Neurological adverse events associated with immune checkpoint inhibitors: a pharmacovigilance study based on the FAERS and JADER databases

Background Neurological adverse events (NAEs) have been reported during immune checkpoint inhibitor (ICI) therapy, but their reporting spectrum and onset patterns across drugs and pharmacovigilance databases remain incompletely characterized. Methods We conducted a pharmacovigilance study of seven ICIs using the FDA Adverse Event Reporting System (FAERS) and the Japanese Adverse Drug Event Report (JADER) database. Primary drug–preferred term (PT) signals were evaluated using whole-database disproportionality analyses. An exploratory FAERS analysis compared PD-L1 with PD-1 inhibitors within study-defined tumor-indicator strata. Factors associated with NAE reporting were assessed in FAERS by univariate and multivariable logistic regression, and time-to-onset (TTO) was summarized using descriptive statistics and Weibull analyses. JADER cemiplimab data were summarized descriptively only because of sparse reporting. Results The analysis included 3,915 NAE reports from FAERS and 2,929 from JADER. The heatmaps displayed 111 positive drug–PT pairs across 31 PTs in FAERS and 49 across 16 PTs in JADER. Myasthenia gravis showed a disproportionality signal for all seven ICIs in non-Japanese FAERS reports and for all six ICIs included in the JADER signal analysis; encephalitis-related terms and immune-mediated myositis also recurred, although drug-specific patterns varied. In FAERS, complete-case regression included 111,879 ICI-related reports with 2,887 NAEs. Age ≥65 years and ICI combination therapy were associated with higher adjusted odds of NAE reporting, whereas PD-L1 and CTLA-4 monotherapy were associated with lower adjusted reporting odds relative to PD-1 monotherapy. The stratified comparison showed lower myasthenia gravis reporting odds for PD-L1 than PD-1 in both strata; encephalitis confidence intervals included 1. Drug-level median TTOs ranged from 16 to 43 days in FAERS and from 16 to 32 days in JADER, with heterogeneous Weibull patterns across drugs and databases. Conclusion ICI-associated neurological reports showed recurring neuromuscular and central nervous system signals, partial qualitative concordance between non-Japanese FAERS reports and JADER, and heterogeneous onset patterns across drugs. The observed reporting associations do not establish comparative clinical safety. Overall, these findings are descriptive and hypothesis-generating and warrant confirmation in studies with exposure denominators, clinically adjudicated diagnoses, and longitudinal follow-up.