We retrospectively evaluated 987 AMI patients who had been admitted to the cardiac intensive care unit at Gyeongsang National University Hospital, Korea, between January 2012 and July 2014. Patients were eligible if they were ≥ 18 years of age with ST-segment elevation AMI and non-ST elevation AMI by significant coronary artery stenosis treated with PCI. The diagnosis and treatment of patients were based on the current American College of Cardiology/American Heart Association guidelines for the management of patients with ST elevation myocardial infarction and non-ST-segment elevation acute coronary syndrome [8]. We excluded patients with AMI by vasospasm. For this study, 145 patients were excluded because of thyroid disease; who had been treated with previous or current thyroid diseases, and who were suspected thyroid disease with extremely abnormal thyroid hormone levels (thyroid stimulating hormone [TSH], > 9 mU/L; or fT4, > 3.0 ng/dL) [3]. During follow-up period, 77 patients dropped out because of no follow-up information. Institutional Review Board approval was obtained by the local ethics committee and informed consent was waived.

Blood sampling for biochemistry including thyroid hormone and brain natriuretic peptide (BNP) was performed as soon as the patients arrived at the ED. All patients had baseline levels of thyroid hormone including total T3, fT4, and TSH determined using an Elecsys electrochemiluminescent immunoassay (Roche Diagnostics Ltd., Mannheim, Germany). The reference ranges for our laboratory were total T3 (80 to 200 ng/dL), fT4 (0.93 to 1.70 ng/dL), and TSH (0.27 to 4.2 mU/L). BNP was measured by Elecsys electrochemiluminescent immunoassay (Roche Diagnostics Ltd., Basel, Switzerland). Other biochemical measurements including creatinine, cardiac specific enzyme (creatine kinase-MB, troponin I), and lipid profile were also evaluated. Glomerular filtration rate (mL/min/1.73 m² ) was calculated by the Modification of Diet in Renal Disease formula.

All eligible patients underwent coronary angiography (CAG) and PCI as coronary reperfusion therapy. CAG was performed by standard technique and significant coronary artery disease was diagnosed visually if luminal diameter narrowing ≥ 50% was present in a major epicardial coronary artery. Left main trunk disease was counted as two-vessel disease and the presence of more than two significant coronary artery diseases was considered multivessel disease.

Echocardiographic assessment was conducted within 24 hours of admission. We measured echocardiographic parameters according to current guidelines of the American Society of Echocardiography. The left ventricular ejection fraction (LVEF) was measured using modified Simpson’s method. As the parameter of left ventricular diastolic function, a ratio of mitral inflow E velocity of mitral annular tissue E’ velocity (E/E’) were calculated. The mean value of septal and lateral E’ was used for AMI patients having regional wall motion abnormalities. The right ventricular systolic pressure (RVSP) was assumed using the peak velocity of tricuspid regurgitation and inferior vena cava.

Follow-up clinical data of 765 PCI-treated AMI patients were obtained from the following sources: reviewing the patients’ hospital medical records, periodically examining the patients on outpatient clinics, and interviewing the patients via telephone. We analyzed the prevalence of 1-year MACCEs, defined as re-hospitalization for HF, nonfatal myocardial infarction or severe angina for coronary revascularization, ischemic cerebrovascular accident, and cardiac death within 12 months. In present study, the definition of cardiac death required the documentation of significant arrhythmia or cardiac arrest, death attributable to congestive HF or myocardial infarction in the absence of any other precipitating factors.

Results for continuous variables are given as the mean ± standard deviation, whereas results for categorical variables are presented as frequencies and percentages. Comparisons between continuous variables were made using Student t test. Comparisons between categorical variables were evaluated using Fisher exact test or Pearson chi-square test, as appropriate. We generated receiver operating characteristic (ROC) curves to define the optimal cut-offs of total T3 and T3/fT4 ratio for predicting HF, and the area under curve (AUC) and its associated 95% confidence interval (CI) were assessed using Medcalc version 13.3.3.0 statistical software (Medcalc, Ostend, Belgium). AUC comparison between categorical variables, high BNP and low T3/fT4, was performed using Medcalc statistical software. Survival analysis for clinical outcome was performed by Kaplan-Meier curve. Variables with p < 0.1 were then entered into the multivariate logistic regression analysis providing odds ratio and 95% CI. A p < 0.05 was considered to indicate a significant difference. All statistical analyses were performed using the SPSS version 21.0 (IBM Co., Armonk, NY, USA).

A total of 765 PCI-treated AMI patients (65.4 ± 12.6 years old, 215 women) were included (Fig. 1). The 143 patients (18.7%) experienced MACCEs for 1-year. The prevalence of 1-year HF requiring re-hospitalization was 3.9% (30 patients) and that of in-hospital and 1-year death caused by cardiac disease was 6.8% (52 patients). In this study, in-hospital cardiac mortality comprised the majority (47 patients, 90.3%) of cardiac death after AMI and pump failure was the main reasons for in-hospital cardiac death. Five patients experienced cardiac death after discharge, three patients were revisit emergency room because of sudden cardiac death, one patient died from late stent thrombosis, and one patient expired due to progressive hypoxemia from intractable HF. We dichotomized into two groups, the HF or non-HF group, and compared the characteristics (Table 1). Compared to those in the non-HF group, the patients in the HF group were older (72.1 ± 12.4 years vs. 65.1 ± 12.6 years, p = 0.030) and had predominance of females (53.3% vs. 27.1%, p = 0.020). The initial-presented Killip classes and BNP levels, as well-known markers of HF of AMI, were significantly different in the two groups. The angiographic results, whether multivessel or infarct related artery disease, were not different in the two groups. In echocardiographic parameters, the baseline LVEF of the HF group were lower than those of the non-HF group (41.9% ± 10.0% vs. 53.5% ± 8.4%, p < 0.001) and the levels of mean E/E’ and assumed RVSP were higher in the HF group than in the non-HF group (Table 1). The patients in the HF group had lower total T3 levels and higher fT4 levels (84.32 ± 21.04 ng/dL vs. 101.20 ± 20.30 ng/dL, p < 0.001; 1.43 ± 0.28 ng/dL vs. 1.28 ± 0.22 ng/dL, p = 0.001, respectively) than those in the non-HF group (Table 1). Similarly, with 1-year HF, the total T3 levels were significantly lower in the patients who had experienced ischemic cerebrovascular accident, cardiac death, and all MACCEs (Fig. 2). The total T3 levels showed a tendency to be lower in patients with 1-year revascularization compared to in patients without. Contrary to total T3, the fT4 levels were only correlated with HF among MACCEs (data not shown).

The ROC curve analysis showed that the cut-off values of total T3 for predicting the 1-year HF was ≤ 85 ng/dL (AUC, 0.734; 95% CI, 0.701 to 0.765; p < 0.001). The sensitivity, specificity, negative predictive value, and positive predictive value were 67%, 78%, 98%, and 11%, respectively. In order to reflect the fT4 change, we assessed the T3/fT4 as one of prognostic parameters. The cut-off value of T3/fT4 for predicting the 1-year HF was calculated as ≤ 60 (AUC, 0.774; 95% CI, 0.743 to 0.803; p < 0.001) (Fig. 3). The predictive values of HF were slightly augmented by using a composite marker, T3/fT4, compared to a single marker, total T3 level (sensitivity 60%, specificity 87%, negative predictive value 98%, positive predictive value 16%). In patients with T3/fT4 ≤ 60, the incidence of 1-year HF was higher (15.5% vs. 1.8%, p < 0.001) (Fig. 4). There were trend of higher 1-year cardiac mortality in patients with T3/fT4 ≤ 60 (10.3% vs. 5.7%, p = 0.060). We evaluated whether the T3/fT4 was an independent parameter for prediction of 1-year HF requiring re-hospitalization using multivariate logistic regression analysis. Among statistically significant parameters (p < 0.010) on univariate analysis, higher BNP, lower LVEF, and lower T3/fT4 were independent predictors (Table 2). The Killip class and mean E/E’ were not significant in multivariate analysis. The total T3 level and the T3/fT4 levels were also a significant predictor for all MACCEs additional to BNP level. Multivariate analysis for all MACCEs showed similar results with 1-year HF (Table 3). However, for all MACCEs, AUC of T3 was larger than that of T3/fT4 (AUC, 0.683; 95% CI, 0.634 to 0.731 for T3) (AUC, 0.647; 95% CI, 0.595 to 0.699 for T3/fT4). T3/fT4 ≤ 60 was showed the additive value to BNP (> 123 pg/mL) in ROC curve analysis (AUC, 0.818; p < 0.001 for BNP) (AUC, 0.865; p = 0.015 for T3/fT4 combined with BNP) (Fig. 5). As a result, low T3/fT4 and high BNP had most powerful predictive value for 1-year HF in patients with AMI.