Predicting obstructive coronary artery disease using carotid ultrasound parameters: A nomogram from a large real‐world clinical data

• DOI: 10.1111/eci.12956
• Publication Date: 2018-05-21T19:29:33Z
• Authors: Na Wu; Xinghua Chen; Mingyang Li; Xiaolong Qu; Yueli Li; Weijia Xie; Long Wu; Ying Xiang; Yafei Li; Li Zhong
• Keywords: Male; Middle Aged; Carotid Intima-Media Thickness; 3. Good health; Nomograms; 03 medical and health sciences; Early Diagnosis; 0302 clinical medicine; Coronary Occlusion; Predictive Value of Tests; Risk Factors; Area Under Curve; Humans; Female; Aged

AbstractBackgroundCarotid ultrasound is a noninvasive tool for risk assessment of coronary artery disease (CAD). There is no consensus on which carotid ultrasound parameter constitutes the best measurement of atherosclerosis. We investigated which model of carotid ultrasound parameters and clinical risk factors (CRF) has the highest predictive value for CAD.Materials and methodsWe enrolled 2431 consecutive patients who have suspected CAD and underwent coronary angiography and carotid ultrasound with measurements of carotid intima‐media thickness (CIMT), total number of plaques and areas of different types of plaques classified by echogenicity.ResultsTotal number of plaques demonstrated the highest incremental prediction ability to predict CAD over CRF (area under the curve [AUC] 0.752 vs 0.701, net reclassification index [NRI] = 0.514, P < .001), followed by area of maximum mixed and soft plaques. CIMT had no significant incremental value over CRF (AUC 0.704 vs 0.701, P = .241; NRI = 0.062, P = .168). The model comprising total number of plaques, areas of maximum soft, hard and mixed plaques plus CRF had the highest discriminatory (AUC = 0.757) and reclassification value (NRI = 0.567) for CAD. A nomogram based on this model was developed to predict CAD. For subjects at low and intermediate risk, the model comprising total number of plaques plus CRF was the best.ConclusionsTotal number of plaques, area of maximum soft, hard and mixed plaques showed significantly incremental prediction ability over CRF. A nomogram based on these factors provided an intuitive and practical method in detecting CAD.