- Email: [email protected]
REVIEW Antiepileptic drugs: safety in numbers?
Seizure 2000; 9: 170–178
doi: 10.1053/seiz.1999.0358, available online at http://www.idealibrary.com on
REVIEW †
Antiepileptic drugs: safety in numbers? JOHN PAUL LEACH
Specialist Registrar in Neurology and Neurophysiology, Walton Centre for Neurology and Neurosurgery, Lower Lane, Fazakerly, Liverpool L9 7LJ, UK
INTRODUCTION Epilepsy is a common neurological condition, with a lifetime prevalence of between 2%1, 2 and 5%3 of the general population, and a point prevalence of between 4 and 80 per 10004 . So common is it among all races that epilepsy may be viewed, not as a rare curse on the individual, but as an all-too frequent price that man and other vertebrates pay for maintaining a plastic and adaptable central nervous system. Within the first few years of the next century, some estimates suggest that there will have been 11 new antiepileptic drugs (AEDs) licensed for use as either add-on or monotherapy since the late 1980s5 . Despite these recent welcome additions to the ranks of available anticonvulsants6 epilepsy remains both stigmatizing and debilitating for many of those with the disorder7 . Are there grounds for optimism about the possibility of treating refractory epilepsy? Previous studies have shown that around one-third of patients will remain refractory to anticonvulsant monotherapy with the established drugs7, 8 . Given this, the new drugs have, with justification, generated a great deal of interest, but despite their use most clinicians would admit that a significant number of patients with epilepsy have seizures that remain unsatisfactorily controlled9 . The challenge for epileptologists over the coming decades is to further reduce the size of this population with refractory disease. The new generation of AEDs are undoubtedly an advance in the treatment of epilepsy (given their advantages of increased tolerability6, 10 and a decreased tendency to cause pharmacokinetic interactions11 ), but it could be argued that a continued search for more new anticonvulsants is an inefficient way of helping our patients. Perhaps it would be more fruitful to look for †
better ways of using the drugs already at our disposal. For many different conditions in many different specialties, the use of drug combinations is the rule rather than the exception. Admittedly, anticonvulsants are potentially neurotoxic8 , can result in severe systemic reactions4 , and have the tendency to sustain multiple pharmacodynamic and pharmacokinetic interactions12 . The same, however, is true of antiparkinsonian drugs13 , and monotherapy would never be advocated as the ideal management strategy for the treatment of Parkinson’s disease13 . In fact here the opposite philosophy applies, where a rational combination of compounds is used at an early stage in an attempt to delay symptomatic progression13 . Since the 1970s, in contrast, anticonvulsant monotherapy has long been regarded as the ideal management strategy. A series of studies by Shorvon and Reynolds14–16 argued against AED polypharmacy on the basis of its strong association with adverse events, they felt, with little evidence of an increase in efficacy. While it seems axiomatic that newly diagnosed epilepsy should be treated with a single drug, the didacticism in favour of monotherapy fails to recognize the realities of everyday practice. One survey has shown17 that even when under review by specialist epilepsy services, almost half the patients required (or, at least, were receiving) long-term treatment with anticonvulsant polypharmacy. At present, even taking into account those compounds currently undergoing phase three testing, it seems unlikely that any of the new AEDs will, by themselves, prove to be the ‘magic bullet’ for epilepsy. In a medical climate so keen on evidence guiding our medical management, it therefore falls to us as epilepsy specialists to formulate a rational strategy for its treatment. Sadly, though, it seems pre-
This was one of the winning entries in the young physician section of the Millenium Gower Prize Competition 1999. 1059–1311/00/030170 + 09
$35.00/0
c 2000 BEA Trading Ltd
Antiepileptic drugs: safety in numbers?
mature to be postulating which AED polypharmacy is best when we still have not fully addressed the issue of whether polypharmacy is superior to monotherapy in the treatment of refractory epilepsy.
THE CASE FOR ANTICONVULSANT MONOTHERAPY The success of AED monotherapy will be enhanced when the drugs are used most appropriately: to state the obvious, each patient should be started on the right drug for their epilepsy at the right dosage. A proper clinical history, with relevant investigations should ensure that patients with a primary generalized epilepsy18 (e.g. juvenile myoclonic epilepsy18 or absence seizures18 ) or benign focal epilepsies of childhood19 receive appropriate initial treatment. Importantly, for example, absence and myoclonic seizures can be exacerbated by carbamazepine20, 21 or vigabatrin20, 21 , so in these circumstances syndromic classification of epilepsy becomes most important, indicating the drugs most likely to protect against all potential seizure types in addition to those which have already revealed themselves. Careful assessment by an experienced epileptologist may allow such a syndromic diagnosis to be made, which can influence the choice of drug, or even indicate that anticonvulsant monotherapy is less likely to be completely successful22 . Phenobarbitone, phenytoin, carbamazepine and valproic acid have shown similar efficacy when used as monotherapy8, 23 in patients with newly diagnosed epilepsy. Some of the newer drugs have been compared with their older counterparts under trial conditions, but many of the key comparisons are still awaited. How good are the new AEDs when they are compared directly with their older counterparts? Brodie et al.10 suggested that lamotrigine was as efficacious as carbamazepine, while being significantly better tolerated. Two further studies comparing lamotrigine with phenytoin24 and carbamazepine25 , however, failed to differentiate them in terms of either efficacy or effectiveness. There remains a need for comparative monotherapy studies of valproate versus lamotrigine. Topiramate has also been assessed as monotherapy26 , but only in comparison with an ‘active placebo’ (a subtherapeutic dose of topiramate). In a study by Kalvainen27 , monotherapy with vigabatrin was better tolerated than with carbamazepine, while lacking its efficacy. Based on the available evidence, the new AEDs are no more efficacious than their older counterparts, so how can our chances of therapeutic success be maximized? Given the incessant calls for ‘evidence’ to guide future practice, we should bear in mind that much of
171
our ‘knowledge’ of AED monotherapy is taken for granted. The superior efficacy of sodium valproate and ethosuximide over phenytoin and carbamazepine in the absence of seizures seems too obvious to demand confirmation in a prospective clinical study, but there are no data differentiating the efficacies of sodium valproate and ethosuximide in the absence of epilepsy28 . Similarly, while no one would argue that for most patients with juvenile myoclonic epilepsy, sodium valproate is the drug of choice29 , no major differences in efficacy between carbamazepine, phenytoin, barbiturates, and sodium valproate in the treatment of partial epilepsies have been fully established8, 23, 30 , in the face of conventional practice in many epilepsy clinics. The limited role for barbiturate drugs is supported in the study by Mattson et al.8 , where they were shown to be less well tolerated than phenytoin and carbamazepine. How should we deal with cases where monotherapy fails to give complete seizure control? Firstly the wide interindividual variation in optimal doses of anticonvulsants31 should be borne in mind: one study32 has shown that increasing the dosage of current therapy to maximally tolerated levels can bring seizure control in almost one-third of patients considered to have seizures ‘refractory’ to monotherapy with phenytoin or primidone. Adverse events may be minimized during dosage escalation by avoiding over-rapid dosage titration or by using controlled-release formulations, particularly with carbamazepine33, 34 and less convincingly with valproic acid35 . Even when the first monotherapy fails, it need not necessarily follow that polypharmacy is an inevitable requirement. Hakkareinen36 investigated the success rate of monotherapy with either carbamazepine or phenytoin when the other had failed, and found that one-third of the treatment failures were successfully controlled by monotherapy with the alternative drug. Similarly, Mattson et al.8 found that 46% of those unsuccessfully treated with an initial monotherapy responded to an alternative monotherapy. In this study, however, some patients were originally treated with phenobarbital or primidone, neither of which would he today accept as reasonable first-line monotherapy. Anticonvulsant monotherapy will be adequate for most patients with newly diagnosed epilepsy, and it is of paramount importance that the chances of monotherapy succeeding are maximized. Even if the first drug fails at maximally tolerated dosages, however, a second monotherapy may well prove to be effective in a substantial minority of patients. In summary, however, although AED monotherapy is good, it is still not good enough.
172
THE CASE FOR AED POLYPHARMACY Notwithstanding the poor image of polypharmacy, monotherapy will fail in a significant proportion of patients, necessitating a trial of polypharmacy with anticonvulsants. Despite the observations of Hakkareinen36 and Mattson8 , we have no randomized trials comparing alternative monotherapy with add-on therapy in groups of patients with poorly controlled epilepsy. In the study by Mattson et al.8 , 39% of monotherapy ‘non-responders’ were helped by the addition of a second drug, with 17% becoming seizurefree. A trial of adjuvant valproic acid in cases where carbamazepine had failed as monotherapy, demonstrated ‘improvement’ in a similar number of patients while on combination therapy, with 17% becoming seizure-free37 . Certainly there is an abundance of trials to prove the superiority of the new antiepileptic drugs over placebo in refractory epilepsy38 , although there would be some grounds for doubting whether all of the trial subjects were previously optimally treated! Despite its efficacy, polypharmacy has long been thought to be fraught with disadvantages: it is held to be responsible for an increased frequency of adverse events17 , pharmacokinetic interactions12 , and teratogenesis39 , while conferring only limited improvements in seizure control16 . In one study40 , monotherapy with phenytoin or carbamazepine produced adverse effects in 28 and 38% of patients respectively, while their combination produced adverse events in around three-quarters of all patients. However, Lammers et al.41 , following a retrospective review, were of the opinion that it was dosage rather than the number of anticonvulsants used that determined the frequency and severity of adverse events. As an ‘irreversible’ adverse event, teratogenicity must bear specific examination. Even as monotherapy, AEDs42, 43 seem to increase the risk of major fetal malformation by a factor of around five43 : the most common AED-related malformations are harelip, cleft palate and cardiovascular anomalies. Such data should be enough to ensure that AED therapy is only begun after careful consideration. Estimates suggest the malformation risk to be 1–2% of pregnancies on AEDs28 . Early screening with fetal ultrasound, serum alpha-fetoprotein and, where indicated, amniocentesis should be offered to women becoming pregnant while on AEDs. There are good reasons why polypharmacy with older anticonvulsants should not be universally successful. Each has a wide range of actions44 , while their anticonvulsant efficacy relies, at least in part, on sodium channel blockade44 . Random combination of these widely active drugs might reasonably be thought to merely increase the degree of sodium channel blockade, while producing a host of more compli-
J. P. Leach
cated biochemical and physiological effects. Mixing drugs with such ‘indiscriminate’ actions as the established anticonvulsants, will almost inevitably lead to problems. The human brain is a delicate instrument, and tampering with a range of neuronal properties may have far-reaching consequences. In fact, considering their potential for pharmacokinetic and pharmacodynamic interactions, it is almost surprising that anyone can tolerate polypharmacy with the older anticonvulsants! It is fair to say that, with the advent of a new generation of anticonvulsants, polypharmacy appears to be more promising. The newer AEDs have more predictable pharmacokinetics than the older agents45 and the incidence and severity of pharmacokinetic interactions with them is markedly reduced6, 45 , which makes their combination less hazardous and complicated11 . In addition, with the probable exceptions of topiramate and gabapentin, the new anticonvulsants are reputed to have a single, well-defined mode of action46 . While there is something appealing in the concept of a pharmacological agent acting on a single process to prevent seizures, some scepticism about the assurances of specificity are warranted, given that after marketing, vigabatrin47, 48 and lamotrigine49 were shown to have additional actions that may contribute to their anticonvulsant effect. Further work on other models has certainly hinted at other modes of action46 , and it should not surprise us if the number of actions of each drug increases with each different model tested. Although careful investigation of the effect of these new anticonvulsants might complicate our thinking about these compounds, it may in return open up further therapeutic possibilities in the treatment of epilepsy. Despite these negative aspects of combinations of the established AEDs, it is still rather puzzling that the inevitability of AED polypharmacy has not inspired any large-scale clinical trials to ascertain which combinations are most effective. In a series of studies in mice, Bourgeois and co-workers50–54 examined the anticonvulsant and neurotoxic results of combining the established anticonvulsants. Anticonvulsant concentrations were standardized to avoid the confounding of the results by the expected pharmacokinetic interactions. Using these methods, it was suggested that combinations of phenytoin and valproic acid had an anticonvulsant effect that was supra-additive (i.e. displayed synergism between the two drugs) while the neurotoxic effects were merely additive45 . Further animal studies suggested that the anticonvulsant benefits of combining valproic acid with carbamazepine51 or ethosuximide52 outweighed the neurotoxic effects, with similar positive effects from a phenytoin/phenobarbital combination54 . In contrast, a valproic acid/phenobarbital51 and phenobar-
Antiepileptic drugs: safety in numbers?
bital/carbamazepine54 combination were more disappointing, with both the neurotoxic and anticonvulsant effects being additive. Of course, none of our patients are mice, so where does this leave us? It may be considered surprising that none of these potentially beneficial results were followed up by proper clinical trials, although by the late 1980s, the newer anticonvulsants had begun to take centre stage as the source of new hope for those with intractable epilepsy. For two decades there has been ongoing debate about the relative merits of polypharmacy and monotherapy. The fact that available monotherapies are not perfect has left clinicians reliant on polypharmacy, and so doctors are left to choose between the theoretically desirable and the practically unavoidable. Given this situation, how can we best plan combination anticonvulsant drug therapy? This question can only be addressed once several smaller problems have been addressed.
How can we develop a rational plan for optimal anticonvulsant polypharmacy?
173
(b) Meta-analysis
Meta-analyses comparing the newer drugs should be interpreted with caution, as each has mostly succeeded in comparing toxic doses of one compound with subtherapeutic doses of another. Other as yet untried meta-analyses, however, are potentially far more productive: prospective anticonvulsants have to undergo clinical testing as add-on therapy, and one efficient method of assessing drug combinations would be to measure the relative benefits of each new treatment in tandem with each established anticonvulsant. At present, plans are underway to carry out such studies [A. Marson, personal communication]. Comparison across trials should be interpreted with great caution, but such analyses may help to highlight any particularly potent combinations. The pharmaceutical industry needs some encouragement to participate, since each company is understandably keen for their product to be seen as a ‘cutting-edge’ compound rather than as a cofactor. Such meta-analyses, however, are of great potential benefit to the industry, for example in rapidly highlighting any potential important adverse pharmacokinetic or pharmacodynamic interactions. (c) Systematic clinical trials involving each drug combination
(a) Chance observation
The history of epilepsy treatments is littered with chance observations, from Hauptmann’s discovery of the anticonvulsant effects of phenobarbital55 to the discovery of the anticonvulsant properties of valproic acid56 and lamotrigine6 . Serendipity is wellrecognized as a ‘method’ of formulating therapeutic advances, being the pay-off for astute clinical observation. Anecdotal reports have suggested specific benefit with particular combinations (e.g. vigabatrin/lamotrigine57 , lamotrigine/valproic acid58 ), but none has been assessed in proper clinical trials. One important point to note is that if clinical observation is to be relied upon to highlight ideal drug combinations, then such opinions should be based on better information. For example, it would be helpful if the data upon which drug licensing applications are based were made available at an early stage, preferably in peer-reviewed journals. It has become apparent that there is a paucity of freely available clinical and preclinical data, even for those drugs on general release. It is less than ideal that data regarding marketed compounds have to be sifted from journal abstracts, and it is probably unhealthy for our understanding of such drugs to be based solely on the subjective (and occasionally selective) data provided by the pharmaceutical industry.
It would be ideal if every anticonvulsant combination could be tested in humans using rigorously designed double-dummy, placebo-controlled studies59 . However, assuming that there are eight available anticonvulsant compounds (a conservative estimate for most countries), this allows 28 combinations of 2-drug and 56 of 3-drug cotreatments60 . Having four treatment arms (two active monotherapies, one placebo, one combination), such trials would require very large numbers of patients, one-quarter of whom would be receiving placebo. In countries outside the USA at least, there is some unease about giving placebo as monotherapy for epilepsy, so in any future trials, common sense may render the placebo arm unnecessary. It is (hopefully) widely recognized that the new drugs are active anticonvulsants, and it could be argued with equal conviction that placebo is not an adequate add-on treatment for refractory epilepsy. The aim of any large trial should not be to ‘reinvent the wheel’ and prove that a combination is better than placebo, but to prove that a particular combination is more beneficial than any other. In an ideal world, with unlimited resources, and an infinite supply of suitable patients, a placebo comparison may be reasonable, but since neither of these conditions apply, direct comparison of anticonvulsants combinations may have to suffice.
174
ity produce an even higher incidence of adverse effects? On suggesting theoretical advantages to some combinations, the next step is to test each potential combination in animal seizure models. Extrapolation of animal effects, although not ideal, is done routinely in the introduction of novel anticonvulsant compounds into early clinical trials. Even once a potentially beneficial combination comes to our attention, the assessment of any positive interaction will not be easy.
ED50 M
Dose of drug Y
One further way of limiting the number of patients required would be to conduct trials only on preselected combinations of drugs, i.e. those that have shown particular promise in clinical practice (e.g. lamotrigine/valproic acid61 ) or those that might, by virtue of their neurochemical properties, seem to be well- suited to combination (Table 1), e.g. vigabatrin/tiagabine, gabapentin/ vigabatrin. It may be possible to predict useful combinations using knowledge of neurophysiology and neurobiochemistry. Further limitation of the number of combinations to be tested will be difficult. Some relatively basic questions remain unanswered: should combined drugs have similar (e.g. two GABAergic drugs or two glutamatergic drugs) or different mechanism of action? Limited studies in animal models62 have suggested that a number of drugs acting on the same system may be preferable, but much work has to be done to examine this fully. It could be argued that multiplicity of action is what has traditionally bedevilled the use of the older anticonvulsants: according to this view, we should be moving away from drugs that have multiple actions (‘intrinsic polypharmacy’) and searching for a pharmacological action on a single system producing an anticonvulsant action. It is probably unrealistic to expect that we will ever have such a drug, and in any case it can be argued that drug combinations give greater flexibility than a single drug with multiple actions, allowing adjustments to be made in the light of any emergent adverse effects.
J. P. Leach
A S
Dose of drug X
ED50
Fig. 1: Three isobolograms showing effects of combining drugs with additive (A0 ), inhibitory (M0 ) and synergistic (S0 ) effects.
How can we prove synergism in drug combinations? Characterizing ancillary actions of the anticonvulsants—is it worth the effort? To assist in the rational choice of drug combinations, it is important to know not only the mechanism of action of each drug, but what their biochemical ‘side effects’ are. Rational planning of anticonvulsant polypharmacy may begin by examining the effects of the new anticonvulsants on the enzymes and substrates of the GABA shunt in rodent cortex (Table 1). This raises the possibility that combining two drugs that inhibit GABA transaminase, vigabatrin and gabapentin, may allow a fuller inhibition of this enzyme while minimizing the inhibition of glutamic acid decarboxylase. This may theoretically remove the ceiling to effective dosage of vigabatrin71 or could even minimize any adverse events72 . Knowledge of the AEDs’ biochemical effects may also provide information on the combinations to avoid. It would be interesting, for example, to assess the effect of a remacemide/vigabatrin combination on whole brain GABA transaminase. Would this combination have any efficacy in animal seizure models? Would a dual depression of glutamic acid decarboxylase activ-
The isobologram outlined by Loewe in 195373 , may be the most effective way of proving synergism in drug combinations (Fig. 1). Each point plotted on an isobologram is an identical treatment effect (in Fig. 1, this is taken as a 50% decrease in seizure frequency {ED50} among a population). Administration of the ED50 dose of either drug alone for the stated treatment effect will, by definition, produce an identical efficacy. Where drugs have additive effects, similar efficacy will be gained by partial replacement of one drug with an equivalent amount of the other, producing a straight line of effect (A0 ). If the compounds interact in an inhibitory manner, then addition of one drug will require higher doses of the other to produce an identical effect, leading to a convex curve plot (M0 ). In contrast, where the drugs act synergistically, addition of one drug will allow a disproportionately greater reduction in the dose of the other drug, leading to a concave curve (S0 ). In practice, the isobologram has been used to assess (and disprove) the presence of any potentiation between felbamate and phenytoin73 . Where the isobologram would be unnecessary, of
Antiepileptic drugs: safety in numbers?
175
Table 1: Effect of various drugs on the enzymes and substrates of the GABA shunt. Activity of Vigabatrin62, 63 Gabapentin64, 65 Topiramate66, 67 Tiagabine62 Levetiracetam68 Remacemide69
GAD
GABA-T
GABA
Levels of Glutamate
Glutamine
↓↓ ←→ ←→ ←→ ←→ ↓
↓↓ ↓↓ ←→ ←→ ←→ ↑
↑↑ ↑ ↑↑ ←→ ←→ ←→
↓↓ ←→ ←→ ? ←→ ↓
↓↓ ←→ ←→ ? ←→ ←→
GABA-T = GABA-transaminase, GAD = Glutamic Acid Decarboxylase, ↓↓ = significant decrease, ↑↑ = significant increase, ←→ = No effect, ↑ = increase of unknown significance, ? = unknown, ↓ = decrease of unknown significance.
course, would be where a drug, independent of any pharmacokinetic interactions, acts on seizure models only as part of a drug combination. This scenario would in itself be an example of pharmacological synergy. The number of potential anticonvulsant dose combinations, and the huge number of subjects needed, may hamper the use of isobolograms in clinical trials, but it should not preclude their use in animal seizure models. Use among animal seizure models would also overcome the obstacle of defining ‘bioequivalence’ among anticonvulsants. In a disease process as unpredictable as epilepsy, it is almost impossible to specify, for example, the dose of valproic acid that is ‘equivalent’ to a particular dose of lamotrigine. In contrast to the clinical setting, where there is a high degree of interindividual variation in pharmacokinetics and pharmacodynamics of each AED, this is relatively easy to define for a particular animal population. To avoid this problem, any large-scale clinical trials may require the standardization of the serum concentrations of anticonvulsants to ensure that patient are undergoing comparable exposure to each drug. Frustratingly, it should be pointed out that there is no proof of a good correlation between serum concentrations of most of the new anticonvulsants and their clinical efficacy74 .
What are the current gold-standard anticonvulsant combinations? At present, there is a lack of consensus on the requirements for an adequate trial of anticonvulsant synergy. There have been a number of studies investigating the efficacy and tolerability of specific anticonvulsant combinations, but none is without design problems. There is currently an urgent need for pragmatic clinical trials examining the benefits of combination therapy. The lack of such trials may in itself be testimony to our lack of understanding of the basic mechanisms of action of these drugs, an area where further
preclinical investigation may be vital. In general, the principles of AED polypharmacy should comply with the points made by Ferrendelli75 . Ideally, drugs used in combination should: •
have a high therapeutic index;
•
not interact pharmacokinetically;
•
have a low tendency to cause adverse events.
This helps us finally outline some of the anticonvulsant combinations that may be most promising in the clinical arena.
• Lamotrigine and valproic acid
Some clinicians have observed that the combination of lamotrigine and valproic acid has particular efficacy, and this was originally thought to be the result of the well-documented inhibition of lamotrigine metabolism caused by valproic acid6 . Brodie et al.58 , as part of an open trial of add-on Lamotrigine with withdrawal of the primary monotherapy, showed a trend towards optimal benefit from a lamotrigine/valproic acid combination, even when lamotrigine concentrations were maintained after the withdrawal of valproic acid. This was in contrast to the findings in patients where the primary monotherapy was carbamazepine or phenytoin, suggesting that valproic acid and lamotrigine may interact synergistically. Trials involving this combination will necessitate the provision of an unblinded observer to ensure that plasma concentrations, of lamotrigine in particular, are comparable between groups. The suggestion that the combination exhibits synergism, however, requires proof in a definitive double-blind trial. • Vigabatrin and gabapentin
Gabapentin is known to potentiate the effects of other AEDs, even when their mode of action do not appear
176
to be related. Vigabatrin and gabapentin inhibit GABA transaminase and have different ancillary effects on the GABA shunt (Table 1). For the reasons discussed above, the combination of these two new anticonvulsants may theoretically have a higher efficacy and tolerability than either as monotherapy. Encouragingly, both drugs have a wide therapeutic index, are free of pharmacokinetic interactions, and are of proven benefit in partial seizures6 . Careful combination may prove efficacious and well tolerated in the treatment of refractory seizures.
CONCLUSION Three decades of opposition to combining AEDs has inhibited the desire to separate the good combinations from the bad. Despite the introduction of new and effective AED monotherapies, polypharmacy with anticonvulsants remains both inevitable and, for some, beneficial. Remarkably there is still no definite clinical evidence of synergism with anticonvulsants, or even evidence that any particular combination is especially efficacious. For the last decade, it seems that each international epilepsy meeting has featured the obligatory lecture on combination therapy with AEDs. In these 10 years, each time ideal combinations are mentioned, the speaker (and it is usually the same speaker!) provides a list of maybes, mights, ifs and buts. In this age of evidence-based medicine, hopefully we can all agree that this is no longer enough. The increasing knowledge of the molecular basis of epilepsy and its treatment should help the search proceed more efficiently, but there is disappointingly little will to push clinical research in this direction. Since the late 1960s there has been an enormous investment of both time and money in developing a number of new, effective, well-tolerated AEDs. While none is a panacea for epilepsy, they are an undoubted advance. It would be, to say the least, unfortunate if we were to miss the opportunity to optimize their use.
REFERENCES 1. Goodridge, D. M. G. and Shorvon, S. D. Epileptic seizures in a population of 6000. 1: Demography, diagnosis and classification. British Medical Journal 1983; 287: 641–644. 2. Goodridge, D. M. G. and Shorvon, S. D. Epileptic seizures in a population of 6000. Treatment and prognosis. British Medical Journal 1983; 287: 645–647. 3. Shorvon, S. D. Epidemiology, classification, natural history, and genetics of epilepsy. Lancet 1990; 336: 93–96. 4. Brodie, M. J. and Dichter, M. Anticonvulsant drugs. New England Journal of Medicine 1996; 334: 168–175. 5. Dichter, M. A. and Brodie, M J. The antiepileptic drugs—2. New England Journal of Medicine 1996; 334: 1583–1590.
J. P. Leach 6. Leach, J. P. and Brodie, M. New antiepileptic drugs: an explosion of activity. Seizure 1995; 4: 5–17. 7. Reynolds, H. E. Changing view of the prognosis of epilepsy. British Medical Journal 1990; 301: 1112–1114. 8. Mattson, R. M., Cramer, J. A., Collins, J. F. et al. Comparison of phenobarbital, phenytoin and primidone in partial and secondary generalised tonic–clonic seizures. New England Journal of Medicine 1985; 313: 145–151. 9. Walker, M. C. and Sander, J. W. A S. The impact of new antiepileptic drugs on the prognosis of epilepsy: seizure freedom should be the ultimate goal. Neurology 1996; 46: 912– 914. 10. Brodie, M. J., Richens, A. and Yuen, A. W. C. Double-blind comparison of lamotrigine and carbamazepine in newly diagnosed epilepsy. Lancet 1995; 345: 476–479. 11. Mawer, G. E. and Pleuvry, B. J. Interactions involving new antiepileptic drugs. Pharmacology Therapy 1995; 68: 209– 231. 12. Brodie, M. J. Drug interactions and epilepsy. Epilepsia 1992; 33: 13–22. 13. Koller, W. C., Silver, D. E. and Liebermain, A. An algorithm for the management of PD. Neurology 1994; 44 (Suppl. 10): 1–51. 14. Shorvon, S. D. and Reynolds, E. H. Unnecessary polypharmacy for epilepsy. British Medical Journal 1977; 1: 1635– 1637. 15. Shorvon, S. D. and Reynolds, E. H. Reduction of polypharmacy for epilepsy. British Medical Journal 1979; 2: 1023– 1025. 16. Reynolds, E. H. and Shorvon, S. D. Monotherapy or polytherapy for epilepsy. Epilepsia 1981; 22: 1–10. 17. Schmidt, D. and Gram, L. Monotherapy versus polytherapy in epilepsy. CNS Drugs 1995; 3: 194–208. 18. Chadwick, D. W. Diagnosis of epilepsy. Lancet 1990; 336: 291–295. 19. Loiseau, P. Benign focal epilepsies of childhood. In: The Treatment of Epilepsy: Principle and Practice (Ed. E. Wyllie). Philadelphia, Lea and Febiger, 1993: pp. 584–594. 20. Gibbs, J., Appleton, R. E. and Rosenbloom, L. Vigabatrin in intractable childhood epilepsy: a retrospective study. Pediatric Neurology 1992; 8: 338. 21. Parker, A. P., Agathonikou, A., Robinson, R. O. and Panayiotopoulos, C. P. Inappropriate use of carbamazepine and vigabatrin in typical absence seizures. Developmental Medicine and Child Neurology 1998; 40: 517–519. 22. Dulac, O. and N’Guyen, T. The lennox gastaut syndrome. Epilepsia 1993; 34 (Suppl. 7): 7–17. 23. Mattson, R. H., Cramer, J. A., Collins, J. F. et al. A comparison of valproate with carbamazepine for the treatment of complex partial seizures with secondarily generalised tonic– clonic seizures in adults. New England Journal of Medicine 1992; 327: 765. 24. Steiner, T. J. and Yuen, A. W. C. Comparison of lamotrigine and phenytoin monotherapy in newly diagnosed epilepsy. Epilepsia 1994; 35 (Suppl. 8): 31. 25. Reunanen, M., Dam, M. and Yuen, A. W. A randomised open multicentre comparative trial of lamotrigine and carbamazepine as monotherapy in patients with newly diagnosed or recurrent epilepsy. Epilepsy Research 1996; 23: 149–155. 26. Sachdeo, R. C., Reife, R. A., Lim, P. and Pledger, G. Topiramate monotherapy for partial onset seizures. Epilepsia 1997; 38: 294–300. 27. Kalviainen, R., Aikia, M., Saukkonen, A. M. et al. Vigabatrin v. carbamazepine monotherapy in patients with newly diagnosed epilepsy: a randomised controlled study. Archives of Neurology 1995; 52: 989–996. 28. Chadwick, D. W. Rational drug therapy for epilepsy. In: The Epilepsies 2 (Eds D. W. Chadwick and R. J. Porter). Boston, Butterworth-Heineman, 1998: pp. 247–266. 29. Verity, C. M., Hosking, G. and Easter, D. J. A multicentre
Antiepileptic drugs: safety in numbers?
30.
31.
32. 33.
34.
35.
36.
37.
38.
39.
comparative trial of sodium valproate and carbamazepine in paediatric epilepsy. Developmental Medicine and Child Neurology 1995; 37: 97–108. Heller, A. J., Chesterman, P., Elwes, R. D. C. et al. Phenobarbitone, phenytoin, carbamazepine, or sodium valproate for newly diagnosed adult epilepsy. Journal of Neurology, Neurosurgery and Psychiatry 1995; 58: 44. Schmidt, D. and Haenel, F. Therapeutic plasma levels of phenytoin, phenobarbital, and carbamazepine: individual variation in relation to seizure frequency and type. Neurology 1984; 34: 1252–1255. Schmidt, D. Single drug therapy for intractable epilepsy. Journal of Neurology 1983; 229: 221–226. McKee, P. J. W., Blacklaw, J., Butler, E. et al. Monotherapy with conventional and controlled-release carbamazepine: a double blind, double-dummy, comparison in epileptic patients. British Journal of Clinical Pharmacology 1991; 32: 99–104. Persson, L. I., BenMenachem, E., Bengtsson, E. et al. Differences in side effects between a conventional carbamazepine preparation and a slow release preparation of carbamazepine. Epilepsy Research 1990; 6: 137–140. Imaizumi, T., lzumi, T. and Fukuyama, Y. A comparative clinical and pharmacokinetic study of a new slow release versus conventional forms of valproic acid in children with epilepsy. Brain and Development 1992; 14: 304–308. Hakkarainen, H. Carbamazepine vs phenytoin vs their combination in refractory epilepsy [abstract]. Neurology 1980; 30: 354. Dean, J. C. and Penry, J. K. Carbamazepine/valproate therapy in 100 patients with partial seizures failing carbamazepine monotherapy: long term follow up [abstract]. Epilepsia 1988; 29: 687. Marson, A. G., Kadir, Z. A., Hutton, J. L. and Chadwick, D. W. The new antiepileptic drugs; a systematic review of their efficacy and tolerability. British Medical Journal 1996; 313: 1169. Nakane, Y., Okuma, T., Takashi, R. et al. Multi-institutional study of the teratogenicity and fetal toxicity of antiepileptic drugs: a report of a collaborative study group in Japan. Epilepsia 1980; 21: 663–680.
40. Beghi, E., Di Masco, R. and Tognoni, G. Drug treatment of epilepsy: outlines, criticisms and perspectives. Drugs 1986; 31: 249–265. 41. Lammers, M. W., Hekster, Y. A., Keysei, A. et al. Monotherapy or polytherapy for epilepsy revisited: a quantitative assessment. Epilepsia 1995; 36: 440–446. 42. Lindhout, D. and Omtzigt, J. G. Pregnancy and the risk of teratogenicity. Epilepsia 1992; 33 (Suppl. 4): 41–48. 43. Samren, E. B., van Duijn, C. M., Koch, S. et al. Maternal use of antiepileptic drugs and the risk of major congenital malformations: a joint European Prospective Study of human teratogenesis associated with maternal epilepsy. Epilepsia 1997; 38: 981–990. 44. Macdonald, R. L. and Kelly, K. M. Antiepileptic drug mechanisms of action. Epilepsia 1995; (Suppl. 2): 2–12. 45. Bourgeois, B. F. D. Important pharmacokinetic properties of antiepileptic drugs. Epilepsia 1995; 36 (Suppl. 5): 1–7. 46. White, H. S. AES Annual Course Booklet. Comparative anticonvulsant and mechanistic profile of the established and newer antiepileptic drugs, 1998: pp. A1–A18. 47. Jung, M. J. and Palfreyman, M. G. Vigabatrin—mechanisms of action. In: Antiepileptic Drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven Press, 1995: pp. 903–913. 48. Leach, J. P., Sills, G. J., Maijid, A. et al. Effects of tiagabine and vigabatrin on GABA uptake into primary cultures of rat cortical astrocytes. Seizure 1996; i: 229–234. 49. Leach, M. J., Lees, G. and Riddall, D. R. Lamotrigine mechanisms of action. In: Antiepileptic drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven
177 Press, 1995: pp. 861–869. 50. Chez, M. G., Bourgeois, B. F., Pippenger, G. E. et al. Pharmacodynamic interactions between phenytoin and valproate: individual and combined antiepileptic and neurotoxic actions in mice. Clinical Neuropharmacology 1994; 17: 32–37. 51. Bourgeois, B. F. Anticonvulsant potency and neurotoxicity of valproate alone and in combination with carbamazepine or phenobarbital. Clinical Neuropharmacology 1988; 11: 348– 359. 52. Bourgeois, B. F. Combination of valproate and ethosuximide: antiepileptic and neurotoxic interaction. Journal of Pharmacology and Experimental Therapeutics 1988; 247: 1128–1132. 53. Bourgeois, B. F. Antiepileptic drug combinations and experimental background: the case of phenobarbital and phenytoin. Naunyn-Schmiebergs Archives of Pharmacology 1986; 333: 406–411. 54. Bourgeois, B. F. and Wad, N. Combined administration of carbamazepine and phenobarbital: effect on anticonvulsant activity and neurotoxicity. Epilepsia 1988; 29: 481–487. 55. Cereghinno, J. J. and Penry, J. K. Introduction. In: Antiepileptic Drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven Press, 1995: pp. 1–11. 56. Meunier, H., Carraz, G., Meunier, Y. et al. Properietes pharmacodynamiques de l’acide n-dipropylacetique. Therapie 1963; 18: 435–438. 57. Stewart, J., Hughes, E., Kirer, S. et al. Combined vigabatrin and lamotrigine for very refractory epilepsy. Seizure 1992; 1 (Suppl. A): 13–39. 58. Brodie, M. J. and Yuen, A. W. C. and the 105 Study Group. Lamotrigine substitution study: evidence for synergism with sodium valproate. Epilepsy Research 1997; 26: 423–426. 59. Richens, A. Rational polypharmacy. Seizure 1995; 4: 211–214. 60. Leach, J. P. Anticonvulsant Polypharmacy: Focus on synergy. CNS Drugs 1997; 8: 366–375. 61. Panayiotopoluos, C. P., Ferrie, C. D., Knott, C. et al. Interaction of lamotrigine and sodium valproate [letter]. Lancet 1993; 341: 445. 62. Klitgaard, H., Knudsen, M. L. and Jackson, H. C. Synergism between drugs with different mechanisms of actions against audiogenic seizures in DBA/2 mice [abstract]. Epilepsia 1993; 34 (Suppl. 6): 93. 63. Leach, J. P., Sills, G. J., Thompson, G. et al. The Neurochemical effects of tiagabine and vigabatrin alone and in combination in mouse cortex. General Pharmacology 1997; 28: 715– 719. 64. Meldrum, B. S. Update on the mechanism of action of antiepileptic drugs. Epilepsia 1996; 37 (Suppl. 6): 4–11. 65. Leach, J. P., Sills, G. J., Butler, E. et al. Neurochemical actions of gabapentin in mouse brain. Epilepsy Research 1997; 27: 175–180. 66. Pettroff, O. A. C., Rothman, D. L., Behai, K. et al. The effect of gabapentin on brain GABA in patients with epilepsy. Annals of Neurology 1996; 39: 95–99. 67. Sills, G. J., Leach, J P., Butler, E. et al. Neurochemical studies with topiramate in mouse brain [abstract]. Epilepsia 1997; (Suppl. 3): 177. 68. Kuzneicky, R. I., Hoby, H., Ho, S. et al. Topiramate increases cerebral GABA concentrations in healthy adults in the acute and chronic state [abstract]. Epilepsia 1998; 39 (Suppl. 6): 71. 69. Sills, G. J., Leach, J. P., Fraser, C. N. et al. Neurochemical studies with the novel anticonvulsant levetiracetam in mouse brain. European Journal of Pharmacology 1997; 325: 35–40. 70. Leach, J. P., Sills, G. J., Butler, E. et al. Neurochemical actions of the desglycinyl metabolite of remacemide hydrochloride (ARL-12495AA) in mouse brains. British Journal of Pharmacology 1997; 121: 923–926.
178 71. McKee, P. J. W., Blacklaw, J., Friel, E. et al. Adjuvant vigabatrin in refractory epilepsy: a ceiling to effective dosage in individual patients? Epilepsia 1993; 34: 937–943. 72. Grant, S. M. and Heel, R. C. Vigabatrin: a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in epilepsy and disorders of motor control. Drugs 1991; 41: 889–926. 73. Swinyard, E. A., Woodhead, J. H. and Wolf, H. H. Use of
J. P. Leach isobolograms in predicting drug interactions. In: Antiepileptic Drug Interactions (Ed. W. H. Pitlick). New York, Publ Demos, 1989: pp. 261–275. 74. Mattson, R. H. Antiepileptic drug monitoring: a reappraisal. Epilepsia 1995; 36 (Suppl. 2): 22–29. 75. Ferrendelli, J. A. Rational polypharmacy. Epilepsia 1995; 36 (Suppl. 2): 115–118.
doi: 10.1053/seiz.1999.0358, available online at http://www.idealibrary.com on
REVIEW †
Antiepileptic drugs: safety in numbers? JOHN PAUL LEACH
Specialist Registrar in Neurology and Neurophysiology, Walton Centre for Neurology and Neurosurgery, Lower Lane, Fazakerly, Liverpool L9 7LJ, UK
INTRODUCTION Epilepsy is a common neurological condition, with a lifetime prevalence of between 2%1, 2 and 5%3 of the general population, and a point prevalence of between 4 and 80 per 10004 . So common is it among all races that epilepsy may be viewed, not as a rare curse on the individual, but as an all-too frequent price that man and other vertebrates pay for maintaining a plastic and adaptable central nervous system. Within the first few years of the next century, some estimates suggest that there will have been 11 new antiepileptic drugs (AEDs) licensed for use as either add-on or monotherapy since the late 1980s5 . Despite these recent welcome additions to the ranks of available anticonvulsants6 epilepsy remains both stigmatizing and debilitating for many of those with the disorder7 . Are there grounds for optimism about the possibility of treating refractory epilepsy? Previous studies have shown that around one-third of patients will remain refractory to anticonvulsant monotherapy with the established drugs7, 8 . Given this, the new drugs have, with justification, generated a great deal of interest, but despite their use most clinicians would admit that a significant number of patients with epilepsy have seizures that remain unsatisfactorily controlled9 . The challenge for epileptologists over the coming decades is to further reduce the size of this population with refractory disease. The new generation of AEDs are undoubtedly an advance in the treatment of epilepsy (given their advantages of increased tolerability6, 10 and a decreased tendency to cause pharmacokinetic interactions11 ), but it could be argued that a continued search for more new anticonvulsants is an inefficient way of helping our patients. Perhaps it would be more fruitful to look for †
better ways of using the drugs already at our disposal. For many different conditions in many different specialties, the use of drug combinations is the rule rather than the exception. Admittedly, anticonvulsants are potentially neurotoxic8 , can result in severe systemic reactions4 , and have the tendency to sustain multiple pharmacodynamic and pharmacokinetic interactions12 . The same, however, is true of antiparkinsonian drugs13 , and monotherapy would never be advocated as the ideal management strategy for the treatment of Parkinson’s disease13 . In fact here the opposite philosophy applies, where a rational combination of compounds is used at an early stage in an attempt to delay symptomatic progression13 . Since the 1970s, in contrast, anticonvulsant monotherapy has long been regarded as the ideal management strategy. A series of studies by Shorvon and Reynolds14–16 argued against AED polypharmacy on the basis of its strong association with adverse events, they felt, with little evidence of an increase in efficacy. While it seems axiomatic that newly diagnosed epilepsy should be treated with a single drug, the didacticism in favour of monotherapy fails to recognize the realities of everyday practice. One survey has shown17 that even when under review by specialist epilepsy services, almost half the patients required (or, at least, were receiving) long-term treatment with anticonvulsant polypharmacy. At present, even taking into account those compounds currently undergoing phase three testing, it seems unlikely that any of the new AEDs will, by themselves, prove to be the ‘magic bullet’ for epilepsy. In a medical climate so keen on evidence guiding our medical management, it therefore falls to us as epilepsy specialists to formulate a rational strategy for its treatment. Sadly, though, it seems pre-
This was one of the winning entries in the young physician section of the Millenium Gower Prize Competition 1999. 1059–1311/00/030170 + 09
$35.00/0
c 2000 BEA Trading Ltd
Antiepileptic drugs: safety in numbers?
mature to be postulating which AED polypharmacy is best when we still have not fully addressed the issue of whether polypharmacy is superior to monotherapy in the treatment of refractory epilepsy.
THE CASE FOR ANTICONVULSANT MONOTHERAPY The success of AED monotherapy will be enhanced when the drugs are used most appropriately: to state the obvious, each patient should be started on the right drug for their epilepsy at the right dosage. A proper clinical history, with relevant investigations should ensure that patients with a primary generalized epilepsy18 (e.g. juvenile myoclonic epilepsy18 or absence seizures18 ) or benign focal epilepsies of childhood19 receive appropriate initial treatment. Importantly, for example, absence and myoclonic seizures can be exacerbated by carbamazepine20, 21 or vigabatrin20, 21 , so in these circumstances syndromic classification of epilepsy becomes most important, indicating the drugs most likely to protect against all potential seizure types in addition to those which have already revealed themselves. Careful assessment by an experienced epileptologist may allow such a syndromic diagnosis to be made, which can influence the choice of drug, or even indicate that anticonvulsant monotherapy is less likely to be completely successful22 . Phenobarbitone, phenytoin, carbamazepine and valproic acid have shown similar efficacy when used as monotherapy8, 23 in patients with newly diagnosed epilepsy. Some of the newer drugs have been compared with their older counterparts under trial conditions, but many of the key comparisons are still awaited. How good are the new AEDs when they are compared directly with their older counterparts? Brodie et al.10 suggested that lamotrigine was as efficacious as carbamazepine, while being significantly better tolerated. Two further studies comparing lamotrigine with phenytoin24 and carbamazepine25 , however, failed to differentiate them in terms of either efficacy or effectiveness. There remains a need for comparative monotherapy studies of valproate versus lamotrigine. Topiramate has also been assessed as monotherapy26 , but only in comparison with an ‘active placebo’ (a subtherapeutic dose of topiramate). In a study by Kalvainen27 , monotherapy with vigabatrin was better tolerated than with carbamazepine, while lacking its efficacy. Based on the available evidence, the new AEDs are no more efficacious than their older counterparts, so how can our chances of therapeutic success be maximized? Given the incessant calls for ‘evidence’ to guide future practice, we should bear in mind that much of
171
our ‘knowledge’ of AED monotherapy is taken for granted. The superior efficacy of sodium valproate and ethosuximide over phenytoin and carbamazepine in the absence of seizures seems too obvious to demand confirmation in a prospective clinical study, but there are no data differentiating the efficacies of sodium valproate and ethosuximide in the absence of epilepsy28 . Similarly, while no one would argue that for most patients with juvenile myoclonic epilepsy, sodium valproate is the drug of choice29 , no major differences in efficacy between carbamazepine, phenytoin, barbiturates, and sodium valproate in the treatment of partial epilepsies have been fully established8, 23, 30 , in the face of conventional practice in many epilepsy clinics. The limited role for barbiturate drugs is supported in the study by Mattson et al.8 , where they were shown to be less well tolerated than phenytoin and carbamazepine. How should we deal with cases where monotherapy fails to give complete seizure control? Firstly the wide interindividual variation in optimal doses of anticonvulsants31 should be borne in mind: one study32 has shown that increasing the dosage of current therapy to maximally tolerated levels can bring seizure control in almost one-third of patients considered to have seizures ‘refractory’ to monotherapy with phenytoin or primidone. Adverse events may be minimized during dosage escalation by avoiding over-rapid dosage titration or by using controlled-release formulations, particularly with carbamazepine33, 34 and less convincingly with valproic acid35 . Even when the first monotherapy fails, it need not necessarily follow that polypharmacy is an inevitable requirement. Hakkareinen36 investigated the success rate of monotherapy with either carbamazepine or phenytoin when the other had failed, and found that one-third of the treatment failures were successfully controlled by monotherapy with the alternative drug. Similarly, Mattson et al.8 found that 46% of those unsuccessfully treated with an initial monotherapy responded to an alternative monotherapy. In this study, however, some patients were originally treated with phenobarbital or primidone, neither of which would he today accept as reasonable first-line monotherapy. Anticonvulsant monotherapy will be adequate for most patients with newly diagnosed epilepsy, and it is of paramount importance that the chances of monotherapy succeeding are maximized. Even if the first drug fails at maximally tolerated dosages, however, a second monotherapy may well prove to be effective in a substantial minority of patients. In summary, however, although AED monotherapy is good, it is still not good enough.
172
THE CASE FOR AED POLYPHARMACY Notwithstanding the poor image of polypharmacy, monotherapy will fail in a significant proportion of patients, necessitating a trial of polypharmacy with anticonvulsants. Despite the observations of Hakkareinen36 and Mattson8 , we have no randomized trials comparing alternative monotherapy with add-on therapy in groups of patients with poorly controlled epilepsy. In the study by Mattson et al.8 , 39% of monotherapy ‘non-responders’ were helped by the addition of a second drug, with 17% becoming seizurefree. A trial of adjuvant valproic acid in cases where carbamazepine had failed as monotherapy, demonstrated ‘improvement’ in a similar number of patients while on combination therapy, with 17% becoming seizure-free37 . Certainly there is an abundance of trials to prove the superiority of the new antiepileptic drugs over placebo in refractory epilepsy38 , although there would be some grounds for doubting whether all of the trial subjects were previously optimally treated! Despite its efficacy, polypharmacy has long been thought to be fraught with disadvantages: it is held to be responsible for an increased frequency of adverse events17 , pharmacokinetic interactions12 , and teratogenesis39 , while conferring only limited improvements in seizure control16 . In one study40 , monotherapy with phenytoin or carbamazepine produced adverse effects in 28 and 38% of patients respectively, while their combination produced adverse events in around three-quarters of all patients. However, Lammers et al.41 , following a retrospective review, were of the opinion that it was dosage rather than the number of anticonvulsants used that determined the frequency and severity of adverse events. As an ‘irreversible’ adverse event, teratogenicity must bear specific examination. Even as monotherapy, AEDs42, 43 seem to increase the risk of major fetal malformation by a factor of around five43 : the most common AED-related malformations are harelip, cleft palate and cardiovascular anomalies. Such data should be enough to ensure that AED therapy is only begun after careful consideration. Estimates suggest the malformation risk to be 1–2% of pregnancies on AEDs28 . Early screening with fetal ultrasound, serum alpha-fetoprotein and, where indicated, amniocentesis should be offered to women becoming pregnant while on AEDs. There are good reasons why polypharmacy with older anticonvulsants should not be universally successful. Each has a wide range of actions44 , while their anticonvulsant efficacy relies, at least in part, on sodium channel blockade44 . Random combination of these widely active drugs might reasonably be thought to merely increase the degree of sodium channel blockade, while producing a host of more compli-
J. P. Leach
cated biochemical and physiological effects. Mixing drugs with such ‘indiscriminate’ actions as the established anticonvulsants, will almost inevitably lead to problems. The human brain is a delicate instrument, and tampering with a range of neuronal properties may have far-reaching consequences. In fact, considering their potential for pharmacokinetic and pharmacodynamic interactions, it is almost surprising that anyone can tolerate polypharmacy with the older anticonvulsants! It is fair to say that, with the advent of a new generation of anticonvulsants, polypharmacy appears to be more promising. The newer AEDs have more predictable pharmacokinetics than the older agents45 and the incidence and severity of pharmacokinetic interactions with them is markedly reduced6, 45 , which makes their combination less hazardous and complicated11 . In addition, with the probable exceptions of topiramate and gabapentin, the new anticonvulsants are reputed to have a single, well-defined mode of action46 . While there is something appealing in the concept of a pharmacological agent acting on a single process to prevent seizures, some scepticism about the assurances of specificity are warranted, given that after marketing, vigabatrin47, 48 and lamotrigine49 were shown to have additional actions that may contribute to their anticonvulsant effect. Further work on other models has certainly hinted at other modes of action46 , and it should not surprise us if the number of actions of each drug increases with each different model tested. Although careful investigation of the effect of these new anticonvulsants might complicate our thinking about these compounds, it may in return open up further therapeutic possibilities in the treatment of epilepsy. Despite these negative aspects of combinations of the established AEDs, it is still rather puzzling that the inevitability of AED polypharmacy has not inspired any large-scale clinical trials to ascertain which combinations are most effective. In a series of studies in mice, Bourgeois and co-workers50–54 examined the anticonvulsant and neurotoxic results of combining the established anticonvulsants. Anticonvulsant concentrations were standardized to avoid the confounding of the results by the expected pharmacokinetic interactions. Using these methods, it was suggested that combinations of phenytoin and valproic acid had an anticonvulsant effect that was supra-additive (i.e. displayed synergism between the two drugs) while the neurotoxic effects were merely additive45 . Further animal studies suggested that the anticonvulsant benefits of combining valproic acid with carbamazepine51 or ethosuximide52 outweighed the neurotoxic effects, with similar positive effects from a phenytoin/phenobarbital combination54 . In contrast, a valproic acid/phenobarbital51 and phenobar-
Antiepileptic drugs: safety in numbers?
bital/carbamazepine54 combination were more disappointing, with both the neurotoxic and anticonvulsant effects being additive. Of course, none of our patients are mice, so where does this leave us? It may be considered surprising that none of these potentially beneficial results were followed up by proper clinical trials, although by the late 1980s, the newer anticonvulsants had begun to take centre stage as the source of new hope for those with intractable epilepsy. For two decades there has been ongoing debate about the relative merits of polypharmacy and monotherapy. The fact that available monotherapies are not perfect has left clinicians reliant on polypharmacy, and so doctors are left to choose between the theoretically desirable and the practically unavoidable. Given this situation, how can we best plan combination anticonvulsant drug therapy? This question can only be addressed once several smaller problems have been addressed.
How can we develop a rational plan for optimal anticonvulsant polypharmacy?
173
(b) Meta-analysis
Meta-analyses comparing the newer drugs should be interpreted with caution, as each has mostly succeeded in comparing toxic doses of one compound with subtherapeutic doses of another. Other as yet untried meta-analyses, however, are potentially far more productive: prospective anticonvulsants have to undergo clinical testing as add-on therapy, and one efficient method of assessing drug combinations would be to measure the relative benefits of each new treatment in tandem with each established anticonvulsant. At present, plans are underway to carry out such studies [A. Marson, personal communication]. Comparison across trials should be interpreted with great caution, but such analyses may help to highlight any particularly potent combinations. The pharmaceutical industry needs some encouragement to participate, since each company is understandably keen for their product to be seen as a ‘cutting-edge’ compound rather than as a cofactor. Such meta-analyses, however, are of great potential benefit to the industry, for example in rapidly highlighting any potential important adverse pharmacokinetic or pharmacodynamic interactions. (c) Systematic clinical trials involving each drug combination
(a) Chance observation
The history of epilepsy treatments is littered with chance observations, from Hauptmann’s discovery of the anticonvulsant effects of phenobarbital55 to the discovery of the anticonvulsant properties of valproic acid56 and lamotrigine6 . Serendipity is wellrecognized as a ‘method’ of formulating therapeutic advances, being the pay-off for astute clinical observation. Anecdotal reports have suggested specific benefit with particular combinations (e.g. vigabatrin/lamotrigine57 , lamotrigine/valproic acid58 ), but none has been assessed in proper clinical trials. One important point to note is that if clinical observation is to be relied upon to highlight ideal drug combinations, then such opinions should be based on better information. For example, it would be helpful if the data upon which drug licensing applications are based were made available at an early stage, preferably in peer-reviewed journals. It has become apparent that there is a paucity of freely available clinical and preclinical data, even for those drugs on general release. It is less than ideal that data regarding marketed compounds have to be sifted from journal abstracts, and it is probably unhealthy for our understanding of such drugs to be based solely on the subjective (and occasionally selective) data provided by the pharmaceutical industry.
It would be ideal if every anticonvulsant combination could be tested in humans using rigorously designed double-dummy, placebo-controlled studies59 . However, assuming that there are eight available anticonvulsant compounds (a conservative estimate for most countries), this allows 28 combinations of 2-drug and 56 of 3-drug cotreatments60 . Having four treatment arms (two active monotherapies, one placebo, one combination), such trials would require very large numbers of patients, one-quarter of whom would be receiving placebo. In countries outside the USA at least, there is some unease about giving placebo as monotherapy for epilepsy, so in any future trials, common sense may render the placebo arm unnecessary. It is (hopefully) widely recognized that the new drugs are active anticonvulsants, and it could be argued with equal conviction that placebo is not an adequate add-on treatment for refractory epilepsy. The aim of any large trial should not be to ‘reinvent the wheel’ and prove that a combination is better than placebo, but to prove that a particular combination is more beneficial than any other. In an ideal world, with unlimited resources, and an infinite supply of suitable patients, a placebo comparison may be reasonable, but since neither of these conditions apply, direct comparison of anticonvulsants combinations may have to suffice.
174
ity produce an even higher incidence of adverse effects? On suggesting theoretical advantages to some combinations, the next step is to test each potential combination in animal seizure models. Extrapolation of animal effects, although not ideal, is done routinely in the introduction of novel anticonvulsant compounds into early clinical trials. Even once a potentially beneficial combination comes to our attention, the assessment of any positive interaction will not be easy.
ED50 M
Dose of drug Y
One further way of limiting the number of patients required would be to conduct trials only on preselected combinations of drugs, i.e. those that have shown particular promise in clinical practice (e.g. lamotrigine/valproic acid61 ) or those that might, by virtue of their neurochemical properties, seem to be well- suited to combination (Table 1), e.g. vigabatrin/tiagabine, gabapentin/ vigabatrin. It may be possible to predict useful combinations using knowledge of neurophysiology and neurobiochemistry. Further limitation of the number of combinations to be tested will be difficult. Some relatively basic questions remain unanswered: should combined drugs have similar (e.g. two GABAergic drugs or two glutamatergic drugs) or different mechanism of action? Limited studies in animal models62 have suggested that a number of drugs acting on the same system may be preferable, but much work has to be done to examine this fully. It could be argued that multiplicity of action is what has traditionally bedevilled the use of the older anticonvulsants: according to this view, we should be moving away from drugs that have multiple actions (‘intrinsic polypharmacy’) and searching for a pharmacological action on a single system producing an anticonvulsant action. It is probably unrealistic to expect that we will ever have such a drug, and in any case it can be argued that drug combinations give greater flexibility than a single drug with multiple actions, allowing adjustments to be made in the light of any emergent adverse effects.
J. P. Leach
A S
Dose of drug X
ED50
Fig. 1: Three isobolograms showing effects of combining drugs with additive (A0 ), inhibitory (M0 ) and synergistic (S0 ) effects.
How can we prove synergism in drug combinations? Characterizing ancillary actions of the anticonvulsants—is it worth the effort? To assist in the rational choice of drug combinations, it is important to know not only the mechanism of action of each drug, but what their biochemical ‘side effects’ are. Rational planning of anticonvulsant polypharmacy may begin by examining the effects of the new anticonvulsants on the enzymes and substrates of the GABA shunt in rodent cortex (Table 1). This raises the possibility that combining two drugs that inhibit GABA transaminase, vigabatrin and gabapentin, may allow a fuller inhibition of this enzyme while minimizing the inhibition of glutamic acid decarboxylase. This may theoretically remove the ceiling to effective dosage of vigabatrin71 or could even minimize any adverse events72 . Knowledge of the AEDs’ biochemical effects may also provide information on the combinations to avoid. It would be interesting, for example, to assess the effect of a remacemide/vigabatrin combination on whole brain GABA transaminase. Would this combination have any efficacy in animal seizure models? Would a dual depression of glutamic acid decarboxylase activ-
The isobologram outlined by Loewe in 195373 , may be the most effective way of proving synergism in drug combinations (Fig. 1). Each point plotted on an isobologram is an identical treatment effect (in Fig. 1, this is taken as a 50% decrease in seizure frequency {ED50} among a population). Administration of the ED50 dose of either drug alone for the stated treatment effect will, by definition, produce an identical efficacy. Where drugs have additive effects, similar efficacy will be gained by partial replacement of one drug with an equivalent amount of the other, producing a straight line of effect (A0 ). If the compounds interact in an inhibitory manner, then addition of one drug will require higher doses of the other to produce an identical effect, leading to a convex curve plot (M0 ). In contrast, where the drugs act synergistically, addition of one drug will allow a disproportionately greater reduction in the dose of the other drug, leading to a concave curve (S0 ). In practice, the isobologram has been used to assess (and disprove) the presence of any potentiation between felbamate and phenytoin73 . Where the isobologram would be unnecessary, of
Antiepileptic drugs: safety in numbers?
175
Table 1: Effect of various drugs on the enzymes and substrates of the GABA shunt. Activity of Vigabatrin62, 63 Gabapentin64, 65 Topiramate66, 67 Tiagabine62 Levetiracetam68 Remacemide69
GAD
GABA-T
GABA
Levels of Glutamate
Glutamine
↓↓ ←→ ←→ ←→ ←→ ↓
↓↓ ↓↓ ←→ ←→ ←→ ↑
↑↑ ↑ ↑↑ ←→ ←→ ←→
↓↓ ←→ ←→ ? ←→ ↓
↓↓ ←→ ←→ ? ←→ ←→
GABA-T = GABA-transaminase, GAD = Glutamic Acid Decarboxylase, ↓↓ = significant decrease, ↑↑ = significant increase, ←→ = No effect, ↑ = increase of unknown significance, ? = unknown, ↓ = decrease of unknown significance.
course, would be where a drug, independent of any pharmacokinetic interactions, acts on seizure models only as part of a drug combination. This scenario would in itself be an example of pharmacological synergy. The number of potential anticonvulsant dose combinations, and the huge number of subjects needed, may hamper the use of isobolograms in clinical trials, but it should not preclude their use in animal seizure models. Use among animal seizure models would also overcome the obstacle of defining ‘bioequivalence’ among anticonvulsants. In a disease process as unpredictable as epilepsy, it is almost impossible to specify, for example, the dose of valproic acid that is ‘equivalent’ to a particular dose of lamotrigine. In contrast to the clinical setting, where there is a high degree of interindividual variation in pharmacokinetics and pharmacodynamics of each AED, this is relatively easy to define for a particular animal population. To avoid this problem, any large-scale clinical trials may require the standardization of the serum concentrations of anticonvulsants to ensure that patient are undergoing comparable exposure to each drug. Frustratingly, it should be pointed out that there is no proof of a good correlation between serum concentrations of most of the new anticonvulsants and their clinical efficacy74 .
What are the current gold-standard anticonvulsant combinations? At present, there is a lack of consensus on the requirements for an adequate trial of anticonvulsant synergy. There have been a number of studies investigating the efficacy and tolerability of specific anticonvulsant combinations, but none is without design problems. There is currently an urgent need for pragmatic clinical trials examining the benefits of combination therapy. The lack of such trials may in itself be testimony to our lack of understanding of the basic mechanisms of action of these drugs, an area where further
preclinical investigation may be vital. In general, the principles of AED polypharmacy should comply with the points made by Ferrendelli75 . Ideally, drugs used in combination should: •
have a high therapeutic index;
•
not interact pharmacokinetically;
•
have a low tendency to cause adverse events.
This helps us finally outline some of the anticonvulsant combinations that may be most promising in the clinical arena.
• Lamotrigine and valproic acid
Some clinicians have observed that the combination of lamotrigine and valproic acid has particular efficacy, and this was originally thought to be the result of the well-documented inhibition of lamotrigine metabolism caused by valproic acid6 . Brodie et al.58 , as part of an open trial of add-on Lamotrigine with withdrawal of the primary monotherapy, showed a trend towards optimal benefit from a lamotrigine/valproic acid combination, even when lamotrigine concentrations were maintained after the withdrawal of valproic acid. This was in contrast to the findings in patients where the primary monotherapy was carbamazepine or phenytoin, suggesting that valproic acid and lamotrigine may interact synergistically. Trials involving this combination will necessitate the provision of an unblinded observer to ensure that plasma concentrations, of lamotrigine in particular, are comparable between groups. The suggestion that the combination exhibits synergism, however, requires proof in a definitive double-blind trial. • Vigabatrin and gabapentin
Gabapentin is known to potentiate the effects of other AEDs, even when their mode of action do not appear
176
to be related. Vigabatrin and gabapentin inhibit GABA transaminase and have different ancillary effects on the GABA shunt (Table 1). For the reasons discussed above, the combination of these two new anticonvulsants may theoretically have a higher efficacy and tolerability than either as monotherapy. Encouragingly, both drugs have a wide therapeutic index, are free of pharmacokinetic interactions, and are of proven benefit in partial seizures6 . Careful combination may prove efficacious and well tolerated in the treatment of refractory seizures.
CONCLUSION Three decades of opposition to combining AEDs has inhibited the desire to separate the good combinations from the bad. Despite the introduction of new and effective AED monotherapies, polypharmacy with anticonvulsants remains both inevitable and, for some, beneficial. Remarkably there is still no definite clinical evidence of synergism with anticonvulsants, or even evidence that any particular combination is especially efficacious. For the last decade, it seems that each international epilepsy meeting has featured the obligatory lecture on combination therapy with AEDs. In these 10 years, each time ideal combinations are mentioned, the speaker (and it is usually the same speaker!) provides a list of maybes, mights, ifs and buts. In this age of evidence-based medicine, hopefully we can all agree that this is no longer enough. The increasing knowledge of the molecular basis of epilepsy and its treatment should help the search proceed more efficiently, but there is disappointingly little will to push clinical research in this direction. Since the late 1960s there has been an enormous investment of both time and money in developing a number of new, effective, well-tolerated AEDs. While none is a panacea for epilepsy, they are an undoubted advance. It would be, to say the least, unfortunate if we were to miss the opportunity to optimize their use.
REFERENCES 1. Goodridge, D. M. G. and Shorvon, S. D. Epileptic seizures in a population of 6000. 1: Demography, diagnosis and classification. British Medical Journal 1983; 287: 641–644. 2. Goodridge, D. M. G. and Shorvon, S. D. Epileptic seizures in a population of 6000. Treatment and prognosis. British Medical Journal 1983; 287: 645–647. 3. Shorvon, S. D. Epidemiology, classification, natural history, and genetics of epilepsy. Lancet 1990; 336: 93–96. 4. Brodie, M. J. and Dichter, M. Anticonvulsant drugs. New England Journal of Medicine 1996; 334: 168–175. 5. Dichter, M. A. and Brodie, M J. The antiepileptic drugs—2. New England Journal of Medicine 1996; 334: 1583–1590.
J. P. Leach 6. Leach, J. P. and Brodie, M. New antiepileptic drugs: an explosion of activity. Seizure 1995; 4: 5–17. 7. Reynolds, H. E. Changing view of the prognosis of epilepsy. British Medical Journal 1990; 301: 1112–1114. 8. Mattson, R. M., Cramer, J. A., Collins, J. F. et al. Comparison of phenobarbital, phenytoin and primidone in partial and secondary generalised tonic–clonic seizures. New England Journal of Medicine 1985; 313: 145–151. 9. Walker, M. C. and Sander, J. W. A S. The impact of new antiepileptic drugs on the prognosis of epilepsy: seizure freedom should be the ultimate goal. Neurology 1996; 46: 912– 914. 10. Brodie, M. J., Richens, A. and Yuen, A. W. C. Double-blind comparison of lamotrigine and carbamazepine in newly diagnosed epilepsy. Lancet 1995; 345: 476–479. 11. Mawer, G. E. and Pleuvry, B. J. Interactions involving new antiepileptic drugs. Pharmacology Therapy 1995; 68: 209– 231. 12. Brodie, M. J. Drug interactions and epilepsy. Epilepsia 1992; 33: 13–22. 13. Koller, W. C., Silver, D. E. and Liebermain, A. An algorithm for the management of PD. Neurology 1994; 44 (Suppl. 10): 1–51. 14. Shorvon, S. D. and Reynolds, E. H. Unnecessary polypharmacy for epilepsy. British Medical Journal 1977; 1: 1635– 1637. 15. Shorvon, S. D. and Reynolds, E. H. Reduction of polypharmacy for epilepsy. British Medical Journal 1979; 2: 1023– 1025. 16. Reynolds, E. H. and Shorvon, S. D. Monotherapy or polytherapy for epilepsy. Epilepsia 1981; 22: 1–10. 17. Schmidt, D. and Gram, L. Monotherapy versus polytherapy in epilepsy. CNS Drugs 1995; 3: 194–208. 18. Chadwick, D. W. Diagnosis of epilepsy. Lancet 1990; 336: 291–295. 19. Loiseau, P. Benign focal epilepsies of childhood. In: The Treatment of Epilepsy: Principle and Practice (Ed. E. Wyllie). Philadelphia, Lea and Febiger, 1993: pp. 584–594. 20. Gibbs, J., Appleton, R. E. and Rosenbloom, L. Vigabatrin in intractable childhood epilepsy: a retrospective study. Pediatric Neurology 1992; 8: 338. 21. Parker, A. P., Agathonikou, A., Robinson, R. O. and Panayiotopoulos, C. P. Inappropriate use of carbamazepine and vigabatrin in typical absence seizures. Developmental Medicine and Child Neurology 1998; 40: 517–519. 22. Dulac, O. and N’Guyen, T. The lennox gastaut syndrome. Epilepsia 1993; 34 (Suppl. 7): 7–17. 23. Mattson, R. H., Cramer, J. A., Collins, J. F. et al. A comparison of valproate with carbamazepine for the treatment of complex partial seizures with secondarily generalised tonic– clonic seizures in adults. New England Journal of Medicine 1992; 327: 765. 24. Steiner, T. J. and Yuen, A. W. C. Comparison of lamotrigine and phenytoin monotherapy in newly diagnosed epilepsy. Epilepsia 1994; 35 (Suppl. 8): 31. 25. Reunanen, M., Dam, M. and Yuen, A. W. A randomised open multicentre comparative trial of lamotrigine and carbamazepine as monotherapy in patients with newly diagnosed or recurrent epilepsy. Epilepsy Research 1996; 23: 149–155. 26. Sachdeo, R. C., Reife, R. A., Lim, P. and Pledger, G. Topiramate monotherapy for partial onset seizures. Epilepsia 1997; 38: 294–300. 27. Kalviainen, R., Aikia, M., Saukkonen, A. M. et al. Vigabatrin v. carbamazepine monotherapy in patients with newly diagnosed epilepsy: a randomised controlled study. Archives of Neurology 1995; 52: 989–996. 28. Chadwick, D. W. Rational drug therapy for epilepsy. In: The Epilepsies 2 (Eds D. W. Chadwick and R. J. Porter). Boston, Butterworth-Heineman, 1998: pp. 247–266. 29. Verity, C. M., Hosking, G. and Easter, D. J. A multicentre
Antiepileptic drugs: safety in numbers?
30.
31.
32. 33.
34.
35.
36.
37.
38.
39.
comparative trial of sodium valproate and carbamazepine in paediatric epilepsy. Developmental Medicine and Child Neurology 1995; 37: 97–108. Heller, A. J., Chesterman, P., Elwes, R. D. C. et al. Phenobarbitone, phenytoin, carbamazepine, or sodium valproate for newly diagnosed adult epilepsy. Journal of Neurology, Neurosurgery and Psychiatry 1995; 58: 44. Schmidt, D. and Haenel, F. Therapeutic plasma levels of phenytoin, phenobarbital, and carbamazepine: individual variation in relation to seizure frequency and type. Neurology 1984; 34: 1252–1255. Schmidt, D. Single drug therapy for intractable epilepsy. Journal of Neurology 1983; 229: 221–226. McKee, P. J. W., Blacklaw, J., Butler, E. et al. Monotherapy with conventional and controlled-release carbamazepine: a double blind, double-dummy, comparison in epileptic patients. British Journal of Clinical Pharmacology 1991; 32: 99–104. Persson, L. I., BenMenachem, E., Bengtsson, E. et al. Differences in side effects between a conventional carbamazepine preparation and a slow release preparation of carbamazepine. Epilepsy Research 1990; 6: 137–140. Imaizumi, T., lzumi, T. and Fukuyama, Y. A comparative clinical and pharmacokinetic study of a new slow release versus conventional forms of valproic acid in children with epilepsy. Brain and Development 1992; 14: 304–308. Hakkarainen, H. Carbamazepine vs phenytoin vs their combination in refractory epilepsy [abstract]. Neurology 1980; 30: 354. Dean, J. C. and Penry, J. K. Carbamazepine/valproate therapy in 100 patients with partial seizures failing carbamazepine monotherapy: long term follow up [abstract]. Epilepsia 1988; 29: 687. Marson, A. G., Kadir, Z. A., Hutton, J. L. and Chadwick, D. W. The new antiepileptic drugs; a systematic review of their efficacy and tolerability. British Medical Journal 1996; 313: 1169. Nakane, Y., Okuma, T., Takashi, R. et al. Multi-institutional study of the teratogenicity and fetal toxicity of antiepileptic drugs: a report of a collaborative study group in Japan. Epilepsia 1980; 21: 663–680.
40. Beghi, E., Di Masco, R. and Tognoni, G. Drug treatment of epilepsy: outlines, criticisms and perspectives. Drugs 1986; 31: 249–265. 41. Lammers, M. W., Hekster, Y. A., Keysei, A. et al. Monotherapy or polytherapy for epilepsy revisited: a quantitative assessment. Epilepsia 1995; 36: 440–446. 42. Lindhout, D. and Omtzigt, J. G. Pregnancy and the risk of teratogenicity. Epilepsia 1992; 33 (Suppl. 4): 41–48. 43. Samren, E. B., van Duijn, C. M., Koch, S. et al. Maternal use of antiepileptic drugs and the risk of major congenital malformations: a joint European Prospective Study of human teratogenesis associated with maternal epilepsy. Epilepsia 1997; 38: 981–990. 44. Macdonald, R. L. and Kelly, K. M. Antiepileptic drug mechanisms of action. Epilepsia 1995; (Suppl. 2): 2–12. 45. Bourgeois, B. F. D. Important pharmacokinetic properties of antiepileptic drugs. Epilepsia 1995; 36 (Suppl. 5): 1–7. 46. White, H. S. AES Annual Course Booklet. Comparative anticonvulsant and mechanistic profile of the established and newer antiepileptic drugs, 1998: pp. A1–A18. 47. Jung, M. J. and Palfreyman, M. G. Vigabatrin—mechanisms of action. In: Antiepileptic Drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven Press, 1995: pp. 903–913. 48. Leach, J. P., Sills, G. J., Maijid, A. et al. Effects of tiagabine and vigabatrin on GABA uptake into primary cultures of rat cortical astrocytes. Seizure 1996; i: 229–234. 49. Leach, M. J., Lees, G. and Riddall, D. R. Lamotrigine mechanisms of action. In: Antiepileptic drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven
177 Press, 1995: pp. 861–869. 50. Chez, M. G., Bourgeois, B. F., Pippenger, G. E. et al. Pharmacodynamic interactions between phenytoin and valproate: individual and combined antiepileptic and neurotoxic actions in mice. Clinical Neuropharmacology 1994; 17: 32–37. 51. Bourgeois, B. F. Anticonvulsant potency and neurotoxicity of valproate alone and in combination with carbamazepine or phenobarbital. Clinical Neuropharmacology 1988; 11: 348– 359. 52. Bourgeois, B. F. Combination of valproate and ethosuximide: antiepileptic and neurotoxic interaction. Journal of Pharmacology and Experimental Therapeutics 1988; 247: 1128–1132. 53. Bourgeois, B. F. Antiepileptic drug combinations and experimental background: the case of phenobarbital and phenytoin. Naunyn-Schmiebergs Archives of Pharmacology 1986; 333: 406–411. 54. Bourgeois, B. F. and Wad, N. Combined administration of carbamazepine and phenobarbital: effect on anticonvulsant activity and neurotoxicity. Epilepsia 1988; 29: 481–487. 55. Cereghinno, J. J. and Penry, J. K. Introduction. In: Antiepileptic Drugs. 4th Edition, (Eds R. H. Levy, R. H. Mattson and B. S. Meldrum). New York, Raven Press, 1995: pp. 1–11. 56. Meunier, H., Carraz, G., Meunier, Y. et al. Properietes pharmacodynamiques de l’acide n-dipropylacetique. Therapie 1963; 18: 435–438. 57. Stewart, J., Hughes, E., Kirer, S. et al. Combined vigabatrin and lamotrigine for very refractory epilepsy. Seizure 1992; 1 (Suppl. A): 13–39. 58. Brodie, M. J. and Yuen, A. W. C. and the 105 Study Group. Lamotrigine substitution study: evidence for synergism with sodium valproate. Epilepsy Research 1997; 26: 423–426. 59. Richens, A. Rational polypharmacy. Seizure 1995; 4: 211–214. 60. Leach, J. P. Anticonvulsant Polypharmacy: Focus on synergy. CNS Drugs 1997; 8: 366–375. 61. Panayiotopoluos, C. P., Ferrie, C. D., Knott, C. et al. Interaction of lamotrigine and sodium valproate [letter]. Lancet 1993; 341: 445. 62. Klitgaard, H., Knudsen, M. L. and Jackson, H. C. Synergism between drugs with different mechanisms of actions against audiogenic seizures in DBA/2 mice [abstract]. Epilepsia 1993; 34 (Suppl. 6): 93. 63. Leach, J. P., Sills, G. J., Thompson, G. et al. The Neurochemical effects of tiagabine and vigabatrin alone and in combination in mouse cortex. General Pharmacology 1997; 28: 715– 719. 64. Meldrum, B. S. Update on the mechanism of action of antiepileptic drugs. Epilepsia 1996; 37 (Suppl. 6): 4–11. 65. Leach, J. P., Sills, G. J., Butler, E. et al. Neurochemical actions of gabapentin in mouse brain. Epilepsy Research 1997; 27: 175–180. 66. Pettroff, O. A. C., Rothman, D. L., Behai, K. et al. The effect of gabapentin on brain GABA in patients with epilepsy. Annals of Neurology 1996; 39: 95–99. 67. Sills, G. J., Leach, J P., Butler, E. et al. Neurochemical studies with topiramate in mouse brain [abstract]. Epilepsia 1997; (Suppl. 3): 177. 68. Kuzneicky, R. I., Hoby, H., Ho, S. et al. Topiramate increases cerebral GABA concentrations in healthy adults in the acute and chronic state [abstract]. Epilepsia 1998; 39 (Suppl. 6): 71. 69. Sills, G. J., Leach, J. P., Fraser, C. N. et al. Neurochemical studies with the novel anticonvulsant levetiracetam in mouse brain. European Journal of Pharmacology 1997; 325: 35–40. 70. Leach, J. P., Sills, G. J., Butler, E. et al. Neurochemical actions of the desglycinyl metabolite of remacemide hydrochloride (ARL-12495AA) in mouse brains. British Journal of Pharmacology 1997; 121: 923–926.
178 71. McKee, P. J. W., Blacklaw, J., Friel, E. et al. Adjuvant vigabatrin in refractory epilepsy: a ceiling to effective dosage in individual patients? Epilepsia 1993; 34: 937–943. 72. Grant, S. M. and Heel, R. C. Vigabatrin: a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in epilepsy and disorders of motor control. Drugs 1991; 41: 889–926. 73. Swinyard, E. A., Woodhead, J. H. and Wolf, H. H. Use of
J. P. Leach isobolograms in predicting drug interactions. In: Antiepileptic Drug Interactions (Ed. W. H. Pitlick). New York, Publ Demos, 1989: pp. 261–275. 74. Mattson, R. H. Antiepileptic drug monitoring: a reappraisal. Epilepsia 1995; 36 (Suppl. 2): 22–29. 75. Ferrendelli, J. A. Rational polypharmacy. Epilepsia 1995; 36 (Suppl. 2): 115–118.