Abstracts

Rationale: The magnitude of the placebo (PBO) effect exhibits variability in randomized clinical trials, and PBO rates have been gradually increasing over time in therapeutic areas including epilepsy.¹ To evaluate the direct treatment effect of an antiepileptic drug (AED), and to make the treatment effect of AEDs comparable across studies, it may be useful to report the PBO-adjusted estimate of treatment effect. Furthermore, because the distribution of seizure frequency outcome data is generally skewed, statistical analysis on skewed data should be considered in the PBO-adjusted estimation. Here we provide details regarding the PBO-adjusted estimation for the 3 double-blind, randomized, PBO-controlled Phase III studies assessing the efficacy of perampanel (PER). Methods: Patients with refractory partial-onset seizures enrolled in the 3 Phase III studies were aged ≥12 yrs and receiving 1-3 concomitant AEDs. Following 6-wk baseline, patients were randomized to 19 wks of once-daily double-blind treatment (6-wk titration, 13-wk maintenance) with PBO or PER 8 or 12mg (Studies 304&305); or with PBO or PER 2, 4 or 8mg (Study 306). These study results were pooled (subjects: PER n=1038; PBO n=442). Efficacy data included median percent reduction from baseline in seizure frequency/28 days. Seizure frequency was also analyzed in patients based on the use of concomitant enzyme-inducing AEDs (EIAEDs). The PER US Prescribing Information (PI) presents the pooled efficacy data of these 3 clinical trials as PBO-adjusted. Hodges-Lehmann (HL) method was used to estimate the median difference between PBO and PER. The HL estimate is a nonparametric method based on ranks for estimating group difference under skewed distribution. The HL estimate and corresponding 95% CI were estimated for each PER dose group. Results: The median percent reduction in seizure rate for PBO group was 21.0%, 9.7% and 10.7% in the 3 studies. This variability suggests that analysis of the PBO-adjusted estimate of treatment effect is warranted. The absolute and PBO-adjusted median percentage of decrease in seizure rate of PER at doses of 2-12mg are depicted in Figures 1A and 1B, respectively. At therapeutic doses (4-12mg), PER significantly decreased seizure frequency using both analyses. The effect of concomitant EIAEDs on the efficacy of PER in the pooled analysis revealed a reduced effect on median percent reduction in the presence of EIAEDs. The results were confirmed in PBO-adjusted analysis. Conclusions: The PBO-adjusted method is useful to directly assess the efficacy of a drug, particularly when assessing efficacy for multiple studies. PER (4-12mg) shows efficacy when median percent reduction in seizure frequency is viewed in absolute numbers as well as PBO-adjusted. Ref: ¹Guekht AB. Epilepsy Behav 2010;17:64. Support: Eisai Inc.