Value of Quantitative Collateral Scoring on CT Angiography in Patients with Acute Ischemic Stroke

Abstract

Background and purpose: Many studies have emphasized the relevance of collateral flow in patients presenting with acute ischemic stroke. Our aim was to evaluate the relationship of the quantitative collateral score on baseline CTA with the outcome of patients with acute ischemic stroke and test whether the timing of the CTA acquisition influences this relationship.

Materials and methods: From the Multicenter Randomized Clinical Trial of Endovascular Treatment of Acute Ischemic Stroke in the Netherlands (MR CLEAN) data base, all baseline thin-slice CTA images of patients with acute ischemic stroke with intracranial large-vessel occlusion were retrospectively collected. The quantitative collateral score was calculated as the ratio of the vascular appearance of both hemispheres and was compared with the visual collateral score. Primary outcomes were 90-day mRS score and follow-up infarct volume. The relation with outcome and the association with treatment effect were estimated. The influence of the CTA acquisition phase on the relation of collateral scores with outcome was determined.

Results: A total of 442 patients were included. The quantitative collateral score strongly correlated with the visual collateral score (ρ = 0.75) and was an independent predictor of mRS (adjusted odds ratio = 0.81; 95% CI, .77-.86) and follow-up infarct volume (exponent β = 0.88; P < .001) per 10% increase. The quantitative collateral score showed areas under the curve of 0.71 and 0.69 for predicting functional independence (mRS 0-2) and follow-up infarct volume of >90 mL, respectively. We found significant interaction of the quantitative collateral score with the endovascular therapy effect in unadjusted analysis on the full ordinal mRS scale (P = .048) and on functional independence (P = .049). Modification of the quantitative collateral score by acquisition phase on outcome was significant (mRS: P = .004; follow-up infarct volume: P < .001) in adjusted analysis.

Conclusions: Automated quantitative collateral scoring in patients with acute ischemic stroke is a reliable and user-independent measure of the collateral capacity on baseline CTA and has the potential to augment the triage of patients with acute stroke for endovascular therapy.

Fig 1.

An example of quantitative collateral capacity scoring. A, An axial plane of a baseline CTA image acquired in the peak venous phase with a right-sided M1 segment occlusion of the MCA territory. B, Segmentation results of automated quantitative collateral assessment of the ipsilateral (red) and contralateral (blue) hemispheres. The quantitative collateral score was 46%. C, 3D representation of the segmented vasculature.

Fig 2.

Case examples of 4 patients with different visually scored collateral grades and corresponding quantitative collateral scores. Each panel shows a maximum-intensity-projection of the CTA image (left) and the segmented vasculature for qCS calculation (right). The automated segmentation on the ipsilateral side is shown in blue and the segmentation on the contralateral side is shown in red. A, Absent collaterals (visual collateral score = 0). CTA of an 83-year-old man with a left-sided M2 occlusion acquired in the early arterial phase. Follow-up infarct volume was 205 mL, and the mRS score was 6. B, Poor collaterals (vCS = 1). CTA of a 79-year-old man with a right-sided M1 occlusion acquired in the equilibrium phase. FIV was 245 mL, and the mRS score was 6. C, Moderate collaterals (vCS = 2). CTA of a 45-year-old woman with a left-sided M1 occlusion acquired in the peak arterial phase. FIV was 24 mL, and the mRS score was 2. D, Good collaterals (vCS = 3). CTA of a 76-year-old woman with a left-sided ICA-T occlusion acquired in the late venous phase. FIV was 48 mL, and the mRS score was 3.

Fig 3.

Distribution of quantitative collateral scores per visual collateral score, ranging from absent collaterals (0% filling of the occluded territory) to good collaterals (100% filling of the occluded territory). The quantitative collateral score was significantly different among all visual collateral score groups, except for absent collaterals (grade 0) versus poor collaterals (grade 1).

Fig 4.

Receiver operating characteristic curve analysis of visual and quantitative collateral scores for discriminating favorable outcome (mRS 0–2) with areas under the curve of, respectively, 0.65 and 0.68 (A) and large infarct (FIV of >90 mL) with areas under the curve of, respectively, 0.69 and 0.71 (B).

Fig 5.

Bar graphs depict the proportion of functional independence (mRS 0–2) by quantitative collateral score strata for CTA image acquisition in the early arterial phase (A) and arteriovenous phase (B), and by visual collateral scores in the early arterial phase (C) and arteriovenous phase (D).