In this prospective study, a total of 410 patients were evaluated and datasets from 340 patients were entered in the final analysis (70 patients were excluded based on the exclusion criteria, Additional file 1: Figure S1). All adult patients (aged 18 years or older) with any gender with a definite STEMI diagnosis for the first time who were admitted because of chest pain within 24 h since the initiation of symptoms at a tertiary heart center (the name of the center is undisclosed for peer review) were consecutively included from April 2020 to September 2020. Patients were diagnosed as STEMI if they had typical ischemic chest pain lasting ≥ 20 min and ST-elevation of 0.25 mA or more at the J point in men below the age of 40 years, ≥ 0.2 mV in men over the age of 40 years, or ≥ 0.15 mV in women in leads V2–V3 and/or ≥ 0.1 mV in other leads (in those without left ventricular hypertrophy or left bundle branch block) on the admission ECG, with/without an increase in cardiac enzyme concentrations (troponin I [CTNI] and/or creatine phosphokinase-MB [CKMB]) [13]. Patients with atrial fibrillation or flutter, left ventricular hypertrophy, right or left bundle branch block, implanted pacemaker or defibrillator, or Wolf-Parkinson-White syndrome were excluded due to imposing additional ECG alteration. Informed consent was obtained from all patients, and the study was conducted in accordance with the declaration of Helsinki [14]. The study protocol was approved by the medical ethics committee of our institution (the name of the institution and ethical code is undisclosed for peer review). The demographics, medical history, laboratory test results, ECG, echocardiographic and angiographic findings were documented for all patients. Patients were categorized into four strata, based on T wave amplitude in aVR lead in their first ECG at admission to the emergency room (Additional file 1: Figure S2): The follow-up after exiting hospitalization was performed by telephone interviews, regular visits, and hospital records evaluation.

The primary endpoint was in-hospital mortality, defined as the occurrence of death due to heart attack/STEMI during the exiting hospital stay. The secondary endpoints were the length of hospital stay (defined as the number of days from admission to discharge/death), the number of patients who develop ventricular tachycardia (VT; sustained and non-sustained) or ventricular fibrillation (VF) during the hospital stay, re-hospitalization (defined as re-admission because of either myocardial infarction or decompensated heart failure until six months from onset of disease), and six months cardiovascular mortality (defined as the occurrence of death due to cardiovascular causes in six months from onset of disease).

All 12-lead ECGs were obtained using MAC 500 ECG machine (GE medical system, USA) with a paper speed of 25 mm/s and standard voltage. ECG analyses were conducted by an attending cardiologist who was blinded to the patients’ clinical status, using digital calipers on a 12-lead ECG and magnified to 200% of normal size. The ECG analysis was repeated by another attending cardiologist and the discrepancies were resolved by consultation with the third cardiologist to mitigate intra-observer variability.

The T-wave amplitude was measured as the value of the largest deflection above and below the baseline in a window spanning from 80 ms after the end of QRS to the end of the T wave. The infarction location was determined based on ECG findings.

In addition to CTNI, creatinine (Cr) measurements, and complete blood tests, routine clinical and paraclinical quantitative measurements were undertaken per standard procedures. Echocardiographic evaluations were conducted at the time of admission to the emergency room. Two-dimensional transthoracic echocardiography was performed once and repeatedly by two attending cardiologists who were blinded to the ECG findings of the patients, using a vivid S5 ultrasound machine (GE medical system, USA) with harmonic imaging. Left ventricular systolic function, defined by the left ventricular ejection fraction (LVEF), was calculated by the biplane Simpson’s method of discs. Coronary angiography was performed on an as-needed case-by-case basis by a team of expert cardiologists, in accordance with Judkins or Amplatz techniques. Multivessel disease was defined as the presence of luminal diameter stenosis of more than 50% in at least two major coronary arteries. Global Registry of Acute Coronary Events (GRACE) score was calculated for all patients [15].

The normal distribution of all variables was tested by the Kolmogorov–Smirnov test. The median and interquartile range (IQR) of quantitative variables (due to non-normal distribution) and frequency and percentage of qualitative variables were reported. Kruskal–Wallis test was utilized to compare quantitative variables between groups. The Chi-square test was used for qualitative variables. Relative risk (RR) at 95% CI was calculated for each T wave amplitude strata. Sensitivity, specificity, negative and positive predictive values (NPV, PPV) were calculated for three cutoff points of T wave amplitude (0, − 1, − 2 mV), using MedCalc software version 19 (MedCalc Software Ltd, Belgium). Multivariable regression analysis was conducted to identify the independent determinants of T wave amplitude in aVR lead. In this analysis, T wave amplitude was considered as a continuous variable and the independent variables were selected from the most clinically relevant variables. All variables in a block were entered in a single step in the regression model (procedure for variable selection = Enter method). Before performing the modeling, continuous variables (including age, LVEF, Cr, BS, Systolic BP, HR) were converted to binary variables at their clinically relevant cut-off points (e.g., patients’ age was converted to the number of patients with the age of ≥ 65 years or < 65 years). Correlation between GRACE score and T wave amplitude on aVR lead was analyzed using Spearman's rank-order correlation test. Youden's J statistic (Youden’s index = sensitivity (%) + specificity (%) − 100) was used to determine the optimal cutoff point. The statistical performance of the GRACE score for prognosis of STEMI was calculated to be compared with that in T wave amplitude and better elucidate the prognostic value of T wave amplitude on aVR lead. For this purpose, receiver operating characteristic (ROC) curves are plotted with the calculation of concordance statistic (C-statistics = area under curve [AUC]) and coordinate points of ROC curve. The statistical analysis was conducted using SPSS version 24 (SPSS co., Chicago). Meta-analysis software (CMA) version 3 was used to subgroup analysis and forest plot. Two-sided P values (p) were reported with a significance level of 0.05. For the clinical outcome and relative risks, the statistical comparisons were conducted first between these four strata (≥ 0, − 1 to 0, − 2 to − 1, and < − 2 mV) with a focus on the group of T wave amplitudes of ≥ 0 mV as the most positive T wave group (Tables 1, 2). Then, three nominal cut-off points of T wave amplitudes (0 mV, − 1 mV, and − 2 mV) were selected and the discrimination power for each was determined (Table 3). Because of the highest Youden’s index at the cutoff point of − 1, subgroup analyses were conducted only at this level.

The baseline clinical and laboratory data of 340 included STEMI patients and four T-wave strata are described in Table 1. T wave amplitudes of ≥ 0, − 1 to 0, − 2 to − 1, and < − 2 mV were detected in 29 (8.5%), 153 (45%), 125 (36.8%), and 33 (9.7%) patients, respectively. Age was significantly different among groups (p, 0.001): patients with a T wave amplitude of ≥ 0 mV had the highest median of age (median [IQR], 70 [63, 78] years in patients with T wave amplitude of ≥ 0 mV vs 64 [54, 72], 56 [48, 67], and 55 [48, 59] years in those with T wave amplitudes of − 1 to 0, − 2 to − 1, and < − 2 mV, respectively). The majority of patients were male (82.9%). There were no significant gender-specific differences between four T wave strata (p, 0.136). The prevalence of diabetes and hypertension (HTN) was significantly different among groups (p, 0.019 and 0.020, respectively) and they were more prevalent in patients with a T wave amplitude of ≥ 0 mV (34.5% and 51.7% in this group, respectively). Moreover, LVEF was significantly lower in patients with a T wave amplitude of ≥ 0 mV (median [IQR], 30 [25, 40]% in patients with T wave amplitude of ≥ 0 mV vs 40 [30, 45]%, 40 [35, 50]%, and 45 [40, 50] % in those with T wave amplitudes of − 1 to 0, − 2 to − 1, and < − 2 mV, respectively; p, 0.001).

An overall, 62 patients (18.2%) developed VT/VF, and 24 patients (7%) died during the hospital stay. In-hospital mortality was higher in patients with a T wave amplitude of ≥ 0 mV compared to other T-wave groups (7 [24.1%] in patients with T wave amplitude of ≥ 0 mV vs 16 [10.4%], 1 [0.8%], and 0 [0%] in those with T wave amplitudes of − 1 to 0, − 2 to − 1, and < − 2 mV, respectively; p 0.001, Table 1). However, no significant difference was observed between groups in terms of VT/VF rate (p 0.801). Length of hospital stay was also significantly different among groups (p 0.001). Patients with a T wave amplitude of ≥ 0 mV had significantly longer stay in hospital compared to other T-wave groups (median [IQR], 7.5 [7, 10] days in patients with a T wave amplitude of ≥ 0 mV vs. 6 [5, 7], 6 [5, 7], and 5 [5, 6] days in those with T wave amplitudes of − 1 to 0, − 2 to − 1, and < − 2 mV, respectively).

All the survived patients (316 cases) were successfully followed for six months. During this period, 70 patients (22.2%) were re-admitted to the hospital and 28 patients (8.8% of followed patients) died due to cardiovascular causes (total six months cardiovascular mortality, 52 cases [15.2%]). Both re-hospitalization and mortality were higher in those with a T wave amplitude of ≥ 0 mV (Table 1).

The relative risks (RR) for aVR lead T wave amplitude thresholds were calculated using the T-wave amplitude in aVR lead 0 to − 1 mV as the reference group since nearly half the population was in this group. The groups of patients with higher T wave amplitude in aVR, had progressively increased relative risk (RR) of in-hospital mortality, re-hospitalization, and six months-cardiovascular mortality (Fig. 1; Table 2). Moreover, there was a significant correlation between GRACE score and T wave amplitude on aVR lead (correlation coefficient [β], 0.354, P < 0.001, Additional file 1: Figure S3).

The multivariate regression analysis identified three independent predictors of T wave amplitude in aVR lead including LVEF > 40% (β − 0.278; 95% CI − 0.477 to − 0.079), age ≥ 60 years (β 0.195; 95% CI 0.025–0.365), HR ≥ 90 (β 0.184; 95% CI 0.006–0.362). Other variables with non-significant correlations with T wave amplitude are described in Additional file 1: Table S1. Associations between T wave amplitude in aVR lead determinants and their correlations with patients' clinical outcomes are depicted in Fig. 2.

T wave amplitude at the cutoff point of 0 mV had high specificity (> 90%), but low sensitivity (< 50%) for prediction of the endpoints (including 6-month cardiovascular mortality, in-hospital mortality, and re-hospitalization (Additional file 1: Table S2 and Table S3). On the other hand, T wave amplitude at the cutoff points of − 1 mV and − 2 mV exhibited high sensitivity (> 90%), but low specificity (< 60%) for the prediction of all study endpoints (Table 3). The cutoff point of − 1 mV had the highest Youden’s index value for all study endpoints. The sensitivity and specificity of T wave amplitude at the cutoff point of -1 mV were 86.54% (95% CI 74.21–94.41) and 52.43% (95% CI 46.49–58.32) for cardiovascular mortality in six months, 95.83% (95% CI 78.88–99.89) and 49.68% (95% CI 44.04–55.33) for in-hospital mortality, and 67.14% (95% CI 54.88–77.91) and 54.47% (48.02–60.81) for re-hospitalization, respectively (Table 3).

Nevertheless, subgroup analysis demonstrated that T wave amplitude at the cutoff point of − 1 mV had higher specificity in those patients with better clinical conditions (age < 65 years, non-diabetic, non-hypertensive, with Cr < 1.2 mg/dl, early presentation [time from symptom to hospital < 12 h, Killip class = 1, or having no PAH or MR, Table 4).

GRACE score indicated higher discriminative potential of prognosis toward the study endpoints (Additional file 1: Figure S4). It had an AUC of 0.928 (95% CI 0.883–0.973) for in-hospital mortality (optimal cutoff point, 167 with sensitivity and specificity of 83.3% and 83.7%, respectively), 0.935 (95% CI 0.906–0.963) for six month-cardiovascular mortality (optimal cutoff point, 141 with sensitivity and specificity of 90.4% and 93.0%, respectively), and 0.757 (95% CI 0.688–0.826) for re-hospitalization (optimal cutoff point, 149 with sensitivity and specificity of 50.0% and 93.3%, respectively).

Although our study was the first to investigate the short and midterm prognosis of STEMI patients with positive T wave in aVR and its discriminative potential, it contains certain limitations. The diagnosis of STEMI was made according to recent guidelines [13]. Neither the invasive procedures (such as tissue histopathology examination), nor provocative testing for ischemia or nuclear magnetic resonance imaging were conducted to validate the diagnosis and the affected myocardium (infarction location). To be more practical in interpretations of ECG records by the clinicians and to make it readily observable independently, especially in an emergency room situation, the smallest unit of T wave amplitude change was considered as 1 mV. T wave amplitude as continuous variables (for values smaller than 1 mV) is questionable relative to commonly used blood tests. Future investigations should verify the sufficiency, accuracy, and precision of the proposed optimal cutoff points for T wave amplitude for a better threshold prognosis, particularly for disconcordants, as well as people with unverified clinical characteristics and treatments in relation to multiple risk factors (Fig. 2; Additional file 1: Table S1). Prognosis of the uncertain anterior STEMI should be confirmed through exclusionary diagnosis of non-ST segment (NSTEMI) cases in high risk people that typically require interventions (e.g., blood thinners, PCI, stenting) with independent variables and those with blockages of multiple coronary arteries and diabetes that CABG rather than angioplasty is recommended.