As a group, individuals with epilepsy have impaired cognitive performance in comparison to healthy subjects matched for age and education (1); however, considerable intersubject variability does exist. Most persons with epilepsy have intelligence in the normal range, and some have superior cognitive abilities. Various factors can have a detrimental effect on cognition in epilepsy, including (a) the cause of seizures; (b) cerebral lesions acquired prior to onset of seizures; (c) seizure type; (d) age at onset of epilepsy; (e) seizure frequency; (f) duration and severity of seizures; (g) physiologic dysfunction (intraictal, interictal, or postictal) resulting from seizures; (h) structural cerebral damage as a consequence of repetitive or prolonged seizures; (i) hereditary factors; (j) psychosocial factors; (k) sequelae of epilepsy surgery; and (l) untoward effects of antiepileptic drugs (AEDs) (2,3).

The etiology of seizures may be one of the strongest factors influencing cognitive abilities (4). Patients with seizures attributable to progressive cerebral degeneration usually exhibit dementia, those with mental retardation have an increased incidence of epilepsy, and those with seizures caused by a focal brain lesion may exhibit a specific neuropsychological pattern of deficits. In contrast, patients with idiopathic epilepsy are more likely to have normal intelligence (4). Seizure type may be strongly associated with cognition (5). Patients with juvenile myoclonic epilepsy usually have normal intelligence, but children with infantile spasms have a poor prognosis. In general, the earlier the age of seizure onset, the more likely it is that a patient will have cognitive impairment. Additionally, patients with mental retardation are more likely to have refractory epilepsy (5,6).

Seizure frequency, duration, and severity may affect cognition in several ways (3,7). Obviously, cognition is impaired intraictally when consciousness is altered during generalized or complex partial seizures. Epileptiform discharges and postictal suppression may impair cognition interictally (8,9). Recent temporal lobe seizures impair consolidation of memory (10). Classic postictal Todd paralysis lasts less than 24 hours, but postictal cognitive dysfunction, such as dysphasia, may persist for several days. Chronic physiologic dysfunction may also exist beyond the area of epileptogenesis. For example, positron emission tomography scans reveal interictal hypometabolism extending to the lateral temporal cortex in patients with epilepsy caused by mesial temporal lobe sclerosis (11). Repetitive or prolonged seizures may permanently damage the cerebral substrate via anoxia, lactic acidosis, or excessive excitatory neurotransmitters. Even temporal lobe seizures of relatively modest frequency over several decades can increase the severity of hippocampal atrophy and reduce cognitive abilities (12,13).

Factors indirectly related to epilepsy may also affect cognition. Hereditary factors strongly influence intelligence. In fact, maternal intelligence quotient (IQ) is the most influential factor overall in predicting a child’s intelligence (14).
Psychosocial factors may adversely affect cognition through such mechanisms as depression or restriction of environmental influences (15). Finally, surgical or pharmacologic treatment of seizures may produce adverse cognitive effects.

Epilepsy surgery does not usually cause any general cognitive decline because dysfunctional tissue is primarily removed (16). Surgery may even result in improved cognition because of the reduction in seizures and AEDs. However, clinically significant postoperative cognitive deficits may occur. For example, left temporal lobectomy may lead to declines in naming and in verbal memory. However, the risks are largely predictable (17,18). Risks are greater if age of epilepsy onset is later or if hippocampal gliosis/atrophy is not present. Verbal memory is at greater risk following left temporal lobectomy if baseline verbal memory is high or functional assessments suggest greater residual preoperative function of the left temporal lobe. Thus, a patient undergoing left temporal lobectomy is at increased risk if the patient has high baseline verbal memory and high memory performance with right intracarotid amobarbital injection and low with left injection. In contrast, a decline in visuospatial memory is inconsistent following right temporal lobectomy. Rarely, unilateral temporal lobectomy has resulted in a severe global anterograde memory disorder. Fortunately, modern advances in preoperative evaluation techniques have minimized this risk. In addition, selective amygdalohippocampectomy may reduce the risk for memory loss compared with standard anterior two-thirds temporal lobectomy (19). In contrast to risks, patients who become seizure free from epilepsy surgery have a significant improvement in their emotional well-being and perceived quality of life (QOL) (20).

AEDs reduce neuronal irritability and thus may reduce neuronal excitability and impair cognition. Because AEDs are the major therapeutic intervention in epilepsy, their cognitive effects are of particular concern to physicians, who must consider the risk-to-benefit ratio of any treatment. Therefore, differentiating the cognitive effects of AEDs and placing them in the proper perspective are important.

Although all AEDs may impair cognition, such side effects are usually modest, as assessed by neuropsychological tests in patients on monotherapy in whom anticonvulsant blood levels are within standard therapeutic ranges (25). Furthermore, the cognitive effects may be partially offset by the reduction in seizures. It is clear that the risk of cognitive side effects rises with polypharmacy and with increasing AED dosages and anticonvulsant blood levels (26). Decreasing the number of AEDs frequently improves cognition and may reduce the number of seizures (27). However, the best drug regimen for an individual patient is the one that best controls seizures with the fewest side effects, and for some patients this regimen may involve polytherapy. Despite the modest cognitive effects of AEDs on formal neuropsychological testing, these effects can be clinically pertinent, as evidenced by the highly significant inverse correlation of neurotoxicity symptoms and QOL scores (28). Despite the absence of overt toxicity on examination, patients who exhibit more symptoms of neurotoxicity have lower perceived QOL.

The differential cognitive effects of AEDs are controversial and certain agents are promoted as being superior in this regard. For the older AEDs, more consistent adverse effects are observed with barbiturates and benzodiazepines, but results are mixed for carbamazepine, phenytoin, and valproate (25,26). Several of the newer AEDs appear to have fewer cognitive side effects than the older AEDs, but the effects of the newer AEDs relative to each other and to older AEDs are not yet fully determined.