Age-accelerated cognitive decline in asymptomatic adults with CSF β-amyloid

Abstract

Objective: Compare cognitive and hippocampal volume trajectories in asymptomatic middle-aged and older adults with positive CSF markers of β-amyloid (Aβ) or tau to adults without an Alzheimer disease (AD)-associated biomarker profile.

Methods: Three hundred ninety-two adults enrolled in a longitudinal cohort study (Wisconsin Registry for Alzheimer's Prevention or Wisconsin Alzheimer's Disease Research Center) completed a lumbar puncture and at least 2 biennial or annual neuropsychological evaluations. Cutoffs for Aβ₄₂, total tau, and phosphorylated tau were developed via receiver operating characteristic curve analyses on a sample of 78 participants (38 dementia, 40 controls). These cutoffs were applied to a separate sample of 314 cognitively healthy adults (mean age at CSF collection = 61.5 years), and mixed-effects regression analyses tested linear and quadratic interactions of biomarker group × age at each visit on cognitive and hippocampal volume outcomes.

Results: Two hundred fifteen participants (69%) were biomarker negative (preclinical AD stage 0), 46 (15%) were Aβ+ only (preclinical AD stage 1), 25 (8%) were Aβ+ and tau+ (preclinical AD stage 2), and 28 (9%) were tau+ only. Both stage 1 and stage 2 groups exhibited greater rates of linear decline on story memory and processing speed measures, and nonlinear decline on list-learning and set-shifting measures compared to stage 0. The tau+ only group did not significantly differ from stage 0 in rates of cognitive decline.

Conclusion: In an asymptomatic at-risk cohort, elevated CSF Aβ (with or without elevated tau) was associated with greater rates of cognitive decline, with the specific pattern of decline varying across cognitive measures.

Figure 1. Biomarker groups and cognitive trajectories

Graphs depict neuropsychological performance on the y-axis for 6 cognitive measures (A–F) and age at each visit (centered on mean age) on the x-axis. Each line depicts the estimated slope for the 4 biomarker groups, adjusting for covariates of sex, education, and practice effects. Higher scores equate better performance on all measures except TMT-B (higher scores = worse performance). Quadratic terms were retained for the RAVLT and TMT-B. Nonsignificant quadratic terms were removed for other outcomes, and linear effects are depicted. Both the Aβ+ only group (orange) and the Aβ+/tau+ group (green) exhibited significantly greater decline than the biomarker-negative group (black). In contrast, the group with only tau+ (blue) did not differ from biomarker-negative individuals. Aβ = β-amyloid; RAVLT = Rey Auditory Verbal Learning Test; TMT-B = Trail Making Test Part B.

Figure 2. Relationships among Aβ₄₂, tau, and longitudinal verbal memory performance

Two-way interaction between age at each visit and Aβ₄₂ (A) or p-tau (B) on memory performance. Figures depict that although performance generally decreases with age, those with low Aβ₄₂ (high brain amyloid) exhibit most rapid decline, whereas the association between age at each visit and memory performance does not vary by p-tau. Facets depict biomarker level by quartile (1 = lowest quartile [0%–25%], 2 = 25%–50%, 3 = 50%–75%, 4 = highest quartile [75%–100%]). Aβ₄₂ = β-amyloid 1–42; p-tau = phosphorylated tau.