Lecanemab disease-progression effects sustain at 36 months in early-symptomatic Alzheimer's cohort

Lecanemab, a monoclonal antibody directed against soluble amyloid-beta protofibrils, demonstrated a 27% slowing of clinical decline at 18 months in the core CLARITY-AD trial (NEJM 2023). A central open question was whether the treatment effect would sustain, attenuate, or diverge over a longer treatment horizon — particularly in the open-label extension where placebo-arm participants crossed over to active treatment. ARIA (amyloid-related imaging abnormalities) — both effusion-type (ARIA-E) and hemosiderin-type (ARIA-H) — remained the principal safety concern.

Across the 36-month observation, the lecanemab arm continued to show a slower decline on CDR-SB than the modelled placebo trajectory derived from the natural-history Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort. The 36-month between-group difference was 0.95 points on CDR-SB (95% CI 0.71-1.19), a clinically meaningful effect size consistent with the 18-month estimate (0.45 points). Importantly, the slope did not attenuate: the second-half slowing rate was comparable to the first-half rate, suggesting durable rather than diminishing treatment effect. Crossover participants (placebo to lecanemab at month 18) showed a partial catch-up effect — their post-crossover decline rate matched the long-term lecanemab rate, but they did not recover the cognitive ground lost during their 18-month placebo exposure. This argues for earlier initiation of treatment. ARIA-E and ARIA-H rates over the full 36 months were 12.6% and 17.3% respectively, comparable to core-trial rates. No new safety signals emerged. Two treatment-related deaths occurred in the extension; both were in APOE-e4 homozygotes with concurrent anticoagulant use, consistent with the established risk profile.

Whether anti-amyloid disease-modifying therapy produces a durable effect — not merely a delay — has been a central question for the field. These 36-month data argue for durability: the slope difference did not narrow over time. If confirmed in further follow-up, the implication is that earlier initiation should produce a larger lifetime cognitive benefit, reshaping diagnostic urgency for early-symptomatic patients.

Participants who completed the 18-month core CLARITY-AD trial were eligible to enter the open-label extension. All extension participants received lecanemab 10 mg/kg biweekly. Primary outcome was the CDR-SB slope across 36 months. Secondary outcomes included ADAS-Cog14, ADCS-MCI-ADL, ARIA incidence, and plasma p-tau217 trajectories. The placebo trajectory beyond month 18 was modelled rather than directly observed: a longitudinal mixed-effects model using ADNI natural-history data matched on baseline CDR-SB, APOE-e4 status, age, and sex was the comparator. Sensitivity analyses used alternative modelling assumptions (linear vs. non-linear progression, alternative reference cohorts). Safety monitoring followed core-trial protocols: MRI at months 3, 6, 12, 18, 24, 30, and 36, plus event-driven scans. ARIA-E and ARIA-H severities were graded centrally.

The most important limitation is the modelled placebo trajectory: comparing observed lecanemab outcomes against an inferred placebo arm is methodologically weaker than a randomised continuation. Sensitivity analyses largely converged but the effect size estimate carries wider uncertainty than the core-trial headline. APOE-e4 homozygotes remain at elevated ARIA risk and excluding them from the trial population (as some clinics now do) trades trial generalisability for safety.

Whether the durability holds beyond 36 months remains unanswered — the extension is ongoing, with a planned 60-month readout in 2028. The threshold at which the slowing effect becomes clinically meaningful for patients (vs. statistically detectable) is also unresolved. Comparative-effectiveness data against donanemab (competing anti-amyloid antibody, Eli Lilly) are not yet available and represent the field's next major question.