Abstracts

Rationale: Randomized, placebo-controlled trials are considered the gold standard of contemporary studies of anti-epileptic drugs (AEDs). These studies typically have a 4- to 8-week baseline (BL) period and 12- to 14-week fixed-dose treatment period. The standard primary analysis in these studies is change from BL in seizure frequency. In patients with epileptic encephalopathies, placebo-controlled trials can be problematic given the long BL period and long duration of exposure to placebo with limitations on altering current medications. A novel time-to-event (TTE) study design has been proposed with the aim of minimizing placebo exposure and overall study duration, while still being sensitive enough to detect a treatment effect. 'Time to BL seizure rate' has previously been tested in several post-hoc analyses. A TTE study design also allows for the BL period to be reconsidered, including the potential to vary the BL duration compared to the traditional design. Study 1 was a positive, double-blind, placebo-controlled study of fenfluramine (FFA, Fintepla, Zogenix) in patients with Dravet syndrome (DS) that followed the 'gold standard' study design. As proof of concept for a 'time to BL seizure' (TTE) design with a short BL, we re-analyzed Study 1 using a 7-day BL period with a TTE analysis. Methods: In Study 1, subjects 2-18 years old with a diagnosis of DS and in whom convulsive seizures (CS) were not completely controlled by their current AED regimen were enrolled. Subjects entered a 6-week BL period to evaluate eligibility. After the 6-week BL, subjects who were eligible were evenly randomized to receive placebo or FFA 0.2 or 0.8 mg/kg/day added on top of stable AEDs during a 14-week treatment period. Because patients with DS typically have frequent seizures, we considered the possibility that a shorter BL period might be feasible in a TTE approach, and we re-analyzed Study 1 by counting the number of seizures during the first 7 days of BL. The number of seizures in this 7-day window was selected as the event (Nth seizure) and was specific for each patient. The TTE analysis during the treatment period was the time for each patient to reach their Nth seizure. Subjects who did not reach their Nth seizure by the end of the treatment period were censored. Results: In Study 1, a total of 119 subjects with DS were randomized to treatment (0.8 mg/kg/day, n=40; 0.2 mg/kg/day, n=39; placebo, n=40). The median TTE was longer in the active groups compared to the placebo group - about 1 week for placebo, 2 weeks for 0.2 mg/kg/day, and 4 weeks for 0.8 mg/kg/day (p<0.001). Censoring due to low seizure rates during the treatment phase was dose-dependent - 0 subjects for placebo, 13% for 0.2 mg/kg/day, and 28% for 0.8 mg/kg/day. Conclusions: Using a 7-day BL duration and a TTE analysis, time to reach the Nth seizure during the treatment period was significantly longer for subjects randomized to FFA and demonstrated a dose-response that was similar to the dose-response observed in the original analysis of Study 1. These data demonstrate that in patients with high seizure burden, a TTE study design with a short (7-day) BL is sensitive to detect a treatment effect that is consistent with the results from the standard 'change from BL analysis' and may be a viable alternative to the 'gold standard' fixed-dose design that minimizes exposure to placebo for patients with high seizure burden. Future analyses may include exploring the data from Study 1 using a variable BL duration to determine the effect on study power. Funding: Zogenix, Inc.