Operative neruosurgery

ANATOMY Surgical Anatomy and Technique MICROSURGICAL ANATOMY OF THE SUPRACEREBELLAR TRANSTENTORIAL APPROACH TO THE POST...

Author: Roham Moftakhar | Yusuf Izci

51 downloads 1611 Views 83MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

ANATOMY Surgical Anatomy and Technique

MICROSURGICAL ANATOMY OF THE SUPRACEREBELLAR TRANSTENTORIAL APPROACH TO THE POSTERIOR MEDIOBASAL TEMPORAL REGION: TECHNICAL CONSIDERATIONS WITH A CASE ILLUSTRATION

Reprint requests: Mustafa K. Bas¸kaya, M.D., Department of Neurological Surgery, University of Wisconsin at Madison, CSC K4/828, 600 Highland Avenue, Madison, WI 53792. Email: [email protected]

OBJECTIVE: Surgical access to the posterior portion of the mediobasal temporal lobe presents a formidable challenge to neurosurgeons, and much controversy still exists regarding the selection of the surgical approach to this region. The supracerebellar transtentorial (SCTT) approach to the posterior mediobasal temporal region can be used as an alternative to the subtemporal or transtemporal approaches. The aim of this study was to demonstrate the surgical anatomy of the SCTT approach and review the gyral, sulcal, and vascular anatomy of the posterior mediobasal temporal lobe. The use of this approach in the resection of a ganglioglioma located in the left posterior parahippocampal gyrus is illustrated. METHODS: The SCTT approach to the posterior parahippocampal gyrus was performed on three silicone-injected cadaveric heads. The gyral, sulcal, and arterial anatomy of the posterior mediobasal temporal lobe was studied in six formalin-fixed injected hemispheres. RESULTS: The SCTT approach provided a direct path to the posterior mediobasal temporal lobe and exposed the posterior parahippocampal gyrus as well as the adjacent gyri in all of the cadaveric specimens. Through this approach, gross total resection of the ganglioglioma was possible in our patient. CONCLUSION: The SCTT approach provided a viable surgical route to the posterior mediobasal temporal lobe in the cadaveric studies. This approach provides an advantage over the subtemporal and transtemporal routes in that there is less temporal lobe retraction.

Received, February 25, 2007.

KEY WORDS: Anatomy, Lingual gyrus, Parahippocampal gyrus, Supracerebellar transtentorial approach

Roham Moftakhar, M.D. Department of Neurological Surgery, University of Wisconsin and Veterans Administration Hospital, Madison, Wisconsin

Yusuf Izci, M.D. Department of Neurological Surgery, University of Wisconsin and Veterans Administration Hospital, Madison, Wisconsin

Mustafa K. Bas¸kaya, M.D. Department of Neurological Surgery, University of Wisconsin and Veterans Administration Hospital, Madison, Wisconsin

Accepted, September 4, 2007.

T

Neurosurgery 62[ONS Suppl 1]:ONS1–ONS8, 2008

he complexity of the gyral and sulcal anatomy of the mediobasal temporal lobe poses a surgical challenge. The mediobasal temporal lobe has been divided into three regions by de Oliveira et al. (3). The anterior segment begins from the anterior end of the rhinal sulcus and continues posteriorly to a transverse line at the level of the inferior choroidal point. The middle segment starts from the inferior choroidal point and continues to the quadrigeminal plate. The posterior segment is an area posterior to the quadrigeminal plate. The posterior portion of the mediobasal temporal region is separated by the anterior calcarine sulcus into superior and inferior parts. The posterosuperior portion includes the isthmus of the cingulate gyrus; the posteroinferior portion includes the lingual (medial temporo-occipital) gyrus (9).

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297030.82996.6E

Several approaches to the posterior mediobasal temporal lobe have been described. One of these approaches, the supracerebellar transtentorial (SCTT) approach, was first reported by Voigt and Yas¸argil (16) in 1976 for the removal of a cavernous angioma in the left parahippocampal gyrus. This approach has several advantages over the subtemporal or transtemporal routes to the posterior mediobasal temporal structures. In the subtemporal approach, retraction, especially of the dominant temporal lobe, could cause temporal lobe injury as a result of direct contusion or venous infarction induced by injury to the vein of Labbe. The transtemporal approach in the dominant temporal lobe may result in speech dysfunction or optic radiation injury.

VOLUME 62 | OPERATIVE NEUROSURGERY 1 | MARCH 2008 | ONS1

MOFTAKHAR ET AL.

FIGURE 1. A left suboccipital craniotomy was performed in the cadaver with extension above the transverse sinus (TS) and one-third past the midline. SSS, superior sagittal sinus; TH, torcular herophili; OS, occipital sinus.

FIGURE 2. The dura was opened over the suboccipital cerebellar surface in the cadaver. SSS, superior sagittal sinus; TH, torcular herophili.

A thorough understanding of the anatomy of the SCTT approach is essential in performing this exposure. Although the operative description of this approach has been well demonstrated in clinical reports (14, 16, 18), a stepwise pictorial demonstration in cadaveric dissections is lacking. Therefore, the present study was conducted to demonstrate the surgical anatomy of this approach, as well as the gyral, sulcal, and arterial anatomy of the posterior mediobasal temporal lobe. In addition, a case in which the SCTT approach was used is illustrated.

MATERIALS AND METHODS A total of three cadaveric heads with veins and arteries infused with colored silicone was examined in this study. Magnifications of ⫻3 to ⫻40 and microsurgical techniques were used to study the SCTT approach in the cadaveric heads. The approach to the posterior mediobasal temporal lobe was examined in three cadavers (six sides). In addition, the gyral, sulcal, and arterial anatomy of the mediobasal temporal region was studied in six hemispheres fixed in 4% formalin. The arteries of the hemispheres were infused with colored silicone.

RESULTS Surgical Anatomy In all cadavers, a midline incision starting 3 cm superior to the inion and extending to C1 was performed. The craniotomy was tailored according to the side of the lesion. In the case of a left-sided approach, suboccipital craniotomy with extension above the transverse sinus was performed two-thirds to the left of the midline and one-third to the right of the midline (Fig. 1). For the right-sided approach, the craniotomy was performed two-thirds to the right and one-third to the left. The extent of the inferior aspect of the craniotomy was tailored so that only the superior half of the dura over the suboccipital surface of the cerebellum was exposed. The foramen magnum was not included in

ONS2 | VOLUME 62 | OPERATIVE NEUROSURGERY 1 | MARCH 2008

FIGURE 3. Cadaveric demonstration showing the tentorium with the bridging vein (BV) cut and the tentorial sinus (TeS) apparent. TS, transverse sinus. any of the bone flaps. The lateral extent of the craniotomy was up to the sigmoid sinus without exposing it. In all cadavers, the dura was opened inferior to the transverse sinus with the transverse sinus used as the base for the dural flap (Fig. 2). The dura over the inferior twothirds of the suboccipital surface of the cerebellar hemisphere was kept intact to prevent outward herniation of the cerebellum during the procedure. The durotomy was performed over the superior one-third of the suboccipital surface of the cerebellar hemisphere (Fig. 2). Under microscopic magnification, arachnoidal dissection of the supracerebellar space was performed on the side of the approach. Laterally located bridging hemispheric veins draining into the tentorial sinus were observed in every cadaver and had to be cut to release the cerebellum from the tentorium. The medially located superior and inferior vermian veins were preserved. During the infratentorial dissection, the superior cerebellar artery and the trochlear nerve were observed over the quadrangular lobule of the cerebellum in every case. Next, the tentorium cerebelli was inspected for venous channels between the leaflets of the tentorium (Fig. 3). In all tentoria, the large tentorial sinuses were mostly located in the medial third of the tentorium. The tentorium was cut

www.neurosurgery-online.com

SURGICAL APPROACH TO THE POSTERIOR MEDIOBASAL TEMPORAL LOBE

FIGURE 4. The location of the left tentorial cut in the middle of the tentorium from lateral to medial is shown by the dashed line in the cadaver.

from lateral to medial, avoiding the venous channels if possible. The location of the tentorial cut was in the middle of the tentorium (Fig. 4). The superior petrosal sinus was not entered in any of the specimens. At this point, the posterior parahippocampal gyrus, collateral sulcus, and fusiform gyrus were visualized in all cadaveric specimens (Fig. 5A). With the midline exposure gained by the SCTT approach, the posterior temporal veins, vein of Rosenthal, and precentral cerebellar vein, all draining into the vein of Galen, were observed (Fig. 5B). Branches of the inferior temporal artery and the parietooccipital artery could be observed in all six approaches. Figure 6 illustrates the trajectory of the SCTT approach to the posterior mediobasal temporal region and final exposure after opening the tentorium.

FIGURE 6. Drawings demonstrating the trajectory of surgery (red arrow) and final exposure of the mediobasal temporal region (inset with asterisk). PG, parahippocampal gyrus.

orly up to the most anterior aspect of the lingual (medial occipitotemporal) gyrus. The posterior extent of the parahippocampal gyrus relative to the posterior end of the cingulate gyrus differed in three of the hemispheres. In three of the six hemispheres, the posterior aspect of the parahippocampal gyrus ended at the posterior end of the cingulate gyrus and the anterior portion of the lingual gyrus (Fig. 7A). In the other three Gyral, Sulcal, and Venous Anatomy of the Posterior hemispheres, the parahippocampal gyrus extended more posMediobasal Temporal Lobe teriorly, passing the posterior end of the cingulate gyrus, and Gyri and Sulci then met with the anterior segment of the lingual gyrus (Fig. 7B). In these cases, the posterior end of the cingulate gyrus In all six cerebral hemispheres, the parahippocampal gyrus ended in the parahippocampal gyrus only and not the lingual formed the medial part of the posterior basal surface of the gyrus. In all hemispheres, three gyri existed on the basal surtemporal lobe. The parahippocampal gyrus continued posteriface of the temporal lobe. The most medial gyrus was the A B parahippocampal gyrus, lateral to the parahippocampal gyrus was the fusiform (lateral occipitotemporal) gyrus, and lateral to the fusiform gyrus was the inferior temporal gyrus. In all hemispheres, the parahippocampal gyrus was separated from the fusiform gyrus by the rhinal sulcus anteriorly and by the collateral sulcus posteriorly. The fusiform gyrus was separated FIGURE 5. A, through the transtentorial approach on the left side, the parahippocampal gyrus (PG), fusiform from the inferior temporal gyrus (FG), and the collateral sulcus (CS) are demonstrated in the cadaver. B, the midline exposure gained by the gyrus by the occipitotemporal left supracerebellar transtentorial approach provides access to midline venous structures. PCV, precentral cerebellar sulcus. Posteriorly, the collatvein; ICV, internal cerebral vein; BVR, basal vein of Rosenthal; MTV, medial temporal vein; TS, transverse sinus. eral sulcus separated the lin-

NEUROSURGERY


MOFTAKHAR ET AL.

A

Illustrative Case

B

History

A 20-year-old woman with a history of tonic clonic seizure since age 7 underwent a left anterior temporal lobectomy for medically resistant seizures at age 17 at another institution. The patient’s seizure frequency did not decrease off medications; however, the patient was seizurefree on two antiepileptics. On postoperative magnetic resonance FIGURE 7. A, the basal surface of the left temporal lobe demonstrating the parahippocampal gyrus (PG), lingual imaging (MRI) scans, a contrastgyrus (LG), and the cingulate gyrus meeting at one point called the isthmus (IS), uncus (U), fusiform gyrus (FG), enhancing lesion in the left posteand collateral sulcus (CS). B, the basal surface of the left temporal lobe demonstrating the posterior end of cingulate rior parahippocampal gyrus was gyrus ending at the PG. This junction would be the IS in this pattern of gyral anatomy. The posterior PG ends at evident, which had not been nothe anterior LG, FG, CS, and splenium of the corpus callosum (SP). ticed previously. The lesion grew on subsequent images. The pagual gyrus from the fusiform gyrus. The collateral sulcus was tient presented to our clinic for evaluation of the lesion. The patient’s neurological examination was unremarkable. present in all hemispheres. In three out of six hemispheres, a

midline sulcus divided the lingual gyrus into two gyri.

Arterial Anatomy In all six hemispheres, the arterial supply to the posterior mediobasal temporal lobe was from the posterior cerebral artery. In four hemispheres, the posterior mediobasal temporal lobe was supplied by one common temporal artery originating from P2 in the ambient cistern. In two of these hemispheres, the common temporal artery divided into two branches; in one hemisphere, it divided into three branches, and in one hemisphere, it did not give off any branches and ended in the calcarine artery. In one hemisphere, an anterior and posterior temporal artery supplied posterior the mediobasal temporal lobe (Fig. 8). In this hemisphere, a calcarine artery originating from P3 was observed. In the last hemisphere, an anterior temporal artery originated from P2 and the posterior temporal artery gave off the calcarine artery.

Preoperative Imaging MRI scans with and without contrast with magnetic resonance venography was performed. In the left posterior parahippocampal gyrus, a 1.7 ⫻ 1.5 cm lesion with lowintensity signal on T1-weighted images and high signal on T2-weighted images was shown. The lesion was enhancing (Fig. 9). The magnetic resonance venographs demonstrated a

Venous Anatomy In all of the supracerebellar approaches, laterally located bridging veins from the tentorial surface of the cerebellum to the tentorial sinuses were observed. In the midline superior vermian, veins that are divided into anterior and posterior groups were present. The anterior group drained into the vein of Galen, and the posterior group drained into the torcula. The major tributaries of the anterior superior vermian vein—the vein of cerebellomesoencephalic fissure, tectal veins, and hemispheric branches from the medial part of the temporal lobe— were observed in all approaches. The basal vein of Rosenthal in the ambient cistern draining into the vein of Galen was present in all specimens (Fig. 5B). The veins of the posterior mediobasal temporal lobe consisted of medial temporal veins that drained into the basal vein of Rosenthal. In all approaches to this region, the medial temporal veins were consistently observed to drain into the basal vein of Rosenthal in the ambient cistern.


FIGURE 8. Arterial supply of the posterior mediobasal temporal lobe is provided by the branches of the quadrigeminal segment (P3) of the posterior cerebral artery. ATA, anterior temporal artery; PTA, posterior temporal artery; CA, calcarine artery; CG, cingulate gyrus; SP, splenium of corpus callosum; LG, lingual gyrus; FG, fusiform gyrus; PG, parahippocampal gyrus.



A

B

C FIGURE 9. Axial (A), coronal (B), and sagittal (C) T1weighted magnetic resonance imaging (MRI) scans of the patient showing the contrastenhancing mass located in the left parahippocampal gyrus.

dominant right transverse sinus. In addition, the previous left anterior temporal lobectomy was demonstrated.

Operation and Postoperative Course (see video at web site) The patient was placed in a sitting position. A midline incision was made starting superior to the inion and extending to C1. A suboccipital craniotomy with extension above the transverse sinus was performed. Two-thirds of the craniotomy was on the left and one-third on the right. The lateral extent of the craniotomy was before the sigmoid sinus. The inferior aspect of the craniotomy was performed to expose the superior half of the suboccipital dura covering the cerebellum. The dura was opened inferior to the transverse sinus with the left and right transverse sinuses as the base. The dura was opened only over the superior one-third of the suboccipital surface of the cerebellum to prevent the outward herniation of the cerebellum. Two maneuvers under the microscope were used to create space between the tentorium and the cerebellar surface. First, the arachnoid between the cerebellum and the tentorial surface was dissected. Second, the laterally located bridging veins from the cerebellar hemisphere to the tentorial sinuses were coagulated and cut. Subsequently, the tentorium was coagulated in the middle and cut from lateral to medial. Bleeding from the tentorial sinuses was controlled with bipolar coagulation. At this time, an intra-axial lesion in the region of the posterior parahippocampal gyrus was observed. Branches of the posterior cerebral artery adjacent and crossing the lesion were skeletonized and mobilized to accommodate

NEUROSURGERY

the gross total removal of the tumor. Pathological examination of the lesion determined the lesion to be a ganglioglioma. Postoperative MRI scans with and without gadolinium demonstrated gross total resection of the tumor, and this remained the case at the 3month follow-up examination (Fig. 10). At the time of the 3-month follow-up examination, the patient was seizure-free and was being tapered off the antiepileptic medications.

FIGURE 10. Three-month postoperative coronal T1-weighted MRI scan with contrast confirming gross total resection of the tumor without recurrence.

DISCUSSION The posterior mediobasal temporal region is of particular surgical significance, not only with regard to epilepsy but also tumor and vascular lesions. The exact functional aspects and neural connections of the posterior mediobasal temporal region remain unclear. The dominant posterior mediobasal temporal region might play a role in language. Lüders et al. (8) found that electrical stimulation of the fusiform gyrus on the dominant side produced a global receptive and expressive aphasia with speech arrest at high stimulus intensities. Therefore, minimizing damage to the posterior mediobasal temporal region, especially on the dominant side, is crucial. Several approaches to the posterior mediobasal temporal structures have been described. The lateral transtemporal approach is used mainly for epilepsy surgery to remove the amygdala and hippocampus (2, 4, 15) or to resect an arteriovenous malformation of the medial temporal lobe (5). One of the drawbacks of this approach is damage to the optic radiation. The subtemporal approach is another alternative to the mediobasal temporal region (6, 10). The disadvantage of the subtemporal approach is the extensive retraction on the temporal lobe, which could lead to vein of Labbe injury and compromise of language areas in the dominant hemisphere. Posterior approaches to the mediobasal temporal surface are the occipital interhemispheric (13) and the SCTT approaches (1, 14, 16, 18). The drawback of the occipital interhemispheric approach is that occipital lobe retraction might lead to postoperative visual deficits, even in the hands of highly skilled surgeons. In a series of seven patients treated for different tumors of the posteromedial temporal lobe by the occipital interhemispheric approach, three patients experienced transient visual disturbances (13). The SCTT approach was devised to avoid risks that might be posed to the temporal or occipital lobe. In Yonekawa et al.’s (18) series of 16 patients with various lesions of the posterior mediobasal temporal lobe approached through the SCTT route, no postoperative complications were reported. We prefer this approach to the posterior mediobasal temporal lobe in general


MOFTAKHAR ET AL.

and especially in our patient who had previously undergone a left-sided anterior temporal lobectomy. Retracting the temporal lobe in a subtemporal approach could present more risk resulting from adhesions. Several important key points exist in performing the SCCT approach successfully as we have shown in our anatomic study and clinical case. We prefer to perform this approach in the sitting position as opposed to prone. With the sitting position, the amount of cerebellar retraction is minimized with the help of gravity; however, the risk of air embolism exists and the arm position is uncomfortable for the surgeon. Kobayashi et al. (7) described the prone Concorde position with the head away from the surgeon. Although this position minimizes the risk of an air embolism and is more comfortable for the surgeon, it presents some disadvantages for the assistant. Also, with the prone Concorde position, more cerebellar retraction is needed compared with the sitting position. The median incision is preferred to the paramedian incision that goes through the suboccipital muscles. In our experience with the median incision, most patients experience less postoperative pain. In our clinical case, the craniotomy was extended superior to the transverse sinus and torcula. This maneuver allows greater exposure of the supracerebellar space and helps minimize cerebellar retraction. In addition, the left transverse sinus was nondominant and could have been divided if additional exposure was needed. In our case, the sinus separated easily from the bone, most likely as a result of the patient’s young age. In elderly patients, in whom the sinus might not separate from the bone as easily, extension of the craniotomy above the transverse sinus might pose risk of injury to the sinus. In that case, the superior aspect of the craniotomy would be just up the inferior margin of the transverse sinus. The dural opening is extended inferiorly just to expose the upper one-third of the suboccipital surface of the cerebellum. This prevents the cerebellum from herniating out during surgery. In our cadaveric studies and clinical case, large bridging veins from the cerebellar hemisphere to the tentorial sinus were cut to make cerebellar retraction less traumatic. In a series of 16 patients, Yonekawa et al. (18) did not report any complications after sacrificing these veins. However, the vermian veins should not be sacrificed to prevent risk of venous infarct and hemorrhage. The tentorium is cut from lateral to medial because starting the cut on the medial side is difficult with the steep angle of the tentorium. During the tentorial cut, cases of arrhythmia and bradycardia attributable to trigeminal nerve irritation have been reported (18). The anesthesiologist should be forewarned of this possibility. Once the tentorium is cut with slight downward retraction, the posterior parahippocampal gyrus, fusiform gyrus, and anterior lingual gyrus could be observed. Care must be taken to preserve the branches of the temporal and parieto-occipital arteries. In the six cerebral hemispheres examined, the architecture of the gyri of the posterior mediobasal temporal lobe varied to some extent. In three hemispheres, the posterior extent of the parahippocampal gyrus ended at the posterior segment of the cingulate gyrus and the anterior aspect of the lingual gyrus.


Yas¸argil (17) defines the segment of the cingulate gyrus where it meets the parahippocampal and lingual gyrus as the isthmus of the cingulate gyrus. In three of the hemispheres that were examined in this study, the isthmus fit the definition given by Yas¸argil (Fig. 7A). However, in the other three hemispheres, the cingulate gyrus ended only in the parahippocampal gyrus and not the lingual gyrus. The gyral variability demonstrates the complicated anatomy of the posterior mediobasal temporal lobe, which requires further study. The vascular anatomy of the posterior mediobasal temporal region is important especially for surgery of the tumors and vascular malformations in this area. Rhoton (12) divides the inferior temporal arteries into five groups based on the branches present and area they supply. The most frequent pattern, referred to as Group 5, is when there are hippocampal, anterior, and posterior temporal branches present but no middle temporal artery. In our examination of the six hemispheres, the most common pattern was Group 5. The medial temporal veins drained the posterior mediobasal temporal lobe. The venous system of the posterior basal temporal lobe consists of lateral and medial groups (11). The lateral group drains into the sinuses in the anterolateral part of the tentorium and the medial group drains into the basal vein. The lateral group of veins draining the posterior basolateral temporal lobe include posterior temporobasal veins, and the medial temporal veins drain the posterior mediobasal temporal lobe. There are some disadvantages to the SCTT approach, making this approach ideal for only some, but not all, lesions of the mediobasal temporal lobe. With this approach, the anterior parts of the mediobasal temporal lobe cannot be accessed. This approach is, however, ideal for the posterior mediobasal temporal lobe. In addition, the difficulty in gaining proximal arterial control, early exposure of the venous drainage, and greater working distance poses a disadvantage to the surgeon. Although the use of this approach in posterior fossa revascularization has been reported (18), the long and narrow working distance possess technical challenges to even the most highly skilled surgeons.

CONCLUSIONS In this study, the posterior mediobasal temporal lobe was accessed easily in all six cadavers using the SCTT approach. This approach is ideal for exposure of this region, with the main advantage being that the temporal lobe does not have to be retracted or traversed. The senior author (MKB) has found the following points useful in performing this approach: 1) the sitting position to decrease retraction of the cerebellum; 2) midline skin incision and subperiosteal muscle dissection as opposed to a paramedian skin incision and traversing the muscle to avoid postoperative muscular pain and to achieve good cervical muscle healing; 3) craniotomy crossing the transverse sinus and torcula for wider exposure in young patients when safe and possible; 4) opening the suboccipital dura in a way not to expose the majority of the cerebellar hemisphere, which avoids herniation of the cerebellum during retraction; and 5) a



thorough knowledge of the microsurgical anatomy of the gyri, sulci, and vascular structures in the mediobasal temporal lobe, which can be supplemented by studying microneurosurgical anatomy in the laboratory.

REFERENCES 1. Campero A, Tróccoli G, Martins C, Fernandez-Miranda JC, Yasuda A, Rhoton AL: Microsurgical approaches to the medial temporal region: An anatomical study. Neurosurgery 59 [Suppl]:ONS279–ONS308, 2006. 2. Clusmann H, Kral T, Gleissner U, Sassen R, Urbach H, Blümcke I, Bogucki J, Schramm J: Analysis of different types of resection for pediatric patients with temporal lobe epilepsy. Neurosurgery 54:847–860, 2004. 3. de Oliveira E, Tedeschi H, Siqueira MG, Ono M, Rhoton AL, Peace D: Anatomic principles of cerebrovascular surgery for arteriovenous malformations. Clin Neurosurg 41:364–380, 1994. 4. Gonçalves-Ferreira A, Miguéns J, Farias JP, Melancia JL, Andrade M: Selective amygdalohippocampectomy: Which route is the best? An experimental study in 80 human cerebral hemispheres. Stereotact Funct Neurosurg 63:182–191, 1994. 5. Heros RC: Arteriovenous malformations of the medial temporal lobe. Surgical approach and neuroradiological characterization. J Neurosurg 56:44–52, 1982. 6. Hori T, Tabuchi S, Kurosaki M, Kondo S, Takenobu A, Watanabe T: Subtemporal amygdalohippocampectomy for treating medically intractable temporal lobe epilepsy. Neurosurgery 33:50–57, 1993. 7. Kobayashi S, Sugita K, Tanaka Y, Kyoshima K: Infratentorial approach to the pineal region in the prone position: Concorde position. Technical note. J Neurosurg 58:141–143, 1983. 8. Lüders H, Lesser RP, Hahn J, Dinner DS, Morris HH, Wyllie E, Godoy J: Basal temporal language area. Brain 114:743–754, 1991. 9. Ono M, Kubik S, Abernathey CD: Atlas of Cerebral Sulci. New York, Thieme, 1990, p 11. 10. Park TS, Bourgeois BF, Silbergeld DL, Dodson WE: Subtemporal transparahippocampal amygdalohippocampectomy for surgical treatment of mesial temporal lobe epilepsy. Technical note. J Neurosurg 85:1172–1176, 1996. 11. Rhoton AL: The cerebral veins. Neurosurgery 51 [Suppl]:S159–S205, 2002. 12. Rhoton AL: The supratentorial arteries. Neurosurgery 51 [Suppl]:S53–S120, 2002. 13. Smith KA, Spetzler RF: Supratentorial–infraoccipital approach for posteromedial temporal lobe lesions. J Neurosurg 82:940–944, 1995. 14. Uchiyama N, Hasegawa M, Kita D, Yamashita J: Paramedian supracerebellar transtentorial approach for a medial tentorial meningioma with supratentorial extension: Technical case report. Neurosurgery 49:1470–1474, 2001. 15. Vajkoczy P, Krakow K, Stodieck S, Pohlmann-Eden B, Schmiedek P: Modified approach for the selective treatment of temporal lobe epilepsy: Transsylvian–transcisternal mesial en bloc resection. J Neurosurg 88:855–862, 1998. 16. Voigt K, Yas¸argil MG: Cerebral cavernous haemangiomas or cavernomas. Incidence, pathology, localization, diagnosis, clinical features and treatment. Review of the literature and report of an unusual case. Neurochirurgia (Stuttg) 19:59–68, 1976. 17. Yas¸argil MG: Topographic anatomy for microsurgical approaches to intrinsic brain tumors, in: Microneurosurgery. New York, Thieme Medical Publisher, Inc., 1994, pp 2–114. 18. Yonekawa Y, Imhof HG, Taub E, Curcic M, Kaku Y, Roth P, Wieser HG, Groscurth P: Supracerebellar transtentorial approach to posterior temporomedial structures. J Neurosurg 94:339–345, 2001.

COMMENTS

T

he work describes an excellent approach to lesions in the posteromesial part of the temporal lobe. The approach avoids the hazardous retraction, risk to the vein of Labbé and other veins on the lower temporal lobe, and damage to the optic radiations and other eloquent areas at risk in the more anterior, lateral, and subtemporal approaches to the area. The authors have provided a good review of

NEUROSURGERY

the operative technique combined with an excellent description of the location and type of lesions for which the approach is suitable. Albert L. Rhoton, Jr. Gainesville, Florida

T

he first I heard about the supracerebellar transtentorial approach to the posterior mediobasal temporal region, namely to the posterior and basal part of the parahippocampal gyrus, was from Professor Yas¸argil during a hands-on course in St. Louis in 1995 (5). I was unaware of the initial reports of this approach (4) and became very impressed with its possibilities. I remember that I immediately associated this ingenious approach with the very creative transsphenoidal approaches to the pituitary developed by the pioneers Schloffer, Halsted, Kanavel, Mixter, Cushing, Hirsh, and Kabler, among others (1). More than anatomical knowledge, the development of such approaches requires an architectural envisioning of the intracranial contents. The infratentorial supracerebellar approach to the pineal region was initially proposed by Krause during the 1920s and brought into microneurosurgery by Stein in the 1980s (2, 3). As Professor Stein emphasized during his lectures, the “cathedral effect” given by the tentorial anatomy together with the descent of the cerebellum, particularly in the sitting position, generates a superb microneurosurgical corridor. This natural atrium that is now also being used to reach the part of the parahippocampal gyrus that lies over the tentorium can be further extended through skilled hands, and, in our opinion, will definitely be further explored with the aid of future surgical tools. The avoidance of damage to the optic radiation and language areas of the dominant hemisphere through transtemporal routes, and the frequent limitations imposed by the vein of Labbé for the subtemporal approaches, justify the development of more experience with this approach to reach lesions of the basal aspects of the inner limbic and paralimbic rings, of the upper brainstem, and of its related cisterns. The augmentation of the natural spaces, in different dimensions, is one of the main aims of surgery. Regarding in particular the approach to the posterior mediobasal temporal region, Moftakhar et al. are right that although an operative description was demonstrated in clinical reports, a stepwise pictorial demonstration in cadaveric dissections, with clear directions for its accomplishment, was still lacking, and they certainly achieved this goal. Guilherme C. Ribas Saõ Paulo, Brazil

1. Landolt AM: History of pituitary surgery, in Greenblatt SH (ed): A History of Neurosurgery. Park Ridge: American Association of Neurological Surgeons, 1997, pp 373–400. 2. Stein BM: The infratentorial supracerebellar approach to pineal lesions. J Neurosurg 35:197–202, 1971. 3. Stein BM: Infratentorial supracerebellar approach, in Apuzzo MLJ (ed): Surgery of the Third Ventricle. Baltimore, Williams & Wilkins, 1987, pp 570–590. 4. Voigt K, Yas¸argil MG: Cerebral cavernous haemangiomas or cavernomas: incidence, pathology, localization, diagnosis, clinical features and treatment. Neurochirurgia (Stuttgart) 19:59–68, 1976. 5. Yas¸argil MG, Olivier A, Spencer D: Epilepsy and Intrinsic Brain Tumors Course. Practical Anatomy and Surgical Education, Saint Louis University School of Medicine, St. Louis, April 28–29, 1995.

W

e commend the authors for this well-written stepwise approach describing the supracerebellar transtentorial trajectory to the pos-


MOFTAKHAR ET AL.

terior mediobasal temporal region. This approach, as originally described by the senior commentator, allows for an alternative trajectory to the posterior mesial temporal lobe region; however, it has a number of limitations. The main issue is the “narrow” surgical trajectory, which can be limiting when one is dealing with vascular lesions. Occasionally, the tentorium may contain large venous channels or “lakes” and therefore, cutting the tentorium through a confined window from below can be a challenge. Finally, the lack of more circumferential vascular control surrounding the lesion in the posterior mesial temporal lobe may pose an additional important limitation. In summary, this is an elegant anatomic trajectory to the posterior mesial temporal region; however, owing to the above-mentioned limitations, in our opinion, it may be considered only in very select lesions within this region. Saleem I. Abdulrauf St. Louis, Missouri

erebellar transtentorial approach, as described here, is very confined and is most suitable for lesions that extend to the surface of the posterior and mediotemporal lobe. The complexity of the gyral and sulcal anatomy of this region also poses a particular challenge in lesions that are subcortical, and the use of frameless stereotaxy is essential to ensure the accuracy of the approach. I would strongly concur with the senior author that the sitting position is preferable to reduce retraction of the cerebellum and that the craniotomy must extend above the transverse sinus and torcular, as upward retraction of the sinus improves the angle of exposure. This is a difficult approach, but probably the most preferable method of accessing lesions in this problematic region. The authors are to be commended for providing this detailed and practical guide to the anatomy that will be invaluable for surgeons who use this approach only infrequently. Andrew H. Kaye Melbourne, Australia

M. Gazi Yas¸argil Little Rock, Arkansas

S

urgical access to lesions within the posterior mediobasal temporal lobe are a major surgical challenge and the authors have described the microsurgical anatomy of the superior cerebellar transtentorial approach to this region. This approach does potentially have advantages over the subtemporal or transtemporal roots to this region, and in particular the avoidance of retraction of the temporal lobe and possible risk of infarction that can be induced by injury to the inferior anastomotic vein of Labbé. The transtemporal approach carries the risk of dysplasia, epilepsy, and damage to the optic radiation. The suprac-

T

he authors provide a comprehensive description of the supracerebellar transtentorial approach to the mediobasal temporal lobe. I generally find it difficult to access supratentorial lesions through the supracerebellar approach, especially if the sitting position is utilized. For these lesions I would favor an occipital interhemispheric approach where the tentorium does not need to be divided. Nevertheless it is a reasonable option to have for lesions in this location. Jeffrey N. Bruce New York, New York

Anatomy Lecture by Dr. Theodorus Hoogeveen, (1773), Nicolaas Rijnenburg. From: WolfHeidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



ANATOMY Surgical Anatomy and Technique

MICROSURGICAL ANATOMY OF THE SAFE ENTRY ZONES ON THE ANTEROLATERAL BRAINSTEM RELATED TO SURGICAL APPROACHES TO CAVERNOUS MALFORMATIONS Rodolfo J. Recalde, M.D. Universidad de Buenos Aires, Hospital Nacional Prof. A. Posadas, Buenos Aires, Argentina

Eberval G. Figueiredo, M.D. Department of Neurological Surgery, University of São Paulo School of Medicine, São Paulo, Brazil

Evandro de Oliveira, M.D., Ph.D. Department of Neurological Surgery, State University of Campinas, Instituto de Ciências Neurológicas, São Paulo, Brazil Reprint requests: Evandro de Oliveira, M.D., Ph.D., Instituto de Ciências Neurológicas, Praça Amadeu Amarau, 27, 5 andar, CEP 01327-010, São Paulo–SP, Brazil. Email: [email protected] Received, February 22, 2007. Accepted, July 10, 2007.

OBJECTIVE: To study the microanatomy of the brainstem related to the different safe entry zones used to approach intrinsic brainstem lesions. METHODS: Ten formalin-fixed and frozen brainstem specimens (20 sides) were analyzed. The white fiber dissection technique was used to study the intrinsic microsurgical anatomy as related to safe entry zones on the brainstem surface. Three anatomic landmarks on the anterolateral brainstem surface were selected: lateral mesencephalic sulcus, peritrigeminal area, and olivary body. Ten other specimens were used to study the axial sections of the inferior olivary nucleus. The clinical application of these anatomic nuances is presented. RESULTS: The lateral mesencephalic sulcus has a length of 7.4 to 13.3 mm (mean, 9.6 mm) and can be dissected safely in depths up to 4.9 to 11.7 mm (mean, 8.02 mm). In the peritrigeminal area, the distance of the fifth cranial nerve to the pyramidal tract is 3.1 to 5.7 mm (mean, 4.64 mm). The dissection may be performed 9.5 to 13.1 mm (mean, 11.2 mm) deeper, to the nucleus of the fifth cranial nerve. The inferior olivary nucleus provides safe access to lesions located up to 4.7 to 6.9 mm (mean, 5.52 mm) in the anterolateral aspect of the medulla. Clinical results confirm that these entry zones constitute surgical routes through which the brainstem may be safely approached. CONCLUSION: The white fiber dissection technique is a valuable tool for understanding the three-dimensional disposition of the anatomic structures. The lateral mesencephalic sulcus, the peritrigeminal area, and the inferior olivary nucleus provide surgical spaces and delineate the relatively safe alleys where the brainstem can be approached without injuring important neural structures. KEY WORDS: Brainstem, Brainstem surgery, Cavernous malformation, Safe entry zones, Surgical anatomy, Surgical approaches, White fiber dissection Neurosurgery 62:ONS9–ONS17, 2008

C

avernous malformations of the central nervous system account for 8 to 15% of all vascular malformations (12, 13). Their occurrence within the brainstem is estimated to be between 9 and 35%, and their surgical treatment poses several risks. The brainstem presents a dense concentration of nuclei and fibers, and their unintentional manipulation may produce neurological morbidity. The presence of functionally intact parenchyma that are in close proximity to the lesions is responsible for such significant morbidity. Criteria for surgical indications include a symptomatic lesion that reaches a pial or ependymal surface and that may be approached without the need to traverse eloquent brainstem tissue (12, 13). However, intrinsic symptomatic lesions may be reached through small neurotomies in “safe entry zones” on

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297062.52433.3F


the brainstem surface (9, 13, 14, 19, 20). These zones represent small regions in which the brainstem may be incised with relative impunity and correspond to areas where critical neural structures are sparse and no perforating arteries are encountered. The elements of the internal anatomy are related to topographic structures on the brainstem surface. Thus, a profound knowledge of both the external and internal anatomy of the cranial motor nuclei and fibers is necessary to optimize the surgical approaches and preserve functionally important brainstem structures. Previous reports have studied safe entry zones in the posterior aspect of the brainstem (1, 2, 9, 17). However, no anatomic assessments have been performed to study safe entry zones in its anterolateral aspect. Many authors have rou-


RECALDE ET AL.

tinely used anatomic landmarks such as the lateral mesencephalicum sulcus (LMS), peritrigeminal area, and olivary body as safe entry zones on the anterolateral surface of the brainstem to approach brainstem cavernous malformations (BCMs). In this study, we performed an evaluation of the intrinsic anatomy of the brainstem using the white fiber dissection technique as related to these safe entry zones. Using this methodology, the successive layers of the intrinsic anatomy of the brainstem are peeled away, and the anatomic relationships are consecutively revealed.

MATERIALS AND METHODS Ten fresh formalin-fixed and frozen brainstems were analyzed (20 halves). Anatomic characteristics of the extrinsic elements as well as white fiber dissection using Ludwig and Klinger’s technique (11) were used to study the intrinsic microsurgical anatomy of the brainstem. Magnification from 3 to 40 was used. Once the pia-arachnoid was removed, we proceded with dissection, using microsurgical tools and wooden tongue depressors as spatulas. The “anterolateral” mesencephalon was defined as the portion of the midbrain located ventral to the LMS, whereas the two anterior quadrants and the part the medulla ventral to the retro-olivary sulcus defined the anterolateral surface of the pons and medulla, respectively. Anatomic landmarks easily recognized on the brainstem surface were selected in the mesencephalon (LMS), pons (peritrigeminal area), and medulla (olivary body). The consecutive layers of the intrinsic anatomy of the brainstem were peeled away to successively expose and allow description of the anatomic relationships. Additionally, surgically relevant measurements were obtained on each level studied.

Mesencephalon FIGURE 1. The lateral mesencephalic sulcus (LMS) and its relationship with adjacent neural structures are shown. The LMS is located in the posterior aspect of the mesencephalon between the cerebral peduncles and the collicular area. It extends longitudinally from the medial geniculate body superiorly and up to the pontomesencephalic sulcus inferiorly. CN, cranial nerve.

The length of the LMS (Fig. 1) and the distance between the pial surface of the LMS to the point where Cranial Nerve (CN) III penetrates the substantia nigra were measured (Figs. 2, 3, and 4).

Pons

After resection of the transverse fibers, the distance between the fifth cranial nerve and the pyramidal fibers as well as the distance between the apparent origin of CN V up to its pontine nucleus were measured (Figs. 3, 4, and 5). The width of the descending fibers was evaluated at three different points: at the level of the pontomesencephalic sulcus, at the level of CN V, and at the level of the pontomedullary sulcus (Figs. 3 and 4).

Medulla In the medulla, the distance of the inferior olivary nucleus from the pia mater to the point where the white fibers enter and exit the nucleus was evaluated (Fig. 6). In addition, the craniocaudal, transverse, and anterodorsal diameters of the olivary bodies were assessed.


FIGURE 2. Cross-section of the midbrain showing the cerebral peduncles (CP) separated from the tegmentum by the substantia nigra (SN). The medial lemniscus (ML) is located posteriorly to the SN. The intrinsic trajectory of the third cranial nerve is dissected bilaterally. There is an alley between the SN and the ML that can be safely approached through the LMS before the optic nerve fibers may be damaged. The black line represents the distance from the LMS to the intrinsic CN III fibers.

FIGURE 3. The descending fibers have been cut at the level of CN V and resected rostrally. The SN as well as the intraneural projection of CN III may be seen. The superior and inferior cerebellar peduncles have been cut. The CN V trajectory up to its nucleus may be appreciated. The white line represents the distance from the LMS to the fibers of CN III.

RESULTS Mesencephalon After performing a myelotomy in this sulcus, we proceeded deeper in the disection, in the direction of the emergence of CN III. The dissection passes through an alley limited laterally by the substantia nigra and medially by the medial lemniscus (Figs. 2 and 7). The red nucleus and the decussation of the superior cerebellar peduncle are reached as the white fiber dissection goes FIGURE 4. Ventral view showdeeper. The fibers of CN III ing the specimen depicted in cross this alley in an almost Figure 3. The crus cerebri (CC) perpendicular fashion, from have been dissected and elevated; the red nucleus to the subthe relationship of CN III with the stantia nigra, and this point SN may be observed. The black corresponds to the anterior line corresponds to the distance limit of a safe surgical from the LMS to the fibers of CN III. approach (Figs. 2, 3, and 4). Going further might damage the CN III fibers. The depth of this corridor is shown in Table 1. The measurements on the mesencephalon were obtained at the level of the LMS. The LMS extends 7.4 to 13.3 mm (9.6 1.41 mm) (Figs. 2, 3, and 4; Table 1) from the medial geniculate body located superiorly to the interpeduncular sulcus. A sec-


MICROSURGICAL ANATOMY OF THE SAFE ENTRY ZONES

ond measurement in the mesencephalon from the LMS to the rootlets of CN III might be found in the depth of the mesencephalon. This distance varied between 4.9 and 11.7 mm (8.2 1.76 mm) (Figs. 2 and 3). According to these data, the LMS offers a mean of 9.6 mm that may be opened, and the surgeon may dissect 8.2 mm deeper without damaging CN III (Fig. 4). However, the lower limit of dissection (4.9 mm) may also be considered as the safest distance when approaching this region (Table 2).

FIGURE 5. Cross-section showing the pons at the level of CN V. The peritrigeminal area (PtrA) is depicted medially to the fibers of CN V, laterally to the pyramidal tract (PT), and ventrally to the motor nucleus (MN CN V) and sensory nucleus (SN CN V) of CN V.

Pons As one proceeds with the myelotomy in the peritrigeminal area, CN V may be seen emerging from the pons as a compact

group of fibers, and continuing its trajectory toward its nucleus (Fig. 5). As the dissection of the transverse fibers of the middle cerebellar peduncle progresses, the fibers of CN V are revealed and might be followed to the nucleus (Figs. 3 and 4). FIGURE 7. Dissection at the level Further dissection consecuof the LMS. The ML fibers may be tively revealed the nucleus seen posteriorly and the CC venof CNs VI, VII, and VIII trally. located posteriorly and medially to the nucleus of CN V. The descendent trajectory of the fibers of CNs VI, VII, and VIII might be disclosed to reach the pontomedullary sulcus. These fibers are located posterior to CN V’s nucleus. The pyramidal tract was also dissected, and the measurements were obtained (Figs. 8 and 9). The distance between the pyramidal fibers and CN V ranged between 3.8 and 5.6 mm (4.64 0.68 mm). Therefore, one can access lesions located medial to CN V by up to 4.64 mm without injuring the pyramidal fibers. The dissection may be carried out 9.5 to 13.1 mm (11.2 1.29 mm) deeper to the nucleus of CN V (Figs. 8 and 9) (Table 1). However, the lower limits of dissection (3.8 and 9.5 mm) may also be considered as the safest distances when approaching this region (Table 2).

Medulla

FIGURE 6. Cross-section showing the medulla at the level of the inferior olivary nucleus (ION). The olivary bodies are surrounded by eloquent anatomic structures ventrally, dorsally, and medially. The PT is located anteriorly to the olivary body, at the ventral surface of the brainstem. The PT is separated from the olivary body by intrinsic fibers of CN XII. The olivary body is bordered medially by the medial lemniscus and the intrinsic fibers of CN XII and posteriorly by the reticular, vestibulospinal, and tectospinal tracts (TT). Dorsally, the olivary bodies are limited by the spinocerebellar tract, descending trigeminal tract, and spinothalamic tract (TST). The nucleus ambiguous (NA) and dorsal motor nucleus (DMN) of CN X lie farther posteriorly. The continuous line illustrates the distance from the pia to the entry and exit points of the fibers in the ION (hilum). This intricate architecture implies that the safest entry zone in the anterolateral surface of the medulla is the ION. HN, hypoglossal nucleus; MLF, medial longitudinal fascicle.

NEUROSURGERY

In the medulla, transverse sections were performed through the inferior olivary nucleus at the most protruding point in the anterior aspect. Its craniocaudal axis averaged 13.5 mm, the transverse axis 7.0 mm, and anterodorsal axis 2.5 mm. Our measurements revealed that the length from the pial surface of the olivary body to the point where the different groups of fibers enter the olivary body ranged between 4.7 and 6.92 mm (5.52 0.5 mm) (Fig. 6). However, the lower limit of dissection (4.7 mm) may be also considered as the safest distance to approach this region (Table 2).

TABLE 1. Measurements in the mesencephalon and ponsa Measurement LMS length

Mean SD (mm)

Range (mm)

9.6 1.41

13.3–7.4

8.02 1.76

4.9–11.7

Superior one-third

12.2 0.97

10.8–13.7

Medial one-third

8.84 0.98

7.5–10.6

Inferior one-third

LMS to CN III Descending fibers of pons

5.6 0.95

4.2–7.5

Deep CN V

11.2 1.29

9.15–13.1

CN V to PT

4.64 0.68

3.1–5.6

a

SD, standard deviation; LMS, lateral mesencephalic sulcus; CN, Cranial Nerve; PT, pyramidal tract.


RECALDE ET AL.

TABLE 2. Summary of the brainstem safe entry zones and their working distancesa Brainstem level Midbrain Pons Medulla a

Safe entry zones

Minimum working distance (mm)

Maximum working distance (mm)

Lateral mesencephalic sulcus

4.9

11.7

8.2 1.76

Peritrigeminal area

9.5

13.1

11.2 1.29

Olivary body

4.7

6.92

5.52 0.5

Mean ⴞ SD (mm)

SD, standard deviation.

FIGURE 8. Anterior view of the brainstem. The transverse fibers have been dissected away and CN V as well as the PT can be observed. CNs VII and VIII present an inferior trajectory coming through the pontomedullary sulcus. The white line shows the distance from the PT to the fibers of CN V. The gray line indicates the distance of the intrinsic trajectory of CN V from the nucleus to the surface of the pons.

DISCUSSION

FIGURE 9. Lateral view showing the brainstem after white fiber dissection. The PT without the transverse fibers as well as the intraneural trajectory of CN V and its nucleus are depicted. The inferior cerebellar peduncle has been cut and, medial to it, the trapezoid body. SCP, superior cerebellar peduncle; ICP, inferior cerebellar peduncle; DN, dentate nucleus; TT, tectospinal tract.

Surgery of the BCM represents a formidable surgical challenge. This region shares the passage of afferent and efferent fibers, the reticular system, cranial nerves, and extrapyramidal nuclei, and their surgical manipulation may confer high levels of surgical morbidity (2, 12–14, 18). Several clinical series have reported acceptable rates of surgical morbidity and mortality for BCMs in the brainstem (9, 10, 13, 14, 17, 20). As more experience has been acquired in treating these lesions, it has become apparent that surgical resection is possible with acceptable risks of postoperative morbidity and mortality when compared with the higher risks of neurological deficits that have been demonstrated after multiple bleedings in this delicate region (2, 12). A better understanding of the surgical anatomy allied with more accurate preoperative studies, including magnetic resonance imaging and frameless stereotactic systems, are factors that have contributed to the contemporary management of BCMs. Surgical treatment is a valuable option for symptomatic patients, particularly when the lesion abuts the pial or ependy-


mal surface. In these cases, the lesion itself provides the surgical access route. Conversely, when only a thin layer of parenchyma separates the lesion from the surface, excision may also be considered if such a rim is situated along the path of a safe entry zone. These areas correspond to regions where critical neural structures are sparse and no perforating arteries are present. Additionally, a safe entry zone should provide satisfactory surgical room and should be located at an appropriate distance from vital structures. With a clear mental image of the internal architecture of the brainstem, a safe entry zone may be used to approach intrinsic lesions, thereby avoiding more perilous areas. The anatomic studies related to surgical access to the brainstem are mostly related to approaches through the floor of the fourth ventricle (1, 2, 10, 12). Numerous reports have described the different cranial base approaches used to reach different regions of the brainstem (12, 15). Although safe entry zones in the anterolateral aspect of the brainstem have been described and used by different authors (1, 2), to the best of our knowledge, no study has practically evaluated the intrinsic microsurgical anatomy of the brainstem related to the safe entry zones on its anterolateral aspect. A comprehensive analysis of the intrinsic anatomy of the brainstem provides a better knowledge of these safe entry areas and their relationships with eloquent internal anatomic structures. The brainstem is densely packed with many vital structures such as long ascending and descending pathways and specific nuclear groups. The core of the brainstem is occupied by the reticular formation. In general, the internal architecture may be divided into two functional regions. During embryogenesis, the neural tube is divided into a dorsal sensory portion (the alar plate) and a ventral motor portion (the basal plate) by a longitudinal indentation along the wall of the neural tube, the sulcus limitans. In maturity, this sulcus continues to be recognizable in the walls of the third and fourth ventricles and the cerebral aqueduct, and it still demarcates a border between sensory and motor structures. The safe entry zones analyzed in this article correspond roughly to the superficial projection of the transition between these two zones. In these regions, the distances between eloquent structures within the brainstem are the largest and provide an optimized surgical working space.

Mesencephalon The midbrain is limited cranially from the diencephalon by the sulcus between the optic tracts and the cerebral pedun-



cles, and separated from the pons by the pontomesencephalic sulcus. It is formed by the cerebral peduncles, the tegmentum, and the tectum. A narrow channel, the aqueduct of Sylvius, traverses the midbrain from top to bottom and divides it into two portions (Fig. 2). The dorsal part is formed by the tectum and the ventral part by the cerebral peduncles. The tegmentum, which is predominantly cellular, is separated from the peduncles by the substantia nigra. The substantia nigra delineates two sulci on the surface of the midbrain: the LMS and the medial sulcus of the cerebral peduncles (Fig. 2). These two sulci indicate the limits between the peduncles and tegmentum on the brainstem surface. The LMS is located in the posterior aspect of the mesencephalon, between the crus cerebri and the collicular area (Fig. 1). It extends vertically from the medial geniculate body superiorly, to the pontomesencephalic sulcus inferiorly. The length of this sulcus is described in Table 1. Lateral to the LMS, the bulging made by the parieto-occipitotemporopontine fibers may be identified. Posteromedially, the lateral lemniscus and the brachium conjunctivum are located inferior and superior, respectively, to the LMS (Figs. 2 and 7). Incision in this entry zone provides a working space that is limited ventrally by the substantia nigra, dorsally by the medial lemniscus, and medially by the fibers of CN III (Figs. 2 and 3). The surgeon may select an extension of nearly 1.0 cm (range, 7.4–13.3 mm) to do the myelotomy and work up to 8.2 mm deeper (range, 4.9–11.7 mm) without injuring the fibers of CN III and keeping a comfortable distance from the motor tracts. However, the lower limit of dissection (4.9 mm) may also be considered as the safest distance when approaching this region. Additionally, this entry zone route is located in a border area of the anterolateral and anteromedial vascular areas as described by Duvernoy (3). It provides a line of access in a hypovascularized region without compromising any important perforator (Fig. 10).

Pons The pons is limited superiorly by the pontomesencephalic sulcus and inferiorly by the pontomedullary sulcus. Its ventral aspect has a convex shape from side to side and from top to bottom and presents horizontal striae resulting from the presence of many transverse fibers (Figs. 5 and 11). These fibers converge ventrally to form a voluminous bundle, the medial cerebellar peduncles, from which emerge the roots of the trigeminal nerve. The apparent origin of CN V is considered to be the limit between the pons and the medial peduncles (Figs. 3–5 and 11). The corticonuclear and pyramidal tracts may be projected onto the surface of the pons. Two imaginary lines project the trajectory of the motor tracts onto the surface of the brainstem. The first line links the most medial point of the pontomesencephalic sulcus to the point where the medial border of the pyramid and the pontomedullary sulcus meet (Fig. 12). The second line links

NEUROSURGERY

the intersection between the lateral aspect of the crus cerebri and the pontomesencephalic sulcus to the point where the lateral border of the pyramid in the medulla and the pontomedullary sulcus meet (Fig. 12). The safe area may be FIGURE 10. Vascular topogradefined in a similar fashion. phy showing the mesencephalon This surgical area, called the with the anteromedial area in peritrigeminal area, has a tridark gray, the anterolateral area angular shape with the base in white, the lateral area in gray, located inferiorly. The medial and the posterior area in light limit is formed by the pyramgray. The relative hypovascularidal tract, the base by the ponized area corresponds to the LMS tomedullary sulcus from the in the surface. lateral aspect of the pyramid to the flocculus, and the lateral limit by a line that goes from Point A and passes through CN V. This triangle is wider in the base and stretches as it extends superiorly (Fig. 12). The floor of the triangle is formed by the most posterior fibers of the middle cerebellar peduncle. Myelotomy in this area carFIGURE 11. Brainstem anterior ries a lower risk of injury to aspect. P, pyramid. any vital structure because one will find the transverse fibers of the middle cerebellar peduncle. The nuclei of CNs VI, VII, and VIII are located posterior and medial to the nucleus of CN V. The trajectory of fibers of CNs VI, VII, and VIII has a descendent direction toward the pontomedullary sulcus. The transverse fibers of the FIGURE 12. On the left side pons are mostly horizontal or (horizontal lines), the PT may be slightly oblique; thus, myeloprojected onto the surface of the tomies in the anterolateral pons. On the right side (vertical aspect of the pons should be lines), the peritrigeminal area performed in a horizontal may be projected with the pondirection to preserve the tomedullary sulcus as the inferior fibers. This area provides a limit, the projection of the pyramsurgical window that measidal tract medially, and a tangenures 4.64 mm (range, 3.8–5.6 tial line passing posterior to CN V laterally. mm) horizontally and 11.2 mm (range, 9.5–13.1 mm) vertically. However, the lower limits of dissection (3.8 and 9.5 mm) may be also considered as the safest distances when approaching this region.


RECALDE ET AL.

Medulla The medulla has paired elongated elevations on its ventral aspect, the pyramids, separated by a longitudinal groove, the anterior median fissure. Oval protuberances, the olivary bodies, are found lateral to the pyramids and to the anterolateral sulcus. Ventral to the olivary bodies, CN XII originates from the anterolateral sulcus. The posterolateral sulcus on the dorsal margin of the olivary bodies gives rise to the rootlets of the accessory nerve, whereas CNs IX and X emerge from the pontomedullary sulcus, lateral to the olivary bodies. The medulla has an oval shape with a hilum located in its inferomedial aspect (17). It has afferent fibers, mostly from the reticular system, and gives rise to axons that cross the midline and course toward the contralateral cerebellum through the inferior cerebellar peduncle, forming the olivocerebellar fibers. The afferents emerge from the reticular formation and pass through the central tegmental fascicle. The medulla also has connections with the spinal cord and the cerebrum through the central tegmental fascicle. The specific function of the inferior olivary nucleus is controversial, and it is thought to be related to some of the functions of the cerebellum. The olivary bodies are limited ventrally by the corticospinal tract, medially by the medial lemniscus and fibers of the hypoglossal nerve, and dorsally by parasympathetic nuclei and fibers (Fig. 6). A group of fibers formed by the reticular and vestibulospinal tracts are placed posterior to the olivary body, and the fibers of the medial and lateral spinocerebellar tract may be found dorsal to them. Between these tracts, the fibers of CN X are encountered coursing toward the retro-olivary sulcus. The reticular and vestibulospinal tracts are interposed between the inferior olivary nucleus and the CN X nucleus. This architecture of the medulla makes a surgical approach quite dangerous because these important structures are densely concentrated in such a small area. There is no evidence demonstrating that isolated lesions of the olivary body may cause permanent deficits. Hence, the retro-olivary sulcus corresponds to the safest approachable area in the anterolateral surface of the medulla. The olivary body offers a surgical space of approximately 13.5 mm in the craniocaudal axis, 7 mm in the transverse diameter, and 2.5 mm in its anterodorsal axis. Our data suggest that the surgeon can reach lesions located in the anterolateral aspect of the medulla up to two-thirds of its transverse diameter.

Surgical Considerations The safe entry zones studied here are usually approached using standard surgical accesses (4–8). The LMS may be approached through subtemporal or supracerebellarinfratentorial routes. The peritrigeminal area may be reached through a subtemporal, retrosigmoid, pretemporal transcavernous approach or through transpetrosal techniques, whereas the olivary bodies are usually approached through a far lateral transcondylar technique. Preoperative magnetic resonance imaging studies may help to select the appropriate approach, which will also depend on the surgeon’s familiarity and expe-


rience in using these techniques. Additionally, frameless stereotactic systems and physiological monitoring may assist in surgical resection and reduce postoperative morbidity.

CONCLUSION Intrinsic BCMs constitute a formidable surgical challenge, and their approach has been considered a journey to terra incognita (16). It requires a meticulous technique and a perfect understanding of the inherent anatomy of the brainstem. Although such anatomy may suffer distortion after a hemorrhage, it is still important to recognize areas where the initial myelotomy will provide enough surgical working space, will not damage vital structures, and will not compromise perforator vessels. This study has demonstrated that a significant extent of the brainstem anatomy may be safely accessed through three entry zones. Use of these accesses does not damage any vital structures located deeper below or perforators, and these entry zones afford satisfactory surgical working space because the local arrangement of the brainstem anatomy provides adequate distances between important neural elements. In the mesencephalon, the LMS confers access to lesions in its posterior twothirds. In the pons, the peritrigeminal area provides a satisfactory approach for anterolateral and deep lesions located up to 10 mm from the pial surface. In the medulla, the olivary nucleus offers an adequate entry zone to reach lesions in its anterolateral aspect. The mean depth that can be safely accessed through the olivary body with a low morbidity rate is 5.52 mm. White fiber dissection improves the understanding of the microsurgical anatomy of the brainstem and offers a better comprehension of this complex structure. This knowledge and its clinical application may help to reduce the surgical morbidity associated with BCMs.

REFERENCES 1. Bricolo A: Surgical management of intrinsic brain stem gliomas. Oper Tech Neurosurg 3:137–154, 2000. 2. Cantore G, Missori O, Santoro A: Cavernous angiomas of the brain stem. Intraaxial anatomical pitfalls and surgical strategies. Surg Neurol 52:84–94, 1999. 3. Duvernoy HM: The Human Brainstem. New York, Springer-Verlag, 1995. 4. Figueiredo EG, Deshmukh P, Zabramski JM, Preul MC, Crawford NR, Siwanuwatn R, Spetzler RF: Quantitative anatomic study of three surgical approaches to the anterior communicating artery complex. Neurosurgery 56:397–405, 2005. 5. Figueiredo EG, Deshmukh P, Zabramski JM, Preul MC, Crawford NR, Spetzler RF: The pterional-transsylvian approach: An analytical study. Neurosurgery 59:263–269, 2006. 6. Figueiredo EG, Gomes M, Vellutini E, Rosemberg S, Marino R Jr: Angioleiomyoma of the cavernous sinus: Case report. Neurosurgery 56:E411, 2005. 7. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Preul MC, Spetzler RF: Anatomical and quantitative description of the transcavernous approach to interpeduncular and prepontine cisterns. Technical note. J Neurosurg 104:957–964, 2006. 8. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Spetzler RF, Preul MC: Comparative analysis of anterior petrosectomy and transcavernous approaches to retrosellar and upper clival basilar artery aneurysms. Neurosurgery 58:13–21, 2006.



9. Fritschi JA, Reulen HJ, Spetzler RF, Zabranski JM: Cavernous malformations of the brain stem. Acta Neurochir (Wien) 130:35–46, 1994. 10. Kyoshima K, Kobayashi S, Gib H, Kuroyanagi T: A study of safe entry zones via the floor of the fourth ventricle for brain-stem lesions. J Neurosurg 78:987–993, 1993. 11. Ludwig E, Klingler J: Atlas Cerebri Humani. Basel, S. Karger, 1956. 12. Porter RW, Detwiler PW, Spetzler RF: Surgical technique for resection of cavernous malformations of the brain stem. Oper Tech Neurosurg 3:124–130, 2000. 13. Porter RW, Detwiler PW, Spetzler RF, Lawton MT, Baskin JJ, Derksen PT, Zabramski JM: Cavernous malformations of the brainstem: Experience with 100 patients. J Neurosurg 90:50–58, 1999. 14. Samii M, Eghbal R, Carvalho GA, Matthies C: Surgical management of brainstem cavernomas. J Neurosurg 95:825–832, 2001. 15. Sarma S, Sekhar LN: Brain stem cavernoma excised by subtemporalinfratemporal approach. Br J Neurosurg 16:172–191, 2002. 16. Spetzler RF: Cavernous malformations of the brainstem. Presented at the Brazilian Congress of Neurosurgery, Florianopolis, Santa Catarina, Brazil, September 14–19, 2006. 17. Strauss C, Lutjen-Drecoll E, Fahlbusch R: Pericollicular surgical approaches to the rhomboid fossa. Part I. Anatomical basis. J Neurosurg 87:893–899, 1997. 18. Testut L: Human Anatomy Treaty [in Spanish]. Barcelona, Salvat, 1964, pp 689–692. 19. Wang CC, Liu A, Zhang J, Sun B, Zhao Y: Surgical management of brain-stem cavernous malformations: Report of 137 cases. Surg Neurol 59:444–454, 2003. 20. Ziyal IM, Sekhar LN, Salas E, Sen C: Surgical management of cavernous malformations of the brain stem. Br J Neurosurg 13:366–375, 1999.

Acknowledgments We thank Pushpa Deshmukh, Ph.D. for help with editing the manuscript and Israel M. de Oliveira for excellent technical assistance.

COMMENTS

A

s our understanding of the microsurgical anatomy of the brainstem improves, neurosurgeons are reconsidering the treatment of lesions once considered inoperable. This article represents a valuable complement to available clinical experience with brainstem cavernous malformations. To be sure, the corridor created by the cavernoma represents the safest entry point (1). However, when lesions do not abut the pia, other corridors need to be considered. In our experience, the lateral entry points between the motor and sensory nuclei described in this manuscript are well tolerated. An important limitation of this study is that the unique anatomy of a brainstem with pathological changes cannot be modeled in a cadaveric analysis. Cranial nerve nuclei and tracts are often distorted. Consequently, the limits of dissection presented in this article represent rough estimates of safe zones and should not be applied strictly. In the future, magnetic resonance imaging tractography may provide detailed information about the location of white matter tracts and assist with preoperative planning for brainstem lesions. We look forward to a follow-up review of the group’s clinical experience. Andrew Little Pushpa Deshmukh Robert F. Spetzler Phoenix, Arizona

fibers dissection technique in frozen human brainstem specimens. The study is focused on three anatomical landmark zones of the ventrolateral surface of the brainstem, namely the lateral mesencephalic sulcus, the pretrigeminal area, and the olivary bodies. Starting from the surface of these areas, they penetrate into the brainstem by dissecting in a three-dimensional way, being able to establish the distance from the surface and the locations of eloquent neurostructures. Now, with the data coming from this study, we know, for example, that by using the lateral mesencephalic sulcus as an entry zone, we find a space limited ventrally by the substantia nigra, dorsally by the median lemniscus, and medially by the third ocular nerve that we can meet at a median distance of 8.2 mm deep, but it is necessary not to go beyond 5 mm to be sure to avoid surgical injury to the ocular motor nerve, because in one case they found the third cranial nerve at a depth of 4.9 mm. The same can also be said of the other functionally important structures that we can meet using the other two “safe” entry zones that are the subjects of this study. We should be grateful to the authors for this straightforward, well-designed, and carefully executed study that allows us to follow the possibility of new paths and know in advance the expected distances of the obstacles to be encountered more precisely, hence certainly helping the surgeon when entering the brainstem. Regarding the clinical application of these data, i.e., direct surgery for removal of purely intrinsic lesions, one must keep in mind, however, that not infrequently the lesion itself, whether an astrocytoma or cavernoma, can move the eloquent structures away from their original site while growing, and therefore the surgeon may find them in a different place from that of the predicted normal anatomy. Although it may be useful to have knowledge of the normal distribution of anatomy as the authors explain to us, for the surgeon to move safely in a place anatomically distorted by a space-occupying lesion, there is also a need to identify the same structures functionally by using intraoperative neurophysiological motor monitoring and mapping (1). With technical advances in magnetic resonance imaging tractography in identifying even corticobulbar tracts, the surgeon will have another form of assistance in performing safe surgery in this risky setting. Albino Bricolo Verona, Italy

1. Sala F, Lanteri P, Bricolo A: Motor evoked potential monitoring for spinal cord and brain stem surgery. Adv Tech Stand Neurosurg 29:133–169, 2004.

R

ecalde et al. report the microanatomy of the brainstem related to the different safe entry zones used to approach intrinsic brainstem lesions. They point out three anatomic landmarks on the anterolateral brainstem surface: lateral mesencephalic sulcus, peritrigeminal area, and olive. They assessed the characteristics of this region in detail using frozen brainstem specimens. In addition, they presented clinical application of this anatomical knowledge. This is an important report to add novel information regarding the surgical treatment for brainstem lesions. We agree with their concept. The lateral mesencephalic sulcus, peritrigeminal area, and olivary body can provide surgical spaces and the relatively safe alleys through which the brainstem can be approached without injuring important neural structures. Yasushi Takagi Nobuo Hashimoto Kyoto, Japan

1. Porter RW, Detwiler PW, Spetzler RF, Lawton MT, Baskin JJ, Derksen PT, Zabramski JM: Cavernous malformations of the brain stem: Experience with 100 patients. J Neurosurg 90:50–58, 1999.

T

he authors have made an important contribution to our knowledge of intrinsic microsurgical anatomy of the brainstem by using a

NEUROSURGERY

T

his is an elegant anatomic demonstration with relevant clinical implications by Recalde et al., reflecting the senior author ’s


RECALDE ET AL.

anatomic school, in the best Rhotonesque tradition. Others had commented on brainstem safe entry zones, including the dorsolateral midbrain, the anterolateral pons just caudal to the fifth nerve root entry zone, and the anterolateral medulla at the anatomic olivary body. The authors describe these areas with supreme anatomic detail, including demonstration of the relative paucity of eloquent tracks and nuclei, and arterial perforator groups, in the vicinity of one of the most “expensive real estates” in the human brain. With regard to surgical exposure of these zones, the authors mention the supracerebellar infratentorial approach to the lateral midbrain. We have more recently favored the far posterior subtemporal approach, with tentorial section, as described by Smith et al. (1). Like the authors, we also use the presigmoid transpetrosal approach with tentorial section to reach the anterolateral pontomesencephalic region and the far lateral transcondylar approach to reach the medullary olivary body. However, the overriding choice of surgical route to brainstem cavernous malformations must remain the area wherein the lesion reaches pial or ependymal surfaces, especially if exophytic, so that there is little or no invasion of any unaffected brainstem. These so-called “safe entry zones” are not a substitute for choosing the area where as little brainstem parenchyma as possible is perturbed to reach the cavernoma. Internal decompression of the lesion is performed and then definition of the perlesional plane and gradual delivery of the lesion, in fragments, through as little a pial or ependymal opening as possible. Yet, when the lesion presents nearest to the surface in more than one area, we have also preferred such lateral safe routes whenever possible, posterolateral to the descending motor tracks, as an alternative to dorsal midline or paramedian approaches, which tend to result in disabling gaze and sensory morbidity. The authors are congratulated on their exquisite demonstration of anatomic subtleties, which will help in identifying the relevant landmarks at surgery, and their boundaries. Detailed anatomic imaging, including the advent of 3-T magnetic resonance will assist in the identification of related landmarks (sulci, nuclei, and tracks) when the surgical approach is planned. Frameless stereotactic image guidance can further enhance the fidelity of intraoperative orientation and navigation in reference to these structures as well as to the lesion. To complement the pathological and anatomic boundaries of the surgical approach, we and others have used a variety of functional monitoring techniques (motor, sensory, and auditory evoked potentials and cranial nerve electromyography) to further enhance the safety of these approaches. However, the alteration or loss of an electrophysiological response should not result in necessarily abandoning the task of lesion excision, nor does it totally predict permanent neurological deficit. But such changes do result in the alteration of specific microsurgical maneuvers during the resection, such as dissection in a specific subzone. This article adds a substantive reference to the seminal literature on brainstem surgery. It will be a much used guide in last-minute planning before execution of these delicate procedures. Issam A. Awad Evanston, Illinois

1. Smith ER, Chapman PH, Ogilvy CS: Far posterior subtemporal approach to the dorsolateral brainstem and tentorial ring: Technique and clinical experience. Neurosurgery 52:364–368, 2003.

I

n this article, the authors have studied the microsurgical anatomy of three safe entry zones into the brainstem for the purpose of removing cavernous malformations of the brainstem in the midbrain, pons, and


medulla. In practice, the majority of brainstem cavernomas are operated on through the area of the brainstem where it comes closest to the pial surface, although this may not always be the safest approach. When the cavernoma comes close to the surface, one may assume that many of the neighboring tracts and nuclei are displaced, similar to what is found with brain mapping of cortical and subcortical lesions. However, at present, we have no reliable way of locating the displaced nuclei and fiber tracts, although advances are being made with diffusion tensor magnetic resonance imaging. The practical difficulty is caused by the presence of the hemosiderin ring around the cavernoma. In the future, I am optimistic that this problem will be solved. The other problem is that the precision of surgery within the brainstem is currently limited by what is possible by human hands and existing microsurgical techniques. We have also not developed techniques for intraoperative monitoring of many of the tracts inside the brainstem. The last issue concerns neural plasticity inside the brainstem. There definitely appears to be some potential for this, as evidenced by the recovery of our patients with brainstem lesions. Perhaps we will find ways of repairing the damaged neural tracts and nuclei in the future. Nevertheless, the results of surgery for brainstem cavernomas are acceptably good (unpublished personal data) in the hands of many expert neurosurgeons. Further advances in technology and biological knowledge will be needed to improve the current results and reduce the patients’ postoperative disabilities. Laligam N. Sekhar Seattle, Washington

R

ecalde et al. present a very important anatomical study of the brainstem that is based on the long-term experience of the senior author, Evandro de Oliveira, with numerous surgical interventions in this area. A modern study design—examination of the macroscopic external and internal anatomy of the brainstem strictly from the microsurgical point of view—has been combined with an old but very efficient method, the white fiber dissection technique. The chief merit of this systematic anatomical study is, in my opinion, the fact that the authors have not only elegantly confirmed the previous empirical experience of other neurosurgeons, namely that surgical manipulation within the brainstem is well-tolerated particularly in the transition zone between sensory and motor structures, but have also provided detailed measurement results that can be useful during surgery. These accurate distance measurements may also be valuable for the planning of the surgical procedure as they can be compared with similar measurements obtained from the magnetic resonance imaging of a specific patient undergoing surgery and later, adjusted to this specific patient, be used intraoperatively. Such measurements can be of great importance when an intrinsic cavernoma has distorted or displaced the local anatomical structures, particularly in patients with a large intralesional hematoma. My own experience with lesions involving the brainstem comprises 103 surgical cases of intrinsic cavernomas and 50 cases of intrinsic gliomas of the brainstem, apart from numerous other extraaxial lesions such as schwannomas, meningiomas, and hemangioblastomas. In a significant number of the patients I have operated on for intrinsic brainstem cavernomas, the vascular malformation was not visible on the surface of the brainstem and in some patients not even a bulging or slight discoloration was present. I consider these the most challenging cases. Moreover, these are the real cases that require a “safe entry zone” because an apparently healthy brainstem has to be opened as closely as possible to a lesion that is invisible on its surface. It is obvious that the decision of where exactly to open the brainstem in such patients requires great experience, profound anatomical knowledge, and careful analysis of preoperative imaging studies before surgery. Computer-



assisted neuronavigation may be of some help in these situations, but the accuracy of this method is not always sufficiently high for complete reliance on this tool. As I have previously mentioned, the anterolateral aspect of the pons that corresponds to one of the three safe entry zones of this study—the peritrigeminal area—tolerates surgical manipulation without the consequence of permanent neurological deficits (1). I have also used other entry zones to the brainstem that turned out to be “safe” as well, for instance, a restricted area at the level of the pontomedullary sulcus between the exit of the rootlets of cranial nerves VII and IX or on the anterior aspect of the crus cerebri just lateral to the oculomotor nerve. It is important to mention, however, that with only very few exceptions I do not “incise” the brainstem in the sense of using a knife or microscissors. In my opinion, a true myelotomy may easily be harmful and should be avoided whenever possible. Instead of incising the brainstem, I just puncture its surface with the thinnest bipolar forceps and then gradually dilate this opening up to 5 or 6 mm in diameter, which may be sufficient to first expose and then remove the underlying intraaxial lesion. Small cottonoids of approximately 3 mm width are helpful for this step. With this technique, superficial structures of the brainstem such as longitudinal or transverse fibers can be temporarily

displaced (dilated) without completely sectioning them. A similar displacement and dilatation of intrinsic anatomical structures is also produced by the cavernoma and intralesional hematoma itself, and, apparently, this is a reversible process as can be seen on many postoperative images. Permanent parenchymal damage may only occur when either a significant number of neurons or fibers are mechanically destroyed by surgical manipulation or important perforating arteries or draining veins (including branches from an associated venous malformation) have been coagulated during surgery. In summary, I consider the present study a most valuable contribution to our knowledge of brainstem anatomy. The results presented here may be of significant help to many neurosurgeons who operate on difficult lesions such as intraaxial vascular malformations or gliomas of the brainstem. Helmut Bertalanffy Zurich, Switzerland 1. Bertalanffy H, Benes L, Miyazawa T, Alberti O, Siegel AM, Sure U: Cerebral cavernomas in the adult: Review of the literature and analysis of 72 surgically treated patients. Neurosurg Rev 25:1–53, 2002.

The Anatomical Theater at Leiden, (1610), copper engraving. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


ANATOMY Technique Assessment

IMAGE-GUIDED LATERAL SUBOCCIPITAL APPROACH: PART 1—INDIVIDUALIZED LANDMARKS FOR SURGICAL PLANNING Alireza Gharabaghi, M.D.

Ramin Shahidi, Ph.D.

OBJECTIVE: Being situated close to the transverse and sigmoid sinus, the asterion has traditionally been viewed as a landmark for surgical approaches to the posterior fossa. Cadaveric studies, however, have shown its variability in relation to underlying anatomic structures. We have used an image-guidance technology to determine the precise anatomic relationship between the asterion and the underlying transverse-sigmoid sinus transition (TST) complex in patients scheduled for posterior fossa surgery. The applicability of three-dimensional (3-D) volumetric image-rendering for presurgical anatomic identification and individualization of a surgical landmark was evaluated. METHODS: One-millimeter computed tomographic slices were combined with venous computed tomographic angiography in 100 patients, allowing for 3-D volumetric imagerendering of the cranial bone and the dural vasculature at the same time. The spatial relationship between the asterion and the TST was recorded bilaterally by using opacity modulation of the bony surface. The location of both the asterion and the TST could be confirmed during surgery in all of these patients. RESULTS: It was possible to accurately visualize the asterion and the sinuses in a single volumetrically rendered 3-D image in more than 90% of the patients. The variability in the anatomic position of the asterion as shown in cadaveric studies was confirmed, providing an individualized landmark for the patients. In this series, the asterion was located from 2 mm medial to 7 mm lateral and from 10 mm inferior to 17 mm superior to the TST, respectively. CONCLUSION: Volumetric image-rendering allows for precise in vivo measurements of anatomic distances in 3-D space. It is also a valuable tool for assessing the validity of traditional surgical landmarks and individualizing them for surgical planning.

Image Guidance Laboratories, Stanford University, Palo Alto, California

KEY WORDS: Asterion, Cranial base, Image-guidance, Sigmoid sinus, Skull base, Surgical anatomy, Threedimensional surgical planning, Transverse sinus

Marcos Tatagiba, M.D., Ph.D.


Eberhard Karls University Hospital, Tübingen, Germany, and International Neuroscience Institute, Hannover, Germany

Steffen K. Rosahl, M.D., Ph.D. Helios Hospital, Erfurt, Germany

Günther C. Feigl, M.D. Eberhard Karls University Hospital, Tübingen, Germany

Thomas Liebig, M.D. Technical University Hospital, Munich, Germany

Javad M. Mirzayan, M.D. Medical University Hospital, Hannover, Germany

Stefan Heckl, M.D. Eberhard Karls University Hospital, Tübingen, Germany

DOI: 10.1227/01.NEU.0000297024.52502.BB

Eberhard Karls University Hospital, Tübingen, Germany

Madjid Samii, M.D., Ph.D. International Neuroscience Institute, Hannover, Germany Reprint requests: Alireza Gharabaghi, M.D., Department of Neurosurgery, University Hospital Tübingen, Charlottenstr. 37/1, 72070 Tübingen, Germany. Email: alireza.gharabaghi@ uni-tuebingen.de Received, February 25, 2006. Accepted, January 26, 2007.

S

urface landmarks are widely used for surgical planning and intraoperative orientation. The asterion, the junction of the lambdoid, parietomastoid, and occipitomastoid sutures (4, 27), has been used in posterior fossa surgery to locate the transverse-sigmoid sinus transition (TST) complex (7, 9, 23–26). Because cadaveric studies have shown considerable variability in the relationship of the asterion to the underlying venous structures (8, 29, 33), its validity as a surgical landmark has been questioned (8, 21). Although most experienced surgeons are aware of the limitations of surgical landmarks to predict the precise location


of hidden intracranial vasculature, neurosurgeons in training may benefit from new means by which surface landmarks can be reliably translated into surgical approaches. Recent advances in the three-dimensional (3-D) rendering of computed tomographic (CT) images (10, 15) have made it possible to visualize minute bone sutures and venous structures of the posterior fossa at the same time (14, 30). We hypothesize that this technology may provide individualized landmarks that allow surgeons to predict the location of the asterion relating to the TST. Therefore, we have preoperatively studied the in vivo relationship of the asterion


INDIVIDUALIZED SURGICAL LANDMARKS

A

B

FIGURE 2. The asterion (red) and the “inner knee” of the transverse–sigmoid sinus transition (TST) junction (orange) are defined, and their distance is measured in each of the triaxial planes (here the axial plane) using the capabilities of the workstation.

FIGURE 1. A and B, by gradually decreasing the opacity of the cranium, it is possible to visualize the relationship between the asterion and the transverse–sigmoid transition (blue). For better orientation, the circle of Willis (red) and a tumor in the cerebellopontine angle (green) are highlighted.

to the underlying TST complex in patients undergoing surgery for cerebellopontine angle lesions and have applied this information to surgical planning.

PATIENTS AND METHODS One hundred patients (52 male, 48 female; mean age, 48 yr [range, 11–78 yr]) scheduled for posterior fossa surgery (88 vestibular schwannomas, 10 meningiomas, two epidermoids) underwent high-resolution CT scans in combination with venous angiography in a multislice scanner (Somatom Plus 4; Siemens, Erlangen, Germany). Nonionic contrast medium (100 ml of 300 mg/ml solution) was injected with a pump into the cubital vein at a rate of 3.5 ml/s. After a patient-specific delay determined by a bolus test (28–37 s), 100 to 120 consecutive scans were obtained in 1-mm thick slices at 1 mm/s table-forward speed (1024 ⫻ 1024 matrix, 120 kV, and 240 mA). The radiation dose was calculated for each patient by using dedicated software (CT-EXPO V 1.2; Medical University Hannover, Hannover, Germany), resulting in a mean effective radiation dose of 7.7 to 11.3 mSv. The slice thickness of 1 mm is routinely acquired in all cases of lateral suboccipital craniotomies, which are referred to our institutions

NEUROSURGERY

independent of the purposes of the present study. We preoperatively evaluate the course of emissary veins in the area of the craniotomy as well as the height of the jugular bulb on the side of the intended transmeatal approach. Moreover, we use these high-resolution images to analyze the distance of both the posterior semicircular canal and the crus commune of the labyrinth to the intrameatal tumor extension in relation to the dorsal wall of the internal auditory canal. For the purpose of this study, the axial source images were transferred to the workstation of a surgical navigation system (Image Guidance Laboratories, Palo Alto, CA). Three-dimensional volumetrically rendered images of the cranium and vasculature were acquired with a minimum density threshold of approximately 60 to 80 HU and a maximum density threshold of 300 to 400 HU. These volumetric 3-D models allowed for visualization of bony landmarks and venous structures at the same time. The spatial relationship between the asterion and the underlying TST complex was recorded on both sides for each patient, yielding a total of 200 evaluated sides. By varying the perspective of the 3-D model interactively and modulating the transparency of the bony surface, accurate matching of the anatomic structures was achieved. The morphometric measurement of the anatomic distances was performed in two steps. First, two target points, the asterion and the inner radius (“inner knee”) of the TST, were defined within the triaxial images and the same 3-D model by opacity modulation of the cranium bone, allowing for simultaneous visualization of the asterion and the underlying venous structures (Fig. 1). To identify the inner knee of the TST, the axial images were viewed in a bottom (caudal) to top (cranial) direction following the course of the sigmoid sinus. As soon as the first slice revealed a connection to the transverse sinus, the image viewer was scrolled one slice back. In this last axial image in which the sigmoid sinus was visible without the transverse sinus, the most posterior and medial part of the sigmoid sinus was defined as the inner


GHARABAGHI ET AL.

knee of the TST (Fig. 2, orange crossbar). In the second step, the distances between the target points were measured in each of the triaxial planes by using specific features of the image-guidance software (Fig. 2). Thereafter, the imaging findings (the asterion is located above/on top/below the TST) were used for surgical planning and were compared with the intraoperative findings, which served as the “gold standard” regarding the spatial relationship of the asterion to the TST. The accuracy of the image-guidance tool was investigated, calculating the intraoperative target registration error by comparing the in situ position of the navigation probe on a nonregistered fiducial or well-defined bony landmark with the localization of the corresponding site in the reconstructed triaxial images. The mean target registration error amounted to 1.5 mm (⫾ 0.6 mm) over all navigated cases.

A

B RESULTS Three-dimensional volumetric image-rendering was capable of accurate in vivo visualization of bony and venous structures at the same time in individual patients. The spatial relationship of the asterion and the underlying TST could be determined 91 times on the left side and 92 times on the right side before posterior fossa surgery. In the remaining patients, the sutures were calcified and not visible in the 3-D reconstructions. The variability of the anatomic position of the asterion shown in cadaveric studies was confirmed in vivo. On the left side, the asterion was located directly over the TST in 59 (65%), above the TST in the supratentorial region in nine (10%), and below the TST complex in 23 (25%) cases, respectively. On the right side, it was located on top of the TST in 68 cases (75%), above the TST in six cases (5%), and below in 18 cases (20%). This information was successfully used for surgical planning of the suboccipital craniotomy. During surgery, the identified asterion could be related to the intraoperative location of the TST. In all of these patients, the preoperatively image-rendered spatial relationship of the asterion to the TST could be confirmed during surgery regarding the qualitative classification (above/on top/below the TST) (Fig. 3; Table 1). In our series, the asterion was located from 2 mm medial to 7.4 mm lateral and from 10 mm inferior to 17 mm superior to the TST, respectively. As a result of this range (⫺2 to 7.4 and ⫺10 to 17), the calculated mean distances of the asterion to the

FIGURE 3. Illustration depicting the lateral view of the TST junction (with mastoid contour for orientation) showing the location of the asterion (91 times for the left [A] and 92 times for the right [B] side) superimposed on it for each side.

TST were 2.3 mm lateral and 2.5 mm superior, although the asterion was located inferior and medial to the TST in the vast majority of patients (Table 1). The measurements for the right and left sides correlated well and did not differ significantly. The mean distances and standard deviations are shown in Figure 4. There was no significant difference for any of the distances with respect to sex, age, or side.

DISCUSSION Our results confirm the wide variability of the asterion described in cadaveric studies (Table 1) (8, 25, 29, 33). Day and

TABLE 1. Summary of studies on the anatomic position of the asterion in relation to the transverse–sigmoid sinus transitiona,b Series (ref. no.)

a b

Study type

Total number of studied sides

Location of the asterion (%) Above TST On TST Below TST L R L R L R

Ribas et al., 1994 (25)

Cadaver

50

2

Day and Tschabitscher, 1998 (8)

Cadaver

200

9

7

78 66

61

17 25

32

Sripairojkul and Adultrakoon, 2000 (29)

Cadaver

100

9

2

58

74

33

24

Uz et al., 2001 (33)

Cadaver

50

2

Present study, 2006

In vivo

200

10

54 5

65

44 75

25

20

TST, transverse–sigmoid sinus transition; L, left; R, right. In the present study, the asterion could be identified in 183 of 200 studied sides.




FIGURE 4. Mean distances (in millimeters) and standard deviations shown between the asterion and the TST in the x-, y-, and z-planes on the left (Xl, Yl, Zl) and on the right sides (Xr, Yr, Zr), respectively. SE, standard error of the mean.

Tschabitscher (8) found the asterion to be located below the TST complex in 25 and 32% of their cadavers for the left and right sides, respectively. Similar results were described by Sripairojkul and Adultrakoon (29), with the asterion located infratentorially in 33 and 24% of their cases, respectively. Uz et al. (33) found this anatomic position of the landmark in 44% of the cases in their data. With the asterion located below the TST in 25 and 20% of the patients on the left and right sides, respectively, our in vivo data correlate well with these studies. Based on this background, relying solely on the asterion as a surgical landmark may be hazardous and bears the risk of damaging the dural sinuses when placing a burr hole directly at this point (8, 29). Other landmarks have been advocated for approaches to the posterior fossa (7, 20, 23, 26, 28, 32). In a cadaver study, Lang and Samii (20) found that if the burr hole was placed 45 mm behind the suprameatal line and 7 mm below the Frankfurt horizontal plane, no sinus opened in 92% of the cases in a series of 37 cases. Rhoton (23) recommended placement of the initial burr hole 2.0 cm below the asterion, two-thirds behind, and one-third in front of the occipitomastoidal suture to avoid the posterior margin of the sigmoid sinus. Day et al. (7) advocated the superior nuchal line connecting the root of the zygoma with the inion as a landmark for the distal transverse sinus and the transverse–sigmoid junction. From their studies in 100 crania, Day and Tschabitscher (8) recommended placement of the burr hole inferior to this line and just behind the ridge that delimits the body of the mastoid bone. Malis (21) suggested drilling 2 cm medial to the mastoid tip. Sekhar (28) advised placing the first burr hole inferior and medial to a line drawn from the inion to the base of the mastoid process. Tubbs et al. (32) recommended using the insertion of the semispinalis capitis muscle as a landmark for the proximal

NEUROSURGERY

transverse sinus. They found that the insertion of the musculus semispinalis capitis corresponded with the transverse sinus in 80 to 93% of cases in a study of 15 cadaveric specimens. Finally, Ribas et al. (26) suggested the most posterior part of the parietomastoid suture and the occipitomastoid suture at the level of the mastoid notch as proper initial burr hole sites. New anatomic studies may bring to light new surface landmarks with less variability (20, 26, 33). The limitation in these studies is that their conclusions are derived from statistical data in a limited number of cadaveric specimens. For clinical practice, as a result of the wide individual variability of the distances, patient-specific anatomic information is desirable. For morphometric investigations with a larger number of cases, it would be helpful to have an accurate methodology that avoids cadaveric dissection by hand. With modern technology, CT and magnetic resonance imaging scans yield reliable morphometric data in individual patients (7). Current advances in image-rendering have made it possible to generate precise 3-D models by the reconstruction of image data (2, 16, 19). Three-dimensional CT angiography has been developed and applied for diagnostic and therapeutic purposes (6, 13, 31). Although image-based stereoscopic virtual reality models are already in use for planning specific surgical procedures and for teaching purposes in neurosurgery (3, 5, 12, 17, 18), most studies that evaluate preoperative neurosurgical planning have focused on 3-D visualization capabilities of the applied software (1, 11, 22, 34, 35). Although several 3-D models simultaneously showing the surface of the cranial bone and venous vasculature of the posterior fossa have been published (14, 30), the gradual, interactive modulation of opacity used to detect minute bone sutures in this study and their relationship to the underlying dural sinuses is a new feature. By using this technique, the spatial relationship of the asterion and the underlying TST complex was clearly identified in more than 90% of the cases. It was the basis for precise anatomic measurements that can be performed in vivo with the option of generating individualized patient-specific morphometric data in a clinical setting. When validated with these data, the asterion can still be used for surgical planning by determining the patient-specific coordinates of this landmark. Alternatively, the sinuses themselves may be turned into landmarks that are projected onto the surface of the cranium by intraoperative image guidance. This would allow the use of CT angiogram rather than magnetic resonance angiogram, thereby avoiding x-ray exposure. With respect to surgical training, however, it must be kept in mind that the new technology cannot replace hands-on anatomic study.

CONCLUSION Volumetric 3-D image-rendering appears to be a valuable tool for the validation of surface anatomic landmarks in individual patients and for basic morphometric investigations. More specifically, a simultaneous visualization and interactive modulation of bony surfaces and vascular structures is a suitable


GHARABAGHI ET AL.

method for determining the relationship of the asterion to the underlying TST complex, thereby providing an individualized landmark for surgical planning. Using this imaging technique, the asterion may be revived as a surgical landmark, provided that its individual relationship to the dural sinuses is known to the surgeon from previous imaging. Interactive opacity modulation in volumetrically rendered 3-D images offers new options for anatomic research and morphometric investigations.

REFERENCES 1. Abrahams JM, Saha PK, Hurst RW, LeRoux PD, Udupa JK: Threedimensional bone-free rendering of the cerebral circulation by use of computed tomographic angiography and fuzzy connectedness. Neurosurgery 51:264–269, 2002. 2. Aoki S, Sasaki Y, Machida T, Ohkubo T, Minami M, Sasaki Y: Cerebral aneurysms: Detection and delineation using 3-D-CT angiography. AJNR Am J Neuroradiol 13:1115–1120, 1992. 3. Balogh A, Preul MC, Schornak M, Hickman M, Spetzler RF: Intraoperative stereoscopic QuickTime Virtual Reality. J Neurosurg 100:591–596, 2004. 4. Bannister LH, Berry MM, Collins P, Dyson M, Dussek JE, Ferguson MWJ (eds): Gray’s Anatomy. New York, Churchill Livingstone, 1995, ed 38, p 56. 5. Bernardo A, Preul MC, Zabramski JM, Spetzler RF: A three-dimensional interactive virtual dissection model to simulate transpetrous surgical avenues. Neurosurgery 52:499–505, 2003. 6. Castillo M, Wilson JD: CT angiography of the common carotid artery bifurcation: Comparison between two techniques and conventional angiography. Neuroradiology 36:602–604, 1994. 7. Day JD, Kellogg JX, Tschabitscher M, Fukushima T: Surface and superficial surgical anatomy of the posterolateral cranial base: Significance for surgical planning and approach. Neurosurgery 38:1079–1084, 1996. 8. Day JD, Tschabitscher M: Anatomic position of the asterion. Neurosurgery 42:198–199, 1998. 9. Hakuba A, Nishimura S, Jang BJ: A combined retroauricular and preauricular transpetrosal–transtentorial approach to clivus meningiomas. Surg Neurol 30:108–116, 1988. 10. Harbaugh RE, Schlusselberg DS, Jeffery R, Hayden S, Cromwell LD, Pluta D, English RA: Three-dimensional computed tomographic angiography in the preoperative evaluation of cerebrovascular lesions. Neurosurgery 36:320–327, 1995. 11. Hayashi N, Endo S, Shibata T, Ikeda H, Takaku A: Neurosurgical simulation and navigation with three-dimensional computer graphics. Neurol Res 21:60–66, 1999. 12. Henn JS, Lemole GM Jr, Ferreira MA, Gonzalez LF, Schornak M, Preul MC, Spetzler RF: Interactive stereoscopic virtual reality: A new tool for neurosurgical education. Technical note. J Neurosurg 96:144–149, 2002. 13. Hsiang JN, Liang EY, Lam JM, Zhu XL, Poon WS: The role of computed tomographic angiography in the diagnosis of intracranial aneurysms and emergent aneurysm clipping. Neurosurgery 38:481–487, 1996. 14. Kaminogo M, Hayashi H, Ishimaru H, Morikawa M, Kitagawa N, Matsuo Y, Hayashi K, Yoshioka T, Shibata S: Depicting cerebral veins by three-dimensional CT angiography before surgical clipping of aneurysms. AJNR Am J Neuroradiol 23:85–91, 2002. 15. Kato Y, Sano H, Katada K, Ogura Y, Hayakawa M, Kanaoka N, Kanno T: Application of three-dimensional CT angiography (3D-CTA) to cerebral aneurysms. Surg Neurol 52:113–122, 1999. 16. Kikinis R, Gleason PL, Moriarty TM, Moore MR, Alexander E 3rd, Stieg PE, Matsumae M, Lorensen WE, Cline HE, Black PM, Jolesz FA: Computerassisted interactive three-dimensional planning for neurosurgical procedures. Neurosurgery 38:640–651, 1996. 17. Kockro RA, Serra L, Tsai YT, Chan C, Sitoh YY, Chua GG, Hern N, Lee E, Hoe LY, Nowinski W: Planning of skull base surgery in the virtual workbench: Clinical experiences. Stud Health Technol Inform 62:187–188, 1999. 18. Kockro RA, Serra L, Tseng-Tsai Y, Chan C, Yih-Yian S, Gim-Guan C, Lee E, Hoe LY, Hern N, Nowinski WL: Planning and simulation of neurosurgery in a virtual reality environment. Neurosurgery 46:118–137, 2000.


19. Koyama T, Okudera H, Kobayashi S: Computer-assisted geometric design of cerebral aneurysms for surgical simulation. Neurosurgery 36:541–547, 1995. 20. Lang J Jr, Samii A: Retrosigmoid approach to the posterior cranial fossa: An anatomical study. Acta Neurochir (Wien) 111:147–153, 1991. 21. Malis LI: Anatomical position of the asterion. Neurosurgery 42:198–199, 1998. 22. Mamata H, Komiya T, Muro I, Matsuyama S: Application and validation of three-dimensional data sets from a phase contrast MR angiography for preoperative computer simulation of brain tumors. J Magn Reson Imaging 10:102–106, 1999. 23. Rhoton AL Jr: Surface and superficial surgical anatomy of the posterolateral cranial base: Significance for surgical planning and approach. Neurosurgery 38:1083–1084, 1996. 24. Rhoton AL Jr: The cerebral veins. Neurosurgery 51 [Suppl 4]:S159–S205, 2002. 25. Ribas GC, Rhoton AL Jr, Cruz OR, Peace D: Temporo-parieto-occipital burrhole sites study and systematized approaches proposal, in Samii M (ed): Skull Base Surgery: Anatomy, Diagnosis and Treatment. Basel, Karger, 1994, pp 723–730. 26. Ribas GC, Rhoton AL Jr, Cruz OR, Peace D: Suboccipital burr holes and craniectomies. Neurosurg Focus 19:E1, 2005. 27. Seeger W: Planning Strategies of Intracranial Microsurgery. Berlin, SpringerVerlag, 1986. 28. Sekhar LN: Anatomical position of the asterion. Neurosurgery 42:198–199, 1998. 29. Sripairojkul B, Adultrakoon A: Anatomical position of the asterion and its underlying structure. J Med Assoc Thai 83:1112–1115, 2000. 30. Suzuki Y, Ikeda H, Shimadu M, Ikeda Y, Matsumoto K: Variations of the basal vein: Identification using three-dimensional CT angiography. AJNR Am J Neuroradiol 22:670–676, 2001. 31. Tampieri D, Leblanc R, Oleszek J, Pokrupa R, Melancon D: Threedimensional computed tomographic angiography of cerebral aneurysms. Neurosurgery 36:749–755, 1995. 32. Tubbs RS, Salter G, Oakes WJ: Superficial surgical landmarks for the transverse sinus and torcular herophili. J Neurosurg 93:279–281, 2000. 33. Uz A, Ugur HC, Tekdemir I: Is the asterion a reliable landmark for the lateral approach to posterior fossa? J Clin Neurosci 8:146–147, 2001. 34. Villavicencio AT, Leveque JC, Bulsara KR, Friedman AH, Gray L: Threedimensional computed tomographic cranial base measurements for improvement of surgical approaches to the petrous carotid artery and apex regions. Neurosurgery 49:342–353, 2001. 35. Zhao JC, Chen C, Rosenblatt SS, Meyer JR, Edelman RR, Batjer HH, Ciric IS: Imaging the cerebrovascular tree in the cadaveric head for planning surgical strategy. Neurosurgery 51:1222–1228, 2002.

COMMENTS

G

harabaghi et al. have used a new concept that is very clever and will improve surgical flow and safety: they present a novel way of using individual landmarks and apply it to the asterion. The findings confirm previous cadaveric data regarding the variability of the relationship between the sinus and the asterion. The main point is, however, a completely new way of approaching this kind of surgical landmark. The authors determined the relationship between the asterion and the sinus in each individual patient (which was possible in 90% of the patients). The surgeon gets a totally renovated means of knowing how to optimize the craniotomy; in posterior fossa surgery, the optimal placement of the craniotomy is even more sensitive than in other locations. Now it becomes much easier to place the craniotomy in perfect relation to the surgical corridor. However, I disagree with the authors’ interpretation and the interpretation of the success rate as the present technique was applicable in 90% of all patients and should not then be regarded as a 99% successful means of identifying the landmark. In addition, the identification of the pre- and intraoperative findings seem to be slightly repetitive as radiological data and navigation were used in both instances.



The conclusion still remains that this technique is clever and creative; the invention is seemingly simple. Even though the technology and patients have been available, nobody else thought of this technique. Tiit Mathiesen Stockholm, Sweden

T

he authors deal with an issue as old as neurosurgery itself—how does one approach the posterior fossa safely and what landmarks does one use for orientation? The asterion has been defined traditionally as such a landmark, even though, as the authors thoroughly demonstrated, multiple studies have proven a variance in relative location between asterion and underlying risk structures. To record the spatial relationship of these structures individually, the authors evaluated the usefulness of combined three-dimensional volumetric rendering of the skull and venous structures in this study. The authors could demonstrate that the method presented was feasible for locating the asterion in more than 90% of the patients evaluated and reconfirmed a wide variability of its relative anatomic location. The visualization software used by the authors allows for the gradual modulation of opacity of the skull, facilitating the detection of bone sutures. This feature obviously enhanced the localization of the asterion and relates to the high detection rates of this landmark in their series. The documented anatomic variability, the related risks of pure landmark-orientation, and the detrimental effects of severing the sinus, leading to air embolism, should propel any efforts that reduce surgical morbidity. As a result, this article offers further proof that image guidance is an essential tool to circumnavigate risk to structures while approaching the actual region of interest in the posterior fossa, whether it may be simply in the planning phase or preferentially intraoperatively, by use of a neuronavigation system. Thomas Gasser Volker Seifert Frankfurt, Germany

NEUROSURGERY

T

he authors add to the extensive literature indicating that the asterion is an unreliable surface landmark for the transverse sinus. I use the bump of bone formed by the intersection of the ridge behind the mastoid groove and the superior nuchal line (understanding this term to mean the superior border of the insertion of the cervical musculature) as a rough landmark for the burr hole for this approach. In my mind, angling the perforator so that the hub of the drill is posteroinferior to the bit is at least as important as the precise location of the burr hole. This angling allows the initial penetration of the bit through the bone (and the potential dural laceration) to be most likely over the posterior fossa dura rather than over the sinus. I am not sure from where the misconception about the value of the asterion as a surface landmark for safe burr hole placement arose. Neither the early descriptions of trephining for septic thrombophlebitis of the lateral sinus (1, 2) nor Dandy’s publications on unilateral suboccipital craniotomy (3, 4) mention the asterion explicitly. The illustrations in these influential publications do sometimes show the asterion as located inferior to the sinus (2, 4); perhaps this is the source of the mistaken belief. The same error can be found in modern illustrations as well (5). Fred G. Barker II Boston, Massachusetts

1. Arbuthnot LW: The treatment of pyaemia consequent upon disease of the middle ear and unassociated with thrombosis of the lateral sinus. Br Med J 1:1480–1481, 1890. 2. Ballance CA: On the removal of pyaemic thrombi from the lateral sinus. Lancet 1:1057–1061, 1890. 3. Dandy W: An operation for the cure of tic douloureux: Partial section of the sensory root at the pons. Arch Surg 18:687–734, 1929. 4. Dandy W: Surgery of the Brain. Hagerstown, WF Prior, 1945. 5. Lang J: Skull Base and Related Structures: Atlas of Clinical Anatomy, Stuttgart, Schattauer, 2001, ed 2.


ANATOMY Complication Avoidance Alireza Gharabaghi, M.D. Department of Neurosurgery, International Neuroscience Institute, Hannover, Germany, and Department of Neurosurgery, Eberhard Karls University Hospital, Tübingen, Germany

IMAGE-GUIDED LATERAL SUBOCCIPITAL APPROACH: PART 2—IMPACT ON COMPLICATION RATES AND OPERATION TIMES

Steffen K. Rosahl, M.D., Ph.D. Department of Neurosurgery, Albert Ludwigs University Hospital, Freiburg, Germany

Günther C. Feigl, M.D. Department of Neurosurgery, International Neuroscience Institute, Hannover, Germany

Sam Safavi-Abbasi, M.D. Department of Neurosurgery, International Neuroscience Institute, Hannover, Germany

Javad M. Mirzayan, M.D. Department of Neurosurgery, International Neuroscience Institute, Hannover, Germany

Stefan Heckl, M.D. Department of Neuroradiology, Eberhard Karls University Hospital, Tübingen, Germany

Ramin Shahidi, Ph.D. Image Guidance Laboratories, Stanford University, Palo Alto, California

Marcos Tatagiba, M.D., Ph.D. Department of Neurosurgery, Eberhard Karls University Hospital, Tübingen, Germany

Madjid Samii, M.D., Ph.D. Department of Neurosurgery, International Neuroscience Institute, Hannover, Germany Reprint requests: Alireza Gharabaghi, M.D., Department of Neurosurgery, University Hospital Tübingen, Charlottenstraße 37/1, 72070 Tübingen, Germany. Email: [email protected] Received, February 25, 2006. Accepted, June 22, 2006.

OBJECTIVE: Image-guidance systems are widely available for surgical planning and intraoperative navigation. Recently, three-dimensional volumetric image rendering technology that increasingly applies in navigation systems to assist neurosurgical planning, e.g., for cranial base approaches. However, there is no systematic clinical study available that focuses on the impact of this image-guidance technology on outcome parameters in suboccipital craniotomies. METHODS: A total of 200 patients with pathologies located in the cerebellopontine angle were reviewed, 100 of whom underwent volumetric neuronavigation and 100 of whom underwent treatment without intraoperative image guidance. This retrospective study analyzed the impact of image guidance on complication rates (venous sinus injury, venous air embolism, postoperative morbidity caused by venous air embolism) and operation times for the lateral suboccipital craniotomies performed with the patient in the semi-sitting position. RESULT: This study demonstrated a 4% incidence of injury to the transverse-sigmoid sinus complex in the image-guided group compared with a 15% incidence in the nonimage-guided group. Venous air embolisms were detected in 8% of the image-guided patients and in 19% of the non-image-guided patients. These differences in terms of complication rates were significant for both venous sinus injury and venous air embolism (P ⬍ 0.05). There was no difference in postoperative morbidity secondary to venous air embolism between both groups. The mean time for craniotomy was 21 minutes in the image-guided group and 39 minutes in non-image-guided group (P ⫽ 0.036). CONCLUSION: Volumetric image guidance provides fast and reliable three-dimensional visualization of sinus anatomy in the posterior fossa, thereby significantly increasing speed and safety in lateral suboccipital approaches. KEY WORDS: Complications, Operation time, Sigmoid sinus, Suboccipital approach, Three-dimensional volumetric image guidance, Transverse sinus, Venous air embolism Neurosurgery 62[ONS Suppl 1]:ONS24–ONS29, 2008

T

he lateral suboccipital approach is a common surgical route to pathologies in the cerebellopontine angle. Its boundaries and most prominent landmarks, the transverse and sigmoid sinus, are hidden behind the cranial bone. Although delicate and fragile, clear identification of the border of the sinuses is crucial for unrestrained exposure in this approach (40, 41, 44). Dural adhesions and emissary veins add to the challenges in dissection and preservation of these vital structures. Injury to the venous sinuses increases the risk of intraoper-


DOI: 10.1227/01.NEU.0000249233.12860.EC

ative venous air embolism (VAE) and potential postoperative morbidity, especially when the operation is performed with the patient in the semi-sitting position (4, 12, 15, 43). With an image-guidance system prepared for volumetrically rendered radiographical images to visualize the sinuses of the posterior fossa three-dimensionally, basing the approach upon the patient’s individual anatomy should be considerably facilitated. However, there is no systematic clinical study available that focuses on the influence of this image-guidance technology on outcome para-


IMAGE-GUIDED LATERAL SUBOCCIPITAL APPROACH

meters. Therefore, we investigated the impact of our navigated approach on complication rates for intraoperative VAE and related postoperative morbidity, as well as on operation time for lateral suboccipital craniotomy in the semi-sitting position.

METHODS Patient Population Two hundred consecutive patients with pathologies involving the cerebellopontine angle were studied retrospectively. The data collection started in November 2005 and included, backward in a consecutive manner, all operations performed before this assigned time. As soon as the data for 100 cases were collected for one of the two groups (image guided versus non-image guided), additional data were collected only for the other group until the number of 100 cases was reached as well. We investigated pre- and postoperative medical information, surgical and clinical reports, anesthesiology and intensive care unit protocols, as well as navigation protocols and surgical video recordings of these cases.

FIGURE 1. The location and course of the transversesigmoid sinus transition in relation to the asterion is drawn on the patient’s skin for the skin incision. The information about these landmarks is obtained by fine-tuned opacity modulation of volumetric threedimensional images and is then transferred to the patient following the probe of the guidance system.

Data Collection The following were included as preoperative variables: patient sex and age, site of the lesion, tumor pathology, as well as years of training of the surgeon who performed the craniotomy. Intraoperative findings that were recorded were use of the navigation system, damage to the transverse and/or sigmoid sinuses and/or to major emissary veins, and air embolism. We also recorded the time necessary to complete the craniotomy, which was defined as starting with the skin incision and ending when the dura mater was opened. In cases in which navigation was used, the time necessary for image rendering and processing was recorded, as was the setup and registration time for the navigation system in the operating room. The indication to use a navigation system was independent of the surgeon or of patient-specific findings. Most of the time, the system was used upon availability. All craniotomies were performed by a neurosurgeon in training under the supervision of a senior neurosurgeon. The surgeons who performed craniotomies were well trained to use the navigation system for preoperative surgical planning and intraoperative guidance. All participating surgeons had performed at least 10 suboccipital craniotomies with and without navigation before participating in this study.

Image Acquisition A computed tomographic (CT) scan (1-mm slice thickness) with additional delayed contrast injection for visualization of the venous system (venous CT angiography [vCTA]) was obtained in all patients. In the image-guided group, seven to nine adhesive skin fiducial markers were placed on the forehead, at the mastoid tips, and in the retroauricular area of each patient before imaging. Software designed for surgical imageguidance (Image Guidance Laboratories, Palo Alto, CA) was used to volumetrically render three-dimensional (3-D) images in which skin, bone, and intracranial structures could be visualized at the same time by opacity modulation. The spatial course of the transverse-sigmoid sinus complex was recorded while varying the perspective of the 3-D model interactively and modulating the translucency of surfaces.

Neuronavigation The patients were brought into the semi-sitting position for surgery. The digital reference frame faced the infrared camera that was mounted on the side of the surgery. The target registration error was

NEUROSURGERY

estimated by comparing the position of the navigation probe on a nonregistered fiducial or well defined bony landmark with the localization of the corresponding site in the reconstructed triaxial images. According to the 3-D renderings, the transverse and sigmoid sinus, the bone sutures, and the position of the mastoid sulcus were drawn onto the patient’s skin for a skin incision (Fig. 1). After soft tissue preparation, image guidance was used to reevaluate the course of the sinuses. In cases in which an orthoclastic craniotomy was performed, the gap was refilled with methylmethacrylate at the end of surgery. In osteoplastic procedures, the first burr hole was placed close to the transverse-sigmoid transition with a safety margin of a few millimeters. After rough bone preparation, the accuracy of the system was always reevaluated to exclude errors from digital reference frame dislocation or shifting of the patient’s head before the more delicate dissection of emissary veins and sinus edges was accomplished (Fig. 2). Microsurgical tumor removal was performed by the senior authors (MS, MT) under electrophysiological neuromonitoring of auditory and facial nerve function.

Anesthesiological Aspects and Monitoring Intraoperative monitoring included electrocardiography and continuous measurement of the arterial blood pressure and superior vena cava pressure. The tip of the central venous catheter was located within the right atrium so as to aspirate any invaded air during embolism. Furthermore, continuous measurement of the end-expiratory CO2 concentration, body temperature, diuresis, and the arterial oxygen saturation (oximetry) were performed. To avoid clinical consequences of air embolism by early recognition, precordial Doppler ultrasonography was initiated because of its high sensitivity. The occurrence of microbubbles in the Doppler and/or a sudden and sustained decrease of the end-expiratory CO2 concentration were defined as VAE. As soon as these signs occurred, the surgical field was irrigated, the jugular veins were compressed, air was aspirated by means of the atrial catheter, the surgical table was inclined cranialward, and the end-expiratory pressure was increased.

Statistical Analysis Statistical analysis was performed using SigmaStat 2.0 software (SPSS, Chicago, IL). T- and χ2 tests were used to compare two inde-


GHARABAGHI ET AL.

was no damage to these venous structures related to drilling. The TST and the edges of the sinuses were precisely exposed in all patients.

Patient Population The mean age for all patients was 47 years (range, 11–78 yr). No significant difference in age was observed between the image-guided group (48 yr; range, 11–78 yr) and the non-imageguided group (46 yr; range, 20–72 yr). There was also no preference of the tumor site in either the image-guided group or the non-image-guided group, with 43 and 47% tumors on the right side, respectively. Female patients comprised 48% of the imageguided group and 54% of the non-image-guided group. The tumor pathologies were similar in both groups, with 88 vestibular schwannomas, 10 meningiomas, and two epidermoids in the image-guided group and 95 vestibular schwannomas and five meningiomas in the non-image-guided group.

Complications

FIGURE 2. The burr hole is placed at the level of the transverse sigmoid transition with a safety margin of approximately 5 mm in this case. After an osteoplastic craniotomy is performed, the accuracy of the system is reevaluated to exclude shifting errors before the delicate sinus edges are dissected.

pendent populations, including their parametric proportions. The level of significance (P) was set at a probability value of less than 0.05.

RESULTS Image Guidance 3-D volumetric image rendering proved to be a reliable method to simultaneously visualize bony and venous structures in each individual patient. In all cases, the intraoperative findings matched the spatial relationships in the 3-D reconstructions. With the location of the landmark structures drawn on the skin, the incision and the craniotomy could be customtailored to each patient’s individual anatomy (Figs. 1 and 2). The real-time display of the transverse and sigmoid sinus complex by the guidance system matched the operative sites with an estimated precision of less than ⫾2 mm in accuracy maps obtained after physical registration. The mean target registration error amounted to 1.4 mm (⫾0.5 mm) for CT scans in all navigated cases. When the burr hole was placed directly below the transverse sinus close to the transverse sigmoid transition (TST), there


Injury to the transverse-sigmoid sinus complex occurred in 4% of the patients in the image-guided group and in 15% of the patients in the non-image-guided group. In all of these cases, sinus injury was followed by VAE. The overall incidence of VAE was 8% in the image-guided group and 19% in the non-imageguided group. These differences in terms of complication rates were significant for both venous sinus injury and VAE (χ2, 7.02; degrees of freedom, 2; P ⬍ 0.05). In all cases, the invaded air during embolism could be recognized by precordial Doppler ultrasonography and aspirated via the central venous catheter, which was located within the right atrium. Nonetheless, VAE caused intraoperative hypotension in one patient in the image-guided group and in two patients in the non-image-guided group. However, these findings were transient and did not lead to postoperative VAErelated morbidity such as pulmonary edema or paradoxical air embolus.

Craniotomy and Image Guidance The surgeons performing the craniotomy had a mean neurosurgical training experience of 3 years (range, 2–4 yr). There was no significant difference between the image-guided (mean, 2.9 yr; range, 2–4 yr) and the non-image-guided groups (mean, 3.1 yr; range, 2–4 yr). The mean time for craniotomy was 21 minutes (range, 14–36 min) in the image-guided group and 39 minutes (range, 19–52 min) in the non-image-guided group. This difference in the mean values of the two groups was significant (t, ⫺2.513; degrees of freedom, 8; P ⫽ 0.036). Preoperatively, there was a mean time of 11 minutes (range, 7–21 min) necessary for image rendering and processing. The mean setup and registration time for the navigation system in the operating room was 14 minutes (range, 10–25 min).

DISCUSSION The most common neurosurgical approaches for lesions in the cerebellopontine angle are based on lateral suboccipital or



retrosigmoid craniotomies (9, 26, 30, 36, 40, 57). With the transverse and sigmoid sinuses, the boundaries of this approach are anatomically well defined (27, 39, 44). As these venous structures are carved inside and hidden behind the cranial bone, superficial bony landmarks have traditionally been used for orientation and localization (10, 16, 24, 28, 37, 44, 51). More recently, cadaveric studies have shown a large variety of superficial landmarks in relation to the transverse and sigmoid sinuses, such as the insertion of the semispinalis capitis muscle, the superior nuchal line, the inion, and the mastoid process, all of which may be applied for localization and protection of the TST (11, 28, 31, 37, 38, 39, 46, 47, 50, 51, 52). The major drawback of using normative data sampled from anatomic dissections is that, lacking precise information on the individual case at hand, the anatomy of the patient can only be estimated. Consequently, neurosurgeons usually work their way to the sinus starting from a distant, but all the more safe, burr hole, never being precisely sure when they will encounter the fragile border of the sinus before actually seeing it. This observation helps to explain one of the findings of our study that the craniotomy took longer in the group of cases performed without image guidance (39 min versus 21 min in the imageguided group). To overcome these shortcomings and to improve approaches to the cranial base, neuronavigation has been evaluated in experimental studies (7, 42, 54), and 3-D CTA has been introduced to surgical planning for cranial base operations (45, 53, 54). With the advent of the 3-D volume rendering technique, these issues can be addressed with the option of gradually changing the opacity of the outer layers, such as the cranial bone, to visualize intracranial structures such as the vascular anatomy (8, 13, 19, 20, 23, 25, 32, 33, 35, 49, 53, 56, 57). Alternatively, the vascular anatomy, e.g., the sinuses themselves, may be turned into landmarks projected onto the surface of the cranium by intraoperative image guidance. This would allow for the changing of image modalities from a CT angiogram to a magnetic resonance angiogram excluding x-ray exposure. Although image-based, stereoscopic virtual reality models are already in use for planning specific surgical procedures and for teaching purposes in neurosurgery (2, 3, 9, 12, 13), most studies that evaluate preoperative neurosurgical planning have focused on 3-D visualization capabilities of the applied software (1, 17, 18, 21, 22, 32, 50, 60). However, the ability to modulate the opacity of the cranial bone to visualize intracranial structures and to use this information intraoperatively has not been explored for its surgical impact during lateral suboccipital approaches. More specifically, there is no other report on the impact of volumetric image guidance on complications during craniotomies in the posterior fossa. In the present study, we identified a significant reduction of intraoperative complications commonly associated with suboccipital craniotomies performed in the semi-sitting position. Both injury to the transverse sigmoid sinus complex and VAE occurred less often in the image-guided group, with incidences of 4 and 8% of the cases, respectively, than in the non-imageguided group, which has incidences of 15 and 19% of the cases,

NEUROSURGERY

respectively. However, all of these complications could be controlled intraoperatively by standard surgical and anesthesiological techniques (e.g., irrigation of the surgical field, closure of the site of air invasion under compression of jugular veins, aspiration of air by means of the atrial catheter) and did not lead to long-term morbidity for the patients in the present series. The rate of VAE in the non-image-guided group of our study (19%) was in the range of reported incidences in the literature, which ranged from 7 to 45% of patients (4, 5, 6, 12, 14, 15, 29, 34, 43, 48, 59). The sites of probable air entrainment are known to be variable. Embolism can occur during surgical procedures and manipulations by invading air entering the open veins of neck muscles or the bone, or the opened venous sinuses or emissary or bridging veins (4, 15). Most of the VAE in this series occurred after transverse or sigmoid sinus opening in four out of eight patients in the image-guided group and in 15 out of 19 patients in the non-image-guided group. Therefore, the reduction of unintended venous sinus opening from 15% in the non-image-guided group to 4% in the image-guided group was paralleled by a decline of VAE from 19 to 8%. Owing to the intraoperative management protocol for VAE, postoperative VAE-related morbidity did not occur in any of the patients. Therefore, the impact of image guidance on intraoperative complications did not affect postoperative morbidity. The mean time for craniotomy was 21 minutes in the image guided group and 39 minutes in the non-image-guided group. Nonetheless, this time-saving impact of volumetric imageguidance for suboccipital craniotomy in our series has to be seen in light of the time necessary to prepare the navigation supported procedure. A mean period of 11 minutes was necessary for image rendering and processing before the operation. In the operating room, an additional mean time of 14 minutes was spent to set up the navigation system and register the patient. Whereas the image preparation time was spent before surgery outside the operating room, the system setup was performed parallel to other surgical preparations and, therefore, did not prolong the overall operation time in most of the cases. However, these additional preparations must be considered when evaluating the impact of image guidance on operation time. The mean time gain of 18 minutes achieved by image guidance was neutralized by the time spent for these preparations in our study. Until now, this setup has been done by the neurosurgeons themselves. When economic aspects and efficiency considerations are taken into account, one should consider the involvement of other staff members, such as neuroradiologists or technicians, in this preparation process.

CONCLUSION The real-time images provided by 3-D volumetric image rendering closely match individual patient anatomy. In the lateral suboccipital approach, the advantage of being able to see beyond tissue barriers before actually dissecting them speeds up the surgical procedure and provides additional safety for the patient with ready and accurate localization of the


GHARABAGHI ET AL.

transverse sigmoid sinus complex. In the future, these findings must be confirmed by prospective and randomized studies, especially in the era of evidence-based medicine.

REFERENCES 1. Abrahams JM, Saha PK, Hurst RW, LeRoux PD, Udupa JK: Threedimensional bone-free rendering of the cerebral circulation by use of computed tomographic angiography and fuzzy connectedness. Neurosurgery 51:264–269, 2002. 2. Balogh A, Preul MC, Schornak M, Hickman M, Spetzler RF: Intraoperative stereoscopic QuickTime Virtual Reality. J Neurosurg 100:591–596, 2004. 3. Bernardo A, Preul MC, Zabramski JM, Spetzler RF: A three-dimensional interactive virtual dissection (IVD) model to simulate transpetrous surgical avenues. Neurosurgery 52:499–505, 2003. 4. Bithal P, Dash HH, Vishnoi N, Chaturvedi A: Venous air embolism: Does the site of embolism influence the hemodynamic changes? Neurol India 51:370–372, 2003. 5. Bithal PK, Pandia MP, Dash HH, Chouhan RS, Mohanty B, Padhy N: Comparative incidence of venous air embolism and associated hypotension in adults and children operated for neurosurgery in the sitting position. Eur J Anaesthesiol 21:517–522, 2004. 6. Black S, Oeckert DB, Oliver WC Jr, Cucchiara RF: Outcome following posterior fossa craniectomy in patients in the sitting or horizontal positions. Anesthesiology 69:49–56, 1988. 7. Brinker T, Arango G, Kaminsky J, Samii A, Thorns U, Vorkapic P, Samii M: An experimental approach to image guided skull base surgery employing a microscope-based neuronavigation system. Acta Neurochir (Wien) 140:883–889, 1998. 8. Villavicencio AT, Gray L, Leveque JC, Fukushima T, Kureshi S, Friedman AH: Utility of three-dimensional computed tomographic angiography for assessment of relationships between the vertebrobasilar system and the cranial base. Neurosurgery 48:327, 2001. 9. Dandy WE: Results of removal of acoustic tumors by the unilateral approach. Arch Surg 42:1026–1033, 1941. 10. Day JD, Kellogg JX, Tschabitscher M, Fukushima T: Surface and superficial surgical anatomy of the posterolateral cranial base: Significance for surgical planning and approach. Neurosurgery 38:1079–1084, 1996. 11. Day JD, Tschabitscher M: Anatomic position of the asterion. Neurosurgery 42:198–199, 1998. 12. Duke DA, Lynch JJ, Harner SG, Faust RJ, Ebersold MJ: Venous air embolism in sitting and supine patients undergoing vestibular schwannoma resection. Neurosurgery 42:1282–1287, 1998. 13. Endo T, Suzuki Y, Ikeda H, Jinbo H, Ikeda Y, Matsumoto K: Evaluation of venous system with 3D-CTA-from the viewpoint of treatment to cerebral aneurysm [in Japanese]. No Shinkei Geka 29:1167–1174, 2001. 14. Giebler R, Kollenberg B, Pohlen G, Peters J: Effect of positive end-expiratory pressure on the incidence of venous air embolism and on the cardiovascular response to the sitting position during neurosurgery. Br J Anaesth 80:30–35, 1998. 15. Gösseln H, Suhr D: What a neurosurgeon should know about anesthesia for posterior fossa tumors, in Raimondi AJ, Choux M, Di Rocco C (eds): Posterior Fossa Tumors. New York, Springer, 1993, pp 86–105. 16. Hakuba A, Nishimura S, Jang BJ: A combined retroauricular and preauricular transpetrosaltranstentorial approach to clivus meningiomas. Surg Neurol 30:108–116, 1988. 17. Hayashi N, Endo S, Shibata T, Ikeda H, Takaku A: Neurosurgical simulation and navigation with three-dimensional computer graphics. Neurol Res 21:60–66, 1999. 18. Henn JS, Lemole GM Jr, Ferreira MA, Gonzalez LF, Schornak M, Preul MC, Spetzler RF: Interactive stereoscopic virtual reality: A new tool for neurosurgical education. Technical note. J Neurosurg 96:144–149, 2002. 19. Hoell T, Oltmanns F, Schilling A, Schernick A, Brock M: The value of 3D magnetic resonance imaging in the planning of neurosurgery. Front Radiat Ther Oncol 33:88–93, 1999. 20. Kaminogo M, Hayashi H, Ishimaru H, Morikawa M, Kitagawa N, Matsuo Y, Hayashi K, Yoshioka T, Shibata S: Depicting cerebral veins by three-


21.

22.

23.

24.

25.

26. 27. 28. 29.

30. 31. 32.

33.

34. 35.

36.

37.

38.

39. 40. 41.

42.

43.

dimensional CT angiography before surgical clipping of aneurysms. Am J Neuroradiol 23:85–91, 2002. Kockro RA, Serra L, Tsai YT, Chan C, Sitoh YY, Chua GG, Hern N, Lee E, Hoe LY, Nowinski W: Planning of skull base surgery in the virtual workbench: Clinical experiences. Stud Health Technol Inform 62:187–188, 1999. Kockro RA, Serra L, Tseng-Tsai Y, Chan C, Yih-Yian S, Gim-Guan C, Lee E, Hoe LY, Hern N, Nowinski WL: Planning and simulation of neurosurgery in a virtual reality environment. Neurosurgery 46:118–137, 2000. Kikinis R, Gleason PL, Moriarty TM, Moore MR, Alexander E 3rd, Stieg PE, Matsumae M, Lorensen WE, Cline HE, Black PM, Jolesz FA: Computerassisted interactive three-dimensional planning for neurosurgical procedures. Neurosurgery 38:640–651, 1996. Koos WT, Perneczky A: Suboccipital approach to acoustic neurinomas with emphasis on preservation of facial nerve and cochlear nerve function, in Rand RW (ed): Microneurosurgery. St. Louis, Mosby, 1985, ed 3, pp 335–365. Korogi Y, Takahashi M, Katada K, Ogura Y, Hasuo K, Ochi M, Utsunomiya H, Abe T, Imakita S: Intracranial aneurysms: detection with three-dimensional CT angiography with volume rendering-comparison with conventional angiographic and surgical findings. Radiology 211:497–506, 1999. Krause F: The exposure of the posterior petrous bone and of the cerebellum [in German]. Beitr Klin Chir 37:728–764, 1903 Lang J: Clinical Anatomy of the Posterior Cranial Fossa and its Foramina. New York, Thieme, 1991. Lang J Jr, Samii A: Retrosigmoid approach to the posterior cranial fossa. An anatomical study. Acta Neurochir (Wien) 111:147–153, 1991. Lodrini S, Montolivo M, Pluchino F, Borroni V: Positive end-expiratory pressure in supine and sitting positions: Its effect on intrathoracic and intracranial pressures. Neurosurgery 24:873–877, 1989. Malis LI: Microsurgical treatment of acoustic neurinomas, in Handa H, Kikuchi H, Yonekawa Y (eds): Microsurgery. Tokyo, Igako Shoin, 1975. Malis LI: Anatomical position of the asterion. Neurosurgery 42:198–199, 1998 (comment). Mamata H, Komiya T, Muro I, Matsuyama S: Application and validation of three-dimensional data sets from a phase contrast MR angiography for preoperative computer simulation of brain tumors. J Magn Reson Imaging 10:102–106, 1999. Marro B, Valery CA, Bitard A, Sahel M, Zouaoui A, Randoux B, Oppenheim C, Marsault C: Intracranial aneurysm on CTA: Demonstration using a transparency volume-rendering technique. J Comput Assist Tomogr 24:96–98, 2000. Matjasko J, Petrozza P, Cohen M, Steinberg P: Anesthesia and surgery in the seated position: Analysis of 554 cases. Neurosurgery 17:695–702, 1985. Nishihara M, Tamaki N: Usefulness of volume-rendered three-dimensional computed tomographic angiography for surgical planning in treating unruptured paraclinoid internal carotid artery aneurysms. Kobe J Med Sci 47:221–230, 2001. Rand RW, Kurze T: Micro-neurosurgical resection of acoustic tumours by a transmeatal posterior fossa approach. Bull Los Angeles Neuro Soc 30:17–20, 1965. Rhoton AL Jr: Surface and superficial surgical anatomy of the posterolateral cranial base: Significance for surgical planning and approach. Neurosurgery 38:1083–1084, 1996 (comment). Ribas GC, Rhoton AL Jr, Cruz OR, Peace D: Temporo-parieto-occipital burrhole sites study and systematized approaches proposal, in Samii M (ed): Skull Base Surgery. Basel, Karger, 1994, pp 723–730. Samii M, Draf W, Lang J: Surgery of the Skull Base. Berlin, Springer, 1989, pp 361–362. Samii M, Draf W, Lang J: Surgery of the Skull Base. Berlin, Springer, 1989, pp 385–387. Samii M, Matthies C: Management of 1000 vestibular schwannomas (acoustic neuromas): Surgical management and results with an emphasis on complications and how to avoid them. Neurosurgery 40:11–23, 1997. Samii A, Brinker T, Kaminsky J, Lanksch WR, Samii M: Navigation-guided opening of the internal auditory canal via the retrosigmoid route for acoustic neuroma surgery: Cadaveric, radiological, and preliminary clinical study. Neurosurgery 47:382–388, 2000. Schwarz G, Fuchs G, Weihs W, Tritthart H, Schalk HV, Kaltenbock F: Sitting position for neurosurgery: Experience with preoperative contrast echocardiography in 301 patients. J Neurosurg Anesthesiol 6:83–88, 1994.



44. Seeger W: Planning Strategies of Intracranial Microsurgery. Wien, SpringerVerlag, 1986. 45. Sekhar LN, Kalia KK, Yonas H, Wright DC, Ching H: Cranial base approches to intracranial aneurysms in the subaarachnoidal space. Neurosurgery 35: 472–483, 1994. 46. Sekhar LN: Anatomical position of the asterion. Neurosurgery 42:198–199, 1998 (comment). 47. Sripairojkul B, Adultrakoon A: Anatomical position of the asterion and its underlying structure. J Med Assoc Thai 83:1112–1115, 2000. 48. Standefer M, Bay JM, Trusso R: The sitting position in neurosurgery: A retrospective analysis of 488 cases. Neurosurgery 14:649–658, 1984. 49. Suzuki Y, Ikeda H, Shimadu M, Ikeda Y, Matsumoto K: Variations of the basal vein: Identification using three-dimensional CT angiography. Am J Neuroradiol 22:670–676, 2001. 50. Tandler J, Ranzi E: Surgical anatomy and operative technique of the central nervous system [in German]. Berlin, Springer-Verlag, 1920, pp 49–53. 51. Tubbs RS, Salter G, Oakes WJ: Superficial surgical landmarks for the transverse sinus and torcular herophili. J Neurosurg 93:279–281, 2000. 52. Uz A, Ugur HC, Tekdemir I: Is the asterion a reliable landmark for the lateral approach to posterior fossa? J Clin Neurosci 8:146–147, 2001. 53. Villavicencio AT, Leveque JC, Bulsara KR, Friedman AH, Gray L: Threedimensional computed tomographic cranial base measurements for improvement of surgical approaches to the petrous carotid artery and apex regions. Neurosurgery 49:342–353, 2001. 54. Vrionis FD, Foley KT, Robertson JH, Shea JJ 3rd: Use of cranial surface anatomic fiducials for interactive image-guided navigation in the temporal bone: A cadaveric study. Neurosurgery 40:755–764, 1997. 55. Wetzel SG, Kirsch E, Stock KW, Kolbe M, Kaim A, Radue EW: Cerebral veins: Comparative study of CT venography with intraarterial digital subtraction angiography. AJNR Am J Neuroradiol 20:249–255, 1999. 56. Wintermark M, Uske A, Chalaron M, Regli L, Maeder P, Meuli R, Schnyder P, Binaghi S: Multislice computerized tomography angiography in the evaluation of intracranial aneurysms: A comparison with intraarterial digital subtraction angiography. J Neurosurg 98:828–836, 2003. 57. Yas¸ argil MG: Microsurgery of cerebellopontine angle tumors, in Plester D, Wende S, Makayama N (eds): Cerebellopontine. Berlin, Springer, 1978, pp 215–257. 58. Young ML, Smith DS, Murtagh F, Vasquez A, Levitt J: Comparison of surgical and anesthetic complications in neurosurgical patients experiencing venous air embolism in the sitting position. Neurosurgery 18:157–161, 1986. 59. Zhao JC, Chen C, Rosenblatt SS, Meyer JR, Edelman RR, Batjer HH, Ciric IS: Imaging the cerebrovascular tree in the cadaveric head for planning surgical strategy. Neurosurgery 51:1222–1228, 2002.

COMMENTS

T

his is a timely analysis of patients who underwent operations either with or without the aid of navigation. Gharabaghi et al. have pro-

NEUROSURGERY

vided a well conducted and extremely painstaking comparison of 100 navigated and 100 non-navigated suboccipital craniotomies in an attempt to provide scientific evidence to justify the use of navigation. As expected, surgery seemed swifter and safer with the aid of navigation, although the actual navigation procedure consumed some of the time that was saved. The authors are to be commended for accepting the difficult and unrewarding task of providing clinical evidence of the efficacy of this navigation technology. Still, I must raise a critical issue. Evidence-based medicine is fashionable, and it is common to demand empirical data for any clinical statement. Any systematic report becomes superior to common sense or “expert opinion.” The findings in this study correlate very well with common sense, but how would we have interpreted the data if the findings had disagreed? It is important to ask whether the investigators actually provided a “critical experiment,” one which, in the words of Karl Popper, would have had the power to corroborate or falsify their implicit hypothesis. Medical science has repeatedly gone astray as analyses have been much more biased than anyone realizes. I fear that this is a typical case in which a retrospective design risks major bias. Thus, I would not accept that the hypothesis has been corroborated in a strictly scientific sense. On the other hand, a prospective randomized trial would be difficult to conduct, and a retrospective trial can hardly be better conducted. This study will serve the medical community as an argument to accept increased costs to improve surgical safety. Tiit Mathiesen Stockholm, Sweden

O

n the basis of the author’s initial publication, which demonstrated the anatomic variability of the asterion in relation to underlying risk structures, the authors now retrospectively evaluate the impact of neuronavigation on the outcome of suboccipital craniotomies. To date the neurosurgical community has been struggling to objectively estimate the value of neuronavigation, even though most of us rely increasingly on this technology in daily practice. Despite the fact that Gharabaghi et al. present retrospective data, their report adds a precious piece to a yet incomplete mosaic of information portraying neuronavigation as a priceless tool, which eventually should encourage hospital administration to invest continuously into advanced surgical technologies, such as navigation and intraoperative imaging. Thomas Gasser Volker Seifert Frankfurt, Germany


ANATOMY Surgical Approach

ENDOSCOPIC SUBLABIAL TRANSMAXILLARY APPROACH TO THE ROSTRAL MIDDLE FOSSA Bonnie C. Ong, H.S.C. University of New South Wales, School of Medical Sciences, Sydney, Australia

Pankaj A. Gore, M.D. Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Michael B. Donnellan, M.B.B.S. Centre for Minimally Invasive Neurosurgery, Prince of Wales Private Hospital, Sydney, Australia

Thomas Kertesz, M.B.B.S. Centre for Minimally Invasive Neurosurgery, Prince of Wales Private Hospital, Sydney, Australia

Charles Teo, M.B.B.S. Centre for Minimally Invasive Neurosurgery, Prince of Wales Private Hospital, Sydney, Australia Reprint requests: Pankaj A. Gore, M.D., 1256 East Sheena Drive, Phoenix, AZ 85022. Email: [email protected] Received, Feburary 24, 2007. Accepted, July 10, 2007.

OBJECTIVE: The rostral middle fossa faces the temporal pole and is the endocranial anterosuperior aspect of the greater wing of the sphenoid. Standard approaches to this region, such as the subtemporal, pterional, or orbitozygomatic approaches, require significant brain retraction or manipulation of the temporalis muscle. We report an endoscopic sublabial transmaxillary approach to this cranial base region that avoids the aforementioned pitfalls. METHODS: Ten adult cadaveric half heads were used to develop the endoscopic approach and to identify the salient surgical landmarks. RESULTS: The approach was divided into three stages: entry into the maxillary sinus, entry into the infratemporal fossa, and entry into the middle fossa. A craniotomy of greater than 20 mm in diameter can be safely created in the rostral middle fossa. When coupled with image guidance, the approach provides the flexibility to tailor the size and location of the middle fossa craniotomy. CONCLUSION: Although endonasal endoscopic approaches are increasing in popularity, the middle fossa has not been adequately accessed with these techniques. The endoscopic sublabial transmaxillary approach provides safe and direct access to the rostral middle fossa, eliminating the need for brain retraction, temporalis muscle manipulation, or an external incision. The approach also permits early devascularization of cranial- or dural-based lesions. KEY WORDS: Anatomy, Cranial base, Endoscopy, Middle fossa, Skull base, Transmaxillary Neurosurgery 62:30–37, 2008

T

DOI: 10.1227/01.NEU.0000297050.89658.61

he temporal pole faces the endocranial anterosuperior aspect of the greater wing of the sphenoid. This region of the cranial base, the rostral middle fossa, is traditionally accessed by a subtemporal approach, a standard pterional approach, or an orbitozygomatic approach. None of these approaches are ideal for small lesions of the rostral middle fossa. The subtemporal approach requires significant brain retraction. The pterional approach and its orbitozygomatic extension approaches permit access with less brain retraction but require elevation of the temporalis muscle with the inevitable potential for atrophy and suboptimal cosmesis. There are alternative approaches to the middle fossa that traverse the nasopharynx and/or the maxillary sinus. Couldwell et al. (5), Sabit et al. (21), and Roche et al. (20) have all described microsurgical approaches, either through sublabial or transfacial incisions,



through the anterior wall of the maxillary sinus to access the middle fossa and surrounding compartments such as the pterygopalatine fossa and infratemporal fossa. Alfieri et al. (1) and Kassam et al. (11) have described an endoscopic endonasal approach to the cavernous sinus, foramen rotundum, and middle fossa dura. We have developed an endoscopic sublabial transmaxillary approach (ESTA) that takes advantage of the anatomically favorable trajectory through the maxillary sinus and anterior infratemporal fossa into the rostral middle fossa. In comparison with the pterional, orbitozygomatic, and subtemporal approaches, the ESTA requires no brain retraction and no temporalis manipulation. Compared with the microsurgical approaches through the maxillary sinus, the ESTA is far less invasive and provides improved visualization at the depth of field. Compared with the endoscopic endonasal approach, the ESTA is technically


ENDOSCOPIC APPROACH TO MIDDLE FOSSA

easier, requires less specialized instrumentation, and provides better access to the lateral rostral middle fossa.

A

B

MATERIALS AND METHODS Ten adult cadaveric half-head specimens were studied, eight of which were fresh-frozen and two of which were embalmed. Four of these were injected with red and blue silicone for visualization of the arterial and venous systems, respectively. Perneczky rod-lens endoscopes (Aesculap AG, Tuttlingen, Germany), 4 mm in diameter, 18 cm in length, with 0- and 30-degree lenses, were used for visualization of the surgical field. Digital still images were obtained through the use of a video-capture system coupled to a three-chip, charge-coupled device camera. A high-speed drill with an M4 shaver handpiece, straight guarded burrs, and angled diamond bit burrs was used for endoscopecontrolled bone work (Medtronic, Jacksonville, FL).

RESULTS The endoscopic sublabial transmaxillary approach to the rostral middle fossa can be considered in three stages: 1) entry into the maxillary sinus, 2) entry into the infratemporal fossa, and 3) entry into the middle cranial fossa.

Stage 1: Entry into the Maxillary Sinus Entry into the maxillary sinus was accomplished using the Caldwell-Luc method. An ipsilateral horizontal incision was made at the junction of the oral mucosa and gingiva, immediately lateral to the canine and extending 15 mm laterally. The upper lip was retracted, and the canine jugum on the anterior surface of the maxilla was identified underlying the medial extent of the incision. Immediately lateral to the most superior aspect of the jugum, a maxillotomy was created with a 4mm osteotome and expanded FIGURE 1. Photograph showing with a 2-mm, 40-degree Kerthe human cranium. The shaded rison rongeur (Codman/Johnarea indicates the placement and son & Johnson, Raynham, approximate size of anterior maxilloMA) to a defect of 15 mm hortomy. IOF, infraorbital foramen; C, izontally and 10 mm vertically canine; AM, anterior maxillotomy. (Fig. 1).

Stage 2: Entry into the Infratemporal Fossa The maxillary sinus is pyramidal in shape, with its apex facing laterally and its base or medial wall adjacent to the nasal cavity. It also consists of anterior, posterior, and lateral walls, as well as a floor. On entry into the maxillary sinus, the mucosa was removed, particularly from the roof, posterior wall, and lateral wall. The infraorbital nerve was observed emerging from the junction of the roof, posterior wall, and lateral wall of the maxillary sinus to course along the roof (Fig. 2).

NEUROSURGERY

FIGURE 2. A, endoscopic view showing the maxillary sinus interior. A 30-degree endoscope was directed cranially. The mucosa has been removed. B, schematic illustration of A showing the infraorbital nerve (IN) emerging at the junction between the roof (R) and posterior wall (PW) of the maxillary sinus and coursing along the roof. The middle superior alveolar nerve (MSAN), responsible for molar innervation, is inconsistently present and is also variable in its site of origin from the infraorbital nerve. OS, maxillary os; LW, lateral wall; MW, medial wall.

The retromaxillary space is composed of the pterygopalatine fossa, located medially, and the infratemporal fossa, located laterally. The infraorbital nerve is the critical landmark in determining the position of the posterior maxillotomy. The origin of the infraorbital nerve indicates the location of the pterygomaxillary fissure, through which the pterygopalatine and infratemporal fossae communicate. In several early dissections, the infraorbital neurovascular bundle was unroofed and mobilized superomedially in its proximal aspect. This allowed access to the contents of the pterygopalatine fossa in addition to the infratemporal fossa. The added exposure of the pterygopalatine fossa was found to be unnecessary to accomplish the craniotomy of the rostral middle fossa. In later specimens, the infraorbital neurovascular bundle was retained within its bony canal. A limited posterior maxillotomy measuring approximately 10 mm in diameter was fashioned immediately inferior to the emergence of the infraorbital neurovascular bundle into the maxillary sinus. The high-speed drill or 4-mm osteotome and 40-degree Kerrison rongeur were used for this purpose. This opening is sufficient to access the rostral middle fossa and avoids the trigeminal maxillary division branches that populate the pterygopalatine fossa.

Stage 3: Entry into the Middle Cranial Fossa The periosteum and adipose and connective tissues underlying the posterior maxillotomy were bluntly dissected in a centrifugal fashion to reveal fibers of the lateral pterygoid muscle converging toward its heads (Fig. 3). The lateral pterygoid muscle consists of two heads: the upper arises from the lower part of the lateral plate of the greater sphenoidal wing and the infratemporal crest, and the lower arises from the lateral surface of the lateral pterygoid plate. The fibers of the upper head run posteroinferolaterally; those of the lower head run horizontally posterolaterally. Toward their point of insertion, the fibers


ONG ET AL.

A

B

FIGURE 3. A, endoscopic view through posterior maxillotomy demonstrating internal maxillary artery and lateral pterygoid muscle within infratemporal fossa. A 30-degree endoscope was directed cranially. B, schematic illustration of A showing the posterior maxillotomy fashioned just inferior and lateral to the infraorbital nerve. The posterosuperior alveolar nerves are responsible for molar innervation and can have a variable origin and course. The retromaxillary fat has been excluded to illustrate the underlying internal maxillary artery. The lateral pterygoid muscle underlies the internal maxillary artery. R, roof; IN, infraorbital nerve; MSAN, middle superior alveolar nerve; LPM, lateral pterygoid muscle; IMa, internal maxillary artery; LW, lateral wall; RMF, retromaxillary fat (partially dissected); P, periosteum. Asterisks, posterior superior alveolar nerve branches.

converge, with those of the upper head inserting primarily into the capsule and articular disc of the temporomandibular joint and those of the lower head inserting onto the pterygoid fovea on the medial aspect of the mandibular condyle. The two heads of the lateral pterygoid were often not discrete from one another. On their identification, the muscle was traced medially and superiorly to its origin on the lateral pterygoid plate and infratemporal surface of the greater wing of the sphenoid. The heads of the pterygoid muscle were elevated from their origin and displaced laterally. The anterior junction of the lateral pterygoid plate and the greater wing of the sphenoid were identified and served as the initial point of entry into the middle fossa (Fig. 4). There is some variation in the size, shape, and position of the lateral pterygoid plate with respect to the greater wing of the sphenoid. In the clinical setting, intraoperative stereotactic image guidance is invaluable in determining the exact location of this entry point. The craniotomy in the floor of the middle fossa was created with a 4-mm osteotome and expanded with a 40-degree Kerrison rongeur (Fig. 5). Alternatively, and more appropriate to the clinical setting, a high-speed drill with a 3.2-mm straight guarded burr was used to initiate the craniotomy, which was then expanded with a 2.9-mm, 15-degree diamond bit burr. A craniotomy with dimensions of 14 mm vertically, 20 mm horizontally, and 22 mm obliquely and located in the most anteromedial aspect of the greater wing of the sphenoid could easily be achieved (Fig. 6). If the endoscopic trajectory with respect to the infraorbital nerve was altered to a slightly inferolateral orientation, the position of the middle fossa craniotomy was more posterolateral (Fig. 7, right). After the craniotomy, the dura was incised to allow visualization of the temporal pole.


DISCUSSION Endoscopic approaches to the cranial base are gaining popularity. The endoscope provides superb illumination, high magnification, a panoramic view, and the ability to “look around corners.” The nasal cavity and nasopharynx compose an ideal anatomic corridor for the endoscope. With the aid of intraoperative image guidance, the entire midline cranial base from the crista galli to the clivus can be accessed (12, 13). Endonasal endoscopic access is also possible to off-midline structures such as the cavernous sinus (3, 11) and the pterygopalatine fossa (1, 4, 8, 18). The middle fossa is a relatively difficult area to access with the endoscope. We present a novel endoscopic technique to access the greater wing of the sphenoid, a region we refer to as the rostral middle fossa. The rostral middle fossa is bounded anteriorly by the superior orbital fissure, medially by the foramen rotundum, posteromedially by the foramen ovale, and posterolaterally by the suture between the greater wing of the sphenoid and the temporal squama. Although this region is often colloquially referred to as the “temporal pole,” “rostral middle fossa” is a more accurate description of its location within the cranial base. The exocranial footprint of the rostral middle fossa encompasses the medial half of the lateral wall of the orbit, the superior aspect of the anterior spine of the pterygoid process, and the roof of the infratemporal fossa medial to the infratemporal crest. The infratemporal fossa underlies the rostral middle fossa. Medially, the infratemporal fossa communicates with the pterygopalatine fossa through the pterygomaxillary fissure. The laterally positioned infratemporal fossa and the medially positioned pterygopalatine fossa bound the maxillary sinus posteriorly. Most neurosurgeons use a subtemporal, pterional, or orbitozygomatic approach to address a lesion of the rostral middle fossa. The subtemporal approach can be accomplished through a small incision just anterior to the tragus and extending rostrally from the level of the zygomatic process. A temporal craniotomy is placed in this location. Despite maximization of methods to promote brain relaxation, it is our experience that significant retraction of the anterior temporal lobe is required to reach the rostral middle fossa. A pterional craniotomy with drilling of the pterion provides access to the rostral middle fossa with less brain retraction than the subtemporal approach. The incision and elevation of the temporalis muscle that is required for a pterional craniotomy often causes significant temporalis atrophy and impaired or painful mastication (6, 17). A pterional craniotomy supplemented with an orbitozygomatic osteotomy (22) provides excellent access to the rostral middle fossa, requiring minimal brain retraction. This technique requires extensive manipulation of the temporalis muscle; temporalis atrophy and at least temporary impairment of mastication are virtually assured. The maxillary sinus provides an alternative anatomic corridor through which to access the infratemporal fossa and the overlying rostral middle fossa. Existing transmaxillary approaches to the retromaxillary region are divided into microsurgical approaches through an anterior maxillotomy and endoscopic endonasal approaches through a medial maxillotomy.



(11), which increases the technical difficulty and the need for specialized instruments. The ESTA to the rostral middle fossa compares favorably with the previously discussed approaches to this region. In comparison with the microsurgical approaches, the sublabial incision and maxillary antrostomy are smaller, reducing the risk of complications such as FIGURE 4. A, isolated sphenoid bone showing the anatomic variation in the anterior junction of the lateral pteryinfraorbital nerve injury and goid plate and the infratemporal surface of the greater wing of the sphenoid. Red line, spine formed by the anterior junction of the lateral pterygoid plate and greater wing of the sphenoid; blue circles, initial entry points of the endooroantral fistula. The cosmetic scopic sublabial transmaxillary approach. B, endoscopic view of pterygoid plate with its anterior spine. C, schematic advantages over the transfacial illustration showing the cranial base region that is readily accessible through the endoscopic sublabial transmaxillary incision are obvious. approach. PP, pterygoid plate; LPM, lateral pterygoid muscle; RMF, retromaxillary fat; ITF, infratemporal fossa; FR, There are several theoretical foramen rotundum. advantages over the endonasal endoscopic approaches as Anterior maxillotomy through a sublabial incision is part of well: 1) endoscopes and instruments angled beyond 30 degrees the otolaryngologists’ armamentarium for treating maxillary are not required; 2) the nervous structures within the pterysinus pathology (15). An extension of this approach through the gopalatine fossa, including the maxillary nerve, the vidian posterior wall of the maxillary sinus provides access to the nerve, and pterygopalatine ganglion, are not manipulated; and pterygopalatine fossa and infratemporal fossa. Before the 3) compartments lateral to the pterygopalatine fossa, i.e., the advent of transnasal endoscopy, this technique was used for rostral middle fossa and infratemporal fossa, are less cumbermaxillary artery ligation and pterygopalatine fossa tumor some to access. These advantages are supported by the work of resection (9). Couldwell et al. (5) used the sublabial transmaxHar-El (9), who published the only previous report in the literillary microsurgical approach coupled with pterygoid plate ature regarding the use of the endoscope-controlled sublabial osteotomy to access the anterior cavernous sinus. transmaxillary approach. The author used the technique in conTransmaxillary approaches through limited transfacial incijunction with endonasal endoscopy to approach the pterysions have been reported by Sabit et al. (21) and Roche et al. gopalatine fossa. Har-El reported that the endoscopic sublabial (20). Sabit et al. (21) described a nasolabial fold incision and an transmaxillary route improves exposure and surgeon comfort en bloc osteotomy of the anterior and lateral maxilla to access in approaching the lateral pterygopalatine fossa and the pterythe infratemporal fossa and middle fossa. Roche et al. (20) gomaxillary fissure. described a nasal crease incision and anterior and posterior maxillotomies to access the infratemporal fossa. Alfieri et al. (1), Cavallo et al. (4), and Kassam et al. (11) have all described an endonasal endoscopic middle turbinectomy approach to the pterygopalatine fossa that traverses the medial and posterior walls of the maxillary sinus. Kassam et al. (11) have detailed the exposure of the cavernous sinus and medial middle fossa (the quadrangular space) through this approach as well as its extension to the infratemporal fossa and lateral middle fossa. Alfieri et al. (1) have also described an endoscopic endonasal inferior turbinectomy transmaxillary approach to the pterygopalatine fossa and infratemporal fossa. The endoscopic endonasal middle turbinectomy approach to the middle fossa described by Kassam et al. (11) is a technically demanding approach that requires a generous nasal sepFIGURE 5. Transmaxillary endoscopic view showing tostomy for approach from the contralateral nostril. The latmiddle fossa dura through craniotomy. R, roof; IN, eral nasal wall and medial pterygoid are traversed to expose infraorbital nerve; MSAN, middle superior alveolar the cavernous sinus, with the foramen rotundum roughly nerve; EAC, ethmoidal air cells; D, dura; RMF, retrodefining the lateral limit of exposure (1, 11). Extension of this maxillary fat. Inset, magnified view of enlarged rostral approach to the infratemporal fossa or the overlying rostral middle fossa craniotomy. middle fossa requires the use of 45- and 70-degree endoscopes

A

NEUROSURGERY

B

C


ONG ET AL.

FIGURE 6. Position of endoscopic sublabial transmaxillary craniotomy within middle cranial fossa is demonstrated. LSW, lesser sphenoidal wing; CNII, second cranial nerve; CP, clinoid process; ICa, internal carotid artery; LPM, lateral pterygoid muscle; MMa, middle meningeal artery; CNIII, third cranial nerve; PTB, petrous temporal bone; TC, tentorium cerebelli.

FIGURE 7. Cranial view showing the cranial base with dura intact. Left, craniotomy created through endoscopic sublabial transmaxillary approach with the posterior maxillotomy as described. Right, craniotomy created through sublabial transmaxillary approach using a more inferolateral endoscopic trajectory.

It is our experience that the ESTA maintains these advantages in accessing the infratemporal fossa and rostral middle fossa. Meningiomas and dumbbell schwannomas are typical tumors of this region that could potentially be addressed by the ESTA. Head and neck tumors that extend to the infratemporal and middle fossa may also be suitable for endoscope-controlled resection through this approach. As an alternative to the transmaxillary technique, we also explored the viability of extending the endoscope-endonasal ipsilateral inferior turbinectomy transmaxillary approach, described by Alfieri et al. (1), to reach the rostral middle fossa. In this approach, the inferior turbinate is resected, and a medial maxillotomy is fashioned in the former location of the inferior turbinate. Once the maxillary sinus is entered, a posterior maxillotomy is created inferior to the infraorbital nerve. The retromaxillary space is bluntly dissected, and the lateral pterygoid plate is identified. A cranial opening is made in the roof of the infratemporal fossa. Although this variant approach avoids a sublabial incision, we found it wanting for several reasons. With the 30-degree endoscope, the nasal septum restricted the degree of lateral angulation that could be achieved for visualization of the infratemporal fossa. Alfieri et al. (1) reported that 45- and 70degree endoscopes are required for appropriate exposure of this region. With the 30-degree endoscope, we had to enter the pterygopalatine fossa and manipulate its contents to gain adequate access to the floor of the rostral middle fossa. Additionally, we were concerned about the clinical consequences of the removal of the inferior turbinate. The inferior turbinate plays a major role in humidifying, warming, filtering, and directing airflow within the nose. Inferior turbinectomy can cause serious problems with breathing and even lead to atrophic rhinitis, resulting in atrophy and crusting of

the remaining nasal mucosa. As a result, the variant approach was discarded. Because this study was conducted solely in cadavers, it fails to take into account the difficulties and complications that could conceivably arise in the clinical setting. The potential clinical pitfalls to the ESTA to the middle fossa include: 1) complications related to the sublabial incision and anterior and posterior maxillotomies; 2) injury to the retromaxillary neurovascular contents; 3) impairment of mastication attributable to manipulation of the lateral pterygoid muscle; and 4) cerebrospinal fluid leak. These are each addressed in turn. The Caldwell-Luc approach (sublabial incision and anterior maxillotomy) has been used by otolaryngologists for more than 100 years to access the maxillary sinus, orbital floor, pterygopalatine fossa, and infratemporal fossa (14, 15). In recent years, its role in the treatment of chronic maxillary sinusitis has been largely supplanted by endonasal endoscopic middle meatus antrostomy. DeFreitas and Lucente (7) reviewed 474 CaldwellLuc operations and reported the rates on long-term complications such as facial (infraorbital) numbness or paresthesias (9%), facial asymmetry (0.6%), oroantral fistula (1%), and tooth devitalization (0.4%). In contrast, Matheny and Duncavage (15) found no long-term complications in a review of 133 patients who underwent the operation. The authors emphasized the importance of careful entry into the maxillary sinus and protection of the infraorbital nerve during periosteal elevation. The Caldwell-Luc approach used by otolaryngologists is an open approach. Our endoscopic modification requires a smaller anterior maxillotomy, and we anticipate that this significantly reduces the risk of facial asymmetry, oroantral fistula, and infraorbital nerve injury resulting in numbness or dysesthesias. Dental denervation is a theoretical problem with both anterior and posterior maxillotomy. The anterosuperior alveolar




nerve innervates the canines and incisors. It descends in the anterior wall of the maxillary sinus and is placed at risk by the anterior maxillotomy. As noted earlier, the reported rates of tooth denervation with anterior maxillotomy are low, ranging from 0 to 0.4% (6, 15). The middle superior alveolar nerve is present in 23 to 57% of cadaver specimens and arises variably from the infraorbital nerve, including from its origin within the posterior wall of the maxillary sinus (2, 19). This origin would potentially place the medial superior alveolar nerve, and, therefore, molar innervation, at risk from the described posterior maxillotomy. The posterosuperior alveolar nerves also have a variable course, running either within the lateral wall of the maxilla or under the mucous membrane of the maxillary sinus and eventually innervating the molars (16). Murakami et al. (16) have reported that very few of the fibers underlying the mucosa communicate with the extramaxillary dental plexus. It can be extrapolated that mucosal extirpation from the maxillary sinus poses little risk of dental denervation. The limited posterior maxillotomy located inferior to the origin of the infraorbital nerve within the maxillary sinus leads directly into the infratemporal fossa. The contents of the pterygopalatine fossa, including the pterygopalatine ganglion, zygomatic nerve, vidian nerve and artery, palatine nerves and arteries, and sphenopalatine artery, are bypassed. The maxillary artery is the only structure of consequence within the anterior infratemporal fossa, and it can be safely ligated as necessary. The mandibular division of the trigeminal nerve is at risk only if the craniotomy within the infratemporal fossa is carried sufficiently posteriorly. Elevation of the lateral pterygoid from its insertion within the infratemporal fossa could lead to problems with mastication. The lateral pterygoids act to advance the mandibular condyles, thereby depressing the mandible and opening the mouth. Unilateral lateral pterygoid dysfunction could result in ipsilateral limitation of opening of the mouth and deviation of the jaw to the contralateral side. Our approach requires only partial elevation of the lateral pterygoid from its bony origin on the greater wing of the sphenoid, and the muscle belly is left intact. We hope that any dysfunction of mastication would be temporary. Cerebrospinal fluid leak is a potential problem with any endoscopic cranial base approach because primary repair of the dural defect is not possible. We anticipate reconstructing the middle fossa defect with a technique similar to that described by Kassam et al. (10). This multilayer technique uses a subdural Duragen (Confluent Surgical, Waltham, MA) inlay graft and an extradural, extracranial acellular dermis onlay graft. Duragen sealant (Confluent Surgical) is applied over the inlay graft. The vascularized lateral pterygoid muscle and retromaxillary fat pad are then positioned over the reconstruction. Any additional dead space within the infratemporal fossa is filled with harvested abdominal fat autograft. If an endoscopic middle meatus antrostomy is added to aid maxillary sinus drainage, a transnasal Foley catheter could be passed into the maxillary sinus and inflated to provide additional temporary tamponade to the reconstruction.

NEUROSURGERY

As with any endoscope-controlled cranial base approach, there is a steep learning curve with the ESTA. The limited size of the anterior maxillotomy requires significant comfort with largely one-handed dissection within the retromaxillary space and with bleeding control, especially if the cranial opening is carried posteriorly toward the pterygoid plexus. A surgeon should be thoroughly familiar with endoscope-controlled techniques as well as the regional anatomy. Practice in the anatomy laboratory is an essential prerequisite to attempting this approach in the clinical setting. Surgeons should also be cognizant of the capabilities and limitations of the ESTA. The approach is best suited for lesions confined to the rostral middle fossa that displace the temporal pole laterally or those that extend inferiorly into the infratemporal or pterygopalatine fossae. Lesions that displace the temporal pole medially and reach the endocranial surface of the calvarium are better accessed by the pterional approach and its extensions.

CONCLUSION The ESTA is a novel method for accessing the rostral middle fossa. Standard open approaches to this region require varying degrees of brain retraction and temporalis muscle disruption. Endonasal approaches to this region are technically challenging as a result of the need for endoscopes and instruments angled greater than 30 degrees and that are best suited to access the middle fossa medial to the foramen rotundum. The ESTA avoids these issues, has no external incision, and allows early devascularization of cranial- and dural-based lesions. With the aid of intraoperative image guidance, the approach can be specifically tailored to address pathology of the rostral middle fossa. On the basis of this cadaver study, we advocate additional development of the ESTA to the rostral middle fossa in the clinical setting.

REFERENCES 1. Alfieri A, Jho HD, Schettino R, Tschabitscher M: Endoscopic endonasal approach to the pterygopalatine fossa: Anatomic study. Neurosurgery 52:374–380, 2003. 2. Carsolio Diaz CM, Escudero Morere PG: Upper and medial alveolar nerves. Study of their frequency and point of origin in 100 cases [in Spanish]. An Fac Odontol 25:5–20, 1989. 3. Cavallo LM, Cappabianca P, Galzio R, Iaconetta G, de Divitiis E, Tschabitscher M: Endoscopic transnasal approach to the cavernous sinus versus transcranial route: Anatomic study. Neurosurgery 56 [Suppl 2]:379–389, 2005. 4. Cavallo LM, Messina A, Gardner P, Esposito F, Kassam AB, Cappabianca P, de Divitiis, Tschabitscher M: Extended endoscopic endonasal approach to the pterygopalatine fossa: Anatomical study and clinical considerations. Neurosurg Focus 19:E5, 2005. 5. Couldwell WT, Sabit I, Weiss MH, Giannotta SL, Rice D: Transmaxillary approach to the anterior cavernous sinus: A microanatomic study. Neurosurgery 40:1307–1311, 1997. 6. de Andrade Júnior FC, de Andrade FC, de Araujo Filho CM, Carcagnolo Filho J: Dysfunction of the temporalis muscle after pterional craniotomy for intracranial aneurysms. Comparative, prospective and randomized study of one flap versus two flaps dieresis. Arq Neuropsiquiatr 56:200–205, 1998. 7. DeFreitas J, Lucente FE: The Caldwell-Luc procedure: Institutional review of 670 cases: 1975–1985. Laryngoscope 98:1297–1300, 1988. 8. DelGaudio JM: Endoscopic transnasal approach to the pterygopalatine fossa. Arch Otolaryngol Head Neck Surg 129:441–446, 2003.


ONG ET AL.

9. Har-El G: Combined endoscopic transmaxillary–transnasal approach to the pterygoid region, lateral sphenoid sinus, and retrobulbar orbit. Ann Otol Rhinol Laryngol 114:439–442, 2005. 10. Kassam A, Carrau RL, Snyderman CH, Gardner P, Mintz A: Evolution of reconstructive techniques following endoscopic expanded endonasal approaches. Neurosurg Focus 19:E8, 2005. 11. Kassam A, Gardner P, Snyderman CH, Mintz A, Carrau R: Expanded endonasal approach: Fully endoscopic, completely transnasal approach to the middle third of the clivus, petrous bone, middle cranial fossa, and infratemporal fossa. Neurosurg Focus 19:E6, 2005. 12. Kassam A, Snyderman CH, Mintz A, Gardner P, Carrau RL: Expanded endonasal approach: The rostrocaudal axis. Part I. Crista galli to the sella turcica. Neurosurg Focus 19:E3, 2005. 13. Kassam A, Snyderman CH, Mintz A, Gardner P, Carrau RL: Expanded endonasal approach: The rostrocaudal axis. Part II. Posterior clinoids to the foramen magnum. Neurosurg Focus 19:E4, 2005. 14. Macbeth R: Caldwell, Luc, and their operation. Laryngoscope 81:1652–1657, 1971. 15. Matheny KE, Duncavage JA: Contemporary indications for the Caldwell-Luc procedure. Curr Opin Otolaryngol Head Neck Surg 11:23–26, 2003. 16. Murakami G, Ohtsuka K, Sato I, Moriyama H, Shimada K, Tomita H: The superior alveolar nerves: Their topographical relationship and distribution to the maxillary sinus in human adults. Okajimas Folia Anat Jpn 70:319–328, 1994. 17. Nitzan DW, Azaz B, Constantini S: Severe limitation in mouth opening following transtemporal neurosurgical procedures: Diagnosis, treatment, and prevention. J Neurosurg 76:623–625, 1992. 18. Pasquini E, Sciarretta V, Farneti G, Ippolito A, Mazzatenta D, Frank G: Endoscopic endonasal approach for the treatment of benign schwannoma of the sinonasal tract and pterygopalatine fossa. Am J Rhinol 16:113–118, 2002. 19. Robinson S, Wormald PJ: Patterns of innervation of the anterior maxilla: A cadaver study with relevance to canine fossa puncture of the maxillary sinus. Laryngoscope 115:1785–1788, 2005. 20. Roche PH, Fournier HD, Laccourreye L, Mercier P: Surgical anatomy of the infratemporal fossa using the transmaxillary approach. Surg Radiol Anat 23:209–213, 2001. 21. Sabit I, Schaefer SD, Couldwell WT: Modified infratemporal fossa approach via lateral transantral maxillotomy: A microsurgical model. Surg Neurol 58:21–31, 2002. 22. Zabramski JM, Kiris T, Sankhla SK, Cabiol J, Spetzler RF: Orbitozygomatic craniotomy. Technical note. J Neurosurg 89:336–341, 1998.

Acknowledgments Bonnie C. Ong, H.S.C., and Pankaj A. Gore, M.D., contributed equally to the creation of this manuscript. We thank Vincent Strack, B.Med.Sc. and Scott Wheatley, H.S.C., for their contributions to specimen preparation.

COMMENTS

D

uring the past decade endoscopic pituitary surgery has greatly boosted the field of cranial base surgery. Starting from the sella, within a few years the boundaries of the endoscopic approach have been gradually extended, first to the areas delimiting the sphenoid sinus cavity, i.e., the superior clivus, the planum sphenoidale, and the cavernous sinus, and then to many different areas of the cranial base not related to the sphenoid sinus cavity, from the olfactory groove down to the craniovertebral junction. Such anatomical study represents a further contribution to the main body of literature regarding the exploration of a possible new application of the endoscopic technique to a limited area of the cranial base, i.e., the rostral middle fossa. To expose this area the authors used a sublabial transmaxillary route, which somehow offers a more straight view on the posterior maxillary sinus wall than the endonasal route. This route can be used together with the endonasal route to improve visualization and handling of instruments during exposure of this deeply located area. Nevertheless, in evaluating the size limits of the craniectomy through this approach,


one has to consider that only very selected lesions can be treated, mainly those creating a surgical corridor through cranial base erosion. However, all of the pros and cons of the established standard approaches to the same area through the conventional procedure should never be forgotten. Therefore, whether new solutions carrying major risks, such as a high incidence of cerebrospinal fluid (CSF) leaks, or even minor complications, are in the best interest of the patient should be carefully weighed when the classic option is usually safe and effective. Paolo Cappabianca Naples, Italy

T

his is a very nice anatomic description of an endoscopic transmaxillary approach to the middle fossa that starts with a sublabial incision and passes through the infratemporal fossa. The sublabial transmaxillary approach, a modification of the Caldwell-Luc maxillotomy is a well-described approach to the cavernous sinus (3) and infratemporal fossa (1). The authors extend this approach to the middle fossa and use an endoscope to make it “minimal access” surgery. Although the authors are correct in stating that the middle fossa is difficult to reach through a more conventional transnasal endoscopic approach, there are several descriptions of the transnasal endoscopic approach to the middle fossa (2, 4). Experienced endoscopic cranial base surgeons call this corridor the “quadrangular space” (2, 4). Contrary to the authors’ claims, the approach can be performed transnasally without removing the inferior turbinate if a wide antrostomy and maxillotomy are performed, followed by a generous septostomy. The exact indications for the sublabial transmaxillary approach are not clear from this article nor is the method of closure to avoid a CSF leak. Whereas the transnasal approach avoids any incisions or risks of dental numbness, the sublabial approach may be more direct and avoids the awkwardness of operating with endoscopes angled more than 30 degrees. It will be of interest to read about the authors’ experience using this approach in clinical practice. Theodore H. Schwartz New York, New York 1. Allen GW, Seigel GJ: The sublabial approach for extensive nasal and sinus resection. Laryngoscope 91:1635–1639, 1981. 2. Cavallo LM, Messina A, Gardner P, Esposito F, Kassam A, Cappabianca P, De Diviitis E, Tschabitscher M: Extended endoscopic endonasal approach to the pterygopalatine fossa: Anatomical study and clinical considerations. Neurosurg Focus 19:E5, 2005. 3. Couldwell WT, Sabit I, Rice D: Transmaxillary approach to the anterior cavernous sinus: A microanatomic study. Neurosurgery 40:1307–1311, 1997. 4. Kassam A, Gardner P, Snyderman C, Mintz A, Carrau RL: Expanded endonasal approach: Fully endoscopic, completely transnasal approach to the middle third of the clivus, petrous bone, middle cranial fossa, and infratemporal fossa. Neurosurg Focus 19:E6, 2005.

T

his article joins other recent anatomic and technical exercises using the endoscopic “minimally invasive” approach to a variety of cranial base tumors. In this instance the target is the floor and mesial aspect of the middle fossa. It would be useful occasionally for cholesteatomas, schwannomas, and CSF leaks. What is described is a difficult pyramidal approach that transgresses territory unfamiliar to most neurosurgeons. Clearly the anatomy, the vessels and nerves in the area, and the physiological effects of the approach must be carefully considered. It is not immediately apparent that this concept will necessarily lead to superior surgical results compared with alternative strategies. Edward R. Laws, Jr. Boston, Massachusetts



T

he authors have demonstrated an endoscopic transmaxillary approach to the medial middle cranial fossa in cadavers. Although it seems feasible, there are several practical difficulties. A major one that the authors have not taken into consideration is troublesome bleeding from the pterygoid venous plexus that is encountered in any work in the middle fossa. Additional difficulties include limited access and ability to expand the approach, the potential for arterial bleeding from the branches of the internal maxillary artery, and CSF leakage. I have also seen patients undergoing transmaxillary approaches with entry into the infratemporal fossa have problems with trismus (owing to pterygoid fibrosis) and chronic maxillary sinusitis. These should be taken into account. Lesions that can be removed by this approach may be medium-sized schwannomas of the infraorbital nerve or cholesterol granulomas in the area. In contrast, an orbitozygomatic or transzygomatic approach to the middle fossa provides much better exposure. In addition, the risk of temporal lobe injury and temporalis damage are minimal with proper care, and there is little to no risk of CSF leakage. This approach requires clinical validation, with long-term follow-up before being recommended for general use. Laligam N. Sekhar Seattle, Washington

Meningiomas, dumbbell schwannomas, and head and neck tumors that extend to the infratemporal and middle fossa are prototypical tumors of this region that could be addressed by the endoscopic sublabial transmaxillary approach. The authors debate in depth the advantages and limits of the technique, comparing it with the more traditional approaches (subtemporal, pterional, and orbitozygomatic) and with endoscopic endonasal approaches. Even if the arguments of the authors are exhaustive, I am still suspicious regarding the Caldwell-Luc approach. There is a possibility of complications such as facial numbness and dysesthesias resulting from infraorbital and alveolar nerve injury, which although rare, are so troublesome and difficult to treat that their importance overshadows the low incidence. Therefore, I agree with Ong et al. that the endoscopic transmaxillary approach is suitable for small lesions of the rostral middle fossa displacing the temporal lobe laterally and extending to the infratemporal and/or pterygopalatine fossa, but I maintain my preference for the endonasal transmaxillary approach. We use an endoscopic endonasal inferior turbinectomy transmaxillary approach and remove part of the ascending branch of the maxillary bone, straightening the surgical corridor and thus avoiding the need for an endoscope angled more than 30 degrees. Clinical experience with patients will determine which method is preferable, the sublabial or the endonasal.

T

he endoscopic sublabial transmaxillary approach presented by Ong et al. is a novel approach to access the rostral middle fossa.

Giorgio Frank Bologna, Italy

The Anatomical Theater at Berlin, (1730), copper engraving, Freidrich Wilhelm Schmidt. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


TUMOR Clinical Study

MEDIAL SPHENOID RIDGE MENINGIOMAS: CLASSIFICATION, MICROSURGICAL ANATOMY, OPERATIVE NUANCES, AND LONG-TERM SURGICAL OUTCOME IN 35 CONSECUTIVE PATIENTS Stephen M. Russell, M.D. Department of Neurosurgery, New York University School of Medicine, New York, New York

Vallo Benjamin, M.D. Department of Neurosurgery, New York University School of Medicine, New York, New York Reprint requests: Vallo Benjamin, M.D., 530 First Avenue, Suite 7W, New York, NY 10016. Email: [email protected] Received, March 11, 2006. Accepted, July 24, 2007.

OBJECTIVE: On the basis of contemporary multiplanar imaging, microsurgical observations, and long-term follow-up in 60 consecutive patients with sphenoid ridge meningiomas, we propose a modification to Cushing’s classification of these tumors. This article will concentrate on patients from this series with global medial sphenoid ridge tumors. METHODS: Data were collected prospectively for 35 patients with global meningiomas arising from the medial portion of the sphenoid ridge that were surgically treated between 1982 and 2002. RESULTS: All patients were followed for the entire length of this study (mean, 12.8 yr). The tumor size ranged from 2 to 8 cm (mean, 4.5 cm). Of the 24 patients with purely intradural tumors, four (17%) had Simpson Grade I and 19 had Simpson Grade II resections; 23 (96%) had gross total resections. Of the 11 patients with tumors extending extradurally (i.e., cavernous sinus), one (9%) patient had a Simpson Grade II resection, whereas nine (82%) had Simpson Grade III resections, with the latter being all visible tumor removed except that in the cavernous sinus. One (9%) of these 11 patients had a gross total resection, and 9 (82%) had radical resections, with the latter defined as total removal of all intradural tumor. The overall morbidity rate was 18%. There was no surgical mortality or symptomatic cerebral infarction. CONCLUSION: An accurate classification of global medial sphenoid meningiomas is mandatory to gain insight into their clinical behavior and for understanding the longterm efficacy and safety of available treatment options. Primary medial sphenoid ridge tumors consistently involve the unilateral arteries of the anterior cerebral circulation, and therefore, the resection of tumor from around these arteries is the most important operative nuance for their safe excision. KEY WORDS: Brain tumor, Meningioma, Operative nuances, Sphenoid ridge Neurosurgery 62[ONS Suppl 1]:ONS38–ONS50, 2008

I

n 1938, Cushing and Eisenhardt (6) reported the first surgical experience with meningiomas of the sphenoid ridge. On the basis of clinical observations and anatomic characteristics obtained during surgery or autopsy in 53 patients, they divided sphenoid ridge meningiomas into four categories: 1) tumors of the deep or clinoidal third, 2) middle-ridge tumors, 3) en plaque pterional tumors, and 4) global pterional tumors. Although, for the most part, this classification remains accurate, a modification on the basis of contemporary multiplanar imaging and microsurgical observations is warranted. Since Cushing’s seminal monograph, comparisons of published experiences regarding surgical treatment of sphenoid ridge meningiomas have been confusing (2, 5, 7, 8, 12, 14, 19,


DOI: 10.1227/01.NEU.0000297019.60125.71

22, 23, 26). These subsequent reports have various insufficiencies and/or inconsistencies, including incomplete clinical, radiographic, anatomic, or surgical data, too few patients, and the introduction and presentation of results using a number of discordant classification systems, thus limiting their comparative utility. The rarity of these tumors, along with patients’ often unsatisfying surgical outcomes, also likely contributed to the infrequent report of these tumors in the literature. On the basis of a 20-year experience with 60 surgically treated patients with sphenoid wing meningiomas, we have modified Cushing’s classification system by dividing these tumors into three groups: Group I, global medial ridge (35 patients); Group II, global lateral ridge (13 patients); and Group


GLOBAL MEDIAL SPHENOID RIDGE MENINGIOMAS

TABLE 1. Classification of sphenoid ridge meningiomas

a

Group

Clinical presentation

I-Medial globala

Visual deficit

Dissection from vascular elements

Pertinent operative nuance

Visual deficit, cerebral infarction

Postoperative concern

II-Lateral global

Seizure

Dissection from cerebral cortex

Cortical deficit, cerebral edema

III-En plaque

Exophthalmus

Removal of involved bone and dura

Persistence of exophthalmus

Group I is divided into two subgroups: IA, without cavernous sinus extension, and IB, with cavernous sinus extension.

III, hyperostosing “en plaque” (12 patients) (Table 1). Group I was also subdivided into two groups: global medial ridge without extradural extension (IA) and global medial ridge with extradural extension into the cavernous sinus (IB). To adequately describe the clinical characteristics and operative nuances, we concentrate on patients with global medial sphenoid ridge tumors (Groups IA and IB). A separate article will address Group II and III lesions.

PATIENTS AND METHODS Data Collection Between March 1982 and April 2002, the clinical, operative, and radiographic data for all surgically treated meningiomas of the sphenoid ridge by the senior author (VB) at New York University Medical Center were prospectively tabulated. This database included 60 consecutively treated patients, with 35 having tumors of the medial sphenoid ridge and the remaining 25 having either lateral sphenoid ridge or en plaque tumors. This latter group of 25 patients was excluded from further analysis. Eight patients having optic canal and/or sheath meningiomas and 13 patients with primary cavernous sinus meningiomas treated during the study period were also excluded. Optic sheath meningiomas originate within the optic canal and, therefore, cause visual loss when the tumor is still quite small. Because optic sheath tumors are usually removed before they grow large enough to envelop the carotid tree, the operative risks and techniques used to remove these tumors are distinct from medial sphenoid ridge meningiomas. Therefore, they were excluded from this study. Data regarding clinical presentations, pre- and postoperative neurological examinations, other surgical and radiation treatments, pre- and postoperative neuro-ophthalmological examinations, surgical and medical complications after resection, and all follow-up outpatient examinations were tabulated. Data from radiographic images, including computed tomographic (CT) scans, magnetic resonance imaging (MRI) scans, and angiograms, were recorded. All patients received preoperative CT and/or MRI scans with contrast. Imaging was also performed immediately after surgery and annually for 5 years, then at longer intervals depending on the presence or absence of residual and/or recurrent tumor. Patients with tumors larger than 4 cm or with extensive involvement of the cavernous sinus underwent preoperative catheter angiography with or without embolization. Copies from all radiographic studies were filed for later reference. For all cases, detailed reports of operative findings were created immediately after surgery using notes, illustrations, and operative videos. Data regarding intracranial vessel attachment, cistern involvement, dural attachment, extent of resection, and venous anatomy were recorded. Considering that only 35 patients were included in our analysis, our results were reported using only descriptive statistics.

NEUROSURGERY

The extent of tumor resection was determined by postoperative contrast-enhanced imaging in combination with microsurgical observations. Gross total tumor resection was defined as complete removal of tumor, including the excision (Simpson Grade I) or thorough coagulation (Simpson Grade II) of any dural involvement (Fig. 1). Gross total intradural tumor resection was defined as total intradural tumor removal, but residual tumor left within the cavernous sinus (Simpson Grade III) (Fig. 2). Tumor in the orbit, sella turcica, bone, and/or sphenoid/ethmoid sinuses was aggressively removed in these patients. Partial intradural tumor resection was defined as residual gross tumor remaining intradural after surgery (Simpson Grade IV).

Microsurgical Technique

(see video at web site) Initially, a lumbar drain was used intraoperatively to withdraw cerebrospinal fluid for brain relaxation. However, with further experience, it was found to be unnecessary because adequate brain relaxation could be obtained by cerebrospinal fluid drainage from the sylvian fissure early in the operation. All patients were positioned supine with the head elevated slightly above the heart to promote venous drainage. The head was rotated 30 degrees opposite the side of the tumor, bringing the sphenoid ridge into a vertical orientation. A vertically oriented sphenoid ridge helped maintain spatial orientation during the initial tumor resection, which was especially useful when large tumors obscured and distorted normal intracranial landmarks. The head was also extended approximately 25 degrees to allow the brain to fall away from the cranial base, thereby minimizing the amount of brain retraction required during the procedure. A frontotemporal (pterional) craniotomy was performed. Using a high-speed pneumatic drill and remaining extradural, we removed the greater wing of the sphenoid ridge to the lateral limit of the lesser wing. The anterior clinoid, optic canal, and orbital bone were not routinely removed at this stage of the operation. Alternatively, if the tumor extended into the optic canal and/or orbit, drilling of the optic canal and bony exposure of the roof and lateral wall of the orbit was performed intradurally after resection of the intracranial portion of tumor. It is our philosophy to identify and decompress the intradural carotid artery before clinoidectomy or removal of orbital/paraclinoid tumor to preclude the risk of inadvertent arterial damage, which may occur when tumor is removed before identifying the carotid artery. After opening the dura, the operating microscope was used for the remainder of the resection (Fig. 3). The arachnoid of the sylvian cistern was opened widely, exposing the lateral extent of tumor in the sylvian fissure. The anterior temporal bridging veins, capping the lateral aspect of the tumor, were coagulated and sectioned. Frontal and temporal lobe retractors were placed, more to protect the cortical surfaces than to actively retract the lobes, which usually do not require retraction after proper patient positioning and spinal fluid release. Single-layered pieces


RUSSELL AND BENJAMIN

the middle cerebral artery (MCA) were then freed from their arachnoid trabeculae up to their point o f i n v o l v e m e n t w i t h t u m o r. Beginning at these distal uninvolved branches of the MCA and progressing proximally towards the cranial base, the tumor was systematically resected to expose and protect the MCA, lenticulostriate perforators, internal carotid artery (ICA) bifurcation, anterior cerebral artery, and the posterior communicating and anterior choroidal arteries. The carotid and its branches were always located within an arachnoid cleft separating the tumor into anteromedial and posterolateral compartments. The optic nerve was then located at its fixed point at the optic foramen. The tumor was then resected along the optic nerve working towards the chiasm. By this stage, the anteromedial component of the tumor had been removed. Under greater magnification, the smaller posterolateral portion of the tumor was then resected, with special attention given to preserving the posterior communicating and anterior choroidal arteries, along with their respective central perforators. Once the bulk of the intracranial tumor was resected, remaining areas of tumor attachment to the dura were removed, and the underlying dura was thoroughly coagulated or excised. When indicated in patients with tumor extension into the orbit, the optic canal and superior and lateral walls of the orbit were unroofed using a 6-mm or smaller carbide FIGURE 1. A representative patient (Patient 19) with a medial sphenoid ridge meningioma (Type IA) who undercutting burr, the dura overlying went gross total resection (Simpson Grade II). A and B, preoperative axial magnetic resonance imaging (MRI) the optic canal was opened, and scans with enhancement demonstrating a meningioma arising from the medial third of the sphenoid ridge with the intraorbital tumor was reextensive arterial encasement (arrowheads). C, lateral internal carotid angiogram revealing an internal carotid sected. If there was tumor extenaneurysm within the tumor (arrow). D and E, coronal MRI scans with enhancement further demonstrating the sion into the cavernous sinus and extent of arterial encasement by the tumor (arrowheads). F, lateral external carotid angiogram revealing a marked the patient was without ophthaltumor blush (arrowheads). This patient underwent endovascular coiling of the intratumoral aneurysm and preopmoplegia preoperatively, this porerative transarterial particulate embolization. G–I, postoperative MRI scans with enhancement documenting a gross tion of tumor was left undistotal tumor resection. Artifact from the aneurysm coils can be seen (asterisk). turbed. However, if the patient was young and had complete ophof Penrose drain and cottonoids were placed below the retractors to prethalmoplegia, resection of the intracavernous tumor was considered vent adhesion and damage to the cortical surfaces during long operations. after preoperative temporary balloon test occlusion of the ICA. After exposure of the lateral side of the tumor through the sylvian The specific technique used for tumor removal involved coagulation of the tumor’s interior with bipolar forceps followed by gutting the fissure, the dural attachment of the tumor on the frontal and temporal tumor with microdissectors, microscissors, and rarely, the ultrasonic sides of the sphenoid ridge was coagulated with bipolar forceps to aspirator. During the initial debulking, 3 to 4 mL of tumor surroundshrink the tumor and decrease its blood supply. The distal branches of




FIGURE 2. A representative patient (Patient 14) with a medial ridge meningioma (Type 1B) who underwent resection with residual tumor in the cavernous and sphenoid sinuses (Simpson Grade III). Preoperative axial (A and B), sagittal (C), and coronal (D and E) MRI scans demonstrating a large medial ridge meningioma with extensive arterial encasement (arrow in B) and distortion (arrowheads in D). Tumor extension into the cavernous sinus, sphenoid sinus, and posterior fossa is present before surgery (A). F, frontal internal carotid angiogram with overlaid venous phase negative revealing a markedly distorted and thinned internal carotid and middle cerebral arteries (arrowheads). Postoperative axial (G and H) and sagittal (I) MRI scans with enhancement documenting the extent of tumor resection. A complete removal of the intradural tumor, yet residual tumor in the right cavernous and sphenoid sinuses is seen (arrowheads in G). ing the vessels would be left undisturbed. After the initial tumor removal that reduces compression of surrounding structures, these residual layers of tumor were then dissected from the compressed arachnoid attached to the blood vessels using countertraction with suction to immobilize the arachnoid/blood vessel tissue, and the bipolar removed the remnants of the tumor. Furthermore, the springaction of the 0.5-mm bipolar and the sharp point of closed microscissors can both be very effective in releasing the neural and vascular structures from the tumor.

NEUROSURGERY

Important for safe tumor dissection was the constant microsurgical surveillance of the arterial tree and maintenance of the arachnoid investment and cerebrospinal fluid interface between the tumor and involved arteries and nerves. Careful attention was used to stay in the “epi-arachnoid” space (9), thereby preserving the arachnoid layers of the cistern walls. This was done by carefully teasing the t u m o r a w a y f ro m a r a c h n o i d layers, which were maintained, where possible, on the surface of involved arteries or perforators. If an intracranial artery directly supplied a portion of the tumor, these neovessels were coagulated from the inner (tumor) side of the arachnoid and cut before traversing the arachnoid, leaving a coagulated stump to prevent tearing the feeder from the parent vessel. Resection of the tumor from the perforators was accomplished with greater magnification and by identifying their origin and following them distally, keeping their protective arachnoid layer under surveillance and intact. Because a major portion of the tumor and its blood supply has been removed by this phase of the operation, coagulation was used more sparingly. The tumor is resected from the anterior visual pathways only after the majority of the intracranial tumor has been removed and the proximal ICA has been identified. The tumor is always retracted away from the optic apparatus, with resection accomplished by holding the arachnoid of the chiasmatic cistern immobile with microforceps to prevent undue manipulation of the optic apparatus.

RESULTS

Patient characteristics and clinical outcomes are presented in Table 2. Karnofsky Performance Scale scores were determined retrospectively and tabulated both preoperatively and at the 3-month postoperative visit. The preoperative patient characteristics of 35 surgically resected medial sphenoid ridge meningiomas are listed in Table 3. The most prevalent neurological finding was impaired visual acuity and/or visual field loss in 30 (86%) patients. Hydrocephalus was present in nine patients (26%) with preoperative mental



a r t e r i a l s u p p l y f ro m t h e external carotid artery u n d e r w e n t p re o p e r a t i v e endovascular embolization. No complications attributable to the embolization occurred in this series. One patient with an incidental posterior communicating artery aneurysm located within the tumor underwent preoperative coiling by our endovascular team. Tumor characteristics, as determined by radiographic imaging and intraoperative observations, are listed in Table 4. Of the 35 patients in this series, 24 (69%) had purely intracranial tumors (Group IA), whereas 11 (31%) had extracranial extension into the cavernous sinus, orbit, and/or infratemporal f o s s a ( G ro u p I B ) . A l l 3 5 (100%) tumors either partially or completely enveloped the ipsilateral ICA and its branches. No patients were lost to follow-up. All patients received contrast enhanced CT and/or MRI scans immediately after surgery. All patients had annual MRI or CT scans with contrastenhancement during outpatient follow-up visits for the first 5 years postoperatively and on a bi- or triannual basis thereafter, depending FIGURE 3. Serial intraoperative photographs illustrating the resection of a large left-sided medial ridge meningioma (see Fig. 2) using a pterional approach. A, sylvian fissure is split widely, exposing the tumor between the frontal and on the state of residual or temporal lobes. Retractors are used to protect the brain during tumor removal. B, after the distal middle cerebral artery recurrent tumor. The average branches are identified, they are exposed proximally by progressive tumor removal. A thin layer of tumor adherent follow-up period for this to these arteries and their lenticulostriate perforators, can be left in place and removed later in the procedure when series was 12.8 years (range, the tumor is mostly decompressed. C, resection of the anterolateral tumor compartment proceeds with the tumor being 1–19 yr). The surgical outremoved from near the cavernous sinus, posterior communicating artery, and lateral brainstem. Care is taken when come is displayed in Table 5. removing the tumor from the arachnoid layers covering the posterior communicating artery and its small thalamoTwenty-three (96%) out of perforators. D, in this patient, residual tumor is left within the cavernous sinus. Resection of the posteromedial tumor 2 4 patients with purely compartment is completed either between the optic nerves or via the opticocarotid triangle. E, after tumor resection, intradural tumors (Group IA) the preserved intracranial arteries and cranial nerves can be identified. F, lower magnification view of the postresechad Simpson Grade I or II tion operative field demonstrating undamaged frontal and temporal lobes, as well as residual tumor in the cavernous sinus. The asterisk marks the opticocarotid triangle. resections (Simpson Grade I, four out of 24 [17%]; Simpson changes, all of whom were treated with elective ventricuGrade II, 19 out of 24 [79%]), whereas only one (9%) out of 11 loperitoneal shunt placement on a separate date before or patients with tumors having extradural extension (Group IB) after tumor resection. Nine patients (26%) with significant had Simpson Grade II removal. In this latter group, nine (82%)




out of 11 patients underwent Simpson Grade III resections where all visible intradural tumor was removed, but tumor in the cavernous sinus remained unresected. Two patients (one in Group IA and one in Group IB) had Simpson Grade IV resections with residual intradural tumor beyond the cavernous sinus; one developed intraoperative cardiac instability precluding a total resection, the second patient was elderly and gross tumor was purposely left adjacent to the cavernous sinus. These two patients, along with three other patients who demonstrated tumor recurrence on serial postoperative imaging, comprised a group of five patients (14%) who underwent conventional external beam radiation (5000 cGy in 30 fractions). Stereotactic radiosurgery was not used for any of the patients. After receiving radiation, no tumor progression occurred in these five patients (average follow-up, 7.1 yr). At the time of pathological analysis, none of the patients had atypical or malignant meningiomas. The overall morbidity rate was 18%. There were no procedure-related deaths (Table 5). When compared with their preoperative status, 33 patients (94%) had improved or unchanged visual acuity and/or visual field examinations, whereas two (6%) had worsened visual acuity. Among the 30 patients with preoperative visual loss, 22 patients (73%) improved, six (20%) were unchanged, and two (7%) worsened. Of note, these latter two patients had markedly diminished visual function present preoperatively. Further damage to their visual pathways resulted from either manipulation of an already compressed optic nerve or from possible inadvertent disruption of its blood supply. An additional patient, with a preoperative visual acuity of 20/800, had her ipsilateral optic nerve intentionally cut to facilitate removal of tumor extending into the sella. None of the five patients with unaffected vision preoperatively demonstrated worsened vision after surgery. One patient with a leftsided tumor who had a previous partial resection at a different institution developed transient hemiparesis after surgery. The etiology of this deficit was uncertain, with no evidence of infarction on postoperative imaging. No diabetes insipidus, infection, or symptomatic cerebral infarction occurred in any patient. There were no surgical deaths; one patient died 6.5 years after resection at the age of 82 from an unrelated disease.

DISCUSSION Classification of Medial Sphenoid Ridge Meningiomas Cushing and Eisenhardt (6) put forth a classification of sphenoid wing meningiomas based on clinical manifestations, macroscopic operative findings, and postmortem examinations in 53 patients. Although their classification remains mostly valid, a revision that incorporates information and experiences gained from the contemporary use of multiplanar MRI and microsurgical technique is warranted. Therefore, on the basis of 60 operative cases of sphenoid wing meningiomas, we propose the following classification: Group I, global medial ridge; Group II, global lateral ridge; Group III, hyperostosing “en plaque.” These three tumor categories were selected because of

NEUROSURGERY

their different clinical presentations, microsurgical approaches for extirpation, and postoperative complications (Table 1). Of note, only two significant modifications were made. First, Cushing’s middle sphenoid ridge group was removed because these tumors do not have distinct radiographic or microsurgical characteristics differentiating them from lateral ridge tumors to warrant a separate grouping. Second, the medial global ridge tumors were divided into two subgroups with (IA) or without (IB) cavernous sinus extension. This subdivision was added because treatment for tumors without cavernous sinus extension is gross total removal; for tumors with cavernous sinus extension, often only the intradural tumor is removed, with the residual cavernous sinus disease usually observed or radiated. Nakamura et al. (18) have recently confirmed the clinical utility of this proposed subdivision. Although further discussion of lateral and hyperostosing sphenoid ridge meningiomas will be reserved for subsequent communications, the evolving confusion over which tumors comprise the medial ridge category will, at this point, be discussed. In agreement with Cushing and Eisenhardt (6), we define a medial sphenoid ridge meningioma as a global meningioma originating from the medial aspect of the sphenoid ridge that envelopes the internal carotid artery (partially or completely) and, when large, its branch vessels with or without cavernous sinus extension. Complete or partial engulfment of the internal carotid artery and its branches is the key requirement, being that this is the feature that differentiates these tumors from many other parasellar meningiomas. In following this definition, optic canal meningiomas, primary cavernous sinus meningiomas with a large global component displacing the intradural vessels, and anterior fossa tumors involving the superior aspect of the sphenoid ridge but projecting solely upward and not involving the carotid tree are excluded; rightfully so, considering they do not have the same prognosis and operative nuances. Of note, certain tumors in the parasellar region defy useful classification, particularly very large (5 cm) meningiomas involving the medial ridge, carotid tree, and cavernous sinus. For these tumors, their origin, cavernous sinus versus medial sphenoid ridge, cannot be determined. However, with the use of MRI, angiography, and operative findings, one can usually categorize most large meningiomas in this region as either cavernous sinus or sphenoid ridge in origin. Proper and consistent classification is mandatory for evaluating new treatment techniques and comparing the published literature regarding medial ridge meningiomas. Unfortunately, this has not been the case.

Pathoanatomic Observations Intracranial venous anatomy likely has an important influence on the direction of tumor growth. As defined by Oka et al. (20), the superficial sylvian vein receives adjacent frontal, temporal, and parietal venous blood and most commonly empties into the sphenoparietal sinus via bridging veins midway or medially along the sphenoid ridge. This bridging vein connection is at the lateral border of the superior orbital fissure, which



TABLE 2. Clinical data from 35 patients with surgically resected medial sphenoid ridge meningiomaa Tumor size (cm)

Dominant hemisphere

Cavernous sinus extension

Previous resection

Visual loss

55

No

No

No

Headache

64

No

Yes

Yes

206

Visual loss

53

Yes

No

Yes

219

Altered mentation

44

No

No

No

49

126

Visual loss

32

No

No

Yes

74

174

Visual loss

22

Yes

No

No

7

52

183

Visual loss

43

Yes

No

No

8

69

194

Altered mentation

65

Yes

No

No

9

43

201

Altered mentation

77

No

Yes

Yes

10

63

225

Visual loss

33

Yes

Yes

No

11

77

78

Altered mentation

43

No

Yes

No

12

68

201

Altered mentation

65

No

No

No

13

62

189

Exopthalmos

55

No

Yes

No

14

64

187

Visual loss

66

Yes

Yes

No

15

68

161

Visual loss

21

Yes

No

No

16

67

177

Visual loss

54

Yes

No

No

17

48

180

Visual loss

66

No

No

No

18

68

180

Visual loss

32

Yes

No

No

19

44

171

Double vision

54

Yes

No

No

20

58

178

Altered mentation

65

No

No

No

21

63

152

Visual loss

32

Yes

No

No

22

66

174

Visual loss

32

Yes

No

No

23

57

168

Headache

53

Yes

Yes

No

24

59

156

Altered mentation

65

No

Yes

Yes

25

62

133

Altered mentation

42

No

No

No

26

58

108

Visual loss

44

Yes

No

No

27

65

130

Headache

55

Yes

Yes

No

28

48

153

Visual loss

21

Yes

No

Yes

29

58

150

Double vision

55

No

Yes

No

30

68

93

Recurrent falls

54

No

No

No

31

66

94

Altered mentation

55

Yes

Yes

No

32

63

69

Visual loss

75

Yes

No

No

33

54

12

Visual loss

53

Yes

No

No

34

65

57

Visual loss

33

No

No

No

35

34

52

Visual loss

43

Yes

No

No

Patient no.

Patient age

Follow-up (mo)

1

66

218

2

55

228

3

60

4

47

5 6

Presenting symptoms

a

MRI, magnetic resonance imaging; Pre, preoperative; post, postoperative; KPS, Karnofsky Performance Scale. Gross total, all tumor removed; gross total intradural, all intradural tumor removed (except cavernous sinus tunor); partial intradural, some intradural tumor remains. c Conventional external beam radiotherapy was used; indications included residual intracranial tumor or tumor growth postoperatively. d Postoperative KPS score recorded at 3 months. b




Preoperative embolization

Surgical complications

MRI extent of resectionb

Simpson grade

Postoperative tumor growth

Postoperative radiationc

Visual outcomes

Pre/ post KPSd

No

None

Gross total

I

No

No

Improved

90/90

No

3rd nerve palsy

Gross total

II

No

No

Unchanged

90/80

No

None

Partial intradural

IV

No

Yes

Improved

80/80

No

None

Gross total

II

No

No

Improved

60/90

No

None

Gross total

II

No

No

Unchanged

90/90

No

None

Gross total

II

No

No

Improved

90/90

No

Visual loss

Gross total

II

No

No

Worse

90/80

No

None

Gross total

II

No

No

Improved

60/80

Yes

Cerebral edema

Gross total intradural

III

Yes

Yes

Unchanged

70/70

No

None


III

No

No

Unchanged

90/90

No

None

Partial intradural

IV

No

Yes

Unchanged

50/60

No

None

Gross total

II

No

No

Unchanged

60/80

No

None


III

No

No

Improved

90/90

Yes

None


III

Yes

Yes

Unchanged

90/90

No

None

Gross total

II

No

No

Improved

90/90

No

None

Gross total

II

No

No

Improved

80/90

Yes

None

Gross total

II

No

No

Improved

80/90

No

None

Gross total

II

No

No

Improved

90/90

No

None

Gross total

II

No

No

Improved

90/90

No

None

Gross total

I

No

No

Unchanged

80/80

No

None

Gross total

II

No

No

Unchanged

90/100

No

Epidural hematoma

Gross total

I

No

No

Improved

90/90

No

None


III

No

No

Improved

90/90

Yes

None


III

Yes

Yes

Unchanged

60/70

No

None

Gross total

II

No

No

Unchanged

60/90

No

Visual loss

Gross total

II

No

No

Worse

90/80

Yes

Temporary hemiparesis


III

No

No

Improved

90/80

No

None

Gross total

II

No

No

Improved

90/100

Yes

None


III

No

No

Improved

90/90

No

None

Gross total

II

No

No

Improved

70/80

Yes

None


III

No

No

Improved

60/no

Yes

None

Gross total

II

No

No

Improved

90/90

Yes

None

Gross total

I

No

No

Improved

90/90

No

None

Gross total

II

No

No

Improved

90/90

No

None

Gross total

II

No

No

Improved

90/90

NEUROSURGERY



TABLE 3. Preoperative characteristics of 35 medial sphenoid ridge meningiomas

TABLE 5. Surgical outcome of 35 medial sphenoid ridge meningiomas

Age (yr)

Follow-up (yr)

60 (34–77)

Sex

Female Male

Resection (Group IA) 29 (83%)

Simpson I

4/24 (17%)

6 (17%)

Simpson II

19/24 (79%)

Simpson IV

1/24 (4%)

Neurological examination

Visual deficit

a

12.8 (1–19)

30 (86%)

Resection (Group IB)

Altered mentation

9 (26%)

Simpson II

Oculomotor palsy

8 (23%)

Simpson III

9/11 (82%)

Hemiparesis

7 (20%)

Simpson IV

1/11 (9%)

Dysphasia

4 (11%)

Complications

Exophthalmos

4 (11%)

Visual loss

2 (6%)

Hydrocephalusa

9 (26%)

Third nerve palsy

1 (3%)

Preoperative embolization

9 (26%)

Cerebral edema

1 (3%)

Previous resection

6 (17%)

Epidural hematoma

1 (3%)

Temporary hemiparesis

1 (3%)

Infarction

—

Infection

—

All nine patients who presented with altered mentation.

TABLE 4. Tumor characteristics of 35 medial sphenoid ridge meningiomas Size (cm)

4.5 (2–8)

Side

Left Right

20 (57%) 15 (43%)

Arterial involvement

Internal carotid

35 (100%)

Middle cerebral

28 (80%)

Posterior communicating/choroidal

19 (54%)

Anterior cerebral

17 (49%)

Basal cistern involvement

Carotid

35 (100%)

Sylvian

31 (89%)

Chiasmatic

31 (89%)

Lamina terminalis

26 (74%)

Crural

25 (71%)

Ambient Interpeduncular Cavernous sinus involvement

7 (20%) 4 (11%) 11 (31%)

corresponds to the lateral extent of Cushing and Eisenhardt’s (6) group of meningiomas originating from the “inner third” of the sphenoid ridge. Although objective pathoanatomic data are not available to support this hypothesis, our intraoperative observations suggest the importance of bridging veins in restricting tumor growth lateral to the temporal pole, thereby encouraging the ultimate involvement of anterior circulation arteries by


Surgical mortality

1/11 (9%)

—

Overall visual outcome

Improved

22 (63%)

Unchanged

11 (31%)

Worsened

2 (6%)

Postoperative radiation

5 (14%)

Recurrence

3 (9%)

tumor. This concept of meningioma growth in the path of least resistance was emphasized by Cushing and Eisenhardt (6). We also found that medial ridge rumors more commonly lift up the frontal lobe than extend into the temporal fossa, further suggesting the importance of anchoring bridging veins in the pathological anatomy of these tumors. Furthermore, bridging veins draining into the sphenoparietal sinus may also help shield the internal carotid artery and its branches from being engulfed by tumors that originate lateral to these veins along the sphenoid ridge. Further study of this concept is needed. It is thought that as meningiomas grow, adjacent arachnoid cisterns become effaced and then obliterated from tumor compression (Fig. 4). As the tumor extends along the path of least resistance, the arachnoid layers remain intact, enveloping adjacent arteries and nerves. Although the cisterns subsequently collapse from progressive tumor compression, spinal fluid flow continues to occur in these potential arachnoidal compartments. However, with long-standing compression, spinal fluid flow will cease, allowing the arachnoid layers to fuse to the enveloped tissue. When this occurs, the tumor will recruit neovasculature from cerebral arteries through the adjacent leptomeninges, with the tumor becoming markedly attached to these structures.



because of no intervening arachnoid membrane, 2) medial ridge tumors without marked attachment to carotid tree secondary to an intervening arachnoid membrane being present, and 3) small tumors solely within the optic canal. In agreement with others (11, 14, 24), we do not use this classification for the following reasons. First, Group 3 tumors (optic canal meningiomas) do not involve the carotid tree and, therefore, should not be included when evaluating medial ridge (clinoidal) meningiomas. Next, although meningiomas originating proximal to the carotid artery’s arachnoid ensheathment (Group 1 tumors) may occur and, in a FIGURE 4. Photomicrographs illustrating the progressive effacement of arachnoid cisterns leading to adhesion of small minority of patients, tumor to intracranial vessels. Images taken from a patient with an incidental medial sphenoid ridge meningioma explain why some tumors are identified at autopsy. A, at the tumor’s edge, the posterior cerebral artery (asterisk in lumen) is partially enveloped markedly attached to the by meningioma; however, an arachnoid cleft remains between the tumor and vessel wall (arrow) (original magniintracranial vessels without fication, 2.5). B, a greater magnification view of the arachnoid cleft (white bar) between tumor (above) and vesa n a p p a re n t i n t e r v e n i n g sel wall (below) seen in A (original magnification, 40). C, in contrast, the anterior cerebral artery (asterisk in arachnoid layer, other factors, lumen) was deeply embedded within the tumor and did not have an intervening arachnoid cleft (original magnifiincluding chronicity of tumor cation, 10). D, greater magnification view of the anterior cerebral artery seen in C revealing tumor (above) adhercompression, tumor conent to, and invading (arrow), the vessel wall (below) (original magnification, 40). All samples were stained with sistency, invasiveness, and hematoxylin and eosin. whether or not the patient has had a previous resection, which four out of the 24 patients did, are probably more imporTaking advantage of this pathological growth pattern, small tant determinants of resectability. All patients in our series had arteries within these arachnoid layers located around the tumor an intervening arachnoid membrane present. Furthermore, the can be preserved. To do this, the arachnoid layer should be presence of intervening arachnoid cannot be reliably detercarefully identified and teased away from the tumor, thereby mined preoperatively, thereby limiting this finding’s relevance visualizing and preserving the delicate vessels within. When to classification and treatment decision making. spinal fluid flow is present in the subarachnoid space, the Risi et al. (23) reported 34 patients undergoing removal of arachnoid usually does not adhere to the tumor and, therefore, “clinoidal” meningiomas, making specific comment on the can be readily separated. However, in areas in which the arachmanagement of cavernous sinus extension. They subclassified noid is scarred to the tumor and spinal fluid ceases to flow, sepglobal medial ridge meningiomas into the following three cataration of the tumor from arachnoid is quite difficult, if not egories: 1) pure clinoidal, 2) clinoidal with lateral extension, impossible. Although uncommon, when a tumor is severely and 3) clinoidal with cavernous sinus extension (with or withattached to an intracranial vessel (13), it may be more prudent out lateral extension). We do not believe the presence of lateral to leave a small piece of the tumor on the vessel wall than to extension warrants a separate subcategory; it does not portend risk damage to the artery. It is our experience that these small a distinct clinical presentation, pathological anatomy, or surgidevascularized tumor remnants tend to involute and do not cal outcome. For similar reasons, dividing a tumor with cavlead to early tumor regrowth. ernous sinus extension into two groups based on the presence Review of the Contemporary Literature of lateral extension along the sphenoid ridge is also of question(1990–Present) able utility. In their series, 15 out of 34 patients had cavernous sinus extension, all of whom underwent attempted surgical Al-Mefty (2) reported 24 patients with clinoidal meninresection. Four of these patients had complete resection of their giomas, subclassifying the tumors into three groups: 1) medial cavernous sinus extension, whereas only partial resection was ridge tumors with marked attachment to the carotid tree

NEUROSURGERY



TABLE 6. Studies of medial sphenoid ridge meningiomas in the peer-reviewed literaturea Series (ref. no.)

No. of patients

Average size

Average follow-up

Mortality

Gross total resection

Improved vision

Recurrence

Cushing and Eisenhardt, 1938 (6)

13

NA

7 yr

15%

31%

8%

36%

Ugrumov et al., 1979 (26)

17

NA

NA

23%

0%

NA

NA

Konovalov et al., 1979 (12)

70

NA

NA

19%

85%

NA

NA

Ojemann, 1980 (19)

13

NA

11.3 yr

0%

0%

15%

31%

7

NA

1–8 yr

43%

29%

17%

25%

9

NA

7 yr

4%

50%

48%

12%

18

NA

NA

27%

NA

NA

25%

Bonnal et al., 1980 (5) Pompili et al., 1982 (22) Fohanno and Bitar, 1986 (8) Al-Mefty, 1990 (2)b

24

NA

4.8 yr

8%

89%

8%

4%

Risi et al., 1994 (23)

34

NA

1.9 yr

6%

59%

32%

21% NA

6

5 cm

3 mo

0%

66%

NA

Lee et al., 2001 (15)

14

3.7 cm

3.1 yr

0%

87%

75%

0%

Abdel-Aziz et al., 2004 (1)c

38

3 cm

8 yr

0%

58%

NA

11%

Present study

35

4.5 cm

12.8 yr

0%

69%

63%

9%

Day, 2000 (7)

a

NA, not available. Study included optic foramen meningiomas. c Study included only meningiomas contacting or infiltrating the cavernous sinus. b

possible in the remaining 11 patients. The surgical outcome in this subset of patients was somewhat poor, including six patients who improved, four who were stable, four who were “aggravated,” and one who died. From this experience, the authors recommended a more conservative approach to the removal of cavernous sinus extension, stating that in cases of “diffuse invasion surgical extirpation should not be attempted because of the high risk of damage to the intracavernous internal carotid artery and cranial nerves encased in tumor.” A report by Lee et al. (14) concentrated on the visual outcome of 15 patients after resection of medial ridge meningiomas. They documented an improvement in preoperative vision in 75% of the patients and attributed their good results to extradural bony decompression of the superior orbital fissure, optic canal, and anterior clinoid, followed by sectioning of the falciform ligament, all before tumor resection. Although releasing the optic nerve to minimize its indirect compression against the falciform ligament during tumor resection may promote good visual outcomes, it is our opinion that gentle microsurgical resection of the tumor and, more importantly, strict preservation of all vessels feeding the optic structures are probably more relevant to visual outcome. Despite patients in our series having larger tumors (mean, 4.5 versus 3.7 cm), their visual outcomes were nonetheless similar to those reported by Lee et al. (improved, 73 versus 75%), despite our not decompressing the optic nerve before tumor removal in any patient. Abdel-Aziz et al. (1) recently reported 38 patients with large sphenoid wing meningiomas involving the cavernous sinus. They concluded that the degree of cavernous sinus involvement determined whether or not it was safe to attempt gross total resection. Specifically, if the tumor had invaded the cavernous sinus medial to the cranial nerves situated in its lateral


wall, complete resection should not be attempted. This recommendation is made in light of the morbidity associated with resecting meningiomas within the cavernous sinus, along with the availability and high efficacy of salvage radiotherapy. We agree with their conclusions and do not recommend the removal of tumor within the cavernous sinus. However, this series is not directly comparable with ours or others in the medial sphenoid ridge literature because the authors do not include medial ridge tumors without cavernous sinus involvement; on the basis of their tumor description and classification, along with the figures presented in their article, many of the meningiomas in their series were likely not sphenoid ridge tumors but rather primary cavernous sinus tumors that extended outside the cavernous sinus and secondarily involved the medial sphenoid ridge.

Operative Results Considering that patients with medial sphenoid ridge meningiomas often do not present until their tumors are greater than 3 cm in diameter, already surrounding or displacing the optic nerves, surgical resection remains a common firstline treatment for these tumors. Despite the methodological disparities in series reported in the literature (Table 6), certain clinical trends can be identified (2–8, 12, 14, 19, 22, 23, 26). For example, a trend toward improved patient outcomes is evident, including reduced mortality, a greater percentage of patients having Simpson Grade I or II tumor resections, and improved vision postoperatively; results likely related to use of the operating microscope, cranial base techniques, modern perioperative intensive care management, and understanding the meningioma-arachnoid interface. We believe that our low incidence of major operative morbidity, including cerebral



infarction, ophthalmoplegia, and cerebrospinal fluid leak, was related directly to our in-depth study and understanding of pathological tumor anatomy preoperatively, along with our conservative approach to resecting extradural tumor in the cavernous sinus, sphenoid/ethmoid sinuses, and bone. Compared with previous reports, the patients in our series harbored larger tumors (mean diameter, 4.5 cm), and it may be extrapolated that patients with smaller lesions may have even better surgical outcomes than reported here because smaller tumors have less involvement of vital structures. Cushing and Eisenhardt (6) noted that “the crux of the removal lies in freeing the growth from its entanglement with the vessels at the carotid bifurcation,” and he cautioned surgeons about the hazards of entering the carotid field. Injury to the vessels of the anterior circulation has been the major cause of operative mortality and morbidity in past surgical series, even after the advent and routine use of the operating microscope. Although two series published in the 1980s had a surgical mortality of 43% (5) and 27% (8), more recent reports have shown lower rates of operative mortality and morbidity, as well as a greater chance for gross total removal (2, 7, 14, 23). For example, Lee et al. (14), reported an 87% gross total rate of resection with no mortality in 15 patients with global medial ridge tumors. Aggressive resection of tumor invading the cavernous sinus is associated with new and permanent cranial nerve palsies, cavernous ICA damage and/or sacrifice, and cerebrospinal fluid leak when the sphenoid or ethmoid sinuses are violated; these risks outweigh the benefits of gross total tumor removal. Therefore, we believe gross total resection of medial ridge meningiomas invading the cavernous sinus, which was present in 11 (31%) of our patients, is not warranted, especially in older patients (18). Regardless of the extent of resection, we recommend patients undergo serial postoperative MRI scans to provide early documentation of any tumor progression or recurrence. Although there is some evidence that cranial base meningiomas recur more frequently than convexity tumors (16, 17), we believe that our conservative approach to removing cavernous sinus tumor extension did not lead to a poor outcome in any of our patients (mean follow-up, 12.8 yr). Our indication for postoperative radiation included: 1) tumor recurrence or growth of residual extradural tumor on postoperative serial imaging, 2) atypical or malignant tumor pathologies, and 3) Simpson Grade IV tumor resections. Immediate postoperative radiation therapy for patients with subtotal benign meningioma resection is controversial (21) and was decided on a patient-by-patient basis. In our series, five patients (14%) were given postoperative radiation with successful tumor control, including three patients with radiographic tumor progression and two patients with partial intradural resections that were given radiation treatment before documented progression. Radiosurgery appears to be at least equally efficacious as external beam radiation for controlling small residual meningiomas involving the cavernous sinus or cranial base (10, 15, 25). In summary, a multidisciplinary approach using both tempered surgical resection and salvage

NEUROSURGERY

postoperative radiation treatment can maximize tumor control while minimizing patient morbidity.

CONCLUSION Resection of medial sphenoid ridge meningiomas has been associated with significant operative morbidity and mortality secondary to vascular injury. However, preoperative recognition of consistent unilateral involvement of the anterior cerebral circulation, with meticulous intraoperative preservation of these major arteries and their branches, can provide a successful and judicious microsurgical excision of the intracranial portion of these tumors. Because of their slow growth and the availability of advanced methods of radiotherapy, the surgeon must weigh the wisdom of attempted complete tumor removal in cases with invasion into the cavernous sinus, orbit, and bone, with its associated risks of significant neurological morbidity and mortality. An accurate classification of global medial sphenoid meningiomas is mandatory to gain insight into their clinical behavior and also for understanding the long-term efficacy and safety of the available treatment options.

REFERENCES 1. Abdel-Aziz KM, Froelich SC, Dagnew E, Jean W, Breneman JC, Zuccarello M, van Loveren HR, Tew JM Jr: Large sphenoid wing meningiomas involving the cavernous sinus: Conservative surgical strategies for better functional outcomes. Neurosurgery 54:1375–1384, 2004. 2. Al-Mefty O: Clinoidal meningiomas. J Neurosurg 73:840–849, 1990. 3. Benjamin V, Nazzaro J: Medial sphenoid ridge meningiomas, in Rengachary SS (ed): Neurosurgical Operative Atlas. Rolling Meadows, American Association of Neurological Surgeons, 1993, pp 285–297. 4. Benjamin V, Russell SM: Surgical management of tuberculum sellae and medial sphenoid ridge meningiomas, in Schmidek H, Roberts D (eds): Schmidek and Sweet’s Operative Neurosurgical Techniques. Philadelphia, Saunders Elsevier, 2005, pp 215–225. 5. Bonnal J, Thibaut A, Brotchi J, Born J: Invading meningiomas of the sphenoid ridge. J Neurosurg 53:587–599, 1980. 6. Cushing H, Eisenhardt L: Meningiomas: Their Classification, Regional Behavior, Life History, and Surgical End Results. Springfield, Charles C. Thomas, 1938. 7. Day JD: Cranial base surgical techniques for large sphenocavernous meningiomas: Technical note. Neurosurgery 46:754–760, 2000. 8. Fohanno D, Bitar A: Sphenoidal ridge meningioma. Adv Tech Stand Neurosurg 14:137–174, 1986. 9. Haines DE, Harkey HL, Al-Mefty O: The “subdural” space: A new look at an outdated concept. Neurosurgery 32:111–120, 1993. 10. Iwai Y, Yamanaka K, Ishiguro T: Gamma knife radiosurgery for the treatment of cavernous sinus meningiomas. Neurosurgery 52:517–524, 2003. 11. Kaye AH: Surgical management of clinoidal meningiomas. Neurosurgery 48:1019–1020, 2001. 12. Konovalov AN, Fedorov SN, Faller TO, Sokolov AF, Tcherepanov AN: Experience in the treatment of the parasellar meningiomas. Acta Neurochir Suppl (Wien) 28:371–372, 1979. 13. Kotapka MJ, Kalia KK, Martinez AJ, Sekhar LN: Infiltration of the carotid artery by cavernous sinus meningioma. J Neurosurg 81:252–255, 1994. 14. Lee JH, Jeun SS, Evans J, Kosmorsky G: Surgical management of clinoidal meningiomas. Neurosurgery 48:1012–1021, 2001. 15. Lee JY, Niranjan A, McInerney J, Kondziolka D, Flickinger JC, Lunsford LD: Stereotactic radiosurgery providing long-term tumor control of cavernous sinus meningiomas. J Neurosurg 97:65–72, 2002. 16. Mathiesen T, Lindquist C, Kihlstrom L, Karlsson B: Recurrence of cranial base meningiomas. Neurosurgery 39:2–9, 1996.



17. Mirimanoff RO, Dosoretz DE, Linggood RM, Ojemann RG, Martuza RL: Meningioma: Analysis of recurrence and progression following neurosurgical resection. J Neurosurg 62:18–24, 1985. 18. Nakamura M, Roser F, Jacobs C, Vorkapic P, Samii M: Medial sphenoid wing meningiomas: Clinical outcome and recurrence rate. Neurosurgery 58:626–639, 2006. 19. Ojemann RG: Meningiomas of the basal parapituitary region: Technical considerations. Clin Neurosurg 27:233–262, 1980. 20. Oka K, Rhoton AL Jr, Barry M, Rodriguez R: Microsurgical anatomy of the superficial veins of the cerebrum. Neurosurgery 17:711–748, 1985. 21. Peele KA, Kennerdell JS, Maroon JC, Kalnicki S, Kazim M, Gardner T, Malton M, Goodglick T, Rosen C: The role of postoperative irradiation in the management of sphenoid wing meningiomas. A preliminary report. Ophthalmology 103:1761–1766, 1996. 22. Pompili A, Derome PJ, Visot A, Guiot G: Hyperostosing meningiomas of the sphenoid ridge—clinical features, surgical therapy, and long-term observations: Review of 49 cases. Surg Neurol 17:411–416, 1982. 23. Risi P, Uske A, de Tribolet N: Meningiomas involving the anterior clinoid process. Br J Neurosurg 8:295–305, 1994. 24. Samii M, Tatagiba M: Surgical management of clinoidal meningiomas. Neurosurgery 48:1020, 2001. 25. Sibtain A, Plowman PN: Stereotactic radiosurgery. VII. Radiosurgery versus conventionally-fractionated radiotherapy in the treatment of cavernous sinus meningiomas. Br J Neurosurg 13:158–166, 1999. 26. Ugrumov VM, Ignatyeva GE, Olushin VE, Tigliev GS, Polenov AL: Parasellar meningiomas: Diagnosis and possibility of surgical treatment according to the place of original growth. Acta Neurochir Suppl (Wien) 28:373–374, 1979.

Acknowledgments We thank Alejandro Berenstein, M.D., who provided the endovascular management, as well as Mark Kupersmith, M.D., who conducted the neuro-ophthalmological examinations for the patients in this series. Their expertise undoubtedly contributed to the patient outcomes achieved.

COMMENTS

I

n this manuscript, Russell and Benjamin have reviewed their results of excision of medial sphenoid wing meningiomas and the long-term follow-up. In general, their results are excellent and commendable. The authors have followed a policy of removing all extracavernous tumor and treating the remainder by nonoperative means. They do not perform an orbital osteotomy, and decompress the optic nerve only if needed. Although the visual results of their patients were good, I wonder if routine optic nerve decompression before the tumor was removed would have improved the results. The authors have proposed their system of classification of these tumors, which is different from that of Al-Mefty and Cushing. I use a classification system based on the expected surgical difficulty, which is useful for treatment planning: 1) medial sphenoid wing meningiomas without cavernous sinus invasion (other than lateral wall) a) without arterial encasement, or b) with arterial encasement; and 2) medial sphenoid wing meningiomas with cavernous sinus invasion (with or without ICA encasement and narrowing). With regard to cavernous sinus invasion, I agree with the policy of leaving tumor behind if more than just the lateral wall is invaded. This is then treated with radiosurgery. For patients in whom radiosurgery has failed and the tumor is growing progressively, I then perform complete tumor excision after internal carotid artery bypass. The authors speculate that the draining veins may influence the patterns of tumor growth. This point is not proven by the authors, and I am skeptical about this theory. Laligam N. Sekhar Seattle, Washington

Dr. Alfred Velpeau at a Demonstration on a Cadaver, Augustin Feyen Perrin at Musée de la Ville. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.




PURELY ENDOSCOPIC RESECTION OF COLLOID CYSTS Jeremy D.W. Greenlee, M.D. Department of Neurosurgery, University of Iowa, Iowa City, Iowa

Charles Teo, M.B.B.S. Centre for Minimally Invasive Neurosurgery, Prince of Wales Hospital, University of New South Wales, Sydney, Australia

Ali Ghahreman, M.B.Ch.B. Centre for Minimally Invasive Neurosurgery, Prince of Wales Hospital, University of New South Wales, Sydney, Australia

Bernard Kwok, M.B.B.S. Centre for Minimally Invasive Neurosurgery, Prince of Wales Hospital, University of New South Wales, Sydney, Australia Reprint requests: Jeremy D.W. Greenlee, M.D., Department of Neurosurgery, University of Iowa, 200 West Hawkins Drive, Iowa City, IA 52242. Email:[email protected] Received, February 2, 2006. Accepted, May 1, 2007.

NEUROSURGERY

OBJECTIVE: To further assess the safety and long-term efficacy of endoscopic resection of colloid cysts of the third ventricle. METHODS: A retrospective review of a series of 35 consecutive patients (18 male, 17 female) with colloid cysts treated by endoscopic surgery was undertaken. RESULTS: The mean patient age was 32.4 years (range, 11–54 yr). Headache was the most common presenting symptom (22 patients). The average tumor size was 18 mm (range, 3–50 mm). The endoscopic technique could not be completed in six patients, necessitating conversion to an open craniotomy and a transcortical approach to the colloid cyst. All patients had histologically confirmed colloid cysts of the third ventricle, and complete resection of the lesion was confirmed macroscopically and radiologically in all patients. There were no deaths. Two patients developed aseptic meningitis without any permanent sequelae. One patient developed unilateral hydrocephalus attributable to obstruction of the foramen of Monro, which was treated with endoscopic septum pellucidotomy. The median follow-up period was 88 months (range, 10–132 mo). There was one asymptomatic radiological recurrence. No seizures occurred after surgery. CONCLUSION: The results of this study support the role of endoscopic resection in the treatment of patients with colloid cysts as a safe and effective modality. In some cases, conversion to an open procedure may be required. Additional follow-up will be required to continue to address the duration of lesion-free survival. KEY WORDS: Endoscopy, Surgical approach, Third ventricle, Tumor Neurosurgery 62[ONS Suppl 1]:ONS51–ONS56, 2008

C

olloid cysts of the third ventricle pose interesting management scenarios for the neurosurgeon, including the decisions of whether or not to treat (23, 24) and which treatment to use when treatment is necessary. These benign neoplasms can be treated either symptomatically, with cerebrospinal fluid diversion alone, or directly, with surgical resection. If a decision is made to remove the lesion, the aims then become complete macroscopic removal without long-term recurrence, restoration of cerebrospinal fluid communication pathways, and minimal morbidity and mortality related to treatment. The primary advantages of endoscopic surgery are variable angles of visualization and illumination compared with those for the operating microscope and minimal invasiveness (9, 27, 31). Such keyhole-style techniques can lead to shorter operative times, less postoperative discomfort, and shorter hospital stays (11, 20). The main disadvantages inherent in endoscopic techniques are the learning

DOI: 10.1227/01.NEU.0000297004.20222.ED

curve encountered to achieve proficiency, difficulties in using both hands simultaneously to operate, and frequent obscuration of the lens (28, 32). As a result, in part because of these disadvantages, the majority of neurosurgeons still use microsurgical open procedures to remove colloid cysts of the third ventricle. Endoscopic removal of colloid cysts has been documented to be efficacious and associated with low morbidity in small series (1, 5, 12–15, 18–20, 25, 26, 29). In some of these series, both purely endoscopic cyst removals and endoscope-assisted microsurgical removal are described; in others, subtotal resections are reported. However, given the relatively recent development of endoscopic techniques, longterm outcomes are not yet known. For example, the series cited above have an average mean follow-up period of 37 months, with only one series reporting patients beyond 5 years. We present this series of patients treated purely endoscopically to add to the longer-term data.


GREENLEE ET AL.

PATIENTS AND METHODS A retrospective review was conducted of a single surgeon’s (CT) series of patients harboring colloid cysts treated by endoscopic removal between 1993 and 2005. Inpatient and outpatient records were reviewed, and the following were recorded for analysis: age, sex, presenting symptoms, neurological findings, radiographic findings, surgical details, length of hospital stay, complications, and long-term outcome. Note was also made of the presence or absence of peri- or postoperative seizures because endoscopic resection of colloid cysts, like transcortical open procedures, requires a corticotomy. Follow-up was designated as the time from the date of surgery to the date of the last clinical evaluation and included magnetic resonance imaging. All patients underwent attempted endoscopic tumor removal using frameless stereotactic guidance and rigid neuroendoscopes and instrumentation (Aesculap Co., Tuttlingen, Germany). Tumors were approached via burr hole craniectomies located approximately 8 cm posterior to the nasion and 7 cm lateral to the midline. As this entry site is not a typical location for ventricular access, frameless stereotaxy is helpful for planning a trajectory to avoid the head of the caudate nucleus. A right-sided approach was used in all but one patient. A ventricular needle was introduced into the frontal horn to confirm a successful trajectory before introduction of the larger diameter endoscope and trocar (6.0 mm) with three parallel working channels. A peel-away sheath can be useful in some patients, owing to the frequent need for the removal and reinsertion of the endoscope from the ventricle. Additionally, in patients for whom cysts are removed intact, the entire endoscope, trocar, and cyst need to be withdrawn together as the cyst will not fit through the 2mm working channel. The cyst and its anatomic relationship to the surrounding structures were then inspected. Monopolar cauterization of choroid plexus overlying the cyst aided visualization and reduced bleeding from choroidal vessels during removal of the cyst wall. For large cysts, deflation of the cyst was necessary to see around and behind it. This was accomplished via cauterization of its surface and incision with sharp scissors followed by controlled, pulsed aspiration of the contents. Pediatric feeding tubes or custom-manufactured rigid suction cannulas worked well for this purpose. After the margins and attachments of the cyst were adequately visualized, the capsule was removed with grasping forceps, scissors, and cautery in various combinations. Bimanual cyst manipulation was possible by use of multiple working channels simultaneously and commercially available instrumentation. We used a two-surgeon technique and have not found rigid fixation of the endoscope and trocar helpful. The micromotion of manually holding the system aids in depth perception, given the two-dimensional view obtained through the endoscope. Bleeding during capsular removal was controlled with irrigation of warm, lactated Ringer’s solution. Cautery was necessary only rarely. After the capsule was removed, the roof of the third ventricle was inspected to ensure that there were no residua. External ventricular drains (EVDs) were not placed at the conclusion of the procedure.

RESULTS Thirty-five patients (17 female, 18 male) who met the inclusion criteria were identified. The average age at the time of surgery was 32.4 years (range, 11–54 yr). The first 20 patients in this series have been described in a previous report but are included here to provide details regarding long-term follow-up (30). The primary reason these patients sought care was headache, which was present in 71% of the patients at the time of presen-


FIGURE 1. Bar graph showing the breakdown of tumor size by number of patients. Inset bar indicates the number of patients with that tumor size who required conversion from the endoscopic technique to an open procedure.

tation. Transient visual obscurations were observed in six patients, and four patients had disturbances in level of consciousness. Two patients noted memory difficulties, two had dizziness, and one reported “occasional blackouts.” In six patients, cysts were found incidentally by referring primary care providers. However, during subsequent detailed neurological questioning, four of these patients were thought to have symptoms from the cyst. The remaining two patients elected for surgical treatment rather than observation. Excluding the three patients who presented with obtundation and required EVDs for treatment of severe hydrocephalus, the only abnormal finding at presentation was impaired shortterm recall in one patient. One patient in whom an EVD was used presented in extremis and awoke after placement of the EVD but had impaired recall before colloid cyst excision. Diagnostic neuroimaging revealed that the diameters of the cysts ranged in size from 3 to 50 mm (mean, 18 mm) (Fig. 1). Five patients were observed to have notable asymmetry between the sizes of the lateral ventricles. In 29 patients, the endoscopic technique was effective, and gross total removal of the cyst was achieved on the basis of intraoperative visual inspection and confirmed with postoperative imaging. In some patients, the cyst wall was opened, the contents were evacuated, and the wall was then removed. In other patients, particularly those with smaller cysts, the cyst was grasped and removed intact. In six patients, the procedure was converted to an open microscopic transcortical removal by fashioning a small craniotomy around the burr hole, and successful removal of the cyst was achieved. Excluding the six patients with these conversions and seven patients for whom data were not available, the operative time averaged 93 (⫾48) minutes. Hospital stays did not exceed 3 days in all but two patients (4 and 8 d). All patients were discharged directly to home. All patients have resumed their preoperative employment status with the exception of one patient who is receiving disability compensation because of short-term memory loss, which resulted from his moribund presentation before surgery. One failure of the endoscopic technique that required conversion to an open procedure occurred because of a technical mal-


ENDOSCOPIC COLLOID CYST RESECTION

Complications observed in the series included aseptic meningitis in two patients, and a trapped ipsilateral lateral ventricle that developed in one patient 1 month after surgery, which required endoscopic fenestration of a thin membrane covering the foramen of Monro and septum pellucidotomy. Aseptic meningitis was confirmed in both patients via lumbar puncture and routine cerebrospinal fluid studies.

DISCUSSION

FIGURE 2. Bar graph showing the length of follow-up for each patient. Stars denote the two patients who have been lost to ongoing follow-up.

function such that endoscopic cautery was not available. For the other five patients who required conversion to the open procedure, a correlation between cyst size and the need for conversion was not apparent (Fig. 1). These patients’ cysts were 9, 9, 10, 40, and 50 mm in diameter. In general, the degree of forniceal adhesion was the most common finding on the operative notes for patients who required conversion to an open procedure. That is, the cysts were thought to be so adherent to adjacent structures that the operating surgeon was uncomfortable with continuing the procedure endoscopically. This situation reflects the relative limitations of currently available endoscopic instrumentation and parallel working-channel trocar systems. In support of the observation of forniceal adhesion is the fact that both patients with preoperative symptoms of memory difficulties required conversion to an open procedure. Interestingly, in these two patients, preoperative asymmetry of the lateral ventricles was also observed on their initial scans. Perhaps as neuroimaging resolution continues to improve, cyst–forniceal relations may be better detected on preoperative scans to assist the surgeon in operative planning. The clinical outcome for the series included a median follow-up period of 88 months (mean, 76 mo; standard deviation, ⫾47 mo; range, 10–132 mo) (Fig. 2). Two patients were lost to ongoing follow-up. Headaches resolved in all but two patients. None of the patients had a new permanent memory deficit after surgery. In the two patients with preoperative impairment, improvement was not seen in one, and only mild improvement was seen in the other. One patient noted temporarily impaired short-term memory for 2 weeks after surgery. Formal neuropsychological testing was not routinely performed for patients in this series. None of the patients required a shunt and none had a clinical seizure. Follow-up neuroimaging has revealed only one small asymptomatic recurrence 2 years after surgery. No bleeding complications or infarctions have been noted on follow-up images.

NEUROSURGERY

Our purpose in presenting this series is not to discuss indications for the treatment of colloid cysts or to make a direct comparison of purely endoscopic versus surgical techniques for their removal. We present our results to provide long-term follow-up for what is now the largest series of patients with these cysts who were treated endoscopically. No reports in the current literature include such follow-up, and it is known that colloid cysts can recur in a very delayed fashion (22). The radiological outcomes of patients in this series are encouraging. With a median follow-up period of 88 months, we reported only a single cyst recurrence. Our series contains two distinct groups of patients, as seen in Figure 2: 18 patients have been followed-up for longer than 88 months without recurrence; follow-up is necessarily shorter in the rest of the patients, who have been treated more recently. These results extend endoscopic outcomes and compare favorably with results from other reports of endoscopic treatment and with those from open microsurgical procedures (2, 8). The efficacy of a purely endoscopic technique to achieve the goal of gross total colloid cyst excision is documented (3, 12, 15–17, 19, 20, 26, 30). These series demonstrate that complete removal can be accomplished, and recurrence, at least in the short-term, is rarely seen. To date, recurrences have been reported only in patients in whom there was a known residual cyst or in whom the cyst wall was not entirely resected and the remnants were coagulated (5, 15, 19). Variations in the use of endoscopic techniques to remove colloid cysts, as reported in the existing literature, include singleportal and biportal endoscopes, single-working-channel and multichannel endoscopes, and endoscopic visualization with bimanual manipulation. Furthermore, some endoscopists advocate cyst evacuation and capsular coagulation rather than radical wall excision (1, 6, 7, 15, 25). Our goal was radical wall excision to prevent late recurrence (Figs. 3 and 4). In rare situations, endoscopic procedures may need to be abandoned with conversion to an open microsurgical resection. In this series, we found no correlation between tumor size and the decision to abandon the endoscopic technique. This need was associated solely with the degree of forniceal adhesion. Adjacent venous structures did not affect the decision for conversion, and there were no bleeding sequelae, even when cysts were grasped and removed intact. Nevertheless, there is a clear need for improved endoscopic instrumentation to allow precise, bimanual dissection. As with any surgical technique, judgment must always be exercised on the basis of all available information, including


GREENLEE ET AL.

A

B

FIGURE 3. A, preoperative coronal T1-weighted postgadolinium magnetic resonance imaging (MRI) scan demonstrating a large cyst causing asymmetric enlargement of the lateral ventricles. B, postoperative image showing no residual cyst and a diminution in ventricular size.

plications were observed. Surgical treatment of colloid cysts of the third ventricle entails a risk of subsequent memory impairment, regardless of the technique used. Published data from series of patients treated with open microsurgery describe memory deficits in up to 26% of the patients (4, 8, 10, 21). Similarly, memory deficits are reported after endoscopic resections (1, 12, 13, 15, 19, 20, 25, 26, 30, 33). Fortunately, in most patients, these deficits are transient. Further studies with detailed neuropsychological testing pre- and postoperatively for these patients will be helpful as we continue to refine our techniques to best serve our patients. These tests are necessary to explore the possibility that the improved visualization offered by angled endoscopes in assessing the cyst–forniceal interface may translate into lower treatment-related memory impairment compared with transcallosal approaches.

CONCLUSION A

B

This series demonstrates that endoscopic removal of colloid cysts is a minimally invasive way to achieve total cyst removal with a low risk of recurrence and low surgical morbidity. Additional follow-up will be necessary to ensure continued long-term radiological and clinical success.

REFERENCES

FIGURE 4. A, preoperative midsagittal T1-weighted MRI scan indicating a small cyst (arrow) in the roof of the third ventricle just posterior to the foramen of Monro. B, corresponding postoperative image showing no residual cyst (arrow).

both pre- and intraoperative data. For example, both patients in this series with preexisting memory difficulties were revealed to have cysts tightly adherent to the fornix, and conversion to an open procedure was necessary in both. Neither deficit resolved completely after surgery. The anatomic relationship of the cyst to the fornix can be confirmed intraoperatively after initial inspection and cyst manipulation (Fig. 5). Our data demonstrate that colloid cysts can be removed endoscopically with low morbidity, irrespective of cyst size. Specifically, no permanent treatment-induced com-

FIGURE 5. Anterolateral endoscopic view of the right foramen of Monro showing a cyst (CC) with an adherent interface (I), including blood vessels continuing from cyst wall onto septum (small arrow), and a free interface (large arrows) with the fornix (F). SV, septal vein; CP, choroid plexus; T, thalamus.


1. Abdou MS, Cohen AR: Endoscopic treatment of colloid cysts of the third ventricle. Technical note and review of the literature. J Neurosurg 89:1062– 1068, 1998. 2. Cabbell KL, Ross DA: Stereotactic microsurgical craniotomy for the treatment of third ventricular colloid cysts. Neurosurgery 38:301–307, 1996. 3. Caemaert J, Abdullah J: Endoscopic management of colloid cysts. Tech Neurosurg 1:185–200, 1996. 4. Camacho A, Abernathey CD, Kelly PJ, Laws ER: Colloid cysts: Experience with the management of 84 cases since the introduction of computed tomography. Neurosurgery 24:693–700, 1989. 5. Decq P, Le Guerinel C, Brugières P, Djindjian M, Silva D, Kéravel Y, Melon E, Nguyen J: Endoscopic management of colloid cysts. Neurosurgery 42:1288–1296, 1998. 6. Deinsberger W, Böker DK, Bothe HW, Samii M: Stereotactic endoscopic treatment of colloid cysts of the third ventricle. Acta Neurochir (Wien) 131:260–264, 1994. 7. Deinsberger W, Böker DK, Samii M: Flexible endoscopes in treatment of colloid cysts of the third ventricle. Minim Invasive Neurosurg 37:12–16, 1994. 8. Desai KI, Nadkarni TD, Muzumdar DP, Goel AH: Surgical management of colloid cyst of the third ventricle—A study of 105 cases. Surg Neurol 57:295–304, 2002. 9. Fratzoglou M, Leite dos Santos AR, Gawish I, Perneczky A: Endoscopeassisted microsurgery for tumors of the septum pellucidum: Surgical considerations and benefits of the method in the treatment of four serial cases. Neurosurg Rev 28:39–43, 2005. 10. Friedman MA, Meyers CA, Sawaya R: Neuropsychological effects of third ventricle tumor surgery. Neurosurgery 52:791–798, 2003. 11. Fries G, Perneczky A: Endoscope-assisted brain surgery: Part 2—Analysis of 380 procedures. Neurosurgery 42:226–232, 1998. 12. Gaab MR, Schroeder HW: Neuroendoscopic approach to intraventricular lesions. J Neurosurg 88:496–505, 1998. 13. Gonzalez-Martinez JA, Zamorano L, Li QH, Diaz FG: Interactive imageguided management of colloid cysts of the third ventricle. Minim Invasive Neurosurg 46:193–197, 2003. 14. Harris AE, Hadjipanayis CG, Lunsford LD, Lunsford AK, Kassam AB: Microsurgical removal of intraventricular lesions using endoscopic visualization and stereotactic guidance. Neurosurgery 56 [Suppl]:125–132, 2005.


ENDOSCOPIC COLLOID CYST RESECTION

15. Hellwig D, Bauer B, Schulte M, Gatscher S, Riegel T, Bertalanffy H: Neuroendoscopic treatment for colloid cysts of the third ventricle: The experience of a decade. Neurosurgery 52:525–533, 2003. 16. Horváth Z, Vetö F, Balás I, Dóczi T: Complete removal of colloid cyst via CTguided stereotactic biportal neuroendoscopy. Acta Neurochir (Wien) 142:539–546, 2000. 17. Jho HD, Alfieri A: Endoscopic removal of third ventricular tumors: A technical note. Minim Invasive Neurosurg 45:114–119, 2002. 18. Kehler U, Brunori A, Gliemroth J, Nowak G, Delitala A, Chiappetta F, Arnold H: Twenty colloid cysts—-Comparison of endoscopic and microsurgical management. Minim Invasive Neurosurg 44:121–127, 2001. 19. King W, Ullman JS, Frazee JG, Post KD, Bergsneider M: Endoscopic resection of colloid cysts: Surgical considerations using the rigid endoscope. Neurosurgery 44:1103–1111, 1999. 20. Lewis AI, Crone KR, Taha J, van Loveren HR, Yeh HS, Tew JM: Surgical resection of third ventricle colloid cysts. Preliminary results comparing transcallosal microsurgery with endoscopy. J Neurosurg 81:174–178, 1994. 21. Mathiesen T, Grane P, Lindgren L, Lindquist C: Third ventricle colloid cysts: A consecutive 12-year series. J Neurosurg 86:5–12, 1997. 22. Mathiesen T, Grane P, Lindquist C, von Holst H: High recurrence rate following aspiration of colloid cysts in the third ventricle. J Neurosurg 78:748–752, 1993. 23. Pollock BE, Huston J III: Natural history of asymptomatic colloid cysts of the third ventricle. J Neurosurg 91:364–369, 1999. 24. Pollock BE, Schreiner SA, Huston J III: A theory on the natural history of colloid cysts of the third ventricle. Neurosurgery 46:1077–1083, 2000. 25. Rodziewicz GS, Smith MV, Hodge CJ Jr: Endoscopic colloid cyst surgery. Neurosurgery 46:655–662, 2000. 26. Schroeder HW, Gaab MR: Endoscopic resection of colloid cysts. Neurosurgery 51:1441–1445, 2002. 27. Schroeder HW, Oertel J, Gaab MR: Endoscope-assisted microsurgical resection of epidermoid tumors of the cerebellopontine angle. J Neurosurg 101:227–232, 2004. 28. Schroeder HW, Oertel J, Gaab MR: Incidence of complications in neuroendoscopic surgery. Childs Nerv Syst 20:878–883, 2004. 29. Souweidane MM: Endoscopic surgery for intraventricular brain tumors in patients without hydrocephalus. Neurosurgery 57 [Suppl]:312–318, 2005. 30. Teo C: Complete endoscopic removal of colloid cysts: Issues of safety and efficacy. Neurosurg Focus 6:e9, 1999. 31. Teo C, Nakaji P: Neuro-oncologic applications of endoscopy. Neurosurg Clin N Am 15:89–103, 2004. 32. Teo C, Rahman S, Boop FA, Cherny B: Complications of endoscopic neurosurgery. Childs Nerv Syst 12:248–253, 1996. 33. Tirakotai W, Schulte DM, Bauer BL, Bertalanffy H, Hellwig D: Neuroendoscopic surgery of intracranial cysts in adults. Childs Nerv Syst 20:842–851, 2004.

COMMENTS

T

his report by Greenlee et al. addresses the long-term recurrence rate of colloid cysts after endoscopic resection. The authors report an impressive recurrence rate of only 3.5% with a mean follow-up of 76 months. Emphasis on surgical philosophy is important, however, in the context of these results. The surgical intent described is complete cyst removal with reliance on a craniotomy when total removal is not possible, an event that accounted for 17% of the reported cases. The approach espoused by the authors is based upon their belief in a hitherto unproven tenant: that recurrence rates are higher when cyst remnants remain after endoscopic resection. Although the concept that total resection is better for controlling tumor recurrence rates is intuitive, the published literature on endoscopic colloid cyst resection is not so definitive. In the 2003 report by Hellwig et al. (2) only 1 recurrence was noted in 20 patients (mean follow-up 64 months) in which “parts of cyst membrane were left behind in all.” On the basis of the current published literature, there is no statistically sound basis for drawing

NEUROSURGERY

the conclusion that cyst remnants that are controlled with coagulation will result in higher recurrence rates. The possible fate of the six patients in whom procedures were converted from an endoscopic resection remains unknown. If the authors instead evacuated the cyst, removed as much of the cyst wall as possible, and used coagulation on any adherent remnants, it is unlikely that recurrence would have been seen. Irrespective of recurrence rates, complete colloid cyst removal also needs to be tempered by the possibility of causing injury when, as described by the authors, “cysts [are] grasped and removed intact.” Of all that I have learned in endoscopic intracranial surgery and applied in more than 100 tumor operations, exercising restraint is paramount. This concern is echoed by preeminent advocates of endoscopic colloid cyst resection in stating that “total or subtotal removal of the capsule . . . may be unnecessary and potentially dangerous” and “cyst evacuation as well as coagulation of the cyst wall seems to be a sufficiently effective alternative to microsurgical resection . . .” (1, 2). I congratulate the authors on excellent control rates over an impressively long follow-up period, but caution the reader contemplating endoscopic colloid cyst resection not to misinterpret the results as being dependent on the approach described. I agree with the philosophy of total removal but not at the expense of increasing risk to the patient. The low reported morbidity in this series is probable testament to a group who have extensive experience with endoscopic neurosurgery for intraventricular brain tumors. Mark M. Souweidane New York, New York

1. Abdou MS, Cohen AR: Endoscopic treatment of colloid cysts of the third ventricle: Technical note and review of the literature. J Neurosurg 89:1062–1068, 1998. 2. Hellwig D, Bauer BL, Schulte M, Gatscherr S, Riegel T, Bertalanffy H: Neuroendoscopic treatment for colloid cysts of the third ventricle: The experience of a decade. Neurosurgery 52:525–533, 2003.

G

reenlee et al. have reviewed a large series of endoscopically treated colloid cysts. The authors have actually strived for radical cyst removal instead of a decompression and a gross total or almost radical removal, which has been more common in endoscopic series. With attempted radicality, recurrence still occurred but only in 1 of 35 patients. One part of the series provides a long follow-up (>80 months) without late recurrences. The endoscopic technique has been refined as described by the authors; it has thus allowed good long-term disease control. The short-term results are also exceptionally good with short hospital stays, discharge directly home and return to previous employment for all except one patient. This is probably the first endoscopic series allowing a long-term assessment and proper comparison to good microsurgical series. The long-term control rate and outcome are well comparable to the zero recurrence rate and excellent outcomes in our transcallosal series, but short-term outcome and rapid return to previous activities seems superior. Naturally, continued follow-up will be necessary as I hope that the one recurrence was an occasional complication rather than an ominous sign of future recurrences; the authors’ method to assure radicality is superior to previously published practices and makes long-term control probable. Despite a transcortical route, the reported epilepsy rate was lower than the expected 5% that is usually quoted for frontal trajectories. It is possible that the figures partly reflect a lack of systematic search for epilepsy because mere clinical follow-up risks omission of some late complications.


GREENLEE ET AL.

Finally, it is necessary to point out that these results probably require exceptional endoscopic skill as results of this quality are rare. And, despite good endoscopy, good microsurgery was also necessary: more than 10% of the procedures needed to be converted to open microsurgery. An important safety issue is the fact that centers dealing with this pathological condition also possess expertise in microsurgical treatment of foraminal cysts. These lesions are rare and benign and affect young adults. It would seem reasonable to have subspecialized treatment reach consistent results with either endoscopy or microsurgery. Tiit Mathiesen Stockholm, Sweden

T

his article fits perfectly with the actual and ongoing discussion: can and should colloid cysts of the third ventricle (“Monroi cysts”) be treated purely endoscopically with a transcortical burr hole puncture approach, or should a microsurgical procedure, perhaps assisted by endoscopy, be preferred (e.g., with a transcallosal midline approach)? The authors describe a relatively high number of 35 patients treated only with the endoscope (the largest number published to date). The results are encouraging and favor endoscopy: in 39 patients, successful surgery was performed with the endoscope alone; in 6 patients the endoscopic procedure had to be converted to microsurgery, apparently via the same (transcortical) approach by enlarging the burr hole to a small craniotomy. In particular, the authors did not experience significant complications owing to the use of the endoscope, and the conversion to microsurgery apparently did not cause any disadvantage for the patients. A small cyst recurrence was seen in only 1 patient; however, the mean follow-up time of 88 months should not be misconstrued: the range from 10 to 132 months does not exclude significant recurrence in some patients within a few years.

Thus, this article corresponds to our experience in 32 cases: the endoscopic technique allows minimally invasive and, in most patients, optically complete resection of colloid cysts of almost all sizes and in any location within the foramen of Monroi and inside the third ventricle. The approach with a simple puncture is less invasive than the microsurgical technique with retractors, and the imaging is far better than with the microscope, including “around the corner,” e.g., for checking membrane remnants inside the third ventricle. However, as the authors admit, the technique needs further improvement: dissection through a small tube–guided uniportal approach is difficult and time consuming, and the limitations of hemostasis are the most frequent reason for conversion to microsurgery. The instrumentation used by the authors should also be discussed: we limit the use of scopes with only 2-mm instrument channels to less complex procedures such as ventriculostomy and septostomy. The minimum manipulation size for tumors and such vascularized cysts in our opinion should be approximately 3 mm (“space scope–type” in an approach tube of ≥6mm inner tube size, not a scope with channels; then peel-away sheaths are not required for frequent “in-and out,” but may be useful for aspiration of cyst contents through the scope tube, and the soft cyst membranes can be pulled out with the scope, leaving the tube in place). An additional small flexible instrument can be used for dissection with fixing of the cyst membrane with the flexible forceps. In our experience with space scope-types, the conversion rate to microsurgery was less than 10 % in our last 27 procedures (2 conversions only). Certainly the endoscopic technique for colloid cysts has left the experimental phase and is at least equivalent to but less invasive than microsurgery. Michael R. Gaab Hannover, Germany

Michelangelo’s Study of Anatomy, (1885), Antonin Mercié. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.


Amin B. Kassam, M.D. Departments of Neurosurgery and Otolaryngology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Daniel M. Prevedello, M.D.

ENDOSCOPIC ENDONASAL PITUITARY TRANSPOSITION FOR A TRANSDORSUM SELLAE APPROACH TO THE INTERPEDUNCULAR CISTERN

Departments of Neurosurgery and Otolaryngology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

OBJECTIVE: The interpeduncular cistern, including the retroinfundibular area, is one of the most challenging regions to approach surgically. The pituitary gland and the infundibulum guard the region when an endonasal route is undertaken. Superior transposition of the pituitary gland and infundibulum is described as a functional means to access this complex region through a fully endoscopic, completely transnasal route. METHODS: Ten consecutive patients in whom a pituitary transposition was performed during an expanded endonasal approach at the University of Pittsburgh Medical Center for resection of retroinfundibular lesions were reviewed. The series consisted of seven men and three women with a mean age of 44.4 years. Pathology consisted of four craniopharyngiomas, four chordomas, and two petroclival meningiomas. RESULTS: Five patients (50%) underwent total resection of the tumor, three patients (30%) underwent near total resection (95% removal), and two patients (20%) had partial resection of petroclival meningiomas with the goal of optic apparatus decompression. All four patients with visual deficits recovered their vision completely. There was no neurological deterioration. Eight patients had normal pituitary function preoperatively, seven of whom (87.5%) had confirmed function preservation postoperatively, with one of these patients experiencing transient diabetes insipidus. The remaining patient with a hypothalamic craniopharyngioma underwent complete resection with obligatory panhypopituitarism and diabetes insipidus. CONCLUSION: Endoscopic endonasal transposition of the pituitary gland and its stalk can provide a valuable corridor to the retroinfundibular space and interpeduncular cistern with pituitary function preservation in the majority of patients. This approach should only be pursued once significant experience with endoscopic endonasal approaches has been acquired.

Ricardo Carrau, M.D.

KEY WORDS: Chordomas, Craniopharyngiomas, Endoscopic cranial base surgery, Expanded endonasal approaches, Hypophysiopexy, Interpeduncular cistern, Pituitary transposition, Transsphenoidal

Department of Neurosurgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Ajith Thomas, M.D. Department of Neurosurgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Paul Gardner, M.D. Department of Neurosurgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Arlan Mintz, M.D. Department of Neurosurgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania

Carl Snyderman, M.D.

Departments of Neurosurgery and Otolaryngology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania Reprint requests: Amin B. Kassam, M.D., Department of Neurosurgery, 200 Lothrop Street, PUH B400, Pittsburgh, PA 15213. Email: [email protected] Received, March 8, 2007. Accepted, May 23, 2007. ONLINE DIGITAL VIDEO

NEUROSURGERY


T

here are few places in the human body that are more compact and complex than the space located directly behind the pituitary gland and infundibulum. The anatomic boundaries of this consist of the optic apparatus and the anterior recess of the third ventricle superiorly, the mamillary bodies and interpeduncular cistern with the basilar artery and posterior cerebral arteries posteriorly, and the posterior communicating artery with its perforators along with the oculomotor nerve bracketing the region laterally (13, 19, 22). The entire region is then guarded anteriorly by the

DOI: 10.1227/01.NEU.0000297013.35469.37

sella and its contents, including the pituitary gland, infundibulum, and the basilar plexus between both cavernous sinuses (Fig. 1) (22, 24, 39, 62). Therefore, accessing this region has proved to be a challenge independent of the approach taken. Conventional anterolateral access to the region through various cranial base approaches has been described, but each of these requires transgressing the lateral contents of the interpeduncular cistern, particularly the oculomotor nerve and posterior communicating artery (19, 28, 33, 42, 47, 53, 57, 59). Despite significant and extensive dissections,


KASSAM ET AL.

the final working space often becomes relatively narrow, representing a small corridor between the optic nerve/tract, Cranial Nerve III, and the branches of the posterior communicating artery. Particularly challenging is visualization of the ipsilateral region directly located under the optic nerve and tract (1, 19). These issues become even more paramount when having to manipulate an optic apparatus that may be ischemic as a result of chronic compression from tumors and lesions located in this space, which can be very intolerant. The cranial–caudal angle approach has a substantial limitation with the optic apparatus positioned between the surgeon and the target. To improve corridors around the parachiasmatic space, critical positioning of the chiasm needs to be considered, specifically whether or not the chiasm is positioned anteriorly (prefixed) or posteriorly (postfixed) (5, 9, 10, 16, 22). To minimize optic apparatus manipulation, a caudal–cranial angle of attack seems ideal. Two corridors facilitate this oblique perspective: a lateral transpetrosal and a midline transsphenoidal approach (1, 19, 22, 37, 39). The lateral corridors rely on many small corridors in between vital structures that form the lateral walls of the cisterns. Alternatively, although the midline transsphenoidal route has the significant advantage of approaching midline lesions from a midline view, thereby avoiding all the important lateral structures and creating a direct infraoptic corridor, it has a signifi-

cant disadvantage: the pituitary gland and infundibulum guard the interpeduncular cistern. Given the four fixed walls of the interpeduncular fossa by critical neurovascular structures discussed previously, tumors in this region are completely guarded along all perimeters by vital structures. Therefore, a decision needs to be made as to which structures are most likely to tolerate manipulation and, in the event that function is compromised, what are the long-term sequelae for that patient. We believe that the pituitary gland is more tolerant of mobilization and, in the event of pituitary dysfunction, can be more readily managed with endocrine replacement. Therefore, we pursued endoscopic transposition as a route to the interpeduncular fossa. In this report, we describe the technical nuances of creating a transdorsum sellae endoscopic approach with transposition of the pituitary gland and infundibulum to gain access to the interpeduncular fossa, and we report our early clinical experience and outcomes.

PATIENTS AND METHODS A retrospective review of the University of Pittsburgh Medical Center’s expanded endoscopic approach database (700 purely endoscopic endonasal procedures) was conducted. Ten consecutive patients undergoing a pituitary transposition were identified, and all of the pertinent records were reviewed (Table 1).

TABLE 1. Ten patients who underwent pituitary transpositiona,b Patient no.

Pathology

Age (yr)/sex

Presenting symptoms

Resection

Septal Symptoms flapc postoperatively

Preoperative Postoperative Postoperpituitary pituitary ative DI function function

Complications

1

Craniopharyngioma

44/M

Tumor growth

Total

Yes

Asymptomatic

Normal

Panhypopit

DI

No

2

Craniopharyngioma

68/F

Bitemporal hemianopsia

Radical subtotal

Yes

Vision normalized

Normal

Normal

No DI

No

3

Petroclival Meningioma

48/M

Headaches

Partial (decompression)

Yes

Asymptomatic

Normal

Normal

No DI

No

4

Petroclival Meningioma

58/F

Superior bitemporal quadrantopia

Partial (decompression)

No

Vision normalized

Normal

Normal

Transient DI CSF leak

5

Craniopharyngioma (recurrent)e

45/M

Right temporal visual deficit

Radical subtotal

Yes

Vision normalized

Panhypopit

Panhypopit

Previous DI No

6

Chordoma

16/M

Headaches

Total

Yes

Asymptomatic

Normal

Normal

No DI

Transient VI CSF leak

7

Chordoma

33/F

Right VI nerve palsy

Radical subtotal

Yes

Normalization

Normal

Normal

No DI

Recurrence, no leak

8

Chordoma

18/M

Headaches

Total

Yes

Asymptomatic

Normal

Normal

No DI

CSF leak

9

Chordoma (recurrent)e

36/M

Right III nerve palsy

Total

Nod

Persistent right III nerve palsy

Panhypopit

Panhypopit

No DI

CSF leak

10

Craniopharyngioma

78/M

Superior left homoTotal nymous quadrantopia

No

Vision normalized

Normal

Normal

No DI

CSF leak

a

DI, diabetes insipidus; Panhypopit, panhypopituitarism; CSF, cerebrospinal fluid. The patients are listed consecutively (most recent case is Patient 1). Presence of nasal septum flap decreased the chance of postoperative CSF leak (P 0.08). d Unsuccessful nasoseptal flap as a result of technical limitations. e Previous surgery in outside institution. b c



TRANSDORSUM SELLAE APPROACH TO THE INTERPEDUNCULAR CISTERN

Surgical Technique

Initiating the Approach Patients are positioned supine on the operating table in Mayfield head holders and attached to the image guidance system with the head in a neutral position and slightly rotated to the right side. The nose is prepared with pledgets soaked with 0.02% oxymetazoline into each naris followed by povidone-iodine solution applied over the nose and upper lip as well as into each nare on a cotton-tipped applicator. The procedure is initiated in the right nare with the removal of the middle turbinate. A posterior septectomy is performed after disarticulating the rostrum from the sphenoid bone, creating a bilateral binarial exposure opening to allow more freedom of movement (26). The middle nare turbinate of the left naris is generally lateralized and not resected. The natural sphenoid ostium on the left side is opened and widened to form one single cavity communicating the entire sphenoid sinus with the posterior aspect of both nasal compartments. The lateral margins of the anterior sphenoidotomies are extended to the level of the medial pterygoid plates and wide bilateral sphenoidotomies are performed. From this point, two surgeons perform the procedure simultaneously. The endoscope is generally positioned by one surgeon in the upper aspect of the right nostril while the other surgeon can use both hands to operate different instruments through both nostrils (binarial technique).

Sphenoid Exposure When the sphenoidotomy is widened enough to include the lateral recess of the sphenoid extending lateral to the carotid canal, the exposure is extended rostrally to expose the posterior cells of the ethmoid sinus to define the junction between the planum sphenoidale and the tuberculum sellae. Care should be taken not to transgress the posterior ethmoidal arteries anteriorly to preserve olfaction. The intrasphenoidal septations are carefully reduced because they often lead to the vertical segment of the cavernous internal carotid artery (ICA). The sphenoid sinus mucosa is removed, and the venous bleeding is controlled by irrigation with warm saline. Finally, the floor of the sphenoid sinus is drilled inferiorly and posteriorly until it is flush with the level of the clivus, allowing adequate space for an angled endoscope to be positioned inferiorly and aiming superiorly to facilitate tumor dissection. This creates the much sought caudal-to-rostral trajectory into the subchiasmatic and retrochiasmatic space.

Dura Mater Exposure and Relevant Anatomy Bone removal over the sellar face is initially extended laterally over the medial portions of each cavernous sinus and rostrocaudally exposing both the superior intercavernous sinus (SIS) and inferior intercavernous sinus (IIS). The tuberculum is thinned using a high-speed drill until the underlying SIS is seen through a small eggshell of residual bone. At this point, the tuberculum sella is then removed, often in one piece, containing both medial clinoids that provide direct access to both medial optic–carotid recesses. The medial clinoids are well described by Professor Rhoton and represent critical elements of this exposure because they represent the anterior–superior lateral border of the sella (48). We have previously reported on the critical role they play in accessing the anterior cranial base as the “key holes” to the cavernous sinus, sella, and opticocarotid cistern (24).

Specific Anatomic Considerations of the Dura In general, the dura mater has two distinct layers in the majority of the locations within the cranium. These layers must be considered separately when trying to understand the relationship of the dura to the surrounding structures along the cranial base. The dural layer in direct

NEUROSURGERY

FIGURE 1. Drawing showing a tumor (Tu) located in the retroinfundibular region of the interpeduncular cistern. Note the complex location of the lesion. The tumor is guarded posteriorly by the brain stem (BS), basilar artery (B), posterior cerebral arteries, and superior cerebellar arteries. Laterally it is surrounded by the optic nerves (ONs) and optic tracts (OT), anterior choroidal arteries, posterior communicating arteries, and III and IV cranial nerves (not shown in the figure). Superiorly the third ventricle (III-v) forms the roof of the region that is occasionally eroded by the lesion. Anteriorly the pituitary gland (PG) with the infundibulum (I) guards the region. The clivus (C) and sphenoid sinus (SS) are shown.

contact with the brain is referred to as the meningeal layer and the corresponding outer layer as the periosteal layer. Therefore, in regions of the cranial base where the dura is not covered by overlying bone, the periosteal layer is absent. This is best exemplified along the superior and lateral portions of the sella, where the lack of bone creates a very unique morphological arrangement of the dura mater. Over the lateral portion of the cavernous sinus, on each side, there is a meningeal layer along the sphenoid ridge. As this then spans medially traveling along the roof of the cavernous sinus and toward the sellar roof, the meningeal layer invaginates into the sella, forming a pouch. As the meningeal layer from both sides progresses centrally and begins to invaginate, a central oval aperture is formed through which the stalk eventually runs (46). Now given that the sella, is completely covered by bone anteriorly, posteriorly, and inferiorly along the sellar floor, the invaginating meningeal layer encounters the periosteal layer in these regions forming the dense double-layered dura mater of the sellar face, which often is interpreted as a single layer (46). Laterally, by virtue of the fact that there is no bone separating the pituitary fossa from the cavernous sinus, the periosteal layer is absent and therefore the meningeal layer alone separates the pituitary gland from the cavernous sinus. The superior, inferior, and posterior intercavernous sinuses are merely spaces formed in between the meningeal and periosteal layers that are present along the points of bony coverage along the sella. These spaces in these specific locations fill with blood forming the superior, inferior, and posterior intercavernous sinuses, thus allowing venous blood to pass from one cavernous sinus to the other. As an aside, this may explain why large venous lakes are in some instances encountered along the sellar face anteriorly during the initial opening. These may well represent anterior spaces between the dural layers that have formed persistent venous channels.


KASSAM ET AL.

FIGURE 2. Intraoperative view with a zero-degree endoscope showing the face of the sella and the planum sphenoidale after removal of the overlying bone. Note the critical landmarks: lateral optic–carotid recesses (LOCR), ICA, the left and right optic nerves (LON, RON), and the middle optic recesses (mOCR). The dura (D) of the suprasellar compartment was opened above the SIS, and the arachnoid (A) was preserved. The opening of the sellar dura is initiated under the SIS, and the pituitary gland (P) is shown.

FIGURE 3. Intraoperative view using a zero-degree endoscope demonstrating coagulation of the SIS with an endobipolar. The bipolars can be seen straddling the SIS located between the sella (S) below and the suprasellar cistern (A) above. Note the dura (D) overlying the cistern was opened without transgressing the arachnoid (A).

Pituitary Transposition Technique (see video at web site) An understanding of the specific anatomic considerations described previously is critical to perform the transposition. The dura mater over the tuberculum, the SIS, and the entire pituitary fossa are now exposed. The dura over the prechiasmatic cistern is opened in a cruciate fashion, taking care to avoid opening the prechiasmatic cistern and contaminating the subarachnoid space, and if at all possible preserving the suprasellar arachnoid membrane (Fig. 2). During the opening, the dura overlying the sella, formed by a dense junction of the two dural layers, is carefully opened along the midline to avoid transgressing a thinner underlying soft tissue layer that forms the pituitary capsule. Once this


FIGURE 4. Intraoperative view with a zero-degree endoscope illustrating the transaction of the previously coagulated SIS with endoscissors. The dura (D) of the suprasellar region was opened, and the arachnoid (A) preserved. At this point, the dura of the face of the sella is also incised with the pituitary gland (P) exposed. The bony resection extends laterally to expose the cavernous sinus (CS).

FIGURE 5. Intraoperative view with a zero-degree endoscope showing the two components of the dura mater located along the face and floor of the sella. The dura here is formed by an inner meningeal dura (MD) and an outer periosteal dura (PD). The intercavernous sinuses run in between both layers as the IIS shown in the picture. Once these layers reach the cavernous sinus, they bifurcate and only the meningeal layer forms the medial wall of the cavernous sinus (CS) along the lateral border of the sella. The pituitary gland is shown with a preserved pituitary capsule (PC).

plane is established, it is followed superiorly underneath the SIS. The SIS is ligated (clipped or preferably coagulated) with a bipolar cautery (Fig. 3) and then transected, communicating the suprasellar and sellar dural openings (Fig. 4). The sellar opening is completely widened laterally and inferiorly in a cruciate manner, allowing visualization of the entire anterior face of the gland. The IIS varies in caliber, and frequently it is very narrow along the midline. If present, it must be transected similarly. As the dissection progresses laterally, one can see that the dense dura of the sellar floor bifurcates in the direction of the cavernous sinus (Fig. 5). This



A

FIGURE 6. Intraoperative view with a zero-degree endoscope showing ligation of the inferior intercavernous sinus by hemoclips. Note the wide exposure required with bony removal undertaken to expose both cavernous sinuses (CS). This creates the complete exposure of the pituitary gland (P) required for transposition. fact is well appreciated when openings near the inferior–lateral corners of the sellar dura result in vigorous venous bleeding from projections of the cavernous sinus in between the two layers of dura. Consequently, the lateral aspect of the IIS often must be clipped rather than just coagulated (Fig. 6). The pituitary gland is then released of the soft tissue attachments along the plane in between the dura that form the medial cavernous sinus wall and the pituitary capsule itself. This separation is performed in a sequence of blunt and sharp dissection. There are numerous fibrous projections connecting the pituitary capsule to the lateral sellar dura or medial cavernous sinus wall. We refer to these as the “pituitary ligaments” (Fig. 7, A and B), and they are analogous to the dentate ligaments that attach the lateral portion of the spinal cord to the dura. These ligaments must be systematically transected along the lateral contour of the gland along with the inferior hypophyseal and McConnel’s arteries, if present, to completely free the gland. It is important to emphasize that the critical step in performing the transposition is the initial opening and establishment of the plane between the layers of dura at the onset and preservation of the pituitary capsule. The pituitary capsule forms a thin but real layer that immediately covers and protects the pituitary gland (46). If an unadvised surgeon opens the pituitary capsule in hopes of seeking the inner dural layer, he or she has gone too deep into the gland; it then becomes very difficult when dissecting laterally to identify the ligaments (soft tissue connections) that run between the lateral portion of the capsule and connect it to the cavernous wall. Therefore, preservation of the pituitary capsule during transposition is essential to avoid gland damage and facilitate dissection. The projection of dura that covers the sella is densely attached to the superior aspect of the capsule of the pituitary gland, and it has to be divided and finally incised in a midline direction, opening the central aperture exposing the stalk (Fig. 8, A–C). The course of the superior and inferior hypophyseal arteries and their relationships with other sellar structures are also important to understand. The inferior hypophyseal artery emerges from the cavernous sinus segment of the ICA at the meningohypophyseal trunk and courses medially toward the gland. It travels within the cavernous sinus to then penetrate the medial wall posteriorly to supply primarily the posterior pituitary gland (29, 48, 60). The superior hypophyseal artery is a medial branch of the internal carotid artery in its paraclinoid seg-

NEUROSURGERY

B

FIGURE 7. A, endonasal cadaveric dissection through a zero-degree endoscope demonstrating a pituitary transposition. The left inferior hypophyseal artery (IHa) was transected. The pituitary gland (PG) is being rotated medially away from the cavernous sinus (CS). The pituitary capsule can be seen covering the underlying pituitary gland (PG). Soft tissue attachments “pituitary ligaments” (PL) can be seen connecting the PC to the CS. Endoscissors are shown transecting the PL along the left side of the sella. The clivus (C) is labeled along the midline for orientation proposes. B, correlative intraoperative endonasal view demonstrating mobilizing the right side of the PG. Note the PL tethering the gland to the CS are identified and released.

ment. It travels along the posterior portion of the SIS at the level of the pituitary aperture. It then travels medially and posteriorly in the direction of the sellar aperture, where it ramifies and supplies the stalk, gland, and chiasm (Fig. 9, A and B). Preservation of the superior hypophyseal artery is critical because it carries the primary supply to the gland (44, 46, 48, 60). Loss of the inferior hypophyseal artery is tolerated because of the anastomotic supply from the superior hypophyseal artery and along the stalk. Once the pituitary aperture is completely opened and the stalk freed, the gland is mobilized superiorly without any resistance. Once the transposition is complete, the infundibulum abuts the ventral


KASSAM ET AL.

A

B

FIGURE 8. A, intraoperative suprasellar view with a zero-degree endoscope showing the pituitary gland (PG) and the pituitary stalk (S) after the opening of the suprasellar and sellar dura, ligation of the SIS, and complete excision of the anterior dural fold (DF) that forms the pituitary aperture. The chiasm (Ch) is visualized superiorly and anteriorly. A small subchiasmatic perforator (SP) branch of the superior hypophyseal artery is shown. B, endonasal cadaveric dissection view using a zero-degree endoscope during a pituitary transposition is shown. At this point, the left lateral aspect of the pituitary gland covered by

surface of the optic chiasm, and the pituitary gland fills the suprasellar space and the prechiasmatic cistern. The gland is covered with fibrin glue to prevent desiccation and to hold its position in the suprasellar space.

Extradural Posterior Clinoidectomies and Dorsectomies (see video at web site) Once the pituitary gland is transposed, the entire posterior wall of the sella is visualized. The dura mater of the sellar floor is coagulated and dissected posteriorly and superiorly, exposing the dorsum sellae. This posterior dura contains the posterior intercavernous plexus (55, 59). This represents a very vascular region, and copious venous bleeding is often encountered during incision of this dura. This venous bleeding is controlled systematically using packing with hemostatic agents. Once complete thrombosis of this venous sinus occurs, then the underlying dorsum sellae and posterior clinoids can be directly exposed. The upper third of the clivus is drilled flat along with the dorsum sellae until it is eggshell-thin, bluelining the dura of the posterior fossa underneath. The residual thin bone of the dorsum is then removed with care to avoid injury to the ICA and to the abducens nerve in the region of Dorello’s canal laterally. The posterior clinoids are also drilled until they can be mobilized medially. The posterior clinoids are very densely attached laterally through the posterior petroclinoid ligaments and anteriorly through the interclinoid ligaments. Thus, for the completion of the removal of the posterior clinoids, sharp dissection to cut these ligaments is often required. It must be stressed at this point that the posterior clinoids must be detached from the posterior carotid canal before mobilization. This requires mobilization of the carotid arteries laterally within the medial boundaries of the cavernous sinus, exposing the underlying carotid canal. Using a high-speed drill with a 1-mm diamond drill bit, an osteotomy is created along the margin of the posterior carotid canal and the posterior clinoid, including a fracture line (Fig. 10, A and B) for subsequent mobilization of the posterior clinoid (Fig. 11). It is imperative


C

its capsule (PC) is totally disconnected from the cavernous sinus (CS) and the superior soft tissue adherence (AI) between the gland and the DF that forms the pituitary aperture is demonstrated. C, endonasal cadaveric dissection using a zero-degree endoscope is shown after releasing the pituitary gland from the DF that forms the aperture. It allows for a suprasellar view. The PG and the S after the opening of the suprasellar and sellar dura are shown. The Ch is visualized superiorly. The location of the left CS and the dorsum sellae (DS) are shown.

not to try to mobilize the posterior clinoid before creating this fracture; otherwise, the posterior carotid canal could move with the posterior clinoid, creating a laceration of the ICA in this segment. The bone resection in this region generates vigorous venous bleeding from the cavernous sinus and from its posterior extensions through the basilar plexus, which is controlled as described previously. Once the posterior clinoids are removed bilaterally and hemostasis is achieved, a complete corridor extending from the top of the planum sphenoidale down through the upper third of the clivus and laterally extending to both carotid arteries within the cavernous sinus is achieved (Fig. 12).

Intradural Dissection (see video at web site) Once the exposure is completed, the dura is opened, providing access to the prepontine and interpeduncular cisterns, providing for direct access to the structures behind Liliequist’s membrane (Fig. 13). The intradural dissection proceeds with strict adherence to the neurosurgical principles. These cisterns contain critical arterial perforating vessels from the posterior circulation along with the third and sixth cranial nerves. Identification of these structures as well as of the posterior communicating artery laterally is recommended before proceeding with the dissection when feasible (Fig. 14). The resection of retroinfundibular lesions is often proceeded with caution, often under the view of 45- or 70-degree endoscopes. The technique varies depending on the pathology of the case. The use of two suctions, applying traction and countertraction during the resection (Fig. 15), is the most commonly used technique to debulk the tumors. Very calcified lesions are more challenging and may require the use of ultrasonic aspirators and/or diamond burr drill bits. If the tumor extends into the anterior recess of the third ventricle, it can then be followed through the interpeduncular fossa and infundibular recess above the level of the mammillary bodies and behind the optic chiasm until the third ventricle is opened. At this point, tremen-



A

A

B B

FIGURE 9. A, schematic drawing showing the sellar region in a frontal view. The pituitary gland (P) is demonstrated in the center attached to the medial wall of the cavernous sinus (CS) by the pituitary ligaments (PLs). The anterior dura covering the pituitary gland was removed, and the pituitary stalk was freed under the chiasm (Ch). The internal carotid arteries are shown on both sides. The inferior hypophyseal arteries (IHa) originate from the meningohypophyseal trunk of the ICA within the CS, and they travel medially and posteriorly to vascularize the inferior posterior third of the gland. The inferior hypophyseal arteries are ligated and cut along with the IIS and the PLs to allow the gland to be mobilized superiorly. The superior hypophyseal arteries (SHa) are preserved, and care should be taken when opening the dural fold of the aperture to avoid injuring them. B, endonasal cadaveric dissection using a zero-degree endoscope after releasing the pituitary gland from the dural fold (DF) that forms the aperture is shown. The SHa runs above the DF, and care should be taken at the last cut when opening the sellar aperture to avoid damaging the SHa. The CS, the IHa, the dorsum sellae (DS), and the clivus (C) are shown. The pituitary stalk (S) can be seen moved to the right side with the pituitary gland (PG) still being tethered by several PLs, preventing complete mobilization.

NEUROSURGERY

FIGURE 10. A, schematic drawing showing the sellar region after the pituitary gland (PG) is transposed superiorly. Before detaching the posterior clinoid process (PC), the dorsum sellae (DS) is drilled. The ICA is retracted laterally within the cavernous sinus (CS), allowing the surgeon to drill the carotid canal with a high-speed drill using a 1-mm diamond drill bit. The ligated IIS is shown. B, intraoperative view using a zero-degree endoscope showing the PC being thinned with a high-speed drill in between the DS and the medial wall of the cavernous sinus (RCS) at the level of the carotid canal. The pituitary gland (P) is transposed superiorly (from, Kassam A, Snyderman CH, Carrau RL, Mintz AH, Gardner PA, Thomas AJ, Prevedello DM: The Expanded Endonasal Approach to the Ventral Skull Base: Sagittal Plane. Tuttlingen, Endo-Press, 2007).

dous care must be exerted to avoid any lateral dissection because the walls of the hypothalamus can be directly seen (Fig. 16, A and B). The presence of the capsule and arachnoid planes will vary depending on the pathology and degree of invasion. The capsule is dissected from the neurovascular tissues. The plane in between the capsule and the arachnoid of the cisterns is identified and followed. Under direct visualization of critical neurovascular structures, the arachnoid bands covering the tumor capsule are sharply dissected with care to preserve the small perforators. It should be noted that sharp dissection is the only technique used in this region. It is our strong recommendation that if this cannot be pursued, the procedure should be terminated.


KASSAM ET AL.

FIGURE 11. Intraoperative view with a zero-degree endoscope showing the pituitary gland (P) transposed superiorly to expose the dorsum sellae (DS). The right posterior clinoid process (PC) is resected after detachment from the posterior carotid canal. The laterally retracted medial wall of the right cavernous sinus (CS) is shown.

FIGURE 12. Schematic drawing showing the direct view of the tumor (Tu) in the interpeduncular cistern that is obtained after the pituitary transposition. Note that the pituitary gland (PG) is elevated and fixed in place with fibrin glue. The dotted lines represent the bone that was removed during the approach (transplanum, transsellar, transclival approaches). The optic nerves (ONs), left optic tract (OT), third ventricle (III-v), brainstem (BS), basilar artery (B), clivus (C), and sphenoid sinus (SS) are shown.

Reconstruction (see video at web site) The defect is reconstructed by placing an initial subdural inlay graft between the brain and the dura mater. We favor the use of collagen matrix (Duragen; Integra Life Sciences, Plainsboro, NJ) because it has good tissue-handling properties and we use it to “reconstruct” the arachnoid layer. Our current technique uses a vascularized pedicle flap that we have previously described (18). This flap is based on the posterior nasal septal artery; it is harvested during the exposure and posi-


FIGURE 13. Intraoperative view using a 45-degree endoscope showing the contents of the interpeduncular cistern after the opening of the retrosellar dura. The basilar artery (B) can be seen at its upper level bifurcating into both posterior cerebral arteries (P1). The superior cerebellar arteries (SCA) originating immediately before the bifurcation. The right third cranial nerve (III) is seen at its origin in between the right P1 and SCA. The tumor (Tu) was invading the floor of the third ventricle. Some small perforators (SP) are seen originating from the left P1 (from, Kassam A, Snyderman CH, Carrau RL, Mintz A, Gardner P, Thomas A, Prevedello DM: The Expanded Endonasal Approach to the Ventral Skull Base: Sagittal Plane. Tuttlingen, Endo-Press, 2007).

FIGURE 14. Intraoperative view using a 45-degree endoscope showing the lateral border of the interpeduncular cistern on the left side. ICA subchiasmatic perforators that vascularize the left optic nerve (LON) and optic chiasm (Ch) are seen. The left posterior communicating artery (Pcomm) is seen joining the posterior cerebral artery (P1/P2 junction). A small perforator (SP) from Pcomm that supplies the left optic tract (LOT) is shown. After the origin of the anterior choroidal artery (covered by Pcomm), the ICA bifurcates into the middle and anterior cerebral arteries. The first segment of the anterior cerebral artery (A1) is seen as it travels over the genu of the left optic nerve/tract. Inferiorly, the basilar artery can be seen with the left superior cerebellar artery (SCA). The left third cranial nerve is seen in between the left SCA and P1.



A

FIGURE 15. Intraoperative view with a 45-degree endoscope positioned in the lower aspect of the sphenoid sinus where a clivectomy was performed aiming superiorly and showing the pituitary gland (PG) transposed. The tumor (Tu) is being resected using the two-suction technique. The right cavernous sinus (CS) is shown. tioned in the nasopharynx or inside the maxillary sinus during the resection. After resection, the nasoseptal flap hinged on the posterior nasal septal artery is then positioned over the dural edges of the defect. Care should be taken to ensure that the flap makes direct and complete contact with the bony edges to promote vascularization and a seal. Also, the flap should be oversized because it contracts over time (Fig. 17, A and B). A very thin layer of dural fibrin sealant is then used to coat the flap “picture frame.” Sealant should not be allowed to migrate between the flap and underlying bone because this will preclude direct contact and vascularization. A balloon is then positioned into the nasal cavity to provide a further buttress to mitigate against flap migration. The balloon consists of a 12-French Foley catheter that is passed from the left nostril across the residual portion of the nasal septum, and the distal portion is placed into the upper nasopharynx across the midline and filled with saline (23). The patient must be examined immediately postoperatively to ensure that the balloon is not causing excessive compression creating visual loss. In patients in whom a cerebrospinal fluid (CSF) leak is detected postoperatively, we advocate early reoperation to repair the leakage as soon as possible to minimize risk of infection. In patients in whom the ventricular system is opened, we use lumbar drainage to provide temporary CSF diversion.

Statistics A Fisher’s exact test was performed as a nonparametric analysis of the relevance of the nasal septum flap in decreasing the rate of CSF leak after the reconstruction of skull base defects based on the small size of the sample (n 10).

RESULTS The cohort consisted of seven men and three women (Table 1). The mean age of the patients was 44.4 years. Pathology consisted of four craniopharyngiomas, four chordomas, and two petroclival meningiomas (Table 2). Five patients (50%) had total resection of the tumor (Figs. 18 and 19), three patients (30%) had near-total resections (95%

NEUROSURGERY

B

FIGURE 16. A and B, intraoperative views with a 45degree endoscope positioned in the lower aspect of the sphenoid sinus where a clivectomy was performed aiming superiorly and showing the third ventricle of two different patients at the end of the tumor resection. Various anatomic structures can be seen in both images as such as the foramen of Monro (FM), the column of the fornix (CF), the anterior commissure (AC), the thalamic mass intermedia (MI), the choroid plexus (CP), and the walls of the hypothalamus (H). The patient in Figure 13 had hydrocephalus preoperatively; body of the fornix (F) is also identified at the level of the septum pellucidum. The internal cerebral veins (ICV) are also prominent and can be easily seen.

volume reduction), and the remaining two patients (20%) had subtotal resections of petroclival meningiomas. In this latter case of subtotal resections, in all patients, complete decompression of the optic apparatus and relief of preoperative symptoms were achieved (Table 2). One patient with a craniopharyngioma was asymptomatic preoperatively with documented tumor growth. All patients with visual deficits preoperatively enjoyed complete recovery postoperatively. One patient with a chondrosarcoma presented with a preoperative abducens nerve palsy. Postoperatively, he had complete recovery of the abducens palsy. The only persistent deficit was in a patient with a recurrent chordoma who had complete third nerve palsy preoperatively and no postoperative recovery.


KASSAM ET AL.

A

A

C

B

D

B

FIGURE 18. Sagittal (A) and coronal (B) preoperative magnetic resonance imaging (MRI) scans postcontrast showing a patient harboring a heterogeneously enhancing retroinfundibular lesion. C and D, postoperative MRI scans with contrast showing a complete resection of the lesion that proved to be an adamantinomatous craniopharyngioma. Sagittal (C) and coronal (D) images showing the pituitary gland (PG) preserved and transposed superiorly.

FIGURE 17. A, intraoperative view with a zero-degree endoscope at the end of the procedure showing when the surgeon is placing the nasoseptal flap (NSF) over the skull base defect. Because it is a vascularized flap, it allows for early adhesion on the inner cranium, preventing CSF leakage. B, an endoscopic view 5 days after the procedure showing the vascularized flap covering the skull base defect. Note that it is live tissue and it has reasonably contracted. The yellow line demarcates the edge of the flap on both images.

Eight patients had normal pituitary function preoperatively. Seven of them (87.5%) had confirmed preserved function postoperatively. One of these patients experienced transient diabetes insipidus (DI) treated with deamino-8-Darginine vasopressin during the hospital stay. The patient who permanently lost gland function had a craniopharyngioma. He was operated on in two stages, with the first stage representing the exposure and transposition and the second the resection. Stages were done 4 days apart. After the first stage, he required a single dose of deamino-8-D-arginine vasopressin on postoperative Day 1; no further treatment was necessary. His anterior gland function was preserved after the first-stage transposition. He underwent complete excision of his tumor, including the portion within the third ventricle. After the second stage, he developed panhypopituitarism and DI.


A

B

C

D

FIGURE 19. Sagittal (A) and coronal (B) preoperative MRI scans postcontrast showing a patient harboring a heterogenic enhancing lesion involving the clivus and expanding superiorly and to the right side of the interpeduncular cistern. C and D, postoperative MRI scans with contrast showing a complete resection of the lesion that proved to be a clival chordoma. Sagittal (C) and coronal (D) images showing the pituitary gland (PG) preserved and transposed superiorly.



TABLE 2. Outcomes in 10 patients who underwent pituitary transposition grouped by pathologya Number of patients

Total resectionb

Vision normalizationc

Craniopharyngiomas

4

2/3 (66.6%)

Chordomas

4

3/4 (75%)

Meningiomas

2

—

Gland function preservationd

CSF leak

3/3 (100%)

2/3 (66.6%)

1/4 (25%)

—

3/3 (100%)

3/4 (75%)

1/1 (100%)

2/2 (100%)

1/2 (50%)

a

CSF, cerebrospinal fluid. Based on intention-to-treat analysis. Specifically, no patients with petroclival meningioma or a recurrent craniopharyngioma were a target for gross total resection. Optic nerve decompression was the primary preoperative goal for these patients. c Only patients with preoperative visual impairment were considered for the denominator. d Only patients with preoperative normal pituitary function were considered for the denominator. b

TABLE 3. Pros and cons of different approaches to the interpeduncular cisterna Pros

a

Cons

Interhemispheric

Lamina terminalis can be opened Midline approach

Craniocaudal view View limited by the optic apparatus Frontal lobe manipulation

Pterional

Standard neurosurgical procedure Sylvian fissure can be split widely

Craniocaudal view Superior view limited by the optic apparatus Angled view of the interpeduncular cistern

Orbitozygomatic

Sylvian fissure can be split widely Improved angle of view with less brain retraction

Craniocaudal view Superior view limited by the optic apparatus Angled view of the interpeduncular cistern

Transcavernous

Sylvian fissure can be split widely Posterior clinoid process can be drilled and it improves the midline view of the prepontine cistern

Craniocaudal view Superior view limited by the optic apparatus Angled view of the interpeduncular cistern Cranial Nerve III is extremely manipulated

Transsylvian translimen insular and subtemporal

Exposes the crural and ambient cisterns Caudal–cranial view

Lateral view Temporal lobe is manipulated

Transpetrosal (combined middle fossa– posterior fossa)

Optimized caudal–cranial view

Lateral view Narrow corridors among Cranial Nerves II, III, IV, and V Pcom, anterior choroidal arteries, and perforators in the way

Transsphenoidal

Optimized caudal–cranial view

View limited by pituitary gland, stalk, and dorsum sellae Risk of CSF leak

Expanded transdorsum sellae with pituitary transposition

Optimized caudal–cranial view Wide view of the interpeduncular cistern

Endoscopic approach (not standard) Risk of CSF leak Risk of pituitary dysfunction

Pcom, posterior communicating; CSF, cerebrospinal fluid.

Two patients who previously had panhypopituitarism had persistent postoperative hormonal deficits needing anterior gland replacement. With regard to posterior gland function, one of these two patients had preoperative DI that persisted postoperatively. The other had intact posterior gland function that remained intact postoperatively. Seven patients had reconstruction of the cranial base using a nasal septum flap, of which two developed CSF leaks postoperatively. All three patients who did not have a vascularized flap for reconstruction of the cranial base defect had a CSF leak

NEUROSURGERY

in the postoperative period. Thus, a total of five patients underwent reoperation for CSF leaks, and no patient developed meningitis. The presence of a vascularized flap suggests a trend to diminish the likelihood of a postoperative CSF leak (P 0.08). With the relatively small sample size of this study, this trend will need to be validated in a larger series to see if statistical significance is reached. One patient had a new transient postoperative deficit. It was a transient VI nerve palsy in a patient with a clival chordoma that completely resolved; therefore, there were no permanent


KASSAM ET AL.

new deficits. One patient had recurrence of a chordoma and underwent a subsequent endoscopic procedure to address it with no complications.

DISCUSSION Cranial Approaches to the Interpeduncular Cistern By virtue of its deep location, narrow confines, and density of critically important vascular structures, the interpeduncular cistern represents one of the most challenging regions to be reached by any neurosurgeon (13, 57). There is a lack of consensus in the literature as to which approach represents the ideal means of accessing this region. As briefly presented in the introduction, there are several established approaches to the interpeduncular cistern, each with their pros and cons (Table 3). The pterional approach can prove to be restrictive in accessing the interpeduncular cistern. Even with a wide opening of the sylvian fissure, there are numerous critical structures interposed between the surgeon and the target. When a standard pterional approach is used, the primary barriers limiting adequate exposure of the interpeduncular cistern are the optic nerves and the chiasm. The deep corridors used during this approach include the opticocarotid, the interoptic, and the carotid oculomotor recesses; they provide access to the lower aspect of the interpeduncular cistern but can be limited when angling superiorly (13, 50). The addition of a cranial base approach such as an orbitozygomatic osteotomy may augment the angulation; however, the approach is still ultimately limited by the position of the optic apparatus and ensuing corridor (57). Sugita et al. (54) noted these issues and subsequently described the need to retract the optic tract and mammillary bodies to reach superior aspects of the interpeduncular cistern. As alternatives, several variations have been forwarded to improve the deeper exposure and available corridors. The carotid oculomotor recess can be expanded by drilling of anterior and posterior clinoid processes, unroofing the optic nerve, and opening the cavernous sinus with mobilization of the ICA (13, 33, 34). Nevertheless, these variations generate a relatively greater morbidity by virtue of their need to mobilize neurovascular structures (13). Most patients who undergo this procedure have transient third nerve palsy (13, 33, 34). These transcavernous approaches do improve the exposure to the lower interpeduncular cistern and the prepontine cistern, which, although of significant value in the case of vascular procedures providing proximal control of the basilar artery, do not compensate for the lack of superior exposure discussed previously. As a result, an alternative means of accessing tumors filling the interpeduncular cistern and extending into the third ventricle involves opening the lamina terminalis through a more anterior subfrontal or interhemispheric approach (11). However, this approach is conversely limited in accessing the inferior portion of the interpeduncular cistern (11, 40). To gain access to the posterior portion of the cistern, several modifications to the pterional approach have been suggested.


The transsylvian translimen insular approach, in which incisions along the mesial temporal structures and the uncinate fasciculus are necessary, and transchoroidal approaches represent such options (57). These approaches, when used to reach the interpeduncular fossa, carry the potential for significant temporal lobe dysfunction (43). In an effort to minimize the impact of these transtemporal approaches, a series of subtemporal approaches have been suggested. The anterior transpetrosal approach proposed by Harsh and Sekhar (20) and the petrosectomy proposed by Kawase et al. (28) can provide access to the prepontine cistern but, again, are limited by the deeper corridors guarded by the optics and oculomotor nerves, thereby again restricting rostral visualization. Furthermore, these approaches may be more limited medially toward the central prepontine and interpeduncular cisterns than the orbitozygomatic–transcavernous approaches (13). Hakuba et al. (19) proposed a further modification of the transpetrosal approach to compensate for this limitation. The posterior transpetrosal approach described by Hakuba et al. in 1985 provides an improved caudal–cranial angle, thereby improving access to the rostral portion of this region. However, adequate visualization of the interpeduncular cistern is only obtained when more aggressive drilling of the petrous bone is performed, threatening hearing and balance functions (56). Additionally, the approach still takes a lateral to medial trajectory. Therefore, a corridor between the optics and oculomotor nerve is still required. Furthermore, visualization of the component under the ipsilateral optic tract can prove to be difficult (19, 51, 56). In the pursuit of a midline corridor to access midline lesions, the transsphenoidal approaches have been advanced. The panoramic and angled views of the endoscope are exploited to create such midline corridors (17, 47).

Extended Transsphenoidal Approaches for Suprasellar Lesions Transsphenoidal surgery for supradiaphragmatic lesions extending into the suprasellar space was pioneered in the early 1980s by Edward Laws (35). Later, Martin Weiss published on the extended approach (8, 9) to achieve access to tumors with suprasellar extension by performing an anterior opening of the planum sphenoidale. The extended transsphenoidal route has been used by a select group of neurosurgeons to approach suprasellar pathologies located adjacent to the tuberculum sellae and planum sphenoidale in the suprasellar space (5, 9, 10, 16, 22, 30, 31, 37). However, the transsphenoidal approach achieved by the microscope is limited in parasellar exposure, particularly for lesions that extend far superiorly, laterally, or posteriorly into the perimesencephalic cisterns (5–7, 9, 10, 22, 30, 37). Some authors have suggested the use of an endoscope to augment a primary extended transsphenoidal microsurgical approach, emphasizing its importance in identifying important structures and thus facilitating fine dissection. However, many have noted that if it is used in conjunction with the microscope, it limits the working space within the speculum for



effective maneuverability of other instruments (5–7, 9, 10, 15, 16, 22, 30, 37).

Expanded Endonasal Purely Endoscopic Approach with Pituitary Transposition The use of the endoscope in cranial base surgery is evolving. As experience is gained, several critical problems such as reconstruction are being resolved, and limitations are being defined. Despite the use of an endoscope coupled with extensive bony removal around the sella turcica, the pituitary gland and its stalk still represent a natural anatomic barrier preventing access to the interpeduncular cistern. However, the superior visualization and anatomic detail afforded by the endoscope allow for the dissection of the pituitary gland and stalk, releasing it from the cavernous sinus and diaphragmatic aperture, respectively. This allows for the transposition of the gland from the sella, allowing for an unparalleled midline corridor into the entire interpeduncular cistern inferiorly and the anterior third ventricle superiorly. A pituitary transposition, or so-called hypophysiopexy, has been previously proposed as part of the resection of pituitary adenomas. It consists of placing an interpositional fat graft between the residual adenoma and the pituitary gland itself, thus potentially protecting the gland in the case of future radiation (4, 38). However, the concept of gland transposition to gain access to deeper underlying structures such as the interpeduncular fossa is unique. The primary goal of the transposition is to obtain adequate access and visualization to the retroinfundibular region. Admittedly, one could obtain the same view and corridor by simply removing the gland and undertake a transsellar approach. For specific types of lesions such as craniopharyngiomas, this may not be an unreasonable consideration because the likelihood of postoperative gland dysfunction is relatively high (62). However, a variety of different lesions may occupy this space such as the two meningiomas and the four chordomas shown in our series. In these situations, gland preservation is a realistic and potentially attainable goal, making transposition a viable option. Even in the case of craniopharyngiomas, gland preservation may still be a consideration depending on the specific site of origin and the individualized goals of surgery. There are situations in which the surgeon must weigh the pros and cons of performing a total transection and removal of the pituitary gland and/or its stalk. This may be even more germane in the pediatric population, in whom there may be a greater desire to preserve gland function. In fact, many have advocated attempted aggressive surgery for craniopharyngiomas but with the goal of stalk and gland preservation (11, 21, 39, 52, 53). If a decision is made to preserve the stalk, dissecting posterior to it becomes challenging independently of the approach selected. Kouri et al. (32) pointed out that the presence of a lesion behind the stalk is one of the major limitations of the extended transsphenoidal approach. We strongly believe that the use of angled endoscopes does not necessarily assuage these issues. In fact, such retroin-

NEUROSURGERY

fundibular dissections using an angled scope without transposition may be potentially dangerous. Specifically, the stalk creates a large pillar in the middle of the corridor, and the gland within the sella forces the surgeon into a small window between the posterior clinoid and the infundibulum. The ability to maneuver instruments becomes very restricted. As a result, there is a greater tendency to pursue blind dissection with curettes and grasping forceps. The consequences can be disastrous. The pituitary transposition represents a logical solution, widening the corridor and allowing for critical dissection of the contents of the interpeduncular cistern under direct visualization. Admittedly, transposition can be extremely challenging, requiring the management of significant venous bleeding; however, the view and corridor obtained are well worth the effort.

Clinical Outcome Measures

Visual Outcomes As previously discussed, the corridor when traveling from conventional lateral approaches is guarded by the visual apparatus; therefore, these approaches are associated with an incidence of postoperative visual decline (14, 54, 57). The incidence of this visual deterioration varies in the literature on the basis of approach used and pathology treated. Fahlbusch et al. (11), emphasizing that care must be taken to avoid damage to the optic pathways during retrochiasmatic removal of craniopharyngiomas, reported 20% of visual deterioration after the bifrontal approach. Van Effenterre and Boch (58) documented a 15% incidence of visual decline postoperatively in craniopharyngiomas. The postoperative visual deterioration in suprasellar meningiomas has been correlated to damage the subchiasmatic perforators and reported to range from 14 to 20% (10, 12, 41, 45, 49, 61). In response, some authors have advocated bicoronal craniotomies with bilateral orbitotomies (anterior transbasal approach) and drilling of the planum sphenoidale to reach the sphenoid sinus allowing for a more caudal trajectory to the chiasm, thus enabling improved visualization in hopes of preserving the subchiasmatic perforators (2). Essentially, this is intended to create a midline corridor. The extended transsphenoidal approach assisted by angled endoscopes has shown a general improvement in visual outcomes for suprasellar disease, perhaps explained by these considerations (10, 15, 16, 37, 39). We believe that the visual outcomes reported in our series, specifically complete recovery in all patients with no incidence of deterioration, are likely related to two primary advantages afforded by the expanded endonasal approach: 1) optimized angle of view of the interpeduncular cistern (caudally and rostrally) allowed by the pituitary transposition without any retraction or manipulation on the optic apparatus; and 2) direct endoscopic visualization of the subchiasmatic perforators, allowing for adequate dissection and protection. We also understand that this is a preliminary study with a relatively small number of patients; larger series are needed to verify these findings.


KASSAM ET AL.

Endocrinological Outcomes Obviously, this is the primary question associated with pituitary transposition. As discussed previously, there is a consideration of simple pituitary transection and exenteration of the contents of the pituitary fossa to facilitate a transsellar approach. We consider this a relatively nihilistic option. As we have demonstrated in this study, pituitary function can be preserved and is more contingent on the goals of surgery and pathology being treated rather than the need to transpose the gland. In the case of noncraniopharyngioma lesions, we were able to preserve the gland in 100% of patients (five of five). In the case of craniopharyngiomas, we were able to preserve gland function in 66% (two of three). The most important determinant of gland function in our experience proved to be the goal of surgery. Specifically, in some situations, a pituitary stalk transection or hypothalamic dissection may be needed. A decision regarding the complete resection of the lesion can only be made in the operating room. Therefore, in the case of retroinfundibular craniopharynigomas, there may be value in transposing the gland at the onset. If an intraoperative decision is made to leave residuals, based on the inability to separate the tumor from the hypothalamus for instance, then gland function may well be preserved. Even in cases in which resecting craniopharyngiomas with the goal of pituitary stalk preservation is advocated, the incidence of postoperative panhypopituitarism is significant and the transsphenoidal route has shown relatively better results than transcranial approaches (3, 5, 9, 11, 15, 16, 21, 22, 36, 39, 53, 58). This becomes an even greater challenge in situations in which the pituitary gland is displaced anteriorly and inferiorly displaced by a retroinfundibular craniopharyngioma, which is reported to occur in 47% of the cases (39). These have been considered to represent a relative contraindication for the transsphenoidal approach, particularly in young patients with normal pituitary function (36). One option for such cases has been to incise the pituitary gland performing the resection of the tumor creating a relatively narrow corridor through the gland (11, 39). Although this technique presents a considerable risk of losing pituitary function, it still has better overall endocrinological results than those found in transcranial surgery (11, 21). However, this alternative does not give an adequate view of the suprasellar region or the interpeduncular cistern, and any attempt to remove the dorsum sellae through this narrow corridor would increase the chance of losing pituitary function. We believe that the transposition provides the best alternative available when an effort to preserve the gland to access retroinfundibular pathology is desirable. Obviously, patients already with panhypopituitarism are unlikely to receive any endocrine benefit from a pituitary transposition. The surgeon might simply disconnect the stalk using sharp transection, avoiding any pulling to preserve the hypothalamic cells connected to the posterior gland, and remove the pituitary gland from the sella. In our series, there were two patients with prior anterior gland dysfunction in whom pituitary transposition was performed. One of the patients did not have DI previously, and we transposed the pituitary gland superiorly,


intending to preserve the posterior gland function; we were successful in doing so. The other patient with panhypopituitarism had associated DI. However, this patient proved to have a healthy pituitary gland during the surgical exposure and also an intact pituitary stalk that we decided to preserve. Both patients did not have any sign of tumor invasion within the pituitary structures. Inexperienced and lacking long-term outcomes with this approach, we opted to preserve the gland and transpose it with the hope that some function might return. Not surprisingly, neither of these two patients had any improvement in their anterior pituitary function. Nonetheless, the patient with normal posterior pituitary function did not have signs of DI postoperatively. Although the main goal of the pituitary transposition is an adequate visualization of the interpeduncular cistern, we are buoyed by the fact that seven of the eight patients with normal preoperative pituitary function (87.5%) maintained their normal pituitary function. This reinforces the fact that the pituitary gland is principally vascularized by the superior hypophyseal artery, and the ligation of the inferior hypophyseal artery, which we performed in every case, does not interfere with the global pituitary function.

Complications

CSF Leaks Since the pioneering work of Laws (35) in 1980, the transsphenoidal approach has evolved with the performance of a more extensive level of arachnoid dissection around the sella, resulting in a greater incidence of CSF leak (3, 9, 10, 15, 22, 25, 37, 39). It has been reported to occur in 2 to 33% of the patients after transsphenoidal craniopharyngioma resection. However, the incidence of CSF leaks is directly proportional to the amount of arachnoid that is disrupted during the dissection and increased in patients in whom the third ventricle is opened. Frank et al. (15) described 30% of CSF leaks after endonasal endoscopic resection of craniopharyngiomas, with all three cases related to the opening of the third ventricle. All of the patients in our series who underwent pituitary transposition had wide arachnoid dissections, and three patients with craniopharyngiomas also had the infundibular recess of the third ventricle opened. In our series, the incidence of CSF leak was higher for the patients in whom the reconstruction was not performed with a vascularized flap. Early in our experience, we were relying on allografts for reconstruction. These grafts were failing under the high pressure and low resistance created by the wide arachnoid openings and communications with the ventricular system. We believe that the development of the nasal septal flap has significantly changed the rate of CSF leaks, which has been dramatically reduced in our practice. For the present small series of 10 patients, the P value did not reach statistical significance (P 0.08). However, it clearly showed a tendency toward a reduced rate. Larger sample sizes will be needed to see if this tendency reaches statistical significance. In fact, as we gained progressive experience with the flap and overcame the learning curve associated with it, the leak rate improved. Of the two flaps that failed early during the recon-



struction, one was noted to have contracted, so we have started to oversize the flap (Fig. 17, A and B). During the other revision, bone wax interfering with revascularization was noted along the margins. The most recent patient who underwent a pituitary transposition had a complete third ventricular dissection, and the nasal septal flap in conjunction with a lumbar drain were used for reconstruction; he had no evidence of a CSF leak. It is important to note that all of the CSF leaks were appropriately managed with no cases of meningitis and no longterm sequelae. We believe CSF leaks must be aggressively treated with early reoperation to avoid continuous exposure of the intradural contents to the nasal cavity. In the case of reexploration of CSF leaks, it is important not to undo the existing reconstruction. The area in question is usually small and focal as a result of graft migration. We recommend placement of a simple augmentation of site with an abdominal fat graft with replacement of the balloon buttress.

Cranial Nerve Deficit There were no new incidences of oculomotor nerve paresis; this is also the result of the midline corridor, which does not require transgressing the plane of the third nerve as it travels from the interpeduncular cistern into the cavernous sinus. We did have one patient with transient abducens nerve palsy. This was encountered during the resection of a clival chordoma, and the sixth nerve had to be dissected free from the tumor along the ventral surface of the lesion along the clivus; we do not believe it was related to the transposition.

CONCLUSION The expanded endonasal approach with pituitary transposition is an efficacious way to expose the retroinfundibular region at the interpeduncular cistern. It has the advantage of being a midline approach to access midline lesions, avoiding all the lateral boundaries of the interpeduncular cistern, which are rich in vital neurovascular structures. In addition, transposition followed by removal of the dorsum sellae and the posterior clinoids allow for direct subchiasmatic/retrochiasmatic access into the heart of any lesion located in the interpeduncular, prepontine cistern as well as the anterior recess of the third ventricle. However, the approach does not change the pathology or goals of surgery, and they must take precedence. Specifically, access does not by itself determine the degree of resection, but rather surgical goals are more influenced by patient factors, biology of the tumor, and adherence. The availability of a consistent form of sealing using a vascularized flap may make this approach potentially more attractive. Obviously, this represents a very early and preliminary experience with the technique, and long-term data with a larger sample size will determine the durability of the approach. It should also be emphasized that in our experience of over 700 purely endoscopic endonasal cases of endonasal surgery over the last decade, this approach represents technically the most demanding of all of the modules we have previously described. We strongly recommend that the endoneurosurgeon only con-

NEUROSURGERY

sider this approach after reasonable experience because the risk of disastrous complications is significant. In the final analysis, the best approach to this region is the approach that the operating surgeon is most comfortable and experienced with. The pituitary transposition may simply add a midline corridor to the surgeon’s armamentarium.

Disclosure AK, CS, and RC are paid consultants for Karl Storz and Stryker Corporations.

REFERENCES 1. Al-Mefty O, Ayoubi S, Kadri PA: The petrosal approach for the total removal of giant retrochiasmatic craniopharyngiomas in children. J Neurosurg 106 [Suppl]:87–92, 2007. 2. Arai H, Sato K, Okuda, Miyajima M, Hishii M, Nakanishi H, Ishii H: Transcranial transsphenoidal approach for tuberculum sellae meningiomas. Acta Neurochir (Wien) 142:751–757, 2000. 3. Chakrabarti I, Amar AP, Couldwell W, Weiss MH: Long-term neurological, visual, and endocrine outcomes following transnasal resection of craniopharyngioma. J Neurosurg 102:650–657, 2005. 4. Couldwell WT, Rosenow JM, Rovit RL, Benzil DL: Hypophysopexy technique for radiosurgical treatment of cavernous sinus pituitary adenoma. Pituitary 5:169–173, 2002. 5. Couldwell WT, Weiss MH, Rabb C, Liu JK, Apfelbaum RI, Fukushima T: Variations on the standard transsphenoidal approach to the sellar region, with emphasis on the extended approaches and parasellar approaches: Surgical experience in 105 cases. Neurosurgery 55:539–550, 2004. 6. de Divitiis E: Endoscopic transsphenoidal surgery: Stone-in-the-pond effect. Neurosurgery 59:512–520, 2006. 7. de Divitiis E, Cappabianca P: Microscopic and endoscopic transsphenoidal surgery. Neurosurgery 51:1527–1530, 2002. 8. Doglietto F, Prevedello DM, Jane JA, Han J, Laws ER: Brief history of endoscopic transsphenoidal surgery—From Philipp Bozzini to the First World Congress of Endoscopic Skull Base Surgery. Neurosurg Focus 19:E3, 2005. 9. Dumont AS, Kanter AS, Jane JA, Laws ER: Extended transsphenoidal approach. Front Horm Res 34:29–45, 2006. 10. Dusick JR, Esposito F, Kelly DF, Cohan P, DeSalles A, Becker DP, Martin NA: The extended direct endonasal transsphenoidal approach for nonadenomatous suprasellar tumors. J Neurosurg 102:832–841, 2005. 11. Fahlbusch R, Honegger J, Paulus W, Huk W, Buchfelder M: Surgical treatment of craniopharyngiomas: Experience with 168 patients. J Neurosurg 90:237–250, 1999. 12. Fahlbusch R, Schott W: Pterional surgery of meningiomas of the tuberculum sellae and planum sphenoidale: Surgical results with special consideration of ophthalmological and endocrinological outcomes. J Neurosurg 96:235–243, 2002. 13. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Preul MC, Spetzler RF: Anatomical and quantitative description of the transcavernous approach to interpeduncular and prepontine cisterns. Technical note. J Neurosurg 104:957–964, 2006. 14. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Spetzler RF, Preul MC: Comparative analysis of anterior petrosectomy and transcavernous approaches to retrosellar and upper clival basilar artery aneurysms. Neurosurgery 58 [Suppl]:ONS13–ONS21, 2006. 15. Frank G, Pasquini E, Doglietto F, Mazzatenta D, Sciarretta V, Farneti G, Calbucci F: The endoscopic extended transsphenoidal approach for craniopharyngiomas. Neurosurgery 59 [Suppl 1]:ONS75–ONS83, 2006. 16. Frank G, Pasquini E, Mazzatenta D: Extended transsphenoidal approach. J Neurosurg 95:917–918, 2001. 17. Fries G, Perneczky A: Endoscope-assisted brain surgery: Part 2—Analysis of 380 procedures. Neurosurgery 42:226–232, 1998. 18. Hadad G, Bassagasteguy L, Carrau RL, Mataza JC, Kassam A, Snyderman CH, Mintz A: A novel reconstructive technique after endoscopic expanded endonasal approaches: Vascular pedicle nasoseptal flap. Laryngoscope 116:1882–1886, 2006.


KASSAM ET AL.

19. Hakuba A, Nishimura S, Inoue Y: Transpetrosal–transtentorial approach and its application in the therapy of retrochiasmatic craniopharyngiomas. Surg Neurol 24:405–415, 1985. 20. Harsh GR, Sekhar LN: The subtemporal, transcavernous, anterior transpetrosal approach to the upper brain stem and clivus. J Neurosurg 77:709–717, 1992. 21. Honegger J, Buchfelder M, Fahlbusch R: Surgical treatment of craniopharyngiomas: Endocrinological results. J Neurosurg 90:251–257, 1999. 22. Kaptain GJ, Vincent DA, Sheehan JP, Laws ER: Transsphenoidal approaches for the extracapsular resection of midline suprasellar and anterior cranial base lesions. Neurosurgery 49:94–101, 2001. 23. Kassam A, Carrau RL, Snyderman CH, Gardner P, Mintz A: Evolution of reconstructive techniques following endoscopic expanded endonasal approaches. Neurosurg Focus 19:E8, 2005. 24. Kassam A, Snyderman CH, Mintz A, Gardner P, Carrau RL: Expanded endonasal approach: The rostrocaudal axis. Part II. Posterior clinoids to the foramen magnum. Neurosurg Focus 19:E4, 2005. 25. Kassam AB, Gardner P, Snyderman C, Mintz A, Carrau R: Expanded endonasal approach: Fully endoscopic, completely transnasal approach to the middle third of the clivus, petrous bone, middle cranial fossa, and infratemporal fossa. Neurosurg Focus 19:E6, 2005. 26. Kassam AB, Snyderman C, Gardner P, Carrau R, Spiro R: The expanded endonasal approach: A fully endoscopic transnasal approach and resection of the odontoid process: Technical case report. Neurosurgery 57 [Suppl]:E213, 2005. 27. Kassam AB, Thomas AJ, Zimmer LA, Snyderman CH, Carrau RL, Mintz A, Horowitz M: Expanded endonasal approach: A fully endoscopic completely transnasal resection of a skull base arteriovenous malformation. Childs Nerv Syst 23:491–498, 2007. 28. Kawase T, Toya S, Shiobara R, Mine T: Transpetrosal approach for aneurysms of the lower basilar artery. J Neurosurg 63:857–861, 1985. 29. Kehrli P, Ali M, Reis M, Maillot C, Dietemann JL, Dujovny M, Ausman JI: Anatomy and embryology of the lateral sellar compartment (cavernous sinus) medial wall. Neurol Res 20:585–592, 1998. 30. Kim J, Choe I, Bak K, Kim C, Kim N, Jang Y: Transsphenoidal supradiaphragmatic intradural approach: Technical note. Minim Invasive Neurosurg 43:33–37, 2000. 31. Kitano M, Taneda M: Extended transsphenoidal approach with submucosal posterior ethmoidectomy for parasellar tumors. Technical note. J Neurosurg 94:999–1004, 2001. 32. Kouri JG, Chen MY, Watson JC, Oldfield EH: Resection of suprasellar tumors by using a modified transsphenoidal approach. Report of four cases. J Neurosurg 92:1028–1035, 2000. 33. Krisht AF: Transcavernous approach to diseases of the anterior upper third of the posterior fossa. Neurosurg Focus 19:E2, 2005. 34. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A strategy for successful outcome using the pretemporal transzygomatic transcavernous approach. Neurosurgery 56 [Suppl]:261–273, 2005. 35. Laws ER Jr: Transsphenoidal microsurgery in the management of craniopharyngioma. J Neurosurg 52:661–666, 1980. 36. Laws ER Jr: Transsphenoidal removal of craniopharyngioma. Pediatr Neurosurg 21 [Suppl 1]:57–63, 1994. 37. Laws ER, Kanter AS, Jane JA, Dumont AS: Extended transsphenoidal approach. J Neurosurg 102:825–828, 2005. 38. Liu JK, Schmidt MH, MacDonald JD, Jensen RL, Couldwell WT: Hypophysial transposition (hypophysiopexy) for radiosurgical treatment of pituitary tumors involving the cavernous sinus. Technical note. Neurosurg Focus 14:E11, 2003. 39. Maira G, Anile C, Albanese A, Cabezas D, Pardi F, Vignati A: The role of transsphenoidal surgery in the treatment of craniopharyngiomas. J Neurosurg 100:445–451, 2004. 40. Maira G, Anile C, Colosimo C, Cabezas D: Craniopharyngiomas of the third ventricle: Trans-lamina terminalis approach. Neurosurgery 47:857–865, 2000. 41. Margalit NS, Lesser JB, Moche J, Sen C: Meningiomas involving the optic nerve: Technical aspects and outcomes for a series of 50 patients. Neurosurgery 53:523–532, 2003. 42. Menovsky T, Grotenhuis JA, de Vries J, Bartels RH: Endoscope-assisted supraorbital craniotomy for lesions of the interpeduncular fossa. Neurosurgery 44:106–112, 1999.


43. Nagata S, Sasaki T: The transsylvian trans-limen insular approach to the crural, ambient and interpeduncular cisterns. Acta Neurochir (Wien) 147:863–869, 2005. 44. Page RB: Pituitary blood flow. Am J Physiol 243:E427–E442, 1982. 45. Park CK, Jung HW, Yang SY, Seol HJ, Paek SH, Kim DG: Surgically treated tuberculum sellae and diaphragm sellae meningiomas: The importance of short-term visual outcome. Neurosurgery 59:238–243, 2006. 46. Peker S, Kurtkaya-Yapicier O, Kiliç T, Pamir MN: Microsurgical anatomy of the lateral walls of the pituitary fossa. Acta Neurochir (Wien) 147:641–649, 2005. 47. Perneczky A, Fries G: Endoscope-assisted brain surgery: Part 1—Evolution, basic concept, and current technique. Neurosurgery 42:219–225, 1998. 48. Rhoton AL Jr: The cavernous sinus, the cavernous venous plexus, and the carotid collar. Neurosurgery 51 [Suppl]:S375–S410, 2002. 49. Rosenstein J, Symon L: Surgical management of suprasellar meningioma. Part 2: Prognosis for visual function following craniotomy. J Neurosurg 61:642–648, 1984. 50. Sano H, Kato Y, Akashi K, Yamaguchi S, Hayakawa M, Arunkumar R, Kanno T: Operation on high-lying basilar bifurcation aneurysms. Surg Neurol 48:458–464, 1997. 51. Seifert V, Raabe A, Zimmermann M: Conservative (labyrinth-preserving) transpetrosal approach to the clivus and petroclival region-indications, complications, results and lessons learned. Acta Neurochir (Wien) 145:631–642, 2003. 52. Shi XE, Wu B, Zhou ZQ, Fan T, Zhang YL: Microsurgical treatment of craniopharyngiomas: Report of 284 patients. Chin Med J (Engl) 119:1653–1663, 2006. 53. Shirane R, Hayashi T, Tominaga T: Fronto-basal interhemispheric approach for craniopharyngiomas extending outside the suprasellar cistern. Childs Nerv Syst 21:669–678, 2005. 54. Sugita K, Kobayashi S, Shintani A, Mutsuga N: Microneurosurgery for aneurysms of the basilar artery. J Neurosurg 51:615–620, 1979. 55. Théron J, Chevalier D, Delvert M, Laffont J: Diagnosis of small and micro pituitary adenomas by intercavernous sinus venography. A preliminary report. Neuroradiology 18:23–30, 1979. 56. Tummala RP, Coscarella E, Morcos JJ: Transpetrosal approaches to the posterior fossa. Neurosurg Focus 19:E6, 2005. 57. Ulm AJ, Tanriover N, Kawashima M, Campero A, Bova FJ, Rhoton A Jr: Microsurgical approaches to the perimesencephalic cisterns and related segments of the posterior cerebral artery: Comparison using a novel application of image guidance. Neurosurgery 54:1313–1328, 2004. 58. Van Effenterre R, Boch AL: Craniopharyngioma in adults and children: A study of 122 surgical cases. J Neurosurg 97:3–11, 2002. 59. Yasuda A, Campero A, Martins C, Rhoton AL Jr, Oliveira E, Ribas GC: Microsurgical anatomy and approaches to the cavernous sinus. Neurosurgery 56 [Suppl]:4–27, 2005. 60. Yasuda A, Campero A, Martins C, Rhoton AL Jr, Ribas GC: The medial wall of the cavernous sinus: Microsurgical anatomy. Neurosurgery 55:179–190, 2004. 61. Zevgaridis D, Medele RJ, Müller A, Hischa AC, Steiger HJ: Meningiomas of the sellar region presenting with visual impairment: Impact of various prognostic factors on surgical outcome in 62 patients. Acta Neurochir (Wien) 143:471–476, 2001. 62. Zuccaro G: Radical resection of craniopharyngioma. Childs Nerv Syst 21:679–690, 2005.

COMMENTS

T

he revival of the transsphenoidal approach by the advent of the endoscope in this field has provided new perspectives in cranial base surgery, with the treatment of lesions once judged amenable to open transcranial surgery only. This is a very interesting proposal to access and manage a difficult surgical target area, i.e., the retroinfundibular fossa. The solution offered by the authors is very well described and detailed with excellent anatomical preparations, illustrative drawings, and intraoperative pictures. Nevertheless, it represents a very sophisticated and technically demanding approach appropriate for expert endoscopic cranial base surgeons only. The great risk related to posterior clinoid dissection maneuvers if arterial bleeding occurs



has to be considered. Furthermore, 5 of 10 patients had a cerebrospinal fluid leak and this remains a major complication to be improved. The authors considered the pituitary gland able to tolerate the manipulation they propose. The results of their series seem to confirm their thoughts; however, longer times and greater numbers of patients are needed for more complete evaluation. In this procedure, inferior hypophyseal arteries are sectioned bilaterally and the blood supply to the gland is no doubt partially compromised. It would be interesting to ask our usual co-workers, the endocrinologists, what they think of such procedures and to know their opinions on early and late postoperative pituitary function. On the other hand, I fully agree with the authors that endocrine replacement therapy can fully compensate for any deficit. Pituitary hypersecretion can be difficult to control, and hypopituitarism is usually counterbalanced with medications. Despite all these difficulties, this procedure seems to provide a wide, midline exposure on the interpeduncular area even compared with transcranial routes. Therefore, in selected patients, transposition of the pituitary gland could represent a useful technique to add to the continuously evolving armamentarium of the endonasal cranial base surgeon. Paolo Cappabianca Naples, Italy

K

assam et al. present a new endoscopic approach through the dorsum sellae to expose the interpeduncular cistern. In patients in whom pituitary functions are able to be preserved, the authors perform a pituitary transposition first. In the authors’ opinion this approach represents, from the technical point of view, the most demanding of all the modules they have described previously. I feel that pituitary transposition is feasible with relatively low risk, but I believe that it is rarely required (only in young people with preserved pituitary function). I believe that the removal of the dorsum sellae and posterior clinoid processes, if the anatomy is normal (craniopharyngiomas), may be extremely risky and difficult. Conversely, when the anatomy is changed by the pathological lesions, such as chordomas, it may be easier and safer, requiring only the widening of the corridor created by the tumor. I appreciate the intense and innovative work of the Pittsburgh group who attempts to standardize new approaches to the cranial base through extended endonasal approaches. It is reasonable, however, to have some doubts about approaches that are so difficult and dangerous that they are rarely reproducible. This is one of the most important characteristics that we ask of any surgical technique, namely, its reproducibility. Giorgio Frank Bologna, Italy

K

assam et al. provide a useful and well-illustrated description of their experience displacing the pituitary gland to perform a transdorsum sellae approach to the interpeduncular cistern. This maneuver increases the working space for the endonasal endoscopic approach through the upper clivus. Superior transposition of the gland with preservation of the superior hypophyseal arteries bilaterally does not appear to compromise gland function. As the authors mention, there are previous descriptions in the literature of pituitary manipulation, either transposition in preparation for radiosurgery (3) or partial resection en route to the suprasellar cistern (1, 4). Nevertheless, the technique described in this article is more elegant because the gland is left intact and the transposition is not merely lateral, away from the cavernous sinus, but superior, which creates a much larger working distance. The authors’ ability to advance this

NEUROSURGERY

concept is no doubt facilitated by the increased visibility provided by the endoscope. In our series of endoscopic cranial base procedures, we have also made use of pituitary transposition and can confirm its utility and the preservation of gland function. However, as we have documented (2), it is possible to reach the interpeduncular cistern from an endonasal approach and preserve the stalk by working above the gland and below the chiasm without performing a transposition. Nevertheless, in certain patients, a full transposition is clearly required, which makes this article a very nice addition to the endoscopic literature. Theodore H. Schwartz New York, New York

1. Fahlbusch R, Honegger J, Paulus W, Huk W, Buchfelder M: Surgical treatment of craniopharyngiomas: Experience with 168 patients. J Neurosurg 90:237–250, 1999. 2. Laufer I, Anand VK, Schwartz TH: Endoscopic, endonasal extended transsphenoidal, transplanum tanstuberculum approach for resection of suprasellar lesions. J Neurosurg 106:400–406, 2007. 3. Liu JK, Schmidt MH, MacDonald JD, Jensen RL, Couldwell WT: Hypophyseal transposition (hypophysopexy) for radiosurgical treatment of pituitary tumors involving the cavernous sinus: Technical note. Neurosurg Focus 14:E11, 2003. 4. Maira G, Anile C, Albanese A, Cabezas D, Pardi F, Vignati A: The role of transsphenoidal surgery in the treatment of craniopharyngiomas. J Neurosurg 100:445–451, 2004.

I

t is true that effective management of some sellar and parasellar lesions may be enhanced by displacement of the pituitary gland to expose a cranial base surgical corridor more effectively. Removal of the bony floor of the sella so that the gland may be displaced inferiorly is useful for many craniopharyngiomas and is a good way to find and to sever the pituitary stalk atraumatically as it enters the gland dorsally. Removal of the dorsum and posterior wall of the sella may also be useful in reaching the retroclinoid and prepontine spaces. This can be done by mobilizing the gland superiorly; however, it is important to realize that there can be luxuriant venous channels behind and around the pituitary gland that are no problem in cadaver dissections but that can produce severe visualization and hemostatic challenges in some patients. No matter in what fashion the pituitary gland is mobilized, it is critical to do so with careful microsurgical dissection and meticulous stepby-step hemostasis, making every effort not to manipulate the pituitary stalk in a forceful manner to avoid retrograde cell death in the hypothalamus. Edward R. Laws, Jr. Boston, Massachusetts

D

r. Kassam et al. present their unique experience of transposing the pituitary gland to reach the interpeduncular cistern via an endonasal endoscopic approach in 10 patients. They nicely describe and illustrate a novel cranial base approach to reach this deep cisternal space that has heretofore been a challenge to access by conventional cranial base routes. As they clearly stress, this technique should not be undertaken without considerable prior experience using extended endonasal approaches to the parasellar area. In addition, absolute indications for pituitary transposition may be somewhat limited. Given the ability to access much of the retroglandular space, including the interpeduncular cistern, from an endonasal approach that courses above or below the pituitary gland, alternative and less technically demanding approaches can be considered. From above, by vertically


KASSAM ET AL.

incising the gland in the paramedian plane, a relatively wide space can be created to the left or right of the infundibulum, giving access to the superior portion of the interpeduncular cistern. From below, by drilling the dorsum sella bone to expose the inferior sellar dura and the bone of the upper clivus, a relatively large exposure of the inferior aspect of the interpeduncular cistern can be achieved. These two techniques fall short of true transposition of the gland but do allow one to safely manipulate the gland, albeit to a lesser degree than described here. With an endoscope-assisted technique, we have used these two approaches, alone or occasionally together, for lesions that involve the interpeduncular cistern with a low rate of new pituitary failure.

Regardless of the approach, once the interpeduncular cistern is reached, and tumor removal is completed, an effective “exit strategy” is essential to seal the large cranial base defect. As Dr. Kassam et al. acknowledge, this aspect of the approach remains a major challenge. Although their postoperative cerebrospinal fluid leak rate is 50% in this small series, it has decreased considerably with the use of the vascularized nasal septal flap technique. It is encouraging to see this evolution of the endonasal technique. Daniel F. Kelly Santa Monica, California

The Anatomical Theater at Padua, (1844). From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



TUMOR Operative Technique

THE JUXTACONDYLAR APPROACH TO THE JUGULAR FORAMEN Michaël Bruneau, M.D. Department of Neurosurgery, Hôpital Lariboisière, Paris, France

Bernard George, M.D. Department of Neurosurgery, Hôpital Lariboisière, Paris, France Reprint requests: Michaël Bruneau, M.D., Department of Neurosurgery, Hôpital Erasme, Route de Lennik, 808, 1070 Brussels, Belgium. Email: [email protected] Received, July 20, 2006. Accepted, July 20, 2007.

OBJECTIVE: We sought to describe the juxtacondylar approach to jugular foramen tumors. METHODS: Through an anterolateral approach, the third segment of the vertebral artery (between C2 and the dura mater) is controlled. The C1 transverse process of the atlas, which is located just inferiorly to the jugular foramen, is then removed. The dissection of the internal jugular vein is performed as high as possible, with control of the IXth, Xth, XIth, and XIIth cranial nerves. If required by a tumor extending into the neck, the internal and external carotid arteries can be exposed and controlled. Through a partial mastoidectomy and after removal of the bone covering the jugular tubercle, the end of the sigmoid sinus and then the posteroinferior part of the jugular foramen are reached. RESULTS: This technique is efficient to expose tumors extending into the jugular foramen. Contrary to the infratemporal approach, it has the main advantage of avoiding petrous bone drilling and associated potential complications. Lower cranial nerves are well exposed in the neck. In patients with schwannomas, complete resection with selective dividing of only the few involved rootlets can be achieved. CONCLUSION: The juxtacondylar approach is an efficient approach to tumors located in the jugular foramen. It necessitates control of the third segment of the vertebral artery but has the advantage of avoiding complications associated with petrous bone drilling. Extension beyond the jugular foramen requires combination with an infratemporal or a retrosigmoid approach. KEY WORDS: Cranial base, Jugular foramen, Skull base, Surgical approach, Tumor, Vertebral artery Neurosurgery 62[ONS Suppl 1]:ONS75–ONS81, 2008

T

umors of the jugular foramen (JF) are rare and most commonly include paragangliomas, schwannomas, and meningiomas (21). Access to the JF is difficult because of its deep location and surrounding structures obscuring the surgical route. Several surgical approaches have been developed to overcome these difficulties. These surgical approaches provide different angles of attack for JF tumors. They have their specific indications according to the tumor type and area of development, as well as their own advantages and disadvantages (Table 1). According to Rhoton (23), JF approaches can be subdivided into three main groups: a lateral group (the postauricular transtemporal approach subdivided in infralabyrinthine, translabyrinthine, and transcochlear approaches); a posterior group (retrosigmoid approach and its more extensive far-lateral and transcondylar variants); and an anterior group (preauricular subtemporal-infratemporal approach). Two other groups also exist but are not suitable alone for lesion resection: the superior group (middle fossa approach); and the inferior group (cervical approach upward to the JF). The standard surgical approach is lateral, the infratemporal transpetrosal approach described by Fisch et al. (9), Fisch and

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297103.19261.79

Mattox (8), and Fisch (7). It permits one to gain superior and lateral access to the JF by drilling the petrous bone. During this procedure, the facial nerve is frequently transposed anteriorly (7–9, 18, 23). This transposition allows the drilling of the bone inferior to the labyrinth (23). Manipulation of the facial nerve exposes the patient to a non-negligible risk of facial nerve palsy (7, 15, 17, 18, 20, 21, 25). To limit the risk of facial nerve palsy, some surgeons advocate keeping the facial nerve in its bony canal if the nerve is not infiltrated by the tumor (3, 12, 16, 19, 21, 22). The retrosigmoid approach is an intradural approach leading to the JF via its medial side and is best indicated for lesions developed in the posterior fossa (24). Whatever the approach used, complex JF tumors are challenging and require combined approaches (1, 2, 4, 18, 20). With increasing experience in vertebral artery (VA) control, we have developed another option, the juxtacondylar approach (11–13). Through an extreme-lateral approach, access is provided to the posteroinferior aspect of the JF, progressing along the lateral wall of the craniocervical junction (10, 13, 14).


BRUNEAU AND GEORGE

TABLE 1. Surgical approaches to the jugular foramen: indications, surgical routes, advantages, and disadvantagesa Anterior Infratemporal fossa approach

a

Lateral Infratemporal approach

Posterior Suboccipital retro- Suboccipital transsigmoid approach condylar approach

Juxtacondylar approach

Indicated for

Tumors extending along the petrous portion of the ICA through the eustachian tube through the cancellous portion of the petrous bone

Tumors extending into the anterosuperior part of the JF

Tumors extending into the posteroinferior part of the jugular foramen

Intradural tumors Intracranial orifice of the JF

Approach through

Tympanic bone

Mastoidectomy and petrous bone

Limited mastoidectomy and jugular tubercle

Lateral suboccipital craniectomy

Occipital condyle

Advantages

Allows access to the middle and upper clivus

Control of major vessels and cranial nerves in the neck; Control of the petrosal; Control of the portion of the ICA; Wide exposure after combination with retrosigmoid craniotomy

Control of major vessels and cranial nerves in the neck; Control of the vertebral artery; Suppression of vascular feeders; No or limited petrous bone drilling

Technical simplicity; Possible extension to the FM and clivus in combination with a far-lateral approach

Access to the FM anterior rim and lower clivus vertebral artery; Excellent exposure of the lower brainstem

Disadvantages

Hearing loss, facial paresis, numbness, jaw malocclusion

Risk of facial palsy, hearing loss, CSF leak, meningitis

Need strict technique for VA exposure; Limited access to the superior and anterior part of the jugular foramen (this limitation is thwarted by combination with an infratemporal approach)

Limited access to the jugular foramen

Craniocervical instability if drilling is too extended; Risk of CSF leak, CN injury

ICA, internal carotid artery; JF, jugular foramen; FM, foramen magnum; CSF, cerebrospinal fluid; VA, vertebral artery; CN, cranial nerve.

PRINCIPLES OF THE JUXTACONDYLAR APPROACH Anatomically, the JF is located exactly above the transverse process of the atlas, approximately 10 mm superiorly (Fig. 1A). Resection of this bone structure after VA control permits the neurosurgeon to reach the JF without petrous bone drilling.

Advantages of the Juxtacondylar Approach For access strictly limited to the JF, petrous bone drilling is limited or not required. Associated complications are then avoided, including hearing loss, cerebrospinal fluid leakage, and facial nerve palsy after manipulation or transposition from the fallopian canal. VA control at the initial step of the surgery permits the surgeon to suppress the vascular tumoral feeders (10). This point is important because JF tumors are often highly vascularized. Embolization by either the endovascular route or direct intratumoral puncture helps to suppress the tumor vascular supply from external carotid artery branches (6, 14). The procedure also allows good exposure and control of the lower cranial nerves. In patients with schwannomas, selective dividing of rootlets bearing the tumor can be achieved.

Limitations The juxtacondylar approach provides access to the inferior and posterior side of the JF. Lesions developed inside the


JF, such as schwannomas, are especially appropriate for this approach. Access to the superior and anterior parts of the JF is limited and requires combination with an infratemporal approach. Both approaches are complementary and sometimes necessary for large JF tumors, especially those extending beyond the JF limits, such as paragangliomas. However, in such cases, the petrous bone drilling may be limited and the facial nerve transposition exceptionally required. Tumors extending in the posterior fossa require combination with a retrosigmoid approach.

SURGICAL ANATOMY The JF is formed by the occipital and temporal bones (Fig. 2). Access to this region is difficult because of its deep location and the important surrounding structures, including the facial nerve superolaterally, the hypoglossal nerve inferomedially, the internal carotid artery anteriorly, the middle ear and the labyrinth and internal auditory canal superiorly, and the VA inferiorly. The JF is composed of two venous and neural parts. The venous part includes two venous channels. The larger venous channel, or sigmoid part, is located posterolaterally and receives the venous flow of the sigmoid sinus (SS) (23). The smaller venous channel, or petrosal part, is located anteromedially and receives mainly the venous flow of the inferior pet-


JUXTACONDYLAR APPROACH TO THE JUGULAR FORAMEN

A

B

gadolinium enhancement helps to define the tumor extension. Angiography is required to study the vascular feeders and occasionally to devascularize the lesion by embolization.

Step 1: Patient Positioning

C

D

The patient is placed supine, with the head slightly extended and rotated toward the opposite side. There is no means to enlarge the space between the C1 transverse process and the mastoid because the atlas follows the head rotation exactly. The head must not be rotated more than 30 degrees because head rotation induces anterior displacement of the atlas, which could limit the surgical access. A few millimeters can be gained by tilting the head inferiorly.

Step 2: Surgical Approach

FIGURE 1. Preoperative imaging of a left-sided vagus nerve schwannoma. The lesion (white asterisk) is located at the level of the left jugular foramen, which is enlarged, and it destroyed the lateral part of the clivus. A, coronal computed tomographic (CT) reconstruction. The transverse process of the atlas (black asterisk) is located 10 mm below the jugular foramen. B, axial contrast-enhanced CT scan. C, preoperative T1weighted contrast enhanced magnetic resonance imaging scan (asterisk), right jugular foramen paraganglioma. D, angiogram showing that the lesion (asterisk) is fed by branches of the external carotid artery. ECA, external carotid artery; ICA, internal carotid artery; Int Max A, internal maxillary artery; Occ A, occipital artery.

rosal sinus; it is also a venous confluent as it receives the blood of the hypoglossal canal, the petroclival fissure, and the vertebral venous plexus (23). The petrosal part empties into the sigmoid part through an opening in the medial wall of the jugular bulb between the glossopharyngeal nerve anteriorly and the vagus and accessory nerves posteriorly (23). The neural part, or intrajugular part, is located at the site of the intrajugular process of the temporal and occipital bones. It contains the glossopharyngeal, vagus, and accessory nerves, which course medially to the internal jugular vein (IJV) after piercing the dura (23).

Surgical Technique

Preoperative Imaging Preoperative imaging generally includes computed tomographic (CT) scans (Fig. 1, A and B), magnetic resonance imaging (Fig. 1C), and angiography (Fig. 1D). CT scanning permits the study of the bone anatomy of the JF, which is often enlarged and eroded by the lesion. Magnetic resonance imaging with

NEUROSURGERY

The skin incision (Fig. 3A) has a question mark-shape and follows the medial border of the sternomastoid (SM) muscle on its upper 6 cm, then curves medially along the mastoid process, and turns around the ear. The temporal muscle is split from the bone behind the ear down to the mastoid process and kept connected with the SM muscle. It will be used as a pediculated flap to close the dura. The SM muscle is then separated from the mastoid process. The dissection is pursued along the IJV lateral border. When dissecting the fat tissue between the IJV and the SM muscle, care must be taken to avoid damage to the spinal accessory nerve (XIth cranial nerve), which crosses the surgical field obliquely and inferiorly (Fig. 3B). The nerve crosses the IJV anteriorly, posteriorly, or on both sides, and then runs usually from the C1–C2 level to the deep aspect of the SM muscle at the C3–C4 level. Muscular contractions induced in its vicinity by bipolar coagulation help to localize it. To open the surgical field, the nerve must be freed as far as possible up to the cranial base. It is then protected and mobilized caudally by traction on the fat pad separated from the deep cervical muscles and wrapped around it. The tip of the transverse process of the atlas can be palpated 15 mm below and in front of the mastoid process tip (Fig. 3B). It is exposed by dividing the attachments of small muscles that insert into it; the minor rectus muscle superiorly, the superior and inferior oblique muscles posteriorly, and the levator scapulae muscle inferiorly. This dissection must be performed cautiously as the VA is just beneath the muscles.

Step 3: Control of the VA V3 Segment The surgical technique for this step has been extensively described previously (5). The VA V3 segment extends from the C2 transverse process up to the dura mater (Fig. 3, C–F). Its course is divided in three portions: a vertical portion between the C2 and C1 transverse processes, a transversary portion inside the C1 transverse foramen, and a horizontal portion above the C1 posterior arch, which finally curves obliquely before piercing the dura mater because of a step at the medial border of the atlas groove. Because the head is slightly rotated, the vertical and horizontal VA segments are stretched and run parallel to the posterior arch of atlas. The key to confident exposure of the VA


BRUNEAU AND GEORGE

is to work in the subperiosteal plane. The VA is enclosed in a periosteal sheath and surrounded by a venous plexus. Splitting the periosteal sheath from the bone with a smooth spatula prevents damage. If venous bleeding occurs, direct bipolar coagulation on the sheath is safe and permits control of bleeding. Following this technique, the VA segment above C1 is separated from the atlas groove. The FIGURE 2. Cadaveric dissection (left side). A, vertebral C dissection of the superior artery (VA) V3 segment is exposed. The interior jugular aspect of this VA portion is vein (IJV) has been dissected up to the cranial base. more difficult because no clear Partial mastoidectomy and suboccipital craniectomy have limit can be identified. The ligbeen performed. The jugular tubercle (JT) is exposed with aments covering the occipithe spatula. B, after removal of the JT with a Kerrison toatlantal joints must be cut rongeur, the jugular bulb (JB), which is the main venous carefully following the VA part of the jugular foramen, is exposed. C, the neural superior aspect. part of the jugular foramen is located anteriorly. This part is composed of Cranial Nerves IX, X, and XI. The C1–C2 segment is Cranial Nerve X lies at the posterior aspect of carotidoalready in view after resection jugular vessels. Cranial Nerve XI crosses the IJV posteof the small muscles that riorly before reaching the posterior aspect of the sterinsert on the C1 transverse nomastoid muscle. The petrous internal carotid artery p ro c e s s . T h i s s e g m e n t i s (ICA) is observed anteriorly after drilling of the temporal petrous bone. CN, cranial nerve. crossed by the second cervical nerve root, which lies just over the posterior aspect of the C1–C2 segment and is a good landmark. It must be preserved unless the periosteal sheath must be opened or control gained down to the C2 transverse process. Cutting this cervical nerve root is responsible for mild sensory deficits of the ear lobe and the jaw angle area. When both VA segments above and below the C1 transverse process have been exposed, the C1 transverse foramen must be opened (Fig. 3F). Again, the periosteum is split with a smooth spatula inside the foramen. The bone can then be resected confidently with a Kerrison rongeur passing between the periosteal sheath and the bone. The VA sheath can then be fully exposed inside the foramen by separating the periosteal sheath in the anterior part of the foramen. The transverse foramen, completely unroofed, is then opened widely until the VA maintains contact only in the concavity of its loop with the lateral part of the C1 posterior arch. In most cases, exposure of the VA is sufficient and the VA can be left in place. The JF can be reached by passing above the VA loop along the lateral aspect of the occipital condyle. If access to the JF must be enlarged, the VA can be transposed out of the C1 transverse foramen so as to expose and, whenever necessary, drill the posterolateral aspect of the C0–C1 and C1–C2 joints and the jugular tubercle.

A

B

FIGURE 3. Surgical steps of the juxtacondylar approach to the jugular foramen.


Step 4: IJV Control The IJV has been exposed during the surgical approach as its crossing by the spinal accessory nerve (Fig. 3B). Its dissection



vertical portion of the intrapetrosal carotid artery, achieved by pursuing the dissection of the cervical internal carotid artery.

Step 8: SS and IJV Occlusions The SS is occluded by packing with Surgicel (Johnson & Johnson, Cargrave, United Kingdom), and then the IJV is ligated with a 3–O silk suture (Fig. 3, F–G). This step is performed without difficulty after vascular control is obtained in the neck and preoperative embolization.

Step 9: Extradural Tumor Resection

FIGURE 4. Perioperative view of the left jugular foramen after the complete removal of a paraganglioma. No petrous bone drilling is required to reach the posteroinferior aspect of the foramen.

must then be pursued upward. Doing so, the IXth, Xth, and XIIth cranial nerves are controlled, the posterior part of the digastric muscle is divided, and the occipital artery is ligated (Fig. 3C).

Step 5: Control of the Internal and External Carotid Arteries This step is required only if the tumor extends into the neck; it does not need to be performed for simple JF exposure (Fig. 3B).

Step 6: Partial Mastoidectomy Using a rongeur and drill, a limited mastoidectomy is performed to expose the distal portion of the SS (Fig. 3F). The remaining bone covering the jugular tubercle and corresponding to the posteroinferior part of the JF is removed with a small Kerrison rongeur and/or drill progressing toward the jugular bulb from both sides (the IJV and the SS sides). The removal of the inferior wall of the JF requires drilling of the jugular tubercle and, occasionally, a small part of the occipital condyle. In practice, opening the JF is often facilitated by the bone erosion and JF enlargement produced by the tumor. The posterosuperior aspect of the JF can also be reached after control of the facial nerve in its third petrosal segment and at the exit of the stylomastoidian foramen. This combined infratemporal approach requires only limited petrous bone drilling because the posteroinferior aspect of the JF has already been opened (Fig. 4).

Step 7: Petrosal Bone Drilling At this step, the lateral part of the posterior fossa and the SS has already been exposed. Drilling of the petrosal bone is performed as extensively as required by the tumor. Opening the inferior part of the JF limits the drilling and permits the facial nerve to remain in the fallopian canal, reducing the risk of facial nerve palsy. The labyrinth is opened, if required, by the tumor extension. The most important point is control of the

NEUROSURGERY

Tumor resection usually begins at the JF level. Afterward, the resection is extended in the SS, IJV, and petrosal bone, and finally along the internal carotid artery.

Step 10: Intradural Tumor Resection If required, the posterior fossa dura mater is incised perpendicularly to the last SS centimeter (Fig. 3H). The incision passes through the SS to reach the JF. A retractor can be placed on the lateral surface of the cerebellum and, under microscopic magnification, the intradural part of the tumor can be resected. The cranial nerve rootlets are found at the opposite aspect of the tumor. Those involved in the tumor must be sacrificed.

Step 11: Closure The mastoid cells are packed with fat graft, muscle, and fibrin glue. If the dura has been opened, a watertight closure is mandatory. A small dural defect usually remains after resection of the intradural tumor component. This hole is closed by use of the temporal muscle flap kept pediculated on the SM muscle. The suture line can be strengthened with fat, muscle, and biological glue. The fat can be taken from the surroundings of the spinal accessory nerve. We often use the digastric muscle, which has been previously mobilized. Finally, we reapproximate the SM muscle to its aponeurosis on the mastoid. Lumbar drainage is also placed for 72 hours to prevent leakage.

Surgical Complications: Management Lower cranial nerve dysfunction can be the cause of severe complications. Swallowing disturbances may lead to inhalation pneumonia and must be anticipated with tracheostomy and gastrostomy. If lower cranial nerves are paralyzed preoperatively, the contralateral nerves may have already established a compensation mechanism and the postoperative swallowing function will not be modified. However, if preoperatively intact lower cranial nerves are damaged during surgery, the patient will experience severe swallowing problems.

CONCLUSION Several surgical approaches have been used to gain access to JF tumors. The juxtacondylar approach permits the neurosurgeon to reach the posteroinferior aspect of the JF via a transcervical route with VA control. It provides perfect access to lesions developed inside the JF. For lesions extending beyond the JF, this approach must be combined with an infratemporal


BRUNEAU AND GEORGE

approach but with reduced drilling of the petrous bone and often no facial nerve transposition.

REFERENCES 1. Al-Mefty O, Fox JL, Rifai A, Smith RR: A combined infratemporal and posterior fossa approach for the removal of giant glomus tumors and chondrosarcomas. Surg Neurol 28:423–431, 1987. 2. Al-Mefty O, Teixeira A: Complex tumors of the glomus jugulare: Criteria, treatment, and outcome. J Neurosurg 97:1356–1366, 2002. 3. Borba LA, Ale-Bark S, London C: Surgical treatment of glomus jugulare tumors without rerouting of the facial nerve: An infralabyrinthine approach. Neurosurg Focus 17:E8, 2004. 4. Bordi LT, Cheesman AD, Symon L: The surgical management of glomus jugulare tumours-description of a single-staged posterolateral combined otoneurosurgical approach. Br J Neurosurg 3:21–30, 1989. 5. Bruneau M, Cornelius JF, George B: Antero-lateral approach to the V3 segment of the vertebral artery. Neurosurgery 58 [Suppl 1]:ONS29–ONS35, 2006. 6. Casasco A, Herbreteau D, Houdart E, George B, Tran Ba Huy P, Deffresne D, Merland JJ: Devascularization of craniofacial tumors by percutaneous tumor puncture. AJNR Am J Neuroradiol 15:1233–1239, 1994. 7. Fisch U: Infratemporal fossa approach for extensive tumors of the temporal bone and base of the skull, in Silverstein H, Norrel H (eds): Neurological Surgery of the Ear. Birmingham, Aesculapius, 1977, pp 34–53. 8. Fisch U, Mattox D: Infratemporal fossa approach type A, in Fisch U, Mattox: Microsurgery of the Skull Base. Stuttgart, Thieme, 1988, pp 136–281. 9. Fisch U, Fagan P, Valavanis A: The infratemporal fossa approach to the lateral skull base. Otolaryngol Clin North Am 17:513–552, 1984. 10. George B: Jugular foramen paragangliomas. Acta Neurochir (Wien) 118: 20–26, 1992. 11. George B: Management of the vertebral artery in skull base surgery, in PJ Donald (ed): Surgery of the Skull Base. Philadelphia, Lippincott Williams and Wilkins, 1998, pp 533–553. 12. George B, Tran PB: Surgical resection of jugulare foramen tumors by juxtacondylar approach without facial nerve transposition. Acta Neurochir (Wien) 142:613–620, 2000. 13. George B, Lot G, Tran Ba Huy P: The juxtacondylar approach to the jugular foramen (without petrous bone drilling). Surg Neurol 44:279–284, 1995. 14. George B, Casasco A, Deffrennes D, Houdart E: Intratumoral embolization of intracranial and extracranial tumors: Technical note. Neurosurgery 35:771– 774, 1994. 15. Green JD, Brackmann DE, Nguyen CD, Arriaga MA, Telischi FF, De la Cruz A: Surgical management of previously untreated glomus jugulare tumors. Laryngoscope 104:917–921, 1994. 16. Inserra MM, Pfister M, Jackler RK: Anatomy involved in the jugular foramen approach for jugulotympanic paraganglioma resection. Neurosurg Focus 17:E6, 2004. 17. Jackson CG, Kaylie DM, Coppit G, Gardner EK: Glomus jugulare tumors with intracranial extension. Neurosurg Focus 17:E7, 2004. 18. Liu JK, Sameshima T, Gottfried ON, Couldwell WT, Fukushima T: The combined transmastoid retro- and infralabyrinthine transjugular transcondylar transtubercular high cervical approach for resection of glomus jugulare tumors. Neurosurgery 59 [Suppl 1]:ONS115–ONS125, 2006. 19. Oghalai JS, Leung MK, Jackler RK, McDermott MW: Transjugular craniotomy for the management of jugular foramen tumors with intracranial extension. Otol Neurotol 25:570–579, 2004. 20. Patel SJ, Sekhar LN, Cass SP, Hirsch BE: Combined approaches for resection of extensive glomus jugulare tumors. A review of 12 cases. J Neurosurg 80:1026–1038, 1994. 21. Ramina R, Maniglia JJ, Fernandes YB, Paschoal JR, Pfeilsticker LN, Neto MC, Borges G: Jugular foramen tumors: Diagnosis and treatment. Neurosurg Focus 17:E5, 2004. 22. Ramina R, Maniglia JJ, Fernandes YB, Paschoal JR, Pfeilsticker LN, Neto MC: Tumors of the jugular foramen: Diagnosis and management. Neurosurgery 57 [Suppl 1]:ONS59–ONS68, 2005. 23. Rhoton AL Jr: Jugular foramen. Neurosurgery 47:S267–S285, 2000. 24. Samii M, Babu RP, Tatagiba M, Sepehrnia A: Surgical treatment of jugular foramen schwannomas. J Neurosurg 82:924–932, 1995.


25. Whitfield PC, Grey P, Hardy DG, Moffat DA: The surgical management of patients with glomus tumours of the skull base. Br J Neurosurg 10:343–350, 1996.

Acknowledgment We thank Marie-Bernadette Jacqmain for providing medical illustrations.

COMMENTS

I

n this manuscript, Bruneau and George illustrate their experience with the use of the juxtacondylar approach or inferolateral approach to lesions of the jugular foramen. They nicely demonstrate this route and the possibilities of obtaining proximal vertebral control and of exposing the lower cranial nerves. One can argue that such exposure can place the above structures at risk; in our experience, however, appropriate identification of each of these structures actually lowers the risks of iatrogenic injury. As the authors acknowledged, the approach to tumors in this region should be tailored to address local tumoral extensions and should be combined with other approaches as needed. Ricardo A. Hanel Robert F. Spetzler Phoenix, Arizona

I

n this article, the authors have elaborated on a juxtacondylar approach to jugular foramen tumors. For a number of years I have used this approach to paragangliomas and schwannomas of the jugular foramen, usually in combination with a transmastoid approach, with or without the displacement of the facial nerve from the fallopian canal (1–3). It has also been used by John Robertson (personal communication, 2000). When the facial nerve is displaced along with the digastric muscle, as first proposed by Derald Brackmann, patients do not experience facial paralysis. For very extensive glomus tumors, the addition of a preauricular subtemporal infratemporal approach may also allow the facial nerve to be retained in its position and improve the exposure and management of the petrous internal carotid artery. More recently, very aggressive embolization of paragangliomas by direct puncture or the endovascular route to render them relatively avascular has made the operation much easier. We have found that monitoring the function of Cranial Nerves X, XI, and XII during surgery also helps to preserve them. Radiosurgery using the CyberKnife or the latest version of the gamma knife is also a treatment option for some patients. Laligam N. Sekhar Seattle, Washington

1. Patel SL, Sekhar LN, Cass SP, Hirsch BE: Combined approaches for resection of extensive glomus jugular tumors. A review of 12 cases. J Neurosurg 80:1026–1038, 1994. 2. Salas E, Sekhar LN, Ziyal IM, Caputy AJ, Wright DC: Variations of the extreme-lateral craniocervical approach: Anatomical study and clinical analysis of 69 patients. J Neurosurg 90[Suppl 2]:206–219, 1999. 3. Sarma S, Sekhar LN, Schessel DA: Non-vestibular schwannomas of the brain: A 7-year experience. Neurosurgery 50:437–439, 2002.

T

his report describes a juxtacondylar approach for the removal of tumors of the jugular foramen. It is an inferolateral approach to the posterior and inferior jugular foramen in contrast with the more commonly used infratemporal approach, which primarily exposes the anterior and superior jugular foramen.



The operation described incorporates the paracondylar extension of the far lateral approach (described by Rhoton [1, p. 635]) in an approach to small tumors of the jugular foramen. As with a standard far lateral approach (1, pp. 627–641), a retroauricular-lateral neck incision that exposes the mastoid process and the transverse process of C1 at the lateral apex of the suboccipital triangle is used to gain access to the vertebral artery above and below the C1 transverse process. The vertebral foramen of the C1 transverse process is opened, and the vertebral artery is bypassed superiorly en route to the jugular process of the occipital bone (1, p. 628, Figure 7-2A, internal view, and p. 638, Fig 7-2K, external view). Detachment of the rectus capitis lateralis muscle exposes this area, the anterior margin of which forms the posterior (and inferior) rim of the jugular foramen. Following the internal jugular vein superiorly from below and the sigmoid sinus (exposed by a limited mastoidectomy [1, pp. 712, 716])

NEUROSURGERY

inferiorly from above leads to the jugular bulb. This exposure can be widened locally by removing the bone of the jugular tubercle and the occipital condyle. Superior extension of the tumor requires further drilling of the petrous temporal bone. After occlusion of the sigmoid sinus and internal jugular vein, extradural and then intradural tumor is removed. This is an anatomically reasonable approach to tumors involving the posterior and inferior jugular foramen. Its utility should be assessed in a series of cases analyzed for indications and outcomes. Griffith R. Harsh IV Stanford, California 1. Rhoton AL: Cranial Anatomy and Surgical Approaches. Schaumburg, Congress of Neurological Surgeons, 2003, pp. 627–641, 712, 716.


TUMOR Instrumentation, Technique, and Technology

THE NEW GENERATION POLESTAR N20 FOR CONVENTIONAL NEUROSURGICAL OPERATING ROOMS: A PRELIMINARY REPORT Vasileios Ntoukas, M.D. Department of Neurosurgery, Johann Wolfgang von Goethe University, Frankfurt am Main, Germany

Rene Krishnan, M.D. Department of Neurosurgery, Johann Wolfgang von Goethe University, Frankfurt am Main, Germany

Volker Seifert, M.D, Ph.D. Department of Neurosurgery, Johann Wolfgang von Goethe University, Frankfurt am Main, Germany Reprint requests: Vasileios Ntoukas, M.D., Department of Neurosurgery, Schleusenweg 2-16, 60528 Frankfurt am Main, Germany. Email: [email protected] Received, January 16, 2007. Accepted, July 10, 2007.

OBJECTIVE: The objective of this work is to present the preliminary clinical experience we acquired in using the new PoleStar generation, N20 (Medtronic Navigation, Louisville, CO), in a modified conventional operating room. METHODS: PoleStar N20 is a 0.15-T, intraoperative scanner combined with both an integrated optical and a magnetic resonance imaging tracking scanner. All standard imaging modes, such as T1, T2, and fluid-attenuated inversion recovery, are available through the magnet. To shield the operating room from radiofrequency interference, a Faraday cage was constructed using a conductive metal mesh installed under the wall decoration. Sixty-one patients, most of whom had gliomas or pituitary adenomas, underwent intraoperative magnetic resonance imaging in our clinic. The extent of resection and the surgical consequences of intraoperative imaging were analyzed. DISCUSSION: The image quality for T1-weighted, gadolinium-enhanced tumors was sufficiently good to enable us to evaluate the extent of tumor resection, whereas the T2weighted image quality must be improved. New technologies, such as high-temperature superconductive coils and ultra-small super-paramagnetic iron particles, e.g., ferumoxtran-10, can lead to a dramatic improvement in image quality, heralding the commencement of the widespread use of intraoperative magnetic resonance imaging. CONCLUSION: The acquisition of the PoleStar N20 opened new horizons in the treatment of our patients. This novel, compact, intraoperative magnetic resonance imaging scanner can be installed in a standard operating room without major modifications. Standard surgical instruments can be used. Intraoperative magnetic resonance imaging provided valuable information that allowed intraoperative modification of the surgical strategy. KEY WORDS: Intraoperative magnetic resonance imaging, Neuronavigation, Resection control Neurosurgery 62[ONS Suppl 1]:ONS82–ONS90, 2008

T

he beginning of intraoperative imaging as a means of quality control, providing the option of evaluating the extent of tumor resection, is now more than 2 decades behind us (1, 8, 10, 13, 14, 15, 22). Ultrasound and computed tomography were the first two imaging tools used by neurosurgeons, and they assisted in achieving greater extension and safer resection during surgery (1, 8, 25). After the first wave of enthusiasm died down, interest within the neurosurgical community decreased because of the lack of satisfactory quality in imaging and in lesion resolution. In the mid-1990s, however, with Black et al.’s (3) introduction of the first intraoperative magnetic resonance imaging (MRI) system (Signa SP; General Electric, Milwaukee, WI) which was developed in collaboration with General Electric


DOI: 10.1227/01.NEU.0000297016.81210.7E

Corporation, the desire for intraoperatively acquired imaging regained widespread interest. Since then, several new high- or low-field intraoperative MRI systems have been introduced, but only a few have managed to attain greater acceptance in neurosurgical operating rooms (7, 12, 16, 17, 18, 23, 24, 26). One of these is the PoleStar low-field system, manufactured by Medtronic Navigation (Louisville, CO), and first introduced by Hadani et al. (5) in 2001. The first PoleStar generation was called N10, and used a 0.12-T scanner. The limitations of this system were discovered during the subsequent years, with the result being that Medtronic Navigation developed the new generation, PoleStar N20, with a slightly stronger field (0.15 T) and better adjusted to the neurosurgical operating room and its needs and requirements.


NEW GENERATION POLESTAR N20

In this article, we introduce this new PoleStar generation, the N20, which has been used in our clinical routine since June 2004, and we report on our preliminary clinical experiences in 61 patients with various intracranial abnormalities.

DESCRIPTION OF INSTRUMENTATION The Device

TABLE 1. PoleStar N10 and N20 in comparison PoleStar N20

PoleStar N10

Field strength

0.15 T

0.12 T

5-G fringe field (radial/axial)

2.2 m

1.5 m

Field of view

16 20 cm

10 15 cm

Gantry driving mechanism

Electric and hydraulic

Electric

Magnets gap

27 cm

25 cm

Front-end low position height

103 cm

95 cm

Gap between gradient coils at shoulders

58 cm

48 cm

The PoleStar N20 is a novel, compact, mobile intraoperative guidance system developed by Medtronic Navigation primarily for intracranial surgery. This MRI scanner is combined with both an integrated optical and an MRI-tracking system and is operated by the neurosurgeon from an in-room computer workstation. The magnet gantry consists of two vertically oriented columns spaced 27 cm apart, which is 2 cm more than in the first generation, PoleStar N10. Gradient coils are mounted on the outside of the columns. At 0.15 T, the strength of the magnetic field is slightly stronger than the N10, which has a strength of 0.12 T. The 5-G fringe field is approximately 1.7 to 2.2 m from the magnets’ isocenter. Table 1 lists some of the new model’s advantages. Ferromagnetic instruments may be brought as close as 25 cm to the magnet without palpable attraction. The installation of the system in a conventional operating room with radiofrequency shielding is, therefore, possible. An important safety factor is that any ferromagnetic tools that are accidentally brought near the magnet will be attracted toward the closer pole and away from the head of the patient. In the operating room, the PoleStar N20 is normally stored in an iron cage (Fig. 1), which allows the room to be used for general neurosurgical procedures. For intraoperative use, the magnet can easily be removed from the cage and positioned underneath the operating table by one person. When the magnet is placed below the level of the operating table, the magnetic field strength in the surgical field is less than 50 g, which permits the use of standard surgical instruments. The gantry is moved in vertical and horizontal directions by means of a remote control device. During surgery, the magnet is usually stored underneath the table. For scanning, the magnet is moved upward to the scanning position, which is memorized, so that the magnet can be placed in exactly the same position for the next scan. This is important for “comparison mode:” the ability to compare exactly the same images acquired just before surgery and during resection control. In the event that the table has been moved to another position during surgery, a mode called “adjust to table” is used to obtain exactly the same scanning position. All of these steps can be controlled entirely at the surgeon’s direction by use of a simple remote control unit. There is no need for the simultaneous presence of a neuroradiologist or an MRI operator.

the relaxing protons. Electronic equipment operated near the scanner may emit radiofrequency noise that is picked up by the receiver-coil of the MRI scanner, adversely affecting the quality of the acquired images. To prevent this interference, a radiofrequency shielding enclosure must be erected around the MRI scanner, enclosing both the scanner and the patient to be scanned. Shielding can be obtained by constructing a Faraday cage that incorporates the entire room by use of a conductive metal mesh installed underneath the wall decoration. Copper shielding can be provided to seal all the openings of the room, including doors. Light and power inlets have to be provided with filters to avoid interference during image acquisition. This is the way we have shielded our operating room (Fig. 2). Another option is the so-called Star Shield. It does not require the entire operating room to be shielded, and it covers only the scanner and the operating table. This type of shield is also made in a copper mesh fabric.

Radiofrequency Shielding

Imaging

MRI scanners function by transmitting radiofrequency signals that are absorbed by the protons of the scanned tissue and by then subsequently receiving the return signals emitted by

According to the PoleStar concept, there is no need to acquire images from the whole brain because PoleStar is not a diagnostic scanner; instead, images are acquired only from the region

NEUROSURGERY

A

B

FIGURE 1. For intraoperative use, the magnet can easily be moved to the operating room (A), but the PoleStar N20 is normally stored in an iron cage (B).


NTOUKAS ET AL.

B

TABLE 2. PoleStar’s various imaging modesa Mode

A

a

Scan time/ slice thickness

T1

1 min / 8 mm, 3.5 min / 4 mm, 7 min / 4 mm, 3.5 min / 4 mm, 6 min / 3 mm, 13 min / 2 mm, 5 min / 8 mm

T2

1 min / 8 mm, 3.75 min / 8 mm, 7 min / 8 mm, 7 min / 5 mm, 13 min / 5 mm, 7 min / 4 mm, 13 min / 4 mm

FLAIR

4 min / 10 mm, 8.5 min / 5 mm

e-steady

8 s / 8 mm, 44 s / 6 mm, 3 min / 4 mm, 5 min / 4 mm

FLAIR, fluid-attenuated inversion recovery.

FIGURE 2. The operating room under construction (A) and the operating room blueprint (B).

of surgical interest. For this reason, the PoleStar N20 produces images with a field of view of 16 20 cm, which will include the cerebral lesion and the surrounding area. All standard imaging modes, such as T1, T2, and fluid-attenuated inversion recovery (FLAIR) imaging, are available through the magnet. A sequence that the vendor has labeled “e-steady” is also available, which combines features of T1- and T2-weighted sequences in its imaging of brain and fluid field spaces respectively. This sequence is generically known as a PSIF (inverted FISP, or “fast imaging-steady state processing”). Scan times range from 8 seconds to 13 minutes and slice thicknesses from 8 to 2 mm. A short, 8-second/8-mm e-steady sequence can be helpful when checking the position of the head inside the magnet, whereas longer sequences, for example 7 minutes/4 mm, are needed to visualize the cerebral lesion and to check the extent of the resection. Table 2 gives an overview of the various imaging modes.

Navigation A fully integrated tool for surgical navigation is included in the PoleStar N20. It is an infrared-based optical system with active cameras (Northern Digital, Waterloo, Canada) and a wand equipped with three reflecting spheres. The navigation camera emits infrared light, which is reflected by special reflective spheres mounted on PoleStar N20 instruments. By receiving the reflected light, the camera can track the location of these instruments. The PoleStar navigation however, does not use registration, i.e., point-to-point matching. The magnet position in space is tracked directly. As long as the patient’s head does not move relative to the patient reference frame, the infrared probe’s location in space will be accurately portrayed on the image. The magnet reference frame (MRF) enables the optical guidance system to track the position of the PoleStar magnets before imaging and to use this information together with the location of the patient reference frame to determine the location and orientation of the acquired slice. The MRF is also used in magnet positioning and in adjusting the gantry scan position in relation to the operating table. The MRF comprises 12 light-


FIGURE 3. Neuronavigation during the surgery.

emitting diodes (LEDs) mounted on the covers of the magnets and six LEDs on each side. To track the MRF, the navigation camera must “see” at least three of the LEDs. After image acquisition, accuracy is confirmed by touching known landmarks in the field of view with the navigation wand. During navigation with the optical wand, the magnet can be in the scan position or stored under the operating table. As has been previously described (5, 21), the system displays three perpendicular images in the multiplanar interactive navigation mode, which are reconstructed from the acquired spatial data. The images intersect at the tip of the wand. When the wand is moved, the images are recomputed and redisplayed in real time (Fig. 3).

PATIENTS AND METHODS Between June 2004, when we began performing intraoperative MRI, and February 2006, a total of 61 patients underwent intraoperative MRI in our clinic. The average age of these patients was 48.3 years. The group consisted of 37 men and 24 women. Histopathological examination revealed the lesions to be pituitary



adenomas (macroadenomas, n 15; microadenomas, n 3), highgrade gliomas (n 26), low-grade gliomas (n 5), meningiomas (n 3), cavernomas (n 2), metastases (n 5), and abscesses (n 2). Most of the pituitary adenomas were hormone-inactive (n 14), whereas three patients presented with acromegaly and one patient with Cushing syndrome. All three patients with meningiomas presented with convexity meningiomas, and either headache (two patients) or seizures (one patient) led to diagnosis. The two cavernomas were superficial and not deep seated. In both cases, headache was the symptom leading to diagnosis. Four patients with metastasis had pulmonary primary tumor and one patient had cerebral metastasis because of melanoma. Transmitted otitis was the cause for brain abscess in both patients. Prone, supine, park-bench, and Fukushima positioning (Fig. 4) can FIGURE 4. Patient positioning in the magnet. easily be performed inside the magnet. The only exception is the half-sitting position, which is not possible because the magnet cannot DISCUSSION be moved as high as is needed. The head is positioned using an MRIcompatible headholder with three-pin fixation, or a simple padded The PoleStar N20 has been successfully integrated and bowl. The time needed for positioning varies from 5 minutes in the case adapted into our neurosurgical routine. At first, however, it of simple supine positioning for pituitary surgery to 25 minutes for a park-bench position. In only one case was it impossible to place the was quite difficult to adapt to having only 27 cm in which to cerebral lesion in the magnet’s field of view. This was an obese man place the head between the two magnet poles or using a less with a pituitary adenoma and a very short neck. flexible head clamp. A learning curve was needed until we After the patient has been positioned as desired, the first scan is were able to position the patient as desired in that short time performed just before surgery. We want to observe by means of this (5–25 min). Since June 2004, we have been using the PoleStar scan if the lesion is placed in the field of view; if it is not, the head must N20 primarily for gadolinium-enhancing gliomas and pituitary be repositioned and another scan performed. At this stage, we usually surgery. During a 20-month period, the number of patients obtain one or two short (a few seconds) scans. The timing of resection (n 61) operated with PoleStar support seems to be quite control was chosen by the neurosurgeon at the stage he had the impressmall. There are two reasons for this. First, in our department, sion of complete tumor removal, or in the case of incomplete tumor we were the only neurosurgeons introduced to and able to use removal, when he thought that no more tumor removal was possible, the new technology; and second, we scheduled patients who e.g., because of an infiltration of eloquent brain areas. were to be operated with PoleStar support on days that were We did not observe any complications attributable to intraoperative MRI. Image quality was sufficient to allow evaluation of the extent of not busy surgical days to avoid time constraints when using tumor resection in the majority of the cases. Intraoperative imaging the new tool. In the beginning, every use of the PoleStar had to revealed remaining tumor in 12 patients (19.6%). This led to further be supervised by Medtronic technicians. tumor removal, resulting in total tumor removal in two patients (3.2%) As far as image quality is concerned, the borders of lowwith high-grade gliomas (glioblastomas) and subtotal tumor removal grade gliomas, which are often observed with good visibility in 10 patients (16.3%) (Figs. 5–7). in preoperative diagnostic T2-weighted images (1.5-T scanDuring resection of a low-grade glioma, we intraoperatively had ner), were unfortunately not easy to identify in low-field intrathe impression of total tumor removal, whereas PoleStar imaging operative T2-weighted images. In addition, FLAIR imaging revealed a suspicious hyperintense rim (Fig. 6A). To validate this infordid not reveal a significant improvement in image quality, mation, we navigated on those images (Fig. 6B). The tip of the navigaeither (Fig. 7). We tried to improve T2-weighted and FLAIR tion wand was beyond the suspicious hyperintense rim. Because of this image quality by extended imaging time or use of e-steady contrary information, we choose not to pursue resection. The surgical data are summarized in Table 3. without success. By comparing the resection control imaging

NEUROSURGERY


NTOUKAS ET AL.

FIGURE 5. Patient with a glioblastoma. 1.5-T diagnostic preoperative (A) and postoperative (E) imaging (T1 with gadolinium). PoleStar N20 imaging

with the postoperative 1.5-T imaging, the inferiority of PoleStar imaging becomes clear again. We are, therefore, skeptical regarding the use of the PoleStar N20 for low-grade gliomas. Comparing the results of intraoperative imaging with pre- and postoperative standard radiological imaging (1.5 T), we found good correlation for gadolinium-enhancing tumors, as shown in Figure 5A. For those tumors, pre- and postoperative use of 1.5-T imaging confirmed the PoleStar N20 imaging. Therefore, we view PoleStar intraoperative imaging for gadolinium-enhancing tumors as a useful tool to help the neurosurgeon achieve a more extended and safer resection. Although other users have suggested reliability problems with the PoleStar, we cannot confirm this (personal communication). We did not notice a system malfunction compelling us to switch to a conventional procedure without intraoperative imaging. The only exception was a patient with a pituitary adenoma and a very short neck. As described previously by other authors, the PoleStar can also be used for frameless tumor biopsy (2, 11) or for acquisition of functional data (6, 20), but neither of these options has been introduced into our operating room. The latter is theoretically possible but, as Schulder et al. (20) have stated, further work is needed to begin testing the concept with patients. Comparing PoleStar with Signa SP, also known as the “double donut,” it becomes clear that two different philosophies comprise each project. The “double donut” is an intermediatefield nonmobile MRI scanner, whereas the PoleStar is a lowfield mobile system. PoleStar is operated by the neurosurgeon


(T1 with gadolinium) for preoperative (B), first (C), and second (D) resection control. D and E, complete tumor resection.

from an in-room computer workstation, whereas radiological support is needed to use the “double donut.” The major advantage of Signa SP is better image quality for low-grade gliomas. High-grade gliomas are comparably visible with both systems. The major disadvantage of Signa SP is the limited space for the surgeon to operate, particularly in transsphenoidal surgery. During surgery, the PoleStar is stored underneath the table so the surgeon can operate freely. Further disadvantages of the Signa SP are the imperative use of MRI-compatible surgical instruments and longer perioperative time. Although most intraoperative MRI systems are designed primarily for neurosurgical use, they are in principal applicable to surgical procedures in many other anatomic regions, and they have been used successfully in otorhinolaryngological and breast surgery procedures (4). PoleStar was designed primarily for neurosurgical use and, therefore, its limited field of view may decrease the ability to use this system in other surgical subspecialties, such as urology. This limitation may negatively influence hospital management regarding the acquisition of this technology. Intraoperative imaging during surgical procedures offers the advantages of more accurate intraoperative lesion localization, intraoperative guidance of the surgical approach, intraoperative assessment of tumor resection, and intraoperative exclusion of procedure-related complications. Of all the various imaging techniques, MRI has become the method of choice for guidance during neurosurgical procedures because of its superior soft tissue contrast and multiplanar imaging ability.



A

TABLE 3. Surgical data Type of surgery

No. of patients

Craniotomy

41

Transsphenoidal

18

Burr hole (abscess drainage)

2

Patient position

Supine

26

Prone

11

Park bench

23

Fukushima

1

No. of scan sessions

Range (minimum–maximum)

3–8

Effect of imaging on surgery

Additional resection

12

Glioma

8

Pituitary adenoma

4

B

FIGURE 6. A, preoperative and resection control PoleStar N20 T2weighted imaging for a low-grade glioma. B, navigation on the suspicious hyperintense rim.

As is true of any new technology, the benefits of intraoperative MRI must be analyzed carefully to ensure appropriate use. Many neurosurgical procedures do not require real-time image guidance and can be performed safely with conventional surgical techniques. Other tumor resections, tumor biopsies, and surgical and interventional procedures benefit distinctly from the sophisticated information provided by intraoperative imaging techniques. In surgery for low-grade gliomas, intraoperative MRI has found general acceptance, whereas its usefulness in monitoring resection of high-grade gliomas remains controversial. As mentioned previously, the borders of low-grade gliomas, which often are seen with good

NEUROSURGERY

FIGURE 7. Preoperative and resection control PoleStar N20 fluidattenuated inversion recovery imaging for a low-grade glioma.

visibility in preoperative diagnostic T2-weighted images (1.5-T scanner), were not easy to identify in PoleStar N20’s intraoperative T2-weighted images. Bearing in mind the importance of intraoperative imaging for low-grade tumors, this issue must be resolved. Our department is, therefore, in close contact and cooperation with Medtronic Navigation in an attempt to improve the imaging quality.


NTOUKAS ET AL.

Even the administration of gadolinium for the visualization of high-grade tumors can cause a problem. For example, induced enhancement of the operative site caused by vascular leak, luxury perfusion, or other changes can cause a major problem with regard to the accurate interpretation of resection control images. Especially if more than one intraoperative assessment of the resection is needed, gadolinium may need to be dosed more than once, which can aggravate that problem. However, Hunt et al. (9) have reported ultra-small superparamagnetic iron particles, such as ferumoxtran-10, which have a long plasma half-life and are trapped by reactive cells within the tumor. These trapped particles provide a method to demonstrate enhancing lesions without the conjuncture of repeated gadolinium administration in the face of blood-brain barrier and vascular injury. However, more studies will be needed to corroborate that statement. The use of high-temperature superconductive (HTS) coils can increase image quality even more (19). Superconductivity, whereby electricity is conducted with greater speed and efficiency at temperatures approaching absolute zero, is one way to increase the signal-to-noise ratio of MRI acquisition, thereby improving image quality. However, maintaining these extremely low temperatures is not practical in an operating room. Use of HTS coil technology can overcome this problem. Supertron Technologies Inc., based on the campus of the New Jersey Institute of Technology, has developed small HTS coils that display dramatic improvements in MRI quality compared with standard copper coils. The HTS coils can also be used to decrease scan times, allowing us to approach real-time intraoperative MRI and to image the entire volume of the brain without enlarging the magnet itself. However, more studies are needed to encourage and support this promising technology. We emphasize that neither ultra-small super-paramagnetic iron particles nor HTS coils have been used in our department. However, we think that these technologies may potentially be helpful to improve image quality. Another important aspect that cannot be overlooked is the question of economy. Comparing different intraoperative imaging systems, it costs approximately $200,000 (US) for ultrasound, $600,000 for computed tomography, and $1 to 7 million for intraoperative MRI. PoleStar N20 implementation costs approximately $1 million. Despite these additional expenses, intraoperative MRI can lead to a significant overall cost reduction in the treatment of selected patients, if a long-term cure can be achieved, if repeat resection can be avoided, or if procedureassociated morbidity can be reduced. That, however, still remains to be proved.

CONCLUSION The acquisition of the PoleStar N20 opened new horizons in the treatment of our patients. This novel, compact, intraoperative MRI-guided system can be installed in a standard operating room without major modifications. The operating room can also be used for general neurosurgical procedures. Standard surgical instruments can be used. A short learning curve was


needed to adjust to the new perioperative circumstances. The image quality for T1 gadolinium-enhancing tumors was sufficiently good to enable us to evaluate the extent of tumor resection, whereas T2-weighted image quality definitely must be improved. Imaging depicting incomplete resection offers the chance of further tumor removal during the same operation. New technologies, such as high-temperature superconductive coils and ultra-small super-paramagnetic iron particles, for example, ferumoxtran-10, can lead to a dramatic improvement in image quality, heralding the commencement of widespread use of intraoperative MRI.

REFERENCES 1. Apuzzo MLJ, Sabshin JK: Computed tomographic guidance stereotaxis in the management of intracranial mass lesions. Neurosurgery 12:277–285, 1983. 2. Bernays RL, Kollias SS, Khan N, Romanowski B, Yonekawa Y: A new artifactfree device for frameless, magnetic resonance imaging-guided stereotactic procedures. Neurosurgery 46:112–117, 200 3. Black PM, Moriarty T, Alexander E, Stieg P, Woodard EJ, Gleason PL, Martin CH, Kikinis R, Schwartz RB, Jolesz FA: Development and implementation of intraoperative magnetic resonance imaging and its neurosurgical applications. Neurosurgery 41:831–845, 1997. 4. Fenchel S, Boll DT, Lewin JS: Intraoperative MR imaging. Magn Reson Imaging Clin N Am 11:421–447, 2003. 5. Hadani M, Spiegelman R, Feldman Z, Berkenstadt H, Ram Z: Novel, compact, intraoperative magnetic resonance imaging-guided system for conventional neurosurgical rooms. Neurosurgery 48:799–809, 2001. 6. Hajnal JV, Collins AG, White SJ, Pennock JM, Oatridge A, Baudouin CJ, Young IR, Bydder GM: Imaging of human brain activity at 0.15T using fluid attenuated inversion recovery (FLAIR) pulse sequences. Magn Reson Med 30:650–653, 1993. 7. Hall WA, Kowalik K, Liu H, Truwit CL, Kucharezyk J: Costs and benefits of intraoperative MR-guided brain tumor resection. Acta Neurochir Suppl 85:137–142, 2003. 8. Hammoud MA, Ligon BL, elSouki R, Shi WM, Schomer DF, Sawaya R: Use of intraoperative ultrasound for localizing tumors and determining the extent of resection: A comparative study with magnetic resonance imaging. J Neurosurg 84:737–741, 1996. 9. Hunt MA, Bagó AG, Neuwelt EA: Single-dose contrast agent for intraoperative MR imaging of intrinsic brain tumors by using ferumoxtran-10. AJNR Am J Neuroradiol 26:1084–1088, 2005. 10. Jolesz FA, Shtern F: The operating room of the future: Report of the National Cancer Institute Workshop, “Imaging-Guided Stereotactic Tumor Diagnosis and Treatment.” Invest Radiol 27:326–328, 1992. 11. Kollias SS, Bernays R, Marugg RA, Romanowski B, Yonekawa Y, Valavanis A: Target definition and trajectory optimization for interactive MR-guided biopsies of brain tumors in an open configuration MRI system. J Magn Reson Imaging 8:143–159, 1998. 12. Levivier M, Wikler D, De Witte O, Van de Steene A, Balériaux D, Brotchi J: PoleStar N-10 low-field compact intraoperative magnetic resonance imaging system with mobile radiofrequency shielding. Neurosurgery 53:1001–1007, 2003. 13. Lunsford LD, Parrish R, Albright L: Intraoperative imaging with a therapeutic computer tomographic scanner. Neurosurgery 15: 559–561, 1984. 14. Nimsky C, Ganslandt O, Fahlbusch R: Functional neuronavigation and intraoperative MRI. Adv Tech Stand Neurosurg 29:229–263, 2004. 15. Nimsky C, Ganslandt O, Kober H, Buchfelder M, Fahlbusch R: Intraoperative magnetic resonance imaging combined with neuronavigation: A new concept. Neurosurgery 48:1082–1091, 2001. 16. Nimsky C, Ganslandt O, Fahlbusch R: Comparing 0.2 Tesla with 1.5 tesla intraoperative magnetic resonance imaging analysis of setup, workflow, and efficiency. Acad Radiol 12:1065–1079, 2005. 17. Nimsky C, Ganslandt O, von Keller B, Fahlbusch R: Intraoperative high-field MRI: Anatomical and functional imaging. Acta Neurochir Suppl 98:87–95, 2006.



18. Rubino GJ, Farahani K, McGill D, Van De Wiele B, Villablanca JP, WangMathieson A: Magnetic resonance imaging-guided neurosurgery in the magnetic fringe fields: The next step in neuronavigation. Neurosurgery 46:643– 654, 2000. 19. Schulder M: Seeing more during brain surgery. UMDNJ Res 4:47–48, 2002. 20. Schulder M, Azmi H, Biswal B: Functional magnetic resonance imaging in a low-field intraoperative scanner. Stereotact Funct Neurosurg 80:125–131, 2003. 21. Schulder M, Catrambone J, Carmel PW: Intraoperative magnetic resonance imaging at 0.12 T: Is it enough? Neurosurg Clin N Am 16:143–154, 2005. 22. Seifert V, Zimmermann M., Trantakis C, Vitzthum HE, Kühnel K, Raabe A, Bootz F, Schneider JP, Schmidt F, Dietrich J: Open MRI-guided neurosurgery. Acta Neurochir (Wien) 141:455–464, 1999. 23. Steinmeier R, Fahlbusch R, Ganslandt O, Nimsky C, Buchfelder M, Kaus M, Heigl T, Lenz G, Kuth R, Huk W: Intraoperative magnetic resonance imaging with the magnetom open scanner: Concepts, neurosurgical indications, and procedures: A preliminary report. Neurosurgery 43:739–748, 1998. 24. Sutherland GR, Kaibara T, Louw D, Hoult DI, Tomanek B, Saunders J: A mobile high field magnetic resonance system for neurosurgery. J Neurosurg 91:804–813, 1999. 25. Unsgaard G, Ommedal S, Muller T, Gronningsaeter A, Nagelhus Hernes TA: Neuronavigation by intraoperative three-dimensional ultrasound: Initial experience during brain tumor resection. Neurosurgery 50:804–812, 2002. 26. Wirtz CR, Tronnier VM, Bonsanto MM, Knauth M, Staubert A, Albert FK, Kunze S: Image guided neurosurgery with intraoperative MRI: Update of frameless stereotaxy and radicality control. Stereotact Funct Neurosurg 68:39–43, 1997.

COMMENTS

T

his article describes early experiences in Frankfurt with the PoleStar N20 intraoperative magnetic resonance imaging (iMRI) system. This compact iMRI is based on a 0.15-T magnet, an increase from the 0.12-T strength of its predecessor. The authors describe some of the functions of the system along with an overview of their imaging and navigation techniques. Over a 20-month period, 61 patients had iMRI-guided craniotomy or transsphenoidal resection using the PoleStar N20. These included 31 patients with glioma, 18 with pituitary adenoma, and 13 with other diagnoses. Intraoperative imaging showed residual tumor in 19.6% of patients; as a result, additional, subtotal removal was performed in 16.3% and total removal was achieved in 3.2%. The authors are frank in discussing the limitations in their experience with this “low-field” iMRI. In particular, they feel that the image resolution is insufficient to guide resection in patients with low-grade gliomas. Undoubtedly, image quality with diagnostic or high-field iMRI will be better. However, the authors’ own intraoperative images (Figs. 6 and 7) appear to quite convincingly demonstrate residual tumor. The logic of intraoperative imaging in neurosurgery would seem to be inescapable. For the foreseeable future, magnetic resonance imaging (MRI) will provide the best imaging for intracranial surgery. We all obtain scans of our patients before and after surgery. Who would not want to know right there in the operating room what the scan looks like and not be surprised the next day or worse, have a patient awaken with a new deficit caused by unnecessarily aggressive surgery? On the other hand, the costs of iMRI systems range from very expensive to very, very expensive. Ntoukas et al. do well to point out that we are still in the early stages of iMRI technological development and clinical assessment. To justify the cost of these tools, rigorous studies will be needed to prove their worth—or not. Michael Schulder Newark, New Jersey

N

toukas et al. present some preliminary data on the application of a low-field 0.15-T mobile magnetic resonance scanner for intraop-

NEUROSURGERY

erative imaging. Despite impaired intraoperative image quality compared with standard pre- and postoperative high-field MRI, they found some benefit for patients with contrast-enhancing tumors. Investigation of iMRI started more than 12 years ago. Meanwhile, common sense indicates that the intraoperative application of imaging acts as immediate intraoperative quality control. The unwanted occurrence of finding residual tumor on a postoperative scan is thus practically eliminated. As a result, the surgical goal of complete or optimal resection can be achieved without any guesswork, if intraoperative imaging quality can compete with the imaging quality of standard pre- and postoperative neuroradiological diagnostic images. Whether this is true for the system presented seems to be doubtful. A detailed analysis comparing the intraoperative findings with postoperative high-field MRI would be necessary to clearly define the applications in which the decreased image quality of the low-field system is still sufficient to evaluate the completeness of a resection. As pointed out by the authors, this seems to be untrue for the low-grade gliomas. Whether it is true for the evaluation of the resection of pituitary adenomas is also doubtful. Of course, the removal of some suprasellar tumor parts may be evaluated reliably by low-field imaging; however, intra- and parasellar evaluation seems not to be possible without doubt. In addition, small remnants of craniopharyngiomas seem not to be depicted reliably by the low-field MRI machines. In contrast to low-field iMRI, high-field iMRI enables intraoperative imaging at a high image quality that is up to the standard of up-to-date pre- and postoperative neuroradiological routine diagnostic images. In addition to achievement of the optimal extent of a resection, e.g., in glioma surgery, simultaneous preservation of functional integrity is possible, as high-field MRI offers various modalities beyond standard anatomical imaging, such as magnetic resonance spectroscopy, diffusion tensor imaging, and functional MRI, which may also be applied intraoperatively. These modalities provide not only data on the extent of resection and localization of tumor remnants but also data on metabolic changes, tumor invasion, and localization of functional eloquent cortical and deep-seated brain areas. It is mandatory to combine the goal of maximum resection with the goal of preservation of function, especially in glioma surgery. Highfield iMRI not only helps to maximize the extent of resection, but in combination with functional multimodality navigation, a minimization of postoperative neurological deficits is possible. MRI-guided removal of local tumor remnants may be critical in an interdisciplinary concept of glioma treatment in which the maximum extent of tumor volume reduction with the fewest neurological deficits is the optimal starting point for further sophisticated adjuvant therapeutic regimens. Christopher Nimsky Erlangen, Germany

T

he authors report on their early experience with the second iteration of a low-field MRI system. They find it to be a usable system, but they have had difficulty applying it to assess extent of resection in lowgrade gliomas—unfortunately, an (if not the most) important indication for intraoperative imaging. We have had better success by using extended, high-resolution T2 imaging with our PoleStar N20 and N10. Unlike the authors, we have had serious reliability problems with the N20 (and N10 since our previous report) over a period of a few years. This situation has led to substantial reduction in use and acceptance of these systems at our institution. Fortunately, recent repairs have led to better usability of our system as this device is really a technological marvel and, when it is functioning, works well as a solution for iMRI in existing operating room suites. Gene H. Barnett Cleveland, Ohio


NTOUKAS ET AL.

W

ith this article, the authors contribute to the growing volume of literature validating and supporting the performance of iMRIguided neurosurgery. It is interesting to note that as other groups are moving toward the investigation of increasing field iMRI-guided neurosurgical systems such as those at 3 T, these authors continue to promote the virtues of an ultra-low 0.15-T imaging system (1, 6). There are clear advantages and disadvantages to any iMRI-guided neurosurgical system that is examined, some of which are valid and some of which are more perceived or theoretical. Some advantages associated with the PoleStar N20 over higher field systems include the lower initial expenditure that is required by the purchasing institution to implement the program, the fact that the technology is primarily directed to neurosurgical usage (this could be considered a disadvantage by some), and the ability to use standard surgical instruments in the magnetic field. Disadvantages are the diminished functionality (magnetic resonance venography, magnetic resonance angiography, brain activation studies [functional MRI], diffusion-weighted imaging, and magnetic resonance spectroscopy) that is possible with high-field systems, longer scan times necessary for image acquisition, the inability to view the entire head on any single scan, and a decreased signal-to-noise ratio with the resultant reduction in image quality. High-field (>1.5 T) iMRI systems are clearly more costly than lowfield systems yet they can be used for routine diagnostic imaging when they are not being used for surgery, preventing the system from being considered a “sunk” cost (2, 5). These systems also require the use of MRI-compatible instrumentation if surgery is performed within the 5G line. Any part of the body can be imaged with a diagnostic quality high-field system, which expands the application of this technology to allow for other surgical services to take advantage of the superb soft tis-

sue visualization that assures that the goals of surgery are accomplished before the completion of the procedure. Although there were early concerns regarding potential safety issues related to performing neurosurgery in a high-field MRI environment, there have been no untoward events reported to date after more than 10 years of surgical experience (4). As a result of the safety of operating in a high-field iMRI environment and the excellent surgical results that have been reported, there are now five iMRI systems in operation in the Twin Cities (3). The work reported by these authors in their first 61 patients further strengthens the rationale for performing neurosurgery using iMRI guidance. Walter A. Hall Syracuse, New York

1. Hall WA, Galicich W, Bergman T, Truwit CL: 3-Tesla intraoperative MR imaging for neurosurgery. J Neurooncol 77:297–303, 2006. 2. Hall WA, Kowalik K, Liu H, Truwit CL, Kucharczyk J: Costs and benefits of intraoperative MR-guided brain tumor resection. Acta Neurochir [Suppl] 85:137–142, 2002. 3. Hall WA, Truwit CL: 3-Tesla functional magnetic resonance imaging-guided tumor resection. Int J Comput Assist Radiol Surg 1:223–230, 2007. 4. Hall WA, Truwit CL: Intraoperative MR-guided neurosurgery. J Magn Reson Imaging 27:368–375, 2008. 5. Kucharczyk J, Hall WA, Broaddus WC, Gillies GT, Truwit CL: Cost-efficacy of MR-guided neurointerventions. Neuroimaging Clin North Am 11:769–774, 2001. 6. Truwit CL, Hall WA: Intraoperative magnetic resonance imaging-guided neurosurgery at 3-T. Neurosurgery 58 [Suppl 2]:ONS338–ONS346, 2006.

WEB SITES OF INTEREST The websites featured in this announcement are provided strictly for informational purposes. NEUROSURGERY assumes no responsibility in regards to the validity of the presented information. 1. The Brain Aneurysm Foundation http://www.bafound.org/ This is a well organized, patient oriented site dedicated to cerebral aneurysmal disease. The Information section contains a series of common definitions and answers to frequently asked questions. The Support/Recovery tab list a number of resources for patients recovering from treatment and an extensive list of regional support groups. 2. The Parkinson’s Disease Foundation http://www.pdf.org/ This site is designed for patients, clinicians, and researchers. The home page displays a series of contemporary news items related to Parkinson’s disease (PD). The site houses literally hundreds of FAQs about PD. It also features an “Ask the Expert” section where inquiries can be submitted. This foundation offers a number of funding opportunities for Parkinson’s disease research. 3. Acoustic Neuroma Association http://www.anausa.org/ The Acoustic Neuroma Association is a member service organization composed mostly of patients with acoustic neruoma. The site has a comprehensive glossary of terms and review of each of the typical treatment options. Although it includes a list of practitioners, the list is not very comprehensive. Still, this is a good resource for patients contemplating or recovering from treatment. JOEL D. MACDONALD, M.D. [email protected]



TUMOR Instrumentation, Technique, and Technology

A COMPARATIVE ANALYSIS OF COREGISTERED ULTRASOUND AND MAGNETIC RESONANCE IMAGING IN NEUROSURGERY Alex Hartov, Ph.D.

Reprint requests: Alex Hartov, Ph.D., Thayer School of Engineering, Dartmouth College, HB 8000, Hanover, NH 03755. Email: [email protected]

OBJECTIVE: This work presents qualitative and quantitative side-by-side comparisons of oblique coregistered magnetic resonance imaging (MRI) scans and ultrasound images obtained during 35 neurosurgical procedures. METHODS: Spatially registered series of ultrasound images were recorded for subsequent off-line evaluation and comparison with corresponding preoperative MRI studies. The degree of misalignment was reduced by reregistering the target volume directly with segmented features. RESULTS: The initial apparent spatial misalignment of the target volume after craniotomy ranged from 0.11 to 8.73 mm (mean, 4.01 mm). After reregistration, the mutual information in overlapping segmented features was increased, presumably evidence of a better alignment locally. Additionally, the degree of feature congruence, which was assessed quantitatively through a convex hull approximation, demonstrated that the ultrasound volume was consistently smaller than its MRI counterpart. CONCLUSION: Although intraoperative ultrasound tends to be difficult to interpret by itself, when accurately coregistered with preoperative MRI scans, its potential utility as a navigational guide is enhanced.

Received, August 14, 2006.

KEY WORDS: Coregistered ultrasound, Intraoperative imaging, Neurosurgery, Three-dimensional tracking, Three-dimensional ultrasound

Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire

David W. Roberts, M.D. Dartmouth Medical School, Lebanon, New Hampshire

Keith D. Paulsen, Ph.D. Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire

Accepted, June 7, 2007.


I

t is now well understood that preoperative imaging studies constitute an increasingly inaccurate basis for neuronavigation as time progresses during a procedure. As surgery proceeds, displacement of the parenchyma resulting from bulging, sagging, retraction, or resection occurs concurrently with the operation. The extent of displacement can be significant, reaching 1 or 2 cm in some studies (12, 15, 30). Coregistration used to align the patient in the world, or operating room (OR), coordinate system with the preoperative imaging reference frame (referred to as magnetic resonance space here for convenience, recognizing that any imaging modality could be used) consists of computing a rigid spatial transformation that merges their coordinates on the basis of a best match of specially designated homologous points, typically fiducial markers on the patient’s skin, defined in both frames of reference (1). The procedure assumes a rigid system, which is a good approximation for the nondeformable cranium but is increasingly erroneous with respect to the parenchyma as the surgery progresses. Intraoperative magnetic resonance imaging (MRI) has been proposed as a method to update the imaging information so that it reflects the current state of the operating field (7, 26, 27, 34). Although able to address the majority of problems result-

NEUROSURGERY

DOI: 10.1227/01.NEU.0000296955.22901.8F

ing from intraoperative motion, this approach suffers from high cost (both capital and recurring) and relatively poor temporal resolution. Open magnet designs used in some installations exhibit poor image quality and significant geometric distortion, which may limit their use for spatial registration. However, large fixed-coil systems generally impede the surgeon’s movement. Coregistered ultrasound (US) has been proposed by several groups as an alternative intraoperative imaging modality that can be used for guidance in neurosurgery (6, 8, 11, 16, 23) and in other procedures in which deformation and motion are problematic (17, 22), e.g., in procedures involving the liver (4, 19). It is significantly less expensive and is generally more accessible to surgeons. However, the contrast resolution, especially in the soft-tissue parenchyma, is much worse than MRI and leads to relatively featureless images. When unaided by anatomic context, US images are more difficult to interpret and use for navigational guidance, which places a premium on maintaining registration with the feature-rich MRI. The extent to which intraoperative US, when coregistered with preoperative MRI, provides adequate neuronavigation in the OR is not well studied. A starting point for such an evalu-


HARTOV ET AL.

ation begins by quantifying the degree of feature correspondence between the two techniques. Additionally, intraoperative US can be used to improve or maintain image registration through realignment of features visible in both image sets after surgery begins, at least until significant brain deformation occurs. In this study, we set out to view side-by-side images of the same features in MRI and US. Because US images are planar (B scans) and MRI scans are three-dimensional (3-D), we recreated oblique sections of the MRI scans that best corresponded to the US image planes. Further details are provided in the Materials and Methods section. Several articles have presented results comparing the appearance of specific tumor types on computed tomographic or MRI scans versus US (2, 20, 21, 29, 31, 32). In most cases, lesions evident on MRI scans are also visible intraoperatively on US. However, most of these studies did not use coregistration but relied on experts to properly identify the corresponding views for comparison. In addition, the motivations for the various studies were different, such as determining the suitability of US for guiding resection (29). Using spatial coregistration, we found it possible to present side-by-side comparisons of US images and their corresponding oblique section in the MRI volume, thus enabling a more quantitative evaluation of the feature correspondence between the pre- and intraoperative imaging modalities. In this work, we present a study of 35 surgical procedures involving tumor removal in which high-resolution preoperative imaging (MRI) was coregistered with intraoperative US. Tumor volumes evident in the two modalities are manually outlined. Mutual information contained in the overlapping volumes is also assessed to quantitatively compare the degree of feature congruence between the two imaging modalities. Fiducial and target registration errors are also estimated in an effort to identify the sources and relative magnitudes of the contributions to feature misalignment between the two imaging studies.

PATIENTS AND METHODS

TABLE 1. Summary of patients analyzed Patient Age no. (yr)/sex

Rigid Body Coregistration All patients had MRI scans within 2 hours before being transported to the OR. MRI studies consisted of 124 axial slices, with 1.5-mm slice spacing and thickness. Each image was 256 ⫻ 256 16-bit square pixels 0.9375 mm on a side. Images were acquired with the use of a head coil. The reconstruction diameter was 240 mm. The MRI scans were T1-weighted spoiled gradient, 3-D volume acquisitions with repetition time values of 25 (34 patients) and 29 (one patient); echo time values of 8 (one patient), 3 (10 patients), and 6 (24 patients); the flip angle was 45 in all cases. The MRI scans used in this study were acquired specifically for the purpose of image-guided surgery, more specifically for the iden-


Location

1

70/M

Glioblastoma multiforme

Left frontal

2

51/M

Anaplastic astrocytoma

Right frontal

3

54/F

Metastasis

Right frontal

4

53/F

Metastasis

Right frontal & left temporal

5

66/M

Meningioma

Right posterior fossa

6

53/F

Meningioma

Left frontal

7

76/M

Metastasis

Right parietal

8

55/F

Metastasis

Left frontal

9

38/M

Choroid plexus papilloma

Left temporal

10

35/F

Cortical dysplasia

Right fronto-temporal

11

52/M


Right temporal

12

62/M

Metastasis

Right posterior temporal

13

61/F

Metastasis

Right occipital

14

20/M

Hemangioblastoma

Right cerebellum

15

50/M

Metastasis

Left frontoparietal

16

60/F

Meningioma and aneurysm

Right sphenoid wing

17

44/F

Metastasis

Left parietal

18

41/F

Meningioma

Left parasagittal

19

38/F

Meningioma

Left anterior fossa

20

50/F

Meningioma

Left fronto-parietal

21

49/F

Ganglioglioma

Right parietal

22

49/F

Meningioma

Right parietal

23

61/F

Meningioma

Left parasagittal

24

18/M

Dysembryoplastic neuroepithelial tumor

Left temporal

25

36/M

Meningioma

Right fronto-parietal

26

31/F


Right parietal

27

52/F

Meningioma

Right posterior fossa

28

55/F

Meningioma

Right frontal

29

62/F

Malignant glioma

Right temporal

30

58/F

Meningioma

Right parafalcine

31

57/M

Oligodendroglioma

Left frontal

32

49/M

Cavernous hemangioma

Left temporal

33

53/M

Meningioma

Right parasagittal

34

61/M


Right frontal

35

50/M

Metastasis

Left parietal

Patient Population We conducted this study with 35 patients (16 men, 19 women) ranging in age from 18 to 76 years (median age, 52 years). Patients presented with an assortment of diagnoses consisting of resectable tumors, except one case involving an epilepsy patient scheduled for cortical electrode placement. Table 1 summarizes the procedures analyzed.

Diagnosis


THREE-DIMENSIONAL ULTRASOUND AND MAGNETIC RESONANCE IMAGING COMPARATIVE ANALYSIS

A

B

FIGURE 1. A, identification of the fiducial markers in the magnetic resonance imaging (MRI) scan. Four markers are intersected by this slice, two of which have their center on that slice (circled). Note also the outlined tumor, a meningioma. B, three-dimensional (3-D) rendition of the markers on a view of the patient’s head (black circles). tification of our fiducial markers. As is our practice for those cases, contrast (gadolinium) was administered to all the patients for these specific studies. Five to 15 markers were placed on the patient’s head for the purpose of coregistration. These markers consisted of 8-mm diameter high-contrast loops mounted on an adhesive pad. The markers were visible in both MRI and computed tomographic scans and were readily identifiable on the patient’s skin in the OR. Before the start of surgery, the preoperative MRI scan was transferred to the OR workstation for processing. The fiducial registration markers were identified manually on all image slices where they appeared, and their coordinates were initially expressed in voxels. Image slices were translated to millimeter spatial coordinates (x, y, z) with the pixel size and slice spacing information in the MRI header. In the OR, the patient’s head was fixed in a Mayfield clamp and oriented appropriately for the procedure. A dedicated tracker was rigidly attached to the clamp so that patient movements were tracked during the procedure (e.g., adjustments of the operating table). The skinattached fiducial markers are located in the OR space using a 3-D tracked stylus. All spatial tracking operations are performed with a Polaris Hybrid system (Northern Digital Inc., London, Canada), a computer, video acquisition capabilities, and a frame grabber. The registration steps are illustrated in Figure 1. With the coordinates of the fiducial markers expressed in both frames of reference (MRI and OR), the rigid transformation that best merges them is computed. The details of this procedure are found in the work of Hajnal et al. (10). Normally, it is necessary to keep track of the correct pairing between MRI-defined fiducial markers and their OR-defined homologs. This, at times, has resulted in errors; even if no errors occur, the process can be time consuming while in the OR. It is particularly challenging in cases in which many markers are identified in the MRI study and only a subset are usable for registration in the OR, when the patient is positioned in the head clamp. To simplify the process, we use an approach that searches the many possible permutations of the MRI and OR points using a genetic algorithm (GA) (25). The GA does not compute the registration; instead, it is used as a method to converge rapidly on an acceptable set or subset of the points that produce a usable registration. In this problem we have to find not only the right sets of points but also

NEUROSURGERY

their correct pairings, which constitutes a potentially very large exhaustive search. The GA is an effective tool to rapidly reduce the search space. In this fashion, we find homologous sets of points that minimize the fiducial registration error (FRE) (9). It converges to the correct homologous sets of points provided that the markers exhibit some asymmetry in their placement, which is not difficult to ensure in practice. The randomized search typically takes between 10 and 200 seconds, using Matlab (Mathworks, Inc., Natick, MA). A graphical display of the results of the registration (Fig. 1B) together with values for FRE and an indication of the markers that are being used allows the user to evaluate the quality of the match.

Intraoperative Image Acquisition After registration, the surgeon performs a craniotomy through which the procedure is planned and US images are recorded. For this study, we used a Sonoline Sienna US system (Siemens Medical Solutions USA, Inc., Malvern, PA) with a curved 5.0 to 8.0 MHz transducer (Model C8–5, Fig. 2A). The scanhead is coupled to a rigidly attached 3-D tracker to which a frame of reference is associated. The tracker consists of a group of infrared light-emitting diodes mounted on a rigid armature, which are sensed by the Polaris system. The Polaris camera, which tracks the tools in the OR, defines the OR frame of reference or space, whereas the tracker attached to the US scanhead defines the US frame of reference. The spatial relationship between any pixel in an US image and the US frame of reference of the scanhead tracker is fixed and can be calibrated. Different methods have been presented to perform the calibration (8, 28); our scheme is inspired by the latter. It is accomplished by taking images of a set of wires arranged into an “N” configuration in a specially constructed tank. The position of the tank in relation to the 3-D tracker (world coordinates) is recorded; as a result, the position of the “N” is completely defined in that space. Similarly, the ratio of the distances between the wire intersections in the US imaging planes defines how far the intersection is along the middle segment of the “N.” With this information it is possible to spatially relate a point in the image to its position in absolute space to calibrate the scanhead-tracker assembly. For each image scale available on the US system (each scale corresponds to a pixel size), we acquire between 50 and 100 images. The calibration algorithm automatically identifies the scale that is displayed in the images. The intersection points of the wires in these images are automatically detected for each image. The transformation matrix relating image pixels to world coordinates is based on a formulation we presented previously (13), modified to enforce uniform scaling and orthogonality using singular value decomposition. The scanhead calibration is performed outside of surgery and needs to be performed only once as long as the rigid coupling between the scanhead and the tracker is not altered. It should also be mentioned that we use a uniform correction factor of 1500/1540 in our calibration meas-


HARTOV ET AL.

US scale is displayed in the individual images, which allowed us to automatically set the pixel sizes by looking up the scale indicator on the images under program control. With each US image obtained during a sweep, a set of 3-D position and orientation measurements was recorded from the Polaris tracker, which provides information on the relative position of the patient, the US tracker, and any tool in the field of view of the infrared tracking cameras. With this information, it is possible to reconstruct the spatial relationship of images from the various sources (MRI and US), as can be seen in Figure 3. FIGURE 2. A, spatially tracked ultrasound (US) scanhead. B, example of a calibration image in which the wire interOur experimental data consections are automatically detected (circles). sisted of the preoperative MRI scan and a series of US images sweeping the extent of the feature B A of interest, with the associated 3-D tracking data. Given the initial registration, we are able to identify the features we wished to compare. Then, after a manual outline of the same features in both modalities, we performed a correction based on the 3-D point sets obtained from the outlines using the iterative closest point (ICP) algorithm (3). Given two point sets representing a sampled irregular surface, the ICP algorithm will attempt to spatially FIGURE 3. A, two approximately perpendicular series of US images displayed in their 3-D spatial arrangement in merge these two sets until the relation to the outline of the patient’s head in the preoperative MRI scan. B, US image (blue-green) overlaid onto distances between the closest the corresponding coregistered oblique MRI slice. Note the tumor, a large meningioma, which is partially intersected point pairs between the sets is by the US beam, with the hyperechoic lower boundary of the tumor aligned in both MRI and US. minimized or close to it. For shapes that exhibit symmetry or regularity (e.g., spheres or parallelograms), the algorithm may fail to urements which are obtained in a water tank. This is intended to comfind the correct rotation; however, if a shape displays sufficient irregpensate for the small differences in the speed of sound in water (∼1500 ularity, such as a segmented brain, the algorithm works well. This m/s) and in brain parenchyma (∼1540 m/s) (18), which would othercorrection corresponds to a local adjustment of the registration based wise result in a 2.6% error in computing distances from the transducer. In the OR, immediately after the removal of the bone flap but with on the feature of interest itself, rather than the entire head of the the dura intact, we acquire several series of US images. In this study, patient. Image pairs are then produced using the corrected registration the surgeon recorded images of the tumor as well as other recognizain which oblique sections of the MRI stack are produced which match ble features, such as sulci, ventricles, the cranium opposite to the surthe corresponding US images. gical opening, and the falx, whenever they were visible. Images were MRI scans and US images do not emphasize the same physical propobtained in groups or series corresponding to deliberate sweeps of the erties of the tissues being imaged and, as a result, have markedly difregion of interest. Two such US series are displayed in their 3-D spatial ferent appearances. As a way to gauge the extent of this dissimilarity arrangement in Figure 3. between the two modalities, we have tabulated two visual assessments US images were acquired using a frame grabber (Model DT3155; for each patient on the basis of a representative set of MRI scans and US Translation, Marlboro, MA) that was connected to the video output of images. In the case of MRI, the images consisted of those slices encomthe US system. Individual images were grayscale 640 ⫻ 480 8-bit pixels. passing the tumor, whereas for US we selected a series of 20 to 25 images that best intersected the tumor. The assessments were conDepending on the scale setting on the US (60 mm, 90 mm, or 120 mm), ducted during one session by an expert surgeon and a nonexpert anathe pixel size was 6.5099 ⫻ 6.363 mm, 4.332 ⫻ 4.228 mm, or 3.17485 ⫻ lyst who conducted the image processing and analysis. The features in 3.25386 mm, respectively. These sizes correspond to the internal calibraquestion were: the tumor’s homogeneity and the sharpness of its tion of the US system and assume propagation in tissue. The adjustable

A

B




A

C

outlines consisting of points expressed in their respective frames of reference were merged into a common coordinate system using the transformation matrix computed in the OR during registration and the tracking data corresponding to each image. The process is illustrated in Figure 4. On the basis of the outlines obtained in this fashion, we applied the ICP algorithm (3) to sets of points representing the features of interest in MRI and US. The transformation produced by the ICP algorithm can be thought of as a correction which better aligns the two sets of contours in 3-D space. It is not our intention to present a new method of registration per se; we are simply seeking to best align the features we are trying to compare in the two imaging modalities. The effect of the correction we are applying is local to the region of interest, with the understanding that the global registration may still be correct in some sense, but that displacement of the region of interest within the cranial cavity may cause some misregistration between its preoperative MRI location and its position expressed in OR space at the time of the procedure. The 4 ⫻ 4 matrix representing the transformation consists of rotation and a translation components:

B

D

FIGURE 4. Coregistered outlines. A cystic ganglioglioma was outlined in the MRI scan and in the US images (A and B). Note that these two views are not coregistered. On the MRI scan, the tumor appears on an axial slice; in the US image, we view it in an oblique image. C and D, two views of the coregistered MRI (red “x”) and US (blue “.”) outlines are shown from one series of US images. The pixel-based outlines result in a higher point density in US than in MRI in this case because of the different voxel/pixel sizes. It should also be noted that the points obtained from the MRI scan (red) in this case are arranged in a regular series of parallel outlines, whereas the US-based points (blue) correspond to a series of oblique images that swept a portion of the region of interest. US-based outlines are not parallel and can intersect each other. The lower two images show both MRI and US point sets coregistered in the same frame of reference.

boundary. For each tumor case (we dropped the one epilepsy case from this comparison), we recorded the tumor’s appearance as homogeneous or heterogeneous in each modality. We also recorded whether the tumor boundary was well defined or diffuse.

Analysis In all the cases presented here, we outlined the feature of interest in the preoperative MRI scan. In most cases, this feature consisted of the contrast-enhanced tumor to be resected. The segmentation was performed by manually outlining the given features. The same features were manually outlined in selected US series as well. Both sets of

NEUROSURGERY

[eq1]

The translation vector (tx, ty, t z ) describes the displacement imparted to the centroid of the corrected outlines with respect to the uncorrected outlines. In this case, we arbitrarily fixed the MRI outlines and let the US outlines match the MRI point cloud with the ICP algorithm as best as possible. If we consider the US acquired during surgery to represent the “true” position of the outlined feature, whereas the MRI outlines represent the position preoperatively, the length of this vector can be viewed as the amount of translation needed to correct the observed displacement. We have tallied the outline displacement vector lengths in all cases. Note that the rotation that was present in the correction matrices, which was always small, was not used in these calculations. In parallel with the outline-based corrections, we computed an estimate of error inside the cranium, Eic, other than at the fiducial markers.


HARTOV ET AL.

Although the target registration error has been proposed as such a measure of accuracy, it is difficult to compute it if one does not have a good estimate of the fiducial localization errors in both frames of reference (MRI and OR spaces in this case). Unlike the FRE, it is not possible to directly measure the target registration error (TRE) because a target is not readily accessible or precisely measurable in patient surgeries (9). Instead, we devised a simple automatic procedure to estimate Eic based on the point midway between the first two fiducials that were used in the registration. Because not all fiducials are usable in the OR, the set of points involved in the registration possesses a randomness that is dependent on the accessibility of a given fiducial marker in the OR. The point pairs of fiducials selected usually (although not always) corresponded to locations on opposite sides of the head. The transformation matrix representing the coregistration operation was applied to the registration points measured in the OR, thereby expressing them in MRI coordinates. This estimation of the Eic is a departure from the use of the TRE that was proposed by Fitzpatrick et al. (9) for example, but it was devised as a practical method to gauge it. The expression given by Fitzpatrick et al., although rigorous, does not lend itself to practical use because it depends on knowing the fiducial location error for each fiducial marker pair. This information is simply not available in the OR. For a given feature, we compared its appearance in MRI relative to US. For selected US series, we computed the oblique MRI intersections corresponding to each US image in the series. The resulting images and their intersections with the slice-based MRI outlines were displayed side by side for comparison. This capability was developed for use in the OR as a visual guide to the level of agreement between the preoperative MRI and the intraoperative US. Because the intersection of the slice-based MRI outlines no longer constituted an ordered and closed set of points for area computation, we applied a convex hull algorithm to the resulting set of points. The same algorithm was applied to the US-based outline in the corresponding image. In principle, if there is no difference in the MRI and US appearances of the selected feature, and if there was no registration error (i.e., the US and oblique MRI match perfectly), the areas of the two convex hull outlines match exactly. The convex hull algorithm is equivalent, conceptually, to stretching a rubber band around a set of arbitrarily distributed points on a plane. The elastic makes contact with a subset of the points defining an outer boundary which encloses all points and is called the convex hull of that set of points. Figure 5 shows a selected US image overlaid on the corresponding oblique MRI slice, the MRI-based outlines as they are intersected, and the convex hull outline. There are instances in which the effective area of the US image (the fan-shaped arc spanned by the acoustic beam) covers only a portion of the feature of interest. This problem, if it were not accommodated, would make it impossible to compare areas with the MRI slices. As a way to remedy this, we projected masks corresponding to the US image extent onto the reconstructed MRI scans for each US scale (depth setting). Hence, if an MRI outline extended outside of the US image range, it was cropped, and only the overlapping areas were compared. Thus, any area defined on the US image will be compared with the area defined in MRI but within the limits of the US mask. To measure the degree of feature correspondence between the MRI and US partial volumes, we used mutual information (MI) between pairs of images to gauge their similarity (33). Here, we also used the masks corresponding to the effective areas in the US images to restrict the zones being matched for MI. The coregistration procedure, based on fiducial markers applied to the head, produces a transformation matrix. Similarly, the ICP algorithm acting on the two sets of outlines produces another transformation matrix registering the two spaces


A

B

FIGURE 5. A, US image overlaid onto the corresponding reconstructed oblique MRI scan. This is a composite image in which the red channel reflects the intensity of the oblique plane MRI scan, whereas the green channel represents the original US image. The spatial relation between the US and MRI scan are known, which allows us to fetch the MRI voxels that correspond to each of the US pixels. We use this overlaying method to display the images in the operating room (OR), together with the MRI scans and US images. B, in this image we show the same MRI oblique plane image as in A, but without the US overlay. The blue pixels represent the intersections of the manually outlined contours of the tumor and cyst that were obtained from the axial images. This is the same case as in Figure 5, a ganglioglioma surrounded by a cyst. These points form a discontinuous dotted contour of the same feature on this oblique plane. The green trace that surrounds all the blue dots is the convex hull for that set of points. The convex hull is the smallest convex outer boundary that encompasses all these points. We used the convex hull to define the volumes to compare and to perform the mutual information (MI) similarity calculations.

(MRI and US), based this time on the feature of interest. The two transformations, unless they are identical (which did not occur in this study), map a given US image to two slightly different oblique MRI scans. The degree to which a US image was similar to an oblique MRI slice obtained from the coregistration or resulting from the ICP procedure can be compared with the MI computed for each image pair. MI is a measure of similarity that is computed by estimating the joint probability distribution function for a pair of images and the resulting marginal probability distribution functions. This is well approximated by using the Kullback-Leibler algorithm (14). The similarity obtained through MI is a representation of the consistency with which a pixel value in one modality is mapped onto another pixel value in a different imaging modality. Thus, two copies of the same image in which the scenes are identical but their grayscale representations are different (e.g., by using a different lookup table) have greater similarity than the same images shifted slightly with respect to each other (for example, see illustrations in Section 3 of Reference 10). In this test, we compared the MI between a series of US images and their corresponding oblique MRI slices as computed from the initial coregistration procedure in the OR (MIcor) and the MI between the same US slice and its corresponding MRI equivalent computed from the outline-based ICP algorithm (MIICP).

RESULTS The accuracy of our registration procedure was evaluated using two measures, the fiducial registration error (FRE) and Eic. The number of fiducial markers varied in each case, with 7 to 15 markers placed on patients for the preoperative MRI scan. In the OR, not all points were accessible for the registration procedure, which results in as few as four markers and as many as



eight being used in the registration computation. For the cases presented here, we used, on average, 4.92 point pairs to compute the registration. The FRE in our cases ranged from 1.69 to 6.74 mm, with a mean of 3.90 mm and a standard deviation (SD) of 1.13 mm. The quantity Eic for all the cases, based on the midpoint between the first two markers, ranged from 0.24 to 2.97 mm, with a mean of 1.50 mm and an SD of 0.58 mm. Using the features outlined in the preoperative MRI scan and the same features outlined in the coregistered US, we found that the centroids of the outlines need to be moved an average of 4.01 mm (minimum, 0.11 mm; maximum, 8.73 mm; SD, 2.50 mm) to line up correctly. We did not find a consistent direction for this correction, in relation to the patient’s position (e.g., the direction of gravity). Using the same MRI- and US-based outlines of the feature of interest, we compared the areas enclosed by the convex hulls in the two modalities based on the initial registration. Here, we found a very clear trend in which the MRI outlines consistently enclosed a larger area than the corresponding US outlines. This trend was observed in 30 (86%) of the 35 patients, with the MRI area 18.9% larger than the US on average (SD, 16.2%). In the 5 (14%) of the 35 cases in which the US areas were larger, the increase averaged 38.8%, with an SD of 33.0%. If we consider the MRI outlines to be the reference, the use of US tends to underestimate the size of the feature of interest in close to 86% of the cases, sometimes by a significant amount. When segregating by tumor types, we observed that tumor sizes in US images were smaller than in MRI scans in most cases of meningiomas (nine out of 13) and in all cases of metastases (nine out of nine). We do not know what accounts for this difference, but a similar finding was reported by Renner et al. (29). The comparison of the areas enclosed by outlines obtained manually depends, to some degree, on a subjective interpretation of the images. In an attempt to produce a more objective, quantitative measure of the degree of feature correspondence, we used the MI between US image series and their corresponding voxels in the MRI scans before and after the ICP-based alignment. We tabulated the MIcor and MIICP values for all cases, on the basis of 20 to 25 US images. Using this table, we tested the null hypothesis that both sets of data had the same mean, using a paired t test for means, assuming the same variances. The result was that we had to reject the null hypothesis (P(T ⬍ ⫽ t) ⫽ 0.0123, α ⫽ 0.05). The data indicate a statistically significant increase in the MI from the initial coregistration (mean MIcor ⫽ 0.458) to the corrected registration (mean MIICP ⫽ 0.500). To gauge the dissimilarity between US and MRI, we simply calculated how often our visual assessments for homogeneity and boundary sharpness agreed between the two modalities. Tumor homogeneity assessments (homogeneous/ heterogenous) agreed between the two modalities only 40% of the time, with the same result for the boundary sharpness assessment (well-defined/diffuse). Furthermore, the agreement in both of these features did not occur in the same cases. Both assessments of features agreed in only 54% of the cases.

NEUROSURGERY

DISCUSSION Rigid registration is imperfect. Initial disagreement may be the entire cause of registration errors, or there may be actual displacement. In previous experiments (5) with a deformable phantom, we were able to repeat the registration procedure with five different computed tomographic scans of the same object. Under these well-controlled laboratory experiments, we obtained FREs between 0.68 and 0.91 mm (mean, 0.83 mm; n ⫽ 5), with a maximum (individual) error of 2.0 mm and a minimum error of 0.92 mm. In their study, Maurer et al. (24) showed that they could achieve accuracies on the order of 0.7 mm. Our mean FRE of 3.90 mm may be the result of not using bone-implanted markers as described previously (24); we use markers attached to the skin. The ability of the skin to stretch will displace the markers and affect registration. Our Eic of 1.5 mm reflects the fact that the alignment error at the midpoint between two selected registration points will tend to be smaller than that recorded at these points because of an inverse lever arm effect (i.e., when moving the end points of a line segment, its midpoint is much less affected than points near the ends). The outline and ICP-based correction resulted in an average displacement of 4 mm, which is comparable with the FRE and significantly greater than the estimated Eic. This magnitude of the initial error cannot be attributed with certainty to the registration (FRE 艐 4 mm) or to actual parenchymal displacement. Nevertheless, it is a measurable disagreement between the preoperative MRI scan and initial intraoperative US images that can be corrected. The correction may be viewed either as a last-step improvement on the overall registration procedure or as an adjustment required to compensate for intervening brain deformation. The disparity between Eic and the computed correction required is likely a function of how we chose to compute the estimate. The error estimate Eic is moderated by the target location midpoint between two fiducial markers, whereas the outline centroids, for which a correction is computed, are not necessarily near that location. We based our study on intraoperative images acquired immediately after the craniotomy. At that stage, no significant deformation other than a possible protruding of the parenchyma through the craniotomy is expected in a typical case. Images acquired at that moment are likely to convey the state of least deformation during that procedure. The correction provided by the ICP algorithm on the outlined region of interest nevertheless showed that some misalignment can exist even at that stage. Later in a procedure, with only a partial tumor remaining, isolating its shape and basing a correction on that information is not likely to improve coregistration. Features outlined in US appeared generally smaller than the corresponding outlines in MRI scans. It is conceivable that this discrepancy is a result of our methodology. We used convex hulls of the outlines rather than the outlines, themselves. We also used masks defining the fan-shaped boundaries of US images to ensure similar coverage in both modalities. Our use


HARTOV ET AL.

of the convex hulls was dictated by the fact that the sets of points constituting the MRI-based outlines came out of order when intersected by an arbitrary plane defined by the corresponding US images. We expect that the convex hulls of the exact same outlines in the two modalities would result in the same computed areas. We, therefore, conclude that the appearance of tumors in US generally results in a boundary that is smaller than that in MRI scans. This should be taken into account by surgeons using US in the OR. It should also be noted that although we used coregistration for the purpose of comparing the spatial locations of the features in both modalities, their outlines are independent of coregistration. This finding confirms the results presented by Renner et al. (29), although their study was not based on coregistered US scans. One could conclude that if US- and MRI-based outlines do not agree on the tumor size, this may account for the disagreement between the two after craniotomy. We used the MI of the enclosed volumes between the two modalities to gauge the effect of using them to reregister and to assess the overall level of feature correspondence between MRI and US. The statistically significant improvement in the MI after correction is an indication that the reregistration step constitutes an improvement. To understand why that would be, one needs to visualize the effect of the ICP algorithm on the data it is given. If one were to start with two identical point sets and then scale one very slightly so that both have exactly the same shape but with one slightly smaller, the ICP would result in the smaller set of points being located inside the larger one, with their centroids colocated. Thus, slight differences in size would not matter in aligning the two sets of points. One could, of course, forego the outlining step and use MI-based optimization on designated voxel sets to attempt the alignment. In our experience, however, this is less reliable than the ICP; the MI-based algorithm does not converge in all cases. In our comparison between the appearances of tumors in MRI scans and US images, we found very little agreement. One would expect that the same tumors imaged on two MRI systems, for example, would result in the same visual assessments close to 100% of the time, both comparing the tumors’ homogeneity and the sharpness of their boundaries. We observed agreement in only 40% of the cases for each comparison. Furthermore, agreement on the homogeneity of two sets of images does not seem to indicate agreement on the sharpness of the tumor boundaries, the two classes of assessments being in agreement only 54% of the time. This disparity reflects the different underlying physical processes on which each modality is based. It is clear that the two imaging modalities reveal very different properties about the tissues that are imaged. US is sensitive to changes in acoustic impedance along the path traveled by the transmitted waves. Compared with MRI, this is not only a different physical property (acoustic impedance) but also a fundamentally different mechanism for “looking” at it. US “sees” better sharp acoustic contrasts; moreover, the orientation of such a boundary will also affect its appearance, with discontinuities perpendicular to the path of propagation best visible. By


contrast, MRI displays intrinsic properties of the tissue, relating to the relative abundance of H nuclei. Quantitative and qualitative findings in any one modality do not extend necessarily to the other. We see the 3-D merging of preoperative MRI scans with intraoperative US as a way to mitigate the uncertainty that is attached to findings in any one modality. Finally, our findings on the respective appearances of tumors in US and MRI are based on a heterogeneous set of cases, with small numbers of each type of tumors, therefore these results should be considered preliminary.

CONCLUSION Imaging data from 35 neurosurgery cases was processed retrospectively to assess initial registration accuracy, its improvement from realignment of feature boundaries, and the degree of feature congruence between preoperative MRI and intraoperative US once optimal alignment has been established. The results show a statistically significant reduction in registration error based on an increase in MI in overlapping feature volumes. Additionally, tumor volumes segmented in MRI were consistently larger than their corresponding regions segmented in US even after the improved registration alignment was achieved, which is consistent with previous findings. Although the initial FRE in this study was relatively large, likely because of the use of scalp placed fiducials, which are susceptible to small scale motion, the US image correction and overall US feature correspondence with MRI effectively reduced the FRE after the start of surgery to a level commensurate with the estimated intracranial error and consistent with high fidelity neuronavigation.

REFERENCES 1. Arun KS, Huang TS, Blostein SD: Least-squares fitting of two 3-D point sets. IEEE Trans Pattern Anal Mach Intell 9:698–700, 1987. 2. Auer LM, van Velthoven V: Intraoperative ultrasound imaging. Comparison of pathomorphological findings in US and CT. Acta Neurochirurgica 104:84–95, 1990. 3. Besl PJ, McKay ND: A method for registration of 3-D Shapes. IEEE Trans Pattern Anal Mach Intel 14:239–256, 1992. 4. Blackall JM, Penney GP, King AP, Hawkes DJ: Alignment of sparse freehand 3-D ultrasound with preoperative images of the liver using models of respiratory motion and deformation. IEEE Trans Med Imaging 24:1405–1416, 2005. 5. Blumenthal T: Quantification of sub-surface brain deformation from spatially-tracked freehand intraoperative ultrasound. Dartmouth College, Thayer School of Engineering, Hanover, New Hampshire, 2004 (dissertation). 6. Bucholz RD, Yeh DD, Trobaugh J, McDurmont LL, Sturm CD, Baumann C, Henderson JM, Levy A, Kessman P: The correction of stereotactic inaccuracy caused by brain shift using an intraoperative ultrasound device, in Troccas J, Grimson E, Mösges AJ (eds): CVRMed-MRCAS’97: First Joint Conference Compter Vision, Virtual Reality and Robotics in Medicine and Medical Robotics and Computer-Assisted Surgery. London, Springer Verlag, Grenoble, 1997, pp 459–466. 7. Clatz O, Delingette H, Talos IF, Golby AJ, Kikinis R, Jolesz FA, Ayache N, Warfield SK: Robust nonrigid registration to capture brain shift from intraoperative MRI. IEEE Trans Med Imaging 24:1417–1427, 2005. 8. Comeau RM, Sadikot AF, Fenster A, Peters TM: Intraoperative ultrasound for guidance and tissue shift correction in image-guided neurosurgery. Med Phys 27:787–800, 2000.



9. Fitzpatrick JM, West JB, Maurer CR: Predicting error in rigid-body, pointbased registration. IEEE Trans Med Imaging 17:694–702, 1998. 10. Hajnal JV, Hill DLG, Hawkes DJ (eds): Medical Image Registration. Boca Raton, CRC Press, 2001. 11. Hammoud MA, Ligon BL, elSouki R, Shi WM, Schomer DF, Sawaya R: Use of intraoperative ultrasound for localizing tumors and determining the extent of resection: A comparative study with magnetic resonance imaging. J Neurosurg 84:737–741, 1996. 12. Hartkens T, Hill DL, Castellano-Smith AD, Hawkes DJ, Maurer CR, Martin AJ, Hall WA, Liu H, Truwit CL: Measurement and analysis of brain deformation during neurosurgery. IEEE Trans Med Imaging 22:82–92, 2003. 13. Hartov A, Eisner SD, Roberts DW, Paulsen KD, Platenik LA, Miga MI: Error analysis for a free-hand 3D ultrasound system for neuronavigation. Neurosurg Focus 6:5, 1999. 14. Haussler D, Opper M: General bounds on the mutual information between a parameter and conditionally independent observations, in Proceedings of the Eighth Annual Conference on Computational Learning. Santa Cruz, 1995, pp 402–411. 15. Hill DL, Maurer CR, Maciunas RJ, Barwise JA, Fitzpatrick JM, Wang MY: Measurement of intraoperative brain surface deformation under a craniotomy. Neurosurgery 43:514–528, 1998. 16. Jödicke A, Deinsberger W, Erbe H, Kriete A, Böker DK: Intraoperative threedimensional ultrasonography: An approach to register brain shift using multidimensional image processing. Minim Invasive Neurosurg 41:13–19, 1998. 17. Kaspersen JH, Sjølie E, Wesche J, Asland J, Odegård A, Lindseth F, Nagelhus Hernes TA: Three-dimensional ultrasound-based navigation combined with preoperative CT during abdominal interventions: A feasibility study. Cardiovasc Intervent Radiol 26:347–356, 2003. 18. Kremkau FW: Diagnostic Ultrasound: Principles and Instruments. Philadelphia, Pa, W.B. Saunders, ed 4, 1993. 19. Lange T, Eulenstein S, Hünerbein M, Lamecker H, Schlag P: Augmenting intraoperative 3D ultrasound with preoperative models for navigation in liver surgery, in Barillot C, Haynor DR, Hellier P (eds): Medical Image Computing and Computer-Assisted Intervention. Vol 3217, Lecture Notes in Computer Science, New York, Springer, 2004, pp 534–541. 20. LeRoux PD, Berger MS, Wang K, Mack LA, Ojemann GA: Low grade gliomas: Comparison of intraoperative ultrasound characteristics with preoperative imaging studies. J Neurooncol 13:189–198, 1992. 21. LeRoux PD, Winter TC, Berger MS, Mack LA, Wang K, Elliott JP: A comparison between preoperative magnetic resonance and intraoperative ultrasound tumor volumes and margins. J Clin Ultrasound 22:29–36, 1994. 22. Leroy A, Mozer P, Payan Y, Troccaz J: Rigid registration of freehand 3D ultrasound and CT-scan kidney images. Medical Image Computing and ComputerAssisted Intervention Lecture Notes in Computer Science, MICCAI. New York, Springer Verlag, 2004, pp 837–844. 23. Letteboer MM, Willems PW, Viergever MA, Niessen WJ: Brain shift estimation in image-guided neurosurgery using 3-D ultrasound. IEEE Trans Biomed Eng 52:268–276, 2005 24. Maurer CR, Fitzpatrick JM, Wang MY, Galloway RL, Maciunas RJ, Allen GS: Registration of head volume images using implantable fiducial markers. IEEE Trans Med Imaging 16:447–462, 1997. 25. Mitchell M: An Introduction to Genetic Algorithms. Cambridge, MIT Press, 1998. 26. Nabavi A, Black PM, Gering DT, Westin CF, Mehta V, Pergolizzi RS, Ferrant M, Warfield SK, Hata N, Schwartz RB, Wells WM, Kikinis R, Jolesz FA: Serial intraoperative magnetic resonance imaging of brain shift. Neurosurgery 48:787–798, 2001. 27. Nimsky C, Ganslandt O, Cerny S, Hastreiter P, Greiner G, Fahlbusch R: Quantification of, visualization of, and compensation for brain shift using intraoperative magnetic resonance imaging. Neurosurgery 47:1070–1080, 2000. 28. Prager RW, Rohling RN, Gee AH, Berman L: Rapid calibration for 3-D freehand ultrasound. Ultrasound Med Biol 24:855–869, 1998. 29. Renner C, Lindner D, Schneider JP, Meixensberger J: Evaluation of intraoperative ultrasound imaging in brain tumor resection: A prospective study. Neurol Res 27:351–357, 2005.

NEUROSURGERY

30. Roberts DW, Hartov A, Kennedy FE, Miga FI, Paulsen KD: Intraoperative brain shift and deformation: A quantitative analysis of cortical displacement in 28 cases. Neurosurgery 43:749–760, 1998. 31. Tronnier VM, Bonsanto MM, Staubert A, Knauth M, Kunze S, Wirtz CR: Comparison of intraoperative MR imaging and 3D-navigated ultrasonography in the detection and resection control of lesions. Neurosurgical Focus 10:E3, 2001. 32. van Velthoven V: Intraoperative ultrasound imaging: Comparison of pathomorphological findings in US Versus CT, MRI and intraoperative findings. Acta Neurochir Suppl 85:95–99, 2003. 33. Wells WM, Viola P, Atsumi H, Nakajima S, Kikinis R: Multi-modal volume registration by maximization of mutual information. Med Image Anal 1:35–51, 1996. 34. Wirtz CR, Bonsanto MM, Knauth M, Tronnier VM, Albert FK, Staubert A, Kunze S: Intraoperative magnetic resonance imaging to update interactive navigation in neurosurgery: Method and preliminary experience. Comput Aided Surg 2:172–179, 1997.

Acknowledgments We gratefully acknowledge support for this work from National Institutes of Health/National Institute of Biomedical Imaging and BioEngineering Grant R01-EB002082.

COMMENTS

A

lthough somewhat technical for the typical reader of NEUROSURGERY, this report by Hartov et al. nicely shows the utility and limitations of coregistering intraoperative ultrasound with magnetic resonance imaging (MRI) scans in craniotomy. Major limitations of most navigation systems in assisting tumor resection are the brain shift from cerebrospinal fluid leakage and local tissue deformations from lesion resection. Intraoperative imaging is used to update navigational data with increasing frequency, often using expensive intraoperative MRI solutions. Ultrasound has been advocated as a substantially less costly means to achieve data updates; however, the images frequently have poor signal-to-noise quality, making their interpretation difficult in some cases. As has been our experience, the authors show that coregistration, even with preoperative MRI scans, can assist with interpretation of ultrasound data although there may be some limitations such as spatial and size fidelity. This application of intraoperative ultrasound is likely to be seen more frequently with commercial systems as surgical navigation continues to evolve. Gene H. Barnett Cleveland, Ohio

H

artov et al. analyzed coregistered ultrasound and MRI scans. Segmented features were used to reduce the degree of misalignment, which amounted up to more than 8 mm in initial maximum error. This article clearly demonstrates that a standard straightforward rigid registration of ultrasound and MRI scans is potentially prone to errors. The authors also emphasize that ultrasound and MRI scans do not show exactly the same results, e.g., when considering the extent of a tumor, which complicates comparative analysis further. Whether it is the lower resolution of the ultrasound images (highest resolution was about 3 ⫻ 3 mm) compared with the MRI scans or a spatial distortion error of the different modalities itself that further complicates image registration is an unsolved question. Nevertheless, the approach presented by the authors seems to be a suitable way to align MRI scans and ultrasound images so that ultrasound can be used as an intraoperative adjunct to preoperative MRI scans. Correct alignment greatly facilitates the potentially difficult inter-


HARTOV ET AL.

pretation of the ultrasound images. However, it has to be emphasized that their method was applied for the preresection situation only, i.e., intraoperative ultrasound was applied just after craniotomy. It will be much more interesting and, of course, more challenging to see how their method will work when parts of a tumor are resected and brain shift phenomena further complicate image registration. Application of intraoperative MRI, on the other hand, has proved to provide images of high quality, compensating for all effects of brain shift. Side-by-side display of pre- and intraoperative images measured at identical scan positions greatly facilitates the evaluation of the completeness of a resection without any guesswork and no registration errors. Christopher Nimsky Erlangen, Germany

H

artov et al. introduce new methods of analysis and preliminary results regarding the use of coregistered intraoperative ultrasound imaging and preoperative MRI. This is the kind of research that is needed as the two most important intraoperative imaging methods, namely ultrasound and MRI, are now being used separately or jointly, including not only preoperative MRI, as in this article, but also intraoperative MRI. The study is also important because it demonstrates the increasing need for academic centers to form multidisciplinary research groups to better understand these often expensive and complex methods that are being used to guide surgical eye-hand interaction. Increasingly, neurosurgeons in their “operating rooms of the future” need to understand images that are often taken in planes oblique to the orthogonal axial, sagittal, and coronal planes, and for which there are no textbook matches. By requesting imaging data (instead of only images or radiological reports), neurosurgeons are making independent judgments as to surgical approach and extent of resection. This is a paradigm shift that started when neurosurgeons first began to independently use real-time intraoperative ultrasound imaging in the early 1980s. It has continued with our use of neuronavigators and most recently, with intraoperative MRI. This report is an effort to more rigorously delineate the extent of tumor to be removed on the basis of two very different physical phenomena. Neither the ultrasound image nor any of the various MRI sequence scans should be relied on as actuality to be robotically applied for the surgical removal of imaged tumors. Although images of meningiomas and even many metastases seem straightforward, images of gliomas are fraught with pitfalls with respect to tumor delineation for purposes of removal. We need more interdisciplinary work such as this to scientifically apply a new concept, namely the “region of surgical interest” (1), describing the approach, the intervention, and the outcome of neurosurgical procedures. John Koivukangas Oulu, Finland 1. Koivukangas J, Katisko J, Yrjänä S, Tuominen J, Schiffbauer H, Ilkko E: Successful neurosurgical 0.23T intraoperative MRI in a shared facility, in Gonçalves V (ed): 12th European Congress of Neurosurgery (EANS), Lisbon. Bologna, Monduzzi Editore, 2003, pp 439–444.

I

n this article, Hartov et al. discuss their experience with coregistration of MRI and ultrasound as a means of intraoperative neuronavigation during the resection of tumors. Tissue deformation occurs throughout a neurosurgical intervention and alters the accuracy of the neuronavigation systems that rely solely on preoperative patient imaging to locate the surgical target, such as a tumor or a functional area. The aim


of this study was to evaluate MRI and ultrasonography alignment accuracy individually and after coregistration. The hypothesis of the study was founded on the well-recognized phenomenon of brain shift during neurosurgical procedures. The use of ultrasound to correlate intraoperative data points with preoperative imaging in an effort to augment neuronavigation is an old technique and has been described previously by several groups (1–3, 7). The authors here present data from a large retrospective case series of patients in which they used ultrasound as an adjunct to preoperative neuronavigation systems at their institution. It is really no surprise that coregistered images provided better alignment (6, 9, 10). One critical finding recognized by this group was that the size of the tumor varied between MRI and ultrasound. Tumors on MRI scans (contrast-enhanced T1-weighted spoiled-gradient) were an average of 18.9% larger. Others have also found that the demarcation of gliomas is not as well defined by ultrasound compared with preoperative computed tomographic or MRI scans (14). This discrepancy is probably a result of differences in how each of the modalities acquires data and develops images. Moreover, if the ultrasound probe is directed more tangentially to the surface of the target and as the distance from the probe to the object increases, the fusion becomes increasingly more inaccurate (12). Thus, although the use of mathematical algorithms allowed for coupling of data sets using mutual information, one fundamental irreconcilable issue that remained was how to compensate for the size and boundaries of the tumor between the two imaging modalities. Which do you trust for resection boundaries? After partial resection, how well can the algorithms work with partial data? It is a starting point and adjustments of imaging techniques and of further refinement of algorithms could no doubt improve the accuracy. It remains to be seen whether ultrasound will play any role in intraoperative neuronavigation for stereotactic surgery of subcortical structures. For this to be feasible, the ultrasound probe would need to be small enough to be maneuvered sufficiently within the confines of the burr hole. The resolution of the ventricular system would need to be on par with that obtained with neonatal ultrasound through the anterior fontanelle. Then a similar algorithm could be used to demarcate the ventricular system on both MRI and ultrasound and thus ultrasound could be used as a means of quality control in the operating room. Surface matching of the ventricular system has previously been performed using ultrasound (1). Alternatively, the duplex and Doppler mode of the ultrasound system can be used to display the intraoperative vascular anatomy that could be integrated into the neuronavigational data set to provide intraoperative image updates (13). As the field of deep brain stimulation continues to expand and intraoperative physiological monitoring of new targets is less well-understood, new means of quality control will be needed and will likely be in the form of intraoperative neuronavigation. There has, of course, been some use of intraoperative MRI scans of these patients with good results (5, 8). However, the capital expenses and logistical difficulty of performing deep brain stimulation under these settings will likely preclude its widespread use. Alternatives such as intraoperative computed tomographic scans, positron emission tomography, functional MRI, and three-dimensional fluoroscopy are avenues that are being considered and pursued (4). Thus, despite the facts that ultrasonography image quality is not always good and orientation remains convoluted in many patients, with the assistance of image coregistration, the orientation of the neuronavigation system is facilitated. Multimodal visualization makes it easy to interpret information from several different image volumes and modalities simultaneously (11). To most efficaciously use new alternatives in neuronavigation, a multidisciplinary approach to the problem is needed. The authors’ development of collaborative efforts



between engineering and neurosurgery should be applauded. Such multidisciplinary approaches are essential for neurosurgical innovations of the future. Julie G. Pilitsis Roy A.E. Bakay Chicago, Illinois 1. Hata N, Suzuki M, Dohi T, Iseki H, Takakura T, Hashimoto D: Registration of ultrasound echography for intraoperative use: A newly developed multiproperty method, in Robb RA (ed): Proceedings of SPIE: Visualization in Biomedical Computing. Bellingham, SPIE Press, 1994, vol 2359, pp 251–259. 2. Hata, N, Dohi T, Iseki H, Takakura K: Development of a frameless and armless stereotactic neuronavigation system with ultrasonographic registration. Neurosurgery 41:608–614, 1997. 3. Hirschberg H, Unsgaard G: Incorporation of ultrasonic imaging in an optically coupled frameless stereotactic system. Acta Neurochir Suppl 68:75–80, 1997. 4. Hunsche S, Sauner D, Maarouf M, Lackner K, Sturm V, Treuer H: Combined x-ray and magnetic resonance imaging facility: Application to image-guided stereotactic and functional neurosurgery. Neurosurgery 60 [Suppl 2]:352–361, 2007. 5. Jolesz FA: Future perspectives for intraoperative MRI. Neurosurg Clin N Am 16:201–213, 2005. 6. Keles GE, Lamborn KR, Berger MS: Coregistration accuracy and detection of brain shift using intraoperative sononavigation during resection of hemispheric tumors. Neurosurgery 53:556–564, 2003. 7. Koivukangas J, Louhisalmi Y, Alakuijala J, Oikarinen J: Ultrasound-controlled neuronavigator-guided brain surgery. J Neurosurg 79:36–42, 1993.

8. Martin AJ, Larson PS, Ostrem JL, Keith Sootsman W, Talke P, Weber OM, Levesque N, Myers J, Starr PA: Placement of deep brain stimulator electrodes using real-time high-field interventional magnetic resonance imaging. Magn Reson Med 54:1107–1114, 2005. 9. Miller D, Heinze S, Tirakotai W, Bozinov O, Surucu O, Benes L, Bertalanffy H, Sure U: Is the image guidance of ultrasonography beneficial for neurosurgical routine? Surg Neurol 67:579–588, 2007. 10. Nagelhus Hernes TA, Lindseth F, Selbekk T, Wollf A, Solberg OV, Harg E, Rygh OM, Tangen GA, Rasmussen I, Augdal S, Couweleers F, Unsgaard G: Computer-assisted 3D ultrasound-guided neurosurgery: Technological contributions, including multimodal registration and advanced display, demonstrating future perspectives. Int J Med Robot 2:45–59, 2006. 11. Rasmussen IA, Lindseth F, Rygh OM, Berntsen EM, Selbekk T, Xu J, Nagelhus Hernes TA, Harg E, Haberg A, Unsgaard G: Functional neuronavigation combined with intra-operative 3D ultrasound: Initial experiences during surgical resections close to eloquent brain areas and future directions in automatic brain shift compensation of preoperative data. Acta Neurochir (Wien) 149:365–378, 2007. 12. Schlaier JR, Warnat J, Dorenbeck U, Proescholdt M, Schebesch KM, Brawanski A: Image fusion of MR images and real-time ultrasonography: Evaluation of fusion accuracy combining two commercial instruments, a neuronavigation system and a ultrasound system. Acta Neurochir (Wien) 146:271–277, 2004. 13. Sure U, Benes L, Bozinov O, Woydt M, Tirakotai W, Bertalanffy H: Intraoperative landmarking of vascular anatomy by integration of duplex and Doppler ultrasonography in image-guided surgery: Technical note. Surg Neurol 63:133–142, 2005. 14. van Velthoven V: Intraoperative ultrasound imaging: Comparison of pathomorphological findings in US versus CT, MRI and intraoperative findings. Acta Neurochir Suppl 85:95–99, 2003.

Andreas Vesalius in Padua, (1859), Edouard Hamman. Marseille, Musée de Beaux-Artes. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


TUMOR Technical Case Report

USEFULNESS OF INTRAOPERATIVE PHOTODYNAMIC DIAGNOSIS USING 5-AMINOLEVULINIC ACID FOR MENINGIOMAS WITH CRANIAL INVASION: TECHNICAL CASE REPORT Yoichi Morofuji, M.D. Department of Neurosurgery, Nagasaki University School of Medicine, Nagasaki, Japan

Takayuki Matsuo, M.D. Department of Neurosurgery, Nagasaki University School of Medicine, Nagasaki, Japan

Yukishige Hayashi, M.D. Department of Neurosurgery, Nagasaki University School of Medicine, Nagasaki, Japan

Kazuhiko Suyama, M.D. Department of Neurosurgery, Nagasaki University School of Medicine, Nagasaki, Japan

Izumi Nagata, M.D. Department of Neurosurgery, Nagasaki University School of Medicine, Nagasaki, Japan Reprint requests: Yoichi Morofuji, M.D., Department of Neurosurgery, Nagasaki University School of Medicine, 1-7-1 Sakamoto, Nagasaki, 852-8501, Japan. Email: [email protected] Received, March 2, 2007. Accepted, June 18, 2007.

OBJECTIVE: We present a case of a meningioma in which photodynamic diagnosis (PDD) using 5-aminolevulinic acid was very useful in identifying the cranial involvement. CLINICAL PRESENTATION: An 83-year-old woman presented with a bony, hard, immobile bulge in her left forehead. Computed tomographic scans showed a thickening in the left frontal bone with a flat mass underneath. Magnetic resonance imaging scans revealed that enhancing lesions spread to the dura mater and subcutaneous tissue around the thickened frontal bone, reaching the upper margin of the left orbit. INTERVENTION: Intraoperative PDD using 5-aminolevulinic acid indicated the optimal extent of the excision by showing clear fluorescence of affected tissues. The tumor was totally resected and diagnosed as an atypical meningioma. Histopathological examination confirmed the consistency of the extent of tumor invasion with affected lesions on PDD. CONCLUSION: To the best of our knowledge, this is the first case demonstrating the efficacy of PDD using 5-aminolevulinic acid for a meningioma with cranial invasion. Additional studies are warranted, as shown in cases of malignant gliomas. KEY WORDS: 5-Aminolevulinic acid, Cranial invasion, Intraoperative photodynamic diagnosis, Meningioma, Skull invasion Neurosurgery 62[ONS Suppl 1]:ONSE102–ONSE104, 2008

P

hotodynamic diagnosis (PDD) using 5-aminolevulinic acid (5-ALA) has been performed much more frequently in recent years and has become an important intraoperative technique. Because of its convenience and usefulness in the field of neurosurgery, PDD is being performed on patients with malignant gliomas at many institutions (2, 4, 7, 8, 9). However, to the best of our knowledge, there have been no studies on PDD using 5-ALA in patients with cranial lesions. Here, we present a case of a meningioma in which PDD using 5-ALA was very useful in identifying cranial involvement.

Clinical Presentation An 83-year-old woman complained of headache and a bulge on the left side of her forehead. She had no notable medical history, including head trauma, and previous cranial magnetic resonance imaging (MRI) scans did not show any abnormalities. She was referred

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297010.50716.05

to our department in June 2006 because the subcutaneous mass (diameter, ∼3 cm) had become painful. It was bony, hard, immobile, and not tender. Cranial computed tomographic (CT) scans showed well demarcated thickening in the left frontal bone and a flat mass exhibiting isodensity immediately underneath. Destruction of both the inner and outer tables of the cranium was confirmed. Contrast-enhanced MRI scans showed even enhancement spreading to the dura mater and subcutaneous tissue around the thickened bone, reaching the upper margin of the left orbit. Diffuse meningeal enhancement was also seen adjacent to the tumor (Fig. 1), suggesting a rapidly progressing meningioma. In order to determine the extent of tumor excision intraoperatively, PDD using 5-ALA was planned and written informed consent was obtained. The protocol of PDD using 5-ALA was approved by the Ethics Committee of Nagasaki University.

VOLUME 62 | OPERATIVE NEUROSURGERY 1 | MARCH 2008 | ONSE102

MOROFUJI ET AL.

A

B

A

FIGURE 1. Preoperative T1-weighted axial (A) and sagittal (B) magnetic resonance imaging (MRI) scans with gadolinium showing a thickened frontal bone. Enhancement was spreading to the dura mater and subcutaneous soft tissue around the thickened bone. Broad meningeal enhancement was also seen.

A

B

FIGURE 2. The tumor fluorescence was evaluated intraoperatively. A, the removed tumor under white light. B, intraoperative photodynamic diagnosis (PDD) confirmed that the tumor itself was highly red fluorescent.

B

FIGURE 4. Postoperative T1-weighted axial (A) and sagittal (B) MRI scans with gadolinium showing no residual tumor.

Intervention Three hours before the induction of anesthesia, 20 mg/kg of 5ALA was administered orally. The patient was kept in a dark room for 48 hours after drug administration to avoid possible skin phototoxicity. After the scalp incision was made, adhesion between the cranial lesion and galea was seen in the area of bone thickening. The cranial lesion, along with the dura mater, was excised as a single mass. Intraoperative PDD showed that the tumor itself was highly fluorescent, but that the dura mater surrounding the tumor was not (Fig. 2). PDD also clearly showed fluorescence from the dipole to the inner table at the stump of the upper orbital margin (Fig. 3), whereas no tumor invasion was observed microscopically. Drilling was performed until the fluorescence disappeared while preserving the outer table. Surgery was completed after confirming the absence of residual fluorescence in the surgical field. The patient was discharged in good health 9 days after surgery. MRI scans performed 9 months after surgery showed no evidence of tumor recurrence (Fig. 4).

Histology

A

B

FIGURE 3. A, the surgical field after removal of the bone flap. B, PDD showing red fluorescence from the dipole to the inner table at the stump of the upper orbital margin.

ONSE102 | VOLUME 62 | OPERATIVE NEUROSURGERY 1 | MARCH 2008

The tumor on the medial surface of the dura mater grew in a turbinated or fascicular manner and obviously involved the cranium. Tumor cells densely existed from the dipole to the outer table of the cranium, and cellular atypia and nuclear fission were seen. The tumor was diagnosed as an atypical meningioma (World Health Organization Grade II). No tumor cells were seen in the surrounding dura mater, including the area where contrast enhancement was observed on preoperative MRI scans. Tumor cells did exist in the fluorescent orbit-


PHOTODYNAMIC DIAGNOSIS FOR CRANIAL INVASION

A

B

C

D

E FIGURE 5. Photomicrographs of the tumor. Microscopic features showed dense tumor cells, cellular atypia, and nuclear fission (A and B). No tumor cells were seen in the nonfluorescent dura mater stump (C) and bone stump (D). Tumor cells were seen in the fluorescent orbit-side stump (E). Hematoxylin and eosin staining; original magnifications, ⫻25 (A, C, D, and E) and ⫻200 (B).

side stump, but none were observed in the non-fluorescent bone stump (Fig. 5).

DISCUSSION First attempted by Moore (6) in 1948, fluorescent dyes have long been used to identify brain tumors. Since then, fluorescent materials and light sources have been developed and improved. At present, PDD using 5-ALA is accepted for its convenience and usefulness, particularly in excising malignant gliomas (2, 7, 8). In addition, randomized, controlled studies on PDD have documented favorable results in glioma excision and progression-free survival (9). To the best of our knowledge, however, no studies on PDD for meningiomas with cranial invasion have been reported. In cranial or brain tumors with cranial involvement, the extent of bone resection is determined mostly on the basis of preoperative imaging and intraoperative findings. However, in clinical settings, it is often difficult to determine the boundary between normal bone and tumor tissues. Therefore, we used PDD in order to identify the extent of cranial invasion in the present case. Because intraoperative PDD showed lesional fluorescence in the operative field, additional bone resection

NEUROSURGERY

was needed, including the superior wall of the left orbit. By confirming the negative fluorescence within the surgical field, the tumor was completely excised. Histopathological examination confirmed that PDD accurately assessed the extent of cranial involvement of the meningioma. Although rapid intraoperative pathological diagnosis of bone lesions without demineralization was not technically feasible, the degree of bone invasion was easily assessed on PDD. Linear thickening and contrast enhancement of the meninges adjacent to a meningioma have been called the “dural tail signs,” “dural thickening,” “flare,” or “meningeal sign.” However, the histological basis of the dural tail with a meningioma is not completely understood. In the present case, preoperative contrast-enhanced MRI scans showed an enhancement of the dura mater adjacent to the meningioma. During the operation, the tumor itself was strongly fluorescent, and the surrounding dura mater was not. Pathological examination confirmed the existence of tumor cells in the fluorescent area, but none were seen in the nonfluorescent dura mater. This suggests that PDD using 5-ALA is quite useful in determining the extent of dura mater involvement in meningioma surgery. 5-ALA has been reported to have several adverse effects, such as skin sensitivity (phototoxicity), nausea, vomiting, and transient liver dysfunction. 5-ALA also produces protoporphyrin IX, which may increase the risk of phototoxic skin reactions within 48 hours of induction. Therefore, after the administration of 5-ALA, we kept the patient in dark surroundings for 48 hours. Thus far, we have used PDD using 5-ALA with a low-dose regimen in 75 cases, but we have never experienced such serious adverse effects. The reliability of PDD using 5-ALA has not been fully verified. Positive reactions to photosensitive materials in nontumor tissue and 85% sensitivity in even malignant gliomas have been reported (1, 3, 5, 8). Nevertheless, to the best of our knowledge, this is the first case demonstrating the usefulness of PDD using 5-ALA for a meningioma with cranial invasion. PDD using 5-ALA is convenient and inexpensive, and, because most adverse reactions are avoidable, it may be applied in diagnosing brain tumors other than malignant gliomas.

REFERENCES 1. Boehncke WH, Rück A, Naumann J, Sterry W, Kaufmann R: Comparison of sensitivity towards photodynamic therapy of cutaneous resident and infiltrating cell types in vitro. Lasers Surg Med 19:451–457, 1996. 2. Haglund MM, Berger MS, Hochman DW: Enhanced optical imaging of human gliomas and tumor margins. Neurosurgery 38:308–17, 1996. 3. Kirdaite G, Lange N, Busso N, Van Den Bergh H, Kucera P, So A: Protoporphyrin IX photodynamic therapy for synovitis. Arthritis Rheum 46:1371–1378, 2002. 4. Kowalczuk A, Macdonald RL, Amidei C, Dohrmann G, Erickson RK, Hekmatpanah J, Krauss S, Krishnasamy S, Masters G, Mullan SF, Mundt AJ, Sweeney P, Vokes EE, Weir BK, Wollmann RL: Quantitative imaging study of extent of surgical resection and prognosis of malignant astrocytomas. Neurosurgery 41:1028–1038, 1997. 5. Messmann H, Kullmann F, Wild T, Knüchel-Clarke R, Rüschoff J, Gross V, Schölmerich J, Holstege A: Detection of dysplastic lesions by fluorescence in a model of colitis in rats after previous photosensitization with 5-aminolaevulinic acid. Endoscopy 30:333–338, 1998.


MOROFUJI ET AL.

6. Moore GE, Peyton WT, French LA: The clinical use of fluorescein in neurosurgery. The localization of brain tumors. J Neurosurg 5:392–398, 1948. 7. Stummer W, Novotny A, Stepp H, Goetz C, Bise K, Reulen HJ: Fluorescenceguided resection of glioblastoma multiforme by using 5-aminolevulinic acidinduced porphyrins: A prospective study in 52 consecutive patients. J Neurosurg 93:1003–1013, 2000. 8. Stummer W, Stocker S, Wagner S, Stepp H, Fritsch C, Goetz C, Goetz AE, Kiefmann R, Reulen HJ: Intraoperative detection of malignant gliomas by 5-aminolevulinic acid-induced porphyrin fluorescence. Neurosurgery 42:518–526, 1998. 9. Stummer W, Pichlmeier U, Meinel T, Wiestler OD, Zanella F, Reulen HJ, ALAGlioma Study Group: Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: A randomised controlled multicentre phase III trial. Lancet Oncol 7:392–401, 2006.

Acknowledgments We thank Keisuke Tsutsumi, M.D., Kentaro Hayashi, M.D., Tomohito Hirao, M.D., Naoe Kinoshita, M.D., and Keisuke Toyoda, M.D., for their help in preparing this manuscript.

COMMENTS

T

he authors discuss a case report of an 83-year-old woman with an anterior frontal fossa cranial base meningioma, for whom they used protoporphyrin 5-aminolevulinic acid (5-ALA)-induced fluorescence to detect the extent of tumor and cranial resection. This interesting technique seems to be reasonably specific for gross tumor tissue, and the authors provide histopathological confirmation that the nonfluorescent dura showed no tumor cells. However, they do not show definitive histopathological confirmation that all the fluorescent tissue was indeed tumor. Previous reports in malignant gliomas (1) have suggested that falsely negative fluorescence may be detected in areas of blood-brain barrier breakdown that are not gross tumor. As such, it remains unclear what the threshold for detection at the tumor margins are with this technique. The goal of this case report, as of others before it, was to intraoperatively detect the extent of tumor infiltration to optimize tumor resection. The problem with using this technique, however, has been that all tumor cells do not take up the fluorescent dye and all the areas of fluorescence are not necessarily tumor cells. This fact has contributed to the failure of photodynamic therapy using porphyrins for malignant gliomas and will also limit the ability to use “photodynamic diagnosis” to delineate with real accuracy the extent of “gross total resection.” Furthermore, unlike infiltrating gliomas, meningiomas are relatively well circumscribed. Therefore, the added value of using photodynamic diagnosis to visualize meningiomas beyond the abnormalities that can already be seen under the operating microscope is unclear. Unlike gliomas, meningiomas and cranial lesions can be distinguished from normal surrounding tissues using the operating microscope. Usually, the neurosurgeon’s ability to detect changes in texture, color, and vascularity that characterize cranial base lesions from surrounding normal tissues using the operating microscope is much greater than the ability to do so for intrinsic brain tumors. The difficulty of resecting cranial base lesions is usually not due to lack of visualization of tumor cells, but usually due to the location of the tumor. Therefore, it is often the functional anatomy (i.e., proximity to cranial nerves or major blood vessels), rather than the presence/absence of microscopic tumor tissue, that guides cranial base surgeons in their resections. Nevertheless, this report suggests that 5-ALA-induced fluorescence can be used in the case of meningioma with cranial invasion. The technique seems to be reasonably safe, and it could be moderately helpful for selected patients with such cranial base lesions. This


single case report, however, is primarily of anecdotal interest. The clinical practicality of this application for cranial base tumors is far from being demonstrated. Linda M. Liau Los Angeles, California

1. Stummer W, Pichlmeier U, Meinel T, Wiestler OD, Zanella F, Reulen HJ: ALAGlioma Study Group: Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: a randomised controlled multicentre phase III trial. Lancet Oncol 7:392–401, 2006.

T

his is a report of the original use of 5-ALA as a diagnostic labeling tool for tumors. There have been previous case reports or small series of patients in whom 5-ALA was used to identify spinal cord tumors (1). The authors have put together a photodynamic diagnosis of a large cranial base meningioma. Because this article is a case report, we do not know the extent of staining in the entire tumor, and the authors did not do a blind analysis of samples taken from different locations which were positive or negative for labeling with 5-ALA. Overall though, it is a nice example of the possibilities of photodynamic diagnosis for cranial base tumors such as meningiomas, but additional studies are necessary to determine whether the staining accurately matches the actual tumor. Michael M. Haglund Durham, North Carolina

1. Arai T, Tani S, Isoshima A, Nagashima H, Joki T, Takahashi-Fujigasaki J, Abe T: Intraoperative photodynamic diagnosis for spinal ependymoma using 5aminolevulinic acid: Technical note [in Japanese]. No Shinkei Geka 34:811–817, 2006.

T

his article is well illustrated, and as a case report, makes a valuable contribution. Because of my prior experience with photoradiation therapy and detection of tumors with hematoporphyrin derivative, I had not been a prior advocate of 5-ALA fluorescent staining for the removal of parenchymal tumors. It seemed clear that tumor bulk would take up the photoactive dye in an inhomogeneous fashion and that infiltrating tumor cells that were partially protected by the bloodbrain barrier often did not. Nonetheless, it is interesting to see how clearly tumor invading bone is identified by this technique. For that reason, I think this case report is of significant interest and may provide practical benefit. Edward R. Laws, Jr. Boston, Massachusetts

T

his technique may prove to be useful. I am still not certain about the sensitivity and specificity of this technique, and a single case is not going to answer that question. However, bone invasion and hyperostosis from meningiomas can be hard to detect during surgery. The authors should extend these studies to include more experience with other patients. Joseph M. Piepmeier New Haven, Connecticut

T

his is an interesting case report describing the use of photodynamic visualization with 5-ALA to guide resection of a meningioma invading the cranium. This technique has been described previously


PHOTODYNAMIC DIAGNOSIS FOR CRANIAL INVASION

for gliomas wherein margins between tumor and brain are less well defined. The benefits of this technique in patients with meningiomas are marginal because the tumor interface with surrounding structures is usually easily discernible. Additional studies, however, may demonstrate its utility in patients with bone invasion such as the one in this case report. Jeffrey N. Bruce New York, New York

NEUROSURGERY

T

he authors have described the use of 5-ALA for “photodynamic diagnosis” to help in the excision of an invasive meningioma. The authors propose that the use of this technique helps to ensure that the resection has been complete. They are to be congratulated on this interesting report, although a single case report does not prove that use of this technique is superior to just a radical resection. Andrew H. Kaye Melbourne, Australia



ENDOSCOPE-CONTROLLED MICRONEUROSURGERY TO TREAT MIDDLE FOSSA EPIDERMOID CYSTS: TECHNICAL CASE REPORT Felipe P. Trivelato, M.D. Division of Neurosurgery, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil

Alexandre V. Giannetti, M.D. Division of Neurosurgery, Federal University of Minas Gerais, Belo Horizonte, Minas Gerais, Brazil Reprint requests: Alexandre V. Giannetti, M.D., Rua Santa Catarina, 1042, Apt. 201, CEP: 30170–080, Lourdes, Belo Horizonte, Minas Gerais, Brazil. Email: [email protected] Received, February 25, 2007. Accepted, September 24, 2007.

NEUROSURGERY

OBJECTIVE: To present an alternative technique of endoscope-controlled microneurosurgery for the treatment of middle fossa epidermoid cysts. METHODS: The three operations described were performed through an approximately 2-cm diameter temporal craniotomy after a straight skin incision was made. Resection was then performed under the magnification of a 30-degree rigid endoscope, which mandated the use of exclusively conventional microsurgical instruments. RESULTS: Total resection was accomplished in all three patients with large middle fossa epidermoid cysts through a small temporal corticectomy, without damage to neurovascular structures. CONCLUSION: This procedure allowed the association of a smaller craniotomy, better cosmetic results, and minor retraction of the brain to wide resection of the tumor and satisfactory functional outcomes. KEY WORDS: Endoscope-controlled, Epidermoid, Middle fossa, Neuroendoscopy Neurosurgery 62[ONS Suppl 1]:ONSE105–ONSE107, 2008

E

pidermoid cysts or tumors are histologically benign, slow-growing congenital neoplasms of the central nervous system that may arise from retained ectodermal implants (1, 15). They account for only 0.2 to 1.8% of all intracranial tumors (1, 3, 4, 6, 17). They are generally found at the base of the cranium, in the sellar or parasellar region and the cerebellopontine angle (3, 4, 6, 9). Most patients become symptomatic in the third or fourth decade of life. Magnetic resonance imaging (MRI) scans show a homogeneous, nonenhancing mass that is nearly isointense to cerebrospinal fluid, and magnetic resonance spectroscopy reveals a lactate peak at 1.3 ppm (16). These radiological characteristics on routine MRI scans make epidermoid and arachnoid cysts similar (6). More sophisticated methods, especially the use of fluid-attenuated inversion recovery, constructive interference in steady state, and diffusion-weighted imaging, display areas of marked hyperintensity in epidermoid tumors and provide definitive radiological diagnosis and important clues about microneurosurgical anatomy. Total microsurgical removal is considered to be the therapy of choice. The standard surgical

DOI: 10.1227/01.NEU.0000297029.82996.B8

approaches to middle fossa lesions are the pterional craniotomy and the temporal minicraniotomy as proposed by Yan and Yu (17). Endoscopes have become far more than simple optical instruments, and increasing numbers of neurosurgical diseases can be approached endoscopically (12, 18). The use of endoscopes may help to reduce retraction and, at the same time, avoid additional dura and bone resection (7). There are some reports about endoscopic microneurosurgery for the treatment of cerebellopontine angle epidermoid cysts. However, in most cases, the microscope was used during resection, whereas endoscopy was reserved for locating tumor remnants (4, 15). We present an alternative management of middle fossa epidermoid cysts taking into consideration the principles of endoscopic and minimally invasive neurosurgery.

Surgical Technique While under general anesthesia, the patient is placed in a supine position with the head turned away from the side of the lesion, parallel to the floor, above heart level, on a horseshoe headrest, and the shoulder is raised on a roll, to prevent jugular compression. A small,


TRIVELATO AND GIANNETTI

A

D

C

B

F

E

FIGURE 1. A, photograph showing the skin incision. B, intraoperative photograph showing the craniotomy measuring approximately 2 cm in diameter. C, endoscopic view after macroscopic tumor debulking. D, tumor removal

straight skin incision is made in the temporal region. The pericranium and the temporalis muscle are cut along that incision. A craniotomy 2 cm in diameter is performed and the dura is opened; this is followed by an underlying corticectomy. Macroscopic tumor debulking is performed, which creates a cavity for the endoscopic surgery. Under the view of a 30degree rigid endoscope with an outer diameter of 4 mm (Karl Storz, Tuttlingen, Germany), and using conventional microsurgical instruments, the mass is removed in a piecemeal fashion. The thickened arachnoid is kept intact in an attempt to avoid aseptic meningitis secondary to the spreading of the cyst contents. A standard closure in layers is made (Fig. 1).

Illustrative Cases

Patient 1 An 18-year-old man presented with a history of complex partial seizures that had occurred since he was 11 years old. His seizures were classified as partial complex with prominent automatisms involving the upper extremities and face, usually preceded by a gustative aura, with occasional secondary tonic-clonic generalization. Despite treatment with 600 mg/day of carbamazepine and 100 mg/day of phenobarbital, the seizures continued at a rate of approximately four per week. The neurological examination was normal. A computed tomographic (CT) scan demonstrated a low-density, homogeneous lesion in the right temporal lobe. An MRI scan revealed a homogeneous, nonenhancing


with conventional instruments (curette). E, after total removal, the internal carotid artery (asterisk) and the optic nerve (arrow) are shown in detail, covered by arachnoid. F, final macroscopic view.

right temporal mass that was hypointense on T1-weighted images and hyperintense on T2-weighted images. It appeared heterogeneous on fluid-attenuated inversion recovery and bright on diffusion-weighted images (Fig. 2). Scalp electroencephalography detected interictal epileptiform discharges in the right temporal area. The patient underwent surgical treatment following the technique described previously. He was discharged 3 days after the surgery with gradual reduction of steroids. After 1 year, he was free from seizures with 1000 mg/day of carbamazepine. A control MRI scan showed no recidivistic or remnant tumor.

Patient 2 A 31-year-old woman with an 18-year history of controlled partial complex epilepsy who was previously taking 300 mg/day of phenytoin and 100 mg/day of phenobarbital presented to the emergency room in status epilepticus after the discontinuation of self-therapy. There were no abnormalities on postictal examination. A CT scan demonstrated a large, hypodense, homogeneous, left temporal lesion reaching the suprasellar cistern and compressing the midbrain, with a slight distortion of the left lateral ventricle (Fig. 3). The patient was surgically treated with the same technique mentioned previously and was discharged 3 days later with steroid prescription and no deficits; she was lost to follow-up.

Patient 3 A 10-year-old boy presented with a history of almost 1 year of headache, nausea, and vomiting. A CT scan showed a right hypodense


ENDOSCOPE-CONTROLLED MICRONEUROSURGERY FOR TEMPORAL EPIDERMOID CYSTS

A

B

C

sion was resected using the same technique as in the previously described patients with the diagnosis of an epidermoid cyst. Recovery was uneventful, and the patient was discharged 3 days later, on steroid therapy. At this time he was asymptomatic, and a control MRI scan showed no tumor 7 months after resection.

DISCUSSION Epidermoid cysts are histopathologically benign, and the goal of the treatment should be complete surgical resection with no damage to vital neurovascular structures (15). It carries significant perioperative risk, including permanent cranial nerve palsies, hydrocephalus, and aseptic meningitis as a result of breakdown products (keratin and cholesterol) produced FIGURE 2. A, preoperative coronal T1-weighted magnetic resonance imaging (MRI) scan showing a hypointense during the desquamation of lesion in the right temporal fossa with mass effect. B, preoperative fluid-attenuated inversion recovery image showepithelial cells (1, 3, 15). ing a heterogeneous lesion. C, preoperative diffusion-weighted MRI scan showing a typical hyperintense epidermoid Complete removal elimitumor. D, postoperative coronal T1-weighted MRI scan of a smaller hypointense area showing resolution of mass nates the chance of epidereffect. E, postoperative fluid-attenuated inversion recovery image showing a hypointense area indicating cerebrospinal moid tumor recurrence and fluid content. F, postoperative diffusion-weighted MRI scan showing no tumor remnant. diminishes the risk of postoperative aseptic meningitis (1, 3, 15). This complication is estimated to occur in 40% of the A B patients who undergo subtotal resection. The recurrence rate is greater than 24% (1). There is no established role for radiotherapy or chemotherapy, even for residual or recurrent lesions. Most authors advocate the pterional approach to middle fossa tumors. Nevertheless, this approach carries some disadvantages in comparison with our proposal; these include larger bone resection, especially in the greater sphenoid wing, and wider detachment of temporal muscle with a possibly unfavorable aesthetic result, which we did not observe in our technique, and a longer operative time (2, 11, 13). Although Yan and Yu (17) described a minicraniotomy, the bone flap diameter was not mentioned, and this technique should be restricted to a particular situation, namely an intracerebral epiFIGURE 3. A and B, computed tomographic (CT) scans showing an dermoid tumor. expansive, well-limited, hypodense, nonenhancing left temporal lesion. In the proposed technique, we perform a corticectomy of approximately 1 cm in a noneloquent area (anterior portion of medial anterior temporal lesion with no mass effect. The patient’s neuthe medial temporal gyrus). Because the endoscope is inserted rological examination was normal, and his symptoms resolved spontainside the tumor cavity, bringing with it the light that will illuneously. The management plan at the time was conservative, with a minate the operating field, this incision can be small. That is the diagnosis of migraine and an incidental arachnoid cyst. principle of endoscope-controlled or -assisted microneuroAfter the patient was asymptomatic for 4 years, a follow-up CT scan surgery (14). In the transsylvian pterional approach, there is no demonstrated an important increase in the lesion size, with distortion of the right lateral ventricle nonenhanced by contrast (Fig. 4). This lecerebral tissue resection. However, the retraction of the tempo-

D

NEUROSURGERY

E

F


TRIVELATO AND GIANNETTI

A

B

CONCLUSION

C

FIGURE 4. A, CT scan demonstrating a right temporal lesion. B, after 4 years, there was an increase in the lesion size. C, diffusion-weighted MRI scan obtained 7 months after surgery demonstrating no tumor remnants.

Although the three cases presented here were not compared with exclusively microsurgical procedures, we recommend endoscope-controlled microneurosurgery as a useful alternative method for epidermoid cyst resection in an attempt to minimize surgical trauma, improve visualization, shorten patient hospitalization time, and reduce surgical expense.

REFERENCES ral lobe that is necessary for external light to penetrate and illuminate the whole field, especially the corners, could be equal or even worse for the temporal lobe. This advantage of the proposed technique does not exist when we face small lesions in the medial area of the middle fossa. In this last case, it is possible that the transsylvian approach could be more appropriate, and the endoscope could be used only at the end of a microsurgical resection (assisted technique). The technique described here was performed only on epidermoid tumors. Because it depends on the presence of a cavity to be performed and bleeding may completely obscure the endoscopic view, we believe it could be used particularly in avascular and/or cystic lesions (3, 8, 15). Endoscope-controlled microneurosurgery is distinguished from endoscope-assisted microneurosurgery in that an operating microscope is not used, whereas both use conventional microsurgical instruments exclusively (12). Treatment with endoscope-assisted and controlled microneurosurgical techniques has already been reported for epidermoid tumors of the posterior fossa, with no reference to middle fossa lesions thus far (4, 15). We think neuroendoscopy can be an important tool for gross total resection of epidermoid cysts with minimal trauma. The advantages of this technique are the use of standard neurosurgical instruments and the freedom of movement, combined with an excellent endoscopic picture. Epidermoid tumors spread along pathways of lower resistance, invaginating into and expanding through the cisterns, sulcus, and subarachnoid space. Without the 30-degree optical view, which provides excellent visualization, the aim of looking into hidden but important corners of the operating field could not be accomplished. Nevertheless, the capsule may be firmly adherent to neurovascular structures, making any attempt at thorough removal risky. Being that they are markedly slow-growing tumors, epidermoid cysts may recur several years later. Therefore, the follow-up period in this study was too short to allow conclusions to be drawn on the superiority of this technique regarding recurrence.


1. Akar Z, Tanriover N, Tuzgen S, Kafadar AM, Kuday C: Surgical treatment of intracranial epidermoid tumors. Neurol Med Chir (Tokyo) 43:275–281, 2003. 2. Badie B: Cosmetic reconstruction of temporal defect following pterional [corrected] craniotomy. Surg Neurol 45:383–384, 1996. 3. Caldarelli M, Massimi L, Kondageski C, Di Rocco C: Intracranial midline dermoid and epidermoid cysts in children. J Neurosurg 100 [Suppl]:473–480, 2004. 4. Darrouzet V, Franco-Vidal V, Hilton M, Nguyen DQ, Lacher-Fougere S, Guerin J, Bebear JP: Surgery of cerebellopontine angle epidermoid cysts: Role of the widened retrolabyrinthine approach combined with endoscopy. Otolaryngol Head Neck Surg 131:120–125, 2004. 5. David EA, Chen JM: Imaging case of the month posterior fossa epidermoid cyst. Otol Neurotol 24:699–700, 2003. 6. Dutt SN, Mirza S, Chavda SV, Irving RM: Radiologic differentiation of intracranial epidermoids from arachnoid cysts. Otol Neurotol 23:84–92, 2002. 7. Fries G, Perneczky A: Endoscope-assisted brain surgery: Part 2—Analysis of 380 procedures. Neurosurgery 42:226–231, 1998. 8. Gaab M, Schroeder HW: Neuroendoscopic approach to intraventricular lesions. J Neurosurg 88:496–505, 1998. 9. Inoue Y, Ohata K, Nakayama K, Haba T, Shakudo M: An unusual middle fossa interdural epidermoid tumor. Case report. J Neurosurg 95:902–904, 2001. 10. Kang SD: Pterional craniotomy without keyhole to supratentorial cerebral aneurysms: Technical note. Surg Neurol 60:457–462, 2003. 11. Kurosaki K, Hayashi N, Hamada H, Hori E, Kurimoto M, Endo S: Multiple epidermoid cysts located in the pineal and extracranial regions treated by neuroendoscopy. Neurol Med Chir (Tokyo) 45:216–219, 2005. 12. Nikolai J, Hopf L, Perneczky A: Endoscopic neurosurgery and endoscopeassisted microneurosurgery for the treatment of intracranial cysts. Neurosurgery 43:1330–1337, 1998. 13. Park J, Hamm IS: Cortical osteotomy technique for mobilizing the temporal muscle in pterional craniotomies. Technical note. J Neurosurg 102:174–178, 2005. 14. Perneczky A, Fries G: Endoscope-assisted brain surgery: Part 1—Evolution, basic concept, and current technique. Neurosurgery 42:219–225, 1998. 15. Schroeder HW, Oertel J, Gaab MR: Endoscope-assisted microsurgical resection of epidermoid tumors of the cerebellopontine angle. J Neurosurg 101:227–232, 2004. 16. von Koch CS, Young G, Chin CT, Lawton MT: Magnetic resonance imaging/spectroscopy of an intraaxial epidermoid: Similarity to an abscess. Case illustration. J Neurosurg 97:492, 2002. 17. Yan PX, Yu CJ: Minicraniotomy treatment of an intracerebral epidermoid cyst. Minim Invasive Neurosurg 47:245–248, 2004. 18. Zhang Y, Wang C, Liu P, Gao X: Clinical applications of neuroendoscopic techniques. Stereotact Funct Neurosurg 75:133–141, 2000.

COMMENTS

T

his is certainly an interesting article. I especially appreciate the correct definitions: “endoscopic surgery” for (coaxial) endoscopic tech-


ENDOSCOPE-CONTROLLED MICRONEUROSURGERY FOR TEMPORAL EPIDERMOID CYSTS

nique, “endoscopy assistance” for the use of the endoscope in addition (and at certain periods of surgery) to the microscope, and “endoscopy controlled,” as in this report, for the use of endoscopic visualization only, but in an “open” (but small) approach using microsurgical dissection instruments and techniques. Endoscopy controlled resection in such temporal lesions has rarely (if at all) been reported; however, this is probably because such epidermoid cysts are quite rare. As cited, our group has resected epidermoid cysts in the posterior fossa: There was no restriction to this area; we simply have not seen epidermoid cysts in the temporal region as those presented in this report. However, whether the technique described is really less invasive than a pterional, perhaps somewhat larger, approach (with a longer skin incision and slightly larger trephination) in endoscopy-assisted surgery has to be discussed: The reason for endoscopy assistance in posterior fossa surgery is the minimally invasive classic retromastoidal approach through a preformed (retropetrosal) fissure combined with the advantage of the (30 and 45 degrees) rigid scope to give an excellent image “around the corner,” e.g., supratentorial or along the trigeminal nerve and to allow a more radical, but atraumatic, resection of recurrence-prone epidermoid cysts in these spaces, which are not properly presented under the microscope. In this report on temporal epidermoid cysts, first, the positioning must be questioned: Head turned 90 degrees away from the side of the lesion only on a horseshoe headrest? Turning the head so far to the side provokes a risk of an increase in intracranial pressure, which makes preparation more damaging, and, furthermore, for endoscopy, the head should be fixed, e.g., in a Mayfield clamp. In our experience, this fixation is absolutely necessary to be sure that no head movement occurs during insertion of the scope, which could lead to brain damage. Thus, I recommend positioning with a maximum 45-degree turn in a Mayfield-like clamp for the pterional approach. Second, and more important, I question the kind of skin incision: The almost vertical incision (only slightly angled from apical-posterior to anterior-basal) is not much shorter than a small curved incision for a small pterional approach. An initial disadvantage of the straight incision is the need to cut through the temporal muscle. The incision

NEUROSURGERY

should be avoided (M.G. Yas¸argil, personal communication), as it results in temporal muscle atrophy. The temporal muscle should, therefore, be cut at its temporal line, only pushed away for a small pterional trephination, and later repositioned; this procedure is really “less invasive,” even with a bit longer curved skin cut. Further, the skin incision is apparently relatively high; as the images after dura opening show, the epidermoid cyst is then approached transcortically through the temporal lobe (even “corticectomy”). However, according to the imaging of the tumors, in all three patients, a transfissural approach seems to be possible—with the opportunity for total tumor removal, using the advantages of the endoscope, without transcortical damage—with an individually adjusted pterional approach, in case one also needs a more temporobasal approach for the use of existent spaces, not going through brain parenchyma. Minimally invasive does not mean small skin incision-small cranial opening; minimally invasive is preservation of important functional structures! Michael R. Gaab Hannover, Germany

T

he authors provide a nice description of the use of an endoscope to minimize the incision required to remove an intraparenchymal epidermoid tumor. In general, endoscopy is useful when one is operating in an air-filled or fluid-filled cavity, such as the sinuses or ventricles. Applications of endoscopy during intraparenchymal surgery are limited because the walls of the resection cavity can obscure the view as the tumor is decompressed, and bleeding from the tumor edges may opacify the lens if the endoscope sits within the cavity. The authors have wisely chosen to apply endoscopy to epidermoid tumors as these tumors are essentially avascular with an excellent plane between the tumor and the parenchyma. Applications of endoscopy in the removal of other more common intraparenchymal tumors are progressing through the use of stereotactically guided tubular retractors which, I predict, will have an increasing role in surgical neuro-oncology. Theodore H. Schwartz New York, New York



ENDOSCOPIC SUPRACEREBELLAR INFRATENTORIAL APPROACH FOR PINEAL CYST RESECTION: TECHNICAL CASE REPORT Pankaj A. Gore, M.D. Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

L. Fernando Gonzalez, M.D. Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Harold L. Rekate, M.D. Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Peter Nakaji, M.D. Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona Reprint requests: Peter Nakaji, M.D., c/o Neuroscience Publications, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013. Email: [email protected] Received, September 25, 2006. Accepted, May 1, 2007.

NEUROSURGERY

OBJECTIVE: Accepted surgical strategies to address symptomatic pineal cysts include transventricular flexible or rigid endoscopy and supracerebellar infratentorial or occipital transtentorial microsurgical approaches. We report the first application of the endoscopic supracerebellar infratentorial approach for the complete resection of a pineal cyst. Unlike transventricular endoscopy, this technique poses no risk to the fornices and can be applied independent of ventricular size. CLINICAL PRESENTATION: A 37-year-old woman sought treatment for intractable headaches. A thorough evaluation revealed only a pineal cyst exerting mass effect on the tectum but causing no hydrocephalus. A period of nonoperative management was unsuccessful, and the patient was referred for surgery. TECHNIQUE: The patient was positioned in the semi-sitting position. The supracerebellar infratentorial corridor was accessed through a burr-hole. The pineal cyst was resected completely via the endoscope. Postoperatively, the patient’s headaches resolved completely. CONCLUSION: The endoscopic supracerebellar infratentorial approach involves minimal brain retraction, poses no risk to the fornices, allows visualization and avoidance of the Galenic veins, and can be performed regardless of the size of the ventricle. Consequently, it is an excellent minimally invasive surgical option for resection or fenestration of symptomatic pineal cysts. KEY WORDS: Endoscopy, Pineal cyst, Supracerebellar infratentorial approach Neurosurgery 62[ONS Suppl 1]:ONSE108–ONSE109, 2008

G

lial cysts of the pineal gland are found in 1.8 to 4.3% of healthy subjects evaluated with magnetic resonance imaging (MRI) scans (5, 10, 21, 30). In young women, this incidence approaches 6% (30). Although usually asymptomatic, pineal cysts can grow sufficiently large to exert mass effect on the tectum, aqueduct, and surrounding venous structures and thereby necessitate treatment (3, 6, 7, 14, 16, 22–24, 32, 34–36). On computed tomographic and MRI scans, pineal cysts may also be indistinguishable from less benign pineal pathological lesions such as pineocytomas and epidermoid cysts (7, 8, 16, 18, 22, 33). Several surgical strategies have been described for the treatment of pineal cysts, including the supracerebellar infratentorial (3, 7, 22–24, 31) and occipital transtentorial (14, 25, 37) microsurgical approaches, the

DOI: 10.1227/01.NEU.0000297005.97350.7D

transventricular (9, 24, 35, 36) endoscopic approach, and stereotactic aspiration (32). We report the first patient whose pineal cyst was entirely excised transendoscopically through the supracerebellar infratentorial approach.

CASE PRESENTATION History A 37-year-old, previously healthy woman sought treatment after having severe, escalating occipital headaches associated with blurry vision and occasional nausea for 2 months. Her neurological examination disclosed nothing abnormal. Pertinent normal findings included the absence of papilledema, nystagmus, or abnormal extraocular movements. MRI examination of the brain demonstrated a 1.0 ⫻ 1.1 ⫻ 1.5-cm ring-enhancing lesion of the pineal region with an anterior nodular component and the suggestion of a small septation (Figs. 1 and 2). On T2-


GORE ET AL.

weighted MRI sequences, the contents of the cyst were slightly hyperintense compared with cerebrospinal fluid (CSF). On fluid-attenuated inversion recovery sequences, the cystic contents were markedly hyperintense compared with CSF (Fig. 3). The cyst exerted mass effect on the superior colliculus, but the aqueduct was patent. No hydrocephalus was present. The imaging findings were most consistent with a pineal cyst. Conservative management with reimaging at 1 year was recommended, and the patient was referred to a neurologist for further evaluation of her headaches. However, no other cause for the patient’s symptoms was identified. Two months later, she returned for treatment because the persistent intractable headaches were affecting her quality of life. Her neurological examination again disclosed nothing abnormal. The patient was offered the option of undergoing an endoscopic supracerebellar infratentorial approach for cyst fenestration, biopsy and attempted resection. An additional rationale for surgery was to obtain histopathological diagnosis of the anterior nodular component of the lesion.

FIGURE 1. Axial postgadolinium T1-weighted magnetic resonance imaging (MRI) scan showing a cystic lesion of the pineal region.

FIGURE 2. Sagittal postgadolinium T1-weighted MRI scan showing a cystic lesion of the pineal region.

Operative Technique The patient underwent preoperative contrast MRI examination of the brain to obtain thin-cut volumetric images. These images were transferred to the StealthStation Treon (Medtronic Navigation, Boulder, CO). The patient was placed in the semi-sitting position, and appropriate precautions for anesthesia were undertaken. A surface-merge registration to the patient was performed. The frameless stereotactic system was used to localize the torcula. A 15-mm burr hole was placed just inferior to the left transverse sinus, about 5 mm to the left of midline. The paramedian location was selected to avoid the occipital sinus. Bone was removed at the rostral margin of the burr hole until the infe-

FIGURE 3. Axial fluid-attenuated inversion recovery MRI scan showing that the cystic contents are markedly hyperintense compared with cerebrospinal fluid.


rior edge of the transverse sinus was exposed. The dura was opened in cruciate fashion. CSF was aspirated from the supracerebellar cistern, and intravenous mannitol was administered to relax the cerebellum. A SureTrak Universal Instrument Adaptor (Medtronic Navigation) was affixed to the 6-mm trocar (Aesculap Instruments Corp., South San Francisco, CA) FIGURE 4. Intraoperative phototo enable neuronavigation with graph showing the endoscopic the endoscope. The 30-degree release of vermis (V) tethered by endoscope, ensheathed within thickened arachnoid. The tentothe trocar, was advanced along rium (T), splenium of corpus calthe supracerebellar infratentorial losum (S), and pulvinar nucleus trajectory. Bridging veins from (P) are visible. the cerebellar hemisphere were coagulated and divided to aid cerebellar relaxation and to eliminate the risk of avulsion. Maintaining the endoscope parallel to the tentorium leads directly to the vein of Galen rather than to the pineal region located caudal to the venous confluence. The endoscope was therefore directed slightly inferior to the tentorium along a trajectory guided by the frameless FIGURE 5. Intraoperative photostereotactic system. There was graph showing the posterior pineal ample room for the endoscope cyst wall before fenestration. along this trajectory, and no additional cerebellar retraction was required. The thick arachnoid tethering the superior vermis was dissected sharply to expose the precentral cerebellar vein (Fig. 4). The pineal cyst was slightly eccentric to the left side, tightly juxtaposed between the left pulvinar nucleus and precentral cerebellar vein. The capsule was cauterized, and the cyst was fenestrated (Figs. 5 and 6). After decompression, the cyst was separated from overlyFIGURE 6. Intraoperative photoing vessels with sharp dissection. graph showing the decompressed Pediatric grasping forceps, incyst wall within the jaws of the serted from the side port of the grasping instrument. trocar, were used to provide traction while endoscopic scissors were wielded through the main port. This “mother-daughter” technique allowed a semblance of “bimanual” dissection (Fig. 7). The cyst was removed piecemeal but completely (Fig. 8). The dural opening was sealed with a piece of Gelfoam (Upjohn Co., Kalamazoo, MI), followed by a layer of DuraSeal (Confluent Surgical, Waltham, MA). The remainder of the closure was accomplished in the standard way. Postoperatively, the patient had no neurological deficit. On postoperative Day 1, she was discharged to her home in excellent condition.


ENDOSCOPIC APPROACH FOR PINEAL CYST RESECTION

An MRI scan performed immediately after the operation confirmed complete resection of the pineal cyst (Fig. 9). Pathological analysis demonstrated that the cyst wall was lined by pineocytes and gliotic tissue with scattered Rosenthal fibers and dystrophic calcifications. At her 2-week and 4-month postoperative visits, the patient reported resolution of her headaches and required no pain medications. At a 15-month follow-up examination conducted on the telephone, the patient remained headache-free and independent of medications.

DISCUSSION As discussed by Horwitz (13), Krause first described the supracerebellar-infratentorial approach to the pineal region in 1913. Stein (31) popularized the approach in the microsurgical era, and it remains a dependable method for accessing the pineal region (31). In the past two decades, neuroendoscopy has come to the forefront for the management of complex hydrocephalus and intracranial cysts. Ruge et al. (29) first reported purely endoscopic fenestration of arachnoid cysts involving the quadrigeminal region via the supracerebellar-infratentorial corridor. In a cadaver-based anatomic study, Cardia et al. (2) demonstrated the viability of the supracerebellar-infratentorial approach for endoscope-assisted techniques. They were able to access not only the pineal region but also the posterior third ventricle via a parapineal entry point. In practice, most endoscopists have avoided use of the supracerebellar approach to pineal cysts, preferring to access the third ventricle either with a flexible endoscope through a precoronal burr hole (9, 24, 35, 36) or with a rigid endoscope through an anterior frontal burr hole (9, 15, 26, 28). The perceived advantages of the flexible endoscope are the ability to fenestrate the pineal cyst and to perform a ventriculostomy in the floor of the third ventricle in one surgical procedure. This technique, however, also has disadvantages. Use of the flexible endoscope is more disorienting to the surgeon and its image quality is substantially inferior. Its use poses a risk of injury to the fornix at the foramen of Monro (12). Transventricular, transforaminal pineal cyst fenestration is also possible with a rigid endoscope. Frameless navigation is used to deterFIGURE 7. Intraoperative photomine an entry point and lingraph showing an example of the ear trajectory that traverses mother-daughter technique. The the foramen of Monro to pineal cyst capsule is under gentle traction from a grasping instruaccess the posterior third ment (not visualized) inserted via ventricle. Whereas use of the trocar side port. The bipolar the rigid endoscope proforceps is simultaneously inserted vides high-quality images, via the main port and is used to there is still a risk to the cauterize the pedicled attachment fornix. Additionally, treatof the capsule. ment of hydrocephalus with

NEUROSURGERY

FIGURE 8. Intraoperative photograph showing the resection bed after gross total removal. The internal cerebral veins (asterisk), the pineal vein (P), and the precentral cerebellar vein (C) are apparent.

FIGURE 9. Postoperative sagittal T1-weighted MRI scan after gadolinium administration showing postoperative changes and confirming complete resection of the cyst.

a third ventriculostomy requires a second precoronal burr hole. Both endoscopic transventricular techniques also require traversal of brain tissue, however noneloquent, with the attendant potential for hemorrhage. The endoscopic supracerebellar infratentorial approach for pineal cysts compares favorably with the endoscopic transventricular approaches. In the sitting position, gravity facilitates an ideal anatomic pathway for supracerebellar endoscopic access to the pineal region. Cerebellar relaxation is aided by CSF diversion, mannitol, and sacrifice of superior bridging veins. The resulting working corridor is 1 to 1.5 cm (17). The rigid endoscope provides excellent illumination and magnification and is easily oriented by the surgeon. The sitting position reduces engorgement of the vein of Galen complex, thereby facilitating sharp dissection of the arachnoid. Michielsen et al. (24) reported that the posterior (pineal) cyst wall is often highly vascularized. The transventricular endoscopic approach is not optimal for the complete resection of cysts because bleeding from the posterior pineal cyst wall is difficult to control and the adjacent vein of Galen complex is poorly visualized. Although endoscopic fenestration of cysts is usually sufficient, symptomatic pineal cysts are known to recur after subtotal resection or simple fenestration (22, 35). The endoscopic supracerebellar infratentorial approach affords more control over the vasculature that invests and abuts the posterior pineal cyst wall. This additional control reduces the chance of uncontrolled hemorrhage and increases the likelihood of attaining complete resection. The feasibility of complete resection also depends on the adherence of the cyst to the tectum (37). The presence of hydrocephalus greatly facilitates the use of flexible or rigid endoscopes in a transventricular approach. However, the size of ventricles in patients with symptomatic pineal cysts can be normal (6–8, 22, 24, 37). The endoscopic supracerebellar infratentorial approach can be used independent of ventricular size. In the absence of hydrocephalus, cyst resection or posterior fenestration into the quadrigeminal cistern is sufficient. If hydrocephalus is present, a “posterior”


GORE ET AL.

third ventriculocystostomy can be performed by fenestrating the anterior wall of the pineal cyst into the third ventricle. Hayashi et al. (11) reported a patient with triventricular hydrocephalus caused by an arachnoid cyst involving the quadrigeminal region. These authors established communication between the third ventricle and the quadrigeminal cistern by fenestrating both the anterior and posterior walls of the cyst. Daniel et al. (4) performed a posterior third ventriculostomy by fenestrating the thinned suprapineal recess in a patient with triventricular hydrocephalus. Both groups used a transventricular endoscopic approach. This anterior-toposterior fenestration can place critical structures, including the vein of Galen complex and major arteries, at risk (4). Theoretically, posterior-to-anterior communication of the quadrigeminal cistern to the third ventricle is safer because the vascular structures are readily visualized and bypassed. The endoscopic supracerebellar infratentorial approach also compares favorably with the open supracerebellar infratentorial approach in the sitting position. Foremost, the latter is associated with a significant risk of venous air embolism (1, 19, 20, 27). Although this complication can also occur with an endoscopic supracerebellar infratentorial approach, it is less likely because the dural sinuses are not exposed. It is also easier to flood the operative field with irrigation fluid and to occlude ingress of air with a simple burr hole. Operating via the microscope on a patient in the semi-sitting position can be fatiguing for the surgeon, whose arms must remain extended and shoulders abducted. The endoscope can be held at the level of the surgeon’s chest and is easily manipulated without strain. Furthermore, the wound associated with an endoscopic supracerebellar infratentorial approach is relatively small, thereby decreasing postoperative pain and morbidity compared with the open approach. The opening and closure proceed more rapidly with the endoscopic supracerebellar infratentorial approach. Given the current technology for endoscopic hemostasis and dissection, use of the endoscopic supracerebellar infratentorial approach should be limited to small or collapsible, relatively avascular lesions such as pineal cysts. With larger or more vascular lesions, microsurgery retains a significant advantage because the instrumentation is superior and true bimanual dissection is possible. As with any endoscopic procedure, the surgeon should always be prepared to proceed with a craniotomy in the event of catastrophic hemorrhage. Use of the endoscopic supracerebellar infratentorial approach is not without pitfalls. Our experience with this and similar procedures has led to several refinements in our preoperative and operative methodology. An anesthesiologist familiar with the sitting position and with management of its potential complications is requisite. Intraoperative frameless navigation is essential to avoid the dural sinuses during placement of the burr hole. To obtain the greatest accuracy for navigation, we advocate that patients in whom an endoscopic supracerebellar infratentorial approach is to be used have skin fiducial markers placed over the occipital region before undergoing the preoperative MRI scan.


The patient is placed in the semi-sitting position with the top of the operating table at least 15 cm below his or her shoulders. This position provides sufficient clearance for the pistol grip of the endoscope to avoid contact with the table. We prefer the Minop 2.7-mm endoscope (Aesculap Instruments Corp.) because of its excellent optics and ergonomics. The scope is ensheathed within a 6-mm trocar, which has a main working channel and two side ports. Instrumentation designed for the pediatric endoscope, such as grasping forceps or scissors (Aesculap Instruments Corp.), can be used from a side port, together with standard endoscopic instrumentation introduced via the main working channel. In this mother-daughter technique, one instrument is used to apply traction while the other is used for dissection. The burr hole is generous and positioned slightly off midline to avoid the occipital sinus. Bone edges are waxed thoroughly. In patients without hydrocephalus, preoperative placement of a lumbar drain greatly aids and hastens cerebellar relaxation. Further relaxation is aided by judicious use of intravenous mannitol and sacrifice of superior bridging veins. The sitting position not only is conducive to cerebellar relaxation but also allows significant reduction of venous pressure and permits an “air” working environment. The former facilitates dissection of the arachnoid around the vein of Galen complex, and the latter makes bleeding points easier to identify and control with the endoscope. Glial cysts of the pineal gland are usually asymptomatic findings that require no intervention (5, 10, 21, 25). Surgical treatment of pineal cysts is generally limited to lesions that are causing hydrocephalus by aqueductal obstruction. Other reported indications include mass effect on juxtapineal structures causing headaches in conjunction with symptoms such as ataxia, motor and sensory deficit and seizures (7, 8, 34), and signs such as diplopia and Parinaud’s syndrome (7, 8, 37). The pathophysiological mechanism behind motor and sensory deficits and seizures is not well understood. There is a challenging subset of patients with a presumed pineal cyst on imaging in the absence of hydrocephalus who present with headaches but for whom there are no clinical findings at physical examination. In one series, four of 33 patients with symptomatic pineal cysts had headaches as the only symptom, with no clinical signs or hydrocephalus (37). The patient we have presented falls into this category. She exhibited paroxysmal headaches, at times accompanied by nausea and blurred vision. We believe these symptoms were probably indicative of intermittent CSF outflow obstruction at the aqueduct caused by compression by the pineal cyst. Wisoff and Epstein (37) described this pathophysiological mechanism and its association with paroxysmal headache. In a series of pediatric patients with these characteristics, Mandera et al. (22) also attributed the headaches to disturbance of flow through the aqueduct and noted gliosis around the aqueduct in 54% of patients. Tamaki et al. theorized that the headaches result from venous outflow obstruction attributable to compression of the vein of Galen by the pineal cyst (33).


ENDOSCOPIC APPROACH FOR PINEAL CYST RESECTION

Nevertheless, a symptomatic pineal cyst in the absence of hydrocephalus is a rare entity. The first line of management should be a conservative approach focused on medical control of the headaches and serial MRI scans to monitor the lesion. If medical management fails, surgical intervention can be considered judiciously. The paroxysmal, as opposed to chronic, nature of the headaches is an important feature of the clinical presentation that favors surgical intervention. There are several reports in the literature regarding the surgical treatment of patients with pineal cysts and headaches in the absence of hydrocephalus (7, 8, 22, 24). In the case we have presented, the durable resolution of symptoms supports our treatment methodology.

CONCLUSION This is the first reported case of a purely endoscopic supracerebellar infratentorial approach used for resection of a pineal cyst. This minimally invasive technique can be used to resect or fenestrate pineal cysts, regardless of whether or not hydrocephalus is present. It presents no risk of forniceal injury, requires minimal brain retraction, and allows visualization and avoidance of the vein of Galen complex. The endoscopic supracerebellar infratentorial approach should be considered as a viable alternative to transventricular flexible endoscopy and microsurgical approaches for the treatment of symptomatic pineal cysts.

REFERENCES 1. Bithal P, Dash HH, Vishnoi N, Chaturvedi A: Venous air embolism: Does the site of embolism influence the hemodynamic changes? Neurol India 51:370–372, 2003. 2. Cardia A, Caroli M, Pluderi M, Arienta C, Gaini SM, Lanzino G, Tschabitscher M: Endoscope-assisted infratentorial-supracerebellar approach to the third ventricle: An anatomical study. J Neurosurg 104:409–414, 2006. 3. Chandy MJ, Damaraju SC: Benign tumours of the pineal region: A prospective study from 1983 to 1997. Br J Neurosurg 12:228–233, 1998. 4. Daniel RT, Lee GY, Reilly PL: Suprapineal recess: An alternate site for third ventriculostomy? Case report. J Neurosurg 101:518–520, 2004. 5. Di Costanzo A, Tedeschi G, Di Salle F, Golia F, Morrone R, Bonavita V: Pineal cysts: An incidental MRI finding? J Neurol Neurosurg Psychiatry 56:207–208, 1993. 6. Engel U, Gottschalk S, Niehaus L, Lehmann R, May C, Vogel S, Jänisch W: Cystic lesions of the pineal region–MRI and pathology. Neuroradiology 42:399–402, 2000. 7. Fain JS, Tomlinson FH, Scheithauer BW, Parisi JE, Fletcher GP, Kelly PJ, Miller GM: Symptomatic glial cysts of the pineal gland. J Neurosurg 80:454–460, 1994. 8. Fleege MA, Miller GM, Fletcher GP, Fain JS, Scheithauer BW: Benign glial cysts of the pineal gland: Unusual imaging characteristics with histologic correlation. AJNR Am J Neuroradiol 15:161–166, 1994. 9. Gaab MR, Schroeder HW: Neuroendoscopic approach to intraventricular lesions. Neurosurg Focus 6:E5, 1999. 10. Golzarian J, Balériaux D, Bank WO, Matos C, Flament-Durand J: Pineal cyst: Normal or pathological? Neuroradiology 35:251–253, 1993. 11. Hayashi N, Endo S, Tsukamoto E, Hohnoki S, Masuoka T, Takaku A: Endoscopic ventriculocystocisternostomy of a quadrigeminal cistern arachnoid cyst. Case report. J Neurosurg 90:1125–1128, 1999. 12. Hayashi N, Hamada H, Umemura K, Kurosaki K, Kurimoto M, Endo S: Selection of surgical approach for quadrigeminal cistern arachnoid cyst [in Japanese]. No Shinkei Geka 33:457–465, 2005.

NEUROSURGERY

13. Horwitz NH: Fedor Krause (1857–1937). Neurosurgery 38:844–848, 1996. 14. Kang HS, Kim DG, Han DH: Large glial cyst of the pineal gland: A possible growth mechanism. Case report. J Neurosurg 88:138–140, 1998. 15. Kim IY, Jung S, Moon KS, Jung TY, Kang SS: Neuronavigation-guided endoscopic surgery for pineal tumors with hydrocephalus. Minim Invasive Neurosurg 43:365–368, 2004. 16. Klein P, Rubinstein LJ: Benign symptomatic glial cysts of the pineal gland: A report of seven cases and review of the literature. J Neurol Neurosurg Psychiatry 52:991–995, 1989. 17. Konovalov AN, Pitskhelauri DI: Infratentorial supracerebellar approach to the colloid cysts of the third ventricle. Neurosurgery 49:1116–1123, 2001. 18. Kurosaki K, Hayashi N, Hamada H, Hori E, Kurimoto M, Endo S: Multiple epidermoid cysts located in the pineal and extracranial regions treated by neuroendoscopy. Neurol Med Chir (Tokyo) 45:216–219, 2005. 19. Leslie K, Hui R, Kaye AH: Venous air embolism and the sitting position: A case series. J Clin Neurosci 13:419–422, 2006. 20. Mammoto T, Hayashi Y, Ohnishi Y, Kuro M: Incidence of venous and paradoxical air embolism in neurosurgical patients in the sitting position: Detection by transesophageal echocardiography. Acta Anaesthesiol Scand 42:643–647, 1998. 21. Mamourian AC, Towfighi J: Pineal cysts: MR imaging. AJNR Am J Neuroradiol 7:1081–1086, 1986. 22. Mandera M, Marcol W, Bierzynska-Macyszyn G, Kluczewska E: Pineal cysts in childhood. Childs Nerv Syst 19:750–755, 2003. 23. McNeely PD, Howes WJ, Mehta V: Pineal apoplexy: Is it a facilitator for the development of pineal cysts? Can J Neurol Sci 30:67–71, 2003. 24. Michielsen G, Benoit Y, Baert E, Meire F, Caemaert J: Symptomatic pineal cysts: Clinical manifestations and management. Acta Neurochir (Wien) 144:233–242, 2002. 25. Mukherjee KK, Banerji D, Sharma R: Pineal cyst presenting with intracystic and subarachnoid haemorrhage: Report of a case and review of the literature. Br J Neurosurg 13:189–192, 1999. 26. Oi S, Shibata M, Tominaga J, Honda Y, Shinoda M, Takei F, Tsugane R, Matsuzawa K, Sato O: Efficacy of neuroendoscopic procedures in minimally invasive preferential management of pineal region tumors: A prospective study. J Neurosurg 93:245–253, 2000. 27. Papadopoulos G, Kuhly P, Brock M, Rudolph KH, Link J, Eyrich K: Venous and paradoxical air embolism in the sitting position. A prospective study with transoesophageal echocardiography. Acta Neurochir (Wien) 126:140– 143, 1994. 28. Pople IK, Athanasiou TC, Sandeman DR, Coakham HB: The role of endoscopic biopsy and third ventriculostomy in the management of pineal region tumours. Br J Neurosurg 15:305–311, 2001. 29. Ruge JR, Johnson RF, Bauer J: Burr hole neuroendoscopic fenestration of quadrigeminal cistern arachnoid cyst: Technical case report. Neurosurgery 38:830–837, 1996. 30. Sawamura Y, Ikeda J, Ozawa M, Minoshima Y, Saito H, Abe H: Magnetic resonance images reveal a high incidence of asymptomatic pineal cysts in young women. Neurosurgery 37:11–16, 1995. 31. Stein BM: The infratentorial supracerebellar approach to pineal lesions. J Neurosurg 35:197–202, 1971. 32. Stern JD, Ross DA: Stereotactic management of benign pineal region cysts: Report of two cases. Neurosurgery 32:310–314, 1993. 33. Tamaki N, Shirataki K, Lin TK, Masumura M, Katayama S, Matsumoto S: Cysts of the pineal gland. A new clinical entity to be distinguished from tumors of the pineal region. Childs Nerv Syst 5:172–176, 1989. 34. Tartara F, Regolo P, Terreni MR, Giovanelli M: Glial cyst of the pineal gland: Case report and considerations about surgical management. J Neurosurg Sci 44:89–93, 2000. 35. Tirakotai W, Schulte DM, Bauer BL, Bertalanffy H, Hellwig D: Neuroendoscopic surgery of intracranial cysts in adults. Childs Nerv Syst 20:842–851, 2004. 36. Turtz AR, Hughes WB, Goldman HW: Endoscopic treatment of a symptomatic pineal cyst: Technical case report. Neurosurgery 37:1013–1015, 1995. 37. Wisoff JH, Epstein F: Surgical management of symptomatic pineal cysts. J Neurosurg 77:896–900, 1992.


GORE ET AL.

COMMENTS

I

n this original article, Gore et al. present us with the first reported case of an entirely endoscopic infratentorial supracerebellar resection of a pineal cyst. In the past, Ruge et al. (1) reported the case of an arachnoid cyst treated by endoscopic fenestration through the same surgical corridor, but no case of a complete resection performed in such a manner has been presented in the international literature so far. We believe their experience is extremely valid and opens the way for further procedures. In this technical note, the authors were confronted with a rather small cyst, exerting a moderate mass effect but not provoking hydrocephalus in a patient complaining of headache. An accurate examination of the preexisting literature and a study of possible physiopathological causes of the symptom allowed the authors to correctly pose an indication for surgery that would otherwise have been controversial. A well-planned surgical maneuver was performed, allowing complete removal of the lesion and minimal invasiveness through a “bimanual” endoscopic technique. However, it is our opinion that a minicraniectomy performed by enlargement of the burr hole by just a few millimeters would not have significantly influenced the invasiveness and the patient’s tolerance of the procedure, permitting bimanual techniques with micro instruments and better management of possible intraoperative complications. Finally, we want to compliment the authors for the innovative and scrupulous work they have performed on the road to reducing invasiveness of neurosurgical procedures. Michelangelo Gangemi Paolo Cappabianca Naples, Italy

1. Ruge J, Johnson RF, Bauer J: Burr hole neuroendoscopic fenestration of quadrigeminal cistern arachnoid cyst: Technical case report. Neurosurgery 38:830–837, 1996.

G

ore et al. describe a purely endoscopic method for resection of a pineal cyst. They highlight important technical features of the procedure including stereotactic guidance for localizing the transverse sinus, the sitting position to facilitate cerebellar relaxation and working in an air medium, and the use of a “bimanual” endoscopic technique. We have used a similar technique for fenestration of quadrigeminal plate arachnoid cysts and have not been impressed with the purported benefits over conventional microsurgery. The endoscopic technique in our experience used a cranial opening similar in size to that of a microsurgical technique, whereas the actual cyst fenestration or resection is more tedious. This debate is reminiscent of the ongoing discussion regarding endoscopic fenestration of middle fossa arachnoid cysts. As a cautionary comment, with the reported reduction in morbidity associated with “minimally invasive” techniques, one needs to remain disciplined in patient selection. Symptoms of headache ascribed to the presence of relatively small pineal cysts without hydrocephalus remains dubious. However, the complete cessation of headaches during the postoperative interval supports the authors’ recommendation. Although it is likely that any procedure that positively affects the risk-benefit analysis may expand surgical indications, the endoscopic surgeon will need to maintain vigilant in patient selection. Last, we respect the authors’ willingness to explore innovative applications of endoscopic surgery in the hopes of creating a greater demand for equipment modification. Jeffrey P. Greenfield Mark M. Souweidane New York, New York


T

his case report is unique in that it describes a purely endoscopic supracerebellar infratentorial approach to resect a pineal cyst. Gore et al. are to be congratulated for a minimally invasive “tour de force.” Several observations come to mind in reviewing this excellent article. The true incidence of pineal cysts in a population is not known, simply because many subjects are probably asymptomatic. Those that come to medical attention often do so for evaluation of headaches or an unrelated issue, such as a head injury. The patient with headaches and a magnetic resonance imaging (MRI) scan consistent with a pineal cyst, without hydrocephalus, becomes a diagnostic dilemma. In my practice, few, if any, patients ever reach the operating room unless there is a question about the pathological condition in the face of growth. Although I remain skeptical about the hypothesis of intermittent aqueduct obstruction in such patients, each surgeon must make a judgment about this sort of complex clinical situation. However, the endoscopic approach may be an excellent alternative to a transventricular approach or an open approach to a pineal tumor in a patient when a biopsy is warranted. The technical issues that the authors have tested and improved upon are the portion of the article that I found most helpful. The selection of the scope, position of patient, and treatment of the veins made the article a useful addition to our repertoire of endoscopic approaches to deep structures. Richard G. Ellenbogen Seattle, Washington

G

ore et al. describe a case report of a patient whose pineal cyst was removed through a supracerebellar infratentorial endoscopic approach. The patient presented with headaches but did not have hydrocephalus. Despite the favorable outcome in this patient, I would strongly urge caution in treating patients with “symptomatic” pineal cysts. Patients are undergoing MRI scans with increasing frequency for nonspecific neurological complaints, such as headache or balance problems. Given the high prevalence of pineal cysts in the normal population, it is likely that many of the patients with these nonspecific complaints will have pineal cysts. In my experience, however, it is extremely rare to encounter a symptomatic cyst without radiographic evidence of hydrocephalus or at least aqueductal compression (best seen on a thincut sagittal MRI scan). Without aqueductal compression, I would recommend extreme caution before advocating surgery in these patients. Conservative management with follow-up MRI scans is preferable. The authors provide a detailed description of their minimally invasive approach with some points deserving emphasis. A cyst is considerably easier to remove than a vascular, solid tumor attached to surrounding structures that would present a much greater challenge with an endoscopic approach. Endoscopically, even a small amount of bleeding can be difficult to control while one is working within the quadrigeminal cistern or third ventricle where there is no soft tissue pressure to help tamponade minor bleeding. Bleeding potentially can occur from small vascular branches over the dorsal surface of the cyst that ultimately form part of the posterior choroidal arteries. In addition, any tears of bridging veins between the cerebellum and tentorium can be problematic. Last, pineal region anatomy can be confusing unless the surgeon is well versed with the surrounding structures because the initial endoscopic trajectory is directed toward the vein of Galen and must therefore be adjusted inferiorly. The quadrigeminal arachnoid is thick, and cisternal vessels are vulnerable while one is attempting to open and dissect the arachnoid. These caveats notwithstanding, the endoscopic approach is provocative, and it will be instructive to see whether a larger series will replicate this excellent outcome. Jeffrey N. Bruce New York, New York


VASCULAR Surgical Anatomy and Technique

TENTORIAL DURAL ARTERIOVENOUS FISTULAE: OPERATIVE STRATEGIES AND MICROSURGICAL RESULTS FOR SIX TYPES Michael T. Lawton, M.D. Department of Neurological Surgery, University of California, San Francisco, San Francisco, California

Rene O. Sanchez-Mejia, M.D. Department of Neurological Surgery, University of California, San Francisco, San Francisco, California

Diep Pham, B.A. Department of Neurological Surgery, University of California, San Francisco, San Francisco, California

Jeffrey Tan, B.A. Department of Neurological Surgery, University of California, San Francisco, San Francisco, California

Van V. Halbach, M.D. Department of Interventional Neuroradiology, University of California, San Francisco San Francisco, California Reprint requests: Michael T. Lawton, M.D., 505 Parnassus Avenue, M-780C, San Francisco, CA 94143–0112. Email: [email protected] Received, March 29, 2007.

OBJECTIVE: Tentorial dural arteriovenous fistulae (DAVF) are rare, have a high risk of hemorrhage, often cannot be obliterated endovascularly, and frequently require microsurgical interruption of the draining vein. We differentiated these fistulae into six types and developed specific operative strategies on the basis of these types. METHODS: During a 9-year period, 31 patients underwent microsurgical treatment for tentorial fistulae: seven galenic DAVF, eight straight sinus DAVF, three torcular DAVF, three tentorial sinus DAVF, eight superior petrosal sinus DAVF, and two incisural DAVF. RESULTS: The posterior interhemispheric approach was used with galenic DAVF; the supracerebellar-infratentorial approach was used with straight sinus DAVF; a torcular craniotomy was used with torcular DAVF; the supratentorial-infraoccipital approach was used with tentorial sinus DAVF; the extended retrosigmoid approach was used with superior petrosal sinus DAVF; and a pterional or subtemporal approach was used with incisural DAVF. Angiographically, 94% of the fistulae were obliterated completely. Four patients had transient neurological morbidity, none had permanent neurological morbidity; and there was no operative mortality (mean follow-up, 4.2 yr). CONCLUSION: Tentorial DAVF can be differentiated on the basis of fistula location, dural base, associated sinus, and direction of venous drainage. The operative strategy for each type is almost algorithmic, with each type having an optimum surgical approach and an optimum patient position that allows gravity to retract the brain, open subarachnoid planes, and shorten dissection times. No matter the type, the fistula is treated microsurgically by simple interruption of the draining vein. KEY WORDS: Arteriovenous malformation, Dural arteriovenous fistula, Microsurgery, Operative approaches, Tentorium Neurosurgery 62:ONS110-ONS125, 2008

DOI: 10.1227/01.NEU.0000297027.98243.21

Accepted, July 31, 2007.

T

entorial dural arteriovenous fistulae (DAVF) are rare and dangerous lesions (2, 3, 5, 14, 15, 24, 31). In a meta-analysis of 377 patients with DAVF reported before 1989, tentorial DAVF were less common than transverse-sigmoid sinus and cavernous sinus DAVF (8 versus 63 and 12%, respectively), but tentorial DAVF had the most aggressive neurological behavior, with 97% causing hemorrhage or progressive focal neurological deficits (2). Tentorial DAVF frequently have angiographic features associated with hemorrhage: retrograde drainage through cortical or subarachnoid veins, deep drainage through the vein of Galen, and venous varices. Consequently, tentorial DAVF are treated aggressively when diagnosed, even in the absence of presenting hemorrhage (24). Endovascular therapy has become the predominant therapy for intracranial DAVF because their arterial supply from the


external carotid artery (ECA) can be embolized safely, and their location on dural venous sinuses facilitates access and occlusion through that sinus (1, 9, 10, 24, 26, 31). The combination of transarterial and transvenous embolization results in high obliteration rates for most DAVF, but tentorial DAVF are an exception. Their arterial supply is extensive, involving meningeal arteries from the internal carotid artery (ICA) and vertebral artery that are difficult to cannulate and riskier to embolize than ECA feeders. Transvenous navigation to deeper locations around the tentorium is difficult. More importantly, tentorial DAVF often drain exclusively to subarachnoid veins rather than to their associated sinus (Borden Type III), which prevents transvenous access (3). Therefore, the management of tentorial DAVF may require microsurgical interruption, unlike most other DAVF (6, 8, 10, 12–14, 23, 26, 30).


TENTORIAL DURAL ARTERIOVENOUS FISTULAE

Operative strategies for A B tentorial DAVF require a thorough understanding of subtle anatomic differences between various types of tentorial DAVF. Published surgical reports are limited to small series that typically include all tentorial-based fistulae and unfortunately confuse the subtypes (10, 14, 15, 24, 31). We have accumulated extensive experience with this lesion. We separate tentorial DAVF into six subtypes, each one having unique pathological anatomy and an optimal surgical approach. FIGURE 1. Types of tentorial dural arteriovenous fistulae (DAVF). A, axial view. B, lateral view. T h e re a re m a n y w a y s t o approach tentorial DAVF, with different craniotomies, surgical trajectories, and positions with venous drainage that can be supratentorial, infratentorial, or both (Fig. 2). Straight sinus DAVF (Type 2) were located in the midline along relative to the tentorium. We present our microsurgical the falcotentorial junction, associated with the straight sinus, with techniques, results, and insights from a consecutive, singledrainage to veins on the undersurface of the tentorium. Torcular DAVF surgeon series of 31 patients.

PATIENTS AND METHODS Types of Tentorial DAVF DAVF located on the tentorium, from their attachment to the clinoid processes and petrous ridges anteriorly to the torcula posteriorly, were included in this analysis. Although transverse-sigmoid sinus DAVF are near the tentorium, they are based on lateral dura overlying cerebellum and temporal lobe, and were excluded. Tentorial DAVF were categorized into six types based on anatomic location, dural base, associated venous sinus, and direction of venous drainage (Fig. 1; Table 1). Galenic DAVF (Type 1) were located in the midline at the posterior margin of the tentorial incisura, associated with the vein of Galen as it enters the anterior falcotentorial junction,

(Type 3) were located in the midline at the posterior margin of the falcotentorial junction, associated with the torcula, with supratentorial venous drainage. Tentorial sinus DAVF (Type 4) were located in the body of the tentorium, associated with the tentorial sinus, with supratentorial drainage to occipital veins. Superior petrosal sinus DAVF (Type 5) were located laterally where the tentorium joins the dura of the middle cranial fossa, associated with the superior petrosal sinus, with infratentorial drainage to the petrosal vein and its tributaries. Incisural DAVF (Type 6) were located along the free edge of tentorium, not clearly associated with a venous sinus, with drainage into supratentorial veins in and around the ambient cistern.

Patients The study was approved by the institutional review board and conducted in compliance with Health Insurance Portability and

TABLE 1. Types of tentorial dural arteriovenous fistulae Dural arteriovenous fistulae

Type

Patients, no. (percentage)

Galenic

1

7 (23%)

Midline

Anterior falcotentorial junction

Vein of Galen

Supra- and infratentorial

Straight sinus

2

8 (26%)

Midline

Middle falcotentorial junction

Straight sinus

Infratentorial

Torcular

3

3 (10%)

Midline

Posterior falcotentorial junction

Torcula

Supratentorial

Tentorial sinus

4

3 (10%)

Paramedian

Tentorium

Tentorial sinus

Supratentorial

Superior petrosal sinus

5

8 (26%)

Lateral

Petrotentorial junction


Infratentorial

Incisural

6

2 (6%)

Paramedian

Tentorial incisura

None

Supratentorial

Total

NEUROSURGERY

Location

Dural base

Venous sinus

Venous drainage

31 (100%)


LAWTON ET AL.

FIGURE 2. Anatomy of tentorial DAVF by type. A, galenic DAVF (Type 1). B, straight sinus DAVF (Type 2). C, torcular DAVF (Type 3). D, tentorial sinus DAVF (Type 4). E, superior sagittal sinus DAVF (Type 5). F, incisural DAVF (Type 6). Insets show the coronal sections through the DAVF. a, artery; PCA, posterior cerebral artery; SCA, superior cerebellar artery; BA,


basilar artery; ADS, artery of Davidoff and Schecter; BVR, basal vein of Rosenthal; ICV, internal cerebral vein; PCV, precentral cerebellar vein. R, right; L, left; PMA, posterior meningeal artery; MMA, middle meningeal artery; ECA, external carotid artery.



Accountability Act regulations. A database of patient information was maintained prospectively, and clinical data were reviewed retrospectively, including medical records, radiographic studies, operative reports, intraoperative photographs, and records from office visits. During a 9-year period between August 1997 and August 2006, 68 patients had intracranial DAVF that required microsurgical treatment by the senior author (MTL), of which 31 had tentorial DAVF (Table 2). The mean patient age was 53 years (range, 25–87 years). There was a strong male predominance, with 21 men and 10 women.

DAVF Characteristics According to our categorization of tentorial DAVF, seven patients (23%) had galenic DAVF, eight (26%) patients had straight sinus DAVF, three patients (10%) had torcular DAVF, three patients (10%) had tentorial sinus DAVF, eight patients (26%) had superior petrosal sinus DAVF, and two patients (6%) had incisural DAVF. Arterial supply to tentorial DAVF originates from six different sources: tentorial artery, middle meningeal artery (MMA), posterior meningeal artery (PMA), meningeal branches from pial arteries, scalp arteries, and miscellaneous ECA branches. First, the tentorial artery, or artery of Bernasconi and Cassinari, is a branch of the meningohypophyseal trunk or, infrequently, the inferolateral trunk, both arising from the cavernous ICA (25). Tentorial artery courses posteriorly along the tentorial incisura to reach the fistula. Second, the middle meningeal artery originates from the internal maxillary artery, passes through the foramen spinosum, and reaches the tentorium directly or can course over the convexity to the midline, where it travels down the falx as a falcine artery to supply falcotentorial fistulae. Third, the posterior meningeal artery originates from the vertebral artery as it pierces the dura at the foramen magnum, courses superiorly along suboccipital dura to the tentorium, and supplies the fistula from behind. Fourth, meningeal branches from pial arteries can supply tentorial DAVF. Tentorial branches from the superior cerebellar artery and posterior cerebral artery, also known as the arteries of Davidoff and Schecter (28, 29), travel through the ambient cistern to the Galenic region and are visualized on vertebral artery angiograms. Similarly, meningeal branches from anteroinferior cerebellar artery and posteroinferior cerebellar artery can travel through the cerebellopontine angle to reach laterally located DAVF. Fifth, occipital artery and occasionally superficial temporal artery supply tentorial DAVF through transosseus perforators that connect to the tentorium. Lastly, a lateral group of miscellaneous arteries from ECA supply tentorial DAVF, including the ascending pharyngeal artery, stylomastoid artery, posterior auricular artery, and the artery of the foramen rotundum. Arterial supply to tentorial DAVF was characterized by analyzing contributions from these six vascular groups.

Treatment and Outcomes Patients were selected for treatment on the basis of clinical presentation with hemorrhage or anatomic features associated with high risk of hemorrhage. Neurological assessments were performed by a nurse clinician under the supervision of a neurologist, preoperatively, postoperatively, and during the follow-up period. The Modified Rankin Scale (MRS) was used to grade outcomes (27). Outcomes were analyzed in terms of MRS on latest follow-up and change in MRS from preoperative to final follow-up MRS. Good outcomes were defined as a final MRS of 0 to 2, and poor outcome as final MRS greater than 2. Improvement was defined as a change MRS of less than or equal to 0 (improved or unchanged), and deterioration was defined as a change in MRS of greater than 0 (deteriorated or dead).

NEUROSURGERY

RESULTS Presentation Seventeen patients (55%) presented with intracranial hemorrhage. Four of these patients presented with obtundation or in coma. The hemorrhage was in the subarachnoid space in 6 patients, in brain parenchyma in 10 patients, and in both locations in 1 patient. Hemorrhagic presentation did not correlate with type of tentorial DAVF. Only two patients (6%) had asymptomatic DAVF that were diagnosed incidentally. The remaining patients had a variety of symptoms, including headache, dizziness, nausea/vomiting, gait instability, pulsatile tinnitus, dysarthria, and myelopathy. Two patients with galenic DAVF had hydrocephalus resulting from compression of the aquaduct of Sylvius by venous varices. Two patients with superior petrosal sinus DAVF presented with pulsatile tinnitus because of the proximity of the fistulae to the auditory system. Venous hypertension, with diffuse cortical venous drainage of the fistula and measured elevations in transverse sinus pressures greater than 40 mmHg, accounted for ischemic symptoms and radiographic finding on brain magnetic resonance imaging in two patients.

Angiographic Anatomy Selective angiography was valuable in defining the anatomy of tentorial DAVF. Arterial supply varied with fistula type (Table 3). Supply from the artery of Davidoff and Schecter was almost pathognomonic of galenic DAVF, participating in all but one of these fistulae (86%) and in only two nongalenic DAVF (one straight sinus and one torcular DAVF). Galenic DAVF, with their deep, central location, were supplied by most vascular groups and from all directions: inferiorly from the superior cerebral artery/posterior cerebral artery, anteriorly from the tentorial artery (57%), superiorly from the MMA/falcine artery (43%), and posteriorly from the PMA (57%) and occipital artery (29%). Straight sinus DAVF were fed primarily by PMA (100%), occipital artery (75%), and tentorial artery (63%). Torcular DAVF attracted robust arterial supply from PMA and occipital artery. All tentorial sinus DAVF were supplied by MMA and scalp arteries. The tentorial artery made its most significant contribution to superior petrosal sinus DAVF, supplying 75% of these lesions. Anteroinferior cerebellar artery did not contributed to these fistulae to the degree that SCA contributed to Galenic fistulae, supplying just one of these eight lesions (13%). Incisural DAVF were supplied by more anterior arteries: MMA, STA, and tentorial arteries. Venous drainage from tentorial DAVF was determined by the patency of the associated venous sinus. According to the Borden classification (3, 7), 26 patients (84%) had Type III DAVF, with venous drainage into subarachnoid veins in a retrograde direction. The other five patients (16%) had Type II DAVF, of which four were superior petrosal sinus DAVF with some shunting into the sinus. The exact anatomy of the vein draining tentorial DAVF was not as critical to surgical planning as the side of the tentorium from which that vein exits. Straight sinus and superior pet-


LAWTON ET AL.

TABLE 2. Summary of patientsa Modified Rankin scale score PreopLate erative

Patient no.

Age/ Sex

Type

1

53/F

1

Hemorrhage

M

III

TA

PIA

Prone

Obliterated

3

1

2

56/M

1

Hydrocephalus

M

III

TA

PIA

Prone

Obliterated

2

1

3

67/F

1

Hemorrhage

M

III

TA

SC-IT; PIA

Prone; Lateral

Obliterated

1

0

4

25/F

1

Hemorrhage

M

III

TA

PIA

Lateral

Obliterated

1

1

5

37/F

1

Hydrocephalus

M

III

TA

PIA

Lateral

Obliterated

1

0

6

52/M

1

Vertigo

M

III

TA

PIA

Lateral

Obliterated

1

0

7

73/F

1

Venous hypertension

M

III

TA

PIA

Lateral

Obliterated

2

0

8

59/M

2

Hemorrhage

M

III

TA

SC-IT

Prone

Obliterated

1

0

Presentation

Side

Borden grade

Embolization

Approach

Position

Angiographic outcome

9

45/M

2

Hemorrhage

M

III

TA

SC-IT

Prone

Obliterated

4

1

10

57/M

2

Hemorrhage

M

III

TA

SC-IT

Prone

Obliterated

1

0

11

55/M

2

Incidental

M

III

TA

SC-IT

Sitting

Obliterated

0

0

12

36/M

2

Hemorrhage

M

III

TA

SC-IT

Sitting

Obliterated

2

0

13

73/F

2

Hemorrhage

M

III

TA

SC-IT

Sitting

Obliterated

3

2

14

70/M

2

Trigeminal neuralgia

M

III

TA

SC-IT

Sitting

Obliterated

1

1

15

68/M

2

Headache

M

III

TA

SC-IT

Sitting

Obliterated

0

0

16

36/M

3

Tinnitus, headache

M

III

TA

Torcular/ ExRS

Lateral

Obliterated

0

0

17

69/M

3

Hemorrhage

M

III

TA

Torcular

Prone

Obliterated

1

0

18

78/F

3

Venous hypertension

M

II

TA

Torcular/ ExRS; SC-IT

Lateral; sitting

Residual

2

0

19

47/M

4

Hemorrhage

L

III

TA

ST-IO

Prone

Obliterated

1

0

20

56/M

4

Hemorrhage

M

III

TA

ST-IO

Prone

Obliterated

4

2

21

71/M

4

Hemorrhage

L

III

TA

ST-IO

Prone

Obliterated

1

1

22

49/M

5

Hemorrhage

L

III

TA

ExRS

Lateral

Obliterated

1

0

23

59/M

5

Hemorrhage

L

III

TA

ExRS

Lateral

Obliterated

5

2

24

49/M

5

Hemorrhage

R

II

TA

ExRS

Lateral

Residual

1

1

25

76/M

5

Hemorrhage

L

II

TA

ExRS

Lateral

Obliterated

1

2

26

72/M

5

Hemorrhage

R

III

TA

ExRS

Lateral

Obliterated

2

0

27

49/F

5

Tinnitus, headache

L

II

TA

ExRS

Lateral

Obliterated

0

0

28

71/F

5

Incidental

L

III

TA

ExRS

Lateral

Obliterated

0

0

29

67/M

5

Tinnitus, headache

L

II

TA

ExRS

Lateral

Obliterated

1

0

30

87/F

6

Hemiparesis

R

III

TA

ExRS; OZ

Lateral; supine

Obliterated

4

2

31

56/M

6

Seizure

L

III

TA

Subtemporal

Supine

Obliterated

2

1

a

M, midline; TA, transarterial; PIA, posterior interhemispheric approach; SC-IT, supracerebellar-infratentorial approach; ExRS, extended retrosigmoid approach; L, left; ST-IO, supratentorial-infraoccipital approach; R, right; OZ, orbitozygomatic-pterional craniotomy.




TABLE 3. Summary of angiographic anatomy Galenic

Straight sinus

Torcular

Tentorial sinus


Incisural

Total

7 (100%)

8 (100%)

3 (100%)

3 (100%)

8 (100%)

2 (100%)

31 (100%)

Tentorial artery

4 (57%)

5 (63%)

2 (67%)

1 (33%)

6 (75%)

1 (50%)

18 (58%)

Middle meningeal artery

3 (43%)

3 (38%)

2 (67%)

3 (100%)

4 (50%)

2 (100%)

16 (52%)

Posterior meningeal artery

4 (57%)

8 (100%)

3 (100%)

2 (67%)

1 (13%)

0 (0%)

18 (58%)

Superior cerebellar or posterior cerebral artery branches

6 (86%)

1 (13%)

1 (33%)

0 (0%)

0 (0%)

0 (0%)

8 (26%)

Posteroinferior or anteroinferior cerebellar artery branches

0 (0%)

2 (25%)

0 (0%)

0 (0%)

1 (13%)

0 (0%)

3 (10%)

Scalp arteries

2 (29%)

6 (75%)

3 (100%)

3 (100%)

2 (25%)

2 (100%)

17 (55%)

External carotid artery branches

1 (14%)

1 (13%)

2 (67%)

0 (0%)

3 (38%)

1 (50%)

8 (26%)

Vein of Galen

3 (43%)

1 (13%)

0 (0%)

0 (0%)

0(0%)

0 (0%)

5 (16%)

Basal vein of Rosenthal

2 (29%)

0 (0%)

0 (0%)

0 (0%)

2 (25%)

0 (0%)

4 (13%)

Petrosal vein

0 (0%)

0 (0%)

0 (0%)

0 (0%)

6 (75%)

0 (0%)

Cortical veins

3 (43%)

0 (0%)

3 (100%)

3 (100%)

0(0%)

2 (100%)

10 (32%)

Cerebellar veins

3 (43%)

7 (88%)

0 (0%)

0 (0%)

1 (13%)

0 (0%)

10 (32%)

Mesencephalic veins

3 (43%)

0 (0%)

0 (0%)

0 (0%)

1 (13%)

0 (0%)

4 (13%)

I

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

II

0 (0%)

0 (0%)

1 (33%)

0 (0%)

4 (50%)

0 (0%)

5 (16%)

III

7 (100%)

8 (100%)

2 (67%)

3 (100%)

4 (50%)

2 (100%)

No. of patients (%) Arterial supply

Venous drainage

6 (19%)

Borden classification

rosal sinus DAVF drained infratentorially. Straight sinus DAVF drained to vermian or superior cerebellar veins, and superior petrosal sinus DAVF drained to the petrosal vein. In contrast, torcular, tentorial sinus, and incisural DAVF drained supratentorially. Torcular DAVF drained to medial and inferior occipital veins along the sagittal and transverse sinuses. Tentorial sinus DAVF drained to infraoccipital veins, while incisural DAVF drained to medial and inferior temporal veins or veins in the ambient cistern. Galenic DAVF drained supratentorially, infratentorially, or in some cases both. The straight sinus was occluded with most of these fistulae, redirecting venous drainage retrograde into the vein of Galen or one of its tributaries (basal vein of Rosenthal or precentral cerebellar vein, but rarely internal cerebral vein). All DAVF were embolized preoperatively from the transarterial route as a presurgical adjunct. Embolic agents included polyvinyl alcohol particles in 67%, coils in 33%, ethanol in 22%, glue in 17%, and combinations of agents in 39%.

Surgical Management Surgical approaches are summarized in Table 4, with twostage procedures required in three patients. The posterior interhemispheric approach (4) with a torcular craniotomy was used for galenic DAVF (Fig. 3). One patient had a supracerebellar-

NEUROSURGERY

26 (84%)

infratentorial approach for this fistula, and the exposure made it difficult to visualize the anterior portion of the fistula. Residual DAVF remained, and a posterior interhemispheric approach was performed as a second stage to occlude the remaining DAVF. Patient positioning for the posterior interhemispheric approach changed from prone early in the series to lateral later in the series, with improved exposure resulting from gravity retraction of the occipital lobe. The supracerebellar-infratentorial approach was used for all straight sinus DAVF (Fig. 4). Early in the series, this procedures was performed with the patient in the prone position (three patients), but this plane was difficult to open in two cases with ruptured DAVF and cerebellar swelling. In one of these cases, the draining vein reruptured with gentle retraction. Subsequently, the supracerebellar-infratentorial approach was performed with patients in the sitting position to allow gravity to retract the cerebellum and open this plane of dissection. Torcular craniotomies were used for torcular DAVF (Fig. 5). Two patients with torcular DAVF also had transverse-sigmoid sinus DAVF, which required that the craniotomy extend laterally to incorporate an extended retrosigmoid approach. In one patient with a Borden Type II torcular DAVF, the sinuses around the torcula were skeletonized and the postoperative


LAWTON ET AL.

TABLE 4. Summary of surgical approaches Galenic/ Type 1

Straight sinus/Type 2

Torcular/ Type 3

Tentorial sinus/Type 4

Superior petrosal sinus/Type 5

Incisural/ Type 6

Total

7

8

3

3

8

2

31

Posterior interhemispheric

7

0

0

0

0

0

7

Supracerebellar-infratentorial

1

8

1

0

0

0

10

Torcular

0

0

3

0

0

0

3

Supratentorial-infraoccipital

0

0

0

3

0

0

3

Extended retrosigmoid

0

0

0

0

8

1

9

Subtemporal or pterional

0

0

0

0

0

2

2

8

8

4

3

8

3

34

Patients, no. Approach

Total

A

E

B

C

F

FIGURE 3. Galenic DAVF (Type 1, Patient 7). A, axial brain magnetic resonance imaging (MRI) (fluid-attenuated inversion recovery [FLAIR] sequence) scan showing increased signal in the left thalamus for retrograde venous drainage from the fistula to the vein of Galen, left basal vein of Rosenthal, and left internal cerebral vein. B, right internal carotid artery angiogram (anteroposterior view) demonstrating arterial supply to the fistula (red asterisk) from the right tentorial artery (dotted arrow) and middle meningeal/falcine artery (solid arrow). Left vertebral artery angiograms (anteroposterior [C], and lateral [D] views) demonstrating arterial supply to the fistula (red asterisk) from the artery of Davidoff and Schecter (solid arrows), and drainage into the vein of Galen and left basal vein of Rosenthal (dotted arrow). E, intraoperative photograph demonstrating the


D

G

posterior interhemispheric approach, with the patient positioned laterally (left side down), gravity retraction of the left occipital lobe, and transection of the left tentorium and falx (at tips of bipolar forceps) to widen the exposure. F, reflection of the dura at the falcotentorial junction with the bipolar forceps visualized the fistula (black asterisk) and the vein of Galen complex (right basal vein of Rosenthal, solid arrow; right internal cerebral vein, dotted arrow; and artery of Davidoff and Schecter, dashed arrow). G, the straight sinus was already occluded, so the fistula (red asterisk) was interrupted with a clip placed on the vein of Galen as it exited the fistula.



A

C

B

A

D

C

B

D

E E

F

F

FIGURE 4. Straight sinus DAVF (Type 2, Patient 15). Left vertebral artery angiograms (anteroposterior [A] and lateral [B] views) demonstrating arterial supply to the fistula (red asterisk) from the posterior meningeal artery (solid arrows). The draining vein coursed anteriorly over the superior surface of the cerebellum, under the apex of the tentorium (not shown). C, intraoperative photograph demonstrating the supracerebellar-infratentorial approach, with the patient in the sitting position, torcular dura pulled superiorly with tacking sutures, and gravity retraction of the cerebellum to open the infratentorial plane. The arterialized vein draining the fistula (black asterisk) was seen exiting the dura (D) and coursing anteriorly to the galenic region (E). F, the fistula was interrupted with a clip on this vein as it exited the tentorial dura.

FIGURE 5. Torcular DAVF (Type 3, Patient 18). Left external carotid artery angiogram (anteroposterior view) (A) and left common carotid artery angiogram (lateral view) (B) demonstrating arterial supply to the fistula (red asterisk) from occipital artery branches and venous drainage into the torcular herophili, tranverse sinuses, retrograde into the superior sagittal sinus. C, venous phase of the left common carotid artery angiogram (lateral view) showed delayed venous outflow, marked venous engorgement, and poor filling of the superior sagittal sinus, consistent with venous hypertension. Computed tomographic angiography reformatted axial (D) and reformatted coronal (E) views demonstrating this abnormal venous anatomy. F, the mean transit time of contrast on computed tomography angiography was markedly increased in the right occipital pole adjacent to the torcular herophili dural arteriovenous fistula. Retrograde venous drainage in cortical veins along the superior sagittal and transverse sinuses was interrupted and the torcular herophili was skeletonized.

angiogram revealed residual DAVF supplied by SCA branches. A supracerebellar-infratentorial approach in the sitting position was used to expose the residual DAVF in a second operation. The supratentorial-infraoccipital approach was used for tentorial sinus DAVF (Fig. 6) (22). This approach exposed the supe-

rior surface of the tentorium between the straight sinus and the superior petrosal sinus, with flexibility in the angle of approach to reach any of these DAVF. One of the DAVF was just off the midline, one was centrally located on the tentorium, and one was laterally located. In all three cases, this approach provided excellent exposure.

NEUROSURGERY


LAWTON ET AL.

A

B

A

C

D

C

FIGURE 6. Tentorial E sinus DAVF (Type 4, Patient 20). Right external carotid artery angiograms (anteroposterior [A] and lateral [B] views) demonstrating arterial supply to the fistula (red asterisk) from the occipital artery (solid arrows) and middle meningeal artery branches (dotted arrows). The draining vein coursed along the undersurface of the occipital lobe to the vein of Galen (dashed arrows). C, intraoperative photograph demonstrating the supratentorial-infraoccipital approach, with the patient positioned laterally (left side down), the table rotated to position the head nearly prone, dura containing the transverse sinus pulled superiorly with tacking sutures, and gravity retraction of the occipital lobe to open the infraoccipital plane. The fistula (black asterisk) and its draining vein were seen under the occipital lobe. The draining vein is easily coagulated and cut (D) and feeding arteries in the tentorium were observed (at tip of sucker) (E).

The extended retrosigmoid approach (21) was used with superior petrosal sinus DAVF (Fig. 7). All patients were positioned laterally, with drilling of the bone overlying the sigmoid and transverse sinuses, opening a small craniotomy in the lateral suboccipital bone, and gently mobilizing the sigmoid sinus anteriorly. Incisural DAVF were exposed either with a pterional approach or a subtemporal approach (Fig. 8). The pterional, transsylvian approach exposed the anterior incisura in one case, whereas the subtemporal approach exposed the posterior incisura in another case, behind the lateral edge of the cerebral peduncle. Incisural DAVF can lie near the medial superior petrosal sinus and be misinterpreted as superior petrosal sinus DAVF. The extended retrosigmoid approach was inadequate to expose the draining vein of an incisural DAVF,


B

E

D

F

FIGURE 7. Superior petrosal sinus DAVF (Type 5, Patient 23). A, brain MRI (axial view, T2-weight image) scan demonstrating a hematoma in the left cerebellar peduncle, surrounding edema, and dilated veins in the cerebellopontine angle. B, left external carotid artery angiogram (lateral view) showing arterial supply to the fistula (red asterisk) from the middle meningeal artery and transosseus perforators. C, the venous phase of the angiogram (anteroposterior view) showed the fistula’s drainage through tortuous and variceal cerebellar veins. D, left internal carotid artery angiogram (lateral view) showing arterial supply to the fistula (red asterisk) from the tentorial artery. E, intraoperative photograph demonstrating the extended retrosigmoid approach, with the patient in the lateral position (right side down), the dura flapped against the transverse and sigmoid sinuses, and exposure of the angle between the petrous bone and the tentorial dura. The vein draining the fistula was visualized at this petrotentorial junction (white asterisk). F, the fistula was interrupted with a clip on the draining vein.

and an orbitozygomatic approach was performed as a second operation to occlude the fistula on the superior surface of the tentorial incisura.

Angiographic Outcomes All patients underwent postoperative angiography, and all but two fistulae were obliterated completely (complete obliteration rate, 94%). One was a superior petrosal sinus DAVF with



A

being followed actively for clinical deterioration or angiographic changes. In patients with cured DAVF, additional angiographic follow-up was not obtained, but 5-year follow-up angiography was recommended.

B

Patient Outcomes

C

E

There was no operative mortality and no new permanent neurological deterioration. Three patients who underwent posterior interhemispheric approaches for galenic DAVF had transient visual field deficits that resolved completely, and one patient who underwent an extended retrosigmoid approach for a downward petrosal DAVF had transient facial weakness (temporary neurological deterioration, 13%). Three patients developed pseudomeningoceles, one of which required a lumboperitoneal shunt. Two other patients required ventriculoperitoneal shunting, one for persistent aquaductal compression from a thrombosed deep venous varix and one for a subarachnoid hemorrhage-induced hydrocephalus. Patients were followed for a mean duration of 4.2 years (range, 1 month to 9 years). At last follow-up, 18 patients (58%) had an MRS score of 0, 8 patients (26%) had an MRS score of 1, and 5 patients (16%) had an MRS score of 2. Five patients who presented after hemorrhage in poor neurological condition improved to good outcomes.

D

F

FIGURE 8. Incisural DAVF (Type 6, Patient 30). A, axial computed tomographic (CT) scan of the brain demonstrating a partially thrombosed venous varix in the right ambient cistern along the tentorial incisura and mass effect on the lateral midbrain. Right external carotid artery angiograms (lateral [B] and anteroposterior [C] views) demonstrating arterial supply to the fistula (red asterisk) from the middle meningeal artery and occipital artery, and drainage through a dilated, tortuous basal vein of Rosenthal. D, intraoperative photograph demonstrating the orbitozygomatic-pterional approach exposing the anterior tentorial incisura, the vein draining the fistula (black asterisk), ambient cistern, internal carotid artery bifurcation (dotted arrow), middle cerebral artery (dashed arrow), and posterior cerebral artery (solid arrow). E, the vein draining the fistula was visualized at the tentorium’s free edge (black asterisk), and F, was occluded with a curved aneurysm clip.

minimal persistent shunting into the superior petrosal sinus. The other was a torcular DAVF with some persistent shunting into the torcula but marked reduction of venous hypertension from 45 to 15 mmHg. These two DAVF were converted from Borden II to Borden I fistulae, with a significant decrease in their hemorrhage risk; both patients have improved and are

NEUROSURGERY

DISCUSSION Tentorial DAVF are complex lesions in deep locations with unusual vascular anatomy and critical surrounding neuroanatomy. The key to treating these lesions is differentiating the six types of tentorial DAVF. Our relatively large surgical experience with these rare lesions has helped us appreciate their nuanced anatomy. Other reports in the literature often confuse their differences by mixing together various types to increase the series’ size (10, 14, 15, 24, 31). Galenic, straight sinus, torcular, and superior petrosal sinus DAVF have all been described previously, although sometimes without clear distinction. For example, galenic and straight sinus DAVF have been reported collectively as “deep venous” DAVF (9, 19). In this report, we have added two other types of fistulae that have not been clearly differentiated: the tentorial sinus DAVF and the incisural DAVF. After a tentorial fistula has been anatomically typed, operative strategy becomes almost algorithmic. Selection of surgical approach is simplified (Fig. 9), and microsurgical interruption of venous drainage is usually straightforward.

Galenic DAVF (Type 1) Galenic DAVF are the most complex of the six types of tentorial fistulae (Fig. 3). The galenic region is the deepest location; the confluence of falx and tentorium creates awkward barriers and surgical blind spots, arterial inflow arrives from all directions, and venous outflow can be difficult to decipher, particularly when veins are tortuous or variceal. Consequently, a panoramic exposure is required that only the posterior interhemipheric approach can provide (4). This approach is performed with the


LAWTON ET AL.

FIGURE 9. Summary of surgical approaches to tentorial DAVF.

patient in the lateral position through a torcular craniotomy that exposes the superior sagittal sinus, both transverse sinuses, and torcula. After opening the dura, the dependent occipital lobe is retracted by gravity, and the interhemispheric fissure opens widely without a retractor. Cuts in the falx above the straight sinus and bilaterally in the tentorium parallel to the straight sinus skeletonize the straight sinus and transform this unilateral, supratentorial exposure into a bilateral, supra-, and infratentorial exposure. Anatomy in front of the fistula in the quadrigeminal and ambient cisterns can be visualized clearly. Skeletonizing the straight sinus also dearterializes galenic DAVF. Transecting the tentorium obliterates supply from the tentorial arteries and ECA branches, whereas transecting the falx obliterates supply from MMA/falcine arteries. Supply from the occipital artery is already obliterated when the scalp flap is elevated, and PMA can be interrupted along its course up the suboccipital dura to the torcular region. However, the goal of occluding the fistula is accomplished by interrupting the venous drainage, not by interrupting the arterial supply. Dural cuts are performed to widen the posterior interhemispheric corridor, visualize the galenic complex, and decipher the venous anatomy, not to de-arterialize the fistula. Interruption of the fistula is completed by placing a clip on the vein draining the fistula, which requires meticulous microsurgical dissection of the venous anatomy. The vein of Galen and internal cerebral veins are most prominent from the perspective of the posterior interhemispheric approach, but the internal cerebral vein is rarely the target for the occluding clip. The basal vein of Rosenthal and the precentral cerebellar vein are more commonly the target for the clip, but are also more difficult to visualize. The basal vein of Rosenthal is identified by dissecting lateral and inferior to the internal cerebral veins


into the ambient cistern. The precentral cerebellar vein lies in a blind spot in the surgical corridor beneath the falcotentorial dura, but this untethered dura can be mobilized aggressively to visualize the cerebellar veins. Veins that are red and dilated declare their participation in draining the fistula, whereas blue veins declare their participation in normal circulation. The connections between fistulous dura and its arterialized draining veins are not always obvious from gross inspection of extravascular anatomy. Clip application is dictated by the straight sinus’ patency, which must be determined preoperatively from the angiogram. All galenic DAVF in our experience were Borden Type III fistulae with retrograde drainage in the vein of Galen or its tributaries. However, not all straight sinuses were occluded in patients with galenic DAVF; some straight sinuses had antegrade flow seen during the venous phase of the angiogram. A patent straight sinus requires vein of Galen preservation and clip application to occlude just the tributary vein draining the fistula, rather than the Galen trunk (Fig. 10). Other tributary veins uninvolved with the fistulous outflow can continue to drain the deep cerebral circulation in an antegrade direction. In contrast, an occluded straight sinus allows clip application directly on the galenic trunk to occlude fistulous outflow, which usually requires less dissection and is easier to decipher. The supracerebellar-infratentorial approach is an alternative approach to galenic DAVF, which has the advantage of positioning the neurosurgeon on the same side of the tentorium as fistulae that drain inferiorly to cerebellar veins. However, the steep pitch of the tentorium at the vein of Galen creates a narrow attic with a very limited view, and a dilated, low-hanging vein of Galen can fill this small field. One patient with an inferiorly draining galenic DAVF was approached early in the series with a supracerebellar-infratentorial approach; the surgical occlusion was incomplete, and she required a posterior interhemispheric approach in a second stage to complete the occlusion.

Straight Sinus DAVF (Type 2) Compared with galenic DAVF, straight sinus DAVF are much simpler: they are not as deeply located, they usually drain out of a solitary vein, and they are exposed by opening the natural subarachnoid plane under the tentorium without skeletonizing sinuses or dissecting venous complexes (Fig. 4). The supracerebellar-infratentorial approach with the patient in the sitting position is the ideal operative approach. With the patient in the sitting position, gravity retracts the cerebellum and opens this plane to easily visualize the fistula, even in patients presenting with hemorrhage that have intraparenchymal clot, cerebellar swelling, and a tight surgical corridor. The supracerebellar-infratentorial approach was performed with three patients in the prone position early in our experience. Retractors are needed when the prone position is used, which risks avulsing the draining vein. This complication occurred in one of these patients, resulting in brisk bleeding from a fistula not yet in view. Subsequently, the sitting position was used in the supracerebellar-infratentorial approach without



A

C

E

B

D

F

vertical and the head is slouched forward to align the plane of the tentorium nearly horizontal. The surgeon sits on a rolling stool and supports his elbows on a free-standing arm brace raised to shoulder level, which relaxes his arms and stabilizes his hands. The dissection and fistula occlusion are usually brief, so this somewhat awkward position for the neurosurgeon is easily tolerated. The suboccipital craniotomy extends from above the transverse sinuses and torcular herophili superiorly to just above, but not to, the foramen magnum. Exposure of the torcular herophili removes the bony ledge that would otherwise obscure the infratentorial plane. Dural tears and venous sinus injury can be dangerous, particularly with patients in the sitting position. Older patients with adherent dura may have sinuses that are not safe to cross with the craniotome, and instead may require a suboccipital craniotomy first, then dissection of the dura from the inner table of cranium under direct visualization. Alternatively, bone overlying the sinuses can be drilled away with a diamond drill bit until the inferior margins of the venous sinuses are seen. We performed this approach in the sitting position in three patients in their 70s without complications from the venous sinuses or air embolism. Once the torcular herophili are exposed, dura is opened in a flap based on the transverse sinuses and tacking sutures that elevate the torcular herophili. The arterialized vein draining the fistula is identified on the cerebellar surface by its red color and traced back to the fistula. Alternatively, the arterialized vein is seen in the subarachnoid space descending from the dura. The draining vein has a thickened wall and distinctive white color with red vaso vasorum. A clip is applied to the draining vein as it exits the tentorial dura, and it is coagulated and cut.

Torcular DAVF (Type 3)

FIGURE 10. A, lateral vertebral artery angiogram demonstrating a galenic DAVF (red asterisk) draining to the vein of Galen, anteriorly to varices in the quadrigeminal cistern, then posteriorly to a superior cerebellar vein. The vein of Galen did not drain the deep cerebral circulation, nor did it connect to the straight sinus, which was patent (solid arrows). B, therefore, the vein of Galen was occluded as it exited the fistula (black arrowhead). C, this patient was positioned prone for the posterior interhemispheric approach, and the distal draining vein was seen below the tentorium on the superior cerebellar surface coursing to the torcular herophili (solid arrows). D, in contrast, this galenic DAVF seen on a lateral vertebral artery angiogram (red asterisk) also had infratentorial venous drainage, but the deep venous circulation drained through a patent vein of Galen and straight sinus (not shown). Therefore, the arterialized vein draining the fistula was isolated (black arrowhead) E, and, occluded with a clip, F, preserving the uninvolved vein of Galen (solid arrows). This patient was positioned laterally, with the right occipital lobe in the dependent position.

further complications. Neurosurgeons often prefer that prone position because it allows them to sit while operating. However, the surgeon can still sit while operating on patients in the sitting position. The patient’s back is positioned nearly

NEUROSURGERY

We encountered just three patients with torcular DAVF, but these few patients demonstrated the difference between treating a Borden Type II and Type III fistula. Type III torcular fistulae that drain exclusively to adjacent veins are treated simply by clipping the arterialized veins as they exit the sinus (Fig. 5). These superficial fistulae are exposed using a torcular craniotomy with the patient positioned prone and with minimal subarachnoid dissection. In contrast, Type II fistulae that drain to torcular sinuses and to adjacent veins are considerably more difficult to treat. Arterialized draining veins are occluded the same way as Type III fistulae, but shunt flow into the torcular sinuses cannot be interrupted without sacrificing a major sinus. Therefore, skeletonization of the torcular herophili is also required with Type II fistulae to interrupt arterial inflow (23). There are a total of eight dural leaflets around the torcula that can harbor arterial supply: falx cerebri, bilateral tentorium (2), bilateral occipital dura (2), bilateral suboccipital dura (2), and falx cerebelli. Complete skeletonization of the torcular herophili requires a total of 12 cuts in these 8 dural leaflets, with 4 of the leaflets requiring two cuts: the occipital dura must be cut along the tranverse sinus and the superior sagittal sinus, and the tentorium must be cut along the straight sinus and the transverse sinus. The torcular craniotomy exposes all of these leaflets, but


LAWTON ET AL.

some additional retraction is needed to make the tentorial incisions along the straight sinus. Our one patient with a Type II torcular fistula presented with venous hypertension measuring 45 mmHg. Draining veins were clipped easily, but aggressive skeletonization of the torcula still did not eliminate all shunt flow to the torcula. Nonetheless, her venous pressures dropped to 10 mmHg, and she improved neurologically.

Tentorial Sinus DAVF (Type 4) We differentiated tentorial sinus DAVF as a distinct but poorly described subtype of tentorial fistulae (Fig. 6). This lesion has not been described in the neurosurgical literature in part because the tentorial sinus is an obscure entity that is often detected but ignored on angiography and MRI. Furthermore, the anatomy of the tentorial sinuses is highly variable, which might make it difficult to recognize different tentorial sinus DAVF as being the same type. Matsushima et al. (16) classified the tentorial sinus into four groups depending on whether venous tributaries originated from the cerebrum (Group I), cerebellum (Group II), the tentorium itself (Group III), or incisura (Group IV). In another anatomic study of 80 cadavers, Muthukumar and Palaniappan (18) found tentorial sinuses in 86% of specimens and classified them into three types depending on their location (medial or lateral) and size (small or large). Miabi et al. (17) used contrast-enhanced magnetic resonance imaging to identify 104 tentorial sinuses in 55 patients and define yet another classification scheme: venous candelabra (Type 1); multiple independent veins (Type II); venous lakes within tentorium (Type III). With so much confusion in the literature concerning a subtle and highly variable venous structure, it is no wonder that the association between the tentorial sinus and paramedian tentorial DAVF has not been firmly established. Although we did not find any description of tentorial sinus DAVF in the neurosurgical literature, it has been reported recently in the radiological literature (11). The three tentorial sinus DAVF encountered in our experience had features implicating the tentorial sinus: none was associated with other dural sinuses (straight, superior petrosal, or transverse sinuses), two were located laterally, where the vein of Labbe might join the tentorial sinus under the temporal and occipital lobes, and one was medially located, where the medial variant of the tentorial sinus has been described. The supratentorial-infraoccipital approach, as described by Smith and Spetzler (22) for posteromedial temporal lobe lesions, was used to expose tentorial sinus DAVF. We prefer to position patients laterally rather than prone, with the head rotated downward to face the floor. A torcular craniotomy is important if the fistula is located medially, because a torcular craniotomy enables the dura to be opened widely and the occipital pole to be mobilized freely. More lateral DAVF near the transverse-sigmoid junction may not need this midline exposure, and a unilateral temporal-occipital craniotomy taken down to or below the transverse sinus would suffice. Occlusion of the fistula is accomplished by interrupting the draining vein as it exits the tentorium.


Superior Petrosal Sinus DAVF (Type 5) Superior petrosal sinus DAVF (Fig. 7) consistently drained infratentorially into the petrosal vein (Dandy’s vein), making the extended retrosigmoid approach an ideal exposure (21). More radical transpetrosal approaches are not necessary because only the draining vein needs to be exposed to interrupt the fistula. The patient is positioned laterally with the head slightly flexed and angled downward toward the floor, optimizing the view along the angle between tentorium and petrous face. We use a C-shaped scalp incision behind the ear, a limited posterior mastoidectomy, skeletonization of the sigmoid sinus with a diamond-bit drill, craniotomy rather than craniectomy, and anterior mobilization of the sigmoid sinus with the dural flap. These maneuvers enhance the exposure of a conventional retrosigmoid approach that does not drill the sigmoid sinus (21). Feeding arteries from the external carotid artery traverse the mastoid and petrous bones to make the drilling bloodier than with other lesions. These transosseus arteries are controlled easily with bone wax or drilling with a diamond bit. If the fistula has ruptured and the cerebellum is swollen, cerebrospinal fluid should be released from the cisterna magna immediately after opening the dura to relax the cerebellum. Microsurgical dissection into the cerebellopontine angle leads to the arterialized petrosal vein, which is frequently variceal because of the high-flow nature of this fistula. The clip is applied as close to the petrous dura as possible, but not so close that closure of the blades avulses the vein. Draining veins distal to the clip should darken after the fistula is interrupted. The venous varice can sometimes hide an additional draining vein coursing medially towards the brain stem, so the varix should be mobilized and this medial territory inspected carefully. Half of these superior petrosal sinus DAVF were Borden Type II fistulae, with drainage medially into the superior petrosal sinus. Patency of this sinus can result in residual, but lowrisk (Borden Type I) shunting after occluding the draining petrosal vein, as in one patient in our experience.

Incisural DAVF (Type 6) Incisural DAVF are the other type of tentorial fistulae not well characterized in the literature. Picard et al. (20) described a “marginal tentorial sinus” that courses along the free edge of the tentorial incisura, which is not present in most people, and receives venous tributaries from the basal vein of Rosenthal and lateral mesencephalic veins (24). We hypothesize that incisural DAVF are associated with this marginal tentorial sinus, and the lack of a clear association between this fistula and venous sinus is caused by the rarity of each entity. In our large experience, this subtype was the least common, with just two patients. Incisural DAVF and tentorial sinus DAVF are similar because both are associated with intrinsic tentorial sinuses, both have variable anatomy, and both drain supratentorially (Fig. 8). However, we differentiated these two types because they require different surgical approaches. Tentorial sinus DAVF are



located infraoccipitally and are exposed through a torcular or occipital craniotomy. In contrast, exposure of incisural DAVF calls for either a more anterior pterional-transsylvian approach or a lateral subtemporal approach. In our experience, the venous drainage from both cases of incisural DAVF coursed posteriorly to the galenic system, but the fistulae were quite anterior, at the level of the uncus or supraclinoid ICA. Incisural DAVF are also near the superior petrosal sinus and can be mistaken for the more common petrosal DAVF. However, the extended retrosigmoid approach used for petrosal DAVF positions the neurosurgeon on the opposite side of the tentorium from an incisural DAVF, and even incising the tentorium as we did in one case was not sufficient to widen the exposure. Therefore, incisural DAVF can be deceptive and require careful analysis of the venous drainage pattern on preoperative angiograms to be certain of their type and location along the incisura. We suspect that some of the difficulty in surgically obliterating superior petrosal sinus DAVF relates to unrecognized differences between superior petrosal sinus and incisural DAVF and the erroneous selection of an infratentorial approach for incisural DAVF.

Limitations We analyzed our experience with tentorial DAVF to provide some appreciation of their anatomic differences and some guidance to neurosurgeons contemplating operative strategies. Tentorial DAVF are rare lesions, and those requiring surgery are rarer still. Consequently, our cohort of patients is small and some types, like the torcular, tentorial sinus and incisural DAVF, have just two or three patients each. We acknowledge that these small numbers may not be conducive to an accurate characterization of their anatomy or clinical behavior. We also recognize the importance of individualizing the selection of surgical approach on the basis of a patient’s unique anatomy and clinical condition. Nonetheless, the selection of surgical approach is critical in treating these lesions safely, and our scheme provides simple recommendations to facilitate operative planning. Our conceptualization of tentorial DAVF types and our surgical algorithm will require prospective assessment.

REFERENCES 1. Anson JA, Spetzler RF: Spinal dural arteriovenous malformations, in Awad IA, Barrow DL (eds): Dural Arteriovenous Malformations. Park Ridge, IL, American Association of Neurological Surgeons, 1993, pp 175–191. 2. Awad IA, Little JR, Akarawi WP, Ahl J: Intracranial dural arteriovenous malformations: Factors predisposing to an aggressive neurological course. J Neurosurg 72:839–850, 1990. 3. Borden JA, Wu JK, Shucart WA: A proposed classification for spinal and cranial dural arteriovenous fistulous malformations and implications for treatment. J Neurosurg 82:166–179, 1995. 4. Chi JH, Lawton MT: Posterior interhemispheric approach: Surgical technique, application to vascular lesions, and benefits of gravity retraction. Neurosurgery 59 [Suppl 1]:ONS41–ONS49, 2006. 5. Cognard C, Gobin YP, Pierot L, Bailly AL, Houdart E, Casasco A, Chiras J, Merland JJ: Cerebral dural arteriovenous fistulas: Clinical and angiographic correlation with a revised classification of venous drainage. Radiology 194:671–680, 1995.

NEUROSURGERY

6. Collice M, D’Aliberti G, Arena O, Solaini C, Fontana RA, Talamonti G: Surgical treatment of intracranial dural arteriovenous fistulae: Role of venous drainage. Neurosurgery 47:56–67, 2000. 7. Davies MA, TerBrugge K, Willinsky R, Coyne T, Saleh J, Wallace MC: The validity of classification for the clinical presentation of intracranial dural arteriovenous fistulas. J Neurosurg 85:830–837, 1996. 8. Goto K, Sidipratomo P, Ogata N, Inoue T, Matsuno H: Combining endovascular and neurosurgical treatments of high-risk dural arteriovenous fistulas in the lateral sinus and the confluence of the sinuses. J Neurosurg 90:289–299, 1999. 9. Halbach VV, Higashida RT, Hieshima GB, Wilson CB, Hardin CW, Kwan E: Treatment of dural fistulas involving the deep cerebral venous system. AJNR Am J Neuroradiol 10:393–399, 1989. 10. Hoh BL, Choudhri TF, Connolly ES, Solomon RA: Surgical management of high-grade intracranial dural arteriovenous fistulas: Leptomeningeal venous disruption without nidus excision. Neurosurgery 42:796–805, 1998. 11. Horie N, Morikawa M, Kitigawa N, Tsutsumi K, Kaminogo M, Nagata I: 2D Thick-section MR digital subtraction angiography for the assessment of dural arteriovenous fistulas. AJNR Am J Neuroradiol 27:264–269, 2006. 12. Kattner KA, Roth TC, Giannotta SL: Cranial base approaches for the surgical treatment of aggressive posterior fossa dural arteriovenous fistulae with leptomeningeal drainage: Report of four technical cases. Neurosurgery 50:1156–1161, 2002. 13. Kiyosue H, Hori Y, Okahara M, Tanoue S, Sagara Y, Matsumoto S, Nagatomi H, Mori H: Treatment of intracranial dural arteriovenous fistulas: Current strategies based on location and hemodynamics, and alternative techniques of transcatheter embolization. Radiographics 24:1637–1653, 2004. 14. Lewis AI, Rosenblatt SS, Tew JM: Surgical management of deep-seated dural arteriovenous malformations. J Neurosurg 87:198–206, 1997. 15. Lewis AI, Tomsick TA, Tew JM: Management of tentorial dural arteriovenous malformations: Transarterial embolization combined with stereotactic radiation or surgery. J Neurosurg 81:851–859, 1994. 16. Matsushima T, Suzuki SO, Fukui M, Rhoton AL, de Oliveira E, Ono M: Microsurgical anatomy of the tentorial sinuses. J Neurosurg 71:923–928, 1989. 17. Miabi Z, Midia R, Rohrer SE, Hoeffner EG, Vandorpe R, Berk CM, Midia M: Delineation of lateral tentorial sinus with contrast-enhanced MR imaging and its surgical implications. AJNR Am J Neuroradiol 25:1181–1188, 2004. 18. Muthukumar N, Palaniappan P: Tentorial venous sinuses: An anatomic study. Neurosurgery 42:363–371, 1998. 19. Ng PP, Halbach VV, Quinn R, Balousek P, Caragine LP, Dowd CF, Higashida RT, Wilson C: Endovascular treatment for dural arteriovenous fistulae of the superior petrosal sinus. Neurosurgery 53:25–33, 2003. 20. Picard L, Bracard S, Islak C, Roy D, Moreno A, Marchal JC, Roland J: Dural fistulae of the tentorium cerebelli. Radioanatomical, clinical and therapeutic considerations [in English, French]. J Neuroradiol 17:161–181, 1990. 21. Quiñones-Hinojosa A, Chang EF, Lawton MT: The extended retrosigmoid approach: An alternative to radical cranial base approaches for posterior fossa lesions. Neurosurgery 58 [Suppl 2]:ONS208–ONS214, 2006. 22. Smith KA, Spetzler RF: Supratentorial-infraoccipital approach for posteromedial temporal lobe lesions. J Neurosurg 82:940–944, 1995. 23. Sundt TM, Piepgras DG: The surgical approach to arteriovenous malformations of the lateral and sigmoid dural sinuses. J Neurosurg 59:32–39, 1983. 24. Tomak PR, Cloft HJ, Kaga A, Cawley CM, Dion J, Barrow DL: Evolution of the management of tentorial dural arteriovenous malformations. Neurosurgery 52:750–762, 2003. 25. Tubbs RS, Hansasuta A, Loukas M, Louis RG, Shoja MM, Salter G, Oakes WJ: Branches of the petrous and cavernous segments of the internal carotid artery. Clin Anat 20:596–601, 2006. 26. Ushikoshi S, Houkin K, Kuroda S, Asano T, Iwasaki Y, Miyasaka K, Abe H: Surgical treatment of intracranial dural arteriovenous fistulas. Surg Neurol 57:253–261, 2002. 27. van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J: Interobserver agreement for the assessment of handicap in stroke patients. Stroke 19:604–607, 1988. 28. Weinstein M, Stein R, Pollock J, Stucker TB, Newton TH: Meningeal branch of the posterior cerebral artery. Neuroradiology 7:129–131, 1974.


LAWTON ET AL.

29. Weinstein MA, Duchesneau PM, Dohn DF: Angiographic identification of the meningeal branch of the posterior cerebral artery. AJR Am J Roentgenol 128:326–327, 1977. 30. Zhang JC, Dion J, Barrow DL: Surgical treatment of intracranial dural arteriovenous fistulas, in Lawton MT, Higashida RT, Gress DR (eds): Controversies in Neurological Surgery: Neurovascular Diseases. New York, Thieme Medical Publishers, 2006, pp 150–156. 31. Zink WE, Meyers PM, Connolly ES, Lavine SD: Combined surgical and endovascular management of a complex posttraumatic dural arteriovenous fistula of the tentorium and straight sinus. J Neuroimaging 14:273–276, 2004.

COMMENTS

T

he authors have elegantly described and classified the various types of dural arteriovenous fistulae (AVF) that may involve the tentorium. Their careful anatomic description of the arterial supply and venous drainage simplifies the surgical decision-making with regard to operative approach and intraoperative strategy. Although each of these fistula types have been described in the literature, Lawton et al. have more clearly defined the differentiating anatomical characteristics of these six types and have introduced a better understanding of the tentorial sinus fistula. Tentorial dural AVFs are associated with an aggressive natural history, and virtually all of them require treatment. In our experience, many of these lesions can be treated by endovascular means. Although arterial embolization alone rarely cures these lesions, transvenous embolization will successfully obliterate many, and transarterial microcatheterization, with “wedging” of the catheter into an arterial feeder and pushing endovascular glue across the fistula into the venous side, will obliterate others. As stated by the authors, many of these lesions are best treated by surgical obliteration of the venous drainage. The actual surgical treatment of the majority of these lesions is very straightforward, requiring only simple interruption of the venous drainage. The primary exception is galenic fistulae, which are more challenging to manage. The most important aspect of operative planning is selection of the appropriate surgical approach that minimizes brain retraction and provides surgical access to the venous drainage. The greatest strength of this publication is the clear description of each of the surgical approaches that is ideal for each subtype of fistula. The only exception I would take to the authors’ strategy is their routine use of preoperative arterial embolization in these patients. The goal of surgical therapy is not to de-arterialize the fistula but to interrupt the venous drainage. It is questionable, in my mind, whether preoperative arterial embolization enhances the surgical procedure in any but a few selected circumstances. Preoperative embolization does, however, expose the patient to two procedures with the attendant risks of each. We have also found that the routine use of intraoperative angiography reliably documents complete obliteration of the fistula and on occasion has demonstrated additional venous drainage that was not recognized intraoperatively. More recently, we have used near-infrared indocyanine green videoangiography in treating dural AVFs and have found it to be a less invasive and highly reliable method to demonstrate intraoperative obliteration. Daniel L. Barrow Atlanta, Georgia

T

his series forms an extremely helpful and well-illustrated guide to tentorial arteriovenous fistulae. The organization of these lesions into six subtypes expands on the previous literature on this topic and helps to explain the anatomical variations seen with dural AVFs. In general, these lesions have a single fistulous connection with a vein that traverses the subarachnoid space. Ligation of that vein is usually cur-


ative for these types of dural AVFs, except when direct shunting into one of the major sinuses is present. As noted by the authors, the operative approach is guided by the most direct route to the principal draining vein and not necessarily to the arterial input. This latter point is in sharp contradistinction to the way the more common congenital cortical arteriovenous malformations are treated surgically. All patients in this series underwent preliminary arterial embolization. This type of embolization is a helpful prelude to surgical treatment but is rarely curative. The possibility that some minority of these fistulae can be cured with transvenous embolization was not discussed in this presentation. In the same way that clipping of the vein is curative, if the vein can be directly catheterized from the dural sinuses or via transarterial catheter placement through the fistula and into the draining vein, then glue placed in the vein can also completely obliterate the fistula without open surgery. Robert A. Solomon New York, New York

L

awton et al. present 31 patients with tentorial dural arteriovenous fistulae (DAVF) treated over a 9-year span. The authors have categorized tentorial DAVF into six types on the basis of the anatomical location of the fistulous point. Accordingly, the type of DAVF dictates which surgical approach should be used. The techniques described are an excellent recapitulation of numerous previous works. The authors also provide an algorithm that simplifies the surgical strategy in approaching tentorial DAVF. Interestingly, 55% of their patients presented with hemorrhage, and 84% demonstrated evidence of cortical venous drainage. These findings bolster the well-accepted belief that DAVF along the tentorium are high-risk vascular lesions that deserve immediate treatment, even when they are asymptomatic. As the authors state, careful identification of the exact point of fistulous connection is the critical first step. In tentorial DAVF, with their numerous feeding arteries and draining veins, selective and even superselective catheter-based angiography may be necessary to delineate the angioarchitecture of the fistula completely. We typically use endovascular techniques as the primary therapy to define the anatomy, to simplify the anatomy with selective transarterial embolization, and to attempt curative embolization via both transarterial and/or transvenous routes. We have discovered that many of these tentorial DAVF can be disrupted purely with endovascular treatment. In fact, the dictum that transarterial embolization cannot disrupt the connection point in DAVF is changing because of growing experience with Onyx. For patients who require microsurgical disruption, we concur with the authors that with a thorough understanding of the angiographic anatomy, the surgeon can take advantage of the appropriate natural corridors to the fistulous point. As the authors state, fistula interrupt i o n , n o t a r t e r i a l d i s c o n n e c t i o n , i s t h e k e y t o t re a t m e n t . Intraoperatively, we look for darkening of the arterialized veins after disruption. If not found, further inspection is required. At our institution, intraoperative angiography is used routinely to verify surgical cure. Because of the dangerous natural history of this type of fistula, we still obtain catheter-based follow-up angiography at 1 and then 5 years postoperatively. We congratulate the authors for their logical surgical strategy, cogent discussion of meticulous techniques, and superb clinical outcomes with these complex vascular lesions. Louis Kim Robert F. Spetzler Phoenix, Arizona



T

he authors retrospectively analyzed 31 patients who underwent microsurgical treatment for tentorial DAVF over a 9-year period. The authors provide a nice breakdown of the locations of the DAVF and surgical strategies for each location. As endovascular techniques have advanced, we have had better experience with complete obliteration of these lesions with newer endovascular agents. However, any potential residue after endovascular treatment should be considered for surgical excision. The authors have clearly defined the strategy of excision to be disconnection of the vein from the dural arteriovenous malformation. Lawton et al. outline the key features of strategies to maintain surgical corridors to allow normal venous drainage to persist while obliterating the retrograde flow through the arterialized vein from the dural arteriovenous malformation. We have developed an algorithm of management similar to those outlined by the authors. In addition to the surgical techniques used, we have made use of stereotactic techniques for tentorial sinus DAVF in an attempt to minimize the operative approach and target the draining vein. With these particular lesions, we have seen the majority of patients present with hemorrhage, given the high incidence of cortical venous hypertension. The authors provide us with a solid outline for conceptualizing and treating this family of dural arteriovenous malformations. Christopher S. Ogilvy Boston, Massachusetts

L

awton et al. presented their extensive experience with the microsurgical treatment of 31 tentorial DAVF. They classified them into six different types (galenic, straight sinus, torcular, tentorial sinus, superior petrosal sinus, and incisural) based on fistula location, dural base, associated sinus, and direction of venous drainage. Furthermore, the authors developed specific operative strategies based on these types, including elegant operative nuances as the lateral positioning for the posterior hemispheric approach, the anterior retraction of the sigmoid sinus for the extended retrosigmoid approach, and the superior retraction of the torcular herophili for the supracerebellar infratentorial approach.

NEUROSURGERY

As recognized by the authors, the main limitation of their work is the relatively small number of patients, particularly for some types of tentorial DAVF. Although the anatomy of the tentorial sinuses is highly variable, the medial tentorial sinus usually receives infratentorial (superior cerebellar surface) veins, whereas the lateral tentorial sinus drains supratentorial (temporobasal and occipitobasal) veins (2). Therefore, if the affected sinus is the medial tentorial sinus, an infratentorial drainage would be expected and the supracerebellar infratentorial approach would be preferred. In addition, the anatomy of the basal vein of Rosenthal is also highly variable (1). Our study on the microsurgical anatomy of the basal vein in 37 injected hemispheres revealed that 32% of hemispheres had a basal vein with predominant anterior drainage (into the sphenoparietal or cavernous sinus) secondary to aplastic or hypoplastic anastomosis between the anterior and middle segments of the basal vein, and 13% had an inferior (into the lateral mesencephalic vein) or tentorial predominant drainage secondary to aplastic or hypoplastic anastomosis between the middle and posterior segments of the basal vein (unpublished results). The knowledge of these common variations may aid in understanding the complex anatomy of those tentorial DAVF with drainage through the basal vein (13% of patients in the series of Lawton et al.). The work completed by Lawton et al. represents the most comprehensive experience on the microsurgical treatment of tentorial DAVF reported to date. Their surgical results are excellent, and therefore this article will be an essential reference for every neurosurgeon facing these challenging lesions. Juan C. Fernández-Miranda Aaron S. Dumont Neal F. Kassell Charlottesville, Virginia 1. Huang YP, Wolf BS: The basal cerebral vein and its tributaries, in Newton TH, Potts DG (eds): Radiology of the Skull and Brain. St. Louis, CV Mosby, 1971, vol 2, book 3, pp 2111–2154. 2. Rhoton AL Jr: The cerebral veins. Neurosurgery 51 [Suppl 4]:S159–S205, 2002.


VASCULAR Instrumentation and Technique

USEFULNESS OF PREOPERATIVE THREE-DIMENSIONAL COMPUTED TOMOGRAPHIC ANGIOGRAPHY WITH TWO-DIMENSIONAL COMPUTED TOMOGRAPHIC IMAGING FOR RUPTURE POINT DETECTION OF MIDDLE CEREBRAL ARTERY ANEURYSMS Kojiro Wada, M.D., Ph.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan

Hirohiko Arimoto, M.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan

Hidenori Ohkawa, M.D., Ph.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan

Toshiki Shirotani, M.D., Ph.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan

Yohsitaro Matsushita, M.D., Ph.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan

Takashi Takahara, M.D. Department of Neurosurgery, Japan Defense Force Central Hospital, Tokyo, Japan Reprint requests: Kojiro Wada, M.D., Ph.D., JDF Central Hospital, Ikejiri 1–2-24, Setagaya, Tokyo 154–8532, Japan. Email: [email protected] Received, January 25, 2007. Accepted, September 4, 2007.

OBJECTIVE: We report the technique of three-dimensional computed tomographic (CT) angiography with a two-dimensional CT image aiding in the early operation of ruptured middle cerebral artery aneurysms. This combined image allows the prediction of the rupture point in the aneurysm and may reduce the risk of rupture during early clipping surgery. METHODS: The findings for 14 patients with 14 middle cerebral artery ruptured aneurysms who underwent subsequent early clipping were analyzed. The average aneurysm size was 8.5 mm, and there were two large and one giant aneurysms. CT examinations were performed by means of a multidetector CT scanner (Aquilion M16; Toshiba Medical Systems, Tokyo, Japan) and reconstructed with a workstation (ZIO M900 QUADRA; Amin Co., Ltd., Tokyo, Japan). We constructed an operating view through three-dimensional CT angiography for a lateral transsylvian approach with a two-dimensional CT image (nonshaded volume-rendering image), which was perpendicular to the direction of the surgical approach. Using this combined image, we predicted the rupture point of the aneurysm and successfully performed clipping surgery through a lateral transsylvian approach. Rupture points were confirmed at the time of surgery. Rupture points of 13 out of 14 aneurysms appeared as we expected, but one differed; all aneurysms were successfully clipped. Thirteen of the 14 patients could be clipped without rupture at surgery, but the remaining patient experienced rupture just after craniotomy. CONCLUSION: The combination of three-dimensional CT angiography and two-dimensional CT images may help improve the surgical outcome by indicating aneurysmal rupture points, leading to the prevention of rupture. KEY WORDS: Aneurysm surgery, Middle cerebral artery, Rupture point, Subarachnoid hemorrhage, Threedimensional angiography Neurosurgery 62[ONS Suppl 1]:ONS126–ONS133, 2008

R

ebleeding of a ruptured aneurysm after the initial hemorrhage is a major factor affecting mortality and morbidity rates. In particular, intraoperative aneurysm rupture was the most common and devastating technical complication that occurred among intraoperative complications (7). The outcome for patients with middle cerebral artery (MCA) aneurysms is significantly worse than that for patients with other anterior circulation aneurysms. This is attributable in part to anatomic aspects and in part to the high incidence of


DOI: 10.1227/01.NEU.0000297061.58271.B4

rupture and bleeding of MCA aneurysms (23). Technical safety increases along with the individual surgeon’s experience (3, 14, 16), although, clearly, this is not the only relevant factor (25). Recently, three-dimensional computed tomographic angiography (3-D CTA) using multidetector computed tomographic (CT) scanning has been developed and has increased the sensitivity to detect aneurysms, providing clearer vascular images for cerebral aneurysm detection. Furthermore, 3-D CTA is


THREE- AND TWO-DIMENSIONAL COMPUTED TOMOGRAPHIC ANGIOGRAPHY FOR ANEURYSM SURGERY

a powerful tool to construct operative images for the operator because 3-D CTA images can be rotated freely and incorporated with cranial information to construct an operative view. Therefore, 3-D CTA has been taking the place of digital subtraction angiography (DSA) for aneurysm diagnosis (1, 5, 8, 15). Three-dimensional CTA has been reported to be superior to DSA in the demonstration of anatomic features of aneurysms and MCA configuration, which include the M1, M2, and M3 segments (24). In particular, 3-D CTA can depict whether the M1 segment is exposed to the aneurysm anteriorly or posteriorly by constructing a surgical-view image. This may lead to a safer operation by making proximal control easier in the event of premature rupture. However, 3-D CTA itself cannot indicate the rupture point of aneurysms. If the rupture point could be indicated before surgery, it would be very helpful to perform successful clipping of aneurysms and to avoid unexpected rerupture. Patients with ruptured aneurysms often show hematoma and clotting on the dome in the area of likely rupture; therefore, the combination of the two-dimensional (2-D) CT image, which can provide hematoma and clot information, and 3-D CTA might allow the identification of the rupture point of MCA aneurysms. Therefore, we predicted rupture points using combined 3-D CTA and 2-D CT images before surgery and confirmed them during surgery.

PATIENTS AND METHODS Patient Assessment We treated 16 patients with subarachnoid hemorrhage caused by MCA ruptured aneurysm from June 2004 to June 2006. Three-dimensional CTA data were obtained, and we detected ruptured aneurysms in 14 patients. The other two patients consequently underwent DSA, but no aneurysm was detected. Therefore, these patients waited for 2 weeks and underwent DSA again, and 2-mm aneurysms were subsequently detected. After that, delayed surgery and clipping were performed. These two patients were excluded from this study. Fourteen patients with subarachnoid hemorrhage (three men and 11 women; mean age, 63 yr; age range, 44–80 yr) with 14 ruptured saccular aneurysms were analyzed as summarized in Table 1. The average aneurysm size was 8.5 mm, including two large and one giant aneurysms.

3-D CTA We used a multidetector CT scanner (Toshiba Aquilon; Toshiba, Inc., Tokyo, Japan) to perform 3-D CTA. Patients were first evaluated with an unenhanced CT scan, followed by an enhanced CT scan. During enhanced examination, mild sedative drugs were administered to the patients. CT scans were performed according to the parameters described previously (15, 19) with slight modification, briefly described as follows: 0.5-mm collimation, 3.5 mm per rotation table increment (3.5 helical pitch), 0.75 seconds/r gantry rotation speed, and 120 kV/300 mA. Before scanning began, 60 ml (350 mg I/ml) of iomeprol contrast medium (Eisai, Tokyo, Japan) was injected through a right antecubital vein by means of a power injector at a rate of 4 ml/second. The initiation of scanning was set with an automatic or visual bolus-tracking program, which was used to start the first arterial phase scan after the injection of contrast medium. Axial slices were reconstructed with a 0.5-mm slice thickness at 0.3-mm inter-

NEUROSURGERY

vals. The images obtained were transferred and processed with a standalone Zio M900 workstation (Amin Co., Ltd., Tokyo, Japan).

Constructing 3-D CTA + 2-D CT Images Three-dimensional CTA images were produced by subtracting bone objects generated by a threshold technique and editing the images manually to remove any remaining bone of the cranial base. Then, a surgical view of the 3-D CTA image was constructed with a frontotemporal cranial window opening. Finally, a nonshaded volume-rendering (VR) image, a 10-mm-thick image was made perpendicular to the direction of the surgical approach. The window level of this nonshaded VR image was set at 40 Hounsfield units and the window width was set at 100 Hounsfield units, the same values as those of conventional 2-D CT images. Thus, we called this nonshaded VR image a 2-D CT image. These 2-D CT images were combined with a 3-D CTA image at the workstation. To use these images, points showing an irregular aneurysmal shape or aneurysmal bleb on 3-D CTA and adjacent thick high-density areas on 2-D CT are expected to indicate the hematoma and clot on the dome in areas of likely rupture. An early operation could be performed using these 3-D CTA findings alone. We performed clipping surgery using a lateral transsylvian approach. Rupture points were confirmed during surgery.

RESULTS Three-dimensional CTA was performed in 14 patients summarized in Table 1. No complications, including rupture, occurred as a result of the enhanced CT study. Rupture points of 13 aneurysms could be predicted, but one aneurysm could not; all were confirmed at the time of surgery. Thirteen patients were clipped successfully without rupture at the time of surgery, but one patient with a Hunt and Hess Grade V aneurysm and a 100-ml hematoma in the temporal lobe ruptured just after craniotomy.

Illustrative Cases

Patient 1 A 56-year-old man presented with a history of severe headaches just before admission and a Hunt and Hess grade of II (18, 22). No neurological deficits were observed. Plain CT scans demonstrated Fisher Grade 3 subarachnoid hemorrhage (SAH) at the basal cistern and a hematoma at the left sylvian fissure (Fig. 1A). The patient was then transferred to our hospital. A three-dimensional CTA study on admission demonstrated an 8-mm aneurysm at a bifurcation of the left MCA with two blebs, anterior and lateral (Fig. 1B). Three-dimensional CTA rotated for the surgical view indicated that the main stem of the MCA (M1) could be identified anterior to the superior branch of the MCA (M2). Three-dimensional CTA combined with a 2-D image visualized the aneurysm, the M1 and M2 segments of the MCA, and the hematoma. This image indicated that the sylvian hematoma faced the lateral bleb of the aneurysm (Fig. 1C). Therefore, the lateral bleb was considered a rupture point. The superior branch of M2 and M1 was separate from the hematoma (Fig. 1C). We operated using a lateral transsylvian approach. The patient underwent early surgery with ipsilateral pterional craniotomy. The distal portion of the superior branch of M2 was recognized and dissected proximally along the superior branch of M2. The bifurcation of MCA was recognized, and then the distal portion of the


WADA ET AL.

TABLE 1. Summary of 14 patients with ruptured middle cerebral artery aneurysms Patient no.

Age (yr)/Sex

Size (mm)

Hunt and Hess grade

Fisher group

Rerupture

Expectation

1

56/M

8

II

3

No

Possible

2

51/F

25

II

3

No

Possible

3

77/F

15

III

4

No

Possible

4

72/F

15

V

4

No

Possible

5

76/F

7

III

3

No

Possible

6

55/F

4

III

3

No

Possible

7

80/F

8

IV

4

No

Possible

8

44/F

5

III

4

No

Possible

9

76/F

7

III

4

No

Possible

10

53/F

3

II

3

No

Possible

11

44/F

6

II

3

No

Possible

12

60/F

5

V

4

Rupture

Possible

13

63/M

6

II

3

No

Possible

14

74/M

5

III

3

No

Impossible

M1 segment was identified anterior to the M2 superior branch. Finally, the inferior neck of the aneurysm was dissected, and the aneurysm was successfully clipped with a 7-mm right-angled titanium clip (B. Braun Aesculap, Tokyo, Japan), without rupture. The expected rupture point of the aneurysm was confirmed as the rupture point (Fig. 1D).

Patient 2 A 51-year-old woman presented with a history of severe headaches and a Hunt and Hess grade of II. No neurological deficits were observed. Plain CT scans demonstrated Fisher Grade 3 SAH at the basal cistern and a hematoma at the left sylvian fissure (Fig. 2A). Threedimensional CTA on admission demonstrated a 25-mm giant aneurysm at an MCA bifurcation, which had two blebs medially and anterolaterally (Fig. 2B). Three-dimensional CTA rotated for the surgical view showed that the M1 branch could be identified anterior to the M2 superior branch. Three-dimensional CTA with a 2-D image visualized the aneurysm, the M1 and M2 segments of the MCA, and the hematoma. This image indicated that the sylvian hematoma facing the lateral bleb was thicker than that of the anterolateral bleb (Fig. 2C). Therefore, the lateral bleb was considered a rupture point. Both superior and inferior branches of M2 and M1 were separate from the hematoma (Fig. 1C). We operated using a lateral transsylvian approach. The patient underwent early surgery with ipsilateral pterional craniotomy. The lateral sylvian fissure was dissected. The distal portion of the superior branch of M2 was recognized and dissected proximally along the superior branch of M2. Then, the medial surface of the aneurysm was dissected to observe the medial neck of the aneurysm. A subfrontal exposure to the carotid cistern was performed, and the left carotid artery and the left M1 trunk were identified. The aneurysm was clipped tentatively (11) and was then dissected totally from the brain. The expected rupture point of the aneurysm was confirmed as the rupture point (Fig. 2D). The neck of the aneurysm was broad and included the origin of both anterior and posterior branches of M2. Therefore, two long Sugita straight 25-mm clips (Mizuho, Inc., Tokyo, Japan) were applied to the dome of the aneurysm to form a proximal M2 portion to avoid narrowing of the M2 origin. Thus, the aneurysm was successfully clipped.


Patient 3 A 77-year-old woman presented with a history of severe headaches just before admission and a Hunt and Hess grade of III. Slight left-sided hemiparesis was observed. Plain CT scans demonstrated Fisher Grade 4 SAH at the basal cistern and a hematoma at the right temporal lobe (Fig. 3A). Three-dimensional CTA on admission demonstrated a large 15-mm aneurysm at a bifurcation of the right MCA (Fig. 3B). Threedimensional CTA rotated for the surgical view showed that the M1 branch could be identified anteromedial to the M2 superior branch. Therefore, we chose a lateral transsylvian approach for clipping. Threedimensional CTA with a 2-D image visualized the aneurysm, the M1 and M2 segments of the MCA, and the hematoma. This image indicated that the temporal hematoma faced the lateral side of the aneurysm (Fig. 3C). The patient underwent early surgery with ipsilateral pterional craniotomy. A major portion of the temporal hematoma was removed through an incision in the superior temporal gyrus, leaving some hematoma and pia remnants around the aneurysm. The distal portion of the superior branch of M2 was recognized and dissected proximally along the superior branch of M2. A part of the medial surface of the aneurysm was recognized. The distal portion of the M1 segment was recognized anteromedial to the M2 superior branch (Fig. 3D). Finally, the superior and inferior necks of the aneurysm were dissected, and the aneurysm was successfully clipped with an 11-mm straight titanium clip (B. Braun Aesculap) through a lateral approach after definite proximal control of the MCA was achieved without rerupture. The expected rupture point of the aneurysm was confirmed as the rupture point (Fig. 3D).

DISCUSSION Conventional catheter angiography, including intra-arterial DSA, is still the “gold standard” for diagnosing cerebral aneurysms. However, it is invasive and involves some risks. The rate of complications has been reported to be 0.25 to 1% and is higher in elderly patients (4, 9, 12, 13). Furthermore, after diag-



tional angiography. CTA is a less invasive technique for visualizing most of the major vasculature of the whole body (21). Furthermore, in published reports, 3-D CTA has been found to be superior to conventional angiography in depicting an aneurysm’s shape, in showing its relation to surrounding vessels and the adjacent cranium, and in demonstrating the anatomic variation of vessels (1). Therefore, 3-D CTA is very valuable for aneurysmal surgery. Thus, we routinely use CTA initially for the detection D C of aneurysms in all patients with SAH. Particularly for early surgery of MCA aneurysms, an understanding of the MCA architecture, the branches surrounding the aneurysm, and the size and shape of the aneurysm are key to successful treatment (6, 17). Three-dimensional CTA has some advantages. It allows the surgeon to rotate the images as a surgical view, facilitating a more accurate understanding of the anatomic relationFIGURE 1. A, plain computed tomographic (CT) scan demonstrating Fisher Grade 3 subarachnoid hemorrhage ships between the M1 and M2 (SAH) at the basal cistern with a hematoma at the left sylvian fissure (black arrow). B, three-dimensional computed branches in the aneurysm neck. tomographic angiography (3D-CTA) study on admission demonstrating a 7-mm aneurysm at a bifurcation of the left Because control of the aneumiddle cerebral artery (MCA) with two blebs, anterior (arrow) and lateral (arrowhead). C, 3D-CTA with a tworysm neck is the most impordimensional (2-D) image visualizing the aneurysm, the M1 and M2 segments of the left MCA, and the Sylvian tant component of clipping the hematoma (arrow). This image indicated that the sylvian hematoma faced the lateral bleb (arrowhead) of the aneurysm; therefore, we considered that the lateral bleb was a rupture point. D, the expected rupture point of the aneurysm and securing it from aneurysm was confirmed as the rupture point during the operation. AN, aneurysm; M1, trunk of the MCA; M2, the circulation, 3-D CTA is very branch of the MCA. valuable for aneurysm surgery. Many MCA aneurysms are multilobulated or bulbous and often have broad necks (3). This nostic and interventional cerebral angiography, embolic events image modality can provide information on the proper clip size are more frequent than the apparent neurological complication and shape or the necessity of multiple clip application. A 2-D CT rate on magnetic resonance imaging scans (2). image perpendicular to the surgical view at the level of the Recently, magnetic resonance angiography and 3-D CTA have aneurysm in addition to the 3-D CTA image provides informabeen used for cerebral aneurysm detection. In the acute stage of tion on the surrounding bone. This information can be used by SAH, magnetic resonance angiography is less appropriate than surgeons to accurately assess how far it is from the sphenoid CTA because of motion artifacts from patient movement, probwing to the aneurysm and how deep a dissection is necessary to lems in monitoring patients in poor condition, and the requirereach the aneurysm from the brain surface. ment that a patient be moved to a magnetic resonance imaging There are three basic approaches to MCA aneurysms: two transtable after examination through unenhanced CT scanning (15). sylvian approaches, which are medial and lateral, and a superior On the other hand, 3-D CTA can be used to examine patients temporal gyrus approach (10, 20, 26). The lateral transsylvian immediately after unenhanced CT scanning is performed and approach is preferred by many neurosurgeons because of its can reduce the examination time, even in the case of uncooperreduced operating time (3). Furthermore, the lateral transsylvian ative or comatose patients. It has been reported that the diagapproach causes less direct compression damage to the brain nostic ability of 3-D CTA is almost the same as that of conven-

A

NEUROSURGERY

B


WADA ET AL.

Three-dimensional CTA rotated for the surgical view depicts the anatomic features of aneurysms and MCA configuration. Therefore, the surgeon is able to accurately understand whether the M1 segment is exposed anteriorly or posteriorly to the aneurysm. This affects the surgeon’s decision regarding proximal-todistal versus distal-to-proximal control. If the M1 segment is exposed anteriorly to the aneurysm in the CTA surgical view, the surgeon can reach the M1 segment to perform a proxD C imal-to-distal approach to the aneurysm. However, if the M1 segment is exposed posteriorly to the aneurysm in the CTA surgical view, the surgeon has some difficulties reaching the M1 segment without distal dissection. In this case, a distalto-proximal approach to the aneurysm is chosen. The direction of the aneurysm also affects the surgical approach. Most (45%) of the bifurcated MCA aneurysms have been reported to be laterally diFIGURE 2. A, plain CT scan demonstrating Fisher Grade 3 SAH at the basal cistern with a hematoma at the left rected, 38% pointed inferiorly, sylvian fissure (black arrow). B, 3-D CTA study on admission demonstrating a 25-mm giant aneurysm at an MCA 15% pointed superiorly, and bifurcation, which showed two blebs, medial and anterolateral. C, 3-D CTA with a 2-D image visualized the only 2% pointed medially (23). aneurysm, the M1 and M2 segments of the MCA, and the sylvian hematoma (arrows). This image indicated that If aneurysms are directed the sylvian hematoma facing the anterolateral bleb was thicker than the medial hematoma; therefore, we considered superiorly, the surgeon prefers that the anterolateral bleb was a rupture point. D, the expected rupture point of the aneurysm was confirmed as the rupture point during the operation. AN, aneurysm; M2, branch of the MCA. to reach the M1 first. If aneurysms are directed laterally or inferiorly, the surgeon prefers resulting from gentle retraction to achieve a good surgical view to reach the M2 first. Then, retrograde dissection of this M2 because this approach gradually allows deeper access after openbranch leads to the M1. Even if using a lateral transsylvian ing a wide view of the shallow brain surface. However, this approach, M1 control before dissection of the aneurysm neck is approach includes the danger of premature rupture of the essential in the case of aneurysm rupture. Gentle brain retracaneurysm because it brings the surgeon to the aneurysm dome tion is needed to reach the M1 segment because too much tenbefore control of the afferent vessel, which is the MCA stem (M1 sion applied to the aneurysm may lead to its premature rupsegment), is ensured. Therefore, it is conceivable that a lateral ture. Therefore, wide dissection of the M2 segment close to the transsylvian approach is less desirable for large or complicated aneurysm neck is necessary. Furthermore, the surrounding MCA aneurysms. Perioperative aneurysm rupture has been arteries and brain should be freed of adhesions from the reported to affect the postoperative outcome. Therefore, avoiding aneurysm neck to reach the M1 segment of the MCA. Because aneurysm rupture during surgery is important. For this reason, 3-D CTA with 2-D CT for surgical field imaging may indicate the medial transsylvian approach is perhaps the most common the rupture point, this would be extremely helpful for the surand basic procedure in surgery for ruptured MCA aneurysms. geon to assess whether proximal control of the M1 segment of However, the distal approach can be indicated, especially when the MCA can be achieved through the lateral transsylvian the M1 segment is long and the fundus of the aneurysm projects approach. If the rupture point of the aneurysm is positioned laterally or inferiorly. laterally or inferiorly, the medial and superior surfaces of the

A

B




However, there are some limitations in performing 3-D CTA and 2-D CT image-guided surgery. We failed to predict the rupture point in one case. This patient had two aneurysms, one of which arose from the left MCA bifurcation and the other from the inferior branch of the MCA. We could not predict the rupture point because we could not detect the density difference nor high-density thickness of either aneurysm using 3-D CTA and 2-D CT imaging. During surgery, no clot was observed around the dome of D C the ruptured aneurysm. Combined 3-D CTA and 2-D CT imaging was ineffective in this patient. Furthermore, 3-D CTA has some disadvantages when compared with conventional angiography. First, perforating arteries are not consistently visualized on 3-D CTA because of their size. Second, small aneurysms (especially aneurysms ⱕ 2 mm) may not be found. Third, the simultaneous demonstration of both arteries and veins often occurs in the FIGURE 3. A, plain CT scan demonstrating Fisher Grade 4 SAH at the basal cistern with a hematoma at the right area of the MCA and posterior temporal tip (black arrow). B, 3-D CTA study on admission demonstrating a large, 15-mm aneurysm at an MCA cerebral artery. The discriminabifurcation. C, 3-D CTA with a 2-D image visualized the aneurysm, the M1 and M2 segments of the MCA, and the tion of arteries from veins is temporal hematoma (arrow). This image indicated that the temporal hematoma faced the lateral side of the aneurysm; therefore, we considered that the lateral side of the aneurysm was the rupture point. D, the expected rupture point difficult. Fourth, 3-D CTA does of the aneurysm was confirmed as the rupture point during the operation. not provide dynamic information on the cerebral circulation such as collateral flow. Lastly, the residual neck or de novo aneurysm after clipping in cases aneurysm dome are safe routes. Therefore, the surgeon can use involving multiple clips or cobalt alloy clips is difficult to visuthe lateral transsylvian approach to reach the proximal M1 segalize because the clips appear as metal artifacts on 3-D CTA (15). ment of the MCA. In our cases, 3-D CTA with 2-D computed tomography for surgical field imaging was able to identify rupture points. Therefore, we could use the lateral transsylvian CONCLUSION approach, and the aneurysm was successfully clipped. DSA remains the standard technique for the investigation of Thus, the fusion VR image produced from 3-D CTA and 2-D SAH. However, we believe that the 3-D CTA technique can CT images could be used to indicate the rupture point of the provide much more information than DSA, and the resulting cerebral aneurysm based on the configuration of the aneuryspreoperative images can lead to more successful surgery. A mal shape, sylvian hematoma, and clot on the dome. We suclarger series is needed to clarify the real efficacy in detecting the cessfully performed 3-D CTA and 2-D CT image-guided surpoint of rupture. gery of acutely ruptured cerebral aneurysms using a lateral

A

B

transsylvian approach. This indicates that 3-D CTA and 2-D CT imaging used in combination is very helpful and may improve the results of surgery for acutely ruptured MCA aneurysms.

NEUROSURGERY

REFERENCES 1. Anderson GB, Steinke DE, Petruk KC, Ashforth R, Findlay JM: Computed tomographic angiography versus digital subtraction angiography for the


WADA ET AL.

2.

3. 4.

5.

6.

7.

8.

9.

10.

11.

12. 13. 14.

15.

16. 17.

18.

19.

20. 21. 22.

23.

24.

diagnosis and early treatment of ruptured intracranial aneurysms. Neurosurgery 45:1315–1320, 1999. Bendszus M, Koltzenburg M, Burger R, Warmuth-Metz M, Hofmann E, Solymosi L: Silent embolism in diagnostic cerebral angiography and neurointerventional procedures: A prospective study. Lancet 354:1594–1597, 1999. Chyatte D, Porterfield R: Nuances of middle cerebral artery aneurysm microsurgery. Neurosurgery 48:339–346, 2001. Cloft HJ, Joseph GJ: Risk of cerebral angiography in patients with subarachnoid hemorrhage, cerebral aneurysms, and arteriovenous malformation: A meta-analysis. Stroke 30:317–320, 1999. Dammert S, Krings T, Moller-Hartmann W, Ueffing E, Hans FJ, Willmes K, Mull M, Thron A: Detection of intracranial aneurysms with multislice CT: Comparison with conventional angiography. Neuroradiology 46:427–434, 2004. Fogelholm R, Hernesniemi JA, Vapalahti MP: Impact of early surgery on outcome after aneurysmal subarachnoid hemorrhage. A population-based study. Stroke 24:1649–1654, 1993. Fridriksson S, Säveland H, Jakobsson KE, Edner G, Zygmunt S, Brandt L, Hillman J: Intraoperative complications in aneurysm surgery: A prospective national study. J Neurosurg 96:515–522, 2002. Futami K, Nakada M, Iwato M, Kita D, Miyamori T, Yamashita J: Simulation of clipping position for cerebral aneurysms using three-dimensional computed tomography angiography. Neurol Med Chir 44:6–13, 2004. Heiserman JE, Dean BL, Hodak JA, Flom RA, Bird CR, Drayer BP, Fram EK: Neurologic complications of cerebral angiography. AJNR Am J Neuroradiol 15:1401–1407, 1994. Heros RC, Ojemann RG, Crowell RM: Superior temporal gyrus approach to middle cerebral artery aneurysms: Technique and results. Neurosurgery 10:308–313, 1982. Kato Y, Sano H, Okuma I, Akashi K, Hayakawa M, Yoneda M, Yoshida K, Kanno T: Pitfalls in aneurysm surgery in acute stages. Neurol Res 19:17–24, 1997. Koenig GH: Rupture of intracranial aneurysms during cerebral angiography: Report of ten cases and review of the literature. Neurosurgery 5:314–324, 1979. Komiyama M, Tamura K, Nagata Y, Fu Y, Yagura H, Yasui T: Aneurysmal rupture during angiography. Neurosurgery 33:798–803, 1993. Leipzig TJ, Morgan J, Horner TG, Payner T, Redelman K, Johnson CS: Analysis of intraoperative rupture in the surgical treatment of 1694 saccular aneurysms. Neurosurgery 56:455–468, 2005. Matsumoto M, Sato M, Nakano M, Endo Y, Watanabe Y, Sasaki T, Suzuki K, Kodama N: Three-dimensional computerized tomography angiographyguided surgery of acutely ruptured cerebral aneurysms. J Neurosurg 94:718– 727, 2001. Maurice-Williams RS, Kitchen ND: Ruptured intracranial aneurysms— Learning from experience. Br J Neurosurg 8:519–527, 1994. McLaughlin N, Bojanowski MW: Early surgery-related complications after aneurysm clip placement: An analysis of causes and patient outcomes. J Neurosurg 101:600–606, 2004. Oshiro EM, Walter KA, Piantadosi S, Witham TF, Tamargo RJ: A new subarachnoid hemorrhage grading system based on the Glasgow Coma Scale: A comparison with the Hunt and Hess and World Federation of Neurological Surgeons Scales in a clinical series. Neurosurgery 41:140–147, 1997. Otawara Y, Ogasawara K, Ogawa A, Sasaki M, Takahashi K: Evaluation of vasospasm after subarachnoid hemorrhage by use of multislice computed tomographic angiography. Neurosurgery 51:939–942, 2002. Pritz MB, Chandler WF: The transsylvian approach to middle cerebral artery bifurcation/trifurcation aneurysms. Surg Neurol 41:217–219, 1994. Prokop M: Multislice CT angiography. Eur J Radiol 36:86–96, 2000. van Gijn J, Bromberg JE, Lindsay KW, Hasan D, Vermeulen M: Definition of initial grading, specific events, and overall outcome in patients with aneurysmal subarachnoid hemorrhage. A survey. Stroke 25:1623–1627, 1994. van Rinne J, Hernesniemi JA, Niskanen M, Vapalahti MP: Analysis of 561 patients with 690 middle cerebral artery aneurysms: Anatomic and clinical features as correlated to management outcome. Neurosurgery 38:2–11, 1996. Villablanca JP, Hooshi P, Martin N, Jahan R, Duckwiler G, Lim S, Frazee J, Gobin YP, Sayre J, Bentson J, Viñuela F: Three-dimensional helical computerized tomography angiography in the diagnosis, characterization, and management of middle cerebral artery aneurysms: Comparison with conventional angiography and intraoperative findings. J Neurosurg 97:1322–1332, 2002.


25. Woodrow SI, Bernstein M, Wallace MC: Safety of intracranial aneurysm surgery performed in a postgraduate training program: Implications for training. J Neurosurg 102:616–621, 2005. 26. Yas¸argil MG: Middle cerebral artery aneurysms, in Yas¸argil MG (ed): Microneurosurgery II: Clinical Considerations, Surgery of the Intracranial Aneurysms and Results. Stuttgart, Georg Thieme Verlag, 1984, pp 72–91.

Acknowledgments We thank Kumiko Suzuki, R.T., and Kouji Yuba, R.T., for their excellent technical assistance.

COMMENTS

T

he authors report an original imaging technique that consists of combining three-dimensional (3-D) and computed tomographic (CTA) and two-dimensional (2-D) computed tomographic (CT) images to predict the rupture point in the aneurysm and therefore reduce the risk of rupture during early clipping surgery. In their experience with 14 ruptured middle cerebral artery (MCA) aneurysms, this modification improved the surgical outcome through indicating aneurysmal rupture points, leading to the prevention of rupture. Multislice helical CTA is currently the primary imaging modality in many neurovascular centers for several reasons. It is noninvasive and performs quick imaging; it has sensitivity and specificity that are comparable to those of digital subtraction angiography in aneurysms larger than 2 mm; and it performs quick reconstruction of 3-D images that allows the surgeon to rotate the images to become surgical view images. As mentioned by the authors, CTA facilitates more accurate understanding of the anatomical relationships between the M1 and M2 branches in the aneurysm neck. Nevertheless, detailed knowledge of the microneurosurgical anatomy of the MCA and its common variants is essential for correct interpretation of CTA images. The bifurcation of the MCA is located at the level of or distal to the limen insulae and proximal to the genu of the MCA in most hemispheres. It is of paramount importance to identify and differentiate early cortical branches or “proximal false bifurcations” from true bifurcations. Early branches typically arise at right angles to the main trunk of the MCA, whereas the true postbifurcation trunks run nearly parallel, diverging only minimally before they reach the genu of MCA (1). The lateral lenticulostriate arteries usually arise proximal to the bifurcation, but in between 17% (1) to 23% (2) of patients, they originate from the postbifurcation part of the MCA. The correct interpretation of the microsurgical anatomy of the MCA shown by the CTA study allows the surgeon to elaborate a mental spatial view of the architecture of the MCA arterial tree in the sylvian fissure and its relation to the sphenoidal ridge, the anatomy of the aneurysm, and, when present, its relation to the hematoma. The combination of 3-D CTA and 2-D CT images as shown by Wada et al. has been proven to be valuable to implementing presurgical mental assimilation of the aneurysmal rupture point. Juan C. Fernández-Miranda Aaron S. Dumont Neal F. Kassell Charlottesville, Virginia

1. Tanriover N, Kawashima M, Rhoton AL Jr, Ulm AJ, Mericle RA: Microsurgical anatomy of the early branches of the middle cerebral artery: Morphometric analysis and classification with angiographic correlation. J Neurosurg 98:1277–1290, 2003.



W

ada et al. report the technique of 3-D CTA with a 2-D CT image aiding in the early operation of ruptured MCA aneurysms. They obtained 3-D CTA images in 14 patients from June 2004 to June 2006. They combined the 2-D CT image with the 3-D CTA image at the workstation to find points showing an irregular aneurysmal shape or aneurysmal bleb on 3-D CTA and adjacent thick high-density areas on 2-D CT images, which indicate the hematoma and clot. With this technique they were able to find the point of rupture in 13 of 14 patients.

The technique proposed in this report could be helpful in surgery for MCA aneurysms, but in our opinion, a larger series of patients is needed to clarify the real efficacy in detecting the point of rupture. This is a welcome technique to improve preoperative planning for MCA aneurysms. Emiliano Passacantilli Roberto Delfini Rome, Italy

The Grand Amphitheater of the School of Surgery in Paris, (1780), copper engraving, Claude-René-Gabriel Poulleu. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


VASCULAR Operative Nuances

EXTERNAL CAROTID ARTERY TO MIDDLE CEREBRAL ARTERY BYPASS WITH THE SAPHENOUS VEIN GRAFT Erica F. Bisson, M.D. Division of Neurosurgery, Department of Surgery, University of Vermont College of Medicine, Burlington, Vermont

Agostino J. Visioni, M.D. Division of Neurosurgery, Department of Surgery, University of Vermont College of Medicine, Burlington, Vermont

Bruce Tranmer, M.D. Division of Neurosurgery, Department of Surgery, University of Vermont College of Medicine, Burlington, Vermont

Michael A. Horgan, M.D. Division of Neurosurgery, Department of Surgery, University of Vermont College of Medicine, Burlington, Vermont Reprint requests: Michael A. Horgan, M.D., Fletcher Allen Health Care, 111 Colchester Avenue, MCHV Campus-Fletcher 5, Burlington, VT 05401. Email: [email protected] Received, September 10, 2006. Accepted, May 29, 2007.

PATIENTS WITH OCCLUSIVE cerebrovascular disease who have failed maximal medical therapy, which consists of antiplatelet agents as well as maximizing modifiable risk factors such as blood pressure, cholesterol, smoking cessation, and obesity, and whose lesions are not amenable or have not responded to the more common vascular procedures (i.e., carotid endarterectomy or stenting) are considered candidates for an extracranial–intracranial bypass. Additionally, for a patient to be a candidate, he/she must have an adequate graft vessel. Typically, this vessel is the superficial temporal artery. However, oftentimes, the superficial temporal artery is an inadequate vessel or the patient requires a high-flow conduit. It is in these patients that use of the saphenous vein should be considered. In this report, we detail the technical aspects of performing an extracranial–intracranial bypass by using a saphenous vein graft. KEY WORDS: Cerebral revascularization, Cerebrovascular occlusive disease, External carotid artery, Extracranial-intracranial bypass, Middle cerebral artery, Saphenous vein Neurosurgery 62[ONS Suppl 1]:ONS134–ONS139, 2008

E

xtracranial–intracranial (EC–IC) bypass for the treatment of ischemic cerebrovascular disease was pioneered by Donaghy (2) and Yas¸argil (9) in 1967. Over the ensuing decade, approximately 2000 patients underwent this operation for transient ischemic attacks, strokes, and stenotic arterial lesions that were inaccessible or inoperable with conventional vascular operations. Yas¸ argil and Yonekawa (10) reported follow-up data on 82 superficial temporal artery (STA) to middle cerebral artery (MCA) bypass operations at their institution. They demonstrated an 87% graft patency rate. However, the complication rate was not insignificant, with morbidity and mortality rates of 25 and 3.5%, respectively (9). Although this series did categorize patients by their preoperative clinical symptoms, the authors did not perform a specific analysis to assess which preoperative characteristics may predict a favorable outcome or decrease the morbidity and mortality rates. Secondary to the increase in popularity of this operation, in 1985, a multi-institutional randomized, controlled clinical trial was published comparing the best medical treatment with aspirin to STA–MCA bypass plus aspirin in a group of patients with one or more transient ischemic attacks or minor strokes. The primary outcome of fatal or nonfatal stroke was found


DOI: 10.1227/01.NEU.0000297014.73587.1D

to be more frequent and earlier, albeit not statistically significant, in patients assigned to the surgical arm, thereby leading to the conclusion that bypass affords no reduction in the risk of stroke (3). Subsequently, EC–IC bypass lost favor as an option for the treatment of cerebrovascular ischemia. However, one criticism of the study was the diverse indications for bypass in the patient population. Many believe that if this procedure was evaluated only in patients with evidence of hypoperfusion, a benefit would have been demonstrated. In lieu of this, Nussbaum and Erickson (7) published data on 20 patients undergoing STA–MCA bypass for occlusive cerebrovascular disease after failing maximal medical therapy. These patients were defined as having continued transient ischemic attacks or strokes despite anticoagulant therapy with or without antiplatelet therapy. Their results underscore the excellent outcomes seen in this particular group of patients, with 17 out of 20 having complete cessation of their preoperative ischemic symptoms. In addition, there is an ongoing trial, Carotid Occlusion Surgery Study, that is specifically looking at the benefits of EC–IC bypass in patients with hypoperfusion. With improvements in surgical technique and technology, options for bypass graft and recipient vessels have increased. The choice of


EXTRACRANIAL–INTRACRANIAL BYPASS WITH SAPHENOUS VEIN GRAFT

donor vessel depends ultimately on both the size of the recipient vessel and the requisite blood flow. Although STA, radial artery, and saphenous vein have been used for this purpose, an individual patient’s anatomy often dictates the choice (5). Advantages of a radial artery graft include a luminal diameter that most closely approximates that of an M2 branch, and it is a physiological conduit for arterial blood. Major disadvantages include the occurrence of spasm, which can be rectified with new techniques of insufflations, and the possibility of the loss of the vaso vasorum, which may lead to further degenerative changes in the wall of the artery itself. Practically speaking, the radial artery is unavailable if the patient has a positive Allen’s test, which occurred in our case (6). Use of the saphenous vein for EC–IC bypass is commonly used in patients with an inadequate STA. Additionally, this is an excellent option in individuals who require a high-flow conduit, as the saphenous vein tends to have significantly higher flow than its arterial counterpart (4). In this report, we detail the operative technique of external carotid to MCA bypass with a reverse saphenous vein graft.

OPERATIVE CONSIDERATIONS Patients with occlusive cerebrovascular disease for whom maximal medical therapy has failed and whose lesions are not amenable or have not responded to the more common vascular procedures (i.e., carotid endarterectomy or stenting) are candidates for EC–IC. Additionally, a patient must demonstrate an inequality of perfusion when compared with the contralateral hemisphere, as well as a lack of adequate cerebral blood flow reserve when challenged with acetazolamide. At our institution, the preoperative evaluation consists of magnetic resonance imaging of the brain, cerebral angiography, and a computed tomographic perfusion study, with and without an acetazolamide challenge, to help predict those patients most likely to benefit from bypass. In particular, patients with a pattern of large vessel stenosis with inadequate collateralization and distal flow are deemed appropriate. Additionally, for a patient to be a candidate, he/she must have an adequate graft vessel. Typically, the superficial temporal artery is used. However, we consider using a saphenous vein graft if 1) the STA is technically not feasible (i.e., the artery has been injured), 2) there is a mismatch between the diameter of the STA and recipient vessel, and 3) a high-flow conduit is needed (the STA is considered a low-flow conduit: 15–25 ml/min versus 70–140 ml/min for saphenous vein). The vein is usually harvested from the lower leg because the diameter of the vessel more closely approximates the cerebrovasculature. However, the vein can be harvested from the thigh if there is excessive tortuosity of the lower saphenous vein or it is not technically feasible.

PATIENT POSITIONING The patient is positioned supine on the operating table. After intubation and induction of general anesthesia, electroen-

NEUROSURGERY

cephalography leads are placed for intraoperative monitoring of burst suppression. The patient’s head is placed into a Mayfield three-point head rest and fixated with the head turned 30 degrees away from the intended side of the bypass for a standard pterional craniotomy. After pin fixation, both the ipsilateral neck and frontotemporal region are prepped for exposure. The bilateral lower extremities are prepped and draped for vein harvest (Fig. 1). Before the start of surgery, the patient is administered prophylactic antibiotics, a loading dose of prophylactic antiepileptics, and dexamethasone.

OPERATING ROOM SETUP To facilitate the head and neck surgical approaches as well as the vein harvest, the patient is turned 180 degrees from the anesthesiologist. For the approach, one surgeon exposes the carotid artery, whereas the other exposes the intracranial compartment. The scrub nurse is positioned opposite the neck dissection to facilitate instrument passing. The operating microscope is placed on the ipsilateral side of the dissection.

SURGICAL TECHNIQUE A question mark-shaped incision is made over the pterional region. The skin and subcutaneous tissue are cut down to the galea and pericranium with a combination of surgical blade and bovie electrocautery. The skin flap and temporalis muscle are elevated with periosteal dissectors, reflected anteriorly, and held in place with fishhooks attached to a Leyla bar. Keyhole and temporal burr holes are placed and a standard pterional bone flap is turned, exposing the dura (Fig. 2). The sphenoid ridge is then drilled to increase the surgical field of view. Dural stay-sutures are spaced evenly around the craniotomy to provide epidural hemostasis. The dura is then opened in a Cshaped fashion. Concomitantly, the exposure of the carotid artery is begun with a transverse skin incision and dissection with electrocautery through the platysma and along the medial border of the sternocleidomastoid muscle. The tissue plane is developed with a combination of Metzenbaum scissors and finger dissection to expose the carotid sheath. The facial vein is identified, running perpendicularly over the carotid bifurcation and subsequently isolated and ligated. The carotid sheath is incised, exposing the common, internal, and external carotid arteries along with the internal jugular vein and vagus nerve. The arteries are circumferentially dissected in preparation for anastomosis with the vein graft (Fig. 3). While the craniotomy and neck dissections are being conducted, the saphenous vein is harvested. At our institution, we use the assistance of the cardiothoracic team to harvest the saphenous vein. Whenever possible, this procedure is performed endoscopically. The details of this procedure have been described previously (1). From the cardiothoracic literature, benefits of endoscopically harvested saphenous vein include decreased complication rates and decreased hospital stay. The


BISSON ET AL.

A

FIGURE 1. Intraoperative photograph showing both of the patient’s legs prepped for saphenous vein harvest.

FIGURE 2. Intraoperative photograph showing a pterional craniotomy exposing the dura.

FIGURE 3. Intraoperative photograph showing the neck dissection exposing the common carotid bifurcation. A vessel loop is placed around the external carotid artery.

theoretical disadvantage of increased trauma to the vein, leading to intimal injury and increased thrombosis rates, has been disproved in several studies that compared open versus endoscopic harvest (8, 11). Once the vein is harvested, it is checked for leaks and placed in heparinized saline. Valvotomy is not performed because the graft is used in a reverse fashion. After the isolation of the external carotid and preparation of the graft, a chest tube is tunneled subcutaneously through a preauricular tract from the neck incision to the cranial incision.


B

FIGURE 4. A, intraoperative photograph showing that a chest tube is passed from the neck to the craniotomy incision in the preauricular space. B, intraoperative photograph showing that the saphenous vein is passed through the chest tube. Note the dotted line drawn on the vein for orientation.

Before passing the vein through the chest tube, we used a marking pen to draw a line on the vein (Fig. 4). This line serves as a reference to prevent twisting of the vein. The microscope is then brought in and standard microsurgical technique is used to open the sylvian fissure widely for clear exposure of the M2 branches. A vessel, usually 1.5 mm in diameter or greater, is chosen as the recipient and is dissected from its surrounding arachnoid. All exposed brain is continuously irrigated and protected with cottonoid patties. Burst suppression with either barbiturates or propofol is initiated, and a dose of 5000 units of heparin is given. Blood pressure is kept normo- to slightly hypertensive. The intracranial portion of the vein is cut at a 45-degree angle to create a fishmouth opening. The anastomotic site on the recipient vessel is excluded from the circulation with small aneurysm clips placed proximally and distally from the site (Fig. 5). An arteriotomy in the recipient vessel is performed via a longitudinal incision (Fig. 6). With 9–O nylon sutures, the vein is anchored to the opening of the recipient vessel at the upper and lower apices. Interrupted 9–O nylon sutures are then placed circumferentially around the remaining portion of the anastomosis (Fig. 7). We use an interrupted suture line rather than a running suture technique, as this helps to prevent constriction at the anastomotic site. The vein graft is continuously irrigated with heparinized saline through the proximal end that has been secured to a microirrigator with a heparin needle. Careful attention is given



A

B FIGURE 7. Artist’s illustration showing anchoring sutures placed at the upper and lower apices of the incision. The anastomosis is then completed with interrupted 9–O nylon sutures.

FIGURE 5. A, intraoperative photograph showing a fish-mouth opening made at the distal end of the vein graft. The adventitia is stripped from the vein to prepare the surface for suturing. B, intraoperative photograph showing the temporary aneurysm clips placed proximally and distally to the anastomotic site. FIGURE 8. Intraoperative photograph showing the proximal end of the graft sutured to the external carotid artery.

FIGURE 6. Intraoperative photograph showing a longitudinal incision made in the recipient vessel.

to checking the integrity of the suture line by back-filling the graft before starting the external carotid anastomosis. After the graft length is tailored, aneurysm clips are placed above and below the anastomotic site on the proximal external carotid. The vessel is then opened with an 11-blade and Potts scissors. The graft is anchored proximally and sewn into place with interrupted 6–O Prolene (Ethicon, Inc., Somerville, NJ) (Fig. 8). Before the final suture is placed, the graft is back-bled. The aneurysm clips are then removed and Doppler ultrasonography is used to verify robust flow through the graft and the recipient vessel.

NEUROSURGERY

The craniotomy and neck dissections are closed simultaneously after irrigation with saline/bacitracin solution. The dura is reapproximated loosely with tacking sutures using 4–O Neurilon (Ethicon, Inc.). The bone flap is replaced after widening of a burr hole for passage of the graft into the extracranial space. Care is taken to avoid kinking or compression of the graft. Both the muscle and galea are closed with interrupted sutures, and the skin is closed with a running 4–O Vicryl (Ethicon, Inc.) rapide suture. For closure of the neck incision, the platysma is closed with a running 3–O Vicryl suture followed by interrupted, inverted 3–0 Vicryl sutures. Dermabond (Ethicon, Inc.) is used to reapproximate the skin. Postoperatively, care is taken to avoid undue pressure on the subcutaneous portion of the graft. Additionally, Doppler recordings are checked routinely. A postoperative angiogram is performed in the first postoperative week to ensure graft patency.

CONCLUSIONS EC–IC bypass using a saphenous vein graft is an excellent treatment option in individuals who would benefit from a


BISSON ET AL.

revascularization procedure but have an inadequate superficial temporal artery or who require a high-flow conduit. Patients with continued symptomatic ischemia despite optimal medical therapy and poor hemodynamic reserve will benefit from this operation.

REFERENCES 1. Alexander MJ, Perna J: Endoscopic saphenous vein graft harvest for extracranial-intracranial bypass procedures. Surg Neurol 63:565–568, 2005. 2. Donaghy RM: Neurologic surgery. Surg Gynecol Obstet 134:269–270, 1972. 3. EC/IC Bypass Study Group: Failure of extracranial-intracranial arterial bypass to reduce the risk of ischemic stroke. Results of an international randomized trial. The EC/IC Bypass Study Group. N Engl J Med 313:1191–1200, 1985. 4. Extracranial-Intracranial Bypass. Available at: http://www.dcmsonline.org/ jax-medicine/1998journals/november98/bypass.htm. Accessed 11/16/07. 5. http://www.aans.org/education/journal/neurosurgical/mar03/14–3-3.pdf. Accessed 10/20/05. 6. Liu JK, Kan P, Karwande SV, Couldwell WT: Conduits for cerebrovascular bypass and lessons learned from the cardiovascular experience. Neurosurg Focus 14:E3, 2003. 7. Nussbaum ES, Erickson DL: Extracranial-intracranial bypass for ischemic cerebrovascular disease refractory to maximal medical therapy. Neurosurgery 46:37–43, 2000. 8. Perrault LP, Jeanmart H, Bilodeau L, Lesperance J, Tanguay JF, Bouchard D, Page P, Carrier M: Early quantitative coronary angiography of saphenous vein grafts for coronary artery bypass grafting harvested by means of open versus endoscopic saphenectomy: A prospective randomized trial. J Thorac Cardiovasc Surg 127:1402–1407, 2004. 9. Yas¸argil MG: Anastomosis between the superficial temporal artery and a branch of the middle cerebral artery, in Yas¸argil MG (ed): Microsurgery Applied to Neurosurgery. Stuttgart, Georg Thieme Verlag, 1969, pp 105–115. 10. Yas¸argil MG, Yonekawa Y: Results of microsurgical extra-intracranial arterial bypass in the treatment of cerebral ischemia. Neurosurgery 1:22–24, 1977. 11. Yun KL, Wu Y, Aharonian V, Mansukhani P, Pfeffer TA, Sintek CF, Kochamba GS: Randomized trial of endoscopic versus open vein harvest for coronary artery bypass. J Thorac Cardiovasc Surg 129:496–503, 2005.

Brain revascularization procedures are mostly performed at present for the treatment of complex aneurysms, cranial base tumors that require carotid artery resection, and moyamoya disease or syndrome. Regarding selection of the conduit, a saphenous vein has as good a patency rate as a radial artery graft. The radial artery has some theoretical advantages over the saphenous vein graft regarding long-term patency and closer matching to brain arteries. However, any long-term advantages in the cerebral circulation remain to be proven, in contrast to the coronary revascularization experience. Regarding technique, I prefer not to anastomose saphenous vein grafts to arteries smaller than 2 mm. The larger the mismatch between the vein graft (which is usually about 4 mm) and the cerebral recipient artery, the greater the turbulence at the anastomotic site and the risk of failure. Regarding suturing, I have not observed any difference between a continuous suturing technique and interrupted sutures in regard to bypass patency or flow. The excimer laser-assisted nonocclusive anastomosis technique of bypass uses a metal ring sutured onto the artery and vein graft, and the excimer laser to punch a hole in the artery must be mentioned as a method of bypass without flow interruption in the intracranial vessel. Laligam N. Sekhar Seattle, Washington

E

xtracranial-intracranial (EC-IC) bypass techniques using the saphenous vein are well established (1–11). At our institution, EC-IC bypass is used for treating aneurysms and tumors. Our preference for constructing a high-flow EC-IC bypass is to harvest a radial artery graft rather than a saphenous vein graft. The reasons include improved patency rates and better and easier anastomosis. Our use of the EC-IC bypass to treat occlusive cerebrovascular disease is currently limited to patients enrolled in the Carotid Occlusive Surgery Study. Francisco Ponce Robert F. Spetzler Phoenix, Arizona

COMMENTS

I

n this article, the authors have illustrated the technique of saphenous vein bypass for brain ischemia. Some comments are pertinent regarding the indications, as well as the operative techniques. For brain ischemia, saphenous vein bypass is currently a rare procedure. After the multicenter randomized Superficial Temporal Artery (STA)-Middle Cerebral Artery (MCA) Bypass Trial demonstrated no benefit in favor of surgery, a revascularization procedure has rarely been used in North America for this indication, but it is still being used in other Asian countries. There are still some patients who may benefit from this procedure, and that possibility is currently being studied through the ongoing Carotid Occlusive Surgery Study trial. In the meantime, clinicians face the dilemma of what to do with patients who have been shown to have brain ischemia and have recurrent transient ischemic attacks or strokes and may not have treatment options other than medical therapy because revascularization procedures are not being approved by Medicare for ischemic indications. When revascularization for ischemia is selected, the first choice is to use the superficial temporal artery as the donor vessel. When this vessel is not available, then the radial artery or the saphenous vein may be selected as conduits. Both of these, and in particular, saphenous vein grafts, have a potential risk of causing hyperemia postoperatively. These types of revascularization have not been proven to prevent strokes any more than the STA-MCA bypass, and additional studies are needed.


1. David CA, Zabramski JM, Spetzler RF: Reversed-flow saphenous vein grafts for cerebral revascularization: Technical note. J Neurosurg 87:795–797, 1997. 2. Friedman JA, Piepgras DG: Current neurosurgical indications for saphenous vein graft bypass. Neurosurg Focus 14:E1, 2003. 3. Friedrich H, Laas J, Walterbusch G, Rickels E: Extra-intracranial bypass procedure with saphenous vein grafts. Thorac Cardiovasc Surg 34:57-62, 1986. 4. Guegan Y, Kerdiles Y, Fardoun R, Pecker J: Extra-intracranial anastomosis using a venous graft. Acta Neurochir (Wien) 59:177–185, 1981. 5. Jafar JJ, Russell SM, Woo HH: Treatment of giant intracranial aneurysms with saphenous vein extracranial-to-intracranial bypass grafting: Indications, operative technique, and results in 29 patients. Neurosurgery 51:138–146, 2002. 6. Kawashima M, Rhoton AL Jr, Tanriover N, Ulm AJ, Yasuda A, Fujii K: Microsurgical anatomy of cerebral revascularization. Part I—Anterior circulation. J Neurosurg 102:116–131, 2005. 7. Liu JK, Kan P, Karwande SV, Couldwell WT: Conduits for cerebrovascular bypass and lessons learned from the cardiovascular experience. Neurosurg Focus 14:E3, 2003. 8. Morgan MK, Ferch RD, Little NS, Harrington TJ: Bypass to the intracranial internal carotid artery. J Clin Neurosci 9:418–424, 2002. 9. Quinones-Hinojosa A, Du R, Lawton MT: Revascularization with saphenous vein bypasses for complex intracranial aneurysms. Skull Base 15:119–132, 2005. 10. Regli L, Piepgras DG, Hansen KK: Late patency of long saphenous vein bypass grafts to the anterior and posterior cerebral circulation. J Neurosurg 83:806–811, 1995.



Michael T. Lawton San Francisco, California

extremely important practice as these procedures often require vessel sacrifice. The same is true for the treatment of complex and giant aneurysms for which great difficulties are still encountered with surgery and endovascular techniques and poor results are shown. The increasing experience with bypass operations has revived an important tool of the neurosurgeon’s technical resources. In selected patients and with adequate technique, external carotid artery-to-MCA bypass is once again gaining popularity in neurosurgical practice as an effective treatment for patients with occlusive cerebrovascular disease. Many techniques were developed since the initial work of Donaghy and Yas¸argil with the STA-MCA bypass in 1967, but, unfortunately, the ideal bypass technique is still a controversial subject. The issue regarding the adequate choice of the appropriate donor vessel, i.e., STA versus radial artery versus saphenous vein in terms of the amount of flow it can deliver, as well as the means to predict how much flow an ischemic area needs or can tolerate is still unanswered. Techniques vary considerably among different groups and also change constantly among the same groups. Consensus is yet to be achieved; nevertheless, the authors are to be commended for their detailed description of their own technique.

reservation of arterial flow in the management of complex cranial base lesions through the use of bypass operations became an

Helder Tedeschi Evandro P. de Oliveira São Paulo, Brazil

11. Story JL, Brown WE Jr, Eidelberg E, Arom KV, Stewart JR: Cerebral revascularization: Common carotid to distal middle cerebral artery bypass. Neurosurgery 2:131–135, 1978.

T

his article describes the technical details of the external carotid artery-to-MCA bypass using a saphenous vein graft. I perform this operation without major differences from the authors’ technique, except that I prefer continuous sutures rather than interrupted sutures. Continuous suturing can be done very quickly, loops of suture can be left loose during the suturing to help visualize the vessel walls clearly, only four knots are needed, and the seal is tight. I have not seen continuous suture constrict the anastomotic site, and, in my experience, a long arteriotomy (approximately three times the width of the recipient artery) is more important than the type of suturing in avoiding anastomotic constriction. For the proximal anastomosis, I like to arteriotomize the ECA with an aortic punch rather than a linear incision because it removes a clean disc of artery wall and ensures a wide opening into the graft.

P

Frontispiece of the Anatomic Compendium, (1717), Johann Georg Puschner. From: WolfHeidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


VASCULAR Technical Note

IMPROVED IMAGE INTERPRETATION WITH COMBINED SUPERSELECTIVE AND STANDARD ANGIOGRAPHY (DOUBLE INJECTION TECHNIQUE) DURING EMBOLIZATION OF ARTERIOVENOUS MALFORMATIONS Tom L. Yao, M.D. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, Tennessee

Eric Eskioglu, M.D. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, Tennessee

Michael Ayad, Ph.D., M.D. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, Tennessee

Arthur J. Ulm, M.D. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, Tennessee

Robert A. Mericle, M.D. Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, Tennessee Reprint requests: Robert A. Mericle, M.D., Department of Neurological Surgery, Vanderbilt University Medical Center, T4224 MCN, Nashville, TN 37232-2380. Email: [email protected] Received, September 28, 2006. Accepted, September 24, 2007. ONLINE DIGITAL VIDEO

NEUROSURGERY

OBJECTIVE: Interpretation of angioarchitecture during embolization of intracranial arteriovenous malformations (AVMs) is critical to optimizing results. We describe an adjunctive technique to aid in the interpretation of AVM embolization and improve safety. METHODS: In the past 100 consecutive patients who underwent AVM embolization by a single surgeon (RAM), each AVM nidus was selectively catheterized and microangiography was performed. After the microcatheter contrast exited the AVM, guiding catheter angiography was performed during the same digital run. The microangiogram was digitally superimposed on the guiding catheter angiogram to delineate important landmarks such as the nidus perimeter, draining veins, and microcatheter tip, which were then drawn on the digital subtraction angiographic monitor with a marking pen in two orthogonal views. RESULTS: Important landmarks were continually visualized during the embolization procedure despite subtracted fluoroscopy (“blank” roadmap). These techniques qualitatively helped to: 1) appreciate the overall size and morphology of the nidus, 2) clearly visualize the safe limits of the embolic injection within the nidus perimeter, 3) clearly visualize draining patterns to help avoid premature venous embolization, 4) decipher small draining veins from arteries, 5) continuously monitor the location and status of the microcatheter tip, and 6) increase the confidence of the surgeon during prolonged embolic injections. CONCLUSION: The double injection technique, with marking pen demarcation of the nidus perimeter, venous drainage, and microcatheter tip position, was qualitatively useful in every case. KEY WORDS: Angioarchitecture, Angiography, Arteriovenous malformation, Double injection angiography, Digital subtraction angiography, Embolization, Superselective angiography Neurosurgery 62[ONS Suppl 1]:ONSE140–ONSE141, 2008

U

nderstanding the detailed vascular anatomy during cerebral angiography is critical for successful intracranial arteriovenous malformation (AVM) embolization. Superselective catheterization and angiography of distal feeding arteries are routinely performed before embolization of intracranial AVMs or other intracranial lesions (1–3). Microangiography provides high-resolution angiographic images of the small portion of the nidus, which was catheterized. However,

DOI: 10.1227/01.NEU.0000297028.75373.7F

because only a small portion of the nidus is visualized with the microcatheter injection, the relationship between the microcatheter angiogram and the overall surrounding cerebrovascular anatomy cannot be visualized. Understanding the relationship between these superselective images and the overall angioarchitecture is a limitation of microangiography that needs improvement. We define the term “angioarchitecture” as the combination of the abnormal AVM vascular anatomy along with the surrounding


YAO ET AL.

normal cerebrovascular anatomy. Traditionally, with printed xray film, this angioarchitecture could be visualized by manually superimposing the film of the microcatheter angiogram with the x-ray film of the guiding catheter angiogram when they are in the same projection and orientation. A number of problems can limit the usefulness of the overlapping film technique, including time consumption, unnecessary film expense, radiation exposure, and potential differences in magnification, projection, and orientation. Additionally, most medical centers are moving toward filmless digital imaging and limiting the use of hard-copy film printers. By using standard digital subtraction angiography (DSA) equipment, a wax pencil or felt-tip marking pen, and standard DSA display monitors, we studied a technique for delineating the angioarchitecture that is especially helpful during prolonged (⬎30 min) embolizations of intracranial AVMs.

MATERIALS AND METHODS (see video at web site) The past 100 consecutive patients who underwent AVM embolization by a single surgeon (RAM) were analyzed. This technique is best used when the patient is under general endotracheal anesthesia. For the purposes of this report, all cases selected involved general endotracheal anesthesia; the same technical development was used in every case. The best angiographic projection must be obtained. It is important that all arterial feeders and draining veins are as distinct from the AVM nidus as possible, thereby allowing best visualization of these key structures throughout the embolization procedure. A microcatheter is first navigated to a position deep within the nidus of an AVM or in a distal pedicle directly entering the AVM. After the best angiographic projection is obtained, warmed contrast is gently injected through the microcatheter during the early segment of a digital run (microangiogram). This microangiography is routinely performed for superselective AVM embolizations. The microcatheter contrast is allowed to exit the AVM, and then a second contrast bolus is injected through the previously placed guiding catheter in the proximal carotid or vertebral artery during the last half of the same digital run (guiding catheter angiogram). As this digitally subtracted double injection run plays through on the monitor, it becomes easier to conceptualize the relationship between the microcatheterized portion of the nidus and the surrounding vascular anatomy visualized by the guiding catheter. It is the ability to see both the microcatheter angiogram and the guiding catheter angiogram at once in the same run that allows improved visualization of the angioarchitecture. This double injection technique can be illustrated on a single still image by “remasking” the microcatheter DSA run at the point of peak opacification and then allowing this inverted “remasked” DSA run (white) to play through to the guiding catheter angiogram (black). These two angiographic images will then be superimposed on each other in a single image (Figs. 1 and 2). This procedure inverts the microcatheter angiogram so it is white, whereas the superimposed guiding catheter angiogram remains black (Fig. 2). This technique offers improved differentiation of the contrast between the two angiography injections while simultaneously integrating the important relationships within the angioarchitecture. These relationships could become even more useful if this technique is combined with three-dimensional angiography, if available.


FIGURE 1. Digital subtraction angiography (DSA) showing a lateral view of an intracranial arteriovenous malformation (AVM) illustrating the double injection technique superimposed on a single frame. The microcatheter (white arrows) was navigated through a feeding artery in this AVM, and the guiding catheter (black arrow) is located in the proximal cervical internal carotid artery (ICA). In the same DSA run, a microcatheter angiogram (white arrowhead) is superimposed on the guiding catheter angiogram (black arrowheads) by remasking the microcatheter angiogram at the point of peak opacification, thereby inverting the black microangiogram to a white angiogram and then allowing the guiding catheter angiogram to play through until it is superimposed.

Once the angioarchitecture of the AVM is understood, the surgeon can use a wax pencil or surgical marking pen to identify and draw on each of the biplane DSA monitors: 1) the perimeter around the nidus, 2) the location of the draining veins, 3) the position of the microcatheter tip, and 4) possibly other adjacent feeding pedicles in two orthogonal projections (Fig. 3). As long as the table, patient, and biplanar image intensifiers remain stationary, these landmarks remain true throughout prolonged injections of the embolic agent through the microcatheter. Multiple new roadmaps can be obtained to subtract previously injected embolic material and to optimally visualize the most recently injected material (Fig. 3). This technique allows the surgeon to maintain confidence that the embolic agent is injected completely within the perimeter of the nidus and is not prematurely entering the draining veins or excessively refluxing on the microcatheter tip. Ethylene vinyl copolymer (Onyx-EV3; Irvine, CA) was used in all embolic procedures reported in this technical note.

RESULTS

When treating complex intracranial AVMs with embolization, understanding the relationships between the abnormal and normal vessels is the key to safely treating the nidus and minimizing complications. It is common practice to use only the superselective angiogram during embolic injections into an AVM (4). We propose an adjunctive technique to improve image interpretation during AVM embolization. Combining microangiography with guide-catheter angiography provides a better understanding of the specific angioarchitecture. Because ethylene vinyl copolymer (EV3) allows prolonged injection times, it is now possible to penetrate portions of the AVM nidus that are not visualized by the superselective microangiogram. In the absence of any other orienting structures, it is sometimes difficult to decipher nidus arteries from branches of normal vessels on the superselective angiogram, and this problem is resolved after the surgical marking pen demarcates the perimeter of the nidus. This perimeter demarcation helps prevent accidental embolization of normal vasculature (Fig. 3). The guidecatheter portion of the same run can determine with full confidence whether or not the microcatheter placement is within the targeted portion of the nidus. In many cases in our


IMPROVED IMAGE INTERPRETATION

A

B

A

B C

D

C FIGURE 2. A, DSA showing Towne’s oblique projection of microcatheter angiogram (black arrowheads). The microcatheter was placed in the left anterior inferior cerebellar artery (AICA) feeding pedicle to a cerebellar AVM. B, DSA lateral projection showing the same microcatheter as in A. C, the same DSA Towne’s oblique projection as A with combined simultaneous double injection. The microangiogram from A (black arrowheads) remains dark but has now been superimposed on the white guiding catheter angiogram from the right vertebral artery (white arrowheads). Note that the DSA microangiogram in this case was “remasked” as described in this technical report. D, the same DSA lateral projection as B with combined simultaneous double injection. The microangiogram from B (black arrowheads) remains dark but has now been superimposed on the white guiding catheter angiogram from the right vertebral artery (white arrowheads). Note that the DSA microangiogram in this case was “remasked” as described in this technical report.

FIGURE 3. A, anteroposterior and lateral DSA projections of a right-sided intracranial AVM showing the double injection technique after this AVM nidus was selectively catheterized. Note the previous (subtracted) embolic cast indicating this is the patient’s second staged embolization procedure (long white arrow). The microcatheter angiogram (white arrowheads) shows white vessels, and the guiding catheter angiogram (black arrowheads) shows black vessels. The important landmarks of this AVM are drawn on both the biplanar orthogonal monitors with a marking pen. The microcatheter (white arrows and drawn as a dotted line), microcatheter tip (circled with marking pen), AVM nidus perimeter (black arrows), and associated large draining vein (black arrow with white outline) are drawn on the DSA monitor so they can be continuously visualized even during new subtracted fluoroscopy (i.e., “blank roadmap”) and new DSA runs. It is critical that the image intensifier, patient, and table remain stationary to ensure accuracy of the monitor markings throughout the embolization. B, fluoroscopic images (same projections as A with important landmarks drawn on the monitors based on the guiding catheter angiogram from A) are shown. These land-

NEUROSURGERY

D

marks remain true despite the absence of intravascular contrast in the image. Note the previous embolic cast indicating that this is the patient’s second staged embolization procedure (long white arrow). Notice this previous embolic cast has been subtracted from A and C. C, newly fluoroscopic subtracted images (“blank roadmaps”) with the same projections as A and B are shown. At that time, the AVM was being injected with the new embolic agent (long black arrow). Because we have already outlined the important landmarks on the monitors, we are able to determine with confidence that the embolic agent is being injected within the perimeter of the AVM in both orthogonal projections. We can also confirm that there is no embolic material approaching the large draining vein (black arrow with white outline) and there is no reflux on the microcatheter tip. D, fluoroscopic images (same projections as A–C) showing the final postembolization result. Note the newly added embolic material from the Stage 2 embolization (difference between B and D). All of the embolic material can be clearly visualized entirely within the nidus perimeter and does not approach the draining vein (black arrow with white outline) or the microcatheter tip (circled with marking pen).


YAO ET AL.

practice, this imaging method has altered the selection of vessels targeted for embolization, allowing us to achieve more desirable outcomes. During prolonged microcatheter injections, it is sometimes difficult to decide when to stop, continue, or alter the force or speed of the injection because the surrounding angioarchitecture is not immediately clear in real-time during the embolization. Because this technique provides biplanar demarcation of key landmarks, including the AVM perimeter on both DSA monitors with a surgical marking pen, we have enhanced confidence that the vessels undergoing embolization are within the perimeter of the nidus, and not in draining veins, therefore improving real-time intraoperative decision making. A guiding catheter angiogram can also be performed intermittently during the embolization procedure, through either pauses in the injection or during the injection, to obtain some of the same information. However, this method does not allow for continuous real-time feedback throughout the embolization and is limited by the volume of contrast used. In our practice, the described techniques help to delineate the angioarchitecture during endovascular treatment of AVMs. Specifically, this technique has assisted us to: 1) appreciate the overall size and morphology of the nidus; 2) clearly visualize the safe limits of the embolic injection within the nidus perimeter; 3) clearly visualize draining patterns to help avoid premature venous embolization; 4) decipher small draining veins from arteries; 5) continuously monitor the location and status of the microcatheter tip to avoid microcatheter reflux; and 6) increase the confidence of the surgeon during prolonged embolic injections by integrating the superselective microcatheter and guiding catheter angiograms to improve understanding of the angioarchitecture, thus allowing for real-time assessment of the embolic injection.

CONCLUSION The additional information gained from this technique has been qualitatively useful in every case, and this technique has altered the treatment plan in all of our patients. The information obtained from the described technique cannot be obtained by computed tomographic angiography, magnetic resonance angiography, or conventional DSA alone; therefore, it is a very useful adjunct to standard microcatheter angiography alone.

REFERENCES 1. Kurata A, Miyasaka Y, Yada K, Kan S: Aneurysmography for visualizing large aneurysms. Neurosurgery 34:745–748, 1994. 2. Turjman F, Massoud TF, Viñuela F, Sayre JW, Guglielmi G, Duckwiler G: Aneurysms related to cerebral arteriovenous malformations: superselective angiographic assessment in 58 patients. AJNR Am J Neuroradiol 15:1601– 1605, 1994. 3. Valavanis A, Yas¸argil MG: The endovascular treatment of brain arteriovenous malformations. Adv Tech Stand Neurosurg 24:131–214, 1998.


4. Willinsky R, TerBrugge K , Montanera W, Wallace C, Aggarwal S: Micro-arteriovenous malformations of the brain: Superselective angiography in diagnosis and treatment. AJNR Am J Neuroradiol 13:325–330, 1992.

Acknowledgments The cerebrovascular group at Vanderbilt University Medical Center thanks, for her tireless work, Rebecca Dettorre, B.A., without whom this article would not be possible.

COMMENTS

Y

ao et al. describe a method of imaging cerebral arteriovenous malformations (AVMs) with a dual injection technique to gain a better understanding of the AVM angioarchitecture and to improve on the safety of endovascular embolization. This technique was probably developed with the availability of Onyx (eV3 Neuro, Irvine, CA) a relatively new liquid embolic agent recently approved for the presurgical embolization of cerebral AVMs. With its cohesive properties as opposed to the adhesive properties of n-butyl cyanoacrylate (Cordis Neurovascular, Miami Flakes, FL), the injection times for Onyx can be prolonged even over 1 hour. The primary determining factors that preclude further injection is the degree of reflux onto the tip of the microcatheter and the tortuosity of the arterial pedicle leading to the nidus. It is also a routine occurrence that Onyx will penetrate into an area of the nidus that was not visualized on the superselective angiogram. In view of this phenomenon, this technique can be useful to the interventionalist. The interventionalist needs to be able to identify in real time whether the liquid embolic is embolizing a draining vein, AVM nidus, or normal vasculature. The embolic material can be pushed through the nidus and can retrogradely fill a completely different arterial pedicle and, if continued, occlude the arterial supply of the normal parenchyma. This method of demarcating the perimeter of the AVM nidus could potentially reduce this likelihood. As the authors point out, there are other methods of determining the relationship of the Onyx cast to the remainder of the nidus. Because of the numerous inject and pause times associated with Onyx embolization, angiographic runs can be performed during the pause periods to generate this information. Overall, this is a technique developed by a center with significant experience in embolizing cerebral AVMs with Onyx and maybe useful to other interventionalists who are just starting their experience with this material. Henry H. Woo Stony Brook, New York

T

his is an interesting technique that may improve the safety of AVM embolization. The introduction of Onyx has changed our ability to perform extensive single injection embolization. As the authors describe, during these procedures, it can become difficult to differentiate a feeding artery from the nidus and draining vein. The extent of feeding artery reflux is also of significant importance. The embolic material becomes an issue for proper visualization of the microanatomy of large lesions. Although this technique may add some extra time to the procedure, particularly if a new projection is desired, it is probably time well spent to prevent nontarget embolization. Sean D. Lavine New York, New York


VASCULAR Case Report

COMBINED ENDOSCOPE-ASSISTED TRANSCLIVAL CLIPPING AND ENDOVASCULAR STENTING OF A BASILAR TRUNK ANEURYSM: CASE REPORT Jean A. Eloy, M.D. Department of Otolaryngology, Head and Neck Surgery, Mount Sinai School of Medicine, New York, New York

Andrea Carai, M.D. Department of Neurosurgery, Mount Sinai School of Medicine, New York, New York

Aman B. Patel, M.D. Department of Neurosurgery, Mount Sinai School of Medicine, New York, New York

Eric M. Genden, M.D. Department of Otolaryngology, Head and Neck Surgery, and Endoscopic Skull Base Surgery Center, Mount Sinai School of Medicine, New York, New York

Joshua B. Bederson, M.D. Department of Neurosurgery, and Endoscopic Skull Base Surgery Center, Mount Sinai School of Medicine, New York, New York Reprint requests: Jean A. Eloy, M.D., Department of Otolaryngology, Head and Neck Surgery, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1189, New York, NY 10029. Email: [email protected] Received, February 14, 2007. Accepted, July 14, 2007.

OBJECTIVE: We describe a patient with a mid-basilar aneurysm treated with combined endoscope-assisted transsphenoidal clipping and endovascular stenting. CLINICAL PRESENTATION: A 28-year-old woman was transferred to the cranial base surgery center with an acute Grade III subarachnoid hemorrhage. Cerebral angiography demonstrated a small basilar trunk aneurysm that was not amenable to acute endovascular treatment. INTERVENTION: The patient underwent sublabial transsphenoidal/transclival endoscope-assisted clipping of the aneurysm and subsequent stenting of the affected segment. The aneurysm was repaired with a low-profile Weck clip (Weck Closure Systems Research, Triangle Park, NC) that permitted a watertight closure of the clival dura using cardiac Medtronic U-clips (Medtronic, Inc., Minneapolis, MN). Postoperatively, the patient had no evidence of cerebrospinal fluid leakage. CONCLUSION: Watertight dural closure was possible due to the use of a lowprofile aneurysm clip that did not protrude through the dural defect, as well as selftying sutures. KEY WORDS: Basilar trunk aneurysm, Endoscopic transsphenoidal clipping, Endovascular stenting Neurosurgery 62[ONS Suppl 1]:ONSE142–ONSE144, 2008

B

asilar trunk aneurysms are amenable to treatment with endovascular techniques (9, 25, 28, 29), but surgery may be preferable in patients with acutely ruptured broadbased or fusiform lesions. Among the different surgical routes described to treat these aneurysms, the transsphenoidal/transclival approach can provide excellent visualization with minimal morbidity in selected cases. However, dural closure may be difficult to obtain, and cerebrospinal fluid (CSF) leakage and meningitis are major complications associated with this approach (32). We report a case of an acutely ruptured basilar trunk aneurysm successfully treated by combined sublabial transsphenoidal/transclival clipping and delayed endovascular stenting. At the time of surgery, a watertight circumferential dural closure was performed, with no evidence of postoperative CSF leakage.

CASE REPORT A 28-year-old woman was referred to our center with Hunt and Hess Grade III subarachnoid hemor-

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297101.34508.43

rhage (Fig. 1A). An external ventriculostomy was placed, resulting in mild improvement of her mental status. Angiography confirmed a 2.5 ⫻ 2.2 ⫻ 1.7–mm basilar trunk aneurysm projecting to the left and posteriorly (Fig. 1B). The small size and broad neck suggested that coil placement within the aneurysm would be difficult and coil prolapse might complicate endovascular treatment. Placement of a basilar artery stent has been reported (14), but we believed that using an open cell stent in the setting of this small aneurysm would not guarantee protection against coil prolapse. The need for antiplatelet agents after stent placement could also pose a risk of rebleeding if the aneurysm could not be coiled safely. In addition, the patient had hydrocephalus requiring external ventriculostomy, which might be complicated by antiplatelet medications. Therefore, stenting was thought to be suboptimal and surgical treatment was chosen. The patient underwent an extended sublabial approach to the sella and sphenoid.

Surgical Technique After lateralization of the inferior and middle turbinates, the sphenoid ostium was localized bilaterally and the sphenoid rostrum was excised. The optic


ELOY ET AL.

FIGURE 1. A, admission computed tomographic scan showing diffuse subarachnoid hemorrhage. B, preoperative angiography demonstrating a left midbasilar aneurysm and adjacent pontine perforating vessels.

nerve and carotid artery locations were delineated. The posterior sphenoid wall and clivus between the carotid arteries was removed with a high-speed drill with a diamond burr (Medtronic Midas Rex Legend, Fort Worth, TX) (Fig. 2A). The clival dura was subsequently opened (Fig. 2B). Venous bleeding from the clival venous plexus was controlled with Surgifoam (Ethicon, Inc., Somerville, NJ). Further dissection exposed the basilar artery with its perforating branches, the anterior cerebellar inferior arteries, and the aneurysm. The arachnoid was then opened, and control of the artery was obtained proximal and distal to the aneurysm (Fig. 2C). The aneurysm appeared to be an excentric fusiform dilation with a thrombosed large dome. The filling portion of the aneurysm seen

on angiography represented only part of the aneurysm neck (Fig. 2D). A curved aneurysm clip was placed across the neck, but the hub of the clip projected well into the sphenoid sinus impeding dural closure (Fig. 3A). The aneurysmal clip was, therefore, replaced with a Weck clip (Weck Closure Systems Research, Triangle Park, NC), which provided a low profile for closure (Fig. 3, B and C). The dural opening was repaired using fascia lata held in place by Weck clips (Fig. 3D) and then sutured to the dura with 18 interrupted self-tying sutures (Medtronic U-clip; Medtronic, Inc., Minneapolis, MN) (Fig. 4, A and B). The anterior sphenoid sinus wall was repaired with fat, septal cartilage, and fibrin glue (Tisseel; Baxter Healthcare Corp., Vienna, Austria). A Valsalva maneuver was performed and showed no CSF extravasation.

Postoperative Course The patient’s postoperative course was stable, without neurological deficits, although she developed vasospasm requiring hypertensive therapy and hydrocephalus requiring placement of a lumboperitoneal shunt. Postoperative angiography showed persistent fusiform dilation in the region of the aneurysm (Fig. 4C). A decision was made to place a Neuroform stent (Boston Scientific, Fremont, CA) across the base of the aneurysm to allow flow diversion and potentially reduce the risk of recurrence or regrowth of the aneurysm. The patient’s poststenting course was unremarkable and she was discharged home in good condition. A delayed angiogram 3 months later demonstrated a slight decrease in the size of the fusiform basilar artery compared to the previous angiogram (Fig. 4D). At the 3-month follow-up examination, she had made a full recovery; at 14 months, she returned to her normal preoperative baseline status.

DISCUSSION

FIGURE 2. Intraoperative photographs showing drilling of the clivus (A), dural opening above and below the aneurysm with intact arachnoid (B), basilar artery exposure proximal and distal to the aneurysm (C), and preparation of the aneurysm’s neck before clip application (D).


Endovascular techniques can be used successfully in basilar trunk aneurysms, but c o i l i n g a c u t e l y ru p t u re d broad-based aneurysms may pose a risk of coil prolapse or rerupture, particularly if the aneurysm is as small as the one we report here. Use of stent- and balloon-assisted coiling has further expanded indications for this technique (3, 27). One of the major issues in stent-assisted coiling is the need for long-term antiplatelet medications, especially in younger patients. Due to the small size and broad neck, coiling was not possible in this case without stenting. However, because of the patient’s recent subarachnoid hemorrhage and ventriculostomy, antiplatelet agents were considered ill-advised since the ability to safely coil is uncertain.


TRANSNASAL CLIPPING AND ENDOVASCULAR STENTING OF A BASILAR ANEURYSM

FIGURE 3. Intraoperative photographs showing the application of the aneurysm clip with the hub protruding into the sphenoid sinus (A), the application of the Weck clip (B), the low-profile Weck clip in place with the aneurysmal clip removed (C), and the fascia lata temporarily held in place with the Weck clips (D).

FIGURE 4. A, dural closure using self-tying U-clips. B, dural repair during Valsalva maneuver. C, postoperative angiogram revealing residual aneurysm filling. D, 3-month postoperative angiogram.

NEUROSURGERY

Numerous surgical routes exist to expose basilar aneurysms (1, 11–13, 17, 18, 20, 23, 30–33). Lawton et al. (26) reported the use of an extended o r b i t o z y g o m a t i c a p p ro a c h f o r aneurysms located at the upper twofifths of the basilar artery, a far lateral approach for those located at the lower two-fifths, and a transpetrosal route for midbasilar lesions. Kawase et al. (21) modified the transpetrosal approach in order to preserve hearing, and Aziz et al. (2) demonstrated such a route to be useful for aneurysms located higher than the floor of the acoustic canal. Nevertheless, the transpetrosal approaches are associated with a 15% chance of CSF leakage and variable hearing preservation (26). Moreover, technical refinement in the surgical armamentarium has expanded the possibilities offered by minimally invasive procedures. Kassam et al. (19) reported the treatment of a vertebral artery aneurysm using a purely transnasal endoscopic approach. A transclival approach to the basilar artery, as we report here, may provide a direct route to the pathology, potentially permitting wide exposure extending from the sella to the foramen magnum, with use of carotid arteries as the lateral limit of dissection (1, 11, 19, 32). In appropriately selected cases, this approach allows direct access to the basilar artery while minimizing cranial nerve and vascular manipulation. The use of angled 30-, 45-, and 70-degree endoscopes can further enhance visualization of the surgical field (4, 8, 10). Despite these potential advantages, difficulty in vascular control in a deep surgical field can represent a major drawback. In this case, proximal and distal control of the basilar artery and the ability to use a twohanded microscopic technique enhanced the safety of the approach. Adequate experience with endoscopy and expertise in nasal anatomy and dissections is also an important part of a successful outcome. Furthermore, depth of the surgical field, lim-


ELOY ET AL.

and atraumatic tissue approximation, eliminating the need for knot-tying. We sutured a fascia lata patch to the dural defect and used 16 U-clips and three Weck clips to obtain complete dural closure. Weck clips are easily applied to keep the patch in a desirable position during definitive closure (Fig. 5). The sphenoid sinus was also packed with layers of fascia, fat, and fibrin glue, and harvested septal cartilage was used to reconstruct the anterior sphenoid wall. No CFS leakage was encountered. We believe that meticulous dural suture was one factor contributing to the good result in this case and might further expand the choice of transclival approaches to the cranial base. FIGURE 5. A–C, postoperative computed tomographic scan reconstructions illustrating the extensive transsphenoidal approach and radial dural closure.

itations of the dural margin, and difficulty manipulating instruments make it difficult to obtain adequate dural closure. To reduce the rate of CSF leakage, a watertight closure of the dura may add a margin of safety in the use of the transsphenoidal/transclival approach for the management of basilar trunk aneurysms. In this case, treatment of the aneurysm was hindered by the hub of the aneurysm clip, which projected into the sphenoid sinus. A much better profile was obtained with a Weck clip positioned ideally to achieve optimal closure. However, the lesser closing strength of the Weck clip may have also resulted in less-than-ideal closure of the aneurysm. In an effort to improve dural closure, many different techniques have been proposed (5–7, 15, 16, 22, 24). Suturing techniques in transsphenoidal/transclival surgery provide excellent closure, but they are technically challenging. Knot-tying can be time-consuming or impossible in the deep and narrow surgical field (34). We addressed these problems by using selftying sutures (U-clips). The device is made of a Nitinol coil, which is a highly elastic alloy that contains thermoshape memory. The U-clip is placed via a conventional curved needle with the use of a standard needle holder and deployed by applying pressure with the needle holder to the release mechanism adjacent to the clip that separates the clip from the suture, thus allowing the Nitinol clip to return to a predetermined closedloop configuration. The Nitinol coil produces a relatively strong


CONCLUSION

The transclival approach to selected basilar trunk aneurysms, combined with endovascular stenting, may be considered in cases not amenable to a single endovascular or surgical approach. Dural closure after transsphenoidal/transclival basilar aneurysm clipping is facilitated by the use of a low-profile clip and self-tying sutures.

REFERENCES 1. Archer DJ, Young S, Uttley D: Basilar aneurysms: A new transclival approach via maxillotomy. J Neurosurg 67:54–58, 1987. 2. Aziz KM, van Loveren HR, Tew JM, Chicoine MR: The Kawase approach to retrosellar and upper clival basilar aneurysms. Neurosurgery 44:1225–1236, 1999. 3. Benitez RP, Silva MT, Klem J, Veznedaroglu E, Rosenwasser RH: Endovascular occlusion of wide-necked aneurysms with a new intracranial microstent (Neuroform) and detachable coils. Neurosurgery 54:1359–1368, 2004. 4. Cappabianca P, Cavallo LM, de Divitiis E: Endoscopic endonasal transsphenoidal surgery. Neurosurgery 55:933–941, 2004. 5. Cappabianca P, Cavallo LM, Esposito F, Valente V, De Divitiis E: Sellar repair in endoscopic endonasal transsphenoidal surgery: Results of 170 cases. Neurosurgery 51:1365–1372, 2002. 6. Cappabianca P, Cavallo LM, Mariniello G, de Divitiis O, Romero AD, de Divitiis E: Easy sellar reconstruction in endoscopic endonasal transsphenoidal surgery with polyester-silicone dural substitute and fibrin glue: Technical note. Neurosurgery 49:473–476, 2001. 7. Cappabianca P, Cavallo LM, Valente V, Romano I, D’Enza AI, Esposito F, de Divitiis E: Sellar repair with fibrin sealant and collagen fleece after endoscopic endonasal transsphenoidal surgery. Surg Neurol 62:227–233, 2004. 8. Cappabianca P, de Divitiis E: Endoscopy and transsphenoidal surgery. Neurosurgery 54:1043–1050, 2004.


TRANSNASAL CLIPPING AND ENDOVASCULAR STENTING OF A BASILAR ANEURYSM

9. CARAT investigators: Rates of delayed rebleeding from intracranial aneurysms are low after surgical and endovascular treatment. Stroke 37:1437–1442, 2006. 10. Catapano D, Sloffer CA, Frank G, Pasquini E, D’Angelo VA, Lanzino G: Comparison between the microscope and endoscope in the direct endonasal extended transsphenoidal approach: Anatomical study. J Neurosurg 104:419–425, 2006. 11. Crockard HA, Koksel T, Watkin N: Transoral transclival clipping of anterior inferior cerebellar artery aneurysm using new rotating applier. Technical note. J Neurosurg 75:483–485, 1991. 12. David CA, Vishteh AG, Spetzler RF, Lemole M, Lawton MT, Partovi S: Late angiographic follow-up review of surgically treated aneurysms. J Neurosurg 91:396–401, 1999. 13. Day JD, Fukushima T, Giannotta SL: Cranial base approaches to posterior circulation aneurysms. J Neurosurg 87:544–554, 1997. 14. Fiorella D, Albuquerque FC, Deshmukh VR, Woo HH, Rasmussen PA, Masaryk TJ, McDougall CG: Endovascular reconstruction with the Neuroform stent as monotherapy for the treatment of uncoilable intradural pseudoaneurysms. Neurosurgery 59:291–300, 2006. 15. Freidberg SR, Hybels RL, Bohigian RK: Closure of cerebrospinal fluid leakage after transsphenoidal surgery: Technical note. Neurosurgery 35:159–160, 1994. 16. Guity A, Young PH: A new technique for closure of the dura following transsphenoidal and transclival operations. Technical note. J Neurosurg 72:824–828, 1990. 17. Hamel W, Grzyska U, Westphal M, Kehler U: Surgical treatment of a basilar perforator aneurysm not accessible to endovascular treatment. Acta Neurochir (Wien) 147:1283–1286, 2005. 18. Heros RC: Lateral suboccipital approach for vertebral and vertebrobasilar artery lesions. J Neurosurg 64:559–562, 1986. 19. Kassam AB, Mintz AH, Gardner PA, Horowitz MB, Carrau RL, Snyderman CH: The expanded endonasal approach for an endoscopic transnasal clipping and aneurysmorrhaphy of a large vertebral artery aneurysm: Technical case report. Neurosurgery 59 [Suppl]:ONSE162–ONSE165, 2006. 20. Kato Y, Sano H, Behari S, Kumar S, Nagahisa S, Iwata S, Kanno T: Surgical clipping of basilar aneurysms: Relationship between the different approaches and the surgical corridors. Minim Invasive Neurosurg 45:142–145, 2002. 21. Kawase T, Toya S, Shiobara R, Mine T: Transpetrosal approach for aneurysms of the lower basilar artery. J Neurosurg 63:857–861, 1985. 22. Kitano M, Taneda M: Subdural patch graft technique for watertight closure of large dural defects in extended transsphenoidal surgery. Neurosurgery 54:653–661, 2004. 23. Krisht AF, Kadri PA: Surgical clipping of complex basilar apex aneurysms: A strategy for successful outcome using the pretemporal transzygomatic transcavernous approach. Neurosurgery 56 [Suppl]:261–273, 2005. 24. Kumar A, Maartens NF, Kaye AH: Evaluation of the use of BioGlue((R)) in neurosurgical procedures. J Clin Neurosci 10:661–664, 2003. 25. Lanzino G, Fraser K, Kanaan Y, Wagenbach A: Treatment of ruptured intracranial aneurysms since the International Subarachnoid Aneurysm Trial: Practice utilizing clip ligation and coil embolization as individual or complementary therapies. J Neurosurg 104:344–349, 2006. 26. Lawton MT, Daspit CP, Spetzler RF: Technical aspects and recent trends in the management of large and giant midbasilar artery aneurysms. Neurosurgery 41:513–521, 1997. 27. Lee YJ, Kim DJ, Suh SH, Lee SK, Kim J, Kim DI: Stent-assisted coil embolization of intracranial wide-necked aneurysms. Neuroradiology 47:680–689, 2005. 28. Molyneux AJ, Kerr RS, Stratton I, Sandercock P, Clarke M, Shrimpton J, Holman R, International Subarachnoid Aneurysm Trial (ISAT) Collaborative Group: International Subarachnoid Aneurysm Trial (ISAT) of neurosurgical clipping versus endovascular coiling in 2143 patients with ruptured intracranial aneurysms: A randomised trial. Lancet 360:1267–1274, 2002. 29. Molyneux AJ, Kerr RS, Yu LM, Clarke M, Sneade M, Yarnold JA, Sandercock P, International Subarachnoid Aneurysm Trial (ISAT) Collaborative Group: 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, 2005. 30. Motoyama Y, Ohnishi H, Koshimae N, Kanemoto Y, Kim YJ, Yamada T, Kobitsu K: Direct clipping of a large basilar trunk aneurysm via the posterior

NEUROSURGERY

31.

32.

33.

34.

petrosal (extended retrolabyrinthine presigmoid) approach—case report. Neurol Med Chir (Tokyo) 40:632–636, 2000. Nutik SL: Pterional craniotomy via a transcavernous approach for the treatment of low-lying distal basilar artery aneurysms. J Neurosurg 89:921–926, 1998. Ogilvy CS, Barker FG, Joseph MP, Cheney ML, Swearingen B, Crowell RM: Transfacial transclival approach for midline posterior circulation aneurysms. Neurosurgery 39:736–742, 1996. Terasaka S, Itamoto K, Houkin K: Basilar trunk aneurysm surgically treated with anterior petrosectomy and external carotid artery-to-posterior cerebral artery bypass: Technical note. Neurosurgery 51:1083–1087, 2002. Vanaclocha V, Sáiz N, Panta F: Repair of dural defects in awkward areastechnical note. Acta Neurochir (Wien) 140:615–618, 1998.

COMMENTS

I

n the kingdom of Kalmon Post, one of the true leaders of the transsphenoidal surgery, a new creature, i.e., the endoscopic transnasal approach to the cranial base, has emerged, rising up from the cooperation between a neurosurgeon (Joshua Bederson) and an otorhinolaryngologist with endoscopic experience (Eric Genden). The results of such a cooperation are clearly visible in this technical case report concerning the endoscopeassisted transclival clipping of a basilar trunk aneurysm. This study points out two topics of tremendous interest: The first one concerns the surgical approach the authors used. The transnasal transclival approach offers a midline trajectory that minimizes manipulation of the brainstem and cranial nerves. None of the surgical routes, as a matter of fact, after dural opening allows obtainment of a lesser distance from the neurovascular structures of the ventral surface of the brainstem. The second topic concerns the use of the endoscope. Despite their endoscope-assisted and not purely endoscopic technique, in the sense that they make use of the transsphenoidal retractor, it cannot be underestimated that the use of the endoscope is critical in illustrating the area where the light of the microscope cannot reach. The advent of the endoscope has renewed interest for the anterior approaches to the clivus and to the craniovertebral junction in the treatment of either extradural or intradural lesions, including vascular lesions, as confirmed by the case illustrated herein. High risk of cerebrospinal fluid leak and the deep and narrow surgical corridor are ancient problems that somehow had restricted the use of anterior approaches to the clivus. They should be nowadays revised considering advantages provided by the endoscope nowadays, which, regardless of the depth of the surgical corridor, brings the eye of the surgeon to the surgical field. Furthermore, new materials and/or techniques for cranial base reconstruction are gradually reducing the rate of postoperative cerebrospinal fluid leaks in endonasal approaches to the cranial base. As recently stated in a publication by the Pittsburgh group (1), the management of vascular lesions via an extended endonasal approach represents the last step of the learning curve, requiring a higher level of experience and technical skill. Although it will be useful to expand that limitation, with growing interest among those who are involved in endonasal surgery to the cranial base, such approaches for vascular lesions should be performed in well-selected patients by surgeons with proper endoscopic training, after adequate experience with surgery limited to the sellar area. Paolo Cappabianca Naples, Italy

1. Snyderman C, Kassam A, Carrau R, Mintz A, Gardner P, Prevedello DM: Acquisition of surgical skills for endonasal skull base surgery: A training program. Laryngoscope 117:699–705, 2007.


ELOY ET AL.

T

his case report is a nice contribution to the acute discussion on extending the endoscopic transnasal approach over the now widely accepted pituitary surgery. Several reports deal with intradural frontobasal and clival tumors, such as meningiomas. But there are only a few attempts of direct basilar aneurysm surgery despite the straight way to almost the whole basilar extension via the transnasal/transclival route. Besides the problems of managing intraoperative hemorrhage via this small, deep approach, one of the main challenges is a reliable dura closure; cerebrospinal fluid leaks after wide dura opening are the most frequent and most difficult complications of intradural surgery via the transnasal route. The authors show a successful technique with a straight approach through the posterior wall of the sphenoid sinus; in this patient, surgery was certainly facilitated by the large sellar-type sphenoidal sinus with a relatively thin posterior wall. Problems with the cavernous sinus were apparently avoided by leaving this area above the approach path, resulting in only minor venous postclival problems. A watertight dural closure could be obtained by using U-clips (Medtronic, Inc., Minneapolis, MN) as self-tying sutures; good results are also reported by using mini vascular clips from vascular surgery; the advantage of the U-clips seems to be the more effective tissue approximation. This article shows the practicability of this approach for such (certainly not too large, but large-neck) basilar aneurysms even after acute subarachnoid hemorrhage. The use of endoscopy-assisted surgery despite this endoscopic only (“endoscopy controlled”) technique in the transnasal-transsphenoidal route is questionable. With the exception of perhaps venous hemostasis of the clival venous plexus by Surgifoam (Ethicon, Inc., Somerville, NJ) and/or the dura closure (where the endoscope may come in conflict with the needle holder for the U-clips), the microscope does not contribute much. Using actual endoscopic techniques (free transnasal or transnasal-assisted with a small speculum), this wide surgical exposure by a large opening of the anterior and posterior sphenoid sinus walls could be achieved with avoidance of the more invasive sublabial route and with about a 30% complication/discomfort rate (septum perforations up to >10%, anosmia 5-6%, par-/anesthesia up to 28%, and epistaxis >10%). Michael R. Gaab Hannover, Germany

T

his case report is provocative because it demonstrates how minimally invasive techniques and current technology might reopen a surgical route long considered too morbid. The authors treated a broadbased basilar trunk aneurysm with transsphenoidal-transclival clipping, endoscopic assistance, dural repair with self-closing U-clips, and postoperative stenting of the parent artery, with a good neurological outcome. Conventional microsurgical approaches would have been far more invasive, with either a transpetrosal approach, an extended orbitozygomatic approach, or an extended retrosigmoid approach; I would have selected one of the latter two. Endoscopy addressed the problem of limited visualization, the Weck clip (Weck Closure Systems Research, Triangle Park, NC) solved the problem of the high-profile aneurysm clip, and U-clips solved the problem of dural repair. The success of these innovative techniques should make us reconsider our aversion to transclival approaches to aneurysms and to minimalism in aneurysm surgery generally. However, it is also important to recognize the pitfalls of such an approach. Weck clips are designed for single applications and are difficult to remove or reposition, the angiographic outcome in this case was imperfect, requiring a stent to improve the result, and minimalistic approaches undoubtedly impair the neurosur-


geon’s ability to deal with an intraoperative rupture or other complications. Therefore, we should be cautious and deliberate in our application of the techniques advanced in this case report, until we have carefully assessed their safety relative to traditional microsurgical techniques. Michael T. Lawton San Francisco, California

T

he authors describe a clever technique to deal with an aneurysm that is known to be of high risk from both an endovascular and surgical standpoint. By combining endovascular techniques with a direct surgical approach, they have provided a nice treatment that has a high chance of being durable in the long term. This question of durability remains open to further consideration, and the authors plan long-term follow-up of this individual. The big problem in the past for direct surgery of fusiform aneurysms has been the risk of recurrence even if an arterial lumen can be reconstructed using aneurysm clips. This report adds further information that perhaps an extravascular and intravascular approach to fusiform aneurysms may be a viable strategy for dealing with the defective arterial wall that typically occurs in this type of aneurysm. The authors are to be commended for their efforts. Christopher S. Ogilvy Boston, Massachusetts

E

loy et al. report the treatment of a ruptured basilar trunk fusiform or dissecting aneurysm via transsphenoidal/transclival endoscopicassisted partial clipping followed by Neuroform stent (Boston Scientific, Fremont, CA) placement in the affected area. The patient had an excellent outcome. As the authors mentioned, stent-assisted coil embolization is not usually considered for ruptured intracranial aneurysms because of the need for aggressive antiplatelet therapy. However, Fiorella et al. (1) have reported two cases of dissecting aneurysms that were successfully treated with Neuroform stent placement, one in the setting of acute subarachnoid hemorrhage and the other in a subacute stage. Additionally, in our own clinical experience, stent placement or stent-assisted coil embolization of ruptured aneurysms in an acute stage may be accomplished safely by using an intravenous half-dose of eptifibatide followed by a loading dose of aspirin and clopidogrel. Although stenting for the treatment of ruptured aneurysms has not been generally accepted, this approach may be one option for surgically challenging ruptured aneurysms. Regarding the surgical approach, we do not agree that the transsphenoidal/transclival approach is the best option for difficult basilar trunk aneurysms such as the one shown in the present case report, given the very small operative field and great difficulty in obtaining proximal flow control. In our opinion, a combined petrosal approach would provide a better option, albeit a more invasive one, than the technique in this report. Endovascular techniques will probably replace surgery for this type of aneurysm; the operation is high risk as is the use of Weck clips on aneurysms. Junichi Yamamoto L. Nelson Hopkins Buffalo, New York

1. Fiorella D, Albuquerque FC, Deshmukh VR, Woo HH, Rasmussen PA, Masaryk TJ, McDougall CG: Endovascular reconstruction with the Neuroform stent as monotherapy for the treatment of uncoilable intradural pseudoaneurysms. Neurosurgery 59:291–300, 2006.


SPINE Anatomic Technique

MODIFICATIONS OF THE TRANSORAL APPROACH TO THE CRANIOVERTEBRAL JUNCTION: ANATOMIC STUDY AND CLINICAL CORRELATIONS A. Samy Youssef, M.D., Ph.D. Department of Neurosurgery, University of South Florida, Tampa, Florida

Bernard Guiot, M.D. Department of Neurosurgery, University of South Florida, Tampa, Florida

Keith Black, M.D. Department of Neurosurgery, Maxine Dunst Brain Tumor Program, Cedars Sinai Medical Center, Los Angeles, California

Andrew E. Sloan, M.D. Department of Neurosurgery, and Case Comprehensive Cancer Center, University Hospitals-Case Medical Center, Cleveland, Ohio Reprint requests: Andrew E. Sloan, M.D., Department of Neurosurgery, University Hospitals Case Medical Center, Hanna House 5th Floor, 1110 Euclid Avenue, Cleveland, OH 44106. Email: [email protected] Received, June 6, 2007. Accepted, October 4, 2007.

OBJECTIVE: This study was designed to more precisely characterize the changes in exposure achieved by modifying the standard transoral approach by sequential mandibulotomy and mandibuloglossotomy with or without palatotomy. METHODS: A series of cadaveric dissections was performed and the operative distance and angle of exposure in both axial and sagittal planes was evaluated for each approach, with and without palatotomy. Intraoperative measurements were made in patients undergoing transoral approaches to assess the validity of the anatomic model. The use of this model was then assessed by a retrospective analysis of a group of 19 patients operated on through transoral approaches between 1991 and 2006. RESULTS: The simple transoral approach exposed the region from the lower third of the clivus to the middle of the C2 vertebral body at an operative distance of 12.9 1.0 cm from the dura. The axial and sagittal angles of exposure were 39.4 3.5 degrees and 36.8 3.5 degrees, respectively. Mandibulotomy significantly increased the sagittal exposure to 59.0 1.0 degrees (P 0.001), exposing the area from the midclivus to the C2–C3 interspace while simultaneously increasing the axial angle of exposure to 51.9 7.4 degrees (P 0.01) and decreasing the operative distance to the dura to 10.7 1.7 cm (P 0.05). Mandibuloglossotomy augmented sagittal exposure to 85.3 0.3 degrees (P 0.001), revealing the region between the upper one-third of the clivus and the C4–C5 interspace (P 0.001) while decreasing the operative distance to the dura to 8.7 0.3 cm (P 0.05). Palatotomy significantly increased the rostral exposure achieved by each approach by 8.5 to 12.3 degrees (P 0.01) without altering caudal or axial exposure or the operative distance. CONCLUSION: The cadaveric data correlated well with intraoperative measurements and the need for modifications of the transoral approach in 15 of the 16 adult patients (93.8%). Pediatric patients, patients with limited mouth opening, elevated craniovertebral junctions, and particularly deep lesions required more extensive exposure. This analysis may be useful for determining the optimal approach for patients undergoing transoral surgery. KEY WORDS: Craniovertebral junction, Glossotomy, Mandibulotomy, Palatotomy, Transoral approach Neurosurgery 62[ONS Suppl 1]:ONS145–ONS155, 2008

T

he transoral approach, first described by Kanavel (19), is the simplest and most commonly used approach for ventral, medial extradural lesions of the craniovertebral junction. Numerous modifications enhancing the transoral exposure have been described, including many variations of the medial labiomandibular glossotomy approach, which consists of mandibulotomy (division of the mandible) (4, 11, 32) and mandibuloglossotomy (division of the mandible and tongue) (3, 21, 25, 30, 36), with or without division of the hard palate (palatotomy). Despite the popularity of the transoral route and its modifications, however, there have been no studies system-

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297065.37186.84

atically comparing the exposures obtained with each technique. Specific indications are poorly defined and frequently contradictory and based largely on small case series (Table 1). Because each sequential modification increases the surgical complexity and the risk of functional and cosmetic complications, the surgeon should choose the simplest procedure that provides adequate surgical exposure. This critical decision should be based on objective data defining the exposure achieved by each modification and confirmed intraoperatively. Ideally, this would minimize intraoperative modification resulting from inadequate exposure (6, 24, 34, 35).


YOUSSEF ET AL.

TABLE 1. Exposure achieved with various modifications of the transoral approach as reported in the literaturea Limits of exposure Approach Transoral

Transoral + palatotomy

Transoral + mandibulotomy

No. of cases

Rostral

Caudal

Menezes et al., 1980 (28)

9

Low clivus

C3

Gilsbach and Eggert, 1983 (13)

10

Lower one-third clivus

C2

Pasztor et al., 1984 (31)

8

Upper one-third clivus

C4

Crockard et al., 1986 (9)

14

NS

C2

Menezes, 1989 (27)

140

Midclivus

C2–C3

Harkey et al., 1990 (16)

4

Low clivus

C1–C2

Cocke et al., 1990 (6)

R

Lower clivus

C2

Crockard and Sen, 1991 (10)

4


C3

Merwin et al., 1991 (29)

16

Low clivus

C2

Crockard, 1993 (8)

R


C2

Shaha et al., 1993 (35)

5

Foramen magnum

C2

Bouthellier et al., 1994 (5)

R

Low clivus

C2

Alonso et al., 1971 (1)

1

Sphenoid sinus

NS

Hayakawa et al., 1981 (17)

3

NS

NS

Crockard, 1985 (10)

R

Sphenoid sinus

C4

Kennedy et al., 1986 (20)

1

Upper clivus

NS


3

Sphenoid sinus

C2–C3

Beals et al., 1992 (3)

1

Midclivus

C1–C2

Lalwani et al., 1992 (24)

R

Sphenoid sinus

C2

Delgado et al., 1981 (11)

R

Lower clivus

C2

Cocke et al., 1990 (6)

1

Upper clivus

C4

Transoral + mandibulotomy +


1

Sella

C2

palatotomy

Grime et al., 1991 (14)

R

Upper clivus

C4

Transoral + mandibulotomy +


1

Lower clivus

C4

glossotomy

Nagib et al., 1990 (30)

1

Midclivus

C3

Shaha et al., 1993 (35)

1

Lower clivus

C2

Wood et al., 1980 (38)

2

Sella

C4

Transoral + mandibulotomy + glossotomy + palatotomy

a

Series (ref. no.)

Arbit and Patterson, 1981 (2)

1

NS

C5

Wood et al., 1990 (39)

9

Sella

C4–C5

Bouthellier et al., 1994 (5)

R

Sphenoid sinus

C4– C5

R, review; NS, not specified. References listed in chronological order by approach.

This study was undertaken to precisely characterize the exposure achieved by the standard transoral approach and successive modifications with mandibulotomy and mandibuloglossotomy. A series of cadaveric dissections was performed and the operative distance and angle of exposure in both axial and sagittal planes were evaluated for each approach, with and without palatotomy. Intraoperative measurements were made in patients undergoing transoral approaches to assess the validity of the anatomic model. The use of this model was then assessed by retrospective analysis of a group of 19 patients who underwent operation through transoral approaches between 1991 and 2006.


MATERIALS AND METHODS Cadaveric Dissections Cadaveric dissections were performed in seven adult specimens to establish the extent of exposure achieved with various modifications of the transoral approach. A standard transoral approach was performed in each cadaver. The specimens were then sequentially modified by mandibulotomy and mandibuloglossotomy to evaluate the change in each successive exposure. The resection of the hard palate was performed only on the right side of each head so that the contribution of palatotomy to each exposure could be assessed. The operative distance was defined as the distance between the retractor (which represents the closest position of the surgeon’s hand) and the


TRANSORAL MODIFICATIONS: ANATOMIC AND CLINICAL CORRELATIONS

operative site (C1 tubercle or dura) and was measured with calipers. The angles of exposure in axial and sagittal planes were measured from vertex and lateral radiographs with a goniometer using spinal needles placed at the superior, inferior, and lateral margins of exposure. Because a right hemipalatotomy was performed in all cadavers, the needle was placed just to the left of midline for determination of the angles of exposure without palatotomy (Fig. 1, A, C, and E) and then moved to the right of midline to determine the sagittal angles with palatotomy (Fig. 1, B, D, and F). All measurements were obtained in triplicate and collected in database format (Microsoft Excel 5.0; Microsoft Inc., Redmond, WA). Statistical significance was determined using the Student’s t test (P 0.05) (Table 2). A transoral approach to the craniovertebral junction was performed in the standard fashion as described by Crockard (8). After opening the mouth, the Crockard self-retaining retractor system (Codman and Shurtleff; Johnson and Johnson Co., Randolph, MA) was inserted into the mouth, and the tongue retractor blade was used to maximally retract the tongue inferiorly. A vertical midline incision in the pharyngeal wall was made over the C1 tubercle. A soft palate incision was also made from the posterior margin of the palatal bone to the superior margin of the uvula, curving gently inferiorly to the right of this structure. The pharyngeal retractor blades were then placed to maximize lateral retraction of the pharynx and rostral retraction of the soft palate. A right hemipalatotomy extending to the transverse palatine suture was made in each specimen with a high-speed drill (AM-8; Midas Rex, Fort Worth, TX). The left side of the hard palate remained intact. The highspeed drill was used to create a trough 15 mm wide in the arch of C1. The odontoid was then hollowed out until only a transparent cortical rim remained. A 2-mm Kerrison rongeur (Codman/Johnson & Johnson, Raynham, MA) and curettes were then used to remove the remaining odontoid peg, and the alar, apical, and transverse ligaments were sharply transected. Bony resection was extended as far rostrally into the clivus and caudally into the cervical vertebrae as the transoral exposure would allow, maintaining a 15-mm bony trough for consistency. Mandibulotomy was performed by incising the lip in the midline through the gingival mucosa to the mandible and then inferiorly to the hyoid as described by Wood et al. (38). A subperiosteal flap was then developed along the midline of the mandible and a stairstep mandibulotomy between the central incisors was made with the oscillating saw. It was not necessary to remove an incisor in any of the specimens. The mucosa in the floor of the mouth was then divided in the midline beneath the tongue. The two halves of the mandible were retracted laterally, and the tongue was depressed into the space created by relaxation of the floor of the mouth. Resection of the clivus and vertebral bodies was extended superiorly and inferiorly to the limits of the exposure. Mandibuloglossotomy extended the exposure by dividing the tongue in the midline raphae from the tip to the circumvallate papilla. The entire floor of the mouth was then divided sharply in the midline. Retractors were used to further separate the mandible and to depress the tongue inferiorly. Bony resection was again extended to the limits of the exposure.

Clinical Experience Intraoperative measurements were made in the same manner as the cadaveric dissections in five adult patients undergoing transoral surgery. The operative distances to the C1 tubercle and the dura were measured, and the sagittal and axial angles of exposure were determined from lateral and vertex skull radiographs, with use of spinal needles to identify the limits of exposure.

NEUROSURGERY

Retrospective Case Study Operative records and radiographic studies were reviewed for 19 consecutive patients who underwent transoral surgery for craniovertebral junction lesions between 1991 and 2006 (Table 3). The pre- and postoperative radiological studies were reviewed, and the approach

A

B

C

D

E

F

FIGURE 1. Lateral skull films showing cadaveric dissections. After unilateral dissection of the right hard palate to the transverse palatine suture, a Crockard retractor and spinal needles were placed to illustrate the rostral and caudal extent of various exposures. A, transoral approach; B, transoral approach with palatotomy; C, transoral approach with mandibulotomy; D, transoral approach with mandibulotomy and palatotomy; E, transoral approach with mandibuloglossotomy; and F, transoral approach with mandibuloglossotomy and palatotomy.


YOUSSEF ET AL.

TABLE 2. Exposure achieved with various modifications of the transoral approach in cadaveric models (n ⴝ 7) and in patients (n ⴝ 5) undergoing transoral surgerya Approach Cadavers

Transoral

(n 7 adults)

Vertebral body (range in cm)

Dura (range in cm)

Palatal split

11.6 1.0

12.9 1.0

Sagittal plane Axial plane (range in (range in degrees) degrees) 36.8 3.5

39.4 3.5

Exposure Cephalad One-third clivus

Caudad C2

(10.5–13.2)

(11.7–14.5)

(26–42)

(34–41.5)

Transmandibular

9.4 1.1b

10.7 1.7b

45.3 1.9

51.9 7.4b

(7–11.7)

(8.9–13)

(33–47)

(43–58)

Mid-two-thirds clivus

IS

Transmandibular

7.3 0.6c

8.7 0.3c

59 1.0b

53.9 5.5

Mid-two-thirds clivus

C4–C5

(6.5–8.1)

(8.4–9.2)

(58–60)

(46–59)

glossotomy

Two-thirds clivus One-third midclivus

Sphenoid sinus

C2 C2–C3

IS

70.9 5.2

C4–C5

(66–77)

IS

85.3 0.3c (85–86) 97.6 1.4 (97–100) Patients

Transoral

(n 5 adults)

12.1 0.2

13.7 0.5

(12.0–12.3)

(13.8–14.1)

38.8 0.9

39.5 4.5

(37.5–40)

(34.0–46.0)

One-third clivus

C1–C2 IS

IS interspace. Measurements were made as noted in “Materials and Methods.” Statistical significance of measurements by Student’s t test was denoted as follows: Differences between transoral approach and transoral mandibulotomy. b P 0.05. c P 0.05. a

suggested by the cadaveric model was then compared with the actual surgical approach used for gross total resection of the lesions.

RESULTS

was 39.4 3.5 degrees. Palatotomy increased rostral exposure to the midclivus and increased the sagittal angle of exposure to 45.3 1.9 degrees without changing the caudal exposure, axial exposure, or operative distance.

Cadaveric Data

Mandibulotomy

Modification of the standard transoral approach with successive mandibulotomy and mandibuloglossotomy reduced operative distances and increased exposure in the axial and sagittal planes. The results of the cadaveric dissections have been summarized in Table 2 and shown diagrammatically in Figures 2, 3, and 4. There was a moderate amount of variation between cadavers, yet the relative change within each cadaver during successive exposures was consistent and reproducible. Because the sequential modifications were measured in each cadaver, each specimen served as its own internal control.

Mandibulotomy improved surgical exposure by allowing lateral retraction of the mandible. This increased mouth opening and allowed the surgeon to work inside the mouth (Fig. 6), thus decreasing the operative distance to the C1 tubercle and dura to 9.4 1.1 and 10.7 1.7 cm (P 0.01 and P 0.05, respectively). This operative distance was 2.2 cm less than the standard transoral approach. The sagittal exposure increased by 22.2 degrees to 59.0 1.0 degrees (P 0.001), exposing the region from the midclivus to the C2–C3 interspace (Fig. 2). The axial angle of exposure also increased by 12.5 degrees to 51.9 7.4 degrees (P 0.01) (Fig. 4). Palatal resection further increased rostral exposure to reveal the upper one-third of the clivus, thereby increasing sagittal exposure without changing operative distance, caudal exposure, or axial exposure.

Transoral Approach The standard transoral approach exposed the region from the lower one-third of the clivus to the middle of the C2 vertebral body. Operative distance was limited to a plane external to the incisors by the patient’s teeth as well by the retractor (Fig. 5). The mean operative distance was 11.6 1.0 cm from the pharyngeal wall over the C1 tubercle and 12.9 1.0 cm to the dura after removal of the odontoid. The angle of exposure in the sagittal planes was 36.8 3.5 degrees, and the axial exposure


Mandibuloglossotomy Mandibuloglossotomy opened the entire floor of the mouth and reduced the operative distance to 7.3 0.6 cm from the vertebral body and 8.7 0.3 cm from the dura, 2.1 and 2.0 cm closer, respectively, compared with mandibulotomy alone (P 0.05). Glossotomy also increased the sagittal angle of expo-



TABLE 3. Exposure achieved with various modifications of the transoral approach as reported in the literaturea Patient no.

Age (yr)/sex

Diagnosis

Location

Operation

Complications

Outcome

1

26/F

Bas. invag.; Down syndrome

Low clivus–C2

TO

None

Good

2

32/M

Bas. invag.; C2 Fx

Good

3

33/M

Bas. invag.; rheumatoid arthritis

4

41/F

Chondrosarcoma

5

63/F

6 7

FM–C2

TO

None

Low clivus–C2

TO

Wound infection

Good

Upper clivus–C3

TO + P+ M+G

VPI

Death


Low clivus–C2

TO

None

Good

28/M

Bas. invag.; congenital

Low clivus–C2

TO

None

Good

41/F


Low clivus–C2

TO

None

Good

8

35/M

Chordoma

Entire clivus–FM

TO + M+G + LFM

None

Good

9

28/M

Bas. invag.; os odontoidium

FM–C2

TO

None

Fair

10

51/F

Bas. invag.; C2 Fx

Fair

11

19/M

Chordoma

12

9/M

13

FM–C2

TO

CSF leak

Low clivus–C2

TO+P

VPI

Fair

Bas. invag.; Klippel Feil

FM–C2

TO; TO + P + M

VPI

Good

19/M


Low clivus–C2–C3 IS

TO + P + M

Sepsis

Fair

14

31/M

Bas. invag.; Klippel Feil

Low clivus–C2

TO

None

Good

15

24/M

Chordoma

Low clivus–C1

TO

None

Good

16

43/M


Low clivus–C2

TO; TO + P + M

VPI

Death

17

10/F

Chordoma

Entire clivus– FM

TO + M+G + LFM

None

Good

Midclivus–C3

TO + P + M

VPI

Good

FM–C2

TO

None

Good

18

3/M

Chordoma

19

17/F


a Bas. invag., basilar invagination; Fx, fracture; FM, foramen magnum; TO, transoral approach; P, palatotomy; M, mandibulotomy; M+G, mandibuloglossotomy; LFM, LeFort I maxillotomy with midface degloving; VPI, velopharyngeal insufficiency; CSF, cerebrospinal fluid. Patients are listed in order of diagnosis and increasing complexity of surgical approach.

sure by 26.3 degrees to 85.3 0.3 degrees (P 0.001) (Fig. 4). This increase in sagittal exposure gave access to the region from the upper one-third of the clivus to the C4–C5 interspace (Fig. 2). The axial angle of exposure increased only 2 degrees to 53.9 5.5 degrees (P 0.05) (Fig. 4). Subsequent palatotomy extended the rostral exposure to the sphenoid sinus; operative distance, caudal and axial exposures were unchanged.

Clinical Experience The intraoperative measurements of operative distance and angle of exposure in axial and sagittal planes from patients undergoing transoral surgery correspond closely to the cadaveric data (Table 2). The operative distances and angles of exposure were well within the range of values obtained in the cadaveric specimens, as were the limits of the exposure.

Retrospective Case Study

FIGURE 2. Comparison showing sagittal exposure achieved by transoral approach, transoral + mandibulotomy, or transoral + mandibuloglossotomy.

NEUROSURGERY

Nineteen patients underwent transoral surgery between 1991 and 2006. The series included 11 males and eight females with a mean age of 31.7 years (range, 3–69 yr). There were eight patients who required modification. These included palatotomy (Patient 11); mandibulotomy and palatotomy (Patients 12, 13, 16, and 18); mandibuloglossotomy and palatotomy (Patient 4); or mandibuloglossotomy with a supplemental LeFort I maxillotomy (Patients 8 and 17). All three (100%) pediatric patients required modifications of the transoral approach. Two of the pediatric patients had extensive chordomas involving the clivus and craniovertebral junction; the third had severe basilar invagination with marked elevation of the craniovertebral junction. In these patients, the combination of limited mouth opening with large, deep lesions made modification of the transoral approach necessary for complete resection. In contrast, only five (31.3%) out of 16 adult patients required modification of the standard transoral approach.


YOUSSEF ET AL.

FIGURE 4. Illustration showing the increase in sagittal and axial exposure and decreased operative distances achieved by sequential mandibulotomy and mandibuloglossotomy in cadaveric dissections.

geal insufficiency manifested as swallowing difficulty and nasal regurgitation, which resolved after 3 months. Patient 17, a 10-year-old girl, had a massive clival chordoma involving the entire clivus down to C2 (Fig. 8). The anatomic data suggested that a simple transoral approach combined with a LeFort I maxillotomy would provide adequate exposure for resection of the lesion. However, intraoperative measurements confirmed that her limited mouth opening, the considerable ventral and lateral extent of the tumor, and severe brainstem compression necessitated a more extensive exposure. Gross total resection was achieved through a transoral approach modified by mandibuloglossotomy, palatotomy, and LeFort I maxillotomy. Although speech was mildly impaired in the immediate postoperative period, it returned to normal by postoperative Day 3. FIGURE 3. Illustration showing the change in sagittal exposure achieved by addition of palatotomy to transoral approach, transoral + mandibulotomy, or transoral + mandibuloglossotomy.

When the approach predicted by the cadaveric model was compared with the technique required for definitive resection, the cadaveric model correctly predicted the necessary approach in 15 of the 16 adult patients (93.8%) and 16 of the 19 patients overall (84.2%). In three patients, two of whom were children, however, intraoperative measurements and clinical judgment demonstrated the need for additional exposure. These are discussed subsequently.

DISCUSSION Cadaveric Model The transoral approach is the preferred midline approach for ventral craniovertebral junction extradural lesions (26). The tran-

Illustrative Cases Patient 12 was a 9-year-old boy with severe congenital basilar invagination with the dens protruding 16 mm above Chamberlain’s line into the posterior fossa (Fig. 7). The patient had a deep bony lesion, marked platybasia, an elevated craniovertebral junction, a small mouth, and a severe fixed cervical flexion deformity. Intraoperative measurements using spinal needles and fluoroscopy demonstrated that the standard transoral approach would not provide adequate exposure of the lesion as predicted by the model. Mandibuloglossotomy was required to achieve brainstem decompression. Postoperatively, the patient did well except for transient velopharyn-


FIGURE 5. Standard transoral approach. The surgeon’s hand is limited by teeth and retractor so that plane of resection is slightly outside plane of the incisors. Note that the surgeon is working at the very end of his instrument (a long bayoneted scalpel).



A

FIGURE 6. Transoral approach with mandibulotomy. Increased exposure allows the surgeon’s hand to work within the patient’s mouth, thus decreasing operative distance.

soral approach through the posterior pharyngeal wall was reserved for drainage of retropharyngeal abscesses until 1960s. Since its first description by Kanavel (19) in 1919, numerous modifications have been used to extend the transoral exposure to the skull base and upper cervical spine. Roux et al. (33) advocated splitting the lower lip and mandible for resection of tumors of the anterior tongue in 1839, and mandibuloglossotomy was described by Kocher (21) in 1911 for tumors of the posterior pharynx and the base of the tongue. Trotter (36) extended the exposure by splitting the tongue sagittally in the median raphae, creating the “median (anterior) translingual pharyngotomy approach,” which was re-introduced by Martin et al. (25) in 1961 as the “median labiomandibular glossotomy.” The palatotomy was first described by Preysing (32) in 1913 for resection of pituitary adenomas, and Alonso et al. (1) applied this technique to transoral surgery for resection of clival chordomas. Although numerous versions of the transoral approach have been described, the exposures achieved by different modifications have not been well defined, and there is little consensus regarding the precise exposure achieved by these modifications or their relative advantages (Table 1). Some have even suggested that the most important indication for a particular procedure was the surgeon’s familiarity with the approach rather than the anatomic characteristics of the lesion itself (18, 26); this paucity of data impedes a systematic approach to the treatment of complex clival and craniovertebral junction lesions.

Transoral Approach We found that the standard transoral approach exposes the region from the middle of the C2 vertebral body to the lower third of the clivus, consistent with several previous reports (5, 10, 13, 16, 17) (Table 1). Shaha et al. (35) reported that the transoral approach did not allow access above the foramen magnum, whereas others report reaching the midclivus (27) or upper third of the clivus (31). Menezes et al. (28) also reported reaching as far caudally as the C2–C3 interspace with a standard transoral approach, whereas others reported reaching as low as C3 (29) or C4 (31). Hadley et al. (15) described their technique that allows

NEUROSURGERY

B

D

C

FIGURE 7. Patient 12. A, preoperative T1-weighted sagittal magnetic resonance imaging (MRI) scan is shown; B, preoperative axial computed tomographic imaging scan (bone windows); C, intraoperative photograph demonstrating stairstep mandibulotomy; and D, postoperative T1-weighted sagittal MRI scan demonstrating excellent brainstem decompression.

exposure from the midclivus to C4 without mandibulotomy or tracheostomy. The lack of agreement on the transoral exposure probably reflects wide variation in surgical technique as well as anatomic variability in individual patients. Palatotomy increases the rostral exposure significantly to expose the midclivus, as previously observed (1, 3, 7, 8, 10, 20, 24).

Mandibulotomy Mandibulotomy shortens the operative distance and increases the angles of exposure in both the axial and sagittal planes. The sagittal exposure extends from the midclivus to the C2–C3 interspace. This range is slightly more than that reported by Delgado et al. (11) but is somewhat less than that reported by Cocke et al. (6). Although others (8) have noted the decrease in operative distance and increase in sagittal angle of exposure achieved with this approach, this has not been precisely quantified, and the advantage of increased axial exposure has not been previ-


YOUSSEF ET AL.

A

B

C D

FIGURE 8. Patient 17. A and B, preoperative T1-weighted sagittal and axial MRI scan. C and D, postoperative T1-weighted sagittal and axial MRI scans showing gross total resection of tumor with brainstem decompression.

ously noted. Vishteh et al. (37) reported an alternative technique through bilateral sagittal split mandibular osteotomies with intraoral incisions and rostrocaudal plane of retraction instead of lateral and enhanced exposure to the craniocervical junction. Palatotomy further increases rostral exposure to the upper third of the clivus without altering operative distance or caudal or axial exposures. The increased exposure of the upper clivus achieved by palatotomy agrees with the subjective observations of previous authors (7, 8, 11, 14). In our experience, mandibulotomy requires little additional time and has not been associated with complications. We routinely perform tracheostomy in conjunction with mandibulotomy, as described elsewhere (8, 10), because of neck swelling in the perioperative period. This is well tolerated, as is the facial incision. Although malocclusion and nonunion of the mandible have complicated mandibulotomy through the transcervical approach (21, 22), these complications have not been reported after transoral surgery (Table 4). Although some authors have reported the need to remove incisors before mandibulotomy, this was not necessary in any of the cadaveric specimens or patients we have treated.

Mandibuloglossotomy Mandibuloglossotomy greatly reduces operative distance and improves the sagittal exposure to reveal the region from


the upper third of the clivus to the C4–C5 interspace. This exposure is greater than that previously reported by others (11, 30, 35). Combining mandibuloglossotomy with palatotomy extends the rostral exposure to the sphenoid sinus without changing operative distance, caudal exposure, or the angle of exposure in the axial plane. The extent of sagittal exposure thus achieved is consistent with previous observations (2, 5, 38, 39). Because the neurovascular structures of the tongue are paired, the midline incision results in minimal loss of blood and function. Speech is impaired for a few days but usually resolves completely. We have not noted any persistent deficits, and none have been reported by others doing the procedure (Table 4) (2, 11, 30, 35, 38, 39). Resection of the hard palate has the advantage of increasing rostral exposure without requiring a facial incision or perioperative tracheostomy. However, it fails to decrease operative distance or angle of exposure in the axial plane or caudal direction. Palatotomy is also associated with a significant risk of velopharyngeal insufficiency, manifested as swallowing difficulty, nasal regurgitation, and hypernasal voice, in over one-third of reported cases (Table 4). In contrast, patients undergoing standard transoral approaches or those modified by mandibulotomy or mandibuloglossotomy without palatotomy have not experienced velopharyngeal insufficiency in our experience. In light of limited reports, this complication appears to be associated with palatal surgery and has been reported in only one patient as a complication of transoral surgery without palatotomy (10). Thus, in patients in whom increased rostral exposure is needed, consideration should be given to extending the standard transoral approach with mandibulotomy or mandibuloglossotomy rather than palatotomy to avoid this complication.

Clinical Experience This model correctly predicted the approach required to achieve a gross total resection in 15 of the 16 adult patients (93.8%) and 16 of the 19 patients (84.2%) overall (Table 3). Three patients with complex craniovertebral junction lesions, limited mouth opening, an elevated craniovertebral junction, and/or deep, wide, calcified lesions required greater modification of the standard transoral approach than predicted by the model. In these patients, the technique of intraoperative assessment of surgical trajectory using fluoroscopic guidance demonstrated whether additional modifications and exposure were required intraoperatively. Limited mouth opening prevented adequate exposure in Patients 12 and 16, necessitating mandibulotomy and palatotomy in both, as well as glossotomy and LeFort I maxillotomy in Patient 17. In addition to having a small mouth, Patient 12 had a fixed cervical flexion deformity (Fig. 7), which further limited mouth opening. This caveat has been noted by Crockard (8), who routinely measures the angle of mouth opening and warns that a standard transoral approach will not be sufficient in those patients with mouth opening less than 25 degrees. The fact that both patients are children also suggests that this model, based on observations in adult cadaveric specimens, may not be applicable to pediatric patients. All three pediatric patients in the current series



TABLE 4. Complications of various modifications of the transoral approach, as reported in the literaturea Approach

Series (ref. no.)

Transoral + palatotomy

Complication VPI

Malocclusion

Nonunion

Impaired speech

Alonso et al., 1971 (1)

1

0

NA

NA

0

Heyakawa et al., 1981 (17)

3

1

NA

NA

0

Kennedy et al., 1986 (20)

1

0

NA

NA

0


3

2

NA

NA

0

Beals and Joganic, 1992 (3)

1

0

NA

NA

0

Current study, 2006

1

1

NA

NA

0

Transoral + mandibulotomy


1

0

0

0

0

+ palatotomy

Current study

3

3

0

0

0

Transoral +

Wood et al., 1980 (36)

2

0

0

0

0

mandibuloglossotomy +

Arbit and Patterson, 1981 (2)

1

0

0

0

0

palatotomy

Wood et al., 1990 (39)

9

3

0

0

0

Current study, 2006 Total percent a

No.

3

1

0

0

0

30

11 (36.7)

0 (0%)

0 (0%)

0 (0%)

VPI, velopharyngeal insufficiency; NA, not applicable.

required modified transoral approaches, but the model correctly predicted the required approach in only one (33.3%). Anatomic anomalies of the craniovertebral junction necessitated a more extensive surgical approach, as in Patients 12 and 16, than that predicted by the cadaveric model. Although these lesions were confined to the craniovertebral junction, platybasia and severe basilar invagination in these patients markedly elevated the craniovertebral junction above its normal location (Fig. 7A). These anatomic abnormalities of the craniovertebral junction required greater rostral exposure than a standard transoral approach would provide. Although limited mouth opening also contributed to the limited exposure in Patient 12 (as previously noted), patients with basilar invagination or other lesions elevating the craniovertebral junction will probably require a more extensive approach than predicted by the cadaveric data, which are based on normal craniovertebral junction anatomy. Lastly, large lesions with extensive brainstem compression, such as those in Patients 8 and 17, are a relative indication for a more extensive approach, particularly when calcified. Mandibulotomy and glossotomy sequentially increase the angles of exposure in the axial plane, thus maximizing exposure of the lateral clivus. More importantly, the significant decrease in the depth of the operative field afforded by these modifications facilitates safer, more extensive brainstem decompression. This is particularly important for the resection of chordomas and chondrosarcomas. Extensive resection of these histologically benign tumors may markedly prolong survival (12).

CONCLUSION We have characterized the exposure achieved by the transoral approach and its modifications by mandibulotomy and

NEUROSURGERY

mandibuloglossotomy with and without palatotomy. Mandibulotomy and mandibuloglossotomy successively decreased operative distance while increasing exposure in the axial and sagittal planes. Palatotomy increased rostral exposure without changing the caudal or axial exposure or the operative distance. Intraoperative measurements may confirm the optimal approach in complex cases. This model correlated well with the transoral modifications required for achieving exposure in 93.8% of adult patients undergoing transoral surgery at our institution. Pediatric patients and patients with limited mouth opening, elevated craniovertebral junctions, or deep, wide, calcified lesions with extensive brainstem compression are apt to need more extensive exposure than predicted by our anatomic data. It is hoped that these findings will facilitate a more systematic approach to transoral surgery.

REFERENCES 1. Alonso WA, Black P, Connor GH, Uematsu S: Transoral transpalatal approach for resection of clival chordoma. Laryngoscope 8:1626–1631, 1971. 2. Arbit E, Patterson RH: Combined transoral and median labiomandibular glossotomy approach to the upper cervical spine. Neurosurgery 8:672–674, 1981. 3. Beals SP, Joganic EF: Transfacial exposure of anterior cranial fossa and clival tumors. BNI Q 8:2–18, 1992. 4. Biller HF, Lawson W: Anterior mandibular-splitting approach to the skull base. Ear Nose Throat J 65:134–141, 1986. 5. Bouthillier A, van Loveren HR, Tew JM: Anterior approaches to the clivus: Classification & indications. Contemp Neurosurg 16:1–8, 1994. 6. Cocke E, Robertson JG, Robertson JT, Crook JP: The extended maxillotomy and subtotal maxillectomy for excision of skull base tumors. Arch Otolaryngol Head Neck Surg 116:92–104, 1990. 7. Crockard HA: The transoral approach to the base of the brain and upper cervical cord. Ann R Coll Surg 67:321–325, 1985. 8. Crockard HA: Transoral approach to intra/extradural tumors, in Sekhar LN, Janecka IP (eds): Surgery of Cranial Base Tumors. New York, Raven Press, 1993, pp 225–234.


YOUSSEF ET AL.

9. Crockard HA, Pozo JL, Ransford AO, Stevens JM, Kendall BE, Essigman WK: Transoral decompression and posterior fusion for rheumatoid atlanto-axial subluxation. J Bone Joint Surg Br 68B:350–356, 1986. 10. Crockard HA, Sen CN: The transoral approach for the management of intradural lesions at the craniovertebral junction: review of 7 cases. Neurosurgery 28:88–98, 1991. 11. Delgado TE, Garrido E, Harwick RD: Labiomandibular, transoral approach to chordomas in the clivus and upper cervical spine. Neurosurgery 8:675–679, 1981. 12. Gay E, Sekhar LN, Rubenstein E, Wright DC, Sen C, Janecka IP, Snyderman CH: Chordomas and chondrosarcomas of the cranial base: Results and follow-up of 60 patients. Neurosurgery 56:887–897, 1995. 13. Gilsbach J, Eggert H: Transoral operations for craniospinal malformations. Neurosurg Rev 6:1999–1209, 1983. 14. Grime PD, Haskell P, Robertson I, Gullan R: Transfacial access for neurosurgical procedures: an extended role for the maxillofacial surgeon. I. The upper cervical spine and clivus. Int J Oral Maxillofacial Surg 20 [Suppl]:285–290, 1991. 15. Hadley MN, Spetzler RF, Sonntag VKH: The transoral approach to the superior cervical spine. J Neurosurg 71:16–23, 1989. 16. Harkey HL, Crockard HA, Stevens JM, Smith R, Ransford AO: The operative management of basilar impression in osteogenesis imperfecta. Neurosurgery 27:782–786, 1990. 17. Hayakawa T, Kamikawa K, Ohnishi T, Yoshimine T: Prevention of postoperative complications after a transoral transclival approach to basilar aneurysms. J Neurosurg 54:699–703, 1981. 18. Holliday MJ, Nachlas N, Kennedy DW: Uses and modifications of the infratemporal fossa approach to skull base tumors. Ear Nose Throat J 65:9–16, 1986. 19. Kanavel AB: Bullet locked between atlas and the base of the skull: Technique for removal through the mouth. Surg Clin 1:361–366, 1919. 20. Kennedy DW, Papel ID, Holliday M: Transpalatal approach to the skull base. Ear Nose Throat J 65:48–60, 1986. 21. Kocher T: Textbook of Operative Surgery. London, Adam and Charles Black, 1911, ed 3. 22. Krespi YP, Har-El G: Surgery of the clivus and the anterior cervical spine. Arch Otolaryngol 114:73–78, 1988. 23. Krespi YP, Sisson GA: Transmandibular exposure of the skull base. Am J Surg 148:534–538, 1984. 24. Lalwani AK, Kaplan MJ, Gutin PH: The transsphenoethmoid approach to the sphenoid sinus and clivus. Neurosurgery 31:1008–1014, 1992. 25. Martin H, Tullefsen HR, Gerold FP: Median labiomandibular glossotomy. Am J Surg 102:753–759, 1961. 26. McDonnell DE: Anterolateral cervical approach to the craniovertebral junction, in Wilkins RH, Rengachary SS (eds): Neurosurgical Operative Atlas. New York, McGraw-Hill, 1991, pp 147–164. 27. Menezes AH: Anterior approaches to the craniocervical junction. Clin Neurosurg 37:756–769, 1989. 28. Menezes AH, VanGilder JC, Graf CJ, McDonnell DE: Craniocervical abnormalities, a comprehensive surgical approach. J Neurosurg 53:444–455, 1980. 29. Merwin GE, Post JC, Sypert GW: Transoral approach to the upper cervical spine. Laryngoscope 101:780–784, 1991. 30. Nagib MG, Wisiol ES, Simonton SC, Levinson RM: Transoral labiomandibular approach to basiocciput chordomas in childhood. Childs Nerv Syst 6:126–130, 1990. 31. Pásztor E, Vajda J, Piffkóp, Horváth M, Gádor I: Transoral surgery for craniocervical space-occupying processes. J Neurosurg 60:276–281, 1984. 32. Preysing H: Contributions to pituitary surgery [in German]. Verh Laryngol 200:51–73, 1913. 33. Butlin HT: Disease of the Tongue. London, Ponden, Cassell and Co., 1900, p 359. 34. Sandor GK, Charles DA, Lawson VG, Tator CH: Transoral approach to the nasopharynx and clivus using the Le Forte I osteotomy with midpalatal split. Int J Oral Maxillofacial Surg 20:352–355, 1991. 35. Shaha AR, Johnson R, Miller J, Milhorat T: Transoral–transpharyngeal approach to the upper cervical vertebrae. Am J Surg 166:336–340, 1993. 36. Trotter W: Operations for malignant disease of the pharynx. Br J Surg 16:485–495, 1929.


37. Vishteh AG, Beals SP, Joganic EF, Reiff JL, Dickman CA, Sonntag VKH, Spetzler RF: Bilateral sagittal split mandibular osteotomies as an adjunct to the transoral approach in the anterior craniovertebral junction. Technical note. J Neurosurg 90:267–270, 1999. 38. Wood BG, Sadar ES, Levine HL, Dohn DF, Tucker HM: Surgical problems of the base of the skull, an interdisciplinary approach. Arch Otolaryngol 106:1–5, 1980. 39. Wood DE, Good TL, Hahn J, Bumphrye F, Bell G, Wood BG: Decompression of the brain stem and superior cervical spine for congenital/acquired craniovertebral invagination: An interdisciplinary approach. Laryngoscope 100:926–931, 1990.

Acknowledgments We thank Mario Ammirati, M.D., for his generous advice and Gina Benke for the photography and graphics. This study was supported by a National Institutes of Health Postdoctoral Interdisciplinary Research Fellowship in CNS Disease (S -T32 MH 19200–05, 19200–06) (AES).

COMMENTS

Y

oussef et al. have provided strong support for proceeding with minimally invasive surgical intervention with limited regard for obesity as a risk factor. This course is, perhaps, counterintuitive. With minimally invasive strategies, however, the risks associated with obesity may be lessened in that the requirements of wound healing are diminished, excessive tissue dissection is limited, and tissue devitalization is henceforth significantly minimized. Their observations should be carefully considered during the strategy determination process. Edward C. Benzel Cleveland, Ohio

T

he authors dissected seven cadavers through the transoral approach and then modified it with a mandibulotomy, mandibuloglossotomy, and palatotomy. The authors clinically correlated the measurements from the cadavers to 19 patients. The cadaveric models correlated well with the intraoperative measurements and with the need to modify the transoral approach in 15 of 16 adult patients. In their 3 pediatric patients, a more extensive approach was applied. The authors concluded that mandibulotomy and mandibuloglossotomy successfully decreased the operative distance while increasing exposure in the axial and sagittal planes. Palatotomy, however, increased rostral exposure without changing the caudal or axial exposure or the operative distance. The authors nicely demonstrate that variations of the transoral approach increased sagittal and axial exposure. Rightfully, they state that having these options available will facilitate preoperative planning for lesions involving the anterior extradural craniovertebral junction. Volker K.H. Sonntag Phoenix, Arizona

T

his is a very well thought out, well-described, and well-illustrated anatomical description of the transoral approach and modifications thereof with particular attention to the limits and advantages of each modification. The discussion regarding the benefits of avoiding a transpalatal approach is relevant, and I will probably alter my practice. I congratulate the authors on a well-done study. Daniel K. Resnick Madison, Wisconsin

I

n this study, Youssef et al. systematically compare the exposures provided by the transoral approach with subsequent extensions of this approach in both cadaveric specimens and patients intraoperatively. They note significant increases in axial and sagittal working angles



and significant decreases in working distance as additional osteotomies (mandibulotomy and palatotomy) and soft tissue dissection (glossotomy) are performed. Having delineated the exposures obtained with each approach, they then also retrospectively evaluate cases of anterior craniocervical compression to determine whether the approach suggested by their model correlates with the actual surgery chosen. As a result, they not only provide a thorough anatomic comparison of the various approaches but also create a simple and relevant tool to aid in deciding which approach to use for pathological lesions in this area. Although considered a safe and effective technique for decompressing the anterior craniocervical junction, the standard transoral approach has limitations. Craniocaudal and lateral exposures are all limited by jaw opening, and the working distance is extended because the surgeon’s hands cannot fit inside the oral cavity. The decision to extend such exposure by using more complex approaches to the anterior craniocervical junction (e.g., mandibulotomy, glossotomy, palatotomy, and maxillotomy) has classically been based on a combination of the patient’s anatomy, lesion size and location, surgical goal (piecemeal versus en bloc resection), and surgeon’s familiarity with the approaches. Although an extended exposure may potentially improve safety by providing greater access to neural and vascular structures, approach-related morbidity may be dramatically increased. Therefore, it is imperative that risks and benefits of each possible approach be thoroughly considered. With advances in radiosurgery for spine tumors and advances in minimally invasive techniques for cranial base decompression, approachrelated morbidity is becoming increasingly relevant to the patient population. In light of such potentially less morbid approaches to the anterior craniocervical junction, this study provides an objective means of further comparing surgical options before definitive treatment. Daniel M. Sciubba Ziya L. Gokaslan Baltimore, Maryland

NEUROSURGERY

T

his is a thoughtful presentation of operative exposures and options available through the transoral operative approach to pathological lesions ventral to the brainstem and proximal to the cervical spinal cord. Youssef et al. have attempted to correlate cadaveric dissection measurements with a review of patients they have treated via the transoral approach. I salute the use of cadaveric dissection to master/perfect operative knowledge, exposure, and techniques and consider the practice good for me and good for our residents-in-training. There is no question that palatotomy or mandibulotomy (with or without glossotomy) will improve the rostral-caudal operative exposure available via this ventral approach. The only questions are 1) whether it is needed, and, if so, 2) when it is needed. We do not learn key answers from this excellent presentation as it is a retrospectively applied assessment of work the authors have performed previously. Nonetheless, for extensive rostral pathological lesions, palatotomy can increase your transoral rostral exposure. For a pathological lesion that extends from the craniocervical junction caudally down the ventral proximal cervical spine, mandibulotomy can increase caudal exposure. In my experience, the use of rostral soft palate retractors and caudal tongue retraction provides virtually all of the exposure needed from the mid-clivus to the top of the C4 vertebral body without having to split the palate or the mandible or perform a tracheostomy. We position ourselves and the operating microscope on the patient’s chest, working upward (almost a transsphenoidal angle) for rostral pathological lesions, and from above the patient’s head working distally for more caudal pathological lesions. This article provides additional information and insights into the operative exposure options available through the transoral operative procedure. Mark N. Hadley Birmingham, Alabama


SPINE Clinical Study

CEREBROSPINAL FLUID-RELATED COMPLICATIONS WITH AUTOLOGOUS DURAPLASTY AND ARACHNOID SPARING IN TYPE I CHIARI MALFORMATION Caitlin E. Hoffman, M.D. Department of Neurological Surgery, Weill Medical College of Cornell University, New York, New York

Mark M. Souweidane, M.D. Department of Neurological Surgery, Weill Medical College of Cornell University, New York, New York Reprint requests: Mark M. Souweidane, M.D., Department of Neurological Surgery, Weill Medical College of Cornell University, 520 East 70th Street, Box 99, New York, NY 10021. Email: [email protected] Received, December 22, 2006. Accepted, July 24, 2007.

OBJECTIVE: Although there is a current consensus that Type I Chiari malformations (CM-I) should be treated only in the setting of symptomatic disease, significant controversy surrounds the most appropriate surgical procedure. Recent enthusiasm for osseous decompression without duraplasty is supported by the purportedly lower morbidity of this approach. Precise rates of morbidity with duraplasty, however, have not been reported. This study is intended to assess the cerebrospinal fluid-related morbidity associated with a patient population treated uniformly with autologous duraplasty for symptomatic CM-I. METHODS: A review of one surgeon’s practice (MMS) from 1997 to 2007 identified patients treated for symptomatic CM-I with osseous decompression and autologous duraplasty. A retrospective chart review was then performed for these patients with an emphasis on cerebrospinal fluid-related complications. RESULTS: Forty patients were treated for CM-I with decompression and autologous duraplasty. Twenty-four patients presented with a preoperative syrinx. The mean age was 13.3 years, and the median age was 12.9 years (range, 3.3–45.8 yr). The mean follow-up period was 11.4 months (range, 1–101 mo). There was no mortality associated with the procedure. Clinical response was observed in 91.8% of patients, with 70.2% experiencing complete symptomatic resolution and 21.6% experiencing partial improvement. Two patients (5.4%) had persistent symptomatic syringomyelia requiring syringosubarachnoid shunting. There was an overall morbidity rate of 2.5% due to one pseudomeningocele treated with a single percutaneous tap. There were no incidences of cerebrospinal fluid leak, meningitis, or postoperative hydrocephalus. CONCLUSION: The cerebrospinal fluid-related morbidity associated with autologous duraplasty for CM-I in a uniformly treated population is negligible. These results challenge the current rationale for a less aggressive surgical approach to CM-I. KEY WORDS: Chiari malformation, Duraplasty, Decompression, Morbidity Neurosurgery 62[ONS Suppl 1]:ONS156–ONS161, 2008

A

lthough the etiology of Type 1 Chiari malformations (CM-I) is diverse, in 1892 Chiari defined the common features of this malformation as “elongation of the tonsils and medial divisions of the inferior lobules of the cerebellum…which accompany the medulla oblongata into the spinal canal” (11, p 708). Resultant obstruction of cerebrospinal fluid (CSF) flow at the craniovertebral (CV) junction and compression of structures within the foramen magnum lead to brainstem compression, cranial nerve dysfunction, scoliosis, and symptomatic hydromyelia (1, 11, 17, 21, 26). The standard of therapy for this disorder, therefore, involves restoration of normal CSF flow surrounding the CV junction (17, 20). Additionally, in their 1991 survey of pediatric neurosurgeons,


DOI: 10.1227/01.NEU.0000297017.58340.CE

Haines and Berger (12) established that surgical intervention for CM-I is typically used only in the setting of symptomatic disease (2, 3, 11). Decompression of the posterior fossa is typically achieved through occipital craniectomy and cervical laminectomy to the level of tonsillar herniation. There is variability, however, in the approach to dural opening for further expansion of the posterior fossa, as well as the extent of intradural procedures and method of dural closure (5–8, 11). Multiple approaches, including dural scoring, dural splitting, and dural opening, are currently used, with significant controversy surrounding selection of the safest and most effective surgery for restoration of normal CSF flow at the CV junction.


CEREBROSPINAL FLUID-RELATED COMPLICATIONS

In light of the effort to move toward a less invasive and purportedly less morbid procedure, there is a current argument for the use of dural sparing surgery based on the assumption that dural opening is associated with frequent CSF-related morbidity (11, 17, 19). Evidence that a lower rate of clinical response and a higher rate of reoperation are associated with dural sparing procedures, however, argues against implementation of dural sparing procedures and supports the use of duraplasty (20). Evaluation of the morbidity associated with these two procedures, in light of conflicting and variable clinical response and surgeons’ preference, is necessary to identify the most appropriate surgical approach to these malformations (23). Therefore, this study presents the CSF-related morbidity associated with autologous duraplasty after osseous decompression in a patient population treated uniformly for symptomatic CM-I.

PATIENTS AND METHODS Patients Fifty-one patients who were treated for symptomatic CM-I with osseous decompression by one surgeon (MMS) from 1997 to 2007 were identified. Of these patients, five were treated with endoscopic third ventriculostomy and were therefore excluded. The dura was closed with graft material other than autologous pericranium in six additional patients. These patients were excluded to establish homogeneity of the surgical procedure. Therefore, there were 40 patients treated uniformly for symptomatic CM-I with occipital craniectomy, cervical laminectomy, and autologous duraplasty. These patients comprised the subject population for data analysis. The Institutional Review Board/Privacy Board at the Weill Medical College of Cornell University granted approval for the review of medical records and conduction of this study. Twenty-four patients presented with a syrinx preoperatively. The presence of syringomyelia did not alter the surgical approach. Effectively, intradural procedures were not performed to treat hydromyelia. The mean age of the patients was 13.3 years (median, 12.9 yr; range, 3.3–45.8 yr). There were two patients older than 20 years of age, indicating a predominantly pediatric subject population. A retrospective chart review was performed for presentation, clinical response, and morbidity. Evaluation of morbidity rates focused on clinically relevant complications related to dural opening, such as pseudomeningocele, postoperative hydrocephalus, CSF leakage, and meningitis. The follow-up periods ranged from 1 to 101 months (mean, 11.4 mo).

Surgical Procedure Preoperative antibiotics were administered to each patient and consisted of intravenous cefazolin (25 mg/kg or 2 g maximum). Corticosteroids were not universally used. Once general anesthesia was administered, patients were placed in a prone position with rigid head fixation. Local subcutaneous anesthesia was used at the proposed site of skin incision. A midline incision extending from the inion to the upper cervical spine was used to perform a standard subperiosteal dissection of muscle from the occipital and cervical region. Muscle attachments were preserved at the superior nuchal line and usually at the C2 lamina. Osseous decompression was achieved with a high-speed air drill, encompassing the inferior aspect of the occipital bone with modest superior extension (approximately 1.5–2.0 cm) and lateral extension to the lateral-most aspect of the foramen magnum and cervical spinal

NEUROSURGERY

canal. The extent of cervical laminectomy was determined by the degree of tonsillar descent on preoperative imaging. A midline durotomy was used with lateral relaxation incisions at the cranial and caudal limits of the opening. The dural opening was performed with microscopic visualization to prevent violation of the arachnoid. Autologous pericranial tissue was harvested from the occipital surface above the superior nuchal line and grafted with monofilament suture. No sealant was used to reinforce the dural suture line, and no forced inspiratory pressure was routinely used to test the integrity of the dural closure. Two central tacking sutures were used between the graft and the fascial planes to enhance restoration of CSF flow behind the tonsils. Patients were routinely imaged with magnetic resonance imaging (MRI) approximately 12 weeks postoperatively. Subsequent imaging was dictated by clinical response.

RESULTS The clinical presentation of our study population is reported in Table 1. Headache, paresthesias, and scoliosis represent the most common symptoms. Patients were evaluated for the degree of clinical response after osseous decompression and autologous duraplasty. Long-term follow-up was available for 37 of the 40 patients. Among these patients, symptomatic improvement was demonstrated in 91.8%; complete resolution of symptoms was reported in 70.2%, and partial resolution in 21.6%. No improvement was demonstrated in two patients (ages 9.4 and 17.8 yr) because of persistent hydromyelia. Both patients had CINE MRI evidence of suboptimal flow across the craniovertebral junction. Treatment options, including reexploration and syringosubarachnoid shunting, were considered and both patients underwent syringosubarachnoid shunting. With follow-up periods of 50 and 34 months, respectively, both patients experienced symptom resolution and underwent no further procedures. The overall morbidity rate related to dural opening was 2.5% as the result of one case of a pseudomeningocele. This patient was treated with a single percutaneous tap and subsequent resolution of the fluid collection. There was a 0% incidence of aseptic or septic meningitis, CSF leak, or postoperative hydrocephalus (Table 2).

DISCUSSION CM-I is a prevalent disease process requiring neurosurgical consultation and intervention (20, 23). Although there is consensus that patients with CM-I should be treated only in the setting of symptomatic disease, the surgical approach remains variable, with posterior fossa decompression and cervical laminectomy serving as the two mainstays of treatment (11). After initial bony decompression, multiple procedures are used, with significant conflict in opinion surrounding the most appropriate method. Current approaches include dural incision without dural closure, dural incision followed by duraplasty, and dural sparing procedures such as dural splitting and dural scoring (11, 16, 18, 20, 23). Although Krieger et al. (16) have reported good clinical response to leaving the dura open, this approach is associated with a high incidence of headache as well as nausea


HOFFMAN AND SOUWEIDANE

TABLE 1. Predominant presenting signs and symptoms Signs and symptoms

No. of patients

Percentage (%)

Headaches/neck pain

29

72.5

8

20.0

11

27.5

Scoliosis Paresthesias: lower extremity Monoparesis: lower extremity

3

7.5

Nystagmus

2

5.0

Apnea

2

5.0

Syncope

2

5.0

TABLE 2. Morbidity associated with autologous duraplasty after osseous decompressiona Morbidity of dural opening CSF leak Aseptic/septic meningitis Pseudomeningocele Hydrocephalus a

Percentage (no.) 0 0 2.5 (1/40) 0

CSF, cerebrospinal fluid.

and vomiting. The most significant conflict, therefore, surrounds the use of dural opening. Historically, dural opening was considered essential for successful posterior fossa decompression because of the need for direct visualization of the level of descent of the cerebellar tonsils, exploration and manipulation of intradural pathology, and further expansion of the posterior fossa dimension (26). The advent of MRI, however, has minimized the need for intraoperative determination of tonsillar descent and extent of cervical laminectomy. In addition, experimentation with different methods of expanding the posterior fossa after bony decompression has led to recent reports of satisfactory results with dural sparing techniques. A review of the current literature comparing the efficacy of dural sparing procedures and duraplasty in CM-I demonstrates a comparable clinical response with both surgeries (Table 3). Multiple authors have evaluated the resolution of symptoms after dural scoring and splitting versus duraplasty and reported clinical improvement in greater than 80% of patients in both groups (9, 10, 11, 13–15, 17, 19, 20, 22, 23, 25, 26). The primary difference between dural opening and dural sparing is evident in the comparison of rates of CSF-related complication, including CSF leakage, meningitis, pseudomeningocele, and postoperative hydrocephalus (Table 4). The morbidity rates without dural opening range from 0 to 10%, with all complications represented by superficial infections (10, 11, 13, 14, 15, 17, 19, 20). In comparison, studies evaluating the morbidity associated with duraplasty reveal a complication rate ranging from 0 to 48%, with the majority of these complications being aseptic meningitis and pseudomeningocele (9, 13, 14, 17, 19, 20, 22, 23).


Although this discrepancy appears high and serves as the present argument for the use of dural sparing surgery, many aspects of these studies indicate the need for further investigation of the true difference in morbidity associated with different approaches to dural opening. The small sample sizes reported in the current series limit the clinical and statistical significance of their results. Tubbs et al. (25) addressed this limitation by reporting a low morbidity rate of 2.3% in a large cohort of 129 patients who all underwent duraplasty following osseous decompression. Although this complication rate was much lower than the mean in the previously published literature, providing evidence in support of dural opening, interpretation of these results is limited by the fact that surgical approach was heterogeneous, involving variable intradural procedures and graft materials. Additional limitations of the current studies addressing dural opening include the wide variability in reported morbidity rates and the variability of intradural manipulation. Feldstein and Choudhri (9) report a complication rate of 0% in seven patients, whereas Munshi et al. (19) report 48% morbidity in 21 patients. This range of observed complications with dural opening could be the result of the small sample size in these studies and the selection bias introduced by the fact that patients with intradural pathology, most commonly syringomyelia, were often chosen for duraplasty. Furthermore, evaluation of the long-term outcome with both procedures may alter the interpretation of the initially reported morbidity rates. Navarro et al. (20) report a 41% complication rate associated with duraplasty versus a 5.6% complication rate without dural opening. This study presents one of the highest reported morbidity rates with duraplasty and, therefore, contributes significantly to the argument against dural opening. Among patients without dural opening in their original procedure, however, there was a greater incidence of reoperation with an associated increase in morbidity of 41.6%. This variability in the reported incidence of CSF-related complications, the greater need for reoperation without duraplasty, and the elevated morbidity reported with reoperation therefore demonstrate the need for a more focused evaluation of the true morbidity associated with duraplasty in the treatment of CM-I. The results of our study provide evidence that patients treated uniformly for symptomatic CM-I with osseous decompression and autologous duraplasty can demonstrate good clinical response with minimal CSF-related morbidity. Symptomatic improvement occurred in 91.8% of patients, with 70.2% of those patients reporting complete resolution of preoperative symptoms. The two patients with stable persistence of their symptoms were found to have syringomyelia and were treated with syringosubarachnoid shunting. In addition, the uniformity of the surgical approach in our patients allows for generalization of these results, with specific implication for future operative decision-making. Lastly, our reported morbidity in one patient (2.5%), represented by a transient and clinically insignificant event, is appreciably lower than the average from published series (12.4%).



TABLE 3. Clinical response and morbidity in patients treated for Chiari type I malformation without duraplasty Series (ref. no.)

No. of patients

Isu et al., 1993 (15)

Symptom resolution, % (no.)

7

86 (6/7)

Complications (%) 0

Hida et al., 1995 (14)

12

Yundt et al., 1996 (26)

3

100 (3/3)

8

86 (7/8)

0

11

73 (8/11)

10 (1/11), 1 infection

Gambardella et al., 1998 (10) Munshi et al., 2000 (19)

83 (10/12)

0 n/a

Genitori et al., 2000 (11)

26

100 (26/26)

8 (2/26), 2 infections

Limonadi and Selden, 2004 (17)

12

100 (12/12)

0

47

72.2 (34/47)

Navarro et al., 2004 (20) Total

126

0

84 (106/126)

6 (3/55)

TABLE 4. Clinical response and morbidity in patients treated for Chiari type I malformation with duraplasty Series (ref. no.) Oldfield et al., 1994 (22)

No. of patients

Symptom resolution, % (no.)

Complications, % (no.)

7

71 (5/7)

n/a

Sahuquillo et al., 1994 (23)

10

80 (8/10)

10 (1/10), aseptic meningitis

Hida et al., 1995 (14)

21

Feldstein and Choudhri, 1999 (9) Munshi et al., 2000 (19) Tubbs et al., 2003 (25)

7 21 129

82 (17/21) 100 (7/7)

10 (2/21), 2 meningitis 0

86 (18/21)

48 (11/23), 2 CSF leaks, 1 aseptic meningitis, 4 pseudomeningoceles, 3 wound infections, 1 pain syndrome

83 (107/129)

2.3 (3/129), 2 extra-axial subdural fluid collections, 1 brainstem compression

Limonadi and Selden, 2004 (17)

12

100 (12/12)

8.3 (1/12), 1 aseptic meningitis

Navarro et al., 2004 (20)

24

68.4 (16/24)

41 (10/24), 4 hygromas, 3 aseptic meningitis, 4 pseudomeningoceles, 1 fourth ventricular hemorrhage

82.2 (190/231)

12.4 (28/226)

Total

231

Interpretation of our results is limited by several factors. The mean follow-up period in this study is 11.4 months. Clinical significance of our lower morbidity rate is unlikely to be influenced by longer observation, as the development of CSFrelated complications is not expected to occur beyond our reported mean. Our study sample was larger than most in the current literature but still represents a relatively conservative number of patients. The presence of syringomyelia did not alter the surgical approach, which may be assumed to affect outcome in light of current investigation into the effect of a syrinx on clinical progression and response rate. Although the presence of a syrinx has been shown to limit clinical improvement, there is no evidence that hydromyelia significantly affects morbidity (11, 19, 20). Additionally, dural opening has not been demonstrated to accelerate or alter resolution of preoperative hydromyelia (20). As a result, although the distinction between patients with and without a syrinx may be necessary in future studies of the efficacy of different degrees of dural manipulation, the fact that syringomyelia did not alter surgical management in this study was unlikely to impact our results. Although a well-designed, prospective, randomized study would be ideal to further elucidate the relative morbidity and

NEUROSURGERY

efficacy of these procedures, it remains unlikely that such a study will be conducted, given current success rates, surgeon preferences, and standardized outcome scales.

CONCLUSION This study maintains the use of duraplasty as a safe and appropriate approach to CM-I and challenges the current rationale for the use of less aggressive and invasive surgeries. In the setting of emerging evidence that dural sparing procedures may be associated with a higher rate of clinical failure and reoperation, investigation of the efficacy of both procedures is necessary to further clarify the most appropriate surgical approach to CM-I, which should be governed by relative rates of morbidity, efficacy, and reoperation.

REFERENCES 1. Bell WO, Charney EB, Bruce DA, Sutton LN, Schut L: Symptomatic ArnoldChiari malformation: Review of experience with 22 cases: J Neurosurg 66:812–816, 1987. 2. Carmel PW: Management of the Chiari malformation in childhood. Clin Neurosurg 30:385–406, 1983.


HOFFMAN AND SOUWEIDANE

3. Carmel PW: The Chiari malformations and syringomyelia, in Hoffman HJ, Epstein F (eds): Disorders of the Developing Nervous System: Diagnosis and Treatment. Boston, Blackwell, 1986, pp 133–151. 4. Chiari H: Concerning alterations in the cerebellum resulting from cerebral hydrocephalus. Pediatr Neurosci 13:3–8, 1987. 5. Danish SF, Samdani A, Hanna A, Storm P, Sutton L: Experience with acellular human dura and bovine collagen matrix for duraplasty after posterior fossa decompression for Chiari malformations. J Neurosurg 104 [Suppl 1]:16–20, 2006. 6. Di Lorenzo N, Palma L, Palatinsky E, Fortuna A: “Conservative” cranio-cervical decompression in the treatment of syringomyelia-Chiari I complex. A prospective study of 20 adult cases. Spine 20:2479–2483, 1995. 7. Dyste GN, Menezes AH: Presentation and management of pediatric Chiari malformation without myelodisplasia. Neurosurgery 23:589–597, 1988. 8. Dyste GN, Menezes AH, VanGilder JC: Symptomatic Chiari malformations. An analysis of presentation, management, and long-term outcome. J Neurosurg 71:159–168, 1989. 9. Feldstein NA, Choudhri TF: Management of Chiari I malformations with holocord syringohydromyelia. Ped Neurosurg 31:143–149, 1999. 10. Gambardella G, Caruso G, Caffo M, Germano A, Rosa GL, Tomasello F: Transverse microincisions of the outer layer of the dura mater combined with foramen magnum decompression as treatment for syringomyelia with Chiari I malformation. Acta Neurochir 140:134–139, 1998. 11. Genitori L, Peretta P, Nurisso C, Macinante L, Mussa F: Chiari type I anomalies in children and adolescents: Minimally invasive management in a series of 53 cases. Child’s Nerv Sys 16:707–718, 2000. 12. Haines SJ, Berger M: Current treatment of Chiari malformations types I and II: A survey of the pediatric section of the American Association of Neurological Surgeons. Neurosurgery 28:353–357, 1991. 13. Hida K, Iwasaki Y, Koyanagi I, Abe H: Pediatric syringomyelia with chiari malformation: Its clinical characteristics and surgical outcomes. Surg Neurol 51:383–391, 1999. 14. Hida K, Iwasaki Y, Koyanagi I, Sawamura Y, Abe H: Surgical indication and results of foramen magnum decompression versus syringosubarachnoid shunting for syringomyelia associated with Chiari I malformation. Neurosurgery 37:673–679, 1995. 15. Isu T, Sasaki H, Takamura H, Kobayashi N: Foramen magnum decompression with removal of the outer layer of the dura as treatment for syringomyelia occurring with Chiari I malformation. Neurosurgery 33:844–850, 1993. 16. Krieger MD, McComb JG, Levy ML: Toward a simpler surgical management of Chiari I malformation in a pediatric population. Pediatr Neurosurg 30:113–121, 1999. 17. Limonadi FM, Selden NR: Dura-splitting decompression of the craniocervical junction: Reduced operative time, hospital stay, and cost with equivalent early outcome. J Neurosurg 101 [Suppl 2]:184–188, 2004. 18. Logue V, Edwards MR: Syringomyelia and its surgical treatment—an analysis of 75 patients. J Neurol Neurosurg Psych 44:273–284, 1981. 19. Munshi I, Frim D, Stine-Reyes R, Weir BK, Hekmatpanah J, Brown F: Effects of posterior fossa decompression with and without duraplasty on Chiari malformation-associated hydromyelia. Neurosurgery 46:1384–1390, 2000. 20. Navarro R, Olavarria G, Seshadri R, Gonzales-Portillo G, McLone DG, Tomita T: Surgical results of posterior fossa decompression for patients with Chiari I malformation. Childs Nerv Sys 20:349–356, 2004. 21. Oakes WJ: Chiari malformations, hydromyelia, syringomyelia, in Wilkins RH, Rengachary SS (eds): Neurosurgery. New York, McGraw-Hill, 1991, pp 2102–2124. 22. Oldfield EH, Muraszko K, Shawker TH, Patronas NJ: Pathophysiology of syringomyelia associated with Chiari I malformation of the cerebellar tonsil: Implications for diagnosis and treatment. J Neurosurg 80:3–15, 1994. 23. Sahuquillo J, Rubio E, Poca MA, Rovira A, Rodriguez-Baeza A, Cervera C: Posterior fossa reconstruction: A surgical technique for the treatment of Chiari malformation and Chiari I/syringomyelia complex—preliminary results and magnetic resonance imaging quantitative assessment of hindbrain migration. Neurosurgery 35:874–885, 1994. 24. Schijman E: History, anatomic forms, and pathogenesis of Chiari I malformations. Childs Nerv Syst 20:323–328, 2004.


25. Tubbs RS, McGirt MJ, Oakes WJ: Surgical experience in 130 pediatric patients with Chiari I malformations. J Neurosurg 99:291–296, 2003. 26. Yundt KD, Park TS, Tantuwaya VS, Kaufman BA: Posterior fossa decompression without duraplasty in infants and young children for treatment of Chiari malformation and achondroplasia. Pediatr Neurosurg 25:221–226, 1996.

COMMENTS

H

offman and Souweidane have insightfully crafted a review to further solidify the foundation for strategy determination in the Chiari I patient population. Their results focus on the younger patient but, nevertheless, are most likely transferrable to the adult population. It is always nice when new publications support one’s bias. This is indeed the case for me regarding this publication. Two factors are emphasized here. First, dural opening and patching are not fraught with significant increased risk. Second, it is likely that the dural relaxation that occurs with duraplasty is associated with improved decompression and better outcomes. These are two compelling reasons to perform duraplasty in this patient population. For their creativity and effort, the authors are to be heartily congratulated. Edward C. Benzel Cleveland, Ohio

T

he authors present a series of 40 patients, mostly pediatric, accrued over a 10-year period, who were treated with a suboccipital decompression and duraplasty for Chiari malformation. In this retrospective review of a relatively small series, cerebrospinal fluid (CSF) complications were rarely seen. The authors conclude that dural sparing procedures may not offer a significant advantage to duraplasty procedures in terms of CSF-related morbidity because such morbidity is rare. No evidence that patient outcomes were improved by opening the dura is presented. This series is best interpreted as demonstrating that there is no single “standard of care” for the operative treatment of Chiari malformations and that the surgeon needs to consider multiple anatomical factors when deciding on an operative strategy. Daniel K. Resnick Madison, Wisconsin

T

he authors conducted a retrospective review of a single surgeon’s case series of foramen magnum decompression for Chiari I malformations. The rationale for the study is that there are a myriad of acceptable surgical procedures for treating this disorder, including purely extradural techniques, dural expansion procedures with a variety of natural and synthetic materials, posterior fossa reconstructions, and procedures involving the resection or manipulation of neural elements. The overall excellent clinical results in this case series have led the authors to conclude that autologous duraplasty is an effective treatment that does not cause an unacceptable rate of surgical morbidity. I think that many neurosurgeons would agree with this conclusion and I certainly use a technique very similar to the one described. It should be noted, however, that this case series is composed largely of pediatric patients (with a median age of 13 years), and the true incidence of complications may indeed be different in the adult population. Michael Y. Wang Miami, Florida

T

he surgical treatment of Chiari type I malformations has not been standardized. Controversy exists concerning the extent of bony



decompression and the need for additional steps such as duraplasty, arachnoid lysis, and reduction of the cerebellar tonsils. Most experienced neurosurgeons tend to use a fixed operative technique, but this approach has limitations and does not address the considerable variability of patient-specific findings. In recent years, osseus decompression with dural sparing has received increasing attention, especially among pediatric neurosurgeons, as a means for reducing CSF leaks and pseudomeningocele formation. The authors of this report present their experience with 40 patients treated by one surgeon with osseus decompression and autologous duraplasty between 1997 and 2007. There was only one CSF-related complication, a pseudomeningocele that resolved after a percutaneous tap, and no instances of CSF wound leakage, meningitis, or postoperative hydrocephalus. The findings in this report demonstrate that autologous duraplasty performed correctly by an experienced neurosurgeon can be associated with negligible complications. However, the measure of an operative procedure includes not only complications but outcome. In this article, there is limited appraisal of the effectiveness of the authors’ technique. For example, two of 24 patients with syringomyelia (8%) had persistent symptomatic syringomyelia that was treated by syringosubarachnoid shunting without consideration of Chiari revision surgery. There is no information about the incidence of persistent tonsillar herniation or persistent asymptomatic syringomyelia. Overall, the strength of this article is that it provides an effective challenge to the notion that duraplasty is an overly aggressive surgical step. Left unsaid, however, is accumulating evidence that optimal strategies

for Chiari surgery require tailored steps to address the patient-specific anatomic and physiological variations of tonsillar herniation. Paolo A. Bolognese Thomas H. Milhorat Manhasset, New York

T

his article represents an honest and accurate assessment of the risks of this more conservative approach. It is class III data, but the authors are to be congratulated for producing arguably the most useful and uniform data on this arachnoid-sparing approach to date. The clinical symptom relief is impressive and it has also been my experience in pediatric populations compared with adult populations. The two patients who had persistent syringes without significant collapse represent surgical failures. A reexploration would be my operation of choice simply because the longevity of a syringosubarachnoid shunt is limited in the best of hands and in the best patients. There may have been no major CSF leaks in this series, but persistent syringes are still a failure of treatment that a more tailored approach may have obviated by treating the intradural scarring that affects an unknown number of patients with Chiari I malformations. I believe that this is a very attractive and safe approach, and the authors present a convincing argument to move to these more conservative approaches. However, concomitant attempts to improve the intradural approach, with deceased CSF complications, will also be necessary for those patients with difficult-to-treat syringomyelia. Richard G. Ellenbogen Seattle, Washington

Instruments of Anatomic Study from Culter Anatomicus, (1665). From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY



CORRECTION OF LATE TRAUMATIC THORACIC AND THORACOLUMBAR KYPHOTIC SPINAL DEFORMITIES USING POSTERIORLY PLACED INTERVERTEBRAL DISTRACTION CAGES Michael Y. Wang, M.D. Department of Neurological Surgery, Keck School of Medicine, University of Southern California, Los Angeles, California

Daniel H. Kim, M.D. Department of Neurological Surgery, Stanford University School of Medicine, Stanford, California

K. Anthony Kim, M.D. Department of Neurological Surgery, Keck School of Medicine, University of Southern California, Los Angeles, California Reprint requests: K. Anthony Kim, M.D., Department of Neurological Surgery, Keck School of Medicine, University of Southern California, 1200 N. State Street, Suite 5046, Los Angeles, CA 90033. Email: [email protected] Received, January 12, 2006. Accepted, July 13, 2007.

OBJECTIVE: To assess the safety and stability of thoracic or thoracolumbar deformity correction from a solely posterior approach with placement of modular anterior cages and posterior segmental fixation in one operation. METHODS: Twenty-eight patients who failed brace trial for 6 months or longer were included in the series. All patients had progressive neurological deficit and/or deformity progression at time of operation. All patients underwent a single operation in the prone position. Segmental fixation was accompanied by anterior column reconstruction using modular cages avoiding nerve root sacrifice. Stackable cages were used for high thoracic deformity. Deformity, Cobb angle, visual analog pain score, and x-ray evaluation of fusion ensued for mean follow-up period of 31 months. RESULTS: Patients achieved a mean sagittal deformity correction of 13.3 degrees ⫾ 7.4 standard deviation. Improved or maintained American Spinal Injury Association scores were noted in all patients. The mean time of operation was 334 minutes ⫾ 85 standard deviation, or 6 to 7 hours. At a mean follow-up of 31 months (range, 12–36 mo), the following complications were noted: subsidence greater than 2.5 mm (n ⫽ 3), cage migration requiring revision (n ⫽ 1), brachial plexopathy from malpositioning (n ⫽ 1), and intraoperative cerebrospinal fluid leak managed via lumbar drain (n ⫽ 2). Plain and dynamic radiographic evidence of maintained deformity correction was noted in 27 patients. CONCLUSION: Delayed kyphotic deformity correction of the thoracolumbar spine is achieved via a posterior-only approach. At a mean follow-up period of 31 months, sagittal angles remained acceptable. Improved fusion criteria and patient numbers will be required to determine fusion and loss of correction rates over time. KEY WORDS: Deformity, Fracture, Kyphosis, Osteotomy, Spinal trauma, Thoracic NeNeurosurgery 62[ONS Suppl 1]:ONS162–ONS172, 2008

F

orceful flexion or axial loading of the human body places the anterior and middle columns of the thoracic and lumbar spine under compressive stress. This can lead to compression and burst fractures, the most common injury patterns seen in spinal trauma. Because of the biomechanics involved, the fractured spine is frequently placed into kyphosis. Left untreated, these injuries may lead to glacial instability with progressive kyphosis (2), progressive neurological injury from spinal cord stretching and compression (1, 30, 39), severe axial pain symptoms from sagittal imbalance, respiratory compromise from rib cage encumbrance, and cosmetic concerns over the kyphus (7).


DOI: 10.1227/01.NEU.0000297069.90551.09

The delayed surgical correction of traumatic deformities is typically performed through combined anterior and posterior approaches. Anterior surgery offers the advantages of straightforward intervertebral distraction for height restoration, direct decompression of the spinal canal, and easier placement of load-bearing grafts with an adequate footprint. However, particularly in patients with previous anterior surgery-related scarring or organ compromise from recent trauma, transthoracic or retroperitoneal approaches may carry the drawbacks of major morbidity and the potential for visceral or vascular injury. Furthermore, the approach is anatomically complicated in certain regions such as the high and midthoracic spine.


CORRECTION OF LATE TRAUMATIC THORACIC AND THORACOLUMBAR KYPHOTIC SPINAL DEFORMITIES

Posterior surgery offers the advantages of multisegmental fixation, a straightforward approach familiar to all spinal surgeons, clear visualization of the neural elements, and biomechanical advantages for the correction of coronal plane deformities. When used alone, however, anterior column reconstruction, which is necessary for the maintenance of sagittal realignment, is difficult to achieve (40, 53). In the delayed setting, operative correction can be even more complicated than for acute injuries (32, 37, 45, 53). In particular, partial fusion of the involved spinal segments may make the deformity correction difficult, requiring a posterior releasing maneuver, followed by anterior height restoration and posterior spinal fixation (10, 11, 48, 54). Although highly effective, this threestep approach carries the drawback of higher operative morbidity. In particular, the technique is associated with prolonged anesthetic times and anterior approach-related sequelae, such as chest tube drainage, post-thoracotomy pain, injury to the vascular structures, pulmonary complications, and diaphragmatic dysfunction (4, 9, 13, 16, 17, 20, 21, 26, 47). Combined kyphoplasty or vertebroplasty with posterior fixation techniques have been attempted and have been of benefit for milder deformities (34, 58). Posterior-only approaches have been devised to enable complete treatment of the problem without violating the body cavities (3, 5, 6, 14, 22, 25, 27, 28, 31, 35, 46, 51, 52, 55, 56). Spinal shortening techniques such at pedicle subtraction osteotomy (PSO) have been developed for posterior-only kyphosis correction (7, 12, 19, 29, 36). The lateral extracavitary and costotransversectomy approaches allow improved access to and reconstruction of the anterior and middle column from behind, and these techniques have been used successfully for treating tumors and acute fractures (8, 43). Advances in interbody technology have now provided additional solutions for the surgeon desiring to correct anterior kyphotic deformities from a posterior-only approach in the setting of delayed trauma. Modularized or expandable cage devices can now be expanded along their sagittal dimension in situ. This method not only simplifies initial interbody cage placement, but it can also aid in the restoration of spinal alignment. This report summarizes our results with the use of stackable carbon-fiber and expandable cages for posttraumatic deformity correction from a posterior approach.

PATIENTS AND METHODS Patient Population Twenty-eight consecutive patients with sagittal spinal deformities after trauma were included in this series. Personal data, type of injury, mechanism of injury, type of fracture and degree of canal compromise, time from injury to surgical procedure, pre- and postoperative American Spinal Injury Association scores, intraoperative blood loss and surgery time, and pre- and postoperative Cobb angles were collected. All patients had remote trauma 6 months or more before surgery and were treated conservatively with bracing or expectant management initially. Surgery was performed for the following reasons: 1) progressive neurological deficit or 2) progressive worsening of the deformity angle over observation period.

NEUROSURGERY

A posterior-only approach was undertaken in the following conditions: previous thoracic cavity trauma or scarring, difficulty of anterior access, and morbid obesity. In patients with a midthoracic progressive kyphosis, thoracic surgeons were consulted with previous computed tomographic chest imaging displaying the proximity of the fracture to the great vessels and heart. The posterior-only approach was advocated to patients for whom thoracic surgeons felt a high morbidity risk in exposure of the region in question; for this reason, five of the eight midthoracic lesions were at the T3–T4 level in patients who were obese or of short stature with barrel chests (Table 1). At the thoracolumbar region, the posterior-only approach was advocated in patients who were morbidly obese and had undergone recent anterior surgery such as exploratory laparotomy post-trauma, which was a common event in our patients as all had experienced recent trauma. Two patients had recent bowel resection secondary to trauma. The risks and benefits of the posterior-only operation versus the anterior-posterior operation, including the possibility of aborting the posterior-only approach and proceeding with the anterior-posterior approach should the posterior-only approach be too difficult to perform intraoperatively, were described to all patients. In all films, comparisons were made with ImageJ, a Java-based version of the public domain National Institutes of Health Image software (Research Services Branch, National Institutes of Health, Bethesda, MD). Plain x-ray enhancement and magnification was performed with Adobe Photoshop 7 (Adobe Systems Inc., San Jose, CA). An inherent ⫾ 2.5 mm inaccuracy was assumed for radiographic measurements, as computed tomographic scales are standardized at a 5 mm width.

Surgical Technique: Decompression and Placement of Pedicle Screws All patients underwent a single surgical approach for 360-degree reconstruction of the spinal column. They were positioned prone on a radiolucent frame. In high thoracic deformity, a standard operating table with Mayfield extension was used. A fluoroscopic C-arm was draped into the sterile field. Linen rolls were used in favor of gel rolls to maximize beam penetration. A standard midline incision was used in all cases, with the exception of the hockey-stick incision used for the lateral extracavitary approach. At the spinal level of the deformity, exposure was taken out laterally beyond the transverse processes to the rib. Pedicle screws were placed at appropriate levels before ventral decompression. Short- versus long-segment fixation was determined at the discretion of the surgeon. In the thoracic spine, long-segment fusion was used to provide adequate screw purchase for rod compression work and to provide adequate long-term posterior column support. In the thoracolumbar junction, long-segment fixation was preferred except in the case of an L2–L4 burst fracture with minimal posterior column injury, where repair of the anterior column sagittal height was considered the main goal and fixation across the thoracolumbar junction was considered less than ideal. Corpectomies were then performed using either bitranspedicular (stackable cages) or the costotransversectomy/lateral extracavitary technique (expandable cages). In brief, a wide laminectomy was performed, exposing the dura and the nerve root around the kyphotic level. The nerve roots were exposed and isolated past the ganglion. The pedicle of the fractured level was hollowed out with a drill, leaving a cortical shell. A small rongeur was used to remove the pedicle walls, taking care, medially, with the thecal sac and the nerve roots above and below. With all bony elements taken down to the pedicle-body junction, the vertebral body was then drilled down using fluoroscopy for depth assessment. Eggshelling of the posterior aspect of the vertebral body



a

Trauma

Trauma

MVA

Trauma

45/M

35/F

55/M

Compression

Burst

Burst

Compression

Compression

Burst

Compression

Compression

Burst

Compression

Compression

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Burst

Dislocation

Dislocation

Dislocation

Type of fracture

L1 and L2

L1

L1

T3–T4

T3–T5

T4

T3 and T4

T3

T12

T3

T3

L1

L1

L1

L1

L1

L1

L1

L1

L1

T12

T12

L1

T12

L1

T9

T11

T11

Level

20%

90%

60%

20%

20%

35%

40%

50%

50%

55%

60%

55%

60%

70%

75%

80%

75%

60%

75%

80%

40%

50%

65%

45%

70%

35%

40%

55%

800

500

850

800

1000

900

1200

900

550

600

400

700

650

700

630

570

580

720

600

510

635

545

710

500

550

615

520

750

400

500

360

360

480

240

400

240

350

300

270

325

305

290

310

340

335

315

340

320

340

335

330

290

315

310

325

345

Canal Blood Operative violation loss (cc) time (mn)

A

A

A

A

B

A

C

D

E

C

D

E

D

D

E

D

D

D

D

D

D

D

D

E

D

C

C

B

Preoperative ASIA

A

A

A

A

B

A

E

E

E

D

E

E

E

E

E

E

E

E

E

E

E

E

E

E

E

C

D

B

Postoperative ASIA

30

30

25

35

35

35

40

45

22

40

43

25

20

17

25

18

22

24

27

18

24

22

20

25

27

10

15

20

Preoperative Cobb

ASIA, American Spinal Injury Association; AVP, automobile versus pedestrian; MVA, motor vehicle accident; MCA, motorcycle accident.

Trauma

Fall

56/M

64/M

MVA

44/M

57/M

MVA

21/F

MVA

MCA

48/M

39/M

MVA

45/M

Fall

MVA

26/F

24/F

MVA

25/M

Fall

MVA

51/M

MVA

MVA

58/M

53/F

MCA

38/M

46/M

Fall

MVA

Fall

21/F

43/M

Plane crash

46/F

54/M

MVA

MCA

50/M

MCA

39/M

24/F

AVP

MVA

42/F

Mechanism

52/M

Age (yr)/sex

TABLE 1. Demographic data on 28 patients who underwent posterior kyphectomy correctiona

20

25

25

22

20

10

25

20

3

18

16

8

10

9

13

9

10

9

12

10

12

9

11

8

12

10

10

8

Postoperative Cobb

10

5

0

13

15

25

25

25

19

22

27

17

10

8

12

9

12

15

15

8

12

11

9

17

15

0

5

12

Correction

brachial plexopathy

CSF leak;

Cage failure; CSF leak

None

None

None

None

T6 radiculopathy

Wound dehiscence

None

Deep venous thrombosis

None

None

None

None

None

None

None

6 mm subsidence

None

None

12 mm subsidence

None

None

8 mm subsidence

None

None

None

None

Complications

WANG ET AL.



was accomplished using a high-speed drill, curettes, rongeurs, and osteotomes. Downward-pushing Epstein curettes were then used to decompress the posterior longitudinal ligament and kyphus away from the ventral spinal cord at the midline. Epidural veins were cauterized with bipolar cautery, and bleeding was controlled with powdered gelatin sponge. The cranial and caudal intervertebral discs were removed with a sharp No. 15 blade followed by presized rasps and curettage. After denuding the endplates, either a stackable or expandable cage was used for interbody distraction (Fig. 1).

A

B

C

D

Surgical Technique: Placement of an Expandable or Stackable Cage Seventeen patients underwent expandable cage correction via the costotransverse/lateral extracavitary approach. For placement of expandable cages, an appropriately sized footprint and height were selected. This was determined by assessing the intraoperative interbody height and the desired degree of height restoration. The telescoping Synex cage (Synthes, Paoli, PA) was then ratcheted into its shortest length and inserted obliquely through this lateral approach into the interbody space. Realignment of the cage along the axis of the spinal column was followed by ratcheting the cage open (Fig. 2). During this maneuver, a connecting rod between the pedicle screws guided the reduction. After near-final cage height determination, the cage was packed with autograft, placed in situ, and fully expanded. Eleven patients underwent stackable cage placement via the transpedicular or costotransverse approach. For patients treated with a stackable cage construct, the intraoperative interbody height was determined and an appropriately sized footprint was selected. The average unitranspedicular insertion width was 12 mm compared with 24 mm or greater costotransverse width. Radiolucent carbon-fiber reinforced polymer stackable cages (Ocelot; DePuy Spine, Johnson & Johnson Co., Raynham, MA) were then placed serially into the intervertebral space and assembled in situ (Fig. 3). The traversing nerve root was preserved. Standard use of the stackable cages dictates that the cages be stacked and secured together with small screws before insertion. In our use for transpedicular insertion, however, each individual cage was placed in situ with the use of an extension, termed an artificial pedicle, that

A

B

C FIGURE 1. A, illustration showing a standard laminectomy. A right pedicle is exposed. Note the proximity of the pedicle to the exiting nerve roots above and below the pedicle. B and C, both pedicles have been drilled down to the vertebral body. A vertebral graft is being placed between the nerve roots on the left side (arrow). Note that the thecal sac is not retracted during the placement of the graft.

NEUROSURGERY

FIGURE 2. Technique of in situ placement of an expandable cage. A, lateral x-ray of a progressive L1 kyphotic deformity. The numbers 2–5 indicate vertebral lumbar levels. B, an L1 corpectomy is performed through a left costovertebral approach. Pedicle screws are placed at this time, and kyphosis correction is undertaken with long-segment fixation. C and D, an expandable cage has been placed and augmented with a MACS TL (Aesculap Inc,. Center Valley, PA) side plate and screw instrumentation. D, the construct in the anteroposterior dimension. Note the placement of the expandable cage below the endplate line of the L2 body, thereby increasing the risk of subsidence. At 20 months, a 12-mm subsidence was noted to the level of the MACS TL traversing screw below. attaches to the left side of the cage. After the cages were placed anterior to the cord, they were stacked with radiographic confirmation. The free ends of the artificial pedicles were then attached to the rod (Fig. 3D). Before placement of the last cage, a thoracic laminar spreader (Kineda System; DePuy Spine, Johnson & Johnson Co.) was placed between the cranial-most surface of the cage construct and the inferior vertebral endplate of the superior vertebra. This maneuver permitted anterior distraction similar to an anterior approach. The final cage is


WANG ET AL.

placed, and the spreader is removed. Having completed a bipedicular take-down, compression along the posterior rod was performed for further kyphosis correction in keeping with traditional PSO technique.

A

B

D C

F E

All cages were prefilled with autograft bone. In five cases with stackable cage use, porous calcium phosphate material (Actifuse; Global Orthopaedic Technology, Sydney, Australia) was also used to fill the cage and for placement in the lateral gutters.

RESULTS Twenty-eight patients (19 men, 9 women) with sagittal spinal deformities after trauma were included in this series. The average patient age was 43 years (range, 21–64 yr). The mechanisms of injury included motor vehicle accident (n ⫽ 17), fall from a height (n ⫽ 5), motorcycle accident (n ⫽ 4), auto versus pedestrian (n ⫽ 1), and plane crash (n ⫽ 1) (Table 1). Seventeen patients (11 men, 6 women; average age, 40 yr) underwent placement of an expandable cage. Eleven patients (eight men, three women; average age, 43 yr) underwent placement of a stackable carbon-fiber reinforced polymer cage. Stackable cages were used in situations requiring a smaller footplate in locations where anterior surgery was not readily accessible. The follow-up period averaged 31 months (range, 12–36 mo) with clinical visits and x-rays (44). The operative time averaged 334 minutes (range, 240–480 min), and the blood loss averaged 678 cc (range, 400–1200 cc). No unexpected events were noted during the follow-up interval. There were no cases of vascular injury, postoperative expanding pneumothorax, or neurological deterioration. Four major types of complications (n ⫽ 7) and four minor types of complications occurred. The following major complications were observed: 1) subsidence greater than 2.5 mm (n ⫽ 3); 2) intraoperative dural violation and cerebrospinal fluid leak (n ⫽ 2); 3) cage migration requiring revision surgery (n ⫽ 1); 4) brachial plexopathy (n ⫽ 1). Three patients developed 6-, 8-, and 8-mm subsidence of graft at 20 months and 6-, 8-, and 12-mm subsidence at 30 months, respectively (Fig. 2C). Assuming a computed tomographic ruler measurement error of ⫾2.5 mm, no gross subsidence was noted in the stackable carbon-fiber cage group, but it was observed in the expandable cage group. Two patients had intraoperative dural violation, both during the drilling of the pedicle. In each instance, the dural violation was on the lateral aspect of the dural sac and was sutured closed, followed by lumbar drain for cerebrospinal fluid diversion for 3 days. No delayed cerebrospinal fluid leak or meningitis was seen.

FIGURE 3. The bitranspedicular technique of in situ placement of modularized stackable cages. Case example of a 46year-old man who fell out of a tree and sustained a T12 burst fracture treated initially with bracing. A, the development of 22-degree local kyphosis with associated pain and myelopathy prompted surgical treatment. Morbid obesity precluded a straightforward anterolateral approach, and a posterior surgical procedure was performed. B, intraoperative view showing three intervertebral cages positioned with the use of “artificial pedicles” (small black arrowheads) and preservation of the exiting root (white arrow). C, intraoperative fluoroscopic view. Note two radiolucent intervertebral cages are placed above the lower vertebral body (VBl) in the intervertebral disk space (double arrow *) followed by anterior intervertebral distraction for anterior column height restoration accomplished with a laminar spreader (**) placed between the cage and upper vertebral body (VBu) endplate. The outline of the radiolucent cages is noted via radio-dense dots that are implanted in the corners of the cages, allowing radiographic visualization. D, final intraoperative view of the construct. E and F, final postoperative x-rays showing short-segment 360-degree reconstruction with a 19-degree correction in sagittal balance. Once again, the presence of the radiolucent stackable cages (arrows) is confirmed with the visualization of the radiodense dots that are seeded into the corners of the cages by the manufacturer. E, pedicle screw heads are marked with an asterisk, artificial pedicle screw heads affixed to rods are marked with a number sign, and the cross-link is marked with a diamond.




One case required immediate surgical revision. A 35-year-old woman was referred to our surgical center in delayed fashion. Despite her intact exam, her magnetic resonance imaging scans showed severe canal stenosis and progression kyphosis, which prompted surgery (Fig. 4, A and B). Intraoperatively, due to the patient’s narrow pedicles, bilateral L1 pedicle subtractions with L1 corpectomy did not allow sufficient width for placement of stackable cages with artificial pedicles. Two undersized stackable cages measuring 10 ⫻ 18 mm each were stacked in vivo without artificial pedicle support through a pedicle subtraction site measuring 12 ⫻ 14 mm with temporary distraction of the L1 nerve roots. Standing anteroposterior and lateral x-ray films were obtained at postoperative Day 4 (Fig. 4D). Cage migration was noted when compared with intraoperative films (Fig. 4, C and D). Flexion and extension films confirmed migration of the cage. The patient was brought back to surgery, and the left tranpedicular approach was widened with muscle retraction to a costotransverse approach. The left L1 nerve root was sacrificed, and a long prestacked cage was placed (Fig. 4E). One patient underwent an 8-hour operation without spinal complications. Postoperatively, he was noted to have a severe right brachioplexopathy (motor score, 0/5) confirmed with nerve conduction studies and electromyographic findings secondary to prolonged malpositioning of the right arm and forearm during the operation. Four minor types of complications occurred, including: 1) wound dehiscence (n ⫽ 1), 2) deep venous thrombosis (n ⫽ 1), 3) superficial chest skin breakdown due to gel rolls (n ⫽ 5), and 4) T6 radiculopathy (n ⫽ 1). One patient developed a small wound dehiscence. Most likely the combined result of diabetes, the 25-cm-long incision, and prolonged supine position, the wound dehiscence was superficial and was treated with packing and intravenous antibiotics. One patient developed symptomatic deep venous thrombosis without pulmonary embolism during the initial 2-week postoperative period. One patient was noted to have a new radioculopathy at the site of graft placement on postoperative Day 1 with benign computed tomographic findings. Stretch injury of the T4 thoracic nerve root was presumed, and a gabapentin regimen was used for 3 weeks with relief and gradually discontinued over 1 month. At 6 months, the patient acknowledged persistence of numbness but denied discomfort. The first five patients who underwent this surgery all have mild degrees of skin breakdown over the area of the chest rolls. After switching from gel chest rolls to linen chest rolls, no further complications were noted. Additionally, xray imaging was enhanced after the removal of the gel rolls. Some degree of neurological dysfunction was present due to spinal cord compression in all but four patients (Table 2). In 15 patients, neurological dysfunction progressed in the late phase after trauma and was due to progressive kyphosis with tenting of the thoracic spinal cord or conus medullaris over a bony angulation. For four cases with compression fractures, associated progressive disc herniations due to the kyphosis contributed to the spinal canal violation. Of the 24 patients with deficits, three were unchanged after surgery. The remainder experienced improvements. Twenty patients advanced one grade, and one improved

NEUROSURGERY

A

B C

E

D

FIGURE 4. Bitranspedicular approach to L1 corpectomy and placement of stackable cages. A, the patient was referred with the following progression kyphotic deformity. B, axial T2-weighted magnetic resonance imaging scan revealing significant canal compromise. C, the patient underwent T10–L2 segmental fixation and L1 corpectomy with placement of stackable cages. D, postoperative Day 5 imaging revealing interval cage migration compared with Day 1 (C). E, the patient underwent revision of cage via a left costotransverse method and sacrifice of the L1 nerve root.

two American Spinal Injury Association grades. None of the patients experienced neurological deterioration.


WANG ET AL.

TABLE 2. Comparison of pre- and postoperative American Spinal Injury Association scoresa Total no. of patients

Preoperative score

Postoperative score A

B

C

D

E

A

1

0

0

0

0

1 2

B

0

1

1

0

0

C

0

0

1

4

1

6

D

0

0

0

0

15

15

E

0

0

0

0

4

4

Total no. of patients

1

1

2

4

20

28

a Shaded area denotes improved American Spinal Injury Association scores in the postoperative setting when compared with preoperative scores.

Kyphotic deformities were measured during the last preoperative visit, immediately postoperative, and at the last followup visit in the clinic using plain x-rays and the Cobb technique. The mean preoperative segmental kyphosis was 26 degrees ⫾ 9.2 standard deviation (SD) (range, 10–45 degrees). The postoperative Day 1 segmental kyphosis was 11.39 degrees ⫾ 4.5 SD (range, 8–25 degrees). The mean 24-month postoperative kyphosis was 12.4 degrees ⫾ 5.8 SD. This resulted in an initial mean correction of 14.3 degrees ⫾ 6.8 SD (range, 0–27 degrees) at postoperative Day 1, 13.0 degrees ⫾ 6.5 SD at a mean of 24 months postoperatively, and 13.3 degrees ⫾ 7.4 SD at a mean of 31 months postoperatively. In all but one patient, there was some improvement in sagittal balance. No statistically significant differences were identified in the complications or outcomes between stackable and expandable cages.

DISCUSSION This report describes our experience with single-stage 360degree reconstruction of late traumatic kyphotic deformities. Whereas multistage reconstructions may be more straightforward, this technique may be preferable in select patients for whom an anterior approach is not feasible. Using the traditional advantages of the far lateral, lateral extracavitary (33, 38, 43), and transpedicular approaches, as well as the rod compressive work of PSO, we used contemporary spinal interbody instrumentation to improve our operative results. These implants have allowed us to overcome problems with interbody geometry to achieve the best anterior column height restoration and support. In addition, distraction directly across the anterior column has allowed us to achieve correction of kyphotic angulations, which can be quite rigid in the delayed setting. This approach allowed access to the high thoracic spine (T2–T5), a region that can be difficult and often inaccessible to treat using anterior or posterolateral approaches. Anteriorly,


the aortic arch and sternum complicate exposure of the spine. Posteriorly, the scapula must be mobilized to allow for adequate lateral access using the lateral extracavitary technique, a maneuver associated with potential shoulder-girdle morbidity. The use of stackable cages via the bitranspedicular access method allowed for placement of larger interbody constructs through smaller apertures, and reconstruction of the vertebrae in this region can be accomplished by a soft tissue exposure limited laterally to the end of the rib heads. Our mean correction of 13 degrees compares favorably to other surgical series using single-stage correction of local kyphosis (4, 5, 23, 24, 41, 42, 49, 55). Been et al. (5) analyzed outcome data after surgical treatment of posttraumatic thoracolumbar kyphosis following a simple Type A fracture. In 10 patients who were treated from the anterior-only approach, correction was noted from 23 degrees preoperatively to 11 degrees at the time of follow-up (mean correction, 12 degrees). In 15 patients treated from the posterior-only approach, correction was noted from 21 degrees preoperatively to 12 degrees at the time of follow-up (mean correction, 9 degrees). Payer (42) reports similar results after transpedicular correction of posttraumatic thoracic fractures. They noted a mean correction of 5 degrees at the time of follow-up. In the thoracic spine, a high degree of correction may not be desired. As was the case in our series, neither Payer (42) nor Been et al. (5) noted loss of deformity correction more than 1 degree over a follow-up period of 15 or more months. The application of these cages allowed for straightforward anterior column restoration and comparable short-term results to the anteroposterior approach in terms of operative time, blood loss, and complications. It can be argued that thoracic nerve root ligation can be performed with impunity, allowing for straightforward placement of traditional interbody grafts. However, we have found it preferable to preserve all of the neural elements. In one patient, however, we did experience postoperative T4 radiculopathy at the level of the graft placement after stackable cage placement. Posterior spinal shortening techniques such as the eggshell osteotomy have been used to achieve corrections of greater than 30 degrees in the thoracic and thoracolumbar spine (15, 59). However, the potential for neurological injury during the wedge removal and osteotomy closure has limited the popularity of this technique. Nevertheless, patients with a greater degree of kyphosis will likely require either osteotomy or combined posterior-anterior-posterior surgery for deformity correction. It should be noted, however, that this technique can be technically demanding for surgeons not familiar with the posterolateral anatomy and decompressions. The surgeon must have a high comfort level in working around the thecal sac and exiting nerve roots to accomplish an adequate corpectomy. The high speed drill, curettes, and rongeurs must be manipulated carefully because there is limited visualization due to the narrow surgical corridor and bone bleeding. Excessive bleeding during the vertebral body resection can be controlled with powdered gel foam used to pack the bony defect. In addition, a generous



geometrically appropriate vertebrectomy must be made, although complete removal of the posterior vertebral body is not mandatory as with wedge closure osteotomy operations. During interbody placement, care must be taken to ensure good contact and a snug fit between the implant and vertebral endplates. A smaller implant foot-plate is often tempting in these cases and must be weighed against subsidence risk. As posterior distraction with pedicle screws is crucial during the interbody placement, the patient’s bone quality is important. Particularly in the high thoracic spine, this technique was advocated only in situations where thoracic surgeons deemed the anterior surgery difficult or improbable for adequate access. Several surgical considerations have cut down our complications. Every patient is now placed on a radiolucent frame with linen rolls in lieu of gel rolls; this maneuver has increased our visualization of the thoracic spine on fluoroscopy and decreased our incidence of skin breakdown. Meticulous arm and forearm positioning is advised before commencement of the operation because these cases can go for 5 to 8 hours. In patients with small pedicles, feeling the inner edge of the pedicle with a laminectomy is advised to avoid lateral dural violation during drilling. In cases in which bilateral pedicle subtraction leads to poor corridor access or implant placement, the surgeon is advised to switch to a costotransverse approach, and, if able, to sacrifice nerve roots for adequate cage placement. Our current technique can be translated into spinal tumors with some caveats. We performed the bitranspedicular access method with stackable cage placement in five patients for alleviation of ventral dural compression from prostate metastasis in the upper and midthoracic spine (Fig. 5). Our initial analysis up to 8 months demonstrates greater blood loss (∼1000–2000 cc), longer operative time (∼8 h), increased deep venous thrombosis rates (80%), and greater difficulty with graft placement secondary to soft vertebral endplates and the need for a more diversified collection of graft footplates and shapes after transpedicular corpectomy. One patient died from acute respiratory distress secondary to pneumonia at 8 months. One patient died of chemotherapy complications at 6 months. One patient represented with adjacent segment tumor progression and kyphosis coupled with extensive muscle atrophy of the thoracic spine likely due to combined surgical and postradiation side-effects at 8 months. Further studies will be needed before the bitranspedicular technique can be translated for tumor and metastasis. During surgery, great care was taken not to disrupt the vertebral endplates to prevent the phenomenon of graft telescoping, typically seen in the intermediate postoperative period, and subsidence. While we were able to show correction and maintenance of sagittal alignment, three patients developed local changes in sagittal balance secondary to subsidence (Fig. 2C). In our series, three patients developed very noticeable subsidence, all of which were associated with the titanium expandable cage. The small footprint, medial posterior placement, and rigid construct of the Synex cage may predispose patients to subsidence (3, 4, 5, 10, 18, 57). We advise caution in using titanium expandable cages in the setting of posttraumatic, osteoporotic, postinfectious, or neoplastic spine as the incidence of subsidence may increase secondary to the above concerns.

NEUROSURGERY

A

B

C

D

E

FIGURE 5. A and B, sagittal and axial T2-weighted magnetic resonance imaging scan demonstrating compression at T3 with involvement of the pedicles with prostate metastasis. C, intraoperative view demonstrating corpectomy after bilateral pedicle subtraction; note that a Woodson tool is seen sliding from the left pedicle site to the right pedicle site anterior to the dura of the spinal cord. D, the pedicle subtraction and nerve root distraction is measured with a caliper for determination of the appropriate size graft placement. E, final intraoperative image after segmental fixation.

In the absence of larger cohort or randomized studies, it is difficult to ascertain the benefits of this approach over other surgical methodologies including the anteroposterior approach. Our study stands as a retrospective case series using modular cages in lieu of the anterior technique with acceptable mean retrospective outcomes at 31 months. We contend that this posterior-only technique is an important addendum to the spine surgeon’s armamentarium for scenarios preventing the surgeon from direct anterior access.


WANG ET AL.

CONCLUSION At the intermediate-term follow-up, we demonstrated the safety and stability of posttraumatic kyphosis deformity corrections performed via a posterior-only 360-degree correction approach. Expandable and stackable cages allowed for anterior column reconstruction and posterior segmental fixation in a single surgical operation through smaller apertures. Patients were spared the additional operative time, blood loss, and morbidity of anterior, traditional posterolateral, and combined approaches. Further follow-up will be needed to assess the integrity of these corrections over time.

REFERENCES 1. Abel R, Gerner HJ, Smit C, Meiners T: Residual deformity of the spinal canal in patients with traumatic paraplegia and secondary changes of the spinal cord. Spinal Cord 37:14–19, 1999. 2. Al-Khalifa FK, Adjei N, Yee AJ, Finkelstein JA: Patterns of collapse in thoracolumbar burst fractures. J Spinal Disord Tech 18:410–412, 2005. 3. An HS, Singh K, Vaccaro AR, Wang G, Yoshida H, Eck J, McGrady L, Lim TH: Biomechanical evaluation of contemporary posterior spinal internal fixation configurations in an unstable burst-fracture calf spine model: Special references of hook configurations and pedicle screws. Spine 29:257–262, 2004. 4. Andress HJ, Braun H, Helmberger T, Schürmann M, Hertlein H, Hartl WH: Long-term results after posterior fixation of thoraco-lumbar burst fractures. Injury 33:357–365, 2002. 5. Been HD, Poolman RW, Ubags LH: Clinical outcome and radiographic results after surgical treatment of post-traumatic thoracolumbar kyphosis following simple type A fractures. Eur Spine J 13:101–107, 2004. 6. Benazet JP, Chamberlin B, Lazennec JY, Saillant G: Surgical treatment of deformities of the thoracic and lumbar spine by posterior spinal osteotomy [in French]. Chirurgie 121:616–621, 1996. 7. Berven S, Kao H, Deviren V, Hu S, Bradford D: Treatment of thoracic pseudarthrosis in the adult: Is combined surgery necessary? Clin Orthop Relat Res Jun:25–31, 2003. 8. Bilsky MH, Boland P, Lis E, Raizer JJ, Healey JH: Single-stage posterolateral transpedicle approach for spondylectomy, epidural decompression, and circumferential fusion of spinal metastases. Spine 25:2240–2250, 2000. 9. Boriani S, Bandiera S, Biagini R, De Iure F, Giunti A: The use of carbon-fiber reinforced modular implant for the reconstruction of the anterior column of the spine. A clinical and experimental study conducted on 42 cases. Chir Organi Mov 85:309–335, 2000. 10. Boriani S, Palmisani M, Donati U, Tella G, De Iure F, Bandiera S, Siccardi G: The treatment of thoracic and lumbar spine fractures: A study of 123 cases treated surgically in 101 patients. Chir Organi Mov 85:137–149, 2000. 11. Bradford DS, Boachie-Adjei O: One-stage anterior and posterior hemivertebral resection and arthrodesis for congenital scoliosis. J Bone Joint Surg Am 72:536–540, 1990. 12. Bridwell KH, Lewis SJ, Rinella AR, Lenke LG, Baldus C, Blanke K: Pedicle subtraction osteotomy for the treatment of fixed sagittal imbalance. J Bone Joint Surg Am 85-A:454–463, 2003. 13. Camargo FP, Cordeiro EN, Napoli MM: Corrective osteotomy of the spine in ankylosing spondylitis. Experience with 66 cases. Clin Orthop Relat Res Jul:157–167, 1986. 14. Chang KW: A reduction-fixation system for unstable thoracolumbar burst fractures. Spine 17:879–886, 1992. 15. Chang KW: Oligosegmental correction of post-traumatic thoracolumbar angular kyphosis. Spine 18:1909–1915, 1993. 16. Channon GM, Jenkins DH: Aggressive surgical treatment of secondary spinal deformity in spina bifida children—is it worth while? Z Kinderchir 34:395– 398, 1981. 17. Chen H, Qui Y, Wang B, Yu Y, Zhu ZZ, Zhu LH: Autogenous tibial strut grafts used in severe kyphoscoliosis: Surgical policies and preliminary results. Chin Med J (Engl) 118:1245–1250, 2005.


18. Coe JD, Warden KE, Herzig MA, McAfee PC: Influence of bone mineral density on the fixation or thoracolumbar implants. A comparative study of the transpedicular screws, laminar hooks, and spine process wires. Spine 15:902– 907, 1990. 19. Danisa OA, Turner D, Richardson WJ: Surgical correction of lumbar kyphotic deformity: Posterior reduction “eggshell” osteotomy. J Neurosurg 91 [Suppl 1]:50–56, 2000. 20. Finkelstein JA, Wai EK, Jackson SS, Ahn H, Brighton-Knight M: Single-level fixation of flexion distraction injuries. J Spinal Disord Tech 16:236–242, 2003. 21. Fountain SS: A single-stage combined surgical approach for vertebral resections. J Bone Joint Surg Am 61:1011–1017, 1979. 22. Gertzbein SD, Harris MB: Wedge osteotomy for the correction of post-traumatic kyphosis. A new technique and a report of three cases. Spine 17:374–379, 1992. 23. Godlewski P, Mazurkieweicz T, Weglowski R: Transpedicular reconstruction of vertebral body height with simultaneous interbody spondylodesis [in Polish]. Neurol Neurochir Pol 38:395–400, 2004. 24. Hardaker WT, Cook WA, Friedman AH, Fitch RD: Bilateral transpedicular decompression and Harrington rod stabilization in the management of severe thoracolumbar burst fractures. Spine 17:162–171, 1992. 25. Heinig JA: Egg shell procedure, in Luque ER (ed): Segmental Spine Instrumentations. Thorofare, Slack, 1984, pp 221–230. 26. Huang TJ, Lubicky JP: Kyphectomy and segmental spinal instrumentation in young children with myelomeningocele kyphosis. J Formos Med Assoc 93:503–508, 1994. 27. Jacobs RR, Casey MP: Surgical management of thoracolumbar spinal injuries. General priniples and controversial considerations. Clin Orthop Oct:22–35, 1984. 28. Kaneda K: Anterior approach and Kaneda instrumentation for lesions of the thoracic and lumbar spine, in Bridewell K, DeWald RL, Hammerberg KW, Lenke LG, Lubicky JP, Neuwirth MG, Spencer DL, Vollmer DG (eds): The Textbook of Spinal Surgery. Philadelphia, Lippincott Williams & Wilkins, 1991, pp 959–990. 29. Kawahara N, Tomita K, Baba H, Kobayashi T, Fujita T, Murakami H: Closingopening wedge osteotomy to correct angular kyphotic deformity by a single posterior approach. Spine 26:391–402, 2001. 30. Knop C, Fabian HF, Bastian L, Blauth M: Late results of thoracolumbar fractures after posterior instrumentation and transpedicular bone grafting. Spine 26:88–99, 2001. 31. Koivikko MP, Myllynen P, Santavirta S: Fracture dislocations of the cervical spine: A review of 106 conservatively and operatively treated patients. Eur Spine J 13:610–616, 2004. 32. Kostuik JP, Matsusaki H: Anterior stabilization, instrumentation, and decompression for post-traumatic kyphosis. Spine 14:379–386, 1989. 33. Larson SJ, Holst RA, Hemmy DC, Sances A: Lateral extracavitary approach to traumatic lesions of the thoracic and lumbar spine. J Neurosurg 45:628–637, 1976. 34. Lee ST, Chen JF: Closed reduction vertebroplasty for the treatment of osteoporotic vertebral compression fractures. Technical note. J Neurosurg 100 [Suppl 4]:392–396, 2004. 35. Lehmer SM, Keppler L, Biscup RS, Enker P, Miller SD, Steffee AD: Posterior transvertebral osteotomy for adult thoracolumbar kyphosis. Spine 19:2060–2067, 1994. 36. Li F, Sagi HC, Liu B, Yuan HA: Comparative evaluation of single-level closing-wedge vertebral osteotomies for the correction of fixed kyphotic deformity of the lumbar spine: A cadaveric study. Spine 26:2385–2391, 2001. 37. Malcolm BW, Bradford DS, Winter RB, Chou SN: Post-traumatic kyphosis. A review of forty-eight surgically treated patients. J Bone Joint Surg Am 63:891–899, 1981. 38. Maiman DJ, Larson SJ, Benzel EC: Neurological improvement associated with late decompression of the thoracolumbar spinal cord. Neurosurgery 14:302–307, 1984. 39. Masini M, Maranhão V: Experimental determination of the effect of progressive sharp-angle spinal deformity on the spinal cord. Eur Spine J 6:89–92, 1997. 40. McLain RF, Sparling E, Benson DR: Early failure of short-segment pedicular instrumentation for thoracolumbar fractures. A preliminary report. J Bone Joint Surg Am 75:162–167, 1993.



41. Oertel J, Niendorf WR, Darwish N, Schroeder HW, Gaab MR: Limitations of dorsal transpedicular stabilization in unstable fractures of the lower thoracic and lumbar spine: An analysis of 133 patients. Acta Neurochir (Wien) 146:771–777, 2004. 42. Payer M: Unstable upper and middle thoracic fractures. Preliminary experience with a posterior transpedicular correction-fixation technique. J Clin Neurosci 12:529–533, 2005. 43. Resnick DK, Benzel EC: Lateral extracavitary approach for thoracic and thoracolumbar spine trauma: Operative complications. Neurosurgery 43:796–803, 1998. 44. Resnick DK, Choudhri TF, Dailey AT, Groff MW, Khoo L, Matz PG, Mummaneni P, Watters WC, Wang J, Walters BC, Hadley MN, American Association of neurological Surgeons/Congress of Neurological Surgeons: Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine: Part 4: Radiographic assessment of fusion. J Neurosurg Spine 2:653–657, 2005. 45. Roberson JR, Whitesides TE: Surgical reconstruction of late post-traumatic thoracolumbar kyphosis. Spine 10:307–312, 1985. 46. Schultz KD, McLaughlin MR, Haid RW, Comey CH, Rodts GE, Alexander J: Single-stage anterior-posterior decompression and stabilization for complex cervical spine disorders. J Neurosurg 93 [Suppl 2]:214–221, 2000. 47. Sengupta DK, Khazim R, Grevitt MP, Webb JK: Flexion osteotomy of the cervical spine: New technique for correction of iatrogenic extension deformity in ankylosing spondylitis. Spine 26:1068–1072, 2001. 48. Shvets AI: Surgical treatment of kyphotic deformities after injuries of the lumbar segment of the spine [in Russian]. Ortop Travmatol Protez 8:13–16, 1990. 49. Strømsøe K, Hern ES, Aunan E: Unstable vertebral fractures in the lower third of the spine treated with closed reduction and transpedicular posterior fixation: A retrospective analysis of 82 fractures in 78 patients. Eur Spine J 6:239–244, 1997. 50. Thomasen E: Vertebral osteotomy for correction of kyphosis in ankylosing spondylitis. Clin Orthop Relat Res Apr:142–152, 1985. 51. Tomita K, Kawahara N, Baba H, Kikuchi Y, Nishimura H: Circumspinal decompression for thoracic myelopathy due to combined ossification of the posterior longitudinal ligament and ligamentum flavum. Spine 15:1114–1120, 1990. 52. Tomita K, Kawahara N, Baba H, Tsuchiya H, Fujita T, Toribatake Y: Total en bloc spondylectomy. A new surgical technique for primary malignant vertebral tumors. Spine 22:324–333, 1997. 53. Vaccaro AR, Silber JS: Post-traumatic deformity. Spine 26 [Suppl 24]:S111– S118, 2001. 54. Verlaan JJ, Diekerhof CH, Buskens E, van der Tweel I, Verbout AJ, Dhert WJ, Oner FC: Surgical treatment of traumatic fractures of the thoracic and lumbar spine: A systematic review of the literature on techniques, complication, and outcome. Spine 29:803–814, 2004. 55. Viale GL, Silvestro C, Francaviglia N, Carta F, Bragazzi R, Bernucci C, Maiello M: Transpedicular decompression and stabilization of burst fractures of the lumbar spine. Surg Neurol 40:104–111, 1993. 56. Wang JC, Boland P, Mitra N, Yamada Y, Lis E, Stubblefield M, Bilsky MH: Single-stage posterolateral transpedicular approach for resection of epidural metastatic spine tumors involving the vertebral body with circumferential reconstruction: Results in 140 patients. Invited submissions from the Joint Section Meeting on Disorders of the Spine and Peripheral Nerves, March 2004. J Neurosurg Spine 1:287–98, 2004. 57. White AA, Panjabi MM: The basic kinematics of the human spine: A review of past and current knowledge. Spine 3:12–20, 1978. 58. Wilhelm K, Stoffel M, Ringel F, Rao G, Rösseler L, Urbach H, Meyer B: Preliminary experience with balloon kyphoplasty for the treatment of painful osteoporotic compression fractures [in German]. Rofo 175:1690–1696, 2003. 59. Wu SS, Hwa SY, Lin LC, Pai Wm, Chen PQ, Au MK: Management of rigid post-traumatic kyphosis. Spine 21:2260–2267, 1996.

COMMENTS

W

ang et al. have presented a dorsal/dorsolateral approach to ventral decompression for significant and relatively fixed kyphotic deformities. A bilateral approach is usually required for such reductions.

NEUROSURGERY

This approach, with a reduction technique that uses stackable cages, is very effective (as the authors have confirmed) for the reduction of fixed kyphotic deformities. Of note, a significant kyphotic deformity anatomically facilitates the lateral/ventral exposure of the spine. Henceforth, ventral decompression is similarly facilitated. The authors’ careful documentation of technique and clinical results warrants significant accolades. Edward C. Benzel Cleveland, Ohio

T

he authors described 28 patients who had a progressive deformity of the thoracic and thoracolumbar spine that led to progressive kyphosis, neurological deficits, or both. They nicely corrected the progression through a posterior approach, performing both fixation and decompression followed by placement of anterior cages via the posterolateral route. They did not have to sacrifice any nerve roots when placing the cages. Surprisingly, no neurological deficits occurred related to placement of these cages. The authors are to be congratulated for their excellent outcomes. As they themselves note, only surgeons with great anatomical knowledge of the thoracic and thoracolumbar spine and excellent expertise in managing complex spinal disorders should attempt to perform this type of surgery. Volker K. H. Sonntag Phoenix, Arizona

W

e have used the extracavitary approach to the thoracolumbar spine, as described by Sanford Larson, in the treatment of spinal deformity for at least 20 years. Using a “lazy S” incision, circumferential osteotomy-including posterior element resection, posterior fixation with safe correction of deformity, and anterior and posterior fusion are accomplished in a logical sequence (Larson and Maiman, Surgery of the Adult Lumbar Spine). If desired, anterior fixation can be accomplished as well. Avoidance of thoracotomy/laparotomy decreases surgical morbidity, if not operative time. Obviously, this is an operation with a considerable learning curve. The authors are to be applauded for the sizable and successful series of patients presented here. I am uncomfortable about two instrumentation issues presented here. The use of short-segment fixation has been considered, and discarded, repeatedly. It makes little sense biomechanically; most clinical series, including our own, have shown a high rate of loss of correction, even in the presence of anterior fixation. Simply stated, the internal stresses on the adjacent segments are unacceptably high. The theoretical value of short segment fixation, which is the incorporation of fewer segments, will probably not be realized until we have spinal fixation devices that produce less internal stress. I also am very concerned about the aggressive utilization of distraction within the disc spaces. Certainly, with a good bilateral extracavitary approach, the spine is loosened enough to allow some anterior distraction. However, when considering the amount of creep that will ultimately occur and the stresses on the spinal cord produced initially, too much distraction is not a good idea. Rather, I would recommend correcting as much deformity as possible posteriorly, with a slight distracting cage adjunct anteriorly. It is certainly better to accept slightly less correction and a happier cord! Again, kudos to the authors on this excellent overall effort. Dennis J. Maiman Milwaukee, Wisconsin


WANG ET AL.

W

ang et al. have demonstrated the feasibility of correcting thoracolumbar and upper thoracic kyphotic deformities and performing 360-degree reconstructions utilizing a posterior approach alone. The surgeon must be comfortable with a narrow and constrained working corridor to utilize this technique. Additionally, cases may be encountered which will require a costotransversectomy to achieve an acceptable outcome. Despite its limitations, this technique will have a role in managing selected patients. Vincent C. Traynelis Iowa City, Iowa

traumatic kyphosis via a posterior approach utilizing expandable or stackable intervertebral cages. The authors report good results with the technique, although follow-up is short. The approach is essentially a lateral extracavitary approach combined with the common use of osteotomy techniques and newer devices such as anterior struts. This approach is a viable option to consider if anterior approaches are not desired, however these procedures are difficult and not necessarily associated with lower morbidity than combined procedures. Surgeons are encouraged to consider all options when dealing with complex deformity cases. This technique represents another tool in the bag.

T

he authors of this paper present a retrospective review of 23 patients who underwent surgery for correction of late post-

Daniel K. Resnick Madison, Wisconsin

INFORMATION FOR SUBSCRIBERS NEUROSURGERY (ISSN 0148-396x), a peer-reviewed journal, is a program of the Congress of Neurological Surgeons and is published monthly by Lippincott Williams & Wilkins, 16522 Hunters Green Parkway, Hagerstown, MD 21740 -2116. Business offices are located at 530 Walnut Street, Philadelphia, PA 19106-3621. Periodical postage paid at Hagerstown, MD, and at additional mailing offices. Copyright © 2007 by the Congress of Neurological Surgeons. POSTMASTER: Send address changes to NEUROSURGERY, P.O. Box 1550, Hagerstown, MD 21740. CNS MEMBERSHIP SERVICES Congress of Neurological Surgeons 10 North Martingale Road, Suite 190 Schaumburg, IL 60173 TEL: 847/240-2500; FAX: 847/240-0804 TOLL-FREE: 877/517-1CNS EMAIL: [email protected]

EDITORIAL CORRESPONDENCE Managing Editor, Rodrick A. Faccio NEUROSURGERY 1420 San Pablo Street, PMB A-106 Los Angeles, CA 90033 TEL: 323/442-3001; FAX: 323/442-3002 EMAIL: [email protected]

ANNUAL SUBSCRIPTION RATES UNITED STATES—$496.00 Individual, $715.00 Institution, $247.00 In-training; REST OF WORLD—$728.00 Individual, $969.00 Institution, $430.00 In-training; Single copy rate $71.00. (All prices include a handling charge.) Subscriptions outside of North America must add $23.00 for airfreight delivery. United States residents of AL, CO, DC, FL, GA, HI, IA, ID, IN, KS, KY, LA, MD, MO, ND, NM, NV, PR, RI, SC, SD, UT, VT, WA, WV add state sales tax. The GST tax of 7% must be added to all orders shipped to Canada (Lippincott Williams & Wilkins’ GST Identification #895524239, Publications Mail Agreement #617687). Subscription prices outside the United States must be prepaid. Prices subject to change without notice. (Visit us online at: www.lww.com ) Members of the Congress of Neurological Surgeons receive the journal as a benefit of membership. Please indicate in-training status and name of institution. Institutional rates apply to libraries, hospitals, corporations, and partnerships of three or more individuals. Subscriptions will begin with currently available issue unless otherwise requested. Copies will be replaced without charge if the publisher receives a request within 90 days of the mailing date. Individual and in-training subscription rates include print and access to the online version. Institutional rates are for print only; online subscriptions are available via Ovid. Institutions can choose to purchase a print and online subscription together for a discounted rate. Institutions that wish to purchase a print subscription, please con-


tact Lippincott Williams & Wilkins, 16522 Hunters Green Parkway, Hagerstown, MD 21740-2116; TEL: 800/638-3030 (outside the United States 301/223-2300); FAX: 301/223-2400. Institutions that wish to purchase an online subscription or online with print, please contact the Ovid Regional Sales Office near you or visit: www.ovid. com/site/index.jsp and select Contact and Locations. ADDRESS FOR NON-MEMBER SUBSCRIPTION INFORMATION, ORDERS, OR CHANGE OF ADDRESS: Lippincott Williams & Wilkins, P.O. Box 1580, Hagerstown, MD 21741-1580; TEL: 800/638-3030 (outside the United States 301/223-2300); FAX: 301/223-2400. IN JAPAN, CONTACT: LWW Igaku-Shoin Ltd., 3-23-14 Hongo, Bunkyo-ku, Tokyo 113-0033; TEL: 81/3-5689-5400; FAX: 81/3-5689-5402. IN BANGLADESH, INDIA, NEPAL, SRI LANKA, AND PAKISTAN, CONTACT: Globe Publications Pvt. B-13, 3rd Floor, A Block, Shopping Complex, Naraina Vihar, Ring Road, New Delhi 110028, India; TEL: 91/11-579-3211; FAX: 91/11-579-8876. REPRINTS: Individual articles are available only from the authors. Reprints in large quantities, for commercial or academic use, may be purchased from the Publisher. For information and prices, call 800/341-2258.

BOUND VOLUMES FOR VOLUMES 58 & 59 will be available in early 2007 to subscribers only (U.S.—$253.00; FOREIGN—$271.00). Orders must be received by the Publisher before December 1, 2006. DISCLAIMER The statements and opinions contained in the articles of NEUROSURGERY are solely those of the individual authors and contributors and not of the CNS or LWW. The appearance of advertisements in the Journal is not a warranty, endorsement, or approval of the products or their safety. The CNS and LWW disclaim responsibility for an injury or property resulting from any ideas or products referred to in the articles or advertisements. INDEXING ABSTRACT SERVICES:The Journal is currently included by the following services in print and/or electronic format: Index Medicus, Current Contents/Life Sciences, Clinical Medicine, and Science Citation Index. COPYRIGHT & PERMISSIONS NEUROSURGERY is copyrighted by the CNS. No portion(s) of the work(s) may be reproduced without written consent from LWW. Permission to reproduce copies of articles for noncommercial use may be obtained from the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923; TEL: 978/7508400. Copyright © 2007 by the Congress of Neurological Surgeons.



THE TRANSLAMINAR APPROACH TO LUMBAR DISC HERNIATIONS IMPINGING THE EXITING ROOT Luca Papavero, M.D. Department of Neurosurgery, University Medical Center Eppendorf, Hamburg, Germany

Niels Langer, M.D. Department of Neurosurgery, University Medical Center Eppendorf, Hamburg, Germany

Erik Fritzsche, M.D. Department of Neurosurgery, University Medical Center Eppendorf, Hamburg, Germany

Pedram Emami, M.D. Department of Neurosurgery, University Medical Center Eppendorf, Hamburg, Germany

Manfred Westphal, M.D. Department of Neurosurgery, University Medical Center Eppendorf, Hamburg, Germany

OBJECTIVE: We undertook a prospective, non-randomized study on the translaminar approach for the treatment of cephalad extruded disc fragments impinging the exiting root. METHODS: Between May 2000 and July 2004, 104 patients (59 men)—presenting with upper lumbar root compression in 74% of the cases —underwent a translaminar approach. The mean age was 57 years (range, 27–80 yr). The lamina was approached either through the conventional subperiosteal route or via a muscle splitting access. Mostly intraforaminal disc fragments were removed through a translaminar hole 10 mm in diameter, and the disc space was cleared in cases of evident perforation of the annulus. Follow-up examinations were performed by an independent observer at 1 and 6 weeks; 3, 6, and 12 months; and once yearly thereafter (mean follow-up period, 32 mo). RESULTS: Extruded (61%) or subligamentous (39%) disc fragments were found intra-operatively. Laminae L4 (44%) and L5 (26%) were mostly involved. In eight cases, the translaminar hole was enlarged to a conventional laminotomy. In 13 patients, the disc space was cleared. The outcomes according to the Macnab criteria were excellent (67%), good (27%), fair (5%), and poor (1%). The incidence of recurrent disc herniations was 7%. Functional radiography performed in the first 20 patients 6 months after surgery and an additional 12 patients complaining of postsurgical back pain excluded any instability. CONCLUSION: The translaminar approach is recommended in disc herniations encroaching the exiting root, as an alternative to the conventional interlaminar route.

Ralph Kothe, M.D.

KEY WORDS: Lumbar disc herniation, Lumbar microdiscectomy, Spinal microsurgery, Translaminar approach

Department of Orthopedic Surgery, University Medical Center Eppendorf, Hamburg, Germany


Reprint requests: Luca Papavero, M.D., Zentrum für Spinale Chirurgie, Klinikum Eilbek, Dehnhaide 120, D-22081 Hamburg, Germany. Email: [email protected] Received, January 12, 2007. Accepted, May 1, 2007.

NEUROSURGERY

T

he anatomic unit of the lumbar spine consists of a vertebral body and the disc below. According to McCulloch (5), we can imagine a three-story anatomic house in which the first story is the disc level, the second story is the foraminal level, and the third story is the pedicle level. Without exception, the second story, i.e., the area between the upper rim of the disc space and the lower border of the cephalad pedicle, is covered by the lamina. In 11% of the patients who presented with lumbar disc herniation, a fragment that extruded cephalad into the spinal or root canal impinged the exiting root (Fig. 1). In 1998, Di Lorenzo et al. (2) proposed treating this “second story”-disc herniation via a direct translaminar approach (TLA), sparing the partial resection of the facet joint and of the yel-

DOI: 10.1227/01.NEU.0000297006.97350.B6

low ligament associated with the conventional interlaminar route. The comments of the reviewers were quite skeptical; the (suspected) technical difficulty of the procedure, the (assumed) inability/impossibility of clearing the disc space, and the potential risk of a fracture of the pars were viewed as important limitations of this new technique. In 2002, Soldner et al. (6) reported their experience in a retrospective study of 30 patients. Although, at that time, the description of the surgical technique was appreciated and the issue of long-term stability was questioned thoroughly. The purpose of this prospective study was to evaluate technical feasibility, clinical outcome, recurrence rate, and postsurgical stability of the TLA in patients with foraminal disc herniations impinging the exiting root.


PAPAVERO ET AL.

Five patients who underwent TLA because of caudally dislocated, so-called “midvertebral body” fragments were evaluated separately from the study group, as were five patients presenting with a second-story recurrent herniation after surgery via an interlaminar access (Fig. 2). Severe spinal canal stenosis and spina bifida were contraindications for TLA. If the bulk of a foraminal disc fragment was more than 50% lateral to a line marking the lateral border of the FIGURE 1. A, sagittal magnetic resonance imaging (MRI) scan showing a “typical” disc fragment (arrow) at the superior facet, it was approached L5–S1 level suitable for treatment with the translaminar approach (TLA). B, coronal MRI scan showing how the exitvia a paraspinal route. ing root L5 is pushed toward the lower rim of the cephalad pedicle. C, axial slice confirming that the fragment (arrow) All patients consented to parimpinges the exiting root L5 within the root canal. ticipate in the study, which was approved by the institutional review board. The patients were informed that the TLA could be PATIENTS AND METHODS widened to a laminotomy with partial facet joint resection corresponding to the conventional interlaminar approach.

A

B

C

Patient Population

Of the 1023 patients who underwent lumbar disc microsurgery between May 2000 and July 2004, 104 (45 women and 59 men) were enrolled in this study. The inclusion criteria were an exiting root syndrome that was resistant to conservative therapy and magnetic resonance imaging (MRI) and computed tomographic scans showing a cephalad extruded disc fragment, ideally pushing the exiting root upward against the lower border of the pedicle. The mean age at time of surgery was 57 years (age range, 27–80 yr). Typically, the patients presented with a history of leg pain that was more severe than back pain. In several patients, root blocks achieved only temporary relief from pain. Motor deficits were present in 90% and sensory deficits in 81% of the cases. Because of the prevalent encroachment of upper lumbar roots, gait disturbance caused by weakness of the quadriceps muscle was a common complaint.

A

B

FIGURE 2. A, “first-story” disc herniation successfully approached via a microsurgical interlaminar approach at the L4–L5 level. B, 3 years later, a large recurrent herniation, a “midvertebral body” fragment, was completely removed via TLA at L4.


Operative Technique For TLA, the same anesthesiological and positioning set-up as for any other lumbar disc surgery would be appropriate. When planning TLA, the following anatomic detail should be kept in mind: the width and the overlapping of the lamina in relation to the disc space increase in the caudal-cephalad direction, whereas the width of the isthmus decreases, meaning that the translaminar hole will be more medial and more oval in the cranial direction (Figs. 3 and 4). Because the lumbar laminae are oblique in the sagittal plane, i.e., they “dive” in the caudalcephalad direction, attention should be paid to compensate, at least partially, for this peculiarity by tilting the operation table in a “head-upward” direction. The advantages of a horizontal target lamina are twofold: the placement of the retractor blade and the drilling of the hole become easier. Because the surgical corridor to the target area is limited, the location of the skin incision must be determined very accurately. It should be centered onto the midpoint of the extruded disc fragment as shown by the FIGURE 3. In millimeters, the MRI scans or as calculated on height of the lamina (white numaxial computed tomographic bers) and its overlapping onto the scans (Fig. 4). disc space increases in the caudadThe targeted lamina can be cephalad direction, whereas the approached via a conventional width of the isthmus decreases subperiosteal route. A Caspar- or (black numbers). Williams-type retractor is then


TRANSLAMINAR APPROACH TO LUMBAR DISC HERNIATIONS

A

B

C

D

E

F

FIGURE 4. A, cranially extruded disc herniation at the L5–S1 level. B, preoperative fluoroscopic labeling of the disc fragment. C, the tubular retractor has been inserted via a transmuscular route; the dissector on the lamina points to the extruded disc fragment. D, the deep layer of the lamina is drilled off with the diamond burr. E, a round hole through the lamina at L5. F, a more oval hole through the lamina at L2.

the pars. The translaminar hole is located just cranially to the proximal insertion of the yellow ligament. The epidural exploration starts with the up and down dissection of the fat along the lateral border of the thecal sac. This dissection should be continued cephalad up to the axilla of the exiting root. Usually at this stage, extruded or subligamentous disc fragment can be mobilized. After decompression, the root slips caudally into the visible field (Fig. 5). The root canal is probed with a doubleangled hook. If an extensive annular perforation is detected, the disc space should be cleared. The wound closure is quite straightforward. Gelfoam (Aesculap, Tuttlingen, Germany) soaked with a long-acting steroid to fill in the hole is optional; this should be avoided if the disc space has been cleared. The patient can be mobilized the day of surgery.

Follow-up Evaluation Patients underwent 6-week, 3-month, 6-month, 1-year, and 2-year follow-up examinations by an independent observer (NL) at our institution. At the time of clinical evaluation, patient- and physician-documented changes in preoperative back and leg pain (Visual Analog Scale) were compared. The neurological function was also assessed. The result was rated as excellent, good, fair, or poor according to the Macnab (4) criteria. In the first 20 patients, biplaFIGURE 5. A, hole through the lamina at L4 on the right side. At least 3 mm of the lateral rim of the pars should nar and lateral functional x-rays be spared. B, the extruded disc fragment lies in between the thecal sac (asterisk) and the exiting root, which is pushed were routinely obtained 6 and 12 cranially and not visible. C, after removal of the disc fragment, the root at L4 slacks down into the surgical field. months after surgery to detect potential segmental instability inserted. An alternative is to use the transmuscular route by bluntly resulting from a fracture of the pars. After that, radiological investisplitting the paravertebral muscles with the index finger. A tubular gations were performed only in symptomatic patients. retractor (15 mm) is inserted and fixed with a holder arm. Regardless of the kind of speculum used, the lateral border of the lamina should RESULTS be visible underneath the retractor valve. At this point at the latest, the lamina should be tilted parallel to the The L4 lamina was the most involved (44%), followed by L5 floor, so that the cutting burr can be held more easily perpendicular to (26%), L3 (22%), and L2 (6%). Two patients required a twothe lamina. With slow, circular movements, a round (L5) or oval (L4 and level procedure. In eight procedures, the translaminar hole was cephalad) hole measuring approximately 10 mm in diameter is drilled enlarged to a laminotomy. The disc space had to be cleared in through three layers: a white layer of outer cortical bone, a red layer of 13 patients (15%). The mean operative time was 60 minutes spongy bone, and a white layer of inner cortical bone. For the sake of (range, 45–115 min). The mean follow-up period was 27 months safety, the inner cortical bone should be drilled with a diamond burr. At (range, 12–50 mo). least 3 mm of the lateral border should be spared to avoid a fracture of

A

NEUROSURGERY

B

C


PAPAVERO ET AL.

DISCUSSION

Clinical Outcome The clinical outcome according to Macnab’s (4) criteria was excellent in 67%, good in 27%, fair in 5%, and poor in 1% of the patients. Four patients affected by a “midvertebral body” herniation and four out of five who underwent operation for recurrent disc herniation made an excellent recovery. The recurrence rate (same level, both sides) in this series was 7%. This rate has to be correlated with the fact that the disc space was cleared in only 15% of the procedures. In these cases, the rupture of the annulus was evident. In four patients, the translaminar hole of the previous procedure allowed for the removal of the recurrent disc fragment as for the clearance of the disc space. In two cases, revision surgery was performed via a conventional interlaminar route. One patient with recurrent disc herniation developed spondylodiscitis and required revision surgery. There were no perioperative deaths or other significant complications. At the beginning of the learning curve, the wrong lamina was approached in two patients, which was corrected intraoperatively. Tilting of the operation table to direct the lamina parallel to the floor further minimized this risk. Dural tears occurred in four patients, and the particularly thin axillary dura was at risk during dissection of adherent disc fragments. Because of the narrow access, gluing a patch on durotomy was the remedy of choice. In eight procedures, the translaminar hole had to be enlarged to a conventional laminotomy. Although not exactly a complication, this change of strategy became necessary whenever a significant perforation of the caudad half of the annulus was detected, especially at the L5–S1 level.

Technical Feasibility If we compare a disc fragment that is extruded cephalad underneath the lamina to a fish underneath the surface of a frozen lake, there are two methods to hook the “fish.” The first is to cross the surface with an icebreaker and to slant to fish. The second option would be to cut a small hole in the ice surface targeted on the fish and to cast the line. The yellow ligament is left untouched, as the translaminar hole is usually located cephalad to its proximal insertion. The epidural fibrosis is minimized (1, 3). The lateral rim of the pars should be carefully spared. The fear that the less invasive novel approach could be counterbalanced by an increased rate of recurrence and postsurgical instability may explain the poor acceptance of this technique. Since the initial description of the TLA, a retrospective study of 30 patients by Soldner et al. (6) has been the only other report on this technique (2). To our knowledge, the present study is the first prospective series. Furthermore, this study involves a large number of patients with a follow-up period long enough to detect postsurgical stability. The translaminar hole was widened into a conventional laminotomy in eight patients. The anatomic reason was mostly a wide perforation of the caudal half of the annulus at the lowest lumbar levels. The overlapping of the disc space by the lamina may be very modest in some patients. Successful TLA requires: 1) meticulous preoperative planning with accurate localization of the disc fragment, 2) fluoroscopic labeling of its location, and 3) microsurgical skills to work through a 10-mm hole, which is facilitated by the use of

Radiographic Outcome To comply with the rigorous German legislation on radiation protection, routine flexion-extension x-rays could be performed at 6 and 12 months after surgery in only the first 20 patients of this series. Furthermore, 12 patients with axial load-dependent back pain underwent this investigation within the first 2 years after surgery. A fracture of the pars or a vertebral slip could be excluded in all the cases. In seven patients, MRI scans showed a decreased disc height. The clinical examination findings suggested facet joint syndrome. Fluoroscopically guided injections with local anesthetics and steroids resolved most of the pain.

A

B

C

FIGURE 6. A, a 65-year-old woman underwent a successful posterior lumbar interbody fusion at the L4–L5 level because of severe load-depending back pain. B, her sudden right-sided L3 pain and motor deficits prompted an MRI scan, which showed a disc herniation at the L3–L4 level (white arrow). C, TLA (white arrow) allowed for removal of the fragment, skipping the scar tissue, with excellent recovery. D, 6 months after surgery, a plain x-ray showed a partial bony closure of the translaminar hole (black arrow). post-OP, postoperative.



TRANSLAMINAR APPROACH TO LUMBAR DISC HERNIATIONS

bayoneted microinstruments. The learning curve should consist of at least 10 procedures.

Revision Surgery The recurrence rate of 7% must be weighed against the fact that the surgical philosophy was based more on fragmentectomy than on discectomy. The median interval of occurrence of recurrent disc herniation was 7 weeks (range, 4 d–53 wk). The review of the surgical reports suggests that, particularly in elderly patients, where the extruded disc tissue consisted mostly of a larger number of minute fragments, the indication to clear the disc space should be considered carefully. The revision surgery had to be performed via the conventional interlaminar route in only two of seven patients. In the five patients who had undergone previous discectomy via the interlaminar route, the recurrent cephalad extruded disc fragment was successfully removed with TLA. In these patients, the interlaminar scar tissue was skipped and the operative time shortened. This was also true in patients who developed a disc herniation in a level adjacent to a lumbar fusion (Fig. 6).

Postsurgical Instability The potential risk of a vertebral slip secondary to the fracture of the pars was not a clinical issue in this study. However, certain safety measures can be recommended: 1) to spare at least 3 mm of the lateral border of the pars, 2) to drill off the translaminar hole as an inverted truncated cone, and 3) to try to catch the most lateral fragments with (double bent) hooks. With this technique, an oval hole 6 mm wide will enable the removal of the extruded disc material through the narrower cephalad laminae.

CONCLUSION TLA is a safe alternative to the interlaminar approach for the removal of cephalad extruded disc fragments impinging the exiting root.

COMMENTS

P

apavero et al. have provided their experience with a translaminar approach to lumbar discectomy in selected patients with rostral herniations. This approach in selected patients seems appropriate. Subsequent instability may be diminished by such an approach, which, however, is yet to be proven. I suspect that the majority of surgeons would continue to use a conventional laminotomy approach to such pathological lesions. The approach presented by Papavero et al., however, should not be forgotten and should be included in our surgical armamentarium. For reviving our interest in this approach, the authors are to be congratulated. Edward C. Benzel Cleveland, Ohio

P

apavero et al. present 104 patients from a nonrandomized prospective study of the translaminar approach to a cephalad displaced disc fragment. The disc fragment was noted not to exceed greater than 50% lateral to the line marking the lateral border of the superior facet. The authors describe use of a conventional subperiosteal or muscle spitting approach to the lamina via a 15-mm port and the placement of a 10-mm translaminotomy. The authors advocate this approach to preserve the ligamenta flava, pars, and medial facet joints. They report a disc herniation recurrence rate of 7%. Four patients were treated by excision of the lesion through the previous translaminar opening, and two patients proceeded to have a conventional intralaminar approach to the disc fragments. By use of Macnab criteria, 94% of these patients reported an excellent or good outcome. The authors revive a novel technique to approach a cephalad extruded disc fragment, as previously described by Di Lorenzo et al. (1) and Soldner et al. (2). The proposal is that stability of the spine at the level addressed will be better maintained with preservation of the yellow ligament, pars, and medial facets. The authors also state that there will be less epidural fibrosis to contribute to complications in the future. The authors’ results of 94% in the excellent or good category using Macnab criteria would be expected when considering excision of disc fragment versus disc herniation. The authors report a 7% recurrence rate, which is slightly higher than the known recurrence rate with conventional techniques. The authors also note conversion to the conventional technique during operative intervention in eight of 104 patients. Odette Harris Atlanta, Georgia Daniel H. Kim Houston, Texas

REFERENCES 1. Aydin Y, Ziyal IM, Duman H, Türkmen CS, Basak M, Sahin Y: Clinical and radiological results of lumbar microdiskectomy technique with preserving of ligamentum flavum comparing to the standard microdiskectomy technique. Surg Neurol 57:5–14, 2002. 2. Di Lorenzo N, Porta F, Onnis G, Cannas A, Arbau G, Maleci A: Pars interarticularis fenestration in the treatment of foraminal lumbar disc herniation: A further surgical approach. Neurosurgery 42:87–90, 1998. 3. Kayaoglu CR, Calikoglu C, Binler S: Re-operation after lumbar disc surgery: Results in 85 cases. J Intern Med Res 31:318–323, 2003. 4. Macnab I: Negative disc exploration. An analysis of the causes of nerve-root envolvement in sixty-eight patients. J Bone Joint Surg Am 53:891–903, 1971. 5. McCulloch JA: Foraminal and extraforaminal lumbar disc herniations, in McCulloch JA, Young PH (eds): Essentials of Spinal Microsurgery. Philadelphia, Lippincott-Raven, 1998, pp 385–387. 6. Soldner F, Hoelper BM, Wallenfang T, Behr R: The translaminar approach to canalicular and cranio-dorsolateral lumbar disc herniations. Acta Neurochir (Wien) 144:315–320, 2002.

NEUROSURGERY

1. Di Lorenzo N, Porta F, Onnis G, Cannas A, Arbau G, Maleci A: Pars interarticularis fenestration in the treatment of foraminal lumbar disc herniation: A further surgical approach. Neurosurgery 42:87–90, 1998. 2. Soldner F, Hoelper BM, Wallenfang T, Behr R: The translaminar approach to canalicular and cranio-dorsolateral lumbar disc herniations. Acta Neurochir (Wien) 144:315–320, 2002.

T

he authors describe a quite interesting novel technique to approach a cephalad extruded disc fragment, reporting their experience with 104 patients. They were able to achieve 67% excellent outcome, according to Macnab criteria, with 7% recurrence and conversion to the conventional technique in eight patients. They assume that the translaminar approach allows maintenance of stability of the spine and prevention of postoperative fibrosis. In my opinion, it is important for the neurosurgical community to know this technique as an option to the conventional technique,


PAPAVERO ET AL.

because it allows the surgeon to perform discectomy and foraminotomy and restore the physiologic periradicular environment combining the preservation of periradicular fat (or free transplant of subcutaneous fat) with the reconstruction of the yellow ligament, thus recreating the interlaminar plane. Oreste de Divitiis Naples, Italy

T

he authors provide a technical report and case series describing a variation on the standard microdiscectomy technique. The patients represent a select group of patients with radiculopathy affecting the

exiting root owing to cranially migrated disc fragments. In this group of patients, a translaminar approach provides direct access to the region of the pathological lesion. I am not convinced that there is any evidence presented to indicate that this procedure provides improved outcomes compared with a standard microsurgical laminotomy beginning at the caudal edge of the lamina and progressing rostrally as necessary for exposure. However, the technical note does help define the surgical “stories” of the lumbar epidural space, and the approach is an elegant technical exercise. Daniel K. Resnick Madison, Wisconsin

Anatomy Lecture by Dr. Cornelis ‘s Gravesande, (1681), Cornelis de Man. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



SPINE Technical Case Report

AN ALTERNATIVE SOURCE OF AUTOGRAFT BONE FOR SPINAL FUSION: THE FEMUR: TECHNICAL CASE REPORT Tann A. Nichols, M.D. Department of Neurological Surgery and Rehabilitation, University of South Florida, Tampa, Florida

H. Claude Sagi, M.D. Florida Orthopedics Institute, Tampa, Florida

Timothy G. Weber, M.D. Indianapolis Orthopedics, Indianapolis, Indiana

Bernard H. Guiot, M.D. Department of Neurological Surgery and Rehabilitation, University of South Florida, Tampa, Florida Reprint requests: Tann A. Nichols, M.D., 4 Columbia Drive, Suite 730, Tampa, FL 33606. Email: [email protected] Received, August 31, 2006. Accepted, May 14, 2007.

OBJECTIVE: Autograft bone obtained from the iliac crest remains the “gold standard” for spinal fusion. For various reasons, including previous harvesting or pelvic dysmorphism, the iliac crest bone graft may not be available to the spinal surgeon. We present a novel use of a common orthopedic procedure, intramedullary reaming, for obtaining autograft for revision spinal fusion. METHODS: A 47-year-old woman presented with failed back syndrome after multiple lumbar surgeries with previous bilateral iliac crest bone harvest. A commercially available reaming system (Synthes Reamer-Irrigator-Aspirator; Synthes USA, West Chester, PA) was introduced into the left intramedullary canal of the femur while the patient remained in the prone position. Using continuous irrigation and aspiration, the reaming debris was collected and used as autograft for the subsequent spinal fusion. RESULTS: The patient underwent a successful L4–L5, L5–S1 transforaminal lumbar interbody fusion with L3–S1 pedicle screw fixation. No complications from the femoral reaming were observed, and 6-month follow-up x-rays demonstrated osseous fusion. CONCLUSION: Femoral reaming provides an alternative source of autograft bone when other sources are unavailable. KEY WORDS: Bone graft, Femoral reaming, Spinal fusion Neurosurgery 62[ONS Suppl 1]:ONSE179, 2008

T

he goal of spinal fusion is the elimination of pain and pathological movement through stable arthrodesis. Autograft bone remains the “gold standard” for spinal fusion because it is the only material with intrinsic osteogenic activity (13, 21). In most cases, adequate quantities of graft material are available from the iliac crests, with or without utilization of graft extenders, to achieve the appropriate volume. However, some patients may yield inadequate autograft from iliac harvest because of variation in body size, shape and volume of the iliac crests, pelvic dysmorphism, or previous harvesting procedures. Unfortunately, the patients who stand to benefit the most from the use of autograft are often those lacking adequate autograft sources. Many of these patients may be candidates for the use of commercially available recombinant osteoinductive factors; however, apprehensions regarding immune sensitization, inflammatory reaction, utilization in oncology


DOI: 10.1227/01.NEU.0000297008.43093.3E

patients, and appropriate resource utilization continue to limit their use (7, 16, 22). In this report, we describe the novel application of intramedullary femoral reaming to obtain adequate autograft for spinal fusion.

CASE REPORT A 47-year-old woman presented to our clinic with a 5-year course of progressive back and leg pain. She had previously undergone an L4–S1 laminectomy with pedicle screw instrumentation and iliac crest bone graft placement. She then developed pseudoarthrosis at the L4–L5 and L5–S1 levels, with a fracture of the right sacral screw. A second surgery, consisting of removal of instrumentation and placement of iliac crest onlay graft, was performed. Two years later, radiographic studies demonstrated continued pseudoarthrosis at the L4–L5 level with adjacent level failure at L3–L4. Both posterior iliac crests showed evidence of previous bone harvest, which was confirmed by review of the previous operative reports. After evaluation and discussion with the patient, she was scheduled for L4–L5 and L5–S1 transforaminal


FEMORAL REAMINGS AS AUTOGRAFT

lumbar interbody fusion with L3–S1 pedicle screw instrumentation and fusion. With the patient’s history of pseudoarthrosis, we thought autologous bone was needed to maximize the probability of a successful fusion. In the absence of available posterior iliac crests, autologous bone was obtained via intramedullary reaming of the femur.

OPERATIVE TECHNIQUE Equipment This technique uses the Reamer-Irrigator-Aspirator (RIA; Synthes USA, West Chester, PA), which was designed for closed-system intramedullary reaming before placement of orthopedic intramedullary nails for fracture fixation (9). The system uses an aggressive end-cutting one-pass reamer head connected to a hand-held reamer via a drive shaft. The drive shaft is encased within a double-cannulated tube that allows constant irrigation at the reamer head to prevent thermal necrosis and constant aspiration of the reaming debris and marrow contents to prevent canal pressurization and fat embolization (Fig. 1). A Redi-Flow Filter-Fine (BioMet, Inc., Warsaw, IN) is placed between the reamer and suction canister to capture the reaming debris.

FIGURE 1. Reamer-Irrigator-Aspirator (RIA) drive shaft. Inset, reamer head.

A

B

Procedure The patient was positioned in the prone position on a Jackson Table (OSI, Union City, CA). The standard prep and drape were modified to include the lateral thigh over the greater trochanter bilaterally. Fluoroscopic imaging was then used to measure the intramedullary canal diameter of the femoral shaft isthmus (narrowest portion) with a radio-opaque measuring device placed over the leg, and the appropriate diameter reamer was selected. It is important to remember that the canal diameter differs significantly in the anteroposterior and lateral dimensions. The appropriate reamer head diameter should be no more than 1 to 2 mm greater than the narrowest canal diameter measured to prevent over-reaming. A 4-cm incision was made on the lateral aspect of the thigh approximately 6 cm above the greater trochanter. The tip of the greater torchanter was then palpated through this incision, and a femoral awl was introduced and directed into the proximal intramedullary canal. (Fig. 2) After this, a 2.5-mm balltipped reaming guide rod was placed into the intramedullary canal to the level of the distal femoral epiphysis. The RIA was introduced over the guide rod and, with continuous closed suction and irrigation, the femoral canal was reamed (Fig. 3). After adequate graft material was obtained, the reamer and guide rod were removed, and the wound was irrigated and closed in routine fashion. The graft material was then stored in moist gauze for use in the interbody cages and as posterolateral onlay graft.

Postoperative Course The patient tolerated surgery well, without complications and with a significant improvement in her preoperative pain and functional level. Full weight-bearing activity was permitted on

NEUROSURGERY

FIGURE 2. A, femoral awl positioned on the greater trochanter (anteroposterior view). B, femoral awl positioned in the femoral canal (lateral view).

the reamed extremity immediately after the procedure. She complained of mild lateral thigh pain at the initial followup visit, which had resolved by the subsequent 6-week evaluation. Six months after surgery, x-rays demonstrated intact instrumentation with a maturing osseous fusion. No complications occurred secondary to the reaming procedure.

DISCUSSION Despite the proliferation of alternative products, including bone morphogenic proteins, platelet-rich plasma FIGURE 3. RIA in the intraderivatives, and demineralmedullary femoral canal (anteroized bone matrix compounds, posterior view). autograft bone remains the “gold standard” for spinal fusion because of a combination of availability, high fusion rates, absence of immunogenic reaction or sensitization, and low cost (13, 21). This bone is usually obtained from either anterior or posterior iliac crests, a very familiar and routine


NICHOLS ET AL.

technique for spinal surgeons. However, this familiarity comes at the price of increased operative time and a significant rate of local site complications, most often persistent pain (6, 11, 14, 18, 19). Multiple sites of autograft bone are available, including posterior iliac crest, anterior iliac crest, posterior elements, adjacent vertebral bodies, and rib; even sternum has been used (1, 8) (D.W. Cahill, personal communication, 2002). Attempts to reduce postoperative pain using alternative methods of harvesting the posterior crest have met with mixed success (3, 17). Many of these sites were previously used or not readily available in this patient. Both posterior crests had been previously harvested and local bone largely resected during previous surgeries. Bone from anterior sources remained an option. However, this would have required an additional positioning step and would have placed the new surgical incision at the harvest site under prolonged pressure during the posterior portion of the surgery. Anterior crest grafts also carry a wellknown and rather high (30–40%) long-term rate of complication, primarily sensory and cosmetic (14). Intramedullary reaming is an established practice within orthopedic surgery before placement of intramedullary nails for long bone fractures of the lower extremities. It has been shown to increase the overall union rate as well as decrease time to union (2, 4). However, the use of this procedure to obtain bone graft material has not, to our knowledge, been described in the literature. Femoral reamings have been shown to have similar in vitro osteogenic properties to iliac cancellous bone (5, 15) and contain viable mesenchymal stem cells (20). Although the fusion rate associated with the use of femoral reamings has not been studied, the similarity in cell function and histology suggest that similar fusion rates are likely. Indeed, in our patient, a maturing fusion mass was noted radiographically at 6 months despite two failed previous attempts at fusion with autograft bone. Femoral canal intramedullary bone harvesting also has the advantage of being accessible from the prone, supine, or lateral positions. The procedure can be performed at a late stage of the surgery, reducing explant time. The volume of bone obtained depends on the size of the patient’s bone and the extent of reaming; however, volumes ranging from 30 to 90 mL can be procured from a single femur (15). The potential benefits of the RIA are balanced by multiple issues not routinely encountered with standard iliac crest bone graft harvesting. First, this procedure, although familiar to many orthopedists, is foreign to the majority of neurosurgeons. However, with proper training, this technique can be easily added to a neurosurgeon’s armamentarium. Strategies for decreasing operative risk involve proper sizing of the reamer head and judicious use of fluoroscopy. The current practice is to select a reamer head no more than 2 mm larger than the smallest diameter of the femoral midshaft isthmus to prevent excessive reaming and cortical thinning. Frequent judicious use of fluoroscopy is necessary for locating and engaging the entry point in the femur, placing the guidewire, measuring canal diameter, and following the progress of the reamer to prevent


errant reaming. Keeping these principles in mind and with repetition and practice, femoral reaming can be safely performed for the acquisition of bone graft material. Femoral reaming carries a specific set of complications separate from the associated spinal operation. Historically, thermal necrosis and microscopic fat emboli have complicated this procedure. The RIA system was specifically designed to reduce intramedullary pressure and heat during reaming, resulting in less thermal damage and a reduction in embolization (10). Additionally, there is a risk of weakening of the femoral shaft secondary to over-reaming, which may result in fracture. After reaming, an intramedullary device is traditionally placed for fracture fixation; thus, iatrogenic fracture is not an issue. However, reaming has been shown to reduce the torsional strength of the femoral shaft (12). Because this procedure is in its infancy with regards to obtaining bone graft, iatrogenic fractures have not been reported in the literature to date. Ultimately, the widespread use of various commercially available bone morphogenic protein products may make autograft and other graft materials a thing of largely historical interest. However, for the near future, autograft will likely remain the workhorse in spinal fusion because of convention, widespread familiarity among spinal surgeons, high fusion rate, and low added risk to the patient. With this in mind, the femoral canal should be considered as an alternative site for autograft acquisition in cases of limited iliac crest bone graft availability.

CONCLUSION Intramedullary femoral reaming in this case was a welltolerated, feasible alternative to iliac crest bone graft for spinal fusion. This technique provides a supply of autologous graft material in patients in whom the standard sites of harvest are inadequate or unavailable and in whom commercially available osteoinductive factors are either unavailable or contraindicated.

Disclosure Timothy G. Weber, M.D., is a paid consultant of Synthes Spine and is engaged in research involving the RIA system.

REFERENCES 1. Arlet V, Jiang L, Steffen T, Ouellet J, Reindl R, Aebi M: Harvesting local cylinder autograft from adjacent vertebral body for anterior lumbar interbody fusion: Surgical technique, operative feasibility and preliminary clinical results. Eur Spine J 15:1352–1359, 2006. 2. Canadian Orthopaedic Trauma Society: Nonunion following intramedullary nailing of the femur with and without reaming. Results of a multicenter randomized clinical trial. J Bone Joint Surg Am 85-A:2093–2096, 2003. 3. David R, Folman Y, Pikarsky I, Leitner Y, Catz A, Gepstein R: Harvesting bone graft from the posterior iliac crest by less traumatic, midline approach. J Spinal Disord Tech 16:27–30, 2003. 4. Forster MC, Aster AS, Ahmed S: Reaming during anterograde femoral nailing: Is it worth it? Injury 36:445–449, 2005. 5. Frölke JP, Nulend JK, Semeins CM, Bakker FC, Patka P, Haarman HJ: Viable osteoblastic potential of cortical reamings from intramedullary nailing. J Orthop Res 22:1271–1275, 2004.


FEMORAL REAMINGS AS AUTOGRAFT

6. Hu RW, Bohlman HH: Fracture at the iliac bone graft harvest site after fusion of the spine. Clin Orthop Relat Res (309):208–213, 1994. 7. McKay B, Sandhu HS: Use of recombinant human bone morphogenetic protein-2 in spinal fusion applications. Spine 27 [Suppl 1]:S66–S85, 2002. 8. Miura Y, Imagama S, Yoda M, Mitsuguchi H, Kachi H: Is local bone viable as a source of bone graft in posterior lumbar interbody fusion? Spine 28:2386–2389, 2003. 9. Morgan SJ, Agudelo JF, Parekh A, Dayton M, Williams A: Preliminary report on the use of the reamer-irrigator-aspirator system (RIA). Presented at the 51st Annual Meeting of the Orthopaedic Trauma Association, Washington DC, February 21–27, 2005. 10. Muller CA, Green J, Sudkamp NP: Physical and technical aspects of intramedullary reaming. Injury 37 [Suppl 4]:S39–S49, 2006. 11. Porchet F, Jaques B: Unusual complications at iliac crest bone graft donor site: Experience with two cases. Neurosurgery 39:856–859, 1996. 12. Pratt DJ, Papagiannopoulos G, Rees PH, Quinnell R: The effects of medullary reaming on the torsional strength of the femur. Injury 18:177–179, 1987. 13. Resnick DK, Choudhri TF, Dailey AT, Groff MW, Khoo L, Matz PG, Mummaneni P, Watters WC, Wang J, Walters BC, Hadley MN, American Association of Neurological Surgeons/Congress of Neurological Surgeons: Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 16: Bone graft extenders and substitutes. J Neurosurg Spine 2:733–736, 2005. 14. Sasso RC, LeHuec JC, Shaffrey C, Spine Interboy Research Group: Iliac crest bone graft donor site pain after anterior lumbar interbody fusion: A prospective patient satisfaction outcome assessment. J Spinal Disord Tech 18 [Suppl]:S77–S81, 2005. 15. Schmidmaier G, Herrmann S, Green J, Weber T, Scharfenberger A, Haas NP, Wildemann B: Quantitative assessment of growth factors in reaming aspirate, iliac crest, and platelet preparation. Bone 39:1156–1163, 2006. 16. Shields LB, Raque GH, Glassman SD, Campbell M, Vitaz T, Harpring J, Shields CB: Adverse effects associated with high-dose recombinant human bone morphogenetic protein-2 use in anterior cervical spine fusion. Spine 31:542–547, 2006. 17. Steffen T, Downer P, Steiner B, Hehli M, Aebi M: Minimally invasive bone harvesting tools. Eur Spine J 9 [Suppl 1]:S114–S118, 2000. 18. Stevens KJ, Banuls M: Iliolumbar hernia following bone grafting. Eur Spine J 3:118–119, 1994. 19. Stevens KJ, Banuls M: Sciatic nerve palsy caused by haematoma from iliac bone graft donor site. Eur Spine J 3:291–293, 1994.

NEUROSURGERY

20. Trinkaus K, Wenisch S, Siemers C, Hose D, Schnettler R: Reaming debris: A source of vital cells! First results of human specimens [in German]. Unfallchirurg 108:650–656, 2005. 21. Vaccaro AR, Chiba K, Heller JG, Patel TCh, Thalgott JS, Truumees E, Fischgrund JS, Craig MR, Berta SC, Wang JC, North American Spine Society for Contemporary Concepts in Spine Care: Bone grafting alternatives in spinal surgery. Spine J 2:206–215, 2002. 22. Walker DH, Wright NM: Bone morphogenetic proteins and spinal fusion. Neurosurg Focus 13:E3, 2002.

COMMENTS

N

ichols et al. have presented a technique for bone graft harvesting that most certainly could play a clinical role in rare cases. Such a clinical circumstance should include at least two criteria. First, all other autograft bone graft sources should have been exhausted. Second, alternatives such as bone morphogenic protein or other bone graft extenders or enhancers should be conceptually eliminated as options. If these parameters can be realized, the technique provided by Nichols et al. may prove to be an invaluable option. For providing this option, the authors are to be congratulated. Edward C. Benzel Cleveland, Ohio

T

he authors present a nicely illustrated case report of cancellous bone harvest through the greater trochanter of the femur. The technique should be able to yield large quantities of high-quality graft material and could be especially useful in situations in which previous iliac crest harvesting had been performed, as in this case report. However, for neurosurgeons unfamiliar with the anatomy of the lower extremities, the risk of vascular or neural injury is not inconsequential as the technique is done with the aid of the fluoroscope. Michael Y. Wang Miami, Florida


PERIPHERAL NERVE Operative Nuances

THE INFRACLAVICULAR APPROACH TO THE BRACHIAL PLEXUS Gabriel C. Tender, M.D. Department of Neurosurgery, Louisiana State University at New Orleans, New Orleans, Louisiana

David G. Kline, M.D. Department of Neurosurgery, Louisiana State University at New Orleans, New Orleans, Louisiana Reprint requests: Gabriel C. Tender, M.D., 3434 Prytania Street, Suite 110, New Orleans, LA 70115. Email: [email protected] Received, February 13, 2007. Accepted, June 11, 2007.

THE INFRACLAVICULAR APPROACH to the brachial plexus is commonly indicated in patients with traumatic injuries and tumors of the brachial plexus elements. We describe the anatomy and operative technique of this approach. KEY WORDS: Brachial plexus, Infraclavicular approach, Nerve action potentials, Sural nerve grafting Neurosurgery 62[ONS Suppl 1]:ONS180–ONS185, 2008

T

he infraclavicular approach is used for lesions involving the brachial plexus distal to the clavicle (1). This approach has been described previously and, hopefully, the online video presentation will aid the novice peripheral nerve surgeon with this often difficult operation. Traumatic injuries, including stretch/contusions, gunshot wounds, lacerations, or iatrogenic injuries to the brachial plexus, may require exploration, nerve action potential recordings, and grafts and/or nerve transfers. Iatrogenic injuries to the infraclavicular plexus are rare and include puncturing of one or more cords or branches during central venous catheterization or brachial plexus anesthetic blocks. The tumors encountered most frequently are schwannomas, neurofibromas, and malignant nerve sheath tumors. Entrapment of the brachial plexus elements at this level is rare.

Surgical Anatomy The normal anatomy of the infraclavicular plexus (1, 5, 6) can be severely altered by scar formation after stretch injuries or previous operations, as well as large tumors. Several anatomic landmarks are particularly important for the surgeon and are presented below. The plexus elements usually encountered at the infraclavicular level are the cords and the origin of the major nerves of the upper extremity. In severe stretch injuries, all or some of the plexus elements may be pulled down inferiorly, and division to cord level may be encountered below the clavicle. The three cords (lateral, medial, and posterior) are named in relation to the axillary artery at the level of pectoralis minor. Anatomic variations are common and definitive identifica-


DOI: 10.1227/01.NEU.0000297012.27846.B9

tion of the individual elements should be undertaken only after most of the infraclavicular plexus is exposed. The lateral cord is usually the first major neural element exposed and runs over the artery from a medial to a lateral direction. The two terminal branches are the musculocutaneous nerve and the contribution to the median nerve. The former gives off branches to the coracobrachialis muscle at the junction of the lateral cord and musculocutaneous nerve and then runs into the biceps muscle. The posterior cord can be found deep to the axillary artery. The thoracodorsal nerve is a relatively small branch that originates from the dorsal aspect of the posterior cord and runs posteriorly to supply the latissimus dorsi muscle. The thoracodorsal nerve usually separates the two subscapularis branches (upper and lower). The two terminal branches are the axillary nerve, which dives down to reach the quadrilateral space, the teres minor and deltoid muscles, and the radial nerve, which runs inferiorly toward the humeral groove. Triceps branches often originate from the proximal radial nerve close to its origin from the posterior cord. One or two subscapularis branches can also originate at this level and/or from the axillary nerve or the thoracodorsal branch. The profunda brachii branch of the axillary artery runs lateral to the radial nerve and medial to the axillary nerve and can be used to identify these two branches. The posterior cord and the axillary nerve are particularly vulnerable to shoulder joint and/or proximal humerus injuries due to their close proximity. The medial cord can be found lateral and somewhat inferior to the axillary vein. Proximal branches from the medial cord


INFRACLAVICULAR APPROACH TO THE BRACHIAL PLEXUS

include the medial pectoral nerves, which join the lateral pectoral nerves from the lateral cord to innervate the pectoralis muscles. The medial brachial and antebrachial cutaneous nerves take off on the ulnar side just before the medial cord divides in its two terminal branches, the ulnar nerve and the contribution to the median nerve. The medial antebrachial cutaneous and ulnar nerves remain medial to the brachial artery as they descend into the upper arm. The contribution to the median nerve crosses the axillary artery from the medial direction to the lateral direction, usually superficial to the artery.

A

Surgical Technique (see video at web site) The infraclavicular approach can be performed either alone (Fig. 1A) or as an extension of the supraclavicular exposure of the brachial plexus (Fig. 1B). The patient, in the supine position, is shifted on the operating table towards the affected side, with the head slightly turned towards the opposite side. The ipsilateral shoulder is elevated with several folded sheets, and the arm is placed on a padded arm board to the side of the patient with only mild abduction. The anterior neck, thorax, shoulder, axilla, and proximal arm are prepared and draped in the usual sterile fashion. If sural nerve grafting is anticipated, one or both legs are similarly prepared. In this case, the leg from which the sural nerve will be harvested must be bent at the knee and medially rotated at the hip in order to provide good exposure of the lateral and posterior sides of the calf. The infraclavicular skin incision is slightly curved and follows the deltopectoral groove and then the medial edge of the biceps muscle (Fig. 1). The plane between the deltoid and pectoralis major muscles can be used for dissection, or, alternatively, the pectoralis major can be split in the direction of its fibers (Fig. 2). The cephalic vein runs longitudinally at this level and can usually be preserved if the dissection is performed inferior and medial to it. Vessels with transverse disposition encountered during exposure can be ligated or coagulated and sharply transected. A Weitlaner retractor can be placed between the deltoid and the pectoralis major muscles to maintain the exposure. The next encountered muscle is the pectoralis minor, which runs transversely to insert on the coracoid process (Fig. 3). This muscle can be isolated by blunt dissection and then transected either by low power electrocautery (Fig. 4) or by sharp dissection followed by bipolar coagulation. The Weitlaner retractor is then repositioned to rest on the divided sides of the pectoralis minor and opened to expose the contents of the axilla. Dissection of the proximal infraclavicular space usually exposes the lateral cord first (Fig. 5). Each element of the plexus is carefully dissected in a circumferential fashion and gently retracted over a Penrose drain to allow for easier proximal and distal dissection. The musculocutaneous nerve takeoff is then identified by following the lateral cord distally. Its site of origin from the lateral cord can be quite variable, and, sometimes, it can even take off after the median nerve formation. Very proximally at its lateral cord origin, the musculocutaneous nerve

NEUROSURGERY

B

FIGURE 1. Drawings illustrating the positioning of the patient and the skin incision for the infraclavicular approach. A, the skin incision follows the deltopectoral groove and then the medial edge of the biceps muscle. B, if supra- and infraclavicular exposure is needed, the skin incision parallels the posterior border of the sternocleidomastoid muscle and angles over the clavicle into the deltopectoral groove. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

usually sends a leash of fine coracobrachialis branches at almost right angles to its course and then dives into the underlying muscle. This is why the musculocutaneous nerve must be dissected through the upper biceps and retracted laterally over a Penrose drain before Weitlaner retractors are placed against the biceps/brachialis muscle. The larger terminal branch of the lateral cord is the contribution to the median nerve, which joins its counterpart from the medial cord to form the median nerve. Shortly after its formation, the median nerve is usually crossed by a transverse vein that can be isolated and transected, although it may appear sizable. Small transverse arteries such as the lateral pectoral or biceps/brachialis vessels can be treated in a similar fashion. The lateral cord also gives one or several pectoralis branches


TENDER AND KLINE

FIGURE 2. Drawing illustrating the pectoralis major muscle being split in the direction of its fibers. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

FIGURE 4. Drawing illustrating the blunt dissection and transection of the pectoralis minor muscle. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

FIGURE 5. Drawing illustrating further exposure of the infraclavicular plexus. The lateral cord and its branches are initially encountered, followed by the axillary artery. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

FIGURE 3. Drawing illustrating the insertion of the pectoralis minor muscle on the coracoid process and the relative position of the underlying brachial plexus elements. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.


that take off from its medial aspect before the bifurcation in the two terminal branches (Fig. 5). The next element to be exposed and mobilized is the axillary artery. Gentle arterial retraction by vasaloops or a vein retractor exposes the underlying and somewhat lateral aspect of posterior cord. Its first branch is the thoracodorsal nerve, which



FIGURE 6. Drawing illustrating the origin and course of the thoracodorsal nerve. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

FIGURE 8. Drawing illustrating the course of the posterior cord and its terminal branches. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

FIGURE 7. Drawing illustrating the complete exposure of the infraclavicular plexus. The two terminal branches of the posterior cord straddle the profunda brachii artery, with the radial nerve medial and the axillary nerve lateral to it, respectively. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.

runs at almost right angles to the cord to reach the deeper latissimus dorsi muscle (Fig. 6). The two terminal branches of the posterior cord straddle the profunda brachii artery (Fig. 7), with the radial nerve running medially to reach the triceps muscle

NEUROSURGERY

FIGURE 9. Drawing illustrating the relative position of the axillary nerve to the shoulder joint. Reproduced from, Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001 (6) with permission from W.B. Saunders Company.


TENDER AND KLINE

(Fig. 8) and the axillary nerve passing laterally under the humeral joint to reach the deltoid muscle (Fig. 9). The last element to be exposed is usually the medial cord. This can be identified by proximally tracing its contribution to the median nerve. Once the medial cord is identified, its other terminal branch, the ulnar nerve, can be found medial and somewhat posterior to the brachial artery. The smaller proximal branches include the medial pectoralis branches, which run transversely to reach the muscle, and the medial antebrachial cutaneous nerves, which run parallel and medial to the ulnar nerve (Fig. 7). Final identification of each neural element is usually possible only after all or most of the elements of the plexus are exposed because anatomic variations are common and stretch injuries or large tumors often distort the normal anatomic relationships. Once exposure of the plexus elements is achieved, intraoperative nerve action potential recordings are performed (7, 8). If nerve transfers and/or nerve grafting are indicated, the affected plexus elements are sectioned and trimmed back to healthy appearing tissue. The proximal and distal stumps are then prepared for grafting by resecting the scarred tissue and creating “fingers” or groups of fascicles that are then sutured proximally and distally to the harvested sural grafts with 7–0 or 8–0 Prolene sutures (Ethicon, Inc., Somerville, NJ). Before closure, meticulous hemostasis is achieved and copious antibiotic irrigation is performed. The pectoralis minor muscle is sometimes but not always reapproximated with Vicryl sutures (Ethicon, Inc., Livingston, United Kingdom), while the split pectoralis major fibers must always be reattached. The subcutaneous tissues are closed with 3–0 resorbable sutures, followed by a 4–0 running subcuticular suture. A soft, mildly compressive dressing is applied for 48 to 72 hours, and the arm is usually placed in a sling for several weeks.

OUTCOMES AND COMPLICATIONS The outcomes after brachial plexus surgery at Louisiana State University have been reported previously (2–4). Stretch injuries are the most frequently encountered lesions, as well as the most difficult lesions to treat. Neurological deficits may improve in the first few months after the traumatic event, but usually stabilize and can be accurately assessed (clinically and electromyographically) at 3 months after injury. The optimal timing for surgical repair of the brachial plexus ranges between 3 and 6 months, although benefits may be obtained for up to 1 year. Outcomes are relatively good for the elements conveying impulses to the large muscles (posterior cord to axillary and radial nerves, lateral cord to musculocutaneous nerve), and suboptimal for the elements innervating the small hand muscles (medial cord to ulnar nerve) (2). Sharp lacerations of the plexus have a better outcome than blunt transections and stretch injuries, especially if the operation is performed within 72 hours and direct reapproximation of the transected nerves is performed (3, 4). The primary intraoperative complications of the infraclavicular plexus approach are vascular (2). In severe stretch injuries,


the axillary/brachial artery and/or vein are intimately attached and very difficult to separate from the scarred plexus elements. Injuries of these major vessels are common and often require the assistance of a vascular surgeon. Of course, there is a risk of damage, or further damage, to elements partially injured or intact preoperatively, particularly when there has been a previous operation or vascular injury and when dissection requires shifting or retraction of elements to reveal the injured ones (e.g., axillary nerve dissection).

CONCLUSION The infraclavicular approach is indicated in most patients with brachial plexus pathology. Surgical exposure in patients with stretch injuries is most difficult because of anatomic distortion and extensive scar formation. Adequate knowledge of the regional anatomic variations and extensive operative experience are helpful in difficult cases.

REFERENCES 1. Kim DH, Cho YJ, Tiel RL, Kline DG: Outcomes of surgery in 1019 brachial plexus lesions treated at Louisiana State University Health Sciences Center. J Neurosurg 98:1005–1016, 2003. 2. Kim DH, Murovic JA, Tiel RL, Kline DG: Infraclavicular brachial plexus stretch injury. Neurosurg Focus 16:E4, 2004. 3. Kim DH, Murovic JA, Tiel RL, Kline DG: Penetrating injuries due to gunshot wounds involving the brachial plexus. Neurosurg Focus 16:E3, 2004. 4. Kim DH, Murovic JA, Tiel RL, Kline DG: Lacerations to the brachial plexus: Surgical techniques and outcomes. J Reconstr Microsurg 21:435–440, 2005. 5. Kline DG, Hudson AR (ed): Brachial Plexus Anatomy and Physiology, in: Nerve Injuries: Operative Results for Major Nerve Injuries, Entrapment, and Tumors. Philadelphia, W.B. Saunders Company, 1995, pp 345–354. 6. Kline DG, Hudson AR, Kim DH: Infraclavicular Plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, W.B. Saunders Company, 2001, pp 19–29. 7. Oberle J, Antoniadis G, Kast E, Richter HP: Evaluation of traumatic cervical nerve root injuries by intraoperative evoked potentials. Neurosurgery 51:1182–1190, 2002. 8. Tiel RL, Happel LT, Kline DG: Nerve action potential recording method and equipment. Neurosurgery 39:103–109, 1996.

COMMENTS

T

ender and Kline continue to add to the operative techniques of peripheral nerve surgery. The infraclavicular approach, the topic of the current report, together with the anterior supraclavicular approach, allows a complete exposure of the brachial plexus for most lesions. The infraclavicular approach allows the exposure of cords, terminal branches, and (with clavicle mobilization and elevation) the divisions. The approach is standard, and the description along with the accompanying illustrations and videos so nicely describe it that much further commentary is not needed. I, therefore, highlight a few points only. I agree with the authors on their statement that the vascular complications are the most common. A suspected vascular injury preoperatively may require the opinion and possibly the assistance of a vascular surgeon. Severe injuries may render the dissection and the identification of neurovascular elements in the scarred region difficult. Vascular injuries are usually related to the axillary artery and vein and their branches more so. As the axillary artery and vein may be intimately related to the posterior surface of the pectoralis minor muscle



in the stretch injury situation, division of this muscle should be done after complete isolation from the surrounding fascia, and it is advisable to peek under the muscle or hold it up with a finger (as illustrated in Figure 4) before dividing it. If resection of the subclavius muscle is needed to allow more proximal exposure, care needs to be taken at its insertion to the inferior surface of the clavicle as a sizable artery and vein can be found and may need to be ligated and divided before they are injured and retracted deeply. We do not reapproximate the pectoralis minor muscle as we believe that the pectoralis major muscle, the overlying fascia, and subcutaneous tissue offer good cover of the repair sites. Rajiv Midha Calgary, Canada

T

ender and Kline provide an excellent guide to the exposure of the infraclavicular plexus. As pointed out by the authors, the essential information presented here has previously been published in Kline et al.’s atlas of peripheral nerve surgery (1). Similarly, the illustrations are largely recycled from this text. The atlas sets a high standard to be met by this article and is the first book that I recommend to residents interested in investing in a reference on how to perform peripheral nerve surgery. Whereas the fundamental techniques of peripheral nerve surgery, soft tissue dissection, external and internal neurolysis, grafting, and nerve action potentials can be taught during residency, the wide variety of approaches to various structures in the peripheral nervous system takes years to master. In light of this fact, any neurosurgeon hoping to incorporate peripheral nerve surgery into their practice should have access to the Atlas. That said, exposures of the brachial plexus above and below the clavicle are the most commonly used approaches other than simple carpal tunnel and ulnar nerve exposures. The demand for wider dissemination of this description justifies the republication of this material. Moreover, much of the “nuance” of this operation can only be grasped by watching the exposure. The accompanying video significantly augments the chapter from the Atlas. Nonetheless, one of the major challenges of peripheral nerve surgery lies in accommodating the frequent anatomical variations that occur naturally as well as in the setting of tumors and trauma. Solving these puzzles provides much of the satisfaction in these patients. Nicholas M. Boulis Cleveland, Ohio

1. Kline DG, Hudson AR, Kim DH: Infraclavicular plexus, in Atlas of Peripheral Nerve Surgery. Philadelphia, WB Saunders, 2001, pp 19–29.

T

he infraclavicular approach is commonly used to address traumatic injuries, iatrogenic injuries, and tumors of the brachial plexus. Despite its varied indications, few surgeons have extensive experience with this approach. Thus, the authors’ extensive experience, docu-

NEUROSURGERY

mented in references 2 to 4, and associated observations are particularly noteworthy. They present a thorough review of the regional anatomy, with specific attention to anatomical landmarks and nuances gained from their experiences with the aim of avoiding complications. They are thorough in outlining details of the operative procedure, from positioning to closure. Most importantly, with their previous experience having added tremendously to the literature, they now highlight relevant outcomes and potential complications. This focus will undoubtedly aid in guiding preoperative, intraoperative, and postoperative management of the affected population. Daniel H. Kim New Orleans, Louisiana

T

his infraclavicular approach by Drs. Tender and Kline along with the previously published supraclavicular and posterior subscapular approaches completes the trilogy on surgical exposures of the brachial plexus. Under normal situations, the infraclavicular approach is relatively quick and simple to perform. The cephalic vein can be retracted, and the internervous plane (deltopectoral interval) is deepened. After taking down the pectoralis minor at its tendinous insertion (for later reapproximation), the surgeon displays the neurovascular elements. Care should be taken during the operative approach to preserve the cephalic vein and thoracoacromial artery, if possible, should secondary reconstruction using free functioning muscle transfer be necessary (such as to provide additional elbow flexion at a later date). Under pathological conditions, the same operative intervention can be treacherous and laborious, especially after previous vascular bypass surgery. Normal planes are distorted. A scarred deltopectoral interval is best identified near the clavicle. A preoperative magnetic resonance angiography scan may be helpful in understanding the vascular anatomy, and intraoperative Doppler ultrasound is indispensable in identifying the axillary artery (“the bouncing nerve”) or the bypass graft. Preplanning may include having a vascular surgeon available as necessary. Exposure may be facilitated by releasing the pectoralis major insertion (and tagging it in anticipation for later reapproximation with heavy sutures). In addition, conjoined tendon can also be released from the coracoid (as long as it is repaired at the end of the case and protected postoperatively) to expose the axillary nerve. The infraclavicular approach is a versatile one. It provides the surgeon with access to the cords and terminal branches that may be needed for nerve repair/reconstruction and/or for intraoperative nerve recordings. For more proximal exposure to the divisions of the brachial plexus, the same incision can be used if the clavicle is mobilized and retracted; the supraclavicular incision can be added for more proximal exposure. For more distal exposure, the extensile approach to the proximal arm can be used. Alternatively, a separate posterior arm exposure can be added to trace the axillary and radial nerves into the proximal arm. All of these approaches (and others) are part of the basic tools of the brachial plexus surgeon. Robert J. Spinner Rochester, Minnesota


PERIPHERAL NERVE Operative Technique

SURGICAL APPROACH TO ULNAR NERVE COMPRESSION AT THE ELBOW CAUSED BY THE EPITROCHLEOANCONEUS MUSCLE AND A PROMINENT MEDIAL HEAD OF THE TRICEPS Olga Gervasio, M.D. Department of Neurosurgery, Bianchi-Melacrino-Morelli Hospital, Reggio Calabria, Italy

Claudio Zaccone, M.D. Department of Neurosurgery, Bianchi-Melacrino-Morelli Hospital, Reggio Calabria, Italy Reprint requests: Olga Gervasio, M.D., Via Cortese n. 3, 89126-Reggio Calabria, Italy. Email: [email protected] Received, January 13, 2007. Accepted, June 11, 2007.

OBJECTIVE: We sought to describe the operative technique in ulnar nerve compression caused by the epitrochleoanconeus muscle and a prominent medial head of the triceps. These anatomic features make the approach to the ulnar nerve at the elbow peculiar and may create technical difficulties during surgical treatment of this area. METHODS: We reviewed patients who underwent surgery for cubital tunnel syndrome between November 1997 and December 2004. The presence of the epitrochleoanconeus muscle with prominent medial head of the triceps occurred in 3.2% of patients. A detailed and illustrated description of the surgical anatomy and the peculiarities of the surgical approach are provided. RESULTS: Epitrochleoanconeus muscle and the prominent portion of the medial head of the triceps were sectioned and removed, and simple decompression of the ulnar nerve was performed. This treatment achieved complete recovery in all of the patients affected by moderate-grade syndrome (Dellon Grade 2 syndrome) who had not shown severe-grade syndrome preoperatively. CONCLUSION: The simple decompression of the ulnar nerve with myotomy or removal of epitrochleoanconeus muscle and the prominent portion of the medial head of the triceps achieved good postoperative results. Experiences from the literature and alternative surgical options are reported. KEY WORDS: Cubital tunnel syndrome, Elbow, Epitrochleoanconeus, Prominent triceps, Ulnar nerve Neurosurgery 62[ONS Suppl ]:ONS186–ONS193, 2008

T

he surgical treatment of cubital tunnel syndrome as it typically occurs is widely described in the literature (2, 8, 15, 25). However, variations of standard muscular anatomy in the region of the medial humeral epicondyle may create technical difficulties during surgical management of this area. In this article, we analyze ulnar nerve compression caused by an aberrant muscle, the epitrochleoanconeus muscle, which occurs frequently, as reported in cadaver and clinical studies (1, 3, 6, 8, 11, 20, 21, 30, 31). In the literature, the nomenclature of this muscle is varied; the terms anconeus-epitrochlearis, subanconeus, or accessory anconeus have also been used to denote the same muscle. There are some reports of ulnar nerve compression at the elbow caused by this muscle (1, 4, 5, 10–12, 14, 17–19, 21, 27–29, 32). In our experience, the epitrochleoanconeus muscle caused compression of the ulnar nerve in 3.2% of the patients who underwent surgery for cubital tunnel syndrome. The presence of this muscle is frequently associated with a prominence of the medial head of the triceps, i.e., the ulnar nerve is completely covered by the triceps up to the


DOI: 10.1227/01.NEU.0000297015.81210.B5

medial humeral epicondyle, at the level of the ulnar canal, and near the olecranon notch. This configuration peculiarity exacerbates the clinical syndrome and the surgical difficulties. In 1986, Dellon (6) demonstrated the presence of the epitrochleoanconeus in 11% of 64 cadaver dissections and of the ulnar nerve beneath the medial head of the triceps in 24% of the dissections. The presence of an association of epitrochleoanconeus muscle with the ulnar nerve completely covered by the medial head of the triceps was statistically significant (P ⬍ 0.001). Another noteworthy feature was a frequent absence of the band of Osborne in this pattern of anatomic variance (6). These anatomic features complicate the approach to the ulnar nerve, and the operative nuances warrant an exhaustive description.

CLINICAL PRESENTATION The clinical picture of this peculiar compressive syndrome is similar to that of common cubital tunnel syndrome. The most common clinical presentation is a moderate-grade syndrome


ULNAR NERVE COMPRESSION CAUSED BY EPITROCHLEOANCONEUS AND TRICEPS

such as Dellon syndrome Grade 2 (7) or equivalent. This grading is probably the result of a slow and progressive mechanism of compression. Muscular hypertrophy, caused by work or sports activities, increases progressively, causing compression of the underlying ulnar nerve. Rarely, the hypertrophy is acute, occurring in muscular edema because of hyperactivity (14). The only peculiar sign of a presence of epitrochleoanconeus muscle is “fullness” of the ulnar groove when examining the elbow. Unfortunately, this characteristic sign is not always present or appreciable; frequently, the muscular anomaly is an operative finding. A clinical history of intense muscular activity (in work or sports) in young patients with cubital tunnel syndrome without other risk factors may sometimes be useful.

PREOPERATIVE WORKUP When the presence of an aberrant muscle is suspected clinically, a radiological investigation is helpful before the surgical procedure. Ultrasonography may show the muscle, which is visible on axial images as a thin hypoechoic lesion on the ulnar nerve (22). Axial magnetic resonance imaging (MRI) is used to identify the epitrochleoanconeus extending from the medial epicondyle to the medial aspect of the olecranon and permits measurement of the diameter and recognition of the size and shape of the muscle. Coronal and sagittal slices allow the identification of muscle position and the measurement in craniocaudal length. Short time inversion recovery images reveal the features of the muscle edema, if present. The relationships among the ulnar nerve, the muscle, and the adjacent soft tissue may be demonstrated with the use of MRI (14). We do not perform an ultrasound screening, because MRI scans provide superior soft-tissue detail, differentiating determinate from indeterminate lesions and providing a diagnosis with a highdegree of certainty. The advantages of MRI include the visualization of the anatomy in multiple planes; knowledge of the anatomic factors is useful for surgical planning, and the surgeon may modify the initial approach to treatment (e.g., an endoscopic procedure may be converted to an open procedure). These muscular anomalies can be readily managed when discovered intraoperatively, but preoperative awareness of anomalies permits the formulation of better surgical planning and optimum organization of the operating room time. A complete preoperative diagnosis permits a more complete informed consent to the surgical procedure to be obtained, and the possibility of informing patient and anesthetist on the modalities of anesthesia. Encountering the variation described during a procedure performed under local anesthesia or sedation may not necessarily require conversion to general anesthesia because extra local anesthesia or sedation may be provided as needed. However, receiving additional anesthesia may be distressing to the patient. Indeed, during our first two procedures (Patients 1 and 2) (Table 1), in which the muscle was an intraoperative finding, extra local anesthesia and sedation were obtained, and psychologically, one of the patients was left with a disagreeable impression of the surgery.

NEUROSURGERY

SURGICAL ANATOMY Standard Anatomy of the Cubital Tunnel The ulnar nerve, which is the terminal branch of the brachial plexus medial cord, approaches the elbow, piercing the medial intermuscular septum (MIS) at approximately the midpoint of the arm, and enters the posterior aspect of the arm. The ulnar nerve then travels behind MIS on the medial head of the triceps and enters the postcondylar groove lateral to the medial epicondyle. At the elbow, the ulnar nerve enters the forearm between the medial epicondyle and olecranon through the cubital tunnel. The roof of the cubital tunnel is a fibrous aponeurosis that thickens to form the cubital tunnel retinaculum or Osborne’s band (also called the ligament of Osborne) (23), which connects the tendinous origin of the humeral and ulnar heads of the flexor carpi ulnaris muscle (FCU). The distal margin blends with the fascia covering the humeral and ulnar heads of the FCU. Osborne’s band is approximately 4 mm to 2 cm wide (from proximal to distal ends) and extends from the medial epicondyle to the tip of the olecranon. The fibers are oriented transversely and are taut in elbow flexion. The cubital tunnel retinaculum (Osborne’s band) is anatomically and functionally discrete from the FCU aponeurosis (8, 20). In some surgery textbooks, Osborne’s band is reported as the proximal thickened edge of the FCU aponeurosis (25); other authors also report an extra structure more proximal to FCU aponeurosis, called the epitrochleoanconeus ligament. This ligament covers the ulnar nerve at this level (2). The medial collateral elbow ligaments and the joint capsule form the floor of the cubital tunnel, whereas the medial epicondyle and olecranon form the walls. Within the tunnel, the ulnar nerve branches to the elbow joint. Exiting the tunnel, the ulnar nerve passes into the forearm between the humeral and ulnar heads of the FCU and branches to the FCU.

Patterns of Anatomic Variation The pattern described previously is the most frequent anatomic variation: the nonsubluxing ulnar nerve approaching the medial humeral epicondyle beneath a thin fascia extending from a normal medial head of the triceps into the MIS, and the presence of Osborne’s band covering the ulnar nerve entering the flexor carpi ulnaris (Fig. 1). Nevertheless, there are two other patterns that are also frequent, as already described by Dellon (6) in 1986; the second is characterized by ulnar nerve subluxation, the medial head of the triceps entering the floor of the cubital tunnel, and the presence of Osborne’s band. The third most common pattern is related to the presence of the epitrochleoanconeus muscle, arising from the medial border of the olecranon and inserting into the medial epicondyle, located near but proximal to the origin of the two heads of the FCU. Proximally to the epitrochleoanconeus, the prominent medial head of the triceps completely covers the ulnar nerve up to medial humeral epicondyle, at the level of the ulnar canal and near the olecranon notch; distally, Osborne’s band is absent (Fig. 2). The epitrochleoanconeus is sometimes incomplete,


GERVASIO AND ZACCONE

TABLE 1. Patient characteristics Dellon syndrome preoperative Grade (7)

Preoperative magnetic resonance imaging scan

Prominent triceps medial head

Tennis player

2

No

Present

Local with sedation

Body builder

2

No

Present

Local with sedation

Right

Photocopying worker

2

Yes

Present

General

42/M

Left

Bricklayer

2

Yes

Present

General

31/M

Right

Painter

2

Yes

Present

General

Patient no.

Age (yr)/sex

Arm

1

33/M

Right

2

34/M

Right

3

31/M

4 5

Occupation or hobby

with mixed muscular and fibrous structure. Thus, the epitrochleoanconeus ligament described in some textbooks may be identified as an epitrochleoanconeus muscle with mainly fibrous structure.

Surgical Relevance of the Anatomic Pattern Including Epitrochleoanconeus Muscle In this anatomic configuration, it is difficult to identify the ulnar nerve proximally because it is covered by the fibers of the prominent medial head of the triceps and, traveling toward the ulnar canal, the epitrochleoanconeus muscle. On the other hand, because of the frequent absence of Osborne’s band, it is easier to identify the ulnar nerve distally, at the confluence of the two heads of the FCU, where it is uncovered until its entry in the FCU, below the fascia of the same. These possible anatomic variations in the cubital tunnel region and their involvement in the

FIGURE 1. Schematic drawing of the most frequent anatomic pattern of the cubital tunnel. The ulnar nerve (UN) lies behind the medial intermuscular septum (MIS) and on a normal medial head of the triceps brachii (TB, medial head). The ulnar nerve enters the postcondylar groove between the medial epicondyle (ME) and olecranon (OL). Osborne’s band (OB) connects the tendinous origin of the humeral and ulnar heads of the flexor carpi ulnaris muscle (FCU) and covers the ulnar nerve entering the FCU, below its fascia. Drawing by Olga Gervasio, M.D.


Anesthesia

surgical exposure are summarized in Table 2.

OPERATIVE TECHNIQUE We prefer to perform the surgical operation with general anesthesia, because the muscular section is painful and operative time is longer compared with the standard procedure of simple decompression. Local anesthesia can be used only with very cooperative patients, with the possible addition of general sedation. A possible alternative is regional block. The first steps of the procedure are similar to a standard simple decompression, which has already been described by the authors in a previous report (9): a slightly curvilinear longitudinal skin incision

FIGURE 2. Schematic drawing of the anatomic pattern of the cubital tunnel in presence of epitrochleoanconeus muscle. The prominent medial head of the triceps (TB, medial head) covers the ulnar nerve (UN) lying behind the medial intermuscular septum (MIS) and entering the postcondylar groove. The epitrochleoanconeus muscle (EA) arises from the medial border of the olecranon (OL) and inserts into the medial epicondyle (ME). The muscle lies near but proximal to the origin of the two heads of the flexor carpi ulnaris muscle (FCU) and covers the ulnar nerve at the postcondylar groove. Osborne’s band is absent. Drawing by Olga Gervasio, M.D.



TABLE 2. Anatomical variations in the cubital tunnel regiona Osborne’s band

Luxation or subluxation of ulnar nerve

Normal

Present

No

Proximally, on the triceps

From proximal to distal

Absent

Entering into floor of the cubital tunnel

Present

Yes

Proximally, on the triceps

From proximal to distal

11

Present

Prominent

Absent

No

Distally, at the FCU confluence

From distal to proximal

15

—

—

—

—

Pattern no.

Frequencyb (%)

Epitrochleoanconeus

1

53

Absent

2

21

3 Miscellaneousc

Triceps medial head

—

Ulnar nerve identification

Optimal direction of ulnar nerve decompression

—

a

FCU, flexor carpi ulnaris. Values were calculated by the authors on the basis of approximate averages of values reported in cadaveric studies in the literature. c Other patterns, including the same described anatomic variations, but different in combinations or other anatomic variations not described in this article. b

is performed (without tourniquet control) crossing an imaginary line between the medial epicondyle and the olecranon, and extending approximately 4 cm proximally and 4 cm distally. The skin incision is placed 2 or 3 cm below the cutaneous projection of the nerve course to avoid contact between the surgical wound and the nerve and the development of fibrous adherences between the nerve and the scar. This procedure also permits the formation of a cutaneous and subcutaneous protective flap over the nerve. The subcutaneous tissue is carefully dissected to identify and preserve the medial cutaneous nerve of the forearm. At this point, instead of exposing the ulnar nerve proximally (the nerve is included in the muscular mass of the triceps), the same nerve is identified more distally, at the level of the confluence of the two heads of the FCU, where it is uncovered for a short section before entering the FCU below its fascia (Fig. 3). Subsequently, the nerve is released from the distal to the proximal while sectioning in this direction the epitrochleoanconeus muscle that is detached from the medial epicondyle and debulked, and also sectioning the fibers of the prominent medial head of the triceps (Fig. 4). The superior surface of the nerve is continuously protected during this procedure. The prominent portion of the medial head of the triceps is separated through a microsurgical dissection from the ulnar nerve and then cut off and removed. The epitrochleoanconeus muscle is excised, and the residual part is coagulated (Fig. 5). Nerve decompression is completed by performing an accurate external neurolysis (paraneurectomy), thereby freeing the nerve in a 360-degree fashion. The skin is retracted proximally and distally, and the nerve is released to the MIS proximally and to the deep flexor-pronator aponeurosis distally. The arcade of Struthers and the MIS are not divided because they rarely compress the ulnar nerve unless it has been previously transposed. Actually, the section of the arcade and the MIS, together with the removal of the epitrochleoanconeus, may cause an iatrogenic subluxation of the nerve. For the same reason, and also to avoid muscular weakness postoperatively, the

NEUROSURGERY

FIGURE 3. Intraoperative photograph (right arm) illustrating an anatomic pattern of cubital tunnel with epitrochleoanconeus muscle. The ulnar nerve (U) is visible only at the level of the confluence of the two heads of the flexor carpi ulnaris muscle (FCU). More proximally, at the level of the medial epicondyle (ME), the ulnar nerve is covered by the prominent medial head of the triceps brachii (TB, mh) and by epitrochleoanconeus muscle (EA). The arrowheads indicate the entrance of the ulnar canal covered by the fibers of the prominent medial head of the triceps, and black lines underline the borders of the epitrochleoanconeus.

confluence of the two heads of the FCU is gently separated, not sectioned. Accurate hemostasis is performed by bipolar coagulation. The more delicate stages of the surgical operation are performed with an operative microscope (Contraves; Carl Zeiss Co., Oberkochen, Germany). We prefer the microscope for better magnification and lighting of the operative field, but this surgical operation may also be performed successfully with the use of loupes. The surgical wound is usually closed with interrupted sutures. A soft, antiseptic dressing is applied to the elbow, which is rested for 7 days. The cutaneous sutures are



of the muscle; thus, they prevent the ulnar nerve from the ulnar canal relocating in subcutaneous position. For the same reason, other authors (12) perform a medial epicondylectomy to prevent the irritation of the ulnar nerve on the medial epicondyle; thus, they carry out an osteotomy of the epicondyle.

CLINICAL CASES

FIGURE 4. Intraoperative photograph (same patient as Fig. 3) showing the epitrochleoanconeus (EA) detached and debulked. Black lines underline the borders of the detached muscle; black arrows indicate the narrowing of the ulnar nerve (U) where it was covered by the muscle. The fibers of the prominent portion of the medial head of the triceps (TB, mh) were sectioned, freeing the ulnar nerve. The arrowheads show the sectioned fibers of the triceps, which covered the entrance of the ulnar canal in the previous figure. FCU, flexor carpi ulnaris muscle; ME, medial epicondyle.

We reviewed the features of patients who underwent operations for cubital tunnel syndrome at the Department of Neurosurgery in Reggio Calabria from November 1997 to December 2004; an epitrochleoanconeus muscle with a prominent medial head of the triceps was present in 3.2% of patients. Among 156 patients who underwent surgery, five were affected with this peculiar compressive syndrome. All of these patients were young men. As is most often reported in the literature, at the time of admission, all of our patients were affected with a moderate-grade syndrome (Dellon syndrome Grade 2) (7). The characteristics of our patients are summarized in Table 1. MRI scans were performed preoperatively in three of the five patients; in two patients, the muscular anomalies were discovered during surgical operation. Anesthesia was local with sedation for two patients and general for three. All patients were subjected to excision of the epitrochleoanconeus muscle and the prominent portion of the medial head of the triceps and to simple decompression of the ulnar nerve (external neurolysis). None of the patients had transposition of the ulnar nerve. This treatment resulted in complete recovery in all patients. However, the patients were affected with moderate-grade syndrome (Dellon syndrome Grade 2) (7); none of the patients had a severe grade syndrome preoperatively.

DISCUSSION The Epitrochleoanconeus Muscle and the Prominent Medial Head of the Triceps in the Literature

FIGURE 5. Intraoperative photograph (same patient as Fig. 3) showing the removal of the prominent portion of the medial head of the triceps (TB, mh), displaced for subsequent cutting. The epitrochleoanconeus muscle (EA) was excised and the residual part was coagulated (encircled). Black arrows indicate the narrowing of the ulnar nerve (U) where it was covered by the epitrochleoanconeus. FCU, flexor carpi ulnaris muscle; ME, medial epicondyle.

removed 15 to 18 days after surgery. Manual laborers return to work approximately 21 days after surgery. Some authors (4, 11, 31) prefer to complete the surgical procedure by anterior subcutaneous transposition because of fear of the subluxation of the ulnar nerve over the medial epicondyle after the removal


The epitrochleoanconeus muscle was first described by Wood (33) in 1868 and, later by LeDouble (16) in 1897 in their works about anomalous muscles in humans. In the world literature, the presence of this muscle in cadaver elbows is reported with a variable incidence ranging from 1 to 30% in different cadaveric series (3, 8, 11, 30, 31). Few clinical studies have examined the incidence of this muscle in the patients who undergo surgery for cubital tunnel syndrome; ulnar nerve compression due to epitrochleoanconeus muscle is reported to range from 5 to 16% in the operative series (4, 18, 28, 29). In our experience, the epitrochleoanconeus muscle has been the cause of ulnar nerve compression in 3.2% of the patients operated on for cubital tunnel syndrome. In all of our patients, this muscle was associated with a prominent medial head of the triceps. To the best of our knowledge, there is only one report of ulnar nerve compression attributable to an association of these two muscular anomalies in the literature (21), but the prominence of the medial head of the triceps is present in the anatomic pattern, including epitrochleoanconeus muscle, as demonstrated



by Dellon (6). Therefore, the association with the prominence of the medial head of the triceps is probably underestimated.

Choice of Surgical Approach It has been debated whether simple decompression of the ulnar nerve with myotomy or excision of epitrochleoanconeus muscle and the prominent portion of the medial head of the triceps could have been effective. In the literature, this procedure is preferred by 67% of the authors (1, 5, 14, 19, 21, 22, 27, 29). The accomplishment of anterior transposition or medial epicondilectomy is preferred by 33% of the authors (4, 11, 12, 31). The authors who performed simple decompression with myotomy or muscular excision alone reported good postoperative results (1, 5, 14, 19, 21, 22, 27, 29). Chalmers (4), Hirasawa et al. (11), and Wachsmuth and Wilhelm (31) performed an anterior subcutaneous transposition because they feared an ulnar nerve subluxation over the medial epicondyle after the removal of epitrochleoanconeus muscle. However, ulnar nerve subluxation is often asymptomatic, as demonstrated by its high incidence in normal cadaver elbows (6). Moreover, it is possible to avoid subluxation by sparing the sections of the MIS and the FCU confluence during simple decompression. Hodgkinson and McLean (12) performed medial epicondylectomy to reduce the irritation of the ulnar nerve on the medial epicondyle after the removal of the epitrochleoanconeus muscle, but this procedure may produce persistent pain and tenderness at the exposed bony surface from which the epicondyle has been removed and the possibility of heterotopic ossification. A good alternative might be submuscular transposition, which includes placing the nerve in a protected position, but this procedure requires extensive tissue dissection and prolonged postoperative elbow immobilization. In our opinion, if a clear compression of the ulnar nerve caused by epitrochleoanconeus muscle and/or prominent portion of the medial head of the triceps is identified during surgical exploration, the need for anterior transposition or medial epicondilectomy may be avoided. The mechanism of nerve compression may implicate hypertrophic changes in muscular bulk caused by intense activity during the patient’s life. Chalmers (4) identified the epitrochleoanconeus muscle as a potential compressive agent because it shows the same relationship that Osborne’s band shares with the ulnar nerve: it overlies the nerve and becomes taut in flexion and relaxed in extension of the elbow. In the absence of pathology of the ulnar canal, compression is caused by the presence of the muscle or its hypertrophy. Therefore, its removal will be sufficient to improve the ulnar neuropathy. The removal of the prominent part of the medial head of the triceps does not seem to compromise triceps function (26).

Technical Difficulties of the Surgical Approach At the elbow, the ulnar nerve covered by epitrochleoanconeus muscle associated with a prominent medial head of the triceps represents the third most common pattern of anatomic variance (6). In cubital tunnel surgery, it is possible to find this anatomic configuration. In this case, the roof of the ulnar canal

NEUROSURGERY

is muscular rather than fibrous. The technical difficulty of the surgical approach compared with the standard approach is to identify the ulnar nerve. In fact, it is preferable to identify the ulnar nerve at the level of the confluence of the two heads of the FCU, where a short segment of the nerve is uncovered, as Osborne’s band is absent in this anatomic pattern (6). After the excision of epitrochleoanconeus muscle, the nerve is decompressed in a distal to proximal direction and separated by microsurgical dissection from the prominent medial head of the triceps. The prominent part of the medial head of the triceps is removed without compromising triceps function. These characteristics make this approach more complex than the standard approach. When this anatomic configuration is suspected clinically, an MRI scan is helpful so that adequate planning of anesthesia and surgical procedure is feasible. Moreover, the presence of the epitrochleoanconeus muscle is reported as a technical difficulty in the endoscopic approach to the cubital tunnel (13). Access to the cubital tunnel is restricted by the muscle mass, so it may be necessary to enlarge the incision.

Phylogenetic Considerations During ulnar nerve dissections of subhuman primates, Dellon (6) demonstrated that in nonanthropoid apes such as the baboon, the ulnar nerve is within the triceps and the epitrochleoanconeus muscle is present. In anthropoid apes, the ulnar nerve is more superficial, covered by a thin sheet of triceps medial head fibers blending into the MIS, the epitrochleoanconeus muscle is present, and Osborne’s band is not well developed (6, 24). During evolution, the medial head of the triceps regressed in humans, leaving a thin layer of fascia from the medial head of the triceps to the MIS, with the ulnar nerve lying below this thin film of fascia. The epitrochleoanconeus muscle became so thin that the fibrous band described by Osborne may be identified as the remnant of the muscle (6, 20). Comparative anatomy studies such as the treatise of Padoa (24) clarified that the epitrochleoanconeus muscle and the prominent medial head of the triceps strengthened elbow adduction and extension for brachiation in climbing. In humans, this function is not required; thus, the epitrochleoanconeus became a potential compressive agent, as did the prominence of the medial head of the triceps.

CONCLUSION The variations of muscular anatomy in the region of the medial humeral epicondyle make the surgical approach to the ulnar nerve at the elbow peculiar. The awareness of possible anatomic patterns and consequent surgical approaches is useful and conveys improved safety for surgical procedures.

REFERENCES 1. Assmus H: Simple decompression of the ulnar nerve in cubital tunnel syndrome with and without morphologic changes. Report of experiences based on 523 cases [in German]. Nervenarzt 65:846–853, 1994. 2. Brunelli G: Ulnar nerve compression at the elbow, in Brunelli GA (ed): Nerve Lesions of the Upper Limb [in Italian]. Pavia, EDIMES, 2004, pp 91–94.



3. Campbell WW, Pridgeon RM, Riaz G, Astruc J, Sahni KS: Variations in anatomy of the ulnar nerve at the cubital tunnel: Pitfalls in the diagnosis of ulnar neuropathy at the elbow. Muscle Nerve 14:733–738, 1991. 4. Chalmers J: Unusual causes of peripheral nerve compression. Hand 10:168–175, 1978. 5. Dahners LE, Wood FM: Anconeus epitrochlearis, a rare cause of cubital tunnel syndrome: A case report. J Hand Surg [Am] 9:579–580, 1984. 6. Dellon AL: Musculotendinous variations about the medial humeral epicondyle. J Hand Surg [Br] 11:175–181, 1986. 7. Dellon AL: Review of treatment results for ulnar nerve entrapment at the elbow. J Hand Surg [Am] 14:688–700, 1989. 8. Doyle JR: Elbow, in Doyle JR, Botte MJ (eds): Surgical Anatomy of the Hand and Upper Extremity. Philadelphia, Lippincott Williams & Wilkins, 2003, pp 365–406. 9. Gervasio O, Gambardella G, Zaccone C, Branca D: Simple decompression versus anterior submuscular transposition of the ulnar nerve in severe cubital tunnel syndrome: A prospective randomized study. Neurosurgery 56:108– 117, 2005. 10. Gessini L, Jandolo B, Pietrangeli A, Occhipinti E: Ulnar nerve entrapment at the elbow by persistent epitrochleoanconeus muscle. Case report. J Neurosurg 55:830–831, 1981. 11. Hirasawa Y, Sawamura H, Sakakida K: Entrapment neuropathy due to bilateral epitrochleoanconeus muscles: A case report. J Hand Surg [Am] 4:181–184, 1979. 12. Hodgkinson PD, McLean NR: Ulnar nerve entrapment due to epitrochleoanconeus muscle. J Hand Surg [Br] 19:706–708, 1994. 13. Hoffmann R, Siemionow M: The endoscopic management of cubital tunnel syndrome. J Hand Surg [Br] 31:23–29, 2006. 14. Jeon IH, Fairbairn KJ, Neumann L, Wallace WA: MR imaging of edematous anconeus epitrochlearis: Another cause of medial elbow pain? Skeletal Radiol 34:103–107, 2005. 15. Kline DG, Hudson AR, Kim DH: Ulnar nerve, in Kline DG, Hudson AR, Kim DH (eds): Atlas of Peripheral Nerve Surgery. Philadelphia, W. B. Saunders Company, 2001, pp 77–85. 16. LeDouble AF: Treatise on Human Muscular System Variations and Their Meaning from a Zoological Anthropological Point of View [in French]. Paris, Schleicher Freres, ed 2, 1897, pp 60–75. 17. Mabin D, Chevalier F, Tea S, Le Saux D: Cubital nerve compression at the elbow by a supernumerary muscle [in French]. Presse Méd 12:2947, 1983. 18. MacNicol MF: The results of operation for ulnar neuritis. J Bone Joint Surg Br 61B:159–164, 1979. 19. Masear VR, Hill JJ, Cohen SM: Ulnar compression neuropathy secondary to the anconeus epitrochlearis muscle. J Hand Surg [Am] 13:720–724, 1988. 20. O’Driscoll SW, Horii E, Carmichael SW, Morrey BF: The cubital tunnel and ulnar neuropathy. J Bone Joint Surg Br 73:613–617, 1991. 21. O’Hara JJ, Stone JH: Ulnar nerve compression at the elbow caused by a prominent medial head of the triceps and an anconeus epitrochlearis muscle. J Hand Surg [Br] 21:133–135, 1996. 22. Okamoto M, Abe M, Shirai H, Ueda N: Diagnostic ultrasonography of the ulnar nerve in cubital tunnel syndrome. J Hand Surg [Br] 25:499–502, 2000. 23. Osborne GV: Surgical treatment of tardy ulnar neuritis. J Bone Joint Surg 39B:782, 1957. 24. Padoa E: Systematics of vertebrates, in Padoa E (ed): Manual of Comparative Anatomy of Vertebrates [in Italian]. 2002, pp 27–61. 25. Rengachary SS: Entrapment neuropathies, in Wilkins RH, Rengachary SS (eds): Neurosurgery (Italian Edition). New York, McGraw-Hill, 1987, vol 3, pp 1–28. 26. Spinner RJ, O’Driscoll SW, Jupiter JB, Goldner RD: Unrecognized dislocation of the medial portion of the triceps: Another cause of failed ulnar nerve transposition. J Neurosurg 92:52–57, 2000. 27. Sucher E, Herness D: Cubital canal syndrome due to subanconeus muscle. J Hand Surg [Br] 11:460–462, 1986. 28. Suden R, Wilhelm A: Proximal ulnar nerve compression syndrome with special reference to the m. epitrochleo-anconaeus [in German]. Handchir Mikrochir Plast Chir 19:33–42, 1987. 29. Vanderpool DW, Chalmers J, Lamb DW, Whiston TB: Peripheral compression lesions of the ulnar nerve. J Bone Joint Surg [Br] 50:792–803, 1968.


30. Von Clemens HJ: About the morphology of epitrocleo anconeus ligament [in German]. Anat Anz 104:343–344, 1957. 31. Wachsmuth W, Wilhelm A: The muscle epitochleoanconaeus and its clinical significance [in German]. Monatsschr Unfallheilkd Versicher Versorg Verkehrsmed 71:1–22, 1968. 32. Wadsworth TG: The external compression syndrome of the ulnar nerve at the cubital tunnel. Clin Orthop Relat Res May:189–204, 1977. 33. Wood J: Variations in human mycology observed during the winter session of 1867–1868 at King’s College, London. Proc R Soc Lond 16:483–525, 1868.

Acknowledgments We thank the translators Giuseppe Gervasio and Liam MacGabhann for their contribution to the preparation of the manuscript.

COMMENTS

I

n this technical review, Gervasio and Zaccone discuss the incidence and relevance of different muscular variations near the retrocondylar groove, including the anconeus muscle and a prominent medial head of the triceps. Their illustrations and operative photographs are excellent, as is their thoughtful analysis presented in Table 2. A reiteration of the vestigial nature of the anconeus muscle is always welcome—this remains a crowd-pleaser with medical students and residents. It is uncertain whether these muscular variations per se compress the ulnar nerve or even predispose it to an entrapment near the elbow. Perhaps they only represent normal muscular variations, presumably more common in weightlifters/heavy workers, which are incidentally encountered during surgery. A study comparing the prevalence of these variations in asymptomatic patients (e.g., by reviewing elbow magnetic resonance imaging scans performed for other indications) to that observed during ulnar nerve decompression would shed some light on this issue. I routinely perform a simple, 180-degree ulnar nerve decompression for this entrapment, using a 1-inch incision near the retrocondylar groove. Once the posterior division of the medial antebrachial cutaneous nerve is identified and preserved, I next palpate the ulnar nerve in or just proximal to the retrocondylar groove, where it is readily identified, either immediately or after some deeper dissection in obese or muscular patients. Once uncovered, I unroof this nerve both distally and proximally, including Osborne’s band (when present), the flexor carpi ulnaris, and the distal flexor-pronator fascia, as well as any muscular bands proximally. I confirm decompression by freely passing my index finger along the ulnar nerve in either direction. If muscular variations are encountered, they are simply released or removed, with or without some extra local or sedation. Considering preoperative magnetic resonance imaging would not change my operative or anesthetic technique, I do not think it is necessary in most patients. Unless the patient has a very sensitive ulnar nerve before surgery, I do not routinely transpose the nerve if intraoperative subluxation is observed after simple decompression. Stephen M. Russell New York, New York

I

n this technical report, the authors provide a detailed description of their surgical approach to a recognized variant in the anatomy of the cubital tunnel region. The epitrochleoanconeus muscle has been known for many years to be a potential site of compression of the ulnar nerve at the elbow. The association of this variant with a prominent medial head of the triceps is clearly described, and the challenges of operative exposure are well detailed. Because this is a very uncommon variant (3.2% of the authors’ series) and its recognition is difficult on the basis



of physical examination, it is likely that the majority of cases will be discovered at the time of surgical intervention. If the surgeon finds it challenging to identify the ulnar nerve in its usual location just proximal to the medial epicondyle and the prominent medial triceps head is covering the nerve, then the authors’ suggestion to find the nerve more distally first can be quite helpful. However, the technique of dissecting a nerve distally to proximally can be hazardous, placing branches at risk. Good lighting is essential and operative loupes are quite helpful (although we rarely use the microscope in these instances unless nerve repair is performed). A few details of management of the entrapped ulnar nerve differ slightly. As the authors have demonstrated in other work, transposition is rarely necessary in routine cases, unless it is a reexploration or there is an anatomic deformity of the elbow joint, e.g., after fracture and/or dislocation. I generally just unroof the nerve rather than perform a full 360-degree external neurolysis and find that symptomatic subluxation of the nerve rarely occurs. We check for subluxation of the nerve after decompression by flexing and extending the elbow and directly observing the nerve prior to closing the incision. Epicondylectomy is generally not necessary and occasionally can be harmful. Eric L. Zager Philadelphia, Pennsylvania

T

he authors have prepared a case series comprised of patients with cubital tunnel syndrome requiring surgical intervention. The source of compression identified in all patients is the anconeus epitrochlearis muscle and an associated prominent medial head of the triceps. They present valuable demographic information regarding potential exacerbating activities, helpful diagnostic radiographs, and a novel surgical technique. The authors have given a thorough historical review of the data and note a varying incidence of the anconeus epitrochlearis muscle alone as a source of ulnar compression, ranging from 1% to 34% in cadaveric studies. The incidence of this variant manifesting clinically and requiring surgical intervention is not noted. The combined presence of anconeus epitrochlearis muscle and an associated prominent medial head of the triceps is admittedly rare: only one report documenting a single case was previously published. No additional data regarding the incidence of this variant manifesting clinically and requiring surgical intervention are available. Thus, this current report of the five patients, all with this anatomic variant, is noteworthy. The primary goal of this report is to present a novel surgical approach. The authors describe microscopic separation and later removal of the prominent portion of the medial head of the triceps to decompress the ulnar nerve. This process is in conjunction with excision of the epitrochleoanconeus muscle followed by accurate external neurolysis. This method/approach is in contrast to other reported methods that include transposition or medial epicondylectomy. Given the reported rarity of this compressive syndrome, it is difficult to understand previous comparisons of surgical techniques in terms of outcomes. Thus, the efficacy of the reported technique cannot be entirely endorsed. Nonetheless, the authors have compiled a cases series that offers descriptive anatomy, effective strategies for diagnosis, and a novel surgical technique to address this compressive syndrome. This report adds significantly to the existing literature regarding this subject. Odette Harris Atlanta, Georgia David G. Kline New Orleans, Louisiana Daniel H. Kim Houston, Texas

NEUROSURGERY

I

have always been interested in anatomic variations, especially as they pertain to the peripheral nerve. Since grade school, I was fortunate to be exposed to clinical and comparative anatomy by my father, as well as his teacher and “second father” Emanuel B. Kaplan, especially as it pertained to peripheral nerves. Over the years, I have observed and described causal relationships of several accessory muscles and nerve compression. I also spent several years researching the role of the medial head of the triceps in a clinical entity called snapping of the triceps and ulnar neuropathy (3). Thus, I was pleased to read this article on ulnar nerve compression at the elbow attributable to an anconeus epitrochlearis muscle associated with a prominent medial head of triceps. The anconeus epitrochlearis is a curious muscle. Although it is present in a small percent of human elbows, its function (i.e., an adductor in other primates) and derivation (i.e., thought to be an extension of the medial triceps) remain uncertain. Its innervation by the ulnar nerve (nota bene, a favorite examination question among hand surgeons) is poorly understood. A beautiful description of the cubital tunnel retinaculum as a remnant of the anconeus epitrochlearis that was put forth by some friends at Mayo Clinic has added to my own understanding of this muscle and its anatomy (2). Entrapment of the ulnar nerve by this muscle has been limited to scattered case reports and small series in the literature. I encounter an anconeus epitrochlearis muscle during ulnar nerve surgery every 18 months or so. Although I was familiar with Lee Dellon’s cadaveric study (1), I had not appreciated the surgical relevance of the association of it with a prominent medial triceps before I reviewed this manuscript. Several weeks ago, I happened to encounter this muscle during an ulnar nerve decompression in a 70-year-old man. Armed with the information that the authors had taught me, I was delighted to see exactly what these authors have reported. I include the intraoperative photographs of my case to reinforce the authors’ findings (Fig. 1). As you will see, my operative pictures appear nearly identical to theirs (apart from the fact these left-sided figures are mirror images of the authors’ right-sided ones). I congratulate the authors on providing some clinically relevant anatomic information that I have already observed and applied in my own practice. Robert J. Spinner Rochester, Minnesota 1. Dellon AL: Musculotendinous variations about the medial humeral epicondyle. J Hand Surg [Br] 11:175–181, 1986. 2. O’Driscoll SW, Horii E, Carmichael SW, Morrey BF: The cubital tunnel and ulnar neuropathy. J Bone Joint Surg Br 73:613–617, 1991. 3. Spinner RJ, Goldner RD: Snapping of the medial head of the triceps and recurrent dislocation of the ulnar nerve. Anatomical and dynamic factors. J Bone Joint Surg Am 80:239–247, 1998.

FIGURE 1. A, at the time of a planned simple decompression of the ulnar nerve through a 5-cm incision, I identified the nerve proximal to the medial epicondyle. The nerve was completely covered by a prominent medial head of triceps (arrowhead on medial edge). The nerve was mobilized and protected in vessel loops. P = proximal; purple ink on medial epicondyle. B, an anconeus epitrochlearis muscle was seen (arrow). C, this variant muscle was resected (asterisk). The ulnar nerve was decompressed slightly further distally through a portion of the flexor carpi ulnaris. The nerve did not subluxate with passive elbow flexion intraoperatively. Neurologic recovery was noted in the recovery room.


PERIPHERAL NERVE Technique Assessment

CLINICAL AND ELECTROPHYSIOLOGICAL COMPARISON OF DIFFERENT METHODS OF SOFT TISSUE COVERAGE OF THE MEDIAN NERVE IN RECURRENT CARPAL TUNNEL SYNDROME Nicolas M. Stütz, M.D. Handcenter, Bad Neustadt, and Department of Plastic and Reconstructive Surgery, Klinikum Nuremberg, Nuremberg, Germany

Andreas Gohritz, M.D. Handcenter, Bad Neustadt, and Department of Plastic and Reconstructive Surgery, Medizinische Hochschule, Hannover, Germany

Alexander Novotny, M.D. Department of Surgery, Klinikum rechts der Isar, Munich, Germany

Udo Falkenberg, M.D. Department of Neurology, Bad Neustadt, Germany

Ulrich Lanz, M.D. Handcenter, Bad Neustadt, Germany

Jörg van Schoonhoven, M.D. Handcenter, Bad Neustadt, Germany Reprint requests: Nicolas M. Stütz, M.D., Department of Plastic and Reconstructive Surgery, Klinikum Nuremberg, Breslauerstraße 201, 90471 Nuremberg, Germany. Email: [email protected] Received, November 8, 2006. Accepted, August 9, 2007.

OBJECTIVE: To evaluate the clinical and electrophysiological results of 26 patients treated with either a hypothenar fat flap or a synovial flap to prevent recurrent scar compression of the median nerve after previously failed carpal tunnel decompression. METHODS: A total of 26 patients underwent flap coverage as a result of a nerve tethering attributable to a position within scar; 15 were covered by a synovial flap and 11 by a hypothenar fat flap. Only patients in whom the median nerve was significantly enveloped in scar tissue were included. All candidates underwent a thorough clinical examination and nerve conduction test. The pre- and postoperative nerve conduction tests and the results of the two groups were statistically compared. RESULTS: The reduction rates of brachial nocturnal pain and pillar pain were 25 and 25%, respectively, in the synovial flap group and 64 and 37%, respectively, in the hypothenar fat flap group. The reduction rates of a positive Tinel’s sign (25%) and a positive Phalen’s test (13%) were lower in the synovial flap group compared with hypothenar fat flap coverage (55% Tinel’s sign, 46% Phalen’s test). Thenar atrophy and paresthesia were reduced in 44 and 62%, respectively, in the synovial flap group and in 46 and 64%, respectively, in the hypothenar fat flap group. The overall patient satisfaction (73%) and the Disabilities of the Arm, Shoulder and Hand score (31 points) appeared superior in the hypothenar fat flap group compared with the synovial flap group (56%; 37 points). Nerve conduction tests demonstrated a significant improvement when comparing the pre- and postoperative measurements in both groups. Distal motor latency decreased in the hypothenar fat flap group from 6.81 ms to 4.92 msec (P ⫽ 0.01; mean value) and in the synovial flap group from 6.04 ms to 4.43 msec (P ⬍ 0.001; mean value). CONCLUSION: Coverage by an ulnar-based hypothenar fat flap appeared to produce superior clinical results compared with coverage with synovial tissue from adjacent flexor tendons, although conclusive statistical evaluation of clinical outcomes was not possible. Further studies to confirm this are warranted. KEY WORDS: Hypothenar fat flap, Median nerve, Recurrent carpal tunnel syndrome, Scar formation, Synovial flap Neurosurgery 62:ONS194–ONS200, 2008

S

urgical decompression of the transverse carpal ligament relieves carpal tunnel syndrome in the vast majority of patients. However, some continue to experience persistent paresthesias, dysesthetic pain, weakness, or focal irritation of the median nerve (5, 23, 24). Complications and treatment failures have been reported to occur in 3 to 19% of the


DOI: 10.1227/01.NEU.0000297051.89658.28

patients in large series, requiring reexploration in up to 12% (1, 3, 14). Causes of discomfort include painful neuroma at the surgical site, incomplete release of structures compressing the median nerve, scarring, or devascularization (1). Even after repeat surgery, some patients continue to experience median nerve irritation. There is evidence that a significant


DIFFERENT METHODS OF SOFT TISSUE COVERAGE OF THE MEDIAN NERVE

proportion of these patients have scarring of the nerve to the palmar and/or radial surface of the carpal canal producing traction and pain (14, 18). Many procedures have been advocated to solve this “problem of recurrence,” although the success of secondary carpal tunnel surgery is usually inferior to the results of primary surgery (7, 11, 13, 16, 18, 20, 30). These methods include external and internal neurolysis, local and distal muscle flaps, fat grafts and flaps, vein wrapping, and synovial flaps (3, 7, 11, 13, 16, 18, 20, 23, 24, 30). The objective of this clinical study was to compare the clinical and electrophysiological results after soft tissue coverage in cases of significant scarring of the median nerve in recurrent carpal tunnel syndrome by two commonly applied methods, the hypothenar fat flap (HFF) or synovial flap (SF).

PATIENTS AND METHODS Patient Selection In a comprehensive retrospective study, patient charts, especially operating reports, were analyzed to determine the intraoperative findings and the indications for revision surgery after carpal tunnel release in a total of 200 patients. In 108 patients, it was found that the flexor retinaculum had been released incompletely. In 12 patients, a nerve laceration had occurred during the primary intervention. Fibrosis was seen as the reason for recurrence in 17 patients, and a tumor was found in the carpal tunnel in four patients. In 13 patients, no specific reason was found for recurrence of symptoms. In 46 of these patients, symptoms were the result of a renewed constriction of the nerve resulting from its position in scar tissue. These results were previously reported by our study group (27). Twenty patients were treated by only external neurolysis. Twenty-six patients in whom the median nerve was found nearly circumscribed or wrapped entirely by scar tissue comprise the current study. These patients could be included in a clinical examination and nerve conduction test (NCT). The diagnosis and documentation of recurrent carpal tunnel syndrome was based on the patient’s history of paresthesia confined to the distribution of the median nerve and pain during the night and on the physical examination, including a positive Tinel’s sign at the carpal tunnel and/or a positive Phalen’s compression test (1, 3, 5, 27). In all patients, carpal tunnel syndrome had been documented clinically and neurologically, symptoms had persisted or recurred, or new symptoms developed after the initial carpal tunnel release. Differences were seen between those with persisting symptoms and those with recurrent symptoms. Symptoms that continued were seen in the group of lacerated nerves and in the group of incomplete retinacular division. Symptoms that recurred after initial relief of pain were more likely to be scar-related, because it takes 3 to 6 months for scar to form and tighten. Patients with a double crush syndrome (i.e., pronator syndrome, cervical disc disease, thoracic outlet syndrome) and patients with more than one revision surgery were excluded from the study.

Objective Assessments Objective assessments were performed by two-point discrimination sensation (each fingertip was measured three times, and the worst value was recorded), measurement of grip strength, and documentation of thenar muscle atrophy. The patients were also asked about pillar pain, severe tenderness, and induration of both “pillars” of the scar

NEUROSURGERY

(9). The validated Disabilities of the Arm, Shoulder, and Hand questionnaire was used to assess the subjective outcome (12). In addition, patients were asked to quantify their satisfaction with the results of their surgery and if they would undergo the same procedure again. In 15 patients, the median nerve was covered with an SF and in 11 patients with a HFF. All 26 patients were available for follow-up approximately 10 to 12 months after revision and simultaneous flap coverage, and all were evaluated by the same observer (UF). All patients had a full NCT preoperatively; thus, the values of a postoperative NCT of the two groups could be statistically compared.

Electrophysiological Methods The electrodiagnostic tests were performed by an experienced electromyographer under standardized conditions in a shielded room. The skin temperature at the wrist was confirmed to be above 32⬚C. Motor nerve conduction velocity, sensory nerve conduction velocity (SNCV), distal motor latencies (DML), and compound muscle action potentials were obtained after distal and proximal stimulation (at the wrist and elbow). The nerves were stimulated supramaximally to localize the site of maximal compound muscle action potentials above the muscle belly with surface disc electrodes. SNCV and compound sensory nerve action potential amplitudes were obtained antidromically using electrodes placed around the index and little fingers. Electromyography was performed using concentric needle electrodes in the abductor pollicis brevis and abductor digiti minimi muscles. The reviewer was blinded for the clinical and electrophysiological testing.

Statistical Evaluation The neurophysiological data were analyzed by an independent statistician (AN). Normal distribution of the preoperative values were checked with the Kolmogorov-Smirnov goodness-of-fit test, and the pre- and postoperative SNCV, motor nerve conduction velocity, and DML values in the two groups were compared with the t test for paired data (the null hypothesis was that the values were comparable). The level of significance was set at a P value of less than 0.05. Analyses were performed using the SPSS 11.5 statistical package (SPSS Corp., Chicago, IL).

Operative Technique Both flaps were performed using a standardized technique, although individual surgical variations cannot be totally excluded in a clinical setup. The surgeons who carried out the revision surgery were skilled hand surgeons with profound experience in both operative flap techniques. All recurrences were treated using an open approach. The carpal tunnel was explored in all cases, and the median nerve had been found severely tethered or wrapped entirely in scar tissue (Fig. 1). An extensive dissection and mobilization was performed, and the adhesions were released (Fig. 2) (27).

Hypothenar Fat Flap The HFF was raised by sharp subcutaneous dissection on the ulnar border of the carpal tunnel incision. Further dissection was carried out ulnarly just beneath the subdermal plexus overlying the hypothenar fat pad to the dermal insertion of the palmaris brevis and until the ulnar nerve and artery were seen. A portion of fat from the hypothenar eminence superficial to the muscle layer was elevated as a pedicled flap and transposed to cover the median nerve and to provide a gliding surface for the nerve (13). Mobilization of the fat pad continued until easy


STÜTZ ET AL.

FIGURE 1. Severe tethering of the median nerve resulted from scar adhesion. The nerve is enveloped in scar tissue.

FIGURE 4. Synovial flap was sutured on the radiopalmar aspect of the carpal tunnel. Asterisk, median nerve; number sign, synovial flap. began with an assessment of the previous surgical incision because wide exposure of the median nerve is necessary to safely and adequately free the scarred nerve. A wide ulnarly based flap of synovial tissue was raised and sutured to the deep side of the radial leaf of the transverse ligament (Fig. 4) (30).

RESULTS

FIGURE 2. Result after external neurolysis of the median nerve.

Twenty-six patients had scarring of the median nerve after open (n ⫽ 21) or endoscopic (n ⫽ 5) decompression of the carpal tunnel and underwent nerve coverage. In 15 patients, nerve coverage was performed by a SF (three were primarily released endoscopically) and in 11 patients by a HFF (in three patients an endoscopic approach had been used during the first operation). The average time interval between the first operation and the revision surgery was 970 days. The mean interval between the first and second operations was 20.6 months in the HFF group and 21.6 months in the SF group (Table 1). Our series included 17 women and nine men, four of whom were in the HFF group and five in the SF group. The mean age was 55 years (range, 29–77 yr) at the time of the first operation.

Electrophysiological Results

FIGURE 3. Elevation of hypothenar fat flap.

transposition over the median nerve was achieved, allowing it to turn the flap like a book page (Fig. 3). Caution was taken not to excessively thin the overlying skin because the fat pad receives its blood supply from segmental arteries arising from the ulnar border of the ulnar artery in Guyon’s canal. The flap was sutured to the deep side of the radial leaf of the transverse ligament.

Synovial Flap The other technique used was a wide ulnarly based pedicle of synovial tissue turned radially to cover the median nerve. The procedure


NCTs demonstrated a significant improvement when comparing the pre- and postoperative measurements in both groups. There was an improvement of the DML in the HFF group from 6.81 to 4.92 msec (P ⫽ 0.01; mean value [MV]) and in the SF group from 6.04 to 4.43 msec (P ⬍ 0.001; MV). Preoperative SNCV in the HFF group were 39.18 m/second and increased to 47.82 m/second (MV) postoperatively and from 38.79 m/second (MV) to 45.21 m/second (MV) in the SF group (Table 1).

Clinical Results Eleven patients in the SF group reported pillar pain (severe tenderness and induration of both “pillars” of the scar) before revision surgery and seven patients reported pillar pain after revision and simultaneous flap coverage. In the HFF group, four out of eight patients experienced a significant improvement after the revision.



TABLE 1. Values of distal motor latency, motor nerve conduction velocity, and sensory nerve conduction velocity in the synovial flap group pre- and postoperativelya Age (yr)

Interval between operations (mo)

DML(msec) prerevision

MNCV(m/sec) prerevision

SNCV(m/sec) prerevision

DML(msec) MNCV (m/sec) postrevision postrevision

SNCV(m/sec) postrevision

Mean

54.5

20.64

6.81

45.67

39.18

4.92

52.40

47.82

Median

54.9

14.00

7.20

45.10

37.80

4.70

51.70

45.50

SD

14.0

16.00

1.91

6.51

5.36

0.71

5.34

7.49

Mean

57.4

21.69

6.04

44.83

38.79

4.43

51.02

45.21

Median

58.0

12.50

5.65

43.25

40.25

4.25

51.60

45.90

6.1

21.63

1.54

4.99

5.60

1.08

4.83

4.84

HFF

SF

SD a

DML, distal motor latency; MNCV, motor nerve conduction velocity; SNCV, sensory nerve conduction velocity; HFF, hypothenar fat flap; SD, standard deviation; SF, synovial flap. DML pre- versus postoperatively, P ⬍ 0.001; MNCV pre- versus postoperatively, P ⬍ 0.001; SNCV pre- versus postoperatively, P ⬍ 0.001; Student’s t test. Values of DML, MNCV, and SNCV in the hypothenar fat flap group pre- and postoperatively: DML pre- versus postoperatively, P ⫽ 0.01; MNCV pre- versus postoperatively, P ⬍ 0.001; SNCV pre- versus postoperatively, P ⫽ 0.021; t test.

TABLE 2. Clinical values of synovial flap and hypothenar fat flapa SF group (n ⴝ 16) Preoperatively

Postoperatively/ percent reduction

Preoperatively

Postoperatively/ percent reduction

Brachial nocturnal pain

11 (69%)

7 (44%)/25%

9 (82%)

2 (18%)/64%

Pillar pain

11 (69%)

7 (44%)/25%

8 (73%)

4 (36%)/37%

Tinel’s sign

12 (75%)

8 (50%)/25%

9 (82%)

3 (27%)/55%

Phalen’s test

a

HFF group (n ⴝ 11)

7 (44%)

5 (31%)/13%

8 (73%)

3 (27%)/46%

Thenar atrophy

14 (88%)

7 (44%)/44%

8 (73%)

3 (27%)/46%

Paraesthesia

16 (100%)

6 (38%)/62%

11 (100%)

4 (36%)/64%

Patient’s satisfaction

9 (56%)

8 (73%)

2 PD (MV)

4.8 mm

4.1 mm

DASH (MV)

37

31

SF, synovial flap; HFF, hypothenar fat flap; 2 PD, two-point discrimination; MV, mean value; DASH, Disabilities of the Arm, Shoulder and Hand.

The preoperative values of the two-point discrimination and the Disabilities of the Arm, Shoulder, and Hand scores were not available because those were not routinely taken before carpal tunnel release. The postoperative two-point discrimination averaged 4.8 mm in the SF group and 4.1 mm in the HFF group. The mean postoperative Disabilities of the Arm, Shoulder, and Hand score in the SF group was 37 (MV; range, 12–65) and 31 in the HFF group (MV; range, 4–89). The symptoms concerning pain during the night (brachialgia nocturna) were reduced in seven out of 11 patients in the SF group and in seven out of nine patients in the HFF group. The positive Phalen’s test remained in five out of seven patients in the SF group and in three out of eight patients in the HFF group. Tinel’s sign persisted in eight patients after revision of 12 patients in the SF group and in three patients after revision

NEUROSURGERY

of nine patients in the HFF group. Preoperatively, all patients reported paresthesia, which persisted after revision in six patients covered with an SF and in four with a HFF. The clinical data are shown in Table 2. Before the revision surgery was performed in seven patients in the SF and HFF groups, severe atrophy of the thenar muscles was seen. Severe atrophy improved in three (SF group) and four patients (HFF group), respectively. Grip strength increased from 21 kg (MV; range, 37, 12–49 kg) to 25 kg (MV; range, 38, 14–52 kg ) in the SF group and from 19 kg (MV; range, 43, 11–54 kg) to 28 kg (MV; range, 48, 15–63 kg) in the HFF group, which was an increase of 72 (SF) versus 75% (HFF) of the strength of the contralateral side. Overall, 56% of the patients with SF coverage and 73% with HFF coverage had a satisfactory subjective outcome and would undergo the same secondary procedure again. No


STÜTZ ET AL.

donor site morbidity was found in either group. In two patients (HFF group), a hypertrophic scar occurred, and in one patient (SF group), delayed wound healing occurred.

DISCUSSION The objective of this study was to compare two different methods of soft tissue coverage of the median nerve performed during the revision surgery after failed carpal tunnel release, the HFF and the SF. Persistent or recurrent symptoms after carpal tunnel surgery are infrequent but challenging clinical problems. A thorough evaluation of these patients is mandatory and must confirm the accuracy of the original diagnosis and rule out the presence of concurrent conditions or disorders that may cause persistent symptoms mimicking carpal tunnel syndrome. Surgical revision should be considered if alternative explanations of the patient’s symptoms cannot be identified and if conservative care is ineffective (4, 21, 29). A frequent cause of failed carpal tunnel surgery, besides incomplete division of the flexor retinaculum, is the position of the median nerve within scar tissue. In this study, we focused on patients requiring revision surgery as a result of complete scarring around the median nerve; the objective was to compare two different methods of soft tissue coverage of the median nerve performed during the revision surgery. A great variety of methods have been recommended to ameliorate the problem of recurrent carpal tunnel syndrome. These mainly aim at providing additional vascular supply to a region of chronic ischemia, allowing for a gliding interface for the median nerve and adding a physical barrier between the nerve and adjacent anatomic structures to prevent renewed constriction (4, 21, 29). The ideal treatment of persistent or recurrent carpal tunnel syndrome has not yet been determined. The reconstruction of the transverse carpal ligament to restore median nerve gliding (9), the pronator quadratus (8), palmaris brevis (25) and the abductor digiti minimi (7) muscles as well as synovium (30) and fat flaps (13) have all been used (4). The use of flaps provides well-vascularized tissue, which covers the median nerve, and the flaps should be thick enough to cushion this vulnerable part. Therefore, neurolysis and nerve wrapping alone, e.g., by a vein, is not performed routinely at our institution. The authors’ preference is the HFF as described by Giunta et al. (13). The HFF provides locally available tissue to prevent further adhesions after the scarred nerve has been released from surrounding tissue. The reason is that the complication rate for performing this flap is very low, and there is enough surrounding tissue to cover and cushion this specific part. Some authors have reported that patients who undergo repeat surgery for recurrent symptoms after incomplete release of the ligament have a better prognosis than those in whom the ligament was completely released at the initial surgery (5). However, others have failed to demonstrate a significant association between this intraoperative finding and the final outcome after repeat surgery (17).


Electrophysiological examinations after successful carpal tunnel decompression surgery can remain abnormal for up to 24 months postoperatively. In both of our groups, the DML had decreased drastically after renewed scar formation around the nerve and was nearly normal 10 to 12 months after revision surgery. A study by Lang et al. (17) reported that 82% of DML improvement occurs within the first month and that the greater the preoperative deficit, the greater the improvement. SNCV and DML have high specificity; DML has a sensitivity of 60 to 80% and SNCV a sensitivity of 40 to 96% (9, 26). However, in accordance with the results reported by Tackmann et al. (28), we agree that electrophysiological testing alone cannot give sufficient information for postoperative decision-making and cannot predict clinical outcome. In some publications, symptom severity scores and functional scores correlate significantly with each other and with general outcome scores, but not with the electrophysiological data demonstrated. Some reports showed a correlation between the SNCV and a symptom severity scale score (28), but other authors denied this (10, 14, 22). However, in specific clinical situations such as the need for and timing of revision surgery, electrophysiological testing adds valuable information and helps if another surgical intervention is required. We conclude that soft tissue coverage seems to be beneficial to prevent recurrent symptoms if the nerve lies directly in the scar. Although conclusive statistical evaluation of clinical outcomes was not possible, coverage by an ulnar-based HFF appeared to produce superior clinical results compared with coverage with synovial tissue from adjacent flexor tendons. Further studies to confirm this are warranted.

REFERENCES 1. Bande S, De Smet L, Fabry G: The results of carpal tunnel release: Open versus endoscopic technique. J Hand Surg [Br] 19:14–17, 1994. 2. Deleted in proof. 3. Botte MJ, von Schroeder HP, Abrams RA, Gellman H: Recurrent carpal tunnel syndrome. Hand Clin 12:731–743, 1996. 4. Cobb TK, Amadio PC, Leatherwood DF, Schleck CD, Ilstrup DM: Outcome of reoperation for carpal tunnel syndrome. J Hand Surg [Am] 21:347–356, 1996. 5. Concannon MJ, Brownfield ML, Puckett CL: The incidence of recurrence after endoscopic carpal tunnel release. Plast Reconstr Surg 105:1662–1665, 2000. 6. Deleted in proof. 7. Cramer LM: Local fat coverage for the median nerve, in Langford LL (ed): Correspondence Newsletter for Hand Surgery. 1985, p 35. 8. Dellon AL, Mackinnon SE: The pronator quadratus muscle flap. J Hand Surg [Am] 9:423–427, 1984. 9. De Smet L, Vandeputte G: Pedicled fat flap coverage of the median nerve after failed carpal tunnel decompression. J Hand Surg [Br] 27:350–353, 2002. 10. Dhong ES, Han SK, Lee BI, Kim WK: Correlation of electrodiagnostic findings with subjective symptoms in carpal tunnel syndrome. Ann Plast Surg 45:127–131, 2000. 11. Duclos L, Sokolow C: Management of true recurrent carpal tunnel syndrome: Is it worthwhile to bring vascularized tissue? Chir Main 17:113–118, 1998. 12. Germann G, Wind G, Harth A: The DASH (Disability of Arm–Shoulder– Hand) Questionnaire—A new instrument for evaluating upper extremity treatment outcome [in German]. Handchir Mikrochir Plast Chir 31:149–152, 1999. 13. Giunta R, Frank U, Lanz U: The hypothenar fat-pad flap for reconstructive repair after scarring of the median nerve at the wrist joint. Chir Main 17:107–112, 1998.



14. Heybeli N, Kutluhan S, Demirci S, Kerman M, Mumcu EF: Assessment of outcome of carpal tunnels syndrome: A comparison of electrophysiological findings and a self-administered Boston questionnaire. J Hand Surg [Br] 27:259–264, 2002. 15. Deleted in proof. 16. Kuschner SH, Brien WW, Johnson D, Gellman H: Complications associated with carpal tunnel release. Orthop Rev 20:346–352, 1991. 17. Lang E, Spitzer A, Pfannmüller D, Claus D, Handwerker HO, Neundörfer B: Function of thick and thin nerve fibers in carpal tunnel syndrome before and after surgical treatment. Muscle Nerve 18:207–215, 1995. 18. Langloh ND, Linscheid RL: Recurrent and unrelieved carpal-tunnel syndrome. Clin Orthop Relat Res 83:41–47, 1972. 19. Deleted in proof. 20. Nancollas MP, Peimer CA, Wheeler DR, Sherwin FS: Long-term results of carpal tunnel release. J Hand Surg [Br] 20:470–474, 1995. 21. O’Malley MJ, Evanoff M, Terrono AL, Millender LH: Factors that determine reexploration treatment of carpal tunnel syndrome. J Hand Surg [Am] 17:638–641, 1992. 22. Padua L, Padua R, Aprile I, Pasqualetti P, Tonali P: Multiperspective followup of untreated carpal tunnel syndrome: A multicenter study. Neurology 56:1459–1466, 2001. 23. Palmer AK, Toivonen DA: Complications of endoscopic and open carpal tunnel release. J Hand Surg [Am] 24:561–565, 1999. 24. Phalen GS: The birth of a syndrome, or carpal tunnel revisited. J Hand Surg [Am] 6:109–110, 1981. 25. Rose EH, Norris MS, Kowalski TA, Lucas A, Flegler EJ: Palmaris brevis turnover flap as an adjunct to internal neurolysis of the chronically scarred median nerve in recurrent carpal tunnel syndrome. J Hand Surg [Am] 16:191–201, 1991. 26. Shurr DG, Blair WF, Bassett G: Electromyographic changes after carpal tunnel release. J Hand Surg [Am] 11:876–880, 1986. 27. Stütz N, Gohritz A, van Schoonhoven J, Lanz U: Revision surgery after carpal tunnel release—Analysis of the pathology in 200 cases during a 2 year period. J Hand Surg [Br] 31:68–71, 2006. 28. Tackmann W, Brennwald J, Nigst H: Sensory electroneurographic parameters and clinical recovery of sensibility in sutured human nerves. J Neurol 229:195–206, 1983. 29. Van de Kar HJ, Jaquet JB, Meulstee J, Molenar CB, Schimsheimer RJ, Hovius SE: Clinical value of electrodiagnostic testing following repair of peripheral nerve lesions: A prospective study. J Hand Surg [Br] 27:345–349, 2002. 30. Wulle C: Treatment of recurrence of the carpal tunnel syndrome. Ann Chir Main 6:203–209, 1987.

Acknowledgment We thank Daniel Fuge, M.D., for his assistance in preparing the photographs.

COMMENTS

T

he optimal treatment of recurrent carpal tunnel syndrome has not yet been determined. Although some surgeons favor simple external neurolysis or internal neurolysis, others favor covering the median nerve with vascularized tissue during reexploration. In this article, the authors compared two useful techniques for treating recurrent carpal tunnel syndrome that was found to be caused by scar formation around the median nerve. They reported results from 15 patients with synovial flap (SF) coverage and 11 patients with hypothenar fat flap (HFF) coverage during revision carpal tunnel surgery. Objective criteria such as nerve conduction studies as well as clinical outcomes were compared. The authors reported a significant improvement when preand postoperative measurements in both groups were compared. They report superior results with the HFF technique. This is a retrospective chart review with a relatively small sample size. The procedures were performed by two different groups of surgeons within the same department. Thus, any extrapolation of the results from this study must be done with caution. For neurosurgeons who do not perform revision

NEUROSURGERY

carpal tunnel surgery with any frequency, this article provides some valuable guidelines for management. Eric L. Zager Philadelphia, Pennsylvania

T

he authors studied two types of soft tissue coverage for median nerves for which prior carpal tunnel release failed. The first group had coverage by a SF and the second group by a HFF. Pre- and postoperative clinical evaluations and nerve conduction tests were compared. It was concluded that a HFF was superior to a SF using these criteria. The authors make the assumption that some type of flap coverage in a revision carpal tunnel surgery is superior to none but, unfortunately, did not do a series of revision surgery without some type of flap. Thus, with or without coverage, a thorough and meticulous external neurolysis of the nerve or a revision carpal tunnel surgery is needed. Oftentimes, a synovectomy as well as resection of another scar adjacent to the nerve and in the region of the carpal tunnel is also needed in my experience. These steps are far more important than creating a flap, whether it be synovium or fat, to cover the nerve. David G. Kline New Orleans, Louisiana

A

lthough open or endoscopic carpal tunnel release is a successful operative procedure, there is still an approximately 10% failure rate. Because it is a commonly performed procedure, a relatively large number of patients who have persistent, recurrent, or new symptoms postoperatively exist. Relatively little is written about these patients. This European group has extensive experience with primary and revision carpal tunnel surgery, which they have carefully documented over the years. This particular retrospective study analyzes the outcomes of 26 patients with recurrent carpal tunnel syndrome attributable to neural fibrosis treated by either a HFF or a SF. Superior results were seen in those treated by the HFF. It is technically relatively easy to perform and has been proved to be reliable by other groups as well. Hand surgeons at my own institution have reported excellent results with this technique combined with microneurolysis (1). The rationale for free or pedicled flaps is logical: to provide a vascularized barrier, potentially preventing readherence and promoting neural gliding. Still other surgeons have advocated a variety of other treatments—including extensive neurolysis without flap coverage (2). The vast majority of neurosurgeons do not have personal experience, training, or expertise in these types of flaps and do not perform them in patients with recurrent carpal tunnel syndrome. A well-designed study to compare flap coverage or neurolysis in this difficult patient population is necessary to demonstrate the benefits of such a procedure and to change our current practice. A previous controlled, blinded study comparing standard closure and hypothenar fat pad flaps did not demonstrate any differences in patients with primary carpal tunnel syndrome (3). The results from a prospective study comparing these different operative strategies in patients with recurrent carpal tunnel syndrome might convince us either to learn these advanced techniques or to refer these patients. Robert J. Spinner Rochester, Minnesota

1. Craft RO, Duncan SFM, Smith AA: Management of recurrent carpal tunnel syndrome with microneurolysis and the hypothenar fat pad flap. Hand 2:85–89, 2007.


STÜTZ ET AL.

2. Duclos L, Sokolow C: Management of true recurrent carpal tunnel syndrome: Is it worthwhile to bring vascularized tissue? Chir Main 17:113–118, 1998. 3. Jones SM, Stuart PR, Stothard J: Open carpal tunnel. Does a vascularized hypothenar fat pad reduce wound tenderness? J Hand Surg [Br] 22:758-760, 1997.

T

hese authors have provided preliminary data supporting the use of a rotational HFF to cover and protect the median nerve for patients with recurrent carpal tunnel syndrome attributable to significant scarring of the median nerve observed at reexploration. In a small number of patients, they documented improved clinical outcomes using this technique versus coverage of the median nerve with local synovium. Although the results are not definitive for methodological reasons,

these authors have extensive experience treating primary and recurrent carpal tunnel syndrome, and therefore their recommendation should be an impetus for further investigation into these techniques, including the potential role of synthetic nerve wraps in the treatment of recurrent carpal tunnel syndrome. I currently favor the HFF as presented by the authors because it has low procedural morbidity, is vascularized tissue, and not only separates the nerve from the surgical field but also and more importantly, in my opinion, provides padding to prevent adherence of the median nerve to scar tissue that may be contiguous with the surface of the palm. Stephen M. Russell New York, New York

Volcher Coiter, (1575), Nicolas Neufchâtel. Nürnberg, Germanisches National Museum. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



FUNCTIONAL AND STEREOTACTIC Technique Assessment

FIDUCIAL VERSUS NONFIDUCIAL NEURONAVIGATION REGISTRATION ASSESSMENT AND CONSIDERATIONS OF ACCURACY Wolfgang K. Pfisterer, M.D. Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Stephen Papadopoulos, M.D. Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Denise A. Drumm, Ph.D. Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Kris Smith, M.D. Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona

Mark C. Preul, M.D. Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona Reprint requests: Mark C. Preul, M.D., Neurosurgery Research Laboratory, Division of Neurological Surgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, 350 West Thomas Road, Phoenix, AZ 85013. Email: [email protected] Received, March 7, 2007. Accepted, May 10, 2007.

NEUROSURGERY

OBJECTIVE: For frameless stereotaxy, users can choose between anatomic landmarks (ALs) or surface fiducial markers (FMs) for their match points during registration to define an alignment of the head in the physical and radiographic image space. In this study, we sought to determine the concordance among a point-merged FM registration, a point-merged AL registration, and a combined point-merged anatomic/surface-merged (SM) registration, i.e., to determine the accuracy of registration techniques with and without FMs by examining the extent of agreement between the system-generated predicted value and physical measured values. METHODS: We examined 30 volunteers treated with gamma knife surgery. The frameless stereotactic image-guidance system called the StealthStation (Medtronic Surgical Navigation Technologies, Louisville, CO) was used. Nine FMs were placed on the patient’s head and four were placed on a Leksell frame rod-box, which acted as a rigid set to determine the difference in error. For each registration form, we recorded the generated measurement (GM) and the physical measurement (PM) to each of the four checkpoint FMs. Bland and Altman plot difference analyses were used to compare measurement techniques. Correlations and descriptive analyses were completed. RESULTS: The mean of values for GMs were 1.14 mm for FM, 2.3 mm for AL, and 0.96 mm for SM registrations. The mean errors of the checkpoints were 3.49 mm for FM, 3.96 mm for AL, and 3.33 mm for SM registrations. The correlation between GMs and PMs indicated a linear relationship for all three methods. AL registration demonstrated the greatest mean difference, followed by FM registration; SM registration had the smallest difference between GMs and PMs. Differences in the anatomic registration methods, including SM registration, compared with FM registration were within a mean ⫾ 1.96 (standard deviation) according to the Bland and Altman analysis. CONCLUSION: For our sample of 30 patients, all three registration methods provided comparable distances to the target tissue for surgical procedures. Users may safely choose anatomic registration as a less costly and more time-efficient registration method for frameless stereotaxy. KEY WORDS: Anatomic landmarks, Fiducial markers, Frameless stereotaxy, Neuronavigation, Stereotaxic registration assessment Neurosurgery 62[ONS Suppl 1]:ONS201–ONS208, 2008

S

urgical navigation systems and, in particular, recent developments in frameless stereotactic systems, have proven to be valuable tools to aid neurosurgeons in planning, exposing, and reaching target structures and in planning the extent of resection of lesions located in the brain. Since the introduction of the first intraoperative frameless stereotactic navigation device in 1986 by Roberts

DOI: 10.1227/01.NEU.0000297007.04975.C6

et al. (19), such systems have become the operative standard for precise localization of anatomic and pathoanatomic targets. A key factor in the advancement of imageguided surgery is the ability to register images to patient imaging data sets (17). Pre- or intraoperative images are loaded into the system, the user defines two volumes (one volume defined by the patient’s radiographic image


PFISTERER ET AL.

space and the other defined by the patient’s physical anatomy space), and the computer then rotates and translates these until an alignment of the head in the physical space accurately matches the coordinates of the radiographic image space. It is essential that corresponding points be precisely identified on the images and on the patient. With most systems, users can choose between either anatomic landmarks (ALs) or surface-merged (SM) fiducial markers (FMs) for their point matching during registration (2, 8, 20, 26). FMs are relatively easy to identify on the image data set and on the patient’s scalp. However, a designated staff member must apply these markers to the patient before the scan, typically on the morning of surgery. Furthermore, skin SM FMs are mobile on the scalp and may lead to an increase in fiducial localization error, which is the distance between the measured position and the true position of the FM. If the patient were to undergo scanning before the day of surgery, that patient would need to have the FMs applied or tagged again in case they were inadvertently removed. The obvious advantage to a method based on AL placement is that the patient does not need any special attention before the scan. Thus, the patient could undergo scanning days or, in some cases, weeks before the surgery without the markers having to be retagged or reapplied. However, with an anatomic method, identification of the exact anatomic reference point may vary from registration to registration (e.g., locating the center of the tragus or the external canthus), as opposed to the logical consistency in identifying the center of an FM. Thus, there remain concerns regarding the accuracy of registration methods for image-guided surgery systems on the basis of ALs. There is accumulating evidence that anatomic systems offer accuracy that is comparable to that of methods based on FM (7, 12, 26). Our hypothesis was that registration based on ALs would be as accurate and, ultimately, as useful overall within a clinical realm as registration based on FMs. On the basis of this assumption, we sought to determine the concordance between predicted and measured anatomic errors as well as the agreement of three point-merge registration methods. We analyzed registration methods using FMs and ALs and considered whether a scalp SM registration could improve a registration based on ALs alone (i.e., in our study, combined anatomic/SM registration). In clinical measurement, comparison of a new technique with an established one is often required to determine whether the procedures agree sufficiently for the new to replace the old. Such investigations are often analyzed inappropriately, notably by using correlation coefficients. The use of such a correlation can be misleading. We used Bland and Altman plots for analysis, which we believe provide a more realistic assessment of repeatability and the relativity of correlation.

PATIENTS AND METHODS This study was performed with 30 patient volunteers who were in the process of undergoing gamma knife treatment for various pathological abnormalities at the Barrow Neurological Institute/St. Joseph’s


FIGURE 1. Setup of a volunteer when registering the fiducial markers (FMs) on the skin. Left, mounted reference frame. Arrow, checkpoint FM on the Leksell frame rod-box.

Hospital and Medical Center in Phoenix, AZ. This clinical protocol was approved by the Institutional Review Board of the Barrow Neurological Institute/St. Joseph’s Hospital and Medical Center and did not add to or subtract from the normal course of gamma knife treatment preparations beyond fiducial and anatomic registration. All measurements were performed during the period after magnetic resonance imaging (MRI) and the patient’s time interval for gamma knife treatment. A Leksell frame (Elekta Instruments, Stockholm, Sweden) was used as a control reference with which to compare the registration methods in the study. An optically based, frameless stereotactic image-guidance system (StealthStation; Medtronic Surgical Navigation Technologies, Louisville, CO) was used. FMs were placed in the usual manner on the patient’s head and on the rod-box of the Leksell frame that was fixed to the patient before MRI. Four FMs, “checkpoint FMs,” placed the frame rod-box acted as a rigid set of points to determine the differences in error between patients registration techniques (Fig. 1). These FMs were placed on the right and left anteriosuperior and posterioinferior corners of the frame rod-box. FMs were self-adhesive magnesium chloride-soaked, donut-shaped markers that are obvious on MRI scans. In the FM method, a second set of nine FMs was placed symmetrically on the patient’s head in accordance with the recommendations of Barnett as follows: (1) a pair just above the lateral aspect of the eyebrows, a pair on the lateral upper forehead, a pair on the asterions, a pair between the area just above the lateral aspect of the eyebrows, and a pair on the vertex or low forehead (with the level depending on the patient’s hairline). For each patient, all FMs were used for registration. For the AL method, registration points were both tragi of the ears, both lateral and medial canthi of the eyes, the nasion, and the lower rim of the nasal septum. All ALs were defined on the triplanar and three-dimensional rendering images on the StealthStation (Fig. 2). The combined anatomic/SM registration was performed with the addition of 40 skin points distributed evenly on the scalp from anterior to posterior and including the nose and cheeks. The patients then underwent MRI examination using the protocol assigned for targeting with the gamma knife system and Stealth imageguidance system. The patient’s image data set was transferred to the StealthStation for preregistration. A StealthStation dynamic reference frame was attached to the Leksell frame (Fig. 1). For registration, a free handheld stereotactic pointing device was used to point to the FM or AL. Each of the three different kinds of point-merged registrations was performed while each patient was in a comfortable position awaiting gamma knife treatment. Registrations were (in order of performance)


NEURONAVIGATION REGISTRATION ASSESSMENT

TABLE 1. Means for generated measurement and physical measurement for each registration methodb Registration method

Mean (SD)

Range

Fiducial

1.14 (0.301)

0.6–1.9

Anatomic

2.30 (0.744)

1.2–4.4

Surface-merged

0.96 (0.276)

0.5–1.7

Fiducial

3.49 (1.09)

1.4–5.8

Anatomic

3.96 (1.68)

1.4–8.4

Surface-merged

3.33 (1.65)

1.2–7.4

Generated measurement

Physical measurement

a b

SD, standard deviation. n = 30 patient volunteers.

RESULTS

FIGURE 2. Medial canthus of the left eye in coronal (top left), sagittal (top right), and axial planes (bottom left). Circles, anatomic landmarks (ALs) for registration. FM, AL, and SM. Generated measurement (GM) was recorded after each registration and was given by the StealthStation as the root mean square error (RMSE) in millimeters. Then, for receiving the physical measurement (PM), we recorded the fiducial registration error in millimeters for the control system (the FMs on the Leksell frame rod-box), according to the method described by West et al. (24). The GM is described as the distance between the measured position of the FM in one space (frameless stereotactic image-guidance system) and its counterpart in the other space (each of the four checkpoint FMs on the Leksell frame rod-box).

Statistical Analyses The main objective of this study was to demonstrate that PM was substantially equivalent to GM. Bland and Altman (3) plots for difference analysis were generated to compare measurement techniques. In this graphical method, the differences (or, alternatively, the ratios) between the two techniques are plotted against the averages of the two techniques. This method is useful for revealing a relationship between the differences and averages, looking for any systematic bias, and identifying possible outliers. If the differences within the mean ⫾ 1.96 (standard deviation [SD]) are not clinically important, the two methods may be used interchangeably. The Bland and Altman plot may also be used to assess the repeatability of a method by comparing repeated measurements using one single method on a series of subjects. The graph can then also be used to determine whether the variability or precision of a method is related to the size of the characteristic being measured. Correlations and descriptive analyses were also completed (5). Statistical analyses were performed with SPSS, Version 12.0 (SPSS Inc., Chicago, IL), and graphical analyses were performed with MDAS, Version 2.0 (EsKay Software, Silver Spring, MD).

NEUROSURGERY

GM and PM recordings were obtained for 30 patients. Overall, the three GM values were smaller than the three PM values, with a mean difference between values of 2.35 mm (SD, 1.01 mm). The mean GM values for this sample were as follows: the FM method mean was 1.14 mm (SD, 0.301 mm; median, 1.10 mm; range, 0.60–1.90 mm), the AL method mean was 2.30 mm (SD, 0.744 mm; median, 2.05 mm; range, 1.20–4.40 mm), and the SM method mean was 0.96 mm (SD, 0.276 mm; median, 0.90 mm; range, 0.50–1.70 mm). The mean values for PM were higher: the FM mean was 3.49 mm (SD, 1.09 mm; median, 3.53 mm; range, 1.4–5.8 mm), the AL mean was 3.96 mm (SD, 1.68 mm; median, 3.78 mm; range, 1.4–8.4 mm), and the SM mean was 3.33 mm (SD, 1.65 mm; median, 2.96 mm; range, 1.2–7.4 mm) (Table 1). We examined the strength of the linear relationship between the two methods through correlation and relative (positive or negative) differences between methods. Significant correlations were found between GM and PM averages for all three registration methods. The correlation between GM and PM recordings was moderately strong for the SM method (r ⫽ 0.72), moderate for the AL method (r ⫽ 0.56), and fair or weak for the FM method (r ⫽ 0.40) when correlated with their respective calculated measures. To examine concordance between the GM and PM methods, a difference plot was used. Inherent imprecision for both methods was assumed; therefore, the difference between each pair of measures (GM and PM) was plotted against the mean of the pair. The methods were considered to be comparable if 95% of the data points were found to lie within ⫾2 SD of the mean differences. All three plots showed linear trends with spread at the higher end of the scale indicating a proportional bias; therefore, a ratio method was used (Figs. 3–5).

GM and PM Values for the FM Method The GM values for the FM method ranged from 0.60 to 1.90 mm (mean, 1.14 mm; SD, 0.301 mm) and were smaller than the PM values (range, 1.43–5.75 mm; mean, 3.49 mm; SD, 1.09 mm).


PFISTERER ET AL.

FIGURE 3. Bland and Altman plot of the differences of generated measurement (GM) and physical measurement (PM) ratios against the average of the two methods for the FM point-merge registration method. The horizontal lines are drawn at the mean difference (Mean Diff) and indicate the limits of acceptability defined as 2 standard deviations (SD) of the mean ratio. Values are given in millimeters. One (3.3%) of the points was outside the ⫾2 SD lines. The plot shows good agreement for the two methods.

FIGURE 5. Bland and Altman plot of the differences of GM and PM ratios against the average of the two methods for the combined point-merge anatomic/ surface-merged registration (SM) method. The horizontal lines are drawn at the mean difference and indicate the limits of acceptability defined as 2 SD of the mean ratio. Values are given as millimeters. Two (6.6%) of the points were outside the ⫾2 SD lines. The plot shows good agreement for the two methods.

TABLE 2. Tolerance: absolute differences between generated and physical measurements of the fiducial markers method Tolerance, mm ⬍1.0

FIGURE 4. Bland and Altman plot of the differences of GM and PM ratios against the average of the two methods for the point-merged AL registration method. The horizontal lines are drawn at the mean difference and indicate the limits of acceptability defined as 2 SD of the mean ratio. Values are given as millimeters. Neither of the points were outside the ⫾2 SD lines. The plot shows good agreement for the two methods.

The mean difference between GM and PM scores was 2.35 mm (SD, 1.01 mm), with a 4.23-mm range (0.52–4.75 mm). Absolute differences between GM and PM methods are shown in Table 2. Correlation between GM and PM methods was weak (r ⫽ 0.40; P ⬍ 0.05). A Bland and Altman plot indicated general agreement between the two methods (Fig. 3). Plot differences indicated that one (3.3%) of the points was outside the ⫾2 SD lines.

GM and PM Values for the AL Method The GM values ranged from 1.2 to 4.4 mm (mean, 2.30 mm; SD, 0.744 mm) and were smaller than the PM values (range,


No. of patients (%) 3 (10.0)

1.0⫹

8 (26.6)

2.0⫹

12 (40.0)

3.0⫹

6 (20.0)

4.0⫹

1 (3.4)

Total

30 (100)

TABLE 3. Tolerance: absolute differences between generated and physical measurements of the anatomical landmarks method Tolerance, mm ⬍1.0

No. of patients (%) 11 (36.7)

1.0⫹

8 (26.7)

2.0⫹

7 (23.3)

3.0⫹

2 (6.7)

4.0⫹

2 (6.6)

Total

30 (100)

1.40–8.40 mm; mean, 3.96 mm; SD, 1.66 mm). The mean difference between GM and PM scores was 1.66 mm (SD, 1.40 mm), with a 5.13-mm range (0.07–5.20 mm). Absolute differences between GM and PM are shown in Table 3. There was a moderate relationship between the GM and PM values (r ⫽ 0.6; P ⬍ 0.01) (Fig. 4). From the plot differences, no point fell outside the ⫾2 SD lines.



DISCUSSION TABLE 4. Tolerance: absolute differences between generated and physical measurements of the surface-merged method Tolerance, mm

No. of patients (%)

⬍1.0

5 (16.7)

1.0⫹

8 (26.7)

2.0⫹

9 (30.0)

3.0⫹

5 (16.7)

4.0⫹

3 (9.9)

Total

30 (100)

GM and PM Values for the SM Method The GM values ranged from 0.50 to 1.7 mm (mean, 0.957 mm; SD, 0.276 mm) and were smaller than the PM values (range, 1.18–7.43 mm; mean, 3.33 mm; SD, 1.65 mm). The mean difference between GM and PM scores was 2.37 mm (SD, 1.47 mm), with a 5.63-mm range (0.40–6.03 mm). Absolute differences between GM and PM are shown in Table 4. The relationship between the GM and PM values was moderately strong (r ⫽ 0.7; P ⬍ 0.001) (Fig. 5). From the plot differences, there were two points (6.6%) that fell outside the ⫾2 SD lines. Overall, the agreement between the GMs and PMs suggested that the measures were relatively similar. For the FM method, the mean ratio or bias was 3.19 mm (SD of difference, 1.22 mm; ⫾1.96 SD limits of agreement, 0.799–5.58 mm); for the AL method, the mean ratio was 0.671 mm (SD of difference, 1.76 mm; limits of agreement, 0.446–3.08 mm), and the mean ratio between GM and PM values for the SM method was 3.44 mm (SD of difference, 1.41 mm; limits of agreement, 1.20–5.67 mm). If differences within ⫾1.96 SD are not clinically important, the methods can be used interchangeably. The SM method demonstrated the greatest mean difference followed by the AL method, and the FM method had the smallest difference between the GMs and PMs.

Anterior/Posterior and Left/Right Checkpoint Comparisons We compared anterior and posterior checkpoint FMs on the Leksell frame rod-box for each method. Mean values for the anterior points were 2.6 mm (SD, 1.12 mm) for the FM method, 3.56 mm (SD, 1.38 mm) for the AL method, and 3.28 mm (SD, 1.45 mm) for the SM method. Posterior mean values were 4.37 mm (SD, 1.39 mm) for the FM method, 4.36 mm (SD, 2.25 mm) for the AL method, and 3.38 mm (SD, 2.02 mm) for the SM method. Comparisons were made between left and right checkpoint FMs as well. The mean values for the right points were 3.61 mm (SD, 1.55 mm) for the FM method, 4.03 mm (SD, 1.95 mm) for the AL method, and 3.39 mm (SD, 1.59 mm) for the SM method. Left point mean values were 3.37 mm (SD, 1.28 mm) for the FM method, 3.89 mm (SD, 1.85 mm) for the AL method, and 3.27 mm (SD, 1.99 mm) for the SM method.

NEUROSURGERY

The principal goal of this project was to determine the accuracy of registration techniques with and without FMs by examining the extent of agreement between the systemgenerated predicted values and measured values. Although comparison and evaluation of intermodality brain image registrations already exists, there is little information comparing different image-to-physical space registrations against registrations of rigid environments with checkpoints (i.e., checkpoint FMs) (23). There is some evidence to support registration methods that use anatomic markers having accuracy comparable to that of registration methods using FMs (7, 12, 26). Others contend that anatomic markers used for registration are more inaccurate than skin-attached markers (8, 20, 21). In addition, SM methods may further improve either anatomic or fiducial registration methods (12). Rapid identification of targets results in shorter surgical time and lower risk of damage to cerebral structures. If we assume that anatomic registration methods are as accurate as the FM registration method, the use of FMs may be redundant and unnecessary, thus we may avoid additional costs and enjoy greater flexibility in scheduling around the time of surgery. Using self-adhesive FMs is the most common method in clinical practice, and errors between 1.5 and 4 mm have been reported by various authors (8, 9, 15, 16, 20, 21, 25). This information corresponds with our own experience with a predicted mean error of 1.14 mm. However, such markers that use adhesive tape may cause skin dislocation, which induces error. In an effort to minimize error caused by dislocation of the skin markers, bone-attached screws have been used with a mean error of 0.23 mm under laboratory conditions (4) and have been tested for increasing the alignment accuracy of overlays between surgical navigation images and microscopic view (6). Maurer et al. (13) tested a multimodal image-to-physical registration using implantable markers. The mean computed tomographic-MRI registration error for their clinical data obtained without rigid head fixation during scanning was 1.4 mm; however, implantable markers are rarely used clinically because of the invasiveness of their application. Other factors that may influence accuracy are related to imaging, such as field of view, slice thickness, and different kinds of sequences (11, 18). There are only two clinical studies and one cadaveric study that describe sufficient accuracy by use of ALs for registration. Vrionis et al. (22) reported a target localization error within 2.44 mm for temporal structures in a cadaveric study. Wolfsberger et al. (26) described a mean RMSE of 3.2 mm using anatomic markers versus 2.9 mm using self-adhesive skin markers. Ganslandt et al. (7) obtained sufficient accuracy by using both fiducial registration and AL SM-fitting computer algorithms with a mean system accuracy of 1.81 mm. The inaccuracy in the use of anatomic registration points is attributed to the difficulty of identifying ALs precisely on an MRI scan and on the patient’s head. Villalobos and Germano (21) reported accuracy of 3.2 mm using anatomic landmarks and one frontal FM, which was inferior to that using only skin


PFISTERER ET AL.

FMs; however, a precise 3-day description of the location of the ALs was not provided. Wolfsberger et al. (26) registered five natural markers: the nasion, both posterior walls of the external acoustic meatus, and both lateral canthi. Anatomic structures such as the external acoustic meatus were used, which are less flexible. A recent study by Mascott et al. (12) reported use of five ALs (bilateral tragus, lateral caruncula [within the medial canthus], and bony bridge of the nose) with a target localization error mean of 5.4 mm (SD, 1.8 mm). We used both tragi of the ears, both the lateral and medial canthi of the eyes, the nasion, and the lower rim of the nasal septum, first, to take points that are easy to find, and second, to avoid coplanar distribution of points in axial and coronal planes, which was thought to be responsible for lower accuracy compared with artificial FMs (10, 14, 21, 24). Some investigators have identified higher intraoperative errors than the RMSE calculated by the computer (12, 21). In our study, the predicted average was lower than the measured average for all three registration methods. Dependence on the navigational system for accurate target localization is important for the neurosurgeon; therefore, we analyzed comparability between GM and PM. We found agreement between the GM and PM methods, which suggests that the surgeon may trust the computed RMSE to obtain safe information about accuracy in navigation. Despite this sphere of accuracy provided by the computer, measurements of checkpoints in different regions may be attributable to different deviations. This accuracy depends first, on the distance from the checkpoint to the center of the anatomic or artificial FMs, and second, on the subjective view of the surgeon during the operation to define the control points “exactly.” The logistical problems associated with the task of placing the FMs as close as possible before imaging can be avoided by using the anatomic registration method. Even an additional so-called “preoperative MRI fiducial study” is unnecessary when the diagnostic MRI scan provides contiguous thin slices that are usable for navigation. Patients could undergo scanning days, if not weeks, before the surgery without the FMs having to be tagged or applied again. There may be reasons to postpone the surgical procedure. In such cases, the MRI navigation study does not need to be repeated. Regarding the mean values and the average difference overall for anterior and posterior checkpoints, the fiducial method values suggested that it was the most accurate for anterior checkpoints, and the SM method values suggested that it was the most accurate for posterior checkpoints. These observations are consistent with suggestions from previous authors that the posterior checkpoints are less accurate than the anterior checkpoints for placement of markers surrounding or nearer to the lesion (1, 24, 26). In contrast to previous studies (10, 21), however, our data suggested that the addition of SM points increases accuracy. This finding is slightly recognizable for the checkpoints closer to the FMs, but significant for the posterior checkpoints, which are more distant from the FMs, meaning that a widespread distribution of SM points helps to reduce inaccuracy, especially in these regions. In reality, neurosurgeons often use certain combinations of scalp-applied FMs, e.g., more


FMs close to or over the region of interest, such as a mass, in the hope of yielding better accuracy, specifically in the area of the mass. We used the same distribution of scalp-applied FMs for each patient, but the more accurate anterior checkpoints compared with the posterior checkpoints reinforce the clinical experience of many neurosurgeons of getting better results in regions that are closer to the reference points. We examined the difference in deviation between left and right for differences between the left checkpoints that are closer to the reference frame of the navigation system than the right checkpoints, but we found nothing significant. The position of the reference frame does not appear to have much influence on the accuracy for different regions of the patient’s head. Agreement between predicted and measured methods for each subject was analyzed through examination of differences or bias and characterization of these differences. The average of the two methods was used as an estimation of the true value (accuracy) because the real value was unknown and there is inherent inaccuracy associated with any method of measurement. The mean difference provided an estimate of whether the two methods, on average, are comparable. A mean difference other than zero indicates a systematic bias in one of the measures. We found a proportional error, as indicated by the difference increasing in value relative to the increase in the mean value. The measure of agreement was determined by calculating the SDs of the mean difference, and the limits of agreement fell within 2 SDs of the mean. Although these data suggest that both methods are roughly comparable, interpretation of results is ultimately based on clinical judgment. Tolerance tables were constructed on the basis of these differences. Our analyses revealed very few outliers. Although we recognize that outliers have an effect on data, we chose to keep the outliers because they reflect clinical reality. Clinicians often wish to have data for which direct measurement without adverse effects is difficult or impossible, and true values remain unknown. Instead, indirect methods of measurement are used, and a new method has to be evaluated by comparison with an established technique rather than with the true quantity. Such is the case for comparison of the fiducial registration method to the AL and SM registration methods. If the new method agrees sufficiently with the old, the old method may be replaced. This scenario is very different from calibration, in which known quantities are measured by a new method and the result is compared with the true value or with measurements made by a highly accurate method, in this case, the “gold standard” fiducial registration method. When two methods are compared, neither provides an unequivocally correct measurement, so the degree of agreement is assessed. It is most unlikely that different methods will agree exactly by yielding an identical result for all individuals. We want to know by how much the new method is likely to differ from the old. If the difference is not enough to cause problems in clinical interpretation, we can replace the old method by the new or use the two interchangeably. How far apart measurements can be without causing difficulties is a question of judgment. Ideally, this value should be defined in advance to help in



interpretation of the comparison of methods and to choose the sample size. All three methods that we analyzed appear to provide sufficient accuracy in assessing distance from the lesion for present neurosurgical procedures. Differences in the anatomic registration method, including SM registration, compared with fiducial registration, were within a mean ⫾1.96 SD according to the Bland and Altman analysis. This result indicates that differences in the anatomic methods, including SM registration, are not clinically important, and the two methods may be used interchangeably with fiducial registration. Users may safely and confidently choose to use anatomic or combined anatomic/SM registrations. This usage translates to practical and economic benefits for the hospital and the patient, including increased efficiency by saving the patient and the hospital time and cost and providing additional comfort to the patient in situations in which imaging can be performed according to the patient’s or clinic’s schedule.

REFERENCES 1. Barnett GH: Fiducial point placement and the accuracy of point-based, rigidbody registration. Neurosurgery 48:816–817, 2001 (comment). 2. Barnett GH, Miller DW, Weisenberger J: Frameless stereotaxy with scalpapplied fiducial markers for brain biopsy procedures: Experience in 218 cases. J Neurosurg 91:569–576, 1999. 3. Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307–310, 1986. 4. Brinker T, Arango G, Kaminsky J, Samii A, Thorns U, Vorkapic P, Samii M: An experimental approach to image guided skull base surgery employing a microscope-based neuronavigation system. Acta Neurochir (Wien) 140:883– 889, 1998. 5. Cohen J, Cohen P: Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale NJ, Lawrence Erlbaum Associates, 1983. 6. Edwards PJ, King AP, Maurer CR, de Cunha DA, Hawkes DJ, Hill DL, Gaston RP, Fenlon MR, Jusczyzck A, Strong AJ, Chandler CL, Gleeson MJ: Design and evaluation of a system for microscope-assisted guided interventions (MAGI). IEEE Trans Med Imaging 19:1082–1093, 2000. 7. Ganslandt O, Behari S, Gralla J, Fahlbusch R, Nimsky C: Neuronavigation: Concept, techniques and applications. Neurol India 50:244–255, 2002. 8. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF: Clinical use of a frameless stereotactic arm: Results of 325 cases. J Neurosurg 83:197–205, 1995. 9. Grunert P, Muller-Forell W, Darabi K, Reisch R, Busert C, Hopf N, Perneczky A: Basic principles and clinical applications of neuronavigation and intraoperative computed tomography. Comput Aided Surg 3:166–173, 1998. 10. Helm PA, Eckel TS: Accuracy of registration methods in frameless stereotaxis. Comput Aided Surg 3:51–56, 1998. 11. Maciunas RJ, Fitzpatrick JM, Gadamsetty S, Maurer CR: A universal method for geometric correction of magnetic resonance images for stereotactic neurosurgery. Stereotact Funct Neurosurg 66:137–140, 1996. 12. Mascott CR, Sol JC, Bousquet P, Lagarrigue J, Lazorthes Y, Lauwers-Cances V: Quantification of true in vivo (application) accuracy in cranial image-guided surgery: Influence of mode of patient registration. Neurosurgery 59 [Suppl]: ONS146–ONS156, 2006. 13. Maurer CR, Fitzpatrick JM, Wang MY, Galloway RL, Maciunas RJ, Allen GS: Registration of head volume images using implantable fiducial markers. IEEE Trans Med Imaging 16:447–462, 1997. 14. Maurer CR, Rohlfing T, Dean D, West JB, Rueckert D, Kensaku M, Shahidi R, Martin DP, Heilbrun MP, Maciunias RJ: Sources of error in image registration for cranial image-guided neurosurgery, in Germano IM (ed): Advanced Techniques in Image-Guided Brain and Spine Surgery. New York, Thieme, 2002, pp 10–36.

NEUROSURGERY

15. Muacevic A, Steiger HJ: Computer-assisted resection of cerebral arteriovenous malformations. Neurosurgery 45:1164–1171, 1999. 16. Olson JJ, Shepherd S, Bakay RA: The EasyGuide Neuro image-guided surgery system. Neurosurgery 40:1092–1096, 1997. 17. Peters TM: Image-guided surgery: From x-rays to virtual reality. Comput Methods Biomech Biomed Engin 4:27–57, 2000. 18. Poggi S, Pallotta S, Russo S, Gallina P, Torresin A, Bucciolini M: Neuronavigation accuracy dependence on CT and MR imaging parameters: A phantom-based study. Phys Med Biol 48:2199–2216, 2003. 19. Roberts DW, Strohbehn JW, Hatch JF, Murray W, Kettenberger H: A frameless stereotaxic integration of computerized tomographic imaging and the operating microscope. J Neurosurg 65:545–549, 1986. 20. Sipos EP, Tebo SA, Zinreich SJ, Long DM, Brem H: In vivo accuracy testing and clinical experience with the ISG Viewing Wand. Neurosurgery 39:194– 202, 1996. 21. Villalobos H, Germano IM: Clinical evaluation of multimodality registration in frameless stereotaxy. Comput Aided Surg 4:45–49, 1999. 22. Vrionis FD, Foley KT, Robertson JH, Shea JJ: Use of cranial surface anatomic fiducials for interactive image-guided navigation in the temporal bone: A cadaveric study. Neurosurgery 40:755–764, 1997. 23. West J, Fitzpatrick JM, Wang MY, Dawant BM, Maurer CR, Kessler RM, Maciunas RJ, Barillot C, Lemoine D, Collignon A, Maes F, Suetens P, Vandermeulen D, van den Elsen PA, Napel S, Sumanaweera TS, Harkness B, Hemler PF, Hill DL, Hawkes DJ, Studholme C, Maintz JB, Viergever MA, Malandain G, Woods RP: Comparison and evaluation of retrospective intermodality brain image registration techniques. J Comput Assist Tomogr 21:554–566, 1997. 24. West JB, Fitzpatrick JM, Toms SA, Maurer CR, Maciunas RJ: Fiducial point placement and the accuracy of point-based, rigid body registration. Neurosurgery 48:810–817, 2001. 25. Wirtz CR, Knauth M, Hassfeld S, Tronnier VM, Albert FK, Bonsanto MM, Kunze S: Neuronavigation—first experiences with three different commercially available systems. Zentralbl Neurochir 59:14–22, 1998. 26. Wolfsberger S, Rössler K, Regatschnig R, Ungersböck K: Anatomical landmarks for image registration in frameless stereotactic neuronavigation. Neurosurg Rev 25:68–72, 2002.

Acknowledgments This work was supported by the Newsome Chair in Neurosurgery Research at the Barrow Neurological Institute, which is held by Mark C. Preul, M.D.

COMMENTS

T

he authors have reviewed a series of patients who underwent direct comparison between fiducial versus nonfiducial neuronavigation registration and have shown no comparable difference between the two methods of registration. The use of nonfiducial anatomic landmarks would save the need for repeating magnetic resonance imaging scans to have fiducial markers present if one already has a magnetic resonance imaging scan without fiducial markers that can be loaded onto neuronavigation systems. Scans for many brain lesions, particularly benign tumors, could be done days before surgery, allowing more flexibility in terms of preoperative preparation. Further studies would help support the conclusions in this article and would be necessary for widespread acceptance of nonfiducial neuronavigation registration. Steven D. Chang Stanford, California

I

n this article, the authors compare fiducial registration with anatomical landmark registration to merged point surface registration using a Stealth system. They found that the concordance between these three methods was reasonably close enough for most frameless stereotactic


PFISTERER ET AL.

procedures. I have no problems with their scientific approach, methods, writing style, or illustrations. And I agree with their conclusion: that nonfiducial registration is probably accurate enough for most intracranial frameless stereotactic procedures. This statement assumes, however, that all who incorporate these conclusions into clinical practice have similar instrumentation and software for merged point surface registration and/or exhibit the same degree of care as the authors in matching patient and imaging defined anatomical points. This, as the authors state, would allow a presurgical data set to be acquired at any time without the hassle of placing stickers on the head and the worry that the patient will remove these before the operation. Patrick J. Kelly New York, New York

T

he authors have taken advantage of the opportunity to determine the application accuracy of a few methods of registration for surgical navigation (frameless stereotaxy) compared with a standard of high-resolution frame-based stereotaxy. Their results show that all three methods for registering the navigation systems are good, with comparable standard deviations. Most previous studies have tried to determine accuracy using cadavers or phantoms, neither of which have the skin mobility of the living patient. Although these results are good, they remain inferior to those attainable with cranium-implanted fiducial markers, which should be considered, particularly when high accuracy is required. Gene H. Barnett Cleveland, Ohio

Anatomy Lecture in Leiden, (1609), copper engraving, Bartholomeus Dolendo. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



FUNCTIONAL AND STEREOTACTIC Technical Report

Neculai Archip, Ph.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Olivier Clatz, Ph.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Stephen Whalen, B.Sc. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Simon P. DiMaio, Ph.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Peter M. Black, M.D., Ph.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Ferenc A. Jolesz, M.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Alexandra Golby, M.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts

Simon K. Warfield, Ph.D. Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts Reprint requests: Ferenc A. Jolesz, M.D, Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115. Email: [email protected] Received, August 25, 2006. Accepted, July 31, 2007.

NEUROSURGERY

COMPENSATION OF GEOMETRIC DISTORTION EFFECTS ON INTRAOPERATIVE MAGNETIC RESONANCE IMAGING FOR ENHANCED VISUALIZATION IN IMAGE-GUIDED NEUROSURGERY OBJECTIVE: Preoperative magnetic resonance imaging (MRI), functional MRI, diffusion tensor MRI, magnetic resonance spectroscopy, and positron-emission tomographic scans may be aligned to intraoperative MRI to enhance visualization and navigation during image-guided neurosurgery. However, several effects (both machine- and patientinduced distortions) lead to significant geometric distortion of intraoperative MRI. Therefore, a precise alignment of these image modalities requires correction of the geometric distortion. We propose and evaluate a novel method to compensate for the geometric distortion of intraoperative 0.5-T MRI in image-guided neurosurgery. METHODS: In this initial pilot study, 11 neurosurgical procedures were prospectively enrolled. The scheme used to correct the geometric distortion is based on a nonrigid registration algorithm introduced by our group. This registration scheme uses image features to establish correspondence between images. It estimates a smooth geometric distortion compensation field by regularizing the displacements estimated at the correspondences. A patient-specific linear elastic material model is used to achieve the regularization. The geometry of intraoperative images (0.5 T) is changed so that the images match the preoperative MRI scans (3 T). RESULTS: We compared the alignment between preoperative and intraoperative imaging using 1) only rigid registration without correction of the geometric distortion, and 2) rigid registration and compensation for the geometric distortion. We evaluated the success of the geometric distortion correction algorithm by measuring the Hausdorff distance between boundaries in the 3-T and 0.5-T MRIs after rigid registration alone and with the addition of geometric distortion correction of the 0.5-T MRI. Overall, the mean magnitude of the geometric distortion measured on the intraoperative images is 10.3 mm with a minimum of 2.91 mm and a maximum of 21.5 mm. The measured accuracy of the geometric distortion compensation algorithm is 1.93 mm. There is a statistically significant difference between the accuracy of the alignment of preoperative and intraoperative images, both with and without the correction of geometric distortion (P ⬍ 0.001). CONCLUSION: The major contributions of this study are 1) identification of geometric distortion of intraoperative images relative to preoperative images, 2) measurement of the geometric distortion, 3) application of nonrigid registration to compensate for geometric distortion during neurosurgery, 4) measurement of residual distortion after geometric distortion correction, and 5) phantom study to quantify geometric distortion. KEY WORDS: Image-guided neurosurgery, Magnetic resonance image geometric distortion, Multimodal registration, Nonrigid registration Neurosurgery 62[ONS Suppl 1]:ONS209–ONS216, 2008

I

nterventional magnetic resonance imaging (iMRI) was introduced to enhance intraoperative visualization and has been found to

DOI: 10.1227/01.NEU.0000297081.51540.45

increase the volume of resected low-grade tumors (2, 3, 15, 23). Several recent studies have demonstrated the effectiveness of iMRI


ARCHIP ET AL.

in achieving gross tumor resection (9, 29). Because functional magnetic resonance imaging (fMRI) of tactile, motor, and language tasks is feasible in patients with cerebral tumors (21), several groups have recently proposed integration of functional data into neuronavigation systems (16). Diffusion tensor imaging (DTI) provides information about normal tumor course, displacement or interruption of white matter tracts around the tumor, and fiber-bundle widening that results from edema or tumor. Consequently, efforts have been made in the last years to combine DTI with neurosurgical navigation systems (24, 34). Therefore, the goal of advanced image-guided neurosurgery systems is to accurately align the preoperatively acquired images (fMRI and DTI) with the images obtained intraoperatively during tumor resection (1). Important geometric distortions often occur in MRI that lead to pixel shifts in the acquired images (21, 35). Several reports describe the importance of these variations, which can be several millimeters in certain areas of the field of view and thus hinder the precise localization of anatomic structures (32). MRI geometric distortion is caused by artifacts that violate the assumptions of spatial encoding in MRI. These artifacts can be categorized into those characteristic of the imaging hardware and those resulting from patient characteristics. The sources of machine-induced magnetic resonance geometric distortion includes static field inhomogeneity, gradient field nonlinearity, and the presence of eddy currents caused by gradient switching. The distortions induced by gradient nonlinearity and main magnetic field nonuniformity are independent of the patient’s position within the scanner. These can be corrected independent of other machine-induced distortions. Sources of patient-induced geometric distortion include magnetic susceptibility effects, chemical shift, and flow. The vendors of clinical MRI systems often provide software for gradient distortion correction. However, recent studies have estimated the residual gradient distortions (6, 25, 36, 37) and have shown that significant distortions are present even after the gradient distortion correction software is applied. Methods for correcting machine-induced geometric distortion are presented by Doran et al. (11), Langlois et al. (17), and Wang et al. (36, 37). The correction of geometric distortion in stereotactic MRI for neurosurgery (10, 20, 38) and bilateral subthalamic stimulation in Parkinson’s disease (19) have also been investigated. Geometric distortion is also commonly encountered when intraoperative MRI scanners are used. For instance, Petersch et al. (24) report maximum distortions of 28 mm (mean, 2.2 mm) that were measured within the field of view in the frequency-encode direction. However, individuals receiving MRIguided therapies would also benefit from correction of patientinduced geometric distortion. To the best of our knowledge, there is no proposed method that can compensate for these distortions during clinical interventions. In this report, we measure the distortion between preoperative and intraoperative images. We also present a novel technique based on a nonrigid registration scheme to reduce the geometric distortion in a 0.5-T open magnet in the context of iMRI-guided neurosurgery. This technique compensates for


both machine- and patient-induced distortions. We describe the technical aspects of our method, its implementation, and its validation. We show that this novel method enables accurate alignment of preoperative datasets to intraoperative images and thus provides neurosurgeons with enhanced information during tumor resection.

PATIENTS AND METHODS Eleven consecutive patients (6 women, 5 men; age range, 28–62 yr; mean, 45.2 yr) with supratentorial gliomas (World Health Organization Grade II, 5 patients; Grade III, 4 patients; Grade IV, 2 patients) were included in our study. All patients underwent surgery at our institution’s intraoperative MRI-guided therapy facility between April 2005 and January 2006 for tumors in and adjacent to eloquent brain areas (such as the precentral gyrus and corticospinal tract for motor function and Broca’s and Wernicke’s areas for language function). For these patients, fMRI and DTI were used for preoperative surgical planning. The study was performed with Institutional Review Board approval and all patients provided informed consent.

Preoperative Imaging Each patient provided informed consent, and the following preoperative MRI protocol was followed several days before surgery was scheduled to occur. We used a 3-T Signa (General Electric, Milwaukee, WI) scanner.

Anatomic Imaging We first obtained whole-brain sagittal three-dimensional spoiled, gradient-recalled images in steady state (slice thickness, 1.3 mm; TE/TR, 6/35 ms; flip angle, 75 degrees; field of view [FOV], 24 cm; matrix, 256 ⫻ 256). Subsequently, we obtained axial T2-weighted fastspin-echo images (slice thickness, 5 mm; TE/TR, 100/3000 ms; FOV, 22 cm; matrix, 512 ⫻ 512).

Functional Magnetic Resonance Imaging Whole-brain functional images were acquired with a T2*-weighted, echo-planar sequence that was sensitive to the blood oxygen leveldependent signal (TR, 2000 ms; TE, 30 ms; matrix, 64 ⫻ 64 ⫻ 6 mm; FOV, 240 mm; imaging, 24 contiguous slices of 5-mm thickness).

Diffusion Tensor Imaging Axial line-scan diffusion images (slice thickness, 5 mm; matrix, 512 ⫻ 512; FOV, 24 cm) and echo-planar DTI (matrix, 128 ⫻ 128; phase FOV, 1.0; FOV, 25.6; slice thickness, 3; B value, 800; directions, 31; number of T2, 1) were acquired to cover the entire region of interest as well as “landmark” regions, i.e., areas where the relevant fiber tracts showed high density (e.g., ventral brainstem for the corticospinal tract and lateral geniculate body for the optic radiation).

Magnetic Resonance Spectroscopy For three patients, we also performed magnetic resonance spectroscopy from the tumor.

Intraoperative Imaging After the patients were positioned for craniotomy and their heads were fixed using a magnetic resonance-compatible carbon fiber Mayfield clamp (Ohio Medical Instruments, Cincinnati, OH), imaging was performed using the following protocol in the vertically open 0.5-T iMRI unit (SignaSP; General Electric Medical Systems)


MAGNETIC RESONANCE IMAGING DISTORTION

with the following parameters. For transverse, sagittal, and coronal T1-weighted, fast spin-echo imaging, the repetition time/echo time (both in milliseconds) was 700/29; FOV, 22 cm; matrix, 256 ⫻ 256; number of signals acquired, 1; section thickness, 3 mm; and intersection gap, 1 mm. For transverse, T2-weighted, fast-spin echo imaging, the repetition time/echo time (both in milliseconds) was 5000/99, FOV, 22 cm; matrix, 256 ⫻ 256; number of signals acquired, 2; section thickness, 3 mm; and intersection gap, 1 mm. For transverse, threedimensional, spoiled gradient-echo imaging, the repetition time/echo time (both in milliseconds) was 15.5/5.2; flip angle, 45 degrees; FOV, 22 cm; matrix, 256 ⫻ 256; number of signals acquired, 1; section thickness, 2.5 mm; and intersection gap, 0 mm.

A

B

C

D

Data Processing Our intraoperative visualization system of multimodal images uses T1-gradient echo (spoiled-gradient echo) sequences to estimate the deformation field that results from brain shift and geometric distortion. We focused our study on the T1-spoiled-gradient echo images. Geometric distortion is present on intraoperative imaging as compared with preoperative 3-T MRI scans (Fig. 1). Accurate alignment of preoperative and intraoperative images requires an algorithm to compensate for the geometric distortion. The geometry of intraoperative images is changed to match the preoperative images. We address this by using a nonrigid registration scheme that was first introduced by our group (8) to compensate for brain shift. This registration scheme uses image features to establish correspondence between images. As used here, it estimates a smooth geometric distortion compensation field by regularizing the displacements estimated at the correspondences. A patient-specific linear elastic material model is used to achieve the regularization. The algorithm can be decomposed into three primary parts. For Part 1, after the preoperative images have been acquired, the patient-specific model is built using image segmentation and mesh generation. Image segmentation is the delineation of structures in the intraoperative data using segmentation strategies that are optimized for the particular type of acquisition. This approach combines the benefits of anatomic information, statistical classification, and elastic matching to achieve results superior to those obtained by any single method alone. Recently, we have also successfully used a method based on a deformable model, which evolves to the brain’s surface by the application of a set of locally adaptive model forces (31). For mesh generation, the tetrahedral discretization (volume mesh) of the segmented intracranial cavity provides the basis for a finite-element method of modeling the physical tissue deformation and serves the function of regularizing of the estimated displacements that were obtained from the blockmatching step of nonrigid registration. The technique used for tetrahedral mesh generation is described by Fedorov et al. (12). It uses implicit representation of the object as input and produces an adaptive tetrahedral mesh specifically suited for applications that exhibit high deformation. An example of a tetrahedral mesh and its corresponding brain segmentation are presented in Figure 2. From these, a patient-specific model is obtained of the brain material using an incompressible linear elastic constitutive equation. Part 2 of the algorithm, which is performed intraoperatively, is the block-matching computation for a set of selected blocks on images. This step estimates a set of displacements across the volume. Part 3 of the algorithm is an iterative hybrid solver that estimates the three-dimensional volumetric deformation field induced by geometric distortion of intraoperative imaging. In this step, the patient-specific model is used to regularize the distortion-compensation field as described in detail by Clatz et al. (8).

NEUROSURGERY

FIGURE 1. Typical magnetic resonance imaging (MRI) scans from neurosurgery patients who participated in our study. A, edges of preoperative 3-T MRI scan are extracted with a Canny operator and are drawn in red. B, intraoperative 0.5-T MRI scan for the same patient, before craniotomy. Significant geometric distortion can be observed. C, the edges of the 3-T MRI images are displayed over the intraoperative 0.5-T image. The evident misalignment is the result of geometric distortion. D, landmarks are selected on the preoperative 3-T image (red) and on the intraoperative image (green). The distance between the two landmarks is the result of geometric distortion.

A

B

FIGURE 2. A, segmentation of brain from the preoperative MRI scan. B, mesh generated from the segmentation (A) is used to create a biomechanical model of the brain. Nonrigid registration algorithms are typically computationally expensive, and parallel computing may be used to accelerate the computation to reduce the computation time to clinically acceptable levels. We have investigated the use of symmetric multiprocessor, cluster, and grid computing hardware to provide accelerated computation (7). Modern hardware enables rapid and effective solution of the system of equations that arises in this approach to geometric distortion.

Geometric Distortion Measurement The Canny edge detector is commonly used in computer vision to locate sharp intensity changes and find object boundaries in an image (5). The Canny edge detector removes the weak edges using a hysteresis threshold. We used it to extract brain edges from the MRI scans as shown in Figure 1A. The edges are distinguished and represented as a set of points.


ARCHIP ET AL.

The Hausdorff metric is a common mathematical measure for comparing two sets of points in terms of their least-similar members. Formally, given two finite point sets: A ⫽ {a1, ... , ap} and B ⫽ {b1, ... , bq} the Hausdorff metric is defined as: H (A, B) ⫽ max{h(A, B), h(B, A)} where: h(A, B) ⫽ maxa⑀A minb⑀B ||a ⫺ b|| and ||⭈|| is the Euclidean norm. The 95% Hausdorff distance is measured between the points on the edges extracted from the two images (the pre- and intraoperative images) with a Canny operator. Ideally, when there are no geometric distortions present, this distance should be zero. Obtaining the 95% Hausdorff value ensures that the outliers are rejected.

RESULTS Our technique was evaluated while treating 11 consecutive neurosurgery patients. The data were transferred, processed, and displayed in the operating room during the neurosurgical procedure. An example of an intraoperative MRI scan obtained after performing compensation for geometric distortion is presented in Figure 3. In Figure 4, alignments among the preoperative DTI and the 3-T T1 MRI scan, the preoperative DTI and the intraoperative 0.5-T T1 MRI scan, and the preoperative DTI and the intraoperative 0.5-T T1 MRI scan after geometric correction are presented. In Figure 5, alignments among the preoperative fMRI and 3-T T1 MRI scans, the preoperative fMRI and intraoperative 0.5-T T1 MRI scans, and the preoperative fMRI and intraoperative 0.5-T T1 MRI scans after geometric correction are presented. The magnitude of the geometric distortion between the images acquired at 0.5 and 3 T is calculated before and after application of the distortion compensation technique using

A

B

FIGURE 3. Compensation for geo- C metric distortion. A, corrected intraoperative 0.5-T image. B, the same image as in A reveals the contours extracted from the original intraoperative image. The differences, and the magnitude of the geometric distortion, can be observed. C, in the corrected intraoperative image, the contours are extracted from the preoperative image. They match well because we have compensated for the distortion.


A

B

FIGURE 4. A, alignment between the preoperative diffusion tensor imaging (DTI) and 3-T T1 MRI scans. B, alignment between the preoperative DTI and intraoperative 0.5-T T1 MRI scans, without correction of geometric distortion. C, alignment between preoperative DTI and intraoperative 0.5-T T1 MRI scans after geometric correction.

C

A

B

FIGURE 5. A, alignment between the preoperative functional magnetic resonance imaging (fMRI) and 3-T T1 MRI scans. B, alignment between the preoperative fMRI and intraoperative 0.5-T T1 MRI scans without correction of geometric distortion. C, alignment between preoperative fMRI and intraoperative 0.5-T T1 MRI scans after geometric correction.

C

anatomic landmarks and the Hausdorff distance as computed between edges extracted from MRI scans using the Canny operator. Overall, the mean magnitude of the geometric distortion was 10.3 mm, with a minimum of 2.91 mm and a maximum of 21.5 mm. The accuracy for our geometric distortioncompensation algorithm, measured based on the 95% Hausdorff distance, was 1.93 mm. There was a statistically significant difference between the accuracy of the alignment of pre- and intraoperative images with and without the compensation of geometric distortion (P ⬍ 0.0004). The complete results are presented in Table 1. The registration results have also been reviewed by a team of neurosurgeons from our department and were judged to be adequate. On a Dell Precision 690n computer (Dell, Round Rock, TX) with four Intel Xeon 5160 processor cores (Intel, Santa Clara, CA) running at 3.0 GHz, execution time is approximately 18 minutes. Additional reductions in computation time are possible if more computers are used.



TABLE 1. Results showing a statistically significant difference between the pre- and postoperative imagesa Geometric distortion magnitude, mm Before After correction correction

Tumor Patient no. Position

Pathology (WHO grade)a

1

Right posterior frontal

Oligoastrocytoma (II)

8.10

1.20

6.75

2

Left posterior temporal

Glioblastoma (IV)

15.10

2.40

6.29

3

Left medial temporal

Glioblastoma (IV)

21.50

3.53

6.09

4

Left temporal

Anaplastic oligoastrocytoma (III)

5.70

1.70

3.35

5

Right frontal


2.91

0.85

3.42

6

Left frontal

Anaplastic astrocytoma (III)

11.10

2.5

4.44

7

Right medial temporal

Anaplastic astrocytoma (III)

20.00

3.14

6.36

8

Right frontal


14.00

2.58

5.42

9

Right frontotemporal


5.57

1.20

4.64

10

Right occipital

Anaplastic oligodendroglioma (III)

2.85

0.85

3.35

11

Left frontotemporal

Oligodendroglioma (II)

7.23

1.34

5.39

10.36

1.93

5.04

Average a

No correction/ correction ratio

WHO, World Health Organization.

In addition to assessing the geometric distortion, we have also measured the magnitude of brain deformation that results from tumor resection. The maximum image deformation resulting from both brain shift and geometric distortion was determined to be 21.3 mm. We have previously described in detail our approach to brain-shift quantification (1). Illustrative images of compensation for geometric distortion and brain shift are presented in Figure 6. We also used a General Electric calibration phantom to measure geometric distortion. The phantom was scanned using the same protocol as applied for the neurosurgery patients. Both 3- and 0.5T images were acquired. Additionally, we scanned the same phantom with a computed tomographic (CT) scan. The disagreement was quantified and represents geometric distortion. The maximum displacement measured with the phantom between the CT scan and the 0.5-T MRI scan was 5 m, and between the CT and the 3-T MRI, it was 1 mm. Results are illustrated in Figure 7.

DISCUSSION Machine-induced MRI distortions have been intensively studied, and several methods to correct them have been proposed. However, patient-induced geometric distortion has received less attention in the context of image-guided neurosurgery. We identified geometric distortion between preoperative 3-T MRI scans and intraoperative 0.5-T MRI scans. Computed tomography provides images with less geometric distortion. However, the present clinical protocol at our institution does not include any CT imaging for brain neurosurgery at the iMRI. Nevertheless, in our study, we demonstrate that we can improve the accuracy for neurosurgical navigation by compensating for geometric distortion of iMRI. Because our 0.5-T iMRI scanner is only for clinical use, we were unable to scan animals or cadavers.

NEUROSURGERY

Instead, we obtained 3- and 0.5-T MRI and CT scans of a General Electric calibration phantom. We measured the differences between these images. Although computed tomography reproduces accurately the geometry of the phantom, we identified differences on 3- and 0.5-T MRI scans. The distortion present on the 0.5-T MRI scan is larger than on the 3-T MRI scan (5 versus 1 mm). Also, based on the anatomic features of the brain MRI scans, qualitative assessment indicates that 3-T images are less distorted than 0.5-T images. Moreover, the 3-T images have higher resolution than the 0.5-T images. Therefore, in our study, we consider the 3-T images as the reference, and modify the geometry of the intraoperative 0.5-T images. We use the phrase “intraoperative imaging” to describe all imaging performed on the 0.5-T MRI machine, and “preoperative imaging” to describe data acquired to plan the surgery using the 3.0-T MRI machine. These two scanners have different geometric distortion properties, with the 0.5-T MRI scanner having the largest distortion, which must be removed for accurate intraoperative navigation. This distortion can change across the course of the surgery as a result of the craniotomy, but after craniotomy, we also see significant soft-tissue deformation. Our focus here is on compensation for the geometric distortion. The geometric distortion associated with intraoperative MRI acquisition is a common problem for all existing open-MRI scanners. It is a consequence of both properties of the patient, such as magnetic susceptibility changes resulting from craniotomy, and properties of the magnet design, such as the homogeneity of the static magnetic field. When using an open-magnet design, obtaining a large region of homogeneous, static, magnetic field is a challenging task. For instance, Petersch et al. (24) report maximum distortions of 28 mm (mean, 2.2 mm) within the FOV in the frequency-encode direction. The scanner


ARCHIP ET AL.

A

B

C

D

A

B

C FIGURE 7. A General Electric calibration phantom was scanned using computed tomography (A), 3T MRI (B), and 0.5-T MRI (C). The distortion (5 mm for the 3-T image and 1 mm for 0.5-T image) is clear.

FIGURE 6. Compensation for geo- E metric distortion and brain shift for intraoperative imaging scans. A, preoperative 3-T MRI scan. B, first intraoperative image before craniotomy with geometric distortion. C, the image in B has been corrected. D, intraoperative image obtained during tumor resection with geometric distortion and brain shift. E, the image in D has been corrected.

used was a Siemens Magnetom Open Viva 0.2-T resistive MRI scanner (New York, NY). The clinical application was radiotherapy treatment of prostate cancer. There are many sources of geometric distortion in magnetic resonance imaging, all of which contribute by fundamentally disrupting the assumption of linear encoding of position in space using frequency and phase encoding in MRI scans. Here, we are concerned with identifying, measuring, and compensating for this geometric distortion. Geometric distortion is present as a result of properties of the scanner; it is subject to and impacts all of the data acquired, even before craniotomy is performed. Soft-tissue deformation such as brain shift occurs only after craniotomy and exists in addition to geometric distortion. Our group has already presented results regarding image registration for brain-shift compensation, for instance, in studies by Nabavi et al. (22) or more recently, by Ruiz-Alzola et al. (27), Clatz et al. (8), and Archip et al. (1). In the present study, we only focus on the distortion in the images before soft-tissue deformation occurs. In our study, we measured distortions as large as 21.3 mm. Therefore, for accurate navigation in neurosurgery and even for data fusion restricted to rigid-body transformation, compensation for this geometric distortion is essential.


In this report, we propose a solution to address these distortions using a nonrigid registration scheme. A robust, accurate, and sufficiently rapid nonrigid registration algorithm was used to compensate for geometric distortion. For each patient, we were able to successfully correct the geometric distortion. A clinically compatible execution time was achieved by using parallel computing and by performing key image-processing steps before the surgery.

CONCLUSION The major contributions of this study are identification of geometric distortion of intraoperative 0.5-T images as compared with preoperative 3-T images, measurement of the size of the geometric distortion, application of nonrigid registration to compensate for geometric distortion during neurosurgery, measurement of the residual distortion after geometric distortion correction, and phantom study to quantify geometric distortion. As clearly demonstrated, after we applied our correction method, the residual geometric distortion apparent in the corrected images was negligibly small for all patients studied. The introduced technology combined with an advanced neurosurgery navigation system enables the use of high-accuracy navigation with preoperative DTI and fMRI scans during tumor resection.

REFERENCES 1. Archip N, Fedorov A, Lloyd B, Chrisochoides N, Golby A, Black PMcL, Warfield SK: Integration of patient specific modeling and advanced image processing techniques for image guided neurosurgery. Presented at the SPIE Medical Imaging, San Diego, CA, February 17–22, 2006. 2. Black PMcL, Alexander E 3rd, Martin C, Moriarty T, Nabavi A, Wong TZ, Schwartz RB, Jolesz F: Craniotomy for tumor treatment in an intraoperative magnetic resonance imaging unit. Neurosurgery 45:423–431, 1999. 3. Black PMcL, Moriarty T, Alexander E 3rd, Stieg P, Woodard EJ, Gleason PL, Martin CH, Kikinis R, Schwartz RB, Jolesz FA: Development and implementation of intraoperative magnetic resonance imaging and its neurosurgical applications. Neurosurgery 41:831–842, 1997.



4. Bradley WG: Achieving gross total resection of brain tumors: Intraoperative MR imaging can make a big difference. AJNR Am J Neuroradiol 23:248–349, 2002. 5. Canny JF: A computational approach to edge detection. IEEE Trans Pattern Analysis and Machine Intelligence 8:679–698, 1986. 6. Chen Z, Ma CM, Paskalev K, Li J, Yang J, Richardson T, Palacio L, Xu X, Chen L: Investigation of MR image distortion for radiotherapy treatment planning of prostate cancer. Phys Med Biol. 21:1393–1403, 2006. 7. Chrisochoides N, Fedorov A, Kot A, Archip N, Black PMcL, Clatz O, Golby A, Kikinis R, Warfield SK: Toward real-time image guided neurosurgery using distributed and grid computing. Presented at the Supercomputing, Tampa, FL, November 11–17, 2006. 8. Clatz O, Delingette H, Talos IG, Golby AJ, Kikinis R, Jolesz FA, Ayache N, Warfield SK: Robust non-rigid registration to capture brain shift from intraoperative MRI. IEEE Trans Med Imaging 24:1417–1427, 2005. 9. Claus EB, Horlacher A, Hsu L, Schwartz RB, Dello-Iacono D, Talos F, Jolesz FA, Black PMcL: Survival rates in patients with low-grade glioma after intraoperative magnetic resonance image guidance. Cancer 103:1227–1233, 2005. 10. Dean D, Kamath J, Duerk JL, Ganz E: Validation of object-induced MR distortion correction for frameless stereotactic neurosurgery. IEEE Trans Med Imaging 17:810–816, 1998. 11. Doran SJ, Charles-Edwards L, Reinsberg SA, Leach MO: A complete distortion correction for MR images: I. Gradient warp correction. Phys Med Biol 50:1342–1361, 2005. 12. Fedorov A, Chrisochoides N, Kikinis R, Warfield SK: An evaluation of three approaches to tetrahedral mesh generation for deformable registration of MR images. Presented at the IEEE International Symposium on Biomedical Imaging: From Nano to Macro, Arlington, VA, April 6–9, 2006. 13. Haberg A, Kvistad KA, Unsgard G, Haraldseth O: Preoperative blood oxygen level-dependent functional magnetic resonance imaging in patients with primary brain tumors: Clinical application and outcome. Neurosurgery 54:902–914, 2004. 14. Hughes DG, Robertson S, Allen PS: Intensity artifacts in MRI caused by gradient switching in an animal-size NMR magnet. Magn Reson Med 25:167–179, 1992. 15. Jolesz FA, Nabavi A, Kikinis R: Integration of interventional MRI with computer-assisted surgery. J Magn Reson Imaging 13:69–77, 2001. 16. Krishnan R, Raabe A, Hattingen E, Szelenyi A, Yahya H, Hermann E, Zimmermann M, Seifert V: Functional magnetic resonance imaging integrated neuronavigation: Correlation between lesion-to-motor cortex distance and outcome. Neurosurgery 55:904–914, 2004. 17. Langlois S, Desvignes M, Constans JM, Revenu M: MRI geometric distortion: A simple approach to correcting the effects of non-linear gradient fields. J Magn Reson Imaging 9:821–831, 1999. 18. Maurer CR Jr, Aboutanos GB, Dawant BM, Gadamsetty S, Margolin RA, Maciunas RJ, Fitzpatrick JM: Effect of geometrical distortion correction in MR on image registration accuracy. J Comput Assist Tomogr 20:666–679, 1996. 19. Menuel C, Garnero L, Bardinet E, Poupon F, Phalippou D, Dormont D: Characterization and correction of distortions in stereotactic magnetic resonance imaging for bilateral subthalamic stimulation in Parkinson disease. J Neurosurg 103:256–266, 2005. 20. Michiels J, Bosmans H, Pelgrims P, Vandermeulen D, Gybels J, Marchal G, Suetens P: On the problem of geometric distortion in magnetic resonance images for stereotactic neurosurgery. Magn Reson Imaging 12:749–765, 1994. 21. Mueller WM, Yetkin FZ, Hammeke TA, Morris GL, Swanson SJ, Reichert K, Cox R, Haughton VM: Functional magnetic resonance imaging mapping of the motor cortex in patients with cerebral tumors. Neurosurgery 39:515–520, 1996 22. Nabavi A, Black PMcL, Gering DT, Westin CF, Mehta V, Pergolizzi RS Jr, Ferrant M, Warfield SK, Hata N, Schwartz RB, Wells WM 3rd, Kikinis R, Jolesz FA: Serial intraoperative magnetic resonance imaging of brain shift. Neurosurgery 48:787–797, 2001. 23. Nimsky C, Ganslandt O, Hastreiter P, Wang R, Benner R, Sorensen AG, Fahlbusch R: Intraoperative diffusion-tensor MR imaging: Shifting of white matter tracts during neurosurgical procedure: Initial experience. Radiology 234:218–225, 2005. 24. Petersch B, Bogner J, Fransson A, Lorang T, Potter R: Effects of geometric distortion in 0.2T MRI on radiotherapy treatment planning of prostate cancer. Radiother Oncol 71:55–64, 2004.

NEUROSURGERY

25. Piepmeier JM, Christopher S, Spencer D, Byrne T, Kim J, Knisel JP, Lacy J, Tsukerman L, Makuch R: Variations in the natural history and survival of patients with supratentorial low-grade astrocytomas. Cancer 38:872–878, 1996. 26. Posse S: Direct imaging of magnetic field gradients by group spin-echo selection. Magn Reson Med 25:12–29, 1992. 27. Ruiz-Alzola J, Westin CF, Warfield SK, Alberola C, Maier SE, Kikinis R: Nonrigid registration of 3D tensor medical data. Medical Image Analysis 6:143–161, 2002. 28. Schenck JF: The role of magnetic susceptibility in magnetic resonance imaging: MRI magnetic compatibility of the first and second kinds. Med Phys 23:815–850, 1996. 29. Schneider JP, Trantakis C, Rubach M, Schulz T, Dietrich J, Winkler D, Renner C, Schober R, Geiger K, Brosteanu O, Zimmer C, Kahn T: Intraoperative MRI to guide the resection of primary supratentorial glioblastoma multiforme—A quantitative radiological analysis. Neuroradiology 47:489–500, 2005. 30. Skare S, Andersson JL: Correction of MR image distortions induced by metallic objects using a 3D cubic B-spline basis set: Application to stereotactic surgical planning. Magn Reson Med 54:169–181, 2005. 31. Smith SM: Fast robust automated brain extraction. Hum Brain Mapp 17:143–155, 2002. 32. Sumanaweera TS, Glover G, Song S, Adler J, Napel S: Quantifying MRI geometric distortion in tissue. Magn Reson Med 31:40–47, 1994. 33. Sun L, Aletras AH, Schmalbrock P, Skinner TE, Chakeres D, Irsik R, Robitaille PM: Water and fat MR imaging with chemical shift selective 3D steady state methods. Magn Reson Med 31:359–364, 1994. 34. Talos IF, O’Donnell L, Westin CF, Warfield SK, Wells WM, Yoo SS, Panych L, Golby A, Mamata H, Maier SE, Ratiu P, Guttmann CG, Black PMcL, Jolesz FA, Kikinis R: Diffusion tensor and functional MRI fusion with anatomical MRI for image guided neurosurgery. Presented at the Sixth International Conference on Medical Image Computing and Computer-Assisted Intervention, Montreal, Canada, November 16–18, 2003. 35. Walton L, Hampshire A, Forster DMC, Kemeny AA: A phantom study to assess the accuracy of stereotactic localization, using T1 weighted magnetic resonance imaging with the Leksell stereotactic system. Neurosurgery 38:170–178, 1996. 36. Wang D, Strugnell W, Cowin G, Doddrell DM, Slaughter R: Geometric distortion in clinical MRI systems Part I: Evaluation using a 3D phantom. Magn Reson Imaging 22:1211–1221, 2004. 37. Wang D, Strugnell W, Cowin G, Doddrell DM, Slaughter R: Geometric distortion in clinical MRI systems Part II: Correction using a 3D phantom. Magn Reson Imaging 22:1223–1232, 2004. 38. Weis J, Ericsson A, Silander HC, Hemmingsson A: Magnetic resonance spectroscopic imaging for visualization and correction of distortions in MRI: High precision applications in neurosurgery. Magn Reson Imaging 16:1265–1272, 1998. 39. Wirtz CR, Knauth M, Staubert A, Bonsanto MM, Sartor K, Kunze S, Tronnier VM: Clinical evaluation and follow up results for intraoperative magnetic resonance imaging in neurosurgery. Neurosurgery 46:1120–1122, 2000. 40. Zhu XX, Macdonald PM: Empirical compensation function for eddy current effects in pulsed field gradient nuclear magnetic resonance experiments. Solid State Magn Reson 4:217–227, 1995.

Acknowledgments We thank GE for the research support. This investigation was supported in part by National Science Foundation Information Technology Research 0426558; National Multiple Sclerosis Society Award #RG 3478A2/2, a research grant from CIMIT; and by National Institutes of Health grants R03 EB-006515, U41 RR-019703, P01 CA-067165, R01 RR021885, R03 CA-126466, P30 HD-018655, and R01 HL-074942.

COMMENTS

M

ultimodality navigation is becoming increasingly popular. Data and images are registered with each other, resulting in a multimodality three-dimensional framework that is used to visualize anatomy, function, metabolism, and other information in the surgical field. This


ARCHIP ET AL.

kind of advanced navigation, even combined with intraoperative imaging, is prone to errors owing to the registration process itself and to varying spatial distortion of the original data from each modality. Archip et al. present a sophisticated approach to deal with this problem by application of a nonrigid registration algorithm to compensate for the geometric distortion. Their challenge was to register various preoperative 3-T data with their intraoperative 0.5-T data. The 0.5-T data seemed to be most distortion vulnerable, which might be a problem of mid- and low-field magnetic resonance scanners in general. The study focuses on the comparison with 0.5-T intraoperative data, which were acquired before craniotomy. These distortions have to be separated from the effects of intraoperative imaging per se caused by intraoperative events such as brain shift and effects due to the air-brain interface during actual intraoperative imaging. The solution presented allows integration of multimodality data that were acquired with different magnetic resonance scanners in the intraoperative setting with reliable accuracy by compensating for the distortion effects of the 0.5-T magnetic resonance scanner. An alternative approach to circumvent this problem is to acquire all different data with the same scanner, preferably in the same setting, so that the patient’s individual effects remain the same. In our setup of intraoperative high-field magnetic resonance imaging (MRI) applying a 1.5-T scanner, functional MRI (fMRI), diffusion tensor imaging (DTI), and magnetic resonance spectroscopy are all performed with the same machine. Imaging after head fixation before surgery is repeated intraoperatively after resection of a tumor to evaluate the extent of resection and to visualize shifted major white matter tracts with intraoperative fiber tracking. A side-by-side display of the identical pre- and intraoperative images, which were measured at the identical slice positions, greatly facilitates image interpretation and reduces the intraoperative time needed for advanced registration algorithms. Nevertheless, nonlinear registration of preoperative data with intraoperative images is an important tool. When preoperative data cannot be obtained easily during surgery, the approach presented allows registering them reliably with intraoperative images. Christopher Nimsky Erlangen, Germany


A

rchip et al. describe a novel system for geometric distortion correction as applied to surgery for brain tumor resection using preoperative imaging (including MRI, fMRI, DTI, magnetic resonance spectroscopy, and positron emission tomography) fused to intraoperative MRI studies. The authors describe a case series of 11 patients undergoing resection of tumors in or near eloquent brain regions. The authors found a 10.3-mm mean magnitude and 21.5-mm maximum magnitude of geometric distortion measured on intraoperative images. By using their compensation algorithm, this was reduced to an average of 1.93 mm. This work is important as it highlights a potential source of bias in combining preoperative studies to interventional imaging and offers a novel solution to this problem. Whether this technology will increase safety, improve outcomes, or prove to be a reasonable surrogate for awake surgery must be determined with further studies. Andres M. Lozano Toronto, Canada

T

his is an excellent article by the group that basically invented intraoperative MRI (iMRI). They have used a clever, nonrigid registration technique to morph the geometrically distorted 0.5-T iMRI T1weighted spoiled gradient echo (SPGR) technique to a less-distorted SPGR image at 3-T. Using a phantom, they show that 3-T SPGR differs from computed tomography (the “gold standard”) by 1 mm, whereas 0.5-T iMRI differs by 5 mm. The reason for doing this is to be able to fuse 3-T preoperative DTI and fMRI to the intraoperative 0.5-T images. I have a relatively minor problem with this article. The DTI and fMRI images are both based on echo planar imaging techniques, which are much more susceptible to geometric distortion than SPGR, particularly at 3-T. This is particularly a problem near the cranial base where diamagnetic susceptibility effects (brain versus air) distort echo-planar–based images. Thus, whereas the SPGR anatomic images may be accurately morphed, neurosurgeons using iMRI should be aware that DTI and fMRI images may not be. William G. Bradley, Jr. San Diego, California


FUNCTIONAL AND STEREOTACTIC Technique Application

LOCATION OF ACTIVE CONTACTS IN PATIENTS WITH PRIMARY DYSTONIA TREATED WITH GLOBUS PALLIDUS DEEP BRAIN STIMULATION Clement Hamani, M.D., Ph.D. Division of Neurosurgery, Toronto Western Hospital, University of Toronto, University Health Network, Toronto, Canada

Elena Moro, M.D., Ph.D. Movement Disorders Center, Division of Neurology, Toronto Western Hospital, University of Toronto, University Health Network, Toronto, Canada

Cindy Zadikoff, M.D. Movement Disorders Center, Division of Neurology, Toronto Western Hospital, University of Toronto, University Health Network, Toronto, Canada

Yu-Yan Poon, R.N. Movement Disorders Center, Division of Neurology, Toronto Western Hospital, University of Toronto, University Health Network, Toronto, Canada

Andres M. Lozano, M.D., Ph.D. Division of Neurosurgery, Toronto Western Hospital, University of Toronto, University Health Network, Toronto, Canada Reprint requests: Andres M. Lozano, M.D., Ph.D., Division of Neurosurgery, Toronto Western Hospital, 399 Bathurst Street WW 4-447, Toronto, ON M5T2S8, Canada. Email: [email protected] Received, December 6, 2006. Accepted, June 19, 2007.

NEUROSURGERY

OBJECTIVE: Deep brain stimulation of the globus pallidus internus has been used for the treatment of various forms of dystonia, but the factors influencing postoperative outcomes remain unknown. We compared the location of the contacts being used for stimulation (active contacts) in patients with cervical dystonia, generalized dystonia, and Parkinson’s disease and correlated the results with clinical outcome. METHODS: Postoperative magnetic resonance scans of 13 patients with cervical dystonia, six patients with generalized dystonia, and five patients with Parkinson’s disease who underwent globus pallidus internus deep brain stimulation were analyzed. We assessed the location of the active contacts relative to the midcommisural point and in relation to the anteroposterior and mediolateral boundaries of the pallidum. Postoperative outcome was measured with the Toronto Western Spasmodic Torticollis Rating Scale (for cervical dystonia) and the Burke-Fahn-Marsden Dystonia Rating Scale (for generalized dystonia) during the last follow-up. RESULTS: We found that the location of the active contacts relative to the midcommisural point and the internal boundaries of the pallidum was similar across the groups. In our series, the contacts used for stimulation were clustered in the posterolateral region of the pallidum. Within that region, we found no correlation between the location of the contacts and postoperative outcome. CONCLUSION: The location of the active contacts used for globus pallidus internus deep brain stimulation was similar in patients with cervical dystonia, generalized dystonia, and Parkinson’s disease. KEY WORDS: Deep brain stimulation, Dystonia, Electrode, Globus pallidus, Movement disorders Neurosurgery 62[ONS Suppl 1]:ONS217–ONS225, 2008

D

eep brain stimulation (DBS) of the globus pallidus internus (GPi) has gained acceptance in recent years as an effective therapy for the treatment of dystonia (2–12, 15–20, 22, 23, 25). With accumulation of experience, some of the factors influencing surgical outcome are now recognized. It seems, for example, that patients with primary forms of dystonia with predominantly phasic movements respond better to the procedure (2, 3, 7, 10, 17, 22, 25). One aspect that has recently been explored, but not fully characterized, is the relationship between the location of the electrodes and clinical outcome. To date, only two studies have systematically addressed this issue (22, 24). They first examined patients with generalized

DOI: 10.1227/01.NEU.0000297026.98243.68

dystonia and concluded that the active contacts were in the posterolateroventral portion of the GPi (24). A second study concluded that location of the contacts relative to the borders of the GPi had a significantly smaller variance in patients with more than 70% of improvement after GPi DBS, compared with patients who improved by less than 50% (22). In this context, the accuracy of lead placement with respect to the internal boundaries of the pallidum did affect outcome. Other articles assessing the location of the electrodes in patients with dystonia have only provided coordinates relative to the midcommissural point but did not attempt to correlate these variables to postoperative outcome (2, 25). As a result, questions such as


HAMANI ET AL.

TABLE 1. Demographics and outcomes in patients with cervical dystonia treated with bilateral GPi stimulationa Patient

Age

1

62

Diagnosis Cervical

Preop TWSTR

Preop severity TWSTR

Postop TWSTR (%)

Postop severity TWSTR (%)

Follow-up (mo)

70.7

29

31.7 (55)

5 (83)

24

Active contact R/L 1–/5–

2

25

Cervical

36

19

4 (89)

4 (79)

18

1–/5–

3

59

Cervical

45.2

24

6 (87)

6 (75)

18

1–/5–

4

39

Cervical

70.2

26

48.5 (31)

14 (46)

42

2–/5–

5

67

Cervical

45.7

20

12 (74)

9 (55)

60

1–/6–

6

37

Cervical

60.7

22

40 (34)

14 (36)

48

1–/5–

7

51

Cervical

51.7

15

35.5 (31)

12 (20)

12

2–3–/5–6–

8

67

Cervical

61.7

26

31 (50)

19 (27)

12

2–/6–

9

33

Cervical

74.5

21

20 (73)

9 (57)

36

1–/5– 1–/5–

b

10

63

Segmental

55

17

23 (58)

14 (18)

6

11

62

Cervical

63

27

28.5 (55)

13 (52)

4

12

50

Cervical

22

17

6 (73)

4 (76)

67

13

55

Cervical

62

26

28.5 (54)

9 (65)

3

22.2. 4.5

24.2 14.0 (58.7 19.6)

26.9 19.6 (53.0 22.5)

26.9 21.6

Mean SD

51.5 13.9

55.3 15.0

1–/5– 2–3–/6–7– 1–/5–

Preop, preoperative; TWSTRS, Toronto Western Spasmodic Torticollis Rating Scale; Postop, postoperative; %, percentage of improvement calculated as (1 preop scores/ postop scores) 100; R/L, right and left electrodes. Results in bold represent mean standard deviation. b Patient with segmental dystonia with a predominant cervical component. a

whether different forms of dystonia respond to stimulation in different pallidal targets, or whether the location of the active contacts in patients with dystonia is similar to that previously used to treat patients with Parkinson’s disease (PD) remain unanswered. To further investigate these issues, we assessed the location of the electrodes in our series of patients with primary dystonia who were treated with bilateral GPi DBS. The coordinates of the contacts relative to standard anatomic landmarks, as well as their location within the boundaries of the pallidum, were calculated and compared among groups of patients with cervical dystonia, generalized dystonia, and PD.

PATIENTS AND METHODS Patients The following inclusion criteria were used in our study: 1) diagnosis of primary dystonia, 2) age older than 17 years, and 3) no previous intracranial surgical procedures. Nineteen of the 25 (76%) patients with primary dystonia treated with bilateral GPi DBS at the Toronto Western Hospital from October 1996 to August 2006 were included in our study. Five patients with primary generalized dystonia were excluded because they had previous pallidotomies. One patient with DYT1-generalized dystonia was excluded because she was 9 years old by the time of surgery. Of the 19 adult patients included, 12 had cervical dystonia, one had segmental dystonia with predominant neck involvement (these two groups were considered together as a primary cervical dystonia group), and six patients had primary generalized dystonia (two were positive and four were negative for the DYT1


gene). Patients with cervical and generalized dystonia, respectively, were assessed before and after surgery (last follow-up) with the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) and the Burke-Fahn-Marsden (BFM) Dystonia Rating Scale, respectively. For patients with generalized dystonia, we considered appendicular scores as the sum of the upper- and lower-extremity BFM scores for each hemibody. Demographics and the average pre- and postoperative TWSTRS and BFM scores are shown in Tables 1 and 2. In addition to patients with primary dystonia, a group of five patients with PD treated with unilateral pallidotomies and contralateral GPi DBS was also included in our analysis. We selected these patients on the basis of the availability of their magnetic resonance imaging (MRI) scans as electronic files in our system. The images of other patients with PD treated with bilateral GPi DBS were no longer accessible in electronic format and could not be sent to our imaging workstation. Postoperative improvement after unilateral GPi DBS in the five patients with PD included in our study was 25.5 21.1%, as assessed with the Unified Parkinson Disease Rating Scale.

Programming of Patients with Dystonia Programming of the stimulators was usually started 3 weeks after surgery. During the first session, each electrode contact was stimulated individually to establish the threshold for side effects (flashes of light with ventral contacts, and muscle contractions and dysarthria with other contacts). Stimulation was then started in a monopolar configuration with the lowest contacts serving as cathodes and the case as the anode. The voltage selected was 0.1 to 0.3 V lower than the threshold for side effects or 3.6 V or less. Two days to 2 weeks later (1 wk average), the patients returned for clinical reassessment. Improvement in dystonia was scored on the basis of their subjective impression as well as TWSTRS and BFM scores (recorded during each programming ses-


LOCATION OF DBS ELECTRODES IN DYSTONIA

TABLE 2. Demographics and outcomes in patients with generalized dystonia treated with bilateral GPi stimulationa Patient

Age

Diagnosis

Preop BFM

1

34

DYT1

71.5

35 (51%)

50

1–/5–

2

46

DYT1

65

25 (63%)

13

2–3–/6–7–

DYT1

68.3 4.6

Mean SD

40.0 8.5

Postop BFM (%)

30.0 7.1 (57.0 8.5%) 3.5 (91%)

Follow-up (mo)

Active contact R/L

31.5 26.2

3

26

DYT1–

40.5

12

2–/5–

4

58

DYT1–

33

5

53

DYT1–

65

16 (51%)

7

1–/5–

17.5 (73%)

36

6

54

DYT1–

51

15 (71%)

1–/5–

3

Mean SD

47.8 14.7

DYT1–

47.3 15.4

13.0 10.6 (71.5 15.2%)

14.5 18.6

2–/6–

Mean SD

45.2 12.6

All

54.3 15.4

18.7 10.6 (66.7 15.2%)

20.2 18.6

Preop, preoperative; BFM, Burke-Fahn-Marsden Dystonia Rating Scale; Postop, postoperative; %, percentage of improvement calculated as (1 preop scores/postop scores) 100; R/L, right and left electrodes. Results in bold represent mean standard deviation.

a

sion). In subsequent visits, homologous contacts (1 and 5; 2 and 6; 3 and 7) were tested in a similar way. After testing all contacts, the ones providing the most striking benefit on dystonia were selected for longterm stimulation. Additional adjustments included increases in voltage and/or pulse width and were made whenever clinical improvement was judged to be unsatisfactory. The most common settings for treating dystonia in our patients were monopolar stimulation at 2.5 to 3.5V, 60 to 90 µs, and 130 Hz.

Location of Active Contacts through MRI Preoperative axial stereotactic T1-weighted or three-dimensional inversion recovery images (in patients operated on before and after 2002, respectively) and postoperative axial T2-weighted images were transferred to a workstation (StealthStation; Medtronic SNT, Louisville, CO). Using the FrameLink 4.1 software (Mach 4.1, StealthStation; Medtronic SNT) these two studies were fused, the fiducials of the frame were recognized, and the mean rod marking error was calculated and registered. Coronal and sagittal planes were reconstructed on the basis of the axial images. The anterior (AC) and posterior commissures (PC) were then targeted in the axial plane of the postoperative scan, and three additional points were plotted in the midline. Thereafter, the images were reformatted parallel to the AC-PC plane and orthogonal to the midline. Pitch, roll, and yaw were corrected in the StealthStation. DBS electrodes were visualized in all three planes, and their tips were targeted. For this study, we defined the tip of the electrode as the most distal portion of the electrode artifact. Although this does not represent the actual tip of the electrode, it was a standard measurement that could be compared across the studied groups. After establishing the location of the tip of the electrodes, the trajectory angle was calculated and the distance between the MRI artifacts of each of the electrode contacts was assessed. For our study, we considered the center of the sphere-shaped artifacts correspondent to the electrode contacts as the center of the contacts (21, 22). The coordinates of the active contacts were then derived relative to the midcommissural point and lateral border of the third ventricle (x, mediolateral plane; y, anteroposterior plane; z, dorsoventral plane). The active contact was defined as the cathode in patients treated with mono- or bipolar stimulation. In patients using double monopolar stimulation, the

NEUROSURGERY

location of the active contact was defined as the midpoint between the two cathodes. The Euclidean distance from either the active contact or the tip of the electrodes to the center of the optic tract in the region closest to the projection of the electrode’s tip was calculated. In a three-dimensional xyz system with p1 being the optic tract (x1, y1, z1) and p2 being either the active contact or the tip of the electrode (x2, y2, z2), the Euclidean distance between p1 and p2 was represented by

兹[(x1 – x2)2 + (y1 – y2)2 + (z1 – z2)2]

(1)

To evaluate the location of the active contacts relative to the pallidal boundaries, we used axial T2-weighted images containing the plane of the center of the active contacts. A straight line was drawn between the anteromedial and posterolateral corners of the GPi (Fig. 1). This was considered to be the anteroposterior (AP) extent of the pallidum in our study (APtotal). A second line was drawn parallel to APtotal from the projection of the anteromedial corner of the pallidum to the center of the active contact (APct) (Fig. 1). The percentage obtained when APct/APtotal 100 was calculated denoted the relative distance of the center of the active contact from the anteromedial corner of the pallidum (%AP). A line perpendicular to APtotal that crossed the center of the active contact was then drawn and used to measure the mediolateral (ML) width of the pallidum (MLtotal) (Fig. 1). The distance from the most lateral point of this line to the center of the active contact was defined as the mediolateral location of the contact (MLct) (Fig. 1). The percentage obtained when MLct/MLtotal 100 was calculated denoted the relative distance of the center of the active contact from the lateral aspect of the pallidum (%ML) (Fig. 1). Note that most of the aforementioned variables assessed were obtained exclusively from postoperative images (x, y, z of the active contacts, Euclidean distances, and the location of the contacts relative to the internal anatomy of the pallidum). Fusion of pre- and postoperative scans was used only to calculate the arc and ring angles of the electrodes relative to the frame. Surgery was always performed in a similar way with burr holes placed approximately 1 to 2 cm anterior to the coronal suture and 2 to 2.5 cm lateral to the midline. In patients with cervical and generalized dystonia, arc angles were 71.7 6.8


HAMANI ET AL.

DYT1- and non-DYT1-generalized dystonia and 2) to compare the clinical outcome assessed in two independent groups. Correlation analysis was used to assess whether or not there was a relationship between the location of the active contacts and outcome. For analysis of variance and Student’s t test, a P value less than 0.05 was considered statistically significant. A correlation was considered to be strong when r 0.75, moderate when r 0.5, weak when r 0.25, and nonexistent when r 0.25. All values in the text are expressed as mean standard deviation.

B A

RESULTS Location of Active Contacts in Different Clinical Conditions FIGURE 1. Location of the active contacts relative to internal pallidal boundaries. A, schematic representation of a section of the GPi on a plane that crossed the center of the active contact (gray circle). A straight line was drawn between the anteromedial and posterolateral corners of the GPi (APtotal). A parallel line was drawn from the projection of the anteromedial corner of the pallidum to the center of the active contact (APct). Note that the actual lines representing MLtotal and MLct would superpose in a real case. These lines were separated in this figure for the sake of clarity. B, schematic representation of the location of the active contacts relative to the boundaries of the pallidum in patients with cervical dystonia, generalized dystonia, and Parkinson’s disease (PD). In this figure, the dorsal ventral location of the contacts was not taken into consideration and all the plots were done in a schematic section corresponding to the level of the intercommissural plane. degrees and 76.7 9.1 degrees, respectively. Ring angles were 89.0 5.1 degrees and 89.2 6.1 degrees, respectively.

Statistical Analysis We used analysis of variance with a Tukey post-hoc analysis to compare the following: 1) the location of the active contacts and tip of the electrodes relative to midcommissural point (MCP), 2) the location of the active contacts relative to the third ventricular wall, 3) the Euclidean distances between the optic tract and either the active contacts or the tip of the electrodes, and 4) the percentage of AP and ML in groups of patients with cervical dystonia, generalized dystonia, and PD. We used the Student’s t test to: 1) compare these same variables in patients with

A summary of our results can be found in Table 3 and Figure 2. Because the location of the active contacts was similar in the right and left hemispheres, data from both sides were grouped for the analysis. We found no significant differences in the location of the active contacts in patients with cervical dystonia, generalized dystonia, and PD (Table 3). Location of the active contacts was also similar in DYT1 and non-DYT1-generalized dystonia (Table 4). Although DYT1+ patients had their contacts placed closer to the midline (P 0.04), this difference was not significant when the width of the third ventricle was taken into account. When the location of the active contacts relative to the internal anatomy of the pallidum was considered, we also found no significant differences across groups (Fig. 1). Active contacts in patients with cervical dystonia, generalized dystonia, and PD were located within 49.0 to 75.6% (average, 59.3 7.0%), 40.0 to 66.0% (average, 54.8 8.8%), and 44.0 to 69.0% (average, 58.0 9.1%) of the APtotal line (P 0.3; 0% being the anteromedial corner and 100% the posterolateral corner of the pallidum), respectively (Fig. 1). In the mediolateral plane, the active contacts in patients with cervical dystonia, generalized dystonia, and PD were located within 54% to 100% (average, 76.0 11.7%), 64% to 98% (average, 84.1 10.7%), and 47% to 89% (average, 80.4 18.7%) of the MLtotal line (P 0.2; 0% being the lateral edge and 100% the medial edge of the pallidum over the line), respectively (Fig. 1). We did not explore

TABLE 3. Location of active contacts relative to the midcommissural point and third ventricular wall in patients with primary dystonia and Parkinson’s disease treated with bilateral globus pallidus internus deep brain stimulation n

x

x (ventricle)

y

z

x (tip)

y (tip)

z (tip)

Cervical dystonia

26

20.3 1.6

18.3 1.9

3.0 1.4

–1.0 2.0

20.2 2.0

0.8 1.6

–6.4 1.9

5.8 2.3

7.3 2.4

Generalized dystonia

12

19.5 1.4

18.1 1.3

3.4 1.1

0 2.2

18.9 1.5

1.1 1.2

–6.0 1.5

6.1 2.2

8.4 2.1

5

21.3 2.2

18.5 2.1

3.8 0.6

–1.9 1.7

20.6 2.1

1.2 2.4

–6.6 2.5

5.3 3.9

8.2 2.6

0.1

0.9

0.4

0.2

0.1

0.8

0.4

0.8

0.4

Parkinson’s disease

P values

ED (tip-ot) ED (ct-ot)

a

n, number of electrodes implanted per group; x, mediolateral coordinate of the active contacts relative to MCP; x (ventricle), mediolateral coordinate of the active contacts relative to the lateral wall of the third ventricle; y, anteroposterior coordinate of the active contacts relative to MCP; z, dorsoventral coordinate of the active contacts relative to MCP; x, y, z (tip), coordinates of the tip of the electrodes relative to MCP; ED, Euclidean distance; tip-ot, from the tip of the electrodes to the optic tract; ct-ot, from the active contacts to the optic tract. Values represent mean standard deviation. P values were obtained with analysis of variance comparing the groups of patients with cervical dystonia, generalized dystonia (as a whole), and Parkinson’s disease.




the dorsoventral location of the active contacts relative to the internal pallidal anatomy because the border between the GPi and the globus pallidus externus could not be clearly delineated in our postoperative MRIs.

A

Correlation between the Location of the Active Contacts and Outcome In patients with cervical dystonia, correlation analysis did not reveal a relationship between the location of the active contacts and outcome, as assessed with the TWSTRS scores (Table 5). In addition, the location of the active contacts was similar in patients with cervical dystonia that had more or less than 50% of improvement in TWSTRS severity scores (Table 6). As for cervical dystonia, correlation analysis in patients with generalized dystonia has also revealed no relationship between the location of the active contacts and outcome (Table 5). Because all patients with generalized dystonia had a satisfactory outcome ( 50% of benefit), we did not subdivide them in groups according to clinical response.

B

C

Relationship between the Contact Being Used and Outcome

FIGURE 2. Location of the active contacts relative to the midcommissural point in patients with generalized dystonia, cervical dystonia, and Parkinson’s disease. A, schematic representation of a coronal view with the mediolateral (x) and dorsal-ventral (z) coordinates of the active contacts. B, schematic representation of an axial view with the mediolateral (x) and anteroposterior (y) coordinates of the active contacts. C, schematic representation of a sagittal view with the anteroposterior (y) and dorsoventral (z) coordinates of the active contacts. Graphs on the right side of the figure represent the correspondent means and standard deviations (bars). Numbers on the graphs represent mm from midcommissural point (MCP). Negative values are inferior and posterior to the MCP.

Most patients in our study were using contact 1 as the cathode (Tables 1 and 2). In cervical dystonia, upper contacts were only used when side effects were noticed with the lower ones. As a result, one could speculate that a better outcome might have been expected in patients using contact 1. In fact, TWSTRS scores in this group of patients improved by 63.1 18.6%, whereas improvement in patients using upper contacts was in the order of 51.8 21.3% (P 0.4). In contrast to these findings, although some patients with generalized dystonia were using upper contacts due to side effects with Contact 1, others had better clinical outcome when receiving stimulation through Contacts 2 and 3. Overall, BFM scores were reduced by 62.7 11.1% in patients using Contact 1 and by 75 14.4% in patients using upper contacts (P 0.3).

DISCUSSION We found that the location of the active contacts for DBS within the GPi was similar in patients with cervical dystonia,

TABLE 4. Location of active contacts relative to the MCP and third ventricular wall in patients with DYT1+ and DYT1- primary generalized dystonia treated with bilateral globus pallidus internus deep brain stimulationa n

x

x (ventricle)

y

z

ED

DYT1

4

18.3 1.1

17.4 0.9

3.0 1.1

–1.0 2.8

7.3 2.3

DYT1–

8

20.1 1.1

18.4 1.4

3.6 1.1

0.5 1.8

9.0 1.9

0.2

0.4

0.4

0.3

P values

0.04

b

a n, number of electrodes implanted per group; x, mediolateral coordinate of the active contacts relative to MCP; x (ventricle), mediolateral coordinate of the active contacts relative to the lateral wall of the third ventricle; y, anteroposterior coordinate of the active contacts relative to MCP; z, dorsoventral coordinate of the active contacts relative to MCP; ED, Euclidean distance. Values represent mean SD. b Statistically significant.

NEUROSURGERY


HAMANI ET AL.

TABLE 5. Correlation analysis of outcome and the location of active contacts relative to the MCP, third ventricular wall, and internal pallidal boundaries in patients with primary dystonia treated with bilateral globus pallidus internus deep brain stimulationa TWSTRS scores

x

x (ventricle)

y

z

ED

%AP

%ML

–0.01

–0.17

0.21

–0.15

–0.13

–0.23

–0.24

BFM scores

0.01

0.21

0.08

0.08

0.2

BFM appendicular scores

0.2

–0.07

0.3

0.03

0.07

0.24 –0.1

0.24 –0.1

Values represent the Pearson correlation coefficient (r). BFM appendicular scores represent the sum of upper and lower extremity scores for the hemibody contralateral to stimulation. a x, mediolateral coordinate of the active contacts relative to MCP; x (ventricle), mediolateral coordinate of the active contacts relative to the lateral wall of the third ventricle; y, anteroposterior coordinate of the active contacts relative to MCP; z, dorsoventral coordinate of the active contacts relative to MCP; ED, Euclidean distance; %AP, location of the active contact relative to the anteromedial and posterolateral corners of the pallidum as seen in postoperative MRI scans (see text for details); %ML, location of the active contact relative to the medial and lateral edges of the pallidum as seen in postoperative MRI scans (see text for details). TWSTRS, Toronto Western Spasmodic Torticollis Rating Scale; BFM, Burke-Fahn-Marsden Dystonia Rating Scale.

TABLE 6. Location of active contacts relative to the midcommissural point and third ventricular wall in patients with cervical dystonia treated with bilateral globus pallidus internus deep brain stimulationa n

x

x (ventricle)

y

z

ED

%AP

%ML

50%

16

20.5 1.6

18.3 1.9

3.0 1.5

–0.7 1.6

7.4 2.7

58.8 7.2

74.8 11.6

50%

10

19.9 1.7

18.1 2.0

2.7 0.9

–1.2 2.6

7.3 2.2

60.3 7.1

77.5 12.2

0.4

0.8

0.6

0.6

0.9

0.6

0.6

P values a

n, number of electrodes implanted per group; x, mediolateral coordinate of the active contacts relative to MCP; x (ventricle), mediolateral coordinate of the active contacts relative to the lateral wall of the third ventricle; y, anteroposterior coordinate of the active contacts relative to MCP; z, dorsoventral coordinate of the active contacts relative to MCP; ED, Euclidean distance; %AP, location of the active contact relative to the anteromedial and posterolateral corners of the pallidum as seen in postoperative MRI scans (see text for details); %ML, location of the active contact relative to the medial and lateral edges of the pallidum as seen in postoperative MRI scans (see text for details). Values represent mean standard deviation.

generalized dystonia, and PD. In addition, we found no correlation between the location of the active contacts and postoperative outcome in these patients. The initial reason that led us to investigate the location of the active contacts in dystonia was our clinical impression that patients with cervical dystonia using upper contacts were doing slightly poorer than those using Contact 1. In this later group, improvement in TWSTRS scores was approximately 20% higher (although not statistically significant) than in patients using upper contacts. In contrast, BFM scores in patients with generalized dystonia using Contact 1 were approximately 15% lower (although also not statistically significant) than in patients being stimulated through upper contacts. Considering this, we initially thought that the pallidal region more suitable for stimulation in cervical dystonia would be ventral to the one for generalized dystonia. Our result, however, did not confirm this hypothesis. A second aspect of our study was to evaluate whether or not part of the variability in postoperative results in patients with dystonia could be related to differences in the location of the active contacts within the pallidum. We found no correlation between the pallidal region being stimulated and outcome. This suggests that other factors, such as the clinical characteristics of the dystonic movements (tonic versus phasic), for example, are probably playing a more influential role in postoperative clinical results. However, we have only explored a


small region of the pallidum. Our active contacts were relatively clustered in the posteroventral portion of the GPi, and we could not test the effects of stimulation in significantly more anterior or lateral regions of the nucleus. In this sense, the only thing we might conclude is that within that region of the pallidum, location of the active contacts did not determine the outcome. In contrast to our findings, Gross et al. (14), for example, have shown that differences in the location of lesions in PD patients treated with pallidotomy correlated with postoperative improvements in specific motor symptoms. In that series, variability in the location of lesions in the anteroposterior and mediolateral planes was much greater than in our study (location of the center of the lesions varied from 8.3 mm anterior to 0.5 mm posterior to MCP and 10.8–19.5 mm lateral to the third ventricular wall) (13). As the target for the placement of the electrodes in dystonia is the posteroventral aspect of the GPi, the location of the electrodes in our series was similar to that reported in the literature (2, 24, 25). Yet, when our results are compared with those published by individual reports, small differences could be observed. For instance, compared with our data, the active contacts in the study by Starr et al. (22) were slightly more anterior relative to the MCP and a little more lateral when the borders of the pallidum were considered. When evaluating the results by Vayssiere et al. (24), however, one can appreciate the variability in the location of active contacts leading to



an excellent clinical outcome, particularly in the mediolateral plane. Considering these results and the variable clinical presentation of the patients, the meaning of small discrepancies in the location of the electrodes observed by different groups is difficult to interpret. Most patients with dystonia in our study were receiving stimulation at 2.5 to 3.5V, 60 to 90 µs, and 130 Hz. Because of the proximity of the active contacts to the internal capsular border, motor contractions were often observed at 2.8 to 4V. Despite this, we continue to place the electrodes in the same way because our patients are having a significant postoperative improvement. In summary, our results showed that the location of the active contacts was similar in patients with generalized and cervical dystonia. Future studies with a greater number of patients are certainly needed to confirm these findings and investigate the relationship between the clinical outcome and best location for electrical stimulation within the pallidum in patients with dystonia.

14.

15.

16. 17. 18.

19.

20.

REFERENCES 1. Andrade-Souza YM, Schwalb JM, Hamani C, Eltahawy H, Hoque T, SaintCyr J, Lozano AM: Comparison of three methods of targeting the subthalamic nucleus for chronic stimulation in Parkinson’s disease. Neurosurgery 56 [Suppl]:360–368, 2005. 2. Bereznai B, Steude U, Seelos K, Bötzel K: Chronic high-frequency globus pallidus internus stimulation in different types of dystonia: A clinical, video, and MRI report of six patients presenting with segmental, cervical, and generalized dystonia. Mov Disord 17:138–144, 2002. 3. Bittar RG, Yianni J, Wang S, Liu X, Nandi D, Joint C, Scott R, Bain PG, Gregory R, Stein J, Aziz TZ: Deep brain stimulation for generalised dystonia and spasmodic torticollis. J Clin Neurosci 12:12–16, 2005. 4. Castelnau P, Cif L, Valente EM, Vayssiere N, Hemm S, Gannau A, Digiorgio A, Coubes P: Pallidal stimulation improves pantothenate kinase-associated neurodegeneration. Ann Neurol 57:738–741, 2005. 5. Chou KL, Hurtig HI, Jaggi JL, Baltuch GH: Bilateral subthalamic nucleus deep brain stimulation in a patient with cervical dystonia and essential tremor. Mov Disord 20:377–380, 2005. 6. Cif L, Valente EM, Hemm S, Coubes C, Vayssiere N, Serrat S, Di Giorgio A, Coubes P: Deep brain stimulation in myoclonus-dystonia syndrome. Mov Disord 19:724–727, 2004. 7. Coubes P, Cif L, El Fertit H, Hemm S, Vayssiere N, Serrat S, Picot MC, Tuffery S, Claustres M, Echenne B, Frerebeau P: Electrical stimulation of the globus pallidus internus in patients with primary generalized dystonia: Long-term results. J Neurosurg 101:189–194, 2004. 8. Diamond A, Shahed J, Azher S, Dat-Vuong K, Jankovic J: Globus pallidus deep brain stimulation in dystonia. Mov Disord 21:692–695, 2006. 9. Eltahawy HA, Feinstein A, Khan F, Saint-Cyr J, Lang AE, Lozano AM: Bilateral globus pallidus internus deep brain stimulation in tardive dyskinesia: A case report. Mov Disord 19:969–972, 2004. 10. Eltahawy HA, Saint-Cyr J, Giladi N, Lang AE, Lozano AM: Primary dystonia is more responsive than secondary dystonia to pallidal interventions: Outcome after pallidotomy or pallidal deep brain stimulation. Neurosurgery 54:613–621, 2004. 11. Eltahawy HA, Saint-Cyr J, Poon YY, Moro E, Lang AE, Lozano AM: Pallidal deep brain stimulation in cervical dystonia: Clinical outcome in four cases. Can J Neurol Sci 31:328–332, 2004. 12. Franzini A, Marras C, Ferroli P, Zorzi G, Bugiani O, Romito L, Broggi G: Long-term high-frequency bilateral pallidal stimulation for neurolepticinduced tardive dystonia. Report of two cases. J Neurosurg 102:721–725, 2005. 13. Gross RE, Lombardi WJ, Hutchison WD, Narula S, Saint-Cyr JA, Dostrovsky JO, Tasker RR, Lang AE, Lozano AM: Variability in lesion location after micro-

NEUROSURGERY

21.

22.

23.

24.

25.

electrode-guided pallidotomy for Parkinson’s disease: Anatomical, physiological, and technical factors that determine lesion distribution. J Neurosurg 90:468–477, 1999. Gross RE, Lombardi WJ, Lang AE, Duff J, Hutchison WD, Saint-Cyr JA, Tasker RR, Lozano AM: Relationship of lesion location to clinical outcome following microelectrode-guided pallidotomy for Parkinson’s disease. Brain 122:405–416, 1999. Hälbig TD, Gruber D, Kopp UA, Schneider GH, Trottenberg T, Kupsch A: Pallidal stimulation in dystonia: Effects on cognition, mood, and quality of life. J Neurol Neurosurg Psychiatry 76:1713–1716, 2005. Houser M, Waltz T: Meige syndrome and pallidal deep brain stimulation. Mov Disord 20:1203–1205, 2005. Krause M, Fogel W, Kloss M, Rasche D, Volkmann J, Tronnier V: Pallidal stimulation for dystonia. Neurosurgery 55:1361–1370, 2004. Kupsch A, Benecke R, Müller J, Trottenberg T, Schneider GH, Poewe W, Eisner W, Wolters A, Müller JU, Deuschl G, Pinsker MO, Skogseid IM, Roeste GK, Vollmer-Haase J, Brentrup A, Krause M, Tronnier V, Schnitzler A, Voges J, Nikkhah G, Vesper J, Naumann M, Volkmann J; Deep-Brain Stimulation for Dystonia Study Group: Pallidal deep-brain stimulation in primary generalized or segmental dystonia. N Engl J Med 355:1978–1990, 2006. Magariños-Ascone CM, Regidor I, Martinez-Castrillo JC, Gómez-Galán M, Figueiras-Mendez R: Pallidal stimulation relieves myoclonus-dystonia syndrome. J Neurol Neurosurg Psychiatry 76:989–991, 2005. Pillon B, Ardouin C, Dujardin K, Vittini P, Pelissolo A, Cottencin O, Vercueil L, Houeto JL, Krystkowiak P, Agid Y, Destée A, Pollak P, Vidailhet M; French SPIDY Study Group: Preservation of cognitive function in dystonia treated by pallidal stimulation. Neurology 66:1556–1558, 2006. Pollo C, Villemure JG, Vingerhoets F, Ghika J, Maeder P, Meuli R: Magnetic resonance artifact induced by the electrode Activa 3389: An in vitro and in vivo study. Acta Neurochir (Wien) 146:161–164, 2004. Starr PA, Turner RS, Rau G, Lindsey N, Heath S, Volz M, Ostrem JL, Marks WJ: Microelectrode-guided implantation of deep brain stimulators into the globus pallidus internus for dystonia: Techniques, electrode locations, and outcomes. J Neurosurg 104:488–501, 2006. Trottenberg T, Volkmann J, Deuschl G, Kühn AA, Schneider GH, Müller J, Alesch F, Kupsch A: Treatment of severe tardive dystonia with pallidal deep brain stimulation. Neurology 64:344–346, 2005. Vayssiere N, van der Gaag N, Cif L, Hemm S, Verdier R, Frerebeau P, Coubes P: Deep brain stimulation for dystonia confirming a somatotopic organization in the globus pallidus internus. J Neurosurg 101:181–188, 2004. Vidailhet M, Vercueil L, Houeto JL, Krystkowiak P, Benabid AL, Cornu P, Lagrange C, Tézenas du Montcel S, Dormont D, Grand S, Blond S, Detante O, Pillon B, Ardouin C, Agid Y, Destee A, Pollak P; French Stimulation du Pallidum Interne dans la Dysonie (SPIDY) Study Group: Bilateral deep-brain stimulation of the globus pallidus in primary generalized dystonia. N Engl J Med 352:459–467, 2005.

COMMENTS

I

n this study of globus pallidus deep brain stimulation (DBS) for primary dystonia, the authors have carefully measured active contact locations using postoperative MRI scans. The active contact locations reported were associated with major improvements in most patients. They show that there was not a systematic difference in optimal active contact locations for cervical versus primary dystonia. They also show that, within the small region of the posterior globus pallidus internus (GPi) covered by their electrodes, lead location did not predict outcome. The study helps to define the region of the GPi that should be targeted to achieve significant clinical benefit in primary dystonia. The active contacts in this study are slightly closer to the pallidocapsular border than those in our own series and were programmed at lower pulse widths but were associated with similar benefits. In some patients with cervical dystonia, we have observed a side effect of pallidal stimulation: a mild bradykinesia in body parts (legs and arms) that are not affected by the dystonia. This has been very


HAMANI ET AL.

annoying to some patients. Such a side effect has not been noticed in generalized dystonia, perhaps because the improvement in dystonia masks any subtle side effects of stimulation on movement times. It is possible that only some subregions of the posterior GPi are associated with this side effect, but this theory remains to be proven. Philip A. Starr San Francisco, California

T

he GPi is the favored target for DBS in the treatment of patients with dystonia (2, 4, 5, 11, 12). There is promising evidence for a striking efficacy of pallidal stimulation in dystonia. The most dramatic results have been obtained with pallidal stimulation to treat primary dystonia linked to genetic mutations, in particular, DYT1, other primary dystonias, and tardive dystonic syndromes. Conflicting results were reported for secondary dystonias. Long-term double-blind studies and a careful assessment of the efficacy are needed. But, first, there is a requirement for a consensus about the precise anatomic position of the optimal lead. Analyses of the intrapallidal localization of the quadripolar contacts associated with optimal benefit have been published infrequently. In fact, there has been a deficiency in much of the DSB literature in this regard. More recent publications have included computationally reformatted postoperative magnetic resonance images and the approximate locations of the contacts being used. These data should be essential in any future publications in the field so as to further delineate the optimal target. In this regard, Hamani et al. add to the literature. In this article, the target is the sensorimotor area of GPi in all cases regardless of indication. This is a large area and the real question is whether there is an optimal target in that area. They did not find a “sweet spot,” but the number of subgroups precluded meaningful statistical power. Although patients with Parkinson’s disease in whom pallidotomy failed but who showed minimal improvement with DBS were included, the comparison of the patients with generalized dystonia and those with cervical dystonia were the focus of the article. Here, comparisons with the literature may shed some light. In a study of 23 adult and pediatric patients with mixed forms of dystonia, Starr et al. (7) implanted DBS leads using MRI-based stereotactic, microelectrode recording, and intraoperative test stimulation to determine thresholds for stimulation-induced adverse effects. Lead locations were measured on computationally reformatted postoperative MRI scans. The active lead locations associated with robust improvement (>50% decrease in the Burke-Fahn-Marsden Dystonia Rating Scale score) were located near the intercommissural plane, at a mean distance of 3.7 mm from the pallidocapsular border. This result is consistent with our finding that the target is more lateral (physiological 21.5 mm rather than 20 mm for pallidotomy) but allows for greater current spread while avoiding capsular side effects (8). Hamani et al. place their lead closer to the capsule, and this placement might explain some of the differences in the two studies. Starr et al. (7) also observed that the spontaneous discharge rates of GPi neurons in dystonia are similar to those of globus pallidus externus (GPe) neurons and the two nuclei must be distinguished by neuronal discharge patterns rather than by rates, verifying a previous report (15). The early models of dystonia suggested that reduced mean discharge rates in GPi lead to disinhibition and increases in activity in the thalamocortical circuit that precipitated the development of dystonia. The clinical improvement in dystonia after pallidotomy was difficult to reconcile with this “rate” hypothesis as the procedure results in a further reduction in pallidal output. This quickly led to the development of an alternative model that, in addition to rate, incorporates changes in pattern and degree of synchronization of neuronal activity (13, 15). Multiple groups have confirmed that the mean GPi neuronal


firing rate in dystonic patients is lower than that in Parkinson’s disease, whereas there is little to no difference in firing rate observed in GPe between the two disorders. This finding has recently been confirmed in cervical dystonia compared with Parkinson’s disease in Toronto (9). In the other referenced manuscript, Vayssiere et al. (11) found that all activated contacts were in the posterolateroventral portion of the GPi. A correlation between the active contact location and the part of the body in which the highest improvement was observed demonstrated that a location more anterior for the inferior limb and one more posterior for the superior limb were delineated for the right side, but not for the left side. Why there are left to right side differences is not clear, but probably this is a sampling error. An optimal cervical stimulation location was not defined. These results may be explained by the somatotopic anatomy of GPi, which is well established (1). There is preservation of GPi somatotopic organization in dystonia (9, 14, 15). When patients with generalized dystonia were compared with patients with segmental craniocervical dystonia, there was no difference in the volumes or separations of leg and arm related territories (14). Hamani et al. similarly did not find a specific area wherein cervical dystonia was maximally improved. This most likely relates to the somatotopic anatomy in which cervical kinestatic cells are diffusely and infrequently found along the posteroventral GPi. Two other studies continue this theme. In a consecutive series of 15 patients with mixed forms of dystonia, the analysis by Tisch et al. (10) of the optimal stimulating contact coordinates with clinical improvement in the Burke-Fahn-Marsden Dystonia Rating Scale score identified two subgroups, distributed along an anterodorsal to posteroventral axis (10). Clinical improvement was greater for posteroventral than for anterodorsal stimulation for the arm and trunk and inversely correlated with the y coordinate. For the leg, posteroventral and anterodorsal stimulation were of equivalent efficacy. Overall clinical improvement was maximal with posteroventral stimulation and inversely correlated with the y (anterior-posterior) coordinate. Similarly, Houeto et al. (3), using a three-dimensional atlas magnetic resonance imaging coregistration method, demonstrated that acute ventral stimulation of the globus pallidus significantly improved the Burke-Fahn-Marsden Dystonia Rating Scale score when the stimulating contact was located in the GPi or medullary lamina in 18 of 21 patients (3). Bilateral acute dorsal pallidal stimulation, primarily localized within the GPe, had variable effects across patients, with half demonstrating slight or no improvement or even aggravation of dystonia compared with baseline. Stimulation of GPe can induce dyskinesias (6). So you do not want to locate the contact too laterally. Undeniably, additional research in this arena is warranted. A larger number of patients are essential but, without more accurate diagnosis and subcategory partition, the patients can not be properly grouped. Multiple problems with inaccuracies in postoperative magnetic resonance localization still remain to be resolved. The irregularities of the nucleus and dissimilar programming philosophies add subtitle problems not yet addressed. So, although the target is the sensorimotor area of GPi, whether the optimal target within that area varies with the explicit type and anatomical localization of the dystonia remains to be resolved. Julie G. Pilitsis Roy A.E. Bakay Chicago, Illinois

1. Chang EF, Turner RS, Ostrem JL, Davis VR, Starr PA: Neuronal responses to passive movement in the globus pallidus internus in primary dystonia. J Neurophysiol 17:S0022–S3077, 2007.



2. Coubes P, Cif L, El Fertit H, Hemm S, Vayssiere N, Serrat S, Picot MC, Tuffery S, Claustres M, Echenne B, Frerebeau P: Electrical stimulation of the globus pallidus internus in patients with primary generalized dystonia: long-term results. J Neurosurg 101:189–194, 2004. 3. Houeto JL, Yelnik J, Bardinet E, Vercueil L, Krystkowiak P, Mesnage V, Lagrange C, Dormont D, Le Bas JF, Pruvo JP, Tezenas du Moncel S, Pollak P, Agid Y, Destee A, Vidailhet M: Acute deep-brain stimulation of the internal and external globus pallidus in primary dystonia: Functional mapping of the pallidum. Arch Neurol 64:1281-1286, 2007. 4. Hung SW, Hamani C, Lozano AM, Poon YY, Piboolnurak P, Miyasaki JM, Lang AE, Dostrovsky JO, Hutchison WD, Moro E: Long-term outcome of bilateral pallidal deep brain stimulation for primary cervical dystonia. Neurology 68:457–459, 2007. 5. Kupsch A, Benecke R, Muller J, Trottenberg T, Schneider GH, Poewe W, Eisner W, Wolters A, Muller JU, Deuschl G, Pinsker MO, Skogseid IM, Roeste GK, Vollmer-Haase J, Brentrup A, Krause M, Tronnier V, Schnitzler A, Voges J, Nikkhah G, Vesper J, Naumann M, Volkmann J: Pallidal deep-brain stimulation in primary generalized or segmental dystonia. N Engl J Med 355:1978–1990, 2006. 6. Mouton S, Xie-Brustolin J, Mertens P, Polo G, Damier P, Broussolle E, Thobois S: Chorea induced by globus pallidus externus stimulation in a dystonic patient. Mov Disord 21:1771-1773, 2006. 7. Starr PA, Turner RS, Rau G, Lindsey N, Heath S, Volz M, Ostrem JL, Marks WJ: Microelectrode-guided implantation of deep brain stimulators into the globus pallidus internus for dystonia: Techniques, electrode locations, and outcomes. Neurosurg Focus 17:E4, 2004. 8. Starr PA, Vitek JL, Bakay RA: Deep brain stimulation for movement disorders. Neurosurg Clin N Am 9:381–402, 1998. 9. Tang JK, Moro E, Mahant N, Hutchison WD, Lang AE, Lozano AM, Dostrovsky JO: Neuronal firing rates and patterns in the globus pallidus internus of patients with cervical dystonia differ from those with Parkinson’s disease. J Neurophysiol 98:720–729, 2007. 10. Tisch S, Zrinzo L, Limousin P, Bhatia KP, Quinn N, Ashkan K, Hariz M: Effect of electrode contact location on clinical efficacy of pallidal deep brain stimulation in primary generalised dystonia. J Neurol Neurosurg Psychiatry 78:1314–1319, 2007.

11. Vayssiere N, van der Gaag N, Cif L, Hemm S, Verdier R, Frerebeau P, Coubes P: Deep brain stimulation for dystonia confirming a somatotopic organization in the globus pallidus internus. J Neurosurg 101:181–188, 2004. 12. Vidailhet M, Vercueil L, Houeto JL, Krystkowiak P, Lagrange C, Yelnik J, Bardinet E, Benabid AL, Navarro S, Dormont D, Grand S, Blond S, Ardouin C, Pillon B, Dujardin K, Hahn-Barma V, Agid Y, Destee A, Pollak P: Bilateral, pallidal, deep-brain stimulation in primary generalised dystonia: A prospective 3 year follow-up study. Lancet Neurol 6:223–229, 2007. 13. Vitek JL: Pathophysiology of dystonia: A neuronal model. Mov Disord 17 [Suppl 3]:S49–S62, 2002. 14. Vitek JL, Bakay RA, Hashimoto T, Kaneoke Y, Mewes K, Zhang JY, Rye D, Starr P, Baron M, Turner R, De Long MR: Microelectrode-guided pallidotomy: technical approach and its application in medically intractable Parkinson’s disease. J Neurosurg 88:1027–1043, 1998. 15. Vitek JL, Chockkan V, Zhang JY, Kaneoke Y, Evatt M, De Long MR, Triche S, Mewes K, Hashimoto T, Bakay RA: Neuronal activity in the basal ganglia in patients with generalized dystonia and hemiballismus. Ann Neurol 46:22–35, 1999.

T

his carefully performed study shows no difference in electrode position within the globus pallidus in patients with good outcomes after DBS performed for the treatment of cervical dystonia, generalized dystonia, and Parkinson’s disease. The authors are properly cautious about generalizing from their data, pointing out the need for further studies with larger numbers of patients, even though the movement disorders group at the University of Toronto has one of the largest experiences with DBS for motor disease. A further conclusion can, perhaps, be drawn from their data. The spatial distribution of effective DBS points extended within the globus pallidus 8 mm in the dorsal-ventral (z) axis and 6 mm in the anteriorposterior (y) axis. It would seem that the critical factor for efficacy of DBS is the total volume of the pallidum that is stimulated, a volume that can be reached from a number of points within the pallidum. Robert G. Grossman Houston, Texas

Sepulchretum Sive Anatomia Practica, (1679). From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


FUNCTIONAL AND STEREOTACTIC Clinical Study

PERCUTANEOUS COMPUTED TOMOGRAPHY-GUIDED RADIOFREQUENCY ABLATION OF UPPER SPINAL CORD PAIN PATHWAYS FOR CANCER-RELATED PAIN Ahmed M. Raslan, M.D. Department of Neurosurgery, Ain Shams University, Cairo, Egypt, and Department of Neurological Surgery, Oregon Health & Science University, Portland, Oregon Reprint requests: Ahmed M. Raslan, M.D., Department of Neurological Surgery, Oregon Health & Science University, Mail Code CH8N, 3303 SW Bond Avenue, Portland, OR 97239. Email: [email protected] Received, November 1, 2006. Accepted, September 24, 2007.

OBJECTIVE: The author presents data to support the continued need for ablative procedures, particularly cordotomy, in the management of cancer-related pain. METHODS: Fifty-one patients with cancer-related body or face pain were treated with computed tomography-guided radiofrequency ablation of the spinothalamic tract or trigeminal tract nucleus in the upper cervical region of the spinal cord. Forty-one patients underwent a unilateral cervical cordotomy, and 10 patients underwent a trigeminal tractotomy–nucleotomy. Three methods to assess patient pain were used: degree of pain relief, Visual Analog Scale, and total sleeping hours. The Karnofsky scale was used to measure the patient’s level of function pre- and postprocedure. RESULTS: After surgical intervention, patients reported initial and 6-months follow-up pain relief as 98 and 80%, respectively. CONCLUSION: Computed tomography-guided ablation of the upper cervical spinal cord is a safe and effective procedure to treat cancer pain involving the body or face. There remains a need for ablative procedures, in particular cordotomy, in the management of cancer-related pain. KEY WORDS: Cancer pain, Cervical cordotomy, Percutaneous computed tomography-guided, Trigeminal tractotomy–nucleotomy Neurosurgery 62[ONS Suppl 1]:ONS226–ONS234, 2008

M

edical advancements in the diagnosis and treatment of cancer continue to prolong the life of patients with cancer. Unfortunately, patients with advanced cancer frequently report pain as a symptom, and as they live longer, the need for effective pain management to improve their quality of life is of paramount importance. Pain incidence varies depending on the neoplasm type, cancer stage, and the extent of spread, which various studies have attempted to define. Reported cancer pain incidence numbers range from 20 to 50% of all patients in early cancer stages to 55 to 95% in later stages (1, 29, 30). For economic, logistic, or medical reasons, many of these patients are often not effectively treated pharmacologically, and surgical procedures remain a viable and feasible option. The upper spinal cord is one of the most important regions of the central nervous system, where destructive pain-relieving procedures are commonly performed, the most common applications being percutaneous cordotomy and trigeminal tractotomy–nucleotomy (TR-NC) (8). However, as a result of the widespread use of intrathecal narcotics, the number of ablative spinal cord procedures performed today to control


DOI: 10.1227/01.NEU.0000297060.73518.DD

pain is on the decline. Of note, opiate availability in Egypt (where this study was undertaken) is limited, and strict opiate-dispensing regulations have contributed to a state of “pain undertreatment.” The environment is such that those patients who successfully obtain prescribed opiates ultimately limit their use. Also, expensive intrathecal opioid pumps are often economically infeasible to both patients and institutions. The purpose of this study was to evaluate the safety and effectiveness of computed tomography-guided ablation of upper spinal cord pathways. The author believes and the results of this study show that computed tomography (CT) guidance increases the safety and precision of ablative spinal cord procedures and justifies revived use to treat cancer pain and perhaps other indications, (e.g., TR-NC for anesthesia dolorosa, which is otherwise very difficult to treat). Additionally, given a situation in which intrathecal opioids are not available or not a viable option, procedures such as CT-guided ablation of upper spinal cord pathways for patients with cancer pain are a feasible option.


PERCUTANEOUS COMPUTED TOMOGRAPHIC-GUIDED RADIOFREQUENCY

PATIENTS AND METHODS Fifty-one patients with cancer-related body or face pain were treated with extracranial ablative neurosurgical procedures. Patients with somatic pain on one side of the body reaching to the midline and below dermatome C5 and patients with visceral pain restricted to one side of the body not reaching the midline were offered cervical cordotomy. A CT-guided unilateral cervical cordotomy is in essence a unilateral spinothalamic tractotomy, and the term “cervical cordotomy” is used throughout the article to refer to unilateral spinothalamic tractotomy. Patients with somatic pain involving one side of the face, tongue, or inner mouth were offered the option of TR-NC. Forty-one patients underwent cervical cordotomy, and ten patients underwent TR-NC. All procedures were performed at the Department of Neurosurgery, Ain Shams University, Cairo, Egypt, by the author between 2000 and 2004. Techniques used were those previously described by Kanpolat and Cosman (7) and Kanoplat et al. (9–12) with use of CT guidance to approach the upper spinal cord between the occiput, C1, and C1–C2. Patients were referred to the author from the Department of Oncology, Ain Shams University, Cairo, Egypt, for surgical pain treatment after primary cancer treatment. For study inclusion, patients had to fulfill the following criteria: 1) histopathological establishment of a cancer diagnosis; 2) pain duration of greater than 1 month; 3) a Karnofsky scale greater than 40 (23); 4) pulmonary function tests with parameters greater than 50% of normal; 5) an estimated life expectancy of more than 3 months; 6) absence of bleeding tendency. Pain intensity was assessed by the surgeon or the surgical team preoperatively using a Visual Analog Scale (VAS), and total duration of sleep uninterrupted by pain was recorded in a patient diary as previously described (27). Patients were further evaluated using the Karnofsky scale (Table 1) (13). Patient postoperative evaluation included VAS, total sleeping hours, and pain scale at postoperative Day 1 and at 1, 3, and 6 months. Scores were recorded as I, no pain; II, almost no pain or partial but satisfactory pain relief; III, partial, nonsatisfactory pain relief; IV, same; V, worse. Patients were also evaluated for possible complications on postoperative Day 1 using an evaluation of pin prick sensation and level of analgesia/anesthesia and neurological evaluation. Postoperative evaluations were performed by either the surgeon(s) or the surgical supportive team, which included a nurse and a medical assistant. The Karnofsky performance scale was used with all patients preoperatively and immediately on postoperative Day 1. It was not used at later follow up as a result of nonspecific scores, which can be affected by conditions such as disease progression and treatment. Therefore, any Karnofsky patient performance scale changes, in this instance, are related directly to the procedure performed. Patients with a Karnofsky scale of less than 40 were considered too sick to tolerate the procedure and were offered other alternatives such as spinal epidural opioids. Pain medications pre- and postprocedure were not tracked in this particular instance (see availability of opioids referred to as the beginning of this article). Intrathecal opioid pumps were reserved only for those patients who initially qualified for cervical cordotomy or TR-NC based on the inclusion criteria but were not suitable candidates as a result of either the type or location of pain. Therefore, for those patients with cancer pain and a life expectancy such that epidural catheter use was infeasible, cordotomy or TR-NC was considered the only viable option.

Surgical Technique Most patients underwent lumbar puncture 30 minutes before the procedure to introduce water-soluble contrast material (8 ml) into the cerebrospinal fluid (CSF) (46 procedures). For five procedures, contrast dye

NEUROSURGERY

TABLE 1. Karnofsky Performance Scale scoresa Percent 100

Description Normal; no complaints; no evidence of disease; able to work

90

Able to carry on normal activity; minor symptoms; able to work

80

Normal activity with effort; some symptoms; able to work

70

Independent; cares for self; unable to carry on normal activity; not able to work

60

Disabled; dependent; requires occasional assistance; cares for most needs

50

Moderately disabled; dependent; requires considerable assistance and frequent care

40

Severely disabled; dependent; requires special care and assistance

30

Severely disabled; hospitalized, death not imminent

20

Very sick; active supportive treatment needed

10

Moribund; fatal processes are rapidly progressing

0

Death

was directly injected through the introducing cannula used for lesioning, avoiding lumbar puncture. However, as a result of higher contrast dye viscosity and slow dye circulation time within the CSF, which was associated with decreased visibility of the outline and true diameters of the cord and dura (Figs. 1 and 2), we ultimately stopped using this approach. Prophylactic antibiotics were administered to all patients.

Percutaneous CT-guided Radiofrequency Unilateral Cervical Cordotomy Forty-one patients underwent CT-guided unilateral cervical cordotomy and were usually admitted on the same day or the evening of the day before the planned procedure. Patients’ anticoagulant or antiplatelet use was stopped preoperatively, and patients fasted for 6 hours before the procedure. Cordotomies were performed with the patient supine and the head immobilized using tape fixed to the gantry of the CT. The mastoid tip was used as an anatomic landmark with the cannula entry site located approximately 1 cm below. The skin entry site was further determined with the aid of CT localization of the space between C1 and C2 and marked on the skin with the CT tube laser. After marking the skin for entry, the surrounding area was prepared in a sterile fashion. CT data acquisition started with a lateral scan followed by axial parallel 2-mm cuts through the lateral space between C1 and C2 where the C2 nerve root exits. The following data were obtained: skin to dura distance, anteroposterior and mediolateral diameter of the cord, and the best slice through which skin entry can be achieved. After local infiltration with 2% Xylocaine (AstraZeneca, Wilmington, DE), a needle electrode system was used (KCTE Kit; Radionics, Inc., Burlington, MA). The KCTE electrode needle tip used was significantly thinner (electrode tip diameter 0.27 mm) than most cordotomy needle tips (0.5 mm) (7). To avoid unintentional penetration of the thecal sac,


RASLAN

FIGURE 1. Computed tomographic (CT) scanning was used to check the cannula before dural puncture; the dye is not yet injected.

FIGURE 2. The electrode is inside the cord. The inadequate mixture of cerebrospinal fluid with dye can be noticed. the cannula was introduced perpendicular to the skin in the coronal plane of the mastoid tip to a depth less than skin to dura distance. Typically, this corresponds to an entry point into the thecal sac in front of the dentate ligament. Adjustment of the needle after placement and revisions may be required before an adequate trajectory is achieved. Before dural puncture, 2 ml of 2% Xylocaine was injected to avoid pain from contact with the C2 ganglion and the dura. After free flow of CSF was obtained, further CT imaging was performed to confirm cannula placement. According to the previously recorded mediolateral diameter of the cord, the exposed electrode tip length was adjusted (the KCTE electrode allows for adjustment of the electrode tip in the range of 2–6 mm) and the electrode then introduced through the cannula. The mediolateral cord diameter was generally 9 mm, and an exposed tip of 4 mm typically penetrated the cord deep enough to cover the anterolateral quadrant. In a situation in which we encountered a spinal cord with a small diameter, the exposed tip was adjusted to not cross the midline. Ideal placement of the KCTE electrode was achieved when the KCTE cannula was just touching the side of the spinal cord in front of the dentate ligament without penetrating the cord and no part of the bare electrode was bridging the distance between the tip of the cannula and the spinal cord. Successful penetration of the spinal cord was confirmed by an impedance jump to greater than 1000 Ω (impedance ⬍400 Ω in the CSF and ⬎1000 Ω inside the spinal cord) (Figs. 1 and 2). Additionally, the patient indicated either a mild pain and/or electric sensation.


CT was continually used to check electrode position in relation to cord geometry to target the area of maximal pain intensity and somatotropic pain fiber arrangement within the spinothalamic tract (anteromedial for the arm, anterolateral for the leg). Cord penetration usually produced some form of sensory phenomena in the contralateral body and increased pain sensation. Impedance monitoring was mainly beneficial in identifying cord penetration and was of limited significance in verification of the target inside the cord. Electrical stimulation was performed through the electrode of the spinothalamic tract, with frequencies adapted both for sensory and motor activation to verify the location of the electrode tip in the intended target. All patients were stimulated with 100 Hz, pulse duration 0.1 ms, as a sensory frequency. No lesion was made until a sensation of warmth or cold and/or painful sensation in the opposite half of the body (as remarked by the patient) after sensory stimulation was obtained with a voltage criteria below 0.5 V. If no response was obtained with these parameters, i.e., sensory response above 0.5 V or sensory response in a localized area (lower limb only), the position of the electrode was adjusted until an adequate response was obtained. All patients were then stimulated with 2 Hz, pulse duration 0.1 ms, as a motor frequency. If no motor response was obtained in the ipsilateral half of the body other than local contraction of the neck musculature with stimulation voltages above 1 V, then the procedure continued. In all patients, the procedure started with a “test” lesion, 60 seconds at 60⬚C, followed by testing the opposite half of the body for decreased pin prick sensation, hypoalgesia, or pain relief. To detect any subtle change in motor power, lesioning was performed with the patient’s ipsilateral leg elevated. If inadequate hypoalgesia and/or pain relief were not obtained (as was the case with most patients), a second lesion, 70 seconds at 70⬚C, was applied and followed by testing with increasing time and degree until hypoalgesia was obtained. Up to three lesions were performed in some patients with a temperature up to 90⬚C for 90 seconds. After successful lesioning, the cannula was removed, and the puncture site was sterilized and covered. All patients were observed overnight for sleep apnea, a recorded complication of cordotomy. Blood pressure and pulse were regularly monitored, and antibiotics and pain killers were administered on postoperative Day 1 to control puncture site and headache pain (commonly reported by patients).

Percutaneous CT-guided Radiofrequency TR-NC Ten patients underwent TR-NC. Preoperative preparation was similar to the cordotomy procedure, with the patient prone (not supine) and the head immobilized using tape to the CT gantry. Skin entry point, which was located 5 to 7 mm lateral of midline and ipsilateral to the painful side, was further determined with the aid of CT cuts, which correspond to the craniocervical junction and marked on the skin with the CT laser. After marking the skin for entry, the surrounding area was prepped in a sterile fashion. CT-guided data acquisition was similar to the cordotomy procedure but with cuts through the craniocervical junction (Fig. 3). The following data were obtained: skin dura distance, anteroposterior and mediolateral diameter of the cord, and the best slice through which skin entry can be achieved. Local infiltration with 2% Xylocaine, the needle electrode system used (KCTE Kit; Radionics, Inc.) and CT checks were the same as for the cordotomy procedure. The target for the percutaneous CT-guided radiofrequency TR-NC was the junction between the ipsilateral one-third and two-thirds length of the cord equator (Fig. 4). Again, cord penetration usually produced pain in the painful area, if a lesion was made in that particular target; electric stimulation also confirmed this phenomenon.



All patients were stimulated with 100 Hz pulse for 0.1 ms as a sensory frequency. No lesion was performed until a severe pain response in the ipsilateral face and/or throat with voltage criteria below 0.5 V was obtained. If no response was obtained with the following parameters: sensory response above 0.5 V or sensory response in a localized area (e.g., upper limb or lower limb), inaccurate electrode placement in the surrounding ascending sensory tracts was indicated, and the position of the electrode was adjusted FIGURE 3. An initial CT slice until an adequate response was scan was used to collect data for obtained. All patients were then tractotomy-nucleotomy. stimulated with 2 Hz, pulse duration 0.1 ms, as a motor frequency. If no motor response was obtained in the ipsilateral half of the body other than local contraction of the neck musculature with stimulation voltages above 1 V, then the procedure continued. Trigeminal tract lesioning can be extremely painful, and pain severity is directly proportional to the temperature used. In one patient, we performed two lesions, a first at 50⬚C for 50 secFIGURE 4. Position of the tractoonds followed by a second at tomy electrode is shown (left). 80⬚C for 80 seconds. In the other nine patients, we performed two lesions, a first at 45⬚C for 180 seconds followed by a second at 55⬚C for 120 seconds. These parameters obviated the need for sedation, although narcotic premedication was not routine, it was available and used by patients unable to tolerate the procedure. The specific parameters used were based on clinical observations and personal communications with Yucel Kanpolat, M.D. (unpublished data). After successful lesioning and cannula removal, patient observation was the same as that used for the cordotomy procedure.

cord diameters at occiput–C1 level, C1–C2 level, and the distance between the skin and the dura were measured; anteroposterior diameter was 5.6 to 10.8 mm (mean, 9.2 mm), mediolateral diameter was 9.3 to 14.2 mm (mean, 10.7 mm), and distance between the skin and the dura was 43 to 56 mm (mean, 50.9 ⫾ 3.4 mm) (Table 2). The total number of cordotomy lesions performed was 87 (average, 2.1 lesion/procedure); average temperature and time were 70.9⬚C and 74.6 seconds, respectively. For the TR-NC procedure, average lesions/procedure, temperature, and time were 2, 51.5⬚C, and 141.5 seconds, respectively. Surgical outcome was evaluated by measuring the patients’ activity and description and expression of pain (Table 3). The mean preoperative Karnofsky Performance Scale (KPS) was 55.5 ⫾ 6.7, and the mean postoperative KPS was 76.9 ⫾ 7.6. Mean preoperative VAS score was 8.5 ⫾ 0.8, and on postoperative Day 1, mean VAS score dropped sharply to 1.2 ⫾ 1.06. The follow-up VAS scores at 1, 3, and 6 months postoperatively were 1.7 ⫾ 1.2, 1.8 ⫾ 1.1, and 2.3 ⫾ 0.6, respectively. Patient total sleeping hours were also significantly improved. Preoperative total sleeping hours in a 24-hour period were 3.25 hours, which increased in the immediate postoperative period to 7 hours. Mean sleeping hours gradually decreased to 4.8 hours at 6 months follow-up but still remained statistically significant compared with the preoperative value. The most common pathology treated was pleural mesothelioma (47%); other pathologies treated are outlined in Table 4. Level and site of decreased pin prick sensation are recorded in Table 5. Decreased pin prick sensation was achieved in the opposite half of the body in all patients with only two exceptions. In these two patients, increased restlessness and patient anxiety rendered the patients uncooperative; to ensure safety, the procedure was halted after a single lesion. Preoperative daily living pain and 6 months postoperative follow-up pain relief were compared (Table 6). Postoperatively, 98% of the patients were Grade I or II, which we categorized as a successful procedure. At 1 month postoperatively, 98% of the

TABLE 2. Summary of data measureda

Statistics

Variable

SPSS for Windows (version 4.0; SPSS, Inc., Chicago, IL) was used to analyze data. Significance levels were P ⬎ 0.05 ⫽ insignificant result and P ⬍ 0.001 ⫽ significant result. Pain evaluations used included Karnofsky scale and VAS and VAS progression over 6 months. A paired t test was applied to compare results for each follow-up period and the scale of the preoperative period. Total sleeping hours were measured, and a paired t test was applied to mean total sleeping hours for each follow-up period and mean total preoperative sleeping hours.

Minimum Maximum

Skin dura

43.0

56.0

Mean ± SD 50.9 ± 3.4

distance (mm)

Cordotomy

50.9 ± 3.7

procedure (mm) TR-NC procedure (mm) Anteroposterior cord

51.4 ± 2.05 5.6

10.8

9.2 ± 1.1

9.3

14.2

10.7 ± 0.8

980

1900

1272.9 ± 182.1

diameter (mm)

RESULTS Most patients were between 50 and 70 years of age (range, 18–82 yr; mean, 52.1 yr) and 33 (64.7%) were male (29, 30). The time from initial reported pain onset to the surgical procedure ranged between 1 and 9 months (mean, 3.8 ⫾ 1.4 mo). Spinal

NEUROSURGERY

Mediolateral cord diameter (mm) Impedance (Ω) a

SD, standard deviation; TR-NC, trigeminal tractotomy–nucleotomy.


RASLAN

TABLE 3. Pain surgery outcome evaluation using scaling systems that measure aspects of patients’ activity, description, and expression of paina Minimum value

Maximum value

Mean

SD

Preoperative

40

70

55.5

6.7

Immediate postoperative

60

90

76.9

7.6

Preoperative

5

10

8.5

0.8


0

6

1.2

1.1

1 month postoperative

0

8

1.7

1.2

33.5

0.001

3 months postoperative

0

8

1.8

1.1

35.8

0.001


1

8

2.3

0.6

36.1

0.001

1.0

Operative period Karnofsky Performance Scale Visual Analog Scale

Total sleeping hours

a b

t-value

P valueb

15.6

0.001

⫺40

0.001

Preoperative

0

5

3.25


5

9

7.1

0.9

18.7

0.001


4

8

5.8

0.9

⫺13.1

0.001


4

8

5.6

0.94

⫺12.0

0.001


2

6

4.8

0.9

⫺6.7

0.001

SD, standard deviation. P ⬍ 0.001 ⫽ statistically significant result.

TABLE 5. Level and site of decreased pin prick sensation

TABLE 4. Summary of cancer patient pathology Pathology Mesothelioma (pleural) Metastatic adenocarcinoma

Number

Percent

1

C5

26

C6

10

C8

2

9.8

Left face

4

7.8

Right face

6

3

5.9

None

2

2

3.9

Total

51

10

19.6

47 3.9

Meningioma (temporal)

1

2

Breast cancer

5

Bronchogenic carcinoma

4

Tongue cancer Multiple myeloma

(gastrointestinal tract)

stomach cancer, basal cell carcinoma of the face, Pancoast tumor) Total

Number

C4

24 2

Others (Ewing sarcoma, prostate cancer,

Level

51

100

patients remained Grade I or II, which decreased to 80% at 6-months follow-up. One patient, who was Grade III at 1month follow-up, dropped to Grade IV at 3-months follow-up and died before the 6-month follow-up contact. Despite two patients having only one lesion performed and the procedure halted, results in terms of pain relief in the first 3 months were not affected and most likely reflect the effect of a single lesion. No complications related to the procedures used were observed. No change in motor power was recorded in any patient. No incidence of sleep apnea was encountered. There were no permanent or serious complications such as respiratory depression and/or weakness in this series. Complications were transient and not severe. Hypotension occurred in three cases (two cordotomies and one TR-NC). All patients


responded to treatment with 24-hour parenteral fluid administration and were discharged without event. Headache was reported in four patients (three cordotomies and one TR-NC). All were treated with analgesics and fluids for 48 hours and discharged without event. Dysesthesia occurred in two patients undergoing cordotomy and persisted in one patient for 3 days and in the other for 2 weeks, which completely resolved by the first month follow-up assessment.

DISCUSSION The concept of sectioning pain-carrying fibers to relieve pain was first proposed by Spiller in 1912 (28). Since that first proposal, a plethora of data detailing pain pathway ablation has been published, from the open sharp knife procedure of Sjöqvist (26), who performed a transverse stab incision of the trigeminal tract to relieve facial pain, to percutaneous procedures using specialized instruments and devices able to produce standardized lesions with electrophysiological parameter



TABLE 6. Patient pain relief postprocedure Scale

Pain




I

No pain

41 (80.4%)

41 (80.4%)

33 (64.7%)

16 (32%)

II

Partial satisfactory pain relief

9 (17.6%)

7 (13.7%)

15 (29.4%)

24 (48%)

III

Partial nonsatisfactory pain relief

1 (2.0%)

3 (5.9%)

1 (2.0%)

7 (14%)

IV

Same, no change in pain

0 (0%)

0 (0%)

2 (3.9%)

3 (6%)

V

Worse pain

0 (0%)

0 (0%)

0 (0%)

0 (0%)

Total a


51 (100 %)

51 (100 %)

51 (100 %)

50 (100 %)a

One patient died by 6-month follow-up evaluation.

measurements such as impedance and temperature for safety and effectiveness of ablation. Complications inherent to conventional open procedures and the poor quality of life of patients for whom these procedures were required led to a percutaneous approach to access spinal cord pain pathways. Mullan et al. (16, 17) performed the first cordotomy procedure with a strontium needle followed by attempts using unipolar anodal electrolytic lesioning; since the pioneering work of Moossy et al. (15) and the first percutaneous radiofrequency cordotomy, neurosurgeons have continued to attempt to improve surgical results with a focus on enhanced lesioning and/or testing the electrophysiological responses of these pathways. However, there remains a dependence on conventional spinal cord image guidance. Lin (14) attempted an alternate route, anterior low cervical cordotomy, and Hitchcock (3) used a stereotactic apparatus to reach the anterolateral quadrant and other targets through the dorsal high cervical region. Trigeminal tractotomy as previously mentioned was first performed by Sjoqvist (26). However, Hitchcock popularized the procedure in the 1970s using an innovative stereotactic approach, which enabled posterior access to the tiny structures of the upper spinal cord and brainstem. Many authors have published their experiences with TR-NC (2–4, 24, 25). Nashold et al. (19) popularized the caudalis dorsal root entry zone procedure, essentially a modified TR-NC with lesion extension to the obex and involvement of the trigeminal nerve caudal structure of the spinal nucleus. This modified procedure was used to treat intractable facial pain conditions such as anesthesia dolorosa and intractable facial cancer pain. The CT-guided tractotomy procedure detailed in this article could be considered a miniaturized caudalis dorsal root entry zone procedure (28). In 1988, Kanpolat et al. (7a) introduced the concept of CT guidance to achieve optimal spinal cord pain pathway targeting. This innovation was aptly described by Osenbach and Burchiel (21) as one of the most important contributions to pain pathway ablation research and one that ultimately led to a resurgent interest in surgery of the spinal cord to treat pain. Spinal cord pain pathways have been the primary target for ablation of pain-carrying fibers with cordotomy, a procedure previously performed frequently at most neurosurgical cen-

NEUROSURGERY

ters. However, the evolving use of intraspinal narcotics has markedly affected the use of spinal cord ablation procedures to such an extent that very few neurosurgical centers in the United States, Europe, or worldwide now use them to manage pain. In essence, the cordotomy procedure could be considered a dying art (28). In the series of 51 patients we describe in this article, we examined the effect of spinal cord pain pathway ablation through radiofrequency lesioning using CT guidance. Targets were the spinothalamic tract and the trigeminal tract nucleus. The clinical and functional outcome of the two procedures was collectively evaluated and analyzed. The same methods of guidance and lesioning were used for all procedures. The selected target differed according to pain type and location. CT guidance offers the following advantages over fluoroscopic-guided cordotomy: 1) measurements of the distance between the skin entry point and the dura are used to avoid unintentional puncture of the dura and the spinal cord; 2) measurements of spinal cord diameter (Table 2) allow the surgeon to determine the electrode length required to penetrate the cord to provide precise needle/target contact (e.g., a cord mediolateral diameter is 10 mm; the electrode length that will penetrate the cord and avoid contralateral lesioning should not be more than 5 mm); 3) precise imaging of the needle–target relationship further ensures target verification along with electrophysiological and clinical verification methods. However, CT-guided procedures do lack the instantaneous feedback of live fluoroscopy that shows electrode penetration depth. In this series, the primary patient pathology was mesothelioma (Table 4), and tumor characteristics make it an ideal candidate for the procedures described. Mesothelioma is a somatic tumor, i.e., causes nociceptive pain, is well circumscribed, and does not exceed the chest wall, and the lungs are often not infiltrated, thus preserving pulmonary function. Similarly, Jackson et al. (6) reviewed 52 patients with mesothelioma and fluoroscopic cordotomy with adequate pain relief, and Price et al. (22) in a prospective study involving 38 patients with thoracic malignancies, including mesothelioma and fluoroscopic cordotomy for cancer pain, found cordotomy to be safe and tolerated, even by patients with impaired respiratory function tests.


RASLAN

Outcome results for TR-NC as reported in the literature range from 60 to 90% excellent pain relief for different pathologies, including intractable trigeminal neuralgia and postherpetic neuralgia along with cancer pain. However, no previous TR-NC study has focused on the results for cancer pain alone (6, 18, 22). Our results showed that initial patient pain relief, postprocedure (Grade I or II), was 98%, which decreased to 80% at 6-months follow-up (Table 6). Of note, 50 of the 51 patients survived to 6-months follow-up. Despite a mean VAS increase from 1.22 in the immediate postoperative period to 2.27 at 6-months follow-up, there was a statistically significant reduction in VAS after cordotomy that remained until the 6month follow-up (Table 3). Total average sleeping hours increased from 3.25 hours before the procedure to 7 hours in the immediate postoperative period and at 6-months followup, total sleeping hours were 4.78 hours, which remained statistically significant (Table 3). Differences between previously published results and those reported in this article may in part be the result of case diversity and/or differing complication incidence (2–5, 8, 10, 12, 19, 24, 26). In a literature review of outcome results for cordotomy (37 series, 5770 cases), pain relief was independently evaluated for patients with cancer and those without cancer. In the patients with cancer, at least 75% of patients reported pain relief for 6 months (28). Mean KPS scores improved immediately postprocedure from 55.5 to 76.9, which was statistically significant. The KPS reflects the overall functional status of the patient and as such was not measured in subsequent follow-up visits because other aspects of disease progression not necessarily related to pain control may skew results. Nauta et al. (20) reported a similar improvement in KPS (50–70) after myelotomy for cancer pain. Overall, our results show effective and safe pain control when CT guidance is used in conjunction with ablative spinal cord procedures to treat cancer pain. These results also support and are in agreement with previously published studies using CT guidance for percutaneous upper cervical spinal cord pain pathway ablation (8–10). The current trend in cancer pain management is the use of “reversible” intrathecal opioids. This has ultimately led to an almost abandonment of ablative spinal cord procedures to treat cancer pain with little published data in the last decade and procedures delegated to a “forgotten art.” However, in situations in which for economic, logistic, or medical reasons intrathecal opioids are not available or not a viable option, the procedures described remain invaluable tools to the surgeon who has the ability to use them.

CONCLUSION On the basis of the results of this and other studies, we strongly believe that ablation of the spinal cord pain pathway, when used with CT guidance, can provide beneficial and safe pain relief to patients with cancer. The total number of studies reported on this particular technique remains small. However, in light of the rapid growth of intraspinal narcotic, deep brain, or spinal cord stimulation used to treat cancer pain, a cost-


effective, safe, successful, rapid procedure that is performed once remains a viable solution, especially when economic and social factors deem other options infeasible.

REFERENCES 1. Bonica JJ, Ventafridda V, Twycross RG: Cancer pain, in Bonica JJ (ed): The Management of Pain. Philadelphia, Lea & Febiger, 1990, p 401. 2. Crue BL, Lasby V, Kenton B, Felsoory A: Needle scope attached to stereotactic frame for inspection of cisterna magna during percutaneous radiofrequency trigeminal tractotomy. Appl Neurophysiol 39:58–64, 1976. 3. Hitchcock E: Stereotactic trigeminal tractotomy. Farmatsiia 19:131–135, 1970. 4. Hitchock ER, Schvarcz JR: Stereotaxic trigeminal tractotomy for post-herpetic facial pain. J Neurosurg 37:412–417, 1972. 5. H o s o b u c h i Y, R u t k i n B : D e s c e n d i n g t r i g e m i n a l t r a c t o t o m y. Neurophysiological approach. Arch Neurol 25:115–125, 1971. 6. Jackson MB, Pounder D, Price C, Matthews AW, Neville E: Percutaneous cervical cordotomy for the control of pain in patients with pleural mesothelioma. Thorax 54:238–241, 1999. 7. Kanpolat Y, Cosman ER: Special radiofrequency electrode system for computed tomography-guided pain-relieving procedures. Neurosurg 38:600–602, 1996. 7a. Kanoplat Y, Atala M, Deda H, Siva A: CT guided extralemniscal myelotomy. Acta Neurochir (Wein) 91:151–152, 1988. 8. Kanpolat Y, Deda H, Akyar S, Bilgic S: CT-guided percutaneous cordotomy. Acta Neurochir Suppl (Wien) 46:67–68, 1989. 9. Kanpolat Y, Deda H, Akyar S, Caglar S: CT-guided pain procedures. Neurochirurgie 36:394–398, 1990. 10. Kanpolat Y, Savas A, Batay F, Sinav A: Computed tomography-guided trigeminal tractotomy–nucleotomy in the management of vagoglossopharyngeal and geniculate neuralgias. Neurosurgery 43:484–489, 1998. 11. Kanpolat Y, Savas A, Caglar S, Akyar S: Computerized tomography-guided percutaneous extraleminiscal myelotomy. Neurosurg Focus 2:e5, 1997. 12. Kanpolat Y, Savas A, Caglar S, Aydin V, Tascioglu A, Akyar S: Computed tomography-guided percutaneous trigeminal tractotomy–nucleotomy. Techniques in Neurosurgery 5:244–251, 1999. 13. Kupers R, Gybels JM: Evaluation of results of pain surgery, in Gildenberg P, Tasker R (eds): Textbook of Stereotactic and Functional Neurosurgery. New York, McGraw Hill, 1998, pp 1311–1319. 14. Lin P: Percutaneous lower cervical cordotomy, in Gildenberg P, Tasker R (eds): Textbook of Stereotactic and Functional Neurosurgery New York, McGraw Hill, 1998, pp 1485–1505. 15. Moossy J, Sagone A, Rosomoff HL: Percutaneous radiofrequency cervical cordotomy: Pathologic anatomy. J Neuropathol Exp Neurol 26:118, 1967. 16. Mullan S, Harper PV, Hekmatpanah J, Torres H, Dobbin G: Percutaneous interruption of spinal-pain tracts by means of a strontium90 needle. J Neurosurg 20:931–939, 1963. 17. Mullan S, Hekmatpanah J, Dobben G, Beckman F: Percutaneous, intramedullary cordotomy utilizing the unipolar anodal electrolytic lesion. J Neurosurg 22:548–553, 1965. 18. Nagaro T, Adachi N, Tabo E, Kimura S, Arai T, Dote K: New pain following cordotomy: Clinical features, mechanisms, and clinical importance. J Neurosurg 95:425–431, 2001. 19. Nashold BS Jr, el-Naggar A, Mawaffak Abdulhak M, Ovelmen-Levitt J, Cosman E: Trigeminal nucleus caudalis dorsal root entry zone: A new surgical approach. Stereotact Funct Neurosurg 59:45–51, 1992. 20. Nauta HJ, Hewitt E, Westlund KN, Willis WD Jr: Surgical interruption of a midline dorsal column visceral pain pathway. Case report and review of the literature. J Neurosurg 86:538–542, 1997. 21. Osenbach R, Burchiel K: Percutaneous cordotomy, in Kaye A, Black P (eds): Operative Neurosurgery. Philadelphia, Churchill Livingstone, 2000, pp 1569–1579. 22. Price C, Pounder D, Jackson M, Rogers P, Neville E: Respiratory function after unilateral percutaneous cervical cordotomy. J Pain Symptom Manage 25:459–463, 2003. 23. Schag CC, Heinrich RL, Ganz PA: Karnofsky performance status revisited: Reliability, validity, and guidelines. J Clin Oncol 2:187–193, 1984. 24. Schvarcz JR: Stereotactic trigeminal tractotomy. Confin Neurol 37:73–77, 1975.



25. Schvarcz JR: Postherpetic craniofacial dysaesthesiae: Their management by stereotaxic trigeminal nucleotomy. Acta Neurochir (Wien) 38:65–72, 1977. 26. Sjöqvist O: Studies on pain conduction in trigeminal nerve. A contribution to the surgical treatment of facial pain. Acta Psychiatr Scand Suppl 17:1–139, 1938. 27. Sorge J, Sittl R: Transdermal buprenorphine in the treatment of chronic pain: Results of a phase III, multicenter, randomized, double-blind, placebo-controlled study. Clin Ther 26:1808–1820, 2004. 28. Tasker R: Percutaneous cordotomy for persistent pain, in Gildenberg P, Tasker R (eds): Textbook of Stereotactic and Functional Neurosurgery. New York, McGraw Hill, 1998, pp 1485–1505. 29. World Health Organization: Cancer Pain Relief. Geneva, World Health Organization, 1986. 30. World Health Organization: Cancer Pain Relief and Palliative Care. Geneva, World Health Organization, 1990.

Acknowledgments I thank Shirley McCartney, Ph.D., and Matthew A. Hunt, M.D., for their valuable contributions to the manuscript.

rior to ITDD in many ways. First, immediate effective analgesia can be obtained with a single, minimally invasive procedure that is generally relatively well tolerated, even in very debilitated patients. After a successful ablative procedure, many patients can significantly reduce and sometimes even eliminate the use of high-dose systemic opioids, which produce significant and often intolerable adverse effects. Whereas administration of ITDD requires continuous and indefinite care, a successful ablative procedure can often reduce or eliminate the need for patients to make repeated trips to health care providers. Finally, a procedure such as a percutaneous cordotomy is arguably less expensive and less of a financial burden on the health care system. I believe that as a specialty, we neurosurgeons (especially those who have a genuine interest and investment in the treatment of chronic pain) should make a concerted effort to educate our oncology colleagues as to the safety and benefits of ablative procedures for the treatment of cancer pain. Hopefully, these techniques will reemerge as vital tools for the treatment of patients with intractable cancer pain, and we can once again train neurosurgical residents in these important procedures. Richard K. Osenbach Durham, North Carolina

COMMENTS

R

aslan presents a contemporary series of spinal and brainstem ablative procedures for the treatment of severe intractable cancer pain of the body and face. This represents fairly substantial experience, considering the general lack of frequency with which ablative procedures in general and these two procedures in particular are currently performed. All procedures were performed using computed tomographic (CT) image guidance, and the author points out the importance of using diametral measurements of the spinal cord and brainstem for accurate positioning of the lesioning electrode. Although Raslan and other authors are staunch proponents of using CT guidance, it is my opinion that percutaneous cordotomy can just as easily, safely, and effectively be performed with the more traditional technique of C-arm fluoroscopy. In my opinion, the most critical information required for a safe and successful procedure is the target confirmation that is obtained with careful intraoperative physiological testing. However, if CT guidance is available, I agree that it is a viable alternative to the more classic approaches. Because of the inclusion criteria for their study, the authors arbitrarily limited the use of cordotomy and trigeminal tractotomy to patients with a life expectancy of more than 3 months. However, I argue that these procedures should be considered even in patients who may be expected to live no more than a few weeks if there is a reasonable possibility that they will result in improvement of the patients’ quality of life and ability to interact with friends and family during their remaining time. Nearly all patients obtained excellent immediate pain relief, which was sustained in 80% of patients who survived up to 6 months. I believe the real value of this report is that it once again provides evidence that percutaneous cordotomy and trigeminal tractotomy can be performed safely and can consistently provide patients with effective analgesia for refractory cancer pain. Before the introduction of intrathecal drug delivery (ITDD) in the early 1980s, ablative procedures such as percutaneous cordotomy were integral tools in the armamentarium of neurosurgeons for the treatment of severe intractable cancer pain. However, because of the widespread use of ITDD (much of which is performed by anesthesia-trained pain specialists), the application of these tried and true techniques has steadily declined. Another consequence of this trend has been that fewer and fewer neurosurgical residents are being trained in these techniques. This situation is truly unfortunate, given the effectiveness of these procedures. Given proper patient selection, I believe that procedures such as percutaneous cordotomy have a number of advantages and are supe-

NEUROSURGERY

T

he author presents the result of 41 CT-guided percutaneous cordotomies and 10 CT-guided trigeminal tractotomies in patients with cancer. Patients’ pain has been evaluated with postoperative pain levels, the Visual Analog Scale, and total sleeping hours. The Karnofsky scale has also been used to measure the level of patient function. The author obtained 98% and 80% pain relief immediately and after 6 months of follow-up, respectively. This is a well-written and clear article that can be followed easily. The author should be congratulated for the high success rate in pain relief. I think cordotomy is the best method for controlling unilaterally localized chronic cancer pain states. The complications drop significantly, to nearly 0%, with CT guidance. The author has made lesions in some patients with a temperature up to 90°C. To avoid complications I usually do not use this high temperature. For recurrence of pain, I prefer to repeat the cordotomy procedure. Most physicians involved in pain management consider destructive procedures only in the terminal stages of malignant diseases. However, I suggest that destructive procedures such as cordotomy and tractotomy should be used as early as possible. If the surgeons can denervate the painful area, the quality of life of patients will improve and they will be in a position to carry on their normal active lives, virtually eliminating dependence on doctors and hospitals. I also strongly recommend that the procedures should be applied only by neurosurgeons, as these unique procedures require not only expert surgical technique, but also extensive knowledge of neuroanatomy and neurophysiology. Yucel Kanpolat Ankara, Turkey

I

n this era of minimally invasive neurosurgery, it is all too easy to forget that some of the most effective treatments for cancer-related pain do not necessarily involve elaborate cocktails of narcotics or expensive intrathecal drug administration systems. The treatment of cancer pain with ablative techniques such as percutaneous cordotomy is a triedand-true technique, with a well-established track record. Moreover, there are many situations in which cancer pain cannot be well controlled with seemingly noninvasive techniques. It is particularly apropos that a contemporary series using CT guidance, instead of the older fluoroscopy guided procedure, be presented. Pioneered by Yucel Kanpolat, CT-guided cordotomy has been used in few centers.


RASLAN

Therefore, Raslan’s large series described in this study is important. In this series, the author examined parameters of lesioning, including needle displacements from the skin during the cervical lesioning procedures. As this is an image-guided procedure, I do not think these are terribly useful parameters, except for their ability to protect against premature entry into the thecal sac. Because the trajectory of the needle largely determines the actual distance to the target, however, it is difficult to know whether practitioners performing these procedures would find such numbers useful. Use of CT guidance, however, is of great benefit, as it allows very precise documentation of localization of the needle, as well as the length of the needle tip entry into the cord. My only other comment is related to the choice of lesioning temperatures in the study. In the cordotomy group, the author describes using temperatures as high as 90° C for 90 seconds in certain patients. This is an intensely hot needle and may reflect less than optimal placement of the needle. I am also concerned that this size of lesion may result in

complications. In addition, the high temperature of the needle is approaching the boiling point of water, which may result in gas formation and a very uneven and unpredictable lesion. Publications on the technique by other groups suggest that between 70 and 80° C for 60 seconds is sufficient for the lesion, although lesioning may need to be repeated after needle repositioning (as it was in this series). In a similar vein, the author reports using an extremely low temperature–long lesion time for trigeminal tractotomy because of the pain associated with higher temperature lesions. Because lesion size probably reaches an asymptote before 120 seconds, it does not make sense that a lesion created after 180 seconds would be any different from one created with a considerably shorter time. In addition, because the accepted isotherm for neuronal cell death is 44° C, it is unlikely the authors obtain much of a lesion with these parameters. Oren Sagher Ann Arbor, Michigan

Johann Jacob Hartlieb, copper engraving, August Christian Fleischmann. From: WolfHeidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



FUNCTIONAL AND STEREOTACTIC Operative Nuances

SPINAL AND NUCLEUS CAUDALIS DORSAL ROOT ENTRY ZONE OPERATIONS FOR CHRONIC PAIN Yucel Kanpolat, M.D. Department of Neurosurgery, Ankara University School of Medicine, Ankara, Turkey

Hakan Tuna, M.D. Department of Neurosurgery, Ankara University School of Medicine, Ankara, Turkey

Melih Bozkurt, M.D. Department of Neurosurgery, Ankara University School of Medicine, Ankara, Turkey

Atilla Halil Elhan, Ph.D. Department of Biostatistics, Ankara University School of Medicine, Ankara, Turkey Reprint requests: Yucel Kanpolat, M.D., Department of Neurosurgery, Inkilap S. 24/2, Kizilay, 06640 Ankara, Turkey. Email: [email protected] Received, September 9, 2006.

OBJECTIVE: Dorsal root entry zone (DREZ) operations came into medical practice after the demonstration of increased electrical activity in the dorsal horn of the spinal cord and brainstem in patients with deafferentation of the central nervous system after injury to these areas. The aim of the study was to describe the technique and the effectiveness of spinal DREZ and nucleus caudalis (NC) DREZ operations, which may be the treatments of choice in unique chronic pain conditions that do not respond to medical therapy or any other surgical methods. METHODS: Fifty-five patients (44 spinal, 11 NC DREZ) underwent 59 (48 spinal, 11 NC DREZ) operations. There were 44 men and 11 women with a mean age of 46.4 years (range, 24–74 yr). The mean follow-up period was 72 months (range, 6 mo–20 yr). Follow-up assessments were performed with clinical examination on the first day and in the sixth and twelfth months postoperatively. Patients’ pain scores and Karnofsky Performance Scale scores were also evaluated pre- and postoperatively. RESULTS: The initial success rates for spinal and NC DREZotomy procedures were 77 and 72.5%, respectively. In the spinal DREZotomy group, mortality occurred in one patient (2.2%). There were two cases of transient muscle weakness (4.4%) and two of cerebrospinal fluid fistulae (4.4%). In the NC DREZotomy group, mortality occurred in one patient (9%). There were two cases of transient ataxia (18%) and two of transient hemiparesis (18%). CONCLUSION: Spinal and trigeminal NC DREZ operations are effective in the treatment of intractable pain syndromes, especially in traumatic brachial plexus avulsions, segmental pain after spinal cord injury, postherpetic neuralgia, topographically limited cancer pain, and atypical facial pain. KEY WORDS: Dorsal horn, DREZ lesioning, Spinal cord, Trigeminal nucleus tract, Trigeminal tractonucleotomy Neurosurgery 62[ONS Suppl 1]:ONS235–ONS244, 2008

DOI: 10.1227/01.NEU.0000297066.44810.46

Accepted, September 4, 2007.

D ONLINE DIGITAL VIDEO

NEUROSURGERY

orsal root entry zone (DREZ) operations came into medical practice after the demonstration of increased electrical activity in the dorsal horn of the spinal cord and brainstem in patients with deafferentation of the central nervous system after injury to these areas (19, 26). Vic-D’Azyr (58) described the Hshaped structure of the central gray matter of the spinal cord. In 1824, Rolando (36) described a gelatinous layer in part of the H-shaped structure. This structure is well known as the second Rexed layer (34). After spinal cord injury, it is believed that injured areas express spontaneous activities. Drake and Stavraky (7) demonstrated deafferentation hyperactivity, and, in 1968, Loeser et al. (26) recorded spontaneous electrical hyperactivity from a paraplegic patient’s

injured spinal cord segment. In animal experiments, hyperactive neurons were observed after dorsal rhizotomies (15). After considering evidence based on the studies demonstrating neuronal activities after deafferentation within the dorsal horn, Sindou and Fischer (47) sectioned the lateral part of the DREZ in a patient with Pancoast tumor in 1972 for the treatment of pain resulting from cancer. Nashold et al. (31) in 1975 planned to make lesions in these spontaneously discharging areas to solve centrally originated pain and applied the procedure to four patients with pain resulting from brachial plexus avulsion. Nashold et al. established the procedure as puncturing of the intermediolateral sulcus by a radiofrequency (RF) electrode system, thereby


KANPOLAT ET AL.

destroying the DREZ. Since that time, this procedure has been carried out by a limited number of neurosurgeons for various indications (8, 10, 12, 13, 29, 31, 39, 48, 57). Nucleus caudalis (NC) DREZ operation is also used to treat various types of deafferentation pain syndromes affecting the face and was first described by Duke’s group (1, 13). According to Olszewski’s (32) detailed description of the anatomy, the target is one of the three parts of the nucleus of the spinal tract that is located in the medulla called the NC. The aim of this study was to describe the technique and the effectiveness of spinal DREZ and NC DREZ operation procedures, which may be the treatments of choice in unique intractable chronic pain conditions that do not respond to medical therapy or any other surgical methods.

Anatomic Description of Spinal DREZ for Surgical Orientation Dorsal rootlets, Lissauer ’s tract, and the dorsal horn together constitute the DREZ (42). According to their level from C1 to the coccygeal region, dorsal rootlets penetrate the intermediolateral sulcus with different patterns and organization, population, geometric shape, and diameter. According to Sindou’s description, each of the C2, C3, and C4 roots divides into approximately four rootlets, which are distinct, wellmarked, and with cylindrical penetration. C5, C6, C7, and C8 roots divide into approximately six rootlets with a diameter of 1.50 mm, making them the thickest together with lumbosacral rootlets. They run together and also have a cylindrical penetration. Thoracal roots divide into approximately five rootlets, which are wide and well-marked. They are very small with a diameter of 0.25 mm and have a filiform penetration. L1, L2, and L3 roots divide into approximately 10 rootlets, and these rootlets have a secondary division and a filiform penetration. L4 and L5, and S1, S2, and S3 divide into approximately seven rootlets and also have a cylindrical penetration. Independent from their level, each rootlet has an approximately 1-mm subpial course within the intermediolateral sulcus (43–45). The DREZ angles also differ. According to Young’s 82 measurements, the mean angle is 30 degrees at C6, 26 degrees at T4, 37 degrees at T12, and 36 degrees at L3 (43). In 1982, Risling and Hildenbrand (35) reported that 30% of the primary nociceptive afferent fibers, whose cell bodies reside in the dorsal root ganglion, enter the ventral roots. Actually, they make a Uturn and go back to the dorsal root to enter the spinal cord dorsally and project onto the dorsal horn (4, 59). The afferent nociceptive fibers, before entering the dorsal horn, bifurcate rostrocaudally or trifurcate rostrocaudal laterally to run for a few segments in a thin tract of small axons capping the dorsal horn and giving off branches into the gray matter at different levels called Lissauer ’s tract. Lissauer ’s tract has the important assignment of modulating the signals transmitting from afferent nociceptive fibers. The lateral part of Lissauer’s tract contains the propriospinal fibers; therefore, destruction of the medial part of Lissauer ’s tract should result in decreased excitability of nociceptive fibers (6). The dorsal


horn plays an important role in modulating the inputs from the peripheral to the central nervous system by inhibiting or increasing the signals. It contains interneurons, which are assigned to modulate the sensorial and propriospinal inputs and axonal prolongations of peripheral nerves. Nociceptive axonal endings terminate in Rexed Levels 1 to 2 and 5. The neurons in these areas begin to fire like those contained in an epileptogenic area after deafferentation, and these neurons may provide the neuropathological basis of central pain. Hyperactivity in the deafferentated neurons has been demonstrated both in animal models (15) and in human models (14, 16, 19). Thus, the DREZotomy operation aims to destroy these hyperactive areas and eliminate central pain.

Anatomic Description of Trigeminal NC for Surgical Orientation Trigeminal afferents that carry the sensations of pain and temperature bifurcate when entering the pons and send a caudal ward branch (called the descending trigeminal tract) into the medulla (22). The descending trigeminal tract overlies the spinal trigeminal nucleus in the posterolateral part of the spinal cord at the cervicomedullary junction. Primary sensory fibers from the VIIth, IXth, and Xth cranial nerves also enter the descending tract of the trigeminal nerve. The three divisions of the trigeminal nerve have a special topographic organization in the descending tract of the nerve. All of the nociceptive afferents of the Vth, VIIth, IXth, and Xth cranial nerves form the descending trigeminal tract. The fibers from the VIIth, IXth, and Xth cranial nerves lie slightly medially behind the tract. Trigeminal afferents that carry the sensations of pain and temperature, unlike other sensory modalities, descend in the spinal trigeminal nucleus (22). The spinocerebellar pathway and the pyramidal tract overlie the NC and spinal trigeminal tracts, and these two important pathways are at significant risk during the procedure. The spinal trigeminal nucleus has three distinct subdivisions along its pontospinal extent: 1) the nucleus oralis, located rostrally between the pons and medulla; 2) the nucleus interpolaris, located intermedially; and 3) the NC, located at the medullispinal junction and extending down to the C2 segment level (Fig. 1). The NC represents the substantia gelatinosa, and there is an extensive overlap between facial and high cervical afferents, where the VIIth, IXth, and Xth cranial nerve afferents also end. The secondary caudalis neurons begin to fire like those contained in an epileptogenic area after deafferentation, and these neurons may provide the neuropathological basis of central pain. Destruction of the NC plays a special role in pain relief (18). There is a topographic representation of the ipsilateral face on the spinal tract of the trigeminal nerve, i.e., the most central areas of the face terminate highest on the NC and the most peripheral areas of the face end lowest. This is referred to as “onion skin organization” and causes the central area of the face to be spared from hypoalgesia after NC DREZ operations if the lesion does not extend above the obex (24, 25, 55).


MICROSURGICAL DORSAL ROOT ENTRY ZONE

FIGURE 1. Nucleus caudalis and related anatomic structures are shown.

Surgical Technique of Spinal DREZ Lesioning (see video at web site) While preparing the patient for the surgical procedure, we obtain the diametric measures of the affected spinal cord by computed tomographic scan. With the help of these dimensions, the depth of the electrodes is arranged. The patient is placed in the prone position under general anesthesia. If the patients are determined to have affected cervical or high thoracic segments, fixation of the head with the Mayfield headholder is necessitated; however, there is no need to immobilize the head in patients requiring lesions below the T4 level. Chest rolls must be optimally placed to decrease the intrathoracic pressure to avoid epidural venous bleeding. The extension of the exposure is directly related to the patient’s symptoms as a result of affected dermatomes. A median vertical skin incision is carried out. The thoracodorsal or dorsolumbar fascia is opened and subperiosteal dissections of the paravertebral muscles are performed. We recommend additional laminectomy of one level both rostrally and caudally. For conus medullaris DREZ lesioning, laminectomy can be performed from T11 to L2. If the patient is young, laminotomy should be considered with a high-speed drill. To preserve the spinal processes and interspinous ligaments, hemilaminectomies must be considered. For unilateral pain, we prefer hemilaminectomy, but limited hemilaminectomy involves a narrow space and limited exposure. Thus, visualization of the root entry zone is mandatory. In particular, in patients with brachial plexus avulsion, there is considerable arachnoid scarring and the DREZ is displaced from its normal position as a result of rotation of the cord. There is a high incidence of adhesions and scar formation, which makes the dissection of the neural structures troublesome. In those cases, it can be problematic to correctly identify the DREZ, in which case bilateral exposure should be considered. This allows identification of the dorsal median sulcus and midline vein and the contralateral normal DREZ. For a safe procedure, normal rootlets are seen bilaterally one level above and below, and tiny vessels entering the sulcus should be observed; coagulation of DREZ is then done in these regions as well (Video 1) (43, 46). The dura is opened vertically just in the line of the roots of the affected side with the help

NEUROSURGERY

of the surgical microscope. The arachnoid is opened directly over the DREZ with the help of micronerve hooks or microscissors. It must be remembered that in the thoracic segments, the main dorsal roots exit through the foramina one or two vertebral levels below, a phenomenon referred to as “blank space” (30). DREZ lesions are made by the DREZ electrodes, limited sulcotomy, and bipolar coagulation. We usually dissect the root entry zone and make the lesions in the lateral part of DREZ by standard RF electrodes. In brachial plexus avulsions, it is not easy to access the avulsed region in the spinal cord. The spinal cord is typically distorted and rotated. The electrode tip is 2 mm in length and 0.25 mm in diameter. Standard RF lesions are made with the parameters of 75 to 80⬚C and 15 seconds for each lesion. The essential part of the technique is keeping the tip of the electrode at 45 degrees and 2-mm depth to the spinal cord. The number of lesions depends on the affected length with 1-mm intervals between each lesioning. An intraoperative monitoring technique by experienced staff is optimal to ensure safety and efficacy of the DREZotomy operation. In our series, an intraoperative monitoring technique could not be used because of technical limitations. This is definitely a shortcoming to be resolved. After all the lesions are made, total hemostasis is essential because blood products may cause secondary tethering of the spinal cord. The dura is then sutured watertight with atraumatic silk, and closure is then completed in order of anatomic layers (Fig. 2).

Surgical Technique of NC DREZ Lesioning (see video at web site) While preparing the patient for the surgical procedure, we obtain the diametric measures of the affected brainstem by magnetic resonance imaging scan and the depth of the electrodes is arranged using these dimensions. After fixation of the

FIGURE 2. Spinal dorsal root entry zone at C5 to C7 level showing configuration of the rootlets emerging from the spinal cord.


KANPOLAT ET AL.

head with the Mayfield headholder, the patient is placed in the prone position under general anesthesia. Chest rolls must be optimally placed to decrease the intrathoracic pressure. The incision is started at the midline just at the external occipital protuberance and extended below the spinous process of C2. The incision is then extended to the deep layers of subcutaneous tissue and nuchal fascia. At the first layer, the trapezius muscle is incised; at the second layer, the semispinalis and splenius capitis muscles are incised; and at the third level, the rectus capitis posterior and oblique muscles are incised. Great care must be taken to avoid injury to the greater occipital nerve while making these incisions. The posterior arcus of C1, laminae, and spinous process of C2 are exposed. After advancing the self-retaining retractor, a suboccipital craniectomy and C1 total laminectomy are performed. In NC DREZotomy operations, C1 hemilaminectomy should be sufficient, but we usually perform total laminectomy for C1. Ligamenta flava are also removed. Dura is incised in Y-shaped configuration with the help of a surgical microscope. If there is a large occipital sinus, the bleeding is prevented by silk sutures. The arachnoid is then opened by a nerve hook and microscissors. The two most important landmarks during the operation are the obex and C2 rootlets, and the surgeon must identify these structures. The sulcus intermediolateralis and cranial roots of the accessory nerve are also important landmarks. DREZ lesions are made using the special NC DREZ electrodes. Electrode tip dimensions are 3 mm in length and 0.25 mm in diameter. It must be kept in mind that the depth of the electrodes is arranged with the help of diametric measurements using imaging modalities. Lesions are created just above the sensory rootlets of the C2 within the intermediolateral sulcus immediately above the obex behind the cranial roots of the accessory nerve. Standard RF lesions are made 20 times with parameters of 75 to 80⬚C and 15 seconds for each lesion (Fig. 3). The essential part of the technique is keeping the tip of the electrode perpendicular to the medulla. Between C2 and the obex, approximately 20 lesions are created (Video 2) (1, 13). An intraoperative monitoring technique by experienced staff is optimal for a safe and effective NC DREZotomy operation. A neurophysiological monitoring technique is useful in more selective lesion-making in the NC, thereby limiting the number of lesions and potentially reducing the morbidity of the procedure (18). In our series, an intraoperative monitoring technique could not be used because of technical limitations and this is definitely a shortcoming to be resolved. The dura is then sutured watertight with atraumatic silk, and closure is completed in order of anatomic layers.

PATIENTS AND METHODS Between 1986 to 2006, 55 patients (44 spinal DREZ, 11 NC DREZ) underwent 59 (48 spinal DREZ, 11 NC DREZ) operations in our department; in all cases, previous medical and surgical operations had failed. Previous surgical interventions of the patients undergoing NC DREZ are summarized in Table 1. There were 44 men and 11 women with a mean age of 46.4 years (range, 24–74 yr). The etiologies of pain in the spinal DREZ group were brachial plexus avulsion in 14


FIGURE 3. Schematic drawing showing nucleus caudalis DREZ lesioning and related anatomic structures. patients, phantom limb pain in four patients, intractable painful spasticity after spinal cord injury in 17 patients, tumor in seven patients, and postherpetic neuralgia in two patients (Table 2). The etiologies of pain in the NC DREZ group were glossopharyngeal neuralgia in two patients, pain resulting from oral and maxillary tumors in three patients, atypical facial pain in four patients, trigeminal neuralgia in one patient, and geniculate neuralgia in one patient. The mean follow-up was 72.6 months (range, 6 mo–20 yr). All follow-up assessments were done with clinical examination on the first day and in the sixth and twelfth months. The psychiatric conditions of the patients were also evaluated. Pain score and life quality were evaluated with the Visual Analog Scale (VAS) and Karnofsky Performance Scale. Pain measurement scale was determined as follows: I, no pain; II, partial satisfactory pain relief; III, partial nonsatisfactory pain relief; and IV, no change in pain. We considered Grade I and II patients as having achieved pain relief after surgical intervention. VAS was used to score the severity of pain and Karnofsky Performance Scale was used to determine the performance status of the patients on the first postoperative day. Wilcoxon signed rank test was used to test the difference between preoperative and postoperative VAS score and Karnofsky Performance Scale. SPSS for Windows 11.5 (SPSS, Inc., Chicago, IL) was used for statistical analysis. A P value less than 0.05 was considered significant.

RESULTS After the spinal DREZotomy procedure, 77% of patients reported initial pain relief (Grade I: 59%, Grade II: 18%, Grade III: 11.5%, Grade IV: 11.5%). After the NC DREZotomy procedure, 72.5% of patients reported initial pain relief (Grade I: 54%, Grade II: 18.5%, Grade III: 18.5%, Grade IV: 9%). One year after surgery, 69% of the patients in the spinal DREZotomy group and 62% of the patients in the NC DREZotomy group still considered their relief satisfactory. Karnofsky Performance Scale was used immediately pre- and postoperatively as a measure of patient performance change. Long-term VAS and Karnofsky scores were not sufficient. Minimum and maximum preoperative scores were 40 and 80, respectively (mean, 55.6 11.0), whereas minimum and maximum postoperative scores



TABLE 1. Previous surgical interventions in the nucleus caudalis dorsal root entry zone groupa Patient no.

a

Intervention for primary pathology

Diagnosis

Intervention for pain

1

Tongue squamous Ca

RT

*

2

Glossopharyngeal neuralgia

3

Parotid gland Ca

4

AFP (trigeminal neuropathic pain)

*

Alcohol injection (2) Frasier operation (2) MVD Tractotomy

5

Glossopharyngeal neuralgia

—

Tractotomy

6

AFP (Trigeminal neuropathic pain)

*

Radiofrequency lesions MVD

7

Invasive pituitary tumor

8


*

Radiofrequency lesions Tractotomy

9


*


10


*


11

Geniculate neuralgia

*

Tractotomy (2)

*

Tractotomy

Surgery

Tractotomy

Transcranial surgery

Tractotomy (2)

Ca, cancer; AFP, atypical facial pain; RT, radiotherapy; MVD, microvascular decompression.

TABLE 2. Pain etiologies of the spinal dorsal root entry zone groupa Pain etiology

No. of patients

Brachial plexus avulsion

14

Phantom limb pain

4

Spinal cord ınjury

17

Tumor

7

Postherpetic neuralgia

2

were 40 and 100, respectively (mean, 73.7 13.8) and the difference was highly significant (P 0.001). The mean preoperative VAS score was 7.41 0.67 with a minimum of 6 and maximum of 9. On the first postoperative day, the score dropped sharply to 1.95 2.38 (P 0.001). Mean, median, minimum, maximum, and standard deviation values of Karnofsky Performance Scale and VAS scores are summarized in Table 3. In the 17 patients with spasticity, a 50% or more decrease in spasticity was achieved in eight (47%). In one case in the spinal DREZotomy group, the patient was discharged from the hospital with normal blood urine levels on the seventh day, but mortality occurred as a result of acute renal failure 20 days after the operation (2.2%). There were two cases of transient muscle weakness (4.4%) and two of cerebrospinal fluid fistula (4.4%). In one patient in the NC DREZotomy group, mortality occurred as a result of massive pulmonary embolism (9%).

NEUROSURGERY

There were two cases of transient ataxia (18%) and two of transient hemiparesis (18%).

DISCUSSION Taylor indicated that 20 to 90% of patients with traumatic brachial plexus avulsion reported deafferentation pain (56). Thus, in these conditions, patients require treatment. Following the influence of Nashold in creating a lesion to the substantia gelatinosa as a therapeutic target for traumatic brachial plexus injury, spinal DREZ lesioning was initiated in such patients with a pain-free rate of 60 to 100% (3, 11, 31, 33, 46, 57). Partial lesion of the brachial plexus, for example, postradiation plexopathy, is preferably treated with stimulation. The probability of degeneration of the targeted fibers by spinal cord stimulation up to the brainstem should be considered (51). Our results are consistent with these in that excellent pain control was achieved, especially in the brachial plexus avulsion group (13 patients [92.8%]) in our series. The common feature of the patients was age with a median age of 31 years in these patients. All 13 patients were able to resume their jobs and their daily living activities. There are primarily two types of chronic pain that disturb the patient after spinal cord injury, radicular pain and burning phantom pain, with an estimated rate of 30 to 45% (28, 37). Nashold reported 50% good and 8.9% fair pain relief in 56 patients with severe spinal cord injury (9). According to the original description by Sindou and Mertens (50) and Sindou et al. (52), in the spinal cord injury group, DREZ lesioning should only be considered as


KANPOLAT ET AL.

TABLE 3. Mean, median, minimum, maximum, and standard deviation values of Karnofsky Performance Scale and Visual Analog Scale scoresa DREZ

a

Mean SD

Preoperative Median (minimum– maximum)

Mean SD

Postoperative Median (minimum– maximum)

P value

VAS

7.41 0.67

7 (6–9)

1.95 2.38

1 (0–8)

0.001

Karnofsky

55.6 11.0

60 (40–80)

73.7 13.8

80 (40–100)

0.001

VAS, Visual Analog Scale; DREZ, dorsal root entry zone; SD, standard deviation.

a treatment option for pain corresponding to the injured segments and performed at the corresponding segmental levels. In our group, microsurgical DREZotomy was performed in these trauma patients, in the subgroup with pain with radiculometameric distribution and from “transition zone or end zone pain” in which pain was reflected to the primary affected level or neighboring levels. Good pain relief was achieved in 47% of the patients. In the postherpetic neuralgia group, Friedman et al. (10) reported good pain relief in eight of 12 patients in the shortterm follow-up. In another study, Friedman and Bullit (8) reported that 25% good pain relief persisted in the long-term follow up. The majority of the postherpetic neuralgias are located at the thoracic level. It should be remembered that the dorsal horn is very narrow at these levels. Hence, because of the close relationship of DREZ and the dorsal column in this area, lesioning should involve the dorsal column and cause complete abolition of the proprioceptive sensations. Although the number of patients with postherpetic neuralgia was small in our series (n 2), complete pain relief was achieved in this group. Jensen and Rasmussen (20) reported phantom limb pain at the stump in nearly 15% of patients. Saris et al. (40) reported 83% pain relief, particularly in patients with radicular symptoms, 29% pain relief in persistent stump pain, and 67% pain relief in phantom type burning pain alone, and they concluded that the rate of response to the procedure differs. In our series, pain relief was achieved in 75% of this group. DREZ lesioning should not be considered as a treatment option based on this indication, and thalamic sensory relay nucleus stimulation for the treatment of phantom limb pain should be considered (60). In the cancer group, less invasive techniques like percutaneous cordotomy and tractotomy–nucleotomy (TR-NC) are applied by a limited number of neurosurgeons (23). In our practice, because the major pathophysiology of pain resulting from cancer is peripheral because of the infiltration of the nociceptive fibers and because it is a less invasive procedure, we prefer TR-NC for craniofacial cancers and percutaneous cordotomy for other cancers located below the C3 level. If these procedures fail, patients with both nociceptive and neuropathic pain are treated with DREZ lesions in our group; however, because the DREZ operation is a weighty operation, the medical status of the patients should be evaluated carefully. If the cancer-originated pain has a neuropathic component, especially in Pancoast tumors, we prefer the DREZotomy operation as a first intervention. Sindou and


Goutelle (48) achieved 85% good pain relief for various types of malignant tumors in their report. If the pain is central pain of a predominantly neuropathic type in the cancer group, the DREZotomy operation should be considered. In our series, pain relief was achieved in 60%. In movement disorders, DREZotomy operation may be suggested as an alternative surgical treatment for patients with segmental dystonia in the extremities (21). Although the authors present a single unique case, this operation should be considered in these situations. In postparaplegic conditions, 10 to 25% of patients with spinal cord and cauda equina injuries develop refractory pain (52). The most valuable reports are established by Sindou’s group. Sindou and Mertens reported on the long-term results of the microsurgical DREZotomy in a series of 44 patients with refractory neuropathic pain secondary to spine injury. They achieved pain relief in 68% of patients who had segmental pain distribution compared with 0% of patients with predominantly infralesional pain (50). In patients with hyperspastic (and painful) paraplegia, Mertens and Sindou (29) achieved reduction in painful spasticity and spasms in 78 and 88%, respectively, in 121 patients with a mean followup of 5 years and 6 months. Sindou and Jeanmonod (49) reported on a series of 53 bedridden patients with harmful spasticity who were treated with microsurgical DREZotomy. Both spasticity and spasms were significantly decreased or suppressed in 75 and 88.2% of the patients, respectively. Sindou et al. (53) reported that beneficial effects on both spasticity and pain led to a gain in functional status in 93% of cases in his series of 16 patients with hemiplegia who experienced harmful spasticity in the upper limb and were treated with selective posterior rhizotomy in the DREZ. In our series, decompression procedure (laminectomy) alone was thought to be sufficient to resolve the spasticity, and when this procedure failed, we performed DREZotomy as a second step. However, our results are not as satisfactory as those of the reported series, and we believe that our technique in spasticity warrants improvement. Regarding the unacceptable mortality in the spinal DREZ lesioning group, we accepted the patient for surgical treatment for long-segment postherpetic neuralgia (Th 1–7). The patient was 72 years old and had chronic renal failure, atherosclerotic cardiac disease, and a 20-year history of diabetes mellitus. We achieved excellent pain control in this patient, but the patient died from acute renal failure 20 days after discharge from the hospital.



Sjöquist (54) demonstrated the trigeminal tract in the brainstem and developed the tractotomy for chronic intractable facial pain. Crue et al. (5), Schwartz (41), and Kanpolat et al. (24, 25) improved the technique by applying stereotactic procedures and lesioning the tractus at the cervicomedullary junction. Hosobuchi and Rutkin (17) conducted trigeminal tractotomy on a 58-year-old woman with atypical facial pain for 15 years. Postoperatively, her pain had completely resolved. In a series of descending trigeminal tractotomy, Lopez et al. (27) obtained associated analgesia of the ipsilateral pharynx, tympanic membrane, and external auditory meatus in five of the six patients, which indicates the close proximity of the pain and temperature fibers of the VIIth, IXth, and Xth cranial nerves to those of the descending trigeminal tract. Bullard and Nashold (2) reported 25 NC DREZ operations for severe facial pain. At the time of discharge, the authors had achieved good to excellent pain relief in 24 of 25 (96%) patients, and 1 year after surgery, 12 of 18 (67%) patients still considered their relief as good to excellent. Rossitch et al. (38) reported five NC DREZ operations for medically refractory facial pain in patients with cancer and achieved 60% pain relief 1 year after surgery. Our results are consistent with those of the Duke groups. Trigeminal tractus and NC, which represent the substantia gelatinosa, are suitable targets for surgical treatment of facial pain. In our experience, the efficiency of interventions to this target (DREZotomy, TR-NC) have been proven, but because computed tomographic scan-guided tractotomy resolves the problem in the majority of cases, it should be the first choice in these patient groups because it is an effective and less invasive alternative to NC DREZ lesioning. TR-NC should be performed first; if it fails, trigeminal NC DREZ operation should be resorted to as a final procedure (23). Regarding the unacceptable mortality in the NC DREZ lesioning group, the decision was taken at the excessive insistence of the patient. The patient was 52 years old and had undergone TR-NC twice because of geniculate neuralgia. The patient had an excessive posterior fossa anomaly. Excellent pain control was achieved, but during the subacute postoperative period, the patient experienced acute pulmonary edema resulting from massive pulmonary embolism and died on the seventh day after surgery.

CONCLUSION Spinal and NC DREZ operations both aim to reduce nociceptive input and/or neural hyperexcitability in the affected area. Spinal and trigeminal NC DREZ operations are effective in the treatment of intractable pain syndromes, especially in traumatic brachial plexus avulsions, segmental pain after spinal cord injury, postherpetic neuralgia, topographically limited cancer pain, and atypical facial pain.

REFERENCES 1. Bernard EJ Jr, Nashold BS Jr, Caputi F, Moossy JJ: Nucleus caudalis DREZ lesions for facial pain. Br J Neurosurg 1:81–91, 1987.

NEUROSURGERY

2. Bullard DE, Nashold BS Jr: The caudalis DREZ for facial pain. Stereotact Funct Neurosurg 68:168–174, 1997. 3. Chen HJ: Dorsal root entry zone lesions in the treatment of pain following brachial plexus avulsion and herpes zoster. J Formosan Med Assoc 91:508–512, 1992. 4. Coggeshall RE, Coulter JD, Willis WD: Unmyelinated fibers in the ventral root. Brain Res 57:229–233, 1973. 5. Crue Bl, Todd EM, Carregal EL: Percutaneous radiofrequency stereotactic trigeminal tractotomy, in Crue BL (ed): Pain and Suffering. Springfield, Charles C. Thomas, 1970, pp 9–79. 6. Denny-Brown D, Kirk EJ, Yanagisawa N: The tract of Lissauer in relation to sensory transmission in the dorsal horn of spinal cord in the macaque monkey. J Comp Neurol 151:175–200, 1973. 7. Drake CG, Stavraky G: An extension of the ‘law of deafferentation’ to afferent neurons. J Neurophysiol 11:229–238, 1948. 8. Friedman AH, Bullit E: Dorsal root entry zone lesions in the treatment of pain following brachial plexus avulsion, spinal cord injury and herpes zoster. Appl Neurophysiol 51:154–169, 1988. 9. Friedman AH, Nashold BS Jr: Pain of spinal origin, in Youmans JR (ed): Neurological Surgery. Philadelphia, W.B. Saunders Co., 1990, pp 3950–3959. 10. Friedman AH, Nashold BS Jr, Ovelman-Levitt J: Dorsal root entry zone lesions for the treatment of post herpetic neuralgia. J Neurosurg 60:1258–1262, 1984. 11. Garcia-March G, Sanchez-Ledesma MJ, Diaz P, Yaqüe L, Anaya J, Gonçalves J, Broseta J: Dorsal root entry zone lesions versus spinal cord stimulation in the management of pain from brachial plexus avulsion. Acta Neurochir Suppl 39:155–158, 1987. 12. Gorecki JP, Burt T, Wee A: Relief from chronic pelvic pain through surgical lesions of the conus medullaris dorsal root entry zone. Stereotact Funct Neurosurg 59:69–75, 1992. 13. Gorecki JP, Nashold BS Jr, Rubin L, Ovelmen-Levitt J: The Duke experience with nucleus caudalis DREZ coagulation. Stereotact Funct Neurosurg 65:111–116, 1995. 14. Guenot M, Bullier J, Rospars J, Lansky P, Mertens P, Sindou M: Single-unit analysis of the spinal dorsal horn in patients with neuropathic pain. J Clin Neurophysiol 20:143–150, 2003. 15. Guenot M, Bullier J, Sindou M: Clinical and electrophysiological expression of deafferentation pain alleviated by dorsal root entry zone lesions in rats. J Neurosurg 97:1402–1409, 2002. 16. Guenot M, Hupe JM, Mertens P, Ainsworth A, Bullier J, Sindou M: A new type of microelectrode for obtaining unitary recordings in the human spinal cord. J Neurosurg 91:25–32, 1999. 17. Hosobuchi Y, Rutkin B: Descending trigeminal tractotomy. Arch Neurol 25:115–125, 1971. 18. Husain AM, Elliott SL, Gorecki JP: Neurophysiological monitoring for the nucleus caudalis dorsal root entry zone operation. Neurosurgery 50:822–827, 2002. 19. Jeanmonod D, Sindou M, Magnin M, Baudet M: Intra-operative unit recordings in the human dorsal horn with a simplified floating microelectrode. Electroencephalogr Clin Neurophysiol 72:450–454, 1989. 20. Jensen TS, Rasmussen P: Phantom pain and related phenomena after amputation, in Wall PD, Malzack R (eds): Textbook of Pain. Edinburgh, Churchill Livingstone, 1999, ed 2. 21. Kahilogullari G, Ugur HC, Savas A, Dirik EB, Akbostanci MC, Elibol B, Kanpolat Y: Management of a hemidystonic patient with thalamotomy, campotomy and cervical dorsal root entry zone operation. Stereotact Funct Neurosurg 83:180–183, 2005. 22. Kanpolat Y: Computed tomography-guided percutaneous trigeminal tractotomy–nucleotomy. Techniques in Neurosurgery 5;3:244–251, 1999. 23. Kanpolat Y, Caglar S, Akyar S, Temiz C: CT-guided pain procedures for intractable pain in malignancy. Acta Neurochir Suppl 64:88–91, 1995. 24. Kanpolat Y, Deda H, Akyar S, Caglar S, Bilgic S: CT-guided trigeminal tractotomy. Acta Neurochir (Wien) 100:112–114, 1989. 25. Kanpolat Y, Savas A, Ugur HC, Bozkurt M: The trigeminal tract and nucleus procedures in treatment of atypical facial pain. Surg Neurol 64 [Suppl 2]:S96–S100, 2005. 26. Loeser JD, Ward AA Jr, Lowell E, White LE Jr: Chronic deafferentation of human spinal cord neurons. J Neurosurg 29:48–50, 1968.


KANPOLAT ET AL.

27. Lopez PD, Salazar FG, Sierra OM, Yague RC, Llaurado G, Lopez E: Trigeminal nucleus caudalis dorsal root entry zone radiofrequency thermocoagulation for invalidating facial pain. Neurochirurgica 14:25–32, 2003. 28. Melzack R, Loeser JD: Phantom body pain in paraplegics: Evidence for a central ‘pain-generating mechanism’ for pain. Pain 4:195–210, 1978. 29. Mertens P, Sindou M: Microsurgical DREZotomy for the treatment of spasticity of the lower limbs. Neurochirurgie 44:209–218, 1998. 30. Nashold BS Jr, El-Naggar A: Dorsal root entry zone (DREZ) lesioning, in Rengachary SS, Wilkins RH (eds): Neurosurgical Operative Atlas. AANS Publications Committee, vol 2, pp 9–24. 31. Nashold BS Jr, Urban B, Zorub DS: Phantom pain relief by focal destruction of the substantia gelatinosa of Rolando, in Bonica JJ, Albe-Fessard D (eds): Advances in Pain Research and Therapy. New York, Raven Press, 1976, vol 1, pp 959–963. 32. Olszewski J: On the anatomical and functional organization of the spinal trigeminal nucleus. J Comp Neurol 92:401–413, 1950. 33. Powers SK, Adams JE, Edwards MS, Boggan JE, Hosobuchi Y: Pain relief from dorsal root entry zone lesions made with argon and carbon dioxide microsurgical lasers. J Neurosurg 61:841–847, 1984. 34. Rexed B: The cytoarchitectonic organization of the spinal cord in the cat. J Comp Neurol 96:414–495, 1952. 35. Risling M, Hildenbrand C: Occurrence of unmyelinated axon profiles at distal, middle and proximal levels in the ventral root C7 of cats and kittens. J Neurol Sci 56:219–225, 1982. 36. Rolando L: Ricerche Anatomiche sulla Strucura del Midollo Spinale [in Italian]. Dizionario Perio dico di Medicina. Torino, Staperia Reale, 1824. 37. Rose M, Robinson JE, Ells P, Cole JD: Pain following spinal cord injury: results from a postal survey. Pain 34:101–102, 1988. 38. Rossitch E Jr, Zeidman SM, Nashold BS Jr: Nucleus caudalis DREZ lesions for facial pain. Br J Neurosurg 1:81–91, 1987. 39. Samii M, Moringlane JR: Thermocoagulation of the dorsal root entry zone for the treatment of intractable pain. Neurosurgery 15:953–955, 1984. 40. Saris SC, Iacono RP, Nashold BS Jr: Successful treatment of phantom pain with dorsal root entry zone coagulation. Appl Neurophysiol 51:188–197, 1988. 41. Schwartz JR: Postherpetic craniofacial dysaesthesia. Their management by stereotactic trigeminal tractotomy. Acta Neurochir 44:35–42, 1977. 42. Sindou M: Etude de la junction radiculo-medullaire posterieure: La radicellotomie posterieure selective dans la chirurgie de la douleur [in French]. Lyon, (dissertation), 1972. 43. Sindou M: Spinal entry zone interruption for persistent pain, in Gildenberg PL, Tasker RR (eds): Textbook of Stereotactic and Functional Neurosurgery. Houston, McGraw-Hill, 1998, pp 1565–1572. 44. Sindou M: Dorsal root entry zone lesions, in Burchiel KH (ed): Surgical Management of Pain. New York, Thieme Medical Publishers, 2002, pp 701–713. 45. Sindou M: Microsurgical drezotomy, in Schmidek HJ, Roberts DW (eds): Operative Neurosurgical Techniques: Indications, Methods and Results. Philadelphia, Elsevier-Saunders, 2006, ed 5, pp 1573–1588. 46. Sindou M, Blondet E, Emery E, Mertens P: Microsurgical lesioning in the dorsal root entry zone for pain due to brachial plexus avulsion: A prospective series of 55 patients. J Neurosurg 102:1018–1028, 2005. 47. Sindou M, Fischer G: La radicellotomie posterieure selective. Premiers resultats dans le chirurgie de la douleur [in French]. Neurochirurgie 20:397–408, 1974. 48. Sindou M, Goutelle A: Surgical posterior rhizotomies for the treatment of pain, in Krayenbühl H (ed): Advances and Technical Standards in Neurosurgery. Wein, Springer-Verlag, 1983, vol 10, pp 147–183. 49. Sindou M, Jeanmonod D: Microsurgical DREZotomy for the treatment of spasticity and pain in the lower limbs. Neurosurgery 24:655–670, 1989. 50. Sindou M, Mertens P: Neurosurgical management of neuropathic pain. Stereotact Funct Neurosurg 75:76–80, 2000. 51. Sindou MP, Mertens P, Bendavid U, Garcia-Larrea L, Mauguiere F: Predictive value of somatosensory evoked potentials for long lasting pain relief after spinal cord stimulation: practical use for patient selection. Neurosurgery 52:1374–1384, 2003. 52. Sindou M, Mertens P, Wael M: Microsurgical drezotomy for pain due to spinal cord and/or cauda equina injuries. Long-term results in a series of 44 patients. Pain 92:159–171, 2001.


53. Sindou M, Mifsud JS, Boisson D, Goutelle A: Selective posterior rhizotomy in the DREZ for treatment of hyperspasticity and pain in the hemiplegic upper limb. Neurosurgery 18:587–595, 1985. 54. Sjoquist O: Studies on pain conduction in trigeminal nerve. A contribution to the treatment of facial pain. Acta Psychiatr Neurol Suppl 17:9–139, 1938. 55. Taren JA, Kahn EA: Anatomic pathways related to pain in face and neck. J Neurosurg 19:116–121, 1962. 56. Taylor PE: Traumatic intradural avulsion of the nerve roots of the brachial plexus. Brain 85:579–602, 1962. 57. Thomas DGT, Jones SJ: Dorsal root entry zone lesions (Nashold’s procedure) in brachial plexus avulsion. Neurosurgery 15:966–968, 1984. 58. Vic-D’Azyr F: Traité d’Anatomie et de Physiologie; Vol 1, Anatomie et Physiologie du Cerveau; Vol 2, Planches Anatomiques. Paris, Didot. 59. Willis WD: The pain system: The neural bases of nociceptive transmission, in Gildenberg PL (ed): The Mammalian Nervous System, Pain and Headache. Basel, Karger, vol 8, 1985. 60. Yamamoto T, Katayama Y, Obuchi T, Kano T, Kobayashi K, Oshima H, Fukaya C: Thalamic sensory relay nucleus stimulation for the treatment of peripheral deafferentation pain. Stereotact Funct Neurosurg 84:180–183, 2006.

COMMENTS

T

he authors present a relatively large series of patients who underwent spinal dorsal root entry (DREZ) procedures for systemic pain or nucleus caudalis DREZ lesioning for intractable facial pain. The single best indication for DREZ lesioning remains without question deafferentation pain caused by brachial plexus avulsion, and in this series 31% of the patients underwent DREZ procedures for this indication. Indeed, DREZ lesioning can be expected to produce significant pain reduction in 80% to 90% of patients with plexus avulsion. I do believe it is important to distinguish whether the pain is from avulsion or from a severe stretch injury. In my opinion, this is more than just an academic exercise, because although DREZ lesioning is highly effective for patients with avulsion injuries, it is relatively ineffective for neuropathic pain after a stretch injury. Over the course of time, it has been well documented that the DREZ procedure is largely ineffective for the diffuse burning pain that often occurs in the anesthetic area below the level of spinal cord injury. However, it has been proven to be effective for the band-like “end-zone” or “transition zone pain” that not uncommonly occurs at the point of transition from anesthesia to more normal sensation. The authors treated two patients with postherpetic neuralgia. Despite the early enthusiasm for treating postherpetic neuralgia with DREZ lesioning, it is not consistently effective and in elderly patients thoracic DREZ lesioning is associated with considerable morbidity. Indeed, the thoracic cord tends to be small, and there is clearly a higher risk of injury to either the dorsal columns or corticospinal pathways. Given the lack of effect and the increased risk, I would rarely recommend a DREZ procedure for postherpetic neuralgia. Interestingly, the largest group of patients in this series who underwent a spinal DREZ procedure had painful spasticity after a spinal cord injury. I must admit that I have never personally used this procedure for spasticity per se or spasticity-related pain. It seems to me that intrathecal baclofen would produce more consistent results. In addition, it is certainly less invasive and associated with less morbidity than a DREZ procedure. Indeed, the authors report that only eight of these 17 patients achieved at least a 50% reduction in spasticity. It would have been nice had the authors provided a change in Ashworth scores for this subgroup of patients. The degree of pain relief in this subgroup of patients is also unclear, and I would argue that the number of patients with spasticity who derived benefit is far less than one would expect for patients treated with intrathecal baclofen. Only 11 patients underwent nucleus caudalis DREZ lesioning. The authors report that four of these patients suffered from atypical facial



pain. Unfortunately, different authors use this particular term to designate different types of pain. In my opinion, and according to the classification of facial pain develop by Burchiel, atypical facial pain is synonymous with a somatoform pain disorder. If, in fact, these patients did have true atypical facial pain, then I would question the use of any surgical procedure, let alone a destructive procedure that carries significant risk. My other observations relate to the authors’ technique. They seem to advocate a unilateral exposure or hemilaminectomy for patients with unilateral pain. Personally, my preference is to perform a bilateral laminectomy in all patients. Not uncommonly, in patients who have sustained nerve root avulsions, there is considerable arachnoid scarring, and the pathological side of the spinal cord can be atrophied and rotated, making identification of the DREZ difficult. A bilateral exposure allows for identification of normal anatomy on the contralateral side and in my opinion is very helpful in correctly identifying the DREZ. Finally, the authors place a significant degree of importance on measuring the diameter of the spinal cord and brainstem for performing spinal and nucleus caudalis DREZ lesioning, respectively. I still have difficulty understanding the rationale for how this measurement is at all helpful. The classic DREZ procedure is an open operation with direct visualization of the spinal cord. The electrodes designed for both spinal and caudalis DREZ have a fixed length of exposed tip that is inserted into either the dorsal horn or the nucleus caudalis at a specified angle. The electrodes have a collar where the insulation begins that prevents the electrode from penetrating too deeply when used properly. Notwithstanding these concerns, Kanpolat et al. have shown that both spinal and nucleus caudalis DREZ procedures can be effective in selected patients with certain intractable pain conditions. I applaud the authors for again providing evidence of the effectiveness of ablative neurosurgical procedures and their important role in the overall management of patients with chronic intractable pain. Richard K. Osenbach Durham, North Carolina

T

he aim of this clinical report was to describe the technique and the effectiveness of spinal DREZ and nucleus caudalis DREZ operations. The authors have accomplished this aim. The relief of pain reported was 97% for the spinal DREZ and 90% for the caudalis DREZ procedures, an improvement over the results in the original reports of Nashold. The authors emphasize the importance of careful anatomical identification of the lesion site and the importance of precise lesion production, both of which determine a successful outcome. The authors use a hemilaminectomy for exposure of the spinal cord in the brachial plexus avulsion because of the spinal cord distortion; in these cases we believe a complete laminectomy is essential for adequate exposure. It is very important to visualize the normal spinal cord adjacent to the trauma, which guides the surgeon in making the DREZ lesions. Although this report does not add any new information on the DREZ operation for these two conditions, the authors do emphasize the importance of careful anatomical location and lesion making using a standard method, which has been proven by past experience. The authors are to be congratulated for reemphasizing the treatment of pain in brachial plexus and intractable facial pain, which until the DREZ operation were untreatable. Blaine S. Nashold, Jr. Durham, North Carolina

NEUROSURGERY

N

euromodulation using neurostimulation, especially spinal cord stimulation, because of its conservative nature, has to be the first option for treating chronic pain. However, spinal core stimulation was revealed to be ineffective for (nociceptive) cancer pain, even topographically limited pain or for neuropathic pain when the corresponding dorsal column or lemniscal fibers are not functionally valid. The latter can be checked by measuring the central conduction time of the somatosensory evoked potentials explored at the level of the pain territory (9). In some of these patients, ablative neurosurgery, namely DREZ lesioning, can be effective when pain generators are located in the dorsal horn or the trigeminal nucleus. In the 1970s, we introduced the concept of surgery in the DREZ (7, 10), first for topographically limited cancer pain in 1972, then for painful spasticity in 1973, and then for peripheral deafferentation pain, specifically, after brachial avulsion, in 1974. We used microsurgical lesioning, under magnified vision, by microcoagulations delivered with a microbipolar forceps after opening of the dorsolateral sulcus. Since then Nashold and Ostdahl (4) and Nashold et al. (5) developed and popularized DREZ lesioning using a radiofrequency electrode as the lesion maker. Other modes of lesioning were introduced later: the laser beam (3, 6, 11) and a microultrasonic probe (1, 2). The authors of this article are using a combination of two methods: microsurgical DREZotomy to opens the dorsolateral sulcus and the radiofrequency electrode as the lesion maker; their results are good. Whatever the method may be, it is of prime importance to be accurate, so that the lesion does not spread to the dorsal column medially or to the pyramidal tract laterally. Precise knowledge of the internal morphology of the spinal cord is mandatory (8). The axis of the dorsal horn is at a 35-degree angle at the cervical level and at a 45-degree angle at the lumbosacral level. The DREZ target is very narrow at the thoracic level, which makes it very sensitive to trespassing of its limits, which increases the risk of deficits. I agree with the authors that best indications for DREZ lesioning are neuropathic pain after avulsion, segmental pain after spinal cord injury, and some topographically limited cancer pain. Regarding trigeminal nucleus lesioning, I do not have experience but do think that indications should be very carefully selected. I am rather reluctant to see it applied to so-called “atypical facial pain.” Whatever the indications may be, DREZ lesioning should be performed with an extremely rigorous technique, as Kanpolat and his team do. Marc Sindou Lyon, France

1. Dreval ON: Ultrasonic DREZ-operations for treatment of pain due to brachial plexus avulsion. Acta Neurochir 122:761–781, 1993. 2. Kandel EL, Ogleznev KJA, Dreval ON: Destruction of posterior root entry zone as a method for treating chronic pain in traumatic injury to the brachial plexus. Vopr Neurochir 6:20–27, 1987. 3. Levy WJ, Nutkiewicz A, Ditmore M, Watts C: Laser induced dorsal root entry zone lesions for pain control: Report of three cases. J Neurosurg 59:884–886, 1983. 4. Nashold BS, Ostdahl PH: Dorsal root entry zone lesions for pain relief. J Neurosurg 51:59–69, 1979. 5. Nashold BS, Urban B, Zorub DS: Phantom pain relief by focal destruction of substantia gelatinosa of Rolando, in Bonica JJ, Albe-Fessard D (eds): Advances in Pain Research and Therapy. New York, Raven Press, 1976, vol 1, pp 959–963. 6. Powers SK, Adams JE, Edwards MS, Boggan JE, Hosobuchi Y: Pain relief from dorsal root entry zone lesions made with argon and carbon dioxide microsurgical lasers. J Neurosurg 61:841–847, 1984. 7. Sindou M: Study of the dorsal root entry zone (DREZ): a target for pain surgery. Lyon, 1972 (dissertation).


KANPOLAT ET AL.

8. Sindou M: Dorsal root entry zone lesions, in Burchiel KJ (ed): Surgical Management of Pain. New York, Thieme Medical Publishers, 2002, pp 701–713. 9. Sindou M, Mertens P, Bendavid U, Garcia-Larrea, Mauguiere F: Predictive value of somato-sensory evoked potentials for long-lasting pain relief after spinal cord stimulation: Practical use for patients selection. Neurosurgery 52:1374–1384, 2003. 10. Sindou M, Quoex C, Baleydier C: Fiber organization at the posterior spinal cord-rootlet junction in man. J Comp Neurol 153:15–26, 1974. 11. Young RF: Clinical experience with radio-frequency and laser DREZ-lesions. J Neurosurg 72:715–720, 1990.

K

anpolat et al. describe their institutional experience with spinal and caudalis DREZ procedures over a 20-year period. This group has championed the use of image-guided ablative procedures for the treatment of chronic pain and has pioneered the use of computed tomography during these procedures to improve localization of the lesions. This particular study, however, outlines their experience and observations on the performance of DREZ procedures in the cervical spine and lower brainstem for a variety of deafferentation syndromes. The article includes a fairly exhaustive review of the literature, as well as a detailed description of their procedural “pearls” in performing these operations. As neurosurgical trainees increasingly lack proper exposure to these important (although admittedly unusual) procedures, it is important that experienced groups such as this one document their technique and outcomes. And although the authors are to be commended for the detailed description of their technique, I do not think the reporting of results in this study allows for proper description of the outcomes. In particular, the use of a Karnofsky score a mere 24 hours after the performance of a major surgical procedure is not terribly meaningful, as this scale is meant to assess the patient’s function in everyday life. The use of a single Visual Analog Scale score at the 24-hour mark is also not terribly helpful, and although the authors describe 1-year outcomes as satisfactory or nonsatisfactory, the assessment of pain relief in the long term should ideally involve a functional or quality-of-life measure as well as a more granular pain relief scale. Oren Sagher Ann Arbor, Michigan

T

he application of DREZotomy was explored in depth by Nashold, and interested students should read his definitive text on the subject. In my practice, which focuses on peripheral nerve and pain surgery, patients with regional neuropathic pain syndromes present commonly. I have reserved DREZotomy for patients with sustained neuropathic pain for which medical and neuromodulation approaches fail. Neuromodulation approaches for appendicular pain syndromes include spinal cord stimulation, peripheral nerve stimulation, and intrathecal delivery of agents with antineuropathic pain activity including bupivacaine, clonidine, and ziconotide (Prialt). Neuromodulation for facial pain can include supra- and infraorbital nerve stimulation, trigeminal ganglion, and root stimulation. Motor cortex stimulation and deep brain stimulation can also be considered as alternatives to either DREZotomy or caudalis DREZ lesioning. We offer neuromodulation ahead of ablative approaches because the potential complications are far less severe. Because of its durability and exceptional efficacy, nerve root avulsion is the one exception to the use of DREZ lesioning as a third-line therapy. In this scenario, it is often extremely difficult to achieve adequate paresthesia coverage of the painful region with spinal cord stimulation for two reasons. First, the same functional alterations in dorsal horn activity that explain the basis for efficacy of the DREZ procedure are


likely to change the somatotopic organization on which we depend to deliver stimulation to specific sensory distributions. Second, the epidural space is often extremely difficult to navigate as a result of the scarring that ensues after avulsion injuries. Longer linear arrays of paddle leads can be used as a prelude to DREZ procedures. That is, one can place the electrode through the same multilevel hemilaminectomy that can later be used for DREZotomy, if the trial of stimulation fails. Because avulsion pain is almost always unilateral, hemilaminectomy provides an excellent means to place these electrodes. In my experience, some patients will accommodate to brachial plexus avulsion pain and others will experience complete or partial spontaneous resolution. For this reason, I will not perform DREZotomy within the first year of avulsion. We have favored hemilaminectomy in most of our patients, because, as pointed out by the authors, one should visualize the intact dorsal rootlets above and below the avulsion as the main geographic landmark. Thus, laminectomy must be performed over many segments regardless of which segments the DREZotomy will be directed toward. Because most of these patients are relatively young, we have been concerned about the long-term risks of disrupting the posterior tension band of the cervical spine. It is tempting, once the full length of avulsion is exposed, to perform DREZ lesions for the full length of the avulsed area. However, neuropathic pain is not always mapped exactly to the full avulsed area. Thus, in the name of minimizing the potential for complications, we tend to perform lesioning only at the segmental levels that correlate with refractory pain. It is critical to grasp two potential pitfalls in DREZotomy for avulsion. The first is the fact that the very common pseudomeningoceles that occur with avulsions often mean that cerebrospinal fluid will begin leaking almost immediately during the laminectomy. This scenario can be challenging because the dural rent is not immediately accessible and may be quite difficult to repair. This repair can often be performed after the dura is opened from the inside. Second, as pointed out briefly by the authors, avulsion often results in rotation of the spinal cord. The contralateral side remains tethered by roots. The avulsed side of the cord may lift up, causing rotation. Because the difference between successful pain relief and long track injury depends entirely on the coronal trajectory of the electrode with respect to the cord, rotation must be taken into account. We prefer preoperative myelograms to assess this rotation as well as the extent and location of pseudomeningoceles. In addition, we have used somatosensory evoked potentials during DREZotomy as a theoretical means to alert us to dorsal column injury during lesioning. Having said this, I note that somatosensory evoked potentials were entirely stable throughout the one operation in which the patient developed a lasting proprioceptive deficit, questioning the utility of this approach. The present article by Kanpolat et al. is a welcomed addition to the neurosurgical pain literature. The implementation of this procedure is now rare. Consequently, few residents are trained in this approach, although it remains an indispensable part of our armamentarium in the treatment of pain. Neuromodulatory solutions are geared to provide patients with a tool to reduce their pain by greater than 50%. In contrast, successful DREZotomy creates a durable cure of avulsion pain with none of the disadvantages of implanted prosthetics. These are the patients who seem to want to name their progeny after me (an action which I usually discourage). Nicholas M. Boulis Atlanta, Georgia


FUNCTIONAL AND STEREOTACTIC New Technology

USE OF AN INTEGRATED PLATFORM SYSTEM IN THE PLACEMENT OF DEEP BRAIN STIMULATORS Gregory G. Heuer, M.D., Ph.D. Center for Functional and Restorative Neurosurgery, Penn Neurological Institute, Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania

Kareem A. Zaghloul, M.D., Ph.D. Center for Functional and Restorative Neurosurgery, Penn Neurological Institute, Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania

Jurg L. Jaggi, Ph.D. Center for Functional and Restorative Neurosurgery, Penn Neurological Institute, Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania

Gordon H. Baltuch, M.D., Ph.D. Center for Functional and Restorative Neurosurgery, Penn Neurological Institute, Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania Reprint requests: Gordon H. Baltuch, M.D., Ph.D., Department of Neurosurgery, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA, 19104. Email: [email protected] Received, October 17, 2006. Accepted, May 31, 2007.

THE PLACEMENT OF deep brain stimulator leads requires a great deal of technology and equipment. We describe our 25-month experience with an integrated platform system, the StimPilot (Medtronic Inc., Minneapolis, MN), for the placement of deep brain stimulator leads. The platform consists of a neuronavigation station, microdrive control, and microelectrode recording display and control. This platform is run from a laptop-sized portable control unit. The unit was used in 147 patients for the placement of 262 leads. Leads were placed into the subthalamic nucleus, ventral intermediate nucleus, globus pallidus interna, and anterior thalamic nucleus. One patient required replacement of one lead during this time frame, with successful reimplantation. No system failures occurred. KEY WORDS: Deep brain stimulator, Microdrive, Microelectrode recordings, Neuronavigation, Platform Neurosurgery 62[ONS Suppl 1]:ONS245–ONS248, 2008

T

he placement of deep brain stimulator (DBS) leads has been shown to be a safe and effective treatment for movement disorders (2, 5, 9, 10). Surgical placement requires a great deal of technical equipment to ensure the correct location (12). Current techniques involve a preoperative magnetic resonance imaging (MRI) scan for target determination (5, 11). This MRI scan is often loaded onto an image guidance workstation platform. Next, the patient has a microelectrode passed into the region of interest, guided by the microelectrode recordings and stereotactic coordinates. For precision, passing of the electrode is often controlled with a microdrive unit. When the region of interest has been determined, a DBS lead is then passed down this tract to the target. Traditionally, placement of the DBS leads required several separate devices, including a neuronavigation station and platform, a microrecording device, and a microdrive device. With recent advances in computer technology, it is now possible to integrate these devices into one system to make this device portable.

DESCRIPTION OF THE SYSTEM The platform-based system, commercially known as StimPilot (Medtronic Inc., Minneapolis, MN), consists of a portable laptop-sized instrument (Fig. 1). The operating system has neuronavigation software, similar to the

NEUROSURGERY

DOI: 10.1227/01.NEU.0000297009.43093.77

FIGURE 1. Intraoperative setup of the platform system for deep brain stimulator lead placement.

StealthStation (Medtronic Inc.) software frequently used in volumetric tumor resections (1, 3, 4, 6). This software allows a three-dimensional MRI, computed tomographic, or MRIcomputed tomographic fused scan to be displayed along with the preplanned stereotactic pathway and has been used for the placement of DBS leads. The software allows the Leksell (Elekta AB, Stockholm, Sweden) frame coordinates and fiducial markers to be loaded into the system. A Schaltenbrand-Wahren atlas can be projected on the loaded image (8) and can be used for indirect targeting of the coordinates of the region of interest (5, 9, 11). If the


HEUER ET AL.

FIGURE 2. Platform display. A, neuronavigation display. Image demonstrating the proposed path (blue cylinder at the bottom right) with the location of vascular structures indicated in purple. B, screen capture from the platform system. Microelectrode recordings are at the top of the display in

region of interest can be visualized, it can be directly targeted on the loaded image. The surgeon then sets the entry point on the scan or inputs the stereotactic coordinates for the entry point on the basis of the frame settings. Once the entry point and target are determined, the platform calculates the path and projects it onto the three-dimensional image (Fig. 2A). The target region is then explored physiologically by obtaining microelectrode recordings (Fig. 2B). In all cases, a tungsten single-ended microelectrode with impedance of 1 MΩ at 1 KHz was used (FHC, Inc., Bowdoinham, ME). The electrode was passed with a microdrive controlled directly by the platform system. Tracings from microelectrode recordings can be displayed on the platform, and the system allows these recordings to be modulated on the workstation. The microelectrode tracks can be superimposed on the projected Schaltenbrand-Wahren atlas and displayed on the workstation along with the real-time projected location along the tract from which the recordings are being taken. Additionally, snapshots of recordings can be saved and displayed on the workstation beside the track registered to the stereotactic coordinate at which they were obtained. The platform system is limited to saving 20-second epochs for data extraction for external research analyses. If continuous recordings are needed, a parallel system must be used to collect the data (7).

CLINICAL APPLICATION The integrated platform system was used for placement of all DBS leads for a period of 25 months. A total of 147 patients were treated with the placement of 262 leads (Table 1). The DBS leads were placed for a number of conditions, including Parkinson’s disease, dystonia, essential tremor, and epilepsy.


green. In the neuronavigation display in the center, the estimated location along track is indicated by a red dot. The microdrive display is at the bottom right. Specific saved microelectrode tracings are shown at the lower left of the display.

TABLE 1. Characteristics and location of deep brain stimulator implants No. Total patients

147

Total implants

262

Sides

Bilateral

115

Right

17

Left

15

Location

Subthalamic nucleus

224

Ventral intermediate nucleus

17

Globus pallidus interna

15

Anterior thalamic nucleus

6

A total of 224 leads were placed into the subthalamic nucleus, 17 into the ventral intermediate nucleus, 15 into the globus pallidus interna, and six into the anterior thalamic nucleus. One patient required replacement of one lead. This patient’s lead was noted to be intraventricular on postoperative imaging. The misplacement was not attributed to the platform system. The patient underwent replacement of the lead the next day without incident. No system failures occurred.

DISCUSSION Proper placement of DBS leads requires a great deal of technological equipment. We feel that the integration of the various


INTEGRATED PLATFORM FOR DEEP BRAIN STIMULATION

technologies into one system provides some advantages over the use of separate units for each integrated component. The components of the platform system have been available previously. For example, software that displays microelectrode recording information, such as Leadpoint (Medtronic Inc.), has been available. Using this software, depth information is displayed along with microelectrode recordings. Neuronavigation software packages, such as StealthStation, that provide a three-dimensional display along with electrode tract information are also available. However, because of cabling issues with the serial interface, it was difficult to use these two technologies at the same time. Specifically, depth information could be displayed on one device but not on both devices simultaneously. On the other hand, the platform system has a display that includes both microelectrode recording information and the proposed location along the implant path on the neuronavigation MRI scan. The ability to coordinate the information provides the surgeon with information in a more user-friendly format. In addition, the amount and size of the hardware needed to provide all component systems can be prohibitive, particularly in a small operating room. The integrated system described is laptop-sized, portable, and self-contained, and little time is required for setup and removal. The system has a small footprint and does not require a specially designed operating room. Also, with this type of system, less equipment is needed overall, with the potential to reduce total capital costs. Currently, the system is available from the manufacturer on a fee-per-case basis. Although this practice may provide a cost advantage to smaller centers, it may increase costs in larger centers with a higher volume of DBS placements. The portability of the device allows it to be transported to and from the office setting and the operating room. This portability allows the surgeon to perform preoperative analysis of neuroimaging studies to facilitate surgical planning. Also, increased portability provides the opportunity for postoperative analysis that may provide useful research information without the need to disrupt the operating room. The system has some limitations. It does not contain a microstimulator device, which can be necessary for DBS placement. Additionally, as stated previously, the system allows only storage of 20-second epochs. If continuous recordings are required, a parallel system must be attached. The integrated platform-based system described in this article is an example of how technologies can be combined to provide the surgeon with a device that has advantages over the separate individual units. As new technologies are developed, systems that can be adapted and integrated will be required to ensure that these technologies are used to their utmost potential.

REFERENCES 1. Barnett GH, Miller DW, Weisenberger J: Frameless stereotaxy with scalpapplied fiducial markers for brain biopsy procedures: Experience in 218 cases. J Neurosurg 91:569–576, 1999.

NEUROSURGERY

2. Deep-Brain Stimulation for Parkinson’s Disease Study Group: Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson’s disease. N Engl J Med 345:956–963, 2001. 3. Golfinos JG, Fitzpatrick BC, Smith LR, Spetzler RF: Clinical use of a frameless stereotactic arm: Results of 325 cases. J Neurosurg 83:197–205, 1995. 4. Haberland N, Ebmeier K, Hliscs R, Grnewald JP, Silbermann J, Steenbeck J, Nowak H, Kalff R: Neuronavigation in surgery of intracranial and spinal tumors. J Cancer Res Clin Oncol 126:529–541, 2000. 5. Jaggi JL, Umemura A, Hurtig HI, Siderowf AD, Colcher A, Stern MB, Baltuch GH: Bilateral stimulation of the subthalamic nucleus in Parkinson’s disease: Surgical efficacy and prediction of outcome. Stereotact Funct Neurosurg 82:104–114, 2004. 6. Lee JY, Lunsford LD, Subach BR, Jho HD, Bissonette DJ, Kondziolka D: Brain surgery with image guidance: Current recommendations based on a 20-year assessment. Stereotact Funct Neurosurg 75:35–48, 2000. 7. Moyer JT, Danish SF, Keating JG, Finkel LH, Baltuch GH, Jaggi JL: Implementation of dual simultaneous microelectrode recording systems during deep brain stimulation surgery for Parkinson’s disease: Technical note. Neurosurgery 60 [Suppl 2]:ONSE177–ONSE178, 2007. 8. Schaltenbrand G, Wahren W: Atlas for Stereotaxy of the Human Brain. Stuttgart, Georg Thieme, 1977. 9. Simuni T, Jaggi JL, Mulholland H, Hurtig HI, Colcher A, Siderowf AD, Ravina B, Skolnick BE, Goldstein R, Stern MB, Baltuch GH: Bilateral stimulation of the subthalamic nucleus in patients with Parkinson disease: A study of efficacy and safety. J Neurosurg 96:666–672, 2002. 10. Starr PA, Vitek JL, Bakay RA: Deep brain stimulation for movement disorders. Neurosurg Clin N Am 9:381–402, 1998. 11. Starr PA, Vitek JL, DeLong M, Bakay RA: Magnetic resonance imaging-based stereotactic localization of the globus pallidus and subthalamic nucleus. Neurosurgery 44:303–314, 1999. 12. Tasker RR: Surgical treatment of the dyskinesias, in Schmidek HH, Roberts DW (eds): Schmidek and Sweet Operative Neurosurgical Techniques. Marion, W.B. Saunders Co., 2000, pp 1601–1624.

COMMENTS

M

icroelectrode-guided deep brain stimulation is equipment intensive. If there is a large microelectrode recording system, large workstation for stereotactic target and trajectory planning, and fluoroscopy, the room is crowded. The authors provide a descriptive summary of the Medtronic StimPilot system that integrates physiological recording and magnetic resonance imaging or computed tomography-based target and trajectory planning. The idea for this system is appealing. We have not used this system for several reasons: 1) Its availability only as a fee-per-case model is not attractive to high-volume centers. 2) The system lacks a microstimulation circuit, which is useful for certain physiological applications (microstimulation-induced visual or sensory phenomena in globus pallidus internus or thalamic mapping, respectively). In addition, microstimulation is useful to “condition” the tips of many types of microelectrodes to bring them into the optimal impedance range. 3) The system is not particularly suited to storage of longer contiguous segments of microelectrode data, which may be of use for physiological research. Philip A. Starr San Francisco, California

T

he report describes the first large case series from an experienced deep brain stimulation center using the integrated recording and image analysis platform from Medtronic. The two components of this device have been available for many years, and this report indicates that the integrated device appears to function as well as the separate components. For centers unwilling to provide capital outlays to purchase devices or for centers that are infrequent users of deep brain


HEUER ET AL.

stimulation, the portable and integrated device used on a fee-for-service basis will probably be attractive. As any user of mobile phones knows, any attempt to integrate too many activities often detracts from some functions for the high-end user. Electrophysiology research will probably still benefit from stand-alone microrecording devices, and the ability to push and manipulate images through a network to a device owned by the center will also have advantages. Yet, for some practitioners, the confidence gained by this report will probably facilitate their incorporation of this technology. Michael G. Kaplitt New York, New York

T

he authors describe their experience with the Medtronic StimPoint system in 147 patients over 25 months. This system is proposed as an improvement over traditional systems in that it combines planning and microelectrode recording into one portable system. There are several advantages. First, all of the components are manufactured by the same company and are therefore compatible. Second, the platform is more compact than systems with separate components, allowing for portability and less equipment in the operating room. Third, it is available to lease and thus allows an opportunity for smaller centers to have a reasonable system without significant overhead. For larger centers and centers with a research focus, however, the system may not be cost-effective or as appropriate. The microelectrode recording capabilities are less than those with stand-alone devices, microstimulation is not available, and data analysis/collection is limited. The idea of an integrated system and the ability to use this system on a lease basis makes business sense on the small scale. These features make the system attractive to hospitals and subsequently may make the procedures more widely available in smaller centers. It remains important, however, for these centers and operating neurosurgeons to

have not only the intraoperative ability to care for these patients but also the perioperative support necessary for successful outcomes. Careful patient selection on the basis of established criteria (2) and vigorous optimization of stimulation parameters are crucial to success (1). Furthermore, when procedures in patients that were deemed treatment failures are evaluated, many failures were the result of inadequate perioperative management (3). Twelve percent of patients had inadequate medication trials and 12% showed significant cognitive decline. In 51% of patients, good outcomes were ultimately seen after intensive reprogramming and medication adjustments. Patients need to have access to a clinician experienced in this perioperative management to ultimately benefit from the procedure as the neurosurgeon may not have the time, experience, or ancillary support to effectively oversee postoperative programming and medication adjustments. Regardless of whether the program is small or large, the procedure requires a high level of expertise to be performed optimally. Julie G. Pilitsis Roy A.E. Bakay Chicago, Illinois

1. Deuschl G, Herzog J, Kleiner-Fisman G, Kubu C, Lozano AM, Lyons KE, Rodriguez-Oroz MC, Tamma F, Troster AI, Vitek JL, Volkmann J, Voon V: Deep brain stimulation: Postoperative issues. Mov Disord 21:S219–S237, 2006. 2. Lang AE, Houeto JL, Krack P, Kubu C, Lyons KE, Moro E, Ondo W, Pahwa R, Poewe W, Tröster AI, Uitti R, Voon V: Deep brain stimulation: preoperative issues. Mov Disord 21:S171–S196, 2006. 3. Okun MS, Tagliati M, Pourfar M, Fernandez HH, Rodriguez RL, Alterman RL, Foote KD: Management of referred deep brain stimulation failures: A retrospective analysis from 2 movement disorders centers. Arch Neurol 62:1250– 1255, 2005.

Anatomy Lecture by Dr. Sebastiaan Egbertz, (1603), Aert Pietersz. Amsterdam, Rijksmuseum. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.



FUNCTIONAL AND STEREOTACTIC Technical Case Reports

METHODS OF SCALP REVISION FOR DEEP BRAIN STIMULATOR HARDWARE: CASE REPORT Alejandro M. Spiotta, M.D. Department of Neurological Surgery, Cleveland Clinic, Cleveland, Ohio

Mark D. Bain, M.D. Department of Neurological Surgery, Cleveland Clinic, Cleveland, Ohio

Milind Deogaonkar, M.D. Department of Neurological Surgery, Center for Neurological Restoration, Cleveland Clinic, Cleveland, Ohio

Warren Hammert, M.D. Department of Plastic Surgery, Cleveland Clinic, Cleveland, Ohio

Armand R. Lucas, M.D. Department of Plastic Surgery, Cleveland Clinic, Cleveland, Ohio

Nicholas M. Boulis, M.D. Department of Neurological Surgery, Center for Neurological Restoration, Cleveland Clinic, Cleveland, Ohio

Ali R. Rezai, M.D. Department of Neurological Surgery, Center for Neurological Restoration, Cleveland Clinic, Cleveland, Ohio Reprint requests: Ali R. Rezai, M.D., Cleveland Clinic, 9500 Euclid Avenue, S-31, Cleveland, OH 44195. Email: [email protected] Received, December 14, 2006. Accepted, September 24, 2007.

NEUROSURGERY

OBJECTIVE: The use of deep brain stimulation (DBS) to treat a variety of disorders has expanded and will result in an increasingly larger number of patients and implanted electrodes. Hardware failure can result from malfunction, lead migration, fracture, and infection. Scalp erosion with exposure of underlying hardware can lead to potential infectious complications and is, in itself, a strong indication for explantation of the neurostimulation system. The patient’s relief of symptoms after DBS will be limited by hardware-related complications and thus, strategies to revise scalp overlying hardware are important in the widespread application of DBS. CLINICAL PRESENTATION: We describe strategies to address complications related to implanted DBS neurostimulator hardware specifically designed to address breach of the integrity of the scalp over the burr hole site. The aim of these approaches is to treat scalp erosion to allow for the reimplantation of previously explanted, infected hardware, or to treat thinned scalp with threatened erosion and prevent the need to remove exposed hardware that is otherwise functioning. INTERVENTION: Two different approaches are presented: 1) a temporoparieto-occipital flap based on the superficial temporal artery with or without scalp expansion, and 2) a scalp fasciocutaneous flap with or without cranioplasty. CONCLUSION: Stimulation of various deep brain targets helps patients with a wide range of diseases. In the future, with continued refinement, hardware complications can be minimized. Until then, novel approaches need to be developed to save DBS systems and provide symptomatic relief to patients. KEY WORDS: Complications, Deep brain stimulation, Scalp revision Neurosurgery 62[ONS Suppl 1]:ONSE249–ONSE250, 2008

T

he use of deep brain stimulation (DBS) has revolutionized the treatment of complex neurological disorders such as Parkinson’s disease, dystonia, and essential tremor. Newer applications of DBS are rapidly emerging and are being refined for conditions such as epilepsy, psychiatric disorders, and chronic pain. These expansions of the therapeutic use of DBS will result in an increasing number of patients and implanted electrodes. However, the widespread application of DBS will be accompanied by an increasing number of complications associated with the implanted hardware. Ultimately, a patient’s relief of symptoms after DBS will be limited by hardware-related complications. Hardware failure can result from malfunction, lead migration, fracture, and infection.

DOI: 10.1227/01.NEU.0000297054.43024.34

Scalp erosion, with exposure of underlying hardware, can lead to infection and is, in itself, a strong indication for explantation of the neurostimulation system. Erosion of the overlying scalp occurs at points of increased tension between the skin and hardware. The reported incidence for scalp erosion is between 1.4 and 8.3% (1, 2, 5, 6). Factors that likely predispose a patient to scalp erosion include scalp thickness and the use of bulky hardware that has sharp edges. Measures to prevent scalp erosion are vital in decreasing morbidity for patients with implanted neurostimulation systems. We present different strategies that incorporate the expertise of neurological and plastic surgery to address complications related to implanted DBS neurostimulator hardware that are specifically designed to address breach of


SPIOTTA ET AL.

the integrity of the scalp at the burr hole site. The aim of these approaches is to treat scalp erosion to allow for the reimplantation of previously explanted infected hardware or to treat thinned scalp with threatened erosion and prevent the need to remove exposed hardware that is otherwise functioning. Two different approaches are discussed by way of case illustrations: 1) a temporoparieto-occipital (TPO) flap that is based on the superficial temporal artery with or without scalp expansion, and 2) a scalp fasciocutaneous flap with or without cranioplasty.

A

B

Illustrative Cases

Patient 1 A 54-year-old woman with Parkinson’s disease underwent bilateral subthalamic nucleus electrode implantation. She experienced relief of symptoms of bradykinesia and shuffling gait. A Navigus burr hole cap (Image Guided Neurologics, Inc., Melbourne, FL) was used to cover the burr hole and secure the electrode. She presented 1 year later, after minor local trauma to the right burr hole cap region resulted in a scalp wound infection. Upon examination, the patient was found to have exposed hardware. She underwent removal of the intracranial lead and internal pulse generator. A large scalp defect remained after debridement of the infected scalp. The scalp was widely undermined at the edges of the defect, and the wound was closed primarily. The scalp overlying resulting the frontoparietal region was thinned and under tension. To provide the patient with the highest likelihood of success for a future implant, a scalp reconstruction was deemed necessary. Intervention. A TPO flap was planned. Jose Juri, M.D., conceptualized and refined this flap over the past 25 years to treat male pattern baldness (3). The flap was designed with a base measuring approximately 3 to 4 cm on the superficial temporal artery (STA) at an angle of 30 to 40 degrees from the orbitomeatal plane (Fig. 1A). The planned flap spanned the entire parietal zone and descended to the occipital region without crossing the midline. The advent of tissue expansion in scalp remodeling allows for larger lengths of TPO flap to be created. Thus, larger defects can be closed while allowing the donor site to be closed with minimal tension. A horseshoe-shaped scalp expander was inserted once the tissue to be used to construct the TPO flap was mapped out. An incision was made over the upper aspect of the flap and dissection was carried out to the galea and superficial temporal fascia. The flap was elevated to the level of the auricular cartilage and posteriorly to the level of the occipital artery. An expander measuring 10 cm in length and 5 cm in width with a volume capacity of 175 mL was placed in the subgaleal space and the port inset inferior to the mastoid process. The port was sutured into position with interrupted 3-0 Vicryl (Ethicon, Inc., Somerville, NJ) to prevent twisting or turning. After this was completed, the scalp incision was approximated with interrupted 3-0 Vicryl for the galea and a running interlocking 3-0 Prolene (Ethicon, Inc.) suture for the skin. The scalp expander was activated over a 2-month period with normal saline to an adequate volume to provide additional expanded scalp (Fig. 2). Transposing the flap was a two-delay procedure performed at 1-week intervals followed by the rotation of the flap. The flap length was up to 25 to 30 cm, as the technique allowed the rotation of tissue previously supplied by the occipital artery upon an STA base. Both delays were performed under local anesthesia. At the first procedure, an incision was made through the skin, subcutaneous tissue, and galea for the first twothirds length of the flap from its base. No undermining was done during this delay. In this manner, an area of unincised galea remained and provided perfusion from the ipsilateral occipital artery (Fig. 1B).


FIGURE 1. A, temporoparieto- C occipital (TPO) flap is designed with its base measuring approximately 2.5 to 4 cm on the superficial temporal artery (STA) at an angle of 30 to 40 degrees from the orbitomeatal plane. B, at the first delay, an incision is carried through skin, subcutaneous tissue, and galea for the first two-thirds length of the flap from its base. No undermining is done during this delay. An area of unincised galea remains and provides perfusion from the ipsilateral occipital artery. C, at the final stage, rotation of the TPO flap is performed 1 week after the second delay to allow for adequate perfusion and viability of the flap. The flap is elevated under the galea in a distal to proximal fashion to its desired location over the frontal convexity. One week later, the distal third of the flap was incised, raised, and undermined superficial to the occipitalis muscle and then returned to its position, thereby isolating it from the occipital artery. At this point, the flap was entirely dependent on perfusion from the STA. The entire length of the flap was closed with nonabsorbable suture. FIGURE 2. An expander measurThe final stage, rotation of the ing 10 cm in length and 5 cm in TPO flap, was performed 1 week width with a volume capacity of up after the second delay under gento 175 mL was placed in the suberal anesthesia. The TPO flap was galeal space with the port inset measured on the basis of the inferior to the mastoid process. The superficial temporal vessels to be port was sutured into position with approximately 4 cm in width and interrupted 3-0 Vicryl to prevent 21 cm in length. An incision was twisting or turning. A tissue scalp made, the previous scalp exexpander was inserted and pander was removed, and the enlarged over 1 to 2 months to a flap was outlined. The superior volume sufficient to provide aspect of the flap was incised enough length on the TPO flap. down to the level of the occipital vessel, and the occipital vessel was ligated. The flap was then raised in a distal to proximal fashion based on superficial temporal vessels, and transposed onto the previous craniotomy defect (Fig. 1C). The donor site was undermined to the level of the angle of the mandible and posterior mid-neck, and the donor site was closed with interrupted 2-0


SCALP REVISION AFTER DEEP BRAIN STIMULATION

Vicryl staples and 3-0 Prolene sutures. The flap was then transposed, the defect was recreated, and the flap was inset with interrupted 4-0 Prolene and 3-0 Vicryl. This flap provided adequate scalp for reimplantation of a neurostimulation system (Fig. 3, A and B). After the remodeled scalp had healed, the patient underwent successful replacement of the right-sided DBS electrode and fixation with a Navigus burr hole cap. She continues to experience relief of symptoms from her Parkinson’s disease.

Patient 2 A 65-year-old Caucasian man presented with medically refractory and disabling tremor. He underwent stereotactic implantation of a DBS electrode in the left ventral intermediate nucleus of the thalamus. The neurostimulation system used was the Activa system (Medtronic, Inc., Minneapolis, MN). The electrode was fixed to the cranium by use of the Medtronic burr hole ring and cap. The patient had excellent relief of the tremor in his right hand after implantation of a deep brain stimulator in the left ventral intermediate nucleus. Five years later, however, he presented with scalp thinning and threatened scalp erosion over the burr hole cap located over the left posterior frontal calvarium. The patient consented to undergo a scalp fasciocutaneous flap and cranioplasty with recontouring of the prominent DBS electrode to prevent subsequent scalp erosion. Intervention. The patient was taken to the operating room and placed in the supine position, and general anesthesia was induced. The scalp was prepared and draped, and local anesthesia was infiltrated into the planned incision, which incorporated the previous incision with extensions both anteriorly and posteriorly as well as medially and laterally at the end. Dissection was made deep down to the electrode to the level of the periosteum. The scalp was elevated around the burr hole cap. The electrodes were identified and protected throughout the entire procedure. Bipolar cautery was used for hemostasis. The periosteum and scar tissue were undermined circumferentially around the burr hole cap. Three Leibinger 1.7-mm-diameter screws (Howmedica Leibinger, Inc., Carrollton, TX) were secured into the outer table of the cranium less than 1 cm lateral to the burr hole cap to provide purchase for the methyl methacrylate to be used in the cranioplasty (Fig. 4). Methyl methacrylate was fashioned to smooth the contour of the burr hole cap and eliminate the abrupt step-off, which was causing the scalp thinning (Fig. 5). The scalp flap was advanced anteriorly and trimmed so that a thicker region of scalp was now overlying the burr

FIGURE 3. Scalp remodeling after the TPO flap provides scalp of sufficient thickness to permit successful reimplantation of the deep brain stimulation electrode and hardware.

NEUROSURGERY

FIGURE 4. The periosteum and scar tissue were undermined extensively in a circumferential pattern around the burr hole cap. Three Leibinger 1.7-mm-diameter screws were secured into the outer table of the cranium, lateral to the burr hole cap, to provide purchase for the methyl methacrylate to be used in the cranioplasty.

hole cap. The flap was then closed in three layers: 4-0 Vicryl was used to close the periosteum, 3-0 Vicryl was used for the deep galeal sutures, and 3-0 Prolene was used for a running cutaneous suture. Postoperatively, the patient did well and was discharged home in stable condition after overnight observation. At the time of discharge, he had minimal incisional pain and intact wounds. At a 2-week follow-up visit, his wounds were healing well, with a healthy flap of tissue covering his Medtronic cap (Fig. 6). His tremor control in the right upper extremity was unchanged.

CONCLUSION Stimulations of various deep brain targets allow neurosurgeons and neurologists to help patients with a wide range of diseases. With the increasing number of patients undergoing DBS procedures, new and unfortunate complications are being encountered. FIGURE 5. Methyl methacrylate Some of the most feared comwas fashioned to smooth the conplications involve lead fracture tour of the burr hole cap. or hardware infection resulting in the need to remove the brain electrode (4, 8). As previously reported, removal of brain electrodes alone can carry significant risk of intracerebral hemorrhage (7). In a review of our own series, we have found a 4.5% risk of hemorrhage associated with lead FIGURE 6. Postoperative photoremoval (unpublished data). graph of the patient demonstrating In addition, the patient is a thicker scalp overlying the burr exposed to all the known risks hole cap in the frontal cranium, associated with reimplantation thereby reducing the profile of the of a DBS system, including hardware. risks of sedation, infection, and intracranial hemorrhage. In this report, we present two strategies used to counteract the burr hole hardware-related complications of DBS. Scalp erosion over neurostimulation hardware results in lead exposure, infection, and, ultimately, lead removal. In the first illustration, a TPO flap was used to provide healthy scalp over the frontal region, thereby allowing reimplantation of a deep brain stimulator in a patient who had a previous, infected system removed. This is a two-delay procedure that involves rotating


SPIOTTA ET AL.

a region of scalp based on a superficial temporary artery base. Scalp expanders provide greater surface area of flap. This is the first report of successful reimplantation of DBS hardware in a patient with a scalp flap. In the second illustration, we present a novel way of prophylactically treating threatened scalp erosion and exposure of the DBS system. This proactive strategy can prevent the need to explant a DBS electrode in a patient who is otherwise enjoying symptom relief. Both techniques presented are relatively safe for the patient and represent a collaborative effort using the expertise of neurological and plastic surgery to salvage a threatened DBS system. In the future, with continued refinement, hardware complications can be minimized. Until then, novel approaches need to be developed to save DBS systems that are providing symptomatic relief to patients with a variety of disorders. Creative procedures such as the two described in this report can reduce hardware-associated morbidity.

REFERENCES 1. Benabid AL, Pollak P, Gervason C, Hoffmann D, Gao DM, Hommel M, Perret JE, de Rougemont J: Long-term suppression of tremor by chronic stimulation of the ventral intermediate thalamic nucleus. Lancet 337:403–406, 1991. 2. Constantoyannis C, Berk C, Honey CR, Mendez I, Brownstone RM: Reducing hardware-related complications of deep brain stimulation. Can J Neurol Sci 32:194–200, 2005. 3. Juri J: Use of parieto-occipital flaps in the surgical treatment of baldness. Plast Reconstr Surg 55:456–460, 1975. 4. Lyons KE, Wilkinson SB, Overman J, Pahwa R: Surgical and hardware complications of subthalamic stimulation: A series of 160 procedures. Neurology 63:612–616, 2004. 5. Medtronic, Inc: Deep Brain Stimulation 3387-89 Lead Kit: Implant Manual. Minneapolis, Medtronic, Inc., 2000. 6. Oh MY, Abosch A, Kim SH, Lang AE, Lozano AM: Long-term hardwarerelated complications of deep brain stimulation. Neurosurgery 50:1268–1274, 2002. 7. Oh MY, Hodaie M, Kim SH, Alkhani A, Lang AE, Lozano AM: Deep brain stimulator electrodes used for lesioning: Proof of principle. Neurosurgery 49:363–369, 2001. 8. Voges J, Waerzeggers Y, Maarouf M, Lehrke R, Koulousakis A, Lenartz D, Sturm V: Deep-brain stimulations: Long-term analysis of complications caused by hardware and surgery—Experiences from a single centre. J Neurol Neurosurg Psychiatry 77:868–872, 2006.

Acknowledgment We are grateful for the work of Maureen Pasuit, B.A., in preparing the artistic illustrations and for the help of Christine Moore, A.A., in preparation of the manuscript.

COMMENTS

T

his is an interesting report of plastic surgery used to treat infected scalp placed over deep brain stimulation (DBS) hardware. The methods, which are precisely described, may help the neurosurgical community and provide a solution, when the infection is such that large amounts of scalp and tissue must be removed, by using skin flaps to cover the exposed material and to close the wound and the scalp defect. This report provides the opportunity to present some comments about the methodology of implanting the hardware: the best treatment is always prevention. The hardware must be reduced in size and have smooth edges to decrease the tension imposed on the scalp as well as smaller volumes,


to reduce scalp erosion. While we are waiting for these improvements, which must be required of manufacturers, some simple precautions can be taken during the procedure to reduce the risk of skin erosion and infection and to avoid having to use extensive surgical procedures such as plastic surgery. The skin incision must never cross the leads or the implanted material. The material from the burr hole to the retroauricular area where the leads become subcutaneous must be placed under the periosteum, therefore preventing the migration of foreign bodies toward the surface of the scalp, as the skin has an emunctory function aimed to expel intruders out of the body. Instead of the burr hole cap, we prefer to fix the electrode by a ligature passed on the rim of the burr hole, using a small, oblique canal, drilled on the edge of the hole and then embedded into dental cement, which provides firm and reliable fixation of the electrode and seals the opening of the skull. When infection involves the intracranial lead at the level of the extracranial part, we have cut the electrode flat at the level of the cement and seal its inner canal with a drop of dental cement. When the electrode has to be reimplanted, the remaining part provides a very reliable basis for retargeting, before it is removed to be replaced by the new electrode. Alim L. Benabid Grenoble, France

C

omplications are the bane of surgical treatments. The study of complications is essential for assessing risk/benefit probabilities and designing techniques to limit their occurrence. Even anecdotal case reports are of value in examining potential solutions. The authors have provided their experience with flap closures after infection and erosion, which is a rare experience. We have not had this experience and to date have successfully treated erosions with subgaleal undermining and plastic rotation closures. Similar to Case 1, we have seen erosion of the Navigus system (Image Guided Neurologies, Inc., Melbourne, FL) only after local trauma. Fortunately, the only requirements were primary closure, antibiotic therapy, and preventing the patient from putting pressure on the wound. The predicament with the Medtronic burr hole cover (Minneapolis, MN) is different. All of our primary burr hole problems (infection and erosions) occurred with this device as opposed to the Navigus system. If there is thinning over an elevated edge of the Medtronic cap, adding cranioplasty bulk and turning a flap would not be our choice. In Case 2, we would have removed the Medtronic cap, secured the lead by one of several low-profile means (cranioplasty of the burr hole or plating the lead), used subgaleal undermining of the scalp for low-tension closure, and used a three-layer (pericranium, galea, and skin) closure over the burr hole. Important tenets for successful management of scalp defects are adequate debridement, establishment of a resilient coverage, and preservation of blood supply. And there are many ways to accomplish these. Serious complications leading to permanent neurological deficits are rare after DBS. However, long-term follow-up demonstrates that hardware complications are relatively common. Some series report that as many as 26% to 30% of patients develop an adverse event, resulting in revision of the implanted hardware (6, 7). Reviews and large long-term follow-up reports suggest lower rates (1, 3, 4, 8–10), and that certainly is our experience. Nevertheless, infections and erosions attributable to DBS hardware can cause considerable morbidity, which may result in prolonged hospital stays and repeated operations. Prevention is the key. Good surgical techniques are the main determinants for avoidance of hardware-related complications. The process starts with planning the scalp incision. Although some will debate whether a straight or curvilin-


SCALP REVISION AFTER DEEP BRAIN STIMULATION

ear incision is best (2, 5), the principle of closure without tension is the real key. The burr hole cover should be low profile and remote from the incision line to lower the risk of erosion and infection. We routinely use a three-layer closure with freeing of the pericranium to oversew the burr hole cover and optimize coverage of this foreign body. In our experience, perioperative antibiotics seem to facilitate low infection rates, especially with foreign bodies. Unfortunately, even after exceptional surgical technique, complications can arise both acutely and later. DBS is a lifetime therapy that requires a lifetime of concern because of the hardware. Refinements in surgical technique, technological improvements, and greater experience with the procedure will probably decrease the incidence of hardware-related complications. The authors are to be congratulated on their efforts to manage these complications. Julie G. Pilitsis Roy A.E. Bakay Chicago, Illinois

5. Kouyialis AT, Boviatsis EJ, Ziaka DS, Sakas DE: Use of a single semilinear incision in deep brain stimulation for movement disorders. Acta Neurochir (Wien) 149:501–504, 2007. 6. Lyons KE, Wilkinson SB, Overman J, Pahwa R: Surgical and hardware complications of subthalamic stimulation: A series of 160 procedures. Neurology 63:612–616, 2004. 7. Paluzzi A, Belli A, Bain P, Liu X, Aziz TM: Operative and hardware complications of deep brain stimulation for movement disorders. Br J Neurosurg 20:290–295, 2006. 8. Rezai AR, Kopell BH, Gross RE, Vitek JL, Sharan AD, Limousin P, Benabid AL: Deep brain stimulation for Parkinson’s disease: Surgical issues. Mov Disord 21 [Suppl 14]:S197–S218, 2006. 9. Seijo FJ, Alvarez-Vega MA, Gutierrez JC, Fdez-Glez F, Lozano B: Complications in subthalamic nucleus stimulation surgery for treatment of Parkinson’s disease: Review of 272 procedures. Acta Neurochir (Wien) 149:867–876, 2007. 10. Voges J, Waerzeggers Y, Maarouf M, Lehrke R, Koulousakis A, Lenartz D, Sturm V: Deep-brain stimulation: Long-term analysis of complications caused by hardware and surgery—Experiences from a single centre. J Neurol Neurosurg Psychiatry 77:868–872, 2006.

I 1. Blomstedt P, Hariz MI: Hardware-related complications of deep brain stimulation: A ten year experience. Acta Neurochir (Wien) 147:1061–1064, 2005. 2. Constantoyannis C, Berk C, Honey CR, Mendez I, Brownstone RM: Reducing hardware-related complications of deep brain stimulation. Can J Neurol Sci 32:194–200, 2005. 3. Hamani C, Lozano AM: Hardware-related complications of deep brain stimulation: A review of the published literature. Stereotact Funct Neurosurg 84:248–251, 2006. 4. Kenney C, Simpson R, Hunter C, Ondo W, Almaguer M, Davidson A, Jankovic J: Short-term and long-term safety of deep brain stimulation in the treatment of movement disorders. J Neurosurg 106:621–625, 2007.

NEUROSURGERY

n this article, Spiotta et al. describe their experience with two plastic surgical techniques, temporoparietal-occipital flap and fasciocutaneous flap, to either salvage implanted DBS hardware that is threatened by scalp thinning or to make feasible reimplantation of DBS hardware after removal of an infected device with significant scalp breakdown. This is a valuable contribution, which delineates effective strategies that may be used by others who face similar circumstances in the future. As the volume of DBS procedures grows, so too will the related complications. The dissemination of successful techniques such as these to deal with those complications is important. Ron L. Alterman New York, New York


GENERAL Concepts and Innovations

QUANTIFICATION OF THE FRONTOTEMPORAL ORBITOZYGOMATIC APPROACH USING A THREE-DIMENSIONAL VISUALIZATION AND MODELING APPLICATION Anthony L. D’Ambrosio, M.D. Department of Neurological Surgery, Columbia University, New York, New York

J Mocco, M.D. Department of Neurological Surgery, Columbia University, New York, New York

Todd C. Hankinson, M.D. Department of Neurological Surgery, Columbia University, New York, New York

Harry R. van Loveren, M.D. Department of Neurological Surgery, University of South Florida, Tampa, Florida

Jeffrey N. Bruce, M.D. Department of Neurological Surgery, Columbia University, New York, New York Reprint requests: Anthony L. D’Ambrosio, M.D., Department of Neurological Surgery, Columbia University, 710 West 168th Street, Neurological Institute, 4th Floor, New York, NY 10032. Email: [email protected] Received, November 27, 2006. Accepted, May 31, 2007.

NEUROSURGERY

OBJECTIVE: We sought to simulate the frontotemporal orbitozygomatic (FTOZ) craniotomy in a three-dimensional virtual environment on patient-specific data and to quantify the exposure afforded by the FTOZ while simulating controlled amounts of brain retraction. METHODS: Four computed tomographic angiograms were reconstructed with commercially available software (Amira 4.1.1; Mercury Computer Systems, Inc., Chelmsford, MA), and virtual FTOZ craniotomies were performed bilaterally (n ⫽ 8). Brain retraction was simulated at 1 and 2 cm. Surgical freedom and projection angle were measured and compared at each stage of the FTOZ. RESULTS: At 1 cm of retraction, surgical freedom increased by 27 ⫾ 14% for the removal of the orbital rim and by 31 ⫾ 18% for FTOZ (P ⬍ 0.01) when compared with frontotemporal (FT) craniotomy. At 2 cm of retraction, surgical freedom increased by 15 ⫾ 5% and 26 ⫾ 8% for the removal of the orbital rim and FTOZ, respectively (P ⬍ 0.01). With increased retraction, surgical freedom increased by 100 ⫾ 26%, 81 ⫾ 15%, and 82 ⫾ 27% for the FT, removal of the orbital rim, and FTOZ craniotomies, respectively (P ⬍ 0.001). Projection angle increased by 24.2% when orbital rim removal was added to the FT craniotomy (P ⬍ 0.01). CONCLUSION: Surgical freedom increases significantly at every step of the FTOZ craniotomy. This effect is less robust when brain retraction is increased. Brain retraction alone has a greater impact on surgical freedom than bone removal alone. Projection angle is significantly increased when orbital rim removal is added to the FT craniotomy. This model overcomes two major limitations of cadaver-based models: quantification of brain retraction and incorporation of patient-specific anatomy. KEY WORDS: Approach, Cranial base, Quantification, Skull base, Surgical simulation, Three-dimensional, Validation Neurosurgery 62[ONS Suppl 1]:ONS251–ONS261, 2008

D

uring the past decade, more attention has been paid to quantifying the additional exposure gained through various cranial base techniques compared with more traditional surgical approaches. Such research efforts are clinically relevant because cranial base techniques have the potential to lengthen operative time and increase morbidity rates and should, therefore, be implemented only when affording true operative benefit. To date, almost every accepted cranial base approach has been objectively quantified to some degree (1, 2, 4–11, 13–18).

DOI: 10.1227/01.NEU.0000297018.65963.9B

Cranial base approach assessment models have been developed in microsurgical cadaver laboratories that use traditional measuring tools (8, 9, 18), frameless navigation systems (1, 10, 11, 13, 15, 17), digitizing probes (4, 6, 7, 14), and robotic microscopes (4, 6, 7, 14) to objectively quantify surgical exposure. Although an invaluable asset to learning microsurgical techniques and understanding complex three-dimensional (3-D) anatomy, fixed human cadaver tissue is limited in its ability to accurately simulate operative exposures because of variations in brain atrophy and tis-


D’AMBROSIO ET AL.

sue fixation (4, 7, 13). As a result, most approach assessment models “maximally retract” the brain and maintain a fixed amount of retraction throughout data acquisition. Despite the obvious effects that variable amounts of brain retraction can have on operative exposure, cadaver-based assessment models have yet to standardize or quantify these effects. Another inherent limitation to studying cadaver specimens is the inability to incorporate clinically relevant intracranial pathology. Current models are unable to account for the potential effects pathological lesions might have on patient-specific anatomy. As a result, cadaver-based models are limited to quantifying surgical access to anatomically normal targets or general regions of the cranial base. Two potentially useful additions to this field of study are standardization of brain retraction and the incorporation of patient-specific anatomy and pathology. In the current study, a cranial base approach simulation and objective quantification model capable of performing these operations are presented. This model simulates various cranial base approach techniques on patient-specific 3-D image reconstructions and quantifies the potential benefits of these maneuvers in an objective and reproducible manner. Methodology for anatomic structure segmentation, 3-D image reconstruction and visualization, surgical approach simulation, and objective quantification techniques are presented. The frontotemporal orbitozygomatic (FTOZ) approach was selected for analysis because it is the most widely studied cranial base approach in the literature. Surgical exposure to specific targets via the FTOZ approach is objectively assessed while simulating measured quantities of brain retraction. For validation purposes, results are compared with those of similar cadaver-based studies designed to objectively quantify the exposure provided by various stages of the FTOZ approach (1, 4, 7, 13).

MATERIALS AND METHODS Data Acquisition Four computed tomographic angiograms (CTAs) were retrospectively selected from an archived clinical database as part of an Institutional Review Board–approved study. CTAs were selected if they were officially read as normal by an attending neuroradiologist and if the patient had no previous intracranial intervention of any kind. Imaging data were acquired on a Siemens Plus4 Volume Zoom scanner (Siemens Medical Solutions USA, Inc., Malvern, PA) according to a standard CTA “Circle of Willis” protocol (1-mm slice thickness; 0-degree angulation; 100 mL of Omnipaque 350 [GE Healthcare, Inc., Princeton, NJ] at 3 mL/s; 15-second scan delay). These data were transferred in the digital image and communication in medicine format to a computer workstation (Dell OPTIPLEX GX280, Dell Computer Corp., Round Rock, TX) for analysis.

Image Segmentation, Reconstruction, and Visualization Individual CTA studies were uploaded into a commercially available 3-D visualization and modeling application (Amira 4.1.1; Mercury Computer Systems, Inc., Chelmsford, MA). With use of various image segmentation modules, the frontal lobes, temporal lobes, cranium, cranial base, and cerebral vasculature were defined on two-dimensional (2-D) axial source images. 3-D surface meshes of each structure were


A

B

FIGURE 1. Image segmentation and model generation. A, anteroposterior (AP) view of a three-dimensional (3-D) model of a patient’s cranium. The computed tomographic data are segmented in a semiautomated fashion and reconstructed in the virtual viewing environment for later analysis. B, AP view of bilateral frontal and temporal lobes in the same patient once the cranium has been removed from the virtual scene.

reconstructed from the segmented data with a volume-rendering module. Once generated, structure meshes could be visualized and manipulated in a 3-D viewing window and image editor. A virtual 3-D model of each patient was created in the viewer and saved for later manipulation (Fig. 1).

Approach Simulation

Building a Preoperative Scene To standardize the frontotemporal (FT) craniotomy, removal of the orbital rim (FTO), and zygomatic arch (FTOZ), specific landmarks were identified on each 3-D model. These landmarks included the supraorbital notch, superior orbital fissure (SOF), inferior orbital fissure (IOF), zygomaticofacial foramen, and root of the zygoma. Relative to the operative corridor provided by a transsylvian FTOZ approach, we identified superficial, deep, high, and low anatomic targets. The midpoint of the anterior communicating artery complex (AComA) was defined as superficial, the basilar artery bifurcation (BA) was defined as deep, the internal carotid artery bifurcation (ICAb) was defined as high, and the anterior clinoid process (ACP) was defined as low. This concept is illustrated in Figure 2. For accuracy, all landmarks and targets were confirmed with coregistered 2-D source image overlays. Desired points were permanently tagged with a point probe function that was used to calculate the exact Cartesian coordinates (x, y, z) for each structure in the 3-D system.

Bone Removal After reconstructing the patient-specific 3-D cranial base models and identifying all anatomic landmarks and targets, we performed virtual craniotomies sequentially on all four patient models bilaterally (n ⫽ 8). First, a standard FT craniotomy was performed with a drawing tool, in which highlighted portions of the cranial mesh were sequentially subtracted from the volume to simulate bone removal. Once the bone flap was subtracted, drilling of the lateral sphenoid wing was simulated by removing bone to the level of the lateral SOF, medially. Temporally, bone was removed to the level of the floor of the middle fossa. Operated cranial models were then saved as FT datasets for later analysis. Next, virtual orbital rim removal was added to each FT dataset. Per protocol, a cut was made in the orbital rim 2 mm lateral to the supraorbital notch (19). A second cut was made from the lateral SOF to the


QUANTIFICATION OF THE FRONTOTEMPORAL ORBITOZYGOMATIC APPROACH

A

A

B

C

D

B

FIGURE 2. Target selection. A two-dimensional (2D) reconstructed computed tomographic angiogram (CTA) in the sagittal plane is displayed in the virtual environment. The image has been windowed to show both bone and blood vessels. The blue (vertical) bar represents the z-axis for the system. The green (horizontal) bar represents the y-axis for the system. A grid overlay is set behind the sagittal image in which each line is 1 cm apart. A, midline sagittal view allowing the operator to label the anterior communicating artery (AComA) as the most superficial target (leftmost red sphere), and the basilar artery bifurcation (BA) as the deepest target (rightmost red sphere), relative to the frontotemporal orbitozygomatic (FTOZ) craniotomy. B, parasagittal view allowing the operator to label the anterior clinoid process (ACP) as the low target (inferior red sphere) and the internal carotid artery bifurcation (ICAb) as the high target (superior red sphere). The 3-D environment provides the exact Cartesian coordinates for all labels generated in the scene.

anterolateral IOF. The lateral orbital rim was then divided by a parallel cut from the anterolateral IOF to the superior aspect of the zygomatic arch (7, 19). Finally, the orbital roof was virtually removed exactly 3 cm back from the anterior aspect of the orbital rim (1, 12, 19). Operated skull reconstructions were then saved as FTO datasets for later analysis. Finally, the zygomatic osteotomy was added to each FTO dataset by making a cut through the malar eminence from the infratemporal fossa to the anterolateral IOF, staying exactly 1 cm above the zygomaticofacial foramen (7). The final cut was made at the root of the zygoma, thereby completely removing the zygomatic arch. Operated cranial reconstructions were then saved as FTOZ datasets for later analysis. A stepwise illustration of the virtual FTOZ is shown in Figure 3.

NEUROSURGERY

FIGURE 3. Stepwise virtual simulation of the FTOZ craniotomy. A, 3-D model of the entire cranium and lobes of the brain were uploaded into the virtual environment. For accurate craniotomy planning, the right superior orbital fissure (SOF), zygomaticofacial foramen, and root of the zygoma were identified and labeled. B, virtual frontotemporal (FT) craniotomy was performed. The lateral sphenoid wing was drilled medially to the level of the SOF, and bone was removed flush to the floor of the middle fossa. The right frontal lobe (yellow) and the right temporal lobe (blue) can be seen. C, the superolateral orbital rim was removed, thereby completing the removal of the orbital rim (FTO) craniotomy. The superior orbital rim was cut 2 mm lateral to the supraorbital notch. Exactly 3 cm of orbital roof was removed. The lateral orbital rim was removed with a cut parallel to the superior margin of the zygoma, exposing the superior and inferior orbital fissures (red spheres). D, zygomatic arch was removed, completing the FTOZ craniotomy. Each virtual craniotomy scene was saved as a separate network for later analysis.

Brain Retraction To standardize brain retraction, the frontal and temporal lobes ipsilateral to the craniotomy were uploaded into the 3-D viewer. On a pure lateral projection, a 2-D measuring grid was inserted as an overlay. With use of a transformation module, the frontal lobe was virtually lifted off of the floor of the anterior fossa in a caudal-cranial direction to 1 and 2 cm of retraction. At the same time, the temporal lobe was virtually pulled back from anterior to posterior to 1 and 2 cm of retraction. Virtual brain retraction was standardized for each lobe independently, as demonstrated in Figure 4.

Microscope Movements To simulate the movement of the operative microscope around a fixed target, a “centerball technique” was implemented. In the 3-D viewer, a specific target was selected with the use of previously recorded Cartesian coordinates. This target was set as the zero axis of rotation for the entire 3-D operative scene. Once defined, the target remained fixed so that the entire operative scene could be freely rotated around the target throughout visualization and data acquisition.


D’AMBROSIO ET AL.

A

B

A

B

C

D

C

D

FIGURE 4. Measured brain retraction. A, FT craniotomy performed in the virtual environment exposing the right frontal (yellow) and temporal (blue) lobes. The scene is being viewed from a pure lateral position. B, a 10-cm ⫻ 10-cm grid was projected over the 3-D models. C, the right frontal lobe was retracted superiorly off of the floor of the anterior fossa by 1 cm. The right temporal lobe was retracted posteriorly, away from the middle temporal fossa by exactly 1 cm. The flattened sphenoid wing and the ACP are visible. D, finally, each lobe was retracted by exactly 2 cm.

FIGURE 5. Projection angle. A, the scene was maintained at 40 degrees of left axial rotation, and the local axis grid was removed for better visualization. B, the 3-D scene was tilted until the target (AComA, single red sphere) could be easily visualized over the superolateral orbital rim. The left frontal and temporal lobes have been removed from the scene for better visualization. C, the projection angle in the oblique plane was measured by generating an angle of vision between the frontal lobe and the orbital roof, through which the target can be visualized. This is represented by a thick white line. The target is at the apex of the projection angle. The centerball sphere has been removed for better visualization. D, the scene has been reset to a true lateral projection. The projection angle to the target can clearly be visualized and recorded.

Data Acquisition

Projection Angle

Surgical Freedom

The projection angle represents a vertical angle of observation in an oblique 2-D plane through which the primary surgeon can visualize a specific target point (1). In the current study, four predefined target points were identified on 3-D reconstructions and verified on coregistered 2-D source images, as previously described. Using the centerball technique, an individual target was defined as the fixed center axis of rotation. Per protocol, the operative scene was initially set in a pure anteroposterior (AP) view. By convention, the AP view was defined as having an axial rotation angle of 0 degrees. The operative scene was then rotated 25 to 40 degrees away from the side of the craniotomy until the primary view was centered over the junction of the superior and lateral orbital rim. The frontal lobe remained fixed at 2 cm of retraction. The head was then tilted until the frontal lobe obstructed visualization of the target of interest. This point was labeled as the upper boundary of the projection angle. The head was then tilted back, maintaining the same degree of axial rotation, until the target became obstructed by the orbital rim. This was marked as the inferior boundary of the projection angle. The vertical angle through which the target could be completely visualized was defined as the projection angle for that target. Projection angle measurements were obtained and directly compared between the FT and FTO craniotomies for all four targets in the same projection plane for each patient. In other words, for each side operated (n ⫽ 8), two craniotomies were performed, brain retraction was fixed at 2 cm, and four targets were approached, resulting in 64 measurements of projection angle in total. This concept is illustrated in Figure 5.

In the current study, surgical freedom is defined as a planar area at the level of the craniotomy through which surgical instruments can be inserted toward a specific target of interest. This objective quantification technique is based on the conical solid method previously described by Schwartz et al. (13). To measure surgical freedom, each of the four previously determined surgical targets were individually identified on the preoperative model (Fig. 6). The target point served as the apex of a cone. Six lines were then circumferentially cast out from the target point, generating unobstructed “lines of sight” to the target. The first line was placed as far into the anteromedial portion of the craniotomy as possible until either the inferior frontal lobe or the medial edge of the craniotomy was encountered. The second line was placed under the frontal lobe in between the first line and the Sylvian fissure. This line simulated the line of sight one might expect to obtain if a retractor blade or instrument were inserted in a subfrontal trajectory toward a target. The third line was placed along the Sylvian fissure to represent exactly 3 cm of fissure splitting. This line was kept fixed throughout all measurements to control for physiologically inaccurate splitting of the fissure. The fourth line was placed at the tip of the temporal lobe to simulate the line of sight one might expect to obtain if a retractor blade or instrument were inserted at the temporal tip on a trajectory toward a specific anatomic target. The fifth line was placed as low in the middle fossa as possible. The sixth line was cast toward the junction of the superior and lateral orbital rim, in between the first and sixth lines, simulating dura reflected over the orbit. This concept is illustrated in Figure 7. An oblique plane was then generated at the surface of the craniotomy, intersecting both the edges of the craniotomy and the lines




A

B

C

D

FIGURE 6. Surgical freedom. For the purpose of illustration, the right FTO virtual craniotomy was performed on a 3-D model of a study patient. The right frontal and temporal lobes were retracted to 2 cm, and the BA was selected as the deep target. A, 2-D coronal reconstructed CTA through the BA and apex was visualized in the 3-D virtual environment. B, coneddown 3-D reconstructed model of the posterior circulation including the BA was inserted. C, the BA was tagged as a target with a red sphere, and the coordinates for the target were recorded. D, the 2-D coronal reconstruction was removed, leaving only the 3-D model of the posterior circulation with the BA target identified in the viewer.

cast from the target of interest. This plane was placed at the level of the craniotomy to incorporate the concept of “depth to target” into the approach assessment. The points of intersection between the oblique plane and the six projected lines were identified and four triangles generated, connecting all points of intersection. Triangles 1 and 2 simulate frontal surgical freedom. Triangles 3 and 4 simulate temporal surgical freedom. The areas of these triangles were calculated by the area of triangles method and were added together to obtain the overall surgical freedom for specific targets under predetermined surgical parameters (i.e., target ⫽ ACP; retraction ⫽ 2 cm; craniotomy ⫽ FTOZ). The superficial intersection plane was kept constant throughout all measurements, allowing accurate comparison of relative values across measurements. This technique is illustrated in Figure 8. The same four targets identified earlier (ACP, AComA, ICAb, and BA) were used as targets for the surgical freedom quantification portion of the study. For each craniotomy (FT, FTO, and FTOZ) in each patient model (n ⫽ 8 sides), the surgical freedom was calculated at 1 and 2 cm of brain retraction. In other words, for each side operated (n ⫽ 8), three craniotomies were performed, two measured amounts of brain retraction were simulated, and four targets were approached, resulting in 92 measurements of surgical freedom in total.

Statistical Analysis A two-tailed Student’s t test was used to determine statistical significance. To account for the multiple comparisons performed, a P value of less than 0.01 was established, a priori, as statistically significant. All statistical analysis was performed with commercially available software (InStat; GraphPad Software, Inc., San Diego, CA).

NEUROSURGERY

FIGURE 7. Surgical freedom. Six lines were cast from the target point, representing unobstructed lines of sight to the target. Line 1 was placed as far into the anteromedial portion of the craniotomy as possible until either the inferior frontal lobe or the medial edge of the craniotomy was encountered. Line 2 was placed under the frontal lobe, in between Line 1 and the Sylvian fissure. This line simulates a frontal retractor or a rigid instrument approaching the basilar apex. Line 3 was placed along the Sylvian fissure to represent exactly 3 cm of fissure splitting. Line 4 was placed at the tip of the temporal lobe to represent a temporal retractor or a rigid instrument approaching the basilar apex. Line 5 was placed as low in the middle fossa as possible. Line 6 was cast over the orbit in between Lines 1 and 6 to represent the dura reflected over the orbit.

RESULTS Projection Angle When analyzed individually, the projection angle for each of the four intracranial targets increased when orbital rim removal was added to the FT craniotomy (FT-FTO). This increase in projection angle was statistically significant when approaching the AComA and ICAb (P ⬍ 0.01). When all targets were combined, the overall projection angle increased by approximately 4 degrees (24.2%). This increase in projection angle was statistically significant (P ⬍ 0.01). The ICAb demonstrated the greater percent increase in projection angle (47.9%, P ⬍ 0.001) compared with the smaller benefit observed for the ACP (6.1%, P ⫽ not significant). Projection angle results are presented in Table 1.

Surgical Freedom: Brain Retraction Fixed at 1 cm When frontotemporal brain retraction was maintained at 1 cm, the overall mean percent increase in surgical freedom was 27 ⫾ 14% when orbital rim removal was added to the FT craniotomy (FT-FTO). When individual targets were analyzed, the greatest increase in surgical freedom was demonstrated at the ICA bifurcation (41 ⫾ 28%) and the smallest benefit was demonstrated at the ACP (20 ⫾ 13%). The increase in surgical freedom was statistically significant across all measurements (P ⬍ 0.01). When the zygomatic arch was removed from the FTO craniotomy (FTO-FTOZ), the mean percent increase in


D’AMBROSIO ET AL.

A

B

TABLE 1. Quantification of projection angles for frontotemporal and orbitofrontal craniotomies obtained at 25 to 40 degrees of axial rotation with brain retraction fixed at 2 cma Projection angle Surgical approach

C

AComA

BA

ICAb

All targets

FT

17.2 ⫾ 2.6 16.1 ⫾ 2.6 12.9 ⫾ 0.8 19.0 ⫾ 2.9 16.3 ⫾ 2.2

FTO

18.3 ⫾ 1.6 20.0 ⫾2.3 15.4 ⫾ 2.8 28.0 ⫾ 5.4 20.4 ⫾ 3.0

% increase

D

ACP

6.1

23.7

19.1

47.9

24.2

a

ACP, anterior clinoid process; AComA, anterior communicating artery complex; BA, basilar apex; ICAb, internal carotid artery; FT, frontotemporal craniotomy; FTO, orbitofrontal craniotomy. Mean values are presented as degrees ⫾ standard deviation.

TABLE 2. Comparison of surgical freedom via frontotemporal craniotomy with or without orbital osteotomy and/or zygomatic osteotomy with brain retraction fixed at 1 cma FIGURE 8. Surgical freedom. The virtual scene has been manipulated to better illustrate the conical volume generated for each target studied under the surgical freedom method. A, a sagittal clipping plane was added to the FTO craniotomy model of the skull, allowing visualization of the basilar apex from an anterior oblique projection. The six lines, or “rays,” illustrated in Figure 7 can be seen originating from the target of interest, which, in this case, is the basilar apex. B and C, to incorporate the concept of surgical depth, an oblique plane cutting through the line probes at the level of the craniotomy was inserted. Linear measurements connecting individual line probes are generated on this plane, creating four triangular areas in between the intersection of the line probes with the plane. This plane intersects all six lines radiating out from the target of interest. D, the line probes have been removed, leaving only four triangular areas. In this example, Triangles 1 and 2 represent frontal surgical freedom, and Triangles 3 and 4 represent temporal surgical freedom. The total area of surgical freedom toward a specific target of interest is then calculated by the area of triangles method.

surgical freedom for all targets was 3 ⫾ 3%. Although a positive benefit was demonstrated for all measurements, it did not reach statistical significance. When we compared the FT craniotomy with the complete FTOZ (FT-FTOZ), the mean percent increase in surgical freedom for all targets was 31 ⫾ 18%. The individual target gaining the greatest benefit was the ICAb (45 ⫾ 31%), and the targets demonstrating the smallest benefit were the AComA (24 ⫾ 17%) and the ACP (25 ⫾ 18%). Once again, the increase in surgical freedom was statistically significant across all measurements (P ⬍ 0.01). These results are shown in Table 2.

Surgical Freedom: Brain Retraction Fixed at 2 cm When frontotemporal brain retraction was maintained at 2 cm, the overall mean percent increase in surgical freedom was 15 ⫾ 5% when orbital rim removal was added to the FT craniotomy (FT-FTO). The greatest increase in surgical freedom was demonstrated at the BA (17 ⫾ 8%), and the least benefit was demonstrated at the ACP (12 ⫾ 9%). This increase in surgical freedom was statistically significant across all measure-


Surgical approach FT-FTO FTO-FTOZ FT-FTOZ

Mean percentage increase in surgical freedom All ACP AComA BA ICAb targets 20 ⫾ 13

22 ⫾ 18

31 ⫾ 23

41 ⫾ 28

4⫾5

2⫾2

4⫾7

2⫾4

27 ⫾ 14 3⫾3

25 ⫾ 18

24 ⫾ 17

36 ⫾ 26

45 ⫾ 31

31 ⫾ 18

a

ACP, anterior clinoid process; AComA, anterior communicating artery complex; BA, basilar apex; ICAb, internal carotid artery; FT, frontotemporal craniotomy; FTO, orbitofrontal craniotomy; FTOZ, frontotemporal orbitozygomatic craniotomy. Mean values are presented as percentages ⫾ standard deviation.

ments (P ⬍ 0.01). When zygomatic arch removal was added to the FTO craniotomy (FTO-FTOZ), a statistically significant increase was found at the ICAb and for all targets combined (P ⬍ 0.01). When we compared the FT craniotomy to the complete FTOZ (FT-FTOZ), the mean percent increase in surgical freedom for all targets was 26 ⫾ 8%. The individual target gaining the greatest benefit was the ICAb (26 ⫾ 13%), and the targets demonstrating the smallest benefit were the AComA (16 ⫾ 7%) and the ACP (15 ⫾ 11%). Once again, the increase in surgical freedom was statistically significant across all measurements (P ⬍ 0.01). These results are shown in Table 3.

Surgical Freedom: Brain Retraction Increased From 1 to 2 cm When bone removal was kept constant and the amount of brain retraction increased from 1 to 2 cm, the mean percent increase in surgical freedom for the FT craniotomy doubled (100 ⫾ 26%) for all targets. When individual targets were approached through FT craniotomy, the BA demonstrated the most benefit from additional brain retraction (120 ⫾ 61%), and the ACP demonstrated the smallest benefit (83 ⫾ 15%). All results reached high statistical significance for FT craniotomy (P ⬍ 0.001). When FTOZ craniotomy was studied, surgical freedom increased for all targets by 81 ⫾ 15% when brain retraction was doubled. The targets demonstrating the greatest benefit



TABLE 3. Comparison of surgical freedom via frontotemporal craniotomy with or without orbital osteotomy and/or zygomatic osteotomy with brain retraction fixed at 2 cma Mean percentage increase in surgical freedom All ACP AComA BA ICAb targets

Surgical approach FT-FTO

12 ⫾ 9

14 ⫾ 6

17 ⫾ 8

16 ⫾ 6

3⫾4

2⫾2

3⫾4

7⫾3

4⫾2

15 ⫾ 11

16 ⫾ 7

21 ⫾ 10

26 ⫾ 13

26 ⫾ 8

FTO-FTOZ FT-FTOZ

15 ⫾ 5

a


TABLE 4. Comparison of surgical freedom via frontotemporal craniotomy with or without orbital osteotomy and/or zygomatic osteotomy with increasing amounts of brain retraction from 1 to 2 cma Surgical approach

Mean percentage increase in surgical freedom All ACP AComA BA ICAb targets

FT 1 cm–FT 2 cm

83 ⫾ 15 111 ⫾ 47 120 ⫾ 61 108 ⫾ 40 100 ⫾ 26

FTO 1 cm–FTO 2 cm

71 ⫾ 13

95 ⫾ 21

94 ⫾ 29

76 ⫾ 16

81 ⫾ 15

FTOZ 1 cm–FTOZ 2 cm 69 ⫾ 11

95 ⫾ 22

92 ⫾ 30

82 ⫾ 16

82 ⫾ 27

a


were the AComA (95 ⫾ 21%) and the BA (94 ⫾ 29%), whereas the least benefit was noted at the ACP (71 ⫾ 13%). All results reached high statistical significance for FTOZ craniotomy (P ⬍ 0.001). When FTOZ craniotomy was studied, surgical freedom increased by 82 ⫾ 27% for all targets when brain retraction was doubled. The targets demonstrating the greatest benefit were the AComA (95 ⫾ 22%) and the BA (92 ⫾ 30%), whereas the least benefit was noted at the ACP (69 ⫾ 11%). All results reached high statistical significance for FTOZ craniotomy (P ⬍ 0.001). The results for each approach and each target are shown in Table 4.

DISCUSSION Model Validation: Comparison with Cadaver-based Studies The primary goal of this study was to demonstrate that the proposed virtual cranial base approach assessment model was capable of generating data similar to the highest-quality objective cranial base approach studies in the current literature. To validate our model, we selected the FTOZ approach for analysis. This approach has been studied extensively and the major-

NEUROSURGERY

ity of quality objective assessment models have been developed around the FTOZ approach (1, 7, 9, 13). Schwartz et al. (13) used frameless stereotactic techniques to quantify the superficial area of exposure gained by removal of the orbital rim and zygomatic arch via the frontotemporal transsylvian approach. This quantification technique is based on the “cone d’approche” described by Emery et al. (3) and the “window of access” described by Honeybul et al. (9). In this model, a cone is generated with its apex at a surgically relevant deep intracranial target. The sides of the cone are bounded by superficial anatomy, blocking visualization of the target of interest. A measurable area is then generated by slicing the base of this cone 10 cm away from the target. This technique allows direct comparison of exposure areas between approaches. In this study, three intracranial targets were assessed. As demonstrated in the current report, Schwartz et al. (13) found that removal of the orbital rim produced a statistically significant increase in the area of exposure to all targets. Additionally, they found that orbital rim removal increased operative exposure to the basilar tip by 28%. This is very similar to the 31% benefit found in our study when the orbital rim was added to the FT craniotomy and brain retraction was fixed at 1 cm. Schwartz et al. (13) also noted that zygomatic arch removal led to additional gain in exposure area; however, this gain did not reach statistical significance over orbital osteotomy alone. The authors hypothesize that the lack of significance was caused not only by the quantity of increase but also by the high variability of increase compared with the FT and FTO craniotomies. In the current report, despite low measurement variability, a statistically significant benefit was not found when the zygomatic arch removal was added to the FTO craniotomy and brain retraction was maintained at 1 cm. However, when brain retraction was increased, a small but statistically significant benefit was found for the ICAb and all targets combined. It is possible that zygomatic arch removal might afford a stronger statistical benefit through a subtemporal corridor; however, only the transsylvian approach was quantified in this study. Gonzalez et al. (7) quantified the operative exposure obtained in the pterional, orbitozygomatic, and modified orbitozygomatic with maxillary extension surgical approaches. In this model, a robotic microscope was used to measure horizontal and vertical angles of attack toward surgically relevant fixed anatomic targets. The microscope was set in a “spherical mode,” which allowed one point to be kept in focus while moving the microscope in any direction. As the surgical approach was extended, the same amount of brain retraction was maintained and the target visualized through the more extensive approach. This technique accounts for additional exposure afforded at the level of the craniotomy; as a result, statistically significant differences between individual approaches were demonstrated. This group used the same technique to compare the angle of approach to the AComA associated with the FTOZ before and after resection of the gyrus rectus (4). This study demonstrated that progressive bone removal produced a significant increase in the vertical angle of attack to the AComA complex for all craniotomies.


D’AMBROSIO ET AL.

We attempted to simulate this quantification technique in our virtual model by introducing the centerball technique. This technique allows the entire operative scene to freely rotate around a fixed target. In our environment, not only can the projection angle be measured, but axial rotation can easily be measured and fixed throughout data acquisition as well. We chose to compare the projection angle to all targets between FT and FTO craniotomies. An axial rotation angle between 25 and 40 degrees was chosen so that the projection angle to the target of interest would fall over the superolateral orbital rim, thereby simulating dura reflected over the orbital roof or orbital contents. We showed an increase in projection angle for all targets; however, this benefit reached statistical significance only for the AComA, ICAb, and all targets combined. If we assume that 2 cm of fixed frontal lobe retraction in our model is similar to gyrus rectus resection in the model of Figueiredo et al. (4), then our results with respect to the orbital osteotomy compare quite well. We demonstrated a projection angle of 16.1 ⫾ 2.6 degrees toward the AComA with FT craniotomy alone. Figueiredo et al. (4) demonstrated a vertical angle of approach measuring 15.3 ⫾ 5.1 degrees for the pterional craniotomy with gyrus rectus resection. We demonstrated a projection angle of 20.0 ⫾ 2.3 degrees with additional orbital rim removal (FTO), which translates into a 23.7% increase versus FT alone (P ⬍ 0.01). Figueiredo et al. (4) demonstrated a vertical angle of approach measuring 18.8 ⫾ 5.8 degrees, translating into a 22.8% increase in vertical angle of approach with removal of the orbital rim and additional gyrus rectus resection.

Overcoming Limitations of Cadaver-based Models: Brain Retraction Two major limitations are repeatedly encountered when critically reviewing accepted objective cranial base approach assessment models: 1) the inherent variability between cadaveric specimens, both in terms of soft-tissue properties and brain atrophy (1, 4, 5, 7, 13); and 2) the inability to incorporate the effects of intracranial pathology on surgical exposure (13). To accurately quantify access to an intracranial target or region through a cranial base approach where brain tissue obstructs a significant proportion of the exposure (i.e., frontotemporal transsylvian), one must attempt to standardize and control brain retraction across measurements. In cadaver-based models, it is possible to control the amount of brain retraction in the same specimen only if all measurements are taken during the same data acquisition session. However, it is not possible to retract a rigid, chemically fixed specimen with no brain atrophy in the exact same fashion as a poorly fixed specimen with significant brain atrophy. Including such specimens in the same anatomic study may result in high levels of measurement variability. To overcome this limitation, we developed a virtual retraction method whereby tissue characteristics and anatomic variations do not affect brain retraction. This technique standardizes and measures the amount of brain retraction across all specimens, allowing one to objectively assess the impact that variable amounts of brain retraction might have on operative exposure.


Our model demonstrated that brain retraction has a dramatic impact on surgical exposure to all targets visualized through the FTOZ craniotomy. As demonstrated with gyrus rectus resection (4), brain retraction had the greatest impact on surgical exposure for the more limited craniotomy. When the FT craniotomy was analyzed, overall surgical freedom increased by 100 ⫾ 26 degrees when the lobes were retracted from 1 to 2 cm. This was highly statistically significant for all targets measured, with the greatest benefit noted at the basilar apex and the smallest benefit found at the anterior clinoid process. The benefit of brain retraction decreased somewhat, but remained highly significant, as the orbital rim and zygomatic arch were sequentially removed. The greatest benefits were noted at the BA and the AComA.

Overcoming Limitations of Cadaver-based Models: Intracranial Pathology The second major limitation to current approach assessment models is the inability to incorporate the effects of intracranial pathology on surgical exposure, which becomes important when approaching pathology that may significantly distort otherwise normal anatomy. As pointed out by Schwartz et al. (13), in many cases of sphenoid wing meningiomas, the lesion will push the brain away from the cranial base, creating a corridor of access and eliminating much of the benefit of orbitozygomatic osteotomy. Unfortunately, this limitation cannot be addressed in cadaver-based approach models. To overcome this limitation, we developed a model based on 3-D reconstructed images of patient-specific data. This model allows the operator to perform various cranial base techniques on visually and spatially accurate models of true patient anatomy. The goal of this study was to determine whether or not our model was capable of performing approach assessment techniques on normal anatomic models similar to cadaver-based models; therefore, we did not choose to use pathological imaging data as a part of this validation study. However, our virtual cranial base approach assessment model is fully capable of analyzing any intracranial pathology of interest.

Virtual Model Capabilities: Target Location and Surgical Exposure Schwartz et al. (13) posed an interesting question that is not directly addressed in their study. They hypothesized that operative exposure will vary according to the specific location of the target being approached, specifically, target height and depth. For example, exposure to a low target, lying along or below a tangent from the sphenoid wing, will benefit less from removal of the orbital rim or zygoma, versus a target lying above this tangent line, such as the AComA. Furthermore, for deeper targets, more surgical hindrance is due to neural structures rather than bone, and the effects of osteotomy on brain position are relatively small. To test this hypothesis, we selected four targets representing high, low, superficial, and deep locations relative to the transsylvian approach. The high target (ICAb) demonstrated the greatest benefit in percent increase in projection angle (47.9%)



among all targets when the orbital rim was removed from the FT craniotomy. This benefit was highly statistically significant (P ⬍ 0.001). The low target (ACP) demonstrated the smallest benefit in percent increase in projection angle (6.1%, P ⫽ not significant) when the orbital rim was removed from the FT craniotomy. This result was again demonstrated when assessing surgical freedom to all targets. With brain retraction fixed at 1 cm, the high target (ICAb) demonstrated the greatest mean percent increase in surgical freedom when the orbital rim was removed from the FT craniotomy (41 ⫾ 28 degrees, P ⬍ 0.005) and when the complete orbitozygomatic osteotomy was added to the FT craniotomy (45 ⫾ 31 degrees, P ⬍ 0.005). These results strongly support the hypothesis that exposure to targets higher in the surgical field stands to benefit more from superficial bone removal at the level of the craniotomy. With regard to target depth, the superficial target (AComA) demonstrated a greater benefit in percent increase in projection angle (23.7%) when compared with the deep target (BA, 19.1%). This increase in exposure reached statistical significance only for the AComA (P ⬍ 0.01). However, this finding was not supported when surgical freedom was assessed for these targets. In fact, the BA demonstrated greater benefit in mean percent increase in surgical freedom than the more superficial AComA as bone was removed more superficially. This pattern was seen when simulated brain retraction was fixed at both 1 and 2 cm. Therefore, our model does not support the concept that deeper targets are afforded less exposure than more superficial targets by superficial bone removal.

Virtual Model Limitations Although the proposed cranial base approach model has the ability to quantify and standardize brain retraction, the simulation of brain retraction is quite crude at this point. This model is not currently able to accurately simulate the placement of a retractor on brain tissue in a visually or mathematically accurate way. This basic simulation may lead to falsely larger operative exposures than one would see in living tissue. To address this, we are incorporating finite element models of the lobes of the brain constructed from magnetic resonance imaging data into future versions of the model. Another limitation to our virtual approach model is the exclusion of cranial nerves, tentorium, muscle, and scalp from our preoperative 3-D models. Incorporating these structures into the operative scene are possible and potentially important when approaching deep targets that may be obstructed by structures other than brain, blood vessels, or bone. To address this issue, we are incorporating coregistered magnetic resonance imaging data capabilities into future versions of the model. Despite these technological limitations, the findings in the current study compare favorably with high-quality, accepted cadaver-based models in the current neurosurgical literature.

CONCLUSION In the current study, a 3-D virtual cranial base approach assessment model is presented and validated through the

NEUROSURGERY

analysis of the FTOZ cranial base approach. The results of this study compare favorably with those from high-quality cadaverbased models. Additionally, this methodology overcomes two major limitations inherent to cadaver-based studies: 1) quantification and standardization of brain retraction and 2) the ability to incorporate patient-specific anatomy and pathology. Through simulated and measured brain retraction, we have eliminated the variability introduced by chemically fixed cadaver tissue and quantified the impact of brain retraction on operative exposure. By designing the model in the virtual environment of a validated 3-D visualization and modeling application, we have introduced the capability to assess the effects of patient-specific anatomy and pathology on operative exposure. This body of work represents the first patient-specific objective cranial base approach assessment model in the literature and is the first step toward improving our ability to tailor surgical strategies on a case-by-case, patient-by-patient basis.

REFERENCES 1. Andaluz N, Van Loveren HR, Keller JT, Zuccarello M: Anatomic and clinical study of the orbitopterional approach to anterior communicating artery aneurysms. Neurosurgery 52:1140–1149, 2003. 2. Deshmukh VR, Figueiredo EG, Deshmukh P, Crawford NR, Preul MC, Spetzler RF: Quantification and comparison of telovelar and transvermian approaches to the fourth ventricle. Neurosurgery 58 [Suppl 2]:ONS202– ONS207, 2006. 3. Emery E, Alaywan M, Sindou M: Respective indications of orbital and/or zygomatic arch removal combined with fronto-pteriono-temporal approaches. 58 cases [in French]. Neurochirurgie 40:337–347, 1994. 4. Figueiredo EG, Deshmukh P, Zabramski JM, Preul MC, Crawford NR, Siwanuwatn R, Spetzler RF: Quantitative anatomic study of three surgical approaches to the anterior communicating artery complex. Neurosurgery 56 [Suppl]:397–405, 2005. 5. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Preul MC, Spetzler RF: Anatomical and quantitative description of the transcavernous approach to interpeduncular and prepontine cisterns. Technical note. J Neurosurg 104:957–964, 2006. 6. Figueiredo EG, Zabramski JM, Deshmukh P, Crawford NR, Spetzler RF, Preul MC: Comparative analysis of anterior petrosectomy and transcavernous approaches to retrosellar and upper clival basilar artery aneurysms. Neurosurgery 58 [Suppl]:ONS13–ONS21, 2006. 7. Gonzalez LF, Crawford NR, Horgan MA, Deshmukh P, Zabramski JM, Spetzler RF: Working area and angle of attack in three cranial base approaches: Pterional, orbitozygomatic, and maxillary extension of the orbitozygomatic approach. Neurosurgery 50:550–557, 2002. 8. Honeybul S, Neil-Dwyer G, Lang DA, Evans BT, Weller RO, Gill J: The extended transbasal approach: A quantitative anatomical and histological study. Acta Neurochir (Wien) 141:251–259, 1999. 9. Honeybul S, Neil-Dwyer G, Lees PD, Evans BT, Lang DA: The orbitozygomatic infratemporal fossa approach: A quantitative anatomical study. Acta Neurochir (Wien) 138:255–264, 1996. 10. Horgan MA, Anderson GJ, Kellogg JX, Schwartz MS, Spektor S, McMenomey SO, Delashaw JB: Classification and quantification of the petrosal approach to the petroclival region. J Neurosurg 93:108–112, 2000. 11. Hsu FP, Anderson GJ, Dogan A, Finizio J, Noguchi A, Liu KC, McMenomey SO, Delashaw JB: Extended middle fossa approach: Quantitative analysis of petroclival exposure and surgical freedom as a function of successive temporal bone removal by using frameless stereotaxy. J Neurosurg 100:695–699, 2004. 12. Lemole GM, Henn JS, Zabramski JM, Spetzler RF: Modifications to the orbitozygomatic approach. Technical note. J Neurosurg 99:924–930, 2003. 13. Schwartz MS, Anderson GJ, Horgan MA, Kellogg JX, McMenomey SO, Delashaw JB: Quantification of increased exposure resulting from orbital rim


D’AMBROSIO ET AL.

14.

15.

16.

17. 18.

19.

and orbitozygomatic osteotomy via the frontotemporal transsylvian approach. J Neurosurg 91:1020–1026, 1999. Siwanuwatn R, Deshmukh P, Figueiredo EG, Crawford NR, Spetzler RF, Preul MC: Quantitative analysis of the working area and angle of attack for the retrosigmoid, combined petrosal, and transcochlear approaches to the petroclival region. J Neurosurg 104:137–142, 2006. Spektor S, Anderson GJ, McMenomey SO, Horgan MA, Kellogg JX, Delashaw JB: Quantitative description of the far-lateral transcondylar transtubercular approach to the foramen magnum and clivus. J Neurosurg 92:824–831, 2000. Tanriover N, Ulm AJ, Rhoton AL, Kawashima M, Yoshioka N, Lewis SB: One-piece versus two-piece orbitozygomatic craniotomy: Quantitative and qualitative considerations. Neurosurgery 58 [Suppl 2]:ONS229–ONS237, 2006. Tanriover N, Ulm AJ, Rhoton AL, Yasuda A: Comparison of the transvermian and telovelar approaches to the fourth ventricle. J Neurosurg 101:484–498, 2004. Youssef AS, Abdel Aziz KM, Kim EY, Keller JT, Zuccarello M, van Loveren HR: The carotid-oculomotor window in exposure of upper basilar artery aneurysms: A cadaveric morphometric study. Neurosurgery 54:1181–1189, 2004. Zabramski JM, Kiris T, Sankhla SK, Cabiol J, Spetzler RF: Orbitozygomatic craniotomy. Technical note. J Neurosurg 89:336–341, 1998.

COMMENTS

T

he authors provide a three-dimensional modeling application: the frontotemporal orbitozygomatic approach. The role of three-dimensional modeling has expanded in neurosurgery with the increasing use of computers in our field. In this article, the authors describe one such method, although its use in actual patients has not been significant, and I am not sure how widely their technique would be used by others on a day-to-day basis in clinical practice. Any preoperative or perioperative modeling that improves patient outcome at a reasonable cost would be desirable. Further work needs to be done in this field of three-dimensional modeling if it is to play a significant role within their surgery. Steven D. Chang Stanford, California

virtual surgical models (1–3) and in using cadaveric anatomic models continually leads us to a reliance on real anatomy or at least on images of real anatomy. We are now fusing real images (fine digital) of the anatomy with imaging. D’Ambrosio et al. should be congratulated for presenting a balanced view that incorporates the strengths and weaknesses of their system. We expect that refinements in their system will allow detailed studies of dissection and anatomy that almost simulate the experience afforded by working with real tissue. The critical point is that we as neurosurgeons do not neglect maintaining contact with real tissue and in the process train our residents and fellows to become “virtual surgeons.” Mark C. Preul Robert F. Spetzler Phoenix, Arizona 1. Balogh AA, Preul MC, Laszlo K, Schornak M, Hickman M, Deshmukh P, Spetzler RF: Multilayer image grid reconstructive technology: Four-dimensional interactive image reconstruction of microsurgical neuroanatomic dissections. Neurosurgery 58 [Suppl 1]:ONS157–ONS165, 2006. 2. Balogh A, Preul MC, Schornak M, Hickman M, Spetzler RF: Intraoperative stereoscopic QuickTime Virtual Reality. J Neurosurg 100:591–596, 2004. 3. Bernardo A, Preul MC, Zabramski JM, Spetzler RF: A three-dimensional interactive virtual dissection model to simulate transpetrous surgical avenues. Neurosurgery 52:499–505, 2003.

T

his report addresses the value of surgical simulation in operative planning and allows the incorporation of the disease state into the image analysis. The findings are not surprising and are an established principle of cranial base surgery. The removal of bone increases the operative exposure. Similar work using cadavers has been published, but, although important, such work does not reflect the patient at hand. The concept described in this report can be augmented to allow surgeons to plan their operations more effectively. In addition, it will allow surgeons to evaluate the sight lines for new surgical approaches. There are limitations as they discuss. Not every element of an exposure can be well simulated at present, including degree of brain retraction.

T

his article is nicely written, especially because many of these virtual systems are difficult to describe as they are inherently visual. D’Ambrosio et al. have rendered a workable system for use in assessment of some maneuvers in the surgical approach. The question is to what extent such virtual systems can replace cadaveric assessment. With refinement and progress, we will be able to take additional structures such as vessels and nerves into account. Currently, this system relies on imaging for segmentation, and the manipulation is based on images. So far such systems have not actually been interactive (i.e., they do not recreate the surgical experience) nor have they been used to quantify approach maneuvers such as resection of more bone or brain retraction. Cadaver-based studies remain the standard. Unfortunately, many institutions do not possess the facilities or expertise to collect and produce finely perfused, vascular-injected specimens for study. Thus, progress with systems such as the one presented by the authors seems necessary and is less costly. D’Ambrosio et al. recognize that although the anatomic variation of the cadaveric specimen is a weakness, at the same time it is a strength for such studies. It accounts for the multitude of soft tissue structures and region through which surgeons must navigate and allows the study of effects related to minor changes in the surgical area, such as slight shifts of brain retraction. We have begun to experiment with methods that incorporate assessment of pathological manipulation into the quantification of cadaveric dissections. Our experience in devising interactive three-dimensional


Douglas Kondziolka Pittsburgh, Pennsylvania

I

n this article, D’Ambrosio et al. have used virtual reality modeling to study the exposures obtained by two different cranial base approaches. This method of study offers some data that are complementary and corroborate those obtained by cadaveric study. However, the information obtained by this method is still not as good as the information obtained in live patients with lesions such as tumors or aneurysms. In patients with aneurysms, brain and brainstem swelling (of varying degrees) may make a difference. In patients with tumors, the tumor itself creates spaces; however, structures such as the tentorium, cavernous sinus, clinoid processes, and petrous bone can make a difference. There are also some relatively expendable areas in the brain such as, the temporal tip or inferior temporal gyrus, and critical areas such as, the hypothalamus, optic nerves and chiasm, and the brainstem, which affect the exposure. The authors are trying to progressively improve their technique, which may play an important role in planning different surgical operations in the future. Laligam N. Sekhar Seattle, Washington

T

he relative barrier to the development of an effective surgical simulation model has not been imaging technology, as three-dimen-



sional imaging technology with computed tomography or magnetic resonance imaging has been quite robust for some time now. Its development is primarily one of information technology, not imaging. This work by D’Ambrosio et al. lays the groundwork for an application of current imaging technology that should be exploited and developed fully in the coming years. It appears that even this relatively crude model provides beneficial effects in the operating room, and there are

probably no insurmountable obstacles to developing the complex refinements of an even more useful surgical simulation model for real patients in a clinical setting. Paul E. Kim Neuroradiologist Los Angeles, California

Lithograph, (1838), Paolo Emilio Morgari from Tavole Anatomiche. From: Wolf-Heidegger G, Cetto AM: Die Anatomische Sektion in Bildlicher Darstellung. Basel, Karger, 1967.

NEUROSURGERY


GENERAL Operative Technique

FRONTOZYGOMATIC TITANIUM CRANIOPLASTY IN FRONTOSPHENOTEMPORAL (“PTERIONAL”) CRANIOTOMY Shaan M. Raza, M.D. Department of Neurosurgery, Division of Cerebrovascular Neurosurgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland

Quoc-Anh Thai, M.D. Department of Neurosurgery, Division of Cerebrovascular Neurosurgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland

Gustavo Pradilla, M.D. Department of Neurosurgery, Division of Cerebrovascular Neurosurgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland

OBJECTIVE: One of the most common problems after frontosphenotemporal, or pterional, craniotomy is the marked depression of the frontozygomatic fossa caused by atrophy of the temporalis muscle. Although temporalis muscle reconstruction techniques have been proposed to prevent this problem, a definitive solution has not been achieved. We report the results of a titanium cranioplasty technique in a prospective series of patients who underwent frontosphenotemporal craniotomy. METHODS: Between April 2002 and June 2006, 209 consecutive patients underwent a frontosphenotemporal craniotomy for aneurysms, vascular malformations, or tumors. At the time of surgery, the patients underwent a frontozygomatic fossa cranioplasty with a titanium plate, to which the temporalis muscle was attached. In this series, 194 patients had documented follow-up periods averaging 9.5 months (range, 1 mo–4 yr; median, 7.5 mo), and the cosmetic results of the cranioplasty have been assessed. RESULTS: The cosmetic outcomes have been outstanding in all patients treated to date. Two patients had the cranioplasty removed due to either orbital pain or local infection secondary to sepsis. CONCLUSION: The frontozygomatic cranioplasty during frontosphenotemporal craniotomy prevents the characteristic depression at the frontozygomatic fossa and accomplishes an outstanding cosmetic result. KEY WORDS: Cranioplasty, Frontosphenotemporal craniotomy, Pterional craniotomy

Rafael J. Tamargo, M.D. Department of Neurosurgery, Division of Cerebrovascular Neurosurgery, The Johns Hopkins University School of Medicine, Baltimore, Maryland Reprint requests: Rafael J. Tamargo, M.D., Division of Cerebrovascular Neurosurgery, Department of Neurosurgery, The Johns Hopkins Hospital, 600 N. Wolfe Street/Meyer 8-181, Baltimore, MD 21287. Email: [email protected] Received, October 22, 2006. Accepted, July 31, 2007.


D

epression of the frontozygomatic fossa after a frontosphenotemporal (FST), or pterional, craniotomy is a common aesthetic defect. This defect is caused by atrophy of the temporalis muscle exacerbated by drilling and bony removal of the greater wing of the sphenoid as well as parts of the frontal bone and squamosal temporal bone. Reconstruction of the temporalis muscle after such bony removal is difficult because of the lack of underlying bony support. Long-term followup of some of these patients reveals that the temporalis muscle appears flattened or depressed because of the lack of skeletal support and muscle atrophy; this is often noted 6 to 12 months postoperatively. An important goal after a successful FST craniotomy is cosmetically superior reconstruction in which there is symmetry of the frontozygomatic fossa. This can be addressed by maintaining the neurovascular supply to


DOI: 10.1227/01.NEU.0000297011.27846.62

the temporalis, maintaining meticulous alignment of the muscle component, and providing a structural support for the muscle itself. Previous reports have addressed some of these key issues. However, there is currently no definitive solution. We report the results of a titanium cranioplasty technique in a series of 209 consecutive patients who underwent FST craniotomy for aneurysms, vascular malformations, or tumors. In addition to the aforementioned methods of careful temporalis muscle dissection and reattachment of the temporalis muscle, the use of a titanium cranioplasty provided the necessary support for the muscle.

PATIENTS AND METHODS Between April 17, 2002 and June 30, 2006, 209 consecutive FST craniotomies were performed by one neurosurgeon (RJT) at our institution; all patients


TITANIUM CRANIOPLASTY

A

E

B

C

F

FIGURE 1. Artist’s rendering of the steps for performing pterional craniotomy and titanium cranioplasty. A, curvilinear scalp incision. B, scalp reflection and temporalis incision. C, reflection of the temporalis with subsequent craniotomy using the Gigli saw (Depuy, Warsaw, IL). D, the

were enrolled in an institutional review board–approved database. The craniotomy was performed the same way each time. In brief, a curvilinear scalp incision was made from the root of the zygoma to the anterior midline edge of the hairline (Fig. 1A). The scalp was reflected with the superficial fascia of the temporalis muscle. The incision for the superficial fascia of the temporalis muscle is shown in Figure 1B. The incision should be below the scalp incision at the root of the zygoma and 1 to 2 cm inferior to the linea temporalis. This preserves the frontal branch of the facial nerve. Then, the temporalis muscle was reflected (Fig. 1C), leaving a cuff for closure. Reflection of the temporalis muscle was performed by dissecting deep to the periosteum of the temporalis muscle in order to avoid injuring the deep temporal arteries and nerves, which can lead to subsequent atrophy. Five burr holes were made (Fig. 1C), and the holes were connected using a Gigli saw (Depuy, Warsaw, IL). After the bone flap was removed, the edge of the remaining temporal squamosal bone was removed with a Leksell rongeur (Aesculap, San Francisco, CA). The greater wing of the sphenoid was drilled medially until the dural sleeve of the orbitomeningeal artery was reached. The orbital roof and inner table of the frontal bone ridges were also drilled (Fig. 1D). The dura was then opened in a curvilinear fashion and reflected anteriorly (Fig. 1D). Sutures were used to secure the dura away from the operative corridor. Toward the conclusion of each surgery, a cranioplasty of the frontozygomatic fossa was performed with a titanium plate (Fig. 2). First, a titanium mesh was cut, as shown in Figure 1E. Once the bone flap was

NEUROSURGERY

D

G

H

orbital roof and frontal bone ridges are drilled and the dura is opened. E, titanium mesh is cut as shown. F, contouring of the mesh to the temporalis fossa. G, the temporalis is sutured to the mesh. H, reapproximation of the temporalis muscle.

in place, the widest section of the cut titanium mesh was secured to the frontozygomatic process to cover the defect from the drilling during the craniotomy. The mesh was contoured to be flush with the bone flap in the temporal fossa (Fig. 1F), and the temporalis muscle was sutured over the mesh (Fig. 1G). The temporalis muscle was secured using the cuff of the temporalis muscle that was left attached to the linea temporalis at the beginning of the procedure. Standard craniotomy closure was performed after the cranioplasty, including closure of the forehead burr hole with a burr hole plate. All patients have been followed postoperatively for aesthetic outcomes and functional assessments; cosmesis was graded according to the scale described in Table 1. The mean duration of follow-up was 9.5 months (n ⫽ 194; range, 1 mo–4 yr), and the median follow-up period was 7.5 months.

RESULTS There were 209 patients with a mean age of 52 years (range, 9–78 yr); 15 patients were lost to follow-up. Of the remaining 194 patients, there were 52 male and 142 female patients. These patients underwent craniotomy for aneurysm (179 patients), tumor (11 patients), or vascular malformations (5 patients). Follow-up of the 195 consecutive FST craniotomies showed a well-healing, aesthetic cranioplasty; 93% of the patients were categorized into Grade A, whereas the remaining patients were con-


RAZA ET AL.

A

B

FIGURE 2. Photographs showing a patient who underwent bilateral craniotomies for aneurysms at different institutions. The patient’s right-sided craniotomy (A) was done at an outside hospital, whereas the patient’s leftsided craniotomy (B) was performed at our institution. As indicated by the arrows, the right-sided craniotomy is marked by temporalis atrophy.

TABLE 1. Grading system for cosmetic outcomes Grade

Components

A

No evidence of surgery with mild prominence of temporalis

B

Slight depression of temporalis fossa

C

Marked depression of temporalis fossa with suggestion of zygomatic process

D

Atrophy of temporalis

E

Atrophy with sliding/rooting of temporalis

sidered to have a Grade B outcome. In comparison with previous FST craniotomies without cranioplasties, patients who had the cranioplasty had a more symmetric appearance. Figure 2 demonstrates the final appearance of a patient in whom a cranioplasty was performed. The frontozygomatic fossa was fuller and resembled the contralateral frontozygomatic fossa. Functionally, there were no problems with the temporalis muscle in terms of strength and range of motion. All wounds healed well, without any problems. There were no wound breakdowns or infections. Two patients required removal of their titanium cranioplasty. One patient developed orbital pain and requested removal of the plate; the patient’s pain persisted despite intervention. A second patient developed sepsis and a focal infection. No other complications have been observed.

DISCUSSION Although several temporalis muscle reconstruction techniques have been described, there is no definitive solution (1–7). One aspect of the reconstruction has focused on the denervation and subsequent atrophy of the temporalis muscle. Several techniques have been published regarding preservation of the vascular and nervous supply of the temporalis muscle (4, 5). Others have focused on the anatomic realignment of the temporalis muscle, and they have suggested varying techniques of reapproximating the temporalis muscle in order to prevent asymmetry (6, 7). In addition to these tech-


niques, however, the cranial defects from the craniotomy have not been addressed. This defect in the sphenoid that results from drilling contributes to temporalis muscle asymmetry despite preservation of vascular and nervous supply and despite correct reapproximation of the temporalis muscle because there is lack of a skeletal foundation upon which the temporalis muscle can rest. As a side note, it is important to note that the Gigli saw is preferred over the pneumatic drill because it creates a thinner cut and a bevel that is otherwise difficult to form with the drill. A thinner cut avoids the creation of a palpable and often visible groove that can occur with realigning the bony edges.

CONCLUSION Titanium cranioplasty is a safe, effective technique of reconstructing the frontozygomatic fossa after an FST craniotomy. During our series of 195 consecutive FST craniotomies, there were no cases of adverse outcome, except for one patient with complaints of orbital pain. There were no instances of infection, wound breakdown, or allergic reaction primarily attributed to the titanium plate and screws. The easily shaped and malleable titanium plates make them versatile and effective as a structural support for the temporalis muscle. A superior aesthetic result is predictably achieved after a cranioplasty with titanium plates. These plates are able to support the temporalis and provide the symmetric contour of the temporal fossa. Reliably producing a symmetric frontal temporal profile is the most important aspect of an aesthetically superior outcome. Therefore, we recommend a titanium cranioplasty for all patients undergoing an FST craniotomy.

REFERENCES 1. Badie B: Cosmetic reconstruction of temporal defect following pterional [corrected] craniotomy. Surg Neurol 45:383–384, 1996. 2. Bowles AP: Reconstruction of the temporalis muscle for pterional and cranioorbital craniotomies. Surg Neurol 52:524–529, 1999. 3. Brunori A, DiBenedetto A, Chiappetta F: Transosseous reconstruction of temporalis muscle for pterional craniotomy: Technical note. Minim Invasive Neurosurg 40:22–23, 1997. 4. Matsumoto K, Akagi K, Abekura M, Ohkawa M, Tasaki O, Tomishima T: Cosmetic and functional reconstruction achieved using a split myofascial bone flap for pterional craniotomy. Technical note. J Neurosurg 94:667–670, 2001. 5. Oikawa S, Mizuno M, Muraoka S, Kobayashi S: Retrograde dissection of the temporalis muscle preventing muscle atrophy for pterional craniotomy. Technical note. J Neurosurg 84:297–299, 1996. 6. Spetzler RF, Lee KS: Reconstruction of the temporalis muscle for the pterional craniotomy. Technical note. J Neurosurg 73:636–637, 1990. 7. Zager EL, DelVecchio DA, Bartlett SP: Temporal muscle microfixation in pterional craniotomies. Technical note. J Neurosurg 79:946–947, 1993.

COMMENTS

T

he authors present a simple but elegant solution to the problem of temporal defects after pterional craniotomy. Although this may appear to be a cosmetic issue, it addresses the psychosocial concern of removing the visual stigmata of illness and treatment, thereby improving patient satisfaction and quality of life. Unfortunately, the cost of the implant can exceed the reimbursement for doing the procedure and


TITANIUM CRANIOPLASTY

may represent a significant portion of the overall expense and reimbursement for the craniotomy. This latter issue represents a paradox and challenge and is an indictment of our current health care socioeconomic environment. T.C. Origitano Maywood, Illinois

I

n this well-written article, Raza et al. present their technique (and results in 209 patients) for titanium cranioplasty for pterional craniotomies to avoid the frontozygomatic area depression resulting from temporal muscle atrophy. The technique essentially allows for the tailoring and cutting of the titanium mesh to fill the defect created by the osteotomies at the anatomic keyhole and the sphenoid wing. I have used similar techniques for closure of orbitozygomatic craniotomies by using titanium mesh reconstruction for the sphenoid wing and keyhole areas, and I agree with the authors that this technique provides superior cosmetic results. In my opinion, the single most important factor for achieving superior cosmetic results is avoidance of the use of a Bovie cautery instrument in the initial dissection of the temporalis muscle as the cauterization of the muscle with this instrument does lead to muscle atrophy. Muscle dissection in the subperiosteal plane can be done sharply (using a periosteal elevator) along the fibers of the muscle and toward the insertion of the temporalis muscle. I commend the authors for achieving good results in this large series with only two cranioplasty-related complications. Saleem I. Abdulrauf St. Louis, Missouri

R

ecently, there have been an increasing number of surgical procedures to access the cranial base; however, the pterional approach still remains the most frequently used. This is a very simple approach that allows wide and safe exposure of the anterior, middle, and, in some cases, posterior fossas. To get a more comfortable exposure with this technique, it becomes necessary to remove the greater wing of the sphenoid bone with a drill or rongeur. This bone management creates a cosmetic defect that is sometimes the main postoperative complaint of patients. In this article, Raza et al. propose a technique to prevent this problem by using titanium mesh, modeled according to the frontoorbitozygomatic contour. They present a series of 194 patients using this procedure with excellent results. When dealing with cosmetic problems of the pterional approach, a surgeon can use this cranioplasty technique, but the increase in the cost of the procedure also has to be considered. Gerardo Guinto-Balanzar Mexico City, Mexico

P

atients undergoing craniotomy have the potential to see the cosmetic results of their surgery (particularly for pterional craniotomy) each time they look in the mirror. Although optimal neurological outcomes are paramount, good cosmesis can aid in patient self-esteem. The trend toward minimal (or no) shaving of hair for these patients is an example of plastic surgical techniques becoming

more widely adopted by the neurosurgical community. Among the more vexing cosmetic issues facing these patients are the atrophic changes affecting the temporalis muscle after craniotomy in this region. The authors stress good technique (preservation of neurovascular structures, leaving cuffs for reattaching the muscle) but also the role of a titanium mesh cranioplasty to improve cosmesis. Albeit costly, the technique may have merit, but attention to detail can go a long way toward improving cosmetic outcome in patients undergoing pterional craniotomy. Gene H. Barnett Cleveland, Ohio

T

he authors present a technique of frontozygomatic cranioplasty using a titanium plate to reconstruct the defect that occurs after a pterional craniotomy (frontosphenotemporal craniotomy). The authors stress the good cosmetic outcomes achieved without any significant morbidities related to the cranioplasty. This article may not seem of significant value to the medical care of patients undergoing surgery; however, I think it is an important reminder to neurosurgeons to pay attention to their cosmetic outcomes. I agree with the importance of reconstructing this region to prevent the “sunken” temporalis muscle problem. In our practice, we utilize a cranial fix device to reconstruct the burr hole at the region of the pterion and further elevate the cranium with “bone source” material (Stryker Instruments). This resulted in a very satisfactory result in our patients similar to what is reported by Raza et al. in this article. We share with the authors the need for drilling the sphenoid aggressively during the sphenoid wing to gain access to the basal cistern and the basal aspect of the brain with minimal or no need for brain retraction. As a result, the cost is the defect produced, which can be reconstructed as mentioned in the article. The benefit is the significant protection of the brain. The authors also stress an important point regarding preservation of the inner temporalis fascia during the dissection process to avoid injury of the blood and nerve supply of the temporalis muscle and its atrophy. This is an important point in view of the fact that several neurosurgeons use the cautery instrument to dissect the temporalis muscle off the temporal bone, which can be injurious to its blood and nerve supply. More recently in our practice we have been shaving the hair only along the incision line. This policy in combination with the proper cosmetic reconstruction leads to a good psychological result in the patient’s perception by minimizing the impact of the surgery. This technique alone leads to patients having a more positive attitude toward earlier mobility and thus leads to faster recovery and faster discharge time from the hospital. With this policy in place, we reviewed our last year’s average hospital stay for patients operated on for unruptured intracranial aneurysms, which, to our pleasant surprise, turned out to be at a mean value of 2.5 days. Ali F. Krisht Little Rock, Arkansas

GENERAL New Technology Koreaki Irie, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Yuichi Murayama, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Takayuki Saguchi, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Toshihiro Ishibashi, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Masaki Ebara, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Hiroyuki Takao, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan

Toshiaki Abe, M.D. Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, Tokyo, Japan Reprint requests: Koreaki Irie, M.D., Division of Endovascular Neurosurgery, Department of Neurosurgery, Jikei University School of Medicine, 3–25–8 Nishishinbashi, Minatoku, Tokyo, #105–8461 Japan. Email: [email protected] Received, January 26, 2007. Accepted, July 10, 2007.

DYNACT SOFT-TISSUE VISUALIZATION USING AN ANGIOGRAPHIC C-ARM SYSTEM: INITIAL CLINICAL EXPERIENCE IN THE OPERATING ROOM INTRODUCTION: DynaCT is a clinical application protocol to create computed tomographic (CT)-like images allowing soft-tissue visualization acquired from an angiography system. A cone beam three-dimensional CT reconstruction is produced from the acquisition of two-dimensional projection images by rotating the c-arm with x-ray source and image receptor around the patient. The purpose of this study is to evaluate the clinical efficacy of DynaCT in the operating room. METHODS: DynaCT was performed in 100 patients undergoing cerebral or spinal interventional procedures in the new-concept operating room. Specially designed AXIOM Artis BA (conventional image intensifier system; 55 patients; Siemens Medical Solutions, Erlangen, Germany) and AXIOM Artis dBA (flat-panel detector; 45 patients; Siemens Medical Solutions) biplane neuroangiographic systems (Siemens Medical Solutions, Germany) were installed in the operating room. The volumetric data set from the AXIOM Artis systems were reconstructed immediately on the three-dimensional workstation in the operating room. We compared DynaCT images with the use of multidetector computed tomography. RESULTS: DynaCT was performed successfully in all patients. High-contrast structures, such as bone, calcified lesions, and metallic materials, were visualized on DynaCT as well as in multidetector computed tomography for each group. Although contrast differentiation of soft tissue such as cerebral cortex, muscle, and hematoma on DynaCT were inferior to multidetector CT scans, the images were sufficiently used as intraoperative CT-like images. However, DynaCT images acquired from flat panel detectorbased systems were found to be superior to those images acquired from image intensifier-based systems. Striking ring artifacts were exhibited and resulted in major limitations in the image intensifier group. CONCLUSION: DynaCT has the potential to be used as a powerful tool for endovascular and neurosurgical procedures and will open new possibilities for neurosurgical management. KEY WORDS: Angiography, Cone beam computed tomography, Endovascular therapy, Flat panel detector, Image, Intraoperative, Three-dimensional Neurosurgery 62[ONS Suppl 1]:ONS266–ONS272, 2008

E

ndovascular treatment is now well established as a management alternative in neurosurgery. Recent advancements in endovascular therapy have resulted from improvements in therapeutic devices, such as microcatheters, embolic coils, stents, and advances in imaging systems. Medical imaging systems have revolutionized clinical diagnosis and therapeutic management in the field of neurosurgery.


DOI: 10.1227/01.NEU.0000297102.11637.35

One of the advancements in imaging systems is soft-tissue visualization (DynaCT), which uses a rotating c-arm system. This imaging technology creates computed tomographic (CT)-like images of soft tissue, such as brain tissue, cerebral cisterns, ventricles, or hematomas, using a rotating c-arm unit, as in digital subtraction angiography (DSA). This technology enables the physician to obtain CT scans during or after angiography without the need


DYNACT IN THE ENDOVASCULAR OPERATING ROOM

TABLE 1. Summary of patient detailsa Diagnosis

II

Number FD

Total

SAH

16

14

30

AVM

12

3

15

Tumor

11

6

17

Carotid stent

4

2

6

Spine

4

5

9

Unruptured aneurysm

2

8

10

Angioma

2

0

2

Skin lesion

2

3

5

grated neurosurgical and endovascular capabilities (8). The facility helps accommodate the expansion of neurosurgery from the full spectrum of conventional surgery into more widely used minimally invasive endovascular neurosurgery. The prototype DynaCT software was originally installed on a conventional II system (AXIOM Artis BA). In addition to the AXIOM Artis BA system, a newly developed FD biplane unit (AXIOM Artis dBA) with new DynaCT software was installed in the second endovascular OR (Fig. 1).

Data Acquisition and Presentation

The II Group (Prototype) The c-arm with a 33-cm image intensifier makes a single rotation around the z-axis of the object. During the 11-second run, the rotation angle is 220 degrees with a 0.8-degree increment, revealing 545 projections with a matrix of 512 ⫻ 512 image elements.

Dural AVF

1

1

2

Abscess

1

1

2

Subdural hematoma

0

1

1

The FD Group (Final Product) The c-arm with a 38 ⫻ 30–cm flat detector rotates 220 degrees with a 0.4-degree increment, revealing 535 projections and a matrix of 1024 ⫻ 1024 image elements in 20 seconds (Fig. 2). The obtained volumetric data set was reconstructed immediately on the dedicated 3-D workstation in the OR. The total time needed to acquire the data set and recon-

Hydrocephalus Total

0

1

1

55

45

100

a

II, image intensifier; FD, flat-panel detector; SAH, subarachnoid hemorrhage; AVM, arteriovenous malformation; AVF, arteriovenous fistula.

to transfer the patient to a CT suite. The most striking feature of this imaging system is its simplicity. DynaCT also enables real-time feedback during endovascular and neurosurgical procedures. The purpose of this study was to evaluate the clinical efficacy and limitations of DynaCT using an angiographic system in neurosurgery.

A

PATIENTS AND METHODS Patient Population This investigation is comprised of 100 studies of 93 patients (50 males, 43 females), whose ages range from 0 to 88 years (mean age, 52 yr), and who underwent DynaCT for cerebrovascular disease, intracranial mass lesion, or spinal disease between March 2004 and October 2005 (Table 1). DynaCT is a commercially-approved software system that produces CT-like three-dimensional (3-D) imaging using a c-arm unit. DynaCT is a 3-D visualization tool obtained from angiographic images acquired through a rotational spin of the c-arm angiographic system but is not considered as a substitute for conventional CT imaging. The study was conducted under the hospital guidelines. The first 55 studies were performed with an image intensifier (II) unit (prototype), whereas the other 45 studies were obtained with a flat-panel detector (FD) unit. During admission, each patient underwent conventional CT scans with a 16-slice multidetector CT scanner (MDCT; SOMATOM Sensation 16; Siemens Medical Solutions, Erlangen, Germany).

B

System and Data Acquisition The imaging systems used were specially designed AXIOM Artis BA (conventional II unit) and AXIOM Artis dBA (FD detector unit) (Fig. 1) biplane neuroangiographic systems (Siemens Medical Solutions). The biplane DSA units were installed in the surgical operating room (OR). This newly designed state-of-the-art neurosurgery suite provides inte-

NEUROSURGERY

FIGURE 1. A, the newly designed state-of-the-art neurosurgery suite with the imaging system (AXIOM Artis dBA; Siemens Medical Solutions, Erlangen, Germany). B, surgical setup during actual case.


IRIE ET AL.

TABLE 2. Comparative evaluations of the image intensifier and flat panel detector groupsa Object

FIGURE 2. Scene of DynaCT operation. The cone beam computed tomographic (CT) system reconstructs three-dimensional images from two-dimensional projection images obtained during rotation of the cone beam x-ray source and detectors around the patient.

struct the images in 512 volume kernel is typically about 5 minutes but varies with protocol details. DynaCT images were displayed on a viewing monitor in front of the surgeon. Two display techniques were used: volume rendering technique for 3-D images, and thick multiplanar reconstruction for two-dimensional (2-D) images. Results are more commonly viewed by slice display via thick multiplanar reconstruction. This technique clearly shows small density differences in structures such as soft tissue.

II group

FD group

Cortical bone

⫹⫹

⫹⫹

Metallic material

⫹⫹

⫹⫹

Calcified lesion

⫹⫹

⫹⫹

Cerebral cortex

⫹

⫹⫹

Muscle

⫹

⫹⫹

Hematoma

⫹

⫹⫹

SAH (Fisher 2)

–

⫹⫹

SAH (Fisher 3)

⫹

⫹⫹

Enhanced tumor

⫹⫹

⫹⫹

Enhanced abscess

⫹⫹

⫹⫹

a II, image intensifier; FD, flat panel detector; ⫹⫹, well visualized; ⫹, visible; –, not visible; SAH, subarachnoid hemorrhage.

RESULTS DynaCT was performed successfully in all patients. Highcontrast structures, such as bone, calcified lesions, and metallic materials, were visualized on DynaCT as well as in MDCT for each group. Although contrast differentiation of soft tissue such as cerebral cortex, muscle, and hematoma on DynaCT were not as good as those obtained from MDCT, the images were acceptable as intraoperative CT-like images to detect small hematoma or SAH. The images in the FD group were found to be superior to those from the II group; striking ring artifacts were observed in the II group. Comparative evaluation of the II (prototype) and FD groups is shown in Table 2. DynaCT images were highly useful in the following conditions: premature rupture of cerebral aneurysms during embolization, ventricular drainage, and spine instrumentation.

Imaging Characteristics

Subarachnoid Hemorrhage Thick, diffuse subarachnoid hemorrhage (SAH) was easily visualized, but thin SAH was more difficult to detect on the II unit. Fisher Grade 2 SAH was clearly visible with the FD system (Fig. 3). The sensitivity of DynaCT for SAH was 54.5% in the II group (six of 11 patients) and 100% in the FD group (14 of 14 patients). Figure 4 demonstrates the rupture of an anterior communicating aneurysm associated with a large frontal arteriovenous malformation (AVM). DynaCT images and 3-D angiograms were obtained with the same c-arm unit.


FIGURE 3. A DynaCT image (A) and multidetector CT (MDCT) image (B) showing Fisher Grade 2 subarachnoid hemorrhage.

Intracerebral Hematoma Small hematomas were not clearly detected in the II group in comparison with conventional CT imaging, although a large hematoma with ruptured AVM was visible. The FD group clearly demonstrated intracranial hematomas (Figs. 5 and 6).

Carotid Stenting DynaCT was performed during carotid artery stenting in six patients. There was clear visualization of the cobalt and tantalum alloy stent. Verification of stent expansion and fixation was smooth and easy (Fig. 7).

Brain Tumors Calcified tumors and tumors that exhibited enhancement with contrast medium could be visualized (Fig. 8). This improved our understanding of complex vascular pathologies and their anatomic relationship to the cranium or ventricles.

Other Applications Intraoperative scanning was performed on eight patients during spinal surgery in this OR. Titanium implants were



Although extravasation was controlled, immediately after embolization, DynaCT showed a significant hematoma in the sylvian fissure and diffuse subarachnoid hemorrhage. Subsequently, hematoma evacuation and surgical clipping were performed in the same endovascular OR. After clipping, DynaCT showed satisfactory evacuation of the hematoma. The patient was discharged 15 days after the procedure without deficit.

DISCUSSION Concept of DynaCT

FIGURE 4. DynaCT images of a ruptured anterior communicating artery aneurysm associated with a large frontal arteriovenous malformation.

FIGURE 5. A DynaCT image (A) and MDCT image (B) showing an intracerebral hematoma associated with ruptured middle cerebral artery aneurysm.

clearly visualized, and we were also able to verify the extent of osseous decompression. The prototype DynaCT on the II-based system was adequate for use in spinal surgery. DynaCT was also performed in the following cases: angioma, brain abscess, facial AVM, dural arteriovenous fistula, and placement of a ventricular drainage system for patients with acute hydrocephalus or high intracranial pressure. The appearance of enhancing brain abscess wall was similar to that of MDCT.

Illustrative Case A 71-year-old woman presented with long-term headache; magnetic resonance angiography demonstrated an unruptured middle cerebral artery aneurysm (Fig. 9). Cerebral angiography was performed, which demonstrated an irregular-shaped, 5-mm aneurysm of the right middle cerebral artery. Endovascular treatment was performed. During placement of the microcatheter, aneurysm perforation occurred. Angiography revealed significant extravasation. Heparin was immediately reversed and a coil was deployed to stop the bleeding. An additional coil was placed inside the aneurysm, resulting in satisfactory occlusion.

NEUROSURGERY

DynaCT uses a cone beam CT technique. 3-D images are reconstructed from 2-D projection images acquired during rotation of the cone beam x-ray source and detector around the patient (2). 3-D rotation angiography (3D-DSA) uses essentially the same imaging technique as cone beam computed tomography (6). Cone beam computed tomography has been developed during the past two decades (2). Until recently, studies of cone beam computed tomography were largely technical and focused on theory and algorithms, with rare clinical studies. This technique had been used primarily for imaging high-contrast structures such as in dental imaging or imaging of contrast medium–filled vascular structures (7). Recent developments of FDs have improved image quality and decreased radiation dose in FDs that have higher dynamic range than charge-coupled device cameras. The FD delivers a digital signal of 14 bits, whereas a charge-coupled device camera delivers only 12 bits (1, 5). In the present study, the II-based system achieved relatively low-contrast resolution in the region of 50 Hounsfield units (HU) for 10mm objects. Kalender (5) reported phantom measurements on the image quality with DynaCT. The flat-panel detector system achieved 10 HU density for a 10-mm object under comparable conditions. Fresh blood is typically in the range of 20 HU. Artifacts, particularly ring artifacts, are a structural limitation in II systems. These artifacts are caused by miscalibration in II response and drift and are not apparent in FD-based DynaCT systems. The development of new software that reduces ring artifacts, scatter artifacts, and truncation artifacts has improved image quality.

Comparison with MDCT We found that in comparison with MDCT, DynaCT was able to scan a wider area in a shorter time and delivered superior quality coronal and sagittal reconstruction images, although artifacts were problematic. Images with low radiographic contrast were of lower quality than those from MDCT. The current version of DynaCT could not emulate the contrast resolution of MDCT; however, at least FD-based DynaCT image quality was sufficient to identify small amounts of SAH or intracerebral hematoma.

Advantages of DynaCT DynaCT has the benefit of enabling immediate availability of CT-like images in the angiography suite without the need to move the patient (3). This new imaging method enables realtime feedback during neurosurgery and enables bleeding to be


IRIE ET AL.

A

B

C

tures were not well visualized with their equipment. The image-acquisition volume was limited to 12 cm3. Our system has more than 30 cm of detectors and provides whole-brain scanning in a single rotation. Another advantage of DynaCT is its potential use as a navigational tool. We use DynaCT during ventriculostomy, during which the location of the drainage tube tip can be confirmed in the OR. DynaCT is also useful for spine instrumentation. Before closing, DynaCT clearly demonstrates the location of the spinal implants. Surgical resection of enhanced malignant brain tumor can also be performed with DynaCT. After tumor removal, DynaCT can demonstrate remnant tumor during surgery.

Limitations The current system still has several limitations in its performance. First, artifacts and calibration problems limit image quality. Second, contrast differentiation is inferior in areas of low radiographic contrast compared with conFIGURE 6. Images of a 19-year-old male with acute headache and visual field defects who had been diagnosed with ventional CT imaging. In ruptured arteriovenous malformation in the right occipital lobe and underwent endovascular treatment. A, MDCT addition, reconstruction still images showing paraventrical hematoma and nidus in the right occipital lobe. B, prototype DynaCT images showing the same lesion with striking ring artifacts. C, flat-panel detector-based DynaCT images 1 year after hemrequires 5 to 10 minutes, and orrhage showing the lesion more clearly. it is still challenging to obtain perfusion imaging with the DynaCT system. Third, the observed during the procedure. This system can aid in diagnostotal radiation dose is 236 mGy in FD-based DynaCT, whereas ing a critical state. Before the introduction of this system, techthe dose in 3D-DSA using the same system is approximately 50 nical complications that occurred during a procedure would mGy. Fourth, the DynaCT procedure has not been accepted as require patients to be transferred immediately to the operative CT scanning under the Japanese insurance system. It is still suite from the angiography suite, increasing the risk of delayed considered fluororadiography. Appropriate procedural coding emergency procedures and adverse clinical outcomes. With the for DynaCT should be considered in the near future. new system, the workflow is seamless and efficient; endovascuIn this investigation, we could not conduct a quantitative lar surgery, diagnosis, and open surgery are all performed in comparative study of MDCT and DynaCT in the clinical setthe same room without transferring the patient. ting. Although a phantom study has already been reported (5), Hott et al. (4) describe their experience using Iso-C c-arm we were unable to conduct a quantitative study in the clinical multiplanar 3-D imaging (Siemens Medical Solutions) for intrasetting. One of the reasons was continuous upgrading of the operative surgical navigation and note the usefulness of Iso-C software to improve ring artifact. Therefore, imaging comparimaging as well as several limitations. Most cases were spinal ison of MDCT and DynaCT was subjective, and a strict quanprocedures, and they noted that intracranial soft-tissue structitative study should be planned in the near future.




A

B

C

D

E

F

FIGURE 7. DynaCT images of a carotid artery stent for a stenotic lesion.

G

FIGURE 8. DynaCT images of a malignant lymphoma on the left frontal lobe.

Feasibility of DynaCT In minimally invasive therapy, easily obtained and accurate intraoperative images are indispensable. DynaCT enables the detection of hemorrhagic complications during neuroendovascular therapy. In neurosurgery, DynaCT will be a useful tool for surgical navigation, stereotactic surgery, and tumor removal. Measurement of cerebral vascular flow might also be possible in the future; DynaCT will be a powerful modality in therapy for cerebrovascular disease.

NEUROSURGERY

FIGURE 9. Example of intraoperative rupture of incidental middle cerebral artery (MCA) aneurysm. A, digital subtraction angiogram showing an irregularly shaped, small MCA aneurysm. B, during placement of the microcatheter inside the aneurysm, perforation occurred and an angiogram demonstrated a large amount of contrast extravasation. C, DynaCT image immediately after coiling demonstrating a marked hematoma in the Sylvian fissure and diffuse subarachnoid hemorrhage. D, postoperative DynaCT image demonstrating near-complete evacuation of the hematoma. E, intraoperative angiogram demonstrating successful clipping without remnant. F, postoperative MDCT image showing compatible findings with DynaCT image. G, surgical setup during this case.

CONCLUSION In this study, DynaCT was successfully used in the surgical OR. This system enables us to expand into a wider range of therapeutic opportunities and detect early clinical complications during endovascular and surgical procedures. Further clinical experience and improvements in hardware and software are necessary to achieve higher contrast resolution such as that in multidetector computed tomography.


IRIE ET AL.

REFERENCES 1. Baba R, Konno Y, Ueda K, Ikeda S: Comparison of flat-panel detector and image-intensifier detector for cone-beam CT. Comput Med Imaging Graph 26:153–158, 2002. 2. Feldkamp LA, Davis LC, Kress JW: A practical cone-beam algorithm. J Opt Soc Am 1:612–619, 1984. 3. Heran NS, Song JK, Namba K, Smith W, Niimi Y, Berenstein A: The utility of DynaCT in neuroendovascular procedures. AJNR Am J Neuroradiol 27:330–332, 2006. 4. Hott JS, Deshmukh VR, Klopfenstein JD, Sonntag VK, Dickman CA, Spetzler RF, Papadopoulos SM: Intraoperative Iso-C C-arm navigation in craniospinal surgery: The first 60 cases. Neurosurgery 54:1131–1137, 2004. 5. Kalender WA: The use of flat-panel detector CT for imaging [in German]. Radiologe 43:379–387, 2003. 6. Kumazaki T: Development of rotational digital angiography and new conebeam 3D image: Clinical value in vascular lesions. Comput Methods Programs Biomed 57:139–142, 1998. 7. Mozzo P, Procacci C, Tacconi A, Martini PT, Andreis IA: A new volumetric CT machine for dental imaging based on the cone-beam technique: Preliminary results. Eur Radiol 8:1558–1564, 1998. 8. Murayama Y, Saguchi T, Ishibashi T, Ebara M, Takao H, Irie K, Ikeuchi S, Onoue H, Ogawa T, Abe T: Endovascular operative suite: Future directions for treating neurovascular disease. J Neurosurg 104:925-930, 2006.

Acknowledgments We thank Akito Hayashi from the Siemens-Asahi Medical Technologies Ltd. AX Business Management Medical Solutions Marketing Division for technical support.

tive imaging serves as a means of immediate quality control. DynaCT seems to be a promising possibility for intraoperative imaging in vascular procedures. Christopher Nimsky Erlangen, Germany

T

his article describes the use of a three-dimensional fluoroscopicbased imaging system that combines computed tomography-like quality images and biplanar angiography in the same operating suite. They describe the use of this technology in 100 consecutive patients and report no complications. The authors provide an example of how this technology altered the endovascular treatment of an intracranial aneurysm. However, although they present a subjective comparison of the quality of the images from DynaCT and the multidetector computed tomography, they lack any objective comparison. This technology is a welcome addition to the burgeoning field of intraoperative three-dimensional imaging, and we look forward to more rigorous analyses of its use. Eric Horn Robert F. Spetzler Phoenix, Arizona

I

rie et al. have presented their experience with the DynaCT for neurosurgical applications. The utility and advantage provided by such technology are significant. They make a good case for expanded use of such imaging technologies. For their meticulous presentation, the authors are to be congratulated.

COMMENTS

I

n the last few years, there has been increasing interest in the use of intraoperative imaging to evaluate the extent of a resection and to update navigational systems compensating for brain shift. Magnetic resonance imaging, computed tomography, and ultrasound are applied for resection control in various setups and imager designs. For neurovascular procedures, however, none of these designs has been proven to be perfect, so intraoperative angiography still remains an important tool in the treatment of complicated vascular malformations. Irie et al. present the implementation of recent, state-of-the-art vascular imaging in an operating theatre. Use of a biplanar angiography unit with flat panel detectors not only allows the obtaining of highresolution digital subtraction angiography images but also reconstructing of computed tomography-like images; this feature is called DynaCT. The configuration with the flat-panel detectors provided better image quality than the older system with the standard image intensifiers. The quality of the DynaCT image cannot yet fully compete with the quality of standard high-resolution 32- or 64-slice computed tomographic scans; however, for the intraoperative visualization of bony structures and for imaging after application of contrast material, DynaCT proved to be of additional value. The authors should be encouraged to analyze the effects of intraoperative imaging with DynaCT on the actual surgical procedure in more detail in the future; it is hoped that this analysis will document a benefit for patients. In addition, manufacturers should be encouraged to further improve image quality so that this technique can be used as a real intraoperative imaging modality to substitute for the placement of standard computed tomographic scanners in an operating room. Intraoperative imaging provides an objective evaluation of the intraoperative situation for the potential benefit of the patient, independent of the technology implemented and actual setup; intraopera-


Edward C. Benzel Cleveland, Ohio

T

he authors have presented us with the results from their extensive experience using a new intraoperative imaging system. The device used is very similar to, but more advanced than, the Siemens Iso-C, which is familiar to many surgeons in the United States. The applications for this type of “real-time” imaging are innumerable, and it is superior to traditional image-guidance systems that rely heavily on preoperative imaging (and are thus susceptible to the problems of reference localization, brain shift, and spinal manipulation). That being said, combination technologies using navigational devices coupled to three-dimensional intraoperative imaging have now become available and offer the best of all possible worlds. As with any new technology, the nuances of device application are critical, and the authors have illustrated some of these points nicely. For example, soft tissue visualization can be problematic, and radioopaque devices in the field of examination can degrade the image quality. In addition, the capital equipment costs associated with these instruments can be prohibitive for smaller hospitals. The authors make a cogent argument with case examples of how the DynaCT can allow for the prompt diagnosis of intraoperative complications and thus minimize the delays and morbidities associated with scanning a patient after the procedure and then returning to the operating theatre. I thus congratulate the authors on their fine work but would like to see a follow-up study comparing their current results with the preDynaCT era to identify in what percentage of patients management was altered. Michael Y. Wang Miami, Florida


GENERAL Technical Note Francesco Biroli, M.D. Division of Neurosurgery, Ospedali Riuniti di Bergamo, Bergamo, Italy

Felice Esposito, M.D., Ph.D. Department of Neurological Sciences, Division of Neurosurgery, Università degli Studi di Napoli Federico II, Naples, Italy

Mario Fusco, M.D. Division of Neurosurgery, Ospedali Riuniti di Bergamo, Bergamo, Italy

Giorgio G. Bani, M.D. Division of Neurosurgery, Ospedali Riuniti di Bergamo, Bergamo, Italy

Antonio Signorelli, M.D. Division of Neurosurgery, Ospedali Riuniti di Bergamo, Bergamo, Italy

Oreste de Divitiis, M.D. Department of Neurological Sciences, Division of Neurosurgery, Università degli Studi di Napoli Federico II, Naples, Italy

Paolo Cappabianca, M.D. Department of Neurological Sciences, Division of Neurosurgery, Università degli Studi di Napoli Federico II, Naples, Italy

NOVEL EQUINE COLLAGEN-ONLY DURAL SUBSTITUTE OBJECTIVE: A watertight and meticulous dural closure is an essential step after intradural neurosurgical procedures. When such a task cannot be performed, dural replacement materials and other adjunctive measures can provide an effective barrier between the subarachnoid compartment and the extradural space. METHODS: We present our experience with a novel collagen-derived dural substitute in a series of 114 patients undergoing a variety of neurosurgical procedures. The patients were clinically or neuroradiologically observed, for immediate and delayed local or systemic complications related to the implant. In three patients who underwent reoperation after decompressive duraplasty and craniectomy for bone flap repositioning, we performed biopsy of the dural implant for histopathological studies. RESULTS: None of the patients experienced local or systemic complications or toxicity related to the dural patch. None of the patients experienced a postoperative cerebrospinal fluid fistula, except one patient who underwent an endoscopic endonasal transsphenoidal marsupialization of a large intrasuprasellar arachnoid cyst; the fistula required reoperation for cerebrospinal fluid fistula repair and intravenous antibiotic therapy for bacterial meningitis. Postoperative magnetic resonance imaging scans showed signs of severe inflammatory response in only one patient who did not present any postoperative clinical symptom or neurological deficits. Three patients underwent reoperation for bone flap repositioning after decompressive craniectomy; in all patients, the dural substitute appeared to have promoted satisfactory dural regeneration, as confirmed by the histological studies. Furthermore, in such patients, no or minimal adherence with the other tissues and the brain cortex was observed. CONCLUSION: This study demonstrates that the new collagen-only biomatrix is a safe and effective dural substitute for routine neurosurgical procedures. The absence of local and systemic toxicity or complications and the scarce promotion of adherences and inflammation make this material appealing for its use as a dural substitute, even in cases in which the necessity of reoperation is foreseen. KEY WORDS: Clinical study, Collagen, Dural repair, Dural substitute, Efficacy Neurosurgery 62[ONS Suppl 1]:ONSE273–ONSE274, 2008

DOI: 10.1227/01.NEU.0000297048.04906.5B0

Luigi M. Cavallo, M.D., Ph.D. Department of Neurological Sciences, Division of Neurosurgery, Università degli Studi di Napoli Federico II, Naples, Italy Reprint requests: Paolo Cappabianca, M.D., Università degli Studi di Napoli Federico II, Department of Neurological Sciences, Division of Neurosurgery, Via Sergio Pansini, 5-80131 Naples, Italy. Email: [email protected] Received, June 4, 2006. Accepted, August 1, 2007.

NEUROSURGERY

I

nadequate closure of the dura at the end of neurosurgical procedures exposes the patient to possible postoperative complications resulting from an abnormal communication between the subarachnoid compartment and the extradural space, such as a cerebrospinal fluid (CSF) fistula, infections, hypertensive pneumocephalus, pseudomeningocele, and other problems that cause longer hospitalization. When germproof, watertight closure of the dural membrane cannot be performed as a result of neoplastic invasion, tearing, or shrinkage during long operations, or in some operations such as for Chiari malformations,

posttraumatic/poststroke decompressive craniectomies, hydromyelia, and cervical myelopathy, a plastic extension of the dura mater may be indicated (8, 11, 21, 22). In common neurosurgical practice, a variety of measures for dural closing have been implemented, such as the use of autologous transplants and homologous, heterologous, and alloplastic grafts and materials. Virtually all of the dural grafts used have been associated with sequelae, some of which have been major (1, 5, 6, 8, 10, 11, 15–18, 20, 22, 23). In recent years, the attention of investigators and neurosurgeons has moved toward research and use of an


BIROLI ET AL.

“ideal dural substitute.” Such a dural substitute should satisfy some requirements to be considered “ideal” for clinical applications (2, 4, 10, 22), including 1) scarce or no immune or inflammatory response; 2) no local or systemic toxicity; 3) promotion of neodura formation and connective architecture restoration; 4) dura-like elasticity and resistance; 5) minimal or no adherence to surrounding tissues; 6) safe sterilization; 7) quick availability, easy handling, and effective application; 8) easy removal when necessary; 9) economic convenience; and 10) adequate scientific studies supporting its use. In these settings, collagen-based heterologous products have become increasingly popular. They have a double advantage (9, 13, 14) in that they do not promote an inflammatory response or a foreign body reaction and the collagen fiber network acts as a matrix for growth of endogenous neodura. We present our experience with a new collagen-based product derived from equine Achilles tendon (Tissudura; Baxter, Vienna, Austria) for dural replacement in a series of 114 patients who underwent a variety of neurosurgical procedures requiring a dural graft implantation and with a minimum follow-up period of 6 months.

MATERIALS AND METHODS At the Division of Neurosurgery of the Ospedali Riuniti di Bergamo (Bergamo, Italy) and the Division of Neurosurgery of the Università degli Studi di Napoli Federico II (Naples, Italy), a total of 129 patients were enrolled between January 2004 and December 2005. The approval of the Committee for Medical Ethics of both institutions was received. Because 15 patients did not have any postoperative neuroimaging study at a minimum of 3 months or any clinical follow-up, only 114 patients were actually considered in the clinical material for this study. The data from the enrolled patients were stored in an ad hoc database created with commercially available software (FileMaker Pro 7.0, Santa Clara, CA). The patients were adequately instructed on the possible necessity of the Tissudura implantation and gave their written informed consent because no clinical studies on the effectiveness of this product in routine neurosurgery are available. Patients who underwent reoperation after the implantation gave their specific informed consent for the removal of a specimen of the implanted graft for pathological study.

Patient Population The patients’ age and sex, general conditions and comorbidities, if any, operative diagnosis, type of neurosurgical procedure, duration of the surgical operation, eventual factors causing perioperative contamination of the site of the implant, length of postoperative hospitalization, and eventual adjuvant therapies—namely, radiotherapy for malignant tumors—were recorded. We decided to use the heterologous dural graft with Tissudura only in patients in whom the common measures for dural closing proved to be ineffective, even a priori or after trying.

nurse should be aware that when dehydrated, it is sticky to the surgical gloves in the presence of blood; he or she should use clean gloves or handle the dehydrated graft with clean instruments. Rehydration in physiological saline for 2 to 5 minutes produces an approximately 2-mm-thick, smooth, transparent film, easy to handle and to cut in the desired shape. The rehydrated graft is not sticky anymore. It has a soft consistency and is relatively elastic and resistant to traction; it can be easily folded and unfolded back to the desired shape, unlike other collagen products, which once folded are difficult to reconstitute into the previous shape. The dural defect to be repaired is measured. Once the graft has been adequately prepared, it is cut to obtain a patch larger than the immediate area of the defect. It is then applied over the defect in either an underlay or an overlay fashion. In general, the underlay position of the patch should be preferred (Fig. 1A). The edges of the patch overlap the edges of the patient’s dural defect and the patch is adjusted with blunt instruments. The application of stitches between the patch and the patient’s dura is generally not required but remains an option. Given that the hydrated Tissudura is a transparent patch, the efficacy of the hemostasis can still be verified after the application. Fibrin glue (Tisseel; Baxter, Vienna, Austria) is then applied over the patch, especially over the overlapping edges (Fig. 1B). Indeed, the use of the fibrin glue is not mandatory or specifically required, but it has been shown that such glue may accelerate the ingrowth of the fibroblasts inside the graft for the neodura formation (10, unpublished data).

Use in Transsphenoidal Surgery The use of Tissudura in transsphenoidal surgery requires some special considerations, which have been reported previously (2–4). Patients undergoing transsphenoidal surgery require sellar dura reconstruction only if an intraoperative CSF leak occurs or if the suprasellar arachnoid cistern falls into the sellar cavity. In cases with a small weeping of CSF without any obvious rent in the suprasellar cistern, packing of the sella and of the sphenoid sinus is performed, according to the technique previously described (2); in these cases, the Tissudura is placed against the arachnoid membrane, followed by the other materials used for the repair. In these cases, we followed the paradigm of graded sellar repair with multiple materials (2–4, 7); first, a layer of Tissudura is placed intradurally against the arachnoid, collagen sponges fill the intrasellar dead space, a solid or semisolid buttress holds the sealing tissue in position and maintains the watertight seal to resist the pulsations of the brain and CSF, and more collagen fleeces fill in the sphenoid sinus. Finally, in patients in whom the suprasellar descends into the sellar cavity, without any evidence of CSF escape, the collagen Tissudura is used to protect and enforce the arachnoidal membrane to prevent postoperative rupture and consequent CSF leak. Fibrin glue is also used.

A

B

Intraoperative Handling Tissudura is a collagen-only dural substitute composed of colloidal collagen (mainly Type I collagen) produced from purified minced Achilles tendon of horses and precipitated in fibrils. The collagen biomatrix product contains 5.6 mg of native collagen fibrils, not chemically crosslinked, per square centimeter. The absence of other proteins is confirmed by immunodiffusion tests (10). Before use, the matrix is a dehydrated sheet with papyraceous consistency. In this form, it is fragile and difficult to handle. The scrub


FIGURE 1. Intraoperative images showing Tissudura graft implant. A, the patch is positioned in an underlay fashion. B, fibrin glue is applied over the patch, especially over the overlapping edges.


COLLAGEN-ONLY FOIL AS A NOVEL DURAL SUBSTITUTE

Postoperative Evaluation Patients underwent postoperative neuroradiological evaluation with contrast-enhanced head magnetic resonance imaging (MRI) scans at a median of 3 months after surgery. Radiological findings suggestive of tissue integration (y ⫽ yes; n ⫽ no) and inflammatory response (0 ⫽ no signs of inflammation; 1 ⫽ scarce; 2 ⫽ moderate; and 3 ⫽ severe) were evaluated. Furthermore, all patients were observed for possible local or systemic signs of product-related toxicity. For patients undergoing reoperation, the macroscopic appearance of the patch (if still present and not reabsorbed), the thickness of the newly developed dura-like tissue, the presence of neoangiogenesis, the presence of adherences, and signs of inflammatory response were evaluated.

RESULTS Patient Population Of the 114 patients who completed the study, 54 (47.4%) were male and 60 (52.6%) female. Their ages ranged between 2 and 80 years (mean, 49.8 yr; median, 54 yr). Fifty-three patients underwent supratentorial craniotomy for brain tumors (both malignant and benign) or nontumoral lesions; 24 patients had a transsphenoidal operation for sellar/perisellar lesions; 15 patients required a life-saving decompressive craniectomy and duraplasty for posttraumatic uncontrolled brain swelling; 13 patients had an operation, either craniotomy or craniectomy, for infratentorial/posterior cranial fossa lesions; and nine patients underwent spinal operations (three had laminectomy for intradural tumors and six had microdiscectomy for lumbar disc herniation). The length of the surgical operation was between 60 and 615 minutes, depending on its complexity (mean, 220.4 min; median, 200 min). In three patients, there was an evident contamination of the surgical field and, therefore, of the site of the implant. The patients were discharged from the hospital after an average of 8.9 days (range, 2–51 d; median, 7 d).

transsphenoidal operation, the Tissudura was used for sellar dural reconstruction.

Outcomes, Complications, and Postoperative Neuroradiological Studies No patients experienced any local or systemic complication or toxicity related to the dural patch. One patient with a large intrasuprasellar arachnoid cyst who underwent endoscopic endonasal transsphenoidal marsupialization of the cysts with subsequent sellar packing and dural and sellar reconstruction experienced postoperative CSF rhinorrhea and bacterial meningitis, which was successfully treated with reoperation for CSF fistula repair and intravenous antibiotic therapy. Based on the 3-month postoperative contrast-enhanced MRI studies available for 92 of the 114 patients (80.7%), most of the patients showed no or only a low-grade inflammatory reaction (62 patients, no visible reaction; 16 patients, scarce reaction; 13 patients, moderate reaction), whereas in one patient, the MRI scan showed a severe inflammatory reaction (Fig. 2). In addition, in all 92 patients, there was a verisimilar integration of the patch with the patient’s tissues. Four patients underwent reoperations: one for CSF leak repair after transsphenoidal operation and three for bone flap

A

B

C

D

Necessity of Dural Grafting The implantation of Tissudura was required for different reasons. For patients undergoing supratentorial craniotomy or posterior fossa surgery for intracranial lesions and for those who had a laminectomy for a spinal intradural lesion, the dural patch application was needed as a dural substitute because 1) the dura of the patient was infiltrated by the tumor (as in cases of malignant tumors with dural involvement or meningiomas), and its removal together with the lesion was required; 2) it was torn during the operation; or 3) at the end of the operation, the patient’s dura mater was retracted and a watertight closure was not possible. In all six patients who underwent lumbar microdiscectomy for disc herniation, there was an unwanted lumbar dural puncture and the patch was applied to prevent a postoperative CSF fistula or postoperative pseudomeningocele. For patients who underwent decompressive durotomy and craniectomy for traumatic brain injury, a plastic extension of the dura was obviously possible only with materials other than the patient’s dura. Finally, for patients who underwent

NEUROSURGERY

FIGURE 2. Contrast-enhanced preoperative axial (A), preoperative coronal (B), postoperative axial (C), and postoperative coronal (D) brain magnetic resonance imaging scans showing a patient with recurrent left frontal convexity meningioma. C and D, a thin hyperintense signal corresponding to the graft implant is consistent with inflammatory reaction.


BIROLI ET AL.

repositioning after decompressive craniectomy. In the patient with CSF rhinorrhea, this occurred a few days after the first operation, and the attachment of the graft could not be evaluated; therefore, in this patient, the Tissudura remained intact. On the contrary, for patients who required bone flap repositioning, such an operation was performed at a median of 7 months after the grafting. In all patients, dura-like tissue formation was noted, but the thickness of the newly developed tissue varied; in two patients, a normotrophic membrane was evident, whereas in one patient, the membrane appeared to be quite thin and transparent. In addition, in all three patients, there was minimal adherence between the neodura and the skin and between the neodura and the brain cortex, thus resulting in only a scarce tendency for scar and adherence formation.

DISCUSSION Tissudura is one of the newest commercially available collagen-derived materials for dural grafting. This material has been studied previously only in animal experiments (10) and for use in endoscopic endonasal transsphenoidal surgery (4). We studied this material in a variety of routine neurosurgical procedures to assess its efficacy and safety as a dural substitute. It is known from preclinical studies (10), from comparisons with other collagen-derived dural substitutes (9, 13, 14), and from our experience (4) with its use that Tissudura has some peculiar characteristics: 1) Tissudura has a lamellar structure (with a virtual absence of pores) that prevents CSF escape, which is different from other collagen sponges that present different grades of porosity. 2) Tissudura promotes neodura formation. In these aspects, it is similar to other collagen sponges (9, 13, 14, 19). Thanks to the presence of natural crosslinks of the collagen fibrils, it favors cell adhesion, proliferation, migration, and differentiation. This facilitates rapid fibroblast ingrowth, collagen formation, angiogenesis, and sealing of dural defects. This advantage is improved by the concomitant use of fibrin glue (3). 3) Adaptability is another advantage of the lamellar structure of Tissudura. Unlike other collagen-derived substitutes, which tend to fold and shrink if wet, Tissudura maintains its dimensions, especially in wet conditions; furthermore, it is flexible, elastic, and easily shaped. 4) It also offers protection against bovine spongiform encephalopathy transmission; the collagen is not a bovine product, but it derives from equine flexor tendon and is inactivated with sodium hydroxide and chloriduric acid. The material is composed mainly of Type I collagen, which is the same collagen subtype that composes the dura mater (12, 13), and collagen fibers may serve as a scaffold for cell ingrowth into the graft and, therefore, neodura formation (4, 9, 12, 13, 14). This fact has been clearly evident in the present study. In fact, all patients who had a reoperation after the dural graft placement showed satisfactory integration of the graft with tissues and evidence of formation of a dura-like membrane with neoangiogenesis formation. Type I collagen is only weakly immunogenic and soluble and thus is associated with scarce adhesions and inflammation. Our results are consistent with such findings as well. Only one patient had neuroradiological


evidence of inflammation but no clinical symptoms or any neurological deficit; this patient was operated on for a recurrent left frontal convexity meningioma. We noted that the patch was quite easy to place and to adjust under the patient’s dural margins when used in an underlay fashion. Furthermore, the patch does not need to be sutured to the patient’s dura, therefore making the operative times shorter. Finally, once put inside, the patch is transparent, thus allowing for evaluation of hemostatic efficacy, even with the dura closed.

CONCLUSION Our data seem to confirm that the advantages of this new dural substitute are at least comparable with those noted in previous studies of similar collagen-derived material. The absence of local and systemic toxicity or complications, together with the scarce promotion of adherences and inflammation, make this material appealing for use as a dural substitute, even in patients for whom reoperation is foreseen. Nevertheless, future studies and longer follow-up are needed to confirm the safety and efficacy of Tissudura.

Disclosure Baxter had no input into the design and results of this study or the production of this article.

REFERENCES 1. Alleyne CH Jr, Barrow DL: Immune response in hosts with cadaveric dural grafts. Report of two cases. J Neurosurg 81:610–613, 1994. 2. Cappabianca P, Cavallo LM, Esposito F, Valente V, De Divitiis E: Sellar repair in endoscopic endonasal transsphenoidal surgery: Results of 170 cases. Neurosurgery 51:1365–1372, 2002. 3. Cappabianca P, Cavallo LM, Valente V, Romano I, D’Enza AI, Esposito F, de Divitiis E: Sellar repair with fibrin sealant and collagen fleece after endoscopic endonasal transsphenoidal surgery. Surg Neurol 62:227–233, 2004. 4. Cappabianca P, Esposito F, Cavallo LM, Messina A, Solari D, di Somma LG, de Divitiis E: Use of equine collagen foil as dura mater substitute in endoscopic endonasal transsphenoidal surgery. Surg Neurol 65:144–149, 2006. 5. Clavel M, Clavel P: Creutzfeldt-Jakob disease transmitted by dura mater graft. Eur Neurol 36:239–240, 1996. 6. Cohen AR, Aleksic S, Ransohoff J: Inflammatory reaction to synthetic dural substitute. Case report. J Neurosurg 70:633–635, 1989. 7. Esposito F, Dusick JR, Fatemi N, Kelly DF: Graded repair of cranial base defects and cerebrospinal fluid leaks in transsphenoidal surgery. Neurosurgery 60 [Suppl 1]:ONS1–ONS9, 2007. 8. Hida K, Yamaguchi S, Seki T, Yano S, Akino M, Terasaka S, Uchida T, Iwasaki Y: Nonsuture dural repair using polyglycolic acid mesh and fibrin glue: Clinical application to spinal surgery. Surg Neurol 65:136–143, 2006. 9. Kelly DF, Oskouian RJ, Fineman I: Collagen sponge repair of small cerebrospinal fluid leaks obviates tissue grafts and cerebrospinal fluid diversion after pituitary surgery. Neurosurgery 49:885–890, 2001. 10. Knopp U, Christmann F, Reusche E, Sepehrnia A: A new collagen biomatrix of equine origin versus a cadaveric dura graft for the repair of dural defects—A comparative animal experimental study. Acta Neurochir (Wien) 147:877–887, 2005. 11. Laun A, Tonn JC, Jerusalem C: Comparative study of lyophilized human dura mater and lyophilized bovine pericardium as dural substitutes in neurosurgery. Acta Neurochir (Wien) 107:16–21, 1990. 12. Maher CO, Anderson RE, McClelland RL, Link MJ: Evaluation of a novel propylene oxide-treated collagen material as a dural substitute. J Neurosurg 99:1070–1076, 2003. 13. Narotam PK, van Dellen JR, Bhoola KD: A clinicopathological study of collagen sponge as a dural graft in neurosurgery. J Neurosurg 82:406–412, 1995.


COLLAGEN-ONLY FOIL AS A NOVEL DURAL SUBSTITUTE

14. Narotam PK, Van Dellen JR, Bhoola K, Raidoo D: Experimental evaluation of collagen sponge as a dural graft. Br J Neurosurg 7:635–641, 1993. 15. Ng TH, Chan KH, Leung SY, Mann KS: An unusual complication of silastic dural substitute: Case report. Neurosurgery 27:491–493, 1990. 16. Parízek J, Meˇricka P, Spacek J, Sercl M, Eliás P, Neˇmecková J, Lichy´ J, Neˇmecek S: Repair of the dura mater in children using xenogenic pericardium [in Czech]. Cas Lek Cesk 128:682–684, 1989. 17. Robertson SC, Menezes AH: Hemorrhagic complications in association with silastic dural substitute: Pediatric and adult case reports with a review of the literature. Neurosurgery 40:201–206, 1997. 18. Simpson D, Robson A: Recurrent subarachnoid bleeding in association with dural substitute. Report of three cases. J Neurosurg 60:408–409, 1984. 19. Stark GB, Hork R, Tanczos E: Biological Matrix and Tissue Reconstruction. New York, Springer-Verlag, 1998. 20. Thadani V, Penar PL, Partington J, Kalb R, Janssen R, Schonberger LB, Rabkin CS, Prichard JW: Creutzfeldt-Jakob disease probably acquired from a cadaveric dura mater graft. Case report. J Neurosurg 69:766–769, 1988. 21. Vanaclocha V, Saiz-Sapena N: Duraplasty with freeze-dried cadaveric dura versus occipital pericranium for Chiari type I malformation: Comparative study. Acta Neurochir (Wien) 139:112–119, 1997. 22. von Wild KR: Examination of the safety and efficacy of an absorbable dura mater substitute (Dura Patch) in normal applications in neurosurgery. Surg Neurol 52:418–425, 1999. 23. Xu BZ, Pan HX, Li KM, Chen XJ, Tian YD, Li YL, Liu J: Study and clinical application of a porcine biomembrane for the repair of dural defects. J Neurosurg 69:707–711, 1988.

Acknowledgments This article has been supported in part by a research grant of the Italian Ministry for Education, University and Research (PRIN #2005069841).

COMMENTS

G

iven the ongoing challenges to prevent postoperative cerebrospinal fluid (CSF) leaks and their attendant complications, any technical advance in this area is always welcome. Biroli et al. have nicely demonstrated that, in their hands, this new equine-derived collagen biomatrix is an effective and safe dural substitute for use in a variety of intracranial and spinal applications. Whereas this equine product and the more commonly used bovine-derived collagen products (e.g., Duragen, Helistat, or Instat) both promote rapid fibroblast in-growth, one of the advantages of this collagen matrix appears to be that it is relatively nonporous compared with the bovine-derived version. Although this claim is not well substantiated in this report, as the authors note, prior studies suggest that this equinederived matrix has minimal porosity, which may translate to a lower risk of postoperative CSF leaks. Whether this collagen product is truly superior to the bovine products, however, remains unproven. Nonetheless, with this report, there is now increasing data that both the equine- and bovinederived collagen versions are reliable dural substitutes, given that they promote rapid formation of a living dural seal characterized by new collagen formation and angiogenesis. Additional advantages over other dural substitutes are that suturing of the matrix is not required and their use often obviates the need for autologous tissue grafts and lumbar CSF diversion. That said, intraoperative dural defects comprise a heterogeneous group of surgical challenges in both size and location. Ultimately, the technical nuances of how these products are used and with what adjuncts will determine the effectiveness of a given CSF leak repair method. Daniel Kelly Santa Monica, California

NEUROSURGERY

W

ith the advent of the extended transsphenoidal approaches for lesions of the anterior cranial base, the problem of repairing defects in the cranial base and preventing CSF leaks has become even more of a challenge. Because these approaches are extracranial and often done with endoscopic techniques, the use of standard measures such as suturing is compromised. Tissue grafts (fat, fascia and muscle are artificial dura) are not always reliable, and the various glue preparations are likewise not universally effective. Methods using mucoperichondrial flaps derived from the turbinates or the nasal septum have been proposed but again are not always reliable. The same is true of the innovative “gasket seal” approach. As we search for an ideal method, the appearance of a new inlay membranous material that is nonallergenic, perfectly sterilizable, and has the quality of self-adherence is most attractive. Surely, if this product becomes available in the United States, it will be widely tested, and it is hoped that the same degree of success as reported in this provocative article will be seen. Edward R. Laws, Jr. Boston, Massachusetts

T

his study evaluates the clinical use of Tissudura, a new product indicated for use in dural reconstruction. Biroli et al. achieved dural closure in 114 patients using this inlay substitute, which is composed of Type I collagen derived from the Achilles tendon of horses. Fifty-three patients underwent supratentorial craniotomy for removal of mass lesions, 24 patients underwent transsphenoidal surgery, 15 patients required expansile duraplasty after decompressive hemicraniectomy, 13 patients underwent surgery in the posterior fossa, and 9 patients underwent spinal surgery. Watertight closure was the rule, even without suturing, as the only CSF leak that occurred was in a high-risk situation (after transsphenoidal decompression of an intrasellar arachnoid cyst). There were no infectious complications related to the graft and, in the three patients who underwent craniotomy reoperations, it appeared that the implant had been replaced by significant ingrowth of native dura. Inflammatory response and adhesions either to underlying brain or to the soft tissues after craniectomy appeared to be minimal. Most collagen matrix dural substitutes seem to achieve watertight closure. Onlay or inlay use saves time in closing and spares the patient the tissue trauma associated with harvesting pericranium, temporalis fascia, or autologous tissue from elsewhere in the body. It remains to be seen whether routine use of dural substitutes is more cost-effective than harvesting autografts. Clear advantages to this product, however, include the ease with which it can be handled and conformed, making it particularly useful at the cranial base and for the challenging dural closures that can be associated with endonasal approaches. Also, the lack of adhesion formation probably saves operating room time and blood loss during the return for replacement of the bone flap after craniectomy. Headache after suboccipital craniotomy, sometimes hypothesized to arise from a local inflammatory response, could conceivably be reduced as well. The product appears to merit further evaluation by other clinical groups in the future. William T. Curry, Jr. Fred G. Barker II Boston, Massachusetts


OPERATIVE NEUROSURGERY TABLE MARCH 2008

OF

VOLUME 62

CONTENTS

BY

TOPIC

OPERATIVE NEUROSURGERY 1

Pages 1–277

Articles are grouped by categories, complete author listings can be found on preceding Table of Contents

ANATOMY

FUNCTIONAL NEUROSURGERY

1

209

9 18 24 30 38 145

Microsurgical Anatomy of the Supracerebellar Transtentorial Approach to the Posterior Mediobasal Temporal Region: Technical Considerations with a Case Illustration: Roham Moftakhar Microsurgical Anatomy of the Safe Entry Zones on the Anterolateral Brainstem Related to Surgical Approaches to Cavernous Malformations: Rodolfo Recalde Image-guided Lateral Suboccipital Approach: Part 1— Individualized Landmarks for Surgical Planning: Alireza Gharabaghi Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi Endoscopic Sublabial Transmaxillary Approach to the Rostral Middle Fossa: Bonnie C. Ong Medial Sphenoid Ridge Meningiomas: Classification, Microsurgical Anatomy, Operative Nuances, and Long-term Surgical Outcome in 35 Consecutive Patients: Stephen M. Russell Modifications of the Transoral Approach to the Craniovertebral Junction: Anatomic Study and Clinical Correlations: A. Samy Youssef

110 126 134 140 142

Microsurgical Anatomy of the Supracerebellar Transtentorial Approach to the Posterior Mediobasal Temporal Region: Technical Considerations with a Case Illustration: Roham Moftakhar Tentorial Dural Arteriovenous Fistulae: Operative Strategies and Microsurgical Results for Six Types: Michael T. Lawton Usefulness of Preoperative Three-dimensional Computed Tomographic Angiography with Two-dimensional Computed Tomographic Image for Rupture Point Detection of Middle Cerebral Artery Aneurysm: Kojiro Wada External Carotid Artery to Middle Cerebral Artery Bypass with the Saphenous Vein Graft: Erica F. Bisson Improved Image Interpretation with Combined Superselective and Standard Angiography (Double-Injection Technique) during Embolization of Arteriovenous Malformations: Tom L. Yao Combined Endoscopic-assisted Transclival Clipping and Endovascular Stenting of a Basilar Trunk Aneurysm: Jean A. Eloy

COMPLICATION 24 156 249

245 249

GENERAL NEUROSURGERY 262 273

Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi Cerebrospinal Fluid-related Complications with Autologous Duraplasty and Arachnoid Sparing in Type I Chiari Malformation: Caitlin E. Hoffman Methods of Scalp Revision for Deep Brain Stimulator Hardware: Alejandro M. Spiotta

18

82 91 102 126 140 201 209 226 251 266

CRANIAL BASE 18 24 30 57 75 251

Image-guided Lateral Suboccipital Approach: Part 1— Individualized Landmarks for Surgical Planning: Alireza Gharabaghi Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi Endoscopic Sublabial Transmaxillary Approach to the Rostral Middle Fossa: Bonnie C. Ong Endoscopic Endonasal Pituitary Transposition for a Transdorsum Sellae Approach to the Interpeduncular Cistern: Amin B. Kassam The Juxtacondylar Approach to the Jugular Foramen: Michaël Bruneau Quantification of the Frontotemporal Orbitozygomatic Approach Using a Threedimensional Visualization and Modeling Application: Anthony L. D’Ambrosio

ENDOSCOPY 30 51 57 105 108 142

Endoscopic Sublabial Transmaxillary Approach to the Rostral Middle Fossa: Bonnie C. Ong Purely Endoscopic Resection of Colloid Cysts: Jeremy D.W. Greenlee Endoscopic Endonasal Pituitary Transposition for a Transdorsum Sellae Approach to the Interpeduncular Cistern: Amin B. Kassam Endoscope-controlled Microneurosurgery to Treat Middle Fossa Epidermoid Cysts: Felipe Trivelato Endoscopic Supracerebellar Infratentorial Approach for Pineal Cyst Resection: Pankaj A. Gore Combined Endoscopic-assisted Transclival Clipping and Endovascular Stenting of a Basilar Trunk Aneurysm: Jean A. Eloy

140 142 266

Improved Image Interpretation with Combined Superselective and Standard Angiography (Double-Injection Technique) during Embolization of Arteriovenous Malformations: Tom L. Yao Combined Endoscopic-assisted Transclival Clipping and Endovascular Stenting of a Basilar Trunk Aneurysm: Jean A. Eloy DynaCT Soft-tissue Visualization Using an Angiographic C-arm System: Initial Clinical Experience in the Operating Room: Koreaki Irie

EPILEPSY 1

N4

162

Correction of Late Traumatic Thoracic and Thoracolumbar Kyphotic Spinal Deformities Using Posteriorly Placed Intervertebral Distraction Cages: Michael Y. Wang

METHDOLOGY IMPROVEMENT 209

Compensation of Geometric Distortion Effects on Intraoperative Magnetic Resonance Imaging for Enhanced Visualization in Image-guided Neurosurgery: Neculai Archip

MINIMALLY INVASIVE 30 51 57 105 108 140 142

Endoscopic Sublabial Transmaxillary Approach to the Rostral Middle Fossa: Bonnie C. Ong Purely Endoscopic Resection of Colloid Cysts: Jeremy D.W. Greenlee Endoscopic Endonasal Pituitary Transposition for a Transdorsum Sellae Approach to the Interpeduncular Cistern: Amin B. Kassam Endoscope-controlled Microneurosurgery to Treat Middle Fossa Epidermoid Cysts: Felipe Trivelato Endoscopic Supracerebellar Infratentorial Approach for Pineal Cyst Resection: Pankaj A. Gore Improved Image Interpretation with Combined Superselective and Standard Angiography (Double-Injection Technique) during Embolization of Arteriovenous Malformations: Tom L. Yao Combined Endoscopic-assisted Transclival Clipping and Endovascular Stenting of a Basilar Trunk Aneurysm: Jean A. Eloy DynaCT Soft-tissue Visualization Using an Angiographic C-arm System: Initial Clinical Experience in the Operating Room: Koreaki Irie

NAVIGATION 24

Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi

NEURONAVIGATION 18 82

Microsurgical Anatomy of the Supracerebellar Transtentorial Approach to the Posterior Mediobasal Temporal Region: Technical Considerations with a Case Illustration: Roham Moftakhar

Image-guided Lateral Suboccipital Approach: Part 1— Individualized Landmarks for Surgical Planning: Alireza Gharabaghi Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi The New Generation PoleStar N20 for Conventional Neurosurgical Operating Rooms: A Preliminary Report: Vasileios Ntoukas A Comparative Analysis of Coregistered Ultrasound and Magnetic Resonance Imaging in Neurosurgery: Alex Hartov Usefulness of Intraoperative Photodynamic Diagnosis Using 5-Aminolevulinic Acid for Meningiomas with Cranial Invasion: Yoichi Morofuji Usefulness of Preoperative Three-dimensional Computed Tomographic Angiography with Two-dimensional Computed Tomographic Image for Rupture Point Detection of Middle Cerebral Artery Aneurysm: Kojiro Wada Improved Image Interpretation with Combined Superselective and Standard Angiography (Double-Injection Technique) during Embolization of Arteriovenous Malformations: Tom L. Yao Fiducial versus Nonfiducial Neuronavigation Registration Assessment and Considerations of Accuracy: Wolfgang K. Pfisterer Compensation of Geometric Distortion Effects on Intraoperative Magnetic Resonance Imaging for Enhanced Visualization in Image-guided Neurosurgery: Neculai Archip Percutaneous Computed Tomography-guided Radiofrequency Ablation of Upper Spinal Cord Pain Pathways for Cancer-related Pain: Ahmed M. Raslan Quantification of the Frontotemporal Orbitozygomatic Approach Using a Threedimensional Visualization and Modeling Application: Anthony L. D’Ambrosio DynaCT Soft-tissue Visualization Using an Angiographic C-arm System: Initial Clinical Experience in the Operating Room: Koreaki Irie

INSTRUMENTATION

266

ENDOVASCULAR

Frontozygomatic Titanium Cranioplasty in Frontosphenotemporal (“Pterional”) Craniotomy: Shaan M. Raza Novel Equine Collagen-only Dural Substitute: Francesco Biroli

IMAGING 24

CEREBROVASCULAR 1

217

Compensation of Geometric Distortion Effects on Intraoperative Magnetic Resonance Imaging for Enhanced Visualization in Image-guided Neurosurgery: Neculai Archip Location of Active Contacts in Patients with Primary Dystonia Treated with Globus Pallidus Deep Brain Stimulation: Clement Hamani Use of an Integrated Platform System in the Placement of Deep Brain Stimulators: Gregory G. Heuer Methods of Scalp Revision for Deep Brain Stimulator Hardware: Alejandro M. Spiotta

91

Image-guided Lateral Suboccipital Approach: Part 1— Individualized Landmarks for Surgical Planning: Alireza Gharabaghi The New Generation PoleStar N20 for Conventional Neurosurgical Operating Rooms: A Preliminary Report: Vasileios Ntoukas A Comparative Analysis of Coregistered Ultrasound and Magnetic Resonance Imaging in Neurosurgery: Alex Hartov

Entire journal content can be viewed online at www.neurosurgery-online.com ARTICLES IN BLUE indicate free access for non-subscribers.

TABLE 201 209 245

OF

CONTENTS

Fiducial versus Nonfiducial Neuronavigation Registration Assessment and Considerations of Accuracy: Wolfgang K. Pfisterer Compensation of Geometric Distortion Effects on Intraoperative Magnetic Resonance Imaging for Enhanced Visualization in Image-guided Neurosurgery: Neculai Archip Use of an Integrated Platform System in the Placement of Deep Brain Stimulators: Gregory G. Heuer

NEW TECHNOLOGY 245 266

Use of an Integrated Platform System in the Placement of Deep Brain Stimulators: Gregory G. Heuer DynaCT Soft-tissue Visualization Using an Angiographic C-arm System: Initial Clinical Experience in the Operating Room: Koreaki Irie

BY

110 145 173 180 186

251

TOPIC

(CONTINUED)

Tentorial Dural Arteriovenous Fistulae: Operative Strategies and Microsurgical Results for Six Types: Michael T. Lawton Modifications of the Transoral Approach to the Craniovertebral Junction: Anatomic Study and Clinical Correlations: A. Samy Youssef The Translaminar Approach to Lumbar Disc Herniations Impinging the Exiting Root: Luca Papavero The Infraclavicular Approach to the Brachial Plexus: Gabriel C. Tender Surgical Approach to Ulnar Nerve Compression at the Elbow Caused by Epitrochleoanconeus Muscle and a Prominent Medial Head of the Triceps: Olga Gervasio Quantification of the Frontotemporal Orbitozygomatic Approach Using a Threedimensional Visualization and Modeling Application: Anthony L. D’Ambrosio

PAIN

SURGICAL TECHNIQUE

194

38

226 235

Clinical and Electrophysiological Comparison of Different Methods of Soft Tissue Coverage of the Median Nerve in Recurrent Carpal Tunnel Syndrome: Nicolas M. Stütz Percutaneous Computed Tomography-guided Radiofrequency Ablation of Upper Spinal Cord Pain Pathways for Cancer-related Pain: Ahmed M. Raslan Spinal and Nucleus Caudalis Dorsal Root Entry Zone Operations for Chronic Pain: Yucel Kanpolat

105 134

PERIPHERAL NERVE

162

180 186

179

194

The Infraclavicular Approach to the Brachial Plexus: Gabriel C. Tender Surgical Approach to Ulnar Nerve Compression at the Elbow Caused by Epitrochleoanconeus Muscle and a Prominent Medial Head of the Triceps: Olga Gervasio Clinical and Electrophysiological Comparison of Different Methods of Soft Tissue Coverage of the Median Nerve in Recurrent Carpal Tunnel Syndrome: Nicolas M. Stütz

RADIOSURGERY 201

Fiducial versus Nonfiducial Neuronavigation Registration Assessment and Considerations of Accuracy: Wolfgang K. Pfisterer

SPINE 145 156 162 173 179 226 235

Modifications of the Transoral Approach to the Craniovertebral Junction: Anatomic Study and Clinical Correlations: A. Samy Youssef Cerebrospinal Fluid-related Complications with Autologous Duraplasty and Arachnoid Sparing in Type I Chiari Malformation: Caitlin E. Hoffman Correction of Late Traumatic Thoracic and Thoracolumbar Kyphotic Spinal Deformities Using Posteriorly Placed Intervertebral Distraction Cages: Michael Y. Wang The Translaminar Approach to Lumbar Disc Herniations Impinging the Exiting Root: Luca Papavero An Alternative Source of Autograft Bone for Spinal Fusion: The Femur: Tann A. Nichols Percutaneous Computed Tomography-guided Radiofrequency Ablation of Upper Spinal Cord Pain Pathways for Cancer-related Pain: Ahmed M. Raslan Spinal and Nucleus Caudalis Dorsal Root Entry Zone Operations for Chronic Pain: Yucel Kanpolat

180 186

194

217 235 262 273

TRAUMA 235

217 226 245 249

Fiducial versus Nonfiducial Neuronavigation Registration Assessment and Considerations of Accuracy: Wolfgang K. Pfisterer Location of Active Contacts in Patients with Primary Dystonia Treated with Globus Pallidus Deep Brain Stimulation: Clement Hamani Percutaneous Computed Tomography-guided Radiofrequency Ablation of Upper Spinal Cord Pain Pathways for Cancer-related Pain: Ahmed M. Raslan Use of an Integrated Platform System in the Placement of Deep Brain Stimulators: Gregory G. Heuer Methods of Scalp Revision for Deep Brain Stimulator Hardware: Alejandro M. Spiotta

SURGICAL APPROACH 1 9 18 24 30 51 57 75 108

Microsurgical Anatomy of the Supracerebellar Transtentorial Approach to the Posterior Mediobasal Temporal Region: Technical Considerations with a Case Illustration: Roham Moftakhar Microsurgical Anatomy of the Safe Entry Zones on the Anterolateral Brainstem Related to Surgical Approaches to Cavernous Malformations: Rodolfo Recalde Image-guided Lateral Suboccipital Approach: Part 1— Individualized Landmarks for Surgical Planning: Alireza Gharabaghi Image-guided Lateral Suboccipital Approach: Part 2— Impact on Complication Rates and Operation Times: Alireza Gharabaghi Endoscopic Sublabial Transmaxillary Approach to the Rostral Middle Fossa: Bonnie C. Ong Purely Endoscopic Resection of Colloid Cysts: Jeremy D.W. Greenlee Endoscopic Endonasal Pituitary Transposition for a Transdorsum Sellae Approach to the Interpeduncular Cistern: Amin B. Kassam The Juxtacondylar Approach to the Jugular Foramen: Michaël Bruneau Endoscopic Supracerebellar Infratentorial Approach for Pineal Cyst Resection: Pankaj A. Gore

Spinal and Nucleus Caudalis Dorsal Root Entry Zone Operations for Chronic Pain: Yucel Kanpolat

TUMOR 1

9

STEREOTAXY 201

Medial Sphenoid Ridge Meningiomas: Classification, Microsurgical Anatomy, Operative Nuances, and Long-term Surgical Outcome in 35 Consecutive Patients: Stephen M. Russell Endoscope-controlled Microneurosurgery to Treat Middle Fossa Epidermoid Cysts: Felipe Trivelato External Carotid Artery to Middle Cerebral Artery Bypass with the Saphenous Vein Graft: Erica F. Bisson Correction of Late Traumatic Thoracic and Thoracolumbar Kyphotic Spinal Deformities Using Posteriorly Placed Intervertebral Distraction Cages: Michael Y. Wang An Alternative Source of Autograft Bone for Spinal Fusion: The Femur: Tann A. Nichols The Infraclavicular Approach to the Brachial Plexus: Gabriel C. Tender Surgical Approach to Ulnar Nerve Compression at the Elbow Caused by Epitrochleoanconeus Muscle and a Prominent Medial Head of the Triceps: Olga Gervasio Clinical and Electrophysiological Comparison of Different Methods of Soft Tissue Coverage of the Median Nerve in Recurrent Carpal Tunnel Syndrome: Nicolas M. Stütz Location of Active Contacts in Patients with Primary Dystonia Treated with Globus Pallidus Deep Brain Stimulation: Clement Hamani Spinal and Nucleus Caudalis Dorsal Root Entry Zone Operations for Chronic Pain: Yucel Kanpolat Frontozygomatic Titanium Cranioplasty in Frontosphenotemporal (“Pterional”) Craniotomy: Shaan M. Raza Novel Equine Collagen-only Dural Substitute: Francesco Biroli

38

51 57 75 82 91 102 105 108 209

226

Microsurgical Anatomy of the Supracerebellar Transtentorial Approach to the Posterior Mediobasal Temporal Region: Technical Considerations with a Case Illustration: Roham Moftakhar Microsurgical Anatomy of the Safe Entry Zones on the Anterolateral Brainstem Related to Surgical Approaches to Cavernous Malformations: Rodolfo Recalde Medial Sphenoid Ridge Meningiomas: Classification, Microsurgical Anatomy, Operative Nuances, and Long-term Surgical Outcome in 35 Consecutive Patients: Stephen M. Russell Purely Endoscopic Resection of Colloid Cysts: Jeremy D.W. Greenlee Endoscopic Endonasal Pituitary Transposition for a Transdorsum Sellae Approach to the Interpeduncular Cistern: Amin B. Kassam The Juxtacondylar Approach to the Jugular Foramen: Michaël Bruneau The New Generation PoleStar N20 for Conventional Neurosurgical Operating Rooms: A Preliminary Report: Vasileios Ntoukas A Comparative Analysis of Coregistered Ultrasound and Magnetic Resonance Imaging in Neurosurgery: Alex Hartov Usefulness of Intraoperative Photodynamic Diagnosis Using 5-Aminolevulinic Acid for Meningiomas with Cranial Invasion: Yoichi Morofuji Endoscope-controlled Microneurosurgery to Treat Middle Fossa Epidermoid Cysts: Felipe Trivelato Endoscopic Supracerebellar Infratentorial Approach for Pineal Cyst Resection: Pankaj A. Gore Compensation of Geometric Distortion Effects on Intraoperative Magnetic Resonance Imaging for Enhanced Visualization in Image-guided Neurosurgery: Neculai Archip Percutaneous Computed Tomography-guided Radiofrequency Ablation of Upper Spinal Cord Pain Pathways for Cancer-related Pain: Ahmed M. Raslan

VASCULAR 134

External Carotid Artery to Middle Cerebral Artery Bypass with the Saphenous Vein Graft: Erica F. Bisson

VIRTUAL NEUROSURGERY 251

Quantification of the Frontotemporal Orbitozygomatic Approach Using a Threedimensional Visualization and Modeling Application: Anthony L. D’Ambrosio

N5

Operative neruosurgery

Operative Attraction

Operative Otorhinolaryngology

AWOL with the Operative

Kempes Operative neurosurgical

AWOL with the Operative

Operative Techniques in Laryngology

Operative Hip Arthroscopy

TeLinde's Operative Gynecology

The Operative (Stratton 3)

Kempes operative neurosurgical

Secret Operative K-13

Mammakarzinom: Operative Behandlungskonzepte

Operative Surgery Manual, 1e

Operative Obstetrics, Second Edition

Operative Pediatric Urology

Principles of Operative Dentistry

TeLinde's Operative Gynecology, 10th Edition

Pickard's Manual of Operative Dentistry

Pickard's manual of operative dentistry

Operative Anatomy of the Heart

Operative Anatomy of the Heart

Morbid Obesity: Peri-Operative Management

Operative Treatment of Elbow Injuries

Operative Urology: At the Cleveland Clinic

Operative Techniques in Laparoscopic Colorectal Surgery

Management of Post-Operative Pain with Acupuncture

Management of Post-Operative Pain with Acupuncture

Laparoscopic Hernia Surgery: An Operative Guide

Operative Dermatologie: Lehrbuch und Atlas, 2.Auflage

Operative neruosurgery

Operative Attraction

Operative Otorhinolaryngology

AWOL with the Operative

Kempes Operative neurosurgical

AWOL with the Operative

Operative Techniques in Laryngology

Operative Hip Arthroscopy

TeLinde's Operative Gynecology

The Operative (Stratton 3)

Kempes operative neurosurgical

Secret Operative K-13

Mammakarzinom: Operative Behandlungskonzepte

Operative Surgery Manual, 1e

Operative Obstetrics, Second Edition

Operative Pediatric Urology

Principles of Operative Dentistry

TeLinde's Operative Gynecology, 10th Edition

Pickard's Manual of Operative Dentistry

Pickard's manual of operative dentistry

Operative Anatomy of the Heart

Operative Anatomy of the Heart

Morbid Obesity: Peri-Operative Management

Operative Treatment of Elbow Injuries

Operative Urology: At the Cleveland Clinic

Operative Techniques in Laparoscopic Colorectal Surgery

Management of Post-Operative Pain with Acupuncture

Management of Post-Operative Pain with Acupuncture

Laparoscopic Hernia Surgery: An Operative Guide

Operative Dermatologie: Lehrbuch und Atlas, 2.Auflage

Recommend Documents