Back

Characterizing spatiotemporal white matter hyperintensity pathophysiology in vivo to disentangle vascular and neurodegenerative contributions

Parent, O.; Alasmar, Z.; Osborne, S.; Bussy, A.; Costantino, M.; Fouquet, J. P.; Quesada, D.; Pastor-Bernier, A.; Fajardo-Valdez, A.; Pichet-Binette, A.; McQuarrie, A.; Maranzano, J.; Devenyi, G. A.; Steele, C. J.; Villeneuve, S.; the PREVENT-AD Research Group, ; the Alzheimers Disease Neuroimaging Initiative (ADNI), ; Dadar, M.; Chakravarty, M.

2025-06-11 neurology
10.1101/2025.06.10.25329342 medRxiv
Show abstract

White matter hyperintensities (WMHs) are neuroimaging markers widely interpreted as caused by cerebral small vessel disease, yet emerging evidence suggests that a subset may have a neurodegenerative etiology. Current imaging methods have lacked the specificity to disentangle biological processes underlying WMHs in vivo. Here, we used voxel-level normative modeling and seven microstructural MRI markers with complementary biophysical sensitivities to generate single-subject high-resolution WMH pathophysiology maps in a large cohort (n=32,526). We calculated data-driven spatial patterns of similar WMHs, revealing distinct periventricular, posterior, and anterior clusters. We identified a reproducible WMH signature linked to dementia and Alzheimers disease, characterized by a posterior predominance and a pathophysiological pattern indicative of selective fiber degeneration. Posterior WMHs connected cortical regions vulnerable to tau pathology. Our framework distinguishes vascular and neurodegenerative contributions of WMHs in vivo, which could alter the course of treatment strategies and provide nuanced interpretations of research findings.

Published in Nature Communications (predicted rank #1) · training set

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.