MRI-derived oxygen extraction fraction is associated with cognitive impairment in cerebral small vessel disease.

BACKGROUND: Conventional MRI markers of cerebral small vessel disease (CSVD) capture accumulated structural injury but may not fully reflect ongoing physiological dysfunction associated with cognitive impairment. AIMS: We investigated whether MRI-derived oxygen extraction fraction (OEF), a non-invasive marker of cerebral oxygen extraction, is associated with cognitive impairment in CSVD beyond conventional structural MRI burden. METHODS: In this hospital-based observational study, 448 adults with CSVD underwent 3T multiecho gradient-echo MRI for whole-brain OEF mapping using a quantitative susceptibility mapping/quantitative blood oxygen level-dependent framework. Participants were classified as having mild cognitive impairment (MCI) or no MCI by a blinded clinical consensus panel. Whole-brain voxelwise analyses compared OEF between groups and examined associations with the composite cognitive score, Fazekas score, and total CSVD score; post hoc ROI-based analyses examined associations with conventional CSVD burden markers and a composite cognitive score. In an exploratory longitudinal subset of 82 participants, we assessed whether baseline OEF or within-subject OEF change was associated with cognitive change. RESULTS: Participants with CSVD-MCI showed higher OEF in distributed cortical-subcortical regions than cognitively preserved participants after adjustment for demographic, vascular, and conventional CSVD burden variables. The largest cross-sectional cluster showed higher OEF in the CSVD-MCI group than in the CSVD-no-MCI group (0.342±0.045 vs. 0.294±0.027). Complementary whole-brain analyses and post hoc analyses of the largest group-difference cluster showed that higher OEF was associated with lower composite cognitive scores after covariate adjustment, whereas no associations with Fazekas score or total CSVD score were identified. In the longitudinal subset, baseline OEF was not associated with follow-up composite cognitive score or cognitive change, whereas greater within-subject OEF increase was associated with greater cognitive decline (standardised β=-0.229, p=0.030). CONCLUSIONS: MRI-derived OEF was associated with cognitive impairment in CSVD independently of conventional structural MRI burden. These findings suggest that OEF may provide complementary physiological information for characterising cognitive vulnerability in CSVD. Larger multicentre longitudinal studies incorporating quantitative perfusion imaging and validated amyloid and tau biomarkers are warranted.