The present study is a retrospective analysis derived from a single-center retrospective observational cohort that originally aimed to investigate the association between the PCI target-vessel (native coronary arteries vs. bypass grafts) and adverse cardiovascular events (NCT05368597). We screened 1,435 consecutive patients with a history of prior CABG who were hospitalized for an ACS and underwent PCI between September 2008 and October 2021. The diagnosis of ACS, including unstable angina (UA), non-ST-segment elevation myocardial infarction (NSTEMI), and ST-segment elevation myocardial infarction (STEMI), was established according to contemporary clinical guidelines (28, 29). For the purpose of this study, the following patients were excluded: those who died prior to discharge, those undergoing hybrid coronary revascularization, those with left ventricular ejection fraction (LVEF) < 30%, renal dysfunction with an estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73m² or chronic dialysis, those who were currently using glucocorticoids for connective tissue disease, those who had active infections on admission, and those lacking relevant baseline data for analysis. A final cohort of 1,208 patients was included in the analysis. The study protocol was approved by the local Institutional Review Board of Beijing Anzhen Hospital, Capital Medical University (No. 2025320x). The study was conducted in accordance with the ethical principles of the Declaration of Helsinki. Given the retrospective nature of the study, the requirement for informed consent was waived.

Comprehensive baseline data were meticulously extracted from the hospital’s electronic medical records system. Collected information included:

Demographics and Vital Signs: Age, sex, body mass index (BMI), systolic blood pressure (SBP), and heart rate at admission.

Medical History: Hypertension, diabetes, heart failure, renal dysfunction, atrial fibrillation, previous myocardial infarction (MI), past PCI, previous stroke, peripheral artery disease (PAD), chronic lung disease, and family history of coronary artery disease (CAD). Heart failure was defined as signs or symptoms of heart failure, treatment for heart failure, or LVEF < 40%. Renal dysfunction was defined as eGFR < 60ml/min/1.73m², calculated by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation. Previous stroke was defined as previous cerebral infarction or hemorrhage.

Clinical Presentation: Type of ACS (UA, NSTEMI, or STEMI). The GRACE (Global Registry of Acute Coronary Events) risk score (based on the GRACE 6-month post-discharge prediction model) was calculated for each patient to standardize the assessment of baseline ischemic risk.

Laboratory Parameters: All laboratory measurements were performed at the central laboratory of Beijing Anzhen Hospital on the first fasting blood samples drawn after admission, following a mandated fasting period of at least 8 hours. These included serum creatinine (SCr), lipid profile (total cholesterol [TC], low density lipoprotein cholesterol [LDL-C], high density lipoprotein cholesterol [HDL-C], triglycerides), high-sensitivity C-reactive protein (hs-CRP), fasting blood glucose (FBG), and HbA1c.

Procedural Details: Time from prior CABG to the index PCI, vascular access site (radial/brachial vs. femoral), and target vessels for PCI (native coronary arteries and/or bypass grafts).

Medication Use: Use of hypoglycemic agents (insulin and oral drugs) before admission and guideline-directed medical therapies at discharge, including antiplatelets, statins, β-blockers, and angiotensin-converting enzyme inhibitors/angiotensin receptor blockers/angiotensin receptor-neprilysin inhibitors (ACEI/ARB/ARNIs).

The key exposure variable, SHR, was calculated using the following formula: admission FBG (AFBG)/[1.59 × HbA1c - 2.59]. Based on the SHR values, the entire study population was stratified into three groups: the lowest tertile (SHR ≤ 0.7700), the middle tertile (0.7700 < SHR ≤ 0.8899), and the highest tertile (SHR > 0.8899).

Information on adverse cardiovascular outcomes was obtained by trained personnel with no knowledge of baseline characteristics through telephone contact with the patients or their family members and was determined by careful review of the corresponding medical records. The primary endpoint was the occurrence of MACCE, a composite of all-cause death, non-fatal stroke, non-fatal MI, or unplanned revascularization during the follow-up period. Secondary endpoints included the individual components of the primary endpoint and a key secondary composite endpoint of all-cause death, non-fatal stroke, or non-fatal MI. The final follow-up was conducted in April 2022.

Continuous variables were presented as mean ± standard deviation (SD) for normally distributed data or median (interquartile range, IQR) for skewed data. Comparisons across three groups were made using one-way ANOVA or the Kruskal-Wallis test, and between two groups were made using t-test or Mann-Whitney U test. Categorical variables were expressed as numbers (percentages) and were compared using the chi-square test or Fisher’s exact test.

The SHR was primarily analyzed as a categorical variable (i.e., the SHR tertiles) and a continuous variable for its association with MACCE. Additionally, receiver operating characteristic (ROC) curve analysis and Youden’s index (sensitivity + specificity − 1) were used to determine the optimal cutoff value of SHR as a continuous variable for predicting the occurrence of MACCE. The cumulative incidence of clinical endpoints over time was visualized using Kaplan-Meier survival curves, and differences among the SHR tertiles were assessed with the log-rank test. To evaluate the independent prognostic value of SHR, both univariate and multivariate Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) with corresponding 95% confidence intervals (CIs) for time to first occurrence of the primary endpoint. Two sets of multivariate models were constructed: one included the GRACE risk score, and a sensitivity analysis excluded it and included its components. Variables with a univariate significance level of < 0.20 were included in the multivariate Cox proportional hazards regression model, while those that might cause internal correlations were excluded. The C-statistic, continuous net reclassification improvement (cNRI), and integrated discrimination improvement (IDI) were calculated to assess the incremental predictive value of adding SHR to the baseline reference prediction models. The model’s discriminatory performance is assessed using the C-statistic, which is equivalent to the area under the receiver operating characteristic curve, with values closer to 1.0 indicating better discrimination. The cNRI quantifies the correct reclassification of patient risk (both with and without events) when the new biomarker is added. The IDI summarizes the overall improvement in predicted probabilities for events and non-events. A positive cNRI and IDI with a P-value < 0.05 indicates a significant improvement in model performance. Furthermore, post-hoc subgroup analyses were performed to assess the consistency of the association between SHR (as a continuous variable) and MACCE across various patient subgroups. The interaction between the subgroup variable and SHR was tested. A two-sided P-value < 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS version 24.0 (IBM Corp., Armonk, New York, US) and R software version 4.1.0 (R Foundation for Statistical Computing, Beijing, China). The calculation of C-statistics and the comparison between two groups were performed using the “survival”, “prodlim” and “survcomp” packages in R; while the calculations of cNRI and IDI were performed using the “survival”, “survC1” and “survIDINRI” packages. The Kaplan-Meier curves were plotted using GraphPad Prism version 7.0 (GraphPad Software Inc., San Diego, California, US).

Of the initial 1,435 patients, only 2 died during hospitalization (the cause of death was related to PCI procedure), and these 2 patients were excluded from this analysis. The baseline characteristics of the 1,208 patients included in the study, categorized into the lowest (n = 402), middle (n = 404), and highest (n = 402) SHR tertiles, are summarized in Table 1. The three groups were well-balanced in terms of age and most traditional cardiovascular risk factors, such as hypertension and smoking status. However, significant differences emerged. The prevalence of diabetes was higher in the lowest and highest SHR tertiles (62.7% and 61.4%, respectively), compared to the middle tertile (47.0%, P < 0.001). Patients in the highest SHR tertile presented with a higher heart rate at admission (69 ± 11 bpm vs. 67 ± 9 and 66 ± 10 bpm in the other groups, P = 0.003) and a more pronounced inflammatory and metabolic profile, characterized by significantly higher levels of hs-CRP, triglycerides, and AFBG (all P < 0.001). Interestingly, HbA1c levels were lowest in the middle SHR tertile. The highest SHR tertile group underwent PCI in graft vessels, particularly saphenous vein grafts (SVGs), more frequently than the other groups (PCI in only graft vessels: 14.4% vs. 7.7% and 8.4%; SVG intervention: 19.9% vs. 11.4% and 13.1%; P < 0.01 for both). Consequently, the use of hypoglycemic agents, both before admission and at discharge (including insulin and oral drugs), was significantly more common in the highest SHR tertile (all P < 0.001). Discharge medications, including antiplatelets, statins, and β-blockers, were otherwise similar across all groups.

Patients were followed up for a median of 1,291 days (750–2008 days). During the period 368 patients (30.5%) experienced MACCE. The baseline characteristics of the study population stratified by MACCE are shown in Table 2. Compared to patients without MACCE, those with MACCE had a significantly higher median SHR (0.8658 vs. 0.8005, P < 0.001). The distribution of MACCE across the SHR tertiles was striking: 20.7% in the lowest, 33.4% in the middle, and 45.9% in the highest tertile. Patients who developed MACCE had a less favorable risk profile at baseline, including higher BMI, SBP, and heart rate at admission. They also had a greater burden of comorbidities, such as hypertension, renal dysfunction, and past PCI. Patients with MACCE had a more atherogenic lipid profile (higher LDL-C and triglycerides, lower HDL-C), and higher levels of hs-CRP and AFBG (all P < 0.05).

The Kaplan-Meier curves vividly illustrated a graded, positive relationship between the SHR tertiles and the follow-up incidence of MACCE over the long-term follow-up. The cumulative incidence of MACCE was significantly higher with each increasing SHR tertile (log-rank P < 0.001; Figure 1A). This pattern was consistent for the key secondary endpoint (log-rank P < 0.001; Figure 1B). The difference in the incidence of MACCE was mainly driven by significant increases in all-cause death (log-rank P = 0.045; Figure 1C), non-fatal stroke (log-rank P = 0.010; Figure 1D), and unplanned revascularization (log-rank P < 0.001; Figure 1F). However, the incidence of non-fatal MI was not significantly different among the SHR tertiles (log-rank P = 0.240; Figure 1E).

In univariate analysis (Table 3), both the middle and highest SHR tertiles were strongly associated with an increased risk of MACCE compared to the lowest tertile (HR 1.557, 95% CI 1.169-2.072, P = 0.002; HR 2.145, 95% CI 1.636-2.812, P < 0.001, respectively). When analyzed as a continuous variable, each unit increase in SHR was associated with a 24% increased risk of MACCE (HR 1.239, 95% CI 1.103-1.392, P < 0.001; Supplementary Table S1).

The prognostic significance of a high SHR remained robust after extensive multivariate adjustment. In the primary model adjusting for the GRACE risk score and other confounders (Table 3), the middle and highest SHR tertiles persisted as powerful, independent predictors of MACCE (adjusted HR: 1.557, 95% CI 1.166-2.079, P = 0.003, and 1.943, 95% CI 1.476-2.557, P < 0.001, respectively). Similarly, SHR as a continuous variable retained its independent association with MACCE risk (adjusted HR: 1.276, 95% CI 1.105-1.474, P = 0.001; Supplementary Table S1). According to the ROC curve analysis and Youden’s index, the optimal cut-off value of SHR for predicting MACCE was 0.7889. Compared with those with SHR < 0.7889, patients with SHR ≥ 0.7889 were at higher risk of MACCE (adjusted HR: 1.694; 95% CI: 1.340-2.141; P < 0.001).

The consistency of the above findings was confirmed in sensitivity analyses that the COX models excluded the GRACE risk score and included its components, where the middle and highest SHR tertiles remained significantly and independently associated with MACCE (adjusted HR: 1.549, 95% CI 1.160-2.068, P = 0.003, and 1.909, 95% CI 1.448-2.517, P < 0.001; Supplementary Table S2). Similarly, SHR as a continuous variable retained its independent association with MACCE risk (adjusted HR: 1.276, 95% CI 1.104-1.474, P = 0.001; Supplementary Table S3).

The addition of SHR to the baseline reference prediction model including GRACE risk score improved model predictive performance markedly (C-statistic: increased from 0.559 [95% CI 0.567-0.631] to 0.626 [95% CI 0.597-0.656], P = 0.002; cNRI: 0.580 [95% CI 0.244-1.177], P = 0.016; IDI: 0.133 [95% CI 0.044-0.229], P = 0.010). Similarly, the addition of SHR had an incremental effect on the predictive ability of the baseline reference prediction model excluding GRACE risk score and including its components for MACCE (C-statistic: increased from 0.615 [95% CI 0.584-0.647] to 0.639 [95% CI 0.610-0.668], P = 0.003; cNRI: 0.556 [95% CI 0.242-1.180], P = 0.014; IDI: 0.118 [95% CI 0.043-0.209], P = 0.004). Compared with the baseline GRACE risk score, the addition of SHR had a significant increase in C-statistic from 0.540 (95% CI 0.506-0.574) to 0.590 (95% CI 0.560-0.620) (P = 0.003), and significant improvement in reclassification as assessed by the cNRI (0.654, 95% CI 0.215-1.048, P = 0.012) and IDI (0.158, 95% CI 0.054-0.270, P = 0.004).

The formula of SHR included AFBG and HbA1c. We compared the predictive abilities of SHR, AFBG and HbA1c for MACCE. The C-statistics of SHR, AFBG and HbA1c were 0.613 (0.569-0.658), 0.608 (0.578-0.639), and 0.550 (0.517-0.583), respectively. Based on the pairwise comparison of the C-statistics, SHR performed the best.

Subgroup analyses were performed to examine the association between SHR (as a continuous variable) and MACCE across various patient strata (Figure 2). The deleterious effect of a higher SHR was consistently observed in most subgroups, including those stratified by sex (male vs. female), BMI (≥ vs. < 24 kg/m²), diabetes status (yes vs. no), clinical presentation (UA vs. acute MI [AMI]), and hs-CRP levels (≥ vs. < 2 mg/L). Significant interactions were detected for age (≥ vs. < 60 years) and hypertension (yes vs. no), indicating that the association between SHR and MACCE was notably stronger in patients younger than 60 years (HR 2.568, 95% CI 1.509-4.370) compared to those aged 60 years or older (HR 1.192, 95% CI 1.044-1.361; P for interaction = 0.010), and in patients without hypertension (HR 2.554, 95% CI 1.461-4.465) compared to those with hypertension (HR 1.197, 95% CI 1.051-1.365; P for interaction = 0.009).

The SHR is a simple, cost-effective, and instantly calculable parameter. Its integration into the initial clinical evaluation of ACS patients with prior CABG could substantially refine early risk stratification. Identifying the highest-risk individuals using SHR could justify several management intensifications, such as: (1) preferential use of potent P2Y12 inhibitors or other antiplatelet strategies in selected cases; (2) earlier and more comprehensive screening for microvascular dysfunction; and (3) closer long-term follow-up for the detection of recurrent ischemia.

Future research should be directed towards several key questions. First, prospective multi-center studies are needed to validate our findings and potentially develop a dedicated risk prediction model incorporating SHR for this specific population. Second, mechanistic studies exploring the pathophysiological links between SHR, endothelial dysfunction, and accelerated atherosclerosis are warranted. Third, future studies should investigate whether patients with elevated SHR benefit from stricter LDL-C targets or earlier initiation of combination lipid-lowering therapy. Of note, nearly all patients in our cohort (99.1%) were discharged on moderate-dose statin therapy. However, the persistent high event rate in patients with elevated SHR suggests that residual metabolic and inflammatory risk may not be fully addressed by standard lipid-lowering regimens. Recent guidelines support the use of more intensive lipid-lowering strategies, including high-dose statins plus ezetimibe and/or proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors, particularly in high-risk subgroups of ACS patients (43, 44). Therefore, whether SHR can identify high-risk individuals who are able to benefit from more intensive lipid-lowering strategies warrants further exploration. Finally, and perhaps most importantly, intervention studies are warranted to investigate whether a treatment strategy targeting stress hyperglycemia—guided by the SHR—can improve outcomes. This could involve protocols for smoother and more controlled in-hospital glucose management or exploring the potential benefits of novel anti-diabetic agents with proven cardiovascular benefits (e.g., sodium-glucose cotransporter 2 [SGLT2] inhibitors, glucagon-like peptide-1 [GLP-1] receptor agonists) initiated early in the post-ACS period in these high-risk patients, as recommended in recent guidelines (43, 44).

The interpretations of our study must be considered in the context of its limitations. The retrospective, single-center design, while allowing for a detailed analysis of a specific cohort, inherently carries the risk of residual confounding despite extensive statistical adjustments (45). The calculation of SHR relied on a single baseline AFBG and HbA1c measurement, which may not fully represent the dynamic nature of glycemic control or glycemic variability during the hospitalization, a factor itself linked to outcomes (46). Details on in-hospital glycemic management (e.g., insulin protocols) and long-term glycemic control after discharge were not available and could represent unmeasured confounders. Finally, as with any single-center study, the generalizability of our findings requires confirmation in larger, prospective, multi-ethnic cohorts.