Real-Time Estimation of Core Infarct in Angiography Using Collateral Flow

Abstract

Background and purpose: In order to attribute a diagnostic value to angiographic runs performed before revascularization, we aimed at developing a regional evaluation of leptomeningeal collateral flow that can be used to detect and predict infarction when performing stroke endovascular procedures.

Materials and methods: We evaluated all consecutive patients treated for occlusions in the anterior circulation in our center between 2009 and 2013, with MRI imaging performed before the endovascular procedure. Two readers performed an evaluation of collateral circulation in 5 cortical regions based on the vascular anatomy. Regional scores were correlated with the presence of infarction in the same cortical sector on pretreatment and follow-up imaging. Global collateral scores for each patient were correlated with infarct volumes.

Results: In 89 patients with 408 cortical regions, we found a significant correlation between the degree of zonal collateral flow and the absence of infarction in the same zone on pretreatment imaging. In a subgroup of 37 recanalized patients (Thrombolysis in Cerebral Infarction scale 3) with 173 cortical zones, retrograde collateral flow to the proximal M4 segment predicted the absence of infarction within the same zone on follow-up imaging (positive predictive value 88.7%). We found good inter-rater agreement for the presence of collateral flow to the M4 proximal segment or further - k = 0.77 (p = 0.05, 95% CI 0.66-0.88). Global collateral scores correlated with infarct volume on initial imaging; all patients with scores ≥4 had infarct volumes ≤70 ml, whereas all patients with global collateral scores ≤1 had infarct volumes ≥70 ml.

Conclusion: Anatomic collateral flow evaluation using the angiographic runs performed during stroke endovascular procedures can provide a real-time estimation of the volume and location of core infarct. For each cortical region, good collateral flow is associated with the absence of infarct on pre-treatment imaging, and is predictive of the absence of infarct on follow-up imaging in recanalized patients.