Abstracts

Rationale: LGS is a severe refractory childhood-onset epilepsy characterized by a triad of multiple seizure types (including drop seizures), intellectual disability, and certain electroencephalographic abnormalities. Study 338 (NCT02834793) was a Phase III study of adjunctive perampanel in patients aged ≥ 2 years with uncontrolled seizures with LGS. Efficacy was assessed primarily on drop seizures, defined as tonic, atonic, or myoclonic seizures that led/could have led to a fall (per Study 338 protocol). Given that the definition of drop seizures has evolved since Study 338 initiation, we hereby report a post hoc efficacy analysis based on the drop seizure definition used in recent Phase III studies of cannabidiol for LGS (tonic, atonic, or tonic-clonic seizures).

Methods: Study 338 enrolled patients with LGS who were receiving 1–4 concomitant anti-seizure medications and had an average of ≥ 2 drop seizures/week during baseline. The study consisted of a randomized, double-blind, placebo-controlled Core Study (6-week Titration, 12-week Maintenance), followed by open-label Extension A (6-week double-blind conversion; 46 weeks Maintenance) and Extension B (for patients in Japan or countries without an Extended Access Program). Primary endpoint: median percentage change from baseline in drop seizure frequency/28 days during Titration and Maintenance Periods. Secondary endpoints: 50% and 75% responder and seizure-freedom rates, and safety outcomes.

Results: The Study 338 per-protocol drop seizures (tonic/atonic/myoclonic) dataset included all 70 patients (perampanel, n=34; placebo, n=36) who were randomized and treated in the Core Study and all 58 patients who entered Extension A. The alternative-definition drop seizures data set (tonic/atonic/tonic-clonic) included a subset of 66 patients (perampanel, n=32; placebo, n=34) in the Core Study, of whom 56 patients were treated in Extension A. In the Core Study, the median percent reduction in drop seizure frequency/28 days was numerically higher for perampanel vs placebo for both drop seizure definitions (tonic/atonic/myoclonic: 23.1% vs 4.5%, P=0.107; tonic/atonic/tonic-clonic, 48.6% vs –0.7%, P=0.001; Figure 1). Similarly, drop seizure responder rates were higher for perampanel vs placebo for both definitions (Figure 2). The efficacy of perampanel was maintained through the Extension Phase for both definitions. Treatment-emergent adverse events occurred in 85.3% of perampanel-treated vs 72.2% of placebo-treated patients, most commonly somnolence.

Conclusions: Numerical reductions in drop seizure frequency were seen with adjunctive perampanel in patients with LGS in the Core Study, regardless of definition (based on seizure types) of drop seizures; these reductions were maintained over the Extension Phase.

Funding: Eisai Co., Ltd., Eisai Inc., and Eisai Ltd.