Dwivedi · BMJ neurology open 2026 · Simulation and longitudinal cohort analysis of published trial data · n=?

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

Cited 0 times in the scientific literature.

Level 3 - non-randomized controlled study

Secondary simulation and longitudinal slope comparison of published trial data across double-blind and open-label extension phases

PubMed 42602524 · doi:10.1136/bmjno-2026-001649 · record verified 2026-08-26

What was done

Authors constructed mixed-effects simulation models calibrated to published means, variances, and attrition patterns from the Clarity-AD (lecanemab) and TRAILBLAZER-ALZ 2 (donanemab) phase 3 trials. They evaluated longitudinal Clinical Dementia Rating-Sum of Boxes (CDR-SB) trajectories, comparing annualized rates of decline between double-blind (DB) and open-label extension (OLE) phases for early-start (active drug in DB) and delayed-start (placebo in DB) cohorts. DB placebo trajectories were also compared to matched untreated historical natural history cohorts.

What was found

In early-start cohorts, annualized cognitive decline was significantly faster during the OLE phase than during the DB treatment phase for both lecanemab (difference in annual slope Δ = 0.53 points; 95% CI 0.40 to 0.66) and donanemab (Δ = 0.66 points; 95% CI 0.41 to 0.91). Delayed-start cohorts showed no meaningful slope differences between phases. Cognitive decline rates in the OLE phase matched those seen in the original DB placebo arms. In the lecanemab trial, DB placebo participants declined more slowly than matched untreated historical controls, while donanemab placebo decline varied by baseline disease severity.

Why it matters

Long-term open-label extension claims of sustained or expanding drug benefit relative to historical controls may be confounded by trial placebo effects and selection bias. When compared directly to within-trial double-blind performance, the slowing of cognitive decline did not persist at the same rate during open-label treatment.

Limits

The study relied on simulations calibrated to published summary-level trial statistics rather than individual patient-level data. Comparisons to external natural history cohorts carry inherent baseline confounding and selection biases. Specific participant counts were not stated in the abstract.

Cited by