Determinants of cerebral microbleed presence and burden in CADASIL.

Background & aims: Cerebral microbleeds (CMBs) are observed in one-third of CADASIL patients and increase with age. Factors associated with CMBs vary across studies, likely due to inadequate handling of their zero-inflated, long-tailed distribution and lack of accounting for measurement variability. We aimed to assess these factors while addressing these limitations. Methods: CMBs were assessed in a large CADASIL cohort, with demographic, vascular, genetic, imaging, and clinical data. White matter hyperintensity volume (WMH), lacune count, and brain parenchymal fraction (BPF) were measured. A Hurdle model was used to distinguish factors associated with CMB presence from those related to lesion counts, while accounting for measurement variability. Three models were built with: (1) demographic/biological factors; (2) MRI markers; and (3) clinical variables. Results: Beyond the consistent effect of the MRI sequence, in Model 1, CMB presence was associated with age, male sex and hypertension, while counts were associated only with age. In Model 2, CMB presence was associated with age, hypertension, and lacune count; counts were higher with WMH and antiplatelet use, and lower with EGFr 1-6 mutations and BPF. In Model 3, CMB presence was associated with age, hypertension, stroke history and gait disturbances; counts were associated with education, lacune count, MMSE ≤24, and hypercholesterolemia measured at the study visit. Conclusions: Hurdle modeling distinguishes determinants of CMB occurrence from accumulation. Hypertension appears to lower the threshold for lesion emergence along aging. CMB burden is associated with the burden of cerebrovascular lesions, brain atrophy and clinical severity, supporting worsening of the microangiopathic process, with specific impact related to mutation type.