Comparative Evaluation of Resting-State and CO 2 -Induced Cerebrovascular Reactivity in Patients with Traumatic Brain Injury

Abstract

Cerebrovascular reactivity (CVR) mapping is a promising biomarker for evaluating vascular dysfunction following traumatic brain injury (TBI). Traditional CVR assessment requires carbon dioxide (CO₂) administration. Assessing CVR from resting-state blood-oxygen-level-dependent (BOLD) sequences (RS-CVR) offers a task-free alternative, but its validity in TBI has not yet been established. We aimed to evaluate whether RS-CVR can reliably detect cerebrovascular impairment in patients with TBI by comparing it with CO₂-inhalation CVR (CO₂-CVR). We enrolled 23 chronic moderate-to-severe TBI patients and 13 healthy controls (HC) who underwent both CO₂-CVR and RS-CVR imaging using BOLD functional magnetic resonance imaging (BOLD fMRI). RS-CVR maps were computed using a voxel-wise general linear model (GLM) across 120 bandpass filters. Spatial correlations between RS-CVR and CO₂-CVR were calculated to identify the optimal frequency bands. Z-score analyses and lesion-based comparisons were performed to assess CVR reductions in TBI. In the TBI cohort, lesion-based CVR was correlated with clinical outcomes using GLM adjusted for age and sex. The highest whole-brain spatial correlation between RS-CVR and CO₂-CVR in HC occurred at [0–116.4 mHz] (r = 0.5239 ± 0.1107). In TBI, the peak correlation slightly shifted to [0–74.5 mHz] (r = 0.5217 ± 0.1108) but remained comparable at [0–116.4 mHz] (r = 0.5093 ± 0.1263). As expected, regions of encephalomalacia, fluid-attenuated inversion recovery hyperintensity, showed CVR reductions on RS-CVR and CO₂-CVR maps, but low CVR was also identified through both methods in normal-appearing brain tissue. Across lesion areas, RS-CVR detected deficits consistent with CO₂-CVR, with mean z-scores of –0.217 ± 0.334 and –0.391 ± 0.294 for encephalomalacia and hyperintensities, respectively. Lesion-based CVR values were associated with clinical outcomes, with both CO₂-CVR and RS-CVR positively correlated with days in the intensive care unit (ICU; p < 0.05) and showing negative associations with Rivermead post-concussion symptoms questionnaire scores, statistically significant for CO₂-CVR (p = 0.031) and trending for RS-CVR (p = 0.089). RS-CVR closely mirrors CO₂-CVR in both global and lesion-specific analyses, validating its use as a noninvasive method for detecting vascular deficits in TBI. This task-free and scalable tool for cerebrovascular assessment offers a valuable approach for characterizing vascular health relevant to TBI prognosis and guiding neurorehabilitation efforts.