Peripheral nervous system safety signals of antibody-drug conjugates: cross-database reproducibility and labeling gaps identified using FAERS, JADER, and CVARD.

Objective: To characterize the spectrum, strength, and cross-database reproducibility of peripheral nervous system (PNS) safety signals associated with antibody-drug conjugates (ADCs), and to examine discrepancies between post-marketing pharmacovigilance signals and current regulatory labeling. Methods: A multi-database pharmacovigilance study was conducted using spontaneous adverse event reports from the FDA Adverse Event Reporting System (FAERS; Q1-2025Q3), the Japanese Adverse Drug Event Report (JADER; 2004Q1-2025 Q3), and the Canada Vigilance Adverse Reaction Database (CVARD; January 1965 - October 2025). PNS-related adverse events were identified using the MedDRA Standardised MedDRA Query "Peripheral neuropathy," supplemented by expert-curated preferred terms. Disproportionality analyses were performed using the reporting odds ratio, proportional reporting ratio, empirical Bayes geometric mean, and information component. Safety signals detected in FAERS were evaluated for reproducibility in JADER and CVARD and compared with PNS events described in current prescribing information. Results: Overall, 31,592 post-marketing reports in which an ADC was recorded as the primary suspect were identified in FAERS, of which 1,395 involved PNS-related adverse events. A broad spectrum of PNS toxicities was observed, with peripheral neuropathy, polyneuropathy, peripheral sensory neuropathy, and peripheral sensorimotor neuropathy being the most frequently reported phenotypes. Consistent and robust associations were observed across all three databases, with reporting odds ratios of 4.62 (95% confidence interval [CI] 4.39-4.87) in FAERS, 42.84 (95% CI 39.56-46.40) in JADER, and 20.79 (95% CI 18.12-23.85) in CVARD. Rare but clinically severe neuropathic conditions, including Guillain-Barré syndrome and chronic inflammatory demyelinating polyradiculoneuropathy, also met predefined signal-detection criteria. Time-to-onset analyses demonstrated substantial heterogeneity, with median onset occurring within approximately 20-40 days for several ADCs, whereas delayed presentations beyond 80-100 days were observed for others. A large proportion of PNS reports were classified as serious, most commonly involving hospitalization. Notably, several PNS signals identified through post-marketing surveillance were not consistently described in current prescribing information. Conclusion: This multi-database pharmacovigilance analysis identified robust and reproducible PNS safety signals associated with ADCs, including clinically relevant events not uniformly described in current regulatory labeling. These findings are hypothesis-generating and support signal prioritization rather than causal inference, underscoring the importance of ongoing post-marketing surveillance to inform regulatory awareness and risk communication as ADC use continues to expand.