincott Williams & Wilkins, Baltimore 44. Sener R N 2003 Diffusion MR changes associated with Wilson disease. Am J Neuroradiol 24:965–967 45. Hadjivassiliou M, Grunewald R, Sharrack B et al 2003 Gluten ataxia in perspective: epidemiology, genetic susceptibility and clinical characteristics. Brain 126:685–691 46. Pfaender M, D’Souza W J, Trost N et al 2004 Visual disturbances representing occipital epilepsy in patients with cerebral calcifications and coeliac disease: a case series. J Neurol Neurosurg Psychiatry 75:1623–1625 47. Polman C H, Reingold S C, Edan G et al 2005 Diagnostic criteria for multiple sclerosis: 2005 revisions to the ‘McDonald criteria.’ Ann Neurol 58:840–846 48. Pretorius P M, Quaghebeur G 2003 The role of MRI in the diagnosis of MS. Clin Radiol 58:434–448 49. Ye G 2006 Multiple sclerosis: the role of MR imaging. Am J Neuroradiol 27:1165–1176 50. Martin R J 2004 Central pontine and extrapontine myelinolysis: the osmotic demyelination syndrome. J Neurol Neurosurg Psychiatry 75: iii22–iii28 51. Christoforidis G A, Spickler E M, Recio M V et al 1999 MR of CNS sarcoidosis: correlation of imaging features to clinical symptoms and response to treatment. Am J Neuroradiol 20:655–669 52. Akman-Demir G, Bahar S, Coban O et al 2003 Cranial MRI in Behçet’s disease: 134 examinations of 98 patients. Neuroradiology 45:851–859

CHAPTER

Cranial and Intracranial Disease: Trauma, Cerebrospinal Fluid Disturbances, Degenerative Disorders and Epilepsy

59

John M. Stevens

Trauma to the skull and brain • Head injury • Primary cerebral damage in closed head injury • Other types of intracranial haemorrhage after closed head injury • Secondary cerebral damage with closed head injury • Other complications with closed head injuries Diseases of the skull—bone disease

Disturbances of the cerebrospinal fluid circulation • Communicating hydrocephalus Degenerative disorders • General aspects The dementias • Movement disorders Epilepsy • Acknowledgements

TRAUMA TO THE SKULL AND BRAIN HEAD INJURY Head injuries are either open (penetrating) or closed (nonpenetrating), the latter being far the more common in civilian practice.The main indication for imaging is suspected intracranial haemorrhage where prompt neurosurgical evacuation may modify outcome. Because it shows haemorrhage particularly well, computed tomography (CT) generally is recommended in preference to magnetic resonance imaging (MRI) for this purpose. Furthermore CT is more widely available on a 24-h basis and is easier to perform following major trauma. Thus there are clear recommendations from the Royal College of Radiologists1 and the National Institute of Clinical Excellence (NICE)2,3 about the indications for and appropriate timing of CT following trauma. During the subsequent clinical course, imaging may be required to assess neurological deterioration or other complications, or perhaps failure to improve, and later to make a final assessment of overall damage for long term prognosis. For many of these less acute indications, MRI may

be preferred. Despite the numerous published guidelines for imaging of the head and cervical spine in trauma, they are only guidelines and many individual brain injury units have their own variations. The principles behind these, however, are simply the application of common sense on a case by case basis.

Skull fractures Detection of fractures of the cranial vault by plain radiography of the skull is now appreciated to be less useful in assessing the probability of intracranial haemorrhage than had been previously suggested. Clinical assessment appears to be a better guide and this, in turn, guides the need for CT. Thus the role of skull radiography has greatly diminished. In any event simple linear fractures are often of little consequence in themselves. Like fractures elsewhere, these may be simple or comminuted. They sometimes branch, and must be distinguished from vascular markings (Fig. 59.1), including the groove in the squamous temporal bone caused by a deep temporal artery4. Acute fractures are usually straighter, more angulated, more radiolucent and do

1342

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

skull base, including the petrous bone, where it may also reveal ossicular dislocation, a treatable cause of traumatic hearing loss. Growing fractures (leptomeningeal cysts) usually occur after severe head injuries early in life4. The dura mater underlying a linear fracture is torn, often with laceration of the underlying brain. Exposure of the remodelling bone to pulsation of the CSF results in progressive widening of the fracture line over weeks or months.

Traumatic haemorrhage

Figure 59.1 Bilateral vault fracture, with fluid level in sphenoid sinus (open arrow). Two fracture lines are seen; the more anterior (upper on this radiograph) is better defined and is therefore on the side nearer the radiographic plate. Apparent islands of bone within (small arrows) are typical of an acute fracture. This radiograph has been obtained with the patient in the supine brow up position.

not have corticated margins. A fracture passing through a sinus or air cell is effectively compound, and of much greater potential significance than a simple fracture. A compound fracture is one in which the cranial cavity is in real or potential communication with the exterior. Depressed fractures (Fig. 59.2) are usually comminuted and often compound; bone fragments embedded in brain substance often are removed or relocated to reduce the risk of post-traumatic epilepsy. Acuteness of the fracture may be determined by demonstrating overlying scalp swelling on CT. Fractures of the skull base are important because of bleeding or leakage of cerebrospinal fluid (CSF); air and fluid within the sphenoid sinus may indicate that the leptomeninges have been torn. CT is extremely helpful in assessment of fractures of the

Trauma may cause bleeding into the scalp, between the cranial vault and the dura mater (extradural—but also termed epidural), between the dura and arachnoid mater (subdural), or into the subarachnoid space, brain or ventricular system.The aim of imaging in the acute stage is to identify patients with intracranial bleeding requiring surgery; they represent less than 1% of patients with well-documented head injury. CT is the imaging procedure of choice, rather than MRI, as haematomas are about the most readily recognizable abnormality on plain CT.

Extradural haemorrhage The acute extradural (or epidural) haematoma is a relatively stereotyped lesion (Figs 59.3 and 59.4). Because the dura mater tends to adhere to the skull, the haematoma is seen on CT sections as a biconvex dense area immediately beneath the skull vault, convex towards both the brain and the vault. The temporoparietal convexity is the most common site, in which lesions are easily detected on axial sections. The haematoma often lies beneath a fracture of the squamous part of the temporal bone. They tend not to cross cranial sutures. Areas of low density within them may indicate continuing bleeding (Fig. 59.4), and add further urgency to the assessment. Frontal, vertical and posterior cranial fossa collections (Fig. 59.5) can be difficult to diagnose; coronal images may be required. Even then, shallow extradural haematomas may be overlooked,

A

Figure 59.2 Stellate comminuted depressed fracture produced by a direct blow. CT volume rendered Image.

Figure 59.3 (A, B) Depressed skull fracture with extradural haematomas; CT. Axial ‘brain and bone windows’ demonstrate the right temporal bone depressed several millimetres with an evident fracture line (arrow) anteriorly. Soft tissue contusion overlying the fracture is also noted. A large extradural collection (curved upper arrow) anteriorly crosses the midline and displaces the falx posteriorly, and there is a second contiguous left frontal extradural collection as well. Other images (not shown) demonstrated bilateral skull fractures. Note the generalized cerebral swelling with complete effacement of the frontal horns.

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Figure 59.4 Acute extradural haematoma. CT. Large biconvex right temporal extradural collection. Note the effacement of the right occipital horn and subtle displacement of the calcified choroid plexus within. There is also a small focus of active/fresh bleeding causing a small round lucency within the high attenuation of the haematoma, close to the expected site of the middle meningeal artery. (Courtesy of Dr Dan Scoffings).

Figure 59.6 Acute extradural haematoma. MRI in a neonate with traumatic delivery. (A) Axial T1-weighted image (750/16). Slightly hyperintense epidural collection (arrow) in the right temporal region. (B) Axial T2-weighted image (3000/120), epidural collection is hypointense and is invisible except for deformation of the underlying cortex. This is the MR signature of deoxyhaemoglobin.

especially when adjacent to contused or haemorrhagic brain. Wide window CT images may help distinguish the intermediate density of the clot from bone and underlying brain. The underlying brain is displaced, but often appears intrinsically normal. MRI can be helpful on occasions (Fig. 59.6).

Subdural haemorrhage

Figure 59.5 Trauma. CT. A biconvex density of blood over the left cerebellar hemisphere indicates an extradural haematoma (thick arrow). A crescent of fresh subdural blood spreads over the left temporal lobe and tracks along the tentorium in a comma shaped fashion (arrowhead); this feature differentiates it from an extradural. Typical sites of haemorrhagic contusions are also seen; gyrus recti and temporal lobe.

Subdural bleeding is often, but not always, associated with damage to the brain, and arises from rupture of veins which cross the subdural space; vault fractures are much less commonly present in patients with subdural haematomas than extradural bleeds. Subdural haematomas are seen most commonly over the cerebral convexities, under the temporal and occipital lobes, or along the falx cerebri. They lie in the virtual space between the dura and arachnoid maters and may be extensive. This is because the blood within them, while under less pressure, is less restricted and tends to spread out over the surface of the brain; bleeding may even extend over an entire cerebral hemisphere. They may follow minor head injuries, and sometimes seem to develop spontaneously, especially in the elderly and in patients with haematological abnormalities. In such situations they are often diagnosed during the investigation of persistent headache, or perhaps transient but repetitive focal neurological deficits. Large ones requiring operative evacuation are usually associated with a reduced conscious state of the patient. On axial CT and MRI, the cerebral surface typically is concave (Fig. 59.7), but on coronal images may appear more convex. Acute lesions are usually hyperdense on CT, but mixed density is also common. They become progressively less dense over time, and typically end up of similar density to CSF within a few weeks or months. During this evolution there is often a period when the attenuation of the haematoma is similar to that of cerebral tissue; the resulting ‘isodense subdural’ haemotoma6 can be difficult to identify (Fig. 59.8) and is a well recognized pitfall on CT which continues to cause problems7. MRI is better at making this diagnosis when these lesions are of some longstanding. Most resolve spontaneously with time, but some

1343

1344

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 59.7 Acute subdural haematoma. CT. Heterogeneous density of irregular shape occupies extra-axial space overlying the left cerebral convexity. There is quite severe mass effect exhibited by effacement of convexity sulci, narrowing of the left sided ventricular system and midline shift. (Courtesy of Dr Dan Scoffings.)

persist, sometimes for years. Occasionally these lesions enlarge progressively at a variable, but usually slow, rate and eventually may require evacuation. The high morbidity of these lesions, particularly in the aged, is due in large part to the associated swelling, contusion or laceration of the underlying brain. It is often evident that midline displacement is greater than would be accounted for by the mass of the haematoma alone. Dilatation of the contralateral ventricle is a bad prognostic sign. The interhemispheric subdural haematoma extends along the falx cerebri and may spread onto the tentorium, giving a characteristic comma shape on axial CT sections (see Fig. 59.5). Coronal CT sections may be useful in distinguishing supraand infra-tentorial bleeding. Sub- or extra-dural collections low in the posterior cranial fossa, which may be life-threatening, may be overlooked, and the presence of unexplained hydrocephalus after acute head injury should prompt thorough examination of that region. The CT attenuation of the blood in a subacute subdural haematoma slowly decreases: as a rule of thumb, it remains

B Figure 59.8 Bilateral isodense subdural haematomas on contrastenhanced CT. The ventricles (A) are slit-like and displaced medially, giving a ‘rabbit’s ears’ appearance. A higher section (B) indicates that the normal grey–white interface lies too near the midline; the cortex appears abnormally thick. On close inspection, both sections show cortical vessels (arrows) displaced away from the cranial vault.

Figure 59.9 Subacute left subdural haematoma: CT. The lesion is of lower attenuation than brain tissue but denser than CSF. The underlying sulci are somewhat effaced and the left ventricle is compressed. (Courtesy of Dr Dan Scoffings.)

denser than the brain for 1 week, and is less dense after 3 weeks (Fig. 59.9). There is thus an interim period of up to 2 weeks when it may be ‘isodense’ with brain (see Fig. 59.8). Not all isodense haematomas are subacute: an acute bleed can be isodense in a very anaemic patient, and if there is continued leakage of venous blood, a chronic haematoma may not be of low density. Indirect signs may then be crucial: midline shift, with compression of the ipsilateral ventricle; contralateral ventricular enlargement; effacement of cerebral sulci, and medial displacement of the junction between the white and grey matter (‘buckling’). Some of these signs may be absent if there are bilateral collections; the frontal horns may then lie closer together than normal, giving a ‘rabbit’s ear’ configuration (see Fig. 59.8). Intravenous contrast medium, by highlighting the vessels on the surface of the brain, may remove any doubts about the extracerebral location of the lesion. The increasing use of MRI for nonacute problems should help overcome this diagnostic problem in the future. Chronic subdural collections are usually biconvex. Their density is less than that of brain, approaching that of CSF. Fluid–fluid levels may be seen between denser blood elements in the more dependent portions and serous fluid above, particularly if haemorrhage has been repeated.The membrane on their deep surface frequently shows contrast enhancement.

PRIMARY CEREBRAL DAMAGE IN CLOSED HEAD INJURY This is commonly associated with intracerebral haemorrhage, usually small and multifocal. An important feature is that the haemorrhages tend to enlarge and become more conspicuous over the initial few days after injury.

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Primary cerebral damage may be described as either superficial or deep. Interestingly these patterns of injury usually are mutually exclusive. Deep damage generally is considered to occur more commonly in high speed accidents and to have a worse prognosis, but exceptions are encountered quite frequently, and as in all forms of head injury, prognosis must be guarded in the initial few months following the injury.

Superficial primary cerebral damage This consists of cerebral contusions (see Fig. 59.5) and cortical lacerations. The underlying white matter also usually is damaged to a variable extent.The lesions usually are quite extensive, and typically involve the inferior parts of the frontal lobes and the anterior parts of the temporal lobes, but they can be found elsewhere. The mechanism is rotation of the brain with respect to the skull, especially the sphenoid ridges and the anterior cranial fossa.The term contra coup contusion often is used because commonly this type of cerebral damage lies diametrically opposite the site of impact, as defined by skull fracture or scalp haematoma. However, the cause is rotation, not linear acceleration of the brain. On CT, contusion appears as superficial low density areas with mild to moderate mass effect, which tends to increase a little in the initial days and subsequently contracts into a region of focal atrophy and sometimes cavitation. Multiple, usually small hyperdense haemorrhages are often present within the low density areas in the early stages. On MRI, they appear as mixed signal lesions, later contracting to regions of persistent mainly cortical cerebral damage. MRI is not much more sensitive in assessing the extent of contusions than CT.

Deep primary cerebral damage This pattern of injury is considerably less common.The mechanism here is differential rates of rotational acceleration within brain substance itself, producing shearing forces which damage axons and microvasculature.The injury is microscopic and may not be detected at all by CT or MRI, unless diffuse atrophy subsequently develops in the brain. Its presence is most often recognized on imaging by the presence of so-called marker lesions. These probably represent small areas of microvascular damage with haemorrhage or infarction, but they are a reliable guide to the presence of diffuse axonal injury, though not to its extent. They are small multifocal lesions, and tend to occur in more or less characteristic sites: high parasagittal cerebral white matter, corona radiata, posterior corpus collosum and subcortical white matter almost anywhere. They usually are not visible on CT unless haemorrhagic; many more may be on MRI, whether haemorrhagic or not. Susceptibility-weighted MRI (T2 ‘star’ acquisitions) often shows still more lesions, even long after the event, as small dark patches of haemosiderin. Characteristically the surrounding brain appears normal. Quantitative diffusion tensor imaging has been considered to demonstrate the axonal damage in a few reported cases, but only when it has been exceptionally severe. When the vascular component of the shearing injury is severe there may be larger haemorrhages in the basal ganglia and elsewhere, a pattern sometimes termed diffuse axonal injury of the brain.

Primary brainstem injuries These usually only occur with deep cerebral damage. The most common is a haemorrhagic lesion in the dorsolateral midbrain. Another is the pontomedullary rent, usually not compatible with life and therefore rarely seen on imaging.

OTHER TYPES OF INTRACRANIAL HAEMORRHAGE AFTER CLOSED HEAD INJURY Subarachnoid haemorrhage Head injury is probably the most common cause overall of subarachnoid haemorrhage. It commonly accompanies superficial cerebral damage, but may be minor and inconspicuous and is usually not recognized radiologically or clinically.

Intraventricular haemorrhage When isolated, intraventricular haemorrhage usually seems to be the result of tears in the attachments of the septum pellucidum to the corpus callosum.

Isolated large intracerebral haemorrhage Although rare, intracerebral haemorrhage is encountered from time to time and is often a source of diagnostic confusion. Cerebral angiography may show a false aneurysm.

SECONDARY CEREBRAL DAMAGE WITH CLOSED HEAD INJURY This results mainly from the effects of raised intracranial pressure, local pressure cones and fluctuations in systemic blood pressure and blood oxygen saturation. A common serious problem in the initial 2 or 3 d after a major head injury is diffuse cerebral swelling due to a increase in the cerebral blood volume, appropriately referred to as hyperaemic brain swelling, and less appropriately as brain oedema. It is a potent cause of raised intracranial pressure in this period and may trigger drastic neurosurgical decompression by wide craniectomies.The appearance of brain substance on CT and MRI is not affected, so it is not directly recognizable by imaging alone. Secondary cerebral damage consists of infarctions and brainstem haemorrhage. Infarcts most commonly occur in the cortical distributions of one or both posterior cerebral arteries, and haemorrhages are most often found in the ventral mid brain and upper ponds. Both probably, but certainly the latter, are due to pressure cones across the tentorial incisura.

OTHER COMPLICATIONS WITH CLOSED HEAD INJURIES Hydrocephalus requiring shunting is an uncommon complication and generally is of a communicating type. Cerebrospinal fluid fistulae are associated with skull base fractures and may present with otorrhoea or rhinorhhoea. Most heal within 10 d or so, but a small number may persist

1345

1346

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

or become intermittent, and eventually require surgical repair. Detailed imaging may be necessary to identify the site of the leak8. CSF leaks are associated with intracranial air, which may be extradural if the meninges are intact, or if they are torn, subarachnoid or intraventricular. These are often referred to as aerocoeles or pneumocephalus. Cranial nerve palsies are usually associated with skull base fractures and are immediate and permanent. Delayed cranial nerve palsies also may occur in the absence of fracture and may be reversible. Virtually any nerve or branch that traverses the skull base can be involved. An arteriovenous fistula may be an immediate or a delayed complication (Fig. 59.10). The most common occurs between the internal carotid artery and the cavernous sinus9, and is

termed a direct carotid cavernous fistula. These usually are torrential, and may require urgent endovascular treatment to preserve vision, sometimes at a time when the overall outcome is in the balance. Alternatively they may be chronic and the patient may present with an alarming engorged exophthalmos.

Penetrating head injuries Although uncommon, such injuries are increasing in frequency in civilian practice. Appearances on imaging vary with the penetrating agent and the trajectory of the cerebral penetration. Intracerebral haemorrhage usually dominates the appearances on imaging. Direct vascular injury is more common than in closed head injury, resulting in larger haemorrhages and more frequent false aneurysms.

B

Figure 59.10 Post-traumatic caroticocavernous fistula. (A–D) Axial contrast-enhanced CT: the right cavernous sinus (A, arrow) is enlarged, and a large enhancing mass runs forwards into the orbit through a widened superior orbital fissure (A, arrowheads). A sigmoid structure (C, open arrow) in the upper part of the right orbit represents the greatly dilated superior ophthalmic vein (cf. normal left side in C, small white arrow). Some of the extraocular muscles are thicker than on the left, and there is marked right proptosis. (E) Intra-arterial DSA, lateral projection, arterial phase, following injection into the right internal carotid artery. Contrast medium floods into the cavernous sinus (S), and drains anteriorly into a grossly dilated superior ophthalmic vein (V); there is also shunting posteriorly and via the inferior petrosal sinus (P). Intracranial arterial filling is poor. (F, G) After therapeutic detachment of a balloon (B) in the cavernous sinus (F, lateral projection), shunting particularly anteriorly, is greatly reduced, and intracranial filling much improved (G).

DISEASES OF THE SKULL—BONE DISEASE Fibrous dysplasia In the skull, fibrous dysplasia takes two main forms.The sclerotic form is the commoner of the two, especially in the polyostotic version of the disease10: it involves the base and/or facial skeleton, which are expanded and dense, sometimes showing the classical featureless ‘ground-glass’ pattern (Fig. 59.11). This is

the most common cause of ‘leontiasis ossea’. In meningiomas, which represent the important differential diagnostic consideration, sclerosis is often more marked than expansion, and extension from the sphenoid bones into the facial skeleton is much less common. Lower density areas, representing cysts or fibrotic masses, within the sclerotic bone are strong evidence

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Figure 59.11 Fibrous dysplasia of the skull base. Dense ‘ground glass’ appearance, with loss of the normal bone texture, extending from the nasion to the clivus into the sphenoid wings on both sides, and into the left maxilla.

Figure 59.12 Osteoporosis circumscripta (Paget’s disease). Extensive loss of bone density affects the lower part of the cranial vault; the margin between abnormal and normal bone is characteristically sharp, as seen in the upper posterior parietal region.

in favour of fibrous dysplasia. Malignant transformation may cause ragged osteolysis. The cystic form of fibrous dysplasia, often an incidental finding, usually produces a small lesion of the skull vault, expanding the outer table and giving a ‘blistered’ appearance. Here the differential diagnosis is from epidermoid, which has finer, better defined borders and a more homogeneously radiolucent centre.

Paget’s disease This condition, of unknown aetiology, is rare before middle age. It affects men more than women, and familial and strong geographical variations in prevalence are also noted. The skull is involved in about two-thirds of cases presenting clinically. A mixture of sclerosis and lysis is most commonly seen, and unless the condition is revealed incidentally, radiological changes are often advanced by the time of diagnosis. An early change is a spotty ‘cotton-wool pledget’ increase in density of the bone, which also becomes thicker. The middle and outer tables are most affected and thickened with course trabeculation. Generalized thickening of the vault may, however, be the first manifestation. Osteoporosis circumscripta may progress to the type described above and is frequently found in association with skeletal Paget changes. A portion of the vault is demineralized, sometimes profoundly so, from the base upwards, with a very sharp border between normal and abnormal bone (Fig. 59.12). Striking resolution can occur with therapy11 or the lesion may progress to classical changes of Paget’s disease. Complications include: • Basilar invagination, due to softening of bone; this occurs in about a third of cases, and can lead to a ‘Tam O’Shanter’ deformity (Fig. 59.13). • Narrowing of basal foramina, producing cranial nerve lesions, especially deafness, and occasionally compression of the medulla oblongata or upper spinal cord.

Figure 59.13 Advanced Paget’s disease, with basilar invagination. Skull lateral projection: gross thickening and alteration of bone texture affects the entire skull. Basilar invagination is manifest as extension of the odontoid peg (outlined in black) above Chamberlain’s line (dotted line).

• Malignancy: osteo- or fibro-sarcomas, which are otherwise extremely rare, arise in about 1% of patients with cranial Paget’s disease, and generally manifest as irregular bone destruction. Imaging reveals a thickened vault, which may appear irregular due to patchy sclerosis. Basilar invagination is readily recognized on sagittal MRI, and can be appreciated on axial images if the entire clivus is seen on one normally angled section, or if the ring of bone surrounding the foramen magnum appears to lie at the centre of the posterior fossa.

Tumours of the skull vault Primary tumours of many kinds can arise in the cranial vault, but most are exceedingly rare.

1347

1348

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Haemangioma Haemangioma of the vault sometimes presents as a lump on the head or local tenderness. It is typified by a wellcircumscribed area of punctate or stellate rarefaction without expansion (Fig. 59.14). Prominent vascular grooves may be present in the vicinity and external carotid arteriography sometimes shows a blush. The tumour is not progressive and treatment is not generally required.

Epidermoids Characteristically, intraosseous epidermoids are well-corticated lesions of the vault, which may expand the inner and outer tables away from each other. Rarely, this appearance may be mimicked by neurofibromas or dermoids.

Osteomas Osteomas are benign condensations of cortical bone which may project external to the skull (exostosis) or towards the cranial cavity (enostosis). Differential diagnosis includes meningioma and fibrous dysplasia; osteomas are typically denser and

Figure 59.14 Haemangioma of skull vault. (A) Lateral skull and (B) CT, bone windows. The well defined lucency in the parietal bone has a typical ‘spoke wheel’ appearance due to prominent vascular impressions.

more circumscribed and do not show abnormal bone texture. Treatment is not usually required.

Tumours Tumours of the skull are usually secondary. Local invasion from a superficial tumour, such as a nasopharyngeal or basal cell carcinoma, may occur. Metastases are common in disseminated malignancy; their characteristics are similar to those found in other bone sites: irregular lysis and/or sclerosis, and multiplicity (Fig. 59.15). In children, focal osteolytic skull defects are a common manifestation of Langerhans’ cell type histiocytosis.

Thinning of the parietal bones An uncommon condition of completely unknown aetiology is bilateral symmetrical thinning of the parietal bones, occurring in elderly patients and more common in men. It does not appear to be related either to generalized bone disease or to cerebral disease, although there may be an increase in deafness. The plain radiological appearance is diagnostic, but may be misinterpreted if the observer is unaware of this condition. There is very marked thinning of the outer table and diploe of the superior portions of both parietal bones. On lateral radiography the markedly thinned bone is well seen, often with a sharp curved inferior border. Although commonly seen in the days of skull radiography, this condition is less widely appreciated on CT.

Figure 59.15 Metastases from carcinoma of the breast; plain skull radiograph, lateral projection. Numerous irregular lytic defects in the cranial vault are associated with large vascular grooves.

DISTURBANCES OF THE CEREBROSPINAL FLUID CIRCULATION In simple terms, CSF is produced within the ventricles by the choroid plexuses, and absorbed at the cranial vertex and in the spinal canal via arachnoid villi. Reduced absorption results in hydrocephalus, ‘water on the brain’. The American neurosurgeon Walter Dandy noted that dye introduced into the ventricles reached the spinal subarachnoid space in some

hydrocephalic patients with what he therefore termed ‘communicating hydrocephalus’, but failed to do so in those with ‘noncommunicating’ or ‘obstructive’ hydrocephalus. The most common identified abnormality in children presenting with large heads and hydrocephalus is aqueduct stenosis (Fig. 59.16); most causes of obstructive hydrocephalus are referred to in

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Figure 59.16 Hydrocephalus secondary to aqueduct stenosis: MRI. (A) Axial proton density and T2-weighted images. Marked enlargement of the lateral ventricles, with a thin ‘halo’ of interstitial oedema. Heterogeneity of the fluid signal within the ventricles is due to pulsatile artefact. (B) Sagittal T1-weighted image demonstrating massive enlargement of lateral ventricles, outpouching of the suprasellar recesses of the third ventricle impinging upon the sella, and ‘ventricularization’ of the proximal aqueduct just above the level of obstruction and above the fourth ventricle. Note the normal size and configuration of the fourth ventricle (arrow).

other sections of this book. Very rarely, as in papilloma of the choroid plexus, hydrocephalus is due to excessive production of fluid. There is recent interest in the concept of often unrecognized venous sinus thrombosis being a contributory factor12.

COMMUNICATING HYDROCEPHALUS In this variety of hydrocephalus, the barrier to absorption of CSF is distal to the foramina of Magendie and Luschka: usually at the tentorial hiatus within the basal cisterns or over the cerebral convexity. In adults, intracranial pressure may be raised, but lumbar CSF pressure can be normal in ‘normal pressure hydrocephalus’. Major causes of communicating hydrocephalus are trauma, subarachnoid haemorrhage and infection, but in a small number of cases the causative factor is not identified; spinal tumours are occasionally incriminated. At CT or MRI, the hallmark of communicating hydrocephalus is ventricular dilatation, often marked and generalized, although the fourth ventricle may be spared. In young children the occipital horns are often most affected, but in adults enlargement of the frontal and temporal horns is more striking. Interstitial cerebral oedema is variable. The basal cisterns and fissures are sometimes prominent, but the cerebral sulci are characteristically not enlarged; they may be small. In adults presenting with dementia, the major alternative cause of ventricular enlargement is cerebral atrophy. The differentiation may seem important, because patients with communicating hydrocephalus have been reported to show a dramatic response to shunting procedures, whereas primary neuronal degeneration and vascular dementia are essentially untreatable. However, differentiation both clinically and by

imaging is difficult and the subject remains controversial. Cisternography, with radionuclides or water-soluble contrast medium and CT, has been employed in such cases; and more recently MRI using flow-sensitive sequences has shown differences in the pattern of movement of the ventricular fluid. Correlation with the results of surgery for any of these is disappointing, most workers now conceding that the best that can be hoped for is a modest improvement in gait, which nonetheless may be significant for care needs.

Ventricular shunting procedures Imaging procedures, such as ultrasound (US) in infants and CT or MRI in adults, are often employed to assess the size of the cerebral ventricles following shunting procedures for communicating or obstructive hydrocephalus. A satisfactory result is indicated by reversion of ventricular size towards normal and, on CT or MRI, a disappearance of interstitial cerebral oedema. Patency of third ventricle ventriculostomy can be adequately assessed by cardiac gated cine-phase contrast MRI.

Complications Those that may be detected include: • Malfunction of the shunt, with failure of the hydrocephalus to resolve. This can simply be because of discontinuity in the extracranial shunt tubing, best assessed by plain radiographs; ventriculo-peritoneal shunts can also lead to a loculated intra-abdominal fluid collection, best assessed by US. • Incorrect placement of the intraventricular catheter. • Haemorrhage, intraventricular, intracerebral or extracerebral. • Subdural effusions, of low density on CT from the outset, which are not uncommon. • Infection: ventriculitis, abscess or subdural empyema.

1349

1350

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

DEGENERATIVE DISORDERS GENERAL ASPECTS These comprise a heterogeneous group of conditions, some common, others rare; their pathological hallmarks are neuronal or axonal degeneration and gliosis. Clinically they are characterized mainly by progressive cognitive decline and movement disorders. The pathological changes are distributed uniformly in some diseases and, in others, nonuniformly—often with characteristic predilection to involve certain regions or structures. Traditionally these diseases are defined in terms of clinical syndromes, believed to reflect the distribution of neuronal damage. In recent years great strides have been made towards unraveling the genetic and molecular basis of some of these conditions which will, in time, become much more precisely defined. Most of these conditions are currently untreatable, but soon this is expected to change, fuelling a research drive to achieve precise clinical diagnosis early, before too much irreversible

damage has occurred. Since it is in the early stages that clinical diagnosis often is most difficult, the search is on for surrogate markers and functional neuroimaging offers considerable promise. In the advanced case, CT and MRI will show atrophy in involved regions and perhaps signal change. In some diseases, but usually only in a minority of cases, the pattern of atrophy can have positive predictive value in the face of clinical ambiguity. Nevertheless, it is still the case that the main role of structural imaging is to exclude other causes, which clinically may resemble degeneration, such as neoplasms or hydrocephalus. The positive yield is vanishingly small, but CT or MRI are recommended at least once in the clinical course, repeated only if intercurrent disease is suspected. An important feature of the degenerative diseases is that they are progressive, and rate of change exceeds that of normal ageing13. It is because structural imaging at one point in time is so often unhelpful that many believe functional imaging has a major role.

THE DEMENTIAS These generally are diseases of the ageing population. After the age of about 50 years the normal brain loses mass and volume at a variable rate, perhaps averaging about 0.1–0.3% per year. Recent advances in quantitative assessment of brain volume have allowed considerable progress in the early diagnosis and assessment of dementias.

Alzheimer’s disease This is the most common cause of primary degenerative dementia, and its incidence increases with age, rising sharply over 70 years. It is a generalized disorder, although in its early stages it may affect the medial part of the temporal lobe, especially the hippocampus, more than elsewhere. The clinical problem often is to distinguish Alzheimer’s disease (AD), a disorder which usually progresses to complete disability within 2–5 years, from the far more benign age related memory loss.

decline18. Perfusion defects are not simply the result of cerebral atrophy, but are greater than can be explained by any atrophy demonstrated on anatomical imaging with CT or MRI. Similarly, [18F]deoxy-d-glucose (FDG)-positron emission tomography (PET) studies show reduced glucose uptake that is not explained by atrophy19. Receptor imaging using radioligands for central benzodiazepine receptors has shown a similar distribution of deficits in AD20.

Frontotemporal dementia

This is usually normal, although occasionally accentuated atrophy medially in the temporal lobe is indicated by widening of the perihippocampal CSF spaces, which should be both bilateral and symmetrical14. There is now much greater understanding of the rate of volume change in many of the dementias15, especially in the hippocampus16 following the seminal work in this area using MRI13. Such studies require meticulous positioning and registration so that subtle volume losses over time can be appreciated.

These comprise less than 10% of the primary degenerative dementias. They include Pick’s disease. Behavioural, motor or speech disorders tend to dominate the early clinical stages rather than memory loss21. Discrimination from a psychiatric disorder often is the clinical issue. Structural imaging is abnormal in up to over 50% of clinically definite cases, and may confirm the diagnosis in situations of clinical ambiguity; imaging is less often helpful in the purely frontal lobe syndromes, such as primary progressive aphasia. Thin section volumetric coronal MRI seems the optimal acquisition to achieve the best positive predictive value, and interpretation may require attention to detail if more subtle changes are to be detected. MRI and even CT show atrophy in the anterior and medial parts of the temporal lobe, which usually is markedly asymmetric (right or left) (Fig. 59.18), and diminishes posteriorly. Asymmetric frontal lobe atrophy may also be present.

Functional imaging

Functional imaging

In established disease, characteristic symmetrical posterior temporal and parietal perfusion defects on regional cerebral blood flow (rCBF) SPECT have a predictive value for the diagnosis of AD of over 80%17 (Fig. 59.17), and the severity of rCBF reduction correlates with the degree of cognitive

Perfusion deficits on rCBF SPECT are predominantly frontal and anterior temporal with preserved perfusion posteriorly. Reduced frontal perfusion is not specific to frontotemporal dementia and can occur in a variety of other conditions, such as schizophrenia, depression, human immunodeficiency virus

Structural imaging

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Figure 59.17 Axial (L) and sagittal (R) 99m Tc HMPAO SPECT in Alzheimer’s disease showing typical perfusion defects in the posterior temporal and parietal regions.

(HIV) encephalopathy, Creutzfeldt–Jacob disease (CJD) and in some cases of AD22.

Lewy body dementia Lewy body dementia is now recognized as the second most common degenerative dementia after AD and it accounts for approximately 20% of all dementia. Neither structural imaging nor rCBF SPECT can reliably distinguish between Lewy body dementia and AD on subjective assessment, as posterior tem-

poroparietal defects occur in both. However, reduced frontal perfusion with HMPOA SPECT and reduced uptake in the cerebellum and visual cortex with FDG-PET is seen in Lewy body dementia compared with AD23.

Vascular dementia This is the clinical diagnosis in about 20% of all dementias. Evidence of ischaemic damage on CT or MRI is mandatory for diagnosis, but such changes, especially in the white matter,

Figure 59.18 Pick’s disease. (A) Axial T2-weighted images and (B) coronal T1-weighted images (from a volumetric acquisition) showing severe rather generalized mainly anterior temporal lobe atrophy, most marked on the left side. Such severe change needs to be distinguished from other causes of brain damage such as head injury and encephalitis. Lack of past history should be decisive, but the pattern of atrophy is usually sufficiently characteristic to distinguish between these conditions.

1351

1352

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

are very common in elderly nondemented subjects and there are no reliable imaging criteria to distinguish between the two. Functional imaging typically shows patchy, cortical and basal ganglia perfusion defects, but it may not always reliably distinguish between vascular dementia and frontotemporal dementia17.

Prion diseases These mainly comprise CJD (sporadic, iatrogenic, familial) and very recently in Europe (especially in the UK) new variant CJD (nvCJD). Rapidly progressive dementia, often with myoclonus, is the usual clinical picture, often preceded by behavioural disturbances, especially in nvCJD. Structural imaging usually appears normal in early stages, but rapidly progressive atrophy soon develops. Symmetrical increases in signal in the putamen and caudate nuclei may be shown by MRI in about 10% of sporadic CJD, and in the posterior part of the thalami in over 50% of nvCJD; when present, these signs can have considerable positive predictive value24.

MOVEMENT DISORDERS Idiopathic Parkinson’s disease and Parkinsonian syndromes Structural imaging has little role here except to exclude other diseases and to demonstrate the extent of the atrophy (Fig. 59.19). Functional imaging provides insights into disease mechanisms in research settings, but the diagnosis still remains mainly clinical. Receptor imaging of the dopaminergic system

Figure 59.19 Multiple system atrophy, cerebellar type (pontoolivocerebellar atrophy). (A) Sagittal T1- and (B) axial T2-weighted MR images showing marked atrophy of the ventral part of the pons and of the ‘pontine nuclei’ and their axons; on axial images signal change as well as atrophy give rise to a ‘hot-cross bun’ appearance, the darker areas representing the preserved corticospinal and lemniscal pathways.

continues to be widely studied in idiopathic Parkinson’s disease and other movement disorders. New functional imaging techniques continue to evolve in this area; some now offer insights into controlling symptoms and future therapeutic innovations.

EPILEPSY Recent onset seizures It is recommended by various bodies1 and by NICE25 that anyone who has an unprovoked seizure should have brain MRI. Intravenous contrast medium is not required routinely even with focal seizures. A progressive cause such as neoplasia will be found in only about 6%, but in countries where tuberculosis and cysticercosis are common, it can be as high as over 50%. Standard MRI examinations are all that are required in this context.

Chronic epilepsy These patients have repeated seizures, usually over years, which usually are well controlled by drug treatment. There may be no detectable underlying structural lesion (cryptogenic epilepsy), or a variety of usually nonprogressive lesions, some of which are rare or do not occur in other contexts.These may be grouped as: • scars—infarcts, trauma • vascular lesions—cavernomas, arteriovenous malformations • malformation of cortical development—various, but especially focal cortical dysplasia, cortical hamartomas (tuberous sclerosis) and neuronal heterotopias (Fig. 59.20) • neoplasms—grade I tumours usually containing mixed glial and neuronal elements: gangliogliomas, dysplastic neuroepithelial tumour

• hippocampal sclerosis—by far the most common and associated with temporal lobe epilepsy (TLE), the most common of the partial epilepsies (Fig. 59.21).

Structural imaging26 In patients with drug-resistant epilepsy, excision of a structural lesion may cure the epilepsy or at least improve seizure control. Focal excision of normal cortical tissue on the other hand, generally does not. The structural lesions associated with chronic epilepsy are shown with great sensitivity by modern MRI, and it is highly likely that any focal cortical excision in MRI-negative patients will yield histologically normal tissue. In most centres this has led to an appropriate emphasis on structural imaging with MRI in preoperative assessment, with resort to functional imaging only when further clarification is required. Routine MRI is usually adequate to detect most lesions, and it is here that T2-weighted lesions with fat suppression (e.g. FLAIR) and high resolution volume acquisitions can be particularly helpful in drawing attention to small cortical abnormalities which may otherwise have gone unnoticed (Fig. 59.22). Acquisition in the coronal plane is mandatory to detect hippocampal sclerosis. The signs of hippocampal sclerosis on MRI are (1) volume loss and (2) increased signal on T2-weighted images

CHAPTER 59

•

CRANIAL AND INTRACRANIAL DISEASE: TRAUMA, CEREBROSPINAL FLUID DISTURBANCES, DEGENERATIVE DISORDERS AND EPILEPSY

Figure 59.20 Neuronal heterotopia. Coronal T1-weighted images from a volumetric acquisition; slice thickness is 1.5 mm. (A) Subependymal (nodular) heterotopia (arrowhead), (B) laminar heterotopia (arrowheads).

(see Fig. 59.21). Outside specialized units, it is signal change that is easiest to detect and most reliable, but it may be important to ensure that the brighter hippocampus is not the larger because then the pathology would not be hippocampal sclerosis. Volume loss therefore is more specific, but it is detected reliably only by thin section volumetric

acquisitions in the coronal plane and requires considerable attention to detail26.

Functional imaging MRI and ictal scalp electroencephalography (EEG) can be discordant or negative in up to 40% of potentially pre-operative

Figure 59.21 Diagnosis of hippocampal sclerosis. (A, B) Normal hippocampus (arrow heads). (A) Body, (B) head. Coronal T1-weighted MRIs from a volumetric acquisition; slice thickness is 1.5 mm. Asymmetry is due to asymmetrical position of each slice with respect to the hippocampus which is essentially unavoidable and usually appears more marked in the head region. (C) T1- and (D) T2-weighted images through the temporal lobes. In (C) the left hippocampus is smaller than the right and (D) is of higher signal. Images (C and D, Courtesy of Dr P. Rich.)

1353

1354

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 59.22 Tuberous sclerosis in a 5-year-old girl with seizures; MRI. Proton density and T2-weighted images (2000/40, 2000/80). Multiple cortical and subcortical hyperintensities represent tubers, with associated demyelination. A single hyperintense subependymal nodule is visible in the right trigone.

cases. Options for further noninvasive investigation include magnetic resonance spectroscopy (MRS), PET/CT, rCBF SPECT, FDG-PET and 11C-flumazenil PET. There is still no clear evidence as to which of these tests is best and most are the subject of ongoing research. Localization of extratemporal epileptogenic foci is less successful with all investigations, including structural MRI.

Calcification of the basal ganglia Calcification is frequently revealed by CT in normal, older people. Causes include hypoparathyroidism, pseudohypoparathyroidism and mitochondrial cytopathy, but most cases are idiopathic and disturbances of calcium metabolism are so rare that biochemical testing is performed only if indicated by other features; some cases are familial (Fahr’s syndrome). In many of these conditions, MRI may show much more extensive increased signal on T1-weighted images than revealed by CT. This is believed to be due to T1 shortening in the hydration shells enclosing microscopic crystal deposits.

ACKNOWLEDGEMENTS In the previous editions, Alison D. Murray, Dawn Saunders and Barton Lane all contributed handsomely.

REFERENCES 1. Royal College of Radiologists 2007 Making the best use of a department of clinical radiology: guidelines for doctors. London, RCR 2. National Institute for Clinical Excellence 2003 Head injury: guidelines. London, NICE http://www.nice.org.uk/guidance/CG4/guidance/pdf/ English/download.dspx 3. Mayor S 2003 NICE recommends greater use of CT imaging for head injuries. BMJ 326: 1414

4. Du Boulay G H 1980 Principles of X-ray diagnosis of the skull. Butterworth, London. 5. Kingsley D P E, Till K, Hoare R D 1978 Growing fractures of the skull. J Neurol Neurosurg Psychiatry 41: 312–318 6. Naidich T P, Moran C J, Pudlowski R M, Naidich J B 1979 CT diagnosis of isodense subdural hematoma. In: Thompson R A, Green J R (eds) Adv Neurol 22: 73–105 7. Boviatsis EJ, Kouyialis AT, Sakas DE. 2003 Misdiagnosis of bilateral isodense chronic subdural haematomas. Hosp Med 64: 374−375. 8. Avdin K, Guven K, Sencer S, Jinkins JR, Minareci O 2004 MRI cisternography with gadolinium-containing contrast medium: its role, advantages and limitations in the investigation of rhinorrhoea. Neuroradiology 46: 75−80 9. Peeters F L M, Kroger R 1979 Dural and direct cavernous sinus fistulas. Am J Roentgenol 132: 599–606 10. Liakos G M, Walker C B, Carruth J A S 1979 Ocular complications of craniofacial fibrous dysplasia. Br J Ophthalmol 63: 611–616 11. Dodd G W, Ibbertson H K, Fraser T R C, Holdaway I M, Wattie D 1987 Radiological assessment of Paget’s disease of bone after treatment with the biphosphonates EHDP and APD. Br J Radiol 60: 849–860 12. Higgins J N, Gillard J H, Owler B K, Harkness K, Pickard J D 2004 MR venography in idiopathic intracranial hypertension: unappreciated and misunderstood. J Neurol Neurosurg Psychiatry 75: 621−625 13. Fox N, Freeborough P, Rossor M N 1996 Visualisation and quantification of rates of atrophy in Alzheimer’s disease. Lancet 348: 94–97 14. De Leon M, Golomb J, George A et al 1993 The radiological prediction of Alzheimer’s disease: the atrophic hippocampal formation. Am J Neuroradiol 14: 897–906 15. Schott J M, Price S L, Frost C, Whitwell J L, Rossor M N, Fox N C 2005 Measuring atrophy in Alzheimer disease: a serial MRI study over 6 and 12 months. Neurology 65: 119−124 16. Fox N L, Warrington E, Freeborough H P et al 1996 Presymptomatic hippocampal atrophy in Alzheimer’s disease. A longitudinal MRI study. Brain 119: 2001–2007 17. Duara R, Barker W, Luis C A 1999 Frontotemporal dementia and Alzheimer’s disease. Differential diagnosis. Dement Geriatric Cogn Disord 10 (Suppl 1): 37–42 18. Bartenstein P, Minoshima S, Hirsch C et al 1997 Quantitative assessment of cerebral blood flow in patients with Alzheimer’s disease by SPECT. J Nucl Med 38: 1095–1101 19. Ibáñez V, Pietrini P, Alexander GE et al 1998 Regional glucose metabolic abnormalities are not the result of atrophy in Alzheimer’s disease. Neurology 50: 1585–1593 20. Varrone A, Soricelli A, Postiglione A et al 1997 Benzodiazepine receptor distribution in Alzheimer’s disease with 123I-iomazenil SPECT. In: De Deyn P P, Dierckx R A, Alave A, Pickut B A (eds) A textbook of SPECT in neurology and psychiatry. John Libbey, London, pp 45–48 21. Neary D, Snowden J S, Gustafson L et al 1998 Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. Neurology 51: 1546–1554 22. Pickut B A, Saerens J, Mariën P et al 1997 SPECT in the differential diagnosis of frontal lobe type dementia and dementia of the Alzheimer type. In: De Deyn P P, Dierckx R A, Alave A, Pickut B A (eds) A textbook of SPECT in neurology and psychiatry. John Libbey, London, pp 3–10 23. Defebvre L J P, Leduc V, Duhamel A et al 1999 Technetium HMPAO SPECT study in dementia with Lewy bodies, Alzheimer’s disease and idiopathic Parkinson’s disease. J Nucl Med 40: 956–962 24. Coulthard A, Hall K, English P T et al 1999 Quantitive analysis of MRI signal intensity in new variant Creutzfeldt–Jacob disease. Br J Radiol 72: 742–748 25. National Institute for Clinical Excellence 2004 Epilepsy. London, NICE http://www.nice.org.uk/CG020 26. Stevens J M 1999 Imaging in epilepsy. Revista di Neuroradiologia 12 (Suppl): 57–62

CHAPTER

The Spine

60

John M. Stevens, Philip M. Rich and Adrian K. Dixon

Methods of examination • Plain radiography of the spine Anatomy • Spina bifida • Invertebral fusions • Transitional vertebrae Magnetic resonance imaging • Volumetric (3D) acquisitions • Fast spin-echo • Fluid attenuation inversion recovery • Phased-array coils • Motion artifacts • Conduct of MRI examination Computed tomography of spine • Positioning Normal anatomy • Vertebrae • Intervertebral discs • Intervertebral foramina • Ligaments and epidural soft tissues • Spinal cord • Spinal nerves Myelography • Special situations • Complications of myelography Spinal angiography • Anatomy Discography Radiological diagnostic approach to common clinical spinal problems • Acute presentations • Chronic presentations Disease processes affecting the spine • Congenital lesions

• Spinal malformations • Lipomyelomeningodysplasias • Neurenteric and other developmental cysts • Diastematomyelia • Chiari malformations • Meningoceles Vertebral fusion anomalies • Skeletal dysplasias • Neurofibromatosis • Spinal instability • Intraspinal arachnoid cyst • Syringomyelia • Degenerative spinal disease • Posterior joints and ligaments • Involvement of neural structures • Imaging • The postoperative spine • Spinal tumours • Extradural tumours • Intradural extramedullary tumours • Intramedullary tumours • Trauma • Imaging • Inflammatory arthropathies Infective disorders of the spine • Vertebral osteomyelitis • Epidural abscess • Paravertebral infections • Spinal meningitis (arachnoiditis) • Myelitis • Vascular lesions • Miscellaneous conditions

1356

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

METHODS OF EXAMINATION Many techniques are available for the investigation of diseases of the spine and its contents. However, all have been superseded by magnetic resonance imaging (MRI): this is now so widely available in most countries that other neuroradiological investigations of the spine are considered only when MRI is either contraindicated or difficult to perform. The one major exception is trauma and multidetector computed tomography (CT) provides exquisite anatomical information and multiplanar viewing on a workstation of the isometric data set. Indeed many spinal lesions now come to light through CT for unrelated symptoms.

PLAIN RADIOGRAPHY OF THE SPINE Plain radiography is still widely used but cannot really be justified any longer as a definitive investigation. However, there are a few situations where plain radiographs remain helpful, the most significant being the localization of trauma and evaluation of instability.

Thoracocervical spine Anteroposterior views These are good for demonstrating overall alignment (in scoliosis), vertebral collapse and the integrity of the pedicles (possible metastasis).

Lateral view This is best obtained as a ‘swimmer’s view’ with one arm elevated above the head and the other down by the side. Following trauma in thick-set individuals, attention to this view may prevent a referral for CT or MRI.

Thoracic spine Anteroposterior and lateral views If there is a scoliosis, the convexity is placed nearer the radiograph so that the divergent rays are more nearly parallel to the disc spaces.The patient may breathe gently during the exposure, blurring out the rib cage.

Lumbar spine Anteroposterior view

Cervical spine Anteroposterior and lateral views

Flexion of the hips and knees will reduce the lumbar lordosis.

These are the standard views although a lateral view with the neck held in the maximal flexion that the patient can manage comfortably is often required (e.g. rheumatoid arthritis). These views may be difficult to acquire following trauma with the patient supine; here it is essential to see the alignment at C7/T1. Figure 60.1 illustrates the importance of high-quality radiography.

Lateral view

Oblique view

Lumbosacral spine Anteroposterior and lateral views

These views display the foramina; those displayed en face are those away from which the head is rotated (cf. lumbar spine). To avoid confusion, each oblique projection must show both right and left side markers. Again the need for such views is questionable in the era of MRI.

Again convexity of any scoliosis is nearer the radiograph.

Anteroposterior oblique view The pedicles can be shown en face at 30–40 degrees of obliquity and neural foramina towards which the trunk is rotated. This can help in possible spondylolisthesis

The lowest disc space is often best assessed separately (by a coned rather overexposed image) from the remainder of the lumbar spine. The sacroiliac joints are best seen on a PA view, where the diverging beam is more aligned to the joints.

Figure 60.1 Lateral plain radiograph of the cervical spine. Patient being examined in the resuscitation unit following trauma. Shoot-through lateral radiographs with the patient supine. (A) Initial attempt. (B) Second radiograph with much improved radiographic positioning allowing demonstration of normal alignment including the all important C7/T1 junction.

CHAPTER 60

• THE SPINE

ANATOMY Some developmental anomalies are very common but most are without clinical significance.

SPINA BIFIDA A wide or narrow midline defect of the posterior arch of the atlas occurs in 6 per cent of the normal population, and of the anterior arch in about 2 per cent. Sometimes both defects coexist, resulting in a bifid atlas1. Failure of fusion of the neural arch of L5 or S1 is even more common, occurring in 20 per cent or more of some populations2.The defect, known as spina bifida occulta when there are no other features, is nearly always narrow and asymmetric.

INTERVERTEBRAL FUSIONS These usually occur only in the cervical region. Generally only one level is involved, C2/3 being the commonest, and this is frequently associated with occipital assimilation of the atlas3 (Fig. 60.2). The AP diameter of congenitally fused vertebrae is often smaller than that of other vertebrae in the region.

Figure 60.2 Lateral plain radiograph of the cervical spine showing the commoner intervertebral fusions: occiput – C1, C2–3 and C6–7.

TRANSITIONAL VERTEBRAE Complete or partial fusion of L5 with the sacrum is seen in more than 6 per cent of the normal population. One or both transverse processes of L5 are enlarged, and the L5–S1 disc space is narrow4. Rudimentary twelfth ribs are usually identifiable. A true extra lumbar vertebra due to lumbarization of S1 is less frequent. The main significance of transitional lumbosacral or thoracolumbar vertebrae is that they may result in a level being wrongly identified pre-operatively. The level of

the iliac crests provides a useful landmark; they usually correlate with the L4/5 disc. Cervical ribs involving the seventh cervical vertebra also occur in about 6 per cent of the normal population. Oblique or lordotic projections and apical chest views may aid good demonstration. A useful feature for differentiating short first ribs from cervical ribs is that the transverse processes of T1 tend to be directed superiorly.

MAGNETIC RESONANCE IMAGING Spatial resolution is a crucial consideration in the spine. Since the signal intensities of disc, bone, cerebrospinal fluid (CSF), spinal cord and epidural fat are very different on most sequences, it is possible to compromise on contrast resolution in the interest of spatial resolution in many applications of spinal MRI. Faster sequences are also possible and speed and throughput are not only economic considerations but also appreciated by patients. All such innovations are easier at high field strength – 1.5 T seems optimal for most applications.

VOLUMETRIC (3D) ACQUISITIONS The deployment of fast imaging techniques has permitted three-dimensional (3D) spatial encoding within a few minutes. The result is multiple contiguous images, no interslice gaps and section thicknesses down to 1 or 2 mm.This is helpful in the evaluation of the intervertebral foramina in the

cervical region. However, the best images of the spinal cord are usually obtained from slightly thicker slices (Fig. 60.3).

FAST SPIN-ECHO In fast spin-echo (FSE) several phase-encoding steps are made at each excitation instead of just one, which greatly reduces acquisition times and permits use of much larger matrices, resulting in twice the resolution in even shorter data acquisition periods. The penalties are slight loss of contrast and increased sensitivity to physiological motion (Fig. 60.4).

FLUID ATTENUATION INVERSION RECOVERY Fluid attenuation inversion recovery (FLAIR) sequences offer improved lesion detection by permitting heavily T2-weighted

1357

1358

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

detecting lesions in cord substance has been disappointing (see ref 15), and standard spin-echo techniques so far are preferable.

PHASED-ARRAY COILS

Figure 60.3 Axial MR image. Axial MRI cervical spine at C5–6, from a gradient-recalled multislice acquisition, slice thickness 4 mm. The internal structure of the spinal cord (surrounded by the white CSF) is quite well shown.

(T2W) acquisitions to be accomplished with suppression of all signal from cerebrospinal fluid. This removes motion artefact from CSF, the CSF appearing paradoxically as signal void on heavily T2W images. However, cysts and cystic lesions within the spinal cord also appear as signal voids. Most other lesions appear as high intensity. Imaging with FLAIR used to be slow and its resolution was low (see ref 14); however, new or fast FLAIR programmes now enjoy widespread clinical use, but sensitivity for

Spinal imaging requires surface coils, and the phased-array configuration enables data to be acquired simultaneously from up to four surface coils in series.This permits imaging of the entire spinal cord over one acquisition period and greatly reduces imaging time.Vertebral level counting, often difficult using single coils in the thoracic region, also becomes simpler (Fig. 60.5). A wide variety of post-processing options has also become available on most workstations. These permit multiplanar reformatting in real time, 3D surface rendering, colour coding and many others. The ability to reconstruct in a curved plane is potentially useful in scoliosis, but its value is limited because most major curvatures occur in more than one plane. In many ways such functions are more relevant to CT than MRI.

MOTION ARTEFACTS These are generated by pulsatile motion of the CSF. This motion has been documented and recently quantified by MRI. At C2/3, inferior movement is estimated at about 0.65 mm during systole5. MRI has shown that the primary driving force behind intracranial and spinal CSF flow is expansion of the brain during systole. The spinal cord and brainstem also descend slightly on systole and oscillate anteroposteriorly6.

Figure 60.4 Fast spin-echo MRIs of the cervical spine from a multislice (2D) sagittal series (TR 4000 ms, TE 80, matrix 512 × 256). The vertebral artery (arrow, right image), the intervertebral canals and dorsal root ganglia (arrows in left image) are shown well, but the spinal cord image (centre) is degraded by motion artefact.

CHAPTER 60

• THE SPINE

spinal cord flattening or exaggerating the degree of compression shown8. Metallic artefacts can be particularly destructive. Many spinal operations utilize metal stabilization devices and patients often require postoperative imaging at some stage. Ferromagnetic materials generate major local artefacts and usually render imaging of the spinal canal impossible. Tiny fragments from drills, invisible on plain radiographs, may also result in devastating artefacts9. Design of implants is also important; numerous titanium devices are now available and it is hoped that these will prove acceptable in the long term10.

CONDUCT OF MRI EXAMINATION

Figure 60.5 Spinal MRI using phased-array coils (spin-echo, TR 600 ms, TE 20). Two coil configurations: (A) cervical and thoracic. (B) Thoracic and lumbar.

In classic studies, Rubin and Enzmann7 showed how such motion generates linear artefacts parallel to the spinal cord/ CSF interface in the phase-encoding direction, producing signal variation in the spinal cord image that could be misinterpreted as an intramedullary lesion. Areas of turbulent CSF flow related to subarachnoid septa can result in signal variation simulating intradural masses or enlarged vessels, particularly in the thoracic spine. Many strategies have been developed to minimize these problems, including altering the direction of the phaseencoding gradient, application of additional motion-nulling gradients and cardiac gating. Spatial presaturation is used to suppress signal arising from moving structures such as the heart. Truncation artefacts are generated at boundaries by image processing. They are particularly significant at the CSF/ spinal cord interface. Such an artefact is one possible cause of the band of high or low signal seen near the centre of the spinal cord in some midsagittal images. It is also one of several sources of error in defining the position of the cord/CSF boundary. The physical dynamics of this boundary are more complex than in CT. Susceptibility effects are another important source of error, generated in particular by gradient-recalled echo techniques. Phantom studies have demonstrated that both electronic and caliper measurements of the midsagittal diameter of the spinal cord can be artificially reduced by over 2 mm in the phase-encoding direction, creating a spurious impression of a

Details of optimal pulse sequence, coil type and strategies for motion suppression are specified by individual manufacturers; the general principles, however, are the same for all equipment. The following is a guide only. Some centres make an effort to reduce the number of sequences used routinely in the interests of saving time. The saving of even a few minutes per patient by omitting one sequence may become significant. Even with well-designed protocols, some patients will have to be recalled for additional sequences.

Cervical spine MRI usually begins with a fast coronal multislice sequence as a localizer. The optimal sagittal plane for the cervical spine is selected from this series, and a multislice high-resolution sagittal series planned. Optimal sequences use matrix sizes of 512 × 256, with spatial presaturation applied over the larynx and carotid arteries. Two sagittal data acquisitions are planned, one strongly T1 weighted and the other T2 weighted. At these matrices, fast spin-echo techniques are required; artefact from CSF motion can be a problem. Axial images are planned from the sagittal data sets, to demonstrate levels of clinical concern or structural abnormality. On modern systems thin T2W images usually suffice, although they may overestimate the degree of compression. The protocol used depends on the clinical problem. In neuroradiology this is usually myelopathy, radiculopathy (often simply arm pain), or both. For myelopathy, a 2D multislice protocol is usually preferable, especially if intrinsic disease such as demyelination is being sought. In the latter, T2 weighting is essential and the best signal-to-noise ratio and shortest acquisition times are often achieved using a gradient-recalled echo technique. Axial fast spin-echo techniques are not usually satisfactory because of unsuppressed CSF motion. When evaluating the degree of cord compression, demonstration of internal structure is less important, and thinner sections can be used to improve precision; a T1-weighted (T1W) spin-echo sequence may be the best option to reduce the possibility of overestimating the degree of compression. When evaluating radiculopathy, the spinal root canals need to be targeted. The best option is a volumetric (3D) acquisition obtained in the axial plane with a T2W gradient-recalled

1359

1360

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

echo sequence data set covering the spine from about C2 to T1–2. Multiplanar reformatting along and orthogonal to the root canals can be performed if so desired.

Thoracic spine

and another T1W axial multislice series acquired, in the same positions as the first using fat signal suppression.This sequence optimizes the evaluation of postoperative scarring so that it can be distinguished more easily from fragments of degenerate fibrocartilage (Fig. 60.8).

The MRI examination proceeds much as for the cervical spine, the commonest indication being a myelopathy. Spatial presaturation is necessary to exclude the heart, and as far as possible the aorta. Cardiac gating via a digital pulse monitor may be desirable. As in the cervical spine, CSF movement is a significant problem (Fig. 60.6) and fast spin-echo techniques are not usually satisfactory in the axial plane.

Lumbar spine Again, the MRI procedure is similar to that followed for the cervical spine. CSF motion is minimal in this region, so fast spin-echo techniques can be used to good advantage even in the axial plane, allowing excellent visualization of the cauda equina (Fig. 60.7). The emphasis of the examination is influenced by the clinical problem. For disc disease and entrapment neuropathies, after the standard T1W and T2W high-resolution sagittal multislice acquisitions, an axial 2D multislice protocol is usually optimal, orientating the slice plane to that of the abnormal or clinically relevant discs, aiming to cover from pedicle to pedicle. The increasing spatial resolution on T2 weighting provided by modern MR systems makes the T1 axial sequence less necessary. For the postoperative lumbar spine, a more elaborate protocol is recommended. Sagittal T1W and T2W multislice sequences are made, followed by axial T1W images through the levels of interest. This is followed by a high-resolution fast spin-echo axial series through the same or more extended levels, to optimize detection of arachnoiditis as a cause for failed lumbar disc surgery. Then Gd-DTPA is given intravenously,

Figure 60.6 Sagittal MRI of thoracic spine using phased-array coils and a fast spin-echo technique (TR 4000 ms, TE 80), slice thickness 3 mm. Turbulent CSF flow posterior to the spinal cord is generating irregular areas of signal loss, simulating intradural masses.

Figure 60.7 High-resolution axial MRI of the lumbar spine using a fast spin-echo multislice technique (TR 4000 ms, TE 80, matrix 512 × 256, slice thickness 4 mm). The normal distribution of the intradural roots is well shown in this series. (A) L3. (B) L3–4 disc. (C) 6 mm below B. (D) L5. The rootlets at each nerve line up in the posterolateral part of the thecal sac and leave via their root sheaths in an anteroposterior orderly sequence. Arrow in (C) = roots entering the L4 root sheaths; arrow in (D) = roots visible within the L5 root sheaths. CSF yields high signal (white).

Figure 60.8 Axial MRI just below the L5–S1 disc, using a spin-echo acquisition (TR 600 ms, TE 20) with fat suppression. IV gadolinium has been given. The patient had had a right partial hemilaminectomy 18 months earlier. The right L5 root is shown embedded in enhancing scar tissue (arrowhead).

CHAPTER 60

• THE SPINE

COMPUTED TOMOGRAPHY OF SPINE CT is still sometimes used for lumbar disc disease and is essential in the evaluation of fractures. Adequate assessment of intradural structures still may require intrathecal contrast medium, though adequate visualization of the spinal cord to exclude compression usually can be achieved without intrathecal contrast medium on modern equipment.

POSITIONING Spinal CT is usually carried out in the supine position, with the spinal column as perpendicular to the plane of section as possible. The neck is moderately flexed to reduce the normal cervical lordosis but artefacts from dental fillings may limit this. Gantry tilt and flexion of the hips was useful in the past but has been superseded by multiplanar reconstructions.

A digital radiograph (‘scanogram’,‘scout film’) of the region to be studied is mandatory for selection of the levels to be examined and for retrospective localization of abnormalities. CT machines are now so fast that the patient can be instructed not to breathe or swallow during the data acquisition. The slice thickness depends on the clinical problem. Softtissue contrast is better on thicker slices. Sections 2–3 mm thick are appropriate for investigating lumbar disc disease.With modern multidetector systems, a volume of data is acquired; the workstation then allows virtually infinite viewing options. When soft-tissue contrast is less important, the thinnest possible sections may be selected and images reconstructed using a bone review algorithm. This provides maximum resolution of skeletal structures. This is also appropriate for most postmyelography CT. Various post-processing techniques allow graphic 3D images (Fig. 60.9).

Figure 60.9 Series of lumbar spine mutiplanar reformatted CT images showing (A) Sagittal reformat showing minor degenerative changes anteriorly on either side of the slightly narrow L3/4 disc. (B) Curved coronal reformat of the image in A demonstrating the entire ‘straightened’ lumbosacral spine; note how this has overcome the lumbar lordosis and the sacral curvature. (C) Surface shaded 3D reconstruction of another patient.

NORMAL ANATOMY VERTEBRAE The thin rim of compact cortical bone is shown well by both CT and MRI. The trabecular pattern of cancellous bone is best shown by high-resolution CT but may also be shown to some extent by good-quality MRI.Vascular channels are often visible, and may simulate undisplaced fractures on CT, but do not usually cause difficulty on MRI. Bone marrow is shown by MRI. Vertebral marrow is haemopoietic, but fatty marrow is usually present as well, constituting up to 50 per cent of spinal marrow volume in adults and increasing with age. Fatty marrow usually predominates

in specific areas of the vertebral body, namely adjacent to the vertebral end-plates and around the basivertebral veins in the central parts of the vertebral body. However, there is great variability, and unusually patchy replacement of red by fatty marrow may be normal. Marrow signal is more uniform, and much less influenced by fat, in children11. Sclerotic bands are often visible at sites of fusion between vertebral components.These are seen most often at the neurocentral junctions, and in the axis. In the latter, these may be three in number: (A) a V-shaped band just below the apex of the dens; (B) a midsagittal band in the rest of the dens; (C) a horizontal plate within the body of the axis demarcating

1361

1362

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

the body of C1 from the body of C2. Remnants of the subdental synchondrosis itself may persist throughout life, and it is common for the normal dens to appear to consist of diffusely denser bone than other vertebral bodies. Cortical bone is thicker in the neural arches and their appendages than in the bodies. The spinal canal is widest and almost circular at C1, and usually narrows in the midcervical levels and slightly widens in the lower, becoming more triangular in shape12. In the thoracic region the canal is almost circular and becomes wider and more triangular, especially in the lower lumbar spine. Although there is considerable individual and regional variation, it is intuitive that an intrinsically narrow canal provides little margin to cope with the inevitable degenerative changes of later life. In the lumbar spine an AP diameter of 11.5 mm or less is worrying13 and an AP diameter of 15 mm plus seems ideal. Of course precise measurements should not constitute the basis of decision making and in some ways cross-sectional areas are more relevant14; in the lumbar spine the cauda equina occupies about 50 mm2 and 100 mm2 is probably needed for reasonable CSF flow. The posterior intervertebral joints are essentially planar in the cervical and thoracic region, and slightly concavo-convex in the lumbar. Maximal rotation of the spine occurs in the thoracic facets. Articular surfaces constitute most of the lateral masses of the cervical vertebrae, but are carried by welldefined articular processes in the thoracic and especially in the lumbar region. Interarticular parts of the laminae can be defined in the lumbar region, and also in the axis.The synovial joint spaces often contain fat, especially near their margins, the fat lying within synovial fringes projecting into the joint cavities. The bone underlying the articular surface of the lateral mass of the axis is thinned focally by the underlying tortuous vertebral artery.

Figure 60.10 Series of four axial CT sections through the lumbar spine extending down to the L4–5 intervertebral disc, showing the intervertebral foramina and contained L4 spinal nerves (arrows). The L4–5 disc is protruding slightly on the left side (black arrowhead).

INTERVERTEBRAL DISCS Disc structure is well shown only in the lumbar region, where the discs are largest. On CT they show an amorphous texture and a relatively high soft-tissue density, usually clearly distinguishable from the fluid-filled thecal sac, extradural roots and vessels and epidural fat (Fig. 60.10). Internal structure and maturation with age are much better shown by MRI (Fig. 60.11). Maturation involves both the nucleus pulposus and annulus fibrosus independently, but usually concomitantly.The annulus thickens due to progressive acquisition of concentric layers of dense, organized collagen, and the nucleus pulposus acquires a transverse plate of dense collagenous tissue that bisects it into upper and lower halves. The sharp distinction in children between the more fibrous outer and the less fibrous inner component diminishes with age15. The disc margins may normally bulge up to 2–3 mm beyond the vertebral margins, especially in children. The posterior surface is normally flat or concave, not convex (except at L5/S1).

Figure 60.11 Midline sagittal T2W MRI of the lumbar spine with good demonstration of conus (normal) and cauda equina. The intervertebral discs down to L2/3 are relatively normal with degenerate discs between L3 and S1. At L4/5 there is extensive herniation of disc material, which is migrating caudally and extending posteriorly through the posterior longitudinal ligament.

CHAPTER 60

• THE SPINE

INTERVERTEBRAL FORAMINA

SPINAL CORD

These are best considered as spinal nerve root canals, of which the interpedicular foramen is only a part. In the C3–7 region, the inferior boundary of each canal is the superior concavity of the transverse process, which cradles the dorsal root ganglion posterior to the foramen transversarium (Figs 60.3, 60.4). The canals are oriented anterolaterally, but nearer the coronal than sagittal plane16.They are only about 5 mm wide in normal people, and can be difficult to image well. Axial images on CT or MRI should be as thin as possible. Volumetric MR acquisitions allow the data to be reformatted in planes parallel and transverse to the canals for full evaluation. However, images acquired in the sagittal plane usually also show the canals quite well, with the ganglia within. In the thoracic region the spinal root canals are short and well seen on images acquired in the sagittal plane. In the lumbar region the root canals are best considered as consisting of two parts: the lateral recess, which lies medial to the superior articular facet and pedicle in the lateral extremity of the spinal canal, and the infrapedicular part. The contents of the intervertebral foramen, which are visible on MRI (and some on CT), are epidural fat and fasciae, radicular arteries and veins, the emerging nerve roots and the dorsal root ganglia14,17,18.

The upper spinal cord may be shown by plain CT, especially in children, but detailed display of the anatomy of the surface of the cord by CT requires intrathecal contrast medium. Assessment of the size and shape of the cord is, at least in theory, easier to achieve on CT myelography than on MRI. With CT, the boundary can be defined by using the most appropriate window settings, determined from a histogram20. Conditions are more difficult to optimize on MRI, where boundaries can be displaced artefactually and there is more variability in the histogram. The normal internal structure of the cord is shown only by MRI; thicker sections (3–4 mm) usually optimize this, and the butterfly-shaped grey matter is shown on both T1W and T2W images (see Fig. 60.3). Midsagittal MRI of the cord often shows a thin longitudinal band in the centre of the cord, or slightly anterior to centre, which is usually mainly or exclusively a truncation artefact, but in some images it may represent the grey commissure. The size of the spinal cord can be quantified by several measurements, probably the most robust being cross-sectional area, but wide variation between the mean values in published normative data emphasizes the practical difficulties involved.

LIGAMENTS AND EPIDURAL SOFT TISSUES The dura mater is in direct contact with the posterior longitudinal ligament throughout most of the spinal canal. The posterior longitudinal ligament blends with the intervertebral discs, but between the discs it is separated from the concave posterior surface of the vertebral bodies by the anterior epidural space. This contains fat, the basivertebral veins and the anterior internal vertebral veins, and is divided into left and right compartments by a tough midline septum that influences the course of disc herniations19. This epidural space is well developed only in the lumbar region. Laterally, the dura mater is usually in contact with the medial extremity of each pedicle, although epidural fat may intervene. Below and between the pedicles, the dura is in contact with the epidural fat in continuity with the intervertebral foramina. Posteriorly, the dura mater is usually loosely attached to the anterior cortex of the spinous processes. Between the upper margins of this cortex there are wedges of epidural fat separating the dura from the laminae, posterior joints and ligamenta flava, sometimes called the interlaminar fat pad14,18 (see Fig. 60.11). In kyphotic regions (usually the thoracic region) the dura is frequently separated from all the posterior elements of the spinal canal by a layer of epidural fat that may be more than 5 mm thick over the most kyphotic region; this normal appearance has been misinterpreted as idiopathic epidural lipomatosis in some clinical contexts. The volume of the spinal canal occupied by epidural fat can vary greatly, and the dura may be separated from all walls of the canal by fat.

SPINAL NERVES High-resolution techniques are necessary to demonstrate the intradural rootlets emerging from the spinal cord and the larger roots of the cauda equina.Thin T2W images on modern machines show the rootlets, which coalesce to form anterior and posterior roots that penetrate the dura through separate ostia; these form CSF-filled pouches, the lateral extremities of which are often called the subarachnoid angles of the root sheaths. The sheaths end proximal to the dorsal root ganglia. Eccentric cystic expansions of the root sheaths are common at all levels, and are often called perineural cysts and sometimes Tarlov cysts21. They may enlarge an intervertebral or sacral foramen; they are particularly frequent, and are largest, on the S2 roots. It remains debatable as to whether these can contribute to symptoms22. The roots of the cauda equina appear as bilateral, often circular bunches near the conus medullaris, as a crescentic aggregate in the posterior part of the thecal sac in the midlumbar spine, and lateral and very peripheral at L5. The roots inferior to S1 are smaller, and often very peripheral, making them difficult to see. Some variation and asymmetry in distribution of roots is acceptable. In the lumbar region, the anterior and posterior roots align in the lateral extremity of the thecal sac. For spinal nerves L1–L5, they run in the subarticular part of the spinal root canal (the lateral recess) and cross the disc space within the thecal sac, emerging with their sheaths near the lower margins of the pedicles of the same number. The first sacral nerve usually leaves the thecal sac above the L5–S1 disc, which it crosses usually enclosed in the longest of the lumbar root sheaths to reach the S1–S2 foramen; variations in anatomy

1363

1364

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

may contribute to the variations in symptoms found between patients with similar disc lesions. See also conjoined roots below.The dorsal root ganglia are readily identified at all levels on MRI, but usually only in the lumbar region by plain CT.

The anterior roots and rootlets can be identified using highresolution MRI.The dorsal root ganglia enhance strongly after IV contrast medium, while normal intradural roots do not, or perhaps only minimally.

MYELOGRAPHY This is rarely indicated nowadays in well-equipped imaging departments, because it is invasive and less informative than MRI. It is only contemplated when MRI is unavailable or contraindicated, or is unsatisfactory due to patient tolerance or unusual shape, such as limb contractures or severe scoliosis. Until the late 1970s, the only contrast agents available were iophendylate oil (Myodil, Pantopaque), which led to considerable long-term arachnoiditis, and air or carbon dioxide. Now there are numerous non-ionic water-soluble contrast media available for intravascular use, but many have not been licensed for intrathecal use. The rate of absorption from the subarachnoid space is extremely variable, but the mean half-life in the body is about 12 h, so that 80–90 per cent is excreted via the kidneys within 24 h. Neurotoxicity is low but not negligible. The maximum concentration recommended for intrathecal use is 300 mg I ml−1, up to a maximum dose of 3 g of iodine. Contrast medium is usually introduced by lumbar puncture using a small 22-gauge needle, L2–3 or L3–4 being the levels to use routinely. The easiest position in which to perform a lumbar puncture is with the patient sitting. However, many experienced myelographers prefer the prone position, with a folded pillow or bolster placed under the abdomen. From this position the patient does not need to be moved before commencing the actual myelogram and it is easy to check the position of the needle with fluoroscopy if the puncture proves difficult. The lateral decubitus position is less often preferred but sometimes it is the only position possible. With the advent of MRI and reduced need for myelography, lateral cervical puncture is rarely performed. Few radiologists outside the major neuroradiological centres will have any experience of it and there is always the potential risk of damage to the spinal cord. Contrast medium dilutes at a variable rate, and in a total study, the first region to be examined is usually demonstrated best. In lumbar myelography the patient is placed prone and the table tilted 45 degrees or more in the head-up position to ensure that the thecal sac is completely filled by contrast medium. Anteroposterior paired 10–15 degrees and 30–40 degrees oblique projections are made, followed by lateral views, which should include most of the sacral canal. Cervical and thoracic myelography are rarely performed nowadays and should only be carried out at specialist centres where radiologists are skilled in these procedures. A myelogram demonstrates the spinal subarachnoid space, which in normal subjects delineates the position of the spinal dura mater. Sometimes spinal nerves at adjacent levels penetrate the dura at a single intervertebral level (conjoined roots). One root sheath is missing at the adjacent intervertebral level, and the thecal sac sometimes shows a long, smooth lateral concavity so that simulation of an elongated epidural mass can be close. However, the resulting conjoined root sleeves

are large, and the composite roots are usually clearly visible. The commonest roots to be conjoined are L5 and S1, but sometimes the L4–5 roots are involved and rarely other levels (Fig. 60.12).

SPECIAL SITUATIONS Myelography can be difficult in certain situations and it is, unfortunately, in just these situations that MR may be problematic and myelography most likely to be considered! In severe multilevel degeneration in the lumbar spine, the spinous processes may remain closely applied to one another no matter how hard the operator tries to separate them by spinal flexion. In these circumstances it is easiest to use an oblique approach to the spinal canal for which fluoroscopic guidance is essential. Spinal stenosis most commonly involves the lower lumbar region, especially L4–5 and L3–4. Therefore it is preferable to perform the lumbar puncture at L2–3, or at L5–S1.

Figure 60.12 Lumbar myelogram, anteroposterior projection showing a relatively common anatomical variant: conjoined nerve roots. The left L4 and L5 roots penetrate the dura mater (arrows) close to the pedicle of L4, resulting in marked asymmetry. Apparent absence of the right L5 root sheath could be confused with compression.

CHAPTER 60

Following laminectomy, the myelographer is advised to avoid attempting to enter the spinal canal through a laminectomy scar wherever possible, even though the available space may seem inviting. Postoperative infection can be chronic and occult, and adhesion of neural structures and arachnoiditis at such sites may be sufficiently common for attempts at puncture to be an excruciatingly painful experience for the patient. It is best to puncture one or two interspaces away from a lumbar laminectomy, even if this means at L1–2. The operator should avoid making a puncture through defects in the neural arches, because the spinal cord may be low and tethered posteriorly in the vicinity of such defects.

COMPLICATIONS OF MYELOGRAPHY Lumbar puncture Between 10 per cent and 35 per cent of patients suffer from headache following simple lumbar puncture. It is generally attributed to loss of spinal fluid through the needle puncture hole. There is evidence that the incidence and severity are reduced if small-gauge needles are used. Infection associated with the puncture is very rare. There have also been reports

• THE SPINE

of intraspinal haemorrhage from lumbar puncture, especially epidural haematomas23, and myelography should not be performed in patients with abnormal haemostasis. Contrast medium ultimately penetrates the Virchow–Robin spaces and freely enters the interstitium of the cerebral cortex, even after lumbar myelography. Side effects do not appear to be reduced by keeping the patient recumbent for a few hours after the myelogram. Back pain and radicular pain occur in up to 30 per cent of cases, sometimes exacerbating presenting symptoms. This is usually only transient, lasting not more than an hour or two. Even with the contrast agents now available, a slight increase in CSF protein with an appearance of few white cells occurs in some cases. Intradural inflammation virtually always used to occur after iophendylate (Myodil) oil myelography, resulting in intradural adhesions recognizable on subsequent water-soluble myelography in 70 per cent of cases24. It is now known that Myodil was the major cause of arachnoiditis after all forms of lumbar disc surgery, not the surgery itself. Adhesive arachnoiditis has not been observed after the clinical use of the non-ionic water-soluble contrast media now available.

SPINAL ANGIOGRAPHY CT angiography can be used as a preliminary screen but formal spinal angiography25 may be needed for a few specific situations and such procedures should generally be carried out in specialist centres where the radiologists are experienced in the procedure. Indications for spinal arteriography include suspected angiomatous malformations or vascular tumours of the spinal cord, meninges or vertebral column. It may follow negative cerebral angiography in the investigation of subarachnoid haemorrhage. Therapeutic embolization may be carried out. Some neurosurgeons require demonstration of major arterial supply to the spinal cord before any surgery that might compromise them (scoliosis correction or costotransversectomy for thoracic disc protrusion, etc.) Since angiography is a costly time-consuming procedure with definite morbidity, no patient should be submitted to it if no action will be taken as a result of the findings. Patients considered unfit for surgery should not have spinal angiography. Nearly always, a positive MRI or myelogram should precede the angiographic search for a spinal cord tumour or arteriovenous malformation. Complications include deterioration in clinical myelopathy, relatively common but usually transient; permanent cord damage is rare.

ANATOMY The spinal cord is supplied via three main longitudinal arterial axes: the midline anterior spinal artery and two postero-

lateral spinal arteries26 (Figs 60.13 and 14). In the cervical region, these arteries arise from the vertebral and deep cervical arteries, themselves branches of the subclavian arteries, while more inferiorly they arise from the posterior intercostal or lumbar arteries. At each vertebral level, a radicular artery runs alongside the nerve root, and some of these, the radiculomedullary arteries, continue to the spinal cord and constitute the major sources of blood for the anterior or posterior spinal arteries. One major radiculomedullary artery (Adamkiewicz) is found in the thoracolumbar region; its level of origin is variable, but it is usually on the left side, between T8 and L1–2 (Fig. 60.15). The number and origins of the posterior radiculomedullary arteries vary, and occasionally a single posterior intercostal artery will give rise to both anterior and posterior radiculomedullary branches. In the cervical region, anterior radiculomedullary arteries arise at every vertebral level, but they vary considerably in size and importance. The anterior spinal artery is by far the most important of the axes, because it supplies the major portion of the cord substance, including the motor cells of the anterior horns. It gives off tiny sulcocommissural arteries that run into the cord; they are not visible at angiography unless pathologically enlarged. Arteries supplying the vertebral column also arise at each vertebral level; their study involves catheterization of the same vessels as arteriography of the spinal cord.

1365

1366

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 60.13 The spinal arteries and their feeding vessels. a = aorta, b = intercostal artery, c = anterior spinal artery, d = posterolateral spinal artery, e = anastomosis at conus medullaris, f = arteria radicularis magna (of Adamkiewicz), g = thoracic and h = cervical radiculomedullary arteries, i = vertebral artery, j = deep cervical artery.

Figure 60.14 Blood supply of the spinal cord. A = anterior spinal artery, B = posterolateral spinal arteries, C = posterior anastomotic artery, D = posterior and E = anterior radiculomedullary arteries, F = sulcocommissural arteries, G = lateral anastomotic artery.

Figure 60.15 Anterior spinal artery (3) in cervical region, fed (A) from vertebral artery (1), and (B) from a deep cervical artery (4), via radiculomedullary arteries (2). The posterior spinal artery (5) is also opacified in the lateral projection (C).

CHAPTER 60

• THE SPINE

DISCOGRAPHY Discography is still employed in some orthopaedic centres, though mainly as a diagnostic preliminary before spinal fusion for the treatment of back pain or neck pain, rather than neural compression syndromes. Although special apparatus has been devised to perform the test, it is simply performed by inserting a 22-gauge spinal needle into the centre of a disc under fluoroscopic control from an oblique approach, anteriorly in the neck and posteriorly in the lumbar spine. The spinal canal and intervertebral canals are avoided. A small amount of non-ionic contrast medium is injected into the central part of the disc

(the nuclear ‘cavity’) and anteroposterior and lateral radiographs obtained. The test is sometimes followed by CT. Discography can be an unpleasant, painful experience for the patient. Artefacts due to inaccurate placement of the needle in the nuclear cavity, or leakage of contrast medium along the needle track, are frequent. Advocates claim that it usually provokes specific pain when the appropriate disc is injected, and may indicate tears in the annulus fibrosus that would not have been shown by any other technique. There is significant uncertainty associated with both these claims.

RADIOLOGICAL DIAGNOSTIC APPROACH TO COMMON CLINICAL SPINAL PROBLEMS Clinical problems may be divided into acute and chronic presentations.

ACUTE PRESENTATIONS 1 Back pain 2 Root syndromes 3 Spinal cord lesions.

Acute back pain Acute back pain and many cases of acute extremity pain of presumed spinal origin are not generally investigated radiologically; rest and conservative treatment are tried first. Plain radiographs may, however, be indicated in cases of pain following significant trauma, and in patients known to develop spinal lesions, such as those with metastatic disease or on steroid therapy. Increasingly MR is being used as the initial investigation for all forms of spinal disease and at an earlier stage in the diagnostic sequence, especially for those clinical situations ‘red flagged’ by guidelines27.

Acute root syndromes Acute lower motor neurone weakness of one arm is uncommonly due to spinal disease in the absence of trauma. Nerve root avulsion is generally not investigated radiologically in the acute stage, but may subsequently require high-resolution MRI. Lumbosacral disc protrusion may present with acute sciatic nerve root compression that is predominantly motor, giving an acute foot drop; acute unilateral leg pain or loss of sphincter function may be present. Although plain CT is nearly as good as MRI at identifying extradural disc protrusion, MRI has become the preferred investigation.

Acute lesions of spinal cord Acute traumatic lesions of the spinal cord are often definitive and permanent. Careful clinical assessment in the early stages may, however, indicate a partial lesion with potential for some recovery. Most patients with major trauma now undergo CT, which may be definitive; MRI may be indicated, especially if plain radiographs and CT are normal.

An acute nontraumatic lesion of the spinal cord is assumed to be compressive until proven otherwise and is a medical emergency, since the degree of recovery may be related to the speed of diagnosis and treatment. A chest radiograph is mandatory, since many cases are neoplastic or infective. Plain radiographs are often still obtained, but MRI of the clinically involved region – and preferably of the whole spine if metastatic disease is suspected – is much the preferred investigation and usually will be all that is required.Water-soluble myelography by the lumbar route supplemented by CT may be necessary if MRI cannot be obtained. Spinal angiography is not indicated unless other imaging findings indicate the presence of a highly vascular lesion. There is probably nothing to be lost in delaying angiography until the acute stage has passed.

CHRONIC PRESENTATIONS 1 Pain 2 Root compression syndromes 3 Cord lesions.

Chronic pain Chronic pain is still widely investigated by plain radiographs, which provide a rapid, cheap overall picture. However plain radiographs are usually noncontributory28. Increasingly MRI is being used at the outset.

Root compression syndromes In chronic sciatica, MRI is the preferred investigation. Myelography may occasionally help to identify the worst affected level in cases with multilevel disease, although most surgical decisions are now made on the basis of noninvasive tests alone. When pain in the arm or shoulder is thought to be of spinal origin, plain radiographs of the cervical spine are appropriate.They may show evidence of degenerative disease, cervical ribs or, more rarely, evidence of neoplasm or infection. Most patients with cervical spondylosis are treated conservatively, although MRI may be required to make therapeutic decisions.

1367

1368

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Chronic cord lesions Three main groups may be identified: extramedullary compressive lesions, intramedullary space-occupying lesions (mostly tumours and syringomyelia) and inflammatory or vascular disease. Chronic compressive lesions are often cervical: spondylosis and cervical disc disease are the most frequent. They are increasingly being investigated by MRI rather than plain radiography. Intrinsic space-occupying lesions of the spinal cord

are rarely evidenced by plain radiographs, but the detection of craniocervical anomalies may orientate the diagnosis towards syringomyelia. MRI is the definitive procedure, including views of the craniocervical junction. If inflammatory or demyelinating cord disease is suspected, MRI of the brain may provide valuable evidence of associated lesions, in addition to MRI of the spine. If MRI is negative, it is extremely unlikely that angiography will be revealing; even demonstration of anterior spinal artery occlusion is unreliable.

DISEASE PROCESSES AFFECTING THE SPINE CONGENITAL LESIONS The commonest congenital malformation found in adults involves the caudal part of the neural tube. A useful descriptive term is lipomyelomeningodysplasia, which emphasizes the varying elements. Other conditions result in a dorsal dermal sinus that runs from a skin dimple to the spinal canal, some terminating in intraspinal dermoids or epidermoids.The neurenteric canal or adhesion may persist, or form at other levels such as the upper cervical region, resulting in neurenteric cysts along the connection between the foregut and spinal canal; a persistent cutaneous communication results in a dorsal enteric fistula. Aberrant neuro-entodermal adhesions are probably also the origin of diastematomyelia. Finally, excessive retrogressive differentiation can lead to varying degrees of sacral and sacrolumbar dysgenesis, often referred to as the caudal regression syndromes.

Figure 60.16 Intramedullary lipoma (arrow) with intact dura in upper thoracic. Sagittal (A) and axial (B) T1W MRIs.

SPINAL MALFORMATIONS Intramedullary lipoma This consists of a mass of adipose tissue located mainly between the posterior columns of the spinal cord. A tongue-like extension along the central canal is often found, in keeping with its embryogenesis. The overlying dura mater usually is intact and the lipoma entirely intradural; however, there may be a dural defect to which cord and lipoma become adherent. Such lesions occur most often near the thoracocervical or craniovertebral junctions (Fig. 60.16). CT and MR demonstrate the fatty nature of the tumour. Nonfatty elements also may be present, resulting in a heterogeneous appearance.

LIPOMYELOMENINGODYSPLASIAS These represent a spectrum of abnormalities ranging from an abnormally low location of the conus medullaris with minimal or even absent lipoma, to massive lipomatous formations involving all elements of the spinal and adjacent subcutaneous tissues (Fig. 60.17). The abnormality may not be apparent clinically, hence the term occult spinal dysraphism. Unlike myelocele, Chiari malformation is present in only about 6 per cent. Patients frequently present in adult life, sometimes only with back pain and minimal neurological signs.

Figure 60.17 Lipomyelomeningodysplasia. Sagittal (A) T1- and fast spin-echo (B) T2W and coronal (C) T2W images showing the lipoma, low position of the spinal cord and a cavity in the distal spinal cord.

Plain radiographs often reveal varying degrees of nonfusion of one or more neural arches and variable expansion of the spinal canal; such incidental findings do not necessitate further investigation unless there are relevant clinical features. Even malformations involving the spinal cord may remain asymp-

CHAPTER 60

tomatic, and the efficacy of prophylactic operations to release tethering remains unclear. MR is the optimal investigation although ultrasound (US) has been advocated in children; MR demonstrates all aspects of the abnormalities including the origin of nerve roots and associated lesions such as cysts in the spinal cord. In over 80 per cent, the spinal cord terminates at or below the level of the third lumbar vertebra, and usually is tethered to the dorsal aspect of the dura, where it fuses with the fatty tumour. Nerve roots issuing from an apparently thickened filum terminale indicate that it contains significant nervous tissue and therefore should not be divided surgically.

NEURENTERIC AND OTHER DEVELOPMENTAL CYSTS Intraspinal neurenteric cysts are intradural, usually unilocular cysts lined by gastrointestinal or bronchial epithelium that occur in either the cervical (often near the craniovertebral junction) or lower thoracic regions29. They compress the spinal cord (usually the anterior aspect) and may invaginate into its substance; occasionally the cord is split into two halves as in diastematomyelia. Plain radiography may show focal expansion of the spinal canal, and thoracic lesions in particular may be associated with butterfly or hemivertebra. On MRI the cyst contents usually yield a slightly higher signal than cerebrospinal fluid, on T1- as well as T2W images, but are clearly demarcated from cord substance. Dermoid and epidermoid cysts are rounded intradural and sometimes intramedullary lesions. Imaging may demonstrate fat within them and calcification.They may be associated with other forms of dysraphism, and in about 20 per cent a dorsal dermal sinus can be traced running obliquely downwards from the lesion to a skin dimple on the lower back, which also may be a source of intradural sepsis30. Ependymal cysts usually occur with other types of dysraphism. They represent little more than focal dilatations of the central canal of the spinal cord, and usually appear as swelling near the lumbar enlargement.

• THE SPINE

Plain radiographs in the region of the abnormality usually show focal expansion of the spinal canal, narrow intervertebral disc spaces and varying degrees of laminar dysplasia and fusion. This combination is very suggestive of the diagnosis, whether or not a bony spur is shown. MRI (and post-myelography) CT show the anatomy well; even plain CT may be diagnostic in some circumstances (Fig. 60.18).

CHIARI MALFORMATIONS These represent a group of abnormalities characterized by dislocation of the hindbrain into the spinal canal. Chiari described four types, but his types III and IV are rare and the Chiari II is seen mainly in paediatric practice. The Chiari I lesion is not really a malformation at all: it may be acquired in conditions associated with raised intracranial pressure (tumours, venous hypertension), lowered intraspinal pressure (lumboperitoneal shunts) and conditions that diminish the volume of the posterior fossa (craniosynostosis, basilar investigation). It may develop during the first 3 years of life32 in the absence of any cause other than probable slower growth of the posterior fossa relative to the hindbrain during this period when both are growing most rapidly. The Chiari I lesion used to be distinguished from coning, by the shape of the herniated tonsils, which were said to be rounded in coning and pointed in Chiari, but it is now clear that both pointed tonsils and medullary elongation may be found in situations associated with coning, and that there are no grounds for distinguishing between the two conditions.

Chiari type I lesion (cerebellar ectopia) This is defined as descent of the otherwise normal cerebellar hemispheres below the foramen magnum, usually involving the tonsils. However, in about 50 per cent, the other component of the hindbrain, the medulla oblongata, shows elongation of the segment between the ponto-medullary junction and dorsal column nuclei, the obex of the fourth ventricle coming to lie in the cervical canal, where it may or may not be overlain by the cerebellar tonsils, the prevalence of medullary elongation increasing with increasing tonsil descent.

DIASTEMATOMYELIA The spinal cord is split into two usually unequal hemicords, each with a central canal and anterior spinal artery, but giving rise to only ipsilateral spinal roots. Any level can be involved, including the filum terminale and medulla oblongata, but most are thoracic31. The cleavage usually extends over several segments and only rarely the hemicords do not re-unite caudally. The hemicords are enclosed in a common dural tube in 50 per cent of cases, usually in the cervical region, but in the remainder each is enclosed within its own dural tube, a bony or cartilaginous spur arising from malformed lamina often lying between them. Abnormal traction may be exerted by such a spur at the point of reunion of the hemicords. Clinical abnormalities are often absent, but in some symptomatic cases progression apparently was halted by excision of tethering spurs.

Figure 60.18 Diastematomyelia. An axial T2W fast spin-echo sequence showing two almost equal-sized hemicords (arrowheads) in a common thecal sac at L3.

1369

1370

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Elongation is sufficient to produce a kink on the posterior surface of the medulla oblongata in about 15 per cent, where the tail of the fourth ventricle rolls down over the upper one or more segments of the spinal cord (Fig. 60.19). The prevalence of the Chiari type I lesion in the normal population has been considerably overestimated on MRI, and is probably under 1 per cent33.

Chiari type II malformations In these malformations the cerebellum is dysplastic. The inferior vermis is everted rather than inverted, so the nodulus becomes the most inferior part of the cerebellum and the fourth ventricle is reduced to a coronal cleft. The cerebellar herniation then consists mainly of the inferior vermis. The medulla oblongata is invariably elongated, usually enough to become kinked. Hydrocephalus and dysplasia of the cerebral hemispheres, cranial vault and meninges are frequent. A meningomyelocele is present in 98 per cent or more of cases, and may play an important role in embryogenesis34. These hindbrain abnormalities may be associated with compression and progressive degeneration in parts of the brainstem, cerebellum and upper spinal cord. In Chiari type I malformations, symptoms commonly do not appear until adult life, and about 50 per cent of symptomatic cases are associated with syringomyelia. Syringomyelia occurs most commonly when the cerebellar tonsils lie between the neural arches of C1 and C2, whereas cerebellar syndromes predominate when the tonsils lie lower than the neural arch of C235.The outcome of foramen magnum decompression is significantly adversely affected by increasing descent of the tonsils35. In the type I lesion, plain radiographs are usually normal. In up to 15 per cent, occipitalization of the atlas and basilar invagination are present, and hindbrain herniation is present in over 50 per cent of cases of symptomatic basilar invagination36. By far the most convenient way to demonstrate these hindbrain abnormalities is by MRI. However, descent of the cerebellum through the foramen magnum of up to 3 mm

seems to be present in up to 20 per cent of normal subjects on midsagittal MRI sections of 4 mm or more in thickness, due to the shape of the foramen magnum and partial volume effects in which the more laterally placed biventral lobules are included in the image when the vallecula is small33. MRI in the coronal plane is more reliable, and the identification of associated features, such as elongation of the upper medulla, can also be helpful.

MENINGOCELES Varying degrees of dural ectasia usually accompany the spinal dysraphisms. Both generalized and focal dural ectasia may occur in systemic disorders such as neurofibromatosis, Ehlers–Danlos and Marfan’s syndromes. It may occur in erosive arthropathies, especially ankylosing spondylitis37, where focal ectasia sometimes forms pockets or saccules invaginating into the walls of the spinal canal, including the vertebral bodies and neural arches. Such lesions also occur idiopathically.

Anterior sacral meningocele This lesion consists of a unilocular, complex lobular or even multilocular presacral cystic mass, containing CSF, which communicates with the intraspinal subarachnoid space. There is a large usually eccentric anterior defect in the lower part of the sacrum, and the sacral canal is expanded. Varying degrees of sacral and coccygeal agenesis may be associated. On plain radiographs, the eccentric anterior sacral defect gives the remaining part of the sacrum a pathognomonic scimitar appearance. The pelvic mass may be shown by US, CT or MRI, the latter invariably demonstrating communications with the sacral canal. Occult intrasacral meningocele is a variant of this condition. The sacral canal is expanded by a meningocele that lies below the normal level of termination of the thecal sac. There is no anterior sacral defect and no intrapelvic extension.

Lateral thoracic meningocele Figure 60.19 Chiari type 1 malformation. T1W sagittal image showing the cerebellar tonsils extending just caudal to the neural arch of C1, and elongation and kinking of the medulla oblongata.

This lesion commonly presents as a paravertebral mass on CXR. It is commonly solitary and usually is found on the right; 70–85 per cent is associated with neurofibromatosis. There is typically an angular kyphoscoliosis towards the side of the meningocele, and pressure erosion of the margins of the relevant intervertebral foramen is evident.

Anterior thoracic meningocele with ventral herniation of spinal cord This is an increasingly recognized condition very occasionally providing explanation for an otherwise unexplained chronic thoracic myelopathy in adults38. It is most readily recognized on midsagittal MRI of the thoracic spine, where the spinal cord is displaced anteriorly in contact with a vertebral body at or very near an intervertebral disc, commonly at about T6. The meningocele may not be easy to show on axial images. Appearances are often misinterpreted as an intradural arachnoid cyst displacing the spinal cord anteriorly, from which the condition needs to be distinguished.

CHAPTER 60

• THE SPINE

VERTEBRAL FUSION ANOMALIES These usually are determined very early in development. The intervertebral discs are narrow and partly bridged by regions where disc material never developed. Fused segments usually also show varying degrees of hypoplasia, and when multiple segments are involved, marked dysplasias such as hemivertebrae are also often present. The term Klippel–Feil syndrome is appropriate when the cervical region is predominantly involved. When only the thoracic spine (which is much more rare), spondylothoracic and spondylocostal dysplasia are more appropriate terms39,40. Other organ systems may be involved, including neural tube and visceral abnormalities.

SKELETAL DYSPLASIAS Achondroplasia Neurological complications are present in 40–50 per cent of cases, the average age of onset being 38 years, rarely under 15 years40.The spinal canal is congenitally narrow and the narrowing is progressive due to degenerative changes and an exaggerated lumbar lordosis. In over 60 per cent of cases, neural compression occurs in the thoracolumbar region, most often of the cauda equina; it is generalized in only 10 per cent. In the remainder, compression occurs mainly in the cervical spine or at the foramen magnum, the latter accounting for virtually all the childhood and especially neonatal presentations.

The mucopolysaccharidoses In the Morquio-Brailsford type (MPS-IV), ligament laxity results in instability and subluxations at the atlanto-axial joint and thoracolumbar junction; the former may require arthrodesis41. Ligamentous thickening is severe in about 50 per cent and is the main cause of upper spinal cord compression. Especially marked and extensive thickening of the dura and extradural tissues has been described in Hurler–Schiei and Maroteaux-Lamy diseases.

Spondyloepiphyseal dysplasia Vertebrae are flattened and enlarged in an AP dimension, especially in a thoracic kyphosis. Severe scoliosis may be present. Neurological complications are uncommon, and usually arise from atlanto-axial instability associated with abnormal ossification of the dens similar to MPS-IV, but usually lacking the soft-tissue thickening. Multiple epiphyseal dysplasia, pseudo-achondroplasia and chondrodysplasia calcificans are also conditions that may have this feature41.

A small number of cases show subluxations, usually involving C1/2 or C2/3. Spinal compression develops in about 16 per cent of cases. A myelopathy developing in childhood may be more likely to be due to spinal deformity than to a neurofibroma or other tumour39.

SPINAL INSTABILITY Craniovertebral junction Isolated atlanto-axial instability is occasionally encountered and may be associated with spinal cord compression. It occurs in about 30 per cent of patients with Down’s syndrome. About 70 per cent of cases are associated with an os odontoideum (Fig. 60.20), and most of the remainder with cranial assimilation of the atlas, which is combined with a C2/3 intersegmental fusion in about 70 per cent of cases. Os odontoideum should probably no longer be regarded as a segmentation anomaly, but rather as abnormal ossification of the dens, secondary to instability and abnormal local stress. Congenital hypoplasia of the dens usually is a misdiagnosis, an os odontoideum in fact being present but overlooked because it is not ossified, small or malplaced. True hypoplasia of the dens only occurs in association with more complex fusion anomalies, especially those that restrict rotation at C1/236.

Spondylolysis Truly congenital spondylolysis is uncommon. It is often associated with other defects, such as absent pedicles, absent superior articular facet, hypoplastic laminae with deviation of the spinous process and hypertrophy of the contralateral pedicle. Such abnormalities may be found in the cervical and lumbar regions, and usually are not associated with neurological symptoms, unless severe subluxation is present, such as sometimes occurs at C2 or C3. The spondylolytis defects, which are relatively common at the lumbosacral junction, especially in young adults, probably originate as stress fractures through the interarticular part of the laminae, resulting in a usually hypertrophic pseudarthrosis. The anatomy is well shown on high-resolution axial and sagittal MRI, or multiple thin-slice CT with reformatting. Special imaging planes may be helpful42.

NEUROFIBROMATOSIS This disease is transmitted as an autosomal dominant disorder, but 50 per cent of cases lack a family history. The type I form is usually associated with skeletal dysplasia. About 50 per cent have a scoliosis that can be very severe, about 10 per cent have dysplastic vertebrae (often consisting of one or more absent or hypoplastic pedicles) and about 10 per cent have dural ectasia.

Figure 60.20 Os odontoideum (os mobile) (arrows) and atlanto-axial instability: cervical myelogram, lateral projections. There was no history of trauma. The pattern of instability, with posterior atlanto-axial subluxation in extension (B), reduced in flexion (A), is common.

1371

1372

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

INTRASPINAL ARACHNOID CYST Extradural arachnoid cysts arise from defects in the dura mater, either congenital or inflammatory (e.g. ankylosing spondylitis); intradural arachnoid cysts arise from arachnoidal duplications or spinal arachnoiditis. Symptoms of pain or neurological disability may arise when the spinal cord or cauda equina are compressed, bearing in mind that size and intracystic pressure may vary considerably; occasionally the spinal cord or roots have herniated through a dural defect and become entrapped. Plain radiographs may show expansion of the spinal canal when the cyst is extradural. MRI shows these lesions well. Signal from fluid in the cyst, and often from the subarachnoid space below it, is usually higher than from CSF elsewhere due to reduced mobility. Effects on the spinal cord are shown, namely compression and rarely myelomalacia or syringomyelia. Small herniations of neural tissue through the dural defects may require thin imaging sections and high-resolution imaging techniques to show them. Aspiration and drainage of arachnoid cysts compressing the spinal cord may be accompanied by immediate and dramatic improvement in clinical condition43. Care must be exercised not to overdiagnose intradural arachnoid cysts in the thoracic region. The retromedullary subarachnoid space in the thoracic spine is commonly wide, and partly loculated by usually incomplete septae; the spinal cord usually is closely applied to the anterior margin of the bony canal and may have a flattened appearance over an exaggerated kyphosis. Perineural arachnoid cysts (Tarlov cysts) occur commonly in the sacrum, especially on the second sacral root. They can be large, multiple and are often associated with eccentric pressure erosion of the sacral canal and are well shown by MRI. Clinical significance, even of large cysts, is doubtful22.

SYRINGOMYELIA The term ‘syringomyelia’ describes conditions in which there is a cavity within the spinal cord, lined mainly by glial tissue and containing fluid that is similar or identical with CSF. It is associated with a number of distinct pathological processes, including cerebellar ectopia and trauma44. Whatever the cause, the cavity seems capable of propagating, probably due to hydrodynamic forces, into normal cord tissue. Usually it involves many segments or the whole spinal cord, but sometimes smaller isolated cavities are found confined to only a few spinal segments, often referred to as fusiform syrinx. The cervical cord is involved most often, although occasionally only the thoracic cord. Only about 10 per cent of cysts extend cranial to C2, where they split into two or deviate to right or left in a plane ventral to the floor of the fourth ventricle. Small cavities usually involve the bases of the posterior columns, and commonly the central cord canal for part of their extent; larger cavities are associated with more extensive loss of cord substance. Double cavities are sometimes present and individual cavities may be multilocular. The spinal cord is enlarged in about 80 per cent, normal in size in 10 per

cent and diffusely atrophic in 10 per cent. The size of the spinal cord may vary in response to posture and respiration. Size variation usually is not associated with changes in clinical state, nor does the severity of clinical disability relate to the size of the cyst and remaining cord substance45. Between 70 per cent and 90 per cent of cases of syringomyelia are associated with cerebellar ectopia, the cerebellar tonsils usually lying at the level of C1 or between C1 and C235. It is postulated that intermittent obstruction of the outlets of the fourth ventricle and of CSF flow across the foramen magnum, combined with a patent communication between the fourth ventricle and the central canal of the spinal cord, together produce secondary degeneration of cord substance by causing intermittent distension of the central canal46. The term syringohydromyelia is often used for this type of cord cavitation. However, this hypothesis has some problems and does not explain all aspects; alternative hypotheses exist47. Other causes of cavitation of the spinal cord can result in appearances indistinguishable from syringohydromyelia. These include intramedullary tumours, chronic spinal cord compression, spinal cord trauma and arachnoiditis. Acute transverse myelitis, infection agents and other inflammatory processes such as sarcoidosis, can result in colliquative necrosis (myelomalacia), which may organize into cavities that propagate into normal cord substance. Finally in about 10–20 per cent of patients with spinal cord cavities, no cause or association can be found. On plain radiography, the spinal canal appears expanded in 30–40 per cent of cases, especially in the cervical region, the highest frequency being in children with Chiari-associated syringomyelia48. Scoliosis often occurs and may be the presenting feature49. Syringomyelia and its cause usually are shown well by MRI. The cavity is well circumscribed and of uniformly abnormal signal, often showing prominent transverse ridges in its wall. In general, the cyst fluid yields a similar signal to CSF on T1W and T2W images (Fig. 60.21). On T2W and other types of images, the signal is more variable; pulsatile cysts show flowrelated signal changes, nonpulsatile cysts do not50. Dynamic MRI, using phase contrast or bolus-tracking techniques, has been used to study CSF movement especially at foramen magnum level51. Mainly in post-traumatic syringomyelia, T2W images have shown a striking increase in signal from the full thickness of an otherwise normal cord or medulla oblongata well beyond the visible extent of the cavity, probably reflecting the extent of actual cord damage. Care must be taken not to exclude an underlying tumour, although these are usually obvious (e.g. Fig. 60.22). On MRI, moderate correlation is found between the presence and location of the cavity and clinical features, but not between clinical severity and size of the syrinx relative to remaining cord substance, or with its degree of distension45,52,53. MRI is good for monitoring the mechanical success of operative strategies54, of which three are in current use, the third being new and controversial, deriving from an alternative theory of causation: (A) foramen magnum decompression; (B) syringosubarachnoid, peritoneal or pleural shunting; (C) lumboperitoneal shunting55,56. All seem equally effective, obtaining and

CHAPTER 60

• THE SPINE

Figure 60.21 Syringomyelia. (A) Mid-sagittal T2W MR showing mild cerebellar ectopia and a syrinx within the spinal cord. (B) 1 year later, after decompression at the foramen magnum.

Figure 60.22 Syringomyelia associated with astrocytoma. (A) Sagittal T1W and (B) T2W MR images showing expanded fluid-filled syrinx associated with a lesion of mixed signal intensity at the C4/5 level. Note also the deformities in the vertebral bodies and posterior elements between C3 and C7. (Courtesy of Dr Justin Cross.)

generally maintaining collapse of the syrinx in 70–80 per cent of cases. Unfortunately, however, clinical outcome and extension of cord cavitation on interval images seem to bear no relation to whether the cavity remains collapsed or not54. Although rarely performed nowadays, standard myelography did provide interesting information about this condition. Symmetrical, smooth fusiform expansion of the spinal cord

usually in the cervical region is shown, usually terminating at C2. However, in about 20 per cent of cases, the spinal cord appears normal or small. When small, the cord is usually flattened in the sagittal plane. Rotating the patient from the prone to supine position may cause considerable changes in cord size and shape. On postmyelography CT, contrast medium should enter all intramedullary cavities, though at a variable rate.

1373

1374

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

DEGENERATIVE SPINAL DISEASE Degeneration of some intervertebral discs is identifiable in virtually everyone over the age of 60 years. It is seen most frequently at multiple levels in the cervical region, at L4/5 and L5/S1 and in the upper lumbar and lower thoracic spine. Mobility of involved segments is related to the propensity to develop degenerative change; discs adjacent to fused spinal segments, from whatever cause, are especially vulnerable. Degeneration is closely associated with fissuring of the annulus fibrosus, usually radial57.The central part of the disc becomes less hydrated, loses volume and its structure disintegrates into disorganized collagen. Transverse fissures may develop, which may intermittently fill with gaseous nitrogen during movement and at other times are filled with fluid (Fig. 60.23). The annulus fibrosus normally may bulge diffusely a little beyond the vertebral margins, especially in children. However, bulging of more than 2–3 mm is usually associated with loosening of the concentric layers of the annulus fibrosus, or radial fissures and accelerated dehydration of the nucleus. Protrusions are focal bulges of the annulus fibrosus associated with radial tears. These can occur from any part of the disc circumference. The commonest are posterior and consist of the following types: (A) posterolateral, the commonest (Figs 60.24 and 60.25); (B) duffuse; (C) midline posterior (this term is used instead of ‘central’, which can be confused with nuclear herniations into the vertebral bodies); (D) lateral, which is lateral to the spinal canal (Fig. 60.26), and can involve the dorsal root ganglion in the intervertebral foramen and (E) far lateral, beyond the foramen and often affecting ventral rami. Herniations are fibrocartilaginous masses attached to, but lying outside the annulus fibrosus.These usually extend cranial or caudal to the disc in the anterior epidural space as migratory fragments usually pass one or other side of the midline septum (see Fig. 60.24). In about 10 per cent, the herniation extends through the posterior longitudinal ligament or its lateral membrane58 and, rarely, sequestrated fragments detach and

Figure 60.23 High signal intensity degenerate disc. Sagittal T2W fast spinecho MRI of the lumbar spine showing high signal (black arrow) throughout most of a degenerate disc (between L2 and L3) associated with an irregular posterior protrusion of disc material. There is no bone destruction or reactive change in the adjacent vertebrae. This disc was not infected.

Figure 60.24 Posterolateral disc protrusion with migratory fragment (arrow). Three axial T1W MRI images at and just below the L5/S1 disc showing the large extruded and migratory disc fragment (arrow) compressing the thecal sac and right S1 root.

Figure 60.25 Axial MR image of the cervical spine in a patient with brachalgia at the C4/5 level. There is a left-of-centre posterior disc lesion impinging on the spinal cord and the left C5 roots.

become freely mobile. The dura itself may be penetrated, usually near a root sheath, and the intradural fragment can closely simulate a neurinoma. Reactive changes occur in the vertebral bodies, both in the peripheral disc margin (osteophytes) and in the cancellous bone adjacent to the vertebral end-plates. In the cervical spine, osteophytes on the disc margins commonly involve the spinal canal; associated enlargement of the uncinate processes of the cervical vertebrae encroaches mainly on the intervertebral canals. Osteophytes generally reduce the range of movement and may result in spontaneous fusion. Elsewhere in the spine,

CHAPTER 60

• THE SPINE

Figure 60.26 Lateral posterior disc protrusion. (A) On the axial T1W MR image there is a large lateral L5/S1 protrusion distorting the left dorsal root ganglion and ventral ramus. (B) On the para-sagittal T1W MR image the left L5 foramen is filled with disc material. (C) Note how deceptively and relatively normal the midline sagittal T2W image appears in this patient. Note also how a lateral disc affects the ventral ramus cranial to the affected disc – unlike the usual disc lesion that affects the root that will emerge at the more caudal level.

disc-related osteophytes usually do not involve the spinal canal, even when large. Modic described three types of reactive changes in the cancellous bone adjacent to the vertebral end-plates: (type 1) in the acute stage of disc disease there is invasion of the cancellous spaces by fibrovascular reactive tissue; in time this leads to (type 2) fatty replacement of red marrow; eventually this leads to (type 3) bony sclerosis59. These changes are exquisitely shown by MRI: (A) type 1 changes yield low signal on T1W and high on T2W; (B) type 2 changes yield high signal on T1W and T2W (unless fat suppressed, when they will yield low signal); (C) type 3 changes yield low signal on all sequences. The vertebral end-plates may fracture and displace into the vertebral body, but the dense compact bone is not destroyed. Occasionally the end-plates become very irregular and the degenerative process progresses to a destructive discovertebral lesion, which may simulate many features of infective spondylitis.The key differentiation is the signal intensity of the disc on T2W – in degenerative change it will be low, whereas in infection it should be high.

usually causing anterior slip of the superior on the inferior vertebra60. Disruption of the collagen and elastin fibres of the accessory and capsular ligaments may result in infiltration by fibrovascular tissue. Sometimes focal or even diffuse cartilaginous or osseous metaplasia occurs. Myxomatous degeneration develops in the capsular ligaments of the posterior joints, which sometimes becomes invaded by hypertrophic and often haemorrhagic synovial tissue continuous with that of the joint. All of the above changes combine to create degenerative spinal stenosis (Fig. 60.27).

POSTERIOR JOINTS AND LIGAMENTS Osteoarthritic changes Osteoarthritic changes develop in the facet joints at all levels, usually in close association with concomitant degeneration in the intervertebral disc.These result in hypertrophy of the articular processes and remodeling, and fragmentation of articular surfaces. The capsule thickens along with accessory ligaments. Because the disc space is narrowed, the flaval ligament thickens and buckles. All these changes encroach on the posterolateral aspect of the spinal canal and on the intervertebral foramina; remodelling and fragmentation can result in facet instability,

Figure 60.27 Degenerative spinal canal stenosis. Sagittal (A), and axial (B) T2W MRI of the lumbar spine showing a severe spinal canal stenosis at L4–5 with evidence of compression of the cauda equina, namely obliteration of CSF signal from the thecal sac at the site of compression (white arrowhead) and some redundant coiling of intrathecal spinal roots above.

1375

1376

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Ossification of the posterior longitudinal ligament (OPLL) involves the mid- and lower cervical region in over 90 per cent of cases. Other ligaments are involved in only 7 per cent and a genetic propensity is present, especially in Japanese men61. Diffuse, segmental and mixed forms are recognized, the segmental type needing to be distinguished from posterior osteophytes.The dense bony mass can be adherent to the dura mater, unlike ordinary osteophytes, and surgical removal is often difficult. Other spinal regions can be involved, and extensive degeneration with exuberant anterolateral ossification, sometimes referred to as diffuse idiopathic skeletal hyperostosis (DISH), may also be associated with OPLL61 (Fig. 60.28).

Ossification of ligamentum flavum Small tongues of ossification commonly extend into these ligaments from the superior borders of the laminae. Much more rarely, a hypertrophic bony mass or masses occur that encroach on the spinal canal; this has been reported most often in or near the thoracolumbar junction, but can occur anywhere62.

Synovial cysts These are uncommon lesions attached to the capsule of the posterior joints. Although they may consist mainly of degenerating synovium, and contain fluid that communicates with joint cavity, frequently they are cartilaginous or myxomatous, and solid rather than cystic. They have been confused with migratory disc fragments from which they should be distinguished by their attachment to the joints, and organized haematomas, because they may contain blood products. They usually occur only in the lower lumbar, or mid- and lower cervical regions63 (Fig. 60.29).

Retro-odontal pseudotumour

Figure 60.28 Ossification of the posterior longitudinal ligament. Sagittal (A) and axial (B), T2W MRIs of the cervical spine showing mild spinal cord compression by the thickened anteriorly located posterior longitudinal ligament (white arrowheads) within the spinal canal (black arrowhead).

The transverse ligament of the atlas and associated ligaments are commonly thickened in the ageing population, and often contain calcification. Rarely, the thickening has been so marked as to simulate a lesion such as meningioma. The masses usually consist of fragmented ligament and degenerate fibrocartilage, and they have also been confused with migratory disc material64.

Figure 60.29 A–C Synovial cyst: an unusual consequence of osteoarthritis of the intervertebral facet joints. Axial (A) and (B) sagittal T2W MR: this lesion of the right L4/5 facet is displacing the flaval ligament and compressing the thecal sac. There is often secondary spinal stenosis at the affected level, as in this case. (C) CT of another patient. Here the lesion arises from a degenerate left-sided facet joint. They are sometimes misinterpreted as an intraspinal tumour, but involvement of the joint is characteristic.

CHAPTER 60

INVOLVEMENT OF NEURAL STRUCTURES All these degenerative processes can result in damage to the spinal cord and spinal roots, due mainly or entirely to mechanical compression. However, the severity of compression is not related linearly to the degree of damage, nor to its clinical effects. Indeed osteophytes in the cervical region are sufficient to cause up to moderate spinal cord compression in about 30 per cent of asymptomatic persons over 50 years of age65,66 and disc protrusions in the lumbar spine sufficient to interfere with spinal nerve roots in nearly 40 per cent of asymptomatic persons67.

Spinal cord In the cervical region, compression is usually intermittent, or intermittently accentuated by neck movement. Cord substance is relatively inelastic, and retains the impression of deforming agents when not in direct contact68. Most often, damage is sustained only when the sagittal diameter of the cord is reduced by more than 50 per cent (Fig. 60.30); pathologically this damage consists of central necrosis involving mainly grey matter, and demyelination associated with axonal damage located mainly deep within the posterior and lateral columns. The mechanism is likely to be either shearing injury to axons in white matter, and to cord microvasculature in the deeper grey matter, analogous to closed head injury69,70, or to compression-induced intermittent ischaemic damage71. It has a distinct propensity to be progressive, and shearing stresses in particular are generated from anterior deforming agents alone, such as a focal angular kyphosis without the necessity for an anteroposterior squeeze72. In the thoracic region, far greater compression is tolerated without the cord being damaged, presumably because of reduced mobility of this part of the spine. The cord frequently becomes focally moulded around the usually calcified fibrocartilaginous masses, which can occupy up to 60 per cent of the spinal canal and be associated with no clinical abnormality73.

Figure 60.30 Cervical spondylotic myelopathy with myelomalacia. Sagittal (A) and axial (B), T2W MR images showing only moderate compression of the spinal cord at C3–4 level, and focal increased signal in the cord substance indicating that damage has occurred. On axial images this often has the appearance of ‘snake eyes’ (black arrowheads) within the spinal cord.

• THE SPINE

Spinal roots In the cervical region, the spinal roots are compressed usually by osteophytes or fibrocartilaginous masses within or close to the entrance of the intervertebral canals. Clinical correlation is close only when these are also large enough to compress the spinal cord (see Fig. 60.30). In the lumbar region, roots usually are compressed by posterolateral disc protrusions or migrating migratory fragments in the anterior epidural space; the root that crosses the abnormal disc to reach the next inferior intervertebral foramen is usually involved, lying in the lateral extremity of the spinal canal under cover of the articular facet and within the thecal sac (except for S1 roots, which usually leave the dura cranial to the L5/S1 disc).The rarer far lateral posterior protrusions compress the dorsal root ganglion ventral ramus exiting cranial to the disc within the intervertebral foramen (see Fig. 60.26). Osteo-arthritic changes in the posterior joints encroach on the lateral part of the spinal canal, and usually displace the dura and its contained root towards the centre of the canal, but occasionally the most laterally lying roots are sufficiently tethered by their sheaths to become entrapped (lateral canal entrapment). The intervertebral foramen becomes distorted and narrowed, and when severe and especially when subluxation is also present, the dorsal root ganglion may be compressed. The cauda equina as a whole may be compressed by large midline disc protrusions, or by posterior and bilateral posterolateral hypertrophic changes in bone and ligaments, usually in combination. Spinal stenosis is usually significant when there is room for only roots and not CSF, and frequently this is accompanied by redundant tortuosity of roots as a consequence of focal entrapment and stretching of these roots, which have a long intradural course (Figs 60.16, 60.19). It is only when these features are present that association with clinical cauda equina syndrome is consistently seen.

Natural history The clinical conditions associated with degenerative lesions of the spine are: cervical spondylotic myelopathy and radiculopathy, sciatica and regional pain syndromes. Whilst relentless progression to severe disability occasionally occurs, much more often clinical features improve or resolve spontaneously in all three conditions. It is now accepted that possibly 60–70 per cent of intraspinal fibrocartilaginous masses of discogenic origin diminish in size or disappear spontaneously over a few weeks or months74, although other much rarer conditions that can simulate migratory disc fragments, such as epidural varices and haematomas, may contribute to these figures. It has been claimed that cervical osteophytes regress after successful interbody fusion, but this has been refuted75. Reduction in cord cross-sectional area by more than 50–60 per cent (down to around 40 mm2) is associated with poor operative outcome in cervical spondylotic myelopathy, with or without operative intervention76.

IMAGING Many of the processes discussed in the preceding section are shown on plain radiographs, namely sclerosis, osteophytes, disc

1377

1378

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

space narrowing, calcified or ossified disc material and ligaments and gaseous nitrogen. A minimum midsagittal diameter of the cervical canal of less than 10 mm (tube-film distance around 2 m) indicates that cord compression is probably present. Alignment abnormalities are well shown on plain radiographs, which also show reducibility of subluxations determined by flexion and extension views. On unenhanced CT in the lumbar region, annular bulges and protrusions, nuclear herniations and migratory fragments, enlarged degenerate ligaments and synovial cysts are usually well shown.These materials usually attenuate X-rays similar to disc substance. Calcification is uncommon; bubbles of nitrogen escaping from the disc via a tear in the annulus and entering the anterior epidural space are well shown only by CT. Spinal canal narrowing by osteoarthritis and soft-tissue thickening is well documented by CT, and usually it is possible to tell which roots are entrapped laterally77.The cross-sectional area of the canal can be measured electronically; wide window settings are needed and only measurements of about 100 mm2 or less show a consistent association with clinical cauda equina entrapment13. In the cervical and thoracic regions, plain CT is of less value and generally is not recommended. However, the anatomy of osteophytes is well shown, and many thoracic disc protrusions are calcified. MRI is the optimal investigation, because effects on neural structures are shown best. The MR signal returned from degenerate discs is usually lower than from healthy discs and best appreciated on T2W images. Occasionally, signal is paradoxically higher than from healthy discs on T1W images, perhaps due to calcium precipitation, and signal may also be higher on T2W images due to the presence of fluid-filled clefts (see Fig. 60.23). The latter appearance needs to be distinguished from infective discitis (see later), especially when it is associated with reactive changes in the adjacent vertebrae; in degeneration, high signal is not uniform throughout the whole disc, and the dark band of cortical bone of the vertebral end-plates is intact though not necessarily undistorted. Tears in the external layers of the annulus fibrosus may be visible as high signal foci on T2W images. Signal from degenerate fibrocartilage within the spinal canal is quite variable on T2W images, and it is common for migratory fragments to be conspicuously brighter than the nucleus of the disc of origin. Migratory fragments can be simulated by the much rarer epidural varices or haematoma78. Reactive changes in the cancellous bone of the adjacent vertebral bodies also return variable signal reflecting their composition: fibrovascular infiltrates (dark on T1W, bright on T2W images), fatty replacement of red marrow (bright on T1, dark or bright on T2), sclerosis (dark on all images)59. Osteophytes show a similar variability in internal structure, although mostly they are sclerotic. Compression of the spinal cord is well shown. This generally is assessed best on axial imaging, but cord flattening may be artefactually exaggerated on T2W and gradient-recalled echo images. Distortion of cord shape may be due to compression alone, or reflect underlying structural damage; when congru-

ous with the shape of the deforming agent, it is probably due to compression and will retain the impression of an impinging agent, even when contact is removed68. Structural damage is usually reflected by signal change within the substance of the spinal cord, and when present, clinical myelopathy usually is also present (Fig. 60.30). Signal change usually is focal and occurs at or slightly caudal to the site of compression. When the distribution can be determined on axial images, it usually seems to involve mainly central areas70, and often has the appearance of bilateral lesions, sometimes likened to snakes’ or cats’ eyes; occasionally these appearances extend over many cord segments. These changes are detected most sensitively by T2W images, and that they do not always indicate permanent damage is demonstrated by their frequent disappearance after decompressive surgery with good outcome; however they usually persist when clinical recovery is minimal or absent79. The spinal nerves are directly shown, and the dorsal root ganglion can often be distinguished from the compressing agent even when entrapped. The ganglia are affected by far lateral disc protrusions, and a ganglion flattened by compression from below may seem enlarged on axial images. Abnormal enhancement has been observed in compressed roots after IV contrast medium, usually focal and mainly extradural, but occasionally extending intradurally for several centimetres80. Myelography is still occasionally requested when multi-level disease is present or MR impossible. Disc protrusions present as indentations of the anterior surface of the thecal sac opposite the disc space; root sheaths of the spinal nerves usually fill normally but are deviated and may appear flattened. Disc herniations are generally larger, and the root sheath exiting from the intervertebral foramen below usually is obliterated. The indentation of the thecal sac is nearly always maximal opposite the level of the disc. Most far lateral posterior disc protrusions will not be shown; these constitute around 10 per cent of disc lesions causing entrapment neuropathies. In lateral recess stenosis the theca usually is displaced medially away from the lateral extremity of the spinal canal, resulting in a constricted or wasted appearance of the thecal sac. Postmyelography CT in the cervical region depicts most accurately the degree of compression of the spinal cord, which can be both under- and overestimated by myelography alone.

THE POSTOPERATIVE SPINE Lumbar disc surgery involves part or complete laminectomy for access, partial facetectomy to free the lateral recess and removal of accessible disc material from spinal canal and disc space, in varying combinations. With improved preoperative imaging, microdiscectomy procedures are increasingly used; these require smaller and smaller laminectomies. However all surgery results in some epidural granulations that eventually mature into fibrous tissue. The importance of this as a cause of recurrent symptoms after operation has been greatly diminished by imaging studies in which both the prevalence and severity of these processes have been shown to be entirely similar in patients who are pain-free after operation81. However, it is still useful to distinguish these reactive processes from recurrent or

CHAPTER 60

residual disc material, the presence of which continues to be an indication for re-operation82. Pseudo-meningoceles sometimes occur, implying that dura was breached during surgery; they usually are not of clinical relevance, but may need to be distinguished from abscess in which communication with the thecal space is not shown. Lumbar disc surgery had long been considered a common cause of lumbosacral adhesive arachnoiditis, but this was largely due to the pre-operative myelography with oil-based contrast medium iophendylate (Myodil, Pantopaque)83. In the cervical region the operation most often performed is anterior spinal fusion, usually combined with a discectomy and removal of osteophytes. Usually either a dowel-shaped (Cloward) or disc-shaped (Smith-Robinson) bone graft from the iliac crest, or an osteoconductive polymer is hammered into the disc space. Sometimes anterior plates or cases are used. The graft should be flush with the anterior surface of the spinal column, the disc space should not be too distracted and the normal spinal curvature should be maintained. In multilevel disease, an extensive laminectomy may be performed, but because this often results in a progressive cervical kyphosis, the operation of laminoplasty has been devised, which widens the sagittal diameter of the spinal canal but removes little or no bone. The latter is often preferred in OPLL, because adequate decompression from the anterior approach can be difficult to achieve in this condition. Plain radiographs may contribute little to the evaluation of the postoperative lumbar spine and, following microdiscectomy, even the operative level can be difficult to identify. However they remain useful in the cervical region to evaluate the stability of a spinal fusion (Fig. 60.31), effectiveness of laminoplasty, or severity of postlaminectomy kyphosis. MRI is the optimal investigation for evaluating the postoperative lumbar spine (Figs 60.32 and 60.33). Recurrent or residual disc material is distinguished reliably from epidural scar and granuloma on post-gadolinium-enhanced images by lack of enhancement82, though disc material is often surrounded by enhancing matter. Enhancement of scar diminishes over 2 years, but generally persists for many years82. Arachnoiditis may be confined to the

• THE SPINE

operation site or be more generalized, and all roots across the involved regions are usually involved83. Rarely one or more roots enhance even intradurally, and often this is the symptomatic and formerly compressed root80. In the cervical region, the discectomy may simulate discovertebral spondylitis, an appearance that may persist for many months. Persistent compression and degenerative changes in the spinal cord may be shown, and sometimes an alternative diagnosis is established by this study, such as multiple sclerosis. Signal change in damaged spinal cord usually regresses when functional outcome is good, but it generally persists when it is poor84.

SPINAL TUMOURS Spinal tumours occur in extradural, intradural extramedullary and intramedullary compartments. For each compartment, the pattern of radiological abnormality is usually distinctive. A list of the more common lesions occurring in each compartment is shown in Table 60.1.

EXTRADURAL TUMOURS The most common extradural tumours are metastases. These usually involve the vertebral bodies and neural arches, but malignant infiltration may spread widely in the epidural space without local bony involvement. Other disseminated malignancies such as myeloma and lymphoma are indistinguishable. Primary bone tumours are less common. Those most often responsible for spinal cord compression are aneurysmal bone cyst, benign osteoblastoma, chordoma and giant cell tumour, the latter occurring mainly in the sacrum, but also in the upper thoracic or upper cervical regions (see Figs 60.34–60.37). Plain radiographs often show evidence of focal bone destruction, and sometimes vertebral collapse; paraspinal masses may be visible, especially with myeloma and extradural neurogenic tumours. Plain CT is far more sensitive at identifying bone destruction, which is not surprising because between 50 per cent and 60 per cent of vertebral body bone

Figure 60.31 Cloward’s anterior spinal fusion: postoperative radiographs in three cases. (A) Satisfactory appearances. (B) The vertebral bodies adjacent to the fusion have partially collapsed, resulting in an angular kyphosis. (C) The bone graft has been extruded anteriorly and the upper vertebra has slipped forwards; the spinal canal remains compromised.

1379

1380

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 60.1 ANATOMICAL LOCATION OF SPINAL LESIONS Extradural Benign Disc prolapse Haematoma Abscess Neurofibroma Osteochondroma Dermoid/epidermoid Vertebral body tumours: Haemangioma Osteoclastoma

Figure 60.32 Epidural scar and residual/recurrent disc protrusion. (A) Axial T1W MRI just below (above) and through (below) the L4/5 disc, showing a large epidural mass (black arrow) on the left side. (B) Images at similar levels made after IV gadolinium and using fat presaturation, showing marked enhancement of most of the epidural mass, but also a central non-enhancing region in contact with the discal margin (arrow). At operation, recurrent disc material was found at this site, embedded in dense fibrous tissue.

Aneurysmal bone cyst Paget’s disease

Malignant Metastases Lymphoma Myeloma Sarcoma Chordoma

Figure 60.33 Extraspinal wound abscess and infected pseudo-meningocele. Axial T1W image at L4/5 made with fat presaturation and IV gadolinium showing a low signal cavity surrounded by a very thick white wall of granulation tissue. Operation revealed a cavity containing fluid infected by Staphylococcus aureus. These appearances can be mimicked by sterile postoperative pseudomeningoceles.

Intradural/extramedullary Neurofibroma Meningioma Dermoid/epidermoid Lipoma also medullary Ependymoma Metastases in CSF: Medulloblastoma Ependymoma Melanoma Carcinoma

Intramedullary Tumours: Ependymoma

mass needs to be destroyed before it becomes evident on plain X-rays. This is particularly true of lesions involving the sacrum. However, the extent of the involvement of the spinal canal is often not shown adequately by plain CT. MRI clearly demonstrates bone destruction and malignant infiltration, intra- and extraspinal masses and spinal cord compression. The entire spinal canal should be examined to reveal the full extent of the disease, and phased array coils are optimal. Although myelography is rarely performed nowadays, the most characteristic feature of an extradural lesion is displacement of the thecal sac away from the bony walls of the spinal canal. When major resections are planned, angiography may be requested to identify the position of the major radiculomedullary arteries. Percutaneous biopsy is often necessary when the diagnosis is not known or there is doubt about tumour transformation (e.g. in lymphoma). As in other situations, it may be performed under fluoroscopic guidance, but CT provides optimal guidance, despite the advances in MRI intervention.

Astrocytoma Glioblastoma Developmental tumours Haemangioblastoma Syringomyelia Myelitis Abscess/granuloma Haematomyelia

INTRADURAL EXTRAMEDULLARY TUMOURS Neurinomas and meningiomas are by far the commonest lesions in this intradural extramedullary location. Significant extradural components are present in about 7 per cent of meningiomas and 30 per cent of neurinomas. Over 80 per cent of meningiomas occur in the thoracic region in middle-

CHAPTER 60

• THE SPINE

Figure 60.34 Chordoma. Axial and sagittal MR showing the typical soft-tissue mass and bone destruction. The cord is markedly displaced to the left. (Courtesy of Dr Justin Cross.)

Figure 60.35 Extraspinal neurofibroma. Axial T2W gradient-recalled echo MRI showing a huge neurofibroma in the left posterior triangle without spinal involvement. (P = posterior.)

Figure 60.36 Osteoclastoma (giant cell tumour) of C2. Sagittal T1W MRIs made after IV gadolinium. C2 is replaced by tumour and the large extraosseous mass is compressing the spinal cord. Note similarity in appearances to the chordoma in Figure 60.34.

aged women and are very rare in the lumbar spine, whereas neurinomas occur at any level, at almost any age and are approximately equal in men and women85. Tumoural calcification may occur in both, but marked calcification is uncommon. Meningiomas may be associated with hyperostosis, but this is seldom conspicuous. Sometimes only diffuse thickening of the spinal roots occurs in neurofibromatosis. However, diffuse enlargement of the spinal roots may also occur in nonneoplastic processes, such as some of the hereditary sensory motor neuropathies. Malignant tumours may disseminate in the subarachnoid space. Cerebral involvement is commonly present as well, and spinal lesions are most often found in the lumbar theca, but can occur anywhere.They are frequently multiple and in some cases there is diffuse pial spread with encasement of the spinal cord or cauda equina. Plain radiographs may reveal expansion of the spinal canal or of one or more intervertebral foramina; this is shown in about 30 per cent of neurinomas, and very rarely with meningiomas. An associated paravertebral mass is far more suggestive of a neurinoma or other neurogenic tumour (see Fig. 60.37). Hyperostosis is uncommon with meningiomas of the spine because the bone is rarely infiltrated. Plain CT will demonstrate bone erosion, sclerosis and extradural extension is good, but purely intradural lesions may not be shown. IV contrast enhancement may be helpful if the level is known. Heavy calcification is rare in meningiomas and neurofibromas; a heavily calcified intraspinal mass is usually found to be extruded disc material and is often seen to be clearly extradural. MRI elegantly demonstrates the intraspinal as well as the extraspinal extent, spinal cord compression and displacement. The signal from spinal meningiomas is usually similar to that from the spinal cord in both T1W and T2W images, whereas that from neurinomas is usually conspicuously higher on T2W images. Even small intradural

1381

1382

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 60.37 Partly intradural and partly extradural right C3 schwannoma. Axial (A) and coronal (B) T1W MRIs of the cervical spine made after the injection of IV gadolinium showing the large lobular schwannoma extending into the spinal canal and compressing the spinal cord (arrow).

lesions are well shown by MRI on the high-resolution MRI techniques now available. IV contrast enhancement may be necessary to demonstrate diffuse pial spread in metastatic disease. However, CSF cytology remains considerably more sensitive in detecting diffuse leptomeningeal involvement86. Myelography, rarely performed nowadays, showed an intradural extramedullary lesion as a rounded, smooth contour mass, expanding rather than compressing the thecal sac and displacing the spinal cord. A plane of cleavage was usually visible between the tumour and the spinal cord.

INTRAMEDULLARY TUMOURS Most of these are gliomas; ependymomas (Fig. 60.38) and astrocytomas occur with about equal frequency in the spinal cord, but ependymomas greatly predominate in the filum terminale, especially in children, where they are usually of the myxopapillary type87. Rarely, ependymomas may arise from the extradural part of the filum terminale and present as a destructive lesion of the sacrum, or a presacral mass, which is prone to metastasize especially to lung. Spinal capillary haemangioblastomas nearly always involve the posterior columns of the spinal cord and abut against a pial surface88. Most are solitary, but up to 30 per cent are associated with similar lesions elsewhere in the central nervous system. Extramedullary and even extradural haemangioblastomas are encountered occasionally in the spine. Metastases may occur within the substance of the spinal cord, and other rare intramedullary neoplasms including primary lymphoma. About 70 per cent of intramedullary tumours are associated with cysts, and three types are described: (A) intratumoral cysts, in which the wall consists of tumour; (B) peritumoral or capping cysts, in which cone-shaped glial-lined cavities extend above and below the tumour for a limited number of spinal segments; (C) syringomyelia, in which extensive cavitation of the spinal cord is present and indistinguishable from other forms of syringomyelia in regions not involved by tumour.

Figure 60.38 Ependymoma of the filum terminate and conus medullaris. Sagittal T2W (A) and (B) T1W post gadolinium-enhanced MRIs of the lumbar spine showing an expansile enhancing intraspinal mass and central signal change in the spinal cord above.

The latter cystic form occurs in approaching 50 per cent of all intramedullary tumours including metastases, but is probably most frequent with haemangioblastoma87,89. Plain radiographs only show expansion of the spinal canal in about 10 per cent of adults, 30 per cent of childhood cases overall, and is most frequent with giant myxopapillary ependymoma of the cauda equina90. MRI is the preferred investigation and intramedullary tumours and associated cysts are usually shown well. Signal from the tumour may be similar to cord substance, though sometimes it is lower on T1W images and frequently somewhat higher on T2W images. They may appear sharply demarcated from normal cord tissue and sometimes seem enclosed in a hypointense capsule.Astrocytoma and ependymoma cannot be distinguished reliably, nor can they be distinguished reliably from many inflammatory processes. Tumour-associated cysts are well-circumscribed tubular or cone-shaped structures of homogeneous signal intensity, low on T1W and usually high on T2W images, but about 15 per cent of apparently cystic regions are found to be solid at surgery91. Syringomyelia secondary to intramedullary tumour is suggested by enlargement or signal change in noncystic regions of the cord and by the appearances after administration of IV gadolinium (Gd-DTPA). Unlike the brain, the great majority of spinal cord gliomas enhance at least partially87. IV contrast enhancement will also help to identify solid tumours, including metastases. The spinal cord appears enlarged, the enlargement usually smooth and uniform as in syringomyelia; however the site of enlargement may be unusual for syringomyelia, and any lobulation, or focal or eccentric expansion, are valuable clues. Spinal angiography often provides definitive diagnosis of spinal haemangioblastoma, but usually is not a necessary prelimi-

CHAPTER 60

nary to operative treatment. The appearance is characteristic, consisting of a usually small, relatively homogeneous, wellcircumscribed dense capillary blush; the one or two arteries commonly supplying the lesion are usually slightly enlarged, and enlarged or normal-sized draining veins opacify only a little earlier than normal. Intra-operative US can distinguish cysts from solid tumour, cysts appearing as circumscribed anechoic areas. US may also assist in identifying the tumour boundary but it is not infallible92.

TRAUMA Clinical aspects Acute post-traumatic myelopathy usually results from burst fractures and fracture-dislocations. If loss of cord function is complete, and remains so for 24 h, functional recovery is unlikely. Management is frequently conservative, and imaging other than CT for diagnosis usually is not indicated. When fracture is absent, pre-existing abnormalities are often present, such as cervical spondylosis or instability such as at the atlantoaxial junction. Also acute, mainly soft-tissue injury occasionally may result in sudden severe cord compression from acute disc herniation or epidural haematoma. In these situations the myelopathy is more often incomplete, and prospects for recovery can be much better. Therefore, early imaging of the spinal canal may be indicated. Many workers also believe that early stabilization and correction in spinal deformity is mandatory to optimize conditions for recovery. In delayed post-traumatic myelopathy, the neurological impairment appears only a few hours after injury, often no explanation is found, but imaging is indicated to exclude compressive lesions such as epidural haematoma or acute disc herniations.When it appears months or years after injury, it can be due to delayed spinal instability, which recent work has shown is most likely to occur with fracture-dislocations that might have been reduced by traction in the early postinjury period; however evidence also is now substantial that post-traumatic deformity, such as angular kyphoses at the fracture site, may result in progressive cord damage manifesting clinically only after many years72. Progressive post-traumatic myelopathy results in worsening of existing disability or ascending functional loss. In such cases, the spinal cord usually is extensively damaged well beyond the site of injury. The entire spinal cord may be involved93,94. This damage manifests as diffuse atrophy with necrosis, cell loss and gliosis, and most frequently the cord is extensively cavitated. Imaging may be indicated because some workers believe it advisable to drain intramedullary cysts or correct spinal deformities and divide adhesions in such cases, in the hope that this may arrest the process. Post-traumatic pseudomeningoceles have occasionally caused a delayed compression myelopathy or neuropathy, but usually they do not, even when large. Acute brachial and lumbar plexus injuries resulting in a flaccid monoplegia in the cervical region are commonly due to avulsion of the spinal roots from the cord, and in the lumbar

• THE SPINE

region to tearing of the nerves at the level of the root sheaths. This is usually associated with rupture of the root sheaths of the affected spinal nerves resulting in pseudo-meningoceles, or less commonly simply in irregular occlusion of the sheaths. Associated fractures may be present, especially in lumbar plexus avulsion injuries, which are seen mainly in children.

IMAGING CT is superior to plain radiographs at detecting fractures and defining the degree of distortion of the spinal canal (Fig. 60.39). More accurate depiction of the true extent of injury will better enable instability to be predicted: fractures involving only the anterior part of the vertebral body are stable, those involving the posterior part of the body and neural arches are potentially unstable. Thin contiguous slices as on modern multidetector systems will demonstrate subtle fractures and provide multiplanar reformatted images, which are essential in assessing vertebral alignment, facet joint integrity and the craniovertebral junction. With mainly ligamentous injuries, such as complete disc rupture or avulsion of the transverse ligament of the atlas, carefully performed CT usually will reveal avulsion fractures. Unlike CT, MRI accurately demonstrates the state of the spinal cord. The spinal cord is abnormal in most patients with acute post-traumatic myelopathy, whether or not a bony injury has occurred95,96. The abnormality may consist of: 1 Diffuse signal increase on T2W images, usually at the site of injury or only for one or two segments beyond 2 Cord swelling, not always present and usually only slight 3 A circumscribed area of low signal within a more extensive area of high signal on T2W images, associated with focal cord swelling, and probably representing intramedullary haematoma. High signal on T1W images may appear, but is infrequent. It is notable that although cord contusions are always haemorrhagic pathologically, this is evident on MRI in only 50 per

Figure 60.39 Burst fracture. Axial CT through L3 showing a comminuted fracture of the body of L3 (arrows) and a large retropulsed square bone fragment (arrowhead) occupying most of the spinal canal. Flaccid paraplegia.

1383

1384

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

cent or less. Major fractures, abnormalities of vertebral alignment and the state of the spinal canal are usually well shown. Acute disc herniations, and other lesions that may acutely compress the spinal cord such as bone fragments or epidural haematoma, also should be shown. However, MRI shows that the spinal cord is not usually significantly compressed after the injury. Animal models that have established the extent of signal change on MRI in the acutely damaged spinal cord is related to the severity of injury. Clinical studies suggest that the extent of signal change has prognostic significance also, and mild or transient loss of function after injury usually is not accompanied by any signal change in the cord. Evidence of haematomyelia has been indicative of poor prognosis. Progression from acute injury to a localized cystic myelopathy has been followed on serial imaging97, and in most cases, this does not lead to further functional loss. Cases with progressive myelopathy usually show extensive signal change throughout the cord on both T1W and T2W images. Cavities generally resemble syringomyelia in such cases, but often are multiple or multilocular. Spinal cord injuries in children have been noted to differ somewhat from adults: children may develop extensive signal change in the cord with minor, remote or no spinal fracture, and any signal change usually is followed by persistent functional loss98. Myelography rarely needs to be performed nowadays, but it may be needed if MR is equivocal for brachial plexus avulsion injuries. The most reliable sign is failure to visualize the intradural rootlets of the avulsed spinal nerves in the cervical region when uninvolved rootlets are clearly visible.

soft-tissue masses contribute to spinal cord compression, and in as many as 30 per cent of cases with myelopathy the subluxation is not reducible99.

Ankylosing spondylitis and other seronegative arthritides Ankylosing spondylitis characteristically involves mainly the spine, but the other syndromes, which include Reiter’s, psoriatic, intestinal or lupus arthritis, rarely do. Even in ankylosing spondylitis, neurological complications are uncommon. They may arise from cord or root compression caused by fractures, atlanto-axial subluxation, discovertebral destructive lesions that are either inflammatory or degenerative in type, and occasionally from focal degenerative or granulomatous soft-tissue masses. The cauda equina syndrome of ankylosing spondylitis probably is an inflammatory radiculopathy, and associations with adhesions between roots and the walls of the thecal sac, and with dural ectasia have been described in some cases. Dural ectasia is rare, and may be either diffuse or saccular, the latter sometimes causing marked focal bony erosions invaginating into neural arches or vertebral bodies37.

Imaging

Erosive arthropathies Rheumatoid arthritis

In atlanto-axial subluxation due to rheumatoid arthritis, the decision to operate, and which operative strategy to employ, are influenced by: 1 Presence and severity of the spinal cord involvement 2 Whether the cord compression would be adequately diminished by regression of the inflammatory soft-tissue masses that often follow arthrodesis 3 Whether the subluxation can be reduced, and the prospect of maintaining that reduction, which in turn is influenced by: a. The integrity of the neural arches, especially of C1 b. The integrity of the lateral masses of C1 and C2.

The prevalence of radiological evidence of spine involvement usually has been found to be up to about 15 per cent, over 70 per cent of these cases involving the craniovertebral junction. However, autopsy studies have indicated an incidence of atlanto-axial involvement in over 80 per cent of rheumatoid arthritics. Subaxial subluxations, usually of multiple levels, are a major feature in 10 per cent. Only 2–6 per cent of cases develop a clinical myelopathy, which almost invariably is due to spinal cord compression, caused by the subluxations (especially the vertical and subaxial). In up to 40 per cent, granulomatous

Plain CT with sagittal reconstructions is better than plain radiographs at demonstrating erosions and the integrity of structural elements. MRI shows all these features, including the soft-tissue masses and state of the spinal cord. Highresolution multiplanar MRI also usually demonstrates the bone adequately, but in some cases the addition of CT may be desirable. Subluxation is best demonstrated by plain radiographs and flexion/extension views are used as a pre-operative investigation.

INFLAMMATORY ARTHROPATHIES

INFECTIVE DISORDERS OF THE SPINE VERTEBRAL OSTEOMYELITIS This is usually based on a destructive discovertebral lesion. However central and subligamentous types are also occasionally encountered involving only the vertebral bodies. Rarely the neural arches are affected. Neurological involvement results from intraspinal extension of infection or more rarely instability, and occurs in less than 1 per cent of cases caused by pyogenic

organisms, but in up to 40 per cent of cases caused by nonpyogenic organisms, of which the commonest is tuberculosis100.

EPIDURAL ABSCESS This may occur from haematogenous dissemination, or has spread from an infected disc or posterior joint, involvement of which may only be noted secondarily or in retrospect. Lesions

CHAPTER 60

are commonly extensive, particularly in children. Acute pain and early onset of paraplegia are characteristic.

PARAVERTEBRAL INFECTIONS Various types of infections near the spine may be associated with subluxations. One of the most notable is atlanto-axial subluxation in the presence of retropharyngeal sepsis.

Imaging MRI is the preferred technique and allows a firm diagnosis of pyogenic discovertebral osteomyelitis at least 2–3 weeks earlier than plain radiographs or CT. The characteristic early signs are: low signal throughout the disc and in adjacent parts of the vertebral bodies on T1W images (Fig. 60.40), and corresponding high signal on T2W images; and thinning, fragmentation and eventual loss of the dark line of the vertebral end-plates above and below the disc. This may need to be distinguished from severe disc degeneration, with high signal clefts in the disc and marked reactive changes in the adjacent vertebrae with irregular end-plates; in degeneration the changes are less uniform, both in disc and vertebral bodies, and actual bone destruction should be absent. Other features, which may be seen in infection, are perispinal extension including paraspinal abscesses (Fig. 60.41) and involvement of the anterior epidural space. Nondiscogenic forms of infective spondylitis present as localized processes, which can be difficult to distinguish from

• THE SPINE

neoplasia, especially metastases. Multiple adjacent vertebrae may be involved in the subligamentous type, which would be unusual for metastases. IV Gd-DTPA results in diffuse enhancement in areas of active infection, often surrounding unenhancing foci of pus. CT characteristically shows punched-out erosions of bone adjacent to the involved disc, giving a moth-eaten appearance, and the lesions may be associated with usually small dense sequestra. Sclerosis is often prominent. Progressive loss of disc space, and thinning progressing to loss of vertebral endplates, followed by bone loss, wedging of adjacent vertebrae, sclerosis and regional subluxations or kyphosis, are the plain radiographic features of infective discovertebral osteomyelitis.

SPINAL MENINGITIS (ARACHNOIDITIS) Inflammation in the subarachnoid space may lead to organizing exudates and permanent intradural adhesions101. The commonest causes of arachnoiditis are iatrogenic, in particular due to previous myelography using iophendylate (Myodil), but this is now becoming a rarity. It usually involves the caudal sac, rarely ascending above the L3/4 disc102. It also may follow other intrathecal injections such as some of the earlier ionic water-soluble agents, steroids or anaesthetic agents. Lumbar disc surgery itself is rarely a cause83. Arachnoiditis also may follow accidental trauma, spinal subarachnoid haemorrhage and is rarely encountered with

Figure 60.40 Discovertebral osteomyelitis. The L4/5 disc space is narrowed and on sagittal T2W MRI there is high signal intensity (pus) within and abnormal signal intensity in the adjacent vertebrae and destruction of the superior surface of L5. On T1W sagittal MR (B) it is difficult to distinguish the infected disc from the infected adjacent vertebrae, both of which are of intermediate signal intensity. On T1W sagittal MR after IV gadolinium (C), there is no enhancement of the pus within the infected disc but the margins of the disc and the infected adjacent vertebrae enhance avidly.

1385

1386

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 60.41 Tuberculosis and osteomyelitis. (A) Coronal T1W and (B) sagittal T2W MR images show a huge prevertebral collection and abnormal vertebrae and abnormal material in the epidural space surrounding and narrowing the cord. There is also an extensive retropharyngeal collection. (Courtesy of Dr Justin Cross.)

intraspinal tumours. Intradural infections, which include spinal tuberculosis and a variety of fungal and parasitic (especially cysticercosis) agents, characteristically cause severe generalized arachnoiditis, as can spinal sarcoidosis. Some cases are unexplained. On rare occasions, the organized exudates become calcified and even ossified (arachnoiditis ossificans). Myelomalacia and syringomyelia often develop in extensive cases.

Imaging MRI is the optimal investigation for demonstrating arachnoiditis and its complications (Fig. 60.42). The signs are tapering or obstruction of the lower end of the subarachnoid space, central clumping and/or peripheral adhesion of roots of the cauda equina, the latter often resulting in the appearance of an empty thecal sac though the walls of the sac commonly appear thickened. Loculation and deformity of the subarachnoid spaces, and irregular deformity of the

spinal cord may be shown with central signal change in severe cases. Care may need to be exercised to distinguish crowding of roots due to intradural adhesions from that due to extradural compression, such as from degenerative spinal disease. Some infectious agents cause diffuse intradural enhancement (Fig. 60.43).

MYELITIS Myelitis usually is due to demyelinating disease, especially multiple sclerosis. About 70 per cent of such patients also have brain lesions. Acute transverse myelitis may be the presentation of other demyelinating diseases such as neuromyelitis optica and acute disseminated encephalomyelitis (ADEM). Sarcoidosis and infections of the spinal cord are uncommon, but a wide range of organisms are documented, which include many of the enteroviruses and recently the T-cell lymphotrophic viruses, especially HTLV-1.

Figure 60.42 Spinal adhesive arachnoiditis. Axial high resolution (fast spin-echo) T2W MRIs of the lumbar spine showing the three main diagnostic features of this condition on MRI. (A) Central clumping of nerve roots. (B) Peripheral adhesion of roots leaving a clear central subarchnoid space. (C) Adhesion of the margins of the thecal sac near the point of exit of the root sheaths (arrows). Compare this with (D), which is normal; here rootlets are clearly seen as they enter the spinal root sheaths on each side.

CHAPTER 60

Figure 60.43 Spinal meningitis, due to Lyme disease. Sagittal T1W MRI of the lumbar spine after IV gadolinium showing diffuse enhancement of the outer surface of the cord and spinal roots.

The pathological picture is usually of inflammatory demyelination that involves several segments. In severe cases, this progresses to necrosis and cavitation. Necrosis is more common with infective causes. Granulomas or abscesses may develop with bacterial and fungal agents. Pyogenic abscesses (pyomyelia) usually arise as complications of spinal dysraphism, such as dorsal dermal sinus. Parasites such as cysticercosis may cause arachnoiditis, but intramedullary cysts sometimes occur, and may be associated with syringomyelia. Granulomas and meningeal inflammation occur in spinal sarcoidosis, which also may result in cord cavitation70. In acute myelitis MRI shows diffuse swelling of the spinal cord and increased signal on T2W images (Fig. 60.44).

Figure 60.44 Acute myelitis, due to multiple sclerosis. Sagittal T2W MRI showing mild expansion of the upper spinal cord (low signal) and signal change (white) within.

• THE SPINE

It commonly extends over multiple segments. In demyelination, over 80 per cent of acute symptomatic cord lesions are shown. Major cord swelling is present in about 30 per cent of acute lesions with extensive diffuse signal changes shrinking over 2–3 months, to leave smaller residual lesions (mainly in the cord white matter); enhancement occurs after IV GdDTPA within the area of more extensive signal change for up to 8 weeks, but does not persist, which may help distinguish primary demyelinating processes from other forms of inflammation such as sarcoidosis. The changes in ADEM often resolve completely. Infections usually produce similar though more florid changes to those seen in acute demyelination, but usually follow a different course on interval imaging and more often result in cavitation and focal atrophy (cysts and arachnoiditis also seen in cysticercosis). MRI of the brain often shows clinically unsuspected lesions, which may help confirm a suspected diagnosis especially of multiple sclerosis, but unsuspected brain lesions may be present in many other conditions, including HTLV-1-associated myelopathy and sarcoidosis.

VASCULAR LESIONS Over 80 per cent of spinal arteriovenous malformations represent arteriovenous fistulas located in the spinal dura mater, usually close to a root sleeve103 (Fig. 60.45). Such fistulae most commonly are located in the thoracic region, but can occur at any level, though in the cervical spine only around the foramen magnum. A fistula usually is supplied by one or two dural branches of a nearby spinal artery, and shunts often via a single vein into intradural veins. In symptomatic lesions, venous drainage usually is both very slow and anomalous, remaining intradural through a greater part of the spinal canal than normal.Venous stagnation is an important cause of the clinical

Figure 60.45 Dural arteriovenous fistula. T2W MRI showing multiple enlarged vessels on the posterior surface of the spinal cord. (Courtesy of Dr Justin Cross.)

1387

1388

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

myelopathy, and in rare cases has resulted in extensive thrombosis of the intraspinal veins with haemorrhagic infarction of the cord: the Foix–Alajouanine syndrome (subacute necrotizing myelitis)104.

Intramedullary arteriovenous malformations Lesions involving multiple small, thin-walled vessels in the form of a poorly circumscribed mass may occur in the spinal cord, as they do in the brain. Sometimes, mainly only in children, a pial arteriovenous fistula is present. These lesions are supplied by radiculomedullary arteries, usually mainly by the sulco-commissural branches of the anterior spinal artery. Very occasionally there are associated vascular malformations in adjacent structures in the same body segment (Cobb’s syndrome). Capillary and cavernous angiomas also occur, and sometimes are multiple, but are less common than in the brain. MRI may demonstrate the following abnormalities in spinal dural fistulae: (A) large intrathecal and extrathecal vessels, which sometimes may be large enough to compress the spinal cord or give its surface a scalloped appearance; (B) thrombosis of intrathecal veins, either spontaneous or as a result of therapeutic embolization; (C) signal changes in the spinal cord are nearly always present in cases with persistent clinical myelopathy, which may diminish or resolve after obliteration of the shunt. Old and recent haemorrhage is seen mainly only with intramedullary malformations. The nidus may be identifiable in some intramedullary lesions and the site of a dural fistula may be shown by dynamic contrast-enhanced MRI105. It should be noted that normal pial veins can be prominent at the level of the lumbar enlargement of the spinal cord. The distribution of abnormally enlarged veins is a poor guide to the location of a dural fistula. Myelography was a sensitive test for detecting enlarged intrathecal veins but is rarely indicated now that MRI can provide high-resolution images of the vessels. However, these lesions are an absolute indication for spinal angiography; no other technique so precisely defines the relevant anatomy (Fig. 60.46). When searching for a dural fistula, imaging should be slow, say one frame every 2 s. A large number of spinal arteries may have to be injected before the lesion is found, and preliminary evaluation by dynamic MRI can be helpful in targeting the search and reducing its extent105. The study should not be regarded as negative unless: (A) all the spinal arteries from the foramen magnum to the coccyx have been opacified adequately, or (B) the veins thought to be abnormal have been opacified and shown to drain normally106. If a lesion is found, adjacent levels also should be injected, and it is advisable to identify the major radiculomedullary arteries in the same spinal region. Flush aortography does not result in adequate opacification of the relevant vessels, and often will not show these fistulae. Embolization of arteries supplying a dural fistula may be feasible, provided the vessel to be embolized can be also shown not to supply the spinal cord. Embolization and surgical excision of intramedullary arteriovenous malformations may also be possible in some cases without precipitating paraplegia.

Figure 60.46 Angiogram showing dural arteriovenous fistula. 1 = intercostal artery; 2 = site of malformation, close to the intervertebral foramen, i.e. lateral to the spinal cord; 3 = draining veins on spinal cord. Frontal projection.

Other vascular malformations Vertebral haemangiomas occur in 10 per cent of the otherwise normal population and are usually asymptomatic. They are not found in children younger than 10 years and occur rarely in the cervical spine. Occasionally, they cause back pain and very occasionally, compression of the spinal cord. Symptomatic lesions usually are associated with vertebral collapse or paraspinal extension. Extraosseous angiomas are rare, but often are associated with nerve roots, and present with radicular pain or dysfunction.

Imaging CT is more capable of demonstrating smaller lesions than plain radiographs. The vertebrae have a characteristic polka dot or stippled appearance. The matrix of the malformation often contains fat, so that the CT numbers are frequently in the negative range. Paraspinal extensions are well shown, especially in symptomatic cases. IV enhancement may show a soft-tissue extension as well, including intraspinal involvement. Extraosseous forms are similar to other extradural lesions. MRI commonly identifies these incidental findings. They often yield high signal because of the fat content. Lesions may be confused with metastases, when multiple or only parts of a vertebra are involved. However, the high signal on T1W images should distinguish them from nonhaemorrhagic metastases, the latter having a signal which is similar to or lower than haematopoietic marrow. MRI also shows soft-tissue extension and spinal cord compression, when present. Negative skeletal scintigraphy may help differentiate from metastases. Only rarely will spinal angiography be indicated to establish a diagnosis. Injection of the relevant intercostal or lumbar arteries shows an intense homogeneous capillary blush which, unlike a normal vertebra, typically does not respect the midline. Therapeutic embolization of symptom-

CHAPTER 60

atic cases may result in regression, and has been used to treat both back pain and neural compression.

Spinal cord infarction Spinal cord infarction is a rare complication of arteriosclerotic vascular disease. It may complicate aortic dissection and surgery for aortic aneurysms. The cause usually is occlusion of the sulco-commissural branches of the anterior spinal artery, the latter remaining patent. Infarction of the adjacent vertebral body may be associated107. Venous infarction seems to occur even less frequently, and has been due to extensive thrombosis of local pial veins in documented cases108. Only MRI demonstrates the pathological changes: signal change, focal or diffuse, associated only with mild swelling is seen, with enhancement in early stages. Rapid evolution over days results in contraction of the area of signal change and thinning of the involved region of the cord.Venous infarction may involve only one side of the spinal cord, such a pattern not being expected from arterial infarction.

Spontaneous epidural haematoma This is a rare, but devastating condition in which the time to act surgically, if appropriate, is short if spinal cord function is to be restored. Acute back pain and progressive flaccid paraplegia over hours is the usual clinical picture, sometimes mistakenly diagnosed as Guillan–Barré syndrome. A cause is rarely, if ever, found even with spinal angiography (which usually is not recommended unless other indications are present); occasionally it complicates metastatic disease. These haematomas are well shown by MRI and should also be clearly visible on modern CT; they are often extensive.

MISCELLANEOUS CONDITIONS Crystal deposition diseases Tumour-like, paraspinal and intraspinal calcified masses sometimes occur in association with similar periarticular masses elsewhere. They may be idiopathic or part of a calcium deposition disease accompanying metabolic disorders such as renal rickets, fluorosis or osseous metaplasia in spinal ligaments, as discussed with degenerative disease. They usually consist of calcium hydroxyapatite. Similar deposits have developed rarely in the cervical spines of patients recovering from major head injuries. Calcium pyrophosphate dyhydrate crystal deposition (pseudogout) may occur in intervertebral discs, especially in the lumbar region and can be associated with discovertebral destructive lesions. Gout rarely affects the apophyseal joints, but when it does, it may produce instability and nerve root compression.

Paget’s disease The axial skeleton is one of the commonest sites of involvement, especially the lumbosacral spine. About 80 per cent of cases are polyostotic, but the cervical vertebrae are only rarely involved. The apophyseal joints usually are normal. The vertebral body is enlarged in only about 20 per cent of cases. Sometimes paraspinal or even intraspinal soft-tissue masses, consisting

• THE SPINE

of poorly mineralized osteoid, develop, perhaps in response to incremental fractures. Spinal cord compression may occur. After medical treatment, the majority of cases show no radiological change, but about 20 per cent have shown increasing sclerosis; regression of paraspinal masses with improvement in neurological state can occur. Sarcomas develop in 0.15 per cent of cases of Paget’s disease, but only about 3 per cent of such sarcomas involve the spine. On MRI, Pagetic bone returns an exceptionally variegated signal, and on CT an expansive mixed lytic and sclerotic appearance.

Subacute combined degeneration of the spinal cord This is a progressive myelopathy presenting with sensory symptoms in the context of dietary deficiency of vitamin B12 or co-proteins. The lateral and dorsal spinal cord columns are affected, and improvement is often only partial after treatment. MRI often shows mild swelling and signal change (often subtle) in earlier stages that regress on treatment. Lesions, which are most often seen between C2 and C5, may enhance after IV contrast medium. However MRI may also remain negative109.

REFERENCES 1. Childers J C Jr, Wilson F C 1971 Bipartite atlas. Review of the literature and report of a case. J Bone Joint Surg 53A:578–582 2. Boon D, Parsons D, Lachmann S M et al 1985 Spina bifida occulta: lesion or anomaly? Clin Radiol 36:159–161 3. McRae D L, Barman A S 1953 Occipitalization of the atlas. Am J Roentgenol 70:23–46 4. Castelliri A E, Goldstein L A, Chan D P K 1984 Lumbosacral transitional vertebrae and their relationship with lumbar extradural defects. Spine 9:493–95 5. Enzmann D R, Pele A J 1993 Cerebrospinal flow measured by phase contrast cine MR. Am J Neuroradiol 14:1301–1307 6. Mikulis D J, Wood M L, Zerdoner O A M, Poncelet B P 1994 Oscillatory motion of the normal cervical spinal cord. Radiology 192:117–121 7. Rubin J B, Enzmann D R 1987 Harmonic modulation of proton MR precessional phase by pulsable motion: origin of spinal CSF flow phenomenon. Am J Roentgenol 148:983–994 8. Youser D M, Janick P A, Atlas S W et al 1990 Pseudoatrophy of the cervical portion of the spinal cord on MR images: a manifestation of truncal artefact? Am J Neuroradiol 11:373–377 9. Yostino M T, Temeltas O M, Carter L P et al 1993 Metallic post operative artefacts on cervical MR. Am J Neuroradiol 14:747–749 10. Williams D F 1994 Editorial. Titanium: epitome of biocompatibility or cause for concern. J Bone Joint Surg 76B:348–349 11. Porter B A, Shields A F, Olson D O 1986 Magnetic resonance imaging of bone marrow disorders. Radiol Clin North Am 24:269–289 12. Dietemann J L, Dogon D, Aubin M L, Manelfe C 1992 Cervico-occipital junction: normal and pathological aspects. In: Manelfe C (ed.) Imaging of the Spine and Spinal Cord. Raven Press, New York, pp 706–715 13. Ullrick C G, Binet E F, Sanecki M G, Kieffer S A 1980 Quantitative assessment of the lumbar spinal canal by computed tomography. Radiology 134:137–143 14. Dixon A K 1986 Who has most epidural fat? Information from computed tomography. Br J Radiol 59:475–480 15. Yu S, Haughton V M, Ho P S et al 1988 Progressive and regressive changes in the nucleus pulposus. Part II. The Adult. Radiology 169:93–97 16. Czervionke L F, Daniels D L, Ho P S et al 1988 Cervical neural foramina: correlative anatomic and MR imaging study. Radiology 169:753–759 17. Raz-Fumagalli R, Haughton V M 1993 Lumbar cribriform fascia: appearance at freezing microtomy and MR imaging. Radiology 187:241–246

1389

1390

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

18. Howicki B H, Haughton V M 1992 Neural foraminal ligaments of the lumbar spine: appearance at CT and MRI. Radiology 183:257–261 19. Schellinger D, Manz H J, Vidic B et al 1990 Disc fragment migration. Radiology 175:831–836 20. Seibert C E, Barnes J, Dreisbach J N et al 1981 Accurate CT measurement of the spinal cord using metrizamide: physical factors. Am J Neuroradiol 2:75–78 21. Wilkins R H 1994 Commentary: Prevalence and percutaneous drainage of cysts of the sacral nerve root sheath (Tarlov cyst). Am J Neuroradiol 15:298–299 22. Van de Kelft E, Van Vyve M 1993 Sacral meningeal cysts and perineal pain. Lancet 341:500–501 23. Stevens J M, Kendall B E, Gedroyc W 1991 Acute epidural haematoma complicating myelography. Case Report. Br J Radiol 64:860–864 24. Johnson A J, Burrows E H 1978 Thecal deformity after lumbar myelography with iophendylate (Myodil) and meglumine iothalamate (Conray 280). Br J Radiol 51:196–202 25. Doppman J L, Di Chiro G, Ommaya A K 1969 Selective arteriography of the spinal cord. Green, St Louis 26. Rodesch G, Lasjaunas P, Berenstein A 1992 Functional vascular anatomy of the spine and spinal cord. Revista di neuroradiologiea 63–66 27. Royal College of Radiologists 2003 Making the best use of a department of clinical radiology: guidelines for doctors. Royal College of Radiologists, London. 5th edn. 28. van den Bosch M A, Hollingworth W, Kinmonth A L, Dixon A K 2004 Evidence against the use of lumbar spine radiography for low back pain. Clin Radiol 59:69–76 29. Brooks B S, Duval E R, El Hammal P et al 1993 Neuro-imaging features of neurenteric cysts: analysis of nine cases and review of the literature. Am J Neuroradiol 14:735–746 30. Rogg J M, Benzil D L, Haas R L, Knucky N W 1993 Intramedullary abscess, an unusual manifestation of a dermal sinus. Am J Neuroradiol 14:1393–1395 31. David K M, Copp A J, Stevens J M 1996 Split cervical spinal cord with Klippel–Feil syndrome: seven cases. Brain 119:1859–1872 32. Huang P P, Constantine S 1994 Acquired Chiari 1 malformations. J Neurosurg 80:1099–1102 33. Savy L, Stevens J M, Taylor D J, Kendall BE 1994 Apparent cerebellar ectopia: a reappraisal using volumetric MRI. Neuroradiology 6:360–363 34. McLone D G, Knepper P A 1989 The cause of Chiari II malformation: a unified theory. Pediatr Neurosci 15:1–12 35. Stevens J M, Serva W, Kendall B E et al 1993 Chiari malformation in adults: relation of morphological aspects to clinical features and operative outcome. J Neurol Neurosurg Psychiatry 56:1072–1077 36. Stevens J M, Kling Chong W, Barber C et al 1994 A new appraisal of abnormalities of the odontoid process associated with atlanto-axial subluxation and neurological disability. Brain 117:133–148 37. Young A, Dixon A K, Getty J, Renton P, Vacher H 1981 Cauda equina syndrome complicating ankylosing spondylitis: use of electromyography and computerised tomography in diagnosis. Ann Rheum Dis 40:317–322 38. Walters M R, Stears J C, Osborn A G 1998 Transdural spinal cord herniation: imaging and clinical spectra. Am J Neuroradiol 19:1335–1336 39. Winter R B 1983 Congenital deformities of the spine. Thiem-Stratton, New York 40. Bethem D, Winter R B, Lutter L et al 1981 Spinal disorders of dwarfism: a review of the literature and report of eighty cases. J Bone Joint Surg 63A:1412–1425 41. Crockard H A, Stevens J M 1995 Craniovertebral function anomalies in inherited disorders. Eur J Pediatr 154:504–512 42. Ulmer J L, Mathews V P, Elster A D et al 1997 Lumbar spondylolysis without spondylolisthesis. Am J Roentgenol 169:233–237 43. Stevens J M, Kendall B E, Davis C, Crockard H A 1987 Percutaneous insertion of the spinal end of a cystoperitoneal shunt as definitive treatment to relieve spinal cord compression from spinal arachnoid cyst. Neuroradiology 29:190–195 44. Potter K, Saifuddin A 2003 MRI of chronic spinal cord injury. Br J Radiol 76:347–352

45. Grant R, Hadley D M, MacPherson P 1987 Syringomyelia-cyst measurement by magnetic resonance imaging and comparison with symptoms, signs and disability. J Neurol Neurosurg Psychiatry 50: 1008–1014 46. Williams B 1972 Pathogenesis of syringomyelia. Lancet i:142–143 47. Ball M J, Dayan A D 1972 Pathogenesis of syringomyelia. Lancet ii:799–801 48. Boijsen E 1954 The cervical spinal canal in intraspinal expansive processes. Acta Radiol 42:101–115 49. Baker A S, Dove J 1983 Progressive scoliosis as the first presenting sign of syringomyelia: report of a case. J Bone Joint Surg 65B:472–473 50. Enzmann D R, O’Donohue J, Rubin J B et al 1987 CSF pulsations within non-neoplastic spinal cord cysts. Am J Neuroradiol 8:517–525 51. Quencer R M, Donovan-Post M J, Hinks R S 1990 Cine MR in the evaluation of normal and abnormal CSF flow: intracranial and intraspinal studies. Neuroradiology 32:371–391 52. Stevens J M, Olney J S, Kendall B E 1985 Post-traumatic cystic and non-cystic myelopathy. Neuroradiology 27:48–56 53. Sherman J L, Berkovich A J, Citrin C M 1986 The MR appearances of syringomyelia: new observations. Am J Neuroradiol 7:985–995 54. Vaquero J, Martinez R, Arlas A 1990 Syringomyelia-Chiari complex. Magnetic resonance imaging and clinical evaluation of surgical treatment. J Neurosurg 73:14–18 55. Milhorat T H, Johnson W D, Miller J I et al 1992 Surgical treatment of syringomyelia based on magnetic resonance imaging criteria. Neurosurgery 31:231–242 56. Vissilouthis J, Panadreon A, Anagnostasas S 1993 Theco-peritoneal shunt for syringomyelia: Report of three cases. Neurosurgery 33: 324–328 57. Yu S, Haughton V M, Ho P S et al 1988 Progressive and regressive changes in the nucleus pulposus. Part II. The adult. Radiology 169: 93–97 58. Schellinger D, Manz H J, Vidic B 1990 Disc fragment migration. Radiology 175:831–836 59. Modic M T, Steinberg P M, Ross J S et al 1988 Degenerative disc disease: assessment of changes in vertebral bone marrow with MR imaging. Radiology 166:193–199 60. Grenier N, Kressel H Y, Schiebler M L et al 1987 Normal and degenerative spinal structures: MR imaging. Radiology 165:517–525 61. Yamashita Y, Takahashi M, Matsuno Y et al 1990 Spinal cord compression due to ossification of ligaments: MR imaging. Radiology 175:843–848 62. Okada K, Oka S, Tohge K et al 1991 Thoracic myelopathy caused by ossification of the ligamentum flavum. Clinico-pathological study and surgical treatment. Spine 16:280–287 63. Jackson D E, Atlas S W, Mani J R, Norman D 1989 Intraspinal synovial cysts: MR imaging. Radiology 170:527–530 64. Crockard J M, Sett P, Geddes J F et al 1991 Damaged ligaments at the craniocervical junction presenting as an extradural tumour: a differential diagnosis in the elderly. J Neurol Neurosurg Psychiatry 54:817–821 65. Yu Y L, Du Boulay G H, Stevens J M, Kendall B E 1986 Computer-assisted tomography in cervical spondylotic myelopathy and radiculopathy. Brain 109:259–278 66. Boden S, McCowin P, Davis D, Dina T, Mark A, Wiesel S 1990 Abnormal MR scans of the cervical spine in asymptomatic patients: a prospective investigation. J Bone Joint Surg 72A:1178–1184 67. Wiesel S, Tsourmas N, Feffer H, Citrin C, Patronas N 1984 A study of the computer assisted tomography: the incidence of positive CAT scans in an asymptomatic group of patients. Spine 9:549–551 68. Stevens J M, O’Dricoll D M, Yu Y L, Ananthapavan A, Kendall B E 1987 Some dynamic factors in compressive deformity of the cervical spinal cord. Neuroradiology 29:136–142 69. Stevens J M 1993 The compressed spinal cord. Current Medical Literature. Med Imaging (R Soc Med) 5:3–8 70. Stevens J M 1995 Imaging of the spinal cord. A review. J Neurol Neurosurg Psychiatry 58:403–408 71. Al-Mefty O, Harkey H L, Marawi I 1993 Experimental compressive cervical myelopathy. J Neurosurg 79:550–561

CHAPTER 60

72. Crockard H A, Heilman A E, Stevens J M 1993 Progressive myelopathy secondary to odontoid fractures: clinical, radiological and surgical features. J Neurosurg 78:579–586 73. Williams M P, Cherryman S R, Husband J E 1984 Significance of thoracic disc herniations demonstrated by MR imaging. J Comput Assist Tomogr 13:211–214 74. Bush K, Cowan N, Katz D, Gishen P 1992 The natural history of sciatica associated with disc pathology. Spine 17:1205–1212 75. Stevens J M, Clifton A G, Whitear R 1993 Appearances of posterior osteophytes after sound anterior interbody fusion in the cervical spine: a high definition computed myelographic study. Neuroradiology B5:227–228 76. Fukushima T, Takaaki I, Taoka Y, Takata S 1991 Magnetic resonance imaging study of spinal cord plasticity in patients with cervical compression myelopathy. Spine 16:534–538 77. Stockley I, Getty C J, Dixon A K et al 1988 Lumbar lateral canal entrapment: clinical, radiculographic and computed tomographic findings. Clin Radiol 39:144–149 78. Gundry C R, Heithoff K B 1993 Epidural haematoma of the lumbar spine: 18 surgically confirmed cases. Radiology 187:427–431 79. Mehali T F, Pezzuti R T, Applebaum B I 1990 Magnetic resonance imaging and cervical spondylotic myelopathy. Neurosurgery 26: 217–227 80. Jinkins J R, Osborn A G, Gerrett D, Hunt S, Story J L 1993 Spinal nerve enhancement with Gd-DTPA: MR correlation with the post-operative lumbar spine. Am J Neuroradiol 14:383–394 81. Boden S D, Davis D O, Dina T S, Parker C P, O’Malley S, Sanner J L, Wiesel S W 1992 Contrast-enhanced MR imaging performed after successful lumbar disc surgery: prospective study. Radiology 182: 59–64 82. Cavanagh S, Stevens J M, Johnson J R 1993 High resolution MRI investigation in recurrent pain after lumbar discectomy. J Bone Joint Surg 75B:524–531 83. Fitt G J, Stevens J M 1995 Post-operative arachnoiditis diagnosed by high resolution fast spin-echo MRI of the lumbar spine. Neuroradiology 37:139–145 84. Clifton A G, Stevens J M, Whitear P W, Kendall B E 1990 Identifiable causes for poor outcome in surgery for cervical spondylosis. Postoperative computed myelography and MR imaging. Neuroradiology 32:450–455 85. Levy W J Jr, Bay J, Dohn J 1982 Spinal cord meningiomas. J Neurosurg 57:804–812 86. Sze G, Krol G, Zimmerman R D, Deck M D F 1988 Malignant extra-dural spinal tumours: MR imaging with Gd-DTPA. Radiology 117:217–223 87. Balenaux D, Parizel P, Bank W O 1992 Intraspinal and intramedullary pathology. In: Manelfe C (ed.) Imaging of the Spine and Spinal Cord. Raven Press, New York, pp 513–564 88. Kaffenberger D A, Shah C P, Murtagh F R, Wilson C, Silbiger M L 1988 MR imaging of the spinal cord haemangioblastoma associated with syringomyelia. J Comput Assist Tomogr 12:495–498

• THE SPINE

89. Williams A L, Haughton V M, Pojunas K W, Daniels D L, Kilgore D P 1987 Differentiation of intramedullary neoplasms and cysts by MRI. Am J Roentgenol 149:159–164 90. Naidich T P, Doundoulakis S H, Poznanski A K 1986 Intraspinal masses: effect of plain spine radiography. Pediatr Neurosci 12:10–17 91. Rubin J M, Aisen A M, Di Pietro M A 1986 Ambiguities in MR imaging of tumoral cysts in the spinal cord. J Comput Assist Tomogr 10:395–398 92. Sosna J, Barth M M, Kruskal J B, Kane R A 2005 Intraoperative sonography for neurosurgery. J Ultrasound Med 24:1671–1682 93. Stevens J M, Olney J S, Kendall B E 1985 Post-traumatic cystic and non-cystic myelopathy. Neuroradiology 27:48–56 94. Falcone S, Quencer R M, Green B A 1994 Progressive post-traumatic myelomalacia myelopathy: imaging and clinical features. Am J Neuroradiol 15:747–754 95. Silberstein M, Nenessy O 1993 Implications of focal spinal cord lesions following trauma-evaluation with magnetic resonance imaging. Paraplegia 31:160–167 96. Kulkarni M R, McArdle C B, Kapanick D 1987 Acute spinal cord injury: MR imaging at 1.5T. Radiology 164:837–843 97. Yamashita Y, Takahiiki M, Matsumoto Y 1990 Chronic injuries of the spinal cord: assessment with MR imaging. Radiology 175:849–854 98. Davies P C, Reisner A, Hudgins P A 1993 Spinal injuries in children: role of MR. Am J Neuroradiol 14:607–617 99. Kendall B E, Stevens J M, Crockard H A 1992 The spine in rheumatoid arthritis. Rev Neuroradiol 5(Suppl 12):23–28 100. Colombo N, Berry I, Norman D 1992 Infections of the spine. In: Manelfe C (ed.) Imaging of the spine and spinal cord. Raven Press, New York 101. Kendall B E, Stevens J M, Thomas D 1991 Arachnoiditis. Curr Imaging 2:113–119 102. Johnson A J, Burrows E H 1978 Thecal deformity after lumbar myelography with iophendylale (Myodil) and meglumine iothalamate (conray 280). Br J Radiol 51:196–202 103. Kendall B E, Logue V 1977 Spinal epidural angiomatous malformations draining into intrathecal veins. Neuroradiology 13:181–189 104. Rodesch G, Berenstein A, Lasjaunias P 1992 Vascular and avascular lesions of the spine and spinal cord. In: Manelfe C (ed.) Imaging of the spine and spinal cord. Raven Press, New York, pp 565–598 105. Thorpe J W, Kendall B E, McMannus D, Millar D H 1994 Dynamic gadolinium-enhanced MRI with detection and localisation of spinal arterio-venous malformations. Neuroradiology 36:522–529 106. Willinsky R, Lasjaunias P, Terbrugge K, Hurth M 1990 Spinal angiography in the investigation of spinal arteriovenous fistula. A protocol with application to the venous phase. Neuroradiology 32:114–116 107. Yuh W T, Marsh C Y, Wang A K 1992 MR imaging of spinal cord and vertebral body infarction. Am J Neuroradiol 13:145–154 108. Henderson F C, Crockard H A, Stevens J M 1993 Spinal cord oedema due to venous stasis. Neuroradiology 35:312–315 109. Locatelli E R, Laurena R, Ballard P, Mark A S 1999 MRI in vitamin B12 deficiency myelopathy. Can Neurol Sci 26:60–63

1391

CHAPTER

The Orbit

61

Tarik F. Massoud and Justin J. Cross

• • • • • • •

Key anatomy of the orbit Imaging the globe Imaging the intraconal compartment Imaging the conal compartment Imaging the extraconal compartment Imaging orbital infection Imaging orbital trauma

The orbit and visual system is a highly complex and intricately organized region of the body, with specialized anatomical features and physiology. There are many diseases that are specific to this region, as well as many neurological diseases that afflict the sensory and motor components of the visual system. It is necessary to resort to cross-sectional and occasional angiographic imaging techniques to diagnose diseases that are not amenable to direct visual inspection and ophthalmoscopy. Computed tomography (CT) and magnetic resonance imaging (MRI) are the most useful techniques for imaging the orbit and its contents. It is usual practice in interpreting CT and MRI findings to ascribe the lesion to a certain compartment within the orbit, thus helping in anatomical interpretation and differential diagnosis. This general approach is useful when a lesion is small and/or confined to one location. However, several diseases are well known to transgress compartments as they progressively spread; the epicentre of such a process can often be assigned to a specific location. CT remains very useful because of the inherent natural contrast provided by the presence of structures with widely different attenuation coefficients (fat, bone, fluid, muscle, adjacent air) within a confined space. A relative advantage over MRI is also the latter’s sensitivity to lid and globe motion. CT is preferred over MRI for detecting small calcified optic nerve meningiomas and in a child with suspected retinoblastoma, where detection of calcification is paramount. Overall, however, MRI is gaining more acceptance because of its greater sensitivity in characterizing diseases of the orbit (especially with use of fat-suppressed T1-weighted imaging, combined head and surface coils and faster data acquisition), and the far greater ability and versatility of MRI in detecting concomitant intracranial (especially intra-axial) abnormalities.

This chapter briefly reviews key anatomical areas of the orbit, followed by discussions of the role of neuroimaging in detection and characterization of diseases of the orbit. Recently, Shields et al1 have published a useful survey of a large number of orbital tumours and other mass lesions. The various anatomical subdivisions within the orbit that help define ‘compartments’ for the purpose of imaging interpretation and differential diagnosis include: the globe and the intraconal, conal and extraconal compartments. When a disease can be multicompartmental, it is discussed in the section of the compartment where it arises or is present most frequently. Only the common diseases of each compartment are discussed; the imaging findings of less common diseases are presented in Tables 61.1–61.3. The imaging findings in orbital infections and trauma are presented separately in the last two sections.

KEY ANATOMY OF THE ORBIT The complex structures of the orbit can conveniently be clustered into several anatomical areas or compartments, including the bony orbit, globe, intraconal compartment (also containing the optic nerve), conal compartment and extraconal compartment. Gentry2 has provided a more detailed account of the normal radiological anatomy of the orbit. The bony orbit comprises four walls, four rims and the apex. The medial wall is formed mostly by the ethmoid bone. This is paper thin, and thus, termed the lamina papyracea. Anterior to this is the lacrimal bone and part of the horizontal portion of the frontal bone. Posterior to this is the maxillary bone and the lesser wing of the sphenoid.The lateral wall is formed mostly by the greater wing of the sphenoid. Anterior to this is the orbital surface of the zygomatic bone, and a small contribution again from part of the horizontal portion of the frontal bone.The superior wall is formed mostly by the orbital surface of the frontal bone. The inferior wall is formed mostly by the maxillary bone.The apex is where all four walls of the orbit converge. The superior orbital fissure is the space between the greater and lesser wings of the sphenoid bone. The inferior orbital fissure is formed by the space between the greater wing of the sphenoid and the posterolateral wall

1394

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 61.1 THE IMAGING FINDINGS OF SOME LESS COMMON DISEASES OF THE GLOBE

Congenital

Pathology

Clinical features

Key imaging findings

PHPV (persistent hyperplastic primary vitreous)

Primary vitreous normally involutes by sixth fetal month, but occasionally persists and undergoes hyperplasia Presents with leukocoria Affects male infants more than female

Microphthalmic globe with enhancing increased density in vitreous humour on CT Soft tissue band from back of lens to posterior globe (follows Cloquet’s canal) Can be unilateral or bilateral

Retinopathy of prematurity

History of prolonged ventilation with high O2 concentration in a premature baby

Bilateral increased density in vitreous Calcification is rare

Pathology shows abnormal proliferation of retinal vascular buds Coat’s disease

Microphthalmia (Fig. 61.4)

Congenital vascular malformation of retina with telangiectasia

Increased density in all or part of vitreous

Exudation from abnormal vessels leads to retinal detachment

Normal-sized globe No calcification

Congenital underdevelopment or acquired diminution in size of the globe

Congenital = small globe in a small orbit Acquired = small, calcified globe

Associated with congenital rubella, PHPV, retinopathy of prematurity and Lowe syndrome Macrophthalmia

Enlargement of the globe

Large globe in a large orbit

Most severe form is called buphthalmos Associated with juvenile glaucoma Coloboma

Defect in the globe, usually near optic nerve head

Small globe with cystic outpouching of vitreous

Involves the sclera, uvea and retina

May be a retro-ocular cyst

Caused by a defect in fetal optic fissure Degenerative

Drusen (Fig. 61.5)

Phthisis bulbi (Fig. 61.6)

Accretion of hyaline material on optic disc

Discrete, flat calcification of optic nerve head

May be asymptomatic or associated with headache or visual field defects

Bilateral in 75 per cent

End-stage injured eye

Collapsed globe May be calcified

Inflammatory

Scleritis

Anterior scleritis presents with pain, erythema, photophobia and tenderness

Thickened enhancing sclera Choroidal detachment may be present

Posterior scleritis is painless and may mimic melanoma Sclerosing endophthalmitis

Tumour

Choroidal haemangioma

2–8-year-old child exposed to soil contaminated by dog faeces

Dense vitreous without a discrete mass

Ingestion of the ova of Toxocara canis results in ophthalmitis

No calcification

Can be isolated or associated with Sturge–Weber syndrome

Lenticular or flat densely enhancing eye wall mass

Benign vascular lesion Medulloepithelioma

Mean age of onset 4 years

Involvement of ciliary body helps differentiate from retinoblastoma

Presents with ciliary body mass, lens coloboma, lens subluxation, cataract, cyclitic membrane and glaucoma

Only 10–15 per cent are calcified

About 50 per cent are teratoid and 50 per cent non-teratoid forms

Rarely may involve optic nerve and other locations in CNS

of the maxillary sinus. The optic canal lies between the two roots of the lesser wing of the sphenoid. The superior orbital rim is formed by the frontal bone; the lateral rim by the zygomatic and frontal bones; the inferior rim by the zygomatic and maxillary bones; and the medial rim by the frontal and maxillary bones.

The globe is mostly fluid filled and is surrounded by three major tissue layers: (A) the sclera and cornea, (B) the uveal tract (choroid, iris and ciliary body) and (C) the retina. The lens and iris divide the globe cavity into an anterior chamber filled with aqueous humor, and a posterior chamber filled with vitreous humor.

CHAPTER 61

• THE ORBIT

Table 61.2 THE IMAGING FINDINGS OF SOME LESS COMMON DISEASES OF THE CONAL AND INTRACONAL COMPARTMENTS Pathology

Clinical features

Key imaging findings

Congenital

Optic nerve hypoplasia

Can be isolated or part of a syndrome (e.g. septo-optic dysplasia)

Decreased size of optic nerve

Inflammatory

Optic neuritis (Fig. 61.13)

About 50 per cent of patients with idiopathic optic neuritis develop multiple sclerosis

Best seen on gadolinium-enhanced T1W imaging If present, look for brain demyelination using T2W imaging

Other causes include sarcoid, radiation, pseudotumour, toxoplasmosis, TB, syphilis, virus infection Tumour

Leukaemia

Reported in 13–16 per cent of cases of leukaemia

Diffuse enlargement of the optic nerve with variable enhancement

More commonly acute lymphoblastic leukaemia but described in AML and adult leukaemias Presents with papilloedema and variable loss of acuity Haemangioblastoma

Associated with von-Hippel Lindau (VHL)

Rarely affects orbit or optic nerve

Progressive loss of vision

Sharply demarcated from nerve

Retinal lesions occur in 60 per cent of patients with VHL

Densely enhancing

Mean age 40–60

Superior orbital masses

More common in women

Tend to invade locally

Presents with proptosis, optic nerve and extraocular dysfunction

Marked contrast enhancement

Usually affect pre-chiasmatic nerve Haemangiopericytoma

Florid blush on angiography Neurofibroma/

About 1 per cent of orbital tumours

Smooth, ovoid, solitary

schwannoma (Fig. 61.14)

Affects young adults

Usually in the superior orbit

Usually presents with proptosis

May be intraconal, extraconal or intramuscular

Neurofibromatosis in 2–18 per cent

Isodense with homogeneous contrast medium enhancement on CT Isointense on T1 and hyperintense on T2WI

Miscellaneous

Raised intracranial pressure (Fig. 16.15)

Papilloedema, loss of venous pulsation

The optic nerve is myelinated by oligodendrocytes, and not Schwann cells like true cranial nerves. It is about 5 cm long and is divided into four segments: intra-ocular, intra-orbital, intracanalicular and intracranial. The subarachnoid space between the dura and pia covering the optic nerve is continuous with that of the suprasellar cistern. The six extra-ocular muscles comprise four rectus muscles and two oblique muscles. The rectus muscles originate from the annulus of Zinn and insert into the sclera of the globe. The superior oblique muscle originates from the lesser wing of the sphenoid, hooks around the trochlea (just posterior to the medial aspect of the superior orbital rim) and the trochlear tendon inserts into the sclera beneath the tendon of the superior rectus muscle. The inferior oblique muscle originates from the anteromedial aspect of the orbital floor near the lacrimal sac and inserts into the sclera after passing below the inferior rectus tendon. The lateral rectus is innervated by the abducens nerve (cranial nerve VI), the superior oblique muscle by the trochlear nerve (IV) and all other muscles by the oculomotor nerve (III). The levator palpebrae superiorus muscle originates from the lesser wing of the sphenoid and

Dilatation of the optic nerve sheath

annulus of Zinn, is located above and parallel to the superior rectus muscle, and inserts into the superior lid retractor system. It is innervated by sympathetic fibres. The superior rectus and levator palpebrae superiorus muscles are difficult to separate on coronal images, and are therefore usually considered as one complex. Thick fascial layers, collectively termed the orbital septum, situated anterior to the globe, roughly divide the more anterior skin and subcutaneous tissues (preseptal compartment) from the more posterior orbit (postseptal compartment). Infection can spread from the preseptal compartment, through an often incomplete inferior portion of the orbital septum, to the postseptal compartment. The latter comprises the globe proper (bulbar compartment) and the retrobulbar compartment, with its separate intraconal, conal and extraconal compartments. Arterial supply to the orbit is mainly via the three large terminal branches of the ophthalmic artery, i.e. the lacrimal, nasociliary and frontal branches, in addition to smaller branches, mainly the central artery of the retina (this also supplies the inner aspect of the optic nerve) and short and long ciliary branches. Valveless veins in the orbit tend to follow arteries

1395

1396

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 61.3 THE IMAGING FINDINGS OF SOME LESS COMMON DISEASES OF THE EXTRACONAL COMPARTMENT Pathology Congenital

Cephalocele

Clinical features

Key imaging findings

Present soon after birth

Soft tissue and CSF continuous with the intracranial contents

Soft mass near medial canthus May be pulsatile and increase with Valsalva Dermoid (Fig. 61.18)

Usually upper outer quadrant of orbit

Usually anterior, between globe and periosteum

Fullness or small lump

Well-defined cystic masses Epidermoid = fluid density, dermoid = fat density on CT May be related to sutures

Lacrimal gland Inflammatory

Postviral

Commonest cause of acute inflammatory enlargement in younger patients

Smooth enlargement of the gland

Sjögren’s syndrome

Decreased lacrimation and dry mouth

Non-specific enlargement of the gland in acute phase

May be primary or secondary to autoimmune connective tissue diseases

Gland may be small in chronic phase Enhancement patchy or absent

Histology = lymphocytic infiltration of gland Mikulicz disease/ syndrome

Mikulicz disease is similar to primary Sjögren’s syndrome

As for Sjögren’s syndrome

Mikulicz syndrome is gland enlargement associated with sarcoid, lymphoma, leukaemia or TB Tumour

Benign mixed tumour (Fig. 61.19)

Same as pleomorphic adenoma Benign

Well-defined, smooth enlargement of gland

Represents about 50 per cent of primary lacrimal gland neoplasms (rest are malignant)

Longstanding so may be bone remodelling May not enhance

Can undergo malignant change Adenoid cystic carcinoma

Most common malignant primary tumour (followed by malignant mixed tumour, adenocarcinoma and mucoepidermoid carcinoma)

Tumour is hard enough to indent globe Gland may have a serrated edge Tendency for perineural spread Enhance well

Lymphoma (NHL)

Lacrimal gland is a common site for NHL in the orbit

Infiltrating mass Enhances well

but demonstrate greater interconnections until they form the superior ophthalmic vein, which drains through the superior orbital fissure into the cavernous sinus, and the inferior ophthalmic vein, which drains through the inferior orbital fissure to the cavernous sinus and the pterygoid plexus.

IMAGING THE GLOBE1 (see also Table 61.1) Retinoblastoma Retinoblastoma is the most common tumour of the globe in children. It occurs in children less than 3 years of age presenting with leukokoria. It derives from primitive photoreceptors or neuronal retinal cells, and histologically resembles other primitive neuroectodermal tumours. Of these tumours, 75 per cent are unilateral unifocal and 25 per cent are bilateral or unilateral multifocal. When it is found bilaterally and in conjunction with a pineoblastoma, it is labelled ‘trilateral retinoblastoma’; 10–40 per cent are familial (autosomal dominant, with the oncogene present on chromosome 13), and these tumours tend to be bilateral and associated with other nonocular tumours. Retinoblastoma is highly malignant and may

spread haematogenously, via lymphatics, or may spread along the optic nerve to the intracranial compartment to give drop metastases in the subarachnoid space. Imaging is crucial for timely management and survival of patients with retinoblastoma3. It is imperative to perform cross-sectional imaging and not just rely on ophthalmoscopy, so as to exclude other retrobulbar tumours with globe invasion, optic nerve invasion by the retinoblastoma and intracranial metastases. CT is the preferred method to image the child with leukokoria because it is sensitive to calcification in retinoblastoma. CT demonstrates clumped or punctate calcification (in 95 per cent of cases) in the posterior part of the globe extending into the vitreous, with minimal enhancement. In advanced cases the tumour may fill the globe (Fig. 61.1). If CT shows calcification in an intra-ocular mass in a child less than 3 years of age, it should be considered a retinoblastoma until proven otherwise. Absence of calcification means this diagnosis is unlikely, since it is rare in other causes of leukokoria. On MRI, retinoblastomas are hyper intense on T1 and hypointense on T2, possibly due to calcification, or some other paramagnetic entity, or tumour protein4. MRI is better for detection of tumour extension both along the optic

CHAPTER 61

Figure 61.1 Huge retinoblastoma with orbital and intracranial involvement. Axial CT image with contrast medium. The right globe is filled by a calcified mass. The optic nerve is also calcified and surrounded by a soft tissue mass that replaces orbital fat and extends through the optic foramen. There is involvement of the suprasellar cistern, temporal lobe, greater wing of sphenoid and temporal fossa.

nerve and intracranially5. The tumour distribution on CT and MRI may help in differentiating recurrent tumours and second primary neoplasms6.

Uveal melanoma Uveal melanoma is the most common primary intra-ocular malignancy in adults. The great majority are located unilaterally in the choroid. It can metastasize to the liver and lung.

• THE ORBIT

Diagnosis is usually performed on ophthalmoscopy and ultrasound (US). CT and MRI are not considered in the routine workup of this disease except when it is not possible to perform adequate ophthalmoscopy, e.g. in the presence of an opaque vitreous or when there are large subendothelial effusion(s). CT and MRI demonstrate a uveal melanoma as a soft tissue mass centred on the outer layers of the globe (Fig. 61.2). This mass bulges inward into the vitreous and may be small and flat or crescentic, or large and sharply demarcated with a ‘mushroom cloud’ appearance. On CT it is hyperdense and enhances after contrast medium administration. On MRI it is hyperintense on T1 and hypointense on T2 in melanotic melanomas due to the presence of paramagnetic melanin with/without haemorrhage. Metastases to the choroid from mucinous adenocarcinomas may have the same signal characteristics. Amelanotic melanomas are hypointense on T1 and hyperintense on T2, as are other tumours. Both types of melanomas enhance on MRI. MRI is better than CT for adequate differential diagnosis with choroidal haemangiomas and choroidal detachments, and for better depiction of episcleral invasion7,8.

Ocular metastases Only 50 per cent of patients with ocular metastases have a known primary tumour, commonly from lung or breast cancer in women and lung and the gastrointestinal tract in men. Ocular metastases occur mostly to the uveal tract (Fig. 61.3). They are seen on CT as small multiple areas of hyperdense thickening, occasionally with subretinal fluid. When found bilaterally in the posterior temporal regions near each macula, it suggests metastases rather than choroidal haemangiomas associated with Sturge–Weber syndrome.

Figure 61.2 Malignant melanoma of the choroid. (A) Oblique sagittal T1-weighted MR image. (B) Oblique coronal T2*-weighted image. There is a nodule applied to the wall of the globe that is hyperintense on unenhanced T1- (black arrow) and hypointense on T2*-weighted (white arrow) images. These signal intensities are in keeping with melanin.

1397

1398

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 61.3 Ocular metastasis from systemic lymphoma. (A) Axial T2, (B) axial T1, (C) axial T1 MRI with IV gadolinium and fat suppression. There is thickening of the wall of the globe with soft tissue and enhancement extending into the vitreous and retrobulbar space.

IMAGING THE INTRACONAL COMPARTMENT2 Pseudotumour Pseudotumour is the commonest cause of intra-orbital mass lesion in adults. Pseudotumour is an idiopathic inflammatory condition with rapid onset9. It usually presents in middle age with unilateral painful ophthalmoplegia, proptosis and chemosis. The more common acute form of the disease results in an early lymphocytic infiltrate. This has a rapid and lasting response to steroids. In the chronic form of pseudotumour there is a poor response to steroids and fibrosis sets in, requiring chemotherapy or radiotherapy as other treatment options. It is probably autoimmune, and in 10 per cent of cases is found with other systemic autoimmune conditions such as Wegener’s granulomatosis, fibrosing mediastinitis, Reidel’s thyroiditis, sclerosing cholangitis, retroperitoneal fibrosis, as well as polyarteritis nodosa, dermatomyositis and rheumatoid arthritis. Tolosa-Hunt syndrome is an idiopathic inflammatory condition similar to pseudotumour affecting the cavernous sinus and orbital apex, also presenting with painful ophthalmoplegia.

The morphological changes apparent on cross-sectional imaging may be found in any structure in the orbit (Fig. 61.7). In order of frequency, the retrobulbar fat, extra-ocular muscles, optic nerve, globe (uveal-scleral area) and the lacrimal gland are affected. Two patterns of the disease may be recognized: (A) the tumefactive type, where there is diffuse involvement of conal/intraconal structures, or (B) the myositic type involving extra-ocular muscles. Involvement of a unilateral single extra-ocular muscle including the tendinous insertion is highly suggestive of pseudotumour rather than thyroid ophthalmopathy. The tumefactive type of pseudotumour may be differentiated from true tumour by the history, biopsy findings and the response to steroids. Pseudotumour enhances after contrast administration, and on CT there may be subtle hyperdensity of intra-orbital fat (dirty fat). Different characteristics of attenuation change on two-phase helical CT and delayed coronal CT can be helpful in differentiating between orbital lymphoma and pseudotumour10. On MRI pseudotumour is hypointense to fat on T2, whereas true tumours are hyperintense to fat on T2. MRI is useful for demonstrating the presence of a variety

CHAPTER 61

• THE ORBIT

Figure 61.4 Congenital microphthalmos. Axial CT image. There is choroidal calcification (large black arrow) with a small globe, a thinned optic nerve (small white arrows) and a small orbit. Note the hypoplastic optic canal (black arrowhead).

Figure 61.7 Orbital pseudotumour. (A) Axial and (B) coronal CT images with IV contrast medium. There is an extra- and intraconal mass with enlargement of the lacrimal gland (large arrow). There is subtle enhancement of the choroid (small arrows). The coronal image shows involvement of the extra-ocular muscles.

Figure 61.5 Drusen. Axial CT. There are small foci of calcification at both optic nerve heads.

Figure 61.6 Phthisis bulbi. Axial CT. The patient had been stabbed in the right eye 2 years previously. The globe is small and densely calcified.

of extra-orbital extensions of orbital inflammatory pseudotumours11.

Lymphoma Orbital lymphoma presents in middle age with painless orbital swelling progressing to proptosis. There is usually no evidence of systemic disease at presentation, although this develops subsequently. Orbital lymphoma is of the B-cell variety (nonHodgkin’s lymphoma [NHL]); Hodgkins disease of the orbit

is rare. Any structure in the orbit may be affected12,13. The lacrimal gland is involved most frequently, then the conal/ intraconal compartment (the superior rectus is the commonest extra-ocular muscle involved)14, and then the optic nerve/ sheath complex, where it may simulate optic nerve meningioma or neuritis15. There is a wide range of radiological findings in orbital lymphoma. It may be a well-defined hyperdense enhancing mass on CT, or can produce diffuse infiltration leading to destruction of the normal anatomical architecture. It however moulds to the contour of the orbit without bone destruction, unless it is very aggressive. The combination of PET/ CT may be useful in the evaluation of orbital neoplasms, especially if lymphoma is suspected16. On MRI lymphoma tends to be hypointense on T1, is usually hyperintense on T2, and enhances (Fig. 61.8). Bilateral orbital masses suggest the diagnosis of lymphoma. It can appear similar to pseudotumour, but lymphoma (and true tumours) tends to be superior in the orbit and the history may be of help. Moreover, comparison of lesion signal intensity to that of extra-ocular muscle appears to be a better alternative than comparison with cerebral gray matter or periorbital fat in differentiating malignant lymphoma from atypical lymphocytic infiltrates17.

Cavernous haemangioma Cavernous haemangioma is the most common orbital tumour. These lesions occur in adults 20–40 years of age and in men more than women. They present with proptosis, but vision is

1399

1400

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 61.8 Lymphoma. (A) Axial T2, (B) axial T1, (C) axial T1 MRI with gadolinium, (D) coronal T1 with gadolinium and fat suppression. There is a diffusely infiltrating mass in the superior right orbit that is isointense to brain on T2- and slightly hypointense on the T1-weighted sequence. The mass extends outside the orbit to involve the temporal fossa. Following gadolinium, there is homogeneous signal enhancement of the mass. On coronal imaging, there is thick meningeal enhancement indicating intracranial spread of lymphoma.

usually unaffected. They are comprised of large endotheliallined vascular spaces with a dense fibrous pseudocapsule, and are usually intraconal in location. They have no recognizable arterial feeders or draining veins, and may have phleboliths and surrounding haemosiderin/ferritin deposition, although intralesional haemorrhage is rare18,19. CT and MRI usually show a sharply demarcated, rounded or oval mass that spares the orbital apex20 (Fig. 61.9). Bony deformity due to erosion may occur, but no bone destruction. CT shows these lesions to be hyperdense, and they enhance.

MRI is better to show the relationship of the optic nerve and extra-ocular muscles. Cavernous haemangioma is usually isointense or hypointense on T1, hyperintense on T2, and enhances. Differences in contrast-enhancement spread patterns may be useful in distinguishing between haemangiomas and other tumours such as orbital schwannomas21.

Capillary haemangioma This lesion is found in infants less than 1 year of age who present with proptosis. It results from proliferation of endothelial

CHAPTER 61

• THE ORBIT

Figure 61.9 Cavernous haemangioma. (A) Coronal T1-weighted MR image with contrast medium. The heterogeneous mass (black arrowhead) lies inferolateral to the optic nerve (arrow). (B) Axial T1-weighted image with gadolinium and fat suppression. The heterogeneous enhancement of the mass (black arrowhead) corresponds to pooling of contrast medium within intratumoral vascular spaces. (C) Sagittal T2*-weighted image. The lesion (black arrowhead) is hyperintense and has a low intensity rim, probably because of a combination of fibrous tissue and haemosiderin/ ferritin in the capsule.

cells with multiple capillaries. They regress spontaneously in the first few years of life. CT and MRI show an enhancing mass that spans the intraconal and extraconal compartments. This mass is not usually sharply demarcated, and indeed its irregular margins may lead to confusion with malignancy. MRI is the imaging tool of choice to show the many punctate hypointense flow voids within such masses.

Optic nerve meningioma Optic nerve meningiomas occur in middle-aged women, and more rarely in children with neurofibromatosis type 2. They are the second commonest primary tumours of the optic nerve after gliomas22,23.They arise from the arachnoid layer of the leptomeninges (meningoblastic rests) surrounding the optic nerve without infiltrating it, and they may be bilateral in neurofibromatosis 1 or 2. Optic nerve meningiomas may spread intracranially, but only to the prechiasmatic optic nerve sheath. Hyperostosis in a remodelled widened optic canal may be present. Optic nerve meningiomas result usually in tubular thickening of the optic nerve/sheath complex rather than fusi-

form or excrescent thickening. Along with this thickening, the optic nerve tends to be seen as separate from tumour. Since these lesions are hyperdense on CT and also enhance strongly they produce a ‘tram track’ sign on axial images and a ‘doughnut sign’ on coronal images. The differential diagnosis of ‘tram tracks’ includes pseudotumour, sarcoid and metastasis. Calcification (psammoma bodies) is common (20–50 per cent of cases). CT may be necessary to detect the linear calcification of early lesions. On MRI, these lesions are hypointense on T1 and hyperintense on T2. They again enhance strongly, especially appreciated with fat suppression (Fig. 61.10). Early meningiomas are thus rendered more conspicuous on T1 fatsuppressed post-contrast images. Overall, MRI is better for orbital apex and intracanalicular lesions because surrounding bone makes this area difficult to outline accurately on CT. The radiological differential diagnosis of optic nerve meningiomas includes haemangioblastomas of the optic nerve in von Hippel–Lindau disease, cavernous haemangiomas and lymphoma. Meningiomas may also arise from the periostium of the orbit (the periorbita) which is in continuity with the intracranial dura mater.

1401

1402

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 61.10 Optic nerve meningioma. (A) Axial T2, (B) axial T1 MRI with gadolinium and fat suppression. There is a mass at the right orbital apex, closely applied to the optic nerve but seen separate to it. On the contrast-enhanced image, ‘tram-track’ enhancement along the nerve can be seen.

Optic nerve glioma Optic nerve gliomas are childhood slow-growing low-grade pilocytic astrocytomas, with 75 per cent of cases occurring at less than 10 years of age22. They cause decreased vision with minimal proptosis. Of patients with neurofibromatosis type 1, 15 per cent have optic nerve or chiasmatic gliomas that may be bilateral24. Bilateral tumours are virtually pathognomonic of neurofibromatosis. Optic nerve gliomas constitute 80 per cent of all primary optic nerve tumours. These lesions result in tortuous thickening of the optic nerve/sheath complex that is most commonly tubular, but may also be fusiform or excrescent. Unlike optic nerve meningiomas, gliomas cannot be separated from the optic nerve itself (Fig. 61.11). Only half these lesions enhance, and they may contain cysts. Calcification is rare except post radiotherapy. MRI is necessary to detect involvement of the rest of the optic pathway because only 25 per cent of optic pathway gliomas are confined to the optic nerves. However, an optic nerve glioma does not itself tend to spread from the optic nerve to the intracranial compartment. MRI shows these lesions as isointense on T1 and hyperintense on T2. However, in neurofibromatosis they may be hypointense on T2 but with a hyperintense rim due to arachnoidal gliomatosis.

Carotid-cavernous fistulae Carotid-cavernous fistulae (CCFs) are fistulae between the carotid siphon and the cavernous sinus.They may occur spontaneously, for example after rupture of a carotid siphon aneurysm or after trauma. Clinically they present with engorgement of the orbit and globe (sclera and conjunctiva), pulsating

exophthalmos, a bruit and eventually glaucoma and visual loss in larger CCFs. Cross-sectional imaging shows signs of orbital venous hypertension with an enlarged engorged superior ophthalmic vein and extra-ocular muscles (Fig. 61.12A). MRI shows signal void in the cavernous sinus and superior ophthalmic vein due to the presence of fast-flowing arterial blood. The enlarged cavernous sinus may be bowed convex to the middle cranial fossa. MR angiography shows filling of the cavernous sinus and superior ophthalmic vein in conjunction with the arterial anterior intracranial circulation. Conventional angiography with injections of internal and external carotid arteries can usually distinguish between ‘direct’ fistulae and ‘indirect’ ones (a dural arteriovenous malformation in the region of the cavernous sinus). Angiography may show filling of the ipsilateral or contralateral cavernous sinus via the intercavernous sinuses, and drainage into ipsilateral or bilateral superior ophthalmic veins (Fig. 61.12B), inferior petrosal sinuses, or even the cortical veins or spheno-parietal sinuses when severe.

Venous varix A venous varix is an enormously dilated vein representing a congenital or acquired (e.g. post-traumatic) venous malformation. It may occur on its own or be associated with an intraorbital or intracranial arteriovenous malformation. Multiple varicosities may be present. Clinically they present with intermittent proptosis upon straining or coughing and retrobulbar pain. On CT a venous varix is seen as an intraconal hyperdense lobulated mass with strong enhancement. Phleboliths and clot

CHAPTER 61

• THE ORBIT

Figure 61.11 Optic nerve glioma. (A) Axial T2, (B) axial T1, (C) axial T1 MRI with gadolinium, (D) coronal T1 with gadolinium and fat suppression. There is smooth expansion of the left optic nerve extending to the orbital apex. On contrast-enhanced imaging, there is homogeneous enhancement of the mass, and the nerve cannot be distinguished from it.

Figure 61.12 Post-traumatic high flow carotico-cavernous fistula. (A) Axial CT with contrast medium shows marked dilatation of the left superior ophthalmic vein (arrow) and moderate dilatation of the right superior ophthalmic vein. (B) Carotid angiography shows early filling of the cavernous sinus and left superior ophthalmic vein (arrow).

1403

1404

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 61.13 Optic neuritis. Coronal T2-weighted image with inversion recovery. There is high signal in the left optic nerve indicating optic neuritis. The patient was symptomatic and had multiple demyelinating lesions in the cerebral white matter.

Figure 61.15 Raised intracranial pressure. Axial T2 MRI. There is dilatation of the optic nerve sheath in a patient with raised intracranial pressure secondary to right transverse sinus occlusion.

phenomena. Clot due to spontaneous thrombosis is common, which can result in variable signal intensity on MRI.

IMAGING THE CONAL COMPARTMENT3 Thyroid ophthalmopathy

Figure 61.14 Intraconal schwannoma. (A) Axial and (B) coronal CT image. There is a large oval mass filling much of the orbit. It is almost isodense compared to muscle with no enhancement following IV contrast medium. The mass had not changed in size over a period of 5 years.

may be present. When small, a varix may not be seen unless it is made to enlarge with a Valsalva manoeuvre or during imaging in the prone position. Orbital phlebography is no longer required for making the diagnosis. MRI may reveal slow flow

Thyroid ophthalmopathy is the commonest cause of unilateral or bilateral exophthalmos in adults. Most patients with thyroid eye disease are hyperthyroid; only 10 per cent are euthyroid (euthyroid ophthalmopathy). It presents with insidious and painless exophthalmos and lid lag. It results from deposits of hygroscopic mucopolysaccharides and infiltration of lymphocytes, mast cells and plasma cells. Of these cases, 85 per cent are bilateral but often asymmetrical25. Axial CT or MRI shows fusiform enlargement and enhancement of the bellies of extra-ocular muscles, with relative sparing of the tendinous insertions (Fig. 61.16). This occurs in order of frequency in the inferior rectus, medial rectus, superior rectus, lateral rectus and oblique muscles, although most commonly all muscles are involved. Only in 10 per cent is there an isolated extra-ocular muscle involved. Importantly, if an isolated lateral rectus belly enlargement is seen, causes other than thyroid ophthalmopathy (e.g. pseudotumour) should be sought. The peak signal intensity from the most inflamed extra-ocular muscle could be the most reliable imaging correlate of clinical disease26. Haemodynamic information obtained by dynamic contrastenhanced MRI may also be useful in evaluating the clinical course of thyroid ophthalmopathy27. For example, the mean of peak enhancement ratio values for the extra-ocular muscles in patients with Graves’ disease tends to decrease according to the severity of the anatomical and clinical changes. The mean

CHAPTER 61

rate of enhancement also decreases according to the severity of the disease. Intra-orbital fat volume is increased, especially in the anteromedial extraconal space, although fat hypertrophy may occur also in other conditions such as following steroid therapy or in Cushing’s disease. In advanced cases the lamina papyracea may show a concavity due to the raised intra-orbital pressure. Orbital apex crowding (due to the enlarged muscles and increased fat content) results in compression of the optic nerve and decreased vision (also exacerbated by the stretching of the nerve).This may require orbital decompression by removal of the medial wall or floor of the orbit. Coronal imaging with either CT or MRI is useful to assess muscle thickness and orbital apex crowding of the optic nerve.

• THE ORBIT

Rhabdomyosarcoma Rhabdomyosarcoma is one of the more common primary malignant tumours of the orbit in children aged 2–5 years. Orbital rhabdomyosarcoma is the most common site of head and neck rhabdomyosarcoma. It is highly malignant and presents with rapidly progressive exophthalmos. It originates from the extra-ocular muscles, the nasopharynx, or the paranasal sinuses. It is usually present in the superomedial orbit and may produce bone destruction. On CT, these bulky aggressive-looking masses are isodense or slightly hyperdense, and show uniform enhancement (Fig. 61.17). On MRI they are of intermediate signal intensity on both T1 and T2 sequences.There is bone destruction in 40 per cent of cases and frequent distortion of the globe.

Figure 61.16 Thyroid ophthalmopathy. (A) Axial and (B) coronal CT imaging. There is generalized enlargement of the bellies of all the extra-ocular muscles, proptosis and increased intraorbital fat.

Figure 61.17 Rhabdomyosarcoma. (A) Axial and (B) coronal CT images with contrast medium. There is a large mass in the superior right orbit which is difficult to separate from the extra-ocular muscles. There is deformity of the posterior wall of the globe and marked proptosis. The mass shows uniform contrast enhancement.

1405

1406

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

IMAGING THE EXTRACONAL COMPARTMENT4 Retrobulbar metastases Most retrobulbar metastases are extraconal in location, and subsequently encroach on the intraconal compartment as they increase in size to produce infiltrating poorly marginated masses. They originate mostly from the greater wing of the sphenoid, resulting in bone destruction. On CT, these lesions are isodense or hyperdense, and enhance. In children, neuroblastoma or Ewing’s sarcoma produce smooth extraconal masses related to the posterior lateral wall of the orbit. They differ from rhabdomyosarcoma because of their baseline hyperdensity and lack of invasion of the preseptal compartment. In adults, an infiltrative retrobulbar mass and

enophthalmos is characteristic of scirrhous carcinoma of the breast.

IMAGING ORBITAL INFECTION Other than infection of the globe with larvae of the nematode Toxocara canis and the long-term development of sclerosing endophthalmitis (presented briefly in Table 61.1), both CT and MRI appearances are nonspecific for particular ocular infections. Ocular Pseudomonas infection may show up as a posterior scleritis cytomegalovirus (CMV) infection in human immunodeficiency virus (HIV)-positive patients and may result in arcuate enhancement of the retina.

Figure 61.18 Dermoid. (A) Axial and (B) coronal CT images with contrast medium. There is a fat density mass in the superolateral left orbit with a thick enhancing capsule. Subtle deformity of the adjacent bone is noted.

Figure 61.19 Benign mixed tumour of the lacrimal gland. (A) Axial T1-weighted MR and (B) coronal CT image with contrast medium (wide window setting). The mass (star) arises from the gland, is homogeneous and does not enhance after contrast medium. There is subtle bony remodelling.

CHAPTER 61

The orbital septum acts as a mechanical barrier to spread of infection into the orbit.This is because the periorbita is reflected onto the anterior aspect of the septum and globe. Thus, preseptal cellulitis is usually confined to the eyelids. On the other hand, much more serious orbital postseptal infection can arise from sinus disease, bacteraemia, trauma, or spread of a serious infection from the skin. On CT and MRI this is seen as illdefined tissue planes in retrobulbar structures, loss of the normal hyperintense signal of orbital fat on MRI, with or without a soft tissue mass. The latter may demonstrate ring enhancement and/or pockets of gas suggestive of abscess formation. A subperiosteal abscess may develop, especially in the presence of ethmoid sinusitis (Fig. 61.20). This can be identified as a soft tissue mass, with or without central fluid, centred on the bony (usually the medial) wall of the orbit and displacing the adjacent extra-ocular muscle whilst preserving a thin layer of extraconal fat.The lamina papyracea may or may not be destroyed. Further evaluation of the intracranial compartment using MRI is useful to exclude spread of infection in the presence of orbital cellulitis and subperiosteal infection.

IMAGING ORBITAL TRAUMA Conventional radiography is no longer used routinely for evaluation of orbital trauma. Instead, CT is the procedure of choice, whilst MRI is reserved for examining the intracanalicular optic nerve, for detecting transection of the optic nerve, for distinguishing haematoma from any other soft tissue mass, and for examining concomitant trauma to the posterior visual pathways and intracranial contents.

• THE ORBIT

There are four categories of orbital fractures to consider: zygomaticomaxillary (trimalar, tripod) fractures, LeFort fractures, orbital blowout fractures (Fig. 61.21) and orbital roof fractures. Radiological evaluation of trauma to the globe is usually necessary in cases of suspected blunt ocular trauma or for determining the presence and location of penetrating foreign bodies. Blunt trauma may uncommonly result in globe rupture or laceration, which is usually evident by loss of the normal globe contour on CT or MRI; dislocation of the lens may also be seen in these cases. More commonly, blunt trauma results in retinal or choroidal detachments. Retinal detachment results in a V-shaped membrane with the apex of the V at the optic disc. The associated subretinal fluid might be hyperintense on T1-weighted MRI because of proteinaceous fluid or haematoma. Choroidal detachments are distinguished by the fact that they do not extend to the optic nerve head because of the tethering effect of vortex veins. Referrals regarding the presence and site of a penetrating intra-ocular foreign body are common. Multidetector CT is usually helpful to demonstrate glass, metal and bone fragments, whereas wood splinters and thorns are difficult to find because they are isodense to soft tissues or may resemble intraorbital air. In such cases, MRI might be useful. However, MRI is contraindicated when there is suspicion of a metallic intra-ocular foreign body because the magnetic field may cause the object to move, which may result in further trauma and intra-ocular haemorrhage.

Figure 61.20 Subperiosteal abscess. (A) Axial and (B) coronal CT images with contrast medium. There is sinusitis affecting the right maxillary and ethmoid sinuses and a subtle rim enhancing fluid collection extending along the medial wall of the orbit indicating a subperiosteal abscess. Note the intact lamina papyracea despite intraorbital extension of infection.

1407

1408

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 61.21 Orbital blowout fracture. (A) Coronal and (B) sagittal reformatted CT images. There is intra-orbital air (arrow) and herniation of fat and inferior rectus into the maxillary sinus (arrowheads).

REFERENCES 1. Shields J A, Shields C L, Scartozzi R 2004 Survey of 1264 patients with orbital tumors and simulating lesions: The 2002 Montgomery Lecture (Pt 1) Ophthalmology 111: 997–1008 2. Gentry L R 1998 Anatomy of the orbit. Neuroimaging Clin N Am 8: 171–194 3. Apushkin M A, Apushkin M A, Shapiro M J, Mafee M F 2005 Retinoblastoma and simulating lesions: role of imaging. Neuroimaging Clin N Am 15: 49–67 4. Gizewski E R, Wanke I, Jurklies C, Gungor A R, Forsting M 2005 T1 Gdenhanced compared with CISS sequences in retinoblastoma: superiority of T1 sequences in evaluation of tumour extension. Neuroradiology 47: 56–61 5. De Graaf P, Barkhof F, Moll A C, Imhof S M, Knol D L, van der Valk P, Castelijns J A 2005 Retinoblastoma: MR imaging parameters in detection of tumor extent. Radiology 235: 197–207 6. Tateishi U, Hasegawa T, Miyakawa K, Sumi M, Moriyama N 2003 CT and MRI features of recurrent tumors and second primary neoplasms in pediatric patients with retinoblastoma. AJR Am J Roentgenol 181: 879–884 7. Blanco G 2004 Diagnosis and treatment of orbital invasion in uveal melanoma. Can J Ophthalmol 39: 388–396 8. Recsan Z, Karlinger K, Fodor M, Zalatnai A, Papp M, Salacz G 2002 MRI for the evaluation of scleral invasion and extrascleral extension of uveal melanomas. Clin Radiol 57: 371–376 9. Jacobs D, Galetta S 2002 Diagnosis and management of orbital pseudotumor. Curr Opin Ophthalmol 13: 347–351 10. Moon W J, Na D G, Ryoo J W et al 2003 Orbital lymphoma and subacute or chronic inflammatory pseudotumor: differentiation with two-phase helical computed tomography. J Comput Assist Tomogr 27: 510–516 11. Lee E J, Jung S L, Kim B S et al 2005 MR imaging of orbital inflammatory pseudotumors with extraorbital extension. Korean J Radiol 6: 82–88 12. Fahim D K, Bucher R, Johnson M W 2005 The elusive nature of primary intraocular lymphoma. J Neuroophthalmol 25: 33–36 13. Neudorfer M, Kessler A, Anteby I, Goldenberg D, Barak A 2004 Coexistence of intraocular and orbital lymphoma. Acta Ophthalmol Scand 82: 754–761

14. Kert G, Clement C I, O’Donnell B A 2004 Orbital lymphoid tumour located within an extraocular muscle. Clin Experiment Ophthalmol 32: 651–652 15. Selva D, Rootman J, Crompton J 2004 Orbital lymphoma mimicking optic nerve meningioma. Orbit 23: 115–120 16. Chan-Kai B T, Yen M T 2005 Combined positron emission tomography/ computed tomography imaging of orbital lymphoma. Am J Ophthalmol 140: 531–533 17. Akansel G, Hendrix L, Erickson B A 2005 MI patterns in orbital malignant lymphoma and atypical lymphocytic infiltrates. Eur J Radiol 53: 175–181 18. Yan J, Wu Z 2004 Cavernous hemangioma of the orbit: analysis of 214 cases. Orbit 23: 33–40 19. Scheuerle A F, Steiner H H, Kolling G, Kunze S, Aschoff A 2004 Treatment and long-term outcome of patients with orbital cavernomas. Am J Ophthalmol 138: 237–244 20. Ansari S A, Mafee M F 2005 Orbital cavernous hemangioma: role of imaging. Neuroimaging Clin N Am 15: 137–158 21. Tanaka A, Mihara F, Yoshura T et al 2004 Differentiation of cavernous hemangioma from schwannoma of the orbit: a dynamic MRI study. AJR Am J Roentgenol 183: 1799–1804 22. Miller NR 2004 Primary tumours of the optic nerve and its sheath. Eye 18: 1026–1037 23. Saeed P, Rootman J, Nugent R A, White V A, Mackenzie I R, Koornneef L 2003 Optic nerve sheath meningiomas. Ophthalmology 110: 2019–2030 24. Thiagalingam S, Flaherty M, Billson F, North K 2004 Neurofibromatosis type 1 and optic pathway gliomas: follow-up of 54 patients. Ophthalmology 111: 568–577 25. El-Kaissi S, Frauman A G, Wall J R 2004 Thyroid-associated ophthalmopathy: a practical guide to classification, natural history and management. Intern Med J 34: 482–491 26. Mayer E J, Fox D L, Herdman G et al 2005 Signal intensity, clinical activity and cross-sectional areas on MRI scans in thyroid eye disease. Eur J Radiol 56: 20–24 27. Taoka T, Sakamoto M, Nakagawa H et al 2005 Evaluation of extraocular muscles using dynamic contrast enhanced MRI in patients with chronic thyroid orbitopathy. J Comput Assist Tomogr 29: 115–120

CHAPTER

Ear, Nose and Throat Radiology

62

Philip Anslow

The ear • The external ear • The middle ear • The inner ear The nose and paranasal sinuses Nasopharynx, oropharynx and larynx • The Nasopharynx • Oral cavity and pharynx • The Larynx • The neck

The introduction of cross-section imaging, notably computed tomography (CT) and magnetic resonance imaging (MRI) has hugely expanded the scope and ability of head and neck radiology to recognize and assess disease. The anatomy of the head and neck is hugely complex; pathological processes make the subject even more complex. Only advanced imaging can make a material contribution to understanding. Imaging investigations involved in the assessment of ear, nose and throat (ENT) cases include: • plain radiographs • conventional tomography • contrast medium examinations (barium swallow, etc.) • ultrasound • radionuclide radiology • angiography • CT • MRI • positron emission tomography (PET).

Plain radiographs and conventional tomography have virtually no place in the assessment of disease in the modern setting. Contrast medium studies still have a major role in the assessment of disorders of swallowing, particularly those where there is some sort of neuromuscular incoordination (multiple sclerosis, myasthenia, stroke). The use of ultrasound and radionuclides is discussed elsewhere in this book; ultrasound in skilled hands is fast becoming the first line investigation for many conditions. Angiography has a highly specific and limited role.This chapter concentrates on the use of CT and MRI in the assessment of disease. In approaching this subject a number of perhaps unfamiliar clinical concepts have to be considered. The modern ENT clinic is equipped with advanced audiological and electrophysiological equipment. This allows deafness, for example, to be differentiated into conductive, where some problem is dampening the proper mechanical conduction of sound, and sensorineural, where there is a lesion in the electronic pathway – cochlea, acoustic nerve or brainstem. Microscopes for the examination of the external ear and fibre-optic endoscopes for the evaluation of the nose, postnasal space and upper airway are routinely available. In such an arena, the detection of disease is usually simple and pathological characterization has been revolutionized by the use of fine needle aspiration cytology at the time of clinic assessment. The main problem facing the ENT surgeon is then one of extent. The lesion is visible, its histology is known—but just how far does it go? This is the role of the radiologist.

THE EAR THE EXTERNAL EAR Anatomy and physiology The external ear canal is 2–3 cm in length. While the outer one-third is lined by conventional desquamating stratified epi-

thelium, the inner two-thirds is lined by highly specialized and unique skin directly attached to periosteum without an intervening dermal layer.This skin does not desquamate and has no hair; instead it has the unique ability of lateral migration and contains highly specialized wax to assist in toilet.

1410

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Pathology

Osteoma of external ear

In the ENT clinic, otitis externa, characterized by irritation and discharge, rarely requires imaging.Three conditions superficially similar to otitis externa and characterized by stenosis of the external auditory meatus (EAM) are frequently referred for imaging: • squamous carcinoma of external ear • ‘malignant’ otitis externa • osteoma of external ear.

These benign tumours may arise spontaneously, but usually occur in those individuals fond of swimming in cold water. They enlarge slowly and usually present late with conductive deafness when the tumour fills the external meatus. CT demonstrates the typical homogeneous, well-defined dense tumour (Fig. 62.2). These tumours are surgically easy to remove but leave exposed bone in the EAM. Due to the unique nature of the skin of the EAM, it is impossible to cover or graft the bone and soft tissue re-stenosis is an almost certain consequence.

Squamous or basal cell carcinoma Clinical diagnosis is easy; extent needs to be defined by crosssectional imaging. High resolution CT (HRCT) will define the extent of bone erosion or destruction. MRI will allow better delineation of the precise extent of the soft tissue mass. Most of the information can be derived from T1- and T2-weighted images; intravenous contrast medium can add information.

Malignant otitis externa This is a poor descriptive term.The pathology is an osteomyelitis of the outer petrous bone (Fig. 62.1). The word ‘malignant’ was coined because of the high mortality of the condition. A typical patient is elderly and diabetic. Pseudomonas is a typical organism. As the disease spreads, the facial nerve is involved, leading to palsy.

Figure 62.1 Elderly diabetic man with a mass raising pinna. (A) CT: note the large and diffuse soft tissue mass, erosion of the cortical bone of the mastoid and the opacity of the mastoid air cells. (B) CT at level of external auditory meatus (EAM) shows an inflammatory mass filling the EAM but not extending into the middle ear cleft.

Figure 62.2 A 50-year-old male patient with a passion for cold water swimming. A ring of small osteoma can be seen in the external auditory meatus (EAM). The removal of such lesions is fraught with difficulty. The skin of the external ear is not found elsewhere in the body and is impossible to replace. (A) Coronal CT of a superiorly placed small osteoma at the junction of the EAM and tympanic membrane. A more sessile osteoma can be seen inferiorly. (B) Coronal CT also shows an osteoma on the posterior wall.

THE MIDDLE EAR Anatomy and physiology The tympanic membrane separates the middle ear from the external ear. The membrane is tough, normally translucent and held in tension by the action of tensor tympani. A small region, the pars flaccida in the most superior part of the drum, is less rigid than the rest of the drum (the pars tensa) and is of importance in the genesis of keratoma (see below). The middle ear cleft is a small air-containing space (approximately 1 ml), connected to the pharynx by the Eustachian tube. Its anatomy is complex and outside the scope of this book. It contains:

CHAPTER 62

• bones: malleus, incus and stapes • muscles: tensor tympani and stapedius • nerves: facial nerve. Medial wall features include: • promontory of cochlea • promontory of horizontal semicircular canal • round and oval windows • facial nerve. Important anatomical relationships include: • mastoid air cells • lateral venous sinus and jugular bulb • carotid canal • intracranial compartment and temporal lobe.

Pathology Common pathological processes requiring imaging include: • chronic suppurative otitis media and keratoma • otosclerosis • venous sinus thrombosis • intracranial complications.

Chronic suppurtive otitis media and keratoma This is regarded as ‘safe’ if there is no evidence of a bone destructive process and ‘unsafe’ if there is.The former requires no radiological intervention.The latter frequently requires intensive investigation using HRCT (preferably in the coronal plane; Fig. 62.3). The primary problem lies with dysfunction of the Eustachian tube. As a consequence of negative pressure within the middle

•

EAR, NOSE AND THROAT RADIOLOGY

ear, the pars flaccida of the tympanic membrane is drawn inwards. When the epithelium of the pars flaccida can desquamate and be cleared by natural processes of external ear toilet, no problem arises.When this process fails, desquamated skin accumulates and forms a ball of skin—a keratoma (more frequently and incorrectly called a cholesteatoma). This can enlarge and cause bone destruction.

Otosclerosis This is divided into two types depending on whether the hearing loss is conductive or sensorineural: Fenestral—This is a process by which the stapes foot plate is fused to the oval window, preventing proper transmission of sound and leading to conductive hearing loss. It can be detected by HRCT (Fig. 62.4). Retrofenestral—This is more properly an osteoporosis of the dense bone around the cochlear and labyrinth, leading to sensorineural hearing loss. Again, it can be demonstrated by HRCT.

Venous sinus thrombosis The lateral venous sinus is an immediate relation of the posterior wall of the mastoid. Posterior extension of sepsis can cause venous thrombosis with consequent intracranial complications, e.g. haemorrhagic infarction.

Intracranial complications These complications are mainly those of sepsis. Infection can spread to cause:

Figure 62.3 Chronic suppurative otitis media. The surgeon sees the ‘tip of the iceberg’ when he looks into the external auditory meatus (EAM). (A) Coronal CT showing erosion of horizontal semicircular canal (HSCC, arrow), with normal (B) for comparison. A large mass can be seen filling the attic. It has eroded into the HSCC, destroyed the ossicles and partially covers the oval window. (C) Coronal CT of erosion of the lateral wall. By comparing this with the normal side (D) the destruction of the lateral attic wall can be easily appreciated.

1411

1412

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.4 A patient with conductive hearing loss. By comparing the normal (A) and abnormal (B) sides on CT, the narrowing of the oval window can be clearly appreciated. This results in fixation of the stapes footplate and conductive hearing loss.

• extradural empyema • subdural empyema • cerebral abscess.

• cochlear implantation • bony dysplasias.

Acoustic neuroma

THE INNER EAR Anatomy and physiology Sound is transmitted from the tympanic membrane, via the ossicles, to the oval window and thence to the perilymph. Mechanical vibrations pass to the apical turn of the cochlear and return via the endolymph to the round window. As the ratio of the area of the tympanic membrane-to-oval window cross-section is 20:1, there is mechanical amplification. Hair cells in the cochlea respond to specific frequencies. Frequencies up to 20 kHz are perceived in the basal turn, down to 20 Hz in the apical turns. The range of intensity of sound perception is huge and is expressed in a logarithmic decibel scale of 0–120 dB. As an illustration, a quiet whisper is 30 dB, a lawnmower runs at 90 dB and a jet plane at takeoff is 120 dB. The electrical output of the hair cells passes to the spiral ganglion contained in the cochlea and then in the cochlear nerves (part of the acoustic nerve) to the brainstem. Within the brainstem, nerve fibres synapse and pass in ipsi- and contralateral pathways to the medial geniculate body and then on to Heschel’s gyrus in the temporal lobe. The semicircular canal system consists of three rings at right angles to each other, and contains fluid which stimulates hair cells. These cells continue to form the superior and inferior vestibular nerves. They pass to the brainstem and are involved in reflexes involved in the control of balance via the flocculonodular lobe of the cerebellum.

Pathology Common disease processes requiring imaging include: • acoustic neuroma • trauma • glomus tumours • facial palsy • congenital malformations

When these occur sporadically they are technically schwannomas. When bilateral, they indicate a diagnosis of neurofibromatosis type II. Any patient with asymmetric sensorineural hearing loss or tinnitus could have an acoustic neuroma. However, in the vast majority the nerve has simply ceased to function. In the small minority with a tumour (perhaps 1 in 40), the size of the mass bears little relationship to the degree of deafness. Indeed a patient with a large tumour may present with signs of brainstem distortion before any features of deafness are present. This is a classical dilemma in ENT practice. A patient with asymmetrical sensorineural hearing loss probably has nothing, but possibly has a tumour which could range in diameter from a few millimetres to a few centimetres. The wider availability of MRI has solved the clinical dilemma—high-resolution MRI can detect the smallest tumours and exclude lesions in the vast majority.The logistical and cost implications of this are, however, highly significant, especially when the yield (often < 1:40) is so small. The choice between high resolution T2-weighted images and gadolinium-enhanced T1-weighted examinations as the optimal method of detection of small intracanalicular tumours remains a subject of much debate (Table 62.1). A contrastenhanced study will detect tiny (1 mm) tumours missed by even the best high resolution T2 study, but their significance remains debatable (Fig. 62.5).

Trauma Skull base fractures involving the petrous bone are uncommon, but are very important to identify because: • there may be an associated CSF leak • the facial nerve may be damaged • the ossicular chain may be disrupted. HRCT allows accurate detection and delineation of fractures. If this is performed, the classical discrimination into longitudinal and transverse subtypes becomes less common because

CHAPTER 62

•

EAR, NOSE AND THROAT RADIOLOGY

Table 62.1 ADVANTAGES AND DISADVANTAGES BETWEEN HIGH RESOLUTION T2-WEIGHTED AND CONTRAST-ENHANCED T1-WEIGHTED MRI

For Against

High resolution T2

Contrast-enhanced T1

Inexpensive

Expensive (cost of contrast medium and additional examination time)

Single sequence

Three sequences pre- and post-contrast and T2 weighted

May miss tiny lesions

Requires injection of contrast medium Requires cover for possible reactions

Figure 62.5 Acoustic neuromas. T2-weighted MRIs showing a small intracanalicular (A) and a larger cerebellopontine angle acoustic neuroma (B).

most are, in fact, complex. The fracture line almost inevitably takes a complex course through a complex bone!

Glomus tumours Glomus jugulare tumours arise from chemoreceptor cells in the jugular bulb. Glomus tympanicum tumours arise close to the tympanic membrane. They both present clinically with tinnitus and as masses in the inferior aspect of the tympanic membrane. Once again, there is the classical clinical dilemma—is this tumour confined to what can be seen down an otoscope or is it the tip of a much larger tumour arising from the jugular bulb and spreading in the skull base? CT and MRI are complementary (Fig. 62.6). CT assesses the bone destruction in the margins of the jugular bulb; fatsuppressed contrast-enhanced MRI assesses the tumour itself. Great care must be taken in the assessment of these lesions. There is a large amount of normal variation in the jugular bulb in terms of size and shape. There is complex turbulence of blood in the jugular bulb which may give rise to artefact on MRI. A glomus tumour is a vascular tumour within a vascular structure and errors of commission and omission can easily be made.

Facial palsy Facial nerve palsy is an extremely unpleasant disorder for the patient. Apart from the problems associated with dribbling and spilling of food and drink, the loss of facial animation can lead to significant social and employment problems. Bell’s palsy is frequently seen clinically but is uncommonly imaged. A typical patient develops a sudden facial paralysis which recovers fully or incompletely after 2–3 months. If imaging is undertaken, pathological enhancement of the nerve is well described. Imaging is mandatory for atypical cases or where there is some unusual historical or clinical feature. High resolution MRI studies may reveal any number of lesions affecting the nerve from without. Intrinsic nerve lesions may demonstrate nerve swelling, signal change on T2-weighted sequences or contrast enhancement.

Congenital malformations Malformations of the cochlea and/or labyrinth may be isolated or part of a more widespread craniofacial syndrome.There may be associated abnormalities of the middle ear, external ear, or pinna. A discussion of this extremely complex subject is outside the scope of this chapter. If a congenital malformation is suggested, it is best to go through a checklist of potential sites

1413

1414

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.6 Glomus jugulare tumours. (A) Note the small soft tissue mass in contact with the inferior aspect of the tympanic membrane on coronal CT (arrow). This cannot be separated from the jugular bulb and there is loss of clarity of the bony wall of the bulb. (B) On axial CT note the lack of clarity of the bony margin of the jugular foramen. (C) The small mass projecting into the hypotympanum can be clearly seen on axial CT. (D) The high signal of the tumour can be clearly seen on axial T2-weighted MRI. Note the normal petrous bone on the other side and which is essentially black on T2-weighted images (bone, air and flowing jugular bulb blood are all black on T2-weighted images). (E) The small projection into the hypotympanum can be readily appreciated on axial T2-weighted MRI.

of abnormality—congenital malformations are rarely solitary if you look hard enough. HRCT is the investigation of choice (Fig. 62.7) and high resolution MRI is complementary. Axial (±coronal) images allow accurate description of the abnormalities, which may include: • cochlea: structure of basal turn; number of turns in modiolus • vestibule: enlargement • semicircular canal: absence; morphological abnormalities

• • • • •

oval and round windows: position; orientation ossicles: fusion; malposition middle ear cleft: orientation; ventilation internal and external meatus: orientation; calibre position of jugular bulb: enlarged or absent; intact or dehiscent.

While it is possible to offer an eponymous title for some of the more common defects, it is preferable for the nonexpert to leave the description in plain English, for all to understand.

Figure 62.7 Congenital malformation of the ear. (A) Axial CT at same level as normal side. The whole of the external ear has failed to develop and is seen as a solid mass of bone. The middle ear is poorly developed and abnormally ventilated. Note also the high jugular bulb. (B) CT of normal anatomy for comparison. Note the well-developed and ventilated external auditory meatus and the cellular and ventilated mastoid.

CHAPTER 62

Cochlear electrode implantation Implantable cochlear electrodes offer a chance of hearing for some individuals, typically those whose hearing has been damaged by childhood meningitis. An array of electrodes (typically 16) is surgically introduced into the cochlea and then connected to a very complex ‘black box’ which is capable of spectral and volume analysis of incoming sound. After effort on behalf of the patient, good hearing can be ‘learned’. Assessment of the patency of the lumen of the cochlea is an essential pre-operative requirement. HRCT will assess bony patency

•

EAR, NOSE AND THROAT RADIOLOGY

and high resolution MRI will then exclude fibrous adhesions within the cochlea, which might prevent passage of the electrode array.

Bony dysplasias The petrous bone, like any other bone, may be involved in dysplasias of which the most common is fibrous dysplasia. In this disorder, the characteristic change in texture with mild expansion is easily diagnostic on HRCT. It can be mimicked by the hyperostosis induced by an intracranial meningioma.

THE NOSE AND PARANASAL SINUSES Anatomy

Physiology

The external nose consists of bone superiorly and a cartilaginous structure inferiorly. Arterial blood supply is derived from facial and ethmoidal arteries and may bleed profusely after trauma. Venous drainage into the angular vein in the medial canthus of the eye explains how nasal sepsis can spread to involve the cavernous sinus and how damage to this results in profuse subcutaneous bleeding around the eye—a ‘black eye’. The integrity of the anterior nasal cavity is maintained by the cartilaginous septum which is supported more posteriorly by the vomer. Posteriorly the cavity continues in the midline on either side of the septum. The medial nasal wall is thus the septum. The lateral nasal wall is complex and supports the three turbinates (superior, middle, inferior) and their associated airway or meatus. Thus the middle meatus lies inferior to the middle turbinate of the lateral wall of the nose. The middle meatus is functionally the most important. It receives mucosal drainage from all the paranasal sinuses except the posterior ethmoid and sphenoid sinuses which drain into the sphenoethmoid recess. The introduction of rigid and flexible endoscopes enables these areas to be directly visualized by the ENT surgeon and has removed the need for plain radiography. The lining of the nose is a pseudostratified ciliated columnar epithelium common to the rest of the respiratory tract. A specialized sensory epithelium lies on either side of the septum immediately beneath the cribriform plate. The neurones are specialized nonmyelinated fibres communicating with the olfactory bulbs superiorly.The most common cause of anosmia (loss of smell) is lack of ventilation of this specialized mucosa by excessive mucus. The paranasal sinuses (frontal, ethmoid, maxillary and sphenoid) are similarly lined by a ciliated epithelium containing numerous goblet cells producing mucus. Mucus is moved from the nose and paranasal sinuses by two mechanisms: 1 Ciliary action—may be absent in congenital disorders such as Kartagener’s syndrome, or after inhalation of toxins such as cigarette smoke. 2 Slime trails—mucus is swept into ‘rivers of slime’ by the action of cilia. Such rivers are visible by nasoendoscopy and pass into the throat.The action of swallowing ‘tugs’ the river of slime and assists nasal clearance of mucosa.

The nose and nasal airway serve a number of functions: • sense (smell) • respiration • air conditioner • immune response to antigens • sound quality.

Smell The sense of smell adds enormously to the quality of life. Smell is a major feature of taste. Anosmia follows when the olfactory mucosa is unventilated (common) or rarely if there is a tumour disturbing the nerves (carcinoma, esthesioneuroblastoma [Fig. 62.8], meningioma).

Respiration The newborn baby is an obligate nasal breather. If there is choanal atresia, the baby may not be able to survive. This reliance on nasal respiration continues into adult life. In adults, mouth breathing is usually only necessary during exertion.

Air conditioning The air conditioning function of the nose involves: • Heat exchange—cold inspired air is warmed; warm expired air transfers its heat back into the nasal mucosa. • Humidification—dry air is almost 100% saturated with water vapour by the time it reaches the alveoli. • Cleaning—larger particles are removed; small dusts pass into the lung. The key to these functions is the mucous blanket. Mucous production and movement is essential for proper nasal function.

Immune response Antibodies in the nasal mucosa are in the first line of defence against air-borne pathogens. Over-reaction to allergens (as in hayfever or asthma) is uncomfortable and unpleasant and leads to mucosal thickening, oversecretion of mucus and airway obstruction.

Speech Listening to anyone with a cold illustrates the importance of the nose in the quality of speech. The nose and sinuses act as a resonant chamber.

1415

1416

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.8 Esthesioneuroblastoma. (A) Contrast-enhanced sagittal T1-weighted MRI of mass based on cribriform plate. Note the extension inferiorly into the nose and superiorly into the brain. (B) Coronal CT shows the bony destruction of the cribriform plate and a soft tissue mass in the ethmoid. The differential diagnosis would include an ethmoid carcinoma.

Radiology and pathology Plain radiographs are of assistance in the detection of gross disease, but interpretation is fraught with difficulty due to the great variation in normal appearance of the paranasal sinuses, and the presence of so many complex overlapping threedimensional structures in a two-dimensional image. This is a classical example of a simple and widely available investigation which requires major expertise for correct interpretation. In any event, the advent of nasoendoscopy has allowed the surgeon direct visual access to structures hitherto invisible and has further reduced the need for plain radiographs. CT and MRI make a large and complementary contribution to the determination of the extent of disease.

Rhinosinusitis This is an extremely common condition which is usually treated medically. Radiological investigation is rarely required unless surgical intervention is contemplated. There are a number of common causes: • allergic—very common. Specific allergens such as pollen precipitate an immune response resulting in mucosal swelling. This may develop into allergic polyposis • vasomotor—a disorder of autonomic regulation of nasal mucus production • infective—all will be aware of the symptoms of the common cold • mechanical—as in deviation of the nasal septum • ciliary disorders—Kartagener’s syndrome • iatrogenic—overuse of nasal decongestants can result in rebound hyperaemia and finally atrophic rhinitis. A proper understanding of nasal physiology and the development of minimally invasive surgery has changed the surgical approach away from invasive and destructive treatments (e.g. radical antrostomy—Caldwell Luc) towards logical approaches to the osteomeatal complex, such as uncinectomy to widen the ostium of the maxillary antrum. Pre-operative assessment of the bony anatomy of the nose is an essential prerequisite, best performed with coronal CT. This assessment should include:

• Identification of congenital variation: deviated nasal septum; hypoplasia and enlargement of normal structures; anomalous air cells, e.g. Haller, Ager nasi. • Identification of extent of disease: which sinuses are involved, which spared; is the osteomeatal complex involved?; is the sphenoethmoidal recess involved?; does disease extend into orbit or cranium? • Identification of bone destruction—this may indicate malignancy • Identification of complications: orbital or intracranial abscess. Almost all this information can be obtained from CT. Given the high inherent contrast in the paranasal sinus system, a low dose technique may be employed and, following the introduction of spiral CT, this investigation is commonly performed at an axial volume acquisition, subsequently reconstructed to give coronal images. However, the sphenoethmoidal recess and the posterior wall of the frontal sinus are best examined in the axial plane. Complications or suspected malignant disease may well require further contrast-enhanced studies employing higher radiographic dose rates to improve soft tissue contrast. Common problems requiring imaging include: • nasal polyposis • antrochoanal polyp • mucocoeles • fractures • epistaxis • nasal and paranasal tumours.

Nasal polyposis (Fig. 62.9) This is a common condition in adults. If it is seen in children, cystic fibrosis is a possible cause, and midline congenital anomalies such as meningocoele or encephalocoele need to be considered and excluded. The aetiology of polyposis is uncertain, but allergy is important. Polyps can be controlled by steroids, but surgery is frequently required. Such polyps can usually be resected using an endoscope, in which case pre-operative assessment as above is sufficient.

CHAPTER 62

•

EAR, NOSE AND THROAT RADIOLOGY

Figure 62.9 Severe nasal polyposis and airway obstruction. Four consecutive CT images, running posterior to anterior. (A) Level of posterior choana. Note the complete opacity of the normal airway. Note also marked streak artefacts from dental fillings. (B) Level of the ostia of the maxillary antra. Note again the complete opacity of the nasal airway. The ostea are widened by benign polyps and the nasal turbinates are partially eroded. (C) Level of the anterior nasal airway. Again there is complete obliteration of the airway by polyps. Note the thinning of the bone of the ethmoids. (D) Anterior nose. The polyps protrude from the sinuses into the anterior nose.

Antrochoanal polyp This is a special unilateral polyp which arises in the maxillary antrum, passes out through the ostium which it enlarges, and then projects backwards into the postnasal space. It causes unilateral nasal obstruction. The radiological hallmark is the enlarged ostium (Fig. 62.10).

Mucocoeles A normal sinus produces mucus, which clears by gravitational drainage and ciliary action. If the sinus ostium becomes blocked and infection does not supervene, the sinus fills with mucus.The mucus acts then as a slow-growing mass lesion, expanding the sinus and thinning the sinus wall—a mucocoele. These lesions are painless and give rise to little or no symptoms until the mass effect becomes a critical issue. Posterior ethmoid mucocoeles may encroach upon the optic nerve and lead to visual failure.

Frontal or anterior ethmoid mucocoeles may extend into the orbit and give rise to proptosis. Diagnosis is best made with CT, which will show the characteristic sinus expansion with a very thin membrane of bone surrounding it. Mucocoeles are usually sterile, but can become infected. This is frequently a dramatic clinical presentation with rapid onset of pain and fever. Clearly an infection can rapidly spread to adjacent tissues and urgent surgical drainage is essential.

Fractures Nasal bones Fractures of the nasal bones are common sequelae of fights. Plain radiographs are sufficient to document the injury, although they are frequently not required in simple fractures.

1417

1418

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.10 Antrochoanal polyp. Four consecutive CT images, running posterior to anterior. These benign lesions arise in the maxillary antrum close to the osteum, pass through the osteum into the middle meatus then turn 90 degrees to run posteriorly along the middle turbinate to the posterior choana. (A) Level of posterior choanal shows the posterior extent of the normal turbinates on the left side and the polyp on the right side. (B) More anterior image showing polyp intimately related to turbinates. (C) Polyp in the maxillary antrum, widening osteum, lying between middle and inferior turbinates. (D) More anterior image showing polyp traversing antrum and nasal airway.

Zygoma, orbit and middle face fractures These are beyond the scope of this chapter. Such fractures are best assessed by HRCT.Three-dimensional CT is of great value to the surgeon planning reconstruction of complex facial fractures.

• • • •

‘Blow-out’ fractures Fist or small high velocity (e.g. squash) ball injuries to the orbit raise the intra-orbital pressure and lead to fracture of the orbital floor. The orbital fat, inferior oblique and inferior rectus muscles will prolapse into the maxillary sinus and become trapped (Fig. 62.11).

Severe uncontrolled epistaxis may be life-threatening and requires drastic intervention. Contrast angiography and selective embolization of bleeding vessels may be life-saving. If such a service is available, the old-fashioned ligation of the external carotid artery will not be required.

Epistaxis does not usually require radiological assessment. If the bleeding is profuse or recurrent, then a source for the bleeding may require investigation. Causes include:

Nasal and paranasal tumours

bleeding polyps. juvenile angiofibroma (male preponderance). carcinoma or other malignancy: nose, sinuses, nasopharynx. Wegener’s granuloma.

Clinical presentations include facial deformity (with or without pain), nerve involvement leading to pain, dysaesthesia and

CHAPTER 62

•

EAR, NOSE AND THROAT RADIOLOGY

Inverting papilloma These frequently present as a unilateral nasal polyp causing obstruction and epistaxis. They are locally invasive tumours and require full surgical excision. Pre-operative assessment is usually satisfactory with CT, although MRI can provide additional information.

Figure 62.11 ‘Blow-out’ fracture. Coronal CT shows normal orbital floor and ventilated antrum on left, depressed floor and blood in antrum on right. The orbit shows some opacity of the fat adjacent to the fracture due to oedema and haemorrhage. The inferior rectus is swollen but remains in the orbit. Untreated, this patient is at risk from disfiguring enophthalmos.

anaesthesia and loss of function (e.g. airway obstruction). The pathological processes include a number of very uncommon disorders which may confuse a pathologist, let alone a radiologist. The radiologist’s job is one of defining extent rather than predicting histological diagnosis. Tumours vary from indolent through to very malignant. Common indications for radiological imaging include: • osteoma • fibro-osseous lesions • inverting papilloma • juvenile angiofibroma • granulomatous conditions • malignant tumours. Osteoma (Fig. 62.12) These are the most common benign tumour. They can arise anywhere but are most frequent in the frontal sinus. They are frequently incidental findings and are usually isolated. They can be part of a genetic disorder such as Gardner’s syndrome. Fibro-osseous lesions (Fig. 62.13) This is a spectrum of pathology with a purely fibrotic lesion at one end and a dysplastic bony lesion at the other. These lesions may produce considerable facial deformity and be very difficult to treat.

Juvenile angiofibroma Adolescent boys with heavy epistaxis characterize the disease. The tumour may be very fibrous in some individuals, more angiomatous in others. HRCT allows accurate diagnosis since these tumours start in the pterygopalatine fissure and enlarge this before they extend.The presence of a nasal mass and a widened fissure is therefore pathognomonic of the condition. It is important to stress that these tumours may spread across the skull base and a combination of CT (to assess bone destruction; Fig. 62.14A) and contrast-enhanced MRI (to assess soft tissue extent; Fig. 62.14B–E) may be required to provide accurate pre-operative assessment. Angiography is not required to establish the diagnosis, but has a valuable role in pre-operative therapeutic embolization to reduce blood loss (Fig.62.14F). Nasal and paranasal sinus granulomas The precise nature of the various granulomas is a matter for a pathologist. These lesions behave in a similar fashion to malignant tumours, but are characterized by small volume destructive processes leading to loss of much of the nasal bony architecture. There may, of course, be considerable coincident systemic disturbance. Malignant tumours The types of tumour are listed above in Table 62.2. It is rarely possible to differentiate between the histological types on imaging. The most common type is squamous carcinoma, followed by adenocarcinoma, adenoid cystic carcinoma and melanoma. Lymphoma can occur as a primary tumour or can occur as a metastasis from a primary lesion elsewhere. As always, the role of radiology is that of defining the extent of disease. Radiological reports should deal obsessively with detail of spread in all planes and with structures involved or destroyed.This level of detail is absolutely necessary if the correct decisions are to be taken about whether a surgical approach is appropriate and, if it is appropriate, exactly which approach is required. The radiological report forms the basis for proper planning and informed consent to proposed surgery. CT and MRI are complementary. CT assesses bone destruction. Multiplanar MRI assesses the soft tissue component more comprehensively and has a major value to the surgeon planning an operation or the radiotherapist planning a portal (Fig. 62.15). Contrast enhancement may be of little help with the definition of the extent of the tumour mass, which is usually easily detected by standard T1 and T2 sequences, but is of enormous value in the detection of spread along nerves.

1419

1420

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.12 Osteoma. (A) A rare use for a plain radiograph. The osteoma can be clearly appreciated, but the image gives no details of its involvement with the intracranial space or orbit. (B,C) The coronal STIR images underestimate the lesion since bone and air are both black on T2-weighted MRI. (D) The axial CT shows extension into the orbit.

Figure 62.13 Fibro-osseous lesions. (A) The milky texture of bone typical of fibrous dysplasia is seen in this CT. (B) Fibrous dysplasia is a lesion which expands bone—note the compression of the pterygopalatine fissure.

CHAPTER 62

•

EAR, NOSE AND THROAT RADIOLOGY

Figure 62.14 Juvenile angiofibroma in a teenager with epistaxis. (A) This axial CT was the first study performed in this patient. The critical finding is widening of the pterygopalatine fissure on the left side. This is virtually diagnostic of an angiofibroma. (B–E) Contrast-enhanced (gadolinium) T1weighted MRIs. (B) Axial image revealing an enhancing, well-defined mass in the nose. Clinically this was visible from an anterior view through the nose and a posterior view with a mirror in the nasopharynx. (C) Axial image. Note the lateral extension of the tumour towards the infratemporal fossa. (D) Axial image. More superiorly the tumour is less well defined and more infiltrative. Note the extension into the pterygopalatine fissure is visible but more easily seen on CT. (E) Coronal image revealing the mass at the level of the posterior choana with its lateral extension. This is an essential observation for surgical planning. (F) Angiography. Lateral superselective injection into the maxillary artery shows the highly vascular nature of the tumour. This investigation was performed before therapeutic embolization and subsequent surgical resection.

Table 62.2 BENIGN AND MALIGNANT NASAL AND PARANASAL SINUS TUMOURS Benign Epithelial tumours

Malignant

Papilloma

Squamous carcinoma

Adenoma

Adenocarcinoma

Inverting papilloma

Melanoma Adenoid cystic carcinoma Malignant salivary tumours

Mesenchymal tumours

Osteoma

Osteogenic sarcoma

Ossifying fibroma complex

Angiosarcoma

Angiofibroma

Chondrosarcoma

Chondroma

Lymphoma

Fibrosarcoma

Rhabdomyosarcoma

1421

1422

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.15 Delineation of tumour extent. The patient complained only of a blocked nose. It is the role of radiology to define extent; consequently sagittal, axial and coronal imaging all have a role to play. If contrast medium is given, it is best to employ a fat-saturation T1-weighted MR technique. (A) Sagittal T1-weighted gadolinium-enhanced MRI. This demonstrates the anterior–posterior extent of the tumour which can be seen obstructing the frontal air sinus anteriorly and the sphenoid sinus posteriorly. (B) Coronal T1-weighted gadolinium-enhanced MRI. Tumour extends laterally to bulge into the maxillary antrum. Superiorly it has invaded into the anterior cranial fossa, but not through dura. Superolaterally, the lamina papyracea of the orbit is breeched. Inferiorly the tumour is in contact with the palate. Medially the tumour has not extended far across the midline. (C) Axial T2-weighted MRI. The complex cystic and solid nature of this tumour can be clearly seen and can be easily differentiated from oedematous mucosa and obstructed sinus. (D) Axial T2-weighted MRI. The extension through the medial wall of the maxillary antrum can be clearly appreciated.

NASOPHARYNX, OROPHARYNX AND LARYNX The ‘head and neck’ is a loose term to describe a specialty dealing with malignant disease of the nasopharynx, oropharynx and larynx. A multidisciplinary team approach including resective and reconstructive surgeons, oncologists, radiologists and radiotherapists is essential. Benign disorders usually fall into the remit of the ENT surgeon but are conveniently included here for completeness sake and to stress that there is frequently a benign differential diagnosis even when malignant disease is suspected.

THE NASOPHARYNX This is a convenient anatomical description of that area of the neck bounded by the skull base superiorly and the palate inferiorly, containing the nasopharyngeal airway in its centre. It is usefully divided into a number of compartments by very complex fascial planes. Each compartment initially contains pathological processes but eventually spread beyond the fascial plane will occur.

CHAPTER 62

It is beyond the scope of this chapter to address the issue of the anatomy (and some of the relevant pathology) of the compartments precisely, but it is convenient to list them and their contents: • mucosal: stratified squamous epithelium; small salivary glands; pharyngeal constrictor muscles; levator palatini muscle • parapharyngeal: parapharyngeal fat • retropharyngeal • prevertebral • masticator: muscles of mastication; mandible and teeth • carotid: carotid artery; jugular vein; vagus and glossopharyngeal nerves; sympathetic trunk • parotid: parotid gland; retromandibular vein; facial nerve.

Pathology Mucosal and parapharyngeal spaces—nasopharyngeal carcinoma Although this is easily the most important of the ‘head and neck’ cancers, it is sadly usually a large tumour at the time of diagnosis. The tumour is able to infiltrate the parapharyngeal fat ‘silently’ and reach large size without symptoms. Both CT and MRI demonstrate the lesion well (Fig. 62.16). Surgery is usually not an option for these tumours and the main thrust of treatment is therefore radiotherapy. As ever, the main role of radiology is in the definition of extent for radiotherapy planning.

Masticator space Asymmetry The development of an asymmetrical face is usually consequent upon growth of a tumour or hypertrophy of muscle. Occasionally atrophy of one side is misinterpreted as overgrowth of the other.Atrophy of muscles of mastication is consequent upon disturbance of their motor innervation—the trigeminal nerve. Sepsis Most sepsis arises around the teeth and is clinically very easy to identify and treat. Occasionally a patient may present

•

EAR, NOSE AND THROAT RADIOLOGY

with trismus and little other symptomatology, when imaging may be diagnostic (Fig. 62.17). Tumours of the muscles of mastication are rare, but are most frequently rhabdomyosarcomas.

Carotid space Vagal neuroma Tumours of the vagus nerve are usually schwannomas and are benign. They may reach a large size due to their very slow rate of growth. These tumours can be identified from their origin within the carotid sheath—they spread the carotid and jugular vessels apart. These tumours are characteristically of high signal on T2-weighted MRI (Fig. 62.18) and enhance avidly after intravenous contrast medium on both CT and MRI. Paraganglionoma Tumours of the chemoreceptor cells have been mentioned above with respect to glomus jugulare tumours. Such tumours can also arise in the carotid body— carotid bulb tumours. In 10% of cases, paraganglionomas are multiple and the neck should always be examined when a glomus jugulare tumour is discovered and the skull base if a carotid bulb tumour is discovered. Vascular anomaly The carotid artery is a common site for atheroma. Occasionally carotid aneurysms may develop spontaneously or as consequence of disease. Large aneurysms may develop layered thrombus and appear very complex on cross-sectional imaging (Fig 62.19), leading to misinterpretation as neoplasms. Lymph node enlargement The carotid sheath is invested with a double layer of fascia and is resistant to invasion by local disease. Lymph nodes within the sheath can enlarge as part of systemic disease (e.g. lymphoma) or when the sheath is directly invaded by tumour.

Figure 62.16 Nasopharyngeal carcinoma. (A,B) Axial T1-weighted MRIs show a large parapharyngeal mass in contact with the deep lobe of the parotid laterally and partially encasing the carotid.

1423

1424

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 62.17 Sepsis. (A,B) Young woman with severe trismus. (A) Axial T2-weighted MRI shows major asymmetry between the masticator space contents on the two sides. The masseter and pterygoid on the right are normal, but on the left there is increased signal reflecting oedema of the muscles. (B) Axial T1-weighted MRI at the same level shows enlargement and reduced density of the muscles on the left side reflecting increased tissue water and muscle swelling. (C) Sepsis following incomplete treatment of a tooth infection. Coronal T1-weighted MRI after gadolinium administration. The whole of the muscle complex on the left side enhances profoundly after administration of contrast medium.

Parotid space—tumours The majority of parotid tumours are benign and are pleomorphic adenomas. Twenty per cent are malignant and are a heterogeneous group of carcinomas, lymphomas, adenolymphomas, etc. As always the role of radiology is in the definition of extent, not histology. Tumours of the deep lobe of the parotid deserve special mention. As they enlarge they spread medially into the parapharyngeal fat, and then are sometimes difficult to differentiate from laterally spreading parapharyngeal tumours. The most crucial structure for the surgeon is the facial nerve which branches within the parotid before innervating the muscles of the face.The relationship of pathology to nerve

is of prime importance if the nerve is to be identified and preserved. Unfortunately, it is rarely possible to identify the nerve using current imaging techniques.

ORAL CAVITY AND PHARYNX The oral cavity includes: • teeth (upper and lower jaw) • bucco-alveolar sulcus (between teeth and cheek) • tongue • floor of mouth • hard palate • salivary gland openings.

CHAPTER 62

•

EAR, NOSE AND THROAT RADIOLOGY

The oral cavity and pharynx have obvious roles in respiration, phonation and mastication. Special function resides in taste (posterior tongue via the glossopharyngeal nerve) and swallowing.

Swallowing

Figure 62.18 Vagal neuroma. Axial T2-weighted MRI shows a welldefined mass of high signal intensity on the right side just posterior and lateral to the flow void of the carotid artery. This is a classical appearance for a vagal neuroma.

The anterior aspect of the tonsil marks the start of the pharynx. The epithelium is a stratified squamous epithelium containing numerous minor salivary glands. Lymphatic drainage is to the ipsilateral nodes of the internal jugular chain with the exception of the anterior mouth which can drain bilaterally. At the junction of the oral cavity and pharynx there is a complete ring of specialized lymphoid tissue—Waldeyer’s ring—which includes: • palatine tonsils (the tonsil) which lie adjacent to the posterior one third of the tongue • adenoids (pharyngeal tonsils) • lingual tonsils. Sensation is derived from the mandibular branch of the trigeminal nerve anteriorly and the glossopharyngeal nerve posteriorly. Motor supply to the tongue is via the hypoglossal nerve.

This is a complex process involving mastication and preparation of food in the mouth under volitional control, followed by a very rapid nonvolitional reflex phase initiated by the elevation of the tongue forcing the food bolus to the pharynx. The reflex includes: • switch from respiration to swallowing: stop breathing; closure of nasopharynx by elevation of soft palate; closure of airway by action of epiglottis • elevation of larynx • constriction of pharyngeal constrictor muscles projecting bolus into the region of the pyriform fossa • relaxation of the cricopharyngeus muscle • oesophageal peristalsis.

Pathology of the oral cavity Peritonsillar abscess—quinsy Usually secondary to streptococcal infection, an abscess here leads to severe pain and trismus. Clinical inspection and needle aspiration/formal drainage usually settles diagnosis and imaging is not required. Imaging with CT or MRI defines the extent and degree of loculation if required.

Carcinoma of the tongue Lesions may be so small that they merely represent enlarged aphthous ulcers, or so large that they infiltrate the tongue widely. They are readily diagnosed by direct inspection and biopsy. The extent is best defined by MRI. Small lesions are simply resected or ablated. Large lesions will require full glossectomy or hemiglossectomy. The latter is a much less disruptive surgical intervention. The tongue can function well from muscular action on one side against a myocutaneous flap replacement. Critical observations include: • Has the midline been crossed? • Does the tumour erode or infiltrate the mandible?

Figure 62.19 Carotid aneurysm. (A) Axial T1-weighted MRI in this patient who presented with a mass in his neck and had two percutaneous biopsies before the true nature of the pathology was realized! (B) Axial T1-weighted MRI at nasopharynx level shows a low signal intensity aneurysm in the lumen and organized thrombus in the wall. (C) Axial T2-weighted MRI close to the skull base shows the lumen reducing as aneurysm approaches the carotid canal.

1425

1426

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

• Does the posterior aspect of the tumour involve the root of the epiglottis and prevent its proper function? A positive answer to these questions will result in radical surgery or radiotherapy.

Pathology of the pharynx Carcinoma of the pharynx These tumours are divided by their anatomical origin, e.g. pyriform fossa carcinoma or supraglottic carcinoma.All tumours infiltrate adjacent structures and along epithelial surfaces. The anatomy of the space between larynx and posterior tongue is difficult and surgical resection and reconstruction may be impossible. The integrity of the epiglottis mechanism is of paramount importance if food or fluid aspiration is to be avoided. This region requires particularly careful assessment so that the correct surgical and oncological decisions can be made.

THE LARYNX Anatomy and physiology The posterior walls of the pharynx are continuous with the larynx. It has two critical functions: 1 Airway protection—the action of the anteriorly situated epiglottis closes the tracheal airway and prevents aspiration of food and fluid.

2 Sound generation—the vocal cords act in a way similar to a ‘reed’ in a wind instrument, generating vibrations (subsequently modified by the action of tongue, palate, mouth, etc.). The cricoid cartilage is the core structure. It resembles a signet ring with its face situated posteriorly. Attached to this, the V-shaped thyroid cartilage sits with its spine anteriorly and its open face posteriorly. The arytenoid cartilages sit within this structure, and superiorly lies the semicircular hyoid bone which links with the tongue. The vocal cords meet in the midline anteriorly and are controlled by the intrinsic muscles of the larynx—cricoarytenoid abductors and thyroarytenoid adductors. The superior laryngeal nerve—a branch of the vagus—conveys sensation and cricothyroid motor function. All other cord function is conveyed by the recurrent laryngeal nerve, also a branch of the vagus. Lymphatic drainage of the supraglottis is to the upper deep cervical nodes and that of the infraglottis is to the internal jugular, peritracheal and mediastinal nodes.

Pathology Carcinoma of the larynx (Fig. 62.20) Tumours arising on the opposing surface of the vocal cords present early due to the effect on the voice.They are small and can be completely removed. More anterior or lateral tumours

Figure 62.20 Laryngeal carcinoma in a 40 year old male smoker. (A,B) Coronal STIR MRI and (C,D) axial T1-weighted fat-saturated MRI after intravenous gadolinium. The coronal images demonstrate the superior and inferior extents while the axial images define the extent more accurately.

CHAPTER 62

may present late and total laryngectomy or radical radiotherapy may be the only option. Critical observations include: • Has the tumour spread anteriorly and escaped through the notch in the thyroid cartilage? • Has the tumour crossed the midline? • Does the tumour extend into the supraglottis? • Does the tumour extend into the upper oesophagus? • Is there invasion of the cartilage structure of the larynx (thyroid, cricoid, arytenoids)?

THE NECK Anatomy and physiology The neck can be divided into two triangles by the sternomastoid muscle: • anterior triangle: sternomastoid posteriorly to mandible superiorly; midline anteriorly • posterior triangle: sternomastoid anteriorly to trapezius posteriorly; clavicle inferiorly.

Pathology The specific regions of the neck have been covered under the various sections of this chapter. The major subject remaining is that of lymphadenopathy.

Lymphadenopathy Generalized lymphadenopathy is a common clinical presentation and the cause is often never known. If widespread and

•

EAR, NOSE AND THROAT RADIOLOGY

painful, glandular fever or other viral infection can be considered. If painless, spread from head and neck malignancy or lymphoma may be the cause. If focal and painful, a bacterial lymphadenitis may be implicated. If the cause is not immediately evident from clinical inspection and history taking, then imaging frequently has a role to play. Palpable nodes will usually be detected by MRI or CT. Coronal STIR sequences are particularly valuable in MRI and contrast-enhanced thin section CT is equally good or even better at detecting small metastatic lesions. Ultrasound (in expert hands) and fine needle aspiration cytology or coreguided biopsy pushes the detection of small metastatic lesions even further. At present, no imaging technique is able to assess micrometastases and it remains common practice to sample the lymph nodes of the cervical chain at the time of surgical resection of tumour to detect these.

SUGGESTED FURTHER READING Swartz J D, Harnsberger H Ric Head and Neck Imaging, Thieme Chung V F H, Tsai S W Naso-Pharyngeal Carcinoma, Amour Publishing O’Donaghy, Bates G, Narulea A. Clinical ENT, Oxford University Press Laing J Clinical Analtomy of the Nose, Nasal Cavity and Paranasal Sinuses. Thieme Lenz M CT and MRI of Head and Neck Tumours. Thieme Rao V, Flanders A, Tom B, Dunitz M MRI and CT Atlas of Correlative Imaging in Otolaryngology Gulya J, Shucknecht H Anatomy of the Temporal Bone with Surgical Implications. Parthanon Publishing Group

1427

CHAPTER

Dental and Maxillofacial Radiology

63

John Rout and Jackie E. Brown

• • • • • • • • • • • • • • • •

Anatomy of teeth and supporting structures Tooth eruption Disturbances in structure of teeth Dental caries Disorders of the pulp Cysts of the jaws Disease of the periodontium Odontomes and odontogenic tumours Imaging in implantology Disorders of bone Tumours of bone Trauma Radiological investigation of maxillary fractures Developmental abnormalities Salivary gland disorders Soft tissues

Dental radiography is performed using two techniques: intraoral and extra-oral radiography. Diagnosing disorders affecting the teeth requires high-definition images that are obtained by using direct exposure (nonscreen) intra-oral film. Bitewing films show the crowns of the upper and lower posterior teeth on one film and are used for the detection of dental caries and the assessment of periodontal disease. Periapical film shows the tooth in its entirety together with some of the apical bone. Occlusal radiographs are larger than periapical films and are mainly used in orthodontic assessment of the upper anterior teeth and for identifying stones in the submandibular duct. A dental panoramic radiograph demonstrates all of the teeth and jaws on one film and is indicated when a wide coverage is required. The technique is based on a form of tomography with the image layer being curved and corresponding to the shape of the jaws. Accurate patient positioning is essential as otherwise the images are prone to distortion. Cephalometric radiography is used in orthodontic assessment and orthognathic surgery planning. The technique uses

a fixed true lateral or PA head position and a long tube–film distance (approximately 2 m) to reduce image magnification. Increasingly, conventional radiographs are being replaced by digital imaging using phosphor stimuable plates or electronic chips such as charged couple devices (CCDs). Guidelines on the most appropriate choice for dental radiographic examination can be found in ‘Selection Criteria for Dental Radiography’1.

ANATOMY OF TEETH AND SUPPORTING STRUCTURES In the primary dentition, there are normally 20 teeth; in adult dentition there are 32. The teeth are identified using one of two systems illustrated in Table 63.1. These are the Zsigmondy system, which uses single digits for the permanent dentition and letters for the primary (deciduous) dentition, and the Federation Dentaire International (FDI) notation, which assigns double digits for each tooth (Figs. 63.1, 63.2).2 All teeth consist of a crown and a root, which may be single or multiple (Fig. 63.3A, B). The crown is covered with a layer of enamel with a composition of 97 per cent mineral, thus being the most radio-opaque tissue in the body. The bulk of the tooth consists of dentine, which is 70 per cent mineralized. The root is covered by a thin layer of cementum, which has a similar radiodensity to dentine and so is indistinguishable from it. Lying within the centre of the tooth is the radiolucent soft tissue of the pulp, which runs from the pulp chamber in the crown along each root canal to the root apex, through which enter the neurovascular bundles. The tooth is supported in the jaws by the periodontal ligament, which consists largely of collagen fibres and appears as a narrow radiolucent line following the contours of the root. These fibres are inserted into a thin layer of dense bone lining the tooth socket (lamina dura), which appears as a linear radio-opaque structure, and is continuous with the cortical bone of the alveolar crest.

1430

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Table 63.1 THE ZIGMONDY (SINGLE DIGIT) AND FDI (DOUBLE DIGIT) SYSTEMS OF TOOTH IDENTIFICATION Permanent dentition UPPER RIGHT

UPPER LEFT

(18)

(17)

(16)

(15)

(14)

(13)

(12)

(11)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

8

7

6

5

4

3

2

1

1

2

3

4

5

6

7

8

8

7

6

5

4

3

2

1

1

2

3

4

5

6

7

8

(48)

(47)

(46)

(45)

(44)

(43)

(42)

(41)

(31)

(32)

(33)

(34)

(35)

(36)

(37)

(38)

LOWER RIGHT

LOWER LEFT

Primary dentition (55)

(54)

(53)

(52)

(51)

(61)

(62)

(63)

(64)

(65)

E

D

C

B

A

A

B

C

D

E

E

D

C

B

A

A

B

C

D

E

(85)

(84)

(83)

(82)

(81)

(71)

(72)

(73)

(74)

(75)

Figure 63.1 Part of a panoramic radiograph showing the permanent dentition of a normal 18-year-old. The teeth in the upper left quadrant have been numbered 1–8. The third molars are unerupted, incompletely formed and impacted.

Figure 63.2 Part of a panoramic radiograph showing the dentition in a 6-year-old child. The deciduous teeth in the upper left quadrant have been labelled A–E. The first permanent molars (labelled 6) and the lower central incisors have erupted. All four primary first molars and the primary lower second molars are carious.

Figure 63.3 (A) Periapical radiograph labelled to show a tooth and its supporting structures. E = enamel, D = dentine, PC = pulp chamber, RC = root canal, PM = periodontal membrane (periodontal ligament space), LD = lamina dura, MF = amalgam filling. (B) Corresponding line diagram.

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

TOOTH ERUPTION Normal eruption The normal eruption times are shown in Table 63.2. The primary teeth erupt between 6–24 months and the permanent teeth between 6–21 years. Root formation is not complete until 1.5–2 years and 2–3 years after eruption for the primary and permanent teeth, respectively.

Disorders of tooth eruption The commonest cause for failure of full eruption is insufficient room in the dental arch to accommodate the erupting tooth. This particularly affects mandibular third molars and to a lesser extent maxillary canines. Alternatively a tooth may be prevented from erupting by, for example, a tumour, cyst or supernumerary tooth. Delayed eruption occurs in certain endocrine disorders, e.g. hypothyroidism and some genetic abnormalities, e.g. Down’s syndrome. Multiple failure of eruption of the permanent dentition is found in cleidocranial dysplasia (Fig. 63.4).

Figure 63.4 Panoramic radiograph of cleidocranial dysplasia in an adult. There are numerous unerupted teeth including several supernumeraries.

Disorders of tooth development Variation in tooth number Absence of one or more teeth is referred to as hypodontia (oligodontia) and as anodontia (rare) where there is complete absence of teeth. Hypodontia most often affects third molars, mandibular second premolars and maxillary lateral incisors (Fig. 63.5). It is seen in association with cleft lip and palate, Ellis-van Creveld (chrondo-ectodermal dysplasia) and facialdigital syndromes. Marked absence of teeth is seen in hydrotic ectodermal dysplasia. Teeth additional to the normal series, hyperdontia, presents as either supplemental or supernumerary teeth. A supplemental tooth is an extra tooth identical in shape and form

Table 63.2 APPROXIMATE DATES OF ERUPTION OF THE PRIMARY AND PERMANENT TEETH Tooth

Designation

Age(months)

Central incisors

A

6–8

Lateral incisors

B

7–10

Canines

C

16–20

First molars

D

10–14

Second molars

E

20–30

Primary dentition

Permanent dentition

Age (years) 6–7

Figure 63.5 Panoramic radiograph showing a marked example of hypodontia involving both the primary deciduous and permanent dentitions in a 6-year-old child. All four lateral incisors are missing from the primary deciduous dentition and 19 permanent teeth are also absent. Note that wisdom tooth formation normally starts between 9 and 13 years of age.

to an adjacent permanent one. The commonest supernumerary teeth are mesiodens and tuberculates, which form in the maxillary midline. Mesiodens, which are small conical teeth, form after the primary upper central incisors and occasionally erupt. Tuberculate supernumeraries develop shortly after the formation of the permanent upper central incisors and usually impede their eruption. Marked hyperdontia in the permanent dentition is seen in cleidocranial dysplasia (Fig. 63.4).

Variation in tooth size A tooth that is larger than normal is termed a macrodont and when smaller, a microdont, the latter being more common and typically affecting maxillary lateral incisors and upper wisdom teeth. Radiotherapy and/or chemotherapy can affect tooth development, resulting in arrested development such that they appear smaller than normal with short, spiculated roots (Fig. 63.6).

Central incisors

1

Lateral incisors

2

7–9

Canines

3

9–12

Variation of tooth form

First premolars

4

10–12

Second premolars

5

10–12

First molars

6

6–7

Second molars

7

11–13

Third molars

8

17–21

Disturbance of tooth form can affect either the crown or root, or both, and may be developmental or acquired. A dens in dente is due to an infolding or invagination of the enamel into the underlying dentine towards the root, creating the appearance of a tooth within a tooth. A markedly hooked root is called dilaceration and typically affects a maxillary

1431

1432

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Dentine

Figure 63.6 Panoramic radiograph of a child aged 12 who received radiation to the neck for lymphoma when aged 5 years. The extent to which the roots have failed to form depends on their stage of development at the time of irradiation; thus the lower incisors, which had nearly completed root formation, are relatively unharmed. The lower first molars are slightly stunted and the premolars and second molars extensively shortened. Such teeth do not suffer from excessive mobility.

central incisor usually following traumatic intrusion of the primary incisor, which displaces the developing permanent incisor. In taurodontism the pulp chamber is markedly elongated. A tooth that appears particularly enlarged may have split during its development or have become fused with an adjacent tooth.

DISTURBANCES IN STRUCTURE OF TEETH Enamel Enamel hypoplasia may affect a single tooth, following a localized periapical infection of its primary precursor (Turner tooth). But a more generalized form occurs as a complication of severe childhood infections or a nutritional deficiency with the manifestation depending on the time of the insult (chronological hypoplasia). Amelogenesis imperfecta is a developmental disorder of enamel formation affecting all or most of the teeth in both dentitions. The enamel may show varying degrees of hypoplasia from being pitted (Fig. 63.7), to almost complete absence of enamel when the crown appears angular. Alternatively, the enamel may be of normal thickness but be hypomineralized such that its radiographic density is similar to that of dentine.

Figure 63.7 Intra-oral (bitewing) radiographs showing marked hypoplasia and pitting of the enamel, whilst the dentine appears normal. Several of the teeth are carious.

Dentinogenesis imperfecta is a developmental anomaly of collagen formation that affects the dentine of both dentitions. The teeth are discoloured, having a brown or purple hue. The enamel chips away and the teeth rapidly wear down. The initial radiographic appearance shows bulbous crowns and large pulp chambers, which soon calcify with abnormal dentine so that little or none of the root canal is visible (Fig. 63.8). Although the teeth may appear sound, they are prone to infection resulting in pulpal necrosis and periapical radiolucencies. The appearance of the bone of the mandible and maxilla remains normal, although type IV is associated with osteogensis imperfecta. Dentinal dysplasia resembles dentinogenesis imperfecta but is less common. In type I the crowns look normal in colour and shape but the roots of the primary and permanent teeth are short and abnormally shaped, and the pulp chambers become obliterated with dentine prior to eruption. In type II, loss of the pulp chamber and narrowing of the root canals occurs after tooth eruption. In some cases the pulp chambers may be thistle shaped.

Cementum Hypercementosis describes the deposition of excessive amounts of cementum typically around the apical half of the root so that it appears bulbous. Usually it is localized to one or two teeth as it is caused by chronic dental infection or occlusal

Figure 63.8 A panoramic radiograph of a young adult with dentinogenesis imperfecta. The teeth have bulbous crowns, short stumpy roots and sclerosis of the root canals.

CHAPTER 63

overstress, but in Paget’s disease of the jaws (Table 63.3) it is widespread.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

ilar dental features are noted in hypophosphatasia but there is often premature loss of the primary teeth. In both conditions the jaws usually appear osteoporotic (osteopenic).

Miscellaneous conditions In hypophosphataemia, the pulps of the primary and permanent teeth are enlarged with pulp horns that extend toward the enamel/dentine junction, making the pulps susceptible to infection so that the teeth frequently become abscessed. Sim-

Table 63.3

DENTAL CARIES Dental caries is caused by microbial action on sugar with the formation of acid, which causes progressive demineralization

DIFFERENTIAL DIAGNOSIS OF LOCALIZED RADIOLUCENT AND RADIO-OPAQUE LESIONS OF THE JAWS

Unilocular radiolucent

Multilocular radiolucent

Radio-opaque

Mixed density

Common

Common

Common

Common

Alveolar abscess

Odontogenic keratocyst

Root fragment

Fibrous dysplasia (early)

Apical granuloma

Central giant cell granuloma

Dense bone island

Cemento-ossifying fibroma

Radicular cyst (apical)

Ameloblastoma

Mandibular torus

Periapical cemento-osseous

Residual cyst

Periapical sclerosing osteitis

Compound odontome

Dentigerous cyst

Hypercementosis

Complex odontome

Nasopalatine duct cyst

Supernumerary tooth

Florid cemento-osseous dysplasia

Odontogenic keratocyst

Sclerosing osteitis

Benign cementoblastoma (cementoma)

Uncommon

Uncommon

Uncommon

Uncommon

Stafne’s bone cavity

Giant cell tumour of hyperparathyroidism

Fibrous dysplasia (late)

Chronic osteomyelitis

Periapical cemento-osseous dysplasia

Osteosarcoma

Fibrous scar Fibrous dysplasia (early)

Ameloblastic fibroma

Periapical cemento-osseous dysplasia (early)

Odontogenic myxoma

Paget’s disease

Florid cemento-osseous dyplasia

Osteomyelitis

Ossifying fibroma (late)

Giant cell tumour of hyperparathyroidism

Compound odontome

Complex odontome

Central giant cell granuloma

Paget’s disease (late)

Ameloblastoma

Osteoma/exostosis

Lateral periodontal cyst Paget’s disease (early) Brown tumour of hyperparathyroidism

Rare

Rare

Rare

Rare

Carcinoma

Aneurysmal bone cyst

Metastatic carcinoma

Haemangioma

Metastatic carcinoma of prostate

Calcifying epithelial-odontogenic tumour (CEOT)

Haemangioma

Cherubism

Cementoblastoma

Calcifying odontogenic cyst

Glandular odontogenic cyst

Osteosarcoma

Osteoradionecrosis

Osteosarcoma Odontogenic myxoma

Chronic sclerosing osteomyelitis

Adenomatoid odontogenic tumour

Burkitt’s lymphoma

Chronic osteomyelitis

Ameloblastic fibro-odontoma

Lymphoma

Osteochondroma

Eosinophilic granuloma Chondroma & chondrosarcoma Neurofibroma Neurilemmoma Odontogenic fibroma Ewing’s tumour Myeloma

1433

1434

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

of the teeth, initially of the enamel, and then the dentine with destruction of their organic components. If left untreated it leads to the break down of the crown and bacterial infection of the pulp. It develops on the occlusal surfaces of the posterior teeth, on the approximal and cervical regions of the crown or root (if exposed), and as recurrent caries beneath restorations. The rate of mineral loss depends on a number of factors including the amount of sugar in the diet, the lack of effective oral hygiene, the presence of areas of food stagnation, and on the health of the individual. It is particularly rapid in those with reduced saliva production following radiation damage to the salivary glands. Radiographic detection of dental decay requires images with good contrast and resolution. Despite its limitations, intra-oral radiographs are valuable in the detection and monitoring of dental decay, particularly on surfaces not easily visualized and also for occult occlusal caries, which can be extensive beneath an apparently intact enamel surface. A carious lesion appears as a radiolucent zone, which represents an area of demineralization. An approximal lesion develops in the enamel just below the contact point with an adjacent tooth and has a triangular shape with the apex pointing towards the dentine. As the lesion progresses, its advancing surface broadens as it extends along the enamel dentine junction and penetrates into the dentine, the margin being ill defined (Fig. 63.9A, B). Adjacent carious lesions commonly develop on contiguous tooth surfaces. If left untreated, the caries reaches the pulp chamber and the weakened crown eventually crumbles away. A similar progression is seen on other tooth surfaces. The radiographic detection of dental caries can be difficult, particularly when the crown remains intact and the carious lesion is only marginally more radiolucent relative to the surrounding dentine.

Figure 63.9 (A) Bitewing radiographs showing gross caries with crown destruction affecting the upper right first and upper left second premolars. There is approximal caries, shown as radiolucencies of the crowns, at the contact points of the other remaining upper premolars, upper and lower right first molars, and occlusal caries in the upper left first molar. (B) Panoramic radiograph of a child aged 8 in the mixed dentition. There is approximal caries in the upper right deciduous first and second molars. There is gross caries distally in the upper left deciduous first molar, which has complete root resorption by the erupting successional premolar. There are retained fragments of the roots of the deciduous first molars. The lower left permanent first molar has gross recurrent occlusal caries beneath a very small restoration. There is less extensive recurrent caries in the lower right permanent first molar. Both these teeth show periapical bone changes, most obviously the widening of the periodontal ligament space on the mesial root of the lower left permanent first molar, consistent with periapical periodontitis.

DISORDERS OF THE PULP Dental decay that extends to the pulp results in inflammation (acute pulpitis), causing severe toothache, but has no radiological manifestations. However, regressive changes of the pulp (chronic pulpitis) due to chronic irritation can appear as calcific deposits, such as pulp stones and sclerosis (narrowing) of the root canals. Generalized pulp sclerosis is a feature of renal osteodystrophy and prolonged corticosteroid therapy. Internal resorption of the root canal or external resorption of the outer root surface may occur following pulp death.

Periapical periodontitis Pulpal necrosis results from acute pulpitis (see earlier) or from dental trauma due to interruption of the pulp’s blood supply. Bacterial action on the necrotic pulp within the root canal leads to the production of endotoxins, which exit the root apex and incite a periapical inflammatory response within the periodontal membrane (periodontitis).When this is acute, the

patient suffers severe discomfort and the offending tooth is tender to touch. Apical periodontitis presents as a widened periodontal ligament space, which appears more prominent than normal. However, the condition may also be chronic and continued progression of the chronic inflammatory process eventually leads to loss of the apical lamina dura and the formation of a discrete periapical radiolucency (Fig. 63.10) due to the formation of a focal inflammatory lesion, either a granuloma, radicular cyst, or chronic abscess. All three conditions have a similar radiographic appearance, that is a periapical radiolucency that is circular or oval, usually well defined, the outline being continuous with the remaining lamina dura around the root. Conversely, low grade stimulation from a nonvital tooth may result in reactive bone formation (sclerosing osteitis), which appears as an irregularly shaped, largely uniform area of dense bone at the root apex (Fig. 63.11).

CHAPTER 63

Figure 63.10 Periapical granuloma at the apex of the grossly decayed upper right lateral incisor. Although well defined, its margins are not corticated. Note the loss of the lamina dura at the tooth apex. There is a similar but smaller lesion at the apex of the exfoliating upper right first premolar root and the upper right central incisor is markedly carious.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

Odontogenic and non-odontogenic cysts have a number of radiological features in common that are characteristic of slow-growing lesions, i.e. they are radiolucent, well defined and often have a cortical margin. With the exception of the odontogenic keratocyst, they have raised intracystic pressure and expand by tissue fluid transudation, and so appear as circular or oval in shape. When large, the bony cortex of the jaws becomes thinned, expanded and then perforated. Jaw cysts tend to displace structures such as tooth roots, unerupted teeth, the inferior alveolar canal, and the antral floor. Radicular cysts are the most common (approximately 65 per cent) of the odontogenic cysts. They are derived from the cell rests of Malassez, which are epithelial remnants of root formation found in the periodontal ligament and develop at the apex of a nonvital tooth (see above periapical periodontitis). Any tooth can be affected but the majority are found on the permanent anterior teeth or first molars. When small (less than 15 mm in diameter) they resemble periapical granulomas but, unlike granulomas, can enlarge well beyond this size (Fig. 63.12). In many cases extraction of the causative tooth brings about resolution, but when this does not happen, the cyst is then termed a ‘residual cyst’. Thus a residual cyst found in an edentulous part of the jaw, has a well-defined, circular radiolucency usually with a cortical margin. Many regress without treatment and some show dystrophic mineralization. A dentigerous cyst (follicular cyst) arises from the reduced enamel epithelium, the tissue which surrounds the crown of an unerupted tooth. It is thus found only on teeth that are buried, particularly mandibular third molars and maxillary canines (Fig. 63.13). Cystic enlargement of the tooth follicle produces a pericoronal radiolucency, which is attached to the tooth at its neck, with the crown appearing to lie within the cyst lumen; however with large cysts this relationship may not be apparent. Odontogenic keratocysts arise from remnants of the dental lamina, the precursor of the tooth germ. The cyst lining has a higher mitotic activity than the oral mucosa and the cyst

Figure 63.11 Periapical radiograph showing sclerosing osteitis. The lower left molar is grossly decayed and its root is surrounded by a zone of radiolucency beyond which the bone is dense as shown by the lack of trabecular spaces.

CYSTS OF THE JAWS Cysts occur in the jaws more frequently than in any other bone because of the numerous epithelial cell residues left after tooth formation. Generally they are slow growing and painless unless infected, however some may reach a considerable size before detection. Cysts of the jaws are divided into odontogenic when they arise from epithelial residues of the tooth-forming tissues, or non-odontogenic, these being uncommon and mainly developmental, arising from epithelium not involved with tooth formation. The four most common odontogenic cysts are the inflammatory radicular (dental) and residual cysts, and the developmental dentigerous cyst and odontogenic keratocyst.

Figure 63.12 Part of a panoramic radiograph showing a corticated radiolucent lesion associated with the carious root of the upper left second premolar, extending into the maxillary antrum above the hard palate, consistent with a radicular cyst. Note the periapical radiolucency (granuloma/abscess) on the upper left second molar.

1435

1436

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.13 Part of a panoramic radiograph of a dentigerous cyst arising on a lower left wisdom tooth, which is unerupted and lying horizontally. It appears as a well-defined, circular radiolucency attached to the tooth at its neck. The inferior alveolar canal has been displaced inferiorly.

is thought to enlarge by mural growth and so behaves more like a benign neoplasm and is now classified as such. It appears as a unilocular or multiloculated, elongated, irregularly shaped radiolucency with a scalloped, well-defined margin (Fig. 63.14). It lacks the more ballooning characteristics of the other odontogenic cysts, which is an important diagnostic feature. Keratocysts occur most often in the lower third molar/ramus region, where they may displace, or occasionally replace an unerupted wisdom tooth and resemble a dentigerous cyst. Recurrence is common (15–20 per cent) so radiographic follow-up is necessary for several years. On CT, attenuation values of cyst fluid are higher than most other jaws cysts due to its high protein (keratin) content, ranging from 30–200 Hounsfield Units, with longstanding, multilocular cysts having the higher value3. Multiple odontogenic keratocysts are a feature of GorlinGoltz syndrome (Fig. 63.15), which also includes multiple basal cell naevi, calcification of the falx, bifid ribs, synostosis of the ribs, kyphoscoliosis, temporal and parietal bossing, hyperptelorism, and shortening of the metacarpals. There are several other less common odontogenic cysts. The lateral periodontal cyst is found on the lateral surface of a vital tooth root between two adjacent teeth, usually the lower incisor or canine teeth. The glandular odontogenic cyst mainly occurs in the anterior body of the mandible, as a multilocular or lobular, often large radiolucency that may cross the midline and has a tendency to recur. The nasopalatine cyst is probably the commonest nonodon-togenic cyst that is believed to arise from epithelial residues in the nasopalatine canal. It appears as a round, welldefined, midline radiolucency between, but not associated with, the upper central incisor teeth. Three lesions that may resemble jaw cysts but have no epithelial lining are sometimes considered with jaw cysts. The solitary bone cyst occurs during the first 2 decades of life,

Figure 63.14 Part of a panoramic radiograph of an odontogenic keratocyst which appears as a loculated radiolucency extending from the condylar neck to the lower first molar region. There is thinning of the bony cortices but no jaw expansion, a feature associated with odontogenic keratocysts. Note the displaced lower right third molar.

Figure 63.15 Part of a panoramic radiograph showing multiple odontogenic keratocysts consistent with Gorlin-Goltz syndrome. All four third molars have been extensively displaced and a lateral facial view showed the upper right one to lie posteriorly close to the orbit.

mainly in the premolar/molar regions of the mandible. Its margin is less well defined than those of odontogenic cysts and its superior border arches up between the roots of the adjacent teeth (Fig. 63.16). Tooth displacement and root resorption is uncommon. At surgery an empty cavity is found, which subsequently heals after bleeding has been induced. The aneurysmal bone cyst is considered to be a reactive lesion of bone and is characterized by a fibrous connective tissue stroma containing many cavernous blood-filled spaces.

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

Figure 63.16 Part of a panoramic radiograph showing a partially corticated radiolucency in the right mandible involving the apices of the second premolar and first and second molars diagnosed as a solitary bone cyst. Note the characteristic scalloping between the roots of the first and second molars.

It is rare and occurs mainly in the young, with over 90 per cent occurring before 30 years of age. It is typically found in the posterior region of the mandible as a well-defined, multilocular, often septated, circular radiolucency. It has a tendency to produce marked cortical expansion. Computed tomography (CT) or magnetic resonance imaging (MRI) shows the presence of fluid levels due to the presence of blood-filled cavities. Stafne’s bone cavity is asymptomatic and typically found in men over the age of 35 years. It forms a depression in the lingual cortex of the mandible just in front of the angle and below the inferior dental canal. Its origin is controversial and it has been postulated that it arises from pressure from the submandibular salivary gland, however whilst some may contain salivary gland tissue a number develop anterior to the gland. On plain radiographs, it appears as a well defined, punched out, dense radiolucency, which rarely exceeds 2 cm in diameter (Fig. 63.17A, B). Its appearance is characteristic and so does not require further imaging or biopsy. However, if CT or MRI is performed, the cavity is often found to contain fat.

DISEASE OF THE PERIODONTIUM There are several disorders that affect the periodontium and these are referred to as periodontal disease. Radiographs, in particular intra-oral views (bitewings and periapical films) are helpful in the assessment of the amount of remaining bone support for the teeth, the pattern of bone loss, the detection of subgingival calculus, and local aggravating factors such as poorly contoured dental restorations. Chronic periodontitis results from the accumulation of dental plaque on the teeth initially causing low-grade inflammation of the gingivae (gingivitis) and progresses to

Figure 63.17 (A) Part of a panoramic radiograph showing a corticated radiolucency between the inferior alveolar canal and the lower border of the mandible due to the presence of a Stafne bone cavity. The 3D CT (B) shows the depression on the lingual aspect of the mandible.

involve the periodontal tissues. Gingivitis has no radiological features. Chronic periodontitis is a painless condition that affects almost all adults and results in slow but gradual horizontal bone loss from the alveolar crest so that the teeth lose their support. Advanced periodontitis affects 10–15 per cent of the population, with smoking being a specific risk factor. A particularly aggressive form called rapidly progressive periodontitis occurs in young adults, where several teeth are affected by vertical bone loss resulting in angular bony defects that extend down towards the tooth apex (Fig. 63.18). Symmetrical widening of the periodontal ligament space affecting several teeth is sometimes seen in progressive systemic sclerosis or as irregular localized widening as an early feature of osteosarcoma of the jaws.

ODONTOMES AND ODONTOGENIC TUMOURS When diagnosing odontogenic disorders, a simple rule is that they occur or are centred upon the tooth-bearing parts of

1437

1438

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.18 Part of a panoramic radiograph of a man aged 40 with smoking related aggressive chronic periodontal disease. There is furcation involvement of the upper left first molar and loss of up to 70 per cent of the bony attachment between the upper right first molar and second premolar and similarly on the upper left side. Combined peridontalendodontic lesions affect both lower first molars.

the jaws, whereas a lesion not involving the jaw alveolus, e.g. those lying below the inferior dental canal, are unlikely to be of odontogenic origin. Odontomes are developmental malformations or hamartomas consisting of dental hard tissues or tooth-like structures. Most are diagnosed in the second decade of life and frequently impede tooth eruption. There are two main types. The compound odontome consists of a collection of small discrete teeth called denticles and is found typically in the anterior region of the maxilla, whereas the complex odontome consists of a randomly arranged mass of enamel, dentine and cementum found predominantly in the lower premolar/molar region. Both types are densely radiopaque due to the presence of tooth enamel and are surrounded by a thin radiolucent capsular space and radio-opaque cortical margin. There are several minor developmental anomalies also classified as odontomes that can resemble teeth, but these are not described. Odontogenic tumours are uncommon, mostly benign and arise from either the odontogenic epithelium, odontogenic epithelium and ectomesenchyme, or primarily ectomesenchyme.The commonest is the ameloblastoma, which accounts for 11 per cent of all odontogenic tumours. It occurs mainly in patients between 30–50 years of age with most (80 per cent) forming in the molar/ramus region of the mandible. When the maxilla is involved, it has the potential to spread insidiously to involve the infratemporal fossa, orbit and skull base, thus a thorough assessment is essential. The ameloblastoma has a variable radiographic appearance being a unilocular or multilocular radiolucency, but typically contains septa or locules of variable size to produce a honeycombed appearance (Fig. 63.19). The margin is well defined, often corticated but when large, produces jaw expansion with perforation of the bony cortex. A useful diagnostic feature is knife-edge resorption of the tooth roots by the tumour, which can be quite marked. The lesion is locally aggressive and requires a wide excision margin, so accurate presurgical assessment is necessary. Contrast-enhanced CT will show the tumour–bone interface but has poor soft tissue delineation. Multislice CT can be used

Figure 63.19 Part of a panoramic radiograph showing an ameloblastoma, which appears as an expansile, multilocular radiolucency involving the left body of the mandible.

to differentiate ameloblastoma from odontogenic keratocysts because of higher density increase during the arterial phase4. On T1-weighted images with gadolinium enhancement and T2-weighted images, there is good conspicuity of the tumour margin with the soft tissues, the lesion having a moderate to high signal.There is a very rare malignant variety in which the ameloblastoma probably undergoes malignant transformation with metastases most often occurring in the lungs5. The unicystic ameloblastoma occurs around the age of 20 years and often causes marked bony expansion (Fig. 63.20). The odontogenic myxoma is a benign but locally aggressive tumour of odontogenic mesenchyme occurring mainly in those younger than 45 years of age. Most occur in the mandible in the premolar molar region. The lesion is usually well

Figure 63.20 An axial CT on bone window settings of a large cystic ameloblastoma of the right side of the mandible showing marked thinning and expansion of the bone. Note the presence of root resorption.

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

defined, unilocular and contains a variable number of internal coarse trabeculations to produce a reticular pattern. There are many types of odontogenic tumour, however two lesions that are well defined and contain variable amounts of focal mineral deposits are the calcifying epithelial odontogenic tumour and adenomatoid odontogenic tumour. The former is more common in men, occurs in middle life and is found mainly in the premolar/molar region of the mandible. The latter mainly affects females in the second decade of life and typically occurs anteriorly, especially in the maxilla and is associated with an unerupted tooth. The cementoblastoma is the only neoplasm of cementum; it is rare and mainly affects young males. It appears as an encapsulated radiopaque mass attached to the root, usually of a lower posterior tooth.

IMAGING IN IMPLANTOLOGY Dental implants are placed into the jaws to replace missing teeth, giving anchorage for dental prostheses.These have gained considerable popularity since the discovery, by Branemark, that endosseous titanium implants could successfully integrate with bone. Imaging plays an essential role in pre-implant assessment, defining the volume and quality of recipient bone, identifying the location of relevant anatomical structures such as the inferior dental canal, the floor of the nose and maxillary antrum, and in assessing the status of adjacent teeth. Postoperatively, imaging is necessary to examine the degree of healing and monitor osseointegration1. Intra-oral periapical and dental panoramic radiography are valuable for initial pre-operative assessment but cannot identify volume of a recipient bone site or accurately localize key structures. Cross-sectional imaging is therefore indicated, particularly in complex cases, taking the form of conventional tomography, CT, MRI or Cone Beam CT (digital volume tomography). Complex tomography can be performed using dedicated multimodal maxillofacial tomography machines (Fig. 63.21) and is now available as an adjunct to conventional panoramic equipment. CT data from multiple contiguous 1–2 mm slices taken parallel to the maxillary hard palate or lower border of mandible are reformatted by dedicated multiplanar reconstruction programmes to give cross-sectional slices crossreferenced with axial plan views of the dental arch. Newly emerging cone-beam CT machines offer a similar imaging capability but with substantially lower radiation exposure, due to use of exclusively hard tissue imaging parameters. While cross-sectional tomography is used for assessment prior to placement of small numbers of implants, CT is normally reserved for complex cases involving implants in multiple quadrants. More recently MRI has been shown to be a feasible alternative to CT6. A stent is generally needed for any form of cross-sectional imaging to indicate chosen implant sites – this may be metallic (tomography), gutta percha or brass (CT), or gadolinium (MRI). In the postoperative phase intraoral periapical radiographs taken perpendicular to the implant, are most useful, monitor-

Figure 63.21 Images reconstructed from fine axial CT slices of an implant site within the mandible (SimPlant). An original axial slice is cross-referenced with a cross-sectional slice, a panoramic-like reconstruction and three-dimensional (3D) surface-rendered views. The software allows precise planning of implant placement to avoid injury to adjacent structures and helps predict the aesthetic outcome. (Courtesy of Mr Sean Goldner).

ing osseointegration and identifying peri-implant bone loss, especially at the neck, which may indicate early failure of integration.

DISORDERS OF BONE Developmental disorders Fibrous dysplasia is a localized abnormality in which cancellous bone is replaced by fibrous tissue containing varying amounts of calcified tissue. When the jaws are affected, the maxilla is involved twice as frequently as the mandible. An immature lesion is largely radiolucent and may mimic a dental cyst, but more frequently fibrous dysplasia presents as a radiopacity, typically having an orange peel or ground-glass texture (Fig. 63.22). On radiographs the margins are usually indistinct, blending in with the normal adjacent bone. It may displace teeth or prevent their eruption. Large lesions produce jaw expansion, with thinning of the bony cortices and displacement of the antral floor. It can resemble both a cementoossifying fibroma, which is better defined, and an osteosarcoma, which produces destructive changes. Cherubism is a rare dysplasia of bone that develops during the first decade of life. It occurs bilaterally in both jaws, but more commonly affects just the mandible. It develops in the posterior aspects of the jaws as a multilocular, honeycombed, expansile radiolucency. Tooth displacement is common. It regresses spontaneously after skeletal growth ceases. Periapical cemento-osseous dysplasia and florid cemento-osseous dysplasia are similar conditions, with the latter being a more extravagant version of the former. They occur mainly in women, particularly of Afro-Caribbean origin, after 25 years of age. Both conditions are characterized by

1439

1440

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.23 Part of a panoramic radiograph of the right mandible of a patient who developed acute osteomyelitis following the extraction of a premolar root 3 weeks previously. There are three sequestra in the alveolar portion of the mandible.

Figure 63.22 Axial CT on bone window setting of fibrous dysplasia of the anterior aspect of the mandible. There is thinning and expansion of the bony cortical plates. The lesion shows areas of low attenuation, lingual to the teeth, due to the presence of fibrous tissue.

the formation of multiple deposits around the tooth roots and the mandible is more frequently involved than the maxilla. In periapical cemento-osseous dysplasia, although the teeth are clinically sound, radiolucent lesions form at the apices of the teeth, which resemble periapical granulomas (see earlier). Gradually cemental-like tissue is deposited within so that it become increasingly radiopaque. When mature it becomes almost totally radiopaque but is surrounded by a thin, peripheral radiolucent zone, which helps distinguish it from sclerosing osteitis. The appearance of florid cemento-osseous dysplasia is similar to periapical cemental dysplasia but the lesions are larger, may produce jaw expansion, are more numerous, and often in both the maxilla and mandible. When extensive the condition may resemble Paget’s disease of bone.

Inflammatory disorders Osteomyelitis of the jaws is uncommon, which is surprising considering the frequency of dental sepsis. It may develop from a dental abscess or complicate tooth extraction. In acute osteomyelitis, there is thinning and discontinuity of the bony trabeculae to produce ill-defined, patchy areas of radiolucency within the cancellous and cortical bone. With time bony sequestrae form and are recognized as irregularly shaped islands of bone set against a region of radiolucency (Fig. 63.23). The features of osteomyelitis are more readily visualized on CT, which also is useful to show periosteal bone formation. On MRI, the marrow usually shows a low signal intensity on T1 and a high signal on T2 weighted images. If the disease becomes chronic, the bone becomes diffusely affected and extensively involved with sclerosis of the marrow spaces. CT will demonstrate the internal structure and the presence of sequestration. In diffuse sclerosing osteomyelitis, the bone becomes increasingly dense and radio-opaque as a result of a proliferative response to low-grade infection. Bone deposition results in obliteration of the marrow spaces with gradual spread through

the mandible, however lytic areas are also seen. MRI shows a thickened cortex. Osteoradionecrosis is an inflammatory condition that can affect the mandible if it is included in the radiation field after a dose of 45–50 Gy has been delivered. It is a clinical diagnosis and presents as areas of exposed necrotic bone. The radiographic features resemble those of chronic osteomyelitis. A similar condition called osteonecrosis may occur in the mandible in patients taking bisphosphonates.

Metabolic, endocrine, and haematological disorders of bone Osteoporosis is found in the jaws as elsewhere in the skeleton. It affects the mandible, which becomes osteopenic, as the marrow spaces enlarge and the trabeculae thin.The cortical outline of the inferior alveolar canal becomes inconspicuous and the lower border of the mandible becomes thinner than normal. Hyperparathyroidism of the jaws results in a general demineralization of the bone, creating a ‘ground glass’ appearance, loss of the lamina dura, formation of brown tumours, and subperiosteal erosions at the angle. Brown tumours develop in the facial bones, particularly in longstanding cases and appear radiolucent, with margins that are often ill defined but can also appear cystic. Haematological replacement disorders may affect the jaws, the radiological manifestations depending on the severity of the condition. In moderate to severe thalassaemia, the jaws become radio-lucent with the presence of coarse trabeculations due to marrow hyperplasia and the maxillary antrum is reduced in size (Fig. 63.24). The skull takes on a granular appearance, with thickening of the diploic spaces and occasionally a ‘hair on end’ appearance. In sickle cell disease similar manifestations are apparent, however several sclerotic areas are seen as a consequence of dystrophic mineralisation of small thrombi. Paget’s disease of bone is sometimes seen in the jaws, the maxilla being more commonly involved than the mandible. The radiographic disease depends on its stage of development progressing from an initial radiolucent stage to assume a more granular or ‘ground glass’ appearance. Later the bony trabeculae become coarse and arranged in a horizontal linear pattern

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

The cemento-ossifying fibroma is a tumour of bone but it can also be considered as a fibro-cemento-osseous lesion.

Its behaviour varies from those showing slow growth to others being quite aggressive. It occurs mainly in young adults, mostly in the body of the mandible. The radiographic appearance depends on its degree of mineralization, and typically contains a wispy or tufted bony trabecular pattern (Fig. 63.26). The lesion is encapsulated and so appears well defined, helping to distinguish it from fibrous dysplasia. Osteomas of the maxillofacial bones and jaws are usually slow-growing, painless and thus discovered by chance. However a large osteoma in the frontal sinuses may obstruct the drainage pathway and cause secondary infection. In the jaws, osteomas more commonly affect the mandible than the maxilla and, whilst any site can be involved, they tend to be found posteriorly on its medial aspect (Fig. 63.27). CT assists in showing the site of origin and provides three-dimensional (3D) topographic detail. Multiple osteomas are a feature of Gardner’s syndrome (familial adenomatous polyposis) and precede the formation of intestinal colonic polyposis. Osteosarcoma is uncommon in the jaws, accounting for only 7 per cent of all osteosarcomas. It tends to be slower growing and occurs about 10 years later than osteosarcoma of the long bones.The mandible is more commonly affected than the maxilla. Maxillary lesions tend to arise from the alveolar ridge, and mandibular ones in the body. It has a destructive appearance and its density varies from being radiolucent, to patchily radio-opaque or predominantly sclerotic. An important early dental radiographic sign is widening of the periodontal ligament space due to tumour spread along the periodontal ligament, however this feature is also seen in other sarcomas (e.g. fibrosarcoma Ewing’s sarcoma). When the periosteum is elevated a ‘hair-on-end’, sunray or onion skin appearance may be visible. CT is required to demonstrate accurately tumour calcification, bone destruction, and bone reaction (Fig. 63.28),

Figure 63.25 Part of a panoramic radiograph of a central giant-cell granuloma of the right side of the mandible, which appears as a welldefined radiolucency containing numerous coarse bony trabeculae. There has been displacement of the premolar teeth posteriorly and the canine anteriorly.

Figure 63.26 Bone window setting of an axial CT of a cementoossifying fibroma of the mandible showing mainly buccal expansion and thinning of both cortical plates, which remain intact. The lesion is of mixed attenuation as it contains areas of fibrosis, mineralization, and coarse bony trabeculations.

Figure 63.24 Panoramic radiograph of a case of thalassaemia. There is marked increase in the height of the mandible, which is composed of coarse trabeculae enclosing large marrow spaces and the small maxillary sinuses. Note the generalized loss of the lamina dura and the periodontal abscess on the distal root of the lower right first molar.

and finally the bone becomes radio-opaque with focal collections of dense bone creating a ‘cotton wool’ appearance. The central giant cell granuloma is probably a reactive lesion and not neoplastic. It most often is detected during the first three decades of life with twice as many being discovered in the mandible as the maxilla. The lesion is usually multilocular, often well defined, lacks cortication but some have ill defined borders suggestive of a destructive lesion. Although largely radiolucent, the internal appearance varies from being almost devoid of any internal structures to those containing wispy septae (Fig. 63.25).

TUMOURS OF BONE

1441

1442

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

nantly in the posterior aspects of the mandible, resulting in loss of the outline of the inferior alveolar canal and destruction of the cortical plates. Typically their outline is moderately to poorly defined, with irregular margins that appear destructive (Fig. 63.29) without new bone formation. Some metastatic deposits are characterized by the development of several areas, often small, of bone destruction. Although most lesions that metastasize to the jaws are lytic, others, notably from the prostate, can produce bone and appear radio-opaque. An extranodal lymphoma can affect the maxilla and posterior aspect of the mandible and generally appears as an ill-defined noncorticated radiolucency (Fig. 63.30). Eosino-

Figure 63.27 Bone window setting of an axial CT showing a dense (compact) osteoma arising from the medial aspect of the ramus of the right mandible.

Figure 63.29 Panoramic radiograph showing metastatic carcinoma of the lung. The patient had his remaining lower teeth extracted 2 weeks previously because of toothache. Note the soft-tissue swelling in the lower right third molar region. The outline of the socket is no longer visible nor is the course of the inferior dental canal. There are multiple punched-out radiolucencies in the right angle (arrows) with erosion of the inferior cortex.

Figure 63.28 Bone window setting of an axial CT of osteogenic sarcoma of the left mandibular ramus. There is bone destruction in the region of the sigmoid notch. The lesion contains areas of neoplastic bone formation and extends medially towards the lateral pterygoid plate, posteriorly to the styloid process, and laterally resulting in facial swelling. (Courtesy of Mr S. Dover, Birmingham).

whilst MRI (T1- and T2-weighted images) will provide better information on the intramedullary and extraosseous components of the tumour7. Primary carcinoma of the overlying oral mucosa can invade the jaws to produce an ill-defined, noncorticated area of bone destruction. Bone destruction around the tooth roots gives an appearance of teeth floating in space.Very rarely, carcinoma may arise from malignant transformation of a cyst lining or epithelial residues within the jaw bone to produce an illdefined osteolytic lesion. It affects older patients usually during in the 6th or 7th decades of life. Metastatic tumour involvement of the jaws is uncommon, being less than 1 per cent. Primary sites include the breast, kidney, lung, colon, and prostate. They form predomi-

Figure 63.30 Extranodal lymphoma of the maxilla shown on a bone window setting axial CT at the level of the alveolus. Although a few areas of the lesion are well defined, the overall appearance is destructive with loss of much of the buccal alveolar plate.

CHAPTER 63

philic granuloma and the other Langerhans’ cell histiocytosis are occasionally found in the jaws, where they may mimic dental periapical or periodontal disease. Multiple myeloma is uncommon in the jaws; the mandible is more frequently affected than the maxilla, with a predilection for the posterior body and angle. Typically the lesions are well defined but lack a cortical margin and so appear punched out, but some have ragged margins.

TRAUMA Teeth The teeth, particularly the anterior ones, are frequently involved in traumatic injuries to the face. They may be partially or completely avulsed or the crowns or roots may be fractured. Crown fractures may just involve the enamel, or the enamel and dentine, or if at a lower level expose the pulp. Root fractures occur less frequently and if undisplaced can be difficult to detect on radiographs when two different angled views may be required. Dental trauma can be associated with a localized dento-alveolar fracture in which a block of bone becomes detached containing several teeth. Intra-oral radiographs best demonstrate traumatic injuries of the teeth.

Fractures of the facial skeleton The facial skeleton is a complex arrangement of bones, air cavities and soft tissues attached to the skull base. Fractures in this region may be confined to the maxilla or involve other parts of the facial skeleton, such as the zygoma and nasal bones. Maxillary fractures are still classified according to the system described by Le Fort in 1900, who defined three principle types of fractures after applying blunt trauma to the faces of cadavers. The classification is contentious because fractures do not always follow the exact pattern he described (Fig. 63.31). Plain films form the initially assessment, but many of these injuries are complicated and require evaluation with axial and coronal CT or cone beam CT.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

In a Le Fort I fracture (Fig. 63.32), the tooth-bearing part of the maxilla is separated from the rest of the maxilla by trauma to the lower part of the face. The fracture line passes above the teeth but below the zygomatric buttress. It runs from the piriform fossa posteriorly to the pterygoid plates and involves the lower part of the nasal septum resulting in detachment of the dentoalveolar fragment from the remainder of the maxilla.The posterior portion may drop inferiorly, resulting in an open bite, and this is a useful diagnostic feature, as the Le Fort I fracture can often be difficult to detect on radiographs. The Le Fort II fracture (Fig. 63.33) runs along the nasal bridge, through the lacrimal bones across the medial orbital walls, and orbital rims wall to involve the anterior and posterolateral wall of the maxillary sinuses and pterygoid plates. The nasal septum is fractured at a variable level. In the Le Fort III or suprazygomatic fracture (Fig. 63.34), there is complete separation of the midface from the cranial base resulting in clinical lengthening of the face. The fracture line runs through the nasal bones, the frontal processes of the maxilla, posterolaterally through the medial and lateral orbital walls, and through the zygomatic arches.The nasal septum is fractured superiorly. In practice, many fractures do not exactly fit these descriptions. Fractures caused by sharp-edged objects can produce comminuted fractures of the maxilla (Fig. 63.35) without affecting the tooth-bearing alveolar bone, and some fractures are not symmetrical. For instance, Le Fort I, II, and III fractures may be unilateral (Figs. 63.35, 63.36), or a Le Fort I or II may

Figure 63.32 Le Fort 1 fracture. The fractures of the lateral walls of the maxillary sinuses have been arrowed. The fractures of the medial walls cannot be seen on this projection.

Figure 63.31 Fracture lines in Le Fort I, II, and III fractures.

Figure 63.33 Le Fort II fracture. The fractures of the inferior margins of the orbits and of the lateral walls of the maxillary sinuses have been arrowed. The fracture of the nasal bone is not visible on this projection.

1443

1444

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.35 3D CT scans showing multiple fractures of the facial bones (Le Fort II and unilateral Le Fort III) and depressed fracture of frontal ethmoidal complex. Note the asymmetrical pattern of the fractures with the more severe injury involving the left side.

Figure 63.34 Combined Le Fort II and Le Fort III fractures. (A) There are fractures of the nasal bones, lateral orbital margins, inferior orbital margins and zygomatic arches. All have been marked with arrows. There is a fluid level in the right maxillary sinus and opacification of the left maxillary sinus. (B) Lateral view of fractures shown in A. There is a fracture of the nasal bones with wide separation (white arrow) and there are fractures of the posterior walls of the maxillary sinuses (black arrow).

Figure 63.36 Coronal CT of a unilateral left-sided Le Fort II fracture. There are fractures of the medial wall of the orbit, the floor and inferior rim of the orbit, and the anterior wall of the maxillary sinus.

coexist with a Le Fort III (Fig. 63.34). Nevertheless the Le Fort classification is still widely used as it allows these complicated fractures to be described simply.

Fractures of the zygomatic complex The zygomatic bone contributes to the lateral and inferior margins of the orbit, the lateral wall of the maxillary sinus, and the anterior end of the zygomatic arch, thus fractures of the zygomatic bone involve all the sites (Fig. 63.37) usually with fractures in the region of the zygomatico-frontal suture, zygomatico-temporal suture, infra-orbital rim and the lateral wall of the antrum. As not all of the fractures are always visible on a single film the presence of even one fracture should raise the suspicion that other fractures are present and, dependent

Figure 63.37 Diagram of the usual sites of fracture of the zygoma and of the zygomatic arch. o = orbit, a = antrum.

CHAPTER 63

upon clinical findings, other imaging may be required. Severe trauma may result in a comminuted fracture of the zygomatic complex (Fig. 63.38A, B). The zygomatic arch may be fractured in association with a fracture of the zygoma as described earlier, or it may be fractured alone as a result of direct trauma to the side of the head and is seen as three points of fracture (Fig. 63.37).

Blow-out fractures of the orbit A blow-out fracture occurs following blunt trauma to the front of the orbit. The inward force of the eyeball produces a temporary increase in pressure within the orbit, resulting in fracture and displacement of the thin bone of the orbital floor, but leaving the orbital rim intact. The orbital soft tissues herniate through the defect into the maxillary sinus.This often shows as soft tissue opacity in the upper aspect of the sinus on an occipitomental radiograph and sometimes a fluid level when blood is present. CT is helpful in defining the extent of the defect of the orbital floor and the involvement of the external ocular muscles in the fracture (Fig. 63.39). Coronal CTs demonstrate soft tissue displacement more clearly than axial. Orbital ultrasound (US) with a curved-array transducer has been shown to be useful in the detection orbital wall and orbital rim fractures8. Failure of recognition of a blow-out fracture or fusion of malpositioned bony fragments may lead to entrapped tissues, fibrosis, and possible diplopia. Orbital floor fractures are often accompanied by fracture of the medial wall (lamina papyracea) of the orbit.

RADIOLGICAL INVESTIGATION OF MAXILLARY FRACTURES The initial radiographic investigation of a patient with a suspected fracture of the maxilla should be limited to a standard occipitomental, a 30-degree occipitomental (both taken postero-anterior (PA) if possible as this provides better bony detail than antero-posterior (AP) projections and with patient cooperation), and a true lateral view centred on the maxilla.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

These radiographs may be sufficient to diagnose and assess the fractures, but other imaging is often necessary, particularly in the more severely injured patient. It should be remembered that recent undisplaced fractures may be not always be demonstrated on routine plain films and soft tissue oedema may obscure detail. Displaced fractures of the orbital mar-gins should show on the occipitomental views. In the Le Fort I and II fractures, there is usually a step or break in the posterior wall of the maxillary sinus on the lateral view (Fig. 63.34). Fractures of the zygomatic arches usually show on occipitomental projections, but may be better visualized on an underpenetrated submentovertical view. Computed tomography (CT) is useful in the diagnosis and assessment of Le Fort II and Le Fort III fractures, nasoorbital fractures, orbital blow-out fractures, skull base fractures and may be helpful with Le Fort I and zygomatic complex fractures. CT demonstrates the amount of displacement or rotation of the fractured fragments, injury to the globes, optic nerve compression from bone fragments, fracture of the cribriform plate, and the possible presence of foreign bodies. Axial slices parallel to the occlusal plane are appropriate for detecting maxillary fractures allowing simple repeatability and limiting artefacts produced by metallic dental restorations. Thin slices are desirable for good bone detail and need not be contiguous unless 3D reconstruction is planned. Coronal slices at right angles to the occlusal plane demonstrate fractures of the orbital floor better than axial scans, but may not be possible in the elderly or uncooperative or recently injured patient. Thin noncontiguous slices (e.g. 2 mm at intervals of 4 or 5 mm) are adequate in most cases and should cover the orbital floors from the orbital rim to the optic foramina. Coronal slices that suffer from artefacts from dental restorations can be reduced by angling away from the true coronal position, or avoided by reconstruction from contiguous axial slices but at the cost of impaired resolution. CT with 3D reconstruction may be used to demonstrate graphically fractures of the facial skeleton (Fig. 63.38B) and is helpful in the management of residual traumatic deformities that require reconstructive surgery. Multislice CT results

Figure 63.38 Comminuted fracture of the left zygoma on axial CT. (A) There are multiple fractures of the anterior, posterolateral and medial walls of the maxillary sinus. There is air in the soft tissues of the cheek and infratemporal fossa. (B) 3D CT reconstruction of a comminuted fracture of the left zygoma (same case as A).

1445

1446

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.39 Coronal CT showing a blow-out fracture of the floor of the right orbit. Fractures of the lamina papyracea (with blood within the ethmoid complex) and orbital floor (with herniation of orbital contents and air within orbital cavity) are arrowed.

in greatly increased speed of data acquisition and reconstruction so that more slices can be recorded with thinner sections than previously with spiral CT. Rapid data acquisition and multiplanar reconstructed images are relevant, particularly for severely injured patients as the result of, for example, blunt trauma from road traffic accidents who may have intracranial pathology or cervical vertebra trauma where there may be patient positioning problems.

Fractures of the mandible Fractures of the mandible tend to occur at specific sites, as shown in Fig. 63.40. They are best demonstrated on a dental panoramic radiograph (Fig. 63.41), (or right and left oblique lateral mandibular views), together with a PA mandible radiograph. However, a fracture in the parasymphyseal region may not show on either of these films but will be demonstrated on intra-oral views, which are useful where the teeth are also

Figure 63.41 Panoramic radiograph showing bilateral fractures of the mandible in the right canine region and left wisdom tooth region.

Figure 63.40 mandible.

Line diagram showing common sites of fracture of the

thought to be fractured. A fracture on one side of the mandible is frequently accompanied by a contralateral fracture, particularly of the condylar neck.

Temporomandibular joint The temporomandibular joint (TMJ) is a complex diarthrodial, synovial joint.It contains the mandibular condyle,which sits in the glenoid fossa when the mouth is closed. Anteriorly lies the articular eminence and posteriorly the external auditory meatus. The joint is divided into an upper and lower joint compartment by a biconcave, fibrocartilagenous disc, which acts as a cushion for the mandibular condyle. The disc lies above the condyle and is attached posteriorly by fibro-elastic tissue to the base of the skull, neck of the mandibular condyle, and elsewhere to the fibrous joint capsule. The capsule is lined by a synovial membrane and encloses the joint. Fibres of the lateral ptergyoid muscle insert into the anterior aspect of the capsule and articular disc as well as to the anterior aspect of the condylar neck.

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

The temporomandibular joint is susceptible to conditions that affect other joints including developmental abnormalities, arthritic, traumatic, and neoplastic disease.

DEVELOPMENTAL ABNORMALITIES These usually affect the temporal and condylar components, mainly consisting of changes in size and form, and usually result in alteration in the growth of the affected side of the mandible. Hypoplasia of the mandibular condyle is failure of the condyle to attain full size. It may be localized to the joint or can be part of a generalized disorder, e.g. first arch syndrome. It results in underdevel-opment of the affected side of the mandible. Condylar hyperplasia causes enlargement of the mandibular condyle due to overactivity of the cartilaginous growth centre. It produces either a posterior open bite on the affected side or a centre line shift of the mandible relative to the maxilla. Radionuclide imaging may be required to demonstrate whether growth of the condylar cartilage is active prior to corrective treatment. In Hurler’s syndrome (gargoylism) the articular surface of the condyle is usually concave instead of convex, an appearance thought specific for this syndrome. A bifid condyle is believed to result from obstructed blood supply or trauma during its development.There are no clinical features and it appears as a notch or indentation on the articular surface of the condyle.

Temporomandibular joint dysfunction Temporomandibular joint dysfunction is a common condition and consists of myofascial pain, resulting in muscle spasm, facial discomfort and/or internal derangement of the articular disc. Myofascial pain occurs in all age groups but is particularly common in young female patients. It presents as muscle tenderness, jaw stiffness and headaches, and is associated with stress and anxiety, tooth clenching, and occlusal disharmony. The condition is a type of fibromyalgia and as the bony tissues are not affected it has no specific radiological changes. Internal derangement is an abnormality of the position of the articular disc, which may also show an altered morphology. The normal disc position in relation to the mandibular condyle is illustrated in Fig. 63.42. When the disc becomes displaced, it does so usually in an anteromedial, medial or sometimes lateral position relative to the condylar head. The condition may be painful and cause clicking or jaw locking, however several studies have shown displaced discs in individuals without symptoms. Discomfort is more prevalent in patients with disc displacement without reduction, particularly in combination with osteoarthrosis and bone marrow oedema. The diagnosis is made from the clinical findings and in most cases the condition improves or resolves with conventional therapy.When this fails and more aggressive treatment is planned or when the diagnosis is uncertain, the disc position can be demonstrated using arthrography or MRI with the latter being the technique of choice9.

Figure 63.42 Diagrammatic representation of the articular disc (shaded). (A) In a normal position in relationship to condylar head, and (B) in an anteriorly displaced position with reduction on opening, and (C) with a nonreducing disc.

The joint can be imaged using protein density or T1 and T2 sequences using 3 mm parasagittal slices, the angulation being determined by the medial angulation of the condylar head. On MRI, the disc appears as a biconcave (bow tie shaped) structure of low attenuation sandwiched between the anterior aspect of the articulating surface of the condyle and the glenoid fossa. Anterior disc displacement with reduction of the disc is shown in Fig. 63.43A and B. When the mouth is opened, the anteriorly displaced disc reduces or moves back to a normal position relative to the condylar head; this manoeuvre often results in a click, which can be apparent to the patient and palpated by the clinician. However, if the disc remains anteriorly displaced (Fig. 63.44) it may interfere with forward translation of the condyle, resulting in locking and restricted mouth opening, and pressure on the disc may cause it to become distorted. T2-weighted images can be used to show joint effusions and inflammatory change. The significance of joint effusion is controversial as it can occur in the nonpainful joint. However, it is observed more often in joints with more advanced stages of disc displacement, i.e. nonreducing discs, and it is thought to represent the presence of synovitis10. Images taken in a coronal plane demonstrate disc displacement medially or laterally; this view is useful to show degenerative changes to the articular surface.The use of plain radiographs to assess joint space as a predictor of disc displacement has been shown to have a low predictive value11.

1447

1448

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

and joint tenderness similar to temporomandibular joint dysfunction (TMJ) but crepitus is often present. The radiographic features are shown diagrammatically (Fig. 63.45), and include flattening and irregularity of the condylar surface, sclerosis, and osteophyte formation, which is seen mainly on the anterosuperior condylar surface (Fig. 63.46). Inflammatory arthritis. Rheumatoid arthritis is a common condition in which the TMJ becomes involved in about half of cases. Symptoms include pain, swelling, and jaw stiffness. Radiographic changes consist of loss of bone density and formation of erosions leading to a somewhat pointed condylar head. T2-weighted coronal MRI images are valuable for demonstrating the presence of joint inflammation. Other

Figure 63.43 Anterior reducible subluxation of disc. Fast-field echo (FFE) MRI parasagittal images showing an anteriorly positioned disk (A), which reduces on opening (B) to lie over the condyle. Figure 63.45 Diagrammatic representation showing various degenerative changes that affect the mandibular condyle.

Figure 63.44 Parasagittal T1-weighted MRI image with the mouth open showing a nonreducing, anteriorly displaced disc.

Arthritides Degenerative joint disease (osteoarthritis) can develop at any age but its incidence increases with age. It is thought to occur when the joint is unable to adapt to remodelling forces.There may be no symptoms or there may be discomfort

Figure 63.46 TMJ. Degenerative arthritis. Parasagittal MR image. Osteophytic change is present at the anterior aspect of the articular surface of the condyle; the articular disc is grossly distorted and lies immediately inferior to the articular eminence.

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

inflammatory arthritides may affect the joint, including systemic lupus, systemic sclerosis, psoriasis, Reiter’s syndrome, juvenile chronic arthritis, and synovial osteochondromatosis, which results in joint swelling and the presence of numerous loose calcific bodies (Fig. 63.47). Juvenile chronic arthritis occurs during the first 2 decades of life and is characterized by intermittent synovial inflammation. It causes pain and tenderness of one or both joints and if severe will affect mandibular growth. The condyle becomes radiolucent and its surface develops erosions and irregularity.

Injury Isolated fractures of the condylar neck can occur but often accompany fractures of the mandible, especially following a blow to the chin, and are usually visible on a dental panoramic radiograph and a PA condylar view.The slender condylar neck acts as a stress breaker, reducing the likelihood of the condyle being driven up into the middle cranial fossa. Fractures of the condylar neck may be simple and undisplaced or displaced with the condyle being pulled forwards and medially by the lateral pterygoid muscle (fracture dislocation). Intracapsular fractures are difficult to demonstrate on plain films and if suspected may require evaluation with CT if symptoms persist. Haemarthrosis and ankylosis may complicate recovery. Acute dislocation of the TMJ occurs following a blow to the mandible when the mouth is open. It is diagnosed from the clinical presentation. The role of radiology is to exclude a fracture or other contributing disease. Recurrent dislocation can develop spontaneously and may be a feature of Marfan’s syndrome and Ehlers–Danlos syndrome. Ankylosis of the TMJ may follow a traumatic haemarthrosis or infective arthritis. When this happens in childhood,

Figure 63.47 Osteochondromatosis of the TMJ. Axial T2W MR view showing mass draped around the right condyle.

Figure 63.48 Bony ankylosis of the left TMJ on axial CT. There are two bone fragments (arrows) between the mandibular condyle and the glenoid fossa, and there is partial bone union between the lateral bone fragment, the condyle, and the glenoid fossa. Both mandibular condyles are rotated due to the fractures of the condylar necks. The patient had developed permanent trismus following an accident in childhood. (Courtesy of Dr Otto Chan).

many cases are complicated by hypoplasia of the condyle secondary to concurrent damage to the epiphyseal growth centre. CT is required to show the extent of the ankylosis (Fig. 63.48). Neoplasms of the temporomandibular joint are uncommon and include osteoma, osteochondroma, chondrosarcomas and rarely, metastatic deposits.

SALIVARY GLAND DISORDERS There are three paired major salivary glands – the parotid, submandibular and sublingual glands – and many minor salivary glands supplying saliva to lubricate, cleanse and aid early digestion within the mouth. The parotid gland lies between the posterior border of the man-dibular ramus and the sternomastoid muscle attaching to the mastoid process. It is enclosed in deep cervical fascia and traversed by the retromandibular vein, external carotid artery, and facial nerve. The retromandibular vein is easily visible on all forms of cross-sectional imaging and indicates the plane of the dividing plexus of the facial nerve lying just laterally. In this plane lies the major intraglandular ductal system, and this divides the gland into a larger superficial and smaller deep portion. While tumours are more common in the superficial lobe, surgical approach to the deep lobe involves dissection of the nerve branches, with attendant risk of nerve damage. The parotid gland drains through Stensen’s duct, running horizontally forward approximately 1 cm below the zygomatic arch on the surface of masseter to turn medially, perforate the buccinator muscle, and emerge on a papilla on the buccal mucosa opposite the 1st maxillary molar tooth.The sharp sigmoid bend in the anterior portion of the parotid duct is a common site for impaction of small salivary stones and is the location of the proposed ‘buccinator window anomaly’, a possible obstructive phenomenon. The submandibular gland wraps around the posterior free border of the mylohyoid muscle, medial to the posterior body of mandible and descends for 2–3 cm into the suprahyoid

1449

1450

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

neck. The main Wharton’s duct passes up from the hilum of the gland, around the posterior margin of mylohyoid, turning anteriorly in the floor of the mouth to open through a small papilla situated on either side of the lingual frenum, behind the lower incisor teeth. The sublingual glands lie anteriorly in the floor of mouth above the mylohyoid muscle. Each gland opens by a single Bartolin’s duct or by multiple ducts into the floor of mouth or terminal part of Wharton’s duct.

Radiological techniques and their application Plain radiographs are of limited value. An intra-oral true mandibular occlusal view detects radio-opaque salivary calculi in the anterior submandibular duct, of which 60–80 per cent are radio-opaque, while a PA view of the cheek may detect the 20–40 per cent of parotid stones that are radio-opaque. A negative result does not preclude obstruction. Sialography has limitations in demonstrating parenchymal disease but remains a highly sensitive test of ductal abnormalities. Cannulation of the parotid duct is normally straightforward but the submandibular duct orifice may be very fine and difficult to identify. The sublingual duct and gland may only be incidentally demonstrated. Direct visualization of ductal filling under fluoroscopy, particularly using digital subtraction techniques (Fig. 63.49), has benefits over the traditional plain film method and water-soluble contrast media should be used to avoid the permanent foreign body reaction sometimes seen when oily media are extravasated from the ductal system. Sialography has no place in the investigation of mass lesions and is indicated primarily for symptoms directly related to the ductal system such as obstruction and sialectasis. Interventional sialography offers a minimally invasive alternative to formal surgical sialadenectomy or sialolithectomy. Small mobile stones may be extracted from the duct system by Dormia basket or balloon catheter, and duct strictures dilated by angioplasty balloon (Figs 63.50A, B; 63.51A, B)12,13.

Figure 63.49 Collection of calculi at hilum of parotid gland. There is minor sialectasis (irregularity of calibre of some intraglandular ducts).

Figure 63.50 (A) An intraoperative sialogram showing an open Dormia basket within the submandibular duct during removal of a salivary stone. (B) Resultant fragmented salivary stone removed from submandibular gland shown in A.

Ultrasound (US) has increasingly become the first-line investigation for masses within the salivary glands and can readily investigate sialadenitis to detect causative stones or resultant duct dilatation, features facilitated by administering a sialogogue prior to investigation (Fig. 63.52). It is highly sensitive for the 70–80 per cent of tumours within the superficial parotid gland when compared with CT, though has limitations when imaging the deep pole. CT is sensitive for the detection of salivary calculi though not resolving enough to show details of duct morphology in obstruction (Fig. 63.53). MRI has largely superseded it for tumour assessment though it can be useful in imaging early involvement of cortical bone.The technique of CT sialography has been displaced by MR sialography. MRI has major advantages for salivary gland imaging, particularly related to its high soft tissue contrast and multiplanar data acquisition. Gadolinium enhancement is not required in uncomplicated cases, but can be of major importance in the assessment of recurrent tumours (Fig. 63.54). Magnetic resonance sialography using heavily T2weighted sequences (Fig. 63.55) (use image of grossly dilated parotid duct) allows noninvasive assessment of obstructive disease by imaging stimulated ductal saliva and correlates well

CHAPTER 63

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

Figure 63.51 (A) Sialogram showing a diffuse stricture at the entrance to the hilum of the parotid gland. (B) Postoperative sialogram showing dilatation of duct stricture following balloon ductoplasty.

Figure 63.52 Ultrasound image of a salivary stone in the proximal portion of the submandibular duct.

with or even improves on conventional sialography in conditions such as Sjögren’s syndrome14,15. Radionuclide radiology is most commonly used to assess glandular function in obstruction and inflammatory conditions. Low resolution and lack of uptake of 99mTc sodium pertechnetate in adenolymphomas (Warthin’s tumours) limits its value in tumour imaging. Positron emission

Figure 63.53 Extensive benign pleomorphic adenoma of the left parotid deep lobe on axial T2W MR.

tomography has been used to image salivary gland tumours but, while being actively taken up by growing neoplasms, is also concentrated in lymphoid tissue and salivary glands. It has higher uptake in both malignant tumours and in the benign Warthin’s tumour16. The most commonly used Figure 63.54 Extensive recurrent parotid adenocarcinoma with intra-cranial extension. Pre- (A) and post- (B) gadolinium T1W coronal MR images. Note intracranial enhancement in (B) suggesting tumour invasion via the foramen spinosum.

1451

1452

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

conditions such as Sjögren’s syndrome (see later). Progression from widened and tortuous ductules to frank cavitation is seen (cavitatory sialectasis).

Inflammatory conditions

Figure 63.55 MR sialography image showing gross dilatation of the main duct and some of the secondary ducts. Areas of low signal in the main duct are due to the presence of several large stones. The distal part of the duct is normal. (Courtesy of Dr M. Becker, Geneva).

glucose analogue is secreted in saliva so small supraglottic, tongue base and laryngeal tumours and tumours measuring less than 1 mm may be missed. PET has strengths in assessing the post-treatment neck but may not distinguish tumour from acute infection or early wound healing17.

Calculi and duct strictures Calculi cause obstruction of the salivary glands typically resulting in mealtime-related swelling of the affected gland and predisposing to infection, sialectasis, and eventual gland atrophy. They are more common in the submandibular duct system (around 85 per cent) where 60–80 per cent are radioopaque and are found particularly at the genu of the submandibular duct – that portion of the duct making an acute bend over the posterior free border of the mylohyoid muscle to enter the gland hilum. Only 20–40 per cent of parotid calculi are radio-opaque and are normally detected in the parotid hilum or main duct overlying the masseter muscle. CT is highly sensitive for small stones but US is a convenient and effective way of detecting the majority of salivary calculi (89–94 per cent sensitivity, 100 per cent specificity), except those lying in the most anterior parts of each duct. Single and multiple calculi may be found in dilated duct segments and may be associated with distally placed strictures. Strictures of the salivary ducts result from inflammation caused by infection or calculi and are best demonstrated by sialography or MR sialography18. These may appear point or diffuse with proximal dilatation of the duct system. Approximately 25 per cent of salivary obstructions are due to strictures. ‘Sialadochitis’ describes the combination of duct dilatation and stenosis that follows obstruction complicated by infection.

Sialectasis Sialectasis develops in sialadenitis and radiologically demonstrates degenerative changes seen within the terminal salivary ducts and acini as a result of obstruction, infection and other

Infective sialadenitis of viral or bacterial origin causes generalized glandular enlargement on cross-sectional imaging with heterogeneous reduced echogenicity on US, increased attenuation on CT, and high signal on T2-weighted MRI (Fig. 63.56). Abscess formation is shown as an ill-defined hypoechoic area on US, with equivalent changes on CT and MR; acute inflammation is also evidenced by inflammatory stranding through the gland to the overlying tissues. Focal chronic inflammatory sialadenitis (Kuttner tumour) presents as a localized area of hypoechoic tissue in the submandibular gland on US and may be mistaken for tumour19. Sialography is indicated in cases of recurrent infection in order to demonstrate any underlying calculus or duct stricture. Sjögren’s syndrome causes damage to intercalated salivary duct walls allowing leakage of contrast media during sialography and creating a characteristically fine punctate sialectasis in approximately 70 per cent of cases. This is evenly distributed throughout salivary gland tissue – normally a parotid gland is chosen to demonstrate involvement. Similar abnormalities have been described in association with other connective tissue disorders such as rheumatoid arthritis, systemic lupus erythematosus, ankylosing spondylitis, Reiter’s disease, polyarteritis nodosa and scleroderma where, in the absence of clinical features of Sjögren’s syndrome, sialographic signs are estimated to exist in 5–15 per cent of cases. MR shows a speckled honeycomb appearance in moderately affected cases on both T1- and T2weighted images, which is said to be specific and similar appearances may be found in the lacrimal glands. MR sialography has shown improved sensitivity and 100 per cent

Figure 63.56 MR sialography image showing chronic sialadenitis of the parotid gland with focal globular high signal areas of sialectasis within the parenchyma of the gland. The main duct appears normal. The submandibular duct and gland (arrow) (seen later) appear normal. (Courtesy of Dr M. Becker, Geneva).

CHAPTER 63

specificity over conventional sialography in diagnosis of Sjögren’s syndrome14. US shows a heterogeneous reticular pattern of small low reflective foci and has a role in monitoring for lymphoma development, which may complicate late Sjögren’s syndrome. Sjögren’s syndrome sufferers have a 44 times greater risk of developing mucosal-associated lymphoid tissue (MALT) lymphoma. Sarcoid results in generalized glandular enlargement with multiple small granulomatous areas of high attenuation on CT (Fig. 63.57), low reflectivity on US and diffuse high signal on MR. There is, in common with all inflammatory conditions, high activity on 67Ga scintigraphy. Human immunodefiency virus (HIV)-associated salivary gland disease is a spectrum of disorders that affects approximately 20 per cent of children and 0.5 per cent of adults with HIV, and includes lymphoepithelial infiltration that may progress to lymphoma. The combination of multiple intra-parotid cysts (which are otherwise rare) and cervical lymphadenopathy should alert the radiologist to this syndrome, which can occur at any stage from early post infection to fullblown acquired immune deficiency syndrome (AIDS). Benign lymphoepithelial infiltration represents a spectrum from reactive to neoplastic change and can occur as an isolated abnormality, but is more commonly a feature of major salivary gland involvement in Sjögren’s syndrome (30–50 per cent). The CT appearance in advanced disease is highly suggestive, with grossly enlarged glands containing multiple well-defined round areas of low attenuation (Fig. 63.58). Ga-citrate scintigraphy has been advocated as a method of assessing the progression of this disease, which is complicated by malignant lymphoma in 5 per cent and anaplastic carcinoma in about 1 per cent of patients. Lymphadenopathy affecting intraparotid nodes is not unusual and may mimic primary salivary tumour.

Salivary gland tumours Most salivary gland tumours are benign and can develop at any age. Of these tumours, 80 per cent are found in the parotid glands, 5 per cent in the submandibular, 1 per cent in the sublingual, and the remaining 15 per cent in the minor salivary glands. The overall incidence of malignancy is 10–20 per cent, and the smaller the major gland the higher the rate of malignancy.

Figure 63.57 Sarcoid of the parotid glands. There is generalized glandular enlargement and multiple small areas of increased attenuation.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

Figure 63.58 Benign lympho-epithelial infiltration of the parotids. The glands are grossly enlarged and contain multiple areas of reduced attenuation.

Benign pleomorphic adenomas (benign mixed tumours) account for around 80 per cent of salivary tumours and typically arise in the superficial portion of the parotid gland, being most common in middle-aged women. These appear uni- or mildly loculated hypoechoic lesions on US and characteristically give a low signal on T1- but high signal on T2-weighted MRI. Adenocarcinomas are usually of distinctive histological appearance, the commonest being adenoid cystic carcinoma and muco-epidermoid carcinoma. These tumours grow slowly and are difficult to eradicate; the adenoid cystic type in particular have a propensity for insidious perineural spread. MRI can usually identify the presence, though not the extent, of perineural spread. Mucoepidermoid tumours are commonest in the minor salivary glands of the palate. Other salivary tumours include the benign adenolymphoma (Warthin’s tumour), lipoma, lymphoma, and tumours of the minor salivary glands. Warthin’s tumours are notably found in the parotid tail of older men, may be multiple (20 per cent) and occasionally bilateral (6.5 per cent). Lipomas give a characteristic hypoechoic appearance with numerous layered highly reflective internal strands on US and markedly low attenuation on CT. Distinction between benign and malignant tumours is based upon criteria, some of which are common to all crosssectional imaging, others being specific to a particular modality. Benignity is best identified by the presence of a capsule or well-defined outline, however notably many salivary malignancies are relatively low grade and are well defined in the early stages. Beyond this, CT may not be particularly discriminating, since many lesions show similarly increased attenuation, though more recently diffusion studies have allowed better differentiation. MRI has a sensitivity of about 75% for identifying benign features; this can be improved by using gaddinium enhancement. Ultrasound normally depicts benign lesions as homogenous, hypoechoic and without regional lymphadenopathy. Additionally, colour flow Doppler indicates vascularity and may

1453

1454

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

present evidence of neo-angiogenesis common in malignant tumours. Concurrent inflammatory change or haemorrhage can be confused with malignancy here. Contrast-enhanced MRI remains preferable for demonstrating recurrent tumour in areas that may be inaccessible to US and that give only nonspecific soft tissue change on CT. Whilst sialography is not recommended for the assessment of mass lesions, the incidental finding of distortion or amputation of intraglandular ducts should arouse concern as to the presence of a benign or malignant tumour, respectively.

Trauma Laceration of the main parotid duct or of a larger intraglandular branch occasionally results from a penetrating facial injury, or is a rare complication of surgery. In recent injury, US may detect a fluid collection and sialography will show extravasation of contrast from the duct system into the soft tissues. Later, healing frequently results in duct stenosis.

features in combination with US-guided fine-needle aspiration cytology (FNAC) gave 100 per cent accuracy compared with CT (77–89 per cent), MR (88 per cent) and US alone (83–98 per cent)22. Infection in the head and neck region is characterized by spread along fascial-bound compartments (mucosal, submandibular, parapharyngeal, carotid, masticator, retropharyngeal, and prevertebral). CT or MRI are useful for the assessment of such spread (Fig. 63.59). Malignant tumours within the oral cavity and environs are predominantly squamous cell carcinomas. MRI has become the first-choice examination for imaging both the extent of the local tumour and any regional lymphadenopathy, but may be oversensitive in the assessment of recurrent disease. MRI is also useful for the assessment of marrow involvement and is steadily encroaching on CT’s superiority in demonstration of cortical bone involvement from tumours such as those in the floor of the mouth (Fig. 63.60). The imaging of

Disorders of function Salivary gland function may be quantified by time–activity curves with 99mTc-pertechnetate scintigraphy and may distinguish between the functioning, the obstructed and the nonfunctional gland. This may supplement sialography or be undertaken when sialography is not possible, and has been used to demonstrate recovery of salivary function following removal of an obstructing calculus20.

SOFT TISSUES Recent improvements in near-field imaging, and in colour flow and power Doppler performance, combined with its established values of high soft-tissue contrast and ease of use in guided biopsy, have increasingly led to US being seen as the first-line investigation for masses in the superficial soft tissues of the maxillofacial region.The potential for operator variability with US is one reason why axial CT remains widely used, as it demonstrates most lesions and assesses their relationship to adjacent structures, although IV contrast is often required for the accurate assessment of cervical lymphadenopathy. MRI, by virtue of its high soft-tissue contrast and multiplanar capabilities, will show most masses with similar or greater ease and has the additional advantage that different sequences may contribute information about the nature of a lesion. US improves on clinical examination of the parotid, submandibular and cervical regions, and is a rapid and accurate means of distinguishing between cervical lymphadenopathy, salivary and other soft tissue neck lesions. Patterns suggestive of malignant involvement of cervical lymph nodes include a round shape (a short axis measurement over 1 cm becoming significant), absence of hilus, irregular outline, heterogeneous internal pattern including coagulation or cystic necrosis, disorganized peripheral colour flow pattern on Doppler US, nodal clusters, and fusion of nodes21. US has been found to be better than CT at detecting malignant cervical nodes but has the disadvantage of being unable to access deep nodes such as those within the retropharyngeal region. Identification of these

Figure 63.59 Abscess in the right oropharyngeal region on axial CT. The patient had received antibiotic treatment for recurrent throat infections and sterile pus was subsequently aspirated from the lesion.

Figure 63.60 Carcinoma of the left side of the tongue, T2-weighted axial MRI. The high signal tumour extends to the midline.

CHAPTER 63

primary lesions is relatively straightforward, but controversy remains surrounding the place of imaging in the assessment of lymph node involvement. Some studies state categorically that MRI lacks sufficient sensitivity and specificity to replace elective neck dissection for both staging and prognostic purposes23, but more recently imaging has shown increased sensitivity in detecting nodal metastases using US and combinations of PET and sentinel node biopsy. MRI has successfully predicted those patients in need of neck dissection24, and has successfully revealed micrometastases (described as intranodal tumour deposits of less than 2 mm diameter at any level of sectioning), which may be undetectable by cross-sectional imaging23,25. Positron emission tomography (PET), utilizing either 18F-fluorodeoxyglucose (FDG-PET) or 11C-tyrosine (TYR-PET) has a similar sensitivity (72 per cent) to CT (89 per cent) for the detection of lymph node metastases, but this may be improved by co-registration with PET (96 per cent sensitivity, 98.5 per cent specificity), or by supplementary sentinel node biopsy24,26. PET and PET/CT currently viewed as a particularly promising technique for the detection of occult primary lesions and recurrent tumours, especially in the post-radiotherapy situation. Recurrent neck disease, even when subclinical, may however be predicted with US and US-guided FNAC. The propensity for regional metastases from squamous cell carcinomas to present late after treatment of the primary lesion is an indication for prolonged follow-up (Fig. 63.61). Lymphomas may arise in Waldeyer’s ring and the salivary glands, including the minor glands. Benign tumours that require imaging in adults are relatively rare other than some dermoids, which present at lines of embryonic fusion; vascular abnormalities, which may grow extremely large leading to secondary growth disturbances; and salivary gland tumours, as already discussed. Colour flow Doppler US and magnetic resonance angiography may be helpful in assessment, but conventional angiography may be necessary, particularly prior to embolization. Phleboliths within such lesions may be apparent on plain films.

Figure 63.61 Metastatic deposit below mandible presenting 7 years post surgical excision of small ipsilateral lower lip squamous carcinoma.

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

A wide spectrum of lesions present in the newborn, infants and children, often benign vascular abnormalities (Figs 63.62, 63.63) but other hamartomas (Fig. 63.64) and malignant tumours may be seen. Thyroglossal and branchial cysts have characteristic positions. Thyroglossal cysts arise from epithelial tissue trapped during the embryonic descent of the thyroid gland, and present as defined midline cystic structures lying on a line between the base of the tongue and the thyroid gland (Fig. 63.65). Branchial cysts arise from epithelium trapped during incorporation of the second branchial arch, and presents as ovoid fluid-containing lesions lying deep to the sternomastoid muscle, protruding anterior to its anterior border. Masseteric enlargement is usually secondary to muscular hypertrophy, which may be unilateral or bilateral, and often

Figure 63.62 Axial MRI of an arterio-venous malformation within the masseter muscle of a 10-year-old. Low signal areas represent calcific deposits and flow voids, suggest a high flow lesion.

Figure 63.63 Lymphangioma in an infant. Coronal STIR MRI.

1455

1456

SECTION 7 •

NEURORADIOLOGY, INCLUDING THE HEAD AND NECK

Figure 63.64 Fibroma of the tongue in an infant. Sagittal T1-weighted MRI.

Figure 63.65 Thyroglossal cyst at base of tongue. Sagittal T1-weighted MRI.

concurrently involves the pterygoid muscles. US is valuable for diagnosis27. Rare causes include malignant lymphoma, leukaemic infiltration, haemangioma and rhabdomyosarcoma. Calcification is occasionally seen in the walls of the facial and lingual arteries in patients with hypercalcaemia or renal failure. Small areas of subcutaneous calcification have been reported in Gorlin’s syndrome and Ehlers–Danlos syndrome.

REFERENCES 1. Faculty of General Dental Practitioners (UK) 2004 Selection Criteria for Dental Radiography, 2nd edn. The Royal College of Surgeons of England, London 2. Browne R M, Edmondson H D, Rout P G J 1995 Atlas of Dental and Maxillofacial Radiology. Mosby-Wolfe, London 3. Yoshiura K, Higuchi Y, Ariji Y et al 1994 Increased attenuation in odontogenic keratocysts with computed tomography: a new finding. Dentomaxillofac Radiol 23: 138–142

4. Kawai T, Murakami S, Kishino M et al 1998 Diagnostic imaging in two cases of recurrent maxillary ameloblastoma: comparative evaluation of plain radiographs, CT and MR images Br J Oral Maxillofac Surg 36: 304–310 5. Verneuil A, Sapp P, Huang C et al 2002 Malignant ameloblastoma: classification, diagnostic, and therapeutic challenges. Am J Otolaryngol 23: 44–48 6. Gray C, Redpath T W, Smith F W 1998 Low-field magnetic resonance imaging for implant dentistry. Dentomaxillofacial Radiol 27: 225–229 7. Lee Y Y, van Tassel P, Nauert C et al 1988 Craniofacial osteosarcomas: plain film, CT, and MRI findings in 46 cases. Am J Roentgenol 150: 1397–1402 8. Jank S, Emshoff R, Etzelsdorfer M et al 2004 Ultrasound versus computed tomography in the imaging of orbital floor fractures. J Oral Maxillofac Surg 62: 150–154 9. Tasaki M M, Westesson P L 1993 Temporomandibular joint: diagnostic accuracy with sagittal and coronal MRI imaging. Radiology 186: 723–729 10. Segami N, Suzuki T, Sato J et al 2003 Does joint effusion on T2 magnetic resonance images reflect synovitis? Part 3. Comparison of histological findings of arthroscopically obtained synovium in internal derangements of the temporomandibular joint. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 95: 761–766 11. Wilson D J 1990 Imaging. In: Norman J E de B, Bramley P (eds). A Textbook and Colour Atlas of the Temporomandibular Joint. Wolfe, London, pp 90–109 12. Brown J E, Drage N A, Escudier M P et al 2002 Minimally invasive radiologically guided intervention for the treatment of salivary calculi. Cardiovasc Intervent Radiol 25: 352–355 13. Drage N A, Brown J E, Escudier M P et al 2002 Balloon dilatation of salivary duct strictures – report on 36 treated glands. Cardiovasc Intervent Radiol 25: 356–359 14. Niemela RK, Takalo R, Paakko E et al 2004 Ultrasonography of salivary glands in primary Sjögrens syndrome. A comparison with magnetic resonance imaging and magnetic resonance sialography of parotid glands. Rheumatology 43: 875–879 15. Ohbayashi N, Yamada I, Yoshino N, Sasaki T 1998 Sjogren syndrome: comparison at assessments with MR sialography and conventional sialography. Radiology 209: 683–688. 16. Uchida Y, Minoshima S, Kawata T et al 2005 Diagnostic value of FDG PET and salivary gland scintigraphy for parotid tumors. Clin Nucl Med 30: 170–176 17. Jones J, Farag I, Hain S F, McGurk M 2005 Positron emission tomography (PET) in the management of oro-pharyngeal cancer. Eur J Surg Oncol 31: 170–176 18. Becker M, Marchal F, Becker C D et al 2000 Sialolithiasis and salivary ductal stenosis: diagnostic accuracy of MR sialography with a threedimensional extended-phase conjugate symmetry rapid spin-echo sequence. Radiology 217: 347–358 19. Ahuja A T, Richards P S, Wong K T et al 2003 Kuttner tumour (chronic sclerosing sialadenitis) of the submandibular gland: sonographic appearances. Ultrasound Med Biol 29: 913–919 20. Makdissi J, Escudier M P, Brown J E et al 2004 Glandular function after intraoral removal of salivary calculi from the hilum of the submandibular gland. Br J Oral Maxillofac Surg 42: 538–541 21. Evans R M, Ahuja A, Rhys Williams S et al 1991 Ultrasound and ultrasound guided fine needle aspiration in cervical lymphadenopathy. Br J Radiol 64(P): 58 [Abstract] 22. Ying M, Ahuja A, Brooks F 2004 Accuracy of sonographic vascular features in differentiating different causes of cervical lymphadenopathy. Ultrasound Med Biol 30: 441–447 23. Wide J M, White D W, Woolgar J A et al 1999 Magnetic resonance imaging in the assessment of cervical nodal metastases in oral squamous cell carcinoma. Clin Radiol 54: 90–94 24. Kovacs A F, Dobert N, Gaa J et al 2004 Positron emission tomography in combination with sentinal node biopsy reduces the rate of elective neck dissections in the treatment of oral and oropharyngeal cancer. J Clin Oncol 22: 3973–3980

CHAPTER 63

25. El-Sayed I H, Singer M I, Civantos F 2005 Sentinel lymph node biopsy in head and neck cancer Otolaryngol Clin North Am 38:145–160, ix–x 26. Schwartz D L, Ford E, Rajendran J et al 2005 FGD-PET/CT imaging for preradiotherapy staging of head-and-neck squamous cell carcinoma. Int J Radiat Oncol Biol Phys 61: 129–136 27. Morse M, Brown E F 1990 Ultrasonic diagnosis of masseteric hypertrophy. Dentomaxillofacial Radiol 19: 18–20

•

DENTAL AND MAXILLOFACIAL RADIOLOGY

FURTHER READING Ahuja A T, Evans R M 2000 Practical Head and Neck Ultrasound, Radionuclide Radiology, Greenwich Medical Media Limited White S C, Pharoah M J 2004 Oral Radiology Principles and Interpretation, 5th edn Mosby

1457

CHAPTER

The Neonatal and Paediatric Chest

64

Isla Lang and Alan Sprigg

The neonatal chest • Radiation risks and radiological techniques • The normal chest, variations and artifacts • Normal lung development • Neonatal respiratory distress The older paediatric patient • Technical factors • The thymus • Tachypnoea—respiratory or cardiac cause?

• • • • • • • •

Infection The radio-opaque chest Imaging the child on the intensive care unit The wheezy chest Congenital chest abnormalities Cystic fibrosis Paediatric interstitial disease Immunodeficiency

THE NEONATAL CHEST Isla Lang

RADIATION RISKS AND RADIOLOGICAL TECHNIQUES ‘Radiation exposure in the first 10 years of life is estimated to produce a risk of total aggregated detriment 5–7 times greater than exposure after the age of 50’1. Radiation in the neonate should be kept to a minimum and the radiographs performed by experienced radiographers. Routine lateral views of the chest are unnecessary. A short exposure time, high kV–low mAs technique with fast film–screen combinations should be used. Accurate collimation of the light beam and lead masking on top of the incubator should be performed. Gonad shielding should be used if an abdominal radiograph is also being performed2.

lung opacities (Fig. 64.1).The ribs have a more horizontal orientation with the usual level of the diaphragm at the sixth to eighth anterior rib. The normal cardiothoracic ratio can be as large as 65% due to the presence of the thymus. The size of the thymus can be highly variable. A normal thymus does not compress or displace other structures (Fig. 64.2). The thymus

THE NORMAL CHEST, VARIATIONS, AND ARTEFACTS The normal appearance of the neonatal chest differs from that of older children. It is almost cylindrical in shape. Small degrees of rotation lead to considerable malprojections of the anterior ribs and sternum. On a rotated chest radiograph sternal ossification centres may simulate healing rib fractures or

Figure 64.1 Rotated radiograph showing the sternal ossification centres simulating healing rib fractures (arrows).

1462

SECTION 8

• PAEDIATRIC IMAGING

Evaluation of tubes and lines

Figure 64.2 Prominent thymus in a premature baby with mild respiratory distress syndrome.

may involute rapidly in association with prenatal or postnatal stress or following the administration of exogenous steroids. It is important to recognize certain artefacts on a neonatal radiograph. The hole in the top of the incubator may mimic a lung cyst or pneumatocele. Redundant skin, particularly in the preterm neonate, can result in a long vertical skinfold which can simulate a pneumothorax (Fig. 64.3). Deep retractions of the lower sternum in respiratory distress can produce a central radiolucency that simulates a pneumomediastinum on frontal radiographs3.

Figure 64.3

Skinfold (arrows) simulating a pneumothorax.

All tubes, catheters, and lines should be evaluated on each radiograph carried out on an infant as they are often the major reason for taking the radiograph. The trachea in the infant is often short and narrow. Optimal positioning for an endotracheal tube (ETT) is approximately 1–1.5 cm above the carina. Considerable changes in ETT position can occur because of head and neck movement. Flexion moves the ETT inferiorly, when the tube may extend into the right main bronchus. Radiographic clues of malposition of the ETT in the oesophagus, for example, are: low position, a tracheal column distinct from the ETT, pulmonary underaeration, air in the distal oesophagus, and gaseous distension of the gastrointestinal tract. Umbilical arterial and venous lines should be differentiated from each other. This is normally possible on an antero-posterior (AP) view of the chest and abdomen but occasionally a lateral view may be required. The umbilical arterial line initially courses caudally through the internal and common iliac arteries to enter the aorta, and lies just lateral to the left side of the spine. The tip should ideally lie between T6 and T10 to avoid the spinal arteries, or at L3–L5 below the level of the bowel and renal arteries. The umbilical vein catheter courses directly cephalad on the right side of the abdomen and enters the left portal vein, at which point it may enter the ductus venosus and then the inferior vena cava. Radiographs should confirm that the tip lies above the liver and has not passed into a tributary vein (Fig. 64.4). Central lines and peripheral vascular lines extending from the periphery also need to be evaluated.These are often small and the position of

Figure 64.4 Tubes and lines. The tip of the umbilical arterial line is at T3 which is high and the top of the umbilical venous line (arrow) lies within the liver. The positions of both were altered following this radiograph.

CHAPTER 64

the tip may be difficult to determine. In this situation a radiograph following contrast medium in the line may be helpful. Feeding may be through a nasogastric or a nasojejunal tube. The position of the tips of these tubes should be recorded as infusion of enteral feeds into a malplaced tube may have serious consequences. Perforation of the oesophagus by a nasogastric tube in the neonate may cause very few symptoms3,4.

NORMAL LUNG DEVELOPMENT The pulmonary tract develops in the third week after conception as an outpouching from the primitive foregut. By the sixth week primitive lung buds are present. Between 6 and 16 weeks there is extensive branching of the respiratory tree. Between 16 and 28 weeks, multiple alveolar ducts arise from the respiratory bronchioles and the primitive alveoli form. True alveolar development extends from approximately 26 weeks gestation to term and continues for the first 2 years after birth, after which alveoli increase in size but not in number. Normal respiration requires production of surfactant. The lamellar bodies that store surfactant appear at around 22 weeks gestation. Surfactant is produced within the endoplasmic reticulum of type 2 pneumocytes within the alveoli from about 24 weeks gestation. The pulmonary surfactant lowers alveolar surface tension. Glucocorticoids and thyroid hormones accelerate surfactant release towards term5.

NEONATAL RESPIRATORY DISTRESS A neonate with respiratory distress clinically presents with tachypnoea, rib recession, and grunting. There are many causes of respiratory distress and a chest radiograph can help to differentiate between them. These include primary pulmonary disease, which can be divided into those that can be treated medically and those requiring surgery. In general, medical conditions in the newborn cause bilateral pulmonary disease, whereas surgical problems are often unilateral and frequently produce contralateral shift of the mediastinum. Cardiac disease is an important cause of respiratory symptomatology in the neonate, for which ultrasound (US) is particularly useful. Extrapulmonary lesions are also important and should be suspected if the chest radiograph is normal or only mildly abnormal in the presence of respiratory distress. These include upper airway obstruction, musculoskeletal abnormalities of the chest, lesions of the diaphragm and abdomen, as well as distant pathology in the central nervous system or metabolic or haematological derangements3.

• THE NEONATAL AND PAEDIATRIC CHEST

normal respiration ensues. If this process is impaired, ‘wet lung’ or transient tachypnoea of the newborn (TTN) develops. It is commonly associated with delivery by caesarean section, prematurity, and some cases of maternal diabetes. The infant is usually tachypnoeic soon after birth and has mild to moderate hypoxia. Radiographs demonstrate prominent pulmonary interstitial markings, fluid in the interlobar fissures and intrapleural space, and slight overaeration (Fig. 64.5). In severe cases the radiographic appearance is that of alveolar oedema or a reticular granular appearance similar to that of respiratory distress syndrome of the newborn, with the only difference being that pulmonary aeration is normal to slightly increased. The changes are usually symmetrical but sometimes the right side is affected more than the left3. The major diagnostic feature is the mild symptomatology, with clinical and radiographic resolution occurring by 48–72 h of age.

Respiratory distress syndrome Respiratory distress syndrome (RDS) remains the most common life-threatening respiratory disorder of newborns. Most neonates with RDS are premature. Approximately 50% of neonates born between 26 and 28 weeks and 20–30% of neonates born at 30–31 weeks of gestation develop RDS6. It occasionally occurs in infants older than 36 weeks gestation. Other risk factors include infants of poorly controlled diabetic mothers, fetal asphyxia, maternal or fetal haemorrhage, and multiple gestations. It is more common and more severe in boys and occurs more commonly in black than in white children. RDS is due to a deficiency of alveolar surfactant.The initial response to a deficiency of surfactant is for the smaller alveoli to

Transient tachypnoea of the newborn Before delivery, the lungs are filled with amniotic fluid. Approximately one-third of fetal lung fluid is cleared out of the airways by the thorax being squeezed in the birth canal, one-third is absorbed by the pulmonary capillaries, and onethird by the lymphatics. Within the first few breaths, it is rapidly cleared and the lungs quickly become fully aerated as

Figure 64.5 Baby with transient tachypnoea of the newborn. Chest radiograph shows hyperinflated lungs, streaky shadowing, and a little fluid in the horizontal fissure.

1463

1464

SECTION 8

• PAEDIATRIC IMAGING

collapse while the larger ones hyperinflate. The lungs become stiff and increased respiratory effort is needed to inflate the lungs. The lung tissue becomes progressively traumatized and there is exudation of plasma from the pulmonary capillaries into the alveolar space. Secondary influx of white cells into the plasma exudate leads to organization of the plasma proteins with the development of a thick membrane composed of protein, white cells, and inflammatory debris.This membrane stains pink with standard histological preparations and gives rise to the alternative name for the condition—hyaline membrane disease. Clinically, infants with RDS will have respiratory distress which worsens during the first 18–24 h of life, with gradual improvement generally starting by the third day. Advances in perinatal medical management that have significantly modified the natural history of RDS include antenatal corticosteroid administration, surfactant replacement therapy, and ventilatory assistance. The chest radiograph will usually be abnormal at 6 h. There is normal to decreased aeration of the lungs as compared to TTN, where there is increased aeration. Initially the patent terminal airways are surrounded by airless alveoli, giving the characteristic fine reticular shadowing seen throughout both lungs with accentuation of the air bronchograms. As the condition progresses, influx of plasma renders the lungs more radio-opaque, the reticulogranular shadowing becomes more confluent, and there is progressive loss of clarity of the diaphragmatic and cardiac contours. Uncomplicated RDS is a bilaterally symmetrical disease: there may be some gradation in radiographic opacification between the upper and lower zones. Asymmetric changes may be seen when some areas have been differentially aerated by, for example, a misplaced ETT, asymmetric surfactant administration, and presence of localized pathology such as interstitial emphysema, pulmonary haemorrhage, or superimposed infection. Clearance of the lungs will depend on how quickly the individual baby is able to synthesize

adequate amounts of endogenous surfactant and may take from 1–2 d to several weeks3–5,7 (Fig. 64.6). Very small infants, less than 25 weeks gestation, often do not develop typical RDS. Their lungs initially appear well aerated radiographically but they often have major problems that may involve the respiratory system, such as apnoea of prematurity, lobar atelectasis, pneumonia, patent ductus arteriosus, and intraventricular haemorrhage. The ventilatory pressures used are often low and complications such as chronic lung disease are relatively common3.

Lung problems associated with neonatal intensive care and management of respiratory distress syndrome Surfactant therapy Surfactant acts synergistically with antenatal corticosteroid therapy to reduce the severity of RDS. Post treatment, the radiographs usually show a marked uniform improvement in lung aeration. Asymmetrical or focal improvement of lung aeration may occur, usually in the central or upper regions of the right lung (Fig. 64.7). Possible explanations for asymmetrical improvement are maldistribution due to malapplication of synthetic surfactant, and focal clearing due to insufficient dose of surfactant. In asymmetric clearing, it is important to exclude a pneumothorax or pneumomediastinum. Pulmonary haemorrhage may occur following surfactant treatment8,9.

Ventilation Many babies receiving intensive care will require some form of positive pressure artificial ventilation. This may be in the form of a moderate elevation of airway pressure through nasal prongs or a facemask (continuous positive airway pressure [CPAP]) or by a cyclical inflation of the lungs through an ETT (intermittent positive pressure ventilation [IPPV]). Pressures in the range of 2–6 cmH2O are used during CPAP and peak pressures between 12 and 26 cmH2O are commonly used during IPPV.

Figure 64.6 Preterm baby with respiratory distress syndrome. (A) Day 1: mild ground-glass appearance within both lungs. (B) Day 3: worsening opacification with air bronchograms.

CHAPTER 64

Figure 64.7 Surfactant therapy. Asymmetric improvement of lung aeration has occurred following surfactant treatment.

Normally ventilation rates approximate the normal respiratory rates of the infant (30–90 breaths min−1) but on occasion, oscillatory ventilation rates of 600–900 cycles min−1 may be utilized. High-frequency ventilation allows adequate gas exchange at lower peak inspiratory pressures and complications of barotrauma resolve more quickly than with conventional ventilation. The efficiency of high-frequency oscillatory ventilation is normally measured by radiological assessment of the degree of lung filling. Frequent chest radiographs may therefore be required for monitoring treatment progress. The right diaphragm should be seen to lie at a level between the posterior ends of the eighth and ninth ribs4,5. Early effects of ventilation and respiratory distress syndrome. Mechanical ventilation is an important risk factor contributing to air leak in premature babies treated for lung disease. Leakage of air from the pulmonary parenchyma into the pleural cavity is the most common, leading to a pneumothorax (Fig. 64.8). The

Figure 64.8 Severe respiratory distress syndrome with bilateral pneumothoraces.

• THE NEONATAL AND PAEDIATRIC CHEST

pneumothorax is often under tension, resulting in contralateral shift of the mediastinum. The pleural air may collect anywhere within the pleural space. Large collections are easily seen. Frequently the pleural air lies anterior and medial to the lung and is more difficult to diagnose, the only sign being increased radiolucency of the ipsilateral hemithorax. There is often increased sharpness of the mediastinal border which, unlike in a pneumomediastinum, extends from the superior extent of the lung to the diaphragm. The thymus is compressed by the pneumothorax rather than being elevated by it, as in a pneumomediastinum. Pleural air may frequently collect at the apex or base of the lung or within an interlobar fissure. Leakage into other adjacent structures may result from a pneumomediastinum when air may dissect outside the mediastinum into the neck, or underneath the lung into an extrapleural location, or into the retroperitoneal space or peritoneal cavity. Less commonly, a pneumopericardium or diffuse vascular air embolism may occur. Air may also leak into the interstitial space and spread throughout the lymphatics and along the perivascular sheaths, producing pulmonary interstitial emphysema (PIE). The radiological appearance is of small bubbles of air radiating out from the hilum. When severe, the lungs are overinflated and cardiac compression may occur (Fig. 64.9). Treatment is difficult; high-frequency ventilation may reduce the incidence of pneumothorax but in some cases PIE may become worse due to further air trapping. Unilateral PIE may improve with decubitus positioning, with the affected side dependent: occasionally obstruction of the bronchus to the affected portion of the lung is necessary. The condition may be associated with later chronic lung disease4,10. Late effects of ventilation and respiratory distress syndrome Bronchopulmonary dysplasia (BPD) or chronic lung disease of prematurity (CLD) is a chronic lung disease that develops in infants treated with positive pressure

Figure 64.9 Pulmonary interstitial emphysema. Small bubbles of air radiate from the left hilum after leaking into the interstitial space. The left lung is overinflated.

1465

1466

SECTION 8

• PAEDIATRIC IMAGING

mechanical ventilation and oxygen. Since the original description of BPD by Northway et al in 1967, the epidemiology has changed significantly11. New treatments and technologies have improved the outcome for premature infants and as a result the number of babies affected by BPD has increased and many need continuous domiciliary oxygen. Classically BPD has been defined clinically as oxygen dependency at 28 days of age associated with an abnormal chest radiograph. However, typical histological changes of BPD may be present at autopsy in infants dying within the first week of life.The above definition also fails to account for differences in gestational age. BPD therefore represents a continuum of lung damage and repair that often begins before the diagnosis is traditionally established. Chronic lung inflammation is caused by a number of factors: barotrauma due to positive pressure ventilation, oxygen toxicity, infection, altered inflammatory response, deficiency of antioxidant defences, and many others. The contribution of different factors varies between patients and a number of less mature infants may develop BPD without ventilatory support or an initial oxygen requirement. Pathologically the condition is characterized by areas of hyperexpansion and atelectasis interspersed with patchy areas of fibrosis. Hyperexpansion may be so severe as to result in moderate-sized pulmonary cysts, and fibrosis may vary from scattered strands of fibrous tissue to complete fibrosis and collapse of entire lobes. Characteristic radiographic appearances are patchy or linear strands of increased density with localized areas of unequal aeration and generalized hyperaeration (Fig. 64.10). Cardiomegaly may occur in severe cases and signifies pulmonary hypertension. The chest radiograph may return to normal in babies whose disease remains mild. Severe BPD may be fatal or lead to debilitating chronic pulmonary insufficiency. Follow-up in babies with BPD shows an increased number of later respiratory infections and an increased incidence of reactive airway disease.

Wilson–Mikity syndrome is described in immature infants who are initially well and do not require ventilation, but who show signs of respiratory distress in the second week12,13. The lungs develop streaky opacification and small cystic lucencies throughout. Respiratory failure may be progressive and symptoms may persist for many months.

Pulmonary agenesis and hypoplasia The interaction between the branching airway and its investing lung is essential for normal lung development. Interference with this development by intrathoracic or extrathoracic compression results in pulmonary agenesis or hypoplasia. Bilateral pulmonary hypoplasia is associated with oligohydramnios which occurs in renal agenesis (Potter’s syndrome) or following early rupture of membranes. Other extrathoracic causes include marked distension of the abdomen by a mass, neuromuscular disease of the chest, or skeletal dysplasias, such as asphyxiating thoracic dystrophy. Intrathoracic causes are usually unilateral, such as congenital diaphragmatic hernia, congenital cystic adenomatoid malformation, or pleural effusions. The severity of the hypoplasia depends on when the compression occurs in utero. The diagnosis is made on the history and plain radiograph findings. It should be suspected when ventilation is much harder than expected. The plain radiograph may show an underlying lesion or the chest may be bell shaped; pneumothoraces may occur as a complication.

Pulmonary haemorrhage Small areas of haemorrhage are frequently seen at autopsy in association with other neonatal lung problems such as RDS, congenital heart disease, aspiration pneumonia, and BPD. Radiographically it is often difficult to distinguish haemorrhage from the underlying disease. Massive pulmonary haemorrhage develops suddenly with the production of blood-stained secretions. On the radiograph there is diffuse homogeneous opacification (Fig. 64.11). In very small premature or severely asphyxiated infants, this is often a terminal event. Less severe pulmonary haemorrhage resolves over several days.

Meconium aspiration syndrome

Figure 64.10 Chronic lung disease. The lungs are hyperexpanded with areas of hyperinflation interspersed with areas of fibrosis. Both lungs were equally affected.

Infants who suffer hypoxic stress in utero may pass meconium into the amniotic fluid which is then inhaled. Meconium consists of hyperosmolar, viscid intestinal secretions. Inhalation of small amounts may be innocuous, while inhalation of large amounts results in patchy widespread collapse and consolidation combined with a severe inflammatory reaction. The viscous inhaled meconium may cause complete bronchial obstruction or a partial occlusion with a ‘ball valve’ effect that may lead to areas of hyperinflation in the peripheral lung. The chest radiograph shows patchy areas of collapse and consolidation with areas of hyperinflation (Fig. 64.12). Pneumothorax and pneumomediastinum are frequent complications which may result in hypoxia and lead to pulmonary artery vasoconstriction, pulmonary hypertension, and rightto-left shunting across the ductus arteriosus (persistent fetal circulation). The best treatment is immediate removal of the

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

ture rupture of membranes are particularly at risk. After the first week of life, hospital infections caused by Gram-negative organisms and Staphylococcus aureus occur more commonly.Viral infections also may cause severe and life-threatening disease. The diagnosis is clinical as radiographic findings vary. The spectrum includes streaky shadows resembling transient tachypnoea of the newborn, perihilar and patchy pulmonary opacities resembling aspiration, and homogeneous opacification resembling RDS. Some pneumonias have characteristic patterns. Pleural effusions frequently occur in group B streptococcal infection. Pneumatocele formation is uncommon in the neonate but may be seen in association with E. coli, Haemophilus influenzae, and now less frequently with Staph. aureus. Chlamydial infection is usually acquired via the birth canal but does not produce pulmonary disease until 4–6 weeks. The baby is often afebrile, may have conjunctivitis and a cough, and the radiograph shows overinflation with marked bilateral symmetric interstitial changes. Viral infections may show marked inflammatory changes with evidence of oedema. Figure 64.11 Pulmonary haemorrhage. Bilateral diffuse opacification within the lungs in a preterm infant. Both lungs were equally affected.

meconium by suction of the airways using an ETT at the time of delivery. If this is not possible, management is often difficult and recovery is often slow. Occasionally treatment with extracorporeal membrane oxygenation (ECMO) is required. The overall mortality is up to 25%.

Neonatal pulmonary infection Infections can be acquired transplacentally, the most common organisms are TORCH (toxoplasma, rubella, cytomegalovirus, and herpes); others include listeriosis, tuberculosis, and congenital syphilis. Peripartum infection is usually by the inhalation of infected amniotic fluid or maternal tract secretions (group B streptococcus, Esherichia coli). Babies born following prema-

Figure 64.12 Meconium aspiration. Bilateral collapse and consolidation within the lungs in a term infant.

Pleural effusions Pleural effusions are an uncommon cause of respiratory distress. They may develop as part of a generalized oedema, as in hydrops fetalis, fetal anaemia, and fetal cardiac failure, or in response to local inflammation following perinatal infections, or due to a defect of lymphatic drainage of the pleural spaces. Chylothorax is the most common cause of an isolated pleural effusion and predominantly affects the right hemithorax. Many different aetiologies have been implicated including congenital anomalies and birth trauma.

Congenital diaphragmatic hernia Congenital diaphragmatic hernia occurs in 1 in 2500 live births. The most common site is posterolateral, to which the inaccurate but established term Bochdalek hernia is applied. It involves the left diaphragm in 70% of cases. Anterior herniation through or adjacent to the foramina of Morgagni usually presents later. The diaphragm initially develops as an incomplete septum in the region of the lower cervical vertebrae and then migrates caudally to produce the pleural spaces. The diaphragmatic septum is derived from several separate elements which fuse between the sixth and seventh weeks of gestation to close the posterolateral diaphragmatic defects that are initially present. Failure of closure allows the developing intra-abdominal contents to become displaced into the thoracic cavity and results in congenital diaphragmatic hernia. In many affected infants the condition is compounded by severe respiratory difficulty secondary to pulmonary hypoplasia, persistent fetal circulation, and a degree of surfactant deficiency. Antenatal US screening allows early diagnosis, counselling, and appropriate labour ward management. Typical postnatal radiograph findings show an opaque hemithorax with deviation of the mediastinum away from the lesion. A nasogastric tube within the stomach, as well as deflating the bowel, is a useful marker for determining the degree of mediastinal shift and the position of the stomach. Once the gastrointestinal

1467

1468

SECTION 8

• PAEDIATRIC IMAGING

tract starts to fill with air, radiolucencies will be seen in the affected hemithorax and there will be progressive deviation of the mediastinum (Fig. 64.13). In addition to the upper alimentary tract, parts of the colon, spleen, kidney, and pancreas can herniate into the chest and their position can be identified by US examination. Malrotation and malfixation of the small bowel are associated problems. The presence of the stomach in the thorax is usually associated with earlier herniation and more severe pulmonary hypoplasia. Radiographically, the differential diagnosis includes congenital cystic adenomatoid malformation and pneumatocele formation. The treatment is surgical repair. The neonate normally undergoes a period of stabilization and assisted ventilation for 24–48 h before correction of the defect is performed. ECMO is occasionally required to support the neonate. Usually congenital diaphragmatic hernia is isolated. Other associated anomalies include congenital heart disease, limb deformities, extra pairs of ribs, and lung sequestration.

Congenital cystic adenomatoid malformation and bronchopulmonary sequestration Congenital cystic adenomatoid malformation (CAM) is a rare lesion characterized by a multicystic mass of pulmonary tissue with proliferation of bronchiolar structures. CAMs are usually unilobar, usually communicate with the normal tracheobronchial tree, and receive their blood supply from a normal pulmonary artery and vein. The natural history and prognosis of these lesions are extremely variable. The prognosis depends on the size rather than the type of lesion, larger lesions having a worse prognosis. Lesions identified prenatally may involute in utero. Prenatal magnetic resonance imaging (MRI) is helpful in the diagnosis of CAM, distinguishing it from a congenital diaphragmatic hernia14,15. Radiographically in the first hours of life, the chest radio-

graph may show a shift of the mediastinum by a soft tissue mass, which represents retained lung fluid within the CAM. The fluid is absorbed and this area is replaced by an air-filled cystic lesion. If the child has severe pulmonary compromise and the contralateral lung is not hypoplastic, early surgical removal may be required. Bronchopulmonary sequestration (BPS) represents a mass of nonfunctioning lung tissue that does not communicate with the normal bronchial tree and receives its vascular supply from the systemic circulation. Lesions may be intralobar or extralobar. The in utero US appearance is of a solid, welldefined highly echo-reflective mass. The anomalous systemic arterial supply is often difficult to visualize despite colour flow Doppler due to its small size and position. These lesions often appear to involute in utero. After birth these babies should be followed into childhood with chest radiographs, as the outcome of lesions seen antenatally but not detectable at birth is uncertain16. These lesions may not be detected antenatally or neonatally and the child presents at an older age with recurrent focal infections, bronchiectasis, haemoptysis, or an asymptomatic pulmonary mass.

Congenital lobar emphysema Congenital lobar emphysema may be a cause of respiratory distress in the neonate. It is characterized by marked overaeration of a single pulmonary lobe, usually an upper lobe, less commonly the right middle lobe. The cause of the abnormality is unknown. Different aetiologies are suggested, including congenital absence of bronchial cartilage leading to bronchomalacia, compression of the bronchus by a vascular sling, reduplication cyst, or inflammation. A primary alveolar abnormality has also been suggested where there is an increase in the size and number of the alveoli in the affected lobe. Immediately after birth, radiologically, the affected lobe may be opaque but will gradually become hyperlucent due to reduced pulmonary vascularity (Fig. 64.14).There will be gross overinflation with compression of the remaining lobes of the lung and contralateral mediastinal shift. Severe respiratory distress may be present and requires urgent lobectomy, but in some cases the respiratory distress is mild, surgery can be delayed, and spontaneous resolution may occur. In cases where the diagnosis is difficult, computed tomography (CT) may be helpful. There is a well-established association between congenital heart disease and infantile lobar emphysema, particularly ventricular septal defects and tetralogy of Fallot3,4.

Persistent fetal circulation

Figure 64.13 Congenital diaphragmatic hernia showing bowel extending from the abdomen in the left hemithorax and shift of the mediastinum to the right side.

Persistent fetal circulation with right-to-left shunt through the patent ductus arteriosus (PDA) is often a life-threatening condition usually seen in term or post-term asphyxiated babies who have depressed respiration. It is also seen in babies with meconium aspiration syndrome, polycythaemia, pulmonary hypoplasia, and myocardial ischaemia. Chest radiographs in idiopathic cases are often normal or sometimes show mild pulmonary oligaemia, retained lung fluid, and cardiomegaly. Echocardiography is essential to demonstrate right-to-left shunting and any cardiac anomalies.

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

Figure 64.14 Congenital lobar emphysema. (A) Age 4 days: opacification in the right upper lobe. (B) Age 3 months: increased lucency and overinflation of the right upper lobe.

Cardiac causes of respiratory distress The most common cardiac cause of respiratory distress is a PDA. Approximately 80% of infants will have a PDA during the first 4 d of life. The PDA only becomes pathological when there is significant right-to-left or left-to-right shunting. The diagnosis is made by echocardiography when several techniques can be used. A PDA can be seen directly with 2D or colour Doppler US and/or the left atrial-to-aortic root diameter can be measured using M-mode US, a non-specific indicator of left heart overload. A ratio greater than 1.4 on day 3 in an otherwise normal heart is indicative of a haemodynamically significant PDA17,18. In very premature infants, pulmonary oedema due to a PDA may form a major part of their pulmonary disease and often occurs early. Most congenital cardiac abnormalities are suspected from the clinical presentation of the neonate and are usually diagnosed on the echocardiogram (see Ch. 23). Several cardiac

conditions such as infradiaphragmatic total anomalous pulmonary venous drainage (TAPVD) causing respiratory distress are difficult to diagnose and may not be detected on the initial echocardiogram.

Late sequelae of neonatal lung disease Radiographic changes in survivors of uncomplicated RDS are uncommon and usually consist of mild linear shadowing representing fibrosis or deep pleural fissuring. Radiographic changes occur in approximately 75% of survivors of neonatal BPD. Most commonly, linear shadowing representing pulmonary fibrosis and pleural fissuring is seen; less commonly areas of irregular aeration and distortion of the bronchovascular architecture. The plain radiograph abnormalities decrease with age. In adolescence, there may be increased bronchial lability and an increase in the AP diameter of the chest is common.

THE OLDER PAEDIATRIC PATIENT Alan Sprigg Respiratory tract disease is common in clinical paediatrics; frontal chest radiographs are widely performed; a lateral view is usually only performed after radiology review. Lower respiratory tract infection (LRTI) is widespread and any inflammatory change in the airways of children will produce symptoms. Whether of viral, bacterial or allergic origin, any narrowing of the small airway in children causes increased resistance to airflow, leading to noisy breathing, wheeze, tachypnoea, and ‘respiratory distress’. Tenacious mucus in

the airways causes lobar collapse at an earlier stage than in adults. Conducted sounds from the upper airways and crying in small children make it difficult for physicians to exclude LRTI on examination and the chest is a common source of ‘sepsis’. Patterns of consolidation or collapse are similar to those in adults and only lesions showing different appearances in paediatrics compared to adult medicine will be considered further here.

1469

1470

SECTION 8

• PAEDIATRIC IMAGING

TECHNICAL FACTORS Poor inspiration may cause significant misinterpretation of a chest radiograph. At least five anterior rib ends should be visualized above the diaphragm before a chest radiograph can be considered adequate (Fig. 64.15). Rotation of the patient causes similar problems in interpretation of mediastinal shift and air trapping as in adult practice. A postero-anterior (PA) erect radiograph taken at full inspiration is optimal, but AP supine views are often obtained in infants. Performing chest radiography on a crying, uncooperative, tachypnoeic child is challenging.

THE THYMUS The mediastinum of infants appears abnormally wide or misshapen, due to the variable appearance of the thymus (Fig. 64.16). Enlargement of the thymus on follow-up radiography may occur after an acute illness—normal ‘thymic rebound’, and may simulate mediastinal mass. The normal thymus is said not to displace the trachea posteriorly unlike a true anterior mediastinal mass.The thymus is of lower attenuation than other mediastinal structures, allowing vessels to be seen through it. US is useful to confirm that mediastinal widening is due to a normal thymus19 before resorting to CT or MRI20.The normal thymus has a uniform reflectivity on US. It may extend to the right or left chest wall or anteriorly to the right or left cardiophrenic angles, simulating consolidation or a mass lesion.

TACHYPNOEA—RESPIRATORY OR CARDIAC CAUSE? Differentiating cardiac from respiratory causes of ‘respiratory distress’ in an infant can be as difficult clinically as it is radiographically. In small children below the age of 1 year, the cardiothoracic ratio may be well over 50% (see above), whereas in older children it should be 50% or less, even though this is a very imprecise rule—as in adults. Significant cardiac disease may exist without a murmur and the heart size may be normal. Cardiac US is performed to exclude symptomatic cardiac disease. Dextrocardia or situs inversus may be seen on a chest radiograph. Cardiac failure in children rarely causes pleural effusions. Mild interstitial oedema leads to peribronchial cuffing, overexpanded lungs with fluid in the horizontal fissure. Abnormal pulmonary vascularity (plethora or oligaemia) may be identifiable, but normal pulmonary vascularity does not exclude valvular stenosis or significant shunts.

INFECTION Suspected infection is the most common reason for a paediatrician requesting a chest radiograph but radiography is seldom helpful in predicting specific bacteriology21. Viral infection is far more common than bacterial pneumonia in children and produces nonspecific changes. The most common finding is of overexpansion due to airway inflammatory

Figure 64.15 Chest radiograph at different phases of respiration. (A) Four anterior rib ends above the diaphragm (expiration). (B) Six anterior rib ends (inspiration). These two radiographs on the same child at the same examination show how poor inspiration may simulate consolidation in the bases and apparent cardiomegaly.

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

B Figure 64.16 Normal thymus. (A) Chest radiograph with the thymus (sail sign) on the right mimicking right upper lobe consolidation. (B) Transverse US of the chest shows a normal thymus (arrow) around the great vessels. The normal thymus has a homogeneous reflectivity.

narrowing response, giving rise to wheeze and tachypnoea. Typical changes include peribronchial thickening, increased linear markings (interstitial infiltrates) radiating outwards from the hila, giving rise to ‘perihilar flaring’, and overexpansion of the lungs with flattening of the diaphragm. Routine follow-up radiography is not necessary if the child has recovered completely clinically22. Radiographic changes may persist for weeks, despite clinical resolution. A follow-up chest radiograph is useful following a complex clinical course, if there are residual clinical symptoms or after extensive lobar collapse/consolidation.The repeat chest radiograph should be delayed for several weeks.

Bronchiolitis23 This is the characteristic ‘viral’ infection of infancy, due to respiratory syncytial virus (RSV) infection, with increased incidence in spring and winter. Small children have limited energy reserves and may present with respiratory failure or respiratory arrest. Secondary bacterial infection may occur. Chest radiographs help detect the complications of bronchiolitis, rather than making the primary diagnosis. Pneumothorax occurs infrequently unless the infant is artificially ventilated. Bronchiolitis may cause a child with pre-existing pulmonary disease (e.g. BPD complicating prematurity) to go into respiratory failure. It is useful to have a ‘baseline’ chest radiograph on all children with chronic lung disease, before they are discharged from the neonatal nursery.

in differential diagnosis but they usually have sharp margins. Round pneumonia resolves rapidly with treatment.

Specific infections Pertussis25 The chest radiograph is usually taken to exclude complications rather than to confirm a clinical diagnosis. A poorly defined cardiac outline due to sublobar consolidation is often seen with Bordetella pertussis infection but it is not specific. Air trapping is seen in the acute phase.

Mycoplasma26 This is an atypical pneumonia and is a cause of segmental consolidation or persistent atelectasis after infancy. Small pleural effusions and a diffuse reticulonodular pattern may be seen, but these appearances are not specific for Mycoplasma pneumoniae.

Round pneumonia24 Children with ‘round pneumonia’ present with acute pyrexia and pleuritic chest pain.The chest radiograph shows a rounded or spherical opacity with poorly defined margins, simulating a pulmonary mass (Fig. 64.17); CT may be needed for clarification. An air bronchogram is uncommon, but the clinical history and the radiographic findings are characteristic. A solitary metastasis or posterior mediastinal tumour may be considered

Figure 64.17 Round pneumonia. There is a rounded opacity at the right base in a pyrexial child. The shadow has ill-defined borders unlike a primary tumour or metastasis.

1471

1472

SECTION 8

• PAEDIATRIC IMAGING

Staphylococcus27 Staphylococcus aureus infection occurs mainly in infancy or in immunocompromised children. In the acute phase it can cause a necrotic cavitating pneumonia. Following clinical resolution thin-walled ‘ghost cavities’ may persist for months (Fig. 64.18). They usually resolve without interference. Pyopneumothorax may occur.

Tuberculosis28 Infection with Mycobacterium tuberculosis in children may show a different pattern from that in adults. It should still be considered in the differential diagnosis of atypical pneumonia, consolidation that is slow to resolve, or persisting hilar adenopathy. The classic pathological description is of primary and secondary tuberculosis. Primary and miliary forms are more commonly seen in young children. In paediatric primary tuberculosis the lung focus may not be evident until it heals. It may present as consolidation with hilar nodal enlargement, persisting lobar collapse/consolidation due to endobronchial disease, or isolated nodal enlargement. Secondary tuberculosis is similar to adults. Miliary tuberculosis can occur at any age. Chest radiography shows a miliary nodular pattern (Fig. 64.19), sometimes with coexisting nodal enlargement or consolidation (Fig. 64.20). The chest radiograph in tuberculous meningitis (TBM) is often normal. Tuberculosis has a higher incidence in children who are malnourished or live in poor circumstances, those from areas where there is a high natural incidence of tuberculosis, and those who are immunocompromised.

Figure 64.19 the lungs.

Miliary tuberculosis. Fine nodular pattern throughout

Hilar adenopathy Hilar enlargement is most commonly due to uncomplicated pulmonary infection rather than tumour or tuberculosis. Persistent hilar enlargement should be further investigated, especially if there are clinical risk factors for tuberculosis. If there is coexistent paratracheal adenopathy this must be

Figure 64.20 Tuberculosis in a 10 year old. A chest radiograph shows right paratracheal adenopathy, right middle lobe and lower lobe consolidation with a coarse miliary pattern throughout the lungs.

investigated further. Sarcoidosis29 is an uncommon cause of hilar adenopathy in children. Paediatric sarcoidosis usually presents with systemic features rather than primary pulmonary disease.

Middle lobe syndrome The right middle lobe frequently collapses due to mucus plugging, especially in asthmatic patients. It is also a common site for foreign bodies.Tuberculosis is now a less common cause of middle lobe disease.

Aspiration pneumonia30–32

Figure 64.18 Staphylococcal infection. Following staphylococcal pneumonia, a thin-walled ‘ghost cavity’ may persist for several months.

Gastro-oesophageal reflux (GOR) is a common problem in paediatrics, especially in children with neuromuscular disorder or cerebral palsy. Nasogastric feeding tubes predispose to GOR. GOR can cause recurrent pneumonia or apnoea. The

CHAPTER 64

affected lobe depends on the child’s posture, the upper lobes being more commonly affected in infants who spend more time supine than in the older child who sits upright.

H-shaped tracheo-oesophageal fistula This, the rarest form of tracheo-oesophageal fistula (TOF), should be considered as a possible cause of recurrent pneumonia and can be demonstrated on a tube oesophogram. However, even good quality oesophogram tube studies may not demonstrate a narrow H-shaped fistula.

Pleural effusion In children, pleural effusions can be due to many causes but are most often due to infection (Table 64.1)—often pneumococTable 64.1

CAUSES OF PLEURAL EFFUSION

Small Bacterial chest infection Empyema

• THE NEONATAL AND PAEDIATRIC CHEST

cal or less frequently staphylococcal, anaerobic, streptococcal, haemophilus, mycoplasma or tuberculous infection. Underlying collapse of a lung may result in elevation of the diaphragm, liver or spleen, making clinical assessment of the extent of the effusion difficult. An effusion is easily confirmed using US, and by defining extent, stranding and depth US guidance can aid the placement of a diagnostic tap or drain. It can influence a decision to perform closed drainage, thoracoscopy, or decortication if there is an empyema (Fig. 64.21).

THE RADIO-OPAQUE CHEST In children infection is the most common cause of the unilaterally opaque chest radiograph. Other causes include bronchial obstruction due to intrinsic tumour (e.g. carcinoid) or extrinsic tumour (e.g. lymphoma/leukaemia nodes), pleural tumour (Askin tumour or other sarcoma), and delayed presentation of a congenital diaphragmatic hernia. A combination of US, CT, or MRI and diagnostic aspiration is used.

Tuberculosis Subdiaphragmatic cause (sepsis or a mass) Postoperative (sympathetic or chylothorax) Thoracic tumour Cardiac failure or fluid overload Hypoproteinaemia (nutritional or nephrotic/gut)

Large (radio-opaque chest): US for assessment As above but especially: Empyema Tumour—lymphoma, leukaemia Infective—consider tuberculosis Diaphragmatic hernia (obstructed)

IMAGING THE CHILD ON THE INTENSIVE CARE UNIT Intensive care chest imaging provides similar problems to adult practice, but respiratory arrest is much more common in paediatrics compared to cardiac arrest. Respiratory failure due to infection or septicaemia is a frequent reason for admission to the intensive care unit (ICU). The small size of the paediatric airway, tracheomalacia, or tenacious secretions cause lobar collapse, often of the right upper lobe. Non-cuffed ETTs are used in children to minimize pressure damage to the trachea but this allows ETT movement

Figure 64.21 Ultrasound in pleural effusions. (A) Chest radiograph of left-sided pleural effusion, elevated left diaphragm and scoliosis to the left. (B) Longitudinal US shows mixed reflectivity with thick internal septa in the fluid on the right of the image, due to a loculated empyema. Thick pus was drained at thoracotomy.

1473

1474

SECTION 8

• PAEDIATRIC IMAGING

with changes in the child’s position—a common cause of a tube migrating from the trachea into a bronchus. Rapid reexpansion usually occurs following correction of ETT malposition (Fig. 64.22), improving gas exchange. Misplaced central venous lines are a significant risk in paediatrics. Peripheral small bore catheters inserted for parenteral nutrition pose a significant risk for cardiac tamponade if their tip is left in the right atrium. Rib fractures are rarely caused by cardiopulmonary resuscitation in children, because their ribs are very compliant33,34. The radiologist may be the first person to identify rib fractures on a chest radiograph. In infants this raises the possibility of nonaccidental injury. If there is no history of significant trauma to explain rib fractures, the findings should be discussed directly with the paediatrician who can ensure the child is protected from further injury.

THE WHEEZY CHEST Asthma The diagnosis of asthma is clinical, not radiographic. Asthma is common but other causes of wheezing are rare (e.g. mediastinal mass, inhaled foreign body, or congenital airway abnormality). Chest radiography may be requested if wheezing is persistent. It should only be performed to exclude complications in a known asthmatic patient, e.g. consolidation with wheeze and pyrexia, or when a child is not responding to bronchodilators. The radiographic changes of asthma are those of overexpansion (flat diaphragm, square chest shape) with bronchial wall thickening or peribronchial cuffing. Right middle lobe collapse is often due to mucus plugging. Air leaks are uncommon.

Mediastinal shift As in adults, mediastinal shift may be due to reduction in volume on one side (e.g. lobar collapse) or increased volume on the other side (e.g. effusion or air trapping). Chest radiography is used to define the side of abnormality and the differential diagnosis (Table 64.2). Uncomplicated asthma does not cause mediastinal shift. Most small children will not cooperate to produce paired inspiration and expiration radiographs to look for mediastinal shift. Fluoroscopy allows assessment of diaphragmatic movement and mediastinal shift with varying phases of respiration, as most small children cry when placed on a screening table! Functional imaging of the abnormal lung is by radionuclide radiology; perfusion (Q) and ventilation (V) scintigrams will assess V/Q mismatching. Nebulized 99mTc-DTPA may precipitate in the bronchi in children, making assessment of regional ventilation difficult. 81mKrypton aerosol ventilation scintigraphy allows a continuous ‘wash-in’ phase assessment without precipitation in the airways, but access to 81mkrypton ventilation scintigraphy is limited by difficulty of supply.

Inhaled foreign bodies35,36 Small children frequently put objects in their mouths.A history of possible inhalation of a foreign body (FB) with the sudden onset of wheezing is an indication for further investigation. There may be no direct history of inhalation of a FB in a child presenting with wheeze or recurrent infection. Less obvious symptoms include persistent cough or recurrent pneumonia. Persisting collapse or consolidation in one lobe should alert the radiologist to the possibility of an endobronchial FB. Bronchoscopy should be performed if there is a good history. Whilst a FB may lodge in any bronchus, the right bronchus intermedius is a common site. The chest radiograph may identify a radio-opaque FB in the airway, air trapping, or lobar collapse, so it is important to review the whole airway with a suspicious history (Fig. 64.23). Unfortunately most inhaled FBs are radiolucent (e.g. plastic toys and peanuts). Comparison of the aeration and perfusion of the same areas of lung on both sides will show areas of air trapping or reflex oligaemia (Fig. 64.24). Fluoroscopy can be Table 64.2

MEDIASTINAL SHIFT

Towards the abnormal side Lobar collapse (mucus plug, foreign body) Pulmonary agenesis or hypoplasia (primary or secondary) Scimitar syndrome (hypogenetic lung) Bronchopulmonary sequestration (basal) Swyer James (MacLeod) syndrome

Away from the abnormal side Pneumothorax Pleural effusion (without collapse)

Figure 64.22 Displaced endotracheal tube (ETT) down the right main bronchus. The tip of the tube (arrow) is distal to the left main bronchus, resulting in complete collapse of the left lung and mediastinal shift to the left. Post right thoracotomy. The left lung completely reinflated 1 h later following repositioning of the ETT.

Air trapping: foreign body, congenital lobar emyphsema (apical), bronchial atresia Cystic congenital adenomatoid malformation Diaphragmatic hernia

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

Figure 64.23 Intubated child with stridor. Endotracheal tube (ETT) and nasogastric tube in place. There is an opaque foreign body (arrow) faintly seen in the right main bronchus, right upper lobe collapse, and right middle lobe consolidation. Arrowhead = nasogastric tube; curved arrow = ETT.

Figure 64.24 Foreign body in the left bronchus (not radio-opaque). Relative oligaemia and small blood vessels at the left lung are due to poor ventilation and reflex vasoconstriction on the left side.

used to demonstrate air trapping (ball valve effect) with mediastinal shift in an uncooperative child (Fig. 64.25). Peanuts are common inhaled FBs that contain oil, causing endobronchial inflammation or stenosis. Peanuts often fragment during bronchoscopic removal, embolizing into smaller bronchi and causing increased consolidation on the post-bronchoscopy radiograph. Petroleum product inhalation is usually due to inappropriate storage in ‘pop’ bottles. An early chest radiograph may be falsely reassuring37. Radiography may show pneumonitis hours later. Pneumatoceles are a late complication.

trachea with respiration and posture. Stridor is often due to viral or bacterial infection. Congenital stridor may be due to tracheal compression by a vascular ring (arch anomaly or pulmonary artery sling). Initial assessment is by contrast medium studies of the oesophagus with AP and lateral views.

Stridor Tracheal ‘buckling’ seen on a frontal chest radiograph is a normal variant due to the flexibility and movement of the infant’s

CONGENITAL CHEST ABNORMALITIES Congenital lobar emphysema This most commonly presents in infancy beyond the neonatal period and involves an upper lobe.The chest radiograph shows mediastinal shift away from the side of the abnormality and reduced vascularity in the affected lobe (Fig. 64.26). CT and V/Q scintigraphy are useful in indeterminate cases, showing

Figure 64.25 Obstructive emphysema due to foreign body. (A) The chest radiograph taken on full inspiration is normal. (B) The expiration film shows a marked mediastinal shift to the left. There was a foreign body in the right main bronchus, causing air trapping due to a ‘ball valve’ effect.

1475

1476

SECTION 8

• PAEDIATRIC IMAGING

Figure 64.26 Congenital lobar emphysema. Eleven-month-old with respiratory distress. (A) Chest radiograph shows mediastinal shift to the left and oligaemia in the right upper lobe. (B) CT shows overexpansion, small pulmonary vessels and oligaemia on the right side. (C) Scintigraphy with 99mTcmacroaggregates shows poor perfusion in the right upper lobe; right posterior oblique view.

extent and reduced perfusion, with absent ventilation on the early phase, delayed entry of isotope activity in the affected lobe and retention in the late phase.

Congenital diaphragmatic anomalies38,39 The diaphragm originates from muscle slips from the ribs fusing into a central tendon and creating separate abdominal and chest cavities. If the central tendon of the diaphragm is abnormally lax, it causes eventration. If there is a diaphragmatic defect a congenital diaphragmatic hernia occurs often presenting in the neonatal period but delayed presentation does occur. Diaphragmatic paralysis may be due to damage to the phrenic nerve during a difficult delivery. Acquired paralysis is uncommon without prior surgical intervention. An abnormal diaphragm may be identified on a chest radiograph taken for other reasons. It may be elevated or have associated mediastinal shift. If there is unilateral diaphragmatic elevation, it is important to exclude a subdiaphragmatic cause (e.g. abdominal mass, subpulmonary effusion, or subphrenic sepsis). US is very useful, as this may show a mass, collection, or impaired diaphragmatic excursion, and can assess paradoxical or asymmetric movement with respiration, which is best assessed on a transverse (slightly coronal) image as this can allow simultaneous visualization of both diaphragms. Fluoroscopy is used to assess both hemidiaphragms as they move with respiration. The lateral position can assess relative movement. Older children can cooperate for the ‘sniff test’ and crying infants produce a rapid inspiratory excursion of the diaphragms at the end of a scream. Unilateral diaphragmatic paralysis involves the whole hemidiaphragm but in eventration there is usually a localized hump in the diaphragm. The differentiation between diaphragmatic hernia and eventration can be difficult especially on the right

side where the liver may act as a central obturator in the hernia, causing a localized hump in the diaphragm. A hernia should be considered in the differential diagnosis of a basal cystic mass in the chest (Table 64.3) and excluded by the use of gastrointestinal contrast studies. Anterior (midline Morgagni) hernias may present in childhood or in adult life. They can be difficult to diagnose on the chest radiograph if the omentum herniates rather than the gas-filled transverse colon. Contrast gastrointestinal studies are useful if there is apparent gut herniation (Fig. 64.27). CT or coronal MRI is useful to confirm herniation of fatty omentum in the differentiation of a paracardiac mass.

Table 64.3

CAUSES OF PAEDIATRIC LUNG CYSTS

Cystic fibrosis Cystic bronchiectasis Bronchopulmonary dysplasia (neonate and older) Tuberculosis (apical thick walled) Mycetoma (apical cyst with contents) Pulmonary abscess (thick wall, fluid level) Empyema (pleural) Streptococcal pneumatocele (thin wall, post infective) Cavitating pneumonia Bronchopulmonary sequestration (basal) Cystic congenital adenomatoid malformation (basal cysts of varying size) Diaphragmatic hernia (basal cysts of similar size) Hiatal hernia (basal posterior) Morgagni hernia (midline anterior) Hydatid disease (in endemic areas) Kerosene inhalation (pneumatocele) Histiocytosis and interstitial disease

CHAPTER 64

Posterior Bochdalek hernias are often diagnosed antenatally or present in the neonate with a large diaphragmatic defect. In the older child the diaphragmatic defect is usually much smaller and the child may present with a loop of gut herniating into the chest through a small defect with subsequent obstruction or gut ischaemia. The clinical presentation is variable with collapse, septicaemia, a cystic chest mass, or a ‘white-out’ chest. A herniated stomach may simulate an isolated cavity in the left lung with a fluid level. Passage of a nasogastric tube can confirm that the stomach is in the chest (Fig. 64.28). If oral contrast studies are performed, they should be followed through until the entire colon has been demonstrated to the rectum. Confirmation that the stomach and small bowel are normally sited does not exclude a colonic hernia. Hiatal hernias are seen in children as well as in adults. Para-oesophageal hernias may present as a cystic mass in the right or left side of the chest.

Congenital cystic adenomatoid malformation40 These commonly present in the neonate, often diagnosed on antenatal US. They may be found in older children on a chest

• THE NEONATAL AND PAEDIATRIC CHEST

radiograph taken for other reasons or who present with symptoms due to obstructive emphysema and mediastinal shift or secondary to infection. A spectrum of cystic change is found but three main forms are described: macrocystic (type 1) (Fig. 64.29), microcystic (type 3), or a mixed pattern (type 2). Congential cystic adenomatoid malformation (CCAM) is included in the differential diagnosis of cystic lesions of the paediatric chest. CCAM shares some histological features with bronchopulmonary sequestration (BPS), which is also a basal disorder.

Bronchopulmonary sequestration41,42 BPS is often diagnosed on obstetric US and overlaps with CCAM. It is due to a developmental abnormality of lobar formation with abnormal vascular supply. It usually presents with a persisting basal opacity on serial radiography, mimicking lower lobe collapse/consolidation or cavitation, and is frequently left sided. It is important to exclude sequestration with a persisting basal lung lesion. Its arterial supply is from the systemic circulation from the aorta above or below the diaphragm. Pre-operative colour Doppler abdominal US may demonstrate the abnormal vessel(s) if they arise from the abdominal aorta, but may miss the arterial supply from the thoracic aorta. CT angiography (CTA) or MR angiography (MRA) can define the arterial feeder(s) and may show anomalous draining veins.

Scimitar syndrome (hypogenetic lung, pulmonary venolobar syndrome) This congenital anomaly shares some features with sequestration except that the lung is normally connected to the bronchial tree, but the vein draining the lobe (usually the right lower lobe) drains into the inferior vena cava or portal vein, rather than to the left atrium. The aberrant basal pulmonary vein may be found coincidentally on a chest radiograph or the child may present with features of a leftto-right shunt. The characteristic appearance on chest radiography is of a small ipsilateral lung with mediastinal shift towards the affected side. An abnormal vessel is usually seen draining down and enlarging towards the diaphragm in the shape of a ‘scimitar’ sword.

Pulmonary vascular malformations43

Figure 64.27 Morgagni diaphragmatic hernia. (A) Chest radiograph shows a cystic structure projected over the heart. (B) A barium follow-through shows the colon in the anterior hernia on the delayed lateral view.

These are similar in appearance to those in adults. Pulmonary or aortic angiography is diagnostic. Radionuclide imaging is used to screen for a right-to-left shunt. Normally, 99mTcmicrospheres injected intravenously do not pass through the lung capillaries. With a right-to-left shunt they can pass through into the systemic circulation and lodge in the kidneys and other organs (Fig. 64.30). Pulmonary systemic shunting may also occur with BPS, scimitar syndrome, bidirectional cardiac shunts, or other arteriovenous malformations.

Pulmonary agenesis and hypoplasia An early insult to lung bud development or vascular supply will interfere with lung development in later pregnancy. Total agenesis of the lung is associated with absence of the pulmonary

1477

1478

SECTION 8

• PAEDIATRIC IMAGING

A

Figure 64.28 Left diaphragmatic hernia. A 2 year old child with acute dyspnoea due to gastric herniation and volvulus. Absent gastric bubble in the abdomen. A nasogastric tube was later passed into the ‘pulmonary cyst’, confirming gastric herniation.

Figure 64.29 Congenital cystic adenomatoid malformation of lung (CCAM). A 4 year old child presented with acute dyspnoea. Chest radiograph shows a cystic lesion in the left lower lobe, confirmed histologically as a CCAM type 1.

Figure 64.30 Pulmonary arteriovenous malformation (AVM). (A) Chest radiograph shows a round opacity (arrow) in the left base posterior to the diaphragm, with large lower lobe vessels. (B) 99mTc-macroaggregate scintigraphy shows normal activity in the lungs and abnormal activity in the kidneys due to shunting from right to left across the AVM. (Courtesy of Professor H.M. Carty, Liverpool, UK.)

artery and bronchial development. Chest radiography demonstrates mediastinal shift and the absence of lung markings on the affected side. Cross-herniation of the normal lung from the opposite side may cause confusion. CT and radionuclide imaging correlated with a recent chest radiograph are helpful. Scintigraphy shows no perfusion or ventilation in the affected lung.Angiography (or echo) confirms a small or absent pulmonary artery on the affected side (Fig. 64.31). Pulmonary hypoplasia is a less severe form of the above and may be unilateral or bilateral. Causes include neuromuscular disease, skeletal dysplasia (e.g. asphyxiating thoracic dystrophy) or pulmonary compression due to oligohydramnios. A chest radiograph shows a ‘bell-shaped’ chest and slender ribs. Unilateral hypoplasia is due to a primary pulmonary embryological defect or is secondary to diaphragmatic hernia or CCAM resection. The acquired form overlaps with the Swyer James (MacLeod) syndrome, which is due to infection of the developing lung in infancy (Fig. 64.32), but unlike Swyer James syndrome, pulmonary hypoplasia shows no evidence of air trapping.

Figure 64.31 Pulmonary agenesis. (A) Chest radiograph. The heart is displaced to the left and the right lung is overexpanded into the left chest, obscuring the heart. (B) A right heart angiogram shows the large right pulmonary artery but no left pulmonary artery, confirming left pulmonary agenesis.

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

Figure 64.32 Hypoplasia of the right lung. (A) The chest radiograph shows mediastinal shift to the right and a small undervascularized right lung. (B) Bronchography shows normal right bronchial branching with attenuation. Normal left bronchi.

THE MEDIASTINUM44–48 The mediastinum may be divided into compartments, as in adult chest disease, but the causes of mediastinal masses are different (Table 64.4). Primary malignancy and acquired vascular disease are rare. Congenital malformations are common. In the radiographic investigation good quality frontal and lateral chest radiographs are essential. US, CT/CTA or MRI/ MRA will give further information on mediastinal masses in the assessment of vascular rings or in congenital malformation (Table 64.5). Many tumours spread across compartment boundaries—an ‘anatomical’ approach to mediastinal pathology is only a guide. The imaging findings must be correlated with clinical data (e.g. blood count to exclude leukaemia and clinical examination to search for other sites of nodal enlargement) before resorting to mediastinal biopsy. Imaging guides the approach of the biopsy by defining the extent of disease and its relation to major vascular and spinal structures. Involvement of pericardium, major vessels or spine may necessitate cardiac or neurosurgical assistance. Anterior mediastinal masses are most commonly nodal (lymphoma or leukaemia) (Fig. 64.33). There may be significant compression of the trachea with an anterior mass (Fig. 64.34)—important information for the anaesthetist. A multilocular cystic mass is usually a hygroma or lymphangioma, but this may have components of haemangioma. Calcification in a middle or anterior mediastinal mass suggests a teratoma or hamartoma, whereas a cystic mass in the middle mediastinum is most commonly a bronchogenic cyst (Fig. 64.35). Bronchogenic cysts usually have a slender connection to the carina and may occur away from the carina, but they only occasionally show air within the cyst. Foregut remnants (bronchogenic cysts, oesophageal duplication, neurenteric cyst) are due to failure of normal embryology as the primitive notocord, oesophagus, and trachea develop and separate.

Table 64.4 CAUSES OF MEDIASTINAL MASS, BY COMPARTMENT Anterior (superior) Normal thymus** Thymic infiltration (leukaemia, lymphoma, histiocytosis)** Nodal mass** Thyroid Cystic lymphangioma/haemangioma Thymic cyst or thymoma Teratoma Plexiform neurofibroma

Anterior (inferior) Cardiac* Pericardial cysts, tumours, and fat pads Morgagni hernia (gut or omentum)

Middle Nodal mass (as above), granulomatous disease (TB)** Vascular and aortic ring anomalies* Bronchogenic cyst (foregut duplication, neurenteric cyst)* Venous anomalies (left superior vena cava) Plexiform neurofibroma

Posterior Sympathetic chain tumours (neuroblastoma ganglioneuroma sequence)** Hiatal and diaphragmatic hernia (intrathoracic kidney)** Spinal related tumours (plus neuroblastoma from abdomen)* Spinal sepsis (staphylococcal, tuberculosis)* Bronchopulmonary sequestration* Oesophageal duplication, neurenteric cyst Neurofibroma Phaeochromocytoma Extension of abdominal pathology (e.g. pancreatitis with pseudocyst) Extramedullary haematopoiesis (haemoglobinopathy) ** = Common; * = occasional; others are uncommon.

1479

1480

SECTION 8

• PAEDIATRIC IMAGING

Table 64.5 INVESTIGATION OF A MEDIASTINAL MASS 1. Plain radiography. Chest PA and lateral, localized spine and rib views if posterior 2. US if mass anterior, pericardial or basal (solid or cystic mass, nodal or vascular/cardiac) 3. CT. Post intravenous contrast medium. Gives details of location and the relationship to vessels and extension. CT essential for lung detail and metastases. 4. MRI. Pre- and post-gadolinium. MRI is used if there is a posterior mass, associated spinal anomaly or clinical evidence of cord involvement. Axial, sagittal and coronal sections are essential.

Additional imaging 5. Contrast studies: Oesophogram if the mass involves the middle or posterior mediastinum or clinical symptoms of dysphagia or stridor. Late image for ? diaphragmatic hernia 6. Radionuclide imaging 99m

Tc-Pertechnetate: duplication cyst of the oesophagus

123

I-MIBG: neuroblastoma-like tumour, whole body

Figure 64.34 Neurofibromatosis (plexiform). There is a large mediastinalmass extending into the neck and causing tracheal compression (arrow).

A middle or posterior mediastinal mass with an associated spinal anomaly may have an extra- or intra-dural extension. MRI is essential. Resection of the mediastinal component without identifying a spinal communication may result in postoperative meningitis or cord compromise. Calcification in a posterior mass suggests a sympathetic chain tumour, commonly a neuroblastoma or ganglioneuroma (Fig. 64.36). Extradural encroachment into the spinal canal often occurs via adjacent foramina at several levels, presenting as paraparesis. Pre-operative MRI is needed to assess extent. Later 123I-MIBG imaging may be needed for staging neuroblastoma. One in 10 of all neuroblastomas occur in the chest. Abdominal neuroblastoma may also extend into the chest, producing a paravertebral mass or effusion.

Vascular anomalies

Figure 64.33 Anterior mediastinal mass. Chest radiograph (A) PA and (B) lateral. Large anterior mediastinal mass displacing the trachea posteriorly (arrow) due to T-cell acute lymphoblastic leukaemia. Similar appearances may be seen with lymphoma.

B

Abnormal vessels seen on CT or US may indicate a primary vascular problem or represent an abnormal vascular supply to or from a congenital abnormality (e.g. scimitar syndrome, pulmonary AVM, or BPS). Abnormal embryologic development of the aortic arch leads to persistence of the primitive aortic arch or its branches. This may give an abnormal mediastinal contour but usually presents with stridor.

Tumours Primary bronchial or pulmonary tumours are rare in paediatric practice compared to adults. Bronchial carcinoid tumour is an infrequent primary tumour occurring in adolescents. Multiple pulmonary metastases are unusually associated with an obvious primary tumour elsewhere (Table 64.6). Chest wall tumours are more common than primary lung tumours. The Askin tumour is a sarcoma arising from the

CHAPTER 64

• THE NEONATAL AND PAEDIATRIC CHEST

Figure 64.35 Bronchogenic cyst. (A) Chest radiograph. Obstructive emphysema in the left chest is due to a central mediastinal mass (arrow). (B) Contrast-enhanced CT (under general anaesthetic) shows a central low attenuation mass (arrow) causing compression of the carina and bilateral lower lobe collapse. Figure 64.36 Neuroblastoma. (A) Chest radiography shows posterior mass on the right in a child with chest pain and no clinical features of infection. (B) CT chest (noncontrast) shows calcification in the mass adjacent to the spine. There was no evidence of extradural extension on MRI.

chest wall, often with rib involvement49 and a pleural effusion. Primary sarcoma arising in the rib often presents late, with a pleural lesion with rib involvement (Fig. 64.37). Skeletal scintigraphy is used to detect multiple sites of tumour or other metastases. Bone involvement of the spine or ribs is also common in mediastinal and chest wall tumours (Table 64.7). Initial staging of mediastinal tumours is performed with contrast-enhanced spiral CT. Percutaneous biopsy is possible for even small lesions50. Spiral CT should be performed before considering thoracotomy for resection of an apparently isolated metastasis or nodule seen on a chest radiograph. Basal atelectasis is a significant problem in the young child that needs anaesthesia for chest CT. Positive end-expiratory

Table 64.6 POTENTIAL SOURCES OF PULMONARY METASTASES Nephroblastoma (Wilms’ tumour) Primary bone sarcoma (Ewing or osteosarcoma) Rhabdomyosarcoma Testicular tumour (in the adolescent)

pressure will help prevent this51 and change in posture from supine to prone may differentiate atelectasis from disease. Peripheral pulmonary metastases are often seen in the posterior inferior recess and may be masked by atelectasis in the supine position.

1481

1482

SECTION 8

• PAEDIATRIC IMAGING

Figure 64.37 Ewing sarcoma. Oblique view of a lobulated, peripheral left chest mass and an abnormal third rib.

Table 64.7

BONE INVOLVEMENT IN THE THORAX

Spine Direct Primary spinal tumours (Ewing/osteosarcoma, osteoid osteoma, aneurysmal bone cyst) Neuroblastoma-like tumour—ribs and spine Sepsis Neurenteric cyst

Early radiographic features include air trapping and bronchial wall thickening, features that are radiographically indistinguishable from asthma or recurrent pneumonia due to other causes. Later a diffuse interstitial pattern, bronchiectasis, and cyst formation occur. The lung disease progresses at a variable rate to an ‘end-stage lung’ in early adult life. Late radiographic features show pulmonary hypertension with hilar enlargement and peripheral oligaemia with a background of chronic lung change (Fig. 64.38), and a barrel-shaped chest. Serial radiography at yearly intervals is used to monitor progression of the pulmonary disease. PA and lateral radiographs are usually performed to allow radiographic ‘scoring’ of progression. Various scoring systems have been devised for clinical and radiographic assessment (Crispin–Norman, Shwachmann)53. A scoring system based on thin section CT has also been used (Bhalla)54. CT will show significantly more change than is detectable on a chest radiograph. The low-dose spiral CT technique minimizes radiation dose55. CT is especially useful when symptoms are disproportionate to chest radiograph changes56. Some children have a ‘target lobe’ where the bronchiectasis and cyst formation is maximal, and where infection seems to recur. Cysts may fill with secretions and show transient filling and clearing on serial radiography, which does not always signify acute infection. Chronic infection on sputum analysis with Staph. aureus is common, and in the later phase with Pseudomonas aeruginosa. Lobar collapse also occurs

Metastatic Leukaemia Neuroblastoma Histiocytosis (LCH) Ewing/osteosarcoma Lymphoma

Ribs All of the above involving spine plus: Actinomyces infection (apical) Osteomyelitis Askin tumour (PNET)

CYSTIC FIBROSIS Cystic fibrosis (CF) is associated with chronic pulmonary sepsis and malabsorption (pancreatic exocrine insufficiency, cirrhosis, and gut involvement). It is an autosomal recessive disease with a carrier frequency of 1 in 25. CF may present as neonatal meconium ileus or in the older child with malabsorption or respiratory problems. The pulmonary features include chronic sepsis, mucus plugging due to abnormally thick sputum, and abnormal cilial motion that results in bronchiectasis, cavitation, and pulmonary fibrosis, and ultimately pulmonary hypertension plus episodic acute infective exacerbations. Other radiographic features include chronic sinusitis, nasal polyposis, nail clubbing, and symptoms of HPOA (hypertrophic pulmonary osteoarthropathy)52.

Figure 64.38 Cystic fibrosis. Chest radiograph shows thick-walled bronchi and cysts in all the lobes with increased reticular pattern. Both lungs are overexpanded with large hila indicating pulmonary hypertension.

CHAPTER 64

secondary to mucus impaction, a finding that alters physiotherapy treatment. Children with CF also have an increased risk of bronchopulmonary aspergillosis. Recurrent, transient, parenchymal opacities together with peripheral and sputum eosinophilia with specific immunology titres are diagnostic57. Haemoptysis may be due to bronchiectasis but occasional massive haemoptysis occurs due to arterial bleeding from an infected cavity or secondary to bronchial arterial hypertrophy due to cor pulmonale. Selective bronchial arteriography and embolization may be life saving. Pneumothorax is uncommon, but may be life threatening.

Bronchiectasis and Kartagener’s syndrome Bronchiectasis is usually idiopathic but also complicates CF, Kartagener’s syndrome, and cilial dysmotility. It is a recognized complication of adenovirus, measles, and pertussis infection. High-resolution spiral CT is more sensitive and much less invasive than bronchography. Kartagener’s syndrome has additional associations with dextrocardia and facial sinusitis. The underlying problem is cilial dysmotility producing inadequate clearance of sputum. This also relates to the failure of the heart to rotate to the usual left-sided position.

PAEDIATRIC INTERSTITIAL DISEASE58 There is a long differential diagnosis for interstitial change in the paediatric chest. A list of common associations is given in Table 64.8. The reader is referred to further texts for details of each syndrome. Clinical correlation is essential, including a drug history, full clinical examination (e.g. iritis, arthropathy, nodal enlargement, etc.) together with pulmonary function testing, biochemical screening (for collagen vascular disease), and then further imaging. If interstitial disease is suspected from a plain chest radiograph, clinical history, or physiological meaTable 64.8 CAUSES OF ‘INTERSTITIAL’ LUNG PATTERN Cystic fibrosis Mycoplasma pneumonia Interstitial oedema (cardiac or fluid overload) Histiocytosis Sarcoidosis Fibrosing alveolitis Idiopathic interstitial pneumonitis Neurofibromatosis

• THE NEONATAL AND PAEDIATRIC CHEST

surements, high-resolution thin-section CT helps determine the presence, extent, and distribution of disease. Atelectasis complicates sedation due to hypostatic change. For spiral CT in small children, sedation or general anaesthesia is often needed. Controlled respiration under general anaesthesia also ensures adequate inspiration, minimizes motion artefact, and also allows scanning in expiration to assess air trapping59. Positive end-expiratory pressure used in anaesthesia minimizes atelectasis. Anaesthetic cooperation is important for optimal paediatric imaging, including endotracheal intubation rather than laryngeal mask airway if suspended respiration or prone positioning is needed. If chest CT is being performed as one of a series of investigations under anaesthesia (e.g. bronchoscopy or MRI), then CT should be performed first, as atelectatic changes often progress with duration of anaesthesia and medical intervention. If the child will stay still but cannot coordinate inspiration/ expiration, then selected sections can be made in both decubitus positions—the dependent lung simulating an ‘expiration phase’. CT also assists in planning the best site for lung biopsy to obtain histology, and confirms associated air trapping, bronchiectasis, and cystic disease60. Lung biopsy should not be obtained from an area of dense fibrosis as the pathologist may only see ‘end-stage diffuse fibrosis’. Fibrosing alveolitis is one of the more frequent causes of paediatric interstitial lung disease. It has a better outlook in children than in adults and may resolve spontaneously or after a short course of treatment. Sarcoidosis is uncommon in children prior to adolescence. ‘Honeycomb lung’ is rare in paediatric practice, but is more commonly seen with histiocytosis (Fig. 64.39) and tuberous sclerosis.

IMMUNODEFICIENCY61–64 The lung is a common site of involvement in the immunocompromised child.There is easy access for inhaled pathogens and the lung is also a common site for haematogenous spread of infection from long tubes used for venous access. The paediatric lung also has a large amount of ‘bronchial associated lymphoid tissue’ (BALT) involved in the immune response. Over half the children dying from acquired immunodeficiency syndrome (AIDS) have pulmonary involvement. Immunocompromise in children usually arises from one of three causes: • congenital or combined immunodeficiency syndromes (SCIDs, etc) • induced immunodeficiency due to steroids or chemotherapy • acquired immunodeficiency syndrome (HIV and AIDS related).

Tuberous sclerosis Rheumatoid and collagen vascular disease Drug reactions Lymphangiectasia Leiomyomatosis Storage disorders (Gaucher, Neimann Pick) Pulmonary haemosiderosis (late, following multiple bleeds)

Pulmonary infection with atypical organisms may occur or the common pathogens may exhibit an abnormal response. Common bacterial (Streptococcus pneumoniae and Haemophilus influenzae) or viral infections (chicken pox or measles) may have a devastating effect on the lung in an immunocompromised child. If virology cannot give a rapid diagnosis, then broncho-alveolar lavage (BAL), transbronchial or imaging-guided

1483

1484

SECTION 8

• PAEDIATRIC IMAGING

in children is not as aggressive as in adults, as it progresses less commonly to lymphoma. The initial symptoms are mild with slowly progressive dyspnoea and hypoxia. Chest radiograph findings include hilar and paratracheal adenopathy, with reticulonodular infiltrates and patchy alveolar opacification. It progresses to cystic and cylindrical bronchiectasis, but can be treated with steroids. There is considerable overlap in the appearances of the various differing infections in the immunocompromised child. The child is usually treated empirically following blood and sputum culture on the basis of the most probable infection. If there is a poor clinical response, a more invasive biopsy may have to be performed to identify the causative organism or pathological process. Figure 64.39 Interstitial disease—histiocytosis. There are multiple lytic bone lesions (especially of the right scapula), mediastinal adenopathy, and interstitial infiltrates.

biopsy should be performed. Alternatively a limited thoracotomy and biopsy is performed to obtain an open lung biopsy for histology and culture. Pneumocystis juroveci (carinii) pneumonia (PCP) infection is as common as in adults.The child is usually pyrexial and hypoxaemic and may progress rapidly to respiratory failure. The chest radiograph shows progressive bilateral infiltrates, evolving to bilateral diffuse airspace (alveolar) opacification. With human immunodeficiency virus (HIV) and immunocompromise tuberculosis may be the first presentation, which may predate the development of ‘AIDS’ by several months. The chest radiograph may not show a miliary pattern even with disseminated disease. Infection with Mycobacterium avium intracellulare (MAI) also occurs, usually with hilar adenopathy, pulmonary infiltrates, and lobar collapse. Fungal infections are also common (Fig. 64.40). Lymphocytic interstitial pneumonitis (LIP) is associated with immune compromise as well as Sjögren’s syndrome, chronic active hepatitis, and myasthenia. The course

Figure 64.40 Mycetoma in an immunosuppressed patient. There are multiple cavitating lesions in the left apex and also right base with internal contents in the apical lesion.

REFERENCES 1. National Radiation Protection Board 1993 Occupational, public and medical exposure, vol 4, no 2. NRPB, Chilton 2. Cook J V, Shah K, Pablot S et al 1998 Guidelines on best practice in the x-ray imaging of children. AGFA, UK, pp 9–10 3. Newman B, Bowen A, Oh K S 1990 A practical approach to the newborn chest. Curr Probl Diagn Radiol 19: 41–84 4. Gibson A T, Steiner G M 1997 Imaging the neonatal chest. Clin Radiol 52: 172–186 5. Agrons G A, Harty M P 1998 Lung disease in premature infants: impact of new treatments and technologies. Semin Roentgenol 33: 101–116 6. Whitsett J A, Pryhuber G S, Rice W R et al 1994 Acute respiratory disorders. In: Avery G B, Fletcher M A, MacDonald M G (eds) Neonatology: pathophysiology and management of the newborn, 4th edn. JB Lippincott, Philadelphia, pp 429–452 7. Pilling D W, Pilling E L The neonatal chest. In: Carty H, Brunelle F, Stringer DA, Kao S C S (eds) 2005 Imaging children, 2nd edn. Churchill Livingstone, New York, pp 1023–1047 8. Dinger J, Schwarze R, Rupprecht E 1997 Radiological changes after therapeutic use of surfactant in infants with respiratory distress syndrome. Pediatr Radiol 27: 26–31 9. Slama M, Andre C, Huon C et al 1999 Radiological analysis of hyaline membrane disease after exogenous surfactant treatment. Pediatr Radiol 29: 56–60 10. Greenhough A, Dixon A K, Roberton N R C 1984 Pulmonary interstitial emphysema. Arch Dis Child 59: 1046–1051 11. Northway W H, Rosan R C, Porter D Y 1967 Pulmonary disease following respirator therapy of hyaline-membrane disease. N Engl J Med 276: 357–368 12. Wilson M G, Mikity V G 1960 A new form of respiratory disease in premature infants. Am J Dis Child 99: 489 13. Grossman H, Berdon W E, Mizaki A et al 1965 Neonatal focal hyperaeration of the lungs (Wilson Mikity syndrome). Radiology 85: 404–417 14. Hubbard A M, Crombleholme T M 1998 Anomalies and malformations affecting the fetal/neonatal chest. Semin Roentgenol 33: 117–125 15. Hubbard A M, Adzick N S, Cromblehome T M 1997 Value of prenatal MR imaging in preparation for fetal surgery of left congenital diaphragmatic hernia. Radiology 203: 636–640 16. Lacey D E, Shaw N J, Pilling D W et al 1999 Outcome of congenital lung abnormalities detected antenatally. Acta Paediatr 88: 454–458 17. Archer N 1993 Patent ductus arteriosus in the newborn. Arch Dis Child 69: 529–532 18. Evans N 1993 Diagnosis of patent ductus arteriosus in the preterm newborn. Arch Dis Child 68: 58–61 19. Carty H 1990 Ultrasound of the normal thymus in the infant: a simple method of resolving a clinical dilemma. Br J Radiol 63: 737–738 20. Siegel M J, Glazer H S, Wiener J I, Molina P L 1989 Normal and abnormal thymus in childhood: MR imaging. Radiology 172: 367–377

CHAPTER 64

21. Wahlgren H, Mortensson W, Eriksson M, Finkel Y, Forsgren M 1995 Radiographic patterns and viral studies in childhood pneumonia at various ages. Pediatr Radiol 25: 627–630 22. Gibson N A, Hollman A S, Paton J Y 1993 Value of radiological follow up of childhood pneumonia. BMJ 307: 1117 23. Simpson W, Hacking P M, Court S D M, Gardner P S 1974 The radiological findings in respiratory syncytial virus infection in children. Pediatr Radiol 2: 155–160 24. Rose R W, Ward B H 1973 Spherical pneumonias in children simulating pulmonary and mediastinal masses. Radiology 106: 179–182 25. Bellamy E A, Johnston I D A, Wilson A G 1987 The chest radiograph in whooping cough. Clin Radiol 38: 39–43 26. Guckel C, Benz-Bohm G, Widemann B 1989 Mycoplasmal pneumonias in childhood. Pediatr Radiol 19: 499–503 27. Macfarlane J, Rose D 1996 Radiographic features of staphylococcal pneumonia in adults and children. Thorax 51: 539–540 28. Cremin B J 1995 Tuberculosis: the resurgence of our most lethal infectious disease—a review. Pediatr Radiol 25: 620–626 29. Lakshmana D N, Hingsbergen E A, Jones J E 1999 Adult diseases in children. Pediatr Radiol 29: 244–254 30. Simpson H, Hampton F 1991 Gastro-oesophageal reflux and the lung. Arch Dis Child 66: 277–283 31. Couriel J M, Bisset R, Miller R, Thomas A, Clarke M 1993 Assessment of feeding problems in neurodevelopmental handicap: a team approach. Arch Dis Child 69: 609–613 32. McVeagh P, Howman-Giles R, Kemp A 1987 Pulmonary aspiration studied by radionuclide milk scanning and barium swallow. Roentgenography. Am J Roentgenol 141: 917–921 33. Spevak M R, Kleinman P K, Belanger P L, Primack C P, Richmond J M 1994 Cardiopulmonary resuscitation and rib fractures in infants—a post mortem radiologic-pathologic study. JAMA 272: 617–618 34. Feldman K W, Brewer D K 1984 Child abuse, cardiopulmonary resuscitation and rib fractures. Pediatrics 73: 339–342 35. Blazer S, Naveh Y, Friedman A 1980 Foreign body in the airway—a review of 200 cases. Am J Dis Child 134: 68–71 36. Svedstrom E, Puhakka H, Kero P 1989 How accurate is chest radiography in the diagnosis of tracheobronchial foreign bodies in children? Pediatr Radiol 19: 520–522 37. Harris V J, Brown R 1975 Pneumatoceles as a complication of chemical pneumonia after hydrocarbon ingestion. Am J Roentgenol Radium Ther 125: 531–537 38. Oh K S, Newman B, Bender T M, Bowen A 1988 Radiologic evaluation of the diaphragm. Radiol Clin North Am 26: 355–364 39. Moccia W A, Kaude J V, Felman A H 1981 Congenital eventration of the diaphragm. Pediatr Radiol 10: 197–200 40. Cohen M, Sprigg A. 2005 Cystic malformations of the lung—a diagnostic approach. Cell Pathol 6: 94–99 41. Felker R E, Tonkin I L D 1990 Imaging of pulmonary sequestration. Am J Roentgenol 154: 241–249 42. Doyle A J 1992 Demonstration of blood supply to pulmonary sequestration by MR angiography. Am J Roentgenol 158: 989–990 43. Chilvers E R, Peters A M, George P, Hughes J M B, Allison D J 1988 Quantification of right to left shunt through pulmonary arteriovenous malformations using 99mTc albumin microspheres. Clin Radiol 39: 611–614 44. King R M, Telander R L, Smithson W A, Banks P M, Han M T 1982 Primary mediastinal tumors in children. J Pediatr Surg 17: 512–520 45. Merten D F 1992 Diagnostic imaging of mediastinal masses in children. Am J Roentgenol 158: 825–832 46. Carty H, Martin J 1993 Review: staging of lymphoma in childhood. Clin Radiol 48: 151–159 47. Carachi R, Burgner D P 1994 Thoracic foregut duplications. Arch Dis Child 71: 395–396

• THE NEONATAL AND PAEDIATRIC CHEST

48. Kincaid P K, Stanley P, Kovanlikaya A, Mahour G H, Rowland J M 1999 Coexistent neurenteric cyst and enterogenous cyst—further support for a common embryologic error. Pediatr Radiol 29: 539–541 49. Fink I J, Kurtz D W, Cazenave L, et al 1985 Malignant thoracopulmonary small-cell (Askin) tumour. Am J Roentgenol 145: 517–520 50. Connolly B L, Chait P G, Duncan D S, Taylor G 1999 CT-guided percutaneous needle biopsy of small lung nodules in children. Pediatr Radiol 29: 342–346 51. Sargent M A, McEachern A M, Jamieson D H, Kahwaji R 1999 Atelectasis on pediatric chest CT: comparison of sedation techniques. Pediatr Radiol 29: 509–513 52. Amodio J B, Berdon W, Abramson S, Baker D 1987 Cystic fibrosis in childhood: pulmonary, paranasal sinus and skeletal manifestations. Semin Roentgenol 22: 125–135 53. Te Meerman G J, Dankert-Roelse J, Martijn A, Van Woerden H H 1985 A comparison of the Shwachmann, Crispin–Norman and Brasfield methods of scoring chest radiographs of patients with cystic fibrosis. Pediatr Radiol 15: 98–101 54. Bhalla M, Turcios N, Aponte V, et al 1991 Cystic fibrosis: scoring system with thin section CT. Radiology 179: 783–788 55. Rogalla P, Stover B, Scheer I, Juran R, Gaedicke G, Hamm B 1998 Lowdose spiral CT: applicability to paediatric chest imaging. Pediatr Radiol 28: 565–569 56. Santamaria F, Grillo G, Guidi G et al 1998 Cystic fibrosis: when should high-resolution CT of the chest be obtained? Pediatrics 101: 908–913 57. Simmonds E J, Littlewood J M, Evans E G V 1990 Cystic fibrosis and allergic bronchopulmonary aspergillosis. Arch Dis Child 65: 507–511 58. Owens C 2004 Radiology of diffuse interstitial pulmonary disease in children. Eur Radiol 14 (suppl): L2–L12 59. Arakawa H, Webb W R 1998 Air trapping on expiratory high-resolution CT scans in the absence of inspiratory scan abnormalities. Am J Roentgenol 170: 1349–1353 60. Wilkinson A G, Paton J Y, Gibson N, Howatson A G 1999 CT-guided 14-G cutting needle lung biopsy in children: safe and effective. Pediatr Radiol 29: 514–516 61. Sotomayer J L, Douglas S D, Wilmott R W 1989 Pulmonary manifestations of immune deficiency diseases. Pediatr Pulmonol 6: 275–292 62. Haller J O, Cohen H L 1994 Pediatric HIV infection: an imaging update. Pediatr Radiol 24: 224–230 63. Haller J O, Ginsberg K J 1997 Tuberculosis in children with acquired immunodeficiency syndrome. Pediatr Radiol 27: 186–188 64. Oldham S A A, Castillo M, Jacobson H L, Mones J M, Saldana M J 1989 HIV-associated lymphocytic interstitial pneumonia: radiologic manifestations and pathologic correlation. Radiology 170: 83–87

Suggested further reading Carty H, Brunelle F, Stringer D A, Kao S C S (eds) 2005 Imaging children, 3rd edn, vol 1. Churchill Livingstone, Edinburgh, chapter 7, pp 1021–1178 A practical paediatric radiology text in two volumes. Duncan A W 1999 Emergency chest radiology in children. In: Carty H (ed) Emergency pediatric radiology. Springer Verlag, Berlin, chapter 2, pp33–116 Kuhn S P, Slovis T L, Haller J O (eds) 2004 Caffey’s pediatric diagnostic imaging, 10th edn, vol 1. Mosby, St Louis, pp 48–103, 765–1126 A paediatric reference text in two volumes. Newman B (ed) 1993 The pediatric chest. Radiol Clin North Am 31: 453–704 An excellent ‘clinics’ covering the pediatric chest. Swischuk L (ed) 1997 Imaging of the newborn infant and young child, 4th edn. William and Wilkins, Baltimore, pp 1–158

1485

CHAPTER

Paediatric Abdominal Imaging

65

Anne Paterson, Louise E. Sweeney and Bairbre Connolly

The neonate • Clinical features of neonatal gastrointestinal disease The infant and older child • Abominal pain • Abdominal manifestations of cystic fibrosis • The immunocompromised child • Miscellaneous gastrointestinal problems The paediatric liver, biliary system and spleen • Liver disorders • Biliary disease • Vascular diseases of the liver • Liver transplantation • The spleen

Paediatric gastrointestinal (GI) radiology is most appropriately considered according to the age of the patient. For this reason, this chapter has been divided into two main sections. The first half of the chapter concentrates on neonatal GI disease, the latter half on GI problems in the infant and older child. Throughout the chapter, the relevant clinical features of each condition are presented, followed by the applicable imaging strategy in each case. Radiological techniques that are peculiar to paediatric radiology are discussed in more detail. The clinical features of neonatal GI disease (Table 65.1) are varied, but may coexist in different conditions. Each problem

is therefore presented under the heading of its most common physical symptoms. The spectrum of GI disease in the infant and older child overlaps with conditions seen in adult patients. This chapter deals primarily with problems specific to paediatrics and the reader is referred to the Chapters on adult GI diseases for further information. Common complaints such as abdominal pain, constipation, abdominal trauma (including nonaccidental injury), nonbilious vomiting, abdominal distension, malabsorption, and GI bleeding are detailed. The GI manifestations of cystic fibrosis, the abdomen in the immunocompromised child, and other miscellaneous GI problems are also discussed. For completeness, this chapter has been expanded to include paediatric hepatic and biliary disease.

Table 65.1 PRESENTING FEATURES OF NEONATAL GASTROINTESTINAL DISEASE Visible defects in the anterior abdominal wall Respiratory distress and choking Nonbilious vomiting Bilious vomiting Abdominal distension Delayed passage of meconium Rectal bleeding Jaundice

THE NEONATE CLINICAL FEATURES OF NEONATAL GASTROINTESTINAL DISEASE Visible abnormalities of the anterior abdominal wall The ventral wall of the embryo is formed during the fourth week of intrauterine development as the cephalic, caudal, and lateral edges of the flat, trilaminar embryonal disc fold in upon themselves. The resultant embryo is cylindrical in shape and

protruding centrally from its ventral surface are the remains of the yolk sac connected to the midgut; a structure that later becomes the umbilical cord. If this complex process is incomplete, then several types of anterior abdominal wall defect may result. Such defects may be detected and characterized by antenatal ultrasound (US). The management and planning of the remainder of the pregnancy and the subsequent delivery of the infant are all affected by the prenatal diagnosis.

1488

SECTION 8

• PAEDIATRIC IMAGING

Omphalocele An omphalocele (exomphalos) is a midline anterior abdominal wall defect through which the solid abdominal viscera and/or bowel may herniate. Larger omphaloceles containing liver tissue are thought to be due to failure of fusion of the lateral body folds. Omphaloceles containing only bowel are thought to arise due to persistence of the physiological herniation of gut after the 10th week of fetal development1. The umbilical cord inserts at the tip of the defect. A giant omphalocele is said to be present when the liver and biliary structures are contained within the herniated membranes. Prognosis is dependent upon associated anomalies. Chromosomal abnormalities are seen in around 50% of patients2. Trisomy 13 and 18 are common. The Beckwith–Wiedemann syndrome has an omphalocele (exomphalos), macroglossia, and gigantism as its primary components (the ‘EMG’ syndrome). Visceral abnormalities are present in up to 70% of cases2.

Gastroschisis The term gastroschisis literally means ‘split stomach’.There is a small defect or split in the ventral abdominal wall, classically to the right side of a normally positioned umbilicus. Gastroschisis typically occurs in the absence of other anomalies and is thought to be due to a localized intrauterine vascular accident, which leads to full-thickness necrosis of a portion of the anterior abdominal wall3. Antenatal US shows (usually only) bowel loops floating freely in the amniotic fluid. There is no covering membrane. Exposure to amniotic fluid damages the bowel; this too may be suspected antenatally if the extruded loops of bowel are dilated and thickened1. Postnatally, this can result in a thick, fibrous ‘peel’ coating the loops of bowel. Short bowel syndrome and intestinal dysmotility are serious consequences of gastroschisis. Associated intestinal atresias and stenoses are secondary to the prenatal ischaemic insult3. Necrotizing enterocolitis (NEC) is reported in up to 20% of patients with gastroschisis3. The morbidity and mortality associated with gastroschisis is mainly due to the additional GI problems, compounded by the prolonged total parenteral nutrition that some patients require; cholestatic liver disease is the major problem. The establishment of enteral feeding often takes longer in preterm infants, resulting in increased hospital stays for this group4. Respiratory embarrassment may occur following repair of the defect. Around one-third of male patients have cryptorchidism, which may result in the testes passing through the abdominal wall defect. Most testes descend to the scrotum after being returned to the abdominal cavity2. Upper GI contrast medium studies in infants with repaired gastroschisis will often demonstrate gastro-oesophageal reflux (GOR), malrotation, dilatation of small bowel loops, and a markedly prolonged transit time. GI transit times greater than 2 d are not uncommon and do not necessarily indicate intestinal obstruction.

Cloacal exstrophy is a rare, midline, infra-umbilical defect that arises due to an abnormality of the caudal body fold. Cloacal exstrophy is more common in boys and is composed of an omphalocele, imperforate anus, spinal dysrhaphism, and

ambiguous genitalia. The caecum opens onto the anterior abdominal wall between two exstrophied hemibladders. Postnatally the bladder and bowel are separated and repaired, and the anterior abdominal wall defect is closed. In the past, genetic male infants with cloacal exstrophy have commonly undergone bilateral orchidectomy and been reared as girls, as the external genitalia are frequently rudimentary. Genetic female infants usually have a double vagina and two hemiuteri. Genital tract imaging and surgery may be delayed until later in childhood. Early postnatal imaging includes US examination of the renal tracts, and upper GI contrast medium studies to detect malrotation and to outline the length of bowel present. Magnetic resonance imaging (MRI) will exclude an associated spinal cord abnormality and delineate the pelvic organs and pelvic floor musculature.

Choking and respiratory distress The neonate with disease affecting the proximal GI tract often presents with respiratory symptoms. These symptoms include choking with feeds, cyanosis, and respiratory distress. Conditions that present in this way include oesophageal atresia (OA) with or without tracheo-oesophageal fistula (TOF), laryngeal clefts, swallowing disorders, diaphragmatic hernias, vascular rings, and gastro-oesophageal reflux (GOR); the latter is by far the most common. The chest radiograph may show airspace disease and atelectasis, should aspiration have occurred.

Oesophageal atresia and tracheo-oesophageal fistula OA with or without a fistulous connection to the trachea occurs in 1 in 3000–4500 live births. It is due to abnormal partitioning of the laryngotracheal tube from the oesophagus by the tracheo-oesophageal septum during the fourth week of gestation. Five different major anomalies result (Fig. 65.1). The atretic segment of the oesophagus tends to be at the junction of its proximal and middle thirds, and a TOF, if present, is usually found proximal to the carina. Occasionally an isolated TOF occurs without OA; this is the H- (or N-) type fistula. Half of all children with OA and TOF have associated congenital anomalies. Features of the VACTERL spectrum (vertebral anomalies, anorectal malformation, cardiovascular malformation, tracheo-oesophageal fistula with oesophageal atresia, renal anomalies, and limb defects) are commonly found, with cardiac and other GI malformations—notably duodenal atresia and stenosis, and anorectal malformations. Trisomy 18, trisomy 21, and Potter’s syndrome are other recognized associations5. A more distal congenital oesophageal stenosis may also be present6. The mortality rates in patients with OA and TOF are usually no longer due to the OA itself, but to the associated malformations, especially congenital cardiac anomalies. The diagnosis of OA may be suspected antenatally, when US demonstrates maternal polyhydramnios. A fetus with OA and no distal fistula will also have absence of the gastric bubble. With a TOF, there may be a normal or slightly small gastric bubble and associated polyhydramnios. The remaining infants present almost immediately in the postnatal period with choking, coughing, cyanosis, and

CHAPTER 65

Figure 65.1 Oesophageal atresia (OA) and tracheo-oesophageal fistula. Diagrammatic representation (A) isolated OA (9%), (B) H-type fistula (6%), (C) OA with distal tracheo-oesophageal fistula (TOF) (82%), (D) OA with proximal TOF (1%), and (E) OA with TOF from both proximal and distal oesophageal remnants (2%).

drooling; these symptoms tend to be exacerbated during attempts to feed the infant. Patients with an H-type fistula are usually symptomatic from birth. The initial radiograph will show the orogastric tube curled in the proximal oesophageal pouch (Fig. 65.2). The diagnosis can be confirmed if air is gently injected via the tube to distend the oesophageal pouch. The lungs may show features of an aspiration pneumonitis. Vertebral anomalies and an abnormal cardiac silhouette may be visible. The presence of gas in the abdomen implies a distal fistula. More recently, the ultrasound appearance of OA with or without TOF has been described in

Figure 65.2 Oesophageal atresia. Supine chest radiograph shows orogastric tube curled in the proximal oesophageal pouch.

•

PAEDIATRIC ABDOMINAL IMAGING

the literature and this is a promising technique for use at the bedside of a sick neonate7. A gasless abdomen is seen with isolated OA and OA with a proximal fistula. Obstruction or atresia of the fistulous tract will also result in a gasless abdomen. In isolated OA, a long gap between the atretic segments is seen in association with 13 pairs of ribs8. The gap between the oesophageal pouches can be assessed following the formation of a feeding gastrostomy. Under fluoroscopic guidance, a Hegar dilator is inserted through the gastrostomy and retrogradely into the distal oesophagus. A Repogle tube is simultaneously used to delineate the superior pouch. As both tubes are radio-opaque, the degree of separation between the pouches is easily visualized. Alternatively, following simultaneous injection of air—into the upper pouch via the indwelling Repogle tube and via the gastrostomy—coronal computed tomography (CT) images can delineate the gap between the oesophageal pouches, which assists surgical management9. Neonates with an H-type fistula commonly have an abdomen distended by gas. The fistula is easy to miss on a routine upper GI study. An oesophogram with the patient placed prone and using a horizontal X-ray beam is recommended. The contrast medium is injected under pressure, via a nasogastric tube, with its tip in the distal oesophagus. The tube is slowly withdrawn under fluoroscopic guidance. The majority of these fistulas are seen at the level of the thoracic inlet (Fig. 65.3). Combined bronchoscopy and oesophagoscopy should be performed if there is a high clinical index of suspicion with negative imaging. Contrast medium is rarely necessary to outline the proximal oesophageal pouch, but is occasionally required to exclude a proximal pouch fistula. Small quantities (1–2 ml) of isotonic, nonionic contrast medium should be used under fluoroscopic guidance; this will do little damage to the lungs should it be aspirated. A proximal fistula can also be demonstrated if air is injected into the Repogle tube during multidetector CT (MDCT). Multiplanar reconstructions and volume rendered images will then define the abnormal anatomy10. Post surgical radiology The immediate complications following OA and TOF repair are recurrence of the TOF, which occurs in up to 10% of patients, and anastomotic breakdown in a further 10–20% of patients11,12. A recurrent fistula may be difficult to demonstrate but is suspected radiologically if the oesophagus is gas filled on the plain radiograph and if contrast medium studies show ‘beaking’ of the anterior oesophageal wall11. Other common complications following OA and TOF repair include anastomotic strictures, disordered oesophageal and more distal GI motility, and GOR. These problems may persist into adulthood and be lifelong. The incidence of strictures is increased in those patients who have had long gap OA with a delayed primary repair; figures of up to 80% have been reported12. Many surgeons will request postoperative oesophograms following OA and TOF repair and before the child is fed.

1489

1490

SECTION 8

• PAEDIATRIC IMAGING

bubble. Postnatally upper abdominal distension and nonbilious vomiting occur. The symptoms may be intermittent and mild when the obstruction is incomplete. With complete obstruction, a plain radiograph will show a dilated stomach with no distal air. The upper GI series will show vigorous gastric peristalsis and consistent filling defects in the antrum or pylorus at the site of the web. An antral web shows the ‘pseudo double bubble’, as barium outlines first the space between the antrum and pylorus, and second the duodenal bulb. At US, webs appear as persistent, linear, echogenic structures arising from the antral or pyloric walls and extending centrally17.

Bilious vomiting Bilious vomiting in the neonate may be the presenting symptom of both ‘medical’ and ‘surgical’ conditions. Medical conditions include functional immaturity of the colon and gastroenteritis. Surgical conditions include obstructions distal to the ampulla of Vater, of which malrotation and midgut volvulus constitute the greatest emergency. Unfortunately, the plain abdominal radiograph cannot always accurately distinguish between the conditions mentioned. If the plain radiograph demonstrates a complete high intestinal obstruction then no further imaging is required. If the radiograph shows a low intestinal obstruction (i.e. distal to the mid ileum), a contrast medium enema is preferred. Should any other plain radiographic findings be present, then an upper GI contrast medium study must be performed. Care must be taken not to overfill the stomach with contrast medium as this can obscure the position of the duodenojejunal flexure. Figure 65.3 H-type tracheo-oesophageal fistula. Upper GI contrast study shows the fistula running obliquely at the level of the thoracic inlet.

Oesophageal strictures in this group of patients can easily and safely be treated by balloon catheter dilatation using fluoroscopic or endoscopic guidance13–15.

Nonbilious vomiting Vomiting is a common problem in children of all ages and its presence as a symptom does not necessarily indicate GI disease. Vomiting is common with infections in any body system, in metabolic disease, disorders of the central nervous system, and as a side-effect of drugs and poisons. Neonatal nonbilious vomiting due to GI causes implies a lesion proximal to the ampulla of Vater and is most frequently due to GOR. A full clinical history and physical examination are the most important factors in determining what imaging investigations (if any) are required.

Obstruction of the stomach Congenital gastric obstruction is rare. It is usually due to a web or diaphragm in the antrum or pylorus. Occasionally a true atresia is present with a fibrous cord uniting the two blind ends. Pyloric atresia is associated with epidermolysis bullosa simplex, when there are mutations in the gene PLEC 116. The diagnosis may be suspected antenatally due to maternal polyhydramnios and a large fetal gastric

Malrotation and volvulus Malrotation is a serious congenital abnormality of the GI tract, as it predisposes to duodenal obstruction and midgut volvulus, which can lead to ischaemic necrosis of the small bowel. Undiagnosed midgut volvulus carries a high mortality rate. In fetal life, the gut begins as a straight, midline tube which as it elongates and develops herniates into the base of the umbilical cord. Between the 6th and 10th weeks of fetal development, the midgut loop rotates 90 degrees anticlockwise around the axis of the superior mesenteric artery (SMA). At this stage the duodenojejunal (proximal) loop thus lies to the right, and the caecocolic (distal) loop to the left side. During the 10th week, the intestines return to the abdominal cavity; the proximal loop of bowel enters first, followed by the distal loop. Both the proximal and distal loops undergo a further 180 degrees of anticlockwise rotation as they return to the abdominal cavity; a total of 270 degrees of rotation. The duodenojejunal loop comes to lie posterior to and the caecocolic loop anterior to the SMA.The duodenojejunal junction (fixed by the ligament of Treitz) should lie in the left upper quadrant of the abdomen, and the ileocaecal junction in the right lower quadrant. The small bowel mesentery extends over the distance from the ligament of Treitz to the ileocaecal junction18,19. Malrotation is a generic term used to describe any variation in the position of the intestines, and is in itself not necessarily symptomatic. Malfixation of the intestines invariably

CHAPTER 65

accompanies malrotation in an attempt to fix the gut in place. Peritoneal (Ladd) bands stretch from the abnormally high lying caecum, across the duodenum to the region of the portahepatis and the anterior and posterior abdominal walls. The Ladd bands can cause duodenal obstruction. In addition the abnormal positions of the duodenojejunal junction and the caecum mean the base of the small bowel mesentery is short. The midgut has a propensity to twist around this narrow base, compromising its vascular supply (midgut volvulus). Most patients have isolated malrotation of the bowel, but there is an increased incidence of associated duodenal stenosis and atresia. Patients with omphalocele, gastroschisis, and congenital diaphragmatic hernia all have malrotation and abnormal fixation of the bowel, although volvulus in these patients is rare after repair of the primary abnormality20. The heterotaxy syndromes, Hirschsprung’s disease, and megacystis–microcolon– intestinal hypoperistalsis (Berdon) syndrome are also associated with malrotation and volvulus. Congenital short bowel without atresia is an extremely rare condition that can occur with malrotation21. Babies with malrotation commonly present within the first month of life, with bilious vomiting. Older children may present with non-specific symptoms of chronic or intermittent abdominal pain, nonbilious emesis, diarrhoea, or failure to thrive. Volvulus, though less common in the older child, still occurs. On physical examination the child may appear relatively well. When bowel ischaemia and necrosis have developed, symptoms of shock supervene. It is important to suspect malrotation and volvulus in a child of any age with bilious vomiting, and to perform an upper GI contrast medium study. There are no specific plain radiographic findings in malrotation, even with volvulus. The radiograph may be completely normal if the volvulus is intermittent or if there is incomplete duodenal obstruction due to a loose twisting of the bowel. If the volvulus is tight, then complete duodenal obstruction results, with gaseous distension of the stomach and proximal duodenum, mimicking the ‘double bubble’ of duodenal atresia. The classical picture is of a partial duodenal obstruction, with mild distension of the stomach and proximal duodenum, with some distal gas (Fig. 65.4). A pattern of distal small bowel obstruction is seen in a closed loop obstruction and represents a more ominous finding. The small bowel loops may be thick walled and oedematous, with pneumatosis sometimes evident.These findings represent small bowel necrosis; the volvulus is so tight that venous occlusion has occurred and the gas cannot be re-absorbed from the bowel lumen. A gasless abdomen is seen if vomiting has been prolonged, and in closed loop obstruction with viable small bowel or massive midgut necrosis. Radiographic examination should cease after the plain radiograph in the neonate with bilious vomiting and a complete duodenal obstruction, or a seriously ill child with obvious signs of peritonism; surgery is indicated. In all other children with bilious vomiting an upper GI contrast medium study is required. The examination is usually performed with barium.

•

PAEDIATRIC ABDOMINAL IMAGING

Figure 65.4 Malrotation and volvulus. Upper GI contrast medium study demonstrates the classical ‘corkscrew’ pattern of the duodenum and jejunum spiralling around the mesenteric vessels.

The intestines in malrotation are malfixed, and the purpose of the upper GI study is to locate the position of the duodenojejunal junction. On a supine radiograph the normal duodenojejunal junction lies to the left of the left-sided pedicles at the height of the duodenal bulb. When malrotation is present, the duodenojejunal junction is usually displaced inferiorly and to the right side. On a lateral view, the junction of the second and third parts of the duodenum is normally retroperitoneal, but it turns sharply anterior in malrotation. The more distal jejunal loops lie to the right of the midline. The caecal pole may lie high, and more to the left side than normal in patients with malrotation. Given that the caecal position is normal in a significant number of patients with malrotation, a barium meal is preferable to a barium enema in this situation. In cases of doubt over the position of the duodenojejunal junction, a delayed radiograph to demonstrate the position of the ileocaecal junction is useful. The normal duodenojejunal junction may be mobile, especially in children younger than 4 months of age. The duodenojejunal junction can be displaced temporarily by a distended colon or stomach, an enlarged spleen, an indwelling naso-enteric tube, or manual palpation. The ‘corkscrew’ pattern of the duodenum and jejunum spiralling around the mesenteric vessels is pathognomonic for midgut

1491

1492

SECTION 8

• PAEDIATRIC IMAGING

volvulus on the upper GI study. When Ladd bands are causing duodenal obstruction rather than volvulus, the duodenojejunal course has been described as ‘Z-shaped’ rather than spiral22,23. US may demonstrate the dilated, fluid-filled stomach and proximal duodenum when obstruction is present. The relationship of the superior mesenteric vein (SMV) to the SMA is abnormal in about two-thirds of patients with malrotation, when the vein lies ventral or to the left of the artery. This sign is neither sensitive nor specific for malrotation, as some patients have abnormally related superior mesenteric vessels in the absence of malrotation24,25. The volvulus itself may be demonstrated with US as the ‘whirlpool sign’. Colour Doppler studies show the SMV spiralling clockwise around the SMA. US is a useful screening tool in the infant with bilious vomiting, but the upper GI study remains the standard technique for the diagnosis of midgut malrotation and volvulus.

Duodenal atresia and stenosis Duodenal atresia is much more common than duodenal stenosis. Both atresia and stenosis are caused by failure of recanalization of the duodenal lumen after the sixth week of fetal life. Duodenal obstruction may also be caused by partial or complete webs or diaphragms. Extrinsic duodenal compression by an annular pancreas or preduodenal portal vein may contribute to the obstruction in some patients. Regardless of the cause, in 80% of cases, the level of the obstruction is just distal to the ampulla of Vater. Associated anomalies occur in the majority of patients with duodenal atresia or stenosis. Down’s syndrome is present in 30% of patients and congenital heart disease in 20%. Malrotation is present in 20–30% of patients and can only be diagnosed before surgery if the duodenal obstruction is partial. Components of the VACTERL association may also be present. Duodenal atresia may be diagnosed antenatally when the dilated stomach and duodenal cap are seen with US in late pregnancy. Maternal polyhydramnios and consequent prematurity are common. Infants otherwise present early in the postnatal period with bilious vomiting and upper abdominal distension. Nonbilious emesis occurs in those infants with a pre-ampullary obstruction. Radiographs show a gas-filled ‘double bubble’ of the stomach and duodenal cap (Fig. 65.5). If the stomach has been decompressed by vomiting or an orogastric tube, then air can be injected via the tube to confirm the diagnosis.With a complete obstruction, no distal gas is seen. If the obstruction is partial or, in the rare cases of a bifid pancreatic duct, straddling the atretic segment, then distal gas will be present. In an upper GI study, duodenal stenosis is seen as a narrowed area in the second part of the duodenum. A duodenal web may be seen as a thin, filling defect extending across the duodenal lumen. If the study is performed via a nasogastric tube, then the pressure of the tube on the obstructing web causes indrawing of the duodenal wall at the site of the web’s attachment (the ‘duodenal dimple’)26. A duodenal web may also be diagnosed using US27.

Small bowel atresia and stenosis Jejunal and ilealatresias have a common aetiology and are due to intrauterine vascular

Figure 65.5 Duodenal atresia. Supine radiograph shows the classical ‘double bubble’ appearance.

problems. The vascular insult may be a primary or secondary event (e.g. due to antenatal volvulus or intussusception). As with the duodenum, atresia is more common than stenosis, and the proximal jejunum and distal ileum are more frequently affected. The ‘apple peel’ syndrome is thought to follow intrauterine occlusion of the distal SMA. There is a proximal jejunal atresia, with agenesis of the mesentery and absence of the mid small bowel. The distal ileum spirals around its narrow vascular pedicle; an appearance that gives the syndrome its name. A malrotated microcolon is also usually present28,29. A second, more complex type of intestinal atresia is the syndrome of multiple intestinal atresias with intraluminal calcifications30,31. The majority of infants with a small bowel atresia present with bilious vomiting in the immediate postnatal period.With more distal atresias abdominal distension is also present. On the plain radiograph, there are dilated loops of small bowel down to the level of the atresia. The loop of bowel immediately proximal to the atresia may be disproportionately dilated and have a bulbous contour. With stenosis rather than atresia, bubbles of distal gas are present. Occasionally fine intraluminal calcifications will be seen with a more distal atresia, though these are not as dense as in the familial multiple atresia syndrome. A meconium peritonitis with calcification of the peritoneum will be present if an intrauterine perforation has occurred.

Abdominal distension Abdominal distension in the neonate may be due to mechanical or functional bowel obstruction, an abdominal mass lesion (Table 65.2), ascites, or a pneumoperitoneum. A supine abdominal radiograph will show the distribution and calibre of the bowel loops, intra-abdominal calcifications (Table 65.3), the presence of pneumatosis or portal venous gas, and possibly a soft tissue mass. A pneumoperitoneum may be detected on

CHAPTER 65

Table 65.2 CAUSES OF A NEONATAL INTRA-ABDOMINAL MASS LESION Complicated meconium ileus Dilated bowel proximal to an obstruction Mesenteric or duplication cyst Abscess Genitourinary causes Hydronephrosis Renal cystic disease Mesoblastic nephroma Wilms’ tumour Adrenal haemorrhage Neuroblastoma Retroperitoneal teratoma Ovarian cyst Hydrometrocolpos Haemangio-endothelioma Hepatoblastoma Choledochal, hepatic, or splenic cysts

Table 65.3 CAUSES OF NEONATAL INTRA-ABDOMINAL CALCIFICATIONS Complicated meconium ileus Intraluminal calcifications Low obstruction Anorectal malformations with a fistula to the urinary tract Adrenal Haemorrhage Neuroblastoma Wolman’s disease Hepatobiliary disease Haemangio-endothelioma Hepatoblastoma TORCH infections Duplication and mesenteric cysts Nephrocalcinosis Intravascular thrombus Teratomas

a supine radiograph, but if small is more likely to be seen on a (preferably left-side down) lateral decubitus or a lateral ‘shoot through’ radiograph with a horizontal beam. US will identify free fluid or the presence of a mass lesion, and is able to confirm the origin of the latter. The majority of neonatal abdominal masses are benign and arise in relation to the genitourinary (GU) tract or are hepatobiliary in origin. Contrast medium studies, scintigraphy, CT, or MRI may follow depending upon the US findings.

Necrotizing enterocolitis (NEC) is the term used to describe the often severe enterocolitis that affects primarily

•

PAEDIATRIC ABDOMINAL IMAGING

premature infants. The increased survival rates of these very low birth weight infants of younger gestational age has led to an increased incidence of NEC in recent years32. The precise aetiology of the condition remains unknown, but immaturity of the gut mucosa and immune response, coupled with ischaemia/hypoxia, are felt to contribute32. Sepsis, early enteral feeding, umbilical arterial and venous cannulation, and maternal cocaine abuse are implicated as additional risk factors,32–34 whilst breastfeeding is associated with a decreased risk of NEC developing32. Mini-epidemics of NEC are known to occur in neonatal intensive care units, suggesting a possible infectious cause. Aside from premature infants and those with low birth weight, NEC is also seen in term infants, particularly those with polycythaemia, cyanotic congenital heart disease, and gastroschisis. In the latter subgroups the condition often develops several weeks after surgery. Initially superficial, the inflammatory process in NEC can extend to become transmural. Diffuse or discrete involvement of the bowel can occur, with the most commonly affected sites being the terminal ileum and colon. Almost half of all cases involve both small and large bowel. The overall mortality rate from NEC is approximately 30%, though this figure is higher in low birth weight infants. The initial clinical symptoms and signs are non-specific and include lethargy, hypoglycaemia, temperature instability, bradycardia, feeding intolerance, increased gastric aspirates, and gastric distension. Disease progression leads to vomiting and diarrhoea (often with the passage of blood or mucus in the stool), and eventually to shock. Severely affected infants may have visibly erythematous anterior abdominal walls, with palpable distended loops of bowel. The initial radiographic features of NEC are non-specific. One of the earliest findings is diffuse gaseous distension of both small and large bowel or isolated gastric distension. If the diameter of a loop of bowel is greater than the width of the L1 vertebral body, then it is likely to be dilated35. Serial radiographs (usually taken every 6–12 h, depending upon the level of clinical suspicion) will demonstrate fixed dilatation of one or more bowel loops, and thickening (oedema) and loss of distinction of the bowel walls as the disease progresses.The use of the contrast medium enema in the acute situation to exclude NEC, when both the clinical and plain radiographic signs are ambiguous, has been suggested by some authors36,37. The risks of sepsis and perforation mean this practice is controversial. A more specific sign of NEC on the plain radiograph is intramural gas (pneumatosis intestinalis), which may be submucosal when it appears as ‘bubbly’ lucencies in the bowel wall or subserosal when linear lucencies are visualized. Not all infants with NEC will have pneumatosis (Table 65.4). More extensive pneumatosis correlates with an increased severity of NEC. Portal venous gas is seen in approximately 10% of cases and is associated with severe NEC, but its presence does not necessarily imply a fatal outcome (Fig. 65.6). The disappearance of intramural or portal venous gas may herald imminent perforation rather than recovery38. There are no definite radiographic signs to help identify those infants most at risk of perforation. Indicators include a solitary, dilated loop of

1493

1494

SECTION 8

• PAEDIATRIC IMAGING

Table 65.4 CAUSES OF PNEUMATOSIS INTESTINALIS IN THE NEONATE AND OLDER CHILD Necrotizing enterocolitis Bowel ischaemia, inflammation, and obstruction Cyanotic congenital heart disease Hirschprung’s disease Gastroschisis Anorectal atresia Inflammatory bowel disease Lymphoma Leukaemia Cytomegalovirus and rotavirus gastroenteritis Colonoscopy Caustic ingestion Short bowel syndrome Congenital immune deficiency states Clostridium infection Chronic granulomatous disease of childhood Chronic steroid use Post hepatic, renal, or bone marrow transplant Collagen vascular disease Graft versus host disease AIDS

bowel present over 24–36 h (the ‘persistent loop sign’)39 and the presence of free intraperitoneal fluid38. US is more sensitive than a plain radiograph in the detection of ascites and portal venous gas.The latter is seen as echogenic particles flowing within the portal vein or focal areas of intrahepatic increased echogenicity. Pneumatosis intestinalis may be recognized with US. The ‘circle sign’ is indicative of bubbles of gas circumferentially in the bowel wall and is seen as a continuous, echogenic ring in cross-section40. Perforation will occur in one-third of children with NEC, most commonly in the ileocaecal region (60% of cases)41. Less than two-thirds of patients with a perforation will have free air visible on a plain radiograph. The supine, cross-table lateral or decubitus view is useful to detect small amounts of free intraperitoneal air, which collects between loops of bowel and is seen as the ‘telltale triangle’42. Almost all patients with NEC who perforate do so within 30 h of diagnosis41 and after this time supine radiographs alone probably suffice for radiographic follow-up. The presence of free intraperitoneal fluid may also be an indicator of perforation but this is seen in only a further 20% of patients38,41. Perforation in infants with NEC is not an absolute indication for surgical intervention. Peritoneal drains are used in the initial resuscitation of these critically ill infants, delaying the need for surgery and allowing time for systemic recovery. In some instances, a peritoneal drain may provide definitive treatment43. A late complication of NEC is stricturing, which can be single or multiple, and occurs in up to a third of patients. The majority of strictures are short, are found in the colon, and are diagnosed up to 3 months following the acute illness44. Contrast medium studies are indicated in infants following surgery, before re-anastomosis of the defunctioned bowel. In rare instances, an acquired intestinal atresia may develop following NEC. Other reported late complications of NEC include abscess formation, enteric fistulas, enterocyst formation, obstruction secondary to adhesions, malabsorption, and short bowel syndrome following surgical resection45–47.

Delayed passage of meconium

Figure 65.6 Necrotizing enterocolitis. Supine radiograph demonstrates multiple dilated loops of bowel, extensive submucosal and subserosal pneumatosis, and portal vein gas.

All infants should pass meconium in the first 24–48 h of life. Failure to do so may be due to an underlying bowel atresia or obstruction (Table 65.5) and leads to progressive abdominal distension. The more common problems encountered include Hirschsprung’s disease, functional immaturity of the colon (meconium plug syndrome, small left colon), meconium ileus and peritonitis, and ileal atresia. In all cases, a supine abdominal radiograph will show the features of a low intestinal obstruction; there will be multiple dilated loops of bowel down to the level of the obstruction. Differentiation between small and large bowel to determine the precise level of the obstruction is virtually impossible in the neonate, given that both may be of similar calibre and that the haustra are poorly developed. A horizontal beam or lateral decubitus view demonstrates fluid levels or confirms perforation and is not always necessary.

CHAPTER 65

Table 65.5 CAUSES OF DELAYED PASSAGE OF MECONIUM IN INFANTS Ileal atresia Meconium ileus Functional immaturity of the colon Colon atresia Anorectal malformations Hirschprung’s disease Megacystis–microcolon–intestinal hypoperistalsis syndrome Extrinsic compression of the distal bowel by a mass lesion Mesenteric cyst Enteric duplication cyst Paralytic ileus, sepsis, drugs, and metabolic upset

Hirschsprung’s disease is a form of functional low bowel obstruction, which is due to failure of caudal migration of neuroblasts in the developing bowel.There is thus an absence of parasympathetic intrinsic ganglion cells in both Auerbach’s and Meissner’s plexuses in the bowel wall. The distal large bowel from the point of neuronal arrest to the anus is aganglionic. The existence of ‘skip lesions’ in Hirschsprung’s disease is extremely unusual48. In about 75% of cases, the aganglionic segment extends only to the rectosigmoid region (short segment disease). Long segment disease involves a portion of the colon proximal to the sigmoid. Variants of Hirschsprung’s disease include total aganglionosis coli and total intestinal Hirschsprung’s disease. Ultrashort segment disease is rare and involves only the anus at the level of the internal sphincter. In short segment disease there is a male preponderance but the sex ratio is equal in long segment disease49. The latter has a strong familial incidence; short segment disease is sporadic. Approximately 5% of children with Hirschsprung’s disease have Down’s syndrome. Other associations with Hirschsprung’s disease include ileal and colonic atresias, cleft palate, polydactyly, craniofacial anomalies, cardiac septal defects and other neurocristopathies50–53. Neonates present with abdominal distension, vomiting (which may be bilious) and failure to pass meconium. Stooling may follow a digital rectal examination or the insertion of a rectal thermometer, before the symptoms recur. Children who present later in childhood are unusual but may do so with a history of chronic constipation and failure to thrive, or rarely with an acute abdomen secondary to colonic volvulus54. Severe bloody diarrhoea, sepsis, and shock are associated with Hirschsprung’s enterocolitis, which occurs in up to 30% of patients in both the pre- and post-operative periods. Enterocolitis is the leading cause of death in Hirschsprung’s disease and has an increased frequency in long segment disease. Other postoperative complications of Hirschsprung’s disease include fistulas and stenoses at the anastomotic site. In the longer term, chronic constipation, soiling, and incontinence are recognized. A suction or full thickness rectal biopsy is required for the definitive diagnosis of Hirschsprung’s disease.

•

PAEDIATRIC ABDOMINAL IMAGING

The abdominal radiograph will typically show a low bowel obstruction, commonly with colonic dilatation out of proportion to the small bowel. The absence of rectal gas is one of the plain radiographic findings in Hirschsprung’s disease but the sign is not specific, being more commonly seen in infants with sepsis and NEC55. About 5% of infants will have a pneumoperitoneum; the perforation occurs most commonly in the ascending colon and may be appendiceal. Perforation is more common in long segment disease. Intraluminal small bowel calcifications may also be present in long segment disease56,57. A water-soluble contrast medium enema should be performed and has a diagnostic accuracy equivalent to barium58. The catheter tip is placed just inside the rectum. It is important that a balloon catheter is not used. A balloon can obscure the diagnostic features or worse, perforate the stiff, aganglionic bowel. The most important radiograph is a lateral view of the rectum during slow filling (Fig. 65.7). In short segment disease the rectum will be narrow and there will be a cone-shaped transition zone to the more proximal, dilated, ganglionated bowel. The radiological transition zone is commonly found to be distal to the pathological transition zone. Irregular contractions may be seen in the denervated rectum. The transition zone may not be present in the neonate, as it takes time for the proximal bowel to dilate. A useful calculation is the recto:sigmoid ratio; the rectum should always be the most distensible portion of the bowel and have a diameter greater than that of the sigmoid colon (recto:sigmoid ratio > 1). In short segment disease this ratio is reversed. Retention of contrast medium above the sigmoid colon on a delayed radiograph after 24 h is a non-specific sign of Hirschsprung’s disease. With coexistent enterocolitis, mucosal oedema, ulceration, and spasm are seen, although an enema would obviously be contraindicated in the infant with fulminant colitis (Fig. 65.8). Giant stercoral ulcers may also be

Figure 65.7 Rectosigmoid Hirschsprung’s disease. Lateral view, contrast medium enema. The cone-shaped transition zone, abnormal rectosigmoid ratio, and tertiary rectal contractions are demonstrated.

1495

1496

SECTION 8

• PAEDIATRIC IMAGING

loss of the normal redundancy of the flexures. Muscle spasm, a pseudo-transition zone, and easy reflux of contrast medium into the terminal ileum are also reported. Occasionally a microcolon is seen49.

Functional immaturity of the colon Immature left colon

Figure 65.8 Hirschsprung’s enterocolitis. Contrast medium enema shows extensive ulceration of the entire colon.

seen in older children with a delayed presentation49. The radiological features of Hirschsprung’s disease may be absent in the neonate and overall the contrast medium enema has a false-negative rate of 20–30%58. In total aganglionosis coli the findings are notoriously unreliable but include shortening of a normal calibre colon, with

(meconium plug syndrome or small left colon) is a relatively common cause of neonatal bowel obstruction. Prematurity, birth by caesarean section, maternal diabetes and drug ingestion, and treatment of mothers with magnesium sulphate during labour have all been reported as risk factors to infants.33,59,60 The condition is not associated with cystic fibrosis. The exact cause of the syndrome remains unknown, but immaturity of the myenteric plexus had been postulated as a theory61. The affected infants present with symptoms and signs of bowel obstruction, though they tend to be less ill than those with mechanical obstruction; the abdomen is less distended and vomiting is not necessarily a prominent feature.There is delayed passage of meconium. The plain radiograph shows distension of both small and large bowel loops to the level of the inspissated meconium plugs. Few fluid levels are seen. The contrast medium enema typically shows a microcolon distal to the splenic flexure, at which point there is an abrupt transition to a mildly dilated proximal colon (this differs from Hirschsprung’s disease in which the transition zone is more gradual and which is uncommon at the level of the splenic flexure) (Fig. 65.9). The rectum is usually distensible in the patient with functional immaturity of the colon and consequently the rectosigmoid

Figure 65.9 Small left colon syndrome. (A) Supine radiograph shows a low obstruction with multiple dilated loops of bowel. (B) Contrast medium enema shows a microcolon distal to the splenic flexure. The transition point is abrupt.

CHAPTER 65

ratio is normal. Discrete plugs of meconium are seen as filling defects in the dilated colon. In the premature infant, the whole colon may be small59. When performed with watersoluble contrast medium, the enema is not only diagnostic, but therapeutic. A variable amount of meconium is typically passed soon after the examination, with a gradual recovery of the infant over the next few hours or days. Bowel perforation has occasionally been reported59. The main differential diagnosis is Hirschsprung’s disease. If there remains doubt as to the diagnosis following the enema, then a suction rectal biopsy is recommended.

Colon atresia is rare when compared with other intestinal atresias, and colonic stenosis is rarer still. The right colon is most commonly affected. As with small bowel atresias, the atresia may take the form of a diaphragm or web, fibrous cord or mesenteric gap defect, with the latter occurring most frequently. Colon atresia is thought to be due to an in utero vascular accident62. More proximal atresias, gastroschisis, and Hirschsprung’s disease may be found in association52,62,63. The abdominal radiograph will show the features of a low intestinal obstruction. If multiple atresias are present, then the bowel will be distended only to the level of the most proximal atresia, giving a misleading initial abdominal radiograph. A contrast medium enema usually demonstrates a distal microcolon, with obstruction to the retrograde flow of contrast medium at the point of the atresia. If a colonic diaphragm or web is present, then the column of barium may terminate in a ‘wind sock’ configuration, as the obstructing membrane balloons into the proximal air-filled colon.

•

PAEDIATRIC ABDOMINAL IMAGING

meconium with gas (Fig. 65.10). This sign is suggestive of, but not diagnostic for, meconium ileus, as it may be seen in other forms of low bowel obstruction. US may be useful in differentiating between meconium ileus and ileal atresia. In meconium ileus, the dilated bowel loops are filled with echogenic material, whereas in ileal atresia the bowel contents are echo poor67. The contrast medium enema in meconium ileus demonstrates a virtually empty microcolon. Reflux of contrast medium into the terminal ileum will show that it too is small in calibre and numerous pellets of meconium are outlined (Fig. 65.11). A spiral appearance of the small bowel loops has been described secondary to intrauterine volvulus in more complicated cases68. More proximal reflux of contrast medium will show the dilated mid ileal loops. The contrast medium enema in uncomplicated meconium ileus may be therapeutic as well as diagnostic. In 1969, Noblett described the nonoperative management of uncomplicated meconium ileus using a Gastrografin enema69. Gastrografin is a water-soluble contrast medium that is hypertonic to plasma and draws water into the bowel lumen by osmosis, softening the viscous meconium and allowing it to be passed. The detergents contained within Gastrografin may also aid in relieving the obstruction.The success rate of Gastrografin enemas in relieving the obstruction in meconium ileus is approximately 60%67,70. The enema can be complicated by perforation in up to 5% of patients70. Due to the hypertonicity of Gastrografin, there is a risk of fluid and electrolyte imbalance in the infant, and the enema should not be performed unless the baby is well hydrated and intravenous fluids are running. Serum electrolytes need to be monitored.

Meconium ileus is a form of distal intestinal obstruction caused by inspissated pellets of meconium in the terminal ileum. Over 90% of infants with meconium ileus have cystic fibrosis and meconium ileus is the presenting feature of cystic fibrosis in 10–15% of affected patients. Of patients with cystic fibrosis, those with the ∆F508 mutation have a higher incidence of meconium ileus64. Black patients with cystic fibrosis have a lower incidence of meconium ileus than white patients64. Children who present with meconium ileus as neonates go on to develop more severe respiratory disease than those without meconium ileus65,66. Over half of the affected infants have uncomplicated meconium ileus. In utero these babies produce meconium which is thick and tenacious, and which fills and distends the small bowel loops. The meconium desiccates in the terminal ileum and becomes impacted, causing a high grade obstruction. A functional microcolon results. Meconium ileus is described as complicated when intra-uterine volvulus, atresia, gangrene, perforation or meconium peritonitis supervene. The presenting clinical symptoms and signs of noncomplicated meconium ileus are vomiting, abdominal distension, and failure to pass meconium. The plain abdominal radiograph will show dilatation of small bowel loops, which are of varying calibre. Fluid levels are scant. Often a ‘soap bubble’ appearance is visible (classically in the right iliac fossa), which is caused by the admixture of

Figure 65.10 Meconium ileus. Supine radiograph shows a low obstruction with multiple, dilated loops of bowel and a ‘soap bubble’ appearance in the right lower quadrant.

1497

1498

SECTION 8

• PAEDIATRIC IMAGING

will present with abdominal distension and delayed passage of meconium. The plain radiograph will show a low obstruction with multiple dilated loops of bowel, one loop of which characteristically has a long air–fluid level. The radiographic findings are similar to those of meconium ileus. The contrast medium enema will outline a microcolon and contrast medium cannot be refluxed into the dilated small bowel. If the atresia has occurred secondary to an ischaemic insult late in the pregnancy, then the colon calibre may be near normal due to the already accumulated succus entericus in its lumen.

Figure 65.11 Meconium ileus. Contrast enema demonstrates the empty microcolon. Contrast refluxes into the narrow terminal ileum, where pellets of meconium are outlined.

Full strength Gastrografin is no longer commonly employed in the management of meconium ileus, rather it is diluted to half strength with saline or water, or another water-soluble contrast medium is substituted70. If the infant’s clinical condition remains stable, then the enema can be repeated as necessary until the obstruction is relieved. Enemas in infants with suspected meconium ileus should only be performed in specialist paediatric centres. If nonoperative treatment fails or complications occur during the enema, then the infant needs surgery. The presence of complicated meconium ileus may be suggested by the plain radiographic findings; intra-abdominal or scrotal calcifications, bowel wall calcification, prominent air–fluid levels, and soft tissue masses. Volvulus of a heavy, meconium-laden loop of bowel is common71.Volvulus can lead to intestinal stenoses, atresias, gangrene and perforation. Perforation of bowel in utero leads to chemical meconium peritonitis. The extruded bowel contents cause an intense inflammatory reaction, with fibrosis and calcification to follow—the characteristic ‘snow storm’ ascites picture on US examinations60. Not all meconium peritonitis is due to meconium ileus, and the associated calcifications are seen more commonly from other causes (e.g. small bowel atresia or intrauterine intussusception)72,73. A meconium pseudocyst is formed in meconium ileus when there is vascular compromise in association with an intra-uterine volvulus. The ischaemic bowel loops become adherent and necrotic, and a fibrous wall develops around them. The wall may calcify and the cyst can have a secondary mass effect upon the adjacent loops of bowel71.

Distal ileal atresia Ileal atresia is due to a prenatal vascular insult. If the atresia is in the distal ileum, then the infant

Anorectal malformations The incidence of anorectal malformations (imperforate anus or anorectal atresia) is 1 in 1500–5000 live births74. The precise aetiology is unknown, but the condition results from failure of descent and separation of the hindgut and the GU tract during the second trimester.The abnormality consists of anorectal atresia, with or without an anomalous connection between the atretic anorectum and the GU tract. Associated congenital anomalies are common. The VACTERL sequence occurs in around 45% of patients, the OEIS complex (omphalocele, bladder exstrophy, imperforate anus and sacral anomalies) in up to 5% of patients74. Between 2 and 8% of patients have Down’s syndrome and these patients tend to have imperforate anus without fistulas74,75. Currarino’s triad is the association between an anorectal malformation (commonly anorectal stenosis), bony sacral anomalies (classically a ‘scimitar sacrum’ with unilateral hypoplasia of the lateral aspect of the vertebral bodies) and a presacral mass lesion (which may be an enteric cyst, teratoma, anterior meningocele or dermoid). Anorectal atresias are classified into high or low lesions depending upon whether the rectum ends above or below the puborectalis sling.The distinction between high and low lesions is clinical rather than radiological, and has important therapeutic and prognostic consequences. In both male and female patients with low lesions, there is usually a visible perineal opening. The orifice may be located more anteriorly than normal (an ectopic anus) and it may be stenotic or covered with a membrane76. Low lesions do not have a communication with the GU tract. Female patients with low lesions will have separate urethral and vaginal orifices with an intact hymen76. Low lesions also include isolated rectal atresia or stenosis. This type of lesion is treated surgically with an anoplasty or dilatation soon after birth. In both male and female patients with high lesions, no visible perineal fistula is present. Male patients will usually have a fistulous tract between the atretic anorectum and the posterior urethra. Less commonly there is a fistula to the bladder or anterior urethra. Female patients have fistulas from the atretic anorectum to the vagina or vestibule. Rarely, in both male and female patients, the rectum ends blindly. High lesions are initially managed with a colostomy, with definitive repair being performed at a later stage. The traditional radiological approach to the baby with imperforate anus is the inverted lateral radiograph. A radioopaque marker is placed over the anal dimple and the distance between the pouch of rectal gas and the marker is measured. False interpretation may occur within the first 24 h of life

CHAPTER 65

when gas may not have reached the rectum, if the infant had not been held prone for long enough, or when meconium is impacted in the distal rectum. Similarly, if the infant is crying or straining, the rectal pouch descends through the levator sling and a high lesion may be misinterpreted as a low one. Transperineal US has been used to measure the distance of the rectal pouch from the perineum77 but interpretation suffers from similar problems to the inverted lateral radiograph. For these reasons, neither of these techniques is recommended. Instead the infant should be clinically assessed to determine whether the lesion is a low or high one (as outlined above). Plain radiographs are more useful if they demonstrate intravesical air (implying a high lesion with a rectovesical or rectourethral fistula in a boy) (Fig. 65.12) or calcified intraluminal meconium (again implying a high lesion in a boy; meconium

Figure 65.12 Colovesical fistula. Supine abdominal radiograph shows intravesical air and vertebral segmentation anomalies in a male infant with a high anorectal malformation. B = bladder.

•

PAEDIATRIC ABDOMINAL IMAGING

calcifies when it comes into contact with urine)78. The plain abdominal radiograph is useful in the identification of any associated bony anomalies of the spine. An augmented pressure colostogram79 is performed in infants with high lesions following the initial formation of a colostomy (Fig. 65.13). A Foley catheter is inserted into the distal segment of colon and its balloon gently inflated to 5 ml. Retrograde traction is then applied to the catheter, so that the balloon seals the stoma. Water-soluble contrast is then hand injected under mild pressure to distend the distal colon and define the fistulous tract. The level of the fistula determines the surgical approach at the time of the definitive repair. Renal US is mandatory as an ancillary investigation in all infants with anorectal atresia. Spinal US should also be performed as spinal cord lesions (e.g. cord tethering) are not uncommon80. Both pre- and post-operative MRI studies may be used to study the pelvic floor musculature and can also reveal any associated renal and spinal malformations.

Figure 65.13 Recto-urethral fistula. Augmented pressure colostogram in a male infant with a high anorectal malformation.

THE INFANT AND OLDER CHILD ABDOMINAL PAIN

Intussusception

Abdominal pain in infants and children is common and is caused by a variety of disorders (Table 65.6). It is better localized in older children than in infants. Plain radiographs of the abdomen are required for suspected mechanical obstruction, perforation and trauma, location of foreign bodies and calculi, and basal pneumonia In non-specific abdominal pain the plain radiographs are frequently unhelpful. US is useful in the diagnosis of a number of GI disorders, including acute appendicitis, intussusception, Henoch-Schönlein purpura (HSP), and mesenteric adenitis, and in excluding other pathology as a cause of abdominal pain. CT is the best investigation for abdominal trauma and is also used in the diagnosis of acute appendicitis in clinically difficult cases.

Intussusception is a common surgical emergency in infants and young children. It consists of a telescoping of a segment of bowel (intussusceptum) into a more distal segment (intussuscipiens). The majority of children are under 1 year of age, with a peak incidence between 5 and 9 months of age. Ileocolic intussusceptions are the most common type. Ileoileocolic, ileo-ileal, and colocolic are much less common. Most (over 90%) have no lead point and are due to lymphoid hypertrophy, usually following a viral infection. Pathological lead points (which include Meckel’s diverticulum, intestinal polyp, duplication cyst, and lymphoma) occur in 5–10% of patients. The clinical presentation is characterized by episodes of intermittent colicky pain. Other common findings include vomiting, ‘red currant jelly’ stools (containing blood and

1499

1500

SECTION 8

• PAEDIATRIC IMAGING

Table 65.6 CAUSES OF ABDOMINAL PAIN IN INFANTS AND CHILDREN Constipation Acute appendicitis Gastroenteritis Intussusception Bowel obstruction Mesenteric adenitis Trauma Perforated Meckel’s diverticulum Cholelithiasis (biliary colic/acute cholecystitis) Henoch–Schönlein purpura Inflammatory bowel disease Acute or chronic pancreatitis Ingested foreign body Peptic ulcer

mucus), and a palpable abdominal mass. Later drowsiness and lethargy develop. A supine abdominal radiograph may detect the intussusception and its complications, or suggest alternative diagnoses (Fig. 65.14). Findings include the presence of a soft tissue mass, distal small bowel obstruction, absence of gas from the colon, difficulty in assessing the position of the caecum, or pneumoperitoneum. Abnormal radiolucencies in the soft tissue mass

are due to mesenteric fat trapped in the intussusception. The abdominal radiograph may be normal in over 50% of cases of intussusception. Ultrasonography is an accurate method for the diagnosis of intussusception, the assessment of bowel viability, and reducibility81. It has also been used to monitor hydrostatic reduction and to avoid the need for a diagnostic enema. A high-frequency linear array transducer should be used. Spontaneous reduction of intussusception has also been documented by US, which shows intussusception as a mass of 3–5 cm in diameter, with a ‘target’ appearance on transverse images creating a ‘sandwich’ longitudinal appearance. Similar findings may occur with bowel wall thickening due to oedema, inflammation, or haematoma (Table 65.7). The presence of a small amount of free intraperitoneal fluid is common. A ‘crescent in doughnut’ appearance on the transverse image represents the apex of the intussusception and its trailing mesentery, which causes a crescentric echogenicity (Fig. 65.15). Enlarged lymph nodes may be seen in the intussusception. Fluid can become trapped between the intussusceptum and intussuscipiens and is associated with vascular compromise82. Doppler interrogation can be used to assess blood flow in the intussusceptum, although nonvisualization of blood flow by colour Doppler ultrasound is not a contraindication to air enema reduction83. US may demonstrate lead points such as a duplication cyst, Meckel’s diverticulum, or lymphoma, which may be missed on fluoroscopy84. Following reduction by contrast medium enema, US is also useful to confirm complete reduction or to show a residual ileocolic or ileo-ileal intussusception in difficult cases. The post-reduction ‘doughnut’ sign may be seen after a successful reduction and represents oedema at the ileocaecal valve. Reduction of the intussusception should only be attempted after surgical consultation. Absolute contraindications to enema reduction include signs of bowel perforation with free intraperitoneal gas on the abdominal radiograph and the presence of septicaemic shock or peritonitis. Pneumatic reduction of intussusception using air under fluoroscopic guidance is the preferred method (Fig. 65.16)85,86. It has been shown to be safe, with a lower absorbed radiation dose compared with barium enema and has largely replaced hydrostatic reduction by barium enema. Pneumatic reduction of intussusception using carbon dioxide has also been described. Hydrostatic reduction with water-soluble contrast medium under US control has the advantage of avoiding radiation exposure and gives a more detailed view of the

Table 65.7 CAUSES OF BOWEL WALL THICKENING ON ULTRASOUND Intussusception Sepsis Typhlitis Trauma Henoch–Schönlein purpura Lymphoma

Figure 65.14 Intussusception. Supine radiograph of the abdomen shows the intussusception as a round soft tissue mass in the right upper quadrant outlined by a faint curvilinear rim of bowel gas. I = intussusceptum.

Fibrosing colonopathy of cystic fibrosis Meckel’s diverticulum

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

Figure 65.16 Intussusception. Air outlines the intussusception in the transverse colon during an air enema.

Acute appendicitis

Figure 65.15 Intussusception. Ultrasound showing (A) crescent in doughnut appearance and (B) fluid trapped between the intussusceptum and intussuscipiens, and some free intraperitoneal fluid.

intussusception87,88. Success rates of 85% or higher have been reported with air and hydrostatic reduction. Delayed attempts at reduction by air enema have been used successfully where the initial air enema has failed89. The main complication of the contrast medium enema is perforation. This complication rate can be kept low when the enema is performed by an experienced radiologist using a pump which can limit the pressure and has a safety release valve. Perforation occurring with an air enema is safer than with liquid enemas. The perforation tends to be smaller and is associated with less faecal spillage and less peritoneal contamination.

Acute appendicitis is the most common indication for emergency abdominal surgery in children. Most children with acute appendicitis who present with typical clinical findings do not require imaging. Approximately 30–40% have atypical findings. Accurate clinical diagnosis can be particularly difficult in very young children and infants. Imaging can help to diagnose or exclude acute appendicitis as a cause of their acute abdominal pain or suggest alternative diagnoses90. Plain radiographs of the abdomen are usually normal in nonperforated appendicitis or demonstrate non-specific findings. US is helpful if the diagnosis is equivocal. This is performed using a linear array high-frequency transducer. The swollen appendix is usually a tubular, fluid-filled structure, which is noncompressible and measures 6 mm or more in diameter (Fig. 65.17). The echogenic mucosal lining may be intact or poorly defined, indicating perforation is imminent. A faecolith can sometimes be present at its tip; this may be detected on an abdominal radiograph (15% of cases; Fig. 65.18) or ultrasonography. A peri-appendicular fluid collection occurs with early perforation. Increased peri-appendiceal echogenicity represents inflammation of the surrounding fat. Colour Doppler usually shows hyperaemia of the appendiceal wall but may be limited if the inflammation is confined to the appendiceal tip91,92. Enlarged reactive lymph nodes are commonly seen. A perforated appendix is indistinguishable on ultrasound from other forms of intraperitoneal abscess.

1501

1502

SECTION 8

• PAEDIATRIC IMAGING

for those cases where the clinical diagnosis is difficult96,97. In the postoperative period both US and CT demonstrate a phlegmon or abscess (Fig. 65.19).The CT appearances of acute appendicitis in children are similar to those in adults98.

Mesenteric lymphadenopathy Mesenteric lymphadenopathy can occur in association with intra-abdominal or pelvic infection, inflammation, or malignancy, and can mimic an acute abdomen. It is a common cause of abdominal pain in children. In most cases the cause is never found. Mesenteric lymphadenitis is a diagnosis made by exclusion. Enlarged lymph nodes in the root of the mesentery may be seen on US in children with acute and chronic recurrent abdominal pain99–102. US is useful in establishing a primary diagnosis in these children.

Henoch–Schönlein purpura (HSP)

Figure 65.17 Acute appendicitis. Ultrasound shows an enlarged appendix with a fluid-filled lumen.

Henoch–Schönlein purpura (HSP) is an idiopathic, systemic vasculitis of unknown aetiology, which is characterized by a purpuric skin rash, abdominal pain, arthralgia, and nephritis. GI involvement occurs in 50–75% of patients who have abdominal pain, vomiting, and bleeding. The vasculitis can involve the bowel wall leading to GI bleeding.The duodenum and upper jejunum are the most common sites of the GI tract to be involved. Plain radiographs of the abdomen may show bowel wall thickening due to haemorrhage and oedema, or small bowel obstruction or perforation. US features include bowel wall thickening, haematoma, ileus, peritoneal fluid, pneumatosis coli, gallbladder wall thickening, and intussusception (Fig. 65.20)103. The majority of intussusceptions in HSP are ileoileal, may be transient, and are not amenable to pneumatic or hydrostatic reduction. Contrast studies may show mucosal thickening, ‘thumb printing’, and separation of bowel loops. In most cases the changes of HSP are completely reversible and healing takes place in 3–4 weeks.

Figure 65.18 Appendicolith and small bowel dilatation in a child with acute appendicitis.

CT has a role in the diagnosis of acute appendicitis93–95 in those patients where US has been difficult because of bowel gas or a retrocaecal appendix. US and CT are both sensitive and specific for the diagnosis of acute appendicitis but they should be reserved

Figure 65.19 Acute appendicitis. CT showing an appendix abscess extending to the anterior abdominal wall. The abscess contains gas and an appendicolith.

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

constipated children have functional constipation, which is not associated with organic abnormalities or drugs. These children have normal ganglion cells; many have emotional problems. The diagnosis of constipation in children is essentially clinical, and imaging is not usually required. The plain abdominal radiograph will demonstrate the degree of faecal loading and dilatation of the large bowel; its routine use in the diagnosis of constipation is not recommended. The presence of faecal loading on the plain radiograph does not necessarily indicate constipation. US can help differentiate a faecal mass from a true mass. Other imaging techniques described for the evaluation of constipation include the measurement of colonic transit time using radio-opaque markers, fluoroscopy, and MRI defaecography, or a scoring system for the evaluation of faecal loading on the plain abdominal radiograph105–107.These are only indicated in a highly selected group of children.

Figure 65.20 Henoch–Schönlein purpura. Transverse ultrasound shows the hypo-echoic thickened bowel wall and echogenic areas in the mucosa representing pneumatosis coli.

Constipation Constipation is a common problem in infants and children and has been defined clinizcally as an alteration in the frequency, consistency, or ease of passage of stool104.This can lead to encopresis or faecal soiling, and occasionally can cause acute, severe abdominal pain. There is also an association with bedwetting, daytime urinary incontinence, and recurrent urinary tract infection. Causes of constipation are listed in Table 65.8. Chronic constipation often follows an inadequately managed acute episode. The majority of Table 65.8

CAUSES OF CONSTIPATION

Functional Neurogenic

Intestinal motility disorders Intestinal motility disorder is a term used to describe a variety of abnormalities that have in common reduced motility of the bowel and no organic occlusion of the bowel lumen. They can be divided into acute and chronic disorders (Table 65.9)105. Acute dysmotility includes paralytic ileus in which there is temporary cessation of peristalsis in the gut. This simulates intestinal obstruction, as there is failure of propagation of intestinal contents. Acute gastro-enteritis can simulate small bowel obstruction by causing a local paralytic ileus, with dilatation of the affected segment of bowel and multiple fluid levels on an erect plain radiograph of the abdomen. Chronic motility disorders include primary abnormalities of the bowel—Hirschsprung’s disease (aganglionosis), hypoganglionosis which is rare and mimics Hirschsprung’s disease, and neuronal intestinal dysplasia, which is a defect of autonomic neurogenesis characterized by an absent or rudimentary sympathetic ganglion innervation of the gut or by hyperplasia of cholinergic nerve fibres and hyperplasia of

Aganglionosis Hypoganglionosis Neuronal intestinal dysplasia Chronic intestinal pseudo-obstruction Disorders of the spinal cord Cerebral palsy Dietary Anal fissures Strictures Endocrine disorders Metabolic disorders

Table 65.9

BOWEL MOTILITY DISORDERS

Acute Paralytic ileus Systemic illness including shock, hypoxia, sepsis Chemical/hormonal upset Functional immaturity of bowel in neonate Meconium plug syndrome Inspissated milk syndrome Small left colon syndrome

Cystic fibrosis

Chronic

Tumours

Hirschprung’s disease

Lymphoma of bowel

Hypoganglionosis

Pelvic rhabdomyosarcoma

Neuronal intestinal dysplasia

Presacral teratoma

Chronic intestinal pseudo-obstruction

Drugs

Nonfunctional

Opiates

Functional

Anticholinergics

Megacystis–microcolon–intestinal hypoperistalsis syndrome

1503

1504

SECTION 8

• PAEDIATRIC IMAGING

neuronal bodies in intramural nerve plexuses108.The majority of cases of neuronal dysplasia present with severe constipation but tend to spontaneously recover colonic motility between 6 and 12 months of age. Plain radiographs of the abdomen may show signs of obstruction of bowel and the barium enema may show dilatation of bowel. Diagnosis is by biopsy.

Table 65.10

Chronic intestinal pseudo-obstruction (CIP) is rare and represents a spectrum of diseases that have in common clinical manifestations consisting of recurrent symptoms mimicking bowel obstruction over weeks or years. The age of presentation varies from newborn to adult. The condition is due to a visceral neuropathy or myopathy, which can be familial or nonfamilial, resulting in a lack of coordinated intestinal motility. Megacystis– microcolon–intestinal hypoperistalsis syndrome is the most severe form of CIP and is usually fatal in the first year of life. Immune deficiencies have also been reported to be linked to CIP. Plain radiographs of the bowel will show dilated loops of bowel with air–fluid levels. A contrast medium enema can exclude mechanical obstruction in children with acute symptoms109.

Bowel wall enhancement

CT SIGNS OF BOWEL INJURY

Free intraperitoneal fluid Bowel wall thickening Mesenteric haematoma Extraluminal air Bowel dilatation Extravasation of oral contrast

Blunt abdominal trauma in children most often results in injury to the liver, spleen, and kidney, and management is conservative in those who are haemodynamically stable. The mechanism of injury is usually a road traffic accident, crush injury, or fall. Surgery is indicated in those who are haemodynamically unstable or have penetrating injuries or bowel rupture. The decision to operate on a child is clinical. Imaging will assist decisions about the intensity of conservative treatment or detect subtle findings in bowel injury and expedite surgical management110. Plain radiographs should include a supine abdominal radiograph and if possible an erect chest radiograph or decubitus abdominal radiograph, with the right side elevated if perforation is suspected. Fractures of the spine, pelvis, and ribs may be identified. CT is indicated in children with severe blunt abdominal trauma (Fig. 65.21, Table 65.10). US is less sensitive than CT

for detection of bowel and visceral injuries, but it has a role where CT is unavailable or may be delayed111,112. The radiological appearances in children are similar to those in adults.The detection of bowel and mesenteric injury can be more difficult. Prompt recognition is important, as surgical intervention may be required.The mechanism of injury is important. A history of bicycle handlebar injury, lap belt injuries, or suspected nonaccidental injury should arouse suspicion of bowel injury. Lap belt injuries are also often accompanied by injury to the lumbar spine due to hyperflexion, resulting in transverse fractures of L1 and L2 or a Chance fracture. The hypoperfusion complex or shock bowel is due to poorly compensated hypovolaemic shock, which results in dilated, fluid-filled loops of bowel, and is a more frequent finding in children than adults. On CT there is intense contrast enhancement of the bowel wall, which may be thickened. The major abdominal blood vessels and kidneys also show intense enhancement, and the calibre of the aorta and inferior vena cava are reduced. Enhancement of the spleen and pancreas is decreased due to splanchnic vasoconstriction (Fig. 65.22). Hepatic periportal low attenuation zones on CT are common in children who have hepatic injury, but also can be seen in children with nonhepatic visceral injury or in the absence of intraabdominal injury113,114. There is high mortality associated with hypoperfusion complex. This abnormality (periportal tracking) is associated with physiological instability and a higher mortality

Figure 65.21 Jejunal perforation. Axial CT in a 13 year old boy following a lap belt injury to the abdomen shows subcutaneous haemorrhage, haemoperitoneum, free gas, dilatation of the small bowel, and bowel wall thickening.

Figure 65.22 Hypoperfusion complex. Axial CT showing periportal low attenuation, small IVC, diminished and patchy enhancement of the spleen, intense enhancement of the adrenal glands and kidneys, and a haemoperitoneum.

Abdominal trauma

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

rate. It is thought to be due to distension of the periportal lymphatics following vigorous hydration of haemodynamically unstable patients. It is not an indication for surgery.

Nonaccidental injury (NAI) Clinical presentation of NAI to the abdomen may be delayed for several days after the trauma has occurred, and is usually with abdominal pain, vomiting, signs of peritonism or obstruction. Bowel injury is more common in NAI than in accidental trauma. The duodenum and proximal jejunum are the most commonly injured parts of the bowel. Perforation may be diagnosed on the plain radiograph. US may demonstrate a duodenal haematoma and more proximal bowel dilatation. In the acute phase, upper GI contrast medium studies will demonstrate the intramural mass, with thickening of the mucosal folds giving a ‘coiled spring’ appearance. Perforation of bowel, mesenteric or visceral injuries, or hypoperfusion syndrome may all be found in association with NAI and are best demonstrated by CT115.

Nonbilious vomiting Hypertropic pyloric stenosis (HPS) The aetiology of HPS is unknown. It presents with projectile vomiting. Associated clinical findings include weight loss, dehydration, and hypochloraemic alkalosis. The age of presentation is typically between 4 and 6 weeks.There is often a family history. It tends to occur in first born male infants (male-to-female ratio 3:1). No imaging is required if an experienced clinician can reliably palpate the hypertrophied pylorus. US has proven to be an accurate method for the diagnosis of HPS and has replaced the barium meal in vomiting infants (Fig. 65.23, Table 65.11)116. Pylorospasm may mimic HPS on US. The measurements vary in pylorospasm with time, which can help distinguish it from HPS117. Occasionally, if US fails to confirm the diagnosis of HPS or if GOR is the likely diagnosis, a barium meal is performed. The pyloric canal is elongated (‘string’ sign) and indents the distal antrum to produce the ‘tit’ deformity (Fig. 65.24). Gastric emptying is delayed and GOR is usually present due to gastric outlet obstruction. Treatment is by pyloromyotomy.

Gastro-oesophageal reflux and hiatal hernia GOR is common in neonates and young children. In the majority of cases the reflux is physiological and is due to a poorly developed lower oesophageal sphincter, with a short intra-abdominal oesophageal course. As the child grows, the lower oesophageal sphincter matures and the GOR tends spontaneously to cease. Pathological reflux occurs when reflux oesophagitis or respiratory symptoms develop. The most frequent presenting symptom of GOR is nonbilious vomiting. Other documented symptoms include failure to thrive and rectal bleeding, and in older children heartburn and dysphagia. Respiratory consequences of GOR include aspiration and pneumonia, exacerbation of reactive airways disease, and laryngospasm. GOR has also been implicated in apnoea and the sudden infant death syndrome118,119. Children with underlying chronic respiratory disease, neurological

Figure 65.23 Hypertropic pyloric stenosis. (A) Longitudinal ultrasound shows the thickened hypo-echoic muscle. The margins of the pyloric canal appear as parallel, curvilinear echogenic lines. (B) Transverse ultrasound of the pylorus with central echogenic mucosa and surrounding hypo-echoic muscle.

impairment or a history of oesophageal atresia repair have a higher incidence of reflux. There is no agreed perfect method for detecting pathological GOR, but the current ‘gold standard’ is 24-h pH probe monitoring. An upper GI series is relatively insensitive for the Table 65.11 ULTRASOUND FEATURES OF HYPERTROPIC PYLORIC STENOSIS Canal length > 16 mm Transverse pyloric diameter > 11 mm Muscle wall thickness > 2.5 mm Pyloric canal does not open Gastric peristalsis is usually increased Poor gastric emptying/large gastric residue

1505

1506

SECTION 8

• PAEDIATRIC IMAGING

Figure 65.24 Hypertropic pyloric stenosis. Barium meal showing the narrow pyloric canal with a double track of barium. The hypertrophied pylorus indents the base of the duodenal cap.

detection of GOR, given the brief duration of the examination. Contrast medium studies are valuable to detect anatomical abnormalities which may be causing the child’s symptoms, such as a hiatal hernia or an antral web, or occur consequent to the symptoms, such as a peptic oesophageal stricture. Hiatal hernias are uncommon in childhood but may be large and associated with severe GOR. A hiatal hernia is sometimes visualized as a retrocardiac mass on the frontal chest radiograph, often with an air–fluid level present. The majority of childhood hiatal hernias are of the sliding type, with the gastrooesophageal junction seen above the level of the oesophageal diaphragmatic hiatus. The para-oesophageal (‘rolling’) type of hiatal hernia is rare in childhood. The stomach is displaced above the diaphragm and lies next to a normally positioned oesophagus. Gastro-oesophageal scintigraphy, capturing images over the entire postprandial period, is more sensitive in the detection of GOR. Scintigraphy is performed using 99mTc-labelled sulphur colloid, which is mixed with formula feed or breast milk and given to the child orally.The number of episodes of GOR and the duration of each episode is documented. US has also been used to detect reflux and to monitor gastric emptying120,121, but is time consuming. Surgery may be required in children who have persistent GOR and symptoms despite optimal medical therapy. The standard operation is the Nissen fundoplication; postoperative fluoroscopy can demonstrate a malpositioned fundoplication, loosening or breakdown of the wrap, oesophageal obstruction and perforation, and gastric volvulus122.

can occur anywhere along the length of the gut but are most frequent in the ileum where they lie along the mesenteric border and share a common muscle wall blood supply. They have a mucosal lining and 43% contain ectopic gastric mucosa. The majority of duplication cysts do not communicate with the GI tract. Duplication cysts may be diagnosed antenatally. Clinically they usually present in the first year of life with vomiting or abdominal pain. Less often presentation is with an asymptomatic palpable abdominal mass or melaena. Infection or haemorrhage into the cyst can cause it to enlarge and suddenly cause pain. A duplication cyst may act as a lead point of an intussusception. An oesophageal duplication can cause stridor. Plain radiographs may show a mediastinal soft tissue mass in the case of an oesophageal duplication cyst or associated vertebral anomalies. An abdominal radiograph may show displacement of bowel loops by a mass or rarely gas might be seen in the cyst if there is communication with the GI tract. US will demonstrate the cyst, which is usually spherical in shape and less often tubular. It will have an inner echogenic mucosal layer and an outer hypo-echoic muscular layer (Fig. 65.25)123. Contrast medium studies may show displaced or obstructed loops of bowel. 99mTc-pertechnetate is taken up by ectopic gastric mucosa and is helpful in diagnosing duplication cysts which present with gastrointestinal bleeding124.

Mesenteric cyst (lymphangioma) Intra-abdominal lymphangiomas are relatively rare and may be found in the mesentery, omentum, or retroperitoneum; the most common site is in the ileal mesentery.The cysts are true congenital abnormalities and arise due to sequestration of the mesenteric lymphatic vessels125.The cysts are lined by endothelial cells. Children usually present in the first decade, with increasing abdominal girth or a palpable abdominal mass. A more acute presentation can occur, with vomiting and abdominal pain following haemorrhage into the cyst, infection, or torsion. The cysts can grow large enough to cause intestinal or ureteric compression.

Abdominal distension Enteric duplication cysts are uncommon congenital anomalies and are due to abnormal canalization of the GI tract. They

Figure 65.25 Duplication cyst of pylorus. Ultrasound showing echogenic mucosal layer and hypo-echoic outer muscular layer.

CHAPTER 65

Plain abdominal radiographs show a soft tissue mass which displaces adjacent bowel loops. Occasionally the cyst wall is calcified. US examination demonstrates a thin-walled, uni- or multi-locular cystic mass that may be adherent to the solid organs and bowel. The cyst wall consists of a single layer, which contrasts with the double layered wall seen with enteric duplication cysts. If the intracystic fluid is chylous, infected, or haemorrhagic, then echogenic debris will be present. CT and MRI more precisely define the anatomical margins of the cyst (Fig. 65.26).The thin septae are more difficult to visualize with CT than US, but may enhance following the administration of intravenous contrast material. The attenuation values of the cyst contents vary; an uncomplicated lesion will have attenuation values similar to those of water. Negative attenuation values may be recorded if the fluid is chylous, and haemorrhagic and infected cysts have increased attenuation values. The MRI characteristics will similarly vary, depending upon the cyst contents.

•

PAEDIATRIC ABDOMINAL IMAGING

Malabsorption Malabsorption is a non-specific finding occurring in many childhood diseases (Table 65.12). When clinically suspected, the diagnosis is made by jejunal biopsy. Only a few of these conditions are important to the paediatric radiologist. Structural abnormalities causing malabsorption include malrotation, blind loop syndrome, or short gut syndrome. Malabsorption secondary to short gut syndrome can occur in a neonate who has had a large length of small bowel surgically excised, e.g. for multiple small bowel atresia. Flocculation, fragmentation, and segmentation of barium are non-specific signs of malabsorption (Fig. 65.27).

Gastrointestinal bleeding GI bleeding in infancy and childhood can involve any part of the intestinal tract. The causes are varied (Table 65.13). The age of the child and clinical presentation may enable a diagnosis to be made or narrow the differential diagnosis and then appropriate imaging will allow a specific diagnosis to be made, e.g. intussusception, HSP, and NEC. In many other cases fibre-optic endoscopy and biopsy can identify the sources of the bleeding. Other techniques that may be required include 99m Tc-sulphur colloid or labelled red blood cell scintigraphy and occasionally angiography if there is continued or recurrent haemorrhage126.

Inflammatory bowel disease Inflammatory bowel disease (IBD) presents in childhood in approximately 25% of cases.The radiology is described in more detail in Chapter 33.The aetiology of IBD in children includes ulcerative colitis, Crohn’s disease, infective colitis, typhlitis, radiation enteritis, graft versus host disease, and Kawasaki syndrome. Often the cause is clinically obvious, e.g. neutropenic colitis. Investigation is similar to adults. US has a greater role in children as bowel imaging is easier127,128. Thickened bowel

Table 65.12

CAUSES OF MALABSORPTION

Cystic fibrosis Coeliac disease Cows’ milk protein sensitivity Disaccharidase deficiency Malrotation Previous gut resection Blind loop syndrome Eosinophilic gastroenteritis Graft versus host disease Immunoglobulin A deficiency syndrome Schwachman–Diamond syndrome Hereditary angioneurotic oedema Intestinal lymphangiectasia Tropical sprue

Figure 65.26 Mesenteric cyst. (A) Upper GI contrast medium study performed in a patient with bilious vomiting shows extrinsic duodenal compression and displacement by a central abdominal mass. (B) The mass is confirmed to be a septated cyst at ultrasound.

Microvillus inclusion disease Zollinger–Ellison syndrome Abetalipoproteinaemia

1507

1508

SECTION 8

• PAEDIATRIC IMAGING

Polyposis syndromes

Figure 65.27 Malabsorption. Small bowel series showing segmentation and flocculation of barium, and mild dilatation of the bowel in a child with coeliac disease.

Intestinal polyposis syndromes are rare in children. They are more commonly found in adult life and are described in Chapter 33.They include familial polyposis coli, Gardner syndrome, Turcot syndrome, juvenile polyposis coli, Peutz–Jegher’s syndrome, and generalized polyposis coli131,132. Most have an autosomal dominant mode of inheritance and many have a significant risk of malignancy. The colon is predominantly affected in polyposis syndromes, with the exception of Peutz– Jegher’s syndrome, which mainly affects the small bowel133. Juvenile polyposis coli is a rare condition in which multiple polyps occur in the colon. The polyps do not have malignant potential but there is an association with a family history of large bowel malignancy. Presentation is in the first decade of life, most often between 4 and 6 years of age. Rectal bleeding is the usual presenting complaint; anaemia, intussusception, and rectal prolapse can occur. The polyps are thought to be mucous retention cysts occurring secondary to inflammation. Double contrast medium GI studies are the gold standard radiological methods for diagnosis. US and MRI can demonstrate polyps and could be used for follow-up, but polyps less than 1.5 cm in diameter may be missed on MRI and clusters of polyps may be seen as a mass lesion132.

Gastrointestinal tumours is the typical finding, but is non-specific. Most children with IBD present with weight loss and diarrhoea. Crohn’s disease may present at any age, often with failure to thrive, abdominal pain, and chronic low grade fever, in addition to the usual symptoms. Endoscopy and barium studies will diagnose most cases of childhood Crohn’s disease. Further imaging with US and radionuclide studies with labelled white cells may be required if endoscopy and barium studies are negative and there is strong clinical suspicion, although CT can be unreliable in the proximal bowel129. CT is useful for demonstrating abdominal abscesses. Gadolinuim-enhanced MRI has also been described in the diagnosis of IBD, though it is not used routinely130.

Table 65.13 CAUSES OF GASTROINTESTINAL BLEEDING Necrotizing enterocolitis Intussusception Midgut volvulus Anal fissure Gastroenteritis Typhilitis Henoch–Schönlein purpura Meckel’s diverticulum Polyposis syndromes Inflammatory bowel disease Haemolytic uraemic syndrome Foreign body Peptic ulceration Vascular malformation

Gastrointestinal tumours are rare in children. Endoscopy may diagnose tumours in the upper GI tract and large bowel. Benign tumours may occur as part of a syndrome, e.g. polyposis syndromes. They may all present in childhood but are more common in adults. Presentation may be with gastrointestinal bleeding, anaemia, and abdominal pain. The neoplasm can act as a lead point for intussusception. Imaging is by double contrast medium barium enema, small bowel series, and/or CT. Lymphoma is the most common primary malignant intestinal tumour in childhood. Involvement of the bowel and mesentery is almost always caused by non-Hodgkin’s lymphoma and is rare in Hodgkin’s disease. It is usually found in the ileocaecal region and may present with abdominal pain, mass, bowel obstruction, weight loss, or intussusception. Bowel involvement with Hodgkin’s disease is usually caused by direct invasion from adjacent involved lymph nodes. US is useful for initial imaging. CT is required for staging; it can identify tumour in the bowel wall and may show additional disease in the abdominal lymph nodes, liver, spleen, or kidneys. Barium studies may show strictures, obstruction, mass lesions, or ulceration (Fig. 65.28)134. Positron emission tomography (PET)– CT has a role in staging and evaluating tumour response to therapy135,136.

Omphalomesenteric (vitelline) duct remnants The omphalomesenteric (vitelline) duct is a normal fetal structure that connects the midgut to the extra-embryonic yolk sac. It runs alongside the allantois (later the urachus/ median umbilical ligament) within the umbilical cord. The omphalomesenteric duct usually involutes in the mid first trimester. Its persistence can give rise to a variety of congenital malformations.

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

Table 65.14 COMPLICATIONS OF OMPHALOMESENTERIC DUCT REMNANTS Diverticulitis ± perforation Bleeding (ectopic gastric mucosa) Enterolith formation Nonhaemorrhagic inflammation and ulceration secondary to the presence of ectopic mucosa Gastric Pancreatic Duodenal Colonic Biliary Intussusception Fibrous bands Obstruction Volvulus Internal hernias Littre’s hernias (prolapse of the diverticulum into an inguinal hernia sac) Impacted foreign bodies Wall inflammation

Figure 65.28 Lymphoma of the large bowel in a 9 year old boy who presented with constipation. Barium enema shows narrow, irregular descending and sigmoid colons.

Perforation Ileal prolapse onto the anterior abdominal wall Omphalomesenteric fistula Umbilical sinus

The majority of symptomatic omphalomesenteric ducts occur in boys and patients most commonly present before the age of 2 years. The clinical symptoms and signs depend upon the underlying abnormality (Table 65.14). If the entire duct remains patent, then there is a discharge of faeces from the umbilicus or the ileum can prolapse onto the anterior abdominal wall. Injection of contrast medium into a discharging umbilical fistula will outline loops of ileum when the omphalomesenteric duct is wholly patent.

Meckel’s diverticulum The most common type of omphalomesenteric duct remnant is the Meckel’s diverticulum, which arises on the antimesenteric border of the ileum. The diverticulum is present in 2–4% of V population. The size of Meckel’s diverticula varies with those greater than 5 cm in length being considered ‘giant’. Most are located within 60 cm of the ileocaecal junction. All the layers of the intestine are contained within their walls and frequently islands of gastric and/or pancreatic mucosa are seen within them. The typical clinical presentation of a child with a Meckel’s diverticulum is with melaena or abdominal pain. Intussusception or a volvulus around a Meckel’s diverticulum presents with symptoms and signs of a small bowel obstruction. Peritonitis may follow perforation of a Meckel’s diverticulum. Meckel’s diverticula which haemorrhage contain ectopic gastric mucosa in 95% of cases137 and 99mTc-pertechnetate scintigraphy can be diagnostic. The sensitivity of this investigation is increased when the patient receives premedication with ranitidine138. Differential diagnoses of a positive result include ectopic mucosa in an enteric duplication cyst, haemangiomata and IBD.

Mucosal polyp, granulation tissue, or cyst at the umbilicus

ABDOMINAL MANIFESTATIONS OF CYSTIC FIBROSIS Cystic fibrosis is an inherited autosomal recessive disease. Pulmonary disease is the predominant cause of mortality. The GI complications of cystic fibrosis result from abnormally viscid secretions within hollow viscera and ducts of solid organs. Bowel obstruction due to meconium ileus or meconium plug syndrome may be present at birth (see above).

Distal intestinal obstruction syndrome Distal intestinal obstruction syndrome (DIOS) presents in 10– 15% of older children with cystic fibrosis as colicky abdominal pain, constipation, and a palpable mass in the right iliac fossa due to impaction of mucofaeculent material in the terminal ileum and right colon139. Abdominal distension, nausea, and vomiting are common. DIOS is potentially fatal and can mimic other causes of abdominal pain, e.g. acute appendicitis, intussusception, or subacute obstruction of small bowel due to stricture or adhesions. Radiographs of the abdomen will show faecal loading of the colon, a bubbly appearance or mass in the right side of the abdomen, and dilated loops of small bowel (Fig. 65.29). Treatment is with oral Gastrografin and, if required, a Gastrografin enema to soften and mobilize the stool.

1509

1510

SECTION 8

• PAEDIATRIC IMAGING

haustration, shortening of the colon, narrowing of the colonic lumen, and nodular thickening of the colonic wall. Treatment is by surgical resection of the affected segment. Histological findings include submucosal fibrosis and fatty infiltration.

Other bowel manifestations The duodenum and small bowel are frequently abnormal in older children with cystic fibrosis. Thickened nodular mucosal folds can be demonstrated on the upper gastrointestinal series. Peptic ulcer, GOR, and associated complications of oesophagitis and oesophageal stricture are well recognized in cystic fibrosis. The US criteria for acute appendicitis are not reliable in children with cystic fibrosis as the appendix in asymptomatic children is often larger than normal144.

Pancreatic abnormalities Pancreatic insufficiency occurs in 80–85% of children with cystic fibrosis and manifests as malabsorption, chiefly of fat and protein. Approximately 30–50% of patients have glucose intolerance and 1–2% require insulin therapy140. Plain radiography of the abdomen may show punctate calcification in the distribution of the pancreas. US may show a small echogenic pancreas. The CT and MRI appearance depends on the amount of fatty replacement and degree of pancreatic fibrosis (Fig. 65.30). Figure 65.29 Distal intestinal obstruction syndrome. Supine abdominal radiograph shows faecal loading, with a bubbly appearance in the right side of the abdomen and dilated loops of small bowel.

Intussusception Intussusception occurs as a complication of cystic fibrosis in 1% of patients. The average age of presentation is 9–12 years, much older than in idiopathic intussusception. It is thought that the thick, putty-like faeces act as a lead point. The intussusceptions are usually ileocolic, although transient ileo-ileal intussusception can also occur. Common presenting symptoms include colicky abdominal pain and vomiting. It can present in a similar way to DIOS and delay the diagnosis. The passage of bloody stool is less common. Clinical examination can be normal. The imaging findings of intussusception in cystic fibrosis are the same as those seen in idiopathic intussusception.

Hepatobiliary disease In children with cystic fibrosis, cirrhosis of the liver may result from impaired bile drainage due to inspissated bile or fibrosis, and the prevalence increases with age. US abnormalities of the liver include a heterogeneous echo pattern, parenchymal nodularity, and periportal increased echogenicity due to fibrosis. The gallbladder may be small and the wall thickened due to chronic cholecystitis. Gallstones are found in up to 10% of children with cystic fibrosis. US findings in portal hypertension include splenic enlargement and gastric varices at the porta hepatitis. Clinical signs of cirrhosis in children

Fibrosing colonopathy Colonic stricture is a recognized complication in children and can lead to intestinal obstruction140. It is due to irreversible and sometimes progressive narrowing of the bowel lumen. High strength pancreatic enzyme supplements have been implicated in the aetiology. The right side of the colon is most frequently affected. Presentation is with signs and symptoms of distal intestinal obstruction. Plain radiographs occasionally can be helpful by demonstrating thickening of the colon wall. MRI can also be used to evaluate patients with fibrosing colonopathy141. US has been shown to be useful in assessing the thickness of the colonic wall142,143. Contrast medium enema findings include loss of

Figure 65.30 Axial CT of the pancreas in child with cystic fibrosis, showing fatty replacement of the pancreas.

CHAPTER 65

may occur late, by which time the disease is advanced. US can detect liver disease before it manifests itself clinically and biochemically145–147.

Intra-abdominal malignancy There have been several reports of GI malignancy occurring in patients with cystic fibrosis, the frequency increasing as the survival of these children improves. The lesions include cancers of the oesophagus, stomach, small bowel, colon, liver, biliary tract, pancreas, and rectum148.

THE IMMUNOCOMPROMISED CHILD Gastrointestinal manifestations of acquired immune deficiency syndrome Opportunistic infections account for most of the GI tract manifestations of acquired immune deficiency syndrome (AIDS) in children149. Primary lymphoma and Kaposi’s sarcoma occur in the adult GI tract but are rare in human immunodeficiency virus (HIV) infected children. The most common symptoms include acute or chronic diarrhoea, failure to thrive, oesophagitis due to candida, and less often cytomegalovirus (CMV) or herpes simplex virus. Imaging with double-contrast medium upper GI studies will accurately diagnose oesophageal lesions of candidiasis. Persistent chronic or recurrent diarrhoea can be caused by a number of organisms, including cryptosporidium, giardia, and Mycobacterium avium–intracellulare complex. Imaging is not usually required. Small bowel obstruction secondary to intussusception can occur in AIDS; a lead point such as lymphoma should be sought. Colitis in children with AIDS is also

•

PAEDIATRIC ABDOMINAL IMAGING

common and is caused by a number of organisms including shigella, salmonella, abdominal tuberculosis, and CMV. Presentation is usually with crampy abdominal pain and fever. Plain abdominal radiographs may show colonic dilatation, oedema, and thumbprinting of the bowel wall, and sometimes pneumatosis intestinalis. Abdominal lymphadenopathy is common in paediatric AIDS. It can be caused by infection, idiopathic lymph node enlargement where no cause is found, or Kaposi’s sarcoma and lymphoma150. US is useful in the evaluation of the abdomen for lymphadenopathy. CT has the advantage that other intra-abdominal organs, the bowel, and mesentery can also be assessed. Section 3 contains further descriptions of the GI tract manifestations of AIDS.

Typhlitis Typhlitis or neutropenic colitis is a complication of treatment of childhood malignancy or bone marrow transplantation. Presentation is with severe abdominal pain, often in the right iliac fossa, diarrhoea and fever. The caecum and ascending colon are most often involved. Bacterial and fungal infection of the bowel wall leads to inflammation and oedema, and occasionally haemorrhage and necrosis. The findings on the plain abdominal radiograph are nonspecific and include thickening of the colonic wall causing thumbprinting, small bowel dilatation, a mass in the right iliac fossa, and occasionally pneumatosis coli. US and CT will show colonic wall thickening, ascites, pericaecal fluid, and an inflammatory mass if present. In children, US is usually preferred because of its portability and lack of ionizing radiation (Fig. 65.31)151–153.

Figure 65.31 Typhlitis. (A) Ultrasound showing thickened bowel wall. (B) Axial CT showing thickening of the wall of the ascending colon.

1511

1512

SECTION 8

• PAEDIATRIC IMAGING

MISCELLANEOUS GASTROINTESTINAL PROBLEMS

hazardous procedure and oesophagoscopy is considered the safest technique for removal of oesophageal foreign bodies160.

Foreign body ingestion

Caustic ingestion

Ingestion of foreign bodies is common in children. The majority will pass uneventfully through the GI tract. The most common site of hold up in the oesophagus is at the level of the cricopharyngeus (80%). Oesophageal foreign bodies may rarely become embedded in the oesophageal wall causing oedema and compression of the trachea. Perforation of the oesophagus is rare. There is usually a history of choking, coughing, excessive salivation, dysphagia, or vomiting. Some children with impacted coins are asymptomatic. Infants may present with stridor and airway obstruction. A lateral radiograph of the soft tissues of the neck and a frontal radiograph of the chest and abdomen will readily demonstrate the majority of opaque foreign bodies (Fig. 65.32)154. If the type of foreign body ingested is known it can be helpful to X-ray the object to assess its density. A barium swallow will be required to make the diagnosis of an impacted nonopaque foreign body155,156. Batteries are important to recognize because if not removed they can corrode and leak causing caustic burns in the stomach157. Ingestion of multiple magnets can cause necrosis, ulceration, and even perforation of bowel when two magnets in separate loops of bowel are attracted together158,159. Removal of foreign bodies from the oesophagus is by oesophagoscopy. Extraction of foreign bodies using a Foley catheter under fluoroscopy has been described, but it is a potentially

The substances most often swallowed include household cleaning products, alkaline caustics, and acids. Caustic substances can cause burns to the mouth, ulceration, necrosis, and perforation of the oesophagus, and later stricture formation is common. Acid gastric secretions tend to neutralize the ingested alkalis and reduce damage to the stomach. In the acute phase in severe cases the chest radiograph may show a dilated air-filled atonic oesophagus. Oesophagoscopy will demonstrate the extent and severity of the oesophageal involvement and imaging is not usually required. If the endoscopic examination was limited because of the risk of oesophageal perforation, upper GI contrast medium studies using water-soluble contrast medium may be required to show the extent of the oesophagitis in the acute phase (Fig. 65.33)161,162. In the healing phase contrast medium studies will demonstrate oesophageal strictures. These may require serial dilatation. Carcinoma arising in caustic strictures can occur many years after the initial injury163.

Figure 65.32

Ingested batteries in the stomach.

Figure 65.33 Corrosive oesophagitis. Upper GI contrast medium study showing oesophageal ulceration and narrowing of the middle third, due to oedema.

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

THE PAEDIATRIC LIVER, BILIARY SYSTEM AND SPLEEN LIVER DISORDERS Imaging techniques Paediatric hepatic and splenic imaging employs many useful and complementary techniques164–167. US is the optimal technique for most paediatric and neonatal conditions of the liver and spleen. New high-resolution US equipment enables exquisite detail of liver anatomy, parenchymal information, as well as vascular/perfusion information with colour, pulse wave, and power Doppler imaging. Choice of probe is age and size dependent (vector 3.5–8 MHz; linear 7–10 MHz for parenchymal abnormalities). Other cross-sectional imaging techniques (CT, MRI) have many advantages for the display of anatomy and disease, but have inherent limitations of radiation, spatial resolution, lack of body fat in neonates, need for sedation and/or availability. Experience with MR venography (MRV), MR angiography (MRA), and MR cholangiography (MRC) in children is still growing167–170. Physiology, anatomy, and functional information is available with radionuclide imaging171–173. The role for PET/CT is still being established. Percutaneous transhepatic (transcholecystic) cholangiography (PTC or PTTC) maintains a role in therapeutic procedures. Since the development of the paediatric duodenoscope, endoscopic retrograde cholangiopancreatography (ERCP) plays a complementary diagnostic and therapeutic role even in the young child174. The diagnostic role of angiography is limited; however, it is essential in select therapeutic interventions. Percutaneous image guided procedures such as biopsy, drainages, biliary stenting, and embolizations are performed with accuracy, safety, and increasing frequency in children.

Diagnostic approach Newborn infants or children require evaluation for a wide variety of congenital and acquired abnormalities, anatomical and functional disorders, infections, and inherited inborn errors of metabolism. The various associations between congenital abnormalities and syndromes must be remembered, e.g. renal cystic disease with hepatic fibrosis, congenital cardiac lesions with left- and right-sided isomerism, and TORCH (toxoplasma, orphan virus, rubella, CMV, and herpes virus) infections with liver, spleen, and brain involvement175. Clear communication and multidisciplinary teamwork with the referring physicians are essential to evaluate appropriately the clinical question. Clinically relevant and timely reports must be provided. In interventional cases, communication regarding preprocedural correction of coagulation abnormalities, diagnostic and therapeutic aims of the procedure, and alternatives is imperative.

The newborn liver is less reflective on US than the echogenic newborn kidney. Commonly by 3–4 months of age this has reverted to the reverse adult pattern. The liver is normally right sided, but may be transverse and symmetric in the midline (bridging) as in polysplenia or asplenia, or left sided as in situs inversus. It is distinguished from the spleen in these situations by its internal portal and hepatic venous anatomy, echogenic portal vein walls, and the location of the gallbladder if present. Congenital agenesis or hypoplasia of a lobe is unusual, but is more common on the left than the right. Acquired atrophy may be due to left portal vein occlusion. The newborn extrahepatic biliary tree is difficult to see without new high-resolution US equipment. The proximal intrahepatic ducts should not exceed 2 mm in diameter. There is a normogram available for normal sized biliary structures according to age. The upper limit of normal for the common bile duct is 2 mm (< 1.27 mm) for the first year, 4 mm (< 3.3 mm) for older children, and less than 7 mm for adolescents/adults177. The normal gallbladder is at least 1.5 cm in neonates, reaching a maximum in children of 7.5 cm in length and 3.5 cm in diameter178. In the neonate, the umbilical vein runs in the falciform ligament from the umbilicus to the anterior surface of the liver, where it divides the medial and lateral segments of the left lobe; it intersects the left portal vein and runs from there via the ductus venosis into the inferior vena cava (IVC) (Fig. 65.34)179. The ductus venosis and umbilical vein are patent in the premature or early newborn (< 48 h) and in two-thirds of those up to 1 week of age; they obliterate within 2–3 weeks and become highly reflective and may calcify. Patency beyond this period or recanalization of a previously obliterated umbilical vein (+ collaterals) is associated with portal hypertension or abnormalities in the venous drainage of the liver. Variations in

Anatomy and normal variants At about 4 weeks of gestation, the caudal portion of the foregut develops as a ventral outgrowth into the liver and biliary tree. The liver develops from the larger cranial portion of the hepatic diverticulum, and the smaller caudal portion develops into the gallbladder and cystic duct176.

Figure 65.34 Liver. Venous anatomy in newborn. Ultrasound sagittal view through the liver showing an umbilical vein catheter in the umbilical vein (single white arrow) approaching the left portal vein (arrowhead). Blood flow is seen by colour Doppler in the ductus venosus (double arrows) as it enters the inferior vena cava (single black arrow).

1513

1514

SECTION 8

• PAEDIATRIC IMAGING

venous anatomy of the liver are common. Preduodenal portal vein (seen on US or CT uncommonly) may be associated with biliary atresia. The normal size of the spleen increases with age, the long axis approximately 6 cm at 3 months, 7 cm at 1 year, 8 cm at 2 years, 9 cm at 4 years, 10 cm at 8 years, and 12–13 cm at 15 years180. Anomalies of splenic size, number, and position are common. The spleen may vary in position, ‘a wandering spleen’. When the spleen has a long mesenteric pedicle it may tort and infarct. Accessory spleens occur in 10% of the normal population: 90% are solitary, 9% are double, and less than 3% are multiple (Fig. 65.35). Accessory spleens, remaining following

splenectomy for haematological problems, may hypertrophy and cause recurrent disease. Multiple functional peritoneal deposits of splenic tissue ‘splenosis’ may occur post traumatic rupture of the spleen or surgical splenectomy. Splenogonadal fusion is a rare variant occurring as a fibrous band or a tongue of ectopic splenic tissue (usually left sided and in boys). The normal spleen on CT has a higher attenuation than liver and enhances heterogeneously in the early phase of enhancement. On MRI, the spleen has lower signal intensity on T1weighted imaging than liver and slightly greater signal intensity than muscle. On T2-weighted images, the spleen has higher signal intensity than the liver. Liver and spleen abnormalities are associated with cardiac isomerisms (right isomerisms with asplenia, midline gallbladder, two right lobes of liver, GI malrotation; left isomerism with polysplenia, midline liver, interruption of the IVC with azygos continuation of the IVC and preduodenal portal vein, biliary atresia, and may be splenogonadal fusion)175. Once one anomaly has been identified, the radiologist should search for other anomalies (Fig. 65.36).

Jaundice

Figure 65.35 Accessory spleens. Patient with left atrial isomerism and numerous splenunculi seen on ultrasound adjacent to the left kidney. This patient also had a midline liver, interruption of the inferior vena cava with hemi-azygos continuation, as well as polysplenia.

Figure 65.36 Congenital abnormalities. (A) Neonate with multiple congenital abnormalities. Note the duodenal atresia with ‘double bubble’ sign, the dextrocardia, the nasogastric tube arrested at the T3 level suggestive of oesophageal atresia, and because of the gas in the stomach it therefore represents a tracheo-oesophageal fistula. (B) This child was found to have an absent portal vein with the superior mesenteric vein draining into the inferior vena cava as seen on the longitudinal ultrasound view.

Prolonged neonatal jaundice (> 10 d) is common. Causes of neonatal jaundice include physiological jaundice of prematurity, breast milk jaundice, ABO incompatibility, and other haemolytic jaundice, and sepsis of any cause. If the cause is clear, the infant does not require liver imaging. Metabolic causes, though uncommon, are an important group in the newborn period (galactosaemia, α1-antitypsin deficiency, cystic fibrosis, etc.). Clinical information regarding the gestational age of the infant, perinatal history, age of onset of jaundice, and the distinction between unconjugated and conjugated hyperbilirubinaemia is required. Unconjugated hyperbilirubinaemia is caused by prehepatic and hepatic forms of liver disease. Conjugated hyperbilirubinaemia (almost always pathological)

CHAPTER 65

can be divided into the extrahepatic obstructive forms—biliary atresia, Alagille syndrome, biliary hypoplasia, choledocal cyst, etc, and the hepatic forms—total parental nutrition (TPN), cholestasis, etc.181,182. Up to 80% of infants referred with persistent jaundice have biliary atresia or neonatal hepatitis. A multidisciplinary approach to the evaluation of neonatal jaundice up to and including liver biopsy is necessary. Jaundice in later life may result from a myriad of congenital, inherited, and acquired causes, which overlap with those seen in adults.

Biliary atresia In the infant with prolonged neonatal jaundice (defined as longer than 10 d), the diagnosis of biliary atresia is urgent, as outcome depends on early diagnosis and surgical establishment of biliary drainage before the onset of cirrhosis183. The distinction between biliary atresia and biliary hypoplasia and neonatal hepatitis can be difficult as imaging features may overlap181. The role for MRI and PTC/PTTC is still evolving. In biliary atresia, the infant presents in the neonatal period with persisting (slightly fluctuating) jaundice, pale stools, and hepatomegaly. It is slightly more common in girls. The aetiology is uncertain (perinatal/embryonic; environmental, infectious, immune and genetic aetiologies have been suggested) but the result is a progressive obliterative inflammatory process, affecting the extrahepatic biliary tree and progressing centrally towards the intrahepatic interlobar ducts184–186. It is classified into different types (Fig. 65.37). On US, the liver becomes large and coarse, with increased periportal reflectivity. Unlike obstructive jaundice in the older child or adult, the biliary tree in biliary atresia does not usually distend or dilate, due to the obliterative inflammatory process. The gallbladder is usually absent or rudimentary; it is visible but small in 20%. The presence of a normal-sized gallbladder, which distends with fasting and contracts with feeding, suggests a diagnosis other than biliary atresia187. The presence of the ‘triangular cord sign’ and absent or small gallbladder are high predictors of biliary atresia. The triangular cord is a highly reflective focus at the hilum of the liver cranial to the portal vein, likely representing the obliterated fibrosed biliary tree. Associated abnormalities occur in 10%: preduodenal portal vein, choledochal cyst, absent IVC, polysplenia, asplenia, trisomy 13, and situs inversus. To distinguish biliary atresia from severe cholestasis, a 99mTc-DISIDA scintigram should be performed in

I 15%

Figure 65.37 Types of biliary atresia.

•

PAEDIATRIC ABDOMINAL IMAGING

conjunction with US. Preparation with phenobarbital (5 mg kg−1 d −1) for 5 d before data acquisition is optimal. 99m Tc-DISIDA is injected intravenously, extracted by the hepatocytes, secreted into bile canaliculi, and excreted into bowel. Sequential imaging during the first 60 min with delayed images at 2, 4, 6, and 24 h is performed. In biliary atresia, the extraction is often normal with hepatic activity seen by 5 min. Failure to show radiopharmaceutical excretion at 24 h, despite good parenchymal extraction, is suggestive of biliary atresia (Fig. 65.38). Radiopharmaceutical within the bowel by 24 h excludes the diagnosis of biliary atresia. The differential diagnosis includes severe hepatocellular dysfunction. Percutaneous liver biopsy is required for definitive diagnosis showing bile duct proliferation, periportal fibrosis, bile plugs, and cholestasis. Traditionally at laparotomy a cholangiogram was performed through a needle placed in the gallbladder or its remnant. Preoperative PTC/PTTC and MRI may play a role in difficult cases. The treatment for biliary atresia is usually a porto-enterostomy (Kasai procedure), whereby a jejunal loop is brought as a Roux-en-Y up to the excavated porta hepatus to allow bile to drain through minute bile remnants or canaliculi into the bowel. The alternative is liver transplantation, which carries significant morbidity and long-term immunosuppression. Although the results of the Kasai procedure are suboptimal long term (at 10 years, > 50% failure, < 20% jaundice free), it remains the primary treatment in patients who present before 60 d188. In late stages of biliary atresia (even post successful Kasai), the liver becomes large and firm with a coarse parenchymal pattern; ultimately cirrhosis, portal hypertension ± varices, splenomegaly, and ascites develop. Persistence of the triangular cord sign on US post Kasai may be a poor prognosticator. Generally, the biliary tree does not dilate; however, epithelial lined bile cysts and lakes may be seen as irregular areas of low reflectivity on US. Ultimately 70–80% of patients with biliary atresia will require liver transplantation.

Biliary hypoplasia (Alagille syndrome) The clinical presentation of biliary hypoplasia is similar to that of biliary atresia but may present later and is of variable severity. It can be an isolated finding, i.e. ‘non-syndromic’ or ‘syndromic’ (previously known as arteriohepatic dysplasia/Alagille

II 10%

III 75%

1515

1516

SECTION 8

• PAEDIATRIC IMAGING

DISIDA) may fail to show excretion into the bowel in about 50% of biliary hypoplasias (Fig. 65.39C). Liver biopsy reveals a paucity in number of intralobular bile ducts per high power field. Cholangiography, intra-operative or by PTC/PTTC, shows patency of the thin spidery ducts (Fig. 65.39D). Management is conservative. A Kasai procedure is not indicated. Late presentation is associated with complications of the disease, e.g. cirrhosis and portal hypertension (PHT) or carcinoma, and may require treatment by liver transplantation. Figure 65.38 Biliary atresia. Radionuclide study of a 2 month old baby boy. 99mTc-DISIDA scintigram. Following preparation with phenobarbital for 5 d, this radionuclide scintigram using 99mTc-DISIDA shows good extraction of the tracer by the liver at 2 min, and no excretion by the biliary tree into the bowel by 6 h (A) or 24 h (B). Biopsy confirmed biliary atresia. Ultrasound showed no gallbladder present.

syndrome), which is associated with forehead bossing, pointed chin, posterior embryotoxin of the eye, butterfly vertebrae, renal anomalies (hypoplastic/dysplastic kidneys and cystic disease), and peripheral pulmonary branch stenosis (Fig. 65.39A). US shows a normal liver, a normal or small gallbladder, and no triangular cord sign (Fig. 65.39B). Radionuclide imaging (using

Choledochal cyst Choledochal cyst is a congenital segmental dilatation of bile ducts which may be detected antenatally. The classic triad (obstructive jaundice, pain, right upper quadrant mass) are the presenting features in one-third of children. Thirty per cent of patients present in the newborn period or infancy with a mass or obstructive jaundice, 50% present by 10 years, and 80% by young adulthood. Presentation with abdominal pain, pale stools, obstructive jaundice, and fever can mimic hepatitis unless a mass is found. The male-to-female ratio is 1:3. Different types of choledochal cyst occur (Type I–V) (Fig. 65.40). The aetiology is unclear; it may relate to the ‘common

Figure 65.39 Alagille syndrome. Biliary hypoplasia. (A) Plain AP radiograph of the spine in a neonate with Alagille syndrome showing numerous hemivertebrae (arrows). (B) Ultrasound of the liver in a 2 month old boy with Alagille syndrome showing a small gallbladder (arrows). (C) Radionuclide scintigraphy in another child shows good extraction of the 99mTc-DISIDA by the liver, and excretion of some tracer into the bowel at 24 h. Both infants had biopsy-proven biliary hypoplasia. (D) Second image of a 2 month old boy with prolonged neonatal jaundice. Pre-operative cholangiogram shows a diminutive or hypoplastic but patent biliary tree consistent with biliary hypoplasia (nonsyndromic). This was confirmed on biopsy.

A

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

Figure 65.40 Types of choledocal cyst.

I 80−90%

IV 10−15%

II 2%

V 1−6%

III 2−5%

channel’ theory (common insertion of pancreatic and common bile duct) (Fig. 65.41) or part of the spectrum of biliary atresia and Type V (Caroli’s), due to a developmental defect during fetal hepatogenesis184,189. On US the biliary tree shows a fusiform or saccular dilatation of varying size involving the common hepatic duct or common bile duct, with or without dilatation of the proximal intrahepatic portions of the bile ducts, with an abrupt transition to normal (Fig. 65.42). The peripheral ducts are not dilated. The gallbladder is usually normal but may be hypoplastic or duplicated. Distinguishing a small fusiform choledochal cyst (Type IV) from a dilated common bile duct due to other causes may be difficult. An abrupt transition between the dilated and nondilated portion of the bile duct favours a choledochal cyst (Fig. 65.41). Sludge or stones may develop within it. Biliary scintigrams (99mTc-DISIDA) may show accumulation of tracer in the choledochal cyst. MRC (fasting) may

satisfactorily demonstrate the bile and pancreatic duct anatomy170. ERCP shows the distal common connection, the cystic dilatation and the abrupt transition to nondilated ducts (Fig. 65.41), whilst PTC and PTTC outline the intrahepatic duct anatomy optimally174. Both, however, are invasive with risks of cholangitis and pancreatis (ERCP), bleeding, and bile leak (PTC + PTTC). Treatment is by surgical excision of the choledochal cyst and porto-enterostomy. Complications of choledochal cyst include ascending cholangitis, bleeding, cirrhosis, stone formation, portal hypertension, and malignancy (incidence increasing with age, up to 40% in adults). The differential diagnosis includes any right upper quadrant cystic mass/obstruction of the biliary tree. Caroli’s disease (Type V choledochal cyst) is rare, characterized by segmental nonobstructive dilatation or ectasia of the intrahepatic bile ducts, probably equivalent to Type V choledochal cyst. The aetiology is uncertain, may be developmental,

1517

1518

SECTION 8

• PAEDIATRIC IMAGING

possibly autosomal recessive, or part of a syndrome with congenital hepatic fibrosis and autosomal recessive polycystic kidney disease190. Patients present in late childhood or early adulthood with cholangitis. Biliary involvement may be diffuse or focal or involve only one lobe (usually left). US shows multiple low reflective cystic areas with a central dot of high reflectivity representing the portal vein branch surrounded by an ectatic bile duct, and sacular or beaded biliary ducts (Fig. 65.43). Heterogeneous areas of high reflectivity in these biliary dilatations may represent stone, sludge, or infection presenting as cholangitis or abscess. ERCP and MRI demonstrate the saccular biliary tree. Cholangiocarcinoma can develop late.

Figure 65.41 ERCP in choledochal cyst. Radiograph of an ERCP examination showing a moderate sized fusiform choledochal cyst in this 16 year old boy. Contrast medium fills the pancreatic duct (arrowhead) and the common channel is seen (arrow).

Total parenteral nutrition cholestasis Although TPN may be life-saving, its use is associated with liver disease. Conjugated hyperbilirubinaemia may occur within 2 weeks of starting TPN, especially in low birth weight infants, infants after extensive bowel resection, and according to the composition of the TPN191. Imaging excludes other causes of jaundice. Ultrasound shows biliary sludge or enlargement of the gallbladder. Cholestasis may resolve if enteral feeding is started; failure to achieve enteral nutrition leads to progressive liver failure and portal hypertension, with some patients ultimately requiring liver transplantation181,182. Neonatal hepatitis The term neonatal hepatitis includes a

Figure 65.42 Choledochal cyst. Ultrasound through the upper abdomen in this 2½ year old girl showing a large choledochal cyst (6.3 cm × 9.4 cm) in longitudinal section and seen in transverse view anterior to the right kidney (arrow) and situated at the portal hepatis.

large group of disorders, the majority of whom are idiopathic. Definitive diagnoses include antenatal infections such as CMV, rubella, enteroviruses, toxoplasmosis, herpes simplex, and spirochaete, as well as metabolic disorders, e.g. cystic fibrosis, α1-antitrypsin deficiency, tyrosinaemia, and galactosaemia182. US features are non-specific and include hepatomegaly, heterogeneous coarse liver parenchyma, and a visible gallbladder (> 1.5 cm) without the triangular cord sign. Cranial ultrasound may identify structural abnormalities or calcification within the brain. Extraction of 99mTc-DISIDA into the liver is often reduced and excretion into the bowel may be reduced proportional to the degree of cholestasis and hepatocellular dysfunction (and excludes biliary atresia). In severe neonatal hepatitis, the cholestasis is so severe that the reduced extraction and excretion make it difficult to distinguish from biliary atresia173. MRC may help. Percutaneous needle biopsy of the

Figure 65.43 Caroli’s disease. (A) Transverse ultrasound image through the liver showing fluid-filled spaces with a central hyperreflective centre or dot (arrows), consistent with bile lakes of Caroli’s disease surrounding a portal vein branch. (B) CT of the same patient showing the low attenuation areas surrounding the portal vein branches (arrow). The whole liver is involved, the right lobe more than the left.

A

B

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

liver is frequently needed to reach a definite diagnosis. Five to 10% develop persistent fibrosis.

Spontaneous perforation of the bile ducts This is an uncommon disorder, which presents in the first three months of life with jaundice, ascites, and pale stools192. The cause is uncertain, and the perforation occurs at the junction of the cystic duct and common hepatic duct, or the cystic duct and gallbladder. Jaundice results from reabsorption of bile from the peritoneum. Liver function tests are usually normal. US shows ascites generalized or localized to the right upper quadrant which may contain debris. Radionuclide scintigrams using 99m Tc-DISIDA are very sensitive at identifying the presence of radionuclide in the peritoneum and occasionally the site of a leak. Aspiration confirms the diagnosis. Surgical or percutaneous drainage is required with spontaneous healing occurring in about 1 month. Haemolytic anaemia Haemolytic anaemia as a cause of jaundice is not confined to the neonatal period. US shows a normal liver, possibly pigment stones in the gallbladder, and splenomegaly.The common causes in infants are haemolytic anaemia of the newborn, Rhesus disease, and autoimmune haemolytic anaemia, and in the older child, the haemoglobinopathies and enzyme deficiencies.

Infectious disorders The majority of viral infections of the liver are diagnosed clinically with serology. Imaging is supportive (US ± CT ± confirmation with biopsy [targeted, percutaneous or transjugular]). In children bacterial liver abscesses occur from (A) bacterial seeding from portal venous drainage of an area of sepsis; (B) direct spread of sepsis from a contiguous focus; or (C) infected traumatic lacerations193,194. Chronic granulomatous disease of childhood (an X-linked recessive disorder with inability of leukocytes to lyse bacteria) may be complicated by liver abscess, as can other immune deficiency states195. In the neonatal period, liver abscesses may occur secondary to necrotizing entercolitis and from extravasation from umbilical vein lines (Fig. 65.44). Many are cryptogenic. The common organisms in liver abscesses include Klebsiella, Staphylococcus aureus, Escherichia coli, aerobic streptococci, and anaerobes. Percutaneous therapy (aspiration + drainage by US or CT, going through a seal of normal liver) has reduced morbidity and mortality from liver abscesses. The lesions may calcify when healed. Nonbacterial abscesses such as amoebiasis and hydatid echinococcus disease are endemic to certain areas and have specific imaging characteristics (see Ch. 35). In children the incidence of opportunistic infections in the liver has increased commensurate with transplantation, chemotherapy, immunosuppression, and HIV. Fungal infections most commonly present as microabscesses or target lesions, and must be differentiated from lymphoproliferative disease (Fig. 65.45). HIV infection in children most commonly causes hepatomegaly,

Figure 65.44 Liver abscess. Premature male infant (860 g) with sepsis secondary to an umbilical line. A 4 × 2 cm ovoid area high in the liver towards the diaphragm showing a rim of increased reflectivity (arrows) and central heterogeneous mixed reflections, and a thin surrounding low reflective rim. This was percutaneously aspirated and drained under ultrasound guidance and pus obtained (Staph. aureus). The child did well following drainage.

with increased or decreased reflectivity of hepatic parenchymal texture, splenomegaly, and focal hypo-echoic lesions representing opportunistic infection or granulomatous foci. Gallbladder wall thickening, gallstones, and sludge are also seen.

Liver tumours Tumours of the liver are uncommon in children, primary hepatic neoplasms accounting for just 0.5–2% of all paediatric

Figure 65.45 Liver graft. Lymphoproliferative disease. Axial CT images through the liver of a reduced liver graft showing several low attenuation small nodules (arrows) scattered throughout the parenchyma. Histology obtained by ultrasound-guided biopsy proved these to be posttransplant lymphoproliferative disease. Fungal infection could look similar.

1519

1520

SECTION 8

• PAEDIATRIC IMAGING

malignancies (solid/nonsolid). They usually present with an incidental lump or mass felt by a parent, or as an incidental finding on ultrasound performed for another reason. They may be benign (one-third) or malignant (two-thirds)196,197. The latter may be primary or secondary.Therapies currently involve combinations of surgery, chemotherapy, and transplant. Categorized by age, haemangiomas, hepatoblastoma, and mesenchymal hamartomas occur in the neonatal period. In the young child, hepatoblastoma, metastases, haemangiomas, and rhabdomyosarcoma occur. In the older child hepatocellular carcinoma, lymphoma, metastases, undifferentiated sarcomas, and adenomas are more common. The benign tumours are commonly vascular in origin, e.g. haemangioma. Although most are asymptomatic, they may present with cardiac failure or thrombocytopenia. Imaging characteristics of many tumours have been described using ultrasound, Doppler, CT, radionuclide radiology, and MRI164,167,197–199. Plain radiographs are usually non-specific but may show calcification ± soft tissue enlargement.

Malignant tumours Hepatoblastoma is the third most common abdominal tumour in infancy (second to Wilms’ tumour and neuroblastoma) with a peak incidence at under 2 years of age and a male preponderance (2:1). It is associated with Beckwith–Wiedemann syndrome (chromosome 11) and familial adenomatus polyposis (chromosome 5). Many are detected as asymptomatic masses; advanced tumours are associated with anorexia, weight loss, pallor, anaemia, and abdominal pain. Twenty per cent have metastases at presentation. Serum α-fetoprotein (AFP) levels are elevated in over three-quarters of cases. Calcification is seen in about 50% on abdominal radiography. Ultrasound shows a heterogeneous mass of mixed high and low reflectivity, which may have calcification, cystic areas of necrosis, and a pseudocapsule (Fig. 65.46). Size varies from small to very large. They may be single, multifocal, or exophytic. The tumour may show pressure on, splaying or infiltration of, hepatic or portal veins and IVC. Increased activity on the angiographic phase and a photopenic area on delayed 99mTc-sulphur colloid scintigraphy are seen. A mixed

Figure 65.46 Hepatoblastoma. Parasagittal view through the upper abdomen on the right through the right lobe of the liver showing a solid mass (arrow) of increased reflectivity when compared to the adjacent normal liver and adjacent posteriorly placed kidney (K). The mass shows compression on the inferior vena cava (thick arrow).

low attenuating lesion ± calcification with peripheral rim enhancement is seen on CT (Fig. 65.47). When large it is difficult to determine the organ of origin, e.g. arising from the liver, kidney, or adrenal. MRI (spin echo, MRA, MRV) assesses the extent of the tumour, its relation to the vascular and liver segmental anatomy, its blood supply, and its resectability. In general, the tumour has heterogeneous low signal on T1-weighted imaging (haemorrhage may be bright), and high signal on T2-weighted imaging with hypo-intense fibrous septae (Fig. 65.47). Angiography and venography are now largely obsolete. The diagnosis is confirmed by percutaneous needle biopsy taking numerous cores from the least necrotic areas (most solid on US) for histology, immunohistochemistry, and molecular studies. The differential diagnosis includes haemangioma, metastatic neuroblastoma, mesenchymal hamartoma, and hepatocellular carcinoma.

Figure 65.47 Hepatoblastoma. (A,B) A 17 month old boy with abdominal distension. A large, low attenuation, solid heterogeneous mass is seen on CT, without calcification, and with patchy enhancement. This mass was histologically shown to be a hepatoblastoma. (C) MRI T1-weighted image shows a low signal, discrete large mass with areas of increased signal consistent with blood. The portal structures are displaced to the right. Low signal septae are seen though the lesion.

CHAPTER 65

Hepatocellular carcinoma This uncommon paediatric tumour may present in the older child (usually older than 4 years) de novo or in the presence of underlying liver disease—previous adenoma, chronic hepatitis, biliary atresia, inborn errors of metabolism (e.g. α1-antitrypsin deficiency), etc. It is the second most common primary malignant hepatic tumour in childhood. Over half of the children will have elevated AFP. The imaging characteristics are detailed in Chapter 35. Other malignant tumours Rare hepatic malignant tumours include fibrolamellar hepatocellular carcinomas occurring in adolescents, a solitary lobulated mass with a central fibrous scar, described fully in the Chapter 35. Undifferentiated embryonal sarcomas occur in older children (6–10 years), and may be solid, cystic, or mixed. They present with pain and swelling. US demonstrates a large solid mass with hypo-echoic areas. CT shows low attenuation mass with septae and fibrous pseudocapsule. Hepatic angiosarcomas, also very rare, are highly malignant, present around 3–4 years of age predominantly in girls with systemic symptoms. US and CT show multiple heterogeneous foci; they enhance centripetally. Survival is poor. Rhadomyosarcoma, a mesenchymal tumour, rarely involves the biliary tree. It presents in children between 2 and 5 years of age with malaise, fever, pain, obstruction, jaundice, and ductal dilatation, with or without haemobilia when it involves a large bile duct.A focal irregularity or large lobulated mass of increased reflectivity arising in the region of the porta hepatis, gallbladder, or biliary tree is detected by US, CT, or MRI. When arising in more peripheral ducts, it may not be considered until a needle biopsy yields the correct diagnosis. The tumour is photopenic on 99mTc-sulphur colloid. Cholangiocarcinoma rarely occurs in children and is described in Chapter 36. Hepatic or splenic involvement with leukaemia and lymphoma is more commonly detected at autopsy than by imaging; it may present as diffuse hepatomegaly or as focal nodules or masses (usually hypo-attenuating and hypo-echoeic). In a child with a known tumour, the differential diagnosis of focal liver or spleen lesions includes metastases, focal infection, or pre-existing incidental focal lesions.The imaging characteristics are similar to those in adults. Wilms’ tumour and neuroblastoma are the most common tumours to metastasize to the liver in children; 60% of neu-

•

PAEDIATRIC ABDOMINAL IMAGING

roblastomas present with metastatic disease. Metastatic neuroblastoma to the liver may be seen as discrete focal lesions, simulating a primary liver tumour or as a diffuse infiltrative multifocal tumour, with areas of low reflectivity ± calcification, giving it an extremely coarse speckled appearance (Fig. 65.48).

Benign tumours Haemangiomas are the most common liver tumours of infancy. The classification of vascular anomalies by Mulliken et al divides lesions into vascular tumours and vascular malformations. Haemangiomas have abnormal parenchyma and fall into the former category. They form a heterogeneous group of benign mesenchymal tumours which present before 6 months of age. They may be focal, multifocal, or disseminated haemangiomatosis (three or more organs involved). Haemangiomas may show hypervascularity of the mass associated with (A) slow flow without shunting; (B) high flow without shunting; (C) arterial-to-venous shunting; (C) venous-to-venous shunting; or (E) mixed shunting200. Most are asymptomatic; others may be associated with high output failure, thrombocytopaenia, and hypothyroidism, and are difficult to treat. US shows single focal or multiple discrete solid hyperechoic masses.When multiple, it is difficult to identify normal liver parenchyma (Fig. 65.49). Vascular spaces with shunting in the lesions may be visible in addition to enlarged hepatic veins. Aortic narrowing below the coeliac artery in the high flow lesions is seen. On unenhanced CT the low attenuation lesions are seen—often of similar attenuation to blood in vessels; following intravenous enhancement, the periphery of the lesions takes up contrast medium first (Figs 65.50, 65.51) Red blood cell scintigraphy may be a useful technique in identifying haemangiomas. MRI shows lesions that are low signal on T1-weighted imaging and high on T2weighted imaging with or without flow voids near or inside the lesions201. Angiography only plays a role in those patients who have failed medical therapy (steroids, vincristine, interferon) and require therapy for their high output failure. Embolization is challenging; its role is to temporize until involution occurs naturally (after a year or so). Multiple approaches may be needed via the hepatic artery, branches of the superior Figure 65.48 Metastatic neuroblastoma. (A) A 3 month old boy with abdominal distension and anaemia. Sagittal ultrasound through the right lobe of the liver shows numerous solid foci of high reflectivity throughout the parenchyma. (B) Axial CT image through the liver (without intravenous contrast medium) shows low attenuating rounded lesions in the liver, becoming confluent in some areas. Biopsy confirmed metastatic neuroblastoma.

1521

1522

SECTION 8

• PAEDIATRIC IMAGING

Figure 65.49 Haemangioma. Midline sagittal view of the upper abdomen in a 6 month old girl in severe congestive cardiac failure. The liver shows numerous rounded lesions of low reflectivity scattered throughout the liver parenchyma. On longitudinal views, the aorta is seen to give off a large coeliac artery to the liver and the aorta below this is significantly smaller. Note the nodules of the liver tumour do not show vascularity on colour Doppler, except for one area adjacent to the coeliac artery which most likely represents volume averaging of a feeding vessel.

mesenteric artery, the portal vein, and hepatic veins. Overaggressive embolization risks hepatic infarction (Fig. 65.52). Mesenchymal hamartoma is rare but is said to be the second most common benign liver tumour or developmental lesion presenting in children, usually younger than 2 years of age (peak 15–22 months). Hamartomas are usually large (5–29 cm, mean 16 cm) and mixed solid cystic, containing a mixture of bile ducts and mesenchyme, multiseptated with a cystic/gelatinous composition202,203. When small they appear more solid (Fig. 65.53). On CT, they are of low attenuation with variable septation and fluid loculation. The tumour is hypovascular, but atrioventricular shunting can occur through enlarged irregular tortuous feeding vessels. MRI shows the multiseptated fluid areas (low on T1- and high on T2-weighted imaging) displacing major vessels. Excision of the tumour is usually therapeutic. Focal fatty infiltration, focal nodular hyperplasia (FNH; benign tumour of epithelial origin), and adenomas are seen far more commonly in adults than children; as a result they may pose a diagnostic dilemma when they do occur (Fig. 65.54). Adenomas occur in children with metabolic disorders (e.g. glycogen storage disorders) or adolescents on the contraceptive pill, and may undergo malignant change.

BILIARY DISEASE In children gallstones are being detected with increased frequency with the widespread application of US. Gallstones are

Figure 65.50 Haemangioma. A 3 month old boy with congestive cardiac failure. (A) Unenhanced CT image shows numerous low attenuation nodules throughout the liver parenchyma, with very little normal parenchyma. (B) Following intravenous contrast medium these nodules enhance intensely from the periphery. The aorta is seen to be of adequate size; below this and distal to the take off of the coeliac axis the aorta became significantly smaller in size (< 50%) (not shown).

Figure 65.51 Haemangioma. Three-phase axial CT through the liver shows the small, low attenuation area posterior to the right portal vein on the enhanced CT (A), which has peripheral to central filling following administration of contrast medium (B,C).

CHAPTER 65

•

PAEDIATRIC ABDOMINAL IMAGING

Figure 65.52 Haemangioma. A 3 month old boy with an asymptomatic mass in the right upper quadrant. (A) Axial CT image shows a large focal low attenuation mass in the right lobe with enhancing peripheral vessels. The mass is largely avascular centrally. (B) Selective coeliac axis angiogram into the hepatic artery shows the large hepatic artery feeding irregular tumour vessels with mass effect. Surgical excision proved this to be a cavernous haemangioma. The patient also had a small cutaneous haemangioma on his neck.

associated with TPN, diuretics, haemolytic anaemias, cystic fibrosis, metabolic disorders, inflammatory bowel disease, short gut syndrome, structural abnormalities of the biliary tree, prematurity, and dehydration. Many children are asymptomatic; the stones resolve or pass spontaneously so treatment is controversial204. Imaging appearances are similar to those in adults (see Ch. 36, The Biliary System). Calculous cholecystitis is rare in children; 50% of those with acute cholecystitis will have no stones evident. Biliary sludge is commonly seen. Kawasaki disease and systemic vasculitis may cause cholecystitis, but more commonly an acute hydrops of the gallbladder.The biliary tree in children may be involved in sclerosing cholangitits and in cystic fibrosis (see above)170.

VASCULAR DISEASES OF THE LIVER

Figure 65.53 Mesenchymal hamartoma. Ultrasound through the right lobe of the liver showing a round septated partly solid, partly cystic mass.

Vascular lesions of the liver are uncommon in children. They may be congenital or acquired—haemangiomas, vascular malformations, portal vein thrombosis, Budd–Chiari syndrome, etc.200,201,205,206.

Figure 65.54 Hepatic adenoma. A 16 year old girl with glycogen storage disease, type I. Axial CT images through the liver in this teenage girl show (A) a large heterogeneous adenoma in the right lobe and (B) a smaller adenoma in the lower aspect of the right lobe (B). This girl underwent successful liver transplantation, and histology of the explanted liver showed both to be adenomas.

1523

1524

SECTION 8

• PAEDIATRIC IMAGING

Portal hypertension Portal hypertension (PHT) is categorized into prehepatic, hepatic, and post-hepatic causes. Prehepatic (extrahepatic) PHT results from portal vein thrombosis and extrinsic occlusion of the portal vein; liver function is normal. Hepatic causes include cystic fibrosis, metabolic liver disease, and numerous causes of cirrhosis, including drug induced. Post-hepatic causes in children include IVC webs, Budd–Chiari syndrome, cardiac causes, and veno-occlusive disease. US (colour and pulsed Doppler) detecting flow and its direction may distinguish the three. MR/MRA/MRV are complementary investigations207.

Portal vein thrombosis Portal vein thrombosis used to be a common sequela of umbilical venous catheterization and umbilical sepsis. In the older child, trauma, pancreatitis, splenectomy, tumours, inflammatory masses, and dehydration are recognized as causes. Late sequelae of portal vein thrombosis include PHT, splenomegaly, ascites, and varices. With US, occlusion of the portal vein may be seen early in its course as a swollen vein containing highly reflective thrombus. Later a leash of collateral vessels is seen surrounding the thrombosed/obliterated/partly recanalized main portal vein, ‘cavernous transformation’. These collateral vessels permit ongoing hepatopetal flow as the liver parenchyma and the sinusoidal pressures are normal. When large, collaterals in cavernous transformation may have mass effect on the biliary tree, pancreatic duct, and the region of the pancreatic head. As PHT develops oesophageal, gastric, or

duodenal varices may be seen. Natural collaterals and shunting develop—splenorenal, umbilical, etc. On CT, the portal vein appears of low density, enhances poorly and is surrounded by peripheral enhancing collaterals. Focal intrahepatic portal vein thrombosis can also occur with thrombosis of the left portal vein leading to focal areas of infarction in the left lobe of the liver, which may become secondarily infected.

Hepatic causes Hepatic causes of PHT, arising from the myriad of causes of fibrosis or cirrhosis, lead to similar sequelae in terms of collaterals, varices (Fig. 65.55), and splenomegaly. Because liver synthetic function may be affected, the prognosis is poorer than with portal vein thrombosis. The imaging features of coarse liver parenchyma, mixed size nodularity, and shrunken liver are similar to the findings in adults.

Budd–Chiari syndrome Budd–Chiari syndrome is rare in children. It presents as an acute, subacute or chronic phenomenon, from a congenital or acquired aetiology, and is due to occlusion of the small centrilobular veins, major hepatic veins, or IVC208. Uncommon causes are a congenital IVC web or membrane, or interruption with agenesis. Acquired causes include thrombosis secondary to hypercoaguable states, vessel wall injury, right heart pathology, tumours, postoperatively or idiopathic. Hepatomegaly, splenomegaly, retrograde portal venous flow, ascites, and large venous collaterals result.

Figure 65.55 Portal hypertension with varices. A 2½ year old boy with portal hypertension secondary to cirrhosis from biliary hypoplasia. (A,B) Barium swallow shows serpiginous filling defects in the lower half of the oesophagus and in the fundus. These are consistent with oesophageal and gastric varices. (C) Ultrasound of the splenic hilum shows numerous rounded vessels of low reflectivity which filled with colour on Doppler interrogation, consistent with splenic varices (arrows).

CHAPTER 65

Veno-occlusive disease (VOD) is allied to classic Budd–Chiari but is an acquired phenomenon, seen in the immunosuppressed child secondary to chemotherapy or bone marrow transplantation. This obstruction occurs at the microscopic level of the central and sublobular veins. A definitive diagnosis requires biopsy,

•

PAEDIATRIC ABDOMINAL IMAGING

frequently via a transjugular route because of coagulopathy during which hepatic venography, free and wedged pressures are performed (Figs 65.56–65.58). Occasionally, hepatic venography confirms the presence of a web or obstruction to the IVC which may be managed by dilatation preferably without stenting.

Figure 65.56 Budd–Chiari syndrome. Ultrasound. (A) Ultrasound images of a teenage girl’s liver with a 2 week history of abdominal distension, on a background history of sagittal sinus thrombosis and lassitude. One hepatic vein is visible with dampened hepatic venous flow. The other hepatic veins are only faintly seen and show no flow on colour or pulse Doppler. (B) Image shows ascites and apparent thickening of the gallbladder wall secondary to the ascites.

Figure 65.57 Budd–Chiari syndrome. CT and MRI. (A,B) CT shows patchy enhancement of the liver parenchyma, absent inferior vena cava (IVC) at the level of the liver, with prominence of the azygos vein (A) (arrow); and more inferiorly enhancement of the caudate lobe with distension of the obstructed IVC (two small arrows) (B). (C,D) Compared with the previous patient, these sagittal (C) and coronal (D) MRI T1-weighted images show a patent IVC (arrow) but absence of any recognizable hepatic veins in this child with more chronic features of Budd–Chiari syndrome.

1525

1526

SECTION 8

• PAEDIATRIC IMAGING

B Figure 65.58 Venacavogram and hepatic venogram. (A) Cavogram in a 12 year old girl with jaundice and abdominal distension. There is compression of the inferior vena cava and its anterior and lateral dimensions, with collateral filling in a retrohepatic and inferohepatic direction and into the caudate lobe. (B) Attempts at cannulation from above, via the jugular vein, into the right or middle hepatic veins was not possible. Canulation into a small left hepatic vein branch shows extensive intraluminal filling defects consistent with clot (arrows) and abnormal staining pattern due to the high pressures, despite the catheter not being wedged. Faint filling of the portal vein is seen (arrowhead).

Interrupted inferior vena cava During embryology, if the right subcardinal vein does not connect appropriately with the liver, the blood from the lower part of the body flows up the azygos vein with interruption of the retrohepatic IVC. US recognition is important in children before future surgery. The IVC is seen to run in a straight posterior direction into the chest, with loss of the normal anterior curve as is usually seen behind the liver where it enters the right atrium in the normal fashion.This is referred to as azygos continuation of the IVC (Fig. 65.59).

phy, demonstrates gas as bright echogenic foci in the main portal vein extending out to the periphery of the liver, with audible high pitched sounds on Doppler (Fig. 65.60).

Congenital and acquired vascular malformations

Portal vein gas is not uncommon in children, occurring usually in the context of necrotizing entercolitis, infectious and inflammatory bowel disorders, uncommonly in severe bowel gangrene or infarction, and occasionally as a benign incidental finding. US, more sensitive than plain radiogra-

Congenital malformations usually present in infancy as an incidental finding on US or with cardiac failure. The acquired types may follow trauma. Intrahepatic shunts are of three varieties: arteriovenous (acquired vascular malformations [AVMs], associated with hereditary haemorrhagic telangiectasia), arterioportal (associated with Ehlers–Danlos, cirrhosis, and traumatic), or venovenous (portcaval or portohepatic), or mixed205,206. Although diagnosed by US and/or MRI, accurate delineation before any embolization requires angiography. Direct percutanous transhepatic coil deposition may be indicated in difficult cases of portal-to-hepatic venous shunting, either in isolation or with haemangiomas. Embolization

Figure 65.59 Interrupted inferior vena cava (IVC). Azygos continuation. Ultrasound of the neonate with colour Doppler shows the straight aorta in the upper abdomen running into the chest, and the IVC continuing in a posterior position as the azygos vein (arrow).

Figure 65.60 Portal venous gas. Ultrasound image of the liver in a newborn with scattered bright reflections throughout the liver parenchyma consistent with gas in the portal venous system.

Portal vein gas

CHAPTER 65

in AVMs may risk arterial ischaemia. Pure venous or venolymphatic malformations are uncommon in the liver.

LIVER TRANSPLANTATION Whole livers (cadaveric) or reduced liver grafts (living or cadaveric donors), in the form of a left lateral segment (II and III or II, III, IV) are used in children209. Imaging of the reduced liver graft requires an understanding of its anatomy: the liver is divided along the division of the segments above, preserving the left portal vein, the left hepatic duct, and an arterial supply to the left lobe.The raw surface of the liver is sealed by various methods (coagulation, cautery, ties, etc.) and transplanted in its orthotopic position with the raw surface facing a right lateral position. Early postoperatively complex collections seen on US may develop posterolaterally along the raw surface, e.g. haematomas and bilomas. Early hepatic artery thrombosis, before any potential for arterial collaterals, is an emergency with probable loss of the graft from acute hepatic necrosis. Careful Doppler study is necessary and possibly angiography. For a full description of complications, see Chapter 35210–213. In the child, the delicate balance of immunosuppression and immune competency has led to the appearance of post-transplant lymphoproliferative disease (PTLD). PTLD may occur in the transplanted organ itself or in a remote organ. The appearance of masses in the liver, spleen, nodes, brain, tonsils, mesentery, or chest raises the suspicion of PTLD. Combined with Epstein–Barr virus serology, biopsy is usually recommended to diagnose PTLD. For a full discussion of imaging of PTLD, see Chapter 72.

THE SPLEEN US is the method of choice for evaluation of spleen size and parenchymal texture. Radiolabelled sulphur colloid is useful in detection of ectopic spleen or asplenia. Coronal MRI is useful when the spleen is very large. The parenchymal texture should be homogeneous. Focal abnormalities suggest neoplasm, cyst, abscesses, or infection214–217. A linear highfrequency probe will help detect subtle abnormalities of parenchymal texture. Homogeneous splenomegaly is associated with portal hypertension, leukaemia, lymphoma, or various haematological disorders, infections, and tumours. MRI enables a full evaluation of focal lesions, as well as an accurate extent of massive splenomegaly (e.g. infiltrative disorders such as Gaucher’s disease)218.

•

PAEDIATRIC ABDOMINAL IMAGING

as well as in those children suffering from chronic granulomatous disease of childhood. These areas may calcify. Multiple lesions in the spleen are seen in HIV infection, cat scratch disease (Fig. 65.61), and histiocytosis. A snowstorm appearance on ultrasound is due to small foci of high reflectivity in Pneumocystis jiroveci (carinii).

Calcification Calcification of the spleen may be seen on plain radiography, CT, or US. Diffuse patchy calcification is seen in the haematological disorders of sickle cell disease and represents splenic infarctions. Calcification of cyst walls occurs in infections such as echinococcal disease. Calcified granulomas also occur in tuberculosis, histoplasmosis, and chronic granulomatous disease, or in opportunistic infections.

Tumours Tumours of the spleen are benign or malignant215–217,219. The benign group includes splenic cysts which are epithelial in nature (e.g. epidermoid, dermoid, transitional cell) or endothelial in nature (e.g. lymphangiomas, haemangiomas, hamartomas). These lesions may reach huge proportions and become symptomatic from capsular distension and pressure. Cholesterol-containing cysts have low level shimmering reflections within them on US (Fig. 65.62). Lymphangiomatous cysts show fine septations within the cyst. A hamartoma is an uncommon solid benign tumour in the spleen in children (Fig. 65.63). Malignant neoplasms of the spleen are usually metastatic or haematogenous in aetiology, e.g. leukaemia and lymphoma. In leukaemia and lymphoma, the spleen can be grossly enlarged homogeneously or heterogeneously with multifocal deposits of reduced reflectivity. Angiosarcoma is the most common malignant tumour of the spleen and presents as a solid heterogeneous mass within the spleen. US-guided needle biopsy of a splenic lesion may be indicated for diagnosis (Fig. 65.63) as the differential diagnosis is wide and more than one pathology can coexist.

Infections In immunocompromised children, small microabscesses, most notably fungal in origin, appear as lesions of low reflectivity with or without a central focus of high reflectivity (target lesion)193. On enhanced CT, these lesions are of low attenuation. US-guided biopsy provides microbial confirmation. Infectious mononucleosis, common in children, normally presents with diffuse splenomegaly. Splenic granulomas can be seen in tuberculosis, histoplasmosis, and coccidioidomycosis,

Figure 65.61 Cat scratch disease. Axial CT through the upper abdomen of a 5 year old girl with fever for 4 weeks shows multiple low attenuating lesions in the spleen. Ultrasound-guided biopsy of these lesions showed cat scratch disease. She improved following treatment with clarithromycin.

1527

1528

SECTION 8

• PAEDIATRIC IMAGING

Figure 65.62 Splenic cyst. Incidental splenic cyst seen in a 5 year old girl with smooth walls, no internal reflections and no flow on colour Doppler. A cholesterol containing cyst looks similar but with low level reflections within it.

Figure 65.63 Leiomyosarcoma secondary to immunosuppression. Ultrasound of spleen showing a rounded lesion of low reflectivity towards the splenic edge (arrow) in a patient after liver transplantation. m = mass, sp = spleen.

Acknowledgements We wish to acknowledge the advice and assistance given by Dr Martin Charron and Dr Philip John.

REFERENCES 1. Emanuel P G, Garcia G I, Angtuaco T L 1995 Prenatal detection of anterior abdominal wall defects with US. RadioGraphics 15: 517–530 2. Weber T R, Au-Fliegner M, Downard C D et al 2002 Abdominal wall defects. Curr Opin Pediatr 14: 491–497 3. Howell K K 1998 Understanding gastroschisis: an abdominal wall defect. Neonatal Netw 17: 17–25

4. Ergün O, Barksdale E, Ergün F S et al 2005 The timing of delivery of infants with gastroschisis influences outcome. J Pediatr Surg 40: 424–428 5. Chittmittrapap S, Spitz L, Kiely E M et al 1989 Oesophageal atresia and associated anomalies. Arch Dis Child 64: 364–368 6. Newman B, Bender T M 1997 Esophageal atresia/tracheoesophageal fistula and associated congenital esophageal stenosis. Pediatr Radiol 27: 530–534 7. Gassner I, Geley T E 2005 Sonographic evaluation of oesophageal atresia and tracheo-oesophageal fistula. Pediatr Radiol 35: 159–164 8. Kulkarni B, Rao R S, Oak S et al 1997 13 pairs of ribs—a predictor of long gap atresia in tracheoesophageal fistula. J Pediatr Surg 32: 1453–1454 9. Luo C-C, Lin J-N, Wang C-R 2002 Evaluation of oesophageal atresia without fistula by three-dimensional computed tomography. Eur J Pediatr 161: 578–580 10. Soye J A, Yarr J, Dick A C et al 2005 Multidetector row computed tomography three-dimensional volume reformatted ‘transparency’ images to define an upper pouch fistula in oesophageal atresia. Pediatr Radiol 35: 624–626 11. Stringer D A, Ein S H 1984 Recurrent tracheo-esophageal fistula: a protocol for investigation. Radiology 151: 637–641 12. Nambirajan L, Rintala R J, Losty P D et al 1998 The value of early postoperative oesophagography following repair of oesophageal atresia. Pediatr Surg Int 13: 76–78 13. Lang T, Hümmer H P, Behrens R 2001 Balloon dilation is preferable to bougienage in children with esophageal atresia. Endoscopy 33: 329–335 14. Yeming W, Somme S, Chenren S et al 2002 Balloon catheter dilatation in children with congenital and acquired esophageal anomalies. J Pediatr Surg 37: 398–402 15. Lan L C L, Wong K K Y, Lin S C L et al 2003 Endoscopic balloon dilatation of esophageal strictures in infants and children: 17 years’ experience and a literature review. J Pediatr Surg 38: 1712–1715 16. Nakamura H, Sawamura D, Goto M et al 2005 Epidermolysis bullosa simplex associated with pyloric atresia is a novel clinical subtype caused by mutations in the Plectin gene (PLEC1). J Mol Diagn 7: 28–35 17. Chew A L, Friedwald J P, Donovan C 1992 Diagnosis of congenital antral web by ultrasound. Pediatr Radiol 22: 342–343 18. Snyder W H Jr, Chaffin L 1954 Embryology and pathology of the intestinal tract: presentation of 40 cases of malrotation. Ann Surg 140: 368–379 19. Filston H C, Kirks D R 1981 Malrotation the ubiquitous anomaly. J Pediatr Surg 16 (Suppl 1): 614–620 20. Levin T L, Liebling M S, Ruzal-Shapiro C et al 1995 Midgut malfixation in patients with congenital diaphragmatic hernia: what is the risk of midgut volvulus? Pediatr Radiol 25: 259–261 21. Tiu C M, Chou Y H, Pan H B et al 1984 Congenital short bowel. Pediatr Radiol 14: 343–345 22. Ablow R C, Hoffer F A, Seashore J H et al 1983 Z-shaped duodenojejunal loop: sign of mesenteric fixation anomaly and congenital bands. Am J Roentgenol 141: 461–464 23. Long F R, Kramer S S, Markowitz R I et al 1996 Radiographic patterns of intestinal malrotation in children. RadioGraphics 16: 547–556 24. Dufour D, Delaet M H, Dassonville M et al 1992 Midgut malrotation, the reliability of sonographic diagnosis. Pediatr Radiol 22: 21–23 25. Zerin J M, Polley T Z Jr 1994 Malrotation in patients with duodenal atresia: a true association or an expected finding on postoperative upper gastrointestinal barium study? Pediatr Radiol 24: 170–172 26. Dwek J R, Teich S 1999 The duodenal dimple: a specific fluoroscopic correlate to the duodenal web. Pediatr Radiol 29: 467–468 27. Yoon C H, Goo H W, Kim E A-R et al 2001 Sonographic windsock sign of a duodenal web. Pediatr Radiol 31: 856–857 28. Schiavetti E, Massotti G, Torricelli M et al 1984 “Apple peel” syndrome. A radiological study. Pediatr Radiol 14: 380–383 29. Seashore J H, Collins F S, Markowitz R I et al 1987 Familial apple peel jejunal atresia: surgical, genetic, and radiographic aspects. Pediatrics 80: 540–544

CHAPTER 65

30. Guttman FM, Braun P, Garance PH et al 1973 Multiple atresias and a new syndrome of hereditary multiple atresias involving the gastrointestinal tract from stomach to rectum. J Pediatr Surg 8:633–640 31. Lambrecht W, Kluth D 1998 Hereditary multiple atresias of the gastrointestinal tract: report of a case and review of the literature. J Pediatr Surg 33: 794–797 32. Noerr B 2003 Part 1. Current controversies in the understanding of necrotizing enterocolitis. Adv Neonatal Care 3: 107–120 33. Carty H, Brereton R J 1983 The distended neonate. Clin Radiol 34: 367–380 34. Albanese C T, Rowe M I 1995 Necrotizing enterocolitis. Semin Pediatr Surg 4: 200–206 35. Edwards D K 1980 Size of gas-filled bowel loops in infants. Am J Roentgenol 135: 331–334 36. Uken P, Smith W, Franken E A et al 1988 Use of the barium enema in the diagnosis of necrotizing enterocolitis. Pediatr Radiol 18: 24–27 37. Kao S C, Smith W L, Franken E A Jr et al 1992 Contrast enema diagnosis of necrotizing enterocolitis. Pediatr Radiol 22: 115–117 38. Leonidas J C, Krasna I H, Fox H A et al 1973 Peritoneal fluid in necrotizing enterocolitis: a radiologic sign of clinical deterioration. J Pediatr 82: 672–675 39. Leonard T Jr, Johnson J F, Pettett P G 1982 Critical evaluation of the persistent loop sign in necrotizing enterocolitis. Radiology 142: 385–386 40. Goske M J, Goldblum J R, Applegate K E et al 1999 The “circle sign”: a new sonographic sign of pneumatosis intestinalis—clinical, pathologic and experimental findings. Pediatr Radiol 29: 530–535 41. Frey E E, Smith W, Franken E A Jr et al 1987 Analysis of bowel perforation in necrotizing enterocolitis. Pediatr Radiol 17: 380–382 42. Seibert J J, Parvey L S 1977 The telltale triangle: use of the supine cross table lateral radiograph of the abdomen in early detection of pneumoperitoneum. Pediatr Radiol 5: 209–210 43. Rovin J D, Rodgers B M, Burns R C et al 1999 The role of peritoneal drainage for intestinal perforation in infants with and without necrotizing enterocolitis. J Pediatr Surg 34: 143–147 44. Daneman A, Woodward S, de Silva M 1978 The radiology of neonatal necrotizing enterocolitis (NEC). A review of 47 cases and the literature. Pediatr Radiol 7: 70–77 45. Paley R H, McCarten K M, Cleveland R H 1979 Enterocolonic fistula as a late complication of necrotizing enterocolitis. Am J Roentgenol 132: 989–990 46. Virjee J P, Gill G J, Desa D et al 1979 Strictures and other late complications of neonatal necrotising enterocolitis. Clin Radiol 30: 25–31 47. Ball T I, Wyly J B 1986 Enterocyst formation: a late complication of neonatal necrotizing enterocolitis. Am J Roentgenol 147: 806–808 48. Haney P J, Hill J L, Sun C C 1982 Zonal colonic aganglionosis. Pediatr Radiol 12: 258–261 49. De Campo J F, Mayne V, Boldt D W et al 1984 Radiological findings in total aganglionosis coli. Pediatr Radiol 14: 205–209 50. Roshkow J E, Haller J O, Berdon W E et al 1988 Hirschsprung’s disease, Ondine’s curse, and neuroblastoma—manifestations of neurocristopathy. Pediatr Radiol 19: 45–49 51. Janik J P, Wayne E R, Janik J S et al 1997 Ileal atresia with total colonic aganglionosis. J Pediatr Surg 32: 1502–1503 52. Siu K L, Kwok W K, Lee W Y et al 1999 A male newborn with colonic atresia and total colonic aganglionosis. Pediatr Surg Int 15: 141–142 53. Amiel J, Lyonnet S 2001 Hirschsprung disease, associated syndromes, and genetics: a review. J Med Genet 38: 729–739 54. Sarioglu A, Tanyel F C, Buyukpamukcu N et al 1997 Colonic volvulus: a rare presentation of Hirschsprung’s disease. J Pediatr Surg 32: 117–118 55. Bradley M J, Pilling D 1991 The empty rectum on plain X-ray. Does it have any significance in the neonate? Clin Radiol 43: 265–267 56. Fletcher B D, Yullish B S 1978 Intraluminal calcifications in the small bowel of newborn infants with total colonic aganglionosis. Radiology 126: 451–455 57. Rosenfield N S, Ablow R C, Markowitz R I et al 1984 Hirschsprung disease: accuracy of the barium enema examination. Radiology 150: 393–400

•

PAEDIATRIC ABDOMINAL IMAGING

58. O’Donovan A N, Habra G, Somers S et al 1996 Diagnosis of Hirschsprung’s disease. Am J Roentgenol 167: 517–520 59. Amodio J, Berdon W, Abramson S et al 1986 Microcolon of prematurity: a form of functional obstruction. Am J Roentgenol 146: 239–244 60. Berrocal T, Lamas M, Gutierrez J et al 1999 Congenital anomalies of the small intestine, colon, and rectum. RadioGraphics 19: 1219–1236 61. Heaton N D, Howard E R, Garrett J R 1991 Small left colon syndrome: an immature enteric plexus. J R Soc Med 84: 113–114 62. Winters W D, Weinberger E, Hatch E I 1992 Atresia of the colon in neonates: radiographic findings. Am J Roentgenol 159: 1273–1276 63. Landes A, Shuckett B, Skarsgard E 1994 Non-fixation of the colon in colonic atresia: a new finding. Pediatr Radiol 24: 167–169 64. Hamosh A, Fitz-Simmons S C, Macek M Jr et al 1998 Comparison of the clinical manifestations of cystic fibrosis in black and white patients. J Pediatr 132: 255–259 65. Evans A K C, Fitzgerald D A, McKay K O 2001 The impact of meconium ileus on the clinical course of children with cystic fibrosis. Eur Respir J 18: 784–789 66. Li Z, Lai H J, Kosorok M R et al 2004 Longitudinal pulmonary status of cystic fibrosis children with meconium ileus. Pediatr Pulmonol 38: 277–284 67. Neal M R, Seibert J J, Vanderzalm T et al 1997 Neonatal ultrasonography to distinguish between meconium ileus and ileal atresia. J Ultrasound Med 16: 263–266 68. Navarro O M, Daneman A, Miller S F 2004 Contrast enema depiction of small-bowel volvulus in complicated neonatal bowel obstruction. Pediatr Radiol 34: 1020–1023 69. Noblett H R 1969 Treatment of uncomplicated meconium ileus by Gastrografin enema: a preliminary report. J Pediatr Surg 4: 190–197 70. Kao S C, Franken E A Jr 1995 Nonoperative treatment of simple meconium ileus: a survey of the Society for Pediatric Radiology. Pediatr Radiol 25: 97–100 71. Leonidas J C, Berdon W E, Baker D H et al 1970 Meconium ileus and its complications. A reappraisal of plain film roentgen diagnostic criteria. Am J Roentgenol Radium Ther Nucl Med 108: 598–609 72. Finkel L I, Slovis T L 1982 Meconium peritonitis, intraperitoneal calcifications and cystic fibrosis. Pediatr Radiol 12: 92–93 73. Lang I, Daneman A, Cutz E et al 1997 Abdominal calcification in cystic fibrosis with meconium ileus: radiologic–pathologic correlation. Pediatr Radiol 27: 523–527 74. Hassink E A, Rieu P N, Hamel B C et al 1996 Additional congenital defects in anorectal malformations. Eur J Pediatr 155: 477–482 75. Torres R, Levitt M A, Tovilla J M et al 1998 Anorectal malformations and Down’s syndrome. J Pediatr Surg 33: 194–197 76. Seibert J J, Golladay E S 1979 Clinical evaluation of imperforate anus: clue to type of anal-rectal anomaly. Am J Roentgenol 133: 289–292 77. Donaldson J S, Black C T, Reynolds M et al 1989 Ultrasound of the distal pouch in infants with imperforate anus. J Pediatr Surg 24: 465–468 78. Berdon W E, Baker D H, Wigger H J et al 1975 Calcified intraluminal meconium in newborn males with imperforate anus. Enterolithiasis in the newborn. Am J Roentgenol Radium Ther Nucl Med 125: 449–455 79. Gross G W, Wolfson P J, Pena A 1991 Augmented-pressure colostogram in imperforate anus with fistula. Pediatr Radiol 21: 560–562 80. Metts J C, III, Kotkin L, Kasper S et al 1997 Genital malformations and coexistent urinary tract or spinal anomalies in patients with imperforate anus. J Urol 158: 1298–1300 81. Daneman A, Navarro O 2003 Intussusception part 1: a review of diagnostic approaches. Pediatr Radiol 33: 79–85 82. Harrington L, Connolly B, Hu X et al 1998 Ultrasonographic and clinical predictors of intussusception. J Pediatr 132: 836–839 83. Harquinet S, Aniishiravani M, Vunda A et al 1998 Reliability of color Doppler and power Doppler sonography in the evaluation of intussuscepted bowel viability. Pediatr Surg Int 13: 360–362 84. Navarro O, Daneman A 2004 Intussusception part 3: diagnosis and management of those with an identifiable or predisposing cause and those that reduce spontaneously. Pediatr Radiol 34: 305–312 85. Gu L, Alton D J, Daneman A et al 1988 Intussusception reduction in children by re-insufflation of air. Am J Roentgenol 150: 1345–1348

1529

1530

SECTION 8

• PAEDIATRIC IMAGING

86. Daneman A, Navarro O 2004 Intussusception part 2: an update on the evolution of management. Pediatr Radiol 34: 97–108 87. Crystal P, Hertzanu Y, Farber B et al 2002 Sonographically guided hydrostatic reduction of intussusception in children. J Clin Ultrasound 30: 343–348 88. Rohrschneider W K, Troger J 1994 Hydrostatic reduction under US guidance. Pediatr Radiol 25: 530–534 89. Navarro O, Daneman A, Chae A 2004 Intussusception: the use of delayed repeated attempts and the management of intussusceptions due to pathologic lead points in pediatric patients. Am J Roentgenol 182: 1169–1176 90. Patriquin H B, Garcier J M, Lafortune M et al 1995 Appendicitis in children and young adults: Doppler sonographic-pathologic correlation. Am J Roentgenol 166: 629–633 91. Kessler N, Cyteval C, Gallix B et al 2004 Appendicitis: Evaluation of, sensitivity, specificity, and predictive values of US, Doppler US and laboratory findings. Radiology 230: 472–478 92. Sivit C, Siegel M, Applegate K et al 2001 When appendicitis is suspected in children. RadioGraphics 21: 247–262 93. Sivit C, Applegate K 2003 Imaging of acute appendicitis in children. Semin Ultrasound CT MR 24: 74–82 94. Sivit C 2004 Imaging the child with right lower quadrant pain and suspected appendicitis: current concepts. Pediatr Radiol 34: 447–453 95. Carty H 2002 Paediatric emergencies: non-traumatic abdominal emergencies. Eur Radiol 12: 2835–2848 96. Taylor G 2004 Suspected acute appendicitis in children: in search of the single best diagnostic test. Radiology 231: 293–295 97. Garcia Pena B M, Cook E F, Mandl K D 2004 Selective imaging strategies for the diagnosis of appendicitis in children. Pediatrics 113: 24–28 98. Fefferman N, Roche K, Pinkney L et al 2001 Suspected appendicitis in children: focused CT technique for evaluation. Radiology 220: 691–695 99. Sivit C J, Newman K D, Chandra R S 1993 Visualization of enlarged mesenteric lymph nodes at US examination. Pediatr Radiol 23: 471–475 100. Karmazyn B, Werner E, Rejaie B et al 2005 Mesenteric lymph nodes in children: what is normal? Pediatr Radiol 35: 774–777 101. Vayner N, Coret A, Polliack G et al 2003 Mesenteric lymphadenopathy in children examined by US for chronic and /or recurrent abdominal pain. Pediatr Radiol 33: 864–867 102. Rathaus V, Shapiro M, Grunebaum M et al 2005 Enlarged mesenteric lymph nodes in asymptomatic children: the value of the finding in various imaging modalities. Br J Radiol 78:30–33 103. Connolly B, O`Halpin D 1994 Sonographic evaluation of the abdomen in Henoch-Schönlein purpura. Clin Radiol 49: 320–323 104. Fotter R 1998 Imaging of constipation in infants and children. Eur J Radiol 8: 248–258 105. Wagener S, Shankar K, Turnock R et al 2004 Colonic transit time— what is normal? J Pediatr Surg 39: 166–169 106. Blethyn A J, Verrier Jones K, Newcombe R et al 1995 Radiological assessment of constipation. Arch Dis Child 73: 255–258 107. Roos J, Weishaupt D, Wildermuth S et al 2002 Experience of 4 years with open MR defaecography: pictoral review of anorectal anatomy and disease. RadioGraphics 22: 817–832 108. Franken E A, Smith W L, Frey E E et al 1987 Intestinal motility disorders of infants and children: classification, clinical manifestations and roentgenology. Crit Rev Diagn Radiol 27: 203–236 109. Goulet O, Jobert-Giraud A, Michel J-L et al 1997 Chronic intestinal pseudo-obstruction syndrome in pediatric patients. Eur J Pediatr Surg 9: 83–90 110. Rance C H, Singh S J, Kimble R 2000 Blunt abdominal trauma in children. J Paediatr Child Health 36: 2–6 111. Emery K, McAneney C, Racadio J et al 2001 Absent peritoneal fluid on screening trauma ultrasonography in children: a prospective comparison with computed tomography. J Pediatr Surg 36: 565–569 112. Strouse P J, Close B J, Marshall K W et al 1999 CT of abdominal and mesenteric trauma in children. RadioGraphics 19: 1237–1250

113. Sivit C J, Taylor G A, Eichelberger M R et al 1993 Significance of periportal low attenuation zones following blunt abdominal trauma in children. Pediatr Radiol 23: 388–390 114. Sivit C J, Eichelberger M R, Taylor G A 1994 CT in children with rupture of the bowel caused by blunt trauma: diagnostic efficacy and comparison with hypoperfusion complex. Am J Roentgenol 163: 1196–1198 115. Rao P 1999 Emergency imaging in non-accidental injury. In: Carty H (ed) Emergency paediatric radiology. Springer, Berlin, pp 348–380 116. Hernanz-Schulzman M 2003 Infantile hypertrophic pyloric stenosis. Radiology 227: 319–331 117. Cohen H L, Zinn H L, Haller J O et al 1998 Ultrasonography of pylorospasm: findings may simulate hypertrophic pyloric stenosis. J Ultrasound Med 17: 705–711 118. Orenstein S R, Orenstein D M 1988 Gastroesophageal reflux and respiratory disease in children. J Pediatr 112: 847–858 [published erratum appears in J Pediatr 1988; 113: 578]. 119. Yulish B S, Rothstein F C, Halpin T C Jr 1987 Radiographic findings in children and young adults with Barrett’s esophagus. Am J Roentgenol 148: 353–357 120. Westra S J, Wolf B H, Staalman C R 1990 Ultrasound diagnosis of gastroesophageal reflux and hiatal hernia in infants and young children. J Clin Ultrasound 18: 477–485 121. Newell S J, Chapman S, Booth I W 1993 Ultrasonic assessment of gastric emptying in the preterm infant. Arch Dis Child 69(1 Spec No): 32–36 122. Trinh T D, Benson J E 1997 Fluoroscopic diagnosis of complications after Nissen antireflux fundoplication in children. Am J Roentgenol 169: 1023–1028 123. Segal S R, Sherman N M, Rosenberg H K et al 1994 Ultrasonographic features of gastrointestinal duplications. J Ultrasound Med 13: 863–870 124. Macpherson R I 1993 Gastrointestinal tract duplications: Clinical, pathologic, and radiologic considerations. RadioGraphics 13: 1063–1080 125. Stoupis C, Ros P R, Abbitt P L et al 1994 Bubbles in the belly: imaging of cystic mesenteric or omental masses. RadioGraphics 14: 729–737 126. Vinton N E 1994 Gastrointestinal bleeding in infancy and childhood. Gastroenterol Clin North Am 23: 93–122 127. Baud C, Saguintaah M, Veyrac C et al 2004 Sonographic diagnosis of colitis in children. Eur Radiol 14: 2105–2119 128. Ali S I, Carty H M L 2000 Paediatric Crohn’s disease. Eur Radiol 10: 1085–1094 129. Grahnquist L, Chapman S C, Hvidsten S et al 2003 Evaluation of 99m Tc-HMPAO leucocyte scintography in the investigation of pediatric inflammatory bowel disease. J Pediatr 143: 48–53 130. Darbari A, Sena L, Argani P et al 2004 Gadolinium-enhanced magnetic resonance imaging: a useful radiological tool in diagnosing pediatric IBD. Inflam Bowel Dis 10: 67–72 131. Hyer W 2001 Polyposis syndromes: pediatric implications. Gastrointest Endosc Clin North Am 11: 659–682 132. Corredor J, Wambach J, Barnard J 2001 Gastrointestinal polyps in children: advances in molecular genetics, diagnosis, and management. J Pediatr 138: 621–628 133. Kurugoglu S, Aksoy H, Kantarci F et al 2003 Radiological work-up in Peutz–Jeghers syndrome. Pediatr Radiol 33: 766–771 134. Cohen M D 1992 Reticulo-endothelial tumours. In: Imaging of children with cancer. Mosby, St Louis, pp 89–126 135. Hudson M M, Krasin M J, Kaste S C 2004 PET imaging in pediatric Hodgkin’s lymphoma. Pediatr Radiol 34: 190–198 136. Wegner E A, Barrington S F, Kingston J E et al 2005 The impact of PET scanning on management of paediatric oncology patients. Eur J Nucl Med Mol Imaging 32: 23–30 137. Khati N J, Enquist E G, Javitt M C 1998 Imaging of the umbilicus and periumbilical region. RadioGraphics 18: 413–431 138. Rerksuppaphol S, Hutson J M, Oliver M R 2004 Ranitidine-enhanced 99m technetium pertechnetate imaging in children improves the sensitivity of identifying heterotopic gastric mucosa in Meckel’s diverticulum. Pediatr Surg Int 20: 323–325

CHAPTER 65

139. Agrons G A, Corse W R, Markowitz R I et al 1996 Gastrointestinal manifestations of cystic fibrosis: Radiologic–pathologic correlation. RadioGraphics 16: 871–893 140. Smyth R L, van Vetzen D, Smyth A R et al 1994 Strictures of the ascending colon in cystic fibrosis and high strength pancreatic enzymes. Lancet 342: 85–86 141. Almberger M, Iannicelli E, Antonelli M et al 2001 The role of MRI in the intestinal complications in cystic fibrosis. J Clin Imaging 25: 344–348 142. Zerin J M, Kuhn-Fulton J, White S J et al 1995 Colonic strictures in children with cystic fibrosis. Radiology 194: 223–226 143. Dialer I, Hundt C, Bertele-Harms R M et al 2003 Sonographic evaluation of bowel wall thickness in patients with cystic fibrosis. J Clin Gastroenterol 37: 55–60 144. Lardenoye S W, Puylaert J B, Smit M J et al 2004 Appendix in children with cystic fibrosis: US features. Radiology 232: 187–189 145. Patriquin H, Lenaerts C, Smith L et al 1999 Liver disease in children with cystic fibrosis: US-biochemical comparison in 195 patients. Radiology 211: 229–232 146. Lenaerts C, Lapierre C, Patriquin H et al 2003 Surveillance for cystic fibrosis-associated hepatobiliary disease: Early changes and predisposing factors. J Pediatr 143: 343–350 147. Williams S M, Goodman R, Thompson A et al 2002 Ultrasound evaluation of liver disease in cystic fibrosis as part of an annual assessment clinic: a nine year review. Clin Radiol 57: 365–370 148. Neglia J P, Fitzsimmons S C, Maisonneuve P et al 1995 The risk of cancer among patients with cystic fibrosis. N Engl J Med 332: 494–499 149. Haller J O, Cohen H L 1994 Gastrointestinal manifestations of AIDS in children. Am J Roentgenol 162: 387–393 150. Andronikou S, Welman C J, Kader E 2002 The CT features of abdominal tuberculosis in children. Pediatr Radiol 32: 75–81 151. McCarville B, Adelman C S, Li C 2005 Typhlitis in childhood cancer. Cancer 104: 380–387 152. Cartoni C, Dragoni F, Micozzi A et al 2001 Neutropenic enterocolitis in patients with acute leukaemia: prognostic significance of bowel wall thickening detected by ultrasonography. J Clin Oncol 19; 756–761 153. Kirkpatrick I D C, Greenberg H M 2003 Gastrointestinal complications in the neutropenic patient : Characterization and differentiation with abdominal CT. Radiology 226: 668–674 154. McGahen E D 1999 Esophageal foreign bodies. Pediatr Rev 20: 129–133 155. Eggli K D, Potter B M, Garcia V et al 1986 Delayed diagnosis of oesophageal perforation by aluminium foreign bodies. Pediatr Radiol 16: 511–513 156. O’Hara S, Donnelly L F, Chuang E et al 1999 Gastric retention of zincbased pennies: radiographic appearance and hazards. Pediatr Imaging 213: 113–117 157. Neilson I R 1995 Ingestion of coins and batteries. Pediatr Rev 16: 35–36 158. Tay E T, Weinberg G, Levin T L 2004 Ingested magnets: the force within. Pediatr Emerg Care 20: 466–467 159. Cauchi J A, Shawis R N 2002 Multiple magnet ingestion and gastrointestinal morbidity. Arch Dis Child 87: 539–540 160. Macpherson R I, Hill J G, Othersen H B et al 1996 Esophageal foreign bodies in children: diagnosis, treatment and complications. Am J Roentgenol 166: 919–924 161. Nagi B, Kochhar R, Thapa B R et al 2004 Radiological spectrum of late sequelae of corrosive injury to upper gastrointestinal tract. A pictorial review. Acta Radiol 45: 7–12 162. Nuutinen M, Uhari M, Karvali T et al 1994 Consequences of caustic ingestions in children. Acta Pediatr 83: 1200–1205 163. Fasulakis S, Andronikou S 2003 Balloon dilatation in children for oesophageal strictures other than those due to primary repair of oesophageal atresia, interposition or restrictive fundoplication. Pediatr Radiol 33: 682–687 164. Donnelly L F, Bisset G S 3rd 1998 Pediatric hepatic imaging. Radiol Clin North Am 36: 413–427

•

PAEDIATRIC ABDOMINAL IMAGING

165. Siegel M J 2001 Pediatric liver imaging. Semin Liver Dis 21: 251–269 166. Hudson-Dixon C M, Long B W, Cox L A 1999 Power Doppler imaging: principles and applications. Radiol Technol 70: 235–243 167. Siegel M J 1995 MR imaging of the pediatric abdomen. Magn Reson Imaging Clin North Am 3: 161–182 168. Guibaud L, Lachaud A, Touraine R et al 1998 MR cholangiography in neonates and infants: feasibility and preliminary applications. Am J Roentgenol 170: 27–31 169. Miyazaki T, Yamashita Y, Tang Y, Tsuchigame T, Takahashi M, Sera Y 1998 Single-shot MR cholangiopancreatography of neonates, infants, and young children. Am J Roentgenol 170: 33–37 170. Chan Y L, Yeung C K, Lam W W, Fok T F, Metreweli C 1998 Magnetic resonance cholangiography–feasibility and application in the paediatric population. Pediatr Radiol 28: 307–311 171. Roca I, Ciofetta G 1998 Hepatobiliary scintigraphy in current pediatric practice. Q J Nucl Med 42: 113–118 172. Nadel H R 1996 Hepatobiliary scintigraphy in children. Semin Nucl Med 26: 25–42 173. Gilmour S M, Hershkop M, Reifen R, Gilday D, Roberts E A 1997 Outcome of hepatobiliary scanning in neonatal hepatitis syndrome. J Nucl Med 38: 1279–1282 174. Cheng C L, Fogel E L, Sherman S et al 2005 Diagnostic and therapeutic endoscopic retrograde cholangiopancreatography in children: a large series report. J Pediatr Gastroenterol Nutr 41: 445–453 175. Applegate K E, Goske M J, Pierce G, Murphy D 1999 Situs revisited: imaging of the heterotaxy syndrome. RadioGraphics 19: 837–852; discussion 853–834 176. Moore K P, Persaud T V N 2002 The developing human, clinically oriented embryology, 6th edn. W B Saunders, Philadelphia 177. McGahan J P, Phillips H E, Cox K L 1982 Sonography of the normal pediatric gallbladder and biliary tract. Radiology 144: 873–875 178. Hernanz-Schulman M, Ambrosino M M, Freeman P C, Quinn C B 1995 Common bile duct in children: sonographic dimensions. Radiology 195: 193–195 179. Schlesinger A E, Braverman R M, DiPietro M A 2003 Pictorial essay. Neonates and umbilical venous catheters: normal appearance, anomalous positions, complications, and potential aid to diagnosis. Am J Roentgenol 180: 1147–1153 180. Rosenberg H K, Markowitz R I, Kolberg H, Park C, Hubbard A, Bellah R D 1991 Normal splenic size in infants and children: sonographic measurements. Am J Roentgenol 157: 119–121 181. Shah H A, Spivak W 1994 Neonatal cholestasis. New approaches to diagnostic evaluation and therapy. Pediatr Clin North Am 41: 943–966 182. Venigalla S, Gourley G R 2004 Neonatal cholestasis. Semin Perinatol 28: 348–355 183. Davenport M 2005 Biliary atresia. Semin Pediatr Surg 14: 42–48 184. Torrisi J M, Haller J O, Velcek F T 1990 Choledochal cyst and biliary atresia in the neonate: imaging findings in five cases. Am J Roentgenol 155: 1273–1276 185. Mack C L, Sokol R J 2005 Unraveling the pathogenesis and etiology of biliary atresia. Pediatr Res 57: 87R–94R 186. Bezerra J A 2005 Potential etiologies of biliary atresia. Pediatr Transplant 9: 646–651 187. Park W H, Choi S O, Lee H J, Kim S P, Zeon S K, Lee S L 1997 A new diagnostic approach to biliary atresia with emphasis on the ultrasonographic triangular cord sign: comparison of ultrasonography, hepatobiliary scintigraphy, and liver needle biopsy in the evaluation of infantile cholestasis. J Pediatr Surg 32: 1555–1559 188. Petersen C 2004 Surgery in biliary atresia—futile or futuristic? Eur J Pediatr Surg 14: 226–229 189. Sherman P, Kolster E, Davies C, Stringer D, Weber J 1986 Choledochal cysts: heterogeneity of clinical presentation. J Pediatr Gastroenterol Nutr 5: 867–872 190. Brancatelli G, Federle M P, Vilgrain V, Vullierme M P, Marin D, Lagalla R 2005 Fibropolycystic liver disease: CT and MR imaging findings. RadioGraphics 25: 659–670 191. Krawinkel M B 2004 Parenteral nutrition-associated cholestasis— what do we know, what can we do? Eur J Pediatr Surg 14: 230–234

1531

1532

SECTION 8

• PAEDIATRIC IMAGING

192. Chardot C, Iskandarani F, De Dreuzy O et al 1996 Spontaneous perforation of the biliary tract in infancy: a series of 11 cases. Eur J Pediatr Surg 6: 341–346 193. Brook I 2004 Intra-abdominal, retroperitoneal, and visceral abscesses in children. Eur J Pediatr Surg 14: 265–273 194. Seeto R K, Rockey D C 1996 Pyogenic liver abscess. Changes in etiology, management, and outcome. Medicine (Baltimore) 75: 99–113 195. Khanna G, Kao S C, Kirby P, Sato Y 2005 Imaging of chronic granulomatous disease in children. RadioGraphics 25: 1183–1195 196. Emre S, McKenna G J 2004 Liver tumors in children. Pediatr Transplant 8: 632–638 197. von Schweinitz D, Gluer S, Mildenberger H 1995 Liver tumors in neonates and very young infants: diagnostic pitfalls and therapeutic problems. Eur J Pediatr Surg 5: 72–76 198. Pobiel R S, Bisset G S, 3rd 1995 Pictorial essay: imaging of liver tumors in the infant and child. Pediatr Radiol 25: 495–506 199. Donnelly L F, Bisset G S 3rd 2002 Unique imaging issues in pediatric liver disease. Clin Liver Dis 6: 227–246, viii 200. Kassarjian A, Dubois J, Burrows P E 2002 Angiographic classification of hepatic hemangiomas in infants. Radiology 222: 693–698 201. Teo E L, Strouse P J, Prince M R 1999 Applications of magnetic resonance imaging and magnetic resonance angiography to evaluate the hepatic vasculature in the pediatric patient. Pediatr Radiol 29: 238–243 202. Koumanidou C, Vakaki M, Papadaki M, Pitsoulakis G, Savvidou D, Kakavakis K 1999 New sonographic appearance of hepatic mesenchymal hamartoma in childhood. J Clin Ultrasound 27: 164–167 203. Murray J D, Ricketts R R 1998 Mesenchymal hamartoma of the liver. Am Surg 64: 1097–1103 204. Keller M S, Markle B M, Laffey P A, Chawla H S, Jacir N, Frank J L 1985 Spontaneous resolution of cholelithiasis in infants. Radiology 157: 345–348 205. Burrows P E, Dubois J, Kassarjian A 2001 Pediatric hepatic vascular anomalies. Pediatr Radiol 31: 533–545 206. Gallego C, Miralles M, Marin C, Muyor P, Gonzalez G, Garcia-Hidalgo E 2004 Congenital hepatic shunts. RadioGraphics 24: 755–772 207. Ohta M, Hashizume M, Tomikawa M, Ueno K, Tanoue K, Sugimachi K 1994 Analysis of hepatic vein waveform by Doppler ultrasonography in 100 patients with portal hypertension. Am J Gastroenterol 89: 170–175 208. Okuda K, Kage M, Shrestha S M 1998 Proposal of a new nomenclature for Budd–Chiari syndrome: hepatic vein thrombosis versus thrombosis of the inferior vena cava at its hepatic portion. Hepatology 28: 1191–1198

209. Kelly D A 1998 Pediatric liver transplantation. Curr Opin Pediatr 10: 493–498 210. Patenaude Y G, Dubois J, Sinsky A B et al 1997 Liver transplantation: review of the literature. Part 3: Medical complications. Can Assoc Radiol J 48: 333–339 211. Patenaude Y G, Dubois J, Sinsky A B et al 1997 Liver transplantation: review of the literature. Part 2: Vascular and biliary complications. Can Assoc Radiol J 48: 231–242 212. Patenaude Y G, Dubois J, Sinsky A B et al 1997 Liver transplantation: review of the literature. Part 1: Anatomic features and current concepts. Can Assoc Radiol J 48: 171–178 213. Westra S J, Zaninovic A C, Hall T R, Busuttil R W, Kangarloo H, Boechat M I 1993 Imaging in pediatric liver transplantation. RadioGraphics 13: 1081–1099 214. Vural M, Kacar S, Kosar U, Altin L 1999 Symptomatic wandering accessory spleen in the pelvis: sonographic findings. J Clin Ultrasound 27: 534–536 215. Paterson A, Frush D P, Donnelly L F, Foss J N, O’Hara S M, Bisset G S 3rd 1999 A pattern-oriented approach to splenic imaging in infants and children. RadioGraphics 19: 1465–1485 216. Karakas H M, Tuncbilek N, Okten O O 2005 Splenic abnormalities: an overview on sectional images. Diagn Interv Radiol 11: 152–158 217. Freeman J L, Jafri S Z, Roberts J L, Mezwa D G, Shirkhoda A 1993 CT of congenital and acquired abnormalities of the spleen. RadioGraphics 13: 597–610 218. Elsayes K M, Narra V R, Mukundan G, Lewis J S Jr, Menias C O, Heiken J P 2005 MR imaging of the spleen: spectrum of abnormalities. RadioGraphics 25: 967–982 219. Thompson S E, Walsh E A, Cramer B C et al 1996 Radiological features of a symptomatic splenic hamartoma. Pediatr Radiol 26: 657–660

SUGGESTED FURTHER READING Carty H, Brunelle F, Stringer D A, Kao S C S (eds) 2005 Imaging children. Churchill Livingstone, Edinburgh Kuhn J P, Slovis T L and Haller J O (eds) 2004 Caffey’s pediatric diagnostic imaging. Mosby, Philadelphia Stringer D A 1989 Pediatric gastrointestinal imaging. BC Decker, Toronto

CHAPTER

Imaging of the Kidneys, Urinary Tract and Pelvis in Children

66

Rose de Bruyn, Isky Gordon and Kieran McHugh

Congenital anomalies • Renal anomalies • Uterus and vagina • Undescended testis Imaging techniques • Ultrasound • Cystography • Nuclear medicine • Urography (plain radiograph and intravenous urogram) • Computed tomography • Magnetic resonance imaging • Interventional procedures Integrated imaging • The neonate • Bladder anomalies • Urethral anomalies • Duplex kidney • Urinary tract infection and vesico-ureteric reflux • Hypertension • Renal cystic disease • Renal transplantation • Tumours • Scrotal masses • Ovarian masses • Presacral masses • Nephrocalcinosis (renal calculus) • Inflammatory diseases of the scrotum

Advances in imaging in infants and children relies heavily on real-time ultrasound (US) for anatomy, and nuclear medicine studies for functional assessment using either dynamic or static agents such as 99mtechnetium-dimercaptosuccinic acid (99mTc-DMSA). Intravenous urography (IVU)

is now only used to outline calyceal anatomy. Micturating cystourethrography (MCUG) remains crucial for the male urethra. Assessment of vesico-ureteric reflux (VUR) can be done using either radionuclide cystography, MCUG, or US. The use of magnetic resonance imaging (MRI) is developing and its role will become clearer in the future. In tumours computed tomography (CT) and MRI have a crucial role. Angiography is reserved for special clinical situations and, although invasive with a high radiation burden, it still remains the gold standard. CT and MR angiography (MRA), however, are emerging as possible replacements for angiography for diagnostic purposes in the main renal artery. Angioplasty, embolization, antegrade studies, and drainage procedures are becoming more common. It should be borne in mind that each imaging technique has its strengths and weaknesses. A guiding principle in paediatrics is to choose the least invasive but most appropriate imaging technique with the lowest radiation dose whenever possible. Consequently, almost all genitourinary studies in children should start with US, with the probable exception of severe abdominal trauma. The attitude of every member of the radiology department should be positive, reassuring, and sympathetic. Explanations to allay anxiety must be offered to the child and parent/carer at all stages of the referral for imaging. Ideally we wish for a cooperative child who can also express his/her fears and anxieties. With sufficient information, the parent/carer can frequently help the child overcome anxiety, especially if he/she remains with the child throughout the examination. With this approach, sedation is rarely used in most examinations of the urinary tract (exceptions being CT and MRI in the young or uncooperative). Sedation is positively discouraged for cystography as the child may be too sleepy to void. No special equipment is required for genitourinary examinations in children. All staff must be appropriately trained.

1534

SECTION 8

•

PAEDIATRIC IMAGING

CONGENITAL ANOMALIES RENAL ANOMALIES Unilateral renal agenesis occurs in around 1 in 1250 live births. Antenatal diagnosis is uncommon, suggesting that agenesis may be an involuted multicystic kidney. It is difficult to differentiate between unilateral agenesis and a small nonfunctioning kidney, especially if it is ectopically located. This is of clinical importance in a girl who is constantly wet and in whom the small kidney drains via an ectopic ureter, and in a child with hypertension, where the small kidney may be the source of excess renin.To exclude a small ectopic kidney requires 99mTcDMSA imaging with anterior views. MRI is useful if there is a dilated ureter but the IVU has no role. Bilateral renal agenesis is incompatible with life. Newborns who die from chronic renal failure do so towards the end of the first weeks of life and those who die in the first hours or day of life usually do so from respiratory causes.

Abnormalities of position An ectopically placed kidney is of little clinical significance, but a malrotated kidney may develop urological complications, be more susceptible to trauma, or indicate that pathology in an adjacent organ is displacing the kidney. In malrotation, the upper pole of the kidney is located more laterally than the lower pole, and the upper pole calyces are therefore lateral to the lower pole calyces. US will ensure that there is no suprarenal mass lesion causing the malrotation. The ectopic kidney may lie within the bony pelvis and present as a mass; its surgical removal owing to the mistaken diagnosis of tumour is well recorded. Occasionally the kidney continues to migrate beyond the normal renal fossa and ends up in the thorax. This anomaly is associated with eventration of the diaphragm so that in some cases the kidney migrates upwards to lie behind the heart in the thorax and may resemble a posterior mediastinal mass on the chest radiograph.

Abnormalities of fusion The duplex kidney is common and can have varied manifestations (see below). The most common fusion abnormality is that of the horseshoe kidney with fusion of the lower poles; this is always associated with malrotation so that the pelves and ureters pass anteriorly over the fused lower poles. The two well-recognized complications are renal pelvis dilatation (with or without obstruction) and renal calculi. The abnormal position also leaves the kidney more susceptible to injury. There is a slightly increased risk of Wilms’ tumour in a horseshoe kidney. In the uncomplicated horseshoe anomaly US may not detect the abnormal axes of the kidneys, and is therefore not reliable to exclude a horseshoe kidney. When a child presents with intermittent pain and/or haematuria and the US diagnosis is that of a horseshoe kidney, calculi may be missed and plain radiographs or, rarely, nonenhanced CT may be useful. Radionuclide imaging (99mTc-DMSA) with an anterior view is useful to show all functioning renal tissue, especially over the spine. MRI can easily demonstrate a horseshoe kidney but usually requires anaesthesia in younger patients.

Figure 66.1 Renal ectopia. Crossed fused renal ectopy demonstrated on an intravenous urogram.

Crossed fused renal ectopia occurs when one kidney is displaced across the midline and is fused inferiorly to the other relatively normally positioned kidney. Both ureters enter the bladder in a normal position (Fig. 66.1). The importance of this condition is that it may present as an abdominal mass or as an obstructive uropathy with a pelvi-ureteric junction (PUJ) obstruction. US reveals an unusually large kidney on the affected side and an absent kidney on the opposite side. A dynamic renogram, such as a 99mTc-mercaptoacetyltriglycine MAG3 study, is required and especially if surgery is being considered for PUJ obstruction. There is an increased incidence of VUR into the crossed kidney and if further anatomical information concerning the ureter and pelves is required an MCUG may prove helpful. The 99mTc-DMSA scintigram shows patchy uptake of the isotope owing to the anatomical abnormality and dysplasia that exists to some degree. A rare fusion anomaly is seen when both kidneys have failed to ascend from the pelvis, which is referred to as a pancake kidney (Fig. 66.2).

UTERUS AND VAGINA In the absence of anti-Müllerian hormone secretion by a fetal testis, the Müllerian ducts meet in the midline by 8 weeks’ gestation and subsequently differentiate into the uterus, cervix, Fallopian tubes, and proximal two-thirds of the vagina. The distal third of the vagina is formed by the urogenital sinus.

CHAPTER 66

Müllerian duct development into the uterus depends on the formation of the Wolffian (mesonephric) duct. Congenital anomalies of the uterus and vagina are uncommon; there is a high incidence of associated renal (50%) and skeletal anomalies (12%). The anomalies usually present as an abdominal or pelvic mass, secondary to obstruction; later they may cause serious complications of sexual function, pregnancy, and labour. The main role of imaging is to document the presence or absence of the uterus and to identify the gonads (Figs 66.3, 66.4). US and sometimes genitography are

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

used routinely, MRI being reserved for cases in which the previous studies were inconclusive. Congenital anomalies of the vagina include complete or partial agenesis, stenosis, septation, and imperforate membrane, all of which cause obstruction (Fig. 66.5). Clinically, the older girl presents with symptoms of menstruation but no bleeding; a palpable mass with haematometrocolpos is found at US (Fig. 66.6). The Mayer–Rokitansky–Kuster–Hauser syndrome is the second most common cause of primary amenorrhoea and includes vaginal atresia, a spectrum of uterine anomalies (hypoplasia or

Figure 66.4 Normal uterus. Sagittal US of a prepubertal uterus in a 6 year old girl. The fundus and cervix are of equal size; there are no endometrial echoes.

Figure 66.2 Pancake kidney demonstrated on an intravenous urogram. Note the patient also has a hypoplastic sacrum.

Figure 66.3 Normal ovaries. Oblique US of the right ovary in an 8 year old girl. This shows the normal appearance of multiple follicles in the ovary.

Figure 66.5 Hydrocolpos. T2-weighted parasagittal MRI showing a markedly distended vagina in a neonate with fluid also seen superiorly in a less distended uterine cavity, seen posterior to the bladder.

1535

1536

SECTION 8

•

PAEDIATRIC IMAGING

V

Figure 66.6 Hydrocolpos. A 10 year old girl had undergone menarche and presented with a 10 d history of spotty vaginal bleeding. (A) A longitudinal midline US image of the pelvis shows a heterogeneous mass in the region of the vagina, clearly separated from the uterus. On physical examination, there was an imperforate hymen which was punctured, releasing 500 ml of blood from the vagina. (B) A follow-up study on the following day showed resolution of the echogenic material, with only a small amount of fluid present within the vagina (v), outlining the uterine cervix.

duplication), and normal Fallopian tubes and ovaries. Imaging assists in the demonstration of uterus didelphys, unilateral hydrometrocolpos, and ipsilateral renal agenesis.

UNDESCENDED TESTIS Cryptorchidism is present in 4% of full-term newborns, but the rate decreases to 0.8% by the first year as many testes descend spontaneously to the scrotal sac. As there is a risk of infertil-

ity and malignancy in the undescended testis, early diagnosis and treatment are important.The ectopic testis may be located anywhere from the retroperitoneum to the scrotum. Most ectopic testes are positioned within the inguinal canal and are easily visualized by US. Ipsilateral renal agenesis is associated with an absent testis so the examination should include renal US. The normal testis has high T2 signal but, contrary to early reports, MRI is poor at locating the undescended testis. Laparoscopy is increasingly used as the definitive investigation for an undescended testis.

IMAGING TECHNIQUES ULTRASOUND US is noninvasive and provides anatomical information of the retroperitoneal, intra-abdominal, and pelvic structures. It is independent of function. In the young child, US should always start with views of the bladder, since the child may micturate at any moment and the distended bladder provides an acoustic window for the lower urinary tract, uterus, adnexae, prostate gland, seminal vesicles, and pelvic musculature and vessels. Following micturition, the residual bladder volume should be estimated. Measurements of bipolar renal lengths should form part of the routine examination and be recorded1,2 (Fig. 66.7). Normative graphs for age, weight, and height, and for the single kidney should be referred to in all reports (Fig. 66.8). In the normal neonatal kidney the renal cortex is more reflective than the adjacent normal liver, the medullary pyramids are hypoechoic and thus appear prominent, and there is an absence of renal sinus fat.These neonatal US appearances tend to become less prominent during the first 6 months of age. US is sensitive in the detection of hydronephrosis, nephrocalcinosis, and renal calculi. Doppler US is important in many conditions, e.g. suspected renal vein thrombosis or immediately following renal transplantation. Visualization of the ovaries by US in girls depends on their location and size, and the age of the child. Ovarian volume

is calculated using the formula according to a prolate ellipse: length × depth × width × 0.523. The mean ovarian volume before 6 years of age is usually 1 cm3.After the age of 6, the ovaries gradually begin to enlarge and there is a marked increase after puberty3. Menstruating adolescents have a mean ovarian volume of 9.8 cm3. In neonates the ovaries can be found anywhere in the abdomen and the pelvis. In childhood and adolescence, the ovaries move posterolaterally to the uterus, but due to the mobile ligamentous attachments may be elsewhere in the pelvis4. The appearance of the ovaries is heterogeneous, secondary to the presence of normal small follicles. Ovarian cysts are seen at all ages. Primordial (nonstimulated follicles) measure less than 9 mm in diameter, while stimulated follicles range between 9 mm and 3 cm in diameter, and these are asymptomatic. The uterus in the neonatal period is quite large (2.3–4.6 cm in length) and thickened with clearly visible endometrial lining caused by the circulating maternal oestrogens. By 2–3 months of age it has changed to a prepubertal configuration with a tubular appearance; the fundus and cervix are the same size and endometrial echoes are no longer present. At puberty, the fundus starts to enlarge and becomes up to three times the size of the cervix. The total uterine length increases to 5–7 cm and the appearance of the endometrial lining will vary with the phase of the menstrual cycle.

CHAPTER 66

13

14 Predicted mean 95% prediction mean

12

12

10

11

9

10

Renal length (cm)

11

Renal length (cm)

Predicted mean 95% prediction mean

13

8 7 6

9 8 7

5

6

4

5

3

4

2 0 2 4 6 8 10 12

5

10

A

15

0

years

months

10

20

30

150

170

190

40 50 Weight (kg)

60

70

80

90

B

Age 13

Predicted mean 95% prediction mean

12 11

Renal length (cm)

10 9 8 7 6 5 4

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

vesicles can be identified in older boys and adolescents, but are difficult to visualize in infants and young boys. The normal testis changes in appearance during childhood. It has a homogeneous texture and is spherical or oval in shape during the neonatal period, measuring 7–10 mm in diameter. The epididymis and mediastinum testis are usually not seen at this point, but are clearly identified by puberty. Testicular size in adolescence ranges from 3 to 5 cm in length and from 2 to 3 cm in depth and width. Testicular flow, as measured by Doppler US, also changes with age. The testis in infants seldom shows colour flow, in spite of optimized slow flow settings. Power-mode Doppler US improves depiction of intratesticular flow in normal pre- and post-pubertal testes, but there are several cases of normal prepubertal testes that show no flow with this technique. The ability to delineate normal and abnormal anatomy is directly related to the skill of the ultrasonographer who must be experienced in examining children and fully conversant with the spectrum of diseases in childhood. A quiet environment, appropriate to the needs of children, leads to better results. One important limitation of US is in the detection of renal scars.

3

CYSTOGRAPHY

2

C

30

50

70

90

110 130 Height (cm)

Figure 66.7 Renal growth chart. Normal renal growth charts for age, weight and height in normal children. Renal patients are often poorly grown so it is best to use charts with all the growth parameters recorded. (Reproduced with permission from Han B K, Babcock D S 1985 Am J Roentgenol 145: 611–616.)

120

Renal length (mm)

100 80 60 40

There are four different methods for undertaking cystography in children. The ideal principle of using the lowest radiation burden and the least invasive technique must be borne in mind, remembering that the MCUG generally has a high radiation dose. All techniques except indirect radio-isotope cystography (IRC) require a bladder catheter. The MCUG, however, is the only technique that can visualize the male urethra adequately. The least invasive method is the IRC and it has a low radiation burden using 99mTc-MAG3; it is also the only physiological examination. However, it requires a toilettrained child. US cystography involving bladder catheterization and the use of US contrast agents is not widely available. Intermittent or transitory VUR may be missed whatever technique is used.

20

Micturating cystourethrogram 0

Figure 66.8 Standards for a single functioning kidney. Plot of mean renal length measurements against weight together with values for 2 SDs above and below mean for a single functioning kidney in children.

This is the definitive method of assessing the lower urinary tract but may reveal upper tract detail if VUR is present. It is necessary in all boys in whom there is any suspicion of urethral pathology; images of the urethra are best achieved during voiding in an oblique projection.

The vagina is identified on US by internal linear bright echoes. Commonly a urine-filled vagina is seen after a girl has voided. On MRI the vagina is best seen on thin axial or sagittal T2-weighted MRI. As with the uterus, the appearance and the thickness of the vaginal epithelium and the signal from the vaginal wall change with the phases of the menstrual cycle. The prostate has an ellipsoid configuration in boys. Its texture is more homogeneous than in adults. The seminal

Indications An MCUG is indicated in all boys with suspected urethral pathology. It is often also done in those under 3 years of age with a urinary tract infection (UTI) who have an abnormal US or DMSA scintigram; when ureteric dilatation is found; for terminal haematuria accompanied by lower urinary tract symptomatology; and also possibly in renal failure of undetermined cause with an abnormal US; in certain voiding problems, particularly when a thick-walled bladder is seen on US; and for infant screening in those who have an older sibling with VUR.

0 3.4

4.5

6.8 7.1

8.4 9.3 10.5 11.8 18.1 16.2 Weight (kg)

17.1 21

1537

1538

SECTION 8

•

PAEDIATRIC IMAGING

Method A small (6F) feeding tube, rather than a balloon catheter, is inserted via the urethra under strict aseptic conditions. If a balloon catheter is already in place, the balloon should be deflated for the cystogram. Oblique voiding images of the male urethra are mandatory without the catheter in situ. An adequate voiding image of the male urethra can be difficult to achieve and so simultaneous videotape recording is useful for review. Dilute contrast medium, ideally warmed to body temperature, is administered either via drip infusion or careful injection; slow filling is ideal. With good urethral distension during voiding, posterior urethral valves will not be obscured by a 6F feeding catheter. If posterior urethral valves are discovered, then the bladder should be left on open catheter drainage so as not to leave a closed system that may become infected. Cyclic or double voiding leads to higher sensitivity in the detection of reflux. Contraindications For the follow-up of known VUR, a radio-isotope cystogram should be carried out. A direct radioisotope cystogram (DIC) is adequate for diagnosing VUR in girls who are not toilet trained, and have a UTI and abnormal 99m Tc-DMSA findings.

Direct radio-isotope cystogram Indications A DIC is indicated whenever renal reflux must be excluded. This group of patients mainly includes young girls with a history of UTI and boys who require follow-up cystography (having previously had an MCUG). There are no contraindications.

Method

Indications The indications are: • whenever renal reflux must be excluded in the older toilettrained child • ureteric dilatation in the toilet-trained child • children with bladder dysfunction, including posterior urethral valves, where the entire nephro-urological system can be evaluated.

Method Following a routine 99mTc-MAG3 scintigram, the child returns to the gamma camera, which has been turned horizontal, when he/she wishes to void. The child sits with the camera at his/her back; often the boys stand. Acquisition starts 30 s before micturition and continues for 30 s afterwards with continuous data acquisition. Processing includes viewing the data in cine mode, drawing regions of interest (ROI) over the bladder and kidneys, as well as creating a compressed image of the entire procedure.

NUCLEAR MEDICINE Dynamic renogram The isotopes used are either 99mTc-DTPA, 123I-hippuran or, commonly, 99mTc-MAG3. Analysis includes renal perfusion, divided renal function (DRF), and drainage of the collecting systems. (A diuretic is required in the presence of a dilated collecting system.) 99mTc-DTPA uptake reflects glomerular function while 99mTc-MAG3 uptake reflects tubular function. In most clinical situations this difference is unimportant.

Indications

This is similar to MCUG—instilling 99mTc-pertechnetate (20 MBq) and warm normal saline until the bladder is full, when micturition should occur. The entire procedure is carried out on top of the gamma camera linked to a computer system with a double disposable nappy on the infant. Both renal areas and the bladder are kept in the field of view. Sedation is never required.

The indications are: • to assess DRF and drainage in suspected obstruction or before surgery on the collecting system • postoperative evaluation of the collecting system • for IRC • following renal transplantation • renography with captopril stimulation for renovascular hypertension.

Ultrasound cystogram

Method

Although described some time ago, this technique is not used frequently as the US contrast agents are not licensed for intravesical or paediatric use in many countries. A bladder catheter is required and it follows the concept of the MCUG. The advantage is lack of irradiation; the disadvantages include the need for catheterization and the logistic difficulty of doing bilateral renal US during voiding.

Indirect radio-isotope cystogram The IRC is physiological, will assess renal and bladder function, does not require a bladder catheter, and will detect reflux under normal voiding conditions. IRC is undertaken following the intravenous injection of diethylene triamine penta-acetate (99mTc-DTPA), 123I-hippuran or, commonly, 99mTc-MAG3. Once the majority of the isotope is in the bladder, the child is asked to void in front of the gamma camera.

Hydration The child should be encouraged to drink from the time of arrival in the department until the actual injection. Voiding before the injection is recommended. An anaesthetic cream should be applied 45 min before the injection. The administered dose is scaled on a body surface basis, with a maximum dose of 80 MBq for 99mTc-MAG3 and 200 MBq for 99m Tc-DTPA, and a minimum dose of 20 MBq for both tracers. The effective dose equivalent (EDE) for a 5 year old child using 99m Tc-DTPA is 0.63 mSv while for 99mTc-MAG3 it is 0.40 mSv. The gamma camera is face up using a low energy, general purpose collimator with the child supine. To reduce movement, sandbags are placed on either side of the child with Velcro straps around the child and the sandbags. The heart, kidneys, and bladder should all be included in the field of view. A frame rate of 10 or 20 s per frame is required and the study should continue for 20 min.

CHAPTER 66

In the presence of a full bladder, drainage from the kidney may be delayed. The use of a diuretic usually causes the child to void, often within 15–20 min; additional data should be routinely acquired after micturition and a change in posture so that analysis of the kidneys can be undertaken when the bladder is empty. The change in posture should be for approximately 5 min before the 1-min late series dynamic data are acquired. −1

Diuretic administration 1 mg kg of frusemide with a maximum dose of 20 mg can be given at appropriate times. Differential renal function estimation The relative function of each kidney is expressed as a percentage of the sum of the right and left kidneys. It is computed from the renogram, during the period 60–120 s from the peak of the vascular curve. The integral and the Patlak–Rutland plot method are the only two internationally recommended techniques. Interpretation of diuretic challenge Simple analysis of the post-frusemide curve only is strongly discouraged since this curve fails to take into consideration the effect of gravity, the bladder status, and the function of that particular kidney. The routine use of the post-micturition image plus assessment of washout relative to the uptake function of the kidney will overcome all these difficulties. Good drainage is easy to define, since the images, curves, and numerical data all reveal little tracer in the kidney and dilated collecting system at the end of the study. However, when drainage is reduced then there is little agreement as to what constitutes moderate drainage or poor drainage. The significance of impaired drainage is also strongly debated and the relationship between impaired drainage and different treatment modalities is not clear. Using pelvic excretion efficiency (PEE), good drainage is defined as a kidney and renal pelvis clearing 84% or more of the isotope which has been taken up by the kidney at the end of the study.

Static renal scintigraphy 99m

Tc-DMSA binds to the proximal convoluted tubules, resulting in fixation of the isotope and an unchanging image over many hours with only 10% of the injected dose excreted in the urine. It provides an accurate image of renal parenchymal function and DRF.

Indications All indications are to detect a focal renal parenchymal abnormality: • in UTI, renal involvement in the acute phase can be diagnosed by 99mTc-DMSA • exclusion of a scar requires a normal 99mTc-DMSA study • when only one kidney has been identified and doubt exists about the presence of a second kidney • in the follow-up of pyelonephritis • assessment of focal parenchymal abnormalities in the presence of hypertension • in children with urinary incontinence and a suspected duplex kidney with ectopic drainage of the upper moiety (this may be a difficult diagnosis)

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

• in children with gross dilatation (e.g. prune belly syndrome)

where assessment of drainage is not possible but DRF is required. There are no contraindications.

Method Posterior and both posterior oblique projections are routine. Anterior views including the empty bladder are essential when an ectopic kidney is to be excluded.The child is kept still with sandbags and Velcro straps. Sedation is almost never used.

UROGRAPHY (PLAIN RADIOGRAPH AND INTRAVENOUS UROGRAM) A plain abdominal radiograph is essential before every IVU, especially in the presence of renal calculi. In the presence of chronic renal failure the femoral epiphyses may slip and renal osteodystrophy should be sought. IVU is now an infrequent examination and may be replaced by MRI in time.

Indications The indications are: • for suspected renal calculi • in suspected autosomal recessive polycystic kidney disease • when an occult duplex kidney is considered, especially if US and 99mTc-DMSA imaging are normal • to define ureteric anatomy in the context of primary enuresis • to define ureteric anatomy in known duplex kidneys • if a small kidney is discovered on US or isotope examination and no VUR is found, then an IVU to show the calyceal anatomy may prove helpful in establishing the cause, e.g. previous ischaemia • occasionally in the post-transplant setting to demonstrate ureteric anatomy.

Method The IVU should be individualized to answer specific clinical questions. To evaluate calyceal anatomy, images approximately 5 min after the injection of contrast medium should be obtained in the coned AP as well as the AP supine renal window view (caudal angle 34 degrees approximately centred on the xiphisternum) as a routine. The latter view will project the bowel gas over the lower abdomen. A US study rather than a post-micturition radiograph is recommended to assess bladder emptying.

COMPUTED TOMOGRAPHY CT generally has a limited but specific role in the evaluation of the paediatric genitourinary tract. The recommended dose of standard nonionic contrast medium, e.g. iohexol 300, for most abdominal studies is 2 ml kg−1.

Indications The indications are: • all patients with a suspected malignant renal or pelvic lesion or suspected abscess after intravenous contrast enhancement

1539

1540

SECTION 8

•

PAEDIATRIC IMAGING

• follow-up of known malignant tumours, although MRI is now

Method

regarded as equally reliable in assessing abdominal masses • major abdominal trauma with suspicion of serious pelvic injury or fracture, or if haematuria is present or bladder rupture is suspected • less common indications may include a suspicion of chronic renal infection or tuberculosis, or of nephrocalcinosis when a US study is inconclusive • CT angiography (CTA) may replace conventional angiography in certain circumstances, e.g. assessment of posttransplant arterial complications or in the evaluation of a suspected large vessel aneurysm.

Coronal imaging produces images akin to an IVU. Contrast enhancement is necessary to detect acute pyelonephritis and roughly to predict tumour vascularity. MRU with heavily T2weighted techniques can be performed as a single-shot, fast spin-echo sequence. If upper tract obstruction is suspected, a prolonged dynamic series with intravenous frusemide after gadolinium injection may eventually replace dynamic renography. During post-processing, signal intensity versus time, similar to a renogram curve, can be plotted utilizing a ROI over the renal pelvis. Additional anatomical information, such as renal pelvis diameter in response to frusemide administration, can be measured simultaneously. MRA techniques employing gadolinium enhancement, rather than nonenhanced time-of-flight or phase-contrast sequences, are more robust and reproducible in the evaluation of the renal vasculature.

Method For the majority of studies intravenous contrast enhancement is so crucial that nonenhanced images are often noncontributory, unless calcification or calculi specifically are being sought. With the new multidetector systems, a low milliampere setting, ideally based on weight, to reduce dose is essential in small patients. In cases of abdominal trauma, delayed imaging after an interval of 10–15 min can be a useful method of detecting extravasation of contrast medium from the genitourinary tract.

MAGNETIC RESONANCE IMAGING All pelvic masses are best assessed with MRI. Spinal MRI is warranted in patients with a suspected spinal cord tethering or a neuropathic bladder. Fat-suppressed gadolinium-enhanced sequences may be as sensitive as 99mTc-DMSA scintigraphy in the detection of acute pyelonephritis. The normal renal medulla on MRI has a longer T1 relaxation time than the cortex such that T1-weighted imaging yields good corticomedullary differentiation. Heavily T2-weighted sequences, e.g. rapid acquisition relaxation enhancement (RARE), can yield an MR urogram (MRU) that is comparable to the standard IVU in terms of anatomical detail. MRU can be performed rapidly, independent of renal function. There is as yet little literature regarding the accuracy of MRA in the assessment of the renal vasculature in children, but it is likely to have a role to play in assessing the first- and second-order branches. Simple and complicated cysts, and haemorrhagic or fat-containing lesions, can be accurately diagnosed with MRI. MRI can replace CT in the assessment of a Wilms’ tumour; however, thoracic CT is needed to exclude pulmonary metastases.

Indications The indications are: • children with a neuropathic bladder or an abnormal spinal radiograph require exclusion of spinal cord pathology, which is best achieved with MRI • to stage the intra-abdominal extent of Wilms’ tumours • evaluation of pelvic tumours • gadolinium-enhanced MRA may have a future role in the assessment of hypertension but conventional arteriography currently remains the reference technique • future indications may include confirmation of renal involvement in acute pyelonephritis.

INTERVENTIONAL PROCEDURES Angiography Angiography with its high radiation dose and invasiveness is reserved for special clinical situations. It is best undertaken by an experienced operator.

Indications The indications are: • hypertension with a high suspicion of renovascular disease, including suspected vasculitis, especially polyarteritis nodosa. Renal vein sampling for renin values is useful in some cases to evaluate which kidney is causing the hypertension • before interventional procedures, e.g. embolization for arteriovenous malformations or balloon dilatation for renal artery stenosis • rarely now in bilateral Wilms’ tumours before surgery • testicular vein embolization for varicocele obliteration.

Method Angiography should use the smallest catheter compatible with good results. Selective arteriography with magnification and oblique views is essential to look for lesions in small renal vessels.

Antegrade pyelogram This investigation should be carried out by an experienced operator in the radiology department or in theatre, usually before surgery, to provide anatomical detail of the renal pelvis and/or ureter unavailable from US or IVU. Occasionally antegrade studies are combined with pressure flow measurements with or without urodynamic studies to determine the physiological significance of a dilated upper urinary tract.

Nephrostomy The placement of a pigtail catheter in a dilated renal pelvis or ureter should be undertaken by an experienced operator under US guidance. The technique in children is similar to that in adults. US guides the needle insertion into, preferably,

CHAPTER 66

a lower pole calyx for contrast delineation of the collecting system. The nephrostomy tube is then placed appropriately. Different sized drainage catheters are used depending on the child’s size and need for drainage. The complications of placing a nephrostomy tube are generally those of catheter placement, extravasation of contrast medium, and leakage of urine.

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Retrograde pyelogram The instillation of dilute contrast medium into a ureter via a catheter inserted into the distal ureteric orifice is usually undertaken by a urologist in the operating theatre. With modern flexible ureteroscopes, the contrast medium may be instilled into the upper ureter or even the renal pelvis. This is done to outline the ureter and its drainage.

INTEGRATED IMAGING THE NEONATE The immature kidney of the neonate is different from that in the older child, both in its US appearance and also in the handling of radio-isotopes or contrast medium. There is difficulty in disposing of an osmotic and/or a sodium load and consequently a greater risk of volume overload with excess intravenous contrast medium. In the neonate and infant there is a relatively large extracellular fluid space so that readily diffusible substances injected intravenously (99mTc-DTPA or iodinated agents for IVU or CT, or gadolinium for MRI) have a large volume of distribution and therefore a relatively low plasma concentration.

Normal imaging High-quality dynamic isotope studies are generally best achieved when some renal maturity has occurred (at about 4 weeks of age). Renal immaturity dictates the use of 99mTcMAG3 since this isotope remains in the blood pool rather than DTPA. Renal immaturity precludes the exclusion of a minor parenchymal abnormality in the first months of life. However, in the presence of renal failure and/or bilateral dilatation, 99mTc-DMSA imaging has proved very helpful in deciding whether both kidneys are equally involved in the disease processes or if all the function is from only one kidney. There is no minimum age in such a clinical situation to undertake 99mTc-DMSA scintigraphy. There is no indication for an IVU in the neonatal period. An IVU in the neonatal period may aggravate renal venous thrombosis (RVT) or medullary necrosis. Even in normal neonates, an IVU may fail to outline the kidneys, especially in the first 48 h of life. MRU has not been fully explored in the newborn.

Prenatal diagnosis of renal/urological abnormality Newborns with a prenatal US diagnosis of a renal tract abnormality do not form a homogeneous group. Postnatal US in the neonatal period guides further imaging following an abnormal prenatal US. A fetus at 18–20 weeks’ gestation with a renal pelvis of 5 mm or greater, which enlarges or remains static, is generally referred for investigation postnatally. In many situations the natural history is unclear and therefore the question arises in many asymptomatic infants whether any surgical correction should be undertaken. Management is complex so a simple attitude cannot be adopted from the data in the literature. An absolute indication for surgery is unequivocal obstruction, e.g. complicated ureterocele, posterior

urethral valve, or major infection complicating a hydronephrosis.

Differential diagnosis The infant with one normal kidney and contralateral hydronephrosis must be distinguished from the infant with bilateral hydronephrosis with or without a normal bladder. Postnatal imaging is heavily dependent on the prenatal US assessment of the urinary tract in respect of how quickly postnatal imaging should be undertaken. The full differential diagnosis is listed in Table 66.1. The most common referral is that of mild pelvic dilatation seen in early pregnancy at around 18 weeks’ gestation: postnatal US is usually normal in these cases. In more specialized centres, isolated unilateral renal pelvic dilatation (RPD) is a more common referral (see section on cystic disorders below for a description of multicystic dysplastic kidney).

Unilateral renal pelvic dilatation RPD or hydronephrosis may be defined as calyceal dilatation plus RPD of greater than 10–15 mm in its AP diameter with no US evidence of a dilated ureter5.This was referred to loosely in the past as PUJ obstruction or PUJ stenosis. On antenatal US the AP renal pelvis is greater than 50% of the longitudinal length of the kidney. This equates to approximately a 5-mm

Table 66.1 DIFFERENTIAL DIAGNOSIS OF PRENATAL HYDRONEPHROSIS Unilateral pathology Renal pelvic dilatation (RPD) Vesico ureteric reflux (VUR) Megaureter (with or without reflux) Multicystic dysplastic kidney Complicated duplex kidney Upper moiety dilatation—either ureterocele or ectopic drainage Lower moiety dilatation—usually VUR but rarely RPD only

Bilateral pathology Bilateral renal pelvic dilatation Bilateral VUR Bilateral megaureter (with or without reflux) Bladder pathology, e.g. neurogenic bladder Bladder outlet pathology (posterior urethral valves) Bilateral complicated duplex kidneys Multicystic kidney on one side and cystic dysplastic kidney on opposite side

1541

1542

SECTION 8

•

PAEDIATRIC IMAGING

pelvis at 20 weeks’ gestation and a 10-mm pelvis in the third trimester. RPD is commonly unilateral but may be bilateral; the importance of this distinction is that if it is unilateral then there is no urgency in terms of investigation. All these infants are asymptomatic. Only about 25% now undergo surgery. Natural history studies suggest that RPD is a relatively benign condition and help define how to investigate and manage these children. Obstruction is not easily defined. The effects of obstruction are hydronephrosis, parenchymal atrophy, and impaired renal function. These changes do not define or predict the potential for progressive renal deterioration. The only accepted diagnosis of obstruction is that if nothing is done either function deteriorates or dilatation increases. If the function and dilatation remain unchanged then that kidney is not obstructed and is in equilibrium. These children with RPD should not be confused with those who have symptoms of loin pain and intermittent PUJ obstruction.

Imaging protocol There is no available test that will predict which kidney with a prenatal RPD will deteriorate if left alone. A US examination should be undertaken 48–72 h after birth to assess calyceal and pelvic dilatation, measure the transverse diameter of the pelvis, and confirm the structural normality of the bladder and opposite kidney. With a secure diagnosis of unilateral calyceal and RPD dilatation, a diuretic MAG3 renogram and follow-up US are required. The exact timing of follow-up will depend on the size of the RPD and DRF. Marked RPD or poor function require closer follow-up than 15–20-mm RPD and/or normal DRF. With stable imaging follow-up is less frequent. Long-term follow-up is recommended until at least 15–20 years of age. There is no role for an MCUG in these children.

reflux stops spontaneously within 2 years in 20%. Postnatal US is normal in the vast majority. Between 30% and 50% of these children had reduced function on isotope renography; however, following UTI this increased to 71% (Fig. 66.9)7. This is the rationale for the use of prophylactic antibiotics in this group of infants. These data suggest that there is a developmental abnormality resulting in both VUR and an abnormal kidney. If the initial US is normal and the child is on chemoprophylaxis then it may be argued that nothing more need be done from an imaging aspect. If the US is abnormal, then further imaging is guided by the abnormality seen.

Megaureter Hydronephrosis with a dilated ureter is dealt with in the same way as pelvic dilatation above (Fig. 66.10). The investigative protocol is similar to that of RPD plus a MCUG to document or exclude VUR. The US must in addition focus on the bladder and vesico-ureteric junction since it is imperative to exclude a small ureterocele or other bladder abnormality. Usually the bladder appears normal on both US and MCUG, in which case there may be a functional abnormality. Management is also complex as some surgeons believe that re-implantation of a ureter into the bladder in a child younger than 1 year of age may result in abnormal bladder function in later life. The role of urodynamics remains uncertain. With a suspected neuropathic bladder, a spinal US in early infancy can exclude spinal cord pathology. With increasing dilatation or fall in function on diuretic 99m Tc-MAG3 studies, the insertion of a double J stent from the ureter into the bladder requires an antegrade study. Follow-up imaging is with US and diuretic 99mTc-MAG3 studies post stent insertion, as well as after stent removal.

Bilateral renal pelvic dilatation

Renal failure

There is uncertainty about the management of bilateral RPD. When and on which side to operate is uncertain and how to judge the results of US and diuretic 99mTc-MAG3 is very difficult. The investigative protocol in this situation should be as for the unilateral RPD dilatation, but here includes an MCUG as well as formal sequential glomerular filtration rate (GFR) estimation. Some surgeons will opt for a nonsurgical approach as long as the degree of dilatation and differential function are stable, while others prefer to operate on the better functioning kidney, and yet others will operate on the worse kidney or the one with the greater dilatation. Until more data become available, no rigid approach can be recommended.

The neonate may be in acute renal failure due to an episode around the time of labour or acquired sepsis resulting in RVT, medullary or acute tubular necrosis (ATN), or any combination of these conditions. The US should identify the size of each kidney, and the echogenecity of the cortex and medulla compared to the liver and spleen. All these disorders result in an echogenic kidney with loss of the normal corticomedullary differentiation. In ATN the kidneys are usually symmetrically abnormal. In RVT they are asymmetrically involved, swollen, with poor corticomedullary differentiation; the interlobular arteries are prominent and often no venous flow can be seen at the renal hilum or in parts of the swollen kidney on Doppler interrogation (Fig. 66.11). Thrombus in the inferior vena cava (IVC) is well recognized. There may be adrenal haemorrhage in this clinical setting, readily seen on US (typically left-sided as the left adrenal vein drains into the left renal vein). The pathogenesis in RVT is dynamic so the US findings change depending on the stage of the pathology. RVT can ultimately result in a small scarred kidney.To assess the long-term damage 99m Tc-DMSA scintigraphy (between 4 and 12 months of age) should be performed.

Normal postnatal ultrasound or mild dilatation This is the largest group of children seen with a prenatal US diagnosis of hydronephrosis. Early studies show that the MCUG revealed no bladder obstruction and a normal bladder wall thickness, but VUR may be present with a marked male preponderance6. Grade IV or V (see Fig. 66.20 for grading)

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

*10 97

Counts/sec

Bladder 49

Left kidney

Right kidney 0

0

F

55 Sec

110

Figure 66.9 Antenatal diagnosis of bilateral hydronephrosis. Due to vesico-ureteric reflux (VUR). This diagnosis was confirmed postnatally. (A) The micturating cystourethrogram (MCUG) shows bilateral VUR with a normal bladder. (B) The calices and pelves on the MCUG appear normal. There was no evidence of bladder outflow obstruction. (C) 99mTc-DMSA scintigram shows a normal left kidney with a small scarred right kidney. The left kidney contributed 76% to overall function and the right kidney 24%. The left kidney took up 34.6% of the injected dose and the right kidney 9.3%. (D) Indirect radio-isotope cystogram (IRC) with 99mTc-MAG3 at age 3 years. The radionuclide may be noted in the bladder and there is still some in both kidneys as well as in the right ureter. This image was obtained 1 h after the intravenous injection of 99mTc-MAG3. (E) IRC at the end of micturition shows the bladder to be almost empty and the ureters empty, but marked activity is noted in both kidneys. (F) The curves show prompt bladder emptying with a bilateral renal reflux of 5 ml into the right kidney and 4 ml into the left kidney. The bladder volume was 55 ml with a voided volume of 40 ml and a urine flow rate of 2 ml s−1.

Associated congenital abnormalities Children with anorectal anomaly, tracheo-oesophageal fistula, or VACTERL association have a high incidence of renal abnormalities such as crossed fused ectopia and a single kidney. All require US, with further imaging dictated by the US findings.

BLADDER ANOMALIES Exstrophy/epispadias Exstrophy and epispadias represent a spectrum of anomalies. Boys are affected three times as frequently as girls.The diagnosis is clinically apparent in cases of exstrophy and in epispadias in the male infant. Female epispadias is rare and may present as an occult cause of enuresis. A measurement of pubic symphysis diastasis can be used as a radiological criterion to estimate the degree of abnormality present in this spectrum. Diastasis of only several millimetres more than the normal 10 mm

suggests epispadias, whereas separation of the pubic bones by more than 25 mm is characteristic of exstrophy (Fig. 66.12). The genital structures are normal, as are the upper urinary tracts initially, though minimal distal ureteric dilatation may be present at birth. Subsequently, following surgery, complications such as VUR or obstruction (which may be silent) must be sought, particularly with any type of diversion. Additionally, the incidence of adenoma or adenocarcinoma at such anastomoses is higher than would be expected, so regular evaluation is important and this is best done using a combination of US and MAG3 to assess the ureters and bladder. Loopograms may also be required following diversion.

Prune belly syndrome (prune belly triad syndrome; abdominal musculature deficiency syndrome) The prune belly syndrome (PBS) is generally readily apparent clinically and consists of absent abdominal wall muscles, undescended testes, and renal dysplasia associated with gross dilatation of the collecting systems. The incidence of the condition

1543

1544

SECTION 8

•

PAEDIATRIC IMAGING

Figure 66.10 Primary megaureter demonstrated on an intravenous urogram. The entire right ureter is dilated, particularly the distal ureter. There is little dilatation of the upper collecting system.

has fallen due to antenatal diagnosis with subsequent termination of pregnancy. Both the pathogenesis and optimal therapy in management are debated. All affected children have a universal mesentery with a mobile bowel. Plain abdominal radiographs show a protuberant abdomen resulting from the lack of abdominal musculature. In the most severely affected patients, there may be urethral obstruction with severe renal dysplasia, while less severely affected individuals demonstrate an array of urinary tract abnormalities. The characteristic features are small kidneys with abnormal minimally dilated calyces and upper ureters, while the lower ureters are tortuous and show disproportionate dilatation. The bladder is thin walled, of large capacity without trabeculation, and has a wide neck. A patent urachus or a urachal diverticulum may be present. The posterior urethra is dilated proximally with a typical conical narrowing and a poor stream in the distal urethra (Fig. 66.13). Occasionally, generalized or focal anterior urethral dilatation is present. Prognosis depends on the degree of renal dysplasia.

Imaging US demonstrates the small lobulated kidneys and upper tracts, but with marked lower ureteric dilatation and a full bladder it may not be able to ‘sort out’ the lower urinary anatomy. 99m Tc-DMSA is valuable for DRF. IVU is rarely indicated

Figure 66.11 Renal vein thrombosis. (A) Longitudinal US of the right kidney in a neonate with haematuria and a palpable kidney. The kidney is large and swollen with an overall increase in the echogenicity and some prominent interlobular vessels. (B) Longitudinal US of the inferior vena cava (IVC). This shows thrombus in the IVC (between calipers). (C) Longitudinal US of the right kidney 6 months later. The kidney is small and echogenic with bright calcified intrarenal vessels.

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Neurogenic bladder

Figure 66.12 Pubic symphysis diastasis. A neonate with a cloacal anomaly. The apparent mass arising out of the pelvis was a hydrocolpos.

Children with this disorder commonly have a deficiency of appropriate detrusor muscle contraction and external sphincter relaxation, resulting in an uncontrolled voiding pattern with incomplete bladder emptying. Meningomyelocele is the most common associated abnormality with significant secondary musculoskeletal and bowel consequences. Defects in the sacrum can also result in a neurogenic bladder. Other abnormalities included under the broad term ‘myelodysplasia’ (including sacral agenesis) may have similar urological implications without obvious musculoskeletal abnormalities.When no spinal abnormality is found, the term non-neurogenic neurogenic bladder is used. Minor degrees of spinal dysraphism, such as spina bifida occulta at the lumbosacral junction, are not associated with neurological defects. The goals in the management of neurogenic bladder are preservation of renal function and attainment of continence. Initially the upper tracts are normal and deterioration of function results from poor bladder emptying with complications of infection, reflux, and hydronephrosis. Current management focuses on regular clean intermittent catheterization, the use of pharmacological agents, and occasionally the use of an artificial urinary sphincter. US, MCUG, and MAG3 scintigraphy are indicated at diagnosis and follow-up. Classification of neurogenic bladder is mainly based on urodynamic results, which have also contributed significantly to the management of the condition. On MCUG the neurogenic bladder may be vertically orientated, thick-walled, and trabeculated with diverticular formation, but mild changes may also be seen early on. Narrowing of the urethra at the external sphincter may be shown and the remainder of the urethra is normal. Spinal US should be done in infants younger than 6 months of age. On plain radiographs it may be difficult to identify a spinal abnormality in the infant. MRI is the best technique to ensure a small mass lesion is not overlooked.

URETHRAL ANOMALIES

Figure 66.13 Prune belly syndrome. Image exposed near the end of micturition on a voiding cystourethrogram. The posterior urethra is dilated proximal to the membranous urethra and the calibre of the latter is normal. Posterior urethral valves are not present.

Delineation of the bladder neck and urethra is best achieved on the MCUG in the oblique projection. In the presence of bladder neck hypertrophy due to urethral obstruction, the posterior aspect of the bladder neck is easy to identify but the anterior boundaries of the bladder neck have no readily demonstrable landmark. When the bladder empties, a contraction ring may be seen, which looks like the bladder neck to the unwary. For these reasons it is often difficult to be certain where the bladder neck finishes and the posterior urethra begins.

Obstruction to bladder outlet and dynamic renal scintigraphy fails to show adequate drainage due to the gross dilation. MCUG is valuable for urethral anatomy but is dangerous if sepsis arises.VUR occurs in about two-thirds of cases.The long-term follow-up is a combination of US and 99mTc-DMSA imaging at intervals, with MCUG being required if bladder outlet symptoms develop.

Anatomical causes of outlet obstruction include posterior urethral valve, urethral dysplasia as seen in certain syndromes (e.g. PBS), anterior urethral valve/diverticulum, and meatal stenosis, as well as post-traumatic strictures of the urethra at any level. Functional obstruction may be secondary to a spinal abnormality and the occult neuropathic bladder.

1545

1546

SECTION 8

•

PAEDIATRIC IMAGING

Intrinsic masses obstructing bladder outflow These include rhabdomyosarcoma of the bladder, vagina, or prostate, and benign tumours such as neurofibroma, hamartoma, and haemangioma. Occasionally, a polyp on a stalk originating in either the posterior urethra or the bladder may cause an obstruction. All these mass lesions may prolapse into the urethra and so cause obstruction. The larger lesions will all be readily identified on US while smaller lesions require an MCUG using dilute contrast medium.

Male urethra Congenital abnormalities affect both the anterior and posterior portion of the urethra, the most common abnormality being hypospadias, which has little radiological importance.

Posterior urethral valves (Fig. 66.14) Rarely these children present with severe UTI and the diagnosis of septicaemia is now more commonly made in early infancy or during fetal development. The presence of bilateral hydronephrosis, a full bladder, and progressive oligohydramnios during pregnancy may be an indication for drainage of the bladder during pregnancy. This needs to be done early if the complications of oligohydramnios, such as pulmonary hypoplasia, are to be prevented. Posterior urethral valves (PUVs) are categorized into either three or four types, depending on the classification used. Detailed descriptions are inappropriate in context, but Types 1

and 3 merit brief mention.Type 1 is the most common, where two folds at the level of the verumontanum have a ventral central slit-like orifice. Type 3 has a pin-point, eccentric orifice which results in forward ballooning of the valve giving a ‘wind in the sail’ appearance on the oblique projection during MCUG. With Type 3 valves there is a high association of renal dysplasia with minimal upper tract dilatation and there may be little improvement in renal function following ablation of this type of valve. The mild degree of dilatation of the collecting systems throughout the natural history of the condition is out of keeping with the degree of chronic renal failure. A rare complication of PUV in the neonatal period is rupture of a calyceal fornix, leading to accumulation of urine (urinoma) around the kidney limited by fascial planes. This urinary leak may result in urinary ascites which has a significant generalized metabolic effect. In male infants the presence of dilated ureters with or without a thick-walled bladder dictates the need for an MCUG to exclude the diagnosis of PUV. US may detect the bladder neck hypertrophy, which is a constant feature of the PUV. On MCUG the visualization of the valve may be difficult unless the correct oblique projection is obtained during micturition. Other signs include dilatation of the posterior urethra proximal to the valve, with a disparity between the respective calibres of the proximal and distal urethra. Bladder neck hypertrophy is invariable, the bladder is usually thick walled and of variable capacity (small or large), and trabeculation may be an early feature. VUR may be seen and is generally associated with poor

Figure 66.14 Bilateral hydronephrosis secondary to posterior urethral valves. (A) Longitudinal US of the left kidney shows hydronephrosis with a bright renal parenchyma. (B) Longitudinal US of the right kidney reveals a similar hydronephrosis to that seen on the left. (C) Longitudinal US of the bladder reveals a thick-walled bladder with a dilated ureter behind the bladder. (D) Transverse US of the bladder once again shows the thick-walled bladder with dilated ureters behind the bladder. Continued

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Figure 66.14 Cont’d (E) Micturating cystourethrogram (MCUG) done on the first day of life reveals bladder neck hypertrophy with a dilated posterior urethra. There is an abrupt change in calibre of the urethra, although the exact outline of the posterior urethral valve cannot be seen. Note that with a 6F feeding tube in the urethra the possibility of obscuring a posterior urethral valve was very small. Reflux into a ureter was also clearly seen. (F) MCUG 3 months after valve resection shows that the bladder neck hypertrophy is less marked and that the differential calibre of the urethra is minimal. Reflux persists; this is on the left. (G) 99mTc-DTPA scintigram. The 1-min image reveals function only in the left kidney. (H) The 20-min image shows progressive accumulation of radionuclide in the left kidney with no function in the right kidney. (I) 10 min after intravenous furosemide. The radionuclide has moved from the dilated left renal pelvis down the ureter into the bladder. (J) Following micturition the bladder has emptied and reflux is clearly seen into the nonfunctioning right kidney. The left kidney as well as the ureter have cleared the radionuclide, suggesting that no obstruction exists.

renal function on the affected side; VUR may cease following valve ablation.

Imaging US usually suggests the diagnosis, typically showing bladder wall thickening with or without hydroureters and echogenic kidneys. US may show a urinoma. The MCUG is definitive. These two examinations allow comprehensive immediate management. A baseline, 99mTc-MAG3 scintigram should be carried out around 12 weeks following valve ablation. Long-term follow-up includes revisualization of the urethra on MCUG after the last instrumentation, as well as US and dynamic diuresis renography at regular intervals. After the age of 3 years the dynamic MAG3 renogram should be completed with an IRC so that the effect may be judged of a full bladder and bladder emptying on the upper tract.

Urethral dysplasia The entire urethra may be dysplastic and this feature is generally associated with dysplasia of the kidneys and consequent renal failure within the first year of life. The urethra may be difficult to catheterize, even using a 6F catheter, and suprapubic cystography may be required to outline the urethra on MCUG. On fluoroscopy, a thin line of contrast medium is seen where the urethra should be, without the normal anatomical landmarks. The entire urethra is of the same calibre. The prognosis depends on the degree of renal dysplasia. In Down’s syndrome narrowing of the distal posterior urethra, not dissimilar to that seen in PBS, is seen. However there is a good stream. The narrowing is concentric and does not act as an obstructive lesion since the children can empty their bladder completely and there is no bladder neck hypertrophy.

1547

1548

SECTION 8

•

PAEDIATRIC IMAGING

In the presence of a single ectopic ureter, the ureter enters below the bladder base and drains a simple (i.e. nonduplex) kidney. There is elongation and narrowing of the posterior urethra between the bladder neck and verumontanum.The ureter is not usually obstructed and may be of normal calibre. The effect on the urethra is variable but is usually obstructive to some degree.

Recto-urethral fistula This abnormality is usually associated with an imperforate anus. A high-pressure loopogram via a colostomy is the best method of demonstrating the passage of contrast medium from the bowel to the posterior urethra (close to the verumontanum). The presence of air in the bladder as shown on plain images confirms the diagnosis of a fistula. Direct rectovesical fistulas are exceedingly uncommon.

Duplication of the urethra This is a rare abnormality in which there is usually complete duplication, with one of the urethras ending as a hypospadias (Fig. 66.15). Pathology within one of the urethras, e.g. a posterior urethral valve, must also be excluded.

Anterior urethral abnormalities This area has been neglected by both radiologists and urologists. High-quality images are essential to exclude a syringocoele or meatal abnormalities.

Traumatic strictures Strictures may be found at any level in the urethra following either instrumentation or external trauma.

and/or obstruction due to a ureterocele (Fig. 66.16).The lower moiety may have reduced function in association with reflux into this moiety or due to obstruction at the PUJ level.

Drainage Incomplete duplication Drainage of the duplex kidney is by definition via two collecting systems. However, the two systems have many possible formations.There may be incomplete duplex, in which the two ureters join at any level above the bladder.This is when ‘yo–yo’ reflux may occur, i.e. when urine refluxes from one moiety down the ureter and then up into the other moiety, rather than going down into the bladder. It is only by using dynamic radionuclide studies that this diagnosis can be made with certainty.

Complete duplication Here the ureters draining the two moieties never join. When both ureters drain into the bladder, the ureter draining the lower moiety is prone to reflux. The ureter draining the upper moiety usually enters the bladder as a ureterocele and is frequently obstructed.The ureterocele may vary in size from being so large that the inexperienced may mistake it for the bladder or it may be collapsed such that it is not recognizable even on a careful US examination. When one moiety drains outside the bladder, it is usually the upper moiety and this ureter can terminate in the urethra or in the vagina in girls. Such ectopic drainage of the ureter is almost always associated with dysplastic function of the upper moiety of the kidney (Fig. 66.17).

Clinical presentation

DUPLEX KIDNEY

• Asymptomatic as an incidental finding • UTI • Pain which can be secondary to an intermittent obstruction

An uncomplicated duplex kidney is a variation of normality. A complicated duplex may be a complete or incomplete duplex. Both moieties may show equal function, and here the duplex kidney is typically longer than normal in bipolar length on US and has divided renal function of over 55% of global function. Reduced function of the upper moiety may be due to dysplasia

at the PUJ level of the lower moiety or due to ‘yo–yo’ reflux with incomplete duplication • Continuous wetting in the girl who has never been dry due to an ectopic insertion of the upper moiety into the vagina • Vaginal prolapse may occur when the ureterocele prolapses out of the bladder. Prolapse of a ureterocele may also result in bladder neck obstruction and may mimic PUV in boys.

Imaging One of the cardinal signs of a duplex system is a change in the axis of the lower moiety. This is best seen in that the calyces of the lower moiety show the lower group medial to the upper group of calyces, giving the lower moiety of the kidney a longitudinal axis pointing to the shoulder on the same side.

Ultrasound

Figure 66.15 Infant with hypospadias. The urethral views of the micturating cystourethrogram demonstrate a double urethra; the dorsal urethra is severely dysplastic and the main channel is through the ventral urethra terminating in the hypospadias.

The US findings are protean depending on the abnormalities present. The upper moiety may be normal, small, and dysplastic, or may be anechoic resembling a ‘cyst’ which is an obstructed ureterocoele. The latter two conditions are generally associated with a dilated ureter.The appearances of a lower moiety in complete duplication may be normal and difficult to recognize or the kidney may simply show two distinct renal pelves.The calyces and pelvis of a lower moiety may be dilated with no ureteric dilatation, which would suggest a PUJ stenosis. Reflux is likely when a dilated ureter is seen.

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Figure 66.16 Ectopic ureterocele. Diagrammatic representation of the anatomical and urographic appearances of an ectopic ureterocele of the left upper moiety without function. Diagnosis of this entity on the urogram depends on recognition of indirect signs: 1 = increased distance from the top of the visualized collecting system to the upper border of the nephrogram; 2 = abnormal axis of the collecting system; 3 = impression upon the upper border of the renal pelvis; 4 = decreased number of calyces compared to the contralateral kidney; 5 = lateral displacement of the kidney and ureter; 6 = lateral course of the visualized ureter; and 7 = filling defect in the bladder. (Redrawn from Lebowitz R L, Avni F E 1980 Misleading appearances in pediatric uroradiology. Pediatr Radiol 10: 15–31.)

Figure 66.17 Duplex anomaly. Anomaly in a 1 year old girl with urinary tract infections. (A) Longitudinal US of the left kidney showing a dilated collecting system in the upper portion, with a normal lower pole. (B) The micturating cystourethrogram shows vesico-ureteric reflux (VUR) into a dilated ureter, which is draining the lower moiety of the left kidney; a small amount of contrast medium is seen in the bladder but no ureterocele is evident. (C) 99mTc-DMSA scintigram in the posterior (prone) projection shows a normal right kidney; the left kidney shows decreased activity in the upper pole with a little abnormal activity on the superomedial aspect. The left kidney contributed 38% to overall function and the right kidney 62%. (D) Left posterior oblique projection of the 99mTc-DMSA scintigram showing to better advantage the defect in the upper pole of the left kidney. This infant had a duplex left kidney with a ureterocele draining the obstructed upper moiety. VUR into the lower moiety had not caused any scarring.

Nuclear medicine The appearances of a duplex kidney vary with the pathology of each moiety. 99mTc-MAG3 will assess function, drainage, and/or reflux, especially with late images. If a moiety is nonfunctional it will not be visualized; this is important when

there is a small severely dysplastic upper moiety. A high index of suspicion when reporting on functional imaging allows the duplex kidney to be recognized easily. With incomplete duplication the upper and lower moieties may be normal or there may be reduced function of either element. With

1549

1550

SECTION 8

•

PAEDIATRIC IMAGING

99m

Tc-MAG3 the drainage from each moiety can be evaluated as well as looking for ‘yo–yo’ reflux.

maldeveloped kidney which previously was considered to be ‘reflux nephropathy’ (Fig. 66.18).

Micturating cystourethrogrophy

Imaging

If a duplex is suspected the technique must be modified; fluoroscopy begins when a small volume of contrast medium has been instilled into the bladder to allow detection of the ureterocele as a filling defect; this will be obliterated once the bladder is full of contrast medium. US, however, is the preferred method for detecting and measuring ureteroceles.VUR may be detected, usually into the lower moiety. A nonrotated frontal image centred on the kidneys is essential. Reflux into the upper moiety is rarely detected.

The aim of imaging is to detect an obstructive uropathy or calculi, and to prevent further infections. A full US examination as soon as possible after diagnosis in the susceptible group is recommended. If the US is normal or abnormal but without dilatation, then a 99mTc-DMSA scintigram is required. If the US shows hydronephrosis a MAG3 diuretic renogram is required. When US shows a dilated ureter or an abnormal bladder then

Intravenous urogram This is only indicated when there is a strong clinical suspicion of a duplex as, e.g. in the girl who is always wet, yet the US and 99mTc-DMSA studies are normal. Since the ureter drains below the bladder base in this situation, there is no point in doing an MCUG. An IVU with special emphasis on the upper poles of the kidneys, including delayed images, may be useful. The only sign of a duplex system may be a little contrast on a late radiograph medial to the upper pole of one of the kidneys. MRU may now be superior in this context.

URINARY TRACT INFECTION AND VESICO-URETERIC REFLUX UTI is common and the imaging that children undergo is important in terms of discomfort, radiation exposure, and resource utilization. An effort must be made not to overburden the radiology department or over investigate the child but at the same time not to overlook treatable pathology or long-term complications. Imaging should therefore be limited to those children with a bacteriologically proven UTI.

Clinical setting Four per cent of all children under 8 years of age will present with a UTI each year, yet the incidence of children requiring dialysis secondary to pyelonephritis is one child per million age-related population. There are little data to support the belief that hypertension is a complication of mild renal scarring. For these reasons the ‘susceptible child’ who has a UTI and requires imaging needs to be identified.These children include those with: febrile infection; recurrent infections; clinical signs (poor urinary stream, palpable kidneys); unusual organisms (non-Escherichia coli) infections; bacteraemia; slow response to antibiotics; or unusual clinical presentation (older boy). VUR is a sign detected in the radiology/nuclear medicine department; it is not a disease. The relationship between VUR and renal damage has been questioned8. A systematic review of the literature in children with UTI showed that only 20% of those with VUR showed renal damage on DMSA, while scarred kidneys were seen in children where no VUR could be demonstrated9,10. Further, 30% of children with a prenatal diagnosis of hydronephrosis already have an abnormal kidney at birth, thus recognizing that a child may have VUR and a

Figure 66.18 Antenatal detection of renal pelvic dilatation (RPD). This baby had antenatal US detection of RPD and this is the postnatal imaging. Longitudinal US showed that the right kidney is larger than the left. Both kidneys had irregular outlines with areas of cortical thinning. (A) Micturating cystogram on the same child showing bilateral reflux into tortuous dilated ureters and pelvicalyceal systems. (B) DMSA scintigram on the same child showing bilateral focal defects in the cortex of both kidneys with the right kidney functionally smaller than the left as seen on the US. This child had never had a urinary tract infection and the defects on DMSA are thought to be related to dysplasia during intrauterine development, which may be associated with vesico-ureteric reflux.

CHAPTER 66

a MCUG plus DMSA imaging is required. If US detects a calculus an abdominal radiograph is indicated (Fig. 66.19). There is growing evidence that gadolinium-enhanced fat-suppressed T1-weighted or inversion recovery (STIR) MRI sequences may be sensitive in the detection of acute pyelonephritic lesions; comparison with 99mTc-DMSA is being undertaken11. In children with recurrent UTI, emphasis must be on the bladder. Frequency void charts and the physiological IRC may help identify the unstable bladder. The follow-up of children with a damaged kidney and VUR who are on long-term antibiotic prophylaxis is unclear10. Imaging should be infrequent and as these children are older they should not have repeat MCUGs. One approach is to only undertake imaging if it will affect treatment, i.e. if a negative

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

isotope cystogram will be the indication to stop long-term prophylaxis. If surgical intervention has been undertaken to stop VUR (Fig. 66.20), then a US and dynamic scintigraphy are essential to ensure that no obstruction has resulted. Current recommendations on this topic from the Royal Colleges of Physicians and Paediatrics and Child Health, and the American Academy of Pediatrics are being seriously questioned10,12,13.

Renal abscess This complication follows acute pyelonephritis and may be first suspected when the child develops a swinging fever with a poor response to antibiotics. Usually the abscess also has a thick wall with heterogeneous internal echogenicity. A 99mTcDMSA scintigram will show a focal defect in a kidney which

Figure 66.19 Renal calculi. (A) Plain abdominal radiograph showing small opacities overlying the left kidney. (B,C) Longitudinal US of the left kidney showing the echogenic foci of the renal calculi casting an acoustic shadow in (B) a calyx and (C) also in the upper ureter. (D) Plain abdominal radiograph showing a left staghorn calculus in a different 4 year old boy.

1551

1552

SECTION 8

•

PAEDIATRIC IMAGING

Figure 66.20 Radiographic grades of reflux. (I) Ureter and upper collecting system without dilatation; (II) mild or (III) moderate dilatation of the ureter and mild or moderate dilatation of the renal pelvis, but no, or only slight, blunting of the fornices; (IV) moderate dilatation and/or tortuosity of the ureter with moderate dilatation of the renal pelvis and calyces and complete obliteration of the sharp angle of the fornices but maintenance of papillary impression in the majority of calyces; (V) gross dilatation and tortuosity of ureters, renal pelvis, and calyces; papillary impressions are not visible in the majority of calyces. (From Levitt S B, Duckert J, Spitzer A, Walker D 1981 Report of the International Reflux Study Committee. Medical versus surgical treatment of primary vesico-ureteral reflux. Reproduced with permission from Pediatrics 67: 392–400© 1981 by the American Academy of Pediatrics.)

generally shows reduced overall uptake of isotope. CT is helpful especially if aspiration or drainage is being considered.

Xanthogranulomatous pyelonephritis Xanthogranulomatous pyelonephritis (XPN) is a rare chronic inflammatory disease of the kidney in childhood. The clinical presentation is with weight loss, failure to thrive, malaise, anaemia, and often an abdominal mass. Urine cultures are positive in around 70% of cases with Proteus species and E. coli being the most common pathogens. Frequently a calculus is evident but XPN is not generally considered to be an infected obstructed kidney but rather a kidney in which there has been an abnormal inflammatory response to infection in the presence of a stone. The imaging findings are usually sufficiently characteristic to allow pre-operative diagnosis and to avoid confusion with a Wilms’ tumour. The diffuse type of XPN, affecting the entire kidney, is more common in childhood. Rarely focal involvement of only part of a kidney is seen. Focal XPN has a female predominance but the diffuse type has an equal sex distribution. The focal type of XPN tends to manifest with a localized intrarenal mass in an otherwise normal kidney. Plain radiographs reveal an abdominal mass with calculi. US demonstrates general renal enlargement in the diffuse form with hypoechoic areas corresponding to inflammatory masses within the kidney. Calcification, particularly at the contracted renal pelvis, is usually prominent. Ureteric calculi may also be visualized. 99mTc-DMSA scintigraphy shows a nonfunctioning kidney in the diffuse form or a photon-deficient area in the focal form. CT after intravenous contrast medium characteristically shows global renal enlargement with prominent low-attenuating abscess cavities. The remaining renal parenchyma can often show some rim enhancement due to the perfusion of the inflamed, nonfunctioning kidney (Fig. 66.21).

The necrotic areas tend to be hyperintense on T2-weighted MRI, with medium signal on T1-weighted sequences, which probably represents the high protein content of the cavities. Perinephric extension is common with hilar or para-aortic adenopathy. The treatment for both types is nephrectomy but this may be difficult due to the surrounding chronic inflammation.There may be a role for pre-operative embolization to reduce peri-operative haemorrhage.

HYPERTENSION The more severe the hypertension and the younger the child, then the more likely it is to be secondary hypertension. Renal disease is the cause of hypertension in over 90% of children after 1 year of age14. Any abnormal kidney may produce renin and so generate hypertension. Renal scarring and glomerular disease are the most common causes, and occasionally PUJ obstruction, neuroblastoma, or Wilms’ tumours present with hypertension. Renovascular disease accounts for approximately 10% of cases, with fibromuscular dysplasia being the most common cause. Other associations are neurofibromatosis, idiopathic hypercalcaemia of infancy, an arteritic illness, or middle aortic syndrome. Phaeochromocytomas, albeit uncommon in childhood, are seen and are often both multiple and extra-adrenal in origin. Essential hypertension is also encountered, usually in milder cases, often with a positive family history of hypertension.

Imaging US will demonstrate the small kidney, the severely scarred kidney, significant hydronephrosis, and both renal and most adrenal tumours. Doppler US of the aorta and renal arteries may reveal aortic narrowing or renal artery stenosis. Further

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Figure 66.21 Xanthogranulomatous pyelonephritis. (A) Longitudinal US of the right kidney in a 2 year old boy showing heterogeneous reflectivity due to xanthogranulomatous pyelonephritis. (B) Axial-enhanced CT study showing calcification at the right renal hilum, renal enlargement, and lowdensity fluid collections within the renal substance. Although there appears to be some enhancement of the renal parenchyma, there was no renal function on isotope studies.

investigation depends upon the abnormal US findings. Doppler US, however, has proved disappointing in the detection of renovascular disease in native renal arteries. A normal US does not exclude renal scarring, renovascular pathology, or a phaeochromocytoma. The differential diagnosis of a small kidney includes renal venous thrombosis, dysplasia, reflux nephropathy, postobstructive atrophy, and arterial pathology. With a normal US examination, a 99mTc-DMSA scintigram is the next recommended investigation. This can detect a focal parenchymal abnormality such as renal scarring. The major limitation of US and 99mTc-DMSA imaging is the possibility of unidentified renovascular disease or phaeochromocytoma. With biochemical evidence of a phaeochromocytoma imaging should be undertaken with a [123I]meta-iodobenzylguanidine ([123I]MIBG) scintigram followed by cross-sectional imaging, either CT or MRI, with specific attention focused on the areas of abnormal [123I]MIBG uptake.With MRI there is no risk of a hypertensive crisis as no iodinated contrast medium is used. To exclude renovascular disease, pre- and post-captopril MAG3 renography is recommended: this will exclude all surgical/angioplasty treatable causes. A negative study will not exclude middle aortic syndrome which requires angiography. Conventional arteriography demands a free flush aortic injection as well as selective renal injections in both AP and oblique projections in order to diagnose both intra- and extra-renal renovascular disease. While digital subtraction angiography is the current reference method, renal MRA with gadoliniumenhanced sequences is promising. However, CT and MRA cannot demonstrate the intrarenal vasculature adequately. It is noteworthy here that intrarenal vascular disease is more common in childhood renovascular hypertension than in adults, underscoring the need to examine the intrarenal vessels in children suspected of renovascular hypertension.

Role of angioplasty Angioplasty is feasible in children with disease of the main or major intrarenal arteries but requires expertise in paediatric angioplasty to minimize complications. The selection of cases

is possibly the most important aspect of angioplasty and this is why some institutions prefer to undertake a diagnostic study first, together with renal vein renin sampling, review of the results of all the imaging, degree of hypertension, and current drug regimen of the child, before undertaking an angioplasty. As angioplasty can cause renal arterial thrombosis, a follow-up DMSA scintigram soon after the procedure is important (Fig. 66.22).

RENAL CYSTIC DISEASE Cystic disease of the kidney is a wide spectrum of different diseases with variable kidney involvement. Although seen at any age, antenatal US allows many now to be detected in utero.

Clinical information No single classification of cystic kidney disease is entirely satisfactory but the most widely acceptable is based on genetics. The genetic disorders are often systemic and there may be more than one abnormal chromosome causing the disorder (Table 66.2). They present at any age, e.g. autosomal dominant polycystic kidney disease (ADPKD). Important factors include family history of renal disease, pedigree, and histopathology (if available) of affected members and consanguinity. US of parents and siblings; renal function; clinical presentation of the child; and evidence of a syndrome together with any abnormalities of the skin, eyes, lungs, cardiac, cerebral, and/or musculoskeletal systems are all important. Nongenetic conditions include both congenital and acquired disorders.

Nongenetic ‘cystic kidney’ (Table 66.3) Dysplasia/cystic dysplasia These terms used by nephrologists, radiologists, pathologists, and urologists have different meanings. Dysplasia is a histological diagnosis based on abnormal metanephric differentiation, with persistence of fetal kidney tissue in the form of nests of metaplastic cartilage associated with primitive ducts. However, most clinicians use the term when US shows a small kidney

1553

SECTION 8

•

PAEDIATRIC IMAGING

Corrected renograms R

1554

0

10

20

30

40

50

60

70

80

C 120

100

L

80

60

40

20

R 0

1.6

26

D

50 BG Subtracted

75

100

Figure 66.22 Hypertension in a 7 year old boy who presented with left facial nerve palsy, a well-recognized presenting feature of hypertension in paediatrics. (A) A 99mTc-DMSA scintigram in February 1987 shows a normal left kidney contributing 69% to overall function and a small unscarred right kidney contributing 31% to overall function. Results of US examination of this boy were normal. (B) Repeat 99mTc-DMSA scintigraphy in March 1987 following oral captopril 90 min before injection of DMSA shows absent function in the right kidney while the left kidney remains normal. (C) The 99m Tc-DTPA renogram curves after background subtraction show a normal left kidney (L); the right kidney (R) shows a decreased uptake of radionuclide with a poor renogram. (D) Following oral captopril the 99mTc-DTPA renal curves reveal no change on the left but that for the right kidney is no longer normal. (E) Free-flush aortic angiogram using DMSA. The aorta is normal; the right renal artery shows narrowing from its origin all the way to the renal hilum (arrow). The left renal artery looks normal apart from narrowing at the origin of the artery inferiorly. (F) Selective left renal arteriogram reveals a normal intra-arterial supply within the left kidney.

Table 66.2 GENETIC CONDITIONS

Table 66.3

Autosomal dominant

Cystic dysplasia or dysplasia

Autosomal dominant polycystic kidney disease

Multicystic dysplastic kidney

Tuberous sclerosis

Multilocular cyst

Medullary cystic disease

Multilocular cystic Wilms’ tumour

Glomerulocystic disease

Localized cystic disease of the kidney

Autosomal recessive Autosomal recessive polycystic kidney disease Juvenile nephronophthisis

NONGENETIC CONDITIONS

Parapelvic cyst Simple cysts Calyceal cyst (calyceal diverticulum) Medullary sponge kidney

Cysts associated with syndromes

Acquired cystic kidney disease (in chronic renal failure)

Chromosomal disorders Autosomal recessive syndromes X-linked syndromes

with loss of the normal corticomedullary differentiation and hyper-reflectivity with a variable number and size of cysts (Figs 66.23, 66.24). Dilatation is not a common feature unless associated with VUR. There is poor function to a varying degree on 99m Tc-DMSA scintigraphy which may also show focal defects that should not be interpreted as a scar since such appearances are seen in children with dysplasia who have never had a UTI. The calyces if outlined by contrast medium at the time of MCUG

show loss of the fornices with resulting blunting. The calyces may be few in number and the renal pelvis may be dilated and rather perpendicular with a tortuous and dilated ureter. If bilateral, the child will go into renal failure but end-stage renal failure is variable and may not come about until late in the second decade of life. Dysplasia is often seen in association with other congenital malformations of the kidney. The most common are the upper moiety of a duplex, posterior urethral valves, and malpositioned kidneys, as in crossed fused renal ectopia, horseshoe kidney, or the pelvic kidney. In many syndromes and genetic disorders the kidneys are affected and are called

CHAPTER 66

Figure 66.23 Renal dysplasia in a boy who presented at birth with rectal atresia. (A) The longitudinal US of the right kidney shows a small (32 mm) bright kidney with some differentiation between the cortex and medulla. A micturating cystourethrogram was carried out (not shown) but no vesico-ureteric reflux was seen. (B) 99mTc-DMSA scintigram reveals two equally functioning kidneys with no focal defects. Differential function: left = 50%, right = 50%. Each kidney, however, only took up 8.7% of the injected dose and the child was noted to be in chronic renal failure. He was classified as having renal dysplasia.

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Figure 66.25 Multicystic dysplastic kidney. Longitudinal US of the right renal area. This shows the right renal fossa filled with multiple cysts of varying sizes and no normal renal parenchyma.

renal parenchyma. Histologically primitive cartilage and ductules are present. There is no increased incidence of VUR. MCDK most commonly presents on antenatal US, is more common in boys, and the natural history is for involution. Uncommonly some persist, particularly those over 5 cm in diameter at diagnosis. A few cases complicated by either hypertension or malignancy are recorded but this is not yet accepted as a true association. Indications for surgery include a mass so large as to impede breathing or feeding, an enlarging mass, or one that by the age of 1 year remains over 5 cm in diameter. MCDK, if bilateral, is incompatible with life. The prognosis depends on the function of the contralateral kidney in which there is a 30% association of PUJ obstruction or ureteric stenosis. A MAG3 scintigram should be undertaken to assess drainage of the opposite kidney if hydronephrosis or a hydroureter is seen. MCUG is not routinely required unless the US examination reveals a contralateral dilated ureter or an abnormal bladder.

Figure 66.24 Cystic dysplasia. (A) Longitudinal US of the right (top) and left (bottom) kidneys. This shows that both kidneys are hyperechoic and contain small cysts (arrowheads). In addition, there is some dilatation of the collecting system on the left. This male infant also had dilated ureters. (B) Longitudinal US of the posterior urethra during voiding shows the dilatation of the posterior urethra. This infant had posterior urethral valves.

Calyceal cysts

dysplastic with or without cysts. Some of the more common syndromes include Meckel’s, Beckwith–Weidemann, Laurence– Moon–Beidl, Opitz Lemi, and oto-brachial-digital syndrome.

This is not a true cystic condition but rather an observation made on US. The importance of a calyceal cyst or diverticulum is the possibility of a calculus developing within it or the theoretical existence of tuberculosis as the cause due to infundibular cicatrization. Further imaging, if any, will depend on the clinical setting.

Multicystic dysplastic kidney

Simple cysts

In the multicystic kidney or multicystic dysplastic kidney (MCDK) there is no normal overall structural pattern to the kidney, which is always nonfunctioning and the ureter is atretic. On US, there is a spectrum of appearances from a small single cyst to a large mass containing multiple, usually anechoic, cysts of varying sizes, often with a dominant large cyst situated peripherally (Fig. 66.25). There is no identifiable

Simple renal cysts in children are extremely rare with a frequency of 0.22% and are usually detected incidentally15. The incidence of cysts is not related to age in this population. Simple cysts are more common following chemotherapy or abdominal radiotherapy in oncology patients. The important differential diagnosis is an abnormal upper moiety of a duplex kidney. US cannot differentiate a simple cyst from a calyceal cyst.

1555

1556

SECTION 8

•

PAEDIATRIC IMAGING

Localized cystic disease of the kidney This condition is recognized as genetically, radiologically, and morphologically distinct from ADPKD. The involved segment of the kidney is usually enlarged, containing multiple, anechoic small cysts which gradually merge into normal renal parenchyma and are not sharply demarcated from the adjacent normal parenchyma. The cardinal feature which helps differentiate this condition from the multilocular cystic Wilms’ tumour is the absence of a surrounding capsule on imaging and histology. The cystic abnormality differs from ADPKD in regard to its localization into one part of the kidney and the absence of cysts in the contralateral kidney. No family history or pattern of inheritance has been reported. The lesions do not appear to be progressive in the few cases that have been reported and left in situ. Hypertension has been documented but the natural history is unclear.

Medullary sponge kidney Medullary sponge kidney is not usually seen in childhood, but when it is the changes are the same as those seen in adults.The US appearances of renal tubular ectasia without calculi are simply an increased echogenicity of the medullae in normal sized kidneys.

Acquired cystic renal disease In patients in chronic renal failure, cysts commonly develop (22%) without any underlying cystic renal disorder. The frequency increases with duration of dialysis with 90% developing cysts if on dialysis for over 10 years. After successful transplantation, the cysts tend to regress in size. Complications include renal cell carcinoma, haemorrhage, or infection in the cysts.

Genetic cystic disease Table 66.4 compares some of the findings in the more common cystic renal disorders in children.

Autosomal dominant polycystic kidney disease (ADPKD) This is rarely seen in childhood and commonly presents after the third decade of life. However, there is considerable phenoTable 66.4

typic variability in the severity of the renal disease with some affected individuals presenting in childhood or being detected antenatally. Prevalence is approximately 1 in 1000 with two or three genetic loci identified; 90% of families have the gene located on the short arm of chromosome 16 (PKD1). The second gene is on chromosome 4 (PKD2). The location of PKD3 has not yet been found. In PKD1 families, 64% of children affected under the age of 10 years will have cysts and 90% between the ages of 10 and 19 years. There may be no family history as spontaneous mutation accounts for many cases. The extrarenal manifestations of ADPKD include cysts in the liver, pancreas, or spleen, again rarely seen in paediatric patients. Subarachnoid haemorrhage due to intracranial aneurysm is also rare in childhood.The occurrence of subarachnoid haemorrhage tends to cluster in families. Congenital hepatic fibrosis is generally associated with autosomal recessive PKD (ARPKD; see below) but some rare cases have been described in association with ADPKD. Ultrasound In the prenatal period, US may demonstrate highly reflective kidneys similar to ARPKD. In infancy the ultrasonic appearances vary, from a normal kidney to a few isolated cysts to a kidney packed full of cysts (Fig. 66.26).Typically the cysts are scattered throughout both the cortex and medulla with unequal involvement of the two kidneys. The intervening renal parenchyma appears normal.When found in the young child (younger than 5 years of age) with no family history, a diagnosis of tuberous sclerosis (TS) must also be considered and actively excluded (Fig. 66.27). Intravenous urogram This probably has little role. MRI may be the examination of choice, especially when tuberous sclerosis is suspected to identify fatty tissue in an angiomyolipoma.

Tuberous sclerosis TS is an autosomal dominant condition with a prevalence of 1 in 10 000.TS is characterized by multiple hamartomas in the brain, skin, heart, kidneys, liver, lung, and bone, and cysts are also seen in the kidneys. Genetic linkage to chromosome 9 is seen in about one-third of families, and to chromosome 16 in the same region as the ADPKD1 gene.

COMPARISON OF RENAL CYSTIC DISEASES

Inheritance

ADPKD

Tuberose sclerosis

ARPKD

MCDK

Simple cyst

D

D

R

None

None Unilateral

Uni- or bi-lateral

Bilateral unequal

Bilateral

Bilateral equal

Uni- or bi-lateral

Kidney size

Normal or large

Normal or large

Very large > 90th centile

Small or large

Normal

Extrarenal manifestations

Cysts in liver spleen pancreas

Cardiac rhabdomyomas, intracranial tubers

Congenital hepatic fibrosis

None

None

Age at presentation

Third decade

Often < 18 months

Neonate and childhood

Antenatal, rare in childhood

Onset in adult life

Cyst size

Visible cysts of variable size

Similar to ADPKD, ± angiomyolipomas

Generally small

Large then often involute

Variable

Diagnosis

US, genetic

US, cardiac echo, cranial MRI

US, IVU, liver biopsy

US, MAG3

US, IVU

Malignancy risk

No

Yes

No

Rare

No

ADPKD = autosomal dominant polycystic kidney disease; ARPKD = autosomal recessive polycystic kidney disease, D = autosomal dominant; MCDK = multicystic dysplastic kidney; R = autosomal recessive

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

although not usually in the paediatric age range. Up to 50% of patients with cardiac rhabdomyomas have TS.

Autosomal recessive polycystic kidney disease ARPKD is a rare genetic disorder with an incidence of approximately 1 in 55 000 live births. The parents are always unaffected. The gene for ARPKD has been located on chromosome 6. Younger children tend to have the more severe renal involvement while in older children hepatic involvement predominates. Less severely affected children may present in later childhood or even in adolescence.

Figure 66.26 Autosomal dominant polycystic kidney disease (ADPKD). Longitudinal US of the right kidney in a child with a strong family history of ADPKD. The kidney contains a number of visible cysts in an otherwise normal looking kidney. The left kidney had very similar appearances.

The incidence of renal manifestations on imaging varies in the reported literature but is estimated to be between 47% and 73%. Angiomyolipomas are the most common lesions found in older patients and they may occur with or without cysts. Renal cysts are found less frequently (18–53%) and occur in younger patients. No imaging modality can currently differentiate cysts in TS from ADPKD. When cysts are found in the young child (younger than 5 years of age) with no family history, a diagnosis of TS must also be considered and actively excluded. Echocardiography and cranial MRI should form part of the diagnostic work-up. Ultrasound On US (Fig. 66.27) the kidneys may have multiple cysts, resembling ADPKD. In later childhood US may demonstrate multiple small rounded echogenic foci throughout the renal parenchyma due to multiple angiomyolipomas. Complications of an angiolipoma when greater than 4 cm in diameter in particular include haemorrhage as renal vascularity is grossly abnormal. Renal cell carcinoma may also occur,

Ultrasound Prenatal diagnosis with US has been reported as early as 14–17 weeks’ gestation but as the appearance of bilateral highly reflective kidneys during the fetal period is not specific to ARPKD, caution is necessary before suggesting this diagnosis on prenatal US. Both kidneys are always equally involved and markedly enlarged, measuring over the 95th centile for age in early infancy. As the infant grows, so the kidneys appear relatively less enlarged. The characteristic appearance is a hyper-reflective cortex and medulla, although variations, with the medulla much brighter than the cortex, may be seen. Using high-frequency probes, small 1–2-mm cysts may be detected in the medulla. Occasionally cysts of less than 2 cm may be found, especially as the child grows. In a small proportion of cases there is an evolving appearance of the kidneys such that in later childhood the kidneys contain large cysts of different sizes with an appearance indistinguishable from ADPKD (see Table 66.4). Most children after the first year of life have some degree of hepatosplenomegaly.The spectrum of hepatic abnormalities may be subtle in the neonate. In the young child the liver is enlarged with increased echogenicity in the periportal region from bile duct proliferation and fibrosis. Single or multiple cysts communicating and closely related to the biliary tree, together with biliary ectasia, may be present. A large spleen and evidence of portal hypertension and varices may be demonstrated with US and Doppler examination, although this is not usually present in the first decade of life.

Figure 66.27 Tuberous sclerosis. (A) Longitudinal US of the right kidney. The left kidney had similar appearances. Both kidneys contain multiple cysts of varying sizes. The appearances on US are indistinguishable from autosomal dominant polycystic kidney disease. This child had the skin stigmata of tuberous sclerosis. (B) Longitudinal US of the right kidney of another patient. There were similar appearances in the left kidney. This shows the more usual appearances of tuberous sclerosis in the kidney with the small echogenic foci of angiomyolipomas.

1557

1558

SECTION 8

•

PAEDIATRIC IMAGING

Intravenous urogram The US appearances are not specific and some conditions may masquerade as ARPKD, such as ADPKD and glomerulocystic disease. The IVU may help differentiate these conditions by demonstrating a streaky nephrogram as contrast medium pools in the ectatic collecting ducts. The calyces may be distorted with full collecting systems. The IVU should be undertaken after 6 months of age since renal immaturity with or without associated renal impairment before this age may result in nonvisualization of the kidneys (Fig. 66.28). 99m Tc-DMSA scintigraphy shows bilateral focal defects in enlarged kidneys with a high background activity. This combination of appearances on US and DMSA are characteristic of ARPKD. The liver has a characteristic appearance. Using HIDA, a hepatobiliary agent which undergoes hepatocyte uptake, transport, and secretion into the bile, the early images reveal an enlarged left lobe of the liver in children over 1 year of age. There may be delayed passage of isotope through the liver, with areas of pooling and prominence of the duct system.This provides a valuable noninvasive method for the demonstration of subclinical biliary ectasia. This appearance has been described in Caroli’s disease. US may fail to show dilatation of the bile ducts.

Juvenile nephronophthisis/medullary cystic disease These are two different terms for conditions that have a different inheritance, different age of onset, and different associations, but a similar renal morphology and imaging. Both

Figure 66.28 Autosomal recessive polycystic kidney disease (ARPKD). (A) Longitudinal US of the right kidney and (B) DMSA scintigram on a patient with ARPKD. US shows the typical appearances of a large hyperechoic kidney. The left kidney appeared similar. The DMSA scintigram shows photon-deficient areas in both the kidneys, particularly in the polar regions. These defects do not correspond to visible cysts on US. The child had no history of urinary tract infection. (C) Intravenous urogram on a patient with ARPKD showing the streaky radiation of contrast medium as it passes through the ectatic tubules. (D) HIDA scintigram: delayed image after 2 h showing retention of tracer within the liver, an enlarged left lobe, and a ‘patchy’ appearance as the tracer has accumulated in the ducts. Similar appearances have been reported in Caroli’s syndrome with congenital hepatic fibrosis.

present with slowly progressive renal failure. Juvenile nephronophthisis (JN) has an autosomal recessive inheritance, presenting in childhood as chronic renal failure. It is characterized by an early concentrating defect, polyuria, polydipsia with growth retardation, and anaemia. The patients are in end-stage renal disease before the age of 25 years. Medullary cystic disease (MCD) shows an autosomal dominant inheritance and presents up to the fourth decade of life. Imaging US reveals two normal-sized kidneys with a globally hyper-reflective appearance. Cysts are not a feature until late in the disease and are then typically corticomedullary. An IVU has no role. In the early stages of the disease when the tubules are affected to a greater degree than the glomeruli, the DMSA scinitigram may fail to show the kidneys yet the 99mTc-DTPA scintigram may be almost normal. The ultimate diagnosis can only be made on biopsy. Extrarenal manifestations reported in JN include skeletal abnormalities, congenital hepatic fibrosis, and mental retardation. None has been reported in MCD. In Jeune syndrome (asphyxiating thoracic dystrophy) children develop renal failure in adolescence similar to that seen in JN.

RENAL TRANSPLANTATION Renal transplantation in very young children is now a wellestablished treatment, giving a significantly improved quality of life. The three most common causes of end-stage renal disease leading to transplantation are renal dysplasia, glomerular disease and pyelonephritis.

CHAPTER 66

Pre-transplantation In children with multiple congenital anomalies, anatomical variations of the aorta and IVC, e.g. high bifurcation, are common. Identification of such anomalies may be achieved using US, but MRI or CT may be required.

Post-transplantation The major imaging techniques used are US and nuclear medicine studies with the occasional need for cross-sectional imaging or interventional techniques.The principles of imaging children post-transplantation are broadly similar to those in adults, with a few exceptions16. The renal artery and vein are often anastomosed to the aorta and IVC rather than to the iliac vessels. US will identify complications, e.g. calyceal, pelvic, or ureteric dilatation, and bladder abnormalities. Colour Doppler examination is useful in the immediate post-transplant phase, especially if the child is anuric; in later follow-up, the results are disappointing with an inability to differentiate between ATN, rejection, and cyclosporin toxicity (Table 66.5). Early reports of the value of the resistive index have not been confirmed and this institution no longer reports any index of flow. 99mTc-DTPA scintigraphy will assess blood flow in the perfusion phase (0–20 s post-injection); the filtration of the kidney (1–2 min post-injection); and then transit through the kidney over 20 min with an additional image following a change in posture and micturition to assess drainage from the calyces to the bladder. The integration of the results of different imaging techniques with clinical data will provide the best results as no single technique allows specific diagnosis of graft dysfunction. The role of MRI has not been fully evaluated. It may well replace the IVU, which is only indicated if a dilated ureter is seen and the creatinine is rising, to assess the exact level of the partial hold-up.

Late dysfunction A late decrease in renal function may be due to rejection or drug toxicity. 99mTc-DTPA scintigraphy may suggest the

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

diagnosis but ultimately all patients have renal biopsies. US will detect hydronephrosis/hydroureter, which if new, requires surgery.Table 66.6 gives the relevant imaging findings in some of the relevant conditions. When children develop hypertension, arterial stenosis must be considered and arteriography is required. This is uncommon with cadaveric donor kidneys as there is a patch of aorta included, but it is more likely with living donor transplant. Catheter angiography remains the reference technique in children, but the results of MRI are promising. The preferred method of treatment is angioplasty. Lymphoceles are readily detected by US as large cystic septated collections but cannot be reliably differentiated from urinomas or large haematomas. A 99mTc-MAG3 scintigram will show whether urine enters the collection (a urinoma). Lymphoproliferative disorder is a well-recognized complication of all transplantation with concomitant immunosuppression. The children may present with local or generalized lymphadenopathy and simply curtailing the immunotherapy may resolve the enlarged lymph nodes. There is a well-recognized association with Epstein–Barr virus. Occasionally children may present with an aggressive lymphoma requiring full chemotherapy.

TUMOURS BENIGN TUMOURS Nephroblastomatosis The term ‘nephrogenic rest’ describes persistence of embryonic renal parenchyma (metanephric blastema) beyond 36 weeks’ gestation when renal development is normally complete. Nephroblastomatosis is the presence of multiple nephrogenic rests. The abnormal foci of persistent nephrogenic cells are regarded as precursor lesions. There is an increased incidence of Wilms’ tumour (‘nephroblastoma’) in children with nephroblastomatosis17. The lesions are found in association with 41% of unilateral Wilms’ tumours, with 94% of metachronous bilateral Wilms’ tumours and in 99%

Table 66.5 IMAGING FINDINGS IN PRIMARY NONFUNCTION Imaging technique

ATN

RVT

RAT

US

Good arterial and venous flow

No venous flow

No venous flow, no arterial flow

US Doppler

Normal arterial and venous flow or damped due to swollen kidney

Damped systolic peak with reverse arterial flow in diastole.

Very late—no flow

99m

Good

Photon-deficient area

Photon-deficient area

99m

Moderate

Very poor

Nil

99m

Very slow and progressive

Often normal

Nil

Tc-DTPA blood flow Tc-DTPA filtration Tc-DTPA transit

ATN = acute tubular necrosis; RAT = renal artery thrombosis; RVT = renal vein thrombosis.

Table 66.6

IMAGING FINDINGS WITH A RISING CREATININE IN RENAL TRANSPLANTS

Imaging technique

Rejection

US

Cyclosporin toxicity

Obstruction

Good arterial and venous flow

Good arterial and venous flow

Dilatation of the calyces, pelvis and ureter

99m

Reduced markedly

Moderate reduction

Normal

99m

Moderate reduction

Reduced markedly

Normal or reduced

Slow, but isotope seen

Slow with retention

Slow and progressively rising in bladder and parenchyma

Tc-DTPA blood flow Tc-DTPA filtration

99m

Tc-DTPA transit

1559

1560

SECTION 8

•

PAEDIATRIC IMAGING

of multicentric or bilateral Wilms’ tumours. The malignant potential of individual lesions is uncertain; however, as only a minority of nephrogenic rests develop into Wilms’ tumours and spontaneous regression may occur. Nephroblastomatosis may either be diffuse or multifocal (Fig. 66.29), although a unifocal form may be found.The diffuse form demonstrates a thick band of reduced reflectivity on US. This abnormal tissue surrounds the renal periphery and is nonenhancing on CT and MRI. In the more common multifocal type, the nephrogenic rests resemble normal renal cortex on all modalities and can be scattered throughout the kidneys; they may be nodular or plaque-like and after contrast medium administration they become hypodense on CT and hypointense on MRI due to poor perfusion in relation to the highly vascular renal cortex. Nephroblastomatosis yields variable signal intensity depending on cellularity and histological characteristics onT2-weighted MRI. In general, the signal intensity of nephroblastomatosis on all sequences, including gadolinium-enhanced images, tends to be relatively homogeneous in contrast to Wilms’ masses which are always heterogeneous in appearance.With enhancement on either CT or MRI, the lesions are more conspicuous, which is important for the detection of small lesions. Despite characteristic imaging appearances, early histological evaluation and serial assessments are warranted because of the known malignant risk. The differential diagnosis of diffuse nephroblastomatosis on US mainly includes renal lymphoma or leukaemia.

removal or capsular penetration. With complete removal there is an excellent prognosis.

Mesoblastic nephroma

Wilms’ tumour or nephroblastoma accounts for up to 12% of all childhood cancers with a peak incidence at around 3 years of age18.The most common presentation is with an asymptomatic abdominal mass. Haematuria, particularly after minor trauma, is another typical clinical manifestation while pain, fever, or hypertension (in up to one-quarter of cases) are unusual but recognized presenting features. Microscopic haematuria is present in 25% of cases. There is equal distribution between the sexes with the highest incidence being in the black population in the USA and Africa. Around 10% of Wilms’ tumours are

Mesoblastic nephroma is the most common renal neoplasm in the first 3 months of life. Commonly it presents as an abdominal mass in a neonate. On US there are areas of heterogeneous reflectivity; the mass is solid but there may be hypoechoic areas due to cystic change or necrosis. Neither US nor CT can reliably distinguish mesoblastic nephroma from Wilms’ tumour. The former tumours show uptake of 99mTc-DMSA. Mesoblastic nephroma does not invade the vascular pedicle nor does it metastasize. Local recurrence may result from incomplete

Multilocular cystic nephroma Multilocular cystic nephroma is an uncommon cystic renal mass, derived from metanephric blastema, occasionally seen in children. It is bimodal, seen more commonly in boys under 4 years and in women in the fifth or sixth decades. It reflects a wide spectrum of histology from the completely benign on the one hand (multilocular renal cyst) to the malignant on the other (multilocular cystic Wilms’ tumour). The imaging is indistinguishable between the different spectrum of histology. There are no associated anomalies. Clinically, the child presents with an abdominal mass. US will show a multilocular renal mass with multiple cysts and septations. A similar appearance is seen on CT and the lesion is nonfunctioning on isotope imaging. The hallmark on imaging is the presence of a capsule. Nephrectomy is curative and is recommended because of the malignant potential.

Angiomyolipoma Angiomyolipoma (see section above on TS) is rarely encountered as an isolated phenomenon in children but more usually represents one of the renal manifestations of tuberous sclerosis.

MALIGNANT TUMOURS Wilms’ tumour

Figure 66.29 Nephroblastomatosis. (A) US demonstrating uniformly hypoechoic lesions surrounding the periphery of an enlarged kidney in a diffuse type of nephroblastomatosis. (B) Delayed post-contrast CT sections demonstrate these lesions, typically showing reduced vascularity in a fairly homogeneous pattern. Dense contrast medium is seen within the normal renal parenchyma.

CHAPTER 66

bilateral, of which two-thirds are synchronous and one-third metachronous. Associated congenital anomalies occur in 15% of children and include cryptorchidism and horseshoe kidney.Certain syndromes have a predisposition to Wilms’ tumour. These include aniridia (absence of ophthalmic iris), Beckwith–Wiedemann (macroglossia, exomphalos, gigantism), hemihypertrophy, Denys–Drash (pseudohermaphroditism), Soto’s (cerebral gigantism), Bloom’s (immunodeficiency and facial telangiectasia) and Perlman’s syndromes. In Denys–Drash syndrome, for example, most but not all patients will develop a Wilms’ tumour, the median age at presentation being 18 months and 20% of cases are bilateral. Routine US screening to detect tumours at an early stage is controversial, because despite 3-monthly US studies large interval tumours may occur. Nephroblastomatosis is commonly found in cases of bilateral masses. A small group of Wilms’ cases are familial. Intralobar nephrogenic rests are also commonly found in such patients but there is no association of familial tumours with bilaterality. Extrarenal Wilms’ tumours are rare, the most common locations are the retroperitoneum, inguinal region and pelvis. Their histology and outlook are identical to those of renal Wilms’ tumours. Wilms’ tumours are solid lesions with a fibrous pseudocapsule and they have variable areas of haemorrhage and necrosis. The tumour may invade the renal vein and IVC with caval extension, often to the right atrium, seen in 4% of patients (Fig. 66.30). Metastases to local para-aortic lymph nodes and haematogenous spread to the lungs and less commonly the liver or skeleton are seen. Epithelial (tubular, glomerular), stromal (spindle, myxoid) and blastemal (small round cells) cell lines are the histological components of nephrogenic rests, fetal kidneys and Wilms’ tumours. When all three are present in malignant masses, the lesions are described as triphasic Wilms’ tumours and are regarded as favourable histology. These lesions lack anaplastic

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

changes. Unfavourable tumours comprise 6% of lesions and typically have hyperchromatic cells with large nuclei. The degree of anaplasia correlates with patient outcome. Minimally anaplastic tumours have a prognosis similar to favourable histology lesions. Clear cell sarcoma (bone metastasizing tumour) and rhabdoid tumour are no longer regarded as Wilms’ variants but as biologically different entities. The (post-surgical) staging of Wilms’ tumour according to the North American National Wilms’ Tumor Study Group is summarized below: • Stage I—tumour confined to the kidney without capsular or vascular invasion. • Stage II—tumour beyond renal capsule, vessel infiltration, biopsy performed before resection or intra-operative tumour rupture. • Stage III—positive lymph nodes in the abdomen or pelvis, peritoneal invasion, or residual tumour at surgical margins. • Stage IV—metastatic disease outside the abdomen or pelvis. • Stage V—bilateral tumours at original diagnosis. The International Society of Paediatric Oncology (SIOP) in Europe generally utilizes the same staging system with one common exception, namely masses that have been biopsied may be regarded as Stage I disease when later excised. As with all abdominal masses, US must be the first radiological method of assessment.The tumour typically is large with a mixture of solid hyperechoic masses and relatively cystic areas; often the cystic components predominate. Normal native renal tissue can be difficult to detect and is typically stretched at the periphery of the lesion. Movement of the mass separate from adjacent organs such as the liver, indicating a lack of direct invasion, can be optimally assessed with US, a phenomenon that is difficult to evaluate on CT or MRI. The renal vein, IVC, liver and opposite kidney should be carefully assessed for spread of disease. US is also the most reliable method of excluding renal vein and IVC thrombus.

Figure 66.30 Wilms’ tumour. (A) Axial CT of the abdomen after intravenous contrast medium enhancement in a 2½ year old boy showing a large mass arising from the right kidney which is of heterogeneous attenuation. The mass is seen to displace the normal enhancing renal parenchyma to the left. (B) Thoracic CT tumour thrombus was seen to extend from the inferior vena cava into the right atrium, causing a filling defect within the heart (thin arrow). Note the enlarged azygos vein (wide arrow) adjacent to the aorta and the large right pleural effusion.

1561

1562

SECTION 8

•

PAEDIATRIC IMAGING

Contrast-enhanced CT or MRI is necessary for further delineation of tumour extent and associated changes. CT is widely available, does not require sedation or anaesthesia, is preferred by surgeons, and is needed anyway to exclude pulmonary metastases.Wilms’ tumours have a heterogeneous appearance with areas of low attenuation. Calcification is unusual, seen in less than 10% of patients.The tumour enhances to a lesser degree than normal renal parenchyma. A so-called claw or beak of normal renal tissue may be seen to be displaced by the tumour. Exclusion of adenopathy, liver abnormalities, peritoneal invasion and contralateral kidney changes should be undertaken. Small superficial or intrarenal nephroblastomatosis lesions are often not identified even with a high quality CT study. Wilms’ masses on MRI are generally hypointense on T1, variably hyperintense on T2, and enhance heterogeneously, and often poorly, with gadolinium administration (Fig. 66.31). Gadolinium is required to assess the contralateral kidney for nephroblastomatosis or another Wilms’ mass. Cross-sectional imaging is playing an increasing role in needle biopsy (14G or 16G cutting needle) planning. A controversy exists with regard to the current role of chest CT in patients at initial diagnosis. The commonly used staging systems are based on frontal and lateral chest radiographs. Although most centres perform chest CT, positive findings are ignored if no lesions can be visualized on a chest radiograph. It is argued that a pre-operative imaging protocol relying on chest radiography alone does not reduce survival and therefore chest CT is not needed. However, in a number of cases, review of a ‘negative initial’ chest radiograph following a positive CT revealed metastasis. For most cancers staging is based on disease distribution and on the presence or absence of disease, not on the imaging modality (chest radiography or CT) by which it is detected.Wilms’ staging, however, still relies solely on plain chest radiographic findings. The typical North American practice is initial surgical removal followed by adjuvant chemotherapy as dictated by the staging at surgery. European oncologists favour initial chemotherapy after biopsy confirmation with later resection. It is not surprising, therefore, as tumour shrinkage is often quite marked, that the percentage of Stage I and II patients is higher when pre-operative chemotherapy is used. The prognosis for Wilms’

tumour patients is excellent and there is little evidence to suggest that the overall relapse-free survival is adversely affected by either approach. The 4-year overall survival rate, and presumed cure, ranges between 86% and 96% for Stages I–III disease, is up to 83% for Stage IV and 70% for Stage V disease. Patients with the much less common diffuse anaplastic Wilms’ tumours have a much poorer outcome, however. Their 4-year survival rates are 45% for Stage III and only 7% for Stage IV disease.

Clear cell sarcoma of the kidney Clear cell sarcoma of the kidney is a distinct entity accounting for 4% of all childhood renal neoplasms with a marked male preponderance. There are no known genetic associations and no reports of bilateral tumours. The peak age of incidence is similar to that of Wilms’ tumour. There are no specific radiological features to help distinguish clear cell sarcoma of the kidney from a Wilms’ tumour; the distinction is histological. Although this neoplasm can metastasize to the lungs, there is a particular predilection for skeletal metastases (over 20%), hence the other known term, bone metastasizing tumour. Once diagnosed, 99mTc-MDP bone scintigraphy is indicated for staging purposes. The discovery of a secondary in the skeleton in a child with a presumed diagnosis of Wilms’ tumour and no evidence of deposits in the lungs should suggest that the primary diagnosis is incorrect and that the tumour is likely to be a clear cell sarcoma of the kidney.

Rhabdoid tumour of the kidney Rhabdoid tumour of the kidney is the most aggressive malignant renal tumour in childhood and accounts for 2% of paediatric renal neoplasms. Most cases are diagnosed in the first year of life. The mass is indistinguishable from a Wilms’ tumour on imaging. A peripheral fluid crescent sign on CT has been described but it is not pathognomic and the relative rarity of the tumour diminishes the positive predictive value of this finding in an individual case. Invasion of the renal vein is common. Metastases to the lungs, liver and brain have been reported. There is also an association with simultaneous primitive neuroectodermal tumours, usually in the posterior fossa. Hypercalcaemia is a recognized finding in rhabdoid tumour but it is not specific as it is also found occasionally in mesoblastic nephroma.

Figure 66.31 Wilms’ tumours. MRI. (A) Axial T2-weighted and (B) axial T1-weighted images, post-gadolinium administration, show a large tumour growing exophytically anterior to the left kidney in a 2 year old girl. (C) Coronal T2-weighted image in another 2 year old showing a heterogeneously hyperintense, large right-sided mass.

CHAPTER 66

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Renal cell carcinoma Renal cell carcinoma rarely presents in the first two decades of life, with less than 1% of all cases in children. A mass or flank pain are more common presenting features than haematuria. The mean age at paediatric presentation is 9 years. A typically solid intrarenal mass cannot be distinguished from a Wilms’ tumour. Occasionally ring-like calcifications are present within the mass but the usual discriminating feature from a possible Wilms’ tumour is the older age of the patient. Metastases to the lungs, liver, skeleton or brain are present in 20% of patients at diagnosis.

Lymphoma and leukaemia Renal involvement with or without retroperitoneal adenopathy is seen in 12% of children with non-Hodgkin’s lymphoma, most commonly B-cell Burkitt lymphoma. Multiple, usually bilateral, nodules are typical, although diffuse infiltration may be seen. There is generally widespread disease elsewhere. Renal enlargement on US with altered echo texture is characteristic of both renal lymphoma and leukaemia.The changes in the kidneys can be quite subtle on CT and may be more conspicuous on MRI, particularly T1-weighted images after gadolinium enhancement. US is recommended in all children with leukaemia/lymphoma before the commencement of chemotherapy to detect tumour infiltration or calyceal dilatation. Before and during initial chemotherapy, a large fluid load is administered which in addition to the excretion of tumour metabolites may result in renal obstruction or uric acid nephropathy; those children with leukaemic or lymphomatous involvement of the kidneys need careful nephrological monitoring.

Figure 66.32 Prostatic rhabdomyosarcoma. MRI sagittal T1weighted images through the pelvis in a 10 year old boy showing a large mass of intermediate signal intensity centred on the prostate (arrowheads). A catheter is seen to traverse the urethra with the balloon in a collapsed bladder. The appearances are those of the bladder/prostate rhabdomyosarcoma.

Rhabdomyosarcoma Rhabdomyosarcoma is the most common malignant neoplasm of the pelvis in children. The genitourinary tract is the second most common site of rhabdomyosarcoma in children after head and neck locations (Fig. 66.32). Although the term suggests a mesenchymal tumour derived from striated muscle, the tumour frequently arises in sites lacking striated muscle. The two major cell types are embryonal and alveolar, of which embryonal is the more common and has a better prognosis. A botryoid subtype of embryonal rhabdomyosarcoma is characterized macroscopically by the presence of grape-like polypoid masses and accounts for 5% of cases. Botryoid tumours classically occur in the vagina, rarely metastasize and respond well to current treatment regimens. In general, pelvic tumours may be very large at presentation and present as abdominal masses (Fig. 66.33). Prostate tumours may cause urinary obstruction and manifest with marked bladder distension or acute retention.The mass is typically solid but heterogeneous on US. Vascularity of the mass lesions can be quite variable, but some tumours are seen to enhance vividly after contrast medium administration on both CT and MRI. In addition to assessment of the primary site, the regional lymph nodes must be evaluated. Although MRI is recommended in general for all pelvic tumours, it does have some pitfalls. Oedema and later non-specific soft tissue thickening after radiotherapy may suggest an increase in tumour size or persisting tumour which can lead to discrepancies between the MRI (and CT) findings and surgical or biopsy results. Coronal

Figure 66.33 Rhabdomyosarcoma. (A) Sagittal nonenhanced T1-weighted images. (B) Sagittal T1-weighted images after gadolinium enhancement showing a large cystic mass lesion due to an embryonal rhabdomyosarcoma in the presacral region extending to the perineum. The mass demonstrates vivid peripheral enhancement after contrast medium administration, but no obvious central enhancement was seen. Note that the bladder and bowel are displaced anteriorly. No intraspinal extension was evident.

or sagittal T1-weighted imaging without contrast can often be the most useful sequences for follow-up. Routine chest CT is also recommended as up to 10% of rhabdomyosarcomas have pulmonary metastases at diagnosis. 99mTc-MDP skeletal scintigraphy is performed in all cases to detect the smaller percentage of patients who have skeletal metastases. Rhabdomyosarcomas in favourable sites such as the vagina have up to 94% 3-year survival. Tumours in the prostate and bladder, however, have a worse prognosis with a 3-year survival of approximately 70%. Prostatic tumours commonly infiltrate locally into the perivesical tissues and into the bladder base and anterosuperior spread into the space of Retzius is also recognized. The goal of therapy for bladder or bladder/prostate tumours (as it is frequently very difficult to tell the exact organ of origin) is survival with an intact bladder.

1563

1564

SECTION 8

•

PAEDIATRIC IMAGING

SCROTAL MASSES Intratesticular benign and malignant tumours are relatively common neoplasms in children. Primary testicular neoplasms include germ cell tumours (most common and often with calcifications), endodermal sinus tumour and embryonal carcinoma. Paratesticular rhabdomyosarcoma accounts for 12% of scrotal tumours in boys. The testes are also secondary sites of disease in children with leukaemia, lymphoma and neuroblastoma, although much less frequently than in adults. Paratesticular rhabdomyosarcoma includes tumours arising in the spermatic cord, testis, epididymis and penis. US is recommended to evaluate scrotal disease. A heterogeneous appearance within the testis with increased flow on Doppler may mimic infection, but clinical presentation is usually with a suspected or palpable mass rather than with signs suggesting infection. The regional lymph nodes, including the para-aortic nodes, require evaluation by both US and CT/MRI. CT studies with oral contrast medium administration are mandatory as retroperitoneal lymph node dissection may depend on the results. Paratesticular rhabdomyosarcoma has a good outlook in children younger than 10 years with a greater than 90% 5-year survival.

OVARIAN MASSES Ovarian cysts Simple ovarian cysts represent the majority of ovarian masses. In the neonate they usually present as abdominal masses because of the small size of the bony pelvis. When very large, they must be differentiated from other lesions such as mesenteric cysts, intestinal duplication and urachal cysts. Spontaneous resolution of some cysts has been reported. The treatment of choice is observation or aspiration if the cyst is very large. In pubertal girls, ovarian cysts result from continuous growth of a follicle after failed ovulation or when it does not involute after ovulation. Most follicular cysts are not loculated, contain clear fluid, Figure 66.34 Ovarian teratomas. (A) Pelvic US in an 8 year old girl showing a mainly anechoic cystic mass, over 10 cm in diameter, with a more solid component posteriorly. (B) Coronal CT reformatted image in a 10 year old girl showing a cystic mass lesion superior to the bladder with (possibly tooth-like) calcification seen peripherally on the right side of the mass.

and measure between 3 and 10 cm. Corpus luteum cysts contain serous or haemorrhagic fluid, and as with follicular cysts, usually involute spontaneously. Most ovarian cysts are asymptomatic, but when complications such as torsion, haemorrhage or rupture occur, patients present with acute abdominal pain, nausea, vomiting and leukocytosis. Torsion of the ovaries and Fallopian tubes results from partial or complete rotation of the ovary on its vascular pedicle. US is helpful in the evaluation of these patients, showing ovarian enlargement and fluid in the cul-de-sac. The presence of a fluid–debris level or septa may be an indication of bleeding or infarction. A more specific sign of torsion is the presence of multiple follicles in the cortical portion of a unilaterally enlarged ovary and a lack of Doppler signal. Physiological or pathological cysts may be found coincidentally in patients with lower abdominal or pelvic pain. Repeating the US study at a different phase of the menstrual cycle is useful.

Ovarian tumours Ovarian neoplasms account for 10% of all childhood tumours, and 10–30% of these are malignant (the most common is malignant germ cell tumours). Tumours include dysgerminoma, immature teratoma, embryonal carcinoma, endodermal sinus tumour and choriocarcinoma. The differentiation between ovarian torsion and a tumour may be difficult. Tumour markers are helpful. Tumours are more common after puberty but can occur at any age, presenting with abdominal pain or as a palpable mass. Mature teratomas and dermoid cysts account for two-thirds of paediatric ovarian tumours and have a wide spectrum of imaging characteristics (Fig. 66.34). The most characteristic appearance on US is that of a mass of low reflectivity with an echogenic mural nodule. Calcifications and fat–fluid levels also assist in the diagnosis. Cystadenomas represent 20% of ovarian tumours in children and are of epithelial origin; they are large tumours, with loculations. Imaging cannot differentiate between malignant and benign cystadenomas. Leukaemia, lymphoma and neuroblastoma are among the primary tumours that metastasize to the ovaries in children.

CHAPTER 66

PRESACRAL MASSES Sacrococcygeal teratoma is the most common presacral tumour and is a congenital lesion containing derivatives of the three germinal layers. These lesions occur more frequently in girls, are mostly nonfamilial, and are classified into four types, according to location. Type I is predominantly external and Type IV entirely presacral with no external component. The lesion may be evident at birth or present later (Type IV). Sacrococcygeal teratomas are attached to the sacrum, with the internal component best depicted on MRI (Fig. 66.35). This finding in addition to solid components allows differentiation from anterior sacral meningocoele.

•

IMAGING OF THE KIDNEYS, URINARY TRACT AND PELVIS IN CHILDREN

Benign teratomas are usually cystic, with calcification and fat. Malignant teratomas are predominantly solid and may invade adjacent structures. All sacrococcygeal teratomas have the potential for later malignant transformation and hence require surgical removal. They can metastasize to the chest. The tumour marker for sacrococcygeal teratoma is α-fetoprotein. Other presacral lesions include anterior myelomeningocoele and neuroenteric cyst, and forms of spinal dysraphism associated with a sacral defect which are well imaged by MRI of the spine. Neuroblastoma and lymphoma are masses less commonly seen in the presacral region.

Figure 66.35 Sacrococcygeal teratomas. (A) Plain abdominal radiograph showing a very large external component to a sacrococcygeal teratoma in a newborn. (B) Sagittal T2-weighted image in another infant showing a large intrapelvic component to the sacrococcygeal teratoma. Note the teratoma, posterior to the uniformly hyperintense bladder, is of intermediate heterogeneous signal, and is invading the sacrum. (C) Axial T1-weighted image in the same patient showing the heterogeneous mass in a presacral location.

1565

1566

SECTION 8

•

PAEDIATRIC IMAGING

NEPHROCALCINOSIS (RENAL CALCULUS)

REFERENCES

US is more sensitive in the early stages of calcium deposition in the kidneys than plain abdominal radiography. Increased medullary echogenicity of the kidneys in children, while nonspecific on US, may be an unexpected finding and indicative of underlying metabolic disease. It always requires a clinical explanation. Iatrogenic nephrocalcinosis (diuretics, vitamin D) is the most common cause in children for an increased echogenicity of the medullary pyramids. If this diagnosis can be excluded a large differential diagnosis remains (Table 66.7). Renal calculi may present with pain, haematuria, or UTI, or they can be asymptomatic. US, a plain abdominal radiograph (for ureteric stones) and a 99mTc-DMSA scintigram to assess the effect in renal function are essential.

1. Dinkel E, Ertel M, Dittrich M et al 1985 Kidney size in childhood. Sonographical growth charts for kidney length and volume. Pediatr Radiol 15: 38 2. Rottenberg G T, de Bruyn R, Gordon I 1996 Sonographic standards for single functioning kidney in children. Am J Roentgenol 167: 1255–1259 3. Bridges N A, Cooke A, Healy M J R et al 1993 Standards for ovarian volume in childhood and puberty. Fertil Steril 60: 456–460 4. Cohen H L, Ticc H M, Mandel F S 1990 Ovarian volumes measures by ultrasound: Bigger than we think. Radiology 177: 189 5. Fernbach S K, Maizels M, Conway J J 1993 Ultrasound grading of hydronephrosis: introduction to the system used by the Society for Fetal Urology. Pediatr Radiol 23: 478–480 6. Dudley J A, Haworth J M, McGraw M E, Frank J D, Tizard E J 1997 Clinical relevance and implications of antenatal hydronephrosis. Arch Dis Child Fetal Neonatal Edn 76: F31–F34 7. Yeung C K, Godley M L, Dhillon H K, Duffy P G, Ransley P G 1998 Urodynamic patterns of infants with normal lower urinary tracts or primary vesico-ureteric reflux. Br J Urol 81: 461–467 8. Moorthy I, Easty M, McHugh K, Ridout D, Biassoni L, Gordon I 2005 The presence of vesicoureteric reflux does not identify a population at risk for renal scarring following a first urinary tract infection. Arch Dis Child 90: 733–736 9. Gordon I, Barkovic M, Pinderia S, Cole T J, Woolf A S 2003 Primary vesicoureteric reflux as a predictor of renal damage in children hospitalized with urinary tract infection: a systematic review and metaanalysis. J Am Soc Nephrol 14: 739–744 10. Verrier Jones K 2005 Time to review the value of imaging after urinary tract infection in infants. Arch Dis Child 90: 663–664 11. Lonergan G J, Pennington D J, Morrison J C et al 1998 Childhood pyelonephritis: comparison of gadolinium-enhanced MR imaging and renal cortical scintigraphy for diagnosis. Radiology 207: 377–384 12. Centre for Reviews and Dissemination, University of York 2004 Diagnosing urinary tract infection (UTI) in the under fives. Effective Health Care, vol 8. London, RSM Press 13. National Institute for Clinical Evidence 2007 Urinary tract infection: diagnosis, treatment and long term management of urinary tract infection in children. Department of Health, London 14. Deal J E, Snell M F, Barratt T M et al 1992 Renovascular disease in childhood. J Pediatr 121: 378–384 15. McHugh K, Stringer D A, Hebert D et al 1991 Simple renal cysts in children: diagnosis and follow-up with ultrasound. Radiology 178: 383–385 16. Baxter G M 2001 Ultrasound of renal transplantation. Clin Radiol 56: 802–818 17. Beckwith J B, Kiviat N B, Bonadio J F 1990 Nephrogenic rests, nephroblastomatosis, and the pathogenesis of Wilms’ tumor. Pediatr Pathol 10: 1–36 18. Strouse P J Pediatric renal neoplasms. Radiol Clin North Am 34: 1081–1100

INFLAMMATORY DISEASES OF THE SCROTUM The more common causes of acute pain and/or swelling in the scrotum include torsion of the testicular appendages, testicular torsion, epididymitis with or without orchitis, trauma, acute hydrocoele, incarcerated hernia and acute scrotal oedema. US is the optimal initial imaging investigation. Doppler examination of the testis in young children however is limited, because even modern high-resolution linear array transducers may not be able to detect slow flow in small testes. Testicular torsion shows alteration in echogenicity compared with the contralateral organ and is a surgical emergency. There is absence of flow on the Doppler studies. There may be some surrounding fluid. Epididymitis presents ultrasonically as a swollen hyperechoic epididymis with normal or increased testicular colour flow. If there is doubt about the US findings, then surgical exploration is required. Acute scrotal oedema presents clinically with abrupt onset of a swollen, painful, red scrotal sac. US confirms the thickening of the scrotal wall but shows normal underlying organs. Table 66.7 CAUSES OF INCREASED MEDULLARY ECHOGENICITY IN CHILDREN Nephrocalcinosis • Noniatrogenic, e.g. idiopathic hypercalcaemia in Williams’ syndrome, absorptive hypercalciuria • Iatrogenic, e.g. treatment for hypophosphataemic rickets or frusemide for bronchopulmonary dysplasia or cardiac failure in a premature infant Tubulopathies, e.g. renal tubular acidosis Protein deposits giving transient increased medullary echogenicity in newborns Vascular congestion, e.g. sickle cell anaemia Infection, e.g. candidiasis and cytomegalovirus Metabolic disease, e.g. urate deposits as in Lesch–Nyhan syndrome. Also seen in tyrosinaemia and glycogen storage disease Oxalosis Cystic medullary renal disease, e.g. autosomal recessive polycystic kidney disease

SUGGESTIONS FOR FURTHER READING Carty H, Brunelle F, Stringer D A, Kao S C S (eds) 2005 Imaging children, 2nd edn. Churchill Livingstone, Edinburgh Kuhn J P, Slovis T, Haller J (eds) 2004 Caffey’s pediatric X-ray diagnosis, 10th edn. Mosby, Philadelphia. Gearhart J P, Garrett R A, Rink R, Mouriquant P 2001 Pediatric urology. WB Saunders, Philadelphia

CHAPTER

Skeletal Radiology in Children: Non-traumatic and Non-malignant

67

Amaka C. Offiah and Christine M. Hall

• • • • • • • •

Constitutional disorders of bone Localized disorders of the skeleton Neurocutaneous syndromes Non-inflammatory disorders Metabolic and endocrine disorders Toxic disorders Haemoglobinopathies Infection of the bones and joints

CONSITUTIONAL DISORDERS OF BONE Nomenclature The constitutional disorders of bone include osteochondrodysplasias and dysostoses. Osteochondrodysplasias consist of dysplasias (abnormalities of bone and/or cartilage growth) and osteodystrophies (abnormalities of bone and/or cartilage texture). Abnormalities in the osteochondrodysplasias are intrinsic to bone and cartilage1, and because of gene expression will continue to evolve throughout the lifespan of the individual. Dysostoses occur as a result of altered blastogenesis in the first 6 weeks of intrauterine life. In contrast to the osteochondrodysplasias, the phenotype is fixed, and previously normal bones will remain so. However, more than one bone may be involved. Inevitably there is some overlap between these groups and, from the radiological point of view, when establishing a diagnosis it is useful to consider them together. Most are genetically determined but some malformation syndromes are the result of environmental effects. The most recent International Classification of Osteochondrodysplasias, published in 20021, expanded on the previous classification. The osteochondrodysplasias have been classified into 33 groups (Groups 1–33), and the dysostoses into three (Groups A–C).

Currently more than 2000 malformation syndromes are recognized and many of these will have skeletal abnormalities. In this chapter, only an approach to diagnosis can be given and only the more common conditions will be used as illustrative examples.

Prevalence Although these conditions individually are rare, collectively they form a large group, which is expensive in both medical resources and human care and commitment. An accurate assessment of the prevalence of patients with malformation syndromes and skeletal dysplasias is difficult to achieve. As a rough estimate, approximately 1% of live births have clinically apparent skeletal abnormalities. This figure does not take into account the large numbers of spontaneous abortions or elective terminations, many of which have significant skeletal abnormalities. Nor does it detect those dysplasias presenting only in childhood, or those relatively common conditions that may never present for diagnosis because they are mild, e.g. hypochondroplasia and dyschondrosteosis, both of which merge with normality in individual cases. At orthopaedic skeletal dysplasia clinics in England and Scotland, approximately 10 000 patients are seen; 6000 of whom will require repeated hospitalization for surgical procedures and some will require more prolonged stays in institutions.

Diagnosis Arriving at an accurate diagnosis of these conditions requires a multidisciplinary approach, i.e. combined clinical, paediatric, genetic, biochemical, radiological and pathological (molecular, cellular and histopathological) input. Rapid advances are being made in the field of gene mapping with many conditions being localized to abnormalities at specific loci on individual chromosomes. Also, ‘families’ of conditions are being recognized with many common clinical and radiological features. One example of this is seen in collagen synthesis. For example, about 90% of cases of osteogenesis

1568

SECTION 8

•

PAEDIATRIC IMAGING

imperfecta are due to mutations in genes responsible for Type I collagen. Classification from genetic mutations has proved of value in determining an underlying causative defect. However, this diagnostic approach does not necessarily arrive at a precise clinical diagnosis, with prediction of natural history, morbidity, and mortality and in individual cases this form of classification is conflicting and indeterminate. The most recent classification of skeletal dysplasias recognizes the radiological features as being paramount in accurate diagnosis. Whilst clinical features, such as cleft palate, deafness and myopia, are of diagnostic importance, the onus is still on the radiologist to evaluate the wealth of findings on the radiographic skeletal survey by careful observation and accurate interpretation, and thereby arrive at an accurate diagnosis. An approach to the radiological interpretation of skeletal surveys performed in the clinical context of a suspected dysplasia is available2. Because of the large number of relatively rare conditions, it is difficult or impossible for an individual radiologist to be familiar with every feature of each disorder. In addition, the radiological features of individual conditions will change with time, from infancy to childhood to adulthood, so that each of these conditions may have many age-dependent features. Examples of this temporal change can be seen in spondyloepiphyseal dysplasia congenita (SEDC). Radiological features in the neonate include absent ossification of the pubic rami and short femoral necks (see Fig. 67.6A). However, at the age of 12 years, the pubic rami are present and there is severe coxa vara (see Fig. 67.6C). Another example is seen in Morquio disease (mucopolysaccharidosis, MPS Type IV) in which the capital femoral epiphyses are well ossified at the age of 2 years, but at 8 years are small and flattened, and by 10 years have typically disappeared. Each radiologist’s personal experience of the individual conditions will be limited because of the vast numbers of conditions involved.Textbooks are also of limited value because of the necessarily restricted number of illustrations and obsolescence. For these reasons, skeletal dysplasias and malformation syndromes as a group lend themselves to computer-based applications and in particular to computer-assisted diagnosis.These tools should be used to help improve our diagnostic abilities. Computer assistance may take the form of menu-driven databases of clinical and radiological features. A number of findings in a particular case can be matched and a group of conditions selected from the database for further consideration before arriving at a diagnosis, e.g. the Radiological Electronic Atlas for Malformation Syndromes (REAMS)3. This approach is really an automated method for cross-referencing gamuts. An alternative method is by means of a knowledge-based expert system in which experts in the field lead the user (the general radiologist) with a series of questions through a differentiation strategy to arrive at a diagnosis. Either system can be linked to a computerized image database capable of illustrating many thousands of images. A specific diagnostic label should only be attached to a patient when it is secure. An inaccurate diagnosis may have a profound effect upon the family in terms of genetic counselling, and upon the patient in terms of management and outcome.

There may be a need to monitor and re-evaluate the evolution of radiological findings over time before establishing a diagnosis. A significant proportion of cases (approximately 30%) are unclassifiable because the combination of findings does not conform to any recognized condition. It is important that data, both clinical (Table 67.1) and radiological and specimens, such as bone, tissue and blood, are stored and that the information can be widely disseminated. Only in this way will it be possible to ‘match’ conditions and to establish the natural history of a disorder.

Prenatal diagnosis In the USA prenatal ultrasound (US) is less formal than in Europe; however, its use has been steadily increasing over the past two decades. By comparison, most pregnant women in the UK undergo prenatal US screening at between 14 and 18 weeks’ gestation. All neonatally lethal skeletal dysplasias may be diagnosed at this stage by demonstrating short limbs, bowed limbs, or a narrow thorax. Where there has been a previously affected sibling, specific malformations such as polydactyly, polycystic kidneys, or micrognathia may be assessed. Skeletal US findings are highly significant but are not very specific, and in general it is unwise to offer a precise diagnosis on the basis of US findings alone. Pregnancy terminations offered on the grounds of such US findings should also have subsequent radiological evaluation to determine the precise diagnosis, and before genetic counselling is offered. Many nonlethal conditions which may present at birth can also be ascertained on prenatal US. Fetal anomaly US examinations are offered to parents of previous babies who have had congenital dysplasias or malformation syndromes, and to at-risk parents with high maternal or paternal ages, or with specific environmental exposures. Occasionally, other imaging techniques may help to confirm a suspected diagnosis prenatally. Maternal abdominal radiographs are not recommended. Too often the fetus is in such a position that diagnostic radiographic details are obscured and misdiagnosis is possible. Nor is the radiation risk to the fetus and mother justified, especially if the fetus is normal or is a potential survivor with a good quality of life. Magnetic resonance imaging (MRI) is being used occasionally for in utero evaluation of specific anomalies when a sibling has suffered from a particular syndrome. Chorionic villus sampling can be used for biochemical evaluation of fetuses at risk from storage disorders in those parents with a previously affected infant. Fetal chromosomal Table 67.1 CLINICAL DATA USED IN THE DIAGNOSIS OF SKELETAL DYSPLASIAS AND MALFORMATION SYNDROMES Stature—proportionate, disproportionate Abnormal body proportion—short trunk, short limbs Abnormal limb segments—rhizomelic, mesomelic, acromelic Local anomalies and deformities—cleft palate, polydactyly Facies—dysmorphology Other—hearing, sight, mental retardation Temporal changes Pedigree

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

analysis from skin biopsy can be assessed when a sibling or carrier parent is affected.

Imaging Making the diagnosis In addition to prenatal US, a skeletal survey should be performed on any pregnancy termination resulting from a US diagnosis of short-limbed dwarfism or significant malformation. Also, a skeletal survey should be performed on any stillbirth. Although this may involve antero-posterior (AP) and lateral ‘babygrams’ of the entire infant, ideally additional views of the extremities should be performed. Spontaneous abortions should also have a radiographic skeletal survey. However, this can rarely be achieved. After birth, a standard full skeletal survey is indicated when attempting to establish a diagnosis for short stature or for a dysmorphic syndrome. This should include: • AP and lateral skull to include the atlas and axis • AP chest • AP pelvis • AP lumbar spine • lateral thoracolumbar spine • AP one lower limb • AP one upper limb • postero-anterior (PA) one hand (usually the left; allows bone age assessment). Occasionally, additional views will be required, particularly with specific clinical abnormalities, and these may include views of the feet, e.g. if polydactyly is present, or views of the cervical spine if cervical instability is suspected with specific diagnoses, or both upper and lower limbs if asymmetry or deformity is a clinical feature. If a diagnosis cannot be established, then (limited) followup skeletal surveys are indicated, e.g. at 1 and 3 years, to evaluate progression and evolution of radiographic appearances. Occasionally a technetium radionuclide skeletal scintigram showing the photon-deficient area of avascular necrosis (AVN) may be of value in differentiating bilateral Perthes disease from the small fragmented capital femoral epiphyses of multiple epiphyseal dysplasias. In this regard, contrast-enhanced MRI is also useful. Conditions with decreased bone density may be assessed and monitored by means of bone densitometry.

Assessing complications When a confident diagnosis is established, further imaging is essential to monitor the progress of potential complications. Complications may result as part of the natural evolution of the condition, but also may occur secondary to medical intervention. Radiography and MRI of the cervical spine in flexion and extension will monitor instability; AP and lateral views of the spine will monitor kyphosis and scoliosis. In some conditions, progressive hip subluxation, genu varum or genu valgum may be problematic. Long limb radiographic views are sometimes used to assess asymmetry, genu varum, and genu valgum, and to monitor progression. Computed tomography (CT) is also of value in the measurement of limb asymmetry.

CT or MRI may monitor the development of hydrocephalus or the presence of neuronal migration defects and structural defects, such as the absence of the corpus callosum. CT may also demonstrate encroachment on the cranial nerve foramina and both CT and MRI are of value in assessing spinal cord compression. US can demonstrate associated organ anomalies, e.g. cystic disease of the kidneys or hepatosplenomegaly and echocardiography may reveal associated intracardiac abnormalities. Arthrography, US, CT and MRI are of value in the assessment of joint problems, particularly when surgical intervention is proposed. Technetium skeletal scintigrams are occasionally used to determine the extent of bony involvement in specific disorders, e.g. the number of exostoses can be demonstrated in diaphyseal aclasis. However, it has been shown that in other patchy disorders, such as fibrous dysplasia, radiographically affected areas may not demonstrate abnormal uptake of radionuclide.

Postoperative imaging US has a place in assessing the development of new bone formation following osteotomies and limb-lengthening procedures and plain radiography and CT in confirming the correct alignment in this situation. All imaging investigations are brought into play in the assessment of a patient following bone marrow transplantation.

Management Only when an accurate diagnosis has been established can the prognosis and natural history of the disorder be given. For example, myopia can be corrected and retinal detachment prevented in Stickler syndrome (hereditary arthro-ophthalmopathy); cord compression can be prevented in conditions with instability in the cervical spine (Table 67.2) or with progressive thoracolumbar kyphosis and spinal stenosis, as seen in achondroplasia. In some conditions, cure may be achieved, e.g. in the severe form of osteopetrosis (which is lethal in childhood and results in cranial nerve compressions leading to blindness in the first year of life), by means of a compatible bone marrow transplant. This not only improves life expectancy and arrests cranial nerve compression, but the radiographic changes revert to normal and the predisposition to fractures resolves. In this condition, the diagnosis needs to be established within the first 6 months of life if most complications are to be prevented. Bone marrow transplantation has also been used with some success in treating selected patients with the mucopolysaccharidoses.The skeletal abnormalities persist but quality of life and life expectancy are improved. This treatment has resulted in iatrogenic manipulation of the natural history of the mucopolysaccharidoses and, with increased life expectancy, later complications, often the result of spinal cord compression, are now being recognized. In many conditions, orthopaedic procedures are invaluable in maintaining or improving mobility. For example, osteotomies prevent or correct dislocations or long bone bowing deformities. Patients with osteogenesis imperfecta may require multiple osteotomies to correct severe deformities, as well as intramedullary

1569

1570

SECTION 8

•

PAEDIATRIC IMAGING

Table 67.2 DISORDERS WITH INSTABILITY IN THE CERVICAL SPINE

Table 67.3 ASYMMETRIC SHORTENING (OR OVERGROWTH)

Cervical spine instability with odontoid peg absence or hypoplasia

Maffucci syndrome

Ollier’s disease (multiple enchondromatosis)

Morquio disease (MPS Type IV)

Polyostotic fibrous dysplasia

Other mucopolysaccharidoses (MPS)

McCune–Albright syndrome

Mucolipidoses (MLS)

Melorheostosis

Achondroplasia

Neurofibromatosis

Down’s syndrome

Diaphyseal aclasis (multiple exostoses)

Metatropic dysplasia

Chondrodysplasia punctata

Kniest dysplasia

Beckwith–Wiedemann syndrome

Diastrophic dysplasia

Klippel–Trenaunay syndrome

Dyggve–Melchior–Clausen disease

Silver–Russell syndrome

Metaphyseal chondrodysplasia, Type McKusick

Dysplasia epiphysealis hemimelica

Spondyloepiphyseal dysplasia congenital (SEDC)

Sturge–Weber syndrome

Multiple epiphyseal dysplasia

Proteus syndrome

Pseudochondroplasia

Hypomelanosis of Ito

Chondrodysplasia punctata

Epidermoid nevus syndrome

Cervical spine instability with cervical kyphosis (C2C3) Diastrophic dysplasia SEDC Lethal Atelosteogenesis Campomelic dysplasia

rodding to reduce fractures, maintain alignment and provide support and stability.These patients suffer from basilar invagination resulting in compression of the brain stem. Surgical intervention may prevent severe neurological impairment (see Fig. 67.19D). Spinal deformities, kyphosis and scoliosis are common in the constitutional disorders of bone. Prevention and treatment consists of spinal bracing and timely arthrodeses or laminectomies for cord compression. Joint replacements may be necessary, especially in those dysplasias, such as multiple epiphyseal dysplasia, in which major involvement of the epiphyses may result in premature osteoarthritis. In some conditions, limb-lengthening procedures may be appropriate to improve mobility. This is usually offered in disorders with asymmetric shortening (Table 67.3; Fig. 67.1), but is sometimes offered to selected patients with achondroplasia (Fig. 67.2) or other short-limbed dysplasias for cosmetic reasons. Achondroplasia has proved particularly amenable to limb-lengthening procedures because redundant soft tissues are a feature of this dysplasia and insufficient soft tissue has proved to be a limiting factor in lengthening procedures in other conditions. An increase of approximately 30% in the length of the long bones may be achieved in achondroplasia, compared with 15% in other disorders. Only with an accurate diagnosis in those conditions presenting at birth can those that are likely to be lethal be predicted. The diagnosis of a lethal dysplasia can prevent unnecessary and distressing prolongation of life, help to reduce parental expectations and anguish and help save on economic resources. Termination of pregnancy may be offered with prenatally diagnosed lethal conditions, or where there is intrauterine evidence of short limbs. When a sibling has suffered from a

Figure 67.1

Ollier’s disease. Leg lengthening with fixator in position.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

Figure 67.2 Achondroplasia in a neonate. (A) Sloping metaphyses, oval transradiant proximal femora, small square iliac wings with horizontal acetabular roofs, and a narrow thorax with short ribs. (B) Short skull base with prominent frontal bone and narrow cervical canal. (C) Small square iliac wings, horizontal acetabular roofs, short sacrosciatic notches, progressive caudal narrowing of the lumbar interpedicular distances and low set sacrum. (D) Mild kyphosis, posterior scalloping of the vertebral bodies and short pedicles with spinal stenosis.

disabling condition associated with particular malformations, these may be specifically looked for prenatally in subsequent pregnancies.This practice is leading to a change in the incidence of certain conditions formerly presenting at birth. With the identification and localization of specific chromosomal abnormalities associated with particular disorders, the development of gene therapy for clinical use poses many challenges and offers great potential for the future. Growth hormone therapy is used in selected disorders to influence final height. Growth hormone stimulates Type I collagen production and is being used, in particular, to augment growth rate in children with osteogenesis imperfecta. Bisphosphonates are pyrophosphate analogues that inhibit osteoclast function. They have been used in osteogenesis imperfecta to improve bone density, and have also been used to treat bone pain and osteopaenia in a variety of rheumatological and dermatological conditions.

Genetic counselling When an accurate diagnosis has been made, meaningful genetic counselling can be given, both to the parents and to the affected individual. Most conditions are inherited in an autosomal dominant or autosomal recessive manner. In conditions with an autosomal dominant inheritance, the affected individual has a one in two chance of passing the same abnormality on to his/her offspring. However, many of these conditions arise as a spontaneous mutation, which means that the parents of the affected individual, who are themselves normal, have an extremely low risk of having another affected child. In autosomal recessive conditions, both the parents are carriers of the disorder, but are not affected themselves, and they have a one in four chance of having another affected child. Other important, although uncommon, modes of inheritance are the result of somatic or gonadal mosaicism. Mosaicism is the presence of at least two cell lines in a single individual

1571

1572

SECTION 8

•

PAEDIATRIC IMAGING

or tissue that derive from a single zygote. Somatic mosaicism for autosomal dominant conditions results in asymmetric or patchy disorders (see Table 67.3). It is thought that when not a mosaic, these disorders are lethal. Clinical evidence of somatic mosaicism includes asymmetry, localized overgrowth, pigmentation and haemangiomas.

Osteochondrodysplasias The clinical and radiographic features of selected osteochondrodysplasias are described in Table 67.4.

Table 67.4 CLINICAL AND RADIOGRAPHIC FEATURES OF SELECTED OSTEOCHONDRODYSPLASIAS Clinical features

Radiological features

Achondroplasia

Common Short limbs, short trunk Narrow thorax with respiratory distress in infancy Bowed legs Prominent forehead with depressed nasal bridge Hydrocephalus and brainstem and spinal cord compression Inheritance: autosomal dominant

‘Bullet-shaped’ vertebral bodies Decrease in interpedicular distance in lumbar spine caudally Short vertebral pedicles Posterior vertebral body scalloping Squared iliac wings with small sciatic notch Flat acetabular roofs Short ribs Short wide tubular bones Relative overgrowth of fibula Large skull vault, relatively short base Small foramen magnum Dilatation of lateral cerebral ventricles V-shaped notches in growth plates (chevron deformity)

Hypochondroplasia

Variable short stature Prominent forehead Inheritance: autosomal dominant

No normal widening in the interpedicular distance in the lumbar spine caudally Short relatively broad long bones Elongation of the distal fibula and ulnar styloid process Variable brachydactyly

Thanatophoric dysplasia

Most common lethal neonatal skeletal dysplasia Short markedly curved limbs Respiratory distress, small thoracic cage Inheritance: sporadic, autosomal dominant mutation

Short ribs with wide costochondral junctions Severe platyspondyly Horizontal acetabular roofs with medial spikes Small sacroiliac notches Marked shortness and bowing of the long bones, ‘telephone receiver femora’ Irregular metaphyses Short broad tubular bones in the hands and feet Small scapulae May be associated with ‘clover leaf skull’

Short limbs Relatively narrow chest Small appendage in coccygeal region (tail) Progressive kyphoscoliosis Progressive change to a relatively short trunk Inheritance: variable, autosomal dominant or recessive

Short tubular bones with marked metaphyseal widening (dumb-bell) Platyspondyly Relatively large intervertebral discs Flat acetabular roofs Short iliac bones Short ribs with anterior widening Progressive kyphoscoliosis Hypoplastic odontoid process

Asphyxiating thoracic dysplasia (Jeune’s; Fig. 67.4)

Often lethal Respiratory problems with long narrow thorax Short hands and feet Nephronophthisis in later life in survivors Inheritance: autosomal recessive

Small thorax with short ribs, horizontally orientated Widened costochondral junctions High clavicles Iliac bones short Horizontal acetabula with medial and lateral ‘spurs’ (trident) Premature appearance of the proximal femoral ossification centres Cone-shaped epiphyses in phalanges May have polydactyly

Ellis–van Creveld (chondroectodermal dysplasia; Fig. 67.5)

Short stature Short limbs, more marked distally Polydactyly

Short ribs in infancy Short iliac wings; horizontal acetabulum with medial and lateral spurs (trident)

Group 1 (achondroplasia group)

Group 3 (metatropic dysplasia group) Metatropic dysplasia (Fig. 67.3)

Group 4 (short rib dysplasias, with or without polydactyly)

Continued

CHAPTER 67

Table 67.4 cont’d

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

CLINICAL AND RADIOGRAPHIC FEATURES OF SELECTED OSTEOCHONDRODYSPLASIAS Clinical features

Radiological features

Hypoplasia of the nails and teeth Ectodermal dysplasia with sparse hair Congenital cardiac defects, atrial septal defect, single atrium Fusion between upper lip and gum Inheritance: autosomal recessive

Premature ossification of the femoral capital epiphyses Pelvis becomes more normal in childhood Laterally sloping proximal tibial metaphysis Exostosis of the medial upper tibial shaft Carpal fusions Cone-shaped epiphyses, middle phalanges Polydactyly of hands and feet

Spondyloepiphyseal dysplasia congenita (SEDC) (Fig. 67.6)

Short stature with short trunk at birth Cleft palate Myopia Maxillary hypoplasia Thoracic kyphosis and lumbar lordosis Barrel-shaped chest Inheritance: autosomal dominant

Ovoid, pear-shaped irregular-sized vertebral bodies in infancy Irregular platyspondyly in later life Odontoid hypoplasia and cervical spine instability Short long bones Absent ossification of epiphyses of knees, talus, calcaneus at birth Pubic and ischial hypoplasia Hands relatively normal Severe coxa vara developing in early childhood Horizontal acetabulum

Spondyloepiphyseal dysplasia tarda (Fig. 67.7). Several types with different modes of inheritance are recognized

Usually presents in adolescence with short stature due to short trunk Joint limitation

Platyspondyly Characteristic mound of bone in central and posterior part of the vertebral end plates in the X-linked dominant type Narrow intervertebral discs Mild to moderate generalized epiphyseal dysplasia Early osteoarthritis

Pseudoachondroplasia (Fig. 67.8)

Short limbs with normal head and face Accentuated lumbar lordosis Genu valgum (knock knees) or genu varum (bow legs) Joint hypomobility Inheritance: autosomal dominant

Platyspondyly with tongue-like anterior protrusion of the vertebral bodies Biconvex configuration of upper and lower vertebral end plates Atlantoaxial dislocation Small femoral capital epiphyses Short iliac bones Wide Y-shaped cartilage Irregular acetabulum Small pubis and ischium Pointed proximal metacarpals Shortening of the tubular bones with expanded, markedly irregular metaphyses Small irregular epiphyses Wide costovertebral joints Relatively long distal fibula

Multiple epiphyseal dysplasia (Fig. 67.9)

Stiffness in joints with limp Early osteoarthritis Mild shortening of limbs Inheritance: autosomal dominant

Delayed ossification and irregularity of the epiphyses of the tubular bones Delayed ossification of carpus and tarsus Short tubular bones of the hands and feet Double-layered patella Only mild irregularity of the vertebral end plates Mild wedging of the vertebral bodies Mild acetabular hypoplasia Early joint degenerative changes

Flat nasal bridge, high arched palate Cutaneous lesions such as ichthyosis Asymmetrical or symmetrical shortening of the limbs Joint contractures Cataracts Inheritance: autosomal dominant, autosomal recessive, X-linked

Stippled calcification in cartilage, particularly around the joints and in laryngeal and tracheal cartilages Shortening, symmetrical or asymmetrical, of the long bones Short digits in some types Vertebral bodies show coronal clefting Eventual disappearance of the stippled calcification in later life Punctate calcification may also be seen in Pacman dysplasia, Zellweger syndrome, fetal alcohol and warfarin embryopathies, and in some chromosomal abnormalities and mucolipidoses (Fig. 67.11)

Group 8 (Type II collagenopathies)

Group 11 (multiple epiphyseal dysplasias and pseudoachondroplasia)

Group 12 (chondrodysplasia punctata/stippled epiphyses) Chondrodysplasia punctata (CDP) (Fig. 67.10)

Continued

1573

1574

SECTION 8

•

PAEDIATRIC IMAGING

Table 67.4 cont’d

CLINICAL AND RADIOGRAPHIC FEATURES OF SELECTED OSTEOCHONDRODYSPLASIAS Clinical features

Radiological features

Short limbs, short stature, presenting in early childhood Genu varum (bow legs) Inheritance: autosomal dominant

Metaphyseal flaring Irregular widened growth plates, most marked at the hips Increased density and unevenness of metaphyses, particularly of upper femora and around knees Large femoral capital epiphyses Coxa vara Femoral bowing Anterior cupping of the ribs Spine normal

Large head Large fontanelles with delay in closure Multiple supernumerary teeth Excessive mobility of the shoulders Narrow chest Inheritance: autosomal dominant

Frontal bossing Wide sutures of the skull Multiple wormian bones Persistently open anterior fontanelle Prominent jaw with multiple supernumerary teeth Variable hypoplasia or pseudoarthrosis of the clavicles Small scapulae Absent or delayed ossification of the pubic bones Hypoplastic iliac wings Short middle phalanges with cone-shaped epiphyses and tapering terminal phalanges Undermodelling of the shafts of the long bones Pseudarthrosis of long bones (rare)

Neonatal Respiratory distress Cleft palate Prenatal bowing of lower limbs Pretibial dimpling

11 pairs of ribs Hypoplastic scapulae Angulation of femora at junction of proximal third and distal two-thirds Angulation of tibiae at junction of proximal two-thirds and distal third

Group 13 (metaphyseal dysplasias) (Figs 67.12–67.14) Metaphyseal chondrodysplasia (Schmid) (Fig. 67.13)

Group 19 (dysplasias with predominant membranous bone involvement) Cleidocranial dysplasia (Fig. 67.15)

Group 20 (bent bone dysplasias) Campomelic dysplasia

Survivors Short stature Learning difficulties Recurrent respiratory infections Inheritance: autosomal dominant (sex reversal has been reported)

Short fibulae Progressive kyphoscoliosis Dislocated hips Deficient ossification of the ischium and pubis Hypoplastic patellae

Group 22 (dysostosis multiplex) Mucopolysaccharidoses (Figs 67.16, 67.17). This group of conditions is characterized by an abnormality of mucopolysaccharid and glycoprotein metabolism. Differentiation between the types is dependent upon laboratory analysis particularly of urine, leukocytes and fibroblastic cultures

Presenting typically in early childhood, the clinical manifestations are variable Short stature isassociated with a distinctive coarse facial appearance with mental retardation, corneal opacities, joint contractures, hepatosplenomegaly and cardiovascular problems Inheritance: autosomal recessive except Type II (Hunter) which is X-linked recessive

Macrocephaly Thick vault with ‘ground-glass’ opacity ‘J’-shaped sella Wide ribs, short wide clavicles, poorly modelled scapulae Ovoid, hook-shaped vertebral bodies with thoracolumbar gibbus Odontoid hypoplasia Iliac wings flared with constricted bases to the iliac bones Small, irregular femoral capital epiphyses Coxa valga Lack of normal modelling of the long bones Thin cortices Coarse trabecular pattern Short wide phalanges with characteristic proximal pointing of the metacarpals Neurological changes include hydrocephalus, leptomeningeal cysts and a variety of abnormalities best demonstrated by MRI

Morquio’s syndrome (MPS-IV)

Normal intelligence Joint laxity Knock knees Short stature

Absent odontoid peg with cervical instability leading to spinal cord compression Platyspondyly Anterior central ‘beak’ or ‘tongue’ of vertebral bodies Iliac wings flared with constricted bases of the iliac bones Progressive disappearance of femoral capital epiphyses Coxa valga Genu valgum Irregular ossification of metaphyses of long bone Small irregular epiphyses Proximal pointing of second to fifth metacarpals Continued

CHAPTER 67

Table 67.4 cont’d

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

CLINICAL AND RADIOGRAPHIC FEATURES OF SELECTED OSTEOCHONDRODYSPLASIAS Clinical features

Radiological features

See Table 67.5.

See Table 67.5.

Additional clinical findings Hydrocephalus Large fontanelles not related to hydrocephalus Kyphoscoliosis Pseudarthrosis Ligamentous laxity Easy bruising Thin skin

Additional radiological findings

Group 24 (dysplasias with decreased bone density) Osteogenesis imperfecta—a group of conditions caused by an abnormality of Type 1 collagen. Several forms are recognized (Table 67.5; Fig. 67.18)

Wormian bones Basilar invagination Tam O’Shanter skull Irregular calcific ‘popcorn’ lesions in metaphyses Hyperplastic callus Severe protrusio acetabuli Lumbar pedicles elongated ‘Codfish’ vertebral bodies Generalized osteoporosis

Group 26 (increased bone density without modification of bone shape) Osteopetrosis (Fig. 67.19). There are several types

Enlargement of the liver and spleen Bone fragility with fracture Cranial nerve palsies Blindness Osteomyelitis Anaemia Inheritance: more severe types are autosomal recessive, although a milder delayed form shows autosomal dominant inheritance

Generalized increase in skeletal density Abnormal modelling of the metaphyses, which are widened with alternating bands of radiolucency and sclerosis Bone within bone appearance Rickets and basal ganglia calcification in a recessive form with carbonic anhydrase deficiency

Pyknodysostosis (Fig. 67.20)

Short limbs Propensity to fracture Respiratory problems Irregular dentition Inheritance: autosomal recessive

Wormian bones Delayed closure of the fontanelles Generalized increase in density of the skeleton Straight mandible Prognathism Deficient ossification of distal phalanges Re-absorption of lateral end of clavicle Pathological fractures

Osteopoikilosis (Fig. 67.21)

Often asymptomatic May present with cutaneous/subcutaneous nodules

Sclerotic foci (islands), especially around the pelvis and metaphyses of long bones

Melorheostosis (Fig. 67.22)

Sclerodermatous lesions over the bony lesions Asymmetry of the affected limbs Vascular anomalies Abnormal pigmentation Muscle contractures and wasting Inheritance: nongenetic

Dense cortical longitudinal hyperostosis of the bones Appearances like wax running down the side of a candle Particularly involves the long bones, less commonly other bones

Muscle weakness Pain in the extremities Gait abnormalities Exophthalmos Inheritance: autosomal dominant

Sclerotic skull base Progressive endosteal and periosteal diaphyseal sclerosis Narrowing of medullary cavity of tubular bones Bone images: increased activity

Usually appears in the first 5 months of life Hyperirritability Soft tissue swelling Inheritance: autosomal dominant suggested in some families (also a lethal recessive form)

Commonly affects mandible, clavicles, ulnas May be asymmetrical Periosteal new bone and cortical thickening When tubular bones are affected, abnormality is limited to the diaphyses

Group 27 (increased bone density with diaphyseal involvement) Diaphyseal dysplasia (Englemann disease)

Group 30 (neonatal severe osteosclerotic dysplasias) Caffey disease (infantile cortical hyperostosis)

Group 31 (disorganized development of cartilaginous and fibrous components of the skeleton) Multiple cartilaginous exostoses (diaphyseal aclasis) (Fig. 67.23)

Proximal pointing of second to fifth metacarpals

Multiple exostoses particularly at the ends of the long bones, ribs, scapulae and iliac bones Secondary deformity and joint limitation Ulnar deviation of the hands Inheritance: autosomal dominant

Multiple flat or protuberant exostoses Secondary joint deformities Exostoses of the iliac crest and scapula Vertebral involvement rare Cranial vault spared Continued

1575

1576

SECTION 8

•

PAEDIATRIC IMAGING

Table 67.4 cont’d

CLINICAL AND RADIOGRAPHIC FEATURES OF SELECTED OSTEOCHONDRODYSPLASIAS Clinical features

Radiological features Long bone exostoses point away from the adjacent joint Short ulna distally (reverse Madelung deformity) Sarcomatous change to be suspected if rapid increase in size, or pain

Enchondromatoses (± haemangiomas) (Fig. 67.24)

Asymmetrical limb shortening Expansion of affected bones Occasional pathological fracture Absence of haemangiomas (Ollier’s disease) Presence of haemangiomas (Maffuci syndrome) Malignancy rare in Ollier’s disease, 15% incidence in Maffuci syndrome Inheritance: nongenetic

Shortening of affected long bones Rounded and streaky radiolucencies, particularly in the metaphyses Expansion of the bone with cortical thinning Areas of calcification within the lesions Pathological fractures Joint deformity Typically markedly asymmetrical Short ulna distally (reverse Madelung deformity) Calcified haemangiomas (Maffuci syndrome, not usually seen until adolescence)

Fibrous dysplasia (Fig. 67.25). May involve one bone only (monostotic) or multiple bones (polyostotic). The association of polyostotic fibrous dysplasia, patchy café-au-lait skin pigmentation and sexual precocity, usually in girls, constitutes the McCune–Albright syndrome

Deformity and pain related to involved bones Pathological fracture Hypophosphataemic osteomalacia or rickets Usually presents in childhood Inheritance: nongenetic

Skull often shows asymmetrical thickening of the vault with sclerosis at the base: multiple rounded areas of radiolucency Obliteration of the paranasal air sinuses Marked facial deformity (leontiasis ossea) Disruption of dentition ‘Ground-glass’ or radiolucent area of trabecular alteration in the long bones associated with patchy sclerosis and expansion, with cortical thinning and endosteal scalloping Pathological fractures and deformities due to softening, e.g. ‘shepherd’s crook’ femoral necks Localized or asymmetrical overgrowth Secondary spinal stenosis

Table 67.5

OSTEOGENESIS IMPERFECTA CLINICAL (BASED ON THE SILLENCE CLASSIFICATION) AND RADIOLOGICAL FINDINGS I

II

III

IV

1:30 000

1:30 000

Rare

Unknown (rare)

Severity

Mild

Lethal

Severe

Mild/moderately severe

Death

Old age

Stillborn

By 30 years

Old age White

Clinical findings Incidence

Sclerae

Blue

Blue

Blue, then grey

Hearing impairment

Frequent

—

Rare

Rare

Teeth (dentinogenesis ) imperfecta

IA Normal

—

Abnormal

IVA normal

IB Abnormal

—

Stature

Normal

—

Short

IVB abnormal Normal/mildly short

Inheritance

Autosomal dominant

Autosomal dominant

Autosomal dominant/ autosomal recessive

Autosomal dominant

Rare

Radiological findings Fractures at birth

60⬚

B

Perkin's line

Putti's Triad 1. Increased acetabular angle Hilgenreiner's line

2. Small epiphysis/ Delayed ossification

Increased 'medial gap'

3. Superolateral displacement (lateral to Perkin's line) Disrupted Shenton's line

D Figure 67.29 Developmental dysplasia of the hip (DDH). (A) Ultrasound of a dislocated hip. (B) Normal Graf angles. (C) AP radiograph of the pelvis with left DDH. (D) Measurements in DDH. (E) AP radiograph of the pelvis showing bilateral DDH. Note the early formation of pseudoacetabulae.

Table 67.7

GRAF ANGLES

Type

α angle (degrees)

β angle (degrees)

Bony roof

Ossific rim

Ia

> 60

< 55

Good

Sharp

Covers femoral head

Mature

Ib

> 60

> 55

Good

Usually blunt

Covers head

Mature

IIa

50–59

> 55

Deficient

Rounded

Covers head

Physiological ossification delay

Cartilage roof

IIb

50–59

> 55

Deficient

Rounded

Covers head

IIc

43–49

< 77

Deficient

Rounded/flat

Covers head

Interpretation

IId

43–49

> 77

Severely deficient

Rounded/flat

Compressed

On point of dislocation

IIIa

< 43

> 77

Poor

Flat

Displaced up

Dislocated

IIIb

< 43

> 77

Poor

Flat

Displaced up

Echo poor Dislocated

Reflective IV

< 43

> 77

Poor

Flat

Interposed

Dislocated

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

Figure 67.31 Focal fibrocartilaginous dysplasia. Note the pathognomonic appearance of the radiolucent band. This condition is usually selfresolving and in the first instance should be managed conservatively

Figure 67.30 Femoral dysplasia. Right coxa vara deformity.

Figure 67.32 Blount’s disease. (A) Plain radiograph, (B) coronal CT and (C) 3D CT reconstruction (posterior view).

1591

1592

SECTION 8

•

PAEDIATRIC IMAGING

and varus of the forefoot. It results from abnormal development around the ninth week of gestation. Aetiological considerations include genetic factors and early amniocentesis (before 11 weeks)5,6. It occurs two to three times more commonly in boys. Useful measurements are summarized in Table 67.8.

Idiopathic avascular necrosis of the femoral head (Perthes disease) Osteonecrosis of the femoral head usually presents with pain or limping between 5 and 8 years of age. It is most often unilateral, but when bilateral (in 15%) it is asymmetrical, helping to distinguish it from an epiphyseal dysplasia. There are four stages of disease: devascularization; collapse and fragmentation; re-ossification; and finally the stage of remodelling. The earliest radiographic feature is that of a radiolucent subchondral fissure—the crescent sign (Fig. 67.33A). Disease progresses with loss of height, fragmentation and sclerosis of

Table 67.8

the femoral head. A coxa magna deformity may ensue, with lateral uncovering of the capital femoral epiphysis. There may be associated irregularity of the acetabular margin. Several radiographic signs and classification systems exist to help in the diagnosis and prognosis of Perthes disease. The extent of subchondral fracture is said to be a good predictor of the final outcome7. More detailed classification systems include that proposed by Catterall8 and later simplified by Salter and Thompson9, and the more recent system proposed by Herring et al10. The last of these is based on the degree of resorption of the lateral femoral head pillar during the fragmentation phase. All systems have been found to be good when used by an experienced observer11. The centre edge angle may be used to assess the degree of femoral head coverage. Perthes disease is occasionally bilateral and synchronous, but it is more usually metachronous (Fig. 67.33B). Synchronous disease should raise the suspicion of an epiphyseal dysplasia.

DIAGNOSIS OF TALIPES

Deformity

DP radiograph

Lateral radiograph

Hind foot varus

Talocalcaneal angle: < 15 degrees

Talocalcaneal angle: < 25 degrees

Midtalar line lateral to first metatarsal base Hind foot valgus

Talocalcaneal angle: > 50 degrees in newborns; > 40 degrees in older children

Talocalcaneal angle: > 50 degrees in newborns; > 45 degrees in older children

Midtalar angel medial to first metatarsal base Hind foot equines

—

Calcaneotibial angle: > 90 degrees Plantar flexion of calcaneus

Hind foot calcaneus

—

Calcaneotibial angle: < 60 degrees Dorsiflexioun of calaneus

Forefoot varus

Narrow with increased overlap of metatarsal bases

Forefoot valgus

Broad with reduced overlap of metatarsal bases

Fifth metatarsal most plantar (normal) First metatarsal most dorsal First metatarsal most plantar

Figure 67.33 Perthes disease at (A) presentation, showing the crescent sign, and (B) 17 months later. Note the metachronous nature of the disease.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

Perthes disease may be suspected from US examination. Capsular distension from a hip effusion persisting for longer than 6 weeks is associated with the development of Perthes disease12. Additionally, irregularity/fragmentation of the capital femoral epiphysis and poor coverage of the femoral head may be demonstrated. While skeletal scintigraphy is highly sensitive and specific for detecting AVN, MRI has now largely replaced it (Fig. 67.34). Signal abnormalities may be detected within the first 3 months of disease onset; however, they are more clearly visualized 3−8 months after onset13. Six patterns of signal abnormality have been described14. In summary, T1-weighted images show low signal intensity, compared to high signal on T2-weighted fatsuppressed/inversion recovery (STIR) sequences. In those with normal signal intensity or complete loss of signal on both sequences (dead bone), intravenous enhancement is not necessary. In those with normal/low T1-weighted and high T2-weighted signal, intravenous contrast medium will identify areas of viable bone. The role of dynamic contrast MRI in the diagnosis of Perthes disease is under evaluation15,16.

Slipped capital femoral epiphysis This is the most common hip disorder of adolescence. There is no change in the relationship between the femoral head and the acetabulum–anterolateral and rotational forces of the hip muscles on the femoral shaft result in anterosuperior translation of the proximal femoral metaphysis relative to the epiphysis. By definition, it occurs through the nonrachitic physis17. It is more common in boys, in those of Afro-American origin, and in the obese.18 The age range for girls is 11–12 years and for boys 12–14 years. It most commonly occurs at the time of the pubertal growth spurt at Risser grade 0, and

is rarely seen in girls after menarche or in boys after Tanner stage 4. Bilateral slips occur in about 25% of Caucasian and up to 50% of Afro-American children (depending on the reported series). When unilateral, the left side is more often involved (65%). Endocrine disorders associated with SCFE include hypothyroidism, growth hormone deficiency, hypogonadism and panhypopituitarism19,20. Clinically SCFE may be classified based either on duration of symptoms—acute (symptoms for less than 3 weeks); chronic (symptoms for more than 3 weeks); or acute on chronic (symptoms for more than 3 weeks with an acute exacerbation). The second system based on patient mobility classifies SCFE as either stable (patient able to walk with or without crutches) or unstable (patient unable to walk with or without crutches)17. The condition is most commonly of chronic onset18. Radiography remains the investigation of choice. AP and frog lateral views of the pelvis are obtained. Observations include disuse osteopenia; a widened growth plate with indistinct borders; and malalignment of the epiphysis and proximal femoral metaphysis (more easily recognized on the frog lateral radiograph). On an AP radiograph malalignment may be more objectively assessed by drawing a line (Klein’s line) along the outer border of the femoral neck—when extended upwards, this line should intersect approximately a sixth of the femoral epiphysis. In cases of SCFE, the Klein’s line may not intersect the proximal femoral epiphysis (Fig. 67.35). The frog lateral radiograph allows an assessment of the severity of the slip according to the method outlined by Southwick21 in which a ‘slip angle’ is calculated (Fig. 67.36). Based on the slip angle obtained, SCFE can be classified as mild (≤ 30 degrees), moderate (31–50 degrees), or severe (≥ 51 degrees)22. Some authors advocate the use of US over radiography for the diagnosis of SCFE23. Findings on US allow classification into stable or unstable SCFE. Absence of a joint effusion together with evidence of remodelling at the physeal–epiphyseal junction (periosteal reaction) implies a stable slip.The presence of a joint effusion with no signs of remodelling implies an unstable slip. Complications of SCFE include chondrolysis (narrowing of the joint space), AVN and osteoarthritis.

Scoliosis The Scoliosis Research Society has defined scoliosis as a lateral curvature of the spine greater than 10 degrees24. Scoliosis may

Figure 67.34 Avascular necrosis. T1 coronal MRI showing bilateral avascular necrosis.

Figure 67.35

Slipped capital femoral epiphysis. Klein’s line.

1593

1594

SECTION 8

•

PAEDIATRIC IMAGING

Figure 67.36 Slipped capital femoral epiphysis. Frog lateral illustrating measurement of slip angle.

Figure 67.37 Scoliosis. AP radiograph of the spine showing measurement of the Cobb angle.

be congenital or idiopathic. In contrast to idiopathic scoliosis, congenital scoliosis is related to a developmental abnormality of the spine. Based on the age of the patient at diagnosis, idiopathic scoliosis may be further subdivided into infantile (onset before 3 years of age), juvenile (between 3 and 10 years of age) and adolescent (from 10 years to skeletal maturity). The majority of children who present with idiopathic scoliosis do so in the adolescent period25. There is a strong hereditary component to idiopathic scoliosis and indeed candidate regions on chromosomes 6, 9 and 16 have recently been identified26. Congenital scoliosis is associated with such vertebral anomalies as hemivertebrae, block vertebrae and butterfly vertebrae. It may be associated with syndromes such as Alagille, Jarcho–Levin,VACTERL, Goldenhar and Klippel–Feil. Other conditions associated with a scoliosis include connective tissue disorders (Marfan’s, Ehlers–Danlos, and homocystinuria), neurological conditions (cerebral palsy, tethered cord, neurofibromatosis), and any cause of a leg length discrepancy. The prognosis (risk of curve progression) depends on the patient’s gender (worse in girls), the severity of the curve, and the child’s growth potential. The latter two may be determined radiographically. The magnitude of the curve is determined by measuring the Cobb angle (Fig. 67.37). An estimation of growth potential can be made by an assessment of the Tanner stage (clinical) and the Risser grade (radiological; Fig. 67.38). The Risser grade is

based on the degree of maturation of the iliac crest apophysis, and gives an estimation of how much growth remains. It has been shown to correlate directly with the risk of curve progression27.

Figure 67.38 Scoliosis. Line diagram of the Risser grades (0 = iliac crest yet to ossify; 4 = full closure of apophysis).

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

Cross-sectional imaging (CT and MRI) is useful for the exclusion of underlying vertebral and spinal cord anomalies; indeed, MRI is advised whenever there is a left thoracic curve, pain, abnormal neurological examination, or other unexpected findings to exclude causes such as tumour, syringomyelia, or spondylolisthesis28.

NEUROCUTANEOUS SYNDROMES Neurofibromatosis This is inherited as an autosomal dominant disorder. The responsible genes have been mapped to chromosomes 17q 21 (NF-I) and 22q12 (NF-II). Clinical features include multiple neurofibromas, schwannomas, axillary freckling, café-au-lait spots and molluscum fibrosum. Up to 85% of patients with neurofibromatosis manifest musculoskeletal abnormality (Fig. 67.39 and 67.40). Other features are listed in Table 67.9.

Tuberous sclerosis Clinical and radiographic features are listed in Table 67.10.

Figure 67.39 Neurofibromatosis Type 1. AP radiograph of the skull showing ‘empty’ left orbit.

Juvenile idiopathic arthritis The International League of Associations for Rheumatology (ILAR) introduced the term juvenile idiopathic arthritis (JIA) in 1997 to replace and unify the previous classifications of juvenile chronic arthritis and juvenile rheumatoid arthritis33.

JIA is arthritis of unknown aetiology occurring before the age of 16 years. It is subdivided to include: ‘oligoarthritis’ (one to four joints affected in the first 6 months of the disease); ‘polyarthritis’ (more than four joints affected within the first 6

Figure 67.40 Neurofibromatosis. (A) Pronounced posterior scalloping (arrowhead) of the lumbar vertebral bodies due to dural ectasia. (B−D) Development of a pseudarthrosis in the distal shaft of the ulna following a fracture of the radius (same patient).

1595

1596

SECTION 8

•

PAEDIATRIC IMAGING

Table 67.9 RADIOGRAPHIC FEATURES OF NEUROFIBROMATOSIS Soft tissues

Focal gigantism (soft tissue overgrowth or plexiform neurofibroma) Neurofibrosarcomas29

Skull

Macrocrania Aplasia/hypoplasia of the sphenoid wings (empty orbit; Fig. 67.39) Hypoplasia of posterosuperior orbital wall (pulsatile exophthalmos) Mesodermal dysplasia (calvarial defects) Neuromas and/or fibromas (with enlarged cranial foramina30)

Spine

Angular kyphoscoliosis Posterior scalloping of the vertebral bodies (dural ectasia) Dumb-bell neurofibromas/lateral meningocoeles

Ribs

‘Ribbon’ ribs (mesodermal dysplasia) Rib notching

Tubular bones

Pseudoarthroses of the tibia, fibula, or clavicle Anteromedial bowing of tibia Fibrous cortical defects (multiple and large) Intraosseous cysts

Table 67.10 CLINICAL AND RADIOGRAPHIC FEATURES OF TUBEROUS SCLEROSIS Clinical features

Radiographic features

Adenoma sebaceum

Renal angiomyofibromas31

Leukoderma

Cardiac myomas

Shagreen patches

Cyst-like phalangeal lesions

Subungual fibromas

Irregular undulating periosteal reaction along metacarpals and other tubular bones

Café-au-lait spots

Bone islands in the vertebral bodies and pedicles

Inflammatory arthropathy

Intracranial calcification32

months); ‘systemic arthritis’ (arthritis accompanied by systemic illness); ‘psoriatic arthritis’; ‘enthesitis related arthritis’ (often HLA B-27 positive); and ‘other arthritis’ (disease that does not fall into the listed groups). The wrist is affected in 61% of patients with polyarticular JIA, being second only to the knee as the most frequently affected joint.The hip and wrist are the most vulnerable joints to radiographically visible destruction. In some children, isolated involvement of the hips with bilateral protrusio acetabuli has been documented. It has been suggested that this isolated inflammatory coxitis may represent a separate subtype of oligoarthritic JIA34. Joint involvement is characterized by synovial inflammation progressing to synovial hyperplasia and pannus formation. Pannus erodes cartilage and bone and leads to articular destruction and ankylosis35. Affected joints are swollen, stiff with reduced motion, painful, erythematous and warm.

Early prediction of the course of disease is vital in JIA in order to instigate appropriate therapy and reduce long-term disability. The best predictors of poor outcome have been shown to be severity and extent of joint involvement at onset of disease; early hip and/or wrist involvement; positive rheumatoid factor; and prolonged active disease36. Radiography (Fig. 67.41A–C), dual energy X-ray absorptiometry (DEXA), ultrasound, CT and MRI are all employed, both for the initial evaluation and diagnosis of patients with JIA and for monitoring response to and complications of therapy. Radiography allows the assessment of soft tissue swelling, osteoporosis, erosions, joint destruction and ankylosis. Early in the course of the disease, the presence of effusions causes the joint spaces to appear widened. As disease progresses in severity, the joint spaces are narrowed (seen in the wrist as loss of height of the carpus) until finally there may be bony ankylosis. Inflammation causes hyperaemia with osteopaenia, relative overgrowth of the femoral condyles and patella and premature fusion of the epiphyses. Pressure from the hypertrophied synovium causes widening of the intercondylar notch. Evaluation of bone density from radiographs is subjective, and reliable detection requires that bone density be reduced to at least 30% its original value37. Other disadvantages of radiography include its low sensitivity for the detection of joint effusions, synovial thickening and differentiation of active from quiescent disease. MRI (Fig. 67.41D) has the advantage of not exposing the child to radiation. Furthermore, it allows improved demonstration of articular cartilage, joint effusions, synovial hypertrophy, fibrocartilaginous structures and muscles. Contrast-enhanced MRI (Fig. 67.41E) allows assessment of the perfusion of cartilage, synovium and bone and is the most sensitive method for determining whether an arthritic condition is present38. Enhancement of thickened synovium suggests active disease and differentiates it from fluid. AVN is a recognized complication of both JIA and steroid therapy. The presence of AVN may be determined using either radiography or MRI. Although it is more sensitive than radiography for the detection of early AVN, MRI is said to be less sensitive for the detection of osteochondral fractures with reported sensitivity and specificity for MRI and radiography cited as 38% and 100%, and 71% and 97%, respectively39. However, of particular significance is the fact that MRI may detect changes of AVN when radiographs appear normal40. A note of caution; standard MRI techniques do not always detect dead bone— this is because at the stage of early marrow necrosis, fat tissue is ‘mummified’ and preserves the same signal intensity as fat41.

Juvenile dermatomyositis Juvenile dermatomyositis (JDM) is a multisystem disease defined as affecting those under 18 years of age, although it more commonly affects children aged 2–15 years. It is of unknown aetiology, but both genetic and infectious agents have been implicated. The disease is characterized by a nonsuppurative inflammation of skin and skeletal muscle and is associated with a typical (pathognomonic) rash.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

Figure 67.41 Juvenile idiopathic arthritis. DP radiographs of the hands showing (A) mild erosive change of carpal bones and (B) severe changes (erosions of carpal bones and phalanges, osteopaenia, loss of height of carpus). (C) AP radiograph of the pelvis with severe loss of joint space on the right. The left hip is less severely affected. (D,E) MRI in same patient as C. (D) STIR image illustrating right hip effusion and avascular necrosis of the right femoral head. (E) T1-weighted image following contrast administration demonstrates bilateral active synovitis

The diagnostic criteria (established by Bohan and Peter42) include this rash and any three of the four following criteria: symmetrical proximal muscle weakness, elevated muscle enzymes, diagnostic histopathology findings and characteristic features on electromyogram (EMG). The latter two are invasive procedures and with the advent of MRI the diagnosis is often based on clinical, laboratory and MRI findings. Disease activity and response to therapy in JDM may be monitored by documenting muscle strength and function, serum muscle enzyme levels, range of joint movement, and physician’s global assessment (PGA) and by MRI. The Paediatric Rheumatology International Trials Organisation (PRINTO) and the Pediatric Rheumatology Collaborative Study Group (PRCSG) have recently published preliminary core sets of measures for disease activity and damage assessment in JDM, which bring together several of the tools listed above. Despite its usefulness they do not include MRI, as it is not universally available43. Radiographs may be normal. Abnormal findings include loss of muscle bulk, disuse osteopaenia and soft tissue calcification (Fig. 67.42). The MRI features of active dermatomyositis are best illustrated on T2-weighted and inversion recovery (STIR) sequences (Fig. 67.43). They include increased signal inten-

Figure 67.42 Juvenile dermatomyositis. Soft tissue calcification. Note also the pamidronate lines (see Fig. 67.44).

1597

1598

SECTION 8

•

PAEDIATRIC IMAGING

Figure 67.43 Juvenile dermatomyositis. (A) T1 axial MRI of the thighs. (B) Axial STIR of the thighs. There is bilateral active myositis with no significant loss of muscle bulk.

sity in affected muscles, perimuscular oedema, enhanced chemical-shift artefact and increased signal intensity in subcutaneous fat. More recently it has been shown that the T2 relaxation time can be used as a quantitative measure of muscle inflammation in JDM44. MRI will also demonstrate loss of muscle bulk with relative and absolute increases in subcutaneous fat, fatty infiltration of muscles, and occasionally soft tissue calcification. The resolution and relapse of signal abnormality during the course of JDM has been documented using serial MRI. However, there have been no controlled studies. Currently there are no recommendations for the timing of MRI in JDM, and it is not known how soon the abnormal signal intensity begins to respond to therapy or when it normalizes. The bisphosphonates are a group of drugs that are pyrophosphate analogues. They act by reducing bone resorption through an inhibitory effect on osteoclast function. Although not yet licensed for children, they are increasingly used in this age group, most commonly to increase bone mass in osteogenesis imperfecta. However, other indications include improvement of bone pain and density in rheumatological conditions such as JIA, JDM and SAPHO (synovitis, acne, palmoplantar pustulosis, hyperostosis and osteitis) syndrome. The radiographic hallmark of bisphosphonate therapy is the presence of dense metaphyseal bands (treatment pulses) alternating with bone of normal density (periods off treatment) (Figs 67.42, 67.44).

decades of life, but the number of affected joints stabilizes by the age of 20. Recurrent episodes of intra-articular bleeding cause villous synovial hypertrophy with accumulation of haemosiderin within macrophages.The arthropathy may progress to cause significant and irreversible cartilage destruction with secondary degenerative disease. Rarely (1–2% of patients) recurrent subperiosteal haemorrhage may become encapsulated and cause bony erosion, giving rise to the so-called haemophilic pseudotumour. This is more common in adult patients45. Based on radiography, five stages of disease may be recognized46; however, in any given patient, the chronology may

NON-INFLAMMATORY DISORDERS Haemophilia In this X-linked recessive disorder, a defect in blood coagulation leads to an increased tendency to haemorrhage. Depending on the severity of the disease (and compliance with therapy), bleeding may be spontaneous or occur following relatively mild trauma. Sites of bleeding include the brain, joints, abdomen and retroperitoneal cavity. Bleeding into the joints is common in haemophilia and usually involves the large joints of the knee (Fig. 67.45), elbow, ankle, hip and shoulder. Haemarthrosis begins in the first two

Figure 67.44 Juvenile dermatomyositis. DP radiograph of the hands showing pamidronate lines in the same patient as shown in Fig. 67.43.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

be demonstrated. Gradient-echo sequences will more readily demonstrate deposits of haemosiderin compared to spin-echo sequences (due to magnetic susceptibility artefact). Threedimensional spoiled gradient-echo and fast spin-echo sequences allow early identification of focal cartilage defects and thinning.

Pigmented villonodular synovitis

Figure 67.45 Bleeding into the joints in haemophilia. AP radiograph of the knees. There is widening of the intercondylar notches, narrow joint spaces and erosive change. The right knee is more severely affected.

not necessarily follow the stages nor is progression to the final stages inevitable in all affected joints. The stages are: • Stage I—soft tissue swelling and/or joint effusion—normal joint surfaces • Stage II—Stage I plus peri-articular osteoporosis—epiphyseal overgrowth • Stage III—erosions, sclerosis and subchondral cysts—joint spaces preserved • Stage IV—Stage III plus focal/diffuse joint space narrowing • Stage V—stiff contracted joint with significant degenerative change. Although CT and US may also be used in the assessment of haemophilic arthropathy, MRI (with its ability to demonstrate early disease, including synovial abnormality, ligamentous tears, peri-articular bleeding, cartilaginous and osseous bruising, erosions and joint space narrowing) is the investigation of choice and classification systems based on MRI findings are currently being developed47–49. Intra-articular haemorrhage will usually be evident as a joint effusion; however, fluid–fluid levels may

Pigmented villonodular synovitis (PVNS) refers to benign villous or nodular proliferation of synovium of uncertain aetiology. It most commonly affects a single joint with the knee being involved in 80%. Although most patients present in the third and fourth decades of life, the disease may also present in childhood (Fig. 67.46). Synovial joints, tendon sheaths, or bursae may be involved. Presentation is with a slow growing painless mass that may be tender to palpation. In long-standing cases there may be destruction of cartilage, secondary degenerative change and pain. Radiographic findings include soft tissue swelling, joint space narrowing and bony erosion (particularly in the hip where subchondral cysts with sclerotic rims are typical). Calcification is rare. Nonenhanced CT will show high attenuation due to the haemosiderin within the mass. The synovial proliferation enhances following administration of contrast medium. Diagnostic MRI features include a nodular lesion with areas of haemosiderin (low signal on all sequences) and haemorrhage (Fig. 67.46C). Joint effusions and bony erosions are well demonstrated. As with CT, contrast enhancement is typical. The differential diagnosis includes haemophilia and synovial haemangioma (rare, phleboliths in soft tissue)50.

METABOLIC AND ENDOCRINE DISORDERS Metabolic disorders Rickets Rickets is osteomalacia in children. There is an excess of unmineralized osteoid. The serum alkaline phosphatase is elevated, while serum and urinary calcium and phosphate levels are low. Serum levels of calcium and phosphate are controlled in part by vitamin D.

Figure 67.46 Pigmented villonodular synovitis. This child was unusual in that he was only 3 years of age at the time of diagnosis. (A) AP and (B) lateral radiographs of the knee showing significant soft tissue swelling. (C) Sagittal T1 MRI illustrating a mixed signal mass with areas of low haemosiderin) and high signal (haemorrhage).

1599

1600

SECTION 8

•

PAEDIATRIC IMAGING

The active form of vitamin D (cholecalciferol) is 1,25-dihydroxy-D3. Hydroxylation of D3 occurs first in the liver at the 25-position, and then in the kidney (regulated by parathyroid hormone) at the 1-position. It acts (with parathyroid hormone) on bone to stimulate the release of calcium and phosphate from osteoclasts; it stimulates intestinal absorption of calcium and phosphate; inhibits secretion of parathormone; and finally stimulates renal tubular re-absorption of phosphate. The focal radiographic features of rickets are best seen at sites of rapid growth [metaphyses and epiphyses of the distal radius (Fig. 67.47A), ulna and femur and proximal humerus and tibia]. Features include widening and cupping of the metaphyses, which have irregular, frayed margins. There is apparent widening of the physis (due to the unossified zone of provisional calcification). The epiphyses have indistinct margins and are relatively osteopaenic. Looser zones may be seen, particularly at the pubic rami, medial margins of the proximal femora, posterior aspects of the proximal ulnae, axillary margins of the scapulae and the ribs. The ‘rachitic rosary’ occurs as a result of expansion of the costochondral junctions. Complications include bowing of the long bones (bone softening, especially of the lower limbs), slipped capital femoral epiphysis and irregular vertebral end plates.

Renal osteodystrophy Chronic renal failure causes rickets as a result of failure of hydroxylation of inactive 1-hydroxy-D3 to the active 1,25dihydroxy-D3 within the renal glomeruli. In chronic renal failure there is retention of phosphate and hypocalcaemia,

Figure 67.47 Rickets. (A) Renal osteodystrophy. (B) Child with Fanconi’s disease. The dense metaphyseal band may be characteristic. (C) Linear sebaceous naevus syndrome. Note the areas of bone sclerosis.

which leads to parathyroid hyperplasia and secondary hyperparathyroidism. In addition there is reduced gastrointestinal absorption of calcium and end-organ resistance to parathormone. Serum phosphate and alkaline phosphatase are elevated, while serum calcium is normal or low. Radiologically, in addition to features of rickets, those of secondary hyperparathyroidism (osteosclerosis, acro-osteolysis and subperiosteal bone resorption) are also present.

Vitamin D-dependent rickets In these autosomal recessive conditions, vitamin D levels are not reduced. Type I is due to a defect in 1-α-hydroxylase, while in Type II vitamin D-dependent rickets there is endorgan resistance to 1,25-dihydroxy-D3. Patients may have alopecia and abnormal dentition.

Vitamin D-resistant rickets In these disorders renal tubular re-absorption of phosphate is defective. Renal excretion of calcium and phosphate is increased. Serum vitamin D levels are normal or even elevated. X-linked hypophosphatasia, vitamin D-resistant rickets with glycosuria (defective glucose and phosphate resorption), Fanconi’s syndrome (Fig. 67.47B) and acquired hypophosphataemic syndrome are the four conditions in which vitamin Dresistant or -refractory rickets may be seen.

Tumour rickets Certain tumours are thought to secrete a phosphaturic substance with consequent elevation in urine phosphate and

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

alkaline phosphatase. Serum calcium is normal. Implicated tumours include haemangiopericytoma, linear sebaceous naevus syndrome (Fig. 67.47C), nonossifying fibroma, giant cell tumour, osteoblastoma, fibrous dysplasia and mixed sclerosing dysplasia. Resection of the tumour leads to resolution of the rickets.

Neonatal rickets Radiographic features of neonatal rickets are not usually seen before 6 months of age. Rickets may develop in the neonate because dietary levels of calcium, phosphate, and vitamin D cannot meet the needs of a rapidly growing skeleton. Rickets occurring in preterm infants on parenteral nutrition is now relatively uncommon as a result of supplements in feeds.

Scurvy In this condition there is deficiency (usually dietary) of vitamin C (ascorbic acid). Infants typically present between 6 and 9 months of age. There is defective osteoid production by osteoblasts with reduced endochondral bone ossification. The bones are osteopenic with relatively dense margins (white lines of scurvy) where mineralization of osteoid continues. In the epiphyses this pencil outline is termed the ‘Wimberger’ sign (Fig. 67.48). Other features include metaphyseal (pelcan) spurs, which may fracture, exuberant periosteal reaction (recurrent subperiosteal bleeding), lucent metaphyseal bands and increased density of the end of the metaphyses (white line of Fraenk) (see Fig. 67.48).

Gaucher’s disease

Figure 67.48 Scurvy. AP radiograph of the knee. Soft tissue calcification is seen in the walls of a large haematoma.

This autosomal recessive storage disorder occurs most frequently in Ashkenazi Jews. The deficient enzyme is glucocerebrosidase. Glucocerebroside accumulates in the reticuloendothelial system and the bone marrow is infiltrated by lipid laden Gaucher cells. Infantile and juvenile forms are associated with mental retardation and early death. A milder adolescent form presents in childhood or early adulthood.

Radiological features include osteopenia, bone infarcts, AVN (particularly of the femoral head; Fig. 67.49A), flattening of the vertebral bodies, which may be significant (vertebra planal; Fig. 67.49B), Erlenmeyer flask deformity of the femora

Figure 67.49 Gaucher’s disease. (A) AP radiograph of the pelvis showing bilateral avascular necrosis. (B) Lateral radiograph of the spine showing vertebra plana. (C) T1-weighted coronal MRI of the femora. Note the Erlenmeyer flask deformity and marrow infiltration.

1601

1602

SECTION 8

•

PAEDIATRIC IMAGING

(Fig. 67.49C) and localized lytic bone lesions (focal deposition of Gaucher cells). Radiology helps to estimate disease burden, detect skeletal complications and monitor response to treatment. In children, physiological conversion of red to yellow marrow may cause confusion when interpreting MRIs. Radiologists should be aware of the patterns of conversion of low signal red marrow to high signal fatty marrow51,52.

ENDOCRINE DISORDERS Hyperparathyroidism In hyperthyroidism serum phosphate is decreased, serum calcium and alkaline phosphatase increased, and urine calcium and phosphate increased. Primary hyperthyroidism is due to a parathyroid adenoma or may occur in multiple endocrine neoplasia and is rare in children. Secondary hyperthyroidism is seen in chronic renal failure and tertiary hyperthyroidism occurs when the parathyroid glands become resistant to the regulatory effects of serum calcium (usually in patients on haemodialysis). Radiographs reveal features either of bone resorption and/ or bone formation. Sites of bone resorption include: • subperiosteal (radial sides of phalanges) • subchondral (sacroiliac joints, acromioclavicular joints, with resorption of the acromial ends of the clavicle and symphysis pubis) • subligamentous (calcaneum at the site of insertion of the Achilles tendon) • trabecular (diploic space causing a ‘salt and pepper’ skull) • intracortical (intracortical tunnelling/striations) • endosteal. Additional findings include osteosclerosis (localized or diffuse), rugger-jersey spine, brown tumours, chondrocalcinosis and soft tissue and vascular calcification. Neonatal hyperparathyroidism (Fig. 67.50) may occur as a primary disorder or as a result of poorly controlled maternal

hypoparathyroidism or pseudo hypoparathyroidism. Radiological features in these infants with failure to thrive include severe osteopaenia with an increased tendency to fracture, coarse trabeculae, metaphyseal cupping and subperiosteal resorption.

Hypoparathyroidism A reduced level of parathormone causes hypocalcaemia, hypophosphataemia and neuromuscular malfunction. It may be idiopathic or occur after surgical removal, disease, or trauma. Radiographs demonstrate osteosclerosis, skull vault thickening, soft tissue calcification, calcification of the basal ganglia, hypoplastic dentition and thickened lamina dura. Less commonly, osteoporosis, dense metaphyseal bands, dense vertebral end plates, premature fusion of the growth plates, vertebral hyperostosis and enthesopathy may occur.

Pseudo hypoparathyroidism and pseudo pseudo hypoparathyroidism Features in common with hypoparathyroidism include osteosclerosis, dense metaphyseal bands and calcification of the soft tissues and basal ganglia. Secondary hyperparathyroidism may be seen in 10% of patients with pseudo hypoparathyroidism (PHP), but is never seen in pseudo pseudo hypoparathyroidism (PPHP). In PHP there is end-organ resistance to normal or increased serum levels of parathyroid hormone. Clinically there is obesity, short stature and a rounded facies. Hypocalcaemia results in muscular tetany. Radiological features include exostoses and brachydactyly [short fourth (and fifth) fingers and toes (Fig. 67.51)] with a positive metacarpal sign (a line joining the heads of the little and middle fingers fails to intersect the head of the fourth metacarpal). Patients with PPHP have the phenotype of PHP; however, serum calcium and phosphate levels are normal. Cone-shaped epiphyses of the tubular bones of the hands and feet fuse prematurely causing brachydactyly.

Figure 67.50 Neonatal hyperparathyroidism. (A) Chest radiograph showing severe osteopenia and metaphyseal spurs of both proximal humeral metaphyses. (B) AP radiograph of the knees. Note the metaphyseal spurs.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

HAEMOGLOBINOPATHIES Sickle cell disease

Figure 67.51 Pseudo hypoparathyroidism. AP radiograph of the hands. Note the short fourth metacarpal.

Hypothyroidism In all types of hypothyroidism (infantile, juvenile and adult) there is either deficiency, or failure, of end-organ response to thyroxine. Radiological features in the paediatric age group include delayed bone age, delay in appearance of secondary ossification centres, epiphyseal fragmentation, stippling and short stature. Additional findings include osteoporosis, multiple wormian bones, delay in closure of the sutures and fontanelles, delayed dentition, dense metaphyseal bands, shortened long bones, increased atlanto-axial distance, kyphosis at the thoracolumbar junction and an increased incidence of slipped capital femoral epiphysis.

TOXIC DISORDERS Fluorosis Although rarely seen in the UK, the incidence of all types of fluorosis in the western world is said to be increasing. Radiologically it manifests as a generalized increase in bone density, particularly of the axial skeleton.There is periosteal proliferation, ligamentous calcification, degenerative enthesopathy and fractures. In children, exposure before the age of 8 years causes patchy opaque areas of the enamel of permanent teeth.

Lead poisoning This manifests radiologically as dense metaphyseal bands. Other features include widened sutures (raised intracranial pressure) and radio-opacities on abdominal radiographs (ingested lead).

Bone pain is the most common reason for hospital admission of patients with sickle cell disease. Thromboembolic infarcts and haemolysis (with chronic anaemia and secondary marrow expansion) is the underlying pathophysiology of the skeletal complications that occur. Acute bony involvement includes bone infarcts, osteomyelitis, stress fractures, vertebral collapse, bone marrow necrosis, orbital compression (infarction of the orbital bone) and dental complications (caries, mandibular osteomyelitis). Chronic complications include osteoporosis (secondary to marrow hyperplasia), AVN, chronic arthritis and growth failure53. Bone infarcts of the small bones of the hands and feet will lead to dactylitis; of the vertebral end plates to the so-called ‘cod fish’ or ‘H-shaped’ vertebrae (Fig. 67.52A); of the epiphyses will lead to joint effusions and AVN (Fig. 67.52B,C). Osteomyelitis is common, with Salmonella species being isolated in up to 70%54. The infection most commonly involves the diaphysis of the humerus (Fig. 67.52A), femur and tibia. The diagnostic challenge is to differentiate acute osteomyelitis from vaso-occlusive disease. Imaging findings may be very similar. The presence of a collection or of a break in the cortex makes infection more likely than simple infarction. US-guided aspiration of subperiosteal collections is said to be useful—it not only provides a sample for laboratory examination, but decompression relieves the pain associated with both conditions55.

Thalassaemia Skeletal changes in thalassaemia arise from the chronic anaemia associated with the condition. In the skull, there is widening of the diploic spaces (low signal on all MRI sequences) with thinning of the outer table of the skull vault. The trabecular markings are oriented perpendicular to the inner and outer tables and on plain radiographs this gives rise to the ‘hair-on-end’ appearance. There is frontal bossing and overgrowth of the facial bones with reduced pneumatization of the paranasal sinuses. The so-called ‘rodent facies’ arises from marrow hyperplasia in the maxillae causing lateral displacement of the orbits and ventral displacement of the central incisors56. In the spine there may be marked osteoporosis and cortical thinning resulting in fractures of the vertebral bodies and platyspondyly. Imaging may reveal paraspinal masses (as a result of extramedullary haematopoiesis). Cord compression can result if these masses extend into the extradural space. MRI findings are secondary to blood transfusion and chelation therapy. There may be expansion of the head and neck of the ribs (Fig. 67.53) and osteoporosis. A rib within a rib appearance may result. Extramedullary haematopoiesis can cause erosions of the inner cortex of the ribs or manifest as a posterior mediastinal soft tissue mass. Premature fusion of the growth plates (particularly of the proximal humerus and distal femur) is a recognized feature. Irregular sclerosis at the metaphyses and anterior rib ends is a recognized complication of treatment with desferrioxamine57.

1603

1604

SECTION 8

•

PAEDIATRIC IMAGING

Figure 67.52 Sickle cell disease. (A) Frontal chest radiograph: heart failure, codfish vertebrae. (B,C) Radiographs of the shoulder (B) and hip (C) showing avascular necrosis.

INFECTION OF THE BONES AND JOINTS Osteomyelitis

Figure 67.53 Thalassaemia. The abdomen shows coarsened trabeculae of the lumbar spine.

Infection may reach the bone through the bloodstream (haematogenous), from direct implantation (e.g. penetrating injury, surgery), or from infection elsewhere adjacent to bone (e.g. soft tissues). Because of the rich blood supply, osteomyelitis of the long bones in children most commonly affects the metaphyses. In infants, metaphyseal vessels penetrate the growth plate, and therefore in this age group there is a higher incidence of epiphyseal and joint involvement. Infection of the bones may be acute, subacute (Brodie’s abscess), or chronic. Acute osteomyelitis is most commonly seen in infants (Staphylococcus aureus, Escherichia coli) and young children (S. aureus, Streptococcus pyogenes, Haemophilus influenzae). Patients with sickle cell disease are more disposed to Salmonella infection. Subacute and chronic osteomyelitis result from incomplete eradication of infection following acute osteomyelitis, or from infection by less virulent organisms. Mycobacterial osteomyelitis occurs from haematogenous spread in a patient with primary tuberculosis. Fungal causes include coccidioidomycosis, bastomycosis and cryptococcosis. In acute disease there is oedema, vascular congestion and thrombosis of small vessels. Clinical features include fever, irritability, lethargy and local signs including swelling, erythema and warmth (inflammation). If not treated promptly (or aggressively), the vascular compromise leads to areas of dead bone (sequestra), which are the hallmark of chronic infection58. Periosteal new bone formation is another feature of chronic

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

osteomyelitis. This new bone (involucrum) encases areas of live bone. Pus may track from the medullary cavity through gaps in the involucrum into the soft tissues. These tracks may eventually penetrate the skin surface (sinus tracts). Radiographic changes (Fig. 67.54A,B) may not be apparent for up to 2 weeks after the onset of disease, and therefore radiographs may appear entirely normal. More specific signs include soft tissue swelling, cortical irregularity (bony destruction) and periosteal reaction. A Brodie’s abscess may be seen as a well-defined lytic lesion with a sclerotic rim. In chronic infection sequestra appear as dense foci, and soft tissue wasting and sinus tracts may be appreciated. Even on appropriate therapy, radiographic signs of improvement may lag behind clinical recovery. Spina ventosa implies tuberculous dactylitis. Radiographically there are cyst-like cavities associated with

diaphyseal expansion. It more commonly affects the bones of the hands than the feet59. High-resolution US provides a simple and noninvasive assessment of infants and children with osteomyelitis. US helps to localize the site and extent of disease, and to confirm the presence and degree of the fluid component of the abscess, and provides guidance for interventional procedures. Chau and Griffiths60 provide a useful review of the US appearances of musculoskeletal infection. Skeletal scintigraphy is useful if multifocal infection is suspected. CT helps to define the extent of cortical destruction and to exclude the presence of sequestra. MRI (Fig. 67.54C,D) has the highest sensitivity and specificity for detecting osteomyelitis in children61. In subacute

Figure 67.54 Osteomyelitis. (A) Right humerus in a patient with sickle cell disease. (B–D) Same patient. (B) Radiograph of the left humerus. (C) T1-weighted coronal MRI of the humerus showing low signal oedema in the medullary cavity. (D) Coronal STIR MRI of the humerus confirming marrow and soft tissue oedema. (E) Lateral ankle: Chevron deformity of the distal tibial metaphysis following meningococcal septicaemia.

1605

1606

SECTION 8

•

PAEDIATRIC IMAGING

infection a characteristic penumbra sign may be recognized62. This consists of a peripheral relatively high signal ring (granulation tissue) surrounding a low signal central zone (abscess cavity). Enhanced fat-suppressed sequences show avid enhancement of the granulation tissue. Contrast medium also helps to identify soft tissue abscesses. On T2-weighted sequences the high signal of reactive oedema may exaggerate the extent of infection. Complications include joint destruction and damage to growth plates (Fig. 67.54E) resulting in limb length discrepancies, angular deformities and early osteoarthritis.

Chronic recurrent multifocal osteomyelitis This is a condition of unknown aetiology characterized by a fluctuating clinical course of relapses and remissions. It affects multiple sites (synchronous or metachronous) and most commonly involves the long bones, clavicle, spine and pelvis. The ribs and sternum may also be affected. No causative agent is found.When associated with acne and palmoplanter pustulosis it is termed SAPHO syndrome. Radiographic features (Fig. 67.55A,B) suggest subacute or chronic osteomyelitis; however, abscess formation, involucra, and sinus tracts are not a feature. In the tubular bones lytic

Figure 67.55 Chronic recurrent multifocal osteomyelitis. (A,B) Hyperostosis of (A) the right clavicle (the left was similarly affected) and (B) the left fibula. (C) T1-weighted coronal MRI showing hyperostosis of the left fibula. (D) Coronal STIR MRI confirming active disease in the left fibula.

CHAPTER 67

•

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

metaphyseal lesions sometimes extending to the diaphysis are typical. Quiescent periods are characterized by bony expansion and sclerosis. In the clavicle, lytic medullary lesions are a feature of active disease, with expansion and sclerosis again being features of a quiescent phase. Chronic recurrent multifocal osteomyelitis (CRMO) manifests in the spine as loss of height of the affected vertebral bodies and is a differential diagnosis of vertebra plana. Bone scintigraphy is useful for the detection of asymptomatic lesions. MRI, by failing to demonstrate abscesses and sinus tracts, excludes chronic osteomyelitis. Active disease is confirmed by the presence of high signal marrow on T2/STIR sequences (Fig. 67.55C,D).

Infective arthritis As with osteomyelitis, infection may spread to joints via the bloodstream, direct inoculation, or spread from a contiguous site. The most common organism is S. aureus. Radiographic features (Fig. 67.56A,B) include soft tissue swelling, oedema and joint effusion. In infants, effusions may lead to dislocation (particularly in the hip joint). Other findings include metaphyseal irregularity and destruction. Septic arthritis in children is a clinical emergency, and early drainage is required to prevent severe bony destruction with resultant shortening and deformity. In this regard US (Fig. 67.56C) is helpful, both to exclude the presence of an effusion (a difference of 1−2 mm between the two sides is suggestive) and to assist in its drainage.

Infection of the spine (discitis and osteomyelitis) Discitis refers to infection of the intervertebral disc space, whereas osteomyelitis of the spine implies pyogenic destruc-

tion of the vertebral body, which may then spread to involve the disc. Differentiation of the two is important, as management may differ. Discitis can involve any spinal level, but most commonly affects the lumbar region in children younger than 5 years of age63. Generally children with discitis are younger, and clinically less toxic than those with osteomyelitis64. Children with either condition will present with back pain, refusal to mobilize, or irritability depending on age at presentation. Although there is much debate as to the aetiology of discitis, a low-grade infection has been postulated. In 70% of cases, the causative organism is not identified. When an organism is cultured, it is most commonly S. aureus. In discitis, radiological changes are confined to the disc and adjacent vertebral end plates, in comparison to osteomyelitis, in which the disease begins in and destroys the vertebral body. Infection may then spread to an adjacent vertebral body either via the intervertebral disc or via the subligamentous spread of pus (e.g. in spinal tuberculosis). In discitis, radiographs/CT of the spine (Fig. 67.57A) show characteristic features, including loss of disc height and irregularity of the adjacent vertebral end plates. Vertebral body height is preserved. It has been suggested that in the clinical context of suspected discitis, further imaging is not required if the radiographs demonstrate characteristic findings. However, if radiographs are normal or equivocal, or the child is toxic (suggesting spinal osteomyelitis), then further imaging is indicated64. MRI (with a sensitivity and specificity of 96% and 93%, respectively) is the next investigation (Fig. 67.57B) and can exclude intraspinal or other soft tissue (e.g. psoas) collections. T1 sagittal and axial and T2 sagittal views are often sufficient. In

Figure 67.56 Infective arthritis. (A) AP radiograph of the pelvis. There is a lytic lesion of the left proximal metaphysis with irregularity and reduced size of the left capital femoral epiphysis. (B) Right shoulder. Osteomyelitis of the humerus is complicated by septic arthritis and a dislocated shoulder. (C) Ultrasound of the hip confirms the presence of an effusion.

1607

1608

SECTION 8

•

PAEDIATRIC IMAGING

Figure 67.57 Discitis. (A) Conventional tomogram showing irregular and indistinct end plates at the affected level with narrowing of the disc space. (B) Sagittal T1weighted MRI showing enhancement of the end plates with a small anterior collection. (C) Axial CT obtained during prone guided biopsy of the affected disc.

equivocal cases, intravenous administration of contrast medium may be helpful. The pattern of enhancement is said to aid in differentiating tuberculous spondylitis (avid, heterogeneous rim enhancement, involving several vertebral bodies) from vertebral osteomyelitis65. However, regardless of the pattern of enhancement, tissue should be obtained for microbiological examination, usually by CT-guided biopsy (Fig. 67.57C).

REFERENCES 1. Hall C M 2001 International nosology and classification of constitutional disorders of bone. Am J Med Genet 113: 65–77 2. Offiah A C, Hall C M 2003 Radiological diagnosis of the constitutional disorders of bone. As easy as A, B, C? Pediatr Radiol 33: 153–161 3. Radiological electronic atlas for malformation syndromes (REAMS). Oxford University Press, Oxford, 2000 4. Cho T-J, Choi H I, Chung C Y et al 2000 The Sprengel deformity: Morphometric analysis using 3D-CT and its clinical relevance. J Bone Joint Surg (Br) 82-B: 711–718 5. Kawashima T, Uhthoff H K 1990 Development of the foot in prenatal life in relation to idiopathic club foot. J Pediatr Orthop 10: 232–237 6. Farrell S A, Summers A M, Dallaire L et al 1999 Clubfoot, an adverse outcome of early amniocentesis: disruption or deformation? CEMAT, Canadian Early and Mid-Trimester Amniocentesis Trial. J Med Genet 36: 843–846 7. Wiig O, Svenningsen S, Terjesen T 2004 Evaluation of the subchondral fracture in predicting the extent of femoral head necrosis in Perthes

8. 9.

10.

11.

12.

13.

14.

15.

16.

disease: a prospective study of 92 patients. J Pediatr Orthop B 13: 293–298 Catterall A 1971 The natural history of Legg-Calve-Perthes disease. J Bone Joint Surg (Br) 53: 37–53 Salter R B, Thompson G H 1984 Legg-Calve-Perthes disease. The prognostic significance of the subchondral fracture and a two group classification of the femoral head involvement. J Bone Joint Surg Am 66: 479–489 Herring J A, Neustadt J B, Williams J J, Early J S, Browne R H 1992. The lateral pillar classification of Legg-Calve-Perthes disease. J Pediatr Orthop 12: 143–150 Wiig O, Terjesen T, Svenningsen S 2002 Interobserver reliability of radiographic classifications and measurements in the assessment of Perthes disease. Acta Orthop Scand 73: 523–530 Eggl H, Drekonja T, Kaiser B, Dorn U 1999 Ultrasonography in the diagnosis of transient synovitis of the hip and Legg-Calve-Perthés disease. J Pediatr Orthop B 8: 177–180 Lahdes-Vasama T, Lamminen A, Merikanto J, Marttinen E 1997 The value of MRI in early Perthes’ disease: an MRI study with a 2-year follow-up. Pediatr Radiol 27: 517–522 Mahnken A H, Staatz G, Ihme N, Gunther R W 2002 MR signal intensity characteristics in Legg-Calve-Perthés disease. Value of fat-suppressed (STIR) images and contrast-enhanced T1-weighted images Acta Radiol 43: 329–335 Sebag G, Ducou Le Pointe H, Klein I et al 1997 Dynamic gadoliniumenhanced subtraction MR imaging—a simple technique for the early diagnosis of Legg-Calve-Perthés disease: preliminary results. Pediatr Radiol 27: 216–220 Lamer S, Dorgeret S, Khairouni A et al 2002 Femoral head vascularisation in Legg–Calve–Perthés disease: comparison of dynamic

CHAPTER 67

17. 18. 19. 20.

21. 22.

23.

24. 25. 26. 27.

28.

29. 30.

31. 32. 33.

34.

35. 36. 37.

38. 39.

40.

41.

•

gadolinium-enhanced subtraction MRI with bone scintigraphy. Pediatr Radiol 32: 580–585 Loder R T 2001 Unstable slipped capital femoral epiphyses. J Pediatr Orthop 21: 694–699 Loder R T 1996 The demographics of slipped capital femoral epiphysis: An international multicenter study. Clin Orthop Relat Res 322: 8–27 Loder R T, Wittenberg B, DeSilva G 1995 Slipped capital femoral epiphysis associated with endocrine disorders. J Pediatr Orthop 15: 349–356 Wells D, King J D, Roe T F, Kaufman F R 1993 Review of slipped capital femoral epiphysis associated with endocrine disease. J Pediatr Orthop 13: 610–614 Southwick W O 1967 Osteotomy through the lesser trochanter for slipped capital femoral epiphysis. J Bone Joint Surg Am 49: 807–835 Boyer D W, Mickelson M R, Ponsetti I V 1981 Slipped capital femoral epiphysis. Long-term follow-up study of one hundred and twenty-one patients. J Bone Joint Surg Am 63: 85–95 Kallio P E, Paterson D C, Foster B K et al 1993 Classification in slipped capital femoral epiphysis. Sonographic assessment of stability and remodelling. Clin Orthop 294: 196–203 Scoliosis Research Society 1976 A glossary of scoliosis terms. Spine 1: 57–58 Reamy B V, Slakey J B 2001 Adolescent idiopathic scoliosis: Review and current concepts. Am Fam Physician 64: 111–116 Miller N H, Justice C M, Marosy B et al 2005 Identification of candidate regions for familial idiopathic scoliosis. Spine 30: 1181–1187 Lonstein J E, Carlson J M 1984 The prediction of curve progression in untreated idiopathic scoliosis during growth. J Bone Joint Surg Am 66: 1061–1071 Oestrich A E, Young L W, Young Poussaint T 1998 Scoliosis circa 2000: radiologic imaging perspective I. Diagnosis and pre-treatment evaluation. Skeletal Radiol 27: 591–605 Vitale M G, Guha A, Skaggs D L 2002 Orthopaedic manifestations of neurofibromatosis in children: an update. Clin Orthop Aug: 107–118 Jacquemin C, Bosley T M, Liu D, et al 2002 Reassessment of sphenoid dysplasia associated with neurofibromatosis type 1. Am J Neuroradiol 23: 644–648 Casper K A, Donnelly L F, Chan B et al 2002 Tuberous sclerosis complex: renal imaging findings. Radiology 225: 451–456 Morris B S, Garg A, Jadhav P J 2002 Tuberous sclerosis: a presentation of less commonly encountered stigmata. Australas Radiol 46: 426–430 Petty R E, Southwood T R, Baum J et al 1998 Revision of the proposed classification criteria for juvenile idiopathic arthritis: Durban, 1997. J Rheumatol 25: 1991–1994 Adib N, Owers K L, Witt J D, Owens C M, Woo P, Murray K J 2005 Isolated inflammatory coxitis with protrusio acetabuli: a new form of juvenile idiopathic arthritis? Rheumatology (Oxf) 44: 219–226 Davidson J 2000 Juvenile idiopathic arthritis: A clinical review. Eur J Radiol 33: 128–134 Ravelli A, Martini A 2003 Early predictors of outcome in juvenile idiopathic arthritis. Clin Exp Rheumatol 21: S89–S93 Garland D E, Lewonowski K, Adkins R H, Stewart C A 1993 Visual versus quantified digital radiographic determination of bone density. Contemp Orthop 26: 591–595 Lamer S, Sebag G H 2000 MRI and ultrasound in children with juvenile chronic arthritis. Eur J Radiol 33: 85–93 Stevens K, Tao C, Lee S-U et al 2003 Subchondral fractures in osteonecrosis of the femoral head: comparison of radiography, CT and MR imaging. Am J Roentgenol 180: 363–368 Brody A S, Strong M, Babikian G et al 1991 Avascular necrosis: early MR imaging and histological findings in a canine model. Am J Roentgenol 157: 341–345 Sebag G, Ducou Le Pointe H, Klein I et al 1997 Dynamic gadoliniumenhanced subtraction MR imaging—a simple technique for the early

SKELETAL RADIOLOGY IN CHILDREN: NON-TRAUMATIC AND NON-MALIGNANT

42. 43.

44.

45. 46.

47.

48.

49.

50. 51. 52. 53. 54. 55.

56. 57.

58. 59. 60. 61. 62. 63. 64.

65.

diagnosis of Legg-Calve-Perthe’s disease: preliminary results. Pediatr Radiol 27: 216–220 Bohan A, Peter J B 1975 Polymyositis and dermatomyositis (part 2). N Engl J Med 292: 403–407 Ruperto N, Ravelli A, Murray K J et al 2003 Preliminary core sets of measures for disease activity and damage assessment in juvenile systemic lupus erythematosus and juvenile dermatomyositis. Rheumatology (Oxf) 42: 1452–1459 Maillard S M, Jones R, Owens C et al 2004 Quantitative assessment of MRI T2 relaxation time of thigh muscles in juvenile dermatomyositis. Rheumatology 43: 603–608 Kerr R 2003 Imaging of the musculoskeletal complications of hemophilia. Semin Musculoskelet Radiol 7: 127–136 Arnold W D, Hilgartner M W 1977 Hemophilic arthropathy. Current concepts of pathogenesis and management. J Bone Joint Surg 59A: 287–305 Kilcoyne R F, Nuss R 2003 Radiological assessment of haemophilic arthropathy with emphasis on MRI findings. Haemophilia 9 (suppl 1): 57–63 Lundin B, Babyn P, Doria A S et al 2005 Compatible scales for progressive and additive MRI assessments of haemophilic arthropathy. Haemophilia 11: 109–115 Doria A S, Lundin B, Kilcoyne R F et al 2005 Reliability of progressive and additive MRI scoring systems for evaluation of haemophilic arthropathy in children: expert MRI Working Group of the International Prophylaxis Study Group. Haemophilia 11: 245–253 Masih S, Antebi A 2003 Imaging of pigmented villonodular synovitis. Semin Musculoskelet Radiol 7: 205–216 Maas M, Poll W L, Terk M R 2002 Imaging and quantifying skeletal involvement in Gaucher disease. Br J Radiol 75 (suppl 1): A13–A24 Bembi B, Ciana G, Mengel E et al 2002 Bone complications in children with Gaucher disease. Br J Radiol 75 (suppl 1): A37–A44 Almeida A, Roberts I 2005 Bone involvement in sickle cell disease. Br J Haematol 129: 482–490 Bennet O M, Namnyak S S 1990 Bone and joint manifestations of sickle cell anaemia. J Bone Joint Surg Br 72: 494–499 Booz M M, Hariharan V, Aradi A J, Malki A A 1999 The value of ultrasound and aspiration in differentiating vaso-occlusive crisis and osteomyelitis in sickle cell disease patients. Clin Radiol 54: 636–639 Tunaci M, Tunaci A, Engin G 1999 Imaging features of thalassaemia. Eur Radiol 9: 1804–1808 Chan Y-L, Pang L-M, Chik K-W, Cheng JCY, Li C-K 2002 Patterns of bone disease in transfusion-dependent homozygous thalassaemia major: predominance of osteoporosis and desferrioxamine-induced bone dysplasia. Pediatr Radiol 32: 492–497 Lazzarini L, Mader J T, Calhoun J H 2004 Osteomyelitis in long bones. J Bone Joint Surg (Am) 86: 2305–2318 Andronikou S, Smith B 2002 Spina ventosa—tuberculous dactylitis. Arch Dis Child 86: 206 Chau C L F, Griffith J F 2005 Musculoskeletal infections: ultrasound appearances. Clin Radiol 60: 149–159 Jorulf K S, Hirsch G 1998 Clinical value of imaging technique in childhood osteomyelitis. Acta Radiol 39: 523–531 Davies A M, Grimer R 2005 The penumbra sign in subacute osteomyelitis. Eur Radiol 15: 1268–1270 Early S D, Kay R M, Tolo V T 2003 Childhood diskitis. J Am Acad Orthop Surg 11: 413– 420 Fernandez M, Carrol C L, Baker C J 2000 Discitis and vertebral osteomyelitis in children: An 18-year review. Pediatrics 105: 1299–1304 Arizono T, Oga M, Shiota E, Honda K, Sugioka Y 1995 Differentiation of vertebral osteomyelitis and tuberculous spondylitis by magnetic resonance imaging. Int Orthop 19: 319–322

1609

CHAPTER

Paediatric Musculoskeletal Trauma and the Radiology of Nonaccidental Injury

68

William H. Ramsden and Padma Rao

Paediatric musculoskeletal trauma • Fractures • Stress and overuse injuries • Epiphyseal injuries • Upper limb injuries

• Fractures of the pelvis and lower limbs • Spinal injuries Radiology of nonaccidental injury • Skeletal and soft tissue injury • Brain injuries

PAEDIATRIC MUSCULOSKELETAL TRAUMA William H. Ramsden

FRACTURES Greenstick (Fig. 68.1) and torus (Fig. 68.2) injuries are both incomplete fractures which commonly occur in children. The bone cortex and periosteum break on the convex side of a greenstick fracture, whilst a torus fracture is diagnosed when the cortex buckles on its concave side. When a long bone bends, rather than breaks, this is known as plastic bowing (Fig. 68.3). It manifests as a subtle increase in curvature rather than a discrete fracture, but histologically multiple oblique microfractures are present at the site of greatest compression1. The forearm bones are most often affected and the fracture may be difficult to appreciate acutely. In addition, the ‘clue’ of periosteal new bone formation may not be visualized during the early healing phase2. It is unusual for balanced forces to cause plastic bowing of both forearm bones and the nonbowed bone may fracture or dislocate.

STRESS AND OVERUSE INJURIES Children may develop similar stress injuries to adults, due to repeated forces acting upon a bone that are less than the force needed to fracture it acutely. In childhood the tibia is most often affected (Fig. 68.4), with significant numbers of fibular and metatarsal fractures3.

The stresses that lead to sports-related injuries in children elicit a different response from the developing skeleton than that seen in adults. In children the main sites affected are the growth plates, muscle insertions and apophyses. Avulsion injuries are responsible for the latter two sites, and are most common around the pelvis and elbow4. Plain radiographs are the initial imaging investigation in both stress and avulsion injuries, and the former may be investigated by both radionuclide radiology and magnetic resonance imaging (MRI). MRI sequences that are particularly sensitive for oedema, such as inversion recovery (STIR), may show widespread changes in bone surrounding a stress injury. This should not deflect the clinician from the diagnosis if clinical circumstances and imaging appearances are otherwise appropriate.

EPIPHYSEAL INJURIES Epiphyseal separations are often the most challenging childhood fractures to investigate. It is important to understand normal epiphyseal anatomy, so that injuries may be detected and normal growth plates correctly identified. The best example of this is at the elbow joint, where epiphyses both appear and fuse in a known sequence. Epiphyseal injuries are grouped according to the Salter– Harris classification (Fig. 68.5), an important guide to both

1612

SECTION 8

Figure 68.1

•

PAEDIATRIC IMAGING

Greenstick fractures of the distal radius and ulna.

Figure 68.3 Plastic bowing of the ulna with dislocation of the radial head (arrow).

Figure 68.2 Torus fractures (arrows) of the necks of the second, third and fourth metatarsals.

treatment and the prognosis of any particular fracture. The grading of injuries runs from I to V, becoming more serious as the numeral rises. In Types I and II, the epiphysis remains intact and these injuries generally have a good prognosis. This contrasts with Types III and IV, where the epiphysis itself is fractured. The Type V fracture is a very rare crushing injury of the physeal cartilage and is more commonly seen in association with Types

Figure 68.4

Healing stress fracture of the proximal tibia.

I–IV injuries than in isolation. Growth disturbance is the main complication that can occur and tends to be more severe with the higher grades. In spite of this classification, prognosis is worse in the lower limb for all grades.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

usually due to falling onto an outstretched hand. The most common fracture type is a greenstick injury centred upon the middle third of the bone. Acromioclavicular joint separation is uncommon in children due to the relative strength of the associated ligaments, and in young children the normal joint may often look ‘wide’. In dislocation of the sternoclavicular joint the medial clavicle is usually displaced anteriorly, but it is the rarer posterior dislocation that is potentially the more serious, because of the proximity of underlying vessels. In such a situation, CT is the imaging method of choice. Although fractures involving the proximal humeral growth plate account for a small minority of physeal injuries, they may have important consequences as this area contributes 80% to the bone’s longitudinal growth. However, it needs to be noted that the normal proximal humeral growth plate has an irregular contour on a standard antero-posterior (AP) radiograph, which may be mistaken for a fracture. Transverse and torus fractures of the proximal humeral metaphysis may also occur, most commonly around the surgical neck. Simple bone cysts are frequently found in this area and pathological fractures often occur. Plain radiographs in the latter situation may demonstrate a characteristic ‘fallen fragment sign’ as cortical bone lies within a fluid-filled cavity.

Distal humeral and elbow trauma Figure 68.5 Salter–Harris classification of fractures involving the epiphyseal plate. [From Carty H (ed) 1999 Emergency pediatric radiology. Springer-Verlag, Berlin.]

Although plain radiographs are the sole imaging method required in the majority of epiphyseal injuries, computed tomography (CT) is useful in planning complex fracture treatment, and MRI is becoming increasingly important in evaluating more difficult cases. MRI is used to visualize nonossified growth plate cartilage, and in very young children the nonossified epiphysis itself. It allows visualization of any associated soft tissue or ligamentous injury, the latter often accompanying epiphyseal injuries at the knee. Although the STIR sequence rapidly localizes injury, conventional spin-echo sequences are needed to separate bone bruises and true fractures. Gradient-echo sequences are particularly good at demonstrating the growth plate as hyperintense compared to surrounding bone.

UPPER LIMB INJURIES Clavicle and proximal humeral fractures The clavicle is the bone most commonly injured at birth and the main associated risk factors are heavy neonates and shoulder dystocia. There is also a risk of brachial plexus damage, although both this and the bony injury generally heal uneventfully. Clavicular fractures are common in childhood,

The six epiphyses surrounding the elbow joint appear and fuse in a predictable sequence commencing with that of the capitellum, followed by the radial head, medial epicondyle of the humerus, trochlea, olecranon and finally, the lateral epicondyle.Any deviation from the above sequence following trauma should initiate efforts to locate an avulsed or malpositioned epiphysis. Good quality frontal and lateral radiographs are the usual initial imaging investigation in elbow trauma and the latter allows detection of elbow joint effusions as they elevate fat pads. Effusions in the absence of visible bony injury may be due to occult fractures, and in children who have undergone subsequent MRI, over 50% were shown to have such injuries. However, in the largest series5 this knowledge had little influence upon treatment or outcome. The lateral radiograph is also particularly important in the detection of supracondylar fractures, and the diagnosis may depend upon visualization of subtle disruption of the normal relationship between the distal humeral metaphysis and capitellum. A line drawn along the anterior cortex of the humerus should pass through the middle third of the capitellum, and if it does not, a supracondylar fracture is likely to be present. Other imaging modalities are particularly useful for demonstrating cartilaginous areas. Ultrasound (US) can demonstrate the relationship of nonossified epiphyses to the metaphyses, and CT can be used to clarify complex epiphyseal injuries before treatment. MRI combines the strengths of both modalities by its excellent demonstration of cartilaginous epiphyses in multiple planes. Supracondylar fractures (Fig. 68.6) are usually due to falling on the outstretched hand. The usual pattern of injury is of a

1613

1614

SECTION 8

•

PAEDIATRIC IMAGING

Figure 68.7 Medial epicondyle epiphysis (arrow) trapped within the elbow joint following avulsion.

Figure 68.6 Supracondylar fracture of the distal humerus with elevation of the posterior fat pad (arrows) indicating a significant joint effusion.

transverse fracture passing just proximal to the ossified capitellum and trochlea with posterior displacement of the distal fragment. Displaced fractures may damage the brachial artery and check radiographs are essential to ensure a satisfactory position after manipulation. Children also sustain condylar and epicondylar fractures, and avulsion of the medial epicondyle carries the significant risk of entrapment of the separated epiphysis within the elbow joint in association with dislocation. The latter may reduce spontaneously or be manipulated, but in either situation the position of the ossified epiphysis must be carefully checked on post-reduction radiographs to ensure it is not intra-articular (Fig. 68.7).

Forearm, wrist and hand fractures The radial and ulnar shafts may fracture together or alone. In a single fracture it is most important to check for dislocation of the noninjured bone, and a line drawn along the longitudinal axis of the radius should bisect the capitellum, whatever the radiographic plane. Monteggia and Galeazzi fractures (Fig. 68.8) may both occur in childhood although the former is less common than in adults. It should also be noted these injuries can result from greenstick or bowing fractures in children (see Fig 68.3). Fractures of both forearm bones occur more commonly, but rarely even these may be associated with radial head dislocation. Greenstick injuries of the distal forearm bones are commonly seen, as are Salter–Harris Type II epiphyseal separations of the distal radius. Distal radial fractures may or may not be accompanied by injuries to the distal ulna, but isolated ulnar fractures are uncommon.

Figure 68.8

Galeazzi fracture dislocation.

Childhood carpal fractures are uncommon but when they do occur, it is the scaphoid that is most often affected. Scaphoid fractures are most commonly seen in adolescents and are extremely rare in younger children. The distal third of the scaphoid is the most common site of injury (Fig. 68.9), in contrast to the waist of the bone in adults. Thus although fracture, nonunion and avascular necrosis occasionally occur6, their incidence is far lower than in older patients.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

Figure 68.9 Fracture of the distal pole of the scaphoid.

Plain radiographs are used for initial evaluation. Traditionally, the patient with persistent symptoms and no demonstrable fracture has been further investigated with follow-up radiographs, and if these too are negative, bone scintigraphy. However, MRI has also been shown to be very useful in the exclusion or confirmation of scaphoid fractures, allowing appropriate management of injured patients and discharge of the remainder7. Metacarpal fractures are more common, with the neck of the little finger metacarpal most frequently affected. However, the most common fracture in children’s hands is a Salter–Harris Type II fracture of the base of the proximal phalanx of the little finger, and this area often harbours subtle greenstick injuries. The middle and distal phalanges are prone to avulsion of their epiphyses, and care must be taken that separated fragments are not missed on inadequate radiographs (Fig. 68.10). Injured digits require good quality AP and lateral radiographs and all the ossified phalangeal epiphyses need to be accounted for, as avulsions may travel a significant distance proximally.

Figure 68.10 Salter–Harris Type II epiphyseal separation of the middle phalanx of the ring finger. The avulsed fragment is not visible on the AP view, but the epiphysis is missing. [From Carty H (ed) 1999 Emergency pediatric radiology. Springer-Verlag, Berlin.]

FRACTURES OF THE PELVIS AND LOWER LIMBS Fractures of the pelvis The system of paediatric pelvic fracture classification, proposed by Torode and Zieg in 19858, attempts to take account of all of a child’s injuries and the associated morbidity and mortality. Pelvic fractures are divided into four types, the less severe comprising avulsion injuries (Type 1) and isolated fractures of the iliac wing (Type 2).Type 3 comprises simple ring fractures and Type 4 ring disruption, the latter having the greatest long-term morbidity9. Avulsion injuries are most commonly seen in adolescents and are usually due to athletic activity. These injuries may be either acute or chronic, the former being due to sudden contraction of a muscle attached to an apophysis and the latter due to repetitive traction. The ischial tuberosity is most commonly affected, as it attaches both the hamstrings and adductors. Other sites frequently affected are the anterior superior and inferior iliac spines, which attach the sartorius and straight head of the rectus femoris, respectively (Fig. 68.11).

Figure 68.11 Avulsion injury. Florid new bone formation (arrow) following an avulsion injury of the reflected head of the rectus femoris muscle in a young footballer.

1615

1616

SECTION 8

•

PAEDIATRIC IMAGING

A standard AP pelvic radiograph will demonstrate most acute avulsion injuries, apart from those affecting the anterior inferior iliac spine, which may require an oblique view. Healing may produce much callus and subtle radiolucency, and periosteal reaction may be seen in chronic avulsion injuries, particularly in the region of the ischial tuberosity. It is important that neither is mistaken for a bone tumour. The higher grades of fracture are usually due to road accidents. Iliac wing fractures (Type 2) are less common than in adults, associated soft tissue injury is uncommon and the overall prognosis is good. Type 3 fractures include those involving the pubic rami and disruption of the pubic symphysis, although the elasticity of a child’s pelvis means that the pubic symphysis may separate to approximately 2.5 cm without sacro-iliac joint instability. This category allows for displaced fractures providing there is no clinical instability of any other pelvic segment. Type 4 fractures comprise those with an unstable segment due to the pelvic ring fracturing at two sites. Unsurprisingly, abdominal and genitourinary injuries are most commonly seen in association with Types 3 and 4 fractures. Plain radiography is the usual initial investigation in suspected pelvic fractures, although it is of limited utility when used purely for screening after trauma10. Children suspected of having serious fractures usually proceed to CT to define better bony injury and/or associated visceral trauma.

Acetabular, hip and femoral fractures Acetabular fractures in children are uncommon, but potentially significant because those that involve the triradiate cartilage may affect subsequent growth. A plain radiograph may be obtained initially, but CT is usually the imaging method of choice. Children who suffer hip dislocation have a lesser probability than adults of sustaining an associated acetabular injury11. However, in those that do, MRI is particularly useful in assessing the degree of posterior acetabular involvement (bony and cartilaginous) and consequently guiding treatment12. Posterior hip dislocation is the most common type and avascular necrosis of the femoral capital epiphysis is a serious complication, particularly if the hip is left unreduced for over 24 h. It is important that a careful check is made for associated fractures of the pelvis, femur and patella. Femoral head and neck fractures are far less common in children than in adults, but complications occur more frequently and tend to be more serious. These include avascular necrosis, premature fusion of the growth plate, varus deformity and nonunion. AP and lateral radiographs are essential to check the position of the femoral capital epiphysis and the integrity of the growth plate. Femoral diaphyseal fractures are not uncommon in children, and surrounding muscle pull may cause angulation and overlap of the opposing bone ends. Despite this, it is essential to obtain radiographs in two planes, otherwise subtle injuries may be missed. Fractures involving the distal femoral growth plate are more likely to lead to problems with subsequent growth disturbance than those at many other locations (Fig. 68.12). This may be

Figure 68.12 Epiphyseal injury of the distal femur. (A) Healing epiphyseal injury (arrows) of the medial aspect of the distal femur. (B) Delayed radiograph demonstrating growth arrest due to premature fusion of the medial half of the epiphysis. [From Carty H (ed) 1999 Emergency pediatric radiology. Springer-Verlag, Berlin.]

due to the significant force needed to disrupt the physis13. Although plain radiographs may reveal the physeal injury, in many cases its full extent is only revealed by MRI, and this additional information will frequently alter subsequent management14.

Fractures, ligamentous and cartilaginous injuries around the knee (including the tibia) Functionally, fractures involving the anterior intercondylar eminence of the tibia (Fig. 68.13) are similar to anterior cruciate ligament disruption, and are usually due to knee hyperextension15. Such injuries may be subtle radiographically, the lateral view revealing a small displaced calcific fragment as the only visible sign of underlying disruption of the epiphyseal cartilage.The lesion may be missed on the AP view as the joint is flexed, and MRI provides the best means of making the

Figure 68.13 Fracture of the anterior intercondylar eminence of the tibia (arrow).

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

diagnosis. The injury is significant as irreducible and displaced fractures require operative fixation13. Hyperextension of the knee can also cause separation of the proximal tibial epiphysis (Fig. 68.14), as the adjacent metaphysis is forced backwards. The usual result is a Salter–Harris Type II fracture which may be difficult to appreciate on plain radiographs if displacement is minimal. Again, MRI provides the best means of diagnosis. Although uncommon, such injuries may be complicated by angular deformity, future leg length discrepancy and damage to the adjacent popliteal artery if displaced. A similar mechanism of injury can cause a transverse fracture of the tibial metaphysis in young children16. This may result in a subtle bony injury causing a child to limp. In children in whom ligamentous or meniscal injury is suspected, MRI is recommended, utilizing sagittal and coronal T1-weighted images, coronal T2-weighted images and sagittal T2-weighted gradient-echo images. Uniform low signal intensity is seen in intact ligaments, whilst possible tears show as areas of increased signal, and definite tears as increased signal with disruption or deformity. Definite meniscal tears are revealed as either deformity or high signal traversing the normally low signal meniscus to reach an articular surface. When high signal within the meniscus does not reach an articular surface, appearances are regarded as equivocal. In children medial meniscal tears are more common than injuries to the lateral meniscus, although both injuries occur significantly less frequently than in adults. A significant number of traumatized lateral menisci will be discoid.The anterior cruciate ligament is injured less frequently than the menisci

Figure 68.14 Salter–Harris Type I epiphyseal separation of the proximal tibia.

and tears of the posterior cruciate ligament are uncommon. Associated injuries may also be demonstrated on MRI; examples including tears of the collateral ligaments, osteochondral fractures and bone marrow oedema17,18. MRI is also useful in the diagnosis of osteochondritis dissecans (Fig. 68.15), which is defined as a subchondral or osteochondral defect due to necrosis of the subchondral bone, possibly following an undiagnosed injury. This may subsequently lead to separation of a subchondral fragment, resulting in an intra-articular loose body. In the knee the condition most frequently involves the medial femoral condyle19, and other sites classically affected include the dome of the talus and capitellum. A plain radiograph of the knee joint will reveal an oval lucency involving the articular surface of one of the femoral condyles with or without a central bony fragment. MRI can be used to predict the stability of such lesions, as unstable fragments may become intra-articular loose bodies. High signal intensity between the fragment and bone, articular fractures, focal osteochondral defects and cysts greater then 5 mm across, deep to the lesion, are all MRI predictors of instability. The patella is infrequently fractured in children, the most common injury types being comminuted and transverse fractures following road traffic accidents. Patellar dislocation usually first occurs in adolescence and is associated with an abnormally flat sulcus between the femoral condyles and patella alta. The injury is associated with a significant number of osteochondral fractures20, derived from either the lateral femoral condyle or the patella itself. Full patellar dislocation will be visible on a standard plain radiographic series. If the dislocation has reduced before the child is examined, a ‘skyline’ view should be obtained to pursue a possible associated osteochondral fracture of the medial patella. As with other osteochondral lesions, they may be subtle or entirely chondral, in which case MRI offers the best means of detection (Fig. 68.16). The classical toddler’s fracture (Fig. 68.17) is an undisplaced, oblique fracture of the distal tibia which may not be visible on initial radiographs. It usually presents in young children as limping or refusal to weightbear. If the initial investigations are

Figure 68.15 Osteochondritis dissecans. Coronal proton density MRI demonstrating characteristic defects in both femoral condyles due to osteochondritis dissecans.

1617

1618

SECTION 8

•

PAEDIATRIC IMAGING

Figure 68.16 Chondral fragment. Coronal T2-weighted MRI (with fat saturation) showing a chondral fragment (arrow) avulsed following patellar dislocation with a ‘bone bruise’ of the adjacent distal femoral epiphysis.

Triplane and Tillaux fractures tend to occur in adolescence, around this time of distal tibial epiphyseal fusion. The former is a Salter–Harris Type IV epiphyseal injury comprising an oblique coronal fracture through the distal tibial metaphysis, which extends horizontally through the lateral part of the physis before running vertically through the epiphysis in the sagittal plane. The lateral ankle radiograph suggests a Salter–Harris Type II fracture due to the extension into the metaphysis whilst the frontal view reveals a Type III injury of the epiphysis. Accurate reduction of such an injury is important and CT with multiplanar reformatting provides the best assessment of the fracture’s extent before reduction (Fig. 68.18). The Tillaux fracture occurs at a similar location and consists of a Salter–Harris Type III injury, which runs through the anterolateral aspect of the distal tibial physis until it reaches the part that has fused. It then passes downwards through the epiphysis into the joint. The fracture is usually demonstrable on plain radiographs (Fig. 68.19), subject to the X-ray beam being parallel, which may require an oblique view. As the majority of the epiphysis has already fused the injury does not cause subsequent growth arrest, but its recognition is important so that joint congruity is restored. CT has been shown to be useful in detecting significant displacement requiring closed or open reduction21.

Fractures of the foot The calcaneus is the most frequently injured tarsal bone in children, but as fractures tend to involve the tuberosity and avoid the posterior facet, most are classified as extra-articular22. They may be very difficult to detect radiographically, and CT or scintigraphy both provide useful means of pursuing the diagnosis further. The most common talar fracture in childhood affects the neck of the bone and displaced injuries carry the risk of subsequent avascular necrosis. However, the high cartilage-to-bone ratio of a child’s talus makes it more resistant to injury than

Figure 68.17 Toddler’s fracture of the distal tibial metaphysis.

negative, a delayed radiograph or scintigraphy may be used to confirm the diagnosis.

Fractures around the ankle Avulsions from the tip of the lateral malleolus and Salter– Harris Type I and II epiphyseal injuries of the distal fibula form the majority of paediatric ankle fractures. Children’s physes are more likely to fail than ligaments and those of the distal tibia and fibula fuse at approximately the same time during adolescence. If only one is fused, the suspicion of an epiphyseal injury is raised.

Figure 68.18 Assessment of the extent of a triplane fracture. (A) Coronal reformat of an axial CT demonstrating epiphyseal and physeal components of a triplane fracture. (B) Saggital reformat of the same case demonstrating metaphyseal extension of the fracture.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

to the forefoot occur more frequently than in the mid and hind foot.

SPINAL INJURIES Normal variants There may be congenital absence of the posterior arch of the atlas and the anterior arch may not ossify until a year of age in normal children. ‘Pseudospread’ of the lateral masses of the atlas may also be seen on an open mouth view in young children. Pseudosubluxation (Fig. 68.21) is the result of ligamentous laxity and horizontally orientated facet joints in young children, allied to their relatively large head size. A line connecting the anterior aspects of the spinous processes of C1 to C3 on a true lateral radiograph may help differentiate true injuries from pseudosubluxation. If this line misses the anterior aspect of the C2 spinous process by 2 mm or more, then the suspicion of a fracture or true subluxation is raised. The latter is very rare, even in severe polytrauma, whilst pseudosubluxation is relatively common in the same context25. Ligamentous laxity also means that the atlanto-axial distance measured on a lateral radiograph may be up to 5 mm in normal children. Figure 68.19 Tillaux fracture (arrows) of the distal tibial epiphysis.

that of an adult23. As with other tarsal bones, CT will best define the acute fracture. The navicular may be subject to avulsion of its dorsal cortex or tuberosity. Fractures of the body of the bone may also occur, sometimes as stress injuries in young athletes. As they are usually orientated in the sagittal plane in the centre of the bone, they may be impossible to detect on plain radiography. Scintigraphy is a useful means of investigating mid foot pain in such patients (Fig. 68.20), and if increased activity is demonstrated in the navicular, CT may be used to confirm a fracture24. Greenstick and torus fractures are the most common types affecting the metatarsals and phalanges, and injuries

Figure 68.20 Navicular fracture. Bone scintigram showing localized increase in activity in the left navicular consistent with an occult fracture.

Injuries to the cervical and thoracic spine Half to two-thirds of injuries to the cervical spine in children are caused by road traffic accidents. In those under 12 years of age the majority of injuries affect the occipito-atlantoaxial segment, which partly explains the higher mortality in younger children, allied to the greater incidences of associated severe head and extraspinal trauma.Younger children (under 8 years) tend to suffer distraction and subluxation injuries whilst fractures occur more commonly in those over 8 years26.

Figure 68.21

Pseudosubluxation of C2 upon C3.

1619

1620

SECTION 8

•

PAEDIATRIC IMAGING

In children, as in adults, the precervical soft tissues visualized on the lateral radiograph should measure 7 mm or less anterior to the second cervical vertebral body. In infants this area should measure no more than three-quarters of the AP diameter of the adjacent vertebral body. The vast majority of children with high cervical injuries will demonstrate widening of the retropharyngeal soft tissues27, although radiographs obtained in flexion may exaggerate such thickening. Rotary subluxation of the atlas upon the axis may or may not follow trauma in children and is a more common occurrence than in adults. The diagnosis is suspected on plain radiographs when the odontoid is positioned asymmetrically between the lateral masses of the atlas. To differentiate this injury from transient torticollis, imaging must demonstrate lack of rotatory movement and/or locking of the lateral masses of the atlas and axis. CT is the best means to demonstrate such locking, and is performed with the head rotated both to the left and right to show that the abnormal relationship between the vertebrae does not change. It also serves to exclude an associated fracture. Atlanto-axial dislocation in children differs from that in adults in that the child’s transverse ligament usually ruptures without an associated odontoid injury, whilst adults more commonly sustain a dens fracture. Suspicion of dislocation is raised when the atlanto-axial distance exceeds 5 mm on the lateral radiograph, or if there is a large difference between flexion and extension. MRI is the method of choice for the demonstration of ligamentous injury and the cord damage that may also occur. Older children tend to sustain injuries to the mid and lower cervical spine which correspond more to the adult pattern and distribution. Spinal cord injury without radiographic abnormality (SCIWORA) may occur following trauma due to the disproportionate flexibility of a child’s spine and spinal cord (Fig. 68.22). Either the cervical or thoracic cord may be affected, with younger children being affected more frequently28. Radiographs are normal, but the underlying cord injury is a significant cause of childhood paraplegia. MRI is required to delineate the cord damage, and in children with neurological signs, it must be obtained before discontinuing spinal immobilization. In a large series of paediatric spinal fractures, the thoracic region (T2–T10) was most frequently injured29. The most common mechanism of trauma to the thoracic spine is hyperflexion, usually due to falls or sporting injury.This usually leads to a stable anterior wedge compression fracture, most commonly in the mid thoracic region. Multiple wedging occurs more commonly than a single fracture, and the diagnosis may be made on plain radiography.

Injuries to the lumbar spine Hyperflexion injuries leading to wedge compression of vertebral bodies also occur in the lumbar region, but the latter area has the further association of fractures from the use of lap seatbelts as restraints in cars (Fig. 68.23). These seatbelts tend to ride up onto the anterior abdominal wall in children, leading to transmission of forces to the abdomen and spine in accidents.This leads to a high incidence of associated injury to the abdominal viscera and spinal fractures that are often more complicated than simple wedge compression.

Figure 68.22 Spinal cord injury without radiographic abnormality (SCIWORA). Sagittal T1-weighted MRI demonstrating injury to the upper cervical cord (arrows) without radiographic abnormality following birth trauma.

Figure 68.23 Lumbar spinal fracture. Sagittal reformat of CT showing lumbar spinal fracture following a seatbelt injury in a child. The fracture has split the spinous process posteriorly and there is a more subtle component affecting the anterior column as well.

A flexion distraction mechanism is responsible, causing ligamentous or bony disruption of the posterior column with an associated injury to the vertebral body and/or disc anteriorly. The injuries most commonly occur between the first and third lumbar vertebrae and are unstable if the posterior elements are disrupted. Conventional abdominal CT has been shown to miss many such injuries30, and plain radiographs of the lumbar spine, in particular the lateral view, are essential in this situation. CT has a definite role in demonstrating associated visceral trauma, and may also be utilized in a dedicated fashion to examine a known spinal fracture. Although the traumatized area may lie inferior to the conus, MRI is useful in assessing any associated damage to the cauda equina or filum terminale.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

RADIOLOGY OF NONACCIDENTAL INJURY Padma Rao The spectrum of injuries encompassed by the terms nonaccidental injury (NAI) or child abuse range from the welldocumented physical and sexual abuse to the less frequently publicized emotional abuse, neglect, deprivation and abandonment. Literature on the presentation and radiology of NAI is extensive and dates back to 1860 when Tardieu published what is thought to be the first article on the concept of the battered child31. Exact figures on prevalence are difficult to establish as there are cultural and geographical variations throughout different societies as to the definition of socially acceptable limits of discipline and parental control. Child abuse is not confined to any one social class or strata of society. The reported incidence of NAI has significantly increased over the past three decades, partly related to society’s greater awareness and understanding of the problem, and also to the expansion in the use of diagnostic imaging in its detection and follow-up32. It is imperative that those involved in abuse cases remain objective in their judgement.The radiologist’s role is to provide an objective assessment of the radiology and this is best carried out in the absence of any contact with the patient or his/her family. Physical signs of abuse are not always present and radiological abnormalities alone, sometimes picked up incidentally, may provide the only evidence and confirmation of abuse.The radiologist may be the first to arouse suspicion of abuse. The radiologist’s role in the management of a child can be summarized as follows: • To document the radiological abnormalities and injuries present. • To recognize specific injuries or patterns of injury as pathognomonic of abuse. • To raise the suspicion of NAI and to perform a complete radiographic skeletal survey to determine the extent of bony and soft tissue injury. • To differentiate the injuries from normal variants in the immature skeleton and to exclude underlying brittle bones or other diseases that may mimic NAI. • To confirm a radiological diagnosis of NAI by an assessment of the number, type and differing ages of fractures with an inappropriate or incompatible history. • To recommend further investigations as appropriate, including repeat radiography and bone scintigraphy to assist in dating fractures and confirming previously undetectable fractures, CT and MRI for intracranial and neurological abnormalities, and specific abdominal investigations to detect visceral injury. Good communication and prompt liaison with the paediatric team, casualty department and social work team is essential in order to ensure appropriate handling of a sensitive topic. If there is any doubt as to the presence or significance of a lesion, a second opinion should be sought from a paediatric radiologist experienced in NAI cases. A wrong diagnosis of NAI can be devastating for the family. A missed diagnosis of NAI may

condemn the child to continued abuse, which may ultimately lead to death. The imaging findings must be considered together with the clinical history and presentation and be correlated with the child’s age and stage of development. The radiologist must be familiar with the typical accidental injuries of childhood and have an understanding of the mechanism involved in inflicting the injuries.There is increasing reliance on multidisciplinary discussion at, for example, case conferences. The clinical presentation is extremely variable and the child may present with the same symptoms and signs as in accidental trauma or medical illness. The initial diagnosis is often that of a medical emergency, such as meningitis or status epilepticus. Symptoms may range from minor vague symptoms (and be dismissed as insignificant) to severe life-threatening shock. The presence of skeletal injuries or typical soft tissue injuries makes a diagnosis of NAI easier but these are not always present. They are often absent in those injuries involving acceleration/deceleration forces alone. Unexplained soft tissue injury in children, especially if extensive or penetrating, should be further investigated for accompanying skeletal injury. Characteristic soft tissue injuries in abuse are bruises, a torn frenulum, cigarette burns, bite marks and scalds, usually of the lower extremities and buttocks. Retinal, subhyaloid and subconjunctival haemorrhages are associated with shaking. Retinal haemorrhage that occurs in normal neonates as a consequence of the high pressures generated during delivery resolves within a few weeks of birth. Accidental bruising usually occurs over bony prominences. In abuse, bruising occurs frequently and multiple bruises of differing ages found in sites unlikely to be affected by accidental injury, such as the flanks or buttocks, are typical of NAI. On plain radiographs there may be swelling and disruption of the normal soft tissue planes at the site of the bruise. Calcified soft tissue haematomata may be apparent. Calcification in the soft tissues around the neck (necklace calcification), represents fat necrosis following strangulation33. The following circumstances should alert the physician to the possibility of NAI: • The presence of an inappropriate, inconsistent, or conflicting history. • The presence of unexplained soft tissue injury. • The presence of a healing fracture or the presentation of the child in a shocked or dehydrated state resulting from a delay in seeking medical help. • Radiological evidence of trauma exceeds that expected from the clinical history. • The presence of skeletal injuries with a high specificity for abuse. Other suspicious circumstances include a failure to thrive of unknown cause suggesting emotional deprivation, poor nutrition and neglect; unexplained abdominal trauma; recurrent pancreatitis; and a history of previous abuse to this child or a sibling.

1621

1622

SECTION 8

•

PAEDIATRIC IMAGING

SKELETAL AND SOFT TISSUE INJURY

Table 68.2

BONE SCINTIGRAM PROTOCOL

Patient preparation

The main determinants of the sequence of diagnostic imaging in abuse are the cardiovascular and neurological status of the child. Stability must be achieved before proceeding with further investigations.

None before injection. After injection, maintain good hydration Empty bladder before delayed images ± Sedation

Radiopharmaceutical

99m

Investigation Skeletal survey

Dose

Weight adjusted percentage of adult dose

A recommended schedule of images is given in Table 68.1. A ‘babygram’ is not an adequate skeletal survey and should not be performed. In cases of sudden infant death syndrome (SIDS), the skeletal survey should be performed and ideally reported by a paediatric radiologist before the post-mortem. Questionable areas on the initial survey should be supplemented by additional views.The skeletal survey should be checked and deemed satisfactory before the child leaves the department. Repeat radiographs after 10–14 d enhance detection of occult fractures, especially of the ribs, and provide additional information on the number, type, and age of the injuries34. The yield from a skeletal survey decreases with increasing age and is of most value in those under 2 years of age35. A skeletal survey should only be performed in older children if clinically it is thought appropriate.

Images

Tc-methyldiphosphonate

Adult dose: 800 MBq

Radioangiogram

If there is an obvious region of interest

Blood pool

Posterior and anterior whole body passes

Delayed

3 h post injection—posterior and anterior whole body passes Anterior and oblique views of ribs Right and left lateral of skull Anterior lower legs: toes turned in to separate bones Anterior and posterior ‘spot’ views of torso Forearms (separating radius and ulna)

Pinhole images

For suspicious area or small region if magnification may assist resolution

Tomography

Rarely performed

Scintigraphy The role of skeletal scintigraphy in suspected or proven NAI is established. A recommended bone scintigram protocol is outlined in Table 68.2. Studies have advocated performing both scintigraphy and a skeletal survey36,37; an increase in sensitivity has been demonstrated of between 25% and 50% in detecting soft tissue as well as bony lesions, assuming adequate patient immobilization and positioning38. Pinhole collimation should be available to resolve occasional difficulties. The urinary bladder should be emptied before the delayed images of the pelvis to resolve superior and inferior rami fractures. The nuclear medicine service needs to have sufficient and specific expertise Table 68.1

SKELETAL SURVEY

AP and lateral chest—lateral to include sternum Upper limbs—AP both forearms, AP both humeri, PA both hands and wrists

in order to interpret the studies accurately and it is therefore recommended that they are performed at a dedicated children’s unit. They should not be performed or interpreted by the occasional practitioner. Familiarity with the normal scintigraphic appearances in children is essential. The normal high uptake in the epiphyses should not be confused with fractures. Physiological periosteal new bone and the periosteal prominence accompanying rapid growth do not show increased isotope uptake. Scintigraphy is most sensitive in detecting rib, scapular, spinal, diaphyseal and pelvic fractures. Sensitivity is lower for flat bones, old healed fractures and metaphyseal fractures38. Scintigraphy is not reliable in identifying skull fractures because of the low lesion-tobackground ratio and the poorer osteoblastic response in linear fractures. Scintigraphy becomes positive within hours of an injury and has been demonstrated as early as 7 h after injury37.

Lower limbs—AP both femora done individually AP both tibiae and fibulae

Ultrasound

AP both ankles, coned, with ankle joints flexed at 90 degrees

US is not routinely performed but may be a useful supplementary technique. It can demonstrate subperiosteal haemorrhage, fracture separation of the epiphysis, costochondral injuries and occult long bone fractures before they become evident radiographically.

Dorsiplantar both feet AP abdomen and pelvis Lateral of thoracolumbar spine Lateral cervical spine AP and lateral skull—add a Townes’ view if there is occipital injury

Dating fractures

Additional views at radiologist’s discretion

Precise dating of fractures is not possible. Estimates are based on the pattern of fracture healing observed in accidental trauma (Table 68.3) when both the time of injury and the interval before investigation are known. Fracture dating is easiest if the time interval between injury and the initial radiography is short. Subtle periosteal new bone may be seen as early as 4 d after a fracture. Fractures with a large amount of periosteal new bone or callus are more than 14 d old.

All films to be checked by a radiologist. It is essential to have the following:

• metal markers on all films • correct patient identification labelling • radiographs should ideally be done on high resolution film but there is increasing use of modern digital equipment.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

Table 68.3 DATING FRACTURES: RADIOGRAPHIC CHANGES IN CHILDHOOD FRACTURES (ADAPTED FROM O’CONNOR J F, COHEN J 1987 DIAGNOSTIC IMAGING OF CHILD ABUSE. WILLIAMS & WILKINS, BALTIMORE, P112) Radiological feature

Early

Peak

Late

Soft tissue resolution

2–5 days

4–10 days

10–21 days

Periosteal new bone

4–10 days

10–14 days

14–21 days

Loss of fracture line definition

10–14 days

14–21 days

42–90 days

Soft callus

10–14 days

14–21 days

2 years to physeal closure

Hard callus

14–21 days

21–42 days

Remodelling

3 months

1 year

Injuries to the neck, larynx and airway Oropharyngeal injuries Injury to the mouth is common in abuse but radiological evidence of orophayngeal injury is uncommon. Pharyngeal perforation may occur secondary to forceful insertion of a blunt or penetrating object or orosexual abuse. The children are usually younger than 1 year old. Retropharyngeal abscess or mediastinitis may result from any cause of retropharyngeal perforation. Perforation may be further complicated by haemorrhage, interstitial emphysema and mediastinal pseudocyst formation39–41. Radiologically, there may be widening of the superior mediastinum on a frontal chest radiograph. A soft tissue lateral radiograph of the neck demonstrates prevertebral soft tissue thickening which may contain gas or a fluid level, or a foreign body may be evident. Oral contrast studies may demonstrate an extraluminal leak. CT and MRI are sensitive for soft tissue abnormalities.

Asphyxiation This most frequently presents as a near miss cot death or SIDS, or recurrent apnoeic attacks. Often asphyxiation occurs in the absence of external visible signs of injury or skeletal injury. Occasionally there may be bruising. There are no specific radiological signs. A chest radiograph may demonstrate aspiration or pulmonary oedema. Both may contribute to the hypoxia. Severe or recurrent episodes of asphyxiation may result in hypoxic ischaemic brain injury.

Skeletal injuries The frequency of skeletal trauma in NAI has been variously reported as between 11% and 55%42. It is most common in children under 3 years of age43,44.The following fractures, which occur infrequently, are considered to have a high specificity for abuse: • metaphyseal, rib and scapular fractures • fractures of the outer third of the clavicle • sternal fractures • spinous process fractures. Other sites and patterns of injury which have moderate specificity for abuse are: • multiple fractures, bilateral fractures and fractures of differing ages • vertebral fractures or subluxation • digital injuries in nonmobile children • spiral fractures of the humerus

• epiphyseal separations • complex skull fractures, i.e. those that are wider than 5 mm, depressed, occipital and growing fractures. Fractures that occur frequently but have a low specificity for abuse are: • mid-clavicular fractures • simple linear skull fractures of the parietal bone • single diaphyseal fractures, with the exception of spiral fractures of the humerus • greenstick fractures. These injuries occur commonly in young mobile children and, if isolated, should not be regarded as significant evidence for abuse. Mid-clavicular and humeral fractures are also wellknown birth injuries. Again, they are isolated and there is an appropriate history.

Metaphyseal fractures Also known as corner or bucket handle fractures (Fig. 68.24), metaphyseal fractures are highly characteristic of and specific for NAI with an incidence of between 11% and 28%43,45.They are most commonly seen in nonmobile abused infants under 18 months of age, and are most frequent around the knees and ankles but also occur at the shoulder, elbow, wrists and hips45. They may be bilateral. Metaphyseal fractures do not usually occur in children over 2 years of age. In these children, trauma to the metaphysis results in diaphyseal or epiphyseal fracture separation. Histological changes consist of a subepiphyseal planar series of microfractures through the metaphyseal primary spongiosa, which results in separation of a part or whole of a mineralized disc and this is the portion evident radiologically42,46. The fractures are transmetaphyseal46. They may occur during shaking when acceleration/deceleration forces are applied directly to the limb, but a further well-recognized mechanism is direct wrenching or twisting of the limb using the extremities as ‘handles’. At the metaphysis, the loosely attached periosteum is frequently stripped by the shearing forces, resulting in subperiosteal bleeding. During healing, this is evident as subperiosteal reaction.With small metaphyseal fractures, periosteal new bone is not seen as the fracture is entirely intracapsular and subperiosteal bleeding does not occur. The most subtle metaphyseal fracture is the metaphyseal lucent line which lies immediately adjacent to the epiphyseal plate (Fig. 68.24E)42,47.

1623

1624

SECTION 8

•

PAEDIATRIC IMAGING

Figure 68.24 Metaphyseal fractures. (A,B) Radiographs of the right femur (A) and both ankles (B) of a 2 month old abused infant demonstrating metaphyseal corner fractures of the distal femur and both distal tibia (arrows). The angled tangential view reveals the ‘bucket handle’ appearance of the fracture. (C) Radiograph of the left ankle of an infant demonstrates a metaphyseal corner fracture of the distal tibia (arrow). (D) Angled tangential view of the right lower limb of an abused infant demonstrates the ‘bucket handle’ appearance (arrow). (E) Radiograph of the right ankle of a 6 week old abused baby with subtle metaphyseal fractures evident as a metaphyseal lucent line (arrow).

Dating metaphyseal fractures may be difficult as they do not always result in callus formation and healing occurs by gradual bone consolidation. Repeated trauma may lead to an irregular fluffy appearance. Metaphyseal fractures are usually asymptomatic and not evident clinically. They are detected incidentally when imaging for other reasons. There are several normal variations in the appearance of the metaphyseal regions that should not be mistaken for abuse, namely, metaphyseal beaks and spurs, the metaphyseal step-off and the proximal tibial cortical irregularity48.

Diaphyseal fractures Diaphyseal fractures are the most common fracture in child abuse and are said to occur four times more frequently in NAI than metaphyseal fractures43. The femur, humerus and tibia are the most frequently injured bones (Fig. 68.25). A variety of frequencies and fracture patterns have been quoted43,49 but no one pattern or type of diaphyseal fracture is specific to NAI. The significance of diaphyseal fractures increases when they are multiple, bilateral, found in a state of healing, are of differ-

ing ages, when there is a fracture through the callus, and when they are associated with fractures that have a high specificity for abuse. Transverse diaphyseal fractures result from direct trauma. The infant may be pulled inappropriately by the limb and the bone fractures due to the weight of the suspended child against the fixed adult hand. Spiral fractures are more common than transverse fractures in both accidental and nonaccidental trauma. They are produced by a twisting/pulling force and are always suspicious of abuse, in particular those of the humerus. It is important to differentiate accidental fractures and normal variants from NAI. In mobile children, the ‘toddlers’ fracture’ (a fine spiral fracture of the tibia; see Fig. 68.17), supracondylar fractures and metaphyseal torus fractures are common accidental injuries of childhood. However, in the appropriate clinical setting, any type of fracture can occur in abuse. The importance of all humeral fractures as indicators of abuse has been emphasized50. Femoral fractures have been said to have a high incidence of abuse in nonmobile children,50,51 but this has recently been debated52. Fractures at the subtrochanteric level are stated to be more common in NAI51.

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

Figure 68.25 Diaphyseal fracture. Radiograph of the left lower leg of a 3 month old abused child demonstrating a mildly diastatic oblique fracture of the distal tibia. There are also metaphyseal corner fractures of the proximal tibia and distal fibula (arrows). A faint metaphyseal lucent line of the distal tibia is also noted.

is abnormal. It is usually seen in fracture repair but can occur in the absence of a fracture as a consequence of a gripping or twisting force44,53,54 or acceleration–deceleration forces alone54,55. In young children, the periosteum has only a loose attachment to the underlying bone and it readily separates following the formation of a subperiosteal haematoma in trauma. This results in periosteal reaction evident radiologically about 7 d post injury. Physiological periosteal new bone shows normal uptake on scintigraphy whereas uptake is increased in traumatic and other causes37. Repetitive injury, failure to immobilize the limb and severe twisting lead to a florid periosteal reaction which may cloak the bone and extend into the epiphysis, and this is characteristic of abuse cases. Clinically, this may present as a hot swollen limb and mimic osteomyelitis, metabolic bone disease, or dysplasia.

Digital injury Common injuries to the hand include bruises, burns and lacerations, and rarely (0–3%) lead to fractures. These usually result from direct injury to the digits from trampling, squeezing, or forced hyperflexion35,55.Accidental fractures in toddlers are usually confined to the phalanges, but in abuse, involvement of the metacarpals and metatarsals is seen.

Vertebral injuries Impaction fractures Impaction buckle fractures occur most commonly at the metadiaphyseal junction of the distal femur and proximal tibia42. They result from impaction forces transmitted when the child is forcibly thumped down on his/her legs on a hard surface. The bone cortex buckles anteriorly and there is an incomplete crush fracture of the bone shaft. They must be differentiated from accidental torus fractures, which also occur at the metadiaphyseal junction, typically at the distal tibia. Impaction fractures also occur in the spine.

Epiphyseal fractures Compared with accidental trauma, the frequency of epiphyseal and true Salter–Harris fractures is rare. Fracture separation of the epiphysis and dislocation is sometimes seen and can be detected by US before becoming radiologically visible44. In children, movement is prevented by the epiphysis being held tightly in position by the periosteum. The proximal femur and humerus are the most common sites42. In humeral fractures, the mechanism of injury is external rotation of the forearm. In abuse the forearm is displaced medially; in accidental injury the displacement is lateral49. Epiphyseal injuries are commonly complicated by orthopaedic deformity and growth disturbances.

Periosteal new bone Physiological periosteal new bone is seen as a normal physiological phenomenon between 6 weeks and 6 months of age, and always has an organized lamellar appearance. It occurs symmetrically, is more common in the lower limbs, and is confined to the diaphysis, never extending to the metaphysis. Outside this age range the presence of periosteal new bone

These occur relatively rarely in NAI. The true incidence is unknown. Unless a lateral spine radiograph is performed they are frequently missed. They are due to hyperflexion and hyperextension forces arising from direct trauma or impaction against a hard surface42, resulting in varying degrees of compression fractures of the vertebral bodies most commonly in the thoracolumbar region. Vertebral injury may be further complicated by rupture of the spinal ligament, vertebral dislocations, disc herniation, avulsion of the posterior elements which, if the injury is high enough, can lead to kyphosis, cord damage and paraplegia. Babies are susceptible to cervical spine damage because of their flexible supporting ligaments and joint capsule. MRI is indicated in the presence of abnormal neurology.

Rib fractures The presence of rib fractures in children, whether solitary or multiple, is highly suspicious of child abuse. Accidental rib fractures are extremely rare even in the presence of major trauma. Once metabolic bone disease, rickets, prolonged parenteral nutrition, jaundice, frusemide therapy, bronchopulmonary dysplasia, osteogenesis imperfecta and bone disease of the premature have been excluded, rib fractures are considered synonymous with abuse56. The incidence of rib fractures in children with proven NAI varies between 5% and 27%57. Eighty per cent of rib fractures in NAI are clinically occult and are not associated with bruising or specific symptoms. They are discovered incidentally during the skeletal survey or when imaging for other problems. Fractures are often multiple and bilateral and affect the necks, posterior shafts and axillae42,44, particularly medial to the costotransverse articulation (Fig. 68.26). Anterior rib fractures are also recognized but are less common.

1625

1626

SECTION 8

•

PAEDIATRIC IMAGING

Figure 68.26 Rib fractures. (A) Chest radiograph of an intubated 3 month old abused infant demonstrating fractures of the angles of the right third, fourth and fifth ribs associated with pulmonary contusion (arrows). (B) Oblique left-sided rib views demonstrating multiple fractures of the angles of the left third to eighth ribs at different stages of healing (arrows). (C) Healing fractures of the posterior aspects of the left seventh and eighth ribs picked up as an incidental finding on a baby who presented with abdominal signs (arrow). Nonaccidental injury was later confirmed.

Rib fractures most commonly occur during episodes of violent shaking whilst the thorax is squeezed and compressed44,45,58. Lateral compression causes posterior fractures, and AP compression, such as from trampling, causes fractures in the mid-axillary line. Often both coexist. They can also result from direct trauma to the chest, such as trampling or kicking, which is often the cause in older children, and are rarely accompanied by underlying lung contusion, but a pleural haematoma may be seen acutely. Unlike adults, the forces sustained in cardiopulmonary resuscitation do not normally cause rib fractures in children59,60. Approximately 90% of rib fractures in abuse occur in children under 2 years of age and most are in those younger than 1 year old. A row of fractures involving the necks and lateral aspects may result in a flail chest. Early after injury, fractures are often difficult to detect. Rib fractures become easier to see during healing44, and a repeat chest radiograph should be obtained 10–14 d later in suspected NAI cases. The more subtle signs of rib fractures are often overlooked, including expansion and widening of the ribs, frequently at the neck.These fractures are pathognomonic of abuse. Fractures of the first rib are rare in both NAI and accidental trauma. If they occur, they are usually sited laterally rather than posteriorly61. The amount of callus formation in healing rib fractures is determined by the degree of bleeding and cortical disruption. The ‘hole in the rib sign’ refers to the cyst-like radiolucencies in the bones arising as sequelae to fractures.

Clavicular and scapular injury The frequency of clavicular fractures in abuse is between 2% and 6%45. In mobile children, the clavicle is frequently injured in accidental trauma, usually in the midshaft, and it is the most frequently injured bone in birth trauma. Birth injuries are usually isolated and periosteal new bone is seen at around 11 d and mature callus at 1 month. In abuse, midshaft fractures also occur but the incidence of fractures to the medial and outer thirds of the clavicle increases. A fracture of the outer third

of the clavicle is highly associated with abuse. Accidental clavicular fractures are rare under the age of about 3 years. They do not always present with immediate symptoms. The child later presents with persistent localized pain or swelling and a radiograph confirms a healing fracture. Scapular and sternal fractures are rare but both are highly specific for abuse. In the scapula, the acromion is the most common site of injury (Fig. 68.27).These fractures may not be evident on radiographs and are only detected on scintigraphy.

Skull fractures Skull fractures in abuse may not necessarily differ from those occurring in accidental trauma, and no specific type of skull fracture is pathognomonic of child abuse62. Any type of fracture can occur (Fig. 68.28). The types of skull fracture that are rare in uncomplicated accidental trauma and may be associated with abuse are listed in Table 68.4. Skull fractures that are suspicious of abuse are those that are multiple and diastatic (Fig. 68.29); of differing ages; complex, involving both sides of the skull; depressed and nonparietal, especially those of the occiput. A depressed occipital fracture is considered specific for abuse. Skull fractures must be differentiated from normal variants and normal sutures that may mimic fractures. Growing fractures of the skull, or traumatic encephaloceles, are sometimes seen in abuse but are not specific to it. Skull fractures are difficult to date as they do not heal by callus formation. A fracture is usually more than 2 weeks old if its edges are rounded and smooth. A scalp haematoma overlying the injury site usually develops immediately or within hours of the injury, thus indicating that an injury is likely to be recent. Most disappear after 3–4 d. A delayed appearance of a haematoma is a recognized although rare occurrence and is due to continuous slow extravasation of blood and CSF. The head injury may be complicated by a nontender swelling beneath the scalp, termed a subepicranial hygroma, which spontaneously disappears with no residual effects. It is due to a collection of CSF beneath the subepicranial aponeurosis63, is generally

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

Figure 68.27 Fracture of the acromion. AP radiograph of the right shoulder and humerus of an abused child demonstrates a fracture of the right acromion process (black arrow). There is also a metaphyseal fracture of the distal humerus (white arrowhead). A thick pathological periosteal reaction of the proximal humerus extending into the metaphysis is evident. [From Rao P 1999 Emergency imaging in non-accidental injury. In: Carty (ed) Medical radiology: Emergency paediatric radiology. SpringerVerlag, Berlin.]

considered to be a benign complication of head injury, and is not associated with any significant intracranial injury. Simple skull fractures in accidental trauma have a very low risk of intracranial sequelae64,65. Most head injuries in young children occur from falls from short distances which impart a predominantly translational (linear) force to the head. In one study, whilst differing fall heights were associated with some differences in the type of injury, most household falls were benign66. Minor accidental falls from heights of 4 feet or less are very rarely associated with serious or fatal injury67–70. In severe injuries or fatalities said to result from falls from heights of less than 4 feet, the putative history is likely to be inaccurate and should be viewed with a high degree of suspicion69. In the presence of major injury, a history of minor trauma is clearly incompatible and the reliability of the history and thus NAI must be queried. Difficulties arise in abused patients as the history is often vague and inadequate. Medically, skull fractures become important if they are depressed, have intracranial penetration with resultant CSF leaks and risk of meningitis, or are basal and often accompanied by brainstem injury. In the absence of a clear history of injury or bleeding disorder, an apparently spontaneous subdural haematoma on CT or MRI in an infant is caused by blood leakage from torn delicate dural veins and is usually caused by a shaking or shaking/ impact injury. It is often accompanied by retinal haemorrhages but may be associated with a skull fracture. A more detailed account of the intracranial manifestations of NAI is given in the section below on brain injuries in NAI.

Differential diagnosis of abuse fractures Several medical conditions may be associated with fractures, periosteal reactions and irregular metaphyses and these may

Figure 68.28 Skull fractures. (A) Lateral skull radiograph demonstrating complex, diastatic temporoparietal skull fractures associated with pneumocephalus. A linear occipital component is also present. (B) Lateral skull radiograph demonstrating multiple skull fractures in the frontal and parietal regions (arrows).

1627

1628

SECTION 8

•

PAEDIATRIC IMAGING

Table 68.4 FEATURES OF SKULL FRACTURES WITH A HIGHER SPECIFICITY FOR ABUSE Complex fractures involving both sides of the skull Multiple fractures Diastatic fractures Fractures of differing ages Depressed fractures especially of the occiput

cause diagnostic confusion. In some, other associated radiological findings enable a differentiation to be made, and in others it is the nonradiological extraskeletal manifestations of the disease that aid the diagnosis (Table 68.5). Confirmation of the underlying medical condition does not automatically exclude NAI; the two may occur concurrently.

Abdominal and thoracic injury These injuries have a lower incidence in NAI compared with skeletal and brain injuries. They are associated with significant morbidity and mortality rates in the order of 50%71. With increasing age, there is a clear change in the site of fatal injury from the brain to the abdomen. The average age for blunt abdominal trauma and fatal visceral injury in abuse is 2 years, i.e. after the child has become mobile. In NAI, the duodenum, pancreas and mesentery are injured more frequently than the liver, kidneys and spleen.The reverse is true for accidental trauma72,73. In children younger than 5 years, in the absence of road traffic accidents, significant

Figure 68.29 Diastatic skull fracture. Lateral projection of the skull of an 8 month old abused infant demonstrating a complex diastatic temporal fracture (arrow).

abdominal injuries are extremely rare. In NAI, injuries occur secondary to direct blunt trauma or to sudden deceleration after the child is thrown. Clinical presentation varies with the severity of injury and the interval between injury and presentation. The child may present acutely with shock, hypovolaemia and peritonitis, or more insidiously with chronic weight loss, malaise,

Table 68.5 DIFFERENTIAL DIAGNOSIS OF NONACCIDENTAL INJURY IN CHILDREN (ADAPTED FROM KLEINMAN P K (ED) 1998 DIAGNOSTIC IMAGING OF CHILD ABUSE. WILLIAMS & WILKINS, P179) Shaft fracture

Periosteal reaction

Metaphyseal abnormality

Nonaccidental injury

+

+

+ ±

Disease Normal bone density Birth trauma

+

±

Congenital indifference to pain

+

+

+

Myelodysplasia

+

+

+

Osteomyelitis

−

+

+

Congenital syphilis

−

+

−

Vitamin A intoxication

−

++

−

Caffey’s disease

−

+

−

Prostaglandin E therapy

−

Metaphyseal and spondylometaphyseal dysplasias

−

−

+

Osteopenic bones Osteogenesis imperfecta

+

−

±

Rickets

+

+

+

Scurvy

−

+

+

Leukaemia

−

+

−

Methotrexate therapy

+

−

±

Menkes’ kinky hair syndrome

−

+

+

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

unexplained mass lesion, or functional organ impairment. The diagnosis is often delayed or missed in these late presentations and is associated with repeated admissions to casualty or visits to the general practitioner and a higher morbidity and mortality. Typical abuse injuries may be an incidental finding detected only when imaging for other reasons, e.g. lower rib fractures on an abdominal radiograph if the child presents to the surgeons with a presumed obstruction or an acute abdomen (see Fig. 68.26C). Vomiting is a nonspecific sign occurring in both acute and chronic presentations. It can be a manifestation of either abdominal or intracranial injury. If investigations fail to identify an intra-abdominal cause, an intracranial cause must be sought.

Investigation Blood pressure is an unreliable predictor of major organ injury in children as a normal blood pressure may be maintained despite significant blood loss, and hypotension may only be apparent when the child is moribund and almost dead72. If the child cannot be stabilized, imaging plays no part and immediate surgery is usually required. Once stable, imaging can proceed.The abdominal radiograph is not a sensitive investigation and demonstrates pneumoperitoneum in less than 30% of patients with visceral rupture. The roles of US and CT in accidental and nonaccidental abdominal trauma in both adults and children have been extensively studied74–76. CT is the ‘gold standard’ and remains the investigation of choice for the complete assessment of the abdomen and thorax. The advantages of US in the acute setting are that it is portable and readily available. It can give good information on solid organ injury and can detect free fluid, but overall assessment of the abdomen is limited. The relative roles of diagnostic peritoneal lavage (DPL) and CT in the assessment of the traumatized patient have been well debated77,78. CT should be performed before DPL as the presence of lavage fluid masks the presence of free fluid on CT, making it difficult to interpret its significance. All CT for trauma should be performed with intravenous contrast medium. Inspection of the images on both soft tissue and wide window settings increases the sensitivity in detecting small amounts of extraluminal air.

inability to void and increasing bladder distension or blood at the urethral meatus79,80. Bruising and/or perineal haematoma and pelvic haematoma are known sequelae. Suspected urethral rupture in boys should be confirmed by retrograde urethography before passing a urinary catheter in order to prevent converting a partial tear to a full tear. Pancreatic injuries Trauma is the most common cause of acute pancreatitis in childhood. Acute haemorrhagic pancreatitis is highly suggestive of NAI once hereditary pancreatitis and accidental injuries have been excluded. The child may present acutely with abdominal pain, vomiting and shock, or more insidiously with weight loss, anorexia and chronic abdominal pain; it is the latter group who may present repeatedly to their general practitioner or casualty and in whom the diagnosis is often missed. More rarely, vomiting and symptoms of bowel obstruction or a palpable mass due to a pancreatic pseudocyst are present81. Due to its retroperitoneal location, signs of pancreatitis are classically absent. The evolution of clinical signs may be slow and associated injuries often mask the few signs that are present. Hence, the diagnosis is frequently delayed, particularly as the rise in serum amylase, although a nonspecific marker, may also be delayed for 24 h or more. In NAI the usual mechanism of injury is a direct blow over the midline where the pancreas crosses the vertebral column. The pancreas is most frequently injured at the junction of the body and tail, and acute pancreatitis, pancreatic abscess, necrosis and pseudocyst are common sequelae. Skeletal and soft tissue injuries may accompany the pancreatitis but the association with medullary fat necrosis at distant sites and lytic osseous lesions are rare. On US, approximately 50% of affected patients demonstrate diffuse/focal gland enlargement and pancreatic duct dilatation (Fig. 68.30), and a decrease in gland reflectivity occurs in the presence of interstitial oedema. Thin-section CT has the highest rate of detection of pancreatic injury although it is quoted as missing one-third of injuries15,82. The CT appearances of acute pancreatitis and its complications are the same

Abdominal injury Solid organ injury In abuse, the liver is injured more frequently than the spleen or kidneys. In accidental trauma, the spleen and kidneys are injured more frequently than the liver.The presentation and radiology of solid organ injuries are essentially the same as those in accidental trauma. Contrast-enhanced CT is the investigation of choice. A detailed account of such injuries is outside the remit of this chapter and the reader is advised to consult specialist texts. Although the overall incidence of renal tract injuries is rare, some injuries are more specific to NAI and include bladder rupture in the absence of a pelvic fracture following a direct blow against a full bladder. Urethral injuries may be associated with sexual abuse, often with associated injuries to the rectum, perineum and genital organs. Common presentations are an

Figure 68.30 Pancreatic duct dilatation. Ultrasound of the pancreas in a young boy performed several weeks after he sustained a direct blow over the midline. In the acute stages the pancreas was enlarged and swollen. During follow-up, the pancreas demonstrates residual duct dilatation (arrow).

1629

1630

SECTION 8

•

PAEDIATRIC IMAGING

as those in other medical aetiologies. CT findings of peripancreatic fluid in the absence of other visceral injury are the best indicator of pancreatic injury83,84. Adrenal injuries These may be incidental findings detected when imaging for other abdominal trauma. Unilateral adrenal haemorrhage tends to be right sided85 and is often clinically silent. Adrenal haemorrhage may be evident clinically when bilateral and acute adrenocortical insufficiency results85. On CT, haemorrhage appears as round or oval low attenuation homogeneous masses within the gland, predominantly in the medulla, but extension into the cortex is seen. There is frequently thickening of the ipsilateral diaphragmatic crus, haemorrhagic infiltration of the surrounding fat and free intraperitoneal fluid or blood. Adrenal injuries may be isolated or occur in association with injuries to the kidneys, spleen and pancreatic tail. Adrenal haemorrhages are not uncommon in the neonate and when present usually resolve within several weeks after delivery86. Injuries to the gastrointestinal tract These severe injuries are often diagnosed late and are thus associated with a high mortality. Bowel injuries are rare in accidental trauma but occur with a relatively high frequency in NAI due to the greater incidence of direct blows across the abdomen.The three main types of injury are mesenteric injuries, bowel disruption from partial or complete transection or perforation and haematoma formation.The mesenteric side of the bowel is more prone to vascular tears and the antimesenteric side to perforation.The duodenum, especially the second and third parts, is the area of bowel most frequently injured in abuse. Injury occurs from direct compression as the

duodenum passes over the vertebral column, from shear injury or from a marked rise in intraluminal pressure when a closed loop is formed between the pylorus and ligament of Treitz87. Intramural haematomas of the duodenum and jejunum are the most frequent findings in bowel trauma in NAI45,88,89. Approximately one-third of patients sustain major vascular injuries. Vascular disruption to the small intestine is secondary to sudden deceleration or results from a compressive force of the fixed duodenum against the spine following blunt trauma. Bleeding into the submucosa and subserosa, secondary to both direct trauma and ischaemic injury, may lead to complete or partial obstruction of the lumen. Clinically, the child usually presents with abdominal pain and vomiting with signs of peritonism and obstruction. In up to 25% of patients, associated biliary and pancreatic injuries occur. The plain abdominal radiograph may demonstrate ischaemic change, proximal bowel obstruction, or free fluid. Upper gastrointestinal contrast studies are very informative. In the acute stages of a duodenal haematoma, an intramural mass with thickening of the proximal folds gives a ‘coiled spring’ appearance. During resolution, signs of previous haemorrhage are seen as localized masses in the duodenal wall with prominent mucosal folds88,90 (Fig. 68.31). The haematoma may act as a lead point and result in intussusception. US may visualize the haematoma directly or demonstrate indirect signs such as proximal duodenal dilatation. Direct trauma may result in bowel rupture and perforation45. The most susceptible parts are the retroperitoneal second and third parts of the duodenum, or at points of mesenteric fixation at or just distal to the ligament of Treitz and the ileocaecal junction. Perforation is the result of both compression and shearing

Figure 68.31 Duodenal haematoma. (A) Upper gastrointestinal contrast study following blunt trauma to the midline demonstrates the haematoma as two low density filling defects in the second part of the duodenum (black arrow) associated with prominent thickened folds giving a coiled spring appearance (white arrow). (B) Resolving duodenal haematoma of the second part of the duodenum seen as localized intramural masses associated with prominent mucosal folds. [From Rao P 1999 Emergency imaging in non-accidental injury. In: Carty (ed) Medical radiology: Emergency paediatric radiology. Springer-Verlag, Berlin.]

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

between opposing surfaces. A clinical presentation of perforation may be delayed and a delay of more than 24 h increases the mortality rate from 5% to 60%. If sufficient free air is present, perforation is readily diagnosed on plain radiography and no other imaging is needed, surgery being the next step. An erect chest or lateral decubitus radiograph can theoretically detect as little as 1–2 ml of free air, but it is not always possible to obtain these in very ill or shocked children, and a horizontal beam lateral or a supine radiograph often has to suffice. CT is again the optimal investigation. A significant amount of unexplained peritoneal fluid is the most frequent sign of bowel rupture. A very sensitive and specific sign of blunt bowel trauma is abnormally intense bowel wall enhancement on contrast-enhanced CT seen in the hypoperfusion complex/shock bowel syndrome91–93. The CT appearances may be confused with bowel rupture, but it is important to distinguish the two as shock bowel requires attention to haemodynamic and fluid balance while rupture requires surgery. Mesenteric injuries include haematoma formation from trauma to small peripheral branches. Severe injuries result from complete avulsion of the superior mesenteric vessels, resulting in mesenteric ischaemia and large quantities of peritoneal blood94. Ischaemic strictures of the bowel are late complications95. On CT there is streaky infiltration of the mesentery, bowel wall thickening, haemoperitoneum and mesenteric haematoma. The stomach is rarely injured in abuse, occurring in less than 1% of cases96, and is usually associated with additional injury to the ribs, spleen and left kidney. Gastric rupture after trauma usually follows a meal when the stomach is full and distended. It is also a rare consequence of failure to pass a nasogastric tube in gastric dilatation secondary to duodenal obstruction or paralysis in acute atony in neglected children. Gastric perforation and rupture are life-threatening events, potentially resulting in septic shock and death45. In 60% of cases, pneumoperitoneum is evident radiographically. In these children there is no need for further imaging and immediate surgery is required. Colonic injury is rare in NAI as the colon is protected from direct trauma by the bony pelvis or by its peripheral location in the abdomen. The types of injury sustained are similar to those in the small intestine. Both the transverse colon and jejunum are subject to the same compressive force as they cross the spine. Colonic haematoma in association with jejunal haematoma has been documented45. Anal and rectal injuries presenting as perineal and pelvic abscesses, faecal peritonitis and rectal tears occur in association with sexual abuse. Colonic perforation may occur from direct stabbing or from forced insertion of foreign bodies.

Thoracic injury Direct contusional injury of the lungs and pleura is rare even in the presence of rib fractures45. Rib fractures or bruising are not always evident and mild injuries may be undetected. In all cases of suspected abuse, if bone scintigraphy is not available, a repeat chest radiograph should be performed after 10–14 d and re-assessed for fractures. Complications of direct thoracic compression, usually as a result of direct punching or stamping, are the same as those in accidental trauma. Pneumothorax and pleu-

ral effusions are not commonly seen in abusive thoracic injury, whereas they are common sequelae in accidental trauma. Contrast-enhanced CT is the investigation of choice and will also determine the presence of any associated diaphragmatic injury. In the presence of both anterior rib fractures and sternal fractures, associated myocardial injury should be suspected. Diaphragmatic injury Rupture may be difficult to diagnose clinically as there may be few physical signs and these may be masked by other injuries. It should be considered if either hemidiaphragm is indistinct. However, the chest radiograph is normal in 10–20% of cases. In intubated patients, the positive pressure ventilation may delay herniation of abdominal contents into the thorax giving a false negative result.The true radiological diagnosis is delayed until extubation when herniation occurs leading to respiratory embarrassment.

Summary Children may suffer from many different types of abuse. As doctors and as radiologists we are concerned with suggesting, diagnosing and confirming NAI. It is incumbent upon all of us to be aware of the possibility of abuse.There may be no skeletal manifestations or external soft tissue markers and radiology may provide the only documentation of abuse. There should be no hesitation in gaining a second opinion if uncertain or in doubtful cases. Good relationship with the paediatricians and support services is essential. Our ultimate aim is the protection of and welfare of the child.

BRAIN INJURIES NAI is the leading cause of serious head injury and death in infants under 2 years of age32,58. External indicators of injury such as skull fracture (see above) or bruising may be absent97. The clinical presentation is often nonspecific. The child may present repeatedly to the general practitioner or casualty department with vague symptoms, such as irritability, lethargy and feeding difficulties. These symptoms are often overlooked due to their lack of specificity but they are often a precursor to more serious NAI. Acute severe presentation, such as fits, vomiting, or collapse may occur98,99. In the absence of bruising the initial presumed diagnosis is often meningitis.

Intracranial manifestations of head injury Most children who suffer nonaccidental brain injury are under 2 years of age.The initial brain injury in head trauma is known as the primary injury.The two main mechanisms of intracranial trauma are impact injuries, including acceleration/deceleration and whiplash shaking injuries. Pathophysiological responses to the primary injury, such as cerebral oedema, cerebral congestion, fall in cerebral blood flow, shock and cerebral vasospasm may result in the secondary injury—hypoxic ischaemic damage99,100. The term ‘whiplash shaken baby syndrome’ describes the association of subdural and/or subarachnoid haemorrhage, massive cerebral oedema, fractured ribs and metaphyseal injury in the absence of external signs of cranial trauma101,102.

1631

1632

SECTION 8

•

PAEDIATRIC IMAGING

Shaking injuries usually occur in children younger than 1 year of age but may occur up to about 2 years, the child being held by the rib cage and squeezed, resulting in fractured ribs and high central venous pressure. Retinopathy, common in NAI, is probably a consequence of high central venous pressure. In older children, who are held by the shoulders or upper arms, subperiosteal bleeding and periosteal reaction in the humeri occur. The young infant has a relatively large head in relation to his/her body, which, when associated with weak neck muscles, results in poor head support and little control. The whiplash movement causes the head to lash to and fro and rotate. Both duration and velocity of the action are important factors.The baby’s brain is relatively small in relation to the size of the cranium and, as the meninges are loose, it starts to move in different directions, which generates shearing forces within the skull and brain. This action is exaggerated by the rotational element which is important in producing the brain injury and associated subdural haemorrhage58,62,101,103–105. Shaking alone can cause brain injury. However, many shaken children are also thrown at the end of the shake, the head impacting against a hard surface and resulting in a shaking–impact injury. The skull fracture is often the last event in the inflicted trauma. Brain injury may be further contributed to by hypoxia sustained as the child’s chest is squeezed during shaking causing impaired ventilation and respiratory embarassment106. Both carotid occlusion secondary to violent neck motion, or direct strangulation106 and suffocation, such as by a hand or pillow in blocking the mouth in order to suppress crying, can contribute to hypoxia107,108. Innocent activity, such as playfully lifting the child above the head, does not result in brain injury. The degree of force needed is such that a neutral observer who saw a child being shaken would recognize the violence of the act and would regard it as dangerous32. The major complications of intracranial injury in abuse are subdural haematoma, cerebral oedema, hypoxic ischaemic encephalopathy, cortical contusions and shearing injuries. Rarer manifestations include intracerebral and intraventricular haemorrhage and extradural haematoma. Cortical contusions occur from impact trauma and may be accompanied by a fracture. If the head is stationary, they occur at the site of impact. If the head is moving, contre coup injuries distant from the site of impact may occur. Contusions tend to occur in the temporal and frontal lobes, particularly the parasagittal cerebral cortex adjacent to the falx100. However, they are more common in older children and in accidental injury, and are less frequent in NAI (Figs 68.32, 68.33).

Subdural haematoma Subdural haematoma is the most common intracranial finding in NAI. The finding of a subdural haematoma has a high specificity for abuse. It results from trauma associated with rotation between the brain and dura causing shearing of the bridging veins in the subdural space109. The force required to produce a subdural haematoma is severe. Accidental causes of subdural haematoma in infants and children are rare, occurring in, for example, major road traffic accidents65,110,111. In these instances the incident has usually been

Figure 68.32 Frontal lobe contusion. Axial unenhanced CT of the brain demonstrating an area of low density in the right frontal lobe consistent with contusion. There is an area of mixed high and low density in the left frontal parasagittal region consistent with haemorrhagic contusion. Associated subdural blood is also noted.

witnessed and a history and other injuries compatible with the degree of trauma are present. Certain features of subdural haematoma are unusual in accidental trauma and are suspicious of NAI. These are: • The presence of sudurals without a skull fracture, implying a shaking injury. • Bilateral subdurals (Figs 68.33, 68.34). • Subdurals of different ages (Fig. 68.35).The density of blood on CT and MRI varies with its age (Table 68.6). • Subdurals in the presence of retinal haemorrhages, implying an acceleration–deceleration force. • Acute interhemispheric fissure subdural or falx haemorrhage (Fig. 68.36). This is evident as a bright and irregularly thickened falx112,113. It must be distinguished from a normal falx which may appear bright against the background of an abnormally low density brain. In accidental trauma, subdural bleeds do not usually extend into the falx. Confusion may occur in differentiating chronic bilateral subdural haemorrhage from cases of unrelated cerebral atrophy or the more common benign enlargement of the extra-axial spaces. In neither of these should blood products be present and the changes are usually symmetrical.

Brain oedema and hypoxic ischaemic encephalopathy The ‘reversal sign’ describes a distinctive CT appearance in a group of children with hypoxic ischaemic cerebral injury and it carries a poor prognosis. It is highly associated with child

Figure 68.33 Bilateral subdural haematoma. (A) Axial T2-weighted image of an abused infant after drainage of bilateral subdural collections demonstrating bilateral frontal, parasagittal haemorrhagic contusions (arrows). Intraventricular blood is also noted in the posterior horn of the left lateral ventricle (arrowhead). (B) Axial T1-weighted image demonstrating bilateral iso-intense subdural haemorrhages and pneumocephalus. There is an extensive area of high signal accompanied by punctate areas of low signal consistent with laminar cortical necrosis demonstrated in the left hemisphere.

Figure 68.34 Bilateral subdural haematoma. (A) Axial unenhanced CT demonstrating a bilateral fresh high density subdural haematoma over both posterior parietal lobes (arrows) and both frontal lobes (arrowheads). (B) Axial unenhanced CT demonstrating bilateral fresh, high density subdural blood (arrows). On the right, there is associated infarction of the brain in the right middle cerebral artery territory associated with mass effect and contralateral midline shift.

1634

SECTION 8

•

PAEDIATRIC IMAGING

Figure 68.35 Subdural haematomas of differing ages. (A) Acute on chronic subdural haematoma. Axial unenhanced CT demonstrating subdural blood of differing ages in the left frontal and temporal regions (arrow). (B) Sagittal T1-weighted MRI slice demonstrating fresh iso-intense subdural haemorrhage over the frontal and parietal convexities (arrows). A small amount of hyperintense subdural blood is also noted in the occipital and posterior parietal region (arrowheads).

Table 68.6 CT AND MRI OF INTRACRANIAL HAEMORRHAGE: CHANGE IN DENSITY/SIGNAL WITH AGE Time

NECT density

CECT density

T1-weighted signal

T2-weighted signal

Gradient-echo signal

4–6 h

Hyper

—

Iso-intense

Hyper

Hypo

7–72 h

Hyper

—

Iso-intense

Hypo

Very hypo

4–7 d

Hyper/iso-intese

—

Iso-intense (centre)

Hypo

Very hypo

Hyper (rim) 1–4 weeks

Iso-intense/hypo

Rim enhancing

Hyper

Hyper

Hypo

Chronic

Hypo

—

Hypo

Hypo

Hypo

NECT = nonenhanced CT, CECT = contrast enhanced CT

abuse113. The CT features of the reversal sign are of diffusely decreased grey and white matter density with decreased or lost grey–white matter differentiation and a relative preservation in density of the thalami, basal ganglia and cerebellum (Fig. 68.37). The appearances are not unique to NAI and can result from a variety of brain insults, such as drowning, fits, status asthmaticus, cardiac arrest and accidental trauma. The evolution of the changes of hypoxic ischaemic encephalopathy is the same whatever the cause.The finding of acute interhemispheric fissure subdural haemorrhage together with the acute reversal sign should be regarded as highly suggestive of a shaking injury of NAI114. Complications of the brain injury include obstructive hydrocephalus from arachnoiditis, communicating hydrocephalus from alteration in the CSF dynamics, cerebral atrophy, cerebral arterial and venous infarction and multicystic encephalomalacia.

Investigations In children, severe intracranial injury can occur in the absence of skull fracture. Skull radiography is not a reliable predictor of intracranial injury. Skull radiography is recommended in children over 2 years of age to confirm a suspected depressed fracture or penetrating injury, or where NAI is suspected, but should be performed in all children under 2 years of age with suspected injury because of the greater probability of NAI in this younger age group66. Any obtunded or neurologically unstable child requires cross-sectional brain imaging. Any child suspected of abuse where a skull fracture is detected, even in the absence of neurological indication, should undergo neuroimaging to confirm or exclude intracranial injury. Multiplanar CT of the brain is the initial investigation of choice. An initial series of images should be obtained without intravenous contrast

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

Figure 68.36 Acute interhemispheric subdural haematoma. (A) Axial unenhanced CT demonstrating acute interhemispheric fissure subdural blood (arrow). Acute on chronic subdural haemorrhage is also present over the left frontal and parietal convexity. (B) Fresh subdural blood is seen tracking along the interhemispheric fissure anteriorly (arrow). (C) Axial unenhanced CT of a different abused baby demonstrates a thickened, bright interhemispheric fissure due to the presence of fresh subdural blood (arrow). The brain is of generalized low density due to associated oedema.

Figure 68.37 Reversal sign in hypoxic ischaemic encephalopathy. (A) Axial unenhanced CT of an abused baby who was shaken, demonstrating the ‘reversal sign’. The brain is oedematous with loss of grey–white matter differentiation and reduced density. There is relative preservation in density of the thalami and basal ganglia. (B) Axial unenhanced CT demonstrating an oedematous, swollen brain with reduced grey–white matter differentiation and mass effect. Small bilateral hyperdense subdural bleeds are present. A right posterior parietal skull vault fracture is present and is associated with an overlying scalp haematoma (arrow).

medium. If an isodense subdural is suspected, intravenous contrast medium may help in its identification. In neonates and small babies with a patent fontanelle, high resolution transfontanellar US will detect subdural collections115 and differentiate them from subarachnoid collections. However, US lacks sufficient parenchymal definition, it is

operator dependent and there are some relatively inaccessible areas of the brain. Thus US is not considered adequate as the sole investigation for the brain. MRI is the imaging method of choice in the stable child for overall full assessment of the intracranial structures, particularly cortical contusions, shearing injuries, small haemorrhages and

1635

1636

SECTION 8

•

PAEDIATRIC IMAGING

early ischaemia and for long-term follow-up. In the early stages, MRI may be very useful if the CT is negative when there is a strong clinical suspicion of abuse. MRI sequence protocols for brain injury may vary between departments with different machines and field strengths. Standard T1- and T2-weighted sequences are needed and modern systems can now provide fast T2-weighted images, with or without fat suppression. There is increasing use of perfusion and diffusion imaging and spectroscopy. Depending on age, inversion recovery techniques (e.g. FLAIR) may be helpful. In the younger patient MR angiographic (MRA) techniques may be contributory when an infarct is suspected.

Acknowledgements W Ramsden would like to thank Miss J Moran for typing the manuscript, and Professor H Carty, Dr M Elias, Dr J Foster and Mr M Hampshire for the loan of illustrations.

REFERENCES 1. Borden S IV 1975 Roentgen recognition of acute plastic bowing of the forearm in children. Am J Roentgenol 125: 524–530 2. Komara J S, Kottamasu L, Kottamasu S R 1986 Acute plastic bowing fractures in children. Ann Emerg Med 15: 585–588 3. Ohta-Fukushima M, Mutoh Y, Takasugi S, Iwata H, Ishii S 2002 Characteristics of stress fractures in young athletes under 20 years. J Sports Med Physical Fitness 42: 198–206 4. Carty H 1994 Sports injuries in children—a radiological viewpoint. Arch Dis Child 70: 457–460 5. Griffith J F, Roebuck D J, Cheng J C Y et al 2001 Acute elbow trauma in children: Spectrum of injury revealed by MR imaging not apparent on radiographs. Am J Roentgenol 176: 53–60 6. Caputo A E, Watson H K, Nissen C 1995 Scaphoid nonunion in a child: A case report. J Hand Surg 20A: 243–245 7. Johnson K J, Haigh S F, Symonds K E 2000 MRI in the management of scaphoid fractures in skeletally immature patients. Pediatr Radiol 30: 685–688 8. Torode I, Zieg D 1985 Pelvic fractures in children. J Pediatr Orthop 5: 76–84 9. Garvin K L, McCarthy R E, Barnes C L, Dodge B M 1990 Pediatric pelvic ring fractures. J Pediatr Orthop 10: 577–582 10. Rees M J, Aickin R, Kolbe A, Teele R L 2001 The screening pelvic radiograph in pediatric trauma. Pediatr Radiol 31: 497–500 11. Hamilton P R, Broughton N S 1998 Traumatic hip dislocation in childhood. J Pediatr Orthop 18: 691–694 12. Rubel I F, Kloen P, Potter H G, Helfet D L 2002 MRI assessment of the posterior acetabular wall fracture in traumatic dislocation of the hip in children. Pediatr Radiol 32: 435–439 13. Zionts L E 2002 Fractures around the knee in children. J Am Acad Orthop Surg 10: 345–355 14. Close B J, Strouse P J 2000 MR of physeal fractures of the adolescent knee. Pediatr Radiol 30: 756–762 15. Accousti W K, Willis R B 2003 Tibial eminence fractures. Orthop Clin North Am 34: 365–375 16. Swischuk L E, John S D, Tschoepe E J 1999 Upper tibial hyperextension fractures in infants: another occult toddler’s fracture. Pediatr Radiol 29: 6–9 17. Zobel M S, Borrello J A, Siegel M J, Stewart N R 1994 Pediatric knee MR imaging: Pattern of injuries in the immature skeleton. Radiology 190: 397–401 18. King S J, Carty H M L, Brady O 1996 Magnetic resonance imaging of knee injuries in children. Pediatr Radiol 26: 287–290 19. Strouse P J, Koujok K 2002 Magnetic resonance imaging of the pediatric knee. Top Magn Reson Imaging 13: 277–294

20. Nietosvaara Y, Aalto K, Kallio P E 1994 Acute patellar dislocation in children: Incidence and associated osteochondral fractures. J Pediatr Orthop 14: 513–515 21. Horn B D, Crisci K, Krug M, Pizzutillo P D, MacEwen G D 2001 Radiologic evaluation of juvenile tillaux fractures of the distal tibia. J Pediatr Orthop 21: 162–164 22. Inokuchi S, Usami N, Hiraishi E, Hashimoto T 1998 Calcaneal fractures in children. J Pediatr Orthop 18: 469–474 23. Ribbans W J, Natarajan R, Alavala S 2005 Pediatric foot fractures. Clin Orthop Rel Res 432: 107–115 24. Prokuski L J, Saltzman C L 1997 Challenging fractures of the foot and ankle. Radiol Clin North Am 35: 655–670 25. Shaw M, Burnett H, Wilson A, Chan O 1999 Pseudosubluxation of C2 on C3 in polytraumatized children—prevalence and significance. Clin Radiol 54: 377–380 26. Kathol M H 1997 Cervical spine trauma. What is new? Radiol Clin North Am 35: 507–532 27. Apple J S, Kirks D R, Merten D F, Martinez S 1987 Cervical spine fractures and dislocations in children. Pediatr Radiol 17: 45–49 28. Martin B W, Dykes E, Lecky F E 2004 Patterns and risks in spinal trauma. Arch Dis Child 89: 860–865 29. Reddy S P, Junewick J J, Backstrom J W 2003 Distribution of spinal fractures in children: does age, mechanism of injury, or gender play a significant role? Pediatr Radiol 33: 776–781 30. Glass R B J, Sivit C J, Sturm P F, Bulas D I, Eichelberger M R 1994 Lumbar spine injury in a pediatric population: Difficulties with computed tomographic diagnosis. J Trauma 37: 815–819 31. Silverman F N 1972 Unrecognised trauma in infants, the battered child syndrome and the syndrome of Ambroise Tardieu. Radiology 104: 337–353 32. American Academy of Pediatrics Committee on Child Abuse and Neglect 1993 Shaken baby syndrome; inflicted cerebral trauma. Pediatrics 92: 872–875 33. Carty H 1993 Case report: Child abuse—necklace calcification—a sign of strangulation. Br J Radiol 66:1186–1188 34. Kleinman P K 1996 Follow up skeletal survey in suspected child abuse. Am J Radiol 167: 893–896 35. Nimkin K, Spevak M, Kleinman P K 1997 Fractures of the hands and feet in child abuse: Imaging and pathological features. Radiology 203: 233–236 36. Mandelstam S A, Cook D, Fitzgerald M, Ditchfield M R 2003 Complementary use of radiological skeletal survey and bone scintigraphy in detection of bony injuries in suspected child abuse. Arch Dis Child 88: 387–390 37. Sty J R, Starshak R J 1983 The role of bone scintigraphy in the evaluation of the suspected abused child. Radiology 146: 369–375 38. Conway J J, Collium M, Tanz R R et al 1993 The role of scintigraphy in detecting child abuse. Semin Nucl Med 23: 321–333 39. McDowell H P, Fielding D W 1984 Traumatic perforation of the hypopharynx—an unusual form of abuse. Arch Dis Child 59: 888–889 40. Kleinman P K, Spevak M R, Hansen M 1992 Mediastinal pseudocyst caused by pharyngeal perforation during child abuse. Case report. Am J Radiol 158: 1111–1113 41. Ablin D S, Reinhart M A 1992 Oesophageal perforation by a tooth in child abuse. Pediatr Radiol 22: 339–341 42. Carty H 1989 Skeletal manifestations of child abuse. Bone 6: 3–7 43. Loder R T, Bookout C 1991 Fracture patterns in battered children. J Orthop Trauma 5: 428–433 44. Carty H 1993 Fractures caused by child abuse. Br J Bone Joint Surg 75: 849–857 45. Kleinman P K 1990 Diagnostic imaging in child abuse. Am J Radiol 155: 703 46. Kleinman P K, Marks S C, Blackbourne B 1986 The metaphyseal lesion in abused infants: a radiological histopathological study. Am J Roentgenol 146: 895–905 47. Kleinman P K, Marks S C 1996 A regional approach to the classic metaphyseal lesion in abused infants: The proximal tibia. Am J Radiol 166: 421–426

CHAPTER 68

•

PAEDIATRIC MUSCULOSKELETAL TRAUMA AND THE RADIOLOGY OF NONACCIDENTAL INJURY

48. Kleinman P K, Belanger P L, Karellas A et al 1991 Normal metaphyseal radiologic variants not to be confused with findings of infant abuse. Am J Roentgenol 156: 781–783 49. King J, Didendorf D, Apthorp J et al 1988 Analysis of 429 fractures in 189 battered children. J Pediatr Orthop 8: 585–589 50. Thomas S A, Rosenfield N S, Leventhal J M et al 1991 Long bone fractures in young children: distinguishing accidental injuries from child abuse. Pediatrics 88: 471 51. Beals R K, Tufts E 1983 Fractured femur in infancy: The role of child abuse. J Pediatr Orthop 3: 583 52. Schwend R M, Werth C, Johnston A 2000 Femur shaft fractures in toddlers and young children: rarely from child abuse. J Pediatr Orthop 20: 475–481 53. Caffey J 1946 Multiple fractures in the long bones of infants suffering from chronic subdural haematoma. Am J Radiol 56: 163–173 54. Chapman S 1990 Radiological aspects of non accidental injury. J R Soc Med 83: 67–71 55. Kleinman P K 1998 Skeletal trauma: general considerations. In: Kleinman P K (ed) Diagnostic imaging in child abuse. Williams and Wilkins, Baltimore, p8 56. Carty H 1995 Radiological features of child abuse. Curr Pediatr 5: 230–235 57. Kleinman P K 1998 Bony thoracic trauma. In: Kleinman P K (ed) Diagnostic imaging in child abuse. Williams and Wilkins, Baltimore, p110 58. Carty H, Ratcliffe J 1995 The shaken infant syndrome. BMJ 310: 344 59. Feldman K W, Brewer D K 1984 Child abuse, cardiopulmonary resuscitation and rib fractures. Pediatrics 73: 339 60. Spevak M R, Kleinman P K, Belanger P L et al 1990 Does cardiopulmonary resuscitation cause rib fractures in infants? Post mortem radiologic–pathologic study. Radiology 177: 162 (abstr) 61. Strouse P J, Owings C L 1995 Fractures of the first rib in child abuse. Radiology 197: 763–765 62. Carty H 1991 The non skeletal injuries of child abuse: The brain. Year Book of Pediatric Radiology, vol 3. Mosby, St Louis 63. Epstein J A, Epstein B S, Small M 1961 Subepicranial hydroma. J Paediatr 4: 562–566 64. Thornbry J R, Campbell J A, Masters S J et al 1983 Skull fracture and low risk of intracranial sequelae in minor head trauma. Am J Radiol 143: 661 65. Harwood-Nash D C, Hendrick E B, Hudson A R 1971 The significance of skull fractures in children. A study of 1187 patients. Radiology 101: 151 66. Duhaime A C, Alario A J, Lewander W J et al 1992 Head injury in very young children: mechanisms, injury types and ophthalmic findings in 100 hospitalised patients younger than two years of age. Paediatrics 90: 179–185 67. Musemeche C A, Barthel M, Cosentino C et al 1991 Paediatric falls from heights. J Trauma 31: 1347 68. Reiber G D 1993 Fatal falls in childhood. How far must children fall to sustain fatal head injury? Report of cases and review of the literature. Am J Forensic Med Pathol 14: 201 69. Chadwick D L, Chin S, Salerno C et al 1991 Deaths from falls in children. How far is fatal? J Trauma 31: 1353–1355 70. Kravitz H, Dreissen G, Gomberg R et al 1969 Accidental falls from elevated surfaces in infants from birth to one year of age. Paediatrics 44: 869–876 71. Cooper A, Floyd T, Barlow B et al 1988 Major blunt abdominal trauma due to child abuse. J Trauma 28: 1483–1487 72. Sivit C, Taylor L, Eichelberger M 1989 Visceral injury in battered children: A changing perspective. Radiology 173: 651–659 73. Filiatrault D, Garel L 1995 Paediatric blunt abdominal trauma—to sound or not to sound. Pediatr Radiol 25: 329–331 74. Turnock R R, Sprigg A, Lloyd D A 1983 CT in the management of blunt abdominal trauma in children. Br J Surg 80: 982–984 75. Sivit C J, Kaufman R A 1995 Commentary: Sonography in the evaluation of children following blunt abdominal trauma: is it to be or not to be? Pediatr Radiol 25: 326–328 76. Jamieson D H, Babyn P S, Pearl R 1996 Imaging gastrointestinal perforation in paediatric blunt abdominal trauma. Pediatr Radiol 26: 188–194

77. Alyono D, Morrow C E, Perry J F Jr 1982 Reappraisal of diagnostic peritoneal lavage criteria for operation in penetrating and blunt trauma. Surgery 92: 751–757 78. Davis J W, Hoyt D B 1990 Complications in evaluating abdominal trauma: DPL versus computerised axial tomography. J Trauma 30: 1506–1509 79. Noe H N, Jerkins G R 1992 Genitourinary trauma. In: Kelalis P P, King L R, Belman A B (eds) Clinical pediatric urology, 3rd edn. WB Saunders, Philadelphia pp1353–1378 80. Goldman S M, Sandler C M, Corriere J N Jr et al 1997 Blunt urethral trauma: A unified anatomical mechanical classification. J Urol 157: 85–89 81. Carty H 1991 The non skeletal injuries of child abuse. Part 2. The body. 1991 Year book of pediatric radiology, vol 3. Mosby, St Louis 82. Sivit C J, Eichelberger M, Taylor L 1992 Blunt pancreatic trauma in children: CT diagnosis. Am J Radiol 158: 1097–1100 83. Jeffrey R B, Federle M P, Grass R A 1983 Computed tomography of pancreatic trauma. Radiology 147: 491–494 84. King L R, Siegel M J, Balge D M 1995 Acute pancreatitis in children: CT findings of intra- and extrapancreatic fluid collections. Radiology 195: 196–200 85. Nimkin K, Teeger S, Wallach M T 1994 Adrenal hemorrhage in abuse children: Imaging and post mortem findings. Am J Radiol 162: 661–663 86. Burks D W, Mirvis S E, Shanmuganathan K 1994 Acute adrenal injury after blunt abdominal trauma: CT findings. Am J Radiol 162: 661–663 87. Cox T D, Kuhn J P 1996 CT scan of bowel trauma in the pediatric patient. Radiol Clin North Am 34: 807–818 88. Kleinman P K, Brill P W, Winchester P 1986 Resolving duodenal–jejunal hematoma in abused children. Radiology 160: 747–750 89. Fulcher A S, Das Narla L, Brewer W H 1990 Gastric hematoma and pneumatosis in child abuse. Case report. Am J Radiol 155: 1283–1284 90. Bratu M, Dower J C, Siegel B et al 1970 Jejunal hematoma, child abuse and Felson’s sign. Conn Med 31: 261–264 91. Taylor G, Fallat M, Eichelberger M 1987 Hypovolaemic shock in children: Abdominal CT manifestations. Radiology 164: 479 92. Hara H, Babyn P S, Bourgeois D 1989 Significance of bowel wall enhancement on CT following blunt abdominal trauma in childhood. J Comput Assist Tomogr 13: 430–432 93. Sivit C J, Eichelberger M R, Taylor G A 1994 CT in children with rupture of the bowel caused by blunt trauma: Diagnostic efficacy and comparison with hypoperfusion complex. Am J Radiol 163: 1195–1198 94. Rizzo M J, Federle M P, Griffiths B G 1992 Bowel and mesenteric injury following blunt trauma: Diagnosis with CT. Am J Radiol 159: 1217–1221 95. Shah P, Applegate K E, Buonomo C 1997 Stricture of the duodenum and jejunum in an abused child. Pediatr Radiol 27: 281–283 96. Brunsting L, Morton J 1987 Gastric rupture from blunt abdominal trauma. J Trauma 27: 887 97. Lloyd D A, Carty H, Patterson M et al 1997 Predictive value of skull radiography for intracranial injury in children with blunt head injury. Lancet 349: 821–824 98. Ludwig S, Warman M 1984 Shaken baby syndrome. A review of twenty cases. Ann Emerg Med 13: 104 99. Brown J K, Minns R A 1993 Non accidental head injury with particular reference to whiplash shaking and medicolegal aspects. Dev Med Child Neurol 35: 849–869 100. Kleinman P K, Barnes P D 1998 Head trauma. In: Kleinman P (ed) Diagnostic imaging of child abuse, 2nd edn. Mosby, St Louis, pp 285–342 101. Caffey J 1972 On the theory and practice of shaking infants. Am J Dis Child 124: 161–169 102. Caffey J 1974 The whiplash shaken baby syndrome: manual shaking by the extremities with whiplash induced intracranial and intraocular bleedings linked with residual permanent brain damage and mental retardation. Paediatrics 54: 396–403 103. Ommaya A K, Faas F, Yamell P 1968 Whiplash injury and brain damage. An experimental study. JAMA 204: 75–79

1637

1638

SECTION 8

•

PAEDIATRIC IMAGING

104. Dykes L 1986 The whiplash shaken infant syndrome. What has been learned? Child Abuse Neglect 10: 211–221 105. Newton R W 1989 Intracranial haemorrhage and non accidental injury. Arch Dis Child 64: 188–190 106. Bird C R, McMahan J R, Gilles F H et al 1987 Strangulation in child abuse: CT diagnosis. Radiology 163: 373–375 107. Krugman R D 1985 Fatal child abuse: Analysis of 24 cases. Paediatrician 12: 68 108. St James-Roberts I 1991 Persistant infant crying. Arch Dis Child 66: 653 109. Guthkelch A N 1971 Subdural haematoma and its relationship to whiplash injuries. BMJ 2: 430–431 110. Boulis Z F, Barnes N R 1978 Head injuries in children—aetiology, symptoms, physical findings and X-ray wastage. Br J Radiol 51: 851–854 111. Bell W O 1995 Paediatric head trauma. In: Arensman R M (ed) Paediatric trauma. Raven Press, New York,101–118 112. Zimmerman R A, Bilaniuk L T, Bruce D 1978 Interhemispheric acute subdural haematoma: a CT manifestation of child abuse by shaking. Neuroradiology 16: 39–40 113. Han K B, Towbin R B, De Courten-Myers G et al 1989 Reversal sign on CT: Effect of anoxic/ischaemic cerebral injury in children. Am J Neuroradiol 10: 1191–1198

114. Rao P, Carty H, Pierce A 1999 The acute reversal sign: Comparison of medical and non-accidental injury patients. Clin Radiol 54: 495–501 115. Lam A H, Cruz G B 1991 Ultrasound evaluation of subdural haematoma. Australas Radiol 35: 330

SUGGESTIONS FOR FURTHER READING Carty H (ed) 1999 Medical radiology: Emergency pediatric radiology. Springer-Verlag, Berlin Kuhn J P, Slovis T L, Haller J O (eds) 2003 Caffey’s pediatric diagnostic imaging, 10th edn. Mosby, Philadelphia, pp 2269–2303 Rodriguez–Merchan E C, Radomisli T E (eds) 2005 Symposium: Pediatric skeletal trauma. Clin Orthop Rel Res 432: 1–131 Rogers L F 2001 Radiology of skeletal trauma, 3rd edn. Churchill Livingstone, New York, pp 111–144 Kleinman P K 1998 Diagnostic imaging of child abuse, 2nd edn. St Louis, Mosby

CHAPTER

Bone Tumours and Neuroblastoma in Children

69

Paolo Toma and Claudio Defilippi Bone tumours • Benign bone tumours • Malignant bone tumours Neuroblastoma

This chapter should be read in conjunction with the descriptions of bone tumours in adults (see Chs 47, 48).

BONE TUMOURS The most common symptom of bone neoplasms is skeletal pain and plain radiography remains the first diagnostic step. Further investigation by computed tomography (CT), magnetic resonance imaging (MRI) and radionuclide radiology may be required, depending on initial interpretation. Due to the relative rarity of bone malignancies in the paediatric age group, the diagnosis is not always appreciated by general radiologists and, in atypical cases, may be delayed due to misinterpretation. However, plain radiographic diagnosis may be impossible during asymptomatic early tumour growth and can remain extremely difficult, even when the tumour is more advanced. Unusual clinical presentation, suspected

metastatic spread from an unknown primary focus, adulttype neoplasms, chemo- and radio-therapy-induced malignancies and tumours associated with genetic syndromes all lead to difficulties. Diagnostic imaging, though playing a crucial role, must be closely coordinated with clinical assessment and histology. It must be remembered that biopsy specimens are small and will only sample a portion of a lesion; thus histopathological diagnosis can be impossible without clinical and radiological data (Table 69.1)1–7. It is increasingly important that all potentially malignant lesions in children are referred to a centre with the necessary expertise and multidisciplinary teams so that discussion about potential biopsy

Table 69.1 ADVANTAGES AND LIMITATIONS OF DIFFERENT IMAGING TECHNIQUES IN EVALUATING MUSCULOSKELETAL NEOPLASMS (modified from ref. 2) Characteristic

Radiography

Radionuclide radiology

CT

Contrastenhanced CT

MRI

Cortical bone

+++

+

+++

+++

++

Bone marrow

−

+

+

+

+++

Soft tissue

+

+

++

++

+++

Intramedullary extent

+

+

++

++

+++

Extramedullary extent

+

+

++

+++

+++

Relationships to neurovascular bundle

−

−

+

++

+++

Calcifications

++

−

+++

+++

++

Skip lesions

+

++

++

++

+++

Differential diagnosis: benign vs malignant

+++

+

+

+

++

Monitoring response to therapy

+

++

++

++

+++

Tumour characterization

+++

+

++

++

++

1640

SECTION 8

•

PAEDIATRIC IMAGING

(and the route of any biopsy) is discussed in advance with the surgeon and oncologists who will be responsible for future care. For example, a biopsy should normally be obtained through tissue planes related to subsequent surgery to avoid the risk of seeding. The aim of initial assessment is to differentiate aggressive from chronic disease and to distinguish benign from malignant lesions.

BENIGN BONE TUMOURS Bone-forming tumours The bone-forming tumours are: • osteoma • enostosis • osteoid osteoma • osteoblastoma.

Osteoid osteoma This is a painful lytic lesion (nidus) with a sclerotic host reaction. Osteoid osteoma occurs in children and adolescents but more than 80% of cases occur in the second decade of life8. A typical patient presents with pain, often worse at night, relieved by aspirin but not anti-inflammatory drugs. Typically the lesions occur in the tubular bones but they may also affect the vertebral appendages. In the tubular bones the lesions are cortically based. The dense sclerotic reaction can mask the nidus on plain radiographs. Intramedullary intracapsular lesions, usually located in the proximal femur at the medial aspect of the femoral neck, may cause growth deformities and overgrowth with limb length discrepancy, muscle atrophy and associated early osteoarthritis. Subperiosteal osteoid osteoma is less common, occurring in the tubular bones and talus9. Typically there is scalloping and irregular bone resorption. Another location is in the posterior elements of the spine, presenting clinically with painful scoliosis. These are seen as dense pedicles or are identified following positive scintigraphy. Very rarely, osteoid osteoma may be polyostotic. The diagnosis (Figs 69.1, 69.2) is based on the plain radiograph, supplemented by CT with direct visualization of the nidus and the surrounding sclerotic reaction. Sometimes, calcified foci may be seen inside the nidus. CT may also reveal surrounding oedema of the soft tissues. MRI can cause confusion; if inflammatory changes are seen near the lesion, the diagnosis may be mistaken for a more aggressive tumour or even infection. The MRI findings are extremely variable and generally non-specific. The nidus has low to intermediate signal intensity on T1-weighted images and low to high signal intensity on T2-weighted images. It may enhance with intravenous paramagnetic contrast medium. Calcification has low signal intensity on all pulse sequences. The nidus may be masked by signal changes due to associated sclerosis, bone marrow oedema and soft tissue inflammation. Some authors5 report a typical tumour pattern on threephase skeletal scintigrams characterized by focal hypervas-

Figure 69.1 Osteoid osteoma. (A) Intracortical osteoid osteoma of the left fibular shaft: typical oval radiolucent area, representing the nidus, with calcification inside and reactive bone sclerosis. (B) Intracortical osteoid osteoma of the left femoral shaft without calcification in the radiolucent area and perifocal osteosclerosis.

Figure 69.2 Osteoid osteoma. Intramedullary pattern. (A) Conventional radiograph shows a minimal peri-ostosis and reactive bone sclerosis (arrow). (B) CT demonstrates the nidus.

cularity and high uptake4. A negative scintigram excludes the presence of osteoid osteoma. Scintigraphy can help to locate the lesion, especially in radiographically difficult areas such as the spine, and to confirm recurrence or incomplete removal if pain persists following surgery. Surgery is the therapy of choice but if the nidus is excised incompletely, there is a risk of recurrence (filming of the specimen to evaluate the complete excision of the nidus is useful). Percutaneous CT-guided excision has been recently suggested with good results and rapid recovery.

Osteoblastoma Osteoblastoma occurs mainly in children and adolescents. It is a rare benign tumour (0.5% of bone tumour biopsies) that

CHAPTER 69

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

is difficult to differentiate histologically from osteoid osteoma while, radiographically, it is quite distinct from osteoid osteoma10,11. Usually monostotic, the main site is the bony spongiosa of the posterior elements of the spine (Fig. 69.3); frequently the location can be either metaphyseal or diaphyseal in the long bones. Osteoblastomas are larger than osteoid osteomas and can reach 2–10 cm in diameter. Radiographically they may be lucent or sclerotic but always have a sclerotic rim, and tend to cause expansion without a large amount of reactive sclerosis.The shape can be fusiform and a real ‘nidus’ is usually not appreciable. Described as nonaggressive, occasionally the cortex may be broken10. Differential diagnosis includes osteosarcoma, if the appearance is aggressive and osteoid osteoma. Usually plain radiography, scintigraphy and CT are more than sufficient for the diagnosis. The MRI appearance is generally non-specific, with a low to intermediate signal intensity on T1-weighted images and an intermediate to high signal intensity on T2-weighted images. MRI elegantly shows the effects on the spinal canal and cord, as well as the intra- and extra-osseous reactive changes and the possible infiltration of adjacent soft tissue12.

pathological fracture. Radiographically these fibrous lesions appear as an osteolytic cystic defect in the metaphysis of a long bone (Fig. 69.4), usually at the site of tendon attachment to the cortical surface. It is well delineated, often multilocular, and has marginal sclerosis1,13. The most frequent sites are the proximal and distal tibia and proximal femur. Eighty-two per cent of all NOFs are detected in children and they usually undergo a spontaneous resolution later in life. Although MRI is not routinely indicated for simple cases of NOF,T1-weighted MRI shows low signal intensity within the lesion compared with the skeletal muscle. On T2-weighted images, the lesion can have variable signal intensity. The lesion usually enhances avidly14. A metaphyseal fibrous cortical defect, which is not a true tumour, grows exclusively in the metaphyseal regions in children and adolescents, most frequently in the tibia and femur. It is a harmless proliferation of fibrous tissue in the periosteum which appears as a cortical bony defect. On plain radiographs, there is a cortical defect separated from the inner bone by a sclerotic rim. On MRI the marginal defect presents as a hypo-intense signal on both T1- and T2weighted sequences15.

Tumours of fibrous tissue origin

Cartilage-forming tumours

The tumours of fibrous tissue origin are: • ossifying bone fibroma (OBF) • nonossifying fibroma (NOF) • benign fibrous histiocytoma • metaphyseal fibrous cortical defect • fibrous bone dysplasia (Jaffe Lichtenstein) (FBD) • osteofibrous bone dysplasia (Campanacci) (OBD)

The cartilage-forming tumours are: • chondroma (enchondroma) (Fig. 69.5) • osteochondroma: • chondroblastoma.

Osteochondroma (exostosis) An osteochondroma is a very common benign neoplasm that results from growth plate cartilage being displaced to the

Nonossifying fibroma and metaphyseal fibrous cortical defects Although histologically different, these two lesions appear radiologically similar and are conventionally treated as one process. NOF is the most common benign tumour in children and is typically encountered during childhood and adolescence. Most lesions are discovered incidentally, often because of a

Figure 69.3 Osteoblastoma. (A) C6 posterior arch is markedly expanded by a large lucent lesion with a sclerotic rim (arrow). (B) CT confirms the lytic nature of the lesion, with lobulated regular margins and internal septa.

Figure 69.4 Nonossifying fibroma with a ‘soap bubble’ appearance eccentrically located in the right distal tibia displaying a characteristic scalloped border.

1641

1642

SECTION 8

•

PAEDIATRIC IMAGING

Figure 69.5 Enchondromatosis. Multiple enchondromas in the right distal forearm.

metaphyseal region16. The underlying bone is normal. The exostosis is covered by a cartilaginous cap which is the source of the growth. Osteochondromas may be flat with a broad base or sessile, referred to as a cauliflower exostosis. Clinically they feel larger than shown on the radiographs due to the unossified cartilage cap. Osteochondromas may grow until skeletal maturity. Growth after maturity suggests malignant degeneration which occurs in 1% of solitary exostoses. Osteochondromas are usually seen as solitary lesions in the first two decades and in a monostotic form in a metaphyseal location adjacent to joints, 36% near the knee and 35% in the extremities. The base is formed of normal bone which has a cartilaginous cap, which in turn can be covered by a bursa due to chronic irritation of the overlying soft tissues which may be displaced. The polyostotic form, called multiple hereditary exostosis or diaphyseal aclasis, is an autosomal dominant disorder. Most osteochondromas are broad based and involve a large portion of the circumference of the metaphysis with a resultant shortening of the limbs and deformity. Plain radiography is diagnostic in the vast majority of cases. On MRI T2-weighted sequences the hyaline cartilage appears hyperintense with a good visualization of its structural layers. MRI demonstrates the cartilaginous lobulated matrix and the thickness of the cartilaginous cap of the tumour (Fig. 69.6). In children and adolescents, the cap may be as thick as 3 cm. Malignant degeneration is very rare in children.

Chondroblastoma This is a rare benign cartilaginous tumour found almost exclusively in the epiphysis and which tends to be seen before epiphyseal closure. It is a monostotic lesion with a cartilaginous matrix, calcification in 50% and no soft tissue involvement.

Figure 69.6 Osteochondroma. (A) ‘Cauliflower’ growth of a chondroma arising from the greater tuberosity of the humerus. (B) MRI (turbo T2-weighted sequence with fat saturation) shows the bright hyperintensity of the cartilaginous cap.

It is normally located in the proximal femur or proximal tibia, is usually lobulated with a sclerotic rim and can be associated with host periosteal reaction. The diagnosis may be suggested on plain radiography by identifying a well-defined radiolucent lesion in a characteristic position. CT is useful to identify any calcified matrix. MRI appearances are non-specific with low to intermediate heterogeneous signal on T2 with a lobulated pattern.The presence of high signal on T2-weighted images depends on noncalcified chondroid matrix. Bone marrow oedema may be present and, if the location is juxtacortical, an intra-articular fluid effusion may be appreciable.

Vascular and other connective tissue tumours These tumours are: • bone haemangioma • bone lymphangioma • massive osteolysis • myofibromatosis • lipoma of bone.

CHAPTER 69

Myofibromatosis

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

Myofibromatosis presents most often in infancy, but can occur even in adults. The solitary or multiple lesions involve skin, subcutaneous tissues, muscles, bones and viscera (particularly lung, heart, gastrointestinal tract and dura). Multicentric visceral involvement, especially of the lung, is considered a poor prognostic indicator associated with early death. The skeletal lesions appear lytic and sharply defined. Sclerotic margins appear later17.

and osteolysis. The lack of clear margins indicates malignancy. MRI identifies the characteristic vascular channels, often seen in the periphery of the mass. The mass is hypervascularized, with a serpentine vascular structure which may have flow voids due to high velocity flow or T2-weighted hyperintensity because of slow flow. The main mass has an intermediate T1-weighted signal and a high T2-weighted signal. Due to its aggressive nature, there is a lack of fatty tissue, in contrast to benign haemangiomas.

Locally aggressive tumours

Tumour-like lesions

These tumours are: • desmoplastic bone fibroma • chondromyxoid fibroma • haemangiopericytoma • osteoclastoma (giant cell tumour).

The tumour-like lesions are: • simple juvenile bone cyst • aneurysmal bone cyst (ABC) • benign fibrous cortical defect: see fibrous tissue tumours • fibrous bone dysplasia (Jaffe Lichtenstein) (FBD) • osteofibrous bone dysplasia (Kempson, Campanacci) (OBD) • eosinophilic granuloma (Langerhans cell histiocytosis).

Chondromyxoid fibroma This is a rare semimalignant bone tumour which accounts for only 0.5% of all bone tumours. It is mainly seen in the second decade of life but cases have been described in the first decade. Typically, chondromyxoid fibromas are located in the metaphyseal region of the long tubular bones, near the growth plate, and can cross into the epiphysis. The lesion is well circumscribed, with an eccentric osteolytic lobulated aspect that bulges outward, thinning the cortex (Fig. 69.7). It is usually benign in character but may cause local invasion with a high recurrence rate if incompletely removed.

Simple bone cyst

This extremely rare soft tissue tumour may invade bone. It occurs in every age group, including children. The tumour arises from pericytes (Zimmerman cells) and grows aggressively with local bone invasion. On a plain radiograph, osteolytic bone destruction is often mixed with sclerosis

Cystic bone lesions are relatively common in children and adults. A juvenile bone cyst or solitary or unicameral cyst is a benign lesion that develops in the centre of the metaphysis, usually of the femur or humerus near to the epiphyseal plate (Fig. 69.8). It expands the bone, thinning the cortex. With maturity, the cyst migrates down the shaft of the bone. It has smoot h margins and, if there has been haemorrhage into the cyst, is filled with clear or sanguinous fluid. The MRI signal depends on the content: usually the signal is intermediate on T1-weighted images and high on T2. It is frequently discovered by chance during incidental radiographs for other purposes. Two-thirds of patients present with a pathological fracture (Fig. 69.8) and, in other symptomatic patients, this pain is thought to be due to microfractures.

Figure 69.7 Chondromixoid fibroma of the distal femoral metaphysis. (A,B) Large osteolytic defect bulging outward.

Figure 69.8 Solitary bone cyst. Large radiolucent area of the right humerus with pathological fracture, and a ‘fallen’ fragment.

Haemangiopericytoma

1643

1644

SECTION 8

•

PAEDIATRIC IMAGING

Aneurysmal bone cyst This is a benign, solitary expansile lesion which occurs in the vertebral appendages, the flat bones and most frequently the metaphysis of the long bones in the femur, tibia and humerus, but may also be found in the tubular bones of the hands and feet. The lesion develops eccentrically without crossing through the growth plate. It consists of a cyst, usually filled with blood. The lesion is frequently multilocular with poorly defined margins. The plain radiographic appearance is usually characteristic in the long bones. In the vertebral column, a destructive expansile lesion is seen in the appendages and may encroach on the spinal canal. Full evaluation of ABCs requires cross-sectional imaging. CT demonstrates the bone destruction and is essential before sclerotherapy. Coronal T2weighted images, with or without fat saturation, are advantageous for visualization of the main cystic lesion, any additional cysts and fluid–fluid levels within the lesion (Fig. 69.9). Contrast-enhanced T1-weighted sequences show enhancement of cyst walls and internal septations18,19. Follow-up studies monitor resolution of these cysts following treatment. In selected cases percutaneous sclerotherapy under fluoroscopy guidance has been used with successful results.

Fibrous bone dysplasia (Jaffe Lichtenstein) This is a malformation which may be mistaken for a bone tumour. FBD is a common osteolytic bone lesion in which the bone is centrally replaced by fibrous tissue20. Local bending or pathological fractures may occur. This is discussed more fully in the skeletal dysplasia section below. Problems with diagnosis mainly arise with the monostotic forms, which on plain radiography show widening of the medullary cavity, bony expansion and a ground-glass appearance (Fig. 69.10). A rare condition, the McCune Albright syndrome, consists of unilateral polyostotic FBD associated with precocious puberty. The most frequent locations for FBD are the ribs, proximal femur, skull, scapula and pelvis.

Osteofibrous bone dysplasia (Kempson, Campanacci) This occurs in the shaft of long tubular bones. It is a multicystic lesion surrounded by a sclerotic bone, with intact cortex and no periosteal reaction. It may be multifocal. On MRI the appearances are those of all fibrous tumours. The plain radiographs are diagnostic (Fig. 69.11). This lesion is often present at birth.

MALIGNANT BONE TUMOURS Osteosarcoma Osteogenic sarcoma (OS) is the most common primary malignant tumour of bone in children and accounts for 15% of all primary bone tumours1. Radiologically OS can be diagnosed when osteoid tissue or bone tissue formation produced by tumour cells is associated with aggressive characteristics21. The aetiology is unknown.The only environmental agent known to cause this tumour is ionizing radiation but radiation-induced OS only accounts for approximately 3% of all OS. The peak

Figure 69.9 Aneurysmal bone cyst. (A) Large well-defined cystic lesion of the left part of the sacrum (arrows), showing the ‘blow out’ pattern, with septa and bridges. (B) MRI: T2-weighted sequence, axial view. Large hyperintense mass, ‘soap bubble’-shaped lesion arising from the right ischiopubic region. Fluid–fluid level (arrows) and hypo-intense septa (arrowheads) are evident.

incidence of OS occurs in the second decade of life, a characteristic that suggests a relationship between rapid skeletal growth and the development of this neoplasm. Furthermore, OS occurs most frequently at sites where the greatest increase in length and size of bone occurs: the metaphyseal portions of the most rapidly growing bones (and parts of bones) in adolescents. Very often, patients have a history of trauma, which brings the problem to clinical attention. The radiographic appearances of OS are variable but the most frequent is that of a destructive lesion in the metaphysis

CHAPTER 69

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

Figure 69.10 Monostotic fibrous dysplasia. (A) Initial ‘candle flame’ aspect (arrow). (B) Cystic-like appearance (arrowhead).

of a long bone, poorly demarcated margins, a Codman's triangle, aggressive periosteal reaction and a soft tissue mass22,23. The lesion may cross the growth plate (Fig. 69.12). CT and MRI provide the required information for staging the disease and monitoring the treatment. On MRI, tumour volume measurement and necrosis grading after chemotherapy are the main prognostic factors in monitoring OS, reflecting reduction in vascularity, peritumoural oedema and neoplastic cell mass. Because of sensitivity to oedema, MRI can overestimate tumour size if performed after biopsy. MRI is superior to CT for detecting marrow extension, relationships with vessels,

Figure 69.11 Osteofibrous dysplasia of the tibial shaft. Multicystic lesion surrounded by sclerotic bone.

Figure 69.12 Osteosarcoma. Mixed sclerotic and lytic lesion of the right humerus, extending into the soft tissues and sparing the growth plate. A Codman triangle is visible at the upper limit of the tumour.

transphyseal spread of tumour, soft tissue component and skip metastases because of its ability to image the tumour in axial, coronal and sagittal planes with high contrast resolution (Fig. 69.13). Thus MRI is the cross-sectional imaging method of choice. Furthermore, it can monitor the response to chemoor radiotherapy demonstrable as changes in the intensity of the signal, in both T2- and enhanced T1-weighted sequences.

Figure 69.13 Osteosarcoma. (A) T1- and (B) T2-weighted MRI sequences of osteosarcoma in the distal femur with transphyseal extension and a small soft tissue component laterally.

1645

1646

SECTION 8

•

PAEDIATRIC IMAGING

Due to the cartilage cap, the rare variant osteochondroma-like parosteal osteosarcoma can be misdiagnosed as a benign lesion at MRI. MRI staging should include T1-weighted sequences following administration of intravenous contrast medium, to differentiate between tumour and oedema. Initial staging also requires CT of the lungs. Scintigraphy is required to look for skeletal metastases. Lung secondaries may also take up the bone-seeking agents. Successful treatment with chemotherapy leads to a decrease in tumour volume and increased signal intensity on T2-weighted images, reflecting both tumour necrosis and oedema. Fast dynamic contrast-enhanced MRI is useful to seek recurrence of the tumour, which shows early enhancement as opposed to slow enhancement of inflammatory tissue. Following treatment, the signal intensity on T2-weighted images decreases due to the replacement of tumour cells by fibrous tissue.

Ewing's sarcoma This tumour was originally described in 1921 by James Ewing as a radiosensitive bone neoplasm, of endothelial origin, growing in the shaft of the long bones (Fig. 69.14). It is generally accepted that the tumour is of neural origin. Ewing's sarcomas (ES) are now accepted as a group of malignant tumours, arising both from bone and extraskeletally, mostly presenting as undifferentiated neoplasms but including more differentiated forms like peripheral primitive neuro-ectodermal tumour (PNET)24. The term PNET includes soft tissue and bone neoplasms previously reported as peripheral neuro-epithelioma, adult neuroblastoma, small cell tumour of the chest wall (Askin tumour) and others. Extra-osseous ES, extra-osseous PNET and bone PNET account for only 13% of all these malignancies. The aetiology and pathogenesis of ES and its variants are unknown.Association with congenital diseases is rare. ES occurs more commonly in adolescents but a second peak of incidence in very young children is well known. It is the second most common bone tumour in children and adolescents, accounting for about 10% overall, with an incidence of 0.6 per million. It is sometimes difficult at presentation to determine whether the tumour is of osseous or extra-osseous origin because of

the large soft tissue component of bone tumours and the osseous involvement in soft tissue malignancies. All bones may be involved, but most of the tumours develop in the long bones and pelvis. Children with pelvic and spinal tumours often have symptoms for several months before a diagnosis is made on plain radiographs. In the pelvis the tumour arises from the iliac bone and a large soft tissue mass is usually present. This, combined with reactive bone, often causes a sclerotic appearance of the iliac blade. Signs of bone destruction are subtle initially and often overlooked. Scintigraphy is indicated in occult bone pain in children and leads to earlier diagnosis. The clinical presentation is non-specific: local pain and swelling can mimic an acute osteomyelitis, delaying the diagnosis. Fever and increased local temperature are often present. Children with a chest wall tumour usually present with pain. The destructive rib lesion is often initially overlooked and by the time of diagnosis there is usually a pleural effusion present. Spread is haematogenous and metastases are found in lungs, bones and bone marrow. Lymph node, liver and skip metastases are rare. The differential diagnosis includes osteomyelitis, other lytic bone lesions, osteosarcoma and metastases from another primary tumour.The routine imaging algorithm of these patients includes plain radiography and MRI of the primary lesion, bone scintigraphy of the skeleton looking for metastatic bone spread and plain radiography and CT for chest evaluation of metastases. The radiographic features include both a poorly defined lytic appearance and sclerotic opacities, often accompanied by severe periosteal reaction (Fig. 69.14).The lytic pattern reflects the severe bone destruction by the neoplasm, while the sclerotic appearance should not be interpreted as tumour bone production, as documented in osteosarcoma, but is a secondary reactive phenomenon. The extent of the disease can be defined both by CT and MRI. On MR images, the presence of high signal intensity in T2 is generally associated with tumour viability or recurrence, while the absence of high T2-weighted signal intensity has a good prognostic value. When a large soft-tissue mass is present it can still be difficult to differentiate between tumour and infection, even with Gadolinium enhancement. The diagnosis is made by biopsy. Clinical management of Ewing's sarcoma is with initial chemotherapy and then resection.

Bone metastases

Figure 69.14 Ewing's sarcoma. Mixed lytic and sclerotic lesion of the third metatarsal bone.

Bone metastases, the most common form of malignant bone tumour in adults, are not common in children. Metastases as the first manifestation of a primary tumour are unusual as the primary is usually evident. Bone and bone marrow are usually involved via a haematogenous pathway.The red marrow is more frequently involved than the yellow marrow and thus long bones are common metastatic sites. Metastatic bone disease in children is most frequently due to leukaemia and neuroblastoma (Fig. 69.15), but lymphoma, ES and OS also metastasize to bone. Bone metastases are also described in rhabdomyosarcoma and medulloblastoma.

CHAPTER 69

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

Figure 69.15 Abdominal neuroblastoma. Contrast-enhanced CT. (A) A prevertebral tumour extends across the midline in the retroperitoneum, displaces and encases the aorta (a) anteriorly and encases the renal arteries (ra). The inferior vena cava (ivc), partially encased, is displaced anterolaterally and compressed by the tumour and nodes. The mass extends to both the renal hili. (B) Infiltrative lesion to the right iliac bone with a large soft tissue component (arrow) that projects inward as a space-occupying mass.

Periosteal reaction and a lytic appearance of lesions are the most common radiographic characteristics in metastatic locations but sclerotic lesions are also reported, particularly in medulloblastoma and leukaemia. Bone changes due to metastatic spread may also include osteoporosis, radiolucent lines, cortical disruption, periosteal thickening and ‘onion skin’ periosteal reaction. Metaphyseal radiolucent lines are typically described in leukaemia and, less commonly, neuroblastoma. A radionuclide skeletal survey confirms secondary bone involve-

ment but MRI is also excellent because of its great sensitivity. All procedures may have to be followed, when clinically indicated, by marrow biopsy.

Rare malignant bone tumours in children The rare malignant tumours are: • chondrosarcoma • primary malignant lymphoma of bone • haemangiosarcoma of bone.

NEUROBLASTOMA Neuroblastoma (NB) is a malignant tumour arising from primordial neural crest cells. These cells migrate during embryogenesis to form the adrenal medulla and the sympathetic ganglia. NB occurs, therefore, in diverse but predictable locations such as the neck, thorax, abdomen and pelvis, arising from the adrenal glands or anywhere along the sympathetic chain. NB arises in the retroperitoneum in 65% of cases and, of these, 40% of lesions occur in the adrenal glands. It is the most common extracranial solid malignancy in children and the third most common paediatric malignancy (the first is central nervous system tumours and the second leukaemia). Overall, NB represents 8–10% of all childhood cancers. Among abdominal neoplasms, NB is the second most common cancer after Wilms’ tumour. The median age at diagnosis is 2 years, with 90% of diagnoses made in children younger than 5 years of age. NB is also seen in the neonatal period and may be identified in the fetus by antenatal imaging25. It may also be present very rarely in adolescents and adults. Three types of neural crest tumours are known, classified according to cellular differentiation. NB is the more common and is characterized by the most primitive and malignant cells, ganglioneuroma, which is completely differentiated and

benign, and ganglioneuroblastoma, which presents intermediate features between NB and ganglioneuroma.

Clinical issues The signs and symptoms of NB reflect the tumour site and extent of disease. The tumours may manifest as an incidental mass, or may cause abdominal pain. Because metastases are so frequently present (skeleton, bone marrow, lymph nodes, liver, and, rarely, lung and brain), the clinical symptoms are often due to metastatic disease. Children may have bone and joint pain, proptosis from orbital metastases, anaemia, weight loss and fever. Horner's syndrome may be the presenting feature of cervical or thoracic NB involving the stellate ganglion. Children may also present with the effects of production of hormones such as catecholamines (hypertension) and vasoactive intestinal polypeptyde (VIP) (intractable watery diarrhoea). Another paraneoplastic syndrome associated with NB is myoclonic encephalopathy of infancy (MEI). The precise pathogenesis of MEI is unknown. NB in infants younger than 1 year of age is often associated with extensive hepatic metastases. Hepatomegaly may be massive despite a small, sometimes not evident, primary

1647

1648

SECTION 8

•

PAEDIATRIC IMAGING

lesion. These children usually have bone marrow lesions and palpable subcutaneous nodules as well (Stage IV-S). This type of NB has a better prognosis and has a tendency to spontaneous regression: the survival probability is about 79%. The very young IV-S patients are at high risk because of severe respiratory complications as a result of massive hepatomegaly. Extradural extension (dumb-bell syndrome) is common with thoracic NB, but rare in abdominal tumours. Approximately 95% of patients with NB have elevated levels of catecholamines [vanillylmandelic acid, homovanilic acid, norepinephrine (noradrenaline), dopamine] in the urine. A combination of a positive bone marrow aspirate and an increase of urinary catecholamine metabolites is sufficient to confirm the diagnosis.

soft tissue mass with calcification in as many as 53% of patients (diffuse, mottled, finely stippled, or coalescent calcifications). Erosion of pedicles is suggestive of intraspinal extension. Radiographs of the long bones in patients with bone metastases may be normal or may show ill-defined areas of bone destruction. A solitary lesion may appear as a lytic, moth-eaten or permeating destructive area, interspersed with sclerotic trabeculae. New periosteal bone formation, often parallel to the shaft, may be present. The radiographic features do indicate malignant growth but are not specific. The most common skeletal sites are the skull and metaphyses of long bones (humerus and femur), but they can be seen in any bone except for the hands, feet and clavicles (see Ch. 48).

Ultrasound

Staging The treatment of NB is determined by the stage of the tumour at the time of diagnosis26.The international NB staging system is based on radiological findings, surgical resectability, lymph node involvement and bone marrow involvement (Table 69.2)27. Several biological markers may also affect prognosis: serum ferritin and serum neuron-specific enolase levels, MYCN oncogene amplification, chromosome 1p deletion (loss of heterozygosity), TRK expression and histopathology (Shimada classification).

Imaging The current evaluation of primary sites (staging and followup) of patients with NB consists of CT or MRI of the primary tumour and meta-iodobenzylguanidine (MIBG) scintigraphy. The evaluation of metastatic disease is based on MRI or CT of the head, orbit and neck; MRI or CT of the abdomen and pelvis; CT of the chest; skeletal scintigraphy with radiographs of abnormal sites; and MIBG scintigraphy. The diagnosis is made most frequently by ultrasound. Usually NB presents as an aggressive tumour with a tendency to invade adjacent structures: the mass surrounds and engulfs, rather than displacing, large vessels such as the aorta, coeliac artery, superior mesenteric artery and vena cava28.

Radiographs Chest radiographs may show a calcified mass. Paravertebral widening in the lower chest may be caused by retrocrural spread. Abdominal radiography demonstrates a nonspecific

Performed to confirm the presence of a mass or for a routine prenatal examination, often suggests the diagnosis of abdominal NB. The mass can have varying echogenicity and is usually heterogeneous and hypervascularized on colour Doppler. Hypo-echoic areas secondary to haemorrhage and necrosis are frequent. Completely cystic NB has been described. Small echogenic foci within the mass represent calcification. The retroperitoneal location of the mass can be confirmed by demonstration of anterior displacement of the aorta and inferior vena cava (Fig. 69.16). NB classically displaces the kidney without distorting the renal collecting system. A preliminary evaluation of intra-abdominal extension of the tumour may be performed and colour Doppler may be used in displaying vessel encasement and displacement. Hepatic metastases have a variable ultrasound appearance. Diffuse infiltration of the liver (Stage IV-S) causes hepatomegaly with heterogeneity of the parenchyma.

Computed tomography and magnetic resonance imaging Although the primary tumour may first be detected by ultrasound, CT or MRI are required as a guide to staging, resectability, prognosis and follow-up. MRI has the advantage at diagnosis that it is excellent at assessing the bone marrow in patients with NB of the abdomen or other sites by revealing areas of abnormal signal intensity (low on T1- and high on T2-weighted images)29. CT of the chest, abdomen and pelvis requires a bolus intravenous administration of contrast medium. Examinations of

Table 69.2 INTERNATIONAL NEUROBLASTOMA STAGING SYSTEM Stage 1

Localized tumour confined to the area of origin; complete gross resection, with or without microscopic residual disease; identifiable ipsilateral and contralateral lymph nodes negative microscopically

Stage 2A

Localized tumour with incomplete gross excision; identifiable ipsilateral and contralateral lymph nodes negative microscopically

Stage 2B

Unilateral tumour with complete or incomplete gross resection with positive ipsilateral regional lymph nodes; contralateral lymph nodes negative microscopically

Stage 3

Tumour infiltrating across the midline with or without regional lymph node involvement; unilateral tumour with contralateral regional lymph node involvement; or midline tumour with bilateral regional lymph node involvement

Stage 4

Dissemination of tumour to distant lymph nodes, bone, bone marrow, liver, skin or other organs (except as defined in Stage 4S)

Stage 4S

Localized primary tumour (as defined for Stages 1, 2A or 2B) with dissemination limited to skin, liver, or bone marrow (< 10% tumour cells, and meta-iodobenzylguanidine scintigram negative in the marrow). Limited to infants younger than 1 year

CHAPTER 69

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

Figure 69.16 Abdominal neuroblastoma. (A) Transverse ultrasound through the right abdomen shows a solid paraspinal mass (arrows) anterior to the right kidney (rk). The aorta (a) and inferior vena cava (ivc) are displaced by the mass. (B) Longitudinal image through the right flank shows the mass (lower arrows) and the stretched inferior vena cava (ivc).

the chest should extend from the lung apices to the lower edge of the liver during relatively early enhancement; the abdominal series should cover the liver in the portal phase and extend down to the pubic symphysis. With MRI the patient must be imaged in at least two planes by T1-weighted spin-echo and fat-saturated T2-weighted fast spin-echo sequences and transverse T1-weighted gradientecho angiographic sequences. The T1-weighted spin-echo sequence must be repeated with fat-suppression after intravenous injection of gadolinium. Usually the masses are large and heterogeneous. On CT, the lesions have an attenuation equal to or less than that of muscle. Tumour calcification is readily detected (calcification is seen by CT in up to 85% of cases). Low-attenuation areas represent regions of necrosis or haemorrhage. NB usually shows mild heterogeneous enhancement, reflecting areas of vascularity alternating to areas of necrosis, haemorrhage and cystic change. At MRI, NB has prolonged T1 and T2. High signal intensity on T1-weighted images represents haemorrhage. After administration of gadolinium the enhancement is heterogeneous. With all techniques, it is essential to assess extent of tumour as clearly as possible. By definition, if the mass crosses the contralateral aspect of the vertebral body, it has crossed the midline: this determination is important in staging. Measurement of the tumour volume is crucial for prognosis and follow-up. As separation of the primary tumour from the accompanying large nodes often is not possible; measurement must include the entire mass. NB often has an invasive pattern of growth encasing the vessels, so it is crucial to define the position of the vessels and to determine the relationships of the tumour tissue to the vessels to define extent of tumour and determine operability (see Fig. 69.15).

The tumour may invade the spinal canal, kidney, or liver. Hepatic metastases and renal atrophy (ischaemia or infarction secondary to encasement or compression of the renal vessels) may be seen. Intraspinal extension may be suspected at CT but cannot be thoroughly evaluated: MRI should be performed on any paraspinal mass suspected of extending into the extradural space (Fig. 69.17). CT or MRI allow the evaluation of skull and iliac bone metastases (see Fig. 69.15). Usually, skull metastases are located in the spheno-orbital region. They appear as an infiltrating mass causing permeative bone destruction and spiculated bony changes, which may extend into the soft tissue of the scalp or push through the inner table of the skull. NB tends to regress in size with treatment, but the regression is frequently incomplete, leaving a small residual soft tissue mass, which is often calcified. Determining whether residual tissue is fibrosis or viable tumour usually requires an MIBG scintigram. The ganglioneuromas have patterns similar to those of NB, mostly occurring in the posterior mediastinum.

Radionuclide radiology CT or MRI is needed to define the local extent of disease, whereas MIBG and skeletal scintigrams are essential for determining distal spreading30. MIBG (131I- or 123I-labelled), an analogue of catecholamine precursors, is taken up by catecholamine-producing cells. In children such uptake is usually specific for NB (primary tumour and metastases). Unfortunately 30% of the primary lesions do not take up MIBG and, in other rare cases, the uptake stops despite the presence of persistent demonstrable disease. However, MIBG scintigraphy remains important in the diagnosis (characterization of

1649

1650

SECTION 8

•

PAEDIATRIC IMAGING

Figure 69.17 Thoracic neuroblastoma with intraspinal extension. MRI. (A–B) axial and coronal T1-weighted images with Gd-DOTA enhancement. (C) sagittal T1-weighted image with Gd-DOTA enhancement. (D) Axial T2-weighted image. Large posterior mediastinal mass (arrows) with anterior and lateral displacement of the aorta (a), which is encased. There is massive tumour extension into the canal through the right neural foramina to displace the spinal cord to the left. Extension into the right paraspinal soft tissues is evident.

the mass, localization of the primary tumour in patients with MEI, and evaluation of the intra- and extra-skeletal extent of disease), and follow-up (new areas of uptake generally indicate new active disease). Two-thirds of patients have metastatic bone disease at diagnosis; therefore, 99mTc-MDP bone scintigraphy should be performed in all patients at diagnosis and follow-up. MIBG uptake in bone is nonspecific because cortical uptake cannot be distinguished from marrow involvement. Positron emission tomography (PET) with [18F]FDG has a major impact on the treatment of adult cancer, whereas the reported experience with tumours of childhood is limited.

PET and MIBG imaging show similar patterns of diffusely abnormal skeletal findings in patients with extensive bone marrow involvement. A major drawback of PET is the lack of visualization of lesions in the cranium because of the high physiological activity in the brain.

Differential diagnosis The differential diagnosis encompasses all paediatric abdominal masses, and particularly Wilms' tumour and neonatal adrenal haemorrhage. Concerning Wilms’ tumour, the mean age at onset is 3 years (for NB it is younger than 2 years), it grows like a ball displac-

CHAPTER 69

ing the vessels (NB surrounds vessels), and it may invade the renal vein and inferior vena cava. It arises from the kidney with the typical claw sign; calcification is uncommon. Lung metastases are present in 20% of cases (very uncommon in NB). The diagnosis of adrenal haemorrhage is usually performed by ultrasound, which identifies the mass and shows, in sequential examinations, decreasing size and cystic evolution. Colour Doppler is useful to show avascularity of adrenal haemorrhage. MRI may show the classic signal intensity pattern of ageing blood products, but diagnostic difficulties can occur because of bleeds of different ages. Other paediatric adrenal tumours, such as phaeochromocytoma and adrenal carcinoma, are much less common. In the neck, firm to hard masses can be cysts, ectopic thymus, abscesses, inflammatory or tumoural adenopathy, benign and malignant tumours, or parotid and thyroid lesions. The most common cancers are lymphomas, thyroid cancers, rhabdomyosarcomas and NB. In the mediastinum, NB is located in the posterior anatomical space. Of the posterior mediastinal masses, approximately 80% are NB or ganglioneuroblastoma and ganglioneuroma. Other abnormalities can be: neurofibroma, neurenteric cyst, meningocele and paraspinal inflammation due to vertebral osteomyelitis.

REFERENCES 1. Adler C P, Kozlowski K 1993 Osteosarcoma. In: Adler C P, Kozlowski K (eds) Primary bone tumors and tumorous conditions in children. Springer Verlag, London 2. Lang P, Johnston J O, Arenal-Romero F, Gooding C A 1998 Advances in MR imaging of pediatric musculoskeletal neoplasms. MRI Clin North Am 6: 579–604 3. Brisse H, Edeline V, Neuenschwander S 2003 Imaging of malignant tumours of the long bones in children: monitoring response to neoadjuvant chemotherapy and preoperative assessment. In: Toma P (ed) Compendium 26th postgraduate ESPR course. Omicron, Genova, pp 75–82 4. Delbeke D, Habibian M R 1988 Non inflammatory entities and the differential diagnosis of positive three phase with bone scintigraphy. Clin Nucl Med 13: 844–851 5. Focacci C, Lattanzi R, Iadeluca M L, Campioni P 1998 Nuclear medicine in primary bone tumours. Eur J Radiol 27: 123–131 6. Fletcher B D 1999 Malignant bone tumors. RSNA special course in pediatric radiology: current concepts in body imaging at the millennium. RSNA Inc, Oak Brook, pp 173–182 7. Vanel D, Verstraete K L, Shapeero L G 1997 Primary tumors of the musculoskeletal system. Radiol Clin North Am 35: 213–237 8. Gitelis S, Schajowicz F 1989 Osteoid osteoma and osteoblastoma. Orthoped Clin North Am 20: 313–325 9. Goldman A B, Schneider R, Pavlov H 1993 Osteoid osteoma of the femoral neck: report of four cases evaluated with isotopic bone scanning CT and MR imaging. Radiology 186: 227–231

•

BONE TUMOURS AND NEUROBLASTOMA IN CHILDREN

10. Lucas D R, Unni K K, McLeod R A, O'Connor M I, Sim F H 1994 Osteoblastoma: clinicopathologic study of 306 cases. Human Pathol 25: 117–134 11. Mitchell M L, Ackerman L V 1986 Metastatic and pseudomalignant osteoblastoma: a report of two unusual cases. Skel Radiol 15: 213–218 12. Woods E R, Martel W, Mandel S H, Crabbe J P 1993 Reactive soft tissue mass associated with osteoid osteoma: correlation of MR imaging features with pathologic findings. Radiology 186: 221–225 13. Blau R A, Zwick D L, Westphal R A 1988 Multiple non ossifying fibroma. J Bone Joint Surg 70: 299–304 14. Ritsch P, Hajek P C, Pechman U 1989 Fibrous metaphyseal defects: magnetic resonance imaging appearance. Skel Radiol 18: 253–259 15. Jee W H, Choe B Y, Kang H S, Suh K J 1998 Non ossifying fibroma: characteristics at MR imaging with pathologic correlation. Radiology 209: 197–202 16. Pazzaglia U E, Pedrotti L, Beluffi G et al 1990 Radiographic findings in hereditary multiple exostoses and new theory of pathogenesis of exostoses. Pediatr Radiol 20: 594–597 17. Eich G F, Hoeffel J C, Tschappeler H, Gassner I, Willi U V 1998 Fibrous tumours in children: imaging features of a heterogeneous group of disorders. Pediatr Radiol 28: 500–509 18. Woertler K 2003 Benign bone tumors and tumor-like lesions: value of cross-sectional imaging. Eur Radiol 13: 1820–1835 19. Mahnken A H, Nolte-Ernsting C C, Wildberger J E et al 2003 Aneurysmal bone cyst: value of MR imaging and conventional radiography. Eur Radiol 13: 1118–1124 20. Jee W H, Choi K H, Choe B Y, Park J M, Shinn K S 1996 Fibrous dysplasia. MRI imaging characteristics with radiologic correlations. Am J Roentgenol 167: 1523–1527 21. Link M P, Eilber F 1997 Osteosarcoma. In: Pizzo P A, Poplack D G (eds) Principles and practice of pediatric oncology. Lippincott-Raven, Philadelphia, pp 889–920 22. Logan M P, Mitchell M J, Munk P L 1998 Imaging of variant osteosarcoma: with an emphasis on CT and MR imaging. Am J Roentgenol 171: 1531–1537 23. Meyer J S, Dormans J P 1998 Differential diagnosis of pediatric musculoskeletal masses. MRI Clin North Am 6: 561–577 24. Horovitz M E, Malawer M M, Woo S Y, Hicks M J 1997 Ewing's sarcoma family of tumors: Ewing's sarcoma of bone and soft tissue and the peripheral primitive neuroectodermal tumors. In: Pizzo P A, Poplack D G (eds) Principles and practice of pediatric oncology. LippincottRaven, Philadelphia, pp 831–863 25. Toma P, Lucigrai G, Marzoli A et al 1994 Prenatal diagnosis of metastatic adrenal neuroblastoma with sonography and MR imaging. Am J Roentgenol 162: 1183–1184 26. Siegel M J, Ishwaran H, Fletcher B D et al 2002 Staging of neuroblastoma at imaging: report of the radiology diagnostic oncology group. Radiology 223: 168–175 27. Brodeur G M, Pritchard J, Berthold F et al 1993 Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment. J Clin Oncol 11: 1466–1477 28. Abramson S J 1999 Neuroblastoma. In: Siegel M J (ed) Special course in pediatric radiology: current concepts in body imaging at the millenium. RSNA Inc, Oak Brook, pp 157–172 29. Couanet D, Geoffray A, Hartmann O et al 1988 Bone marrow metastases in children's neuroblastoma studied by magnetic resonance imaging: advances in neuroblastoma research. Prog Clin Biol Res 271: 547–555 30. Kushner B H 2004 Neuroblastoma: a disease requiring a multitude of imaging studies. J Nucl Med 45: 1172–1188

1651

CHAPTER

Paediatric Neuroradiology

70

Roxana S. Gunny and W. K. ‘Kling’ Chong

• • • • • • • • • • • • • • •

Normal brain maturation Brain structural malformations Neurocutaneous syndromes Spinal malformations Inborn metabolic brain disorders Craniosynostosis Neonatal nasal obstruction: nasal cavity stenosis/atresia Brain tumours Cerebrovascular disease and stroke Hypoxic–ischaemic injury in the developing brain Miscellaneous acquired toxic or metabolic disease Intracranial and intraspinal infections Trauma Hydrocephalus Summary

NORMAL BRAIN MATURATION Brain maturation is assessed by observing tissue characteristics related to myelination, as well as variations in morphology. Most of the changes associated with myelination occur in the first 2 years of life, but other changes in tissue characteristics occur as the brain matures. Gyral and sulcal development mainly occurs in utero or in the premature brain, but other morphological changes are observable later in life.

Normal myelination The extent of myelination of the infant brain is assessed by magnetic resonance imaging (MRI) according to specific milestones analogous to the normal milestones of clinical development. As the brain matures there is progressive T1 and T2 shortening of the white matter due to changes in the lipid and water content of developing myelin and packing of white matter tracts1. This follows a centrifugal posterior-to-anterior and inferior-to-superior pattern and is virtually complete by the age of 2 years (Fig. 70.1). Advanced MRI techniques show progressive reduction in water diffusion, increased fractional anisotropy (assessed by diffusion tensor imaging), and increased magnetization transfer 2–4. Brain myelination is detected in

grey matter earlier on T2-weighted fast spin-echo (FSE) and in the white matter tracts earlier on T1-weighted spin-echo (SE) or inversion recovery (STIR) sequences. Most myelination occurs post-term in the first 8 months, although it may extend into adulthood. The brain should appear virtually fully myelinated on T2-weighted sequences by 2 years, with almost an adult appearance on T1-weighted sequences by 10 months (Fig. 70.2). The newborn has limited motor function but a well-developed sensory system. Thus the myelination pattern seen at birth at full term is primarily in the sensory tracts. During the first 6 months of life the process of myelination is easiest to follow on T1-weighted images, where the myelinated areas appear bright.T2-weighted images are less sensitive and it takes much more myelin to produce a hypointense signal within the white matter. During this period T2-weighted images show only subtle myelination. At full term, T1-weighted images should show high signal in the dorsal medulla and brain stem, the cerebellar peduncles, a small part of the cerebral peduncles, about a third of the posterior limb of the internal capsule, the central corona radiata, and the deep white matter in the region of the pre- and post-central gyrus5. Progression of myelination is seen in the optic radiations during the first months of life. The internal capsule will demonstrate T1 shortening within the anterior limb by 3 months, while on T2-weighted images the hypointensity due to myelin is not seen until about 8 months of age. The splenium of the corpus callosum on T2-weighted images becomes hypointense at 3 months of age. The hypointense signal extends anteriorly along the body and genu, and the complete corpus callosum is myelinated at 6 months6. After 6 months the signal pattern on T1-weighted images becomes less precise, and after 10 months the brain is fully myelinated by T1 criteria. T2-weighted images are then used to assess the myelination from 6 months to 24 months of age, when the signal pattern generally is fully mature and has a completely adult pattern, though the milestones of myelination are much more imprecise than during the first 6 months of life. On T2 weighted images the first signs of mature subcortical white matter are found around the calcarine fissure at 4 months and in the pre- and post-central gyri at 8 months.

1654

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.1 Normal brain development with age seen on T2- (top row) and T1-weighted (bottom row) MRI. On T2-weighted images myelination at term and at 2.5 months is seen centrally as signal hypointensity within the posterior limbs of the internal capsules. This progresses from posterior to anterior with age, as does myelination of the corpus callosum. Myelination also progresses centrally to peripherally. The sagittal T1-weighted image shows progressive bulking up of the corpus callosum. By 6 months the corpus callosum should reach approximately its normal childhood size. The splenium is slightly enlarged with respect to the genu.

Regions of persistent hyperintensity on T2-weighted sequences known as the ‘terminal myelination zones’7 may be seen within the peritrigonal areas well into adulthood, and can be distinguished from white matter disease by the presence of a rim of normal myelinated brain between these areas and the ventricular margin. Other areas may also persist as regions of signal hyperintensity beyond 2 years, e.g. in the frontotemporal subcortical white matter, and should not be mistaken for disease (Fig. 70.3).

Normal gyration

Figure 70.2 Normal brain myelination at 3 and 12 months. On T1weighted MRI at term, T1 shortening is seen within the posterior limb of the internal capsule. This progresses posteriorly to anteriorly and centrally to peripherally until by 12 months the brain appears fully myelinated.

By 10 months the occipital subcortical white matter appears isointense with the overlying grey matter and finally shows mature hypointense signal around 1 year of age. This process proceeds anteriorly and by 18 months has finally reached the most frontal parts and the frontal poles of the temporal lobes.

Gyration is the process by which the individual gyri and sulci of the cerebral hemispheres form. The MRI appearances lag behind the extent of gyral formation seen at the same age at post-mortem. The surface of the cerebral hemispheres is initially smooth, with the interhemispheric fissure and Sylvian fissures having already formed by 16 weeks gestation. Other primary sulci, such as the callosal sulcus and parieto-occipital fissure, are recognizable at 22 weeks gestation, followed by the cingular and calcarine sulci. The central sulcus is seen in most infants by 27 weeks. Gyration then continues into the postterm period in a standardized and consistent sequence, beginning with the sensorimotor regions and visual pathways, areas that are also myelinating at the same time. The slowest regions of gyration are also those with the slowest myelination, such

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.3 Normal terminal myelination zones compared to pathological states. (A) T2-weighted MRI shows signal hyperintensity adjacent to the trigones of the lateral ventricles but with a rim of darker signal between this and the ventricular margins. This is the normal appearance of the terminal myelination zones. (B) Periventricular signal hyperintensity extending down to the ventricular margin. The ventricles are dilated posteriorly and there is irregular scalloping of the ventricular margins in keeping with white matter volume loss. These are the typical features of periventricular leukomalacia due to hypoxic ischaemia. (C) Peritrigonal and splenial signal abnormality. The ventricles are not dilated. This is the typical pattern of adrenoleukodystrophy.

as the frontal and temporal poles. By term the gyral pattern is nearly the same as the appearance in adults with further deepening of the sulci occurring post-term. The Sylvian fissures are also wider and vertically oriented and these continue to mature post-term.8,9

Other post-natal maturational changes Development of the corpus callosum begins with the posterior genu, body, and splenium, then the anterior genu and rostrum. All these components are present by 20 weeks gestation; however, it continues to grow in length and thickness through the rest of the fetal period and post-term. The adult appearance with full thickness of the corpus callosum is achieved by 8–10 months of age, and bulking up of the splenium as the visual pathways mature occurs by 4–6 months10. In the adult there are several regions where there is relative T2 hypointensity, considered to be due to the normal deposition of iron; these are the basal ganglia, particularly the globus pallidus, substantia nigra, and red nucleus. In the infant the basal ganglia begin to appear relatively T2 hypointense to cortex by about 6 months of age due to myelination, but the putamen and globus pallidus are isointense to each other and the internal capsule. They then become relatively bright with respect to white matter as this begins to myelinate. By 9 or 10 years there is a second stage of T2 shortening in the globus pallidus, substantia nigra and red nucleus, which reduces further during the second decade11. The dentate nuclei show similar though less marked changes by about age 15 years.This phase is due to iron deposition which continues throughout adult life. In normal infants up to the age of 2 months the anterior pituitary gland has a convex upper border and is of relatively high T1-weighted signal12. From 2 months the pituitary gland has a flat surface and is isointense with grey matter. It slowly grows during childhood and ranges from 2 to 6 mm in vertical diameter until puberty when it enlarges again.

BRAIN STRUCTURAL MALFORMATIONS Cerebellar malformations Cerebellar hypoplasia The cerebellum may be small due to lack of formation or cerebellar atrophy. Cerebellar hypoplasia an acquired insult, such as infection (especially congenital cytomegalovirus); an inborn error of metabolism, such as a glycosylation disorder; or unknown cause. Clinically the child may present with variable hypotonia or ataxia but these features may be very mild and there is no clear correlation between the severity of the imaging findings and the clinical presentation. Although the underlying aetiology is usually not elucidated, in around 10% of children the diagnosis includes carbohydrate-deficient glycoprotein syndrome, infantile neuroaxonal dystrophy (Fig. 70.4), pontocerebellar hypoplasia, spinocerebellar atrophies, Friedrich’s ataxia and other rare syndromes13. The pons may also be small which may simply reflect the lack of synaptic connections from the hypoplastic cerebellum. The posterior fossa is of normal size and the cerebellum is of normal signal intensity.

Dandy–Walker complex This encompasses a spectrum of cystic posterior fossa malformations from the complete Dandy–Walker malformation to persistent Blake’s pouch and mega cisterna magna, all of which have in common an apparently focal extra-axial cerebrospinal fluid (CSF) collection which is continuous with the fourth ventricle and variable cerebellar hypoplasia14. Children with any of these developmental anomalies may present as incidental findings or with developmental delay, seizures and hydrocephalus.The Dandy–Walker malformation and its variants are associated with hydrocephalus and other midline anomalies, and can be an indicator for underlying clinical syndromes and chromosomal abnormalities.The classical Dandy–Walker malformation is characterized by cystic dilatation of the fourth ventricle, which almost fills the entire enlarged posterior fossa;

1655

1656

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.5 Dandy–Walker malformation. (A,B) The fourth ventricle opens into a large posterior fossa cyst. There is associated hydrocephalus. (C) The cerebellum is hypoplastic and a thin rim of cerebellar tissue is seen forming the wall of the posterior fossa cyst (arrow). The vein of Galen, straight sinus, and venous confluence are elevated above the level of the lambdoid suture. Figure 70.4 Child with infantile neuroaxonal dystrophy. The cerebellar folia are underformed and the fissures are widened. The brainstem is also small.

the cerebellar vermis is hypoplastic and rotated or aplastic and the tentorium and venous confluence of the torcula are elevated (Fig. 70.5)15. Torcular–lambdoid inversion may be seen on plain radiography, computed tomography (CT) and MRI. At the mildest end of the spectrum, the mega cisterna magna is seen as an incidental finding of doubtful clinical significance and consists of an infracerebellar CSF collection (or normal cisternal space), occasionally with an enlarged posterior fossa but a normal cerebellum and fourth ventricle.These need to be distinguished from posterior fossa arachnoid cysts. Such arachnoid cysts do not communicate with the fourth ventricle. They may occasionally cause sufficient mass effect to warrant surgical intervention. The presence of crossing vessels and the falx cerebelli favours the mega cisterna magna over arachnoid cyst.

Joubert’s syndrome and related disorders The typical clinical syndrome is of hyperpnoea, abnormal eye movements and ataxia that is associated with a particular pattern of cerebellar dysgenesis16. The typical imaging findings may occasionally be seen in clinically normal children or in children without the full triad of clinical features, and there is some controversy as to the exact definition of this condition. In a few cases a specific genetic locus has been identified, and many syndromes with additional features such as renal cysts, ocular abnormalities, liver fibrosis, hypothalamic hamartoma and polymicrogyria have been classified with this anomaly, so that detection of the typical midbrain

changes should prompt investigation for these. It is probably best to think of it as a generalized developmental disorder of the midbrain and hindbrain. The imaging findings reflect a failure of formation of the decussation of the superior cerebellar peduncles, lack of the pyramidal decussations and other anomalies of the midbrain crossing tracts and their nuclei. On cross-sectional imaging the fourth ventricle is enlarged with a ‘batwing appearance’ and there is a cleft in the vermis. The midbrain is small. The ‘molar tooth’ appearance seen on axial images arises from the lack of the superior cerebellar decussation and the superior peduncles also appear enlarged (Fig. 70.6)17.

Rhombencephalosynapsis and other cerebellar malformations Rhombencephalosynapsis is a very rare cerebellar malformation in which the cerebellar hemispheres are fused across the midline and there is hypoplasia or aplasia of the vermis18. Again, the clinical presentation is variable. It is associated with other midline supratentorial anomalies such as absence of the septum pellucidum and corpus callosum and holoprosencephaly. L’Hermitte-Duclos or dysplastic cerebellar gangliocytoma is a developmental mass lesion with a distinctive radiological appearance in which there is enlargement of the cerebellar cortex usually affecting one hemisphere. On MRI there is a nonenhancing mass with diffusely enlarged cerebellar folia19. Pial enhancement may be demonstrated. There are many other forms of nonspecific cerebellar dysgenesis for which as yet there are no universally accepted classification systems. The Chiari malformations are discussed separately below as they represent separate entities.

CHAPTER 70

Figure 70.6 Child with Joubert’s syndrome. (A) Typical batwing appearance to the fourth ventricle (arrow) and (B) prominent superior cerebellar peduncles with failure of the normal midline decussation (arrow). This gives the typical ‘molar tooth’ appearance. The midbrain is hypoplastic in this condition.

Cerebral malformations20–22 The earliest malformations to appear relate to the formation of the neural tube, i.e. abnormalities of dorsal induction or cranial dysraphism (3–4 weeks gestation). Anencephaly, cephalocele and Chiari II (Arnold–Chiari) malformation are consequences of abnormalities of dorsal induction. The events following the formation of the neural tube are termed ventral induction, when the two separate cerebral hemispheres are formed (5–8 weeks). The holoprosencephalies are all abnormalities of ventral induction. The structures in the posterior fossa are also formed during this period. Neurones form and proliferate in the subependymal layer of the lateral ventricles known as the germinal matrix from around 7 weeks gestation. The neurones subsequently migrate peripherally along radially oriented microglia to form the layers of the cerebral cortex from 2 to 5 months’ gestation, the deeper layers forming first.

Disorders of dorsal induction Anencephaly is the most common cerebral malformation in the fetus and is incompatible with life. Most anencephalics are stillborn, but a few survive for a few days. Anencephaly in which there is no cerebral cortex should be distinguished from gross hydrocephalus in which there is a very thin, barely visible cerebral cortical mantle. Cephalocele. A cephalocele is an extracranial protrusion of intracranial structures through a congenital defect of the skull and dura mater (Fig. 70.7). Unlike myelomeningoceles in the spine there is usually no skin defect.When the cephalocele contains only leptomeninges and CSF it is a meningocele, and when it also contains neural tissue, which is usually abnormal and nonfunctioning, it is an encephalocele.When this includes part of the ventricle it is called an encephalocystocele. The cephalocele may be palpable clinically and may be pulsatile. The majority of congenital cephaloceles occur in the occipital and frontal regions. The primary role of imaging is to establish the presence of neural tissue, other intracranial malformations and hydroceph-

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.7 Parieto-occipital cephalocele with herniation of the brain and meninges through a calvarial defect. Most of the herniated component is in the form of a cerebrospinal fluid-containing meningocele.

alus, preferably by MRI. Small meningoceles may not require surgery since their size may decrease with time, producing the appearance of an ‘atretic meningocele’. The detection and localization of vascular structures is important before any neurosurgical intervention. Herniated neural tissue may have an obliterated blood supply and hence show signs of ischaemia. Chiari II malformation (Arnold–Chiari). This is discussed here as a congenital malformation of the hindbrain that, for practical purposes, is always associated with myelomeningocele. Affected children usually present with hydrocephalus following repair of the myelomingocele after birth. Other symptoms of complications of the malformation include upper airway problems, such as apnoea and stridor, and feeding problems, such as dysphagia. The Chiari II malformation consists of inferior displacement of the cerebellum, pons, medulla oblongata and cervical cord. Associated features include an inferiorly displaced and elongated fourth ventricle, beaking of the tectum, flattening of the ventral pons and low attachment of the tentorium23,24. The tentorial incisura is enlarged and the cerebellum herniates superiorly into the supratentorial space. The falx is partially absent or fenestrated with consequent interdigitation of gyri across the midline, and the massa intermedia of the thalami is enlarged.The foramen magnum is enlarged and ‘shield-shaped’ (Fig. 70.8). Other malformations that may be associated with the Chiari II malformation include a lacunar skull dysraphism (lukenschadel), disorders of neuronal migration, malformation of the corpus callosum, a dorsal midline cyst and absence of the septum pellucidum. The diagnosis can be readily made with CT by identifying the wide tentorial incisura, typical configuration of the foramen magnum and the small fourth ventricle and posterior fossa. Interdigitation of the cerebral hemispheres may be also identified. MRI is the best investigation to show complications, which include hydrocephalus, an isolated fourth ventricle, hydrosyringomyelia and compression of the craniocervical junction.

1657

1658

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.8 Chiari II malformation. (A) The posterior fossa is enlarged and ‘shield shaped’. The fourth ventricle is small and slit-like. (B) The cerebellum towers superiorly through the tentorium and there is interdigitation of a cerebral gyrus through the fenestrated falx (arrow).

The fourth ventricle in Chiari II malformation should be slitlike: a normal or enlarged ventricle suggests hydrocephalus or that the ventricle may be isolated (Fig. 70.9).

Disorders of ventral induction Holoprosencephaly is a midline malformation of ventral induction of the anterior brain, skull and face, resulting from the failure of the embryonic prosencephalon to undergo segmentation and cleavage into two separate cerebral hemispheres. Holoprosencephaly is associated with chromosomal abnormalities, facial clefting and various teratogenic factors including maternal diabetes.

Figure 70.9 Chiari II malformation. The fourth ventricle, which should normally be small and slit-like in this condition, is enlarged, indicating hydrocephalus. There is cascading tonsillar tissue herniating through the foramen magnum (white arrow). Beaking of the tectal plate is also seen (black arrow), as well as a cervical spinal cord syringomyelic cavity.

Alobar holoprosencephaly is the severest form and affected fetuses are often spontaneously aborted.The ventricular system with a single cavity is in continuity with a large dorsal cyst (Fig. 70.10). Hydrocephalus prevails, causing macrocrania in spite of microcephaly. Semilobar holoprosencephaly is less severe and there is less reduction in brain volume. Posteriorly the interhemispheric fissure is partially formed. The frontal lobes are fused, but the thalami are partially separated. There is still a single ventricle, but there is an indication of a third ventricle. The corpus callosum may be partially formed. Lobar holoprosencephaly is associated with mild (or absent) facial malformations and intellectual abilities that range from mild impairment to normal. The brain is generally of normal volume and shows almost complete separation into two hemispheres, though in the depth of the frontal lobes there is continuous cerebral cortex between the two lobes.

Malformations of commissural and related structures Agenesis of the septum pellucidum. Absence of the septum pellucidum is not a severe malformation, but should be recognized as an indicator of other cerebral malformations. Associated malformations are septo-optic dysplasia, agenesis of the corpus callosum, holoprosencephaly, Chiari II malformation, schizencephaly and other migration disorders. Septooptic dysplasia or de Morsier’s syndrome includes the triad of hypopituitarism, hypoplasia of the optic nerves and absence of the septum pellucidum (Fig. 70.11). The syndrome is not rare but the clinical manifestations are quite variable. On coronal T1-weighted MRI it is important to assess the size of the optic nerves and chiasm, as well as the pituitary gland. Commissural agenesis or dysgenesis. The major interhemispheric commissural connections are the corpus callosum and anterior and posterior commissures. The corpus callosum may be partially formed (dysgenetic) or completely absent. The anterior part (posterior genu and anterior body) is formed before the posterior part (posterior body and splenium). Thus a small or absent genu or body, with an intact

Figure 70.10 CT brain of this infant shows that the cerebral hemispheres have failed to form and there is no interhemispheric fissure or corpus callosum. Instead there is a thin pancake of cerebral tissue crossing the midline anteriorly (arrowhead) and a single holoventricle continuous with a large dorsal cyst. The midbrain and deep grey structures are fused into a single indiscriminate mass (arrow).

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.11 Septo-optic dysplasia in a child with HESX1 gene mutation. (A) The septum pellucidum is absent and the frontal horns have a typical box-like configuration. (B) The posterior pituitary gland is ectopic (arrow). (C,D) The right optic nerve is small (arrows).

splenium and rostrum, indicates secondary destruction rather than abnormal development. The structure develops between about 7 and 20 weeks gestation, which parallels the development of the rest of the cerebrum and the cerebellum; abnormalities of the corpus callosum are therefore commonly associated with other congenital malformations of the brain, such as the Chiari II malformation, the Dandy–Walker malformation, lipoma, abnormalities of neuronal migration and organization, dysraphic anomalies, encephaloceles, septo-optic dysplasia, ocular anomalies and other midline facial anomalies25. There are many well-defined syndromes in which callosal abnormalities feature, including Aicardi’s syndrome character-

ized by seizures, intellectual impairment and brain abnormalities. Callosal agenesis is a feature of oculocerebrocutaneous syndrome or Delleman’s syndrome (Fig. 70.12). Dysgenesis of the corpus callosum is also frequently seen in fetal alcohol syndrome. The presence of parallel lateral ventricles with the third ventricle seen between them and without the normal convergence of the bodies of the ventricles indicates callosal agenesis and can be detected on axial imaging. It is thought that the interhemispheric cyst frequently seen with this malformation originates from the herniated third ventricle but has lost continuity with this structure. On sagittal MRI, partial

Figure 70.12 Child with skin lesions and right orbital cyst in oculocerebrocutaneous syndrome. (A) There is callosal agenesis with dorsal interhemispheric cysts. (B) The right cerebral hemisphere is dysplastic with thickening of the cortex and an indistinct grey–white matter junction (arrow). There is a cyst expanding the orbit with a small calcified globe seen inferiorly (arrowhead). (C) An associated Dandy–Walker posterior fossa malformation is also present.

1659

1660

SECTION 8

•

PAEDIATRIC IMAGING

or complete absence of the corpus callosum must be distinguished from a generally thinned one (which is more likely to be an acquired abnormality, e.g. ischaemia). In callosal agenesis the vertically oriented sulci extend right down to the ventricle and there is no horizontally running cingulate sulcus. Presence of one midline anomaly, such as callosal agenesis (Fig. 70.13), should prompt the reporting radiologist to look carefully for other midline anomalies that are frequently associated.

Malformations of neuronal migration and cortical organization Abnormalities of the cerebral cortex are a common finding in children with developmental delay and children with partial epilepsy. The cortex is formed from neuroblasts that are generated in the germinal matrix, an ependymal layer of the ventricular wall. The neuroblasts migrate to the surface of the brain along radially oriented glia, passing neurones previously laid down to form the layers of cortex. Thus the six layers of the cortex are formed with the youngest neurones on the surface and the oldest ones adjacent to the subcortical white matter. The migration of the neuroblasts starts at about week 7 of gestation, is most intense during weeks 15–17 and is largely complete by weeks 23–24. When the process of neuronal migration is disturbed, there is a migration disorder resulting in morphologically abnormal cortex.

Schizencephaly Schizencephaly is a defect that involves the complete cerebral mantle and connects the calvarium and the outer surface of the brain with the lateral ventricles. The defect is a cleft

Figure 70.13 Callosal agenesis. (A) Axial T2-weighted MRI shows separated ventricles with parallel orientation. The superior part of the third ventricle is just seen. (B) Sagittal T1-weighted MRI through the midline confirms callosal agenesis. There is no cingulate sulcus and the vertically oriented cerebral sulci extend right down to the third ventricle. This finding is associated with other midline anomalies such as a frontoethmoidal cephalocele (arrow), seen also on the axial T2-weighted MRI (arrow, C). (D) The optic chiasm is absent.

lined by grey matter and leptomeninges, which differentiates it from a transmantle infarction in which the defect is lined by white matter. The schizencephaly may have an ‘open lip’ with a wide open defect (Fig. 70.14), or a ‘closed lip’ when the cleft is closed but lined with grey matter entirely into the ventricle. The convolutional pattern of the cortex adjacent to the clefts is abnormal and consists of polymicrogyria. The clinical features are variable, depending on the size and location of the lesion. Severe seizures are quite common, as is spasticity. Children with bilateral clefts have severe mental and psychomotor developmental delay. Wide clefts usually correlate with moderate to severe developmental delay, while children with narrow or closed-lipped lesions may only have hemiplegia and/or seizures. The location of the lesion is typically central, involving the pre- and post-central gyri. However, the clefts may also be found in parasagittal, frontal or occipital sites when the clinical manifestations are often mild. In most cases the diagnosis is made with CT, but this may not detect all cases with closed lips, which are best detected using coronal T1-weighted MRI, preferably a three-dimensional (3D) volume acquisition. MRI also shows the abnormal appearance of the cortical mantle along the cleft and the cortex appearing thicker than normal owing to the presence of polymicrogyria.The contralateral hemisphere may also have developmental abnormalities, such as polymicrogyria and subependymal heterotopia. CT may show subependymal or parenchymal calcification in many cases, which suggests that one cause of schizencephaly may be intrauterine infection with cytomegalovirus.

CHAPTER 70

Figure 70.14 Schizencephaly with a grey matter-lined cleft (arrows) extending from the leptomeningeal surface through the brain parenchyma to the ventricular margin.

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.15 Classical with lissencephaly. This child was ‘floppy’ at birth and then had developmental delay, seizures, and strabismus. MRI shows a smooth gyral pattern which is slightly more developed frontally in keeping with classical lissencephaly (LIS1 mutation). The cerebral cortex is generally thin and there is a band of arrested neurones deep to the ‘cell sparse zone’. The Sylvian fissures are vertically oriented and extend into a vertical cleft.

Lissencephaly–agyria–pachygyria The lissencephaly–agyria–pachygyria group includes the severest forms of abnormal neuronal migration. Lissencephaly literally means ‘smooth brain’. The brain has very few or no gyri, opercularization (development of the Sylvian fissures) is abnormal, and the Sylvian fissures are shallow. Other associated features are agenesis or hypoplasia of the corpus callosum or septum pellucidum. Agyria refers to the total absence of a convolutional pattern, i.e. there are no gyri or sulci, and is synonymous with ‘complete lissencephaly’. In pachygyria, the gyri are relatively few and are unusually broad and flat. The entire brain is not affected and therefore it is also known as ‘incomplete lissencephaly’. Macroscopically and on standard thick-slice MRI, pachygyria may be difficult to differentiate from polymicrogyria, but the latter may be distinguished more easily on higher resolution imaging, e.g. with volumetric T1-weighted imaging (e.g. MPRAGE, SPGR, FLASH). In complete lissencephaly the brain surface is smooth and the Sylvian fissures are wide and vertically orientated.The cortex is thin; there is a ‘cell-sparse zone’ of white matter adjacent to it and a broad band of grey matter, the ‘arrested neurones’ that have failed to migrate to the cortex, deep to it (Fig. 70.15). The gyral pattern of the brain resembles the appearance of the 23–24-week normal fetal brain. The implication of this is that lissencephaly is unlikely to be reliably diagnosed on early fetal MRI, though there may be clues to the diagnosis before this, such as the immature appearance of the Sylvian fissures, cell-sparse zone and the broad band of arrested neurones. With incomplete lissencephaly there is an antero-posterior (AP) gradation of gyral development. In the LIS1 mutation (chromosome 17) the fronto-temporal gyri are more developed than the parieto-occipital gyri and in the X-linked form the posterior gyri are more developed. Cobblestone lissencephaly is the result of overmigration of neurones, and is characterized by thick meninges adherent to the smooth cortical surface. There is no recognizable lamination of the cortex. Large areas of heterotopia are prominent and delay in myelination is frequently present. The congeni-

tal muscular dystrophies (CMDs) are a heterogeneous group characterized clinically by hypotonia at birth, muscle weakness and joint contractures, developmental delay and seizures. They may demonstrate ‘cobblestone lissencephaly’, e.g. in Walker– Warburg syndrome with coexistent ocular abnormalities and Fukayama CMD with relatively normal eyes.

Grey matter heterotopia Grey matter heterotopia refers to the occurrence of grey matter in an abnormal position anywhere from the subependymal layer to the cortical surface, but the term is usually reserved for ectopic neurones in locations other than the cortex. Its most common clinical presentation is as a seizure disorder. Heterotopia can be subependymal, focal subcortical or band formed, or parallel to the ventricular wall (double cortex). They are isointense with cortical grey matter on all imaging sequences and do not enhance after the intravenous infusion of paramagnetic contrast agents. Subependymal heterotopia are smooth and ovoid, with their long axis typically parallel to the ventricular wall, quite different from subependymal hamartomas in tuberous sclerosis which are irregular and have their long axis perpendicular to the ventricular wall (Fig. 70.16). Hamartomas are more heterogeneous depending on any calcification, gliois, etc., and do not have signal characteristics of grey matter. They may also enhance after the intravenous infusion of a paramagnetic contrast agent. Small isolated areas of heterotopia may be seen occasionally as incidental findings in normal patients. Focal subcortical heterotopia produces variable motor and intellectual impairment, depending on the size and location of the lesions. The overlying cortex is thin with shallow sulci. The foci may be isolated or may coexist with other malformations such as schizencephaly, microcephaly, polymicrogyria, dysgenesis of the corpus callosum, or absence of the septum pellucidum.

1661

1662

SECTION 8

•

PAEDIATRIC IMAGING

Transmantle cortical dysplasia

Figure 70.16 Subependymal grey matter heterotopia (A) and subependymal hamartomas of tuberous sclerosis (B). (A) Multiple subependymal continuous ‘nodules’ running along the ventricular margin with signal intensity isointense to grey matter. (B) Scattered nodules which project into the ventricles and with variable signal intensity. Some are markedly hypointense in keeping with calcification. Note also the multiple regions of cortical and subcortical white matter abnormality with slight mass effect in keeping with cortical tubers.

In transmantle cortical dysplasia, abnormal cells extend all the way from the wall of the ventricle to the cortex. The presenting signs and symptoms relate to the size and location of the lesion. Patients with involvement of just a small part of the cortex may have a normal neurological examination, though most patients with transmantle cortical dysplasia have complex partial epilepsy. Large lesions are quite characteristic on neuroimaging but smaller ones may be very difficult to detect. Typical features include blurring of the junction between the cortex and the white matter, and hyperintensity (compared to the white matter on T2) extending from the lateral ventricular wall to the blurred cortex (Fig. 70.18). There is a correlation between abnormal venous drainage and dysplastic cortex so it can be useful to look for large cortical vessels on MRI in the search for the abnormal area of cortex. The lesion may sometimes be calcified and seen as a radially oriented band of T1 shortening extending down towards the ventricle. Occasionally the only clue to its presence may be a focus of calcification seen on CT.

Hemimegalencephaly Band heterotopia or ‘double cortex’ is predominantly seen in girls (> 90%). Although the imaging appearances are severe, the degree of developmental delay may be quite variable. Some children may even be normal except for a relatively mild seizure disorder. On imaging, band heterotopia appears as a homogeneous band of grey matter between the lateral ventricle and the cerebral cortex, separated from both by a layer of white matter. The overlying cortex is usually of normal thickness but has shallow sulci. Band heterotopia may be complete or partial, the latter predominantly involving the frontal lobes.

Polymicrogyria This is considered to be a disorder of neuronal organization, occurring after neuronal migration. The extent of polymicrogyric cortex may vary from small, isolated, unilateral areas to larger areas of bilateral disease (Fig. 70.17). The appearances on imaging may also vary from apparently broad, thickened gyri mimicking pachygyria, to clearly overconvoluted multiple gyri, to subtle lesions that are difficult to detect even when using high-resolution 3D MRI sequences.

Figure 70.17 Extensive bilateral cerebral hemisphere polymicrogyria. Virtually no normal cortex is seen. At first glance the cortex appears thickened but closer inspection reveals an overconvoluted gyral pattern and a ‘lumpy bumpy’ grey–white matter interface (including regions marked by white arrows). The Sylvian fissure is abnormally oriented with a parietal cleft that extends posteriorly (black arrow).

This is a structural malformation due to defective neuronal proliferation, migration and organization, leading to hamartomatous overgrowth of all or part of one hemisphere. It may occur in isolation or in association with syndromes such as proteus, epidermal naevus and Klippel–Trenaunay–Weber, syndromes neurofibromatosis type 1 (NF1) and tuberous sclerosis. The affected hemisphere contains regions of pachygyria, polymicrogyria, heterotopia, as well as dysmyelination and gliosis. Usually (but not always) the hemisphere is enlarged, there is diffuse cortical thickening, white matter signal abnormality and there may be calcification. The ipsilateral ventricle is enlarged and there is a very characteristic configuration of the frontal horn which is straight and pointed (Fig. 70.19). Occasionally this may be the only imaging clue to an underlying malformation.

Chiari I malformation This may be considered a form of hindbrain deformation rather than a true malformation and is characterized by tonsillar descent. It may be an acquired condition and has occasionally been observed either to improve or worsen without interven-

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

most frequently seen are NF1, tuberous sclerosis, neurofibromatosis type 2 (NF2), von Hippel–Lindau disease and Sturge– Weber syndrome26,27.

Neurofibromatosis type 1

Figure 70.18 Child with focal seizures and right fronto-parietal lobe cortical dysplasia. (A) Axial T2-weighted MRI shows blurring of the right frontal grey–white matter junction and more extensive associated white matter signal hyperintensity involving most of the hemisphere, including the parietal lobe at this level (arrow). (B) Coronal T1 SPGR image (part of the volumetric dataset) also shows grey–white matter junction signal abnormality (arrow) and more subtle white matter change.

tion. Clinical symptoms are more likely when there is greater than 5 mm descent below the foramen magnum, and therefore descent below this level is considered to be clinically significant. However, MRI does not predict who is symptomatic. Children between 5 and 15 years have greater tonsillar descent up to 6 mm as a normal finding compared to children under 5 years or adults. There may be an associated syringomyelia. Symptoms include cough-induced headache, cranial nerve palsies and disassociated peripheral anaesthesia.

NEUROCUTANEOUS SYNDROMES The neurocutaneous syndromes or phakomatoses are congenital malformations affecting particularly structures of ectodermal origin, i.e. the nervous system, skin and eye. The

The most common neurocutaneous syndrome is NF1 with an incidence of 1 in 3000–4000 births. As well as being one of the most common inherited central nervous system (CNS) disorders, it is the most common autosomal dominant condition, due to a mutation on chromosome 17 which encodes for the tumour suppressor gene product neurofibromin, and the most common inherited tumour syndrome. Half are new mutations. The diagnosis is made on the basis of at least two major criteria (Table 70.1)28. Minor criteria are supportive of the diagnosis. CNS tumours in NF1 include visual pathway gliomas, plexiform neurofibromas and cranial and peripheral nerve gliomas.These may be diagnosed radiologically without recourse to biopsy and the diagnostic criteria allow the diagnosis of NF1 to be made purely on neuroimaging or as an adjunct to clinical findings. Visual/optic pathway gliomas (OPGs) are usually WHO Grade I pilocytic astrocytomas and are the most common brain abnormality in NF1, occurring in up to 15% of patients. Most of these are diagnosed in childhood but only half are symptomatic. OPGs in NF1 are more likely to affect the optic nerves rather than the chiasm and post-chiasmatic pathways, as opposed to non-NF1 OPGs, and are also associated with a better prognosis. Once the tumour involves the chiasm and hypothalamus there is a risk of precocious puberty and a greater risk of visual deterioration29. They have a wide spectrum of biological behaviours ranging from static or minimal growth in most to rapidly increasing size in a minority. Fusiform expansion of the nerve and widening of the optic foramen may be detected on CT, along with the very characteristic sphenoid wing dysplasia and plexiform neurofibroma. Although CT can detect the intra-orbital involvement of OPG, this involves irradiating the eye and is less sensitive than MRI for delineating tumour within the chiasm and intracranial extension. Other orbital features seen in NF1 include dilatation of the optic nerve sheaths due to dural ectasia and intra-orbital extension of plexiform neurofibroma. Variable

Table 70.1 DIAGNOSTIC CRITERIA FOR NEURO-FIBROMATOSIS TYPE 1 Major criteria

Figure 70.19 Hemimegalencephaly with overgrowth and enlargement of the left cerebral hemisphere. There is thickening of the cortex with broad, thickened gyri, underdeveloped sulci, and extensive white matter signal abnormality. Note the typical appearance of the frontal horn and genu of the corpus callosum, which is straight and points anteriorly (arrow).

Minor criteria

Café-au-lait spots

Small stature

Freckling in the inguinal or axillary areas

Macrocephaly

One plexiform neurofibroma or two neurofibromas of any type*

Scoliosis*

Visual pathway glioma*

Pectus excavatus*

Two or more lisch nodules of iris

‘Hamartomatous lesions’ of NF1*

Distinctive osseous lesion, e.g. sphenoid dysplasia or thinning of cortex*

Neuropsychological abnormalities

First-degree relative with neurofibromatosis type 1 (NF1) *

Radiologically detectable features.

1663

1664

SECTION 8

•

PAEDIATRIC IMAGING

extension into the chiasm, the lateral geniculate bodies or optic radiations is best detected on MRI. A suggested NF1 imaging protocol includes orbital MRI with axial and coronal dual-echo STIR, coronal and T1-weighted pre- and fatsaturation post-gadolinium images, all with a slice thickness of 3 mm or less. Images obtained with a fat-saturation pulse allow elimination of chemical shift artefact at the interface of the optic nerve sheath complex and intra-orbital fat, making assessment easier, and contrast medium improves visualization of the normal intra-orbital optic nerves. The whole of the optic nerve may be expanded or there may be subarachnoid extension of tumour around a normal sized nerve. Tumour infiltration within the nerve causes variable enhancement of expanded nerve within the optic nerve sheath, whereas if the tumour is predominantly subarachnoid, a rim of enhancing tumour around a minimally enhancing nerve is detected. It is important to identify the expanded nerve within the optic nerve sheath in order to distinguish it from NF1-associated dural ectasia in which the optic nerve sheath is expanded by CSF rather than tumour. Generally these optic pathway tumours are kept under observation unless symptomatic or progressive (≤ 5%). Spontaneous involution of tumour is also well recognized. There is an increased risk of other CNS tumours in NF1 with OPG. These are usually WHO Grade I pilocytic astrocyomas occurring in 1–3% of patients, particularly within the cerebellum and brainstem, although other low grade and higher grade tumours also occur. In the brainstem they are usually less aggressive than non-NF1 brainstem astrocytomas and are more likely to be seen in the medulla and midbrain (e.g. tectal plate gliomas) than the pons. A tectal plate tumour may cause aqueduct stenosis and hence hydrocephalus. These brainstem NF1 tumours are often biologically benign and may regress spontaneously, and therefore clinical management is generally not aggressive unless clinical/radiological tumour progression is seen. Another characteristic lesion of NF1 is the so-called ‘hamartomatous’ changes of NF1, also known as ‘unidentified bright objects (UBOs)’, ‘neurofibromatosis bright objects (NBOs)’ or areas of myelin vacuolation. These are seen in 60–80% of NF1 cases, depending on the age at which the

Figure 70.20 Child with neurofibromatosis type 1 (NF1). (A,B) There are very characteristic lesions within the lentiform nuclei, brainstem, and midbrain, the socalled ‘hamartomatous’ lesions or ‘unidentified bright objects’ of NF1 (arrows). They are hyperintense on T2-weighted imaging, with minimal mass effect. Basal ganglia lesions may demonstrate some T1 shortening, as in this case, or are hypointense on T1-weighted imaging. (C) There are also bilateral optic nerve gliomas extending into the optic chiasm (arrows).

child is imaged, and in 95% of children with NF1 and OPG, and may have an impact on cognitive function30. They are few in number before the age of 4 years, increase in number and volume between 4 and 10 years and then decrease in the second decade, being rare over the age of 20. Therefore, they are rarely seen in the adult NF1 population. They appear as multiple T2 hyperintense lesions with minimal mass effect and no contrast enhancement in typical sites such as the pons, cerebellar white matter, internal capsules, basal ganglia, thalami and hippocampi. They have normal signal on T1-weighted images, apart from lesions in the basal ganglia which are often slightly T1 hyperintense (Fig. 70.20). Generally these features distinguish them from gliomas. Astrocytomas, however, may develop in the areas involved by UBOs and radiologically it is not always possible to distinguish them. Enhancement and increasing mass effect are suspicious for tumour development, in which case the involved areas should be kept under regular imaging review. Other non-CNS tumours recognized in NF1 include phaeochromocytoma, carcinoid, rhabdomyosarcoma and childhood chronic myeloid leukaemia. Plexiform neurofibromas are one of the main diagnostic criteria of NF1. These are multinodular lesions formed when tumour involves either multiple trunks or multiple fascicles of a large nerve.Typical locations include the orbit growing along the ophthalmic division of the trigeminal nerve in association with progressive sphenoid wing dysplasia. They are hypodense on CT and generally do not enhance with contrast agents. On MRI they have a more heterogeneous appearance, of low T1-weighted signal intensity and T2-weighted hyperintensity, with variable contrast enhancement, although at least part of the tumour normally enhances. Extension occurs along the nerve pathways into the pterygomaxillary fissure, orbital apex/superior orbital fissure and cavernous sinus (Fig. 70.21). Other characteristic sites include the lumbosacral and brachial plexi. There is a malignant potential with transformation to fibrosarcoma quoted at between 2% and 12%. Neurofibromas are more homogeneous and well-defined lesions which cause diffuse expansion of nerves. It may be possible to distinguish plexiform from other types of neurofibroma radiologically. The former are more diffuse lesions. Neurofibromas appear on MRI as nodules seen along the spinal nerves of the cauda

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.21 Infant with neurofibromatosis type 1. The diagnosis was made from these images. (A) There is sphenoid wing dysplasia causing expansion of the middle cranial fossa (arrow) and absence of the lateral orbital wall, which causes the ‘bare orbit’ sign (arrow) on AP plain radiographs. (B) There is an associated extensive plexiform neurofibroma involving the deep and superficial fascial spaces of the neck, tongue, and orbit. (C,D) This is almost indistinguishable on imaging from multiple cranial nerve fibromas involving the left cavernous sinus (arrows) and middle cranial fossa. Note the neurofibroma has extended through the foramen ovale and is elevating the dura, seen as a black line (arrowhead).

equina and, as they enlarge, they extend out through the neural exit foramina, enlarging them. They may have a central region of T2 hypointensity producing a target appearance. Malignant transformation in these is less common, and cannot be excluded radiologically, but suggestive features are local pain, rapid growth, large size, and internal heterogeneity. NF1 is associated with some characteristic bone dysplasias, including lambdoid sutural dysplasia, thinning of long bone cortices and kyphoscoliosis with a high thoracic acute curve. One of the most common is sphenoid wing dysplasia marked by a bone defect which allows herniation of the temporal lobe through the orbit. On plain radiography this produces the ‘empty’ or ‘bare’ orbit (Fig. 70.21A). Clinically there may be pulsatile exophthalmos due to transmission of CSF pulsations. The globe may also be affected. As well as kyphoscoliosis due to a primary skeletal dysostosis, there may be dural ectasia with vertebral scalloping and lateral meningoceles containing CSF. Scoliosis may be seen in association with an intrinsic spinal cord tumour or peripheral nerve neurofibroma.

Tuberous sclerosis Tuberous sclerosis (TS) is a multisystem genetic neurocutaneous syndrome characterized by hamartomas, cortical tubers and benign neoplastic lesions (giant cell astrocytomas), with an incidence of around 1 in 5800 live births. The most frequently affected organs are the skin, brain, retina, lungs, heart, skeleton and kidneys, but the few manifestations that are associated with the reduced life expectancy seen in this condition are, in order of highest to lowest frequency, neurological disease (seizures

and subependymal giant cell tumour), renal disease (angiomyolipoma and renal cell carcinoma), pulmonary disease (lymphangioleiomyomatosis and bronchopneumonia) and cardiovascular disease (rhabdomyosarcoma and aneurysm). TS is autosomal dominant with a high level of penetrance and variable phenotypic expression; 60–70% of cases are sporadic and two-gene mutations have been identified, TSC1 and TSC2, encoding protein products with a tumour suppressor function. The classical clinical presentation of TS is the triad of intellectual impairment, epilepsy and adenoma sebaceum, but there is a wide phenotypic range. There are a large number of diagnostic primary, secondary and tertiary criteria for TS, some of which are radiological and which categorize TS as ‘definite’, ‘probable’ or ‘possible’. Radiological primary criteria are the presence of calcified subependymal nodules while noncalcified subependymal nodules and tubers, cardiac rhabdomyoma and renal angiomyolipoma are secondary criteria32,33. In 80% of patients with TS, infantile spasms or myoclonic seizures are the presenting symptom. Conversely, 10% of children with infantile spasms will have evidence of TS, so structural MR neuroimaging is indicated in these children. Ocular manifestations of TS include retinal hamartomas seen near the optic disc in 15% and are often bilateral and multiple. On CT they appear as nodular masses originating from the retina and when calcified may be difficult to distinguish from retinoblastomas unless there are also calcified subependymal nodules. Subretinal effusions may also be detected. Micro-ophthalmia and leukocoria are other features. The intracranial manifestations include subependymal hamartomas or nodules (SENs), subependymal giant cell

1665

1666

SECTION 8

•

PAEDIATRIC IMAGING

astrocytomas (GCAs), radially oriented linear bands and cortical tubers (Fig. 70.22). SENs are the most common lesion and are seen in 88–95% of individuals with TS. They may be calcified which is a useful diagnostic feature on CT or T2* MRI; calcification increases with age and is rarely detected under the age of 1 year33. Histologically they are indistinguishable from GCAs. The latter are determined by location in the caudothalamic groove adjacent to the foramen of Monro, progressive growth on serial imaging and the presence of hydrocephalus. Contrast enhancement may be seen in both SENs and GCAs. The lack of myelination in infants helps to identify white matter anomalies, which become less visible as myelination progresses34. SENs and linear parenchymal tuberous sclerosis lesions in infants under 3 months old are hyperintense on T1-weighted images and hypointense on T2-weighted images as opposed to the reverse pattern of signal intensity in older children and adults.

Sturge–Weber syndrome Sturge–Weber syndrome is a congenital syndrome characterized by a port-wine naevus on the face and ipsilateral leptomeningeal angiomas with a primarily parieto-occipital distribution35.

Bilateral involvement may occasionally occur. The clinical manifestations include the onset of focal seizures, appearing during the first year of life, and developmental delay with progressive hemiparesis, hemianopsia and intellectual impairment. The seizures progressively become refractory to medication. The leptomeningeal angiomas cause abnormal venous drainage with chronic ischaemia, leading ultimately to cortical atrophy and calcification, the latter feature being usually very prominent. By 2 years of age, skull radiographs may reveal ‘tramline calcifications’ within the cortices. In early imaging the brain may look normal on CT as well as on MRI without intravenous contrast enhancement. The involved hemisphere progressively becomes atrophied and the pial angioma is seen as diffuse pial enhancement of variable thickness (Fig. 70.23)36. The ipsilateral white matter appears hypointense on T1-weighted images and hyperintense on T2weighted images. Other findings include enlargement of the ipsilateral choroid plexus and dilatation of transparenchymal veins that communicate between the superficial and deep cerebral venous systems. In ‘burnt out’ cases the pial angioma may no longer be detected after contrast enhancement, leaving only a chronically shrunken and calcified hemisphere.

Neurofibromatosis type 2 This is located to an abnormality on chromosome 22 and occurs in 1 in 50 000 live births. Nearly all have bilateral vestibular schwannomas, other tumours such as meningiomas and other cranial and peripheral nerve schwannomas and ependymomas, including spinal tumours (Fig. 70.24). While in adults hearing loss is a common presentation, seizures and facial nerve palsy are more common in children.

Other neurocutaneous syndromes

Figure 70.22 Intracranial manifestations of tuberous sclerosis. (A) Multiple tubers involving the cortex and subcortical white matter. Bilateral lesions are seen at the foramina of Monro, in keeping with giant cell astrocytomas (arrows). (B) Subependymal nodules project into the ventricles, some of which are markedly hypointense, in keeping with calcification (arrowhead).

Figure 70.23 Sturge–Weber syndrome. (A) Coronal T1 postcontrast image shows an enhancing pial angioma overlying the right cerebral hemisphere which is atrophic. The right choroid plexus is enlarged. Foci of signal hypointensity within the gyri and adjacent white matter are due to calcification. (B) Axial T2-weighted image shows in addition prominent superficial cortical veins and ependymal veins (arrows). (C) Axial post-contrast T1-weighted image shows bilateral choroidal angiomas (arrows) in addition to the pial angioma.

These include hypomelanosis of Ito in which hypomelanotic skin lesions are associated with polymicrogyria, heterotopias and callosal dysgenesis, basal cell naevus syndrome and PHACES (posterior fossa malformations, facial haemangiomas, arterial anomalies, cardiac and eye anomalies and sternal cleft) syndromes. Neurocutaneous melanosis is a rare syndrome in which giant congenital melanocytic naevi on the skin are associated with intracranial melanosis (Fig. 70.25). Neuroimaging

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.24 Neurofibromatosis type 2 (NF2). (A) Bilateral cerebellopontine angle masses extending into the internal auditory meati and causing expansion (arrow) in a child with NF2 and bilateral acoustic neuromas. (B) Trigeminal schwannomas extending into the cavernous sinus on the right. The arrow indicates the cisternal segment of the right trigeminal nerve. (C) Sagittal T1-weighted images show expansion of the neural exit foramina by enhancing nerve or nerve sheath tumours (arrow). (D) Axial image shows a mainly dural mass extending out through the neural foramen with a small intradural component (arrow).

Figure 70.25 Child with giant pigmented naevus and neurocutaneous melanosis. (A) Initial coronal T1-weighted MRI shows the typical regions of T1 shortening within the amygdalae of the mesial temporal lobes. (B,C) Axial T1-weighted MRI obtained 2 years later shows hydrocephalus in addition to the regions of melanin deposition within the mesial temporal lobes and cerebellum. These do not show any enhancement. There is a posterior fossa arachnoid cyst, a described association. (D) Within the spine there is diffuse pial enhancement over the spinal cord in addition to focal haemorrhagic and partly enhancing extramedullary lesions, indicating malignant melanoma, which was subsequently confirmed histologically.

detects the clusters of melanocytes by the melanin that they are associated with, therefore appearing as regions of T1 shortening on MRI in characteristic locations: the anterior and mesial temporal lobe, cerebellum and pons. CT may detect areas of increased density but these are much more difficult to appreciate. Diffuse melanosis with intracranial and intraspinal leptomeningeal spread may occur and therefore hydrocephalus. Degeneration into malignant melanoma may occur.

SPINAL MALFORMATIONS Normal development The spinal cord forms during three embryological stages known as gastrulation (2–3 weeks), primary neurulation (3–4 weeks) and secondary neurulation (5–6 weeks). During gastrulation the embryonic bilaminar disc consisting of epiblast and hypoblast is converted to a trilaminar disc by migration of cells from the

1667

1668

SECTION 8

•

PAEDIATRIC IMAGING

epiblast through Hensen’s node, a focal region of thickening occurring at the cranial end of the midline ‘primitive streak’ of the disc. This results in the midline notochord and a layer that will form the future mesoderm. During primary neurulation the notochord induces the overlying ectoderm to become neurectoderm and form the neural plate. Subsequent folding and bending occurs until the margins unite to form the neural tube. The cranial end closes at day 25 while the caudal end closes a couple of days later. Finally the caudal cell mass arises from the primitive streak and undergoes retrogressive differentiation with cavitation. This is the origin of the fetal neural tissue and vertebrae distal to S2, and will become the conus medullaris. A focal expansion of the central canal known as the terminal ventricle occurs as a result of incomplete retrogressive differentiation. This may be seen as a normal asymptomatic finding in young children and may persist in a small minority into adulthood. It is seen on all post-mortem studies but is bigger in those detectable on MRI. Spinal dysraphisms can result from abnormalities occurring during any of these periods37–39.

Definitions Spinal dysraphisms may be open, in which case nervous tissue is exposed, or closed, in which case the defect is covered by skin, although a cutaneous lesion such as a dimple, sinus, hairy naevus or haemangioma may be seen as a marker of an underlying defect in 50% of these cases38,39. Spina bifida refers to the failure of fusion of the posterior spinal bony elements. The neural placode is a flat segment of unneurulated nervous tissue that may be seen at the end of the spinal cord or at an intermediate position along its course. Tethered cord syndrome is a clinical diagnosis of progressive neurological deterioration (usually leg weakness, deformities such as scoliosis or foot abnormalities, loss of bladder and bowel function), presumed to be due to traction damage on the tethered cord. Although there may be some suggestive features such as an associated spinal cord syrinx, this is not a diagnosis made radiologically; the position of the conus or neural placode following successful surgical ‘untethering’ always remains unchanged. The level of the spinal cord termination has a Gaussian distribution. It is a popular misconception that the spinal cord lies lower in the neonate and continues to rise as the vertebral column grows during childhood. In fact most authors agree that it has reached its adult position by term, and in 98% lies above L2/3, the majority lying between T11/12 and L1/240,41. The spinal cord termination should be considered convincingly abnormal if seen at or below L3.

Open spinal dysraphism Most open spinal dysraphisms (OSDs) are myelomeningoceles and these are virtually always associated with Chiari II malformation (Fig. 70.26). The neural placode protrudes beyond the level of the skin and there is an expanded CSF-containing sac lined by meninges38,39. A small proportion of OSDs are myeloceles where the placode is flush with the surface and there is no meningocele component. Both disorders result from defective closure of the primary neural tube and persistence of unneurulated nervous tissue in the form of the neural

Figure 70.26 Repaired myelomeningocele in a child with Chiari II malformation. (A) The neural placode terminates inferiorly in the meningeal sac (arrow). The lumbosacral posterior vertebral elements have not formed. (B) Post-repair the skin is continuous over the defect.

placode, usually at the lumbosacral level at the spinal cord termination. Nerve roots arising from the everted ventral surface of the placode cross the widely dilated subarachnoid spaces of the meningocele to enter the neural exit foraminae. The posterior elements of the vertebral column and any other mesenchymal derivatives, such as the paravertebral muscles, remain everted. Hemimyelomeningoceles and hemimyeloceles may occur with diastematomyelia (split cord syndrome) and may be associated with an asymmetric skin abnormality.The clinical presentation is also very asymmetrical. These can be explained embryologically by failure of primary neurulation of one hemicord in addition to a gastrulation abnormality. Myelomeningoceles are operated on soon after birth as if untreated the exposed neural tissue is prone to ulceration and infection. In some centres in utero repair has been correlated with subsequent failure to develop the typical hindbrain malformation of Chiari II, although other abnormalities such as the enlarged massa intermedia and falx fenestration persist. Hydrocephalus and the need for surgical drainage may also be delayed and even reduced. Hydrocephalus usually develops 2–3 days post-neonatal repair but may occur pre-operatively. Other causes of post-operative deterioration include re-tethering of the spinal cord and the development of a syrinx which may later cause scoliosis42.

Closed spinal dysraphism Closed spinal dysraphisms (CSDs) may be associated with cutaneous stigmata or a mass43,44. This may be a subcutaneous lipomatous mass overlying the spinal defect, as in lipomyeloceles and lipomyelomeningoceles, due to abnormal primary

CHAPTER 70

neurulation in which there is premature disjunction of the cutaneous ectoderm from the adjacent neuroectoderm, allowing mesenchymal elements to come into contact with the open neural tube. In a lipomyelocele the junction between the placode and the lipoma lies within the spinal canal (Fig. 70.27), while in lipomyelomeningoceles it lies outside. Typically the placode is rotated to one side while the meninges are herniated to the other. A posterior meningocele consists of herniation of a CSF sac lined by dura and arachnoid through the spinal defect, resulting in a clinically apparent mass covered by skin. These are mainly lumbosacral but may be seen at any level. Anterior meningoceles are typically presacral, are seen with caudal agenesis and are present in older children and adults with low back pain and bladder/bowel disturbance. The terminal myelocystocele is a rare condition associated with syndromes such as VACTERL in which the central canal is dilated by a large hydromyelic cavity that herniates into a posterior meningocele through the posterior spinal bony defect. The cavity is usually discontinuous with the meningocele and lies inferiorly and posteriorly. CSDs without a mass include simple intramedullary and intradural lipomas. These typically occur along the posterior midline in a subpial juxtamedullary location at the cervicothoracic level. Embryologically they are also the result of premature disjunction and the lipoma fills in the gap between the unopposed folds of the neural placode. On MRI they have the signal characteristics of fat, including signal suppression on STIR sequences. The filum terminale lipoma is considered to arise from a disturbance of caudal regression. Fatty thickening of the filum terminale is detected on MRI, and may be more easily seen on axial T1-weighted sequences. This is estimated to occur in 1.5–5% of the normal adult population and may be considered a normal variant in the absence of the clinical tethered cord

•

PAEDIATRIC NEURORADIOLOGY

syndrome. The ‘tight’ filum terminale, also due to abnormal retrogressive differentiation, is a short, thick filum greater than 2 mm in diameter associated with clinical tethering and a lowlying spinal cord.

Dorsal dermal sinus A dermal sinus is an epithelial-lined opening on the skin with variable fistulous extension to the dural surface, typically seen in the lumbosacral region and often associated with cutaneous stigmata, such as hairy naevus and capillary haemangioma43,44. Embryologically it arises from a failure of disjunction of neuroectoderm from cutaneous ectoderm. Dermal openings seen at the sacrococcygeal level are directed inferiorly below the thecal sac and are known as sacrococcygeal pits. They do not require further imaging. Those seen above the intergluteal cleft pass superiorly and may form a fistulous connection with the dural sac and warrant further investigation. On MRI they are seen as a thin linear strip of tissue hypointense to adjacent fat (Fig. 70.28). US may provide useful information about the intradural extension of the dermal sinus tract and the mobility of the conus45.

Diastematomyelia The split notochord syndromes are disorders of notochord midline integration. In diastematomyelia the spinal cord is split in two, with each hemicord having one anterior and one posterior grey matter horn46,47. In type I diastematomyelia there is a craniocaudal spectrum of abnormality ranging from partial clefting cranially to two complete dural sacs separated by an osteocartilaginous spur inferiorly. There may be plain radiographic features including scoliosis or hemivertebrae or bifid/ fused vertebrae at the level of the bony spur. The bony spur in type I diastematomyelia is completely extradural, usually midline, though it may be seen coursing obliquely from the

Figure 70.27 Closed spinal dysraphism. The spinal cord is too low and the neural placode terminates at the lumbosacral junction in a lipomyelocele (black arrows). There is an associated spinal cord syringomyelic cavity (white arrows). The posterior elements are deficient and everted.

1669

1670

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.28 Dorsal dermal sinus. (A) The termination of the spinal cord is very low and there is a syringomyelic cavity with internal septations. In addition there is a more focal cyst within the conus confirmed at surgical untethering to be a dermoid cyst (arrow). (B,C) There is an associated dermal sinus track (arrows).

posterior vertebrae to the laminae, and may be complete or incomplete. It is seen at the caudal end of the split and the hemicords fuse tightly just below it. On MRI marrow signal may be detected within it and distinguish it from a simple fibrous band. Above it the split is much longer. In rare cases, the separate dural sacs may terminate distally with different fates, one as a OSD and the other as a CSD. In diastematomyelia type II (Fig. 70.29) there is a single dural sac in which the two hemicords lie, and a fibrous septum, seen as a band of T2 hypointensity, may be seen passing intradurally between the two hemicords.There may be no septum or there may only be partial clefting of the cord. The conus is often low and there may be a tight or fatty filum. Both forms of diastematomyelia may be associated with hydromyelia.

Figure 70.29 Cervical spinal cord diastematomyelia type II with associated craniocervical meningocele. (A) Sagittal T2-weighted MRI appears to show signal abnormality and thinning of the spinal cord, and is the clue to the diastematomyelia seen on the axial images. (B,C) Axial T2-weighted MRI shows that the cord has split into two hemicords. The apparent signal abnormality is in fact normal cerebrospinal fluid interspersed between the two hemicords. These re-unite inferiorly. The meningocele is seen herniating through a bony defect in the vertebral posterior elements.

Neurenteric cysts The severest and rarest form of notochordal midline integration anomaly occurs with dorsal enteric fistulas and neurenteric cysts48. The cysts are usually seen intradurally anterior to the spinal cord and are derived from endodermal remnants trapped between a split notochord. They have signal characteristics of CSF or of proteinaceous fluid with T1 shortening (Fig. 70.30). The fistula connects the dorsal skin with bowel across a duplicated spine.

Disorders of the caudal cell mass/caudal regression syndrome The last group of developmental spinal abnormalities is those affecting development of the caudal cell mass49. Caudal agenesis and the rarer condition of segmental spinal dysgenesis are considered to occur as a result of apoptosis of notochordal cells

Figure 70.30 Neurenteric cyst ventral to the medulla and upper cervical spinal cord, displacing them posteriorly, and with the vertebral arteries displaced around it. The cyst is hyperintense on the T2-weighted images and there is also some mild T1 shortening (arrow).

which have not formed in their correct craniocaudal position. In caudal agenesis there is a severe abnormality which results in absence of the vertebral column at the affected level, as well as a truncated spinal cord, imperforate anus and genital anomalies. It may be seen with OEIS (omphalocele, exstrophy,

CHAPTER 70

imperforate anus, spinal defects), VACTERL and the Currarino triad in which there is partial sacral agenesis (or sometimes a ‘scimitar’-shaped sacrum), anorectal malformation and a presacral mass which may be a teratoma or anterior meningocele50–52. In type I caudal agenesis, which affects secondary neurulation and formation of the caudal cell mass, but also primary neurulation of the distal true notochord, there is a high (often at T12) abrupt spinal cord termination with a characteristic wedge-shaped configuration and variable coccygeal to lower thoracic vertebral aplasia (Fig. 70.31). Clinically the neurological deficit due to absence of the distal spinal cord is stable. In type II caudal agenesis the true notochord is not affected and only the caudal cell mass is involved. The vertebral aplasia is less extensive with up to S4 present as the last vertebra. It may be difficult to detect partial agenesis of the conus because it is stretched and tethered to a fatty filum terminale, lipoma, lipomyelomeningocele, or anterior sacral meningocele. Finally, in segmental spinal dysgenesis there is a segmental abnormality affecting the spinal cord, segmental nerve roots and vertebrae, and associated with a congenital paraparesis and lower limb deformities. On imaging there is an acute angle kyphus, and the spine and spinal cord in the most severe cases may appear ‘severed’. In less severe cases the cord is focally hypoplastic.

Figure 70.31 Caudal regression syndrome. The spinal cord is truncated with a typical blunt edge seen at the inferior margin of T12. The thecal sac terminates at the superior margin of L4. The sacrum distal to S2 is agenetic (arrows).

•

PAEDIATRIC NEURORADIOLOGY

INBORN METABOLIC BRAIN DISORDERS Metabolic disorders may be inborn or acquired. Inborn errors of metabolism may result from an enzyme deficiency leading to the build-up of a directly toxic metabolite, or have an indirect toxic effect by activation or inhibition of another metabolic pathway, leading to increased levels of a different toxic metabolite. The inborn errors of metabolism may be subdivided according to a number of different classification systems, one of which is radiological. One system classifies them by cellular organelle involvement into mitochondrial (usually meaning disorders of mitochondrial energy metabolism), lysosomal and peroxisomal disorders. Another scheme classifies them by the biochemical enzyme pathway affected, e.g. the organic acidaemias, aminoacidopathies, or disorders of heavy metal metabolism. In the final classification scheme neuroimaging can be used to help classify them according to white or grey matter involvement, specifically by MRI. Interested readers can also refer to Van der Knaap and Valk53 and Patay54. There are some general principles that may be helpful in detecting and characterizing the radiological features. The abnormalities of metabolic disease are characteristically bilateral and symmetrical. Assessment on MRI should include analysis of grey and white matter structures. When describing white matter abnormalities it is helpful to describe them in terms of general location (lobar involvement, centripetal/centrifugal distribution and AP gradient), juxtacortical U-fibre involvement, involvement of deep white matter structures such as the internal capsule, corpus callosum and white matter tracts such as the pyramidal tracts as they descend from the motor strip (precentral gyrus) through the posterior limbs of the internal capsules and cerebral peduncles to the decussation within the medulla and then into the spinal cord. Assessment of the grey matter should include analysis of the cerebral and cerebellar cortex and basal ganglia and thalami for signal abnormality, swelling and volume loss. Signal changes include T2 and T1 prolongation, but faint T1 shortening within the basal ganglia may be seen also when there is calcification. Other deep grey structures include the red nuclei, and subthalamic and dentate nuclei. Calcification is much better assessed on CT. Macrocephaly is a useful clinical pointer to diseases such as megalencephalic leukodystrophy with subcortical cysts (MLC), Canavan’s and Alexander’s leukoencephalopathies, glutaric aciduria type I, GM2 gangliosidosis and l-2 hydroxyglutaric aciduria. An assessment of the degree of myelination is useful as delay or hypomyelination is frequently seen in metabolic disorders. Although often nonspecific it is a helpful clue to an underlying neurometabolic condition. Myelination requires active energy-dependent metabolism and therefore may also be seen with cardiorespiratory illness. However, hypomyelination may also be seen as a general marker of developmental delay and correlates well with clinical developmental milestones. In premature infants the appropriate MRI markers for corrected age should be assessed before deciding that myelination is immature. There are also specific inherited hypomyelination

1671

1672

SECTION 8

•

PAEDIATRIC IMAGING

disorders that may be detected on MRI as a myelination pattern which is immature for the child’s age. These include Pelizaeus Merzbacher disease, in which T2-weighted imaging often shows more severe hypomyelination compared to T1weighted imaging (Fig. 70.32). Serial imaging should be assessed for progressive cerebral atrophy. This may involve white matter, shown as ventricular enlargement and thinning of the corpus callosum, cortical grey matter shown as sulcal widening, or deep grey matter with atrophy of the specific basal ganglia and thalamic structures. Pathognomonic imaging patterns are seen in X-linked adrenoleukodystrophy (ALD), Alexander’s disease, glutaric aciduria type I, Canavan’s disease, l-2 hydroxyglutaricaciduria, neonatal maple syrup urine disease and MLCs, of which classic X-linked adrenoleukodystrophy is the most common leukodystrophy of children and affects 1 in 20 000 boys. It is due to a defect in a peroxisomal membrane protein leading to defective incorporation of fatty acids into myelin. Screening of family members of affected cases for the specific gene defect should be considered. Clinically boys present between the ages of 5 and 10 years with learning difficulties, behavioural problems, deteriorating gait and impaired visuospatial perception. Adrenal insufficiency may precede the CNS presentation or may be absent. Without bone marrow transplantation (which replaces the defective gene), the disease progresses to spastic paraparesis, blindness and deafness. Lorenzo’s oil may also delay disease progression. Imaging features are of low attenuation on CT and T2-weighted hyperintensity on MRI in the posterior central white matter, specifically the splenium and peritrigonal white matter progressing to the corticospinal tracts and visual and auditory pathways. The regions of T2 signal abnormality

Figure 70.32 Pelizaeus Merzbacher disease in a 2 year old child with nystagmus and developmental delay. (A) Axial T2-weighted image shows dark regions of myelination in the posterior limbs of the internal capsules, splenium, and genu of the corpus callosum, and within the parieto-occipital lobe white matter. The rest of the cerebral hemispheres are not myelinated (the white matter is too bright). This pattern is severely delayed and equivalent approximately to an age of 6 months. (B) Paradoxically coronal T1-weighted MRI shows much more advanced myelination which appears virtually complete, with T1 shortening extending right out into the subcortical white matter.

show increased diffusion.The leading edge of the demyelination enhances, where there is active inflammation and disruption of the blood–brain barrier (Fig. 70.33). MR spectroscopy may detect early changes and may in the future guide early bone marrow transplantation before overt and irreversible changes in the white matter have occurred. Adrenomyeloneuropathy is a variation on ALD which presents in young adults or adolescents, usually boys, with progressive paraparesis and cerebellar signs. It has a less specific radiological pattern causing diffuse disease of the white matter, and it more commonly involves the cerebellum and less frequently the cerebral hemispheres. In Alexander’s disease, which has a neonatal, juvenile and adult form, imaging shows extensive white matter abnormality beginning in the frontal and periventricular white matter (Fig. 70.34). Large cystic cavities are seen within the frontal and temporal regions. The basal ganglia may also be involved. Contrast enhancement may be seen along the ventricular ependyma. l-2 hydroxyglutaricaciduria is a slowly progressive disorder which is usually discovered in childhood or early adulthood, although it is likely to have started earlier than this. The clinical presentation is nonspecific with learning difficulties, epilepsy and pyramidal and cerebellar signs.The MRI findings show white matter involvement with peripheral involvement, particularly of the subcortical U fibres, internal, external and extreme capsules, sparing of the periventricular white matter and corpus callosum, and with a slight frontal predominance. There is macrocephaly. There is also grey matter involvement affecting the basal ganglia and sparing the thalami (Fig. 70.35). Maple syrup urine disease (MSUD) is an autosomal recessive disorder in which an enzyme deficiency leads to an accumulation of amino acids (leucine, isoleucine and valine) and their metabolites. MLC is a recently identified autosomal recessive leukodystrophy with macrocephaly. Suggestive MRI patterns include methylmalonic acidaemia in which there is bilateral symmetrical involvement of the globus pallidus with sparing of the thalami and the rest of the basal ganglia (Fig. 70.36). The cerebral cortex is also normal. In the acute stage there is swelling and oedema of these structures while in the chronic phase there is imaging evidence of atrophy and gliosis. Bilateral pallidal involvement is also seen as T2 hyperintensity in other rarer inborn errors of metabolism, such as GAMT (guanidinoacetate methyltransferase deficiency), Kearns–Sayer syndrome and some acquired and toxic disorders, such as kernicterus and carbon monoxide poisoning (Fig. 70.36). Disorders of heavy metal metabolism include Wilson’s and Menke’s diseases, both disorders of copper metabolism, molybdenum cofactor deficiency and disorders of magnesium and manganese metabolism. Wilson’s disease results from defective extracellular copper transport and leads to multi-organ copper deposition. Hyperintensities on T2-weighted MRI are seen in the basal ganglia, midbrain and pons, thalami and claustra, and there is T1 shortening in the basal ganglia and thalami, as in other hepatic encepha-

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.33 Adrenoleukodystrophy. MRI of a 6 year old boy with increasing gait disturbance and impaired vision demonstrates peritrigonal and splenial signal abnormality (increased signal on T2-weighted images and low signal on T1weighted images), and (A,B,C, arrows) marginal enhancement at the leading edges where there is active inflammation, typical of adrenoleukodystrophy (D, arrow).

Figure 70.34 Child with Alexander’s disease. Head circumference charts showed the child had macrocephaly. (A) Axial T2-weighted MRI shows extensive bilateral symmetrical deep and subcortical white matter signal hyperintensity with a frontal predominance and mild swelling. (B) Sagittal T1-weighted MRI shows corresponding low signal in the affected areas without evidence of cavitation but in keeping with oedema.

lopathies. Menke’s disease is X-linked and affects transcellular copper metabolism at the level of the cell membrane. There is a systemic failure of copper-requiring enzymes, particularly those of the cytochrome c-oxidase system. They

acquire connective tissue defects with ‘kinky hair’, inguinal herniae, hyperflexible joints and bladder diverticula. In the brain there is progressive cerebral atrophy which may allow subdural collections of CSF or subdural haematomas (and therefore mimics of nonaccidental head injury). The basal ganglia may also show T1 shortening. Children develop a severe cerebral vasculopathy in which vessels are tortuous and prone to dissection. Disorders of cellular organelle function include mitochondrial, lysosomal and peroxisomal disorders. Mitochondria are involved in energy metabolism; lysosomes in the degradation of macromolecules, e.g. those involved in the maintenance of cell membrane integrity such as lipids and lipoproteins; and peroxisomes have a role in both catabolic and anabolic metabolism. Mitochondrial disorders include those of mitochondrial energy metabolism affecting oxidative phosphorylation, fatty acid oxidation and ketone metabolism. Respiratory chain disorders affect the respiratory chain, a multicomplex protein on the inner membrane of the mitochondria which has an integral role in oxidative phosphorylation and tend to be multisystem diseases. In the brain they may result in multiple cerebral infarcts in nonvascular territories. Leigh’s disease can

1673

1674

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.35 L-2 hydroxyglutaricaciduria axial T2-weighted MRI. Coronal FLAIR and show extensive white matter signal hyperintensity involving mainly deep and subcortical white matter but with relative sparing of the periventricular white matter, and bilateral pallidal involvement. The dentate nuclei are also abnormal.

metabolic demand, e.g. febrile illness. Patients may also have a cardiomyopathy and endocrinopathies. Imaging features are those of cerebral infarcts in nonvascular territories and symmetrical basal ganglia calcification. Lysosomal disorders include Krabbe’s disease and metachromatic leukodystrophy. Imaging features which might suggest Krabbe’s disease are white matter changes, more severely posteriorly and centrally; basal ganglia and thalamic involvement, including T2 hypointensity; cerebellar white matter abnormality, sparing the dentate nuclei; and involvement of the pyramidal tracts within the brainstem. Peroxisomal disorders include Zellweger’s syndrome and Xlinked ALD. Imaging features of Zellweger’s syndrome include severely delayed myelination, periventricular germinolytic cysts, peri-Sylvian polymicrogyria and grey matter heterotopias. When this combination is seen in the clinical context of severe hypotonia with visual and hearing deficits, seizures, hepatomegaly and jaundice, the pattern may be considered pathognomonic. Figure 70.36 Examples of signal abnormality affecting both globus pallidi. (A) Kernicterus. (B) Methylmalonicacidaemia. (C) Kearns–Sayer syndrome. This child also has hyperintense signal in both caudate nuclei and thalami, left frontal white matter and (D) Dorsal midbrain and cerebellar dentate nuclei (arrows).

be caused not only by respiratory chain defects but also by enzyme disorders such as those of pyruvate and tricarboxylic acid metabolism. Bilateral typically symmetrical signal change is seen within the brainstem, deep cerebellar grey matter, subthalamic nuclei and basal ganglia (Fig. 70.37). The midbrain changes have been described as a ‘panda face’ with involvement of the substantia nigra and tegmentum, and the medulla is frequently involved and may account for the apneoic episodes often seen clinically. Cerebral grey matter infarction may also be seen. MELAS (mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes) is another mitochondrial disease, typically occurring between the ages of 4 and 15 years but which can occur at any age. Acute metabolic decompensation may be provoked by any insult which causes increased

CRANIOSYNOSTOSIS55 Craniosynostosis is a disorder of growth, one of the manifestations of which is premature closure of one or more calvarial or skull base sutures. The three broad categories of craniosynostosis are the simple nonsyndromic type, usually involving one suture; complex syndromic forms involving many sutures; and secondary craniosynostosis due to disrupted growth caused by a wide range of insults such as drugs and metabolic bone disease or secondary to an underlying small brain, as in chronic, treated hydrocephalus or any other cause of microcephaly. The most common type of primary craniosynostosis is simple sagittal synostosis. The diagnosis is made initially by clinical assessment of the skull shape. Imaging provides confirmatory evidence and information regarding the skull base and orbits, and is important in the assessment of intracranial complications of craniosynostosis, such as hydrocephalus and visual failure. Standard radiographs will allow assessment of the coronal, sagittal, lambdoid and metopic sutures on the AP view; lambdoid and sagittal sutures on the Towne’s view; and coronal and lambdoid sutures

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.37 Leigh’s disease. There is bilateral symmetrical signal hyperintensity on the coronal FLAIR (A) and axial T2 images (D), (E) matched by hypointensity on the coronal T1-weighted MRI (B) affecting the midbrain, pons, and medulla. Symmetrical contrast enhancement (C), (F) indicates breakdown of the blood–brain barrier in keeping with active disease.

on the lateral view in addition to assessment of the skull shape, foramen magnum and fontanelles. The affected suture may be absent, indistinct, show bridging sclerosis or a heaped up or beaked appearance, but also may appear normal if the synostosis is fibrotic not bony. Skull growth decreases perpendicular to the suture and increases parallel to it.Therefore, greater weight is given to the skull shape than the radiological evidence of direct sutural involvement. Conversely, if the sutures are not clearly visualized but the skull shape is normal, then craniosynostosis is unlikely. Sagittal synostosis produces an elongated head shape called scaphocephaly (Fig. 70.38). Bicoronal synostosis causes brachycephaly or foreshortening in the AP direction which

is accompanied by lateral elevation of the sphenoid wings producing the characteristic ‘harlequin’ deformity, upward slanting of the petrous apices and hypertelorism. Unicoronal synostosis causes anterior plagiocephaly or asymmetrical skull deformity and may be associated with compensatory growth on the unaffected side resulting in frontoparietal bossing (Fig. 70.39). Metopic synostosis causes trigonocephaly or ‘keel deformity’ and an AP view may show parallel, vertically oriented medial orbital walls. True unilateral lambdoid synostosis, the rarest form of monosutural synostosis, causes posterior plagiocephaly. This should be distinguished from the much more common positional or deformational plagiocephaly in which the suture is normal, and which is seen more often since the

Figure 70.38 Sagittal synostosis. (A) Brain CT and (B) lateral scout view showing the typical ‘boat-shaped’ skull or scaphocephaly of sagittal synostosis.

Figure 70.39 Unicoronal synostosis. (A) Axial CT and (B) 3D surfaceshaded reformat show the asymmetrical head shape of left frontal plagiocephaly due to unicoronal craniosynostosis, with bossing seen on the right side.

1675

1676

SECTION 8

•

PAEDIATRIC IMAGING

1992 recommendation of the American Academy of Pediatrics to place newborns supine rather than prone in their cots to reduce the incidence of sudden infant death. In this case the skull deformity is caused by the child lying on one side in preference to the other, and may also be seen in torticollis or developmental delay. On plain radiographs and CT the lambdoid sutures appear open. Imaging of the spine may reveal segmentation anomalies, e.g. C5 and C6 in Apert’s syndrome, and atlanto-occipital assimilation and basilar invagination in Apert’s and Crouzon syndrome. CT is more sensitive and specific than plain radiographs for detecting radiological evidence of craniosynostosis, such as sclerosis and bony bridging, and CT venography may be helpful to assess the jugular foramina, variations in venous anatomy and patency of the venous sinuses. 3D CT surface-shaded bone and soft tissue reconstructions or maximum intensity projections (MIPs) with a low milliampere technique may also be acquired, preferably in the craniofacial unit where treatment is being considered. The sutures should be assessed on both the axial 2D CT and 3D CT as either technique may miss relevant diagnostic features. There is also increasing prenatal diagnosis of craniosynostosis, particularly the syndromic forms, by ultrasound and fetal MRI. In Apert’s syndrome there are features of brachycephaly due to bicoronal synostosis, a wide open midline calvarial defect from the root of the nose to the posterior fontanelle in what would normally be the sagittal and metopic sutures and anterior fontanelle (Fig. 70.40). The sutures never form properly and instead bone islands appear within the defect, eventually coalescing to bony fusion by about 36 months. There is also hypertelorism with shallow anterior cranial fossae, depressed cribriform plate, as well as maxillary hypoplasia causing midface retrusion with exorbitism. Indeed the globe may actually sublux onto the cheek. The hand and feet deformities distinguish

this condition from most other syndromic craniosynostoses and include syndactyly, phalangeal fusion and short radially deviated thumb producing the ‘mitten’ or more severe ‘hoof ’ hand. Children with Crouzon’s syndrome demonstrate a more complex syndromic synostosis involving the coronal, sagittal, metopic and squamosal sutures with early rather than late fontanelle closure. There is no midline calvarial defect but there is maxillary hypoplasia, hypertelorism, exorbitism and dental malocclusion. The limbs are usually clinically normal. CT will detect any underlying structural brain abnormality. Although the brain usually appears structurally normal, midline anomalies such as callosal and septum pellucidum agenesis, limbic system abnormalities in Apert’s, ventriculomegaly or distortion of the posterior fossa and skull base causing Chiari I malformations (tonsillar descent) may be detected. Tonsillar herniation is more frequently seen in Crouzon’s syndrome, probably because there is more frequent skull base sutural synostosis in these conditions compared to Apert’s. Predicting raised intracranial pressure is known to be difficult by imaging, and correlation with clinical assessment is extremely important. Sometimes, however, direct invasive intracranial pressure (ICP) monitoring will be required. Hydrocephalus, seen in 4–25% of craniosynostosis and more commonly in the syndromic forms, may be multifactorial; possible aetiologies include tonsillar herniation and altered craniocervical junction CSF dynamics and venous hypertension due to venous foraminal narrowing, such as the jugular foramina, which ultimately may lead to venous occlusion and the development of venous collateral pathways such as enlargement of the stylomastoid emissary veins. Hydrocephalus is the most sensitive radiological indicator (see below) but only detects 40% of these children with raised ICP. Finally CT may also be useful to assess the airway. Midface hypoplasia, small maxilla with dental overcrowding and basilar kyphosis may contribute to nasopharyngeal obstruction. Deviation of the nasal septum and choanal atresia may also be detected.

NEONATAL NASAL OBSTRUCTION: NASAL CAVITY STENOSIS/ATRESIA The most common cause of neonatal nasal obstruction is mucosal oedema, followed by choanal atresia, skeletal dysplasias and congenital dacrocystocele due to distal nasolacrimal duct obstruction.

Choanal atresia

Figure 70.40 Apert’s syndrome. (A,B) 3D CT surface-shaded display shows the wide open defect of the sagittal suture and brachycaphaly with bicoronal synostosis typical of Apert’s syndrome. The coronal sutures appear fused and are ridged. (C,D) Plain radiographs of the hands show the ‘mitten hand’ appearance with syndactyly and shortened metacarpals.

Choanal atresia/stenosis, a congenital malformation of the anterior skull base characterized by failure of canalization of the posterior choanae, is the most common form of nasal cavity stenosis. It may be bony and/or fibrous in nature, unilateral presenting in later childhood with chronic nasal discharge and bilateral presenting in newborns with respiratory distress, particularly during feeding and which is a surgical emergency. Bilateral forms are more likely to be syndromic (50%) than unilateral forms and common associations are Crouzon’s, Treacher–Collins, CHARGE and Pierre–Robin syndromes.

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

The CHARGE syndrome describes the association of colobomas of the eye, heart defects, atresia of the choanae, retardation of growth and development, genitourinary anomalies and ear anomalies. The atresia is best evaluated on CT by direct axial and direct coronal imaging on a bone algorithm after administration of a nasal decongestant. The nasal cavity appears funnel shaped with a fluid level proximal to the obstruction. The posterior vomer is thickened and the nasal septum is deviated to the side of the stenosis. A bony, fibrous or membranous bridging bar across the posterior choana is seen (Fig. 70.41).

Skeletal dysplasias Fibrous dysplasia is a benign congenital disorder in which bone is gradually replaced by fibrous tissue. McCune–Albright syndrome is a subtype of polyostotic fibrous dysplasia in which there is pituitary hypersecretion (hence precocious puberty but also Cushing’s syndrome, etc.) and café-au-lait spots. Cherubism refers to fibrous dysplasia of the mandible and maxilla, and unlike the other forms of fibrous dysplasia is an inheritable condition. Typically the mandibular condyles are spared. The teeth may be displaced, impacted, resorbed, or appear to be floating. Clinical symptoms other than cosmetic deformity relate to the site of bony involvement; hence cranial nerve impingement caused by narrowing of the skull base foramina, exophthalmos and optic foraminal narrowing caused by orbital involvement are all seen. On CT the typical lesion is a region of bony expansion with a ‘ground-glass’ appearance, but lesions may be lytic or sclerotic, or a combination of all three (Fig. 70.42).

BRAIN TUMOURS56 The United Kingdom Children’s Cancer Study Group (UKCCSG) guidelines57 At initial presentation all paediatric brain tumours should have brain and whole spine MRI, including contrast-enhanced images. A suggested protocol includes T2-weighted FSE axial and coronal images of the brain, T1-weighted SE pre- and post-contrast imaging in at least two orthogonal planes, and post-contrast imaging of the whole spine. Ideally spine imaging should be performed at the outset; but if the need for surgery

Figure 70.42 Polyostotic fibrous dysplasia with diffuse calvarial expansion, mixed lytic lesions, and sclerosis. The optic nerve canals are markedly narrowed (arrows).

is immediate, then postoperative spinal imaging (pre- and postcontrast T1-weighted MRI to allow exclusion of T1 shortening due to post-surgical blood products) should be performed. The immediate postoperative MRI should be performed within 48 h (or 72 h maximum, according to the UKCCSG guidelines), the rationale being that post-surgical nodular enhancement which mimics tumour will not be seen before then. Post-surgical linear enhancement may be seen within this time period and indeed intra-operatively, and should therefore be interpreted with caution. The management strategy and frequency of subsequent surveillance imaging is determined by the tumour histology and presence of residual or recurrent tumour58–60.

Posterior fossa tumours The most common intra-axial cerebellar tumours in children are medulloblastoma, pilocytic astrocytoma, ependymoma and atypical teratoid/rhabdoid tumour, of which medulloblastoma and astrocytoma are the most common. Cerebellar haemangioblastoma may be seen in the context of von Hippel–Lindau disease but otherwise is an unusual tumour in the paediatric age group. Other posterior fossa tumours include brainstem gliomas and extra-axial tumours, such as dermoid and epidermoid cysts, schwannoma, neurofibroma, meningioma and skull base lesions, such as Langerhans’ cell histiocytosis, Ewing’s sarcoma (Fig. 70.43) and glomus tumours.

Cerebellar tumours

Figure 70.41 Choanal atresia. Axial skull base CT in a child with chronic nasal discharge shows right-sided choanal atresia. There is bony narrowing of the funnel-shaped posterior right choana down to a bony bridging bar (arrows) and pooling of secretions proximally.

Clinically all of these tumours may present as a ‘posterior fossa’ syndrome with lethargy, headache and vomiting due to hydrocephalus and/or direct involvement of the brainstem emetic centre. Before the fontanelles have closed infants may present with macrocephaly and sunsetting eyes.Truncal and gait ataxia is seen more often in older children and adults. Medulloblastomas are highly malignant small, round cell tumours.They are slightly more common than pilocytic astrocytomas in most pathological series, are more common in boys, and account for 30–40% of posterior fossa tumours. They are

1677

1678

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.43 Ewing’s sarcoma. This unusual posterior fossa tumour is also hypointense on the T2-weighted images but it is extradural. The dura is seen as a hypointense layer around the tumour and it erodes the right petrous bone. There are retained mastoid secretions. This was a Ewing’s sarcoma of the skull base.

also associated with some rare oncogenetic disorders such as Li–Fraumeni syndrome, Gorlin’s or basal cell naevus syndrome (with falcine calcification), Turcot and Cowden syndromes. They are aggressive, high-grade tumours (WHO Grade IV) and tend to have a shorter onset of symptoms, typically shorter than 1 month, compared to other cerebellar tumours.The peak age of presentation is 7 years but they have a wide range and may be seen from the neonatal period to late adulthood.There is a second peak in young adults presenting with the ‘desmoplastic’, less aggressive type of medulloblastoma, which is seen more frequently in the cerebellar hemisphere than the vermis. Occasionally there may be symptoms and signs or imaging evidence of intracranial or intraspinal leptomeningeal metastatic disease at presentation, a feature observed more commonly than with other posterior fossa tumours. The typical appearance of the childhood medulloblastoma on CT is of a hyperdense midline vermian mass abutting the roof of the fourth ventricle, with perilesional oedema, variable patchy enhancement and hydrocephalus.The brainstem is usually displaced anteriorly rather than directly invaded. Cystic change, haemorrhage and calcification are frequently seen. On MRI, the mass is hypointense or isointense compared to grey matter.The CT finding of hyperdensity and MRI finding of T2 hypointensity, supported by the presence of restricted diffusion on diffusion-weighted imaging, are in our experience the most reliable observations in prospectively differentiating medulloblastoma (and atypical rhabdoid tumour which on imaging appears identical to medulloblastoma) from ependymoma or other posterior fossa tumours (Fig. 70.44, Table 70.2)61. Both the CT hyperdensity and MRI T2 signal hypointensity reflect the increased nuclear-to-cytoplasmic ratio and densely packed cells of the tumour, and are particularly useful in differentiating medulloblastomas with ‘atypical’ features, such as a lateral site involving the foramen of Luschka or extrusion through the foramen of Magendie, which are more commonly seen with ependymomas. Medulloblastomas demonstrate restricted diffusion and reduced N-acetyl asparatate (NAA) peak with an increased choline-to-creatine ratio, and occasionally lactate and lipid peaks on MR spectroscopy.

Figure 70.44 Medulloblastoma. (A) CT and (B–D) axial T2, ADC, and diffusion MRI show a mixed solid and cystic mass within the right cerebellopontine angle encroaching on the pons and fourth ventricle and causing hydrocephalus. The solid component is hyperdense on CT, hypointense on the T2-weighted sequence, and demonstrates restricted diffusion in keeping with a cellular tumour. Despite some less typical features, such as lateral site (more usually seen in older patients and associated with the desmoplastic variant) and cystic components, on the basis of the signal characteristics this was correctly diagnosed as a medulloblastoma.

Table 70.2 DIFFERENTIAL DIAGNOSIS OF POSTERIOR FOSSA TUMOUR WITH CT HYPERDENSITY AND T2 HYPOINTENSITY Medulloblastoma/primitive neuroectodermal tumour /atypical teratoid/ rhabdoid tumour Choroid plexus carcinoma Ewings’ sarcoma Chondrosarcoma Chordoma Lymphoma Langerhans’ cell histiocytosis

In medulloblastoma, both intracranial and intraspinal subarachnoid dissemination should be actively looked for, and is seen in a third of cases at presentation, most often occurring as irregular, nodular leptomeningeal enhancement, (Fig. 70.45). Imaging is reported as being more sensitive than CSF cytology and false positives can be avoided by pre-operative imaging of the brain and spine. Occasionally, enhancement may not always be detected or may be very mild, making the detection of both disseminated disease and residual or recurrent tumour on surveillance imaging more difficult. Other features of leptomeningeal disease include sulcal and cisternal effacement and communicating hydrocephalus; thickening, nodularity and clumping of nerve roots; and pial ‘drop’ metastases along the surface of the spinal cord.

CHAPTER 70

Figure 70.45 Medulloblastoma. Same patient as in Figure 70.44 with posterior fossa medulloblastoma has evidence of disseminated metastatic disease. There is nodular enhancement over the conus medullaris and a mass within the thecal sac in addition to pial enhancement over the midbrain and cerebellar folia (arrows).

Standard treatment for medulloblastoma is by surgical resection, with adjuvant craniospinal radiotherapy for those over 3 years of age (as the infant brain is more susceptible to radiation effects) and chemotherapy.The 5-year survival varies from 50% to 80%. Favourable prognostic factors include complete surgical resection, lack of CSF dissemination at presentation, onset in the second decade, female gender and lateral location within the cerebellar hemisphere. Surveillance imaging detects recurrences earlier than clinical presentation, allows earlier therapeutic intervention, and correlates with increased survival.

•

PAEDIATRIC NEURORADIOLOGY

There remain regional differences in outcome in the UK, emphasizing the importance of managing children with these tumours in a paediatric neuro-oncology centre with multidisciplinary support of neurosurgery, radiology and pathology. Atypical teratoid/rhabdoid tumours are unusual malignant tumours with a poor prognosis. For practical purposes, the imaging features are indistinguishable from medulloblastoma/ primitive neuroectodermal tumour (PNET). They are more aggressive tumours, are often large at the time of presentation and occur in slightly younger children, typically under the age of 2 years. The next most common cerebellar tumour in children after medulloblastoma, and accounting for 30–40%, is the cerebellar low-grade astrocytoma (CLGA), which in most cases (85%) is a pilocytic tumour (WHO Grade I) and in up to 15% is a more diffuse fibrillary type with a higher histological grade. The 5-year survival for cerebellar pilocytic astrocytoma is in excess of 95% in most reported series, including the original description by Cushing in 1931. The CLGA is a well-circumscribed, slowly growing lesion of children and young adults, though it is occasionally seen in older people.There is an association with NF1 as there is for other neuro-axis pilocytic astrocytomas. The duration of symptom onset is more insidious than that of medulloblastomas, typically being intermittent over several months. On CT and MRI the tumour is typically a cerebellar vermian or hemispheric tumour which is cystic with an enhancing mural nodule.The solid component is hypointense to isodense on CT, hyperintense on T2-weighted FSE and hypointense on T1-weighted sequences reflecting the hypocellular and loosely arranged tumoral architecture62. The solid component is highly vascular with a deficient blood–brain barrier and therefore enhances avidly and homogeneously (Fig. 70.46). Figure 70.46 Cerebellar hemispheric tumour in a child with a history of ataxia, nausea, and vomiting over several months. (A,B,D,E) Axial T2, coronal FLAIR, coronal and sagittal T1 enhanced MRI show a left cerebellar hemispheric tumour with a large cystic component and solid homogeneously enhancing component which is bright on T2-weighted sequences (compare with the images of posterior fossa medulloblastoma, Figures 70.44 and 70.45). The solid component is not restricted on the diffusion-weighted image (C) and ADC map (F) compared to medulloblastoma and there is free diffusion in the cystic component.

1679

1680

SECTION 8

•

PAEDIATRIC IMAGING

Occasionally the pilocytic astrocytoma may present with diffuse nodular enhancement of the leptomeninges, indicating intracranial or intraspinal pial dissemination.This is typically seen with WHO Grade I tumours, does not imply a higher grade tumour and like the tumour primary tends to grow slowly. Ependymomas (WHO Grade II) account for approximately 10% of paediatric posterior fossa tumours and the posterior fossa is the most common site for ependymomas in children. Their mean age at presentation is 6.4 years, with a range of 2 months to 16 years. They are well-defined tumours which typically originate from the floor or the roof of the fourth ventricle, extend into the cerebellopontine angle and extrude through the foramina of Luschka and Magendie (Fig. 70.47). Perivascular pseudorosettes and ependymal rosettes are the cardinal histopathological features. They are more cellular than CLGAs but usually demonstrate CT and MRI features consistent with a higher water content than medulloblastomas. Therefore they are still hypodense or isodense on CT, hypointense on T1 and isodense to hyperintense on MRI. Foci of microcystic change, haemorrhage and calcification are common features. They can also disseminate throughout the neuro-axis by leptomeningeal spread, although this is much less common then for medulloblastoma. Incomplete resection, age under 3 years and anaplastic histopathological features correlate with a worse prognosis. The 5-year progression-free survival is 50% but worse under the age of 2 years. Children with von Hippel–Lindau disease may occasionally present with cerebellar haemangioblastoma (WHO Grade I), which in the sporadic form is usually a tumour of adults. It consists of a rich capillary network in addition to large vacuolated stromal cells. On imaging haemangioblastomas may mimic a CLGA with an intensely enhancing mural nodule and a cystic component. The tumour abuts the pial surface but unlike the CLGA it may be associated with prominent vascular flow voids. Haemorrhage and frank necrosis may occur but are less common. Most brainstem tumours in children are astrocytomas. Medullary tumours may present with symptoms and signs of raised cranial pressure and cranial nerve dysfunction. The two major groups are diffuse tumours and focally exophytic tumours63.

Figure 70.47 Ependymoma. (A–C) Axial T2, enhanced T1, and coronal FLAIR images showing a solid and microcystic fourth ventricular tumour extending out through the foramina of Luschka, Magendie, and the foramen magnum (arrows), the typical features of an ependymoma.

Diffuse tumours extend up and down the brainstem and are seen best as ill-defined signal hyperintensity on T2-weighted (including FLAIR) images in association with expansion of the brainstem. Their enhancement if present is usually minimal unless the tumour has been irradiated (Fig. 70.48). Focal exophytic tumours are usually dorsally exophytic, well-defined tumours, which do not extend along white matter tracts in the same ways as diffuse astrocytomas (hence their exophytic growth), but often enhance. They are usually Grade I pilocytic or Grade II astrocytomas and are associated with a better prognosis than diffuse astrocytomas, particularly focal tumours within the midbrain tectum64. Occasionally, however, anaplastic gangliogliomas or PNET tumours may appear like this.

Suprasellar tumours Some knowledge of the typical clinical presentation of certain suprasellar tumours can be very helpful in differentiating them even before imaging. For example, hypopituitarism is more likely to be seen with craniopharyngioma, delayed puberty with hypothalamic astrocytoma or Langerhans’ cell histiocytosis, and occasionally pituitary adenoma, precocious puberty due to hypothalamic infundibular lesions, such as Langerhans’ cell histiocytosis, germinoma, craniopharyngioma, hamartoma and non-neoplastic granulomatous disease, such as sarcoidosis and tuberculosis. Large suprasellar mass lesions have the ability to produce hydrocephalus by obstruction at the foramen of Monro, and visual field defects by compression or involvement of the optic chiasm.

Craniopharyngioma Although histologically benign (WHO Grade I), this tumour is associated with significant morbidity and mortality because of its site and often large size at presentation. It is the most common suprasellar tumour in children, accounting for 1–3% of intracranial tumours of all ages, but usually occurring from age 5 to 15 years. Most craniopharyngiomas have a suprasellar mass and a smaller intrasellar component. In 5% of cases it is purely intrasellar and may be difficult to distinguish from a Rathke’s cleft cyst. Classically it appears as a calcified, mixed

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

cally as appearing like ‘machine oil’. The cystic components are of increased signal on T2-weighted FSE. MR spectroscopy demonstrates high lipid peaks. Surgical resection is often incomplete (in > 20%) because of the close adherence of the tumour to the optic chiasm; despite this long-term survival is good (> 90% in children).

Hypothalmic–optic pathway glioma

Figure 70.48 Diffuse brainstem astrocytoma. Note the mass effect within the pons distorting the fourth ventricle and the encasement of the basilar artery (arrow).

Table 70.3 LESIONS

DIFFERENTIAL OF ENHANCING INFUNDIBULAR

Germinoma Langerhans’ cell histiocytosis Lymphocytic hypophysitis Granuloma (tuberculosis, sarcoid)

cystic and solid tumour with enhancement of the solid component and the cyst wall (Fig. 70.49). Large tumours may cause hydrocephalus and compression and distortion of the optic chiasm. The cystic components may extend behind the clivus and into any of the cranial fossae. On T1-weighted sequences the cyst may demonstrate T1 shortening due to proteinacaeous components, which have been described macroscopi-

Astrocytomas of the optic chiasm and hypothalamus account for 10–15% of supratentorial tumours in children. Optic chiasm tumours often extend into the hypothalamus and vice versa; hence they are discussed here together. Optic nerve tumours are usually pilocytic astrocytomas with a very indolent course, while hypothalamic/chiasmatic tumours may be of higher histological grade and more aggressive biological behaviour. CT and MRI both define involvement of the optic nerves. CT can detect expansion of the optic canal. However, MRI is the best modality for delineating expansion of the chiasm and hypothalamus and involvement of the posterior visual pathways (Fig. 70.50). Tumour appears isointense to hypointense on T1-weighted imaging and hyperintense on T2-weighted imaging.There may be diffuse fusiform expansion of the nerve from subarachnoid dissemination of tumour around the optic nerve. Enhancement is variable. The main differential diagnosis is from craniopharyngioma, which tends to present later, is usually calcified and is adherent to the chiasm rather than arising from it and causing expansion.

Infundibular tumours These include germinomas and Langerhans’ cell histiocytosis, both of which cause expansion and enhancement of the pituitary infundibulum (Fig. 70.51). Onset of diabetes insipidus appears to correlate with absence of high T1 signal in the posterior pituitary Figure 70.49 Craniopharyngiomas in two children. (A–C) The first child has a large suprasellar, pre-pontine and middle cranial fossa tumour which is causing considerable mass effect on the brainstem and is encasing the basilar artery (arrowheads). There are solid enhancing and calcified components (arrows). The cystic components are of higher density on CT and there is T1 shortening on MRI in keeping with proteinacous contents. (D–F) The second child has a smaller suprasellar lesion, which is also calcified (arrowhead). The optic chiasm is clearly separate from the lesion and is draped over the top (E,F).

1681

1682

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.52 Hypothalamic hamartoma. Coronal T1 and sagittal T1-weighted post-contrast MRI shows a nonenhancing lesion arising from the floor of the third ventricle posterior to the pituitary infundibulum and projecting inferiorly into the suprasellar cistern (arrows).

Figure 70.50 Hypothalmic–optic pathway glioma. (A) Coronal T1-weighted, (B) sagittal, and (C) coronal enhanced T1-weighted MRI, and (D) coronal FLAIR show an optic chiasm glioma. The chiasm is not identified separately from the tumour.

Figure 70.51 Langerhans’ cell histiocytosis. (A) Axial T2-weighted image. (B) Sagittal T1-weighted post-contrast image. Suprasellar T2 hypointense and enhancing mass (arrows) with associated oedema extending superiorly along white matter tracts in a child with multisystem Langerhans’ cell histiocytosis.

gland. The hypothalamic hamartoma (not a tumour by definition) presents either with precocious puberty or with gelastic seizures.These may be sessile or pedunculated and are well-defined lesions arising from the floor of the third ventricle and extending inferiorly into the suprasellar or interpeduncular cistern. They are typically isointense to grey matter on T1- and T2-weighted MRI and do not enhance (Fig. 70.52).

Pituitary tumours Pituitary adenomas are uncommon in children but may present in adolescents, and account for 2% of all pituitary adenomas.The most common functional tumours are prolactinomas and corticotrophin- and growth hormone-secreting tumours. A quarter of paediatric pituitary tumours are nonfunctioning.

The imaging appearances are the same as in adults. Rathke’s cleft cysts are also rare in children.

Pineal region tumours The pineal region is a descriptive term and encompasses the posterior third ventricle and its contents, including the pineal gland itself, the tectal plate and aqueduct, the posterior septum pellucidum, corpus callosum and thalami, internal cerebral veins in addition to the quadrigeminal cistern containing the posterior cerebral arteries, vein of Galen and straight sinus. Lesions may arise from any of these components and therefore the differential diagnosis of a pineal region tumour is wide. The most common lesions are germ cell tumours (GCTs), followed by primary pineal gland masses. Gliomas are also relatively common lesions at this site, usually derived from adjacent brain parenchyma. Pineal region tumours can cause hydrocephalus by obstruction of the cerebral aqueduct. Direct compression or invasion of the tectal plate, specifically the colliculi, may cause failure of upward gaze and convergence (Parinaud’s syndrome). They may also cause precocious puberty.

Central nervous system germ cell tumours CNS GCTs are primarily tumours of the young, over 90% occurring in the under 20 age group and with a peak incidence from age 10 to 12 years.They are more common in Asia but in the West account for 1% of intracranial neoplasms in children. Germinoma is the most common type of CNS GCT, followed by nonsecreting teratoma. Other GCTs include rarer secreting forms such as yolk sac, embryonal and choriocarcinoma, and may be of mixed cellular types, the latter associated with the worst prognosis. Germinomas are characteristically found in the midline, in the pineal or suprasellar regions. A minority of lesions may be found in the basal ganglia, thalami, or cerebellum. Pineal and suprasellar lesions may be synchronous, and when so, are pathognomonic. Most pineal region GCTs occur in boys and suprasellar GCTs in girls. Pathologically the tumour is solid and consists of large, glycogen-rich germ line cells with variable desmoplastic stroma and lymphocytic infiltrate. Necrosis

CHAPTER 70

and haemorrhage are unusual with the exception of lesions seen in the basal ganglia. Germinomas are classified as malignant tumours but respond extremely well to radiotherapy and may melt away over just a few days of treatment. Overall they are more common in young adolescent males. On CT they appear as a hyperdense, solid mass within the posterior third ventricle and enhance avidly and homogeneously. They tend to engulf the pineal gland which may be calcified. Occasionally this may be difficult to differentiate from intrinsic tumoral calcification more suggestive of a teratoma or primary pineal gland tumour, such as a pineoblastoma, although both of these lesions are typically more heterogeneous with haemorrhage and cyst formation. On MRI the cellularity of the tumour is reflected by T2 hypointensity relative to grey matter. The tumour may be less homogeneous than on CT and cystic change may be detected, although enhancement remains a marked feature. Germinomas demonstrate restricted diffusion. Synchronous germinomas in typical midline sites, such as the suprasellar region, and evidence of early CSF dissemination should be actively looked for. Benign teratomas are very heterogeneous, mixed cystic, solid and well-defined masses characterized by calcification and fat. Enhancement is not usually seen unless there are areas of malignant degeneration. Other GCTs are also heterogeneous in appearance and do not contain fat, but the individual tumour types are not distinguishable radiologically.

Primary pineal tumours: pineoblastoma and pineocytoma Pineoblastomas are malignant (WHO Grade IV), small, round cell tumours which histologically are similar to medulloblastomas and share similar imaging characteristics in terms of hyperdensity on CT and are hypointense to isointense signal on T2-weighted FSE relative to grey matter. They may contain areas of calcification and rarely haemorrhage. The solid parts of the tumour enhance intensely. They may be distinguished from Grade II pineocytomas by the age at presentation, as they occur most frequently in the first two decades of life compared to pineocytomas which are tumours of young adults; by their size (> 3 cm); and by their relative T2-weighted hypointensity (Fig. 70.53).

•

PAEDIATRIC NEURORADIOLOGY

Supratentorial hemisphere tumours Overall supratentorial tumours occur as frequently as posterior fossa tumours in the paediatric age group but are more common under the age of 2 and over the age of 10, while posterior fossa tumours are more common from the ages of 2 to 10.The presenting symptoms are due to a large mass-occupying lesion and include headaches and vomiting. In the under 2 year old age group the tumour may be very large at the time of presentation; the fontanelles are open so that infants present with increasing head size as often as they do with hydrocephalus. Seizures are seen more particularly with temporal and frontal cortex lesions.

Astrocytomas The most common paediatric cerebral hemispheric tumour is the astrocytoma, accounting for 30% of supratentorial brain tumours in children. In children under 2 years, other diagnoses should be considered; teratomas and desmoplastic infantile gangliogliomas are seen in neonates and PNET, atypical teratoid/rhabdoid tumours, and ependymomas are seen in slightly older infants. Some children who present with a longer history and refractory epilepsy may have low-grade, indolent glioneuronal tumours, including the dysembryoplastic neuroepithelial tumour. As with cerebellar astrocytomas, hemispheric astrocytomas may be cystic with an enhancing mural nodule, entirely solid with variable enhancement or solid with a necrotic centre. On CT the solid part of the hemispheric astrocytoma is isodense or hypodense and on MRI T2-weighted sequences it is hyperintense, helping to distinguish these tumours from small, round cell tumours such as supratentorial PNET. They may rarely be multicentric. The histological grade cannot be reliably determined by the radiological features. Contrast enhancement can be seen in both low- and high-grade tumours. A simple cyst with nonenhancing walls, minimal surrounding oedema and a single enhancing mural nodule is more likely to be a pilocytic astrocytoma. Lesions containing haemorrhage and associated with marked adjacent oedema are more likely to be of higher grade. Glioblastoma is seen in children and in children is associated with a better prognosis than in adults. The pleomorphic

Figure 70.53 Pineoblastoma. (A) Axial, (B) sagittal T2-weighted fast spin-echo, and (C) enhanced sagittal T1-weighted MRI in a 2 year old girl show a pineal region tumour effacing the tectal plate. The hydrocephalus is treated by a frontal extraventricular drain (track through the genu of the corpus callosum marked by the arrows). The tumour is hypointense on the T2-weighted images with rings of lower signal consistent with calcification and haemorrhagic products, and peripheral rim enhancement, and was confirmed as a pineoblastoma.

1683

1684

SECTION 8

•

PAEDIATRIC IMAGING

xanthoastrocytoma is an uncommon astrocytoma of children and young adults. It is usually a WHO Grade II lesion, though may have anaplastic features.Typically it has a superficial location within the temporal lobe and consists of a cyst and enhancing mural nodule with minimal adjacent oedema. Radiologically it is often indistinguishable from ganglioglioma and some other glioneuronal tumour subtypes.

Ependymonas Supratentorial ependymomas occur more commonly in boys with a peak incidence between the ages of 1 and 5 years, and again are similar to the posterior fossa ependymoma. However they are more usually extraventricular in site and therefore CSF dissemination is less common than with their posterior fossa equivalent.They are isodense to hyperdense, well-defined lesions on CT with variable enhancement of the solid component. Tumour heterogeneity with cystic areas and foci of calcification are common (Fig. 70.54). Haemorrhage may also be seen.

Primitive neuroectodermal tumours PNET (WHO Grade IV) is a cellular tumour of primitive cell types and may be seen in children as young as 4 weeks to 10 years with a mean age of 5.5 years. A rare subtype that includes more neuronal cells is known as cerebral neuroblastoma. PNET is found more frequently in the posterior fossa than in the supratentorial compartment. Poor prognostic factors for PNET include age under 2 years and a supratentorial location. The 5-year survival drops from 85% for tumours in the posterior fossa to 34% for supratentorial tumours. They are frequently haemorrhagic, necrotic and have foci of calcification. The solid parts of the tumour are hyperdense on CT

and hypointense on T2-weighted FSE. There is always some enhancement although the degree is variable. Although the tumour may appear well-defined radiologically, tumour cells are likely to extend beyond the apparent margins. Widespread dissemination of tumour both through the CSF space and to the lungs, liver and spleen, is not infrequent. Medulloepithelioma is a rare, highly malignant tumour usually affecting children between the ages of 6 months and 5 years. Solid components are hypercellular; therefore, the density and signal characteristics may be similar to PNET.They are large at the time of presentation with extensive areas of haemorrhage and necrosis but may be distinguished from other tumours, including PNET, by their lack of contrast enhancement.

Desmoplastic infantile gangliomas These typically present as a massive cyst with an enhancing cortically-based mural nodule in an infant with increased head size and bulging fontanelle. Adjacent dural enhancement may also be seen.The cyst, which is hypointense on T1, hyperintense on T2 and may contain septations, usually does not enhance.

Choroid plexus tumours These are ‘cauliflower-like’ tumours arising from the epithelium of the choroid plexus and are more frequently benign (papilloma) than malignant (carcinoma). Both types may disseminate within the CSF space. They are the most common brain tumour in children under 1 year and present with hydrocephalus, possibly due to CSF hypersecretion or obstructive hydrocephalus due to haemorrhage, arachnoiditis or carcinomatosis in carcinomas. For papillomas the 5-year survival approaches 100% while for carcinomas it ranges from 25% to 40% with a higher survival if the tumour is completely resected. On CT they are seen as hyperdense or isodense, lobulated ‘frond-like’, avidly and homogeneously enhancing masses with punctate calcifications, occasionally with haemorrhage and with hydrocephalus.The typical site is the trigone of the lateral ventricle, while in older children the cerebellopontine angle or fourth ventricle may be involved (Fig. 70.55). On MRI the papillary appearance is more readily appreciated: they are more mottled and isointense or hypointense on T1-weighted imaging with intense enhancement. Vascular flow voids, usually from choroidal arteries, are often seen in association, and arterial embolization may be considered before surgery in an attempt to reduce the vascularity of the tumour. Haemorrhage and localized vasogenic oedema are suggestive of carcinoma with invasion but the two histological types cannot be reliably distinguished on imaging. Other intraventricular tumours include meningiomas, which are rare in children outside NF2 and ependymomas.

Dysembryoplastic neuroepithelial tumours Figure 70.54 Supratentorial grade II ependymoma. The left cerebral hemisphere tumour extends across the midline into the right ventricle. (A) CT (post partial debulking) shows it is heavily calcified and (B–D) MRI (axial T2 enhanced, sagittal T1 and axial T1-weighted MRI ) show a mixed cystic and solid heterogeneously enhancing tumour. There is associated hydrocephalus.

Dysembryoplastic neuroepithelial tumour (DNT) is a WHO Grade I benign tumour which classically presents with complex partial seizures in children and young adults under the age of 20. It is a cortically based lesion which may have associated foci of cortical dysplasia.

CHAPTER 70

Figure 70.55 Choroid plexus tumours. (A,B) Axial T2, enhanced sagittal T1-weighted MRI in a boy aged 6 months demonstrate a lobulated homogeneously enhancing intraventricular tumour with relative T2 hypointensity, in keeping with a highly cellular tumour closely related to the choroid plexus. The ependyma is enhancing and the interface between tumour and adjacent brain is indistinct. Histology confirmed the tumour to be a choroid plexus carcinoma and this was subsequently embolized before surgery. There is a communicating hydrocephalus which may be due to increased cerebrospinal fluid production or to proteinaceous/haemorrhagic exudate. (C,D) A different child with choroid plexus papilloma. Choroid plexus tumours although commonly seen within the trigone of the lateral ventricle may arise from anywhere within the ventricular system. This child has a frondy choroid plexus papilloma arising within the third ventricle and extending superiorly through the foramina of Monro. In this case the hydrocephalus was obstructive.

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.56 Dysembryoplastic neuroepithelial tumour (DNT) in a child with long-standing refractory focal epilepsy referred to the epilepsy surgical programme. Several MRI examinations over 3 years had shown no change in the appearances of this right inferior parietal lobe lesion. The lesion is well-defined, cortically based, extends towards the ventricular margin, and has the typical lobulated internal architecture of DNT.

children and a single episode may be multifactorial. Associated factors, some of which may be amenable to treatment, include congenital heart disease, anaemias, prothrombotic disorders such as protein C and S deficiencies, hyperhomocystinaemia, lipid abnormalities, recent infections and respiratory chain disorders (Fig. 70.57). Equally, long-term follow-up has shown a poor outcome with some degree of dependency in

On imaging it appears as a well-defined cortically based lesion with a characteristic ‘bubbly’ internal structure, minimal mass effect and no associated vasogenic oedema (Fig. 70.56). There may be some adjacent bony scalloping consistent with a long-standing lesion and a third of lesions demonstrate calcification. On MRI they are hypointense on T1 and have a hyperintense rim on FLAIR or proton density-weighted imaging. Most tumours do not enhance and if present, enhancement is faint and patchy.

CEREBROVASCULAR DISEASE AND STROKE Stroke is an important paediatric illness with an incidence of around 2 cases per 100,000 children per annum. The aetiology of paediatric stroke is significantly different from adult stroke, as large artery atherosclerosis, cardio-embolic and small vessel disease combined account for only 10% of cases in children65. There have been several misconceptions regarding stroke in children. Paediatric stroke was said to be idiopathic, associated with a good prognosis with low recurrence rates and good recovery of motor function and school performance, and was minimally investigated on the assumption that this would not affect management. Recent work has shown that there are many different aetiologies associated with stroke in

Figure 70.57 Strokes occurring in mitochondrial cytopathy. Bilateral symmetrical lentiform and caudate calcification and extensive cerebral infarction crossing arterial territories in a child with a mitochondrial disorder (MELAS).

1685

1686

SECTION 8

•

PAEDIATRIC IMAGING

60% of affected children. Although difficult to quantify due to selection bias, the risk of recurrence for ischaemic stroke ranges from 5% to 20%. Children who have a stroke at a young age have a worse physical and intellectual outcome and behavioural problems may be significant. Hence children with acute stroke should be referred to, or have their management discussed with a paediatric neurologist, and thoroughly investigated on each occasion to detect all potential risk factors. Cross-sectional brain imaging is mandatory in children presenting with clinical stroke. Brain MRI is recommended and should be performed as soon as possible after presentation. If MRI is not available within 48 h, CT is an acceptable initial alternative. Imaging should be undertaken urgently in those who have a depressed conscious level or whose clinical status is deteriorating66. The purpose of neuroimaging is to confirm the diagnosis and exclude alternative treatable lesions, as well as to help understand the underlying cause, guide treatment and monitor progression of the disease. The radiological findings of brain parenchymal involvement in stroke are not significantly different from findings in adults. There is however some evidence that restricted diffusion may occur for a shorter time in younger children and pseudonormalization may occur earlier than in adults. Vascular abnormalities in paediatric ischaemic stroke are common67. These are more frequently intracranial than extracranial, involve the anterior rather than posterior circulation and typically consist of occlusion of proximal large arteries, i.e. middle cerebral artery (MCA), anterior cerebral artery (ACA) and terminal internal carotid artery (ICA). Sickle cell disease, a chronic haemolytic anaemia in which abnormal haemoglobin (HbS) forms with a tendency to cause red blood cells to distort and block small blood vessels, is the most common cause of ischaemic stroke in children worldwide. Even within this single disease entity there are several factors that contribute to stroke. These include a cerebral vasculopathy known as moya moya disease, an underlying predilection to infection, tissue hypoxia and precipitation of sickling vasoocclusive crisis resulting from chronic anaemia, high white cell count, adenotonsillar hyperplasia causing obstructive sleep apnoea, cardiomegaly and a generalized procoagulant state. Stroke in sickle cell disease is clinically apparent in 9% of children with sickle cell disease under the age of 20 years, but silent infarction occurs in as many as 25%. The imaging pattern of infarction is typically of arterial watershed infarction between the major cerebral arterial territories. A fifth of strokes in sickle cell disease may be haemorrhagic. The diploic space of the calvarium may be diffusely thickened due to increased haematopoiesis in order to compensate for the chronic haemolytic anaemia. The brain MRI and circle of Willis MR angiogram (MRA) may show evidence of a cerebral vasculopathy known as moya moya syndrome. In this there is typically progressive stenosis of the terminal ICA and proximal segments of the major intracranial arteries (Fig. 70.58). There is a predilection for the anterior circulation. As the stenosis progresses, increased flow occurs through proximal collateral vessels, particularly the lenticulostriate and thalamoperforator arteries, resulting in the

Figure 70.58 Sickle cell disease and moya moya syndrome. Child with extensive frontal, deep and posterior watershed infarction (A). (B) Shows extensive perimesencephalic ‘moya moya’ collaterals (arrow) and attenuated right middle cerebral artery (MCA) flow voids. (C) Compressed maximum intensity projection image shows narrowed terminal internal carotid artery (ICA), reduced filling of right MCA, and A1 segment of the anterior cerebral artery. There is an aneurysm at the A1/anterior communicating artery (ACOM) junction (arrow).

‘puff of smoke’ appearance to which moya moya refers. These are seen as multiple small flow voids within the basal ganglia. Prominent transmedullary veins and pial enhancement may also be seen. Other acquired vascular abnormalities include aneurysms and small arteriovenous malformations. Referral to a neurosurgical paediatric centre for consideration of external-to-internal carotid (EC–IC) bypass may be indicated. At this point cerebral perfusion imaging to detect ‘diffusion–perfusion mismatch’ may be helpful in order to assess for critical ischaemia or to select which hemisphere should be revascularized first. This involves a bolus of intravenous contrast medium for MRI perfusion studies and it is important that the child with sickle cell disease is well hydrated for this. Surgical EC–IC bypass may be performed indirectly by mobilization of part of the temporalis muscle with its blood supply, the superficial temporal artery (STA) and laying it onto the pial surface of the brain, or by direct end-to-side anastomosis between the STA and distal MCA branch. Some centres use multiple calvarial burr holes alone to promote superficial angiogenesis and collateral revascularization. Post-operative MRA and perfusion imaging after successful EC–IC bypass should show flow through the STA and increased collateral flow within the distal MCA branches. Perfusion imaging will show increased cerebral blood volume and flow to the revascularized hemisphere. Moya moya accounts for up to 30% of cerebral vasculopathy in paediatric stroke but it is not unique to sickle cell disease, being also idiopathic, secondary to NF1, cranial irradiation, Down’s syndrome, human immunodeficiency virus (HIV) and even tuberculous meningitis. Postinfective angiitis associated with varicella zoster, in which the terminal ICA and proximal MCA are usually affected and there is infarction of the basal

CHAPTER 70

ganglia, is relatively common (Fig. 70.59). MRA in the majority (but not all) shows evidence of stabilization or remodelling and improvement of the angiographic appearances at 6-month follow-up. Arterial dissection may occur intracranially but is also seen, as it is in young adults especially, not infrequently within the cervical arteries.Vertebral arterial dissection occurs most commonly as the vertebral artery exits the transverse foramen of C2 before passing posterolaterally over the lateral masses of C1 to enter the foramen magnum, and ICA dissection occurs above the bifurcation (Fig. 70.59C,D). Dissections may involve intracranial and extracranial arteries. There may be other rarer causes of cervical arterial disease with more diffuse involvement proximal to the bifurcation, e.g. that seen with the connective tissue disorders such as Marfan’s, Ehlers–Danlos type IV, osteogenesis imperfecta type I, autosomal dominant polycystic kidney disease, fibromuscular dysplasia, or other causes of cystic medial necrosis and Menke’s disease. Cervical arterial dissection may occur without an antecedent history of trauma. As this is a potentially treatable cause of stroke68, we advocate in all paediatric stroke noninvasive imaging by MRA not only of the intracranial circle of Willis but also of the entire great arteries of the neck from their origins at the aortic arch to their intracranial terminations. Radiological criteria for the diagnosis of dissection are visualization of an intimal flap or a double lumen in the wall of the artery. These pathognomonic signs are detected in fewer than 10% of adult dissections by catheter angiography which remains the gold standard for diagnosis. However, tapering arterial occlusion, the ‘string sign’ or ‘rat’s

•

PAEDIATRIC NEURORADIOLOGY

tail’ appearance (Fig. 70.59C,D), aneurysmal dilatation of the artery and eccentric mural thrombus are other less specific signs. Vein of Galen aneurysmal malformations (VGAMs) are a rare cause of paediatric stroke but have a wide range of clinical presentations and account for 30% of vascular malformations in children. Their recognition is important as there is increasing evidence that appropriate endovascular treatment by neuroradiologists with specialist skill in dealing with paediatric high flow AV shunts in a multidisciplinary setting with access to neonatologists, anaesthetists and neurologists is associated with improved outcome. VGAMs are unique congenital malformations of the intracranial circulation characterized by an enlarged midline venous structure, a persistent embryological remnant, with multiple arteriovenous communications leading to aneurysmal dilatation, presumably secondary to high arteriovenous flow. Neonates may present with severe cardiac failure. In infants and children who present with VGAMs the degree of shunting is much smaller and they may have evidence of hydrocephalus with cerebral atrophy. Older children may have headaches and seizures, and also are more likely to develop intracerebral or subarachnoid haemorrhage. MRI will demonstrate the dilated venous sac, location of fistulous connections and the arteries involved, venous drainage and any evidence of thrombus within the venous sac (Fig. 70.60). MRI can determine the extent of parenchymal damage, including focal infarctions and generalized cerebral atrophy, although CT is more sensitive for the parenchymal calcification which accompanies chronic venous hypertension. Angiography is the gold standard of diagnosis and ideally should be performed at the time of endovascular treatment. In imaging follow-up, evidence of significant arteriovenous shunting, progressive cerebral damage/atrophy and jugular vein occlusion as a chronic effect of venous hypertension should be sought.

HYPOXIC–ISCHAEMIC INJURY IN THE DEVELOPING BRAIN

Figure 70.59 Vasculitis. (A,B) In a child with chicken pox vasculitis, axial T2 and maximum intensity projection (MIP) image of time of flight (TOF) circle of Willis MRA show a mature proximal left middle cerebral artery branch infarct with reduced and turbulent flow (arrow) and reduced distal filling. Arterial dissection. (C,D) In a 14 year old girl with acute onset left hemiplegia, large right middle cerebral artery territory infarct, and internal carotid artery (ICA) dissection is seen. On the axial T2-weighted image there is an eccentric filling defect within the lumen of the right internal carotid artery which is narrowed, confirming the presence of dissection (arrow). The MIP image of the TOF MRA shows a typical rat’s tail appearance of a tapering stenosis distal to the right ICA bifurcation (arrow).

Well-recognized patterns of brain injury have been attributed to hypoxic–ischaemic injury and are believed to vary according to the nature and severity of the insult and the degree of maturity of the developing brain69–72. The term partial hypoxic–ischaemic injury is used to describe an episode or episodes of hypoxia or hypoperfusion to the developing brain, whilst profound hypoxic–ischaemic injury is used to describe a briefer episode of anoxia or circulatory arrest. Injuries occurring in the first and early part of the second trimester of pregnancy are expected to result in brain malformations and will not be discussed further in this section. Injuries occurring later will be discussed below.

Preterm patterns The so-called ‘preterm’ patterns of hypoxic–ischaemic injury tend to be seen in brains of about 20–35 weeks gestational age and are characterized clinically by a neonatal encephalopathy. Few survive a profound hypoxic–ischaemic injury, but if they do, the pattern of injury appears predominantly to affect

1687

1688

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.60 Vein of Galen malformation (partly occluded by embolization glue) showing residual flow through the promesencephalic vein, containing a combination of glue and thrombus, via the falcine sinus (arrow) towards the venous confluence. Arterial supply is via a number of choroidal vessels.

the thalami with relative sparing of the other deep grey matter structures. Partial hypoxic–ischaemic injury is believed to result in the most common pattern seen in this age group, which is that of periventricular leukomalacia (PVL), germinal matrix or periventricular haemorrhage and intraventricular haemorrhage (GM/IVH), also described as periventricular haemorrhagic infarction (PVHI)71,72. In extreme cases, cystic encephalomalacia may be seen. Outcome is determined by the degree of brain injury and is also influenced by any complications and the effectiveness of any intervention, such as CSF diversion procedures for hydrocephalus. The physiological conditions necessary for the development of PVL are thought to be present from 25 to 34 weeks gestational age, the condition being most frequent in the older group, 30–33 weeks gestational age. The precise mechanisms responsible for these lesions are not fully understood, but this condition is usually considered to be a complication of prematurity and is probably multifactorial. The clinical picture is spastic diplegia or quadriplegia, often with visual impairment. Mental retardation is usually absent or mild, except in very severe cases, as are seizures73. The simplest theory suggests that there is cerebral hypoperfusion and hypoxia causing ischaemic infarction, which in 20% may be complicated by secondary haemorrhage following reperfusion of the damaged areas. The parts of the immature brain most sensitive to insufficient cerebral perfusion or hypoxia are found in the periventricular white matter, which is therefore the most common location of PVL. Factors such as respiratory problems, sepsis, necrotizing enterocolitis, feto-maternal haemorrhage, or hypoglycaemia are associated with PVL. PVL can also be seen in mature newborns but the early stages of damage are not seen as the lesion occurs in utero and is well into the sequence of pathological development by birth at term. PVL results in infarction with oedema seen in the periventricular region. This may be seen as increased echogenicity on US. The damaged tissue undergoes cystic degeneration 10–20 d after the insult. Small, often confluent, cysts form in the periventricular white matter; these are usually transient and subsequently collapse. The detection of these cysts is the most reliable US finding of PVL in its early development (Fig. 70.61).As the cysts collapse, atrophy of the damaged brain tissue

Figure 70.61 Early sign of periventricular leukomalacia. (A) US shows periventricular echolucencies (arrows), one of the earliest signs of periventricular leukomalacia. (B,C) On T1-weighted MRI sequences, these are seen posteriorly in the peritrigonal area and are lined by small focal regions of T1 shortening in keeping with haemorrhage (arrows).

follows and this process is first detected by the demonstration of secondary ventricular dilatation; in more severe cases there is a more generalized loss of brain tissue, particularly white matter. Ventricular dilatation beyond normal limits is usually detectable by US or CT 4–8 weeks after the injury, depending on the severity of the lesions, and persists throughout life as permanent tissue loss. The features of end-stage PVL result from the decreased amount of periventricular white matter adjacent to the trigones. There is ventricular dilatation with irregular ventricular margins and the distribution is characteristically worst in the parieto-occipital regions with sparing of the frontal and temporal regions. Injury to the remaining white matter is more difficult to detect and MRI is most reliable in demonstrating these end-stage changes of PVL, 1–2 years after the injury, when the myelination process is complete or almost complete. MRI then shows abnormal signal in the remaining periventricular white matter (Fig. 70.62).

CHAPTER 70

•

PAEDIATRIC NEURORADIOLOGY

which may communicate with the ventricle or focal dilatations of the ventricles.

Term patterns

Figure 70.62 End-stage changes of periventricular leukomalacia. A child with spastic diplegia scanned much later in life (15 years) has the typical chronic changes of periventricular leukomalacia. There is posterior periventricular increased signal on T2-weighted images and enlargement of the ventricles posteriorly with irregular, scalloped margins, indicating white matter loss. The corpus callosum seen on the sagittal T1-weighted MRI is markedly thinned, particularly affecting the posterior body.

GM/IVH is also common in the immature brain but is possibly less significant in terms of subsequent handicap. The cause of IVH is also thought to be fluctuations in cerebral perfusion. The germinal matrix is in a process of involution after 24 weeks gestational age and its fragile vessels rupture easily. Its proximity to the lateral ventricle, from which it is separated only by ependyma, frequently results in rupture of haemorrhage into the ventricular system. In some immature neonates with IVH, the amount of blood is excessive and it dilates the ventricle, causing congestion in the periventricular white matter, venous infarction and secondary haemorrhage. The lesions are often unilateral and anterior, and tend to occur in the group of neonates younger than 30 weeks gestational age.The findings are well seen on US which is used for grading the severity of disease. Later, resolution of the parenchymal haemorrhage results in either paraventricular cavities

The ‘term’ patterns of hypoxic–ischaemic injury tend to be seen in brains of about 36–42 weeks gestational age at the time of the insult. The pattern that is attributed to profound hypoxic–ischaemic injury characteristically affects the brain regions that are most metabolically active and therefore most selectively vulnerable at the time of insult. These are the posterolateral putamina, ventrolateral thalami and adjacent capsular white matter. The hippocampi, peri-Rolandic (motor and sensory) cortex and visual cortex are also often affected, and the changes are typically bilateral and symmetrical. The cerebellar vermis is also recognized as selectively vulnerable in this context. This pattern is often matched with the clinical picture of dyskinetic or dystonic cerebral palsy70,72. The injuries attributed to partial hypoxic ischaemia are seen in a parasagittal distribution, typically involving a combination of cortex and subcortical white matter, and most often across the frontoparietal regions. Whilst usually bilateral, this pattern is not uncommonly asymmetric. A characteristic region of involvement is the posterior part of the Sylvian fissures. More characteristically, the greatest injury occurs at the base of the gyri, within the depths of the sulci, resulting in focal atrophy in these areas and a pattern recognized as ulegyria (Fig. 70.63). As with the preterm brain, more prolonged insults are thought to result in cystic encephalomalacia (Fig. 70.64). The predominant involvement of the cerebral hemipheres with relative sparing of the posterior fossa structures is a pattern that favours hypoxic–ischaemic injury over other causes of global brain injury at term, such as perinatal/neonatal infection. The common clinical sequelae of this type of injury are microcephaly with severe mental retardation and spastic quadriplegia which may be asymmetric74.

MISCELLANEOUS ACQUIRED TOXIC OR METABOLIC DISEASE Kernicterus is the result of the toxic effect of neonatal unconjugated hyperbilirubinaemia on the brain. The brain regions Figure 70.63 Hypoxic ischaemia at term, imaged in childhood. The gyri are thinner at their bases than at their apices. This is known as ulegyria and dates the hypoxic–ischaemic event to term. Note the relative preservation of the cerebellum and brainstem.

1689

1690

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.64 Prolonged hypoxic ischaemia resulting in multicystic encephalomalacia. There is cystic cavitation of most of the white matter with grossly thinned corpus callosum, leaving only a very thin rim of preserved cortical mantle.

that have selective susceptibility include the globus pallidi, subthalamic nuclei and hippocampi, as well as cranial nerves VIII, VII and III. Neonates with bilirubin encephalopathy have a depressed conscious level, hypotonia and seizures with opisthotonus. Delayed effects include extrapyramidal signs (athetosis), deafness, gaze palsies and developmental delay. In the acute stage there is signal abnormality with bilateral symmetrical T2 prolongation and swelling of the globus pallidi. There may be T1 shortening also. Eventually the pallidal changes will progress to atrophy with variable persistent gliotic change. US and CT are initially normal although later there may be evidence of calcification. Hypoglycaemia may also present with a nonspecific encephalopathy subsequently leading to seizures.This is a problem particularly with neonates who have immature enzyme pathways and relatively poor glycogen reserves, particularly in the context of increased requirements due to sepsis or associated hypoxia–ischaemia but may also be the end result of some inborn errors of metabolism or hyperinsulinism. Imaging studies may show evidence of a diffuse encephalopathy but the findings are typically most severe in the parieto-occipital regions. There is T1 and T2 prolongation with swelling affecting the cortex and subcortical white matter with variable restricted diffusion (Fig. 70.65), progressing to the chronic sequelae of cerebral infarction with evidence of gliosis, cavitation and atrophy. Pallidal damage may also occur. Hypernatraemia is most commonly seen in premature infants, particularly if there is additional dehydration, e.g. due to diarrhoea. Affected infants have a depressed conscious level and irritability. There is an osmotic water shift from the intracellular to the extracellular space resulting in interstitial oedema, manifest by T1 and T2 prolongation with increased diffusion, but also in parenchymal haemorrhage as the brain

Figure 70.65 Neonate with seizures and hypoglycaemia shows (A,B) increased signal within the parieto-occipital cortex and white matter, with patchy loss of the normal cortical low signal (arrow). (C) Diffusionweighted image and (D) ADC map show that the diffusion changes are a mixture of restricted (black arrows) and increased diffusion (white arrows). All of these changes, including the T2 and restricted areas of diffusion, completely resolved on follow-up.

shrinks and bridging dural veins are torn as they are pulled away from the calvarium. Toxic exposure should be considered in children, particularly adolescents, with acute neurological symptoms and bilateral symmetrical grey matter involvement. Toxins include toluene or other organic solvents, cyanide and carbon monoxide poisoning.

CHAPTER 70

INTRACRANIAL AND INTRASPINAL INFECTIONS Congenital infections (TORCH) These infections are acquired in utero or during passage through the birth canal; bacterial infections spread from the cervix to the amniotic fluid while toxoplasmosis, rubella, cytomegalovirus (CMV), syphilis and human immunodeficiency virus (HIV) spread via the transplacental route, and herpes simplex virus (HSV) is acquired from direct exposure to maternal type II herpetic genital lesions during delivery. The stage of brain development judged by the gestational age is more important than the actual organism in determining the pattern of CNS injury. Therefore in utero infections acquired before 16–18 weeks, when neurones are forming within the germinal matrix and migrating to form the cerebral cortex, produce lissencephaly and a small cerebellum. Spontaneous abortion is also a frequent outcome during this time. Between 18 and 24 weeks, when the cortical neurones are organizing but the immature brain is unable to mount an inflammatory response, the infective insult may produce localized dysplastic cortex and porencephaly. This is seen as a smooth-walled cavity isointense to CSF on all sequences in continuity with the ventricular system and without evidence of gliosis. From the third trimester onwards the insult results in asymmetrical cerebral damage with gliosis, cystic change and calcification. CMV is the most common cause of serious viral infection in fetuses and neonates in the West, occurring in up to 1% of all births. It is acquired transplacentally and the vertical transmission rate is 30–40%. The classical manifestations of CMV disease at birth include hepatosplenomegaly, petechiae, thrombocytopenia, microcephaly, chorioretinitis and sensorineural deafness occurring in up to 10% of CMV infection, but there is also an increased risk of developing deafness and other neurological deficits up to 2 years after exposure. The mechanism of injury may be due to a direct insult to the germinal matrix cells leading to periventricular calcification, cortical malformations with microcephaly and cerebellar hypoplasia, or due to the virus causing a vascular insult. Transfontanelle cranial US may demonstrate branching curvilinear hyperechogenicity in the basal ganglia, or ‘lenticulostriate vasculopathy’, which may also be seen with other congenital infections, hypoxic–ischaemia and trisomy 13 and 21. Infants affected in the second trimester have lissencephaly with a thin cortex, hypoplastic cerebellum, ventriculomegaly and periventricular calcification, which is more reliably detected on CT than MRI. Those injured later, probably during the period of cortical organization in the second trimester, have polymicrogyria, with less ventricular dilatation and cerebellar hypoplasia, and later infection produces parenchymal damage, large ventricles, calcification and haemorrhage without an underlying structural brain malformation. Temporal pole cysts are also a feature. Toxoplasmosis is a protozoan infection caused by ingestion by the mother of Toxoplasma gondii oocytes in undercooked meat. The transmission rate is high and increases from 30% at 6 months’ gestation to approaching 100% at term. The incidence of congenital infection is approximately 1 in 1000 to 1

•

PAEDIATRIC NEURORADIOLOGY

in 3400 but it accounts for 1% of all stillbirths.When the CNS is involved the infection may cause a granulomatous meningitis or diffuse encephalitis. Most infants at birth are asymptomatic, although sequelae such as seizures, hydrocephalus and chorioretinitis may appear later. Imaging features of microcephaly and parenchymal calcification are similar to CMV infection although cerebellar hypoplasia and polymicrogyria are not seen, and the ventriculomegaly may be due to an active ependymitis causing obstructive hydrocephalus rather than diffuse cerebral damage (Fig. 70.66).The severity of brain involvement correlates with earlier maternal infection. The CNS involvement in HSV is a rapidly disseminating diffuse encephalitis unlike the pattern of involvement in children or adults in which disease starts within the mesial temporal lobes and spreads within the limbic system. CT and MRI of neonatal infection show widespread asymmetrical regions of hypodensity or T2 hyperintensity mainly in the white matter. As the disease progresses there is increasing swelling and cortical involvement (CT cortical hyperdensity and T1/T2 shortening on MRI) and meningeal enhancement. Subsequent loss of brain parenchyma occurs early on, often as early as the second week, eventually resulting in profound cerebral atrophy, cystic encephalomalacia and calcification. Congenital rubella is now very rare in the West following the introduction of mass immunization programmes, but immigrant populations remain at risk as do populations where uptake of the MMR vaccine is low. In the first 8 weeks, cataracts, glaucoma and cardiac malformations occur while in the third trimester infection may be asymptomatic. Brain imaging appearances demonstrate similar changes to other congenital infections depending on the timing of the insult. It is estimated that over 60 million people worldwide are infected with HIV. Almost half are women and vertical transmission of HIV accounts for 90% of newly diagnosed cases. Children with congenitally acquired AIDS usually present between the ages of 2 months and 8 years with nonspecific signs such as hepatosplenomegaly and failure to thrive. Neonatal presentation is unusual. Affected children may develop a

Figure 70.66 CT of a neonate with congenital TORCH infection. Both the globes are small and calcified (phthisis bulbi). There is a Dandy– Walker malformation and hydrocephalus with transependymal oedema.

1691

1692

SECTION 8

•

PAEDIATRIC IMAGING

progressive HIV encephalopathy in which dementia, spasticity and increasing head size occur. A more static form is also seen in which cognitive and motor developmental delay predominate. Global atrophy and bilateral basal ganglia calcification are the most common imaging findings. Diffuse symmetrical periventricular and deep white matter abnormalities are seen in almost half of children with HIV encephalopathy and are usually associated with mild atrophy. HIV may also cause corticospinal tract degeneration.

Meningitis Meningitis is the most common infection in childhood. The diagnosis is made not on imaging but on the presence of clinical symptoms and signs and the results of lumbar puncture. Indeed in uncomplicated meningitis the imaging is usually normal, and the role of neuroimaging is to detect the complications of meningitis. Neuroimaging is indicated when the diagnosis is unclear, if the meningitis is associated with persistent seizures or focal neurological deficits, if symptoms or signs suggest raised intracranial pressure, or when recovery is unduly slow.

Pathophysiology Organisms reach the meninges by five main routes: direct spread from an adjacent infection, especially otitis media and sinusitis; haematogenous spread; rupture of a superficial cortical abscess; passage through the choroid plexus; or from direct penetrating trauma. Cerebral infarction (venous and arterial) is seen in 30% of neonates with bacterial meningitis. Infection spreads along the adventitia of penetrating cortical vessels within the periventricular spaces. Arterial thrombosis may arise from the resulting arterial wall inflammation and necrosis, or from a similar process affecting the arteries that traverse basal meningitic exudates. Venous thrombosis and subsequent infarction is particularly common in the presence of subdural empyemas due to veins becoming thrombosed as they traverse the infected subdural space. Extension of the infection through thrombosed vessels into the brain parenchyma can result in cerebritis and abscess formation. Fibropurulent exudates in the basal cisterns, ventricular out-

let foramina or over the brain convexity result in hydrocephalus due either to the obstruction of CSF flow or failure of resorption. Ventriculitis occurs in about 30% of children with meningitis and is particularly common in neonates with ependymal changes seen in severe or prolonged meningitis. Later on, ventricular enlargement may persist due to damage to the adjacent periventricular white matter and parenchymal loss. Neonatal meningitis has two distinct clinical presentations. The first presents in the first few days of life with overwhelming generalized sepsis often in association with complicated labour, such as premature rupture of membranes. The second develops after the first week with milder systemic sepsis but with more meningitic features. In older infants and children acute bacterial meningitis has a mortality of 5% in the developed world which rises to between 12% and 50% in developing countries where there is a high incidence of permanent neurological sequelae.

Uncomplicated meningitis Although neuroimaging is not performed for this indication alone and is usually normal, occasionally meningeal enhancement may be seen on CT or MRI. MRI is more sensitive than CT but the sensitivity of either/both is insufficient to warrant imaging as a diagnostic test for meningitis. Imaging findings are more useful for chronic and granulomatous meningitides where dense enhancing basal exudates may be seen within the cisternal spaces. Recurrent meningitis is unusual and full neuro-axis imaging is often applied to identify underlying risk factors (see below).

Imaging of complications (Table 70.4) Hydrocephalus may be detected on CT or MRI and may reflect a combination of obstructed CSF flow and impaired absorption. Ependymitis may cause debris/haemorrhage within the ventricular system resulting in obstructive hydrocephalus at the foramen of Monro, cerebral aqueduct and fourth ventricular outlet foramina. Purulent exudates may impair CSF absorption within the subarachnoid space resulting in communicating hydrocephalus.

Table 70.4 INTRACRANIAL COMPLICATIONS OF MENINGITIS IN INFANTS Pathology

Imaging

Cerebritis

Diffuse hypodensity (CT), hyperintensity (T2-weighted MRI) involving cortex and white matter, gyral swelling, ill-defined enhancement

Abscess formation

Peripheral rim enhancement surrounding central necrotic cavity, adjacent oedema

Effusion

Cerebrospinal fluid density/signal subdural collection, no pathological enhancement

Empyema

Higher density (CT), restricted diffusion (MRI) subdural collection with pachymeningeal/dural enhancement

Deep venous thrombosis

Hyperdense expanded venous sinus (CT), lack of T2 flow void, expanded sinus (MRI), variable haemorrhagic venous infarction

Cavernous sinus thrombosis

Expanded cavernous sinus, filling defects on CTV, signal drop off MRV

Arterial thrombosis

Large arterial territory infarct, basal ganglia/thalamic small perforating arterial territory infarcts

Ventriculitis

Debris within ventricular system, hyperdense ependyma (pre-contrast), ependymal contrast enhancement, ventricular isolation

Hydrocephalus

Obstructive intraventricular (foramen of Monro, cerebral aqueduct), obstructive extraventricular, communicating

Deafness

CT/MRI evidence of labyrinthitis ossificans

CHAPTER 70

Sterile subdural effusions are often seen as a complication of meningitis, particularly in neonates with Streptococcus pneumoniae or Haemophilus influenzae. They are not empyemas, do not need to be surgically treated and will regress as the meningitis is treated. On imaging, the subdural collections have density and signal characteristics similar to CSF on CT and MRI, though they may be slightly hyperintense to CSF on MRI.Enhancement is not usually seen; however, leptomeningeal (pial) enhancement, to be distinguished from pachymeningeal (dural) enhancement, may be seen as a result of underlying brain infarction or due to leptomeningeal inflammation. Subdural empyemas may require urgent surgical drainage to prevent further cerebritis and cerebral infarction.These appear as more proteinaceous subdural collections (increased density on CT, intermediate T1 signal intensity relative to CSF, T2 hyperintensity) with pachymeningeal (dural) and leptomeningeal enhancement (Fig. 70.67). Imaging evidence for ventriculitis, usually spread via the choroid plexus, comes from the finding of debris layered posteriorly within the ventricular system and ependyma which are hyperdense on CT. Hydrocephalus may be seen and there may be isolation of various components of the ventricular system

•

PAEDIATRIC NEURORADIOLOGY

resulting in some parts draining adequately while others remain dilated as debris obstructs the CSF outlet foramina. Infection may extend directly into the adjacent brain parenchyma causing cerebritis. Thrombosis of deep venous sinuses and cortical veins may occur, particularly in children with dehydration (Fig. 70.68). The symptoms of deep venous sinus thrombosis, such as headache, impaired consciousness and prolonged fitting, cannot be distinguished from the symptoms of the underlying meningitis and neuroimaging is required for confirmation. Cavernous sinus thrombosis is uncommon, is more commonly seen with paranasal sinus, dental or orbital infection, and tends to present with ophthalmoplegia due to involvement of the cranial nerves II, IV, and VI as they pass through the cavernous sinus. Direct involvement of the cavernous carotid artery by infection may produce a mycotic aneurysm. In generalized sepsis, the sagittal and transverse sinuses are most commonly involved. Deep venous sinus thrombosis may be seen as a hyperdense expanded sinus on unenhanced CT. As the thrombus becomes less dense, intravenous enhancement may demonstrate the ‘empty delta’ sign as a filling defect within the sinus lumen. A dense cortical vein may also be seen on CT but neither CT nor MRI can reliably detect cortical vein thrombosis.

Figure 70.67 Bilateral subdural empyemas. There is leptomeningeal and pachymeningeal enhancement (arrows) most marked over the right cerebral convexity and extending back to the vertex (on the sagittal view). There is enhancing debris within the subdural space and the signal is slightly increased compared to cerebrospinal fluid. The source of infection was from the frontal sinus (arrow).

Figure 70.68 Venous sinus thrombosis in a child with recent history of nausea and vomiting. (A) CT shows hyperdense thrombus within the vein of Galen just reaching the internal cerebral veins (arrow). There is diffuse cerebral swelling with more hypodense change and swelling affecting the left hemisphere and thalami (B) Sagittal T1-weighted MRI confirms the diagnosis with T1 shortening in keeping with methaemoglobin in the internal cerebral veins and vein of Galen (arrow). (C) The ADC map shows patchy restricted diffusion (low signal) (arrow) within the deep white matter in keeping with infarction.

1693

1694

SECTION 8

•

PAEDIATRIC IMAGING

The diagnosis of deep venous sinus thrombosis may be missed in up to 40% of patients at CT; MRI with MR venography (MRV) is more sensitive. In the subacute phase thrombus is seen as T1 shortening within an expanded sinus and this finding may be sufficient to make the diagnosis. In the acute phase when the thrombus is isointense on T1-weighted imaging and hypointense on T2-weighted imaging it can be mistaken for flowing blood, although the sinus, in the latter, should not be expanded. MRV is useful to confirm absence of flow in the thrombosed sinus. Venous infarction may occur in up to 40% of children with deep venous sinus thrombosis. Venous infarcts are often bilateral, do not conform to an arterial territory but to the territory of venous drainage, and are frequently haemorrhagic. They are parasagittal when the superior sagittal sinus is involved, thalamic when the internal cerebral veins or straight sinus/vein of Galen are involved, and temporal lobe when the transverse or sigmoid sinus or vein of Labbé (one of the deep superficial venous system anastomoses connecting the middle cerebral vein to the transverse sinus) are involved. On diffusion imaging there is a mixture of restricted and free diffusion even when nonhaemorrhagic. Arterial infarction may be seen as wedge-shaped cortical and white matter hypodensity conforming to a major arterial territory on CT or as T2 hyperintensity and T1 hypointensity with cortical highlighting on MRI. The basal meningitis may occlude small perforating branches from the circle of Willis causing small infarcts in the region of the deep grey nuclei (basal ganglia and thalami). Labyrinthitis ossificans, the most common cause of acquired deafness in childhood, is one of the sequelae of bacterial meningitis resulting from direct spread of infection from the meninges into the inner ear. Faint enhancement of the membranous labyrinth may be seen on enhanced T1-weighted images in acute infection. In some children inflammation persists; fibrosis and ossification subsequently develop and may be detected on high-resolution CT of the temporal bone, as increased density within the membranous labyrinth. High-resolution T2-weighted MRI (e.g. performed with 3DFT-CISS imaging) may be more sensitive than CT to the changes of labyrinthitis ossificans, and T2 signal drop-off may detect the fibrous stage before ossification when children are still suitable for cochlear implantation. Once the typical appearances of diffuse labyrinthitis ossificans develop, cochlear implantation is much more difficult (Fig. 70.69).

Tuberculous infection Tuberculosis (TB) has risen in incidence recently, and while it remains a serious problem in children, there does not appear to have been an increase in tuberculous meningitis in children as yet, possibly due to targeted immunization of at-risk immigrant populations. TB may cause meningitis, cerebritis and abscess formation (tuberculomas). With leptomeningeal disease, which may be seen without evidence of miliary TB elsewhere, thick enhancing purulent soft tissue exudates may be seen in the subarachnoid space, particularly in the basal cisterns, and associated with hydrocephalus, basal ganglia and thalamic infarcts. Larger

Figure 70.69 Child with recent meningitis and new sensorineural deafness. (A) High-resolution axial CT through the petrous bones shows increased density in keeping with calcification within the lateral semicircular canals (arrows). (B) Coronal and (C) axial CISS MRI shows reduced T2 signal within all the semicircular canals (arrows), particularly the left, confirmed on the oblique axial views (C). There is reduced signal within the cochlea, again worse on the left (arrows). These are typical features of labyrinthitis ossificans.

major arterial branch cortical infarcts are seen less frequently. Tuberculomas are seen as solid or ring-enhancing lesions, particularly at the grey–white matter junction.

Bacterial infection: cerebritis and abscess formation Cerebritis is the earliest stage of purulent brain infection, may be focal or multifocal, and may resolve or evolve into frank abscess formation. Predisposing conditions include middle ear, dental and paranasal sinus infection, penetrating injury, postoperative complication or dermal sinus, immune deficiency, and any cause of arteriovenous shunting (e.g. cyanotic congenital heart disease). Fungal infection may also cause brain abscesses, particularly in immunocompromised children, but in this case the lesion may not be able to encapsulate because this requires the ability to mount an adequate immune response. Radiological differentiation between cerebritis and abscess formation is important because cerebritis may respond to antibiotics while an abscess may require surgical drainage as an adjunct. During the cerebritis stage CT and MRI show an ill-defined area of oedema with swelling with or without variable ill-defined enhancement and haemorrhagic transformation. As the infection becomes more established, focal areas may become walledoff and abscess formation occurs. On imaging this appears as a space-occupying lesion with a central region of pus manifest as low density on CT, or T2 hyperintensity on MRI, surrounded by an enhancing wall. The wall of the cavity may demonstrate T1/T2 shortening. Typically there is surrounding oedema. The central region of pus shows restricted diffusion and this may

CHAPTER 70

help to differentiate abscesses from necrotic or cystic tumours which usually demonstrate central areas of increased diffusion. Imaging can be used to monitor response to treatment although there may be a lag time between the resolution of imaging findings and clinical response.

Neurocysticercosis Cysticercosis occurs from ingestion of the encysted form of Taenia solium (porcine tapeworm). Humans act as the intermediate host and neurological disease occurs as the host mounts an inflammatory response and the parasites die. In children a typical presentation is with parenchymal cysts associated with seizures, headache and focal neurological deficits. These may occur anywhere within the brain but typically at the grey– white matter interface. Calcification may occur and may be punctate within the solid component or may occur within the wall of the cyst. Perilesional oedema is the result of the inflammatory response to the dying larva. Lesions at various stages of development may be seen. Therefore active lesions seen as regions of oedema may coexist with ring or solid enhancing lesions with or without oedema, and with foci of calcification without oedema or enhancement which are burnt out lesions following the death of the parasite. There are also leptomeningeal, intraventricular and racemose forms of neurocysticercosis. Leptomeningeal disease is demonstrated on CT or MRI by soft tissue filling the basal cisterns with marked contrast enhancement. Granulomata with variable calcification may be seen within the subarachnoid space. Hydrocephalus and brain infarcts are also complications. Intraventricular cysticerci are important to detect because of the risk of acute onset hydrocephalus and sudden death. MRI is more sensitive than CT at identifying the cysts with their scolex. In the racemose form there are multilobular cysts without a scolex within the subarachnoid space, typically in the cerebellopontine angles, suprasellar region and basal cisterns and Sylvian fissures. They may demonstrate enhancement and may coexist with leptomeningitis.

Viral encephalitis The typical pattern of viral encephalitis is that of patchy and asymmetrical disease with a predeliction for grey matter. The anterior temporal and inferior frontal cortical regions are a classical location for herpes encephalitis. Another characteristic herpes virus pattern is the involvement of the hippocampi and cingulate gyrus as a limbic encephalitis. More widespread hemispheric involvement is seen with a variety of enteroviruses, echovirus and cocksackie virus in particular. A more unusual pattern of deep grey matter and upper brainstem disease is seen with rare types of viral encephalitis. Patchy enhancement with oedema has been seen with Epstein–Barr virus infection and may also be mimicked by mycoplasma infection.Thalamic and upper brainstem involvement, occasionally with haemorrhagic change, is a feature of Japanese encephalitis and the related West Nile virus. Acute cerebellitis presents infrequently in children with sudden onset of truncal ataxia, dysarthria, involuntary eye move-

•

PAEDIATRIC NEURORADIOLOGY

ments due to a swollen cerebellum, and nausea, headache and vomiting due to resulting hydrocephalus. Fever and meningism may also be present. The many causes of a swollen cerebellum include infectious (e.g. pertussis), post-infectious (e.g. acute disseminated encephalomyelitis [ADEM]), toxic, such as lead poisoning and vasculitis. Imaging shows effacement of the cerebellar fissures, enlarged cerebellum, signal abnormality affecting the cortex and white matter, and variable hydrocephalus. Although the cerebellar swelling may resolve spontaneously, surgical CSF diversion may be necessary as a temporary measure.

Infection in immunocompromised children Children with primary, acquired or iatrogenic immunodeficiencies are vulnerable to opportunistic and unusual organisms. Examples include fungal infections such as aspergillosis, actinomyces and unusual bacterial infections such as atypical mycobacteria. The host’s ability to develop an inflammatory response may be limited and the interpretation of images should be considered in this context. Contrast enhancement remains a useful hallmark. A diffuse pattern of nodular leptomeningeal and parenchymal disease may be seen. Conversely, focal large parenchymal abscesses or mycetomas may develop.

Spinal infections The patterns of discitis and spondylitis in children are similar to those of young adults and do not need to be discussed in detail in this chapter. However, special mention should be made of congenital spinal abnormalities (see Chapter 51) that may harbour or increase the risk of spinal infection.These children may present with recurrent meningitis. Spinal imaging may reveal a dorsal dermal sinus tract or an intraspinal dermoid.

Brain and cord inflammation Children may develop CNS disease known as acute disseminated encephalomyelitis (ADEM) following an infection, usually viral, or vaccination.This is assumed to be a post-infectious inflammatory immune-mediated phenomenon. Affected children present with focal neurological deficits, headache, fever and altered consciousness following a recent infection. Classically ADEM is a monophasic disease occurring at multiple sites within the brain and spinal cord. Most children recover completely, although 10–30% will have a permanent neurological deficit. Occasionally ADEM may present with relapses occurring within a few months of the original presentation, but these are still considered as part of a monophasic inflammatory process. When relapses occur that are more disseminated in time or place the diagnosis of multiple sclerosis (MS) may be made. On follow-up imaging, children with ADEM have no new lesions and complete or partial resolution of the majority of old lesions, while in MS there are new lesions which may or may not be symptomatic. On MRI multiple asymmetrical areas of demyelination seen as increased signal intensity on T2-weighted imaging with swelling occur within the subcortical white matter of both hemispheres and may also involve the cerebellum and spinal cord. Cortical and deep grey matter may also be involved but to a lesser extent (Fig. 70.70). Diffusion-weighted imaging

1695

1696

SECTION 8

•

PAEDIATRIC IMAGING

Figure 70.70 Acute disseminated encephalomyelitis in an 11 year old boy with recent viral illness and acute impaired consciousness. MRI shows bilateral, asymmetrical, mainly subcortical, cerebral hemisphere white matter lesions but also involving the cortex and deep grey matter. The cerebellum and spinal cord are also involved. There is focal swelling of the involved regions. The imaging features are typical of acute disseminated encephalomyelitis. The child made a full recovery.

shows increased free diffusion within the lesions. Occasionally, a fulminant haemorrhagic form may develop. Periventricular and callosal lesions (such as Dawson’s fingers) are more in keeping with MS lesions while cortical abnormality is not seen with MS and deep grey matter involvement, though seen in both, is more frequent in ADEM. Other differentials include viral encephalitis, which typically is more cortically based, and vasculitis (the lesions should also show restricted diffusion).

TRAUMA Birth trauma Extracranial haemorrhage may be seen as a consequence of birth trauma and more often with instrumental delivery. It does not usually require imaging but may be detected on images done for intracranial assessment. Subgaleal haemorrhage may occur deep to the scalp aponeurosis. As there is a large potential space the extent of haemorrhage may occasionally be severe, requiring transfusion. Cephalhaematoma is a traumatic subperiosteal haemorrhage between the vault and the outer layer of periosteum and is therefore confined by the sutures. It may calcify peripherally where the periosteum calcifies.Very rarely forceps may cause skull fractures.These may be linear, depressed or ‘ping pong’ (depressed without a fracture line, but to be distinguished from an asymptomatic parietal depression which is a normal variant) with a cephalhaematoma overlying them. Caput succadaneum is due to diffuse oedema and bleeding under the scalp and is more commonly seen with prolonged vaginal or Ventouse delivery. Birth trauma can result in a subdural haematoma, which may be seen even without instrumental delivery. Most are small, clinically silent and infratentorial, although they may extend above the tentorium cerebelli, and resolve spontaneously within 4 weeks. Occasionally they are large and associated with extensive intracranial haemorrhage and hydrocephalus. However, small posterior fossa subdural bleeds are common incidental findings on MRI when newborn infants are imaged for clinical CNS illness. The spinal cord may be injured by distraction during a difficult delivery, usually a breech delivery, typically affecting the lower cervical and upper thoracic regions.The cord can be tran-

sected while the soft and compliant spine remains undamaged. A more common birth-related neurological injury is brachial plexus damage secondary to traction of the shoulder during a breech delivery of the head. Brachial plexus MRI may show CSF signal pseudomeningoceles around the avulsed nerve roots.

Growing skull fractures Growing skull fractures are seen when there is a dural tear deep to the fracture and usually also when there is localized brain parenchymal damage, which may be associated with a focal neurological deficit or seizures (Fig. 70.71). CSF pulsation may keep the tear open, preventing healing of both the dura and the fracture. The fracture margins become progressively widened on serial radiographs and are bevelled and sclerotic. It usually occurs in children under 1 year with 90% occurring under the age of 3. A leptomeningeal cyst with arachnoid adhesions can cause further pressure erosion.

Spinal trauma The investigation of spinal trauma in children requires knowledge of the normal appearances of the developing spine on plain radiographs. The distance between the anterior arch of

Figure 70.71 Large cerebral parenchymal defect in association with frontal bone defect which increased in size following a frontal bone linear fracture. Note the well corticated margins of the fracture.

CHAPTER 70

C1 and the dens can be up to 5 mm in normal children, and there is increased mobility not only between C1 and C2 but also between C3 and C4. The prevertebral soft tissues should not exceed two-thirds of the width of the C2 vertebral body. Increased mobility in many children allows anterior ‘pseudoluxation’ of C2 on C3 by up to 4 mm but the posterior elements should always remain correctly aligned and the distance should reduce in extension. There is generalized ligamentous laxity of the immature spine. This may allow significant injury to the spinal cord in the absence of detectable bony injury causing SCIWORA (spinal cord injury without radiographic abnormality). In the appropriate clinical setting a proper evaluation may therefore require carefully performed flexion and extension views, and MRI to look for soft tissue and cord injury.

Atlanto-axial rotatory fixation This entity is discussed in this section for convenience, although there is often little or no history of trauma in these cases. Torticollis is common in children. Most cases are acute and the symptoms disappear without treatment within a week. Rarely it may be caused by rotatory fixation between C1 and C2, and with variable degrees of subluxation, and may occur within a normal range of movement and without subluxation. Atlanto-axial rotatory fixation should be suspected if the symptoms of torticollis persist for more than 2 weeks. The diagnostic test (CT is best) must prove that there is a fixed relationship between C1 and C2 in all positions, and in particular, on turning the head to the opposite side of the clinical presentation. Atlanto-axial rotatory fixation is present if there is rotation between C1 and C2 which remains constant throughout all positions. The treatment is aggressive with traction; if it is unsuccessful, the atlanto-axial rotatory fixation will result in a permanent rotatory malalignment requiring surgery. Secondary degenerative changes and eventually fusion may occur.

•

PAEDIATRIC NEURORADIOLOGY

repeated trauma also causes shearing injuries in the brain, often located in the subcortical region at the junction of white and grey matter. MRI detects these injuries more frequently than CT. Brain oedema is a very important feature of the shaken baby syndrome, the mechanism of which is hypothesized to be due to hypoxic–ischaemic injury. The oedema is usually massive with reduced or absent grey–white matter differentiation, and is worst in the parieto-occipital regions. The basal ganglia, brainstem and cerebellum are, in turn, relatively preserved.The outcome of brain injury caused by shaking when severe brain oedema is present is usually grim.There is rapid destruction of brain tissue and significant atrophy becomes obvious after 2–3 weeks. In severe cases the end result is multicystic encephalomalacia, and microcephaly with marked mental and motor disability (Fig. 70.72).

HYDROCEPHALUS The term ‘hydrocephalus’, literally ‘water on the brain’, is unfortunately a nonspecific term which refers to any condition in which the ventricles are enlarged, including cerebral atrophy. The clinical and radiological challenge is to identify those forms that are likely to benefit from intervention. Some qualifying terms have been used to distinguish some of these entities. The terms communicating and noncommunicating

Nonaccidental head injury Imaging of the brain is particularly important for the diagnosis of the so-called ‘shaken baby syndrome’ (see also Chapter 68).The triad of retinal haemorrhages, subdural haemorrhages and encephalopathy is accepted as a useful marker for this condition. The mechanism for injury is thought to be that of vigorous or violent shaking or such shaking followed by an impact injury. The emphasis is on rapid and alternating forces of acceleration and deceleration acting on an unsupported head. The strength of force required to cause these injuries is unknown and not easy to study scientifically. However, experienced workers would agree that such injuries cannot occur from the normal handling of an infant or young child. Subarachnoid haemorrhage, cerebral contusions, lacerations or ‘splits’ in the brain parenchyma and diffuse cerebral oedema with little evidence of external impact injury are supportive hallmarks of the shaken baby syndrome. During shaking, the brain also comes into contact with the inner surface of the skull, and there may also be a final impact, and cerebral contusions are seen. The mechanism behind the

Figure 70.72 Brain injury in ‘shaken baby syndrome’. (A) The initial CT shows posterior interhemispheric subdural and intraventricular blood. There are bilateral hypodense subdural collections overlying both hemispheres. Extensive bilateral hypodense parenchymal lesions consistent with contusions or infarction are seen. (B) Follow-up MRI shows evolution of the parenchymal damage with marked atrophy and large chronic subdural collections. (C) Coronal and (D) sagittal T1-weighted sequence shows the subdural collections have signal slightly greater than cerebrospinal fluid in keeping with proteinaceous contents from previous haemorrhage. There is cortical T1 shortening suggesting the parenchymal lesions are regions of cerebral infarction.

1697

1698

SECTION 8

•

PAEDIATRIC IMAGING

hydrocephalus are used to indicate extraventricular obstructive hydrocephalus and intraventricular obstructive hydrocephalus, respectively. These need to be distinguished from ventriculomegaly due to underlying lack of brain parenchyma, through atrophy or primary lack of brain development; some authors have suggested avoiding the use of the term hydrocephalus altogether for these cases. The concepts fail to hold true in a sizeable minority of cases, where the changes following intervention may not appear to follow expectations. For these, an appreciation of the complexities of CSF physiology may be helpful. Whilst it is generally accepted that those forms of hydrocephalus that are likely to benefit from interventions are caused by an imbalance between CSF production and absorption, the exact pathophysiology of these mechanisms remains poorly understood. CSF is produced by the choroid plexuses and is mostly absorbed through the brain and cord with a lesser contribution from the arachnoid granulations. There is a net flow of CSF through the ventricular system, from lateral to third to fourth ventricles, which is influenced by cerebrovascular pulsations. Post-haemorrhagic and post-infective hydrocephalus account for a significant proportion of hydrocephalus presenting in neonates and infants. Apart from these, congenital causes include aqueductal stenosis/gliosis, Chiari II malformation (usually occurring after repair of the associated lumbosacral myelomeningocele) and other malformations such as the Dandy–Walker malformation. Rarer causes include congenital midline tumours and vascular (e.g. vein of Galen) malformations. Before closure of the fontanelles (i.e. up to 2 years), the most reliable clinical sign is progressive macrocephaly documented by serial head circumference measurements. Caution should be exercised in the case of an asymmetrical skull or preferential growth in one direction due to craniosynostosis, which may give a spuriously enlarged head circumference without hydrocephalus. Alongside an increasingly disproportionate head circumference which crosses centiles on growth charts, other features of hydrocephalus in this age group include frontal bossing, calvarial thinning, presence of a tense, bulging anterior fontanelle, sutural diastasis and enlargement of scalp veins. Sunsetting eyes with failure of upward gaze, lateral rectus palsies and leg spasticity due to stretching of the corticospinal tracts as they descend from the motor strip around the ventricles are also seen. In older children posterior fossa neoplasms and aqueduct stenosis are the most common causes of hydrocephalus. Early morning headache, nausea, vomiting, papilloedema, leg spasticity, cranial palsies and alterations in conscious level are the dominant clinical features as the skull is rigid. The fontanelles have fused and the sutures are fusing, and therefore increasing head circumference, fontanelle bulging and sutural diastasis are not features. Imaging indicators of intraventricular obstructive (noncommunicating) hydrocephalus include dilatation of the temporal horns disproportionate to lateral ventricular dilatation, enlargement of the anterior and posterior recesses of the third ventricle with inferior convexity of the floor of the third ventricle, transependymal (periventricular interstitial) oedema and

bulging of fontanelles (Fig. 70.73). The sulcal spaces, major fissures and basal cisterns are small or obliterated. A careful survey of the regional ventricular dilatation may reveal the location as well as the cause of obstruction. Other features, such as changes in the configuration of the frontal horns of the lateral ventricles, specifically widening of the radius of the frontal horn, and a decrease in the angle it makes with the midline plane, are less useful. Further features classically described in chronic hydrocephalus, such as erosion of the dorsum sellae and copper beaten skull, are even less reliable. Extraventricular obstructive (communicating) hydrocephalus, however, may reveal a range of findings from ventricular and sulcal prominence to ‘normal’ CT/MRI appearances. It may have a variety of causes ranging from haemorrhage and proteinaceous cellular debris with infection and disseminated malignancy to venous hypertension from impaired venous drainage to impaired arterial compliance from diffuse arteriopathies. In some cases, a combination of both forms of obstructive hydrocephalus would be expected. Assessment of the cortical sulci for effacement disproportionate to ventricular size in children can be difficult and it requires some knowledge of the normal appearances for age; in normal younger children, under two, the ventricles and subarachnoid spaces are more prominent, and should not be misinterpreted as ‘atrophic’ or due to hydrocephalus. Knowledge of the head size is important in making this distinction, being large in hydrocephalus and benign enlargement of the subarachnoid spaces, or small if there is cerebral atrophy. Again it may not be possible to be certain about the findings in the absence of serial measurements documenting trends. In benign enlargement of the subarachnoid spaces, there is believed to be a mismatch between the rate of skull growth and rate of growth of the developing brain. The child is neurologically normal. There is rapid skull growth, often to above the 95th centile by the time of presentation, but this then stabilizes and both the size of the subarachnoid spaces and head size have usually normalized by age 2 years. The extra space anterior to the cerebral convexities is distinguishable from subdural collections by lack of mass effect on the adjacent brain and the presence of crossing

Figure 70.73 Child with mucopolysaccharidosis type I and hydrocephalus. The temporal horns and anterior recesses of the third ventricle are dilated. There is transependymal oedema and the cerebral sulci are effaced.

CHAPTER 70

veins within the subarachnoid space.The importance of its recognition is to indicate that no intervention is required, with the expectation that the situation will normalize with age. In the second decade the hemispheric cortical sulci become much less conspicuous and the ventricles less prominent, which should equally not be misinterpreted as due to cerebral oedema or the presence of a ‘tight’ brain. In this case in the normal child the basal cisterns should not be effaced. An attempt should be made to identify a structural cause for the hydrocephalus. The narrowest parts of the ventricular system are the most susceptible to mechanical obstruction, i.e. the foramina of Monro, cerebral aqueduct and fourth ventricular outflow foramina. Masses causing obstructive hydrocephalus at the foramina of Monro include superior extension of suprasellar tumours and arachnoid cysts, colloid cysts, which may cause sudden acute and life-threatening hydrocephalus, and giant cell astrocytomas in tuberous sclerosis. Masses effacing the cerebral aqueduct include tectal plate gliomas, superior extension of midline posterior fossa tumours and brainstem diffuse astrocytomas and inferior extension of pineal region tumours. Atrial diverticula from dilated lateral ventricles may herniate into the quadrigeminal and supracerebellar cisterns and may also compress the tectum.These may be distinguished from large arachnoid cysts using multiplanar MRI to confirm continuity with the ventricular system. More diffuse patterns of obstruction, including isolation of pockets of CSF, may be caused by haemorrhage, infection, disseminated tumour and reparative fibrosis. Clues to the underlying cause include evidence of haemosiderin staining or superficial siderosis in previous haemorrhage; focal atrophy or porencephaly from previous haematomas; the presence of hyperdense exudates in the basal cisterns, subarachnoid space and pial enhancement with infection or pial neoplastic dissemination. Aqueduct stenosis is one of the most common causes of hydrocephalus in children but may present at any time from birth to adulthood. It may be developmental or acquired gliosis secondary to infection or intraventricular haemorrhage. Classically, the lateral and third ventricles are dilated and the fourth ventricle is of normal size, and there is no evidence of a tectal plate tumour. On sagittal MRI there is focal narrowing of the cerebral aqueduct, normally proximally at the level of the superior colliculi or at the intercollicular sulcus with posterior displacement of the tectal plate. Proximal to this the aqueduct is dilated (Fig. 70.74). Occasionally a congenital web may be seen as a very thin sheet of tissue across the distal aqueduct. Overproduction of CSF is a very rare cause of hydrocephalus, most often seen with choroid plexus papillomas. These tumours may cause hydrocephalus by other mechanisms, e.g. obstructive hydrocephalus of the lateral ventricle due to mass effect at the body/trigone or foramen of Monro of the lateral ventricle, and haemorrhage within the ventricular system. A highly proteinaceous exudate produced by the tumour can cause communicating hydrocephalus due to impaired extraventricular drainage of CSF. Spinal cord tumours may be associated with hydrocephalus due to proteinaceous exudates or pial dissemination of disease, both causing impaired absorp-

•

PAEDIATRIC NEURORADIOLOGY

Figure 70.74 Child with aqueduct stenosis with a history of prematurity. There is dilatation of the lateral ventricles and proximal cerebral aqueduct (arrow). The hydrocephalus is long-standing as there is white matter volume loss and the anterior recesses of the third ventricle are not bulging. There is more focal damage in the left frontal lobe, probably due to previous germinal matrix haemorrhage. A posterior fossa arachnoid cyst is also noted.

tion of CSF. Indeed in the cases of unexplained hydrocephalus, spinal imaging is recommended to exclude a possible occult intraspinal lesion. Chiari II malformation is a common cause of hydrocephalus presenting early in life. The underlying aetiology is disputed; it may not be due to simple craniocervical junction obstruction but due to the displacement of the fourth ventricular outlet foramina within the spinal canal inferior to the foramen magnum, i.e. within the spinal canal, where the capacity for CSF absorption is reduced. Some specific features should be looked for when assessing these children for hydrocephalus or shunt malfunction. Children with occipital headaches at night probably have some degree of shunt malfunction which may present with potentially fatal signs of brainstem compression and apnoeic attacks. Also, the fourth ventricle should be slit-like, so the presence of a normal sized or enlarged fourth ventricle suggests a shunt malfunction or isolation of the fourth ventricle requiring diversion. Finally, hydrocephalus may be seen as a consequence of raised intracranial venous pressure. Examples include syndromic skull base abnormalities in craniosynostosis(es) and leading to stenosis of the jugular outflow foramina. Vascular causes include venous thrombosis, the vein of Galen aneurysmal malformation and dural arteriovenous shunts. Persistent untreated raised intraventricular pressure may result in secondary parenchymal damage affecting adjacent white matter and resulting in impaired cognitive function, permanent spasticity and blindness. As well as by removing the obstructive lesion, hydrocephalus can be treated by CSF diversion. This can be performed initially by external ventricular drainage, or more permanently by ventriculoperitoneal, ventriculoatrial shunting, or third ventriculostomy. Here a surgical defect is made in the floor of the third ventricle allowing CSF drainage into the suprasellar cistern. Shunt malfunction from obstruction of the shunt by choroid plexus or glial tissue and subsequent ventricular dilatation

1699

1700

SECTION 8

•

PAEDIATRIC IMAGING

is best assessed by comparison with baseline or previous imaging. In addition to a recurrence of the signs of hydrocephalus, there may be the appearance of fluid tracking along the length or the shunt tubing. Assessment of shunt tubing integrity may be made by plain radiograph imaging of the tube from the skull to the abdomen/pelvis. The scout view of a brain CT may also be useful to pick up disrupted shunt tubing or disconnection of the valve. As well as separation of the fragments, calcification may be seen at either end of the shunt disruption due to inflammation and fibrotic change where the shunt has become tethered and then distracted. Knowledge of the type of shunt inserted, or discussion with the neurosurgeon who inserted it, should allow distinction of the normal radiolucent components of the shunt (often as it exits the calvarium) from shunt distraction. Patency of a third ventriculostomy can be inferred on MRI by visualizing the surgical defect and detecting rapid CSF flow through it into the suprasellar cistern, manifest as large hypointense flow voids on T2-weighted imaging. This may be aided by imaging without any flow compensation. The incidence of shunt infection has declined over the years and in most neurosurgical centres runs at around 1–5%, being slightly higher in infants. Imaging may detect evidence of ventriculitis: enlarged ventricles with hyperdense ependyma on CT, ependymal enhancement and debris within the ventricular system.This may progress to cerebritis with devastating consequences for the developing brain parenchyma. Other shunt complications include the development of abdominal ascites, pseudocyst formation and perforated abdominal viscus. Rarely, shunted hydrocephalic patients may become symptomatic without ventricular dilatation and develop the ‘slit ventricle’ syndrome. This has a number of potential causes which include overdrainage of CSF, stiff, poorly compliant ventricles which allow raised intraventricular pressure without ventricular dilatation, intermittent shunt malfunction and headaches unrelated to the shunt.

SUMMARY Paediatric neuroradiology is a subspecialist field drawing from knowledge and expertise in neuroimaging, paediatrics, embryology and genetics. Some skill is required in adapting neuroimaging techniques to suit the paediatric population, especially in terms of sedation/general anaesthesia. The range of conditions encountered is very different from that seen in adults. The management aims and options are also often different from adult cases.

REFERENCES 1. Barkovich A J 2000 Concepts of myelin and myelination in neuroradiology. Am J Neuroradiol 21: 1099–1109 2. Wimberger D M, Roberts T P, Barkovich A J et al 1995 Identification of ‘premyelination’ by diffusion weighted MRI. J Comput Assist Tomogr 19: 28–33 3. Huppi P S, Maier S E, Peded S et al 1998 Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 44: 584–590

4. Engelbrecht V, Rassek M, Preiss S et al 1998 Age-dependent changes in magnetization transfer contrast of white matter in the pediatric brain. Am J Neuroradiol 19:1923–1929. 5. Barkovich A J 1998 MR of the normal neonatal brain: assessment of deep structures. Am J Neuroradiol 19: 1397–1403 6. Barkovich A J, Kjos B O, Jackson D E et al 1988 Normal maturation of the neonatal and infant brain: MR imaging at 1.5T. Radiology 166: 173–180 7. Yakovlev P I, Lecours A R 1967 The myelogenetic cycles of regional maturation of the brain. In: Minkowski A (ed) Regional development of the brain in early life. Oxford: Blackwell, pp 3–70 8. Van der Knaap M S, Van Wezel-Miejler G, Barth P G et al 1996 Normal gyration and sulcation in preterm and term neonates: appearance on MR images. Radiology 200: 389–396 9. Martin E, Kikinis R, Zuerrer M et al 1988 Developmental stages of human brain: an MR study. J Comput Assist Tomogr 12: 917–922 10. Barkovich A J, Kjos B O 1988 Normal postnatal development of the corpus callosum as demonstrated by MR imaging. Am J Neuroradiol 9: 487–491 11. Aoki S, Okada Y, Nishimura K et al 1989 Normal deposition of brain iron in childhood and adolescence: MR imaging at 1.5T. Radiology 172: 381–385 12. Cox T D, Elster A D 1991 Normal pituitary gland: changes in shape, size and signal intensity during the first year of life at MR imaging. Radiology 179: 721–724 13. Raemakaers V T, Heimann G, Reul J et al 1997 Genetic disorders and cerebellar structural abnormalities in childhood. Brain 120: 1739–1751 14. Tortori-Donati P, Fondelli M, Rossi A et al 1996 Cystic malformations of the posterior cranial fossa originating from a defect of the posterior membranous area: mega cisterna magna and persisting Blake’s pouch: two separate entities. Childs Nerv Syst 12: 203–308 15. Hart M N, Malamud N, Ellis W G 1972 The Dandy–Walker syndrome. A clinicopathological study based on 28 cases. Neurology 22: 771–780 16. Joubert M, Eisenring J J, Robb J P et al 1969 Familial agenesis of the cerbellar vermis. A syndrome of episodic hyperpnea, abnormal eye movements, ataxia and retardation. Neurology 19: 813–825 17. Kendall B, Kingsley D, Lambert S R et al 1990 Joubert syndrome: a clinical–radiological study. Neuroradiology 31: 502–506 18. Toelle S P, Yalcinkaya C, Kocer N et al 2002 Rhombencephalosynapsis: clinical findings and neuroimaging in 9 children. Neuropediatrics 33: 209–214 19. Meltzer C C, Smirniotopoulos J G, Jones R V 1995 The striated cerebellum: an MR imaging sign in L’Hermitte-Duclos disease (dysplastic gangliocytoma). Radiology 194: 699–703 20. Barkovich A J 2002 Magnetic resonance imaging: role in the understanding of cerebral malformations. Brain Dev 24: 2–12 21. Wright L B, James C A, Glasier C M 2001 Congenital cerebral and cerebrovascular anomalies: magnetic resonance imaging. Top Magn Reson Imaging 12: 361–374. 22. Barkovich A J 2005 Congenital malformations of the brain and skull. In: Pediatric neuroradiology, 3rd edn. Philadelphia: Lippincott, Williams & Wilkins, ch 3 23. Naidich T P, McLone D G, Fulling K H 1983 The Chiari II malformation: Part IV. The hindbrain deformity. Neuroradiology 25: 179–197 24. Wolpert S M, Anderson M, Scott R M et al 1987 Chiari II malformation: MR imaging evaluation. Am J Roentgenol 149: 1033–1042 25. Hetts S W, Sherr E H, Chao S et al 2006 Anomalies of the corpus callosum: an MR analysis of the phenotypic spectrum of associated malformations. Am J Roentgenol 187: 1343–1348 26. Barkovich A J 2005 The phakomatoses. In: Paediatric neuroimaging, 3rd edn. Philadelphia: Lippincott, Williams & Wilkins, pp 440–504 27. Herron J, Darrah R, Quaghebeur G 2000 Intra-cranial manifestations of the neurocutaneous syndromes. Clin Radiol 55: 82–98 28. Neurofibromatosis. In: National Institutes of Health Consensus Development Conference. Arch Neurol 1988;575–578 29. Balcer L J, Liu G T, Heller G et al 2001 Visual loss in children with neurofibromatosis type 1 and optic pathway gliomas: relation to

CHAPTER 70

30.

31. 32.

33.

34. 35.

36.

37.

38.

39.

40.

41.

42. 43. 44. 45. 46.

47. 48. 49. 50.

51.

tumor location by magnetic resonance imaging. Am J Ophthalmol 131: 442–445 Goh W H, Khong P L, Leung C S et al 2004 T2-weighted hyperintensities (unidentified bright objects) in children with neurofibromatosis 1: their impact on cognitive function. J Child Neurol 19: 853–858 Roach E S, Sparagana S P 2004 Diagnosis of tuberous sclerosis complex. J Child Neurol 19: 643–649. Roach E S, Gomez M R, Northrup H 1998 Tuberous sclerosis complex consensus conference: revised clinical diagnostic criteria. J Child Neurol 13: 624–628 Kingsley D, Kendall B, Fitz C 1986 Tuberous sclerosis: a clinicoradiological evaluation of 110 cases with particular reference to atypical presentation. Neuroradiology 28: 171–190 Baron Y, Barkovich A J 1999 MR imaging of tuberous sclerosis in neonates and young infants. Am J Neuroradiol 20: 907–916 Braffman B, Naidich T P 1994 The phakomatoses: Part II. von Hippel– Lindau disease, Sturge–Weber syndrome, and less common conditions. Neuroimaging Clin N Am 4: 325–348 Marti -Bonmarti L, Menor F, Mulas F 1993 The Sturge–Weber syndrome: correlation between the clinical status and radiological CT and MRI findings. Childs Nerv Syst 9: 107–109 Naidich T P, Blaser S I, Delman B N et al 2002 Embryology of the spine and spinal cord, in Proceedings of the 40th Annual Meeting of the American Society of Neuroradiology, Vancouver, BC, May 13–17, pp 3–13 Barkovich A J 2005 Congenital anomalies of the spine. In: Barkovich A J (ed) Pediatric neuroimaging, 4th edn. Philadelphia: Lippincott Williams & Wilkins, 2005 Rossi A, Gandolfo C, Morana G, et al 2006 Current classification and imaging of congenital spinal abnormalities. Semin Roentgenol 41: 250–273 Rowland Hill C A, Gibson P J 1995 Ultrasound determination of the normal location of the conus medullaris in neonates. Am J Neonat Radiol 16: 469–472 Beek FJ, de Vries L S, Gerards L J, Mali W P 1996 Sonographic determination of the position of the conus medullaris in premature and term infants. Neuroradiology 38 (Suppl 1): S174–177 McLone D G, Dias M S 1991–1992 Complications of myelomeningocele closure. Pediatr Neurosurg 17: 267–273 Drolet B 1998 Birthmarks to worry about. Cutaneous markers of dysraphism. Dermatol Clin 16: 447–453 Thompson D 2000 Hairy backs, tails and dimples. Curr Paediatr 10: 177–183 Hughes J A, De Bruyn R, Patel K, Thompson D 2003 Evaluation of spinal ultrasound in spinal dysraphism. Clin Radiol 58: 227–233 Pang D, Dias M S, Ahab-Barmada M 1992 Split cord malformation. Part I: a unified theory of embryogenesis for double spinal cord malformations. Neurosurgery 31: 451–480 Pang D 1992 Split cord malformation. Part II: clinical syndrome. Neurosurgery 31: 481–500 Faris J C, Crowe J E 1975 The split notochord syndrome. J Pediatr Surg 10: 467–472 Duhamel B 1961 From the mermaid to anal imperforation: the syndrome of caudal regression. Arch Dis Child 36: 152–155 Carey J C, Greenbaum B, Hall B D 1978 The OEIS complex (omphalocele, exstrophy, imperforate anus, spinal defects). Birth Defects Orig Artic Ser 14: 253–263 Currarino G, Coln D, Votteler T 1981 Triad of anorectal, sacral, and presacral anomalies. Am J Roentgenol 137: 395–398

•

PAEDIATRIC NEURORADIOLOGY

52. Pang D 1993 Sacral agenesis and caudal spinal cord malformations. Neurosurgery 32: 755–779 53. Van Der Knaap M S, Valk J 2005 Magnetic resonance imaging of myelination and myelination disorders, 3rd edn. Berlin: Springer Verlag 54. Patay Z 2005 Metabolic disorders. In: Carty H et al (eds) Imaging children, 2nd edn, vol 2. Edinburgh: Churchill Livingstone, pp 1899–1956 55. Hayward R, Jones B, Dunaway D, Evans R (eds) 2004 The clinical management of craniosynostosis. Clin Dev Med 163 56. World Health Organization 2000 Classification of tumours. Pathology and genetics. In: Kleihues P, Cavenee W (eds) Tumours of the nervous system. IARC Press Cambridge 57. Griffiths P D 1999 A protocol for imaging paediatric brain tumours. United Kingdom Children’s Cancer Study Group (UKCCSG) and Societe Francaise D’Oncologie Pediatrique (SFOP) Panelists. Clin Oncol (R Coll Radiol) 11: 290–294 58. Saunders D E, Phipps K P, Wade A M et al 2005 Surveillance imaging strategies following surgery and/or radiotherapy for childhood cerebellar low-grade astrocytoma. J Neurosurg 102 (2 Suppl):172–178. 59. Saunders D E, Hayward R D, Phipps K P et al 2003 Surveillance neuroimaging of intracranial medulloblastoma in children: how effective, how often, and for how long? J Neurosurg 99: 280–286. 60. Good C D, Wade A M, Hayward R D et al 2001 Surveillance neuroimaging in childhood intracranial ependymoma: how effective, how often, and for how long? J Neurosurg 94: 27–32. 61. Rumboldt Z, Camacho D L, Lake D et al 2006 Apparent diffusion coefficients for differentiation of cerebellar tumors in children. Am J Neuroradiol 27: 1362–1369. 62. Arai K, Sato N, Aoki J et al 2006 MR signal of the solid portion of pilocytic astrocytoma on T2-weighted images: is it useful for differentiation from medulloblastoma? Neuroradiology 48: 233–237. 63. Fischbein N J, Prados M D, Wara W et al 1996 Radiologic classification of brain stem tumors: correlation of magnetic resonance imaging appearance with clinical outcome. Pediatr Neurosurg 24: 9–23. 64. Bowers D C, Georgiades C, Aronson L J 2000 Tectal gliomas: natural history of an indolent lesion in pediatric patients. Pediatr Neurosurg 32: 24–29. 65. Chong W K, Saunders D S, Ganesan V 2004 Stroke is an important paediatric illness. Pediatr Radiol 34: 2–4. 66. Paediatric Stroke Working Group 2004 Stroke in childhood. Clinical guidelines for diagnosis, management and rehabilitation. London: Clinical Effectiveness and Evaluation Unit, Royal College of Physicians. 67. Ganesan V, Prengler M, McShane M A, Wade A M, Kirkham F J 2003 Investigation of risk factors in children with arterial ischemic stroke. Ann Neurol 53: 167–173. 68. DeVeber G 2005 In pursuit of evidence-based treatments for paediatric stroke: the UK and Chest guidelines. Lancet Neurol 4: 432–436. 69. Barkovich A, Truwit C 1989 Brain damage from perinatal asphyxia: correlation of MR findings with gestational age. Am J Neuroradiol 11: 1087–1096 70. Barkovich A 1992 MR and CT evaluation of profound neonatal and infantile asphyxia. Am J Neuroradiol 13: 959–972 71. Barkovich A, Sargent S 1995 Profound asphyxia in the premature infant: Imaging findings. Am J Neuroradiol 16: 1837–1846 72. Krageloh-Mann I 2004 Imaging of early brain injury and cortical plasticity. Exp Neurol 190 (Suppl 1): S84–90 73. Krageloh-Mann I, Peterson D, Hagberg G, et al 1995 Bilateral spastic cerebral palsy – MRI pathology and origin. Analysis from a representative series of 56 cases. Dev Med Child Neurol 37: 379–397

1701

CHAPTER

Imaging of the Endocrine System

71

S. Aslam A. Sohaib, Andrea Rockall, Jamshed B. Bomanji, Jane Evanson and Rodney H. Reznek

• • • • • • • •

The hypothalamic–pituitary axis The thyroid gland The parathyroid glands Pancreatic and gastrointestinal endocrine disorders Carcinoid tumours The adrenal glands The female reproductive system The male reproductive system

Endocrine disorders result from excessive or deficient hormone production, which may be continuous or intermittent and may affect a single or multiple hormones. Hypofunction results from destruction or replacement of a gland by a tumour, infection, autoimmune processes, haemorrhage or infarction, or by compression from an adjacent mass lesion. Hyperfunction may be caused by hypertrophy, microadenoma, adenoma (single or multiple), or carcinoma of the endocrine gland. Over-production of a hormone may also arise from an ‘ectopic’ site. In most cases, this is due to a tumour that is usually, but not always, malignant (e.g., adrenocorticotropic hormone [ACTH] secretion by small-cell lung cancer). The diagnosis of endocrine disease is usually based on clinical presentation and subsequent assay of the appropriate hormone. It is rare for radiological signs to suggest an unsuspected endocrine disease. There are few radiological signs that confirm the diagnosis when biochemical tests are equivocal. The challenge for imaging in endocrine disease lies in identifying the abnormality leading to the endocrine disorder. Although radiological investigations of endocrine disease now centre mostly on cross-sectional imaging techniques, a functional abnormality often needs to be corroborated by radionuclide imaging, venous sampling, or biochemistry. Cross-sectional imaging techniques are unable to distinguish between a functioning and a nonfunctioning tumour. Imaging of the endocrine system differs in several ways from imaging in other body systems. Although a correlation

between anatomical abnormality and functional abnormality is important, the degree of endocrine abnormality does not correlate with the size of the macroscopic lesion. Very small lesions can give rise to profound biochemical disturbance; attention to technique is therefore essential, as a small lesion may be missed. Furthermore, an endocrine disorder may arise from one of several sites within the endocrine system in addition to ectopic sites; for example, Cushing’s syndrome may arise from anywhere within the hypothalamic–pituitary–adrenal axis or from an ectopic site. Detection of the source of Cushing’s syndrome may require imaging of several different anatomical regions using multiple imaging techniques. Due to the diversity of the endocrine system and its pathology, the imaging needs to be tailored to the endocrine gland or disease in question, as discussed in the sections below.

THE HYPOTHALAMIC–PITUITARY AXIS ANATOMY AND PHYSIOLOGY The hypothalamus is a thin layer of tissue that forms the floor and lower lateral walls of the third ventricle. It extends from the lamina terminalis and optic chiasm anteriorly to the mammillary bodies posteriorly (Fig. 71.1). The inferior surface of the hypothalamus between these structures is called the tuber cinereum. A swelling on its surface, the median eminence, is continuous with the infundibulum (pituitary stalk). The pituitary stalk tapers smoothly as it travels inferiorly from the hypothalamic origin to the pituitary insertion. The normal stalk is 3–3.5 mm wide at its origin and 2 mm wide near its insertion1. The hypothalamus controls pituitary function through direct neuronal links with the posterior lobe of the pituitary (neurohypophysis) and vascular links with the anterior lobe (adenohypophysis). Within the medial parts of the hypothalamus are a number of named hypothalamic nuclei (e.g., supraoptic and paraventricular nuclei), which manufacture hormones. Axons from the supraoptic and paraventricular nuclei transport

1706

SECTION 9

•

MISCELLANEOUS DISORDERS

6 5 4 3

2

1

Figure 71.1 Normal pituitary gland. Unenhanced sagittal T1-weighted spin-echo images of a normal pituitary gland and hypothalamus. The anterior pituitary gland (1), posterior pituitary bright spot (2), pituitary stalk (3), tuber cinereum (4), mammillary bodies (5), sphenoid sinus (6) and optic chiasm (7) are identified.

vasopressin (antidiuretic hormone; ADH) and oxytocin down the stalk into the posterior lobe of the pituitary. Hypothalamicreleasing factors are delivered to the anterior lobe of the pituitary by a system of hypophyseal veins, which run from the hypothalamus to the vascular sinusoids of the anterior pituitary. The pituitary gland is composed of two lobes that are physiologically, embryologically and anatomically distinct.The anterior lobe is the largest part and comprises 75% of the total volume. It arises from an invagination of Rathke’s pouch from the fetal nasopharynx. It is composed of glandular cells that produce numerous hormones. Its predominant blood supply is from the hypophyseal-portal system, which also serves as the pathway by which hypothalamic-releasing hormones reach the pituitary. In adults, the anterior pituitary appears homogeneous and isointense to grey matter on T1- and T2weighted MRI. The posterior lobe of the pituitary occupies the posterior third of the sella and arises embryologically from a down growth of the hypothalamus. It is characteristically of high signal on T1-weighted MRI sequences, i.e., ‘posterior pituitary bright spot’ (see Fig. 71.1).This high signal is ascribed

to the vasopressin complex present within the axon terminals. Absence of this high signal is reported in patients with central diabetes insipidus2. However, the high signal may not be seen in all normal subjects3. An ectopic position of the posterior pituitary may be a normal variant, but can be associated with pituitary dwarfism (growth hormone [GH] deficiency)4. The pituitary gland occupies the sella turcica, which is a cup-shaped depression in the basisphenoid bone. The roof of the sella is formed by the diaphragma sella, a fold of dura, which is perforated to allow passage of the pituitary stalk. Above the diaphragma sella lies the suprasellar cistern the optic chiasm and the anterior cerebral arteries. The lateral walls of the pituitary fossa are formed by the cavernous sinuses containing the internal carotid arteries and cranial nerves III, IV, the first and second divisions of V and VI. Behind the sella is the pontine cistern containing the basilar artery.The cavernous sinus, pituitary gland and stalk lie outside the blood–brain barrier and, therefore, show significant enhancement after the administration of intravenous contrast medium. The hypothalamus plays an important role in the regulation of the autonomic nervous system, temperature, appetite and sleep patterns. It also influences most endocrine glands through its effect on the pituitary gland (Table 71.1).

IMAGING TECHNIQUES MRI and CT are the main techniques for evaluating the sella and juxtasellar region, with MRI being the technique of choice. Current imaging protocols for the hypothalamic–pituitary axis (HPA) typically utilize sagittal and coronal T1-weighted spin-echo sequences. In order to achieve high spatial resolution, thin slices (3 mm), and a small field of view (16–20 cm) is used. Alternatively a 3D-spoiled gradient-echo volume acquisition can be used. This results in thinner slices (1.5 mm) with good signal-to-noise ratio and can be reconstructed in all three planes. Susceptibility artefacts at air/bone interfaces may occur with these sequences. Dynamic T1-weighted spin-echo acquisition within 1–2 minutes after intravenous gadolinium slightly improves the sensitivity of MRI for the detection of small pituitary adenomas5. T2-weighted sequences are not as sensitive as contrast enhanced T1-weighted sequences for the detection of adenomas, but can occasionally prove useful. There is still a role for CT in imaging the pituitary. There are a number of patients who cannot undergo MRI, and CT is also useful for detection of calcification, the presence of which

Table 71.1 PITUITARY HORMONES AND THEIR REGULATION Pituitary lobe Anterior

Posterior

Hormone

Releasing factor

Inhibitory factor

Target organ

Corticotropin

CRH

ADH

Adrenal gland

FSH, LH

GnRH

–

Gonads

GH

GHRH

Somatotstatin

Liver, bone, fat

Prolactin

VIP

Dopamine

Breast

TSH

TRH

Somatotstatin

Thyroid

ADH

–

–

Kidney

Oxytocin

–

–

Breast, uterus

CHAPTER 71

may influence the differential diagnosis. CT may contribute to preoperative planning particularly in regard to the pneumatization and anatomy of the sphenoid air sinus. Pituitary CT can be performed in the axial plane with thin (1 mm) contiguous sections following intravenous contrast medium. Images can then be reformatted in the coronal and sagittal planes. The role of inferior petrosal venous sampling in pituitary tumour localization is discussed elsewhere.

DISORDERS OF THE HYPOTHALAMIC– PITUITARY AXIS Dysfunction of the HPA typically results from intrinsic disorders of the HPA itself although lesions in adjacent structures (e.g., cavernous sinus) may involve the HPA. Hyperfunction of the HPA can give rise to a number of well-described syndromes depending on which hormone is being hypersecreted (Tables 71.2–71.7). Pituitary hyperfunction is most commonly from a pituitary adenoma or occasionally from other pituitary lesions (e.g., hypophysitis). Hypofunctioning states results from destruction of the HPA leading to the under production of hormone. Total or partial hypopituitarism may occur in patients with pituitary adeno-

Table 71.2 RADIOLOGICAL FEATURES OF ACROMEGALY

•

IMAGING OF THE ENDOCRINE SYSTEM

Table 71.4

RADIOLOGICAL FEATURES OF HYPOPITUITARISM

Skull

Unfused sutures

Skeleton

Small but normal proportions (Lorain dwarf) Slender bones Small pituitary fossa Unfused epiphyses

Table 71.5 RADIOLOGICAL FEATURES OF HYPERTHYROIDISM (THYROTOXICOSIS) Skull

Exophthalmos

Skeleton

Osteopenia Cortical striation—acropachy In childhood, early appearance and accelerated growth of ossification centres

Heart

Cardiac enlargement Cardiac failure

Thymus

Enlargement

Table 71.6 RADIOLOGICAL FEATURES OF HYPOTHYROIDISM (CRETINISM AND JUVENILE MYXOEDEMA) Skull

Delayed closure of fontanelles Relatively large sella Poorly developed paranasal sinus Usually brachycephalic Dentition delayed: dental caries Wormian bones

Skeleton

Vault thickened Paranasal and mastoid air cells enlarged Pituitary fossa enlarged Floor of the fossa asymmetrical or ballooned

Dwarfism Increased density

Ossification centres

Retarded growth Multicentric and irregular Bilateral and symmetrical

Mandible

Prognathism with increased angle

Epiphyses

Spine

Kyphosis Enlarged vertebral bodies Posterior scalloping of vertebral bodies

Delayed fusion and appearance Inhomogeneous epiphyses Fine or coarse stippling Fragmentation

Chest

Increased antero-posterior diameter Ribs increased in calibre and length

Spine

Hands

General enlargement Enlarged bases of phalanges and terminal tufts, spade-like Enlarged muscle attachments

Kyphosis Flattening of bodies Increase in width of intervertebral space Bullet-shaped vertebral bodies, usually L1 and L2

Feet

Thickening of ‘heel pad’: M > 23 mm, F > 21.5 mm

Long bones

Short Dense transverse bands at metaphyseal ends

Long bones

Thickened by periosteal new bone formation

Pelvis

Narrow with coxa vara

Joints

Widening of joint spaces due to thickened cartilage Premature degeneration (OA) changes (shoulders, hips, knees) Chondrocalcinosis

Table 71.7

Skull

Soft tissues

RADIOLOGICAL FEATURES OF MYXOEDEMA

Heart

Enlargement Pericardial effusion Heart failure

Body cavities

Pleural effusion Ascites

Gastrointestinal tract

Abnormalities of oesophageal peristalsis Decreased incidence of peristalsis Constipation ‘Pseudo-obstruction’

Enlarged heart, kidneys, liver Calcification of pinna of ears

Table 71.3 RADIOLOGICAL FEATURES OF CUSHING’S SYNDROME Skull

Pituitary fossa usually normal

Skeleton

Osteoporosis Vertebral collapse Kyphosis Concave vertebral margins Wedged vertebral bodies Rib fractures—multiple, painless with excess callus Necrosis of femoral heads Secondary osteoarthritis

mas, with parasellar diseases, in those who have had pituitary surgery or radiation, or following head injury. The systemic radiological changes in adults with hypopituitarism are minimal (see Table 71.4). The role of imaging in disorders of the HPA is to identify and characterize structural lesions, which may have caused the

1707

1708

SECTION 9

•

MISCELLANEOUS DISORDERS

endocrine abnormalities. It is important first to identify the primary anatomical site of the lesion. Lesions may be intrasellar, suprasellar, or combined intra- and suprasellar.

Intrasellar lesions Normal variants: The normal pituitary size and shape varies with age and sex. The pituitary gland in neonates is rounder, brighter and relatively larger during the first 2 months of life than in later childhood. In adults, the average height of the pituitary is between 3.5 and 8 mm and is slightly larger in women than men. In pubertal girls and pregnant women the gland enlarges and often appears spherical or upwardly convex and may occasionally reach 10–11 mm in height. The anterior pituitary may also appear hyperintense in pregnant or postpartum women. Pubertal boys may have a slightly rounded, upward convex pituitary that measures up to 6– 7 mm in height. The pituitary gland normally decreases in size with ageing, though there may be a slight increase in size in perimenopausal women6. In addition to the normal physiological changes in the pituitary with age and sex, the pituitary may enlarge during the administration of exogenous oestrogens, in central precocious puberty, with hypothalamic tumours, ectopic tumours secreting hypothalamic-releasing factors, and secondary to end organ failure, e.g., in hypothyroidism. An empty sella is usually an incidental finding thought to be caused by a congenital defect in the diaphragm sellae, which allows pulsating cerebrospinal fluid (CSF) to expand the sella. This leads to compression and atrophy of the pituitary gland and expansion of the sella turcica. As the anterior pituitary has a large functional reserve, endocrine dysfunction only occurs if less than 10% of the gland remains active. On CT and MRI the pituitary gland is flattened to a thin rim along the floor of the sella turcica. The pituitary stalk remains in its normal position, a finding that differentiates an empty sella from a cystic tumour in which the stalk is displaced or obliterated. Occasionally intrasellar herniation of the optic nerve, chiasm, or tract into the empty sella is seen. An acquired empty sella may occur with anterior pituitary haemorrhage secondary to pituitary gland necrosis following peripartum circulatory collapse (Sheehan syndrome) or from infarction of a pituitary adenoma. Adenomas are the most common primary neoplasms of the pituitary. They are benign and slow growing. Those smaller than 10 mm in diameter are termed microadenomas (Fig. 71.2) and those larger than 10 mm termed macroadenomas (Figs 71.3, 71.4). Approximately 75% of patients with pituitary adenomas will have symptoms of hormone excess, while the remainder (those with nonfunctioning tumours) present with clinical findings related to tumour mass effect (e.g., headache, visual field defects, cranial nerve palsies). Pituitary adenomas are classified on the basis of electron microscopic and immunohistochemical criteria. On high-resolution CT, pituitary adenomas are typically hypodense compared with the normal gland on both unenhanced and contrast-enhanced images. Similarly, on MRI, 80–90% of microadenomas appear as a focal hypointense lesion compared with the normal gland

Figure 71.2 Pituitary microadenoma. Coronal T1-weighted spin echo images before (A) and after (B) gadolinium enhancement demonstrates a microadenoma (arrow) in the left side of the gland. The adenoma enhances less avidly than the rest of the gland.

on unenhanced T1-weighted images. After gadolinium, the adenoma enhances less avidly than the rest of the gland. The use of gadolinium increases the sensitivity of MRI for the detection of adenomas. Although up to 50% of microadenomas are hyperintense on T2-weighted sequences, overall T2-weighted sequences are less sensitive than T1-weighted sequences for detection of adenomas and are not used routinely. Other evidence of an adenoma includes focal erosion of the sella floor or focal convexity of the superior surface of the gland. Tilting of the pituitary stalk, which was once thought to be indirect evidence of an adenoma, is now considered to be unreliable as a diagnostic sign owing to the wide range of

Figure 71.3 Pituitary macroadenoma. Unenhanced coronal T1weighted spin-echo image demonstrates a pituitary macroadenoma. The presence of tumour tissue (white arrow) lateral to the black flow-void of the left carotid artery (*) indicates involvement of the cavernous sinus; the right cavernous sinus is also involved. A small suprasellar nodule of tumour deforms the left optic nerve (black arrow).

CHAPTER 71

•

IMAGING OF THE ENDOCRINE SYSTEM

may be hyperintense on T1- and T2-weighted sequences. The lack of contrast enhancement and absence of calcification usually allows distinction from craniopharyngiomas. Occasionally a Rathke’s cleft cyst can enlarge and compress the chiasm.

Suprasellar lesions

Figure 71.4 Pituitary macroadenoma with suprasellar extension. Unenhanced coronal T1-weighted spin-echo image demonstrates a pituitary macroadenoma extending into the right cavernous sinus (curved arrow) and extending upward into the suprasellar cistern to impinge on the optic chiasm (open arrows). The high signal from the tumour indicates the presence of recent haemorrhage.

normal variation7. Macroadenomas have similar signal characteristics to microadenomas, however, it is often not possible to identify any remaining normal pituitary tissue with larger lesions. Macroadenomas may grow upwards to compress the optic nerves and chiasm and/or may extend downward into the sphenoid sinus. It is important to diagnose cavernous sinus invasion by a pituitary adenoma, but unfortunately neither CT nor MRI has proved highly accurate in its preoperative detection. Encasement of the internal carotid artery is conclusive evidence. The identification of any tumour lying lateral to the lateral tangent of the intra and supracavernous internal carotid artery on coronal MRI is highly suggestive of cavernous sinus involvement. Cystic degeneration and haemorrhage are not uncommon in larger tumours but are also seen in microadenomas. Large pituitary adenomas are prone to develop infarction or haemorrhage owing to their tenuous blood supply. While this may present with fulminant and life-threatening pituitary apoplexy, it is far more frequently a subclinical phenomenon, being detected only at surgery or MRI. On MRI, over half of pituitary adenomas treated with bromocriptine will reveal evidence of haemorrhage that is clinically silent8. Postoperative imaging of the sella is difficult. Changes in its anatomy and that of its surroundings, the surgical technique used, inflammation and the quality and quantity of the implanted synthetic and organic materials are all factors that impede assessment of residual tumour. In practice, baseline imaging 3–4 months after pituitary surgery can give an accurate assessment of residual tumour. Rathke’s cleft cyst: Asymptomatic non-neoplastic pituitary cysts are found in approximately 11% of autopsies9. They arise from remnants of epithelium from Rathke’s pouch. They are typically in the centre of the gland, although they can be laterally located or even suprasellar. Many of these cysts are isointense with CSF, however, if they contain proteinaceous fluid they

The most common lesion in the suprasellar region is the pituitary macroadenoma with suprasellar extension (see intrasellar adenomas above). Other common lesions include craniopharyngiomas, meningioma, hypothalamic/chiasmatic glioma and aneurysm. Uncommon lesions include metastases, meningitis and granulomatous disease. Craniopharyngiomas are encapsulated cystic tumours of the sella or suprasellar region (Fig. 71.5), which originate from epithelial remnants of Rathke’s pouch. There is a bimodal age distribution; one peak between 5 and 10 years and the second smaller peak between 50 and 60 years. Typical presentation in childhood is with headache and raised intracranial pressure. Visual impairment and hypopituitarism may be present. MRI reveals a predominantly suprasellar mass that has both cystic and solid elements. The cystic portions of the tumour are characteristically hyperintense on T1- and T2-weighted sequences. The signal characteristics are due to the cholesterol crystals present in the cyst fluid, at surgery this fluid appears like ‘engine-oil’. The solid portions of the tumour enhance with gadolinium. These tumours are locally invasive, hard to resect and have a tendency to recur. Calcification is typical and is easier to identify on CT, appearing as an area of hypointensity on MRI. Less typical appearances of craniopharyngioma include purely solid masses and entirely intrasellar lesions. Craniopharyngiomas in the older age group are more likely to be solid, less likely to be calcified and have a better prognosis. Parasellar meningiomas can arise from any dural surface in the sellar region, This includes the anterior clinoid processes, the diaphragma sella, tuberculum or dorsum sellae, or

Figure 71.5 Craniopharyngioma. Enhanced sagittal T1-weighted spin echo images demonstrates a recurrent craniopharyngioma (curved arrow). The mass is suprasellar and clearly separate from the normal pituitary gland (open arrow). There is patchy enhancement but no cystic component is visible.

1709

1710

SECTION 9

•

MISCELLANEOUS DISORDERS

the cavernous sinuses. Their appearances on CT and MRI are similar to those of meningiomas elsewhere. A clear plane separating the tumour from the pituitary gland is often visible at high-resolution MRI. Meningiomas of the cavernous sinuses may cause constriction of the internal carotid artery, a feature not seen in pituitary adenomas. Primary hypothalamic tumours, such as gliomas, are uncommon. It is often difficult to identify whether a tumour in this site has arisen from the optic chiasm or the hypothalamus, particularly if it is a large mass. Chiasmatic gliomas are found mainly in children and may be associated with neurofibromatosis type 1. These tumours have a tendency to spread along the optic pathways, and T2-weighted sequences should be used to assess signal change along the optic radiation.These tumours are usually low grade, may be cystic, may or may not enhance and do not calcify. Germinomas are solid tumours arising from germ cells and typically occurring between the ages of 5 and 25 years. Although the pineal is the most common location for these lesions, they can also occur primarily in the suprasellar region and may be multicentric (Fig. 71.6). Diabetes insipidus and visual impairment are often the presenting feature. On MRI they appear as homogeneous solid masses often involving the pituitary stalk and optic chiasm. They are slightly hypointense on T1-weighted sequences and slightly hyperintense on T2-weighted sequences. Enhancement is prominent and homogeneous. Metastases to the sellar region account for approximately 1% of sellar masses and they most commonly arise from lung, breast, kidney, gastrointestinal tract, or nasopharyngeal tumours. There is usually clinical evidence of widespread metastatic disease. Metastases are usually suprasellar or combined intra- and suprasellar in position.They show varying degrees of enhancement and calcification is extremely rare. Vascular lesions: Aneurysms of the supraclinoid internal carotid artery are important to recognize. MRI shows a typical area of black flow void (on spin-echo sequences). Partially thrombosed aneurysms may have more complex signal characteristics. Aneurysms of the cavernous portion of the internal carotid artery can expand the cavernous sinus and extend into the pituitary fossa.

Suprasellar arachnoid cysts show smooth margins and have CSF signal intensity. They do not calcify or enhance. They may enlarge displace and compress the third ventricle. On fluid attenuation inversion recovery (FLAIR) and diffusion-weighted sequences they are identical to CSF within the ventricles. Hypothalamic (tuber cinerum) hamartoma is a congenital non-neoplastic heterotopia. The clinical presentation is usually that of precocious puberty although gelastic (laughing) seizures may occur. On MRI the hamartoma is seen as a sessile or pedunculated well-defined lesion arising from the tuber cinereum (Fig. 71.7). This is isointense to grey matter on T1weighted images and isointense (or slightly hyperintense) to grey matter on T2-weighted images10.There is no calcification or enhancement. Other causes of infundibular masses include Langerhans cell histiocytosis, sarcoid lymphoma and metastases. Involvement of the infundibulum with one of these processes usually enlarges it to a diameter greater than 4.5 mm or greater than that of the basilar artery1. CT or MRI shows an enhancing, usually homogeneous mass.

Figure 71.6 Germinoma. Coronal T1-weighted spin echo image after intravenous injection of gadolinium demonstrates homogeneously enhancing masses (arrow) in the (A) suprasellar and (B) pineal regions. The masses were isointense prior to contrast (not shown)

Figure 71.7 Hypothalamic hamartoma. Unenhanced sagittal T1-weighted spin echo image demonstrates a hypothalamic hamartoma (arrow). This arises from the inferior surface of the tuber cinereum and is isointense with other grey matter.

THE THYROID GLAND ANATOMY AND PHYSIOLOGY The thyroid gland is located in the anterior part of the neck with the lateral lobes lying on either side of the trachea joined across the midline by the isthmus. Each lobe measures approximately 40 × 20 × 20 mm. The pretracheal fascia binds the thyroid to the trachea, and it moves with the trachea on swallowing. An accessory lobe, the pyramidal lobe may be present in up to 50% of the population and usually arises from the thyroid isthmus11.

CHAPTER 71

The thyroid produces two major hormones, thyroxine (T4) and tri-iodothyronine (T3).The production of these hormones from the thyroid is regulated via thyroid-stimulating hormone (TSH) secreted by the anterior pituitary, which in turn is regulated by TSH-releasing hormone (TRH) released from the hypothalamus.

IMAGING TECHNIQUES Ultrasound Ultrasound (US) provides the best anatomical representation of the thyroid gland. Using high-resolution (7.5–10 MHz) probes, modern machines provide excellent spatial resolution and allow nodules as small as 2–3 mm to be detected.

Radionuclides Radionuclide imaging of the thyroid provides functional information that complements the anatomical information obtained from US. Several radionuclide-labelled agents are currently used to image the thyroid. Technetium-99m (99mTc) pertechnetate is the most widely used thyroid-imaging agent and is used as a first-line diagnostic tool in the evaluation of the thyroid. The pertechnetate anion is actively concentrated across the epithelium of a number of tissues via the same ion channel as the iodide anion. In addition to the thyroid gland, this includes the choroid plexus, salivary glands and gastric mucosa. Although pertechnetate is concentrated within the thyroid, unlike iodide it is not incorporated into thyroglobulin. Pertechnetate is excreted unchanged by the kidneys, salivary glands and gut. Before imaging, markers are placed on any palpable abnormalities, the cricoid and suprasternal notch. Imaging begins 20 minutes after the intravenous injection of sodium pertechnetate. Anterior, left and right oblique views are obtained and uptake of the injected activity is measured (normal 0.4–4.0%). The uptake of pertechnetate provides a map of the trapping function of thyroid tissue. Usually, uniform uptake of pertechnetate is seen throughout both lobes, but sometimes it concentrates in the isthmus or in the pyramidal lobe. Radioisotopes of iodine permit tracer studies of the entire metabolic pathway of iodine within the thyroid. This includes trapping, organification, coupling, hormone storage and secretion. Oral iodides, radiographic contrast media and iodinecontaining drugs compete with these radioisotopes and should be avoided before studies with radioisotopes of iodine. Iodine123 (123I) decays by electron capture with a single gamma-ray photon of 159 keV (which is ideal for imaging) and releases no beta particles. Its low radiation dose and short half-life (13 h) makes it an excellent radiotracer for functional evaluation of the thyroid. The disadvantages of 123I are its limited availability and high cost of production, as it requires a cyclotron.123I is given intravenously and imaging is performed 2–6 hours and 18–24 hours after administration, at which time most of the iodine within the thyroid is present as radio-iodotyrosine residues on thyroglobulin, reflecting hormonogenesis. [F-18]-2-fluoro-2-deoxyglucose ([F-18]FDG) has recently become the most commonly used radiopharmaceutical for clinical positron emission tomography (PET) studies in cancer

•

IMAGING OF THE ENDOCRINE SYSTEM

patients. FDG is transported into cancer cells like glucose. The more aggressive thyroid cancers, which often do not accumulate radio-iodine, do accumulate [F-18]FDG. The main clinical indication is in patients with residual or recurrent thyroid cancer with raised thyroglobulin and a negative radio-iodine whole-body survey. [F-18]FDG-PET imaging may also be helpful in anaplastic and some medullary carcinomas of the thyroid. Other radiotracers used in thyroid imaging:Thallium201 (201Tl) uptake is seen in various thyroid disorders, including cancers, subacute and chronic thyroiditis and Graves’ disease. Its main clinical application is in patients with residual or recurrent thyroid cancer with raised thyroglobulin and a negative radio-iodine whole-body survey. In 29% of such patients, 201Tl can detect residual disease. 99mTc-methoxyisobutylisonitrile (99mTc-MIBI) is also concentrated by thyroid tissue, and its role in thyroid imaging is similar to that of 201Tl. 99mTc-MIBI images have better resolution and are easier to interpret than 201Tl-scintigrams. However, because of considerable hepatic and abdominal activity with 99mTcMIBI, 201Tl is preferred if these areas are suspected sites of disease. 99mTc-tetrofosmin has similar imaging properties to 99m Tc-MIBI. 201Tl, 99mTc-MIBI and 99mTc-tetrofosmin do not require the withdrawal of thyroid-suppressive treatment unlike radio-iodine. In the investigation of medullary thyroid cancer 123I-metaiodobenzylguanidine (123I-MIBG), pentavalent 99m Tc-DMSA and indium-111-diethylenetriaminepentaacetic acid-octreotide (111In-DTPA-octreotide) may be used12. However, with rising calcitonin levels and negative 123I-MIBG, pentavalent 99mTc-DMSA and 111In-DTPA-octreotide imaging, [F-18]FDG-PET is useful.

Computed tomography and magnetic resonance imaging MRI and CT have a limited role in the evaluation of thyroid disease. They may occasionally be used to demonstrate the extent of local invasion or local recurrence of thyroid malignancy and to detect the presence of retrosternal and retrotracheal extension of thyroid enlargement.

THYROID DISORDERS Thyroid nodules One of the most common clinical problems is in patients presenting with a solitary palpable thyroid nodule. Its importance is due to the fact that thyroid cancer most often presents as a solitary thyroid nodule. However, thyroid nodules are extremely common: 4–7% of adults have palpable nodules13. The majority are benign, due to cysts, thyroiditis, adenomas, or colloid nodules. Investigation is directed towards detecting the 5–10% of palpable thyroid nodules that are malignant12. US, often the first imaging investigation, is of value in establishing whether a palpable nodule is solitary, cystic or part of a multinodular process such as multinodular goitres. In approximately 25% of patients, clinically solitary palpable nodules are shown by US to be multiple. Multinodular goitres have a lower incidence of malignancy (1–6%) than is associated with a single ‘cold’ nodule that is malignant (15–25% of cases).

1711

1712

SECTION 9

•

MISCELLANEOUS DISORDERS

US cannot reliably distinguish between benign and malignant thyroid nodules, but is however extremely effective in distinguishing between solid and cystic lesions. Purely cystic lesions with no soft tissue component are very rarely cancers, and in most institutions are managed with fine-needle aspiration. Certain US features increase the likelihood of malignancy of a thyroid nodule; notably, invasion into surrounding tissues, which results in loss of clear definition of the tissue planes. US is more sensitive than clinical examination in the detection of enlarged cervical nodes. Cervical lymph nodes, infiltrated by papillary carcinoma of the thyroid, may be entirely cystic and mimic other cystic masses of the neck, such as a branchial cyst, while calcification may be seen within nodes invaded by medullary carcinoma. Microcalcification smaller than 2 mm in diameter with acoustic shadowing suggests malignancy, as it is observed in about 60% of carcinomas but only 2% of benign nodules14. Ultrasound-guided fine-needle aspiration for cytology (FNAC) is the most reliable nonoperative method for obtaining a definitive diagnosis and is important in the selection of patients for surgery. Overall, the sensitivity of FNAC for the detection of malignancy in both cystic and solid lesions ranged from 90 to 100%. However, false-positives occur quite frequently, so the specificity is only 55% for solid nodules and 52% for cystic lesions15. Other drawbacks of FNAC are that it may be less sensitive for cystic papillary carcinomas15 and that it is not possible to differentiate between follicular adenoma and adenocarcinoma, as histologically the distinction relies on documenting capsular or venous invasion. Many centres now perform US-guided core biopsy for these reasons. Radionuclide thyroid scintigraphy is used to determine the functional status of nodules. Nodules may be ‘cold’, ‘warm’, or ‘hot’, depending on the uptake of tracer compared with normal thyroid. Pitfalls inherent in the use of these terms are well recognized: a superficial cold nodule may appear to be functioning because normal thyroid tissue behind it is visualized; or a superficial functioning nodule may appear hot because of photons emanating from the deeper, normal tissue. Thyroid cancers concentrate less radio-iodine (only 1%) than normal thyroid tissue and hence appear ‘cold’ in the presence of normal thyroid tissue (Fig. 71.8). Rarely, well-differentiated thyroid cancers have preserved trapping function, but virtually no capacity for iodine organification, therefore giving the appearance of function on early imaging (trapping phase) but becoming ‘cold’ on delayed images. This drawback can be overcome by acquiring delayed images of functioning nodules. Most ‘cold’ nodules are adenomas, colloid nodules, or foci of thyroiditis or rarely intrathyroid lymph nodes, lymphomas, or metastases. Approximately 10–25% of nonfunctioning (‘cold’) solitary nodules are malignant. ‘Cold’ nodules always require further assessment, usually by fine-needle aspiration cytology, core biopsy and/or excision biopsy. This combined imaging approach reduces significantly the rate of unnecessary surgery for benign nonfunctioning nodes. Functioning nodules account for approximately 30% of solitary nodules, and one third to one half are autonomous. In euthyroid individuals, the autonomous and nonautono-

Figure 71.8 Thyroid carcinoma presenting as a cold nodule. (A) 99mTcpertechnetate scintigram shows a ‘cold’ nodule in the lower pole of the right lobe of the thyroid gland. (B) Transverse ultrasound corresponding to the lesion in (A) shows an inhomogeneous nodule (arrows).

mous nodules can be differentiated by the administration of T3, which uniformly suppresses the uptake of iodide by both the TSH-dependent nodule and the extranodular normal tissue, but not by the autonomous nodule. The diagnosis of autonomous function can also be made by the demonstration of suppressed extranodular tissue on repeating the data acquisition following the administration of TSH. Approximately 10–20% of euthyroid subjects with autonomous nodules become hyperthyroid within 6 years, of whom half show an elevation in T3 only (‘T3 toxicosis’). Scintigraphic progression of the disorder (progressive suppression of extranodular uptake) parallels the decline in serum TSH concentration, but hyperthyroidism rarely occurs until a nodule reaches 30 mm in diameter. Malignant autonomous nodules usually produce hyperthyroidism. The ‘hot’ appearance of a nodule is not synonymous with autonomy; indeed, it often represents a spared focus of normal thyroid tissue (or a TSH-dependent nodule) in a gland otherwise involved in a destructive process (typically Hashimoto’s thyroiditis). Diagnosis can be made by the T3-suppression test. Multinodular goitre is generally, but not always, a benign disorder, and dominant cold nodules, especially enlarging ones, require further evaluation by aspiration biopsy. Autonomous foci are common and, when extensive, ultimately result in hyperthyroidism (toxic multinodular goitre, Plummer’s disease), one of the nodules often being dominant.

Thyroid cancer Thyroid cancer is uncommon and accounts for only 0.5% of all cancer deaths12. All forms of thyroid cancer occur more frequently in women than men and the peak incidence is seen in young women aged 25–35 years. Radiation exposure is the most important aetiological factor, but other risk factors include iodine excess, genetic predisposition and alcohol excess. The main histological subtypes of thyroid cancers are papillary, follicular, medullary and anaplastic cancers. Papillary cancers are the most common type and account for 50–80% of all thyroid malignancy. They are characteristically low-grade tumours with good prognosis. These tumours

CHAPTER 71

are usually multicentric histologically and spread to lymph nodes early, but distant metastases are rare. Metastatic lymph nodes may be normal in size, cystic, calcified, haemorrhagic, or may contain colloid. Papillary carcinomas concentrate radio-iodine. Follicular cancers account for 10–40% of thyroid cancers. They are slow-growing but have a tendency to spread via the bloodstream and disseminate to bone, liver, or lung and only rarely metastasize to regional lymph nodes. Hurthle cell tumours comprise 20% of follicular carcinomas; these are more common in women, they usually do not take up radio-iodine and metastases are frequently undetected. In such cases, 201Tl, 99m Tc-MIBI or 99mTc-tetrofosmin may be useful. Anaplastic thyroid cancers are undifferentiated tumours and are the most malignant. They have a poor prognosis and are usually treated with radiotherapy rather than surgery. Lymphatic metastases occur in the majority of patients and are often necrotic. These neoplasms do not concentrate radioiodine. On US they are often hypoechoic, and on CT punctate calcification and necrosis are frequently present. Medullary carcinomas originate from the parafollicular C cells. Medullary carcinomas are usually solitary lesions that may invade locally, spread to regional nodes, or undergo haematogenous spread to lung, bone, or liver and may be sporadic or familial. The familial form occurs with multiple endocrine neoplasia (MEN) type II syndrome (Table 71.8) or with other endocrine neoplasms. Patients may present with ectopic hormone production, and calcitonin is normally elevated. Unlike papillary and follicular tumours, medullary tumours do not concentrate radio-iodine but radionuclides specific for neuroendocrine tissue such as 123I-MIBG and somatostatin analogues may be used to evaluate patients with medullary cancer (Fig. 71.9). Lymphomas account for 10% of thyroid malignancies and are mostly non-Hodgkin type. They occur in one third of patients with Hashimoto’s thyroiditis. Imaging cannot distinguish between lymphoma and thyroiditis16. Patients usually present with a rapidly enlarging thyroid mass. They tend to involve the nodes and spread to the gastrointestinal tract.

Table 71.8

MULTIPLE ENDOCRINE NEOPLASIA (MEN)

MEN Type I Parathyroid adenoma or hyperplasia (hyperthyroidism) Pancreatic islet cell tumours Pituitary adenoma

MEN Type IIA Parathyroid hyperplasia Adrenal medullary tumour Medullary thyroid cancer

MEN Type IIB Medullary thyroid cancer Adrenal medullary tumour Marfanoid features, multiple cutaneous and mucosal neuromas, neurofibromas and gastrointestinal ganglioneuroma

•

IMAGING OF THE ENDOCRINE SYSTEM

Figure 71.9 MEN Type IIB. 111In-pentetreotide image shows a patient with medullary carcinoma of the thyroid (curved arrow) and bilateral glomus tumours (small arrows) and a paraganglioma (large arrow) in lower mediastinum.

Thyroid lymphoma presents most commonly as a solitary nodule (80%) or as multiple nodules.The nodule is hypoechoic on US17, hypodense on CT, and of high signal intensity on T2weighted MRI18. Necrosis and calcification is uncommon. Metastatic tumour presenting on the thyroid is uncommon. Clinically, the most commonly detected primary is renal cancer19. The diagnosis of thyroid malignancy is confirmed by cytology or histology from a thyroid nodule or after the removal of a cervical lymph node. Imaging is useful to delineate the extent of the tumour, detect metastatic disease and for followup of treated patients to detect local recurrence and distant spread. The main factors determining the prognosis include the patient’s age, size of the tumour, degree of differentiation, invasion and nodal and distant metastases.

Treatment and follow-up of thyroid cancer Patients with thyroid cancer are usually treated with a total thyroidectomy. This enables subsequent follow-up with 131Iscintigraphy and serum thyroglobulin measurement. Subtotal operations are followed by 131I-ablation therapy. Clinical evaluation, repeat scintigraphy and serum thyroglobulin measurement are performed 6–12 months later to assess the adequacy of ablation and to detect and treat metastatic disease under elevated TSH drive. Whole-body iodine imaging is performed after discontinuing thyroid hormone (for 4 weeks for T4 or for 2 weeks for T3) and establishing that TSH concentrations are >30 µmol L−1 20. It is also possible to image these patients without discontinuing thyroid medication using recombinant TSH, although slightly better results are achieved

1713

1714

SECTION 9

•

MISCELLANEOUS DISORDERS

by suspending conventional thyroid medication.131I-wholebody imaging can identify most functioning metastases (Fig. 71.10), which are usually in the neck, lungs, or bone20. Lung metastases are clearly seen on 131I-imaging. The scintigraphic detection of metastases is increased when the patient is imaged about 7 days after a therapeutic dose of 131I. The sensitivity and specificity of 131I-scintigraphy for detecting metastatic disease due to differentiated thyroid cancers is 42–62% and 99– 100%, respectively21,22. Both the sensitivity and specificity of serum thyroglobulin are in the range 55–78%, depending on the presence or absence of anti-thyroglobulin antibody23. In patients with raised thyroglobulin concentrations and a negative 131I-whole-body survey, the sensitivity of 201Tl, 99mTcMIBI, 99mTc-tetrofosmin and [F-18]FDG-PET is 49%, 88%, 89% and 94% respectively23–25. High specificity has also been reported with [F-18]FDG-PET. In patients with medullary carcinoma, following removal serum calcitonin concentrations are used to diagnose recurrence. When elevated, pentavalent 99mTc-DMSA or 111In-DTPA-octreotide or 123 I-MIBG have been used for the detection and documentation of recurrent or residual disease12. In cases of 123I-MIBG-positive disease, therapy with 131I-MIBG can be instituted. In the detection of recurrent disease, MRI has a positive predictive value of 82% and a negative predictive value of 86% in the neck26. Like CT, false-positive results can be caused by inflammatory lymphatic hyperplasia and granulation tissue. Nevertheless, there is a distinct advantage to using MRI rather than US or CT in the detection of recurrent disease26. US is useful during follow-up to detect neck masses or lymph nodes.

Screening for thyroid cancer Following radiation therapy to the neck, there is an increased risk of papillary and follicular thyroid carcinoma. For example, in patients who have received radiotherapy for Hodgkin’s disease, the risk of developing thyroid cancer is about 15%27. The carcinoma is unlikely to develop within 5 years of exposure to radiation; therefore, beyond 5 years, US provides a useful,

sensitive and relatively inexpensive means of screening a population at risk for thyroid cancer27.

Hyperthyroidism Hyperthyroidism is most commonly due to Graves’ disease or autonomous toxic nodules (Plummer’s disease). Other causes of hyperthyroidism include iodide-induced hyperthyroidism, thyroiditis and rarely hyperthyroidism secondary to a pituitary, hypothalamic, or ectopic source such as struma ovarii. Hyperthyroidism is diagnosed clinically and biochemically. Imaging may be helpful in identifying the cause and in the follow-up of patients managed with radio-iodine. Graves’ disease is an autoimmune disorder in which a circulating immunoglobulin, produced largely by intrathyroidal lymphocytes, stimulates the thyroid gland by binding to TSH receptors. If this sequence of events occurs in a thyroid that is normal to begin with, the result is diffuse toxic goitre. Scintigraphy reveals high and uniform uptake of tracer in the enlarged gland (Fig. 71.11). At US the gland is often diffusely hyperechoic without discrete nodules.There is a characteristic increase in thyroid vasculature on colour Doppler (see Fig. 71.11) especially in patients with active hyperthyroidism. Approximately 5–10% of the population have underlying thyroid disease to start with, and Graves’ disease may be superimposed upon solitary thyroid nodules, multinodular goitre, or Hashimoto’s thyroiditis, and the imaging will reflect this. The combination of Graves’ disease and a functioning nodule is called Marine– Lenhart syndrome. This entity is important as 50% of the nodules are radioresistant: this gives rise to unusual scintigraphic findings following 131I-therapy. Toxic multinodular goitre (Plummer’s disease) may present either as a toxic goitre with hyperfunctioning nodules and suppressed stroma or as diffuse toxicity of the thyroid stroma with intervening nodules. A single toxic nodule shows high uptake of tracer and the remaining normal thyroid tissue may show poor or virtually no activity on the images. Toxic multinodular goitre that is not associated with any of the autoimmune manifestations of Graves’ disease is the most common cause of toxicosis in elderly patients.

Hypothyroidism

Figure 71.10 Well-differentiated thyroid carcinoma imaged 3 days after 131Iodine (150 mCi, 5.55 GBq). Imaging shows increased uptake in thyroid bed, right hilar region and left upper chest (arrows). Salivary activity is also noted which is a normal feature (curved arrow).

Hypothyroidism may be primary (endogenous thyroid disorder), secondary (decreased TSH production due to pituitary disease) or tertiary (decreased TRH secretion from hypothalamic disorders). The most common causes of primary hypothyroidism include surgery or radiation ablation for Graves’ disease or thyroid cancer, primary idiopathic hypothyroidism, Hashimoto’s thyroiditis (see below), iodine deficiency and congenital hypothyroidism. In neonates with suspected primary congenital hypothyroidism based on screening (heel-stick blood levels of T4 and TSH), the 99mTc-pertechnetate scan is a useful adjunct procedure. The 99mTc-pertechnetate helps to confirm the presence of a normal gland in patients with a falsepositive screening test and to differentiate between the three sub-groups of primary congenital hypothyroidism: (A) nonvisualization of functioning thyroid tissue; (B) hypoplastic and ectopic glands; and (C) dyshormonogenesis28.

CHAPTER 71

•

IMAGING OF THE ENDOCRINE SYSTEM

ment of large nodules within the gland increase the possibility of superimposed non-Hodgkin lymphoma. Silent or painless thyroiditis is another autoimmune disorder in which radio-iodine uptake is initially very low, but in the recovery (hypothyroid) phase it may be normal to elevated. The scintigram in mild, subacute (viral) thyroiditis generally shows patchy uptake in an enlarged and tender thyroid gland. With extensive disease, the uptake is markedly suppressed by both the disruptive effect of inflammation and the ensuing hyperthyroidism. Riedel’s thyroiditis is characterized by a fibrosing reaction that destroys the thyroid. The thyroid is hypoechoic on US and hypodense on CT29. On MRI, the thyroid is of low signal intensity on T1- and T2-weighted images29. In acute suppurative thyroiditis the affected portion of the gland is enlarged and heterogeneous on cross-sectional imaging. Focal abscess may develop with disease progression, and there may be associated myositis and cellulitis in the adjacent neck. Imaging is used to exclude a fistula (from the pyriform sinus) as a cause of the thyroiditis.

Congenital disorders

Figure 71.11 Graves’ disease. (A) Longitudinal ultrasound with colour Doppler shows diffusely increased vascularity of the thyroid gland. (B) 99mTc-pertechnetate imaging shows a diffusely increased tracer uptake.

Thyroiditis Hashimoto’s thyroiditis (chronic lymphocystic thyroiditis) is an autoimmune destructive disorder. On scintigraphy, there is no typical pattern. Uptake is most commonly heterogeneous and patchy, but may be uniformly increased or decreased. At US the gland shows a variety of patterns. The thyroid may be normal or enlarged in size and is diffusely abnormal with a heterogeneous echo texture. There may be numerous poorly defined hypoechoic regions separated by fibrous strands. Discrete nodules and adjacent adenopathy are less commonly seen. End-stage disease may show a fibrotic gland that is small, ill defined and heterogeneous. US is performed to follow up Hashimoto’s disease to check for malignancy and the develop-

During fetal development, the thyroid descends from the base of the tongue (foramen caecum) to the low anterior neck. The thyroid is attached to the tongue base by the thyroglossal duct. During the caudal descent, the duct elongates and then degenerates and abnormal development or aberrant descent results in a spectrum of congenital lesions. Ectopic thyroid tissue may occur anywhere from the foramen caecum, along the thyroglossal tract to the pretracheal, mediastinal, pericardiac area. Residual tissue at the foramen caecum is called a ‘lingual thyroid’.A lingual thyroid is rarely associated with the presence of a normally positioned thyroid gland. Following total thyroidectomy, an ectopic focus within the thyroglossal tract near the centre of the hyoid bone is commonly seen, and must not be mistaken for a metastasis of thyroid cancer. Hypoplasia or agenesis of a lobe almost always involves the left side30; conversely, a small right lobe is usually the result of disease rather than anatomical variation. Incomplete degeneration of a thyroglossal cyst may result in a persistent fistulous tract or thyroglossal duct cyst. More than half of these cysts have normal thyroid tissue in their walls11. Thyroglossal cysts are infrahyoid in 65%, suprahyoid in 20% and in the hyoid region in 15%11. Uncomplicated cysts have the appearances of a typical cyst on cross-sectional imaging. Occasionally, if the cyst is infected, haemorrhagic, or has a high protein content, it may appear as hyperdense on CT, of high signal intensity on T1- and T2-weighted MRI (Fig. 71.12), and have internal echo at US. Rarely, thyroglossal cyst may undergo malignant change usually into a papillary carcinoma, which may be suspected if a soft tissue component exists within or around the cyst.

THE PARATHYROID GLANDS ANATOMY AND PHYSIOLOGY The parathyroid glands are derived from the endoderm of the third and fourth pharyngeal pouches. There are usually four

1715

1716

SECTION 9

•

MISCELLANEOUS DISORDERS

Pseudo-pseudohypoparathyroidism is a rare disorder exhibiting the same skeletal syndrome as pseudohypoparathyroidism but with normal blood chemistry.

Hyperparathyroidism

Figure 71.12 Thyroglossal duct cyst. (A) Axial T1-weighted spin-echo and (B) T2-weighted fast spin-echo images show a well-defined high signal intensity thyroglossal duct cyst just anterior to the hyoid bone.

glands, which lie behind the thyroid, but there may be as few as two or as many as six. Approximately 2.5% of the population have five glands. Each parathyroid gland measures approximately 5 × 3 × 1 mm and weighs 35 mg. Compared with the thyroid, the parathyroid glands are hypoechoic on US and have lower attenuation on CT than the thyroid. Approximately 80% of abnormal parathyroid glands are situated in and immediately adjacent to the thyroid. Ectopic glands may be found anywhere from the level of the third pharyngeal pouch, just behind the angle of the mandible at the level of the hyoid bone, down to the aortic root. Up to 5% may be in the posterior mediastinum. Parathormone (PTH), together with vitamin D and calcitonin, maintains normal serum calcium levels by acting on bone, kidney and intestine.

PARATHYROID DISORDERS Hypoparathyroidism Hypoparathyroidism is most commonly due to accidental removal of the parathyroid glands or their ablation by radio-iodine therapy. Less frequently, hypoparathyroidism occurs in association with autoimmune disorders, including Addison’s disease, pernicious anaemia and chronic thyroiditis. Rarely, aplasia or hypoplasia of the parathyroids occurs. Congenital causes of hypoparathyroidism include di George syndrome, which is associated with agenesis of the thymus. The skeleton is usually normal in patients with hypoparathyroidism and the principal radiological manifestation of the disease is calcification of the basal ganglia and areas of the cerebrum and cerebellum. Ectopic calcification may also occur in periarticular or subcutaneous tissue. Pseudohypoparathyroidism is an inherited disorder in which the biochemical finding of hypocalcaemia and hyperphosphataemia are similar to that found in hypoparathyroidism but do not respond to treatment with parathormone. Affected individuals are short in stature with short metacarpals, metatarsals and phalanges. The fourth and fifth metacarpals in particular are affected. The teeth are hypoplastic with defective enamel, and calcification may be present in the basal ganglia, cerebellum and skin. Children are moon-faced and may have deformities due to chronic tetany.

In hyperparathyroidism, increased concentrations of parathyroid hormones cause hypercalcaemia and increased urinary calcium excretion. This may result in nephrocalcinosis, renal calculi and bone disease. The causes of primary hyperparathyroidism are, in order of frequency: single adenoma (80%), hyperplasia (15%), carcinoma (4%) and multiple adenoma (1%). Very rarely, ectopic excretion occurs, particularly with bronchial carcinoma. Secondary hyperparathyroidism results from parathyroid hyperplasia in response to hypocalcaemia. Tertiary hyperparathyroidism follows prolonged hypocalcaemia (as in chronic renal failure) and an autonomous adenoma forms. The diagnosis of hyperparathyroidism is based on biochemical findings. Imaging is performed purely to localize parathyroid tumours and abnormal glands before surgery. The effects of hyperparathyroidism on the skeleton are discussed elsewhere.

Localization of abnormal parathyroid gland The need for preoperative localization depends on the approach of individual surgeons. An experienced surgeon can achieve a cure rate of approximately 95% in primary hyperparathyroidism following initial exploration of the neck. Under these circumstances, preoperative localization is unlikely to add significantly to these results, or to reduce the operation time31. Other authors recommend unilateral neck dissection if one abnormal gland is demonstrated on preoperative images, only proceeding to bilateral exploration if the second ipsilateral gland appears abnormal at the time of operation32. A unilateral neck dissection allows a considerable reduction in operating time. It is advocated that a localization study should be performed in those patients who require reoperation, when the surgical cure rate without preoperative localization is 30–40% lower than in patients being operated on for the first time. Even though an increased proportion of abnormal glands will lie in an ectopic site, the pathological gland is still most likely to be located in the neck. In these patients, imaging before surgery improves the results by identifying the location of a parathyroid tumour33,34. On US, parathyroid adenomas typically appear as oval, well-defined anechoic or hypoechoic masses posterior to the thyroid gland. They tend to be of the order of 10 mm in diameter but can grow to a very large size (4–5 cm). As glands become larger, they are more likely to be multilobulated and to contain echogenic areas, cysts and calcifications. There are several problems in the localization of parathyroid adenomas by US. Retropharyngeal, retro-oesophageal and mediastinal parathyroid glands, which account for 5–15% of all glands, are not easily accessible by US. Occasionally, in primary hyperplasia, one gland may predominate; simulating a solitary parathyroid adenoma. An exophytic thyroid nodule can be indistinguishable from a parathyroid nodule and a

CHAPTER 71

PTH assay on fine-needle aspiration of the nodule will often resolve this problem. Other false-positive US diagnoses can result when hyperplastic lymph nodes, prominent longus colli muscles and sympathetic ganglia are misinterpreted as parathyroid adenomas. Intraoperative US has been shown to be of value during reoperation for primary hyperparathyroidism, assisting the surgeon in localizing the abnormal glands and reducing the operating time. Computed tomography does not have any advantage over ultrasound in detecting an adenoma located in a normal parathyroid site, but is very useful for detecting abnormalities lying in sites inaccessible to ultrasound, such as behind the trachea, in the mediastinum, or in an undescended inferior parathyroid gland in the carotid sheath. These are all sites where adenomas are missed at initial neck exploration. CT is, therefore, a very useful localization procedure when performed before reoperation for persistent hyperparathyroidism. It is usually successful at identifying low cervical adenomas, but only 50% of mediastinal adenomas are detected. Magnetic resonance imaging is increasingly being used for localizing abnormal parathyroid glands. Parathyroid adenomas typically are isointense to muscle on T1-weighted images, and of high signal intensity on T2-weighted images. However, approximately 30–40% of abnormal parathyroid glands do not

•

IMAGING OF THE ENDOCRINE SYSTEM

have conventional signal characteristic35 and must be identified on the basis of abnormal size and location. Radionuclide imaging with 201Tl or 99mTc-MIBI, using a subtraction technique whereby the image of the thyroid gland as shown by 99mTc or 123I is removed, is a noninvasive technique for tumour localization (Fig. 71.13).99mTcMIBI can also be used on its own to detect adenomas (Fig. 71.14); early (15 min post injection) and delayed (90 min post injection) images are acquired for this purpose.99mTcMIBI is more sensitive than 201Tl in preoperative data acquisition and in patients undergoing reoperation. Cumulative analysis of 99mTc-MIBI scintigraphy in patients with hyperparathyroidism demonstrates a sensitivity of 90% for single adenomas, 55% for abnormal glands in patients with multiglandular disease and 75% for recurrent hyperparathyroidism; the specificity for primary adenomas has been shown to be 98%36. The availability of 99mTc-MIBI single photon emission tomography (SPET)/CT and intraoperative detection of adenomas using hand-held gamma probes is becoming more popular, especially in patients who have undergone previous neck exploration. Autografted normal parathyroid glands can also be imaged with 99m Tc-MIBI37. In patients where surgical re-exploration is being planned, but with negative conventional scintigraphy, 11 C-methionine PET/CT should also be considered. Figure 71.13 Parathyroid adenoma. (A) Left top figure shows 99mTc-pertechnetate scintigram, which shows uptake by thyroid tissue only. Right top figure shows 99mTcMIBI with uptake in both thyroid and parathyroid tissue. Left bottom figure shows the subtraction image locating the parathyroid adenoma behind the lower pole of the right lobe of the thyroid gland. (B) Longitudinal ultrasound confirms the parathyroid adenoma (arrow) posterior to the lower pole of the right lobe of the thyroid gland (arrowheads).

1717

1718

SECTION 9

•

MISCELLANEOUS DISORDERS

less sensitive than MRI in localizing mediastinal parathyroid adenomas in most studies.

PANCREATIC AND GASTROINTESTINAL ENDOCRINE DISORDERS DIABETES MELLITUS Diabetes mellitus is by far the most common endocrine disorder. It is characterized by chronic elevation of blood glucose level, for which a cause is not usually identified. Occasionally, diabetes mellitus is secondary to damage to the pancreas such as chronic pancreatitis and haemochromatosis. The diagnosis of diabetes is made on fasting blood-glucose measurement. Patients with diabetes have many radiological features (Table 71.9). These are attributable to complications from diabetes.

Table 71.9

RADIOLOGICAL FEATURES OF DIABETES MELLITUS

Cardiovascular system

Figure 71.14 Multiple parathyroid adenoma. Left top figure shows the early 99mTc-MIBI scintigram, which shows uptake by thyroid and parathyroid tissue. The retention of tracer in parathyroid is significantly longer than normal thyroid tissue and the right top figure shows retention of 99mTc-MIBI in two parathyroid adenomas (arrows) near the lower pole of the thyroid gland. Left bottom figure shows conventionally obtained 99mTc-pertechnetate scintigram 20 minutes after injection for comparison.

Arterial calcification: aortic, iliac, femoral, carotid Myocardial infarction Heart failure Atheromatous plaques, stenosis Obstruction

Respiratory system

Infections

Genitourinary system

Pyelonephritis Papillary necrosis Emphysematous interstitial nephritis Perirenal abscess

Angiography and venous sampling are usually used as a last resort in difficult cases.The ablation of parathyroid adenomas has been reported following successful selective angiography.

Nephrosclerosis—small contracted kidneys Cystitis, cystitis emphysematosa Neuropathic bladder

The accuracy of imaging techniques in localizing parathyroid disease There is great variation in the reported accuracy of imaging techniques in the localization of parathyroid disease. The overall sensitivity for identifying parathyroid adenomas with 99m Tc-sestamibi ranges from 80 to 95%, which is significantly higher than that of 201Tl34,36,38,39. The specificity is high at 91–98%. The sensitivity of radiopharmaceutical techniques to demonstrate hyperplasia is lower (55%) than that for the detection of parathyroid adenomas36,40, and the sensitivity for the detection of pathology at reoperation is 75%36. Modern US using high-resolution (7.5–10 MHz) probes is only slightly less sensitive than nuclear medicine; the highest reported sensitivity for identifying an adenoma is 82%26,41. The specificity, however, is high, ranging from 78 to 100% 40 . As with all imaging techniques, the sensitivity of US falls to around 60% in identifying hyperplastic glands and to about 50% when localizing parathyroid pathology following a previous operation34,41. The sensitivity of CT in identifying parathyroid adenomas is similar to that of US, with specificities ranging from 92 to 95%34,38. After failed exploration of the neck, sensitivity of 44–50% has been reported.The results for MRI are similar to those for CT34,38,40. CT appears to be

Ureteric reflux Gas in pelviureteric systems Vas deferens calcification Skeletal system

Osteomyelitis, secondary to overlying skin necrosis May have gas in soft tissues Charcot’s or neuropathic foot Periarticular bone fragmentation with soft tissue calcification Resorption of terminal tufts of terminal phalanges Localized osteoporosis, especially feet Subluxations Fragmentation of femoral heads Generalized retarded maturation of bones

Gastrointestional system

Gastroparesis Acute gastric dilatation Colonic pseudo-obstruction Small bowel malabsorption pattern

Gall bladder

Emphysematous cholecystitis

Neonates

Large infants Hyaline membrane disease

CHAPTER 71

Many are related to the duration of the disease and are the result of degenerative cardiovascular lesions in the eyes, heart, brain, kidneys, bladder, peripheral nervous system and musculoskeletal system. Other radiological abnormalities are due to infective, neuropathic, or metabolic changes. The role for radiology in patients with diabetes depends on the medical problems of the individual patients.

ISLET-CELL TUMOURS Islet-cell tumours are a range of rare neoplasms of neuroendocrine origin arising in or near the pancreas. The normal islets cells of Langerhans in the pancreas contain B cells (which secrete insulin), A cells (glucagon), D cells (somatostatin), D1 cells (pancreatic polypeptide, PP) and D2 cells (vasoactive intestinal peptide, VIP). The majority (85%) of islet-cell tumours secrete one or more of these hormones, or other substances not normally found in the adult pancreas (although often present in the fetal pancreas), notably gastrin.

Functioning tumours Functioning tumours are named according to the main hormone produced, and patients usually present with the clinical features of this excessive hormone secretion. These tumours may be either benign or malignant, solitary or multiple, or form part of the multiple endocrine neoplasia syndrome (MEN). The diagnosis is almost always made biochemically and the role of imaging is to localize the tumour prior to surgery and to look for evidence of malignancy or metastases. Insulinomas are the most frequent functioning pancreatic tumours and account for 50% of all islet-cell tumours42. They produce spontaneous hypoglycaemia, which is relieved by glucose and associated with high levels of plasma insulin and C peptide. Insulinomas are usually solitary and benign: 10% are malignant, 10% are multiple and 4% are associated with MEN I, in which case multiple tumours are usually present. Insulinomas are usually very small: 90% are less than 2 cm and 50% less than 1.3 cm in diameter43. When multiple, the individual lesions are smaller (mean diameter 9 mm versus 13 mm)44. Malignant insulinomas tend to be larger than benign ones (2.5–12 cm diameter)43. Gastrinomas are the second most common functioning islet-cell tumour of the pancreas accounting for 20% of all islet-cell tumours42. They give rise to Zollinger–Ellison syndrome, which comprises increased gastric acid secretion, diarrhoea and peptic ulceration. The diagnosis is established by the demonstration of a raised fasting serum-gastrin level with high basal gastric acid output. Over 90% of gastrinomas are found in the ‘gastrinoma triangle’ bounded by the third part of the duodenum, the neck of the pancreas and the porta hepatis45. Gastrinomas are multiple in 20–40% of patients and often extrapancreatic, with 20% found in the duodenum. Gastrinomas are frequently malignant with metastatic spread occurring to the liver and local lymph nodes. They tend to be small: 38% of pancreatic and all duodenal tumours are less than 1 cm in diameter at diagnosis. One third of cases are associated with MEN I in which multiplicity is the rule and there is a tendency to recurrence.

•

IMAGING OF THE ENDOCRINE SYSTEM

Glucagonomas are rare, cause nonketogenic diabetes mellitus and a characteristic migrating, necrolytic rash, as well as stomatitis, diarrhoea and venous thrombosis.The tumours have an average diameter of 4–7 cm at diagnosis and are malignant in approximately 60% of cases. VIPomas produce watery diarrhoea, hypokalaemia and achlorhydria (Verner–Morrison syndrome). The tumours range in diameter from 2 to 7 cm at diagnosis. The site of the tumour is intrapancreatic in 90% of cases, with the remainder (mainly gangliomas or ganglioneuroblastomas) originating in the sympathetic chain or adrenal medulla. Most extrapancreatic tumours are benign but 50% of pancreatic VIPomas are malignant. Somatostatinomas are very rare, slow growing tumours, which produce the clinical triad of gallstones, diabetes mellitus and steatorrhoea.These tumours arise either from the pancreas or duodenum. Duodenal somatostatinomas occur in association with neurofibromatosis and are usually periampullary in position.

Nonfunctioning tumours Nonfunctioning tumours account for 15% of pancreatic neuroendocrine tumours. They do not usually present until the tumour is large enough to cause symptoms from mass effect. The tumours tend to be large, solid and malignant but are usually slow growing.

Imaging islet-cell tumours Transabdominal ultrasound usually demonstrates an isletcell tumour as a well-circumscribed mass of lower echogenicity and finer echo texture than normal pancreas (Fig. 71.15). There may be a hyperechoic rim and larger tumours may show evidence of necrosis or calcification. Occasionally masses, especially gastrinomas, may be hyperechoic. Some lesions are isoechoic and are seen due to a hypoechoic halo around the lesion or due to the distortion of the gland. Liver

Figure 71.15 Insulinoma. Transverse ultrasound through the upper abdomen shows a small hypoechoic insulinoma (arrow) in the neck of the pancreas.

1719

1720

SECTION 9

•

MISCELLANEOUS DISORDERS

metastases are mostly hyperechoic or heterogeneous, and larger lesions may show evidence of cystic degeneration43. The overall detection rate for insulinomas is approximately 25–63% but can be up to 70% if the pancreas is adequately visualized. The reported detection rate for gastrinomas is worse, with only 30% of lesions identified46. The sensitivity is better for the detection of intrapancreatic gastrinomas than extrapancreatic lesions. The low sensitivity of transabdominal US is due to overlying bowel gas, deep location of the pancreas and limited resolution of with 3.5–5 MHz probes. Endoscopic ultrasound (EUS) is superior to transabdominal US due to the use of high-frequency probes (7.5–12 MHz) placed in close proximity to the pancreas and duodenum (Fig. 71.16). In experienced hands, it has a high sensitivity in detecting tumours of the pancreatic head but has been less successful for lesions of the pancreatic tail and duodenum wall. Sensitivities as high as 79–100% have been reported for intrapancreatic lesions47. The procedure is relatively invasive and the equipment and expertise not widely available. However, in many centres, it is now considered a first-line investigation in patients with biochemical evidence of insulinoma. Intraoperative ultrasound (IOUS), like EUS, is highly operator-dependent, but in experienced hands it is extremely effective for the identification of insulinomas at surgery. It is much less effective, however, for the identification of gastrinomas, since it is unable to identify extrapancreatic tumours of this type. Intraoperative US does not replace preoperative localization of islet-cell tumours but is used as an adjunct to palpation. Computed tomography is the most widely used method for localizing islet-cell tumours. Most islet-cell tumours are isodense on unenhanced CT and will not be seen without intravenous administration of contrast medium unless they are large enough to deform the pancreatic outline. Occasional small tumours may be apparent as a hyperdense mass. Calcification may occur in up to 20% and is more common in malignant than benign tumours. Larger lesions may show central necrosis and are also more likely to calcify. On contrast-enhanced CT, islet-cell tumours are generally seen as a rounded area that enhances more than the surrounding pancreas (Fig. 71.17), although hypodense lesions do occur48. Hepatic metastases are seen as low-attenuation areas

Figure 71.16 Insulinoma on endoscopic ultrasound. (A) Endoscopic ultrasound shows a hypoechoic tumour (arrow) in the tail of the pancreas, which is seen on the (B) contrast-enhanced CT as a bulge in the tail of the pancreas (arrowheads).

on unenhanced CT. Central necrosis is common in larger lesions and calcification may occur. Following intravenous contrast medium injection, the lesions, which are highly vascular, enhance. Solitary lesion may be indistinguishable from a haemangioma. Biphasic enhanced CT, with thin primary collimation in the arterial phase followed by parenchymal phase imaging has a very high sensitivity (94%) for the detection of small pancreatic islet-cell tumours when compared with multiphasic CT without thin sections (57%) or conventional CT imaging (29%)42,49. Combining biphasic thin collimation CT with EUS has been reported to have a sensitivity of 100%42. Magnetic resonance imaging is increasingly being used in the localization of islet-cell tumours. Studies comparing MRI and CT have shown that MRI has greater sensitivity than CT, particularly for smaller islet-cell tumours. In general, islet-cell tumours are of lower signal intensity on T1-weighted images and higher signal intensity on T2-weighted images than normal pancreas (Fig. 71.18)50. Arteriography With the development of cross-sectional imaging, arteriography has decreased in importance in localizing pancreatic neuroendocrine tumour. Meticulous arteriographic technique should include selective arteriography of the superior mesenteric artery, the splenic artery, the gastroduodenal artery, the dorsal pancreatic artery, the common hepatic artery and any accessory hepatic arteries. Hepatic arteriography is performed both to demonstrate the arterial anatomy for the surgeon and to identify any liver metastases. Islet-cell tumours typically appear as a well-circumscribed blush in the arterial and early venous phase (Fig. 71.18). Abnormal feeding vessels may be seen in large tumours.Vascular liver lesions may be seen in the case of a malignant tumour.This technique is very operator-dependent. This may partly account for the widely varying reported sensitivities of selective angiography, which for insulinoma is between 43% and 94%51 and for gastrinomas between 64% and 100%. In patients with multiple tumours, the sensitivity for individual lesions is lower. Lesions may be missed on angiography because they are too small or hypovascular or are obscured by the blush of adjacent bowel or spleen. False positives result from misinterpretation of blush of adjacent bowel, splenunculus, or angioma. Venous sampling allows functional radiological localization of insulinomas and gastrinomas. In arterial stimulation and venous sampling (ASVS), selective pancreatic arterial injections of a secretagogue (calcium for insulinoma and secretin for gastrinoma) are made and hepatic venous effluent is sampled.When arteries supplying the tumour are injected, there is a rise in the hepatic venous hormone concentration. ASVS detects lesions not seen on cross-sectional studies52. Arteriography is usually combined with ASVS to give a high sensitivity for the detection and localization of functioning tumours. In cases where arteriography is negative, ASVS will localize the tumour to a region of the pancreas. ASVS is less reliant on operator expertise and consequently has a high reported sensitivity of 80–90%51. Transhepatic portal venous sampling (TPVS) is performed by transhepatic catheterization of the right portal vein. Samples for hormonal analysis are obtained from the splenic

CHAPTER 71

•

IMAGING OF THE ENDOCRINE SYSTEM

Figure 71.17 Insulinoma. (A) Contrast enhanced spiral CT demonstrating a small brightly enhancing insulinoma (curved arrow) in the head of the pancreas. (B) Axial T1-weighted spin echo MR image and (C) T2-weighted fast spin echo image show the corresponding lesion (arrow) on MRI of low signal intensity on the T1- and high signal intensity on the T2-weighted images. (D) A coeliac arteriogram in the same patient shows the vascular blush of the insulinoma (arrow).

vein, superior and inferior mesenteric veins, portal trunk and pancreatic veins. TPVS only localizes the tumour to a region of the pancreas. TPVS is an invasive procedure with a risk of hepatic arteriovenous fistulae, haemobilia and superior mesenteric artery occlusion and significant mortality. It is reserved

for patients with suspected insulinoma if all other preoperative tests are negative51 and is seldom used nowadays. Scintigraphy for neuroendocrine tumours of the pancreas and bowel can be performed using radiolabelled somatostatin analogues and vasoactive intestinal peptide (123I-VIP). The

Figure 71.18 A small nonfunctioning islet-cell tumour in a patient with MEN type 1 syndrome. Axial T1-weighted spinecho (A), T2-weighted fast spin-echo (B), and fat-suppressed T1-weighted spin echo image before (C) and after (D) intravenous gadolinium administration show the typical signal intensity pattern for islet-cell tumour on the various sequences in addition to the marked enhancement following intravenous gadolinium.

1721

1722

SECTION 9

•

MISCELLANEOUS DISORDERS

radiolabelled somatostatin analogue 111In-pentetreotide is used to image a variety of somatostatin receptor-positive tumours. The main advantages of scintigraphy are its ability to image the whole body and to detect tumours or their metastases as small as 1 cm in diameter, especially in areas not under clinical suspicion. These imaging techniques can also be used to monitor the effects of therapy53. These small tumours can also be located at surgery using hand-held gamma probes. Table 71.10 summarizes the proportions of tumours that show uptake on scintigraphy. More recently 68Ga-octreotate, 11 C-5-hydroxytryptophan and 11C-l-dihydroxyphenylalanine PET/CT have been introduced to improve tumour diagnosis and localization.

Localization of islet-cell tumours The localization of islet-cell tumour presents a challenge to the radiologist as several imaging techniques are capable of demonstrating the tumour and no one technique is superior to the others. A rational approach to the localization of these tumours requires careful consideration of cost, sensitivities and availability of the imaging techniques. In most cases, initial imaging with a combination of US and CT or MRI will demonstrate the tumour and hepatic metastases. If these tests are negative or equivocal arteriography (with ASVS) is the next line of investigation. If the tumour remains undetected, further investigation depends on local expertise. EUS has emerged as a highly sensitive test for the detection of small pancreatic tumours and may also demonstrate extrapancreatic gastrinomas. IOUS is a useful adjunct to palpation at the time of surgery. Somatostatin-receptor imaging is useful in somatostatin receptor positive tumour not detected on other imaging techniques.

CARCINOID TUMOURS Carcinoid tumours are neuroendocrine neoplasms occurring most commonly within the gastrointestinal system, less frequently in the respiratory tract and thymus and very rarely at other sites. Hormonal products can be demonstrated within cellular neurosecretory granules in all carcinoid tumours, but systemic manifestations of hormonal secretion are rare. A small number of patients may present with symptoms, related to the release of 5-hydroxytryptamine (5-HT) and other vasoactive metabolites, which include skin flushing, intestinal hypermotility bronchoconstriction, cardiac valvular fibrosis

and liver enlargement. Carcinoid syndrome is present in cases with hepatic metastases. Carcinoid tumours may also produce ACTH resulting in Cushing’s syndrome, most commonly due to lung carcinoid. The resulting occult ectopic ACTH can often be extremely difficult to localize as sampling of the systemic veins is usually negative54. Carcinoid tumours often have a relatively indolent course and prolonged survival is not uncommon even in the presence of distant metastases with reported survival rates of more than 80% at 5 years. Gastrointestinal carcinoids are found in the appendix (45%), the small bowel (25%) and colorectum (25%). Symptoms result from local fibrosis with obstruction or from endocrine effects; the latter usually from an ileal carcinoid after it has metastasized to the liver. Gastrointestinal tumours are sometimes diagnosed on barium examinations as intramural tumours or constrictive lesions. On CT, although the primary tumour may not be seen, mesenteric infiltration and fibrosis produce a characteristic appearance of a soft tissue mass in the mesentery with a radiating stellate pattern of linear densities. Carcinoid liver metastases can be shown by US or CT or MRI, angiography being reserved for patients in whom hepatic arterial embolization is planned. Hepatic metastases are typically hypervascular and are optimally depicted during the arterial phase of a contrast-enhanced CT or MRI (Fig. 71.19) and may calcify.These tumours may appear isoattenuating when compared with adjacent hepatic parenchyma during the portal venous phase of enhancement. In our experience, MRI following administration of MnDPDP, a liver-specific contrast agent, increases the ability to detect neuroendocrine liver metastases with a high level of interobserver reproducibility. Bronchial carcinoids represent 1% of pulmonary neoplasms and appear as well-circumscribed nodules or masses on chest radiographs. Patients with bronchial carcinoid present with chest symptoms such as cough, recurrent infection, or haemoptysis. Approximately 2% of bronchial carcinoid tumours present with an endocrine disturbance, and approximately 1%

Table 71.10 PROPORTIONS OF TUMOURS THAT SHOW UPTAKE ON SCINTIGRAPHY Type of tumour Gastrinomas

111

In-pentetreotide (%) 80

123

I-VIP (%) –

Glucagonomas

95

–

Carcinoid

86

85

Insulinomas

61

82

Somatostatinomas VIPomas

100

–

80

100

Figure 71.19 Carcinoid liver metastases. ‘Arterial’ phase contrast enhanced CT shows multiple highly vascular lesions typical of metastases from a primary neuroendocrine tumour.

CHAPTER 71

of cases of Cushing’s syndrome are secondary to bronchial carcinoid through the ectopic production of ACTH or corticotropin-releasing hormone (CRH). Bronchial carcinoids represent the most common source of ectopic ACTH production in patients presenting with Cushing’s syndrome54.The causative bronchial carcinoid is usually very small and is often difficult to identify. Biochemically, the syndrome is often difficult to distinguish from a pituitary cause of excess ACTH, hence the term occult ectopic ACTH production54. Thymic carcinoids usually manifest as an anterior mediastinal mass in adults. The majority of patients present with symptoms related to compression or invasion of the mediastinal structures. Approximately 50% of thymic carcinoids are functionally active resulting most commonly in Cushing’s syndrome. CT and MRI may be helpful in the evaluation of patients with occult, hormonally active lesions55. Radionuclide imaging plays an important role in the localization and management of these tumours.The ability of these tumours to concentrate 123I- or 131I-MIBG allows scintigraphy to be performed with a cumulative sensitivity of 71%56. The most common midgut carcinoids (appendix and distal ileum) probably concentrate the radiolabelled MIBG more readily than those in the foregut and hindgut. The main use for 123 I-MIBG scintigraphy is as a prelude to 131I-MIBG therapy, for which good palliative results are obtained.111In-pentetreotide has also been used to image these tumours, which has a higher sensitivity (86%) than MIBG. Recently, lutetium-177octreotate (a somatostatin analogue) has been used for targeted radionuclide therapy of these tumours.

THE ADRENAL GLANDS ANATOMY AND PHYSIOLOGY The right adrenal gland lies posterior to the inferior vena cava and the left adrenal gland anteromedial to the upper pole of the left kidney (Fig. 71.20). The adrenal glands have an arrowhead configuration, with a body and medial and lateral limbs.

Figure 71.20 Normal adrenal glands. Contrast enhanced CT of normal adrenal glands (arrows).

•

IMAGING OF THE ENDOCRINE SYSTEM

The normal adrenals extend 2–4 cm in the craniocaudal direction, and the thickness of the adrenal body and limbs does not exceed 9–10 mm and 4–5 mm, respectively57. The adrenal glands have an inner medulla and outer cortex. The outer zone of the cortex, the zona glomerulosa, secretes mineralocorticoids (largely aldosterone), which controls salt and water metabolism. The inner cortical zones, the zona fasciculata and zona reticularis, secrete glucocorticoids and androgens responding to ACTH from the hypothalamic-pituitary axis, which is in turn inhibited by the cortisol. The medulla, derived from the primitive neuroectoderm, consists of chromaffin cells, which produce catecholamines, adrenaline (epinephrine) and noradrenaline (norepinephrine). Their metabolites (excreted in urine) include vanillyl mandelic acid (VMA). The medulla is almost entirely localized within the body of the adrenal gland.

TECHNIQUES FOR INVESTIGATING THE ADRENAL GLANDS Ultrasound is seldom used in the detection of adrenal mass lesions in adults as the normal adrenal glands and small adrenal lesions remain difficult to visualize, especially in obese patients. Computed tomography is currently the technique of choice in detecting and evaluating adrenal lesions. Thin collimation allows accurate density measurements of adrenal lesions. Intravenous administration of contrast medium helps to distinguish the adrenal glands from adjacent vessels and to assess the vascularity of an adrenal mass, and to characterize individually detected masses. Magnetic resonance imaging offers an alternative method for imaging the adrenal gland. Imaging protocols usually consists of axial T1- and T2-weighted sequences. Imaging in the coronal and sagittal planes helps to identify invasion into adjacent structures by large mass lesions. Chemical-shift imaging sequences are performed to differentiate between benign and malignant adrenal masses. Intravenous contrast media are administered to differentiate between solid and cystic masses and to assess the vascularity of a lesion. Radionuclide imaging: Radiopharmaceuticals can be grouped into two categories—adrenocortical and adrenomedullary imaging agents. Radiolabelled analogues of cholesterol, such as 131I-6β-iodomethyl-19-norcholesterol (NP-59), are used to identify and localize masses that result in adrenal cortical dysfunction. MIBG, an analogue of guanethidine, is concentrated in sympathoadrenal tissue, and is thus used to image adrenomedullary disorders.131I- and 123IMIBG have the advantage of being able to screen the whole body for sympathomedullary tissue. Venous sampling is extremely accurate in the preoperative localization of the source of abnormal hormone secretion. It is, however, an invasive procedure, requiring long fluoroscopy times and high radiation exposures. Even in experienced hands, failure to catheterize the adrenal veins occurs in 10–30% of cases. Complications are frequently seen and include adrenal infarction, adrenal vein thrombosis, adrenal haemorrhage, hypotensive crises and adrenal insufficiency. Venous sampling

1723

1724

SECTION 9

•

MISCELLANEOUS DISORDERS

is best reserved for patients in whom findings are equivocal on cross-sectional imaging. Percutaneous adrenal biopsy has an accuracy of 80–90%58. Minor complications include abdominal pain, haematuria, nausea and small pneumothoraces. Major complications that necessitate treatment occur in 3–5% of cases and include pneumothoraces that require intervention and haemorrhage. Biopsy of an unsuspected phaeochromocytoma can be life-threatening due to a catecholamine ‘storm’. The complication rate varies with the approach used, but does not appear to be related to the size of the needle used58.

IMAGING OF ADRENAL LESIONS Adrenocortical hyperfunction59 Endogenous Cushing’s syndrome

Figure 71.21 Bilateral adrenal hyperplasia (arrows) in a patient with Cushing’s disease.

Endogenous Cushing’s syndrome is the result of an excess of ACTH production in 75% of cases and excess cortisol from an adrenal neoplasm in 25% of cases. The diagnosis of Cushing’s syndrome depends predominantly on the clinical and biochemical findings. The aims of imaging are to identify the source of excess ACTH production either from a pituitary tumour or an ectopic ACTH source, and to localize or exclude a unilateral adrenal mass. The appearance of the adrenal gland in Cushing’s syndrome will depend on the aetiology of the syndrome60,61. ACTH-dependent Cushing’s syndrome A pituitary cause (i.e., Cushing’s disease) is responsible in approximately 85% of cases and the remainder are due to ectopic ACTH secretion, which may be overt or occult. Overt ectopic ACTH production is usually clinically apparent and presents with a short history, lacks the clinical features of Cushing’s syndrome, and is most commonly from a small-cell lung cancer. Occult ectopic ACTH-secreting tumours present with a more chronic clinical picture, often indistinguishable from Cushing’s syndrome due to pituitary or adrenal adenoma. The source of occult ectopic ACTH is most commonly a bronchial carcinoid tumour; less frequently an islet-cell tumour of the pancreas, phaeochromocytoma, medullary thyroid carcinoma, or thymic carcinoid tumours are the source. In ACTH-dependent Cushing’s syndrome, the adrenals show hyperplastic changes (Fig. 71.21) in 71% of cases, with the glands largest in patients with ectopic ACTH production60.

Adrenocortical adenomas account for 10–20% of cases of Cushing’s syndrome. Functioning adrenal cortical adenomas responsible for Cushing’s syndrome are usually larger than 2 cm in diameter (Fig. 71.22). The contralateral gland is usually normal, but occasionally is atrophic because of reduced ACTH secretion. On cross-sectional imaging, most adenomas are homogeneous, although some may be heterogeneous because of focal areas of necrosis or haemorrhage, which is seen in larger masses. Adrenal carcinomas account for 10–15% of cases of Cushing’s syndrome. Most of these tumours exceed 6 cm in diameter at the time of presentation and are readily detected on US, CT, or MRI (Figs 71.23, 71.24). On CT, these tumours are heterogeneous, with areas of necrosis and calcification. Approximately 15% of carcinomas are less than 6 cm in diameter and may resemble adenomas. Adrenal carcinomas may invade adjacent organs, extend into the inferior vena cava, and involve lymph nodes and metastases to the liver, lung and bone. Functional imaging using [F-18]FDG-PET/CT can be used to map the full extent of spread. Rare adrenal causes of Cushing’s syndrome61 include primary pigmented nodular adrenocortical disease (which may be associated with cardiac myxomas, skin pigmentation and testicular tumours in Carney complex) and primary macronodular hyperplasia.

Figure 71.22 Cushing’s syndrome due to adrenocortical adenoma. (A) Contrast enhanced CT, (B) T1-weighted spin-echo and (C) T2weighted fast spin-echo images all showing a 3 cm adenoma (arrows) in the left adrenal gland.

CHAPTER 71

Figure 71.23 Adrenal carcinoma. Contrast-enhanced CT demonstrating a large partly calcified, inhomogeneously enhancing left adrenal cancer (arrows).

•

IMAGING OF THE ENDOCRINE SYSTEM

the most accurate means of localizing aldosteronomas, with a sensitivity approaching 100% and a positive predictive value of 90%65. Adrenal scintigraphy with a cholesterol-based radiopharmaceutical such as 131I-6β-iodomethyl-19-norcholesterol (NP59) and dexamethasone suppression can distinguish between a unilateral adenoma and bilateral hyperplasia. As with CT, the sensitivity of the technique is dependent in part on the size of the adenoma and its accuracy compares unfavourably with that of CT. Because of this, and also because the procedure is time-consuming and expensive, it is seldom used. In practice, when CT or MRI identifies a solitary adenoma, further imaging before surgery is unnecessary.Venous sampling is reserved for cases in whom both adrenal glands appear normal (as a micronodule may be undetected), where there are bilateral nodules that may be due to nodular hyperplasia or multiple adenomata, and where there is disagreement between the imaging and biochemistry findings.

Primary hyperaldosteronism

Adrenogenital syndrome

Primary hyperaldosteronism (Conn’s syndrome) results from a benign cortical adenoma (Fig. 71.25) in 79% of cases, bilateral adrenal hyperplasia (Fig. 71.26) in 20% of cases, and is rarely due to an adrenocortical carcinoma. Imaging plays an important role in distinguishing between the causes of primary hyperaldosteronism, as surgery should be considered for adenomas and bilateral hyperplasia is best treated medically. Adrenocortical adenomas (Conn’s tumour) tend to be small, with an average size of approximately 1.6–1.8 cm. These adenomas are of low density on CT (less than 10 HU) because of high cytoplasmic lipid content. They do not enhance significantly after intravenous injection of contrast medium and rarely calcify. On MRI, most adenomas appear slightly hypointense or isointense relative to the liver on T1-weighted images and slightly hyperintense or isointense relative to hepatic parenchyma on T2-weighted images (see Fig. 71.25)62. The sensitivity of CT in detecting these adenomas is 70–90%; its specificity exceeds 90%63. MRI is probably as sensitive and as specific as CT62,64. When successful, venous sampling is

Androgen-producing tumours of the adrenals are usually carcinomas, less commonly adenomas. They usually exceed 2 cm in diameter and have the same characteristics described elsewhere for adenomas and carcinomas66. In patients with virilism, imaging is used to detect a tumour in the adrenals, ovaries, or testes. The most common adrenal cause of excess androgens in childhood is congenital adrenocortical hyperplasia. This usually results from an inborn enzyme deficiency, which causes a partial block to the synthesis of adrenocortical steroids and an increase in the production of androgens. The compensatory elevation of ACTH results in gross enlargement of the adrenals, easily recognizable on CT or MRI. Long-term stimulation can result in transformation of a hyperplastic nodule into an adenoma or even a carcinoma58.

Adrenal medullary tumours Phaeochromocytomas are catecholamine-producing tumours that arise from the paraganglion cells anywhere in the

Figure 71.24 Adrenal carcinoma on MRI. (A) Axial T1-weighted spin-echo, (B) axial and (C) sagittal T2-weighted fast spin-echo images show a large left adrenal cancer (arrows). The size of the lesion and the high signal intensity on the T2-weighted images are typical of an adrenal carcinoma.

1725

1726

SECTION 9

•

MISCELLANEOUS DISORDERS

Figure 71.25 Conn’s syndrome due to an adrenocortical adenoma. (A) T1-weighted spin-echo and (B) T2-weighted fast spin-echo images showing the typical appearance of a small aldosterone-producing adenoma (arrow) in the right adrenal gland.

autonomic nervous system. Phaeochromocytomas originate in the adrenal medulla in 90% of cases; most extra-adrenal phaeochromocytomas arise in the paravertebral sympathetic ganglia, in the organ of Zuckerkandl or, rarely, in the bladder67. About 90% of adrenal medullary phaeochromocytomas are functional, but only 50% of sympathetic and 1% of para-

sympathetic tumours produce excess catecholamines. Ninety per cent of phaeochromocytomas are sporadic, the remaining 10% being associated with neuroectodermal disorders such as neurofibromatosis, tuberous sclerosis, von Hippel–Lindau syndrome, or multiple endocrine neoplastic syndrome. Multiple phaeochromocytomas occur in up to 10% of sporadic cases, but in as many as 30% when associated with a neuroectodermal syndrome. Approximately 10% of phaeochromocytomas are malignant. On unenhanced CT, phaeochromocytomas appear as round masses of similar density to surrounding soft tissue structures. Phaeochromocytomas frequently undergo marked necrosis, so that the mass may have a fluid-filled centre (Fig. 71.27). Calcification occurs occasionally and is usually speckled. In sporadic cases of phaeochromocytoma, the tumour tends to be large with an average size of 5 cm58. The lesions are often smaller when detected in patients with multiple endocrine abnormalities as they are often specifically sought. Phaeochromocytomas show intense enhancement following intravenous injection of contrast medium. Unlike ionic contrast media, adrenergic blockade to prevent hypertensive crisis is not required with non-ionic contrast media68. The overall accuracy of CT in detecting adrenal phaeochromocytomas is very high, with a sensitivity of 93–100% and a positive predictive value greater than 90%. On MRI, phaeochromocytomas are hypointense on T1weighted images and usually markedly hyperintense on T2weighted images (Fig. 71.28). Although their appearance on T2-weighted images is typical it is not specific, as there is some similarity in appearance with adrenal metastasis. Atypical signal intensity on T2-weighted images is seen in 35% of phaeochromocytomas58. CT and MRI are equally accurate for identifying adrenal phaeochromocytomas. On US, phaeochromocytomas are well-defined, round or ovoid masses with uniform reflectivity. Large tumours frequently undergo haemorrhage or necrosis with loss of homogeneity (see Fig. 71.27). The accuracy of US in detecting phaeochromocytomas is less than CT, MRI, or MIBG scintigraphy, particularly for small or ectopic tumours. MIBG scintigraphy is especially useful for the detection of ectopic phaeochromocytomas and also for the identification of metastatic or locally recurrent disease, because, unlike US, CT, or MRI, it is inherently a whole-body imaging technique. Phaeochromocytomas are seen as an abnormal focal area of increased activity. The sensitivity of MIBG scintigraphy in detecting functioning phaeochromocytomas is slightly less than 90%, and its specificity exceeds 90%. However, a positive MIBG study always requires correlation with CT or MRI. In metastatic disease, 131I-MIBG therapy provides an important palliative therapeutic option. Neuroblastoma—See paediatric chapter (Chapter 69).

Hypoadrenalism

Figure 71.26 Conn’s syndrome due to bilateral nodular adrenal hyperplasia. Contrast-enhanced CT shows small nodules in the adrenal glands (arrows).

Hypoadrenalism is usually due to primary idiopathic atrophy, in western countries, and imaging of the adrenals is seldom undertaken. However, imaging can be useful in identifying a potentially treatable cause. In patients with idiopathic atrophy, the glands are extremely small and may be difficult

CHAPTER 71

•

IMAGING OF THE ENDOCRINE SYSTEM

Figure 71.27 Phaeochromocytoma with necrosis. (A) A longitudinal US examination showing an adrenal mass (arrows) lying above the kidney (*) and behind the liver. (B) Contrast-enhanced CT shows a necrotic right adrenal tumour (arrow). Axial (C) T1-weighted spin-echo and (D) T2-weighted images showing the typical signal intensity pattern of a phaeochromocytoma (arrow) in the right adrenal gland. (E) Posterior coronal view from an 131I-MIBG scintigram showing the increased focal uptake in a phaeochromocytoma (arrow) in the right adrenal gland.

to identify. Small adrenal glands may also be seen in chronic tuberculosis (see below). Enlarged glands are most commonly due to tuberculosis, but occasionally may be due to amyloid, sarcoidosis, metastatic disease, haemochromatosis, fungal infection, or adrenal haemorrhage. If Addison’s disease develops suddenly then imaging techniques may be useful to confirm the presence of bilateral haemorrhage.

Figure 71.28 Phaeochromocytoma on MRI. Axial (A) T1-weighted spin-echo and (B) T2-weighted fast spin-echo images showing a right adrenal phaeochromocytoma (arrow). The very high signal intensity is typical of a phaeochromocytoma.

Adrenal disorders not resulting in altered hormonal activity Non-hyperfunctioning adrenal adenomas are detected incidentally in up to 5% of CT examinations58. The number and size of these nodules increase with age and are most common in obese diabetics and elderly women. These ‘silent’ adenomas are thought to represent non-neoplastic overgrowth of adrenocortical cells of the zona fasciculata. The appearance of non-hyperfunctioning and hyperfunctioning adenomas on cross-sectional imaging is similar61. Primary adrenocortical carcinomas are rare and highly malignant. Ninety per cent produce steroids but only half cause symptoms related to excess hormone production. Functioning carcinomas most commonly result in Cushing’s syndrome, but virilization or feminization and hyperaldosteronism may also occur. Most carcinomas are usually large (> 6 cm in diameter) at the time of diagnosis and are more commonly seen on the left; only about 10% are bilateral. The imaging characteristics of carcinomas are described above. Adrenal metastases are most commonly from tumours of the lung, kidney, melanoma, breast, digestive tract and ovary. Adrenal metastases very rarely result in hypoadrenalism. The presence of an adrenal mass in a patient with known malignancy does not necessarily indicate metastatic disease, because 40–50% are nonmetastatic and represent adenomas. Metastases tend to be larger than adenomas, less well-defined, inhomogeneous and have a thick, irregular enhancing rim after injection of intravenous contrast medium. They are more commonly unilateral than bilateral. On MRI, they are typically hypointense compared with the liver on T1-weighted images and relatively hyperintense on T2-weighted images. The adrenal glands may show diffuse

1727

1728

SECTION 9

•

MISCELLANEOUS DISORDERS

enlargement without evidence of metastases in patients with known malignancy69,70. Primary adrenal lymphoma is very rare, tends to be extranodal and has a poor prognosis. The adrenal glands are enlarged but maintain their adreniform shape. The adrenal glands are more commonly involved in widespread lymphoma, being detected on imaging in about 4% of cases of non-Hodgkin lymphoma, but at autopsy in up to 25% of cases. On CT, secondary adrenal lymphoma usually appears as solid homogeneous masses of soft tissue density, which enhance slightly after intravenous injection of contrast medium. Calcification is unusual without previous radiotherapy. Adrenal cysts are uncommon, usually unilateral and occur more often in women than men. Adrenal cysts are usually endothelial or epithelial in origin but may also be parasitic or pseudocysts. Pseudocysts are thought to represent the result of previous haemorrhage. Adrenal cysts have the same characteristics on imaging as cysts seen elsewhere in the body (Fig. 71.29), i.e., they are well-defined, thin-walled fluid-filled structures that show no enhancement after intravenous injection of contrast medium. Calcification is seen in 15% of adrenal cysts, which is often peripheral and curvilinear. On MRI, adrenal cysts are usually markedly hypointense on T1-weighted images and markedly hyperintense on T2-weighted images. However, the presence of proteinaceous material, infectious debris, or haemorrhage within the cyst can cause increased signal intensity on T1-weighted images. Adrenal myelolipomas are rare, benign neoplasms composed of fat and bone-marrow tissue in varying proportions. Most are asymptomatic and nonfunctioning and identified incidentally on imaging. However, there are isolated reports of endocrine abnormalities associated with myelolipomas71. Haemorrhage or necrosis within the tumour may cause pain and also lead to discovery of these lesions. The diagnosis of these lesions is based on the demonstration of fat within an adrenal mass (Fig. 71.30). However, the proportion of fat within the lesion may vary and those with only a small proportion of fat may be difficult to differentiate from other adrenal masses72. Nevertheless, the use of narrow collimation on CT will usually allow the demonstration of any fat that is present. The fat component of these lesions can also be readily shown on MRI, with signal intensity similar to fat on all pulse sequences. Infection: In granulomatous infections (tuberculosis, histoplasmosis, or blastomycosis), there is usually bilateral but asymmetric involvement of the adrenal glands. In active infection the adrenals are enlarged and inhomogeneous in density on CT particularly after contrast medium enhancement, when numerous small nonenhancing areas, of caseous necrosis, can often be seen (Fig. 71.31). Calcification may be seen in the acute phase or during healing and long-standing infection results in atrophic glands. Percutaneous biopsy is often required to confirm the diagnosis of granulomatous disease and to identify the organism. In AIDS patients with extrapulmonary Pneumocystis carinii infection, there may be punctuate or coarse calcification in the adrenal gland as well as in the spleen, liver, kidney and lymph nodes. Adrenal abscesses are rare, most being found in

Figure 71.29 Adrenal cyst. (A) Contrast-enhanced CT showing the typical appearance of a cyst (arrow) within the left adrenal gland. (B) T1-weighted spin-echo image shows the low signal intensity of a simple cyst (arrow). (C) T2-weighted fast spin-echo image shows the uniformly high signal intensity of a fluid filled lesion (arrow).

neonates with pre-existing adrenal haemorrhage. The abscess appears as a thick-walled cystic lesion. Adrenal haemorrhage is usually clinically silent. It is seen on CT in 2% of patients who sustain severe trauma. The most common CT feature of adrenal injury is a round or oval adrenal haematoma (83%), followed by uniform adrenal enlargement (9%) or diffuse irregular haemorrhage obliterating the gland (9%). Non-traumatic adrenal haemorrhage is usually associated with anticoagulants (or other bleeding disorders), stress caused by surgery, sepsis (in particular meningococcal) and hypotension. Approximately 80% of adrenal haemorrhages are unilateral and most commonly on the right side (Fig. 71.32). Adrenal haemorrhage usually resolves but occasionally the haematoma liquefies and persists as a pseudocyst. The size and appearance of an adrenal haematoma varies with its age. On CT, in the acute or subacute phase, the enlarged adrenals are of increased density

CHAPTER 71

Figure 71.30 Myelolipoma. (A) A longitudinal ultrasound showing a hyperechoic mass (arrow) lying above the right kidney (*) and behind the liver. (B) Contrast-enhanced CT showing large right adrenal mass (arrow) consisting predominantly of fat. (C) Corresponding T1-weighted MRI showing a mass consists predominantly of fat of similar intensity to the surrounding perirenal fat. (D) STIR sequence showing suppression of the signal from fat within the mass (arrow). (After Peppercorn P D, Reznek R H 1997 State-of-the-art CT and MRI of the adrenal gland. Eur Radiol 7: 822–836, with permission).

(50–70 HU) and follow-up studies show reductions in the density and size of the lesion. Calcification may develop a few months after the adrenal haemorrhage. On MRI, intracellular deoxyhaemoglobin in the acute phase causes signal loss on T1- and T2-weighted images. Subacutely, methaemoglobin causes bright signal intensity on T1-weighted images. Chronic adrenal haemorrhage often has low signal intensity on T1- and T2-weighted images because of calcification and haemosiderin deposition.

The ‘incidentally’ discovered adrenal mass Adrenal masses are found incidentally in up to 5% of abdominal CT examinations. In the assessment of an incidentally discovered adrenal mass, the initial step is biochemical evaluation.The type of biochemical abnormality then dictates further management. It is important to note that if there is no history of malignancy, a unilateral non-hyperfunctioning adrenal mass rarely, if ever, represents a metastasis unless there is a known underlying malignancy, and the differential diagnosis rests usually between adenoma and carcinoma. Though small adrenal masses are more likely to be benign, it has been shown that size alone is poor at discriminating between adenomas and non-adenomas73. The measurement of the CT attenuation value of the mass can be helpful. Unlike malignant lesions, benign adenomas generally contain intracellular lipid and thus have a lower CT attenuation value. An attenuation value of less than 10 HU on non-contrastenhanced CT has an extremely high specificity and acceptable

•

IMAGING OF THE ENDOCRINE SYSTEM

Figure 71.31 Adrenal gland tuberculosis. (A) Nonenhanced CT showing bilateral adrenal masses (curved arrows) with punctuate calcification in the left adrenal gland (open arrow). (B) Following intravenous injection of contrast media there are the typical nonenhancing areas within the gland (curved arrows) corresponding to multiple small caseating granulomata. (After Peppercorn P D, Reznek R H 1997 State-of-the-art CT and MRI of the adrenal gland. Eur Radiol 7: 822–836, with permission).

sensitivity for a benign cortical adenoma. Much greater overlap is seen after enhancement. However, malignant masses tend to have a rapid loss of attenuation value following intravenous contrast material injection.The percentage absolute and relative enhancements on the delayed imaging at 10 and 15 minutes have been shown to be useful in discriminating benign from malignant adrenal mass. Absolute percentage loss of enhancement of greater than 60% at 10 minutes has 100% specificity for the diagnosis benign adenoma74.

Figure 71.32 Adrenal haemorrhage. Non-enhanced CT showing bilateral enlargement due to adrenal haemorrhage (arrow). An area of high attenuation due to bleeding is demonstrated within the right adrenal gland (arrowhead).

1729

1730

SECTION 9

•

MISCELLANEOUS DISORDERS

Chemical shift imaging on MRI may also be used to characterize adrenal masses (Fig. 71.33). As with CT, the characterization is based on the detection of intracellular lipid in benign adenomas. When sufficient loss of signal has been demonstrated in adrenal masses on chemical shift imaging, a specificity of 100% has been reported for benign adenomas58. Cholesterol-based radiopharmaceutical imaging can also be used in evaluating an incidental adrenal mass. Cholesterol-based scintigraphy may show increased uptake in a nonfunctioning adenoma, whereas distorted, absent or reduced uptake suggests pathology such as carcinoma or other spaceoccupying lesions. Recently, PET with [F-18]FDG has also been reported to be able to differentiate between benign adenomas and metastases.

THE FEMALE REPRODUCTIVE SYSTEM The ovaries have both germinal and hormonal functions, which are modulated by hypothalamic and pituitary hormones. The adult ovaries function cyclically with changes in hormone levels. Ovarian endocrine dysfunction may present with failure to develop secondary sex characteristics, amenorrhoea, galactorrhoea, or infertility. Amenorrhoea, failure of ovulation, may be either primary (i.e., where no menstruation has previously occurred) or secondary (i.e., when there is loss of menstruation following previous periods). In the investigation of amenorrhoea, pregnancy must be excluded before recourse to other biochemical or imaging investigations. In primary amenorrhoea, imaging plays an important role in the exclusion of gross urogenital anatomical defects before proceeding to more complex biochemical assessment.

Figure 71.33 Chemical shift imaging to evaluate incidentally discovered adrenal masses. In-phase (A) and out-of-phase (B) images in a benign adrenal adenoma (arrow). The out-of-phase image shows marked loss of signal intensity within the adenoma indicating presence of intracellular lipid in keeping with a benign adenoma. In-phase (C) and out-of-phase (D) images in an adrenal metastasis from a colon cancer (arrow). The out-of-phase images show that there has been no loss of signal intensity and, therefore, the presence of intracellular fat has not been demonstrated.

Turner’s syndrome is the classical form of ovarian dysgenesis. These patients have a 45-XO karyotype and elevated folliclestimulating hormone (FSH) concentrations. They have ‘streak ovaries’ and a small uterus. Patients with Turner’s syndrome may also have a high arched palate, a webbed neck, widelyspaced nipples, poorly developed breasts, cubitus valgus, short fourth metacarpals, beaked vertebral bodies and osteoporosis. Pelvic gonadal dysgenesis may occur with a normal karyotype; in Swyer syndrome, an XY karyotype is associated with gonadal dysgenesis.The importance of recognizing this condition lies in the 25% incidence of associated ovarian tumours, such as dysgerminomas, which occur up to the age of 20 years. Annual pelvic US is recommended in these patients to detect any such ovarian tumour.

FUNCTIONING OVARIAN TUMOURS Ovarian tumours that cause virilization include Sertoli– Leydig cell tumours, Brenner tumours and the rare lipoid cell tumours. Granulosa-theca cell tumours are the most common oestrogen-producing tumours but may also release testosterone. Cross-sectional imaging techniques show these tumours are predominantly solid or solid and cystic75.There are no specific features to help distinguish between functioning ovarian tumours or with other ovarian neoplasm.

THE MALE REPRODUCTIVE SYSTEM Like the ovaries, the testes have both germinal and hormonal functions, which are modulated by the hypothalamic and pituitary hormones. Gonadotropin-releasing hormone (GnRH) from the hypothalamus controls FSH and luteinizing hormone (LH) secretion. In men FSH is responsible for initiating

CHAPTER 71

spermatogenesis and LH stimulates Leydig cells to produce androgens. Primary abnormalities of the testes are best imaged using US and MRI may be of benefit where US is inconclusive. Abnormalities of testicular function have different consequences depending on the phase of life in which they first become manifest, i.e., in fetal development, puberty, or adult life.

ANDROGEN EXCESS Primary tumours of either adrenal glands or testes, or congenital adrenal hyperplasia may give rise to excessive androgen production in childhood, resulting in precocious puberty. Primary germ cell tumour (seminoma or teratoma) or nongerminal (interstitial or Sertoli cell) tumour may also cause androgen excess.

ANDROGEN DEFICIENCY Androgen deficiency may result either from primary testicular failure or secondary to abnormality in the hypothalamic–pituitary axis (see above). In primary testicular disorders, serum testosterone levels are decreased and gonadotropin levels are elevated.The diagnosis is made on clinical grounds, biochemical assay, and the demonstration of chromosomal abnormality. Imaging has a minimal role in the diagnosis and management of these patients.

REFERENCES 1. Tien R D, Newton T H, McDermott M W, Dillon W P, Kucharczyk J 1990 Thickened pituitary stalk on MR images in patients with diabetes insipidus and Langerhans cell histiocytosis. Am J Neuroradiol 11: 703–708 2. Sato N, Ishizaka H, Yagi H, Matsumoto M, Endo K 1993 Posterior lobe of the pituitary in diabetes insipidus: dynamic MR imaging. Radiology 186: 357–360 3. Brooks B S, el Gammal T, Allison J D, Hoffman W H 1989 Frequency and variation of the posterior pituitary bright signal on MR images. Am J Roentgenol 153: 1033–1038 4. Kelly W M, Kucharczyk W, Kucharczyk J et al 1988 Posterior pituitary ectopia: an MR feature of pituitary dwarfism. Am J Neuroradiol 9: 453–460 5. Bartynski W S, Lin L 1997 Dynamic and conventional spin-echo MR of pituitary microlesions. Am J Neuroradiol 18: 965–972 6. Tsunoda A, Okuda O, Sato K 1997 MR height of the pituitary gland as a function of age and sex: especially physiological hypertrophy in adolescence and in climacterium. Am J Neuroradiol 18: 551–554 7. Ahmadi H, Larsson E M, Jinkins J R 1990 Normal pituitary gland: coronal MR imaging of infundibular tilt. Radiology 177: 389–392 8. Yousem D M, Arrington J A, Zinreich S J, Kumar A J, Bryan R N 1989 Pituitary adenomas: possible role of bromocriptine in intratumoral hemorrhage. Radiology 170: 239–243 9. Teramoto A, Hirakawa K, Sanno N, Osamura Y 1994 Incidental pituitary lesions in 1,000 unselected autopsy specimens. Radiology 193: 161–164 10. Hahn F J, Leibrock L G, Huseman C A, Makos M M 1988 The MR appearance of hypothalamic hamartoma. Neuroradiology 30: 65–68 11. Loevner L A 1996 Imaging of the thyroid gland. Semin Ultrasound CT MR 17: 539–562 12. Hussain H K, Britton K E, Grossman A B, Reznek R H 2004 Thyroid cancer. In: Husband J E, Reznek R H (eds) Imaging in Oncology. Taylor and Francis, London, pp 669–710

•

IMAGING OF THE ENDOCRINE SYSTEM

13. Hegedus L 2004 Clinical practice. The thyroid nodule. N Engl J Med 351: 1764–1771 14. Clark O H, Duh Q Y 1991 Thyroid cancer. Med Clin North Am 75: 211–234 15. de los Santos E T, Keyhani-Rofagha S, Cunningham J J, Mazzaferri E L 1990 Cystic thyroid nodules. The dilemma of malignant lesions. Arch Intern Med 150: 1422–1427 16. Shibata T, Noma S, Nakano Y, Konishi J 1991 Primary thyroid lymphoma: MR appearance. J Comput Assist Tomogr 15: 629–633 17. Hopkins C R, Reading C C 1995 Thyroid and parathyroid imaging. Semin Ultrasound CT MR 16: 279–295 18. Takashima S, Ikezoe J, Morimoto S H et al 1988 Primary thyroid lymphoma: evaluation with CT. Radiology 168: 765–768 19. Haugen B R, Nawaz S, Cohn A et al 1994 Secondary malignancy of the thyroid gland: a case report and review of the literature. Thyroid 4: 297–300 20. Maisey M N 1998 Thyroid imaging. In: Grossman A B (ed) Clinical Endocrinology. Blackwell Science Ltd, London, pp 298–314 21. Grunwald F, Schomburg A, Bender H et al 1996 Fluorine-18 fluorodeoxyglucose positron emission tomography in the follow-up of differentiated thyroid cancer. Eur J Nucl Med 23: 312–319 22. Dietlein M, Scheidhauer K, Voth E, Theissen P, Schicha H 1997 Fluorine18 fluorodeoxyglucose positron emission tomography and iodine-131 whole-body scintigraphy in the follow-up of differentiated thyroid cancer. Eur J Nucl Med 24: 1342–1348 23. Chung J K, So Y, Lee J S et al 1999 Value of FDG PET in papillary thyroid carcinoma with negative 131I whole-body scan. J Nucl Med 40: 986–992 24. Alam M S, Kasagi K, Misaki T et al 1998 Diagnostic value of technetium99m methoxyisobutyl isonitrile (99mTc-MIBI) scintigraphy in detecting thyroid cancer metastases: a critical evaluation. Thyroid 8: 1091–1100 25. Gallowitsch H J, Mikosch P, Kresnik E, Unterweger O, Gomez I, Lind P 1998 Thyroglobulin and low-dose iodine-131 and technetium99m-tetrofosmin whole-body scintigraphy in differentiated thyroid carcinoma. J Nucl Med 39: 870–875 26. Auffermann W, Clark O H, Thurnher S, Galante M, Higgins C B 1988 Recurrent thyroid carcinoma: characteristics on MR images. Radiology 168: 753–757 27. Healy J C, Shafford E A, Reznek R H et al 1996 Sonographic abnormalities of the thyroid gland following radiotherapy in survivors of childhood Hodgkin’s disease. Br J Radiol 69: 617–623 28. Sfakianakis G N, Ezuddin S H, Sanchez J E, Eidson M, Cleveland W 1999 Pertechnetate scintigraphy in primary congenital hypothyroidism. J Nucl Med 40: 799–804 29. Perez F F, Cordido C F, Pombo F F, Mosquera O J, Villalba M C 1993 Riedel thyroiditis: US, CT, and MR evaluation. J Comput Assist Tomogr 17: 324–325 30. Melnick J C, Stemkowski P E 1981 Thyroid hemiagenesis (hockey stick sign): a review of the world literature and a report of four cases. J Clin Endocrinol Metab 52: 247–251 31. Krubsack A J, Wilson S D, Lawson T L et al 1989 Prospective comparison of radionuclide, computed tomographic, sonographic, and magnetic resonance localization of parathyroid tumors. Surgery 106: 639–644; discussion 644–646 32. Lucas R J, Welsh R J, Glover J L 1990 Unilateral neck exploration for primary hyperparathyroidism. Arch Surg 125: 982–984; discussion 984–985 33. Al-Sayer H M, Krukowski Z H, Williams V M, Matheson N A 1985 Fine needle aspiration cytology in isolated thyroid swellings: a prospective two year evaluation. Br Med J (Clin Res Ed) 290: 1490–1492 34. Levin K E, Gooding G A, Okerlund M et al 1987 Localizing studies in patients with persistent or recurrent hyperparathyroidism. Surgery 102: 917–925 35. Lee V S, Spritzer C E 1998 MR imaging of abnormal parathyroid glands. Am J Roentgenol 170: 1097–1103 36. Pattou F, Huglo D, Proye C 1998 Radionuclide scanning in parathyroid diseases. Br J Surg 85: 1605–1616

1731

1732

SECTION 9

•

MISCELLANEOUS DISORDERS

37. Sierra M, Herrera M F, Herrero B et al 1998 Prospective biochemical and scintigraphic evaluation of autografted normal parathyroid glands in patients undergoing thyroid operations. Surgery 124: 1005–1010 38. Kohri K, Ishikawa Y, Kodama M et al 1992 Comparison of imaging methods for localization of parathyroid tumors. Am J Surg 164: 140–145 39. Denham D W, Norman J 1998 Cost-effectiveness of preoperative sestamibi scan for primary hyperparathyroidism is dependent solely upon the surgeon’s choice of operative procedure. J Am Coll Surg 186: 293–305 40. Price D C 1993 Radioisotopic evaluation of the thyroid and the parathyroids. Radiol Clin North Am 31: 991–1015 41. Miller D L, Doppman J L, Shawker T H et al 1987 Localization of parathyroid adenomas in patients who have undergone surgery. Part I. Noninvasive imaging methods. Radiology 162: 133–137 42. Noone T C, Hosey J, Firat Z, Semelka R C 2005 Imaging and localization of islet-cell tumours of the pancreas on CT and MRI. Best Pract Res Clin Endocrinol Metab 19: 195–211 43. King C M, Reznek R H, Dacie J E, Wass J A 1994 Imaging islet cell tumours. Clin Radiol 49: 295–303 44. Gorman B, Charboneau J W, James E M et al 1986 Benign pancreatic insulinoma: preoperative and intraoperative sonographic localization. Am J Roentgenol 147: 929–934 45. Power N, Reznek R H 2002 Imaging pancreatic islet cell tumours. Imaging 14: 147–159 46. London J F, Shawker T H, Doppman J L et al 1991 Zollinger-Ellison syndrome: prospective assessment of abdominal US in the localization of gastrinomas. Radiology 178: 763–767 47. McLean A M, Fairclough P D 2005 Endoscopic ultrasound in the localisation of pancreatic islet cell tumours. Best Pract Res Clin Endocrinol Metab 19: 177–193 48. Smith T R, Koenigsberg M 1990 Low-density insulinoma on dynamic CT. Am J Roentgenol 155: 995–996 49. King A D, Ko G T, Yeung V T, Chow C C, Griffith J, Cockram C S 1998 Dual phase spiral CT in the detection of small insulinomas of the pancreas. Br J Radiol 71: 20–23 50. Owen N J, Sohaib S A, Peppercorn P D et al 2001 MRI of pancreatic neuroendocrine tumours. Br J Radiol 74: 968–973 51. Jackson J E 2005 Angiography and arterial stimulation venous sampling in the localization of pancreatic neuroendocrine tumours. Best Pract Res Clin Endocrinol Metab 19: 229–239 52. Pereira P L, Roche A J, Maier G W et al 1998 Insulinoma and islet cell hyperplasia: value of the calcium intraarterial stimulation test when findings of other preoperative studies are negative. Radiology 206: 703–709 53. Sandler M P, Delbeke D 1993 Radionuclides in endocrine imaging. Radiol Clin North Am 31: 909–921 54. Vincent J M, Trainer P J, Reznek R H et al 1993 The radiological investigation of occult ectopic ACTH-dependent Cushing’s syndrome. Clin Radiol 48: 11–17 55. Hanson J A, Sohaib S A, Newell-Price J et al 1999 Computed tomography appearance of the thymus and anterior mediastinum in active Cushing’s syndrome. J Clin Endocrinol Metab 84: 602–605 56. Bomanji J, Conry B G, Britton K E, Reznek R H 1988 Imaging neural crest tumours with 123I-metaiodobenzylguanidine and X-ray computed tomography: a comparative study. Clin Radiol 39: 502–506 57. Vincent J M, Morrison I D, Armstrong P, Reznek R H 1994 The size of normal adrenal glands on computed tomography. Clin Radiol 49: 453–455

58. Sohaib S A, Reznek R H 2000 Adrenal imaging. BJU Int ; 86 (suppl 1): 95–110 59. Sohaib SA, Rockall AG, Reznek RH. Imaging functional adrenal disorders. Best Pract Res Clin Endocrinol Metab 2005; 19: 293–310 60. Sohaib S A, Hanson J A, Newell-Price J et al 1999 The CT appearance of the adrenal glands in adrenocorticotrophic hormone (ACTH)-dependent Cushings syndrome. Am J Roentgenol 172: 997–1002 61. Rockall A G, Babar S A, Sohaib S A et al 2004 CT and MR imaging of the adrenal glands in ACTH-independent Cushing syndrome. Radiographics 24: 435–452 62. Sohaib S A, Peppercorn P D, Allan C et al 2000 MRI in primary hyperaldosteronism (Conn’s syndrome). Radiology 214: 527–531 63. Lingam R K, Sohaib S A, Vlahos I et al 2003 CT of primary hyperaldosteronism (Conn’s syndrome): the value of measuring the adrenal gland. Am J Roentgenol 181: 843–849 64. Lingam R K, Sohaib S A, Rockall A G et al 2004 Diagnostic performance of CT versus MR in detecting aldosterone-producing adenoma in primary hyperaldosteronism (Conn’s syndrome). Eur Radiol 14: 1787–1792 65. Sheaves R, Goldin J, Reznek R H et al 1996 Relative value of computed tomography scanning and venous sampling in establishing the cause of primary hyperaldosteronism. Eur J Endocrinol 134: 308–313 66. Hanson J A, Weber A, Reznek R H et al 1996 Magnetic resonance imaging of adrenocortical adenomas in childhood: correlation with computed tomography and ultrasound. Pediatr Radiol 26: 794–799 67. Sahdev A, Sohaib A, Monson J P, Grossman A B, Chew S L, Reznek R H 2005 CT and MR imaging of unusual locations of extra-adrenal paragangliomas (pheochromocytomas). Eur Radiol 15: 85–92 68. Mukherjee J J, Peppercorn P D, Reznek R H et al 1997 Pheochromocytoma: effect of nonionic contrast medium in CT on circulating catecholamine levels. Radiology 202: 227–231 69. Vincent J M, Morrison I D, Armstrong P, Reznek R H 1994 Computed tomography of diffuse, non-metastatic enlargement of the adrenal glands in patients with malignant disease. Clin Radiol 49: 456–460 70. Jenkins P J, Sohaib S A, Trainer P J, Lister T A, Besser G M, Reznek R 1999 Adrenal enlargement and failure of suppression of circulating cortisol by dexamethasone in patients with malignancy. Br J Cancer 80: 1815–1819 71. Jenkins P J, Chew S L, Lowe D G, Reznek R H, Wass J A 1994 Adrenocorticotrophin-independent unilateral macronodular adrenal hyperplasia occurring with myelolipoma: an unusual cause of Cushing’s syndrome. Clin Endocrinol (Oxf) 41: 827–830 72. Musante F, Derchi L E, Zappasodi F et al 1988 Myelolipoma of the adrenal gland: sonographic and CT features. Am J Roentgenol 151: 961–964 73. Korobkin M, Brodeur F J, Yutzy G G et al 1996 Differentiation of adrenal adenomas from nonadenomas using CT attenuation values. Am J Roentgenol 166: 531–536 74. Al Hawary M M, Francis I R, Korobkin M 2005 Non-invasive evaluation of the incidentally detected indeterminate adrenal mass. Best Pract Res Clin Endocrinol Metab 19: 277–292 75. Sohaib S A, Husband J E, Reznek R H 2004 Ovarian cancer. In: Reznek RH, Husband JE, editors. Imaging in Oncology. Taylor and Francis, London, 429–466 76. Peppercorn P D, Reznek R H 1997 State-of-the-art CT and MRI of the adrenal gland. Eur Radiol 7: 822–836

CHAPTER

Reticuloendothelial Disorders: Lymphoma

72

Sarah J. Vinnicombe and Rodney H. Reznek

• • • • • • • • • •

Epidemiology Histopathological classification Staging, investigation and management Lymph node disease in lymphoma Extranodal manifestations of lymphoma Burkitt lymphoma Lymphoma in the immunocompromised Monitoring response to therapy Surveillance and detection of relapse Conclusion

nal large B-cell lymphoma, which has a peak incidence between 25 and 35 years and mantle cell lymphoma, which is more common in those over 60 years.

Infectious agents

EPIDEMIOLOGY

Oncogenic lymphotrophic DNA and RNA viruses have been implicated in many NHL types.The single most important agent in this regard is the herpes virus, Epstein–Barr virus (EBV). The EBV genome was first detected in cultured African Burkitt lymphoma cells and is known to be present in over 90% of such cases. EBV is important as a trigger for lymphoproliferations/lymphomas occurring in congenital immunodeficiencies, iatrogenically immunosuppressed organ transplant recipients, patients receiving maintenance chemotherapy and patients receiving combined immunosuppressive therapy for collagen disorders. EBV is also found in HD (mostly the mixed cellularity type) and patients who have had infectious mononucleosis are at increased risk of HD.The retrovirus human lymphotropic virus type 1 (HTLV-1) is implicated in the causation of adult T-cell lymphoma, which is endemic in certain areas of the globe, particularly in East Africa, the Caribbean, southwest Japan and New Guinea. Human herpesvirus 8 (HHV-8) has been implicated as a cause of primary effusion lymphoma, a rare type of large cell lymphoma confined to serous-lined body cavities, which occurs with highest frequency in the HIV-positive population. Bacterial overgrowth can also promote lymphomagenesis. In gastric lymphoma of mucosa-associated lymphoid tissue (MALT) type, Helicobacter pylori infection has been shown to be necessary for the development and early proliferation of the lymphoma.

Age

Immunosuppression

Hodgkin’s disease has a peak incidence in the 20–30 year age group, with a second, smaller peak in the elderly population that is becoming less evident, partly because of a refinement of the classification of NHL. The incidence of NHL increases exponentially with age after 20 years. The subtypes of lymphoma encountered differ in frequency between the adult and paediatric groups, with a strong bias towards precursor B- and Tlymphoblastic lymphoma and Burkitt lymphoma in childhood. Lymphomas with less typical age distribution include mediasti-

A variety of lymphomas are associated with pre-existing immunosuppression.The degree of immunosuppression is important in determining the lymphoma type that may emerge. In organ-specific autoimmune diseases, such as Hashimoto’s thyroiditis and Sjögren’s syndrome, extranodal marginal zone lymphomas of MALT type can arise within the affected organ. In severe immunodeficiency states, such as the congenital immunodeficiencies, AIDS and after organ transplantation, the lymphomas are very often EBV-driven large B-cell lymphomas.

The lymphomas are a complex group of diseases divided into two broad groups: Hodgkin’s disease (HD), or Hodgkin’s lymphoma (HL) and lymphomas other than Hodgkin’s disease, the non-Hodgkin’s lymphoma (NHL). The overall incidence of NHL is increasing globally, with age-adjusted incidence rates for NHL being highest in more developed countries. Recent figures from the USA give an incidence of 15.5 per 100 000 persons per year, representing a 73% increase since the early 1970s1. This is due in part to secondary lymphoma arising in the setting of AIDS, but a steady increase had been noted before the AIDS epidemic. In the corresponding period, the incidence of HD has remained relatively steady at around 3.7 per 100 000.

1734

SECTION 9

•

MISCELLANEOUS DISORDERS

In the setting of systemic collagen diseases, there is an increase in haematological malignancy and patients receiving immunosuppressive therapy for these conditions are at still greater risk. The types of haematological malignancy that arise are quite varied, but there is a slight excess of myeloma and small B-lymphocytic lymphoma.

Table 72.1 WHO CLASSIFICATION OF LYMPHOID NEOPLASMS

Genetic factors

Mature B-cell neoplasms

It is known that the risk of developing haematological malignancy is greater in patients with a family history of disease. This increased risk does not extend to the histological type or lineage of the tumours in question, such that one family member may have HD whereas a relative may have NHL or myeloid leukaemia.

Gender and race It has long been recognized that there is a slight predominance of NHL and HD in men (1.1–1.4 to 1).The incidence of NHL and HD varies by race, with a higher frequency in whites than blacks or Asians. Certain NHL types cluster according to race, for example the natural killer (NK) T-cell lymphomas are most frequently encountered in oriental populations.

Neoplasm B-CELL NEOPLASMS

Non-Hodgkin’s lymphoma Many of the difficulties that beset early taxonomists in the classification of NHL have been overcome with improved immunological and molecular methods of diagnosis, resulting in a new classification scheme, the Revised European– American Lymphoma (REAL) classification. This depended on a triad of morphology, immunophenotype and molecular methods for defining disease entities, as well as clinical features2, differentiating it from earlier morphologically based classifications. The scheme forms the backbone of the second WHO classification of tumours of haematopoietic and lymphoid tissues3. A summary of the WHO classification is given in Table 72.1. The WHO classification stratifies neoplasms by lineage into distinct disease entities and is a real advance in the ability to identify disease accurately and consistently.

Hodgkin’s disease Biological studies have shown that Hodgkin’s disease is a true lymphoma, many pathologists preferring the term Hodgkin’s lymphoma (HL). The defining cell of HD is the Reed– Sternberg cell, a large, binucleated blast cell. Mononuclear counterparts are called Hodgkin cells. The Reed–Sternberg cells and their variants are a minority population accounting for less than 5% of the volume of any involved lymph node. The balance is made up of reactive non-neoplastic T cells, histiocytes, plasma cells, eosinophils and fibroblasts, varying in proportion according to the histological subtype. Since the 1960s, so-called classical HD has been subclassified into four histological types, indicated below along with their relative frequencies in western populations:

85.0

Precursor B-cell neoplasms Precursor B-lymphoblastic leukaemia/lymphoma CLL/small lymphocytic lymphoma

6.7

B-cell prolymphocytic leukaemia Lymphoplasmacytic (lymphoplasmacytoid) lymphoma Splenic marginal zone lymphoma

1.2 1 (i.e. nonambulatory) advanced stage III or IV presence of >1 extranodal site of disease.

Treatment is nearly always with combination chemotherapy, since around 80% of patients will have advanced disease (stage III or IV) at presentation. Radiotherapy alone is considered for the small proportion of patients with stage I disease and no adverse factors, in whom surgical excision alone is considered inappropriate.

In HD, lymph node involvement is usually the only manifestation of disease, whereas in NHL nodal disease is frequently associated with extranodal sites of tumour. At presentation, differences exist between the patterns of lymph node involvement in HD and NHL. Lymph nodes tend to be larger in NHL than HD; indeed in nodular sclerosing and lymphocytedepleted HD, nodal enlargement may be minimal. Typically, involved nodes tend to displace adjacent structures rather than invade them, except in the case of diffuse large B-cell lymphoma, which is often locally aggressive.

Imaging nodal disease At present, size is the only criterion by which lymph nodes demonstrated on CT or MRI are considered to be involved, though clustering of multiple small nodes, for example within the anterior mediastinum or the mesentery, is suggestive. A maximum short-axis diameter of 10 mm is taken to be the upper limit of normal, depending upon the exact site within the neck, thorax, abdomen, or pelvis. Thus within the neck, the jugulodigastric node can measure up to 13 mm short axis diameter, whereas those in the gastrohepatic ligament and porta hepatis are considered abnormal if they measure more than 8 mm in diameter. Retrocrural nodes greater than 6 mm are taken as enlarged7 and in the pelvis the upper limit of normal is regarded as 8 mm8. Lymph nodes at some sites, such as the splenic hilum, presacral and perirectal areas, are not usually visualized on cross-sectional imaging and, when demonstrated, are likely to be abnormal whatever the size. Enlarged lymph nodes in both HD and in NHL are usually homogeneous and of soft-tissue density on CT. They may show mild or moderate uniform enhancement after intravenous injection of contrast medium. Calcification is uncommon but may

CHAPTER 72

be seen on post-treatment scans. Necrosis is rarely seen in large nodal masses in both HD (particularly nodular sclerosing HD) and aggressive NHL, more frequently after treatment. On MRI, lymph nodes are easily identified as relatively low to intermediate signal intensity masses on T1-weighted images, of intermediate to high signal on T2-weighted images and which may have very high signal intensity on short-tau inversion recovery (STIR) sequences. Though the signal intensity of involved nodes and the presence of necrosis do not appear to have much prognostic significance, there is some evidence that within large lymphomatous masses, heterogeneous T2 signal at MRI or heterogeneous enhancement at CT are associated with a worse outcome.

Choice of imaging technique The ability of CT to demonstrate enlarged lymph nodes throughout the body and detect associated lesions in softtissue structures, together with a high level of reproducibility, have all contributed to CT becoming the treatment of choice for the staging and follow-up of lymphoma. Ultrasound will readily show lymph node enlargement in the coeliac regions, splenic hilum and porta hepatis9. Frequently, however, the entire retroperitoneum cannot be shown, limiting its value in staging.Typically on ultrasound, lymphomatous nodal involvement produces uniform hypoechoic lobulated masses. The pattern of vascular perfusion as demonstrated by power Doppler interrogation may suggest the diagnosis, lymphomatous nodes having rich central and peripheral perfusion. The main value of ultrasound in lymphoma lies not in routine staging but in confirming that a palpable mass is in fact nodal, or in solving specific problems related to detection of involvement of tumour in the liver, spleen, or kidney. Although the accuracy of MRI in detecting lymph node involvement is equal to that of CT (and indeed better in some areas such as the supraclavicular fossa and within the pelvis), it has no particular advantage over CT in this respect and its role is essentially adjunctive, to solve problems in the identification of lymph node pathology or in monitoring response to treatment10–12. Although lymphangiography was shown to be equal, or slightly superior, to CT for detecting nodal lymphoma in the 1980s, it had many disadvantages, which have led to the almost universal abandonment of the technique. Radioisotope studies using gallium-67 (Ga-67) and the positron emitter, 2[F-18]fluoro-2-deoxy-d-glucose ([F-18]FDG), can demonstrate viable tumour cells within nodes with high sensitivity. However, the accuracy of gallium-67 is dependent upon several factors including cell type, location and the size of the lesion. Its accuracy is greater for HD and high-grade NHL than for other forms and decreases for lesions measuring less than 2 cm, even with optimal technique. It performs poorly below the diaphragm because of normal splenic and bowel activity and a significant number of lymphomas are non-gallium-avid. For these reasons, it has no role as an isolated tool in the staging of lymphoma. Numerous studies have shown that positron emission tomography (PET) using [F-18]FDG (FDG-PET) is at least as accurate as CT in the depiction of nodal and extranodal disease13,14 and more sensitive than Ga-67. It results in clinically

•

RETICULOENDOTHELIAL DISORDERS: LYMPHOMA

significant upstaging in up to 10–20% of patients and as a result, changes in therapy, particularly in HD15,16. In NHL, it indicates tumour burden as well as the presence or absence of extranodal disease, though occasional false negative studies occur with very low grade lymphomas such as MALT types. Despite the proven advantages of FDG-PET, until recently few authorities would have recommended its use as an isolated staging tool. However, the advent of combined PET-CT will potentially revolutionize staging, since for the first time it is possible to assess both morphological and functional abnormalities accurately.

Neck Between 60 and 80% of patients with HD present with cervical lymphadenopathy. The spread of the disease is most frequently to contiguous nodal groups, with involvement of the internal jugular chain first and spread to other deep lymphatic chains in the neck, supraclavicular fossae and axillae. Patients with supraclavicular or bilateral neck adenopathy are at increased risk of infradiaphragmatic disease. Cervical adenopathy is less common in NHL, although involved nodes may be much larger. Extranodal disease with or without associated adenopathy is common,Waldeyer’s ring being the site most frequently involved. Approximately 40–60% of patients who present with head and neck involvement will have disseminated NHL. Involved nodal groups tend to be noncontiguous. Lymph nodes greater than 10 mm in diameter are generally considered enlarged. Minimally enlarged discrete lymph nodes seen in the neck in patients with lymphoma usually have a well-defined contour, but once tumour has broken beyond the confines of the node, the fat planes between the nodal mass and adjacent structures are lost. Central necrosis within a lymph node is rarely seen. Imaging with CT or MRI has a useful role in evaluating the neck in patients with lymphoma, as it may identify enlarged nodes which are impalpable. It is useful to assess treatment response, particularly in patients treated with radiotherapy, where post-treatment fibrosis renders clinical assessment difficult and it may identify recurrence in such patients.

Thorax Nodes within the thorax are involved at the time of presentation in 60–85% of patients with HD and 25–40% of patients with NHL17,18. Nodes larger than 1 cm short axis diameter at CT or MRI are considered enlarged. Any intrathoracic group of nodes may be affected, but all the mediastinal sites other than paracardiac and posterior mediastinal nodes are more frequently involved in HD than NHL. The frequency of nodal involvement in HD is as follows18: • prevascular and paratracheal—84% (Figs 72.1, 72.2) • hilar—28% (Fig. 72.2) • subcarinal—22% (Fig. 72.2). Other sites, which account for about 5% of all involved nodal groups, include the aortopulmonary window, the anterior diaphragmatic group, the internal mammary (Fig. 72.1) and the posterior mediastinal group (Fig. 72.2). In NHL involvement of the hilar and subcarinal groups is rarer, occurring in 9%

1737

1738

SECTION 9

•

MISCELLANEOUS DISORDERS

Figure 72.1 Internal mammary lymphadenopathy. Unenhanced CT showing marked enlargement of the internal mammary lymph nodes bilaterally. Note the minimal bilateral axillary and paratracheal lymph node enlargement and small right pleural effusion.

and 13% respectively, whereas superior mediastinal nodes are involved in 35%19. The great majority of cases of HD show enlargement of two or more nodal groups, whereas only one nodal group is involved in up to half of the cases of NHL. Almost all patients with nodular sclerosing HD have disease in the anterior mediastinum. Hilar nodal enlargement is rare without associated mediastinal involvement, particularly in HD. The posterior mediastinum is infrequently involved, but if disease is present in the lower part of the mediastinum, contiguous retrocrural disease is likely. Although paracardiac nodes are rarely involved at presentation in HD, they may become important as sites of recurrence, as they are not included in the classical ‘mantle’ radiation field. In most cases lymphadenopathy is bilateral but asymmetrical and nodes may be discrete or matted together. In HD and NHL, large anterior mediastinal masses usually represent thymic infiltration as well as a nodal mass (Fig. 72.3). Cystic change can be recognized at CT and MRI with large anterior mediastinal masses, especially in mediastinal large Bcell lymphoma (previously called sclerosing large B-cell lymphoma). A large anterior mediastinal mass in HD is recognized as an adverse prognostic feature. As indicated above, such a feature alters management strategy and defines the need for more aggressive initial therapy. In 10% of patients with HD, CT will demonstrate unsuspected mediastinal nodal enlargement despite a normal chest radiograph and these patients have a poorer prognosis. As a result, CT of the chest will change stage and alter management in up to 25% of patients with HD. The therapeutic impact is less in patientswith NHL.

Abdomen and pelvis

Figure 72.2 Middle mediastinal nodal disease. (A) Contrast-enhanced CT showing marked enlargement of the paratracheal and precarinal group of nodes, extending laterally into the left aortopulmonary nodes and continuing inferiorly into the subcarinal group (B) in the same patient.

The pattern of the disease below the diaphragm is markedly different in HD and NHL. At presentation the retroperitoneal nodes are involved in 25–35% of patients with HD and 45–55% of patients with NHL20. Mesenteric lymph nodes are involved in more than half the patients with NHL and less than 5% of patients with HD20,21. Other sites, as in the porta hepatis and around the splenic hilum, are also less frequently involved in HD than NHL (Fig. 72.4). In HD, nodal spread is predictably from one lymph node group to another through directly connected lymphatic pathways22. Nodes are frequently of normal size or only minimally enlarged23. Spread from the mediastinum occurs through the lymphatic vessels to the retrocrural nodes, coeliac axis and so on. Around the coeliac axis, multiple normal-sized nodes may be seen, which can be difficult to evaluate because involved, normal-sized nodes are frequent in HD23. The coeliac axis, splenic hilar and porta hepatis nodes are involved in about 30% of patients and splenic hilar nodal involvement is almost always associated with diffuse splenic infiltration (Fig. 72.4). In the porta hepatis (Fig. 72.4), the node of the foramen of Winslow (porta caval node), lying between the portal vein and the inferior vena cava, is important, as it is often overlooked and may be the only site of disease relapse. This node has a triangular or lozenge shape; its normal transverse diameter is up to 3 cm and in the anteroposterior plane is approximately 1 cm.

CHAPTER 72

•

RETICULOENDOTHELIAL DISORDERS: LYMPHOMA

Figure 72.3 Mediastinal masses in lymphoma. (A) Contrast-enhanced CT showing a large anterior mass in a young patient with Hodgkin’s disease. No other disease was demonstrated. (B) Contrast-enhanced CT in a patient with mediastinal diffuse large B-cell lymphoma (non-Hodgkin’s). The mass is involving the anterior and middle mediastinum. This subtype of non-Hodgkin’s lymphoma carries a poor prognosis and, as in this patient, often involves the abdominal viscera.

Figure 72.4 Upper abdominal lymph node enlargement. (A) Contrast-enhanced CT showing enlarged lymph nodes in the gastrohepatic ligament. Minimal lymph node enlargement (exceeding 6 mm) is also seen in the right retrocrural region. (B) Contrast-enhanced CT showing lymph node enlargement in the region of the splenic hilum around the coeliac axis and in the porta hepatis. Multiple small, nonenhancing foci are demonstrated within the spleen, typical of splenic involvement.

In NHL, nodal involvement is frequently noncontiguous and bulky and is more frequently associated with extranodal disease. Discrete mesenteric nodal enlargement or masses may be seen with or without retroperitoneal nodal enlargement. Large volume nodal disease in both mesentery and retroperitoneum may give rise to the so-called ‘hamburger’ sign, in which a loop of bowel is compressed between two large nodal masses. In NHL, regional nodal involvement is frequently seen in patients with primary extranodal lymphoma involving an abdominal viscus. Involved nodes tend to enhance uniformly

and the presence of central necrosis or multilocular enhancement should suggest an alternative diagnosis such as tuberculosis or atypical infection. In the pelvis, all nodal groups may be involved in both HD and NHL. Presentation with enlarged inguinal or femoral lymphadenopathy is seen in less than 20% of HD, but when it does occur, careful attention must be paid to the pelvic nodes on imaging, as these will be the next contiguous site of spread. In patients with massive pelvic disease, MRI is helpful for delineating the full extent of tumour and the effect on the adjacent organs.

1739

1740

SECTION 9

•

MISCELLANEOUS DISORDERS

EXTRANODAL MANIFESTATIONS OF LYMPHOMA Involvement of extranodal sites by lymphoma usually occurs in the presence of widespread advanced disease elsewhere. Such secondary involvement occurs in both HD and NHL and it is a recognized adverse prognostic feature. However, in approximately 30–40% of cases, primary involvement of an extranodal site occurs, with lymph node involvement limited to the regional group of nodes—stages I–IIE. Both primary and secondary extranodal involvement is more common in NHL than HD. As primary extranodal HD is extremely rare, rigorous exclusion of disease elsewhere is essential before this diagnosis can be made. The incidence of extranodal involvement in NHL depends on a number of factors, including the age of the patient, the presence of pre-existing immunodeficiency and the pathological subtype of lymphoma. There is an increased incidence of extranodal disease in children, especially in the gastrointestinal tract, the major abdominal viscera and extranodal locations in the head and neck,24 and in the immunocompromised host. In both these patient groups, the high incidence of extranodal involvement is a reflection of the fact that such lymphomas are usually of the more aggressive histological subtypes. As the frequency of NHL is increasing (both in the general population and in the immunocompromised), the incidence of extranodal disease is rising faster than that of nodal disease25. For example, primary lymphomas of the CNS and orbit are increasing in frequency at a rate of 10% and 6%, respectively, per annum. Of the various pathological forms of NHL, mantle cell (a diffuse low-grade B-cell lymphoma), lymphoblastic lymphomas (80% of which are T-cell), Burkitt (small cell noncleaved) and MALT lymphomas demonstrate a propensity to arise in extrandal sites. CT generally performs well in the depiction of extranodal disease, though there are certain instances where ultrasound or MRI are preferred. FDG-PET is more sensitive than CT because of its ability to identify splenic and bone marrow infiltration14, though limited availability has precluded its use in this regard.

Thorax Pulmonary parenchymal involvement

parenchyma, hence its paramediastinal or perihilar location. In this circumstance there is no effect on stage; the ‘E’ lesion. Patients with HD presenting with an intrapulmonary lesion in the absence of demonstrable mediastinal disease are unlikely to have lymphomatous disease of the lung, unless there has been previous mediastinal or hilar irradiation, when recurrence may be confined to the lungs. Conversely, in NHL, nodal disease is absent in 50% of those patients with pulmonary or pleural involvement. As nodal disease progresses or relapses, lung involvement becomes commoner in HD and NHL, such that 30–40% of patients with HD have pulmonary involvement at some stage during the course of the disease27. Radiographic appearances The radiographic appearances

are extremely variable, but the commonest pattern is that of one or more discrete nodules, with or without cavitation, which tend to be less well defined and less dense than those of primary or metastatic carcinoma, which they otherwise resemble (Figs 72.5, 72.6)28.The disease tends to spread along lymphatic channels and involves lymphoid follicles around bronchovascular divisions, resulting in peribronchial nodula-

Figure 72.5 Hodgkin’s disease with cavitating lung lesions. Chest radiograph showing bilateral hilar lymph node enlargement and multiple cavitating lung nodules (arrows).

Several categories of lymphomatous lung involvement can be identified: • that associated with existing or previously treated intrathoracic nodal disease • that associated with widespread extrathoracic disease • primary pulmonary HD • primary pulmonary NHL. Some authors also separate out AIDS-related lymphomas (ARL), which commonly affect the lungs and the post-transplant lymphoproliferative disorders (PTLDs)26. There is lung involvement at presentation in just under 4% of patients with NHL, but in approximately 12% of patients with HD. It usually occurs by direct extension of nodal disease, which is generally evident radiographically, into the adjacent

Figure 72.6 Pulmonary involvement in a patient with Hodgkin’s disease. CT performed at the time of presentation, showing widespread ill-defined intrapulmonary nodular shadowing scattered throughout both lungs. Early cavitation can be seen in a peripheral nodule in the left lung.

CHAPTER 72

tion spreading out from the hila, which can result in streaky shadowing visible on chest radiographs and at CT. A more unusual pattern is for the lymphoma to fill the pulmonary acini, producing rounded or segmental areas of consolidation with air bronchograms (Fig. 72.7). Small nodulations along the bronchial wall may enable differentiation from infective consolidation. A rare pattern of disease is widespread interstitial reticulonodular shadowing, producing a lymphangitic picture. Another rare manifestation is atelectasis, which results from endobronchial lymphoma rather than extrinsic compression by nodal disease. Several problems exist in the differential diagnosis of pulmonary involvement in lymphoma, which includes druginduced changes, the effect of radiotherapy and opportunistic infection during or following chemotherapy, particularly in patients with antecedent immunosuppression.

Primary pulmonary lymphoma Primary pulmonary lymphoma accounts for less than 1% of all lymphomas and is usually low-grade B-cell NHL, arising from MALT or bronchus-associated lymphoid tissue (BALT). BALT lymphomas are low-grade lymphomas of small lymphocytic type which tend to occur in the fifth to sixth decades, have an indolent course with 5-year survivals of over 60% and tend to remain extranodal, although lymph node involvement can occur with advanced disease29. Many patients will have a prior history of inflammatory or autoimmune disease, such as Sjögren’s syndrome, collagen vascular disease and dysgammaglobulinaemia30. The imaging findings are nonspecific with the single commonest manifestation being a solitary nodule. Multiple nodules, or one or more rounded or segmental areas of consolidation31, are also seen. These can persist unchanged for long periods. Pleural effusions are seen in up to 20% of cases30. In the remaining 15–20% of patients, primary lung lymphoma is due to high-grade NHL. The most common finding on a chest radiograph is of a solitary or multiple pulmonary nodules, which characteristically grow rapidly.

Figure 72.7 Lung involvement in Hodgkin’s disease. CT showing peribronchial streaking and consolidation radiating from both hila. This pattern reflects spread along the peribronchial lymphatics.

•

RETICULOENDOTHELIAL DISORDERS: LYMPHOMA

Primary pulmonary HD is extremely rare. The most frequently described finding is single or multiple nodules with upper zone predominance and a relatively high incidence of cavitation.

Pleural disease Pleural effusions, usually accompanied by mediastinal lymphadenopathy, occur at presentation in 10% of patients with NHL and 7% of patients with HD (Fig. 72.8)32, though they may be detected on CT in 50% of patients with mediastinal nodal disease. They are usually exudates secondary to central lymphatic or venous obstruction, rather than direct malignant involvement and therefore clear promptly with treatment of the mediastinal disease. Pulmonary involvement need not be present. Focal pleural masses do occur at presentation but are more commonly seen in recurrent disease, when they are generally accompanied by an effusion32.

Pericardium and heart Direct pericardial and cardiac involvement can occur with high-grade peripheral T-cell and large B-cell lymphomas. Such direct spread to involve the heart is rare, except in patients with AIDS related lymphoma (ARL) and post-transplant lymphoproliferative disorders (PTLD) who may present with acute onset of heart block, tamponade, or congestive cardiac failure. Pericardial effusions occur in 6% of patients with HD at the time of presentation and are associated with large masses adjacent to the heart. For staging, effusions are regarded as evidence of pericardial involvement. Small pericardial effusions are often seen at CT during treatment, the aetiology of which is unclear. They usually resolve with time, although some pericardial thickening may remain.

Thymus It is rare for HD to arise primarily in the thymus, but involvement of the thymus by HD in association with enlarged mediastinal nodes occurs in 30–50% of patients at presentation. Mediastinal large B-cell lymphoma characteristically involves the thymus, occurring typically in young women

Figure 72.8 Pleural disease in lymphoma. CT showing a typical appearance of pleural involvement in a patient with NHL. There is a lobulated mass encasing the right hemithorax involving all the pleural surfaces, encasing and narrowing the superior vena cava.

1741

1742

SECTION 9

•

MISCELLANEOUS DISORDERS

between the ages of 25 and 40 years (Fig. 72.3B). Rapidly growing bulky disease is typical and up to 40% have superior vena caval obstruction, which is rare with other lymphomas. On CT, differentiation of enlarged mediastinal lymph nodes from thymic involvement is often difficult as the thymus involved by lymphoma usually has a homogeneous soft-tissue density or a heterogeneous nodular appearance. On MRI as well, the gland is often of mixed signal intensity similar to that of involved nodes. Nevertheless, thymic involvement usually retains the shape of the normal thymus and has a smooth contour, whereas nodal masses are usually lobulated. CT or MRI may occasionally demonstrate cysts up to 3 cm in diameter, which can persist or even increase in size following regression of the rest of the involved gland with successful treatment. Calcification may be present at the outset or may develop during treatment33,34. Benign thymic rebound hyperplasia can develop after completion of chemotherapy, which can be difficult to differentiate from recurrent disease. Unfortunately, functional imaging with gallium-67 or FDG-PET may not always differentiate between the two and clinical correlation combined with follow-up studies may be necessary.

Chest wall In HD, spread into the chest wall usually occurs by direct infiltration from an anterior mediastinal mass, especially from an internal mammary chain. In HD or NHL, chest wall masses can also spread from axillary or supraclavicular nodes, or arise de novo within the chest wall. Bony destruction is rare and should suggest a diagnosis of carcinoma or infection rather than lymphoma.Thoracic wall disease is better shown by MRI than CT, particularly on T2-weighted or STIR sequences, where there is excellent contrast between the mass and normal low signal intensity muscle. This facilitates more accurate planning of radiotherapy portals35.

Breast Lymphoma of the breast is usually associated with widespread disease elsewhere. Primary NHL of the breast is rare, accounting for approximately 2% of all lymphomas and under 0.5% of all malignant breast neoplasms.The age distribution is bimodal, with the first peak occurring during pregnancy and lactation, often high-grade or Burkitt and affecting both breasts diffusely with an inflammatory picture. Mammography shows thickening of the skin and trabeculae36. There is a second peak at around 50 years when patients present with discrete masses, solitary in 66% of cases. Bilateral disease occurs in up to 13%. The masses are usually fairly well defined and there is little accompanying architectural distortion or skin thickening. Secondary involvement is characterized by multiple nodules, with associated large-volume adenopathy. Calcification has not been described in primary or secondary disease.

Hepatobiliary system and spleen Liver Liver involvement is present in up to 15% of adult patients with NHL at presentation, though this figure is higher in the paediatric population and with recurrent disease. In HD, liver involvement occurs in about 5% of patients at presentation, almost invariably in association with splenic HD.True primary hepatic lymphoma is rare but the incidence is rising, up to 25% of affected patients being hepatitis B or C positive. Pathologically, diffuse microscopic infiltration around the portal tracts is the most common form of involvement. CT and MRI are therefore insensitive in the detection of liver involvement. However, hepatomegaly strongly suggests the presence of diffuse infiltration (in contradistinction to the significance of splenomegaly). Larger focal areas of infiltration are only present in 5–10% of patients with hepatic lymphoma. Crosssectional imaging may demonstrate miliary nodules or larger solitary or multiple masses, resembling metastases (Fig. 72.9).

Figure 72.9 Lymphomatous infiltration of the liver. (A) Transverse ultrasound of the liver, showing multiple large hypoechoic lesions corresponding to the large focal masses demonstrated on contrast-enhanced CT (B).

CHAPTER 72

On ultrasound they are usually well defined and hypoechoic; on CT they are hypodense before and after intravenous injection of contrast medium; and on MRI they are of higher signal intensity than the surrounding parenchyma on T2-weighted sequences. Superparamagnetic iron oxide particles can increase the conspicuity of such focal deposits. Occasionally, especially in children, periportal infiltration is manifest as periportal low attenuation tissue at CT (Fig. 72.10). Non-Hodgkin’s lymphoma of the bile ducts and gallbladder is rare but occurs with relatively high frequency in patients with ARL.

Spleen The spleen is involved in 30–40% of patients with HD at the time of presentation, usually in the presence of nodal disease above and below the diaphragm (stage III), but in a small proportion it is the sole focus of intra-abdominal disease. In the majority of patients, the involvement is microscopic and diffuse and thus particularly difficult to identify on cross-sectional imaging, as splenomegaly does not necessarily indicate involvement; 33% of patients have splenomegaly without infiltration and, conversely, 33% of normal-sized spleens are found to contain tumour following splenectomy. Measurements of splenic volume and splenic indices are not generally utilized. Focal splenic deposits occur in only about 10–25% of cases which, when they are more than 1 cm in diameter, can be demonstrated on any form of cross-sectional imaging (Fig. 72.4B). Up to 40% of patients with NHL have splenic involvement at some stage. Imaging findings include a solitary mass, military nodules, or multiple masses, all of which tend to have a nonspecific appearance. The differential diagnosis of multiple masses includes opportunistic infection and granulomatous disease. In early studies, the sensitivity of ultrasound and CT for the detection of splenic involvement was extremely low (about 35%), although in a more recent study ultrasound was more sensitive than CT in detecting nodules down to 3 mm in size

•

RETICULOENDOTHELIAL DISORDERS: LYMPHOMA

(63 vs. 37%)37. Detection of small focal nodules has improved with the advent of multidetector CT (MDCT) and powered contrast medium injection, with optimal splenic parenchymal opacification. MRI with superparamagnetic iron oxide may improve the diagnostic accuracy but is seldom undertaken to assess splenic status. However, FDG-PET can detect splenic disease more accurately than CT or gallium scintigraphy. In the past, the poor sensitivity of imaging for the detection of splenic involvement in HD necessitated staging laparotomy with splenectomy but the development of effective combination chemotherapy with good salvage regimens has led to this practice being abandoned. Primary splenic NHL is rare, accounting for 1% of all patients with NHL. Patients present with splenomegaly, often marked and focal masses are usual. Splenic involvement is also a particular feature of certain other pathological subtypes of NHL, such as mantle cell lymphoma and splenic marginal zone lymphoma. Infarction is a well-recognized complication.

Gastrointestinal tract The gastrointestinal tract is the commonest site of primary extranodal NHL, accounting for 30–45% of all extranodal presentations. It is the initial site of lymphomatous involvement in up to 10% of all adult patients and up to 30% of children24. As elsewhere, primary HD of the gastrointestinal tract is most unusual. Secondary involvement of the gastrointestinal tract is extremely common, usually from direct extension from involved mesenteric or retroperitoneal lymph nodes and consequently multiple sites of involvement occur. Primary lymphomas arise from lymphoid tissue of the lamina propria and the submucosa of the bowel wall and occur most frequently below the age of 10 years (usually Burkitt lymphoma, BL) and in the sixth decade (MALT type and enteropathyassociated T-cell type). Primary gastrointestinal lymphoma is usually unifocal; accepted criteria for the diagnosis of primary disease include:

Figure 72.10 Periportal lymphoma. (A) Contrast-enhanced CT in a child with NHL, showing infiltration of low-density lymphomatous tissue from the porta hepatis, encasing the main portal vein, extending alongside the right portal vein. A solitary focal abnormality is seen posteriorly within the liver. (B) A follow-up after chemotherapy shows complete resolution of the disease.

1743

1744

SECTION 9

•

MISCELLANEOUS DISORDERS

• an absence of superficial or intrathoracic lymph node enlargement • no involvement of the liver or spleen • a normal white cell count • no more than local regional lymph node enlargement. In both primary and secondary cases, the stomach is most frequently involved (50%), followed by the small bowel (35%) and large bowel (15%).

organs is unusual, unlike in gastric carcinoma39, though it may occur in DLBCL in Table 72.1. In gastric MALT lymphomas, mural thickening may be minimal; CT is then of limited value and endoscopic ultrasound is more useful in staging, prognostication and assessment of response (Fig. 72.12). Low-grade MALT lymphoma is more likely to cause shallow ulceration and nodulation, whereas high-grade lymphoma can produce more massive gastric infiltration and polypoid masses.

Small bowel Stomach Primary lymphoma accounts for about 2–5% of all gastric tumours38. It originates in the submucosa, affecting the antrum more commonly than the body and the cardia. Radiologically the appearances reflect the gross pathological findings; common appearances are multiple nodules, some with central ulceration, or a large, fungating lesion with or without ulceration. About a third of patients have diffuse infiltration with marked thickening of the wall and narrowing of the lumen, sometimes with extension into the duodenum indistinguishable from linitis plastica. Only about 10% are characterized by diffuse enlargement of the gastric folds, similar to the pattern seen in hypertrophic gastritis (Fig. 72.11). As the disease originates in the submucosa, the signs described above are best demonstrated on barium studies or endoscopically, but these investigations do not reflect the true extent of gastric wall thickening and accompanying nodal involvement, which is well shown on CT. Typically, infiltration of adjacent

Figure 72.11 Gastric lymphoma. Double-contrast barium meal study (with compression) showing marked rugal hypertrophy. Note areas of ulceration.

Lymphoma accounts for up to 50% of all primary tumours of the small bowel, occurring most frequently in the terminal ileum and becoming progressively less frequent proximally, such that duodenal lymphomas are rare (Fig. 72.13). In children the disease is almost exclusively ileocaecal. Most are of Bcell lineage.The disease is multifocal in up to 50% of cases and because the disease originates in the lymphoid follicles, mural thickening with constriction of segments of bowel is typical. Patients commonly present with obstructive symptoms. Bowel wall thickening is well demonstrated on CT (Fig. 72.14).With progressive tumour spread through the submucosa and muscularis mucosa, long tube-like segments result. Aneurysmal dilatation of long segments of bowel also develop, presumably due to infiltration of the autonomic plexus. Such alternating areas of dilatation and constriction are a common manifestation of small bowel infiltration. Occasionally the lymphomatous infiltration is predominantly submucosal, which results in multiple nodules or polyps of varying size scattered throughout the small bowel, but predominantly in the terminal ileum. It is this form of lymphoma that typically results in intussusception, usually in the ileocaecal region. This is the commonest cause of intussusception in children older than 6 years. Barium studies typically

Figure 72.12 MALT lymphoma. Endoscopic ultrasound showing a narrow sheet of low echogenic tissue in the submucosa (arrowed). (Image courtesy of Dr A. McLean, Department of Diagnostic Imaging, St Bartholomew’s Hospital, London.)

CHAPTER 72

•

RETICULOENDOTHELIAL DISORDERS: LYMPHOMA

Figure 72.14 Small bowel involvement in lymphoma. CT showing marked nodular thickening of multiple loops of ileum secondary to infiltration by low-density lymphomatous material.

Figure 72.13 Non-Hodgkin’s lymphoma involving the duodenum. (A, B) Diffuse concentric gross thickening of the duodenal wall due to lymphomatous infiltration. Note multiple prominent lymph nodes in the mesentery and retroperitoneum.

most arising in the caecum and rectum. The most common pattern of the disease is a diffuse or segmental distribution of small nodules 0.2–2.0 cm in diameter, typically with intact mucosa (Fig. 72.15). A less common form of the disease is a solitary polypoid mass, often in the caecum, indistinguishable from carcinoma on imaging unless there is concomitant involvement of the terminal ileum, which is more suggestive of lymphoma. In advanced disease, there may be marked thickening of the colonic or rectal folds resulting in focal strictures, fissures or ulcerative masses with fistula formation. In the rectosigmoid, lymphomatous strictures are generally longer than carcinomatous strictures and irregular excavation of the mass strongly suggests lymphoma. Involvement of the anorectum is a feature of ARL. Patients usually present with obstruction and rectal bleeding.

Oesophagus show multiple polypoid filling defects, with or without central ulceration and irregular thickening of the valvulae. Enteropathy associated T-cell lymphoma and immunoproliferative small intestinal disease (alpha-chain disease) commonly present with clinical and imaging features of malabsorption, but acute presentations with perforation are common. Often the whole small intestine is affected, especially the duodenum and jejunum. In the small bowel (and colon), MALT lymphoma is manifest as mucosal nodularity, which can be appreciated in barium studies. Secondary invasion of the small bowel is commonly seen when large mesenteric lymph node masses cause displacement, encasement or compression of the bowel. Peritoneal disease identical to that seen with ovarian carcinoma generally occurs late in advanced disease, though it may be seen at presentation in BL.

Colon and rectum Primary colonic lymphomas are usually of Burkitt or MALT subtypes, but account for under 0.1% of all colonic neoplasms,

Involvement of the oesophagus is extremely unusual and begins as a submucosal lesion, usually in the distal third of the oesophagus, resulting in smooth luminal narrowing with intact

Figure 72.15 Involvement of large bowel in non-Hodgkin’s lymphoma. CT showing marked and extensive diffuse, uniform thickening of the wall of the transverse colon (arrow). Mesenteric nodes can also be identified. There is also very marked thickening of the wall of the ascending colon (curved arrow) and caecum.

1745

1746

SECTION 9

•

MISCELLANEOUS DISORDERS

overlying mucosa. Later, ulceration can develop. Secondary involvement by contiguous spread from adjacent nodal disease is more common but rarely results in dysphagia.

Pancreas Primary pancreatic lymphoma accounts for only 1.3% of all pancreatic malignancies and 2% of patients with NHL. Intrinsic involvement of the pancreas usually results in a solitary mass lesion, usually in the head of the pancreas, indistinguishable from primary adenocarcinoma on US, CT, or MRI40. Biliary or pancreatic ductal obstruction can occur. Calcification and necrosis are rare. Less commonly, diffuse uniform enlargement of the pancreas is seen. As elsewhere, involvement is far more common in NHL than in HD. Secondary pancreatic involvement is seen in association with disease elsewhere and usually results from direct infiltration from adjacent nodal masses, either focal or massive.

Genitourinary tract Although the genitourinary tract is not commonly involved at the time of presentation (