Free water levels in normal-appearing white matter predict vascular lesion progression in individuals with dementia

Abstract

Introduction: Cerebral small vessel disease (SVD) is a common co-pathology in elderly and individuals with dementia. Neuroimaging markers of SVD include white matter hyperintensities (WMH) and MRI-visible perivascular spaces (PVS). However, the mechanisms underlying changes in these markers over time, whether ischemic or beta-amyloid (Aβ)-related, remain elusive. Here, we evaluated the effects of microstructural injury in the normal-appearing white matter and Aβ in the cerebral cortex on the progression of WMH and PVS over three years. Methods: Data was obtained from two independent cohorts: (i) TRIAD, comprising cognitively normal, MCI, and AD dementia participants (baseline: N=199, follow-up year 1 and 2: N=102 and 62); and (ii) MITNEC-C6, comprising “real-world” patients with mixed dementia and moderate-to- severe WMH burden (baseline: N=52, 2 years follow-up: N=25). We quantified global WMH and PVS volumes from FLAIR and T1w MRI. At baseline, we examined associations between these volumes and diffusion MRI-derived free water. Longitudinally, we employed linear [mixed-effect] models to investigate the relation of WMH or PVS volume changes over time with baseline free water, using cortical Aβ-PET, age, sex, and APOE-ε4 as covariates. Results: In TRIAD and MITNEC-C6 respectively, mean ages were 72±6 and 77±8 y, 60% and 42% were female, and 41% and 48% were Aβ-positive. At baseline, higher free water in normal- appearing white matter was associated with higher WMH volume (β_TRIAD=+0.34±0.06, P_TRIAD<0.001 and β_MITNEC=+0.31±0.14, P_MITNEC=0.03) as well as total PVS volume (β_TRIAD=+0.53±0.06, P_TRIAD<0.001 and β_MITNEC=+0.30±0.13, P_MITNEC=0.03).Longitudinally, faster WMH progression was predicted by higher baseline free water in normal- appearing white matter (P_TRIAD<0.001 and P_MITNEC=0.01) in Aβ-positive but not negative individuals (Figure 1). Conversely, higher free water in normal-appearing white matter was moderately related to slower PVS progression (P_TRIAD=0.03 and P_MITNEC=0.004). Discussion: Our findings support a key role for free water and Aβ in predicting the progression of volumetric MRI-based markers of SVD. Future investigations will explore the spatial relationships involved. Additionally, further validation of our longitudinal segmentation tools for PVS is necessary to support a potential link between free water and PVS temporal dynamics.