Cancer among epileptic patients exposed to anticonvulsant drugs.

Cancer incidence among 8,004 patients hospitalized for epilepsy between 1933 and 1962 in the Filadelfia treatment community in Denmark was compared to that of the general population. Patients received powerful and prolonged treatment with phenobarbital, phenytoin, and other anticonvulsants. This new survey extends the follow-up from 1976 through 1984. Among 7,864 patients with epilepsy not known to have received radioactive Thorotrast, record linkage with national cancer incidence files identified 789 cancers, compared to 664 expected [relative risk (RR) = 1.19; 95% confidence interval = 1.11-1.27]. Significant risks were found for cancers of the brain and central nervous system (RR = 5.7; n = 118) and the lung (RR = 1.4; n = 106). The excess numbers of brain cancer were concentrated within 10 years of hospitalization (RR = 20.7; n = 80) and decreased significantly over time, which suggests that brain tumors account for the seizure disorder and are not due to phenobarbital exposure as suggested by some epidemiologic studies. No overall risk was apparent when brain cancers were excluded (RR = 1.03). Because bladder cancer was significantly decreased (RR = 0.6; n = 18), the excess risk of lung cancer may not have been related to the "anecdotal" heavy smoking reported among confined groups of epileptic patients in the early years of the study period. The incidence of malignant melanoma was also significantly low (RR = 0.5; n = 7), which suggested limited exposure to sunlight among confined patients. The risk of non-Hodgkin's lymphoma was increased, but not significantly (RR = 1.4; n = 16), which is interesting in view of previous reports suggesting an association with phenytoin. Overall, these data provide little evidence that phenobarbital and phenytoin are carcinogenic to humans, but the excess risks of lung cancer and non-Hodgkin's lymphoma among epileptic patients in our study deserve further evaluation.