Abstracts

Rationale: High baseline seizure frequency was previously identified as a factor associated with poor seizure control; identification of clinical factors associated with improved response may help to inform clinicians on optimal treatment. This post hoc analysis aimed to identify predictive clinical factors associated with response to adjunctive perampanel in patients with FOS in the prospective, observational AMPA Study (NCT04257604; Study 501).

Methods: Adult and adolescent patients aged ≥ 12 years with inadequately controlled FOS, with/without focal to bilateral tonic-clonic seizures (FBTCS), while receiving 1–3 anti-seizure medications (ASMs) were prescribed adjunctive perampanel in line with the approved indication. The treating physician’s decision to prescribe perampanel was made before and independently of their decision to include the patient in the study. The primary endpoint was median percent change in seizure frequency at 6 months (secondary endpoint, Month 12); other secondary endpoints included retention, 50% and 75% responder (assessed between 6–12 months or at the end of treatment for early discontinuations) and seizure-freedom rates, and monitoring of treatment-emergent adverse events (TEAEs). This analysis assessed variables influencing 50% and 75% responder rates using logistic regression to determine odds ratios (ORs) and 95% confidence intervals (CIs). Goodness of fit was measured by area under the receiver operating characteristic (ROC) curve. Baseline seizure frequency, seizure history (simple FOS, complex FOS, and FBTCS), age at diagnosis, time since diagnosis, idiopathic etiology, enzyme-inducing ASM (EIASM) status, and number of baseline ASMs were assessed individually as continuous and/or categorical variables in univariate analyses. Multivariate analyses were performed using both forwards and backwards selection and included all variables.

Results: Of the 234 patients who were treated with adjunctive perampanel, 200 patients had baseline and relevant post-baseline seizure data and were included in this analysis (Intent-to-Treat Analysis Set). In univariate analyses, there was borderline evidence that concomitant EIASM use reduced the chances of 50% response (OR 0.564; 95% CI 0.311, 1.022; P=0.0592). In multivariate analyses, presence of simple FOS (OR 2.119; 95% CI 1.023, 4.390; P=0.0433), and non-EIASMs (OR 0.559; 95% CI 0.293, 1.064; P=0.0767) were the best predictors of 50% response (Figure 1). For 75% response, both univariate and multivariate analyses showed some evidence that concomitant EIASM use reduced the chances of response (Figure 2). No other variables influenced response (Figures 1 and 2).

Conclusions: In the AMPA Study, the best predictive factors for 50% response were non-EIASM use and presence of simple FOS at baseline; the best predictive factor for 75% response was non-EIASM use at baseline.

Funding: Eisai s.r.l.