Supplementary Material for: The Prognostic Value of CT Angiography and CT Perfusion in Acute Ischemic Stroke

Background: CT angiography (CTA) and CT perfusion (CTP) are important diagnostic tools in acute ischemic stroke. We investigated the prognostic value of CTA and CTP for clinical outcome and determined whether they have additional prognostic value over patient characteristics and non-contrast CT (NCCT). Methods: We included 1,374 patients with suspected acute ischemic stroke in the prospective multicenter Dutch acute stroke study. Sixty percent of the cohort was used for deriving the predictors and the remaining 40% for validating them. We calculated the predictive values of CTA and CTP predictors for poor clinical outcome (modified Rankin Scale score 3-6). Associations between CTA and CTP predictors and poor clinical outcome were assessed with odds ratios (OR). Multivariable logistic regression models were developed based on patient characteristics and NCCT predictors, and subsequently CTA and CTP predictors were added. The increase in area under the curve (AUC) value was determined to assess the additional prognostic value of CTA and CTP. Model validation was performed by assessing discrimination and calibration. Results: Poor outcome occurred in 501 patients (36.5%). Each of the evaluated CTA measures strongly predicted outcome in univariable analyses: the positive predictive value (PPV) was 59% for Alberta Stroke Program Early CT Score (ASPECTS) ≤7 on CTA source images (OR 3.3; 95% CI 2.3-4.8), 63% for presence of a proximal intracranial occlusion (OR 5.1; 95% CI 3.7-7.1), 66% for poor leptomeningeal collaterals (OR 4.3; 95% CI 2.8-6.6), and 58% for a >70% carotid or vertebrobasilar stenosis/occlusion (OR 3.2; 95% CI 2.2-4.6). The same applied to the CTP measures, as the PPVs were 65% for ASPECTS ≤7 on cerebral blood volume maps (OR 5.1; 95% CI 3.7-7.2) and 53% for ASPECTS ≤7 on mean transit time maps (OR 3.9; 95% CI 2.9-5.3). The prognostic model based on patient characteristics and NCCT measures was highly predictive for poor clinical outcome (AUC 0.84; 95% CI 0.81-0.86). Adding CTA and CTP predictors to this model did not improve the predictive value (AUC 0.85; 95% CI 0.83-0.88). In the validation cohort, the AUC values were 0.78 (95% CI 0.73-0.82) and 0.79 (95% CI 0.75-0.83), respectively. Calibration of the models was satisfactory. Conclusions: In patients with suspected acute ischemic stroke, admission CTA and CTP parameters are strong predictors of poor outcome and can be used to predict long-term clinical outcome. In multivariable prediction models, however, their additional prognostic value over patient characteristics and NCCT is limited in an unselected stroke population.

背景：计算机断层扫描血管造影（CTA）和计算机断层扫描灌注（CTP）是急性缺血性卒中诊断中的关键工具。本研究旨在探讨CTA和CTP对临床预后的预测价值，并确定其在患者特征和非对比增强CT（NCCT）之外是否具有额外的预测价值。方法：纳入了荷兰急性卒中前瞻性多中心研究中的1374例疑似急性缺血性卒中患者。其中60%的病例用于预测因子的推导，剩余40%用于验证。计算了CTA和CTP预测因子对不良临床预后（改良Rankin量表评分3-6）的预测值。通过计算优势比（OR）评估了CTA和CTP预测因子与不良临床预后的关联。基于患者特征和NCCT预测因子的多变量逻辑回归模型建立后，随后加入CTA和CTP预测因子。通过曲线下面积（AUC）值的增加来评估CTA和CTP的额外预测价值。通过评估区分度和校准来执行模型验证。结果：不良预后发生在501名患者（36.5%）中。所评估的每项CTA指标在单变量分析中均强烈预测预后：Alberta卒中计划早期CT评分（ASPECTS）≤7在CTA原始图像上的阳性预测值（PPV）为59%（OR 3.3；95% CI 2.3-4.8），存在近端颅内闭塞的PPV为63%（OR 5.1；95% CI 3.7-7.1），脑膜周围侧支循环不良的PPV为66%（OR 4.3；95% CI 2.8-6.6），以及颈动脉或椎基底动脉狭窄/闭塞>70%的PPV为58%（OR 3.2；95% CI 2.2-4.6）。CTP指标亦然，其PPV分别为大脑血容量图上的ASPECTS≤7的65%（OR 5.1；95% CI 3.7-7.2）和平均传输时间图上的ASPECTS≤7的53%（OR 3.9；95% CI 2.9-5.3）。基于患者特征和NCCT测量的预后模型对不良临床预后具有高度预测性（AUC 0.84；95% CI 0.81-0.86）。向该模型中加入CTA和CTP预测因子并未提高其预测价值（AUC 0.85；95% CI 0.83-0.88）。在验证队列中，AUC值分别为0.78（95% CI 0.73-0.82）和0.79（95% CI 0.75-0.83）。模型的校准令人满意。结论：对于疑似急性缺血性卒中患者，入院时的CTA和CTP参数是不良预后的强有力预测因子，可用于预测长期临床预后。然而，在多变量预测模型中，它们在未选择卒中人群中对患者特征和NCCT的额外预测价值有限。