We prospectively enrolled adult AMI patients admitted to the coronary care unit (CCU) between June 1, 2017, and May 31, 2020, at Zhongshan Hospital Fudan University. AMI was defined as when there is acute myocardial injury with clinical evidence of acute myocardial ischemia and with detection of a rise and/or fall of cardiac troponin (cTn) values with at least one value above the 99th percentile URL and at least one of the following: symptoms of myocardial ischemia; new ischemic electrocardiograph changes; development of pathological Q waves; imaging evidence of new loss of viable myocardium or new regional wall motion abnormality in a pattern consistent with an ischemic etiology (18). The exclusion criteria were as follows: inability to provide informed consent, clinical data were not available, and coronary angiography results were not consistent with the diagnosis of myocardial infarction.

The studies involving human participants were reviewed and approved by the Ethics Committee of Zhongshan Hospital, Fudan University (Shanghai, China, approval number: B2022–375R). The patients provided written informed consent to participate in this study.

Blood samples were taken after 12 h of overnight fasting on the second day of admission to CCU. All biomarkers were analyzed at the Department of Laboratory Medicine, Zhongshan Hospital, Fudan University. Levels of IL-1β, sIL-2R, IL-6, IL-8 and IL-10 were determined using the chemiluminescence method on automatic chemiluminescence immune analyzer IMMULITE 1,000 (Siemens, Germany). The lower detection limit of IL-1β was 5.0 pg/ml; the upper detection limit of sIL-2R was 7,500 U/ml; the lower detection limit of IL-6 was 2 pg/ml, and the upper detection limit was 1,000 pg/ml; the upper detection limit of IL-8 was >7,500 pg/ml; the lower detection limit of IL-10 was 5 pg/ml, and the upper detection limit was 1,000 pg/ml. When the actual test result is lower than the detection limit, we handle it with the lowest detection limit value. Levels of cardiac troponin T (cTnT) and N-terminal pro-brain natriuretic peptide (NT-proBNP) were determined with electrochemiluminescence on Cobas e 801 (Roche Diagnostics, Germany) and the inter-assay and intra-assay coefficient of variation were <10%. Levels of high sensitivity C reactive protein (hs-CRP) were determined with immunoturbidimetry on Cobas c 702 (Roche Diagnostics, Germany) and the inter-assay and intra-assay coefficient of variation were <5%. Routine biochemical analyses, including total cholesterol, high-density lipoprotein (HDL) cholesterol (mg/dl) and fasting blood sugar were determined by conventional laboratory assays. The diagnostic criteria for diabetes are random blood glucose ≥11.1 mmol/L or fasting blood glucose ≥7.0 mmol/L or oral glucose tolerance test 2-hour blood glucose ≥11.1 mmol/L or glycated hemoglobin A1c ≥ 6.5% (19).

Patients were usually followed up regularly at the outpatient clinic every one to three months after discharge. After all patients were enrolled, the investigators followed up once a year at a centralized time. A composite endpoint of major adverse cardiac events (MACEs) was defined as cardiovascular death, MI, unscheduled revascularization ≥3 months after the index infarction, rehospitalization for heart failure, or stroke. Furthermore, long-term data on composite endpoints were collected after a median of 2.2 years of follow-up.

We used the Kolmogorov-Smirnov test to assess the normality of the variables. Continuous variables were presented as means with standard deviations or medians with interquartile ranges (IQR), and we used independent t-tests or rank-sum tests to compare variables between groups. Qualitative variables are presented as frequencies and corresponding percentages, and we used chi-squared or Fisher's exact tests to compare variables between groups. Associations between interleukins, clinical variables and endpoints were assessed by univariate cox regression analysis. A multivariable cox regression was used to study the association of interleukin with endpoints adjusted for cardiovascular risk factors.

Several risk factors for adverse cardiovascular outcomes in patients with AMI have been identified, and various risk models have been established, such as, Thrombolysis in Myocardial Infarction (TIMI) score (20, 21), and Global Registry of Acute Coronary Events (GRACE) score (22). However, different risk models were developed based on the characteristics of the respective study populations. We thus constructed a “multivariate plus traditional risk factor” model regarding the above risk model, taking into account the population characteristics and sample size of this study. This model included the following factors: heart rate, systolic blood pressure, smoker, diabetes mellitus, hypertension, hypercholesterolemia, coronary artery disease, heart failure, history of stroke, history of percutaneous transluminal coronary intervention (PCI) or coronary artery bypass grafting (CABG), renal insufficiency, anemia, total cholesterol, HDL cholesterol, fasting blood sugar. We also measured biomarkers commonly used to reflect cardiac function, including cTnT, hs-CRP and NT-proBNP. As a result of skewness, the following continuous variables were logarithmically transformed with the natural logarithm: cTnT, hs-CRP, NT-proBNP. We included all covariates with a p-value ≤0.05 in univariate analysis and other clinically relevant variables in multivariable models. We used a backward stepwise regression approach to produce the final model. The variance inflation factor (VIF) was used to measure the severity of multicollinearity in regression analysis. The following covariates were included in the model to explore the predictive value of interleukin for endpoint events at 1-year follow-up: age (>65 vs. ≤65 years), heart rate (>100 vs. ≤100 b.p.m.), systolic blood pressure (<100 vs. ≥100 mmHg), renal insufficiency (eGFR <60 vs. ≥60 ml/min), and anemia (yes vs. no). The following covariates were included in the model to explore the predictive value of interleukin for endpoint events at long-term follow-up: age (>65 vs. ≤65 years), heart rate (>100 vs. ≤100 b.p.m.), history of PCI or CABG (yes vs. no), anemia (yes vs. no), renal insufficiency (eGFR <60 vs. ≥60 ml/min), and fasting blood sugar. We used multivariable cox regression to study the association of interleukin with endpoints adjusted for cardiovascular risk factors. sIL-2R and IL-8 remained significantly associated with endpoints after adjusting for cardiovascular risk factors in multivariate Cox regression. We used linear regression analysis to analyze the correlation between sIL-2R and IL-8. We transformed sIL-2R and IL-8 from continuous variables to grouped variables. Cutoff values for the variable groupings were obtained by receiver operator characteristic curve (ROC) analysis, and interleukins were classified as high (greater than the cutoff value) and low (less than or equal to the cutoff value) interleukins levels. Baseline characteristics of the study population stratified by sIL-2R and IL-8 levels were analyzed. Multivariate cox regression was used to analyze the predictive value of high interleukin levels (vs. low interleukin levels) for endpoints. Moreover, we also plotted Kaplan-Meier curves based on the follow-up (years) to estimate the cumulative survival between groups of high and low interleukin levels by log-rank test without adjustment. GRACE score may affect the prognosis of MI patients (22). We compared the predictive value of GRACE score with sIL-2R, IL-8 for prognosis. We assessed the ability of sIL-2R and IL-8 by themselves or in combination with GRACE score to discriminate patients with or without a composite endpoint by the area under the receiver-operating curve. We also evaluated the benefits of adding sIL-2R and IL-8 to the GRACE score using continuous and category-based net reweighting indices (NRI). All p-values were two-sided, and a p-value ≤0.05 was considered statistically significant. We analyzed the data using SPSS software (version 26.0, IBM Inc., Armonk, NY, United States) and R software, (version 4.2.1; R Core Team).

A total of 176 patients with AMI were screened for this study, and 3 cases were excluded (clinical data were not available,1; the final diagnosis was myocarditis, 2). A total of 173 patients were included in this study. The median (IQR) age was 64 (15) years, and 147 (85.0%) were male. Median follow-up was 2.2 years (maximum 5.2 years) (Table 1). A total of 24 patients (13.8% of the total population) experienced MACEs during 1-year follow-up, including 9 cardiovascular deaths (1 history of MI), 13 reinfarctions (3 history of MI), 1 hospitalization for heart failure and 1 stroke. During long-term follow-up (median follow-up, 2.2 years), MACEs occurred in 40 (23.1%) patients, including 12 cardiovascular deaths (1 history of MI, 2 history of strokes, 1 history of stroke and MI), 24 reinfarctions (9 histories of MIs), 1 MI,1 hospitalization for heart failure and 2 strokes (1 history of stroke and MI).

Patients who experienced MACEs during 1-year or long-term follow-up tended to be older and more often had cardiovascular risk factors including rapid heart rate (>100 b.p.m.), renal insufficiency (eGFR <60 ml/min), anemia and higher fasting blood sugar, compared with patients without MACEs. Patients who experienced MACEs during 1-year or long-term follow-up had significantly higher GRACE scores and higher levels of NT-proBNP than those without MACEs (Table 1).

In univariable Cox regression analyses, IL-1β, sIL-2R, IL-8 and IL-10 levels were significantly elevated in patients who experienced MACEs during the first year and long-term follow-up compared with patients without (Table 1). After adjustment for clinical characteristics in multivariable Cox regression models, sIL-2R and IL-8 remained significantly elevated in patients who experienced MACEs during the first year (Supplementary Table S1) and long-term follow-up (Supplementary Table S2). Furthermore, multivariable Cox regression analysis showed that among the five interleukins studied, only sIL-2R and IL-8 were independently associated with MACEs during follow-up (Supplementary Table S1, S2). Linear regression analysis showed that serum sIL-2R was positively associated with IL-8 levels (p = 0.006), but the model fit was poor (R-squared 0.044) (Supplementary Figure S1). We then convert sIL-2R and IL-8 from continuous variables to grouped variables. Cutoff values were calculated by ROC analysis for dichotomizing IL-8 into high (greater than the cutoff value 32.5 pg/ml) and low levels (less than or equal to 32.5 pg/ml) for 1-year follow-up. The same approach was used to classify sIL-2R as high sIL-2R (greater than the cutoff value 807 U/ml) and low sIL-2R (less than or equal to 807 U/ml).

Baseline characteristics of the included patients based on levels of sIL-2R (high vs. low) and IL-8 (high vs. low) were shown in Supplementary Table S3. Compared with patients with low sIL-2R levels, patients with high sIL-2R levels tended to be older and had significantly higher total cholesterol, fasting blood sugar, hs-CRP levels, and more often had hypertension, heart rate (>100 b.p.m.), diabetes mellitus, renal insufficiency and anemia. Compared with patients with low IL-8 levels, patients with high IL-8 levels had significantly higher fasting blood sugar and NT-proBNP levels, and more often had coronary artery disease.

Patients with high sIL-2R levels (higher than the cutoff value) had a significantly higher risk of MACEs during 1-year follow-up (Figure 1) with a hazard ratio (HR) of 7.7 (95% CI: 3.3–18.0; p < 0.001) (Table 2). This association remained significant after adjustment for relevant significant cardiovascular risk factors (adjusted HR: 3.5; 95% CI: 1.4–9.0; p < 0.001) and all cardiovascular risk factors (adjusted HR: 3.7; 95% CI: 1.2–11.9; p = 0.028) (Table 2). Also, patients with high sIL-2R levels had a significantly higher risk of MACEs during long-term follow-up (HR: 2.9; 95% CI: 1.4–5.7; p = 0.003), which remained significant after adjustment for different cardiovascular risk factors (Supplementary Figure S2 and Table S4). The results of VIF between the factors were all <2 in model 1 and <3 in model 2.

Patients with high IL-8 levels (higher than the cutoff value) had a significantly higher risk of MACEs during 1-year follow-up (Figure 2) with an HR of 4.8 (95% CI: 2.1–10.7; p < 0.001) (Table 3). This association remained significant after adjustment for relevant significant cardiovascular risk factors (adjusted HR: 4.4; 95% CI: 1.9–9.9; p < 0.001) and all cardiovascular risk factors (adjusted HR: 6.2; 95% CI: 2.2–17.5; p = 0.006) (Table 3). Also, patients with high IL-8 levels had a significantly higher risk of MACEs during long-term follow-up (HR: 3.3; 95% CI: 1.8–6.4; p < 0.001), which remained significant after adjustment for different cardiovascular risk factors (Supplementary Figure S3 and Table S5). The results of VIF between the factors were all <2 in model 1 and <3 in model 2.

Multivariable Cox regression analysis models including cardiovascular risk factors, cTnT levels, hs-CRP levels, and NT-proBNP levels showed that high levels of sIL-2R and IL-8 remained associated with an increased risk of MACEs during the first year's follow-up (Figure 3), as well as during the long-term follow-up (Supplementary Figure S4). ROC analysis regarding predictive accuracy for MACEs during the first-year follow-up showed that the area under the curve (AUC) was 0.763 for GRACE scores (0.64–0.88, p < 0.001), 0.66 (0.54–0.79, p = 0.011) for sIL-2R levels and 0.69 (0.56–0.82, p < 0.001) for IL-8 levels (Figure 4). ROC analysis did not find any significant predictive value of cTnT levels, hs-CRP levels, and NT-proBNP levels for MACEs during the first-year follow-up (Supplementary Figure S5). The sIL-2R combined with IL-8 levels had further better predictive value for MACEs than sIL-2R, IL-8 alone during the first-year follow-up, with an AUC of 0.720 (0.59–0.85, p < 0.001) (Figure 4).

The GRACE score combined with sIL-2R, IL-8 and sIL-2 combined with IL-8 further improved the predictive power for MACEs during the first-year follow-up, with AUCs of 0.768 (0.645–0.891, p < 0.001), 0.786 (0.666–0.905, p < 0.001) and 0.788 (0.666–0.910, p < 0.001), respectively. Adding sIL-2R, IL-8, sIL-2R combined with IL-8 to GRACE score for predicting MACEs during the first year's follow-up reclassified a large proportion of patients to the correct risk stratum. The NRI values of adding sIL-2R, IL-8, and sIL-2R combined with IL-8 to the GRACE score were 0.0766 (p = 0.355), 0.167 (p = 0.129), and 0.208 (p = 0.029) respectively (Figure 4). The addition of sIL-2R combined with IL8 to the GRACE score led to a significant improvement in the predictive power of the new model, contributing to a 20.8% increase in the proportion of correct classifications. We also compared typical model performance metrics between the different models, and the results for Akaike information criterion (AIC) and Bayesian information criterion (BIC) were shown in the Supplementary Table S6.