Introduction
Acute myocardial infarction (AMI) is the cause of a major disease burden worldwide, and many known challenges exist in the management of AMI [
1‐
2]. A significant portion of the patients who present to the emergency department exhibit symptoms that may be suggestive of AMI, making this a major condition encountered in daily clinical practice. Currently, percutaneous coronary intervention (PCI) is the standard of care employed to relieve myocardial ischemia resulting from the target stenosis, but accumulating evidence indicates that as many as 20% of AMI patients have a worse prognosis after successful PCI [
3‐
6]. Therefore, the identification of potentially modifiable and residual risk factors would be helpful for AMI patients at high risk for a poor prognosis after stent implantation. Moreover, precise and early identification of patients at low risk for AMI based on such risk factors would be of profound clinical importance for fast initiation of individual comprehensive therapy and risk factor management [
7].
Recent research has established that insulin resistance (IR) can be quantified using the triglyceride-glucose (TyG) index, an alternative biomarker that is both accessible and reliable [
8]. Additionally, substantial statistical evidence indicates that the TyG index correlates with the progression of acute coronary syndromes (ACS) [
8‐
12]. Furthermore, the TyG index has been shown to be a valid and effective prognostic factor in patients with ACS, with research showing that the TyG index can be used for risk stratification of ACS patients [
8‐
12]. The molecular mechanisms underlying the predictive role of the TyG in coronary atherosclerosis have also been studied [
8].
The pathology of IR is characterized by a complex interplay among factors such as metabolic disturbance, endothelial dysfunction, exaggerated inflammatory response, and smooth muscle cell dysfunction. These factors collectively contribute to the dysregulation of glucose metabolism and may increase the risk of various adverse outcomes [
8]. Moreover, clinical data from direct comparisons indicate that the TyG index is more accurate than the homeostasis model assessment estimated insulin resistance (HOMA-IR) index and, thus, could provide incremental prognostic information for clinical outcomes in patients with coronary artery disease (CAD) [
8,
13−
14]. In addition, because the TyG index incorporates two metabolic indicators, fasting blood-glucose (FBG) and fasting triglyceride (TG), that can be readily measured, this biomarker offers the benefits of a low cost-to-benefit ratio, common availability in clinical practice, and being easily generalized for practical utilization.
Given that AMI is a complex disorder involving multiple pathophysiological mechanisms, binary diagnosis of coronary artery disease does not reflect the complexity of AMI or cardiovascular risk stratification [
7]. Hence, a quantitative model for determining the risk of ACS based on multi-modal data is needed. However, the effectiveness of integrating the TyG index with electronic health records (EHR) data in a prediction model that can provide individualized risk stratification for ACS patients has not been determined. Therefore, we investigated the long-term prognostic value of the TyG index combined with EHR data among patients with new-onset ST-elevation myocardial infarction (STEMI) after PCI.
Discussion
Our prediction nomogram for the probability of MACEs in patients with new-onset STEMI at 2, 3 and 5 years after emergency PCI combines a biomarker of IR, the TyG index, with several traditional clinical risk factors (age, diabetes mellitus, and current smoking). This nomogram constructed in a development cohort was validated to be effective in an independent validation cohort according to AUC values, Harrell’s C-indexes, calibration curves and DCA. The good predictive performance of our prediction nomogram suggests that the variables included provide new insight into the risk stratification of STEMI, which could support the development of novel strategy.
Atherosclerosis is a dynamic process that evolves throughout the course of ACS and is promoted by multiple factors related to inflammation and the immune system, imbalance of the autonomic nervous system and metabolic disturbance [
19‐
24]. Therefore, the occurrence, development and clinical outcome of ACS are influenced by the interactions of multiple systemic factors rather than any single factor. As noted elsewhere, Zheng et al. [
25] constructed a simple-to-use nomogram, integrating three clinical features that are both straightforward to obtain and routinely collected in clinical practice, evaluating cardiac mortality in non-ST-segment elevation ACS patients after PCI. Consistently, emerging evidence demonstrates that a prediction nomogram considering five conventional clinical features that are easy to obtain in daily clinical work in cardiovascular risk assessment can provide clinicians with a non-invasive and simple method for assessing MACE risk in patients with ACS who are scheduled for emergency coronary angiography [
26]. Undeniably, single-modality data alone cannot systematically and comprehensively depict the degree of coronary atherosclerosis in patients, especially for new-onset STEMI patients who have undergone emergency complete and successful revascularization. Therefore, approaches to understanding and managing ACS have evolved from focusing on “vulnerable plaques” to “vulnerable patients” who are suffering from an atheromatous-plaque burden, metabolic disturbance, as well as inflammatory and oxidative stress responses leading to plaque disruption to build a platform for better risk stratification for ACS patients [
22]. Systematic ACS risk assessment is a process for identifying individuals who are at an increased risk of an adverse outcome, and such assessment could inform clinical decision-making by identifying high-risk patients who may require extra care and resources [
22,
26‐
28]. Notably, residual cardiovascular risk is a persisting clinical problem that limits the benefit some patients can receive from the long-term success of PCI [
26‐
27,
29−
30]. Therefore, in the present study, a nomogram was developed for risk prediction in patients with STEMI treated with PCI that incorporates routine clinical data collected from EHR and a biomarker of metabolism, the TyG index.
The TyG index, which incorporates two major metabolism biomarker subtypes, is considered to be a representative biomarker of IR [
8,
31]. Consistent clinical data have demonstrated that the TyG index offers prognostic efficacy in various cardiovascular and cerebrovascular diseases, including ACS, heart failure, atrial fibrillation and stroke [
8,
32‐
33]. Emerging evidence demonstrates that a longitudinal correlation exists between the baseline TyG index and cardiovascular disease [
34]. Another multi-center study found that a higher TyG index may identify patients with a high risk of multi-vessel CAD [
35]. Additionally, the TyG index was shown to serve as a risk management strategy for CAD patients with different glucose metabolic states that can help determine the severity of CAD and facilitate healthcare decision-making [
35]. A retrospective cohort study confirmed that an increased TyG index was independently related to a poor outcome in ACS patients with diabetes [
36], and based on clinical data from 1282 diabetes patients with new-onset stable CAD over a long follow-up, Jin et al. observed a close correlation between the TyG index and adverse cardiovascular outcomes [
37]. Our overall findings in the present study are in line with the previous literature indicating that the TyG index could benefit prediction strategies for long-term prognosis in ACS patients [
8,
36]. However, previous studies included patients with a PCI history or prior MI that might affect the TyG index [
36]. Therefore, in order to reduce the confounding effects of such factors in the current study, patients with previous CAD diagnoses were strictly excluded. Moreover, the benefit of incorporating the TyG with other clinical risk factors for risk stratification of patients had not been reported. Our prediction nomogram can play a valuable role in clinical work based on its ease of use in a wide variety of health care systems [
18]. The results of the present study support the incremental prognostic value for long-term prognosis of adding the TyG index into a model that includes well-established risk factors among ACS patients [
36‐
37]. Our prediction nomogram on the basis of the TyG index as a metabolism-related biomarker combined with EHR data offers a relatively straightforward clinical tool that can be obtained in little time using a simple intake form. Nevertheless, it is universally acknowledged that any prognostic model should only be accepted into clinical practice when it performs well and achieves the desired goals among researchers and physicians [
18]. External validation, preferably in multiple, disparate datasets, is the “gold standard” and frequently utilized to determine whether a model performs well and should be incorporated into the clinical setting. In the present study, the development and independent validation cohorts were sourced from different centers of patients, and our external validation demonstrated that the prediction nomogram had satisfactory consistency for the prediction of MACEs. However, the prevalence of a strongly weighted prognostic risk factor, hypertension, was lower in the.
development cohort compared than in the external validation cohort. Nonetheless, our model performed well, displaying good discrimination as reported by C-index values of 0.772 for the development set and 0.736 for the validation set. Additionally, our model showed good performance based on calibration curves. The DCA also provided clinical evidence that the clinical accuracy of the prediction nomogram in both the development and external validation cohorts. Still, replication and further validation of our findings in other well-defined populations is necessary to ensure the robustness of our model.
Despite the close association between the TyG index and adverse cardiovascular outcomes, the underlying mechanism remains unclear. Pathologically, IR contributes to the interaction between an imbalance in glucose metabolism and the inflammatory response to oxidative stress [
38]. IR is known to disrupt lipid homeostasis, which can induce endothelial injury and result in the initiation of atherosclerosis [
38‐
39]. Additionally, IR is thought to be related to increased production of pro-inflammatory cytokines, which can lead to further myocardial injury in established ischemic myocardium, and in turn, increased myocardial oxygen consumption could eventually lead to a reduction in the compensatory capacity in non-infarcted muscles [
8,
39]. Therefore, the TyG index could be considered an indicator of residual cardiovascular risk that is further aggravated in STEMI patients with post-PCI progression. Accordingly, the developed prediction nomogram incorporating the TyG index with well-established risk factors may offer increased discriminatory ability and accuracy for STEMI patients after PCI who are at early risk for long-term poor prognosis, and the timely warning provided by this nomogram could in turn inform clinical decision-making and help healthcare providers to allocate resources effectively.
Study limitations
First, the present study had a small sample size, and thus, it is essential that a large prospective cohort study be conducted to confirm the prediction nomogram’s predictive performance. Second, some heterogeneity in the clinical and demographic data derived from the two centers and incomplete clinical data for some patients could have resulted in some confounding factors being missed that may have affected the conclusions of the study. Third, due to the different methods used by the two centers to measure myocardial injury markers (e.g., troponin), this study did not consider these variables. Fourth, the prediction results from the nomogram were assumed to remain constant over time. However, in reality, disease outcomes may vary due to improved treatment, early detection and changes in its natural history. Therefore, the performance of the nomogram may become less accurate over time.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.