Emilio Perucca MD, PhD*
René H. Levy PhD**
* Department of Internal Medicine and Therapeutics, Clinical Pharmacology Unit, University of Pavia, Pavia, Italy
** Professor and Chair, Department of Pharmaceutics; and Professor of Neurological Surgery, University of Washington School of Pharmacy and Medicine, Seattle, Washington
Up to the early 1970s, the use of combination therapy was standard practice among neurologists treating people with epilepsy: for example, a survey conducted in four European countries in 1975 showed that each patient received an average of 3.2 drugs, of which more than two-thirds were anticonvulsants (1). This practice was based on the neverproven assumption that the simultaneous prescription of two or more antiepileptic drugs (AEDs) ensured synergistic therapeutic activity while protecting against the risk of excessive toxicity. Phenobarbital and phenytoin were by far the most commonly coprescribed drugs, to the extent that many AED preparations available on the market at the time contained fixed-ratio combinations of these agents (2). There were other reasons for prescribing more than one drug at the same time. Because of the lack of broad-spectrum agents, patients with multiple seizure types often required combination therapy to obtain complete control of their seizures, the best example being the use of ethosuximide and phenobarbital to suppress absence and tonic-clonic seizures, respectively. It was also not uncommon to prescribe additional central nervous system (CNS) medications, such as bromide, atropine, caffeine, and amphetamine, in an attempt to potentiate antiepileptic efficacy or to counteract the sedative effects of first-line AEDs (2).
Unfortunately, the goal of seizure freedom with little or no toxicity was seldom achieved with these therapies. Some patients failed to achieve seizure control when excessively low dosages were prescribed, and the subsequent attempt to use higher dosages of multiple medications often resulted in unacceptable toxicity. In a comprehensive review from those times, Reynolds (3) was alarmed by the frequency and severity of side effects of AED therapy and identified polytherapy and use of high dosages as the main factors responsible for iatrogenic disease in patients with chronic epilepsy. This realization paved the way to a series of investigations in which the advantages of monotherapy became all too evident. It was fortunate that this development coincided with the availability of novel information on the importance of pharmacokinetic principles and on value of serum AED level monitoring, which allowed the use of individual anticonvulsants in a more efficient way.
In a landmark study, Shorvon and coworkers (4) assessed prospectively the value of phenytoin and carbamazepine monotherapy, assisted by serum level monitoring, in 51 patients with newly diagnosed partial or generalized tonic-clonic seizures. After a follow-up of 28 months (for phenytoin) or 12 months (for carbamazepine), 76% to 88% of these patients had their seizures completely controlled, a finding leading to the conclusion that “polypharmacy is largely, and possibly totally, unnecessary in newly diagnosed adult epileptics.” Many subsequent studies have confirmed that monotherapy is effective and well tolerated both in adults and in children. Although response rates vary greatly in relation to seizure type and syndromic form, between 50% and 90% of patients with newly diagnosed epilepsy can have their seizures fully controlled using one appropriate drug at individualized dosages (6,7, 8, 9, 10, 11, 12, 13). Other studies have shown that, even in patients with chronic refractory epilepsy, reduction of polypharmacy often can be achieved successfully without deterioration in seizure control, and with appreciable benefit in terms of a lessened burden of side effects (14).
Based on the foregoing evidence, no physician currently will question the principle that the treatment of epilepsy should be optimally started with a single drug (15,16). Realization of the many advantages of monotherapy (Table 8.1), however, should not lead one to consider combination therapy as an evil to be avoided at all costs. Indeed, it has been convincingly shown that not all patients can be successfully managed with monotherapy, and in some situations
the simultaneous use of more than one drug is necessary to obtain the best clinical response (17,18).
TABLE 8.1. ADVANTAGES OF MONOTHERAPY
The present chapter provides a critical overview of the current role of combination therapy in the treatment of epilepsy. An attempt is made to identify situations in which multiple drug therapy is indicated and to provide information about specific drug combinations that may be particularly useful, as well as combinations that should be preferably avoided. Finally, brief consideration is given to the potential implications of combining AED therapy with other medications that a patient may require for unrelated medical conditions, such as hypertension, infection, or the need for contraception. Because the use of multiple drug therapy involves the possibility of pharmacokinetic and pharmacodynamic interactions, the basic principles underlying the mechanisms, prediction, and management of such interactions are briefly discussed.
WHEN SHOULD COMBINATION ANTIEPILEPTIC DRUG THERAPY BE USED?
As discussed earlier, there is consensus that combination AED therapy should be reserved for those patients whose seizures cannot successfully be controlled with a single drug. However, this statement fails to identify the precise moment at which combination therapy should actually be introduced in the treatment algorithm. Should AED combinations be used in patients who fail to respond to maximally tolerated doses of one initially prescribed AED, or should these combinations be reserved for those patients whose seizures persist despite sequential use of two or more AEDs, each given as monotherapy? There seems to be considerable variation in the attitude of prescribing physicians toward these strategies. For example, in a survey in 14 Mediterranean countries, the proportion of neurologists opting for combination therapy (instead of using alternative monotherapy) when initial monotherapy had failed ranged from 23% in France to 67% in Syria: in many countries, including Turkey, Greece, and Italy, physicians were almost equally divided in their attitude toward early introduction of polypharmacy (19).
A review of evidence from prospective studies provides valuable clues regarding when combination drug therapy should be preferentially tried. In what is perhaps the most frequently quoted abstract in the history of epileptology, Hakkarainen (20) randomized a total of 100 patients with newly diagnosed convulsive seizures to either carbamazepine or phenytoin and found that, after 1 year, 50 patients continued to have seizures while they were receiving the allocated treatment. When these patients were switched to monotherapy with the alternative drug, 17 (34%) became seizure free. Of the 33 patients who were refractory to both phenytoin and carbamazepine as monotherapy, only five (15%) had their seizures controlled when the two drugs were tried together. These results clearly indicate that a substantial proportion of patients refractory to an initial drug can have their seizures controlled by switching to alternative monotherapy, and only a few patients do well with combination therapy. Although the design of this study may be criticized on the grounds that carbamazepine and phenytoin, sharing a similar mechanism of action and CNS side effect profiles, may not be ideal drugs to combine, other studies support the conclusion that alternative monotherapy has significant merits in patients refractory to a single drug. In a randomized comparison of vigabatrin and carbamazepine in patients with newly diagnosed partial epilepsy, 11 of 25 (44%) patients who failed to respond to initial monotherapy had their seizures fully controlled when they were switched to monotherapy with the alternative drug (21). Only five of the 14 patients refractory to two sequential monotherapies had their seizures controlled by the same two drugs in combination. In a larger, single-center observational study, 67 of 248 patients (27%) refractory to initial monotherapy were rendered seizure free with a second or third drug used as monotherapy, and only 12 had their seizures controlled by combination therapy (13). An interim analysis of an ongoing randomizing study comparing add-on therapy with alternative monotherapy also failed to show major outcome differences between these two strategies (22). In other studies in which patients refractory to initial treatment were switched to combination therapy (18,23, 24, 25,26), response rates are generally comparable to those described for patients managed with alternative monotherapy, but the burden of side effects tends to be greater in patients receiving more than one drug (6). Thus, it is clear that alternative monotherapy is associated with a significant probability of therapeutic success when the initially prescribed AED has failed, and therefore it should be the preferred strategy in these patients. Although it could be argued that the early addition (rather than the substitution) of a second drug will allow more rapid achievement of seizure control in the small subgroup of patients who do require combination therapy, such a policy would expose to a high risk of adverse effects many patients whose seizures could be managed with a single drug.
Based on the evidence discussed earlier, it appears reasonable to restrict the use of combination therapy to those patients in whom sequential therapies with at least two
appropriate AEDs, each prescribed at the maximally tolerated dosage (25), have failed. The value of adding a second and, sometimes, even a third drug in patients with longstanding refractory epilepsy is documented by many placebo-controlled add-on trials of newer AEDs (27, 28, 29, 30), even though one cannot exclude that, at least in some of these patients, an improvement in seizure frequency could have been obtained by simply increasing the doses of baseline medication. In general, between 20% and 50% of patients with refractory partial epilepsies or symptomatic generalized epilepsies are expected to benefit from AED combinations (18,29), although the actual proportion of those who will be free of seizures will be considerably smaller, typically less than 20%. Earlier, more aggressive use of combination therapy may be justified in occasional cases, for example, in patients with notoriously refractory epilepsy syndromes.
Although the benefit of combination therapy in a subgroup of patients with difficult-to-treat epilepsies cannot be questioned, one should be cautious about the risk of overtreatment (16,31). Use of more than one drug, especially when high doses are administered, leads to a greater burden in terms of side effects (32), and it is important to evaluate in the individual patient whether the price paid in terms of greater toxicity is justified by the improvement in seizure control. In patients with chronic epilepsy, seizure frequency fluctuates over time, and it is not uncommon for a second AED to be added during a period of spontaneous exacerbation: under these conditions, the subsequent improvement in seizure frequency may be related to spontaneous amelioration (the so-called regression to the mean), rather than to the effect of the added drug (33). Because of this situation, the need for maintaining combination therapy should be reassessed at regular intervals, and monotherapy should be reinstituted whenever appropriate. Moreover, in some patients, the addition of a second drug may cause a paradoxical increase in seizure frequency as a manifestation of drug toxicity (23,34): failure to recognize this may lead to a vicious circle whereby a further increase in AED load produces even less seizure control.
ARE SOME ANTIEPILEPTIC DRUG COMBINATIONS BETTER THAN OTHERS?
From a theoretical standpoint, combining two drugs may lead to additive, supraadditive (synergistic), or infraadditive effects. An AED combination would be desirable if it produces supraadditive antiepileptic efficacy in the presence of simply additive toxicity or additive efficacy in the presence of infraadditive toxicity. In a series of careful studies, Bourgeois and coworkers (35, 36, 37, 38, 39) looked at pharmacodynamic effects of various combinations of conventional AEDs in animal models and failed to provide evidence of definitely favorable interactions, except for valproate-ethosuximide (38) and, possibly, valproate-carbamazepine (39) combinations, which were associated with additive anticonvulsant activity and infraadditive toxicity. Results of studies of some of the newer AEDs suggest supraadditive anticonvulsant effects with some combinations (40,41), but the clinical relevance of these findings is difficult to assess, partly because we lack reliable models for predicting human neurotoxicity. Animal studies may be useful to generate hypotheses, but the ultimate demonstration of the usefulness of specific combinations can come only from the clinic.
Although it has been proposed that combining AEDs with different mechanisms of action should be therapeutically beneficial (42), in practice, our knowledge of the mechanisms of action of the various drugs, most of which have more than one primary action, is too incomplete to allow a rational application of this approach (27). Thus, AEDs are usually combined mainly on empirical grounds, by taking into consideration several general rules (Table 8.2). Clinical experience does suggest that some AED combinations may have a superior therapeutic index compared with others, the evidence being particularly convincing for valproate combined with ethosuximide in patients with refractory absence seizures (43) and for valproate combined with lamotrigine in patients with various refractory seizure
types (44, 45, 46, 47). With these combinations, pharmacodynamic mechanisms of interaction are assumed to be at play, although pharmacokinetic changes may simultaneous occur. In a trial that clearly illustrates the complexities of these interactions, Pisani and coworkers (47) evaluated prospectively 13 patients with refractory complex partial seizures who had failed to respond to maximally tolerated dosages of either valproate or lamotrigine given separately. When the two drugs were given together, four patients became seizure free, and an additional four patients experienced seizure reductions of 62% to 78%. Although the addition of valproate to lamotrigine initially produced an increase in serum lamotrigine levels, the appearance of side effects, especially tremor, required a reduction in dose of both medications, and seizure control was finally achieved at serum drug concentrations that were lower than those achieved before combination therapy was instituted. Potential advantages have been claimed for other combinations (Table 8.3), but the evidence is mostly anecdotal, and interindividual variation can be considerable.
TABLE 8.2. CRITERIA FOR SELECTION OF ANTIEPILEPTIC DRUG COMBINATIONS
TABLE 8.3. SOME ANTIEPILEPTIC DRUGS COMBINATIONS CLAIMED TO BE PARTICULARLY ADVANTAGEOUS
COMBINATION THERAPY AND DRUG INTERACTIONS
As discussed earlier, concomitant use of medications leads to the possible occurrence of drug interactions. Traditionally, these are classified into two groups: (a) pharmacokinetic interactions, which involve a change in the absorption, distribution, or elimination of the affected drug; and (b) pharmacodynamic interactions, which are thought to result in a change in pharmacologic response at the site of action. Interaction may occur between or among two or more AEDs, when they are used in combination therapy, or between AEDs and other drugs used for unrelated conditions. The following sections provide a brief overview on some important issues related to the mechanisms and management of drug interaction in patients with epilepsy.
Unlike pharmacokinetic interactions, which usually involve an easily measured change in the blood concentration of the affected drug, pharmacodynamic interactions manifest themselves by a change in pharmacologic response that may be difficult to characterize objectively. This situation explains why these interactions have been investigated incompletely, even though their clinical relevance is probably considerable. As mentioned earlier, two drugs may interact pharmacodynamically, leading to additive, supraadditive, or infraadditive effects, at the level of both therapeutic response and toxicity (38,39).
Pharmacodynamic AED interactions may be adverse, neutral, or beneficial. Possibly favorable interactions are described earlier in the discussion of the preferential use of certain AED combinations. Adverse pharmacodynamic interactions, possibly explained by additive neurotoxicity (48), most often involve the appearance of CNS side effects in patients receiving polytherapy, even when the doses and blood levels of individual AEDs are in the low range. Just as some combinations seem to produce better therapeutic effects than others (43, 44,45, 46, 47,49, 50, 51, 52, 53, 54), evidence also indicates that certain specific combinations may be more likely to cause tolerability problems. For example, several studies have documented that the addition of lamotrigine to carbamazepine often results in symptoms suggestive of carbamazepine toxicity: this was initially explained as the possible result of a rise in serum carbamazepine-10,11-epoxide levels, but more recent studies indicate that the levels of this metabolite are unaffected by lamotrigine, and an adverse pharmacodynamic interaction is responsible for these manifestations (55).
Knowledge of the mechanisms underlying pharmacokinetic drug interactions expanded in the 1990s, and consequently
it has become possible to develop a rational approach to their prediction. These interactions have been traditionally divided in three categories: interactions based on absorption, those based on distribution, and those based on elimination or metabolism. For AEDs, little information exists on interactions occurring within the lumen of the gastrointestinal tract. At the level of distribution, there is a significant body of literature on protein binding displacement, probably because of a lingering misconception that increases in the unbound fraction of a drug would result in increases in its unbound concentration and possibly enhanced toxicity. Eventually, it became apparent that AEDs have low extraction ratios, and thus changes in unbound fraction do not affect unbound concentrations. At the level of metabolism, significant qualitative predictions of changes in concentration of an AED became possible with a knowledge of the hepatic (and intestinal, if appropriate) enzymes responsible for its metabolic clearance (56). To characterize the effects of an AED on other drugs, it is necessary to define its inhibition and induction spectra.
ANTIEPILEPTIC DRUGS AS SUBSTRATES
Except for gabapentin, levetiracetam, and vigabatrin commonly used AEDs are metabolized by microsomal enzymes such as cytochromes P450 and glucuronyl transferases. Carbamazepine is a substrate for cytochromes CYP3A4 and some CYP2C isoforms, and this information is sufficient to explain increases in carbamazepine levels associated with coadministration of macrolide antibiotics (e.g., triacetyl oleandomycin, erythromycin, clarithromycin, roxithromycin, josamycin, azithromycin, flurithromycin, ponsinocycin, and spiramycin), diltiazem, verapamil, ketoconazole, danazol, propoxyphene, fluoxetine, fluvoxamine, and viloxazine (57). Induction of those isoforms (CYP3A4 and possibly CYP2Cs) explains the decreases in carbamazepine levels caused by phenytoin, phenobarbital, primidone, felbamate, and rifampicin.
Most of the fate of phenytoin is controlled by two enzymes, CYP2C9 and CYP2C19, and most of the drugs that elevate phenytoin levels are known inhibitors of one or both enzymes: amiodarone, phenylbutazone, miconazole, and sulfonamides inhibit CYP2C9, and cimetidine, felbamate, fluoxetine, omeprazole, ticlopidine, viloxazine, oxcarbazepine and topiramate inhibit CYP2C19, whereas fluconazole inhibits both enzymes (58). CYP3A4 contributes to the elimination of tiagabine and zonisamide (59,60), and induction of this isozyme explains the increase in the clearance of these drugs in patients receiving concomitant treatment with phenytoin, carbamazepine, phenobarbital and primidone. Although only a small fraction of the dose of topiramate is eliminated by oxidative metabolism in healthy subjects, induction of metabolism explains why the total clearance of topiramate is increased twofold when this drug is coadministered with phenytoin (61).
A glucuronosyltransferase, UGT1A4, plays an important role in the metabolism of lamotrigine, and most of a lamotrigine dose is recovered in urine in the form of a 5-N-glucuronide and a 2-N-glucuronide. This enzyme is inhibited by valproate, and it is induced (possibly with other glucuronosyltransferases) by phenytoin and carbamazepine. Coadministration of valproate is associated with appreciable elevations in lamotrigine levels, and, conversely, coadministration of inducers such as phenytoin, carbamazepine, or phenobarbital results in decreases in lamotrigine concentrations (62). Both types of interactions usually require dosage modifications.
Similarly, a significant portion of valproate clearance is controlled by glucuronidation, and this explains the decreases in valproate levels associated with phenytoin, carbamazepine, and phenobarbital (63). Oxcarbazepine is essentially a prodrug for the active 10-hydroxy metabolite resulting from reduction by a cytosolic enzyme. This metabolite is primarily glucuronidated, and its levels are also reduced in the presence of enzyme inducers (64).
ANTIEPILEPTIC DRUGS AS INHIBITORS
Some evidence indicates that oxcarbazepine topiramate, felbamate, valproate, and carbamazepine behave as weak inhibitors of the metabolism of phenytoin through the enzyme CYP2C19. This inhibition results in relatively small increases in phenytoin levels with appreciable intersubject variability.
Valproate is also a mild inhibitor of microsomal epoxide hydrolase, and its coadministration with carbamazepine is associated with elevations in carbamazepine-10,11-epoxide (65). The interaction of valproate with phenobarbital is associated with elevations in phenobarbital levels, probably caused by inhibition of uridine diphosphate glucosyltransferase (66). As mentioned earlier, another valproate inhibition interaction that requires attention is based on inhibition of UGT1A4 (and possibly other glucuronyl transferases), resulting in elevations in lamotrigine levels that usually require dosage adjustments (62).
ANTIEPILEPTIC DRUGS AS INDUCERS
In the field of drug interactions, phenobarbital, phenytoin, and carbamazepine rank among the most potent inducers of drug metabolism. They affect the disposition of many drugs, principally those that are substrates of CYP3A4 and CYP2C enzymes, as well as drugs metabolized by glucuronidation. The administration of phenobarbital, phenytoin, and carbamazepine is usually associated with decreases in levels of carbamazepine (heteroinduction and autoinduction) and other CYP3A4 substrates such as cyclosporine, oral contraceptives and corticosteroids. Oxcarbazepine,
topiramate, and felbamate have a narrower spectrum of enzyme inducing activity, but they have also been found to reduce the blood levels of steroid oral contraceptives. As indicated earlier, the two AEDs that are eliminated principally by conjugation, valproate and lamotrigine, are induced by phenobarbital, phenytoin, and carbamazepine.
ROLES OF TRANSPORTERS
In the last few years, evidence has accumulated that transporters play a critical role in the disposition of some drugs and are involved in drug interactions. For example, P-glycoprotein, which had been initially characterized for its role in multidrug resistance in cancer chemotherapy, has been recognized to interact with many drugs (67). It functions as an efflux pump located in several tissues (intestinal wall, liver and biliary system, blood-brain barrier, kidney, and placenta). Digoxin is a substrate for P-glycoprotein, and inhibition of this pump by ketoconazole or quinidine is the probable mechanism for digoxin toxicity associated with these drug combinations. The roles of transporters in the disposition of AEDs is just emerging.