Clinical Practice Outcomes With Lecanemab for Alzheimer Disease

Lecanemab, an antiamyloid monoclonal antibody, received US Food and Drug Administration approval in July 2023 to treat amyloid-positive early stages of Alzheimer disease (AD) and has since been adopted in many countries. The primary objective was to assess amyloid-related imaging abnormality (ARIA) incidence and serious adverse events. Secondary objectives included identifying ARIA predictors and evaluating its effect on 12-month cognitive change.

We performed a retrospective, single-site observational study in patients treated with lecanemab at the Duke University (May 2023-June 2025). Eligible patients met National Institute on Aging-Alzheimer's Association criteria for mild cognitive impairment or mild AD with confirmed amyloid pathology. Treatment protocols and MRI interpretations were standardized.

Among 230 patients (68% MCI; mean age 73.6; 50.9% female; 67.4% allele producing the ε4 type of apolipoprotein E [APOE ε4] carriers), the mean follow-up was 396 days (SD 183). Forty-nine patients (21% [95% CI 16.3%-27.3%]) discontinued treatment. ARIA occurred in 56 patients (24.3% [95% CI 19.1%-30.5%]). ARIA with edema (ARIA-E) and ARIA with microhemorrhages (ARIA-MH) generally had a peak incidence rate at ~10 weeks, with a smaller ARIA-E peak at ~25 weeks; most moderate-to-severe events occurred between 12 and 24 weeks. Clinically significant adverse events occurred in 72 patients (31.3% [95% CI 25.5%-37.8%]), including 5 deaths (2 ARIA-related). Twelve-month cognitive changes did not significantly differ by ARIA status, although there was a significant decline of 0.107 [95% CI 0.04-0.17] units per month in the Montreal Cognitive Assessment score. ARIA-E risk was elevated in APOE ε4 carriers (OR 4.81 [95% CI 1.17-21.45] for 2 copies, p = 0.014) and in men (OR 2.40 [95% CI 0.88-7.25], p = 0.075). Baseline ptau-181/amyloid beta 42, APOE ε4, and Fazekas scores showed high specificity (>90%) but low sensitivity (<25%) for ARIA. Initiation of treatment earlier in the disease course was associated with fewer ARIA events, particularly among APOE ε4 noncarriers and those with lower baseline amyloid burden.

Despite the limitations of a single-site cohort, these findings inform real-world lecanemab safety and efficacy. Dual ARIA-E peaks at ~10 and ~25 weeks underscore the need for ongoing MRI monitoring. Although several markers showed modest negative predictive value, none reliably predicted ARIA. Improved markers to predict and monitor ARIA are needed to optimize cognitive outcomes.