Modeling small vessel disease burden and vascular dementia likelihood from NACC data

INTRODUCTION: Cerebrovascular pathology contributes to dementia but lacks standardized quantification for multicenter datasets. We developed harmonized indices to quantify small vessel disease (SVD) and vascular cognitive impairment (VCI) likelihood using National Alzheimer's Coordinating Center Neuropathology 11 data. METHODS: Six cerebrovascular domains were scored (0 to 3) and aggregated into a SVD index (0 to 18), with proportional scaling when ≥5 domains were available. A vascular cognitive impairment neuropathology guideline (VCING)-Lite proxy substituted global for occipital-specific ratings. A principal component analysis-derived global-cognition factor and clinical dementia status were modeled against vascular indices. Coherence was tested across complete, domain-complete, and Alzheimer's disease neuropathologic change (ADNC)-limited subsets. RESULTS: SVD scores were consistent across datasets, with lower values only in the ADNC-limited subset (p < 0.003). SVD-VCING-Lite correlations remained strong after adjustment (r = 0.72, p < 0.001). Both were associated with worse global cognition and higher odds of dementia. DISCUSSION: This framework links cerebrovascular pathology to in-life outcomes and enables reproducible modeling of vascular contributions in mixed dementias.