Abstract Background and aims Cerebrovascular reactivity (CVR) is impaired in cerebral small vessel disease (CSVD), commonly assessed using CO₂ inhalation during MRI. However, this is methodologically challenging and uncomfortable. We assessed whether resting-state BOLD may provide a non-invasive alternative. Methods The crossover-design OxHARP trial tested three weeks of sildenafil versus placebo on MRI-CVR, assessed by 6%-CO₂ gas challenge. Resting-state, high-frequency BOLD-MRI (TR=0.4s, multiband 6) was also acquired with concurrent measurement of end-tidal CO2. After motion correction, high-pass filtering, brain extraction and voxel-wise correction for temporal delays, CVR was determined as correlation between standardised BOLD per mmHg change in etCO2, within grey (GM) and white matter (WM). Agreement between scans was determined by ICC and linear regression, and effects of drug treatment by mixed-effects linear models. Results Across 93 scans, there was significant agreement between challenge-CVR and resting-state CVR in GM (r=0.629, p0.001, ICC=0.773) and WM (r=0.598, p0.001, ICC=0.749). There was a similar voxel-wise pattern of increased CVR on sildenafil with both acquisitions, but the average difference by ROIs was only significant with challenge-CVR (Challenge: GM p=0.0108, WM p=0.0020; Resting-state: GM p=0.20, WM: p=0.099). Conclusions Optimised resting-state scanning demonstrates strong agreement with CO₂ gas challenge measures of CVR and may provide a feasible alternative for large-scale CVR assessment, or in populations where individuals are less able to tolerate a gas challenge. However, gas challenges remain optimal for cleanly detecting drug-induced CVR changes in clinical trials. Conflict of interest Vanessa Hyde: nothing to disclose Figure 1 - belongs to Results Figure 2 - belongs to Results
Hyde et al. (2026) studied this question.