FAERS database reveals latent signals of neuro-ophthalmic toxicity associated with immune checkpoint inhibitors: feature mining and clinical warnings.

OBJECTIVE

To perform signal mining of neuro-ophthalmic immune-related adverse events (irAEs) associated with immune checkpoint inhibitors (ICIs) using the FDA Adverse Event Reporting System (FAERS), and to synthesize their clinical manifestations and temporal patterns, thereby providing evidence-based insights for optimizing clinical drug safety.

METHODS

Reports of neuro-ophthalmic irAEs associated with ICIs were extracted from the FAERS (Q1 2011 to Q4 2025). Signal detection was performed using disproportionality analysis methodologies, specifically the Reporting Odds Ratio (ROR) and Information Component (IC) algorithms, with predefined statistical thresholds for signal significance.

RESULTS

A total of 521 reports identified neuro-ophthalmic irAEs with ICIs as primary suspect drugs, predominantly affecting males and elderly patients At the High-Level Group Term (HLGT) level, positive signals were detected for pembrolizumab and avelumab. Preferred Term (PT) analysis revealed six significant signals, with eyelid ptosis being the most frequently reported event. The median time to onset was 28-38 days, with acute presentations (≤24 hours post-infusion) observed for pembrolizumab and nivolumab; both agents exhibited elevated case fatality proportions, with a higher proportion of male patients among fatal cases.

CONCLUSION

This study delineated potential neuro-ophthalmic positive disproportionality signals across seven ICIs, revealing demographic variations in signal reporting frequency with higher susceptibility among males and elderly patients. Pembrolizumab and avelumab were associated with higher reporting frequencies of neuro-ophthalmic adverse events, which were related to poor visual outcomes. Consequently, prompt recognition and management of neuro-ophthalmic irAEs are imperative to enhance ICI safety profiles in clinical practice.