Small vessel dysfunction at 7T MRI locally predicts white matter damage progression in CADASIL

Abstract

Cerebral small vessel diseases (cSVDs) contribute significantly to stroke and dementia. Advanced 7 T MRI techniques have revealed small vessel dysfunction in cSVD patients, linked to global white matter damage cross-sectionally. However, it remains unclear whether these vascular deficits predict progressive tissue damage. This longitudinal study examined the spatial relationship between local vascular function and white matter damage progression in patients with Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL). Twenty-two patients underwent baseline small vessel function assessment using 7 T MRI. Voxelwise blood-oxygenation level-dependent cerebrovascular reactivity (BOLD-CVR) to a hypercapnic was evaluated. White matter changes were assessed on 3 T MRI over two years, analyzing mean diffusivity changes and conversion of normal-appearing white matter to white matter hyperintensities (WMH). Results showed significant global increases in white matter damage over time. Voxelwise analysis revealed that lower baseline BOLD-CVR magnitude and higher dispersion were associated with increased white matter damage and WMH progression at specific locations at follow-up. However, whole-brain vascular function measures did not predict white matter changes at a global level. These findings suggest that local vascular function plays a key role in white matter damage progression in CADASIL, highlighting the importance of regional vascular health in cSVDs.

Keywords: CADASIL; MRI; Small vessel disease; cerebrovascular reactivity; white matter damage.

Figure 1.

Registration-overview linking longitudinal white matter damage to baseline small vessel function. (a) DWI scans were registered to FLAIR scans at baseline and follow-up (b) using linear affine registration with ‘FSL FLIRT.’ (d) Follow-up FLAIR scans were then registered to baseline FLAIR (c) using non-linear deformation with the symmetric normalisation method of the ANTs software package.^, (e) The functional 7T BOLD-CVR data (acquired voxel size 2 mm isotropic; upsampled to 1mm isotropic) were registered to baseline FLAIR data (voxel size 1 mm isotropic) using linear registration with ‘FSL epi_reg’ and trilinear interpolation. ANTs: advanced normalization tools; BOLD: blood-oxygenation level-dependent; CVR: cerebrovascular reactivity; DWI: diffusion-weighted imaging; FLAIR: fluid-attenuated inversion recovery; FSL: FMRIB software library; MD: mean diffusivity; NAWM: normal-appearing white matter; WMH: white matter hyperintensity.

Figure 2.

White matter hyperintensity changes after a two-year follow-up. (a) The top row shows a frequency map of WMH of the whole cohort at baseline and after two-year follow-up, including a frequency map of the locations with incident WMH and (b) FLAIR scans of an individual participant are shown in the bottom row for baseline on the left and follow-up in the middle. The arrows indicate new WMH lesions appearing after two-year follow-up. BOLD-CVR: blood-oxygenation level-dependent cerebrovascular reactivity; FLAIR: fluid-attenuated inversion recovery; GM: gray matter; NAWM: normal-appearing white matter; WMH: white matter hyperintensity; WM: white matter.

Figure 3.

Voxelwise association between baseline BOLD-CVR and longitudinal ΔMD. Voxelwise linear mixed model associations between baseline BOLD-CVR and longitudinal ΔMD are shown for NAWM in FLAIR space (voxel size 1 mm isotropic). Associations are shown for all NAWM at baseline (a), separated into NAWM that remained stable (green) and progressed to WMH (purple) after a two-year follow-up (b). Unstandardised regression coefficients (β) are shown for each association with 95%-CI; *: all were significant at p < 1.0 × 10⁻¹⁵. The shaded area represents the standard error from the regression fit. BOLD: blood-oxygenation level-dependent; CI: confidence interval; CVR: cerebrovascular reactivity; MD: mean diffusivity; NAWM: normal-appearing white matter.