Peri-coronary fat inflammation predicts proximal atherosclerotic plaque formation associated with LAD myocardial bridge
10.3760/cma.j.cn112150-20240709-00549
- VernacularTitle:基于冠状动脉周围脂肪预测左冠状动脉前降支心肌桥近端动脉粥样硬化斑块形成的研究
- Author:
Suyu LI
1
;
Fan ZHOU
;
Zhihan XU
;
Yanchun CHEN
;
Qian CHEN
;
Yunyan SU
;
Yun FENG
;
Haitao ZHU
;
Longjiang ZHANG
Author Information
1. 南京大学医学院附属金陵医院 东部战区总医院放射诊断科，南京 210002
- Publication Type:Journal Article
- Keywords:
Coronary CT angiograph;
Myocardial bridging;
Atherogenesis;
Adipose tissue;
Inflammation;
Machine learning
- From:
Chinese Journal of Preventive Medicine
2025;59(5):604-612
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the correlation between peri-coronary fat attenuation index (FAI) and plaque formation in patients with myocardial bridge (MB) of the left anterior descending artery (LAD) using coronary computed tomography angiography (CCTA) and to develop an optimal predictive model to explore the potential application of FAI in the primary prevention of MB related atherosclerosis.Methods:In this retrospective study, prediction models associated with perivascular fat inflammation were developed and validated using both logistic regression and machine learning (ML) algorithm. A training dataset was collected from 253 patients who underwent ≥2 coronary computed tomography angiography (CCTA) with ≥3 months intervals from one tertiary hospital from January 2007 to April 2021 and had baseline CCTA showing no plaques in LAD MB. The median follow-up time was 3.2 years. According to the same criteria, a total of 75 LAD MB patients from four other hospitals were included to form an independent external validation dataset, with a median follow-up time of 1.8 years. Receiver operating characteristic (ROC) curve analysis with integrated discrimination improvement (IDI) and category net reclassification index (NRI) were used to compare the performance of the predictive models.Results:62 patients (24.5%) in the training dataset had proximal plaque formation in LAD MB, while 22 patients (29.3%) in the external validation dataset had plaque formation during the follow-up period. Baseline FAI within the longitudinal distance equal to 30 mm proximal to the MB entrance was an independent predictor ( OR=1.068, P=0.046). According to the model results, ROC curves were plotted. The AUC of Model 1 was 0.822, and the AUCs of Model 2 and 1 were 0.821 and 0.591 in the training dataset. After the DeLong test, the AUC of Model 1 was superior to that of Model 2 ( Z=2.839, P=0.005) and Model 1 ( Z=6.124, P<0.001). These findings were further validated in the external validation dataset, where ML-model 3 yielded the best predictive performance, outperforming the logistic regression-based Model 2 (categorical NRI=0.359, P=0.048; IDI=0.108, P=0.046). Conclusion:FAI measured within the 30 mm proximal to the entrance of MBs due to its prone to plaque development is an independent predictor for atherosclerotic plaque formation. The ML-prediction model based on a decision tree algorithm combines FAI, MB anatomical features, and patient risk factors, which is beneficial for patients undergoing routine CCTA examination to identify inflamed coronary arteries in advance and guide the clinical adoption of more targeted preventive treatment, including anti-inflammatory treatment.
