Subjective cognitive decline is associated with longitudinal cerebral blood flow reductions and gray matter atrophy in older adults

Abstract

AbstractBackgroundSubjective cognitive decline (SCD), a potential harbinger of Alzheimer’s disease (AD), has been cross‐sectionally associated with neurodegeneration and cerebrovascular disease markers on brain MRI. However, longitudinal associations are less clear with few existing studies that yield inconsistent findings. Here, we related baseline SCD to longitudinal structural and hemodynamic markers of brain health in older adults free of clinical dementia and stroke.MethodsVanderbilt Memory & Aging Project participants (n=299,72±7 years, 58% male) completed a baseline SCD protocol and multimodal 3T brain MRI, including T 1‐weighted imaging (grey matter volumes) and pseudo‐continuous arterial spin labeling (cerebral blood flow (CBF)) at baseline and multiple follow‐ups (18‐months, 3‐years, 5‐years). Baseline SCD (total and subdomain memory, language, and executive functioning scores) was longitudinally related to global and regional grey matter volumes and CBF (mL/100g/min). Linear mixed effects models, with random intercepts and slopes and a follow‐up time interaction, covaried for baseline age, sex, race/ethnicity, education, diagnosis, mood, and apolipoprotein E (APOE)‐e4 status. Grey matter volume models covaried for intracranial volume. CBF models covaried for associated tissue volume.ResultsMean follow‐up time was 3.8 years. Higher total SCD predicted hippocampal volume reductions (p=0.04), increased inferior lateral ventricular volume (p<0.001; Figure 1A), and global CBF reductions (p<0.001; Figure 1B). Higher scores within each SCD domain predicted hippocampal volume reductions, increased inferior lateral ventricular volume, and global CBF reductions (p‐values<0.04). Higher executive functioning‐SCD related only to frontal and occipital CBF reductions (p‐values<0.003; Figure 1C) but not temporal CBF or lobar atrophy (p‐values>0.07). Language‐SCD predicted reduced frontal CBF (p=0.01; Figure 1D) and temporal lobe volume (p=0.02). Surprisingly, memory‐SCD was not associated with temporal lobe volume or CBF (p‐values>0.05). Sensitivity analyses excluding outliers revealed the same associations.ConclusionsHigher baseline SCD predicted multiple pathways of longitudinal brain changes, including cerebral hemodynamics and volume loss within areas first affected by AD. Executive functioning‐SCD predicted hemodynamic changes and language‐SCD related to alterations within the frontal and temporal lobes. Results suggest the relevance of considering the type of SCD, as specific domains may predict different pathologies over time with important prognostic implications. Future research is needed to understand the clinical consequences of these findings.