Associations between MRI-visible perivascular spaces, brain atrophy, white matter hyperintensities, and speeded executive function, in neurodegenerative and cerebrovascular diseases.

Abstract

Introduction: MRI-visible perivascular spaces (PVS) are a neuroimaging feature of cerebral small vessel disease and are commonly observed in patients with cerebrovascular and neurodegenerative disease. However, it is unclear whether PVS burden is associated with cognition. The aim of this study was to investigate the potential associations between PVS volumes, brain atrophy, white matter hyperintensities (WMH), and speeded executive function, in patients from the Ontario Neurodegenerative Disease Research Initiative (ONDRI). Methods: ONDRI participants (n=333) clinically diagnosed with Alzheimer's disease/mild cognitive impairment (ADMCI), frontotemporal dementia (FTD), and cerebrovascular diseases (CVD), with clinical, neuropsychological, MRI, plasma biomarkers (glial fibrillary acidic protein (GFAP); neurofilament light, (NfL); p-tau181; Aβ42/40), and available apolipoprotein E epsilon 4 (APOE E4) status, were included in this analysis. MRI-based measurements for brain parenchymal fraction (BPF), lacunes, PVS and WMH were extracted using the ONDRI imaging pipeline. The neuropsychological test scores were standardized (z-transformed), speeded executive function z- scores were computed as the mean of digit symbol modalities test and trail making test part B timed z-scores. The variables selected in bivariate analysis (p<0.2) were introduced in a linear regression model with speeded executive function as the dependent variable, with and without BPF and WMH. To assess the effect of PVS volumes on processing speed through BPF and WMH volumes, a mediation analysis was applied. Results: Speeded executive function was significantly associated with PVS (β=-0.115, p=0.04), GFAP (β standardized effect=-0.231, p=0.002), and the clinical diagnostic cohort (β=-0.160, p=0.004), the adjusted R2 of the model was 0.104 (p<0.001). However, it was not significantly associated with PVS (p=0.691) when BPF and WMH volumes were included in the model, suggesting the relationship was mediated by these factors. In the mediation analysis, the total model effect of PVS on speeded executive function was significant (effect=-0.1067, p=0.042), due to an indirect effect of PVS, mediated by BPF (effect=-0.0718) and WMH (effect=-0.0256). Discussion: Our preliminary results suggest that the relationship between PVS volume and speeded executive function is mediated by global atrophy and WMH in neurodegenerative and cerebrovascular disease. Future analyses will further examine the role of plasma biomarkers and other imaging markers such as cerebral microbleeds.