Development and validation of a computed tomography myocardial perfusion imaging radiomic model for major adverse cardiovascular events prediction: a multicenter study.
Authors
Affiliations (9)
Affiliations (9)
- Department of Radiology, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, #85 Wujin Rd, Shanghai, China.
- Faculty of Medical Imaging Technology, College of Health Science and Technology, Shanghai Jiao Tong University School of Medicine, China.
- Department of Radiology and Tianjin Key Laboratory of Functional Imaging, Tianjin Medical University General Hospital, Tianjin, 300052, China.
- Department of Cardiology, The Second Affiliated Hospital of Harbin Medical University, Harbin, China.
- Institute of Diagnostic and Interventional Radiology, Shanghai Jiao Tong University Affiliated Sixth People's Hospital, #600, Yishan Rd, Shanghai, China.
- Imaging Diagnosis and Treatment Center, Xi'an International Medical Center Hospital, Xi'an, Shaanxi,710100, China.
- Department of Radiology, Sichuan Provincial People's Hospital, School of Medicine, University of Electronic Science and Technology of China, Chengdu, China.
- Department of Radiology, The Affiliated Huai'an No.1 People's Hospital of Nanjing Medical University, 1# Huanghe West Road, Huaiyin District, Huai'an, 223300, Jiangsu Province, China.
- Department of Radiology, The First Affiliate Hospital with Nanjing Medical University, Nanjing, 210029, China.
Abstract
Accurate prediction of major adverse cardiovascular events (MACE) is crucial for risk stratification in patients with suspected coronary artery disease. CT myocardial perfusion imaging (CT-MPI) provides various parameters, which may help comprehensively characterize perfusion features. This study aimed to develop a combined model, including clinical risk factors, coronary atherosclerotic characteristics, and radiomic features derived from CT-MPI, to predict MACE. 784 patients who underwent coronary CT angiography (CCTA) and CT-MPI from eight hospitals were retrospectively enrolled. Radiomic analysis was performed on eight perfusion parameter maps. Three prediction models were established accordingly: Model 1 (clinical risk factors and coronary atherosclerotic characteristics), Model 2 (incorporating myocardial blood flow values upon Model 1), and Model 3 (integrating radiomic scores upon Model 2). The C-indices for Model 3 in the training, internal validation, and external validation sets were 0.898 (95% confidence interval [CI]: 0.856-0.947), 0.844 (95% CI: 0.780-0.908), and 0.840 (95% CI: 0.791-0.889), respectively, demonstrating significant improvements over Model 1 and Model 2 (all p < 0.05). In the external validation set, Model 3 had the largest time-dependent areas under the curve (AUC) values for 1-, 3-, and 5-year MACE prediction (0.890 [95% CI: 0.831-0.948], 0.880 [95% CI: 0.823-0.938], and 0.837 [95% CI: 0.726-0.949]), compared to Model 1 and Model 2. The radiomic features from multiparametric CT-MPI maps simultaneously captured perfusion features associated with MACE at both macrovascular and microvascular levels. The combined model exhibited improved MACE prognostic performance compared with conventional models while maintaining high interpretability.