The Impact of the Stress Hyperglycemia Ratio on Short-term and Long-term Poor Prognosis in Patients With Acute Coronary Syndrome: Insight From a Large Cohort Study in Asia

Cardiovascular disease is the leading cause of death (1). With the standardization of drug therapy and the development of coronary interventional therapy, the mortality rate of cardiovascular disease has decreased dramatically (2,3). Nevertheless, a growing number of patients suffer from metabolic disorders due to excessive food intake and unhealthy lifestyles (4–6), which could exacerbate adverse cardiovascular outcomes, especially coronary heart disease (7–11).

A number of previous studies have shown that acute hyperglycemia at admission is independently associated with a poor early and late prognosis in acute coronary syndrome (ACS) patients, especially those diagnosed with acute myocardial infection (AMI) (12–15). However, the admission blood glucose (ABG) level cannot entirely reflect the acute hyperglycemic state, which could also be affected by the chronic glucose level. Hence, some researchers have proposed a novel marker to reflect true acute hyperglycemic status, i.e., the stress hyperglycemia ratio (SHR) (16). Recently, many investigators have explored the short-term prognostic value of this ratio in subjects with ACS or AMI (17–21). However, it is unknown whether the SHR is independently related to a poor long-term prognosis in ACS patients. Therefore, we performed this study to evaluate the correlation between the SHR and adverse cardiovascular events in ACS patients treated with drug-eluting stent (DES) implantation.

This investigation was a prospective, observational cohort study at Fuwai Hospital, National Center for Cardiovascular Diseases. The study process was in accordance with the Declaration of Helsinki and was authorized by the Fuwai Hospital Ethics Review Committee. All subjects were informed and signed an informed consent form.

From 1 January 2013 to 31 December 2013, a total of 10,112 patients who underwent DES implantation at Fuwai Hospital were consecutively screened. Patients meeting the following criteria were included: 1) age between 18 and 80 years and 2) ACS treated with DES implantation. Patients meeting the following criteria were excluded: 1) admission hemoglobin (Hb) <100 g/L, 2) ABG <3.90 mmol/L, 3) severe chronic renal insufficiency (estimated glomerular filtration rate [eGFR] <30 mL/min/1.73 m²), 4) history of erythropoietin application or recent blood transfusion, 5) history of malignant tumor, and 6) missing crucial laboratory data (glucose on admission or glycated hemoglobin [HbA_1c]) or incomplete 1-month or 2-year follow-up data. Finally, 5,562 ACS patients treated with DES implantation were successfully enrolled. The specific process of population enrollment is illustrated in Supplementary Fig. 1.

We prospectively gathered the baseline demographic and clinical data for all patients. Demographic data included age, sex, BMI, concurrent diseases, smoking status, family history of premature coronary artery disease (CAD), and previous myocardial infarction (MI) or revascularization history (percutaneous coronary intervention [PCI] or coronary artery bypass grafting [CABG]). Clinical data consisted of the main diagnosis on admission; physical, imaging, and laboratory examination findings; and drug regimen at discharge. The glycemic status on admission was assayed with a LABOSPECT 008 system (Hitachi, Tokyo, Japan), and the HbA_1c value was measured with high-performance liquid chromatography (Tosoh G8 HPLC Analyzer; Tosoh Bioscience, Tokyo, Japan).

After the coronary intervention was completed, the characteristics of the coronary disease, including the number of stenotic vessels, unusual types of coronary stenosis, the SYNergy between percutaneous coronary intervention with TAXus and cardiac surgery (SYNTAX) score (22), and data related to stent implantation, were analyzed and recorded by two coronary intervention specialists who were blinded to the baseline data (23). And when the patients were discharged, we recorded the in-hospital adverse cardiovascular events. Diabetes was recorded if the patient was previously diagnosed with diabetes, used (currently or previously) oral hypoglycemic drugs or insulin, or had HbA_1c >6.5%. Stroke was defined as a previous history of cerebral bleeding, ischemic stroke, or transient ischemic attack. Moreover, we categorized the diagnosis on admission as unstable angina (UA), non-ST-segment elevation MI, or ST-segment elevation MI (STEMI) according to the latest relevant guidelines (24–26). Estimated average chronic glycemic level was calculated with the formula [(28.7 × HbA_1c %) − 46.7] (27), and SHR was defined as glucose on admission divided by the estimated average chronic glycemic value (16). Left main (LM) disease was determined by >50% stenosis in the LM coronary artery. Three-vessel disease was defined as >50% stenosis in three vessels located in different epicardial vascular systems (e.g., the left anterior descending branch, left circumflex branch, and right coronary artery). The type of implanted stents included homogeneous first-generation DES (paclitaxel-eluting stent or sirolimus-eluting stent), homogeneous second-generation DES (zotarolimus-eluting stent or everolimus-eluting stent), and hybrid first-generation and second-generation DES.

The patients were followed up at 1, 6, 12, and 24 months after discharge through telephone, correspondence, and outpatient visits. We defined the primary end point as major adverse cardiovascular and cerebrovascular events (MACCE) at the 2-year follow-up, including all-cause death, nonfatal MI, nonfatal stroke, and target vessel revascularization (TVR). The secondary end points were defined as major adverse cardiovascular events (MACE) (a composite of cardiac death, nonfatal MI, and TVR) at 2-year follow-up, cardiac death, and nonfatal MI at two-year follow-up, and in-hospital cardiac death and nonfatal MI. All-cause death was defined as death from any cause, including cardiac death and noncardiac death. Cardiac death was defined as death from any heart disease. MI and stroke were determined by the criteria mentioned above. We defined TVR as ischemic-driven or clinic-driven revascularization of the target vessel. The participants underwent follow-up for at least 24 months regardless of the occurrence of the primary end point unless they died, were lost to follow-up, or dropped out of the study.

SPSS Statistics (version 26; SPSS, Chicago, IL), R (version 4.1.0), and Stata/MP (version 14; StataCorp) were used to perform the statistical analyses and create the tables and figures. We use means ± SD to describe continuous variables with a normal distribution and median (interquartile range) to describe other measurement data without a normal distribution. Categorical variables are expressed by quantities and percentages. We applied ANOVA to explore differences in measurement data among the five groups. If continuous variables did not have a normal distribution or homogeneity of variance, the rank sum test was used to evaluate differences among the five groups. We used the χ² test to assess categorical variables. Moreover, the log-rank test and Kaplan-Meier (K-M) survival analyses were performed to explore differences in event-free survival among the different groups. In addition, multivariable Cox regression and logistic regression were applied to identify independent risk factors of adverse cardiovascular outcomes and the corresponding hazard ratios (HRs) and odds ratios (ORs) after adjustment for confounding factors including age, sex, BMI, diagnosis on admission, family history, previous MI, previous PCI, previous CABG, hypertension (HTN), hyperlipidemia (HL), diabetes, previous stroke, smoking status, LM disease, three-vessel disease, chronic total occlusion (CTO) disease, in-stent restenosis (ISR) disease, SYNTAX score, and type, number, length, and diameter of stents, intra-aortic balloon pump (IABP) application, diastolic blood pressure (DBP), left ventricular ejection fraction (LVEF), low T3 syndrome, eGFR, triglyceride (TG), total cholesterol (TC), LDL cholesterol (LDL-C), HDL cholesterol (HDL-C), Hb, uric acid, hs-CRP, and oral drugs (dual antiplatelet therapy [DAPT], statin, β-blocker, ACE inhibitor [ACEI]/angiotension II receptor blocker [ARB], and calcium-channel blocker [CCB]) after discharge according to clinical experience. Moreover, an additional confounding factor (glucose-lowering therapy) was adjusted in the population with diabetes. In addition, restricted cubic spline (RCS) analyses with five knots, at the 5th, 27.5th, 50th, 72.5th, and 95th centiles, were performed to explore the characteristics of the correlation between SHR and the end points. In the RCS model, we also adjusted for other confounding factors as mentioned above. Moreover, considering the associations between SHR and adverse events were approximately linear below and above the SHR value corresponding to an HR equal to 1 (it was 0.78 in the overall population), we calculated HRs and ORs per SD increase in SHR using a linear model. Subgroup analyses were also performed to assess the influence of SHR on cardiovascular prognosis in different subgroups stratified by diabetes status and primary diagnosis. Finally, we performed sensitivity analyses to test the robustness of our findings. Cox logistic least absolute shrinkage and selection operator (LASSO) regression and logistic LASSO regression were used to select variables from the variable pool including HbA_1c and ABG to avoid influence from collinearity (28,29). A two-sided P < 0.05 was defined as the statistical threshold for all analyses.

A total of 5,562 ACS patients were enrolled in our study. Mean ± SD age was 58.01 ± 10.10 years, 4,265 (76.7%) patients were men, obese patients (BMI ≥28 kg/m²) accounted for 23.9%, 3,873 (69.6%) patients were identified as having UA, 1,689 (30.4%) patients were diagnosed with AMI, 3,585 (64.5%) patients suffered from HTN, 3,706 (66.6%) subjects were diagnosed with HL, 2,184 (39.3%) subjects were diagnosed with diabetes, and >59% of patients were current smokers. All participants were separated into five groups based on SHR (quintile 1 [n = 1,113], SHR ≤0.70; quintile 2 (n = 1,112), 0.70 < SHR ≤ 0.75; quintile 3 (n = 1,113), 0.75 < SHR ≤ 0.81; quintile 4 (n = 1,112), 0.81 < SHR ≤ 0.90; and quintile 5 (n = 1,112), SHR >0.90). Detailed baseline data are shown in Table 1. We found statistically significant differences among the five groups in age, sex, BMI, diagnosis on admission, family history of premature CAD, diabetes, history of previous MI or PCI, CTO disease, IABP, SYNTAX score, DBP, LVEF, low T3 syndrome, eGFR, triglyceride, TC, LDL-C, HbA_1c, ABG, uric acid, hs-CRP, Hb, ACEI/ARB use, CCB use, and glucose-lowering therapy at discharge. Further analysis showed the lowest BMI in quintile 1 among the five groups and a higher prevalence of low T3 syndrome in quintile 1 than in quintile 3. The percentage of patients diagnosed with AMI in quintile 5 was significantly higher than that in any other group. Moreover, the HbA_1c level and the percentage of patients with diabetes in quintiles 1 and 5 were notably higher than those in quintile 3. In addition, the comparison between the final population and population excluded for missing HbA_1c or ABG measurements is illustrated in Supplementary Table 1.

A total of 5,562 patients completed the 1-month and 2-year follow-up examinations; the median follow-up time was 28.32 months (interquartile range 25.82, 30.65). A total of 643 MACCE were recorded: there were 72 cases of all-cause death, including 37 cases of cardiac death, and 259 cases of nonfatal MI, 97 cases of stroke, and 264 cases of TVR (Supplementary Table 2). K-M survival analyses showed a significant difference in the incidence of MACCE, MACE, and cardiac death and MI among the five groups at the 2-year follow-up, with the lowest MACCE rate in quintile 3 (all P values <0.001). Details of the K-M survival analyses are presented in Fig. 1. Then, results of the RCS analyses indicated that there were U-shaped associations of the SHR with the MACCE rate and MACE rate at 2-year follow-ups even after adjustment for other confounding factors (all P values for nonlinearity <0.05), while there were J-shaped associations of SHR with in-hospital cardiac death and MI and that at 2-year follow-up (all P values for nonlinearity <0.05; when SHR was <0.78, HR and OR changed slowly, while they increased sharply when SHR was >0.78). The value of the SHR corresponding to the lowest risk of every end point on multivariate-adjusted RCS analyses was 0.78 for the overall population (Fig. 2). Cox regression analyses and logistic regression were performed to determine independent risk factors and calculate the HRs and ORs of the SHR for the primary end point and each secondary end point. The results showed that in comparisons with subjects with an SHR of 0.76–0.81 (quintile 3), the multivariable-adjusted HR for MACCE at the 2-year follow-up was 1.39 (95% CI 1.07–1.80) for subjects with an SHR <0.71 (quintile 1), 1.32 (1.01–1.71) for subjects with an SHR from 0.82 to 0.90 (quintile 4), and 1.49 (1.15–1.94) for subjects with an SHR >0.91 (quintile 5) (all P values <0.05). Moreover, when SHR was <0.78, the HR per SD increase in predicted MACCE was 0.84 (0.75–0.95), while the HR per SD increase in predicted MACCE was 1.13 (1.05–1.21) if SHR was >0.78. Detailed data of the association of SHR with the different end points are shown in Table 2 and Fig. 2.

Subgroup analyses were performed for evaluation of the association of SHR with the long-term primary end point and each secondary end point in different populations according to principal diagnosis (UA or AMI) and diabetes status (diabetes or no diabetes). The results indicated J-shaped associations of SHR with in-hospital cardiac death and nonfatal MI and that at 2-year follow-up remained robust in different populations, while the U-shaped associations of the SHR with MACCE and MACE were blunt in the subgroup without diabetes (when SHR was below the reference value, the 95% CI of the HRs per SD increase in predicted MACCE and MACE included 1, and the P values were >0.05). Details of the RCS analysis in different populations can be found in Supplementary Figs. 3–6. To further test the robustness of the U-shaped and J-shaped association between SHR and poor long-term prognosis, we carried out sensitivity analysis, where the LASSO regression was used to selected variables for adjusting. We obtained four group variables for the corresponding four end points (Supplementary Fig. 2). The results of RCS and multivariable Cox regression referring to the variables selected from the lasso regression suggested that there were also U-shaped or J-shaped associations between SHR and long-term adverse cardiovascular events (Supplementary Table 3 and Supplementary Fig. 2).

In the current study, the association between SHR and adverse cardiovascular events in ACS patients undergoing PCI was evaluated, revealing the following two findings: 1) SHR was independently associated with short-term and long-term adverse cardiovascular outcomes in ACS patients treated with DES implantation and 2) the associations were U shaped or J shaped and the HRs for adverse cardiovascular prognosis significantly increased when SHR was >0.78.

Stress hyperglycemia has been proven to be a strong predictor of a poor prognosis in ACS patients, particularly patients with AMI (12,14,15,30–32). On the one hand, stress hyperglycemia reflects the seriousness of an emergency and poor glucose control to some extent. On the other hand, it may exacerbate acute cardiac disease in various ways, including compounding microvascular obstruction (13), attenuating endothelium-dependent vasodilation (33), impairing platelet nitric oxide responsiveness (34), and promoting other mechanisms of vascular damage mediated by hyperglycemia. However, the real stress state cannot be entirely reflected by the ABG. The chronic glucose level, which can be determined by the equation [(28.7 × HbA_1c [%]) − 46.7], should not be neglected (27). As such, Roberts et al. (16) proposed the concept of the SHR and proved that SHR was a better biological marker for critical disease than absolute hyperglycemia on admission. Some recent studies have explored the value of SHR for predicting a poor prognosis in patients with AMI, especially a poor short-term prognosis. Simsek et al. (17) carried out a retrospective study including 905 STEMI patients; the primary end point was defined as the incidence of no-reflow after primary PCI, and the findings suggested that SHR was a strong predictor of no-reflow. Marenzi et al. (18) consecutively enrolled 1,553 AMI patients, and the primary end point was defined as a combination of cardiogenic shock, acute pulmonary edema, and in-hospital mortality. They found that SHR was a better biomarker of in-hospital mortality and morbidity than absolute glucose level on admission. In two recent studies investigators assessed the value of SHR for predicting long-term adverse cardiovascular events in patients with AMI or undergoing PCI. Kojima et al. (14) recruited 6,287 STEMI subjects who were safely discharged. The end points were all-cause mortality and readmission because of heart failure, and the median follow-up time was 1,522 days. They found that the incidence of a long-term adverse prognosis in the highest SHR quartile (quartile 4) was significantly higher than that in the lower SHR quartiles (quartiles 1–3) in patients without diabetes. Yang et al. (20) retrospectively enrolled 4,362 subjects undergoing PCI from the CathOlic Medical Center percutAneous Coronary inTervention Registry (COACT) study. MACCE was defined as the primary end point, and median follow-up time was 2.5 years. The results showed that the HR for the highest SHR quartile (quartile 4) for long-term MACCE was 1.31 (95% CI 1.05–1.64) when the lower SHR quartiles (quartiles 1–3) served as a reference.

To our knowledge, the current study may be the first to show the U-shaped or J-shaped association between SHR and early and late poor prognosis in ACS patients treated with PCI. In our study we consecutively recruited 5,562 ACS patients treated with PCI in Asia, and the median follow-up period was 28.32 months. The results suggested that the incidence of long-term MACCE and MACE in the lowest SHR quintile (quintile 1) and the incidence in the highest SHR quintile (quintile 5) were significantly higher than that in the median SHR quintile (quintile 3), and the HRs for short-term and long-term cardiac death and MI increased significantly when the SHR was >0.78, even after adjustment for other confounding factors. Further RCS analysis showed that the correlation was U shaped or J shaped, and the outcomes of sensitivity analyses illustrated the robustness of the findings. The results of subgroup analyses were largely consistent with those of the overall population, except in the subgroup without diabetes. In view of all P values for interaction >0.05, the different outcomes in these two subgroups might be due to an insufficient sample size. The findings of the current study are to some extent consistent with those of previous studies. In the study of Roberts et al. (16), the correlation between SHR and rate of critical illness was approximately J shaped. In addition, the results of the studies of Kojima et al. (14) Yang et al. (20) revealed a difference in the incidence of long-term adverse outcomes between the highest SHR quartile (quartile 4) and the lower SHR quartiles (quartiles 1–3), which is in accordance with the outcomes of our study to some degree. However, they did not explore the potential differences among the lower SHR quartiles (quartiles 1–3) and might ignore the nonlinear analysis of the correlation of the SHR and a poor prognosis. Considering the negative impact of the SHR on acute cardiac diseases as mentioned above, it is not hard to understand the J-shaped association of the SHR with short-term or long-term cardiac death and MI and that SHR 0.78 is an inflection point. However, the underlying mechanisms of the U-shaped association of the SHR with MACCE or MACE in ACS patients remain uncertain and might include the following mechanisms. First and foremost, it is a natural law that stress response including stress hyperglycemia will happen to people if they are under stress, which are largely mediated by the hypothalamic-pituitary-adrenal axis and the sympathoadrenal system (35). Some studies suggested that mild-to-moderate stress hyperglycemia is a protective factor during stress, especially ischemia, by the following ways. In an animal model, stress hyperglycemia can enhance cardiac output and improve survival (36). In addition, for ischemic cells, moderate stress hyperglycemia can promote cells using glucose more effectively (37). Moreover, Malfitano et al. (38) found that stress hyperglycemia can upregulate cell survival factors such as vascular endothelial growth factor and hypoxia inducible factor-1α, reduce cell apoptosis, decrease infarction size, and improve cardiac systolic function in a Wistar rats MI model. Thus, mild-to-moderate stress hyperglycemia in the current study might play a protective role against adverse cardiovascular and cerebrovascular events when SHR is <0.78. Besides, baseline data showed that the incidence of diabetes and insulin use was higher in quartile 1 than in quartile 3. In view of the characteristics of lower ABG, higher HbA_1c level, and higher insulin use ratio, we inferred that the patients in quartile 1 might undergo more hypoglycemic episodes as a result of the improper use of insulin or hypoglycemic agents, which has been shown to increase the risk of death and cardiovascular events (39,40), although we excluded subjects with ABG <3.9 mmol/L. In our study, we found that the inflection point of SHR for poor prognosis was ∼0.78. Therefore, SHR >0.78 may truly indicate stress hyperglycemia. SHR values <0.78 indicate chronic hyperglycemia (high HbA_1c) with current good glycemic control or exceeding glycemic control (low ABG). It is therefore reasonably likely that the relation between SHR and any outcome will have a J- or U-shaped relation: the risk increases in all situations that depart from the linear correlation between ABG and HbA_1c in both directions. For outcomes more correlated with acute response, the curve will be J shaped. For outcome more correlated to chronic hyperglycemia, the curve will be U shaped and tend to be symmetric. In the future, a more large-scale, prospective, cohort study should be implemented to determine the threshold for the diagnosis of stress hyperglycemia by SHR and explore its predictive value for cardiovascular outcomes in ACS or MI patients. Finally, in combining the higher incidence of diabetes in quartile 1 and the blunt U-shaped association in the subgroup without diabetes, although the P for interaction was >0.05, the U-shaped association between SHR and long-term MACCE or MACE may to some extent be caused by diabetes status.

This is the first study focused on the role of SHR in ACS patients treated with DES implantation. Moreover, we evaluated the nonlinear correlation between SHR and early and late adverse cardiovascular events and, for the first time, propose a U-shaped or J-shaped association between SHR and a poor prognosis in patients with ACS. This research also has some shortcomings. For instance, we did not use the oral glucose tolerance test to further diagnose possible new-onset diabetes. Moreover, we could not assess cardiac troponin levels because different departments in our hospital tested for different biomarkers, including troponin I (TNI), high-sensitivity troponin I (hs-TNI), and troponin T (TNT), in 2013. In addition, we cannot entirely exclude the possibility of unmeasured or unknown confounding factors that may account for the associations observed in this study.

There were U-shaped associations of SHR with MACCE rate and MACE rate at 2-year follow-ups and J-shaped associations of SHR with in-hospital cardiac death and MI in hospital and that at 2-year follow-up in ACS patients treated with DES implantation, and the inflection point of SHR for poor prognosis was 0.78. More large-scale, multicenter research should be performed in the future to assess the predictive value of SHR in ACS patients; additionally, the underlying mechanisms of the U-shaped and J-shaped association need to be further studied.

Acknowledgments. The authors thank all the investigators and subjects who participated in this project.

Funding. This work was supported by National Key R&D Program of China grant 2020YFC2004700; National Natural Science Foundation of China grants 81825003, 91957123, 81800327, and 81900272; and Beijing Nova Program grant Z201100006820002 from Beijing Municipal Science & Technology Commission.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.Y., C.S., and Y.-D.T. participated in the study design. Y.Z., C.L., J.G., and X.M. participated in data collection. J.Y., K.Z., and W.W. performed the statistical analysis. J.Y. drafted the manuscript. All authors read and approved the final manuscript. Y.-D.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.