Multiomics Profiling Supports Cathepsin B as a Protective Factor in White Matter Hyperintensities and Cognition

White matter hyperintensities (WMH) are a common brain magnetic resonance imaging marker of cerebral small vessel disease and a major risk factor for stroke and dementia, yet the molecular mechanisms governing lesion formation remain incompletely understood. We sought to identify circulating proteins that lie upstream of WMH and influence related clinical manifestations.

We conducted a 4-stage proteogenomic study. Stage 1 tested a WMH polygenic risk score against 2923 plasma proteins in 52 560 UK Biobank participants (mean age, 56.8 years, 53.9% female, 94.0% White), and candidate proteins against WMH volume in the neuroimaging subset (n=5443; mean age 54.2 years, 53.5% female). Stage 2 applied bidirectional Mendelian randomization, colocalization, and local genetic correlation using cis protein quantitative trait locus and cis expression quantitative trait locus instruments from plasma, vascular, and brain tissue. Stage 3 analyzed single-nucleus RNA sequencing of 8 postmortem periventricular white matter samples (4 WMH lesions, 4 controls; mean age, 81 years; 50% female). Stage 4 tested circulating and genetically predicted protein levels against general cognitive ability.

Stages 1 and 2 converged on CTSB (cathepsin B) and ECHDC3 (enoyl-coenzyme A hydratase domain-containing 3). Higher plasma CTSB was associated with lower WMH volume (β=-0.09 [95% CI, -0.13 to -0.05]; n=5260), and Mendelian randomization estimates were directionally consistent across all 5 instrument sets (plasma protein quantitative trait locus β=-0.052 [95% CI, -0.064 to -0.040]), with concordant effects on white matter microstructure, perivascular spaces, and lacunar stroke. Single-nucleus RNA sequencing showed CTSB downregulation within lesions, predominantly in astrocytes (log₂ fold change=-1.70) and oligodendrocytes (log₂ fold change=-0.48; both false discovery rate P<1×10^-20). Higher plasma CTSB (β=0.061 [95% CI, 0.010-0.112]; n=5193) and a cis protein quantitative trait locus polygenic score (β=0.031 [95% CI, 0.009-0.054]; n=8884) predicted better cognitive performance. ECHDC3 showed no consistent association beyond stage 1.

Multimodal proteogenomic evidence nominates CTSB as a lesion-suppressed protein associated with lower WMH burden and better cognition, and a candidate therapeutic target for cerebral small vessel disease.