Divergent patterns of cognitive decline in preclinical Alzheimer's disease: implications for secondary prevention trials.

Background: Alzheimer's disease biomarkers in cognitively unimpaired older adults are associated with later cognitive and clinical decline, yet substantial heterogeneity in the timing and rate of decline remains insufficiently characterized. This study aims to identify subgroups of cognitive decline among biomarker-defined cognitively unimpaired adults and determine baseline predictors of heterogeneity in preclinical Alzheimer's disease progression. Methods: Longitudinal data was analyzed from the Anti-Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4) Study, which enrolled amyloid-positive participants, and the parallel LEARN Study, which enrolled amyloid-negative individuals meeting all other A4 criteria. Participants completed baseline amyloid PET, plasma P-tau217, structural MRI, and serial cognitive assessments. Latent Class Mixed-Effects Models (LCMMs) were used to identify distinct cognitive trajectory classes. Associations between class membership and demographic, clinical, and biomarker characteristics were evaluated. The primary outcome was longitudinal change in the Preclinical Alzheimer Cognitive Composite (PACC). Findings: Three cognitive trajectory classes were identified: stable, slow decliners, and fast decliners. Higher plasma P-tau217, smaller hippocampal volume, and elevated tau PET were associated with greater odds of belonging to declining classes. Among amyloid-positive individuals, approximately 70% were classified as stable over the observed follow-up interval. These stable individuals likely contribute little to the power of preclincal Alzheimer's trials. Interpretation: Latent class modeling reveals marked heterogeneity in preclinical cognitive trajectories, even among individuals with biomarker evidence of Alzheimer pathology. The high proportion of stable individuals, though consistent with the long presymptomatic interval, has important implications for prevention trial design, particularly regarding inclusion criteria, outcome measures, and treatment effect assumptions. Identifying subgroups of decline may improve prognostic modeling and guide enrichment strategies for precision secondary prevention trials. Funding: US National Institutes of Health, Eli Lilly, Alzheimer's Association, Foundation for the National Institutes of Health, GHR Foundation, Davis Alzheimer Prevention Program, Yugilbar Foundation, Avid Radiopharmaceuticals, Cogstate, Albert Einstein College of Medicine, Foundation for Neurologic Diseases, and Epstein Family Foundation. Research in Context: Evidence before this study: We searched PubMed, MEDLINE, and Google Scholar for articles published between January 1, 2000, and December 31, 2025, using the terms "preclinical Alzheimer's disease," "cognitive decline," and "latent class." Previous studies have identified patterns of cognitive decline in aging populations, but there is limited understanding of the predictive value of recently developed biomarkers of Alzheimer's pathology, and we found no studies examining the impact on secondary prevention trial design.Added value of this study: This study is the first to apply latent class analysis to data from the A4 and LEARN studies, identifying distinct classes of cognitive decline in individuals at risk for Alzheimer's disease. Our findings reveal significant heterogeneity in cognitive trajectories, suggesting that preclinical Alzheimer's disease is not a uniform process. These insights could inform personalized intervention strategies and improve the design of future clinical trials.Implications of all the available evidence: Our results underscore the importance of considering individual variability in cognitive decline when developing and testing interventions for preclinical Alzheimer's disease. Recognizing the diverse patterns of cognitive decline can enhance the precision of clinical trial designs and lead to more effective, tailored therapeutic approaches. Future research should focus on validating these findings in larger, more diverse cohorts and exploring the underlying mechanisms driving the observed heterogeneity.