Abstract 1910: Hyperperfusion in Acute Ischemic Stroke: Serial Arterial Spin-Labeled MRI of Reperfusion and Hemorrhagic Transformation

Background: Arterial spin-labeled (ASL) MRI facilitates repeated noninvasive evaluation of cerebral blood flow without the use of contrast. Hyperperfusion may be readily detected with ASL and serial imaging may therefore chronicle the dynamics of territorial perfusion from acute to chronic phases after stroke. We characterized hyperperfusion on ASL in a prospective series of acute ischemic stroke patients, describing the clinical correlates, time course and association with reperfusion hemorrhage. Methods: A consecutive series of acute ischemic stroke patients admitted during a 1-year period were evaluated with pseudo-continuous ASL with background suppressed 3D GRASE (delay=2s, matrix=64x64; 26 slices, resolution 3.4x3.4x5mm, scan time 4min). Post-processed ASL CBF maps were visually inspected for detection of hyperperfusion. DSA measures of collaterals and reperfusion were scored when available and hemorrhagic transformation (HT) was graded on GRE in all 198 cases. Univariate and multivariate statistical analyses delineated clinical correlates, timing and other imaging features of hyperperfusion. Results: Among 198 patients, mean age was 69.4±15.7 years and 48.5% were women. Among 77 with serial ASL MRI, interval from initial to follow-up MRI was median 25.0 (IQR 10.3-53.9) hours. Hyperperfusion was detected in 15/198 (7.6%) patients at baseline and 30/77 (39.0%) at follow-up. Trajectories included 7/77 (9.1%) with hyperperfusion at both baseline and follow-up and 38/77 (49.4%) showing hyperperfusion at any timepoint during admission. Hyperperfusion correlated with achievement of reperfusion among patients undergoing endovascular therapy (OR 6.5, 95% CI 1.82-23.25, p=0.018) and history of atrial fibrillation (OR 4.4, 95% CI 1.9-10.6, p<0.001). Analysis of the 42 cases with DSA revealed that hyperperfusion was most common in patients with poor collateral grade followed by more complete TICI reperfusion scores. Overall, HT affected 57/198 (28.8%), including 35/198 (17.7%) HI1, 11/198 (5.6%) HI2, 8/198 (4.1%) PH1 and 3/198 (1.5%) PH2. Multivariate analyses revealed that hyperperfusion at any timepoint was a potent predictor of HT (OR 52.6, 95%CI 12.4-222.6, p<0.001). Conclusions: Hyperperfusion in acute ischemic stroke is frequently demonstrated by ASL MRI, providing novel insight on the dynamics of reperfusion and HT. Hyperperfusion increases the risk of HT 50-fold, likely due to autoregulatory loss. Poor collaterals and sudden reperfusion in vulnerable cases such as those with atrial fibrillation may herald hyperperfusion and HT.