ABSTRACT
Objectives
Metabolic dysregulation and adiposity‐related alterations contribute to coronary atherosclerosis, yet the interrelations among cardiometabolic markers and epicardial adipose tissue index (EATi) remain unclear. This study aimed to investigate the associations of the triglyceride–glucose (TyG) index, cardiometabolic index (CMI), and EATi with coronary artery disease (CAD), and to explore potential metabolic–adiposity pathways.
Methods
A total of 120 patients with CAD and 60 age‐ and sex‐matched controls were included. Clinical and biochemical data were collected. The TyG index and CMI were calculated, and EAT volume (EATV) was quantified by coronary computed tomography angiography (CCTA) and indexed to body surface area (EATi). Associations among variables were assessed using correlation analysis and multivariable logistic regression. Mediation and receiver operating characteristic (ROC) analyses were performed to assess potential pathways and diagnostic performance.
Results
Fasting glucose, systolic blood pressure, triglycerides, TyG, CMI, and EATi were significantly higher in the CAD group (all p < 0.05). TyG (OR = 2.688, p = 0.002), CMI (OR = 3.339, p = 0.005), and EATi (OR = 2.284, p = 0.002) were independently associated with CAD. Mediation analysis demonstrated a significant indirect effect via CMI, whereas no mediation effect was observed for EATi. Combined assessment of TyG, CMI, and EATi demonstrated improved diagnostic performance for CAD (AUC = 0.847).
Conclusion
The combined evaluation of TyG, CMI, and EATi improves the identification of CAD. CMI mediates the association between TyG and CAD, while EATi independently contributes to CAD risk, suggesting distinct metabolic and adiposity‐related mechanisms.
Keywords: cardiometabolic index, coronary artery disease, epicardial adipose tissue, mediation analysis, triglyceride‐glucose index
This study demonstrates that the TyG index, CMI, and EATi are independently associated with CAD. CMI partially mediates the relationship between TyG and CAD, whereas EATi contributes through an independent pathway. These findings suggest that integrating metabolic and adiposity‐related markers may improve cardiovascular risk assessment.
1. Introduction
Coronary artery disease (CAD) remains the leading cause of morbidity and mortality worldwide, continuing to impose substantial clinical and economic burdens despite ongoing advances in imaging, pharmacotherapy, and revascularization strategies [1]. The persistent risk of adverse cardiovascular events highlights the need for improved understanding of early cardiometabolic alterations that accompany coronary atherosclerosis, particularly among individuals without overt symptoms.
Epicardial adipose tissue (EAT), a metabolically active visceral fat depot surrounding the coronary arteries, has attracted increasing attention for its potential role in atherosclerotic disease. As an endocrine and paracrine organ, EAT secretes numerous cytokines, chemokines, and adipokines capable of influencing vascular inflammation and plaque vulnerability. Accumulating evidence demonstrates that increased EAT volume (EATV) or thickness is associated with the presence and severity of CAD, subclinical atherosclerosis, and high‐risk plaque features on coronary computed tomography angiography (CCTA) [2, 3, 4]. Given the close association of EATV with body size parameters such as body surface area (BSA) and BMI, the EAT index (EATi) provides a more standardized measure of epicardial adiposity.
In parallel, metabolic indices such as the triglyceride–glucose (TyG) index and cardiometabolic index (CMI) have emerged as simple, noninvasive surrogates for insulin resistance, visceral adiposity, and atherogenic dyslipidemia. The TyG index has been associated with coronary artery calcification, atherosclerotic burden, and cardiovascular events in diverse populations. CMI integrates the triglyceride‐to‐high‐density lipoprotein cholesterol (TG/HDL‐C) ratio with anthropometric parameters, such as waist‐to‐height ratio (WHtR), and has been validated as a sensitive marker of visceral adiposity and cardiovascular risk [5, 6, 7].
Although the associations of EATi, TyG, and CMI with CAD have been reported individually, the extent to which these markers may complement one another in reflecting coronary risk remains insufficiently explored. Recent studies suggest that integrating systemic metabolic markers with imaging‐derived indicators of pericoronary adiposity may offer complementary information about coronary atherosclerosis, particularly in individuals without overt clinical manifestations [8]. Furthermore, cardiometabolic derangement and epicardial fat accumulation may independently contribute to atherosclerosis progression through distinct yet complementary metabolic and inflammatory pathways.
Accordingly, this study aimed to investigate the associations of EATi, TyG index, and CMI with CAD and to explore potential metabolic–adiposity pathways through which these markers may independently relate to disease presence. By examining their combined performance and interrelationships, this study seeks to provide a multidimensional perspective on cardiometabolic alterations relevant to coronary atherosclerosis.
2. Materials and Methods
2.1. Data This Retrospective Case‐Control Study Included Patients Source and Study Population
This retrospective case–control study enrolled patients hospitalized at the Second Hospital of Hebei Medical University between June and December 2023 who were diagnosed with CAD based on the 2019 European Society of Cardiology Guidelines for the diagnosis and management of chronic coronary syndromes [9]. A total of 120 consecutive patients with CAD were enrolled in the case group. The control group consisted of 60 healthy individuals who underwent routine physical examinations at the hospital's health examination center during the same period, matched with the case group for age, gender, height, and weight. The mean age of the CAD group was 59.77 ± 7.25 years, with 74 males (61.7%), while the control group had a mean age of 59.40 ± 5.03 years, with 33 males (55.0%). Inclusion criteria were as follows ① age ≥ 18 years; ② CAD confirmed on CCTA (≥ 50% luminal stenosis); ③ matched controls undergoing routine health examinations with negative CCTA findings; ④ availability of complete clinical, biochemical, and imaging data. Exclusion criteria as follows: ① Presence of structural or congenital heart disease; ② History of heart failure or severe valvular disease; ③ Acute or chronic infections, malignancies, or autoimmune diseases; ④ Severe hepatic or renal dysfunction, or thyroid disease; ⑤ Poor‐quality CT images or incomplete EAT measurements; ⑥ Missing clinical records or laboratory biochemical parameters.
2.2. Data Collection
Baseline demographic and clinical information for all participants, including gender, age, height, weight, smoking history, alcohol consumption, and past medical history (cardiovascular disease, hypertension [HTN], diabetes mellitus [DM], and hyperlipidemia), as well as systolic blood pressure (SBP) and diastolic blood pressure (DBP), were collected from the electronic medical record system and the Health Management Center of the Second Hospital of Hebei Medical University.
Fasting venous blood samples (5 mL) were collected from each participant in the morning after at least 8 h of overnight fasting. Biochemical parameters, including total cholesterol (TC), triglycerides (TG), high‐density lipoprotein cholesterol (HDL‐C), low‐density lipoprotein cholesterol (LDL‐C), and fasting blood glucose (FBG), were measured using an automated biochemical analyzer in the hospital's central clinical laboratory following standard operating procedures.
2.3. Definitions
Body mass index (BMI) was calculated as: BMI = weight (kg) ÷ height2 (m2). The triglyceride‐glucose (TyG) index was calculated as ln (TG [mg/dL] × FBG [mg/dL]/2). Waist circumference (WC) was obtained from medical records. According to standard clinical practice, WC is measured in centimeters (cm) at the midpoint between the lower margin of the last palpable rib and the top of the iliac crest. WHtR was calculated as: WHtR = WC (cm) ÷ height (cm). CMI was calculated as: CMI = (TG [mmol/L] ÷ HDL‐C [mmol/L]) × WHtR. HTN was defined as SBP ≥ 140 mmHg, DBP > 90 mmHg, or ongoing treatment with antihypertensive medications. DM was defined as FBG level ≥ 126 mg/dL or treatment with hypoglycemic medications. Hyperlipidemia (HLD) was defined as TC ≥ 220 mg/dL, TG ≥ 150 mg/dL, LDL‐C ≥ 160 mg/dL, HDL‐C ≤ 40 mg/dL, or ongoing treatment with lipid‐lowering medications.
2.4. CCTA Examination
All CCTA examinations were performed using a dual‐source CT scanner (Somatom Definition Flash; Siemens Healthcare, Germany) with a gantry rotation time of 330 ms and a tube voltage of 120 kV. Patients with a heart rate > 60 beats/min received 25–75 mg of oral metoprolol (AstraZeneca, Shanghai, China) to maintain a target heart rate. All patients were given sublingual nitroglycerin immediately before scanning. Scans were performed in the supine position, covering the region from the tracheal carina down to the cardiac apex. After a preliminary coronary calcium scan, contrast volume was determined based on patient weight. Iopamidol, a nonionic contrast agent, was administered at 5.0–5.5 mL/s, followed by 40 mL of saline at the same rate. CCTA was acquired using a prospective or retrospective ECG‐gated protocol with a trigger threshold of 100 HU.
2.5. EATV Measurement
EATV was quantitatively measured from CCTA images on a dedicated workstation (Syngo.via VB20V, Siemens Healthcare, Germany). The measurement protocol followed our previously published methods [10, 11]. Briefly, EAT was defined as adipose tissue located between the myocardium and the visceral pericardium, with a CT attenuation threshold range from −190 to −30 Hounsfield units (HU), a range commonly used for fat segmentation that enables reliable discrimination from adjacent myocardium and contrast‐enhanced blood.
The measurement boundaries were set from the level of the pulmonary artery bifurcation to the apex of the left ventricle. The pericardial contour was manually delineated on axial images at 3–5 slice intervals. Given the anatomical continuity of the pericardium, this interval has been shown to provide reliable volumetric estimation while improving segmentation efficiency [12, 13]. The ROIs were then interpolated using a spline‐based segmentation algorithm based on a densitometric threshold. Adipose voxels within the defined attenuation range were then automatically identified and quantified to calculate the total EATV. EATi was calculated by dividing the measured EATV for each patient by the patient's BSA. BSA was calculated using a modified Stevenson formula: BSA (m2) = (84.4673 × height (cm)^0.6997 × weight (kg)^0.4176)/10,000, where height is expressed in centimeters and weight in kilograms.
All EATV measurements were independently performed by two experienced radiologists blinded to clinical and laboratory information; disagreements were adjudicated by a senior radiologist. Interobserver and intra‐observer reproducibility were assessed using intraclass correlation coefficients (ICCs) and Bland–Altman analysis.
2.6. Statistical Analysis
All statistical analyses were performed using SPSS version 22.0 (IBM Corp., Armonk, NY, USA). The distribution of continuous variables was assessed using the Kolmogorov–Smirnov test. Normally distributed variables (DBP, TC, TG, TyG index, and WHtR) were expressed as mean ± SD and compared using the independent samples t‐test. Non‐normally distributed variables (including EATi, age, height, weight, BMI, WC, SBP, FBG, CMI, HDL‐C, and LDL‐C) were presented as median and interquartile range (M [Q1, Q3]) and analyzed using the Mann–Whitney U test. Categorical variables were described as frequencies and percentages (n, %) and compared using the chi‐square (χ 2) test.
Pearson correlation analysis was performed to assess the relationships among the TyG index, EATi, and CMI. The coefficient of determination (R 2) was also calculated to reflect the proportion of variance explained. Multicollinearity was evaluated by calculating the tolerance and variance inflation factor (VIF), with VIF < 10 and tolerance > 0.1 indicating acceptable collinearity. To account for differences in measurement scales among the TyG index, CMI, and EATi, these variables were standardized using Z‐transformation prior to regression analyses. A multivariate logistic regression analysis with forward stepwise selection (Forward LR) was applied to identify independent predictors of CAD.
To explore mediating mechanisms linking the TyG index to CAD, parallel multiple mediation analysis was conducted using the PROCESS macro (Model 4). CMI and EATi were entered as parallel mediators. Indirect effects with 95% confidence intervals (CIs) were estimated via bias‐corrected bootstrapping with 5000 iterations. Receiver operating characteristic (ROC) curve analyses were performed to compare the predictive performance of TyG, CMI, EATi, and their combined model for CAD. All statistical tests were two‐tailed, with a p‐value < 0.05 considered statistically significant.
3. Results
The intra‐ and inter‐observer ICCs for EATV were 0.941 (95% CI: 0.848–0.974) and 0.968 (95% CI: 0.934–0.984), respectively. Bland‐Altman analysis demonstrated mean biases (95% LoA) of 4.18 (−11.16 to 19.53) cm3 and 1.59 (−11.29 to 14.47) cm3 for intra‐ and inter‐observer agreement, respectively (Figure 1).
Figure 1.
Bland‐Altman plots for EATV reproducibility.
A total of 180 subjects were enrolled in this study, including 120 patients with CAD and 60 healthy controls. Baseline demographic, anthropometric, and most clinical characteristics, including gender, age, BMI, and various cardiovascular risk factors, showed no significant differences between the CAD and control groups (p > 0.05 for all, Table 1), indicating good comparability between the two study populations. Compared with the control group, the CAD group showed significantly higher SBP (p = 0.037), FBG (p = 0.009), and TG levels (p < 0.001), while HDL‐C levels were significantly lower (p = 0.002). No significant differences were observed in TC or LDL‐C between the two groups (all p > 0.05).
Table 1.
Comparative analysis of clinical, laboratory, and quantitative CT data between two groups of patients (χ ± SD or M [Q1, Q3]).
| All patients (n = 180) | CAD (n = 120) | Control (n = 60) | t/Z/χ 2 | p value | |
|---|---|---|---|---|---|
| Male gender (n, %) | 107 (59.4) | 74 (61.7) | 33 (55.0) | 0.737 | 0.390a |
| Age (year) | 60.00 (56.00, 64.00) | 60.50 (55.25, 65.00) | 59.50 (56.00, 64.00) | 0.635 | 0.525b |
| Height | 167 (162, 171) | 167 (160, 171.75) | 168 (163, 171) | 0.417 | 0.677b |
| Weight | 70 (64, 78) | 70 (64, 76) | 71 (62.88, 80) | 0.020 | 0.984b |
| BMI (kg/m2) | 25.41 (23.63, 27.55) | 25.39 (23.68, 27.55) | 25.50 (23.00, 27.57) | 0.563 | 0.573b |
| WC (cm) | 89.73 (68.91, 99.06) | 89.41 (69.82, 99.27) | 90.12 (67.89, 98.06) | 0.533 | 0.594b |
| WHtR | 0.52 ± 0.09 | 0. 52 ± 0.09 | 0.51 ± 0.11 | 0.480 | 0.632c |
| Smoking (n, %) | 130 (72.2) | 88 (73.3) | 42 (70.0) | 0.222 | 0.638a |
| CAD history (n, %) | 11 (5.7) | 36 (30.0) | 13 (21.7) | 1.402 | 0.236a |
| Drinking history (n, %) | 58 (32.2) | 36 (30.0) | 22 (36.7) | 0.814 | 0.367a |
| HTN (n, %) | 110 (61.1) | 72 (60.0) | 38 (63.3) | 0.187 | 0.665a |
| DM (n, %) | 72 (40.0) | 56 (46.7) | 16 (26.7) | 3.110 | 0.078a |
| HLD (n, %) | 76 (42.2) | 52 (43.3) | 24 (40.0) | 0.182 | 0.670a |
| SBP (mmHg) | 130 (120, 143) | 130.00 (121.00, 145.00) | 128.50 (119.25, 138.00) | 2.082 | 0.037b * |
| DBP (mmHg) | 81.13 ± 10.84 | 82.16 ± 11.14 | 79.08 ± 9.98 | 1.804 | 0.073c |
| FBG (mmol/L) | 6.14 (5.69, 6.83) | 6.19 (5.74, 7.32) | 5.94 (5.56, 6.42) | 2.595 | 0.009b * |
| TG (mmol/L) | 1.54 ± 0.48 | 1.69 ± 0.45 | 1.24 ± 0.38 | 6.617 | 0.000c ** |
| HDL‐C (mmol/L) | 1.13 (0.99, 1.33) | 1.10 (0.91, 1.30) | 1.20 (1.07, 1.40) | 3.093 | 0.002b * |
| LDL‐C (mmol/L) | 2.75 (2.11, 3.50) | 2.67 (1.02, 3.50) | 3.03 (2.26, 3.49) | 1.297 | 0.195b |
| TC (mmol/L) | 4.47 ± 1.05 | 4.37 ± 1.15 | 4.64 ± 0.81 | 1.652 | 0.100c |
| TyG index | 8.74 ± 0.40 | 8.88 ± 0.35 | 8.45 ± 0.32 | 7.784 | 0.000c ** |
| CMI | 0.67 (0.46, 0.90) | 0.79 (0.58, 0.95) | 0.48 (0.36, 0.70) | 5.714 | 0.000b ** |
| EATi (cm3/m2) | 60.51 (55.77, 68.13) | 63.84 (55.79, 71.25) | 58.20 (55.76, 63.11) | 2.819 | 0.005 b * |
Note: a: χ 2 test, b: nonparametric rank sum test (Mann−Whitney U test), c: independent sample t‐test.
Abbreviations: BMI, body Mass Index; CMI, cardiometabolic risk index; DBP, diastolic blood pressure; DM, diabetes mellitus; EATi, epicardial adipose tissue volume index; FBG, fasting blood glucose; HDL‐C, high density lipoprotein cholesterol; HLD, hyperlipidemia; HTN, hypertension; LDL‐C, low density lipoprotein cholesterol; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; TyG index, triglyceride glucose index; WC, waist circumference; WHtR, waist‐to‐height ratio.
p < 0.05;
p < 0.001.
Regarding metabolic indices, the TyG index, CMI, and EATi were all significantly higher in the CAD group compared to the control group (all p ≤ 0.01, Table 1). Specifically, the TyG index was 8.88 ± 0.35 versus 8.45 ± 0.32; t = 7.784, p < 0.05, CMI was 0.79 (0.58, 0.95) vs 0.48 (0.36, 0.70) (Z = −5.714, p < 0.001), and EATi was 63.84 (55.79, 71.25) versus 58.20 (55.76, 63.11) cm3/m2 (Z = −2.819, p < 0.05).
3.1. Correlation and Collinearity Analysis Among Variables
To evaluate the intercorrelations among variables and assess the presence of multicollinearity within the multivariate logistic regression analysis, Pearson correlation analysis was first performed. The TyG index showed strong positive correlations with TG (r = 0.840, p < 0.01, adjusted R 2 = 0.704) and CMI (r = 0.607, p < 0.01, adjusted R 2 = 0.365). Furthermore, TG and CMI were highly correlated (r = 0.712, p < 0.01, adjusted R 2 = 0.504), while no statistically significant correlation was observed between EATi and other biochemical parameters (all p > 0.05) (Table 2). Subsequently, multicollinearity diagnostics were conducted within the multivariable logistic regression analysis. The VIF values for all independent variables were less than 10. Specifically, the VIF values were 1.593 for the TyG index, 1.583 for CMI, and 1.012 for EATi. In addition, all tolerance values exceeded 0.1, further confirming the stability and reliability of the regression analysis (Table 3).
Table 2.
Pearson correlation coefficients among TyG index, CMI, EATi, and other metabolic parameters.
| Variable | EATi | TG | TyG index | FBG | CMI |
|---|---|---|---|---|---|
| EATi | 1 | 0.071 | 0.110 | 0.087 | 0.075 |
| TG | 0.071 | 1 | 0.840** | 0.088 | 0.712** |
| TyG index | 0.110 | 0.840** | 1 | 0.580** | 0.607** |
| FBG | 0.087 | 0.088 | 0.580** | 1 | 0.102 |
| CMI | 0.075 | 0.712** | 0.607** | 0.102 | 1 |
p < 0.01 indicates statistical significance.
Table 3.
Collinearity diagnostics for the multivariable logistic regression analysis.
| Variable | Unstandardized coefficient (B) | SE | Standardized coefficient (Beta) | t‐value | p value | Tolerance | VIF |
|---|---|---|---|---|---|---|---|
| Constant | 0.555 | 0.182 | 3.049 | 0.003 | |||
| EATi | 0.007 | 0.002 | 0.182 | 2.851 | 0.005 | 0.988 | 1.012 |
| TyG index | 0.491 | 0.096 | 0.409 | 5.108 | 0.000 | 0.628 | 1.593 |
| CMI | 0.137 | 0.095 | 0.115 | 1.435 | 0.153 | 0.632 | 1.583 |
Note: Dependent variable = CHD group (1 = CHD, 0 = control). No multicollinearity detected (all VIF < 10 and Tolerance > 0.1).
3.2. Multivariate Logistic Regression Analysis for Predicting the CAD
Multivariate logistic regression analyses showed that the TyG index, CMI, and EATi were all significantly associated with CAD after mutual adjustment (Table 4). The TyG index demonstrated a robust association with CAD (OR = 2.688 95% CI: 1.421–5.084, p = 0.002). CMI also remained significantly associated with CAD (OR = 3.339, 95% CI: 1.431–7.789, p = 0.005). In addition, EATi, as an imaging‐derived marker of pericoronary fat burden, showed an independent association with CAD (OR = 2.284, 95% CI: 1.341–3.891, p = 0.002). The Nagelkerke R 2 for this multivariable model was 0.445. Sensitivity analyses revealed that the predictive value of EATi was highly robust; models using either absolute EATV or EATi (without BMI adjustment) yielded consistent effect estimates and identical directions of association (Supporting Information S1: Table 1). Furthermore, after further adjusting for BMI in the multivariable model, EATi remained a potent and independent predictor of CAD (OR = 2.341, 95% CI: 1.356–4.043, p = 0.002), whereas BMI itself showed no significant association with CAD risk (p = 0.599; Supporting Information S1: Table 1).
Table 4.
Multivariate logistic regression analysis for independent predictors of CAD.
| Variable | B | SE | OR (95% CI) | p value |
|---|---|---|---|---|
| TyG index | 0.989 | 0.325 | 2.688 (1.421–5.084) | 0.002 |
| CMI | 1.206 | 0.432 | 3.339 (1.431–7.789) | 0.005 |
| EATi | 0.826 | 0.272 | 2.284 (1.341–3.891) | 0.002 |
Note: All variables were standardized (Z‐scores) before inclusion in the model. p < 0.05 was considered statistically significant.
Abbreviations: B, regression coefficients; CI, confidence interval; OR, odds ratios; SE, standard errors.
3.3. Mediation Analysis
A parallel multiple mediation analysis (Model 4) was conducted to examine the potential mediating roles of CMI and EATi in the association between the TyG index and CAD. The TyG index demonstrated a significant direct association with CAD (direct effect: 3.189, 95% CI: 1.707–4.671). Regarding indirect effects, a statistically significant mediation pathway was observed through CMI (B = 1.171, 95% CI: 0.224–2.609), whereas the pathway via EATi was not statistically significant (B = 0.245, 95% CI: –0.135–0.795). The total indirect effect was significant (B = 1.416, 95% CI: 0.331–3.052). Pairwise comparisons showed no significant difference between the CMI and EATi pathways (B = –0.926, 95% CI: –2.324–0.094) (Table 5, Figure 2).
Table 5.
Parallel mediation analysis of TyG index on CAD via EATi and CMI.
| Path number | Effect (log‐odds) | BootSE | 95% CI |
|---|---|---|---|
| Direct effect | 3.189 | 0.756 | 1.707−4.671 |
| Ind 1 | 0.245 | 0.235 | 0.135−0.795 |
| Ind 2 | 1.171 | 0.613 | 0.224−2.609 |
| Total | 1.416 | 0.688 | 0.331−3.052 |
| (C1) | −0.926 | 0.623 | 2.324−0.094 |
Note: All effects are expressed in a log‐odds metric. Ind 1 = TyG → EATi → CAD; Ind 2 = TyG → CMI → CAD. The 95% CI was calculated using the percentile bootstrap method (5000 samples). An effect is considered statistically significant when the 95% CI does not contain zero.
Figure 2.
Mediation Analysis of CMI and EATi in the association between the TyG index and CAD (Model 4). Values on the paths are regression coefficients (B) expressed in a log‐odds metric. Solid lines represent significant pathways, while the dashed line indicates a non‐significant pathway (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001.
Specifically, the TyG index was significantly associated with CMI (B = 0.608, p < 0.001), and CMI was significantly associated with CAD (B = 1.926, p = 0.033). In contrast, although EATi was significantly associated with CAD (B = 0.073, p = 0.001), its association with the TyG index was not statistically significant (B = 3.376, p = 0.141). (Supporting Information S1: Table 2, Figure 2).
3.4. Predictive Performance of TyG INDEX, CMI, EATI, and Their Combined Assessment for CAD
ROC curve analysis demonstrated that both the TyG index and CMI exhibited good discriminatory ability for CAD, with comparable performance (TyG: AUC = 0.801 [95% CI: 0.736–0.866]; CMI: AUC = 0.762 [95% CI: 0.685–0.838]; DeLong test, Z = 1.172, p = 0.2413). In contrast, EATi showed a significantly lower discriminative ability, AUC = 0.629 (95% CI: 0.550–0.708) (Figure 3).
Figure 3.
ROC curve analysis was performed to assess the discriminative power of individual predictors and their combined assessment for CAD. Combined represents TyG + EATi + CMI.
When the three indicators were evaluated in combination, the discriminatory performance was further improved, yielding an AUC of 0.847 (95% CI: 0.788–0.906), with a sensitivity of 80.0% and a specificity of 76.7%. DeLong's tests indicated that the combined assessment was significantly superior to the TyG index alone (Z = 2.581, p = 0.0099), CMI alone (Z = 3.142, p = 0.0017), and EATi alone (Z = 4.945, p < 0.001). Additionally, both the TyG index and CMI demonstrated significantly better discriminatory performance than EATi (TyG vs. EATi: Z = 3.185, p = 0.0014; CMI vs. EATi: Z = 2.289, p = 0.0221).
4. Discussion
This study demonstrated that the TyG index, CMI, and EATi were all independently associated with CAD, although EATi was not significantly correlated with TyG or CMI. The combined evaluation of these markers improved the discrimination of CAD compared with each individual indicator (AUC = 0.847). The multivariable model incorporating these variables showed a Nagelkerke R 2 of 0.445, suggesting that these markers collectively account for a moderate proportion of the variance in CAD risk.
Mediation analysis further suggested that CMI may partially mediate the association between the TyG index and CAD, whereas EATi is independently associated with CAD. The absence of a significant indirect effect through the EATi pathway indicates that metabolic and adiposity‐related measures contribute to CAD through partially independent mechanisms. These findings suggest that systemic cardiometabolic dysfunction and epicardial fat accumulation may operate through parallel, rather than strictly sequential, pathways in the development of coronary atherosclerosis.
Consistent with this framework, patients with CAD in the present study exhibited significantly higher TyG index and CMI values, reflecting the presence of metabolic dysregulation and visceral adiposity commonly observed in atherosclerotic disease. Both indices have been widely reported as indicators of insulin resistance, dyslipidemia, and adverse cardiometabolic profiles, which are closely linked to the development and progression of coronary atherosclerosis [5].
The TyG index, derived from the product of fasting plasma glucose and triglyceride levels, has been recognized as a reliable surrogate marker for IR and metabolic dysfunction. In this study, the TyG index demonstrated a significant and independent association with both the presence of CAD. Each 1‐standard‐deviation increase in TyG was associated with markedly higher odds of having CAD, supporting its relevance to cardiometabolic disturbances that accompany atherosclerotic processes. These findings are consistent with previous research showing that elevated TyG values correlate with coronary artery calcification, overall atherosclerotic burden, and adverse cardiovascular outcomes [14, 15, 16].
Recent observational and cohort studies further reinforce these associations. For example, Liu et al. reported that higher TyG levels were linked to increased cardiovascular events and all‐cause mortality, independent of traditional risk factors [17]. Other studies have shown that TyG is related not only to the presence of coronary plaques but also to plaque severity, including multivessel involvement and complex lesion characteristics [18].
Several biological mechanisms may underlie these observations. Insulin resistance is known to reduce nitric oxide bioavailability, impair endothelial function, and promote oxidative stress and low‐grade inflammation. It also contributes to dysregulated lipid metabolism, including elevated VLDL and reduced HDL concentrations, which collectively facilitate lipid infiltration, macrophage recruitment, and foam‐cell formation—early hallmarks of atherosclerotic plaque development. These interconnected pathways help explain the observed relationship between TyG and CAD in this study [19, 20, 21].
Taken together, the TyG index appears to reflect key metabolic disturbances relevant to coronary atherosclerosis and may provide clinically useful information in understanding cardiometabolic alterations among individuals at risk for CAD [22]. While our findings support its potential value, further prospective studies are necessary to clarify its role across diverse populations and to better define how TyG may contribute to a broader, multidimensional assessment of cardiometabolic health.
The CMI, which integrates lipid metabolism through the triglyceride‐to‐HDL cholesterol ratio and central adiposity via the waist‐to‐height, has gained increasing recognition as a comprehensive indicator of metabolic burden and cardiovascular risk [23]. As a composite measure, CMI reflects the combined effects of dyslipidemia and visceral adiposity—two key contributors to cardiometabolic dysfunction. Elevated CMI levels have been associated with HTN, type 2 diabetes, dyslipidemia, and non‐alcoholic fatty liver disease (NAFLD) in multiple populations, underscoring its relevance as a marker of systemic metabolic disturbance [24, 25, 26].
In the present study, CMI demonstrated an independent association with CAD. A one‐standard‐deviation increase in CMI was associated with a 3.34‐fold higher likelihood of CAD, supporting its ability to capture the synergistic influence of visceral adiposity and metabolic dysregulation on atherogenesis. This finding aligns with previous research showing that higher CMI levels correlate with CAD presence even after adjustment for traditional cardiovascular risk factors [27]. Furthermore, mediation analysis identified CMI as a key intermediary in the association between the TyG index and CAD, suggesting that metabolic abnormalities reflected by TyG may promote atherosclerosis partially through pathways linked to central adiposity. This highlights the potential value of CMI in characterizing metabolic–adiposity interactions relevant to coronary disease.
EAT has also emerged as a critical contributor to CAD pathogenesis because of its close anatomical proximity to the coronary arteries and its capacity to secrete inflammatory adipokines. Functionally, EAT is a metabolically active organ that secretes pro‐inflammatory cytokines and adipokines, which directly impact coronary vasculature and contribute to atherosclerosis [28, 29]. Previous studies have demonstrated consistent associations between increased EATV and both the presence and severity of CAD [30, 31]. Consistently, EATi was significantly higher among patients with CAD, supporting the concept that EAT is not merely a marker of adiposity but a metabolically active tissue involved in local vascular inflammation and plaque development.
In addition, EATi showed no significant correlation with the TyG index or CMI, suggesting that EATi may act as a relatively distinct pathogenic component rather than a mere downstream manifestation of systemic metabolic and adiposity disturbances. Consistently, EATi did not significantly mediate the TyG–CAD pathway. This lack of association might be explained by the complex, non‐linear relationship between insulin resistance and epicardial fat. EATi may promote atherosclerosis through mechanisms independent of systemic metabolism, such as localized paracrine signaling and direct anatomical interactions with coronary vessels. Recent evidence linking EAT to coronary microvascular dysfunction [32] provides a plausible perspective on the independent deleterious impact of EATi observed here. Finally, qualitative features of epicardial fat, such as inflammatory activity, may have greater implications for coronary plaque vulnerability than fat volume alone [33, 34]. a factor not assessed in this study but worthy of further investigation.
5. Limitations of the Study
This study has several limitations. First, its retrospective case–control design prevents causal inference, and the observed associations should be interpreted as exploratory. Although multivariable adjustments were performed to minimize confounding, strict 1:1 matching between cases and controls was not applied. Prospective studies with more rigorous matching strategies are needed to further validate these findings and clarify the temporal relationships among TyG, CMI, EATi, and CAD. Second, the study population consisted of hospitalized patients with CAD, which may limit the generalizability of the findings to broader or asymptomatic populations. Larger and more diverse cohorts would improve external validity. Third, EATi was assessed using CCTA, which quantifies fat volume but does not capture qualitative inflammatory or compositional features that may also influence CAD risk. Additional imaging modalities could provide a more comprehensive evaluation. Fourth, anthropometric data, particularly WC, were derived from routine clinical records and may be subject to variability in measurement and recording. Therefore, WC‐related indices should be interpreted with caution. Finally, other potentially relevant biomarkers—such as inflammatory cytokines or genetic factors—were not assessed. Future studies incorporating a wider range of metabolic, inflammatory, and imaging indicators may help refine the understanding of cardiometabolic pathways related to CAD.
6. Future Research Directions
Future studies should employ longitudinal designs to clarify the temporal relationships among TyG, CMI, EATi, and CAD progression, as the present cross‐sectional analysis cannot establish causality. More refined imaging approaches—such as MRI or PET—may help characterize qualitative features of epicardial fat and further validate its role in coronary atherosclerosis. In addition, interventional studies targeting metabolic dysfunction or reducing epicardial fat, including pharmacologic or lifestyle strategies, are warranted to determine whether modifying these markers can reduce CAD risk. These efforts may ultimately contribute to more precise and individualized cardiovascular prevention strategies.
7. Conclusions
This study demonstrates that the TyG index, CMI, and EATi are each independently associated with obstructive CAD, and that their combined assessment improves the identification of disease presence. The observed statistical pathways suggest that metabolic and adiposity‐related disturbances may contribute to coronary atherosclerosis through partially independent mechanisms, with CMI potentially acting as an intermediary link between systemic metabolic dysfunction and CAD, whereas EATi appears to be independently associated with disease rather than serving as a downstream mediator. Collectively, these findings suggest the potential value of integrated metabolic and adiposity markers in CAD assessment, pending validation in prospective, multicenter studies.
Author Contributions
Gao Z.H. was responsible for conceptualization, supervision, project administration, and critical review of the manuscript. Yin Y.L. and Zuo Y.Q. performed the methodology, software, formal analysis, data curation, and visualization. Yang X. and Xu L. handled the investigation and resources. Jia M. and Zhai S.M. prepared the original draft. Jia M. and Zhai S.M. contributed equally to this work. All authors contributed to data interpretation, critically revised the manuscript for important intellectual content, and approved the final version for submission.
Funding
The authors have nothing to report.
Ethics Statement
This retrospective study was approved by the Research Ethics Committee of the Second Hospital of Hebei Medical University (Approval No. 2025‐R206). The requirement for informed consent was waived due to the retrospective nature of the study, in accordance with national legislation and institutional requirements.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Supporting File: clc70375‐sup‐0001‐Supplementary_Materials.docx.
Min J., Shumei Z., Yuqiang Z., et al., “Association of the TyG Index, Cardiometabolic Index, and Epicardial Adipose Tissue With Coronary Artery Disease,” Clinical Cardiology 49 (2026): e70375. 10.1002/clc.70375.
Jia Min and Zhai Shumei have contributed equally to this work.
Data Availability Statement
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Lindstrom M., DeCleene N., Dorsey H., et al., “Global Burden of Cardiovascular Diseases and Risks Collaboration, 1990‐2021,” Journal of the American College of Cardiology 80, no. 25 (2022): 2372–2425, 10.1016/j.jacc.2022.11.001. [DOI] [PubMed] [Google Scholar]
- 2. Tesche C., Bauer M. J., Straube F., et al., “Association of Epicardial Adipose Tissue With Coronary CT Angiography Plaque Parameters on Cardiovascular Outcome in Patients With and Without Diabetes Mellitus,” Atherosclerosis 363 (2022): 78–84, 10.1016/j.atherosclerosis.2022.10.006. [DOI] [PubMed] [Google Scholar]
- 3. Napoli G., Pergola V., Basile P., et al., “Epicardial and Pericoronary Adipose Tissue, Coronary Inflammation, and Acute Coronary Syndromes,” Journal of Clinical Medicine 12, no. 23 (2023): 7212, 10.3390/jcm12237212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Khan I., Hondros C. A. B., Hanseth S., et al., “Epicardial Adipose Tissue Volume, Compositional Plaque Progression, and Vulnerability in Nonobstructive Coronary Artery Disease,” Coronary Artery Disease 36, no. 7 (2022): 580–586, 10.1097/MCA.0000000000001539. [DOI] [PubMed] [Google Scholar]
- 5. Liang S., Wang C., Zhang J., et al., “Triglyceride‐Glucose Index and Coronary Artery Disease: A Systematic Review and Meta‐Analysis of Risk, Severity, and Prognosis,” Cardiovascular Diabetology 22, no. 1 (2023): 170, 10.1186/s12933-023-01906-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Wang W., Yang J., Wang K., et al., “Association Between the Triglyceride‐Glucose Index and In‐Hospital Major Adverse Cardiovascular Events in Patients With Acute Coronary Syndrome: Results From the Improving Care for Cardiovascular Disease in China (CCC)‐Acute Coronary Syndrome Project,” Cardiovascular Diabetology 23, no. 1 (2024): 170, 10.1186/s12933-024-02270-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Lazzer S., D'Alleva M., Isola M., et al., “Cardiometabolic Index (CMI) and Visceral Adiposity Index (VAI) Highlight a Higher Risk of Metabolic Syndrome in Women With Severe Obesity,” Journal of Clinical Medicine 12, no. 9 (2023): 3055, 10.3390/jcm12093055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mohammadian Khonsari N., Khashayar P., Shahrestanaki E., et al., “Normal Weight Obesity and Cardiometabolic Risk Factors: A Systematic Review and Meta‐Analysis,” Frontiers in Endocrinology 13 (2022): 857930, 10.3389/fendo.2022.857930. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Knuuti J., Wijns W., Saraste A., et al., “2019 ESC Guidelines for the Diagnosis and Management of Chronic Coronary Syndromes,” European Heart Journal 41, no. 3 (2020): 407–477, 10.1093/eurheartj/ehz425. [DOI] [PubMed] [Google Scholar]
- 10. Gao Z., Zuo Y., Jia L., et al., “Association Between Epicardial Adipose Tissue Density and Characteristics of Coronary Plaques Assessed by Coronary Computed Tomographic Angiography,” International Journal of Cardiovascular Imaging 38, no. 3 (2022): 673–681, 10.1007/s10554-021-02428-4. [DOI] [PubMed] [Google Scholar]
- 11. Zhihong G., Yuqiang Z., Linyi J., et al., “Correlation Analysis Between Epicardial Adipose Tissue and Acute Coronary Syndrome,” Scientific Reports 15, no. 1 (2025): 3015, 10.1038/s41598-025-87594-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Milanese G., Silva M., Bruno L., et al., “Quantification of Epicardial Fat With Cardiac CT Angiography and Association With Cardiovascular Risk Factors in Symptomatic Patients: From the ALTER‐BIO (Alternative Cardiovascular Bio‐Imaging Markers) Registry,” Diagnostic and Interventional Radiology 25, no. 1 (2019): 35–41, 10.5152/dir.2018.18037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Shmilovich H., Dey D., Cheng V. Y., et al., “Threshold for the Upper Normal Limit of Indexed Epicardial Fat Volume: Derivation in a Healthy Population and Validation in an Outcome‐Based Study,” American Journal of Cardiology 108, no. 11 (2011): 1680–1685, 10.1016/j.amjcard.2011.07.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Saffar Soflaei S., Salehi‐Sangani P., Fallahi Z., et al., “Triglyceride‐Glucose Index Association With Severity of Coronary Artery Disease,” Endocrinol Diabetes Metab 8, no. 2 (2025): e70025, 10.1002/edm2.70025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ormazabal V., Nair S., Elfeky O., Aguayo C., Salomon C., and Zuñiga F. A., “Association Between Insulin Resistance and the Development of Cardiovascular Disease,” Cardiovascular Diabetology 17, no. 1 (2018): 122, 10.1186/s12933-018-0762-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Gounden V., Devaraj S., and Jialal I., “The Role of the Triglyceride‐Glucose Index as a Biomarker of Cardio‐Metabolic Syndromes,” Lipids in Health and Disease 23, no. 1 (2024): 416, 10.1186/s12944-024-02412-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Liu X., Tan Z., Huang Y., et al., “Relationship Between the Triglyceride‐Glucose Index and Risk of Cardiovascular Diseases and Mortality in the General Population: A Systematic Review and Meta‐Analysis,” Cardiovascular Diabetology 21, no. 1 (2022): 124, 10.1186/s12933-022-01546-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Siverio‐Morales O., Mora‐Fernández C., Hernández‐Carballo C., et al., “Predictive Value of Triglyceride‐Glucose Index for the Evaluation of Coronary Artery Disease Severity and Occurrence of Major Adverse Cardiovascular Events,” American Journal of Physiology‐Heart and Circulatory Physiology 328, no. 1 (2025): H14–H20, 10.1152/ajpheart.00684.2024. [DOI] [PubMed] [Google Scholar]
- 19. Adams‐Huet B. and Jialal I., “An Increasing Triglyceride‐Glucose Index Is Associated With a Pro‐Inflammatory and Pro‐Oxidant Phenotype,” Journal of Clinical Medicine 13, no. 13 (2024): 3941, 10.3390/jcm13133941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Paneni F., “Role of Oxidative Stress in Endothelial Insulin Resistance,” World Journal of Diabetes 6, no. 2 (2015): 326–332, 10.4239/wjd.v6.i2.326. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Kosmas C. E., Bousvarou M. D., Kostara C. E., Papakonstantinou E. J., Salamou E., and Guzman E., “Insulin Resistance and Cardiovascular Disease,” Journal of International Medical Research 51, no. 3 (2023): 3000605231164548, 10.1177/03000605231164548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Scott D. A., Ponir C., Shapiro M. D., and Chevli P. A., “Associations Between Insulin Resistance Indices and Subclinical Atherosclerosis: A Contemporary Review,” American Journal of Preventive Cardiology 18 (2024): 100676, 10.1016/j.ajpc.2024.100676. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Chen X., Zhao Y., Sun J., Jiang Y., and Tang Y., “Identification of Metabolic Syndrome Using Lipid Accumulation Product and Cardiometabolic Index Based on NHANES Data From 2005 to 2018,” Nutrition & Metabolism 21, no. 1 (2024): 96, 10.1186/s12986-024-00864-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Guo T., Zhou Y., Yang G., Sheng L., and Chai X., “Association Between Cardiometabolic Index and Hypertension Among US Adults From NHANES 1999‐2020,” Scientific Reports 15, no. 1 (2025): 4007, 10.1038/s41598-025-87029-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Song J., Li Y., Zhu J., et al., “Non‐Linear Associations of Cardiometabolic Index With Insulin Resistance, Impaired Fasting Glucose, and Type 2 Diabetes Among US Adults: A Cross‐Sectional Study,” Front Endocrinol (Lausanne) 15 (2024): 1341828, 10.3389/fendo.2024.1341828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Cheng L., Wu Q., and Wang S., “Association Between Cardiometabolic Index and Hepatic Steatosis and Liver Fibrosis: A Population‐Based Study,” Hormones 23, no. 3 (2024): 477–486, 10.1007/s42000-024-00572-9. [DOI] [PubMed] [Google Scholar]
- 27. Wei Q., Cheng X., Li M., Wu K., Chen M., and Zhang D., “Associations Between the Cardiometabolic Index and Atherosclerotic Cardiovascular Disease Across Different Glucose Metabolism Statuses: Insights From Nhanes, 1999–2020,” Lipids in health and disease 24, no. 1 (2025): 93, 10.1186/s12944-025-02508-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Iacobellis G., “Epicardial Fat Links Obesity to Cardiovascular Diseases,” Progress in Cardiovascular Diseases 78 (2023): 27–33, 10.1016/j.pcad.2023.04.006. [DOI] [PubMed] [Google Scholar]
- 29. Li C., Liu X., Adhikari B. K., et al., “The Role of Epicardial Adipose Tissue Dysfunction in Cardiovascular Diseases: An Overview of Pathophysiology, Evaluation, and Management,” Front Endocrinol (Lausanne) 14 (2023): 1167952, 10.3389/fendo.2023.1167952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Zhou H., Pan Y., Du J., Liang F., Ma X., and Lv D., “Association of Epicardial Fat Volume With the Severity of Coronary Artery Disease: A Preliminary Study on Risk Prediction of Obstructive Coronary Heart Disease,” BMC Cardiovascular Disorders 25, no. 1 (2025): 293, 10.1186/s12872-025-04743-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Yu W., Liu B., Zhang F., et al., “Association of Epicardial Fat Volume With Increased Risk of Coronary Artery Disease in Chinese Patients With Suspected Coronary Artery Disease,” Journal of the American Heart Association 10, no. 6 (2021): e018080, 10.1161/JAHA.120.018080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Abusnina W., Merdler I., Cellamare M., et al., “Epicardial Fat Tissue: A Potential Marker for Coronary Microvascular Dysfunction,” Journal of the American Heart Association 14, no. 3 (2025): e038484, 10.1161/JAHA.124.038484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Goeller M., Achenbach S., Cadet S., et al., “Pericoronary Adipose Tissue Computed Tomography Attenuation and High‐Risk Plaque Characteristics in Acute Coronary Syndrome Compared With Stable Coronary Artery Disease,” JAMA Cardiology 3, no. 9 (2018): 858–863, 10.1001/jamacardio.2018.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Nerlekar N., Brown A. J., Muthalaly R. G., et al., “Association of Epicardial Adipose Tissue and High‐Risk Plaque Characteristics: A Systematic Review and Meta‐Analysis,” Journal of the American Heart Association 6, no. 8 (2017): e006379, 10.1161/JAHA.117.006379. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting File: clc70375‐sup‐0001‐Supplementary_Materials.docx.
Data Availability Statement
The data supporting the findings of this study are available from the corresponding author upon reasonable request.