Perampanel in routine clinical use across Europe: Pooled, multicenter, observational data

ObjectiveTo pool observational data on the routine use of perampanel to obtain information on real-world outcomes and data in populations typically underrepresented in clinical trials. MethodsIndividual-level data of people with epilepsy treated with perampanel at 45 European centers were merged into a single dataset. Prespecified outcomes were: 1-year retention rate, 1-year seizure freedom rate (duration6months), and incidence of treatment-emergent adverse events (TEAEs). In addition, relationships were explored with logistic regression analyses. ResultsThe full analysis set comprised 2396 people: 95% had focal seizures; median epilepsy duration was 27years; median number of concomitant antiepileptic drugs (AEDs) was 2; and median prior AEDs was 6. One-year retention rate was 48% (1117/2332; 95% confidence interval [CI] 46-50%), and 1-year seizure-free rate (6-month duration) was 9.2% (74/803; 95% CI 7-11%). Median treatment duration was 11.3months (1832 patient-years); median dose was 8mg. In 388 individuals with available data at 3, 6, and 12months, responder rates were 42%, 46%, and 39%, respectively. During the first year, TEAEs were reported in 68% of participants (1317/1497; 95% CI 66-70%). Logistic regression found higher age at perampanel initiation was associated with higher seizure-free rate, and higher number of prior AEDs with lower seizure-free rate and lower rates of somatic TEAEs. In 135 individuals aged 65years, 1-year retention rate was 48% and seizure-free rate was 28%. SignificanceAcross a large, treatment-resistant population, add-on perampanel was retained for 1year by 48% of individuals, and 9% were seizure-free for 6months. TEAEs were in line with previous reports in routine clinical use, and less frequent than in the clinical trial setting. No new or unexpected TEAEs were seen. Despite the limitations of observational studies, our data indicate that some individuals may derive a marked benefit from the use of perampanel.