Food Chemical Safety, Volume I: Contaminants (Woodhead Publishing in Food Science and Technology)

Food chemical safety Volume 1: Contaminants

Related titles from Woodhead’s food science, technology and nutrition list: Food chemical safety Volume 2: Additives (ISBN: 1 85573 563 6) This volume provides comprehensive information about additives in the food industry. The book opens with an explanation of risk analysis and analytical methods in relation to the use of additives in food products. This is followed by full details of relevant EU and USA regulations. The second part of the book provides information about specific subjects including flavourings, sweeteners and colourings. Making the most of HACCP (ISBN: 1 85573 504 0) Based on the experience of those who have successfully implemented HACCP systems, this book will meet the needs of food processing businesses at all stages of the HACCP system development. The collection is edited by two internationally recognised HACCP experts and includes chapters reflecting the experience of both major companies such as Cargill, Heinz and Sainsbury and the particular challenges facing SMEs. The scope of the book is truly international with experiences of HACCP implementation in countries such as Thailand, India, China and Poland. Foodborne pathogens (ISBN: 1 85573 454 0) A practical guide to identifying, understanding and controlling foodborne pathogens. This book relates current research to practical strategies for risk analysis and control and is designed for both microbiologists and non-specialists, particularly those concerned directly with food processing operations. The first part of the book examines specific microbiological hazards. This is followed by an examination of risk assessment and the concluding section provides a guide to controlling pathogens throughout the supply chain from farmer to consumer. Details of these books and a complete list of Woodhead’s food science, technology and nutrition titles can be obtained by: • visiting our web site at www.woodhead-publishing.com • contacting Customer services (email: [email protected]; fax: +44 (0) 1223 893694; tel.: +44 (0) 1223 891358 ext. 30; address: Woodhead Publishing Limited, Abington Hall, Abington, Cambridge CB1 6AH, England)

If you would like to receive information on forthcoming titles in this area, please send your address details to: Francis Dodds (address, tel. and fax as above; e-mail: [email protected]). Please confirm which subject areas you are interested in.

Food chemical safety Volume 1: Contaminants Edited by David H. Watson

Published by Woodhead Publishing Limited Abington Hall, Abington Cambridge CB1 6AH England www.woodhead-publishing.com Published in North and South America by CRC Press LLC 2000 Corporate Blvd, NW Boca Raton FL 33431 USA First published 2001, Woodhead Publishing Limited and CRC Press LLC ß 2001, Woodhead Publishing Limited The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publishers cannot assume responsibility for the validity of all materials. Neither the authors nor the publishers, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from the publishers. The consent of Woodhead Publishing Limited and CRC Press LLC does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited or CRC Press LLC for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress. Woodhead Publishing Limited ISBN 1 85573 462 1 CRC Press ISBN 0-8493-1210-8 CRC Press order number: WP1210 Cover design by The ColourStudio Project managed by Macfarlane Production Services, Markyate, Hertfordshire ([email protected]) Typeset by MHL Typesetting Limited, Coventry, Warwickshire Printed by TJ International, Padstow, Cornwall, England

Contents

List of contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

xi

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Watson, Food Standards Agency, London 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Veterinary drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Persistent environmental chemicals . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Processing contaminants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Migration from materials and articles in contact with food . . 1.7 Naturally occurring toxicants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Control measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.9 Current and future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.10 Dedication and acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.11 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 2 4 6 7 8 8 9 11 11 11

Part I Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

2

Risk analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. R. Tennant, Consultant, UK 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Hazard identification in the food supply chain . . . . . . . . . . . . . . . 2.3 Dose-response characterisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Exposure analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Risk evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Methods for risk management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

15 15 16 19 21 28 29

vi

Contents 2.7 2.8 2.9

3

4

5

Future trends in risk analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sources of further information and advice . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32 35 35

Analytical methods: quality control and selection . . . . . . . . . . . . . . R. Wood, Food Standards Agency, London 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Legislative requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 FSA surveillance requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Laboratory accreditation and quality control . . . . . . . . . . . . . . . . . 3.5 Proficiency testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Standardised methods of analysis for contaminants . . . . . . . . . . . 3.8 The future direction for methods of analysis . . . . . . . . . . . . . . . . 3.9 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Information for potential contractors on the analytical quality assurance requirements for food chemical surveillance exercises .......................................................

37 37 38 41 41 47 53 57 61 62 64

Molecular imprint-based sensors in contaminant analysis . . . . . P. D. Patel, Leatherhead Food Research Association 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 The principles of molecularly imprinted polymer-based techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 The development and application of MIP-based sensors . . . . . 4.4 Case studies: contaminant analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Sources of further information and advice . . . . . . . . . . . . . . . . . . . 4.7 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

71

Bioassays in contaminant analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . L. A. P. Hoogenboom, State Institute for Quality Control of Agricultural Products (RIKILT), Wageningen 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Dioxins and the DR-CALUX bioassay . . . . . . . . . . . . . . . . . . . . . . 5.3 The use of bioassays for other groups of compounds . . . . . . . . 5.4 Future developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

71 73 76 79 84 86 87 88

91 92 100 102 102 102

Contents Part II Particular contaminants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

7

8

9

vii 107

Veterinary drug residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. N. Dixon, Food Standards Agency, London 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Control of veterinary products in the UK . . . . . . . . . . . . . . . . . . . . 6.3 Chemical substances commonly used in veterinary medicines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Surveillance for veterinary drug residues . . . . . . . . . . . . . . . . . . . . 6.5 Analytical methods employed in drug residues surveillance . 6.6 Results of surveillance for veterinary drug residues in the UK (1998) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.7 Potential effects on human health of veterinary drug residues in food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.8 Current issues relating to residues of veterinary drugs in food in the UK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.10 Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

109

Inorganic contaminants in food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. Harrison, Food Standards Agency, London 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Metals and metalloids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 Nitrate and nitrite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

148

Environmental organic contaminants in food . . . . . . . . . . . . . . . . . . . N. Harrison, Food Standards Agency, London 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Aromatic hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Polycyclic aromatic hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Dioxins and PCBs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Chlorinated hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.6 Phthalic acid esters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.7 Endrocrine disrupters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

169

Chemical migration from food packaging . . . . . . . . . . . . . . . . . . . . . . . L. Castle, Ministry of Agriculture, Fisheries and Food, York 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Chemical migration and the main factors that control it . . . . . 9.3 The range and sources of chemicals in food packaging that pose a potential risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Research on health issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5 Regulatory context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

193

109 112 117 132 134 138 143 144 146 147

148 150 163 165

169 171 172 175 182 184 185 186

193 195 199 205 206

viii

Contents

9.6 9.7 9.8

Migration testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Case studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Suggested further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

209 212 217

10 Pesticides I. Shaw and R. Vannoort, Institute of Environmental Science and Research, Christchurch 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Monitoring pesticides in food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 High risk groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4 The UK’s approach to pesticide surveillance . . . . . . . . . . . . . . . . 10.5 Findings from the UK pesticide monitoring scheme . . . . . . . . . 10.6 Human exposure monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.7 Should we ban pesticides? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

218

11 Mycotoxins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. E. Smith, University of Strathclyde, Glasgow 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Health implications of mycotoxins . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Analytical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Application of HACCP systems to reduce mycotoxin presence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Prevention and control of mycotoxins . . . . . . . . . . . . . . . . . . . . . . . 11.6 Conclusion and future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

238

Part III Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The international regulation of chemical contaminants in food ....................................................... T. Berg, Danish Veterinary and Food Administration, Soborg 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 The nature of international regulation: Codex Alimentarius . 12.3 Decision making and enforcement mechanisms . . . . . . . . . . . . . 12.4 The Codex General Standard on Contaminants and Toxins in Food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.5 Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.6 Sources of further information and advice . . . . . . . . . . . . . . . . . . . 12.7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

218 225 228 229 231 234 236 236

238 241 246 250 253 256 257 261

12

263 263 265 268 271 274 276 277

Contents The regulation of chemical contaminants in foodstuffs in the European Union . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. A. Slorach, National Food Administration, Uppsala 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Scientific advisory committees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3 Pesticide residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4 Veterinary drug residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5 Mercury and histamine in fishery products . . . . . . . . . . . . . . . . . . 13.6 Other chemical contaminants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.7 Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

13

Contaminant regulation and management in the United States: the case of pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. K. Winter, University of California, Davis 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 Pesticide regulation in the US . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3 Regulatory monitoring of pesticides in the US . . . . . . . . . . . . . . . 14.4 Managing pesticides in foods in the US . . . . . . . . . . . . . . . . . . . . . 14.5 Improving the management of pesticides in foods . . . . . . . . . . . 14.6 Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.7 Further information and advice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.8 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

279 279 280 282 284 287 288 289 290

14

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

295 295 297 298 301 305 310 311 311 314

This page intentionally left blank

Contributors

Chapter 1

Chapter 3

Dr David Watson Food Standards Agency Room 212 Ergon House 17 Smith Square London SW1P 3WG

Dr Roger Wood Food Standards Agency c/o Institute of Food Research Norwich Research Park Colney Norwich NR4 7UA

Tel: +44 (0)20 7238 6250 Fax: +44 (0)20 7238 6124 E-mail: david.Watson@food standards.gsi.gov.uk

Tel: +44 (0)1603 255298 Fax: +44 (0)1603 507723 E-mail: roger.wood@foodstandards. gsi.gov.uk

Chapter 2 Dr D. R. Tennant 14 St Mary’s Square Brighton East Sussex BN2 1FZ Fax: +44 (0)1273 276358 E-mail: [email protected]

Chapter 4 Dr Pradip Patel Leatherhead Food RA Randalls Road Leatherhead Surrey KT22 7RY Tel: +44 (0)1372 822200 Fax: +44 (0)1372 386228 E-mail: [email protected]

xii

Contributors

Chapter 5

Chapter 9

Dr Ron Hoogenboom Department of Food Safety and Health State Institute for Quality Control of Agricultural Products (RIKILT) PO Box 230 6708AE Wageningen The Netherlands

Dr Laurence Castle Ministry of Agriculture, Fisheries and Food Central Science Laboratory Sand Hutton York YO41 1LZ

Tel: +31 317 475623 Fax: +31 317 417717 E-mail: L.A.P.Hoogenboom@RIKILT. DLO.NL

Chapter 6 Dr S. N. Dixon Food Standards Agency PO Box 31037 Room 409b Ergon House 17 Smith Square London SW1P 3WG Tel: +44 (0)20 7238 6358 Fax: +44 (0)20 7238 6402 E-mail: Stephen.Dixon@food standards.gsi.gov.uk

Chapters 7 and 8 Dr N. Harrison Food Standards Agency Room 234 Ergon House 17 Smith Square London SW1P 3JR Fax: +44 (0)20 7238 5331 E-mail: Nigel.Harrison@food standards.gsi.gov.uk

Tel: +44 (0)1904 462 540 Fax: +44 (0)1904 462 133 E-mail: [email protected]

Chapter 10 Professor Ian Shaw and Dr Richard Vannort Food Safety Programme Manager Institute of Environmental Science and Research PO Box 29-181 Christchurch New Zealand Tel: +64 3 351 6019 Fax: +64 3 351 0010 E-mail: [email protected]

Chapter 11 Emeritus Professor J. E. Smith Department of Bioscience and Biotechnology University of Strathclyde Glasgow G1 1XW Fax: +44 (0)141 553 4124 (via secretary) E-mail: [email protected]

Contributors xiii

Chapter 12

Chapter 14

Dr Torsten Berg Division of Chemical Contaminants Ministry of Food, Agriculture and Fisheries Danish Veterinary and Food Administration 19 Morkhoj Bygade DK-2860 Soborg Denmark

Professor Carl K. Winter Department of Food Science and Technology University of California One Shields Avenue Davis CA 95616-8598 USA

Tel: +45 33 95 60 00 Fax: +45 33 95 60 01 E-mail: [email protected]

Chapter 13 Dr S. Slorach Swedish National Food Administration Box 622 75126 Uppsala Sweden E-mail: [email protected]

Tel: (1) 530 752 5448 Fax: (1) 530 752 3975 E-mail: [email protected]

This page intentionally left blank

1 Introduction D. Watson, Food Standards Agency, London

1.1

Background

Scientific knowledge about chemical contamination of food has grown considerably in recent years. Only a few years ago I wrote that ‘the study of chemical contaminants in food is still a relatively young science’ (Watson, 1993). Since then this area of science has continued to develop, in particular becoming an established part of regulatory reviews of food safety across the world. Chapters 12 to 14 of this book describe this major new development. Chapters 2 to 5 detail growing development of practical methods of detecting, monitoring and managing chemical contamination of food. Practical methods of working on these substances have developed steadily. The middle chapters (6 to 11) summarise and review information about the different types of chemical contaminants. The main groups of chemical contaminants that can be found in food share the following characteristics: • They are not intentionally added to food. • Contamination can happen at one or more stages in food production. • Illness is likely to result if consumers ingest enough of them.

The first of these points distinguishes chemical contaminants from other chemicals in food, e.g. vitamins and additives. The wide range of possible sources of chemical contamination has major resource implications, particularly in controlling chemicals that find a wide range of uses, for example pesticides as opposed to veterinary drugs. In order to ensure consumer and worker protection very careful attention must be given at all stages in food production, unless it is known that contamination with a particular chemical cannot occur at some stages.

2

Food chemical safety

We know most about residues in food of chemicals such as pesticides and veterinary drugs that are used in food production. Companies that produce these chemicals are generally required to convince the licensing authorities that there will not be unsafe levels of residues in food, if the products containing the chemicals are used as instructed. This requirement has generated a huge amount of information, some of it now in the public domain. Less is known about toxins that occur naturally in food, although there has been much chemical research on toxins produced by fungi (mycotoxins).

1.2

Pesticides

Many pesticides have been tested for in food. Organochlorine chemicals, such as DDT, have been included in surveys for residues of pesticides in food for many years. The detection of DDT and related compounds in the environment in the 1960s led to concern that their environmental persistence might cause widespread and lasting damage to ecosystems. Indeed their presence in the environment was linked with reduction in eggshell thickness and hence reduced breeding success of birds of prey. Such concerns led to surveillance for residues of persistent organochlorine pesticides in environmental and food samples. Group detection of these compounds by a common method of analysis aided this surveillance. It also generated data on chemically related compounds that are not pesticides, notably polychlorinated biphenyls (PCBs; see section 1.4), and demonstrated environmental contamination by phthalate esters (section 1.6). The latter esters probably contaminated food samples in the laboratory. PCBs proved to be very persistent contaminants in the environment. Environmentally persistent organochlorine pesticides such as DDT have largely been replaced as insecticides by organophosphorus compounds. This change has brought its own problems. Concerns about possible effects of organophosphorus pesticides on users is leading to their replacement by yet other pesticides such as pyrethroids that are thought to be less hazardous to man and the environment. In the case of pyrethroids their ‘green’ image derives from their origins in ancient usage of chrysanthemum (pyrethrum) products for a variety of purposes. Natural origin of chemicals in food does not mean that the chemicals are safer than residues of man-made ones in what we eat. The control of pesticide residues in food is done via the law in many countries. There is little variation between national laws in this area, but the way the law is applied differs between countries. As for pesticide residues, so also for other aspects of the chemical safety of food, global harmonisation of food law is a distant prospect. This is despite considerable effort in some groups of countries to achieve consistency, for example in the European Union, the MERCOSUR group (Argentina, Brazil, Paraguay and Uruguay) and the Australia New Zealand Food Authority. The FAO/WHO Codex Committee on Pesticide Residues (CCPR) and the Joint Meeting on Pesticide Residues (JMPR) have worked for many years towards globally accepted standards for pesticide

Introduction

3

residues in food, but only recently have these started to be adopted in countries. For all pesticides in food it is essential to check that any residues in food are within maximum residue limits or other standards provided these are based on extensive toxicological testing of the chemicals involved. Maximum residue limits for pesticide residues in food are set by many internationally recognised authorities, notably the European Union, the US Food and Drugs Agency, and the world bodies CCPR and JMPR. A wide range of practical steps can be taken to control pesticide residues in food, including the following: • Make information available to consumers so that they can make an informed choice. It is essential that consumers are properly informed about the food that they purchase. Apart from the fact that this is one of their basic rights, communication on food safety that is poor or absent can lead to ‘scare stories’ in the media with consequential concern for consumers and money lost by food suppliers. • Control the availability and usage of pesticides. A particularly effective method in theory, since it applies at source. This is not necessarily true in practice. If there is an urgent need to control a pest or even where the need is not pressing but the user is several stages divorced from the consumer, there is always the potential for more pesticide to be used than is actually needed, or for the pesticide to be misused (e.g. applied too close to harvest or slaughter). • Limit food contamination by pesticides present in the environment. It is important to recognise that the use of pesticides near to crops and farm animals, and in factories concerned with food production, can lead to residues in food. Obviously this can be particularly difficult to detect if surveillance for residues looks mainly for those pesticides used directly on crops or farm animals. The remedy is to extend surveillance and to remind users to be very careful to avoid adventitious contamination of food at all stages of production right through to marketing. • Police limits for pesticides in food. Global standard setting may be advancing but it is essential that there is effective surveillance. This must include action as well as monitoring. • Advice of the type noted above on avoiding adventitious contamination of food needs to be made both nationally and by management in companies involved with food. The control of pesticide residues in food could include HACCP (Chapter 11) to provide an important element of prevention. Other preventative methods include testing of incoming supplies of raw and other materials, and direction by retailers to suppliers that they must use only a defined list of pesticides in specified ways. • Halt the supply of contaminated food. This rather draconian measure can and has been used in extreme cases. In the UK there are measures in place to do this in the Food and Environment Protection Act 1985. • Apply an open and objective system of controlling the use, safety and availability of pesticides.

4

Food chemical safety

Only where all of these approaches are used is there likely to be the chance of convincing consumers that they are protected against unsafe levels of pesticides in food. Many consumers have now been influenced by the media to think that all residues of pesticides in food are unsafe. This has contributed to the growth in demand for organic food.

1.3

Veterinary drugs

Like pesticides, veterinary drugs have an important role to play in reducing disease and suffering but their use has been brought into question by concerns about residues of them in food. The fact that their use can lead to residues in the food supply was recognised rather later than for pesticide residues. Nevertheless, methods of analysis and surveillance for veterinary drug residues are now well established in many parts of the world. The main classes of veterinary drugs used in farm animals are: • Ectoparasiticides used to control flies, ticks and other skin parasites. These fall under the general heading of pesticides (section 1.2 above). • Antimicrobial agents which are used to treat and prevent diseases caused by bacteria and fungi. • Anthelmintic agents used against worms and flukes (hence the name – helminths include liver flukes). • Anabolic agents to promote growth. This group includes hormones and some antimicrobial substances. • Tranquillisers and beta-agonists which have been used to reduce the risk of harm to animals being taken to slaughter. • Coccidiostats to treat and prevent coccidial parasites in the GI tract.

Antimicrobial agents are of two general types – antibiotics and chemotherapeutic agents. The first of these inhibits microbial growth, the second kills the micro-organisms. Residues of antibiotics have been screened by testing of milk, meat or kidney samples to see if they inhibit microbial growth. This simple, direct approach has been used widely. The testers need to take account of possible false positive results if the sample contains a natural inhibitor, and the variable sensitivity of bacteria used in the test to different antibiotics. The use of chemical methods to measure residues of antibiotics has not been as popular, mainly because they are more labour intensive and hence more expensive than microbial inhibition tests. However, some antibiotics, such as sulphonamides, are not readily detected by bacterial tests. Chemical methods are essential for chemotherapeutic antimicrobial agents, such as chloramphenicol, since they are not detected effectively by bacterial inhibition. Residues of antimicrobial agents tend to be present at their highest levels in the liver and kidney, as well as the site of injection (if that is the route of application). The liver is the main site of biochemical modification of antimicrobial substances, as the body tries to convert them to less toxic

Introduction

5

compounds. Indeed, the body tries to do this to all substances that are foreign to the body. The kidney, as one of the two main sites of chemical excretion, the other being the GI tract, is therefore a good organ to test at slaughter for the presence of residues of veterinary drugs including antimicrobial agents. Residues of veterinary drugs are usually at lower levels in meat than in kidney or liver. Anthelmintic agents are a smaller, less diverse group than antimicrobial agents. They include levamisole and the benzimidazoles, and the newer avermectins. The benzimidazoles include thiabendazole which has also been used as a pesticide on plants and as a food preservative. Chemical methods have been used widely to quantify the residues of these substances in food, as part of national control programmes. They are an important sub-group of the antiparasitic agents that many consider to be essential for good animal husbandry. Anabolic agents are used to promote growth. They include hormones and some antimicrobial substances. The latter are used to inhibit the growth of harmful organisms in the GI tract. They are used at low levels in feeding stuffs and are thought unlikely to lead to residues. Hormonal anabolic agents are banned from use in the European Union. They are either synthetic and therefore clearly detectable, or natural and therefore more difficult to distinguish from hormones that are endogenously found in animal products. The use of immunoassay revolutionised the detection of synthetic hormone residues, much as it did in dope testing of athletes. The potential for consumer exposure to residues of tranquillisers and betaagonists has been tested intensively over the past twenty years. This followed evidence of illegal drug availability and usage. These substances could have immediate effect on health if consumed in large amounts. Action was taken to protect consumers. Surveillance continues to ensure that no detectable contamination of the food supply occurs with these substances (see, for example, VMD, 1995). Coccidiostats are used in both poultry and large farm animals to treat and prevent coccidial parasites in the GI tract. Residue testing for coccidiostats has required a lot of work to develop chemical methods of testing. Veterinary drug residues are controlled in similar ways to pesticide residues at national and international levels. The FAO/WHO Codex Committee on Residues of Veterinary Drugs in Food first met in 1986. It elaborates maximum limits, drawing on the toxicological advice of the Joint Expert Committee on Food Additives and Contaminants. National governments and the same international groupings noted above (section 1.2) have developed similar approaches to setting standards for veterinary drug residues in food, notably assessing residues data for tissues from animals treated under controlled conditions and applying surveillance to assess consumer exposure to residues. Where there are differences in approach from country to country, as for pesticide residues, they tend to be in how the law is applied. This makes it particularly important to monitor imported as well as homegrown animal products, although within supra-national alliances such as the

6

Food chemical safety

European Union the emphasis is on surveillance of home-grown products by each member state rather than checking each others’ exports. The sources of veterinary drug residues in food are clearly more limited in variety than those of pesticides. However, little attention seems to have been given to possible contamination of the environment and hence food by animal waste products containing residues of veterinary drugs. Residues of drugs and their metabolites in faeces could, in theory at least, lead to contamination of soil and hence crops following muck spreading.

1.4

Persistent environmental chemicals

Polychlorinated biphenyls (PCBs; section 1.2) and dioxins have been most widely studied in this category. PCBs were used in a wide variety of industrial applications, for example as dielectrics in transformers. But they are very persistent contaminants, both in the general environment and in human fat. In theory the routes of entry into food are: • uptake from the environment by food producing animals, particularly those with high fat content (as PCBs are lipophilic) • direct contamination of food or animal feed following an industrial accident • migration from packaging into food, just as other chemicals in packaging can migrate into food (section 1.6).

In practice there is little evidence for the migration of PCBs from packaging (JFSSG, 1999a). There is considerably more evidence for the other two routes leading to PCBs in food. Indeed, historical trends can be drawn up for residues of PCBs in human fat, breast milk and fish. There has been a very gradual decline in levels of these organochlorine compounds in the environment, food and human tissues. This is entirely consistent with the persistence of these compounds. There has been considerable work on the individual congeners of PCBs, so much so that analysis for these is now commonplace. Toxicological review of these substances has also been extensive. This has identified dioxin-like PCBs, probably the first time that such close parallels have been drawn between the toxicological and other properties of two major groups of chemical contaminants in food. Dioxins in food and the environment have been intensively studied over the past twenty years (see for example Steering Group on Chemical Aspects of Food Surveillance, 1992a). The term ‘dioxins’ has come to be used for polychlorinated dibenzo-p-dioxins and in some cases also polychlorinated dibenzofurans. Both of these are ubiquitous environmental contaminants. They are highly resistant to breakdown in the environment. They are particularly difficult to study because of the large number of substances involved and the very low levels of detection needed. Nevertheless, much surveillance work has been done on polychlorinated dibenzo-p-dioxins, particularly 2,3,7,8-tetrachlorodibenzo-p-

Introduction

7

dioxin (TCDD). A system of toxic equivalency has been developed to allow comparison of survey results with a complex and growing body of literature on the toxicology of these compounds. The sources of food contamination with dioxins are now known to be many more than was originally asserted when claims were made that incinerators were the main source of dioxins in the environment. Indeed known sources of dioxins now include vehicle exhausts, domestic coal fires, manufacture and use of organic chemicals, and metallurgical processors. Two main types of contamination of food appear to be involved: atmospheric deposition and spreading of sludge, in both cases on farmland. Other environmental contaminants in food include several metals and a yet to be defined number of the organic chemicals used by industry and by-products of industrial activity. Metal contamination of food can occur in a wide variety of ways, including environmental and other sources such as canning. There has been a huge amount of analytical work on metals such as lead in food. Indeed a large part of the periodic table has been covered. Early work on metals identified that analytical quality assurance is a key tool in the surveillance of food for chemical contaminants. It also led to the development of toxicological standards which can be used to define whether or not surveillance results show there is a hazard to consumer health. Both of these types of approaches are now standard in the best surveillance programmes, whether they are on contaminants or additives in food. Surveillance and other research on residues of industrial organic chemicals in food is at a much earlier stage than work on metals in food. From the list of about 50,000 such chemicals one needs to identify those that are likely to contaminate food and pose a risk to consumers. This is by no means an easy task. The sort of factors that one might consider are as follows (Steering Group on Food Surveillance, 1988): • production volume (e.g. exclude chemicals that are manufactured in very small amounts) • pattern of usage • potential for release into the environment • persistence in the food chain • toxicity following oral ingestion.

1.5

Processing contaminants

It has been very difficult to predict which chemicals might be formed during food processing and might pose a hazard to consumers. It cannot be assumed that this class of substances does not exist. There are already a few established examples: • There is evidence that carcinogenic N-nitrosamines can be formed during the production of alcoholic beverages, fermented foods and cured meats (Steering Group on Chemical Aspects of Food Surveillance, 1992b).

8

Food chemical safety

• Carcinogenic polycyclic aromatic hydrocarbons can contaminate smoked food (Bartle, 1991), although the main dietary sources of these compounds in the UK appear to be early in food production. • 3-Monochloropropane-1,2-diol (3-MCPD) and ethyl carbamate have both also proven to be unwanted contaminants that are formed during food processing (JFSSG, 1999b; Food Standards Agency, 2000).

Each of these process contaminants was discovered by studies on the foods in question, either by chance or after research on chemical contamination of the environment. A more systematic approach is needed to identify processing contaminants. If this can be achieved, and this is no easy task, work on the above processing contaminants has shown how they can be researched so that levels are reduced and hence consumers protected.

1.6

Migration from materials and articles in contact with food

Early work on phthalate esters (section 1.2) and several monomers such as styrene, which are used to make plastics, demonstrated that chemical migration can occur from packaging into food. There has been a huge amount of practical work on this over the last thirty years (Gilbert, 1997), much of it on plastics. Thus there are now in place detailed controls on this aspect of plastics in the European Union (EU) and the USA. The controls in the EU have been fully implemented in Great Britain (FCM Unit, 2000). Less is known about chemical migration from other packaging materials. Paper and board have been subjected to surveillance which so far has shown that some chemicals can migrate from it into food (e.g. diisopropylnaphthalenes; JFSSG, 1999c) whilst others probably do not (e.g. PCBs; JFSSG, 1999a). The reason that some chemicals migrate and others do not is unclear at present. It is not due to a layer of packaging between the food and the contaminated packaging. Nor does it appear to be due to a fundamental difference in volatility or other physical property of the chemical migrants. Research on chemical migration from other types or components of packaging material – e.g. glass, wood, cork, coatings, adhesives – has been carried out sporadically. There is now a concerted UK programme to study chemical migration from these so that problems can be identified and dealt with in a consistent way (Working Party on Chemical Contaminants from Food Contact Materials and Articles, 1999).

1.7

Naturally occurring toxicants

These are of three types:

Introduction

9

• toxins produced by microbial contamination of food and raw materials used in food production • toxins produced by crops (in some cases at least to protect the plants from insects) • toxins ingested by food-producing animals.

The first category includes toxins produced by fungi (mycotoxins) and bacteria. The second group includes a wide range of food-producing plants. The third is a small group of marine toxins, mostly produced by dinoflagellate algae, which find their way up the food chain and hence onto our plates. These algal toxins and those produced by bacteria are unusual chemical contaminants in that they are quick to take effect. Most other chemical contaminants, including mycotoxins and inherent toxins in crops, would take some time to cause illness. It is perhaps not surprising that more effort has been concentrated on these quick-acting toxins, than on many of the slower ones from plants or fungi. Considerable progress can be made in protecting consumers from many natural toxins in food by applying straightforward good agricultural practices and through the careful handling of food. For example, crop rotation can reduce mycotoxin contamination, as can keeping stored grain and seeds dry; and bacterial toxins are much less likely to be found in food if HACCP is applied in food production. More sophisticated measures are needed to protect consumers from inherent toxins in crops. Plant breeding can lead to higher as well as lower levels of toxins. Careful testing of new varieties for toxins associated with the plant family involved is a key measure that should be adopted. Some wild forms of food plant, for example potatoes, contain very high levels of toxins (glycoalkaloids in the case of potatoes) which have been reduced over the centuries by selective plant breeding. It is not inconceivable that consumer demand for more ‘natural’ foods might lead to reintroduction of high toxin levels by crossing existing cultivars with wild varieties. Similarly the main line of defence against the ‘red tides’ of algae that cause toxin contamination of bivalve molluscs is to assay the amount of toxin in recently gathered molluscs. If the level is unsafe consumers must be warned not to gather the respective species of shellfish. Much research has been done on the incidence of marine toxins around the world (Leftley and Hannah, 1998). It is now essential to use this and other information on natural toxins in food to protect consumers.

1.8

Control measures

As noted above, general good hygienic practice can go a long way in many cases towards minimising chemical contamination of food. It is not always necessary to resort to sophisticated, expensive methods of analysis to reduce consumer exposure to toxic chemicals in food, although knowing what is there does help considerably, and in most cases provides one with an edge over what can be a series of complex problems.

10

Food chemical safety

Much emphasis has been placed in this chapter on the use of survey information as a source of information in controlling chemical contamination of food. This has helped governments around the world to assess problems in this area. They have found that surveillance can stimulate action as well as press coverage. The key is to ensure that action is taken when problems are found. In no particular order the main options are: • • • • •

control the availability and usage of man-made contaminants limit or eliminate the source of contamination police limits advise halt the supply of contaminated food.

As noted in section 1.2 all of these can be applied to pesticides. There is less room for manoeuvre in the case of natural contaminants, where few limits have been set and advice can become very complicated. In between the extremes of advice and further legislative burden, most chemical contaminants can be controlled by using at least two or three of the above options. The key is defining as exactly as possible what the problem is and then taking the most appropriate action. In many cases a balance must be drawn between competing factors. For example, reduction in nitrosamine levels in cured meats must be done without prejudicing protection provided by nitrite against Clostridial contamination, although nitrite is the precursor of nitrosamines in cured meat. The slowly lengthening list of international limits for chemical contaminants in food promises to provide some more flexibility for government control, if not for food producers. International procedures have developed considerably over the past ten years at committees reporting to the Codex Alimentarius Commission and in other multinational trade arrangements, notably in MERCOSUR (the Southern Common Market; section 1.2), the European Union (EU) and the Australia–New Zealand Food Authority. There has been significant progress in agreeing international standards for residues of pesticides and veterinary drugs. Progress has been much slower worldwide on setting limits for mycotoxins and metals in food. Work has yet to begin in earnest on defining controls across the world on chemical migration from packaging, although this is well advanced for plastics in the EU and the USA. The respective committees in the Codex Alimentarius for chemical contaminants are as follows: • Pesticides: Codex Committee on Pesticide Residues (CCPR) and Joint Meeting on Pesticide Residues (JMPR). • Veterinary drug residues: Codex Committee on Residues of Veterinary Drugs in Food (CCRVDF) and the Joint Expert Committee on Food Additives and Contaminants (JECFA). • Other chemical contaminants including natural toxins: Codex Committee on Food Additives and Contaminants (CCFAC) and JECFA.

Introduction

11

The secretariats of these Codex Committees are based at the Food and Agriculture Organisation in Rome. The secretariat of JECFA is based in Geneva at WHO headquarters.

1.9

Current and future trends

The control of chemical contamination of food is clearly developing. If there is a key part of this process at present it is the international harmonisation of controls. The Codex Committee on Food Additives and Contaminants (CCFAC) is actively developing a Codex General Standard for Contaminants and Toxins in Food (Chapter 12). This followed a suggestion from the UK delegation, in 1991, that there was a need to develop a Codex philosophy on contaminants. Work on this Standard has accelerated the use of maximum limits for contaminants in food. It has also stimulated work on a general code of practice for source-directed measures to reduce contamination. Effort now needs to be applied to completing position papers and standards for specific contaminants. At the moment attention at the CCFAC is focused on a relatively small number of contaminants, notably a few of the better-studied mycotoxins (aflatoxins, ochratoxins, fumonisin, zearalenone and patulin), PCBs, dioxins and lead. Indeed this short list may prove rather too long if the respective reviews at CCFAC do not make progress. Delay in agreeing standards would send out the wrong signals about the utility of what is a potentially very valuable Codex Standard in a growing international debate on controlling chemical contaminants in food.

1.10

Dedication and acknowledgement

To Linda. I acknowledge colleagues and friends in the Food Standards Agency, and elsewhere. The views expressed in this chapter are those of the author and not of his employer.

1.11

References

(1991) Analysis and occurrence of polcyclic aromatic hydrocarbons in food, in Food contaminants: sources and surveillance, publ. RSC, Cambridge, UK. FCM UNIT (2000) Explanatory note on the legislation in Great Britain, available from the Food Standards Agency, Room 216, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6528, fax +44 (0)20 7238 6124. FOOD STANDARDS AGENCY (2000) Survey for ethyl carbamate in whisky, Food Surveillance Information Sheet 2/00, available from the Food Standards BARTLE, K.D.

12

Food chemical safety

Agency, Room 303b, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6245/6150, fax +44 (0)20 7238 6330 or email: information [email protected]. GILBERT J. (ed.) (1997) Food packaging: ensuring the safety and quality of foods, ed. J Gilbert, Food Additives and Contaminants 14, publ. Taylor and Francis, Basingstoke, Hants, UK. JFSSG (1999a) Survey of retail paper and board food packaging materials for polychlorinated biphenyls (PCBs), Food Surveillance Information Sheet 174, available from the Food Standards Agency, Room 303b, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6245/6150, fax +44 (0)20 7238 6330 or email: [email protected]. JFSSG (1999b) Survey of 3-monochloropropane-1,2-diol in soya sauce and similar products, Food Surveillance Information Sheet 187, available from the Food Standards Agency, Room 303b, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6245/6150, fax +44 (0)20 7238 6330 or email: [email protected]. JFSSG (1999c) Diisopropylnaphthalenes in food packaging made from recycled paper and board, Food Surveillance Information Sheet 169, available from the Food Standards Agency, Room 303b, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6245/6150, fax +44 (0)20 7238 6330 or email: [email protected] LEFTLEY, J.W. and HANNAH, F. (1998) Phycotoxins in seafood, pp. 182–224 in Natural toxicants in food, CRC/Sheffield Academic Press, ed. D. H. Watson, ISBN 1-85075-862-X. STEERING GROUP ON CHEMICAL ASPECTS OF FOOD SURVEILLANCE (1992a) Dioxins in food, Food Surveillance Paper No. 31, publ. HMSO, London. STEERING GROUP ON CHEMICAL ASPECTS OF FOOD SURVEILLANCE (1992b) Nitrate, nitrite and N-nitroso compounds in food: second report, Food Surveillance Paper No. 32, publ. HMSO, London. STEERING GROUP ON FOOD SURVEILLANCE (1988) Food surveillance 1985 to 1988, Food Surveillance Paper No. 24, publ. HMSO, London. VMD (THE VETERINARY MEDICINES DIRECTORATE) (1995) Annual report on surveillance for veterinary residues in 1995, publ. VMD, Weybridge, UK. WATSON, D.H. (1993) Preface to Safety of chemicals in food: chemical contaminants, Woodhead Publ., Cambridge, UK, ISBN 0-13-787862-1. WORKING PARTY ON CHEMICAL CONTAMINANTS FROM FOOD CONTACT MATERIALS

(1999) Review of current research projects, available from the Food Standards Agency, Room 216, P.O. Box 31037, London SW1P 3WG, Tel. No. +44 (0)20 7238 6528, fax +44 (0)20 7238 6124.

AND ARTICLES

Part I Analytical methods

This page intentionally left blank

2 Risk analysis D. R. Tennant, Consultant, UK

2.1

Introduction

We all share the expectation that food will be safe to eat. However, the opportunities for food to become contaminated by chemicals at some stage in its production are legion. Nevertheless, incidents of chemical contamination are very rare and this is testimony to systems for risk assessment and risk management that are applied by food producers, processors and retailers. Any reliable system for assessing and controlling chemical risks must contain six key elements whose relationships are described in Fig. 2.1: 1. 2.

Hazard identification. It is necessary to be aware of what chemical contaminants might occur in a particular foodstuff and the nature of the harmful consequences to human health that might be associated with them. Dose-response characterisation. Different chemicals will be associated with different toxicological end-points and the risk of any individual experiencing toxicity is related to the dose that they receive. Very often it is

Fig. 2.1

Food chemical risk assessment and risk management.

16

3.

4.

5.

6.

Food chemical safety possible to identify a dose level below which the probability of anyone experiencing an adverse effect is very low or zero. Exposure analysis. The amount of any chemical that an individual is exposed to will depend upon the levels that occur in food and the amounts of those foods that are consumed. Different population groups will often have different levels of exposure and it is therefore necessary to identify such sub-groups. Risk evaluation. If a dose level at which no adverse effects are experienced has been identified then it is necessary to identify any population subgroups whose exposure might exceed that level. Risk evaluation should aim to quantify the level of risk that any such populations are exposed to. Risk management. If any population sub-group has been identified as being potentially at risk then measures to control the risk must be assessed and introduced. Any benefits associated with the foods affected must be taken into account and the costs associated with alternative methods of control evaluated. Risk communication. Where there are potential risks associated with chemical contaminants in food other interested parties, in particular any sub-groups particularly affected, must be informed.

The remaining sections of this chapter will provide detailed information about each of these six steps. However, specific strategies for risk assessment and management may need additional elements, depending on the nature of the hazard, the foods in which it may occur and other specific conditions.

2.2

Hazard identification in the food supply chain

Chemical contamination can occur at any point in the food chain from the field through to the point of consumption (Fig. 2.2). Once contamination has occurred it is usually expensive and technically difficult to remove it. Processing may further complicate the situation, by concentrating the contaminant in a particular fraction such as vegetable oil. It is therefore vital to identify all potential sources of contaminants in order to prevent contamination from occurring. 2.2.1 Primary food production In the field, soil can sometimes be a significant source of contamination. For example, in areas of mineralisation, heavy metals may occur naturally at elevated levels in soils. If there has been mining activity in the past this can exacerbate the problem by the spreading of mine spoil at the surface. Animals, particularly cattle, which tend to tug at forage rather than biting it, can ingest significant amounts of soil and metals can accumulate in tissues such as the liver. Plants may also take up elements from the soil that can accumulate in

Risk analysis 17

Fig. 2.2

Opportunities for chemical contamination in the food chain.

edible parts. Where mineral workings drain to the sea, metals can precipitate where chemical conditions change in estuarine environments. Shellfish caught from such estuaries may also have elevated levels of heavy metals. Soils may also become contaminated with industrial pollutants or with agricultural chemicals. For example, fields located close to industrial plants such as incinerators or metal smelters can gradually accumulate residues of combustion products and other chemicals from the fall-out from smoke plumes. Organo-chlorine pesticides, which are now largely banned, can persist in soils for many years and nitrates used in fertilisers can accumulate in soils which, under certain growth conditions, can result in high levels in certain crops. Animal feeds appear to be particularly vulnerable to chemical contamination. Chemical hazards associated with fungal toxins (mycotoxins) were first identified when poultry were adversely affected. Mycotoxins such as aflatoxins

18

Food chemical safety

and ochratoxins were found to be carcinogenic and to occur at low levels in a variety of plant and animal-derived foods. Many national authorities have taken steps to control the levels of mycotoxins in food. On occasions, animal feed has been suspected of deliberate contamination. Incidents involving contamination of animal feed by industrial by-products such as polycylic aromatic hydrocarbons (PAHs) and combustion products such as dioxins are not uncommon. A problem with animal feed is that there is sometimes inadequate control over the provenance of feed constituents. For example, spent cooking oil from food-processing plants is a legitimate feed component. Unfortunately, the temptation for the unscrupulous to dispose of other unwanted oils in this way is too great for some. In many cases such adulterants are probably diluted to such an extent that they are undetectable by conventional chemical analyses. Nevertheless, they may still represent a longterm cumulative hazard to consumers of products from animals fed on such material. 2.2.2 Food processing It is often necessary to process food before it is suitable for human consumption. Grain must be ground into flour, milk churned into butter, barley and hops brewed into beer, for example. Simple contamination might arise from direct contact with containers and tools used in food processing if they are not made from suitable materials. Machinery lubricants and coolants sometime leak and they can find their way into food. 2.2.3 Retailing and consumption Foods are frequently packaged before being put on sale and constituents of the packaging or inks, dyes and glues used in packaging present the potential for migration into food. During transport, food and other materials can become inadvertently intermingled thus presenting a further hazard. In the home there are further potential hazards such as contamination from storage vessels and cooking containers and utensils. The process of cooking can alter food so that new chemical substances are formed. Cooking meat so that it is well browned (for example by roasting, grilling or frying) can produce heterocyclic amines, which are potentially carcinogenic. Hazards can occur right up to the point of consumption. For example, ceramic glazes with a high lead content can be leached out by acidic foods such as wine. 2.2.4 Hazard characterisation Once potential hazards have been identified the nature of the hazard must be characterised. Initially the nature of any toxicological damage should be identified. For example, cadmium, which can be present at high levels in certain soils and sediments, can cause damage to the kidneys, whereas polycyclic

Risk analysis 19 aromatic hydrocarbons, which are pollutants produced by high-temperature combustion, are carcinogenic. The time-scale over which contaminants can exert an adverse effect is also of importance. For carcinogens such as PAHs and mycotoxins, it is the long-term cumulative dose that is most important. Conversely for organophosphorus pesticide residues, one meal might be sufficient to cause some inhibition of the cholinesterase enzyme.

2.3

Dose-response characterisation

The severity of any adverse effect associated with a chemical contaminant is usually directly related to the dose. Severity can be measured as the degree of damage to an individual or the probability of being affected, or a combination of these effects. For a substance that causes tissue damage such as cadmium, increasing the total dose will tend to increase the degree of damage to the kidneys as measured by the loss of proteins in the urine. For carcinogens, where just one molecule has the theoretical ability to induce a tumour, increasing the dose increases the probability that an individual will contract a tumour. In either case there may be some threshold level below which no effects are observed. 2.3.1 Thresholded end-points For many substances the body’s own mechanisms for de-toxification and repair mean that low doses of some chemicals can be tolerated without experiencing any adverse effects. However, once a certain threshold has been exceeded then the degree of adverse effect is related to the dose. The highest dose at which no adverse effects are observed in the most susceptible animal species is identified at the No Observed Adverse Effect Level (NOAEL). The NOAEL is used as the basis for setting human safety standards for contaminants, Provisional Tolerable Weekly Intakes (PTWIs) or Tolerable Daily Intakes (TDIs).1 The PTWI is defined as the amount that an individual can ingest weekly over a lifetime without appreciable health risk. It is related to the NOEAL so that: PTWI (mg/kg bw/week) = NOEAL (mg/kg bw/day)  UF1  UF2  7 Where: UF1 = Uncertainty factor to allow for extrapolation from animal species to humans. UF2 = Uncertainty factor to allow for inter-individual variability in humans. Uncertainty factors usually have a default value of 100 so that the PTWI is usually equal to the NOEAL  700. If human data are available then UF1 is usually taken to be one. Intakes that exceed the PTWI will not necessarily result in any adverse effect because the uncertainty factors are designed to be conservative. In practice it is

20

Food chemical safety

probable that most people could exceed the PTWI by a considerable margin before suffering any harm. Nevertheless, the probability that an individual will suffer harm (risk) increases once the PTWI is exceeded and so this must be balanced against the costs of control. The level of risk below the PTWI is never quite zero because there is always a residual risk that relates to the lack of absolute certainty in the methods used for toxicological testing. 2.3.2 Non-thresholded end-points Some chemical contaminants are believed to have no threshold below which toxic effects are observed. The most common group of hazards in this respect are genotoxic carcinogens. It is generally understood that the risk associated with a very low dose of a carcinogen is proportionate to the risk associated with a higher dose.2 However, we are all subject to continual low doses of carcinogens in our diet that are derived from natural constituents of plants and man-made chemical contaminants probably make up a very small fraction of our total carcinogen load. The indication is that some protective mechanism exists to neutralise small doses of genotoxic carcinogens. This may be a metabolic process or related to the consumption of natural protective chemicals in the diet. The mechanisms for this effect are still poorly understood but it is clear that if they did not exist then the incidence of diet-related cancers would be much higher than it is. Non-thresholded chemicals that are not carcinogens are less frequently identified. For many years lead was considered to be thresholded because its effects on haemoglobin synthesis were not seen at low doses. However, recent work into the effects of lead on mental development suggest that there may be no threshold for this end-point. Food is a relatively minor source of lead exposure compared with air and dust in urban environments. For chemicals that relate to toxicological end-points that do not show thresholds it is not possible to identify a NOAEL or PTWI. In such cases it is desirable to estimate the level of risk associated with a given level of exposure. 2.3.3 Quantitative Risk Assessment In recent years it has become increasingly apparent that for chemical contaminants that are abundant in the environment a more sophisticated approach to dose-response characterisation is required. There is increasing evidence that small but significant sub-populations are exposed to intakes that exceed PTWIs and most people are exposed to potential carcinogens through their diet. In such cases the PTWI concept is redundant because it is necessary to assess the actual levels of risk to which individuals are exposed in order to introduce proportionate control measures. Simply knowing that the hazard exists is not sufficient. Quantitative Risk Assessment (QRA) techniques were pioneered in the USA from the 1960s after introduction of the now repealed Delaney

Risk analysis 21 amendment of 1954 which effectively prohibited the presence of any carcinogen in food.3 Many carcinogenic chemicals are not genotoxic and can exhibit threshold effects. Such substances are demonstrably carcinogenic if sufficiently large doses are given to animals to overcome any defence mechanisms. However, under Delaney these mechanistic arguments could not be taken into account. US risk analysts responded by creating a QRA approach that could be extrapolated to estimate hypothetical risks at the very low doses likely to be encountered by the public. Early QRA methods aimed to define a mathematical relationship between dose and risk of carcinogenesis. They were criticised because they seemed to be forcing biological data into mathematical models rather than developing mathematical models that fitted the data. More recent developments such as ‘physiologically-based pharmacokinetic’ (PB-PK) models aim to take physiological and biological mechanisms better into account (for further information about these developments see section 2.6.1 of this chapter). QRA modelling is still under development and is not widely used outside the USA.

2.4

Exposure analysis

Having established some information about the relationship between dose and response it is necessary to determine the levels of actual doses to the human population. Two pieces of information are vital for this: • Occurrence data: the concentrations of the chemical in the foods of concern including, if relevant, the frequency of occurrence. • Food consumption data: the amounts of the affected foods eaten including, if necessary, consumption by sub-groups.

Intake can then be calculated using a relatively simple equation: Occurrence (mg/kg)  Consumption (kg/week) Intake (mg/kg bw/week) ˆ Bodyweight (kg) Many factors can influence the accuracy of intake estimates and it is of primary importance to ensure that the assumptions made and data used are relevant to the specific risk analysis.4 The selection of inappropriate data and methods can easily lead to estimates of intake that are orders of magnitude greater or less than real levels. A particular question is the selection of statistics to represent a particular population. In the past a population average figure would have been typically used but this approach could very easily underestimate intakes of consumers at the upper end of the distribution of intake levels. Modern guidelines demand that particular sub-groups (such as children) are considered and the intakes at the upper end of the intake range (commonly 90th or 97.5th percentiles). This means that the distributions of contaminant concentrations in food and food consumption patterns must be taken into account.

22

Food chemical safety

2.4.1 Occurrence data Comprehensive data on the levels and frequency of occurrence of chemical contaminants in food are extremely expensive to obtain and are thus relatively rare. Nevertheless there are some reliable data sources and it is usually possible to acquire some data. Unfortunately, it is usually only possible to obtain data on average levels from many published sources. Even if ranges are published with the mean, interpretation is extremely difficult. In the UK the Joint Food Science and Safety Group of the Department of Health and the Ministry of Agriculture, Fisheries and Food have published the results of many analyses for chemical contaminants in food carried out under their Food Surveillance Programme. In many cases the raw data from these surveys are available for analysis. Table 2.1 lists the results of analyses for lead in some samples of cow, sheep and pig kidney obtained in Scotland and England.5 There are clear differences between species and some evidence of differences between sampling locations. What is not clear is the extent to which the variability observed is due to real and consistent differences between species and location or to normal biological variation. For the purpose of risk analysis it is necessary to determine lead concentrations that can be used to represent the distribution of observed values in the above intake calculation. If the average of all the data is used it will overlook differences between species. If the average by species is used it may overlook inter-regional differences. For example, it is possible that the high concentrations in the data set represent cattle in geographic locations in Scotland and England where consumers are consistently exposed to such levels. However, if such high values were used in the risk assessment they would not represent UK consumers as a whole. This example is used to illustrate the difficulty of completing an exposure analysis unless the context of the analysis is fully understood. In this case the differences are relatively Table 2.1

Lead in animal kidneys in England and Scotland (mg/kg)

Location

Cow

Sheep

Pig

North England North England North England North England Scotland Scotland Scotland Scotland S/SE England S/SE England S/SE England S/SE England S/SE England Mean

0.14 0.16 0.08 0.1 0.09 0.35 0.12 0.22 0.07 0.19 0.11 0.35 0.11 0.16

0.08 0.04 ND 0.04 ND 0.21 0.09 ND 0.03 0.12 0.16 0.14 0.23 0.09

ND ND ND ND ND ND ND ND ND ND ND ND ND ND

Risk analysis 23 minor. In other circumstances an inappropriate choice of data could lead to considerable errors. 2.4.2 Food consumption data There are many sources of data on food consumption although not all are necessarily appropriate for risk assessment. The most readily available data are Food Balance Sheets (FBSs) which are prepared globally every year by the UN Food and Agriculture Organisation (FAO).6 These list the domestic production, imports, exports and non-food uses for major raw food commodities for each country together with the calculated per capita annual consumption. Such data are invaluable for making comparisons between national diets since they provide a good indication of the types of food being consumed in each country. They are of limited value for risk assessment since they give no indication of the range of food consumption patterns within the country. Food consumption surveys conducted at the household level provide more information about the distribution of consumption levels. If data on the composition of the household by age and sex are available, modelling can provide some basic information about consumption of individuals. However, for reliable estimates of food consumption by individuals the weighed diary method is probably the best. In this type of survey respondents are asked to weigh and record everything that they eat for the period of the survey. Subjects are usually selected from geographical regions and at different times of year so that the survey is as representative as possible. The principal disadvantage of this type of survey is that it can only cover a few consecutive days and so food consumption over longer time-scales cannot be determined without a supplementary questionnaire about frequency of consumption. The use of weighed dietary survey data can be illustrated by extending the example of lead in kidney used above. Tables 2.2 and 2.3 give the average weekly consumption of cows’, sheep and pigs’ kidney by adults7 and preschool children8 in the UK. These figures represent those individuals who reported consumption of one of these foods during the survey and the per capita average for all individuals in the survey, whether they consumed kidney or not. Since the proportion who reported consumption is less than Table 2.2

Consumption of kidney by UK adults Food consumption, g/day Consumers only Per capita % Mean 90th Mean 90th Consuming percentile percentile

Bovine kidney 14.5 Ovine kindey 1.4 Porcine kidney 1.9 All kidney 17.4

6.05 16.14 6.15 7.01

10.37 40.45 14.17 12.61

0.88 0.22 0.12 1.22

IC IC IC IC

24

Food chemical safety

Table 2.3

Consumption of kidney by UK pre-school children Food consumption, g/day Consumers only Per capita % Mean 90th Mean 90th Consuming percentile percentile

Bovine kidney Ovine kindey Porcine kidney All kidney

3.6 0.2 0.1 4.0

3.67 7.56 0.46 3.81

6.71 15.78 0.52 8.02

0.13 0.02 0.00 0.15

IC IC IC IC

20%, the per capita consumption figures will always be at least five times smaller. Similarly, children tend to eat less of certain foods, such as kidney, than adults. Because all of the data from the original survey are available, it is possible to determine percentiles of the distribution of possible values as well as averages. The choice of figure that is taken to represent kidney consumption will depend on the question being asked. The difference between the per capita average kidney consumption for pre-school children (0.15 g/person/week) and the 90th percentile for adults who are lamb’s kidney consumers (40.5 g/person/week) could have a significant effect on the result of any intake calculation. 2.4.3 Estimating intakes To calculate estimates of intake it is necessary to multiply levels in food with food consumption. As the foregoing sections indicate, this is not a simple matter since great care must be exercised in deciding which level in food to use and which level of consumption. Combining averages provides a simple solution but gives no information about the higher levels of intake to which individuals could be exposed. Combining average contaminant concentrations with high-level consumption might be satisfactory if distribution of the foods of concern is such that over a long time period any individual’s exposure will tend to average out. However, if contamination is localised then it may be necessary to consider using high-level contaminant data with high-level consumption. This worst-case approach would overestimate intakes for the vast majority of consumers. Tables 2.4 and 2.5 provide estimates of intake based on the average of the lead concentrations in Table 2.1 and kidney consumption distributions summarised in Tables 2.2 and 2.3. Intake figures tend to follow food consumption patterns although for pork kidney the intake is zero because no lead was detected in it. As expected, when calculated on a per person per week basis children have lower intakes of lead than do adults. However, when consumption is corrected for individual bodyweight, as is necessary for comparison with a PTWI, children’s intake of all lead is about three times that of

Risk analysis 25 Table 2.4

Potential intakes of lead from kidney by UK adults

Pb, mg/kg Bovine kidney 0.16 Ovine kindey 0.09 Porcine kidney 0.00 All kidney

Pb, mg/kg Bovine kidney Ovine kindey Porcine kidney All kidney

0.16 0.09 0.00 17.4%

Lead intake, ng/person/day Consumers only Per capita Mean 90th Mean 90th percentile percentile 968 1453 0 925

1659 3640 0 1747

141 20 0 161

IC IC IC IC

Lead intake, ng/kg bodyweight/day Consumers only Per capita Mean 90th Mean 90th percentile percentile 14 21 0 13

23 51 0 23

2 0 0 2

IC IC IC IC

adults. This could be significant if children’s intakes were a cause of particular concern. However, if cumulative intake over the longer term was a concern then elevated intake during childhood might be of lesser importance. There is clearly considerable scope for producing a wide variety of intake estimates for any given scenario. It is therefore vital that the underlying Table 2.5

Potential intakes of lead from kidney by UK pre-school children

Pb, mg/kg Bovine kidney Ovine kindey Porcine kidney All kidney

0.16 0.09 0.00 4.0%

Pb, mg/kg Bovine kidney Ovine kindey Porcine kidney All kidney

0.16 0.09 0.00 4.0%

Lead intake, ng/person/day Consumers only Per capita Mean 90th Mean 90th percentile percentile 588 681 0 576

1073 1420 0 1284

21 2 0 23

IC IC IC IC

Lead intake, ng/kg bodyweight/day Consumers only Per capita Mean 90th Mean 90th percentile percentile 42 51 0 41

68 106 0 83

2 0 0 2

IC IC IC IC

26

Food chemical safety

toxicological concerns and the nature of food consumption data are fully understood if an intake estimate that is relevant to the particular situation is to be provided.

2.4.4 Probabilistic intake modelling The approaches for estimating intakes described thus far have relied on being able to extract single figure statistics from data sets to represent those data. In reality all data will show a distribution of values and this distribution tends to be overlooked using conventional methods. This may be unimportant if longterm intakes are relevant to the risk assessment since in the long term data tend to average out. However, when acute intakes are of interest the problem is more complex. It would be possible for a very high level consumer of a particular foodstuff to select, unknowingly, a food item that contains the maximum amount of contaminant. This worst-case scenario may be very unlikely and it might be imprudent to base a risk assessment on it. Yet considering averages alone would underestimate the true situation which lies somewhere between these extremes. Probabilistic (also known as Monte Carlo) modelling provides a means of making use of all of the data from the distribution of contaminant levels and food consumption to predict the probability of a given level of intake occurring. Probabilistic modelling works by taking a random sample from the distribution of contaminant levels and combining it with a random sample from the distribution of food consumption levels. It is well illustrated by the estimation of acute intakes of an organophosphate pesticide from individual apples by children. Figure 2.3 shows the distribution of an organophosphate pesticide in individual apples harvested after treatment with the agent.9 The distribution is skewed so that the mean is 0.69 mg/kg whilst the median is 0.53 mg/kg. The highest value is 3.9 mg/ kg but this represents only one out of 54 samples. It would be possible to make an exposure estimate based on either the mean, median or highest value but this would not represent the real distribution of all possible values. Figure 2.4 is the distribution of apple consumption by occasion by UK preschool children aged 1½ to 4½ years. Only raw apple consumption is considered and it is assumed that each eating occasion represents one apple or less. The data show a typically skewed distribution where the mean value is 65 g whilst the median is 59 g. Interestingly, the modal (most frequently reported) value is exactly 100 g and represents nearly 10% of the 1299 data points. This is probably due to carelessness by the reporting adults and illustrates the need to maintain caution when interpreting such data. The maximum value is 217 g but since this survey is based on a sample of only four days it is unlikely that this level of consumption would be sustained over a longer period. A better estimate of high-level intake over the longer term might be the 97.5th percentile, which is 150 g. Nevertheless, for organophosphate pesticides the concern relates to possible inhibition of the cholinesterase

Risk analysis 27

Fig. 2.3

Incidence of pesticide residues in individual apples.

enzyme and this can occur very rapidly. In this case the amount of pesticide residue ingested on a single occasion or over one day is more relevant to the risk assessment. In the Monte Carlo analysis samples are drawn at random from the residue distribution and then from the apple consumption distribution to provide the data points for the intake distribution. This sequence is repeated several thousand times until a smooth intake distribution curve is produced. The intake distribution shown in Fig. 2.5 represents 20,000 samples drawn from the pesticide residues and apple consumption distributions shown in Figs 2.3 and 2.4. The bars represent the relative frequency of each intake level and the line is the cumulative frequency distribution. The distribution is very skewed and it can be seen that the cumulative frequency is nearing 100% when only the mid-point of the distribution is being approached. This means that very high intakes are relatively rare occurrences.

Fig. 2.4

Daily apple consumption by UK pre-school children.

28

Food chemical safety

Fig. 2.5

Potential intake of an OP pesticide by UK pre-school children.

The mean intake is 0.044 mg and the 97.5th percentile is 0.171 mg. The maximum possible value (0.85 mg) does not appear in the distribution indicating its relative improbability (1 in 20,000). The US Environmental Protection Agency currently uses the 99th percentile,10 which for this distribution is 0.491 mg. It should be noted that these figures relate only to consumers of apple and assume that all apples are treated with this particular product. If the proportion of the population who do not eat raw apple and the proportion of apples that are actually treated with this particular product are taken into account, then intakes will be correspondingly lower at any given percentile. Monte Carlo analysis does not in itself provide a solution for risk managers. All the method can do is to provide the best possible representation of the real situation. The acceptability of a given proportion of consumers being exposed to a given residue level will depend on such factors as the nature of the hazard and the size of any safety margins, if present. For example, occasional minor stomach upsets would be far more acceptable than seizure or sudden death.

2.5

Risk evaluation

Risk evaluation is an apparently simple task of comparing an estimate of intake with the PTWI. If intakes are below the PTWI then there is no risk whereas if they exceed the PTWI then some risk management action may be required. For non-thresholded contaminants risk is assumed to be proportional to intake and therefore intakes should be as low as practically achievable. In practice risk evaluation is a far less certain science. A vital and often over-looked aspect of risk evaluation is ensuring that the estimate of intake corresponds to the PTWI so that like is being compared with like. For example, toxicological end-points are frequently time-related. On rare

Risk analysis 29 occasions, such as in the case of the bacterial botulinum toxins, a single dose can rapidly lead to poisoning and even death. For other end-points, such as carcinogenicity or kidney damage caused by cadmium exposure, it is the cumulative dose over long periods that is significant. However, risk evaluations cannot be conveniently divided into acute and chronic scenarios. In the case of the neurological and behavioural consequences of lead exposure, sustained exposure during a particularly vulnerable phase of life (early childhood) is regarded as being critical. For teratogenic substances the mother’s intake during a particular stage of foetal development can be critical whilst exposure at any other time is insignificant.

2.6

Methods for risk management

The outputs from risk assessment will normally include information about the relationship between dose and risk and estimates of levels of doses and thus risks in the population. For contaminants that have a toxicological threshold the Provisional Tolerable Weekly Intake (PTWI) might be defined and the number of consumers who have the potential to exceed this level of intake quantified. If a PTWI cannot be established (such as for genotoxic carcinogens) then it may be possible to quantify the proportion of a population exposed to a given level of risk by using QRA methods. If QRA methods cannot be applied then a qualitative assessment can be made such as to reduce intake levels to as low as is reasonably practicable. In either case it is the function of risk management to identify an optimal course of action to minimise the risk to consumers. 2.6.1 Contaminants with toxicological thresholds In many cases the risk assessment might indicate that the proportion of consumers with the potential to exceed the PTWI is zero and that there is no need for any action to be taken to control risks. However, even in these circumstances other risk management activities, such as risk communication, might be appropriate. If there is the potential for consumers to exceed the PTWI then it will be necessary to consider measures to control risks. The simplest approach would be to identify the level of contamination in a foodstuff that would cause a highlevel consumer to exceed the PTWI and to introduce legislation or some other form of control to establish that as the maximum permitted concentration. In practice there may be multiple routes of potential exposure and so all potentially affected foods should be taken into account. Setting maximum permitted concentrations for all affected foods would then require that the PTWI be apportioned between foods according to the potential intake from each food. This method assumes that a high-level consumer always consumes foods that contain the contaminant at the maximum permitted level. In reality this is an

30

Food chemical safety

unlikely scenario and so the method would lead to unnecessarily conservative restrictions. Alternative methods for setting standards where several possible food routes are involved include probabilistic modelling which can take into account the proportion of a particular foodstuff that might be contaminated at a particular level. In making risk management decisions it is necessary to take into account information other than technical data. For example, if several different foods are affected by contamination and action needs to be taken to reduce intakes then it may be much cheaper to reduce levels in one food (perhaps by making small modifications to processing variables) than another (where, for example, only suspending supply would suffice). In such cases it would be necessary to take into account the cost-effectiveness of different control options before identifying an optimal risk management strategy. 2.6.2 Contaminants without toxicological thresholds Contaminants such as carcinogens that are assumed to have no toxicological threshold are usually managed by setting maximum permissible concentrations at the lowest levels practically achievable. In practice this is often interpreted as being the lowest concentration that can be readily detected using current analytical methods. This is because any lower level would be unenforceable. Whilst providing a practical solution the levels selected are actually quite arbitrary, depending as they do on the best analytical performance available at any given time. As analytical methods improve so maximum permitted levels would be expected to decrease. For non-thresholded contaminants some mechanism is required that will allow the benefits in terms of reduced risks and costs associated with control to be taken into account. The costs of control will include enforcement costs as well as costs to producers in reaching ever stricter standards. Ultimately these costs will be borne by consumers in taxes, increased prices or reduced choice. Economic theory dictates that there must be a point where the extra increase in the cost of control is not justified by the corresponding increase in benefit (reduction in risk). This optimal point will differ for each contaminant according to the technology needed to control it, the nature of the hazard, and the relationship between dose and risk. It is in this latter context that quantitative risk assessment (QRA) becomes critical (see section 2.3.4 of this chapter). 2.6.3 Consumer perceptions of risk In making risk management decisions it is important to take into account nontechnical factors in addition to scientific and economic information. Recent crises in the food industry have indicated that consumers’ perceptions about risks are driven factors that would not be considered in conventional risk assessments. Research has shown that factors such as whether sub-groups

Risk analysis 31 (particularly children) might be affected, whether the hazard is familiar, if there are effects on the environment or if risks and benefits are equitably shared can determine consumers’ reactions to an issue. Risk managers must be aware that in the event of a crisis consumers’ perceptions about risks can have as great or greater impact on the outcome than the real food safety issues. 2.6.4 Decision analysis Decisions about risk need to take into account a wide range of quantitative and qualitative factors if they are to reflect both the true nature of the risk and the social context in which it is expressed. It is particularly difficult to balance scientific facts against consumer demands. A current example of this problem underlies a trade dispute between the EU and USA. Independent experts and the WHO/FAO Joint Expert Committee on Food Additives and Contaminants have consistently advised that the use of certain naturally occurring hormones as growth promoters in animal production does not present a health hazard to consumers. The US Food and Drugs Administration endorses this view and has approved the use of these substances in American agriculture. In the EU the Medicines Control Agency has imposed a moratorium on the use of these substances, largely because of consumer objections.11 The consequence of this is that imports of hormone-treated meat from the USA are prohibited in Europe. The UN World Trade Organisation (WTO) has judged in favour of the US position because the WTO is allowed only to take scientific data into account in their decisions. More recently the EU Scientific Committee on Veterinary Measures Relating to Public Health has concluded that no threshold levels can be defined for the endocrine, developmental, immunological, neurobiological, immunotoxic, genotoxic and carcinogenic effects associated with hormone residues in bovine meat and meat and that the available data do not enable a quantitative estimate of the risk. It is yet to be known whether the WTO will be persuaded by this argument. In reality, decisions about public health are often made on the basis of politics rather than science. If this were not so, more resources would be committed to controlling food poisoning relating to micro-organisms and less to relatively minor health hazards such as pesticides and environmental contaminants. Political decision-making needs to balance the needs of public health against legitimate consumer expectations. Recent trends have seen more open acknowledgement of the need to balance scientific and social factors. For example, in the UK the Food Advisory Committee which advised food ministers on food safety issues is comprised of a wide range of expertise including toxicologists, chemists, food technologists, economists, and representatives of consumer organisations and the food industry. The Committee is thus able to provide a balanced view which takes all interests into account.

32

Food chemical safety

2.6.5 Risk communication An important part of the risk management process involves informing consumers, industry and other stakeholders of the decisions made by regulatory authorities. However, this is a narrow view of risk communication which does not take into account the potential for dialogue between interested parties that can result in better decision-making. Understanding how consumers view the risks associated with chemical contaminants in food can help to avoid either under- or over-regulation. There are more ways of controlling exposures to chemical contaminants in food than by simply imposing maximum permitted concentrations. Providing advice and information about routes of contamination may allow food producers to take simple precautions to control contamination. Whilst each producer might introduce similar methods they will not all necessarily be able to reduce contaminant levels to the same level. However, the overall effect on the food supply might be sufficient to lower consumers’ intakes to acceptable levels. This principle is enshrined in the Hazard Analysis Critical Control Points (HACCP) approach which is used to control microbiological hazards. Here the emphasis is on exchanging information about good practice rather than imposing absolute limits. Risk communication can of itself be a useful risk management tool. Regulatory authorities have for many years gathered information about the levels of chemical contamination in foods. However, such technical information is difficult to disseminate and many consumers and food producers are probably unaware that it takes place. Some authorities have taken the decision that the brand names associated with foods should be released along with the data. This ‘name and shame’ approach will allow consumers to take avoiding action if they wish. The consequence will probably be that those food producers and retailers with a good brand image to protect will make more stringent efforts to identify and control chemical contaminants in their products.

2.7

Future trends in risk analysis

Risk analysis, like most scientific disciplines, is subject to continual evolution. This means that methods and concepts that are in common use today may well become discredited and obsolete in the future. This presents a problem for risk managers because it creates the impression that everything done before was somehow ‘wrong’. In fact most changes are gradual and tend to take effect at the margins rather than overturning all previous assessments. An example of this is the introduction of acute risk assessments for pesticide residues. It must not be assumed that all pesticides now present acute risks for consumers. In fact only a minority even hold the potential for acute effects and of these probably only a small proportion will require any action to prevent the possibility of adverse events. Similarly the development of PB-PK methods in dose-response modelling and aggregate exposure estimates will probably affect only a small proportion of the chemical contaminants previously assessed.

Risk analysis 33 2.7.1 Physiologically-based pharmacokinetic (PB-PK) modelling In section 2.3 of this chapter the present approach to characterisation of doseresponse relationships was described. In most cases it is necessary to extrapolate from animal species that are used in testing to humans. It may also be necessary to extrapolate from experimental conditions to real human exposures. At the present time default assumptions (which are assumed to be conservative) are applied to convert experimental data into predictive human risk assessments. However, the rates at which a particular substance is adsorbed, distributed, metabolised and excreted can vary considerably between animal species and this can introduce considerable uncertainties into the risk assessment process. The aim of PB-PK models is to quantify these differences as far as possible and so to be able to make more reliable extrapolations. The PB-PK model is based on a mathematical representation of the physiological and biological structure of the species being described. Certain physiological parameters such as blood flow into an organ are specific for each species and so a multi-compartmental model can be assembled which can be used to predict the behaviour of a chemical over a wide range of conditions. Data from animal studies can be used to investigate the absorption, distribution, metabolism and elimination of a particular substance in a given species. Once the behaviour of the chemical in the animal model is fully understood the physiological parameters can be altered so that they represent a human. This will allow the behaviour of the chemical in the human system to be predicted. The use of biological markers such as levels of the chemical in blood and urine can be used to validate the model, particularly if human volunteer studies are available. PB-PK modelling allows further refinement of the dose-response evaluation by partitioning the relationship into pharmacokinetic (exposure vs. tissues dose) and pharmacodynamic (tissue dose vs. toxic response) components. This allows the uncertainties associated with each component to be assessed separately and adds accuracy to the overall animal to man extrapolation. Future developments of PB-PK modelling may allow specific sub-populations such as the newborn or individuals with metabolic variations to be taken into account. However, before this can be done there will need to be considerable growth in the amounts of physiological, pharmacokinetic and pharmacodynamic information available. 2.7.2 Aggregate exposure assessment Until very recently the risks associated with different types of chemicals such as food additives, pesticides, environmental contaminants and natural constituents of food were assessed and managed separately. However, a particular substance might fall into two or more of these categories and so the opportunity for simultaneous exposure might be overlooked. Furthermore, exposure to a chemical could occur through diet, drinking water, air pollution or dermal absorption. Aggregate exposure assessment aims to take all of the possible sources and routes of exposure into account in a realistic manner and thereby obtain a better overall estimate of risk. Initiatives have been set up in both the

34

Food chemical safety

consumer affairs directorate of the European Commission and the Environmental Protection Agency in the USA. Aggregate exposure assessment is naturally more complex than the methods used for dietary risk assessment. In the simplest analysis a worst case can be established for each source and exposure route and then summed to give a total exposure. If this were below any threshold of concern such as the PTWI then no further action would be required. However, if the total worst case exposure was above a PTWI then it is unlikely to reflect the real situation since the probability that any individual would be exposed to each source by each route at the maximum level is very remote. When the EPA considered exposures to insecticide residues in the home they identified at least six possible sources and routes; these are given in Table 2.6. Their original approach apportioned the acceptable daily intake (ADI) between the various routes but it soon became clear that this was unrealistic because an individual was unlikely to be exposed via all routes on any one day. The EPA’s present strategy is to develop an approach called micro-exposure event modelling. Micro-exposure event modelling is based on statistical data on the frequencies and levels of contamination of food, water, etc. and on behavioural information about the frequency of use of lawn/pet/timber treatments, etc. The combined data are assembled in a probabilistic model called ‘LIFELINE’ which is able to predict the frequency and level of exposure to a group of hypothetical individuals over their lifetime.12 The model is also able to take account of the relative proportions of different types of accommodation, the incidence of pet ownership or any other data that will affect real levels of exposure. The output from the LIFELINE model allows the exposures of individuals in a population to be modelled over any interval from a single occasion to a lifetime. Micro-exposure event modelling combined with probabilistic modelling provides a great opportunity to assess real aggregate exposures in the real world. However, the method is highly dependent on the availability of complete and accurate information about levels of contamination and human behaviour. Whilst some of this is available, particularly in the context of pesticides, for other chemicals it may be a long time before this approach can be employed.

Table 2.6 Source

Route

Residues in food Residues in drinking water Residues in water Pet treatments Lawn treatments Timber treatments

oral oral inhalation/dermal (volatilisation in showers, etc.) oral/inhalation/dermal oral/inhalation/dermal oral/inhalation/dermal

Risk analysis 35

2.8

Sources of further information and advice

World-Wide-Web resources The EU Commission Directorate General on Health and Consumer Protection World Health Organisation UK Pesticides Safety Directorate UK Food Standards Agency Joint FAO/WHO Expert Committee on Food Additives and Contaminants Institute of Food Science and Technology (IFST) International Society of Exposure Analysis (ISEA)

http://europa.eu.int/_omm./dgs/ health_consumer/index_en.htm http://www.who.int/ http://www.pesticides.gov.uk/ http://www.foodstandards.gov.uk/ http://www.inchem.org/ aboutjecfa.html http://www.ifst.org/ http://www.ISEAweb.org

Publications The Economics of Food Safety. Elsevier Applied Science Publishers Ltd (1991). ISBN 0-444-01614-7. TENNANT, D. R. Food Chemical Risk Analysis. Blackie Academic and Professional, Chapman and Hall, London (1997). ISBN 0-412-723107. CASWELL, J. A.

2.9 1. 2. 3.

4.

5. 6.

References and STEADMAN, J. H. (1990). Criteria for setting quantitative estimates of acceptable intakes of chemicals in food in the UK. Food Additives and Contaminants 7/3, 287–302. MCDONALD, A. L., FIELDER, R. J., DIGGLE, G. E., TENNANT, D. R. and FISCHER, C. E. (1996). Carcinogens in food: priorities for regulatory action. Human and Experimental Toxicology 15 739–46. LOVELL, D. P. and THOMAS, G. (1997). Quantitative risk assessment. pp. 57– 86 in TENNANT, D. R. (ed.) Food Chemical Risk Analysis. Blackie Academic and Professional, Chapman and Hall, London. ISBN 0-412723107. DOUGLASS, J. S. and TENNANT, D. R. (1997). Estimations of dietary intake of food chemicals. pp. 195–215 in TENNANT, D. R. (ed.) Food Chemical Risk Analysis. Blackie Academic and Professional, Chapman and Hall, London. ISBN 0-412-723107. MAFF UK Food Surveillance Information Sheets. Number 160, September 1998. Lead, cadmium copper and zinc in offals. Can be down-loaded from the FAO website at: http://apps.fao.org/lim500/ nph-wrap.pl?FoodBalanceSheet&Domain=FoodBalanceSheet RUBERY, E. D., BARLOW, S. M.

36 7. 8. 9. 10. 11.

12.

Food chemical safety and WISEMAN, M. (1990). The Dietary and Nutritional Survey of British Adults. HMSO, London. GREGORY, J. R., COLLINS, D. L., DAVIES, P. S. W., HUGHES, J. M. and CLARKE P. C. (1995). National Diet and Nutrition Survey: Children aged 1½ to 4½ years. HMSO, London. ISBN 0-11-691611-7. PESTICIDES SAFETY DIRECTORATE (1998). The Occurrence of Unit to Unit Variability of Pesticide Residues in Fruit and Vegetables. PSD, UK. OFFICE OF PESTICIDE PROGRAMS (1998). Guidance for Submission of Probabilistic Exposure Assessments to the Office of Pesticides Programs’ Health Effects Division. US Environmental Protection Agency. Opinion of the Scientific Committee on Veterinary Measures Relating to Public Health: Assessment of potential risks to human health from hormone residues in bovine meat and meat products (30 April 1999). http://europa.eu.int/comm/dg24/health/sc/scv/out21_en.html For more information contact the lifeline web page at Hampshire Research: http://www.hampshire.org/welcome.html GREGORY, J., TYLER, H.

3 Analytical methods: quality control and selection R. Wood, Food Standards Agency, London

3.1

Introduction

It is now universally recognised as being essential that a laboratory produces and reports data that are fit-for-purpose. For a laboratory to produce consistently reliable data it must implement an appropriate programme of quality assurance measures; such measures are now required by virtue of legislation for food control work or, in the case of the UK Food Standards Agency (FSA), in their requirements for contractors undertaking surveillance work. Thus customers now demand of providers of analytical data that their data meet established quality requirements. These are further described below. The significance of the measures identified are then described and some indications are given as to the future of analytical methodology within the food laboratory. These are then discussed. Methods of analysis have been prescribed by legislation for a number of foodstuffs since the UK acceded to the European Community in 1972. However, the Community now recognises that the quality of results from a laboratory is equally as important as the method used to obtain the results. This is best illustrated by consideration of the Council Directive on the Official Control of Foodstuffs (OCF) which was adopted by the Community in June, 1989.1 This, and the similar Codex Alimentarius Commission requirements, are described below. As a result of this general recognition there is a general move away from the need to prescribe all analytical methodology in detail towards the prescription of the general quality systems within which the laboratory must operate. This allows greater flexibility to the laboratory without detracting from the quality of results that it will produce.

38

Food chemical safety

3.2

Legislative requirements

3.2.1 The European Union For analytical laboratories in the food sector there are legislative requirements regarding analytical data which have been adopted by the European Union. In particular, methods of analysis have been prescribed by legislation for a number of foodstuffs since the UK acceded to the European Community in 1972. However, the Union now recognises that the competency of a laboratory (i.e. how well it can use a method) is equally as important as the ‘quality’ of the method used to obtain results. The Council Directive on the Official Control of Foodstuffs which was adopted by the Community in 19891 looks forward to the establishment of laboratory quality standards, by stating that ‘In order to ensure that the application of this Directive is uniform throughout the Member States, the Commission shall, within one year of its adoption, make a report to the European Parliament and to the Council on the possibility of establishing Community quality standards for all laboratories involved in inspection and sampling under this Directive’ (Article 13). Following that, in September 1990 the Commission produced a Report which recommended establishing Community quality standards for all laboratories involved in inspections and sampling under the OCF Directive. Proposals on this have now been adopted by the Community in the Directive on Additional Measures Concerning the Food Control of Foodstuffs (AMFC).2 In Article 3 of the AMFC Directive it states: 1.

2.

3.

Member States shall take all measures necessary to ensure that the laboratories referred to in Article 7 of Directive 89/397/EEC1 comply with the general criteria for the operation of testing laboratories laid down in European standard EN 450013 supplemented by Standard Operating Procedures and the random audit of their compliance by quality assurance personnel, in accordance with the OECD principles Nos. 2 and 7 of good laboratory practice as set out in Section II of Annex 2 of the Decision of the Council of the OECD of 12 Mar 1981 concerning the mutual acceptance of data in the assessment of chemicals.4 In assessing the laboratories referred to in Article 7 of Directive 89/ 397/EEC Member States shall: (a) apply the criteria laid down in European standard EN 45002;5 and (b) require the use of proficiency testing schemes as far as appropriate. Laboratories meeting the assessment criteria shall be presumed to fulfil the criteria referred to in paragraph 1. Laboratories which do not meet the assessment criteria shall not be considered as laboratories referred to in Article 7 of the said Directive. Member States shall designate bodies responsible for the assessment of laboratories as referred to in Article 7 of Directive 89/397/EEC. These bodies shall comply with the general criteria for laboratory accreditation bodies laid down in European Standard EN 45003.6

Analytical methods: quality control and selection 4.

39

The accreditation and assessment of testing laboratories referred to in this article may relate to individual tests or groups of tests. Any appropriate deviation in the way in which the standards referred to in paragraphs 1, 2 and 3 are applied shall be adopted in accordance with the procedure laid down in Article 8.

In Article 4, it states: Member States shall ensure that the validation of methods of analysis used within the context of official control of foodstuffs by the laboratories referred to in Article 7 of Directive 89/397/EEC comply whenever possible with the provisions of paragraphs 1 and 2 of the Annex to Council Directive 85/591/EEC of 23 December 1985 concerning the introduction of Community methods of sampling and analysis for the monitoring of foodstuffs intended for human consumption.7 As a result of the adoption of the above directives legislation is now in place to ensure that there is confidence not only in national laboratories but also those of the other Member States. As one of the objectives of the EU is to promote the concept of mutual recognition, this is being achieved in the laboratory area by the adoption of the AMFC directive. The effect of the AMFC Directive is that organisations must consider the following aspects within the laboratory: its organisation, how well it actually carries out analyses, and the methods of analysis used in the laboratory. All these aspects are inter-related, but in simple terms may be thought of as: • becoming accredited to an internationally recognised standard; such accreditation is aided by the use of internal quality control procedures • participating in proficiency schemes, and • using validated methods.

In addition it is important that there is dialogue and co-operation by the laboratory with its customers. This is also required by virtue of the EN 45001 Standard at paragraph 6, and will be emphasised even more in future with the adoption of ISO/IEC Guide 17025.8 The AMFC Directive requires that food control laboratories should be accredited to the EN 45000 series of standards as supplemented by some of the OECD GLP principles. In the UK, government departments have nominated the United Kingdom Accreditation Service (UKAS) to carry out the accreditation of official food control laboratories for all the aspects prescribed in the Directive. However, as the accreditation agency will also be required to comply with the EN 45003 Standard and to carry out assessments in accordance with the EN 45002 Standard, all accreditation agencies that are members of the European Cooperation for Accreditation of Laboratories (EA) may be asked to carry out the accreditation of a food control laboratory within the UK. Similar procedures will be followed in the other Member States, all having or developing equivalent organisations to UKAS. Details of the UK requirements for food control laboratories are described later in this chapter.

40

Food chemical safety

3.2.2

Codex Alimentarius Commission: guidelines for the assessment of the competence of testing laboratories involved in the import and export control of food The decisions of the Codex Alimentarius Commission are becoming of increasing importance because of the acceptance of Codex Standards in the World Trade Organisation agreements. They may be regarded as being semilegal in status. Thus, on a world-wide level, the establishment of the World Trade Organisation (WTO) and the formal acceptance of the Agreements on the Application of Sanitary and Phytosanitary Measures (SPS Agreement) and Technical Barriers to Trade (TBT Agreement) have dramatically increased the status of Codex as a body. As a result, Codex Standards are now seen as de facto international standards and are increasingly being adopted by reference into the food law of both developed and developing countries. Because of the status of the CAC described above, the work that it has carried out in the area of laboratory quality assurance must be carefully considered. One of the CAC Committees, the Codex Committee on Methods of Analysis and Sampling (CCMAS), has developed criteria for assessing the competence of testing laboratories involved in the official import and export control of foods. These were recommended by the Committee at its 21st Session in March 19979 and adopted by the Codex Alimentarius Commission at its 22nd Session in June 1997;10 they mirror the EU recommendations for laboratory quality standards and methods of analysis. The guidelines provide a framework for the implementation of quality assurance measures to ensure the competence of testing laboratories involved in the import and export control of foods. They are intended to assist countries in their fair trade in foodstuffs and to protect consumers. The criteria for laboratories involved in the import and export control of foods, now adopted by the Codex Alimentarius Commission, are: • to comply with the general criteria for testing laboratories laid down in ISO/ IEC Guide 25: 1990 ‘General requirements for the competence of calibration and testing laboratories’8 (i.e. effectively accreditation) • to participate in appropriate proficiency testing schemes for food analysis which conform to the requirements laid down in ‘The International Harmonised Protocol for the Proficiency Testing of (Chemical) Analytical Laboratories’11 (already adopted for Codex purposes by the CAC at its 21st Session in July 1995) • to use, whenever available, methods of analysis that have been validated according to the principles laid down by the CAC • to use internal quality control procedures, such as those described in the ‘Harmonised Guidelines for Internal Quality Control in Analytical Chemistry Laboratories’.12

In addition, the bodies assessing the laboratories should comply with the general criteria for laboratory accreditation, such as those laid down in the ISO/IEC Guide 58:1993: ‘Calibration and testing laboratory accreditation systems – General requirements for operation and recognition’.13

Analytical methods: quality control and selection

41

Thus, as for the European Union, the requirements are based on accreditation, proficiency testing, the use of validated methods of analysis and, in addition, the formal requirement to use internal quality control procedures which comply with the Harmonised Guidelines. Although the EU and Codex Alimentarius Commission refer to different sets of accreditation standards, the ISO/IEC Guide 25: 1990 and EN 45000 series of standards are similar in intent. It is only through these measures that international trade will be facilitated and the requirements to allow mutual recognition to be fulfilled will be achieved.

3.3

FSA surveillance requirements

The Food Standards Agency undertakes food chemical surveillance exercises. It has developed information for potential contractors on the analytical quality assurance requirements for food chemical surveillance exercises. These requirements are describe below but reproduced as an appendix to this chapter; they emphasise the need for a laboratory to produce and report data of appropriate quality. The requirements are divided into three parts dealing with: • Part A: quality assurance requirements for surveillance projects provided by potential contractors at the time that tender documents are completed and when commissioning a survey. Here information is sought on: – the formal quality system in the laboratory if third-party assessed (i.e. if UKAS accredited or GLP compliant) – the quality system if not accredited – proficiency testing – internal quality control – method validation. • Part B: information to be defined by the FSA customer once the contract has been awarded to a contractor, e.g. the sample storage conditions to be used, the methods to be used, the IQC procedures to be used, the required measurement limits (e.g. limit of detection (LOD), limit of determination/ quantification (LOQ), and the reporting limits) • Part C: information to be provided by the contractor on an on-going basis once contract is awarded – to be agreed with the customer to ensure that the contractor remains in ‘analytical control’.

3.4

Laboratory accreditation and quality control

Although the legislative requirements apply only to food-control laboratories, the effect of their adoption is that other food laboratories will be advised to achieve the same standard in order for their results to be recognised as equivalent and accepted for ‘due diligence’ purposes. In addition, the Codex

42

Food chemical safety

requirements affect all organisations involved in international trade and thus provide an important ‘quality umbrella’. As shown above, these include a laboratory to be third-party assessed to international accreditation standards, to demonstrate that it is in statistical control by using appropriate internal quality control procedures, to participate in proficiency testing schemes which provide an objective means of assessing and documenting the reliability of the data it is producing and to use methods of analysis that are ‘fit-for-purpose’. These requirements are summarised below and then described in greater detail later in this chapter. 3.4.1 Accreditation The AMFC Directive requires that food-control laboratories should be accredited to the EN 45000 series of standards as supplemented by some of the OECD GLP principles. In the UK, government departments will nominate the United Kingdom Accreditation Service (UKAS) to carry out the accreditation of official food-control laboratories for all the aspects prescribed in the Directive. However, as the accreditation agency will also be required to comply to EN 45003 Standard and to carry out assessments in accordance with the EN 45002 Standard, any other accreditation agencies that are members of the European Co-operation for Accreditation of Laboratories (EA) may also be nominated to carry out the accreditation. Similar procedures will be followed in the other Member States, all having or developing equivalent organisations to UKAS. It has been the normal practice for UKAS to accredit the scope of laboratories on a method-by-method basis. In the case of official food-control laboratories undertaking non-routine or investigative chemical analysis it is accepted that it is not practical to use an accredited fully documented method in the conventional sense, which specifies each sample type and analyte. In these cases a laboratory must have a protocol defining the approach to be adopted which includes the requirements for validation and quality control. Full details of procedures used, including instrumental parameters, must be recorded at the time of each analysis in order to enable the procedure to be repeated in the same manner at a later date. It is therefore recommended that for official food-control laboratories undertaking analysis, appropriate methods are accredited on a generic basis with such generic accreditation being underpinned where necessary by specific method accreditation. Food-control laboratories seeking to be accredited for the purposes of the Directive should include, as a minimum, the following techniques in generic protocols: HPLC, GC, atomic absorption and/or ICP (and microscopy). A further protocol on sample preparation procedures (including digestion and solvent dissolution procedures) should also be developed. Other protocols for generic methods which are acceptable to UKAS may also be developed. Proximate analyses should be addressed as a series of specific methods including moisture, fat, protein and ash determinations.

Analytical methods: quality control and selection

43

Where specific Regulations are in force then the methods associated with the Regulations shall be accredited if the control laboratory wishes to offer enforcement of the Regulations to customers. Examples of these are methods of analysis for aflatoxins and methods of analysis for specific and overall migration for food contact materials. By using the combination of specific method accreditation and generic accreditation it will be possible for laboratories to be accredited for all the analyses of which they are capable and competent to undertake. Method performance validation data demonstrating that the method was fit-for-purpose shall be demonstrated before the test result is released and method performance shall be monitored by on-going quality-control techniques where applicable. It will be necessary for laboratories to be able to demonstrate quality-control procedures to ensure compliance with the EN 45001 Standard,3 an example of which would be compliance with the ISO/AOAC/IUPAC Guidelines on Internal Quality Control in Analytical Chemistry Laboratories.12 3.4.2 Internal quality control (IQC) IQC is one of a number of concerted measures that analytical chemists can take to ensure that the data produced in the laboratory are of known quality and uncertainty. In practice this is determined by comparing the results achieved in the laboratory at a given time with a standard. IQC therefore comprises the routine practical procedures that enable the analyst to accept a result or group of results or reject the results and repeat the analysis. IQC is undertaken by the inclusion of particular reference materials, ‘control materials’, into the analytical sequence and by duplicate analysis. ISO, IUPAC and AOAC INTERNATIONAL have co-operated to produce agreed protocols on the ‘Design, Conduct and Interpretation of Collaborative Studies’14 and on the ‘Proficiency Testing of [Chemical] Analytical Laboratories’.11 The Working Group that produced these protocols has prepared a further protocol on the internal quality control of data produced in analytical laboratories. The document was finalised in 1994 and published in 1995 as the ‘Harmonised Guidelines For Internal Quality Control In Analytical Chemistry Laboratories’.12 The use of the procedures outlined in the Protocol should aid compliance with the accreditation requirements specified above. Basic concepts The protocol sets out guidelines for the implementation of internal quality control (IQC) in analytical laboratories. IQC is one of a number of concerted measures that analytical chemists can take to ensure that the data produced in the laboratory are fit for their intended purpose. In practice, fitness for purpose is determined by a comparison of the accuracy achieved in a laboratory at a given time with a required level of accuracy. Internal quality control therefore comprises the routine practical procedures that enable the analytical chemist to accept a result or group of results as fit-for-purpose, or

44

Food chemical safety

reject the results and repeat the analysis. As such, IQC is an important determinant of the quality of analytical data, and is recognised as such by accreditation agencies. Internal quality control is undertaken by the inclusion of particular reference materials, called ‘control materials’, into the analytical sequence and by duplicate analysis. The control materials should, wherever possible, be representative of the test materials under consideration in respect of matrix composition, the state of physical preparation and the concentration range of the analyte. As the control materials are treated in exactly the same way as the test materials, they are regarded as surrogates that can be used to characterise the performance of the analytical system, both at a specific time and over longer intervals. Internal quality control is a final check of the correct execution of all of the procedures (including calibration) that are prescribed in the analytical protocol and all of the other quality assurance measures that underlie good analytical practice. IQC is therefore necessarily retrospective. It is also required to be as far as possible independent of the analytical protocol, especially the calibration, that it is designed to test. Ideally both the control materials and those used to create the calibration should be traceable to appropriate certified reference materials or a recognised empirical reference method. When this is not possible, control materials should be traceable at least to a material of guaranteed purity or other well characterised material. However, the two paths of traceability must not become coincident at too late a stage in the analytical process. For instance, if control materials and calibration standards were prepared from a single stock solution of analyte, IQC would not detect any inaccuracy stemming from the incorrect preparation of the stock solution. In a typical analytical situation several, or perhaps many, similar test materials will be analysed together, and control materials will be included in the group. Often determinations will be duplicated by the analysis of separate test portions of the same material. Such a group of materials is referred to as an analytical ‘run’. (The words ‘set’, ‘series’ and ‘batch’ have also been used as synonyms for ‘run’.) Runs are regarded as being analysed under effectively constant conditions. The batches of reagents, the instrument settings, the analyst, and the laboratory environment will, under ideal conditions, remain unchanged during analysis of a run. Systematic errors should therefore remain constant during a run, as should the values of the parameters that describe random errors. As the monitoring of these errors is of concern, the run is the basic operational unit of IQC. A run is therefore regarded as being carried out under repeatability conditions, i.e. the random measurement errors are of a magnitude that would be encountered in a ‘short’ period of time. In practice the analysis of a run may occupy sufficient time for small systematic changes to occur. For example, reagents may degrade, instruments may drift, minor adjustments to instrumental settings may be called for, or the laboratory temperature may rise. However, these systematic effects are, for the purposes of IQC, subsumed into the

Analytical methods: quality control and selection

45

repeatability variations. Sorting the materials making up a run into a randomised order converts the effects of drift into random errors. Scope of the guidelines The guidelines are a harmonisation of IQC procedures that have evolved in various fields of analysis, notably clinical biochemistry, geochemistry, environmental studies, occupational hygiene and food analysis. There is much common ground in the procedures from these various fields. However, analytical chemistry comprises an even wider range of activities, and the basic principles of IQC should be able to encompass all of these. The guidelines will be applicable in the great majority of instances although there are a number of IQC practices that are restricted to individual sectors of the analytical community and so not included in the guidelines. In order to achieve harmonisation and provide basic guidance on IQC, some types of analytical activity have been excluded from the guidelines. Issues specifically excluded are as follows: • Quality control of sampling. While it is recognised that the quality of the analytical result can be no better than that of the sample, quality control of sampling is a separate subject and in many areas not yet fully developed. Moreover, in many instances analytical laboratories have no control over sampling practice and quality. • In-line analysis and continuous monitoring. In this style of analysis there is no possibility of repeating the measurement, so the concept of IQC as used in the guidelines is inapplicable. • Multivariate IQC. Multivariate methods in IQC are still the subject of research and cannot be regarded as sufficiently established for inclusion in the guidelines. The current document regards multianalyte data as requiring a series of univariate IQC tests. Caution is necessary in the interpretation of this type of data to avoid inappropriately frequent rejection of data. • Statutory and contractual requirements. • Quality assurance measures such as pre-analytical checks on instrumental stability, wavelength calibration, balance calibration, tests on resolution of chromatography columns, and problem diagnostics are not included. For present purposes they are regarded as part of the analytical protocol, and IQC tests their effectiveness together with the other aspects of the methodology.

Recommendations The following recommendations represent integrated approaches to IQC that are suitable for many types of analysis and applications areas. Managers of laboratory quality systems will have to adapt the recommendations to the demands of their own particular requirements. Such adoption could be implemented, for example, by adjusting the number of duplicates and control material inserted into a run, or by the inclusion of any additional measures favoured in the particular application area. The procedure finally chosen and its accompanying decision rules must be

46

Food chemical safety

codified in an IQC protocol that is separate from the analytical system protocol. The practical approach to quality control is determined by the frequency with which the measurement is carried out and the size and nature of each run. The following recommendations are therefore made. (The use of control charts and decision rules are covered in Appendix 1 to the guidelines.) In all of the following the order in the run in which the various materials are analysed should be randomised if possible. A failure to randomise may result in an underestimation of various components of error. Short (e.g. n < 20) frequent runs of similar materials Here the concentration range of the analyte in the run is relatively small, so a common value of standard deviation can be assumed. Insert a control material at least once per run. Plot either the individual values obtained, or the mean value, on an appropriate control chart. Analyse in duplicate at least half of the test materials, selected at random. Insert at least one blank determination Longer (e.g. n > 20) frequent runs of similar materials Again a common level of standard deviation is assumed. Insert the control material at an approximate frequency of one per ten test materials. If the run size is likely to vary from run to run it is easier to standardise on a fixed number of insertions per run and plot the mean value on a control chart of means. Otherwise plot individual values. Analyse in duplicate a minimum of five test materials selected at random. Insert one blank determination per ten test materials. Frequent runs containing similar materials but with a wide range of analyte concentration Here we cannot assume that a single value of standard deviation is applicable. Insert control materials in total numbers approximately as recommended above. However, there should be at least two levels of analyte represented, one close to the median level of typical test materials, and the other approximately at the upper or lower decile as appropriate. Enter values for the two control materials on separate control charts. Duplicate a minimum of five test materials, and insert one procedural blank per ten test materials. Ad hoc analysis Here the concept of statistical control is not applicable. It is assumed, however, that the materials in the run are of a single type. Carry out duplicate analysis on all of the test materials. Carry out spiking or recovery tests or use a formulated control material, with an appropriate number of insertions (see above), and with different concentrations of analyte if appropriate. Carry out blank determinations. As no control limits are available, compare the bias and precision with fitness-for-purpose limits or other established criteria. By following the above recommendations laboratories would introduce internal quality control measures which are an essential aspect of ensuring that data

Analytical methods: quality control and selection

47

released from a laboratory are fit-for-purpose. If properly executed, quality control methods can monitor the various aspects of data quality on a run-by-run basis. In runs where performance falls outside acceptable limits, the data produced can be rejected and, after remedial action on the analytical system, the analysis can be repeated. The guidelines stress, however, that internal quality control is not foolproof even when properly executed. Obviously it is subject to ‘errors of both kinds’, i.e. runs that are in control will occasionally be rejected and runs that are out of control occasionally accepted. Of more importance, IQC cannot usually identify sporadic gross errors or short-term disturbances in the analytical system that affect the results for individual test materials. Moreover, inferences based on IQC results are applicable only to test materials that fall within the scope of the analytical method validation. Despite these limitations, which professional experience and diligence can alleviate to a degree, internal quality control is the principal recourse available for ensuring that only data of appropriate quality are released from a laboratory. When properly executed it is very successful. The guidelines also stress that the perfunctory execution of any quality system will not guarantee the production of data of adequate quality. The correct procedures for feedback, remedial action and staff motivation must also be documented and acted upon. In other words, there must be a genuine commitment to quality within a laboratory for an internal quality control programme to succeed, i.e. the IQC must be part of a complete quality management system.

3.5

Proficiency testing

Participation in proficiency testing schemes provides laboratories with an objective means of assessing and documenting the reliability of the data they are producing. Although there are several types of proficiency testing schemes they all share a common feature: test results obtained by one laboratory are compared with those obtained by one or more testing laboratories. The proficiency testing schemes must provide a transparent interpretation and assessment of results. Laboratories wishing to demonstrate their proficiency should seek and participate in proficiency testing schemes relevant to their area of work. The need for laboratories carrying out analytical determinations to demonstrate that they are doing so competently has become paramount. It may well be necessary for such laboratories not only to become accredited and to use fully validated methods but also to participate successfully in proficiency testing schemes. Thus, proficiency testing has assumed a far greater importance than previously. 3.5.1 What is proficiency testing? A proficiency testing scheme is defined as a system for objectively checking laboratory results by an external agency. It includes comparison of a

48

Food chemical safety

laboratory’s results at intervals with those of other laboratories, the main object being the establishment of trueness. In addition, although various protocols for proficiency testing schemes have been produced the need now is for a harmonised protocol that will be universally accepted; the progress towards the preparation and adoption of an internationally recognised protocol is described below. Various terms have been used to describe schemes conforming to the protocol (e.g. external quality assessment, performance schemes, etc.), but the preferred term is ‘proficiency testing’. Proficiency testing schemes are based on the regular circulation of homogeneous samples by a co-ordinator, analysis of samples (normally by the laboratory’s method of choice) and an assessment of the results. However, although many organisations carry out such schemes, there has been no international agreement on how this should be done – in contrast to the collaborative trial situation. In order to rectify this, the same international group that drew up collaborative trial protocols was invited to prepare one for proficiency schemes (the first meeting to do so was held in April 1989). Other organisations, such as CEN, are also expected to address the problem. 3.5.2 Why proficiency testing is important Participation in proficiency testing schemes provides laboratories with a means of objectively assessing, and demonstrating, the reliability of the data they produce. Although there are several types of schemes, they all share a common feature of comparing test results obtained by one testing laboratory with those obtained by other testing laboratories. Schemes may be ‘open’ to any laboratory or participation may be invited. Schemes may set out to assess the competence of laboratories undertaking a very specific analysis (e.g. lead in blood) or more general analysis (e.g. food analysis). Although accreditation and proficiency testing are separate exercises, it is anticipated that accreditation assessments will increasingly use proficiency testing data. 3.5.3 Accreditation agencies It is now recommended by ISO Guide 25,8 the prime standard to which accreditation agencies operate, that such agencies require laboratories seeking accreditation to participate in an appropriate proficiency testing scheme before accreditation is gained. There is now an internationally recognised protocol to which proficiency testing schemes should comply; this is the IUPAC/AOAC/ ISO Harmonised Protocol described below. 3.5.4

ISO/IUPAC/AOAC International Harmonised Protocol For Proficiency Testing of (Chemical) Analytical Laboratories The International Standardising Organisations, AOAC, ISO and IUPAC, have co-operated to produce an agreed ‘International Harmonised Protocol For

Analytical methods: quality control and selection

49

Proficiency Testing of (Chemical) Analytical Laboratories’.11 That protocol is recognised within the food sector of the European Community and also by the Codex Alimentarius Commission. The protocol makes the following recommendations about the organisation of proficiency testing, all of which are important in the food sector. Framework Samples must be distributed regularly to participants who are to return results within a given time. The results will be statistically analysed by the organiser and participants will be notified of their performance. Advice will be available to poor performers and participants will be kept fully informed of the scheme’s progress. Participants will be identified by code only, to preserve confidentiality. The scheme’s structure for any one analyte or round in a series should be: • • • • • • • •

samples prepared samples distributed regularly participants analyse samples and report results results analysed and performance assessed participants notified of their performance advice available for poor performers, on request co-ordinator reviews performance of scheme next round commences.

Organisation The running of the scheme will be the responsibility of a co-ordinating laboratory/organisation. Sample preparation will either be contracted out or undertaken in house. The co-ordinating laboratory must be of high reputation in the type of analysis being tested. Overall management of the scheme should be in the hands of a small steering committee (Advisory Panel) having representatives from the co-ordinating laboratory (who should be practising laboratory scientists), contract laboratories (if any), appropriate professional bodies and ordinary participants. Samples The samples to be distributed must be generally similar in matrix to the unknown samples that are routinely analysed (in respect of matrix composition and analyte concentration range). It is essential they are of acceptable homogeneity and stability. The bulk material prepared must be effectively homogeneous so that all laboratories will receive samples that do not differ significantly in analyte concentration. The co-ordinating laboratory should also show the bulk sample is sufficiently stable to ensure it will not undergo significant change throughout the duration of the proficiency test. Thus, prior to sample distribution, matrix and analyte stability must be determined by analysis after appropriate storage. Ideally, the quality checks on samples referred should be performed by a different laboratory from that which prepared the sample,

50

Food chemical safety

although it is recognised that this would probably cause considerable difficulty to the co-ordinating laboratory. The number of samples to be distributed per round for each analyte should be no more than five. Frequency of sample distribution Sample distribution frequency in any one series should not be more than every two weeks and not less than every four months. A frequency greater than once every two weeks could lead to problems in turn-round of samples and results. If the period between distributions extends much beyond four months, there will be unacceptable delays in identifying analytical problems and the impact of the scheme on participants will be small. The frequency also relates to the field of application and amount of internal quality control that is required for that field. Thus, although the frequency range stated above should be adhered to, there may be circumstances where it is acceptable for a longer time scale between sample distribution, e.g. if sample throughput per annum is very low. Advice on this respect would be a function of the Advisory Panel. Estimating the assigned value (the `true' result) There are a number of possible approaches to determining the nominally ‘true’ result for a sample but only three are normally considered. The result may be established from the amount of analyte added to the samples by the laboratory preparing the sample; alternatively, a ‘reference’ laboratory (or group of such expert laboratories) may be asked to measure the concentration of the analyte using definitive methods or thirdly, the results obtained by the participating laboratories (or a substantial sub-group of these) may be used as the basis for the nominal ‘true’ result. The organisers of the scheme should provide the participants with a clear statement giving the basis for the assignment of reference values which should take into account the views of the Advisory Panel. Choice of analytical method Participants can use the analytical method of their choice except when otherwise instructed to adopt a specified method. It is recommended that all methods should be properly validated before use. In situations where the analytical result is method-dependent the true value will be assessed using those results obtained using a defined procedure. If participants use a method that is not ‘equivalent’ to the defining method, then an automatic bias in result will occur when their performance is assessed. Performance criteria For each analyte in a round a criterion for the performance score may be set, against which the score obtained by a laboratory can be judged. A ‘running score’ could be calculated to give an assessment of performance spread over a longer period of time.

Analytical methods: quality control and selection

51

Reporting results Reports issued to participants should include data on the results from all laboratories together with participants’ own performance score. The original results should be presented to enable participants to check correct data entry. Reports should be made available before the next sample distribution. Although all results should be reported, it may not be possible to do this in very extensive schemes (e.g. 800 participants determining 15 analytes in a round). Participants should, therefore, receive at least a clear report with the results of all laboratories in histogram form. Liaison with participants Participants should be provided with a detailed information pack on joining the scheme. Communication with participants should be by newsletter or annual report together with a periodic open meeting; participants should be advised of changes in scheme design. Advice should be available to poor performers. Feedback from laboratories should be encouraged so participants contribute to the scheme’s development. Participants should view it as their scheme rather than one imposed by a distant bureaucracy. Collusion and falsification of results Collusion might take place between laboratories so that independent data are not submitted. Proficiency testing schemes should be designed to ensure that there is as little collusion and falsification as possible. For example, alternative samples could be distributed within a round. Also instructions should make it clear that collusion is contrary to professional scientific conduct and serves only to nullify the benefits of proficiency testing. 3.5.5 Statistical procedure for the analysis of results The first stage in producing a score from a result x (a single measurement of analyte concentration in a test material) is to obtain an estimate of the bias, thus: bias ˆ x

X

where X is the true concentration or amount of analyte. The efficacy of any proficiency test depends on using a reliable value for X. Several methods are available for establishing a working estimate of X^ (i.e. the assigned value). Formation of a z-score Most proficiency testing schemes compare bias with a standard error. An obvious approach is to form the z-score given by: z ˆ …x X^ †= where  is a standard deviation.  could be either an estimate of the actual variation encountered in a particular round (~s) estimated from the laboratories’

52

Food chemical safety

results after outlier elimination or a target representing the maximum allowed variation consistent with valid data. A fixed target value for  is preferable and can be arrived at in several ways. It could be fixed arbitrarily, with a value based on a perception of how laboratories should perform. It could be an estimate of the precision required for a specific task of data interpretation.  could be derived from a model of precision, such as the ‘Horwitz Curve’.15 However, while this model provides a general picture of reproducibility, substantial deviation from it may be experienced for particular methods. Interpretation of z-scores If X^ and  are good estimates of the population mean and standard deviation then z will be approximately normally distributed with a mean of zero and unit standard deviation. An analytical result is described as ‘well behaved’ when it complies with this condition. An absolute value of z (jzj) greater than three suggests poor performance in terms of accuracy. This judgement depends on the assumption of the normal distribution, which, outliers apart, seems to be justified in practice. As z is standardised, it is comparable for all analytes and methods. Thus values of z can be combined to give a composite score for a laboratory in one round of a proficiency test. The z-scores can therefore be interpreted as follows: jzj < 2 ‘Satisfactory’: will occur in 95% of cases produced by ‘well behaved results’ 2 < jzj < 3 ‘Questionable’: but will occur in 5% of cases produced by ‘well behaved results’ jzj > 3 ‘Unsatisfactory’: will only occur in 0.1% of cases produced by ‘well behaved results’ Combination of results within a round of the trial There are several methods of combining the z-scores produced by a laboratory in one round of the proficiency test described in the Protocol. They are: The sum of scores, SZ = Rz The sum of squared scores, SSZ = Rz2 The sum of absolute values of the scores, SAZ = R jzj All should be used with caution, however. It is the individual z-scores that are the critical consideration when considering the proficiency of a laboratory. Calculation of running scores Similar considerations apply for running scores as apply to combination scores above.

Analytical methods: quality control and selection

3.6

53

Analytical methods

Analytical methods should be validated as fit-for-purpose before use by a laboratory. Laboratories should ensure that, as a minimum, the methods they use are fully documented, laboratory staff trained in their use and control mechanisms established to ensure that the procedures are under statistical control. The development of methods of analysis for incorporation into International Standards or into foodstuff legislation was, until comparatively recently, not systematic. However, the EU and Codex have requirements regarding methods of analysis and these are outlined below. They are followed by other International Standardising Organisations (e.g. AOAC International (AOACI) and the European Committee for Standardization (CEN)). 3.6.1 Codex Alimentarius Commission This was the first international organisation working at the government level in the food sector that laid down principles for the establishment of its methods. That it was necessary for such guidelines and principles to be laid down reflects the confused and unsatisfactory situation in the development of legislative methods of analysis that existed until the early 1980s in the food sector. The ‘Principles for the Establishment of Codex Methods of Analysis’16 are given below; other organisations which subsequently laid down procedures for the development of methods of analysis in their particular sector followed these principles to a significant degree. They require that preference should be given to methods of analysis the reliability of which have been established in respect of the following criteria, selected as appropriate: • specificity • accuracy • precision; repeatability intra-laboratory (within laboratory), reproducibility inter-laboratory (within laboratory and between laboratories) • limit of detection • sensitivity • practicability and applicability under normal laboratory conditions • other criteria which may be selected as required.

3.6.2 The European Union The Union is attempting to harmonise sampling and analysis procedures in an attempt to meet the current demands of the national and international enforcement agencies and the likely increased problems that the open market will bring. To aid this the Union issued a Directive on Sampling and Methods of Analysis.7 The Directive contains a technical annex, in which the need to carry out a collaborative trial before it can be adopted by the Community is emphasised. The criteria to which Community methods of analysis for foodstuffs should now conform are as stringent as those recommended by any international

54

Food chemical safety

organisation following adoption of the Directive. The requirements follow those described for Codex above, and are given in the Annex to the Directive. They are: 1.

2.

3. 4.

Methods of analysis which are to be considered for adoption under the provisions of the Directive shall be examined with respect to the following criteria: (i) specificity (ii) accuracy (iii) precision; repeatability intra-laboratory (within laboratory), reproducibility inter-laboratory (within laboratory and between laboratories) (iv) limit of detection (v) sensitivity (vi) practicability and applicability under normal laboratory conditions (vii) other criteria which may be selected as required. The precision values referred to in 1 (iii) shall be obtained from a collaborative trial which has been conducted in accordance with an internationally recognised protocol on collaborative trials (e.g. International Organisation of Standardization ‘Precision of Test Methods’).17 The repeatability and reproducibility values shall be expressed in an internationally recognised form (e.g. the 95% confidence intervals as defined by ISO 5725/1981). The results from the collaborative trial shall be published or be freely available. Methods of analysis which are applicable uniformly to various groups of commodities should be given preference over methods which apply to individual commodities. Methods of analysis adopted under this Directive should be edited in the standard layout for methods of analysis recommended by the International Organisations for Standardization.

3.6.3 Other organisations There are other international standardising organisations, most notably the European Committee for Standardization (CEN) and AOACI, which follow similar requirements. Although CEN methods are not prescribed by legislation, the European Commission places considerable importance on the work that CEN carries out in the development of specific methods in the food sector; CEN has been given direct mandates by the Commission to publish particular methods, e.g. those for the detection of food irradiation. Because of this some of the methods in the food sector being developed by CEN are described below. CEN, like the other organisations described above, has adopted a set of guidelines to which its Methods Technical Committees should conform when developing a method of analysis. The guidelines are:

Analytical methods: quality control and selection

55

Details of the interlaboratory test on the precision of the method are to be summarised in an annex to the method. It is to be stated that the values derived from the interlaboratory test may not be applicable to analyte concentration ranges and matrices other than given in annex. The precision clauses shall be worded as follows: Repeatability: The absolute difference between two single test results found on identical test materials by one operator using the same apparatus within the shortest feasible time interval will exceed the repeatability value r in not more than 5% of the cases. The value(s) is (are): . . . Reproducibility: The absolute difference between two single test results on identical test material reported by two laboratories will exceed the reproducibility, R, in not more than 5% of the cases. The value(s) is (are): . . . There shall be minimum requirements regarding the information to be given in an Informative Annex, this being: Year of interlaboratory test and reference to the test report (if available) Number of samples Number of laboratories retained after eliminating outliers Number of outliers (laboratories) Number of accepted results Mean value (with the respective unit) Repeatability standard deviation (sr) (with the respective unit) Repeatability relative standard deviation (RSD r) (%) Repeatability limit (r)w(with the respective units) Reproducibility relative standard deviation (sR) (with the respective unit) Reproducibility relative standard deviation (RSD R) (%) Reproducibility (R) (with the respective unit) Sample types clearly described Notes if further information is to be given.

3.6.4 Requirements of official bodies Consideration of the above requirements confirms that in future all methods must be fully validated if at all possible, i.e. have been subjected to a collaborative trial conforming to an internationally recognised protocol. In addition this, as described above, is now a legislative requirement in the food sector of the European Union. The concept of the valid analytical method in the food sector, and its requirements, is described below.

56

Food chemical safety

3.6.5 Requirements for valid methods of analysis It would be simple to say that any new method should be fully tested for the criteria given above. However, the most ‘difficult’ of these is obtaining the accuracy and precision performance criteria. Accuracy Accuracy is defined as the closeness of the agreement between the result of a measurement and a true value of the measureand.18 It may be assessed with the use of reference materials. However, in food analysis, there is a particular problem. In many instances, though not normally for food additives and contaminants, the numerical value of a characteristic (or criterion) in a Standard is dependent on the procedures used to ascertain its value. This illustrates the need for the (sampling and) analysis provisions in a Standard to be developed at the same time as the numerical value of the characteristics in the Standard are negotiated to ensure that the characteristics are related to the methodological procedures prescribed. Precision Precision is defined as the closeness of agreement between independent test results obtained under prescribed conditions.19 In a standard method the precision characteristics are obtained from a properly organised collaborative trial, i.e. a trial conforming to the requirements of an International Standard (the AOAC/ISO/IUPAC Harmonised Protocol or the ISO 5725 Standard). Because of the importance of collaborative trials, and the resource that is now being devoted to the assessment of precision characteristics of analytical methods before their acceptance, they are described in detail below. Collaborative trials As seen above, all ‘official’ methods of analysis are required to include precision data. These may be obtained by subjecting the method to a collaborative trial conforming to an internationally agreed protocol. A collaborative trial is a procedure whereby the precision of a method of analysis may be assessed and quantified. The precision of a method is usually expressed in terms of repeatability and reproducibility values. Accuracy is not the objective. Recently there has been progress towards a universal acceptance of collaboratively tested methods and collaborative trial results and methods, no matter by whom these trials are organised. This has been aided by the publication of the IUPAC/ISO/AOAC Harmonisation Protocol on Collaborative Studies.14 That Protocol was developed under the auspices of the International Union of Pure and Applied Chemists (IUPAC) aided by representatives from the major organisations interested in conducting collaborative studies. In particular, from the food sector, the AOAC International, the International Organisation for Standardisation (ISO), the International Dairy Federation (IDF), the Collaborative International Analytical Council for Pesticides (CIPAC), the Nordic Analytical Committee (NMKL), the Codex Committee on Methods of

Analytical methods: quality control and selection

57

Analysis and Sampling and the International Office of Cocoa and Chocolate were involved. The Protocol gives a series of 11 recommendations dealing with: • • • • • • • •

the components that make up a collaborative trial participants sample type sample homogeneity sample plan the method(s) to be tested pilot study/pre-trial the trial proper.

3.6.6 Statistical analysis It is important to appreciate that the statistical significance of the results is wholly dependent on the quality of the data obtained from the trial. Data that contain obvious gross errors should be removed prior to statistical analysis. It is essential that participants inform the trial co-ordinator of any gross error that they know has occurred during the analysis and also if any deviation from the method as written has taken place. The statistical parameters calculated and the outlier tests performed are those used in the internationally agreed Protocol for the Design, Conduct and Interpretation of Collaborative Studies.14

3.7

Standardised methods of analysis for contaminants

There are many organisations that publish standardised methods of analysis for contaminants, such methods normally having been validated through a collaborative trial organised to conform to one of the internationally accepted protocols described previously. Such organisations will include AOACI, the European Organisation for Standardisation (CEN) and the Nordic Committee for Food Analysis (NMKL). Within Europe, the most important of these international standardising organisations is probably CEN. CEN has a technical committee dealing with horizontal methods of analysis in which both additive and contaminant methods of analysis are discussed (TC 275). The methods of analysis for contaminants within its work programme are outlined below. This is given by Working Group. The titles under the Working Group heading refer to the work item (topic area) of that Working Group. The Working Groups not listed (e.g. 1, 2, etc.) are concerned with additive methods of analysis. Work programme of CEN TC 275 Working Group 3: Pesticides in Fatty Foods Work Item A: determination of pesticides and polychlorinated biphenyls (PCBs): Part 1: general considerations Part 2: extraction of fat, pesticides and PCBs and determination of fat content

58

Food chemical safety

Part 3: clean-up methods Part 4: determination, confirmatory tests, miscellaneous. Work programme of CEN TC 275 Working Group 4: Pesticides in Non-Fatty Foods Work Item A: multiresidue methods for the gas chromatographic determination of pesticide residues: Part 1: general considerations Part 2: methods for extraction and clean-up Part 3: determination and confirmatory tests. Work Item B: determination of dithiocarbamate and thiuram disulfide residues: Part 1 spectrometric method Part 2: gas chromatographic method Part 3: xanthogenate method. Work Item C: determination of bromide residues: Part 1: determination of total bromide as inorganic bromide Part 2: determination of bromide. Work Item D: determination of N-methyl carbamate residues. Work Item E: determination of benomyl, carbendazim, thiabendazole and thiophanate-methyl. Work programme of CEN TC 275 Working Group 5: Biotoxins Work Item A: determination of aflatoxin B1 and/or the sum of B1, B2, G1 and G2 in cereals, shell fruits and derived products – high-performance liquid chromatographic method with postcolumn derivatisation and immunoaffinity column. Work Item B: determination of ochratoxin A in cereals and cereal products: Part 1: HPLC method with silica gel clean-up Part 2: HPLC method with bicarbonate clean-up. Work Item C: determination of ochratoxin A in cereals and cereal products – HPLC method with immunoaffinity clean-up. Work Item D: determination of patulin content. Work Item E: determination of fumonisins. Work Item F: criteria of analytical methods for mycotoxins – CEN-Report.

Analytical methods: quality control and selection

59

Work Item G: determination of domoic acid in mussels. Work Item H: determination of aflatoxin B1 and total aflatoxins by immunoaffinity column clean-up and HPLC in fig paste, pistachios, peanut butter and paprika powder. Work Item I: determination of okadaic acid and dinophysis toxin in mussels by HPLC. Work Item J: determination of saxitoxin and dicarbamoyl saxitoxin in mussels by HPLC. Work Item K: determination of aflatoxin M1 in liquid milk. Work programme of CEN TC 275 Working Group 6: Microbiology Work Item A: enumeration of Staphylococcus aureus: Part 1: colony count technique with confirmation of colonies (ISO/DIS 68881: 1997) Part 2: colony count technique without confirmation of colonies (ISO/DIS 6888-2: 1997). Work Item B: horizontal method for the detection of coagulase positive Staphylococci (Staphylococci aureus and other species). Work Item C: horizontal method for the detection and enumeration of Listeria monocytogenes: Part 1: detection method Part 2: enumeration method. Work Item D: enumeration of Clostridium perfringens – colony count technique. Work Item E: horizontal method for the detection of Salmonella. Work Item F: general guidance for enumeration of Bacillus cereus – colony count technique at 30ºC. Work Item G: detection of thermotolerant Campylobacter. Work Item H: detection of Yersinia enterocolitica. Work Item I: preparation of the test sample, of initial suspension and of decimal dilutions, for microbiological examination: Part 1: general rules for the preparation of the initial suspension and of decimal dilutions

60

Food chemical safety

Part 2: specific rules for the preparation of the test samples and initial suspension of meat and meat products Part 3: specific rules for the preparation of the test samples and initial suspension of milk and milk products Part 4: specific rules for the preparation of the test samples and initial suspension of fish products Part 5: specific rules for the preparation of the test samples and initial suspension of products other than milk and milk products, meat and meat products and fish products. Work Item J: general guidance for microbiological examinations. Work Item K: validation of alternative microbiological methods. Work Item L: guidelines on quality assurance and performance testing of culture media: Part 1: quality assurance of culture media in the laboratory Part 2: performance testing Part 3: practical implementation of the general guideline on quality assurance of culture media in the laboratory. Part 4: performance testing of culture media. Work Item M: horizontal method for the detection of Escherichia coli O 157. Work Item N: horizontal method for the enumeration of Bacillus cereus. Work Item O: guidelines on quality assurance and performance testing of culture media (to be elaborated as European Prestandards C02/97, C03/97): Part 1: quality assurance of culture media in the laboratory Part 2: practical implementation of the general guidelines on quality assurance of culture media in the laboratory Part 3: performance testing. Work programme of CEN TC 275 Working Group 10: Determination of Trace Elements Work Item A: determination of trace elements – general considerations. Work Item B: determination of mercury by CVAAS after pressure digestion. Work Item C: determination of lead and cadmium by ETAAS after dry ashing. Work Item D: performance criteria and general considerations. Work Item E: pressure digestion.

Analytical methods: quality control and selection

61

Work Item F: determination of lead, cadmium, chromium and molybdenum by ETAAS after pressure digestion. Work Item G: determination of lead, cadmium, zinc, copper, iron, chromium and nickel after dry ashing. Work Item H: determination of lead and cadmium by ETAAS after microwave digestion. Work programme of CEN TC 275 Working Group 11: Genetically Modified Organisms Work Item A: detection of genetically modified organisms and derived products – sampling. Work Item B: detection of genetically modified organisms and derived products – nucleic acid extraction. Work Item C: detection of genetically modified organisms and derived products – qualitative nucleic acid based methods. Work Item D: detection of genetically modified organisms and derived products – protein-based methods.

3.8

The future direction for methods of analysis

There is current discussion on an international basis whereby the present procedure by which specific methods of analysis are incorporated into legislation are replaced by one in which method performance characteristics are specified. This is because by specifying a single method: • the analyst is denied freedom of choice and thus may be required to use an inappropriate method in some situations • the procedure inhibits the use of automation • it is administratively difficult to change a method found to be unsatisfactory or inferior to another currently available.

As a result the use of an alternative approach whereby a defined set of criteria to which methods should comply without specifically endorsing specific methods is being considered and slowly adopted in some sectors of food analysis. This approach will have a considerable impact on the food analytical laboratory. There are a number of issues that are of concern to the food analytical community of which analysts should be aware. These are outlined briefly below.

62

Food chemical safety

3.8.1 Measurement uncertainty It is increasingly being recognised both by laboratories and the customers of laboratories that any reported analytical result is an estimate only and the ‘true value’ will lie within a range around the reported result. The extent of the range for any analytical result may be derived in a number of different ways, e.g. using the results from method validation studies or determining the inherent variation through different components within the method, i.e. estimating these variances as standard deviations and developing an overall standard deviation for the method. There is some concern within the food analytical community as to the most appropriate way to estimate this variability.

3.8.2 In-house method validation There is concern in the food analytical community that although methods should ideally be validated by a collaborative trial, this is not always feasible for economic or practical reasons. As a result, IUPAC guidelines are being developed for in-house method validation to give information to analysts on the acceptable procedure in this area. These guidelines should be finalised by the end of 2001.

3.8.3 Recovery It is possible to determine the recovery that is obtained during an analytical run. Internationally harmonised guidelines have been prepared which indicate how recovery information should be handled. This is a contentious area amongst analytical chemists because some countries of the organisations require analytical methods to be corrected for recovery, whereas others do not. Food analysts should recognise that this issue has been addressed on an international basis.

3.9 1. 2. 3. 4.

References EUROPEAN UNION,

Council Directive 89/397/EEC on the Official Control of Foodstuffs, O.J. L186 of 30.6.1989. EUROPEAN UNION, Council Directive 93/99/EEC on the Subject of Additional Measures Concerning the Official Control of Foodstuffs, O.J. L290 of 24.11.1993. EUROPEAN COMMITTEE FOR STANDARDIZATION, General Criteria for the Operation of Testing Laboratories – European Standard EN 45001, Brussels, CEN/CENELEC, 1989. ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT , Decision of the Council of the OECD of 12 Mar 1981 Concerning the Mutual Acceptance of Data in the Assessment of Chemicals, Paris, OECD, 1981.

Analytical methods: quality control and selection 5. 6. 7.

8. 9. 10. 11.

12. 13. 14. 15. 16. 17. 18. 19.

EUROPEAN COMMITTEE FOR STANDARDIZATION,

63

General Criteria for the Assessment of Testing Laboratories – European Standard EN45002, Brussels, CEN/CENELEC, 1989. EUROPEAN COMMITTEE FOR STANDARDIZATION, General Criteria for Laboratory Accreditation Bodies – European Standard EN45003, Brussels, CEN/CENELEC, 1989. EURPEAN UNION , Council Directive 85/591/EEC Concerning the Introduction of Community Methods of Sampling and Analysis for the Monitoring of Foodstuffs Intended for Human Consumption, O.J. L372 of 31.12.1985. INTERNATIONAL ORGANIZATION FOR STANDARDIZATION , General Requirements for the Competence of Calibration and Testing Laboratories ISO/IEC 17025, Geneva, ISO, 1999. CODEX ALIMENTARIUS COMMISSION, Report of the 21st Session of the Codex Committee on Methods of Analysis and Sampling – ALINORM 97/ 23A, Rome, FAO, 1997. CODEX ALIMENTARIUS COMMISSION, Report of the 22nd Session of the Codex Alimentarius Commission – ALINORM 97/37, Rome, FAO, 1997. INTERNATIONAL UNION OF PURE AND APPLIED CHEMISTRY , The International Harmonised Protocol for the Proficiency Testing of (Chemical) Analytical Laboratories, ed. Thompson M and Wood R, Pure Appl. Chem., 1993 65 2123–2144 (also published in J. AOAC International, 1993 76 926–940). INTERNATIONAL UNION OF PURE AND APPLIED CHEMISTRY, Guidelines on Internal Quality Control in Analytical Chemistry Laboratories, ed. Thompson M and Wood R, Pure Appl. Chem., 1995 67 649–666. INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, Calibration and Testing Laboratory Accreditation Systems – General Requirements for Operation and Recognition – ISO/IEC Guide 58, Geneva, ISO, 1993. HORWITZ W, ‘Protocol for the Design, Conduct and Interpretation of Method Performance Studies’, Pure Appl. Chem, 1988 60 855–864 (revision published 1995). HORWITZ W, ‘Evaluation of Analytical Methods for Regulation of Foods and Drugs’, Anal. Chem., 1982 54 67A-76A. CODEX ALIMENTARIUS COMMISSION, Procedural Manual of the Codex Alimentarius Commission – Tenth Edition, Rome, FAO, 1997. INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, Precision of Test Methods – Standard 5725, Geneva, ISO, 1981 (revised 1986 with further revision in preparation). INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, International Vocabulary for Basic and General Terms in Metrology – 2nd Edition, Geneva, ISO, 1993. INTERNATIONAL ORGANIZATION FOR STANDARDIZATION, Terms and Definitions used in Connections with Reference Materials – ISO Guide 30, Geneva, ISO, 1992.

64

Food chemical safety

Appendix: Information for potential contractors on the analytical quality assurance requirements for food chemical surveillance exercises Introduction The FSA undertakes surveillance exercises, the data for which are acquired from analytical determinations. The Agency will take measures to ensure that the analytical data produced by contractors are sufficient with respect to analytical quality, i.e. that the results obtained meet predetermined analytical quality requirements such as fitness-for-purpose, accuracy and reliability. Thus when inviting tenders FSA will ask potential contractors to provide information regarding the performance requirements of the methods to be used in the exercise, e.g. limit of detection, accuracy, precision etc., and the quality assurance measures used in their laboratories. When presenting tenders laboratories should confirm how they comply with these specifications and give the principles of the methods to be used. These requirements extend both to the laboratory as a whole and to the specific analytical determinations being required in the surveillance exercise. The requirements are described in three parts, namely: • Part A: Quality assurance requirements for surveillance projects provided by potential contractors at the time ROAMEs are completed and when commissioning a survey • Part B: Information to be defined by the FSA customer once the contract has been awarded – to be agreed with contractor • Part C: Information to be provided by the contractor on an on-going basis once contract is awarded – to be agreed with the customer.

Each of these considerations is addressed in detail below. Potential contractors are asked to provide the information requested in Part A of this document when submitting ROAME forms in order to aid the assessment of the relative merits of each project from the analytical/data quality point of view. This information is best supplied in tabular form, for example that outlined in Part A, but may be provided in another format if thought appropriate. The tables should be expanded as necessary. Parts B and C should not be completed when submitting completed ROAME forms. Explanation of Parts A, B and C of Document Part A Part A describes the information that is to be provided by potential contractors at the time that the ROAME Bs are completed for submission to the Group. Provision of this information will permit any FSA ‘Analytical Group’ and customers to make an informed assessment and comparison of the analytical quality of the results that will be obtained from the potential contractors bidding for the project. Previously potential contractors have not been given defined

Analytical methods: quality control and selection

65

guidance on the analytical quality assurance information required of them and this has made comparison between potential contractors difficult. Part A is supplied to potential contractors at the same time as further information about the project is supplied. The list has been constructed on the premise that contractors will use methods of analysis that are appropriate and accredited by a third party (normally UKAS), participate in and achieve satisfactory results in proficiency testing schemes and use formal internal quality control procedures. In addition, Parts B and C are made available to the potential contractors so that they are aware of what other demands will be made of them and can build the costs of providing the information into their bids. Part B This section defines the analytical considerations that must be addressed by both the customer and contractor before the exercise commences. Not all aspects may be relevant for all surveys, but each should be considered for relevancy. Agreement will signify a considerable understanding of both the analytical quality required and the significance of the results obtained. Part C This section outlines the information that must be provided by the contractor to a customer on an on-going basis throughout the project. The most critical aspect is the provision of Internal Quality Control (IQC) control charts thus ensuring that the customer has confidence that the contractor is in ‘analytical control’. By following the above the FSA customers will have confidence that the systems are in place in contractors with respect to analytical control and that they are being respected. It is appreciated that not all aspects outlined in Parts A, B and C will be appropriate for every contract but all should be at least considered as to their appropriateness. Contents of Parts A, B and C of document Part A Potential contractors should provide the information requested below. Please provide the information requested either in section 1 or in section 2 and then that in sections 3 to 5. Section 1: Formal quality system in the laboratory if third party assessed (i.e. if UKAS accredited or GLP compliant) Please describe the quality system in your laboratory by addressing the following aspects:

66

Food chemical safety

• To which scheme is your laboratory accredited or GLP compliant? • Please describe the scope of accreditation, by addressing: 1. the area that is accredited 2. for which matrices, and 3. for which analytes or supply a copy of your accreditation schedule. • Do you foresee any situation whereby you will lose accreditation status due to matters outside your immediate control, e.g. closure of the laboratory?

Section 2: Quality system if not accredited Please describe the quality system in your laboratory by addressing the following aspects: • Laboratory Organisation: 1. Management/supervision 2. Structure and organisational chart 3. Job descriptions if appropriate. • Staff: 1. Qualifications 2. Training records 3. Monitoring of the analytical competency of individual staff members. • Documentation 1. General lab procedures 2. Methods to be used (adequate/detailed enough to control consistent approach). • Sample Preparation 1. Location 2. Documented procedures 3. Homogenisation 4. Sub-sampling 5. Sample identification 6. Cross-contamination risk 7. Special requirements. • Equipment Calibration 1. Frequency 2. Who 3. Records 4. Marking. • Traceability 1. Who did what/when 2. Equipment – balances etc. 3. Sample storage/temperature 4. Calibration solutions: how prepared and stored. • Results/Reports 1. Calculation checks

Analytical methods: quality control and selection

67

2. Typographic checks 3. Security/confidentiality of data 4. Software usage/control 5. Job title of approved signatory. • Laboratory Information management System Please outline the system employed. • Internal Audits 1. Audit plan 2. Frequency 3. Who carries out the audit? 4. Are internal audit reports available? 5. What are the non-compliance follow-up procedures? • Sub-contracting 1. In what circumstances is sub-contracting carried out? 2. How is such sub-contracting controlled and audited?

Section 3: Proficiency testing Please describe the arrangements for external proficiency testing in your laboratory by addressing the following aspects: • Do you participate in proficiency testing schemes? If so, which schemes? • Which analyte/matrices of the above schemes do you participate in? • What are your individual proficiency scores and their classification, (e.g. zscores or equivalent), over the past two years, for the analyte/matrices of relevance to this proposal? • What remedial action do you take if you should get unsatisfactory results?

Section 4: Internal quality control Please describe the IQC measures adopted in your laboratory by addressing the following: • • • • • • • •

What control samples do you use in an analytical run? Do you follow the Harmonised Guidelines?1 What IQC procedures are in place? Do you use Certified Reference Materials (CRMs), and if so, how? For example, specify the concentration(s) matrix type(s) etc. Which appropriate CRMs do you use? Do you use In-House Reference Materials (IHRM) and how are they obtained? For example, specify the concentration(s) matrix type(s). Are they traceable? For example, to CRM, a reference method, interlaboratory comparison, or other. What criteria do you have regarding reagent blanks?

1 ‘Guidelines on Internal Quality Control in Analytical Chemistry Laboratories’, ed. M. Thompson and R. Wood, Pure Appl. Chem., 1995, 67, 649–66.

68

Food chemical safety

• • • •

What action/warning limits are applied for control charts? What action do you take if the limits are exceeded? Do you check new control materials and calibration standards? If so, how? Can we see the audit of previous results – what actions have been taken or trends observed? • Do you make use of duplicate data as an IQC procedure? • How frequently are control materials (CRMs, blanks, IHRM etc.) incorporated in the analytical run? • Do you randomise your samples in an analytical run? (including duplicates). Section 5: Method validation Please describe the characteristics of the method of analysis you propose to use in the survey by addressing the following: • What methods do you have to cover the matrix and analyte combinations required? • Do you routinely use the method? • Is the method accredited? • Has the method been validated by collaborative trial (i.e. externally)? • Has the method been validated through any In-House Protocol? • Is it a Standard (i.e. published in the literature or by a Standards Organisation) Method? • Please identify the performance characteristics of the methods, i.e. 1. LOQ 2. LOD 3. Blanks 4. Precision values over the relevant concentration range expressed as relative standard deviations 5. Bias and recovery characteristics including relevant information on traceability. • Do you estimate measurement uncertainty/reliability? • Do you normally give a measurement uncertainty/reliability when reporting results to your customer?

Part B The FSA customer is to consider and then define the following in consultation with the contractor: 1. 2. 3. 4.

What analysis is required for what matrices. The sample storage conditions to be used. Are stability checks for specific analytes undertaken? The methods to be used and a copy of Standard Operating Procedures (SOPs) where accredited, including any sampling and sample preparation protocols, to be supplied to the customer. The IQC procedures to be used. In particular the following should be considered:

Analytical methods: quality control and selection

69

• • • •

5. 6. 7.

8. 9. 10. 11. 12. 13. 14.

the use of the International Harmonised Guidelines for IQC the use of control charts randomisation within the run the composition of the analytical run (e.g. the number of control samples, and in particular the number of blanks, spikes, IHRMs etc.) • the reference materials to be used • the determination of recoveries on each batch using procedures as described in the International Harmonised Guidelines with all results to be corrected for recovery except where otherwise specified (i.e. for pesticides) and for the recovery data quoted to be reported. The measurement limits (i.e. limit of detection (LOD): limit of determination/quantification (LOQ), and reporting limits, etc.). The maximum acceptable measurement reliability (also known as measurement uncertainty) for each analytical result. The treatment of individual results with respect to uncertainty, reliability, i.e. as (a) x  y g/kg where y is the measurement reliability (i.e. as if the sample were to be a ‘historic’ surveillance result), or (b) not less/more than x g/kg where x is the analytical result determined less the measurement reliability (i.e. as if the sample were to be an ‘enforcement style’ result) when assessing compliance with a (maximum) limit. Whether there are to be action limits whereby the customer is immediately notified of ‘abnormal’ results. The procedures to be used for confirmation of ‘abnormal’ results, e.g. those that exceed any defined statutory limit. The procedures to be used if qualitative analysis is to be undertaken. The consistent way of expressing results, e.g. (a) on a wet (as is) basis, on a dry weight basis or on a fat weight basis, and (b) the reporting units for specific analytes to be used throughout survey (i.e. mg/kg etc.). The time interval for customer visits (e.g. once every three months, or as otherwise appropriate) and for submission of control charts. Whether there are any possibilities of developing integrated databases between customer and major contractors. If not, the customer to provide reporting guidelines. The procedure for logging in of samples and traceability of sample in the laboratory. The security of samples within the laboratory.

Part C The following are to be provided by the contractor on an on-going basis throughout the contract to confirm that the contractor remains in ‘analytical control’. 1. Copies of the control charts and duplicate value control charts or other agreed measures to monitor IQC.

70

Food chemical safety

2.

Records of action taken to remedy out-of-control situations to be provided at the same time with control charts. Where action limits have been identified in Part B (see para. 8), the results of samples that exceed the action limits are to be sent to the customer as soon as available. Any relevant proficiency testing scheme results obtained during the course of the survey.

3. 4.

4 Molecular imprint-based sensors in contaminant analysis P. D. Patel, Leatherhead Food Research Association

4.1

Introduction

Sample extraction and preparation remain the most time-consuming and errorprone steps in the analytical process, but these are crucial procedures because food scientists need to isolate and concentrate a wide variety of analytes from complex and varied matrices. Advances in sample extraction and preparation in chemical analysis have only in the past several years been given critical consideration as an important component in obtaining reliable and robust analytical results. This situation is also true of analyses carried out in other industrial sectors (e.g. chemical and microbiological contaminants in food, agriculture and environment). A typical analytical process for PCBs and dioxins involves extraction of the contaminants from a bulk matrix into a solvent (e.g. liquid-liquid extraction with hexane), sample clean-up to remove the bulk co-extractants and allow separation and concentration of the target analyte (e.g. chromatography using fluorisil, silica, alumina and activated carbon) and, finally, identification and quantification (e.g. by GC-electron capture detector or GC-MS). The general steps in the analytical process are equally applicable to other contaminants (e.g. pesticides, antibiotics, b-agonists and mycotoxins). The selected analytical procedures are based on the requirements that must be met for the analytes in question. This includes high sensitivity and low limits of detection (i.e. ppt/subppt levels, owing to the extreme toxicities of some of these compounds), high selectivity (i.e. distinction required between PCBs and dioxin from other coextractants and, possibly, interfering compounds present at much higher concentrations), high specificity (a differentiation of various congeners is desired (e.g. 17 2,3,7,8 substituted PCDDs) and high accuracy and precision (i.e.

72

Food chemical safety

the determination should give a valid estimate of true concentration in the sample). However, in practice, many of the classical techniques can be extremely labour-intensive (and hence costly), cumbersome, lengthy and not wholly reliable. With the current strict regulatory framework (encompassing modern QA systems and product specifications that include trace level measurements of contaminants) under which a variety of foods is produced, there is a dire need for analytical techniques that are rapid, quantitative, reliable, robust and cost-effective. Recently, solid phase extraction (SPE), based on a solid/liquid partition in polypropylene or polystyrene columns or SPE discs containing one or more sorbents, has been widely adopted, replacing many conventional methods of extraction, isolation and concentration of numerous contaminants, including antibiotics, -agonists, mycotoxins, pesticides and tainting compounds (Pimbley and Patel, 2000). Their major advantages include high speed (e.g. using SPE discs), better reproducibility than conventional adsorbents, versatility (applicable to wide-ranging analytes, including hydrophobic, hydrophilic and ionic types), low cost through mass production techniques, price competition and reduction in solvent usage. The SPE systems have been made highly specific and selective for analytes by introduction of antibody-linked sorbents (immunoaffinity). The major advantages of this system include reduction in coextraction of potential cross-reacting or interfering species, concentration of the test analyte on the sorbent, which results in increased sensitivity, and the fact that the subsequent detector need not be highly specific. The immunoaffinity systems have been widely used in selective areas of chemical and microbiological contamination, including analysis of mycotoxins (Dragacci and Fremy, 1996), b-agonists (Haines and Patel, 1998) and microbial pathogens (e.g. Salmonella, Listeria and E. coli O157; Cudjoe et al., 1993). An alternative to the immunoaffinity systems is the use of molecular imprinted polymers (MIPs). Unlike biological antibodies in the immunoaffinity systems, MIPs are based on chemical synthetic polymers, which are highly stable (i.e. do not degrade by enzymic action, processing conditions such as heat and low pH, and organic solvents), can be regenerated extensively, require lowcost reagents to prepare MIPs and can be reproducibly prepared in larger quantities (Owens et al., 1999). This chapter on MIPs addresses the area related to SPE as part of the overall analytical process and covers: (i) some basic principles of MIPs and MIP-based techniques in which MIPs are largely used as sorbents packed in chromatographic columns; (ii) examples of the usage of MIPs as physicochemical recognition elements in sensors with a number of transduction elements (e.g. optical and electrochemical); (iii) some case-study examples of the application of MIPs in contaminant analysis (e.g. antibiotics and clenbuterol), largely based on results obtained in the EC collaborative project (FAIR CT96-1219) co-ordinated by the Leatherhead Food RA Ltd; and (iv) some perspectives on future developments in this field, together with further sources of information for the readers.

Molecular imprint-based sensors in contaminant analysis 73

4.2 The principles of molecularly imprinted polymer-based techniques Molecular imprinting involves preparation of a ‘mould’ comprising polymerisable functional monomers around a print molecule (e.g. Penicillin V, see Fig. 4.1). Initially, the monomer (methacrylic acid) is allowed to establish bond formation with the print molecule and the resulting complexes or adducts are then co-polymerised with cross-linkers (e.g. ethylene glycol dimethacrylate) into a rigid polymer. The print molecule is extracted to leave specific recognition sites in the polymer that are structurally and functionally complementary to the print molecule. The resulting so-called ‘plastic antibodies’ can be functionally as effective as the biological antibodies. For details regarding the structural basis of the imprinting process, refer to Yu and Mosbach (1998). In practice, the molecular imprinting process involves the following steps. 1.

The print molecule (e.g. clenbuterol) is dissolved in a porogen (e.g. acetonitrile) together with either one or two monomers (e.g. methacrylic acid and 4-vinylpyridine). This allows non-covalent complexes to form

Fig. 4.1

Schematic representation of a molecular imprinting process.

Table 4.1

Examples of solid-phase MIP particles for contaminant analysis

Contaminant

Functional monomer

Test format

Sensitivity

Reference

Diaminonaphthalenes in solution

Acrylic acid

NS

Matsui et al., 1993

4-Nitrophenol in river water

10 ug/l

Masque´ et al., 2000

Atrazine herbicide in solution

4-Vinylpyridine (4-VP) MAA

1 ug/ml

Muldoon and Stanker, 1995

Atrazine in bovine liver

MAA

0.02 ppm in meat

Muldoon and Stanker, 1997

Atrazine in solution Atrazine in solution

MAA MAA

0.1 mM 0.25 uM

Matsui et al., 1995 Siemann et al., 1996

Triazine in solution

DEAEM

0.01 mM

Piletsky et al., 1997

Chlorotriazine in environmental samples (natural waters and sediment samples)

MAA

In-situ prepared MIPpacked HPLC column MIP-based SPE clean-up, then RP-HPLC MIP-based competitive inhibition assay (molecularly imprinted sorbent assay) MIP-based SPE clean-up, then HPLC MIP-based HPLC column MIP-based HPLC and radio-ligand assay MIP-based competitive fluorescent assay MIP-based SPE cartridge with LC diode array detection

0.05–0.2 ug/L depending on the type of pesticide

Ferrer et al., 2000

2,4-dichlorophenoxyacetic acid herbicide in solution Simazine herbicide in water

4-VP

Ametryn and other triazines herbicides in tap water Chloramphenicol (CL) in serum

NS

Penicillin V, Penicillin G and oxacillin in solution Clenbuterol in solution

MAA and 4-VP

Listeria monocytogenes cells partitioned between organic/ aqueous medium

Poly(allylamine) used in presence of diacid chloride

NS = Not specified.

MAA

Diethylamino-ethyl methacrylate (DEAEM)

MAA

MIP-based radioligand assay MIP-based SPE and HPLC MIP-based cartridge Competitive displacement of CL-methyl red dye conjugate from CLimprinted polymer packed in HPLC column MIP-packed HPLC column MIP-packed HPLC column Polymeric microcapsules (bacteria-mediated lithography)

30 ng/ml

Haupt et al., 1998

NS

Matsui et al., 1997

< 1 ug/L. However, low recovery (10–40%) 3 ug/ml

Ferrer and Barcelo´, 1999

NS

Skudar et al., 1999

NS

Crescenzi et al., 1998

NS

Aherne et al., 1996 Whitcombe et al., 1997

Levi et al., 1997

76

2.

3.

4.

Food chemical safety between the chemical functionalities of the monomers and complementary ionic groups of the print molecule. In the next step, an excess of cross-linking monomer (e.g. trimethylolpropane trimethacrylate or ethylene glycol dimethacrylate) is added together with an initiator (e.g. 2,2’-azoisobutyronitrile), which induces the polymerisation process. Under nitrogen and high temperature, the polymerisation process results in the formation of a rigid mass of polymer. The mass of polymer is then ground and wet-sieved, normally through 25 m or less mesh with acetone. The collected particles are sonicated and allowed to settle in acetone to eliminate any fines. The solvent is then evaporated, leaving behind particles of size 10–25 m. In most reported applications, MIP particles are then packed into stainless steel columns for HPLC analysis (Yoshizako et al., 1998). However, it should be noted that MIPs have also been used in other applications, including specific sorbents in SPE (Ferrer and Barcelo´, 1999; Stevenson, 1999), support for electrophoretic separations, as membranes and as synthetic models of biological antibodies in sensors and affinity assays (Owens et al., 1999; Surugiu et al., 1999; Takeuchi and Haginaka, 1999). The final step is to extract the print molecule before the MIP particles can be used as solid phase for separation and concentration studies. A range of extractants have been used in this step, e.g. methanol containing triethylamine or acetic acid are added until a steady baseline is obtained.

Molecular imprinting is a relatively new technique for generating solid phase molecular imprinted polymer (MIP)-based affinity systems, but already many applications have been demonstrated (particularly involving organically soluble low molecular weight compounds), including drugs, herbicides, sugars, nucleotides, amino acids and proteins (Bru¨ggemann et al., 2000). Table 4.1 shows examples of MIPs reported for contaminant analysis.

4.3

The development and application of MIP-based sensors

The term sensor has been defined as a device or system – including control and processing electronics, software and interconnection networks – that responds to a physical or chemical quantity to produce an output that is a measure of that quantity. A biosensor comprises two distinct elements: a biological recognition element (e.g. antibodies, enzymes, lectins, receptors and microbial cells) and, in close contact, a signal transduction element (e.g. optical, amperometric, acoustic and electrochemical) connected to a data acquisition and processing system. Thus, the signal generated as a result of analyte interaction with the biological element is converted to a quantifiable signal, e.g. optical or electrochemical. For a review of biosensor applications, the reader is referred to Patel (2000). MIPs are increasingly being used in combination with a range of transducer elements

Molecular imprint-based sensors in contaminant analysis 77 (fibre optic and electrochemical) for the rapid detection and quantification of analytes (Dickert and Hayden, 1999; Yano and Karube, 1999). Some examples of reported applications of MIP-based sensors, also referred to as biomimetic sensors, for contaminant analysis are shown in Table 4.2. Piezoelectric transducer-based MIP sensors are devices based on materials such as quartz crystals (quartz cystal microbalance, QCM) that resonate on application of an external alternating electric field. The frequency of the resulting oscillation is a function of the mass of the crystal. Thus, interaction of an analyte in a sample with the corresponding MIP, previously immobilised to a quartz crystal, will increase the overall mass that is measured as a change in the frequency of oscillation. In the surface acoustic wave (SAW) device, an acoustic wave travels over the surface of the crystal that is coated with the specific MIP. A change in the frequency will occur as the analyte in a sample binds to the corresponding MIP. These MIP-based sensors have been reported for the determination of a range of polycyclic aromatic hydrocarbons (e.g. anthracene, chrysene, pyrene, perylene and benzoperylene) and o-xylene in water (Dickert et al., 1999; Dickert and Hayden, 1999). Two types of electrochemical MIP-based sensor have been reported for the analysis of herbicides atrazine (Piletsky et al., 1995) and 2,4-dichlorophenoxyacetic acid (2,4-D; Kro¨ger et al., 1999). In the atrazine sensor, polymeric membranes containing molecular recognition sites for atrazine were prepared by a molecular imprinting process comprising radical polymerisation of diethyl aminoethyl methacrylate and ethylene glycol dimethacrylate in the presence of atrazine as template. The subsequent electrochemical measurement was carried out in a cell with two platinum electrodes separated by the imprinted membrane. A change in membrane electroresistance was measured by applying a small amplitude alternating voltage with a varying frequency generated by a low-frequency wave form generator to the electrodes, both with and without atrazine, and recorded as a function of time. Atrazine was detected in the range 0.01–0.50 mg/l with no interference from the herbicide simazine. In the 2,4-D sensor, a suspension of 2,4-D imprinted polymer or control polymer particles was air-dried onto the working electrode of the screen-printed electrodes system (electrodes were printed onto polyester sheets). The particles were immobilised using an over-layer of agarose gel. The measurement of 2,4-D was carried out by a competition assay as follows. The electrodes were placed in solutions containing the electroactive probe (homogentisic acid) and the analyte (2,4-D). The non-related electroactive probe had previously been shown to bind effectively to the analyte-imprinted particles, but not to the control particles. After incubation, the excess reactants were washed off the electrodes and the bound probe then quantified by differential-pulse voltammetry using an electrochemical analyser. The peak current generated was directly proportional to the concentration of probe bound to the electrode and inversely proportional to the analyte concentration in the sample. 2,4-D was detected in the range approximately 0.1–100 M, taking into account the background response due to the control particles.

Table 4.2

Examples of MIP-based sensors for contaminant analysis

Contaminant

Functional monomer

Test format

Sensitivity

Reference

o-xylene in solution

NS Methacrylic acid (MAA) or DEAEM 4-VP

QCM: 1 ng SAW: 1 pg 0.01 mg/L

Dickert and Hayden, 1999

Atrazine in solution

MIP-coupled QCM or SAW MIP-based conductimetric sensor MIP-based electrochemical detection on disposable screenprinted electrode MIP-based sensors (QCM, SAW and optical transducers) MIP-based electrochemical sensor

1 uM

Kro¨ger et al., 1999

30 ng/L (ppt level)

Dickert et al., 1999

0.5 ug/ml

Unpublished

2,4-dichlorophenoxy-acetic acid in solution Polycyclic aromatic hydrocarbons (PAHs) in water

Phloroglucinol and triisocyanate

Clenbuterol in solution

MAA

QCM: Quartz crystal microbalance, SAW: Surface acoustic wave.

Piletsky et al., 1995

Molecular imprint-based sensors in contaminant analysis 79

4.4

Case studies: contaminant analysis

Most of the work reported in the literature is related to academic development of MIPs and MIP-based sensors for wide-ranging analytes covering a number of major industrial sectors, including pharmaceutical, agriculture, food and environment. Application of MIPs and MIP-based sensors to real industrial matrices is beginning to increase in order to realise their true potential for commercial exploitation. In this context, MIP-based SPE techniques have been applied to several types of biological sample, including detection of atrazine in chloroform extracts of bovine liver homogenate, propranolol in dog plasma, rat bile and human urine, tamoxifen in human plasma and urine, 7hydroxycoumarine in urine, and nicotine in chewing gum (Owens et al., 1999; Sellergren, 1999). An MIP-based bulk acoustic wave (BAW) sensor has been developed for the estimation of caffeine in human serum and urine (Liang et al., 1999). In this section, two examples of recent application of MIPs to contaminant analysis in the food industry, developed as part of the EC project FAIR CT961219, are considered. 4.4.1 MIP-based extraction of clenbuterol from food The details regarding preparation of clenbuterol imprinted polymers, HPLC columns and detection have been described previously (Crescenzi et al., 1998). A typical chromatogram showing the resolution of clenbuterol and timolol from a mixture at pH values 2.0 and 3.4 is shown in Fig. 4.2. In terms of the selectivity of the stationary phase, expressed as separation factor , the values at pH 2.0 and pH 3.4 were 3.1 and 14.4, respectively. For control particles, the  value was 1. The following procedure was used to determine the potential value of the MIP for extraction of clenbuterol from foodstuffs. Each of beef and lamb kidney samples were spiked with a high level (500 g/ml) of clenbuterol. Unspiked food samples were used as negative controls. Briefly, a sample (1 g) of finely chopped food was mixed with 3 ml of the extraction buffer (acetonitrile:0.001M phosphate buffer pH 3.4, 70:30) containing 1 mg of a non-specific protease. The mixture was homogenised and allowed to incubate for 1 h at 55ºC with shaking. The mixture was cooled to 4ºC, vortexed for 30 s and centrifuged at room temperature for 15 min at 4000 rpm. A portion (2.4 ml) of the supernatant was carefully transferred into a clean vial and mixed with 0.2 ml of the protease inhibitor solution. Finally, the solution was made up to 3.2 ml using the extraction buffer, giving a final concentration of 0.25 g/ml. The extracted samples (in mobile phase: acetonitrile:0.001M phosphate buffer pH 3.4, 70:30) were injected (20 l per sample, 1 ml/min flow rate) onto the clenbuterol-MIP and blank polymer packed HPLC columns. The bound fraction was eluted in the mobile phase and the elution profile monitored at 208– 300 nm.

80

Food chemical safety

Fig. 4.2 Chromatograms of timolol (a) and clenbuterol (b) mixture on polymer imprinted against clenbuterol (EC project FAIR CT96-1219).

Figures 4.3 and 4.4 show the elution profiles of the spiked and control food extracts in the clenbuterol and blank polymer columns. With the blank polymer, the food matrix and clenbuterol were eluted from the HPLC column within 6 min of the sample injection, compared with retention of clenbuterol up to 18 min after sample injection. The retention times for the clenbuterol spiked in both beef and lamb kidney were 18 min 14 s and 18 min 48 s, respectively, indicating reproducible performance of the MIP column. 4.4.2 MIP-based extraction of -lactam antibiotics from milk The details regarding preparation of oxacillin imprinted polymers, HPLC columns and detection have been described previously (Bru¨ggemann et al.,

Molecular imprint-based sensors in contaminant analysis 81

Fig. 4.3 Clenbuterol MIP (a) and control blank MIP (b) based HPLC profiles showing resolution of clenbuterol from spiked (—) and unspiked (- - -) kidney samples (EC project FAIR CT96-1219).

82

Food chemical safety

Fig 4.4 Clenbuterol MIP (a) and control blank MIP (b) based HPLC profiles showing resolution of clenbuterol from spiked (—) and unspiked (- - -) beef samples (EC project FAIR CT96-1219).

Molecular imprint-based sensors in contaminant analysis 83 2000). A typical chromatogram showing the resolution of various -lactam antibiotics from a mixture at pH 3.5 is shown in Fig. 4.5. In terms of the selectivity of the stationary phase, the separation factors  were 1, 2.15 and 1.54 for oxacillin, penicillin G and penicillin V, respectively. The procedure for extraction of the antibiotic oxacillin using oxacillin imprinted polymer involved the following steps: (i) spiked milk samples

Fig. 4.5 Chromatogram representing print molecule (oxacillin) and two other -lactam antibiotics (penicillin G and V) on oxacillin MIP (a) compared with the control blank MIP (b). Reprinted from J. Chromatogr. A, 889, Bru¨ggemann et al. ‘New configurations and applications of molecularly imprinted polymers’, 15–24 (2000), with permission from Elsevier Science.

84

Food chemical safety

containing 0 and 50g/ml oxacillin were centrifuged at 11,000 rpm for 20 min and the pellet discarded; (ii) the supernatants were adjusted to pH 4.5 with 5N hydrochloric acid and then centrifuged as before; (iii) the pH of the supernatants were raised to pH 7.5 with 5N NaOH and methanol added (2 parts to 1 part aqueous phase); (iv) the supernatants were extracted twice with chloroform, each for 10 min under slight agitation; (v) the top aqueous/methanol phases were dried under a stream of nitrogen; (vi) the dried extracts were reconstituted in the mobile phase acetonitrile:0.001 M phosphate buffer, pH 3.5 (50:50) and then injected (20 l per sample, 0.5 ml/min flow rate) onto the HPLC column; and (vii) the bound fraction was eluted in the mobile phase and the elution profile monitored at 208–300 nm. Figure 4.6 shows the elution profiles of the unspiked milk sample (a) and milk spiked at 50 g/ml (b) using oxacillin polymer packed columns. The bulk milk was eluted from the column within 15 min, whereas the oxacillin was retained until after approximately 38 min from the time of the sample injection.

4.5

Future trends

Overall, MIPs are not as yet widely used for routine applications in analytical laboratories because of several significant limitations. These are as follows: (i) unless the template (print material) is inexpensive, preparation of MIPs can be very costly since only about 20% of the starting polymer is employed as HPLC stationary phase, owing to losses during grinding and sieving processes; (ii) robustness and sensitivity are still not adequate, owing to inefficient total removal of the print molecules during MIP preparation, which can result in gradual leaching of the template; (iii) the MIP-based sensors also have lengthy response times and are not particularly easy to prepare reproducibly; and (iv) MIPs are produced at the laboratory level and procedures now need to be developed for scaling up to commercial production. The combination of MIPs with a range of tranducers to produce sensors is expected to be a powerful realtime analytical technology. Already, the potential of this has been demonstrated as described previously, but their true value will be realised only when the basic limitations described previously are satisfactorily addressed. The majority of applications of MIPs are directed to low molecular weight, organically soluble analytes of interest in the pharmaceutical and medicare industries (e.g. chiral drug separation or controlled release of drug, Allender et al., 2000). Applications of MIPs in other industrial sectors will increase, e.g. agriculture, environment and food. In addition, several newer approaches to MIP production (e.g. surface imprinting) are expected to allow MIPs to be prepared in aqueous media for aqueous-soluble analytes, and extend the scope to macromolecules (e.g. proteins, DNA, polysaccharides) and particles (e.g. microorganisms). An interesting new approach to generating synthetic materials capable of specific recognition of macromolecules (e.g. proteins) for use in similar areas of application to conventional MIPs has been reported by Shi and

Molecular imprint-based sensors in contaminant analysis 85

Fig. 4.6 Oxacillin MIP-based HPLC profiles showing resolution of oxacillin from spiked (a) and unspiked (b) milk samples (EC project FAIR CT96-1219).

Ratner (2000). In that study, protein-binding nanocavities were prepared on a polysaccharide-like surface using a novel radiofrequency plasma deposition of thin films. The proteins included albumin, immunoglobulin, fibrinogen, lysozyme and alpha-lactalbulin. Protein-specific nanometer-sized ‘pits’ were clearly demonstrated using techniques such as electron microscopy, mass spectrometry and radioligand binding. It was observed, however, that the structurally unstable protein alpha-lactalbumin exhibited weaker template recognition than the ‘robust’ proteins (e.g. lysozyme). Some other future developments in this field are described below.

86

Food chemical safety

4.5.1 Magnetic MIPs Immunomagnetic particles have been widely used in molecular biology and an increasing number of products are becoming available for food and clinical microbiologists (Cudjoe et al., 1993). Recently, magnetic MIPs have been developed for the beta-blocker (S)-propranolol in which the MIPs retained high binding capability to the print molecule (S)-propranolol compared with control non-imprinted polymer whilst exhibiting superparamagnetic properties (Ansell and Mosbach, 1998). 4.5.2 Catalytic MIPs (also referred to as enzyme mimics or plastizyme) Since MIPs are highly stable and can be sterilised, they are valuable for use in biotransformation processes (Ramstro¨m and Mosbach, 1999). The application of MIP in catalytic reaction has been demonstrated with reference to the enzymic condensation of Z-L-aspartic acid with L-phenylalanine methyl ester to give Zaspartame (Ye et al., 1999). In this study, when the product-imprinted polymer was present, a considerable increase (40%) in product yield was found. 4.5.3 MIP-based sensors and molecularly imprinted sorbent assays These types of MIP-based assay allow direct analyte recognition and estimation from a test sample. To date, several applications of non-radioligand-based competitive assays and MIP-based sensors for contaminant analysis have been reported using MIPs as alternatives to antibodies (Tables 4.1 and 4.2). Already, several other transducers have been combined with MIPs for sensor development, including a surface plasmon resonance sensor for theophylline, caffeine and xanthan (Lai et al., 1998). However, most of these applications have been restricted to assay development using standard solutions and determination of assay characteristics (e.g. sensitivity, selectivity, specificity and cross-reactivity). To be of commercial value, application of the MIP-based sensors and ligand binding assays will need to be demonstrated in real matrices (e.g. foodstuffs, agricultural products, potable and river waters, serum and urine). In addition, the MIP-based systems would have to be at least equivalent to the modern analytical techniques (e.g. HPLC, GC-MS and ELISAs) in terms of performance characteristics and cost of analysis.

4.6

Sources of further information and advice

There is a vast amount of published information available for anyone interested in molecular imprinting, from historical perspective to current and future research and development in the field. This is complemented with proactive workshops, conferences and discussions groups around the world. Much of the information is also available on the Internet, together with other electronic products (e.g. CD-ROMs). In this context, further information can be obtained in

Molecular imprint-based sensors in contaminant analysis 87 the form of reviews (Anon., 1999; Liu et al., 1999; Ramstro¨m and Mosbach, 1999; Mosbach and Haupt, 1998) and books (Reid et al., 1998; Sellergren, 2000). The proceedings of a recent MIP 2000 workshop (first international workshop on molecularly imprinted polymers), held at the Cardiff University on 3–5 July 2000, will be published in a special issue of Analytica Chimica Acta. This will be a valuable resource to find out the current state of the art in the field, major international groups working in the area and developments for real-world use. A range of research and interest groups also exist which proactively encourage and promote molecularly imprinted polymer and related fields internationally. The Molecularly Imprinted Materials for Integrated Chemical Sensors (MIMICS) is a project that is funded by the European Community (Brite-Euram BE-95-1745), with eight pan-European organisations. Details can be found at the following Web site: http://inn7201.casaccia.enea.it/index.html. The site also provides links to major international groups working in the molecular imprinting field (e.g. Lund University, Cambridge University, University of California and University of Kalmar) as well as many references related to the MIP technology. The Society for Molecular Imprinting (c/o Pure and Applied Biochemistry, Lund University, P.O. Box 124, SE-221 00, Lund, Sweden; http:// www.ng.hik.se/~SMI) provides the latest in upcoming international conferences, workshops and venues on molecular imprinting and also new articles written in the field, current job postings, international members and addresses. The LFRA Ltd is an international independent membership-based organisation providing research in all aspects of food science and technology, and information including food legislation and trends in marketing world-wide. Further information on the activities and membership details can be found on our Web sites (www.lfra.co.uk and www.foodindustryweb.com). The author’s multidisciplinary team has considerable expertise in innovative analytical technologies covering many areas of food chemistry, food microbiology, immunology, sensor technology and miniaturisation. For full details and published references, the reader is referred to the following Web site: www.foodconsulting-lfra.com.

4.7

Acknowledgements

All technical work reported on the development of MIPs for clenbuterol and lactam antibiotics was carried out on a recent European-funded project (FAIR CT96-1219), involving partners from University of Lund, University of Rome, Merck Eurolab (Prolabo), UNIR Association and Leatherhead Food Research Association Ltd (LFRA, Co-ordinator). The technical work on application of clenbuterol and -lactam-MIPs to foods was carried out by Mr John Haines and Miss Francesca Aulanta of LFRA Ltd.

88

4.8

Food chemical safety

References

and VULFSON E N (1996), ‘Bacteria-mediated lithography of polymer surfaces’, J Am Chem Soc, 118, 8771–8772. ALLENDER C J, RICHARDSON C, WOODHOUSE B, HEARD C M and BRAIN K R (2000), ‘Pharmaceutical applications for molecularly imprinted polymers’, Int J Pharm, 195, 39–43. ANON. (1999), ‘Several review articles on MIP technology’, Trends Anal Chem, 18, 137–204. ANSELL R J and MOSBACH K (1998), ‘Magnetic molecularly imprinted polymer beads for drug radioligand binding assay’, Analyst, 123, 1611–1616. ¨ GGEMANN O, HAUPT K, LEI Y, YILMAZ E and MOSBACH K (2000), ‘New BRU configurations and applications of molecularly imprinted polymers, Review’, J Chromatogr A, 889, 15–24. CRESCENZI V, MASCI G, FONSI, M and CASATI G (1998), ‘Molecularly imprinted polymers binding clenbuterol’, Presentation at the American Chemical Society meeting, held in Boston, USA (August 1998). CUDJOE K S, PATEL P D, OLSEN E, SKJERVE E and OLSVIK Ø (1993), ‘Immunomagnetic separation techniques for the detection of pathogenic bacteria in foods’, in Kroll R G, Gilmour A and Sussman M, New Techniques in Food and Beverage Microbiology, Oxford, Blackwell Scientific Publications, 17–31. DICKERT F L and HAYDEN O (1999), ‘Molecular imprinting in chemical sensing’, Trends Anal Chem, 18, 192–199. DICKERT F L, TORTSHANOFF M, BULST W E and FISCHERAUER G (1999), ‘Molecularly imprinted sensor layers for the detection of polycyclic aromatic hydrocarbons in water’, Anal Chem, 71, 4559–4563. DRAGACCI S and FREMY J M (1996), ‘Application of immunoaffinity column cleanup to aflatoxin M1’, J Food Protect, 59, 1011–1013. ´ D (1999), ‘Validation of new solid-phase extraction FERRER I and BARCELO materials for the selective enrichment of organic contaminants from environmental samples’ Trends Anal Chem, 18, 180–192. FERRER I, LANZA F, TOLOKAN A, HORVATH V, SELLERGREN B, HORVAI G and ´ D (2000), ‘Selective trace enrichment of chlorotriazine pesticides BARCELO from natural waters and sediment samples using terbuthylazine molecularly imprinted polymers’, Anal Chem, 72, 3934–3941. HAINES J H and PATEL P D (1998), ‘A brief evaluation of the performance of three commercial ELISAs for the analysis of -agonists in foods’, Leatherhead Food RA Scientific and Tech. Notes No. 124. HAUPT K, DZGOEV A and MASBACH K (1998), ‘Assay system for the herbicide 2, 4–dichlorophenoxyacetic acid using a molecularly imprinted polymer as artificial recognition element’, Anal Chem, 70, 628–631. ¨ GER S, TURNER A P, MOSBACH K and HAUPT K (1999), ‘Imprinted polymerKRO based sensor system for herbicides using differential-pulse voltametry on screen-printed electrodes’, Anal Chem, 71, 3698–3702. AHERNE A, ALEXANDER C, PAYNE M J, PEREZ N

Molecular imprint-based sensors in contaminant analysis 89 and POLSKY B (1998), ‘Surface plasmon resonance sensors using molecularly imprinted polymers for sorbent assay theophylline, caffeine and xanthan’, Can J Chem, 76, 265. LEVI R, MCNIVEN S, PILETSKY S, CHEONG S H, YANO K and KARUBE I (1997), ‘Optical detection of chloramphenicol using molecularly imprinted polymers’, Anal Chem, 69, 2017–2021. LIANG C, PENG H, BAO X, NIE L and YAO S (1999), ‘Study of a molecular imprinting polymer coated BAW bio-mimic sensor and its application to the determination of caffeine in human serum and urine’, Analyst, 124, 1781–1785. LIU Q, ZHOU Y X and LIU Y T (1999), ‘Recent progresses in research on molecular imprinting sensors’, Chin J Anal Chem, 27, 1341–1347. MASQUE´ N, MARCE´ R M, BORULL F, CORMACK P A G and SHERRINGTON D C (2000), ‘Synthesis and evaluation of a molecularly imprinted polymer for selective on-line solid-phase extraction of 4-nitrophenol from environmental water’, Anal Chem, 72, 4122–4126. MATSUI J, KATO T, TAKEUCHI T, SUZUKI M, YOKOYAMA K, TAMIYA E and KARUBE I (1993), ‘Molecular recognition in continuous polymer rods prepared by a molecular imprinting technique’, Anal Chem, 65, 2223–2224. MATSUI J, MIYOSHI Y, DOBLHOFF-DIER O and TAKEUCHI T (1995), ‘A molecularly imprinted synthetic polymer receptor selective for atrazine’, Anal Chem, 67, 4404–4408. MATSUI M, OKADA M, TSURUOKA T and TAKEUCHI T (1997), ‘Solid-phase extraction of a triazine herbicide using a molecularly imprinted synthetic receptor’, Anal Commun, 34, 85–89. MOSBACH K and HAUPT K (1998), ‘Some new developments and challenges in non-covalent molecular imprinting technology’, J Mol Recognit, 11, 62– 68. MULDOON M T and STANKER L H (1995), ‘Polymer synthesis and characterisation of a molecularly imprinted sorbent assay for atrazine’, J Agric Food Chem, 43, 1424–1427. MULDOON M T and STANKER L H (1997), ‘Molecularly imprinted solid phase extraction of atrazine from beef liver extracts’, Anal Chem, 69, 803. OWENS P K, KARLSSON L, LUTZ E S M and ANDERSSON L I (1999), ‘Molecular imprinting for bio- and pharmaceutical analysis’, Trends Anal Chem, 18, 146–154. PATEL P D (2000), ‘(Bio)sensors for measurement of analytes implicated in food safety: A review’, Leatherhead Food RA Scientific and Tech. Notes No. 125. PILETSKY S A, PILETSKAYA E V, ELGERSMA A V, YANO K and KARUBE I (1995), ‘Atrazine sensing by molecularly imprinted membranes’, Biosens Bioelectro, 10, 959–964. PILETSKY S A, PILETSKAYA E V, EL’SKAYA A V, LEVI R, YANO K and KARUBE I (1997), ‘Optical detection system for triazine based on molecularlyimprinted polymers’, Anal Lett, 30, 445–455. LAI E P C, FARFARA A, VANDERNOOT V A, KONO M

90

Food chemical safety

and PATEL P D (2000), ‘Agrifood applications of solid-phase extraction: A review’, Leatherhead Food RA Scientific and Tech. Notes No. 197. ¨ M O and MOSBACH K (1999), ‘Synthesis and catalysis by molecularly RAMSTRO imprinted materials’, Curr Opin Chem Biol, 3, 759–764. REID E, HILL H M and WILSON I D (1998), Drug development assay approaches, including molecular imprinting and biomarkers – SP 226. London, Royal Society of Chemistry. SELLERGREN B (1999), ‘Polymer- and template-related factors influencing the efficiency in molecularly imprinted solid-phase extractions’, Trends Anal Chem, 18, 164–175. SELLERGREN B (2000), Molecularly imprinted polymers. Man-made mimics of antibodies and their application in analytical chemistry. London, Elsevier Science. SHI H and RATNER B D (2000), ‘Template recognition of protein-imprinted polymer surfaces’, J Biomed Mater Res, 49, 1–11. SIEMANN M, ANDERSSON L I and MOSBACH K (1996), ‘Selective recognition of the herbicide atrazine by noncovalent molecularly imprinted polymers’, J Agric Food Chem, 44, 141–145. ¨ M O (1999), ¨ GGEMANN O, WITTLESBERGER A and RAMSTRO SKUDAR K, BRU ‘Selective recognition and separation of -lactam antibiotics using molecularly imprinted polymers’, Anal Commun, 36, 327–332. STEVENSON D (1999), ‘Molecular imprinted polymers for solid-phase extraction’, Trends Anal Chem, 18, 154–159. SURUGIU I, YE L, YILMAZ E, DZGOEV A, DANIELSSON B, MOSBACH K and HAUPT K (1999), ‘An enzyme-linked molecularly imprinted sorbent assay’, Analyst, 125, 13–16. TAKEUCHI T and HAGINAKA J (1999), ‘Separation and sensing based on molecular recognition using molecularly imprinted polymers’, J Chromatogr B Biomed Sci Appl, 728, 1–20. WHITCOMBE M J, ALEXANDER C and VULFSON E N (1997), ‘Smart polymers for the food industry’, Trends Food Sci and Technol, 8, 140–145. YANO K and KARUBE I (1999), ‘Molecularly imprinted polymers for biosensor applications’, Trends Anal Chem, 18, 199–204. ˚ NSSON M O and MOSBACH K (1999), ‘Use of ¨ M O, ANSELL R J, MA YE L, RAMSTRO molecularly imprinted polymers in a biotransformation process’, Biotechnol Bioeng, 64, 650–655. YOSHIZAKO K, HOSOYA K, IWAKOSHI Y, KIMATA K and TANAKA N (1998), ‘Porogen imprinting effects’, Anal Chem, 70, 386. YU C and MOSBACH K (1998), ‘Insights into the origins of binding and the recognition properties of molecularly imprinted polymers prepared using an amide as the hydrogen bonding functional group’, J Mol Recognit, 11, 69–74. PIMBLEY D

5 Bioassays in contaminant analysis L. A. P. Hoogenboom, State Institute for Quality Control of Agricultural Products (RIKILT), Wageningen

5.1

Introduction

Since ancient history, humankind has relied on bioassays to determine the safety of food and environment. In medieval times, food tasters were employed to ensure that food was free of poisons. Miners used small birds to detect the possible presence of toxic gases in mining tunnels. With increasing knowledge about the responsible toxicants, improvements in analytical chemistry, combined with the need to reduce animal experiments, we now rely on chemical methods aimed at the detection of compounds by their physicochemical properties. The use of animal bioassays is more or less restricted to the testing of the safety of specific substances, thereby supported by in vitro models with mammalian and prokaryotic cells. However, even today, bioassays with mice and rats are still the only reliable way to detect paralytic and diuretic poisons in shellfish,1,2 and the neurotoxins produced by Clostridium botulinum.3 Fish assays are widely used for testing the quality of drinking water. However, despite the rapid improvements in analytical chemistry, at the same time we start to realize that these methods may no longer be sufficient to deal with the often very complex mixtures of chemicals or ever changing chemical structures of toxicants present as residues in our food chain. Furthermore, there is a strong need for rapid screening assays that can be used for extensive monitoring programmes. Bioassays with pro- or eukaryotic cells capable of detecting compounds based on their effects, offer a possible solution. For the detection of antibiotics in milk and meat, a number of different tests are used for the screening4,5 and in many cases, chemical identification of the responsible substances is no longer required. Recent advances in cell biology and in particular biotechnology have

92

Food chemical safety

allowed the development of a new generation of bioassays, based on the possibilities to introduce specific properties and reporter genes into stable cellular systems. This chapter will describe this new generation of bioassays and demonstrate their advantages, especially when used in combination with sensitive analytical methods. This will be demonstrated by the experiences obtained with the so-called DR-CALUX assay, a bioassay used for the detection of dioxins. The inclusion of bioassays in modern test strategies will allow rapid screening and detection of new, possibly unknown, agonists and help to evaluate the possible health hazards involved with the presence of such compounds in the food chain.

5.2

Dioxins and the DR-CALUX bioassay

5.2.1 Development of the assay After the discovery of dioxins in the food chain, it became clear that it would be impossible to set up large monitoring programmes for this group of compounds. The major reason for this was the very expensive and laborious analytical procedure required to detect 17 different 2,3,7,8-chlorinated dibenzo-p-dioxins (PCDDs) or furans (PCDFs) at the pg/g level. This can only be achieved after extensive clean-up and by using a high resolution mass spectrometer. A set of so-called toxic equivalency factors (TEFs), ranging from 1 to 0.0001, has been developed in order to express the concentrations of each of the congeners into one figure, which represents the group as if it was only the most toxic congener 2,3,7,8-tetrachloro-dibenzo-p-dioxin (TCDD).6 A typical limit for food is the maximum residue limit of 6 pg i-TEQ/g fat for dairy products in the Netherlands. Even lower limits or action levels are used in several other countries. Levels like this are required to meet the tolerable daily intake of 1–4 pg per kg body weight as set by the WHO. It has now become clear that many other substances like the planar non-ortho and mono-ortho PCBs and most likely also some of the brominated polyaromatic hydrocarbons should be included in these limits. New analytical methods are required to include these compounds in the analysis. In response to the limited analytical capacity, bioassays with mammalian cells have been developed, initially based on the known effects of these compounds. Receptor assays have been developed based on the binding of dioxins to a specific receptor (Ah-receptor) in the cells. The so-called ERODassay measures the deethylation of ethoxy-resorufin by certain cytochrome P450 enzymes, following the binding of dioxins to the cytosolic Ah-receptor, the binding of the complex to specific sites (DREs) in the DNA and the increased transcription of the gene encoding for the enzyme. Assays based on this principle are used for determining the levels of dioxin-like compounds in environmental samples like sewage sludge7 and sediments.8 A major drawback of this system is the possible inhibition of the enzyme by many different compounds, including natural occurring substances. The specificity of the test

Bioassays in contaminant analysis 93

Fig. 5.1 Principle behind the CALUX bioassay for Ah-receptor agonists. Following binding of the agonist to the Ah-receptor, the complex will be transported to the nucleus and bind to a so-called dioxin responsive element, resulting in the increased transcription of the luciferase gene and production of luciferase. Following incubation this enzyme can subsequently be measured in cell lysates by a light producing reaction.

was therefore tremendously increased by the development of a cell-line which contains the reporter gene luciferase under control of a murine DRE.9, 10, 11 In response to dioxins, this H4IIE rat hepatoma cell-line will synthesize luciferase in a dose-dependent way, which can subsequently be quantified by an enzymatic light producing reaction (Fig. 5.1). Figure 5.2 presents a typical dose-response curve, showing an increased luciferase production at concentrations as low as 0.5 pM. Since the test can be performed in 96 well-plates, a response is obtained with less than 50 fg TCDD. In principle, the amount of dioxins can be quantified by comparison of the response in the test with the calibration curve for TCDD. Several other dioxin and PCB congeners have been tested and were shown to give a response that reflects the differences in the TEF values (Fig. 5.3). However, congeners with a low TEF value showed a relatively low response in the test. This is similary true for 1,2,3,7,8-PeCDD which TEF value was recently adjusted from 0.5 to 1, and which is often a relatively important contributor to the total dioxin content. As a result the test may underestimate the total TEQ content, if calculations were based on the calibration curve for TCDD. However, in general it is evident that the bioassay obeys the TEQ principle and that the result will reflect the total TEQ content of the sample. 5.2.2 Validation for milk fat Following the succesful development of the cells, a rapid clean-up procedure for fat samples was developed, based on the use of an acid silica column. Using

94

Food chemical safety

Fig. 5.2 Dose-response curve for 2,3,7,8-tetrachloro-p-dibenzodioxin (TCDD) in the CALUX bioassay. The concentration (expressed as TCDD) can subsequently be determined by comparing the response obtained with a sample extract with the calibration curve.

Fig. 5.3 Comparison of the relative response of a number of dioxins and non-ortho (#126, 169) and mono-ortho (#105, 118 and 156) PCBs in the CALUX bioassay and the TEF values established by the WHO.6 Compounds were selected based on their relative importance (contribution to total TEQ levels) in food samples. A major difference between the WHO-TEF values and the i-TEF values used in previous studies is the increase of the TEF for 1,2,3,7,8-penta-PCDD from 0.5 to 1, which is not supported by the CALUX assay.

Bioassays in contaminant analysis 95 Table 5.1 Reproducibility of the CALUX assay with milk fat samples. Spiked samples were tested singly in three independent test series (adapted from 12) Sample number

GC/MS CALUX determined dioxin content* determined (pg/g) level series 1 series 2 series 3 Mean (pg i±SD TEQ/g)

CV (%)

(%)

1 2 3 4 5 6

1 3 6 9 12 15

97 4 54 10 27 11

80 110 75 83 78 87

0.0 3.5 6.9 7.1 12.3 14.6

1.0 3.2 4.6 7.2 8.1 11.8

1.3 3.2 2.1 8.3 7.8 12.8

0.8±0.7 3.3±0.2 4.5±2.4 7.5±0.7 9.4±2.5 13.1±1.4

Recovery

*

CALUX determined levels were corrected for the blank sample being respectively 6.3, 2.0 and 3.5 pg/g fat for series 1, 2 and 3 respectively. In addition values were corrected for the difference between relative responses in the CALUX assay and i-TEF values.

dimethylsulphoxide as intermediate, the extracted dioxins are transfered to the tissue culture medium and subsequently added to the cells. After exposure for 24 hours the luciferase concentration in the cells is determined. The test was validated for milk fat using a number of samples spiked at 1 to 15 pg i-TEQ/g fat (1.2–17.5 pg WHO-TEQ/g) with a mix containing the 17 congeners at equal amounts.12 Table 5.1 shows the reproducibility obtained in three independent tests with these samples. Concentrations were calculated based on the TCDD calibration curve, and subsequently corrected for the 15% difference between CALUX and i-TEF values. The calculated limit of detection was around 1 pg iTEQ/g fat, explaining the high variation obtained with the lowest sample. When calculated in i-TEFs, the recovery varied between 70% and 103%. These results demonstrate the suitability of the test to screen milk fat samples. Another important conclusion from these studies was the need to include reference samples with levels around 0 and the residue limit, in order to control for possible impurities introduced with the chemicals, recovery losses and differences in TEF values. Based on this approach, dioxin-like compounds were measured in oil obtained from a large number of different fish and shellfish products. Levels up to 100 pg i-TEQ were detected but based on general agreements these should be corrected for the sometimes very low oil levels in, for example, shellfish. As shown in Fig. 5.4, a good correlation was obtained with the combined dioxin and non-ortho PCB contents in these oils, although in a few cases relatively large differences were observed. Although this might be caused by high levels of mono-ortho PCBs (not included in GC/MS measurements), it cannot be excluded that other compounds are responsible for this effect.

96

Food chemical safety

Fig. 5.4 Comparison of CALUX determined dioxin levels and combined GC/MS determined levels of dioxins and non-ortho PCBs (77, 126 and 169) in fish and shellfish oil. Since oil levels vary widely in these samples, dioxin levels are normally expressed on a pg/g product base.

5.2.3 Citrus pulp incident Following the succesful validation of the test for milk fat, the bioassay was first used in the food and feed area during the Brazilian citrus pulp incident. Increasing milk levels in German cows were traced back to the use of citrus pulp that had been mixed with contaminated lime. Pulp samples of 5 g were extracted and cleaned by the same procedure as used for the milk fat. A rapid comparison between CALUX and GC/MS data showed that the assay was capable of selecting the highly contaminated samples, using a cut-off value of 5000 pg iTEQ/kg. Most samples contained levels higher than this limit and required GC/ MS confirmation. At the end of the crisis the limit was officially set at 500 pg iTEQ/kg, based on the detection limit of the GC/MS method. The test procedure was subsequently optimized and validated. Based on the consideration that an increased response is not necessarily caused by dioxins or dioxin-like PCBs, and that samples with an increased response would still have to be confirmed by GC/ MS, it was decided to switch to a screening approach. This approach is based on the comparison of the response obtained with test sample with that of a reference sample, containing 400 pg i-TEQ/kg. Table 5.2 shows the results obtained with 71 citrus pulp samples containing GC/MS determined levels between 100 g/l. All wines sampled directly from the bottle contained < 250 g/l.12 The contribution from deposition of airborne lead on soil and crops to lead in diets is estimated to be between 13% and 31% for children. For individual plants a high percentage of lead may derive from aerial deposition (40–100%). Where crops are contaminated by lead from the air and soil, much of this may be removed by washing and other normal culinary practices. 7.2.2 Cadmium The use of cadmium by man is relatively recent and it is only with its increasing technological use in the last few decades that serious consideration has been given to cadmium as a possible food contaminant. Cadmium is naturally present in the environment: in soils, sediments and even in unpolluted seawater. It is closely related chemically to zinc and is found wherever zinc is also found. Thus most commercially available zinc compounds will contain cadmium at low levels. Cadmium and its compounds have been used widely in industry with concomitant environmental pollution. Cadmium is emitted to air by mines, metal smelters and industries using cadmium compounds for alloys, batteries, pigments and in plastics, although many countries have stringent controls in place on such emissions. Within the European Union, as part of a Community Action Programme to combat environmental pollution by cadmium, Directive 89/677/EEC came into force in 1992.13 This banned the use of cadmium in pigments, stabilisers and plating, except in cases where no suitable alternative is available or where cadmium is used for safety reasons such as in the nuclear industry. The results of studies on animals show that cadmium is an extremely toxic metal. Cadmium is poorly excreted by the human body and although only 5– 10% of that ingested is absorbed, it does accumulate in the body over time with renal damage being caused by long-term exposure.14 One sign of this damage is proteinuria (the appearance of increased levels of unaltered proteins in the

Inorganic contaminants in food

153

urine). Cadmium and its compounds can also give rise to carcinogenic effects in animals, although the evidence for similar effects in humans is not conclusive. As cadmium is ubiquitous in the environment, all food is exposed to and contains cadmium. The principal route of absorption for non-industrially exposed populations is from food, although cigarette smoking can be an important source. The main sources of cadmium contamination of the UK food supply are phosphate fertilisers, atmospheric deposition and sewage sludge. There are currently no UK limits for cadmium in food. However, the existing recommendation of the UK Food Advisory Committee is that food containing levels of cadmium not acceptable in its country of origin should not be admitted to the UK. Cadmium is present at low concentrations in most foods, with those that are consumed in larger quantities making the largest contribution to the population dietary exposure. Although the concentrations of cadmium in food in the UK are generally low, some foods of minor dietary importance such as shellfish or kidney often contain levels in excess of 0.5 mg/kg. Plant-based foods rarely contain more than 0.2 mg/kg on a fresh weight basis, although some root crops such as carrots and parsnip, and some leafy crops, such as spinach and lettuce, tend to contain more cadmium than other plant foods. This is also true of cereals, which indicates that plants tend to take up cadmium from the soil, unlike lead. Most fish taken around the UK coast contain little cadmium, with the average being less than 0.2 mg/kg. The results of a routine surveillance exercise are given in Table 7.2.15 The 144 samples analysed were collected between 1995 and 1997. It was found that marine fish had concentrations of less than the LOD of 0.01 mg/kg, while shellfish contained higher concentrations than do most other foods, but, with the exception of lobster, whelks and crabs, shellfish from unpolluted waters rarely have an average cadmium concentration greater than 1 mg/kg. However, the body meat of crabs (brown meat) may often contain higher concentrations. The level of cadmium in animal tissues other than offal is usually low, average concentrations being less than 0.05 mg/kg. Individual samples of kidneys may contain levels higher than 0.5 mg/kg. Animals grazing on land contaminated with cadmium will often have meat with normal levels of cadmium, while the level in offal would usually be significantly higher.16 Table 7.3 presents the results from the TDS for cadmium between 1976 and 1997. The results show that for the average person in the UK, the dietary exposure to cadmium has remained roughly the same over this period at approximately 0.018–0.02 mg/day, but with a slight downward trend.10 These are upper-bound estimates, that is in the estimation of dietary intake it is assumed that levels in food that are less than the limit of detection are equal to the limit of detection. These estimates therefore overestimate the actual intake to some extent. For mean and 97.5th percentile adult consumers, the dietary exposures from the 1997 UK TDS were 0.014 mg/day and 0.024 mg/day respectively. These exposures to cadmium from food and other sources (e.g. air, water and smoking)

154

Food chemical safety Table 7.2 Cadmium in fish and shellfish (mg/ kg fresh weight of edible portion) Fish

Mean concentration

Haddock Cod Herring Mackerel Plaice Whiting Red fish Cod fish fingers Brown shrimps Cockles Crab Lobster Mussels Pink shrimps Queen scallops Squid Scallops Scampi Winkles

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.04 0.07 1.5 0.12 0.19 0.05 0.17 0.01 0.54 0.06 0.19

Note: Data from reference 15.

can be compared with the JECFA PTWI of 0.007 mg/kg bodyweight, which is equivalent to 0.06 mg/day for a 60 kg person. Although the margin of safety between the PTWI and the dietary exposure is smaller than with most other metals, it is reassuring that the trend is for lower exposures. The most recent data for dietary exposures to cadmium are similar to those found elsewhere. For example, a study carried out by the European Union in 1995 and 1996 found that dietary exposures in the 15 member States of the Union ranged from 0.007 mg/ day to 0.057 mg/day.17 Other comparable dietary exposures are 0.015 mg/day for the USA,9 0.024 mg/day for Canada,7 and 0.028 mg/day for New Zealand.18 However, the results of duplicate diet studies have shown that there are localised high intakes by consumers in certain areas or by consumers of certain foods. In the old mining village of Shipham in Somerset, UK, where the cadmium levels in some vegetable samples were more than 1 mg/kg, the dietary exposure to cadmium of the study population was about double the average weekly dietary intake and some individuals exceeded the PTWI for cadmium.19 Table 7.3 day)

Population dietary exposures to cadmium from UK Total Diet Studies (mg/

1976

1979

1982

1985

1988

1991

1994

1997

0.02

0.017

0.018

0.018

0.019

0.018

0.014

0.012

Inorganic contaminants in food

155

There was no evidence that any of the residents of the village had suffered adverse health effects related to cadmium. 7.2.3 Arsenic Arsenic is a metalloid, but it is generally included in work on metals in food. It is rarely found as the free element in the natural environment, but more commonly as a component of sulphur-containing ores in which it occurs as metal arsenides. Arsenic is present in rocks, soils, water and living organisms in concentrations in the parts per million (mg/kg) range. Interest in the contamination of food with arsenic in the UK arose from a serious outbreak of arsenical poisoning in northern England in 1900 owing to beer made with glucose. It was established that the outbreak arose chiefly as a result of the use of arsenical pyrites to manufacture sulphuric acid used in production of the glucose. A Royal Commission set up in 1903 to consider the case concluded that maximum levels of arsenic in food and liquids should be established. In the UK, the general limit for food is 1 mg/kg, and 0.1 mg/kg for ready-to-drink beverages.20 However, naturally present arsenic in fish and edible seaweed or products containing fish or edible seaweed is specifically excluded, as are hops or hop concentrates intended for commercial brewing and any food with an arsenic level limited by other regulations. Arsenic is present in food in different forms (species) which vary in toxicity, with inorganic forms considered to be the most toxic. This is reflected in the JECFA PTWI of 0.015 mg/kg bodyweight (equivalent to 0.12 mg/day for a 60 kg adult), which applies to inorganic arsenic only.21 Most arsenic in the diet is present in the less toxic organic forms. However, it can be difficult to distinguish analytically between the different forms of arsenic and for this reason most surveys have measured total arsenic. The weekly dietary exposure to arsenic can be calculated from an analysis of the various food groups of the TDS and the most recent data cover 1976–97.10 The major source of arsenic in the diet is fish which, in 1997, accounted for 94% of the total dietary exposure. An important consideration when assessing the significance of these dietary exposures is the finding that arsenic is almost entirely present in fish as the organic chemical arsenobetaine, which has been found not to be metabolised in man.22, 23 Fish are known to accumulate arsenic and a study carried out in 1998 found appreciable quantities of total arsenic in all samples analysed.15 The mean concentration of arsenic in samples of fresh marine fish landed in UK ports in 1995–1997 ranged between 1.9 mg/kg and 8.4 mg/kg. An earlier survey found that fish that live on or close to the sea bed, such as plaice, dabs, flounders and skate, tend to have higher levels of arsenic than other fish,24 and this was confirmed by the later work, where the highest level was found in plaice. Arsenic levels in shellfish show more variation, ranging from 1.3 mg/kg to 30 mg/kg. High levels are frequently found in crab, in which the white meat generally contains more arsenic than the brown meat. Lobsters contained similar levels of arsenic to crabs, with the highest levels found in pink shrimps.

156

Food chemical safety

Table 7.4

Population dietary exposures to arsenic from UK Total Diet Studies (mg/day)

1976

1979

1982

1985

1988

1991

1994

1997

0.075

n.d.

0.09

n.d.

n.d.

0.07

0.063

0.065

Note: n.d. ˆ not determined

The total arsenic, that is inorganic plus organic, content of seaweed-based dietary supplements has been determined.25 Samples of kelp powder and tablets were analysed in 1987 for both total arsenic and inorganic arsenic (unstable organic species were also included). Levels from 7 to 45 mg/kg of total arsenic were found, with an outlying value of 120 mg/kg. The inorganic arsenic content was between 0.01 to 0.45 mg/kg, with the same outlying sample containing 50 mg/kg. A more recent study on a range of dietary supplements carried out between 1995 to 1998 found total arsenic levels ranging from below the LOD of 0.005 mg/kg to 7.3 mg/kg in a sample of a supplement produced from greenlipped mussels.26 It can be seen from Table 7.4 that the population dietary exposure to total arsenic for UK consumers has remained fairly constant. The dietary exposures for mean and 97.5th percentile adult UK consumers in 1997 were 0.12 mg/day and 0.42 mg/day respectively. Although these exposures are similar or greater than the JECFA Provisional Tolerable Daily Intake (PTDI) for inorganic arsenic of 0.002 mg/kg bodyweight, which is equivalent to 0.12 mg/day for a 60 kg person, these are estimated for total arsenic. As has been stated before, most of the arsenic in the diet is from fish and most of the arsenic in fish is in the less toxic organic forms. 7.2.4 Mercury Mercury is the only metallic element that is liquid at room temperature. It is ubiquitous throughout the environment. It occurs mainly in the form of the ore cinnabar (mercury (II) sulphide), but there are also at least thirty minerals in which the metal is found. Exposure to mercury vapour over periods of months or years can result in chronic poisoning. Exposure to inorganic mercury compounds can result in bleeding from the gastrointestinal tract and kidney damage, followed by death from uraemia. Organomercury compounds have a wide variety of effects, and range in value from therapeutic agents to lethal chemicals. Of the organomercury compounds, the alkylmercurials are considered to be the most toxic. There have been a number of incidents involving alkylmercurials, of which methylmercury is the most intensively studied as it is thought to be the main agent of Minimata disease. In this incident, children of consumers of fish from Minimata Bay in Japan suffered from damage to the central nervous system, resulting in disturbances in vision, hearing, muscle function and mentality.27 It was later determined that there had been discharges of methylmercury in effluent from a chemical plant to the bay.

Inorganic contaminants in food

157

The main sources of exposure to mercury for the general population are from the diet and dental amalgam.21, 28 The main dietary source of mercury is fish and this has led to interest in potential exposure to mercury on the neurological development of children from populations with high fish consumption.29 There is no statutory control of mercury in most food in the UK, but the levels of mercury in fish are controlled by European Commission Decision 93/351/EEC which sets an average limit for mercury in fish of 0.5 mg/kg or 0.5 ppm (part per million).30 This average limit is, however, increased to 1.0 mg/kg or 1 ppm for the edible parts of the predatory and bottom-dwelling species listed in the Annex to the Decision. Investigations into the presence of mercury in food have been carried out in a comparatively small number of foods in the UK since 1966. Mercury is included in the analytes measured in the samples of the Total Diet Study. The estimated dietary exposures to total mercury (organic and inorganic) for the general population as determined from the UK TDS (Table 7.5) have remained fairly constant between 0.002 and 0.005 mg/day. The dietary exposures of mean and 97.5th percentile consumers in the UK in 1997 were 0.0031 mg/day and 0.0064mg/day.10 These may be compared with the JECFA PTWI for mercury of 0.005 mg/kg bodyweight/week (of which no more than two-thirds should be methyl mercury),31 which is equivalent to 0.043 mg/day for a 60 kg adult. The dietary exposures are similar to those in the USA (0.008 mg/day)8 and the Netherlands (0.002 mg/day)7 but lower than New Zealand (0.013 mg/day).18 The highest concentrations of mercury are found in the fish group of the TDS, which contributes about 33% to the overall dietary exposure. Higher than usual concentrations of mercury can sometimes be found in the cereals and the meat groups, which indicates that contamination of these foods can occur. Although the use of organomercury compounds as seed dressings is now no longer permitted in the UK, it has been known for treated grain to be milled for human consumption in the past and there have been incidents in Iraq, Pakistan and Guatemala. A survey by the Working Party on the Monitoring of Foodstuffs for Heavy Metals in 1984 found that of 17 samples of the bread and cereals group analysed, none contained mercury in excess of 0.001 mg/kg.17 A hundred samples of grain analysed at the same time contained concentrations of less than 0.005 mg/kg in most cases, although four samples contained between 0.013 to 0.017 mg/kg total mercury. It is not known for sure why these levels were found,

Table 7.5 day)

Population dietary exposures to mercury from UK Total Diet Studies (mg/

1976

1979

1982

1985

1988

1991

1994

1997

0.005

0.004

0.003

n.d.

n.d.

0.002

0.004

0.003

Note: n.d. ˆ not determined.

158

Food chemical safety

but similar concentrations have been known to accumulate in grain due to atmospheric contamination. Mercury has been shown to accumulate in eggs when organomercury compounds are present in cereal used as feed. Egg white normally contains higher levels of total and organic mercury than the yolk owing to an association of mercury with the protein ovalbumin. In 1984, 75 samples of retail chicken eggs were analysed for total mercury and inorganic mercury.17 The mean mercury concentration was about 0.004 mg/kg (range < 0.0005 to 0.029 mg/kg). An extreme consumption of eggs would be about 500 g/week (about one-and-ahalf eggs a day) and taking the mean level results in a mercury exposure to 0.01 mg/week. Mercury is normally present in fish as it is ubiquitous in the environment. Fish and shellfish caught off the coasts of England, Wales and Scotland have been regularly monitored for mercury since 1973; the latest data are for 1996.32 These data are used to meet UK obligations in the European Community and to the Oslo and Paris Conventions. Examples of these data are those collected for Liverpool Bay, into which the Mersey flows. The Mersey is subject to discharges (from chlor-alkali plants) some of which can contain mercury. Data are submitted annually to the EC to indicate compliance with the Environmental Quality Standard (EQS) of 0.3 mg/kg mercury wet weight in fish flesh. Between 1977 and 1984 there was a significant reduction in the amount of mercury discharged into the marine environment: the data show a 52% reduction in mean mercury concentrations in all species of fish in the Liverpool Bay area. This decrease has been continued in recent years with the mean concentration decreasing from 0.23 mg/kg to 0.14 mg/kg between 1985 and 1994. Thus, it is apparent that, overall, levels of mercury in fish in Liverpool Bay are decreasing and easily comply with the EQS. As it is known that fish is a major source of mercury in the diet, extensive surveillance work has been carried out. Predatory species such as tuna and swordfish tend to accumulate relatively high levels. During 1973–74, a survey conducted by MAFF determined that mean mercury concentrations in imported fish were in general less than 0.1 mg/kg, although halibut and tuna were exceptions with mean concentrations of 0.16 mg/kg and 0.12 mg/kg respectively.27 Most of the species analysed had concentrations of less than 0.05 mg/kg. A further survey of frozen, tinned or dried imported fish in 1983 showed that with the exception of tuna and one sample of sardines, levels were all below 0.05 mg/kg, confirming the earlier findings. The mean level of tuna (tinned) was 0.12 mg/kg (range: 0.04 mg/kg to 0.22 mg/kg). Although this level was higher than those for the other species of fish, it is similar to the concentration found in the earlier survey indicating no large change in mercury concentrations. A recent survey on marine fish landed in UK ports found a range of 0.016 mg/kg to 0.14 mg/kg.15 Similarly, levels in shellfish ranged between 0.01 mg/kg and 0.29 mg/kg.

Inorganic contaminants in food

159

Table 7.6

Population dietary exposures to tin from UK Total Diet Studies (mg/day)

1976

1979

1982

1985

1988

1991

1994

1997

4.4

3.2

3.1

1.7

n.d.

5.3

2.4

1.8

Note: n.d. ˆ not determined.

7.2.5 Tin Tin is widely distributed in nature at low concentrations in plants, both marine and land, and animals. It has extensive uses in industry, where organotin compounds are used as heat stabilisers in plastics, salts in glazes in porcelain and oxide coatings used to reduce abrasion of glass containers. The main source of tin in the diet is from tin-plated steel used in the manufacture of cans for foods and beverages. High tin concentrations in food may cause short-term acute health effects in some people, including stomach upsets, abdominal cramps, nausea and/or diarrhoea. These short-term effects may occur in some individuals at concentrations above 200 mg/kg.33 The UK Tin in Food Regulations 1992 limit the maximum amount of tin in food sold in the UK to 200 mg/kg.34 Fortunately, concentrations of tin in most foods are well below 10 mg/kg, although canned foods may contain higher concentrations as a result of slow dissolution of the tin coating used on some cans to protect the steel body of the can from corrosion. Tomato-based products tend to have high levels of tin as nitrate in the food accelerates corrosion of the tin. A survey of canned tomato products sold in the UK in 1998/1999 found that 98% of the products tested were below the 200 mg/kg limit.35 The JECFA PTWI for tin is 14 mg/kg bodyweight, equivalent to 120 mg/day for a 60 kg adult. Average dietary exposure has remained in the range 1.7–5.3 mg/day (Table 7.6). The dietary exposure estimates for mean and 97.5th percentile consumers were 1.9 mg/day and 6.3 mg/day respectively, well below the PTWI. 7.2.6 Aluminium Aluminium is the third most abundant element in the earth’s crust and is used widely in the manufacture of construction materials, wiring, packaging materials and cookware. The metal and its compounds are used in the paper, glass and textile industries as well as in food additives. Despite the abundance of the metal, its chemical nature effectively excludes it from normal metabolic processes. This is due largely to the low solubility of aluminium silicates, phosphates and oxides that result in the aluminium being chemically unavailable. However, it can cause toxic effects when there are raised concentrations of aluminium in water used for renal dialysis. These effects are not seen when aluminium is at the concentrations usually present in drinking water. There is currently much activity to examine the factors that influence uptake of aluminium from the diet.

160

Food chemical safety

Table 7.7 day)

Population dietary exposures to aluminium from UK Total Diet Studies (mg/

1976

1979

1982

1985

1988

1991

1994

1997

n.d.

n.d.

n.d.

n.d.

3.9.

10

11

3.4

Note: n.d. ˆ not determined.

In the 1997 TDS, the highest concentrations were found in the bread (6.6 mg/ kg) and fish (6.1 mg/kg) food groups.10 The concentration found in bread may reflect the use of permitted aluminium-containing additives. The largest contributions to dietary exposure were from beverages, bread and cereal products; beverages owing to their high consumption and cereal products owing to the use again of aluminium additives. Aluminium has been measured in a range of crops.36 A large variation in levels was apparent for the various species. The highest concentrations were found in lettuce (6.2–810 mg/kg), parsnips (6.0–82 mg/kg) and Brussels tops (7.7–116 mg/kg). Aluminium has been included in the UK TDS only since 1988, owing to the difficulty of obtaining reliable analytical data in the presence of environmental levels of aluminium (Table 7.7). Dietary exposures for mean and 97.5th percentile for UK consumers in 1997 were 3.2 mg/day and 5.7 mg/day respectively,10 similar to the population exposure. All the estimated exposures were well below the JECFA PTWI of 7 mg/kg bodyweight, which is equivalent to 60 mg/day for a 60 kg adult. The population estimate is similar to dietary exposures reported for other countries. A dietary exposure in the USA of 11.5 mg/day was reported for 14–16-year old males,8 while dietary exposures in an Italian TDS were between 2.3–6.3 mg/day.37 7.2.7 Copper Copper is an essential element for all plants and animals, but can be toxic at high levels of exposure. It is widely distributed and always present in food. Animal offal, which is the major contributor to dietary exposure to copper, various shellfish and nuts contain, on average, more than 20 mg/kg. This was the limit recommended for food by the Food Standards Committee in 1956. Milk contains little copper, usually less than 0.1 mg/kg, and fresh fish and alcoholic drinks contain less than 1 mg/kg of copper. Table 7.8

Population dietary exposures to copper from UK Total Diet Studies (mg/day)

1976

1979

1982

1985

1988

1991

1994

1997

1.8

n.d.

1.3

1.3

n.d.

1.4

1.2

1.2

Note: n.d. ˆ not determined.

Inorganic contaminants in food

161

The population daily exposure to copper from UK Total Diet Study (TDS) samples (Table 7.8) has remained fairly constant, being 1.8 mg/person in 1976 and 1.2 mg/person in 1997.10 The mean and 97.5th percentile exposures were 1.4 mg/ day and 3.2 mg/day respectively. These latter values are well below the JECFA PTDI of 0.5 mg/kg bodyweight, equivalent to 30 mg/day for a 60 kg adult.38 7.2.8 Other elements Most of the surveys on metals and metalloids in food have concentrated on those elements that are known to be toxic, or where there are possible concerns about their levels in food. In the course of collecting the data, information on other metals is often collected in addition. Other metals that have been included in the UK Government’s surveillance are zinc, antimony, chromium, cobalt, indium, nickel, thallium and tin. Zinc As with copper, zinc is an essential element for all plants and animals. It is necessary for the correct function of various enzyme systems. However, excessive intakes of zinc can have long-term effects as they can interfere with absorption of copper and iron in the diet, and may result in anaemia.39 JECFA has recommended a PTDI of 1 mg/kg bodyweight, equivalent to 60 mg/day for a 60 kg adult.38 In food, the major contributors to the diet are meat and its products, from which zinc is readily absorbed. Liver, with concentrations of around 62 mg/kg, contains the highest levels of any meat products, with other tissues having values of a half to a third of this figure. The second greatest source of zinc is cereals. Concentrations of zinc in whole cereal products are similar to those in meat. The mean and high-level dietary exposures to zinc in the UK in 1997 were 11 and 20 mg/day respectively.10 The time trend for the population dietary exposure is given in Table 7.9. Antimony Compounds of antimony are used as fire retardants in plastics and paper, and for veterinary purposes. The metal is found in specialised alloys such as white metal bearings and pewter, which is an alloy of tin, antimony (up to 7.5%) and copper. Concentrations in food are low, generally in the range < 0.01 to 0.08 mg/kg, but have been found to be higher in samples of aspic jelly and cream of tartar.40 The Table 7.9

Population dietary exposures to zinc from UK Total Diet Studies (mg/day)

1976

1979

1982

1985

1988

1991

1994

1997

10

n.d.

10

10

n.d.

10

8.4

8.4

Note: n.d. ˆ not determined.

162

Food chemical safety

estimated population dietary exposure from the UK TDS was estimated to be 0.002 mg/day in 1985. In comparison, the average and mean dietary exposures in the 1994 TDS were both 0.003 mg/day. The corresponding high level (97.5th percentile) dietary exposure estimate was 0.004 mg/day.41 Chromium Chromium is used in the manufacture of stainless steel and other specialist steels, and non-ferrous alloys. Chromate salts are used as tanning agents, pigments, catalysts, corrosion inhibitors and in electroplating solutions. Although stainless steel is inert, the dissolution of chromium from this steel is likely to be the major source of chromium in food. Chromium is an essential element for man. The minimal requirement for man is estimated to be about 1 g/day. As the absorption of inorganic chromium (Cr III) is about 0.5% of a given dose, and the absorption of organically bound chromium is even higher, a dietary exposure to 200 g chromium/day will provide the estimated requirement. The US National Academy of Sciences has recommended a dietary intake of 50 to 200 mg/day for adults.42 From the 1997 UK TDS, the mean and 97.5th percentile consumer dietary exposures were 0.1 mg/day and 0.17 mg/day respectively. There is no JECFA PTWI or PTDI for chromium. Cobalt Cobalt is used in the manufacture of alloys and in nuclear technology. Cobalt compounds are also included in trace element supplement preparations for ruminants. The cobalt concentrations in vegetables and other foods have been found to be between < 0.01 and 0.83 mg/kg, while levels in milk are between 0.0002 and 0.06 mg/kg.40 Cobalt concentrations in dietary supplements ranged between