Predicting Coronary Artery Stenosis Severity in Coronary Heart Disease: A Combined Model of Triglyceride-Glucose Index and Carotid Ultrasound Radiomics.

AIMS/BACKGROUND: Precise evaluation of coronary lesion severity is pivotal for the clinical management of coronary heart disease (CHD). However, because the invasive coronary angiography (ICA) method, which is considered the gold standard for diagnosis, is relatively invasive, it is not widely used in early risk stratification. This study evaluated the performance of a predictive model based on a combined triglyceride-glucose (TyG) index and ultrasound radiomics characteristics of carotid plaques for evaluating coronary artery lesion severity in patients with CHD. METHODS: From January 2020 to October 2024, 381 CHD patients were diagnosed by ICA at Yichang Central People's Hospital. Radiomics features were extracted from manually divided regions of interest (ROIs) in carotid plaque ultrasound images. A two-stage feature selection was carried out to identify the essential features, and then logistic regression was employed to construct a radiomics score (Rad score). Clinical data, such as the TyG index, were also included in the construction of the combined prediction model, and the predictive performance of this model was evaluated on both the training and test sets. RESULTS: Multivariate logistic regression shows that the TyG index (odds ratio [OR] = 3.82, 95% confidence interval [CI]: 2.13-6.84), sex (OR = 1.80, 95% CI: 1.08-3.00), and hypertension status (OR = 1.81, 95% CI: 1.14-2.87) are all independent predictors of coronary lesion severity. Based on 17 essential radiomics features, the Rad score model reached an area under the curve (AUC) of 0.673 (95% CI: 0.613-0.734) for the training set and an AUC of 0.686 (95% CI: 0.567-0.806) for the test set. The combined model including sex, TyG index, hypertension status, and Rad score performed better, with AUC values of 0.823 (95% CI: 0.777-0.870) for the training set and 0.730 (95% CI: 0.616-0.844) for the test set. CONCLUSION: The combined model of the TyG index and ultrasound radiomics features for carotid plaques can be used to predict the severity of coronary artery lesions in patients with CHD. Non-invasively, it can be used to assess the risk of coronary artery disease and tailor a plan for other treatment options.