The Assessment of infarct core on non contrast computed tomography versus computed tomography perfusion imaging using RAPID software in patients with large vessel occlusion

Abstract

Objective: To evaluate whether infarct core on non-contrast computed tomography (NCCT) and CT perfusion (CTP) correspond and equally predict infarct extent after thrombectomy for ischemic stroke. Methods: From our prospective database, we selected consecutive patients with acute middle cerebral artery occlusion undergoing successful thrombectomy (TICI 2b/3) between May-‘16 and February-‘18. All patients had baseline NCCT, CTP with RAPID post-processing software, and 24h NCCT. Two raters independently assessed the infarct cores using the Alberta Stroke Program Early Computed Tomography Score (ASPECTS) on each modality in a blinded fashion: 1) baseline NCCT, 2) CTP (CBF <30% of normal tissue), and 3) 24h NCCT. We calculated the interrater agreement, correlation between baseline NCCT-ASPECTS, CTP-ASPECTS, and 24h NCCT-ASPECTS; and the positive predictive value (PPV) of NCCT- and CTP-ASPECTS. Results: We studied 86 patients: 63% (54/86) women, median age 74 (IQR62-82), median baseline NCCT-ASPECTS 9 (IQR 8–10), median CTP-ASPECTS 9 (IQR 8–10), and mean CTP-infarct volume 14 28cc. Interrater agreement was substantial for NCCT-ASPECTS (weighted kappa: 0.71) and almost perfect for CTP-ASPECTS (weighted kappa: 0.94). There was substantial correlation between baseline and 24h NCCT-ASPECTS (Spearman’s rho: 0.62; p < 0.001), but poor correlation between baseline NCCT-ASPECTS and CTP-ASPECTS (Spearman’s rho: 0.18; p ¼ 1.0). CTP-ASPECTS and CTP-core volume did poorly correlate with 24h NCCT-ASPECTS (Spearman’s rho: 0.21; p ¼ 0.06 and -0.16; p ¼ 0.15 respectively). PPV of baseline NCCT-ASPECTS was 81%, while that of CTPASPECTS was 53%. Conclusion: In this series, infarct core on RAPID-CTP did not correlate with the baseline-NCCTand tended to depict a larger core than the final infarct.