To ensure the comprehensiveness of our study, we performed a thorough search in the following databases: PubMed, Embase, Cochrane Library, and Web of Science. The search encompassed the entire period from the establishment of these databases until October 1, 2023. Our search involved a mix of subject terms and free-text terms. The terms we used included “Coronary intervention”, “Percutaneous Coronary Intervention”, “PCI”, “coronary intervention angioplasty”, “Triglycerides”, “Triacylglycerol”, “triacyl glyceride”, and “TyG”. The specific methods for conducting the search are provided in Supplementary Material S1. Furthermore, we conducted a manual search of the references cited in published systematic reviews to ensure that we included all relevant literature.

The inclusion criteria for this study were as follows: (1) Participants: The study included patients who underwent percutaneous coronary intervention (PCI) for the treatment of coronary artery disease. (2) The exposure variable of interest was the TyG index, which was calculated using the formula Ln [fasting triglyceride (mg/dl) × fasting plasma glucose (mg/dl)/2]. (3) Study design: Only observational studies were considered for inclusion in this analysis. (4) The primary outcome measure was major adverse cardiovascular events (MACEs), which included all-cause death, cardiac death, myocardial infarction, revascularization, stroke, and heart failure. Secondary outcomes of interest included the occurrence of new-onset atrial fibrillation, in-stent restenosis (ISR), and reduction of quantitative flow ratio. (5) The included studies reported relative risks (RR), odds ratios (OR), and hazard ratios (HR) along with their corresponding 95% confidence intervals (CIs). Alternatively, studies provided raw data that could be used for the calculation of these effect measures.

The following studies were excluded: (1) The included sources will consist of reviews, case reports, research plans, and conference papers. (2) Clinical trials, animal or in vitro studies will be considered for inclusion. (3) Examples of publications that will be excluded are duplicate publications or studies with unavailable full texts. (4) Any literature that lacks information on outcome indicators will not be included in the analysis.

The literature screening process was carried out by two independent reviewers, HM TAO and L TAO. In case there were any differing opinions during the screening, the reviewers engaged in a constructive exchange of ideas to reach a consensus. Alternatively, if needed, a third reviewer named Wang was involved to provide a final decision.

HM TAO and L TAO, two independent reviewers, extracted information from the included literature. They focused on various details, including the first author, publication year, country, basic characteristics of the study population, exposure levels, follow-up duration, and outcome indicators. Their analysis covered a wide range of information.

The two independent reviewers, Wang and Kong, conducted the quality assessment of the included studies using the Newcastle-Ottawa Scale (NOS) (30). The assessment criteria for retrospective studies consist of eight aspects. These include the adequacy of case definition, representativeness of cases, selection of controls, definition of controls, comparability of cases and controls based on design or analysis, determination of the extent of exposure, the shared method for determining cases and controls, and comparison of non-response rates in cases and controls. For prospective studies, the evaluation encompasses several aspects as well. These include the representativeness of the exposure cohort, selection of the non-exposure cohort, confirmation of exposure, absence of the outcome of interest at the study's outset, comparability of cohorts through design or analysis, evaluation of outcomes, determination of whether the follow-up duration is sufficient to obtain outcomes, and adequacy of follow-up. Each aspect is assigned a score ranging from 0 to 1, except for comparability, which can receive up to 2 points. Therefore, a study is considered of high quality if it achieves a total score of 6 points or more out of 9.

We performed a meta-analysis using Stata 17.0 to examine a collection of diverse studies. To address inconsistencies in outcome measures, we used a conversion formula to transform odds ratios (OR) into risk ratios (RR), and aggregated RR and hazard ratios (HR) as effect sizes. To assess heterogeneity among studies, we employed the Q test and calculated the I² statistic. For studies with no significant heterogeneity (I² < 50% and P > 0.1), we applied the fixed-effects model (Mantel-Haenszel method) for the meta-analysis. Conversely, for studies exhibiting significant heterogeneity, we utilized the random-effects model (Der Simonian-Laird method). We further explored the extent and sources of heterogeneity through subgroup analyses and regression analyses, considering factors such as diabetes background and follow-up duration. Additionally, we conducted a sensitivity analysis to assess the robustness of the meta-analysis results. To evaluate publication bias in the included literature, we created a funnel plot and employed either Egger's or Begg's test for statistical assessment, making sure to have a minimum of 5 studies. If significant publication bias was found, we used the trim-and-fill method to estimate the potential impact on the findings.

After eliminating duplicate and irrelevant studies, a total of 2101 papers were obtained from the initial database search. The titles and abstracts of 612 articles were then examined. Ultimately, 17 studies (11–27) were included in this study. The search process for relevant studies is illustrated in Figure 1.

The studies included in the analysis involved a total of 22,467 participants, with 17,376 being males and 5,100 being females. The average age of the participants in these studies ranged from 58 to 66.6 years. The sample sizes of the studies varied, ranging from 214 cases to 5,489 cases. Ten studies explicitly mentioned that the patients included in their research underwent percutaneous coronary intervention (PCI) for acute coronary syndrome (ACS). Four studies focused on ISR, eleven studies examined the occurrence of post-PCI MACE, one study investigated new-onset atrial fibrillation, and one study explored the reduction in quantitative flow ratio. The follow-up duration ranged from 6 months to 5 years, with varying durations in different studies. Table 1 provides detailed information on the key characteristics of the studies included in this analysis.

Based on the NOS scale, the quality assessment reveals that 16 of the 17 studies included in the analysis obtained a score of 6 or above. For more detailed information can be found in Supplementary Table S2.

To assess the influence of each study on the aggregated results, we conducted a sensitivity analysis using a leave-one-out method. Our primary outcome, MACE, was the focus of this analysis. Interestingly, none of the individual studies exerted a significant influence on the overall findings. This finding further supports the reliability of the overall results obtained from this meta-analysis.

To ensure the validity of the results obtained from the meta-analysis, we employed a variety of techniques. These included the utilization of a funnel plot, as well as the implementation of Egger's and Begg's tests, to identify any possible publication bias pertaining to Major Adverse Cardiovascular Events (MACE). The results obtained from these analyses revealed no significant evidence of publication bias for MACE.

Our research has demonstrated the significant value of the TyG index in predicting the risk of major adverse cardiovascular events (MACE) in patients who have undergone percutaneous coronary intervention (PCI). This finding is consistent with previous studies that have investigated the relationship between the TyG index and coronary artery disease (CAD), including its association with subclinical atherosclerosis, hypertension, coronary artery calcification, and arterial stiffness (8, 20, 31, 32). While PCI involves procedures such as stent implantation or balloon dilation, which improve blood flow and clinical outcomes, it is essential to also consider the occurrence of post-PCI cardiovascular adverse events. The heightened risk of MACE observed in patients with elevated TyG index levels could be attributed to insulin resistance (IR) (33, 34). Insulin resistance and disturbances in glucose metabolism contribute to oxidative stress, inflammation, and dysfunctional immune regulation, thereby exacerbating the development of arteriosclerosis and promoting plaque formation. Inflammation, triggered by insulin resistance and subsequent disturbances in glucose metabolism, may also play a role in the occurrence of in-stent restenosis (ISR), a common complication following stent implantation (35, 36). Additionally, the TyG index has demonstrated varying predictive abilities for different cardiovascular adverse events (19). This variation may provide valuable insights into the underlying mechanisms that contribute to its predictive ability.

Previous studies have demonstrated that the TyG index was associated with the prognosis of patients diagnosed with ACS or chronic coronary syndrome (CCS) (10). However, in response to an acute syndrome, patients with ACS may experience abnormal blood glucose levels due to stress-induced hyperglycemia (SIH) (37). SIH could result in the misbalance of oxidative equilibrium, trigger inflammation, and cause endothelial dysfunction, which can further worsen the prognosis (38). Additionally, the fluctuations in fasting blood glucose (FBG) levels can affect the calculation of the TyG index. Although researchers have recognized SIH as a potential confounding factor that may impact the predictive ability of the TyG index, there is currently a lack of relevant studies on this topic (39). In this study, we found a significant association between the TyG index and the risk of MACE in patients diagnosed with ACS. Further research is needed to investigate the predictive ability of the TyG index for eliminating factors with collinearity, such as SIH. Furthermore, subgroup analysis revealed that the predictive ability of the TyG index was comparable between patients with and without diabetes. This could be because both groups, diabetic and non-diabetic, share certain risk factors for insulin resistance (18).

Previous research has highlighted a limitation in using the TyG index at a specific time to accurately predict disease progression. This limitation arises from the variability of triglycerides and glucose levels (40). For instance, the fluctuation of glucose levels caused by stress-induced hyperglycemia (SIH) can impact the TyG index and introduce a confounding factor when assessing a patient's metabolic status (39, 41). Currently, some studies suggest the importance of dynamic monitoring of the TyG index or the use of cumulative TyG index. These approaches have been found to have greater clinical predictive value in the prognosis of coronary artery disease (20, 42).

Furthermore, subgroup analysis based on the duration of follow-up was conducted for the consideration of various factors such as endothelial dysfunction, smooth muscle cell proliferation, and the infiltration of inflammatory cells, all of which result from prolonged insulin resistance. Results have revealed that the predictive ability of TyG index was comparable between patients with varying durations of follow-up. Most studies had similar duration of follow-up, one research conducted by Luo et al., have reported the occurrence of MACE at different time points after PCI, but they did not analyze the change in the predictive ability of the TyG index based on different follow-up durations (16). Considering the long-term damage on cardiac function caused by insulin resistance and insufficient insulin signaling, it is essential for future research to explore the utility of dynamic monitoring of the TyG index or cumulative TyG index in patients who have undergone PCI. Additionally, it is necessary to clarify the time-dependence variation curve of the predictive value of the TyG index.

The predictive accuracy of the TyG index for CAD has been demonstrated to be higher compared to other indicators such as triglyceride (TG), atherogenic index of plasma (AIP), and the lipoprotein combine index (LCI) (6). Moreover, integrating the TyG index into the GRACE risk score has been found to enhance the prognostic predictive capability for patients who underwent PCI (18, 43) according to previous research. While additional studies are required, our research has contributed to the understanding of the predictive value of the TyG index and its potential utility in managing the prognosis of patients who underwent PCI.

Our meta-analysis conducted a comprehensive analysis of the studies regarding the correlation between the TyG index and the prognosis of post-PCI patients. The findings of this study provide a solid evidence-based foundation in the field of medicine, which holds great potential in enhancing the postoperative conditions of PCI patients. However, it is important to acknowledge the existence of certain limitations in this study. Firstly, the HR and RR values were extracted from various multivariate analyses to account for the confounding factors present in each study. However, it should be noted that the confounding factors were not entirely consistent across all included studies. Secondly, most of the studies included in this meta-analysis predominantly involved Asian individuals. It is plausible that the pathogenesis of those undergoing PCI in different regions may exhibit variations, thus impacting the generalizability of our results. Nonetheless, our findings suggest a clinical value of the TyG index in predicting the prognosis of post-PCI patients.