Existing symptom-based and structural brain imaging methods have been of limited help in outcome prognostication or making return-to-play/duty decisions. Now recognized as a functional disorder,¹ concussion has been shown to alter cerebral hemodynamics^2,3 and mitochondrial function in both animals and humans.^4,5 Quantitative functional MRI (qfMRI) and gas control provides an opportunity to directly image cerebrovascular physiology and may be a more sensitive modality to investigate the pathophysiology of concussion.Athletes in high-impact (i.e., hockey, football, soccer and rugby) and non-impact (i.e., volleyball and rowing) sports are being recruited during season training from Queen’s University varsity teams through attending staff and physicians. Athletes with concussion are evaluated in the acute (< 7 days), early (14 days) and late (3 months, 1 year) post-injury stages. Here we report preliminary data on 3 non-concussed (NC) and 3 concussed (CI) athletes over acute and early time points. Subjects are scanned on a 32-channel receiver coil [Tim-Trio, Siemens, Erlangen, Germany] at Queen’s Centre for Neuroscience Studies. Prior to scanning, participants are fitted with a sealed rebreathing circuit and given the opportunity to become familiar with breathing tasks, cued by a visual and auditory metronome. Dual-Echo pseudo-Continuous Arterial Spin Labelling (DE-pCASL)⁶ data is acquired throughout a computer controlled breathing challenge (RespirAct^TM, Thornhill Research Inc)⁷ delivered as follows: A block of 2 min hypercapnia at 10 mmHg increase in end-tidal partial pressure in CO₂ (P_ETCO₂) above the subject’s resting level, preceded and followed by 2 min at resting P_ETCO₂, under maintained iso-oxia (Total = 6 min). Following standard anatomical imaging (T₁-MPRAGE), and acquisition of an M₀ image for CBF quantification (DE-PCASL; TR=15s, PLD=4000ms), the breathing challenge is performed under DE-pCASL acquisition (TR/TE₁/TE₂=4000/10/30ms, 3.9 mm isotropic, PLD=1000ms, tag=1.665s). Data analysis is performed using SPM12, FSL and MATLAB. Pre-processing includes motion correction, de-trending, spatial smoothing, slice time correction, and re-alignment of MR signal data with the end-tidal respiratory traces recorded by the RespirAct^TM. Whole brain (WB), gray matter (GM) and frontal cortex (FC) masks are generated using a combination of intensity-based thresholding and manual delineation; the FC mask is based on anatomical landmarks of inferior and middle frontal gyri. Block designs are used to model BOLD-CVR and quantitative CBF maps. CBF is quantified based on the recommendations found in the ASL white paper.⁸ MRI data is also modeled using neurocognitive scores, including ImPACT (pre-season/72 hrs post-concussion) and MOCA (24 hrs post-MRI sessions 1 & 2).Whereas MOCA and ImPACT index scores were normal (“Average” to “Very Superior” on classification ranges) across all subjects and time points, Fig1 and Table1 show typical abnormal patterns in BOLD, baseline perfusion and CVR observed in concussed (CI) versus non-concussed (NC) subjects. A reduction in BOLD-CVR is generally observed 7 days post-concussion (D versus A; blue arrow); BOLD-CVR grossly recovers after 14 days (G versus D; red arrow), yet abnormalities remain (heterogeneous coverage in G). The greatest inter-subject differences are observed in the FC (Table1). A closer look at the underlying baseline perfusion maps (B) indicates a hyperemic response 7 days post-injury (E; red arrow & Table1), which does not resolve to normal levels, even after 14 days (H; red arrow & Table1). Consequently, sustained impaired CVR is observed (F & I versus C; blue arrows & Table1). A pattern where BOLD is restored to normal levels (G; red arrow) with persistently low CVR (I; blue arrow) might be indicative of a rising hypometabolic response. This potential homeostatic safety mechanism of the brain to spare healthy tissue would further corroborate reports of altered isometabolism under hypercapnia.⁹ Attribution of vascular impairment to altered CBF is confounded by normal variations within/between subjects, as well as across anatomical region. To reduce such subjective assessment, baseline and CVR reference atlases will be generated from co-registering normative maps in 40 NC athletes.¹⁰In this study, standard concussion assessment tools (i.e., MOCA, ImPACT) failed to index the extent of brain injury. Indications of impaired cerebrovascular regulation, through changes in BOLD, baseline perfusion and reactive capacity, were observed across athletes in the acute and early stages post-concussion. Clinical integration of these qfMRI measures sensitive to brain dysfunction following concussion may lead to accurate prognostication, tailored therapy and improved health outcomes.