A total of 1574 consecutive patients with a confirmed diagnosis of STEMI undergoing primary PCI in Beijing Chaoyang Hospital Heart Center between January 2014 and June 2017 were retrospectively enrolled. The diagnostic criteria of STEMI were in correspondence with the third universal definition of myocardial infarction (MI) [14]: (1) a significant increase or decrease in cardiac biomarkers,(2) persistent ischemic symptoms (within 12 h of presentation), and(3) a newly developed left bundle branch block pattern, or a new ST elevation in two or more contiguous leads, with readings of at least 0.2 mV in leads V1, V2, and V3 or at least 0.1 mV in the remaining leads. Exclusion criteria were a life expectancy of less than 6 months, a history of coronary artery bypass, or conditions resulting in a severe impairment of quality of life.

All the general clinical information, demographics, and treatment records were retrospectively collected from the medical database of Beijing Chaoyang Hospital. The potential inflammation status that complicated the condition of STEMI patients, including a history of chronic obstructive pulmonary disease (COPD), asthma, rheumatoid arthritis, gout, psoriasis, chronic kidney disease, and obstructive sleep apnea hypopnea syndrome (OSAHS), was identified. Laboratory tests, including white blood cell count, hemoglobin, platelet count, low-density lipoprotein, triglyceride, creatinine, fast glucose, brain natriuretic peptide (BNP), and high-sensitivity C-reactive protein (hs-CRP), were recorded within 12 h after admission. The cardiac troponin I (CTNI) and creatine kinase–myocardial band (CKMB) were consecutively censored with an interval of 6 h until the peak was obtained. The ESR was tested within 12 h after PCI by the instruments (Monitor 100, Italy), which is in accordance with the standard Westergren method. Echocardiography was also performed within 12 h postoperatively and left ventricular ejection fraction (LVEF) was recorded.

In this study, coronary angiography was reviewed and assessed by two specialists according to standards. Both investigators were blinded to the outcomes and management of medication during follow-up. Coronary single vessel disease was defined as vessel stenosis > 50% in a major coronary artery (e.g., left anterior descending coronary artery, left circumflex coronary artery, or right coronary artery) and/or in its main branches. Multiple vessel disease was defined as stenosis > 50% in at least two major coronary arteries.

Subsequently, the SYNTAX score was measured from the initial diagnostic angiogram according to the online SS calculator (www.​syntaxscore.​com). All lesions ≥1.5 mm in diameter and stenosis ≥50% were scored and the completely occluded infarct-related artery with TIMI (Thrombolysis in Myocardial Infarction) flow less than 1 was scored as a total occlusion within 3 months. Due to the SYNTAX score available for the stable CAD, patients with history of CABG (coronary artery bypass grafting) were excluded. Subsequently, SSII was calculated on two anatomical variables (SS and left main CAD) and six clinical variables (age, gender, COPD, peripheral arterial disease, creatinine clearance, and LVEF) using an automatic online calculation system.

All subjects were followed up by telephone contact or scheduled outpatient visits to track disease development and occurrence of major clinical adverse cardiovascular events. The duration of follow-up for all patients was up to 52 months. The MACE was set as the primary endpoint of this study, which was mainly defined as a combination of cardiovascular death, rehospitalization for heart failure, recurrent nonfatal MI, repeated coronary revascularization, and nonfatal stroke.

This study was approved by the institutional review board of Beijing Chaoyang Hospital and performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Written informed consent forms were obtained from all patients or their legal relatives.

The normality of data distribution was tested by Shapiro–Wilk test analysis. Continuous variables were presented as mean ± standard deviation (mean ± SD) or median (interquartile range) and categorical variables as frequency (percentage). Student t-test and one-way analysis of variance (ANOVA) test were measured as appropriate for parametric data. Mann–Whitney U or Kruskal–Wallis nonparametric tests were performed on nonnormally distributed variables. The Pearson χ² test or Fisher’s exact test was used to determine the differences of categorical variables, as appropriate. The Kaplan–Meier survival curves were applied to assess the incidence of MACE and the log-rank test was used to determine the intergroup differences. Univariate and multivariate Cox proportional hazard regression analyses were performed to identify predictors for adverse clinical outcomes. To evaluate the incremental prognostic value of the addition of ESR to the SSII risk system, several analytical approaches were applied to compare the changes when ESR was added in the study: (i) The receiver operating characteristic (ROC) curves were performed to assess the predictive value of SSII alone and in combination with ESR, respectively, and (ii) in order to analyze the degree to which the addition of ESR improved the predictive ability of the SSII model, the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were performed [15]. NRI is fit for the models reconstructed for participants negative or positive to events and quantifies the correct movement of categories – upwards for events and downwards for nonevents. Therefore, the event NRI (NRIe) was set as the net percentage of persons with the event of interest correctly assigned a higher predicted risk, while nonevent NRI (NRIne) as the net percentage of persons without the event of interest correctly assigned a lower predicted risk. Conversely, IDI is fit for the difference between average sensitivity and “1-specificity” for models with or without ESR and measures enhancement in average sensitivity without sacrificing the average specificity of the new model [15].(iii) A nested model was constructed to test whether a model combining the two factors could offer a better prognostic value by using the likelihood ratio test. Akaike’s information criteria (AIC) evaluated the probability that a given model is the “best” fitting model of those studied: a lower value of AIC often indicates a better fit [16]. Statistical analyses were performed using STATA (version 15.0). All statistical tests were two-tailed, and a probability value of ≤0.05 was considered statistically significant.

After screening all angiographies, there were 485 STEMI and multivessel disease patients without CABG. According to the exclusion criteria, a total of 483 consecutive patients were finally enrolled and completed the follow-up in this study. The patients were divided into two groups according to the level of ESR (normal group, ESR ≤15 mm/h; and elevated group, ESR > 15 mm/h) and three groups in line with the tertiles of SSII (low SSII group, score ≤ 22; moderate SSII group, score between 22 and 33; and high SSII group, score ≥ 33) at baseline. A summary of the measurements is presented in Table 1. Patients in the elevated ESR group were older and mostly female; presented with lower body mass index (BMI), lower counts of white blood cells and platelets, lower hemoglobin, and higher hs-CRP level; and complicated with a higher prevalence of a prior history of smoking, hypertension, diabetes mellitus, COPD, and peripheral vascular disease (PVD). The elevated ESR group presented with higher SSII than that in the normal ESR group (p < 0.001), whereas the SYNTAX score did not show any difference. No significant difference was found in the inflammatory status, expect for a higher prevalence of chronic kidney disease among the elevated ESR group (p < 0.001).

As is usually reported in similar cohorts of patients, the average age, BMI, admission heart rate, white blood cell counts, and creatinine were significantly different among the three SSII groups. Patients in the high SSII group had a higher prevalence of diabetes mellitus, hypertension, smoking, COPD, and PVD than the prevalence in those in the low SSII group (Fig. 1 and Supplementary Table 1). Compared with the low SSII group, lower LVEF (p < 0.001), higher ESR level (p < 0.001), and lower levels of cholesterol (p = 0.01) and low-density lipoprotein (p = 0.02) were significantly exhibited by the high SSII group. There were no significant differences in other characteristics or variables among the three groups.

A total of 139 (27.8%) subjects reached the clinical endpoint, including 28 (5.8%) cardiovascular deaths, 21 (4.4%) recurrent MIs, 65 (13.5%) repeat revascularizations, 43 (8.9%) heart failures, and 6 (1.2%) nonfatal strokes. The overall MACE was significantly higher for the elevated ESR group compared with those in the normal group (46.0% vs. 22.8%, respectively, p < 0.001). Patients in the elevated ESR group were prone to suffer a higher frequency of mortality (9.7% vs. 4.5%, p = 0.03), heart failure (21.8% vs. 4.5%, p < 0.001), revascularization (19.4% vs. 11.4%, p = 0.03), and nonfatal stroke (1.6% vs. 0.6%, p = 0.04). Additionally, when stratified by the tertiles of the SSII risk system, the incidence of MACE was significantly higher in the high SSII group (37.7% vs. 25.5% vs. 17.0%, p < 0.01), particularly in mortality (10.6% vs. 3.85% vs. 0.0%, p < 0.001) and heart failure (16.1% vs. 5.4% vs. 1.0%, p < 0.001), when compared with those in moderate SSII group and low SSII group, respectively.

Subsequently, the cumulative incidences of MACE among the two ESR groups and three SSII groups were assessed by a Kaplan–Meier survival curve (Fig. 2). It was demonstrated that patients with elevated ESR had a significantly higher occurrence of MACE compared to those in the normal level. In contrast to those with moderate and high SSII, decreased event-suffer probability was observed in the low SSII group. A log-rank test on the curve identified significant differences between the two ESR groups (χ² = 31.97, p = 0.01) and among the three SSII groups (χ² = 15.99, p < 0.001) in the cumulative incidence of MACE in patients with STEMI and multivessel disease.

The Cox proportional hazard regression analysis results are shown in Table 2. Univariate analysis revealed that both SSII (HR = 1.038, 95% CI = 1.024–1.053, p < 0.001) and ESR level (HR = 1.027, 95% CI = 1.022–1.041, p < 0.001) were associated with an increased risk of MACE. We took into account the multicollinearity between SSII and its continuous parameters (SS, age, LVEF, and creatinine); hence, those parameters did not enter into the multivariate regression analysis following SSII. After adjustment for other potential confounders including BMI, admission heart rate, admission systolic pressure, cholesterol, low-density lipoprotein, hs-CRP, and a history of hypertension, diabetes mellitus, hyperlipidemia, or smoking, both SSII and ESR were still independent powerful predictors for the incidence of MACE in patients with STEMI and multivessel disease (ESR, HR = 1.022, 95% CI = 1.012–1.033, p < 0.001; and SSII, HR = 1.033, 95% CI = 1.017–1.049, p < 0.001).

By using the likelihood ratio, the nested models were used to classify which one could provide better positive predicted probability on different clinical outcomes. The combined model provided a better fit with a lower AIC than that in models containing SSII alone (MACE, 1284.235 vs. 1299.304, p < 0.001; cardiovascular death, 295.6914 vs. 299.6875, p < 0.01; heart failure, 438.9405 vs. 461.0, p < 0.001) (Table 3).

The additive prognostic value of ESR in addition to SSII was evaluated by an increase in the area under the ROC curve (AUC) (Fig. 3 and Supplementary Figs. 1 and 2). The AUC of the combined models regarding MACE, cardiovascular death, and heart failure was 0.683, 0.8802, and 0.8467, respectively, of which all were higher than that for SSII alone (MACE, 0.653, p = 0.04; cardiovascular death, 0.8546, p = 0.22; heart failure, 0.7743, p < 0.01).

Models combining ESR and SSII significantly improved the net reclassification in predicting compound events and heart failure (Table 4). The combined model analysis results revealed a significant improvement in reclassification by 6.5% for those with MACE and by 38.4% for those without, with both reaching a significant statistical difference (p < 0.001). Therefore, the combination of SSII and ESR is likely to predict a lower risk of MACE than that in SSII alone, and subsequently provide a 31.9% improvement in net reclassification in the nonevent group. Meanwhile, in the setting of cardiovascular death, all measures of discrimination and reclassification, the combination of SSII and ESR resulted in significant improvement by 66.8 and 20.9% in the nonevent and event groups, respectively (NRIne, 0.668; NRIe, 0.209; NRI, 0.877, p < 0.001). IDI analysis results also indicated that the model predictive value on MACE and heart failure, but not cardiovascular death, was significantly improved by the addition of ESR to the SSII system (MACE, 0.033, p < 0.001; cardiovascular death, 0.120, p = 0.57; heart failure, 0.099, p < 0.01) (Table 4).

In this retrospective cohort study from a single center, some limitations should be cautioned. First, due to the relatively small number of eligible patients with STEMI and multivessel disease from only one cardiac center, our findings are inevitably affected by a selection bias. Further large multicenter studies should be conducted. Second, although there is a statistical significance between the two ESR groups, a low incidence of non-fatal stroke cases probably contributed to some restrictions on the evaluation of cerebrovascular events in this study. It is less reasonable that the anatomical SYNTAX score and derivative SSII have access to the prevalence of non-stroke patients. Third, we did our best to complete the follow-up, but it is inevitable that a few admitted to other hospitals are missing because of the fluent population in Beijing. Finally, as STEMI patients accompanied with inflammation disorders were not excluded in this study, it is uncertain that the elevated inflammation marker is potentially influenced by inflammation status. However, whether the patients were complicated with other inflammation disorders, the ESR could assess the extent of inflammation activity and predict coronary atherosclerosis and cardiac mortality.