PERAMPANEL-BTRE-META
Efficacy and Safety of Perampanel in Patients With Brain Tumor-Related Epilepsy: A Systematic Review and Meta-Analysis
Bottom Line
Across 8 observational studies (n=382), perampanel achieved seizure freedom in 42% and ≥50% seizure reduction in 46% of BTRE patients, with adverse events in 26% (mostly mild-to-moderate) and no signal of an effect on tumor progression — supporting perampanel as a mechanistically targeted, non–enzyme-inducing option for BTRE.
Major Points
- Systematic review and meta-analysis of 8 observational studies (published 2017–2024) including 382 adults with brain tumor-related epilepsy treated with perampanel
- Pooled seizure freedom rate 42% (95% CI 27–59%; I²=65.7%, 7 studies); ≥50% seizure reduction 46% (95% CI 31–62%; I²=84.3%, 8 studies)
- Pooled oncologic progression 39% (95% CI 17–65%; I²=0%, 5 studies) — consistent with natural tumor history, no evidence of perampanel effect on tumor behavior
- Pooled adverse-event rate 26% (95% CI 17–38%; I²=61.6%, 8 studies); most frequent: dizziness, somnolence, fatigue
- Meta-regression: retrospective design associated with higher seizure-freedom rates (coef 1.03, p=0.035); AEs lower in retrospective cohorts (coef −1.03, p=0.040) and higher in IDH-mutant cases (coef −1.57, p=0.044); sample size and mean age were not associated with any outcome
- Mechanistic rationale: perampanel is a selective noncompetitive AMPA-receptor antagonist targeting glutamate-mediated peritumoral hyperexcitability that drives BTRE
- Non–enzyme-inducing profile makes perampanel attractive for co-administration with antineoplastic therapies (minimal drug–drug interactions)
- Limitations: all included studies were single-arm observational series, small samples, short/variable follow-up, English-only, and heterogeneous outcome definitions — precluding causal inference
Design
Study Type: Systematic Review and Meta-Analysis
Randomization:
Blinding: Not applicable (meta-analysis of single-arm observational studies)
Enrollment Period: Studies published 2017–2024; database search inception to April 2025
Follow-up Duration: Variable across included studies (not standardized)
Centers: 8
Countries: Italy, USA, Austria, Japan, France
Sample Size: 382
Analysis: Random-effects meta-analysis of single proportions (REML estimator); leave-one-out sensitivity, GLMM sensitivity, meta-regression; Egger test and funnel plots for publication bias; PROSPERO CRD420251113758; PRISMA 2020
Inclusion Criteria
- Prospective or retrospective observational study published in English
- Adult patients diagnosed with brain tumor-related epilepsy (BTRE) — epileptic seizures in association with a primary or secondary brain tumor
- Evaluated perampanel as monotherapy or adjunctive antiseizure medication
- Single-arm data (no control group required)
- Reported at least one efficacy or safety outcome (seizure freedom, ≥50% seizure reduction, oncologic progression, or adverse events)
Exclusion Criteria
- Non-original publications (reviews, editorials, commentaries, conference abstracts, theses)
- Animal or preclinical studies
- Publications with overlapping populations (retained the most complete/updated dataset)
- Non-English publications
Arms
| Field | Perampanel (pooled single-arm) |
|---|---|
| Intervention | Perampanel, monotherapy or adjunctive; mean daily dose ~3.4–7.5 mg/day across studies |
| Duration | Variable across included studies |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Seizure freedom — proportion of patients seizure-free during follow-up (pooled random-effects meta-analysis; 7 studies) | Primary | N/A (single-arm) | 42% pooled | N/A | |
| ≥50% reduction in seizure frequency (8 studies) | Secondary | N/A | 46% (95% CI 31%–62%); I²=84.3% | N/A | |
| Oncologic progression during perampanel therapy (5 studies) | Secondary | N/A | 39% (95% CI 17%–65%); I²=0.0% (no signal of perampanel effect on tumor behavior) | N/A | |
| Any adverse event (8 studies) | Secondary | N/A | 26% (95% CI 17%–38%); I²=61.6% | N/A | |
| Seizure freedom heterogeneity (I²) | Secondary | N/A | 65.7% (substantial); leave-one-out omitting Coppola 2020 resolved heterogeneity to I²=0% | N/A | |
| Meta-regression — study design as moderator of seizure freedom | Secondary | N/A | Coefficient 1.03 (retrospective > prospective) | 0.0345 | |
| Meta-regression — study design as moderator of AEs | Secondary | N/A | Coefficient −1.03 (lower AE rates in retrospective cohorts) | 0.0398 | |
| Meta-regression — IDH mutation status as moderator of AEs | Secondary | N/A | Coefficient −1.57 (higher AE probability in IDH-mutant) | 0.0441 | |
| Any adverse event (pooled) | Adverse | 26% (95% CI 17–38%); I²=61.6% | |||
| Most common AEs | Adverse | Dizziness, somnolence, fatigue (mild-to-moderate; consistent with known perampanel safety profile) | |||
| Serious AEs | Adverse | Not systematically reported across studies | |||
| Discontinuation for AEs | Adverse | Not pooled | |||
| Meta-regression — sample size as moderator of AEs | Adverse | Coefficient −0.60, p=0.32 (NS) | |||
| Meta-regression — mean age as moderator of AEs | Adverse | Coefficient −0.97, p=0.17 (NS) | |||
Subgroup Analysis
Leave-one-out sensitivity confirmed robustness of all outcomes (seizure freedom 0.37–0.50; ≥50% reduction 0.38–0.49; oncologic progression 0.32–0.44; AEs 0.26–0.30). GLMM sensitivity analyses produced comparable pooled estimates. Meta-regression covariates examined: study design, sample size, mean age, IDH mutation status.
Criticisms
- All included studies were single-arm observational series — no control groups, precluding causal inference or estimation of relative treatment effects
- Small individual sample sizes (median cohort ~12; range 5–272) and relatively short follow-up periods
- Substantial statistical heterogeneity for seizure freedom (I²=65.7%) and ≥50% reduction (I²=84.3%)
- Heterogeneity in outcome definitions, follow-up duration, and adverse-event reporting across studies (measurement/reporting bias)
- Inconsistent reporting of perampanel dose, treatment duration, concomitant ASMs, and molecular tumor markers (IDH1, MGMT)
- Restriction to English-language publications introduced potential language bias
- Wide confidence intervals for some pooled estimates reflecting limited data
- Modest number of studies (k<10) limits reliability of publication-bias assessment (funnel plot, Egger test) and statistical power of meta-regression
- JBI critical appraisal flagged incomplete/non-consecutive inclusion of participants in several retrospective series (selection bias)
- One study (Nakai 2024) contributed the majority of patients (272/382, 71%) and dominated pooled AE and ≥50%-reduction estimates
Funding
No targeted funding reported. Authors report no relevant disclosures.
Based on: PERAMPANEL-BTRE-META (Neurology Clinical Practice, 2026)
Authors: Da Silva AMP, Gonçalves OR, Tudella GCN, et al.
Citation: Neurol Clin Pract 2026;16(4):e200636. DOI: 10.1212/CPJ.0000000000200636
Content summarized and formatted by NeuroTrials.ai.