One of the most important obstacles in defining the progressive cognitive and behavioral decline of epilepsy is the need to separate toxicity of antiepileptic drugs (AEDs) from the effects of seizures. Studies that demonstrate stable or improving cognitive function over time may actually reflect benefits gained from seizure control. The AED that provides better seizure control for an individual may be more likely to provide a cognitive or behavioral benefit than another AED that has a statistically better cognitive-behavioral profile in controlled clinical studies.115

Cognitive benefits

The cognitive benefits of treating epilepsy are supported by several studies. Seidenberg et al.27 showed increased Wechsler verbal, performance, and full-scale IQ measurements in patients with improved seizure control but no change in patients with poor seizure control. In another study, neuropsychological testing showed that unmedicated epilepsy patients performed significantly worse than normal controls on fine motor, attention, and cognitive tasks.116

Cognitive effects of toxicity

Bourgeois et al.,23 however, found that toxic drug levels and early age of onset were the factors most closely correlated with poor cognitive performance. Seizure control, therefore, should not be achieved at the price of repeated episodes of drug toxicity. Another study showed that treatment with AEDs, especially phenobarbital, caused decreased Wechsler Intelligence Scale for Children–Revised scores.117

Effects of individual AEDs

Debate concerning the precise cognitive and behavioral effects of different AEDs is far from resolved. In many studies, differences among drugs may reflect differences among the patients taking them rather than differences in the drugs themselves. Although study groups may appear similar, important intrinsic differences may exist between them that can account for different test scores. Dodrill47 noted that when two equally efficacious AEDs are prescribed for two patients with the same seizure type, differences between the patients likely account for different usage of medication. These patient differences may include

• ability to deal with complex AED regimens
• general intelligence
• financial status versus cost of medication
• emotional factors
• distance from the clinic where serial laboratory testing is conducted

In the Holmfrid study, 100 children with epilepsy were seizure-free for 1 year on monotherapy with carbamazepine, phenytoin, or valproic acid.118 The AEDs were withdrawn over a 3-month period, and the children were reevaluated 3 to 4 months later. Significant improvement attributable to drug withdrawal was noted on only the psycho- motor speed test, suggesting a limited role for AEDs on cognitive function. This study showed, however, that phenytoin patients experienced greater cognitive impairment on tests of motor and mental speed than did carbamazepine patients.

Another drug withdrawal study involved children treated with carbamazepine, valproate, or phenytoin as monotherapy.119 Withdrawal of the AED produced improvement in binary choice and visual search tasks, but similar improvements were noted in the normal control group. The performance of the children taking carbamazepine was similar to that of the children in the control group both before and after drug withdrawal. Children taking phenytoin, however, performed poorly both before and after drug discontinuation. A similar impairment, but to a lesser degree, was noted with the valproate group.

Other findings from studies of individual AEDs:

• Phenytoin impairs memory and cognition, and this impairment resolves on drug withdrawal.126
• When tested after 12 months of treatment, subjects taking valproic acid had higher cognitive scores than those taking carbamazepine or ethosuximide.120
• Carbamazepine was less likely to cause cognitive impairment than was phenobarbital.121
• Children taking phenobarbital had lower full-scale and performance IQ scores than did those on valproate.122
• Baseline WAIS full-scale IQ and WAIS performance IQ scores were lower for a group taking phenobarbital than for a group taking valproate. The scores improved over time for the valproate group but not for the phenobarbital group123.
• Increased phenytoin and phenobarbital and decreased folic acid were the most important correlates of cognitive dysfunction.41
• Patients taking phenobarbital performed less well on digit symbol testing but on no other neuropsychological tests when compared to those taking carbamazepine.124
• Patients taking phenobarbital for 8 to 12 months had no significant differences in IQ from those taking placebo, but memory was affected in a serum-dependent manner and comprehension was negatively affected with longer duration of treatment.125

Issues in assessing cognitive side effects

The role of polytherapy is another consideration in assessing cognitive side effects of AEDs. Studies show that reduction of a multidrug regimen to monotherapy has beneficial effects on cognitive functioning and mood.127 However, studies that suggest greater cognitive impairment with polypharmacy usually involve conversion to monotherapy of groups of patients doing poorly on polytherapy.115 Dodrill47 found that seizure frequency is positively correlated with the number and amount of prescribed AEDs, so that polypharmacy is used to treat sicker patients.

Several methodologic problems are present in the study of cognitive side effects of AEDs128:

• lack of prospective studies
• wide cognitive fluctuations in epilepsy
• study of acute rather than chronic effects of the drug
• use of normal volunteers rather than patients with seizures
• attribution of cognitive change to AEDs by default
• testing of multiple drugs simultaneously
• comparison of drugs at noncomparable doses
• testing of performance postictally
• emphasis on statistical significance over clinical relevance
• attempts to analyze more factors than is warranted by sample size

These problems must be addressed before the role of AEDs in cognitive decline can be definitively stated.

Adapted from: Devinsky O and Tarulli A. Progressive cognitive and behavioral changes in epilepsy. In: Devinsky O and Westbrook LE, eds. Epilepsy and Developmental Disabilities. Boston: Butterworth-Heinemann; 2001;133–149. With permission from Elsevier (www.elsevier.com).
Reviewed and revised May 2004 by Steven C. Schachter, MD, epilepsy.com Editorial Board.

Back to top