Comparison of double-blind and open-label decline rates in lecanemab and donanemab trials in Alzheimer's disease.

Background: Phase 3 trials of the anti-amyloid monoclonal antibodies lecanemab and donanemab in Alzheimer's disease demonstrated modest slowing of cognitive decline over 18 months. Subsequent open-label extensions (OLE) suggested greater long-term benefit based on comparisons with historical untreated cohorts, which are vulnerable to selection and attrition bias. Methods: We simulated longitudinal Clinical Dementia Rating-Sum of Boxes trajectories using mixed-effects models calibrated to reproduce published means, variances and attrition patterns from the Clarity-AD and TRAILBLAZER-ALZ 2 trials, separately for early- and delayed-start cohorts and in combined analyses. Observed annualised slopes were also directly compared between double-blind (DB) and OLE phases. Placebo effects were estimated by comparing DB placebo arms with matched untreated historical cohorts derived from published natural history data. Results: Simulations closely reproduced reported trajectories. In both trials, the early-start cohorts showed significantly faster annualised decline during OLE than during DB treatment (lecanemab ∆=0.53; 95% CI 0.40 to 0.66; donanemab ∆=0.66; 95% CI 0.41 to 0.91), whereas delayed-start cohorts showed no meaningful phase-related differences. OLE decline rates approximated those observed in DB placebo groups. In the lecanemab trial, DB placebo participants declined more slowly than matched untreated controls; decline in the donanemab trial varied by baseline severity. Conclusions: Despite selective retention favouring slower progressors, both lecanemab and donanemab accelerated cognitive decline during OLE phases and paralleled placebo-level declines.