Abstracts

Rationale: To investigate remission and relapse in a refractory epilepsy population followed prospectively. Methods: In 2000 we identified a cohort of 246 patients at the University of Pennsylvania Epilepsy Center who met a pre-specified definition of drug refractory epilepsy (failure of at least two antiepileptic drugs and seizure frequency greater than once per month). Median age at entry was 40, mean duration of epilepsy was 25 years, and mean duration of intractability was 20 years. We have now followed this cohort for over six years. Follow-up methods included chart review, and for those not seen within the last year, additional attempts at direct contact using last known phone number. We used Kaplan-Meier methods to estimate the cumulative risk of achieving a 12-month seizure remission and subsequent relapse. Cox regression analysis was used to evaluate clinical predictors for seizure remission and relapse. Results: At 7 years, the estimated cumulative incidence of 12-month seizure remission was 35% (95% confidence interval, 26-44%) for all cases and 33% (95% confidence interval, 25-44%) when limited to those treated only with medication. For the subset of patient who underwent surgery, the cumulative risk for 12-month seizure remission at 4 years was 45% (95% confidence interval, 16-41%). The cumulative risk of relapse in the entire cohort was 56% (95% confidence interval, 42-70%) by 2 years after achieving a one year remission and 71% (95% confidence interval, 55-86%) by 5 years. Negative predictors of remission included mental retardation, symptomatic generalized epilepsy, duration of intractability, and number of antiepileptic medications failed. The only statistically significant negative predictor of relapse was localization related epilepsy. Drug discontinuation was not a predictor of relapse. Almost half of the subjects who relapsed had two seizures or fewer. Conclusions: Among patients with refractory epilepsy, the cumulative risk for remission continues to be substantial over a long period of follow up (5% per year over 7 years). Following remission, the cumulative risk of relapse in this population is also significant with a large proportion occurring even after the first year of remission. Interestingly, approximately four fifths of relapses after a significant remission occur within the first two years after achieving a one year seizure remission. Epileptologists should be cautious when counseling drug refractory patients about possible cure or when recommending return to driving.