Triglyceride-glucose index for predicting repeat revascularization and in-stent restenosis in patients with chronic coronary syndrome undergoing percutaneous coronary intervention

This was a retrospective observational study. A total of 1414 consecutive patients diagnosed with CCS in Fuwai Hospital, Chinese National Center for Cardiovascular Diseases, were enrolled from March 2017 to December 2017. The exclusion criteria were as follows: (1) age less than 18 years; (2) non-significant stenosis (< 50%); (3) no DES implantation; (4) history of coronary artery bypass grafting (CABG); (5) lack of follow-up coronary angiography (CAG)/coronary CT angiography (CCTA); (6) lack of baseline measurements for TyG index; (7) suspected familial hypertriglyceridemia (plasma triglycerides ≥ 500 mg/dL); (8) suspected ISR on CCTA but the absence of CAG confirmation at follow-up; (9) obvious ischemia-driven symptoms but refused coronary evaluation at follow-up; (10) obvious progression of CAD on CAG/CCTA but refused coronary intervention at follow-up. The detailed screening process is shown in Fig. 1. This study was performed in line with the Declaration of Helsinki and was authorized by the Ethics Committee of Fuwai Hospital. All participants provided written/oral informed consent.

Clinical data, including demographics, medical history, and laboratory tests, were collected from medical records. Laboratory tests consisting of fasting blood glucose (FBG), triglyceride (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), plasma creatinine (Pcr), high-sensitive C-reactive protein (hs-CRP) and N-terminal pro-B-type natriuretic peptide (NT-proBNP) were performed under standardized instructions and assaying system (LABOSPECT 008, Hitachi, Tokyo, Japan). Glycated hemoglobin (HbA1c) was determined by high-performance liquid chromatography (G8, TOSOH, Tokyo, Japan). All blood samples were collected after overnight fasting before CAG. Therefore, all indicators were at the same temporal window for each participant. The left ventricular ejection fraction (LVEF) was measured by echocardiography using the biplane Simpson method. The angiographic data was obtained from the cardiac catheterization laboratory records.

The definition of CCS complied with the current guideline of the European Society of Cardiology [2]. Body mass index (BMI) was calculated as weight (kg)/height’s square (m²). Diabetes mellitus was defined as a past diagnosis with diabetes, FBG ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, or use of antidiabetic drugs. Hypertension was defined as a history of hypertension with anti-hypertensive drugs or repeated blood pressure ≥ 140/90 mmHg. Dyslipidemia was diagnosed as LDL-C ≥ 3.4 mmol/L, HDL-C < 1.0 mmol/L, TG ≥ 1.7 mmol/L, or use of lipid-lowering treatment. As previously reported, the TyG index was calculated as Ln (fasting triglycerides (mg/dL) × fasting plasma glucose (mg/dL)/2) [11]. The estimated glomerular filtration rate (eGFR) was estimated according to the modified Modification of Diet in Renal Disease equations that are based on the Chinese population: 186 × Pcr^−1.154 × age^−0.203 × 0.742 (if female) × 1.233 (if Chinese) [16].

The primary endpoint was a composite of PCI complications, including repeat revascularization and ISR. The secondary endpoints were repeat revascularization and ISR. Strategies for repeat revascularization at follow-up comprise DES, bare metal stenting, plain balloon angioplasty, drug-coated balloon angioplasty, and CABG. The repeat revascularization was determined by experienced interventional cardiologists regarding individual risk and patients’ decisions. To avoid counting endpoints in patients with staged PCI, we excluded the records of staged PCI and collected the endpoints under the judgment of experienced cardiologists. The ISR was defined as the presence of significant diameter stenosis (≥ 50%) at the segment inside the stent or involving its 5-mm edges.

All patients underwent follow-up CAG or CCTA in Fuwai Hospital after the baseline successful PCI. The follow-up period lasted until October 2022. Considering the high specificity of CCTA, we mainly included patients diagnosed with the absence of ISR by CCTA. For the patients with suspected ISR, a CAG confirmation was required. Importantly, the follow-up CAG and CCTA were interpreted by a panel of dependent experienced radiologists and cardiologists. The coronary revascularization and periprocedural management were performed according to current guidelines and regulations of our center. All patients have received standard secondary prevention recommended by current guidelines [2].

Continuous variables were described as mean ± standard deviation if consistent with a normal distribution, otherwise as median and interquartile range (IQR). Categorical variables were presented as numbers and percentages (%). All participants were stratified into three groups: T₁ (TyG index < 8.60), T₂ (8.60 ≤ TyG index < 9.07), and T₃ (TyG index ≥ 9.07) in accordance with the TyG index tertiles. The Analysis of variance was applied to analyze the difference in continuous variables between groups, while the Kruskal–Wallis test was used to analyze the difference in categorical variables.

Multiple imputation was used in the imputation of the missing data by Gibbs sampling [17, 18]. The target variables containing missing data were predicted based on the whole dataset. The process was iterated until all the missing values converged, and five imputed datasets with chained equations were created using R multivariate imputation by chained equation software package [19].

Log-rank tests and Kaplan–Meier methods were performed to explore differences in event rates between TyG tertiles and plot time-to-event curves. P values for pairwise comparisons were corrected for multiple testing by the Benjamin-Hochberg algorithm. The Cox proportional hazards regression analysis was used to estimate the hazard ratio (HR) and the 95% confidence interval (CI) of the TyG index in developing the primary and secondary endpoints. In the multivariable model, the following confounders were chosen considering the clinical importance: age (continuous), sex, BMI, previous PCI, presence of peripheral artery disease (PAD), presence of multivessel CAD, eGFR (continuous), hs-CRP (continuous), presence of lesion’s length ≥ 20 mm, stent length (continuous). Trend analyses were conducted by entering the tertiles of the TyG index as a continuous variable and rerunning the corresponding regression models. Moreover, the RCS were used to examine the possible nonlinear relationships of the TyG index with the primary and secondary endpoints. To balance best fit and overfitting in the RCS, the Akaike information criterion was used [20]. The potential non-linearity of the RCS was tested through a likelihood ratio test that compares the model presenting only a linear term with the model presenting linear and cubic spline terms [18, 21, 22]. The median of the TyG index was assigned as the reference value.

All statistical analyses were performed in R software (version 4.1.2). A two‐sided P value < 0.05 was considered statistically significant.

The average age of the 1414 CCS participants undergoing DES-PCI was 58.04 ± 0.25 years, and 1103 (78.00%) were men. Compared with the T₁ group, T₂ and T₃ group participants tended to have higher BMI (P < 0.001). The proportions of cardiometabolic risk factors, including diabetes, hypertension, family history of CAD, and the use of insulin and oral hypoglycemic drugs, were higher in the T₃ group than in the other two groups (all P < 0.05). Moreover, the participants of the T₃ group had higher TC, LDL-C, triglycerides, FPG, HbA1c, and Pcr (all P < 0.001) (Table 1).

Table 2 presents the baseline characteristics according to the primary endpoint. A total of 548 participants were observed experiencing repeat revascularization and/or ISR during the follow-up. Compared with participants without endpoints, participants with an endpoint event had higher BMI and greater proportions of diabetes, previous MI, previous PCI, multi-vessel CAD, and use of insulin and oral hypoglycemic drugs (all P < 0.05). Besides. participants with an endpoint event showed elevated concentrations of LDL-C, FPG, HbA1c, and statistically significant different LVEF (all P < 0.05). Furthermore, participants experiencing a primary endpoint had higher levels of TyG, and the participants with an endpoint event tended to be in the T₃ group (both P < 0.05). Compared to those without an endpoint event, patients with the primary endpoint had higher proportions of restenotic lesions, chronic total occlusions, lesions > 20 mm long, number of stents, and smaller diameter of stent (all P < 0.05).

The respective baseline information according to the secondary endpoints is shown in Additional file 1: Tables S1–S2. More details about the angiographic findings and medications at charge among the TyG index tertiles are shown in Additional file 1: Table S3.

The median follow-up duration of the patients for the primary endpoint was 60 months (IQR 56–64 months). During the follow-up period, 548 (38.76%) patients had the primary endpoint. Results of the Kaplan–Meier survival analyses are presented in Fig. 2. It’s indicated that the incidence of the primary endpoint in T₃ was significantly higher than that in the T₁ (log-rank test, overall P = 0.0028, adjusted pairwise P values between T₁ and T₃ = 0.0021), which was driven by the increase in repeat revascularization (log-rank test, overall P = 0.0047, adjusted pairwise P values between T₁ and T₃ = 0.0058) as well as ISR (log-rank test, overall P < 0.001, adjusted pairwise P values between T₁ and T₃ = 0.0007). Table 3 describes the results of the multivariable Cox proportional hazards regression analysis. The unadjusted model 1 indicated that the TyG index was statistically significantly associated with the primary endpoint, and the T₃ was at elevated risk for an endpoint event. After adjusting for age, sex, BMI, previous PCI, presence of PAD, presence of multivessel CAD, hs-CRP, eGFR, and presence of lesion’s length ≥ 20 mm in model 2, the TyG index as a continuous variable was an independent predictor for the primary endpoint (HR, 1.201; 95% CI 1.047–1.377; P = 0.009). Taking the T₁ as a reference, the risk of the primary endpoint was 1.338-fold higher (HR, 1.338; 95% CI 1.078–1.659; P = 0.008) in the T₃. After further adjusting for stent length in model 3, the TyG index as a continuous variable remained independently associated with the primary endpoint (HR, 1.191; 95% CI 1.038–1.367; P = 0.013). Using T₁ as the reference, the risk of participants experiencing the primary endpoint increased by 31.9% (HR, 1.319; 95% CI 1.063–1.637; P = 0.012) in the T₃. The trend analyses for the three models were all statistically significant (all P for trend < 0.05).

The association of the TyG index with the primary endpoint in model 3 was visualized in Fig. 3a using RCS. A linear and dose–response relationship between the TyG index and the primary endpoint was observed (non-linear P = 0.373, P overall = 0.035).

A sum of 499 participants experienced repeat revascularization. The TyG index, as a continuous variable, was found to be significantly associated with repeat revascularization in model 1 and independently predicted repeat revascularization in model 2 and model 3 (all P < 0.05). Using the T₁ as the reference, each additional unit increase in the TyG index was found to increase the risk for repeat revascularization by 29.2% (HR, 1.292; 95% CI 1.030–1.622; P = 0.027) and 27.4% (HR, 1.274; 95% CI 1.015–1.600; P = 0.037) in model 2 and model 3, respectively. Besides, there was statistical significance in the trend analyses (all P for trend < 0.05). The relation of the TyG index with repeat revascularization in model 3 was fitted by RCS (Fig. 3b). Similarly, the TyG index was related to the risk of repeat revascularization in a linear and dose–response relationship (non-linear P = 0.251, P overall = 0.048).

There were 230 participants having ISR. The overall association between the TyG index and the risk of ISR was statistically significant (P overall = 0.002). After adjusting for potential confounders in model 3, the result of the RCS revealed a linear association between the TyG index and ISR (non-linear P = 0.463) (Fig. 3c). Moreover, the TyG index at 8.596 was indicated to be the reference value for the dose–response relationship that participants with a TyG index higher than 8.596 were associated with higher risk for ISR (HR, 1.738; 95% CI 1.250–2.417; P = 0.001). Detailed results of Cox regression analyses for the secondary endpoints are shown in Additional file 1: Tables S4, S5.