Does diabetes status modify the association between the triglyceride-glucose index and major adverse cardiovascular events in patients with coronary heart disease? A systematic review and meta-analysis of longitudinal cohort studies | Cardiovascular Diabetology

Protocol registration

This study was conducted in strict accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [18]. The protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO) under the registration number CRD: 420251018545.

Eligibility criteria

Two independent investigators screened the titles and abstracts of studies retrieved from the databases to identify eligible publications. The inclusion criteria were as follows:

1. 1.

   Study design: Only prospective or retrospective cohort studies were included;

2. 2.

   Population: Participants had to be aged ≥ 18 years, and all subjects were clearly diagnosed with CHD;

3. 3.

   Diabetes status: The diabetes status of participants had to be clearly defined, distinguishing between those with and without diabetes, or the study must have provided subgroup analyses on the basis of diabetes status. Diabetes had to be diagnosed according to internationally recognized criteria (e.g., fasting plasma glucose, oral glucose tolerance test, or HbA1c);

4. 4.

   Exposure: Studies were required to evaluate the TyG index as an exposure factor. The TyG index could be analysed as a categorical variable (e.g., quartiles or specific thresholds) or a continuous variable, and HRs with corresponding 95% confidence intervals (CIs) had to be reported;

5. 5.

   Outcomes: Studies had to report at least one of the following primary outcomes: adverse cardiovascular events, all-cause mortality, nonfatal myocardial infarction, nonfatal stroke, or revascularization.

Studies were excluded if they met the following criteria:

1. 1.

   Full text not available;

2. 2.

   Missing critical data;

3. 3.

   Failure to report the association between the TyG index and outcomes (HR and 95% CI) specifically for participants with or without diabetes;

4. 4.

   Duplicate publications.

Literature search

To identify all relevant studies, we systematically searched PubMed, the Cochrane Library, Web of Science, and Embase using a combination of Medical Subject Headings (MeSH) and free-text terms from database inception to March 13, 2025. We included longitudinal studies investigating the associations between the TyG index and adverse cardiovascular outcomes in patients with CHD.

The exposure of interest was the TyG index. The search terms included “triglyceride-glucose index,” “TyG index,” and “TyG.”

The study population included patients with CHD, and the search terms included “coronary heart disease,” “CHD,” “coronary artery disease,” “CAD,” “acute coronary syndrome,” “ACS,” “myocardial infarction,” “MI,” “ST-segment elevation myocardial infarction,” “STEMI,” “non-ST-segment elevation myocardial infarction,” “NSTEMI,” and “unstable angina” (Table S1).

To ensure a comprehensive search, outcome-related terms (e.g., major adverse cardiovascular events, all-cause mortality) were not included in the search strategy. However, all retrieved titles and abstracts were carefully screened by the investigators to identify studies that met the inclusion criteria. Additionally, the reference lists of all included studies were manually reviewed to identify any potentially relevant studies that might have been missed.

Study selection and data extraction

Two independent investigators (YGC and YXZ) screened the titles and abstracts of all the retrieved studies to identify those that met the inclusion criteria. In cases of disagreement, a third investigator was consulted to reach a consensus. The full texts of the selected studies were then reviewed to determine the final set of studies included in the meta-analysis.

An electronic data extraction form was developed to collect the following information: first author’s name, year of publication, study location, total number of participants, number of outcome events, number of participants with diabetes, number without diabetes, HRs with 95% CIs for the highest versus lowest TyG index category stratified by diabetes status, HRs with 95% CIs for the TyG index as a continuous variable stratified by diabetes status, and adjusted confounding variables. Two independent researchers (YGC and SCX) extracted the data via this approach. For studies that did not focus exclusively on diabetic or nondiabetic populations, subgroup results stratified by diabetes status were extracted whenever available.

Quality assessment

The quality of the included studies was independently assessed by two investigators (YYD and SLH) using the Newcastle–Ottawa Scale (NOS) [19]. On the basis of the NOS score, study quality was categorized as high (≥ 8 points), moderate (5–7 points), or low (< 5 points).

Statistical analysis

Data analysis was performed using Stata software (version 17.0) and R software (version 4.4.1). The consistency of the literature selection, data extraction, and quality assessment between the two independent reviewers was evaluated using Cohen’s kappa statistic. The interpretation of the kappa values followed the standard proposed by Landis and Koch[20]: < 0 indicated poor agreement, 0.00–0.20 slight agreement, 0.21–0.40 fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 substantial agreement, and 0.81–1.00 almost perfect agreement. Hazard ratios (HRs) and corresponding 95% confidence intervals (95% CIs) were used as pooled effect estimates for the TyG index, whether treated as a categorical or continuous variable. Heterogeneity across studies was assessed using the I² statistic. If I² > 50%, a random-effects model (DerSimonian–Laird method) was used for pooling; if I² ≤ 50%, a fixed-effects model (Mantel–Haenszel method) was applied. Subgroup analyses and meta-regression were conducted to further explore potential sources of heterogeneity. Following the method proposed by Greenland and Longnecker [21] and extended by Orsini et al. [22], a dose‒response meta-analysis was conducted using a restricted cubic spline (RCS) model to evaluate the potential nonlinear association between the TyG index and the risk of MACEs. The RCS model is a flexible regression approach that fits piecewise cubic polynomials between predefined knots, allowing the estimation of smooth, potentially nonlinear exposure–outcome relationships. In this analysis, three knots were placed at fixed percentiles (commonly the 10th, 50th, and 90th percentiles) of the TyG index distribution, following standard methodological guidelines. Nonlinearity was tested by assessing whether the coefficient of the second spline term was significantly different from zero using a Wald chi-square test. A statistically significant result indicated a deviation from linearity, supporting the application of the spline-based model.

Publication bias was initially assessed by visual inspection of funnel plot symmetry and further examined using Egger’s test and Begg’s test for quantitative analysis. Sensitivity analysis was conducted by sequentially removing individual studies to evaluate the stability of the results. A two-sided p value of < 0.05 was considered statistically significant.

Literature selection process and results

A total of 982 articles were initially identified through searches of the PubMed, Cochrane Library, Web of Science, and Embase databases. After removing duplicates, 589 articles remained. The titles and abstracts of these articles were screened, and studies that were irrelevant to the topic or failed to meet the inclusion criteria were excluded. A total of 88 articles were selected for full-text evaluation. After a thorough review and strict screening of the full texts, 36 articles met the inclusion criteria. Ultimately, 36 studies [10,11,12,13, 15,16,17, 23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51] were included in the meta-analysis (Fig. 1). The kappa value for literature selection between the two reviewers was 0.838, indicating an almost perfect level of agreement.

Study characteristics and quality assessment

The publication years of the included studies ranged from 2020 to 2025. The number of participants varied from 231 to 101,113, with follow-up durations ranging from 12 to 85 months. The majority of studies were conducted in China. Among the 36 included studies, 12 studies [23, 24, 28,29,30,31,32,33, 35,36,37,38, 51] exclusively included diabetic patients, 4 studies [41, 42, 45, 46] exclusively included nondiabetic patients, and the remaining 19 studies [10,11,12,13, 15,16,17, 25,26,27, 34, 39, 40, 43, 44, 47,48,49,50] included both diabetic and nondiabetic patients, reporting results separately for each group. With respect to the TyG index, 12 studies [11, 12, 24, 27, 31, 33, 35, 38, 45, 49,50,51] analysed the TyG index as a categorical variable (e.g., divided into quartiles or tertiles), whereas 15 studies [10, 13, 15,16,17, 23, 25, 26, 30, 34, 41,42,43,44, 47] treated the TyG index as a continuous variable. Nine studies [28, 29, 32, 36, 37, 39, 40, 46, 48] conducted both categorical and continuous analyses. The kappa value for data extraction between the two reviewers was 0.7500, indicating substantial agreement. The Newcastle‒Ottawa Scale (NOS) scores of the included studies ranged from 6 to 9, with 22 studies classified as high-quality and 14 studies classified as moderate-quality (Table S3). The kappa value for the study quality assessment between the two reviewers was 0.778, indicating a substantial level of agreement.

Association between the TyG index and MACEs risk in diabetic patients

Among diabetic patients, a higher TyG index was significantly associated with an increased risk of MACEs. This conclusion was derived from 29 studies, including 14 studies [11, 12, 29, 31,32,33, 35,36,37,38,39,40, 48, 49] that treated the TyG index as a categorical variable and 22 studies [10, 13, 15,16,17, 23, 25,26,27,28,29,30, 32, 34, 36, 37, 39, 40, 43, 44, 47] that analysed it as a continuous variable. The included studies, published between 2020 and 2025, involved a total of 161,273 participants, of whom 117,827 were diabetic, and reported 7539 MACEs. Pooled analysis of categorical data revealed that patients with higher TyG index values had a significantly elevated risk of MACEs (HR = 1.98, 95% CI 1.61–2.43, P < 0.001, I² = 86.6, Cochran’s Q test: P < 0.001) (Fig. 2A). Similarly, the analysis of TyG as a continuous variable demonstrated consistent results (HR = 1.57, 95% CI 1.38–1.78, P < 0.001, I² = 70.9, Cochran’s Q test: P < 0.001) (Fig. 2B). The funnel plots for both categorical and continuous analyses of the TyG index appeared asymmetric (Figs. S1 and S2), and Egger’s test indicated potential publication bias, with p values of 0.022 and 0.019, respectively.

Association between the TyG index and the risk of MACEs in nondiabetic patients

Among nondiabetic patients, a higher TyG index is significantly associated with an increased risk of MACEs. This conclusion is derived from 22 studies published between 2020 and 2025, encompassing 131,681 participants, 45,639 of whom were nondiabetic, with 6,178 MACEs reported during the follow-up period. Among these studies, 8 [11, 12, 39, 40, 45, 46, 48, 49] analysed the TyG index as a categorical variable, and 17 [10, 13, 15,16,17, 25,26,27, 34, 39,40,41,42,43,44, 46, 47] analysed it as a continuous variable. The pooled HR for studies treating the TyG index as a categorical variable was 1.65 (95% CI 1.33–2.05, P < 0.001, I² = 86.5, Cochran’s Q test: P < 0.001) (Fig. 3A), whereas the HR for studies using the TyG index as a continuous variable was 1.74 (95% CI 1.46–2.06, P < 0.001, I² = 75.9, Cochran’s Q test: P < 0.001) (Fig. 3B), indicating a consistent positive association. Funnel plots for both types of analyses appeared symmetrical upon visual inspection (Figs. S2 and S4), and Egger’s test yielded p values of 0.095 and 0.146, respectively, suggesting no significant publication bias.

Association between the TyG index and all-cause mortality

A total of 8 studies were included in the analysis, with publication years ranging from 2020 to 2025 and follow-up durations between 12 and 36 months. Of these, 6 studies [31, 35,36,37, 50, 51] focused on diabetic patients, whereas 3 studies [36, 45, 50] focused on nondiabetic patients. These studies involved 11,645 patients with CHD, including 7611 diabetic patients and 4034 nondiabetic patients. During the follow-up period, 923 all-cause deaths were reported. Among diabetic patients, a higher TyG index was significantly associated with an increased risk of all-cause mortality, with the highest vs. lowest pooled HR of 1.74 (95% CI 1.45–2.08, P < 0.01, I² = 45.9, Cochran’s Q test: P = 0.100) (Fig. 4A). Similarly, among nondiabetic patients, a higher TyG index was also significantly associated with an increased risk of all-cause mortality, with the highest vs. lowest pooled HR of 1.50 (95% CI 1.18–1.90, P < 0.001, I² = 38.9, Cochran’s Q test: P = 0.195) (Fig. 4B).

Association between the TyG index and nonfatal myocardial infarction

A total of 6 studies published between 2020 and 2022 were included in the analysis, with follow-up durations ranging from 24 to 48 months. Among them, 4 studies [31, 35,36,37] focused on patients with diabetes, and 2 [45, 46] focused on nondiabetic individuals. These studies collectively involved 10,004 patients with CHD, including approximately 6839 with diabetes and 3165 without diabetes. During the follow-up period, a total of 450 nonfatal myocardial infarction events were recorded. In diabetic patients, a higher TyG index was significantly associated with an increased risk of nonfatal myocardial infarction, with the highest vs. lowest pooled HR of 2.05 (95% CI 1.52–2.77, P < 0.001, I² = 0.00, Cochran’s Q test: P = 0.518) (Fig. 5A). Similarly, in nondiabetic patients, a higher TyG index was also significantly associated with an increased risk of nonfatal myocardial infarction, with the highest vs. lowest pooled HR of 2.46 (95% CI 1.11–5.47, P = 0.027, I² = 59.3, Cochran’s Q test: P = 0.117) (Fig. 5B).

Association between the TyG index and nonfatal stroke

A total of 5 studies published between 2015 and 2025 were included in the analysis, with follow-up durations ranging from 24 to 48 months. These studies included 9206 patients with CHD, including 6041 with diabetes and 3165 without diabetes. Among the included studies, 4 focused on diabetic patients, and 2 focused on nondiabetic patients. During the follow-up period, a total of 226 nonfatal stroke events were reported. In patients with diabetes, a higher TyG index was significantly associated with an increased risk of nonfatal stroke, with the highest vs. lowest pooled HR of 1.73 (95% CI 1.12–2.66, P = 0.013, I² = 0.00, Cochran’s Q test: P = 0.667) (Fig. 6A). However, in nondiabetic patients, no statistically significant association was observed between a higher TyG index and the risk of nonfatal stroke, with the highest vs. lowest pooled HR of 1.66 (95% CI 0.88–3.12, P = 0.118, I² = 0.00, Cochran’s Q test: P = 0.764) (Fig. 6B).

Association between the TyG index and revascularization

A total of 5 studies published between 2020 and 2023 were included in the analysis, with follow-up durations ranging from 18.83 to 48 months. Among these, 3 studies [24, 35, 36] focused on diabetic patients, and 2 [45, 46] focused on nondiabetic patients. In total, 6528 patients with CHD were included, comprising 3363 with diabetes and 3165 without diabetes. During the follow-up period, 735 patients underwent revascularization. In patients with diabetes, a higher TyG index was significantly associated with an increased risk of revascularization, with an HR of 2.52 (95% CI 1.26–5.04, P < 0.009, I² = 91.2, Cochran’s Q test: P < 0.001) (Fig. 7A). Similarly, in nondiabetic patients, a higher TyG index was also significantly associated with an increased risk of revascularization, with an HR of 2.09 (95% CI 1.57–2.76, P < 0.001, I² = 0.00, Cochran’s Q test: P = 0.342) (Fig. 7B).

Dose‒response meta-analysis

Owing to limitations in the available data, the dose‒response meta-analysis was conducted only among diabetic patients to evaluate the association between the TyG index and the risk of MACEs. The analysis demonstrated a significant nonlinear association between the TyG index and MACEs risk (χ² = 17.21, df = 1, P_nonlinearity < 0.0001). Specifically, each one-unit increase in the TyG index was associated with a 1.97-fold increased risk of MACEs (HR = 1.97, 95% CI = 1.17–2.77, P_{dose–response} < 0.0001) (Fig. 8).

Sensitivity analysis and subgroup analysis

Sensitivity analysis was conducted using the leave‒one‒out method to evaluate the robustness of the association between the TyG index and the risk of MACEs in patients with CHD across different diabetes statuses. The results showed that, regardless of diabetes status, the overall risk estimates changed minimally after each individual study was sequentially excluded, suggesting that the findings were relatively stable and robust (Figs. S5–S8).

A series of subgroup analyses were conducted on the basis of sample size, ACS presence, PCI status, follow-up duration, study quality, and median age of participants. Regardless of whether the TyG index was analysed as a categorical or continuous variable, a higher TyG index tended to be associated with an increased risk of MACEs among both diabetic and nondiabetic patients with CHD. These subgroup findings are generally consistent with the main results and suggest the potential of the TyG index as a potentially reliable marker for predicting the risk of MACEs in patients with CHD. Furthermore, meta-regression analysis suggested that differences in study quality might be a potential source of heterogeneity. Notably, regardless of the analytic type of the TyG index, studies with higher methodological quality appeared to show reduced heterogeneity in both diabetic and nondiabetic patients (Tables 1 and S4–S6).