Neurodegeneration, Alzheimer's disease biomarkers, and longitudinal verbal learning and memory performance in late middle age

Abstract

This study examined the effect of neurodegeneration, and its interaction with Alzheimer's disease (AD) cerebrospinal fluid biomarkers, on longitudinal verbal learning and memory performance in cognitively unimpaired (CU) late middle-aged adults. Three hundred and forty-two CU adults (cognitive baseline mean age = 58.4), with cerebrospinal fluid and structural MRI, completed 2-10 (median = 5) cognitive assessments. Learning and memory were assessed using the Rey Auditory Verbal Learning Test (RAVLT). We used sequential comparison of nested linear mixed effects models to analyze the data. Model selection preserved a significant ptau₁₈₁/Aβ₄₂ × global atrophy × age interaction; individuals with less global atrophy and lower ptau₁₈₁/Aβ₄₂ levels had less learning and delayed recall decline than individuals with more global atrophy and/or higher levels of ptau₁₈₁/Aβ₄₂. The hippocampal volume × age × ptau₁₈₁/Aβ₄₂ interaction was not significant. Findings suggest that in a sample of CU late middle-aged adults, individuals with AD biomarkers, global atrophy, or both evidence greater verbal learning and memory decline than individuals without either risk factor.

Keywords: Cerebrospinal fluid biomarkers; Global atrophy; Hippocampus; Preclinical Alzheimer's disease; Verbal learning; Verbal memory.

Figure 1:

Immediate recall (sum of learning trials 1 through 5) from the Rey Auditory Verbal Learning Test. Lines depict model-predicted age trajectories for three values of ptau₁₈₁/Aβ₄₂: red represents +1 standard deviation from the mean; gray represents the mean value; blue represents −1 standard deviation from the mean. Each panel reflects the model fit at a particular value of global atrophy (columns; −1, 0, +1 SD from the mean) and hippocampal volume (rows; sim.). Model predictions were made assuming a male participant with 16.15 years of education (the mean level) and no prior exposure to the battery. Confidence bands reflect the standard error of prediction for each line. The overlaid scatter represents raw individual test score measurements within nine (3×3 panel) predictor value bins, grouped such that Z_predictor ≤ −0.5 (top/left), −0.5 < Z_predictor ≤ 0.5 (center/center), and Zpredictor > 0.5 (bottom/right).

Figure 2:

Delayed recall from the Rey Auditory Verbal Learning Test. Lines depict model-predicted age trajectories for three values of ptau₁₈₁/Aβ₄₂: red represents +1 standard deviation from the mean; gray represents the mean value; blue represents −1 standard deviation from the mean. Each panel reflects the model fit at a particular value of global atrophy (columns; −1, 0, +1 SD from the mean) and hippocampal volume (rows; sim.). Model predictions were made assuming a male participant with 16.15 years of education (the mean level) and no prior exposure to the battery. Confidence bands reflect the standard error of prediction for each line. The overlaid scatter represents raw individual test score measurements within nine predictor value bins, grouped such that Z_predictor ≤ −0.5 (top/left), −0.5 < Z_predictor ≤ 0.5 (center/center), and Z_predictor > 0.5 (bottom/right).