Introduction: Good collateral flow is an independent predictor of reperfusion that can be used to extend the treatment window in the new era of endovascular therapies for patients with acute ischemic stroke (AIS). Using a multiparametric approach, we aimed to identify perfusion parameter/s that can represent the extent of collaterals in comparison to CTA.
Methods: AIS patients with anterior circulation large vessel occlusion who had baseline CTA and CT perfusion were included. CT perfusion data were processed by Bayesian method to generate arterial tissue delay (ATD) maps at thresholds of 2 & 6 seconds. The volume of mild delayed perfusion (Vol-ATD>2sec), moderate delayed (Vol-ATD2-6sec) and critical delayed perfusion (Vol-ATD>6sec) in addition to corresponding rCBV and rCBF were calculated. Baseline CTA collaterals were scored using an established scoring scale1 and dichotomized to poor or good. The association of perfusion parameters and status of collaterals was assessed by repeated measure of analyses and receiver operating characteristic (ROC).
Results: In 28 patients included, 16 had good collaterals on CTA. After controlling for age, sex, baseline NIHSS and type of treatment, multivariate logistic regression analysis identified rCBV (p<0.001) and ATD2-6sec (p=0.003), but not rCBF, Vol-ATD>2sec or Vol-ATD>6sec, as independent predictors of good collaterals. ROC analysis showed AUC of 0.88 (sensitivity/specificity: 75%/100%) for rCBV and AUC of 0.84 (sensitivity/specificity: 93%/67%) for Vol-ATD2-6sec. We defined a perfusion collateral index (PCI) calculated from Vol-ATD2-6sec x its rCBV, that remained an independent predictor of good collaterals with improved diagnostic accuracy over each measure alone resulting in nominal AUC of 1 (sensitivity/specificity: 100%/100%).
Conclusions: Multiparametric CT perfusion can be used to assess the status of collaterals in patients with AIS. Perfusion collateral index (PCI) defined as Vol-ATD2-6sec x rCBV is a new perfusion index with a nominal diagnostic accuracy of 100% compared to baseline CTA to predict status of collaterals in our small cohort. Our results need to be validated in a larger prospective cohort.