Abstract：
BACKGROUND: Most pretest probability (PTP) tools for obstructive coronary artery disease (CAD) were Western -developed. The most appropriate PTP models and the contribution of coronary artery calcium score (CACS) in Asian populations remain unknown. In a mixed Asian cohort, we compare 5 PTP models: local assessment of the heart (LAH), CAD Consortium (CAD2), risk factor-weighted clinical likelihood, the American Heart Association/American College of Cardiology and the European Society of Cardiology PTP and 3 extended versions of these models that incorporated CACS: LAH(CACS), CAD2(CACS), and the CACS-clinical likelihood.
RESULTS: The study cohort included 771 patients referred for stable chest pain. Obstructive CAD prevalence was 27.5%. Calibration, area under the receiver-operating characteristic curves (AUC) and net reclassification index were evaluated. LAH clinical had the best calibration (χ2 5.8; P=0.12). For CACS models, LAH(CACS) showed least deviation between observed and expected cases (χ2 37.5; P<0.001). There was no difference in AUCs between the LAH clinical (AUC, 0.73 [95% CI, 0.69-0.77]), CAD2 clinical (AUC, 0.72 [95% CI, 0.68-0.76]), risk factor-weighted clinical likelihood (AUC, 0.73 [95% CI: 0.69-0.76) and European Society of Cardiology PTP (AUC, 0.71 [95% CI, 0.67-0.75]). CACS improved discrimination and reclassification of the LAH(CACS) (AUC, 0.88; net reclassification index, 0.46), CAD2(CACS) (AUC, 0.87; net reclassification index, 0.29) and CACS-CL (AUC, 0.87; net reclassification index, 0.25).
CONCLUSIONS: In a mixed Asian cohort, Asian-derived LAH models had similar discriminatory performance but better calibration and risk categorization for clinically relevant PTP cutoffs. Incorporating CACS improved discrimination and reclassification. These results support the use of population-matched, CACS-inclusive PTP tools for the prediction of obstructive CAD.
RESULTS: The study cohort included 771 patients referred for stable chest pain. Obstructive CAD prevalence was 27.5%. Calibration, area under the receiver-operating characteristic curves (AUC) and net reclassification index were evaluated. LAH clinical had the best calibration (χ2 5.8; P=0.12). For CACS models, LAH(CACS) showed least deviation between observed and expected cases (χ2 37.5; P<0.001). There was no difference in AUCs between the LAH clinical (AUC, 0.73 [95% CI, 0.69-0.77]), CAD2 clinical (AUC, 0.72 [95% CI, 0.68-0.76]), risk factor-weighted clinical likelihood (AUC, 0.73 [95% CI: 0.69-0.76) and European Society of Cardiology PTP (AUC, 0.71 [95% CI, 0.67-0.75]). CACS improved discrimination and reclassification of the LAH(CACS) (AUC, 0.88; net reclassification index, 0.46), CAD2(CACS) (AUC, 0.87; net reclassification index, 0.29) and CACS-CL (AUC, 0.87; net reclassification index, 0.25).
CONCLUSIONS: In a mixed Asian cohort, Asian-derived LAH models had similar discriminatory performance but better calibration and risk categorization for clinically relevant PTP cutoffs. Incorporating CACS improved discrimination and reclassification. These results support the use of population-matched, CACS-inclusive PTP tools for the prediction of obstructive CAD.
摘要：
背景：大多数用于阻塞性冠状动脉疾病（CAD）的预测概率（PTP）工具是西方开发的。亚洲人群中最合适的PTP模型和冠状动脉钙积分（CACS）的贡献仍然未知。在一个混合的亚洲队列中，我们比较了5种PTP模型：心脏局部评估（LAH），CAD联盟(CAD2),危险因素加权临床可能性，美国心脏协会/美国心脏病学会和欧洲心脏病学会PTP以及这些模型的3个扩展版本，其中包含CACS：LAH（CACS），CAD2(CACS),和CACS临床可能性。
结果：研究队列包括771例因稳定型胸痛转诊的患者。阻塞性CAD患病率为27.5%。校准,评估受试者工作特征曲线下面积（AUC）和净重新分类指数。LAH临床校准最好（χ25.8；P=0.12）。对于CACS模型，LAH（CACS）显示观察到的病例与预期病例之间的偏差最小（χ237.5；P<0.001）。LAH临床之间的AUC没有差异(AUC，0.73[95%CI，0.69-0.77])，CAD2临床（AUC，0.72[95%CI，0.68-0.76])，危险因素加权临床可能性(AUC，0.73[95%CI：0.69-0.76)和欧洲心脏病学会PTP(AUC，0.71[95%CI，0.67-0.75]）。CACS改善了LAH(CACS)的辨别和重新分类(AUC，0.88；净重新分类指数，0.46),CAD2（CACS）（AUC，0.87；净重新分类指数，0.29)和CACS-CL(AUC，0.87；净重新分类指数，0.25）。
结论：在亚洲混合队列中，亚洲衍生的LAH模型具有相似的辨别性能，但对于临床相关的PTP截止值具有更好的校准和风险分类。合并CACS改善了歧视和重新分类。这些结果支持使用人口匹配，包含CACS的PTP工具用于预测阻塞性CAD。
结果：研究队列包括771例因稳定型胸痛转诊的患者。阻塞性CAD患病率为27.5%。校准,评估受试者工作特征曲线下面积（AUC）和净重新分类指数。LAH临床校准最好（χ25.8；P=0.12）。对于CACS模型，LAH（CACS）显示观察到的病例与预期病例之间的偏差最小（χ237.5；P<0.001）。LAH临床之间的AUC没有差异(AUC，0.73[95%CI，0.69-0.77])，CAD2临床（AUC，0.72[95%CI，0.68-0.76])，危险因素加权临床可能性(AUC，0.73[95%CI：0.69-0.76)和欧洲心脏病学会PTP(AUC，0.71[95%CI，0.67-0.75]）。CACS改善了LAH(CACS)的辨别和重新分类(AUC，0.88；净重新分类指数，0.46),CAD2（CACS）（AUC，0.87；净重新分类指数，0.29)和CACS-CL(AUC，0.87；净重新分类指数，0.25）。
结论：在亚洲混合队列中，亚洲衍生的LAH模型具有相似的辨别性能，但对于临床相关的PTP截止值具有更好的校准和风险分类。合并CACS改善了歧视和重新分类。这些结果支持使用人口匹配，包含CACS的PTP工具用于预测阻塞性CAD。
