Perivascular spaces and cognitive performance in the Northern Manhattan Study

Abstract

AbstractBackgroundPerivascular spaces (PVS) are fluid‐filled compartments surrounding brain small blood vessels, serving as conduits for fluid transport, exchange between CSF and interstitial fluid, and clearance of waste products. PVS are visible on MRI and associated with cerebral small vessel disease (SVD), and research is needed to understand their role as markers of SVD and potential contributors to vascular cognitive impairment. PVS have been reported in Alzheimer's disease, but less is known regarding their association with cognition in diverse populations. We tested the hypothesis that PVS are associated with cognition, independent of demographics and other markers of SVD, including white matter hyperintensity volume, silent brain infarcts, and cerebral microbleeds.MethodWe included 1,201 participants from the prospective Northern Manhattan Study. PVS were identified as parenchymal hypointensities <3mm on axial T1‐weighted MRI without a hyperintense rim on T2‐weighted or FLAIR images, that followed the orientation of perforating arteries. A semiquantitative PVS score represented the degree of brain involvement (range=0‐28). Generalized linear models were used to assess associations with four cognitive domains (memory, executive function, processing speed, language) and global cognition (mean of domains) over two assessments carried out an average 6 years apart. Model 1 adjusted for demographics; Model 2 also included vascular risk factors; Model 3 also included burden of SVD.ResultAmong 1,201 participants (mean age=71±9 years; 60% women, 66% Hispanic, 18% Black, 10±5 years education; 68% with hypertension, 19% diabetes, 65% hypercholesterolemia), PVS scores ranged from 0‐22 with 55% having a PVS score >4. PVS score was not associated with global cognition z‐score at first neuropsychological assessment, nor change in global cognitive performance between the two assessments. Higher PVS score was associated with worse processing speed at first assessment in models 1 and 2 (Model 2 beta=‐0.015, p=0.01). A higher PVS score was associated with greater decline in memory, adjusting for demographics (model 1 beta=‐0.015, p=0.01), but this relationship was attenuated and not significant after adjusting for modifiable vascular risk factors.ConclusionPVS are prevalent and associated with worse processing speed independent of age and vascular risk factors. Vascular risk factors largely explained variability in the association between PVS and memory decline.