Abstract W P33: Functional MRI and CBF Responses to Transient Oxygen Challenges in Acute Ischemic Stroke Patients

Background: There is increasing interest in using oxygen challenges during MRI to image metabolic dysfunction in acute ischemic stroke. Methods: Subjects enrolled in a prospective MRI study of patients imaged within 48 h of stroke onset underwent arterial spin labeling (ASL) MRI for 10-12 min, while receiving a transient normobaric oxygen (NBO) challenge, according to this protocol: (1) Room air (RA) for 2-5 min; (2) NBO at 10-12 L/min via nonrebreather mask for 4 min; (3) RA for 2-4 min. ASL data were motion corrected and used to calculate CBF. Changes of blood-oxygen level dependent (BOLD) and perfusion MRI signal intensities during NBO were calculated (along with Z-scores) using FEAT. Mean changes within the DWI lesion (Core), ipsilateral normal tissue (IPS) and contralesional hemisphere (CNL) were compared. Results: Patient characteristics (N=19) were age 62±13 years, median admission NIHSS 12 [IQR 3-12], time-to-MRI 33±13 h, median Core volume 31.2 [IQR 14.3-68] cc. 6 patients were imaged after tPA therapy. One subject’s CBF data were unusable due to artifacts. Both hyperemia (N=8) and hypoperfusion (N=12) were observed in areas in and around the DWI lesion. Baseline CBF in Core was significantly higher than in CNL (P=0.01) and IPS (P=0.03), indicative of hyperemia. No significant difference was found for perfusion change in response to NBO (mean Z-score < 1.1 for all regions). In contrast, strong positive and negative BOLD responses were found both ipsilaterally and contralaterally. Negative BOLD responses were significantly smaller (P<0.01) in the Core (Z-score 1.3±1.6) than in either IPS (2.5±1.6) or CNL (2.5±1.4). Positive BOLD responses did not significantly differ across regions, but had high Z-scores (Core: 5.2±2.6, IPS: 5.8±2.7, CNL:5.8±2.9), indicating strong response to NBO. Discussion: Our results confirm previous reports in both human and experimental stroke models of strong increases in BOLD signal in response to NBO challenges. In addition, we found negative BOLD responses both ipsilesionally and contralesionally, often in conjunction with hyperemia, perhaps suggesting a steal effect. Future research may further elucidate the complexities of oxygen metabolism in stroke, perhaps encouraging the development of novel therapies.