Baseline Amyloid PET Centiloid Score and Risk of Amyloid-Related Imaging Abnormalities in Patients Treated With Lecanemab.
BACKGROUND AND PURPOSE Amyloid-related imaging abnormalities (ARIA) are a known complication of anti-amyloid monoclonal antibody therapy for Alzheimer's disease. Centiloid score, derived from amyloid PET/CT, provides a quantitative assessment of baseline amyloid burden; however, its relationship with ARIA risk remains uncertain. This study evaluated whether baseline Centiloid score was associated with ARIA in patients treated with lecanemab and included a secondary exploratory analysis of regional amyloid uptake. MATERIALS AND METHODS We retrospectively identified patients treated with lecanemab within a single academic health system who underwent pretreatment florbetaben amyloid PET/CT. Centiloid scores were generated using a standardized processing pipeline. Multivariable logistic regression evaluated the association between baseline Centiloid and ARIA, adjusting for age, sex, APOE ε4 carrier status, and baseline cerebral microbleed presence. ARIA-H and ARIA-E subgroups were evaluated separately using Firth penalized logistic regression. A secondary Cox proportional hazards analysis accounted for unequal follow up durations. Exploratory regional analysis compared baseline amyloid uptake in regions that subsequently developed ARIA-E with that in mirrored contralateral regions. RESULTS The primary analysis included 41 patients with ARIA and 72 controls who completed at least 14 lecanemab infusions without ARIA. Baseline Centiloid was not independently associated with ARIA (adjusted OR per 10-Centiloid increase, 1.01 [95% CI, 0.89-1.15]; P = .90), ARIA-H (OR, 1.01 [95% CI, 0.89-1.14]; P = .91), or ARIA-E (OR, 1.02 [95% CI, 0.87-1.21]; P = .78). In the Cox analysis of 41 patients with ARIA and 100 patients without ARIA, baseline Centiloid was not associated with time to ARIA (adjusted HR, 1.01 [95% CI, 0.91-1.11]; P = .91). Among 19 patients with ARIA-E included in the regional analysis, baseline uptake was modestly greater in regions that subsequently developed ARIA-E than in mirrored contralateral regions for SUVmax (3.44 ± 0.78 versus 3.25 ± 0.77; P = .02) and SUVmean (2.60 ± 0.56 versus 2.45 ± 0.64; P = .01). CONCLUSIONS Baseline global Centiloid was not independently associated with ARIA in patients treated with lecanemab. Exploratory regional analysis demonstrated modest differences in baseline amyloid uptake in regions that subsequently developed ARIA-E, warranting confirmation in larger prospective studies.