Prediction of coronary atherosclerosis progression in type 2 diabetes mellitus based on AI-derived CCTA parameters and clinical factors: a follow-up study.
Authors
Affiliations (4)
Affiliations (4)
- Department of Medical imaging, The First Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, China.
- Department of Medical imaging, The First Hospital of Jiaxing, Jiaxing, Hanzhou, China.
- Department of Medical imaging, The First Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, China. [email protected].
- Department of Medical imaging, The First Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, China. [email protected].
Abstract
To develop a predictive model for coronary atherosclerosis progression in patients with type 2 diabetes mellitus (T2DM) based on Artificial intelligence(AI)-derived coronary computed tomography angiography (CCTA) parameters combined with clinical indicators. This retrospective study enrolled 114 patients with T2DM and non-obstructive coronary artery disease (1%-49% stenosis) who underwent CCTA at our hospital between September 2019 and September 2024. After follow-up of 1-5 years, patients were stratified into a progression group (n = 48) and a control group (n = 66). Progression was defined as the occurrence of acute myocardial infarction, revascularization, or stenosis progression to ≥ 50% on follow-up CCTA. Clinical data and AI-derived quantitative parameters of pericoronary adipose tissue(PCAT) and coronary plaques were compared. A predictive model was then developed using stepwise logistic regression and its performance evaluated. The progression group had significantly higher glycated hemoglobin (HbA1c) and troponin, and lower high-density lipoprotein cholesterol (HDL-C) (all P < 0.05).CCTA analysis revealed that the progression group had a significantly greater pericoronary fat attenuation index around the left anterior descending (LAD-FAI) and left circumflex (LCX-FAI) arteries, as well as increased plaque length, total plaque volume, and specific plaque component volumes (all P < 0.05). Multivariate logistic regression analysis identified HbA1c, LAD-FAI, and plaque length as independent predictors of progression. A nomogram based on these predictors achieved an AUC of 0.8332. LAD-FAI and plaque length and HbA1c are independent predictors for the progression of non-obstructive coronary lesions in T2DM patients. The proposed nomogram may provide a noninvasive and cost-effective tool for early risk stratification and preventive management.