How reliable is early treatment response in predicting long-term seizure outcome?

Assessing if an early therapeutic response to an antiepileptic drug (AED) is of prognostic value for long-term outcome is of considerable clinical interest. To that end, we evaluated the likelihood that a patient who becomes seizure-free after 6 months of treatment with a single AED would lose that response at 12 months, or vice versa. In a post hoc analysis, data from five comparative, double-blind, single-drug studies designed to evaluate the efficacy of treatment of patients with partial seizures with oxcarbazepine versus carbamazepine, phenobarbital, phenytoin, and valproate for approximately 1 year were pooled to investigate same-patient seizure outcome at 6 and 12 months. These studies had similar dosing regimens and included a wide range of patients with newly diagnosed and chronic epilepsy. The main finding in the population included in the analysis is that those patients who are seizure-free at 6 months have a 90% chance of being seizure-free at 12 months, whereas those not seizure-free at 6 months have only a 45% chance of being seizure-free at 12 months (chi(2)=118.716, P<0.000001, odds ratio=11.23 with 95% confidence limits 6.8-18.7). In a worst-case assessment, those not seizure-free at 6 months have only an 18% chance of being seizure-free at 12 months (chi(2)=408.105, P<0.000001, odds ratio=41.23 with 95% confidence limits 26.4-65.85). Failure to maintain the response in 10% of patients, including 4% with two or more seizures, was noted with all AEDs studied here and in patients with newly diagnosed as well as chronic epilepsy. Among patients with seizures in Months 1-6, those with newly diagnosed epilepsy became seizure-free more often over time than those with chronic epilepsy. The main conclusion is that response at 6 months is an excellent predictor of response at 12 months.