Real-world safety and efficacy of lecanemab in the treatment of Alzheimer's disease: a systematic review and meta-analysis.

BACKGROUND

Despite the approval of lecanemab for Alzheimer's disease (AD), evidence regarding its performance outside controlled trial settings remains scarce, leaving its real-world safety and clinical utility largely unknown.

METHODS

We searched PubMed, Embase, and Web of Science (through October 1, 2026) for real-world studies of lecanemab in AD. Outcomes included ARIA, discontinuation, infusion-related reactions, serious adverse events, MMSE, MoCA, CDR-SB, and amyloid PET. Random-effects models with REML and Hartung-Knapp adjustment were used. Subgroup analyses were stratified by region, APOE ε4 genotype, baseline CDR score, microhemorrhage, and white matter hyperintensity. Sensitivity analyses included leave-one-out and Egger's test. Scenario-based sensitivity analyses were also performed for overlapping cohorts. This study was registered with PROSPERO (No. CRD420261333493).

RESULTS

Thirty-seven studies were included. For overlapping cohorts, the largest dataset per outcome was retained after prioritizing peer-reviewed reports. Pooled incidence of any ARIA was 15.1% (US vs Asian studies 19.5% vs 9.7%), ARIA-E 6.5% (10.6% vs 2.5%), ARIA-H 11.5% (15.2% vs 7.0%), and symptomatic ARIA 1.3% (3.3% vs 0.1%). APOE ε4 homozygotes showed the highest rates across all three subtypes (any ARIA 34.1%, ARIA-E 10.3%, ARIA-H 18.9%), followed by heterozygotes (14.5%, 4.7%, 12.0%) and non-carriers (11.0%, 2.9%, 6.7%). Baseline microhemorrhage was associated with higher any ARIA (23.4% vs 11.5%) and ARIA-H (19.3% vs 4.2%), but not ARIA-E (5.4% vs 5.7%). ARIA incidence did not differ significantly by baseline CDR severity or deep white matter hyperintensity burden. Except for MoCA at ≤ 3 months (MD 0.515, 95% CI 0.451-0.578; 2 studies), no statistically significant changes were observed in cognitive outcomes. Amyloid PET decreased at 6 months (Centiloid MD -28.453, 95% CI -37.708 to -19.198) and 12 months (SUVR MD -0.171, 95% CI -0.291 to -0.052). Discontinuation rates were 3.5% (ARIA-related), 6.2% (adverse event-related), and 12.2% (total); infusion-related reactions were 18.5% and serious adverse events 2.8%. Sensitivity analyses supported robustness.

CONCLUSIONS

This is the first large-scale systematic review and meta-analysis of real-world studies of lecanemab reporting safety and efficacy outcomes. Safety outcomes particularly ARIA may be influenced by APOE ε4 genotype and baseline microbleeds. Cognitive outcomes lacking control data cannot confirm efficacy and longer-term cross-regional controlled studies are needed.