Perfusion-Based Collaterals in Prediction of Infarct Growth in Patients with Acute Ischemic Stroke: A Multicenter...

In patients with acute ischemic stroke (AIS), robust collaterals are associated with lower rates of infarct growth. Perfusion-based collateral matrices, including hypoperfusion intensity ratio (HIR), relative CBV (rCBV) index, and perfusion collateral index (PCI), have been used successfully to assess collaterals in patients with AIS. We aimed to assess and compare the diagnostic ability of these perfusion-based collateral indices in the prediction of infarct growth in patients with AIS after successful reperfusion.

In this retrospective multicenter study, patients with anterior circulation large vessel occlusion were included if they had pretreatment perfusion (CT or MR), successful reperfusion (modified TICI ≥ 2b), and follow-up MRI within 24-48 hours from the date of treatment. The primary outcome assessed was substantial infarct growth, which was defined as a final infarct volume of ≥10 mL compared with the baseline ischemic core volume. Perfusion-based collateral indices including HIR (volume of time-to-maximum [Tmax] >10 seconds/volume of Tmax >6 seconds) and rCBV index (mean of rCBV values within the volume of Tmax > 6 seconds) were calculated using RAPID software (Version 5.0.4), and PCI (volume of delay^{2-6 seconds} multiplied by its corresponding mean rCBV) was calculated using Olea software (SP.23). The association between perfusion-based collateral indices, along with demographic and clinical variables for determination of infarct growth, was tested using appropriate univariate analysis. Multivariable logistic regression using a stepwise backward selection approach was performed to identify independent predictors of infarct growth.

Among 116 included patients, 58 (50%) had infarct growth ≥10 mL. Infarct growth volume (median, interquartile range) was: 11, 2.6-34.7 mL. Poor collaterals determined by PCI and rCBV index was significantly (P < .01) associated with infarct growth, whereas HIR was not (P = .83). Following multivariate logistic regression, 3 variables remained as independent predictors of infarct growth: baseline National Institutes of Health Stroke Scale (OR = 1.10; 95% CI: 1.02-1.15; P = .005); rCBV index (OR = 0.005; 95% CI: 0.00-0.14; P = .002); and PCI (OR = 0.99; 95% CI: 0.98-0.99; P = .01).

Perfusion-based collateral indices that integrate CBV (rCBV index and PCI) outperformed delay-based HIR in the prediction of substantial infarct growth in patients with AIS, and may play an important role in treatment decision-making, patient transfer decisions, and prognostication.