Antiepileptic Drugs Combination Therapy and Interactions

This page intentionally left blank

This book reviews the use of antiepileptic drugs focussing on the interactions between these drugs, and between antiepileptics and other drugs. These interactions can be beneficial or can cause harm. The aim of this book is to increase awareness of the possible impact of combination pharmacotherapies. Pharmacokinetic and pharmacodynamic interactions are discussed sup- ported by clinical and experimental data. The book consists of five parts covering the general concepts and advantages of combination therapies, the principles of drug interactions, the mechanisms of interactions, drug interactions in specific populations or in patients with co-mor- bid health conditions, concluding with a look at the future directions for this field of research. The book will be of interest to all who prescribe antiepileptics to epileptic and non-epileptic patients, including epileptologists, neurologists, neuropediatricians, psychiatrists and general practitioners.

Antiepileptic Drugs

Combination Therapy and Interactions

Edited by

Jerzy Majkowski
The Foundation of Epileptology, Warsaw

Blaise F. D. Bourgeois Harvard Medical School, USA

Philip N. Patsalos Institute of Neurology, UK

and

Richard H. Mattson
Yale University School of Medicine, USA

cambridge university press

Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo Cambridge University Press

The Edinburgh Building, Cambridge cb2 2ru, UK

Published in the United States of America by Cambridge University Press, New York

http://www.cambridge.org

Information on this title: http://www.cambridge.org/9780521822190

This publication is in copyright. Subject to statutory exception and to the provision of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.

First published in print format 2005

isbn-13 978-0-511-12593-5 eBook (NetLibrary) isbn-10 0-511-12593-3 eBook (NetLibrary)

isbn-13 978-0-521-82219-0

hardback

isbn-10 0-521-82219-x

hardback

Cambridge University Press has no responsibility for the persistence or accuracy of urls for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

Part I

1 2

Part II

4 5 6 7 8

Page ix xiii xv xvii

16 26

57 Influence of food and drugs on the bioavailability of antiepileptic drugs 93

Carlos A. Fontes Ribeiro

Interactions between antiepileptic drugs 111 Bernhard Rambeck and Theodor W. May

Interactions between antiepileptic and non-antiepileptic drugs 139 Jerzy Majkowski and Philip N. Patsalos

Contents

List of contributors Foreword Giuliano Avanzini Foreword Torbjörn Tomson Acknowledgements

Introduction

Combination therapy of diseases: general concepts

Emma Mason and Philip A. Routledge

Combination therapy with antiepileptic drugs: potential advantages and problems
Richard H. Mattson

Pharmacogenetic aspects

Matthew C. Walker, Michael R. Johnson and Philip N. Patsalos

Pharmacokinetic interactions

Pharmacokinetic principles and mechanisms of drug interactions

Philip N. Patsalos

Predictability of metabolic antiepileptic drug interactions

Edoardo Spina, Emilio Perucca and Rene Levy

vi Contents

Part III

9 10 11 12 13

Part IV

14 15 16 17

18 19

Pharmacodynamic interactions 179 Pharmacodynamic principles and mechanisms of drug interactions 181

Blaise F. D. Bourgeois

Methods for assessing pharmacodynamic interactions 193 Blaise F. D. Bourgeois

Experimental studies of pharmacodynamic interactions 208 Stanislaw J. Czuczwar

Clinical studies of pharmacodynamic interactions 228 John R. Pollard and Jacqueline French

Clinical studies of pharmacodynamic interactions
between antiepileptic drugs and other drugs 241 Gaetano Zaccara, Andrea Messori and Massimo Cincotta

Drug interactions in specific patient populations
and special conditions 255

Antiepileptic drug interactions in children 257 Olivier Dulac, Elizabeth Rey and Catherine Chiron

Antiepileptic drug interactions in the elderly 273 Jeannine M. Conway and James C. Cloyd

Antiepileptic drug interactions in pregnancy 294 Mark S. Yerby

Antiepileptic drug interactions in handicapped and
mentally retarded patients 325 Matti Sillanpää

Antiepileptic drugs and sex steroids 341 Richard H. Mattson

Antiepileptic drug interactions in patients requiring
psychiatric drug treatment 350 Michael R. Trimble and Marco Mula

Antiepileptic drugs in non-epileptic health conditions:
possible interactions 369 Jerzy Majkowski

Drug monitoring in combination therapy 392 Walter Fröscher

vii Contents

Part V

24 25

Cognitive side-effects due to antiepileptic drug
combinations and interactions 403 Albert P. Aldenkamp, Mark de Krom, Irene Kotsopoulos and Jan Vermeulen

Conclusions and future perspectives 419

Selection of drug combinations in clinical practice:
current and future perspectives 421 Jerzy Majkowski

Future research: an experimental perspective 441 Rob A. Voskuyl, Daniel M. Jonker and Fernando H. Lopes da Silva

Future research: a clinical prospective 458 Carlos A. Fontes Ribeiro

Index 475

List of contributors

ProfessorAlbertP.Aldencamp DrMarkdeKrom

Department of Neurology, University Hospital of Maastricht, PO Box 21, NL 2100 AB, Heeze, The Netherlands

Professor Blaise F. D. Bourgeois

Children’s Hospital – HU2, Harvard Medical School, 300 Longwood Avenue, Boston,
MA 02115, USA

Dr Catherine Chiron

Hospital Necker-Enfants Malades, 149 Rue de Sevres, Paris 75015, France

Dr Massimo Cincotta

Unit of Neurology, Santa Maria Nuova Hospital, Florence, Italy

Dr James C. Cloyd

College of Pharmacy, University of Minnesota, Room 7101, Weaver Densford Hall, 308 Harvard St SE, Minneapolis, MN 55455 0353, USA

Dr Jeannine M. Conway

College of Pharmacy, University of Minnesota, 7-170 WDH, 308 Harvard St SE, Minneapolis, MN 55455, USA

Professor Stanislaw J. Czuczwar

Department of Pathophysiology, Medical Academy, Jaczewskiego 8, 820-090 Lublin, Poland

Department of Neurology, University Hospital of Maastricht, PO Box 21, NL 2100 AB, Heeze, The Netherlands

Professor Olivier Dulac

Hospital Necker-Enfants Malades, 149 Rue de Sevres, Paris 75015, France

Professor Carlos A. Fontes Ribeiro

Department of Pharmacology, Faculty of Medicine, 3000 Coimbra, Portugal

Professor Jacqueline French

Department of Neurology, 3 West Gates, Hospital of the University of Pennsylvania, 3400 Spruce St, PA 19104, USA

Professor Walter Fröscher

Department of Neurology and Epileptology, Die Weissenau (Department of Psychiatry I), University of Ulm, D-88214 Ravensburg, Germany

Dr Michael R. Johnson

Division of Neurosciences and Psychological Medicine, Imperial College London, Charing Cross Hospital, London,
W6 8RP, UK

Dr Daniel M. Jonker

Epilepsy Institute of the Netherlands (SEIN), Achterweg 5, NL 21 03 SW,
Heemstede, The Netherlands

x List of contributors

Dr Irene Kotsopoulos

Department of Neurology, University Hospital of Maastricht, PO Box 21, NL 2100 AB, Heeze, The Netherlands

Professor Rene Levy

Department of Pharmaceutics, School of Pharmacy, University of Washington, Health Sciences Center H-Wing, Suite 272, Seattle, WA 98195, USA

Professor Fernando H. Lopes da Silva

Epilepsy Institute of the Netherlands (SEIN), Achterweg 5, NL 21 03 SW, Heemstede, The Netherlands

Professor Jerzy Majkowski

Diagnostic and Therapeutic Center for Epilepsy, Foundation of Epileptology, ul Wiertnicza 122, 02-952 Warsaw, Poland

Dr Emma Mason

Department of Pharmacology, Therapeutics and Toxicology, Wales College of Medicine, Cardiff University, Heath Park, Cardiff, CF14 4XN, UK

Professor Richard H. Mattson

Department of Neurology, Yale University 701 LC1, 33 Cedar Street, New Haven,
CT 06510, USA

Dr Theodor W. May

Biochemisches Labor der Gesellschaft für Epilepsieforchung, Maraweg 13, D-33617 Bielefeld, Germany

Dr Andrea Messori

Unit of Pharmacy, Careggi Hospital, Florence, Italy

Dr Marco Mula

Amadeo Avogadro University,
C.so Mazzini, 18 28100 Novara, Italy

Professor Philip N. Patsalos

Pharmacology and Therapeutic Unit, Department of Clinical and Experimental Epilepsy, Institute of Neurology, Queen Square, London, WC1N 3BG; The National Society for Epilepsy, Chalfont St Peter, UK

Professor Emilio Perucca

Clinical Pharmacology Unit, University of Pavia, Piazza Botta 10, I 27100 Pavia, Italy

Dr John R. Pollard

Department of Neurology, 3 West Gates, Hospital of the University of Pennsylvania, 3400 Spruce St, PA 19104, USA

Dr Bernhard Rambeck

Biochemisches Labor der Gesellschaft für Epilepsieforschung, Maraweg 13, D-33617 Bielefeld, Germany

Dr Elizabeth Rey

Hoˆ pital Saint Vincent de Paul, Paris, France

Professor Philip A. Routledge

Department of Pharmacology, Therapeutics and Toxicology, Wales College of Medicine, Cardiff University, Heath Park, Cardiff, CF14 4XN, UK

Professor Matti Sillanpää

Department of Public Health, 20014 Turku University, Turku, Finland

Professor Edoardo Spina

Department of Clinical and Experimental Medicine and Pharmacology, University of Messina, Policlinico Universitario, Via Consolare Valeria, 98125 Messina, Italy

Professor Michael R. Trimble

The National Hospital for Neurology and Neurosurgery, Institute of Neurology, Queen Square, London, WC1N 3BG, UK

xi List of contributors

Dr Jan Vermeulen

Epilepsy centre SEIN, Heemstede, The Netherlands

DrRobA.Voskuyl

LACDR, Division of Pharmacology, Gorlaeus Laboratories, Postbus 9502, 2300 RA, Leiden, The Netherlands

Dr Matthew C. Walker

Pharmacology and Therapeutic Unit, Department of Clinical and Experimental Epilepsy, Institute of Neurology,
Queen Square, London, WC1N 3BG, UK

Dr Mark S. Yerby

North Pacific Epilepsy Research, 2455 NW Marshall St, Ste 14, Portland, OR 97201, USA

Dr Gaetano Zaccara

Department of Neurology, Ospedale S.M. Nuova, Piazza S.M. Nuova 1, 50124 Florence, Italy

Foreword

It is my special pleasure to introduce this book about the principles on which to base combination antiepileptic drug (AED) therapy and its related problems.

As reviewed in the excellent opening chapter by Mason and Routledge, therapeutic strategies involving the combination of different drugs are currently used to treat hypertension, infectious diseases and cancer in an attempt to enhance efficacy, reduce unwanted side effects and decrease the of probability of developing resistance. However, their disadvantages may exceed their benefits. First of all, drug toxicity may actually be increased by combination therapy as a result of negative pharmacody- namic interactions and the increased probability of idiosyncratic reactions. Secondly, the management of combination therapy is complicated by pharmacokinetic interac- tions. Thirdly, the risks of non-compliance and medication error are significantly greater with a multiple drug regimen.

How these general concepts apply to pharmacological antiepileptic therapy is dealt with by the most authoritative specialists in the first three parts of the book, which give considerable space to pharmacokinetic and pharmacodynamic inter- actions, while the fourth part develops these questions further with special regard to the patients’ age, associated health problem (neurological or general), and sexual life (contraception, pregnancy, etc.). The reader is thus guided in understanding the rationale for combining AEDs, and made aware of the caveats that need to be taken into account.

In an ideal situation, we should consider AED combinations in such a way as to ensure that each pharmacological ingredient targets a specific epileptogenic mecha- nism. Unfortunately, our current understanding of the basic mechanisms of epilep- togenesis and drug activity is still too limited to make such rational polypharmacy feasible. However, the favourable effects of some combinations based on traditional or newly developed AEDs (or both) is documented in the literature and here criti- cally reviewed. This information is relevant and important when choosing the drug combinations to be prescribed to patients failing to respond to single drug regimens on the basis of exploiting the potential synergies of different drugs.

It is worth noting that the availability of newly developed AEDs has made multiple drug regimens increasingly frequent in clinical practice because, until the

xiii

xiv Foreword

efficacy and tolerability of a given new drug are fully understood it would be inap- propriate (and in many instances illegal) to use it as a first choice monotherapy. A good knowledge of the advances and drawbacks of combination therapy is essen- tial for the everyday use of new AEDs.

Appropriate attention is given to the pharmacogenetic aspects underlying the variables that may influence AED responses and interaction profiles, such as metabolism, pharmacokinetics and pharmacodynamics, and there is a critical dis- cussion of the usefulness and pitfalls of genetic screening. Pharmacogenetics and pharmacogenomics are currently seen as speculative perspectives, but it is worth bearing in mind that it is already possible to characterize individuals on the basis of the polymorphisms of genes encoding drug metabolic enzymes, even though the relevance of this approach to the clinical use of combination therapy has not yet been assessed.

This book will stimulate new thoughts and ideas, and I am sure that all of its readers will learn something even about what at first glance may seem familiar sub- jects. For instance, although I was of course aware that most drug formulations con- tain multiple ingredients, it had not occurred to me that this makes the very concept of monotherapy rather relative as the active principle may make up as little as 8% of a tablet’s weight, with the rest consisting of coating and binding agents, fillers, dyes, preservatives, and solubilising and disintegrating ingredients which, however rarely, may give rise to dose-related or idiosyncratic reactions in susceptible subjects.

In summary, this book will provide readers an updated account of the state of the art and an appraisal of the exciting perspectives of an important aspect of pharmacological antiepileptic therapy. The editors (Jerzy Majkowski, Blaise Bourgeois, Philip Patsalos and Richard Mattson) wrote some of the critical chap- ters themselves, but also gathered a highly authoritative group of other scientists in order to cover the field comprehensively. In thanking them for this, I wish the book the success it deserves.

Giuliano Avanzini President of the International League Against Epilepsy

Foreword

Drug interactions may be regarded as a stimulating challenge by the pharmacologist but by the physician responsible for management of the patient, interactions are often considered cumbersome and a vexing factor complicating treatment. Drug interactions are particularly common in the treatment of patients with epilepsy. Although monotherapy has been the favoured treatment strategy for the last 25 years or so, up to 50% may not achieve satisfactory seizure control while on the first drug they have been prescribed. A high proportion of these patients will eventually end up taking a combination of different antiepileptic drugs. Until now, the selec- tion of drug combinations has more often been the result of chance or the physi- cian’s individual preferences rather than being rational or evidence-based. Given the long duration of epilepsy treatment, most patients will frequently be prescribed drugs for other conditions too. Conventional antiepileptic drugs have been among the most prone to pharmacokinetic interactions, and pharmacodynamic inter- actions occur whenever two drugs are used together. For all these reasons, the topic of combination therapy and drug interactions is of great importance and up-to-date knowledge is an essential basis for a rational approach to the pharmacological treat- ment of people with epilepsy.

The editors of the current book on Antiepileptic drugs: combination therapy and interactions have managed to gather an international group of experts to cover these and related issues in a comprehensive volume. The reader is provided the relevant general background, along with in-depth coverage of pharmacokinetic and phar- macodynamic interactions as well as interactions in specific patient populations. It is made clear that while pharmacokinetic interactions in most cases are negative, recent advances in our understanding of drug metabolism enable us to predict and avoid adverse interactions. Drug level monitoring can help us manage those inter- actions that cannot be avoided. Pharmacodynamic interactions are not always adverse. Some are advantageous, improving the therapeutic index, and could be exploited to the benefit of our patients. This volume, which should be of interest to all physicians engaged in the treatment of patients with epilepsy, shows how far we have advanced from the level where interactions could be regarded as just an

xvi Foreword

unwieldy factor complicating pharmacotherapy. Instead, the data provided will hopefully serve as a platform for more rational and effective therapeutic strategies in the future for epilepsy patients in need of combination therapy.

Torbjörn Tomson Chairman, Commission on Therapeutic Strategies, International League Against Epilepsy

Acknowledgements

This book has arisen as a result of the activities of the International League Against Epilepsy’s (ILAE) Sub-commission of Polytherapy and Drug Interactions of the Commission on Therapeutic Strategies. Under the auspices of the Sub-commission, three international pre-congress symposia/satellite symposia were organised during the 1st European Congress of Epileptology, Oporto, Portugal, 1994; the 3rd European Congress of Epileptology, Warsaw, Poland, 1998; and the XIV Conference on Epilepsy, Warsaw, Poland, 2000. These symposia brought together internation- ally recognised experts in the field of antiepileptic drug interactions and it is these experts that have contributed the chapters that constitute this book. The activities of the Sub-commission of Polytherapy and Drug Interactions and thus this book would not have been possible without the generous financial support of the ILAE, Polish Society of Epileptology, Abbott, Aventis, GlaxoSmithKline, Hoechst Marion Roussel, Novartis, Ortho (Johnson and Johnson) and Pfizer.

J. Majkowski, B.F.D. Bourgeois, P.N. Patsalos and R.H. Mattson

Part I

Introduction

Historical aspects

Combination therapy of diseases: general concepts

Emma Mason and Philip A. Routledge
Department of Pharmacology, Therapeutics and Toxicology, Wales College of Medicine, Cardiff University, Cardiff, UK

Many drugs are excellent when mingled and many are fatal

Homer 950 EC

Combination therapy has been used since therapeutics was first practiced. The physician or asu of Mesopotamia in 1700 BC used combinations of several plants, minerals and animal products in concoctions, salves and fomentations (Lyons and Petrucelli, 1987). We know little of the efficacy or toxicity of these combined med- ications. However, the Babylonian code of Hammurabi states that a doctor who causes the death of a patient or loss of an eye should lose his hands. It would not have been surprising if such stringent punishments encouraged the use of a large num- ber of non-toxic (and possibly non-efficacious) medicines. At least this would have ensured that the physician could continue to be able to mix his own preparations.

Since many early drugs were of plant origin, the use of single herbal preparations containing many potentially active ingredients resulted in combination therapy, albeit often unknowingly. Thus cannabis, advocated by the Red Emperor (Shen Nung) around 2800 BC contains around 30 cannabinoid compounds, and debate still rages today as to whether cannabis has greater therapeutic efficacy than single cannabinoid therapy (e.g. with delta-9 tetrahydrocannabinol) in certain medical conditions. Traditional Chinese medicines continue to be used regularly by up to half the population of China (Encyclopaedia Britannica, 1999), and contain several constituents prescribed in individualized doses in a bespoke fashion. The patient takes these ingredients home and boils them in a soup, before consuming the broth.

In 1753, the Scottish physician and sailor, James Lind described one of the first controlled trials of drug therapy in history, which he had performed 6 years earlier. He administered a combination treatment for scurvy containing nutmeg, garlic mustard seed, rad. raphan, balsam of Peru and gum myrrh to two sailors for 6 days.

4 Emma Mason and Philip A. Routledge

It is not surprising that the sailors who improved most were not these two individ- uals, but two others given another ‘combination therapy’ – two oranges and a lemon (Lind, 1753)!

The deliberate combination of medicines continued to be practiced right through into the nineteenth and twentieth centuries, although not embraced by all physicians. William Withey Gull (1816–1890) particularly condemned prescriptions con- taining multiple drugs. He was a passionate advocate of the scientific basis of med- icine and stated that ‘The road to a clinic goes through the pathologic museum and not through the apothecary’s shop’. Drug combinations were often contained in med- icines, the contents of which were kept secret from the patient. Dr Pierce’s Pleasant Purgative Pills were said to combine the active principles of several unspecified vegetable compounds which ‘in some inexplicable manner, gradually changed cer- tain morbid conditions of the system, and established a healthy condition instead’ (Pierce, 1891). Dr Pierce did not patent his proprietary medicines as ‘cure-alls’, but others did patent theirs, since there was little or no government regulation of ingredients or need to verify claims of therapeutic efficacy. It was not until 1938, a year after 105 people died due to an elixir of sulfonamide made up of 70% diethylene glycol that the US government legislation was introduced to ensure labeling of all

ingredients and prevention of false claims of efficacy (Routledge, 1998a).
The issue of toxicity of ingredients, which still occurs today (Stephens, 1998) is a reminder that most formulations of medicines contain several ingredients, some of which may rarely cause either dose-related (Type A) or idiosyncratic (Type B) toxicity in certain susceptible individuals. Thus, the active principle may constitute as little as 8% of the weight of a typical tablet, and the remainder may include coat- ing and binding agents, fillers, dyes, preservatives, solubilizing and disintegrating agents (Freestone, 1969). To this extent, combination therapy with several compounds occurs when only one medicine is prescribed, although the other ingredients are inactive in most individuals. However, changes to the formulation may affect bio- availability, and were responsible for an outbreak of phenytoin (diphenylhydantoin) toxicity in Australia when lactose was substituted for calcium sulfate as an excipient

(Tyrer et al., 1970).
A scientific basis for the value of combination therapy was established in the

1940s. Waksman had discovered streptomycin as the first compound to be effective in the treatment of tuberculosis (Waksman, 1949). Indeed the efficacy of strepto- mycin in tuberculosis was the subject of the first published randomized controlled trial in medicine (Medical Research Council (MRC), 1948). It was soon realized that streptomycin monotherapy required the use of large doses, which could cause sig- nificant toxicity. The emergence of streptomycin resistance was also soon recognized, and combination therapy was seen to be a possible answer to this serious problem. Thus a trial of para-aminosalicylic acid and streptomycin in pulmonary tuberculosis

5 Combination therapy of diseases: general concepts

Table 1.1 Principles for the development of chemotherapeutic regimens in oncology

. 1 Each single agent should have activity against the disease

. 2 The agents should have different mechanisms of action

. 3 The agents should have non-overlapping toxicity profiles

. 4 The regimen should combine cell cycle specific and cell cycle non-specific agents

found a reduction in streptomycin resistance from 67% in the streptomycin-only group to 10% in those treated with both agents concomitantly (MRC, 1950).

It soon became clear that similar principles applied to the treatment of malignant cells as to slow-growing pathogenic bacteria such as Mycobacterium tuberculosis. This led not only to the use of combination chemotherapy of cancer according to specific principles shown in Table 1.1 (Muggia and Von Hoff, 1997). The first three of these principles are generally applicable to combination therapy in other condi- tion, although some exceptions will be highlighted in this chapter. Before discussing the possible advantages and disadvantages of combination therapy, it is important to define and discuss two terms that have been used in this context, sometimes interchangeably.

Polypharmacy

The term polypharmacy has been in use in medicine for around 40 years. One of the first occasions on which it was used was in the context of multiple drug admin- istration versus hypnosis for surgical patients (Bartlett, 1966). This early paper did not make any suggestion that polypharmacy was a bad practice, but a subsequent review of polypharmacy in America highlighted the potential problems that polypharmacy could produce (Hudson, 1968). Indiscriminate polypharmacy has been identified as a major medical problem in some developing countries and a challenge for the World Health Organisation’s action program on essential drugs (Hogerzeil et al., 1993).

The strict definition of the word in The New Shorter Oxford Dictionary (1993) is ‘the use of several drugs or medicines together in the treatment of disease’. However this initially rather non-judgemental definition is immediately qualified with the rider ‘frequently with the suggestion of indiscriminate, unscientific or excessive pre- scription’. Other authors have assumed that the administration of an excessive number of drugs is implicit in the definition (Online Medical Dictionary, 1997). This has led to the use of the term rational polypharmacy to distinguish the appropriate use of drug combinations from indiscriminate use of several medi- cines concurrently (Kalviainen et al., 1993; Reus, 1993; Wolkowitz, 1993). Thus

6 Emma Mason and Philip A. Routledge

polypharmacy tends to be a pejorative term for excessive irrational drug use, although the drugs may be being used for a range of medical conditions rather than for a single disease.

Polytherapy

The first record of the use of this term listed on Medline was just over 20 years ago (1978) in the context of epilepsy management (Deisenhammer and Sommer, 1978). Since then it has been used predominantly in this therapeutic area, and largely by German, Italian, Spanish and French authors. It has not entered general use in the UK, where combination therapy is generally the preferred term for use of more than one drug for the same condition. The definition in the Online Medical Dictionary is ‘A therapy that uses more than one drug’. It thus differs from polyphar- macy in that it normally refers to the use of drugs for the same medical condition rather than for a group of existing medical conditions. In the following discussion, we will treat the term polytherapy as synonymous with combination therapy, a term that is more widely accepted across the spectrum of therapeutics and throughout Europe and the USA.

Epidemiology of combination therapy

Although, around 10% of the general population take more than one prescribed medicine, the incidence of combination therapy is even greater in the elderly, in females and in those who have had recent hospital admission (Nobili et al., 1997; Teng Liaw, 1997). Stewart and Cooper reviewed a number of studies and con- cluded that patients aged over 65 years use on average 2–6 prescribed medications and 1–3.4 non-prescribed medications (Stewart and Cooper, 1994).

The effects of multiple drug administration on the incidence of adverse drug reac- tions were first studied by May and co-workers in 10 518 patients hospitalized on a general medical service during a 5-year period (May et al., 1977). Their data sug- gested a disproportionately increased risk of adverse drug reactions for patients, the more drugs they were receiving. A significant proportion of these adverse drug reac- tions were due to adverse interactions between two or more co-prescribed agents.

In a case-control study by Hamilton and co-workers (who over the 3-year period 1993–1995 studied more than 157 000 patients in the USA) the drug com- bination most often associated with hospital admission was angiotensin convert- ing enzyme (ACE) inhibitors co-prescribed with potassium replacement therapy. Combinations with inhibitors of drug metabolism (particularly macrolide anti- biotics such as erythromycin) formed the next most frequent group of agents asso- ciated with increased hospitalization (Hamilton et al., 1998).

7 Combination therapy of diseases: general concepts

Advantages of combination therapy

Efficacy can be enhanced by combination therapy

One of the first indications that the use of more than one agent could be more effective than the use of either agent as monotherapy was in the treatment of severe infections (e.g. bacterial endocarditis) with combinations of penicillin and an aminoglycoside (Wilson et al., 1978). It later became clear that this synergism was achieved by a dual action on bacterial growth. Penicillins inhibited cell wall synthesis while the aminoglycoside inhibited protein synthesis. Synergism was also demon- strated between loop diuretics and thiazides, since each acted at a different site on the nephron to reduce sodium and water reabsorption. This combination (e.g. frusemide and metolazone) is still used to produce diuresis in resistant congestive cardiac fail- ure. Thus combination therapy normally involves the use of two or more drugs with different mechanism of action, and therefore normally from different drug classes.

The effects of some drug combinations are merely additive rather than synergis- tic. Nevertheless, the combination produces more efficacy than the use of each single agent alone and this can be of therapeutic benefit. Patients may now leave hospital after acute myocardial infarction on a beta-blocker, ACE inhibitor, antiplatelet agent (e.g. aspirin) and lipid lowering agent (e.g. statin), all having been shown individually to provide secondary preventive benefit in this situation. In heart failure, ACE inhibitors, beta-blockers and spironolactone have been shown to reduce mortality when added to standard therapy. Ischemic heart disease, heart failure and hypertension are heterogeneous diseases with multiple mechanisms contributing to their pathogenesis. It is therefore not surprising that more than one mechanism of action (and therefore more than one drug) may be needed to treat the underlying problems. In addition, several of these chronic diseases result in multiple end-organ damage and several drugs may be needed to treat the mul- tiple pathologies associated with them.

Monotherapy is effective in only around 50% of hypertensive patients, but efficacy can be increased to around 80% with the judicious use of combination therapy (Mancia et al., 1996). The need for combination therapy is also demon- strated by the hypertension optimal treatment (HOT) study. Depending on the target blood pressure, up to 74% of patients needed more than one drug to achieve the required blood pressure (Hansson et al., 1998; Opie, 1998). It is also interesting to note that in this study, patients randomized to acetylsalicylic acid had signifi- cantly reduced rates of major cardiovascular events. Thus combined antihyperten- sive and antiplatelet therapy is valuable, even though these drugs are producing their beneficial effects in completely different ways.

Anticonvulsant drugs are also thought to have a range of different mechanisms of action, but that the same principles should also apply. Even with carefully instituted

8 Emma Mason and Philip A. Routledge

and monitored monotherapy, only 70–80% of patients will achieve satisfactory control of their epilepsy (Jallon, 1997) so that combination therapy may be an option that should be considered.

Combination therapy may help to reduce the incidence and/or severity of adverse drug reactions

Dose-related (Type A) adverse drug reactions are thought to make up around 75% of all adverse drug reactions (Routledge, 1998b). Combinations of medicines with different spectra of adverse drug reactions may therefore allow reduction of dose of each compound to levels that are less likely to produce clinically relevant toxic- ity. This principle (i.e. that the agents should have non-overlapping toxicity) is one of the underlying reasons for the general use of combination chemotherapy in cancer (Muggia and Von Hoff, 1997). In the case of tuberculosis, the use of triple and quadruple antituberculous chemotherapy has allowed some potentially toxic agents (e.g. ethambutol and pyrazinamide) to be used at lower and therefore safer doses than previously. This approach has also allowed shorter treatment courses, thus reducing duration of exposure to risk of toxicity. In hypertension combina- tions of low doses of two agents from different classes have been shown to provide additional antihypertensive efficacy, thereby minimizing the likelihood of dose- dependent adverse effects.

Combination therapy can prevent the development of resistance

The experience of treatment of tuberculosis indicated that combination therapy might help to prevent the emergence of resistant bacteria. Chambers and Sande (1996) have elegantly argued that if spontaneous mutation were the major mecha- nism by which bacteria acquired antibiotic resistance, combination chemotherapy should be effective. They illustrate their argument with the example of a micro- organism that has a frequency of development of resistance to one drug of 10 7 and to a second drug of 10 6. In this case, the probability of independent mutation of resistance to both drugs in a single cell would be the product of the two frequencies (i.e. 10 13) making the likelihood of development of resistance extremely small. Such arguments clearly apply to other situations such as oncology where the devel- opment of resistance can otherwise limit drug efficacy. They are less relevant to the treatment of other diseases.

Disadvantages of combination therapy

The evidence for the benefits of combined therapy is often poor

At the beginning of the last century, therapeutics was based more on the experi- ence of others, rather than on firm evidence. Thus Wilson was able to state that

9 Combination therapy of diseases: general concepts

although many remedies had been tried and were still in favor for the treatment of epilepsy, the only ones that have any effect are the bromides of potassium, sodium and ammonium. It is interesting to note that ‘the best results seem to follow the administration of all three in a combined dose’ (Wilson, 1912).

In diabetes, the benefits of combination therapy with a biguanide (e.g. met- formin) and sulfonylurea (e.g. glibenclamide), in patients with Type 2 (non-insulin- dependent) diabetes who are inadequately controlled with either agent alone, have been claimed for 40 years. The mechanism of action of these two drug classes is dif- ferent. Biguanides such as metformin (which first became available in Europe in 1957), work by increasing the action of insulin in peripheral tissues and reducing hepatic glucose output due to inhibition of gluconeogenesis. Sulfonylureas act pri- marily by potentiating glucose-stimulated insulin release from functioning pancre- atic islet beta-cells (O’Meara et al., 1990), although studies of insulin secretion at the same plasma glucose concentrations before and during long-term sulfonylurea therapy have shown increased beta-cell sensitivity to glucose and continuously aug- mented insulin secretion (Gerich, 1989). However, the evidence for combined ther- apy sulfonylurea/biguanide was relatively sparse for many years, and rested largely on a single non-randomized observational trial of 108 sulfonylurea failures (Clarke and Duncan, 1965). It was only 30 years later that controlled trials confirmed the benefits of this combination of agents (Hermann et al., 1994; DeFronzo, 1995).

In his article on rational polypharmacy in epilepsy, Richens points out that few randomized placebo-controlled studies have been undertaken to compare the rela- tive merits of monotherapy and combination therapy with respect to seizure control (Richens, 1995). Evidence-based medicine should play an important role in the ther- apeutics of epilepsy, as it has increasingly done in other areas of disease management.

Toxicity may be greater with combination therapy than monotherapy

One of the principles of combination therapy in cancer is that the agents should have non-overlapping toxicity. Clearly this is not always possible, even in oncology, since many anti-cancer drugs share similar toxicity profiles (e.g. myelotoxicity). It may also be difficult to achieve in other therapeutic areas.

It is possible that combination therapy is a risk factor in the production of sud- den unexpected death in epilepsy, although the use of more than one drug may just reflect the severe unstable nature of the epilepsy in such individuals (Nilsson et al., 2001). It is also possible that combination therapy is associated with greater risk of anticonvulsant embryopathy in infants exposed to anticonvulsant drugs in utero, (control frequency 8.5%, monotherapy 20.6%, combination therapy 28.0%) so that the risks from each agent in this situation may be additive (Holmes et al., 2001).

Non-steroidal anti-inflammatory drugs (NSAIDs) used in the treatment of arthritis can increase the risk of peptic ulcer by around four-fold in patients aged

10 Emma Mason and Philip A. Routledge

65 years or older (Griffin et al., 1991). Corticosteroids are also used in some patients with arthritis, particularly rheumatoid arthritis. Piper and colleagues, using the same design and patient database as Griffin, showed that the estimated relative risk for the development of peptic ulcer disease among current users of oral corticosteroids (but not NSAIDs) was 1.1 (i.e. a 10% increase in risk). However, patients concurrently receiving corticosteroids and NSAIDs had a risk for peptic ulcer disease that was 15 times greater than that of non-users of either drug (Piper et al., 1991).

Similarly, compared with non-users of either drug, the relative risk of hemorrhagic peptic ulcer disease among current users of both anticoagulants and NSAIDs was 12.7 (95% confidence interval, 6.3–25.7)(Shorr et al., 1993). However, the prevalence of NSAID use among anticoagulant users was 13.5%, the same as in those who were not using anticoagulants. Thus toxicity of drug combinations may sometimes be syner- gistic and be greater than the sum of the risks of toxicity of either agent used alone.

Enhanced toxicity of drug combinations may sometimes be due to pharmaco- kinetic interaction. Herpes Zoster infections are not uncommon in immuno- compromised patients, and anti-viral agents may be required. Unfortunately 19 people with cancer and Herpes Zoster died in Japan in 1993 because of fluoro-pyrimidine toxicity, caused by the inhibition of 5-fluorouracil metabolism by the metabolite of a new anti-viral agent, sorivudine. Sixteen of the deaths occurred after the drug had been licensed, illustrating that not all drug interactions may be recognized before marketing and widespread exposure to the offending combination of agents occurs (Watabe, 1996).

In 1997, Mibefradil (Posicor) was marketed in the USA and Europe for the treat- ment of hypertension and angina as an exciting new molecule that selectively blocked T-calcium channels (Frishman, 1997). It was already known before market- ing that mibefradil inhibited the metabolism of three potentially toxic agents, astem- izole, cisapride and terfenadine. Soon further clinically significant interactions with cyclosporin and tricyclic antidepressants were being reported. It was known that mibefradil could inhibit the action of cytochrome P450 3A4 and thus reduce the clearance of other drugs that were metabolized by this enzyme. In December 1997, because of seven reports of statin-induced rhabdomyolysis in patients receiving simvastatin and mibefradil, lovastatin and simvastatin were added to the list of those that should never be co-administered with mibefradil. This was of particular importance, since hypertension and hypercholesterolemia are important and often co-existing risk factors for ischemic heart disease.

Finally, as a result of the number of serious interactions, the manufacturers announced the withdrawal of mibefradil from the market in 1998; almost exactly a year after the drug had been given marketing approval (Po and Zhang, 1998). Recently Wandel and co-workers have used a human intestinal cancer-derived cell which expresses P-glycoprotein to show that mibefradil is not only a substrate for

11 Combination therapy of diseases: general concepts

P-glycoprotein, but may well be a potent inhibitor of this efflux pump mechanism (Wandel et al., 2000). Thus its combined effects on CYP3A and P-glycoprotein could explain the magnitude of the effect of its interactions with other drugs. Thus the clinical significance of potential interactions may not be fully realized until after marketing.

Combination therapy may be associated with increased risk of non-compliance (non-concordance)

Compliance with therapy is an essential prerequisite of obtaining the benefits of the drugs. The use of combination therapy means that the patient has to take more tablets, unless the drugs have been formulated in a combined preparation. If two drugs are being used in combination, the dose of each should be adjusted to achieve optimal benefit. Thus, patient compliance is essential, yet more difficult to achieve. If patients perceive that they are being overmedicated, they self-report that their com- pliance falls (Fincke et al., 1998). Polypharmacy may thus result in poor compliance, which may itself result in failure of therapy. This mechanism has been reported to be a problem in individuals with epilepsy (Lambie et al., 1981), and an important fac- tor precipitating admission to hospital for seizure (Lambie et al., 1986).

To obviate the problem of multiple medication use, many fixed-dose drug com- binations are marketed. The use of such combinations is advantageous only if the ratio of the fixed doses corresponds to the needs of the individual patient. In the USA, a fixed-dose combination of drugs is considered a ‘new drug’ and as such must be approved by the Food and Drug Administration (FDA) before it can be marketed, even though the individual drugs are available for concurrent use. To be approved, certain conditions must be met. Either the two drugs must act to achieve a better therapeutic response than either drug alone (e.g. many antihypertensive drug combinations); or one drug must act to reduce the incidence of adverse effects caused by the other (e.g. a diuretic that promotes the urinary excretion of K combined with a K -sparing diuretic) (Nies and Spielberg, 1996).

Combination therapy may be associated with an increased risk of medication error

Misuse of medications is a major cause of morbidity and mortality. Patients’ med- ication bottles and their reported use of medications were compared with physi- cians’ records of outpatients in Boston, Massachusetts. Discrepancies were present in 239 patients (76%). The 545 discrepancies in these patients were the result of patients taking medications that were not recorded (n 278 [51%]); patients not taking a recorded medication (n 158 [29%]) and differences in dosage (n 109 [20%]). Older age and polypharmacy were the most significant correlates of discrepancy (Bedell et al., 2000).

12 Emma Mason and Philip A. Routledge

Conclusions

Combination therapy is an essential therapeutic tool, although one that can all too often be misused, to the detriment of the patient. The efficacy of many treatment schedules can be enhanced by combination therapy, and this approach may help to reduce the incidence and/or severity of adverse drug reactions. In cancer and anti- infective chemotherapy, combination therapy can prevent or at least delay the development of resistance.

However there is often a dearth of robust evidence for the benefits of certain drug combinations. Increased toxicity, sometimes as a result of direct interaction, is also a possibility. Finally combination therapy may be associated with an increased risk of non-compliance (non-concordance). A good working knowledge of the pharmacology of the drugs prescribed, and the potential for interaction is an important part of obtaining the benefits of combination therapy and minimizing toxicity. In addition, the risk of medication error in patients on multiple medicines means that physicians should check medication lists with patients carefully. Since patients are major stakeholders in the prescribing process, they should be encour- aged to engage in a ‘prescribing partnership’. They can help in the monitoring of therapy by alerting physicians, pharmacists and other healthcare professionals to problems that occur, especially when new drugs are introduced or doses of existing agents are changed (Seymour and Routledge, 1998).

REFERENCES

Bartlett EE. Polypharmacy versus hypnosis in surgical patients. Pac Med Surg 1966; 74: 109–112. Bedell SE, Jabbour S, Goldberg R, et al. Discrepancies in the use of medications: their extent and

predictors in an outpatient practice. Arch Intern Med 2000; 160: 2129–2134.
Chambers HF, Sande MA. Antimicrobial agents: general considerations. In Goodman and

Oilman’s The Pharmacological basis of Therapeutics, 9th edn. USA: McGraw Hill, 1996. DeFronzo RA, Goodman A and the Multicenter Metformin Study Group. Efficacy of metformin in patients with non-insulin-dependent diabetes mellitus. New Engl J Med 1995; 335: 541–549. Deisenhammer E, Sommer R. Blood level of antiepileptic drugs and therapeutic effect in mono-

and polytherapy. Nervenarzt 1978; 49: 674–677.
Duncan LJ, Clarke BF, Munro JF. The treatment of diabetes mellitus with oral hypoglycaemic

drugs. Curr Med Drug 1965; 5: 23–32.
Encyclopaedia Britannica on CD ROM, Multimedia Edition. 1999.
Fincke BG, Miller DR, Spiro A III. The interaction of patient perception of overmedication with

drug compliance and side effects. J Gen Intern Med 1998; 13: 182–185.
Freestone DS. Formulation and therapeutic efficacy of drugs used in clinical trials. Lancet 1969;

2: 98–99.

13 Combination therapy of diseases: general concepts

Frishman WH. Mibefradil: A new selective T-channel calcium antagonist for hypertension and angina pectoris. J. Cardiovasc Pharmacol Ther 1997; 2: 321–330.

Gerich JE. Oral hypoglycaemic agents. New Engl J Med 1989; 321: 1231–1235.
Griffin MR, Piper JM, Daugherty JR, et al. Nonsteroidal anti-inflammatory drug use and increased risk for peptic ulcer disease in elderly persons. Ann Intern Med 1991; 114: 257–263. Hamilton RA, Briceland LL, Andritz MH. Frequency of hospitalization after exposure to known

drug–drug interactions in a Medicaid population. Pharmacotherapy 1998; 18: 1112–1120. Hansson L, Zanchetti A, Carruthers SG, et al. Effects of intensive blood-pressure lowering and low-dose aspirin in patients with hypertension: principal results of the Hypertension Optimal

Treatment randomised trial. Lancet 1998; 351: 1755–1762.
Hermann LS, Schersten B, Bitzen PO, et al. Therapeutic comparison of metformin and

sulfonylurea, alone and in various combinations. Diabetes Care 1994; 17: 1100–1109. Hogerzeil HV, Bimo, Ross-Degnan D, Laing RO, et al. Field tests for rational drug use in twelve

developing countries. Lancet 1993; 342: 1408–1410.
Holmes LB, Harvey EA, Coull BA, et al. The teratogenicity of anticonvulsant drugs. New Engl J

Med 2001; 344: 1132–1138.
Hudson RP. Polypharmacy in twentieth century America. Clin Pharmacol Ther 1968; 9: 2–10. Jallon P. Monotherapy and polytherapy use of anti-epileptic drugs. Development of views. Rev

Neurol (Paris) 1997; 153(Suppl. 1): 829–833.
Kalviainen R, Keranen T, Riekkinen Sr PJ. Place of newer antiepileptic drugs in the treatment of

epilepsy. Drugs 1993; 46: 1009–1024.
Lambie DG, Johnson RH, Stanaway L. Prescribing patterns for epilepsy. New Zeal Med J 1981;

93: 15–19.
Lambie DG, Stanaway L, Johnson RH. Factors which influence the effectiveness of treatment of

epilepsy. Aust NZ J Med 1986; 16: 779–784.
Lind J. A Treatise of the Scurvy. In three parts. Containing an inquiry into the nature, causes and

cure, of that disease; together with a critical and chronological view of what has been pub- lished on the subject. Edinburgh: Printed by Sands, Murray and Cochran for A. Kincaid and A. Donaldson, 1753.

Lyons AS, Petrucelli RJ. Medicine: An Illustrated History. New York: Harry N Abrams, Inc., 1987. Mancia G, et al. Guidelines for the treatment of hypertension: a commentary. Curr Ther Res

1996; 57: 3–15.
Materson BJ, Reda DJ, Cushman WC. Department of Veterans Affairs single-drug therapy of

hypertension study. Revised figures and new data. Department of Veterans Affairs Cooperative

Study Group on Antihypertensive Agents. Am J Hypertens 1995; 8: 189–192.
May FE, Fuller S, Stewart RB. Drug use and adverse drug reactions prior to and during hospital-

ization. J Amer Pharm Assoc 1977; 17: 560–563.
Medical Research Council. Streptomycin treatment of pulmonary tuberculosis: a Medical

Research Council investigation. Br Med J 1948; 2: 769–782.
Treatment of pulmonary tuberculosis with streptomycin and para-aminosalycilic acid. Br Med J 1950; 2: 1073–1083.

Muggia FM, Von Hoff DD. Malignant diseases. In Avery’s Drug Treatment, 4th edn. T. M. Speight, N. H. G. Holford, eds. Auckland: Adis International, 1997.

14 Emma Mason and Philip A. Routledge

Nies AS, Spielberg SP. Principles of therapeutics. In Goodman and Oilman’s The Pharmacological Basis of Therapeutics, 9th edn. USA: McGraw Hill, 1996.

Nilsson L, Bergman U, Diwan V, et al. Antiepileptic drug therapy and its management in sudden unexpected death in epilepsy: a case-control study. Epilepsia 2001; 42: 667–673.

Nobili A, Tettamanti M, Frattura L, et al. Drug use by the elderly in Italy. Ann Pharmacother 1997; 31: 416–422.

O’Meara NM, Shapiro ET, van Cauter E, et al. Effect of glyburide on b-cell responsiveness to glu- cose in non-insulin-dependent diabetes mellitus. Am J Med 1990; 89(Suppl. 2A): 11S–16S.

Online Medical Dictionary. http://www.medic8.com/MedicalDictionary.htm
Opie LH. Principles of combination therapy for hypertension. What we learn from HOT and

other studies – a personal point of view. Cardiovasc Drug Ther 1998; 12: 425–429.
Pellock JM. Antiepileptic drug therapy in the United States: a review of clinical studies and

unmet needs. Neurology 1995; 45: 817–824.
Pierce RV. The Peoples Common Sense Medical Adviser in Plain English or Medicine Simplified,

26th edn. Buffalo, NY: World’s Dispensary Printing Office and Bindery, 1891.
Piper JM, Ray WA, Daugherty JR, et al. Corticosteroid use and peptic ulcer disease: role of non-

steroidal anti-inflammatory drugs. Ann Intern Med 1991; 114: 735–740.
Po AL, Zhang WY. What lessons can be learnt from withdrawal of mibefradil from the market?

Lancet 1998; 351: 1829–1830.
Reus VI. Rational polypharmacy in the treatment of mood disorders. Ann Clin Psychiatry 1993;

5: 91–100.
Richens A. Rational polypharmacy. Seizure 1995; 4: 211–214.
Routledge P. 150 years of pharmacovigilance. Lancet 1998a; 351: 1200–1201.
Routledge P. In Detection of New Adverse Reactions, 4th edn. M. D. B. Stephens, J. C. C. Talbot,

P. A. Routledge, eds. London: Macmillan Reference Ltd., 1998b.
Seymour RM, Routledge PA. Important drug–drug interactions in the elderly. Drug Ageing 1998;

6: 485–494.
Shorr RI, Ray WA, Daugherty JR, et al. Concurrent use of nonsteroidal anti-inflammatory drugs

and oral anticoagulants places elderly persons at high risk for hemorrhagic peptic ulcer dis-

ease. Arch Intern Med 1993; 153(14): 1665–1670.
Stephens M. In Detection of New Adverse Reactions, 4th edn. M. D. B. Stephens, J. C. C. Talbot,

P. A. Routledge, eds. London: Macmillan Reference Ltd., 1998.
Stewart RB, Cooper JW. Polypharmacy in the aged. Practical solutions. Drug Aging 1994;

4: 449–461.
Teng Liaw ST. Drug interactions among the elderly. Aust Fam Phys 1997; 26: 355–357.
The New Shorter Oxford Dictionary. L. Brown, ed. Oxford: Oxford University Press, 2001.
Tyrer JH, Eadie MJ, Sutherland JM, et al. Outbreak of anticonvulsant intoxication in an Australian

city. Br Med J 1970; 4: 271–273.
Waksman SA (ed.). Streptomycin: Nature, and Practical Applications. Baltimore: The Williams &

Wilkins Co., 1949.
Wandel C, Kim RB, Guengerich FP, et al. Mibefradil is a P-glycoprotein substrate and a potent

inhibitor of both P-glycoprotein and CYP3A in vitro. Drug Metab Dispos 2000; 28: 895–898.

15 Combination therapy of diseases: general concepts

Watabe T. Strategic proposals for predicting drug–drug interactions during new drug develop- ment: based on sixteen deaths caused by interactions of the new antiviral sorivudine with 5-fluorouracil prodrugs. J Toxicol Sci 1996; 21: 299–300.

Wilson A. Diseases of the brain and nervous system. In The Modern Physician. London: Caxton Publishing Company, 1912.

Wilson WR, Geraci IE, Wilkowske CJ, et al. Short-term intramuscular therapy with procaine penicillin plus streptomycin for infective endocarditis due to viridans streptococci. Circulation 1978; 57: 1158–1161.

Wolkowitz OM. Rational polypharmacy in schizophrenia. Ann Clin Psychiat 1993; 5: 79–90.

Combination therapy with antiepileptic drugs: potential advantages and problems

Richard H. Mattson
Department of Neurology, Yale University, New Haven, CT, USA

Rationale for combination therapy

Antiepileptic drug (AED) treatment of epilepsy to prevent or minimize recurrent epileptic seizures begins with the use of a single agent as monotherapy. The pri- mary reason why two or more drugs are used together is the failure of monotherapy to control the seizures. Depending on the type of seizures and epilepsy syndrome, control may be complete or very poor. In adult-onset seizures control varies between 35% and 60% for partial seizures and 10% and 20% higher for tonic–clonic seizures after 1-year follow-up (Mattson et al., 1985, 1992, 1996; Richens et al., 1994; Heller et al., 1995; Kwan and Brodie, 2000). The long-term response is less favor- able for some patients because breakthrough seizures occur over time although others enter remission (Richens et al., 1994; Heller et al., 1995). Seizures associated with idiopathic generalized epilepsies are usually more easily controlled.

When seizures continue despite increasing doses of the initial AED to the maxi- mum that can be tolerated, a second drug is usually added to the first in an effort to achieve better control. In those patients with particularly refractory seizures/ epilepsy, three and even four AEDs are occasionally employed. The overall indica- tions for and selection of combined therapy as well as the associated problems encountered are issues of importance.

When initial monotherapy fails to provide adequate seizure control despite being optimally given, an alternative AED is added and, when possible, is titrated up gradually. Dose increases are made as tolerated and as needed to obtain control. The intent is to taper the first AED to again achieve monotherapy. If addition of a second drug fails, another is recommended as an alternate until it is clear that con- trol cannot be achieved with use of a single agent. In actual practice it is common for the second AED to be added without tapering of the first. At times, the patient may be unwilling to change the medical regimen if complete or significantly

17 Combination therapy with antiepileptic drugs

improved control has been achieved with a combination of two or more AEDs for fear of a recurrent seizure with all the attendant medical and social complications.

Background

The use of combination therapy for management of medical or psychiatric prob- lems extends far back, even before the time when pharmacodynamically active products became available. When Locock and later others used bromides for treat- ment of seizures, they often combined multiple agents, although only bromides ultimately proved effective. Turner (1907) stated ‘perhaps the drug most frequently used as a substitute for, or as an adjuvant to, the bromides is borax (sodium biborate)’. Others included belladonna, zinc salts and opium. Similarly, when Hauptmann (1912) first introduced phenobarbital (PB), it was often used in combination with the bromides. This pattern has persisted with virtually every new AED introduced. The addition was usually made to improve seizure control. After the introduction of phenytoin (PHT), Yahr and colleagues (1953) reported that PHT was more suc- cessful than PB but the combination produced the best control of seizures in patients not controlled by either alone. Indeed, a product became available from Parke-Davis known as phelantin that contained 100 mg of dilantin and 32 mg of PB. New onset patients could be put on the combination without ever trying monotherapy. The lack of dosing flexibility together with studies by Reynolds and co-workers (1981) as well as Schmidt (1983) in the early 1980s emphasized that monotherapy was as effective as polytherapy in the majority of patients and was associated with fewer adverse effects. A shift to the use of monotherapy fol- lowed and has remained the accepted principle to the present time.

Potential advantages of combination therapy

It is assumed that combined AEDs work to increase efficacy (Table 2.1) either by an additive and/or synergistic effect or by achieving infra-additive adverse effects (Bourgeois and Dodson, 1988; Chapter 9, this book) allowing a higher dose to be administered. Unfortunately, more often such combinations add adverse effects at the same time and fail to improve the overall outcome or success.

Table 2.1 Advantages for combinations of AEDs

Broader spectrum
Additive efficacy Complementary mechanisms Decreased adverse effects Counteracting adverse effects

18 Richard H. Mattson

Different seizure types

Combination therapy is clearly indicated when two or more seizure types exist that fail to respond to any one agent. For example, until the introduction of valproate (VPA) with its broad spectrum of action it was necessary to combine both an anti- absence drug, ethosuximide, with another AED effective against tonic–clonic seizures, such as PB or PHT, for patients with generalized idiopathic epilepsy having both seizure types. With the introduction of new AEDs since the 1990s, many of which have broad-spectrum efficacy, the need for such combination therapy is less frequent.

Different antiepileptic mechanisms

The concept of rational polytherapy is based on the realization that most AEDs have different mechanisms and act, at least in part if not primarily, at one site to produce an anti-seizure effect. PHT, for example, has well defined effects at the sodium channel to block high-frequency discharge of action potentials. PB has an action at the gamma amino butyric acid (GABA)-A receptor enhancing chloride flux. This effect increases hyperpolarization leading to inhibition of neuronal depolarization. This concept of combining complementary mechanisms is ‘rational’ and, as noted below, is one of the most commonly used. Similar principles would favor other combinations such as PHT or other AEDs active at the sodium channel (carbamazepine, CBZ; lamotrigine, LTG) with GABA-active drugs such as barbitu- rates, vigabatrin (VGB) or tiagabine (TGB). Combinations might also include drugs whose mechanism is unclear or unknown (VPA; levetiracetam, LEV), or multiple (topiramate, TPM; felbamate, FBM). By this reasoning it would not be rational to combine AEDs with similar mechanism such as PHT with CBZ or oxcarbazepine (OXC) with CBZ. However, some evidence (below) suggests these latter combina- tions may be effective. As with all combinations, the actual evidence favoring ‘rational’ polytherapy is lacking and the concept remains theoretical.

Additive efficacy/infra-additive adverse effects

A third reason for combined AED therapy is to achieve infra-additive adverse effects and equal or better efficacy. By selecting AEDs with different adverse effect profiles, it might be expected that efficacy would be additive while adverse effects would remain tolerable. For example, dose-related adverse effects of CBZ often first appear as dizziness or visual dysfunction (blurring or diplopia) whereas PB causes sedation and cognitive compromise as doses increase. In theory, giving modest doses of both drugs should provide additive efficacy while keeping the AED levels sufficiently low to remain under the threshold for tolerability problems. In contrast, monotherapy doses increased to achieve comparable efficacy would usually double the adverse effects. Other combinations can be readily considered using this logic.

19 Combination therapy with antiepileptic drugs

Counteracting adverse effects

A fourth potential advantage of combination therapy is to add efficacy at the same time using AEDs with counteracting adverse effects. For example, combining TPM with VPA would utilize a drug causing weight loss with one causing weight gain. The help in ameliorating tremor by TPM would also decrease this side effect of VPA.

Pharmacoeconomic benefits

Although combination therapy usually adds to the cost of drugs, it can be theo- rized that a combination of PB with low doses of any other AED would be less expensive than high doses of any other drug even in monotherapy due to the very low cost of PB. Another potentially less costly combination is the use of low dose VPA with LTG. The relatively less costly VPA markedly inhibits LTG clearance, making it possible to give a half or less of LTG, the more expensive drug, and achieve comparable LTG blood levels to what would be obtained if giving double the dose as monotherapy.

Evidence of benefits of combination therapy versus monotherapy

Although many theoretical advantages can be proposed for combination therapy as noted above, it must be emphasized that there is no evidence to prove a benefit. That is not to say there is no benefit. It only emphasizes that there is a need for evidence. The only prospective, randomized, double-blind comparison between monotherapy and combination AED treatment was been carried out and published by Deckers and colleagues (2001). They compared CBZ monotherapy to VPA combined with CBZ in patients with new onset epilepsy. Doses were selected to reflect comparable ‘drug loads’ and were intended to be low. Adverse effects were the primary out- come. No significant difference was found for adverse effects (or control) although withdrawals showed a trend favoring combination therapy. Unfortunately, the num- ber of patients entered (130) was too small to detect possible clinically meaningful differences. Although this design using new onset epilepsy patients is of interest, it is not the setting in which combination therapy is commonly employed. It might be a concept to consider when initiating therapy despite the many reasons to avoid combinations as noted below.

In fact, combination therapy is almost always selected when monotherapy has failed to control seizures in a more refractory population. No prospective efficacy studies have been conducted comparing monotherapy to combinations of AEDs in patients not controlled on monotherapy when titrated to maximally tolerated doses.

The typical clinical trial design for licensing of a new AED adds an investigational AED to a regimen of one or more drugs that failed monotherapy. All approved new

20 Richard H. Mattson

AEDs have demonstrated improved control by some efficacy outcome measure. However, compared to the placebo control groups, the add-on investigational group has always been associated with more adverse effects. In addition, the designs do not increase doses of pre-study medication in the placebo group to an amount producing comparable amounts of adverse effects. If improved efficacy could still be detected in such a setting, it would be strong evidence for greater effect for combination therapy.

A prospective combination trial was attempted in the original Veterans Administration (VA) study comparing CBZ, PB, PHT and primidone (PRM). Patients failing acceptable control on monotherapy despite maximally tolerated doses on initial or a second alternate drug were randomized to a two-drug com- bination. Unfortunately only 89 patients entered this protocol and had a 1-year follow-up. Nine of the patients (11%) were fully controlled. Although this is a small number, it provided evidence of increased efficacy. However, a quantitative measure of adverse effects (Cramer, 1983) showed scores higher than the mono- therapy groups, suggesting better control came at least in part at the cost of more side effects.

In an often cited abstract Hakkarainen (1980) reported the results of a group of 100 patients randomized to either CBZ or PHT. After a year of treatment one half were controlled. Those failing were crossed to the other drug for the next year and another 17% came under control. Those still not controlled were placed on the combination and another 15% achieved remission. This work was never published in full text to allow scrutiny of the methods and results. A limitation of interpreting studies such as those above to show efficacy of combination therapy is the fact that some spontaneous remission occurs in epilepsy and inclusion of a parallel group maintained on monotherapy would be needed to demonstrate a true difference.

Evidence that AED combinations are more effective than monotherapy also can be inferred from the repeated observations that testing of new AEDs for regulatory approval is carried out by showing efficacy of an added drug compared to placebo as add-on to failed treatment with one or more drugs. However, such trials inevitably show more adverse effects of some type than the placebo group. Other observa- tions suggesting added efficacy of AED combinations are common in epilepsy monitoring units. In an effort to record events on CCTV/EEG, AEDs are com- monly reduced sequentially. The occurrence of attacks after one or more drugs is removed and another continues to be administered implies the drug removed was contributing to seizure control.

Potential problems with combined AEDs

Problems with combination AED therapy are given in Table 2.2.

21 Combination therapy with antiepileptic drugs

Table 2.2 Problems with combination AED therapy

Increased adverse effects Pharmacokinetic interactions
New active metabolites
Choice of combination
Method of initiation/discontinuation

Additive adverse effects

Add-on trials for licensing of all the new AEDs have demonstrated a statistically significant improvement in the percentage of patients achieving a 50% or greater reduction in seizures compared to placebo. Although this seems to provide clear evidence of better control with use of combined agents, virtually all trials reveal more adverse effects in the arm with an add-on drug than in the placebo arm. It is likely that drug combinations with similar adverse effects of central nervous sys- tem type are more likely to become poorly tolerated. For example, adding LTG to CBZ in clinical trials caused dizziness in 38% of patients, an additive adverse effect common to both AEDs, whereas when studied as monotherapy only 8% reported dizziness. The increased side effects may be difficult to attribute to any of a combi- nation of drugs.

In addition to increased additive or supra-additive adverse effects from pharma- codynamic mechanisms, CBZ and LTG in combination were found to have similar dose-related central nervous system (CNS) side effects and often caused dizziness, ataxia, and visual complaints when used together. Similarly LTG and VPA often increase tremor well above what is seen in monotherapy.

Pharmacokinetic interactions

All the older AEDs (CBZ, PB, PRM, PHT, VPA) are associated with potentially clin- ically important interactions when used in combination (Perucca et al., 2002). VPA inhibits PB, PRM and CBZ epoxide metabolism leading to increased blood levels with associated side effects. VPA also inhibits the clearance of LTG at times, leading to rapid elevation of LTG levels and increased risk of a hypersensitivity reaction. PB, PRM and PHT induce the clearance of CBZ and VPA such that the elimination half-life of these drugs is approximately half of what is found when the drugs are given as monotherapy. Unless more frequent dosing is given (with increased chance of non-compliance), or extended release formulations are used, peak and trough effects can lead to swings from side effects to insufficient control. Similar effects result when these inducing drugs are combined with some of the newer AEDs (LTG; TPM; zonisamide, ZNS). These problems are sufficient that the

22 Richard H. Mattson

text, Antiepileptic Drugs (Levy et al., 2002), devotes a chapter to this topic for each of the AEDs.

Active metabolites

Combinations of AEDs may produce pharmacodynamically active metabolites not present in clinically relevant concentrations when drugs are used as monotherapy. These include the conversion of PRM to PB in much greater proportion when co-administered with PHT. The consequence is that giving PRM in such a com- bination essentially means the PRM is little more than a more costly pro-drug for PB. CBZ is metabolized into the 10–11 epoxide (CBZ-E), a pharmacodynamically active product. The quantities are usually sufficiently low to be of minimal clinical effect when CBZ is used as monotherapy. When PHT is co-administered, the con- version to CBZ-E is enhanced. If VPA is combined with CBZ, inhibition of CBZ-E hydrolase occurs and levels of the CBZ-E may rise to clinically meaningful amounts. These changes may contribute to efficacy and, perhaps more importantly, to side effects.

VPA given in moderate dose is primarily metabolized to the 2-ene derivative in the mitochondria. When used at higher doses, and especially if co-administered with enzyme-inducing drugs such as PB or PHT, significant metabolism occurs in the hepatic CYP 450 system causing omega oxidation and producing putatively hepatotoxic and teratogenic products.

Selection of AED combinations

The principles that are used in selection of an added drug are different mecha- nisms and/or different adverse effects expectation, with the goal of an overall greater efficacy without a parallel increase in intolerable adverse effects. However, it must be re-emphasized that no clinical data from controlled randomized studies exist to address this theoretical issue. CBZ, LTG, or PHT, sodium channel active drugs having primarily vestibulo-cerebellar dose-related adverse effects, should be an appropriate combination with GABA-active drugs such as VGB or TGB with adverse effects of sedation or cognitive type.

CBZ, LTG, and PHT work at least in part by action at the sodium channel to pre- vent rapid neuronal firing and seizure spread. VGB or TGB act to increase GABA inhibitory effect and presumably provide different and complementary action. Some support of this concept was reported in the study of Tanganelli and Regestra (1996) in a comparative trial of CBZ or VGB alone or in combination. Although this is a ‘rational’ combination, it implies that we understand the mechanism by which the AEDs work. CBZ and PHT are both thought to function by preventing rapid firing due to action at the sodium channel. Consequently, combining both drugs should not be useful if the first was maximally given. In fact, however, this ‘non-rational’

23 Combination therapy with antiepileptic drugs

combination has been effective in clinical practice going back to early reports by Troupin and Hakkarainen (Dodrill and Troupin, 1977; Hakkarainen, 1980). Similar experience has shown that the combination of two closely related AEDs, CBZ and OXC may prove more effective that either used alone (Barcs et al., 2000).

Problems with the process of combining AEDs

When a decision is made to add a second or third AED after monotherapy has failed an adequate trial, the decision needs to be made not only what drug should be selected but how the drug should be given. Questions arise concerning initial dose, titration rate and target dose. Clinical responses of achieving seizure control or, more frequently, limitations of tolerability are the main guidelines. Adverse effects may appear as the second (or third) AED is titrated up. It is unclear whether the escalation of the add-on drug should be slowed/reversed or whether the base- line drug dose should be decreased to allow higher doses of the add-on AED. The adverse effects may be attributed erroneously to the add-on AED. For example, sedation was often observed when VPA was combined with PB. Evidence made clear that the side effect often was due to marked elevation of PB levels as a conse- quence of inhibition of PB metabolism by VPA rather than a direct effect of VPA.

Combinations of AEDs may double the cost of using monotherapy with a few exceptions mentioned above. In some cases combining an enzyme-inducing drug such as PHT with CBZ or VPA increases the clearance, often requiring a much larger dose to achieve blood levels comparable to those achieved with monotherapy. An even greater expense can be incurred by combining one of these older enzyme- inducing drugs with one of the costlier new AEDs, LTG, OXC, TPM or ZNS.

Summary

Expense

The failure of monotherapy to prevent seizures in 20–60% of patients (depending on seizure and epilepsy type) has led to combinations of AEDs to achieve better control. Although persuasive evidence indicates such treatment may improve con- trol, the benefit is usually modest and adverse effects are almost always increased for both pharmacokinetic and pharmacodynamic reasons. No adequate random- ized prospective clinical trials have compared combination of AED treatment with monotherapy in either new onset or refractory epilepsy. The absence of evidence does not mean combination therapy is not helpful, but until such evidence becomes available, treatment decisions unfortunately must be based on Level III and IV evidence.

24 Richard H. Mattson

REFERENCES

Barcs G, Walker EB, Elger CE, et al. Oxcarbazepine placebo-controlled, dose-ranging trial in refractory partial epilepsy. Epilepsia 2000; 41: 1597–1607.

Bourgeois BFD, Dodson WE. Antiepileptic and neurotoxic interactions between antiepileptic drugs. In Antiepileptic Drug Interactions. W. H. Pitlick, ed. New York: Demos, 1988: 209–219.

Brodie MJ, Yuen AW. Lamotrigine substitution study: evidence for synergism with sodium val- proate? 105 study group. Epilepsy Res 1997; 26: 423–432.

Cramer J. A method for quantification for the evaluation of antiepileptic drug therapy. Neurology 1983; 33(Suppl. 1): 26–37.

Deckers CLP, Hekster YA, Keyser A, et al. Monotherapy versus polytherapy for epilepsy: a multi- center double-blind randomized study. Epilepsia 2001; 42: 1387–1394.

Dodrill CB, Troupin AS. Psychotropic effects of carbamazepine in epilepsy: a double-blind com- parison with phenytoin. Neurology 1977; 27: 1023–1028.

Hakkarainen H. Carbamazepine vs diphenylhydantoin vs their combination in adult epilepsy. Neurology 1980; 30: 354.

Hauptmann A. Luminal bei epilepsie. Muenchener Medizinsche Wochenschrift 1912; 57: 1907–1909.

Heller AJ, Chesterman P, Elwes RDC, et al. Phenobarbitone, phenytoin, carbamazepine or sodium valproate for newly diagnosed adult epilepsy: a randomized comparative monother- apy trial. J Neurol Neurosurg Psychiatr 1995; 58: 44–50.

Kwan P, Brodie MJ. Comparison of carbamazepine, phenobarbital, phenytoin and primidone in partial and secondarily generalized tonic–clonic seizures. New Engl J Med 1985; 313: 145–151. Kwan P, Brodie MJ. Early identification of refractory epilepsy. New Engl J Med 2000; 342:

314–319.
Levy RH, Mattson RH, Meldrum BS, Perucca E (eds.). Antiepileptic Drugs, 5th edn. Philadelphia:

Lippincott Williams and Wilkins, 2002.
Mattson RH, Cramer JC, Collins JF, et al. A comparison of valproate with carbamazepine for the

treatment of complex partial seizures and secondarily generalized tonic–clonic seizures in

adults. New Engl J Med 1992; 327: 765–771.
Mattson RH, Cramer JC, Collins and the VA Cooperative Epilepsy Study Group. Prognosis for

complete control of complex partial and secondarily generalized tonic–clonic seizures.

Neurology 1996; 47: 68–76.
Perucca E, Levy RH. Combination therapy and drug interactions. In Antiepileptic Drugs, 5th edn.

R. H. Levy, R. H. Mattson, B. S. Meldrum, E. Perucca, eds. Philadelphia: Lippincott Williams

and Wilkins, 2002: 96–102.
Reynolds EH, Shorvon SD. Monotherapy or polytherapy for epilepsy? Epilepsia 1981; 22: 1–10. Richens A, Davidson DLW, Cartlidge NEF, et al., on behalf of the EPITEG Collaborative Group.

A multicentre comparative trial of sodium valproate and carbamazepine in adult onset

epilepsy, J Neurol Neurosurg Psychiatr 1994; 57: 682–687.
Schmidt D. Reduction of two-drug therapy in intractable epilepsy. Epilepsia 1983; 24: 368–376. Stephens LJ, Brodie MJ. Seizure freedom with more than one antiepileptic drug. Seizure 2002;

11: 349–351.

25 Combination therapy with antiepileptic drugs

Tanganelli P, Regestra G. Vigabatrin vs carbamazepine monotherapy in newly diagnosed focal epilepsy: a randomized response conditional cross-over study. Epilepsy Res 1996; 25: 257–262. Turner WA. Epilepsy – The Study of the Idiopathic Disease. London: McMillan and Co., Limited,

1907: 234–236.
Yahr MD, Sciarra D, Carter S, et al. Evaluation of standard anticonvulsant therapy in three hun-

dred nineteen patients. JAMA 1953; 150: 663–667.

Pharmacogenetic aspects

Matthew C. Walker1, Michael R. Johnson2 and Philip N. Patsalos1

1 Pharmacology and Therapeutic Unit, Department of Clinical and Experimental Epilepsy, Institute of Neurology, Queen Square, London, UK

2 Division of Neurosciences and Psychological Medicine, Imperial College London, Charing Cross Hospital, London, UK

Introduction

Pharmacogenetics and pharmacogenomics are fields which show how the genetic make-up of an individual can influence drugs effects. In epilepsy it is one part of a number of influences that determine drug responsiveness. Other contributors are age, sex, concomitant medication, other illnesses and cause and type of epilepsy. The cause and type of epilepsy may have a complex interaction with the genetics of drug response, as the genes that contribute to epilepsy can directly affect drug responsiveness (see below), and epilepsy itself may influence genetic expression. The observation that inherited differences can affect drug disposition, adverse effects and responsiveness is not new. The observation that there are slow meta- bolizers of phenytoin was made in the 1960s (Kutt et al., 1964), and later this was noted to be an inherited familial trait (Vasko et al., 1980; Vermeij et al., 1988).

The human genome project will undoubtedly revolutionize the practice of med- icine. The relatively small number of human genes (approximately 30 000–40 000; International Human Genome Sequencing Consortium, 2001) and the growth of rapid sequencing technology has brought the possibility of complete genome screening closer to reality. Variation in these genes, environmental factors, and their joint interactions determine our individual response to drugs. Human genetic variation mostly consists of single nucleotide polymorphisms (SNPs) and small insertion or deletion (INDELS) polymorphisms. Over 1.4 million SNPs were identified in the initial sequencing of the human genome (International SNP Map Working Group, 2001). Most of these lie in non-coding regions of the genome, with fewer (approximately 60 000) identified within exons (coding regions of the genes). Between any two genomes there are an estimated 2.3 million variants and on a population level, up to 10 million variant positions with a frequency of more than 1%. Due to linkage disequilibrium, certain patterns of SNPs within a gene are found within specific populations (Salisbury et al., 2003), which may enable a reduction in the number of SNPs that need to be genotyped in order to screen for

27 Pharmacogenetic aspects

the association of variability in a gene with disease and drug response. Although technology has advanced, in many instances tests of the gene product (e.g. enzyme activity) rather than for the gene itself may be cheaper, more reliable and more relevant (see review by Streetman et al., 2000).

Genetic polymorphisms can influence antiepileptic drug (AED) responses and, during polytherapy, their interaction profile by influencing metabolism, central nervous system penetration, pharmacodynamics and adverse events. We will con- sider the evidence for each of these in turn before reviewing the use and pitfalls of genetic screening.

Metabolism

Lipophilic drugs cannot be easily eliminated from the body, and thus are biotrans- formed to more hydrophilic compounds that are then easily excreted. This bio- transformation involves either modification of functional groups (phase I) or conjugation with hydrophilic moieties (phase II). Both of these systems are under extensive genetic control. Most of our presently available AEDs are metabolized by the cytochrome P450 (CYP) system. The CYP system consists of a number of dif- ferent enzymes, and the classification of these, adopted in 1996, was into CYP{num- ber}{letter}{number}*{number} groups (Nelson et al., 1996). The first number groups into families which have greater than 40% protein sequence homology, the subsequent letter into subfamilies that have greater than 55% homology, the sec- ond number into members of subfamilies that are encoded by a particular gene, and the number following the ‘*’ represents specific alleles of that gene. Four isoenzymes (CYP3A4, CYP2D6, CYP2C9 and CYP1A2) are known to be responsible for the metabolism of 95% of all drugs, and there are extensive pharmacogenetic poly- morphisms for each of the enzymes. Three isoenzymes (CYP2C9, CYP2C19 and CYP3A4) are of particular importance in relation to AED metabolism and interac- tions (Rendic and Di Carlo, 1997). Indeed the two enzymes that have received the most attention have been CYP2C9 and CYP2C19. CYP2C9 is the dominant enzyme in the metabolism of phenytoin, and the two alleles CYP2C9*2 and CYP2C9*3 have impaired enzymatic activity compared to CYP2C9*1 (Aithal et al., 1999); those with either a CYP2C9*2 or CYP2C9*3 allele need a phenytoin dose that is 30% lower than those who have only CYP2C9*1 (van der Weide et al., 2001). Due to impaired enzymatic activity, those patients with a CYP2C9*2 or CYP2C9*3 allele are more likely to experience metabolic interactions during combination therapy with phenytoin and an interacting drug (Meyer, 2000). CYP2C19 is also involved in the metabolism of phenytoin, but to a lesser degree and consequently CYP2C19 polymorphism has less of an effect on phenytoin metabolism and its propensity to interact with concomitant drugs. CYP2C19 is, however, the dominant enzyme in the

28 Matthew C. Walker et al.

metabolism of phenobarbitone, and CYP2C19 allelic variation has been associated with decreased metabolism and also an increased propensity for metabolic inter- actions. Decreased metabolism is especially common in the Japanese population where 8% of patients with epilepsy may be poor phenobarbitone metabolizers (Mamiya et al., 2000) and may be more prone to metabolic interactions.

Like CYP-mediated reactions, glucuronidation processes are susceptible to inhi- bition and induction.

Phase II metabolism is also subject to genetic variation. Uridine glucuronyl trans- ferases (UGTs) are a family of enzymes that catalyze the process of glucuronidation and comprise two distinct families, UGT1 and UGT2, with eight isoenzymes identi- fied in each family. The UGT1A4 isoenzyme plays an important role in the glu- curonidation of lamotrigine (Green et al., 1995), whereas the isoenzyme isoforms catalyzing the glucuronide conjugation of valproic acid have not yet been elucidated. Patients with Gilbert’s syndrome (unconjugated hyperbilirubinaemia due to a muta- tion in a gene coding for UGT) have over 30% lower clearances and higher half-lives for lamotrigine when compared to healthy volunteers (Posner et al., 1989). Certain drug interactions with lamotrigine can be explained by the glucuronidation pathway, such as the reduction of lamotrigine serum concentrations by oral contraceptives (Sabers et al., 2001) and the potential reductions in olanzipine glucuronidation by lamotrigine (Linnet, 2002). Such interactions are likely to be affected by polymor- phisms and mutations in the genes coding for UGT, although this remains to be tested.

Genotyping to determine drug metabolism probably has a limited role in epilepsy for two main reasons:

1 for many AEDs, there is not a clear relationship between plasma concentrations and efficacy/adverse events;

2 AEDsaretitratedupslowlyandconcomitantbloodlevelmonitoringoftengives an accurate idea if patients are slow or fast metabolizers.

This contrasts with the now commonly used screening of children for thiopurine S-methyltransferase deficiency before beginning mercaptopurine treatment for acute lymphoblastic leukemia (McLeod and Siva, 2002). In these cases the children are given acute courses of a drug whose efficacy and side-effect profile is closely related to plasma concentrations. An exceptional use for genotyping for drug metab- olism may come into use for AEDs which have potential metabolites that are toxic (see below), and geneotyping may prove useful in predicting drug–drug interactions.

Central pharmacokinetics

The point of action for AEDs is the brain, and so AEDs have to be able to cross the blood–brain barrier. Transport proteins regulate the flux of drugs across the

29 Pharmacogenetic aspects

blood–brain barrier. Many of these proteins belong to the ATP-binding cassette family of membrane transporters of which P-glycoprotein is the most extensively studied (Lee et al., 2001; Sisodiya, 2003). P-glycoprotein at the blood–brain barrier limits the accumulation of specific drugs in the central nervous system by trans- porting the drugs out of the brain. The role of such transporters in epilepsy remains uncertain (Sisodiya, 2003). This is partly because there is at present no consensus on which AEDs are transported by these proteins (see, for example Potschka et al., 2001 and Owen et al., 2001). Nevertheless, upregulation of these proteins is associ- ated with drug resistant epilepsy in both humans and animal models (Sisodiya, 2003). Furthermore, a specific SNP in the gene encoding P-glycoprotein, ABCB1, has a strong association with AED resistance (Siddiqui et al., 2003). This SNP is in a non-coding portion of the gene and thus its functional significance is uncertain – it is probable that it is associated with a separate functional SNP in an exon (Siddiqui et al., 2003). This raises a problem with the use of SNPs in order to deter- mine biological function, as they could be associated with SNPs elsewhere in the gene or even on other genes and thus unless a change of function of the gene prod- uct is demonstrated, such SNPs should only be used as biological markers as they may not be causal. The use of such markers for drug resistance could be useful for determining early referral for surgery, the spectrum of drug responsiveness or even the use of concomitant blockers of such transporters. In addition, the finding that carbamazepine may inhibit P-glycoprotein, albeit at high concentrations (Weiss et al., 2003), raises the possibility that certain AED interactions could be explained by competitive inhibition of these drug transporters. In such instances, polymor- phisms could determine the degree to which such interactions occur.

Pharmacodynamics

There is, at present, scant human evidence that genotype contributes to AED responsiveness, despite considerable evidence that receptor and channel subtypes determine drug pharmacodynamics. Genes that determine the type of epilepsy can influence drug pharmacodynamics by two specific mechanisms. First the epilepsy type and the pathophysiological substrate of the epilepsy could influence drug pharmacodynamics and secondly a genetic mutation could lead to both a channel that is ‘responsible’ for the epilepsy and also particularly sensitive/resistant to spe- cific drugs. Thus, the first of these influences can be illustrated by the idiopathic generalized epilepsies which are largely genetically determined. Despite the like- lihood that there are many genes determining the subtype and expression of these epilepsies, they are characterized by seizures with similar pathophysiological sub- strates. Thus absence seizures are generated within a recurrent loop between the thalamus and neocortex, and their generation is dependent upon oscillatory

30 Matthew C. Walker et al.

behavior mediated by gamma amino butyric acid (GABA)A receptors, GABAB receptors, T-type calcium channels and glutamate receptors (Crunelli and Leresche, 2002). One hypothesis is that hyperpolarization of the thalamocortical neurons in the thalamus mediated by GABAergic inhibition leads to activation of T-type calcium currents which open on neuronal depolarization, resulting in repetitive spiking that activates neurons in the neocortex which in turn stimulate the thalamic reticular nucleus leading to GABAergic inhibition of the thalamocortical (relay) neurons, and so the cycle continues (Danober et al., 1998; Huguenard, 1999). The pathophysiological substrates of absence seizures lead to specific pharmaco- dynamic actions that may largely be independent of the genetic defects underlying the generation of such seizures. Within this circuit, clonazepam preferentially inhibits the thalamic reticular neurons, perhaps due to the higher expression of 3- containing GABAA receptors (Browne et al., 2001). Ethosuximide, a drug whose main action may be on T-type calcium channels, has a specific action on absence seizures. Drugs that increase ambient GABA, such as tiagabine and vigabatrin, and GABAB receptor agonists can hyperpolarize thalamocortical neurons and so can have a pro-absence effect (Danober et al., 1998). Also certain other drugs such as carbamazepine and phenytoin can worsen absence seizures; the mechanism of this is unknown, but does not seem to be a class effect, as lamotrigine, a drug that also inhibits sodium channels (see below) has an antiabsence effect (Frank et al., 1999).

That genes that determine specific epilepsies could also influence drug respon- siveness has been well documented recently. Autosomal dominant frontal lobe epilepsy is an epilepsy that can result from a mutation in the gene for the 4 sub- unit of the nicotinic receptor. How this mutation results in the epilepsy remains a topic for speculation, but an interesting observation is that this mutation also ren- ders the receptor more sensitive to carbamazepine (Picard et al., 1999), and this tallies with clinical experience as carbamazepine is a very effective treatment in this disorder. A note of caution needs to be raised here: a mutation of a specific chan- nel does not necessarily mean that drugs acting at that channel are more likely to be effective. Thus benign neonatal convulsions result from mutations in KCNQ2 and KCNQ3 potassium channels (these channels make up the M potassium current – a potassium current that is ‘switched off’ by muscarinic receptor activation; Tatulian et al., 2001). A facile interpretation is that drugs that act at these potas- sium channels are likely to be most effective in this epilepsy, and such a drug exists; retigabine (Tatulian et al., 2001). Yet one could equally expect drugs that act at muscarinic receptors to be effective, and with further thought, and the realization that epilepsy is a network phenomenon that involves a multitude of receptors and channels, one could predict the efficacy of drugs acting at quite separate targets. In fact this epilepsy responds very well to a range of conventional AEDs. Nevertheless certain genetic defects could prevent the efficacy of certain drugs. An interesting

31 Pharmacogenetic aspects

finding is that of a mutation of the subunit of the GABAA receptors underlying absence epilepsy with febrile seizures in a large family (Wallace et al., 2001). This mutation, along with other mutations in the same subunit, possibly results in seizures by decreasing the function of GABAA receptors containing this subunit (Baulac et al., 2001; Bianchi et al., 2002). Yet this mutation also renders the recep- tors benzodiazepine insensitive, and thus possibly makes this a benzodiazepine- resistant epilepsy (Wallace et al., 2001).

Genetic differences could also affect the channels to which specific drugs are tar- geted, and may be independent of those genes that are contributing/determining the epilepsy. Voltage-dependent sodium channels and GABAA receptors are two of the main targets for presently available AEDs, and the effect of drugs on these tar- gets is subtype dependent. Since drug action is critically dependent on subunit composition, it is easy to appreciate how genetic polymorphisms could have a strong influence on drug effects. We will use these two targets as illustrations of how genetic differences can influence drug effects, and how those genes that deter- mine the epilepsy syndrome could similarly affect drug responsiveness.

Voltage-gated sodium channels are responsible for the rising phase of the action potential in excitable cells and membranes, and are thus critical for action poten- tial generation and propagation (Catterall, 2000). The sodium channel exists in three principle conformational states:

. 1 at hyperpolarized potentials the channel is in the resting closed state;

. 2 with depolarization the channels convert to an open state that conducts
sodium ions;

. 3 the channel then enters a closed, non-conducting, inactivated state, this inacti-
vation is removed by hyperpolarization.
In this manner, depolarization results in a transient inward sodium current that

rapidly inactivates.

The sodium channel consists of a 260-kDa subunit that forms the sodium selective pore. This subunit consists of four homologous domains (I–IV) that each consist of six -helical transmembrane segments (S1–6). The highly charged S4 segments are responsible for voltage-dependent activation. A ‘hinged lid’ con- sisting of the intracellular loop connecting domains III and IV that can only close following voltage-dependent activation provides the mechanism of inactivation (Catterall, 2000).

In the central nervous system, the subunit is associated with two auxiliary subunits ( 1 and 2) that influence the kinetics and voltage dependence of the gating. There are at least 10 different sodium channel isoforms (Nav1.1–1.9 and Nax). Five of these isoforms are present in the central nervous system – Nav1.1–1.3, Nav1.5 (in the limbic system) and Nav1.6; these isoforms have some functional differences

32 Matthew C. Walker et al.

that are of physiological importance. Certain receptor subtypes, such as Nav1.6 are more prone to late openings following a depolarization that can lead to persistent sodium currents that can contribute top burst firing. Sodium channels are additionally modulated by protein phosphorylation, which can affect the peak sodium current, and the speed and voltage dependence of channel inactivation (Catterall, 2000).

Many drugs including certain anesthetics and antiarrhythmics exert their thera- peutic effect by preferential binding to the inactivated state of the sodium channel (Catterall, 2000). This has two effects: first to shift the voltage dependence of inac- tivation towards the resting potential (i.e the channels become inactive at lower membrane potentials), and secondly to delay the return of the channel to the rest- ing, closed conformation following hyperpolarization. Phenytoin, lamotrigine and carbamazepine have a similar mode of action (Lang et al., 1993; Kuo, 1998). All bind in the inner pore of the sodium channel, and their binding is mutually exclu- sive (Kuo, 1998). There are, however, differences in the fashion in which drugs interact with adjacent amino acids that can partly explain drug specific effects (Ragsdale et al., 1996; Liu et al., 2003); AEDs perhaps have more complex interac- tions with surrounding amino acids than do local anesthetics (Liu et al., 2003), and will have their effects modified by a greater number of possible polymorphisms. Indeed, mutations of single amino acids affect the binding of individual drugs to different degrees, indicating that these drugs interact in an overlapping, but non- identical, manner with a common receptor site (Ragsdale et al., 1996). Sodium channels from patients with refractory temporal lobe epilepsy may be selectively resistant to carbamazepine (Remy et al., 2003).

There are other drugs such as valproate that inhibit rapid repetitive firing (McLean and Macdonald, 1986), but act at a different site from the site on which carbamazepine, lamotrigine and phenytoin act (Xie et al., 2001). Thus there could be single amino acid substitutions that would affect sodium channel inhibitors (but not necessarily all drugs acting on that channel), and also amino acid substi- tutions that could result in resistance to specific drugs.

GABAA receptors are the target for a number of AEDs since alterations in GABAA receptor-mediated transmission have been implicated in the pathogenesis of epilepsy. GABAA receptors are mainly expressed post-synaptically in the brain (pre-synaptic GABAA receptors have been described within the spinal cord). GABAA receptors are constructed from five of at least 16 subunits, grouped in seven classes: , , , , , and (Mehta and Ticku, 1999). This permits a vast number of putative recep- tor isoforms. The subunit composition determines the specific effects of allosteric modulators of GABAA receptors, such as neurosteroids, zinc and benzodiazepines (Mehta and Ticku, 1999). Importantly the subunit composition of GABAA recep- tors expressed in neurons can change during epileptogenesis, and these changes influence the pharmacodynamic response to drugs (Brooks et al., 1998). GABAA

33 Pharmacogenetic aspects

receptor activation results in the early rapid component of inhibitory transmission. Since GABAA receptors are permeable to chloride and, less so, bicarbonate, the effects of GABAA receptor activation on neuronal voltage are dependent on the chloride and bicarbonate concentration gradients across the membrane (Macdonald and Olsen, 1994). In neurons from adult animals, the extracellular chloride con- centration is higher than the intracellular concentration resulting in the equilib- rium potential of chloride being more negative than the resting potential. Thus GABAA receptor activation results in an influx of chloride and cellular hyperpolar- ization. This chloride gradient is maintained by a membrane potassium/chloride co-transporter, KCC2 (Rivera et al., 1999). Absence of this transporter in immature neurons results in a more positive reversal potential for chloride, and thus GABAA receptor activation in these neurons produces neuronal depolarization (Ben-Ari et al., 1994; Rivera et al., 1999). Under these circumstances GABAA receptors can mediate excitation rather than inhibition. Thus the expression of KCC2 could influence the response to drugs acting at GABAA receptors, and importantly the expression of KCC2 can be modified by epileptogenesis. Thus, polymorphisms in genes that do not directly code for the GABAA receptor could influence the phar- macodynamic response of drugs acting on this receptor.

Benzodiazepines are specific modulators of GABAA receptors and act at GABAA receptors that contain an 1, 2, 3 or 5 subunit in combination with a subunit (Mehta and Ticku, 1999). Drugs acting at the benzodiazepine site have different affinities for the different subunit-containing GABAA receptors, and this speci- ficity can affect pharmacodynamic response (McKernan et al., 2000). This is due perhaps to the varied distribution of these receptors in the brain. Thus the 1 subunit- containing receptors seem to have mainly a sedative effect, and are perhaps respon- sible for this side effect of benzodiazepines (McKernan et al., 2000). This may also explain why zolpidem, a drug that has great affinity for GABAA receptors contain- ing the 1 subunit has marked sedative effects and weak anticonvulsant efficacy (Crestani et al., 2000). More selective ligands could thus result in benzodiazepine agonists that have less sedative effect and greater anticonvulsant potential. Importantly, single amino acid substitutions rendering certain subunits insensitive to benzodiazepines can thus radically alter the profile of these drugs. Importantly, a mutation in the subunit has been found to underlie a specific epilepsy syn- drome in some families, and this mutation renders the GABAA receptors benzo- diazepine insensitive. It is likely that a range of polymorphisms in the GABAA receptor are likely to underlie the range of clinical responses to these drugs (adverse effects, efficacy, tolerance etc.). Even drugs that are less selective than benzodiazepines (e.g. barbiturates) still show some preference for certain GABAA receptor subtypes.

The manner by which polymorphisms and receptor expression can affect drug interactions is an unexplored area, but it is easy to speculate that certain

34 Matthew C. Walker et al.

pharmacodynamic interactions are receptor subtype dependent. For example, the enhancement of the action of tiagabine (increasing extracellular GABA) on GABAA receptors by a benzodiazepine would require the presence of benzodiazepine- sensitive GABAA receptors, and the extent of such an interaction could thus be deter- mined by polymorphisms and mutations in specific receptor subunits.

Adverse events

The spectrum of adverse events may also depend upon receptor and channel poly- morphisms, but most of these are likely to be dose related and inconsequential. There is unlikely to be a role for screening for these. There are, however, serious adverse events with AED use that may benefit from pharmacogenetic screening. Pharmacogenetic screening of serious adverse events that result from drug metabo- lites is potentially a powerful application. Felbamate has three primary metabo- lites, 2-hydroxy, p-hydroxy, and monocarbamate metabolites (Kapetanovic et al., 1998). The monocarbamate metabolite is eventually metabolized to a carboxylic acid (3-carbamoyl-2-phenylpropionic acid), which is the major metabolite of fel- bamate in humans (Kapetanovic et al., 1998). Metabolism of the monocarbamate metabolite can also result in the formation of a reactive aldehyde, atropaldehyde that could be responsible for aplastic anemia and hepatic damage associated with this drug. Enzymatic defects in the metabolism of the monocarbamate metabolite may result in the overproduction of atropaldehyde, or defects in the conjugation of atropaldehyde with glutathione (and thus detoxification) could lead to its accu- mulation; screening for these defects could result in identification of those who are susceptible to the serious adverse effects of felbamate, and could result in the wider use of a potentially very effective drug. Many adverse effects such as rash have an immunological basis, and these are frequently associated with human leukocyte antigen (HLA) type, providing a possible method of screening for other idiopathic adverse events. The association with HLA does not always indicate an immuno- logical basis as HLA-determining genes on chromosome 6 can be in linkage dis- equilibrium with other genes (i.e. the occurrence of a specific HLA type may increase the chance of a specific polymorphism in a neighboring but distinct gene) (see, for example Pirmohamed et al., 2001).

There could also be a place for screening for chronic adverse events. Reduced folate levels have been associated with chronic AED treatment. A possible conse- quence of this is hyperhomocysteinaemia. Hyperhomocysteinaemia is associated with vascular disease and so a prediction would be that AED therapy, through reducing folate levels, would increase homocysteine levels and result in an increase in cardiovascular disease. This could explain the increased incidence of cardio- vascular disease in patients with epilepsy. Such associations have been described,

35 Pharmacogenetic aspects

and indeed it has been noted that patients receiving phenytoin and carbamazepine, who are homozygous for the thermolabile genotype of methylenetetrahydrofolate reductase gene (MTHFR), are at significant risk of hyperhomocysteinaemia (Yoo and Hong, 1999). Similarly gum hypertrophy with phenytoin is a problem that may occur in up to 50% of the patients treated with phenytoin. In cats the main metabolite of phenytoin, p-hydroxyphenol-5-phenylhydantoin (p-HPPH), has been shown to induce gingival overgrowth (Hassell and Page, 1978), and it may be that those who produce higher concentrations of p-HPPH (i.e. the fast metabolizers) have a higher incidence of gum hypertrophy. Undoubtedly there are other genetic factors at play that may affect fibroblasts and gingival inflammation (Seymour et al., 1996).

Lastly there may be strong genetic determinants in AED teratogenicity. Defects in detoxification pathways such as epoxide hydrolase, which detoxifies epoxides, have been implicated in increasing the risk of fetal malformations. One of the oxi- dized products of phenytoin is an arene oxide (epoxide). It has been proposed that these arene oxide metabolites can covalently bind to cell macromolecules, resulting in cell death, hypersensitivity and even birth defects (Spielberg et al., 1981; Strickler et al., 1985). A defect in epoxide hydrolase has been proposed to increase the risk of fetal malformations (Strickler et al., 1985; Buehler et al., 1990). Furthermore developmental homeobox genes may play an important role, as it has been shown that certain mutations result in an increased chance of valproate-associated mal- formations in mice (Faiella et al., 2000).

The predisposition to the formation of toxic metabolites, and an enhanced sus- ceptibility to the adverse effects of these metabolites will undoubtedly lead to enhanced toxicity with specific drug combinations. Conversely, there may be cer- tain drug combinations that could be protective in that they may reduce the serum concentrations of responsible metabolites. Identification of relevant polymor- phisms may help tailor AED therapy and drug combinations in pregnant women.

Misconceptions about the use of genetic tests

We have shown above that genetic polymorphisms may have a profound effect on drug responsiveness and drug interactions. How useful will genetic testing be? The purpose of a diagnostic test is to provide increased certainty of the presence or absence of a disease. Pharmacogenetic tests are performed in an attempt to predict the therapeutic or adverse consequences of a drug in an unexposed individual. As such, they are screening tests not diagnostic tests and whilst screening tests may have health benefits, the harm that can result from inappropriate tests or their inappropriate interpretation is well documented (Sackett, 1991; Grimes and Schulz, 2002). Here we review the utility of genetic testing to predict drug response

36 Matthew C. Walker et al.

Table 3.1 Discriminative value of genetic test (T) for drug outcome (D)

Disease outcome

99 980

1000

98 980

T (T) T

1019 98981 100 000

Test

The background risk of SJS or TEN with lamotrigine is estimated at 2 in 10 000 (see text). D : presence; D : absence of SJS/TEN following 100000 hypothetical exposures. T refers to a or test result.

Sensitivity (probability of T in people with D ): 19/20 Specificity (probability of T in people without D ): 98 980/99 980 PPV (probability of D in people with T ): 19/1019 NPV (probability of of D in people with T ): 98 980/98 981

0.95
0.99
0.02
0.99 999

(pharmacogenetics) in the context of the intrinsic epidemiological constraints on genetic tests. We do not consider other important aspects of genetic testing includ- ing ethical, legal and social implications (Rothstein and Epps, 2001).

The principles of screening tests are best considered by means of an example. Consider an hypothetical genetic test for predicting the risk of Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) in response to lamotrigine. The precise risk of SJS/TEN following lamotrigine use is unknown. Such data are uncertain because of an inability to control for confounding, low risk and (rela- tively) small numbers of exposed individuals. Observational studies by prescrip- tion event monitoring in general practice, however, suggest an approximate risk of 0.1/1000 patient months of exposure, with most cases occurring in the first 2 months of use (Mackay et al., 1997; Rzany et al., 1999). Suppose we have a genetic test for identifying patients at risk of SJS/TEN with lamotrigine that has sensitivity 95% and specificity 99%. Intuitively one might consider this an ‘excellent’ test, but how will such a test perform in clinical practice?

The effectiveness of a screening test can be evaluated using a 2 2 table that relates test result to drug outcome. The ability of our hypothetical test to dis- criminate those at risk of SJS/TEN from those not at risk is illustrated in Table 3.1. Table 3.1 shows how the four indices of a test’s validity, sensitivity, specificity and positive and negative predictive value are calculated. For the clinician, who wishes to predict the probability of a patient developing SJS/TEN with lamotrigine, the key index is the positive predictive value (PPV – the probability of the disease given a positive test result). In the example cited (Table 3.1), it can be seen that although

37 Pharmacogenetic aspects

1 0.8 0.6 0.4 0.2 0

Positive test result

Negative test result

0 0.2 0.4 0.6 0.8 1 Pre-test probability (prior risk)

Figure 3.1 Influence of pre-test probability (prior risk) on the probability of a disease after a negative and positive screening test result (test sensitivity is 95% and specificity 99%)

the test has both high sensitivity (95%) and high specificity (99%), the predictive value of a positive test result (PPV) is only 2%. Although the probability of SJS/TEN following a positive test result has risen substantially from 2 in 10 000 to 2 in 100, it remains the case that 98% of patients testing positive will not develop SJS/TEN. This example illustrates how, in low prevalence settings, even good tests may have low predictive value. Thus, for a test with a sensitivity 95% and a specificity 99%, PPV only exceeds 90% when prior risk exceeds 1 in 11 (Figure 3.1). If prior risk is not considered when interpreting the result of the test, a posi- tive result might deny some epilepsy patients the opportunity of an appropriate treatment.

The above discussion illustrates how the probability of a specific drug outcome after a screening test is dependent on prior risk. Knowledge of prior risk is there- fore critical for interpreting the result of a screening test. Yet, even for serious adverse drug reactions (ADRs), an accurate estimate of prior risk may not be avail- able. Co-medication, co-morbidity, age, sex, weight, duration of treatment, renal and liver function, under- and over-reporting as well as misdiagnosis all confound the accurate assessment of prior risk. Moreover, in routine clinical practice the dependency of test performance on prior risk is frequently under-appreciated, resulting in badly interpreted test results (Johnson et al., 2001). Without accurate prevalence data, the clinical utility of a predictive genetic test will have to be deter- mined by prospective randomized controlled trial. Yet methodology for the evalu- ation of new diagnostic techniques remains poorly defined, it is less advanced than that relating to the assessment of new therapies and there are no formal standards for the acceptance of new diagnostic procedures (Knottnerus, 2002).

Post-test probability

38 Matthew C. Walker et al.

The predictive value of a pharmacogenetic test can also be viewed from a genetic epidemiological perspective. Whilst some have argued that genetic testing will be widely used to predict a person’s probability of developing a disease, others have pointed to limitations based on the low magnitude of relative risk and incomplete penetrance associated with various genotypes in the general population (Holtzman and Marteau, 2000; Vineis et al., 2001). Using simple epidemiological principles, Holtzman and Marteau (2000) demonstrated that under most conditions, com- mon genotypes associated with common human diseases will have little predictive power. We can apply similar principles when considering the potential for phar- macogenetic tests to yield clinically useful predictive value. The PPV of a test for a genetic susceptibility factor (this might denote the alleles that a person possesses at a single gene locus on homologous chromosomes or a complex genomic profile) is a function of the frequency of the genetic factor in the population, its relative risk and the prevalence of the drug outcome (Lilienfeld and Lilienfeld, 1980; Khoury et al., 1985; Holtzman and Marteau, 2000). This can be appreciated if we consider a gene test (or genomic profile) for a drug response (adverse or therapeutic) with a prior risks prevalence of 1 in 100 and 1 in 10 (Table 3.2). Thus the drug response of interest may occur in persons with a specific genotype (G) as well as in persons without that genotype (1 G). Individuals without the specific genotype may still experience the drug response of interest due to locus and allelic heterogeneity, environmental variation and/or stochastic factors.

If r is the risk associated with exposure to genotype, and r the risk associated with non-exposure, then the relative risk for the drug response conferred by the susceptibility phenotype (R) r /r . The prevalence of the drug response (D) will include cases that arise from exposure to the genotype (G r ) as well as cases that arise from unrelated mechanisms ((1 G) r ). Substituting PPV for r , this can be re-written as:

D G PPV (1 G) PPV/R
which, solved for
PPV DR 100/G(R 1) 1 (expressedasa%)

The PPV of a test based on a genetic susceptibility factor can now be estimated across a range of D, R and G values. Table 3.2 lists PPV across a range of G and R values for a drug response with a prevalence of 1 in 100 and 1 in 10. For an outcome with prior risk of 1 in 100, it can be seen that only when the frequency of the susceptibility fac- tor is low, and the genotype relative risk is high, will PPV be high. Whilst this ‘Mendelian’ situation may account for some drug responses, it seems just as likely that genotypes will confer lower relative risk for a specific drug response and thus lower PPV. Where the prevalence of the outcome of interest is lower (as, for example

39 Pharmacogenetic aspects

Table 3.2 PPV of a screening test for a genetic susceptibility factor (genotype) for a drug response (adverse or therapeutic) with a prior risk of 1 in 100 and 1 in 10

Genotype relative risk
frequency 2 5 10 20 50 2 5 10 20 50

Genotype

0.001 2.0 0.01 2.0 0.1 1.8 0.3 1.5

5.0 9.9 4.8 9.2 3.6 5.3 2.3 2.7

19.6 47.7 16.8 33.6 6.9 8.5 3.0 3.2

PPV: 1 in 100 (%)

PPV: 1 in 10 (%)

20.0 50.0 99.1
19.8 48.1 91.7
18.2 35.7 52.6 69.0 84.7 15.4 22.7 27.0 29.8 31.8

in the risk of SJS/TEN with lamotrigine), then PPV will be even lower. Of course, where the outcome is more prevalent (e.g., 1 in 10 exposures, Table 3.2), then PPV will be higher. In this situation, whilst there may be no point in testing people if the prevalent outcome is a beneficial drug response, there may be value in identifying people without genetic susceptibility to a prevalent harmful response.

What is clear, at the moment however, is that there is a lack of data on which to base predictions regarding the potential utility of pharmacogenetic testing. Epidemio- logical considerations such as those above highlight that if genetic tests for drug outcomes (therapeutic or adverse) are to become widely used, they will need to be validated, easy to use, unambiguous, and provide a significant improvement over current clinical practice. Physicians are used to working within established risk scenarios, and may not adapt easily to genetically altered benefit–risk trade-offs. Clinical and cost effectiveness of pharmacogenetic tests may need to be established in prospective randomized trials and their use may require new professional stan- dards of testing and test interpretation. The degree to which pharmacogenetic tests become integrated into routine clinical practice will be determined as much by epidemiological constraints as the important legal, ethical, social and commercial aspects of genetic testing.

REFERENCES

Aithal GP, Day CP, Kesteven PJL, et al. Association of polymorphisms in the cytochrome P450 CYP2C9 with warfarin dose requirement and risk of bleeding complications. Lancet 1999; 353: 717–719.

Baulac S, Huberfeld G, Gourfinkel-An I, et al. First genetic evidence of GABA(A) receptor dys- function in epilepsy: a mutation in the gamma2-subunit gene. Nat Genet 2001; 28: 46–48.

40 Matthew C. Walker et al.

Ben-Ari Y, Tseeb V, Raggozzino D, et al. Gamma-aminobutyric acid (GABA): a fast excitatory transmitter which may regulate the development of hippocampal neurons in early postnatal life. Prog Brain Res 1994; 102: 261–273.

Bianchi MT, Song L, Zhang H, et al. Two different mechanisms of disinhibition produced by GABAA receptor mutations linked to epilepsy in humans. J Neurosci 2002; 22: 5321–5327. Brooks KA, Shumate MD, Jin H, et al. Selective changes in single cell GABA(A) receptor subunit

expression and function in temporal lobe epilepsy. Nat Med 1998; 4: 1166–1172.
Browne SH, Kang J, Akk G, et al. Kinetic and pharmacological properties of GABAA receptors in single thalamic neurons and GABAA subunit expression. J Neurophysiol 2001; 86: 2312–2322. Buehler BA, Delimont D, van Waes M, et al. Prenatal prediction of risk of the fetal hydantoin

syndrome. New Engl J Med 1990; 322: 1567–1572.
Catterall WA. From ionic currents to molecular mechanisms: the structure and function of

voltage-gated sodium channels. Neuron 2000; 26: 13–25.
Crestani F, Martin JR, Mohler H, et al. Mechanism of action of the hypnotic zolpidem in vivo.

Br J Pharmacol 2000; 131: 1251–1254.
Crunelli V, Leresche N. Childhood absence epilepsy: genes, channels, neurons and networks. Nat

Rev Neurosci 2002; 3: 371–382.
Danober L, Deransart C, Depaulis A, et al. Pathophysiological mechanisms of genetic absence

epilepsy in the rat. Prog Neurobiol 1998; 55: 27–57. FaiellaA,WernigM,ConsalezGG,etal.Amousemodelforvalproateteratogenicity:parentaleffects,

homeotic transformations, and altered HOX expression. Hum Mol Genet 2000; 9: 227–236. Frank LM, Enlow T, Holmes GL, et al. Lamictal (lamotrigine) monotherapy for typical absence

seizures in children. Epilepsia 1999; 40: 973–979.
Green MD, Bishop WP, Tephley TR. Expressed human UGT1.4 protein catalyzes the formation

of quaternary ammonium-linked glucuronides. Drug Metab Disp 1995; 23: 299–302.
Grimes DA, Schulz KF. Uses and abuses of screening tests. Lancet 2002; 359: 881–884.
Hassell TM, Page RC. The major metabolite of phenytoin (Dilantin) induces gingival over-

growth in cats. J Periodont Res 1978; 13: 280–282.
Holtzman NA, Marteau TM. Will genetics revolutionize medicine? New Engl J Med 2000; 343:

141–144.
Huguenard JR. Neuronal circuitry of thalamocortical epilepsy and mechanisms of antiabsence

drug action. Adv Neurol 1999; 79: 991–999.
International Human Genome Sequencing Consortium. Initial sequencing and analysis of the

human genome. Nature 2001; 409(6822): 860–921.
International SNP Map Working Group. A map of human genome sequence variation contain-

ing 1.42 million single nucleotide polymorphisms. Nature 2001; 409: 928–933.
Johnson MR, Good CD, Penny WD, et al. Playing the odds in clinical decision making: lessons from berry aneurysms undetected by magnetic resonance angiography. Br Med J 2001; 322: 1347–1349. Kapetanovic IM, Torchin CD, Thompson CD, et al. Potentially reactive cyclic carbamate metabo- lite of the antiepileptic drug felbamate produced by human liver tissue in vitro. Drug Metab

Dispos 1998; 26: 1089–1095.
Khoury MJ, Newill CA, Chase GA. Epidemiological evaluation of screening for risk factors:

application to genetic screening. Am J Public Health 1985; 75: 1204–1208.

41 Pharmacogenetic aspects

Knottnerus JA, van Weel C, Muris JWM. Evaluation of diagnostic procedures. Br Med J 2002; 324: 477–480.

Kuo CC. A common anticonvulsant binding site for phenytoin, carbamazepine, and lamotrigine in neuronal Na channels. Mol Pharmacol 1998; 54: 712–721.

Kutt H, Wolk M, Scherman R, et al. Insufficient parahydroxylation as a cause of diphenylhydan- toin toxicity. Neurology 1964; 14: 542–548.

Lang DG, Wang CM, Cooper BR. Lamotrigine, phenytoin and carbamazepine interactions on the sodium current present in N4TG1 mouse neuroblastoma cells. J Pharmacol Exp Ther 1993; 266: 829–835.

Lee G, Dallas S, Hong M, et al. Drug transporters in the central nervous system: brain barriers and brain parenchyma considerations. Pharmacol Rev 2001; 53: 569–596.

Lilienfeld AM, Lilienfeld DE. Foundations of Epidemiology, 2nd edn. New York: Oxford University Press, 1980.

Linnet K. Glucuronidation of olanzapine by cDNA-expressed human UDP-glucuronosyltrans- ferases and human liver microsomes. Hum Psychopharmacol 2002; 17: 233–238.

Liu G, Yarov-Yarovoy V, Nobbs M, et al. Differential interactions of lamotrigine and related drugs with transmembrane segment IVS6 of voltage-gated sodium channels. Neuropharmacology 2003; 44: 413–422.

Macdonald RL, Olsen RW. GABAA receptor channels. Ann Rev Neurosci 1994; 17: 569–602. Mackay FJ, Wilton LV, Pearce GL, et al. Safety of long term lamotrigine in epilepsy. Epilepsia

1997; 38: 881–886.
Mamiya K, Hadama A, Yukawa E, et al. CYP2C19 polymorphism effect on phenobarbitone.

Pharmacokinetics in Japanese patients with epilepsy: analysis by population pharmaco-

kinetics. Eur J Clin Pharmacol 2000; 55: 821–825.
McKernan RM, Rosahl TW, Reynolds DS, et al. Sedative but not anxiolytic properties of benzo-

diazepines are mediated by the GABA(A) receptor alpha1 subtype. Nat Neurosci 2000; 3:

587–592.
McLean MJ, Macdonald RL. Sodium valproate, but not ethosuximide, produces use- and

voltage-dependent limitation of high frequency repetitive firing of action potentials of mouse

central neurons in cell culture. J Pharmacol Exp Ther 1986; 237: 1001–1011.
McLeod HL, Siva C. The thiopurine S-methyltransferase gene locus – implications for clinical

pharmacogenomics. Pharmacogenomics 2002; 3: 89–98.
Mehta AK, Ticku MK. An update on GABAA receptors. Brain Res Rev 1999; 29: 196–217.
Meyer UA. Pharmacogenetics and adverse drug reactions. Lancet 2000; 356: 1667–1671.
Nelson DR, Koymans L, Kamataki T, et al. P450 superfamily: update on new sequences, gene

mapping, accession numbers and nomenclature. Pharmacogenetics 1996; 6: 1–42.
Owen A, Pirmohamed M, Tettey JN, et al. Carbamazepine is not a substrate for P-glycoprotein.

Br J Clin Pharmacol 2001; 51: 345–349.
Picard F, Bertrand S, Steinlein OK, et al. Mutated nicotinic receptors responsible for autosomal

dominant nocturnal frontal lobe epilepsy are more sensitive to carbamazepine. Epilepsia 1999;

40: 1198–1209.
Pirmohamed M, Lin K, Chadwick D, et al. TNFalpha promoter region gene polymorphisms in

carbamazepine-hypersensitive patients. Neurology 2001; 56: 890–896.

42 Matthew C. Walker et al.

Posner J, Cohen AF, Land G, et al. The pharmacokinetics of lamotrigine (BW430C) in healthy subjects with unconjugated hyperbilirubinaemia (Gilbert’s syndrome). Br J Clin Pharmacol 1989; 28: 117–120.

Potschka H, Fedrowitz M, Loscher W. P-glycoprotein and multidrug resistance-associated pro- tein are involved in the regulation of extracellular levels of the major antiepileptic drug carba- mazepine in the brain. Neuroreport 2001; 12: 3557–3560.

Ragsdale DS, McPhee JC, Scheuer T, et al. Common molecular determinants of local anesthetic, antiarrhythmic, and anticonvulsant block of voltage-gated Na channels. Proc Natl Acad Sci USA 1996; 93: 9270–9275.

Remy S, Gabriel S, Urban BW, et al. A novel mechanism underlying drug resistance in chronic epilepsy. Ann Neurol 2003; 53: 469–479.

Rendic S, Di Carlo FJ. Human cytochrome P450 enzymes: a status report summarizing their reactions, substrates, inducers, and inhibitors. Drug Metab Rev 1997; 29: 413–580.

Rivera C, Voipio J, Payne JA, et al. The K /Cl co-transporter KCC2 renders GABA hyperpolar- izing during neuronal maturation. Nature 1999; 397: 251–255.

Rothstein MA, Epps PG. Ethical and legal implications of pharmacogenomics. Nat Rev Genet 2001; 2: 228–229.

Rzany B, Correia O, Kelly JP, et al. Risk of Stevens–Johnson syndrome and toxic epidermal necrolysis during first weeks of antiepileptic therapy: a case control study. Lancet 1999; 353: 2190–2194.

Sabers A, Buchholt JM, Uldall P, et al. Lamotrigine plasma levels reduced by oral contraceptives. Epilepsy Res 2001; 47: 151–154.

Sackett DL. Clinical Epidemiology: A Basic Science for Clinical Medicine. Boston: Little Brown, 1991. Salisbury BA, Pungliya M, Choi JY, et al. SNP and haplotype variation in the human genome.

Mutat Res 2003; 526: 53–61.
Seymour RA, Thomason JM, Ellis JS. The pathogenesis of drug-induced gingival overgrowth.

J Clin Periodontol 1996; 23: 165–175.
Siddiqui A, Kerb R, Weale ME, et al. Association of multidrug resistance in epilepsy with a poly-

morphism in the drug-transporter gene ABCB1. New Engl J Med 2003; 348: 1442–1448. Sisodiya SM. Mechanisms of antiepileptic drug resistance. Curr Opin Neurol 2003; 16: 197–201. Spielberg SP, Gordon GB, Blake DA, et al. Anticonvulsant toxicity in vitro: possible role of arene

oxides. J Pharmacol Exp Ther 1981; 217: 386–389.
Streetman DS, Bertino Jr JS, Nafziger AN. Phenotyping of drug-metabolizing enzymes in adults:

a review of in-vivo cytochrome P450 phenotyping probes. Pharmacogenetics 2000; 10:

187–216.
Strickler SM, Dansky LV, Miller MA, et al. Genetic predisposition to phenytoin-induced birth

defects. Lancet 1985; 2: 746–749.
Tatulian L, Delmas P, Abogadie FC, et al. Activation of expressed KCNQ potassium currents and

native neuronal M-type potassium currents by the anti-convulsant drug retigabine. J Neurosci

2001; 21(15): 5535–5545.
van der Weide J, Steijns LS, van Weelden MJ, et al. The effect of genetic polymorphism of

cytochrome P450 CYP2C9 on phenytoin dose requirement. Pharmacogenetics 2001; 11: 287–291.

43 Pharmacogenetic aspects

Vasko MR, Bell RD, Daly DD, et al. Inheritance of phenytoin hypometabolism: a kinetic study of one family. Clin Pharmacol Ther 1980; 27: 96–103.

Vermeij P, Ferrari MD, Buruma OJ, et al. Inheritance of poor phenytoin parahydroxylation capacity in a Dutch family. Clin Pharmacol Ther 1988; 44: 588–593.

Vineis P, Schulte P, McMichael AJ. Misconceptions about the use of genetic tests in populations. Lancet 2001; 357: 709–712.

Wallace RH, Marini C, Petrou S, et al. Mutant GABA(A) receptor gamma2-subunit in childhood absence epilepsy and febrile seizures. Nat Genet 2001; 28: 49–52.

Weiss J, Kerpen CJ, Lindenmaier H, et al. Interaction of antiepileptic drugs with human P-glyco- protein in vitro. J Pharmacol Exp Ther 2003; 307: 262–267.

Xie X, Dale TJ, John VH, et al. Electrophysiological and pharmacological properties of the human brain type IIA Na channel expressed in a stable mammalian cell line. Pflügers Arch 2001; 441: 425–433.

Yoo JH, Hong SB. A common mutation in the methylenetetrahydrofolate reductase gene is a determinant of hyperhomocysteinemia in epileptic patients receiving anticonvulsants. Metabolism 1999; 48: 1047–1051.

Part II

Pharmacokinetic interactions

Pharmacokinetic principles and mechanisms of drug interactions

Philip N. Patsalos

Pharmacology and Therapeutics Unit, Department of Clinical and Experimental Epilepsy, Institute of Neurology, London, UK
The National Society for Epilepsy, Chalfont St Peter, UK

Introduction

In recent years, many of the fundamental principles and concepts of pharmacoki- netics have emerged from studies with antiepileptic drugs (AEDs). Pharmacoki- netics describes how a drug is absorbed, distributed, metabolized, and ultimately excreted from the body. These characteristics will determine not only the ease of clinical use of the drug (e.g. how it is prescribed) and whether or not a patient will comply with its prescription, but also the pharmacokinetics of a drug has a direct impact on a drug’s efficacy. During combination therapy with AEDs and indeed with AEDs and other drugs, there is potential for interference in pharmacokinetic processes and these interactions can be of major clinical significance. In this chap- ter, we review the various pharmacokinetic principles that are important to drug interactions and relate these to the major mechanisms of drug interactions.

Mechanisms of drug interactions

There are two basic types of drug interaction, pharmacokinetic and pharmacody- namic. Pharmacokinetic interactions are associated with changes in drug disposi- tion, which are readily measured in that changes in drug concentrations in plasma occur. These interactions, which in fact can be associated with a change in plasma concentration of either the drug or its metabolite(s) or both, involve a change in the absorption, distribution, or elimination of the affected drug and account for most known interactions (Patsalos and Perucca, 2003a). Pharmacodynamic interactions are also important but are less well recognized and occur between drugs that have similar or opposing pharmacological mechanisms of action. These interactions take place at the cellular level where drugs act, leading to additive, supra-additive, or infra-additive effects in relation to a therapeutic

48 Philip N. Patsalos

response or drug toxicity. Pharmacodynamic interactions are not associated with any change in the plasma concentration of either drug and are reviewed in detail in Chapter 9.

As pharmacokinetic interactions can occur during any stage of drug disposition (i.e. during absorption, distribution, metabolism, or elimination) these stages are discussed in greater detail below.

Absorption

Absorption is the entry of drug molecules into the systemic circulation via the mucous membranes of the gut or lungs, via the skin, or from the site of an injection. Although drug interactions with AEDs are rare during absorption, such interac- tions can be important in some cases. For example, when phenytoin is ingested with certain nasogastric feeds, it is thought to bind to constituents of the feeding formu- las to form insoluble complexes that cannot be absorbed (Bauer, 1982; Hatton, 1984; Worden et al., 1984). Therefore, phenytoin absorption is impaired. Another example is that of antacids which have been shown to reduce the absorption of some AEDs (e.g. phenytoin, phenobarbitone, carbamazepine, and gabapentin) by decreasing the acidity of the stomach (Patsalos and Perucca, 2003b).

Another, useful, interaction is that with activated charcoal which both impairs drug absorption and adsorbs drug secreted into the intestine. This interaction is exploited clinically to hasten the elimination of phenobarbitone, phenytoin, and carbamazepine in overdose patients (Neuvonen et al., 1978; Neuvonen and Elonen, 1980; Mauro et al., 1987; Weichbrodt and Elliot, 1987).

In recent years, evidence has accumulated that transporters, particularly P-glycoprotein, may play an important role in the gastrointestinal absorption of many drugs (Lin and Yamazaki, 2003), including digoxin (Hoffmeyer et al., 2000) and cyclosporine (Fricker et al., 1996; Lown et al., 1997). Whether P-glycoprotein contributes to the gastrointestinal absorption of AEDs is unknown. As the dis- tribution of P-glycoprotein varies significantly across the gastrointestinal tract, its role and contribution to drug absorption may vary for different drugs (Cox et al., 2002). Furthermore, the expression of P-glycoprotein in many tissues, including the gut, is subject to inhibition and induction by co-administered drugs, and many inhibitors and inducers of the cytochrome P450 (CYP) isoenzyme CYP3A4 may inhibit or induce P-glycoprotein activity (Wacher et al., 1995; Jette et al., 1996; Schuetz et al., 1996; Verschraagen et al., 1999). Therefore, overall, based on these observations, it cannot be excluded that some AED interactions currently ascribed to other mechanisms could in fact be mediated by modulation of P-glycoprotein function at the level of drug absorption or distribution. This possibility needs to be investigated.

49 Pharmacokinetic principles and mechanisms of drug interactions

Distribution

Distribution is the movement of drug molecules between the various water, lipid, and protein compartments in the body, including the movement of drugs to their sites of action, metabolism, and elimination. Interactions involving the distribu- tion of drugs are difficult to ascertain. For example, during combination therapy with vigabatrin and phenytoin, phenytoin plasma concentrations are reduced by approximately 30%. Although the mechanism of this interaction is unknown, it is thought to involve an effect on phenytoin distribution (Tonini et al., 1992).

Drug distribution is affected by protein binding in the circulation and the pri- mary proteins to which drugs bind are albumin and -glycoprotein, with albumin being by far the most important in relation to AEDs. Since the non-protein-bound drug concentration is that that is available for distribution in the body in general, and in relation to AEDs for distribution into the brain, and is pharmacologically active, plasma protein binding is important. Therefore, interactions involving competition between two drugs for plasma-protein-binding sites may affect drug distribution. However, these interactions are only important for drugs which are highly protein bound ( 90%), and among AEDs, only phenytoin, valproic acid, diazepam, and tiagabine have this characteristic (Perucca, 2001; Table 4.1).

Competition of drugs for albumin binding sites depends on both the affinity and the concentration of the two drugs. Drugs with lower affinity and lower con- centration will be displaced. The most commonly occurring plasma-protein- binding displacement interaction involving AEDs is the displacement of phenytoin by valproic acid (Patsalos and Lascelles, 1977a; Perucca et al., 1980). As the free frac- tion of phenytoin increases, total systemic clearance also increases, leading to a decline in total phenytoin concentration. Unbound (pharmacologically active) drug concentrations are dependent on drug dose and hepatic intrinsic clearance. Therefore, although at steady state a displacement interaction may transiently increase the unbound concentration of phenytoin, the concentration should return to its pre-interaction value, assuming there has not been any alteration in hepatic intrin- sic clearance (e.g. due to concurrent inhibition). Thus, although typically this interaction results in a fall in total phenytoin concentration while the concentra- tion of free, pharmacologically active, phenytoin is usually unaltered (Tsanaclis et al., 1984), in some patients a modest rise in free phenytoin concentration may actually be seen, due to a concomitant inhibition of phenytoin metabolism by val- proic acid (Patsalos and Lascelles, 1977b). Awareness of this interaction is impor- tant for interpretation of plasma drug concentration measurements since in this setting the “therapeutic” range of total plasma phenytoin concentrations is shifted towards lower values and therapeutic and toxic effects will occur at total drug concentrations lower than usual. Patient management may best be guided by monitoring free unbound phenytoin concentrations (Patsalos, 2001, 2002).

50 Philip N. Patsalos

Table 4.1 Some pharmacokinetic characteristics the various AEDs

Undergoes metabolic Undergoes renal Elimination

AED % bound transformation elimination half-life (h)a

Carbamazepine 75 Yes No Clobazam 85 Yes No Clonazepam 85 Yes No Diazepam 98 Yes No Ethosuximide 0 Yes No Felbamate 25 Yes Yes Gabapentin 0 No Yes Lamotrigine 56 Yesd No Levetiracetam 0 Yese Yes Oxcarbazepineb 40 Yes Yes Phenobarbitone 50 Yes Yes Phenytoinc 90 Yes No Primidone 25 Yes Yes Tiagabine 98 Yes No Topiramate 15 Yes Yes Valproic acid 90 Yes No Vigabatrin 0 No Yes Zonisamide 60 Yes Yes

16–24 10–58 19–40 24–48 40–60 13–23 5–9 22–38 6–8

5–30 80–100 7–42 8–12

5–8 19–25 8–18

5–7

57–68

aValues relate to patients co-administered with non-interacting drugs. bRefers to the mono-hydroxy metabolite of oxcarbazepine.
cDose or plasma concentration dependent.
dRefers to glucuronide metabolite of lamotrigine.

eMetabolism is non-hepatic.

Tolbutamide and phenylbutazone also interact with phenytoin by simulta- neously displacing phenytoin from its protein-binding site and inhibiting its metab- olism (Tassaneeyakul et al., 1992). Thus, the same precautions described above for valproic acid would also apply.

AEDs that are not protein bound (Table 4.1; ethosuximide, gabapentin, leve- tiracetam, and vigabatrin) would not be susceptible to protein-binding displace- ment interactions.

Distribution of AEDs from the blood compartment to the brain is very neces- sary for a successful therapeutic outcome. There is evidence that the efflux of some AEDs, including carbamazepine, felbamate, lamotrigine, phenobarbitone, and phenytoin, across the blood–brain barrier is mediated by p-glycoprotein (Potschka and Loscher, 2001; Potschka et al., 2001, 2002; Rizzi et al., 2002). Furthermore, p-glycoprotein overexpression in brain tissue may limit the penetration of AEDs to

51 Pharmacokinetic principles and mechanisms of drug interactions

their sites of action and may be a mechanism of pharmacoresistance in epilepsy (Sisodiya, 2003). Therefore, the possibility exists that AEDs may compete for trans- port across the blood-brain barrier via p-glycoprotein mechanisms.

Metabolism

Metabolism is the most important mechanism of elimination and accounts for the majority of clinically relevant drug interactions with AEDs. By far the most impor- tant system for AED metabolism is that involving the CYP system (e.g. carba- mazepine, phenobarbitone, phenytoin, tiagabine, topiramate, zonisamide, and felbamate). However, metabolic pathways such as conjugation involving uridine glucuronyl transferases (UGTs) (e.g. lamotrigine and valproic acid) and -oxidation (e.g. valproic acid) are also important.

CYP enzymes are a major component of the mixed function oxidase system that is located in the smooth endoplasmic reticulum of the cells of almost all tissues. The highest concentrations of CYP enzymes are found in the liver and four of these isoenzymes (CYP3A4, 50%; CYP2D6, 25%; CYP2C9, 15%; CYP1A2, 5%) are known to be responsible for the metabolism of 95% of all drugs (Spatzenegger and Jaeger, 1995). Furthermore, 50–70% of all drugs might be substrates for CYP3A4 and three isoenzymes (CYP3A4, CYP2C9 and CYP2C19) are of particular impor- tance in relation to AED interactions (Rendic and Di Carlo, 1997). CYP3A4 and CYP2C9, which are responsible for the metabolism of carbamazepine and pheny- toin respectively, are susceptible to induction and inhibition by many compounds and carbamazepine is capable of inducing its own metabolism (autoinduction) via its action on the CYP3A4 isoenzyme. If two drugs are metabolized by, or act upon, the same isoform of CYP, then drug interactions are more likely. Phenobarbitone, primidone, phenytoin, and carbamazepine are inducers of CYP isoenzymes, whereas valproate is an inhibitor (Mather and Levy, 2000).

The UGT family of enzymes are involved in the catalysis of glucuronidation processes and comprise two distinct families, UGT1 and UGT2. To date eight isoenzymes have been identified in each family. The glucuronidation of lamotrig- ine is by the UGT1A4 isoenzyme, whereas the isoenzyme isoform catalysing the glucuronide conjugation of valproic acid has not yet been identified (Green et al., 1995). Glucuronidation processes, just like those mediated by CYPs, are susceptible to inhibition and induction.

Of all AEDs, phenytoin has the greatest propensity to interact. Phenytoin binds loosely to CYP isoenzymes and consequently it is easily displaced from its binding sites by other drugs. Consequently, its metabolism is readily inhibited. Furthermore, the fact that the metabolism of phenytoin is saturable makes phenytoin particu- larly susceptible to problematic interactions. As the metabolism of phenytoin is

52 Philip N. Patsalos

primarily via the isoenzyme CYP2C9 (responsible for approximately 80% of the metabolism of phenytoin) whilst the isoenzyme CYP2C9 contribution is limited (responsible for the remaining 20%) the clinical significance of an interaction will very much depend on which isoenzyme is involved. Thus, amiodarone, which interacts with CYP2C9, will have a greater effect on the plasma concentration of phenytoin compared with cimetidine, which interacts with CYP2C19.

By far the most important pharmacokinetic interactions with AEDs are those which are related to induction or inhibition of drug metabolism (Anderson, 1998; Patsalos and Perucca, 2003a). Enzyme inhibition is the phenomenon by which a drug or its metabolite(s) blocks the activity of one or more drug-metabolizing enzymes resulting in a decrease in the rate of metabolism of the affected drug. This, in turn, will lead to increased plasma concentrations of the affected drug and, possibly, clinical toxicity. Inhibition is usually competitive in nature and dose dependent, and tends to begin as soon as sufficient concentrations of the inhibitor are achieved, with significant inhibition being often observed within 24h after addition of the inhibitor (Anderson, 1998). However, the time scale of the maximal pharmacological potentiation consequent to an inhibitory interaction depends on the elimination half-life of the affected drug with potentiation of drug activity occurring more quickly if the drug has a short half-life. As a rule, a new steady-state plasma concentration will be achieved at a time that is equivalent to five half-life values of the affected drug (Figure 4.1). For example, lamotrigine has a half-life

Inhibition

t1⁄2 4 h Steady-state

achieved at 20 h after inhibition

t1⁄2 2 h

Steady-state achieved at 10h after inhibition

0 4 8 12 16 20 24 28 32 36 40

Time (h)

Figure 4.1 A schematic showing how the elapsed time after an inhibitory interaction and maximal pharmacological potentiation is dependent on the elimination half-life (t1/2) of a drug. A new steady-state plasma concentration is typically achieved at a time representing five t1/2 values

Plasma drug concentration

53 Pharmacokinetic principles and mechanisms of drug interactions

value of approximately 1.5 days, and therefore its maximal pharmacological poten- tiation occurs 7.5 days later (Table 4.1). In contrast, the maximal pharmacological potentiation of phenobarbitone will occur 20 days later because its half-life is longer (approximately 4 days). If drug interactions result in an increased plasma concentration of a drug or its active metabolite, then the patient may experience toxicity and side effects, in which case it may be necessary to reduce the dose of the affected drug. However, in some patients, an increase in plasma drug concentration may actually enhance the therapeutic response, particularly if the concentration was previously sub-therapeutic. An extended half-life may also mean that the fre- quency of dosing can be reduced, which may actually help to improve compliance.

Enzyme induction, which is the consequence of an increase in the synthesis of CYP isoenzymes in the liver and in other tissues resulting in an increase in enzyme activity, becomes apparent more slowly than that of inhibition (Perucca et al., 1984; Su et al., 1998). The elevated enzyme activity, in turn, results in an increase in the rate of metabolism of drugs, which are substrates of those isoenzymes, leading to a decrease in plasma concentration of the affected drug. In this setting the pharma- cological effect of the drug will be reduced. However, if the affected drug has a phar- macologically active metabolite (e.g. the epoxide of carbamazepine), induction can result in increased metabolite concentration and seizure control may continue to be effective, but the possibility of an increase in drug toxicity is also greater. A further example involves the induction of disopyramide and amiodarone by enzyme- inducing AEDs whereby formation of an active metabolite complicates dosage requirements after induction has occurred (Aitio et al., 1981; Nolan et al., 1990). As enzyme induction requires synthesis of new enzymes, the time course of induction is dependent on the rate of enzyme synthesis and degradation and the time to reach steady-state concentrations of the inducing drug. Thus, the time course of induc- tion is usually dose dependent and gradual (Perucca, 1987; Patsalos et al., 1988).

It should be remembered that both enzyme induction and enzyme inhibition are reversible processes and that upon the removal of an interacting drug, drug dosage re-adjustments will be necessary.

AEDs that are not subject to hepatic metabolism (gabapentin, levetiracetam, and vigabatrin) would not be susceptible to metabolic interactions (Table 4.1).

Elimination

Elimination is the removal of drug molecules from the body by excretion, usually by the kidneys, or by biotransformation/metabolism (primarily CYPs), mainly in the liver. Excretion is important for water-soluble drugs and the water-soluble metabo- lites of lipid-soluble drugs. Conjugation by UGT isoenzymes usually results in the production of pharmacologically inactive and less lipid-soluble metabolites, which

54 Philip N. Patsalos

are often excreted in the urine or in the bile. Although drug interactions affecting renal excretion are rare with AEDs, AEDs that undergo extensive renal elimination in unchanged form may be susceptible to interactions affecting the excretion process, particularly when the latter involves active transport mechanisms or when the ion- ized state of the drug is highly sensitive to changes in urine pH (Bonate et al., 1998). For example, probenecid increases the plasma concentration of penicillin by com- peting for the same active transport system in the kidneys and consequently reduces the renal excretion of penicillin (Hansten, 1998). Also, agents which cause alkaliniza- tion of urine, reduce the reabsorption of phenobarbitone from the renal tubuli and consequently enhance its elimination (Powell et al., 1981). The latter interaction is exploited therapeutically in severe cases of barbiturate intoxication. It should be borne in mind that although vigabatrin, gabapentin, levetiracetam, topiramate and felbamate are renaly excreted, it has not been established whether or not this occurs by active transport systems (Table 4.1). Nevertheless, other drugs that are similarly excreted could potentially interact with these AEDs.

REFERENCES

Aitio ML, Mansury L, Tala E, et al. The effects of enzyme induction on the metabolism of disopy- ramide in man. Br J Clin Pharmacol 1981; 11: 279–285.

Anderson GD. A mechanistic approach to antiepileptic drug interactions. Ann Pharmacother 1998; 32: 554–563.

Bauer LA. Interference of oral phenytoin absorption by continuous nasogastric feeding. Neurology 1982; 32: 570–572.

Bonate PL, Reith K, Weir S. Drug interactions at the renal level. Implications for drug develop- ment. Clin Pharmacokinet 1998; 34: 375–404.

Cox DS, Scott KR, Gao H, et al. Effect of P-glycoprotein on the pharmacokinetics and tissue dis- tribution of enaminone anticonvulsants: analysis by population and physiological approaches. J Pharmacol Exp Ther 2002; 302: 1096–1104.

Fricker G, Drewe J, Huwyler J, et al. Relevance of P-glycoprotein for the enteral absorption of cyclosporine A: in vitro-in vivo correlation. Br J Pharmacol 1996; 118: 1841–1847.

Green MD, Bishop WP, Tephly TR. Expressed human UGT1. 4 protein catalyzes the formation of quaternary ammonium-linked glucuronides. Drug Metab Disp 1995; 23: 299–302.

Hansten PD. Understanding drug interactions. Sci Med 1998; January/February: 16–25.
Hatton RC. Dietery interaction with phenytoin. Clin Pharm 1984; 3: 110–111.
Hoffmeyer S, Burk O, von Richter O, et al. Functional polymorphisms of the human multidrug

resistance gene: multiple sequence variations and correlation of one allele with P-glycoprotein

expression and activity in vivo. Proc Natl Acad Sci USA 2000; 97: 3473–3478.
Jette L, Beaulieu E, Leclerc JM, et al. Cyclosporin A treatment induces overexpression of

P-glycoprotein in the kidney and other tissues. Am J Physiol 1996; 270: F756–F765.

55 Pharmacokinetic principles and mechanisms of drug interactions

Lin JH, Yamazaki M. Role of P-glycoprotein in pharmacokinetics. Clin Pharmacokinet 2003; 42: 59–98.

Lown KS, Mayo RR, Leichtman AB, et al. Role of intestinal P-glycoprotein (mdr1) in interpatient variation in the oral bioavailability of cyclosporin. Clin Pharmacol Ther 1997; 62: 248–260. Mather GG, Levy RH. Anticonvulsants. In Metabolic Drug Interactions, R. H. Levy,

K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia, USA: Lippincott

Williams & Wilkins, 2000; 217–232.
Mauro LS, Mauro VF, Brown DL, et al. Enhancement of phenytoin elimination by multiple-dose

activated charcoal. Ann Emerg Med 1987; 16: 1132–1135.
Neuvonen PJ, Elfving SM, Elonen E. Reduction of absorption of digoxin, phenytoin and aspirin

by activated charcoal in man. Eur J Clin Pharmacol 1978; 13: 213–218.
Neuvonen PJ, Elonen E. Effect of activated charcoal on absorption and elimination of phenobarbitone, carbamazepine and phenylbutazole in man. Eur J Clin Pharmacol 1980; 17:

51–57.
Nolan PE, Marcus FI, Karol MD, et al. Effect of phenytoin on the clinical pharmacokinetics of

amiodarone. J Clin Pharmacol 1990; 30: 1112–1119.
Patsalos PN. Therapeutic drug monitoring in epilepsy – principles and concepts. Epilepsy Monit

2001; 5: 1–6.
Patsalos PN. Therapeutic drug monitoring in epilepsy – the established and the new antiepilep-

tic drugs. Epilepsy Monit 2002; 6: 1–8.
Patsalos PN, Lascelles PT. Effect of sodium valproate on plasma protein binding of diphenyl-

hydantoin. J Neurol Neurosurg Psychiat 1977a; 40: 570–574.
Patsalos PN, Lascelles PT. In vitro hydroxylation of diphenylhydantoin and its inhibition by

other commonly used anticonvulsants. Biochem Pharmacol 1977b; 26: 1929–1933.
Patsalos PN, Lascelles PT. Clinically important drug interactions in epilepsy: general features and

interactions between antiepileptic drugs. Lancet Neurol 2003a; 2: 347–356.
Patsalos PN, Perucca E. Clinically important drug interactions in epilepsy: interactions between

antiepileptic drugs and other drugs. Lancet Neurol 2003b; 2: 473–481.
Patsalos PN, Duncan JS, Shorvon SD. Effect of the removal of antiepileptic drugs on antipyrine

kinetics in patients taking polytherapy. Br J Clin Pharmacol 1988; 26: 253–259.
Perucca E. Clinical implications of hepatic microsomal enzyme induction by antiepileptic drugs.

Pharmacol Ther 1987; 33: 139–144.
Perucca E. The clinical pharmacology and therapeutic use of the new antiepileptic drugs. Fund

Clin Pharmacol 2001; 15: 405–417.
Perucca E, Hebdige S, Frigo GM, et al. Interaction between phenytoin and valproic acid: plasma

protein binding and metabolic effects. Clin Pharmacol Ther 1980; 28: 779–789.
Perucca E, Hedges A, Makki KA, et al. A comparative study of the enzyme inducing properties of

anticonvulsant drugs in epileptic patients. Br J Clin Pharmacol 1984; 18: 401–110.
Potschka H, Loscher W. Multidrug resistance-associated protein is involved in the regulation of

extracellular levels of phenytoin in the brain. Neuroreport 2001; 12: 2387–2389.
Potschka H, Fedrowitz M, Loscher W. P-glycoprotein and multidrug resistance-associated pro- tein are involved in the regulation of extracellular levels of the major antiepileptic drug carba-

mazepine in the brain. Neuroreport 2001; 12: 3557–3560.

56 Philip N. Patsalos

Potschka H, Loscher W. P-Glycoprotein-mediated efflux of phenobarbital, lamotrigine, and felba- mate at the blood-brain barrier: evidence from microdialysis experiments in rats. Neurosci Lett 2002; 327: 173–176.

Powell JR, Nelson E, Conrad KA, et al. Phenobarbital clearance, elimination with alkaline diure- sis, and bioavailability in adults. Clin Pharmacol Ther 1981; 29: 273.

Rendic S, Di Carlo FJ. Human cytochrome P450 enzymes: a status report summarizing their reactions, substrates, inducers, and inhibitors. Drug Metab Rev 1997; 29: 413–580.

Rizzi M, Caccia S, Guiso G, et al. Limbic seizures induce P-glycoprotein in rodent brain: func- tional implications for pharmacoresistance. J Neurosci 2002; 22: 5833–2539.

Schuetz EG, Beck WT, Schuetz JD. Modulators and substrates of P-glycoprotein and cytochrome P4503A coordinately up-regulated these proteins in human carcinoma cells. Mol Pharmacol 1996; 49: 311–318.

Sisodiya SM. Mechanisms of antiepileptic drug resistance. Curr Opin Neurol 2003; 16: 197–201. Spatzenegger M, Jaeger W. Clinical importance of cytochrome P450 in drug metabolism. Drug

Metab Rev 1995; 27: 397–417.
Su T, He W, Gu J, et al. Differential xenobiotic induction of CYP2A5 in mouse liver, kidney, lung,

and olfactory mucosa. Drug Metab Dispos 1998; 26: 822–824.
Tassaneeyakul W, Veronese ME, Birkett DJ, et al. Co-regulation of phenytoin and tolbutamide

metabolism in humans. Br J Clin Pharmacol 1992; 34: 494–498.
Tonini M, Gatti G, Manzo L, et al. Vigabatrin does not affect the intestinal absorption of pheny-

toin in rat duodeno-jejunal loops in situ. Pharmacol Res 1992; 26: 201–205.
Tsanaclis LM, Allen J, Perucca E, et al. Effect of valproate on free plasma phenytoin concentra-

tions. Br J Clin Pharmacol 1984; 18: 17–20.
Verschraagen M, Koks CHW, Schellens JHM, et al. P-glycoprotein system as a determinant of

drug interactions: the case of digoxin-verapamil. Pharmacol Res 1999; 40: 301–306.
Wacher VJ, Wu CY, Benet LZ. Overlapping substrate specificities and tissue distribution of cytochrome P4503A and P-glycoprotein: implications for drug delivery and activity in cancer

chemotherapy. Mol Carcinog 1995; 13: 129–134.
Weichbrodt GD, Elliot DP. Treatment of phenytoin toxicity with repeated doses of activated

charcoal. Ann Emerg Med 1987; 16: 1387–1389.
Worden Jr JP, Wood Jr CA, Workman CH. Phenytoin and nasogastric feedings. Neurology 1984;

34: 132.

Predictability of metabolic antiepileptic drug interactions

Edoardo Spina1, Emilio Perucca2 and Rene Levy3

1 Section of Pharmacology, Department of Clinical and Experimental Medicine and Pharmacology, University of Messina, Messina, Italy

2 Clinical Pharmacology Unit, University of Pavia, Pavia, Italy
3 Department of Pharmaceutics, University of Washington, Seattle, WA, USA

Principles of drug metabolism

Many drugs are lipid soluble, weak organic acids or bases that are not readily elim- inated from the body, being reabsorbed into the blood from the glomerular filtrate. Metabolic processes are necessary to convert a drug into one or more metabolites which are chemically different from the parent compound, but generally more polar and water soluble, facilitating their excretion in urine or bile. Although metabolism usually results in inactivation or detoxification, many drug metabolites have pharmacological activity. Metabolites may occasionally be much more active than the parent compound (which then may be designated as a prodrug), they may exert effects similar to or different from those of the parent molecule, or they may be responsible for toxic effects (Perucca and Richens, 1995). When metabolites are active, termination of their action occurs by further biotransformation or by direct excretion of the metabolite in urine or bile.

The chemical reactions involved in the biotransformation of drugs are catalyzed by various enzyme systems and are conventionally divided into phase I (function- alization) and phase II (conjugation) biotransformation reactions, which may occur in series. Phase I reactions involve the addition of a polar functional group (e.g. a hydroxyl group) or the deletion of a non-polar alkyl group (e.g. N-demethylation) by oxidation, reduction, or hydrolysis. In phase II or conjugation reactions, the drug or the phase I metabolite are covalently attached to a water-soluble endogenous substrate (e.g. glucuronic acid, acetic acid, sulfate, amino acids or glutathione), usually resulting in an inactive, easily excretable compound.

The liver is usually the main organ responsible for phase I and phase II reactions, but other organs such as the gastrointestinal tract, the kidney, the lungs, the brain, the blood, the skin and the placenta may also contribute to metabolism. In the hepatocyte, phase I oxidative enzymes are located almost exclusively in the smooth

58 Edoardo Spina et al.

endoplasmic reticulum, along with the phase II enzyme, glucuronyltranferase. Other phase II enzymes responsible for conjugation reactions are found predominantly in the cytoplasm.

Major drug-metabolizing enzymes

Knowledge of the main enzyme systems involved in the biotransformation of antiepileptic drugs (AEDs) is essential for understanding the principles and mech- anisms of metabolically based drug interactions involving these drugs.

The cytochrome P450 system

The cytochrome P450 (CYP) system constitutes a superfamily of isoenzymes that are responsible for the oxidative metabolism of many endogenous (e.g. steroids, prostaglandins and fatty acids) and exogenous compounds (e.g. many drugs). These isoenzymes are haemoproteins located in the membranes of the smooth endoplasmic reticulum in the liver and in many extrahepatic tissues (Guengerich, 1997a), and they are subdivided into families, subfamilies and isoenzymes accord- ing to a nomenclature system based on amino acid sequence homology (Nelson et al., 1996). Each enzyme is designated with the root CYP followed by a first Arabic number indicating the ‘family’ ( 40% sequence identity within family members), a capital letter designating the ‘subfamily’ ( 59% sequence identity within sub- family members), and a second Arabic number representing individual iso- enzymes. The major CYP enzymes involved in drug metabolism in humans belong to families 1, 2 and 3, which together represent approximately 70% of the total CYP content in human liver (Shimada et al., 1994). The most important isoforms play- ing a major role in the biotransformation of therapeutic agents are CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP2E1 and CYP3A4, and these will be discussed in more detail below. Each CYP isoform is a specific gene product and possesses a characteristic but relatively broad spectrum of substrate specificity. Different CYP isoforms may display overlapping substrate specificities.

There is a large variability in the expression and activity of these isoenzymes, which may lead to interindividual differences in drug exposure. Such a variability results from genetic, pathophysiological and environmental factors, including con- comitant administration of other drugs. A number of genes coding for CYP iso- forms have variant alleles resulting from mutations, and these mutations can result in enzyme variants with higher, lower or no activity, or in the very absence of the enzyme. The existence of mutated alleles in at least 1% of the population is referred to as genetic polymorphism (Meyer, 1994). The CYP polymorphisms that have the greatest clinical implications are CYP2D6, CYP2C9 and CYP2C19.

59 Predictability of metabolic antiepileptic drug interactions

CYP1A2

In recent years, the major CYP isoenzymes have been characterized at the molecu- lar level and their different substrates, inhibitors and inducers have been identified (Rendic and Di Carlo, 1997). As indicated in Table 5.1, the majority of AEDs are metabolized by CYP enzymes and some may also inhibit or induce, to varying degrees, one or more of these isoforms. The activity of CYP enzymes can be evaluated in vitro and in vivo. In vitro studies provide a screening method for evaluating drug affinities as substrates, inhibitors or inducers. In vivo studies include phenotyping and/or genotyping tests. Phenotyping tests are based upon administration of a single dose of a probe compound to an individual, followed by measurement of urinary or plasma concentrations of the test compound and its major metabolite(s). The ratio of parent drug/metabolite (metabolic ratio, MR) is used as a measure of the activity of the enzyme responsible for the formation of that metabolite. Geno- typing is performed by using polymerase chain reaction (PCR)-based assays and restriction fragment length polymorphism (RFLP) analyses and allows detection of allelic variants for the genes coding for the polymorphic enzymes.

CYP1A2 accounts for approximately 13% of total hepatic CYPs and represents the primary enzyme responsible for the metabolism of many drugs, including phenacetin, paracetamol, tacrine, theophylline, caffeine, clozapine and olanzapine (Miners and McKinnon, 2000). Although CYP1A2 has not been found to play a major role in the metabolism of any AED, it does contribute to a minor extent to carbamazepine metabolism (Patsalos et al., 2002). Phenacetin and theophylline are frequently used as in vitro probes for CYP1A2, and caffeine is also widely used as a marker for CYP1A2 activity in vivo. Though CYP1A2 activity does not seem to be polymorphically distributed, it shows large interindividual variability.

Furafylline and -naphtoflavone are potent selective inhibitors of CYP1A2 and, therefore, may be used in vitro to evaluate the contribution of this isoform in drug- metabolizing pathways. Fluvoxamine is also a potent, but not selective, inhibitor of CYP1A2. The activity of CYP1A2 is induced by polycyclic aromatic hydrocarbons (including those found in charcoal-broiled foods and cigarette smoke), rifampicin, omeprazole and, possibly, by phenobarbital, phenytoin and carbamazepine (Guengerich, 1997a). Two polymorphisms have been reported which seem to enhance the inducibility of CYP1A2 (Nakajima et al., 1999; Sachse et al., 1999), but the clinical implications of this observation have not been clarified.

CYP2C9 and CYP2C19

The human CYP2C subfamily, which accounts for approximately 20% of total CYPs expressed in human liver, includes at least four members: CYP2C8, CYP2C9, CYP2C18 and CYP2C19 (Rettie et al., 2000). The relative contributions of these

Table 5.1 Substrates, probe drugs, inhibitors and inducers of the major CYP isoforms involved in drug metabolism

Isoenzymes

Substrates

Probe drugs

Inhibitors Furafylline

Inducers

CYP1A2

Antidepressants: amitryptyline, clomipramine, imipramine, fluvoxamine, mirtazepine

In vitro
Phenacetin O-deethylation Theophylline 8-hydroxylation In vivo
Caffeine

Cigarette smoke Rifampicin Carbamazepine Barbiturates Phenytoin Charcoal-broiled meat

CYP2C9

tamoxifen, R-warfarin
AEDs: phenytoin, phenobarbital

In vitro
Phenytoin p-hydroxylation Tolbutamide 4-hydroxylation S-w(s)-warfarin hydroxylation In vivo
Diclofenac, Losartan S-warfarin
Tolbutamide

Sulfaphenazole Amiodarone Fluconazole Miconazole Valproic acid Fluoxetine Fluvoxamine

Rifampicin Barbiturates Phenytoin Carbamazepine

CYP2C19

AEDs: S-mephenytoin, methylphenobarbital, phenytoin, diazepam

In vitro S-mephenytoin 4 -hydroxylation In vivo S-mephenytoin Omeprazole

Omeprazole Ticlopidine Fluvoxamine Felbamate Topiramate (weak)

Rifampicin Barbiturates Phenytoin Carbamazepine

CYP2D6

Antidepressants: amitryptyline, clomipramine, imipramine, desipramine, nortriptyline, fluoxetine, paroxetine, fluvoxamine, citalopram, venlafaxine, mianserine, mirtazepine

In vitro

Quinidine Propafenone Thioridazine Perphenazine

No inducer known

Antipsychotics: clozapine, olanzapine, haloperidol Methylxanthines: theophylline, caffeine Miscellaneous: phenacetin, paracetamol, tacrine,

-naphtoflavone Fluvoxamine Ciprofloxacin Clarithromycin

NSAIDs: diclofenac, ibuprofen, naproxen, piroxicam, celecoxib

Miscellaneous: S-warfarin, tolbutamide, losartan, torasemide, fluvastatin

Antidepressants: amitryptyline, clomipramine, imipramine, citalopram, moclobemide

Miscellaneous: omeprazole, propranolol, proguanil, R-warfarin

Dextrometorphan O-demethylation Debrisoquine

CYP2E1

Ethanol, halotane, dapsone, isoniazid, chlorzoxazone, felbamate, phenobarbital

In vitro

Disulfiram

Ethanol Isoniazid

CYP3A4

AEDs: carbamazepine, ethosuximide, tiagabine, zonisamide, some benzodiazepines (e.g. alprazolam, midazolam, triazolam)

In vitro

Ketoconazole Itraconazole Fluconazole Erythromycin Troleandomycin Ritonavir Indinavir Fluvoxamine Grapefruit juice

Carbamazepine Barbiturates Phenytoin Rifampicin

Antipsychotics: thioridazine, perphenazine, fluphenazine, zuclopenthixol, haloperidol, risperidone, clozapine, olanzapine, chlorpromazine

4-hydroxylation
Bufuralol 1 -hydroxylation In vivo
Debrisoquine Dextrometorphan Sparteine
Metoprolol
Desipramine

Opioids: codeine, dextromethorphan, tramadol -blockers: alprenolol, bufuralol, metoprolol,

Fluoxetine Paroxetine Haloperidol

propanolol, timolol, pindolol Antiarrhythmics: encainide, flecainide,

propafenone, sparteine
Miscellaneous: debrisoquine, phenformin

Antidepressants: amitryptyline, clomipramine, imipramine, sertraline, nefazodone, mirtazepine

Midazolam 1 -hydroxylation Erythromycin N-demethylation Testosterone 6 -hydroxylation In vivo

St. John’s wort Glucocorticoidsa Oxcarbazepinea Topiramatea Felbamatea

Antipsychotics: clozapine, risperidone, quetiapine, ziprasidone, haloperidol

Nifedipine Midazolam Cortisol Dapsone Erythromycin

Calcium antagonists: diltiazem, nifedipine, verapamil, felodipine

Miscellaneous: cisapride, terfenadine, astemizole, cyclosporine, tacrolimus, erythromycin, clarithromycin, tamoxifen, amiodarone, quinidine, itraconazole, ketoconazole, indinavir, ritonavir

a Weaker or tissue-selective inducers.

Chlorzoxazone 6-hydroxylation

In vivo

Chlorzoxazone

62 Edoardo Spina et al.

CYP2D6

isoforms to total CYP2C content in human liver are about 60% for 2C9, 35% for 2C8, 4% for 2C18 and 1% for 2C19. Of these isoforms, CYP2C9 and CYP2C19 seem to be the most important for drug metabolism, although CYP2C8 should not be neglected as it contributes to the metabolism of carbamazepine (Kerr et al., 1994). As CYP2C9 and CYP2C19 show a 91% identity in amino acid sequence, many sub- strates of CYP2C9 are also metabolized, at least in part, by CYP2C19.

CYP2C9 plays an important role in the oxidation of many drugs, including pheny- toin, phenobarbital, S-warfarin, tolbutamide, losartan, fluvastatin and many non- steroidal anti-inflammatory agents such as diclofenac, ibuprofen and piroxicam (Perucca and Richens, 1995; Guengerich, 1997a). CYP2C9 is polymorphically expressed in humans. To date, three different allelic variants have been identified, that code for enzymes with different catalytic activity (Miners and Birkett, 1998). The fre- quencies of the defective alleles CYP2C9*2 and CYP2C9*3 vary between 8% and 12% and 3% and 8%, respectively among Whites, but they are somewhat lower in Orientals and black Africans. Subjects carrying two mutated alleles for CYPC9*3 lack almost completely CYP2C9 activity, and, therefore, are unable to metabolize important CYP2C9 substrates such as phenytoin and S-warfarin (Brandolese et al., 2001). While sulfaphenazole is the prototypic inhibitor for CYP2C9, other inhibitors include val- proic acid, amiodarone, fluconazole and miconazole.

CYP2C19 is involved to a significant extent in the biotransformation of methylphenobarbital, phenytoin, omeprazole, proguanil, citalopram and tricyclic antidepressants (demethylation reactions) (Guengerich, 1997a). However, the pro- totype substrate for this isoform is the S-enantiomer of mephenytoin, which undergoes p-hydroxylation at position 4 on its aromatic ring. The exclusive partic- ipation of CYP2C19 in this metabolic pathway is the basis for the use of S-mephenytoin as an in vitro and in vivo probe for CYP2C19 activity. CYP2C19 also exhibits an important genetic polymorphism. The frequency of the poor metabo- lizer (PM) phenotype varies from approximately 3% in Whites to 12–25% in many Asian populations, while in black Africans PM frequencies vary between 4% and 7% (Goldstein, 2001). The major defective alleles responsible for the PM pheno- type are CYP2C19*2, the most common among Whites and Orientals, and CYP2C19*3, found at a frequency of about 12% among Orientals, but almost absent among Whites. The activity of CYP2C19 may be inhibited by felbamate, omeprazole, ticlopidine, fluvoxamine and, possibly, topiramate.

Inducers of the activity of CYP2C isoforms include barbiturates, phenytoin, carbamazepine and rifampicin.

Although expressed at low levels (2% of hepatic CYPs) compared with other human CYPs, CYP2D6 plays an important role in the biotransformation of a large

63 Predictability of metabolic antiepileptic drug interactions

CYP2E1

CYP3A4

number of drugs (Zanger and Eichelbaum, 2000). To date, however, none of the major AEDs has been found to be metabolized to a significant extent by CYP2D6. Debrisoquine, sparteine, dextromethorphan and desipramine have been vali- dated as probe drugs for CYP2D6. This enzyme exhibits an important genetic polymorphism. PMs lack CYP2D6 activity and represent approximately 3–10% of Whites, but only 1–2% of Orientals (Evans et al., 1980). Among extensive meta- bolizers (EMs), the catalytic activity varies largely, and a subgroup of subjects with extremely high enzyme activity have been classified as ultrarapid metabolizers (UMs) (Johansson et al., 1993). The CYP2D6 gene is extremely polymorphic with more than 70 allelic variants described so far (Bertilsson et al., 2002). Three major mutated alleles, CYP2D6*3, CYP2D6*4 and CYP2D6*5, account for 90–95% of the PM alleles in Whites, and CYP2D6*4 is the most common allele associated with the PM phenotype in Whites (allele frequency of about 21%). CYP2D6*4 is almost absent in Orientals, which may account for the low incidence of PMs in these pop- ulations. On the other hand, the high frequency (up to 50%) of the CYP2D6*10 allele among Orientals, and its absence among Whites, may explain the slightly lower CYP2D6 activity found in Oriental EMs compared to Whites. The frequency of the CYP2D6*5 allele, with deletion of the entire CYP2D6 gene, is about 4–6% and is similar in different ethnic populations. Individuals heterozygous for the defect alleles have lower enzyme activity than homozygous EMs. On the other hand, alleles with duplication or multiduplication of a functional CYP2D6*2 gene are associated with an increased CYP2D6 activity: the frequency of this condition varies from 1–2% in Swedes to up to 7–10% in Spaniards and Southern Italians

(Bertilsson, 2002).
Quinidine, fluoxetine, paroxetine and different phenothiazines are potent

inhibitors of CYP2D6. In contrast to all other CYPs involved in drug metabolism, CYP2D6 does not appear to be inducible, an important consideration in predict- ing interactions caused by AEDs.

CYP2E1, which represents approximately 7% of total human hepatic CYPs, is of greater importance in toxicants’ metabolism than in drug metabolism (Raucy and Carpenter, 2000). CYP2E1 is responsible for the metabolism of ethanol, halotane and dapsone, and plays a minor role in the oxidative biotransformation of felbamate and phenobarbital. Chlorzoxazone has been suggested as probe drug for CYP2E1. CYP2E1 activity is inhibited by disulfiram and is induced by ethanol and isoniazid.

The CYP3A subfamily, which includes the isoforms 3A4, 3A5 and 3A7, is the most abundant in human liver, accounting for approximately 30% of total CYP content.

64 Edoardo Spina et al.

CYP3A4 is the predominant isoform in adults, is present both in the liver and in the small intestine, and participates in the biotransformation of more than 50% of all eliminated by metabolism drugs (Wrighton and Thummel, 2000). CYP3A4 is the primary enzyme responsible for the metabolism of carbamazepine, ethosuximide, tiagabine and zonisamide, and it is also involved in the biotransformation of felba- mate (Perucca and Richens, 1995). Other drugs primarily metabolized by this isoform include immunosuppressants (e.g. cyclosporin and tacrolimus), triazolobenzodi- azepines (e.g. alprazolam, midazolam and triazolam), non-sedating antihistamines (e.g. terfenadine and astemizole), calcium antagonists (e.g. diltiazem, verapamil, nifedipine and other dihydropyridines), cholesterol lowering drugs (e.g. simvastatin and lovastatin), antiarrhythmics (e.g. amiodarone and quinidine), and several steroids (e.g. cortisol, ethinylestradiol and levonogestrel). Index reactions for CYP3A4 activity in vitro include midazolam and triazolam 1- and 4-hydroxylation, nifedipine dehydrogenation and testosterone 6 -hydroxylation. Cortisol, nifedipine, erythro- mycin and midazolam have been used as in vivo probes.

The hepatic and enteric location of CYP3A4 makes it well suited to play a sig- nificant role in first-pass (or presystemic) drug metabolism. Furthermore, the con- siderable overlap in substrate selectivity and tissue localization of CYP3A4 and P-glycoprotein, an intestinal transport protein located in the small bowel, has led to the hypothesis that this transporter and enzyme pair act as a co-ordinated barrier against xenobiotics at the intestinal level (Schuetz et al., 1996). Although CYP3A4 drug-metabolizing activity varies more than 20-fold among individuals, it has a unimodal distribution and does not appear to be subject to genetic poly- morphism. Its wide interindividual variability is caused, at least in part, by modu- lation of CYP3A4 activity by many environmental compounds, including dietary constituents and medications.

Compounds that inhibit CYP3A4 activity include azole antimycotics (e.g. keto- conazole and itraconazole), macrolide antibiotics (e.g. erythromycin and trolean- domycin), HIV protease inhibitors (e.g. ritonavir and indinavir), nefazodone and some of the furanocoumarin dimers found in grapefruit juice (Guengerich, 1997b). The hepatic and, possibly, the intestinal CYP3A4 isoforms are induced by glucocorticoids (e.g. dexamethasone), rifampicin, phenobarbital, phenytoin and carbamazepine. Felbamate, oxcarbazepine and topiramate appear to exert a selec- tive inducing effect on CYP3A4 activity, at least in some tissues.

Recent studies indicate that CYP3A5 can account for more than 50% of total CYP3A hepatic and jejunal content in 30% of Whites and 50% of African Americans. This behavior has been associated with the CYP3A5*1 wild-type allele (Lamba et al., 2002). Such individuals will exhibit more variability in clearance of CYP3A substrates and more variability in drug interactions since CYP3A5 appears less inhibitable than CYP3A4.

65 Predictability of metabolic antiepileptic drug interactions

Epoxide hydrolases

Epoxide hydrolases (EHs) belong to the group of hydrolytic enzymes which also includes esterases, proteases, dehalogenases and lipases. EHs catalyze a specialized form of hydrolysis, called hydration, where water is added to a compound without causing its cleavage into separate components (Omcienski, 2000). These enzymes hydrate epoxides and arene oxides to dihydrodiol and diol-epoxide metabolites.

Many epoxide intermediates, formed during oxidation of xenobiotics and endogenous substances, are reactive electrophilic species that may act as critical initiators of cellular damage through protein and RNA adduction as well as genetic mutation. Through inactivation of epoxides, EHs are usually implicated in detoxi- fication processes, although in certain instances they may be involved in bio- activation. Five classes of EHs have been described: (a) cholesterol oxide hydrolase; (b) hepoxylin A3 hydrolase; (c) leukotriene A4 hydrolase; (d) soluble EH; and (e) microsomal EH.

Microsomal EH catalyzes the trans-addition of water to a broad range of epoxides and arene oxides derived from xenobiotics, resulting in the formation of dihydro- diol products. This enzyme exhibits a broad-substrate specificity and plays a role in the metabolism of some AEDs. Phenobarbital, phenytoin and carbamazepine, in particular, are metabolized by CYP isoenzymes to epoxide intermediates, which have been implicated in idiosyncratic adverse drug reactions and teratogenicity. Since these epoxide intermediates can be substrates for microsomal EH, it has been hypothesized that EH enzymatic status may modulate the individual susceptibility to adverse drug reactions (Lindhout, 1992).

Unlike other epoxides, the 10,11-epoxide metabolite of carbamazepine is chem- ically stable and retains anticonvulsant activity. The clearance of this metabolite is controlled by microsomal EH activity. In vitro and in vivo interaction studies with carbamazepine-10,11-epoxide have indicated that valpromide, valnoctamide and, to a lesser extent, valproic acid are inhibitors of microsomal EH (Kerr et al., 1989; Pisani et al., 1993). Another AED, progabide, has been reported to inhibit microso- mal EH both in vivo and in vitro (Kroetz et al., 1993). The activity of microsomal EH may be moderately induced by phenobarbital, phenytoin and carbamazepine.

Uridine diphosphate-glucuronosyltransferases

Uridine diphosphate (UDP)-glucuronosyltransferases (UDPGTs) are a subset of enzymes belonging to the superfamily of UDP-glycosyltransferases (UGTs) (Liston et al., 2001). These enzymes, which catalyze the glucuronidation of a large number of endobiotics and xenobiotics, are located in the endoplasmic reticulum, mainly in the liver, but also in the kidney, intestine, skin, lung, prostate and brain.

Glucuronidation, which is the most common pathway in phase II drug metabo- lism, involves the transfer of the glucuronyl moiety of uridine diphosphate glucuronic

66 Edoardo Spina et al.

acid (UDPGA) to the substrate, with subsequent release of UDP. While most sub- strates undergo glucuronide conjugation after phase I reactions, in some cases, i.e. morphine and valproic acid, direct conjugation proceeds without phase I func- tionalization of the parent compound. In general, glucuronidation leads to formation of water-soluble inactive metabolites, but active and reactive glucuronide metabolites have also been described, as in the case of morphine.

In recent years, at least 33 families within the UGTs superfamily have been identified and classified by a nomenclature similar to that used for the CYP system (Mackenzie et al., 1997). Various UDPGTs have been characterized and assigned to the UGT1 and UGT2 gene families. Among the isoforms of the UGT1 family, UGT1A3 is involved in the O-glucuronidation of valproic acid and UGT1A4 has been found to be the major isoform responsible for the N-glucuronidation of lam- otrigine (Dickins and Chen, 2002) and retigabine (Hiller et al., 1999). Among the isoforms of the UGT2 family, the UGT2B7 variant also appears to contribute to the O-glucuronidation of valproic acid (Jin et al., 1993).

In contrast to extensive documentation for CYP-mediated drug interactions, there are fewer data on interactions involving glucuronidation. Any substrate of UGT has the potential to competitively inhibit glucuronidation of other substrates metabo- lized by the same enzyme. Unlike the CYP system, no specific inhibitors of individ- ual UGT isoforms have been identified. Valproic acid has been reported to inhibit several glucuronidation reactions, while phenobarbital, phenytoin, carbamazepine and, to a lesser extent, oxcarbazepine may act as inducers (Perucca and Richens, 1995; Perucca, 2001). In particular, phenobarbital appears to induce UGT1A1, the major enzyme responsible for the glucuronidation of bilirubin, ethinylestradiol and the opioids buprenorphine, nalorphine and naltrexone (Bock et al., 1999).

Enzyme induction and enzyme inhibition

Drug interactions involving CYP isoforms and other drug-metabolizing enzymes may result from one of two processes, enzyme induction or inhibition.

Enzyme induction

The activity of drug-metabolizing enzymes in the liver and/or extrahepatic tissues may be increased (‘induced’) by chronic administration of several exogenous agents including drugs, industrial contaminants, dietary or voluctuary substances, as well as by endogenous compounds (Guengerich, 1997b). Although induction involves predominantly CYP isoenzymes, other enzymes including microsomal EH and UGTs may be affected. Morphologically, enzyme induction may be associated with a proliferation of the smooth endoplasmic reticulum and hepatic hypertrophy. From a biological point of view, induction is an adaptive response that protects the

67 Predictability of metabolic antiepileptic drug interactions

cells from toxic xenobiotics by increasing the detoxification activity. Therefore, it is to be expected that induction will result in decreased concentration of an active compound. However, for those agents that are inactive but are biotransformed to active metabolites, enzyme induction may paradoxically increase pharmacological or toxicological activity.

Enzyme induction is the consequence of an increased concentration of the enzyme protein (Lin and Lu, 1998; Thummel et al., 2000). In most cases, this involves an enhanced protein synthesis resulting from an increase in gene transcription, usu- ally mediated by intracellular receptors. However, enzyme induction may also occur by an inducer-mediated decrease in rate of enzyme degradation, mainly through protein stabilization. Each inducer has its own specificity in inducing a given range of drug-metabolizing enzymes, and several mechanisms of induction are often acti- vated by a single agent, but to a different extent. Five main mechanisms of induction have been established to date (Fuhr, 2000). The two best known are the polycyclic aromatic hydrocarbon type and the phenobarbital type of induction.

Induction mediated by the aryl hydrocarbon receptor

Polycyclic aromatic hydrocarbons such as benzo(a)pyrene and 3-methylcholanthrene are environmental contaminants formed by incomplete combustion of organic matter, i.e. cigarette smoke and charcoal-broiled beef. These agents selectively induce CYP1A1 and CYP1A2, but they can also stimulate the activity of other enzymes such as UGTs. The mechanism of this type of induction has been well characterized and involves a sequence of events: initial binding of the inducer to the intracellular aryl hydrocarbon (Ah) receptor, dissociation of heat-shock 90 proteins from the receptor, translocation of the receptor–ligand complex into the nucleus, binding to the Ah receptor nuclear translocator (Arnt), binding of the Ah–Arnt complex to response elements on the CYP1A genes, resulting in increased gene transcription (Sogawa and Fujii-Kuriyama, 1997). In addition to polycyclic aromatic hydrocarbons, certain constituents of cruciferous vegetables, and certain drugs, such as omeprazole and rifampicin, appear to induce CYP1A enzymes by the same mechanism (Fuhr, 2000).

The constitutive androstane receptor and phenobarbital-type induction

Phenobarbital is recognized as the prototype of a class of agents known to induce drug metabolism (Perucca and Richens, 1995). Many other compounds, including phenytoin, primidone, carbamazepine, rifampicin and the oxazaphosphorines cyclophosphamide and ifosfamide have been shown to stimulate drug-metabolizing enzymes with an induction pattern which overlaps, at least in part, with that of barbiturates. Early investigations in liver microsomes from individuals exposed to phenobarbital and in primary cultures of human hepatocytes have documented

68 Edoardo Spina et al.

the ability of phenobarbital to induce CYP enzymes, but the specific isoforms induced could not be identified. More recently, with the improvement in culture techniques and the development of isoform-specific reagents, it has been possible to demonstrate that the cluster of enzymes induced by phenobarbital and related agents appears to include several CYPs such as CYP2C subfamily members, CYP3A4, CYP2B6, possibly CYP1A2, but not CYP2D6 (Fuhr, 2000). In addition, micro- somal EH and some UGTs appear to be induced by these agents. Thus, the drugs metabolized by enzymes subject to phenobarbital-type induction include a major fraction of all drugs undergoing biotransformation. Endogenous compounds, such as cortisol, testosterone and vitamin D3, are also susceptible to induction by pheno- barbital and related agents (Perucca, 1978).

Recent evidence suggests that the orphan receptor constitutive androstane receptor (CAR) is the molecular target and mediator of phenobarbital-type induc- tion (Sueyoshi et al., 1999). It should be pointed out that the molecular mechanism of phenobarbital-type induction may show partial overlap with that of the preg- nane X receptor (PXR), which mediates CYP3A4 induction by rifampicin and glucocorticoids.

Induction mediated by the PXR

This type of induction, previously called the rifampicin/glucocorticoid-type induction, has as target CYP3A4 enzymes, mainly in the gut. Induction involves the binding of CYP3A4 inducers, including several steroids, rifampicin and pheno- barbital, to the human PXR (Fuhr, 2000).

Enzyme induction by ethanol

The ethanol-type induction is probably limited to a single target, CYP2E1. Unlike other types of induction mediated by intracellular ‘receptors’, ethanol-type induc- tion occurs through protein stabilization mediated by the binding of the inducers to the active site of the enzyme (Gonzalez et al., 1991). Ethanol-type inducers sta- bilize the enzyme by protecting it from degradation, resulting in accumulation of the enzyme. Inducers of CYP2E1 are often substrates of the same enzyme and include ethanol, isoniazid and many organic solvents such as acetone, benzene and carbon tetrachloride.

Induction caused by peroxisome proliferators

This type of induction is mediated by binding to two peroxisome proliferator- activated receptors (PPARs), PPAR and PPAR . PPAR controls the transcription of genes encoding for enzymes mediating the metabolism of lipoproteins and fatty acids, while PPAR is involved in adipogenesis. Typical peroxisome proliferator inducers are members of the classes of fibrates and glitazones (Fuhr, 2000).

69 Predictability of metabolic antiepileptic drug interactions

Enzyme induction as a cause of drug interactions

Enzyme induction is a slow regulatory process, which is dose and time dependent (Perucca, 1978, 1987; Perucca et al., 1984). In other words, the extent of induction is generally proportional to the dose of the inducing agent and, since the process usually requires synthesis of new enzyme, it occurs with some delay after the expo- sure to the inducing agent. In practice, the time required for induction depends on the time to reach the steady state of the inducing agent (approximately five elimi- nation half-lives) and the rate of synthesis of the enzyme(s). Similarly, the time course of de-induction is also gradual and depends on the rate of degradation of the enzyme and the time required to eliminate the inducing drug. Either of these two processes could be the rate-limiting step. As far as classical enzyme-inducing AEDs are concerned, induction by phenobarbital is usually manifest after approx- imately 1 week, with maximal effect occurring after 2–3 weeks following initiation of therapy. De-induction follows a similar time course (Anderson and Graves, 1994; Anderson, 1998). With phenytoin, maximal induction or de-induction occur approximately 1–2 weeks after initiation or removal of therapy respectively (Anderson and Graves, 1994; Anderson, 1998). Carbamazepine is the only AED which significantly induces its own metabolism (autoinduction) and, as a result of this, its plasma clearance more than doubles during the initial weeks of therapy. The time course of carbamazepine autoinduction should be completed within approximately 3–5 weeks (Anderson and Graves, 1994; Anderson, 1998).

Enzyme induction may have a profound impact on the pharmacokinetics of drugs metabolized by the susceptible enzyme(s) (Perucca, 1978). Elevated enzyme concentrations in the eliminating organ(s) generally result in an increase in the rate of metabolism of the affected drug, leading to a decrease in serum drug con- centrations and, possibly, decreased clinical efficacy. If the affected drug has an active metabolite, induction can result in increased metabolite concentrations and possibly enhanced toxicity. There are three different situations where enzyme induction plays a role in therapeutic decision-making: addition of a medication when an inducer is already present, addition of an inducer to an existing therapy, and removal of an inducer from chronic therapy. In the first two cases a higher dose of the affected drug will be needed to achieve or maintain clinical efficacy, while a reduction of the dose of the affected drug may be necessary to prevent tox- icity after removal of the inducer. The magnitude and timing of these interactions are critical to allow clinicians to adjust dosages in order to maintain therapeutic effects and prevent toxicity.

Enzyme induction represents a common problem in the management of epilepsy. Based on their enzyme-inducing properties, phenobarbital, phenytoin and carbamazepine have been reported to increase the clearance or reduce the thera- peutic efficacy of many different compounds including other AEDs (Perucca, 1982;

70 Edoardo Spina et al.

2001). As a general rule, these compounds will induce the biotransformation of any drug that is primarily metabolized by CYP3A4, CYP2C9, CYP2C19 and, possibly, CYP1A2 (see Table 5.1). The possibility of induction of CYP1A2 by carbamazepine is supported by evidence that this agent increases the metabolic clearance of CYP1A2 substrates such as olanzapine and R-warfarin, and increases the percentage of labelled caffeine exhaled as carbon dioxide, a marker of CYP1A2 activity in vivo (Parker et al., 1998). Because the induction profiles of phenobarbital, phenytoin and carba- mazepine are not fully overlapping, stimulation of the metabolism of all drugs listed in Table 5.1 may not necessarily be observed with each of these AEDs. Moreover, in some cases enzyme induction and inhibition may occur at the same time, complic- ating the prediction process. In any case, clinically relevant interactions should be expected when enzyme-inducing agents are co-administered with drugs with a low therapeutic index such as warfarin, oral contraceptives or cyclosporin (Anderson, 1998). When active metabolites are formed, enzyme induction may result in potenti- ation of therapeutic and/or toxic effects. For example, the enhanced hepatotoxicity of valproic acid in children concurrently treated with enzyme inducers could be explained by accelerated formation of reactive oxidation products (Kondo et al., 1990).

In addition to classical enzyme-inducing AEDs, some newer agents, namely fel- bamate, oxcarbazepine and topiramate, may produce significant enzyme induc- tion, though the spectrum of enzymes induced by these agents appears to be more restricted. In particular, felbamate may induce the activity of CYP3A4 (Glue et al., 1997), as indicated by a decrease in the plasma concentrations of CYP3A4 sub- strates such as carbamazepine (Fuerst et al., 1988), ethinylestradiol and gestodene (Saano et al., 1995). Unlike carbamazepine, oxcarbazepine is not subject to auto- induction, but it may selectively induce the specific isoforms of the CYP3A group involved in the metabolism of oral contraceptives (Fattore et al., 1999) and dihy- dropyridine calcium antagonists (Zaccara et al., 1993). In addition, oxcarbazepine may also induce UGTs, as suggested by a significant acceleration of lamotrigine clearance (May et al., 1999). Topiramate is also a weak inducer of CYP3A4, because at dosages above 200mg/day it may decrease plasma concentrations of ethinylestradiol by approximately 30% with a risk of failure of contraception (Rosenfeld et al., 1997). At lower dosages, topiramate does not appear to affect the metabolism of steroid contraceptives (Doose et al., 2003), reinforcing the impor- tant concept that enzyme induction is a dose-dependent phenomenon. Recent evi- dence indicates that topiramate, at higher dosages, may induce CYP3A4 by activation of PXA (Nalloni et al., 2003)

Enzyme inhibition

Enzyme inhibition is the most common mechanism underlying drug interactions. A large number of compounds may inhibit the activity of drug-metabolizing

71 Predictability of metabolic antiepileptic drug interactions

enzymes, in particular with CYPs. As a consequence of enzyme inhibition, the rate of metabolism of a particular agent is decreased, resulting in increased plasma drug concentrations and potential enhancement of its pharmacological effects.

The mechanisms of enzyme inhibition include reversible inhibition, slowly reversible inhibition and irreversible inhibition (Lin and Lu, 1998; Thummel et al., 2000). In reversible inhibition, the normal function of the enzyme is restored after the inhibitor has been eliminated from the body. In contrast, the loss of enzyme activity caused by irreversible inhibition persists even after the elimination of the inhibitor, and de novo biosynthesis of new enzyme is the only means by which activity can be restored.

Reversible inhibition

This type of enzyme inhibition is probably the most common and, kinetically, it can be subdivided further into competitive, non-competitive and uncompetitive inhibition (Lin and Lu, 1998). Competitive inhibition involves a mutually exclu- sive competition between the substrate and the inhibitor for binding to the cat- alytic site of the enzyme. Competitive inhibitors can be non-substrates with nevertheless high binding affinity: binding of the inhibitor prevents the substrate from binding to the active site of the enzyme and, therefore, the substrate cannot be metabolized. This inhibition can be reversed by increasing the concentration of the substrate. In the case of non-competitive inhibition, the inhibitor binds to another site of the enzyme and the inhibitor does not affect the binding of the sub- strate, but formation of the enzyme–inhibitor complex results in loss of enzyme activity. Uncompetitive inhibition occurs when the inhibitor does not bind to the enzyme, but to the enzyme–substrate complex, and again formation of the enzyme– substrate–inhibitor complex results in loss of enzyme activity.

Slowly reversible inhibition

Several drugs undergo metabolic activation by CYP enzymes to form inhibitory metabolites. These metabolites can form stable complexes with the prosthetic haem of CYPs, the so-called metabolic intermediate (MI) complexes, so that the CYP isoform is sequestered in a functionally inactive state (Lin and Lu, 1998). While in vitro MI complexation can be reversed, in vivo the MI complex is usually so stable that the CYP involved in the complex is not available for drug metabo- lism, and the activity can be restored only after synthesis of new enzyme. The effect of this inhibition may, therefore, persist well after the elimination of the interacting drug. Troleandomycin and erythromycin are probably the best-known macrolide antibiotics involved in the formation of MI complexes. These two agents are asso- ciated with a clinically significant inhibition of the CYP3A4-mediated metabolism of carbamazepine (Spina et al., 1996). Hydrazine derivatives represent another

72 Edoardo Spina et al.

class of compounds that may form stable complexes with the haem of CYP enzymes. Among these agents, isoniazid may cause a significant inhibition of pheny- toin metabolism (Patsalos et al., 2002), probably through MI complexation with CYP enzymes involved in its biotransformation.

Irreversible inhibition

Some drugs are oxidized by CYPs to reactive intermediates that then cause irre- versible inactivation of the enzyme (Lin and Lu, 1998). As metabolic activation is required for enzyme inactivation, these agents are classified as mechanism-based inactivators or suicide inhibitors. This inactivation of CYPs may result from irre- versible alteration of the haem or protein, or a combination of both. Classical examples of compounds that alkylate the prosthetic haem group and inactivate the enzyme include olefins, acetylenes and dihydropyridines. Chloramphenicol pro- vides perhaps the best example of a drug causing irreversible (suicide) inactivation of CYP through protein modification.

Enzyme inhibition as a cause of drug interactions

Competitive inhibition is typically a rapid and dose-dependent process (Anderson and Graves, 1994; Anderson, 1998). The initial effect usually occurs within 24 h from the addition of the inhibitor, though the time to reach maximal inhibition will depend on the elimination half-lives of the affected drug and of the inhibiting agent. When the inhibitor is withdrawn, restoration of baseline (pre-interaction) conditions is also dependent on the rates of the elimination of the affected drug and of the inhibitor. With non-competitive and uncompetitive inhibition, the time course of the interaction may be more complex, and a significant role may be played by the turnover (re-synthesis) rate of the enzyme.

Inhibitors of drug metabolism usually interfere with only a limited number of isoenzymes and, therefore, they may be used to discriminate between different isoenzymes (Guengerich, 1997b). Compounds acting as inhibitors of different CYPs are listed in Table 5.1. Some potent inhibitors of a given enzyme are substrates of the same enzyme, but this is generally not the case. For example, quinidine is a potent inhibitor of CYP2D6, but it is metabolized by CYP3A4. Inhibition of non- oxidative phase I and conjugating phase II enzymes has also been documented.

Among AEDs, those acting most commonly as enzyme inhibitors are valproic acid and felbamate (Perucca and Richens, 1995). Valproic acid is considered as a broad-spectrum inhibitor of various enzymes. In particular, studies in human liver microsomes demonstrated that, at clinically relevant concentrations, valproic acid competitively inhibits CYP2C9 activity, inhibits only weakly CYP2C19 and CYP3A4, and it has no appreciable effect on CYP2D6 and CYP2E1 (Wen et al., 2001). This is

73 Predictability of metabolic antiepileptic drug interactions

consistent with clinical evidence that valproic acid may significantly increase the plasma concentrations of CYP2C9 substrates such as phenytoin and phenobarbi- tal (Scheyer, 2002). Studies in human liver microsomes also indicated that valproic acid inhibits EH, which explains its ability to increase the plasma concentration of carbamazepine-10,11-epoxide in carbamazepine-treated patients (Kerr et al., 1989). Valproic acid also has an important inhibitory effect on UGTs, as indicated by its ability to inhibit in vivo the glucuronide conjugation of lamotrigine, lorazepam and zidovudine, as well as the N-glucosidation of phenobarbital (Liston et al., 2001). The specific UGT isoform involved in these metabolic reactions is known only for lamotrigine, whose glucuronidation is metabolized by UGT1A4.

Felbamate is a selective inhibitor of CYP2C19 (Glue et al., 1997), which is con- sistent with the observation that felbamate reduces the clearance and increases the plasma concentration of phenytoin (Fuerst et al., 1988). Moreover, felbamate has been reported to increase the plasma concentrations of phenobarbital (Gidal and Zupanc, 1994), clobazam, carbamazepine-10,11-epoxide (concomitantly with a reduction in plasma carbamazepine levels) and valproic acid (Patsalos et al., 2002): with the possible exception of the increase in carbamazepine-10,11-epoxide, which may be related to induction of carbamazepine metabolism, these interac- tions are also ascribed to inhibition of the metabolism of the corresponding com- pounds, though the precise molecular mechanisms have not been elucidated. In the case of valproic acid, there is evidence that the increase in its plasma levels after addition of felbamate can be ascribed at least in part to inhibition of mitochondrial -oxidation (Hooper et al., 1996).

Other AEDs may at times act as enzyme inhibitors. Topiramate has been reported to be a modest inhibitor of the activity of CYP2C19 in vitro, though at con- centrations higher than those usually found in therapeutic practice (Sachdeo et al., 2002). Whether this mechanism is responsible for the moderate rise in plasma phenytoin concentration which is seen in a small subset of phenytoin-treated patients given topiramate is unclear. Other inhibitors of CYP2C19 are carba- mazepine and oxcarbazepine: in particular, CYP2C19 inhibition explains the abil- ity of oxcarbazepine, especially when used at high dosages, to increase by up to 40% the plasma concentrations of phenytoin (Patsalos et al., 2002). Phenobarbital concentrations may also be increased by oxcarbazepine, though to a lesser extent compared with phenytoin. Interestingly, oxcarbazepine is an inducer of UGT and CYP3A4, as demonstrated by its ability to increase the metabolism of lamotrigine and oral contraceptives respectively (Perucca, 2001): this illustrates the important concept that a compound may act as an inducer or as an inhibitor depending on which isoenzyme is being considered. There are also situations where a drug may induce and inhibit the same isoenzyme simultaneously: for example, at low doses phenobarbital tends to induce the metabolism of phenytoin, probably through

74 Edoardo Spina et al.

induction of CYP2C9 and/or CYP2C19, whereas at higher doses it may competi- tively inhibit phenytoin metabolism (Patsalos et al., 2002). The extent of these differential interactions may vary across individuals, which may explain the unpredictable and bi-directional changes in plasma phenytoin concentration after addition or removal of phenobarbital therapy.

Most AEDs undergo extensive biotransformation, and their metabolism is, therefore, vulnerable to inhibition by a large number of competitive substrates and enzyme inhibitors.

In vitro systems for testing drug metabolism and metabolic drug interactions

The potential for metabolic drug interactions is an important aspect to be consid- ered during the development of new drugs. In the past, most drug interaction studies were performed relatively late in phase II and III studies, and investigations were focused on compounds chosen for their likelihood of concurrent use. Since susceptibility to involvement in drug interactions is an undesirable property of a drug, information on this should ideally be obtained already, in the preclinical phase. In recent years, different in vitro techniques have been developed and have become widely used as screening tools to predict potential drug interactions before a drug reaches the clinical phases of development. The techniques used for in vitro assess- ments are described concisely in the sections below. For more comprehensive infor- mation, the reader is referred to specialized reviews (Ring and Wrighton, 2000).

Enzyme-based techniques

Initially, the simplest approach to the in vitro study of drug metabolism was through use of purified enzymes. One could determine whether a drug is a substrate of a specific isoenzyme, and the ability of the drug to inhibit the same isoenzyme can be evaluated by investigating its effect on the in vitro biotransformation of a probe substrate. However, the complexity of the purification techniques required to isolate these enzymes and the need for detergents, lipids and other enzymes (e.g. cytochrome b5 and P450 reductase) in the incubation system may limit the possi- ble extrapolation of results obtained with purified enzyme systems to the in vivo situation (Ring and Wrighton, 2000).

Recent advances in molecular biology have allowed isolation of cDNA encoding for drug-metabolizing enzymes. In these systems, the cDNA encoding for a specific enzyme is transfected into a cell host (e.g. Escherichia coli, yeast, insect cells) and the expressed enzymes can be isolated and utilized in metabolic studies (Ring and Wrighton, 2000). Although recombinant human enzymes are routinely used, they have the same limitations as purified enzymes when trying to extrapolate results to the in vivo situation.

75 Predictability of metabolic antiepileptic drug interactions

Microsomes or other subcellular fractions prepared from human livers represent a ready source of enzymes responsible for drug metabolism and, therefore, a pri- mary tool for in vitro interaction studies (Ring and Wrighton, 2000). Human liver samples, frozen and stored at approximately 80°C, retain their metabolic poten- tial over a long period of time. These microsomal preparations contain the various human cytochromes in proportion to their quantitative representation in human liver in vivo. In these systems, the contribution of a given isoenzyme to the metab- olism of a test drug can be assessed by using different approaches such as the inves- tigation of changes in biotransformation rate after addition of a specific inhibitor of that isoenzyme. Likewise, the potential ability of the test drug to act as an enzyme inhibitor can be investigated by assessing its effect on isoenzyme-specific metabolic pathways of probe substrates added in the system. The data obtained with microsomes from human liver may have a greater relevance to the in vivo sit- uation than those obtained through the use of isolated enzyme systems (Ring and Wrighton, 2000). This is mainly due the similarity of the lipid and enzyme envi- ronment to the in vivo situation. It should be noted that microsomal fractions may also be prepared from tissues other than the liver, in order to investigate extrahepatic drug metabolism.

Cell-based techniques

The two cell-based systems most commonly utilized to study drug metabolism include cultured hepatocytes and liver slices. The use of an intact cell system is, at least in theory, ideal because of its greater physiologic relevance to the intact organism, as it contains both phase I and II enzymes along with the appropriate cofactors found in vivo. A major advantage of this system, with special reference to primary cultures of human hepatocytes, is the possibility of studying the induc- tion potential of a test compound, an effect which cannot be evaluated in in vitro enzyme systems (Lin and Lu, 1998; Ring and Wrighton, 2000).

Primary cultures of human hepatocytes as a tool to predict enzyme-inducing potential at the preclinical level are advantageous over the use of animal models in vivo because interspecies differences in substrate specificity and regulation of expression preclude extrapolation of animal data to humans. On the other hand, a drawback in the use of primary hepatocytes is the requirement for fresh human tissue.

Prediction of metabolic drug interactions based on in vitro data

Two complementary approaches have been developed to predict potential drug interactions in vivo based on in vitro data: (a) identification of the enzymes (CYP isoforms or other drug-metabolizing enzymes) responsible for the biotransformation

76 Edoardo Spina et al.

of a test drug; (b) determination of the potential of the test drug to inhibit or induce the activities of the various drug-metabolizing enzymes. The first approach allows prediction of interactions affecting the metabolism of the test compound (i.e. interactions affecting the test drug as a substrate), the second allows prediction of any effect that the test compound may have on the metabolism of other drugs (i.e. interactions where the test drug may act as an inducer or an inhibitor).

The test drug as a substrate (target for interactions)

Prediction of interactions that may affect the test drug requires (a) knowledge of the enzyme systems responsible for its biotransformation, and (b) knowledge of the influence of other drugs on such enzyme systems.

Identification of the enzymes responsible for the metabolism of the test drug

Identification of the isoenzyme(s) responsible for the metabolism of a given drug is the major prerequisite for rational prediction of metabolic drug interactions. To this purpose, however, it is also important to determine the relative contribution of each isoenzyme, and related metabolic pathways, to the overall elimination of that drug in vivo. Apart from prediction of drug interactions, this information may be used to anticipate the possible occurrence of genetic polymorphisms (in the case of involvement of CYP2D6, CYP2C9 or CYP2C19), as well as the likelihood of sub- stantial extrahepatic contributions to drug metabolism, as most frequently seen with CYP3A substrates that may be biotransformed in part in the gastrointestinal mucosa (Dresser et al., 2000).

Information on the CYP isoforms responsible for the oxidative metabolism can be obtained by using a general in vitro strategy (Lin and Lu, 1998; Ring and Wrighton, 2000). This may involve assessment of: (a) catalytic activity in human liver microsomes; (b) correlation of this activity with markers for known CYP iso- forms; (c) catalytic activity in cDNA-based vector systems; (d) catalytic activity in purified enzymes; (e) effects of selective inhibitors; and (f) immunoinhibition with monoclonal or polyclonal antibodies against various CYP isoforms. Each approach has its advantages and disadvantages, and a combination of approaches is usually required.

Usually, studies begin with a kinetic analysis of the in vitro formation of metabolites in human liver microsomes. These analyses allow determination of the oxidative metabolite(s) of the test drug, and of the range of enzymes that may be able to form a particular metabolite (Lin and Lu, 1998). The formation of each of the metabolites is determined over a wide range of substrate concentrations. The apparent kinetic parameters such as KM (Michaelis–Menten constant, representing the concentration of the substrate that results in half-maximal velocity) and Vmax (maximal velocity of the reaction) for the enzyme(s) responsible for the formation

77 Predictability of metabolic antiepileptic drug interactions

of a particular metabolite may then be calculated. In this system, a first indication of the isoenzyme(s) involved in the production of the metabolite may subsequently be obtained through studies correlating the formation rate of the metabolite to the activities of various enzyme isoforms in microsomes from different individuals. Isoform-selective catalytic activities for the major CYPs involved in drug metabo- lism are reported in Table 5.1. Identification of the isoenzyme(s) responsible for the formation of the metabolite may also be obtained by using cDNA expressed enzymes or purified enzymes. Another approach to determine the role of the vari- ous drug-metabolizing enzymes is through use of isoenzyme-specific inhibitors. The capacity of a relatively specific chemical inhibitor (Table 5.1) to inhibit the biotransformation of a given drug to its initial metabolite constitutes evidence supporting the participation of the corresponding isoenzyme. A similar approach to confirm the role of specific enzymes makes use of antibodies with relatively spe- cific inhibitory activity against individual isoenzymes.

In order to estimate the contribution of a given CYP isoform to total drug clear- ance, the information obtained in vitro must be combined with the results from preliminary in vivo quantitative metabolic studies (sometimes carried out with radiolabeled drug) that measure the fraction of dose eliminated by each pathway (including renal excretion).

Predicting interactions affecting the metabolism of the test drug

Once the contribution of different isoenzymes to the metabolism of a given drug has been elucidated, prediction of interactions affecting the clearance of that drug can easily be made. This prediction is based on existing knowledge of the influence that other drugs have on the activity of the same isoenzymes. Moreover, if the influence of a potential interfering agent is not known, this can be easily tested in the in vitro systems described above.

The main isoenzymes responsible for the metabolism of most AEDs have been identified (Riva et al., 1996; Anderson, 1998) and available data are summarized in Table 5.2. Carbamazepine may serve as an example of how this information can be applied to prediction of drug interactions (Levy, 1995). Identification of CYP3A4 as the primary catalytic enzyme for the main clearance pathway of carbamazepine allows the anticipation that any compound known to inhibit CYP3A4 activity at therapeutically meaningful concentrations has the potential to decrease carba- mazepine clearance and to increase plasma carbamazepine concentration at steady state. The validity of this prediction is supported by a large bulk of experimental and clinical studies. For example, different compounds known to inhibit CYP3A4 activity such as the calcium-channel blockers diltiazem and verapamil, the macrolide antibiotics troleandomycin and erythromycin, the antidepressants viloxazine and nefazodone, and the antifungals ketoconazole and fluconazole, have been reported

Table 5.2 Elimination pathways for the major AEDs. Fraction of absorbed dose cleared by metabolic and renal elimination refers to average values described for patients on monotherapy. CYP isoforms responsible for metabolic clearance of each drug are shown in brackets (bold characters identify enzymes involved in metabolic pathways responsible for a major proportion of total drug clearance)

Drug

Fraction cleared by CYPs

UGTs Other enzymes Renal

Carbamazepine Ethosuximide Felbamate Gabapentin Lamotrigine Levetiracetam Oxcarbazepinea Phenobarbital Phenytoin Tiagabine Topiramate Valproic acidb Vigabatrin Zonisamide

75% (CYP3A4, 70% (CYP3A4) 15% (CYP3A4, Nil

CYP2C8, CYP1A2) CYP2E1)

15% Negligible 5% Nil Negligible 20% 10% 25% (hydrolysis) 50% Nil Nil 100% 80% Negligible 10% Nil 30% (hydrolysis) 70% 70% Nil 30% Negligible 25% (N-glucosidation) 25% Nil Negligible 5% Nil Not identified 2% Nil Not known 75% 40% 35% ( -oxidation) 5% Nil Nil 100% Negligible 20% (acetylation) 30%

Nil
Nil
5%
30% (CYP2C9, 90% (CYP2C9, 95% (CYP3A4) 25%

CYP2C19, CYP2E1) CYP2C19)

10% (CYP2C9, CYP2A6, CYP2B6) Nil
50% (CYP3A4, CYP2C19, CYP3A5)

a Data refer to the active monohydroxycarbazepine derivative (MHD). Oxcarbazepine is transformed to MHD by ketoreduction catalyzed by cytosolic arylketone reductase.

b Fractions metabolized through various pathways are dose dependent.

79 Predictability of metabolic antiepileptic drug interactions

to cause a clinically significant elevation in plasma carbamazepine concentration (Perucca, 1982; Patsalos et al., 2002). Similarly, it is known that CYP3A4 activity is stimulated by enzyme-inducing AEDs such as phenobarbital and phenytoin: this allows prediction of the ability of these compounds to increase carbamazepine clearance and to reduce plasma carbamazepine concentration in patients with epilepsy (Spina et al., 1996).

There can be many other examples of how knowledge of the isoenzymes respon- sible for the metabolism of an AED can be used to predict interactions affecting the plasma clearance of that drug in vivo. In the case of phenytoin, which is a substrate of CYP2C9 and CYP2C19, clinically documented examples of interactions consis- tent with inhibition of CYP2C9 are those caused by amiodarone, phenylbutazone, propoxyphene, miconazole and valproic acid, while interactions probably due to inhibition of CYP2C19 are caused by ticlopidine, fluoxetine, omeprazole and fel- bamate (Raguenau-Majlessi et al., 2002).

When utilizing only in vitro findings to predict in vivo changes in the pharma- cokinetics of the affected drug, it is important to remember that the extent of inhi- bition or induction of a given pathway as assessed on the basis of in vitro data does not necessarily imply that the total clearance of the affected substrate in vivo will be affected to the same extent. In fact, any change in clearance of the affected sub- strate will also be influenced by other factors, including the degree of inhibition or induction of the affected pathway in vivo (which may not necessarily correspond to the in vitro situation, due to intervention of confounding variables); the contri- bution of the affected pathway to the overall elimination of the substrate; the phar- macokinetic characteristics of the substrate and its route of administration; any influence that the interfering drug may have on alternative metabolic pathways of the substrate. A more detailed discussion of how these factors impact on the pre- diction process, including potential pitfalls, is provided in the section ‘Crucial fac- tors in predicting in vitro–in vivo correlations’.

The test drug as a cause of interactions affecting the metabolism of other drugs

The first step in predicting what effect a test compound may have on the metabo- lism of other drugs consists in the investigation of the influence of that compound on the activity of the various drug-metabolizing isoenzymes. This information is then interpreted by taking into account existing knowledge on the range of drugs which are substrates of the affected enzymes.

Assessment of the influence of the test drug on the activity of drug-metabolizing isoenzymes

In vitro approaches similar to those described in the section ‘Identification of the enzymes responsible for the metabolism of the test drug’ are applicable to the eval- uation of drugs as potential inhibitors of specific enzyme isoforms. Using human

80 Edoardo Spina et al.

liver microsomes or individual enzymes, a series of drugs and/or their metabolites can be screened relatively quickly to determine quantitatively their potency in inhibiting reactions considered to reflect specifically the activity of individual enzyme isoforms. One approach involves the use of a fixed concentration of the probe sub- strate incubated with variable concentrations of the potential inhibitor (Greenblatt et al., 1998). Evaluation of the decrease in metabolite formation rate as a function of the inhibitor’s concentration allows calculation of the 50% inhibitory concen- tration (IC50), i.e. the inhibitor’s concentration at which the reaction rate is reduced by 50%. IC50 values are independent of the specific biochemical mechanism of inhibition and, therefore, they are suitable for comparing the relative potency of a series of inhibitors. On the other hand, when inhibition is competitive, IC50 values depend on substrate concentration: therefore, they cannot be directly applied to in vitro–in vivo scaling models, except when inhibition is established as having a non-competitive mechanism. A second approach to the assessment of inhibitory interactions is based on calculation of the inhibition constant (Ki), which reflects inhibitory potency in a reciprocal fashion (Segel, 1975). Determination of Ki is more time and labour consuming, since it requires the study of multiple substrate concentrations and multiple inhibitor concentrations. Ki is model dependent, since it depends upon the specific mechanism of inhibition, which may not be established. However, Ki is independent of substrate concentration and can be used under some defined conditions for the quantitative in vitro–in vivo scaling of drug interactions. Although Ki is less than or equal to IC50 as a general rule, Ki will be equal to IC50 if inhibition is non-competitive, or if inhibition is competitive and the substrate concentration is far below the reaction KM (Segel, 1975). Both Ki and IC50 provide similar estimates of relative inhibitory potency for a series of inhibitors of a specific reaction.

As discussed in the section ‘Cell-based techniques’, in vitro systems can also be used to estimate enzyme-inducing potential. These experiments are far more com- plex, time consuming and expensive, as they involve the use of primary cultures of hepatocytes (Li et al., 1997). Evaluation of changes in the activity of specific isoen- zymes can be obtained by applying the techniques described in the sections above.

Predicting effects of the test drug on the metabolism of other drugs

The effects of the major AEDs on various drug-metabolizing enzymes are summa- rized in Table 5.3. Once it has been established that an AED inhibits or induces the activity of a given isoenzyme, then one can predict that the metabolism of sub- strates of the same isoenzyme will be correspondingly affected. A list of substrates of individual CYP isoenzymes is reported in Table 5.1: for example, if a drug inhibits CYP1A2, then one can predict that the CYP1A2-mediated pathways of substrates such as amitryptyline, fluvoxamine, mirtazepine, clozapine, olanzapine,

81 Predictability of metabolic antiepileptic drug interactions

Table 5.3 Effects of AEDs on the most common drug-metabolizing enzyme systems

Drug Effect Enzymes involved

Carbamazepine

Ethosuximide Felbamate

Gabapentin Lamotrigine Levetiracetam Oxcarbazepine

Phenobarbital/ primidone

Phenytoin

Tiagabine Topiramate

Valproic acid

Vigabatrin

Zonisamide

Inducer

None (?) Inhibitor

Inducer None Negligible None Inhibitor Inducer

Inducer

None Inhibitor Inducer

Inhibitor

None

None (?)

CYP1A2, CYP2B6, CYP2C, CYP3A4 Microsomal EH
UGT

CYP2C19 -oxidation CYP3A4

UGT (weak autoinduction)

CYP2C19
CYP3A4 (weaker induction compared

with carbamazepine)
UGT (weaker induction compared

with carbamazepine)
CYP1A2, CYP2B6, CYP2C, CYP3A Microsomal EH
UGT
CYP1A2, CYP2B6, CYP2C, CYP3A4 Microsomal EH
UGT

CYP2C19 (weak inhibition)
CYP3A4 (weaker induction compared

with carbamazepine) CYP2C9

Microsomal EH UGT

haloperidol, theophylline, caffeine and phenacetin will be inhibited. The extent of inhibition will depend on the inhibiting potency and on the concentration (dosage) of the inhibitor, but the concentration of the substrate may also play a role. Similar considerations apply to predictions of drug interactions mediated by enzyme induction.

A good correlation between the ability to inhibit various CYPs in vitro and the in vivo inhibitory interaction profile has been established for a number of AEDs,

82 Edoardo Spina et al.

including valproic acid (Scheyer, 2002) and felbamate (Glue et al., 1997). As dis- cussed in the section ‘Predicting interactions affecting the metabolism of the test drug’, it should be noted that the predicted extent of inhibition or induction of a given pathway does not necessarily imply that the total clearance of the affected substrate in vivo will be affected to the same extent. In vivo changes may be influ- enced by a number of variables such as the action of other metabolites, the acces- sibility of the inhibitor or inducer to the enzyme, the contribution of the affected pathway to the overall elimination of the affected drug, the pharmacokinetics char- acteristics of the affected drug and its route of administration, and any influence that the interfering drug may have on alternative metabolic pathways. These factors are discussed concisely in the section below.

Crucial factors in predicting in vitro–in vivo correlations

Information on the drug-metabolizing enzyme systems and their substrates, inhibitors and inducers may be of a great value for clinicians to anticipate and eventually avoid potential interactions. Co-administration of two substrates of the same enzyme, or co-administration of a substrate with an inhibitor or an inducer, entails the possibility of a drug interaction. As a consequence, plasma concentra- tions of the co-administered drugs may be increased or decreased, resulting in clin- ical toxicity or diminished therapeutic effect. Dosage adjustments may then be required to avoid adverse effects or therapeutic failure. However, not all theoreti- cally possible drug interactions that are predicted from in vitro studies will occur in vivo, and some may not be clinically significant anyway. As suggested by Sproule et al. (1997), different aspects including drug-related, patient-related and epidemi- ological factors must be taken into account when evaluating the potential occur- rence, extent and clinical significance of a metabolic drug interaction.

With respect to prediction of whether an interaction will occur in the clinical situation, it should be pointed out that, although it is relatively easy to assess a drug interaction in vitro, the correct interpretation and extrapolation of in vitro data to the in vivo situation may be complicated by various factors and require a good understanding of pharmacokinetic principles (Bertz and Granneman, 1997; Lin and Lu, 1998; Levy and Trager, 2000). One of the most important factors to be considered is whether in vitro drug interaction studies have utilized clinically relevant con- centrations of inhibitor (or inducer) and substrate. While the use of suprathera- peutic concentrations may obviously result in a drug interaction in vitro but not in vivo, it may not be easy to determine whether a given range of concentrations tested in vitro is therapeutically relevant. For example, reference to drug concen- trations measured at steady state in patients receiving therapeutic dosages may not provide an adequate estimate of the concentration of the interacting (or affected)

83 Predictability of metabolic antiepileptic drug interactions

drug at the site of metabolism in vivo, due to the confounding effect of binding to proteins, transport systems and presence of other interfering endogenous and exogenous metabolites. A potentially important factor affecting in vitro drug inter- action studies is represented by the protein concentration of microsomes. The Ki values of an inhibitor may be overestimated at high microsomal protein concen- trations as a result of the depletion of the inhibitor by non-specific binding to microsomal proteins and/or microsomal metabolism. Moreover, the specificity of chemical inhibitor probes is of concern for the interpretation of in vitro studies. No inhibitory probe is completely specific for its corresponding isoform and all ultimately become non-specific at high concentrations. In view of these consider- ations, no prediction can be expected to be 100% accurate, and both false positive and false negatives need to be anticipated.

While the above limitations should be understood, it is nevertheless true that consideration of a number of factors is essential in maximizing the probability of making accurate predictions about the occurrence, and potential clinical importance, of specific drug interactions. These factors will be briefly discussed in the remaining part of this chapter.

The therapeutic index of the substrate

In general, interactions affecting medications with a narrow therapeutic index (e.g. phenytoin, anticoagulants, immunosuppressants or anticancer drugs) are more likely to be clinically relevant than interactions affecting drugs with a broad mar- gin of safety (e.g. gabapentin, penicillin). In fact, given the same degree of inhibi- tion or induction, any change in the plasma levels of the affected substrate is more likely to result in toxic or subtherapeutic values if the substrate has a narrow ther- apeutic index. Of course, the importance of the interaction will also vary depend- ing on whether at baseline the concentration of the affected agent was near the threshold associated with toxicity or therapeutic failure.

Extent of metabolism of the substrate through the affected enzyme

For interactions involving inhibition of drug metabolism, a clinically important change in the plasma concentration of the affected drug can only be expected if the inhibited pathway contributes to a major extent to total drug clearance. For example, inhibition of a pathway which accounts for only 10% of total drug clearance will only increase the concentration of the affected drug by no more than 10%. It should be noted that the relative contribution of a given metabolic pathway to the overall elimination of a drug may vary across individuals, an observation which may explain why some interactions show considerable interindividual variability in their occurrence or extent (Gatti et al., 2001). In some cases, the relative contri- bution of a given isoenzyme to total drug clearance is concentration dependent.

84 Edoardo Spina et al.

A good example for this is phenytoin, whose major metabolic pathway, p– hydroxylation, is mediated by CYP2C9 and, to a lesser extent, CYP2C19 (Bajpai et al., 1996). At high concentration, the activity of CYP2C9 becomes saturated and the contribution of CYP2C19 to the metabolism of the drug will correspondingly increase. Therefore, a significant impact of CYP2C19 inhibitors on phenytoin dis- position will only be expected to occur at higher concentrations, when the CYP2C19- mediated pathway becomes increasingly important for the elimination of the drug.

For interactions involving enzyme induction, the situation is totally different from that described for enzyme inhibition. In fact, there is theoretically no limit to the increase in the efficiency of a given metabolic pathway when the corresponding isoenzyme(s) have been induced. In other words, enzyme induction could trans- form an initially minor metabolic pathway into a major contributor to the overall elimination of the drug, with a consequent important increase in total drug clearance.

The principles summarized above are well illustrated by the metabolic interac- tions described for felbamate and topiramate. Since CYP3A4 plays only a minor role in the metabolism of felbamate, inhibitors of this isoform would be expected to have only minimal effects on the overall clearance of this drug and, in line with this prediction, felbamate pharmacokinetics have been found not to be signifi- cantly affected by the potent CYP3A4 inhibitor erythromycin (Glue et al., 1997). On the other hand, the total plasma clearance of felbamate is significantly increased and its plasma concentrations are significantly decreased by concomi- tant treatment with the CYP3A4-inducers phenytoin, phenobarbital and carba- mazepine. A similar situation is observed with topiramate, a drug which in healthy subjects is primarily excreted unchanged in urine. Because metabolism is of minor importance in the overall clearance of topiramate, no significant changes in its plasma concentration are expected when an enzyme inhibitor is added for patients receiving topiramate monotherapy. On the other hand, metabolic elimination becomes an important determinant of topiramate clearance in patients treated with enzyme-inducing AEDs, an observation which explains the ability of the latter to decrease plasma topiramate concentration by 40–50% (Perucca and Bialer, 1996). It should be noted that, theoretically, the plasma concentration of fel- bamate and topiramate could be significantly affected by an enzyme inhibitor only when the latter is added on as a third agent in a patient who is already taking an enzyme inducer. This is because it is only in enzyme-induced patients that the con- tribution of metabolism to the overall clearance of these drugs becomes clinically significant.

Role of metabolites

A factor to be considered is whether metabolites have any enzyme inducing or inhibiting effects independent of those of the parent drug. For example, if a

85 Predictability of metabolic antiepileptic drug interactions

metabolite has an inhibiting effect on a given isoenzyme that is not shared by the parent drug, in vitro experiments designed to test the enzyme-inhibiting potential of the parent drug may fail to identify a clinically important interaction.

As discussed above, many metabolites are biologically active and this needs to be considered when predicting the clinical consequence of a drug interaction. If the affected drug has a pharmacologically active (or toxic) metabolite, enzyme inhibition may paradoxically result in decreased pharmacological (or toxicological) effect, while the reverse will be true for enzyme induction. It is also important to consider what influence the interaction is expected to have on the subsequent bio- transformation of the metabolites.

Pharmacokinetic characteristics of the drug and route of drug administration

An example of how pharmacokinetic characteristics can influence the conse- quences of metabolic drug interactions has already been provided in the section ‘Extent of metabolism of the substrate through the affected enzyme’ when discussing the implications of the concentration-dependent pharmacokinetics of phenytoin.

An even more important aspect to be considered is whether the affected drug shows a low or a high extraction ratio in the organ (usually the liver) responsible for its metabolism. In the case of highly extracted drugs, clearance is mainly deter- mined by the blood flow through the eliminating organ, and changes in enzyme activity will have little or no effect on their pharmacokinetics after parenteral administration. However, if metabolism takes place mainly in the liver or in the gut, enzyme induction or inhibition can have a marked effect on the first-pass extraction of these agents and, hence, on their oral bioavailability. These consider- ations provide an explanation for the marked reduction in the bioavailability of high clearance drugs such as ethinylestradiol (Perucca, 1982), lidocaine (Perucca and Richens, 1979) and nisoldipine (Michelucci et al., 1998) in patients taking the enzyme-inducers phenobarbital, carbamazepine or phenytoin.

The pharmacokinetics of drugs which show a low metabolic clearance are not influenced by changes in blood flow, and their plasma concentration is largely determined by drug-metabolizing enzyme activity irrespective of the route of intake. Therefore, enzyme induction and inhibition are expected to affect the steady-state plasma concentration of these drugs after both parenteral and oral administration. For a detailed discussion of these principles, the reader is referred to the seminal work of Wilkinson and Shand (1975).

Complex or biphasic interactions

Enzyme induction and inhibition are not mutually exclusive and may occur at the same time. The ability of a given compound to act as an inducer and as an inhibitor

86 Edoardo Spina et al.

at the same time provides an explanation for the inconsistent and apparently contradictory nature of certain drug interactions. As discussed above, for example, phenobarbital may either decrease or increase the plasma concentration of pheny- toin depending on whether induction or inhibition of phenytoin metabolism prevails in an individual patient (Perucca, 1982). Even more complex is the interaction between phenytoin and warfarin. When phenytoin is started in a patient stabilized on warfarin therapy, phenytoin may initially competitively inhibit the metabolism of warfarin because both phenytoin and S-warfarin are substrates for CYP2C9 and phenytoin has a KM (and therefore a Ki) within its therapeutic range. After an initial increase, the plasma concentration of S-warfarin will then decline within 1–2 weeks because of CYP2C9 induction (Cropp and Bussey, 1997).

Even more complex situations may be observed when other mechanisms of interaction, e.g. altered gastrointestinal absorption, drug displacement from bind- ing sites, or pharmacodynamic interactions, occur simultaneously with changes in enzyme activity. Other complex situations arise in patients receiving combin- ations of three or more drugs, and in this case direct and indirect interactions may become difficult to predict. At times, interactions may actually cancel out reciprocally: for example, the clearance of lamotrigine is markedly enhanced by co-administration of enzyme-inducing AEDs (phenobarbital, carbamazepine and phenytoin) and inhibited by valproate. However, patients receiving lamotrigine in a triple therapy regimen that includes valproate and an enzyme inducer show lam- otrigine clearance values comparable with those observed in patients on lamotrig- ine monotherapy (Jawad et al., 1989).

Other sources of variability

There is a large intersubject variability in the extent and clinical relevance of meta- bolic drug interactions. As discussed above, enzyme induction and inhibition are usually dose dependent, and differences in dosage (or plasma concentration) of the interfering drug are important in determining the occurrence or extent of a drug interaction. Additional sources of variability relate to interindividual differences in the contribution of specific metabolic pathways to overall drug clearance. Age has also been reported to affect response to drug interactions: for example, it has been suggested that the elderly may be less sensitive to enzyme induction (Twum- Barima et al., 1984), even though in a recent study auto- and heteroinduction of carbamazepine metabolism was not found to differ between elderly patients and younger adults (Battino et al., 2003). The role of confounding factors (e.g. the additional influence of enzyme inducers or inhibitors found in the diet or in voluc- tuary substances) also varies considerably across individuals.

In patients receiving drugs metabolized by a polymorphic enzyme, the effects of inhibitors or inducers may vary between phenotypes/genotypes. EMs are generally

87 Predictability of metabolic antiepileptic drug interactions

more susceptible to enzyme inhibition or induction than PMs. For inhibition, this has been most clearly documented for CYP2D6: interactions caused by potent inhibitors of this isoform, i.e. quinidine, are not observed in PMs, who show a genetically determined lack of functional CYP2D6 in their liver (Steiner et al., 1987). Likewise, PMs for CYP2C19 and CYP2C9, which play a role in the metabo- lism of phenytoin, are not expected to be vulnerable to the inhibition of phenytoin metabolism caused by selective inhibitors of the corresponding enzymes. More complex effects can be expected when the genotype/phenotype influences suscep- tibility to drug interactions in an indirect way: for example, the enzyme-inducing effects of 40mg/day omeprazole (a CYP2C19 substrate) on CYP1A2 activity only occurs in PMs for CYP2C19, because only these subjects achieve plasma omeprazole concentrations which are sufficiently high to cause enzyme induction (Rost et al., 1992).

For any given extent of interaction, clinical consequences also vary widely across individuals. As discussed above, interactions are more likely to be clinically signif- icant when the plasma concentration of the affected drug at baseline is closest to the threshold for toxicity or therapeutic failure. Pharmacodynamic factors affect- ing response to any drug concentration are also important. Elderly patients in gen- eral are more prone to adverse drug interactions, not only because they more frequently receive multiple drug therapy but also because they may show increased pharmacodynamic sensitivity to drugs.

Conclusions

Metabolic drug interactions may have important clinical consequences. In the case of AEDs, these interactions are particularly common, due to the fact that many of these agents are potent inducers (or in some case, inhibitors) of the drug- metabolizing enzymes, and they are usually administered chronically, often in combination therapy. In recent years, an improved understanding of the nature of the main isoenzymes responsible for drug metabolism, coupled with advances in methodology for the in vitro assessment of metabolic reactions and interactions, has resulted in major breakthroughs in our ability to predict the occurrence and the in vivo implications of drug interactions. While the methodology still requires some refinement to improve the predictive power, available knowledge is already applied successfully not only in drug discovery (through design and selection of new agents devoid of undesirable interaction potential) and in drug development (though rational identification of drug interactions to be assessed in the clinical setting), but also in making informed decisions when adding or withdrawing co-medications in routine clinical practice.

88 Edoardo Spina et al. REFERENCES

Anderson GD. A mechanistic approach to antiepileptic drug interactions. Ann Pharmacother 1998; 32: 554–563.

Anderson GD, Graves NM. Drug interactions with antiepileptic agents: prevention and manage- ment. CNS Drugs 1994; 2: 268–279.

Bajpai M, Roskos LK, Shen DD, et al. Roles of cytochrome P4502C9 and cytochrome P4502C19 in stereoselective metabolism of phenytoin to its major metabolite. Drug Metab Dispos 1996; 24: 1401–1403.

Battino D, Croci D, Rossini A, et al. Serum carbamazepine concentrations in elderly patients: a case-matched pharmacokinetic evaluation. Epilepsia 2003; 44: 923–929.

Bertilsson L, Dahl ML, Dalen P, et al. Molecular genetics of CYP2D6: clinical relevance with focus on psychotropic drugs. Br J Clin Pharmacol 2002; 53: 111–122.

Bertz RJ, Granneman GR. Use of in vitro and in vivo data to estimate the likelihood of metabolic pharmacokinetic interactions. Clin Pharmacokinet 1997; 32: 210–258.

Bock KW, Gschaidmeier H, Heel H, et al. Functions and transcriptional regulation of PAV- inducible human UDP-glucuronosyltransferases. Drug Metab Rev 1999; 31: 411–422.

Brandolese R, Scordo MG, Spina E, et al. Severe phenytoin intoxication in a subject homozygous for CYP2C9*3. Clin Pharmacol Ther 2001; 70: 391–394.

Cropp JS, Bussey HI. A review of enzyme induction of warfarin metabolism with recommenda- tions for patient management. Pharmacotherapy 1997; 17: 917–928.

Dickins M, Chen C. Lamotrigine. Chemistry, biotransformation and pharmacokinetics. In Antiepileptic Drugs, 5th edn. R. H. Levy, R. H. Mattson, B. S. Meldrum, E. Perucca, eds. Philadelphia: Lippincott Williams & Wilkins, 2002: 370–379.

Doose DR, Wang SS, Padmanabhan, et al. Effect of topiramate or carbamazepine on the phar- macokinetics of an oral contraceptive containing norethindrone and ethinyl estradiol in healthy obese and non-obese female subjects. Epilepsia 2003; 44: 540–549.

Dresser GK, Spence D, Bailey DG. Pharmacokinetic–pharmacodynamic consequences and clin- ical relevance of cytochrome P450 3A4 inhibition. Clin Pharmacokinet 2000; 38: 41–57.

Evans DAP, Mahgoub A, Sloan TP, et al. A family and population study of the genetic poly- morphism of debrisoquine oxidation in a white British Population. J Med Genet 1980; 17: 102–105.

Fattore C, Cipolla G, Gatti G, et al. Induction of ethinylestradiol and levonorgestrel metabolism by oxcarbazepine in healthy women. Epilepsia 1999; 40: 783–787.

Fuerst RH, Graves NM, Leppik IE, et al. Felbamate increases phenytoin but decrease carba- mazepine concentrations. Epilepsia 1988; 29: 488–491.

Fuhr U. Induction of drug metabolizing enzymes: pharmacokinetic and toxicological conse- quences in humans. Clin Pharmacokinet 2000; 38: 493–504.

Gatti G, Furlanut M, Perucca E. Interindividual variability in the metabolism of antiepileptic drugs and its clinical implications. In Interindividual Variability in Human Drug Metabolism. G. M. Pacifici, O. Pelkonen, eds. London: Taylor and Francis Ltd, 2001: 157–180.

Gidal BE, Zupanc ML. Potential pharmacokinetic interaction between felbamate and pheno- barbital. Ann Pharmacother 1994; 28: 455–458.

89 Predictability of metabolic antiepileptic drug interactions

Glue P, Banfield CR, Perhach JL, et al. Pharmacokinetic interactions with felbamate: in vitro–in vivo correlation. Clin Pharmacokinet 1997; 33: 214–224.

Goldstein JA. Clinical relevance of genetic polymorphism in the human CYP2C subfamily. Br J Clin Pharmacol 2001; 52: 349–355.

Gonzalez FJ, Ueno T, Umeno M, et al. Microsomal ethanol oxidizing system: transcriptional and posttranscriptional regulation of cytochrome P450, CYP2E1. Alcohol Alcohol 1991; 1(Suppl.): 97–101.

Greenblatt DJ, Von Moltke LL, Harmatz JS, et al. Drug interactions with newer antidepressants: role of human cytochromes P450. J Clin Psychiat 1998; 59(Suppl. 15): 19–27.

Guengerich FP. Comparisons of catalytic selectivity of cytochrome P450 subfamily enzymes from different species. Chem Biol Interact 1997a; 106: 161–182.

Guengerich FP. Role of cytochrome P450 enzymes in drug–drug interactions. Adv Pharmacol 1997b; 43: 7–35.

Hiller A, Nguyen N, Strassburg CP, et al. Retigabine N-glucuronidation and its potential role in enterohepatic circulation. Drug Metab Dispos 1999; 27: 605–612.

Hooper WD, Franklin ME, Glue P, et al. Effect of felbamate on valproic acid disposition in healthy volunteers: inhibition of beta-oxidation. Epilepsia 1996; 37: 91–97.

Jawad S, Richens A, Goodwin G, et al. Controlled trial of lamotrigine (Lamictal) for refractory partial seizures. Epilepsia 1989; 30: 356–363.

Jin C, Miners JO, Lillywhite KJ, et al. Complementary deoxyribonucleic acid cloning and expres- sion of a human liver uridine diphosphate-glucuronosyltransferase glucuronidating car- boxylic acid-containing drugs. J Pharmacol Exp Ther 1993; 264: 475–479.

Johansson I, Lundqvist E, Bertilsson L, et al. Inherited amplification of an active gene in the cytochrome P450 CYP2D locus as a cause of ultrarapid metabolism of debrisoquine. Proc Natl Acad Sci 1993; 90: 11825–11829.

Kerr BM, Rettie AE, Eddy AC, et al. Inhibition of human liver microsomal epoxide hydrolase by valproate and valpromide: in vitro/in vivo correlation. Clin Pharmacol Ther 1989; 46: 82–93. Kerr BM, Thummel KE, Wurden CJ, et al. Human liver carbamazepine metabolism: role of CYP3A4

and CYP2C8 in the 10,11-epoxide formation. Biochem Pharmacol 1994; 47: 1969–1979.
Kondo T, Otani K, Hirano T, et al. The effects of phenytoin and carbamazepine in serum concentra-

tions of mono-unsaturated metabolites of valproic acid. Br J Clin Pharmacol 1990; 29: 116–119. Kroetz DL, Loiseau P, Guyot M, et al. In vivo and in vitro correlation of microsomal epoxide

hydrolase inhibition by progabide. Clin Pharmacol Ther 1993; 54: 485–497.
Lamba JK, Lin YS, Schuetz EG, et al. Genetic contribution to variable human CYP3A-mediated

metabolism. Adv Drug Deliv Rev 2002; 54(10): 1271–1294.
Levy RH. Cytochrome P450 isozymes and antiepileptic drug interactions. Epilepsia 1995;

36(Suppl. 5): S8–S13.
Levy RH, Trager WF. From in vitro to in vivo: an academic perspective. In Metabolic Drug

Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds.

Philadelphia: Lippincott Williams & Wilkins, 2000: 21–27.
Li AP, Maurel P, Gomez-Lechon MJ, et al. Preclinical evaluation of drug–drug interaction poten-

tial: present status of the application of primary human hepatocytes in the evaluation of cytochrome P450 induction. Chem Biol Interact 1997; 107: 5–16.

90 Edoardo Spina et al.

Lin JH, Lu AYH. Inhibition and induction of cytochrome P450 and the clinical implications. Clin Pharmacokinet 1998; 35: 361–390.

Lindhout D. Pharmacogenetics and drug interactions: role in antiepileptic-drug-induced terato- genesis. Neurology 1992; 42: 43–47.

Liston HL, Markowitz JS, Devane L. Drug glucuronidation in clinical psychopharmacology. J Clin Psychopharmacol 2001; 21: 500–515.

Mackenzie PI, Owens IS, Burchell B, et al. The UDP glycosyltransferase gene superfamily: rec- ommended nomenclature update based on evolutionary divergence. Pharmacogenetics 1997; 7: 255–269.

May TW, Rambeck B, Jurgens U. Influence of oxcarbazepine and methsuximide on lamotrigine concentrations in epileptic patients with and without valproic acid co-medication: results of a retrospective study. Ther Drug Monit 1999; 21: 175–181.

Meyer UA. The molecular basis of genetic polymorphism of drug metabolism. J Pharm Pharmacol 1994; 46(Suppl. 1): 409–415.

Michelucci R, Cipolla R, Passatelli D, et al. Reduced plasma nisoldipine concentrations in phenytoin-treated patients with epilepsy. Epilepsia 1998; 37: 1107–1110.

Miners JO, Birkett DJ. Cytochrome P4502C9: an enzyme of major importance in human drug metabolism. Br J Clin Pharmacol 1998; 45: 525–538.

Miners JO, McKinnon RA. CYP1A. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 61–73.

Nakajima M, Yokoy T, Mizutani M, et al. Genetic polymorphism in the 5 -flanking region of human CYP1A2 gene: effect on the CYP1A2 inducibility in humans. J Biochem 1999; 125: 803–808.

Nallani SC, Glauser TA, Hariparsad N, et al. Dose-dependent induction of cytochrome P450 (CYP) 3A4 and activation of pregnane X receptor by topiramate. Epilepsia 2003; 44: 1521–1528. Nelson DR, Koymans L, Kamataki T, et al. P450 superfamily: update on new sequences, gene

mapping, accession numbers and nomenclature. Pharmacogenetics 1996; 6: 1–42.
Omcienski CJ. Epoxide hydrolases. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins,

2000: 205–214.
Parker AC, Pritchard P, Preston T, et al. Induction of CYP1A2 activity by carbamazepine in chil-

dren using the caffeine breath test. Br J Clin Pharmacol 1998; 45: 176–178.
Patsalos PN, Froscher W, Pisani F, et al. The importance of drug interactions in epilepsy therapy.

Epilepsia 2002; 43: 365–385.
Perucca E. Clinical consequences of microsomal enzyme induction by antiepileptic drugs.

Pharmaco Ther 1978; 2: 285–314.
Perucca E. Pharmacokinetic interactions with antiepileptic drugs. Clin Pharmacokinet 1982; 7: 57–84. Perucca E. Clinical implications of hepatic microsomal enzyme induction by antiepileptic drugs.

Pharmacol Ther 1987; 33: 139–144.
Perucca E. The clinical pharmacology and therapeutic use of the new antiepileptic drugs. Fund

Clin Pharmacol 2001; 15: 405–417.
Perucca E, Bialer M. The clinical pharmacokinetics of the newer antiepileptic drugs. Focus on

topiramate, zonisamide and tiagabine. Clin Pharmacokinet 1996; 31: 29–46

91 Predictability of metabolic antiepileptic drug interactions

Perucca E, Richens A. Reduction of oral availability of lignocaine by induction of first-pass metabolism in epileptic patients. Br J Clin Pharmacol 1979; 8: 21–31.

Perucca E, Richens A. Biotransformation. In Antiepileptic Drugs, 4th edn. R. H. Levy, R. H. Mattson, B. Meldrum, eds. New York: Raven Press, 1995: 31–50.

Perucca E, Hedges A, Makki KA, et al. A comparative study of the enzyme inducing properties of anticonvulsant drugs in epileptic patients. Br J Clin Pharmacol 1984; 18: 401–410.

Pisani F, Haj-Yehia A, Fazio A, et al. Pharmacokinetics of valnoctamide in epileptic patients and its interaction with carbamazepine: in vitro/in vivo correlation. Epilepsia 1993; 34: 954–959.

Raguenau-Majlessi I, Bajpai M, Levy RH. Phenytoin and other hydantoins: interactions with other drugs. In Antiepileptic Drugs, 5th edn. R. H. Levy, R. H. Mattson, B. S. Meldrum, E. Perucca, eds. Philadelphia: Lippincott Williams & Wilkins, 2002: 581–590.

Raucy J, Carpenter SP. CYP2E1. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 95–113.

Rendic S, Di Carlo FJ. Human cytochrome P450 enzymes: a status report summarizing their reactions, substrates, inducers, and inhibitors. Drug Metab Rev 1997; 29: 413–580.

Rettie AE, Koop DR, Haining RL. CYP2C. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 75–86.

Ring BJ, Wrighton SA. Industrial viewpoint: application of in vitro drug metabolism in various phases of drug development. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 29–39.

Riva R, Albani F, Contin M, et al. Pharmacokinetic interactions between antiepileptic drugs: clin- ical considerations. Clin Pharmacokinet 1996; 31: 470–493.

Rosenfeld WE, Doose DR, Walker SA, et al. Effect of topiramate on the pharmacokinetics of an oral contraceptive containing norethindrone and ethinyl estradiol in patients with epilepsy. Epilepsia 1997; 38: 317–323.

Rost KL, Brosicke H, Brockmoller J, et al. Increase of cytochrome P450IIA activity by omepra- zole: evidence by the 13C-(N-3-methyl)-caffeine breath test in poor and extensive metaboliz- ers of S-mephenytoin. Clin Pharmacol Ther 1992; 52: 170–180.

Saano V, Glue P, Benfield CR, et al. Effects of felbamate on the pharmacokinetics of a low-dose combination oral contraceptive. Clin Pharmacol Ther 1995; 58: 523–531.

Sachdeo RC, Sachdeo SK, Levy RH, et al. Topiramate and phenytoin pharmacokinetics during repet- itive monotherapy and combination therapy to epileptic patients. Epilepsia 2002; 43: 691–696. Sachse C, Brockmoller J, Bauer S, et al. Functional significance of a C → A polymorphism in intron 1

of the cytochrome P450 CYP1A2 gene tested with caffeine. Br J Clin Pharmacol 1999; 47: 445–449. Scheyer RD. Valproic acid: drug interactions. In Antiepileptic Drugs, 5th edn. R. H. Levy, R. H. Mattson, B. S. Meldrum, E. Perucca, eds. Philadelphia: Lippincott Williams & Wilkins,

2002: 801–807.
Schuetz EG, Beck WT, Schuetz JD. Modulators and substrates of P-glycoprotein and cytochrome

P4503A coordinately up-regulate these proteins in human colon carcinoma cells. Mol Pharmacol 1996; 49: 311–318.

92 Edoardo Spina et al.

Segel IH. Enzyme Kinetics. New York, NY: John Wiley & Sons, 1975.
Shimada T, Yamazaki H, Mimura M, et al. Interindividual variations in human liver cytochrome

P450 enzymes involved in oxidation of drugs, carcinogens, and toxic chemicals: studies with human liver microsomes of 30 Japanese and 30 Caucasians. J Pharmacol Exp Ther 1994; 270: 414–423.

Sogawa K, Fujii-Kuriyama Y. Ah receptor, a novel ligand activated transcription factor. J Biochem 1997; 122: 1075–1079.

Spina E, Pisani F, Perucca E. Clinically significant pharmacokinetic drug interactions with car- bamazepine: an update. Clin Pharmacokinet 1996; 3: 198–214.

Sproule BA, Naranjo CA, Bremner KE, et al. Selective serotonin reuptake inhibitors and CNS drug interactions: a critical review of the evidence. Clin Pharmacokinet 1997; 33: 454–471.

Steiner E, Dumont E, Spina E, et al. Inhibition of desipramine 2-hydroxylation by quinidine and quinine in rapid and slow hydroxylators. Clin Pharmacol Ther 1987; 44: 431–435.

Sueyoshi T, Kawamoto T, Zelko I, et al. The repressed nuclear receptor CAR respond to pheno- barbital in activating the human CYP2B6 gene. J Biol Chem 1999; 274: 6043–6046.

Thummel KE, Kunze KL, Shen DD. Metabolically-based drug–drug interactions: principles and mechanisms. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 3–19.

Twum-Barima Y, Finnigan T, Habasch AI, et al. Impaired enzyme induction by rifampicin in the elderly. Br J Clin Pharmacol 1984; 17: 595–596.

Wen X, Wang JS, Kivisto KT, et al. In vitro evaluation of valproic acid as an inhibitor of human cytochrome P450 isoforms: preferential inhibition of cytochrome P450 2C9 (CYP2C9). Br J Clin Pharmacol 2001; 52: 547–553.

Wilkinson GR, Shand DG. A physiological approach to hepatic drug clearance. Clin Pharmacol Ther 1975; 18: 377–390.

Wrighton SA, Thummel KE. CYP3A. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 115–133.

Zaccara G, Gangemi PF, Bendoni L, et al. Influence of single and repeated doses of oxcarbazepine on the pharmacokinetics profile of felodipine. Ther Drug Monit 1993; 15: 39–42.

Zanger UM, Eichelbaum M. CYP2D6. In Metabolic Drug Interactions. R. H. Levy, K. E. Thummel, W. F. Trager, P. D. Hansten, M. Eichelbaum, eds. Philadelphia: Lippincott Williams & Wilkins, 2000: 87–94.

Influence of food and drugs on the bioavailability of antiepileptic drugs

Carlos A. Fontes Ribeiro
Department of Pharmacology, Faculty of Medicine, University of Coimbra, Coimbra, Portugal

Introduction

Whenever two or more agents are used in combination the potential for interactions can occur. These interactions can occur at the pharmacodynamic and/or pharmaco- kinetic level. Pharmacokinetic interactions are by far the most frequent and result in the modification of blood or tissue drug concentration as a consequence of alterations in absorption, distribution, metabolism, or elimination of a drug.

Drugs with a narrow therapeutic range or low therapeutic index are more likely to be associated with clinically important interactions. As far as antiepileptic drugs (AEDs) are concerned, they may interact with each other when used in combina- tion therapy, and with other non-epilepsy-related drugs or with over-the-counter medications. Furthermore, food and many excipient components of pharmaceutical formulations may also interact with AEDs. This chapter deals with interactions which occur before (pharmaceutical interactions) and during (pharmacokinetic interactions) absorption.

General principles

Since many AEDs are sparingly soluble in aqueous solutions, they are sensitive to any effects that alter solubility, dissolution, or gastrointestinal motility. The delivery of drugs into the circulation may be altered by physicochemical interactions that occur prior to absorption. For example, drugs may interact in an intravenous solution to produce an insoluble precipitate or may be damaged by light (Figueiredo et al., 1993). Moreover, in the gut, drugs may chelate with metal ions or adsorb to resins. Thus the absorption of a particular drug is profoundly influenced by a great number of factors, which can be classified as follows:

(a) chemical characteristics and formulation, (b) food and fluid intake,

94 Carlos A. Fontes Ribeiro

Absorption, regardless of the site, is dependent upon drug solubility. Drugs administered in an aqueous solution are more rapidly absorbed than those admin- istered in an oily solution, a suspension, or in a solid form because they mix more readily with the aqueous phase at the absorptive site. For those drugs administered as a solid form, the rate of dissolution may be the limiting factor in their absorption.

Large fluid intake results in faster emptying of the stomach due to the distension of the stomach wall (Deutsch et al., 1991). Thus, a drug that is ingested with large volume of fluid will travel faster into the small intestine, ensuring a better and more complete absorption.

Whilst the effect of large fluid volume on the absorption of a drug is predictable, the effect of food is unpredictable. As a general rule, after the ingestion of solid food, the emptying time of the stomach is decreased and intestinal motility and splanchnic blood flow are increased. However, an increase in the extent of drug dissolution in the stomach, as a result of meal prolongation of gastric residence time, does not appear to contribute substantially to fed-state increases in drug plasma concentrations that are observed when a lipid meal is co-administered (Miles et al., 1997). One hypothesis is that the solid meal may enhance the pancreatic secretion thus providing a greater fluid volume for drug dissolution in the small intestine (Miles et al., 1997).

Several drugs can also interfere with the physiologic conditions and function of the gastrointestinal tract and therefore alter the absorption of other drugs. These interactions might be the consequence of altered pH, decreased or increased motility, toxic effects on mucosa and changes in splanchnic blood flow. Therefore, antacids could raise the pH of gastric juice; metoclopramide and other gastrokinetic drugs (cisapride and domperidone) accelerate stomach emptying; propantheline retards stomach emptying; laxative agents decrease the intestinal transit time; cytostatic agents and antibiotics can damage the intestinal mucosa or the normal bacterial flora. Although all of these changes may result in a modified rate and/or extent of absorption, it is not possible to predict whether or not the interaction will be of clinical significance.

Another mechanism of interaction is via the cytochrome P450 (CYP) isoen- zymes that are present in the gut and which contribute to the first-pass metabolism of some drugs. For instance, the isoenzyme CYP3A4 is abundant in the gut and can be stimulated by carbamazepine, phenytoin, phenobarbital and primidone thus reducing the plasma concentrations of drugs that are metabolized by CYP3A4. In contrast, CYP3A4 can be inhibited by acetazolamide, macrolide antibiotics, isoniazid, metronidazole, certain antidepressants, verapamil, diltiazem, cimetidine, danazol and

95 Influence of food and drugs on the bioavailability of AEDs

propoxyphene (Spina et al., 1996), increasing plasma concentrations of drugs that are metabolized by CYP3A4.

Disease states also influence absorption of drugs: in diseases accompanied by decreased motility of the stomach the absorption of drugs is generally delayed or reduced, while in diseases with faster gastric emptying absorption is enhanced. Subjects with an ileojejunal bypass are likely to require increase oral dosages (e.g. phenytoin) to achieve an optimal plasma concentration (Kennedy and Wade, 1979). When blood flow is reduced, as can occur in cardiac failure or shock, or after drugs, the rate of absorption is generally diminished, although the extent is unpredictable. In contrast, increased blood flow may serve to augment absorption.

On the other hand, if optimizing drug therapy aims at achieving and maintaining therapeutic and safe drug concentrations, the sustained release formulations can be useful. The sustained release formulations are designed to be absorbed by an efficient gastrointestinal system that is not limited to certain sites along the gas- trointestinal tract. Nevertheless, in the gastrointestinal tract some interactions can occur, namely with drugs which modify intestinal motility. Thus it is possible for all the drug dose, which is encapsulated in the sustained release formulation, to be released at once through some accidental chemical or physiological mechanism. In this setting, the patient could be in danger of a drug overdose (Bialer, 1992). If the drug has a long half-life the probability of interactions during absorption is lower, as was verified for topiramate.

Regarding the formulation and administration of AEDs by the rectal route, there are only a few studies. Generally, drug administration by the rectal route is not acceptable to patients, particularly since absorption can be interrupted by defaeca- tion (de Boer et al., 1982).

Intramuscular drug absorption can be slow, erratic and incomplete, and this has been particularly demonstrated for phenytoin (Tuttle, 1977). Factors which play a role in the bioavailability of these medications include the water solubility of the drug, dispersion of the injected solution and blood flow at the muscle site.

Finally, chronovariability in absorption–elimination parameters (such as peak concentration and peak time) has been observed for many AEDs. Sometimes these changes had been attributed to interactions with food or drugs. The fasting-induced increase in hepatic glucuronidation during the night and the relative inactivity of the gut during this period may explain variations in circulation plasma drug concentra- tions (Chaudhary et al., 1993). Loiseau et al. (1982) found diurnal variations in steady-state plasma concentrations of valproic acid, when administered by the oral route. Similar findings were observed by Yoshiyama et al. (1989) who reported that Cmax tended to be higher and Tmax shorter in the morning than in the evening. Such circadian variations of pharmacokinetic parameters have also been shown for carba- mazepine (Bruguerolle et al., 1981) but not for phenytoin (Petker and Morton, 1993).

96 Carlos A. Fontes Ribeiro

Interactions with the established AEDs Phenytoin

Phenytoin is the most studied of the AEDs, principally because it has been in use the longest. Much of the knowledge about this drug may be applied to other AEDs. Phenytoin, a weak acid with a pKa of 8.3, is practically insoluble in water. The salt is readily soluble in water but in the acidic medium of the stomach it precipitates after dissolving (Levy, 1976). In relation to the parenteral formulation of phenytoin, despite the fact that this phenytoin salt is water soluble, it precipitates in a large- volume, glucose-containing fluid.

The factors influencing the absorption rate are the particle size, nature of the filler, and whether the free acid or the sodium salt of phenytoin is the active ingre- dient. Thus, the rate of absorption varies considerably among dosage forms. Numerous studies (Cacek, 1986) have shown that phenytoin products from different manufacturers vary in absorption rate and differ in the time to reach maximum concentration, even if the area under the curve (AUC) is adequate. Many generic preparations are more rapidly absorbed and may produce an intolerable fluctua- tion in the plasma phenytoin concentration. In fact, differences in absorption sig- nificant enough to be associated with clinical toxicity have occurred with changes in the excipient (Bochner et al., 1972; Carter et al., 1981) – for instance, when calcium sulfate dihydrate was replaced by lactose, since the calcium compound interferes with phenytoin absorption. Therefore, changes in dosage form or manufacturer should be avoided once a patient’s dosage requirements have been established, as a relatively small decrease or increase in bioavailability can greatly alter the steady- state plasma concentration during chronic administration.

Phenytoin suspensions of the acid have limited clinical utility for two reasons. First, unless well dispersed, precipitation of the drug in the bottle gives rise to doses lower-than-expected initially and higher-than-expected as the container is emp- tied. Second, the usual methods for measurements of liquids, especially with tea- spoons, are inexact. When the phenytoin suspension is put into a unit-dose package, it is important to state on the label whether or not rinsing of the container is needed to ensure proper delivery of the intended dose.

The time to reach the maximum phenytoin plasma concentration after a single oral dose increases with the dose – the greater the dose, the longer the time to reach the peak (Tozer and Winter, 1990). The greatly increased peak time with dose is probably a consequence of two mechanisms. One is the relatively low solubility, slow dissolution and continued absorption of the drug; the other is the capacity- limited metabolism that is associated with phenytoin.

Burstein et al. (2000) reported that the absorption of phenytoin from polyethylene glycol rectal suppositories in healthy subjects is highly variable and unpredictable. Thus, this formulation is not recommended.

97 Influence of food and drugs on the bioavailability of AEDs

Food effect on phenytoin absorption

Food has been found to have variable but modest, usually enhancing, effects on phenytoin absorption (Cacek, 1986). The mechanisms include increased phenytoin dissolution in the stomach, saturation of the first-pass mechanisms, and increased splanchnic blood flow (Melander et al., 1979). High-fat meals appear to increase phenytoin bioavailability (Sekikawa et al., 1980), probably due to a combination of stimulation of bile flow and accelerated dissolution of phenytoin particles or by the delay of the gastric emptying time caused by the fat intake (Hamaguchi et al., 1993). A protein-rich diet had the same effect on phenytoin acid but not on the sodium salt (Kennedy and Wade, 1982). Food may also enhance the delivery of phenytoin from prodrug formulations (e.g. 3-pentanonoyloxymethyl-5,5-diphenylhydantoin and 3-octanoyloxymethyl-5,5-diphenylhydantoin; Stella et al., 1999).

Grapefruit juice, which inhibits the intestinal CYP3A4, does not affect the oral bioavailability of phenytoin (Kumar et al., 1999). A possible explanation for this may relate to the fact that only a small amount (dose) of grapefruit juice was ingested by the subjects investigated in the study and also because phenytoin is not a substrate of CYP3A4.

The absorption of phenytoin is significantly impaired when given concurrently to epileptic patients receiving continuous nasogastric feeds (nutritional formulae). Substantial reduction in steady-state phenytoin plasma concentration have been reported in neurosurgery patients and in normal subjects (Bauer, 1982; Prichard et al., 1987). The most likely mechanism is a reduced bioavailability due to rapid gastro- intestinal transit. In addition, it has been demonstrated that the presence of caseinate salts and calcium chloride may decrease phenytoin absorption (Smith et al., 1988). Binding of phenytoin to the nasogastric tube apparatus has been largely excluded, since the tube is flushed after dosing (Cacek et al., 1986). Decreased plasma phenytoin concentration associated with enteral feeding formulations may increase the risk of seizures (Au Yeung and Ensom, 2000).

Phenytoin and gastrointestinal diseases

The bioavailability of phenytoin may be reduced by gastrointestinal diseases, par- ticularly those associated with increased intestinal motility. Thus, in cases of severe diarrhea, malabsorption syndromes, or gastric resection, decreased bioavailability should be considered (Tozer and Winter, 1990).

Drugs which may affect the gastrointestinal absorption of phenytoin

Activated charcoal

The absorption of phenytoin was almost completely prevented when given just before the oral ingestion of activated charcoal (Nation et al., 1990). When a single

98 Carlos A. Fontes Ribeiro

dose of activated charcoal was administered 1 h after a dose of phenytoin, there was still an estimated 80% reduction in absorption (Welling, 1984).

Antacids and inhibitors of gastric hydrochloric acid secretion

Studies regarding the effect of antacids on the disposition of phenytoin have pro- duced conflicting results. Overall, it appears that the influence of antacids is variable, both between antacid preparations and between subjects (Kutt, 1984). Moreover, both the timing of antacid dosing and the volume of antacid used may also con- tribute to this variability (D’Arcy and McElnay, 1987). The magnesium-containing antacids primarily increase gastric pH that enhances the solubility of weak acids and reduces the absorption rate from the stomach as it increases the ionization of the drug (Kutt, 1989). Aluminium-containing antacids, in addition, prolong gastric emptying time which under these circumstances further slows the rate of absorp- tion (Marano et al., 1985). Chelation or adsorption of phenytoin into the calcium- containing preparations has been suspected (Kutt, 1989; Nation et al., 1990). On the whole, antacids containing aluminium hydroxide, magnesium hydroxide, and calcium carbonate decreased the bioavailability of phenytoin. It is generally rec- ommended that, if antacids are to be used in patients receiving phenytoin, the administration of the two agents should be separated by a few hours.

Omeprazole does not affect the single-dose kinetics of phenytoin in healthy vol- unteers (Bachmann et al., 1994). However, Prichard et al. (1987) have reported that the extent and rate of oral absorption of phenytoin is increased during omeprazole therapy. The mechanism of this is unknown but may relate to changes in gastric pH. Multiple doses (for 7 days) of pantoprazole were without effect on the rate or the extent of single-dose phenytoin absorption (Middle et al., 1995). The hydrogen receptor antagonist cimetidine may increase the bioavailability of orally administered phenytoin (Hetzel et al., 1981) through inhibition of CYP isoenzymes, although additional factors relating to absorption may also be involved.

Sucralfate

Concurrent administration of sucralfate significantly reduced the AUC of phenytoin (Hall et al., 1986). Further studies are required to assess the effect of long-term sucralfate administration on phenytoin plasma concentrations.

Theophylline

It was suggested by Hendeles et al. (1979) that theophylline decreased the absorp- tion of phenytoin when the two agents were administered at the same time.

Antineoplastic therapy

Some antineoplastics (e.g. cisplatinum, vinblastine, and bleomycin) impair the gastrointestinal absorption of phenytoin (Sylvester et al., 1984).

99 Influence of food and drugs on the bioavailability of AEDs

Other drugs

Erythromycin, clarithromycin, and roxithromycin may increase the bioavailability of phenytoin (al-Humayyd, 1997). This effect may be due to an increased gastro- intestinal motility induced by these macrolide antibiotics and subsequent augmented phenytoin absorption.

Co-administration of ciprofloxacin and phenytoin revealed a significant decrease in steady-state maximum and minimum concentrations and in the area under the plasma time concentration curve (Islam et al., 1999). This finding warrants close monitoring of levels when these two agents are given simultaneously.

An approximately 30% reduction in dietary fat absorption induced by orlistat administered at doses of 120 mg three times daily did not significantly alter the pharmacokinetics of a single 300 mg oral dose of phenytoin in healthy volunteers (Melia et al., 1996).

Interactions during phenytoin parenteral administration

Although phenytoin sodium could be given both intravenously and intramuscularly, both of these routes of administration have limitations.

The major disadvantage of the intravenous route is the requirement for slow administration of the propylene glycol/alcohol diluent which is adjusted to pH 12 with sodium hydroxide (Tozer and Winter, 1990). This vehicle is required to maintain phenytoin in solution at a concentration of 50 mg of the sodium salt per milliliter. Due to the inconvenience of administering the drug slowly, there is often a desire to give phenytoin with other intravenous fluids. If phenytoin admixtures are to be used, only normal saline or lactated Ringer’s solution should be used, since admix- tures with other solutions could result in phenytoin precipitation (Tozer and Winter, 1990).

The intramuscular route of administration should be avoided because phenytoin precipitates at the site of injection. Consequently, absorption from the injection site tends to be rather erratic and slow, often continuing for 5 days or more (Tozer and Winter, 1990).

Phenytoin actions affecting the pharmacokinetics and/or pharmacodynamics of other drugs

It seems that phenytoin does not alter the absorption of other drugs. However, epileptic patients receiving phenytoin have been reported to exhibit a significantly smaller diuretic response to furosemide (Williamson, 1986). Furthermore, the time to peak diuretic response was considerably delayed in these patients. This was attrib- uted to delayed oral absorption of furosemide, perhaps the result of a phenytoin- induced decrease in the spontaneous activity of gastrointestinal smooth muscle (Williamson, 1986). However, other factors may be involved, such as the reduc- tion of the sensitivity of the renal tubule to the diuretic action of furosemide

100 Carlos A. Fontes Ribeiro

(Ahmad, 1974). In contrast with the observations with furosemide, Keller et al. (1981) have reported that pre-treatment with phenytoin did not alter the disposi- tion of orally administered hydrochlorothiazide. There is some evidence that pheny- toin treatment may decrease the gastrointestinal absorption of thyroxin and folic acid (Nation et al., 1990). Phenytoin can act as folate antagonist and precipitate folic acid deficiency (Matsui and Rozovski, 1982). Finally, Rowland and Gupta (1987) suggested that the treatment with phenytoin leads to decreased gastrointestinal absorption of cyclosporine.

Carbamazepine

The gastrointestinal absorption of carbamazepine formulations is slow, erratic and unpredictable (Morselli, 1989). The mechanisms that are associated with these characteristics may be:

(a) lowwatersolubility( 200mg/ml)andotherphysicochemicalpropertiesofthe molecule (a neutral drug which cannot be converted to a soluble salt), leading to a very slow dissolution rate in gastrointestinal fluid,

(b) anticholinergic properties of the drug which may become more evident during prolonged treatment and which modify its gastrointestinal transit time (Morselli, 1989).

It has been suggested that the rate and extent of its absorption may be dose- dependent.

Carbamazepine usually peaks 3–8 h after oral dosing, but the addition of propylene glycol, polysorbate, or ethanol can accelerate the absorptive process and reduce the time to peak to 1.5–4 h and increase its bioavailability (Leppik and Wolff, 1993). There is evidence that the dissolution rate of tablets can be affected by moisture (Wang et al., 1993). Furthermore, absorption of the suspension is more rapid than that of tablets, resulting in peak concentrations at 1–3 h (Morselli, 1989). Therefore, liquid oral carbamazepine dosage formulations are typically associated with a doubl- ing in their oral bioavailability compared with tablet formulations (Brewster et al., 1997). However, the relative bioavailability of carbamazepine suspension with enteral or nasogastric feeding administration is slightly diminished and generally slower than during fasting (Bass et al., 1989). Changes in gastric pH induced by ranitidine in healthy adults did not affect the bioavailability of carbamazepine (Dalton et al., 1985).

Carbamazepine induces the CYP3A4 catalyzed sulfoxidation of omeprazole, apparently without major clinical implication, and it has no or less effect on hydroxylation via the CYP2C19 (Bertilsson et al., 1997). CYP3A4 isoenzyme exists in the gut and liver. Carbamazepine half-life and 24h post dose concentration increased significantly during erythromycin administration (Miles and Tennison,

101 Influence of food and drugs on the bioavailability of AEDs

1989). These effects are not only due to changed absorption but also to inhibition of metabolic pathways (Miles and Tennison, 1989).

Aminophylline reduced the bioavailability of carbamazepine which may be of clinical significance (Kulkarni et al., 1995); 400 mg pentoxifylline administered at 22:00 h reduced the rate but not the extent of carbamazepine absorption (Poondru et al., 2001). Interestingly, these effects were not observed when pentoxifylline was administered at 10:00 h.

Valproic acid

Valproic acid is a branched-chain fatty acid which is rapidly and completely absorbed once it is released from its pharmaceutical formulation. In spite of differ- ences in populations and pharmaceutical formulations, the absolute bioavailability of valproate is consistently found to be close to unity. This observation indicates that valproate is not subject to a first-pass effect which is consistent with its low metabolic clearance.

Meals can have a profound effect on the time to peak concentration for the enteric- coated tablets; however, the long peak times represent delayed, rather than pro- longed, absorption. Ramadan, with its changes in eating and rest/activity rhythms, significantly influences the pharmacokinetics of valproic acid. A significant decrease in the bioavailability of valproic acid was found at the end of the 3rd week of Ramadan, compared to the control period (Aadil et al., 2000).

Carbapenem antibiotics induce a decrease in plasma concentration of valproic acid in epileptic patients (Torii et al., 2002). By using Caco-2 cell monolayers, the influence of carbapenems was tested on the transepithelial transport of valproic acid (Torii et al., 2002); it was found that carbapenems may inhibit the absorption of valproic acid at the basolateral membrane of intestinal epithelial cells. The same authors had veri- fied that imipenem inhibits the intestinal absorption of valproic acid but not through an inhibition of a carrier-mediated transport of valproic acid (Torii et al., 2001).

Repeat charcoal administered several hours after sodium valproate ingestion appears not to impair the absorption of valproic acid or indeed its pharmaco- kinetics (al-Shareef et al., 1997). Aminophylline also seems not to alter the pharmaco- kinetic parameters of valproic acid (Kulkarni et al., 1995).

Phenobarbital

Phenobarbital has a pKa of 7.2 and is more water soluble than phenytoin or carba- mazepine. Early work on the rate and extent of absorption of phenobarbital indicated the potential for dissolution-rate-limited absorption after oral administration (Rust and Dodson, 1989). More recent studies have found that phenobarbital (acid and tablets) is absorbed rapidly and completely. The absolute bioavailability of phenobarbital has been found to be close to unity.

102 Carlos A. Fontes Ribeiro

The bioavailability of phenobarbital appears to be greater in protein malnourished subjects (Syed et al., 1986). Activated charcoal reduces phenobarbital absorption, a characteristic that is exploited clinically in the early treatment of phenobarbital over- dose (Neuvonen and Elonen, 1980; Welling, 1984). However, colestipol hydrochlo- ride, a hypocholesterolemic bile acid-binding anion-exchange polymer, does not change phenobarbital absorption (Phillips et al., 1976).

Is has been suggested that the absorption of griseofulvin may be reduced by phe- nobarbital (Riegelman et al., 1970), perhaps as a result of diminished dissolution. Phenobarbital may cause a modest reduction of cimetidine absorption (Somogyi and Gugler, 1982), mainly due to induction of its gastrointestinal metabolism (Somogyi et al., 1981). Patients receiving phenobarbital have been reported to exhibit a significantly smaller diuretic response to furosemide and this may be the consequence of reduced absorption (Williamson, 1986).

Ethosuximide

Ethosuximide is relatively water soluble and is rapidly absorbed from tablets. The time required to reach peak plasma concentration is less than 3 h (Chang, 1989). Due to its very low clearance, no first-pass effect is expected. Ethosuximide has not been associated with any interactions at the gastrointestinal site of absorption.

Interactions with other AEDs

Over the past few years, eight new AEDs (felbamate, gabapentin, lamotrigine, oxcar- bazepine, topiramate, zonisamide, vigabatrin, and levetiracetam) have reached the market and are licensed for clinical use. Due to the risks associated with the use of an unproved new drug as monotherapy, current guidelines for AED trials require that the test drug be evaluated as add-on therapy. Thus, drug interactions are impor- tant considerations. Very dramatic pharmacokinetic interactions were observed with some new AEDs that were evaluated during the 1980s. For example, nafimi- done is a potent inhibitor of both carbamazepine and phenytoin (Leppik et al., 1993); the inhibition is of such magnitude that clinical toxicity is observed, and this limited the development of the drug. Another example is that of MK-801 (Leppik et al., 1993). These examples underscore the need for evaluating pharma- cokinetic interactions in the early stage of new AED development. However, in general, AED interactions with food are not studied during preclinical studies (phases I–III) and therefore information in this regard is sparse.

Vigabatrin

Vigabatrin is a synthetic gamma aminobutyric acid (GABA) derivative which was designed to increase brain GABA concentrations by inhibiting GABA transaminase,

103 Influence of food and drugs on the bioavailability of AEDs

the enzyme responsible for the breakdown of GABA. Vigabatrin is a racemic mixture but only the S( ) enantiomer is pharmacologically active. However, the R enan- tiomer does not interfere with the disposition of the S enantiomer, nor does it undergo chemical inversion in vivo (Haegle and Schechter, 1986; Richens, 1989; Rey et al., 1992).

The bioavailability of vigabatrin is considered to be at least 60–80% (Haegle and Schechter, 1986). The AUC for fasted and fed volunteers is not significantly differ- ent, indicating that food does not affect the extent of absorption (Frisk-Holmberg et al., 1989). Overall the interaction potential of vigabatrin is minimal.

Tiagabine

This AED, a nipecotic acid derivative, increases brain GABA concentrations through inhibition of GABA re-uptake (Natsch et al., 1997). After oral ingestion tiagabine is rapidly absorbed with peak plasma concentrations occurring within 1 h. Its bioavailability is 90% (Jansen et al., 1995). Whilst the rate of tiagabine inges- tion is slowed by food co-ingestion (Tmax increases from 0.9 to 2.6 h), the extent of absorption remains the same (Mengel et al., 1991). To date there are no data on the effects of drugs on the absorption of tiagabine.

Felbamate

Gabapentin

Gabapentin is a GABA-related amino acid with properties of an amino acid, but unlike GABA it readily penetrates the blood–brain barrier. Gabapentin is a substrate of intestinal large neutral amino acid carriers (Gidal et al., 1998b). A consequence of this type of transport is the dose-dependent oral absorption of gabapentin, with saturation at high doses (McLean, 1994). Thus, the bioavailability of gabapentin which is reported to be only 35% at a steady dosage of 1500mg t.i.d., may be improved by ingesting the drug more frequently (e.g. from t.i.d. to q.i.d.; Gidal et al., 1998a).

High-protein meals do not seem to interfere with the absorption of gabapentin in spite of the fact that amino acids could interfere with the carrier system (Benetello et al., 1997). In contrast, a trend was noted for a modest increase in both Cmax and AUC values when gabapentin was ingested with a fat-free chocolate pudding

Felbamate is a lipophilic dicarbamate which is only very slightly soluble in water. After oral ingestion felbamate is rapidly absorbed with a bioavailability of over 90% (Shumaker et al., 1990). Food co-ingestion has no significant effect on either the rate or extent of absorption of felbamate (Graves et al., 1989; Leppik et al., 1993). To date there are no data on the effects of drugs on the absorption of felbamate.

104 Carlos A. Fontes Ribeiro

(Gidal et al., 1998b), which led these authors to state that dietary macronutrient composition (i.e. protein) may favourably influence gabapentin absorption. However, this conclusion is not in accordance with the transport of gabapentin though an amino acid carrier. Overall, it can be concluded that the bioavailability of gabapentin is not significantly affected by food.

An interaction between gabapentin and antacids containing aluminium and magnesium hydroxide has been reported (Turnheim, 2004). The gastrointestinal absorption of gabapentin appears to be reduced and typically gabapentin plasma concentrations are approximately 15% lower; this interaction is not considered to be of clinical significance. To date there are no other data on the effects of drugs on the absorption of gabapentin.

Lamotrigine

Lamotrigine is a phenyltriazine derivative which was initially developed as an antifo- late compound. Following oral ingestion, lamotrigine is rapidly well absorbed with peak plasma concentrations occurring at 1–3 h post ingestion (Cohen et al., 1987; Yuen, 1991; Leppik et al., 1993). The absolute bioavailability of lamotrigine after a 75-mg oral dose is 98 5% (Yuen, 1991). Whereas food co-ingestion slightly delays the occurrence of the peak plasma lamotrigine concentration, it does not affect the extent of absorption (Goa et al., 1993). To date there are no data on the effects of drugs on the absorption of lamotrigine.

Topiramate

Topiramate is a sulfamate-substituted monosaccharide which is structurally dis- tinct from other AEDs. It is rapidly absorbed, with peak plasma concentrations occurring within 2 h to 4 h after oral ingestion. The bioavailability of topiramate is estimated to be 81–95% (Easterling et al., 1988). Co-administration with food moderately slows absorption (11–13% decreased mean maximum absorption) whereas the extent of absorption is unaffected (Doose et al., 1996). Thus topiramate can be ingested without due regard to meal times. To date there are no data on the effects of drugs on the absorption of topiramate.

Oxcarbazepine

Oxcarbazepine, a keto compound chemically related to carbamazepine, has a similar therapeutic profile to that of carbamazepine but is associated with an improved tolerability profile (Jensen and Dam, 1990). Following oral ingestion, oxcarbazepine is rapidly absorbed with peak plasma concentrations of its pharmacologically active metabolite (a monohydroxylated derivative), occurring 4–6 h later. Its bioavailability is 89% (Feldmann et al., 1978). After a fat- and protein-rich breakfast there was a moderate increase in the monohydroxylated derivative AUC (16%) and Cmax (23%)

105 Influence of food and drugs on the bioavailability of AEDs

values but with no changes in Tmax and terminal half-life values (Degen et al., 1994). These changes should be of little therapeutic consequence. To date there are no data on the effects of drugs on the absorption of oxcarbazepine.

Zonisamide

Zonisamide is a benzisoxazole compound which is structurally different to other AEDs. Absorption is rapid after oral ingestion with peak plasma concentrations occurring after 2.4–3.6 h. The bioavailability of zonisamide is estimated to be 65%. The bioavailability of zonisamide is unaffected by food co-ingestion although there is a delay in peak plasma concentration values to 4–6 h. To date there are no data on the effects of drugs on the absorption of zonisamide.

Levetiracetam

Levetiracetam is the S enantiomer of the ethyl analog of piracetam and as such is structurally unrelated to other AEDs. The absorption of levetiracetam after oral ingestion is rapid with peak plasma concentrations occurring approximately 1 h later. Its bioavailability is considered to be essentially 100% (Patsalos, 2002). Although food co-ingestion slows the rate of absorption of levetiracetam, the extent is unaf- fected (Patsalos, 2003). To date there are no data on the effects of drugs on the absorption of levetiracetam.

REFERENCES

Aadil N, Fassi-Fihri A, Houti I, et al. Influence of Ramadan on the pharmacokinetics of a single oral dose of valproic acid administered at two different times. Method Find Exp Clin Pharmacol 2000; 22: 109–114.

Ahmad S. Renal insensitivity to frusemide caused by chronic anticonvulsant therapy. Br Med J 1974; 3: 657–659.

al-Humayyd MS. Pharmacokinetic interactions between erythromycin, clarithromycin, roxi- thromycin and phenytoin in the rat. Chemotherapy 1997; 43: 77–85.

al-Shareef A, Buss DC, Shetty HG, et al. The effect of repeated-dose activated charcoal on the phar- macokinetics of sodium valproate in healthy volunteers. Br J Clin Pharmacol 1997; 43: 109–111. Au Yeung SC, Ensom MH. Phenytoin and enteral feedings: does evidence support an inter-

action? Ann Pharmacother 2000; 34: 896–905.
Bachmann KA, Sullivan TJ, Jauregui L, et al. Drug interactions of H2-receptor antagonists. Scand

J Gastroenterol 1994; 206(Suppl.): 14–19.
Bass J, Miles MV, Tennison MB, et al. Effects of enteral tube feeding on the absorption and phar-

macokinetic profile of carbamazepine suspension. Epilepsia 1989; 30: 364–369.
Bauer LA. Interference of oral phenytoin absorption by continuous nasogastric feedings. Neurology

1982; 32: 570–572.

106 Carlos A. Fontes Ribeiro

Benetello P, Furlanut M, Fortunato M, et al. Oral gabapentin disposition in patients with epilepsy after a high-protein meal. Epilepsia 1997; 38: 1140–1142.

Bertilsson L, Tybring G, Widen J, et al. Carbamazepine treatment induces the CYP3A4 catalysed sulphoxidation of omeprazole, but has no or less effect on hydroxylation via CYP2C19. Br J Clin Pharmacol 1997; 44: 186–189.

Bialer M. Pharmacokinetic evaluation of sustained release formulations of antiepileptic drugs. Clinical implications. Clin Pharmacokinet 1992; 22: 11–21.

Bochner F, Hooper WD, Tyrer J, et al. Factors involved in an outbreak of phenytoin intoxication. J Neurol Sci 1972; 16: 481–487.

Brewster ME, Anderson WR, Meinsma D, et al. Intravenous and oral pharmacokinetic evaluation of a 2-hydroxypropyl-beta-cyclodextrin-based formulation of carbamazepine in the dog: com- parison with commercially available tablets and suspensions. J Pharm Sci 1997; 86: 335–339.

Bruguerolle B, Valli M, Bouyard L, et al. Circadian effect on carbamazepine kinetics in rat. Eur J Drug Metab Pharmacokinet 1981; 6: 189–193.

Burstein AH, Fisher KM, McPherson ML, et al. Absorption of phenytoin from rectal supposito- ries formulated with a polyethylene glycol base. Pharmacotherapy 2000; 20: 562–567.

Cacek AJ. Review of alterations in oral phenytoin bioavailability associated with formulations, antacids and food. Ther Drug Monit 1986; 8: 166–171.

Cacek AT, DeVito JM, Koonce JR. In vitro evaluation of nasogastric administration methods for phenytoin. American J Hosp Pharm 1986; 43: 689–692.

Carter BL, Garnett WR, Pellock JM, et al. Effect of antacid on phenytoin bioavailability. Ther Drug Monit 1981; 3: 333–340.

Chaudhary A, Lane RA, Woo D, et al. Multiple-dose lorazepam kinetics: shuttling of lorazepam glucuronide between the circulation and the gut during day- and night-time dosing intervals in response to feeding. J Pharmacol Exp Ther 1993; 267: 1034–1038.

Chang T. Ethosuximide: absorption, distribution, and excretion. In Antiepileptic Drugs. R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. New York: Raven Press, 1989: 671–678. Cohen AF, Land GS, Bramer DD, et al. Lamotrigine, a new anticonvulsant: Pharmacokinetics in

normal humans. Clin Pharmacol Ther 1987; 42: 535–541.
Dalton MJ, Powell JR, Messenheimer JA Jr. Ranitidine does not alter single-dose carbamazepine

pharmacokinetics in healthy adults. Drug Intell Clin Pharm 1985; 19: 941–944.
D’Arcy PF, McElnay JC. Drug–antacid interactions of clinical importance. Drug Intel Clin Pharm

1987; 21: 607–617.
de Boer AG, Moolenaar F, de Leede LG, et al. Rectal drug administration: clinical pharmaco-

kinetic considerations. Clin Pharmacokinet 1982; 7: 285–311.
Degen PH, Flesch G, Cardot JM, et al. The influence of food on the disposition of the antiepilep-

tic oxcarbazepine and its major metabolites in healthy volunteers. Biopharm Drug Dispos

1994; 15: 519–526.
Deutsch T, Ludwig E, Gráber H, et al. Basic clinical pharmacokinetics. In Human Pharmacology.

H. Kuemmerle, T. Shibuya, J.-P.Tillement, eds. Amsterdam: Elsevier Science Publishers, 1991:

307–535.
Doose DR, Walker SA, Gisclon LG, et al. Single-dose pharmacokinetics and effect of food on

the bioavailability of topiramate, a novel antiepileptic drug. J Clin Pharmacol 1996; 36: 884–891.

107 Influence of food and drugs on the bioavailability of AEDs

Easterling DE, Zakszewski T, Moyer MD, et al. Plasma pharmacokinetics of topiramate, a new anticonvulsant in humans. Epilepsia 1988; 29: 662.

Feldmann KF, Brechbuhler S, Faigle JW, et al. Pharmacokinetics and metabolism of GP 47680, a compound related to carbamazepine, in animals and man. In Advances in Epileptology. H. Meinardi, A. J. Rowan, eds. Amsterdam: Swets and Zeitlinger, 1978: 290–294.

Figueiredo A, Fontes Ribeiro CA, Gonçalo M, et al. Experimental studies on the mechanisms of tiaprofenic acid photosensitization. J Photochem Photobiol B: Biol 1993; 18: 161–168.

Frisk-Holmberg M, Kerth P, Meyer P. Effect of food on the absorption of vigabatrin. Br J Clin Pharmacol 1989; 27: 23S–25S.

Gidal BE, DeCerce J, Bockbrader HN, et al. Gabapentin bioavailability: effect of dose and fre- quency of administration in adult patients with epilepsy. Epilepsy Res 1998a; 31: 91–99.

Gidal BE, Maly MM, Kowalski JW, et al. Gabapentin absorption: effect of mixing with foods of varying macronutrient composition. Ann Pharmacother 1998b; 32: 405–409.

Goa KL, Ross SR, Chrisp P. Lamotrigine—A review of its pharmocological properties and clini- cal efficacy in epilepsy. Drugs 1993; 46: 152–176.

Graves NM, Ludden TM, Holmes GB, et al. Pharmacokinetics of felbamate, a novel antiepileptic drug: Application of mixed-effect modelling to clinical trials. Pharmacotherapy 1989; 9: 372–376. Haegle KD, Schechter PJ. Kinetics of the enantiomers of vigabatrin after an oral dose of the race-

mate or the active S-enantiomer. Clin Pharmacol Therap 1986; 40: 581–585.
Hall TG, Cuddy PG, Glass CJ, et al. Effect of sucralfate on phenytoin bioavailability. Drug Intell

Clin Pharm 1986; 20: 607–611.
Hamaguchi T, Shinkuma D, Irie T, et al. Effect of a high-fat meal on the bioavailability of pheny-

toin in a commercial powder with a large particle size. Int J Clin Pharmacol Ther Toxicol 1993;

31: 326–330.
Hendeles L, Wyatt R, Weinberger M, et al. Decreased oral phenytoin absorption following con-

current theophylline administration. J Allergy Clin Immunol 1979; 63: 156.
Hetzel DJ, Bochner F, Hallpike JF, et al. Cimetidine interaction with phenytoin. Br Med J 1981;

282: 1512.
Islam AF, Garg SK, Bhargava VK. Effect of ciprofloxacin on steady state pharmacokinetics of

phenytoin in rabbits. Ind J Exp Biol 1999; 37: 86–88.
Jansen JA, Oliver S, Dirach J, et al. Absolute bioavailability of tiagabine. Epilepsia 1995; 36(Suppl.

3): S159.
Jensen PK, Dam M. Oxcarbamazepine. In Comprehensive Epileptology. M. Dam, L. Gram, eds.

New York: Raven Press, 1990: 621–629.
Keller E, Sulzer U, Brennes M. Disposition of hydrochlorothiazide during phenytoin treatment.

Klin Wochenschr 1981; 59: 1223–1224.
Kennedy MC, Wade DN. Phenytoin absorption in patients with ileojejunal bypass. Br J Clin

Pharmacol 1979; 7: 515–518.
Kennedy MC, Wade DN. The effect of food on the absorption of phenytoin. Aust NZ J Med 1982;

12: 258–261.
Kulkarni C, Vaz J, David J, et al. Aminophylline alters pharmacokinetics of carbamazepine but

not that of sodium valproate – a single dose pharmacokinetic study in human volunteers. Ind J Physiol Pharmacol 1995; 39: 122–126.

108 Carlos A. Fontes Ribeiro

Kumar N, Garg SK, Prabhakar S. Lack of pharmacokinetic interaction between grapefruit juice and phenytoin in healthy male volunteers and epileptic patients. Method Find Exp Clin Pharmacol 1999; 21: 629–632.

Kutt H. Interactions between anticonvulsants and other commonly prescribed drugs. Epilepsia 1984; 25(Suppl. 2): S118–S131.

Kutt H. Phenytoin – Interactions with Other Drugs. In Antiepileptic Drugs. R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. New York: Raven Press, 1989: 215–232.

Leppik IE, Wolff DL. Antiepileptic medication interactions. Neurol Clin 1993; 11: 905–921. Leppik IE, Graves N, Devinsky O. New antiepileptic medications. Neurol Clin 1993; 11:

923–949.
Levy G. Clinical implications of interindividual differences in plasma protein binding of drugs

and endogenous substances. In The Effect of Disease States on Drug Pharmacokinetics.

L. Z. Benet, ed. Washington: American Pharmaceutical Association, 1976: 137–151.
Loiseau P, Cenraud B, Levy RH, et al. Diurnal variations in steady-state plasma concentrations

of valproic acid in epileptic patients. Clin Pharmacokinet 1982; 7(6): 544–552.
Marano AR, Caride VJ, Prokop EK. Effect of sucralfate and an aluminum hydroxide gel on gas-

tric emptying of solids and liquids. Clin Pharmacol Ther 1985; 37: 629–632.
Matsui MS, Rozovski SJ. Drug–nutrient interaction. Clin Ther 1982; 4: 423–440.
McLean MJ. Clinical pharmacokinetics of gabapentin. Neurology 1994; 44(Suppl. 5): S17–S22. Melander A, Brante G, Johansson O, et al. Influence of food on the absorption of phenytoin in

man. Eur J Clin Pharmacol 1979; 15: 269–274.
Melia AT, Mulligan TE, Zhi J. The effect of orlistat on the pharmacokinetics of phenytoin in

healthy volunteers. J Clin Pharmacol 1996; 36: 654–658.
Mengel HB, Gustavson LE, Soerensen HJ, et al. Effect of food on the bioavailability of tiagabine

HCl. Epilepsia 1991; 32(Suppl. 3): 6.
Middle MV, Muller FO, Scall R, et al. No influence of pantoprazole on the pharmacokinetics of

phenytoin. Int J Clin Pharmacol Ther 1995; 33: 304–307.
Miles MV, Tennison MB. Erythromycin effects on multiple-dose carbamazepine kinetics. Ther

Drug Monit 1989; 11: 47–52.
Miles C, Dickson P, Rana K, et al. CCK antagonist pre-treatment inhibits meal-enhanced drug

absorption in dogs. Regul Pept 1997; 68: 9–14.
Morselli PL. Carbamazepine absorption, distribution and excretion. In Antiepileptic Drugs.

R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. New York: Raven Press,

1989: 473–490.
Nation RL, Evans AM, Milne RE. Pharmacokinetic drug interactions with phenytoin (Part I).

Clin Pharmacokinet 1990; 18: 37–60.
Natsch S, Hekster YA, Keyser A, et al. Newer anticonvulsant drugs: role of pharmacology, drug

interactions and adverse reactions in drug choice. Drug Saf 1997; 17: 228–240.
Neuvonen PJ, Elonen E. Effect of activated charcoal on absorption and elimination of phenobarbitone, carbamazepine and phenylbutazone in man. Eur J Clin Pharmacol 1980; 17:

51–57.
Patsalos PN. Pharmacokinetic profile of levetiracetam: toward ideal characteristics. Pharmacol

Ther 2002; 85: 77–85.

109 Influence of food and drugs on the bioavailability of AEDs

Patsalos PN. The pharmacokinetic characteristics of levetiracetam. Method Find Exp Clin Pharmacol 2003; 25: 123–129.

Petker MA, Morton DJ. Comparison of the effectiveness of two oral phenytoin products and chronopharmacokinetics of phenytoin. J Clin Pharm Ther 1993; 18: 213–217.

Phillips WA, Ratchford JM, Schultz JR. Effects of colestipol hydrochloride on drug absorption in the rat. J Pharm Sci 1976; 65: 1285–1291.

Poondru S, Devaraj R, Boinpally, RR, et al. Time-dependent influence of pentoxifylline on the pharmacokinetics of orally administered carbamazepine in human subjects. Pharmacol Res 2001; 43: 301–305.

Prichard PJ, Walt RP, Kitchingman GK, et al. Oral phenytoin pharmacokinetics during omepra- zole therapy. Br J Clin Pharmacol 1987; 24: 543–545.

Rey E, Pons G, Olive G. Vigabatrin: clinical pharmacokinetics. Clin Pharmacokinet 1992; 23: 267–278. Richens A. Vigabatrin. In Antiepileptic Drugs. R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry,

F. E. Dreifuss, eds. New York: Raven Press, 1989: 937–946.
Riegelman S, Rowland M, Epstein WL. Griseofulvin–phenobarbital interactions in man. J Amer

Med Assoc 1970; 213: 426–341.
Rowland M, Gupta SK. Cyclosporin–phenytoin interaction: re-evaluation using metabolite data.

Br J Clin Pharmacol 1987; 24: 329–334.
Rust RS, Dodson WE. Phenobarbital: absorption, distribution, and excretion. In Antiepileptic

Drugs. R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. New York: Raven

Press, 1989: 293–304.
Sekikawa H, Nakano M, Takada M. Influence of dietary components on the bioavailability of

phenytoin. Chem Pharm Bull 1980; 22: 2443–2449.
Shumaker RC, Fantel C, Kelton E, et al. Evaluation of the elimination of [14C] felbamate in

healthy men. Epilepsia 1990; 31: 642.
Smith OB, Longe RL, Altman RE, et al. Recovery of phenytoin from solutions of caseinate salts

and calcium chloride. Am J Hosp Pharm 1988; 45: 365–368.
Somogyi A, Thielscher S, Gugler R. Influence of phenobarbital treatment on cimetidine kinetics.

Eur J Clin Pharmacol 1981; 19: 343–347.
Somogyi A, Gugler R. Drug interaction with cimetidine. Clin Pharmacokinet 1982; 7: 23–41. Spina E, Pisani F, Perucca E. Clinically significant pharmacokinetic drug interactions with car-

bamazepine. An update. Clin Pharmacokinet 1996; 31: 198–214.
Stella VJ, Martodihardjo S, Rao VM. Aqueous solubility and dissolution rate does not adequately

predict in vivo performance: a probe utilizing some N-acyloxymethyl phenytoin prodrugs.

J Pharm Sci 1999; 88: 775–779.
Syed GB, Sharma DB, Raina RK. Pharmacokinetics of phenobarbitone in protein energy malnu-

trition. Dev Pharmacol Ther 1986; 9: 317–322.
Sylvester RK, Lewis FB, Caldwell KC, et al. Impaired phenytoin bioavailability secondary to cis-

platinum, vinblastine, and bleomycin. Ther Drug Monit 1984; 6: 302–305.
Torii M, Takiguchi Y, Saito F. Inhibition by carbapenem antibiotic imipenem of intestinal

absorption of valproic acid in rats. J Pharm Pharmacol 2001; 53: 823–829.
Torii M, Takiguchi Y, Izumi M. Carbapenem antibiotics inhibit valproic acid transport in Caco-

2 cell monolayers. Int J Pharm 2002; 233: 253–256.

110 Carlos A. Fontes Ribeiro

Tozer TN, Winter ME. Phenytoin. In Applied Pharmacokinetics – Principles of Therapeutic Drug Monitoring. W. E. Evans, J. J. Schentag, W. J. Jusko, eds. San Francisco: Applied Therapeutics, 1990: 275–314.

Turnheim K. Drug interactions with antiepileptic agents. Wien Klin Wochenshr 2004; 116: 112–118.

Tuttle CB. Intramuscular injections and bioavailability. Am J Hosp Pharm 1977; 34(9): 965–968. Wang JT, Shiu GK, Ong-Chen T, et al. Effects of humidity and temperature on in vitro dissolu-

tion of carbamazepine tablets. J Pharm Sci 1993; 82: 1002–1005.
Welling PG. Interactions affecting drug absorption. Clin Pharmacokinet 1984; 9: 404–434. Williamson HE. Interaction of furosemide and phenytoin in the rat. Proc Soc Exp Biol Med 1986;

182: 322–324.
Yoshiyama Y, Nakano S, Ogawa N. Chronopharmacokinetic study of valproic acid in man: com-

parison of oral and rectal administration. J Clin Pharmacol 1989; 29: 1048–1052.
Yuen WC. Lamotrigine. In New Antiepileptic Drugs (Epilepsy Research Suppl. 3). F. Pisani,

E. Perucca, G. Avanzini, eds. Amsterdam: Elsevier, 1991: 115–123.

Interactions between antiepileptic drugs

Bernhard Rambeck and Theodor W. May
Biochemisches Labor der Gesellschaft für Epilepsieforschung, Maraweg 13, Bielefeld, Germany

Summary

Old and new antiepileptic drugs (AEDs) are associated with a wide range of phar- macokinetic drug–drug interactions. The classic AEDs exert important inducing and inhibiting effects on old and new AEDs.

Phenobarbital (PB) concentrations are significantly increased by valproic acid (VPA) and to a variable degree also by phenytoin (PHT). PHT levels may be decreased or increased by PB, depending on the PB concentration. The protein binding of PHT is decreased by VPA. Enzyme-inducing AEDs decrease primidone concentrations, but increase the levels of its metabolite PB. Carbamazepine (CBZ) concentrations are decreased by PB and PHT, whereas its metabolite CBZ-10,11-epoxide (CBZ-E) may be increased by VPA. Concentrations of VPA are considerably decreased by enzyme- inducing AEDs such as PB, PHT or CBZ. Sulthiame, a rarely used AED, increases PHT levels. Methsuximide (MSM), another rarely used AED, inhibits the metabolism of PB and PHT, but induces the metabolism of lamotrigine (LTG) and oxcarbazepine (OXC).

New AEDs exert relatively few inhibiting or inducing effects on the classic AEDs and hardly any on the new AEDs. However, felbamate (FBM) increases concentra- tions of PHT, PB, VPA and of CBZ-E, but reduces concentrations of CBZ. OXC (and some other new AEDs) may also increase PHT, whereas vigabatrin reduces the serum levels of PHT by approximately 20%. OXC has less pronounced enzyme-inducing effects than CBZ; however, topiramate (TPM) and LTG may be lowered by OXC.

On the other hand, enzyme-inducing AEDs reduce serum concentrations of FBM, LTG, tiagabine (TGB), TPM, zonisamide (ZWS) and to a minor extent of 10-hydroxy- carbazepine, the clinically relevant metabolite of OXC. VPA markedly increases LTG and FBM. In comparison to other AEDs the potential for clinically relevant inter- actions associated with gabapentin and levetiracetam is low.

Introduction

Antiepileptic therapy has been associated with a wide range of drug–drug inter- actions. Classical pharmacokinetic interactions are enzyme induction, enzyme

111

112 Bernhard Rambeck and Theodor W. May

inhibition and displacement from protein binding. From the pharmacological point of view monotherapy with AEDs is often considered as the treatment regime of choice for epileptic patients in order to avoid undesirable consequences of drug interactions such as side effects by increased AED concentrations or inefficacy of the therapy due to decreased serum levels. But for clinical reasons, in practice, many patients have to be treated with AED combinations. Furthermore, newly introduced AEDs are licensed usually as comedication.

Correspondingly, the knowledge of pharmacokinetic interactions is most impor- tant. Countless papers have been published about interactions of AEDs. However, many of these interactions are hardly clinically relevant in so far as they concern only weak influences or they have no practical consequences. This overview will deal especially with the clinically important interactions of AEDs. Of course, the extent and the significance of an interaction can vary individually, as it often depends not only on the relative dosages of the interacting drugs, but also on previous drug exposure and on pharmacogenetic factors.

Benzodiazepines are not regarded in this review. The effects of these drugs are minimal as they usually occur only in relatively low concentrations in the serum compared to AED concentrations. Possibly enzyme-inducing drugs may reduce their serum concentrations, but there are hardly any investigations on this topic.

Interactions between classic AEDs (phenobarbital, phenytoin, primidone,

carbamazepine, valproic acid, ethosuximide, methsuximide) and other AEDs

Phenobarbital

Phenobarbital (PB) is about one-third metabolized to a p-hydroxylated derivative. It is partially (50–60%) bound to serum proteins.

Effect of phenobarbital on other drugs

PB is the prototype among inducers of the hepatic mixed-function oxidase system. Numerous studies have been performed showing that PB decreases concentrations of other concomitantly given AEDs. Particularly impressive is the effect of PB on CBZ and VPA. A typical investigation, which documents the influence of PB on CBZ metabolism, is a detailed study with data of 609 epileptic patients (Rambeck et al., 1987). PB decreases CBZ levels by about 34% when compared to levels of patients on CBZ alone. The inducing effect was thereby comparable with that of PHT and primidone (PRM).

PB shows not only inducing effects, but also inhibiting effects on some enzyme sys- tems. In some cases, such as for the influence of PB on PHT, results are controversial as two apparently contradictory mechanisms, competitive metabolic inhibition

113 Interactions between antiepileptic drugs

and enzyme induction may play a role (Inoue and Chambers, 1985). There are studies which show that PB tends to lower PHT levels when the two drugs are used simultaneously (Abarbanel et al., 1978) and others which demonstrate a significant increase of PHT in the presence of PB, but returning to prevalues some weeks later (Müller et al., 1977). Another study with stable isotope tracer techniques con- cluded that PB does not alter PHT steady-state concentration or kinetics (Browne et al., 1988a). A statistical investigation of a large collective with 1992 epileptic patients indicated that low levels of PB induce PHT metabolism and thereby decrease PHT concentrations, but higher PB levels inhibit PHT metabolism in a competitive man- ner and thereby increase PHT concentrations (May et al., 1982).

PB increases the clearance of VPA. For example, in a representative study with 259 epileptic patients, VPA levels were about 24% lower when VPA was given con- comitantly with PB than when it was given alone (May and Rambeck, 1985). The inducing effect in this case was smaller than that of CBZ and PHT, where reductions of 34% and 50% respectively were found.

PB shows its inducing effect also in the presence of some new AEDs. PB reduces LTG levels considerably. A typical study with data of 302 epileptic patients docu- mented that LTG levels are decreased by PB by 48% (May et al., 1996a). The induc- ing effect was somewhat stronger than that of CBZ (43%) but smaller than that of PHT (68%). TGB levels are reduced by PB (see section on TGB), and TPM metab- olism is increased (see section on TPM); furthermore, PB induces the metabolism of ZNS (see section on ZNS). There is also a study (Tartara et al., 1993) which indicates that the biotransformation of OXC and its metabolite 10-hydroxy-carbazepine (or monohydroxy-derivative, MHD) may be accelerated by concomitant treatment with PB, but the magnitude of this effect is unlikely to be of great clinical signifi- cance. PB does not seem to influence FBM levels (Kelley et al., 1997) or gabapentin (GBP) levels (Hooper et al., 1991).

Effect of other drugs on phenobarbital

The metabolism of PB itself is inhibited by some other AEDs when given in combi- nation. VPA increases PB concentrations and often thereby causes side effects such as sedation and drowsiness. In a study with 186 epileptic patients, PB levels were about 40% higher when VPA was given additionally. This effect was independent whether PB was directly given or occurred as a metabolite of PRM (Rambeck et al., 1979). Wilder et al. (1978) documented a comparable influence for 25 epileptic adults. Various studies showed a reduced hydroxylation of PB (Bruni et al., 1980) and a prolongation of the half-life of PB by VPA by about 50% (Patel et al., 1980; Kapetanovic et al., 1981).

As PHT and PB are metabolized by the same phenyl hydroxylating enzyme system, PHT may inhibit PB metabolism in a competitive manner. PB concentrations are

114 Bernhard Rambeck and Theodor W. May

then increased (Windorfer and Sauer, 1977). Correspondingly, Duncan et al. (1991) found a decrease in PB concentrations when the concomitant PHT med- ication was stopped. A study with 121 patients (Eadie et al., 1976) failed to find any significant elevations of plasma levels attributable to PHT. But there is also a study (Encinas et al., 1992) which concludes that PHT may interact with PB as an inducer or an inhibitor of metabolism depending on the length of treatment with the combination of the two drugs.

An important interaction is the competitive inhibition of PB metabolism by methsuximide or its clinically relevant metabolite N-desmethyl-MSM, whereby PB concentrations are increased by about 40% (Rambeck, 1979).

FBM increases PB levels by a reduction of its p-hydroxylation (Reidenberg et al., 1995a; Glue et al., 1997). Furthermore, OCBZ may increase PB but to a minor extent (Barcs et al., 2000). The other new AEDs do not show clinically relevant influences on PB; these facts are discussed in the respective sections.

Phenytoin

Effect of phenytoin on other drugs

PHT has enzyme-inducing properties and decreases drug concentrations of con- comitant AEDs.

CBZ metabolism is increased by PHT to a considerable extent. The above- mentioned study with 609 epileptic patients indicated a reduction of CBZ levels by 40% (Rambeck et al., 1987). As already mentioned the interaction between PHT and PB is controversial as different effects may play a role. Some studies found an increase of PB concentrations by about 30% (Windorfer and Sauer, 1977; Duncan et al., 1991), others found no influence of PHT on PB levels (Eadie et al., 1976). PHT induces the metabolism of PRM to its important metabolite PB. The ratio between PRM and its metabolite PB which is usually about 1:1 in monotherapy is then changed to about 1:4 (Fincham et al., 1973).

Addition of PHT to a VPA therapy leads to a considerable decrease in VPA levels. An analysis of data from 259 epileptic patients on polytherapy with AEDs indicated that PHT was a strong inducer (reduction 50%) of VPA levels (May and Rambeck, 1985).

LTG levels are also reduced by PHT. This was shown in the already mentioned study with 302 epileptic patients on LTG where LTG levels were reduced by 68%

PHT is nearly completely metabolized to p-hydroxy-PHT and glucuronidated derivatives. It is bound to serum proteins to about 92%. Due to its saturable non- linear Michaelis–Menten kinetics, even moderate influences of other drugs on its metabolism may lead to considerable increases of PHT serum concentration.

115 Interactions between antiepileptic drugs

when patients were on PHT comedication (May et al., 1996a). PHT exerts its induc- ing effect also on some other new AEDs such as TGB (see section on TGB), TPM (see section on TPM) and ZNS (see section on ZNS).

The influence of PHT on FBM is not quite clear. In a study by Kelley et al. (1997) PHT increased the clearance of FBM by about 40%, whereas Troupin et al. (1997) found no appreciable changes in FBM clearance for comedication with PHT.

Effect of other drugs on phenytoin

The metabolism of PHT itself may be increased or decreased by comedicated drugs; in some cases even by the same substance, depending on the serum concentration of the interacting drug. These effects have been discussed exemplarily for the influ- ence of PB on PHT in the section ‘Effect of PB on other drugs’.

An investigation by Browne et al. (1988b) of six otherwise healthy men found that CBZ increases PHT serum concentrations. Concomitant therapy with MSM often leads to a remarkable increase in PHT concentrations (mean 78%) and thereby disturbing side effects may occur (Rambeck, 1979). Sulthiame also inhibits PHT metabolism and increases PHT levels (Hansen et al., 1968). Although today sulthiame is only rarely used, this interaction is noteworthy as it may induce severe side effects. Furthermore, this was one of the first important drug interactions observed in the treatment of epilepsy.

The interaction of PHT with VPA is somewhat complex as it primarily concerns protein binding. VPA displaces PHT from serum proteins and increases the free fraction of this drug from normally 8% in the absence of VPA to 20%, depending on the VPA concentration (May et al., 1991). But, as the total concentration of PHT decreases, the actually important free concentration of PHT often remains unchanged. Lai and Huang (1993) concluded that there are at least two mecha- nisms involved in this interaction. Whereas VPA displacing PHT from plasma pro- tein decreased the total drug concentration of PHT, the enzyme inhibition by VPA increased both the total and unbound concentration of PHT. A detailed analysis of data from 237 patients on PHT with and without VPA comedication indicated a significant decrease in total PHT concentration by VPA (Rambeck et al., 1979). The interaction between PHT and VPA may even be time dependent as the plasma concentration of VPA fluctuates during the day, resulting in variable displacement of PHT from its protein binding (Riva et al., 1985; May and Rambeck, 1990).

The new AEDs show no or only small effects on PHT metabolism. LTG does not influence the disposition of PHT (Grasela et al., 1999) and no significant effect by ZNS on the serum concentration or protein binding of PHT was found (Tasaki et al., 1995). As expected, the addition of levetiracetam (LEV) did not bring about clinically relevant changes in PHT pharmacokinetic parameters (Browne et al., 2000).

116 Bernhard Rambeck and Theodor W. May

Total PHT plasma concentrations increased with coadministered FBM (Fuerst et al., 1988), accordingly the PHT dosage should be reduced by about 20% (Sachdeo et al., 1999). OXC also seems to inhibit the metabolism of PHT (Barcs et al., 2000). Some studies showed that vigabatrin (VGB) decreases serum PHT con- centrations, but the mechanism is unknown (Gatti et al., 1993).

Primidone

Effect of primidone on other drugs

As PRM is metabolized to a great extent to PB, it shows the same influences on other drugs as PB itself. This means that it decreases levels of VPA, CBZ, LTG and many other drugs.

Effect of other drugs on primidone

When discussing influences of other AEDs on PRM metabolism two effects have to be considered. Primarily, the degradation of PRM to PB is induced by drugs such as PHT or CBZ and furthermore other comedicated AEDs may increase the result- ing metabolite PB (Porro et al., 1982).

In the first days of a PRM monotherapy, only PRM is found in the serum. Then the auto-induction of its own metabolism leads to increasing PB concentrations. After some weeks, in steady-state conditions, PB/PRM ratios of about 1:1 are reached. In the presence of other inducing AEDs such as PHT or CBZ the PB/PRM ratio is fur- ther increased to 5:1. This has been documented in various studies (Fincham et al., 1973; Schmidt, 1975).

But, it must also be considered that the same drugs that increase PB levels also increase levels of PB occurring as a metabolite of PRM. This has been shown for PHT (Lambie and Johnson, 1981), MSM (Rambeck, 1979) and VPA (Rambeck et al., 1979). A further increase of the PB/PRM ratio to 10:1 may be the consequence. In such cases, it is questionable how far or whether the anticonvulsant effect of a PRM therapy is still exerted by PRM itself or more or less by PB.

Carbamazepine

CBZ is largely metabolized to CBZ-E and then to CBZ-10, 11-diol. CBZ-E seems to contribute to the side effects of a CBZ therapy whilst the diol is physiologically inac- tive. CBZ and CBZ-E are bound by about 40% to serum proteins.

PRM is metabolized to PB and phenyl-ethyl-malonamide (PEMA). The ratio of PRM to PB and PEMA depends not only on auto-induction but also on induction by other AED.

117 Interactions between antiepileptic drugs

Effect of carbamazepine on other drugs

CBZ has enzyme-inducing properties and correspondingly decreases concentra- tions of other concomitantly given AEDs.

The influence of CBZ on PHT is not quite clear. Lai et al. (1992) showed in a study with volunteers that CBZ may decrease PHT levels possibly by decreased bioavailability of PHT when CBZ was co-administered. As mentioned above CBZ may induce the metabolism of PRM.

VPA is considerably reduced by CBZ. In a study with 259 patients on VPA, CBZ reduced VPA levels by about 34%. Its inducing effect was larger than that of PB but smaller than that of PHT (May and Rambeck, 1985). Comparable results were found in a study of Reunanen et al. (1980) with epileptic patients and in a study of Bowdle et al. (1979) with healthy volunteers.

CBZ reduces LTG levels (Bartoli et al., 1997; Battino et al., 1997). In a study of 302 patients on LTG, patients on CBZ comedication had LTG levels that were about 50% lower than that of patients on LTG monotherapy. The inducing effect was comparable with that of PB but less than that of PHT (May et al., 1996a). Furthermore, CBZ reduces FBM levels (Kelley et al., 1997; Troupin et al., 1997), TGB levels (Brodie, 1995; So et al., 1995; Snel et al., 1997), TPM levels (Sachdeo et al., 1996) and ZNS concentrations (Ojemann et al., 1986). GBP concentrations are not influenced (Radulovic et al., 1994).

Effect of other drugs on carbamazepine

Besides the impressive inducing effect of CBZ it has to be borne in mind that CBZ itself is subject to enzyme induction. Various studies have documented that simul- taneously given AEDs reduce CBZ concentrations.

An investigation by Michele et al. (1985) with 58 patients showed that PB reduces CBZ levels to a considerable extent. Christiansen and Dam (1973) showed in 123 epileptic patients that PB and PHT reduce CBZ concentrations. In a study with 609 epileptic patients on CBZ therapy (Rambeck et al., 1987), the mean serum concen- tration of CBZ was reduced when given in combination with PHT by 42%, with PB by 34% and with VPA by 17%.

Besides the inducing effect on CBZ metabolism some drugs inhibit the degra- dation of CBZ-E. In the above-mentioned study (Rambeck et al., 1987) the mean concentration of CBZ-E was increased by VPA ( 45%), PRM ( 19%) and a com- bination of the latter ( 67%) compared to CBZ monotherapy. These effects are reflected by the ratios between CBZ and its CBZ-E. In CBZ monotherapy a ratio of about 7:1 is found in adults (Rambeck et al., 1987). In the presence of inducing AEDs the ratio is lowered to 3:1, and in the presence of inducing AEDs in combi- nation with VPA it is 2:1. VPA appears to inhibit the conversion of CBZ-E to the trans-diol derivative and furthermore the glucuronidation of this CBZ-10,11-diol

118 Bernhard Rambeck and Theodor W. May

(Bernus et al., 1997). In the special case of adding CBZ to a basic VPA therapy, the inhibiting effect of VPA on the metabolism of CBZ-E is particularly impressive, especially in children. CBZ-E concentrations of up to 13 g/ml have been observed, accompanied by side effects such as vomiting and tiredness, although the CBZ levels were in the usually accepted effective range (Rambeck et al., 1990). After a few days the CBZ-E concentration decreases, but CBZ/CBZ-E ratios of 3:1 remain.

FBM appears to induce CBZ metabolism and decrease CBZ levels (Liu and Delgado, 1997), whereby CBZ-E levels are increased (Wagner et al., 1993). In a study by Jedrzejczak et al. (2000), VGB increased CBZ concentrations. There was no significant change in the serum concentrations of CBZ when LTG was added to a CBZ therapy (Eriksson and Boreus, 1997; Gidal et al., 1997b; Besag et al., 1998). Data regarding the influence of LTG on CBZ-E are conflicting. TPM (Sachdeo et al., 1996) and GBP (Radulovic et al., 1994) also do not influence CBZ levels.

Valproate

Effect of valproate on other drugs

As already discussed, VPA shows an inhibiting effect on the CBZ metabolite CBZ-E. When VPA is given in combination with CBZ, CBZ-E is increased (Sälke-Treumann et al., 1988; Rambeck et al., 1990; Bernus et al., 1997).

VPA increases PB levels by about 40% (Rambeck et al., 1979). Regarding PHT, there seem to be two mechanisms involved in the interaction of VPA with PHT. Whereas VPA displacing PHT from the plasma protein decreased the total drug concentration of PHT, the enzyme inhibition by VPA increased both the total and unbound concentration of PHT (Lai and Huang, 1993).

VPA increases levels of LTG in an impressive manner (Yuen et al., 1992; Anderson et al., 1996; May et al., 1996a; Battino et al., 1997; Kanner and Frey, 2000). In our study with 302 epileptic patients the LTG levels of patients on a combina- tion of LTG with VPA were increased by a factor of 3.6 in comparison to patients on LTG monotherapy (May et al., 1996a). This could benefit the patient with epilepsy not only by attaining higher plasma LTG concentrations with ‘standard’ dosages of LTG, but also possibly by achieving better seizure control through pro- viding a less variable peak-to-trough fluctuation in LTG concentrations as a result of extending the half-life of LTG (Morris et al., 2000).

VPA decreased the clearance of FBM by about 21% (Kelley et al., 1997). Although VPA seems to decrease the protein binding of TGB, the relevance of this effect is unclear. VPA does not influence the metabolism of other new AEDs.

VPA is metabolized to a series of saturated and unsaturated carbonic acids and glucuronidated derivatives. It is largely bound to serum proteins.

119 Interactions between antiepileptic drugs

Effect of other drugs on valproate

Inducing AEDs such as PB (May and Rambeck, 1985), PHT (May and Rambeck, 1985) or CBZ (May and Rambeck, 1985; Yukawa et al., 1997) decrease VPA levels considerably. In accordance with our own observations (Mataringa et al., 2002) Besag et al. (2001) reported that MSM also significantly decreases VPA levels. Besides the inducing effects of other AEDs on VPA, it has to be considered that the kinetics of VPA is non-linear, resulting in a lower than proportional increase of the serum concentration when increasing the dose. These two facts are the reason why in polytherapy even with high dosages of up to 6 g VPA per day, morning concen- trations higher than 100 g/ml are rarely exceeded.

Ethosuximide (ESM) seems to reduce VPA levels by an unknown mechanism (Sälke-Kellermann et al., 1997).

VPA levels rose by 12.7% when FBM was added (Wagner et al., 1994; Hooper et al., 1996; Siegel et al., 1999). In a study with human volunteers, the addition of LTG was associated with a small but significant decrease in steady-state VPA plasma concentration (Anderson et al., 1996). Mataringa et al. (2002) observed also a slight decreasing effect of LTG on VPA ( 7%) in a retrospective study. However, in clinical studies such an effect was not documented (Jawad et al., 1987; Eriksson et al., 1996). The effect of TPM on VPA kinetics seems to be negligible (Rosenfeld et al., 1997). GBP (Radulovic et al., 1994) or VGB (Armijo et al., 1992) do not influ- ence the kinetics of VPA.

Ethosuximide

ESM is a simple aliphatic compound which is metabolized to hydroxylated com- pounds. It is not bound to proteins. Besides a weak decreasing effect on VPA, ESM does not influence other drugs.

The metabolism of ESM itself may be induced to some degree by PB and PHT, but this is hardly of clinical relevance (Sälke-Kellermann et al., 1997).

Methsuximide

MSM is rapidly metabolized to the therapeutically active derivative N-desmethyl- MSM and then to hydroxylated and glucuronidated derivatives.

Effect of methsuximide on other drugs

MSM inhibits the metabolism of PHT and PB. In a study with 94 epileptic patients MSM increased concentrations of PB by 38%, of PB as metabolite of PRM by 40% and of PHT by 78%, in many cases with ensuing side effects (Rambeck, 1979).

But MSM also has enzyme-inducing effects and lowers LTG (May et al., 1999; Besag et al., 2000), VPA (Besag et al., 2001) and TPM levels (May et al., 2002).

120 Bernhard Rambeck and Theodor W. May

Effect of other drugs on methsuximide

PB and PHT can increase concentrations of N-desmethyl-MSM, the metabolite of MSM, in a competitive manner as these substances are metabolized by the same hydroxylating liver enzymes (Rambeck, 1979).

Interactions between new AEDs and other AEDs

In the last decade, a series of new AEDs have become available for the treatment of epileptic patients. One of the basic reasons to develop new AEDs was the aim of finding agents which are not or only to a small degree interactive with other drugs; but this aim has only partially been reached.

Felbamate

Effect of felbamate on other drugs

Early studies (Wilensky et al., 1985; Fuerst et al., 1988) with only a few patients showed that adding FBM resulted in an increase in PHT concentrations and a small decrease in CBZ concentrations. These effects were also found in a clinical trial with FBM by Graves et al. (1989) where 32 patients received concomitant PHT and CBZ treatment. All patients required a PHT dose reduction of 10–30% during FBM treatment to maintain stable PHT concentrations. CBZ serum concentra- tions decreased (mean 1.3 g/ml) in nearly all patients. Theodore et al. (1991) also found a significant reduction (24%) of CBZ concentrations in a clinical study with FBM. Albani et al. (1991) reported on a controlled trial where FBM was added to a stable CBZ monotherapy of 22 patients. CBZ total concentrations were lower during FBM treatment (mean reduction 25%). Wagner et al. (1993) evaluated the effect of FBM on CBZ and its major metabolites during a trial in 26 patients. Mean CBZ concentrations decreased from 7.5 g/ml during placebo treatment to 6.1 g/ml during FBM treatment. Mean CBZ-E concentrations increased from 1.8 to 2.4 g/ml. The effects of FBM on the kinetics of PB and its hydroxylated metabolite were assessed in a study with 24 healthy volunteers by Reidenberg et al. (1995a). FBM increased the area under the curve (AUC) of PB by 22% and the maximum concentration (Cmax) by 24%.

Wagner et al. (1994) showed that VPA doses may require reduction when FBM is added to a regimen of VPA. Co-administration of FBM increased the mean AUC, Cmax and average steady-state concentrations (from 67 to 103 g/ml) of VPA in 10 epileptic patients who received FBM in addition to a stable VPA dosage. This effect

FBM is partly bound to plasma proteins (24–35%) and eliminated by renal excre- tion, hydroxylation and conjugation.

121 Interactions between antiepileptic drugs

has also been documented by Hooper et al. (1996) in a study of 18 healthy volunteers.

FBM has only a small increasing effect (Colucci et al., 1996) or no effect on LTG (Gidal et al., 1997a). Reidenberg et al. (1995b) found no clinically relevant inter- actions between FBM and VGB in a study of 18 healthy volunteers. The influence of FBM on the multiple dose kinetics of monohydroxy and dihydroxy metabolites of OCBZ was assessed in healthy volunteers (Hulsman et al., 1995). FBM had no effect on MHD kinetics.

Effect of other drugs on felbamate

PHT and CBZ induce the metabolism of FBM resulting in lower than expected steady-state concentrations. Wagner et al. (1991) performed a controlled discon- tinuation study of PHT and CBZ in five patients with FBM. As PHT dosages were reduced, FBM clearance decreased by 21% and as the CBZ dosages were reduced, FBM clearance decreased by an additional 16.5%.

In a study by Kelley et al. (1997), PB had no influence on FBM, and VPA reduced the clearance of FBM by about 21%.

Reidenberg et al. (1995b) did not find any clinically relevant influence of VGB on FBM. Furthermore, LTG has no influence on FBM (Troupin et al., 1997). However, a study indicated that the half-life of FBM is increased by GBP via an unknown mechanism (Hussein et al., 1996).

Gabapentin

GBP shows dose-dependent absorption kinetics. It is not bound to plasma proteins and it is eliminated unchanged in the urine.

Effect of gabapentin on other drugs

The US Gabapentin Study Group (1994) found no influence of GBP on CBZ, PHT and VPA concentrations in a study with GBP as add-on therapy.

When administered over a period of 3 days, GBP had no statistically significant effect on PB concentrations in 12 healthy volunteers (Hooper et al., 1991). Clinical studies have also documented a lack of interaction between GBP and PB (Crawford et al., 1987; Goa and Sorkin, 1993). Radulovic et al. (1994) investigated the effect of GBP co-administration for more than 3 days on steady-state CBZ concentrations (12 epileptic patients) and for more than 5 days on VPA concentrations (14 epilep- tic patients). Mean CBZ and CBZ-E and mean VPA concentrations before, during and after GBP administration were not significantly different.

Crawford et al. (1987) performed a dose-ranging study with 300, 600 and 900 mg/day GBP as add-on therapy. No significant drug interactions were seen,

122 Bernhard Rambeck and Theodor W. May

although there was a trend towards elevation of serum PHT concentration in patients taking 900 mg/day of GBP.

There is also a case report about a considerable PHT increase after the addition of low doses of GBP (300 and 600mg/day) to PHT with CBZ and clobazam as comedication (Tyndel, 1994). The authors conclude that the unusual step of adding GBP to three AEDs may have allowed this unusual interaction. But, it seems rather problematic to draw such a conclusion from a single clinical observation with few serum level determinations since, for example, irregular drug intake prior to addition of GBP may also result in an increase of serum concentrations.

As mentioned above, GBP might elevate FBM levels.

Effect of other drugs on gabapentin

The above-mentioned investigation by Hooper et al. (1991) found no statistically significant influences of PB on GBP kinetics. There are no special studies about PHT, but according to our own experience PHT does not significantly influence GBP concentrations. In the study by Radulovic et al. (1994), GBP pharmacokinetic parameters during CBZ or VPA co-administration were similar to data reported in healthy subjects. The authors conclude that no pharmacokinetic interaction exists between CBZ or VPA and GBP.

Lamotrigine

LTG is about 55% bound to to plasma proteins and is extensively metabolized by glucuronidation.

Effect of lamotrigine on other drugs

Concentrations of concomitant VPA, PHT or CBZ were unaltered by 1 week of LTG administration in 22 patients examined by Jawad et al. (1987). Loiseau et al. (1990) reported on a controlled add-on trial of LTG in 23 patients. Concentrations of PHT, CBZ and PB remained unchanged. Sander et al. (1990) also performed a controlled add-on trial of LTG in 21 epileptic patients. Serum concentrations of CBZ, PHT, VPA and PB were unaffected by LTG treatment. Jawad et al. (1989) assessed the antiepileptic effects of LTG in a crossover trial in 24 adult patients. No statistically significant changes in concentrations of PHT, CBZ, PRM or PB were found between the two treatment periods. Schapel et al. (1993) performed a con- trolled trial of LTG as add-on therapy in 41 patients. Concomitant AEDs (CBZ, PHT and VPA) concentrations were virtually unchanged. Moreover, no clinically important changes in plasma concentrations of CBZ, VPA, ESM and PB were observed in epileptic children during LTG therapy (Eriksson et al., 1996).

In contrast, an interaction between LTG and CBZ metabolism resulting in an increase of CBZ-E of 45% was reported by Warner et al. (1992). These observations

123 Interactions between antiepileptic drugs

are at variance with those of Wolf (1992). He added LTG to a subtoxic, just toler- ated dose of CBZ in nine patients. Cerebellar toxicity developed in eight of them. In the total group, a small (about 10%) but significant increase of CBZ-E was found, whereas no consistent change could be detected in CBZ. The increase in the CBZ-E, however, was too small and too inconsistent to explain the toxicity in all cases. These results indicate that the interaction of CBZ and LTG may be primarily phar- macodynamic rather than pharmacokinetic. Pisani et al. (1994) found no effect of LTG on CBZ-E. They compared the pharmacokinetics of a single dose of 100 mg CBZ- E in 10 patients on chronic LTG monotherapy and in 10 drug-free healthy control sub- jects. CBZ-E kinetic parameters were similar in subjects on LTG and in controls.

Effect of other drugs on lamotrigine

Binnie et al. (1986) reported on short-term effects of a single dose of LTG in 16 persons with epilepsy. Comedication with CBZ and/or PHT reduced the elimina- tion half-life to a mean of 15 h and comedication with VPA prolonged the half-life to a mean of 59 h. In a study by Jawad et al. (1987), patients receiving LTG together with enzyme-inducing AEDs showed as LTG plasma elimination half-life of 14 7 h (mean SD). Those receiving LTG plus an inducing AED plus VPA exhibited a mean LTG half-life of 30 10 h.

Yuen et al. (1992) studied six healthy volunteers who received LTG as a single dose alone or together with VPA. Concomitant administration of VPA reduced LTG total clearance by approximately 21% and increased the elimination half-life and AUC. Renal elimination of LTG was not impaired.

May et al. (1996a) studied the influence of comedication on LTG concentrations in 588 blood samples of 302 epileptic patients. The LTG serum concentration in relation to LTG dose per body weight (level-to-dose ratio, LDR, g/ml per mg/kg) was calculated and compared for different drug combinations. The results showed that comedication had a highly significant influence on the LTG serum concentrations. The mean LDR for LTG was as follows: 0.32 (LTG PHT) 0.52 (LTG PB) 0.57 (LTG CBZ) 0.98 (LTG monotherapy) 0.99 (LTG VPA PHT) 1.67 (LTG VPA CBZ) 1.80 (LTG VPA PB) 3.57 (LTG VPA). The considerable influence of various AED and their combinations on LTG concentra- tions is shown in Figure 7.1. It is interesting that a comparable study by Battino et al. (1997) with 482 LTG determinations form 106 epileptic patients found nearly the same values. The LDR of LTG for patients on VPA was 3.2, for patients on enzyme-inducing drugs 0.6 and on VPA in combination with enzyme-inducing drugs 1.9. These data furthermore were confirmed in a prospective study with epileptic children (Bartoli et al., 1997).

As already mentioned, several studies (May et al., 1999; Besag et al., 2000) found that MSM lowers LTG levels by about 50–70%.

124

Bernhard Rambeck and Theodor W. May

Significantly higher

n.s. different from LTG mono

Significantly lower

Figure 7.1

(LTG monotherapy 100%) 450

400

350

300

250

%
200

150 100 50 0

Influence of PHT, PB, CBZ, MSM, GBP, OCBZ, VPA and of their combinations on serum concentrations of LTG. LTG monotherapy is taken as 100% (n.s.: not significantly; bars 95% confidence intervals); data of 302 patients (May et al., 1996a)

The decreasing effect of OXC on LTG levels (29%) is less than that of CBZ but statistically significant (May et al., 1999). FBM and TPM (Berry et al., 1998; Doose et al., 2003) have no important influence on LTG.

Oxcarbazepine

OXC is the 10-keto analogue of CBZ. OXC is a prodrug for MHD, and is rapidly converted to this substance. MHD is approximately 40% bound to serum proteins and is excreted mainly by direct conjugation to glucuronic acid.

Effect of oxcarbazepine on other drugs

McKee et al. (1994) investigated the interaction between OXC and other AEDs in three groups of 12 epileptic patients taking CBZ, VPA or PHT as monotherapy. No differences in the median AUC at steady-state of CBZ and its metabolite CBZ-E, as well as VPA and PHT, were observed during additional treatment with OXC at steady-state compared with the AUC calculated for the placebo phase, suggesting an absence of metabolic interference with these AEDs. In contrast, Barcs et al. (2000) found in an OXC dose-ranging trial a slight decrease in CBZ levels of 13%,

PHT PB PHT

PHT CBZ MSM

PB CBZ PB

CBZ LTG

GBP OXC

VPA PHT VPA CBZ

VPA PB VPA

125 Interactions between antiepileptic drugs

an increase in PB levels of 15% and an increase of PHT levels of 40% in patients with high MHD concentrations, compared to placebo. The in vitro study by Lakehal et al. (2002) indicated that MHD inhibited CYP2C19-mediated PHT metabolism at therapeutic concentrations. Thus, administration of OXC with CYP2C19 sub- strates with narrow therapeutic ranges should be done cautiously.

Battino et al. (1992) investigated changes of unbound and total VPA concentra- tions after replacement of CBZ with OXC in four epileptic patients. In confirmation of the above results, total and free VPA concentrations rose when the medication was switched from CBZ to OXC. Houtkooper et al. (1987) also observed a statisti- cally significant increase of concomitant VPA and PHT concentrations in a crossover trial with 48 patients when CBZ was replaced by OXC. The increase in the serum concentrations during OXC therapy can be explained by a decrease in the prior enzyme induction caused by CBZ (Houtkooper et al., 1987).

The inducing properties of OXC on the metabolism of LTG and TPM are less pronounced than that of CBZ but the inducing effect is statistically significant (May et al., 1999, 2002). A mean decrease in LTG levels of about 30% compared to LTG monotherapy was found. A comparable effect was also found on TPM metabolism, patients on OXC comedication had about 30% lower TPM levels than patients on TPM monotherapy (May et al., 2002).

Effect of other drugs on oxcarbazepine

Kumps and Wurth (1990) analyzed the concentrations of MHD and of the inactive metabolite CBZ-diol in 15 epileptic patients, six of them receiving PB and/or PHT as comedication. The results indicate that MHD concentrations are unaffected by the comedication, but oxidation of MHD to its inactive metabolite may be induced. However, this seems to be of little clinical significance. In the above- mentioned study of McKee et al. (1994), patients taking CBZ or PHT had lower MHD concentrations compared with control patients without CBZ or PHT, the difference being small and statistically significant only for the CBZ-treated group. VPA had no effect at all in this study. In contrast, the OXC dose-ranging study of Barcs et al. (2000) found that patients receiving concomitant treatment with PHT and PB had statistically lower MHD levels than patients not receiving these AEDs.

The absence of an effect of VPA was confirmed by Tartara et al. (1993). The kinetics of OXC and MHD after a single oral OXC dose were comparable in healthy control subjects and in epileptic patients treated with VPA. However, in patients on PB the AUC values of both OXC and MHD were lower and the MHD half-life marginally shorter than in controls. But the magnitude of this effect was judged to be only of minor clinical significance. In combination with VPA the free fraction of MHD (64%) was slightly, but significantly, higher than in monotherapy (52%) with OXC (May et al., 1996b).

126 Bernhard Rambeck and Theodor W. May

The study of Hulsman et al. (1995) documented the absence of an influence of FBM on MHD and its metabolite. Further observations indicate that LTG and GBP have no influence on MHD (Sallas, 1999; Viola et al., 2000).

Vigabatrin

Effect of vigabatrin on other drugs

In an early double-blind study (Grant and Heel, 1991) of VGB in epileptic patients, serum concentrations of PHT were about 20% lower during VGB treatment than during placebo, but concentrations of other concomitant AEDs did not change. In a study with 89 epileptic patients, Browne et al. (1989) also found a statistically significant decrease of 20% in PHT concentrations when VGB was added. Furthermore, minor decreases in PB (7%) and PRM (11%) were observed. Dalla Bernardina et al. (1995) performed a study in 46 epileptic children. Serum con- centrations of associated AEDs (CBZ, PB and VPA) showed no significant changes, except for PHT which decreased from 19.3 8.0 to 11.9 5.2 g/ml on VGB treatment. The effect of VGB on PHT has been further studied by Rimmer and Richens (1989). When VBG was added to the PHT therapy of eight epileptic patients, mean plasma PHT concentrations fell significantly by 23% during the 5th week. No change was found in plasma protein binding of PHT, the urinary ratio of PHT to its metabolite p-hydroxy-PHT, and the antipyrine clearance before and at the end of the treatment period. It is not clear why the fall in PHT levels may show a delay of a few weeks. This slight, but unequivocal, effect was confirmed in 21 epileptic patients by Gatti et al. (1993). By switching from oral to intravenous PHT for 5 days before and after combined treatment with VGB and by measuring p-hydroxy-PHT, it could be demonstrated that the oral availability of PHT is unaf- fected by VGB. So the mechanism of the VGB-induced decrease in serum PHT is still unclear. A dose–response study of VGB in 20 children aged 2 months to 18 years also showed a modest decrease in PHT plasma levels (Herranz et al., 1991), but no changes in CBZ and VPA levels.

Armijo et al. (1992) investigated the effects of adding VGB to the antiepileptic regimens of 16 children. In the eight patients receiving VPA, no significant changes of VPA concentrations were observed.

Furthermore, several controlled trials have shown that VGB has no significant effect on serum concentrations of CBZ and VPA (Gram et al., 1985), CBZ, PB, PHT and VPA (Loiseau et al., 1986) or CBZ and PB (Cocito et al., 1989).

On the other hand, Jedrzejczak et al. (2000) found in a study with 66 epileptic patients a small increasing influence (of about 10%) of VGB on CBZ. Some patients

VGB does not bind to plasma proteins, does not appear to undergo metabolic transformation and is excreted extensively in urine in its unchanged form.

127 Interactions between antiepileptic drugs

responded with adverse, toxic symptoms. Also Sanchez-Alcaraz et al. (2002) reported higher CBZ concentrations during comedication with VGB compared to CBZ mono- therapy in 15 patients.

As already mentioned, the study of Reidenberg et al. (1995b) found no clinically relevant influence of VGB on FBM in healthy volunteers.

Effect of other drugs on vigabatrin

In the study of Armijo et al. (1992) no differences were found in VGB concentra- tions between patients with and without VPA. In a retrospective study (Armijo et al., 1997), patients with and without enzyme-inducing AEDs (PHT, PB and CBZ) had comparable VGB levels. One study (Sanchez-Alcaraz et al., 1996) reported on a small decreasing influence of CBZ.

An investigation by Reidenberg et al. (1995b) concluded that FBM does not influence the inactive R( )-VGB enantiomer, but produced a 13% increase in AUC and an 8% increase in urinary excretion of the active S( ) enantiomer.

Topiramate

Effect of topiramate on other drugs

TPM has no significant or only little effect on the serum concentrations of CBZ or its metabolite CBZ-E (Sachdeo et al., 1996) or on PB, PRM and LTG, except for an occasional moderate increase in plasma PHT levels (Walker and Patsalos, 1995), and a small mean decrease of VPA levels, but this is hardly clinically relevant (Rosenfeld et al., 1997).

LTG does not influence TPM levels to a clinically relevant extent (Berry et al., 1998; Doose et al., 2003).

Effect of other drugs on topiramate

The elimination half-life of TPM of approximately 20–30 h may be shortened con- siderably in the presence of concomitant treatment with enzyme inducers such as PB, PHT or CBZ and lead to a decrease in TPM levels (Sachdeo et al., 1996; Glauser et al., 1999; Rosenfeld et al., 1999; May et al., 2002). Furthermore, MSM and to a lesser degree OXC reduces TPM levels (May et al., 2002). VPA (Rosenfeld et al., 1999), LTG and GBP (Contin et al., 2002; May et al., 2002; Doose et al., 2003) have no significant influence on TPM.

TPM is only 15% plasma protein bound and it is mainly excreted unchanged in the urine (80%), but significant metabolism occurs when TPM is administered in conjunction with enzyme-inducing AEDs.

128 Bernhard Rambeck and Theodor W. May

Tiagabine

Effect of tiagabine on other drugs

TGB does not influence serum concentrations of other AEDs, as was shown in studies for CBZ and PHT (Gustavson et al., 1998a), VPA (Gustavson et al., 1998b) and for CBZ, PHT, VPA, VGB (Richens et al., 1995). This lack of interactions is under- standable because of its low concentration, in the nanogram range.

Effect of other drugs on tiagabine

Enzyme inducers such as CBZ, PB and PHT reduce the elimination half-life of TGB considerably (So et al., 1995; Snel et al., 1997).

Levetiracetam

LEV is a new AED with a nearly ideal pharmacokinetic profile. It shows a high bioavailability, linear and time-invariant kinetics, minimal protein binding and a low metabolism to an inactive metabolite.

In some clinical trials, the addition of LEV increased PHT levels to variable degrees in a few patients (Sharief et al., 1996; Patsalos, 2000), but this effect could not be confirmed by trials with deuterium-labeled PHT (Browne et al., 2000). Besides this unexplained effect no clinically relevant interactions are known. Perucca et al. (2000) found no interactions between other AED and LEV. However, more recent studies indicate that enzyme-inducing AEDs (May et al., 2003; Perucca et al., 2003) and OCBZ (May et al., 2003) slightly decrease LEV concentrations.

Zonisamide

ZNS is rapidly and completely absorbed. It is approximately 50% bound to pro- teins and has a relatively long half-life of about 63–69 h. It is partly metabolized with non-linear kinetics.

Effect of zonisamide on other drugs

Conflicting results have been found regarding the influence of ZNS on comed- icated AEDs. Sackellares et al. (1985) showed a consistent rise in concentrations of the comedication, particularly of CBZ, when ZNS was administered to 10 adult patients in a pilot study. In contrast, in a study by Minami et al. (1994) the average LDR of CBZ was lower in patients with ZNS than in patients without ZNS. Other studies could not demonstrate a relevant influence of ZNS on concentrations or protein binding of concomitant AEDs such as CBZ, PHT, PB, PRM or VPA (Schmidt et al., 1993) and PHT or VPA (Tasaki et al., 1995).

TGB is 96% protein bound. It is metabolized in the liver and only small portions are excreted unchanged.

129

Table 7.1 Pharmacokinetic interactions of AEDs

Effect of/on … CBZ PB

PHT PRM ESM MSM ↓/↑ ↓ PB ↑ ↓

VPA VGB ↓↓ (↓)

GBP LEV /↓

TGB TPM

LTG

OXCe FBM ZNS ↓↓↓↓↓

CBZ
PB ↓↓a
PHT ↓↓a /↑
PRM ↓↓ CBZ-E ↑
ESM ↓

MSM
VPA
VGB
GBP (↑) LEV (↑) TGB TPM ↑/↑↑

↓↓

↓

↓↓ ↓↓ ↑↑

↓ CBZ-E ↑↑ /↑

f
(↑)

LTG (CBZ-E ↑) OXC /↓ /↑ ↑/↑↑ FBM ↓CE↓↓↑↑↑↑ ZNS ↑/↓ CE ↑ (↑)

(↑)

↑↑ ↑↑ ↑↑ ↓c (↓) ↓/↓↓

/↑ g

↓/↑b ↓ ↑ ↓PB↑ ↓ ↑

↓↓ ↓↓ ↓↓

/↓ /↓ /↓

↓↓ ↓↓ ↓↓ ↓ ↓↓ ↓↓ ↓↓ ↓

↓↓ ↓↓ ↓↓ ↓↓

/↓ /↓ ↓↓ /↓ ↓↓ ↓↓ ↓

↓ PB ↑ (↓) PB ↑ (↓)

(↓)
/↓
↑↑ /↑

: no relevant or statistically significant interaction. ↑ and ↓: increase and decrease, respectively, of serum concentrations mostly without clinical relevance.
↑↑ and ↓↓: clinically relevant increase and decrease, respectively, of serum concentrations. Different symbols (e.g. /↑): indication of inconsistent or contradictory observations. Arrows in parentheses: indication of interactions based on case reports or on a small number of patients. Empty cells: no data available.
a CBZ decreases, ratio CBZ-E/CBZ increases.
b Dependent on concentration of PB.
c VPA decreases PHT total concentration; however, as VPA simultaneously increases the free fraction of PHT, these effects cancel each other to a great extent.
d VPA probably increases the free fraction of TGB.
e Data regarding the clinically relevant 10-hydroxy-carbazepine (MHD).
f VPA slightly increases the free fraction of 10-hydroxy-carbazepine (MHD).
g VPA slightly increases the free fraction of ZNS.
For clarity, bromide, sulthiame and benzodiazepines are not listed. Clinically relevant interactions of bromide with other AEDs are unlikely and were not reported. Sulthiame can increase concentrations of PHT markedly; other clinically relevant interactions with sulthiame are unknown. For benzodiazepines, relatively few interactions with other AEDs are reported.

/↓ ↓↓

130 Bernhard Rambeck and Theodor W. May

Effect of other drugs on zonisomide

The study of Shinoda et al. (1996) indicated that enzyme-inducing AEDs (PB, PHT and CBZ) reduce the ZNS LDR. Ojemann et al. (1996) investigated the influence of CBZ and PHT on kinetics of a single dose of ZNS in epileptic patients. Plasma half- life of ZNS was significantly higher in patients on CBZ (36.4 h) than in those on PHT therapy (27.1 h), but both values were shorter than half-life values (50–68 h) usually found after administration of single oral doses on ZNS in healthy volunteers.

VPA has no clinically relevant influence on ZNS levels (Shinoda et al., 1996). Therapeutic implications

The experience of several decades with the classic AEDs has shown that interactions may have severe clinical consequences. However, in the case of the pharmaco- kinetic interactions of the new AEDs their clinical importance is less clear. This is because the relevance of serum concentrations and of therapeutic drug monitor- ing for avoidance of side effects or for reduction of seizures has not yet been defin- itively established for most of the new AEDs. For FBM, LTG, OXC, TPM and GBP, a relation between serum concentrations and antiepileptic effect probably exists, but further studies are necessary to clarify this important topic. This seems also to be true for the correlation between serum concentrations and side effects of FBM, LTG and OXC. In contrast, in the case of VGB the antiepileptic effect and side effects seem to be unrelated to the serum concentration of the drug.

Reports on AED interactions usually focus on the increase or decrease of serum concentrations. But, it should be borne in mind that interactions moreover influ- ence the whole pharmacokinetic properties of an AED. For example, changes of the half-life time also affect daily fluctuations of serum levels and furthermore the time to reach steady-state concentrations or the speed of elimination after with- drawal of a drug.

The pharmacokinetic interactions (summarized in Table 7.1) of the old and new AEDs have been investigated by many studies. Most interactions correspond to the pharmacokinetic properties of the compounds, but it should be borne in mind that rare interactions may also play an important role in the individual.

REFERENCES

Abarbanel J, Herishanu Y, Rosenberg P, et al. In vivo interaction of anticonvulsant drugs. The mathematical correlation of plasma levels of anticonvulsant drugs in epileptic patients. J Neurol 1978; 218: 137–144.

Albani F, Theodore WH, Washington P, et al. Effect of felbamate on plasma levels of carbamazepine and its metabolites. Epilepsia 1991; 32: 130–132.

131 Interactions between antiepileptic drugs

Anderson GD, Yau MK, Gidal BE, et al. Bidirectional interaction of valproate and lamotrigine in healthy subjects. Clin Pharmacol Ther 1996; 60: 145–156.

Armijo JA, Arteaga R, Valdizan EM, et al. Coadministration of vigabatrin and valproate in chil- dren with refractory epilepsy. Clin Neuropharmacol 1992; 15: 459–469.

Armijo JA, Cuadrado A, Bravo J, et al. Vigabatrin serum concentration to dosage ratio: influence of age and associated antiepileptic drugs. Ther Drug Monit 1997; 19: 491–498.

Barcs G, Walker EB, Elger CE, et al. Oxcarbazepine placebo-controlled, dose-ranging trial in refractory partial epilepsy. Epilepsia 2000; 41: 1597–1607.

Bartoli A, Guerrini R, Belmonte A, et al. The influence of dosage, age, and comedication on steady state plasma lamotrigine concentrations in epileptic children: a prospective study with preliminary assessment of correlations with clinical response. Ther Drug Monit 1997; 19: 252–260.

Battino D, Croci D, Granata T, et al. Changes in unbound and total valproic acid concentra- tions after replacement of carbamazepine with oxcarbazepine. Ther Drug Monit 1992; 14: 376–379.

Battino D, Croci D, Granata T, et al. Lamotrigine plasma concentrations in children and adults: influence of age and associated therapy. Ther Drug Monit 1997; 19: 620–627.

Bernus I, Dickinson RG, Hooper WD, et al. The mechanism of the carbamazepine–valproate interaction in humans. Br J Clin Pharmacol 1997; 44: 21–27.

Berry DJ, Besag C, Pool F, et al. Does topiramate change lamotrigine serum concentrations when added to treatment? An audit of a dose-escalation study. Epilepsia 1998; 39(Suppl. 6): 56–57. Besag FM, Berry DJ, Pool F, et al. Carbamazepine toxicity with lamotrigine: pharmacokinetic or

pharmacodynamic interaction? Epilepsia 1998; 39: 183–187.
Besag FM, Berry DJ, Pool F. Methsuximide lowers lamotrigine blood levels: a pharmacokinetic

antiepileptic drug interaction. Epilepsia 2000; 41: 624–627.
Besag FMC, Berry DJ, Vasey M. Methsuximide reduces valproic acid serum levels. Ther Drug

Monit 2001; 23: 694–697.
Binnie CD, van Emde Boas W, Kasteleijn-Nolste-Trenite DG, et al. Acute effects of lamotrigine

(BW430C) in persons with epilepsy. Epilepsia 1986; 27: 248–254.
Bowdle TA, Levy RH, Cutler RE. Effects of carbamazepine on valproic acid kinetics in normal

subjects. Clin Pharmacol Ther 1979; 26: 629–634.
Brodie MJ. Tiagabine pharmocology in profile. Epilepsia 1995; 36(Suppl. 6): S7–S9.
Browne TR, Szabo GK, Evans J, et al. Phenobarbital does not alter phenytoin steady-state serum

concentration or pharmacokinetics. Neurology 1988a; 38: 639–642.
Browne TR, Szabo GK, Evans J, et al. Carbamazepine increases phenytoin serum concentration and

reduces phenytoin clearance. Neurology 1988b; 38: 1146–1150.
Browne TR, Mattson RH, Penry JK, et al. A multicentre study of vigabatrin for drug-resistant

epilepsy. Br J Clin Pharmacol 1989; 27(Suppl. 1): S95–S100.
Browne TR, Szabo GK, Leppik IE, et al. Absence of pharmacokinetic drug interaction of leve-

tiracetam with phenytoin in patients with epilepsy determined by new technique. J Clin

Pharmacol 2000; 40: 590–595.
Bruni J, Wilder BJ, Perchalski RJ, et al. Valproic acid and plasma levels of phenobarbital.

Neurology 1980; 30: 94–97.

132 Bernhard Rambeck and Theodor W. May

Christiansen J, Dam M. Influence of phenobarbital and diphenylhydantoin on plasma carba- mazepine levels in patients with epilepsy. Acta Neurol Scand 1973; 49: 543–546.

Cocito L, Maffini M, Perfumo P, et al. Vigabatrin in complex partial seizures: a long-term study. Epilepsy Res 1989; 3: 160–166.

Colucci R, Glue P, Holt B, et al. Effect of felbamate on the pharmacokinetics of lamotrigine. J Clin Pharmacol 1996; 36: 634–638.

Contin M, Riva R, Albani F, et al. Topiramate therapeutic monitoring in patients with epilepsy: effect of concomitant antiepileptic drugs. Ther Drug Monit 2002; 24(3): 332–337.

Crawford P, Ghadiali E, Lane R, et al. Gabapentin as an antiepileptic drug in man. J Neurol Neurosurg Psychiat 1987; 50: 682–686.

Dalla Bernardina B, Fontana E, Vigevano F, et al. Efficacy and tolerability of vigabatrin in chil- dren with refractory partial seizures: a single-blind dose-increasing study. Epilepsia 1995; 36: 687–691.

Doose DR, Brodie MJ, Wilson EA, et al. Topiramate and lamotrigine pharmacokinetics during repetitive monotherapy and combination therapy in epilepsy patients. Epilepsia 2003; 44: 917–922.

Duncan JS, Patsalos PN, Shorvon SD. Effects of discontinuation of phenytoin, carbamazepine, and valproate on concomitant antiepileptic medication. Epilepsia 1991; 32: 101–115.

Eadie MJ, Lander CM, Hooper WD, et al. The effects of phenobarbitone dose on plasma pheno- barbitone levels in epileptic patients. Proc Aust Assoc Neurol 1976; 13: 89–96.

Encinas MP, Santos Buelga D, Alonso Gonzalez AC, et al. Influence of length of treatment on the interaction between phenobarbital and phenytoin. J Clin Pharm Ther 1992; 17: 49–50.

Eriksson AS, Hoppu K, Nergardh A, et al. Pharmacokinetic interactions between lamotrigine and other antiepileptic drugs in children with intractable epilepsy. Epilepsia 1996; 37: 769–773.

Eriksson AS, Boreus LO. No increase in carbamazepine-10,11-epoxide during addition of lam- otrigine treatment in children. Ther Drug Monit 1997; 19: 499–501.

Fincham RW, Schottelius DD, Sahs AL. The influence of diphenylhydantoin on primidone metabolism. Trans Am Neurol Assoc 1973; 98: 197–199.

Fuerst RH, Graves NM, Leppik IE, et al. Felbamate increases phenytoin but decreases carba- mazepine concentrations. Epilepsia 1988; 29: 488–491.

Gatti G, Bartoli A, Marchiselli R, et al. Vigabatrin-induced decrease in serum phenytoin concen- tration does not involve a change in phenytoin bioavailability. Br J Clin Pharmacol 1993; 36: 603–606.

Gidal BE, Kanner A, Maly M, et al. Lamotrigine pharmacokinetics in patients receiving felba- mate. Epilepsy Res 1997a; 27: 1–5.

Gidal BE, Rutecki P, Shaw R, et al. Effect of lamotrigine on carbamazepine epoxide/carba- mazepine serum concentration ratios in adult patients with epilepsy. Epilepsy Res 1997b; 28: 207–211.

Glauser TA, Miles MV, Tang P, et al. Topiramate pharmacokinetics in infants. Epilepsia 1999; 40: 788–791.

133 Interactions between antiepileptic drugs

Glue P, Banfield CR, Perhach JL, et al. Pharmacokinetic interactions with felbamate. In vitro-in vivo correlation. Clin Pharmacokinet 1997; 33: 214–224.

Goa KL, Sorkin EM. Gabapentin – a review of its pharmacological properties and clinical poten- tial in epilepsy. Drugs 1993; 46: 409–427.

Gram L, Klosterskov P, Dam M. -Vinyl GABA: a double-blind placebo-controlled trial in par- tial epilepsy. Ann Neurol 1985; 17: 262–266.

Grant SM, Heel RC. Vigabatrin – a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in epilepsy and disorders of motor control. Drugs 1991; 41: 889–926. Grasela TH, Fiedler-Kelly J, Cox E, et al. Population pharmacokinetics of lamotrigine adjunctive

therapy in adults with epilepsy. J Clin Pharmacol 1999; 39: 373–384.
Graves NM, Holmes GB, Fuerst RH, et al. Effect of felbamate on phenytoin and carbamazepine

serum concentrations. Epilepsia 1989; 30: 225–229.
Gustavson LE, Boellner SW, Granneman GR, et al. A single-dose study to define tiagabine

pharmacokinetics in pediatric patients with complex partial seizures. Neurology 1997; 48:

1032–1037.
Gustavson LE, Cato A, Boellner SW, et al. Lack of pharmacokinetic drug interactions between

tiagabine and carbamazepine or phenytoin. Am J Ther 1998a; 5: 9–16.
Gustavson LE, Sommerville KW, Boellner SW, et al. Lack of a clinically significant pharmacoki-

netic drug interaction between tiagabine and valproate. Am J Ther 1998b; 5: 73–79.
Hansen JM, Kristensen M, Skovsted L. Sulthiame (ospolot) as inhibitor of diphenylhydatoin

metabolism. Epilepsia 1968; 9: 17–22.
Herranz JL, Arteaga R, Farr IN, et al. Dose–response study of vigabatrin in children with refrac-

tory epilepsy. J Child Neurol Suppl 1991; 2: S45–S51.
Hooper WD, Kavanagh MC, Herkes GK, et al. Lack of a pharmacokinetic interaction between

phenobarbitone and gabapentin. Br J Clin Pharmacol 1991; 31: 171–174.
Hooper WD, Franklin ME, Glue P, et al. Effect of felbamate on valproic acid disposition in

healthy volunteers: inhibition of beta-oxidation. Epilepsia 1996; 37: 91–97.
Houtkooper MA, Lammertsma A, Meyer JW, et al. Oxcarbazepine (GP 47.680): a possible alter-

native to carbamazepine? Epilepsia 1987; 28: 693–698.
Hulsman JA, Rentmeester TW, Banfield CR, et al. Effects of felbamate on the pharmacokinetics

of the monohydroxy and dihydroxy metabolites of oxcarbazepine. Clin Pharmacol Ther 1995;

58: 383–389.
Hussein G, Troupin AS, Montouris G. Gabapentin interaction with felbamate. Neurology 1996;

47: 1106.
Inoue F, Chambers DN. The effect of phenobarbital on the pharmacokinetics of phenytoin – a

case report. Can J Hos Pharm 1985; 38: 147–148.
Jawad S, Yuen WC, Peck AW, et al. Lamotrigine: single-dose pharmacokinetics and initial 1 week

experience in refractory epilepsy. Epilepsy Res 1987; 1: 194–201.
Jawad S, Richens A, Goodwin G, et al. Controlled trial of lamotrigine (lamictal) for refractory

partial seizures. Epilepsia 1989; 30: 356–363.
Jedrzejczak J, Dlawichowska E, Owczarek K, et al. Effect of vigabatrin addition on carba-

mazepine blood serum levels in patients with epilepsy. Epilepsy Res 2000; 39: 115–120.

134 Bernhard Rambeck and Theodor W. May

Kanner AM, Frey M. Adding valproate to lamotrigine: a study of their pharmacokinetic inter- action. Neurology 2000; 55: 588–591.

Kapetanovic IM, Kupferberg HJ, Porter RJ, et al. Mechanism of valproate–phenobarbital interac- tion in epileptic patients. Clin Pharmacol Ther 1981; 29: 480–486.

Kelley MT, Walson PD, Cox S, et al. Population pharmacokinetics of felbamate in children. Ther Drug Monit 1997; 19: 29–36.

Kumps A, Wurth C. Oxcarbazepine disposition: preliminary observations in patients. Biopharm Drug Dispos 1990; 11: 365–370.

Lai ML, Lin TS, Huang JD. Effect of single- and multiple-dose carbamazepine on the pharmaco- kinetics of diphenylhydantoin. Eur J Clin Pharmacol 1992; 43: 201–203.

Lai ML, Huang JD. Dual effect of valproic acid on the pharmacokinetics of phenytoin. Biopharm Drug Dispos 1993; 14: 365–370.

Lakehal F, Wurden CJ, Kalhorn TF, et al. Carbamazepine and oxcarbazepine decrease phenytoin metabolism through inhibition of CYP2C19. Epilepsy Res 2002; 52: 79–83.

Lambie DG, Johnson RH. The effects of phenytoin on phenobarbitone and primidone metabo- lism. J Neurol Neurosurg Psychiat 1981; 44: 148–151.

Liu H, Delgado MR. Effect of valproate and felbamate on carbamazepine and its metabolite in epileptic children. J Epilepsy 1997; 10: 37–41.

Loiseau P, Hardenberg JP, Pestre M, et al. Double-blind, placebo-controlled study of vigabatrin (gamma-vinyl GABA) in drug-resistant epilepsy. Epilepsia 1986; 27: 115–120.

Loiseau P, Yuen AW, Duche B, et al. A randomised double-blind placebo-controlled crossover add-on trial of lamotrigine in patients with treatment-resistant partial seizures. Epilepsy Res 1990; 7: 136–145.

Mataringa MI, May TW, Rambeck B. Does lamotrigine influence valproate concentrations? Ther Drug Monit 2002; 24: 631–636.

May T, Stenzel E, Rambeck B. Phenytoin serum concentration in epileptic patients: influence of therapeutic and physiological factors. Nervenarzt 1982; 53: 291–296.

May T, Rambeck B. Serum concentrations of valproic acid: influence of dose and comedication. Ther Drug Monit 1985; 7: 387–390.

May T, Rambeck B. Fluctuations of unbound and total phenytoin concentrations during the day in epileptic patients on valproic acid comedication. Ther Drug Monit 1990; 12: 124–128.

May TW, Rambeck B, Nothbaum N. Nomogram for the prediction of unbound phenytoin con- centrations in patients on a combined treatment of phenytoin and valproic acid. Eur Neurol 1991; 31: 57–60.

May TW, Rambeck B, Jürgens U. Serum concentrations of lamotrigine in epileptic patients: the influence of dose and comedication. Ther Drug Monit 1996a; 18: 523–531.

May TW, Rambeck B, Sälke-Kellermann A. Fluctuations of 10-hydroxy-carbazepine during the day in epileptic patients. Acta Neurol Scand 1996b; 93(6): 393–397.

May TW, Rambeck B, Jürgens U. Influence of oxcarbazepine and methsuximide on lamotrigine concentrations in epileptic patients with and without valproic acid comedication: results of a retrospective study. Ther Drug Monit 1999; 21: 175–181.

May TW, Rambeck B, Jürgens U. Serum concentrations of topiramate in patients with epilepsy: influence of dose, age, and comedication. Ther Drug Monit 2002; 24: 366–374.

135 Interactions between antiepileptic drugs

May TW, Rambeck B, Jürgens U. Serum concentrations of levetiracetam in patients with epilepsy: influence of dose and comedication. Ther Drug Monit 2003; 25: 690–699.

McKee PJ, Blacklaw J, Forrest G, et al. A double-blind, placebo-controlled interaction study between oxcarbazepine and carbamazepine, sodium valproate and phenytoin in epileptic patients. Br J Clin Pharmacol 1994; 37: 27–32.

De Michele G, Brescia Morra V, Pisanti N, et al. Carbamazepine serum levels in epileptics. Influence of age, sex, body weight and interaction with phenobarbital. Acta Neurol (Napoli) 1985; 7: 228–234.

Minami T, Ieiri I, Ohtsubo K, et al. Influence of additional therapy with zonisamide (excegran) on protein binding and metabolism of carbamazepine. Epilepsia 1994; 35: 1023–1025.

Morris RG, Black AB, Lam E, et al. Clinical study of lamotrigine and valproic acid in patients with epilepsy: using a drug interaction to advantage? Ther Drug Monit 2000; 22: 656–660.

Müller FO, Aucamp AK, Hundt HK, et al. Evaluation of serum levels during prolonged combi- nation therapy with phenytoin and phenobarbitone. S Afr Med J 1977; 52: 356–358.

Ojemann LM, Shastri RA, Wilensky AJ, et al. Comparative pharmacokinetics of zonisamide (CI- 912) in epileptic patients on carbamazepine or phenytoin monotherapy. Ther Drug Monit 1986; 8: 293–296.

Patel IH, Levy RH, Cutler RE. Phenobarbital–valproic acid interaction. Clin Pharmacol Ther 1980; 27: 515–521.

Patsalos PN. Pharmacokinetic profile of levetiracetam: toward ideal characteristics. Pharmacol Ther 2000; 85: 77–85.

Perucca E, Gidal BE, Ledent E, et al. Levetiracetam does not interact with other antiepileptic drugs. Epilepsia 2000; 41(Suppl. 7): 254–255.

Perucca E, Gidal BE, Baltes E. Effects of antiepileptic comedication on levetiracetam pharmaco- kinetics: a pooled analysis of data from randomized adjunctive therapy trials. Epilepsy Res 2003; 53: 47–56.

Pisani F, Xiao B, Fazio A, et al. Single dose pharmacokinetics of carbamazepine-10,11-epoxide in patients on lamotrigine monotherapy. Epilepsy Res 1994; 19: 245–248.

Porro MG, Kupferberg HJ, Porter RJ, et al. Phenytoin: an inhibitor and inducer of primidone metabolism in an epileptic patient. Br J Clin Pharmacol 1982; 14: 294–297.

Radulovic LL, Wilder BJ, Leppik IE, et al. Lack of interaction of gabapentin with carbamazepine or valproate. Epilepsia 1994; 35: 155–161.

Rambeck B. Pharmacological interactions of mesuximide with phenobarbital and phenytoin in hospitalized epileptic patients. Epilepsia 1979; 20: 147–156.

Rambeck B, Boenigk HE, May T. Pharmacological influence of valproate on phenobarbitone and phenytoin serum concentrations in epileptic patients. Nervenarzt 1979; 50: 743–746.

Rambeck B, May T, Juergens U. Serum concentrations of carbamazepine and its epoxide and diol metabolites in epileptic patients: the influence of dose and comedication. Ther Drug Monit 1987; 9: 298–303.

Rambeck B, Sälke-Treumann A, May T, et al. Valproic acid-induced carbamazepine-10,11-epox- ide toxicity in children and adolescents. Eur Neurol 1990; 30: 79–83.

Reidenberg P, Glue P, Banfield CR, et al. Effects of felbamate on the pharmacokinetics of pheno- barbital. Clin Pharmacol Ther 1995a; 58: 279–287.

136 Bernhard Rambeck and Theodor W. May

Reidenberg P, Glue P, Banfield C, Colucci R, et al. Pharmacokinetic interaction studies between felbamate and vigabatrin. Br J Clin Pharmacol 1995b; 40: 157–160.

Reunanen MI, Luoma P, Myllyla VV, Hokkanen E. Low serum valproic acid concentrations in epileptic patients on combination therapy. Curr Ther Res Clin Exp 1980; 28: 456–462.

Richens A, Chadwick DW, Duncan JS, et al. Adjunctive treatment of partial seizures with tiagabine: A placebo-controlled trial. Epilepsy Res 1995; 21: 37–42.

Rimmer EM, Richens A. Interaction between vigabatrin and phenytoin. Br J Clin Pharmacol 1989; 27(Suppl. 1): 27S–33S.

Riva R, Albani F, Contin M, et al. Time-dependent interaction between phenytoin and valproic acid. Neurology 1985; 35: 510–515.

Rosenfeld WE, Liao S, Kramer LD, et al. Comparison of the steady-state pharmacokinetics of topiramate and valproate in patients with epilepsy during monotherapy and concomitant therapy. Epilepsia 1997; 38: 324–333.

Rosenfeld WE, Doose DR, Walker SA, et al. A study of topiramate pharmacokinetics and tolera- bility in children with epilepsy. Pediatr Neurol 1999; 20: 339–344.

Sachdeo RC, Sachdeo SK, Walker SA, et al. Steady-state pharmacokinetics of topiramate and car- bamazepine in patients with epilepsy during monotherapy and concomitant therapy. Epilepsia 1996; 37: 774–780.

Sachdeo R, Wagner ML, Sachdeo S, et al. Coadministration of phenytoin and felbamate: evidence of additional phenytoin dose-reduction requirements based on pharmacokinetics and tolera- bility with increasing doses of felbamate. Epilepsia 1999; 40: 1122–1128.

Sackellares JC, Donofrio PD, Wagner JG, et al. Pilot study of zonisamide (1,2-benzisoxazole- 3-methanesulfonamide) in patients with refractory partial seizures. Epilepsia 1985; 26: 206–211.

Sälke-Kellermann RA, May TW, Boenigk HE. Influence of ethosuximide on valproic acid serum concentrations. Epilepsy Res 1997; 26: 345–349.

Sälke-Treumann A, Rambeck B, May TW, et al. Nebenwirkungen durch carbamazepin-epoxid bei kindern. Epilepsie-Blätter 1988; (1): 39–41.

Sallas W. Population pharmacokinetics analysis of oxcarbazepine (trileptal) in children with epilepsy. Epilepsia 1999; 40(Suppl. 7): 102.

Sanchez-Alcaraz A, Quintana B, Rodriguez I, et al. Plasma concentrations of vigabatrin in epileptic patients. J Clin Pharm Ther 1996; 21: 393–398.

Sanchez-Alcaraz A, Quintana MB, Lopez E, et al. Effect of vigabatrin on the pharmacokinetics of carbamazepine. J Clin Pharm Ther 2002; 27: 427–430.

Sander JW, Patsalos PN, Oxley JR, et al. A randomised double-blind placebo-controlled add-on trial of lamotrigine in patients with severe epilepsy. Epilepsy Res 1990; 6: 221–226.

Schapel GJ, Beran RG, Vajda FJ, et al. Double-blind, placebo controlled, crossover study of lamotrigine in treatment resistant partial seizures. J Neurol Neurosurg Psychiat 1993; 56: 448–453.

Schmidt D. The effect of phenytoin and ethosuximide on primidone metabolism in patients with epilepsy. J Neurol 1975; 209: 115–123.

Schmidt D, Jacob R, Loiseau P, et al. Zonisamide for add-on treatment of refractory partial epilepsy: a European double-blind trial. Epilepsy Res 1993; 15: 67–73.

137 Interactions between antiepileptic drugs

Sharief MK, Singh P, Sander JWAS, et al. Efficacy and tolerability study of ucb L059 in patients with refractory epilepsy. J Epilepsy 1996; 9: 106–112.

Shinoda M, Akita M, Hasegawa M, et al. The necessity of adjusting the dosage of zonisamide when coadministered with other anti-epileptic drugs. Biol Pharm Bull 1996; 19: 1090–1092. Siegel H, Kelley K, Stertz B, et al. The efficacy of felbamate as add-on therapy to valproic acid in

the Lennox–Gastaut syndrome. Epilepsy Res 1999; 34: 91–97.
Snel S, Jansen JA, Mengel HB, et al. The pharmacokinetics of tiagabine in healthy elderly volun-

teers and elderly patients with epilepsy. J Clin Pharmacol 1997; 37: 1015–1020.
So EL, Wolff D, Graves NM, et al. Pharmacokinetics of tiagabine as add-on therapy in patients

taking enzyme-inducing antiepilepsy drugs. Epilepsy Res 1995; 22: 221–226.
Tartara A, Galimberti CA, Manni R, et al. The pharmacokinetics of oxcarbazepine and its active metabolite 10-hydroxy-carbazepine in healthy subjects and in epileptic patients treated with

phenobarbitone or valproic acid. Br J Clin Pharmacol 1993; 36: 366–368.
Tasaki K, Minami T, Ieiri I, et al. Drug interactions of zonisamide with phenytoin and sodium

valproate: serum concentrations and protein binding. Brain Dev 1995; 17: 182–185.
The US Gabapentin Study Group. The long-term safety and efficacy of gabapentin (neurontin)

as add-on therapy in drug-resistant partial epilepsy. Epilepsy Res 1994; 18: 67–73.
Theodore WH, Raubertas RF, Porter RJ, et al. Felbamate: a clinical trial for complex partial

seizures. Epilepsia 1991; 32: 392–397.
Troupin A, Montouris G, Hussein G. Felbamate: therapeutic range and other kinetic informa-

tion. J Epilepsy 1997; 10: 26–31.
Tyndel F. Interaction of gabapentin with other antiepileptics. Lancet 1994; 343: 1363–1364. Viola MS, Bercellini MA, Saidon P, et al. Pharmacokinetic variability of oxcarbazepine in epilep-

tic patients. Medicina Buenos Aires 2000; 60(6): 914–918.
Wagner ML, Graves NM, Marienau K, et al. Discontinuation of phenytoin and carbamazepine in

patients receiving felbamate. Epilepsia 1991; 32: 398–406.
Wagner ML, Remmel RP, Graves NM, et al. Effect of felbamate on carbamazepine and its major

metabolites. Clin Pharmacol Ther 1993; 53: 536–543.
Wagner ML, Graves NM, Leppik IE, et al. The effect of felbamate on valproic acid disposition.

Clin Pharmacol Ther 1994; 56: 494–502.
Walker MC, Patsalos PN. Clinical pharmacokinetics of new antiepileptic drugs. Pharmacol

Therapeutics 1995; 67: 351–384.
Warner T, Patsalos PN, Prevett M, et al. Lamotrigine-induced carbamazepine toxicity: an inter-

action with carbamazepine-10,11-epoxide. Epilepsy Res 1992; 11: 147–150.
Wilder BJ, Willmore LJ, Bruni J, et al. Valproic acid: interaction with other anticonvulsant drugs.

Neurology 1978; 28(9 Pt 1): 892–896.
Wilensky AJ, Friel PN, Ojemann LM, et al. Pharmacokinetics of W-554 (ADD 03055) in epilep-

tic patients. Epilepsia 1985; 26: 602–606.
Windorfer Jr A, Sauer W. Drug interactions during anticonvulsant therapy in childhood:

diphenylhydantoin, primidone, phenobarbitone, clonazepam, nitrazepam, carbamazepine

and dipropylacetate. Neuropädiatrie 1977; 8: 29–41.
Wolf P. Lamotrigine: preliminary clinical observations on pharmacokinetics and interactions

with traditional antiepileptic drugs. J Epilepsy 1992; 5: 73–79.

138 Bernhard Rambeck and Theodor W. May

Yuen AW, Land G, Weatherley BC, et al. Sodium valproate acutely inhibits lamotrigine metabo- lism. Br J Clin Pharmacol 1992; 33: 511–513.

Yukawa E, Honda T, Ohdo S, et al. Detection of carbamazepine-induced changes in valproic acid relative clearance in man by simple pharmacokinetic screening. J Pharm Pharmacol 1997; 49: 751–756.

Interaction between antiepileptic and non-antiepileptic drugs

Jerzy Majkowski1 and Philip N. Patsalos2

1 Center for Epilepsy Diagnosis and Treatment Foundation of Epileptology, Warsaw, Poland
2 Pharmocology and Therapeutics Unit, Department of Clinical and Experimental Epilepsy, Institute of Neurology,

London; The National Society for Epilepsy, Chalfont St Peter, UK

Introduction

Clinically important drug interactions occur essentially at two levels – at the phar- macokinetic level and at the pharmacodynamic level (Patsalos et al., 2002; Patsalos and Perucca, 2003a). By far the most important interactions are pharmacokinetic in nature and this is partly due to the fact that they are particularly prevalent in relation to antiepileptic drug (AED) use and also because they are more readily detected and quantitated. Whilst pharmacodynamic interactions are also of clini- cal significance they are less well documented and indeed difficult to quantitate. Pharmacokinetic interactions are associated with a change in blood concentration as a consequence of alterations in absorption, protein binding, distribution, metab- olism or elimination of a drug.

Since AEDs are frequently used for years, decades or even throughout a patient’s life, it is inevitable that drugs for the treatment of concurrent diseases will be co-prescribed. In this setting the potential for interactions is high and there are many such interactions that have been described (Patsalos and Perucca, 2003b). By far the most important and clinically significant interactions occur either as the conse- quence of hepatic enzyme inhibition or hepatic enzyme induction of cytochrome P450 (CYP) isoenzymes. Enzyme induction results in reduction in blood concen- trations and possibly a loss of an adequate therapeutic response whilst enzyme inhi- bition results in an elevation in blood concentrations and possibly toxicity.

The characterization of the isoenzymes involved in the metabolism of indivi- dual drugs during the past decade has greatly enhanced our ability to predict whether or not a metabolic interaction will occur and this is covered in more detail in Chapter 5. In clinical practice it is best to avoid prescribing drugs that have a high propensity to interact. However, it is sometimes necessary to co-prescribe such drugs. In this setting, it is advisable to undertake therapeutic monitoring and

139

140 Jerzy Majkowski and Philip N. Patsalos

to measure plasma drug concentrations, particularly after an interacting drug is introduced or withdrawn but also when a dosage change has occurred. Occasionally, it may be necessary to measure the free (pharmacologically active) concentration in plasma so as to aid a dose change and dose optimization. This would apply to drug interactions that involve the displacement of a drug that is highly protein bound ( 90%) from its plasma protein-binding site combined with an inhibition of its metabolism (e.g. phenytoin (PHT) and phenylbutazone).

In this chapter, clinically significant interactions between AEDs and non-AEDs are described. Interactions between AEDs and oral contraceptives, and between AEDs and psychoactive drugs are not described here as they are discussed in detail in Chapters 16 and 19, respectively. The interactions are presented in alphabetical order and are divided into those affected by a particular AED and those that affect the AED. However, in some instances we discuss interactions within a drug class. With regard to the new AEDs, because of the scarcity of available information, all interactions are highlighted regardless of whether or not a significant interaction was identified. In contrast, non-interaction drug combinations with the estab- lished AEDs are not reported.

It should be remembered that for interactions that are associated with an increase in clearance, a reduction in plasma concentrations and a reduction in area under the concentration versus time curve (AUC) values would probably require that a dose increase be undertaken so as to maintain an adequate therapeutic response. Conversely, interactions that are associated with a decrease in clearance, an increase in plasma concentrations and an increase in AUC values would proba- bly require that a dose reduction be undertaken so as to prevent drug toxicity. In both settings it is appropriate that patients are closely monitored and that plasma concentrations are measured.

Carbamazepine

Carbamazepine is extensively metabolized to carbamazepine-10,11-epoxide and then to carbamazepine-10,11-diol by CYPP450 enzymes. The formation of the epoxide is mediated primarily via CYP3A4, with some contribution by CYP2C8, whilst the metabolism of the epoxide is via the enzyme epoxide hydrolase. Plasma protein binding is 70%.

Interactions affecting carbamazepine

Antibiotics

As the macrolide antibiotics are metabolized by CYP3A4 they have the propensity to interact with carbamazepine. The interactions can be classified into three

141 Interaction between antiepileptic and non-antiepileptic drugs

groups according to their risk of interaction with AEDs (Periti et al., 1992). The first group comprises clarithromycin, erythromycin and troleandomycin and these drugs have a high propensity to inhibit the metabolism of carbamazepine (Babany et al., 1988). Typically, plasma carbamazepine concentrations increase by up to four-fold (Mesdjian et al., 1980; Majkowski, 1995).

The second group compromises flurithromycin, josamycin, midecamycin, mio- camycin and roxithromycin. These antibiotics are less potent CYP3A4 inhibitors and are usually associated with only a modest increase in plasma carbamazepine concen- trations (Albin et al., 1982; Vincon et al., 1987; Barzaghi et al., 1988; Couet et al., 1990; Levy, 1995). For example, the addition of clarithromycin (500mg/day) to carbamazepine, can result in an increase in plasma carbamazepine concentration of 30–50%, and a concurrent decrease of carbamazepine-epoxide concentrations (Albani et al., 1993; O’Connor and Fris, 1994; Yasui et al., 1997). In one study roxi- thromycin was not associated with any interaction (Saint-Salvi et al., 1987).

Azitromycin, dirithromycin, rokitamycin and spiramycin comprise the third group of macrolide antibiotics and these do not have any effect on CYP3A4 and therefore do not interact with carbamazepine (Periti et al., 1992; Principi and Esposito, 1999).

Antiviral agents

The antiviral agents delavirdine, indinavir and ritonavir are potent CYP3A4 inhibitors. Thus, their co-administration with carbamazepine can result in carba- mazepine toxicity. Indeed, there are reports of ritonavir causing a two- to three- fold increase in plasma carbamazepine concentration (Burman and Orr, 2000; Garcia et al., 2000; Kato et al., 2000; von Moltke et al., 2000; Mateu-de Antonio et al., 2001). Carbamazepine toxicity was similarly observed in a patient taking ritonavir and efavirenzin in combination (Burman and Orr, 2000).

Cimetidine

Cisplatin

During combination therapy with cimetidine, carbamazepine intoxication has been reported. However, the interaction does not occur consistently and is proba- bly of little clinical significance since the effect on carbamazepine is small (17% increase in plasma concentrations) and possibly transient (Dalton et al., 1986; Spina et al., 1996).

It has been reported that a young woman with epilepsy had seizures during anti- neoplastic therapy and that the seizures were the consequence of a reduced plasma carbamazepine concentration (Neef and de Voogd-van der Straaten, 1988). The mechanism of this interaction may be induction of metabolism or an increased volume of distribution.

142 Jerzy Majkowski and Philip N. Patsalos

Danazol

Diltiazem

Co-administration of danazol with carbamazepine results in a clinically significant increase (50–100%) in plasma carbamazepine concentrations. Moreover, danazol inhibits carbamazepine-epoxide elimination via an action on epoxide hydrolase and this too can contribute to the associated carbamazepine toxicity (Krämer et al., 1986; Zielinski et al., 1987; Hayden and Buchanan, 1991; Spina et al., 1996).

Plasma carbamazepine concentration can increase by up to 50% during combina- tion therapy with diltiazem (Brodie and MacPhee, 1986; Eimer and Carter, 1987; Bahls et al., 1991; Maoz et al., 1992). Diltiazem is metabolized to two metabolites (N-desmethyl-diltiazem and N,N-didesmethyl-diltiazem) and both are potent inhibitors of CYP3A4-mediated testosterone-6- -hydroxylation (11 and 200 times, respectively), compared to that of diltiazem. This would suggest that the major contribution to this interaction is the consequence of the two metabolites.

Fluconazole

Isoniazid

Many imidazole antifungals are potent inhibitors of CYP isoenzymes and these drugs commonly interact with carbamazepine. Fluconazole is a strong inhibitor of carbamazepine metabolism and a mean 120% increase in plasma carbamazepine concentration has been reported (Nair and Morris, 1999). During combination therapy carbamazepine intoxication can occur.

Isoniazid can inhibit the metabolism of carbamazepine, via an action on CYP3A4, resulting in elevated plasma concentrations and associated toxicity (Valsalan and Cooper, 1982). The clearance of carbamazepine can be decreased by up to 45% (Block, 1982; Wright et al., 1982; Spina et al., 1996).

Ketoconazole

Like fluconazole, ketoconazole is also a strong inhibitor of carbamazepine meta- bolism. The administration of ketoconazole to patients taking carbamazepine has been found to result in a significant (mean 29%) increase in plasma carba- mazepine concentrations and possibly in carbamazepine intoxication (Spina et al., 1997).

Metronidazole

The metabolism of carbamazepine can be inhibited by metronidazole. This results in an increase in plasma carbamazepine concentration and possible adverse events (Patterson, 1994).

143 Interaction between antiepileptic and non-antiepileptic drugs

Nicotinamide

Nicotinamide has been reported to increase plasma carbamazepine concentrations (Bourgeois et al., 1982).

Propoxyphene

Propoxyphene appears to reduce the activity of CYP3A4 and consequently inhibits the metabolism of carbamazepine (Abernethy et al., 1985). Thus during combina- tion therapy, plasma carbamazepine concentrations can increase by 45–77% (Dam and Christensen, 1977; Hansen et al., 1980). In addition, plasma carbamazepine- epoxide concentrations are significantly reduced (Bergendal et al., 1997).

In healthy volunteers, single doses of the antimalarial agent quinine have been reported to increase plasma carbamazepine concentrations (Amabeoku et al., 1993).

Ticlopidine

Verapamil

Quinine

Ticlopidine increased plasma carbamazepine concentrations which resulted in symptoms of neurological intoxication in a patient with epilepsy undergoing coro- nary stenting (Brown and Cooper, 1997).

Verapamil is extensively metabolized in the liver to several metabolites by numer- ous CYP isoenzymes: CYPCA4, CYP2C8 and CYP1A2 (Kroemer et al., 1993; Spina et al., 1996; Tracy et al., 1999). Verapamil inhibits CYP3A4 and it has been reported that plasma carbamazepine concentrations can increase by a mean of 46% result- ing in neurotoxicity (MacPhee et al., 1986). Interestingly, an increase in free carba- mazepine concentrations can also occur (mean rise of 33% in five of six patients) (MacPhee et al., 1986).

Interactions affected by carbamazepine

Carbamazepine is a potent hepatic enzyme inducer and, as well as inducing its own metabolism via an action on CYP3A4, it also induces the metabolism of many other drugs that are CYP3A4 substrates. There is also evidence to suggest that it induces CYP2C9, CYP2C19 and CYP1A2.

Antihypertensive drugs

Carbamazepine enhances the metabolic clearance of the -adrenoceptor blocking agents propranolol, metropronol and alprenolol, and the dihydropyridine calcium antagonists nimodipine, nifedipine, felodipine and nisoldipine as well as vera- pamil (Tartara et al., 1991; Flockart and Tanus-Santos, 2002). In relation to

144 Jerzy Majkowski and Philip N. Patsalos

nimodipine, nifedipine, felodipine and nisoldipine, the magnitude of the inter- action is so substantial (e.g. with nimodipine, plasma concentrations can decline seven-fold) that the usefulness of these agents in patients co-medicated with carbamazepine, and indeed other enzyme inducing AEDs, is questionable (Tartara et al., 1991).

Cyclosporin

Dicoumarol

Doxycycline

Fentanyl

Indinavir

The half-life of the antibiotic doxycycline is reduced two-fold when co-adminis- tered with carbamazepine (Penttila et al., 1974).

The anaesthetic fentanyl is primarily metabolized by CYP3A4 and its metabolism is enhanced by carbamazepine. Consequently, induction of anaesthesia requires substantially higher doses of fentanyl in patients taking carbamazepine (Tempelhoff et al., 1990; Feierman and Lasker, 1996).

In one case report, the addition of carbamazepine (200 mg/day) to indinavir treat- ment (800 mg t.i.d.) resulted in a reduction in plasma indinavir concentration by up to 16 times (Bonay et al., 1993). This interaction has recently been reported in another case report (Hugen et al., 2000).

Cyclosporin is metabolized by CYP3A4 and consequently during combination therapy with carbamazepine the metabolism of cyclosporin A is enhanced (Alvarez et al., 1991). Typically, plasma cyclosporin concentrations can be expected to decline by 65% (Cooney et al., 1995).

Carbamazepine reduces the anticoagulant effect of dicoumarol by enhancing its metabolism, possibly via an action on CYP2C9 (Freedman and Olatidoye, 1994). Overall, whenever there is a change in carbamazepine therapy (and indeed that of any other enzyme inducing AED; see sections later) it is advisable to monitor internationalized normalized ratio (INR) because all anticoagulants are associated with a narrow therapeutic ratio (Cropp and Bussey, 1997).

Itraconazole

Co-administration of carbamazepine with itraconazole results in a clinically sig- nificant reduction in plasma itraconazole concentrations (Bonay et al., 1993).

145 Interaction between antiepileptic and non-antiepileptic drugs

Methotrexate

The clearance of methotrexate is significantly enhanced by carbamazepine, result- ing in a clinically significant reduction in the therapeutic efficacy of methotrexate (Relling et al., 2000).

Phenprocoumon

Carbamazepine induces the metabolism of phenprocoumon and consequently reduces its anticoagulant effect (Schlienger et al., 2000). Overall, whenever there is a change in carbamazepine therapy (and indeed that of any other enzyme induc- ing AED; see sections later) it is advisable to monitor INR because all anticoagu- lants are associated with a narrow therapeutic ratio (Cropp and Bussey, 1997).

Rocuronium

Carbamazepine, through its induction of CYP3A4, CYP2C19 and CYP1A2, enhances the metabolism of rocuronium, and some other neuromuscular blocking agents, and therefore reduces their efficacy (Soriano et al., 2000).

Carbamazepine enhances the metabolic clearance of a variety of steroids including prednisolone, methylprednisolone and dexamethasone (Spina et al., 1996).

Teniposide

Vincristine

Warfarin

Steroids

Carbamazepine enhances the clearance of teniposide and consequently reduces the efficacy of teniposide (Relling et al., 2000).

During co-medication with carbamazepine, vincristine clearance was increased by 63% when compared to a control group (Villikka et al., 1999). As vincristine is metabolized in part by CYP3A4, induction of this isoenzyme by carbamazepine is the most likely explanation of this interaction.

The metabolism of warfarin is significantly enhanced by carbamazepine and this is associated with an increase in prothrombin time and a reduced anticoagulant effect (Schlienger et al., 2000). The interaction is mediated via an action on CYP2C9, although some induction of CYP3A4 may also occur (Rettie et al., 1992; Kunze et al., 1996). Overall, whenever there is a change in carbamazepine therapy (and indeed that of any other enzyme inducing AED; see sections later) it is advis- able to monitor INR because all anticoagulants are associated with a narrow therapeutic ratio (Cropp and Bussey, 1997).

146 Jerzy Majkowski and Philip N. Patsalos

Ethosuximide

Ethosuximide is eliminated primarily by metabolism with 30–60% of an adminis- tered dose recovered in urine. Metabolism is primarily mediated by CYP3A and to a lesser extent by CYP2E and CYP2B/C. Approximately 20% of an administered dose is excreted unchanged in urine. Ethosuximide is not bound to plasma proteins.

Interactions affecting ethosuximide

Isoniazid

Isoniazid may increase plasma ethosuximide concentrations resulting in clinical signs of intoxication (Van Wieringen and Vrijlandt, 1983).

Rifampicin

Interactions affected by ethosuximide

There are no clinical data to suggest ethosuximide induces or inhibits the meta- bolism of other non-AEDs.

Felbamate

Approximately 50% of an administered dose is metabolized to form two hydroxy- lated metabolites. Felbamate is a substrate of CYP3A4 and CYP2E1. Approximately 40–50% of an absorbed dose is excreted unchanged in urine. Plasma protein binding is 23%.

As a new AED, knowledge of the interaction profile of felbamate with non-AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting felbamate

Erythromycin, a potent CYP3A4 inhibitor is without effect on the metabolism of felbamate and indeed plasma felbamate concentrations are not significantly affected during combination therapy (Glue et al., 1997).

Interactions affected by felbamate

To date there are no clinical data to suggest that felbamate induces or inhibits the metabolism of other non-AEDs. However, interactions may conceivably occur with drugs that are substrates for the same isoenzymes as occur with felbamate.

In adult healthy volunteers, rifampicin has been observed to decrease plasma etho- suximide concentrations by induction of its metabolism (Bachmann and Jauregui, 1993).

147 Interaction between antiepileptic and non-antiepileptic drugs

Gabapentin

Gabapentin is not metabolized and is exclusively eliminated as unchanged gaba- pentin in urine. It is not protein bound. Consequently, gabapentin should have little propensity to interact with other drugs and indeed this is the case.

As a new AED, knowledge of the interaction profile of gabapentin with non- AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting gabapentin

Antacids

Antacids containing aluminium and magnesium hydroxide reduce the absorption of gabapentin by approximately 15% (Busch et al., 1992). This interaction is of little clinical significance.

Cimetidine

Interactions affected by gabapentin

To date there are no clinical data to suggest that gabapentin affects the metabolism of other non-AEDs.

Lamotrigine

Lamotrigine undergoes extensive metabolism via glucuronidation and the pri- mary metabolite is N-2 glucuronide (71% of dose). Glucuronidation is a major conjugation reaction that is catalyzed by a number of different isoforms of uridine 5 diphosphate (UDP)-glucuronosyl transferase (UGT). The N-2 glucuronidation of lamotrigine is catalyzed by UGT1A4. Plasma protein binding is 50%.

As a new AED, knowledge of the interaction profile of lamotrigine with non- AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting lamotrigine

Acetaminophen (paracetamol)

Since acetaminophen is excreted by glucuronidation, as is indeed lamotrigine, it was anticipated that an interaction between the two drugs would occur. In a healthy volunteer study acetaminophen enhanced the clearance of lamotrigine (15%) and decreased AUC (20%) and half-life values (15%) (Depot et al., 1990).

Cimetidine appears to decrease plasma gabapentin concentrations by approxi- mately 15%. The mechanism appears to be renal in nature and is considered not to be of clinical significance.

148 Jerzy Majkowski and Philip N. Patsalos

Bupropion

Cimetidine

Rifampicin

In healthy volunteers, bupropion was observed not to interact with lamotrigine (Odishaw and Chen, 2000).

In healthy volunteers, cimetidine was observed not to interact with lamotrigine (Ebert et al., 2000).

Rifampicin is a potent inducer of CYPP450 and of the UGT enzyme system. In adult healthy volunteers, rifampicin enhanced the clearance of lamotrigine and the amount of lamotrigine excreted as a glucuronide was increased by 36% when com- pared to placebo (Ebert et al., 2000). The corresponding half-life and AUC values for lamotrigine were significantly reduced (60% and 56%, respectively) compared to placebo.

Interactions affected by lamotrigine

To date there are no clinical data to suggest that lamotrigine affects the metabolism of other non-AEDs.

Levetiracetam

Levetiracetam undergoes minimal metabolism with approximately 30% being metabolized non-hepatically in blood to an inactive metabolite. Furthermore, the elimination of levetiracetam is predominantly renal with approximately 70% of a levetiracetam dose excreted unchanged in urine. It is not protein bound, conse- quently levetiracetam should have little propensity to interact with other drugs and indeed this is the case.

As a new AED, knowledge of the interaction profile of levetiracetam with non- AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting levetiracetam

Digoxin

No clinically relevant effect of digoxin on the pharmacokinetics of levetiracetam was observed in a study of 11 healthy adult volunteers (Levy et al., 2000).

Probenecid

The pharmacokinetics of levetiracetam were unaffected by co-administration of probenecid (Patsalos, 2000). However, the plasma concentration of the primary

149 Interaction between antiepileptic and non-antiepileptic drugs

Warfarin

pharmacologically inactive metabolite of levetiracetam (ucb LO 57) was increased 2.5-fold. The clinical significance of the latter effect is unknown.

The co-administration of warfarin with levetiracetam did not result in any significant change in the pharmacokinetics of levetiracetam (Ragueneau-Majlessi et al., 2001).

Interactions affected by levetiracetam

Digoxin

Warfarin

Plasma digoxin concentrations are not significantly affected by levetiracetam (Levy et al., 2000).

Levetiracetam does not alter the anticoagulant effect or the pharmacokinetics of warfarin (Ragueneau-Majlessi et al., 2001).

Oxcarbazepine

Although oxcarbazepine is clinically related to carbamazepine, its pharmaco- kinetic and interaction profiles are substantially different. Oxcarbazepine under- goes rapid and extensive metabolism to its pharmacologically active metabolite, 10-hydroxycarbazepine, which is subsequently eliminated by glucuronidation or undergoes hydroxylation to form a dihydrodiol metabolite. Only the latter reac- tion depends on CYP isoenzymes (Patsalos and Duncan, 1993; Baruzzi et al., 1994; Tecoma, 1999). Oxcarbazepine can induce CYP3A4 and CYP3A5 activities and inhibit CYP2C19. Plasma protein binding is 40%.

As a new AED, knowledge of the interaction profile of oxcarbazepine with non- AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting oxcarbazepine

Cimetidine

Dextropropoxyphene

In a study of eight patients taking oxcarbazepine, dextropropoxyphene adminis- tration was without effect on steady-state plasma 10-hydroxycarbazepine concen- trations (Mogensen et al., 1992).

Co-administration of cimetidine and oxcarbazepine in healthy volunteers was not associated with any significant change in the pharmacokinetics of oxcarbazepine (Keränen et al., 1992b).

150 Jerzy Majkowski and Philip N. Patsalos

Erythromycin

Co-administration of erythromycin and oxcarbazepine to eight healthy volunteers over a 1-week period did not result in any major change in oxcarbazepine or 10-hydroxycarbazepine pharmacokinetic parameters (Keränen et al., 1992a).

Verapamil

Interactions affected by oxcarbazepine

Felodipine

Warfarin

The potential interaction between verapamil and oxcarbazepine was investigated in 10 healthy volunteers (Krämer et al., 1991). A 20% decrease in 10-hydroxycarbazepine AUC values was observed and the investigators concluded that this interaction may be of clinical relevance in some patients.

The potential interaction of oxcarbazepine and felodipine, a calcium antagonist, was studied in eight healthy volunteers (Zaccara et al., 1993). It was observed that the bioavailability of felodipine was reduced by 28% but the clinical relevance of this observation is as yet not clear.

The influence of oxcarbazepine on the anticoagulant effect of warfarin was studied in 10 adult healthy volunteers (Krämer et al., 1992). Oxcarbazepine was without any significant effect as measured by prothrombin time.

Phenobarbital

Phenobarbital is extensively metabolized to two major metabolites, p-hydroxy- phenobarbital, and 9-D-glucopyranosylphenobarbital. CYP2C9 plays a major role in the metabolism of phenobarbital with minor metabolism by CYP2C19 and CYP2E1. Phenobarbital is a potent enzyme inducer. Plasma protein binding is 50%.

Interactions affecting phenobarbital

Activated charcoal

Co-administration of activated charcoal with phenobarbital results in reduction of phenobarbital absorption (Neuvonen and Elonen, 1980). This interaction is used clinically to treat those patients that have overdosed with phenobarbital. After phenobarbital intravenous administration, repeated administration of charcoal over 2–4 days results in an increased phenobarbital clearance (60–270%) whilst half-life values are decreased 2.5–8-fold (Berg et al., 1982, 1987; Frenia et al., 1996).

151 Interaction between antiepileptic and non-antiepileptic drugs

This effect is considered to be the consequence, in part, of impaired phenobarbital enterohepatic recirculation (Wakabayashi et al., 1994).

Chloramphenicol

Ethanol

Co-administration of chloramphenicol and phenobarbital results in a significant decrease (40%) in phenobarbital clearance (Koup et al., 1978).

During chronic use, alcohol enhances the metabolism of phenobarbital (Sands et al., 1993). In contrast, the acute effect of alcohol is to inhibit the metabolism of phenobarbital (Forney and Hughes, 1964).

Interactions affected by phenobarbital

Cefotaxime

Cimetidine

Cyclosporin

Dexamethasone

The administration of dexamethasone intravenously to asthmatic patients receiv- ing phenobarbital was associated with significantly shorter dexamethasone half- life values (45%) and increased clearance (87%), when compared to values before phenobarbital administration (Brooks et al., 1972).

Felodipine

High dose phenobarbital administration to children along with -lactam anti- biotics results in toxic exanthematous skin reactions in about 50% of cases (Harder et al., 1990). This potentiating of antibiotic-related allergic features is particularly prevalent in children receiving phenobarbital and cefotaxime in combination. This interaction may be pharmacodynamic in nature.

Phenobarbital enhances the metabolism and clearance (15%) of cimetidine (Somogyi et al., 1981).

Phenobarbital can significantly enhance the metabolism of cyclosporin in a dose- dependent manner (Carstensen et al., 1986; Nishioka et al., 1990). Withdrawal of phenobarbital from a paediatric renal transplant patient resulted in a 70% reduc- tion in cyclosporin clearance (Burckart et al., 1984).

The clearance of felodipine can be significantly enhanced ( 9-fold) by pheno- barbital co-administration (Capewell et al., 1988).

152 Jerzy Majkowski and Philip N. Patsalos

Fentanyl

Folic acid

Ifosfamide

Itraconazole

Phenobarbital decreases plasma itraconazole concentrations (Bonay et al., 1993).

Metronidazole

Phenobarbital enhances the metabolism of metronidazole resulting in significant decreases in metronidazole half-life (23%) and AUC (30%) values (Eradiri et al., 1988). This drug interaction is associated with clinical failure of metronidazole treatment in women with vaginal trichomoniasis and gardiasis (Mead et al., 1982; Gupte, 1983).

Methylprednisolone

The co-administration of methylprednisolone with phenobarbital results in phar- macokinetic changes to methylprednisolone, which are similar in magnitude to that described for dexamethasone (Stjernholm and Katz, 1975; Wassner et al., 1976).

Nifedipine

Nimodipine

Phenobarbital enhances the metabolism of fentanyl and decreases its plasma con- centration (Tempelhoff et al., 1990).

As phenobarbital is a potent hepatic enzyme inducer, it enhances the metabolism of folic acid and, typically, folate concentrations can be reduced by 24% during long-term treatment (Reynolds, 1974). Consequently, folic acid supplementation is mandatory if one is to avoid adverse foetal outcome (e.g. neural tube defects) that is associated with low plasma folate concentrations (Kishi et al., 1997).

A reversible toxic encephalopathy was reported in a girl with epilepsy who was tak- ing phenobarbital. The symptoms presented after a single dose of ifosfamide/ mesna and the authors concluded that it was the consequence of an interaction with phenobarbital (Ghosn et al., 1988).

The clearance of nifedipine can be significantly (270%) enhanced by phenobar- bital co-administration (Schellens et al., 1989).

The clearance of nimodipine can be significantly enhanced (nine-fold) by pheno- barbital co-administration (Tartara et al., 1991). Thus clinically relevant reduced nimodipine efficacy can be observed.

153 Interaction between antiepileptic and non-antiepileptic drugs

Prednisolone

Phenobarbital enhances the metabolism of prednisolone resulting in significantly shorter (32%) prednisolone half-life values and increased (44%) clearance values. Co-administration of prednisolone with phenobarbital in patients with rheuma- toid arthritis has resulted in shorter (25%) half-life values and marked worsening of clinical symptoms (Brooks, S., et al., 1972; Brooks, P., et al., 1976).

Teniposide

Theophylline

Tirilazad

Phenobarbital enhances (34%) the clearance of theophylline in older children and adults (Landay et al., 1978; Saccar et al., 1985; Yazdani et al., 1987). However, in premature neonates this interaction does not occur (Kandrotas et al., 1990).

Phenobarbital significantly decreases plasma concentrations of tirilazad by 50–69% (Fleishaker et al., 1996).

Phenobarbital co-administered with teniposide results in a two- to three-fold increase in clearance of teniposide (Baker et al., 1992). The resultant reduced effi- cacy of teniposide (Relling et al., 1994) is the consequence of induction of CYP3A4 and possibly CYP3A5 (Relling et al., 2000).

Tolbutamide

Phenobarbital increases the free fraction of tolbutamide by displacing it from plasma protein-binding sites (Fernandez et al., 1985). The clinical significance of this interaction is not established.

Verapamil

Warfarin

The clearance of orally ingested verapamil can be enhanced five-fold during combi- nation therapy with phenobarbital. When verapamil was administered intravenously, the clearance of verapamil was enhanced two-fold (Rutledge et al., 1988).

Co-administration of warfarin and phenobarbital results in a significant increase ( 50%) in warfarin clearance and a decrease in its half-life ( 40%) (Orme and Breckenridge, 1976). These changes are accompanied by a 25% reduction in pro- thrombin time which may persist for 3–4 weeks after phenobarbital discontinua- tion (Udall, 1975; Cropp and Bussey, 1997). This prolonged effect requires that patients are closely monitored.

154 Jerzy Majkowski and Philip N. Patsalos

Phenytoin

PHT is eliminated almost entirely by metabolic transformation. Metabolism is via the isoenzymes CYP2C9 and CYP2C19. PHT is an enzyme inducer (CYP2A, CYP2C and CYP3A) and has a high propensity to interact with other drugs. Plasma protein binding is 92%.

Interactions affecting phenytoin

Activated charcoal

Acyclovir

Co-administration of activated charcoal with PHT results in reduction of PHT absorption (Welling, 1984). This interaction is used clinically to treat those patients that have overdosed with PHT.

Co-administration of PHT and acyclovir may result in a reduction in PHT plasma concentrations (Permeggiani et al., 1995).

Amiodarone

Antacids

Amiodarone is potent inhibitor of CYP2C9. In healthy subjects, amiodarone has been observed to increase the half-life of PHT several fold (Nolan et al., 1989), whilst in patients, plasma PHT concentrations have been increased two- to three- fold (McGovern et al., 1984), resulting in possible PHT intoxication.

The gastrointestinal absorption of PHT may be reduced by co-ingestion with antacids such as aluminium or magnesium hydroxides and calcium bicarbonate. This interaction is avoided if the ingestion of PHT and the antacid is separated by a few hours. Sucralfate, a complex of aluminium hydroxide and sulphated sucrose, which has minimal antacid properties but acts by protecting the mucosa from acid–pepsin attack, can similarly impede the absorption of PHT.

Antineoplastic agents

It has been reported that antineoplastic agents such as adriamycin, bleomycin, cis- platin or vinblastine can decrease plasma PHT concentrations (Bollini et al., 1983; Sylvester et al., 1984; Neef and de Voogd-van der Straaten, 1988). It has been reported that a young woman with epilepsy had seizures during antineoplastic therapy and that the seizures were the consequence of a reduced (37%) plasma PHT concentration (Neef and de Voogd-van der Straaten, 1988). The mechanism of this interaction may be induction of metabolism or an increased volume of

155 Interaction between antiepileptic and non-antiepileptic drugs

distribution. In contrast, tamoxifen has been associated with increased plasma PHT concentrations and signs of PHT toxicity (Rabinowicz et al., 1995).

Bishydroxycoumarin

Plasma bishydroxycoumarin concentrations can increase in some patients co-administered with PHT (Skovsted et al., 1974).

Calcium channel blockers

Whilst verapamil and nifedipine have little or no effect on plasma PHT concentra- tions, diltiazem may cause an elevation and cause PHT intoxication in some patients (Bahls et al., 1991).

Chloramphenicol

Whilst chloramphenicol may cause only modest elevations of plasma PHT con- centrations in some patients, it may produce marked elevations in others (Koup, 1978; Nation et al., 1990).

Cimetidine

Disulfiram

Ethanol

Fluconazole

Cimetidine inhibits the metabolism of PHT thereby increasing plasma PHT con- centrations and this may result in clinical intoxication (Salem et al., 1983; Phillips and Hansky, 1984; Levine et al., 1985).

Disulfiram inhibits the metabolism of PHT and increases its plasma concentra- tion. This can result in signs of PHT intoxication in the majority of patients (Olesen, 1967; Levy, 1995). In healthy volunteers, disulfiram was shown to reduce PHT clearance by 30% (Svendsen et al., 1976).

Chronic use of alcohol decreases plasma PHT concentrations, probably as a conse- quence of enzyme induction (Sandor et al., 1981), whereas occasional moderate or heavy alcohol consumption can result in an increase in PHT plasma concentration and this can result in PHT toxicity (Kutt, 1984).

Fluconazole inhibits both CYP2C9 and CYP2C19 activities and consequently would be expected to inhibit the metabolism of PHT. Indeed there are several case reports, both of healthy volunteers and patients, that describe significant increases in plasma PHT concentrations and toxicity during combination therapy with

156 Jerzy Majkowski and Philip N. Patsalos

fluconazole (Howit and Oziemski, 1989; Mitchell and Holland, 1989; Blum et al., 1990; Lazar and Wilner, 1990; Cadle et al., 1994; Levy, 1995).

PHT metabolism is inhibited by isoniazid. In patients taking isoniazid and PHT, significant PHT accumulation with consequent intoxication has been reported in 10–15% of patients (de Wolff et al., 1983; Witmer and Ritschel, 1984). This inter- action would be particularly prevalent in patients that exhibit slow acetylation. In the Groote Schuur Hospital, South Africa, where 74% of patients with epilepsy are taking PHT, it has been observed that 12% of patients have plasma PHT con- centrations in the toxic range because they are taking antituberculous medication of which the primary drug is isoniazid (Walubo and Aboo, 1995).

Miconazole

Omeprazole

In healthy subjects, the co-administration of omeprazole with PHT has been shown to result in a significant increase in PHT plasma concentrations (Gugler and Jensen, 1985; Prichard et al., 1987).

Phenylbutazone

When PHT and phenylbutazone are co-administered, the half-life of PHT is sig- nificantly increased and this may be accompanied by clinical intoxication (Levy, 1995). The mechanism of this interaction involves the displacement of PHT from plasma albumin binding sites and a concurrent inhibition of PHT metabolism (Skovsted et al., 1974). Thus, the interaction can present as an increase in the free pharmacologically active concentration of PHT in the absence of a change in the total PHT concentration. Dosage adjustment may be needed and should be based on the measurement of free PHT concentrations.

Propoxyphene

Isoniazid

Rifampin

Propoxyphene inhibits the metabolism of PHT via an action on CYP2C9 (Levy, 1995). The consequent increase in plasma PHT concentrations can result in intox- ication (Dam et al., 1980; Kutt, 1984).

Rifampin may significantly increase the clearance of PHT by as much as two-fold and consequently decrease plasma PHT concentrations (Kay et al., 1985). It should

As miconazole is an inhibitor of CYP2C9, it inhibits the metabolism of PHT resulting in elevated plasma PHT concentrations and symptoms of PHT toxicity (Rolan et al., 1983; Levy, 1995).

157 Interaction between antiepileptic and non-antiepileptic drugs

be noted that rifampin minimizes the inhibitory effect of isoniazid on PHT, even in patients that are slow acetylators.

Salicylates

Sulfonamides

Numerous bacteriostatic sulpfonamides (sufadiazine, sulfamethiazole, sulfa- methoxazole and sulfaphenazole) are inhibitors of PHT metabolism and can decrease its clearance and prolong its half-life (Molhom Hansen et al., 1979). Sulfaphenazole is particularly potent in this regard.

Ticlopidine

Tolbutamide

Tolbutamide displaces PHT from its plasma protein-binding sites and this can result in lower plasma PHT concentrations (Wesseling and Molsthurkow, 1975). However, as free PHT concentrations are unaffected, this interaction is not of clin- ical significance.

Interactions affected by phenytoin

Acetaminophen (paracetamol)

PHT enhances the metabolism of acetaminophen and reduces its plasma concen- tration (Nation et al., 1990).

Chloramphenicol

During combination therapy with PHT and chloramphenicol, plasma chloram- phenicol concentrations have been observed to decline significantly (Krasinski et al., 1982).

Cyclophosphamide

It has been reported that PHT increases the clearance of both the R– and S-isomers of cyclophosphamide by 100% and 150%, respectively (Williams, M., et al., 1999).

Although salicylates can displace PHT from its plasma protein-binding site so that the unbound fraction of PHT is increased from 10% to 16%, the concurrent increase in PHT clearance makes this interaction of little clinical significance for the majority of patients (Fraser et al., 1980).

Ticlopidine is a potent CYP2C19 inhibitor. Consequently, when co-administered with PHT the clearance of PHT is decreased and PHT intoxication can occur (Privitera and Welty, 1996; Klaassen, 1998; Denahue et al., 1999).

158 Jerzy Majkowski and Philip N. Patsalos

Cyclosporin

Dexamethasone

The metabolism of dexamethasone is substantially enhanced by PHT, probably via enzyme induction. In one study the elimination half-life of dexamethasone was reduced from 3.5 to 1.8 h (Chalk et al., 1984). In another patient study, plasma dexamethasone concentrations were reduced by 50% (Wong et al., 1985).

Dicoumarol

Digitoxin

Digoxin

PHT significantly enhances the metabolism of cyclosporin and reduces its maxi- mal plasma concentration as well as AUC and half-life values, resulting in a reduc- tion of the clinical efficacy of cyclosporin (Freeman et al., 1984).

PHT can decrease blood dicoumarol concentrations, probably via induction of metabolism (Hansen, J., et al., 1971). Overall, whenever there is a change in PHT therapy (and indeed that of any other enzyme-inducing AED) it is advisable to monitor INR because all anticoagulants are associated with a narrow therapeutic ratio (Cropp and Bussey, 1997).

In some patients, PHT is associated with a modest reduction in plasma digitoxin concentrations (Solomon et al., 1971).

In healthy volunteers, PHT increased digoxin clearance by 27% and this was asso- ciated with a significant decrease in its half-life (Rameis, 1985).

Disopyramide

Although PHT enhances the metabolism of disopyramide and therefore reduces plasma disopyramide concentrations, the fact that plasma concentration of its pharmacologically active metabolite is also increased, may not necessarily result in a loss of effectiveness (Aitio et al., 1981).

Doxycycline

Fluconazole

PHT enhances the metabolism of doxycycline and decreases its plasma concentra- tion (Neuvonen et al., 1975).

Plasma fluconazole concentrations are substantially reduced during co-medication with PHT probably via induction of fluconazole metabolism (Tucker et al., 1992).

159 Interaction between antiepileptic and non-antiepileptic drugs

Folic acid

Furosemide

Itraconazole

In healthy volunteers, PHT has been observed to decrease itraconazole AUC values by 93% and half-life values by 83% (Tucker et al., 1992; Ducharme et al., 1995).

Ketoconazole

Plasma ketoconazole concentrations are substantially reduced during co-medication with PHT probably via induction of ketoconazole metabolism (Tucker et al., 1992).

Methadone

Meperidine (pethidine)

During combination therapy with PHT and meperidine, the half-life of meperi- dine was reduced by 30% and the AUC of its primary metabolite was increased (Pond and Kretschzmar, 1981).

Methotrexate

PHT increases the clearance of methotrexate. This interaction has been reported to compromise the efficacy of methotrexate in the treatment of lymphoblastic leukemia in children (Relling et al., 2000).

Mexiletine

As PHT is a potent hepatic enzyme inducer, it enhances the metabolism of folic acid (Lewis et al., 1995). Consequently, folic acid supplementation is mandatory if one is to avoid adverse fetal outcomes (e.g. neural tube defects) that are associated with low plasma folate concentrations (Kishi et al., 1997).

The diuretic effect of furosemide is reduced when PHT is co-administered. The interaction is primarily due to a reduction in furosemide absorption from the alimentary tract but a pharmacodynamic interaction in the kidneys may also occur (Ahmad, 1974).

Plasma methadone concentrations were decreased by 50% during co-medication with PHT (Tong et al., 1981). In this setting patients may experience symptoms of methadone withdrawal.

In healthy volunteers, PHT increased the metabolism of mexiletine and reduced the AUC of mexiletine by 55% (Begg et al., 1982).

160 Jerzy Majkowski and Philip N. Patsalos

Misonidazole

PHT enhances the metabolism of misonidazole and reduces its half-life. Therefore, lower plasma misonidazole concentrations are achieved which may serve to reduce the toxicity of misonidazole whilst not reducing its effectiveness as a therapeutic adjunct in radiation therapy (Williams, K., et al., 1983).

Nisoldipine

Praziquantel

PHT induces the metabolism of praziquantel, a drug used to treat neurocysticer- cosis. The interaction results in a two- to three-fold reduction in plasma prazi- quantel concentrations (Bittencourt et al., 1992).

Prednisolone

Quinidine

PHT enhances the clearance of prednisolone and consequently reduces the effec- tiveness of this corticosteroid (Nation et al., 1990).

PHT decreases the half-life of quinidine by 50% (Nation et al., 1990).

In patients with epilepsy, PHT has been observed to significantly enhance the metabolism of nisoldipine with a mean reduction in nisoldipine AUC values of 90% (Nation et al., 1990; Michelucci et al., 1996).

Rocuronium

In patients taking PHT chronically, muscle relaxation after rocuronium adminis- tration was only achieved at higher doses of rocuronium and also it was necessary to administer rocuronium more frequently (Soriano et al., 2000). This effect is considered to be the consequence of enzyme induction.

Teniposide

Theophylline

In healthy volunteers, PHT administration (300–400 mg/day) was associated with an enhanced clearance and a 40% reduction in theophylline half-life values after intravenously administered theophylline (Jonkman and Upton, 1984; Sklar and Wagner, 1985).

PHT induces the metabolism and enhances the clearance of teniposide and this interaction is of clinical significance (Baker et al., 1992; Relling et al., 2000).

161 Interaction between antiepileptic and non-antiepileptic drugs

Tirilazad

Vecuronium

Warfarin

In patients taking PHT chronically, muscle relaxation after vecuronium adminis- tration was only achieved at higher doses of vecuronium and it was necessary to administer vecuronium more frequently (Platt and Thackery, 1993). This effect is considered to be the consequence of enzyme induction.

The effect of PHT on warfarin is variable. Overall, the observed interaction involves a reduction in warfarin blood concentrations, via hepatic induction of warfarin metabolism. However, an increase in anticoagulant effect has been reported in some patients (Nappi, 1979). Overall, whenever there is a change in PHT therapy (and indeed that of any other enzyme-inducing AED) it is advisable to monitor INR because all anticoagulants are associated with a narrow therapeutic ratio (Cropp and Bussey, 1997).

In healthy volunteers PHT enhanced the clearance of tirilazad by 92% (Fleishaker et al., 1998). The mechanism of this interaction is probably enzyme induction.

Primidone

Primidone is metabolized to two pharmacologically active metabolites, namely phenylethylmalonamide and phenobarbital. Phenobarbital, the primary metabo- lite, subsequently undergoes oxidation to form p-hydroxyphenobarbital. Primi- done, via its metabolite phenobarbital, is an enzyme inducer. Plasma protein binding of primidone is 15%. The interactions of primidone are primarily those involving phenobarbital.

Interactions affecting primidone

Acetazolamide

Isoniazid

Acetazolamide may impair the absorption of primidone (Syverson et al., 1977). Similar effects can be expected with other drugs that alter gastric pH (antacids) or motility.

Isoniazid decreases the conversion of primidone to phenobarbital resulting in increased plasma primidone concentrations. The interaction is considered to be a consequence of CYP inhibition (Sutton and Kupferberg, 1975).

162 Jerzy Majkowski and Philip N. Patsalos

Nicotinamide

Nicotinamide decrease the conversion of primidone to phenobarbital resulting in increased plasma primidone concentrations. The interaction is considered to be a consequence of CYP inhibition (Bourgeois et al., 1982).

Interactions affected by primidone

Folic acid

The absorption of folic acid appears to be hindered by primidone (Reynolds et al., 1972).

Tiagabine

Interactions affecting tiagabine

Cimetidine

Erythromycin

The effect of erythromycin on the pharmacokinetics of tiagabine in 13 healthy vol- unteers was investigated and it was observed that erythromycin was without effect (Thompson et al., 1997).

Other drugs

Tiagabine is extensively metabolized by CYP3A and is also extensively protein bound (98%).

As a new AED, knowledge of the interaction profile of tiagabine with non-AEDs is rather limited. Only interactions with specific drugs have been investigated.

A multiple-dose crossover study of the effect of cimetidine on the pharmaco- kinetics of tiagabine showed a small ( 5%) increase in tiagabine plasma concen- trations (Mengel et al., 1995). This is not considered to be of clinical significance.

The effects of numerous other drugs on the pharmacokinetics of tiagabine have been investigated. Triazolam (Richens et al., 1998), ethanol (Kastberg et al., 1998), theophylline (Mengel et al., 1995), digoxin (Snel et al., 1998) or warfarin (Mengel et al., 1995) showed no effect.

In vitro studies have shown that tiagabine is displaced from its protein- binding sites by the analgesics naproxen and salicylate (Brodie, 1995; Gustavson and Mengel 1995; Patsalos et al., 2002). The clinical significance of these interac- tions is not known.

163 Interaction between antiepileptic and non-antiepileptic drugs

Interactions affected by tiagabine

Digoxin

Ethanol

Tiagabine was without effect on the pharmacokinetics of digoxin in a series of 13 healthy volunteers (Snel et al., 1998).

Tiagabine was without effect on the pharmacokinetics of ethanol in a series of 20 healthy volunteers (Mengel et al., 1995; Kastberg et al., 1998).

Theophylline

Triazolam

Warfarin

Tiagabine was without effect on the pharmacokinetics of theophylline in healthy volunteers (Mengel et al., 1995).

Tiagabine was without effect on the pharmacokinetics of triazolam in healthy vol- unteers (Mengel et al., 1995).

The pharmacokinetics of warfarin are unaffected by warfarin (Mengel et al., 1995). Tiagabine does not appear to displace other highly protein-bound drugs, such as amitriptyline, tolbutamide and warfarin, from their plasma protein-binding sites

(Brodie, 1995).

Topiramate

In the absence of hepatic enzyme inducers, only 40% of topiramate is metabolized, whilst in the presence of inducers this value is doubled. Although the specific CYP isoenzymes responsible for the metabolism of topiramate have not been identified, it is evident that isoenzymes induced by carbamazepine (CYP3A4) and PHT (CYP2C9 and CYP2C19) play a major role. Elimination occurs both via hepatic metabolism and renal excretion. Plasma protein binding is 10%.

As a new AED, knowledge of the interaction profile of topiramate with non-AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting topiramate

There have been no clinical studies to investigate the effect of non-AEDs on the pharmacokinetics of topiramate.

164 Jerzy Majkowski and Philip N. Patsalos

Interactions affected by topiramate

Digoxin

In a study of 12 healthy volunteers the pharmacokinetics of a single oral dose of digoxin were compared during monotherapy and in combination with topiramate (Liao and Palmer, 1993). Digoxin plasma concentrations were reduced by 16% and clearance was increased by 13% during topiramate administration when com- pared with administration of digoxin alone.

Valproic acid

The metabolism of valproic acid is both extensive and complex in that it involves multiple metabolic pathways, including – and -oxidation, CYP2A6, CYP2C9, CYP2C19 and CYP2B6 isoenzymes and glucuronidation by UGT. To date, in excess of 25 metabolites of valproic acid have been identified. Valproic acid is 92% protein bound.

Interactions affecting valproic acid

Cholestyramine

There is a report suggesting that cholestyramine may decrease valproic acid plasma concentrations during combination therapy (Malloy et al., 1996).

Cimetidine

Cisplatin

Doxorubicin

Ibuprofen

Doxorubicin (adryamicin) can decrease plasma valproic acid concentrations (Neef and de Voogd-van der Straaten, 1988).

In vitro data show that ibuprofen can significantly displace valproic acid from its plasma protein-binding sites and increase the free concentration of valproic acid

Cimetidine inhibits the metabolism of valproic acid and increases its plasma con- centrations (Webster et al., 1984).

It has been reported that a young woman with epilepsy presented with seizures during antineoplastic therapy and that the seizures were the consequence of a reduced valproic acid plasma concentration (Neef and de Voogd-van der Straaten, 1988). The mechanism of this interaction may be induction of metabolism or an increased volume of distribution.

165 Interaction between antiepileptic and non-antiepileptic drugs

Isoniazid

(Dasgupta and Volk, 1996). The clinical significance of this interaction is not known.

Isoniazid may increase valproic acid plasma concentrations resulting in clinically significant intoxication (Jonville et al., 1991).

Ketoconazole

In vitro data show that ketoconazole can significantly displace valproic acid from its plasma protein-binding sites and increase the free concentration of valproic acid (Dasgupta and Luke, 1997). The clinical significance of this interaction is not known.

Mefenamic acid

In vitro data show that mefenamic acid can significantly displace valproic acid from its plasma protein-binding sites and increase the free concentration of val- proic acid (Dasgupta and Volk, 1996). The clinical significance of this interaction is not known.

Methotrexate

Methotrexate significantly decreases (75%) plasma valproic acid concentrations (Schroder and Ostergaard, 1999).

In vitro data show that naproxen can significantly displace valproic acid from its plasma protein-binding sites and increase the free concentration of valproic acid (Dasgupta and Volk, 1996). The clinical significance of this interaction is not known.

Rifampicin

Salicylic acid

Salicylic acid displaces valproic acid from its protein-binding sites on albumin and consequently higher unbound concentrations occur (Fleitman et al., 1980; Abbott et al., 1986). In addition, salicylic acid inhibits the metabolism of valproic acid (Schobben et al., 1978; Goulden et al., 1987). The combination of these two effects can result in elevated valproic acid plasma concentrations and consequent toxicity.

Naproxen

Rifampicin enhances the metabolism of valproic acid and its clearance can increase by 40%.

166 Jerzy Majkowski and Philip N. Patsalos

Tolbutamide

Tolmetin

Tolbutamide can displace valproic acid from its plasma protein-binding sites and increase the free concentration of valproic acid (Fernandez et al., 1985). The clinical significance of this interaction is not known.

In vitro data show that tolmetin can significantly displace valproic acid from its plasma protein-binding sites and increase the free concentration of valproic acid (Dasgupta and Volk, 1996). The clinical significance of this interaction is not known.

Interactions affected by valproic acid

Warfarin

Valproic acid can displace warfarin from its plasma protein-binding sites but this is not considered to be of clinical significance (Panjehshahin et al., 1991).

Zidovudine

Vigabatrin

Vigabatrin is not metabolized and is exclusively eliminated as unchanged viga- batrin in urine. It is not protein bound. Consequently vigabatrin should have little propensity to interact with other drugs and indeed this is the case.

As a new AED, knowledge of the interaction profile of vigabatrin with non-AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting vigabatrin

To date there have been no reports of non-AEDs affecting the pharmacokinetics of vigabatrin.

Interactions affected by vigabatrin

To date there have been no reports of vigabatrin affecting the pharmacokinetics of non-AEDs.

The clearance of zidovudine is significantly reduced by valproic acid resulting in elevated plasma (Lertora et al., 1993) and cerebrospinal fluid concentrations (Akula et al., 1997). The probable mechanism of this effect is by inhibition of zidovudine glucuronidation (Lertora et al., 1993).

167 Interaction between antiepileptic and non-antiepileptic drugs

Zonisamide

Zonisamide undergoes extensive metabolism, via CYP3A4, and approximately 30% of zonisamide is excreted in urine as unchanged zonisamide. Plasma protein binding is 50%.

As a new AED, knowledge of the interaction profile of zonisamide with non- AEDs is rather limited. Only interactions with specific drugs have been investigated.

Interactions affecting zonisamide

Sulfonamides

In vitro studies show that sulpfonamides can readily displace zonisamide from its binding to erythrocytes (Matsumoto et al., 1989) but not from albumin (Matsumoto et al., 1983). The clinical significance of this interaction is not known.

Other drugs

Interactions affected by zonisamide

To date there are no reports of zonisamide affecting the pharmacokinetics of non-AEDs.

REFERENCES

Abbott FS, Kassam J, Orr JM, et al. The effect of aspirin on valproic acid metabolism. Clin Pharmacol Ther 1986; 40: 94–100.

Abernethy DR, Greenblatt DJ, Morse DS, et al. Interaction of propoxyphene with diazepam, alprazolam and lorazepam. Br J Clin Pharmacol 1985; 19: 51–57.

Ahmad S. Renal insensitivity to furosemide caused by chronic anticonvulsant therapy. Br Med J 1974; 3: 657–659.

Aitio ML, Mansbury L, Tala E, et al. The effect of enzyme induction on the metabolism of disopyramide in man. Br J Clin Pharmacol 1981; 11: 279–286.

Akula SK, Rege AB, Dreisbach AW, et al. Valproic acid increases cerebrospinal fluid zidovudine levels in a patient with AIDS. Am Med Sci 1997; 313: 244–246.

Albani F, Riva R, Baruzzi A. Clarithromycin–carbamazepine interaction: a case report. Epilepsia 1993; 34: 161–162.

Albin H, Vincan G, Pehourcq F, et al. Influence of josamycin treatment on carbamazepine kinetics (in French). Therapie 1982; 37: 151–156.

In vitro studies have shown that the metabolism of zonisamide is inhibited in descending order of potency by ketoconazole, cyclosporin, dihydroergotamine, itraconazole, miconazole, triazolam and fluconazole (Sugihara et al., 1996). The clinical significance of these interactions is not known.

168 Jerzy Majkowski and Philip N. Patsalos

Alvarez JS, Del Castillo JAS, Ortiz MJA. Effect of carbamazepine on cyclosporin blood level. Nephron 1991; 58: 235–236.

Amabeoku GJ, Chikuni O, Akino C, et al. Pharmacokinetic interaction of single doses of quinine and carbamazepine, phenobarbitone and phenytoin in healthy volunteers. East Afr Med J 1993; 70: 90–93.

Babany G, Larrey D, Pessayre D. Macrolide antibiotics as inducers and inhibitors of cytochrome P450 in experimental animals and man. Prog Drug Metab 1988; 11: 61–98.

Bachmann KA, Jauregui L. Use of simple clearance estimates of cytochrome P450 substrates to characterize human hepatic CYP status in vivo. Xenobiotica 1993; 3: 307–315.

Bahls FH, Ozuma J, Ritchie DE. Interactions between calcium channel blockers and anticonvul- sants carbamazepine and phenytoin. Neurology 1991; 41: 470–472.

Baker DK, Relling MV, Pui CH, et al. Increased teniposide clearance with concomitant anticon- vulsant therapy. J Clin Oncol 1992; 10: 311–315.

Baruzzi A, Albani F, Riva R. Oxcarbazepine: pharmacokinetic interactions and their clinical rel- evance. Epilepsia 1994; 35(Suppl. 3): S14–S19.

Barzaghi N, Gatti G, Crema F, et al. Effect of flurithromycin, a new macrolide antibiotic, on carbamazepine disposition in normal subjects. Int J Clin Pharmacol Res 1988; 8: 101–105.

Begg EJ, Chinwah PM, Webb C, et al. Enhanced metabolism of mexiletine after phenytoin administration. Br J Clin Pharmacol 1982; 14: 219–223.

Berg MJ, Berlinder WG, Goldberg MJ, et al. Acceleration of the body clearance of phenobarbital by oral activated charcoal. New Engl J Med 1982; 307: 642–644.

Berg MJ, Rose JQ, Wurster DE, et al. Effect of charcoal and corbitol–charcoal suspension on the elimination of intravenous phenobarbital. Ther Drug Monit 1987; 9: 41–47.

Bergendal L, Friberg A, Schaffrath AM, et al. The clinical relevance of the interaction between carbamazepine and dextropropoxyphene in elderly patients in Gothenburg, Sweden. Eur J Clin Pharmacol 1997; 53: 203–206.

Bittencourt PRM, Gracia CM, Martins R, et al. Phenytoin and carbamazepine decrease oral bioavailability of praziquantel. Neurology 1992; 42: 492–496.

Block SH. Carbamazepine-isoniazid interaction. Paediatrics 1982; 69: 494–495.
Blum RA, Wilton JH, Hilligross DM, et al. Effect of fluconazole on disposition of phenytoin. Clin

Pharmacol Ther 1990; 47: 182.
Bollini P, Riva R, Albani F, et al. Decreased phenytoin level during antineoplastic therapy: a case

report. Epilepsia 1983; 24: 75–78.
Bonay M, Jonville-Bera AP, Diot P, et al. Possible interaction between phenobarbital, carba-

mazepine and itraconazole. Drug Saf 1993; 9: 309–311.
Bourgeois BF, Dodson WE, Ferrendelli JA. Interactions between primidone, carbamazepine and

nicotinamide. Neurology 1982; 32: 1122–1126.
Brodie MJ. Tiagabine pharmacology in profile. Epilepsia 1995; 36(Suppl. 6): S7–S9.
Brodie MJ, MacPhee GJ. Carbamazepine neurotoxicity precipitated by diltiazem. Br Med J 1986;

292: 1170–1171.
Brooks S, Werk E, Ackerman S, et al. Adverse effects of phenobarbital on corticosteroid metabo-

lism in patients with bronchial asthma. New Engl J Med 1972; 286: 1125–1128.

169 Interaction between antiepileptic and non-antiepileptic drugs

Brooks P, Buchanan W, Grove M, et al. Effects of enzyme induction on metabolism of pred- nisolone: clinical and laboratory study. Ann Rheum Dis 1976; 35: 339–343.

Brown RI, Cooper TG. Ticlopidine–carbamazepine interaction in a coronary stent patient. Can J Cardiol 1997; 13: 853–854.

Burckart G, Venkataramanan R, Starzl T, et al. Cyclosporin clearance in children following organ transplantation. J Clin Pharmacol 1984; 24: 412.

Burman W, Orr L. Carbamazepine toxicity after starting combination antiretroviral therapy including ritonavir and efavirenz. AIDS 2000; 14: 2793–2794.

Busch JA, Radulovic LI, Bockbrader HN, et al. Effect of Maalox TC on single-dose pharmaco- kinetics of gabapentin capsules in healthy subjects. Pharm Res 1992; 9(Suppl. 10): S315.

Cadle RM, Zenon III GJ, Rodriguez-Barradas MC, et al. Fluconazole-induced symptomatic phenytoin toxicity. Ann Pharmacother 1994; 28: 191–195.

Capewell S, Freestone S, Critchley J, et al. Reduced felodipine bioavailability in patients taking anticonvulsants. Lancet 1988; 480–482.

Carstensen H, Jacobsen N, Dieperink H. Interaction between cyclosporin A and phenobarbi- tone. Br J Clin Pharmacol 1986; 21: 550–551.

Chalk JB, Ridgeway K, Brophy T, et al. Phenytoin impairs the bioavailability of dexamethasone in neurological and neurosurgical patients. J Neurol Neurosurg Psychiatr 1984; 47: 1087–1090. Coniglia AA, Garnett WR, Pellock JH, et al. Effect of acute and chronic terfenadine on free and

total serum concentration in epileptic patients. Epilepsia 1989; 30: 611–616.
Cooney GF, Mochon M, Kaiser B, et al. Effects of carbamazepine on cyclosporine metabolism in

pediatric renal transplant recipients. Pharmacotherapy 1995; 15: 353–356.
Couet W, Istin B, Ingrand I, et al. Effect of ponsinomycin on single-dose kinetics and metabolism

of carbamazepine. Ther Drug Monit 1990; 12: 144–149.
Cropp JS, Bussey HJ. A review of enzyme induction of warfarin metabolism with recommenda-

tions for patient management. Pharmacotherapy 1997; 17: 917–928.
Dalton MJ, Powell JR, Messenheimer Jr JA, et al. Cimetidine and carbamazepine: a complex drug

interaction. Epilepsia 1986; 27: 553–558.
Dam M, Christensen J. Interaction of propoxyphene with carbamazepine. Lancet 1977; 2: 509. Dam M, Christensen JM, Brandt J, et al. Antiepileptic drugs: interaction with dextro-

propoxyphene. In Antiepileptic Therapy: Advances in Drug monitoring. S. I. Johannessen,

P. L. Morselli, C. E. Pippenger, et al., eds. New York: Raven Press, 1980: 299–304.
Dasgupta A, Luke M. Valproic acid–ketoconazole interaction in normal, hypoalbuminemic, and uremic sera: lack of interaction in uremic serum caused by the presence of inhibitor. Ther

Drug Monit 1997; 19: 281–285.
Dasgupta A, Volk A. Displacement of valproic acid and carbamazepine from protein binding in

normal and uremic sera by tolmetin, ibuprofen, and naproxen: presence of inhibitor in uremic

serum that blocks valproic acid–naproxen interactions. Ther Drug Monit 1996; 18: 284–287. Denahue S, Flockart DA, Albernethy DR. Ticlopidine inhibits phenytoin clearance. Clin

Pharmacol Ther 1999; 66: 563–568.
Depot M, Powell JR, Messenheimer Jr JA, et al. Kinetic effects of multiple oral doses of aceta-

minophen on a single oral dose of lamotrigine. Clin Pharmacol Ther 1990; 48: 346–355.

170 Jerzy Majkowski and Philip N. Patsalos

de Wolff F, Vermeij P, Ferrari MD, et al. Impairment of phenytoin parahydroxylation as a cause of severe intoxication. Ther Drug Monit 1983; 5: 213–215.

Ducharme MP, Slaughter RL, Warbasse LH. Itraconazole and hydroxyitraconazole serum concen- trations are reduced more than tenfold by phenytoin. Clin Pharmacol Ther 1995; 58: 617–624. Ebert U, Thong NQ, Oertel R, et al. Effects of rifampicin and cimetidine on pharmacokinetics and

pharmacodynamics of lamotrigine in healthy subjects. Eur J Clin Pharmacol 2000; 56: 299–304. Eimer M, Carter BL. Elevated serum carbamazepine concentrations following diltiazem initia-

tion. Drug Intel Clin Pharm 1987; 21: 340–342.
Eradiri O, Jamali F, Thomson A. Interaction of metronidazole with phenobarbital, cimetidine,

prednisone, and sulfasalazine in Crohn’s disease. Biopharm Drug Dispos 1988; 9: 219–227. Feierman DE, Lasker JM. Metabolism of fentanyl, a synthetic opioid analgesic, by human liver

microsomes: role of CYP3A4. Drug Metab Dispos 1996; 24: 932–939.
Fernandez MC, Erill S, Lucena MI, et al. Serum protein binding of tolbutamide in patients

treated with antiepileptic drugs. Clin Pharmacokin 1985; 10: 451–455.
Fleishaker JC, Pearson LK, Peters GR. Gender does not affect the degree of induction of tirilazad

clearance by phenobarbital. Eur J Clin Pharmacol 1996; 50: 139–145.
Fleishaker JC, Pearson LK, Peters GR. Induction of tirilazad clearance by phenytoin. Biopharm

Drug Dispos 1998; 19: 91–96.
Fleitman JS, Bruni J, Perrin JH, et al. Albumin-binding interactions of sodium valproate. J Clin

Pharmacol 1980; 20: 514–517.
Flockart DA, Tanus-Santos JE. Implications of cytochrome P450 interactions when prescribing

medication for hypotension. Arch Intern Med 2002; 162: 405–412.
Forney RB, Hughes FW. Meprobamate, ethanol or meprobamate–ethanol combinations on per-

formance of human subjects under delayed autofeedback (DAF). J Psychol 1964; 57: 431–436. Fraser DG, Ludden TM, Evens RP, et al. Displacement of phenytoin from plasma binding sites by

salicylate. Clin Pharmacol Ther 1980; 27: 165–169.
Freedman MD, Olatidoye AG. Clinically significant drug interactions with anticoagulants. Drug

Saf 1994; 10: 381–394.
Freeman DJ, Laupacis A, Keown A, et al. Evaluation of cyclosporin and phenytoin interaction

with observations on cyclosporin metabolites. Br J Clin Pharmacol 1984; 18: 887–893.
Frenia ML, Schauben JL, Wears RL, et al. Multiple-dose activated charcoal compared to urinary alkalinization for the enhancement of phenobarbital elimination. J Toxicol Clin Toxicol 1996;

34: 169–175.
Garcia BA, Latorre IA, Porta EJ, et al. Protease inhibitor-induced carbamazepine toxicity. Clin

Neuropharmacol 2000; 23(4): 216–218.
Ghosn M, Carde P, Leclerq B, et al. Ifosfamide/mesna related encephalopathy: a case report with

a possible role of phenobarbital in enhancing neurotoxicity. Bull Cancer 1988; 75: 391–392. Glue P, Banfield CR, Perhach JL, et al. Pharmacokinetic interactions with felbamate: in vitro–in

vivo correlation. Clin Pharmacokinet 1997; 33: 214–223.
Goulden KI, Dooley IM, Camfield PR, et al. Clinical valproate toxicity induced by acetylsalicylic

acid. Neurology 1987; 37: 1392–1394.
Gugler R, Jensen JC. Omeprazole inhibits oxidative drug metabolism. Gastroenterology 1985; 89:

1235–1241.

171 Interaction between antiepileptic and non-antiepileptic drugs

Gupte S. Phenobarbital and metabolism of metronidazole. New Engl J Med 1983; 308: 529. Gustavson LE, Mengel HB. Pharmacokinetics of tiagabine; a -aminobutyric acid-uptake

inhibitor, in healthy subjects after single and multiple doses. Epilepsia 1995; 36: 605–611. Hansen BS, Dam M, Brandt J, et al. Influence of dextropropoxyphene on steady state serum levels and protein binding of three anti-epileptic drugs in man. Acta Neurol Scand 1980; 61: 357–367. Hansen JJM, Siersbaek-Nielsen K, Kristensen M, et al. Effect of diphenylhydantoin on the metab-

olism of dicoumarol in man. Acta Med Scand 1971; 189: 15–19.
Harder S, Schneider W, Bae ZU, et al. Undesirable drug reactions in simultaneous administration of

high-dosage phenobarbital and beta-lactam antibiotics. Klinische Padiatrie 1990; 202: 404–407. Hayden M, Buchanan N. Danazol–carbamazepine interaction (Letter). Med J Aust 1991; 155: 851. Howitt KM, Oziemski MA. Phenytoin toxicity induced by fluconazole. Med J Aust 1989; 151:

603–604.
Hugen PW, Burger DM, Brinkman K, et al. Carbamazepine–indinavir interaction causes anti-

retroviral therapy failure. Ann Pharmacother 2000; 34(4): 465–470.
Jonkman JHG, Upton RA. Pharmacokinetic drug interactions with theophylline. Clin Pharma-

cokinet 1984; 9: 309–334.
Jonville AP, Gauchez AS, Autret E, et al. Interaction between isoniazid and valproate: a case of

valproate overdosage. Eur J Clin Pharmacol 1991; 40: 197–198.
Kandrotas R, Cranfield T, Gal P, et al. Effect of phenobarbital administration on theophylline

clearance in premature neonates. Ther Drug Monit 1990; 12: 139–143.
Kastberg H, Jansen JA, Cole G, et al. Tiagabine: absence of kinetic or dynamic interactions with

ethanol. Drug Metabol Drug Interact 1998; 14: 259–273.
Kato Y, Fujii T, Mizoguchi N, et al. Potential interaction between ritonavir and carbamazepine.

Pharmacotherapy 2000; 20: 851–854.
Kay L, Kampmann JP, Svendsen TL, et al. Influence of rifampin and isoniazid on the kinetics of

phenytoin. Br J Clin Pharmacol 1985; 20: 323–326.
Keränen T, Jolkkonen J, Jensen PK, et al. Absence of interaction between oxcarbazepine and

erythromycin. Acta Neurol Scand 1992a; 86: 20–23.
Keränen T, Jolkkonen J, Klosterskov Jensen P, et al. Oxcarbazepine does not interact with cimeti-

dine in healthy volunteers. Acta Neurol Scand 1992b; 85: 239–242.
Kishi T, Fujita N, Eguchi T, et al. Mechanism for reduction of serum folate by antiepileptic drugs

during prolonged therapy. J Neurol Sci 1997; 145: 109–112.
Klaassen SL. Ticlopidine-induced phenytoin toxicity. Ann Pharmacother 1998; 32: 1295–1298. Koup JR. Interaction of chloramphenicol with phenytoin and phenobarbital: a case report. Clin

Pharmacol Ther 1978; 24: 571–575.
Koup JR, Gibaldi M, McNamara P, et al. Interaction of chloramphenicol with phenytoin and

phenobarbital. Clin Pharmacol Ther 1978; 24: 571–575.
Krämer G, Theisohn M, von Unruh GE, et al. Carbamazepine–danazol drug interaction: its

mechanism examined by a stable isotope technique. Ther Drug Monit 1986; 8: 387–392. Krämer G, Tettenborn B, Flesch G. Oxcarbazepine–verapamil drug interaction in healthy volun-

teers. Epilepsia 1991; 32(Suppl. 1): S70–S71.
Krämer G, Tettenborn B, Klosterskov Jensen P, et al. Oxcarbazepine does not affect the anticoa-

gulant activity of warfarin. Epilepsia 1992; 33: 1145–1148.

172 Jerzy Majkowski and Philip N. Patsalos

Krasinski K, Kusmiesz M, Nelson JD. Pharmacologic interactions among chloramphenicol, phenytoin and phenobarbital. Pediat Infect Dis 1982; 1: 232–235.

Kroemer HK, Gautier JC, Beaune P, et al. Identification of P450 enzymes involved in metabolism of verapamil in humans. Naunyn Schmiedebergs Arch Pharmacol 1993; 348: 332–337.

Kunze KL, Wienkers LC, Thummel KE, et al. Warfarin-fluconazole: I. Inhibition of the human cytochrome P450-dependent metabolism of warfarin by fluconazole: in vitro studies. Drug Metab Dispos 1996; 24: 414–421.

Kutt H. Interactions between anticonvulsants and other commonly prescribed drugs. Epilepsia 1984; 25(Suppl. 2): S118–S131.

Landay R, Gonzalez M, Taylor J. Effect of phenobarbital on theophylline disposition. J Aller Clin Immunol 1978; 62: 27–29.

Lazar JD, Wilner KD. Drug interactions with fluconazole. Rev Infect Dis 1990; 12: S327–S333. Lertora JJ, Greenspan DL, Rege AB, et al. Valproic acid inhibits glucuronidation of zidovudine

(AZT) in HIV-infected patients. Clin Pharmacol Ther 1993; 53: 197.
Levine M, Jones MW, Sheppard I. Differential effect of cimetidine on serum concentrations of

carbamazepine and phenytoin. Neurology 1985; 35: 562–565.
Levy RH. Cytochrome P450 isozymes and antiepileptic drug interactions. Epilepsia 1995;

36(Suppl. 5): S8–S13.
Levy RH, Ragueneau-Majlessi I, Baltes E. Repeated administration of the novel antiepileptic

agent levetiracetam does not alter digoxin pharmacokinetics and pharmacodynamics in

healthy volunteers. Epilepsia 2001; 46: 93–99.
Lewis DP, Van Dyke DC, Willhite LA, et al. Phenytoin–folic acid interactions. Ann Pharmacother

1995; 29: 726–735.
Liao S, Palmer M. Digoxin and topiramate drug interaction study in male volunteers. Pharm Res

1993; 10(Suppl. 1): S405.
MacPhee GJ, McInnes GT, Thompson GG, et al. Verapamil potentiates carbamazepine neuro-

toxicity: a clinically important inhibitory interaction. Lancet 1986; 1: 700–703.
Majkowski J. Carbamazepine and erythromycin interaction resulting in myoclonic seizures in

epileptic patients. Case report. Epileptologia 1995; 3: 57–62.
Malloy MJ, Ravis WR, Pennell AT, et al. Effect of cholestyramine resin on single dose valproate

pharmacokinetics. Int J Clin Pharmacol Ther 1996; 34: 208–211.
Maoz E, Grossman E, Thaler M, et al. Carbamazepine neurotoxic reaction after administration

of diltiazem. Arch Intern Med 1992; 152: 2503–2504.
Mateu-de Antonio J, Grau S, Gimeno-Bayon JL, et al. Ritonavir induced carbamazepine toxicity.

Ann Pharmacother 2001; 35: 125–126.
Matsumoto K, Miyazaki H, Fujii T, et al. Absorption, distribution and excretion of 3-(sul-

famoyl[14C]methyl)-1,2-benzisoxazole (AD-810) in rats, dogs and monkeys and of AD-810 in

men. Arzneimittelforschung 1983; 33: 961–968.
Matsumoto K, Miyazaki H, Fujii T, et al. Binding of sulfonamides to erythrocyte proteins and

possible drug–drug interaction. Chem Pharm Bull (Tokyo) 1989; 37: 2807–2810.
Mattson RH, Gallagher BB, Reynolds EH, et al. Folate therapy in epilepsy: a controlled study.

Arch Neurol 1973; 29: 78–81.

173 Interaction between antiepileptic and non-antiepileptic drugs

McGovern B, Geer VR, Laraia PJ, et al. Possible interaction between amiodarone and phenytoin. Ann Intern Med 1984; 101: 650–651.

Mead P, Gibson M, Schentag J, et al. Possible alteration of metronidazole metabolism by pheno- barbital. New Engl J Med 1982; 306.

Mengel H, Jansen JA, Sommerville KW, et al. Tiagabine evaluation of the risk of interaction with theophylline, warfarin, digoxin, cimetidine, oral contraceptives, traizolam, or ethanol. Epilepsia 1995; 36(Suppl. 3): S160.

Mesdjian E, Dravet C, Cenraud B, et al. Carbamazepine intoxication due to triacetylolean- domycin administration in epileptic patients. Epilepsia 1980; 21: 489–496.

Michelucci R, Cipolla G, Passarelli D, et al. Reduced plasma nisoldipine concentrations in phenytoin-treated patients with epilepsy. Epilepsia 1996; 37: 1107–1110.

Mitchell AS, Holland JT. Fluconazole and phenytoin: a predictable interaction. Br Med J 1989; 298: 1315.

Mogensen PH, Jorgensen L, Boas J, et al. Effects of dextropropoxyphene on the steady-state kinetics of oxcarbazepine and its metabolites. Acta Neurol Scand 1992; 85: 14–17.

Molholm Hansen J, Kampmann JP, Siersbaek-Nielsen K, et al. The effect of different sulfon- amides on phenytoin metabolism in man. Acta Med Scand 1979; 624(Suppl. 1): S106–S110. Nair DR, Morris HH. Potential fluconazole-induced carbamazepine toxicity. Ann Pharmacother

1999; 33: 790–792.
Nappi J. Warfarin and phenytoin interaction. Ann Intern Med 1979; 90: 852.
Nation RL, Evans AM, Milne RW. Pharmacokinetic interactions with phenytoin. Clin Pharma-

cokinet 1990; 18: 37–60.
Neef C, de Voogd-van der Straaten I. An interaction between cytostatic and anticonvulsant

drugs. Clin Pharmacol Therapeut 1988; 43: 372–375.
Neuvonen PJ, Elonen E. Effect of activated charcoal on absorption and elimination of pheno-

barbitone, carbamazepine and phenylbutazone in man. Eur J Clin Pharrnacol 1980; 17: 51–57. Neuvonen PJ, Penttila O, Lehtovaara R, et al. Effects of antiepileptic drugs on the elimination of

various tetracycline derivatives. Eur J Clin Pharmacol 1975; 9: 147–154.
Nishioka T, Ikegami M, Imanishi M, et al. Interaction between phenobarbital and ciclosporin fol-

lowing renal transplantation: a case report. Hinyokikia Kiyo Acta Urol Jap 1990; 36: 447–450. Nolan PE, Marcus FI, Hoyer GL, et al. Pharmacokinetic interaction between intravenous pheny-

toin and amiodarone in healthy volunteers. Clin Pharmacol Ther 1989; 46: 43–50.
O’Connor NK, Fris J. Clarithromycin–carbamazepine interaction in a clinical setting. Am Board

Family Pract 1994; 7: 489–492.
Odishaw J, Chen C. Effects of steady-state bupropion on the pharmacokinetics of lamotrigine in

healthy subjects. Pharmacotherapy 2000; 20: 1448–1453.
Olesen OV. The influence of disulfiram and calcium carbamide on the serum diphenylhydan-

toin. Arch Neurol 1967; 16: 642–644.
Orme M, Breckenridge A. Enantiomers of warfarin and phenobarbital. New Engl J Med 1976;

295: 1482.
Panjehshahin MR, Bowmer CJ, Yates MS. Effect of valproic acid, its unsaturated metabolites and

some structurally related fatty acids on the binding of warfarin and dansylsarcosine to human albumin. Biochem Pharmacol 1991; 41: 1227–1233.

174 Jerzy Majkowski and Philip N. Patsalos

Patsalos PN. Pharmacokinetic profile of levetiracetam: toward ideal characteristics Pharmacol Ther 2000; 85: 77–85.

Patsalos PN, Duncan JS. Antiepileptic drugs: a review of clinically significant drug interactions. Drug Saf 1993; 9: 156–184.

Patsalos PN, Elyas AA, Ratnaraj N, et al. Concentration-dependent displacement of tiagabine by valproic acid. Epilepsia 2002; 43(Suppl. 8): S143.

Patsalos PN, Froscher W, Pisani F, et al. The importance of drug interactions in epilepsy therapy. Epilepsia 2002; 43: 365–385.

Patsalos PN, Perucca E. Clinically important interactions in epilepsy: general features and inter- actions between antiepilepticdrugs. Lancet Neurol 2003a; 2: 347–356.

Patsalos PN, Perucca E. Clinically important interactions in epilepsy: interactions between antiepileptic drugs and other drugs. Lancet Neurol 2003b; 2: 473–481.

Patterson BD. Possible interaction between metronidazole and carbamazepine. Ann Pharmacother 1994; 28: 1303–1304.

Penttila O, Neuvonen PJ, Aho K, et al. Interaction between doxycycline and some antiepileptic drugs. Br Med J 1974; 2: 470–472.

Periti P, Mazzei T, Mini E, et al. Pharmacokinetic drug interactions of macrolides. Clin Pharmacokinet 1992; 23: 106–131.

Permeggiani A, Riva R, Posar A, et al. Possible interaction between acyclovir and antiepileptic treatment. Ther Drug Monit 1995; 17: 312–315.

Pessayre D, Larrey D, Funck-Brentano C, et al. Drug interactions and hepatitis produced by some macrolide antibiotics. J Antimicrob Chemother 1985; 16(Suppl. A): S181–S194.

Phillips P, Hansky J. Phenytoin toxicity secondary to cimetidine administration. Med J Aust 1984; 141: 602.

Platt PR, Thackery NM. Phenytoin-induced resistance to vecuronium. Anesth Intens Care 1993; 21: 185–191.

Pond SM, Kretschzmar KM. Effect of phenytoin on meperidine clearance and normeperidine formation. Clin Pharmacol Ther 1981; 30: 680–686.

Posner J, Webster H, Yuen AWC. Investigation of the ability of lamotrigine, a novel antiepileptic drug, to induce mixed function oxygenase enzymes. Br J Clin Pharmacol 1991; 32: 658.

Prichard PJ, Walt RP, Kitchingman GK, et al. Oral phenytoin pharmacokinetics during omepra- zole therapy. Br J Clin Pharmacol 1987; 24: 543–545.

Principi N, Espasito S. Comparative tolerability of erythromycin and newer macrolide anti- bacterials in paediatric patients. Drug Saf 1999; 20: 25–41.

Privitera M, Welty TE. Acute phenytoin toxicity followed by seizure breakthrough from a ticlopidine–phenytoin interaction. Arch Neurol 1996; 53: 1191–1192.

Rabinowicz AL, Hinton DR, Dyck P, et al. High-dose tamoxifen in treatment of brain tumours: interaction with antiepileptic drugs. Epilepsia 1995; 36: 513–515.

Ragueneau-Majlessi I, Levy RH, Meyerhoff C. Lack of effect of repeated administration of leve- tiracetam on the pharmacodynamics and pharmacokinetic profile of warfarin. Epilepsy Res 2001; 47: 55–63.

Rameis H. On the interaction between phenytoin and digoxin. Eur J Clin Pharmacol 1985; 29: 49–53.

175 Interaction between antiepileptic and non-antiepileptic drugs

Relling MV, Nemec J, Schuetz EG, et al. O-demethylation of epipodophyllotoxins is catalyzed by human cytochrome P450 3A4. Mol Pharmacol 1994; 45: 352–358.

Relling MV, Pui CH, Sandlund JT, et al. Adverse effect of anticonvulsants on efficacy of chemotherapy for acute lymphoblastic leukaemia. Lancet 2000; 356: 285–290.

Rettie AE, Korzekwa KR, Kunze KL, et al. Hydroxylation of warfarin by human cDNA-expressed cytochrome P-450; a role for P-4502C9 in the etiology of (S)-warfarin–drug interactions. Chem Res Toxicol 1992; 5: 54–59.

Reynolds EH. Chronic antiepileptic toxicity: a review. Epilepsia 1974; 16: 319–352.
Reynolds EH, Mattson RH, Gallagher BB. Relationship between serum and cerebrospinal fluid anticonvulsant drug and folic acid concentrations in epileptic patients. Neurology 1972; 22:

841–844.
Richens A, Marshall RW, Dirach J. Absence of interaction between tiagabine, a new antiepileptic

drug, and the benzodiazepine triazolam. Drug Metabol Drug Interact 1998; 14: 159–177.
Roe TF, Podosin RL, Blaskovics ME. Drug interaction: diazoxide and diphenylhydantoin.

J Pediatr 1975; 87: 480–484.
Rolan PE, Somogy AA, Drew MR, et al. Phenytoin intoxication during treatment with parenteral

miconazole. Br Med J 1983; 287: 1760.
Rutledge DR, Pieper JA, Mirvis DM. Effects of chronic phenobarbital on verapamil disposition

in humans. J Pharmacol Exp Ther 1988; 246: 7–13.
Saccar C, Danish M, Ragni M, et al. The effect of phenobarbital on theophylline disposition in

children with asthma. J Allergy Clin Immunol 1985; 75: 716–719.
Saint-Salvi B, Tremblay D, Surjus A, et al. A study of the interaction of roxithromycin with theo-

phylline and carbamazepine. J Antimicrob Chemother 1987; 20(Suppl. B): S121–S129.
Salem RB, Breland BD, Mishra SK, et al. Effect of cimetidine on phenytoin serum level. Epilepsia

1983; 24: 284–288.
Sandor P, Sellers EM, Dumbell M, et al. Effect of long and short term alcohol on phenytoin kinet-

ics in chronic alcoholics. Clin Pharmacol Ther 1981; 30: 390–397.
Sands BF, Knapp CM, Ciraulo DA. Medical consequences of alcohol–drug interactions. Alcohol

Health Res World 1993; 17(4): 316–320.
Schellens JH, van der Wart JH, Brugman M, et al. Influence of enzyme induction and inhibition

on the oxidation of nifedipine, sparteine, mephenytoin and antipyrine in humans as assessed

by a “cocktail” study design. J Pharmacol Exp Ther 1989; 249: 638–645.
Schlienger R, Kurmann ML, Drewe J, et al. Inhibition of phenprocoumon anticoagulation by

carbamazepine. Eur Neuropsychopharmacol 2000; 10: 219–221.
Schroder H, Ostergaard JR. Interference of high-dose methotrexate in the metabolism of val-

proate? Pediatr Hematol 1994; 11: 445–449.
Schobben F, Vree TB, van der Kleijn D. Pharmacokinetics, metabolism and distribution of 2-N–

propyl pentanoate (sodium valproate) and the influence of salicylate comedication. In Advances

in Epileptology. H. Meinardi, A. J. Rowan, eds. Amsterdam: Swets & Zeitlinger, 1978: 271–277. Sklar SJ, Wagner JC. Enhanced theophylline clearance secondary to phenytoin therapy. Drug

Intell Clin Pharm 1985; 19: 34–36.
Skovsted L, Hansen JM, Kristensen M, et al. Inhibition of drug metabolism in man. In Drug

Interactions. P. L. Morselli, S. Garattini, S. N. Cohen, eds. New York: Raven Press, 1974: 81–90.

176 Jerzy Majkowski and Philip N. Patsalos

Snel S, Jansen JA, Pedersen PC, et al. Tiagabine, a novel antiepileptic agent: lack of pharmacoki- netic interaction with digoxin. Eur J Clin Pharmacol 1998; 54: 355–357.

Solomon HM, Reich S, Spirt N, et al. Interaction between digitoxin and other drugs in vitro and in vivo. Ann NY Acad Sci 1971; 79: 362–369.

Somogyi A, Thielscher S, Gugler R. Influence of phenobarbital on cimetidine kinetics. Eur J Clin Pharmacol 1981; 19: 343–347.

Soriano SG, Kaus SJ, Sullivan LJ, et al. Onset and duration of action of rocuronium in children receiving chronic anticonvulsant therapy. Pediatr Anaesth 2000; 10: 133–136.

Spina E, Pisani F, Perucca E. Clinically significant pharmacokinetic drug interactions with car- bamazepine: an update. Clin Pharmacokinet 1996; 31: 198–214.

Spina E, Arena D, Scordo MG, et al. Elevation of plasma carbamazepine concentrations by keto- conazole in patients with epilepsy. Ther Drug Monit 1997; 19: 535–538.

Stjernholm M, Katz F. Effects of diphenylhydantoin, phenobarbital, and diazepam on the metabo- lism of methylprednisolone and its sodium succinate. J Clin Endocrinol Metab 1975; 41: 887–893. Sugihara K, Kitamura S, Tatsumi K. Involvement of mammalian liver cytosols and aldehyde oxi-

dase in reductive metabolism of zonisamide. Drug Metab Dispos 1996; 24: 199–202.
Sutton G, Kupferberg HJ. Isoniazid as an inhibitor of primidone metabolism. Neurology 1975;

25: 1179–1181.
Svendsen TL, Kristensen M, Hansen JM, et al. The influence of disulfiram on the half-life and

metabolic clearance rate of diphenylhydantoin and tolbutamide in man. Eur J Clin Pharmacol

1976; 9: 439–441.
Sylvester RK, Lewis FB, Caldwell KC, et al. Impaired phenytoin bioavailability secondary to cis-

platinum, vinblastin and bleomycine. Ther Drug Monit 1984; 6: 302–305.
Syverson GB, Morgan JP, Weintraub M, et al. Acetazolamide-induced interference with primi-

done absorption. Arch Neurol 1977; 34: 80–84.
Tartara A, Galimberti CA, Manni R, et al. Differential effects of valproic acid and enzyme-

inducing anticonvulsants on nimodipine pharmacokinetics in epileptic patients. Br J Clin

Pharmacol 1991; 32: 335–340.
Tecoma ES. Oxcarbazepine. Epilepsia 1999; 40(Suppl. 5): S37–S46.
Tempelhoff R, Modica PA, Spitznagel EL. Anticonvulsant therapy increases fentanyl require-

ments during anaesthesia for craniotomy. Can J Anaesth 1990; 37: 327–332.
Thompson MS, Groes L, Schwietert HR, et al. An open label sequence listed two period crossover pharmacokinetic trials evaluating the possible interaction between tiagabine and erythromycin during multiple administration to healthy volunteers. Epilepsia 1997;

38(Suppl. 3): S64.
Tong TG, Pond SM, Kreek MJ, et al. Phenytoin induced methadone withdrawal. Ann Intern Med

1981; 94: 349–351.
Tracy TS, Korzekwa KR, Gonzalez FJ, et al. Cytochrome P450 isoforms involved in metabolism

of the enantiomers of verapamil and norverapamil. Br J Clin Pharmacol 1999; 47: 545–552. Tucker RM, Denning DW, Hanson LH, et al. Interaction of azoles with rifampin, phenytoin, and

carbamazepine: in vitro and clinical observation. Clin Infect Dis 1992; 14: 165–174.
Udall J. Clinical implications of warfarin interactions with five sedatives. Am J Cardiol 1975; 35:

67–71.

177 Interaction between antiepileptic and non-antiepileptic drugs

Valsalan VC, Cooper GL. Carbamazepine intoxication caused by interaction with isoniazid. Br Med J 1982; 285: 261–262.

Van Wieringen A, Vrijlandt CM. Ethosuximide intoxication caused by interaction with isoniazid. Neurology 1983; 33: 1227–1228.

Villikka K, Kivisto KT, Maenpaa H, et al. Cytochrome P450-inducing antiepileptics increase the clearance of vincristine in patients with brain tumors. Clin Pharmacol Ther 1999; 66: 589–593. Vincon G, Albin H, Demotes-Mainard F, et al. Effects of josamycin on carbamazepine kinetics.

Eur J Clin Pharmacol 1987; 32: 321–323.
von Moltke LL, Durol AL, Duan SX, et al. Potent mechanism based inhibition of human CYP3A

in vitro by amprenavir and ritonavir: comparison with ketoconazole. Eur J Clin Pharmacol

2000; 56: 259–261.
Wakabayashi Y, Maruyama S, Hachimura K, et al. Activated charcoal interrupts enterohepatic

circulation of phenobarbital. J Toxicol Clin Toxicol 1994; 32: 419–424.
Walubo A, Aboo A. Phenytoin toxicity, due to concomitant antituberculosis therapy. South Afr

Med J 1995; 85: 1175–1176.
Wassner S, Pennisi A, Malekzadeh M, et al. The adverse effect of anticonvulsant therapy on renal

allograft survival: a preliminary report. J Pediatr 1976; 88: 134–137.
Webster LK, Michaly GW, Jones DB. Effect of cimetidine and ranitidine on carbamazepine and

sodium valproate pharmacokinetics. Eur J Clin Pharmacol 1984; 27: 341–343.
Welling PG. Interactions affecting drug absorption. Clin Pharmacokinet 1984; 9: 404–434. Wesseling H, Molsthurkow I. Interaction of diphenylhydantoin (DPH) and tolbutamide in man.

Eur J Clin Pharmacol 1975; 8: 75–78.
Williams K, Begg E, Wade D, et al. Effects of phenytoin, phenobarbital and ascorbic acid on

misonidazole elimination. Clin Pharmacol Ther 1983; 33: 314–321.
Williams ML, Wainer IW, Embree L, et al. Enantioselective induction of cyclophosphamide

metabolism by phenytoin. Chirality 1999; 11: 569–574.
Witmer DR, Ritschel WA. Phenytoin and isoniazid interaction: a kinetic approach to manage-

ment. Drug Intell Clin Pharm 1984; 18: 483–486.
Wong DD, Longenecer RG, Liepman M, et al. Phenytoin–dexamethasone: a possible drug–drug

interaction. J Am Med Assos 1985; 254: 2062–2063.
Wright JM, Stokes EF, Sweeney VP. Isoniazid-induced carbamazepine toxicity and vice versa: a

double drug interaction. New Engl J Med 1982; 307: 1325–1327.
Yasui N, Otani K, Kaneko S. Carbamazepine toxicity induced by clarithromycin coadministra-

tion in psychiatric patients. Intern Clin Psychopharmacol 1997; 12: 225–229.
Yazdani M, Kissling GE, Tran TH. Phenobarbital increases the theophylline requirement of pre-

mature infants being treated of apnea. Am J Dis Child 1987; 141: 97–99.
Zaccara G, Gangemi PF, Bendoni L, et al. Influence of single and repeated doses of oxcarbazepine

on the pharmacokinetic profile of felodipine. Ther Drug Monit 1993; 15: 39–42.
Zielinski JJ, Lichten EM, Haidukewych D. Clinically significant danazol–carbamazepine interac-

tion. Ther Drug Monit 1987; 9: 24–27.

Part III

Pharmacodynamic interactions

Pharmacodynamic principles and mechanisms of drug interactions

Blaise F. D. Bourgeois
Harvard Medical School, Division of Epilepsy and Clinical Neurophysiology, Children’s Hospital, Boston, MA, USA

Distinction between pharmacodynamic and pharmacokinetic drug interactions

The term pharmacokinetics refers to mostly quantitative assessments of what hap- pens to a drug in the body following its administration by any route. The processes that are assessed include absorption into the blood, serum protein binding, distri- bution into various tissues or compartments, biotransformation, and elimination from the body. The drug can be eliminated unchanged through the kidneys or in the form of a conjugate or metabolite following conjugation or enzymatic bio- transformation in the liver. Metabolites can be eliminated through the kidneys or through the bile. Pharmacokinetics rely on measurements of drug concentrations in various body fluids or tissues (in practice mostly in the blood, urine, or saliva) and assessments of changes in these concentrations over time. The clinical rele- vance of pharmacokinetics is based on the fact that optimal treatment with a drug requires achieving and maintaining certain levels in the target organs, and corre- sponding levels in the blood.

The term pharmacodynamics refers to qualitative and quantitative assessments of all possible effects of a drug in various organs of the body. These effects may include:

. 1 one or more desirable therapeutic effects (e.g. seizure reduction, prevention of migraine headaches or a positive psychotropic effect);

. 2 one or more undesirable/harmful adverse effects (e.g. sedation or an allergic reaction);

. 3 side effects that may be either desirable or undesirable (e.g. weight loss);

. 4 side effects that are neither desirable nor undesirable (e.g. elevation of gamma-
glutamyl transpeptidase (gamma-GT), lowering of bilirubin levels).

Some desirable and undesirable pharmacodynamic drug effects can be assessed quantitatively (e.g. seizure reduction, excessive weight gain, and hyponatremia), some can be assessed semi-quantitatively (e.g. decreased seizure severity, sedation

181

182 Blaise F. D. Bourgeois

or gum hypertrophy), and some can virtually not be assessed quantitatively (mostly idiosyncratic reactions). Obviously, pharmacodynamics are more complex and more difficult to assess than pharmacokinetics. Also, since many pharmacodynamic effects of drugs are related to concentrations, pharmacodynamic observations may be influenced by pharmacokinetics. However, pharmacodynamics have no influ- ence on pharmacokinetics, with the exception of hepatic enzyme induction and inhibition.

Based on the above concepts, there is a fundamental difference between phar- macokinetic and pharmacodynamic interactions. Pharmacokinetic interactions consist of alterations in the concentration of a drug that are caused by the presence of another drug in the body. This may include competition for absorption, dis- placement from protein-binding sites, enzymatic induction, enzymatic inhibition, or competition for renal excretion. Pharmacokinetic interactions are relatively easy to assess quantitatively. They are mostly undesirable but, when they are known, they can be anticipated and corrected. Any pair of drugs may or may not have pharmacokinetic interactions.

Pharmacodynamic interactions consist of the quantitative or qualitative alter- ations of any effect of a drug on any organ when these alterations are caused by the presence of another drug in the body. Pharmacodynamic interactions are much more difficult to assess quantitatively. They can be desirable (enhancement of a therapeutic effect or reduction of an adverse effect) or undesirable (enhancement of an adverse effect or reduction of a therapeutic effect). However, even when they are known and predicted, pharmacodynamic interactions cannot be influenced, corrected, or avoided. By altering drug concentrations, pharmacokinetic interac- tions may cause apparent pharmacodynamic interactions in the absence of a true pharmacodynamic interaction. However, pharmacodynamic interactions will never cause an apparent pharmacokinetic interaction.

Types of pharmacodynamic interaction

In order for two drugs to have a pharmacodynamic interaction, they have to share at least one common pharmacodynamic property or, more specifically, they have to share an identifiable clinical effect. Just as pharmacokinetic interactions may result in a drug concentration that is greater or smaller than expected, a pharma- codynamic interaction may result in a measurable response that is greater or smaller than expected. In general, it is assumed that each drug alone could elicit that response to some extent. However, it is conceivable that a specific effect of a drug could be enhanced or inhibited by another drug that does not have that par- ticular effect by itself, even in the absence of a pharmacokinetic interaction. Nevertheless, for most pairs of drugs that do not share a common effect, the

183 Pharmacodynamic principles and mechanisms of drug interactions

Table 9.1 Basic pharmacodynamic interactions

Additive C A B

Supra-additive Infra-additive Indifferent

C A B
C A B
C A and/or B

A, effect of drug A alone; B, effect of drug B alone;
C, combined effect of drugs A and B administered together; A B, expected sum of individual effects of drugs A and B.

pharmacodynamic interaction will be absent or indifferent. For instance, in the absence of a pharmacokinetic interaction, an antiepileptic drug (AED) is unlikely at any dose to alter the antimicrobial effect of an antibiotic, and an antibiotic is unlikely to alter the seizure protection provided by an AED. Of course, if that antibiotic is known to lower the seizure threshold by itself, it could diminish the seizure protec- tion provided by the AED and this would represent a pharmacodynamic interaction.

The various types of pharmacodynamic interaction are listed in Table 9.1. If the combined effect C of drug A and B administered together corresponds to the expected sum of the effects of drug A alone and drug B alone, the interaction is said to be additive. If the combined effect is greater than the expected sum, the interac- tion is said to be supra-additive. A supra-additive effect is also called potentiation and the terms can be considered to be synonymous. The term synergism is used by some synonymously with a supra-additive effect, but it has been argued that syn- ergism means literally that drugs just work together. Therefore, the term synergism should be used preferably for any type of pharmacodynamic interaction that is not indifferent. When the combined effect of two drugs is greater than that of each drug alone at the same concentration, but less than the expected sum of the two actions, the pharmacodynamic interaction is said to be infra-additive. As there is no other reason why a drug should be less effective in combination than when it is given alone, this type of pharmacodynamic interaction is also called antagonistic. This implies that at least one of the two drugs somehow decreases the effectiveness of the other. Antagonism may exist between two drugs with a common pharmaco- logical effect and, of course, between two drugs with an opposite pharmacological effect (for instance elevation and lowering of the seizure threshold).

These definitions raise one obvious question: what is the expected sum of the effects of two drugs that are administered together and how is it determined? The difficulty of quantifying individual and combined drug actions is the main reason why pharmacodynamic interactions are much more difficult to assess than the pharmacokinetic interactions. For most AEDs the relationship between dose and level is linear, such as the relationship between single dose and peak level, or

184 Blaise F. D. Bourgeois

(a) Drug dose (b) Drug dose or concentration

Figure 9.1 Relationship between (a) drug dose and drug level and (b) drug dose or drug concentration and the magnitude of response to the drug

between maintenance dose and steady-state level (the main exception to this rule is phenytoin). If the administered dose is doubled, this will result in a level that is twice as high. In contrast, the magnitude of the response to a drug as a function of its dose or of its concentration usually follows a sigmoid curve (see Figure 9.1). Therefore, at twice the dose or level of drug A, the magnitude of the response is not twice as high. It may be less or it may be greater. Similarly, the magnitude of the response to combined doses of two drugs that produce the same effect individually will not be twice the magnitude of the individual effect. These issues will be addressed in detail in the next chapter devoted to the methods for assessing phar- macodynamic interactions.

Clinical significance of pharmacodynamic interactions

Whenever a patient takes two or more medications simultaneously, there is the potential for some type of pharmacodynamic interaction. If, in the absence of a pharmacokinetic interaction, any clinical response to one of the drugs is enhanced or reduced by another drug, a pharmacodynamic interaction can be assumed. For the desirable primary effect of either drug, and for the desirable and undesirable secondary effect of either drug, the interaction can be additive, supra-additive or infra-additive. The clinical spectrum of possible pharmacodynamic interactions is summarized in Table 9.2:

. 1 The common primary therapeutic effect of two drugs can be enhanced when they are administered together. An obvious example would be further seizure reduction when a second AED is added to the first.

. 2 A common adverse effect of two drugs can be enhanced when they are adminis- tered together. An example would be increased sedation when a second poten- tially sedative AED is added to the first.

Drug level

Drug response

185 Pharmacodynamic principles and mechanisms of drug interactions

Table 9.2 Clinical spectrum of pharmacodynamic interactions

1 Enhancement of common (primary) therapeutic effect
2 Enhancement of common adverse effect
3 Reduction of or less than additive common therapeutic effect
4 Reduction of or less than additive common adverse effect
5 Enhancement of a therapeutic effect that is not shared by the drugs 6 Enhancement of an adverse effect that is not shared by the drugs
7 Reduction of a therapeutic effect that is not shared by the drugs
8 Reduction of an adverse effect that is not shared by the drugs

. 3 The common primary therapeutic effect can be reduced, or less than additive, when two drugs are administered together. An example would be a lack of fur- ther seizure reduction, or even an increase in seizure frequency, when two AEDs are administered together, compared with each one administered alone.

. 4 A common adverse effect of two drugs can be reduced, or less than additive, when they are administered together. For example, there may be no increase in sedation when a second potentially sedative AED is added to the first.

. 5 The therapeutic effect of a drug could be enhanced by a drug that does not by itself possess this property. For instance, the seizure protection by an AED could be enhanced by adding another drug for which no antiepileptic efficacy has been demonstrated.

. 6 Inversely, an adverse effect of a drug could be enhanced by the addition of a sec- ond drug that does not by itself cause this adverse effect. For instance, the inci- dence of liver failure could be higher when a drug is combined with other drugs that do not cause liver failure.

. 7 The therapeutic effect of a drug can be reduced after the addition of a second drug that does not share this therapeutic effect or that has an opposite effect. For example, the seizure frequency can increase when an AED is combined with a drug that potentially can lower the seizure threshold, such as certain psycho- active drugs.

. 8 Finally, an adverse effect of a drug can be reduced by the addition of a drug that does not share this side effect, or that has an opposite effect. For example, the seda- tive effect of an AED could be reduced after the addition of a psychostimulant.

Whenever two or more drugs are taken simultaneously by a patient, more than one pharmacodynamic interaction may occur, and any combination of the interactions listed in Table 9.3 is possible. Whether or not a drug combination is therapeutically more desirable than the individual drugs taken alone will ultimately not depend on a single pharmacodynamic interaction between the drugs, but on the ultimate clin- ical result of all the pharmacodynamic interactions that exist between the drugs.

186 Blaise F. D. Bourgeois

Table 9.3 Desirable and undesirable pharmacodynamic interactions

Potential advantages of drug combinations

Better effectiveness (higher-therapeutic index) Milder or absent (subthreshold) side effects Broader spectrum of seizure control

Potential disadvantages of drug combinations

Potentiation of side effects (lower-therapeutic index) or more, different side effects

Idiosyncratic toxicity Seizure exacerbation

Desirable and undesirable pharmacodynamic interactions

Desirable interactions

Potential clinical advantages and disadvantages of drug combinations over single drug therapy are listed in Table 9.3. Although enhancement of the primary thera- peutic efficacy seems to be the obvious pharmacodynamic interaction that will render a particular drug combination desirable, other considerations will have to be included in order to take into account clinical realities. For instance, in the treat- ment of epilepsy, further seizure reduction could be achieved by increasing the dose of an AED in monotherapy. However, even if the efficacy of the drug contin- ues to increase as its dose is increased, there will come a point when the patient will no longer tolerate a further dosage increase because of dose-related adverse effects. The maximal tolerated dose is an easily defined clinical therapeutic endpoint. The clinical value of a drug will be determined not only by its efficacy, but also by its tolerability. This relationship between efficacy and tolerability can be expressed as the therapeutic index (for instance the ratio between toxic dose and therapeutic dose). It can also be expressed as clinical effectiveness, which reflects both efficacy and tolerability (Deckers et al., 2000). If a drug is more efficacious at a high dose but not tolerated, it will not be more effective at a higher dose. These considera- tions that apply to increases in the dosage of a single drug also apply to the addi- tion of a second drug. In order for a combination of two drugs to be more desirable than either drug taken alone, the combination has to be more effective than either drug alone. In other words, either the combination provides better seizure protec- tion at the maximal tolerated dose, or it is better tolerated at the same level of seizure protection than either drug alone. In both cases, the combination can be said to be more effective or to have a better therapeutic index. Whether or not this is the case for a certain combination of two drugs will depend on the ultimate result of all possible pharmacodynamic interactions that occur between the two drugs. Specifically, this may be the case if the common therapeutic effect (for

187 Pharmacodynamic principles and mechanisms of drug interactions

instance seizure protection) is supra-additive and the dose-related adverse effects are additive or infra-additive, or if the therapeutic effect is additive and the dose- related adverse effects are infra-additive. If, however, the therapeutic effect and the adverse effects are both supra-additive, or both additive, the combination is unlikely to be clinically more effective than either drug taken alone.

A drug combination could be superior to either drug used alone by causing milder side effects or no side effects, even if the actual seizure protection is not better. The reason is that all AEDs share an antiepileptic effect, whereas they do not share all of their adverse effects. In addition, many side effects are dose related and may occur only once a certain dosage threshold is attained. When two AEDs are combined, a certain degree of seizure protection could be achieved at a dose of both drugs that is below their individual threshold for this specific side effect. That same degree of seizure protection with either drug alone might require a dose that is above their threshold dose. This represents a concept that is opposite to the widespread concept of high-dose monotherapy, namely low-dose polytherapy. For example, a patient may become seizure free on valproate, but only at a dose that causes thrombocy- topenia and tremor. In the same patient, topiramate alone may then fully control the seizures, but only at a dose that causes undesired weight loss or word finding difficulties. It is conceivable that this patient’s seizures might be controlled on valproate and topiramate in combination at lower doses that cause none of these side effects. This concept of low-dose polytherapy is supported by the literature analysis of Deckers et al. (1997). These authors concluded that it is the total drug load of a patient that determines the number of adverse effects, and not just the number of AEDs that the patient is taking (see section ‘Undesirable interactions’).

An obvious potential advantage of combining AEDs is a broadening of the spec- trum of activity. This applies only to patients who have more than one seizure type and in whom no single drug is fully effective against all seizure types and also well tolerated. For example, in patients with juvenile myoclonic epilepsy, the generalized tonic–clonic seizures might come under full control with either valproate, lamotrigine or topiramate, but the myoclonic seizures may persist. Inversely, clonazepam has been shown to be more effective against the myoclonic seizures than against the generalized tonic–clonic seizures in patients with juvenile myoclonic epilepsy (Obeid and Panayiotopoulos, 1989). In some patients, only a combination of clonazepam with one of the above three drugs may provide full control of the generalized tonic–clonic and the myoclonic seizures.

Undesirable interactions

There can be little doubt that one of the main disadvantages of antiepileptic com- bination therapy is an increase in the intensity or the number of side effects. In general, decreasing the number of AEDs will be associated with a decrease in side effects. This decrease in side effects involves a reduction in their severity, in their

188 Blaise F. D. Bourgeois

number, or both. Several studies have suggested that a reduction in the number of AEDs reduces the overall occurrence of side effects, in particular the sedative effects and the dose-related neurological side effects in general (Fischbacher, 1982; Bennett et al., 1983; Schmidt, 1983; Theodore and Porter, 1983; Albright and Bruni, 1985; Pellock and Hunt, 1996). Interestingly, there was little or no increase in seizure fre- quency among the patients enrolled in these studies, and a reduction in seizures was actually not uncommon. When patients undergoing a temporal lobe resection were randomized to ongoing polytherapy or reduction to carbamazepine monotherapy, the seizure recurrence rate was the same for both groups, but side effects were more common in the polytherapy group (30%) than in the monotherapy group (10%). Also, controlled monotherapy trials with some of the newer AEDs have shown lower incidence of side effects than for the same drug in add-on trials.

Deckers et al. (1997) studied the relationship between AED polytherapy and adverse effects by analyzing published data from a literature review. They intro- duced a concept that they called total AED load. This concept is based on the ratio PDD/DDD, or prescribed daily dose (PDD) divided by the usual or defined daily dose (DDD). The total antiepileptic load in a patient is then calculated as the sum of the PDD/DDD ratios for all AEDs taken by the patient. For instance, if a patient takes 1.5 times the usual dose of two different AEDs, that patient’s total drug load is three, whereas a patient taking the usual dose of one drug has a total drug load of one. This type of analysis takes into account not only the number of drugs taken by a patient, but also the total relative dosage of these drugs. In 15 selected articles, the authors found a relationship between total drug load and number of adverse effects, but they found no relationship between just the number of AEDs prescribed and these adverse effects. This finding was later confirmed in a randomized study (Deckers et al., 2001). In this study, 130 adult patients with untreated generalized tonic–clonic and/or partial seizures were randomized to equal drug loads of monotherapy with carbamazepine, 400mg/day, or combination therapy with carbamazepine 200 mg/day and valproate 300 mg/day. The study was designed to detect differences in neurotoxicity, and no such difference was found between the two groups. There was also no difference in efficacy, but this was not the primary outcome variable.

In addition to the dose-related central nervous system side effects of AEDs, there is no doubt that eliminating drugs from the regimen will eliminate the various individual and specific side effects of those drugs that are discontinued, such as excessive weight gain or tremor from valproate, behavioural problems from leve- tiracetam, or cognitive impairment from topiramate.

Exacerbation of side effects by combination therapy is not limited to central nervous system toxicity of AEDs. For instance, ammonia levels following a first dose of valproate were significantly higher than baseline in patients treated with

189 Pharmacodynamic principles and mechanisms of drug interactions

phenobarbital, phenytoin, or both, whereas ammonia levels did not differ from baseline in patients receiving no other medication (Zaccara et al., 1985). Also, the rates of fatality from valproate hepatotoxicity have been found to be substantially higher in polytherapy than in monotherapy for all age groups (Bryant and Dreifuss, 1996). In patients less than 3 years old, the rate was 1 in 618 on polytherapy, whereas there was no death in this age group among 4533 patients on monotherapy. For all ages, the rate of death was 6–7 times higher on polytherapy than on monotherapy.

At times, combining two AEDs may increase the likelihood of an idiosyncratic toxic reaction. For instance, treatment with valproate can be associated with an acute encephalopathy characterized by a change in mental status that can evolve to stupor or coma, as well as by seizure exacerbation (Sackellares et al., 1979; Marescaux et al., 1982). It has been shown that this encephalopathic reaction to valproate is more likely to occur in the presence of another AED, and it is invariably reversible after valproate is discontinued. This reaction can also subside after another AED is removed from the drug regimen, although that drug itself may never have been associated with such an encephalopathic reaction (Sackellares et al., 1979; Marescaux et al., 1982).

Antiepileptic combination therapy can at times cause seizure exacerbation. As mentioned above, a reduction in the number of AEDs has been at times found to be associated with a decrease rather than an increase in seizure frequency. Besides a spontaneous fluctuation, there are possible explanations for this observation:

(a) Seizure aggravation by AEDs, paradoxical intoxication. There is a growing body of literature supporting the notion that certain drugs can cause or aggravate certain seizures in certain types of epilepsy. This is particularly common in generalized epilepsies. For instance, carbamazepine can cause or aggravate absence seizures (Snead and Hosey et al., 1985; Liporace et al., 1994), myoclonic seizures (Shield and Saslow, 1983), seizures in patients with Lennox–Gastaut syndrome or with benign rolandic epilepsy (Corda et al., 2001). It can also aggravate or cause de novo generalized spike-wave discharges in the electroen- cephalogram (EEG) (Talwar et al., 1994). In patients with juvenile myoclonic epilepsy, myoclonic seizures have been shown to be potentially exacerbated by carbamazepine and phenytoin (Genton et al., 2000), and by lamotrigine (Biraben et al., 2000). In the Lennox–Gastaut syndrome, certain seizures can be aggra- vated by carbamazepine, phenytoin, gabapentin, vigabatrin, benzodiazepines and lamotrigine (Guerrini et al., 1999). In patients with severe myoclonic epilepsy of infancy (Dravet syndrome), seizure aggravation was reported with carbamazepine, lamotrigine and vigabatrin (Guerrini et al., 1998). The higher the number of AEDs taken by a patient, the more likely it is that one of them may actually be exacerbating certain seizures and that its elimination might lead to improved seizure control.

190 Blaise F. D. Bourgeois

(b) Pharmacodynamic antagonism. Whenever a patient takes two or more AEDs, there is a pharmacodynamic interaction. In relation to seizure protection, this interaction can be purely additive, it can be supra-additive (this represents poten- tiation), or it can be infra-additive (this represents antagonism). In case of antag- onism, one drug actually prevents or decreases the efficacy of the other drug. There is experimental and clinical evidence suggesting that antagonism may exist between AEDs. In an animal model, the combined seizure protection provided by carbamazepine and lamotrigine was found to be infra-additive (Czuczwar, S. J., personal communication, 2002). In a study of the efficacy of vigabatrin in children with infantile spasms (Elterman et al., 2001), the efficacy of vigabatrin was reduced in patients taking valproate and in those taking carbamazepine, and it was even lower in those taking valproate and carbamazepine (Shields, W. D., personal communication, 2002). Based on such evidence, it is conceivable that removing an AED involved in an antagonistic antiepileptic pharmacodynamic interaction will result in improved seizure control.

Clinical relevance of pharmacodynamic interactions: monotherapy versus

combination therapy

There has been a shift in practice regarding the treatment of epilepsy with drug com- binations or with one drug alone. After decades during which patients were treated with multiple drugs, monotherapy has been considered to be the gold standard for over 20 years (Deckers et al., 2001). More recently, the concept of rational polytherapy has been proposed and debated. The clinical significance of pharmacodynamic interactions and their advantages and disadvantages have been discussed in detail ear- lier in this chapter. There are three potential advantages of combination drug therapy:

1 better seizure control with similar or fewer side effects,
2 same seizure control with fewer side effects,
3 reduction of two or more different seizure types that respond only to different

drugs.

Clinical studies of pharmacodynamic interactions between AEDs are discussed in Chapter 13. There is a paucity of clinical studies documenting the superiority of specific AED combinations. Whether or not to use combination therapy and the selection of a combination will often have to be based on an educated guess or on careful clinical observations in each individual patient (Meinardi, 1995). Consider- ations may include the mechanism of action of the drugs, the clinical spectrum of activity, and potential pharmacokinetic interactions. It has been suggested that drugs to be combined should have different mechanisms of action which would be complementary (Perucca, 1995; Macdonald, 1996). Although elegant, this hypothesis

191 Pharmacodynamic principles and mechanisms of drug interactions

has never been proven experimentally or clinically. A literature review of data in animals and in humans was used to determine whether appropriate AED combi- nations can be selected on the basis of their mechanism of action (Deckers et al., 2000). There was some evidence that efficacy could be enhanced by combining a sodium channel blocker with a drug enhancing GABAergic inhibition, or by com- bining two gamma amino butyric acid (GABA) mimetic drugs, or by combining an alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) antagonist with an N-methyl-D-aspartate (NMDA) antagonist. At the present time, the basis for choosing a drug combination based on the mechanisms of actions is purely hypothetical and no specific combination can be recommended. When a patient has two or more different seizure types that cannot be controlled by one drug alone, two drugs can be selected according to their spectrum of efficacy. Although the absence of pharmacokinetic interactions between two drugs will certainly make it easier and safer to use them together, the interactions are known and predictable, and therefore largely correctable. Therefore, pharmacokinetic interactions should not be a reason to avoid a potentially beneficial drug combination. Finally, as discussed earlier, there are arguments in favor of the concept of low-dose polytherapy as opposed to the common practice of high-dose monotherapy. The rationale for this concept is that AEDs share an antiepileptic effect, but do not necessarily share their side effects.

In conclusion, rational polytherapy can rarely be predicted. In any given patient, a rational AED combination will have been identified if the patient does better in terms of seizure control versus side effects while taking drugs A and B together (at any doses) than the patient had done on drug A alone and on drug B alone at their respective optimal doses. There may be instances in which it would be appropriate to maintain a drug combination beyond the above definition. For instance, a patient may respond partially to a first drug and may experience further improvement after addition of a second drug, or the patient becomes seizure free after addition of the second drug, despite lack of response to the first drug. It is understandable in such a case that the patient and the physician may be reluctant to make any change.

REFERENCES

Albright P, Bruni J. Reduction of polytherapy in epileptic patients. Arch Neurol 1985; 42: 797–799.

Bennett HS, Dunlop T, Ziring P. Reduction of polypharmacy for epilepsy in an institution for the retarded. Dev Med Child Neurol 1983; 25: 735–737.

Biraben A, Allain H, Scarabin JM, et al. Exacerbation of juvenile myoclonic epilepsy with lamot- rigine. Neurology 2000; 55: 1758.

Bryant III AE, Dreifuss FE. Valproic acid hepatic fatalities. III. U.S. experience since 1986. Neurology 1996; 46: 465–469.

192 Blaise F. D. Bourgeois

Corda D, Gelisse P, Genton P, et al. Incidence of drug-induced aggravation in benign epilepsy with centrotemporal spikes. Epilepsia 2001; 42: 754–759.

Deckers CL, Hekster YA, Keyser A, et al. Reappraisal of polytherapy in epilepsy: a critical review of drug load and adverse effects. Epilepsia 1997; 38(5): 570–575.

Deckers CL, Czuczwar SJ, Hekster YA, et al. Selection of antiepileptic drug polytherapy based on mechanisms of action: the evidence reviewed. Epilepsia 2000; 41: 1364–1374.

Deckers CL, Hekster YA, Keyser A, et al. Monotherapy versus polytherapy for epilepsy: a multicenter double-blind randomized study. Epilepsia 2001; 42: 1387–1394.

Elterman RD, Shields WD, Mansfield KA, et al. Randomized trial of vigabatrin in patients with infantile spasms. Neurology 2001; 57: 1416–1421.

Fischbacher E. Effect of reduction of anticonvulsants on wellbeing. Br Med J 1982; 285: 423–424. Genton P, Gelisse P, Thomas P, et al. Do carbamazepine and phenytoin aggravate juvenile

myoclonic epilepsy? Neurology 2000; 55: 1106–1109.
Guerrini R, Belmonte A, Genton P. Antiepileptic drug-induced worsening of seizures in children.

Epilepsia 1998; 39: S2–S10.
Guerrini R, Belmonte A, Parmeggiani L, et al. Myoclonic status epilepticus following high-

dosage lamotrigine therapy. Brain Dev 1999; 21: 420–424.
Liporace JD, Sperling MR, Dichter MA. Absence seizures and carbamazepine in adults. Epilepsia

1994; 35: 1026–1028.
Macdonald R. Is there a mechanistic basis for rational polypharmacy? Epilepsy Res 1996; 11: 79–93. Marescaux C, Warter JM, Micheletti G, et al. Stuporous episodes during treatment with sodium

valproate: report of seven cases. Epilepsia 1982; 23: 297–305.
Meinardi H. Use of combined antiepileptic drug therapy. In Antiepileptic Drugs, 4th edn.

R. H. Levy, R. H. Mattson, B. S. Meldrum, eds. New York: Raven Press, 1995: 91–97.
Obeid T, Panayiotopoulos CP. Clonazepam in juvenile myoclonic epilepsy. Epilepsia 1989; 30:

603–606.
Pellock JM, Hunt PA. A decade of modern epilepsy therapy in institutionalized mentally

retarded patients. Epilepsy Res 1996; 25(3): 263–268.
Perucca E. Pharmacological principles as a basis for polytherapy. Acta Neurol Scand 1995;

162(Suppl.): 31–34.
Sackellares JC, Lee SI, Dreifuss FE. Stupor following administration of valproic acid to patients

receiving other antiepileptic drugs. Epilepsia 1979; 20: 697–703.
Schmidt D. Reduction of two-drug therapy in intractable epilepsy. Epilepsia 1983; 24: 368–376. Shield WD, Saslow E. Myoclonic, atonic, and absence seizures following institution of

carbamazepine therapy in children. Neurology 1983; 33: 1487–1489.
Snead OC, Hosey LC. Exacerbation of seizures in children by carbamazepine. New Engl J Med

1985; 313: 916–921.
Talwar D, Arora MS, Sher PK. EEG changes and seizure exacerbation in young children treated

with carbamazepine. Epilepsia 1994; 35: 1154–1159.
Theodore WH, Porter RJ. Removal of sedative-hypnotic antiepileptic drugs from the regimen of

patients with intractable epilepsy. Ann Neurol 1983; 13: 320–324.
Zaccara G, Paganini M, Campostrini R, et al. Effect of associated antiepileptic treatment on

valproate-induced hyperammonemia. Ther Drug Monit 1985; 7: 185–190.

Methods for assessing pharmacodynamic interactions

Blaise F. D. Bourgeois
Harvard Medical School, Division of Epilepsy and Clinical Neurophysiology, Children’s Hospital, Boston, MA, USA

Experimental methods

Basic principles

Overall, it is much easier to assess and quantify pharmacokinetic interactions than pharmacodynamic ones. In the case of pharmacokinetic interactions, one drug will alter the pharmacokinetics of another drug. Changes in pharmacokinetic parameters can be assessed quantitatively by single dose pharmacokinetic studies, by changes in steady-state levels, or by changes in protein binding, etc. Measuring levels and calculating pharmacokinetic parameters is relatively straightforward. Assessing a pharmacodynamic interaction between two drugs requires a valid quanti- tative measurement of a specific drug effect for the two drugs individually, as well as a quantitative measurement of the effect of the two drugs administered together. Finally, it is necessary to determine the nature of the pharmacodynamic interaction that has occurred between the two drugs. Also, before the two drugs are adminis- tered together, one has to determine for each of the two drugs the appropriate dose to be administered for an assessment of the pharmacodynamic interaction to be possible. Once the response to the two drugs given together has been measured, the interaction has to be analyzed and categorized according to its type. As discussed and defined in Chapter 9 (see Table 9.1), there are four possible types of pharma- codynamic interactions: additive, supra-additive (potentiation), infra-additive (antagonism), and indifferent. Methods have been developed that make it possible to determine the type of interaction in experimental animal models. None of these methods can be applied directly to clinical studies.

It is important to realize that determining the type of pharmacodynamic interac- tion is only the first step. Whether the pharmacodynamic interaction (for instance seizure protection by two drugs) is additive or supra-additive may be of no inter- est whatsoever unless the therapeutic relevance of this interaction can be assessed.

193

194 Blaise F. D. Bourgeois

For instance, does the fact that the combined seizure protection achieved by two drugs in combination is supra-additive compared to their individual effects have therapeutic relevance? Not necessarily. One could envision that the same seizure protection that is provided by the two drugs together could possibly be achieved by administering sufficiently high doses of either one of the two drugs alone. However, the limiting factor to progressive increases of the dose of the drugs (alone or in combination) will be the dose-related toxicity. Consequently, the therapeutic relevance of a pharmacodynamic interaction between two drugs providing seizure protection will depend not only on the nature of their antiepileptic interaction, but also on the type of their pharmacodynamic interaction in relation to their dose- related neurotoxicity. If the neurotoxic interaction is also supra-additive, it may be that the seizure protection afforded by the two drugs together at their subtoxic doses is no better than the seizure protection afforded by either drug alone at its subtoxic dose. In other words, what is really relevant about the pharmacodynamic interactions between two drugs is how and to what extent the therapeutic index of the combination differs from the individual therapeutic indices of the two drugs. As early as 1955, while discussing the concepts delineated by Loewe, Weaver et al. (1955) addressed this issue quite appropriately:

Loewe (1953) has examined the characteristics of the dose–effect relationship of combined drugs acting in an additive manner and has given attention to the meaning of the common terms which are used to describe deviations from simple additive effects. He indicates that the terms synergism and antagonism, and analogous terms for supra-additive and infra-additive effects of combined drugs are usually imaginary terms and are meaningful only when they are clearly defined … He further states that it is more important to study the ratio between the intensities of various effects of the same combination, i.e., to know whether these ratios (margins of safety or therapeutic indices) assume a larger or smaller value for the combinations than for the components.

Isobolographic analysis

The concept of the isobolographic analysis has been developed about half a cen- tury ago (Loewe, 1953; Hewlett, 1969). The isobolographic analysis is currently widely used to determine the various types of pharmacodynamic interaction in experimental animal models, in particular for antiepileptic drugs (see also Chapter 11). It is a relatively simple and accurate method, which can be represented as a diagram in which concentrations or doses of drugs a and b used in a given experi- ment are plotted. The principle and the name of the method are best understood by analyzing additive interactions (Figure 10.1-I). The effective concentrations or doses of drugs a and b administered alone are plotted as A and B, respectively. This could be, for instance, the minimal effective dose or plasma concentration (MED,

195 Methods for assessing pharmacodynamic interactions

Drug b Drug b

BB Responders

Responders

1⁄2A A Drug a Supra-additive

1⁄2B 1⁄4B

(I) Drug b

1/6B

1⁄2A 3⁄4A A Drug a Additive

(II)

Responders

Responders BB

5/6B

(III)

Figure 10.1

A Drug a

Indifferent

Isobolographic analysis of different types of pharmacodynamic interactions. The abscissa and the ordinate represent the dose or concentration of drugs a and b. A and B represent the effective doses or concentrations of drugs a and b. See text for additional explanations

Drug b

1⁄2A A Drug a Infra-additive

(IV)

MEC), or the median effective dose or concentration (ED50, EC50) against maximal electroshock (MES). If the interaction between drugs a and b is purely additive, 1⁄2 of A combined with 1⁄2 of B will achieve the same effect as A or B alone. Similarly, 3⁄4 of A and 1⁄4 of B will also achieve the same effect as A or B alone. In both cases, the plot of the doses or concentrations of drugs a and b will fall on the straight line connecting A and B. Whenever plots of effective doses or concentrations of drugs a and b administered together fall on this straight line, the pharmacodynamic interaction is additive. This additive interaction implies that 1⁄2 of A can replace 1⁄2 of B, and 1⁄4 of A can replace 1⁄4 of B. Therefore, the ‘drug bolus’ consisting of 1⁄2A plus 1⁄2B, or 3⁄4A plus 1⁄4B, is equivalent in efficacy to A or B alone. Hence the term isobolo- gram. The straight line between A and B represents the isobole for additive inter- action. Any dose or concentration pair of a and b that plots above this line will be effective (responders) and any pair that plots below this line will be ineffective (non-responders).

196 Blaise F. D. Bourgeois

If a pharmacodynamic interaction is supra-additive, the bolus of drugs a and b administered together that will be necessary to achieve efficacy may be smaller than would be expected from an additive interaction. Therefore, the line that defines the interface between responders and non-responders is a curve that bends downward below the straight isobole for additive interaction (Figure 10.1-II). For instance, 1⁄2 of A with only 1⁄6 of B may be effective. Any dose or concentration pair of a and b that plots above this downward curving line will be effective (responders) and any pair that plots below this downward curving line will be ineffective (non-responders).

Inversely, if a pharmacodynamic interaction is infra-additive, the bolus of drugs a and b administered together that will be necessary to achieve efficacy may be greater than would be expected from an additive interaction. Therefore, the line that defines the interface between responders and non-responders is a curve that bends upward above the straight isobole for additive interaction (Figure 10.1-III). For instance, 1⁄2 of A and 5⁄6 of B may be required to achieve the response provided by A or B alone. Any dose or concentration pair of a and b that plots above this upward curving line will be effective (responders) and any pair that plots below this upward curving line will be ineffective (non-responders).

Finally, an interaction can be indifferent. In this case, the drugs do not act together at all and no amount of drug b will replace any amount of drug a. Therefore, the drug combination will be effective only if the amount of drug a is A, or the amount of drug b is B (Figure 10.1-IV). If the administered amounts of drugs a and b are smaller than A and B, respectively, there will be no response.

The isobolographic analysis can be applied in at least two different ways (Figure 10.2). One application consists of plotting doses or concentrations for individual animals receiving different doses and different ratios of drugs a and b (Figure 10.2-I). On this diagram, responders and non-responders must be identified as such. In Figure 10.2-I, one can see that there are several responders whose plots fall below the isobole for additive interaction. This is evidence that this particular interaction is supra-additive. An example of such an application is provided by a study on the anticonvulsant interaction between phenytoin and phenobarbital (Masuda et al., 1981).

The isobolographic analysis can also be applied by using a single plot for values obtained from a group of animals. For instance, once the median effective dose (ED50) has been determined for drugs a and b, the ED50 can be determined for the combination of the two drugs. In order to do so, the two drugs must be adminis- tered together at increasing doses until the ED50 in the combination is determined. For this purpose, it is crucial to maintain constant ratios of doses or concentrations of drugs a to b at any level. It is probably best to choose a ratio of dose a:b that is equal to ED50a:ED50b. Such an example is provided in Figure 10.2-II. If the ED50 of the two

197 Methods for assessing pharmacodynamic interactions

Drug b B

(I)

Figure 10.2

Drug b ED50b

1⁄2

Drug a

ED50a Drug a

(II)

Two types of application of the isobolographic analysis. (I) Scatter plot of the results
from individual animals, supra-additive interaction (closed symbols represent responders,

open symbols represent non-responders). (II) Determination of median effective dose (ED50) of the combination of drugs a and b, using a dosage ratio A:B. Point 1: ED50
in case of additive interaction. Points 2 and 3: ED50 in case of supra-additive (2) and infra- additive (3) interaction. Symbols A, B, a and b as in Figure 10.1. See text for additional explanations

drugs in combination is equal to 1/2 of each drug’s respective ED50, the interaction is additive and the point will fall on the straight isobole line for additive interaction (point 1). If the interaction is supra-additive, smaller doses will be sufficient to achieve the same effect, and the ED50 of the two drugs in combination will plot below the straight isobole (point 2). Inversely, if the interaction is infra-additive, larger doses will be necessary and the combined ED50 values will plot above the straight isobole (point 3). Again, the dose ratio of drug a to b in combination does not have to be equal to the ratio of the respective ED50 or equivalent values, but the ratio must be constant throughout the dosage range used to determine the ED50 or equivalent value of the two drugs in combination.

As discussed earlier, the practical relevance of pharmacodynamic interactions may be limited to their effect on the therapeutic index. The therapeutic index of individual drugs and of drug combinations can be expressed as a ratio of a certain toxic dose or concentration divided by the effective dose or concentration, for instance TD50/ED50. The isobolographic analysis in its traditional form does not allow comparison of the therapeutic index of a drug combination with the thera- peutic indices of individual drugs. In order for that, it is necessary that first, the dose or concentration ratio of the two drugs be similar when efficacy and toxicity are measured and, secondly, that the sum of the two drug doses or concentrations be used. For this purpose, a modified version of the isobolographic analysis was

198 Blaise F. D. Bourgeois

(I)

ai
Bb

b (%) 0 20 40 60 80 100

a (%) 100 80 60 40 20 0

(II)

Figure 10.3 Modified form of the isobolographic analysis taking into account the sum of the amounts of drugs a and b (ordinate), and the relative concentrations of the drugs a and b (abscissa). A and B are equivalent effective amounts of the drugs. The curve drawn between points A and B corresponds to an additive interaction according to Eq. (10.1). Reproduced, with permission, from Bourgeois, 1986

developed (Bourgeois, 1986) (Figure 10.3). This modified version takes into account the dose or concentration ratio (abscissa) and the sum of the doses or con- centrations of the two drugs (ordinate). The sum of the doses or concentrations in case of additive interaction would then be:

ai bi A 1 ri 1 ri R

(10.1)

where ai and bi are the median effective or toxic concentrations of drugs a and b, respectively, in a given combination i, A and B are the corresponding effective or toxic concentrations of drugs a or b given alone, ri is the ratio bi/ai (concentration ratio), and R is A/B (potency ratio). When A B, the line formed by all values for additive interaction at various concentration ratios is no longer straight. An example of such an application is provided in Figure 10.4 for the combination of carba- mazepine and phenobarbital (Bourgeois and Wad, 1988). As can be seen, the inter- actions are additive for seizure protection (lower points) as well as for neurotoxicity (upper points). In this study, the therapeutic index of phenobarbital was 1.6, the therapeutic index of carbamazepine was 4.4, and the therapeutic index of the com- bination was 2.8 (higher than for phenobarbital but lower than for carbamazepine).

Sum a b

199 Methods for assessing pharmacodynamic interactions

200

150

100

0
CBZ (%) 0

25 50 75 100

PB(%)100 75 50 25 0

Figure 10.4

Median effective brain concentrations against maximal electroshock (circles) and median toxic brain concentrations (squares) for phenobarbital (PB) alone (left) and for carbamazepine (CBZ) alone (right), as well as for the sum of the two drugs in combination. Solid lines represent expected values for purely additive interaction according to Eq. (10.1), and vertical bars represent 95% confidence limits. Both the anticonvulsant and the neurotoxic interactions are purely additive. Reproduced, with permission, from Bourgeois and Wad, 1988

A different example is provided in Figure 10.5 (Bourgeois, 1988). In this case, the anticonvulsant interaction between valproate and ethosuximide (lower points) was purely additive, whereas the neurotoxic interaction (upper points) was infra- additive. In this study, the therapeutic index of valproate alone was 1.8, the therapeu- tic index of ethosuximide alone was 2.4, and the therapeutic index of the combination was 3.1, superior to both individual values.

A similar diagram expressing the effective dose or concentration (ordinate) as a function of the drug concentration ratio (abscissa) was later proposed by Levasseur et al. (1998). Their mathematical analysis also includes a quantification of the intensity of the pharmacodynamic interaction.

Other methods

Besides the isobolographic analysis, various other methods have been used to assess pharmacodynamic interactions. One of them is the fractional effective con- centration index (Elison et al., 1954; Kerry et al., 1975). The first step for this method is to determine the fractional effective concentration (FEC) for drugs a

Brain concentration ( mol/kg)

200 Blaise F. D. Bourgeois

3000

2500

2000

1500

1000

500

0
ESM (%) 0 25

VPA (%) 100

50 75 100

75 50 25 0

Figure 10.5

Median effective brain concentrations against pentylenetetrazole (circles) and median toxic brain concentrations (squares) for valproate (VPA) alone (left) and ethosuximide (ESM) alone (right), as well as for the sum of the two drugs in combination. Solid lines represent expected values for purely additive interaction according to Eq. (10.1), and vertical bars represent 95% confidence limits. The anticonvulsant interaction is purely additive, whereas the neurotoxic is clearly infra-additive. Reproduced, with permission, from Bourgeois, 1988

and b. The FEC is the ratio between the effective amount of a drug used in combi- nation with a second drug and the effective amount of the drug used alone. For instance, it could be the ratio between the median effective concentration (EC50) of the drug in the presence of the other drug, divided by the corresponding EC50 of the drug alone (FECa EC50a in combination with drug b/EC50a alone). The sum of the FEC value for drugs a and b represents the FEC index. It has been suggested that an FEC index of 0.7–1.3 can be considered to represent an additive interaction (Kerry et al., 1975). FEC index values below 0.7 are indicative of a supra-additive interaction, and FEC index values above 1.3 are indicative of infra-additive inter- actions. An example of additive interactions by FEC index is provided in Table 10.1. This FEC index analysis is based on the same study as Figure 10.4, and addresses the seizure protection and the neurotoxicity of carbamazepine and phe- nobarbital, alone and in combination (Bourgeois and Wad, 1988). The isobolo- graphic analysis had revealed a purely additive interaction for both seizure protection and neurotoxicity (Figure 10.4). Analysis of the interaction using the

Brain concentration ( mol/kg)

201 Methods for assessing pharmacodynamic interactions

Table 10.1 Fractional effective concentration (FEC) and FEC indices of phenobarbital (PB) and carbamazepine (CBZ)

FECa
PB CBZ FEC indexb

MES 39.5 0.41 97.1

Rotorod 107.5 0.68 157.9

11.3 0.45 0.86 25.2

35.5 0.32 1.00 111.3

a FEC: EC50 or TC50 in combination/EC50 or TC50 alone. b FEC index: sum of FEC values for PB and CBZ. A value

of 1.0 0.3 indicates an additive interaction, lower values being indicative of synergism and higher values indicating antagonism.

Table 10.2 FEC and FEC indices of valproate (VPA) and ethosuximide (ESM)

FECa
VPA ESM FEC indexb

PTZ 225.0 0.41 549

Rotorod 921.7 0.93 986.9

507.8 0.61 1.02 826.3

1348.6 0.67 1.60 2001.0

a FEC: EC50 or TC50 in combination/EC50 or TC50 alone.
b FEC index: sum of FEC values for VPA and ESM. A value of 1.0 0.3

indicates an additive interaction, lower values being indicative of synergism and higher values indicating antagonism.

FEC index also indicates additive interaction for efficacy and neurotoxicity, both FEC indices being 0.7 and 1.3 (Table 10.1). Another example is provided in Table 10.2. The data are from the same study as Figure 10.5, and are based on seizure protection and neurotoxicity of valproate and ethosuximide, alone and in combination (Bourgeois, 1988). The isobolographic analysis had revealed an additive anticonvulsant interaction and an infra-additive neurotoxic interaction (Figure 10.5). The FEC indices confirm these findings. The FEC index of 1.60 for neurotoxicity is considered to be in the infra-additive range.

Another method of analysis of pharmacodynamic interactions consists of admin- istering an inactive dose of one drug and determining the effect of this inactive dose on the potency of the other drug. Examples and definitions of ineffective

202 Blaise F. D. Bourgeois

doses may include the dose that is effective for 1% or less of the animals, i.e. ED1 or ED0.1 (Bourgeois et al., 1983), or a dose that is ineffective in a group of animals (for instance 25–50 animals) (Gordon et al., 1993; Borowicz et al., 1999). The ED1 or the ED0.1 can be determined by extrapolation of the probit analysis used to determine the ED50 of a compound.

When drug a is administered at a sub-effective dose together with drug b, the potency of drug b can be modified, such as a significant decrease of its ED50 compared to the administration of b alone. This has been shown to be the case for the effect of nicotinamide on the anticonvulsant activity of phenobarbital (Bourgeois et al., 1983). It has also been shown that inactive doses of phenytoin, valproate, carba- mazepine, and phenobarbital can significantly lower the ED50 of felbamate (Gordon et al., 1993), or that a sub-protective dose of melatonin can enhance the effect of carbamazepine and phenobarbital on the electroconvulsive threshold in mice (Borowicz et al., 1999). Such an effect has been considered to represent evidence of potentiation. Whether or not this is valid will be addressed in the section on ‘Methodological pitfalls’.

Methodological pitfalls

Considering how complex and elaborate the assessment of pharmacodynamic interactions can be, it is not surprising that there may be several methodological pitfalls. One potential pitfall has already been alluded to earlier in this chapter, mainly attributing a certain relevance to a given pharmacodynamic interaction where there may be none. A good example would be the finding of a supra-additive anti- convulsant interaction between two drugs being interpreted as an argument for the combined use of these two drugs. In fact, the finding may be totally irrelevant, unless it can be shown that the combination has a superior therapeutic index.

Another potential methodological pitfall is the use of drug doses to quantify a pharmacodynamic interaction between two drugs when there is also a pharmaco- kinetic interaction between the drugs. What is interpreted as a pharmacodynamic interaction may actually only be a pharmacokinetic interaction. The interaction between phenytoin and phenobarbital is a good example. It has been concluded from several studies, based on the analysis of doses, that the anticonvulsive phar- macodynamic interaction between phenytoin and phenobarbital is supra-additive (Chen and Ensor, 1954; Weaver et al., 1955; Wallin et al., 1970; Consroe et al., 1977). However, there is a pharmacokinetic interaction between the two drugs. It was shown independently in rats (Leppik and Sherwin, 1977) and in mice (Bourgeois, 1986) that brain concentrations of phenytoin following a single dose are higher in relation to the dose when phenytoin is administered with phenobarbital than when it is administered alone. If doses only are analyzed, this may lead to the

203 Methods for assessing pharmacodynamic interactions

conclusion that the interaction between the two drugs is supra-additive, because of the higher brain concentration of phenytoin when the combination is tested. In rats and in mice (Leppik and Sherwin, 1977; Bourgeois, 1986), assessment of the anticonvulsive pharmacodynamic interaction based on brain concentrations was shown to be consistent with a purely additive interaction.

Using an ineffective or sub-protective dose of one drug and measuring its effect on the potency of another drug has been presented earlier as one possible method for the assessment for pharmacodynamic interactions. In particular, it has been concluded that, if an inactive dose of a drug significantly reduces the ED50 of another drug, this represents a supra-additive interaction. The reasoning is that adding 0 to any number should not increase that number. In the case of pharma- codynamic interactions between drugs this conclusion is open to criticism and may actually be wrong. The main reason is that the relationship between dose and response (or concentration and response) is not a straight line, but a sigmoid curve (see Chapter 9, Figure 9.1). Also, based on the isobolographic analysis, 1⁄4 of the ED50 of drug a can be replaced by 1⁄4 of the ED50 of drug b, etc. Yet, it may be that 1⁄4 of the ED50 of either drug is in itself ineffective. Therefore, a dose that is inef- fective in itself is not necessarily ineffective when added to another drug, or when added to another dose of the same drug for that matter, even if the interaction is purely additive. Let us assume that the anticonvulsant ED50 of a drug is 50 mg/kg and that a dose of 10 mg/kg of that drug is found to be ineffective. However, if one adds the ineffective dose of 10mg/kg to the ED50 of 50mg/kg of the drug, the resulting dose of 60 mg/kg will protect 50% of the animals. Inversely, one could argue that, since 10 mg/kg is ineffective, 40 mg/kg should be just as effective as 50mg/kg, which is obviously not true. Although the method of assessing the change in potency of one drug by an ineffective dose of another drug may not help to distinguish between a supra-additive and an additive interaction, this approach is still valid for the study of drug combinations. The reason is that this method does allow an assessment of the effect of one drug on the therapeutic index of another drug.

Clinical methods

Basic principles

The difficulties encountered in the assessment of pharmacodynamic interactions in experimental animals are compounded when interactions are to be studied clin- ically in patients, especially for antiepileptic drugs. The isobolographic analysis can be applied clinically under certain circumstances, but it is difficult to apply to patients with epilepsy. The interaction between anesthetics has been studied in

204 Blaise F. D. Bourgeois

patients using the isobolographic analysis. For instance, the propofol–thiopental hypnotic interaction was analyzed in patients undergoing eye surgery (Vinik et al., 1999). The abolition of the ability to open the eyes on command was used as an endpoint. The ED50 was determined by probit analysis for thiopental alone, for propofol alone, and for three different dosage ratios of the two drugs. All ED50 val- ues were on the straight isobole, indicating a purely additive interaction. In this study, the FEC index analysis was also used, resulting in values equal to or close to 1.0.

Populations of patients with seizures are not homogeneous and cannot be com- pared with groups of healthy animals in whom standardized seizures are elicited, e.g. with MES or pentylenetetrazole. Therefore, values for median effective doses or for median toxic doses, or equivalent values, cannot be determined in patients with seizures, and isobolograms are therefore difficult or impossible to create. A more reproducible and quantifiable endpoint would actually be the maximal tolerated dose (MTD) or sub-toxic dose. In the end, determining the type of phar- macodynamic interaction between two antiepileptic drugs (mainly whether it is supra-additive or not) in patients may be a moot point. As stated earlier, what is of interest is not so much the type of interaction, but the practical relevance of the interaction. Translated into clinical terms, the practical relevance is whether a cer- tain combination of two antiepileptic drugs, compared to each drug alone, can provide better seizure protection with the same level of toxicity or the same degree of seizure protection with fewer side effects. Therefore, clinical studies should be designed to address this issue rather than whether the anticonvulsant interaction between two antiepileptic drugs is supra-additive or not.

Trial designs

The discussion on designs of clinical trials to study pharmacodynamic interactions between antiepileptic drugs will be based on the assumption that the studies are to be clinically relevant. They should be aimed at demonstrating that, compared with the individual drug effects, a given combination offers better seizure protection at the same level of toxicity or the same seizure protection with less toxicity. Possible study designs will be divided into four groups: optimal, probably valid, questionably valid, and invalid.

Optimal design

Among a cohort of patients with uncontrolled seizures, a group is to be identified whose seizures are not fully controlled at the MTD of drug a in monotherapy. The MTD is a dose that causes no persistent side effects and is just below a dose that does cause persistent side effects. This group of patients should be switched to monotherapy with drug b. The dose of drug b should be increased until seizures subside, or to the MTD. Those patients who do not benefit significantly from drug

205 Methods for assessing pharmacodynamic interactions

b, even at the MTD, should have drug a re-introduced while maintained on drug b. The dose of drug b may be maintained at the MTD or, if necessary, it may be low- ered somewhat in order to allow an increase in the dose of drug a. To what extent the combination of drugs a and b is superior will be determined by the percentage of patients receiving this combination that will have a 50% reduction in their seizure frequency. As can be seen, the essential component of an optimal study design is that all patients receive an appropriate monotherapy trial with both drugs before being treated with a combination.

Probably valid design

. 1 Theinitialstepherewouldalsobetoidentifypatientswhoseseizureshavefailed to come under control at the MTD of drug a. At that point, drug b is added to drug a, if necessary to the MTD. In those patients who experience a 50% reduction in seizure frequency after the addition of drug b, an attempt is made to gradually taper and discontinue drug a. To what extent the combination of drugs a and b is superior will be determined by the percentage of patients whose seizure control deteriorates as drug a is tapered or discontinued and whose seizure control improves after drug a is re-introduced.

. 2 Patients are identified whose seizures are not controlled on monotherapy at the MTD. Some may be on drug a, some on drug b, and some on drug c. In these patients, drug d is added as a second drug and the dose of drug d is increased as necessary and as tolerated. It is conceivable that, after the addition of drug d, a sub- stantial number of patients on drug b, for example, may experience a 50% seizure reduction whereas few or none of the patients taking drug a or c benefit. This would represent fairly good evidence that drugs b and d may have a favorable pharmaco- dynamic interaction profile and that they represent a desirable drug combination. An effect of drug d alone is unlikely with the present design, if a good response after the addition of drug d is not observed in patients taking drug a or c, assuming that the patient groups do not differ significantly in terms of their seizure types.

Questionably valid design

. 1 Clinical studies have been carried out for the use of designs outlined above as optimal or probably valid; the only difference being that patients were not on monotherapy and then on a combination of only two drugs, but took additional baseline antiepileptic drugs. Even though these baseline antiepileptic drugs remain constant, they introduce different variables and more potential for pharmacody- namic interactions.

. 2 Pharmacodynamic interactions between antiepileptic drugs can also be assessed by using a design based on the concept of the isobolographic analysis. As in the optimal design described above, patients are identified who have failed to

206 Blaise F. D. Bourgeois

Invalid design

respond to the MTD of drugs a or b in monotherapy. Trough serum levels of drugs a and of drug b are determined at the MTD. The patients then receive drugs a and b in combination at doses that are adjusted to achieve 1/2 of the trough serum levels that were reached at the MTDs. Theoretically, if the antiepileptic pharmacodynamic interaction between the two drugs is supra-additive, the patients as a group should experience a reduction in their seizure frequency. However, this method does not necessarily address the issue of the clinical supe- riority of the combination, i.e. whether the combination has a better efficacy to toxicity ratio than either drug alone.

At times, conclusions regarding the value of drug combinations have been drawn from studies that were not designed to properly address this question. For instance, if patients improve with the addition of drug b after failure of drug a, this cannot be interpreted as evidence that this improvement is due to a combination of the two drugs. The improvement could just as well been entirely due to the effect of drug b only, in which case it might be maintained after discontinuation of drug a. Also, the combination cannot be assessed unless the doses of the two drugs in monotherapy have been increased to the MTD. Since the combination may be at the MTD, it is possible that improved seizure control could also have been achieved at the MTD of the two drugs in monotherapy. Finally, as in experimental studies, possible pharmacokinetic interactions between drugs have to be taken into account and they must be corrected or compensated.

REFERENCES

Borowicz KK, Rafal K, Gasior M, et al. Influence of melatonin upon the protective action of con- ventional anti-epileptic drugs against maximal electroshock in mice. Eur Neuropsychopharmacol 1999; 9: 185–190.

Bourgeois BFD. Antiepileptic drug combinations and experimental background: the case of phe- nobarbital and phenytoin. Naunyn-Schmiedeberg’s Arch Pharmacol 1986; 333: 406–411.

Bourgeois BFD. Combination of valproate and ethosuximide: antiepileptic and neurotoxic interaction. J Pharmacol Exp Ther 1988; 237: 1128–1132.

Bourgeois BFD, Wad N. Combined administration of carbamazepine and phenobarbital: effect on anticonvulsant activity and neurotoxicity. Epilepsia 1988; 29: 482–487.

Bourgeois BFD, Dodson WE, Ferrendelli JA. Potentiation of the antiepileptic activity of pheno- barbital by nicotinamide. Epilepsia 1983; 23: 238–244.

Chen G, Ensor CR. A study of the anticonvulsant properties of phenobarbital and dilantin. Arch Int Pharmacodyn 1954; 100: 234–248.

Consroe P, Wolkin A. Cannabidiol – antiepileptic drug comparisons and interactions in experi- mentally induced seizures in rats. J Pharmacol Exp Ther 1977; 201: 26–32.

207 Methods for assessing pharmacodynamic interactions

Elison G, Singer S, Hitchings GH. Antagonists of nucleic acid derivatives. VIII. Synergism in combinations of biochemically related antimetabolites. J Biol Chem 1954; 208: 477–488.

Gordon R, Gels M, Wichmann J, et al. Interaction of felbamate with several other antiepileptic drugs against seizures induced by maximal electroshock in mice. Epilepsia 1993; 34: 367–371.

Hewlett PS. Measurement of the potencies of drug mixtures. Biometrics 1969; 25: 477–487. Kerry DW, Hamilton-Miller JMT, Brumfitt W. Trimethoprim and rifampicin: in vitro activities

separately and in combination. J Antimicrob Agents Chemother 1975; 1: 1417–1427.
Leppik IE, Sherwin AL. Anticonvulsant activity of phenobarbital and phenytoin in combination.

J Pharmacol Exp Ther 1977; 200: 570–575.
Levasseur LM, Delon A, Greco WR, et al. Development of a new quantitative approach for the

isobolographic assessment of the convulsant interaction between pefloxacin and theophylline

in rats. Pharmaceut Res 1998; 15: 1069–1076.
Loewe S. The problem of synergism and antagonism of combined drugs. Arzneimittelforsch

1953; 3: 285–290.
Masuda Y, Utsui Y, Shiraishi Y, et al. Evidence for a synergistic interaction between phenytoin and

phenobarbital in experimental animals. J Pharmacol Exp Ther 1981; 217: 805–811.
Vinik R, Bradley E, Kissin I. Isobolographic analysis of propofol–thiopental hypnotic interaction

in surgical patients. Anesth Analg 1999; 88: 667–670.
Wallin RF, Blackburn WH, Napoli. Pharmacological interactions of albutoin with other anticon-

vulsant drugs. Pharmacol Exp Ther 1970; 174: 276–282.
Weaver LC, Swinyard EA, Woodbury LA, et al. Studies on anticonvulsant drug combinations:

phenobarbital and diphenylhydantoin. J Pharmacol Exp Ther 1955; 113: 359–370.

Experimental studies of pharmacodynamic interactions

Stanislaw J. Czuczwar

Department of Pathophysiology, Medical University and
Isotope Laboratory, Institute of Agricultural Medicine, Jaczewskiego, Lublin, Poland

Introduction

In most patients, the therapy of newly diagnosed epilepsy is initiated with a single antiepileptic drug. Approximately 60–70% of patients may experience a reasonable seizure control with monotherapy (Sander et al., 1993; Czuczwar and Patsalos, 2001). However, monotherapy is not sufficient for the remainder of epileptic patients. Therefore, experimental background information may be helpful for an epileptologist to know what drug combinations can be considered preferentially for combination therapy or for controlled clinical trials. Animal studies evaluate the com- binations of conventional antiepileptic drugs or combinations of a conventional antiepileptic drug with a novel (or potential) antiepileptic drug. The protective effect of a drug combination may be quantified with the isobolographic method on the basis of equieffective doses of individual drugs administered alone or in combina- tion (Tallarida, 1992; Tallarida et al., 1989). An alternative method evaluates the effect of one antiepileptic drug given in sub-protective doses upon the ED50 value (the effective dose of a drug necessary to protect 50% of the animals) of another drug against experimental seizures. The ED50 value of the second drug in combi- nation with sub-protective doses of the first antiepileptic drug is compared to the control ED50 value, estimated for the second drug alone, according to the method of Litchfield and Wilcoxon (1949).

Interactions between conventional antiepileptic drugs

As already mentioned, the experimental background may provide clues regarding which drug combinations may actually have a significant therapeutic potential. There have been many experimental studies dealing with combinations of conventional antiepileptic drugs. For instance, Bourgeois (1986; 1988a, b), Bourgeois and Wad (1984), and Chez et al. (1994) studied the interactions between conventional antiepileptic drugs in two rapid and simple screening convulsive tests – the maximal

208

209 Experimental studies of pharmacodynamic interactions

electroshock and the pentylenetetrazol test. Practically, all existing conventional and novel antiepileptic drugs are effective in at least one of these tests, except for leve- tiracetam. According to Löscher and Schmidt (1988), maximal electroshock-induced seizures in rodents provide a good experimental model for generalized tonic–clonic convulsions while the pentylenetetrazol test may be regarded as a model for myoclonic seizures in humans.

On the basis of brain concentrations of phenytoin and phenobarbital, Bourgeois (1986), using the isobolographic analysis, concluded that the interaction between these antiepileptic drugs was purely additive against maximal electroshock in mice, while their neurotoxicity was infra-additive. However, because of the very poor therapeutic index of phenobarbital in this model, the therapeutic index of phenytoin alone was still better than the therapeutic index of the combination. For seizure protection, a purely additive interaction between phenytoin and phenobar- bital, based on their brain concentration in rats, was also found by Leppik and Sherwin (1977). There are other reports pointing to a synergy between these two antiepileptic drugs in rodents. However, they were based on the analysis of doses (Chen and Ensor, 1954; Weaver et al., 1955). On the other hand, an apparent syn- ergy was found between phenobarbital and phenytoin in mice and rabbits with the use of maximal electroshock, and these results were verified with both plasma and brain concentrations of the antiepileptic drugs. However, the neurotoxicity of this combination was not evaluated (Masuda et al., 1981).

Anticonvulsant efficacy and neurotoxicity of another combination of conven- tional antiepileptic drugs, carbamazepine and phenobarbital, was evaluated in mice against maximal electroshock by Bourgeois and Wad (1988). Brain concentrations of these drugs were taken into consideration. No supra-additive interaction was found. An additive effect was evident for both the anticonvulsant and the neuro- toxic activity. In another model of experimental epilepsy – penicillin-induced epileptic foci in cats – no potentiation could be demonstrated between carba- mazepine and phenobarbital (Monaco et al., 1985). Also, only additive effects were reported when valproate was combined with phenobarbital or carbamazepine in maximal electroshock test in mice. Considering the neurotoxic effects of these com- binations, additive and infra-additive interactions were evident, respectively (Bourgeois, 1988a). With the use of the same experimental approach, Chez et al. (1994) provided evidence for a supra-additive anticonvulsant interaction between valproate and diphenylhydantoin (while neurotoxicity was simply additive) that may be interpreted in terms of a potential benefit for antiepileptic treatment. Also, another combination of conventional antiepileptic drugs, valproate and ethosux- imide, was found potentially beneficial in the pentylenetetrazol test in mice (Bourgeois, 1988b). Although isobolographic analysis of effective brain concentra- tions of both drugs was indicative of an additive anticonvulsant interaction, a less

210 Stanislaw J. Czuczwar

Table 11.1 Antiepileptic drugs – mechanisms of action

Enhancement of Blockade of Na Blockade of T-type Blockade of other GABA-mediated

Antiepileptic drug channels Ca2 channels Ca2 channels events

Benzodiazepines Carbamazepine Ethosuximide Felbamate Gabapentin Lamotrigine Phenobarbital Phenytoin Tiagabine Topiramate Valproate Vigabatrin

↑ ↑↑

↑

↑↑ ↑ ↑↑

↑ ↑

↑↑

↑

↑↑ ↑ ↑ ↑↑

↑↑

Data are from (see for review) Löscher (1998), Urbañska et al. (1998), Deckers et al. (2000), and Czuczwar and Patsalos (2001). Only the mechanisms were considered which are evident within therapeutic drug concentrations.
For the influence of antiepileptic drugs on glutamate-mediated events see Table 11.3.

GABA, -amino butyric acid. ↑, effective; ↑↑, very effective.

than additive neurotoxic interaction was found. These interactions resulted in a bet- ter therapeutic index for the combined treatment than for either drug alone.

A question that has been debated is whether there might be a general rule on how to combine antiepileptic drugs based on their mechanisms of action. Accord- ing to Deckers et al. (2000), combining a sodium channel blocker (mechanisms of action of antiepileptic drugs are listed in Table 11.1) with a GABAergic drug seems more efficacious than two sodium channel blockers. Experimental data pro- vided by Czuczwar et al. (1981) seem to support such a hypothesis. These authors observed a potent enhancement of diazepam’s anti-pentylenetetrazol effect in mice by diphenylhydantoin, which is completely inactive in this seizure model. Although the plasma concentrations of these antiepileptic drugs were not measured, a phar- macokinetic mechanism does not seem probable since this very potent interaction was not observed against bicuculline- or isoniazid-induced seizures in mice (Czuczwar et al., 1981). This may also point to different mechanisms of action of conventional antiepileptic drugs, which may result in a potentiation in some mod- els of experimental epilepsy. On the other hand, some other models may require the involvement of different mechanisms.

211 Experimental studies of pharmacodynamic interactions

Interactions between conventional and newer antiepileptic drugs

Shank et al. (1994) studied the protection offered by a newer antiepileptic drug, topiramate, alone and combined with standard antiepileptic drugs, phenytoin, phenobarbital, and carbamazepine against maximal electroshock-induced seizures in mice. Topiramate was combined with a conventional antiepileptic drug at fixed ratios (0.75/0.25, 0.50/0.50, and 0.25/0.75) of their respective ED50 values. To plot a dose–response curve, multiple doses of each combination were used. The results provided evidence that the combination of topiramate with phenytoin was addi- tive in terms of anticonvulsant activity. However, a synergy was observed when topiramate was combined with either carbamazepine or phenobarbital. The sec- ond experimental approach (effect of sub-protective doses) was used to study the interactions between felbamate and carbamazepine, phenytoin, phenobarbital, or valproate against maximal electroshock in mice (Gordon et al., 1993). It was evident that all conventional antiepileptic drugs in non-effective doses in this seizure test reduced the ED50 value of felbamate (42.9 mg/kg) – carbamazepine (4 mg/kg) by 70%, phenytoin (6 mg/kg) by 60%, phenobarbital (4 mg/kg) by 45%, and valproate (150mg/kg) by 69%. It is noteworthy that the protective index of felbamate, defined as its TD50/ED50, was significantly elevated after combinations with each standard antiepileptic drug (TD50 is the dose of a drug necessary to cause neuro- toxicity in 50% of the animals). Similarly to the former studies, a pharmacokinetic mechanism was unlikely to account for the observed interaction. Conversely, doses of felbamate sub-protective against electroconvulsions failed to affect the ED50 values of carbamazepine, phenytoin, phenobarbital and valproate against maximal electroshock in mice (Borowicz et al., 2000c). This may emphasize the importance of dose ratios in the final quantitative analysis of an interaction between antiepileptic drugs. In fact, such a dose dependence was observed by Shank et al. (1994) with topiramate and conventional antiepileptic drugs. Swiader et al. (2000) combined topiramate (in sub-protective doses of 2.5 and 5 mg/kg in relation to the electroconvulsive threshold in mice) with conventional antiepileptic drugs. The convulsive test was maximal electro-shock. A possible pharmacokinetic inter- action was identified on the basis of measurements of the free-plasma concentra- tions of the antiepileptic drugs. Topiramate’s ED50 against maximal electroshock was 62.1mg/kg. The most remarkable interaction was observed when topiramate (5 mg/kg) was co-administered with carbamazepine (its ED50 value was reduced by 41%). In the case of phenobarbital and phenytoin, the ED50 reductions were 30% and 28%, respectively. Much weaker effect was observed for the combination of topiramate (5 mg/kg) with valproate (its ED50 value was decreased by only 18%). However, topiramate (5 mg/kg) elevated the free-plasma concentration of carba- mazepine by 47%. Thus, a pharmacokinetic factor is apparently responsible for the

212 Stanislaw J. Czuczwar

observed potentiation of the protective effect of carbamazepine. The free-plasma concentrations of the remaining antiepileptic drugs were not affected by topira- mate. Although the interaction of topiramate with valproate was not remarkable (but still statistically significant) in terms of the anticonvulsant activity, the com- bined treatment did not disturb motor coordination or long-term memory of mice evaluated in the chimney test and passive avoidance task, respectively. In contrast, valproate alone at its ED50 value of 248mg/kg against maximal electroshock impaired both motor performance and long-term memory (Swiader et al., 2000). In the pentylenetetrazol test in mice, pronounced anticonvulsant activity was noted when topiramate was administered together with clobazam or phenobarbital, limited and/or variable effects being observed for its combinations with valproate, primi- done, and ethosuximide (Sills et al., 1999). Another newer anti-epileptic drug, gabapentin, at a sub-protective dose of 25 mg/kg, reduced the ED50 values of major conventional antiepileptic drugs: carbamazepine (by 28%), phenytoin (by 52%), phenobarbital (by 58%), and valproate (by 28%) against maximal electroshock in mice. In no case were the free-plasma concentrations of the conventional antiepilep- tic drugs affected by gabapentin. Therefore, a pharmacokinetic interaction is not probable (Czuczwar et al., 1999). Isobolographic analysis revealed distinctly supra- additive interactions for the combinations of gabapentin with carbamazepine, valproate, phenytoin, or phenobarbital, since experimentally evaluated ED50 values were much lower than the additive ED50 values theoretically calculated from the line of additivity for the respective combinations. A pharmacokinetic interaction was at least partially involved in the interactions between gabapentin and pheno- barbital. The adverse effects of the respective drug mixtures were only additive which suggests that the combinations are potentially promising for clinical studies (Borowicz et al., 2002b). Gabapentin was also evaluated in this respect in a model of reflex epilepsy, sound-induced seizures in DBA/2 mice (De Sarro et al., 1998). At a non-protective dose of 2.5 mg/kg, gabapentin enhanced the protective activity of car- bamazepine, diazepam, phenytoin, phenobarbital, and valproate. The most remark- able potentiation of the anticonvulsant effect occurred for diazepam, phenobarbital, and valproate. In addition, the therapeutic indices of the combined treatments were better than for the respective antiepileptic drugs alone. A possible pharmacokinetic mechanism may be excluded because gabapentin did not significantly affect the plasma concentration of the antiepileptic drugs. Some combinations between newer and conventional antiepileptic drugs are listed in Table 11.2.

Interactions between newer antiepileptic drugs

Only limited experimental data are available on this issue. De Sarro et al. (1998) studied gabapentin and its combinations with felbamate or lamotrigine. However,

213 Experimental studies of pharmacodynamic interactions

Table 11.2 Interactions between conventional and novel antiepileptic drugs in the maximal electroshock-induced convulsions in mice

Novel antiepileptics
Conventional antiepileptic drug Felbamate Gabapentin Topiramate

Carbamazepine Phenobarbital Phenytoin Valproate

0 ↑↑b ↑↑a,b 0 ↑↑b ↑↑b 0 ↑↑b ↑↑b 0 ↑↑b ↑↑

Novel antiepileptic drugs were given at non-protective doses, evaluated in the threshold electroconvulsive test.
Data are from Shank et al. (1994), Borowicz et al. (2000c, 2002b), and Swiader et al. (2000). See text for the adverse potential of these combinations.
0, no interaction; ↑, positive or additive interaction; ↑↑, very potent (or supra-additive) interaction. a Pharmacokinetic interaction was found. b Isobolographic analysis was performed.

the results were not as remarkable as in the case of gabapentin combined with diazepam, phenobarbital, and valproate. Topiramate co-administered with felba- mate or tiagabine demonstrated convincing efficacy against pentylenetetrazol in mice. Combinations of topiramate with gabapentin, vigabatrin, lamotrigine, or remacemide were completely without effect in this seizure model (Sills et al., 1999). Stephen et al. (1998) tested the intriguing hypothesis of whether two drugs ineffective against pentylenetetrazol might be effective when combined. Actually, lamotrigine and topiramate, fulfilling these criteria, provided a strong protection in the pentylenetetrazol test.

Interactions of antiepileptic drugs with excitatory amino acid antagonists

N-methyl-D-aspartate receptor antagonists
Endogenous excitatory amino acids, mainly glutamate or aspartate, have been shown to play an important role in the induction of seizure activity (Meldrum, 1984). Also, clinical data indicate that a number of cases of human epilepsy are accompa- nied by elevated concentrations of excitatory amino acids in plasma (Huxtable et al., 1983; Janjua et al., 1992). In the early 1980s, intensive experimental studies were initiated on the possible anticonvulsant activity of ionotropic glutamate receptor antagonists. Results from various models of experimental epilepsy pro- vided a good deal of data confirming this hypothesis (Czuczwar and Meldrum,

214 Stanislaw J. Czuczwar

Table 11.3 Antiepileptic drugs and receptors for excitatory amino acids

Antiepileptic drug NMDA receptor AMPA/KA receptor mGluR

Benzodiazepines Carbamazepine Ethosuximide Felbamate Lamotrigine Phenobarbital Phenytoin Tiagabine Topiramate Valproate Vigabatrin

0 0 ND 0 ND 0 0 ND 0 ND 0* 0* ND 0 ND 0 ND ND ND ND 0 ND 0 ND ND ND ND

For review see Löscher (1998), Urbañska et al. (1998), Deckers et al. (2000), and Czuczwar and Patsalos (2001).
0, no effect; , inhibition of receptor-mediated events; ND, not determined.
*Inhibition of glutamate release was found in vitro but not in vivo.

1982; Croucher et al., 1982; Czuczwar et al., 1985; Smith et al., 1991; Turski et al., 1990, 1992). Ionotropic glutamate receptor antagonists block two major groups of receptors: those sensitive to N-methyl-D-aspartate (NMDA receptors) and those sen- sitive to -amino-3-hydroxy-5-methyl-isoxazole-4-propionate/kainate (AMPA/KA or non-NMDA receptors; Watkins et al., 1990). Both groups of receptors control different ion currents – excitation of NMDA receptors is associated with an influx of calcium and sodium ions into a neuron whilst non-NMDA receptors prefer- entially affect sodium-gated channels (Monaghan et al., 1989). Moreover, it has been suggested that some antiepileptic drugs interact with glutamate receptors (see Table 11.3).

Utilizing the method of Litchfield and Wilcoxon (1949), a number of NMDA or non-NMDA receptor antagonists were tested for their ability to interact with conventional antiepileptic drugs (Table 11.4). D-3-(2-carboxypiperazine-4-yl)- 1-propenyl-1-phosphonic acid (D-CPP-ene; a competitive NMDA receptor antag- onist – 1 mg/kg) considerably enhanced the protective activity of carbamazepine, diazepam, phenytoin, phenobarbital, or valproate against maximal electroshock- induced seizures in mice without any effect upon their plasma concentrations (Zarnowski et al., 1994a). Except for carbamazepine, combinations of other antiepileptic drugs with D-CPP-ene resulted in serious impairment of motor coor- dination and long-term memory. A very good correlation between the experimen- tal studies and clinical data needs to be emphasized. D-CPP-ene (as an adjuvant

215 Experimental studies of pharmacodynamic interactions

Table 11.4 Influence of NMDA or AMPA/KA receptor antagonists on the anticonvulsant activity of conventional antiepileptic drugs against maximal electroshock-induced seizures in mice

Excitatory amino acid

receptor antagonist (mg/kg) Phenobarbital Diphenylhydantoin Carbamazepine Valproate

CGP 37849 (0.25) CGP 37849 (1.0) D-CPP-ene (1.0) GYKI 52466 (5) LY 300164 (2) Memantine (0.5) NBQX (10) Procyclidine (10)

NT NT 53 53 58 50 91(NS) 51 65 70 NT NT 59 53 75 69

NT 43 66 NT 63 60 36 32 68 41 NT 55 74 59 75 75

Table data indicate reductions of the ED50 values of antiepileptic drugs (in %) after combinations with excitatory amino acid receptor antagonists. ED50s of antiepileptic drugs alone are ascribed to 100%. Excitatory amino acid receptor antagonists were given at doses ineffective upon the convulsive threshold. Data are from Czechowska et al. (1993), Pietrasiewicz et al. (1993), Zarnowski et al. (1993; 1994a, b), Borowicz et al. (1995), and Czuczwar et al. (1998c).

NS, not significant; NT, not tested.

antiepileptic drug) was also given to patients with complex partial seizures (Sveinbjornsdottir et al., 1993). This combined therapy induced a number of severe adverse reactions in epileptic patients, including poor concentration, ataxia, amnesia, and sedation. Interestingly, no therapeutic improvement with D-CPP-ene was noted, in contrast to findings in the animal study (Zarnowski et al., 1994a). A pos- sible explanation for this discrepancy is that an experimental animal model for complex partial seizures in man is the amygdala-kindled seizure model in rats (Löscher et al., 1986). NMDA receptor antagonists are not very potent in this experimental model, which may result in the poor anticonvulsive effects of D-CPP- ene in patients with complex partial seizures. Memantine or procyclidine, when combined with conventional antiepileptic drugs, considerably disturbed motor coordination and long-term memory in mice, although the protection offered by the antiepileptic drugs was potentiated (Urbañska et al., 1992; Zarnowski et al., 1994b). Some other NMDA receptor antagonists possessed much better profile of activity in this regard. For instance, D,L-(E)-2-amino-4-methyl-5-phosphono-3- pentenoate (CGP 37849) and its ethylester (CGP 39551) increased the protective action of valproate against maximal electroshock in mice, maximally by 57% and 55%, respectively. It is remarkable that the combinations with valproate were free from adverse effects upon motor performance and long-term memory, which was not the case with valproate alone at its ED50 against maximal electroshock. Again,

216 Stanislaw J. Czuczwar

no pharmacokinetic factor, at least in terms of the plasma concentration of val- proate, seems to be involved (Czechowska et al., 1993). Similar results were observed, in terms of the anticonvulsant activity, when these NMDA receptor antagonists were combined with carbamazepine, phenytoin, and phenobarbital (Pietrasiewicz et al., 1993). Only combinations with phenytoin were devoid of adverse effects. The recently studied NMDA receptor antagonist CPP and its active D( ) isomer potentiated the anti-electroshock efficacy of all four conventional antiepileptic drugs with no adverse potential being observed for carbamazepine, phenytoin, and phenobarbital (Borowicz et al., 2000a). Also, the combinations with valproate were superior to valproate alone in this respect, since valproate alone at its ED50 against maximal electroshock-induced seizures produced impair- ment of motor coordination and long-term memory. Combination with CPP revealed only motor impairment (Borowicz et al., 2000a). It seems reasonable to state that any future therapy of seizures with NMDA receptor antagonists may result in a problem of serious side effects. This was studied in detail by Löscher and Hönack (1991) who showed that amygdala-kindled rats were much more suscepti- ble to adverse activity of NMDA receptor antagonists than naive (non-epileptic) rats. Combined treatment with antiepileptic drugs together with NMDA receptor antagonists may help to partially overcome this problem, especially when there is a potent interaction in terms of anticonvulsant activity. Usually, the adjuvant antiepileptic drugs are used in lower doses than those necessary to produce a pro- tective effect per se. This procedure also leads to reductions of the ED50 values of conventional antiepileptic drugs. Some of the experimental data cited above indi- cate that a number of combinations may be actually free from undesired adverse reactions. Moreover, low-affinity NMDA receptor antagonists possess a lower- adverse effect potential. A good example is remacemide, effective against both experimental and human seizures and well tolerated by epileptic patients (Bialer et al., 1999).

AMPA/KA receptor antagonists

1-(4-Aminophenyl)-4-methyl-7,8-methylenedioxy-5H-2,3-benzodiazepine hydrochloride (GYKI 52466; a non-competitive antagonist of AMPA/KA recep- tors), at the sub-effective dose of 5 mg/kg potentiated the anticonvulsant action of carbamazepine, phenytoin, and valproate, but not that of phenobarbital, against maximal electroshock-induced seizures in mice (Borowicz et al., 1995). The GYKI 52466-induced enhancement was very significant, the respective ED50 values of these antiepileptic drugs being diminished by 64%, 59%, and 68%, respectively. The non-NMDA receptor antagonist did not affect the free-plasma concentration of the affected antiepileptic drugs. No effective combination of GYKI 52466 with the antiepileptic drugs resulted in undesirable effects. Combination of GYKI 52466

217 Experimental studies of pharmacodynamic interactions

(up to 10mg/kg) with conventional antiepileptic drugs in the pentylenetetrazol test was much less remarkable. This non-NMDA receptor antagonist proved inef- fective when combined with clonazepam, ethosuximide, and phenobarbital. Only a combination with valproate was quite effective (Czuczwar et al., 1998a). However, this combination resulted in a mnemonic effect. Promising effects were obtained with the competitive antagonist of AMPA/KA receptors, 2,3-dihydroxy- 6-nitro-7-sulfamoylbenzo(F)-quinoxaline (NBQX) at 10mg/kg against maximal electroshock-induced seizures. This excitatory amino acid receptor antagonist potentiated the protective activity of conventional antiepileptic drugs, including phenobarbital. A pharmacokinetic interaction was considered unlikely. Practically, combinations of NBQX with antiepileptic drugs did not produce side effects, the only exception being one with valproate (Zarnowski et al., 1993). A very promising substance among AMPA/KA receptor antagonists is 7-acetyl-5-(4-aminophenyl)- 8,9-dihydro-8-methyl-7H-1,3-dioxolo(4,5H)-2,3-benzodiazepine (LY 300164) which was studied in combination with conventional antiepileptic drugs against maximal electroshock, pentylenetetrazol, and amygdala-kindled seizures. In a sub- protective dose of 2 mg/kg, LY 300164 reduced the ED50 values of carbamazepine, clonazepam, phenytoin, phenobarbital, and valproate against maximal elec- troshock in mice very significantly (Czuczwar et al., 1998c; Borowicz et al., 1999). Side effects of clonazepam, phenobarbital, and valproate alone were more pro- nounced than those of the respective combinations of these antiepileptic drugs with LY 300164 (Czuczwar et al., 1998c; Borowicz et al., 1999). In the pentylenete- trazol test, LY 300164 increased the anticonvulsant-protective potential of val- proate and ethosuximide, and these combinations were free from adverse effects (Czuczwar et al., 1998b). A very potent interaction was found for LY 300164 and benzodiazepine derivatives, clonazepam and diazepam in amygdala-kindled rats. The combination of clonazepam (in a non-protective dose of 0.001 mg/kg) with LY 300164 (in a sub-protective dose of 2mg/kg) resulted in an anticonvulsant effect comparable to that provided by clonazepam alone at 0.1 mg/kg (Borowicz et al., 1999). Similar effects were observed when LY 300164 was combined with diazepam (Borowicz et al., 2000b). The combination of this benzodiazepine (at 1.25mg/kg) with LY 300164 (at 2mg/kg) provided a protection against seizures comparable to that of diazepam alone at 10–20mg/kg. Also, the combinations were devoid of adverse effects whilst diazepam alone very potently disturbed motor coordination and long-term memory in amygdala-kindled rats (Borowicz et al., 2000b). Among the remaining conventional antiepileptic drugs, only a com- bination of valproate with LY 300164 (at 2 mg/kg) resulted in protective activity against amygdala kindling (Borowicz et al., 2001). In no instance did LY 300164 affect the free-plasma concentration of antiepileptic drugs. The inter- actions of LY 300164 with conventional antiepileptic drugs in the kindling model

218 Stanislaw J. Czuczwar

Table 11.5 Combined treatment of selective antagonists of NMDA and AMPA/KA receptors, LY 235959 and LY 300164, with conventional antiepileptic drugs in amygdala-kindled seizures in rats

Protective activity Adverse effects
(mg/kg) LY 235959 LY 300164 LY 235959 LY 300164

Antiepileptic drug

Diazepam (1.25) 0 Phenobarbital (15) 0 Phenytoin (40) ↑ Carbamazepine (15) 0 Clonazepam (0.001) 0 Valproate (75) 0

↑↑ NT 0
0 NT NT 0 ↑↑ NT 0 NT NT ↑↑ NT 0 ↑↑ NT 0

LY 235959 and LY 300164 were administered intraperitoneally (i.p.) in a sub-protective dose of 2 mg/kg, 15 min prior to the convulsive test. Antiepileptic drugs were also given i.p. in sub-effective doses, diazepam, clonazepam, carbamazepine, valproate – 30 min; phenobarbital – 60 min; phenytoin – 120 min before the test.
Data are from Borowicz et al. (1999, 2000b, 2001).
0, no interaction or side-effect; ↑, positive interaction; ↑↑, very potent interaction; NT, not tested.

of epilepsy are shown in Table 11.5. Generally, AMPA/KA receptor antagonists dis- play less adverse potential than NMDA receptor antagonists (Parada et al., 1992; Danysz et al., 1994). This may be relevant in terms of their possible clinical use as adjuvant antiepileptic drugs in cases where monotherapy fails.

Ligands of metabotropic glutamate receptors

In the early 1990s, experimental evidence indicated that metabotropic glutamate (mGlu) receptors (mGluRs) participate in the generation of seizure activity (Sacaan and Schoepp, 1992; McDonald et al., 1993; Tizzano et al., 1993). It was later eluci- dated that different ligands of mGluRs were effective anticonvulsant agents. For example, (S)-4-carboxy-3-hydroxyphenylglycine (an antagonist of mGlu1a and agonist of mGlu2 receptors), administered intracerebrally, inhibited sound-induced seizures in mice. This effect could be probably ascribed to a reduced release of glu- tamate because this process seems to be controlled by mGluRs (Thomsen et al., 1994). This mGluR ligand was also effective in other experimental seizure types, including pentylenetetrazol-induced and electrically-induced convulsions. However, at effective anticonvulsant doses, the substance significantly impaired motor coor- dination (Dalby and Thomsen, 1996). A number of other agents interacting with mGluRs proved to exert anticonvulsant effects (for review see Urbañska et al., 1998). Recently, an agonist of mGlu2 receptors has been available. 2-Aminobicyclo- (3,1,0)hexane-2,6-dicarboxylate (LY 354740) has a unique property among the mGluR ligands – it can easily enter the brain after peripheral administration. The

219 Experimental studies of pharmacodynamic interactions

substance proved effective against pentylenetetrazol- and picrotoxin-induced con- vulsions and potentiated the anticonvulsant efficacy of diazepam (but not that of ethosuximide or valproate) against pentylenetetrazol. Interestingly, apart from the potentiation of the activity of diazepam, LY 354740 reduced the free-plasma con- centration of this antiepileptic drug (Klodziñska et al., 2000).

Blockade of all ionotropic receptors for glutamate – a new therapeutic possibility?

Löscher et al. (1993) were first to report on a clearly synergistic effect of NBQX combined with an NMDA receptor antagonist against amygdala-kindled seizures in rats. Also, Czuczwar et al. (1995) examined NMDA receptor antagonists (dizocilpine and D-CPP-ene) and AMPA/KA receptor antagonists (NBQX and GYKI 52466) in this regard, finding a strong interaction in terms of anticonvulsant activity. Some combinations were devoid of adverse effects (Czuczwar et al., 1995).

Interaction of antiepileptic drugs with voltage-dependent calcium channel inhibitors

There is no doubt that calcium channels are involved in the generation of seizure activity (Pumain et al., 1984; Speckmann et al., 1993). A hypothesis that voltage- dependent calcium channel inhibitors may be effective anticonvulsants was chal- lenged by Desmedt in the 1970s but it was later confirmed (Desmedt et al., 1976; De Sarro et al., 1988; Jagiello-Wójtowicz et al., 1991). This was followed by attempts to test the combinations of calcium channel inhibitors with antiepileptic drugs. Flunarizine (at a sub-protective dose of 20mg/kg) considerably decreased the ED50s of carbamazepine (by 51%) and valproate (by 54%) against electrically induced convulsions in mice. The ED50 value for phenytoin was reduced by 24%. Nimodipine was considerably weaker in this regard. None of these calcium channel blockers affected the plasma concentrations of these antiepileptic drugs and, generally, no adverse effects were observed (Czuczwar et al., 1992). Also, the anti-electroshock activity of phenytoin and carbamazepine was potentiated by nifedipine and diltiazem, but the activity of phenobarbital and valproate was not influenced (Czuczwar et al., 1990a). Interestingly, verapamil was completely inactive in this respect, both in the maximal electroshock and pentylenetetrazol test (Czuczwar et al., 1990a, b). This clearly indicates that the calcium channel inhibitor-induced hypotension is probably not involved in their interaction with antiepileptic drugs. The lack of effect of verapamil to modulate the anticonvulsant potential of antiepileptic drugs may be associated with its poor penetration through the blood–brain barrier (Hamann et al., 1983). Some conventional antiepileptic drugs

220 Stanislaw J. Czuczwar

were also affected by calcium channel inhibitors in the pentylenetetrazol test in mice. These were ethosuximide and, to a lesser degree, valproate and phenobarbital (Czuczwar et al., 1990b; Gasior et al., 1996). The combination of nimodipine with ethosuximide or valproate, however, resulted in motor impairment (Gasior et al., 1996). It is noteworthy that flunarizine, although potently increasing the protective efficacy of conventional antiepileptic drugs against electrically induced convulsions (Czuczwar et al., 1992), was completely ineffective in the pentylenetetrazol test in mice (Gasior et al., 1996). Amlodipine reduced the ED50 values of carbamazepine, phenobarbital, and valproate against maximal electroshock in mice, but the protec- tive activity of phenytoin was not affected. Since amlodipine elevated the free-plasma concentration of carbamazepine, this effect is the consequence of a pharmacokinetic interaction. Combinations of amlodipine with conventional antiepileptic drugs caused a strong motor impairment. Also, co-administration of amlodipine with phe- nobarbital or valproate resulted in a potent mnemonic effect (Kamiñski et al., 1999). In the pentylenetetrazol test, this calcium channel inhibitor enhanced the protective action of ethosuximide, phenobarbital, and valproate without affecting their plasma concentrations. Again, the combined treatment produced a considerable impair- ment of motor coordination in mice (Kamiñski et al., 2001).

Although many calcium channel inhibitors actually potentiated the anticonvulsant activity of conventional antiepileptic drugs, in many cases significant side effects were evident. In this context, experimental data may help to choose an appropriate calcium channel inhibitor for the treatment of cardiovascular diseases in epileptic patients. One has to consider that there are even certain calcium channel inhibitors, for instance niguldipine, which were shown to impair the anticonvulsant activity of carbamazepine and phenobarbital against maximal electroshock in mice or amygdala-kindled seizures in rats (Borowicz et al., 1997; 2002a). Consequently, some calcium channel inhibitors may be counteracted in epileptic patients.

Recent data by Swiader et al. (2002) indicated that flunarizine potentiated the protective activity of LY 300164 against maximal electroshock-induced convul- sions in mice, presumably via a pharmacodymanic mechanism. This combination was also free of adverse effects. Among other calcium channel inhibitors, nifedipine did not modify the anticonvulsant activity of LY 300164, while nicardipine signifi- cantly raised its free-plasma concentration. Also, flunarizine was the only calcium channel inhibitor that could be shown to enhance the anticonvulsant action of another AMPA/KA receptor antagonist, GYKI 52466 (Gasior et al., 1997).

Concluding remarks

Experimental data may provide a good background for the add-on treatment of epilepsy. It is evident from the data presented above that some combinations of

221 Experimental studies of pharmacodynamic interactions

antiepileptic drugs are promising, although the results of experimental studies can only be extrapolated with caution to the clinical setting. A considerable amount of experimental data are in agreement with what is observed in epileptic patients. A good example is D-CPP-ene and its adverse potential in rodents and epileptic patients, already discussed above (Sveinbjornsdottir et al., 1993; Zarnowski et al., 1994a). It is also worth stressing that the psychotomimetic activity of dizocilpine (a non-competitive antagonist of NMDA receptors) found in epileptic patients (Porter, 1990) was also evident in amygdala-kindled rats (Löscher and Hönack, 1991). However, it needs to be taken into consideration that antiepileptic drugs may undergo different metabolism in experimental animals and epileptic patients. For instance, topiramate was documented to increase the free-plasma concentra- tion of carbamazepine in mice (Swiader et al., 2000) while this effect was appar- ently not confirmed in epileptic patients (Bourgeois, 1996). However, most studies on the use of antiepileptic drugs are carried out acutely in rodents while epileptic patients receive a chronic antiepileptic therapy. It is widely known that carba- mazepine and phenytoin are cytochrome P450 inducers when administered chron- ically, and this may be a reason for some discrepancies. Consequently, all experimental suggestions require careful clinical verification.

According to Majkowski (1994) and Deckers et al. (2000), the new antiepileptic drugs are currently used mainly as add-on therapy. Nevertheless, rational polytherapy with new antiepileptic drugs is likely to become increasingly widespread. It will remain a challenge for pharmacologists to provide experimental data on interactions between newer antiepileptic drugs. So far, such evidence is only fragmentary. Detailed interactions between antiepileptic drugs in both experimental and clinical conditions were also reviewed by Fröscher (1994) and Deckers et al. (2000). The experimental background for the evaluation of synergistic and additive effects of antiepileptic drugs given in combination was discussed by Czuczwar (1998) and Deckers et al. (2000).

In summary, monotherapy is recommended for the treatment of epilepsy, prefer- entially among newly diagnosed patients. However, in patients who are resistant to monotherapy, combination therapy may be beneficial. Experimental studies provide evidence that a combination of two antiepileptic drugs may produce antagonistic, additive, and supra-additive (synergistic) anticonvulsant effects. A drug combination producing a supra-additive seizure protection should be of clinical interest. However, if in addition to the enhanced protective efficacy against seizures there is also supra- additive toxicity, the protective index (and hence the effectiveness of the drug combi- nation) may be equal or even inferior, when compared with each drug alone.

Two main experimental approaches for studying drug interactions exist. The isobolographic analysis may be employed when antiepileptic drugs are used at active doses against seizures. A shift of the dose–response curve for an antiepileptic drug in the presence of an adjuvant (usually in sub-protective doses) may also

222 Stanislaw J. Czuczwar

indicate which combinations to choose for clinical evaluation. Existing experimen- tal evidence points to a favorable synergistic interaction between valproate and phenytoin (or ethosuximide) or topiramate and carbamazepine (or phenobarbi- tal), or felbamate and all major antiepileptic drugs. However, the anticonvulsant potency of carbamazepine, phenytoin, phenobarbital, and valproate was not affected by felbamate at sub-protective doses against maximal electroshock in mice. This may indicate that synergism is encountered at only some drug concentration ratios. Considerable enhancement of the protective activity of conventional antiepileptic drugs by some calcium channel inhibitors and excitatory amino acid antagonists has also been demonstrated. The experimental data may be helpful for choosing drug combinations potentially beneficial in epileptic patients. However, final con- clusions have to be based on appropriate clinical trials.

REFERENCES

Bialer M, Johanessen SI, Kupferberg HJ, et al. Progress report on new antiepileptic drugs: a sum- mary of the fourth Eilat conference (EILAT IV). Epilepsy Res 1999; 34: 1–41.

Borowicz KK, Gasior M, Kleinrok Z, et al. The non-competitive AMPA/kainate receptor antago- nist, GYKI 52466, potentiates the anticonvulsant activity of conventional antiepileptics. Eur J Pharmacol 1995; 281: 319–325.

Borowicz KK, Gasior M, Kleinrok Z, et al. Influence of isradipine, niguldipine and dantrolene on the anticonvulsive action of conventional antiepileptics in mice. Eur J Pharmacol 1997; 323: 45–51. Borowicz KK, Luszczki J, Szadkowski M, et al. Influence of LY 300164, an antagonist of AMPA/

kainate receptors, on the anticonvulsant activity of clonazepam. Eur J Pharmacol 1999; 380:

67–72.
Borowicz KK, Kleinrok Z, Czuczwar SJ. Influence of D( )CPP and ( )CPP upon the protective

action of conventional antiepileptic drugs against electroconvulsions in mice. Pol J Pharmacol

2000a; 52: 431–439.
Borowicz KK, Kleinrok Z, Czuczwar SJ. The AMPA/kainate receptor antagonist, LY 300164,

increases the anticonvulsant effects of diazepam. Naunyn-Schmiedebergs Arch Pharmacol

2000b; 361: 629–635.
Borowicz KK, Stasiuk G, Teter J, et al. Low propensity of conventional antiepileptic drugs for

interaction with felbamate against maximal electroshock-induced seizures in mice. J Neural

Transm 2000c; 107: 733–743.
Borowicz KK, Kleinrok Z, Czuczwar SJ. Glutamate antagonists differentially affect the protective

activity of conventional antiepileptics against amygdala-kindled seizures in rats. Eur Neuro-

psychopharmacol 2001; 11: 61–68.
Borowicz KK, Kleinrok Z, Czuczwar SJ. Niguldipine impairs the protective activity of carba-

mazepine and phenobarbital against amygdala-kindled seizures in rats. Eur Neuropsychopharmacol 2002a; 12: 225–233.

223 Experimental studies of pharmacodynamic interactions

Borowicz KK, Swiader M, Luszczki J, et al. Effect of gabapentin on the anticonvulsant activity of antiepileptic drugs against maximal electroshock in mice – an isobolographic analysis. Epilepsia 2002b; 43: 956–963.

Bourgeois BFD. Antiepileptic drug combinations and experimental background: the case of phe- nobarbital and phenytoin. Naunyn-Schmiedebergs Arch Pharmacol 1986; 333: 406–411.

Bourgeois BFD. Anticonvulsant potency and neurotoxicity of valproate alone and in combina- tion with carbamazepine or phenobarbital. Clin Neuropharmacol 1988a; 11: 348–359.

Bourgeois BFD. Combination of valproate and ethosuximide: antiepileptic and neurotoxic interaction. J Pharmacol Exp Ther 1988b; 247: 1128–1132.

Bourgeois BFD. Drug interaction profile of topiramate. Epilepsia 1996; 37(Suppl. 2): S14–S17.

Bourgeois BFD, Wad N. Individual and combined antiepileptic and neurotoxic activity of carbamazepine and carbamazepine-10,11-epoxide in mice. J Pharmacol Exp Ther 1984; 231: 411–415.

Chen G, Ensor CR. A study of the anticonvulsant properties of phenobarbital and dilantin. Arch Int Pharmacodyn Ther 1954; 100: 234–238.

Chez MG, Bourgeois BFD, Pippenger E, et al. Pharmacodynamic interactions between pheny- toin and valproate: individual and combined antiepileptic and neurotoxic actions in mice. Clin Neuropharmacol 1994; 17: 32–37.

Croucher MJ, Collins JF, Meldrum BS. Anticonvulsant action of excitatory amino acid antago- nists. Science 1982; 216: 899–901.

Czechowska G, Dziki M, Pietrasiewicz T, et al. Competitive antagonists of NMDA receptors, CGP 37849 and CGP 39551, enhance the anticonvulsant activity of valproate against electro- convulsions in mice. Eur J Pharmacol 1993; 232: 59–64.

Czuczwar SJ. Experimental background for synergistic and additive effects of antiepileptic drugs. Epileptologia 1998; 6(suppl. 2): 21–29.

Czuczwar SJ, Meldrum BS. Protection against chemically induced seizures by 2-amino-7- phosphono-heptanoic acid. Eur J Pharmacol 1982; 83: 335–338.

Czuczwar SJ, Patsalos PN. The new generation of GABA enhancers. CNS Drugs 2001; 15: 339–350.

Czuczwar SJ, Turski L, Kleinrok Z. Diphenylhydantoin potentiates the protective effect of diazepam against pentylenetetrazol but not against bicuculline and isoniazid-induced seizures in mice. Neuropharmacology 1981; 20: 675–679.

Czuczwar SJ, Cavalheiro EA, Turski L, et al. Phosphonic analogues of excitatory amino acids raise the threshold for maximal electroconvulsions in mice. Neurosci Res 1985; 3: 86–90.

Czuczwar SJ, Chodkowska A, Kleinrok Z, et al. Effects of calcium channel inhibitors upon the effi- cacy of common antiepileptic drugs. Eur J Pharmacol 1990a; 176: 75–83.

Czuczwar SJ, Malek U, Kleinrok Z. Influence of calcium channel inhibitors upon the anticonvul- sant efficacy of common antiepileptics against pentylenetetrazol-induced convulsions in mice. Neuropharmacology 1990b; 29: 943–948.

Czuczwar SJ, Gasior M, Janusz W, et al. Influence of flunarizine, nicardipine and nimodipine on the anticonvulsant activity of different antiepileptic drugs in mice. Neuropharmacology 1992; 31: 1179–1183.

224 Stanislaw J. Czuczwar

Czuczwar SJ, Borowicz KK, Kleinrok Z, et al. Influence of combined treatment with NMDA and non-NMDA receptor antagonists on electroconvulsions in mice. Eur J Pharmacol 1995; 281: 327–333.

Czuczwar SJ, Gasior M, Kamiñski R, et al. GYKI 52466 (1-(4-aminophenyl)-4-methyl-7,8-meth- ylenedioxy-5H-2,3-benzodiazepine hydrochloride) and the anticonvulsive activity of conven- tional antiepileptics against pentetrazol in mice. Mol Chem Neuropathol 1998a; 33: 149–162.

Czuczwar SJ, Kamiñski R, Gasior M, et al. LY 300164, a novel non-NMDA receptor antagonist, potentiates the anticonvulsive activity of antiepileptic drugs. In 4th Congress of the European Society for Clinical Neuropharmacology. A. D. Korczyn, ed. Bologna: Monduzzi Editore, 1998b: 35–39.

Czuczwar SJ, Swiader M, Kuzniar H, et al. LY 300164, a novel antagonist of AMPA/kainate recep- tors, potentiates the anticonvulsive activity of antiepileptic drugs. Eur J Pharmacol 1998c; 359: 103–109.

Czuczwar SJ, Kamiñski R, Kleinrok Z, et al. Influence of gabapentin on the anticonvulsive activ- ity of conventional antiepileptic drugs in mice. Epilepsia 1999; 40(suppl. 2): 125.

Dalby NO, Thomsen C. Modulation of seizure activity in mice by metabotropic glutomate receptor ligands. J Pharmocol Exp Ther 1996; 276: 516–522.

Danysz W, Essman U, Bresink I, et al. Glutamate antagonists have different effects on sponta- neous locomotor activity in rats. Pharmacol Biochem Behav 1994; 48: 111–118.

Deckers CL, Czuczwar SJ, Hekster YA, et al. Selection of antiepileptic drug polytherapy based on mechanisms of action; the evidence reviewed. Epilepsia 2000; 41: 1364–1374.

De Sarro GB, Meldrum BS, Nistico G. Anticonvulsant effects of some calcium entry blockers in DBA/2 mice. Br J Pharmacol 1988; 93: 247–256.

De Sarro GB, Spagnolo C, Gareri P, et al. Gabapentin potentiates the antiseizure activity of cer- tain anticonvulsants in DBA/2 mice. Eur J Pharmacol 1998; 349: 179–185.

Desmedt LKC, Niemegeers CJE, Janssen PAJ. Anticonvulsant properties of cinnarizine and flunarizine in mice. Arzneim Forsch 1976; 25: 1408–1413.

Fröscher W. Synergistic effects of drug combinations. Epileptologia 1994; 2(suppl. 1): 23–32. Gasior M, Kamiñski R, Brudniak T, et al. Influence of nicardipine, nimodipine and flunarizine on the anticonvulsant efficacy of antiepileptics against pentylenetetrazol in mice. J Neural

Transm 1996; 103: 819–831.
Gasior M, Borowicz K, Kleinrok Z, et al. Anticonvulsant and adverse effects of MK-801, LY

235959, and GYKI 52466 in combination with Ca2 channel inhibitors in mice. Pharmacol

Biochem Behav 1997; 56: 629–635.
Gordon R, Gels M, Wichmann J, et al. Interaction of felbamate with several other antiepileptic

drugs against seizures induced by maximal electroshock in mice. Epilepsia 1993; 34: 367–371. Hamann SR, Todd GD, McAllister RG. The pharmacology of verapamil. Tissue distribution of

verapamil and norverapamil in rat and dog. Pharmacology 1983; 27: 1–8.
Huxtable RJ, Laird H, Lippincott SE, et al. Epilepsy and the concentration of plasma amino acids

in humans. Neurochem Int 1983; 5: 125–135.
Jagiello-Wójtowicz E, Czuczwar SJ, Chodkowska E, et al. Influence of calcium channel blockers

on pentylenetetrazol and electroshock-induced convulsions in mice. Pol J Pharmacol 1991; 43: 95–101.

225 Experimental studies of pharmacodynamic interactions

Janjua NA, Itano T, Kugoh T, et al. Familial increase in plasma glutamic acid in epilepsy. Epilepsy Res 1992; 11: 37–44.

Kamiñski R, Jasiñski M, Jagiello-Wójtowicz E, et al. Effect of amlodipine upon the protective activity of antiepileptic drugs against maximal electroshock-induced seizures in mice. Pharmacol Res 1999; 40: 319–325.

Kamiñski R, Mazurek M, Turski WA, et al. Amlodipine enhances the activity of antiepileptic drugs against pentylenetetrazole-induced seizures. Pharmacol Biochem Behav 2001; 68: 661–668.

Klodziñska A, Bijak M, Chojnacka-Wójcik E, et al. Roles of group II metabotropic glutamate receptors in modulation of seizure activity. Naunyn-Schmiedebergs Arch Pharmacol 2000; 361: 283–288.

Leppik IE, Sherwin AL. Anticonvulsant activity of phenobarbital and phenytoin in combination. J Pharmacol Exp Ther 1977; 200: 570–575.

Litchfield JT, Wilcoxon F. A simplified method of evaluating dose–effect experiments. J Pharmacol Exp Ther 1949; 96: 99–113.

Löscher W, Jäckel R, Czuczwar SJ. Is amygdala kindling in rats a model for drug-resistant partial epilepsy? Exp Neurol 1986; 93: 211–226.

Löscher W, Schmidt D. Which animal models should be used in the search for new antiepileptic drugs? A proposal based on experimental and clinical considerations. Epilepsy Res 1988; 2: 145–181.

Löscher W, Hönack D. Anticonvulsant and behavioral effects of two novel competitive N-methyl-D- aspartic acid receptor antagonists, CGP 37849 and CGP 39551, in the kindling model of epilepsy. Comparison with MK-801 and carbamazepine. J Pharmacol Exp Ther 1991; 256: 432–440.

Löscher W, Rundfeldt C, Hönack D. Low doses of NMDA receptor antagonists synergistically increase the anticonvulsant effect of the AMPA receptor antagonist NBQX in the kindling model of epilepsy. Eur J Neurosci 1993; 5: 1545–1550.

Löscher W. New visions in the pharmocology of anticonvulsion. Eur J Pharmacology 1998; 342: 1–13. Majkowski J. Interactions between new and old generations of antiepileptic drugs. Epileptologia

1994; 2(suppl. 1): 33–42.
Masuda Y, Utsui Y, Shiraishi Y, et al. Evidence for a synergistic interaction between phenytoin and

phenobarbital in experimental animals. J Pharmacol Exp Ther 1981; 217: 805–811. McDonald JW, Fix AS, Tizzano JP, et al. Seizures and brain injury in neonatal rats induced by

1S,3R-ACPD, a metabotropic glutamate receptor agonist. J Neurosci 1993; 13: 4445–4455. Meldrum B. Amino acid neurotransmitters and new approaches to anticonvulsant drug action.

Epilepsia 1984; 25: S140–S149.
Monaco F, Sechi GP, Russo A, et al. Comparison of carbamazepine and phenobarbital given in

combination in experimental epilepsy. Epilepsia 1985; 26: 103–108.
Monaghan DT, Bridges RJ, Cotman CW. The excitatory amino-acid receptors: their classes,

pharmacology and distinct properties in the function of the central nervous system. Ann Rev

Pharmacol Toxicol 1989; 29: 365–402.
Parada J, Czuczwar SJ, Turski WA. NBQX does not affect learning and memory tasks in mice: a

comparison with D-CPP-ene and ifenprodil. Cognitive Brain Res 1992; 1: 67–71. Pietrasiewicz T, Czechowska G, Dziki M, et al. Competitive NMDA receptor antagonists enhance

the antielectroshock activity of various antiepileptics. Eur J Pharmacol 1993; 250: 1–7.

226 Stanislaw J. Czuczwar

Porter RJ. New antiepileptic agents: strategies for drug development. Lancet 1990; 336: 423–424. Pumain R, Kurcewicz I, Louvel J. Fast extracellular calcium transients: involvement in epileptic

processes. Science 1984; 222: 177–179.
Sacaan AI, Schoepp DD. Activation of hippocampal metabotropic excitatory amino acid recep-

tors leads to seizures and neuronal damage. Neurosci Lett 1992; 139: 77–82.
Sander JW. Some aspects of prognosis in the epilepsies: a review. Epilepsia 1993; 34: 1007–1016. Shank RP, Gardocki JF, Vaught JL, et al. Topiramate: preclinical evaluation of a structurally novel

anticonvulsant. Epilepsia 1994; 35: 450–460.
Sills GJ, Butler E, Forrest C, et al. Combination studies with the novel anticonvulsant topiramate in

the pentylenetetrazol seizure model. Epilepsia 1999; 40(Suppl. 2): 128.
Smith SE, Durmuller N, Meldrum BS. The non-N-methyl-D-aspartate receptor antagonists,

GYKI 52466 and NBQX are anticonvulsant in two animal models of reflex epilepsy. Eur J

Pharmacol 1991; 201: 179–183.
Speckmann EJ, Stroub H, Köhling R. Contribution of calcium ions to the generation of epilep-

tic activity and antiepileptic calcium antagonism. Neuropsychobiology 1993; 27: 122–126. Stephen LJ, Sills GJ, Brodie MJ. Lamotrigine and topiramate may be a useful combination. Lancet

1998; 351: 958–959.
Sveinbjornsdottir S, Sander JWAS, Upton D, et al. The excitatory amino acid antagonist D-CPP-

ene (SDZ EAA-494) in patients with epilepsy. Epilepsy Res 1993; 16: 165–174.
Swiader M, Kotowski J, Gasior M, et al. Interaction of topiramate with conventional antiepilep-

tic drugs in mice. Eur J Pharmacol 2000; 399: 35–41.
Swiader M, Borowicz KK, Porebiak J, et al. Influence of agents affecting voltage-dependent cal-

cium channels and dantrolene on the anticonvulsant action of the AMPA/kainate receptor

antagonist LY 300164 in mice. Eur Neuropsychopharmacol 2002; 12: 311–319.
Tallarida RJ. Statistical analysis of drug combinations for synergism. Pain 1992; 49: 93–97. Tallarida RJ, Porecca F, Cowan A. Statistical analysis of drug–drug and site–site interactions with

isobolograms. Life Sci 1989; 45: 947–961.
Thomsen C, Klitgaard H, Sheardown M, et al. (S)-4-carboxy-3-hydroxyphenylglycine, an antag-

onist of metabotropic glutamate receptor (mGluR)1a and an agonist of mGluR2, protects

against audiogenic seizures in DBA/2 mice. J Neurochem 1994; 62: 2492–2495.
Tizzano JP, Griffey KI, Johnson JA, et al. Intracerebral 1S,3R-1-aminocyclopentane-1,3-dicar- boxylic acid ((1S,3R)-ACPD) produces limbic seizures that are not blocked by ionotropic glu-

tamate receptor antagonists. Neurosci Lett 1993; 162: 12–16.
Turski WA, Urbañska E, Dziki M, et al. Excitatory amino acid antagonists protect mice against

seizures induced by bicuculline. Brain Res 1990; 514: 131–134.
Turski L, Jacobsen P, Honoré T, et al. Relief of experimental spasticity and anxiolytic/anticonvulsant

actions of the -amino-3-hydroxy-5-methyl-4-isoxazole-propionate antagonist 2,3-dihydroxy-

6-nitro-7-sulfamoyl-benzo(F)-quinoxaline. J Pharmacol Exp Ther 1992; 260: 742–747. Urbañska E, Dziki M, Czuczwar SJ, et al. Antiparkinsonian drugs memantine and tri- hexyphenidyl potentiate the anticonvulsant activity of valproate against maximal electroshock-

induced seizures. Neuropharmacology 1992; 31: 1021–1026.
Urbañska EM, Czuczwar SJ, Kleinrok Z, et al. Excitatory amino acids in epilepsy (invited review).

Restor Neurol Neurosci 1998; 13: 25–39.

227 Experimental studies of pharmacodynamic interactions

Watkins JC, Krogsgaard-Larsen P, Honoré T. Structure–activity relationships in the development of excitatory amino acid receptor agonists and competitive antagonists. Trend Pharmacol Sci 1990; 11: 25–33.

Weaver LC, Swinyard EA, Woodbury LA, et al. Studied on the anticonvulsant drug combinations: phenobarbital and diphenylhydantoin. J Pharmacol Exp Ther 1955; 113: 359–370.

Zarnowski T, Kleinrok Z, Turski WA, et al. 2,3-Dihydroxy-6-nitro-7-sulfamoyl-benzo(F)quinox- aline enhances the protective activity of common antiepileptic drugs against maximal elec- troshock-induced seizures in mice. Neuropharmacology 1993; 32: 895–900.

Zarnowski T, Kleinrok Z, Turski WA, et al. The competitive NMDA antagonist, D-CPP-ene, potentiates the anticonvulsant activity of conventional antiepileptics against maximal elec- troshock-induced seizures in mice. Neuropharmacology 1994a; 33: 619–624.

Zarnowski T, Kleinrok Z, Turski WA, et al. The NMDA antagonist procyclidine, but not ifen- prodil, enhances the protective efficacy of common antiepileptics against maximal elec- troshock-induced seizures in mice. J Neural Transm 1994b; 97: 1–12.

Clinical studies of pharmacodynamic interactions

John R. Pollard and Jacqueline French Department of Neurology, University of Pennsylvania, Philadelphia, PA

Introduction

This chapter addresses the clinical impact of pharmacodynamic (PD) interactions of antiepileptic drugs (AEDs) and the strategies that have been used to discover these interactions. Particular attention is paid to the limitations of available studies and the chapter concludes with a summary of expert opinion about optimal study design for identifying PD interactions. For the purposes of this chapter, the defini- tion of a PD interaction is the interaction of two drugs causing a greater or less than expected effect or side-effect in the absence of a pharmacokinetic interaction.

Polypharmacy has undergone a renaissance since the early 1980s (Goldsmith and de Bittencourt, 1995). The old arguments against combination therapy were predi- cated upon the observation that refractory patients placed in polytherapy were experiencing increased adverse events without better efficacy (Schmidt, 1982). Since then, the advent of monitoring AED levels and a deeper understanding of the mechanisms of AED action have led to more effective use of rational polypharmacy. A combination of drugs can now be used which suppress excitation, enhance inhi- bition, and work by other novel mechanisms, thus providing a previously lacking theoretical construct for the assertion that the efficacy of combinations of drugs can be additive or supra-additive. In addition, monitoring of AED levels can limit pharmacokinetic variation that often used to cause adverse events when drugs were combined. The clinician’s goal is to identify combinations that improve effec- tiveness, a goal that could be achieved more often if natural synergies could be identified.

For any given effect or side effect, there are four possible outcomes of PD inter- actions. The first is additivity, which indicates that there is no change in the effect expected from each drug. The second is supra-additivity, a state in which a given combination results in an effect which is greater than that expected from simple additivity. The third possibility is antagonism, which is a combination that does not have at least the total effect that each medicine would be expected to have on its

228

229 Clinical studies of pharmacodynamic interactions

own. Lastly, there is aberrancy, in which a combination results in completely differ- ent effects. Alternative nomenclature of positive- or negative-PD interactions has been used in some literature. These terms imply a deviation in one direction or the other away from the additive state.

Deckers et al. (2000) define effectiveness as ‘a measure encompassing both efficacy and tolerability’, and PD interactions can affect both. This chapter will address the evidence for additivity, supra-additivity, antagonism and aberrancy for various combinations as they relate to both efficacy and tolerability. Also, an attempt will be made to summarize the experiences to date with trial design quantifying PD interactions and to highlight the designs that have the best chance of providing clinically useful knowledge. Of note, the works of Deckers et al. (2000, 2003) and Bourgeois (2002) are excellent recent reviews that summarize the results of studies relevant to PD interactions.

Positive-PD interactions: efficacy

The following section will outline trials that have provided, or have attempted to provide, relevant data on PD interactions that impact on efficacy. In assessing the validity of these studies, several issues should be considered.

Should trials be sequential or parallel?

Many of the studies discussed below have been sequential – each patient must ‘fail’ on monotherapy of one or two drugs, which are then combined, to determine whether the combination succeeds where monotherapy failed. The advantage to this approach is that each patient can be pushed to individual maximal tolerated dose. This ensures that the monotherapy was a true failure, rather than a failure to achieve the proper dose. The disadvantage is that studies designed in this way are long, leading to dropouts, which may bias the outcome.

What is the impact of drug load on PD interactions?

One major problem with many studies of AED combinations is that drug load is not taken into account. Clearly, the same adverse events would not occur when two drugs are given at high doses, as when they are combined in lower doses. Deckers has suggested that toxicity may be a result of total drug load, rather than the com- bination of two drugs per se. He uses a prescribed daily dose/defined daily dose (PDD/DDD) calculation to determine drug load. In a review, he points out that most studies of add-on therapy do not provide information about doses of back- ground drugs, making it difficult to determine total drug load (Deckers et al., 1997a). However, this concept of total drug load toxicity may not be true for all drugs. A drug that was pharmacodynamically benign might be able to be added to

230 John R. Pollard and Jacqueline French

any existing drug combination without causing problems. If a drug proves partic- ularly tolerable, this concept can be used to evaluate combinations that would allow the average patient to exceed normal drug loads (Deckers et al., 1997b).

Is the goal of the study improved efficacy or lowering of toxicity?

In most of the trials discussed below, two drugs are combined at standard doses, to produce additive efficacy. In some cases, however, the goal of combination therapy may be a reduction in toxicity rather than improved efficacy. Most AEDs demon- strate both increase in efficacy as well as toxicity as dosage increases. Even standard dosages may produce undesirable dose-related side effects. Therefore, it may be useful to demonstrate that lower than standard doses of two drugs can be com- bined to produce the efficacy of either drug at higher (and presumably less well- tolerated) doses in monotherapy. This approach would only be useful if reduced toxicity could be demonstrated, since efficacy presumably is no better than monotherapy.

What type of outcome analysis should be employed?

Seizure freedom is the ultimate goal of any epilepsy therapy. Many studies have focused on this outcome measure in combination trials. Often, seizure freedom is the only outcome measure provided. While this is useful, it may be misleading. For example, by random chance, some patients may improve while others deteriorate. In this case, reporting seizure freedom only might give an appearance of benefit, where none exists.

Definitive data supporting the presence of additive or supra-additive PD inter- actions are difficult to find and several obstacles will be illustrated in the examples below. A caveat to the following presentation of the available data concerning addi- tivity comes from Patsalos who suggests that there is a possibility that ‘some of these therapeutic enhancements result from pharmacokinetic interactions taking place in the central brain compartment, rather than as a result of PD interactions …’ (Patsalos et al., 2002). Nevertheless, for clinical purposes, any synergistic result is still important.

Add-on placebo-controlled trials

The most common studies of additive effects of AEDs are the randomized placebo- controlled add-on studies of the new AEDs. The design timeline is shown in Figure 12.1. The patients enter these trials on a variety of baseline drugs, typically with a maximum of two allowed. Increasing doses of the study drugs are employed, often leading to an incremental decrease in seizure frequency (see Figure 12.2) (Cramer et al., 1999). These studies suggest that the study drug does indeed have at least an additive effect on efficacy. Thus the entire generation of newer AEDs that were all

231 Clinical studies of pharmacodynamic interactions

Dose 2

Dose 1

Placebo

Taper and follow-up

Baseline

Titration

Treatment

Figure 12.1 Double-blind placebo-controlled trial schema

40 30 20 10

600 2400 Oxcarbazepine

1000 3000 Levetiracetam

100 200 400 Zonisamide

Efficacy of new antiepileptic drugs

Figure 12.2 Fifty per cent seizure reduction in placebo-controlled add-on trials of three new AEDs (with placebo rate subtracted) (after Cramer et al., 1999)

tested using this type of study design probably have at least an additive effect on efficacy. Unfortunately, one cannot glean specific information about which combi- nations were most effective, because of the small number of patients on each base- line AED. It is also difficult to establish definitively that the effect is truly additive, because of problems establishing the efficacy of a baseline drug a study patient received. A very conservative interpretation of these types of study would question whether these add-on studies simply show that the new drug was effective while the older baseline drug was not.

Other studies

Dean and Penry (1998) studied the combination of carbamazepine (CBZ) and valproate (VPA) using 100 patients who had failed monotherapy with CBZ. This

% of patients with 50% seizure reduction

232 John R. Pollard and Jacqueline French

study showed good success. However, it is possible that VPA monotherapy would have worked just as well, thus complicating the interpretation of these data as being supportive of an additive effect on efficacy. Of note, Harden et al. (1993) presented a smaller number of patients with a similar study design and result.

A trial of phenobarbital (PB) compared to phenytoin (PHT) and the PHT/PB combination was done in a non-randomized fashion in neonates with refractory seizures (Painter et al., 1999). This study suggested that the combination therapy made an additional 12–17% seizure free. As with the aforementioned studies, a monotherapy of the second drug was not tried, so it is not clear what the result would have been with just substituted monotherapy. A similar study design was used by Murri and Iudice (1995) in an add-on study of vigabatrin added to CBZ. There was a dropout rate of 30% but a substantial number of patients became seizure free.

Tanganelli and Regesta (1996) performed a study that used patients with newly diagnosed epilepsy, a good way to avoid the difficulties with establishing the base- line efficacy of each drug. Vigabatrin and CBZ were studied separately and in com- bination. The patients were randomly assigned to either drug and then titrated until they were either seizure free or experienced toxicity. Patients who became toxic before achieving adequate control were switched to the other medication. Combination therapy was attempted only for those patients who failed monother- apy. A total of 51/58 patients completed the study, and no data is available for the ones who did not complete. Approximately half of the patients responded well to their initial therapy. Of the non-responders 45% had good control with the cross over drug. The combination therapy had good results as well, with 5/14 patients (35%) becoming seizure free. There was no statistically significant difference between the efficacy of the two drugs, but the study was not powered to necessar- ily detect a small difference in efficacy between the drugs. This study shows an additive effect for efficacy when these two drugs are combined, because many patients who did not respond to either drug separately became seizure free with the combination. This study design was also used by Hakkaraninen (1980) using CBZ and PHT. In this study of 100 patients, presented in abstract form only, 5/33 patients (15%) who failed sequential monotherapy became seizure free on combi- nation therapy.

Walker and Koon (1988) tried a slightly different study design. They compared CBZ, VPA and the combination in series, dropping those patients who responded well from the next study arm. Again some patients became seizure free on the com- bination. This is relatively good evidence for an additive effect of the two drugs, but the data may be challenged because of the sequential study design.

In another classic add-on study, ethosuximide was added to VPA for control of absence seizures (Rowan et al., 1983). Five patients were involved in the study, and

233 Clinical studies of pharmacodynamic interactions

Week

Group

1st

2nd

3rd

4th

Latin square design

E. PB group: 3/8 and Diphenylhydantoin (DPH) group: 3/8 F. PB group: 3/4
G. DPH group: 3/4
H. Blank

Figure 12.3 Comparison of PB and DPH (after Gruber et al., 1956)

all became seizure free. Two of these patients had been refractory to ethosuximide monotherapy, so these results also support at least additivity for efficacy.

The issue of using sub-toxic doses of two drugs to reduce side effects was explored in several interesting studies. An oft-cited study by Gruber et al. (1956) compared PB and PHT in what today would be considered an unusual design, a latin square (see Figure 12.3). Patients were on their own baseline medication for 3 days of the week and then were given the study drug for 4 days. Given the long half-lives of both study drugs, it is not clear if adequate washout time was given. The study results suggested that 50 mg of either drug daily was just as efficacious as 25 mg of both drugs in combination. This study design is similar to the isobolo- grams done when studying PD interactions in animals (Chapter XI).

In patients with newly diagnosed epilepsy, Deckers et al. (2001) compared full dose CBZ, full dose VPA, and a half drug load of both. No difference was found in overall neurotoxicity or efficacy as measured by seizure frequency. It should be kept in mind, however, that newly diagnosed patients are not as sensitive to efficacy differences between regimens, and are usually responsive to lower doses of med- ication. No study arm was included with half dose of either drug alone. If we assume that half dose of either AED would translate into less effectiveness than the full dose of either, this study supports the notion that these two AEDs have an additive PD effect with respect to efficacy. This type of study using the concept of drug load may be invaluable for future studies.

Using a latin square design similar to the Gruber study mentioned above, Cereghino et al. (1975) compared CBZ, PHT and PB alone and in various combinations. The groups were not assigned randomly, but instead were divided into groups the authors thought were equivalent. As in the Gruber study, the PB arms probably were not given adequate washout time. In addition one criterion for inclusion in the study was that each patient had to be refractory to CBZ treatment, thus complicating the interpretation by raising the possibility that the CBZ was not working at all in some patients. Nonetheless, in terms of total seizure frequency, the combination of

234 John R. Pollard and Jacqueline French

all the three drugs was the superior condition for controlling seizures, while the group on combination PB and PHT had the most frequent seizures. Despite many limitations, this study may show an additive effect.

Kwan and Brodie (2000) compared add-on therapy to substitution therapy in refractory patients. In this prospective chart review, patients had similar rates of effectiveness when converted to a sequential therapy or an add-on. The authors observed ‘more patients became seizure free when the combination involved a sodium channel blocker and a drug with multiple mechanisms of action compared to other combinations.’ These data, while relatively underpowered, would support the theory that PD interaction that results in increased efficacy will likely be a result of targeting multiple points along the pathways of excitatory and inhibitory action (Goldsmith and de Bittencourt, 1995).

Certain specific combinations have been suggested as being more successful than others. Stephen et al. (1998) presented three cases where topiramate was added to lamotrigine and the patients became seizure free. There are several studies of the lamotrigine and VPA combination, and these provide the best evidence that there is a supra-additive effect from certain combinations of AEDs. The first notable study was by Brodie and Yuen (1997) (see Figure 12.4). Three hundred and forty- seven patients with any type of refractory epilepsy on monotherapy (VPA, CBZ, PHT) received add-on lamotrigine in addition to their previous drug. If patients had a 50% reduction in seizures, then the first drug was withdrawn. The lamot- rigine produced seizure reduction in a proportion of patients when it was added to each of the tested baseline medications. When the primary drug was withdrawn, seizure frequency declined slightly in the PHT and CBZ groups, possibly as a result

10 8 6 4 2

0 Baseline

Lamotrigine substitution trial

Add-on Withdrawal Study phase

Monotherapy

PHT
CBZ
Sodium VPA

Figure 12.4 Results of lamotrigine substitution for each of the above AEDs (after Brodie et al., 1997)

Median monthly seizure count

235 Clinical studies of pharmacodynamic interactions

of removal of the hepatic enzyme-inducing effect of these medications, and a resulting rise in lamotrigine levels. In contrast, when the VPA was withdrawn, there was an increase in seizure frequency, despite the fact that the lamotrigine serum levels were higher as a result of dosage adjustments. This effect suggests that there may be at least an additive effect of these two drugs and possibly even PD supra-additivity. Unfortunately, this part of the study had relatively few subjects due to substantial numbers of patients who dropped out. Of course, this dropout effect may account for the apparent improvement, as those who were doing better would be more likely to remain. Another weakness of this study is the possible selection bias of the primary treatment drug.

Kanner and Frey (2000) specifically studied the combination of lamotrigine and VPA and controlled for pharmacokinetic interactions. The study evaluated 27 patients with partial epilepsy and one with generalized epilepsy who were refractory to treatment on at least three AED. All patients were on lamotrigine monotherapy at sub-toxic doses and then had VPA added. The average seizure free duration was 6.2 months on combination but only 2.1 months on monotherapy. One limitation is that the enrolled patients were selected specifically because they were refractory to lamotrigine monotherapy. These results, from a well-controlled study, again indi- cate the possibility that an additive or even a supra-additive PD interaction may exist between these two drugs both in efficacy and side effects.

Negative-PD interactions: efficacy

Antagonistic PD interactions for efficacy exist when a combination of two medicines does not have the efficacy that each would be expected to have on its own. In the study by Brodie and Yuen (1997) described above, a group of refractory patients who were taking CBZ or PHT as primary drugs had lamotrigine added on and then the primary drug withdrawn. As noted, during the combination period, patients had more seizures than during lamotrigine monotherapy. Although this result may reflect antagonism for efficacy, it is plausible that pharmacokinetic, rather than PD interac- tions resulted in a spurious result. However, had pharmacokinetic interactions not been a factor, this study design would have been ideal for identifying PD antagonism.

A few case reports have suggested that the combination of VPA and clonazepam can induce status epilepticus, a result that could be defined as the worst case scenario for antagonistic PD interaction for efficacy. However, other studies with larger numbers of patients showed no episodes of status (Rosenberry et al., 1979; Mireles and Leppik, 1985). It is possible the surprising dearth of data showing antagonistic effects of AED on efficacy may reflect a certain reality. Some have maintained that ‘PD interactions (regarding efficacy) … are probably unidirectional

236 John R. Pollard and Jacqueline French

and lead only to increased effects’ (Reife, 1998). However, another possibility is that the proper studies to look for this type of interaction have not been done. Even drug combinations that produce improvement in many patients may produce worsening in some. Somerville et al. (2002) looked at seizure worsening in pooled data from randomized adjunctive trials. He found that more patients worsened when tiagabine was added than when placebo was added, even though tiagabine caused a greater overall seizure reduction than placebo. This would indicate a bimodal dis- tribution, with some patients improving, and others worsening. This indicates that PD interactions are not always unidirectional.

PD interactions: side effects

As noted above, side effects are often dose-related. Negative-PD interactions, also called supra-additivity for side effects, may occur when two drugs with similar side-effect profiles exceed the threshold for that side effect in combination but not individually. The possibility exists of discovering combinations of drugs that have additivity for efficacy permitting the use of doses below the threshold for side effects. A study by Lammers et al. (1995) used a quantitative assessment of adverse effects for patients on monotherapy vs. polytherapy. Interestingly, the study showed that as an aggregate measure, adverse events were no more frequent in either group. This suggests that it is possible that specific combinations of medications may offer extra efficacy without producing extra side effects. An alternative explanation for these results is that measuring the percentage of people who suffer from a given side effect may not be the best measure. Some subjects may have experienced wors- ening of side effects with the combination of medicines, but this would not have been detected by the measurements used in this study.

Several studies of specific AED combinations have demonstrated an increase in side effects. In the study by Kanner and Frey (2000) described above, the combina- tion of VPA and lamotrigine caused an increase in the number of patients com- plaining of tremor to 55%. This combination of VPA and lamotrigine also caused a notable increase in the fraction of patients experiencing tremor in a study by Pisani et al. (1999). It is unclear whether the increase was additive or supra-additive. Another example is the studies by Tanganelli and Regesta (1996) and Murri and Iudice (1995) discussed above, in which the combination of vigabatrin and CBZ led to increase in side effects such as weight gain and ataxia. As another example of a possible combination-specific interaction, in a small case series, four patients on polytherapy that included CBZ were started on levetiracetam and experienced side effects characteristic of CBZ toxicity. All the patients responded to decreasing the dose of one of the drugs, but no levels were drawn (Sisodiya et al., 2002). This interaction has not been confirmed by other investigators.

237 Clinical studies of pharmacodynamic interactions

PD interactions may also increase the likelihood of non-dose-related side effects and serious idiosyncratic reactions. Osteopenia has been reported to occur in highest incidence among patients taking more than one enzyme-inducing AED (Farhat et al., 2002). Hepatic toxicity is significantly more common in patients tak- ing valproic acid in combination therapy than in monotherapy, and this effect becomes even more pronounced in the young. The incidence of VPA-induced hepatic failure increases from 1/2000 in children under 2 years old on monother- apy, to 1/200 in those on polytherapy. The cause of this interaction is unknown (Dreifuss et al., 1987).

Another dramatic PD interaction is the development of side effects that are not described for either drug in isolation. In one descriptive paper, three patients developed new-onset chorea, and all were on a combination of PHT and lamotrig- ine (Zaatreh et al., 2001). The chorea resolved in all these patients with tapering of one medication. Although this side effect has been described for AEDs, it was unusual that this combination appeared in all three cases of chorea seen at an epilepsy clinic and represents an aberrant PD interaction for side effects.

Trial designs

The problem of designing the ideal trial to assess PD interactions has been addressed by several authors. Pledger (1989) suggests that the most straightforward and ethical design would involve a baseline medication that had no interactions with the two drugs to be studied (X and Y). All patients would be on the baseline drug and then groups would receive X, Y or X Y as add-on therapy. However, even the author notes that this study design would probably be prohibitively large. Deckers et al. (2003) suggest another paradigm that might be less costly. Patients would be evalu- ated on polytherapy while in the midst of switching monotherapies. He argues that this would provide useful clinical information, and provide information about PD interactions. Additionally since there is very little evidence for negative-PD interac- tions for efficacy, if a given combination is evaluated and proved not to have higher efficacy than the primary monotherapy, then the secondary therapy likely does not work. This would save the patient from an ineffective second monotherapy. Bourgeois (2002) suggests the optimal model would be to give drug X to maximally tolerated dose, then give drug Y to maximally tolerated dose as monotherapy, then a combination of both. While potentially valid, one must consider the likelihood of spontaneous regression/remission when analyzing such a trial.

Bourgeois has also discussed the impact of drug load, as it relates to PD interac- tions. He states that while drug load can be used in an isobologram fashion to give half doses of each drug, he considers this option suboptimal. A patient who is not tolerating maximal doses of drug X may be tried randomly in two arms of a

238 John R. Pollard and Jacqueline French

trial: half dose X half dose Y; or convert to drug Y titrated up from half dose. This type of trial was attempted by Deckers et al. (2001) but no attempt was made to discover if half dose of drug Y was as effective as the combination.

Bourgeois lists other designs as most likely valid such as: failure of drug X, improvement after addition of drug Y, and then worsening after elimination of drug X; or adding drug D to drugs X, Y, or Z and obtaining significantly better results with one of the combinations.

In summary, the total database of proven PD interactions is far from complete. To date, the best data for a potentially supra-additive effect on efficacy are for the combination of lamotrigine and VPA. Studies undertaken in the future should ideally address many of the difficulties identified above. These include: using uni- versally accepted measures of efficacy, inefficacy, and side effects; accounting for dropouts; using the concept of drug load; and performing well-controlled studies that rule out pharmacokinetic interactions. For many reasons, whether cost or ethics or unavailability of patients, we are unlikely to gain the insights into PD interactions that the perfect studies would afford.

REFERENCES

Bourgeois, BFD. Antiepileptic drug combinations: when two are better than one. Epilepsy Network News 2002; 9(1): 1–3, 6.

Brodie MJ, Yuen AWC, 105 Study Group. Lamotrigine substitution study: evidence for synergism with sodium valproate. Epilepsy Res 1997; 26: 423–432.

Cereghino JJ, Brock JT, Van Meter JC, et al. The efficacy of carbamazepine combinations in epilepsy. Clin Pharmacol Ther 1975; 18: 733–741.

Cramer JA, Fisher R, Ben-Menachem E, et al. New antiepileptic drugs: comparison of key clini- cal trials. Epilepsia 1999; 40(5): 590–600.

Dean JC, Penry JK. Carbamazepine/valproate therapy in 100 patients with partial seizures failing carbamazepine monotherapy: long term follow up. Epilepsia 1988; 29: 687.

Deckers CL, Hekster YA, Keyser A, et al. Reappraisal of polytherapy in epilepsy: a critical review of drug load and adverse effects. Epilepsia. 1997a; 38(5): 570–575.

Deckers CL, Hekster YA, Keyser A, et al. Drug load in clinical trials: a neglected factor. Clin Pharm Ther 1997b; 62: 592–595.

Deckers CLP, Czuczwar SJ, Hekster YA, et al. Selection of antiepileptic drug polytherapy based on mechanisms of action: The evidence reviewed. Epilepsia 2000; 41(11): 1364–1374.

Deckers CLP, Hekster YA, Keyser A, et al. Monotherapy versus polytherapy for epilepsy: a multi- center double-blind randomized study. Epilepsia 2001; 42(11): 1387–1394.

Deckers CLP, Genton P, Sills GJ, Schmidt D. Current limitations of antiepileptic drug therapy; a conference review. Epilepsy Res 2003; 53: 1–17.

Dreifuss FE, Santilli N, Langer DH, et al. Valproic acid hepatic fatalities: a retrospective review. Neurology 1987; 37(3): 379–385.

239 Clinical studies of pharmacodynamic interactions

Farhat G, Yamout B, Mikati MA, et al. Effect of antiepileptic drugs on bone density in ambula- tory patients. Neurology 2002; 58(9): 1348–1353.

Goldsmith P, de Bittencourt PRM. Rationalized polytherapy for epilepsy. Acta Neurol Scand Suppl 1995; 162: 35–39.

Gruber Cm, Mosier JM, Grant P, et al. Objective comparison of phenobarbital and diphenylhy- dantoin in epileptic patients. Neurology 1956; 6: 640–645.

Hakkaraninen H. Carbamazepine vs. diphenylhydantoin vs their combination in adult epilepsy. Neurology 1980; 30: 354.

Harden CL, Zisfein J, Atos-Radzion EC, et al. Combination valproate-carbamazepine therapy in partial epilepsies resistant to carbamazepine monotherapy. J Epilepsy 1993; 6(2): 91–94.

Kanner AM, Frey M. Adding valproate to lamotrigine: a study of the pharmacokinetic inter- action. Neurology 2000; 55: 588–591.

Kwan P, Brodie MJ. Epilepsy after the first drug fails: substitution or add-on? Seizure 2000; 9(7): 464–468.

Lammers MW, Hekster YA, Keyser A, et al. Monotherapy of polytherapy for epilepsy revisited: a quantitative assessment. Epilepsia 1995; 36(5): 440–446.

Mireles R, Leppik IE. Valproate and clonazepam comedication in patients with intractable epilepsy. Epilepsia 1985; 26(2): 122–126.

Murri L, Iudice A. Vigabatrin as first add-on treatment in carbamazepine-resistant epilepsy patients. Acta Neurol Scand Suppl 1995; 162: 40–42.

Painter MJ, Scher MS, Stein AD, Aet al. Phenobarbital compared with phenytoin for the treat- ment of neonatal seizures. New Engl J Med 1999; 341(7): 485–489.

Patsalos PN, Froscher W, Pisani F, et al. The importance of drug interactions in epilepsy therapy. Epilepsia 2002; 43(4): 365–385.

Pisani F, Oteri G, Russo MF, et al. The efficacy of valproate-lamotrigine comedication in refractory complex partial seizures: evidence for a pharmacodynamic interaction. Epilepsia 1999; 40(8): 1141–1146.

Pledger GW. Drug interactions in clinical trials: statistical considerations. In Antiepileptic Drug Interactions, Pitlick, WH. ed. New York: Demos, 1989.

Reife RA. Assessing pharmacokinetic and pharmacodynamic interactions in clinical trials of antiepileptic drugs. In Antiepileptic Drug Development. Advances in Neurology, Vol. 76. J. French, M. A. Dichter, eds. Philadelphia: Lippincott-Raven, 1998.

Rosenberry KR, Korberly BH, Graziani LJ. Combination of clonazepam and sodium valproate in the treatment of refractory epileptic seizures in 12 children. Am J Hosp Pharm 1979; 36(6): 736, 738. Rowan AJ, Meijer JWA, de Beer-Pawlikowski N, et al. Valproate-ethosuximide combination therapy

for refractory absence seizures. Arch Neurol 1983; 40: 797–802.
Schmidt D. Two antiepileptic drugs for intractable epilepsy with complex-partial seizures.

J Neurol Neurosur Psychiat 1982; 45: 1119–1124.
Sisodiya SM, Sander JWAS, Patsolos PN. Carbamazepine toxicity during combination therapy

with levetiracetam: a pharmacodynamic interaction. Epilepsy Res 2002; 48: 217–219. Somerville ER. Aggravation of partial seizures by antiepileptic drugs: is there evidence from clin-

ical trials? Neurology 2002; 59(1): 79–83.
Stephen LJ, Sills GJ, Brodie MJ. Lamotrigine and topiramate may be a useful combination. Lancet

1998; 351: 958–959.

240 John R. Pollard and Jacqueline French

Tanganelli P, Regesta G. Vigabatrin vs. carbamazepine monotherapy in newly diagnosed focal epilepsy: a randomized response conditional cross-over study. Epilepsy Res 1996; 25: 257–262.

Walker JE, Koon R. Carbamazepine versus valproate versus combined therapy for refractory par- tial complex seizures with secondary generalization. Epilepsia 1988; 32(5): 693.

Zaatreh M, Tennison M, D’Cruz O, et al. Anticonvulsants-induced chorea: a role for pharmaco- dynamic drug interaction? Seizure 2001; 10(8): 596–599.

Clinical studies of pharmacodynamic interactions between antiepileptic drugs and other drugs

Gaetano Zaccara1, Andrea Messori2 and Massimo Cincotta1

1 Unit of Neurology, Santa Maria Nuova Hospital, Florence, Italy 2 Drug Information Centre, Careggi Hospital, Florence, Italy

Introduction

Pharmacodynamic (PD) drug–drug interactions can occur when a patient receives concomitant treatment with two or more drugs. In general, the clinical effect resulting from PD interactions can be either advantageous or disadvantageous. A few studies in animal models have addressed the therapeutic or adverse synergistic effects of antiepileptic drugs (AEDs) (Meinardi, 1995). In humans, formal studies aiming to prove PD interactions between AEDs and other drugs are rare.

In this field, one of the most studied PD interactions is that occurring between flumazenil and benzodiazepines (BZD). Flumazenil is a specific and competitive antagonist of central BZD receptors, reversing all effects of BZD ago- nists. For this reason, incremental intravenous bolus injections of flumazenil are effective and well tolerated in the diagnosis and treatment of BZD overdose; treat- ment with flumazenil results in complete awakening with restoration of upper air- way protective reflexes (Weinbroum et al., 1997). However, withdrawal symptoms and even seizures can be observed after administration of flumazenil in long- term BZD users; these symptoms may be avoided by a slow titration of flumazenil dose.

Alcohol is another substance whose PD interactions with sedative drugs have often been studied. Sedation, which is a typical adverse effect of many AEDs, is increased by the concomitant administration of alcohol in a way that has been described in different studies as either synergistic or additive (Kastberg et al., 1998).

In this chapter, we discuss in more detail the clinical data concerning PD inter- actions of AEDs with antidepressants (ADs), antipsychotics (APs), central nervous system (CNS) stimulants, anesthetic agents, analgesics and anti-inflammatory drugs.

241

242 Gaetano Zaccara et al.

PD interactions with ADs

AEDs and ADs are often co-administered. In fact, the lifetime prevalence of major depression reported in epileptic patients is remarkably higher than in the general population (8–48% vs. 6–17%, respectively) (Lambert and Robertson, 1999). It has been hypothesized that common pathogenetic mechanisms may predispose to depression in some patients with certain types of epilepsy (Jobe et al., 1999).

Experimental and clinical data suggest that AEDs and ADs have similar mecha- nisms of action which could result in favorable and/or unfavorable PD interactions depending on the particular agents involved. Recently, biological psychiatrists have assessed the potential usefulness of AEDs in the treatment of affective disorders. Furthermore, some data suggest that ADs can have anticonvulsant and proconvul- sant properties. Finally, many drugs of these two classes share similar adverse effects which are worsened by the concomitant administration.

AEDs and affective disorders

In some patients, AEDs may precipitate mood disorders. The probability of devel- oping such adverse events is highest with the combination of barbiturates and vigabatrin (VGB) and very low with the combination of carbamazepine (CBZ) and valproate (VPA). Brent et al. (1987) found that the prevalence of depression and suicidal ideation was higher in adolescents and children taking phenobarbital (PB) than in age-matched subjects treated with CBZ. Furthermore, a meta-analysis of clinical studies performed with new AEDs shows that, in patients taking GB, the percentage of withdrawal due to depression was significantly higher than in patients treated with placebo (Marson et al., 1997).

However, the psychiatric prognosis of patients affected by epilepsy is likely to be improved by the use of AEDs. In fact, a better seizure control can have an indirect positive effect on the predisposition to mood disorders. In addition, the estab- lished positive psychotropic effects of some AEDs in non-epileptic psychiatric con- ditions suggest that AEDs could also directly improve the mood of epileptic patients (beyond their influence on seizure activity). In this context, the choice of the appropriate AED in individual patients should not merely be guided by the efficacy of the drug, but also by its AD properties and by its adverse effect profile.

Combination of AEDs and AD in the treatment of affective disorders

In an open, pivotal study, the effect of low doses of CBZ combined with low doses of amitriptyline has been evaluated in patients with major depression (Dietrich and Emrich, 1998). The particularly good results of this drug association have been postulated as a typical example of PD interaction. The authors hypothesize that the mood is regulated by two distinct groups of functional subsystems in the CNS.

243 Clinical studies of PD interactions between AEDs and other drugs

While ADs affect the most important neurotransmitter systems (which are assumed to be directly involved in endogenous depression), CBZ would affect some regulatory mechanisms between temporal cortex and amygdala which cou- ple cognition and perception with emotions. Simultaneous targeting of these two functional subsystems would cause a favorable PD interaction and a potentiation of the AD effect of these drugs.

Various new AEDs have been also used in the treatment of mood disorders. Lamotrigine (LTG) is the most widely studied and has proven efficacy in acute bipolar depression and in the long-term treatment of bipolar depression (Yatham et al., 2002). Recently, in a placebo-controlled double-blind study, LTG was added to paroxetine in depressed patients and appeared to accelerate the onset of action of the AD (Normann et al., 2002). One can therefore speculate that LTG also has favorable PD interactions with some ADs. CBZ and VPA are used for the prophy- laxis of bipolar disorders in combination with lithium. It is known that in these disorders, monotherapy is associated with a high failure rate. In contrast, the com- bination of lithium with CBZ or VPA has been reported to be highly effective (Post et al., 1996). Some double-blind and open studies have revealed that lithium and CBZ have additive effects (Kramlinger and Post, 1989). Similar results have been observed for the combination of lithium plus VPA and, in this case, a synergistic effect has been proposed (Salomon et al., 1998). These results are of particular interest because the combination of lithium and ADs gave different results. In fact, in a multi-center study which compared the efficacy of lithium, imipramine and the combination of lithium plus imipramine, the failure rate was similar for treat- ments with lithium and with lithium plus imipramine (Prien et al., 1984). Some experimental data suggest that lithium and VPA have a true positive PD inter- action; they are thought to down-regulate the expression of a protein involved in synaptic transmission which seems to be involved in stabilizing recurrent mood episodes. These two drugs act at different biochemical levels and they can therefore be synergistic (Lenox et al., 1996).

ADs effects on seizure threshold

Shortly after the introduction to the market of the tricyclic antidepressants (TCAs), seizures were reported in people taking these drugs. The most clear-cut situation in which ADs show an effect on seizure activity is overdose; in such a condition the incidence of seizures ranges from 4 to 20% with a mean overall incidence of 8.4% (Pisani et al., 1999). Maprotiline and amoxapine appear to be more frequently asso- ciated with seizures. With TCAs, seizures are reported in 3–8% of cases. Finally, the cumulative evidence from published reports shows that selective serotonin- reuptake inhibitors (SSRIs) are much less likely to cause seizures in overdose and that trazodone is the safest agent in this respect (Alldredge, 1999).

244 Gaetano Zaccara et al.

Table 13.1 Incidence of seizures induced by AD drugs

Drug Dose (mg/day) Seizure incidence (%)

Imipramine

Amitriptyline Bupropion

Clomipramine Maprotiline Fluoxetine Fluvoxamine Viloxazine

High ( 200) 0.6 Moderate (50–600) 0.3 Low ( 200) 0.1 High ( 200) 0.06 Moderate–low ( 200) 0.00 High ( 450) 2.19 Moderate–low ( 450) 0.44 Wide range 0.5 Wide range 0.4 20–60 0.2 100 0.2 150–800 0.13

Source: From Pisani et al. (1999), with some modifications.

The incidence of seizures occurring with therapeutic doses of ADs varies from 0.1 to 4% (Pisani et al., 1999). This incidence differs slightly from the annual inci- dence of first seizures in the general population (which has been estimated to be from 0.073 to 0.086%). However, under these circumstances, a clear dose-related effect has been observed for some ADs. For example, Peck et al. (1983), through the analysis of almost one hundred studies on imipramine, found that the overall inci- dence of seizures was 0.33%. However, seizures occurred in 0.10% of patients when the drug was prescribed at doses of 200mg/day or less and in 0.63% of patients treated with doses greater than 200 mg/day. The incidence of AD-related seizures for some ADs is reported in Table 13.1.

Interestingly, some AEDs may also have similar proconvulsant characteristics, particularly in overdose. For example, an increased frequency of partial seizures can be the primary manifestation of intoxication with CBZ or phenytoin (PHT) (Perucca et al., 1998). CBZ, which has the chemical structure of a TCA, may worsen epilepsy in several conditions even at therapeutic dosages. In particular, this agent may precipitate or exacerbate a variety of seizures in patients with gen- eralized epilepsies. Similar paradoxical proconvulsant effects have also been described, although less frequently, with other traditional and new AEDs (Perucca et al., 1998).

In spite of frequent observations of seizures induced by ADs in non-epileptic patients, the few studies in which an AD has been administered to epileptic patients show that the seizure control was improved in most cases (Alldredge, 1999). This effect might be secondary to an attenuation of emotional triggers for seizures or to

245 Clinical studies of PD interactions between AEDs and other drugs

enhancing the effectiveness of concomitant AED therapy through pharmacokinetic interactions. However, direct anticonvulsant effects of ADs have been shown in ani- mal as well as in some human studies (Alldredge, 1999). In a small double-blind cross-over study, imipramine at a dose of 25 mg/day was effective in the treatment of absences and myoclonic–astatic seizures (Hurst, 1984). In a more recent add-on open-label study, 17 non-depressed patients with drug-resistant complex partial seizures were treated with fluoxetine (Favale et al., 1995). Six patients became seizure free for 8 months, while the remaining patients experienced an average 30% reduction in seizure frequency. An effect against partial seizures has also been reported with doxepin (Pisani et al., 1999). For a more detailed review, see Alldredge (Alldredge, 1999). All of these data suggest that some TCAs and some SSRIs, at a certain dose, may exert an inhibitory action on neural excitability. It seems that the most important factor in determining the direction of a given AD in terms of inhibition or excitation is drug dosage. It would be interesting to explore possible favorable interactions between AEDs and ADs on different epileptic syndromes.

Adverse effects

PD interactions can also cause the appearance or the worsening of some adverse effects. Sedation may be particularly troublesome in patients taking AEDs, particularly barbiturates or BZD. This adverse effect can be aggravated by the co-administration of most of the older ADs, especially TCAs, mianserin, trazodone and mirtazapine (Lambert and Robertson, 1999). Patients with epilepsy often complain of memory disturbances and some AEDs, such as barbiturates and topiramate (TPM) are known to have deleterious effects on memory. The association of these drugs with older TCAs (especially amitriptyline, which has strong anticholinergic effects), mianserin, and trazodone has been found to produce cognitive impairment and therefore should be avoided (Lambert and Robertson, 1999).

Theoretically, monoamine oxidase (MAO) inhibitors should not be co-admin- istered with CBZ because this may precipitate a hypertensive crisis. However, this event has not been observed in practice. In contrast, a case has been described of a toxic serotonin syndrome attributed to the concomitant use of fluoxetine and CBZ in a patient with an affective disorder (Lambert and Robertson, 1999). Finally, CBZ and, more frequently, oxcarbazepine have been associated with hyponatremia. This meta- bolic effect has also been documented in patients taking SSRI (Bouman et al., 1998). Therefore, attention should be paid when SSRI are co-administered with CBZ or oxcarbazepine, particularly in elderly patients also treated with diuretic drugs.

In summary ADs and AEDs share several clinical effects. The factors which determine the direction of the effect (pro- or anticonvulsant) may be the dosage of drugs and the epileptic syndrome. PD and also pharmacokinetic interactions (some AEDs induce the metabolism of ADs and in turn are inhibited by these

246 Gaetano Zaccara et al.

drugs) might potentiate or change the direction of these effects and therefore should be investigated.

PD interactions with AP drugs

AP and AEDs are frequently co-administered and PD interactions concerning their effects on psychosis and seizure threshold are possible. Epidemiological studies have identified a variety of psychoses in about 7–8% of patients with epilepsy. The risk for this adverse effect seems to be higher in patients with temporal lobe epilepsy. In a prospective study of psychosis and epilepsy, children with temporal lobe epilepsy had a 10% chance of developing interictal psychoses during a 30-year follow-up compared with a mean incidence of psychosis of 0.8% in the general population. On the other hand, patients with schizophrenia appear to be more prone to seizures than the general population. This vulnerability can be related both to neuropatho- logic substrate of schizophrenia and to the exposure to psychotropic medications that lower the seizure threshold (Torta and Keller, 1999).

AEDs and psychosis

Neurobiological hypotheses of epileptic psychoses are focused on the neuropatho- logic alterations observed in epilepsy and on the neurophysiologic modifications of various neurotransmitter systems (particularly 3,4-dihydroxyphenylalanine, DOPA) induced by the epileptic discharge (Torta and Keller, 1999). In general, the overall psychiatric prognosis of epilepsy is thought to be improved by the use of AEDs. However, in many situations, interictal psychoses can be induced or aggra- vated by some AEDs. Mechanisms related to these adverse events are represented by forced normalization1, folate deficency, drug toxicity and abrupt withdrawal of a drug. Ethosuximide is associated with forced normalization and psychosis both in children and in adults (Torta and Keller, 1999). Psychoses are also described with PHT when serum level is above 35 mg/l (McDanal and Bolman, 1975). Among the new AEDs, VGB and TPM are more frequently responsible for psychotic distur- bances. In patients included in controlled clinical studies, the incidence of this complication was 3.4% with VGB (Ferrie et al., 1996) (which is higher than the incidence of psychosis in patients treated with placebo: 0.6%) and 3% with TPM (Shorvon, 1996).

APs effects on seizure threshold

As far as convulsant effects of APs are concerned, a report of seizures induced by chlorpromazine appeared in the literature within the first year of the introduction

1 Thetermforcednormalization(Landolt,1958)indicatestheappearanceofapsychosisinanepilepticpatient in whom the abnormal EEG became normal as a result of anticonvulsant treatment.

247 Clinical studies of PD interactions between AEDs and other drugs

Table 13.2 Incidence of seizures induced by APs

Drug Dose Seizure incidence (%)

Phenothiazines

Clozapine

Olanzapine Quietapine Risperidone

High ( 1 mg/day) 9.0 Moderate 0.7 Low ( 200 mg/day) 0.3 High (600–900 mg/day) 4.4 Moderate (300–599 mg/day) 2.7 Low ( 299 mg/day) 1.0 Wide range 0.9 Wide range 0.9 Wide range 0.3

Source: Data from Logothetis (1967) and Alldredge (1999).

of this drug in clinical practice (Zaccara et al., 1990). Subsequent studies showed that phenothiazines were able to produce convulsions. In a study of hospitalized psychi- atric patients, Logothetis (1967) found that the incidence of spontaneous seizures was 1.2% among 859 patients under treatment with phenothiazines. The incidence increased to 9% among patients receiving large therapeutic doses of these agents, while only 0.5% of patients treated with low or moderate doses had seizures. Patients with organic brain diseases were at higher risk. Seizures were generally observed at the onset of therapy or after a sudden increase in the dose.

To date, almost all of the APs introduced in clinical practice are known to induce seizures in predisposed subjects. In this respect, the aliphatic phenothiazines (e.g. chlorpromazine, promazine and triflupromazine) imply a higher risk of this adverse event than the phenothiazines bearing a piperazine or piperidine moiety (Zaccara et al., 1990). The degree of the epileptogenic power of a neuroleptic seems to be related to the ratio between the blockage of D2 dopaminergic receptor (which is convulsant) and the blockage of D1 receptor (anticonvulsant). It seems also to be associated with the agent’s antihistaminergic activity. In general, the more prominent the sedative properties of an individual AP, the higher its epilep- tic potential. However, as with ADs, the low incidence of reported cases does not allow an accurate assessment of the relative seizure risk. It has been suggested that, among traditional APs, haloperidol, fluphenazine, molindone, pimozide and trifluoperazine have a lower rate of seizures during therapeutic use and should be preferred in patients with epilepsy (Alldredge, 1999).

Convulsant effect of atypical APs (clozapine, olanzapine, quetiapine and risperidone)

Among the atypical APs (AAPs), clozapine carries the highest seizure risk (see Table 13.2). The occurrence of seizures appears to be dose-related, possibly

248 Gaetano Zaccara et al.

occurring at a dosage rate of 0.7% per 100 mg. At higher doses, seizure risk rises and reaches 5% at doses of 600–900 mg/day (Alldredge, 1999). In schizophrenic patients, the drug causes electroencephalographic (EEG) abnormalities typically characterized by background slowing in the theta and often the delta range. Bilateral spike, polyspike and slow wave discharges have also been described (Malow et al., 1994). Antiepileptic treatment is indicated in patients experiencing seizures with clozapine. Olanzapine has a binding profile similar to that of clozapine but, despite their similarity, the two drugs demonstrate a strong clinical difference concerning induction of seizures (Table 13.2). As far as quietapine is concerned, no difference in the incidence of seizures was observed between patients treated with this drug and those given placebo (incidence of 0.4% and 0.5%, respectively). Finally, as far as risperidone and sertindole are concerned, only a few patients with seizures have been reported (Alldredge, 1999; Torta and Keller, 1999).

In summary, with the exception of clozapine, the new APs are less prone to induce seizures than the traditional ones. Nevertheless, caution is recommended in using these drugs in patients with a history of seizures or with a lowered seizure threshold.

Adverse effects

Clozapine causes agranulocytosis in about 0.4% of patients (Lader, 1999). Similar figures have also been reported with some AEDs. Incidence values for aplastic anemia have been published for CBZ (39 cases per million) and felbamate (FBM) (127 cases per million) (Kaufman et al., 1997). These values are consistently higher than the overall incidence in the general population which is two cases per million per year (Kaufman et al., 1997). Therefore, concomitant administration of clozap- ine and other AEDs to patients at high risk of developing aplastic anemia (partic- ularly FBM) should be avoided.

The association of low-potency sedative APs and sedative AEDs (e.g. barbiturates, BZD) may precipitate or aggravate sedation. All neuroleptics cause weight gain and this adverse effect is more evident for AAPs olanzapine and clozapine. Some AEDs (VPA and VGB) cause weight gain too. Therefore, the choice of the appropriate asso- ciation between AEDs and APs should also take into account this aspect. Finally, one case has been described, in the literature, of catatonia-like events apparently induced by the association of VPA, sertraline and risperidone. A complex PD interaction has been advocated to explain this rare adverse effect (Lauterbach, 1998).

PD interactions with stimulants of the CNS

All CNS stimulants produce a dose-related excitation of the CNS which can lower seizure threshold (Zagnoni and Albano, 2002). In fact, seizures are frequently

249 Clinical studies of PD interactions between AEDs and other drugs

observed during overdose with these drugs. In a retrospective study of seizures associated with poisoning or drug intoxication, CNS stimulants were involved in 29% of the cases (Olson et al., 1994). However, in some patients, amphetamines (whose effect is to increase dopaminergic transmission) may reduce seizure activ- ity and improve EEG (Zaccara et al., 1990).

A reduced level of vigilance, which can be induced by AEDs and particularly by barbiturates, can worsen seizure frequency in some types of epilepsy (Papini et al., 1984). In addition, sedative effects of traditional AEDs can exacerbate the over- activity and aggressiveness of some epileptic patients (Viani et al., 1977). Based on these considerations, the use of amphetamines, which improve vigilance and con- trast sedation, was proposed as a comedication in some epileptic disorders. According to this hypothesis, a propylhexedrine salt of PB (barbexaclone, an amphetamine-like molecule) has been used in the treatment of epilepsy. The aim of this association was to determine a favorable PD interaction characterized by potentiation of the anticonvulsant effects and antagonism of the sedative effects of PB. The safety of the use of barbexaclone in epileptic patients has been docu- mented only by a few open studies conducted in small patient groups (Visintini et al., 1981) even though more studies document that amphetamines have bene- ficial effects on attention deficits in epileptic patients (Gross-Tsur et al., 1997). However, even at low doses, CNS stimulants can have proconvulsant activities. In 234 non-epileptic children with attention deficit and hyperactive disorders, seizures occurred in 2% of the stimulant-treated group (a rate higher but not par- ticularly alarming given that an estimated 1% of unselected children have seizures). Instead, in a subgroup of patients with epileptiform discharges in the EEG, seizures were observed in 20% of cases (Hemmera et al., 2001).

Adverse effects

Dyskinesia is a rare adverse effect of many AEDs. Choreoathetosis, dystonia and orofacial dyskinesias have been described. PHT is the AED most frequently impli- cated although cases have also been described with CBZ, gabapentin (GBP), FBM, VPA and LTG (Zaatreh et al., 2001). Young subjects with organic brain abnormali- ties are at higher risk. It has been postulated that PHT may cause chorea through enhancement of central dopaminergic pathways in the basal ganglia. Since amphe- tamines increase dopaminergic transmission and may cause dyskinesias, the associ- ation of amphetamine-like stimulants with some AEDs in high-risk patients can increase the risk of this adverse effect (personal unpublished observation).

In conclusion, at low doses, stimulants are co-administered with AEDs in the epileptic patient and seem to have beneficial PD interactions. However, in some patients with a lower seizure threshold or exposed to high doses, these substances have proconvulsant effects.

250 Gaetano Zaccara et al.

PD interactions with anesthetic agents

Some anesthetic agents are used in the treatment of status epilepticus and therefore have strong anticonvulsant properties. However, selected agents used during gen- eral anesthesia are reported to be epileptogenic (Zaccara et al., 1990). For example, etomidate and enflurane enhance epileptiform activity in the EEG and have been exploited for their ability to elucidate epileptogenic regions during seizure surgery. Methohexital, a short acting barbiturate, has also been used to enhance epileptiform activity on the EEG, although it paradoxically functions as an anticonvulsant.

Among anesthetic agents, lidocaine is of particular interest. This agent has a concentration-dependent effect on seizures. At concentrations between 0.5 and 5.0 mg/l, lidocaine can effectively suppress seizures in animal models of epilepsy and in clinical practice (DeToledo, 2000). In fact, it has been used in the treatment of convulsive status epilepticus and epilepsia partialis continua. Levels above 8–9mg/l, however, selectively block inhibitory mechanisms and may induce seizures. This bimodal response has been clearly demonstrated in experimental models of epilepsy and in healthy volunteers (DeToledo, 2000).

Treatment of anesthetic-induced convulsions can be particularly difficult because of many possible unfavorable PD interactions. AEDs (particularly barbi- turates) may seriously exacerbate circulatory and respiratory depression caused by anesthetics. Particular attention should be paid to depressive effects on the myocardium which are potentiated by co-administration of anesthetics and barbi- turates. Since severe hypoxia, hypercapnia and lactic acidosis occur concomitantly with anesthetic-induced convulsions and may be aggravated by many AEDs, treat- ment with succinylcholine and simultaneous ventilation should be the immediate treatment of choice to stop convulsions rapidly. However, because this procedure has no effect on cortical electric seizure activity, a co-treatment with BZD is also required (Zaccara et al., 1990).

PD interactions with analgesic and anti-inflammatory agents

AEDs, analgesics and anti-inflammatory drugs can often be co-administered for the treatment of some forms of pain. Neuropathic pain is not a specific entity, but comprises a variety of pain states with differing sensitivities to varying pharmaco- logical interventions (MacPherson, 2000).

AEDs in the treatment of pain

Abnormal ectopic impulse generation represents an important pathophysiologic mechanism of neuropathic pain. This abnormal impulse generation in injured nerves may depend on changes in the cell membrane Na channels. Furthermore,

251 Clinical studies of PD interactions between AEDs and other drugs

hypofunction of GABA-ergic inhibitory mechanisms and/or hyperfunction of glu- tamatergic excitatory mechanisms has been hypothesized to explain the diffusion of pain from the peripheral pain generator into the CNS (Bonezzi and Demartini, 1999). It has been observed that CBZ is effective in reducing acute pain but seems ineffective for continuous pain (Bonezzi and Demartini, 1999). GBP is now con- sidered a first-line medication in the treatment of several neuropathic syndromes. VPA and, more recently, LTG have given encouraging results (MacPherson, 2000).

However, several pain mechanisms may be operant in the same neuropathic dis- order and, therefore, it is often useful to associate drugs with different mechanisms of action. AEDs may be associated with opioids, ADs, alfa2-adrenergic agonists and non-steroidal anti-inflammatory drugs (MacPherson, 2000).

Effect of analgesic and anti-inflammatory drugs on seizure threshold

When administered in the CNS, both morphine and other opioid peptides can evoke an epileptiform activity in the EEG. These abnormalities are probably mediated by specific opioid receptors and are antagonised by opiate antagonists, such as naloxone (Tortella et al., 1979). However, in some experimental models, morphine has an anti- convulsant effect (Nowack et al., 1987). In the current practice, opioids have a low potency to induce seizures. This does not apply to pethidine. This drug may cause agi- tation, restlessness and seizures which have been postulated to be due to accumulation of the N-demethylated metabolite norpethidine. However, opioid-induced neuro- toxicity, which comprises cognitive failure, organic hallucinations and seizure activity, can result from therapy with any of the opioids, including morphine, fentanyl and hydromorphone (MacPherson, 2000). Seizures can also be observed after salicylate intoxication (Zaccara et al., 1990). In this circumstance, intravenous diazepam is con- sidered the drug of choice. The proconvulsant effect of ADs has already been described.

Opioids in valproic acid overdose

Recently, a few cases have been described in which naloxone has been successfully used to reverse CNS depression associated with acute VPA overdose (Roberge and Francis, 2002). In conclusion, co-administration of AEDs with analgesics and/or anti-inflammatory drugs and/or ADs can be useful in the treatment of neuro- pathic pain. Since this condition has different pathogenetic mechanisms, it is often necessary to administer drugs with different actions to target pain generation mechanisms at many levels and minimize adverse effects.

Conclusions

Although many AEDs are widely used in combination with other drugs (ADs, analgesics) to treat various diseases, a scarce knowledge has been gained on the PD

252 Gaetano Zaccara et al.

interactions of these drugs. There are hints that a true synergistic effect between some AEDs and ADs or analgesics can take place in the treatment or prophylaxis of mood disorders and in the treatment of neuropathic pain, respectively. In the field of epilepsy one can speculate that, in particular cases, the combination of AEDs with other drugs might improve seizure control. In addition, the association of stimulants with AEDs could be useful to antagonize some adverse effects (i.e. sedation). Further clinical studies are needed to verify these hypotheses.

REFERENCES

Alldredge BK. Seizure risk associated with psychotropic drugs: clinical and pharmacokinetic considerations. Neurology 1999; 53(Suppl. 2): S68–S75.

Bonezzi C, Demartini L. Treatment options in postherpetic neuralgia. Acta Neurol Scand 1999; 173: 25–35.

Bouman WP, Pinner G, Johnson H. Incidence of selective serotonin reuptake inhibitor (SSRI) induced hyponatremia due to the syndrome of inappropriate antidiuretic hormone (SIADH) secretion in the elderly. Int J Geriatr Psychiatry 1998; 13: 12–15.

Brent DA, Crumrine PK, Varma RR. Phenobarbital treatment and major depressive disorder in children with epilepsy. Pediatrics 1987; 80: 909–917.

DeToledo JC. Lidocaine and seizures. Ther Drug Monit 2000; 2: 320–322.
Dietrich DE, Emrich HM. The use of anticonvulsants to augment antidepressant medication.

J Clin Psychiatry 1998; 59(Suppl. 5): 51–58.
Favale E, Rubino V, Mainardi P, et al. Anticonvulsant effect of fluoxetine in humans. Neurology

1995; 45:1926–1927.
Ferrie CD, Robinson RO, Panayiotopoulos CP. Psychotic and severe behavioural reactions with

vigabatrin: a review. Acta Neurol Scand 1996; 93: 1–8.
Gross-Tsur V, Manor O, van der Meere J, et al. Epilepsy and attention deficit hyperactivity dis-

order: is methylphenidate safe and effective? J Pediatr 1997; 130: 40–44.
Hemmera SA, Pasternak JF, Zecker SG, et al. Stimulant therapy and seizure risk in children with

ADHD. Pediatric Neurology 2001; 24: 99–102.
Hurst DL. The use of imipramine as an anticonvulsant for minor motor seizures. Ann Neurol

1984; 18: 394.
Jobe PC, Dailey JW, Wernicke JF. A noradrenergic and serotonergic hypothesis of the linkage

between epilepsy and affective disorders. Crit Rev Neurobiol 1999; 13: 317–356.
Kastberg H, Jansen JA, Cole G, et al. Tiagabine: absence of kinetic or dynamic interactions with

ethanol. Drug Metabol Drug Interact 1998; 14: 259–273.
Kaufman DW, Kelly JP, Anderson T, et al. Evaluation of case reports of aplastic anaemia among

patients treated with felbamate. Epilepsia 1997; 38: 1265–1269.
Kramlinger KG, Post RM. The addition of lithium carbonate to carbamazepine: antimanic effi-

cacy in treatment resistant mania. Acta Psychiatr Scand 1989; 79: 378–385.

253 Clinical studies of PD interactions between AEDs and other drugs

Lader M. Some adverse effects of antipsychotics: prevention and treatment. J Clin Psychiatry 1999; 60(Suppl. 12): 18–21.

Lambert MV, Robertson MM. Depression in epilepsy: etiology, phenomenology, and treatment. Epilepsia 1999; 40(Suppl. 10): S21–S47.

Landolt H. In Lectures on Epilepsy. L. de Haas, ed. Asterdam: Elsevier, 1958: 91.
Lauterbach EC. Catatonia-like events after valproic acid with risperidone and sertraline.

Neuropsy Neuropsychol Behav Neurol 1998; 11: 157–163.
Lenox RH, McNamara RK, Waterson JM, et al. Myristoylated alanine-rich C kinase substrate

(MARKS): a molecular target for the therapeutic action of mood stabilizers in the brain? J Clin

Psychiatry 1996; 57(Suppl. 13): 23–31.
Logothetis J. Spontaneous epileptic seizures and electroencephalographic changes in the course

of phenothiazine therapy. Neurology 1967; 17: 869–877.
MacPherson. The pharmacological basis of contemporary pain management. Pharmacol Ther

2000; 88: 163–185.
Malow BA, Reese KB, Sato S, et al. Spectrum of EEG abnormalities during clozapine treatment.

Electroencephalogr Clin Neurophysiol 1994; 91: 205–211.
Marson AG, Kadir ZA, Hutton JL, et al. The new antiepileptic drugs: a systematic review of their

efficacy and tolerability. Epilepsia 1997; 38: 859–880.
McDanal CE, Bolman WM. Delayed idiosyncratic psychosis with diphenylhydantoin. J Am Med

Assoc 1975; 231: 1063.
Meinardi H. General principles. Use of combined antiepileptic drug therapy. In Antiepileptic Drugs,

4th edn. R. H. Levy, R. H. Mattson, S. Meldrum, eds. New York: Raven Press, 1995: 91–97. Normann C, Hummel B, Scharer LO, et al. Lamotrigine as adjunct to paroxetine in acute depres-

sion: a placebo-controlled study. J Clin Psychiatry 2002; 63: 337–344.
Nowack WJ, Johnson RN, Hanna GR. Observations of the effect of morphine on thalamo-

cortical excitability in the cat. Epilepsia 1987; 28: 457–462.
Olson KR, Kearney TE, Dyer JF, et al. Seizures associated with poisoning and drug overdose.

Am J Emerg Med 1994; 12: 392–395.
Papini M, Pasquinelli A, Armellini M, et al. Alertness and incidence of seizures in patients with

Lennox-Gastaut syndrome. Epilepsia 1984; 25: 161–167.
Peck AW, Stern WC, Watckinson C. Incidence of seizures during treatment with tricyclic anti-

depressant drugs and bupropion. J Clin Psichiatry 1983; 44: 197–201.
Perucca E, Gram L, Avanzini G, et al. Antiepileptic drugs as a cause of worsening seizures.

Epilepsia 1998; 39: 5–17.
Pisani F, Spina E, Oteri G. Antidepressant drugs and seizure susceptibility: from in vitro data to

clinical practice. Epilepsia 1999; 40(Suppl. 10): S48–S56.
Post RM, Ketter TA, Pazzaglia PJ, et al. Rational polypharmacy in the bipolar affective disorders.

In Rational Polypharmacy. J. E. Leppik, ed. Philadelphia: Elsevier Science, 1996: 153–180. Prien RF, Kupfer DJ, Mansky PA, et al. Drug therapy in the prevention of recurrences in uni- polar and bipolar affective disorders: report of the NIMH collaborative study group comparing lithium carbonate, imipramine, and lithium carbonate–imipramine combination. Arch Gen

Psychiatry 1984; 41: 1096–1104.

254 Gaetano Zaccara et al.

Roberge RJ, Francis EH. Use of naloxone in valproic acid overdose: case report and review. J Emerg Med 2002; 22: 67–70.

Salomon DA, Keitner GI, Ryan C, et al. Lithium plus valproate as mainteinance polypharmacy for patients with bipolar I disorders: a review. J Clin Psychopharmacol 1998; 18: 38–49.

Shorvon SD. Safety of topiramate: adverse events and relationships to dosing. Epilepsia 1996; 37(Suppl. 2): S18–S22.

Torta R, Keller R. Behavioral, psychotic, and anxiety disorders in epilepsy: etiology, clinical fea- tures, and therapeutic implications. Epilepsia 1999; 40(Suppl. 10): S2–S20.

Tortella FC, Moreton J, Khazan N. Electroencephalographic and behavioral effects of D-ala-methionine-enkephalinamide and morphine in rat. J Pharmacol Exp Therap 1979; 206: 636–646.

Viani F, Avanzini G, Baruzzi A, et al. Long-term monitoring of antiepileptic drugs in patients with the Lennox-Gastaut syndrome In Epilepsy: The VIIIth International Symposium. J. K Penry, ed. New York: Raven Press, 1977: 131–138.

Visintini D, Calzetti S, Mancia D. Barbexaclone in the treatment of the epilepsies. Riv Patol Nerv Ment 1981; 102: 29–37.

Weinbroum AA, Flaishon R, Sorkine P, et al. A risk-benefit assessment of flumazenil in the man- agement of benzodiazepine overdose. Drug Saf 1997; 17: 181–196.

Yatham LN, Kusumakar V, Calabrese JR, et al. Third generation anticonvulsants in bipolar dis- order: a review of efficacy and summary of clinical recommendations. J Clin Psychiatry 2002; 63: 275–283.

Zaatreh M, Tennison M, D’Cruz N, et al. Anticonvulsants-induced chorea: a role for pharmaco- dynamic drug interaction? Seizure 2001; 10: 596–599.

Zaccara G, Muscas GC, Messori A. Clinical features, pathogenesis and management of drug- induced seizures. Drug Safety 1990; 5: 109–151.

Zagnoni PG, Albano C. Psychostimulants and epilepsy. Epilepsia 2002; 43(Suppl. 2): 28–31.

Part IV

Drug interactions in specific patient populations and special conditions

Antiepileptic drug interactions in children

Olivier Dulac1, Elizabeth Rey2 and Catherine Chiron1

1 Hôpital Necker-Enfants Malades, Paris, France 2 Hôpital Saint Vincent de Paul, Paris, France

Introduction

Many clinical practitioners are of the opinion that the optimal treatment of epilepsy is best achieved by use of antiepileptic drugs (AEDs) that have several modes of action, and therefore the drugs that are the most effective in this regard are AEDs such as carbamazepine (CBZ), valproate (VPA) or topiramate (TPM) whose efficacy relates to several modes of action. Thus, from the pharmacody- namic point of view, these AEDs when prescribed as monotherapy in fact comprise polytherapy regimens. On the other hand, because of metabolism, many AEDs reach the brain as combinations of the parent drug and their metabolite(s) and this too can be considered a form of polytherapy. For example, CBZ, which is meta- bolized to a pharmacologically active metabolite CBZ-epoxide, readily enters the brain where it exerts pharmacological effects. Thus polytherapy at the brain level can in fact be distinguished from polytherapy at the oral level. The same applies to clobazam (CLB) whose metabolism is inhibited by stiripentol leading to signifi- cant increase of CLB and norCLB with far better tolerability (Perez et al., 1999; Chiron et al., 2000). This is also observed with VPA for which the proportion of toxic 4-ene-VPA is decreased (Levy et al., 1987). A further consideration is that the metabolic pathway may vary according to age. Therefore, in infants, the hydroxy- lation of both diazepam and nordiazepam is very limited, combined with low glucoronidation capacity which generates major hypotonia (Morselli et al., 1973).

In clinical practice, access to plasma level monitoring has demonstrated that meta- bolic interactions are very complex, and contribute to frequent and often insidious side effects including paradoxical increases in seizure frequency (Reynolds and Shorvon, 1981). Thus, insidious occurrence of increased plasma concentration may generate severe toxicity. This is the case for the combination of phenytoin (PHT) with phenobarbital (PB) in children, a combination that results in unpredictable plasma concentrations and carries a risk of increased toxicity to PHT with cerebellar atrophy, due to progressive accumulation of PHT. Another example of the increased toxicity of combined drugs compared to monotherapy is the combination of lamotrigine (LTG)

257

258 Olivier Dulac et al.

with VPA that produces the highest incidence of skin rash, although this combination is, from the therapeutic point of view, particularly efficacious (Brodie and Yuen, 1997). On the other hand, a supra-additive effect has been observed with LTG and VPA in combination, that could be the consequence of a pharmacokinetic, due to metabolic interaction in the liver, or a pharmacodynamic, due to some modification of the action of the molecules inside the brain, interaction, or both (Pisani et al., 1999).

Therefore, the issue of mono- versus polytherapy covers a wide range of con- cepts, and proper analysis requires us to take into account the whole pathway from oral administration, through liver metabolism, to the mode of action within the brain, and also pharmacokinetic differences according to age. However, because of to the lack of insight into the mechanism of action of most AEDs, it is not possible to predict the benefit versus negative effects of the various combinations.

Few studies have been performed to address this issue, primarily because of its complexity and the significant number of factors that are involved. Animal studies have provided evidence of some specific AED combinations that are indeed addi- tive, whilst others were supra-additive and still others infra-additive (Bourgeois, 1988). However, these studies were based on acute AED administration and could not take in account the effect of chronic administration, which may modify the supply of drug to the brain because of metabolic interactions in the liver that need at least several days, often a few weeks, to take place after onset of therapy. Pragmatic clinical studies have been performed, but the results obtained could be misleading if not interpreted properly. Thus, no significant difference was found between VPA mono- and polytherapy (Deckers et al., 2001). Clinical experience shows that mono- and polytherapy do not have the same value according to the type of epilepsy, and to the type of polytherapy. Thus, the type of epilepsy needs to be taken in account, and also the type of drug.

Specificity of epilepsy in pediatrics is its considerable heterogeneity with a grow- ing number of epilepsy syndromes identified. The latter is combined with more or less specific response to drugs or to drug combinations for each given syndrome (Luna et al., 1989; Roger, 1992; Schlumberger et al., 1994). This variable response to drugs includes a risk for worsening of seizure frequency and severity that needs to be taken in account, even when addressing the issue of drug combinations (Perucca et al., 1998). The situation is complicated by the fact that a patient may switch over time from one syndrome to another as an effect of age or as a consequence of treat- ment. Infantile spasms in a patient with focal malformation respond to vigabatrin (VGB), but in approximately 50% of cases the child is left with focal seizures (Lortie et al., 1993). The addition of CBZ raises the risk of relapse of spasms that again dis- appear with cessation of CBZ (Talwar et al., 1994; unpublished data).

A very particular aspect that impacts on our knowledge of AEDs and the best use we can make of them, relates to the strategy of their development by the

259 Antiepileptic drug interactions in children

pharmaceutical industry, which is guided by registration body requirements, and both ethical and marketing considerations. Drugs are first developed for adults that suffer from partial epilepsy. Then, if it appears useful to adults, the drug is tested in children, with a clear preference for what is considered as the most intractable conditions, par- tial epilepsy and Lennox–Gastaut syndrome. However, compounds are tested first as add-on because it is given to patients with resistant epilepsy for which it is not possi- ble to withdraw the previous treatment. For this reason the drug reaches the market with an indication restricted to polytherapy. It is only later, and with major method- ological difficulties, that studies permit the efficacy in monotherapy to be demon- strated. Often, a few years later, it appears that the AED is much better tolerated in mono- than in polytherapy. The best example is VPA for which several years were needed before the compound was widely and legally used in monotherapy, whereas is was clear that polytherapy had contributed to fatal hepatic toxicity (Dreifuss et al., 1987, 1989; Bryant and Dreifuss, 1996). Such drawbacks of polytherapy also apply to the therapeutic aspect itself: in one open study performed soon after the launch of VPA in polytherapy, patients with idiopathic generalized epilepsy still suffering from tonic–clonic seizures when treated with the combination of VPA with PB, experienced disappearance of seizures just by withdrawing PB, without any modification of the dose of VPA (Dulac et al., 1982). The growing interest for evidenced-based medicine and restrictions given to the use of drugs out of the strict legal indications may there- fore contribute to a somewhat vicious use of medication because it is paradoxically only legal to use the compound in its most hazardous condition, polytherapy, before studies demonstrate efficacy in monotherapy. In the present state of knowledge, it is therefore reasonable with drugs that have no monotherapy claim, to start the medica- tion as add-on therapy, as legally required, but then to go to monotherapy as soon as a clear benefit has been obtained. Nevertheless, there are individual cases in which the combination is more effective than monotherapy.

Interactions between AEDs

In this chapter we will review the characteristics of the various interactions between the various AEDs, including those that are in development, and what is presently known regarding their mechanism; we will then highlight the benefits this knowl- edge can offer to optimize the treatment for each type of epilepsy in children.

Clinically relevant metabolic and pharmacodynamic interactions

Phenobarbital, phenytoin and carbamazepine

PB, PHT and CBZ are potent inducers of hepatic metabolizing enzymes and con- sequently any comedication compound that undergoes hepatic metabolism will

260 Olivier Dulac et al.

Valproate

need to be administered at a higher dose so as to achieve an adequate therapeutic response. PHT generates particular difficulties since it follows a similar metabolic pathway to that of PB and may therefore compete on the catabolism enzymatic activity, with unpredictable results. In particular, accumulation of PHT with toxic effects in the cerebellum and in peripheral nerves may occur insidiously. Because of the metabolism of CBZ to a pharmacologically active metabolite, CBZ-epoxide, the administration of CBZ in effect represents two compounds. Thus a CBZ-epoxide plasma level of over 2.2 mg/l combined with CBZ may be toxic in children, whereas each single component seems to be better tolerated; no major side effect was observed when patients achieved similar plasma concentrations of CBZ-epoxide when administered alone (Schoeman et al., 1984). Based on these observations, CBZ was administered in combination with the experimental compound stiripen- tol, a compound that inhibits the metabolism of CBZ in the liver, thus decreasing the formation of CBZ-epoxide and increasing the plasma concentration of CBZ. This rational polytherapy results in better tolerability and better therapeutic effect (Tran et al., 1996; Perez et al., 1999).

VPA being a metabolic inhibitor requires that drugs administered in combination are administered at lower doses. This applies particularly to PB, PHT, CBZ, LTG and ethosuximide (ESM). For PHT, the total plasma concentration is reduced but the free fraction is not affected, and therefore the dose should not be altered. For CBZ, the clearance of CBZ-epoxide is reduced resulting in poor tolerability particularly in relation to cognitive function, thus necessitating CBZ dose reduction. The combina- tion of VPA with PB results in a decrease of VPA and an increase of PB plasma con- centrations. This does have some clinical relevance since it could explain the disappearance of tonic–clonic seizures that occurs upon PB withdrawal and without modification of VPA dose (Dulac et al., 1982). For the combination of VPA with LTG, the risk is that of skin rash, when this combination is introduced too rapidly. However, this combination has the advantage that lower LTG doses are needed and therefore treatment costs are reduced. In addition, this combination has been associ- ated with a positive pharmacodynamic interaction (Pisani et al., 1999) as well as a therapeutic synergism (Brodie and Yuen, 1997). With clonazepam, reduced wakeful- ness may contribute to the precipitation of status epilepticus in intractable epilepsy with myoclonic seizures (Covanis et al., 1982). VPA also has intrinsic metabolites; 2-ene-VPA that has been shown to be more effective than VPA (Loscher and Nau, 1985), and 4-ene-VPA that may be involved in hepatic toxicity (Nau et al., 1984).

Most new compounds have reduced metabolic interactions with comedication. However, VGB does reduce the clearance of PHT, and thus PHT plasma concen- trations may become toxic when removing VGB (Luna et al., 1989).

261 Antiepileptic drug interactions in children

Oxcarbazepine

Oxcarbazepine (OXC) is associated with few metabolic interactions; with PHT, plasma PHT concentration can be increased by 40% (Sallas et al., 2003) and with LTG, plasma LTG concentration is decreased by 33% (May et al., 1999). OXC may induce the metabolism of other non-antiepileptic comedication.

Gabapentin

Felbamate

Lamotrigine

Gabapentin (GBP) is not metabolized, and does not affect liver enzymes and conse- quently GBP does not affect the metabolism of drug comedication. However, there may be an interaction with felbamate (FBM) at the level of kidney excretion, result- ing in 50% increase in felbamate half-life values (Hussein et al., 1996).

FBM is, from the pharmacokinetic point of view, a particularly complex com- pound since it increases the clearance of CBZ and CBZ-epoxide, and reduces that of VPA, PB and PHT. This could lead to toxic plasma PHT concentrations. The metabolism of FBM is enhanced by enzyme-inducing drugs. The combination of FBM with VPA is useful in the treatment of Lennox–Gastaut since in one series the frequency of drop attacks was reduced by 40% (Siegel et al., 1999).

Although topiramate (TPM) is mainly excreted through the kidney, this com- pound is sensitive to the enzyme-inducing AEDs which enhance its hepatic metabo- lism two-fold (Dooley et al., 1999). The increase in behavioral disorders that have been associated with LTG comedication, are likely to be the consequence of a phar- maco-dynamic interaction (Gerber et al., 2000). Also, TPM may inhibit PHT metab- olism. Overall, the tolerability of TPM is clearly far better as a monotherapy regimen compared to when administered in combination, particularly in combination with CBZ or VPA.

LTG is particularly sensitive to the metabolic effect of comedication, both of inducer and inhibitor compounds: its elimination half-life is reduced by PB and CBZ but increased by VPA. When starting LTG in combination with VPA, the plasma concentration tends to rise more quickly than when it is given alone, and this increases the risk for skin rash. Indeed, before this pharmacokinetic effect was identified, we experienced a 10% rate of skin rash when adding LTG to VPA (Schlumberger et al., 1994), that decreased to 1% when the dose was titrated more slowly (Besag et al., 1995). The combination of LTG with CBZ is poorly tolerated in terms of vigilance, and produces the effects of overdosage with CBZ (Besag et al., 1995).

262 Olivier Dulac et al.

Pragmatic aspects of treatment

First-line treatment

At this stage, there is no longer any place for polytherapy. A number of AEDs have now been shown to be effective as monotherapy for various types of epilepsy, in which they may therefore be administered as first-line drug. This is the case for CBZ (Glauser, 2000), VPA (Dulac et al., 1982), OXC (Serdaroglu et al., 2003), and LTG (Ueberall, 2001) which are effective in partial epilepsy. This similarly applies to VGB in infantile spasms, not only those due to tuberous sclerosis (Chiron et al., 1997) but whatever the etiology (Appleton et al., 1999; Elterman et al., 2001). For childhood absence epilepsy, the effects of VPA and ESM seem to be interchange- able (Sato et al., 1982). For juvenile absence epilepsy, the risk of generalized tonic– clonic seizures in combination with absences is an indication not to restrict to ESM monotherapy and, for LTG, no controlled trial has confirmed the effect as monotherapy in absence epilepsy. For idiopathic generalized epilepsy with tonic–clonic and/or myoclonic seizures, controlled trials with VPA monotherapy are only available in adults (Turnbull et al., 1982). The occurrence of repeat tonic–clonic seizures between 2 and 5 years of age in a previously normal child is most likely to be the first expression of myoclonic–astatic epilepsy, that contra- indicates the use of CBZ, and indicates VPA, although VPA is likely to soon prove to be insufficient in monotherapy (Dulac et al., 1998).

Epilepsy resistant to a first-line monotherapy

Epilepsy resistant to a first-line monotherapy requires a switch to a second monotherapy. However, it remains unclear whether there should be an immediate switch with withdrawal of the previous AED, or a progressive addition of a second AED followed by removal of the first as soon as benefit from the second AED is confirmed. In practice, before new data are available, it seems reasonable to decide according to each specific condition: for epilepsy syndromes or seizure types for which the presently administered drug is determined to be worsening, or does not seem appropriate because it is known to comprise a sizeable risk of worsening; for those for which a given AED seems more appropriate; and for the conditions in which the first AED did not give any clear benefit; a simple switch over a couple of weeks should be undertaken. In cases with apparently partial effects, the addition of the second AED should be chosen before returning to monotherapy, because removing the previous AED could generate withdrawal effects if the new AED is not sufficiently effective. In addition, one needs to take into account the poten- tial metabolic interactions between the first and second AED and therefore adapt the dose of the previous drug, and one needs also to adapt the pharma- cokinetics of drug withdrawal to the type of AED and to the duration of previous

263 Antiepileptic drug interactions in children

treatment: even if ineffective; a previous treatment with PB, VGB, PHT or CBZ lasting several months could generate dependency, and therefore require very slow withdrawal.

Epilepsy resistant to a second AED

The use of a third AED in the treatment of epilepsy resistant to a second AED in adults is known to be associated with little benefit. Therefore, for partial epilepsy, provided that the diagnosis of the type of epilepsy is correct, that the treatment is properly given and at a proper dose and that there is no underlying progressive dis- ease (Aicardi, 1988), it seems reasonable to consider surgery at this point (Kwan and Brodie, 2000). In children, no such data are available. Typically, controlled add-on trials for partial epilepsy with a new AED report rates of seizure freedom of approximately 5–15%: 10% for VGB (Luna et al., 1989), 14% for TPM (Ritter et al., 2000), and 14% for OXC (Rey et al., 2004), but only 3% for gabapentin (Appleton et al., 2001). However, for generalized epilepsy, the potential benefit depends on the type of syndrome. In absence epilepsy, combining LTG with VPA was associated with significant benefit (Pisani et al., 1999), and the same applies to myoclonic– astatic epilepsy (Dulac et al., 1998).

Treatment according to the type of epilepsy or epilepsy syndrome

Cryptogenic or symptomatic partial epilepsy

In cryptogenic or symptomatic partial epilepsy, whatever the age, monotherapy has a place of choice, with no significant difference of benefit with VPA or CBZ in terms of efficacy following a first seizure (Verity et al., 1995) . However, tolerability seems to be slightly better with the former (Chaigne and Dulac, 2003). In addition, there is a mild restriction about the use of CBZ according to age and the type of epilepsy. In infancy, the risk of secondary development of infantile spasms, follow- ing partial epilepsy, is such that unless there is focal lesion usually not combined with infantile spasms, such as Sturge–Weber disease, the use of CBZ should be avoided. When, in childhood, cryptogenic or symptomatic focal epilepsy is com- bined with major spike wave activity, CBZ could contribute to the generation of continuous spike waves in slow sleep (Corda et al., 2001). Nevertheless, in the chronic condition, CBZ is more efficacious than VPA in preventing the recurrence of focal seizures. Lack of response to this first AED indicates the need to switch to the alternate AED. However, the indication could depend the topography of the epilepsy focus. Thus, OXC may be more efficacious in temporal lobe epilepsy, TPM seems more efficacious in motor seizures generated by the motor strip, whereas LTG seems useful in frontal lobe seizures, namely when combined with VPA.

264 Olivier Dulac et al.

Infancy

For patients with transient effect of CBZ, the addition of stiripentol may produce remarkable effects (unpublished data).

Dravet syndrome

In infancy, Dravet syndrome may worsen with the addition of CBZ, PB (Thanh et al., 2002), LTG (Guerrini et al., 1998), or VGB (Lortie et al., 1993). Bromide is widely used in Germany and in Japan. Treatment strategy for this disorder clearly improved with the introduction of a new concept of polymedication. Indeed, nei- ther VPA nor PB succeed in preventing status epilepticus that contributes largely to worsening of the condition (Casse-Perrot et al., 2001). The combination of CLB with stiripentol has been shown to prevent the occurrence of status epilepticus in young children, and to significantly reduce seizure frequency (Chiron et al., 2000). A pragmatic study has shown that this combination should be administered as soon as possible in the course of the disease (Thanh et al., 2002). The addition of TPM also reduces seizure frequency, although the occurrence of status epilepticus cannot be prevented, and although in many instances this drug does not permit withdrawal of VPA (Coppola et al., 2002). Nevertheless, it would become possible to withdraw stiripentol and CLB in the second decade, thus reducing the polyther- apy to a combination of VPA and TPM.

Infantile spasms

In infantile spasms, monotherapy either with VGB or steroids seems to be the treat- ment of choice. When the first drug is not efficient, the alternate may prove effec- tive. No controlled study has questioned an eventual benefit from the combination of both. Patients with previous psychomotor retardation but no neuroradiological abnormality were found not to respond to VGB or steroids alone, although half these patients became spasm-free with the combination of both over several months (Villeneuve et al., 1998). In intractable cases, low doses of LTG combined with VPA seem to be of benefit in a small proportion of patients (Cianchetti et al., 2002). In contrast, this is not observed when LTG is combined with CBZ Veggiotti et al., 1994). Clinical practice shows that felbamate may be effective, but no data could show specific benefit of a combination of this compound compared to its administration as monotherapy. However, since for regulatory purposes initial studies were performed as add-on, the AED is registered for add-on administration. A number of patients with infantile spasms exhibit focal seizures, either in combi- nation with the spasms or as residual phenomena after the disappearance of the spasms with the therapy. In these cases, not only is the combination of CBZ ineffec- tive, but it could also precipitate the relapse of spasms (Talwar et al., 1994). LTG could be helpful in these cases, after the age of 2 years (Veggiotti et al., 1994).

265 Antiepileptic drug interactions in children

Benign partial seizures in infancy and benign myoclonic epilepsy in infancy

Benign partial seizures in infancy and benign myoclonic epilepsy in infancy are easily controlled by monotherapy with, respectively, VPA or either VPA, ESM or a benzodiazepine, primarily CLB.

Childhood

Absences

In childhood, absences usually respond to monotherapy with either VPA or ESM. The older concept that PB should be added because of the risk of occurrence of tonic–clonic seizures is no longer valid since it is clear that very few patients with childhood absence epilepsy do exhibit tonic–clonic seizures and PB may worsen the absences. A majority of patients non-responsive to either VPA or ESM may respond to add-on LTG. Whether they would respond to monotherapy LTG remains to be determined, and usually it is after a few months of the combination with VPA that a progressive reduction to monotherapy could be attempted. However, because the dose required for duotherapy with VPA is lower, one tends to maintain the combi- nation. The dose of both ESM and LTG should be halved, and the introduction of LTG should be undertaken slowly because of VPA comedication. A combination of all three AEDs, VPA, ESM and LTG may occasionally be helpful.

Benign partial epilepsy

Benign partial epilepsy responds well to various monotherapies. The real need with this condition is to make a clear diagnosis in order to reduce, as much as pos- sible, the indication for any AED medication which, in practice, is required in less than one-third of such patients (Ambrosetto and Tassinari, 1990). It is clear that when a treatment is needed, VPA, sulthiame, CBZ and CLB are all very efficacious. The only restriction is that associated with CBZ, which has a small risk of con- tributing to the occurrence of continuous spike waves during slow sleep (Corda et al., 2001), as has also been reported with LTG (Battaglia et al., 2001).

Myoclonic–astatic epilepsy

Myoclonic–astatic epilepsy is resistant to any monotherapy. VPA is usually admin- istered when the first, tonic–clonic seizure presents between 2 and 5 years of age (Kaminska et al., 1999). Seizure recurrence or the additional occurrence of myoclonic seizures would necessitate the addition of LTG, with the restrictions and cautions mentioned earlier. Because of the long time lag to reaching proper dosage, it is preferable to start adding this compound as soon as the diagnosis becomes likely, based on the recurrence of tonic–clonic seizures with generalized spike

266 Olivier Dulac et al.

waves in this age range, before myoclonic–astatic seizures do occur. The addition of ESM may be useful, in case of very rapid increase of seizure frequency, before the dose of LTG can reach sufficient levels, and when myoclonic seizures or absences persist. In this setting, LTG dosage needs to be reduced by half because of VPA comedication. TPM may be useful in some cases, but the combination with VPA should be avoided because of potential side effects. Indeed speech, which is often affected in this type of epilepsy, is sensitive to TPM, and this is a very specific effect of this AED (Aldenkamp et al., 2000). The addition of clonazepam to VPA may eventually precipitate status epilepticus (Covanis et al., 1982). CLB does not seem to carry this risk. Thus, many patients end up being treated with three AEDs.

Lennox–Gastaut syndrome

Lennox–Gastaut syndrome is rarely controlled with a single AED. As soon as the syndrome is suspected, LTG should be added to VPA. Indeed, LTG has been shown to be effective in this condition, as add-on therapy, with a sizeable number of patients becoming seizure-free (Motte et al., 1997). Since absences are one compo- nent of the syndrome, there is likely a pharmacodynamic interaction of both AEDs as in absence epilepsy (Perucca et al., 1998). Persistent seizures are then an indica- tion to add FBM, with the usual biological, hepatic and hematological follow-up monitoring constraints. Although TPM was also shown to significantly reduce seizure frequency in this condition (Sachdeo et al., 1999), no patient became seizure-free, thus polytherapy with this AED would come later in the treatment algorithm. The use of PHT in the polytherapy is more rare, with the aim of reducing the frequency of generalized, namely tonic seizures.

Continuous spike waves in slow sleep

Continuous spike waves in slow sleep are rarely controlled by benzodiazepine or sulthiame monotherapy (Rating et al., 2000). Adding ESM may occasionally be useful. The addition of other conventional AEDs, namely CBZ, PB or PHT is more often deleterious than useful. Even the occurrence of additional focal seizures is not an indication for this type of medication, since it may aggravate the condition (Perucca et al., 1998). Few patients have benefited from TPM (Mikaeloff et al., 2003). At this point, steroids are the most helpful. Whether a benzodiazepine should then be maintained in combination with steroids is not clear.

Combining AEDs with non-AED drugs

Combining AEDs with non-AED drugs needs special attention. In infants treated with VPA, the administration of acetylsalicylic acid should be limited to situations in which there is absolute need. Indeed, this combination is associated with a high

267 Antiepileptic drug interactions in children

risk of liver failure (Dreifuss et al., 1989). Macrolides should not be given with CBZ, because they reduce its clearance and may produce insidious toxicity (Mesdjian et al., 1980). Among old generation AEDs, CBZ, PHT and PB induce the activity of several enzymes involved in drug metabolism leading to decreased plasma concen- tration and reduced pharmacological effect of drugs, which are substrates of the same enzymes. This occurs with immunosuppressive drugs including cyclosporine (Yusof 1988; Baciewicz and Baciewicz, 1989; Wasfi and Tanira, 1993; Cooney et al., 1995), tacrolimus (Thompson and Mosley, 1996), sirolimus (Fridell et al., 2003), glucocorticoids, tricyclic antidepressants such as imipramine, amitriptyline, some antipsychotic agents such as haloperidol, chlorpromazine, clozapine (Facciola et al., 1998; Lane et al., 1998), antiarrhythmic agents including disopyramide, lido- caine, propranolol (Vu et al., 1983), doxycycline and acetaminophen (Douidar and Ahmed, 1987). In contrast, the new AEDs OXC, GBP, LTG, levetiracetam, and TPM are not hepatic enzyme-inducing drugs and are not reported to be involved in such drug interaction. However, an interaction between cyclosporine and OXC (Rosche et al., 2001) was suspected in a single case study with decrease of the cyclosporine plasma concentration. Regarding enzyme inhibition the clearance of CBZ was decreased by clozapine (Langbehn and Alexander, 2000), and that of PHT was decreased by cimetidine (Rafi et al., 1999) leading in both cases to increase in the plasma concentration of the AED. Similarly, the plasma concentration of PHT is significantly increased by fluconazole (Cadle et al., 1994). Prediction of drug inter- action is difficult because enzyme induction or inhibition may coexist and many other factors are involved in determining whether a clinically significant drug interaction will occur or not. Furthermore most of these data are only case reports. Thus, available data should not be regarded as exhaustive.

Interactions with AEDs and chemotherapeutic drugs (CTDs), although poorly documented, do also occur. The coadministration of AED and a CTD may lead either to reduced activity or increased toxicity of an AED. Lowered plasma con- centrations of PHT were reported with seizure recurrence during administration of cisplatin or vinca alkaloid, and a 25% decrease in VPA plasma concentration was observed with high-dose infusion of methotrexate in children. Increased toxicity due to higher plasma-PHT concentration was reported when this AED was coad- ministered with 5-fluouracil. Although this comedication may not be relevant in children, it points out that one should be cautious when CTDs and AEDs have to be administered concomitantly. The effect of drug interaction may also lead to a reduced activity or an increased toxicity of a CTD: the clearance of vincristine was increased by 63% with the coadministration of enzyme-inducing AEDs, however, the impact on the efficacy of vincristine was not investigated. A faster clearance was observed for teniposide with a lower efficacy in children who received PHT, PB or CBZ. Increased toxicity was reported with the coadministration of VPA, cisplatin

268 Olivier Dulac et al.

and etoposide. Pharmacokinetic interactions may be suspected when AED and CTD drugs share a common metabolic pathway (Vecht et al., 2003).

Conclusion

Although the rule of monotherapy as the strategy of choice clearly applies to the majority of pediatric patients suffering from epilepsy, it remains difficult to main- tain it for patients with pharmacoresistant epilepsy. In addition, there is clear advantage of comedication in a restricted number of specific types of epilepsy. This may be the strict rational polytherapy. However, too few structured studies have been performed to validate this concept. In all cases, any study design of this type should take into account the epilepsy syndrome. Finally, there is a sizeable number of patients for whom polytherapy is by no means rationally designed, but imposed by the story of the epilepsy, because there can be a significant increase in seizure frequency at any attempt to reduce the polytherapy. Furthermore, episodes of status epilepticus may even occur, if a vigorous reduction of AEDs is attempted. In these patients, polytherapy can be considered a failure, and its reduction should be tried at regular intervals in order to diminish the risk of insidious side effects of the combination.

REFERENCES

Aicardi J. Clinical approach to the management of intractable epilepsy. Dev Med Child Neurol 1988; 30: 429–440.

Aldenkamp AP, Baker G, Mulder OG, et al. A multicenter, randomized clinical study to evaluate the effect on cognitive function of topiramate compared with valproate as add-on therapy to carbamazepine in patients with partial-onset seizures. Epilepsia 2000; 41: 1167–1178.

Ambrosetto G, Tassinari CA. Antiepileptic drug treatment of benign childhood epilepsy with rolandic spikes: is it necessary? Epilepsia 1990; 31: 802–805.

Appleton RE, Peters AC, Mumford JP, et al. Randomised, placebo-controlled study of vigabatrin as first-line treatment of infantile spasms. Epilepsia 1999; 40: 1627–1633.

Appleton R, Fichtner K, LaMoreaux L, et al. Gabapentin as add-on therapy in children with refractory partial seizures: a 24-week, multicentre, open-label study. Dev Med Child Neurol 2001; 43: 269–273.

Baciewicz AM, Baciewicz Jr FA. Cyclosporine pharmacokinetic drug interactions. Am J Surg 1989; 157: 264–271.

Battaglia D, Iuvone L, Stefanini MC, et al. Reversible aphasic disorder induced by lamotrigine in atypical benign childhood epilepsy. Epileptic Disord 2001; 3: 217–222.

Besag FM, Wallace SJ, Dulac O, et al. Lamotrigine for the treatment of epilepsy in childhood. J Pediatr 1995; 127: 991–997.

269 Antiepileptic drug interactions in children

Bourgeois BF. Problems of combination drug therapy in children. Epilepsia 1988; 29(Suppl. 3): S20–S24.

Brodie MJ, Yuen AW. Lamotrigine substitution study: evidence for synergism with sodium valproate? 105 Study Group. Epilepsy Res 1997; 26: 423–32.

Bryant III AE, Dreifuss FE. Valproic acid hepatic fatalities. III. U.S. experience since 1986. Neurology 1996; 46: 465–469.

Cadle RM, Zenon III GJ, Rodriguez-Barradas MC, et al. Fluconazole-induced symptomatic phenytoin toxicity. Ann Pharmacother 1994; 28: 191–195.

Casse-Perrot C, Wolf P, Dravet C. Neuropsychological aspects of severe myoclonic epilepsy in infancy. In I. Jambaque, M. Lassonde, O. Dulac, eds. Neuropsychology of Childhood Epilepsy. New York: Kluwer Academic/Plenum Press, 2001: 131–140.

Chaigne D, Dulac O. Carbamazepine versus valproate in partial epilepsies of childhood. Advances in Epileptology. New York: Raven Press, 2003: 17.

Chiron C, Dumas C, Jambaque I, et al. Randomized trial comparing vigabatrin and hydrocortisone in infantile spasms due to tuberous sclerosis. Epilepsy Res 1997; 26: 389–395.

Chiron C, Marchand MC, Tran A, et al. Stiripentol in severe myoclonic epilepsy in infancy: a ran- domised placebo-controlled syndrome-dedicated trial. STICLO study group. Lancet 2000; 356: 1638–1642.

Cianchetti C, Pruna D, Coppola G, et al. Low-dose lamotrigine in West syndrome. Epilepsy Res 2002; 51: 199–200.

Cooney GF, Mochon M, Kaiser B, et al. Effects of carbamazepine on cyclosporine metabolism in pediatric renal transplant recipients. Pharmacotherapy 1995; 15: 353–356.

Coppola G, Capovilla G, Montagnini A, et al. Topiramate as add-on drug in severe myoclonic epilepsy in infancy: an Italian multicenter open trial. Epilepsy Res 2002; 49: 45–48.

Corda D, Gelisse P, Genton P, et al. Incidence of drug-induced aggravation in benign epilepsy with centrotemporal spikes. Epilepsia 2001; 42: 754–759.

Covanis A, Gupta AK, Jeavons PM. Sodium valproate: monotherapy and polytherapy. Epilepsia 1982; 23: 693–720.

Deckers CL, Hekster YA, Keyser A, et al. Monotherapy versus polytherapy for epilepsy: a multi- center double-blind randomized study. Epilepsia 2001; 42: 1387–1394.

Dooley JM, Camfield PR, Smith E, et al. Topiramate in intractable childhood onset epilepsy – a cautionary note. Can J Neurol Sci 1999; 26: 271–273.

Douidar SM, Ahmed AE. A novel mechanism for the enhancement of acetaminophen hepato- toxicity by phenobarbital. J Pharmacol Exp Ther 1987; 240: 578–583.

Dreifuss FE, Santilli N, Langer DH, et al. Valproic acid hepatic fatalities: a retrospective review. Neurology 1987; 37: 379–385.

Dreifuss FE, Langer DH, Moline KA, et al. Valproic acid hepatic fatalities. II. US experience since 1984. Neurology 1989; 39: 201–217.

Dulac O, Steru D, Rey E, et al. Sodium valproate (Na VPa) monotherapy in childhood epilepsy. Arch Fr Pediatr 1982; 39: 347–352.

Dulac O, Plouin P, Shewmon A. Myoclonus and epilepsy in childhood: 1996 Royaumont meeting. Epilepsy Res 1998; 30: 91–106.

270 Olivier Dulac et al.

Elterman RD, Shields WD, Mansfield KA, et al. Randomized trial of vigabatrin in patients with infantile spasms. Neurology 2001; 57: 1416–1421.

Facciola G, Avenoso A, Spina E, et al. Inducing effect of phenobarbital on clozapine metabolism in patients with chronic schizophrenia. Ther Drug Monit 1998; 20: 628–630.

Fridell JA, Jain AK, Patel K, et al. Phenytoin decreases the blood concentrations of sirolimus in a liver transplant recipient: a case report. Ther Drug Monit 2003; 25: 117–119.

Gerber PE, Hamiwka L, Connolly MB, et al. Factors associated with behavioral and cognitive abnormalities in children receiving topiramate. Pediatr Neurol 2000; 22: 200–203.

Glauser TA. Expanding first-line therapy options for children with partial seizures. Neurology 2000; 55: S30–S37.

Guerrini R, Dravet C, Genton P, et al. Lamotrigine and seizure aggravation in severe myoclonic epilepsy. Epilepsia 1998; 39: 508–512.

Hussein G, Troupin AS, Montouris G. Gabapentin interaction with felbamate. Neurology 1996; 47: 1106.

Kaminska A, Ickowicz A, Plouin P, et al. Delineation of cryptogenic Lennox–Gastaut syndrome and myoclonic astatic epilepsy using multiple correspondence analysis. Epilepsy Res 1999; 36: 15–29.

Kwan P, Brodie MJ. Early identification of refractory epilepsy. New Engl J Med 2000; 342: 314–319.

Lane HY, Su KP, Chang WH, et al. Elevated plasma clozapine concentrations after phenobarbital discontinuation. J Clin Psychiatr 1998; 59: 131–133.

Langbehn DR, Alexander B. Increased risk of side-effects in psychiatric patients treated with clozapine and carbamazepine: a reanalysis. Pharmacopsychiatry 2000; 33: 196.

Levy R, Loiseau P, Guyot A, et al. Effects of stiripentol on valproate plasma level and metabolism. Epilepsia 1987; 28(5): 605.

Lortie A, Chiron C, Mumford J, et al. The potential for increasing seizure frequency, relapse, and appearance of new seizure types with vigabatrin. Neurology 1993; 43: S24–S27.

Loscher W, Nau H. Pharmacological evaluation of various metabolites and analogues of valproic acid. Anticonvulsant and toxic potencies in mice. Neuropharmacology 1985; 24: 427–435. Luna D, Dulac O, Pajot N, et al. Vigabatrin in the treatment of childhood epilepsies: a single-

blind placebo-controlled study. Epilepsia 1989; 30: 430–437.
May TW, Rambeck B, Jurgens U. Influence of oxcarbazepine and methsuximide on lamotrigine

concentrations in epileptic patients with and without valproic acid comedication: results of a

retrospective study. Ther Drug Monit 1999; 21: 175–181.
Mesdjian E, Dravet C, Cenraud B, et al. Carbamazepine intoxication due to triacetylolean-

domycin administration in epileptic patients. Epilepsia 1980; 21: 489–496.
Mikaeloff Y, Saint-Martin A, Mancini J, et al. Topiramate: efficacy and tolerability in children

according to epilepsy syndromes. Epilepsy Res 2003; 53: 225–232.
Morselli PL, Principi N, Tognoni G, et al. Diazepam elimination in premature and full term

infants, and children. J Perinat Med 1973; 1: 133–141.
Motte J, Trevathan E, Arvidsson JF, et al. Lamotrigine for generalized seizures associated with

the Lennox–Gastaut syndrome. Lamictal Lennox–Gastaut Study Group. New Engl J Med 1997; 337: 1807–1812.

271 Antiepileptic drug interactions in children

Nau H, Loscher W, Schafer H. Anticonvulsant activity and embryotoxicity of valproic acid. Neurology 1984; 34: 400–401.

Perez J, Chiron C, Musial C, et al. Stiripentol: efficacy and tolerability in children with epilepsy. Epilepsia 1999; 40: 1618–1626.

Perucca E, Gram L, Avanzini G, et al. Antiepileptic drugs as a cause of worsening seizures. Epilepsia 1998; 39: 5–17.

Pisani F, Oteri G, Russo MF, et al. The efficacy of valproate–lamotrigine comedication in refract- ory complex partial seizures: evidence for a pharmacodynamic interaction. Epilepsia 1999; 40: 1141–1146.

Rafi JA, Frazier LM, Driscoll-Bannister SM, et al. Effect of over-the-counter cimetidine on phenytoin concentrations in patients with seizures. Ann Pharmacother 1999; 33: 769–774. Rating D, Wolf C, Bast T. Sulthiame as monotherapy in children with benign childhood epilepsy

with centrotemporal spikes: a 6-month randomized, double-blind, placebo-controlled study.

Sulthiame Study Group. Epilepsia 2000; 41: 1284–1288.
Rey E, Bulteau C, Motte J, et al. Multicenter study of oxcarbazepine pharmacokinetics and

long-term tolerability as open-label, add-on therapy in children with refractory epilepsy. J Clin

Pharmacol 2004; 44: 1290–1300.
Reynolds EH, Shorvon SD. Monotherapy or polytherapy for epilepsy? Epilepsia 1981; 22:

1–10.
Ritter F, Glauser TA, Elterman RD, et al. Effectiveness, tolerability, and safety of topiramate in

children with partial-onset seizures. Topiramate YP Study Group. Epilepsia 2000; 41(Suppl. 1):

S82–S85.
Roger J. Epileptic Syndromes in Infancy, Childhood and Adolescence. M. Bureau, C. Dravet,

F. Dreifuss, A. Perret, P. Wolf, eds. London: John Libbey, 1992.
Rosche J, Froscher W, Abendroth D, et al. Possible oxcarbazepine interaction with cyclosporine

serum levels: a single case study. Clin Neuropharmacol 2001; 24: 113–116.
Sachdeo RC, Glauser TA, Ritter F, et al. A double-blind, randomized trial of topiramate in

Lennox–Gastaut syndrome. Topiramate YL Study Group. Neurology 1999; 52: 1882–1887. Sallas WM, Milosavljev S, D’Souza J, et al. Pharmacokinetic drug interactions in children taking

oxcarbazepine. Clin Pharmacol Ther 2003; 74: 138–149.
Sato S, White BG, Penry JK, et al. Valproic acid versus ethosuximide in the treatment of absence

seizures. Neurology 1982; 32: 157–163.
Schlumberger E, Chavez F, Palacios L, et al. Lamotrigine in treatment of 120 children with

epilepsy. Epilepsia 1994; 35: 359–367.
Schoeman JF, Elyas AA, Brett EM, et al. Correlation between plasma carbamazepine-10,

11-epoxide concentration and drug side-effects in children with epilepsy. Dev Med Child

Neurol 1984; 26: 756–764.
Serdaroglu G, Kurul S, Tutuncuoglu S, et al. Oxcarbazepine in the treatment of childhood

epilepsy. Pediatr Neurol 2003; 28: 37–41.
Siegel H, Kelley K, Stertz B, et al. The efficacy of felbamate as add-on therapy to valproic acid in

the Lennox–Gastaut syndrome. Epilepsy Res 1999; 34: 91–97.
Talwar D, Arora MS, Sher PK. EEG changes and seizure exacerbation in young children treated

with carbamazepine. Epilepsia 1994; 35: 1154–1159.

272 Olivier Dulac et al.

Thanh TN, Chiron C, Dellatolas G, et al. Long-term efficacy and tolerance of stiripentol in severe myoclonic epilepsy of infancy (Dravet’s syndrome). Arch Pediatr 2002; 9: 1120–1127.

Thompson PA, Mosley CA. Tacrolimus–phenytoin interaction. Ann Pharmacother 1996; 30: 544. Tran A, Vauzelle-Kervroedan F, Rey E, et al. Effect of stiripentol on carbamazepine plasma con-

centration and metabolism in epileptic children. Eur J Clin Pharmacol 1996; 50: 497–500. Turnbull DM, Rawlins MD, Weightman D, et al. A comparison of phenytoin and valproate in

previously untreated adult epileptic patients. J Neurol Neurosurg Psychiatr 1982; 45: 55–59. Ueberall MA. Normal growth during lamotrigine monotherapy in pediatric epilepsy patients –

a prospective evaluation of 103 children and adolescents. Epilepsy Res 2001; 46: 63–67.
Vecht CJ, Wagner GL, Wilms EB. Interactions between antiepileptic and chemotherapeutic

drugs. Lancet Neurol 2003; 2: 404–409.
Veggiotti P, Cieuta C, Rex E, et al. Lamotrigine in infantile spasms. Lancet 1994; 344: 1375–1376. Verity CM, Hosking G, Easter DJ. A multicentre comparative trial of sodium valproate and

carbamazepine in paediatric epilepsy. The Paediatric EPITEG Collaborative Group. Dev Med

Child Neurol 1995; 37: 97–108.
Villeneuve N, Soufflet C, Plouin P, et al. Treatment of infantile spasms with vigabatrin as first-line

therapy and in monotherapy: apropos of 70 infants. Arch Pediatr 1998; 5: 731–738.
Vu VT, Bai SA, Abramson FP. Interactions of phenobarbital with propranolol in the dog. 2.

Bioavailability, metabolism and pharmacokinetics. J Pharmacol Exp Ther 1983; 224: 55–61. Wasfi IA, Tanira MO. The effect of chronic administration of cyclosporin A on phenytoin

pharmacokinetic parameters in the rat. Life Sci 1993; 52: 199–204.
Yusof WZ. Gingival hyperplasia: an intra-oral side effect of phenytoin, nifedipine and cyclo-

sporine therapies. Singapore Med J 1988; 29: 498–503.

Antiepileptic drug interactions in the elderly

Jeannine M. Conway and James C. Cloyd

Epilepsy Research and Education Program, College of Pharmacy, University of Minnesota, Minneapolis, MN, USA Department of Experimental and Clinical Pharmacology, University of Minnesota, Minneapolis, MN, USA

Introduction

The elderly ( 65 years) are the fastest growing segment of the population in developed countries. In the USA, older adults presently comprise 13% of the pop- ulation and are projected to increase to 20% within the next 20 years. Similar demographics exist for many European countries. With advancing age comes increasing morbidity, medication use, and adverse drug reactions. Over two-thirds of older adults have one or more chronic medical problems (Hoffman et al., 1996). As a consequence more elderly take medications than others and the elderly take more drugs per person. In the USA, almost 90% of community-dwelling elderly take one or more medications (Guay et al., 2003). Antiepileptic drugs (AEDs) are fre- quently prescribed in the elderly due to the high prevalence of AED-treatable neu- ropsychiatric disorders in this age group. For example, epilepsy is twice as common in those 65 years (1.5%) than in younger adults (Hauser, 1997). An estimated 1.6% of community-dwelling elderly take one or more AEDs (Nitz et al., 2000).

AED use is even greater among elderly nursing home residents. Based on two national surveys, approximately 10–11% of elderly nursing home residents take at least one AED and within this group 14–19% are on two or more AEDs including combinations known to interact (Schachter et al., 1998; Garrard et al., 2000).

Factors contributing to AED interactions in the elderly

There are several factors associated with AED therapy in the elderly that substan- tially increase the risk of clinically significant drug interactions. These include mul- tiple medication use including many drugs with a high potential for interactions, altered sensitivity to drug action, and age-related changes in drug disposition.

Pharmacoepidemiology

The probability of an interaction increases significantly with the number of medica- tions a person takes (Nolan and O’Malley, 1989). Community-dwelling elderly take

273

274 Jeannine M. Conway and James C. Cloyd

3.1–7.9 prescription and non-prescription medications whereas nursing home residents take an average of 7.2 maintenance and pro re nata (PRN) medications (Beers et al., 1993; Stewart, 2001). The most common types of medication used include cardiovascular, gastrointestinal, central nervous system (CNS), analgesic, and vitamin agents, all of which have the potential to interact with other medications (Guay et al., 2003).

The AEDs most commonly prescribed for older patients also have the greatest potential for drug interactions. In a survey based on 1995 data, the vast majority of community-dwelling elders on an AED were on one or more of the following: phenytoin (PHT), carbamazepine (CBZ), phenobarbital (PB), and valproic acid (VPA) (Nitz et al., 2000). More recent studies in nursing home residents reveal a similar pattern although there is greater use of gabapentin (GBP) and clonazepam (Schachter et al., 1998; Garrard et al., 2003). Between 14% and 19% of nursing home elderly, who receive at least one AED, receive two or more AEDs, with the most fre- quently occuring combinations being PHT and CBZ, PB or VPA (Schachter et al., 1998; Garrard et al., 2000). All these AED combinations are known to interact with each other.

The types of co-medication used by community-dwelling elderly taking AEDs have not been characterized, but AED use in nursing home elderly has been exten- sively studied. Elderly nursing home residents on an AED take more medications than other elderly residents. In one study, those receiving an AED were on 5.6 maintenance medications versus 4.6 for all other elderly residents (Lackner et al., 1998). The most commonly prescribed co-medications in this study included CNS drugs, cardiovascular agents, and anticoagulants, all of which have the potential to interact with AEDs (Figure 15.1).

Age-related alterations in pharmacodynamics

The elderly exhibit altered pharmacodynamics resulting in greater sensitivity to both pharmacological and toxicological drug effects. This can produce either a more nar- row therapeutic range or a shift downward in the lower and upper limits of the range. Older persons on AEDs appear to be more sensitive to drug effects even when con- centrations are controlled. Ramsay et al. analyzed the effect of advancing age on occur- rence of adverse effects in a controlled clinical trial comparing the safety and efficacy of CBZ and VPA (Ramsay et al., 1994). They found that patients over 65 years of age experienced side effects at CBZ and VPA concentrations of 50% and 20%, respectively, lower than in younger patients. In the face of increased sensitivity to pharmacological and toxicological effects, elderly patients are more likely to experience a clinically sig- nificant pharmacodynamic drug interaction with certain drug combinations. For example, elderly taking both PB for epilepsy and a benzodiazepine for sleep are more likely to have cognitive impairment than with either drug taken alone (Michelucci and

275 Antiepileptic drug interactions in the elderly

Drug category Antidepressant

Antipsychotics Benzodiazepines Thyroid supplement Antacid AEDs Calcium channel blockers Warfarin Cimetidine

12.7

14 8

12 7

5.9 2.5

Figure 15.1

Medication use in elderly nursing home residents on AEDs (adapted from Lackner et al., 1998)

0 5 10 15 20 25 30 AED recipients (%)

Tassinari, 2002). A pharmacodynamic interaction resulting in decreased effectiveness of AED therapy occurs when medications that lower seizure threshold are added to a patient’s regimen, as may occur with antipsychotics (Lader, 1999).

Age-related alterations in pharmacokinetics

The most common AED interactions in older patients are associated with either an increase or decrease in one or both interacting drugs. The elderly are particularly sus- ceptible to pharmacokinetic interactions due to age-related changes in drug disposi- tion (Mayersohn, 1992) (Table 15.1). Medical conditions common in the elderly such as cardiovascular, renal or gastrointestinal diseases further alter drug disposi- tion. Advanced age is associated with increased gastric pH, diminished gastrointesti- nal fluids, slower intestinal transit, and reduced absorptive area. Each of these changes can affect either or both the rate and extent of absorption. Age-related reduction in intestinal and hepatic blood flow, intestinal drug transport and metab- olism, and hepatic metabolism can also affect the systemic bioavailability of some drugs. Gastric pH and intestinal transit time may exhibit intra-patient day-to-day variability while other processes tend to slowly decline. Transporter proteins, such as P-glycoprotein, which are located in intestinal enterocytes, facilitate efflux of certain drugs thereby reducing bioavailability. It is not known if advancing age alters the activity of transporter enzymes. Age-related alterations in absorption are most likely to affect slowly absorbed AEDs, particularly those administered as solid dosage

276 Jeannine M. Conway and James C. Cloyd

Table 15.1 Age-related changes in physiology

Absorption

Gastrointestinal blood flow Gastric pH
Gastric emptying Intestinal motility

Distribuition

Lean body mass Body fat
Plasma albumin

Metabolism

Liver mass
Hepatic blood flow
Oxidative metabolism Conjugation metabolism Induction of microsomal enzymes

Excretion

Kidney mass
Renal blood flow Glomerular filtration Filtration fraction

Decreased Decreased Decreased Decreased

Decreased
Increased
Unchanged or decreased

Decreased
Decreased
Decreased by 1%/year; age: 40 years Unchanged or decreased
Decreased?

Decreased
Decreased
Decreased by 1%/year; age: 40 years Decreased

Adapted from Cloyd and Conway (2002).

forms, extended release formulations, or drugs absorbed by active transport (GBP). A recent report described highly variable PHT concentrations collected from 56 eld- erly nursing home residents on constant maintenance doses of PHT with no changes in interacting co-medication. Although there is no direct evidence, an alteration in bioavailability is the most likely explanation for this phenomenon (Birnbaum et al., 2003). The addition of drugs that affect AED absorption can compound the effects of age-related changes in gastrointestinal function. For example, calcium-containing products are commonly used in elderly patients, particularly women. These products can chelate with PHT resulting in decreased bioavailability (Cacek, 1986).

Older persons undergo a gradual reduction in serum albumin: by age 65 years, many individuals have low normal albumin concentrations or are frankly hypo- albuminemic (Wallace and Verbeeck, 1987). Albumin concentration may be further reduced by conditions such as malnutrition, renal insufficiency, and rheumatoid arthritis. As serum albumin levels decline, the greater the likelihood that drug binding will decrease. This has the effect of lowering the total serum drug concen- tration while unbound serum drug concentration remains unchanged. The elderly

277 Antiepileptic drug interactions in the elderly

are more susceptible to protein-binding interactions due to lower serum albumin levels and the use of multiple medications, many of which are highly bound to serum proteins. Protein-binding competitive displacement interactions, which are the most common, occur when the unbound concentration of the displacer drug or its binding affinity is greater than that of the displaced drug (MacKichan, 1992). Displacer drug concentrations are often higher in elderly patients due to reduced clearance. Hence elderly patients are likely to have a displacement interaction and the extent of displacement will be greater than in younger patients. Protein-binding interactions complicate interpretation of total serum AED concentrations. Measurement of unbound AED concentrations may be useful in assessing the clin- ical significance of this type of interaction.

Several studies have shown that both hepatic and renal drug clearances decline at a rate of 10% per decade of life beginning at the age 40 years (Mayersohn, 1992). Age-related decreases in clearance result in higher drug concentrations when stan- dard doses are used in older persons. Most interactions with drugs that inhibit clear- ance are concentration dependent. When a standard dose of an inhibiting drug is given to an elderly patient, its concentration will be higher and, hence, its inhibition of the affected drug’s clearance will be greater than in a younger adult. If the elderly patient is also taking a standard dose of the affected drug, the greater decrease in its clearance results in a further increase in concentration that was already elevated as compared to a younger adult on the same dose. As a result drug interactions that are clinically important in younger adults will have an even greater impact in the elderly; and drug combinations not known to interact in younger adults may be clinically important in older patients. Even when dosage adjustments are made to the inhibitor and the affected drug, the concentration of the latter can still increase if the inhibitor concentration approaches or exceeds its Ki. In this situation, the increased pharma- codynamic sensitivity in the elderly can result in an adverse drug interaction although the rise in the concentration of the affected drug is limited.

Drug interactions occurring as a result of induction of hepatic metabolism follow a similar pattern. In most cases, induction of hepatic metabolism is concen- tration dependent although there is some controversy as to whether the elderly respond to inducers to the same extent as younger adults (Mayersohn, 1992). Con- sequently, standard doses of the inducer may result in higher concentrations that, in turn, may cause a greater extent of induction.

Finally, management of pharmacokinetic interactions must consider both initiation and discontinuation of drug therapy. If appropriate dosage adjustments have been made to control drug concentrations in the presence of an interaction, the concentra- tion of the affected drug will fall or rise once the inhibitor or inducer is withdrawn. As the elderly are likely to have a more narrow therapeutic range for many medications, the clinical impact of withdrawing an interacting drug can also be significant.

278 Jeannine M. Conway and James C. Cloyd

AEDs versus other drug combinations

The elderly frequently take numerous medications for a variety of medical condi- tions. The use of polypharmacotherapy leaves the elderly patient at an increased risk for adverse events. There are many medications that are frequently prescribed for ailments that the elderly experience and unfortunately this chapter cannot address every possible drug interaction.

Antihypertensives

Carbamazepine

Diltiazem and verapamil are inhibitors of cytochrome P450 (CYP) 3A4, which is the major metabolic elimination pathway for CBZ (Ma et al., 2000). This inhibition may lead to increased CBZ blood concentrations and neurotoxicity (Macphee et al., 1986; Eimer and Carter, 1987; Beattie et al., 1988; Bahls et al., 1991; Shaughnessy and Mosley, 1992). CBZ is a potent inducer of CYP3A4 (Luo et al., 2002). As a result, any medica- tions that are metabolized via that pathway are likely to be affected. CBZ also induces CYP1A2, CYP2C9, and to a variable degree CYP2C19. Patients will likely require increased doses of affected antihypertensives to decrease blood pressure adequately.

Phenorbital

Phenytoin

Other

Diltiazem has been demonstrated to cause PHT toxicity, most likely due to enzyme inhibition although the exact mechanism is unclear (Bahls et al., 1991; Clarke et al., 1993). Careful monitoring for PHT toxicity is recommended if diltiazem is pre- scribed. As PHT may cause an induction in metabolism, antihypertensives that are metabolized by CYP P450 may be affected.

There are no known drug interactions with antihypertensives with the following AEDs: felbamate (FBM), GBP, lamotrigine (LTG), levetiracetam (LEV), oxcarbazepine (OXC), tiagabine (TGB), topiramate (TPM), VPA, and zonisamide (ZNS).

PB is a significant enzyme inducer of CYP3A4, CYP2C8, CYP2C9, and CYP2C19 (Gerbal-Chaloin et al., 2001; Raucy et al., 2002; Edwards et al., 2003). Any anti- hypertensive medication that is metabolized via these metabolic enzymes is likely to fall victim to increased metabolism. Hence, if an elderly patient is on PB they may require higher doses of their antihypertensive to get a therapeutic response. Conversely, if PB therapy is initiated, the existing antihypertensive medication may lose its efficacy.

279 Antiepileptic drug interactions in the elderly

Antihyperlipidemics

Carbamazepine

CBZ is a substrate and inducer of CYP3A4 (Luo et al., 2002). A recent study demon- strated that CBZ significantly increases the clearance of simvastatin. This resulted in an 80% decrease in patient exposure to simvastatin (Veor et al., 2004). While there are no published reports of drug interactions with other antihyperlipidemics it may be inferred depending on their metabolic pathway including atorvastatin and lovastatin (Wang et al., 1991; Mazzu et al., 2000; Paoletti et al., 2002). CBZ co-medication may increase a patient’s dose requirement in order to get an adequate therapeutic response.

Phenobarbital

Atorvastatin, lovastatin, and simvastatin are substrates of CYP3A4 (Wang et al., 1991; Mazzu et al., 2000; Paoletti et al., 2002). PB co-medication may increase a patient’s dose requirement in order to get an adequate therapeutic response.

Phenytoin

Other

Atorvastatin, lovastatin, and simvastatin are substrates of CYP3A4 (Wang et al., 1991; Mazzu et al., 2000; Paoletti et al., 2002). PHT co-medication may increase a patient’s dose requirement in order to get an adequate therapeutic response.

There are no known drug interactions with antihyperlipidemics with the following AEDs: FBM, GBP, LTG, LEV, OXC, TGB, TPM, VPA, and ZNS.

Anticoagulants/antiplatelets

Carbamazepine

Warfarin is metabolized via several CYP P450 enzymes (Kaminsky and Zhang, 1997). If a patient is stabilized on CBZ and warfarin therapy is initiated, the patient will require a larger warfarin dose than patients not receiving CBZ will require (Ross and Beeley, 1980; Kendall and Boivin, 1981; Massey, 1983). If a patient is sta- bilized on both CBZ and warfarin and the CBZ is discontinued, the patient is very likely to experience an increase in their internationalized normalized ratio (INR) that would place the patient at an increased risk of bleeding (Denbow and Fraser, 1990).

Felbamate

There are no known documented drug interactions between anticoagulants and FBM, but FBM inhibits CYP2C19, which may increase the effect of warfarin, result- ing in an increased risk of bleeding (Glue et al., 1997).

280 Jeannine M. Conway and James C. Cloyd

Oxcarbazepine

There are no known documented drug interactions with anticoagulants or antiplatelets, but OXC inhibits CYP2C19 and induces CYP3A4, which may alter the metabolism of warfarin (Trileptal, 2001). Caution should be taken if adding OXC to a medication regimen that includes warfarin.

Phenobarbital

The clearance of warfarin is increased when a patient is also on PB (Udall, 1975; Mungall et al., 1985). Management of this interaction is similar to the interaction of warfarin with CBZ. If PB is added to a medication regimen including warfarin, the practitioner needs to monitor for a decrease in INR and efficacy. If PB is removed from a stable medication regimen including warfarin, the practitioner needs to monitor for an increased INR and an increased risk of bleeding. The dose of warfarin will need to be appropriately decreased.

Phenytoin

Topiramate

Valproate

There are no known drug interactions between TPM and anticoagulants or antiplatelets. TPM is a weak inhibitor of CYP2C19 and an inducer of CYP3A4 (Benedetti, 2000). Caution should be taken when prescribing TPM with clopido- grel, ticlopidine, and warfarin, as the presence or absence of drug interactions is not established.

Aspirin may displace VPA from protein-binding sites and inhibit metabolism (Goulden et al., 1987). Patients should be monitored for VPA toxicity while taking aspirin. VPA may displace warfarin from protein-binding sites in vitro (Depakote, 2002). Caution should be taken when prescribing VPA with warfarin.

The interaction between PHT and warfarin is unpredictable, with reports of increased and decreased effects of warfarin (Nappi, 1979; Levine and Sheppard, 1984; Panegyres and Rischbieth, 1991). The initiation of PHT may cause warfarin to be displaced from protein-binding sites, followed by an increase in the metabolism of warfarin (Levine and Sheppard, 1984). Caution must be used when managing a patient on warfarin and PHT. Ticlopidine inhibits CYP2C19 that may cause inhibi- tion of the metabolism of PHT, resulting in toxicity (Klaassen, 1998; Donahue et al., 1999). Aspirin may cause protein-binding displacement of PHT at doses that exceed 650 mg every 4 h (Leonard et al., 1981). There is no evidence that aspirin dosed at 81–325 mg per day should cause a clinically significant interaction.

281 Antiepileptic drug interactions in the elderly

Other

Analgesics

Carbamazepine

Fentanyl is a substrate of CYP3A4 and the addition of CBZ may reduce its effec- tiveness (Labroo et al., 1997; Duragesic, 2001). There is no documented interaction between CBZ and valdecoxib. Valdecoxib is metabolized via CYP3A4, CYP2C9, and glucuronidation (Bextra, 2002). It is an inhibitor of CYP2C19 and a weak inhibitor of CYP3A4 and CYPC9 (Bextra, 2002). It may be hypothesized that CBZ may induce the metabolism of valdecoxib, reducing its effectiveness, or valdecoxib may inhibit the metabolism of CBZ, resulting in toxicity. The drug interaction between propoxyphene and CBZ has been reported numerous times in the literature (Kubacka and Ferrante, 1983; Yu et al., 1986; Oles et al., 1989; Allen, 1994; Bergendal et al., 1997). Propoxyphene appears to inhibit the metabolism of CBZ resulting in toxicity. This combination should be cautiously used. CBZ increases the metabo- lism of tramadol that may decrease its efficacy at usual doses (Ultram, 2000).

Lamotrigine

Phenobarbital

There is no documented drug interaction between celecoxib and PB. Celecoxib is metabolized via CYP2C9 and PB induces CYP2C9 (Tang et al., 2000; Raucy et al., 2002). It is possible that PB will reduce the efficacy of celecoxib. There is no docu- mented drug interaction between valdecoxib and PB. Valdecoxib is metabolized via CYP3A4, CYP2C9, and glucuronidation (Bextra, 2002). It is an inhibitor of CYP2C19 and a weak inhibitor of CYP3A4 and CYP2C9 (Bextra, 2002). PB may induce the metabolism of valdecoxib, reducing its effectiveness, or valdecoxib may inhibit the metabolism of PB resulting in toxicity. Propoxyphene may cause up to a 20% increase in PB blood concentrations (Hansen et al., 1980). Patients should be monitored for toxicity.

Phenytoin

There are no known drug interactions between anticoagulants/antiplatelets and the following AEDs: GBP, LTG, LEV, TGB, or ZNS.

Only one interaction with analgesics has been reported with LTG. One study examined the pharmacokinetics of a single dose of LTG following multiple doses of acetaminophen. The investigators found that it appears that acetaminophen increases the clearance of LTG (Depot et al., 1990). It is not clear if this interaction is clinically significant.

PHT induces the metabolism of acetaminophen, resulting in increased clearance and a decrease in the duration of analgesia (Miners et al., 1984). Aspirin may cause

282 Jeannine M. Conway and James C. Cloyd

Valproate

Other

displacement of PHT from its protein-binding sites at high doses ( 650 mg every 4 h), however lower-dose aspirin ( 650 mg every 4 h) should not be very prob- lematic (Leonard et al., 1981). There is no documented drug interaction between valdecoxib and PHT. Valdecoxib is metabolized via CYP3A4, CYP2C9, and glu- curonidation (Bextra, 2002). It is an inhibitor of CYP2C19 and a weak inhibitor of CYP3A4 and CYP2C9 (Bextra, 2002). PHT may induce the metabolism of valde- coxib, reducing its effectiveness, or valdecoxib may inhibit the metabolism of PHT resulting in toxicity. Propoxyphene inhibits CYP2C9 and may increase PHT serum concentrations resulting in increased toxicity (Levy, 1995).

Aspirin may displace VPA from protein-binding sites and inhibit metabolism (Goulden et al., 1987). Patients should be monitored for VPA toxicity while taking aspirin.

There are no known drug interactions between analgesics with the following AEDs: FBM, GBP, LEV, OXC, TGB, TPM, and ZNS.

Gastrointestinal agents

Carbamazepine

Cimetidine is a modest inhibitor of CYP3A4 (Martinez et al., 1999). The addition of cimetidine may result in CBZ toxicity.

Gabapentin

Phenobarbital

There are no known drug interactions between gastrointestinal agents and PB. Interactions may be possible depending on the metabolism of the gastrointestinal agents and the pathways that PB induces.

Phenytoin

Concomitant use of antacids (Maalox®) has been shown to decrease the absorp- tion of GBP by 20% (Neurontin, 2002). The manufacturer recommends taking antacids and GBP at least 2 h apart.

Antacids, when taken simultaneously with PHT, may create an insoluble complex resulting in decreased or erratic absorption of PHT (Carter et al., 1981; McElnay et al., 1982). To avoid any potential interaction, it is recommended that patients take antacids and PHT at least 2 h apart. Cimetidine appears to cause a decrease in the clearance of PHT resulting in toxicity (Algozzine et al., 1981; Hetzel et al., 1981; Bartle et al., 1983; Frigo et al., 1983). The mechanism of the drug interaction is

283 Antiepileptic drug interactions in the elderly

Valproate

Other

likely to be the inhibition of CYP2C19 (Furuta et al., 2001). Caution should be taken when prescribing cimetidine with PHT. There is potential for a drug interaction between omeprazole and PHT (Prichard et al., 1987). Omeprazole is a potent inhibitor of CYP2C19 that may cause inhibition of PHT metabolism resulting in toxicity (Furuta et al., 2001). Careful monitoring is warranted.

One study, of six subjects, demonstrated decreased clearance of a single dose of VPA when cimetidine was also administered (Webster et al., 1984). It is not known if cimetidine interacts with multiple doses of VPA, but caution should be taken if prescribing cimetidine and VPA.

There are no known drug interactions with gastrointestinal agents and the follow- ing AEDs: FBM, LTG, LEV, OXC, TGB, TPM, and ZNS.

Endocrine/metabolic agents

Carbamazepine

There is no documented interaction between hormone replacement therapy and CBZ but it is well known that CBZ increases the metabolism of hormones (Ramsay and Slater, 1991). It would be expected that women who choose hormone replace- ment therapy might require higher doses of hormones for control of menopausal symptoms. There is no documented interaction with pioglitazone and CBZ. Pioglitazone is metabolized via CYP3A4 and may undergo increased metabolism secondary to CBZ induction (Actos, 2002).

Felbamate

Lamotrigine

Oxcarbazepine

There is no documented interaction between hormone replacement therapy and OXC. There is one study that examined the effect of OXC on oral contraceptives that

There is no documented interaction between hormone replacement therapy and FBM. There was one study that examined the effect of FBM on oral contraceptives and it was found that FBM decreases hormone blood levels (Saano et al., 1995). It is unknown if FBM would also adversely affect blood levels of hormone replacement therapy.

There is no documented interaction between hormone replacement therapy and LTG but it has been demonstrated that oral contraceptives increase the clearance of LTG by as much as 50% (Jaben et al., 2003). LTG clearance may be increased by hormone replacement therapy.

284 Jeannine M. Conway and James C. Cloyd

found OXC decreases hormone blood levels (Fattore et al., 1999). It is unknown if OXC would also adversely affect blood levels of hormone replacement therapy.

Phenobarbital

Both glypizide and tolbutamide are CYP2C9 substrates (Kidd et al., 1999; Kirchheiner et al., 2002). There is no documented interaction with PB and glypizide or tolbutamide. PB may cause some enzyme induction altering the metabolism of both agents (Gerbal-Chaloin et al., 2001). Close monitoring of therapeutic response and glucose levels is warranted. There is no documented interaction between hor- mone replacement therapy and PB. There is evidence that PB induces the metabo- lism of hormones and monitoring of therapy is recommended (Ramsay and Slater, 1991). There is no documented drug interaction between pioglitazone and PB. Pioglitazone is metabolized via CYP3A4 and may be susceptible to increased metab- olism secondary to PB co-medication (Actos, 2002; Luo et al., 2002; Edwards et al., 2003). There is no documented drug interaction between rosiglitazone and PB. Rosiglitazone is metabolized via CYP2C8, a metabolic pathway induced by PB (Baldwin et al., 1999; Gerbal-Chaloin et al., 2001).

Phenytoin

Topiramate

Other

There is no documented drug interaction between hormone replacement therapy and TPM. There is one study that documented a 14–33% increase in clearance of ethinyl estradiol when administered with TPM (Rosenfeld et al., 1997). The for- mulation ethinyl estradiol used in this study is an oral contraceptive. It may be extrapolated that TPM may increase the clearance of other estrogen supplements.

There are no known drug interactions between endocrine/metabolic agents and the following AEDs: GBP, LEV, TGB, VPA, and ZNS.

There is no documented interaction between PHT and glypizide or tolbutamide. Both glypizide and tolbutamide are CYP2C9 substrates (Kidd et al., 1999; Kirchheiner et al., 2002). PHT may cause some enzyme induction altering the metabolism of both agents. Close monitoring of therapeutic response and glucose levels is warranted. There is no documented drug interaction between hormone replacement therapy and PHT. However, there is evidence that PHT increases the clearance of oral contracep- tives (Coulam and Annegers, 1979; Mattson et al., 1986). It may be extrapolated that PHT will increase the clearance of other hormone replacement therapy. There is no documented drug interaction between pioglitazone and PHT. Pioglitazone is metab- olized via CYP3A4 and may be susceptible to increased metabolism secondary to PHT co-medication (Actos, 2002; Luo et al., 2002).

285 Antiepileptic drug interactions in the elderly

Respiratory agents

There are no established interactions between inhaled respiratory medications and the AEDs.

Phenobarbital

There is a study of six adults that demonstrated an increased clearance of theo- phylline when they were also receiving PB (Landay et al., 1978). It is unknown what the significance of this drug interaction is in the elderly. It may be inferred that they would require larger doses of theophylline while being treated with PB.

Phenytoin

CNS agents

Carbamazepine

There are numerous antidepressants whose metabolism is increased by CBZ, including tricyclic antidepressants, bupropion, mirtazepine, and sertraline (Leinonen et al., 1991; Wellbutrin, 1999; Sitsen et al., 2001; Pihlsgard and Eliasson, 2002). Several antidepres- sants may cause an elevation of CBZ blood concentrations including fluoxetine, flu- voxamine, and nefazodone. Fluoxetine has been shown to cause inhibition of CBZ in one study of six subjects (Grimsley et al., 1991), while another study of eight subjects showed no change in CBZ pharmacokinetics (Spina et al., 1993). Fluvoxamine was hypothesized to cause inhibition of metabolism of CBZ in three cases (Fritze et al., 1991), while a study of seven subjects showed no change in CBZ pharmacokinetics when fluvoxamine was added (Spina et al., 1993). Nefazodone is a CYP3A4 inhibitor (Rotzinger and Baker, 2002). The metabolism of CBZ is decreased when nefazodone is added to a patient’s regimen (Laroudie et al., 2000). Patients should be monitored for signs of CBZ toxicity and increased blood concentrations if nefazodone is prescribed to a patient also on CBZ. The metabolism of olanzapine, an atypical antipsychotic, is increased by CBZ by approximately 40%, which may not be clinically significant since olanzapine has a wide therapeutic range (Lucas et al., 1998; Olesen and Linnet, 1999; Linnet and Olesen, 2002). When CBZ was added to a regimen containing haloperidol, the clearance of haloperidol increased by 60% (Jann et al., 1985). The resulting increase in clearance may lead to treatment failure due to insufficient efficacy (Hesslinger et al., 1999). The metabolism of risperidone may be increased by CBZ (Ono et al., 2002). Alternatively risperidone may modestly increase plasma concentrations of CBZ and its metabolite CBZ-epoxide (CBZ-E) (Mula and Monaco, 2002). Monitoring the efficacy

There are several case reports and studies that have demonstrated that PHT increases the clearance of theophylline (Miller et al., 1984; Sklar and Wagner, 1985; Adebayo, 1988). Patients on PHT may require increased doses of theophylline to get an adequate response.

286 Jeannine M. Conway and James C. Cloyd

of both agents is warranted. There are two case reports of quetiapine being added to a CBZ regimen that was thought to cause toxicity secondary to an increase in CBZ-E (Fitzgerald and Okos, 2002). The metabolism of quetiapine may be increased by CBZ (DeVane and Nemeroff, 2001). CBZ increases the metabolism of ziprasidone but it is not clear if the increase is clinically significant (Miceli et al., 2000).

Donepezil, a reversible inhibitor of acetylcholinesterase, used to treat dementia, is metabolized by CYPZD6, CYP3A4 and glucuronidation (Aricept, 2002). Galantamine, also a reversible inhibitor of acetylcholinesterase used to treat dementia, is metabolized via CYP2D6 and CYP3A4 (Reminyl, 2003). It is expected that CBZ may increase their clearance.

Lamotrigine

Phenobarbital

There are no well-documented drug interaction studies done between antidepres- sants and PB. Since it is established that PB induces CYP2C9, CYP2C19, and CYP3A4, interactions may be inferred by looking at the metabolic pathway of the antidepressant being prescribed (Glue et al., 1997; Raucy et al., 2002; Edwards et al., 2003). The metabolism of clozapine is increased by co-medication with PB (Facciola et al., 1998). It has also been reported that PB may increase the metabolism of haloperidol (Linnoila et al., 1980). An increase in metabolism may result in a loss of efficacy without adequate dose adjustments. It is expected that PB may increase the clearance of donepezil and galantamine (Aricept, 2002; Reminyl, 2003).

Phenytoin

There are two case reports of patients receiving LTG to which sertraline was added to therapy resulting in toxicity secondary to increased LTG blood concentrations (Kaufman and Gerner, 1998). It was hypothesized that a glucuronidation pathway interaction may be the cause of the reaction, but further research is needed. Patients on LTG and sertraline should be monitored for signs of toxicity secondary to LTG.

There are several case reports of the addition of fluoxetine to a regimen with PHT resulting in PHT toxicity (Jalil, 1992; Woods et al., 1994). An in vitro study examining the effect of fluoxetine on PHT metabolism demonstrated that fluoxetine inhibited CYP2C9 resulting in impaired metabolism of PHT (Nelson et al., 2001). Fluvoxamine inhibits CYP2C19, which may result in PHT toxicity, and doses may need to be appro- priately adjusted (Schmider et al., 1997; Hemeryck and Belpaire, 2002). Tricyclic anti- depressants may cause inhibition of CYP2C9 and CP2C19 resulting in an increased risk of PHT toxicity (Shin et al., 2002). Monitoring of blood concentrations and dose adjustments of PHT may be necessary. Sertraline has been associated with increased PHT toxicity in a report of two elderly patients (Haselberger et al., 1997). In vitro data also demonstrated that sertraline has the potential to inhibit CYP2C9 (Schmider

287 Antiepileptic drug interactions in the elderly

Valproate

Other

et al., 1997; Nelson et al., 2001). The addition of PHT to quetiapine resulted in a five- fold increase in the metabolism of quetiapine (Wong et al., 2001). Patients should be monitored for a loss of efficacy if PHT is added to a regimen containing quetiapine. It is expected that PHT may increase the clearance of donepezil and galantamine (Aricept, 2002; Reminyl, 2003). It has been reported that PHT decreases the efficacy of levodopa therapy in patients with Parkinson’s; as a result, larger doses of levodopa may be necessary (Mendez et al., 1975).

There are no well-documented clinically significant drug interactions between CNS medications and VPA.

There are no documented drug interactions between CNS agents and the follow- ing AEDs: FBM, GBP, LEV, OXC, TGB, TPM, and ZNS.

Conclusion

AED interactions in the elderly are common and often lead to serious adverse events. A growing number of elderly are taking AEDs, usually in combination with other medications. Older patients appear to be more sensitive to adverse effects even when drug concentrations are controlled. Age-related changes in drug disposition and the use of multiple medications greatly increase the risk of clinically significant inter- actions in older patients. A number of AEDs either induce or inhibit drug metabo- lizing enzymes and, in turn, their metabolism is affected by many co-medications. Clinicians and older patients need to recognize that the addition or discontinuation of medications can place the patient at risk of an adverse event due to a drug inter- action. An understanding of the principles that determine interactions and the phar- macokinetics of specific AEDs and other medications permits prospective assessment of the risk of an interaction when a drug is added or stopped. This allows clinicians to avoid interactions by selecting an alternate medication or rationally managing an interaction when it cannot be avoided. Several of the newer AEDs do not appear to interact with other medications, while others are affected by enzyme induction of inhibition but do not appear to alter the disposition of co-medications. Thus, the newer AEDs may be particularly useful in older patients.

REFERENCES

Actos (2002). Package Insert. Takeda Pharmaceuticals America, Inc.
Adebayo GI. Interaction between phenytoin and theophylline in healthy volunteers. Clin Exp

Pharmacol Physiol 1988; 15(11): 883–887.

288 Jeannine M. Conway and James C. Cloyd

Algozzine GJ, Stewart RB, Springer PK. Decreased clearance of phenytoin with cimetidine. Ann Intern Med 1981; 95(2): 244–245.

Allen S. Cerebellar dysfunction following dextropropoxyphene-induced carbamazepine toxicity. Postgrad Med J 1994; 70(828): 764.

Aricept (2002). Package Insert. Eisai Co, Ltd.
Bahls FH, Ozuna J, Ritchie DE. Interactions between calcium channel blockers and the anti-

convulsants carbamazepine and phenytoin. Neurology 1991; 41(5): 740–742.
Baldwin SJ, Clarke SE, Chenery RJ. Characterization of the cytochrome P450 enzymes involved

in the in vitro metabolism of rosiglitazone. Br J Clin Pharmacol 1999; 48(3): 424–432.
Bartle WR, Walker SE, Shapero T. Dose-dependent effect of cimetidine on phenytoin kinetics.

Clin Pharmacol Ther 1983; 33(5): 649–655.
Beattie B, Biller J, Mehlhous B, et al. Verapamil-induced carbamazepine neurotoxicity. A report

of two cases. Eur Neurol 1988; 28(2): 104–105.
Beers MH, Fingold SF, Ouslander JG, et al. Characteristics and quality of prescribing by doctors

practicing in nursing homes. J Am Geriatr Soc 1993; 41(8): 802–807.
Benedetti MS. Enzyme induction and inhibition by new antiepileptic drugs: a review of human

studies. Fundam Clin Pharmacol 2000; 14(4): 301–319.
Bergendal L, Friberg A, Schaffrath AM, et al. The clinical relevance of the interaction between

carbamazepine and dextropropoxyphene in elderly patients in Gothenburg, Sweden. Eur J

Clin Pharmacol 1997; 53(3–4): 203–206.
Bextra (2002). Package Insert. Pfizer.
Birnbaum A, Hardie NA, Leppik IE, et al. Variability of total phenytoin serum concentrations

within elderly nursing home residents. Neurology 2003; 60: 555–559.
Cacek AT. Review of alterations in oral phenytoin bioavailability associated with formulation,

antacids, and food. Ther Drug Monit 1986; 8(2): 166–171.
Carter BL, Garnett WR, Pellock JM, et al. Effect of antacids on phenytoin bioavailability. Ther

Drug Monit 1981; 3(4): 333–340.
Clarke TA, Waskell LA. The metabolism of clopidogrel is catalyzed by human cytochrome P450

3A and is inhibited by atorvastatin. Drug Metab Dispos 2003; 31(1): 53–59.
Clarke WR, Horn JR, Kawabori I, et al. Potentially serious drug interactions secondary to high- dose diltiazem used in the treatment of pulmonary hypertension. Pharmacotherapy 1993;

13(4): 402–405.
Cloyd J, Conway J. Age-related changes in pharmacokinetics, drug interactions, and adverse

effects. In Clinical Neurology of the Older Adult. J. Sirven, B. Malamut, eds. Philadelphia, PA:

Lippincott Williams & Wilkins, 2002: 29–44.
Coulam CB, Annegers JF. Do anticonvulsants reduce the efficacy of oral contraceptives? Epilepsia

1979; 20(5): 519–525.
Denbow CE, Fraser HS. Clinically significant hemorrhage due to warfarin–carbamazepine inter-

action. South Med J 1990; 83(8): 981.
Depakote (2002). Package Insert. Abbott Laboratories.
Depot M, Powell JR, Messenheimer JA, et al. Kinetic effects of multiple oral doses of

acetaminophen on a single oral dose of lamotrigine. Clin Pharmacol Ther 1990; 48(4): 346–355.

289 Antiepileptic drug interactions in the elderly

DeVane CL, Nemeroff CB. Clinical pharmacokinetics of quetiapine: an atypical antipsychotic. Clin Pharmacokinet 2001; 40(7): 509–522.

Donahue S, Flockhart DA, Abernethy DR, et al. Ticlopidine inhibits phenytoin clearance. Clin Pharmacol Ther 1999; 66(6): 563–568.

Duragesic (2001). Package Insert. Janssen.
Edwards RJ, Price RJ, Walts PS, et al. Induction of cytochrome P450 enzymes in cultured

precision-cut human liver slices. Drug Metab Dispos 2003; 31(3): 282–288.
Eimer M, Carter BL. Elevated serum carbamazepine concentrations following diltiazem initia-

tion. Drug Intell Clin Pharm 1987; 21(4): 340–342.
Facciola G, Avenoso A, Spina E, et al. Inducing effect of phenobarbital on clozapine metabolism

in patients with chronic schizophrenia. Ther Drug Monit 1998; 20(6): 628–630.
Fattore C, Cipolla G, Gatti G, et al. Induction of ethinylestradiol and levonorgestrel metabolism

by oxcarbazepine in healthy women. Epilepsia 1999; 40(6): 783–787.
Fitzgerald BJ, Okos AJ. Elevation of carbamazepine-10,11-epoxide by quetiapine. Pharma-

cotherapy 2002; 22(11): 1500–1503.
Frigo GM, Lecchini S, Caranaggi M, et al. Reduction of phenytoin clearance caused by cimeti-

dine. Eur J Clin Pharmacol 1983; 25(1): 135–137.
Fritze J, Unsorg B, Lanczik M, et al. Interaction between carbamazepine and fluvoxamine. Acta

Psychiatr Scand 1991; 84(6): 583–584.
Furuta S, Kamada E, Suzuki T, et al. Inhibition of drug metabolism in human liver microsomes

by nizatidine, cimetidine and omeprazole. Xenobiotica 2001; 31(1): 1–10.
Garrard J, Cloyd J, Gross C, et al. Factors associated with antiepileptic drug use among elderly

nursing home residents. J Gerontol A Biol Sci Med Sci 2000; 55(7): M384–M392.
Garrard J, Harms S, Hardie N, et al. Antiepileptic drug use in nursing home admissions. Annals

of Neurology 2003; 54: 75–85.
Gerbal-Chaloin S, Pascussi JM, Pichard-Garcia L, et al. Induction of CYP2C genes in human

hepatocytes in primary culture. Drug Metab Dispos 2001; 29(3): 242–251.
Glue P, Banfield CR, Perhach JL, et al. Pharmacokinetic interactions with felbamate. In vitro–in

vivo correlation. Clin Pharmacokinet 1997; 33(3): 214–224.
Goulden KJ, Dooley JM, Camfield PR, et al. Clinical valproate toxicity induced by acetylsalicylic

acid. Neurology 1987; 37(8): 1392–1394.
Grimsley SR, Jann MW, Carter JG, et al. Increased carbamazepine plasma concentrations after

fluoxetine coadministration. Clin Pharmacol Ther 1991; 50(1): 10–15.
Guay D, Artz M, Hanlon J, et al. The pharmacology of aging. In Brocklehurst’s Textbook of Geriatric Medicine. R. Tallis, H. Fillit, eds. London, UK: Churchill Livingstone, 2003: 155–161. Hansen BS, Dam M, Brandt J, et al. Influence of dextropropoxyphene on steady state serum levels and protein binding of three anti-epileptic drugs in man. Acta Neurol Scand 1980; 61(6): 357–367. Haselberger MB, Freedman LS, Tolbert S. Elevated serum phenytoin concentrations associated

with coadministration of sertraline. J Clin Psychopharmacol 1997; 17(2): 107–109.
Hauser W. Epidemiology of seizures in the elderly. In Seizures and Epilepsy in the Elderly.

A. Rowan, R. Ramsay, eds. Boston, MA: Butterworth-Heinemann, 1997: 7–20.
Hemeryck A, Belpaire FM. Selective serotonin reuptake inhibitors and cytochrome P-450 medi-

ated drug–drug interactions: an update. Curr Drug Metab 2002; 3(1): 13–37.

290 Jeannine M. Conway and James C. Cloyd

Hesslinger B, Normann C, Langosch JM, et al. Effects of carbamazepine and valproate on haloperidol plasma levels and on psychopathologic outcome in schizophrenic patients. J Clin Psychopharmacol 1999; 19(4): 310–315.

Hetzel DJ, Bochner F, Hallike JF, et al. Cimetidine interaction with phenytoin. Br Med J (Clin Res Ed) 1981; 282(6275): 1512.

Hoffman C, Rice D, Jung HY. Persons with chronic conditions. Their prevalence and costs. J Am Med Assoc 1996; 276(18): 1473–1479.

Jalil P. Toxic reaction following the combined administration of fluoxetine and phenytoin: two case reports. J Neurol Neurosurg Psychiatry 1992; 55(5): 412–413.

Jann MW, Ereshefsky L, Jaklod SR, et al. Effects of carbamazepine on plasma haloperidol levels. J Clin Psychopharmacol 1985; 5(2): 106–109.

Kaminsky LS, Zhang ZY. Human P450 metabolism of warfarin. Pharmacol Ther 1997; 73(1): 67–74. Kaufman KR, Gerner R. Lamotrigine toxicity secondary to sertraline. Seizure 1998; 7(2): 163–165. Kendall AG, Boivin M. Warfarin-carbamazepine interaction. Ann Intern Med 1981; 94(2): 280. Kidd RS, Straughn AB, Mayer MC, et al. Pharmacokinetics of chlorpheniramine, phenytoin,

glipizide and nifedipine in an individual homozygous for the CYP2C9*3 allele.

Pharmacogenetics 1999; 9(1): 71–80.
Kirchheiner J, Bauer S, Meineke I, et al. Impact of CYP2C9 and CYP2C19 polymorphisms on

tolbutamide kinetics and the insulin and glucose response in healthy volunteers. Pharma-

cogenetics 2002; 12(2): 101–109.
Klaassen SL. Ticlopidine-induced phenytoin toxicity. Ann Pharmacother 1998; 32(12): 1295–1298. Kubacka RT, Ferrante JA. Carbamazepine–propoxyphene interaction. Clin Pharm 1983; 2(2): 104. Labroo RB, Paine MF, Thummel KE, et al. Fentanyl metabolism by human hepatic and intestinal

cytochrome P450 3A4: implications for interindividual variability in disposition, efficacy, and

drug interactions. Drug Metab Dispos 1997; 25(9): 1072–1080.
Lackner TE, Cloyd JC, Thomas LW, et al. Antiepileptic drug use in nursing home residents: effect

of age, gender, and comedication on patterns of use. Epilepsia 1998; 39(10): 1083–1087. Lader M. Some adverse effects of antipsychotics: prevention and treatment. J Clin Psychiatry

1999; 60(Suppl. 12): S18–S21.
Landay RA, Gonzalez MA, Taylor JC. Effect of phenobarbital on theophylline disposition.

J Allergy Clin Immunol 1978; 62(1): 27–29.
Laroudie C, Salazar DE, Cosson JP, et al. Carbamazepine–nefazodone interaction in healthy

subjects. J Clin Psychopharmacol 2000; 20(1): 46–53.
Leinonen E, Lillsunde P, Laukkanen V, et al. Effects of carbamazepine on serum antidepressant

concentrations in psychiatric patients. J Clin Psychopharmacol 1991; 11(5): 313–318. Leonard RF, Knott PJ, Rankin GO, et al. Phenytoin–salicylate interaction. Clin Pharmacol Ther

1981; 29(1): 56–60.
Levine M, Sheppard I. Biphasic interaction of phenytoin with warfarin. Clin Pharm 1984; 3(2):

200–203.
Levy RH. Cytochrome P450 isozymes and antiepileptic drug interactions. Epilepsia 1995;

36(Suppl. 5): S8–S13.
Linnet K, Olesen OV. Free and glucuronidated olanzapine serum concentrations in psychiatric

patients: influence of carbamazepine comedication. Ther Drug Monit 2002; 24(4): 512–517.

291 Antiepileptic drug interactions in the elderly

Linnoila M, Viukari M, Vaisanen K, et al. Effect of anticonvulsants on plasma haloperidol and thioridazine levels. Am J Psychiatry 1980; 137(7): 819–821.

Lucas RA, Gilfillan DJ, Bergstrom RF. A pharmacokinetic interaction between carbamazepine and olanzapine: observations on possible mechanism. Eur J Clin Pharmacol 1998; 54(8): 639–643.

Luo G, Cunningham M, Kim S, et al. CYP3A4 induction by drugs: correlation between a pregnane X receptor reporter gene assay and CYP3A4 expression in human hepatocytes. Drug Metab Dispos 2002; 30(7): 795–804.

Ma B, Prueksaritanont T, Lin JH. Drug interactions with calcium channel blockers: possible involvement of metabolite-intermediate complexation with CYP3A. Drug Metab Dispos 2000; 28(2): 125–130.

MacKichan J. Influence of protein binding and use of unbound (free) drug concentrations. In Applied Pharmacokinetics. W. Evans, J. Schentag, W. Jusko, eds. Vancouver, WA: Applied Therapeutics, Inc., 1992.

Macphee GJ, McInnes GT, Thompson GG, et al. Verapamil potentiates carbamazepine neuro- toxicity: a clinically important inhibitory interaction. Lancet 1986; 1(8483): 700–703.

Martinez C, Albet C, Agundez JA, et al. Comparative in vitro and in vivo inhibition of cytochrome P450 CYP1A2, CYP2D6, and CYP3A by H2-receptor antagonists. Clin Pharmacol Ther 1999; 65(4): 369–376.

Massey EW. Effect of carbamazepine on Coumadin metabolism. Ann Neurol 1983; 13(6): 691–692. Mattson RH, Cramer JA, Dorney PD, et al. Use of oral contraceptives by women with epilepsy.

J Am Med Assoc 1986; 256(2): 238–240.
Mayersohn M. Special Pharmacokinetic Considerations in the Elderly. In Applied Pharmacokinetics.

W. Evans, J. Schentag and W. Jusko, eds. Vancouver, WA: Applied Therapeutics, Inc., 1992. Mazzu AL, Lasseter KC, Shamblen EC, et al. Itraconazole alters the pharmacokinetics of atorvas- tatin to a greater extent than either cerivastatin or pravastatin. Clin Pharmacol Ther 2000;

68(4): 391–400.
McElnay JC, Uprichard G, Collier PS. The effect of activated dimethicone and a proprietary

antacid preparation containing this agent on the absorption of phenytoin. Br J Clin Pharmacol

1982; 13(4): 501–505.
Mendez JS, Cotzias GC, Mena I, et al. Diphenylhydantoin. Blocking of levodopa effects. Arch

Neurol 1975; 32(1): 44–46.
Miceli JJ, Anziano RJ, Roberge L, et al. The effect of carbamazepine on the steady-state pharmaco-

kinetics of ziprasidone in healthy volunteers. Br J Clin Pharmacol 2000; 49(Suppl. 1): S65–S70. Michelucci R, Tassinari C. Benzodiazepines-adverse effects. In Antiepileptic Drugs. R. Levy, R. Mattson, B. Meldrun, E. Percucca, eds. Philadelphia, PA: Lippincott Williams & Wilkins, 2002. Miller M, Cosgriff J, Kwong T, et al. Influence of phenytoin on theophylline clearance. Clin

Pharmacol Ther 1984; 35(5): 666–669.
Miners JO, Attwood J, Birkett DJ. Determinants of acetaminophen metabolism: effect of induc-

ers and inhibitors of drug metabolism on acetaminophen’s metabolic pathways. Clin Pharmacol

Ther 1984; 35(4): 480–486.
Mula M, Monaco F. Carbamazepine–risperidone interactions in patients with epilepsy. Clin

Neuropharmacol 2002; 25(2): 97–100.

292 Jeannine M. Conway and James C. Cloyd

Mungall DR, Ludden TM, Marshall J, et al. Population pharmacokinetics of racemic warfarin in adult patients. J Pharmacokinet Biopharm 1985; 13(3): 213–227.

Nappi JM. Warfarin and phenytoin interaction. Ann Intern Med 1979; 90(5): 852.
Nelson MH, Birnbaum AK, Remmel RP. Inhibition of phenytoin hydroxylation in human liver microsomes by several selective serotonin re-uptake inhibitors. Epilepsy Res 2001; 44(1):

71–82.
Neurontin (2002). Package Insert. Pfizer, Inc.
Nitz N, Garrard J, Harms S, et al. Prevalence of antiepileptic drug use among medicare benefici-

aries. Epilepsia 2000; 41(Suppl. 7): S251.
Nolan L, O’Malley K. Adverse drug reactions in the elderly. Br J Hosp Med 1989; 41(5): 446, 448,

452–457.
Oles KS, Mirza W, Penry JK. Catastrophic neurologic signs due to drug interaction: Tegretol and

Darvon. Surg Neurol 1989; 32(2): 144–151.
Olesen OV, Linnet K. Olanzapine serum concentrations in psychiatric patients given standard

doses: the influence of comedication. Ther Drug Monit 1999; 21(1): 87–90.
Ono S, Mihara K, Suzuki A, et al. Significant pharmacokinetic interaction between risperidone and carbamazepine: its relationship with CYP2D6 genotypes. Psychopharmacology (Berl)

2002; 162(1): 50–54.
Panegyres PK, Rischbieth RH. Fatal phenytoin warfarin interaction. Postgrad Med J 1991;

67(783): 98.
Paoletti R, Corsini A, Bellosta S. Pharmacological interactions of statins. Atheroscler Suppl 2002;

3(1): 35–40.
Pihlsgard M, Eliasson E. Significant reduction of sertraline plasma levels by carbamazepine and

phenytoin. Eur J Clin Pharmacol 2002; 57(12): 915–916.
Prichard PJ, Walt RP, Kitchingman GK, et al. Oral phenytoin pharmacokinetics during omepra-

zole therapy. BrJ Clin Pharmacol 1987; 24(4): 543–545.
Ramsay RE, Slater JD. Effects of antiepileptic drugs on hormones. Epilepsia 1991; 32(Suppl. 6):

S60–S67.
Ramsay R, Rowan A, Slater J, et al. Effect of age on epilepsy and its treatment: results of the VA

cooperative study. Epilepsia 1994; 35(Suppl. 8): 91A.
Raucy JL, Mueller L, Duan K, et al. Expression and induction of CYP2C P450 enzymes in pri-

mary cultures of human hepatocytes. J Pharmacol Exp Ther 2002; 302(2): 475–482.
Reminyl (2003). Package Insert. Janssen Pharmaceutical Products. L.P., Titusville, NJ.
Rosenfeld WE, Doose DR, Walker JA, et al. Effect of topiramate on the pharmacokinetics of an

oral contraceptive containing norethindrone and ethinyl estradiol in patients with epilepsy.

Epilepsia 1997; 38(3): 317–323.
Ross JR, Beeley L. Interaction between carbamazepine and warfarin. Br Med J 1980; 280(6229):

1415–1416.
Rotzinger S, Baker GB. Human CYP3A4 and the metabolism of nefazodone and hydroxyne-

fazodone by human liver microsomes and heterologously expressed enzymes. Eur Neuro-

psychopharmacol 2002; 12(2): 91–100.
Saano V, Glue P, Banfield CR, et al. Effects of felbamate on the pharmacokinetics of a low-dose

combination oral contraceptive. Clin Pharmacol Ther 1995; 58(5): 523–531.

293 Antiepileptic drug interactions in the elderly

Schachter SC, Cramer GW, Thompson GD, et al. An evaluation of antiepileptic drug therapy in nursing facilities. J Am Geriatr Soc 1998; 46(9): 1137–1141.

Schmider J, Greenblatt DJ, Von Moltke LL, et al. Inhibition of CYP2C9 by selective serotonin reuptake inhibitors in vitro: studies of phenytoin p-hydroxylation. Br J Clin Pharmacol 1997; 44(5): 495–498.

Shaughnessy AF, Mosley MR. Elevated carbamazepine levels associated with diltiazem use. Neurology 1992; 42(4): 937–938.

Shin JG, Park JY, Kim MJ, et al. Inhibitory effects of tricyclic antidepressants (TCAs) on human cytochrome P450 enzymes in vitro: mechanism of drug interaction between TCAs and phenytoin. Drug Metab Dispos 2002; 30(10): 1102–1107.

Sitsen J, Maris F, Timmer C. Drug–drug interaction studies with mirtazapine and carbamazepine in healthy male subjects. Eur J Drug Metab Pharmacokinet 2001; 26(1–2): 109–121.

Sklar SJ, Wagner JC. Enhanced theophylline clearance secondary to phenytoin therapy. Drug Intell Clin Pharm 1985; 19(1): 34–36.

Spina E, Avenoso A, Pollicino AM, et al. Carbamazepine coadministration with fluoxetine or fluvoxamine. Ther Drug Monit 1993; 15(3): 247–250.

Stewart RB. Drug use in the elderly. In Therapeutics in the Elderly. J. Delafuente, R. Stewart, eds. Cincinnati: Harvey Whitney, 2001: 235–256.

Tang C, Shou M, Mei Q, et al. Major role of human liver microsomal cytochrome P450 2C9 (CYP2C9) in the oxidative metabolism of celecoxib, a novel cyclooxygenase-II inhibitor. J Pharmacol Exp Ther 2000; 293(2): 453–459.

Trileptal (2001). Package Insert. Novartis.
Ucar M, Neuronen M, Luurila H, et al. Carbamazepine markedly reduces serum concentrations

of simvastatin and simvastatin acid. Eur J Pharmacol 2004; 59: 879–882.
Udall JA. Clinical implications of warfarin interactions with five sedatives. Am J Cardiol 1975;

35(1): 67–71.
Ultram (2000). Package Insert. Ortho-McNeil Pharmaceutical, Inc.
Wallace SM, Verbeeck RK. Plasma protein binding of drugs in the elderly. Clin Pharmacokinet

1987; 12(1): 41–72.
Wang RW, Kari PH, Lu AY, et al. Biotransformation of lovastatin. IV. Identification of cytochrome

P450 3A proteins as the major enzymes responsible for the oxidative metabolism of lovastatin

in rat and human liver microsomes. Arch Biochem Biophys 1991; 290(2): 355–361.
Webster LK, Mihaly GW, Jones DB, et al. Effect of cimetidine and ranitidine on carbamazepine

and sodium valproate pharmacokinetics. Eur J Clin Pharmacol 1984; 27(3): 341–343. Wellbutrin (1999). Package Insert. Glaxo Smith-Kline.
Wen X, Wary JS, Bockman JT, et al. Genfibrozil is a potent inhibitor of human cytochrome

P4502C9. Drug Metab Dispos 2001; 29: 1354–1361.
Wong YW, Yeh C, Thyrum PT. The effects of concomitant phenytoin administration on the

steady-state pharmacokinetics of quetiapine. J Clin Psychopharmacol 2001; 21(1): 89–93. Woods DJ, Coulter DM, Pillans P. Interaction of phenytoin and fluoxetine. NZ Med J 1994;

107(970): 19.
Yu YL, Huang CY, Chin D, et al. Interaction between carbamazepine and dextropropoxyphene.

Postgrad Med J 1986; 62(725): 231–233.

Antiepileptic drug interactions in pregnancy

Mark S. Yerby
North Pacific Epilepsy Research, Oregon Health Sciences University, Portland, Oregon, USA

Scope of the problem

Women with epilepsy require chronic antiepilepsy drugs (AEDs) to prevent seizures, maintain their function and health. Unlike most young women they are unable to discontinue their medications if they become pregnant, for to do so increases their risk of seizures, personal injury, miscarriage and developmental delay in the offspring. With a prevalence of between 0.6% and 1.0% and an esti- mated 40% of those with epilepsy being women of childbearing years one can see that the potential public health impact is significant. Most women with epilepsy have healthy children but there is an increased risk for congenital malformations, fetal loss, developmental delay and neonatal hemorrhage. Maternal epilepsy is a contributor but the use of AEDs is a significant confounder. To make matters more complicated 86% of pregnant women take medications during pregnancy. A sur- vey by the World Health Organization of 14 778 women in 22 countries reported that of the 86% of women taking medications during pregnancy the average number of prescriptions was 2.9 (range of 1–15). This study did not evaluate over-the-counter medications. The preponderance of prescriptions, 73%, were written by obstetricians (Collaborative Drug Use in Pregnancy, 1991).

When evaluating AED use in pregnancy one is hampered by the lack of knowl- edge of specific co-medications, even though it is clear that this is a common event. While monotherapy with AEDs is a goal of epilepsy management, it is not always an obtainable one. Polytherapy is also more common with the “newer” post-1993 introduction AEDs, because all were initially approved for use as adjunctive therapy. The pregnancy outcome of greatest interest is congenital malformations. While there are substantial data on this outcome, most are in the form of case series and case reports, and accurate rates and risks cannot be determined. Other adverse outcomes are at least as common in terms of incidence (developmental delay, fetal loss), but have received significantly less attention.

Let us review some of the clinically important issues surrounding pregnancy and AED exposure.

294

295 Antiepileptic drug interactions in pregnancy

Antiepileptic drugs and hormonal contraceptives

A discussion of pregnancy needs to be preceded by reviewing the problems of contra- ception. Oral contraceptives have not been associated with exacerbation of epilepsy (Mattson et al., 1986). The effectiveness of hormonal contraceptives can, however, be reduced by enzyme-inducing AED (carbamazepine, phenytoin, phenobarbital, felbamate, topiramate). Hormonal contraceptives come in three formulations:

• oral (estrogen–progesterone combinations or progesterone only);
• subcutaneous (levonorgestrel) or intrauterine (progestasert) implants; • injectable (depoprovera).

All three forms can be adversely impacted by enzyme-inducing AED.
AEDs may lower concentrations of estrogens by 40–50%. They also increase sex hormone-binding globulin (SHBG), which increases the binding of progesterone and reduces the unbound fraction. The result is that hormonal contraception is

less reliable with enzyme-inducing AEDs.
The low- or mini-dose oral contraceptives are therefore to be used with caution.

As it is the progesterone not the estrogen that inhibits ovulation, using higher-dose estrogens alone may not be effective. The more rapid clearance of the oral contra- ceptive when used in conjunction with an enzyme-inducing AED will reduce the likelihood of unwanted side effects from higher-dose tablets.

Failure of implantable hormonal contraceptives has also occurred (Shane- McWhorter et al., 1998). Mid-cycle spotting or bleeding is a sign that ovulation is not suppressed. If this occurs alternative or supplementary methods of contracep- tion are required. Contraceptive failure may not always be predictable, even when mid-cycle spotting does not occur. Failure of basal body temperature to rise at mid-cycle can be used to document ovulatory suppression.

Medroxyprogesterone injections should be given every 10 instead of 12 weeks to women on enzyme-inducing AED. This shorter cycle is less likely to result in unin- tended pregnancy (Crawford, 2002a).

For multiparous women with epilepsy, intrauterine devices may be an excellent contraceptive choice. Alternatively non-enzyme-inducing AEDs may need to be considered (valproate, lamotrigine, gabapentin or zonisamide). A recent report suggests that topiramate at doses of 200 mg a day lacks enough enzyme induc- tion to effect hormonal contraceptives. Higher doses however do reduce ethinyl estradiol concentrations by 18% on 200mg, 21% with 400mg and 30% with 800 mg of topiramate a day (Doose et al., 2002).

The importance of the potential impact of enzyme-inducing AEDs cannot be underestimated. In a survey of 294 general practices in the General Practice Research Database, 16.7% of women aged 15–45 with epilepsy were taking an oral

296 Mark S. Yerby

contraceptive. Two hundred were on an enzyme-inducing AED and 56% on low estrogen ( 50 g) hormonal contraceptives (Shorvon et al., 2002).

There has been at least one circumstance in which oral contraceptives effect AED concentration. Sabers and colleagues (2003) have demonstrated a marked reduc- tion in lamotrigine concentrations when oral contraceptives are also taken. The average plasma concentration in 22 women on lamotrigine monotherapy and an oral contraceptive was 13 mol/l. In a similar group of women on lamotrigine monotherapy with no oral contraceptive use, the plasma concentrations averaged 28 mol/l: a significant reduction in AED concentration of over 50%. It has been suggested that oral contraceptives may induce the metabolism of glucuronidated drugs such as lamotrigine.

Maternal complications associated with AED

Seizures not infrequently worsen during pregnancy. One-quarter to one-third of woman with epilepsy (WWE) will have an increase in seizure frequency during pregnancy. This increase is unrelated to seizure type, duration of epilepsy, or seizure frequency in a previous pregnancy. In a recent series of 215 pregnancies in WWE an increase in seizures during the first trimester occurred in 30% of monotherapy- and 43% of polytherapy-treated women. One in 8 or 12.5% had to be hospitalized for their seizures during the pregnancy (Cahill et al., 2002).

Plasma concentrations of anticonvulsant drugs decline as pregnancy progresses, even in the face of constant and in some instances increasing doses (Tomson et al., 1994; Rodriguez-Palomares et al., 1995; Tomson et al., 1997). Although reduction of plasma drug concentration is not always accompanied by an increase in seizure frequency, virtually all women with increased seizures in pregnancy have sub- therapeutic drug levels (Dansky et al., 1982; Janz, 1982; Schmidt, 1982; Schmidt et al., 1983; Otani, 1985). The decline of anticonvulsant levels during pregnancy is largely a consequence of decreased plasma protein binding (Perruca, 1982; Yerby et al., 1985; Tomson et al., 1994), reduced concentration of albumin and increased drug clearance (Dam et al., 1979; Nau et al., 1981; Janz, 1982; Philbert and Dam, 1982). The clearance rates are greatest during the third trimester.

Kaarkuzhali and colleagues (2002) found that a majority of their pregnant patients on carbamazepine, phenytoin or phenobarbital required numerous dose adjustments during pregnancy to maintain therapeutic levels. Fifty percent of the pregnancies had breakthrough seizures when the levels fell below the therapeutic range. It is therefore imperative to monitor AED levels at least monthly and adjust dosage to maintain therapeutic levels (Levy and Yerby, 1985). Table 16.1 summa- rizes some of the pharmacokinetics of anticonvulsant drugs during pregnancy.

Less is known about the kinetics of the newer AEDs in pregnancy. A report demonstrates that lamotrigine clearance increases by 50% during pregnancy

297 Antiepileptic drug interactions in pregnancy

Table 16.1 Pharmacokinetic data for first generation AEDs

Percent decrease

Percent free fraction
Anticonvulsant third trimester Normal Maternal Neonatal

Total level by

Carbamazepine 40 Ethosuximide ? Phenobarbital 55 Phenytoin 56 Primidone 55 Derived phenobarbital 70 VPA 50

22 25 35 90 ? ? 51 58 66

9 11 13

? ? ? 75 80 ? 9 15 19

and that the clearance changes occur relatively early in pregnancy. Eleven of 12 pregnancies required increased doses of lamotrigine to maintain therapeutic levels during pregnancy (Tran et al., 2002).

Polycystic ovaries

A great deal of confusing literature has been written about the effect of AEDs on the development of polycystic ovaries (PCO). Ovarian cysts are found in approxi- mately 6.6% of women of childbearing age. Most of these cysts (over 80%) will disappear within 3 months (Borgfeldt and Andolf, 1999). Multiple or PCO are more commonly found in women taking hormonal contraceptives with progesterone, and women who are infertile. The rates vary but average between 10% and 20% (Borgfeldt and Andolf, 1999).

The polycystic ovarian syndrome (PCOS) is a specific disturbance of neuro- endocrine function defined as no or irregular menses (oligomenorrhea), elevated levels of male sex steroid hormones (hyperandrogenism) without evidence of other disturbances such as hyperprolactinemia, thyroid dysfunction or 21-hydroxylase deficiency. It is uncommon as occurs in approximately 6.5% of women of repro- ductive age (Asuncion et al., 2000). It is associated with sustained release of gonadotropic-releasing hormone (GnRH) and lutenizing hormone (LH), and affected women are often overweight, have elevated serum lipids and are less sen- sitive to insulin. PCOS is also seen in higher than expected rates in mothers (24–52%) and sisters (32–66%) of women with this disorder leading some to believe that it is a genetic disorder (Govind et al., 1999; Kahsar-Miller et al., 2001).

Isojarvi and colleagues (1993) demonstrated an excess of menstrual abnormalities in WWE taking valproic acid (VPA) compared to other AEDs. They also stated that

298 Mark S. Yerby

80% of women taking valproate before the age of 20 developed PCO. Despite this observation other researchers have not found a consistent association between specific AED or epilepsy types and PCOS (Chappell et al., 1999; Bilo et al., 2001; Genton et al., 2001). Women with bipolar disorder are often treated with valproate and do not have an increase in PCOS (Rasgon et al., 2000).

Fetal complications associated with AED

A number of adverse outcomes of pregnancy are known to occur more often in infants of mothers with epilepsy (IME). Of the three major variables – maternal epilepsy, maternal seizures during gestation, and AEDs – it is not always possible to determine which is the most significant. For the outcome congenital malforma- tions, AEDs appear to be a significant risk factor. A recent epidemiological study in Iceland suggests that untreated women with epilepsy have approximately the same rate of malformations in their offspring as do treated mothers, 4.8 vs. 5.9%, respec- tively. This suggests that a portion of the increased risk is secondary to maternal epilepsy itself (Olafsson et al., 1998). On balance however, malformation rates are twice that seen in the general population, and the proportion of women with epilepsy who are untreated is so small that it is clinically insignificant.

Congenital malformations are defined as a physical defect requiring medical or surgical intervention, and resulting in a major functional disturbance. Congenital anomalies in contrast are defined as deviations from normal morphology that do not require intervention. It is uncertain whether these aberrations represent distinct entities or a spectrum of physiological responses to insult to the devel- oping fetus: malformations at one extreme and anomalies at the other. For the purposes of this review, congenital malformations and anomalies will be discussed separately.

IME, exposed to anticonvulsant drugs in utero, are twice as likely to develop malformations as infants not exposed to these drugs. Malformation rates in the general population range from 2% to 3%. Reports of malformation rates in vari- ous populations of exposed infants range from 1.25% to 11.5% (Fedrick, 1973; Nakane et al., 1980; Philbert and Dam, 1982; Kelly, 1984a; Steegers-Theunissen et al., 1994; Jick and Terris, 1997; Olafsson et al., 1998; Kaneko et al., 1999; Vajda et al., 2002). These combined estimates yield a risk of malformations in a pregnancy of a WWE of 4–6%. Cleft lip, cleft palate, or both, and congenital heart disease account for many of the reported cases. Orofacial clefts are responsible for 30% of the increased risk of malformations in these infants (Kelly, 1984a; Friis et al., 1986; Abrishamchian et al., 1994).

The increased rate of malformations in the offspring of mothers with epilepsy appears to be related to AED exposure in utero. Evidence to support this association comes from four observations.

299 Antiepileptic drug interactions in pregnancy

. 1 Comparisons of the malformation rates in the offspring of mothers with epilepsy treated with AEDs as opposed to those with no AED treatment reveal consistently higher rates in the children of the treated women (South, 1972; Speidel and Meadow, 1972; Lowe, 1973; Monson et al., 1973; Annegers et al., 1978; Nakane et al., 1980).

. 2 Mean plasma AED concentrations are higher in mothers with malformed infants than mothers with healthy children (Dansky et al., 1980).

. 3 Infants of mothers taking multiple AEDs have higher malformation rates than those exposed to monotherapy (Nakane, 1979; Lindhout et al., 1984).

. 4 Maternal seizures during pregnancy do not appear to increase the risk of con- genital malformations (Fedrick, 1973).

Majewski and co-workers (1980) described increased malformation rates and cen- tral nervous system injury in IMEs exposed to maternal seizures. More recently, Lindhout and co-workers (1992) described a marked increase in malformations amongst infants exposed to first trimester seizures (12.3%) compared to fetuses that were not subject to any maternal seizures (4.0%). Malformations were more often observed in infants exposed to partial seizures than to generalized tonic– clonic seizures. Nonetheless, most investigators have found that maternal seizures during pregnancy had no impact on the frequency of malformations, development of epilepsy or febrile convulsions (Annegers et al., 1978; Nakane et al., 1980).

A variety of congenital malformations have been reported in children of mothers with epilepsy, and every anticonvulsant medication has been implicated in their development. Cleft lip and/or palate, and congenital heart disease account for a majority of reported cases (Elshove and Van Eck, 1971; Anderson, 1976; Annegers et al., 1978). Orofacial clefts are relatively common malformations in the general population, occurring with a frequency of 1.5 per 1000 live births. IME have a rate of orofacial clefting of 13.8 per 1000, a nine-fold increase in risk (Kelly, 1984a; Kallen, 1986a). Early observations that persons with clefting of the lip or palate were twice as likely to have family members with epilepsy as controls suggested that orofacial clefts were associated with epilepsy (Friis et al., 1981). Subsequent studies of the prevalence of facial clefts in the siblings and children of 2072 persons with epilepsy found observed or expected ratios increased only for maternal epilepsy. The risk was greater if AEDs were taken during pregnancy (4.7) than if no AED treatment was used (2.7). The authors concluded that there was no evidence that epilepsy itself contributed to the development of orofacial clefts (Friis et al., 1986). Israeli researchers have found that children with cleft lip or palate are four times as likely to have a mother with epilepsy as the general population, and mothers with epilepsy are six times as likely to bear a child with an orofacial cleft as non-epileptic women (Gatoh et al., 1987). Orofacial clefts account for 30% of the excess of congenital malformations in IMEs.

300 Mark S. Yerby

Congenital heart defects are the second most frequently reported teratogenic abnormality associated with AEDs. IME have a 1.5–2% prevalence of congenital heart disease, a relative risk (RR) of three-fold over the general population (Kallen, 1986b). Anderson (1976) prospectively studied maternal epilepsy and AED use in 3000 children with heart defects at the University of Minnesota. Eighteen IMEs were identified. Twelve of these had ventricular septal defects; 9 of the 18 children had additional non-cardiac defects, 8 of which were orofacial clefts.

No AED can be considered absolutely safe in pregnancy, but for the vast majority of drugs no specific pattern of major malformations has been identified (Kallen, 1986b). This lack of a particular or characteristic pattern of defects has been cited as evidence that AEDs are not teratogenic. When phenobarbital is given during pregnancy for conditions other than epilepsy, no increase in malformation rates has been demonstrated (Shapiro et al., 1976). Phenobarbital has been demonstrated to be relatively teratogenic in mono- and polytherapy. Five of 79 phenobarbital monotherapy-exposed pregnancies were associated with major malformations (proportion 6.3%; 95% confidence interval (CI): 2.1–14.2%). When compared to the background rate (1.62%), there was a significantly increased risk for major malformations, with a RR of 3.8 (95% CI: 1.7–9.0%). A two-fold increase in risk was found when phenobarbital was compared to three other frequently used AED monotherapies (RR 2.2; 95% CI: 0.9–5.2%) (Holmes et al., in press).

Mechanisms of teratogenicity

Epoxides

A hypothesis that metabolites of AEDs are responsible for malformations has been developed on the basis of the following observations:

. 1 an arene oxide metabolite of phenytoin or other AED is the ultimate teratogen;

. 2 a genetic defect in epoxide hydrolase (arene oxide detoxifying enzyme) system
increases the risk of fetal toxicity;

. 3 free radicals produced by AED metabolism are cytotoxic;

. 4 ageneticdefectinfreeradicalscavengingenzymeactivity(FRSEA)increasesthe
risk of fetal toxicity.

A large number of drugs can be converted into epoxides, in reactions that are cat- alyzed by the microsomal monoxygenase system (Jerina and Daly, 1974; Sims and Grover, 1974). Arene oxides are unstable epoxides formed by aromatic compounds. Various epoxides are electrophilic and may elicit carcinogenic, mutagenic and other toxic effects by covalent binding to cell macromolecules (Nebert and Jensen, 1979; Shum et al., 1979). Epoxides are detoxified by two processes:

1 conversion to dihydrodiols catalyzed by epoxide hydrolase in the cytoplasm, 2 conjugation with glutathione (GSH) in the microsomes.

301 Antiepileptic drug interactions in pregnancy

Epoxide hydrolase activity has been found in the cytosol and the microsomal sub- cellular fraction of adult and fetal human hepatocytes. Epoxide hydrolase activity in fetal liver is lower than that of adults (Pacifici et al., 1983). One-third to one-half of fetal circulation bypasses the liver, resulting in higher direct exposure of extra- hepatic fetal organs to potential toxic metabolites (Pacifici and Rane, 1982).

Phenytoin teratogenicity

Formation of arene oxides by phenytoin

Arene oxides are obligatory intermediates in the metabolism of aromatic compounds to transdihydrodiols. Phenytoin forms a transdihydrodiol metabolite (Chang et al., 1970). This metabolite is also formed by neonates exposed to phenytoin in utero (Horning et al., 1974). In vitro studies have shown that an oxidative (NADPH/02 dependent) metabolite of phenytoin binds irreversibly to rat liver microsomes (Martz et al., 1977). This binding is increased by an inhibitor of epoxide hydrolase (trichloroponene oxide, TCPO) and decreased by GSH (Martz et al., 1977; Pantarotto et al., 1982; Wells and Harbison, 1985). Using human lymphocytes to assess cell defense mechanisms against toxicity, Spielberg et al. (1981) showed that cytotoxicity was enhanced by inhibitors of epoxide hydrolase.

Phenytoin birth defects and lymphocyte cytotoxicity

Strickler et al. (1985) examined lymphocytes of 24 children exposed to phenytoin during gestation and lymphocytes from their families using the Spielberg test of cytotoxicity (Spielberg et al., 1981). Lymphocytes were incubated with phenytoin in a mouse microsomal system. A positive response was defined as increase in cell death over baseline. Cells from 15 children gave a positive response. Each positive child had a positive parent (as many mothers as fathers), and a positive response was highly correlated with major birth defects. The authors concluded that a genetic defect in arene oxide detoxification increased the risk of the child having major birth defects (Strickler et al., 1985).

Phenytoin birth defects and epoxide hydrolase activity

In 1985, Buchler reported epoxide hydrolase activity in skin fibroblasts of a pair of dizygotic twins exposed to phenytoin in utero. The infant who had more features of the fetal hydantoin syndrome (FHS) showed lower epoxide hydrolase activity. Although this finding supports the epoxide hydrolase hypothesis, it should be noted that a full report of the experimental details has not yet appeared.

The evidence that epoxide metabolites of phenytoin are teratogenic can be sum- marized as follows. Phenytoin has an epoxide metabolite that binds to tissues.

302 Mark S. Yerby

Inhibition of the detoxifying enzyme epoxide hydrolase increases the rate of orofacial clefts in experimental animals, lymphocyte cytotoxicity, and the binding of epoxide metabolite to liver microsomes.

These facts cannot completely explain the teratogenicity seen in phenytoin or other AEDs. The lymphocyte cytotoxicity seen with epoxide metabolites correlates with major but not minor malformations (Dansky et al., 1987). Dysmorphic abnormalities have been described in siblings exposed to ethotoin in utero. Ethotoin is not metabolized through an arene oxide intermediate (Finnell and DiLiberti, 1983). Embryopathies have been described with exposure to mephenytoin, which also does not form an arene oxide intermediate (Wells et al., 1982). Trimethadione is clearly teratogenic but has no phenyl rings and thus cannot form an arene oxide metabolite. Therefore, an alternate mechanism must exist.

Free radical intermediates of AEDs and teratogenicity

Some drugs are metabolized or bioactivated by co-oxidation during prostaglandin synthetase (PGS)-catalyzed synthesis of prostaglandins. Such drugs serve as electron donors to peroxidases, resulting in an electron-deficient drug molecule, which by definition, is called a free radical. In the search for additional electrons to complete their outer ring, free radicals can covalently bind to cell macromolecules, including nucleic acids (DNA, RNA), proteins, cell membranes and lipoproteins to produce cytotoxicity.

Phenytoin is co-oxidated by PGS, thyroid peroxidase and horseradish peroxidase producing reactive free radical intermediates that bind to proteins (Kubow and Wells, 1989). Phenytoin teratogenicity can be modulated by substances that reduce the formation of phenytoin-free radicals. Acetylsalicylic acid (ASA) irreversibly inhibits PGS, caffeic acid is an antioxidant, alpha-phenyl-N–t-butylnitrone (PBN) is a free radical spin-trapping agent. Pretreatment of pregnant mice with these compounds reduces the number of cleft lip or pathies secondary to phenytoin in their offspring (Wells et al., 1989).

GSH is believed to detoxify free radical intermediates by forming a non-reactive conjugate. N-acetylcysteine (NACl) a GSH precursor, decreases phenytoin- induced orofacial clefts and fetal weight loss in rodents (Wong and Wells, 1988). 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) inhibits GSH reductase, an enzyme necessary to maintain adequate cellular GSH concentrations, and increases pheny- toin embryopathy at doses at which BCNU alone has no embryopathic effect (Wong and Wells, 1989). The metabolism of phenytoin or other AEDs to free rad- ical intermediates may be responsible for the teratogenicity seen in IMEs. Twenty- six children with myelomeningocele and their parents were studied by Graf and colleagues (1995). They were found to have significantly lower antioxidant enzymes, particularly GSH peroxidase, than controls.

303 Antiepileptic drug interactions in pregnancy

Neural tube defects and AEDs

Antiepileptic drugs as a group do not produce any specific pattern of major mal- formations. A possible exception to this is the association of sodium valproate and carbamazepine with neural tube defects (NTDs). Robert and Guibaud (1982) were the first to make this association while working in a birth defects registry in the Rhone Alps region of France. They reported NTDs in IME exposed to VPA. Other studies have revealed an association between carbamazepine exposure in utero and NTDs (Rosa, 1991; Little et al., 1993). Subsequent evaluations of these exposures identify spina bifida aperta (SB) as the specific NTD associated with the VPA or carbamazepine exposure (Lindhout et al., 1992). Methodologic problems make frequency estimates imprecise since most published data are case reports, case series or very small cohorts from registries that were not designed to evaluate preg- nancy outcomes. The prevalence of SB with valproate exposure is approximately 1–2% (Lindhout and Schmidt, 1986) and with carbamazepine 0.5% (Rosa, 1991; Hiilesmaa, 1992). A recent prospective study in the Netherlands, however, found IME exposed to valproate had a 5.4% prevalence rate of SB. Average daily valproate doses were higher in the IME with SB (1640 136 mg/day) than in the unaffected IME (941 48 mg/day). Another group of investigators has found that valproate doses of 1000 mg/day or plasma concentrations of 70 g/ml are unlikely to cause malformations (Kaneko et al., 1999). Both groups recommend that valproate dose be reduced whenever valproate must be used in pregnancy (Omtzigt et al., 1992; Kaneko et al., 1999). It has also been suggested that multiple daily doses or the use of extended release formulations may reduce the peak plasma concentrations and thus reduce the risk of malformations.

NTDs are uncommon malformations occurring in 6/10 000 pregnancies. Spina bifida and anencephaly are the most commonly reported NTD and affect approxi- mately 4000 pregnancies annually resulting in 2500–3000 births in the US each year (Mullinare and Erickson, 1997; Honein et al., 2001). The types of NTD asso- ciated with AED exposure are primarily myelomeningocele and anencephaly, which are the result of abnormal neural tube closure between the third and fourth weeks of gestational age.

Previous thinking about NTD visualized the fusion of the neural tube as one in which the lateral edges met in the middle and fused both rostrally and caudally similar to a bi-directional zipper. Recent studies have suggested there are multiple sites for neural tube closure (Van Allen et al., 1993; Golden and Chenroff, 1995) and that different etiologies may result in different types of abnormality.

There are differences in specific sites and timing of each individual closure region. The majority of human NTD can be explained by failure of one or more clo- sure sites. Anencephaly with frontal and parietal defects is due to failure at closure site two. Holocrania which also involves defects of the posterior cranium to the

304 Mark S. Yerby

foramen magnum is due to failure of closure of areas two and four. Lumbar spina bifida results from failure of closure one. The development of closure sites appears to be under genetic control and also affected by environmental factors. In twins, concor- dance rates are only 56% for anencephaly and 71% for spina bifida. In Great Britain there is a male preponderance of lumbar spina bifida and female preponderance of holocrania and anencephaly. Even VPA appears to have species differential effects being associated with spina bifida in humans and exencephaly in mice (Seller, 1995).

A number of risk factors are associated with NTDs. A previous pregnancy with NTD is the strongest association, with a RR of 10. There are strong ethnic or geo- graphic associations with NTDs. Rates per 1000 are 0.22 for Whites, 0.58 for persons of Hispanic descent and 0.08 for persons of African descent. The incidence of NTDs in Mexico is 3.26/1000, for Mexican-born persons living in California 1.6/1000 and for US-born persons of Mexican descent 0.68/1000 (Harris and Shaw, 1995). Diabetic mothers have 7.9 times the rates of NTDs in their offspring (Becerra et al., 1990). Deficiencies of GSH, folate, vitamin C, riboflavin, zinc, cyancobalamin, sele- nium and excessive exposure to vitamin A have been associated with NTD. Higher rates are seen in children of farmers, cleaning women and nurses (Matte et al., 1993; Blatter et al., 1996). Pre-pregnancy weight has also been demonstrated to be a fac- tor. Werler and colleagues (1996) compared RR for NTD in control women weigh- ing 50–59 kg and found the RR increased to 1.9 in women weighing 80–89 kg and 4.0 for those weighing over 110 kg. AEDs may be a necessary but not sufficient risk factor for the development of NTDs.

Folate deficiency as a potential mechanism of AED teratogenicity

Folate is a coenzyme necessary for the development of white and red blood cells, and proper function of the central nervous system. Normal concentrations are typically measured in the serum (plasma folate 6–20 ng/ml) and erythrocytes (red blood cell folate, RBCF 160–640 ng/ml). Low levels of folate are associated with hyperhomocysteinemia and concentrations required to prevent this are 6.6 ng/ml for SF and 140 ng/ml for RBCF.

Deficiencies of folate have been implicated in the development of birth defects. Dansky et al. (1987) found significantly lower blood folate concentrations in women with epilepsy with abnormal pregnancy outcomes. Co-treatment of mice with folic acid, with or without vitamins and amino acids, reduced malformation rates, and increased fetal weight and length in mice pups exposed to phenytoin in utero (Zhu and Zhou, 1989). Biale and Lewenthal (1984) reported a 15% malformation rate in IMEs with no folate supplementation, whereas none of 33 folate-supplemented children had congenital abnormalities. Eight trials have demonstrated that pre- conceptual folate reduces the risk of recurrence of neural tube defects in women with a previous affected pregnancy (Table 16.2).

305 Antiepileptic drug interactions in pregnancy

Table 16.2 Pre-conceptual folate, after Lewis et al. (in press)

Authors Study type N Dose of folate Results

Smithells et al., 1983

Seller and Nevin, 1984

Mulinare et al., 1988

Milinsky et al., 1989

MRC, 1991

Czeizel and Dudas, 1992

Werler et al., 1993

Werler et al., 1996

Non-randomized, controlled

Non-randomized

Case–control

Cohort

Randomized, double blind, controlled

Randomized, controlled

Case–control Case–control

Fully supplemented 454 Partially
supplemented 519 Unsupplemented 114

Unsupplemented 543 Supplemented 421

Case 181 Control 1480

23 491 1195

Case 2420 Control 2333

Case 436 Control 2615

Case 604 Control 1658

0.36 mg

Multivitamins with folate

4.0 mg

0.8 mg

Folate supplements

86% risk reduction

Risk reduction

60% risk reduction

71% risk reduction 72% risk reduction

No defects
with folate supplementation

60% risk reduction

Folate did not decrease rates in women >70 kg

Unfortunately pre-conceptual folate supplementation may not be protective for women with epilepsy. Craig and colleagues (1999) reported a young woman whose seizures were controlled for 4 years by 2000 mg of VPA a day. Though she took 4.0 mg of folic acid a day for 18 months prior to her pregnancy she delivered a child with a lumbosacral NTD, a ventricular and atrial septal defect, cleft palate and bilateral talipes. Two Canadian women delivered children with NTD despite folate supplementation. One taking 3.5 mg folic acid for 3 months prior to conception and 1250 mg of VPA aborted a child with lumbosacral spina bifida, Arnold Chiari malformation and hydrocephalus. A second woman who took 5.0 mg of folic acid had one spontaneous abortion of a fetus with an encephalocele and two therapeu- tic abortions of fetuses with lumbosacral spina bifida (Duncan et al., 2001). These cases might have been predicted given the demonstrated failure of folate to reduce

306 Mark S. Yerby

NTD and embryotoxicity in vitro and in vivo in rodent models (Hansen and Grafton, 1991; Hansen et al., 1995). In fact not all research supports the association with folate deficiency and malformations. Mills et al. (1992) found no difference between serum folate levels in mothers of children with NTD and controls. A number of other studies also failed to demonstrate a protective effect of pre-conceptual folate (Laurence et al., 1981; Winship et al., 1984; Vergel et al., 1990; Bower and Stanley, 1992; Kirke et al., 1992; Friel et al., 1995). These studies are problematic due to small sample sizes, failure to document folate supplementation and recall bias in the retrospective investigation.

There is evidence to suggest that women with similar folate intake may have dif- ference serum concentrations due to differences in folate metabolism. Absorption does not account for the difference in plasma concentration between cases and controls (Davis et al., 1995).

New AED in pregnancy

A number of new AEDs have been marketed since 1993. Gabapentin, felbamate, lamotrigine, levetiracetam, oxcarbazepine, tiagabine, topiramate and zonisamide are all now available in the US. The numbers of reported exposed pregnancies with these drugs is very low, and unfortunately not large enough for one to determine if there is an increased risk of adverse outcome with fetal exposure to these com- pounds. We know that lamotrigine and levetiracetam concentrations decline during pregnancy and expect that this is also true for the other new AEDs (Tomson et al., 1997). This is what we know to date.

Gabapentin

Lamotrigine

Despite its extensive use for a variety of conditions little has been published about its effect on pregnancy outcomes. A large post-marketing surveillance study of 3100 English patients taking this drug identified 11 pregnancies and no malformations (Wilton and Shakir, 2002). Dr. Georgia Montouris (2002) has collected 51 pregnan- cies from 39 women with epilepsy. The malformation rate was 4.5%. Eighty-seven percent of the pregnancies were live births, there were 11.3% miscarriages and 2% therapeutic abortions.

The International Lamotrigine Pregnancy Registry has identified 334 pregnancies reported in women taking lamotrigine in the first trimester. One hundred and sixty eight of these were with monotherapy, 166 with polytherapy. There is a sig- nificant difference in malformation rates when lamotrigine is used in monotherapy (1.8%), polytherapy with VPA (10%) and polytherapy without VPA (4.3%) (Tennis and Eldridge, 2002).

307 Antiepileptic drug interactions in pregnancy

Lamotrigine clearance increases early in pregnancy and continues to accelerate through all three trimesters in most women taking this medication. In at least one case report the apparent clearance increased by 700% (Pennell et al., 2002).

Lamotrigine crosses the placenta and at delivery fetus and mother have similar plasma concentrations. Elimination in infants appears to be rather slow. Seventy- two hours postpartum infant plasma levels are 75% that of the mother. Median milk/plasma (M/P) ratios are 0.61 (Ohman et al., 2000).

Oxcarbazepine

In the first 12 reported cases of pregnancy with oxcarbazepine there have been nine live births and three spontaneous abortions (Friis et al., 1993). In a prospective study of 11 pregnancies one child with spina bifida exposed to oxcarbazepine in polytherapy was reported. The manufacturer has been notified of five cases of fetal malformations in the post-marketing period. One was a cardiac defect and there were three cleft palates and one facial dysmorphism. Three of the five were exposed to AED polytherapy. The drug has been available in Europe for 10 years, but an accu- rate denominator is not available thus we are unable to calculate rates. In a recent prospective report of 42 pregnant women taking oxcarbazepine, 25 on mono- therapy and 17 on polytherapy, no malformations were seen in the monotherapy group. A child with a ventricular septal defect was exposed to oxcarbazepine and phenobarbital (Rabinowicz et al., 2002). Oxcarbazepine crosses the placenta with equivalent maternal and fetal cord levels (Myllynen et al., 2001).

Topiramate

Zonisamide

We have little information of the number of pregnancies with topiramate exposure. In the clinical trials there were 28 reported pregnancies with one malformation and two children with anomalies. All of these were polytherapy cases. Post-marketing surveillance has collected 139 reports of pregnancy. These included 87 live births, 23 therapeutic abortions, 29 cases lost to follow-up and five cases of hypospadias. Topiramate crosses the placenta with cord and maternal plasma levels being equiv- alent at delivery. M/P concentration ratios average 0.86. Infant elimination appears to be substantial with little measurable drug found in plasma of breast fed infants 2–3 weeks postpartum (Ohman et al., 2002).

There have been 26 reported pregnancies with zonisamide exposure. Two of the 26 (7.7%) had congenital malformations. One child was also exposed to phenytoin and the other to both phenytoin and VPA (Kondo et al., 1996).

Zonisamide also freely crosses the placenta with transfer rates of 92%. Though data is available from only two children M/P ratios are 0.8 and elimination half-life ranges from 61 to 102 h (Kawada et al., 2002).

308 Mark S. Yerby

Syndromes of anomalies

In distinction to malformations, which are deformities of anatomy requiring med- ical or surgical intervention to maintain a functionally healthy person, anomalies are abnormalities of structure, which, while varying from the norm, do not consti- tute a threat to health. Patterns of anomalies in IMEs have been noted with certain AED exposure. Five clinical syndromes have been reported in IMEs: fetal trimetha- dione syndrome, FHS, a primidone embryopathy, a fetal valproate syndrome and a fetal carbamazepine syndrome.

Fetal trimethadione syndrome

FHS

In 1970, German and colleagues described a case of a WWE treated with trimetha- dione who had had four unsuccessful pregnancies. After trimethadione was dis- continued, she went on to have two healthy children. Her physician then surveyed trimethadione-exposed infants delivered at New York Hospital between 1946 and 1968. The records of 278 women with epilepsy were reviewed and, of these, 14 had taken trimethadione during pregnancy. Only 2 of these 14 children were normal. One had multiple hernias and diabetes; 8 had developmental defects; 3 were spon- taneously aborted and only 3 of the 14 actually survived infancy.

The peculiar facial characteristics of these children were delineated by Zackai et al. (1975), who noted that not only were these children short in stature and suffering from microcephaly, they had V-shaped eyebrows epicanthal folds, low set ears, anteriorly folded helices, and irregular teeth. Other abnormalities were often frequent: inguinal hernias, hypospadias and simian creases. Feldman et al. (1977) reviewed 53 pregnancies in which trimethadione was used. In 46 of these (87%), there was fetal loss or the development of a congenital malformation. Follow-up studies of the surviving children have reported significant rates of mental retarda- tion (Goldman et al., 1986).

The most famous and controversial of the dysmorphic syndromes associated with AEDs is the FHS. It was first reported by Loughnan et al. (1973), who described seven infants exposed to hydantoin in combination with a barbiturate, in utero. The children displayed hypoplasia and irregular ossification of the distal pha- langes. In 1974, Barr and co-workers reported distal digital hypoplasia (DDH) in eight children exposed to phenytoin and phenobarbital. The syndrome was given its name by Hanson and Smith (1975), who reported five IMEs who had been exposed to hydantoin in utero. The infants had multiple systemic abnormalities of the face, cranium, and nails, DDH, intrauterine growth retardation, and mental deficiencies. Only one of the five was exposed to phenytoin monotherapy. Of the others, three were exposed to phenobarbital, one to mephobarbital and one to a

309 Antiepileptic drug interactions in pregnancy

combination of phenobarbital, phensuximide and mephenytoin. Despite the multiplicity of exposures, the authors noted the resemblance to the fetal alcohol syndrome and described their cases as suffering from FHS.

Subsequent work by Hanson’s group found that approximately 11% of infants exposed to hydantoin in utero demonstrated the complete syndrome, and an addi- tional 30% would have some anomalous components (Hanson et al., 1976). Many of the features of the syndrome appear to be subjective, but some investigators believe that DDH is a unique and relatively constant feature (Kelly et al., 1984b).

The prevalence and significance of the dysmorphic features of FHS remain unclear. Researchers at the University of Virginia followed 98 women with epilepsy who took phenytoin during pregnancy and found that 30% of their offspring had DDH with no other features of FHS (Kelly et al., 1984b). Gaily et al. (1988a) reported a prospective study of 121 IMEs at the University of Helsinki, 82 of whom were exposed to phenytoin. None of the children had FHS. Hypertelorism and DDH were the only dysmorphic features associated with phenytoin exposure. In our own experience following 64 IMEs, no children with FHS were seen. Dysmorphic features could be seen with any drug exposure (Yerby et al., 1992).

Hanson (1986) feels that there are three components to the syndrome:

1 abnormal growth,
2 abnormal performance,
3 dysmorphic cranial facial features.

An unexpected sequela of the syndrome may be an increased risk of cancer. Four cases of neuroblastoma associated with the FHS have been described since 1976, although all children were also exposed in utero to primidone or phenobarbital. There have also been reports of carcinoma, ganglioneuroblastoma, Wilms’ tumor, a melanotic neuroectodermal tumor and a malignant mesenchymona in children with FHS (Ehrenband and Chaganti, 1981).

The contention that FHS results in abnormal performance or mental deficiency is not supported by subsequent research. Of 103 IMEs exposed to phenytoin, only 1.4% displayed mental deficiency on the Wechsler Preschool and Primary Scale of Intelligence or Leiter International Performance Scale, not significantly different from the general population (Gaily et al., 1988b).

Gaily’s work suggests that there is a genetic component that permits expres- sion of the FHS. Children of mothers with epilepsy who are not exposed to AEDs in utero have frequencies of dysmorphic abnormalities intermediate to those children exposed to AEDs and controls. Dizygotic twins exposed to hydantoins in utero have been shown to display discordant dysmorphism (Phelan et al., 1982; Buchler, 1985). If the first child in a family has FHS, the chance of a second such child is 90%, compared to the 2% chance of having a second child with FHS if the

310 Mark S. Yerby

first is normal (Van Dyke et al., 1988). Such observations suggest that hydantoin exposure may be a necessary but not sufficient cause of infant dysmorphism.

Krauss and co-workers (1984) described four siblings with features of FHS. The first two were exposed to both phenytoin and primidone in utero. In an attempt to prevent further fetal injury, Krauss discontinued the phenytoin and the patient was treated with primidone monotherapy. Two subsequent pregnancies resulted in chil- dren with similar dysmorphic features to their elder siblings.

Primidone embryopathy

Five years before Krauss’ report, Rudd and Freedom (1979) had described craniofacial abnormalities in children exposed to primidone in utero. These children had hirsute foreheads, thick nasal roots, antiverted nostrils, long philtrum, straight thin upper lips and hypoplastic nails. These children were also likely to be small for their gestational age and have psychomotor retardation and heart defects (Gustavson and Chen, 1985).

Fetal valproate syndrome

Reports of dysmorphic children exposed to valproate in utero had previously been made by other investigators (Dalens et al., 1980; Clay et al., 1981), but it was DiLiberti et al. (1984) who described a specific fetal valproate syndrome. They reported seven infants exposed to VPA in utero who had facial abnormalities characterized by interiorepicanthal folds, a net nasal bridge, an upturned nose, a long upper lip, a thin vermillion border, a shallow philtrum and downturned mouth. These children also had abnormalities of their distal digits, and they tended to have long thin over- lapping fingers, toes and hyperconvex nails. Subsequent reports of valproate-exposed infants having radial ray aplasia have also been made.

The prevalence of this syndrome has not yet been established. Jaeger-Roman et al. (1986) described it in 5 of 14 children exposed to valproate monotherapy. In this same group, 43% of the children were distressed at labor, and 28% had low Apgar scores and other major malformations. High doses of valproate were associated with drug withdrawal, hypotonia, and motor and language delay. In a review of 344 women who took valproate during the first trimester of pregnancy, Jeavons (1984) described a 19.8% rate of abnormal deliveries, but no evidence of a dose–response effect with valproate exposure.

Felding and Rane (1984) described an infant with severe congenital liver disease after in utero exposure to VPA and phenytoin. Ardinger and co-workers (1988) reported craniofacial dysmorphism in 19 children exposed to valproate in utero and confirmed the features described by DiLiberti. They also found a large propor- tion of these infants had postnatal growth deficiency and microcephaly, particularly if the children were exposed to polytherapy. The association of valproate with spina bifida is discussed further on.

311 Antiepileptic drug interactions in pregnancy

Benzodiazepine syndrome

Infants exposed to benzodiazepines in utero are at greater risk for intrauterine growth retardation, dysmorphic features and neurological dysfunction. Seven of 37 infants exposed to benzodiazepine drugs in utero were described as hypotonic and hyperexcitable, with dystonic postures and choreoathetotic movements (Laegreid et al., 1987). Delayed hand–eye coordination, psychomotor slowing and a learning disability were also noted. Four infants had major malformations and dysmorphic faces with wide-set eyes, epicanthal folds, upturned noses, dysplastic oracles, high- arched palates, webbed necks and wide-spaced nipples (Laegreid et al., 1987). In a survey of 278 women whose infants had congenital malformations, children with a history of diazepam exposure in the first trimester had a four-fold increase in cleft lip and/or palate (Safra and Oakley, 1975).

Carbamazepine syndrome

The most recently described syndrome of anomalies associated with AED exposure is the carbamazepine syndrome. One group of investigators has described cranio- facial defects (upslanting palpebral fissures, epicanthal folds, short nose, long philtrum), hypoplastic nails, and microcephaly, in 37 IMEs exposed to carbama- zepine monotherapy (Jones et al., 1989). The authors used the Bayley Scale of Infant Development, the Stanford-Binet IV, and the Wechsler Scale of Preschool and Primary Intelligence in their evaluations and found a 20% rate of developmental delay in 25 children of mothers taking carbamazepine monotherapy. They used an unconventional one standard deviation from the mean to define delay, however.

A case of DDH in an IME exposed to carbamazepine monotherapy had been described earlier (Niesen and Froscher, 1985), but that child was otherwise normal. Low birth weight has been reported with in utero exposure to carbamazepine monotherapy (Kallen, 1986b). A reduction in fetal head circumference has been noted in IMEs exposed to carbamazepine (Hiilesmaa et al., 1981). While smaller than control children, the head sizes were still within the normal range. Subsequent studies on the same clinical population failed to find differences in head circumfer-

ence as the children matured (Granstrom, 1987).

Newer AED and anomalies

There have been case reports of anomalies associated with exposure to the newer (introduced after 1993), AEDs, but no drug-specific syndrome of anomalies described. Three children exposed to lamotrigine and VPA have been reported to have dysmorphic facial features of broad nasal bridge, low set ears and hyper- telorism. One child was karyotyped as 47, XXX and another simply had epicanthal folds (GlaxoSmithKline, 2002).

312 Mark S. Yerby

Clinical and laboratory evidence clearly supports the association of certain anti- convulsants with teratogenic effects, especially facial and distal digital anomalies. However, the existence of drug-specific syndromes is doubtful. Facial dysmor- phism is difficult to quantify and clearly is not drug specific. Infants of epileptic mothers with similar dysmorphic features have been described in the pre- anticonvulsant era (Baptist, 1938; Philbert and Dam, 1982). Follow-up of these infants into adult life has yet to be accomplished, and therefore the significance of these anomalies is unclear. Gaily et al. (1988a) followed a cohort of IMEs to 51⁄2 years of age. These children had more minor anomalies characteristic of FHS than control children but so did their mothers. Only hypertelorism and digital hypopla- sia were associated with phenytoin exposure. Certain anomalies, particularly epi- canthal folds, appeared to be associated with maternal epilepsy, not with AED exposure.

The hypothesized association of dysmorphic features with mental retardation (Kelly, 1984a) has not been confirmed (Hutch et al., 1975; Granstrom, 1982). In the few cases that have been followed into early childhood, the dysmorphic features tend to disappear as the child grows older (Janz, 1982). Mental deficiency was found in only 1.4% of IMEs followed to 51⁄2 years of age (Gaily et al., 1988b). Exposure to AEDs below toxic concentrations or to maternal seizures did not increase the risk of lower intelligence. No association between features of FHS and mental retardation could be demonstrated.

The primary abnormalities in these syndromes involve the midface and distal digits. A retrospective study spanning 10 years of deliveries in Israel found hyper- telorism to be the only anomaly seen more often in IME than in controls (Neri et al., 1983). This was associated with all AEDs except primidone. A prospective study of 172 deliveries of IMEs evaluated eight specific AEDs and other potential con- founding factors and found no dose-dependent increase in the incidence of mal- formations associated with any individual AED. Furthermore, no specific defect could be associated with individual AED exposure (Kaneko et al., 1988). It has been suggested that, since a variety of similar anomalies of the midface and distal digits are seen in a small proportion of children exposed to anticonvulsants in utero, a better term for the entire group of abnormalities would be fetal anti- convulsant syndrome or AED embryopathy (Dieterich et al., 1980; Vorhees, 1986; Huot et al., 1987).

Neonatal complications associated with AED

A unique neonatal hemorrhagic phenomenon has been described in the IME. It differs from other hemorrhagic disorders in infancy in that the bleeding tends to occur internally during the first 24 h of life. It was initially associated with exposure to phenobarbital or primidone, but has subsequently also been described in children

313 Antiepileptic drug interactions in pregnancy

exposed to phenytoin, carbamazepine, diazepam, mephobarbital, amobarbital and ethosuximide (Van Creveld, 1957; Mountain et al., 1970). One group of investigators suggests that vigabatrin may also increase the risk of neonatal hemorrhage (Howe et al., 1999). Prevalence figures are as high as 30% but appear to average 10%. Mortality is high, over 30%, because bleeding occurs within internal cavities and is often not noticed until the child is in shock.

The hemorrhage is the result of a deficiency of vitamin K-dependent clotting factors II, VII, IX and X. Anticonvulsants can act like warfarin, and inhibit vitamin K transport across the placenta. This results in the increase in an abnormal pro- thrombin induced by vitamin K absence of factor II (PIVKA-II). Maternal coagu- lation parameters are invariably normal. The fetus, however, will demonstrate increased levels of PIVKA, diminished clotting factors, and prolonged prothrom- bin and partial thromboplastin times. PIVKA-II has been demonstrated in 54% of infants exposed to AEDs in utero compared to 20% of controls (P 0.01), and maternal vitamin K concentrations are lower in WWE than those untreated though PIVKA is rarely detectable in mothers (Cornelissen et al., 1993).

This phenomenon can be prevented by maternal ingestion of oral vitamin K in the last month of gestation (Deblay et al., 1982; Crawford, 2002b). I use 10 mg/day of oral vitamin K. Routine intramuscular administration of vitamin K at birth is not adequate to prevent hemorrhage within the first 24 h of life.

The prevalence of AED-associated neonatal hemorrhage is unclear. One report states it is 1.6 times as common in IME as controls (Speidel and Meadow, 1972). A more recent prospective study followed 667 IME and 1334 controls and found neonatal bleeding in 5 of 667 (0.7%) IME and 5 of 1334 (0.4%) of controls. While more prevalent there was no statistical difference between the groups. The authors concluded that there was no increased risk for neonatal bleeding in the IME (Kaaja et al., 2002). I would point out that the sample size for a low frequency outcome such as this may need to be larger and there was clearly a trend for more bleeding in the IME.

Developmental complications associated with AED

Developmental delay

IME have been reported to have higher rates of mental retardation than controls. This risk is increased by a factor of two- to seven-fold according to various authors (Speidel and Meadow, 1972; Hill et al., 1974). None of these studies controlled for parental intelligence, although differences in IQ scores at age 7 between groups of children exposed (full-scale IQ, FSIQ 91.7) or not exposed (FSIQ 96.8) to phenytoin reached statistical significance, the clinical significance of such difference is unknown (Hill and Tennyson, 1982).

314 Mark S. Yerby

We have found that IME display lower scores in measures of verbal acquisition at both 2 and 3 years of age. Though there was no difference in physical growth parameters between IME and controls, IME scored significantly lower in the Bailey Scale of Infant Development’s mental developmental index (MDI) at 2 and 3 years. They also performed significantly less well on the Bates Bretherton early language inventory (P 0.02) and in the Peabody Picture Vocabulary’s scales of verbal rea- soning (P 0.001) and composite IQ (P 0.01), and they displayed significantly shorter mean lengths of utterance (P 0.001) (Leavitt et al., 1992).

Polytherapy-exposed infants performed significantly less well on neuropsycho- metric testing than those exposed to monotherapy. Socioeconomic status had the strongest association with poor test scores, but maternal seizures during pregnancy was also a significant risk factor (Losche et al., 1994).

Leonard et al. (1997) has in part addressed the question of whether maternal seizures or in utero exposure to AEDs are responsible for the developmental delay seen. A group of children of mothers with epilepsy followed to school age were found to have a rate of intellectual deficiency of 8.6%. The Wechsler Intelligence Scale for Children revealed significantly lower scores for children exposed to seizures during gestation (100.3), than for children whose mother’s seizures were controlled (104.1) or controls (112.9). All AEDs are clearly not created equal and Koch and co-workers (1999) have demonstrated that primidone, particularly when used in polytherapy, is associated with lower Wechsler score of intelligence.

Conclusion

The potential interactions of AEDs in pregnant women with epilepsy can be char- acterized by those effecting the mother, and those effecting the fetus. While preg- nancy, maternal seizures and AEDs pose risks for successful pregnancy outcome, the majority of patients can and do have healthy children. Physicians cannot elim- inate risk, but can reduce it. Pre-conceptual folic acid is an approved intervention but may not prevent all malformations. Though there are no head to head studies of the safety of AEDs in pregnancy some principles have been clearly established. Monotherapy is safer than polytherapy. Phenobarbital is no safer than, and proba- bly more hazardous than, other AEDs in monotherapy. VPA has in addition to the underlying increased risk for malformations an additional risk for development of NTDs. The newer AEDs have theoretical advantages over older ones in terms of malformations but the sample sizes collected to date are not adequate to determine relative safety. Malformations are not the only adverse outcome that one should be concerned about. Developmental delay is, in terms of magnitude, as significant as birth defects. There is no drug-specific syndrome of anomalies but a tendency for all AEDs to cause facial dysmorphism, which is a relatively transient condition.

315 Antiepileptic drug interactions in pregnancy

Given the nature of the data available to date clinical judgement in determining the most effective AED for the seizure type and using the lowest effective dose is still the best approach.

REFERENCES

Abrishamchian AR, Khoury MJ, Calle EE. The contribution of maternal epilepsy and its treat- ment to the etiology of oral clefts: a population based case–control study. Genet Epidemiol 1994; 11(4): 343–351.

Anderson RC. Cardiac defects in children of mothers receiving anticonvulsant therapy during pregnancy. J Pediatr 1976; 89: 318–319.

Annegers JF, Hauser WA, Elveback LR, et al. Congenital malformations and seizure disorders in the offspring of parents with epilepsy. Int J Epidemiol 1978; 7: 241–247.

Ardinger HH, Atkin JF, Blackston RD, et al. Verification of the fetal valproate syndrome pheno- type. Am J Med Genet 1988; 29: 171–185.

Asuncion M, Calvo RM, San Millan JL, et al. A prospective study of the prevalence of the poly- cystic ovary syndrome in unselected Caucasian women from Spain. J Clin Endocrinol Metab 2000; 85: 2434–2438.

Barr M, Pozanski AK, Schmickel RD. Digital hypoplasia and anticonvulsant during gestation: a teratogenic syndrome? J Pediatr 1974; 84: 254–256.

Becerra JE, Khoury MJ, Cordero JF, et al. Diabetes mellitus during pregnancy and the risks for specific birth defects: a population based case control study. Pediatrics 1990; 85: 1–9.

Biale Y, Lewenthal H. Effect of folic acid supplementation on congenital malformations due to anticonvulsant drugs. Eur J Obstet Gynecol Reprod Biol 1984; 18: 211–216.

Bilo L, Meo R, Valentino R, et al. Characterization of reproductive endocrine disorders in women with epilepsy. J Clin Endocrinol Metab 2001; 86: 2950–2956.

Blatter BM, Roeleveld N, Zielhuis GA, et al. Spina bifida and prenatal occupation. Epidemiology 1996; 7: 188–193.

Borgfeldt C, Andolf E. Transvaginal sonographic ovarian findings in a random sample of women 25–40 years old. Ultrasound Obstet Gynecol 1999; 13(5): 345–350.

Bower C, Stanley FJ. Periconceptual vitamin supplementation and neural tube defects: evidence from a case control study in Western Australia and a review of recent publications. J Epidemiol Commun Health 1992; 46: 157–161.

Buchler BA. Epoxide hydrolase activity and the fetal hydantoin syndrome. Clin Res 1985; 33: A129.

Cahill, WT, Kovilam OP, Pastor D, et al. Neurologic and fetal outcomes of pregnancies of mothers with epilepsy. Epilepsia 2002; 43(Suppl. 7): 289.

Chang T, Savory A, Glazko AJ. A new metabolite of 5,5-diphenylhydantoin (Dilantin). Biochem Biophys Res Commun 1970; 38: 444–449.

Chappell KA, Markowitz JS, Jackson CW. Is valproate pharmacotherapy associated with poly- cystic ovaries? Ann Pharmacother 1999; 33: 1211–1216.

316 Mark S. Yerby

Clay SA, McVie R, Chen HC. Possible teratogenic effect of valproic acid. J Pediatr 1981; 98: 828. Collaborative Drug Use in Pregnancy: an international survey on drug use in pregnancy. Int J

Risk Safety Med 1991; 1: 1.
Cornelissen M, Steegers-Theunissen R, Kollee L, et al. Increased incidence of neonatal vitamin K

deficiency resulting from maternal anticonvulsant therapy. Am J Obstet Gynecol 1993; 168

(3 Pt 1): 923–928.
Craig J, Morrison P, Morrow J, et al. Failure of preconceptual folic acid to prevent a neural tube

defect in the offspring of a mother taking sodium valproate. Seizure 1999; 8: 253–254. Crawford P. Epilepsy and pregnancy. Seizure 2002a; 11(Suppl. A): 212–219.
Crawford P. Interactions between antiepileptic drugs and hormonal contraception. CNS Drug

2002b; 16(4): 263–272.
Czeizel AE, Dudas I. Prevention of the first occurrence of neural tube defects by preconceptional

vitamin supplementation. New Engl J Med 1992; 327: 1832–1835.
Dalens B, Raynaud EJ, Gaulme J. Teratogenicity of valproic acid. J Pediatr 1980; 97: 332–333. Dam M, Christiansen J, Munck O, et al. Antiepileptic drugs: Metabolism in pregnancy. Clinical

Pharmacokinetics 1979; 4: 53–62.
Dansky LV, Andermann E, Sherwin AL, et al. Maternal epilepsy and congenital malformations:

a prospective study with monitoring of plasma anticonvulsant levels during pregnancy.

Neurology 1980; 3: 15.
Dansky LV, Andermann E, Andermann F, et al. Maternal epilepsy and congenital malformations:

Correlation with maternal plasma anticonvulsants levels during pregnancy. In Janz D, Dam M, Richens A, Bossi L, Helge H, Schmidt D, eds. Epilepsy, Pregnancy and the Child. New York: Raven Press, 1982: 251–258.

Dansky LV, Strickler SM, Andermann E, et al. Pharmacogenetic susceptibility to phenytoin teratogenesis. In The XIth Epilepsy International Symposium. Advances in Epileptology, vol. 16. P. Wolf, M. Dam, D. Janz, F. E. Dreifuss, eds. New York: Raven Press, 1987: 555–559.

Davis BA, Bailey LB, Gregory JF, et al. Folic acid absorption in women with a history of preg- nancy with neural tube defect. Am J Clin Nutr 1995; 62: 782–784.

Deblay MF, Vert P, Andre M, et al. Transplacental vitamin K prevents hemorrhagic disease of infants of epileptic mothers. Lancet 1982; 1: 1247.

Dieterich E, Steveling A, Lukas A, et al. Congenital anomalies in children of epileptic mothers and fathers. Neuropediatrics 1980; 11: 274–283.

DiLiberti JH, Farndon PA, Dennis NR, et al. The fetal valproate syndrome. Am J Med Genet 1984; 19: 473–481.

Doose DR, Jacobs D, Squires L, et al. Oral contraceptive-AED interactions: no effect of topiramate as monotherapy at clinically effective dosages of 200 mg or less. Epilepsia 2002; 34(Suppl. 7): 235.

Duncan S, Mercho S, Lopes-Cendes I, et al. Repeated neural tube defects and valproate monotherapy suggest a pharmacogenetic abnormality. Epilepsia 2001; 42(6): 750–753.

Ehrenband LT, Chaganti RSK. Cancer in the fetal hydantoin syndrome. Lancet 1981; 1: 97. Elshove J, Van Eck JHM. Aangeboren misvorminge, met name gespleten lipmet zonder gespleten verhemelte, bij kinderen van moeders met epilepsie. Nederland T Geneesk 1971; 115(33):

1371–1375.

317 Antiepileptic drug interactions in pregnancy

Fedrick J. Epilepsy and pregnancy: a report from the Oxford record linkage study. Br Med J 1973; 2: 442–448.

Felding I, Rane A. Congenital liver damage after treatment of mother with valproic acid and phenytoin? Acta Pediatr Scand 1984; 73: 565–568.

Feldman GL, Weaver DD, Lovrien EW. The fetal trimethadione syndrome: report of an additional family and further delineation of this syndrome. Am J Dis Child 1977; 131(13): 89–92.

Finnell RH, DiLiberti JH. Hydantoin induced teratogenesis: are arene oxide intermediates really responsible? Helv Paediatr Acta 1983; 38: 171–177.

Freil JK, Frecker M, Fraser FC. Nutritional patterns of mothers of children with neural tube defects in Newfoundland. Am J Med Genet 1995; 55: 195–199.

Friis ML, Breng-Nielsen B, Sindrup EH, et al. Facial clefts among epileptic patients. Arch Neurol 1981; 38: 227–229.

Friis ML, Holm NV, Sindrup EH, et al. Facial clefts in sibs and children of epileptic patients. Neurology 1986; 38: 346–350.

Friis ML, Kristensen O, Boas J, et al. Therapeutic experiences with 947 epileptic out-patients in oxcarbazepine treatment. Acta Neurol Scand 1993; 87(3): 224–227.

Gaily E, Granstrom ML, Hiilesmaa V, et al. Minor anomalies in offspring of epileptic mothers. J Pediatr 1988a; 112: 520–529.

Gaily E, Sorsa EK, Granstrom ML. Intelligence of children of epileptic mothers. J Pediatr 1988b; 113: 677–684.

Gatoh N, Millo Y, Taube E, et al. Epilepsy among parents of children with cleft lip and palate. Brain Dev 1987; 9: 296–299.

Genton P, Bauer J, Duncan S, et al. On the association between valproate and polycystic ovary syndrome. Epilepsia 2001; 42: 295–304.

German J, Kowal A, Ellers KH. Trimethadione and human teratogenesis. Teratology 1970; 3: 349–362.

GlaxoSmithKline. Lamotrigine Pregnancy Registry Interim Report 2002.
Golden JA, Chenroff GF. Multiple sites of anterior neural tube closure in humans: evidence from

anterior neural tube defects (anencephaly). Pediatrics 1995; 95: 506–510.
Goldman AS, Zachai EH, Yaffe SJ. Environmentally induced birth defect risks. In Teratogen

Update. J. L. Sever, R. L. Brent, eds. New York: Liss, 1986: 35–38.
Govind A, Obhrai MS, Clayton RN. Polycystic ovaries are inherited as an autosomal dominant

trait: analysis of 29 polycystic ovary syndrome and 10 control families. J Clin Endocrinol Metab

1999; 84(1): 38–43.
Graf WD, Pippenger CE, Shurtleff DB. Erythrocyte antioxidant enzyme activities in children

with myelomeningocele. Dev Med Child Neurol 1995; 37: 900–905.
Granstrom ML. Development of the children of epileptic mothers, preliminary results from the

prospective Helsinki study. In Janz D, Dam M, Richens A, Bossi L, Helge H, Schmidt D (eds)

Epilepsy Pregnancy and the Child. New York: Raven Press, 1982: 403–408.
Granstrom ML. Early postnatal growth of the children of epileptic mothers. In The XVIth Epilepsy International Symposium. Advances in Epileptology, vol. 16. P. Wolf, M. Dam, D. Janz,

F. E. Dreifuss, eds. New York: Raven Press, 1987: 573–577.

318 Mark S. Yerby

Gustavson EE, Chen H. Goldenhar syndrome, anterior encephalocele and aqueductal stenosis following fetal primidone exposure. Teratology 1985.

Hansen DK, Grafton TF. Lack of attenuation of valproic acid induced effects by folinic acid in rat embryos in vitro. Teratology 1991; 43: 575–582.

Hansen DK, Grafton TF, Dial SL, et al. Effect of supplemental folic acid on valproic acid induced embryotoxicity and tissue zinc levels in vivo. Teratology 1995; 52: 277–285.

Hanson JW. Teratogen update: fetal hydantoin effects. Teratology 1986; 33: 349–553.
Hanson JW, Smith DW. The fetal hydantoin syndrome. J Pediatr 1975; 87: 285–290.
Hanson JW, Myrianthopoulos NC, Sedgwich MA, et al. Risks to the offspring of women treated

with hydantoin anticonvulsants with emphasis on the fetal hydantoin syndrome. J Pediatr

1976; 89: 662–668.
Harris JA, Shaw GM. Neural tube defects – why are rates high among population of Mexican

descent? Environ Health Perspect 1995; 103(Suppl. 6): 163–164.
Hiilesmaa VK. Pregnancy and birth in WWE. Neurology 1992; 42(Suppl. 5): 8–11.
Hiilesmaa VK, Teramo K, Granstrom ML, et al. Fetal head growth retardation associated with

maternal antiepileptic drugs. Lancet 1981; 2: 165–167.
Hill RM, Verniaud WM, Horning MG, et al. Infants exposed in utero to antiepileptic drugs. A

prospective study. Am J Dis Child. 1974; 127: 645–653.
Hill RM, Tennyson L. Premature delivery, gestational age, complications of delivery, vital data

at birth on newborn infants of epileptic mothers: review of the literature. In Epilepsy, Pregnancy and the Child. D. Janz, L. Bossi, M. Dam, et al., eds. New York: Raven Press, 1982: 167–173.

Hill D, Pond DA, Mitchell W, et al. Personality change following temporal lobectomy for epilepsy. J Ment Sci 1957; 103: 18–27.

Holmes LB, Wyszynski DF, Lieberman E. The AED pregnancy registry. A six year experience. Arch Neurol 2004; 61: 673–678.

Honein MA, Paulozzi LJ, Mathews TJ, et al. Impact of folic acid fortification of the US food sup- ply on the occurrence of neural tube defects. J Am Med Assoc 2001; 285(23): 2981–2986.

Horning MG, Stratton C, Wilson A, et al. Detection of 5-(3-4)-diphenylhydantoin in the newborn human. Anal Lett 1974; 4: 537–582.

Howe AM, Oakes DJ, Woodman PD, et al. Prothrombin and PIVKA-II levels in cord blood from newborn exposed to anticonvulsants during pregnancy. Epilepsia 1999; 40(7): 980–984.

Huot C, Gauthier M, Lebel M, et al. Congenital malformations associated with maternal use of valproic acid. Can J Neurol Sci 1987; 14: 290–293.

Isojarvi JI, Laatikainen TJ, Pakarinen AJ, et al. Polycystic ovaries and hyperandrogenism in women taking valproate for epilepsy. New Engl J Med 1993; 329(19): 1383–1388.

Jaeger-Roman E, Deichl A, Jakob S, et al. Fetal growth, major malformations, and minor anom- alies in infants born to women receiving valproic acid. J Pediatr 1986; 108: 997–1004.

Janz D. Antiepileptic drugs and pregnancy: altered utilization patterns and teratogenesis. Epilepsia 1982; 23(Suppl. 1): 853–863.

Jeavons PM. Non dose related side effects of valproate. Epilepsia 1984; 25(Suppl. 1): 550–555. Jerina DM, Daly JW. Arene oxides: a new aspect of drug metabolism. Science 1974; 185: 573.

319 Antiepileptic drug interactions in pregnancy

Jick SS, Terris BZ. Anticonvulsants and congenital malformations. Pharmacotherapy 1997; 17(3): 561–564.

Jones KL, Lacro RV, Johnson KA, et al. Pattern of malformations in the children of women treated with carbamazepine during pregnancy. New Engl J Med 1989; 320: 1661–1666.

Kaaja E, Kaaja R, Matila R, et al. Enzyme-inducing antiepileptic drugs in pregnancy and the risk of bleeding in the neonate. Neurology 2002; 58(4): 549–553.

Kallen B. A register study of maternal epilepsy and delivery outcome with special reference to drug use. Acta Neurol Scand 1986a; 73(3): 253–259.

Kallen B. Maternal epilepsy, antiepileptic drugs and birth defects. Pathologia 1986b; 78: 757–768. Kaneko S, Otani K, Fukushima Y, et al. Teratogenicity of antiepileptic drugs: analysis of possible

risk factors. Epilepsia 1988; 29: 459–467.
Kaneko S, Battino D, Andermann E, et al. Congenital malformations due to antiepileptic drugs.

Epilepsy Res 1999; 33(2–3): 145–158.
Kahsar-Miller MD, Nixon C, Boots LR, et al. Prevalence of polycystic ovary syndrome (PCOS) in

first-degree relatives of patients with PCOS. Fertil Steril 2001; 75(1): 53–58.
Kawada K, Itoh S, Kusaka T, et al. Pharmacokinetics of zonisamide in perinatal period. Brain Dev

2002; 24: 95–97.
Kelly TE. Teratogenicity of anticonvulsant drugs I. Review of literature. Am J Med Genet 1984a;

19: 413–434.
Kelly TE, Edwards P, Rein M, et al. Teratogenicity of anticonvulsant drugs II. A prospective study.

Am J Med Genet 1984b; 19: 435–443.
Kirke PN, Daly LE, Elwood JH. A randomized trial of low dose folic acid to prevent neural tube

defects. Arch Dis Child 1992; 67: 1442–1446.
Koch S, Titze K, Zimmermann RB, et al. Long-term neuropsychological consequences of

maternal epilepsy and anticonvulsant treatment during pregnancy for school age children and

adolescents. Epilepsia 1999; 40: 1237–1243.
Kondo T, Kaneko S, Amano Y, et al. Preliminary report on teratogenic effects of zonisamide in

the offspring of treated women with epilepsy. Epilepsia 1996; 37(12): 1242–1244.
Krauss CM, Holmes LB, Van Lang QC, et al. Four siblings with similar malformations after expo-

sure to phenytoin and primidone. J Pediatr 1984; 105: 750–755.
Krishnamurthy KB, Sundstrom DT, Beaudoin JM, et al. Pregnant women with epilepsy taking

older anticonvulsant medications must have drug levels checked frequently to avoid seizures.

Epilepsia 2002; 43(Suppl. 7): 232–233.
Kubow S, Wells PG. In vitro bioactivation of phenytoin to a reactive free radical intermediate

by prostaglandin synthetase, horseradish peroxidase, and thyroid peroxidase. Mol Pharmacol

1989; 35: 504–511.
Laegreid L, Olegard R, Wahlstrom J, et al. Abnormalities in children exposed to benzodiazepines

in utero. Lancet 1987; 1: 108–109.
Laurence KM, James J, Miller MH, et al. Double blind randomized controlled trial of folate treat-

ment before conception to prevent recurrence of neural tube defects. Br Med J 1981; 282:

1509–1511.
Leavitt AM, Yerby MS, Robinson N, et al. Epilepsy and pregnancy: developmental outcomes at

12 months. Neurology 1992; 42(Suppl. 5): 141–143.

320 Mark S. Yerby

Leonard G, Andermann E, Pitno A, et al. Cognitive effects of antiepileptic drug therapy during pregnancy on school age offspring. Epilepsia 1997; 38(Suppl. 3): 170.

Levy RH, Yerby MS. Effect of pregnancy on antiepileptic drug utilization. Epilepsia 1985; 26(Suppl. 1): 525–557.

Lindhout D, Schmidt D. In utero exposure to valproate and neural tube defects. Lancet 1986; 1: 1392–1393.

Lindhout D, Hoeppener RJEA, Meinardi H. Teratogenicity of antiepileptic drug combinations with special emphasis on epoxidation (of carbamazepine). Epilepsia 1984; 25: 77–83.

Lindhout D, Omtzigt JG, Cornel MC. Spectrum of neural tube defects in 34 infants prenatally exposed to antiepileptic drugs. Neurology 1992; 42(4 Suppl. 5): 111–118.

Little BB, Santos-Ramos R, Newell JF, et al. Megadose carbamazepine during the period of neu- ral tube closure. Obstet Gynecol 1993; 82(4 Pt 2 Suppl.): 705–708.

Losche G, Steinhausen H-C, Koch S, et al. The psychological development of children of epilep- stic parents. II. The differential impact of intrauterine exposure to anticonvulsant drugs and further influential factors. Acta Paediatr 1994; 83(9): 961–966.

Loughnan PM, Gold H, Vance JC. Phenytoin teratogenicity in man. Lancet 1973; 1: 70–72. Lowe CR. Congenital malformations among infants born to epileptic women. Lancet 1973; 1:

9–10.
Majewski F, Raft W, Fischer P, et al. Zur tertogenitat von anticonvulsiva. Deut Med Wochenschr

1980; 105: 719–723.
Martz F, Failinger C, Blake D. Phenytoin teratogenesis: correlation between embryopathic effect

and covalent binding of putative arene oxide metabolite in gestational tissue. J Pharmacol Exp

Ther 1977; 203: 231–239.
Matte TD, Mulinare J, Erickson JD. Case–control study of congenital defects and parental

employment in health care. Am J Ind Med 1993; 24: 11–23.
Mattson RH, Cramer JA, Darney PD, et al. The use of oral contraceptives by women with

epilepsy. J Am Med Assoc 1986; 256(2): 238–240.
Milinsky A, Jick H, Jick SS, et al. Multivitamin/folic acid supplementation in early pregnancy

reduces the prevelance of neural tube defects. J Am Med Assoc 1989; 262: 2847–2852.
Mills JL, Tuomileho J, Yu KF, et al. Maternal vitamin levels during pregnancies producing infants

with neural tube defects. J Pediatr 1992; 120: 863–871.
Monson RR, Rosenberg L, Hartz SC. Diphenylhydantoin and selected congenital malformations.

New Engl J Med 1973; 289: 1049–1052.
Montouris G. Safety of gabapentin treatment during pregnancy. Epilepsia 2002; 43(Suppl. 7):

234.
Mountain KR, Hirsh J, Gallus AS. Maternal coagulation defect due to anticonvulsant treatment

in pregnancy. Lancet 1970; 1: 265–268.
MRC Vitamin Study Group. Prevention of neural tube defects: Results of the Medical Research

Council Vitamin Study. Lancet 1991; 338: 131–137.
Mulinare J, Cordero JF, Erickson JD, et al. Periconceptional use of multivitamins and the occur-

rence of neural tube defects. J Am Med Assoc 1988; 260: 3141–3145.
Myllynen P, Pienimaki P, Jouppila P, et al. Transplacental passage of oxcarbazepine and its

metabolites in vivo. Epilepsia 2001; 42(11): 1482–1485.

321 Antiepileptic drug interactions in pregnancy

Nakane Y. Congenital malformations among infants of epileptic mothers treated during preg- nancy. Folia Psychiatr Neurol Japon 1979; 33: 363–369.

Nakane Y, Okuma T, Takahashi R, et al. Multi-institutional study on the teratogenicity and fetal toxicity of antiepileptic drugs: report of a collaborative study group in Japan. Epilepsia 1980; 21: 663–680.

Nau H, Rating D, Koch S, et al. Valproic acid and its metabolites: placental transfer, neonatal pharmacokinetics, transfer via mother’s milk and clinical status in neonates of epileptic mothers. J Pharmacol Exp Ther 1981; 219: 768–777.

Nebert DW, Jensen NM. The Ah locus: genetic regulation of the metabolism of carcinogens, drugs, and other environmental chemicals by cytochrome P-450 mediated mono-oxygenases. CRC Crit Rev Biochem 1979; 6: 401–437.

Neri A, Heifetz L, Nitke S, et al. Neonatal outcomes in infants of epileptic mothers. Eur J Obstet Gynecol Reproduct Biol 1983; 16: 263–268.

Niesen M, Froscher W. Finger and toenail hypoplasia after carbamazepine monotherapy in late pregnancy. Neuropediatrics 1985; 16: 167–168.

Ohman I, Vitols S, Tomson T. Lamotrigine in pregnancy: pharmacokinetics during delivery, in the neonate, and during lactation. Epilepsia 2000; 41(6): 709–713.

Ohman I, Vitols S, Luef G, et al. Topiramate kinetics during delivery, lactation, and in the neonate: preliminary observations. Epilepsia 2002; 43(10): 1157–1160.

Olafsson E, Hallgrimsson JT, Hauser WA, et al. Pregnancies of women with epilepsy: a popula- tion-based study in Iceland. Epilepsia 1998; 39(8): 887–892.

Omtzigt JG, Los FJ, Hagenaars AM, et al. Prenatal diagnosis of spina bifida aperta after first trimester valproate exposure. Prenat Diagn 1992; 12(11): 893–897.

Otani K. Risk factors for the increased seizure frequency during pregnancy and the puerpurium. Fol Psychiar Neurol Japan 1985; 39: 33–44.

Pacifici GM, Rane A. Metabolism of styrene oxide in different human fetal tissues. Drug Metab Dispos 1982; 10: 302–305.

Pacifici GM, Colizzi C, Giuliani L, et al. Cytosolic epoxide hydrolase in fetal and adult human liver. Arch Toxicol 1983; 54: 331.

Pantarotto C, Arboix M, Sezzano P, et al. Studies on 5,5-diphenylhydantoin irreversible binding to rat liver microsomal proteins. Biochem Pharmacol 1982; 31: 1501–1507.

Pennell PB, Montgomery JQ, Clements SD, et al. Lamotrigine clearance markedly increases during pregnancy. Epilepsia 2002; 43(Suppl. 7): 234–235.

Perruca E, Crema A. Plasma protein binding of drugs in pregnancy. Clinical Pharmacokinetics. 1982; 7: 336–352.

Phelan MC, Pellock JM, Wance WE. Discordant expression of fetal hydantoin syndrome in heteropaternal dizygotic twins. New Engl J Med 1982; 307: 99–101.

Philbert A, Dam M. The epileptic mother and her child. Epilepsia 1982; 23: 85–99.
Rabinowicz A, Meischenguiser R, Ferraro SM, et al. Single-center 7 year experience of oxcar-

bazepine exposure during pregnancy. Epilepsia 2002; 43(Suppl. 7): 208–209.
Rasgon NL, Altshuler LL, Gudeman D, et al. Medication status and polycystic ovary syn- drome in women with bipolar disorder: a preliminary report. J Clin Psychiatr 2000; 61:

173–178.

322 Mark S. Yerby

Rodriguez-Palomares C, Belmont-Gomez A, Amancio-Chassin O, et al. Phenytoin serum con- centration monitoring during pregnancy and puerperium in Mexican epileptic women. Arch Med Res 1995; 26: 371–377.

Robert E, Guibaud P. Maternal valproic acid and congenital neural tube defects. Lancet 1982; 11: 937.

Rosa FW. Spina bifida in infants of women treated with carbamazepine during pregnancy. New Engl J Med 1991; 324: 674–677.

Rudd NL, Freedom RM. A possible primidone embryopathy. J Pediatr 1979; 94: 835–837. Sabers A, Ohman I, Christensen J, et al. Oral contraceptives reduce lamotrigine plasma levels.

Neurology 2003; 61(4): 570–571.
Safra MJ, Oakley GP. Association between cleft lip with or without cleft palate and prenatal expo-

sure to diazepam. Lancet 1975; 2: 478–480.
Schmidt D, Canger R, Avanzini G, et al. Change of seizure frequency in pregnant epileptic

women. J Neurol Neurosurg Psychiatr 1983; 46: 751–755.
Seller MJ, Nevin NC. Periconceptional vitamin supplementation and the prevention of

neural tube defects in south-east England and Northern Ireland. J Med Genet 1984; 21(5):

325–330.
Seller MJ. Recent developments in the understanding of the aetiology of neural tube defects. Clin

Dysmorphol 1995; 4: 93–104.
Shane-McWhorter L, Cerveny JD, MacFarlane LL, et al. Enhanced metabolism of levonorgestrel

during phenobarbital treatment and resultant pregnancy. Pharmacotherapy 1998; 18(6):

1360–1364.
Shapiro S, Slone D, Hartz SC, et al. Anticonvulsant and parental epilepsy in the development of

birth defects. Lancet 1976; 1: 272–275.
Shorvon SD, Tallis RC, Wallace HK. Antiepileptic drugs: coprescription of proconvulsant drugs

and oral contraceptives: a national study of antiepileptic drug prescribing practice. J Neurol

Neurosurg Psychiatr 2002; 72(1): 114–115.
Shum S, Jensen NM, Nebert DW. The Ah locus: in utero toxicity and teratogenesis associated

with genetic differences in B(a)P metabolism. Teratology 1979; 20: 365–376.
Sims P, Grover PL. Epoxides in polycyclic aromatic hydrocarbon metabolism and carcinogene-

sis. Adv Cancer Res 1974; 20: 165.
Smithells RW, Nevin NC, Seller MJ. Further experience of vitamin supplementation for preven-

tion of neural tube defect recurrences. Lancet 1983; 1: 1027–1031.
South J. Teratogenic effects of anticonvulsants. Lancet 1972; 2: 1154.
Speidel BD, Meadow SR. Maternal epilepsy and abnormalities of the fetus and newborn. Lancet

1972; 2: 839–843.
Spielberg SP, Gordon GB, Blake DA, et al. Anticonvulsant toxicity in vitro: possible role of arene

oxides. J Pharmacol Exp Ther 1981; 217: 386–389.
Steegers-Theunissen RP, Reiner WO, Borm GF, et al. Factors influencing the risk of abnormal

pregnancy outcomes in epileptic women: a multicenter prospective study. Epilepsy Res 1994;

18(3): 261–269.
Strickler SM, Dansky LV, Miller MA, et al. Genetic predisposition to phenytoin-induced birth

defects. Lancet 1985; 2: 746–749.

323 Antiepileptic drug interactions in pregnancy

Tennis P, Eldridge RR. Scientific Advisory Committee. International Lamotrigine Pregnancy Registry. Epilepsia 2002; 43(10): 1161–1167.

Tomson T, Lindbom U, Ekqvist B, et al. Epilepsy and pregnancy: a prospective study of seizure control in relation of free and total plasma concentrations of carbamazepine and phenytoin. Epilepsia 1994; 35: 122–130.

Tomson T, Ohman I, Vitols S. Lamotrigine in pregnancy and lactation: a case report. Epilepsia 1997; 38(9): 1039–1041.

Tran TA, Leppik IE, Blesi K, et al. Lamotrigine clearance during pregnancy. Neurology 2002; 59(2): 251–255.

Vajda FJ, O’Brien TJ, Hitchcock A, et al. The Australian Registry of antiepileptic drugs in preg- nancy: multivariate regression analysis demonstrating an increased risk for valproate, with a dose-dependent relation. Epilepsia 2002; 43(Suppl. 7): 211.

Van Allen MI, Kalousek DK, Chernoff GF, et al. Evidence for multi-site closure of the neural tube in humans. Am J Genet 1993; 47: 723–743.

Van Creveld S. Nouveaux aspects de la maladie hemorragique du nouveau ne. Ned Tijdschr Geneeskd 1957; 101: 2109–2112.

Van Dyke DC, Hodge SE, Helde F, et al. Family studies in fetal phenytoin exposure. J Pediatr 1988; 113: 301–306.

Vergel RG, Sanchez LR, Heredero BL, et al. Primary prevention of neural tube defects with folic acid supplementation: Cuban experience. Prenat Diagn 1990; 10: 149–152.

Vorhees CV. Developmental effects of anticonvulsants. Neurotoxicology 1986; 7: 235–244.
Wells PG, Harbison RD. Significance of the phenytoin reactive arene oxide intermediate, its oxepin tautomer, and clinical factors modifying their roles in phenytoin-induced teratology. In T. M. Hassell, M. C. Johnston, K. H. Dudley, eds. Phenytoin-Induced Teratology and Gingival

Pathology. New York: Raven Press, 1980: 83–108.
Wells PG, Kuper A, Lawson JA, et al. Relation of in vivo drug metabolism to stereoselective

fetal hydantoin toxicology in mouse: evaluation of mephenytoin and its metabolite, nirvanol.

J Pharmacol Exp Ther 1982; 221: 228–234.
Wells PG, Zubovits JT, Wong ST, et al. Modulation of phenytoin teratogenicity and embryonic

covalent binding by acetylsalicylic acid, and alpha-phenyl-N–t-butylnitrone implications for

bioactivation by prostaglandin synthetase. Toxicol Appl Pharmacol 1989; 97: 192–202.
Werler MM, Shapiro S, Mitchell AA. Periconceptual folic acid exposure and risk of occurrent

neural tube defects. J Am Med Assoc 1993; 269: 1257–1261.
Werler MM, Louik C, Shapiro S, et al. Prepregnant weight in relation to risk of neural tube

defects. J Am Med Assoc 1996; 275: 1089–1092.
Wells PG, Harbison RD. Significance of the phenytoin reactive arene oxide intermediate, its

oxepin tautomer, and clinical factors modifying their roles in phenytoin-induced teratology. In Hassell TM, Johnson MC, Dudley KH, eds. Phenytoin–induced teratology and gingival pathology. New York: Raven Press, 1980; 83–108.

Wilton LV, Shakir S. A post-marketing surveillance study of gabapentine as add on therapy for 3,100 patients in England. Epilepsia 2002; 43(9): 983–992.

Winship KA, Cahal DA, Weber JCP, et al. Maternal drug histories and central nervous system anomalies. Arch Dis Child 1984; 59: 1052–1060.

324 Mark S. Yerby

Wong M, Wells PG. Effects of N-acetylcysteine on fetal development and on phenytoin terato- genicity in mice. Teratogenesis Carcinog Mutagen 1988; 8: 65–79.

Wong M, Wells PG. Modulation of embryonic glutathione reductase and phenytoin teratogenicity by 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU). J Pharmacol Exp Ther 1989; 250: 336–342. Yerby MS, Koepsell T, Daling J. Pregnancy complications and outcomes in a cohort of women

with epilepsy. Epilepsia 1985; 26: 631–635.
Yerby MS, Leavitt A, Erickson D, et al. Antiepileptics and the development of congenital anom-

alies. Neurology 1992; 42(Suppl. 5): 132–140.
Zackai EH, Mellman WJ, Neiderer B, et al. The fetal trimethadione syndrome. J Pediatr 1975; 87:

280–284.
Zhu M, Zhou S. Reduction of the teratogenic effects of phenytoin by folic acid and a mixture of

folic acid, vitamins, and amino acids: a preliminary trial. Epilepsia 1989; 30: 246–251.

Antiepileptic drug interactions in handicapped and mentally retarded patients

Matti Sillanpää
Departments of Child Neurology and Public Health, University of Turku, Turku, Finland

Introduction

Epilepsy in the mentally retarded differs from epilepsy in the mentally normal patient in relation to etiology, seizure types, epilepsy syndromes, choice of anti- epileptic drugs, identification of their side effects and treatment outcome. Con- sequently, a successful antiepileptic drug therapy is a demanding task in terms of choice of drug therapy, combinations of drugs and side effects in mentally retarded patients compared with mentally normal people. Adverse effects and inter- actions between different antiepileptic drugs are a potential risk in the presence of many and difficult-to-treat seizure types, leading to frequent polytherapy. There is also an increased risk of interactions between antiepileptic drugs and other drugs because of the increased incidence of co-morbidity among these patients.

In patients who are handicapped or mentally retarded, there is no evidence that pharmacokinetic drug interactions per se are quantitatively or qualitatively different from those seen in otherwise normal epilepsy patients. However, it is the context of the treatment of their epilepsy that puts a different emphasis on the potential for inter- actions. These patients are characterized by an increased incidence of co-morbidity that may require treatment with other medications. Their epilepsies are generally more refractory to treatment and antiepileptic drug combinations are more likely to be used. Also, central nervous system (CNS) toxicity of drugs may be more prominent in mentally retarded patients, and this may include neurotoxic pharmacodynamic interactions between antiepileptic drugs as well as pharmacodynamic interactions between antiepileptic drugs and other psychotropic drugs. As a group, these patients may be particularly vulnerable to the problems associated with polytherapy. The main purpose of this chapter is not to provide an exhaustive discussion of possible pharma- cokinetic interactions that are discussed elsewhere in this book, but to emphasize the context in which pharmacokinetic and pharmacodynamic interactions are likely to occur during the treatment of epilepsy in handicapped and mentally retarded patients.

325

326 Matti Sillanpää

Epidemiology of epilepsy in the mentally retarded

Epilepsy occurs in approximately 15% of patients with mild mental retardation (IQ 50–69) (Blomquist et al., 1981; Drillien et al., 1966; Hagberg et al., 1981) and 30% of those with severe mental retardation (IQ 50) (Corbett, 1993; Drillien et al., 1966; Gustavson et al., 1977a, b). In institutionalized patients with mostly severe or profound mental retardation, the prevalence of epilepsy ranges from 35% to 60% (Iivanainen, 1974; Illingworth, 1959; Mariani et al., 1993). The age at the onset of the epilepsy does not differ from that in the general population (Forsgren et al., 1990; Goulden et al., 1991; Richardson et al., 1980). However, children with severe mental retardation were found to have a significantly earlier seizure onset than those with a mild mental retardation (Steffenburg et al., 1996).

Table 17.1 shows several lesional, developmental, chromosomal and metabolic conditions in which epilepsy is associated in up to 100% of the cases. The etiology of severe mental retardation is reportedly prenatal in 55–78%, perinatal in 8–15%,

Table 17.1 Occurrence of epilepsy in certain syndromes with MR

Syndrome Prevalence (%) Author(s) and year

Cerebral palsy and MR Mitochondrial disorders Polymicrogyria Tuberous sclerosis

Chromosomal anomalies

Angelman syndrome

Rett syndrome Wolf–Hirschhorn syndrome Fragile-X syndrome Prader–Willi syndrome Down syndrome
Klinefelter syndrome

Metabolic disorders

Peroxisomal diseases 80

Krabbe’s disease
Biotinidase deficiency 50–75 Disorders of urea cycle 60

28–38 Goulden et al. (1991), Sillanpää (1978) 96–100 Hirano and Pavlakis (1994)
90 Kuzniecky et al. (1993)
90 Barkovich and Kjos (1992), Fois et al. (1988)

84–90 Cassidy and Schwartz (1998), Viani et al. (1995), Zori et al. (1992)

75–80 Hagberg (1996), Perry (1991) 70 Jennings and Bird (1981)
25 Wisniewski et al. (1991) 15–20 Bray et al. (1983)

6–12 Stafstrom et al. (1991), Veall (1974)
2–10 Becker et al. (1996), Nielsen and Pedersen

(1969), Zuppinger et al. (1967)

Garcia-Alvarez et al. (1997)

50–75

MR, mental retardation.

327 AED interactions in handicapped and mentally retarded patients

Table 17.2 Failure to recognize epileptic seizures in the mentally retarded

Seizures with vertigo
Seizures with paresthesias
Seizures with visceral disturbances
Seizures with headache
Seizures with loss of emotional control
Partial seizures with other clinical manifestations Supplementary sensorimotor area seizures Simple partial seizures
Absence seizures
Drop attacks
Automatisms

postnatal in 1–12% and unknown in 13–22% (Goulden et al., 1991; Gustavson et al., 1977b; Hagberg and Kyllerman, 1983; Linna, 1989). In patients who have a mild mental handicap, the corresponding figures are 23–43%, 7–18%, 4–5% and 43–55%, respectively. In many patients, the etiology is still unknown but probably prenatal (Blomquist et al., 1981; Hagberg and Kyllerman, 1983).

Problems in diagnosing epilepsy

The diagnosis of epileptic seizures may be difficult in mentally retarded patients, because they cannot in many cases express themselves and therefore fail to tell about their perceived symptoms (Table 17.2). Also, in these patients, motor automatisms are not easily distinguished from stereotypic movements, nor are nocturnal seizures easy to separate from parasomnias. Table 17.3 lists the most important non-epileptic conditions which may lead to a misdiagnosis of epilepsy.

Intractability of seizures

The main groups of reasons for intractability of seizures are related to actions by the physician, to the patient, to the epilepsy itself and to the drugs (Table 17.4). The type of epilepsy may be a priori intractable. Epileptic and non-epileptic seizures may be intermingled in the same patient. Certain antiepileptic medications, at therapeutic or at toxic doses, may cause or aggravate seizures. Remote sympto- matic etiology, abnormal neurological status, occurrence of status epilepticus and poor short-term effect of drug therapy have been shown to be independent pre- dictors of intractability (Kwan and Brodie, 2000; Sillanpää, 1993).

328 Matti Sillanpää

Table 17.3 Differential diagnosis of non-epileptic seizures in the mentally retarded

Cardiovascular mechanisms

Infantile syncope
• Breath-holding spells

– Cyanotic infantile syncope

– Reflex anoxic seizures • Syncope in older children

Paroxysmal movement disorders

Infantile jitteriness
Benign myoclonus of early infancy Hyperekplexia
Gastroesophageal reflux
Paroxysmal dystonia/choreoathetosis Shuddering attacks
Stereotypic movements
Alternating hemiplegia of childhood Masturbation
Stool withholding activity and constipation

Psychological disorders

Psychogenic or pseudoseizures Hyperventilation Münchhausen by proxy

Migraine and migraine equivalents

Recurrent abdominal pain Basilar migraine

Sleep disorders

Arousal disorders REM sleep disorders

REM, rapid eye movement.

Drug interactions and adverse effects

Most of the untoward effects are not as readily recognized in mentally retarded as in mentally normal patients. These patients may also be at higher risk for certain adverse effects of antiepileptic therapy, such as reduced bone density (Andress et al., 2002; Tolman et al., 1975). Pharmacokinetics of antiepileptic drugs may be affected in many ways. Administration of the drugs may be complicated by the reluctance of the patient to take the pills, or decreased absorption due to slow bowel movements and constipation. Elimination of drugs metabolized by the liver may also be altered due to changes in genetic capacity, especially in inborn errors of neurometabolism

329 AED interactions in handicapped and mentally retarded patients

Table 17.4 Intractability of epilepsy in the mentally retarded

Physician related

Incorrect diagnosis Misclassification of epilepsy
Failure to recognize all seizure types Failure in choice of drug
Failure to recognize seizure freedom

Patient related

Non-compliance

Epilepsy related

Severe early infantile encephalopathies Minor motor seizures
Complex partial seizures
Atonic seizures

Multiple seizure types
Organic etiology of epilepsy Progressive etiology of seizures Non-epileptic seizures Concomitant non-epileptic seizures

Drug related

Problems in ingestion of drug
Lack of good early effect of therapy Side effects of single drug therapy Side effects of polytherapy
Drug interactions
Deviating drug kinetics

involving the liver. Epilepsy in mental retardation commonly presents with several seizure types, drug resistance, concomitant psychiatric symptoms and syndromes with various enzyme abnormalities, which increase the risk of interactions. Often, polytherapy in mentally retarded patients with epilepsy can be reduced successfully (Bennett et al., 1983). In a 10-year study in 244 institutionalized patients, the per- centage of patients receiving monotherapy could be increased from 36.5% to 58.1% with no observed loss in seizure control (Pellock and Hunt, 1996). Whenever poly- therapy is reduced, it is important to keep in mind that existing pharmacokinetic interactions are reversible upon removal of the drug responsible for the interaction.

Phenobarbital

Phenobarbital (and other barbiturates) has been used for almost one century for its good anticonvulsive efficacy. Phenobarbital (and primidone, the main active

330 Matti Sillanpää

metabolite of which is phenobarbital) is considered to typically affect cognition, behavior and affect in mentally normal people. Combination of valproate with phenobarbital therapy results in elevated phenobarbital levels, due to inhibition of phenobarbital hydroxylation, with subsequent somnolence and even coma or hyperkinesis, aggressive bursts and insomnia (Bruni et al., 1980). Inversely, pheno- barbital accelerates the metabolism of valproate, thus lowering valproate levels in relation to the dose. The metabolism of cimetidine, used against peptic ulcer, which is not so uncommon in the mentally retarded, may be induced by phenobarbital with subsequent decreased blood levels (Somogyi and Gugler, 1982). Because of its potential adverse effects, phenobarbital cannot be recommended as the first or sec- ond choice of drug for epileptic seizures associated with mental retardation.

Phenytoin

Valproate

Along with phenobarbital, phenytoin was for decades the most important tool against seizures in the mentally retarded. Phenytoin therapy is not easily managed because of its saturation kinetics, marked differences in attaining steady-state lev- els in the blood and in other features of metabolism, and certain pharmacokinetic interactions which may in some cases result in toxic levels of phenytoin. Combined with primidone, phenytoin may cause phenobarbital intoxication by causing a marked rise in the ratio of phenobarbital to primidone (Fincham and Schottelius, 1989).

The most serious groups of side effects include neurological adverse effects. Brain damage, which is commonly associated with mental retardation, and phenytoin in polytherapy further increase the risk for neurological adverse effects at therapeutic or even low levels of plasma phenytoin (Iivanainen, 1998). A chronic and in the mentally retarded often irreversible syndrome of phenytoin encephalopathy was seen in 28% (Iivanainen et al., 1977).

Phenytoin can no longer be recommended as the first or second drug of choice against epileptic seizures associated with mental retardation. This is particularly true when the patient has primary locomotion disorder or evidence of cerebellar disease.

Valproate is a major antiepileptic drug with a broad spectrum, which is an advan- tage because it can cover several types of seizure so typical of the many mentally retarded. Seizure freedom is achieved by 20–70% of children with mental retarda- tion and infantile spasms (Friis, 1998), and one-fifth of those with Lennox–Gastaut syndrome (Covanis et al., 1982; Henriksen and Johannessen, 1982) become seizure- free on a high-dosage valproate monotherapy. Valproate may have a clinically sig- nificant displacing effect on phenytoin and can cause phenytoin intoxication due to high free levels of phenytoin, even in the presence of therapeutic total levels (Wilder and

331 AED interactions in handicapped and mentally retarded patients

Rangel, 1989). Valproate can significantly elevate levels of phenobarbital (also derived from primidone), ethosuximide and lamotrigine. The risk of death from liver failure is highest in children who are less than 2 years of age, especially among those with mental retardation, genetic metabolic disorders, brain injury or a family history of severe hepatic disease and/or who are receiving valproate in polytherapy (Bryant III and Dreifuss, 1996).

Carbamazepine

Carbamazepine is effective against focal and generalized seizures. It is not effective against atypical absence, atonic and myoclonic seizures, and may even cause or increase these seizures, which are common in mentally retarded patients. Neurotoxicity is for the most part dose related. Though negative behavioral effects are in general fewer on carbamazepine than on phenytoin, phenobarbital or prim- idone, they may occur in mentally retarded patients and particularly in those with brain damage and those with pre-existing behavioral problems (Alvarez et al., 1998; Friedman et al., 1992; Reid et al., 1981).

Carbamazepine levels are lower but carbamazepine-epoxide concentrations are higher in combination therapy with phenobarbital, phenytoin, primidone and val- proate than in monotherapy. But carbamazepine and epoxide levels do not appear to be affected by newer anticonvulsants. Increasing displacement of carbamazepine from plasma proteins increases free fraction of carbamazepine during valproate co-medication (Haidukewych et al., 1989). In case of co-medication with felbamate, lamotrigine, phenobarbitone, phenytoin, primidone, progabide and valnoctamide, carbamazepine-epoxide concentrations may reach toxic levels. Carbamazepine combined with valproate appears to have synergistic effects in frontal and temporal focal seizures (Gupta and Jeavons, 1985).

Oxcarbazepine

Oxcarbazepine is similar to carbamazepine in its mode of action and efficacy against epileptic seizures. Few data are available on its efficacy in people with men- tal retardation. Given as adjunctive therapy for difficult-to-treat patients with mental retardation, a 50% or greater decrease in seizure frequency has been achieved in 50–60% of patients (Gaily et al., 1997; Sillanpää and Pihlaja, 1988/1989; Singh and Ramani, 2001). The tolerability of oxcarbazepine is better and interactions are less frequent than those observed with carbamazepine, with the exception of higher frequency of hyponatremia. Electrical status epilepticus in sleep may occur during oxcarbazepine therapy in the mentally retarded. Oxcarbazepine has not shown any significant autoinduction or interactions with other drugs (Baruzzi et al., 1994), and may therefore be a useful drug for polytherapy in the treatment of difficult- to-treat seizures.

332 Matti Sillanpää

Benzodiazepines

Benzodiazepines are in most cases used as an adjunctive therapy, for example in children with Lennox–Gastaut syndrome or other epilepsy types with mental retardation. Clinically relevant interactions of benzodiazepines are rare, if any. In some patients, however, adjunctive therapy with clonazepam may cause toxic lev- els of phenytoin (Isojärvi and Tokola, 1998). The incidence of tolerance is higher in patients with clonazepam-treated West syndrome or Lennox–Gastaut syndrome than in epilepsy with typical absence seizures (Specht et al., 1989). Interactions with other drugs are based on pharmacodynamic influences. A combination with other CNS-depressant drugs may increase depression (Haefely, 1989).

Vigabatrin

Vigabatrin proved to be an efficient drug against difficult-to-treat seizures in peo- ple with mental retardation (Pitkänen et al., 1993) and particularly in children with infantile spasms, with a 50% or greater decrease in seizure frequency in two- thirds (Chiron et al., 1991). Vigabatrin does not cause excessive behavioral distur- bances in mentally retarded patients (Pitkänen et al., 1993). Hyperactive agitation or aggression, on the other hand, have been observed in up to 15–26% of pediatric patients (Dulac et al., 1991; Uldall et al., 1991). Myoclonic jerks may be provoked by vigabatrin, necessitating discontinuation of the drug (Dean et al., 1999).

The good efficacy of vigabatrin on seizures (Kälviäinen et al., 1995) is shadowed by recent observations of visual field constriction, which occurs in one-third (Kälviäinen and Nousiainen, 2001), is caused by accumulation of vigabatrin in the retina (Sills et al., 2001), and appears irreversible (Nousiainen et al., 2001). The benefits, however, outweigh the risks and the therapy can be continued under strict clinical control (Paul et al., 2001). This is particularly true for infantile spasms due to tuberous sclerosis (Harding, 1998). Vigabatrin has not been found to be involved in any pharmacokinetic interaction.

Lamotrigine

The antiepileptic efficacy of lamotrigine is similar to that of other major antiepilep- tic drugs in placebo-controlled studies. In a retrospective evaluation of 44 institu- tionalized patients with mental retardation (Gidal et al., 2000), lamotrigine, added to other antiepileptic drug therapy, decreased seizure frequency by 50% or more in 55% of the patients with mental retardation (Beran and Gibson, 1998). Addition of lamotrigine to carbamazepine may accentuate or cause carbamazepine side effects, such as dizziness, diplopia and sedation which are subjective symptoms, and may present as behavioral disturbances in the mentally retarded (Besag et al., 1998a). The most important effect of other antiepileptic drugs is inhibition of lamotrigine metabolism by valproate and the acceleration of lamotrigine metabolism by

333 AED interactions in handicapped and mentally retarded patients

enzyme-inducing antiepileptic drugs. Methsuximide lowers lamotrigine to a clini- cally significant extent and this must be considered in the dosing of lamotrigine (Besag et al., 1998b). Several other papers have reported favorable effects on seizure frequency (Buchanan, 1995), cognition and behavior (Meador and Baker, 1997), and quality of life (Nadarajah and Duggan, 1995) and less successful involvement of behavior (Beran and Gibson, 1998; Davanzo and King, 1996).

Gabapentin

Gabapentin has been shown to be effective as an adjunct on refractory partial- onset seizures. Eleven (42%) of 26 children with mental retardation experienced a 50% or greater decrease in seizure frequency on gabapentin add-on therapy. The response did not differ from that of mentally normal study subjects (Khurana et al., 1993; Mikati et al., 1998). Gabapentin has an effect on focal seizures but not on myoclonic, atonic or absence seizures. With regard to adverse effects, 16% of 110 mentally retarded people showed aggressiveness, 15% had increase in seizure frequency, and 9% had ataxia or lethargy (Mayer et al., 1999). Mikati et al. (1998) reported behavioral adverse changes in 58% of 26 mentally retarded children. In one study, gabapentin was shown to extend the elimination half-life of felbamate by a 50% (Hussein et al., 1996). No other interactions involving gabapentin have been described.

Tiagabine

Topiramate

Tiagabine, another GABAergic antiepileptic drug, is in many respects similar to gabapentin. According to a meta-analysis (Marson et al., 1997), the chance for at least 50% reduction in seizure decrease was three-fold with add-on tiagabine than without. No separate data on mentally retarded patients are so far available. Lack of clinically relevant cognitive adverse effects may encourage tiagabine trials in mentally retarded individuals. On the other hand, dizziness, asthenia, nervousness, abnormal thinking, depression, aphasia and abnormal abdominal pain are signifi- cantly more common compared with placebo. Three patients with tiagabine- associated encephalopathy have been reported (So et al., 2001). The elimination of tiagabine is accelerated by enzyme-inducing antiepileptic drugs, but tiagabine does not seem to affect the pharmacokinetics of any other drugs.

The meta-analysis of six pooled double-blind, placebo-controlled studies on the effectiveness of topiramate on partial-onset seizures (Reife et al., 2000) showed that the seizure frequency had decreased by at least 50% in 43% of 527 patients, compared with 12% on placebo. In 98 patients with Lennox–Gastaut syndrome,

334 Matti Sillanpää

topiramate decreased the frequency of both drop attacks and generalized tonic– clonic seizures in every third patient, whereas the same effect occurred in only every eighth patient on placebo (Glauser et al., 2000). Interactions affecting other drugs are negligible due to predominantly renal excretion and low protein binding, but the half-life of topiramate is shortened by enzyme-inducing antiepileptic drugs such as carbamazepine, phenytoin, phenobarbital and primidone. Side effects during topiramate therapy include dizziness, fatigue, visual disturbance, diplopia, ataxia, psychomotor slowing, weight decrease and, in rare cases, renal stones, and hypohidrosis.

Felbamate

When felbamate was launched, it soon appeared effective in patients with, among other conditions, the Lennox–Gastaut syndrome and infantile spasms (The Felbamate Study Group, 1993). Dangerous adverse effects, mainly aplastic anemia and liver failure, have greatly restricted its use. Felbamate has also several inter- actions with other drugs. It increases significantly carbamazepine epoxide, pheno- barbital, phenytoin and valproate plasma levels and decreases total carbamazepine concentrations. Both carbamazepine and phenytoin induce felbamate metabolism and hence increase its clearance. For the moment, felbamate can only be used in well-selected patients under strict and individualized control.

Zonisamide

Few data are available on the efficacy of zonisamide in patients with mental retar- dation and refractory seizures. Iinuma et al. (1998) reported a more than 50% decrease in seizure frequency in 67% of mentally normal and in 41% of retarded patients. Adverse effects were as common in the retarded as in the mentally normal children (27% vs. 30%). The most common untoward effect was aggravation of seizures, which was more common in the mentally normal than in the retarded (28% vs. 18%), and drowsiness. No data on antiepileptic drug interactions were reported. Zonisamide does not induce or inhibit other drugs but its half-life is shortened in humans by enzyme-inducing drugs such as carbamazepine, pheno- barbital, phenytoin and valproate (Sackellares et al., 1985).

Levetiracetam

Levetiracetam, a novel broad-spectrum antiepileptic drug, is effective against focal and generalized seizures. In three multicenter, double-blind, placebo-controlled studies (Ben-Menachem and Falter, 2000; Cereghino et al., 2000; Shorvon et al., 2000), about one-third of 904 patients with partial-onset, drug-refractory seizures achieved an at least 50% overall decrease in seizure frequency. The tolerability was

335 AED interactions in handicapped and mentally retarded patients

good. According to existing data, no interactions can be anticipated in clinical use (Patsalos, 2000). The place of levetiracetam in the treatment of seizures in the handicapped and mentally retarded patient remains to be established.

Outcome of epilepsy in the mentally retarded

The outcome of drug therapy may be difficult to assess in the mentally retarded, for example in patients with infantile spasms. Video electroencephalograph (EEG) monitoring may be needed for this purpose. Most of the few population-based studies dealing with the prognosis of epilepsy in the mentally retarded show a less favorable seizure outcome (seizure freedom in 38–46%) than in mentally normal patients (65–89%) (Aicardi, 1986; Brorson and Wranne, 1987; Wakamoto et al., 2000). In another recent long-term follow-up prospective study (Sillanpää, to be published), 34% of patients with epilepsy and mental retardation and 67% of patients with uncomplicated epilepsy became seizure-free. The prognosis is better, the higher the intelligence level (Goulden et al., 1991; Rowan et al., 1980; Sillanpää, to be published). Additional predictors of poor outcome are symptomatic etiology, association of cerebral palsy and perinatal brain injury.

Conclusion

Epilepsy is a common concomitant disorder in people with mental retardation. The diagnosis of epilepsy may be more difficult, because epilepsy in the mentally retarded often presents with several seizure types. The differential diagnosis between epileptic and non-epileptic events may also at times cause difficulties. In many patients, epileptic and non-epileptic seizures may co-occur. The effects of medication are difficult to evaluate, not least due to impaired abilities of these individuals to express themselves about perceived side effects. Video EEG moni- toring may be needed. The responses to antiepileptic drugs may be different from that in mentally normal individuals. Numerous attempts in individual patients to attain seizure freedom or an acceptable level of seizure frequency mostly result in polytherapy and increasing adverse effects. These side effects may result from a dif- ferent susceptibility of the brain to the drugs, to pharmacokinetic interactions, or to a greater susceptibility to pharmacodynamic interactions. To avoid or minimize these effects, the drugs should be as few as possible and a conversion to monother- apy with a broad-spectrum drug should be preferred when feasible. This seems to be particularly important in patients with mental retardation. The impact of the newer antiepileptic drugs may consist of a better tolerability with fewer interactions.

336 Matti Sillanpää

REFERENCES

Aicardi J. Epilepsy in Children. New York: Raven Press, 1986.
Alvarez N, Besag F, Iivanainen M. Use of antiepileptic drugs in the treatment of epilepsy in

people with intellectual disability. J Intellect Disabil Res 1998; 42(Suppl. 1): 1–15.
Andress DL, Ozuna J, Tirschwell D, et al. Antiepileptic drug-induced bone loss in young male

patients who have seizures. Arch Neurol 2002; 59: 781–786.
Annegers JF, Rocca WA, Hauser WA. Causes of epilepsy: contributions of the Rochester

epidemiology project. Mayo Clin Proc 1996; 71: 570–575.
Barkovich AJ, Kjos BO. Gray matter heterotopias: MR characteristics and correlation with devel-

opmental and neurologic manifestations. Radiology 1992; 182: 493–499.
Baruzzi A, Albani F, Riva R. Oxcarbazepine: pharmacokinetic interactions and their clinical sig-

nificance. Epilepsia 1994; 35(Suppl. 3): 14–19.
Becker KL, Hoffman DL, Albert A, et al. Klinefelter’s syndrome: clinical and laboratory findings

in 50 patients. Arch Int Med 1996; 118: 314–321.
Ben-Menachem E, Falter U. Efficacy and tolerability of levetiracetam 3000 mg in patients with

refractory partial seizures: a multicenter, double-blind, responder-selected study evaluating

monotherapy. Epilepsia 2000; 41: 1276–1283.
Bennett HS, Dunlop T, Ziring P. Reduction of polypharmacy for epilepsy in an institution for the

retarded. Dev Med Child Neurol 1983; 25: 735–737.
Beran RG, Gibson RJ. Aggressive behaviour in intellectually challenged patients with epilepsy

treated with lamotrigine. Epilepsia 1998; 39: 280–282.
Besag FMC, Berry DJ, Pool F, et al. Carbamazepine toxicity with lamotrigine: pharmacokinetic

or pharmacodynamic interaction? Epilepsia 1998a; 39: 183–187.
Besag FMC, Berry DJ, Pool F. Methsuximide lowers LTG blood levels: a pharmacokinetic AED

interaction (abstract). Epilepsia 1998b; 39(Suppl. 2): 25–26.
Blomquist HK, Gustavson KH, Holmgren G. Mild mental retardation in children in a Northern

Swedish county. J Ment Defic Res 1981; 25: 92–109.
Borusiak P, Korn-Merker E, Holert N, et al. Hyponatremia induced by oxcarbazepine in children.

Epilepsy Res 1998; 30: 241–246.
Bray GA, Dahms WT, Swerdloff RS, et al. The Prader–Willi syndrome: a study of 40 patients and

a review of the literature. Medicine (Baltimore) 1983; 62: 59–80.
Brorson LO, Wranne L. Long-term prognosis of childhood epilepsy: survival and seizure prog-

nosis. Epilepsia 1987; 28: 324–330.
Bruni J, Wilder BJ, Perchalski RJ, et al. Valproic acid and plasma levels of phenobarbital.

Neurology (Minneap) 1980; 30: 94–97.
Bryant III AE, Dreifuss FE. Valproic acid hepatic fatalities. III. U.S. experience since 1986.

Neurology 1996; 46: 465–469.
Buchanan N. The efficacy of lamotrigine on seizure control in 34 children, adolescents and

young adults with intellectual and physical disability. Seizure 1995; 4: 233–236.
Cassidy SB, Schwartz S. Reviews in Molecular Medicine: Prader–Willi and Angelman Syndromes, Disorders of Genomic Imprinting. Baltimore: Williams & Wilkins, Medicine, 1998:

140–151.

337 AED interactions in handicapped and mentally retarded patients

Cereghino JJ, Briton V, Abou-Khalil B, et al. Levetiracetam for partial seizures: results of a double-blind, randomized clinical trial. Neurology 2000; 55: 236–242.

Chiron C, Dulac O, Beaumont D, et al. Therapeutic trial of vigabatrin refractory infantile spasms. J Child Neurol 1991; 6: 52–59.

Commission on Classification and Terminology of the ILAE. Proposal for revised clinical and electroencephalographic classification of epileptic seizures. Epilepsia 1981; 22: 489–501.

Corbett J. Epilepsy and mental handicap. In A Textbook of Epilepsy. J. Laidlaw, A. Richens, D. Chadwick, eds. Edinburgh: Churchill Livingstone, 1993: 631–636.

Covanis A, Gupta AK, Jeavons PM. Sodium valproate: monotherapy and polytherapy. Epilepsia 1982; 23: 693–720.

Davanzo PA, King BH. Open trial lamotrigine in the treatment of self-injurious behaviour in an adolescent with profound mental retardation. J Child Adolesc Psychopharmacol 1996; 6: 273–279. Dean C, Mosier M, Penry K. Dose–response study of vigabatrin as add-on therapy in patients

with uncontrolled complex partial seizures. Epilepsia 1999; 40: 74–82.
Drillien CM, Jameson S, Wilkinson EM. Studies in mental handicap. Part I: Prevalence and dis-

tribution by clinical type and severity of defect. Arch Dis Childh 1996; 41: 528–538.
Dulac O, Chiron C, Luna D, et al. Vigabatrin in childhood epilepsy. J Child Neurol 1991;

6(Suppl. 2): 30–37.
The Felbamate Study Group in Lennox–Gastaut Syndrome. Efficacy of felbamate in childhood

epileptic encephalopathy (Lennox–Gastaut syndrome). New Engl J Med 1993; 328: 29–33. Fincham RW, Schottelius DD. Primidone: interactions with other drugs. In Antiepileptic drugs, 3rd edn. R. Levy, R. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. New York: Raven

Press, 1989: 413–422.
Fois A, Tomaccini D, Balestri P, et al. Intractable epilepsy: etiology, risk factors and treatment.

Clin EEG 1988; 19: 68–73.
Forsgren L, Edvinsson SO, Blomquist HK, et al. Epilepsy in a population of mentally retarded

children and adults. Epilepsy Res 1990; 6: 234–248.
Friedman LD, Kastner T, Plummer AT, et al. Adverse behavioral effects in individuals with men-

tal retardation and mood disorders treated with carbamazepine. Am J Mental Defic 1992; 96:

541–546.
Friis ML. Valproate in the treatment of epilepsy in people with intellectual disability. J Intellect

Disabil Res 1998; 42(Suppl. 1): 32–35.
Gaily E, Granström ML, Liukkonen E. Oxcarbazepine in the treatment of early childhood

epilepsy. J Child Neurol 1997; 12: 496–498.
Garcia-Alvarez M, Nordli DR, De Vivo DC. Inherited metabolic diseases. In Epilepsy: A

Comprehensive Textbook. E. Engel, T. A. Pedley, eds. Philadelphia: Lippincott-Raven Press,

1997: 2547–2562.
Gidal BE, Walker JK, Lott RS, et al. Efficacy of lamotrigine in institutionalized, developmentally

disabled patients with epilepsy: a retrospective evaluation. Seizure 2000; 9: 131–136.
Glauser TA, Levisohn PM, Ritter F, et al. Topiramate in Lennox–Gastaut syndrome: open-label treat-

ment of patients completing a randomized controlled trial. Epilepsia 2000; 41(Suppl. 1): 86–90. Goulden KJ, Shinnar S, Koller H, et al. Epilepsy in children with mental retardation: a cohort

study. Epilepsia 1991; 32: 690–697.

338 Matti Sillanpää

Gupta AK, Jeavons PM. Complex partial seizures: EEG foci and response to carbamazepine and sodium valproate. J Neurol Neurosurg Psychiatr 1985; 48: 1010–1014.

Gustavson KH, Hagberg B, Hagberg K, et al. Severe mental retardation in a Swedish county. I. Epidemiology, gestational age, birth weight and associated CNS handicaps in children born 1959–70. Acta Paediatr Scand 1977a; 66: 373–379.

Gustavson KH, Holmgren G, Jonsell R, et al. Severe mental retardation in children in a Northern Swedish county. J Ment Defic Res 1977b; 21: 161–180.

Haefely W. Benzodiazepines, mechanisms of action. In Antiepileptic drugs. R. H. Levy, R. H. Mattson, B. Meldrum, J. K. Penry, F. E. Dreifuss, eds. 3rd edn. New York: Raven Press, 1989; 721–734.

Hagberg B. Rett syndrome: clinical peculiarities and biological mysteries. Acta Paediatr 1996; 84: 971–976.

Hagberg B, Kyllerman M. Epidemiology of mental retardation – a Swedish survey. Brain Devel 1983; 5: 441–449.

Hagberg B, Hagberg G, Lewerth A, et al. Mild mental retardation in Swedish school children. II. Etiologic and pathogenetic aspects. Acta Paediatr Scand 1981; 70: 445–452.

Haidukewych D, Zielinski JJ, Rodin EA. Derivation and evaluation of an equation for prediction of free carbamazepine concentration in patients comedicated with valproic acid. Ther Drug Monit 1989; 11: 528–532.

Harding GF. Severe persistent visual field constriction associated with vigabatrin. Benefit risk ratio must be calculated from individual patients. Br Med J 1998; 316: 232–233.

Henriksen O, Johannessen SI. Clinical and pharmacokinetic observations on sodium valproate: a five-year follow-up study of 100 children with epilepsy. Acta Neurol Scand 1982; 65: 504–523. Hirano M, Pavlakis SG. Mitochondrial myopathy, encephalopathy, lactic acidosis, and strokelike

episodes (MELAS): current concepts. J Child Neurol 1994; 9: 4–13.
Hussein G, Troupin AS, Montouris G. Gabapentin interaction with felbamate. Neurology 1996;

47: 1106.
Iinuma K, Minami T, Cho K, et al. Long-term effects of zonisamide in the treatment of epilepsy

in children with intellectual disability. J Intellect Disabil Res 1998; 42(Suppl. 1): 68–73. Iivanainen M. A Study of Origins of Mental Retardation. Clinics in Developmental Medicine No.

51. Spastics International Publications. London: William Heinemann Medical Books, 1974. Iivanainen M. Phenytoin: effective but insidious therapy for epilepsy in people with intellectual

disability. J Intellect Disabil Res 1998; 42(Suppl. 1): 24–31.
Iivanainen M, Viukari M, Helle EP. Cerebellar atrophy in phenytoin-treated mentally retarded

epileptics. Epilepsia 1977; 18: 375–386.
Illingworth RS. Convulsions in mentally retarded children with or without cerebral palsy. J Ment

Defic Res 1959; 3: 88–93.
Isojärvi JIT, Tokola RA. Benzodiazepines in the treatment of epilepsy in people with intellectual

disability. J Intellect Disabil Res 1998; 42(Suppl. 1): 80–92.
Jennings MT, Bird TD. Genetic influences in the epilepsies. Am J Dis Child 1981; 135: 450–457. Kälviäinen R, Nousiainen I. Visual field defects with vigabatrin: epidemiology and therapeutic

implications. CNS Drug 2001; 15: 217–230.
Kälviäinen R, Äikiä M, Saukkonen AM, et al. Vigabatrin vs. carbamazepine monotherapy in

patients with newly diagnosed epilepsy. A randomized, controlled study. Arch Neurol 1995; 52: 989–996.

339 AED interactions in handicapped and mentally retarded patients

Khurana DS, Riviello JJ, Helmess S, et al. Efficacy of gabapentin therapy in children with refrac- tory partial seizures. J Pediatr 1993; 128: 829–833.

Kuzniecky R, Andermann F, Guerrini R. Congenital bilateral perisylvian syndrome: study of 31 cases. Lancet 1993; 341: 608–612.

Kwan P, Brodie MJ. Early identification of refractory epilepsy. New Engl J Med 2000; 342: 314–319. Linna SL. Prevalence, Aetiology, Associated Handicaps and Self Care Ability in 5–19-Old Severely

Mentally Retarded. A doctoral thesis. Oulu: University of Oulu Printing Center, 1989. Mariani E, Ferini-Strambi L, Sala M, et al. Epilepsy in institutionalized patients with encephalopathy: clinical aspects and nosological considerations. Am J Ment Retard 1993;

98(Suppl.): 27–33.
Marson AG, Kadir ZA, Hutton JL, et al. The new antiepileptic drugs: a systematic review of their

efficacy and tolerability. Epilepsia 1997; 38: 859–880.
Mayer T, Schutte W, Wolf P, et al. Gabapentin add-on treatment: how many patients become

seizure-free? An open-label multicenter study. Acta Neurol Scand 1999; 99: 1–7.
Meador KJ, Baker GA. Behavioral and cognitive effects of lamotrigine. J Child Neurol 1997;

12(Suppl.): 44–47.
Mikati MA, Choueri R, Khurana DS, et al. Gabapentin in the treatment of refractory partial

epilepsy in children with intellectual disability. J Intellect Disabil Res 1998; 42(Suppl. 1): 57–62. Nadarajah J, Duggan L. A missed diagnosis. A missed opportunity for community integration.

Seizure 1995; 4: 151–153.
Nielsen J, Pedersen E. Electro-encephalographic findings in patients with Klinefelter’s syndrome

and the XXY-syndrome. Acta Neurol Scand 1969; 45: 87–94.
Nousiainen I, Mäntyjärvi M, Kälviäinen R. No reversion in vigabatrin-associated visual field

defects. Neurology 2001; 57: 1916–1917.
Patsalos PN. Pharmacokinetic profile of levetiracetam: toward ideal characteristics. Pharmacol

Therapeut 2000; 85: 77–85.
Paul SR, Krauss GL, Miller NR, et al. Visual function is stable in patients who continue long-term

vigabatrin therapy: implications for clinical decision making. Epilepsia 2001; 42: 525–530. Pellock JM, Hunt PA. A decade of modern epilepsy therapy in institutionalized mentally

retarded patients. Epilepsy Res 1996; 25: 263–268.
Perry A. Rett syndrome: a comprehensive review of the literature. Am J Ment Retard 1991; 3: 275–290. Pitkänen A, Ylinen A, Matilainen R, et al. Long-term antiepileptic efficacy of vigabatrin in drug-

refractory epilepsy in mentally retarded patients. A 5-year follow-up study. Arch Neurol 1993;

50: 24–29.
Reid AH, Naylor GJ, Kay DS. A double blind placebo controlled crossover trial of carbamazepine

in overactive severely mentally handicapped patients. Psychol Med 1981; 11: 109–113.
Reife R, Pledger G, Wu SC. Topiramate as add-on therapy: pooled analysis of randomized con-

trolled trials in adults. Epilepsia 2000; 41(Suppl. 1): 66–71.
Richardson SA, Koller H, Katz M, et al. Seizures and epilepsy in a mentally retarded population

over the first 22 years of life. Appl Res Ment Retard 1980; 1: 123–138.
Rowan AJ, Overweg J, Sadikoglu S, et al. Seizure prognosis in long-stay mentally subnormal

epileptic patients: interrater EEG and clinical studies. Epilepsia 1980; 21: 219–225.
Sackellares JC, Donofrio PD, Wagner JG, et al. Pilot study of zonisamide (1,2-benzisoxazole- 3-methanesulfonamide) in patients with refractory partial seizures. Epilepsia 1985; 26: 206–211.

340 Matti Sillanpää

Shorvon SD, Lowenthal A, Janz D, et al. Multicenter double-blind, randomized, placebo- controlled trial of levetiracetam as add-on therapy in patients with refractory partial seizures. European Levetiracetam Study Group. Epilepsia 2000; 41: 1179–1186.

Sillanpää M. CP-vammaisen ennuste sosiaalilääketieteelliseltä kannalta (Prognosis of cerebral palsies with special reference to socio-medical aspects). Kansanterveystiet Julk M33. Turku: Gill OY (in Finnish, with English summary), 1978.

Sillanpää M. Remission of seizures and predictors of intractability in long-term follow-up. Epilepsia 1993; 34: 930–936.

Sillanpää M, Pihlaja T. Oxcarbazepine (GP 47680) in the treatment of intractable seizures. Acta Paediatr Hung (1988/1989); 29: 359–364.

Sills GJ, Patsalos PN, Butler E, et al. Visual field constriction: accumulation of vigabatrin but not tiagabine in the retina. Neurology 2001; 57: 196–200.

Singh BK, Ramani V. Oxcarbazepine in adults with developmental disabilities. Epilepsia 2001; 42(Suppl. 7): 186–187.

So EL, Fessler AJ, Cascino GD, et al. Tiagabine-associated encephalopathy. Epilepsia 2001; 42(Suppl. 7): 261.

Somogyi A, Gugler R. Drug interaction with cimetidine. Clin Pharmacokinet 1982; 7: 23–41. Specht U, Boenigk HE, Wolf P. Discontinuation of clonazepam after long-term treatment.

Epilepsia 1989; 30: 458–463.
Stafstrom CE, Patxot CE, Gilmore HE, et al. Seizures in children with Down syndrome: etiology,

characteristics and outcome. Dev Med Child Neurol 1991; 33: 191–200.
Steffenburg U, Hagberg G, Kyllerman M. Characteristics of seizures in a population-based series

of mentally retarded children with active epilepsy. Epilepsia 1996; 37: 850–856.
Tolman KG, Jubiz W, Sannella JJ, et al. Osteomalacia associated with anticonvulsant drug ther-

apy in mentally retarded children. Pediatrics 1975; 56: 45–51.
Uldall P, Alving J, Gram L, et al. Vigabatrin in pediatric epilepsy – an open study. J Child Neurol

1991; 6(Suppl. 2): 38–44.
Veall RM. The prevalence of epilepsy among mongols related to age. J Ment Def Res 1974; 18:

99–106.
Viani F, Romeo A, Viri M, et al. Seizure and EEG patterns in Angelman’s syndrome. J Child

Neurol 1995; 10: 467–471.
Wakamoto H, Nagao H, Hayashi M, et al. Long-term medical, educational` social prognoses of

childhood-onset epilepsy: a population-based study in a rural district of Japan. Brain Dev

2000; 22: 246–255.
Wilder BJ, Rangel RJ. Clinically relevant antiepileptic drug interactions. In Antiepileptic Drug

Interactions. W. H. Pitlick, ed. New York: Demos Publications, 1989: 65–75.
Wisniewski KE, Segan SM, Miezejesji CM, et al. The Fra(X) syndrome: neurological, electro-

physiological and neuropathological abnormalities. Am J Med Genet 1991; 38: 476–480.
Zori RT, Henrickson J, Woolven S, et al. Angelman syndrome: clinical profile. J Child Neurol

1992; 7: 270–280.
Zuppinger K, Engel E, Forber AP, et al. Klinefelter’s syndrome, a clinical and cytogenetic study in

24 cases. Acta Endocrinol 1967; 34(Suppl. 113): 5–48.

Antiepileptic drugs and sex steroids

Richard H. Mattson
Department of Neurology, Yale University School of Medicine, New Haven, CT, USA

Background

In 1972 Kenyon sent a letter to the British Medical Journal describing a patient with epilepsy treated with phenytoin (PHT) who became pregnant despite taking usual amounts of oral contraceptive (OC) pills (Kenyon, 1972). She astutely attributed the contraceptive failure to an inductive effect of the PHT on the metabolism of the sex steroid hormones. This observation was soon confirmed by others (Coulam and Annegers, 1979; Janz and Schmidt, 1974) and the underlying mech- anisms were further elucidated (Back, 1980). All the older antiepileptic drugs (AEDs), carbamazepine (CBZ), phenobarbital (PB), PHT and primidone (PRM) except valproate (VPA) (Crawford 1986; Sonnen, 1983) were found to have similar effects (Mattson et al., 1986; Schmidt, 1981). In contrast most of the new AEDs with the exception of felbamate (FBM), oxcarbazepine (OXC) and topiramate (TPM) do not change the metabolism of the OCs. Parenteral formulations (intra- muscular (i.m.) depot, subcutaneous implant and dermal patch) of contraceptive female sex hormones have also been reported to be subject to increased clearance.

The effect of the AEDs on testosterone metabolism has also indicated changes occur although the evidence of clinical effects is less easily assessed than an unplanned pregnancy. Conversely, except for lamotrigine (LTG) the OCs do not appear to change the pharmacokinetics of AEDs.

Frequency and importance of interactions

Following the initial case report by Kenyon, three other cases of OC failure were cited by Janz and Schmidt (1974). By 1983 Sonnen found that 52 cases had been reported. He concluded that the incidence was probably much higher because 12 women in his own small population had experienced an unplanned pregnancy while using OC pills when taking AEDs (Sonnen, 1983). Although the effect of

341

342 Richard H. Mattson

CBZ, PB and PHT became established, the exact risk of an unplanned likelihood could only be estimated. The probability approximates that of condom use, a five- to ten-fold increase relative to use of OCs in women not receiving e

deemagclinic

Antiepileptic Drugs Combination Therapy and Interactions