Small vessel disease neuroimaging markers contribute robustly and independently to longitudinal cognitive decline in older adults

Abstract

AbstractBackgroundCerebral small vessel disease (SVD) is visible on brain magnetic resonance imaging (MRI) as white matter hyperintensities (WMHs), perivascular spaces (PVS), cerebral microbleeds (CMBs), and lacunes. This study expands our prior cross‐sectional work and examines how these four markers contribute to longitudinal cognitive trajectory to assess whether these markers represent common versus unique pathways of injury.MethodsVanderbilt Memory and Aging Project participants free of baseline clinical stroke and dementia (n=327, 73±7 years, 41% female) completed multimodal 3T brain MRI to quantify WMHs, PVS, CMBs, and lacunes at baseline and neuropsychological assessment at multiple time points (baseline, 18‐month, 3‐year, 5‐year). Linear mixed‐effects regression models related each baseline SVD marker to longitudinal neuropsychological trajectory through interaction with follow‐up time, adjusting for age, sex, race/ethnicity, education, Framingham Stroke Risk Profile, diagnosis, apolipoprotein E‐e4 status, and intracranial volume.ResultsMean follow‐up time was 3.9±1.1 years. Increases in each SVD marker related to faster episodic memory decline (p‐values≤0.04). Increased lacunes (p=0.04), WMHs (p≤0.008), and CMBs (p≤0.001) related to information processing speed decline. Increased WMHs (p=0.02) and CMBs (p=0.03) related to executive function decline. Increased WMHs related to language decline (p=0.02). Increased CMBs related to visuospatial decline (p=0.03). In combined models simultaneously considering multiple statistically significant SVD markers in relation to cognitive outcomes, WMHs independently predicted faster information processing speed (p=0.03), executive function (p=0.05), and episodic memory decline (p=0.02) beyond all other SVD biomarkers. CMBs independently predicted information processing speed decline beyond lacunes and WMHs (p‐values≤0.04). PVS independently predicted episodic memory decline beyond lacunes and CMBs (p=0.03).ConclusionsIncreased SVD, including PVS, CMBs, lacunes, and especially WMHs, was associated with faster longitudinal cognitive decline. In competitive models, WMHs consistently predicted faster decline in multiple domains despite inclusion of other significant SVD predictors. In a subset of models, CMBs and PVS independently predicted faster longitudinal cognitive decline yet lacunes did not. Thus, WMHs, CMBs, and PVS may each reflect a unique pathway of injury whereas lacunes may not. Funding: F31‐AG066358, T32‐AG058524, Alzheimer’s Association IIRG‐08‐88733, R01‐AG034962, R01‐AG056534, K24‐AG046373.