Effects of Antiepileptic Drugs on Cognition

This article reviews our knowledge about a specific subgroup of central nervous system (CNS)-related chronic side effects of treatment with antiepileptic drugs (AEDs) (i.e., cognitive side effects). Although the severity of cognitive side effects is considered to be mild to moderate for most of the AEDs, their impact may be substantial in some patients when critical functions are involved, such as learning in children or driving capacities in adults (often requiring milliseconds precision), or when functions are impaired that are already vulnerable, such as memory function in the elderly. The interest in the cognitive side effects of AED treatment is of recent origin, and the first studies are from the early 1970s (1,2), probably stimulated by the widening range of possibilities for drug treatment during that period. In the period January 1970 through December 1998, 1,357 articles on cognitive effects of AEDs were published in peer-reviewed journals. This database (3) is used to give an overview of the effects per drug. We decided, however, to disregard some of the studies in this database, using basic methodologic criteria mentioned in our review (3). We first concentrated on the drugs that are commonly used in clinical practice: phenobarbitone (PB), phenytoin (PHT), carbamazepine (CBZ), and valproate (VPA). With respect to these drugs, three types of data were evaluated: Absolute drug effects, which are the effects of drug treatment against no-treatment (nondrugs) in the same subjects, which represents the most valuable information. Relative drug effects (i.e., the comparison of the cognitive side effects of a drug with the effects of another drug). Although this is relevant clinical information, this comparison does not rule out the possibility that two drugs are not different because they both impair cognitive function to the same extent. Dose effects (i.e., the relationship between cognitive side effects and dosing of an AED). Finally we inspected the available data on the newer AEDs: oxcarbazepine (OCBZ), lamotrigine (LTG), gabapentin (GBP), vigabatrin (VGB), tiagabine (TGB), topiramate (TPM), levetiracetam, and rufinamide. For PB, only one study (4) is available (after applying our criteria) allowing the evaluation of absolute effects (i.e., the differences between PB and a nondrug condition). This study shows serious memory impairment (short-term memory recall) in 19 patients with epilepsy. For PHT, five studies are available (5–9) comparing PHT with a nondrug condition. These studies all reveal PHT-induced cognitive impairment in the areas of attention, memory and especially mental speed. The magnitude of the reported effects is moderate to large. A caveat is, however, that all these studies were carried out in normal volunteers, which opens the possibility that these effects represent short-term outcomes of the drug. For CBZ, there is no consistent report about absolute cognitive effects. Two studies, one in normal volunteers (6) and one in patients with epilepsy (10) report “no cognitive impairment” compared with a nondrug condition. This is challenged by the group of Meador et al. (8,9) that reported mild impairments of memory, attention, and mental speed, largely the areas that may also be affected by PHT. For VPA, four studies (11–13) allow the interpretation of absolute effects and show mild to moderate impairment of psychomotor and mental speed. For PB, comparisons with other AEDs are available from four studies (14–17), all with patients with epilepsy. One of these shows more impairment for PB than for PHT or CBZ on visuomotor and memory tests (17), and two other studies show convincing and clinically highly relevant impairments of intelligence scores after long-term PB treatment in comparison with VPA (14,15). Only the study by Meador et al. (1990) (16) does not show differences between PB and PHT or CBZ. For PHT, the results of head-to-head comparisons are somewhat more confusing. Using an ingenious long-term treatment and withdrawal design Gallassi et al. (17) found more cognitive impairment than that with CBZ. Conversely, no differences with CBZ, VPA, and even with PB are reported (8,9,16,18). For CBZ, again we have to consider conflicting results of the Italian study by Gallassi et al. (17), showing a more favorable profile compared with PHT and PB, and the United States–based study by Meador et al. (8,9,16) that showed no differences with PHT and PB. Finally, for VPA, the comparison with other drugs shows lower performance of memory and visuomotor function compared with CBZ (17) and a favorable profile compared with PB on tests for intelligence (14,15). One study does not show a difference with PHT (18). The cognitive side effects of AEDs do not develop only at higher doses: there is indeed no significant correlation between dose and cognitive impairment for PHT in a large cohort study (19) and nor for VPA (20,21). This is in accordance with the finding that, generally, side effects of VPA are not related to dose or serum level (22). This also concurs with the absence of any differences between conventional and controlled-release formulations for VPA (23). Only for CBZ are some effects of dose reported, with improvement of cognitive function at a higher dose (24) and with the use of controlled-release formulations (25). One study (26) evaluated the effects of switches between different generic formulations of CBZ, revealing no large effects. No information is available for PB. The relatively short period of introduction of most of the newer AEDs illustrates that we cannot expect many cognitive studies with these AEDs that fulfil the aforementioned criteria. Most of the available data for these drugs were obtained in add-on polytherapy designs, the gold standard type of design to test the efficacy of a new drug in early clinical studies. It is remarkable that no normal-volunteer studies are available for any of these drugs. Nonetheless we may summarize the available information. Most of the information presented here is based on the exchange of data during the international workshop Cognitive Effects of the Newer Antiepileptic Drugs, organized during the International Epilepsy Congress in Sydney, 1995 (27). For vigabatrin (VGB), the absence of cognitive side effects (compared with the existing first-line drugs) is claimed, but the data come from only one center: Riekinnen et al. in Finland (28,29), and are not based on high-powered studies. A few additional studies did, however, not pass our aforementioned criteria, mostly because they were carried out in polytherapy designs. In clinical practice, the drug does not seem to have a large difference from the cognitive profile of CBZ or VPA. Mood effects may occur, however (30), secondarily also affecting cognitive performance. Anecdotal clinical information suggests no cognitive impairment for lamotrigine (LTG), and in some patients, even improvement of performance is reported (31), which may be in line with the claimed psychotropic effect of the drug. There is, however, no empiric evidence from controlled cognitive studies to support this claim. Marciani et al. (32) studied the cognitive effects of LTG in 11 patients in an uncontrolled study. There is also some anecdotal information in the report by Meador and Baker (1997) (33). Finally, the study by Martin et al. (34) did not show cognitive effects at short term, but this was a low-powered study. None of these studies is conclusive. The claim of absence of cognitive effects with oxcarbazepine (OCBZ), a compound related to CBZ, is based on two studies (35,36). The latter study reported cognitive improvement (focused attention and speed) in 12 patients in comparison with a nondrug condition, but did not control for the beneficial cognitive effects of improved seizure control. For tiagabine (TGB), a favorable cognitive profile is reported. Dodrill et al. (37) showed no cognitive effects in monotherapy with TGB in low or high doses, but some evidence for mood effects in add-on treatment with TGB at higher dosing, possibly related to titration speed. In the polytherapy study by Kalviainen et al. (38), no cognitive effects were found. With gabapentin (GBP), Leach et al. (39) studied 21 patients in an add-on polytherapy study after 4 weeks of adjunctive therapy and found no change in psychomotor and memory tests. Drowsiness was more often found in higher dosing (2,400 mg). Mortimore et al. (40) did not find a difference between continued polytherapy or an add-on with GBP in measures of quality of life. Martin et al. (34) used an acute dose and rapid titration in six volunteers and did not find cognitive effects of GBP. These studies do not show convincing evidence for cognitive effects of GBP but are not conclusive, as they were performed in polytherapy designs that do not allow conclusions about a specific drug. For topiramate (TPM), recent data are available. In the study by Martin et al. (34), an acute dose of 200 mg TPM and rapid dose escalation to 400 mg in 4 weeks were used. Interestingly, this was very reminiscent of the dose escalation used in the initial randomized clinical trials of TPM, which was associated with somnolence, psychomotor slowing, speech disorders, and concentration and memory difficulties. Martin et al. showed neuropsychometric changes commensurate with these CNS effects. In a randomized controlled trial using dose escalations of 50 mg weekly, Meador (41) reported a much more limited effect of TPM on cognitive function. Of a test battery of 23 variables, four variables measuring attention, vigilance, and word naming revealed only mild to moderate changes. The sample size of the Meador study adds to its validity: 155 patients with epilepsy. In a recent study comparing TPM with VPA (42), 25-mg increments per week were used. In this study, the preliminary analyses showed no difference between VPA and TPM on the major cognitive variables after 20 weeks of treatment. This demonstrates that TPM entails a risk for cognitive impairment that can be prevented using gradual titration. For the drugs that are still in the experimental phase, we have information about rufinamide (CGP 33.101) showing improvement of cognitive function in a lower dose (improvement on reaction-time tests) and a possibility of impairment of short-term memory at a higher dose. Full data are not yet available, although the study had an impressive sample size (>200) (43). For (UCB L059) levetiracetam, a nootropic effect is claimed. We only have data from a small pilot study that does not allow definite conclusions (44). All established AEDs have been reported to be associated with absolute cognitive side effects (i.e., all the investigated drugs have effects when compared with no treatment). These effects are definitely large for PB and possibly larger for PHT than for CBZ or VPA. But even these last two drugs, generally considered to be drugs with a safe cognitive profile, have cognitive effects, mostly resulting in a mild general psychomotor slowing. The respective differences between the four investigated AEDs can be considered as small, with the exception of the cognitive effects of PB, which has positively a less favorable cognitive profile when compared with PHT, VPA, and CBZ. Clinical anecdotal information thus far does not show large differences between the newer AEDs and the established drugs.