Comprehensive diffusion MRI analysis reveals key microstructural alterations underlying white matter hyperintensities

Abstract

Cerebral small vessel disease (CSVD) is a prevalent age-related condition and a major contributor to cognitive decline, dementia, and stroke. Its most common MRI marker, white matter hyperintensities (WMH), reflects heterogeneous microstructural alterations, including demyelination, inflammation, and axonal loss. Monitoring of individual and regional variability associated with WMH in vivo provides critical insights into pathophysiological processes associated with aging and CSVD. However, diffusion MRI studies of WMH remain limited by the use of dichotomous lesion-based analyses, insufficient correction for free-water contamination, and an overreliance on fractional anisotropy. We analyzed longitudinal ADNI3 data from 216 cognitively healthy older adults with WMH over a two-year period using three complementary approaches: (1) lesion-gradient-based analyses to characterize microstructural alterations within WMH and the surrounding normal-appearing white matter; (2) an aging-adapted tractography pipeline to examine diffusion profiles along major white matter bundles using tractometry; and (3) data-mining approaches applied to bundle profiles to identify the diffusion metrics most strongly associated with WMH burden. Lesion-gradient-based analyses revealed pronounced microstructural gradients extending from WMH into the surrounding normal-appearing white matter, which remained largely stable over time. Tractometry and data-mining analyses showed that the free-water fraction as well as tissue radial and mean diffusivities were strongly associated with WMH volume. Notably, the FP-Growth data-mining approach further revealed specific combinations of co-occurring diffusion alterations associated with WMH burden, offering a level of interpretability beyond what conventional correlation-based analyses can provide. Together, these findings indicate that WMH-related pathology extends beyond visibly affected tissue, with a progressive accumulation of free water within the tissue over time, accompanied by subtle alterations in tissue microstructure consistent with mild myelin-related changes, without evidence of substantial axonal compromise. Furthermore, this study suggests that FA, even when corrected for free-water, is not the optimal metric for assessing WMH-related microstructural alterations.