Background: Cognitive composites commonly serve as primary outcomes in Alzheimer's disease (AD) secondary prevention trials.
Objective: To evaluate the association between amyloid (Aβ) burden level (+/-) and performance on three separate composite endpoints: Preclinical Alzheimer's Cognitive Composite (PACC), PACC+Semantic Fluency (PACC5), and Repeatable Battery for Neuropsychological Status (RBANS).
Design: Screening data from the randomized, double-blind, placebo-controlled, phase 2b/3 atabecestat EARLY study in preclinical AD participants were used in this analysis.
Setting: The EARLY study was conducted at 143 centers across 14 countries.
Participants: 3,569 cognitively unimpaired older adults (Clinical Dementia Rating of 0; aged 60-85 years) screened for inclusion in the EARLY study with Aβ status and at least PACC or RBANS at screening were included. Participants were categorized as those with non-pathological Aβ levels (Aβ-, n=2,824) and those with pathological Aβ levels (Aβ+, n=745) based on florbetapir uptake or levels of cerebrospinal fluid Aβ1-42.
Measurements: Analysis of Covariance models controlling for age, sex, and education were used to examine the difference in PACC, PACC5, and RBANS between Aβ groups. Nonparametric bootstrap was used to compare sensitivity of composites to differentiate between Aβ status.
Results: Of 3,569 participants, 2,116 were women (59%); 3,006 were Caucasian (84%); mean (SD) age was 68.98 (5.28) years. Aβ+ participants performed worse versus Aβ- participants on all cognitive composites though the magnitude of the Aβ effect was generally small. The Aβ+/- effect size for the PACC (Cohen's d=-0.15) was significantly greater than the RBANS (d=-0.097) while the PACC5 effect size (d=-0.139) was numerically larger than the RBANS. When examining subscores from the composites, memory tests (i.e., Free and Cued Selective Reminding Test, Figure Recall) and speed of processing (i.e., Digit-Symbol/Coding on the PACC/RBANS) exhibited the largest Aβ+/- effect sizes.
Conclusions: Cross-sectional relationships between Aβ and cognition among clinically unimpaired older adults are detectable on multi-domain cognitive composites but are relatively small in magnitude. The Aβ+/- group effect was statistically larger for PACC and marginally larger for PACC5 versus RBANS. However, interpretation of composite sensitivity to Aβ status cross-sectionally cannot be generalized to sensitivity to change over time.
Keywords: Amyloid; PACC; RBANS; cognition; preclinical AD.