Leptomeningeal Collateral Status by Signal Variance in Perfusion Magnetic Resonance Imaging: Association With Initial Stroke Severity and Early Functional Outcome After Thrombectomy

Abstract

BackgroundCollaterals are the main determinants of the severity of cerebral ischemia and control the pace of the ischemic tissue damage in acute ischemic stroke. Assessment of collateral status remains a major challenge in stroke imaging. We evaluated a signal variance–based collateral vessel index in perfusion‐weighted imaging (CVIPWI) in terms of its association with initial stroke severity, presence of a mismatch for endovascular thrombectomy (EVT), and early functional outcome in patients with large‐vessel occlusion.MethodsT2*‐weighted time series from dynamic susceptibility contrast perfusion imaging were processed to calculate the CVIPWI. Ischemic cores were segmented automatically on apparent diffusion coefficient maps. The relationship between collateral status and the fulfilment of mismatch criteria for EVT as well as the association between the CVIPWIand functional outcome in patients undergoing EVT were analyzed. Furthermore, spatial patterns of pial collateralization were investigated.ResultsA total of 156 patients with large‐vessel occlusion were included in the final analysis. Higher CVIPWIand thus better collateral supply was associated with lower baseline National Institutes of Health Stroke Scale and smaller baseline infarct volumes (P=0.022 andP=0.002, respectively), and the CVIPWIvaried significantly among groups according to fulfillment of mismatch criteria for EVT (P<0.001). In patients undergoing EVT (n=105), the CVIPWIwas an independent predictor of favorable functional outcome (modified Rankin scale score of 0–2) at discharge in multivariate analysis (P=0.031). In patients with EVT who had successful reperfusion (n=79), good collateral status was associated with a higher rate of early neurological improvement (P=0.026) and better functional outcome at discharge (P=0.04) in shift analysis.ConclusionSignal variance–based CVIPWIrepresents a semiquantitative and objective, thus observer‐independent parameter for direct assessment of collateral status with clinical relevance. Its use may inform clinical decision‐making and may be of interest for clinical stroke trials.