Cerebrovascular Reactivity to Hypoxic and Hypercapnic Gas Challenges Using 7T BOLD fMRI

Background. Cerebrovascular reactivity (CVR) is a sensitive marker of vascular health, most commonly measured with BOLD fMRI during hypercapnic gas challenges. Hypoxia is a complementary but under-characterized vasoactive stimulus. We characterize a combined hypercapnic-hypoxic gas-challenge paradigm at 7 Tesla BOLD fMRI and its cross-session reproducibility. Methods. Eleven healthy adults were scanned twice at 7 Tesla during an 18-minute multi-echo, multiband BOLD acquisition, receiving fixed-inspired hypercapnic (5% CO2) and hypoxic (10% O2) gas boluses separated by medical-air recovery periods, with end-tidal gas and peripheral saturation monitoring. Hypercapnic and hypoxic CVR were estimated using three trace-based general linear models: end-tidal gas, peripheral saturation, and Severinghaus-derived arterial saturation; and a finite impulse response model. Cross-session reproducibility was assessed by region-of-interest intraclass correlation. Results. The paradigm produced robust, condition-specific gray-matter BOLD responses that reproduced across sessions. The Severinghaus-derived saturation regressor generalized significantly better than end-tidal and peripheral regressors under cross-session validation and outperformed the finite impulse response benchmark in 58% of gray-matter voxels. Reproducibility across cortical, subcortical, and brainstem atlases was moderate-to-high (median cortical intraclass correlation 0.71). Conclusion. A combined hypercapnic-hypoxic paradigm yields reproducible voxelwise cerebrovascular reactivity maps at 7 Tesla; Severinghaus-derived arterial saturation is the most generalizable regressor for hypoxic reactivity.