Quantifying reperfusion of the ischemic region on whole-brain computed tomography perfusion

Abstract

To derive the reperfusion index best predicting clinical outcome of ischemic stroke patients, we retrospectively analysed the acute and 24-h computed tomography perfusion data of 116 patients, collected from two centres equipped with whole-brain computed tomography perfusion. Reperfusion index was defined by the percentage of the ischemic region reperfused from acute to 24-h computed tomography perfusion. Recanalization was graded by arterial occlusive lesion system. Receiver operator characteristic analysis was performed to assess the prognostic value of reperfusion and recanalization in predicting good clinical outcome, defined as modified Rankin Score of 0-2 at 90 days. Among previous reported reperfusion measurements, reperfusion of the Tmax>6 s region resulted in higher prognostic value than recanalization at predicting good clinical outcome (area under the curve = 0.88 and 0.74, respectively, p = 0.002). Successful reperfusion of the Tmax>6 s region (≥60%) had 89% sensitivity and 78% specificity in predicting good clinical outcome. A reperfusion index defined by Tmax>2 s or by mean transit time>145% had much lower area under the curve in comparison to Tmax>6 s measurement (p < 0.001 and p = 0.003, respectively), and had no significant difference to recanalization at predicting clinical outcome (p = 0.58 and 0.63, respectively). In conclusion, reperfusion index calculated by Tmax>6 s is a stronger predictor of clinical outcome than recanalization or other reperfusion measures.

Keywords: Acute stroke; brain imaging; brain ischemia; cerebral blood flow measurement; reperfusion.

Figure 1.

Threshold setting on perfusion maps. For each perfusion map, one threshold delineates one ischemic region, and the setting of 20 threshold levels results in 20 ischemic regions. Each ischemic region corresponds to one reperfusion index.

Figure 2.

Box plot of reperfusion index against perfusion thresholds for each map. Reperfusion index measured by the restoration of the Tmax prolonged regions (a), by the restoration of the MTT prolonged regions (b), by the restoration of the CBF abnormality regions (c), and by the restoration of the CBV abnormality regions (d). MTT: mean transit time; CBF: cerebral blood flow; CBV: cerebral blood volume.

Figure 3.

Comparison of the reperfusion index measured by Tmax>6 s, by Tmax>2 s, and by MTT>145%. (a) Box plot shows that reperfusion index of the ischemic region measured by Tmax>6 s has higher value than that by Tmax>2 s and by MTT>145%. (b) ROC analysis shows that reperfusion index measured by Tmax>6 s has larger AUC than Tmax>2 s and MTT>145% measurement in predicting good clinical outcome (mRS of 0–2). Reperfusion of the Tmax>6 s region also has larger AUC than recanalization. MTT: mean transit time; ROC: receiver operator characteristic; AUC: area under the curve.

Figure 4.

Flow diagram of the patient selection process.

Figure 5.

A stroke case showing reperfusion is better than recanalization in predicting clinical outcome. Female, 31, admitted to John Hunter Hospital with baseline NIHSS 10. Acute brain scanning shows occlusion of middle cerebral artery (MCA, green arrow) on CTA and corresponding ischemic region on CTP (red area delineated by Tmax>6 s). 24-h scanning shows the persistence of occlusion on MCA (blue arrow), but the reduction of ischemic region on CTP. The patient has no occlusion recanalization but achieves 81.05% reperfusion. The good reperfusion explains the small infarction on 24-h NCCT (yellow circle). It also explains her clinical improvement at 24 h (NIHSS = 4) and good clinical outcome at 90 days (mRS = 1). CTP map is generated from standard SVD. NIHSS: National Institutes of Health Stroke Scale; MCA: middle cerebral artery; CTA: computed tomography angiography; CTP: computed tomography perfusion; NCCT: non-contrast computed tomography; SVD: singular value decomposition.