Basic clinical characteristics such as age, gender, total cholesterol, triglyceride, HDL, LDL, systolic blood pressure, diastolic blood pressure, diabetes, smoking history and medication were drawn into the present study. Clinical characteristics of all the patients and divided by evidence of LV remodeling, and by experience of endpoint are summarized in Table 1. The concentrations of cTnT and NT-proBNP on admission, LVEDV and LVEF at follow-up, ΔLVEDV, ΔLVESV and ΔLVEF levels were significantly higher in patients with remodeling than those without (p < 0.05). There were no significant differences in any other clinical characteristics between the remodeling and non-remodeling group. NT-proBNP levels on admission, ΔLVEDV, ΔLVESV and ΔLVEF levels were significantly higher in patients in the experienced endpoint group than those in the no endpoint group (p < 0.05). There were no significant differences in any other clinical characteristics between the experienced endpoint and no endpoint group.

To assess the value of plasma miR-208b and miR-34a for predicting LV remodeling after AMI, we categorized all the patients into remodeling (n = 116) and non-remodeling (n = 243). LV remodeling was defined as at least 10% increase from the baseline in LVEDV during follow-up. Plasma miRNA levels were determined using qRT-PCR. The results were summarized in Table 2. Independent-samples T test showed statistical differences of plasma miR-208b and miR-34a levels between remodeling and non-remodeling group. Results of 2^−ΔΔCt analysis showed that plasma concentrations of miR-208b and miR-34a were both markedly elevated in patients with remodeling relative to those without, fold 4.81 (±3.57) and 3.26 (±2.42) respectively (Figure 1A,B).

ROC curve analysis was performed to evaluate the predictive power of plasma miR-208b and miR-34a for LV remodeling after AMI. The ability to discriminate the patients with remodeling from those without was determined according to the AUC of 0.780 (95% confidence interval, 0.734–0.822; p < 0.001) for miR-208b and 0.738 (95% confidence interval, 0.689–0.783; p < 0.001) for miR-34a, respectively. The combination of miR-208b and miR-34a showed an AUC of 0.812 (95% confidence interval, 0.767–0.851; p < 0.001). The AUC measured for NT-proBNP was 0.704 (95% confidence interval, 0.654–0.751; p < 0.001) (Figure 2A). The comparison of ROC curves between miR-208b and Nt-proBNP was of borderline statistical significance (z = 1.96, p = 0.05). MiR-34a was not superior to NT-proBNP in the prediction (z = 0.875, p = 0.381). However, the combination of the two miRNAs showed a superior predictive power compared to Nt-proBNP (z = 2.605, p = 0.009).

Cardiogenic death or development of heart failure during follow-up was considered as primary endpoint. A total of 83 patients (23.1%) experienced the primary endpoint. As shown in Figure 1C,D, levels of miR-208b and miR-34a were both elevated in patients who experienced primary endpoint, compared to those with no endpoint (p < 0.05), fold 4.21 (±4.55) and 3.12 (±2.86) respectively. The ability of plasma miR-208b, miR-34a and combined miRNAs to discriminate the experienced endpoint group from those with no endpoint was evaluated by the ROC curve with an AUC of 0.737 (95% confidence interval, 0.689–0.782; p < 0.001), 0.642 (95% confidence interval, 0.590–0.691; p < 0.001) and 0.777 (95% confidence interval, 0.731–0.819; p < 0.001), respectively. The AUC measured for NT-proBNP was 0.669 (95% confidence interval, 0.617–0.717; p < 0.001) (Figure 2B). Comparison of ROC curves showed that neither of the single miRNAs was superior to NT-proBNP in predicting endpoint after AMI (miR-208b vs. NT-proBNP: z = 1.526, p = 0.127; miR-34a vs. NT-proBNP: z = 0.629, p = 0.529). While the combined miRNAs showed a higher predictive power than NT-proBNP (z = 2.496, p = 0.013). Multivariate logistic regression analysis showed that Odds Ratios (95% confidence interval), adjusted for age, gender, current smoking, cTnT, NT-proBNP, and time from AMI onset to sampling were 17.91 (2.07–98.81, p = 0.003) for miR-208b, 4.18 (1.36–12.83, p = 0.012) for miR-34a and 18.73 (1.96–101.23, p = 0.000) for the combination of miR-208b and miR-34a, respectively.

Reclassification analyses were performed to evaluate the added value of each miRNA and both miRNAs over a multi-parameter clinical model including age, gender, current smoking, cTnT, NT-proBNP, and time from AMI onset to sampling. Patients were categorized into low (<10%), intermediate (10%–30%) and high (>30%) probability of remodeling groups. Results showed that the capacity to reclassify patients originally misclassified by the multi-parameter clinical model into remodeling and no remodeling was 9.5% (NRI = 0.095, p = 0.039) for miR-208b, 6.5% (NRI = 0.065, p = 0.177) for miR-34s and 11.7% (NRI = 0.117, p = 0.025) for the combination of miRNAs, respectively (Tables 3–5). Again, the IDI calculated for miR-208b, miR-34a and the combination of miRNAs was 0.042 (p = 0.013), 0.008 (p = 0.150) and 0.077 (p = 0.002), respectively.

Three hundred and fifty-nine consecutive patients diagnosed with AMI were enrolled in this study from the First Affiliated Hospital of Zhengzhou University (Zhengzhou, China) between December 2011 and October 2012. Patient characteristics are described in Table 1. Diagnosis of AMI was based on combination of several parameters: (1) ischemic symptoms; (2) increased levels of myocardial necrosis markers (troponins (cTns) and creatine kinase (CK)) to above twice the upper limit of the normal range; (3) pathological Q wave; and (4) ST-segment elevation or depression [39]. Patients with previous MI or PCI, hematological diseases, acute or chronic infection, significant renal or hepatic dysfunction, and known or treated malignancies were excluded. Hypertension was defined according to the presence of elevated systolic (>140 mmHg) and/or diastolic (>90 mmHg) blood pressure or the current use of antihypertensive drugs. Definition of diabetes mellitus was based on clinical features and requirement of dietary treatment and/or medical therapy to control blood glucose levels. Hyperlipidaemia was defined as serum total cholesterol levels ≥5.2 mmol/L, or triglycerides ≥1.7 mmol/L, or low density cholesterol ≥2.6 mmol/L, or use of statin medication [40]. Patient was considered as a smoker if he/she was smoking at the current moment or was a smoker in the past.

Echocardiographic studies were completed by three cardiologists. However, the same operator analyzed the same patient during the hospitalization (approximately 1–4 days after admission) and at the six-month follow-up. Patients were categorized according to whether they demonstrated LV remodeling post AMI, as assessed from the change (ΔLVEDV) in LVEDV between admission and follow-up. LV remodeling was defined as at least 10% increase from the baseline in LVEDV during follow-up [41]. For the assessment of the prognostic value of circulating miRNA levels, the cardiogenic death (n = 19) or development of heart failure (clinical diagnosis (n = 25), an ejection fraction <40% (n = 61)) during follow-up was considered as primary endpoint. Subjects were classified as heart failure when they met the Framingham criteria for the clinical diagnosis, and if circulating NT-proBNP was above the age-related cutoff points published by Januzzi et al. [42].

Fasting venous blood samples from AMI patients were collected in tubes containing EDTA—K₂ on admission (a median of 6 (range 2–10) h after the onset of symptoms). Samples were centrifuged at 3000× g for 10 min at 4 °C, then the supernatant was isolated and centrifuged at 12,000× g for 10 min at 4 °C. Plasma was collected and stored in aliquots at −80 °C until analysis.

Two hundred microliters of plasma were spiked with miScript miRNA mimic SV40 (Qiagen, Hilden, Germany, 2 μM, 1 μL per 100 μL plasma). Total RNA was extracted from these plasma samples using TRI Reagent BD (MRC, TB-126, Cincinnati, OH, USA) according to the manufacturer’s protocol and dissolved in 10 μL diethylpyrocarbonate (DEPC)-treated water. The concentration and quality of the RNA samples was determined using NanoDrop spectrophotometer (NanoDrop, Thermo Fisher Scientific, Waltham, MA, USA).

Total RNA (0.5 μg) from each sample was reverse transcribed using miRNA-specific stem-loop RT primer. The PrimeScript^OeR RT reagent Kit With gDNA Eraser (TaKaRa, DRR047S, Tokyo, Japan) was used according to the kit procedures. In brief, the 20 μL reactions were incubated for 15 min at 42 °C, followed by 5 s at 85 °C, and the resulting cDNA was stored at −20 °C for the following quantitative real-time PCR. qRT-PCR was performed on ABI 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) using SYBR^OeR Premix Ex Taq™ II (TaKaRa, DRR820S, Tokyo, Japan) according to the manufacturer’s instructions. Briefly, the thermal cycling consisted of an initial denaturation at 95 °C for 30 s, followed by 40 cycles of 95 °C for 3 s and 60 °C for 30 s. A melt curve was performed after each reaction. All PCR reactions yielded a single peak on the melt curve, indicating acceptable specificity of the primers. Assays were performed in triplicate. Resultant miRNA levels were normalized using spiked-in SV40. Data were analyzed with 7500 Fast System SDS Software version 1.4.0.25 (Applied Biosystems, Foster City, CA, USA) with the automatic C_t setting for assigning baseline and threshold for C_t determination. The C_t values greater than 36 were considered as not expressed. MiRNA relative expression levels were calculated by the equation of 2^−ΔCt, (ΔC_t = C_{t target} − C_{t spiked-in SV40}). Fold change in miRNA levels were calculated using the 2^−ΔΔCt method [43].

SPSS Statistics software version 20.0 (IBM, Armonk, NY, USA) was used for the statistical analysis. All data were subjected to a normality test (Kolmogorov-Smirnov). Continuous data are presented as mean ± SD or median with interquartile range. Categorical variables are presented as counts and percentage. Independent-samples t-test and Mann-Whitney U test were used to compare two groups of continuous variables and Chi-Square test for categorical variables. Receiver operating characteristic (ROC) curve analysis and the comparision of the derived area under the curve (AUC) were performed using Medcalc software version 13.0.2 to estimate the predictive power of biomarkers. Prediction of miRNA levels for cardiogenic death or development of heart failure was assessed by multivariate logistic regression. Reclassification analyses were performed, evaluating the predictive value for the identification of remodeling of each miRNA and both miRNAs together with a multi-parameter clinical model including age, gender, current smoking, cTnT, NT-proBNP, and time from AMI onset to sampling. Logistic regression models were used for patient classification. The net reclassification index (NRI) and the integrated discrimination improvement (IDI) [44] were calculated to validate the added predictive value of the miRNAs. All p-values were two-tailed and a level of p < 0.05 was considered statistically significant.

The protocol of this study was carried out according to the principles of the Declaration of Helsinki and approved by the Medical Ethics Committee of the First Affiliated Hospital of Zhengzhou University. Written informed consent was obtained from all the participants before enrolment.