Predicting Cognitive Decline Using Baseline Data in Cognitively Unimpaired Older Adults: Results from the A4 Trial
Babak Khorsand1, Elham Ghanbarian1, Laura Rabin2, Ali Ezzati1
1Department of Neurology, University of California, Irvine, 2Brooklyn College of CUNY
Objective:
To identify factors predicting cognitive/functional decline on the CDR over a 5-year period in cognitively unimpaired older adults.
Background:
The clinical and biological heterogeneity observed in Alzheimer’s disease makes the design of trials and inclusion of appropriate participants difficult. Identifying methods to discern more homogenous subgroups of participants can assist in enriching trials.
Design/Methods:
Participants included 1169 amyloid-positive participants enrolled in the A4 trial (receiving solanezumab, N=578, or placebo, N=591), and 538 amyloid-negative individuals enrolled in the LEARN observational study. We used multiple logistic regression models to examine the separate and joint predictive value of demographics, APOE4 genotype, neuropsychological tests, amyloid PET SUVR, and plasma phosphorylated tau 217 (P-tau217) levels in predicting individuals experiencing cognitive/functional decline (defined as an increase of 0.5 or more in the CDR-global score after 240 weeks).
Results:
Participants had an average age of 70.5 ± 4.2 years, with 60.3% being female. The addition of P-tau217 and PACC to the base model (demographics and APOE4 status) significantly improved the predictive performance across all study arms. For the base model plus P-tau217, the AUC values were: Solanezumab (AUC = 0.81±0.11), Placebo (AUC = 0.78±0.11), and LEARN (AUC = 0.76±0.15). Similarly, for the base model plus PACC, the AUC values were: Solanezumab (AUC = 0.72±0.12), Placebo (AUC = 0.76±0.11), and LEARN (AUC = 0.74±0.15). The models that included all covariates achieved the highest predictive performance across the study arms, with AUC values of: Solanezumab (AUC = 0.82 ± 0.1), Placebo (AUC = 0.83 ± 0.1), and LEARN (AUC = 0.80 ± 0.13).
Conclusions:
These findings underscore the importance of baseline neuropsychological scores and plasma P-tau217 in predicting cognitive decline. Predictive models using these practical measures can improve clinical trial design and optimize participant selection, enhancing the effectiveness of interventions.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.