Comparison of safety of lecanemab and donanemab: a real-world disproportionality analysis using the FDA adverse event reporting system.

Background: Lecanemab and donanemab are anti-amyloid-β (Aβ) monoclonal antibodies recently approved for the treatment of Alzheimer's disease (AD). Although both agents have demonstrated therapeutic potential, their post-marketing adverse event reporting profiles remain insufficiently characterized and compared in spontaneous reporting systems. This study aimed to systematically compare adverse event signals associated with these two drugs using the FDA Adverse Event Reporting System database, with sex-stratified and sensitivity analyses performed to support the main findings. Methods: FAERS reports up to the fourth quarter of 2025 were retrospectively analyzed. Reports in which lecanemab or donanemab was recorded as the primary suspect drug were included. AEs were classified by system organ classes (SOCs) and preferred terms (PTs) according to the Medical Dictionary for Regulatory Activities (MedDRA). Signal detection was performed using four algorithms: reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network (BCPNN), and multi-item gamma Poisson shrinker (MGPS). Sex-stratified analysis, sensitivity analysis after excluding reports involving concomitant medications, and time-to-onset (TTO) analysis based on the Weibull shape parameter model were also conducted. False discovery rate (FDR) correction was applied to analyses involving multiple P values. Results: A total of 3,640 AE reports were identified, including 2,602 for lecanemab and 1,038 for donanemab. Nervous system disorders was the most frequently reported SOC and the only SOC meeting the positivity criteria across all four algorithms for both drugs. At the PT level, the frequently reported events for both drugs were mainly concentrated in amyloid-related imaging abnormality (ARIA)-related events, including ARIA with oedema/effusion (ARIA-E) and ARIA with microhaemorrhages/haemosiderin deposition (ARIA-H), headache, and infusion-related reactions. The strongest disproportionality reporting signals were mainly observed for ARIA-related preferred terms and cerebral microhaemorrhage. In separate FAERS-based disproportionality analyses, lecanemab showed stronger disproportionality reporting signals for ARIA-H, whereas donanemab showed stronger disproportionality reporting signals for ARIA-E. In addition, drug-specific PT signal distributions differed between the two agents. Several potential novel PT signals were also identified. Sex-stratified analysis suggested differences in the distribution of certain PT signals between female and male reports. Sensitivity analysis supported the stability of most core signals. The median TTO was 46 days for lecanemab and 31 days for donanemab, and Weibull analysis suggested different temporal patterns of AE occurrence. Conclusion: Using four signal detection algorithms, this study compared real-world adverse event reporting profiles associated with lecanemab and donanemab. The two drugs showed both shared and distinct adverse event reporting profiles, particularly in ARIA subtype-related disproportionality signals, drug-specific PT signals, potential novel reporting signals, sex-stratified reporting patterns, and TTO characteristics. These findings provide pharmacovigilance evidence for targeted safety monitoring and hypothesis generation, but should not be interpreted as causal associations or true incidence estimates.