Factors associated with the presence, and longitudinal progression of, perivascular spaces: An exposome-wide analysis in the UK Biobank.

Abstract

Background: Perivascular spaces (PVSs) are markers of cerebral small vessel disease (cSVD). While some vascular risk factors are known, broader determinants of regional PVS burden and progression remain unclear.

Aim: This study aimed to identify exposome-wide determinants of cross-sectional regional PVS burden in the basal ganglia (BG) and centrum semiovale (CSO) and examine associations of identified exposures with PVS progression.

Methods: We analyzed 44,938 UK Biobank participants (4568 with repeat imaging after 2.6 ± 1.0 years). Of 3767 candidate exposures across eight domains, 1289 were retained after preprocessing. The cross-sectional Exposome-Wide Association Study (ExWAS) used negative binomial regression with false discovery rate (FDR) correction. Surviving exposures were further examined with weighted quantile sum (WQS) regression to assess relative importance, Conway-Maxwell-Poisson longitudinal models, and structural equation modeling (SEM) linking exposures to cognition via PVS.

Results: ExWAS identified 80 exposures for BG-PVS and 89 for CSO-PVS (P_{FDR-corrected} < 0.01). Higher diastolic blood pressure (BG: incident rate ratio (IRR) = 1.04, 95% confidence interval (CI) = 1.03-1.05; CSO: IRR = 1.07, 1.05-1.08) and greater left ventricular (LV) myocardial mass (BG: IRR = 1.05, 1.04-1.07; CSO: IRR = 1.09, 1.06-1.11) were associated with greater burden. WQS weights were dominated by cardiovascular exposures in BG (54.5%) but were more broadly distributed in CSO, including cardiovascular (31.9%), sociodemographic (20.9%), and environmental (12.7%). Longitudinally, cardiovascular and adiposity exposures were nominally associated with faster BG-PVS progression, with stronger effects in females. SEM demonstrated cardiovascular exposures linked to cognition through BG-PVS, but not CSO-PVS.

Discussion: Determinants of PVS burden and progression showed regional heterogeneity: BG-PVS is more strongly associated with cardiovascular exposures and cognition, while CSO-PVS shows a broader, more heterogeneous pattern. These findings support the importance of regional PVS assessment and suggest cardiovascular risk management may be particularly relevant to BG-PVS-related cSVD.