FMRI Approaches to Mapping Cerebrovascular Reactivity: Comparison of Gradient Echo BOLD and Spin Echo BOLD With Arterial Spin Labeling

Purpose: To compare markers of cerebrovascular reactivity (CVR) measured with Gradient Echo (GE, T2*w) and Spin-Echo (SE, T2w) BOLD fMRI with quantitative physiological CVR measured using ASL. Methods: CVR, the ability of blood vessels to dilate and increase cerebral blood flow (CBF), can be investigated with ASL fMRI but has a low signal-to-noise ratio (SNR). GE-BOLD provides semi-quantitative CVR assessment, has high SNR but is influenced by baseline deoxyhemoglobin and the macrovascular venous compartment. SE-BOLD is selectively more sensitive to microvasculature, mitigating contamination from larger vessels. We developed a pCASL GE-BOLD/SE-BOLD MRI sequence and collected all three fMRI modalities in 20 healthy subjects during a hypercapnic breath-holding task. Results: CVRs in GM were CVRASL = 5.3%/mmHg ± 1.8%/mmHg (mean ± SD), CVRGE-BOLD = 0.18% BOLD/mmHg ± 0.05% BOLD/mmHg, and CVRSE-BOLD = 0.09% BOLD/mmHg ± 0.03% BOLD/mmHg. The ground truth of physiological CVR was represented by the ASL measurements with higher SNR (CVRASL). CVRGE-BOLD and CVRSE-BOLD correlated across subjects with CVRASL in GM, with SE showing a stronger correlation (CVRBOLD vs. CVRASL:r = 0.54, p < 0.05 for GE-BOLD and r = 0.70, p < 10−3 for SE-BOLD). Region of interest (ROI) analysis based on previously reported macrovascular venous density maps derived from SWI showed that the lower correlation for GE-BOLD was driven by GM ROIs with high venous density. Similar results were observed for spatial correlations (across regions) of group average maps of CVR (CVRBOLD vs. CVRASL:r = 0.44 for GE-BOLD and r = 0.58 for SE-BOLD, p's < 10−3). Conclusions: BOLD fMRI provides a semi-quantitative CVR assessment. Using SE-BOLD may be advisable, as it approximates physiological CVR more closely than GE-BOLD due to reduced sensitivity to larger veins while maintaining a similar group-level sensitivity.