Introduction

Sudden cardiac death (SCD) remains a leading cause of mortality in chronic heart failure patients with reduced ejection fraction (HFrEF), despite sufficient medication.1,2 The main cause of SCD is ventricular tachyarrhythmias. Therefore, identifying HFrEF patients at high risk of SCD is important.

Cardiac I-123-metaiodobenzylguanidine (MIBG) imaging provides prognostic information in patients with HFrEF.3,4,5,6,7,8,9,10,11,12,13 AdreView Myocardial Imaging for Risk Evaluation in Heart Failure (ADMIRE-HF) study provided a prospective validation of the independent prognostic value of MIBG scintigraphy in assessing patients with HFrEF.8 Furthermore, ADMIRE-HF risk score is a novel index that combines clinical characteristics and MIBG imaging variables to provide individualized estimates of serious arrhythmic risk in patients with HFrEF.14 However, there is no information available on the external validation of the ADMIRE-HF score for the prediction of SCD in HFrEF patients.

Early repolarization pattern (ERP) in the standard 12-lead electrocardiogram (ECG) is characterized by an elevation of the J point, which is due to QRS slurring or notching in at least two inferior or lateral leads.15 Although ERP was considered benign for decades, previous clinical studies reported that patients with idiopathic ventricular fibrillation had a high prevalence of ERP.16,17,18 Furthermore, in a case–control study, ERP was more common in patients with previous myocardial infarction who had ventricular arrhythmic events.19 Pei et al. and we have recently shown that ERP in the inferior leads is associated with an increased risk of SCD in HFrEF patients.20,21 However, no information is available on the long-term value of combining ADMIRE-HF risk score and ERP for predicting SCD in HFrEF patients.

The aim of this study was to perform the external validation of ADMIRE-HF risk score and compare the prognostic values of ADMIRE-HF risk score and ERP for predicting SCD in patients with HFrEF.

Methods

We enrolled 111 consecutive HFrEF outpatients with radionuclide left ventricular ejection fraction (LVEF) < 40%, from 1995 October to 2000 December. Heart failure was diagnosed from the clinical signs and symptoms according to the Framingham criteria.22 Patients were required to be stable for at least 3 months on conventional therapy. Patients were excluded from the present study if they had significant renal dysfunction, insulin-dependent diabetes mellitus, or autonomic neuropathy. None of the patients had had an implantable cardioverter-defibrillator (ICD), biventricular pacemaker, or biventricular defibrillator (CRT-D) at enrollment. We also excluded 18 patients with bundle branch block and three patients with a pacemaker rhythm due to the inability to assess ERP.

The final study population included 90 HFrEF patients. The mean patient age was 65 ± 11 years. Of the 90 patients, 68 were men and 22 were women. Heart failure was due to ischemic heart disease in 50 patients and idiopathic dilated cardiomyopathy in 40 patients. The average New York Heart Association (NYHA) functional class was 2.0 ± 0.6 (class I, 11%; class II, 68%; class III, 21%). The LVEF was 31 ± 7%. 5 patients received ICD treatment during the follow-up period. Informed consent was obtained from each patient, and all procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. At entry, all patients underwent cardiac MIBG imaging, 12-lead electrocardiography, 24-hour Holter ECG monitoring, and echocardiography. In addition, a venous blood sample was drawn.

All patients were assessed while resting in the supine position by myocardial imaging with I-123 MIBG (FUJIFILM RI Pharma Co., Ltd, Tokyo, Japan) using a dual-head gamma camera (Prism 2000, Picker, Bedford, OH, USA) equipped with a low-energy, high-resolution, parallel-hole collimator. A 111 MBq dose of I-123 MIBG was injected intravenously with the patient at rest after an overnight fast. Initial and delayed images were obtained in the anterior chest view at 20 and 200 minutes after isotope injection, respectively. As previously described,4 two independent observers, who were unaware of the clinical status of the patients, assessed the cardiac MIBG uptake. The MIBG heart to mediastinum (H/M) ratio was determined by dividing the counts/pixel in a visually drawn heart region of interest by the counts/pixel in a 7 × 7-pixel upper mediastinum region of interest. After considering the radioactive decay of I-123, the cardiac MIBG washout rate was calculated from the initial and delayed images. ADMIRE-HF score was derived from the sum of the point values of the following parameters as previously reported: LVEF (< 25%, 5 points), MIBG H/M uptake ratio on the delayed image (< 1.6, 12 points), and systolic blood pressure (< 120 mmHg, 3 points; 120-139 mmHg, 0 points; > 140 mmHg, − 3 points). The score had values ranging from − 3 to 20. According to the previous study, study patients were classified into three groups; low (< 4), intermediate (4-15), and high (> 15) ADMIRE-HF risk score groups.14

We obtained a standard resting 12-lead ECG at enrollment. ERP was defined as J-point elevation ≥ 0.1 mV above the baseline in at least two inferior or lateral leads. The anterior precordial leads (V1-V3) were excluded to eliminate patients with potential Brugada syndrome. Two cardiologists (Y.F. and M.K.), who were blinded to the patient’s clinical information, independently reviewed the 12-lead ECG of all subjects. The concordance rate was 90.2% for the diagnosis of ERP. If their opinions differed, a third cardiologist (T.Y.) blindly reviewed the ECG findings and confirmed the presence or absence of ERP. The RR interval, PR interval, QRS duration and QT interval were measured automatically. QT interval was corrected for heart rate using Bazett’s formula: QTc = QT/RR1/2 (ms), where QTc is the corrected QT interval. Patients also underwent 24-hour Holter ECG recording with a Marquette Electronics 8000 Holter monitoring system (Marquette Electronics, Milwaukee, WI, USA). The severity of ventricular arrhythmias was classified according to Lown’s grade, the total number of ventricular premature contractions per day, and nonsustained ventricular tachycardia (defined as ≥ 5 consecutive premature ventricular beats lasting < 30 seconds).

Blood sampling was done from an intravenous cannula after the patients had rested for ≥ 30 minutes in the supine position. The samples were used to assess plasma noradrenaline concentration, serum creatinine, sodium, potassium, and uric acid levels. The plasma noradrenaline concentration was determined using high-performance liquid chromatography at Shionogi Biomedical Laboratories (Osaka, Japan). Two-dimensional echocardiography was performed with a Toshiba SSH-380A recorder equipped with a 2.5 or 3.75 MHz transducer. The standard technique was employed to determine the size of the LV and left atrium. The LV dimension was measured at end-diastole (defined as the onset of the R-wave) just below the level of the mitral leaflets through the standard left parasternal window. The left atrial dimension was measured as the distance from the leading edge of the posterior left wall at end-diastole.

All study patients were followed up prospectively in our hospital at least once a month by clinicians who were blinded to the cardiac MIBG imaging results. The primary endpoint was SCD, defined as witnessed cardiac arrest or death within 1 hour of the onset of acute symptoms, unexpected or unwitnessed death in a patient who was known to have been well within the previous 24 hours. The secondary endpoint was defined as severe arrhythmic events (SAE) such as SCD, sustained VT and appropriate ICD therapy, which are following ADMIRE-HF study. Survival data were obtained by medical records in our hospital, direct contact with patient’s primary physicians, or by telephone interview of patients or their family or mail by dedicated coordinators.

The data are presented as the mean ± standard deviation or median with first quartile to third quartile range. Continuous variables were compared using Student’s t test or Mann–Whitney U test Wilcoxon rank sum test based on their distribution, and the differences in discrete variables were compared using a Fisher’s exact test. In the univariate Cox proportional hazards regression model, the association of the baseline patient characteristics with SCD was assessed, and we used the multivariate Cox proportional hazards regression model to assess the prognostic value of ADMIRE-HF risk score and ERP, adjusting for the clinical variables (age, gender, NYHA functional class, and diabetes mellitus). Cardiac event-free rates were calculated using the Kaplan–Meier method, and differences between groups were detected using the log-rank test. All statistical analyses were performed using MedCalc, Version 16.1.2 (MedCalc Software bvba, Ostend, Belgium). p values < 0.05 were considered significant.

Results

During a median follow-up of 7.5 (4.5-12.0) years, 22 patients had SCD. The baseline characteristics of patients with and without SCD are listed in Table 1. There were no differences in age, gender, the proportion of patients with ischemic heart disease, NYHA functional class, heart rate, blood pressure, drug use, left ventricular end-diastolic dimension, left atrial dimension, serum creatinine, sodium, potassium, or uric acid levels between patients with and without SCD. The LVEF tended to be lower in patients with SCD. Patients with SCD had a significantly higher plasma concentration of noradrenaline.

Table 1 Clinical characteristics in patients with and without sudden cardiac death
Full size table

As for MIBG parameters, patients with SCD had a significantly lower H/M ratio on the early and delayed images, and a greater washout rate than those without SCD (Table 2). ADMIRE-HF score was significantly higher in patients with than without SCD (12 [2-15] vs 3 [0-8.5], p = 0.0171) (Figure 1). On the other hand, 12 of 90 study patients had ERP. The ERP-positive rate was significantly higher in patients with than without SCD. The ERP-positive rate in both the inferior and lateral leads was significantly higher in patients with than without SCD, whereas no significant difference was observed in the ERP-positive rate in the inferior or lateral lead only between the two groups. No significant differences were found in the other parameters in 12-lead standard ECG and Holter ECG monitoring between the patients with and without SCD, except for Lown’s grade and VPC / hour.

Table 2 MIBG and ECG findings in patients with and without sudden cardiac death
Full size table
Figure 1
figure 1

ADMIRE-HF risk score in HFrEF patients with and without sudden cardiac death (SCD)

Full size image

Kaplan-Meier analysis revealed that SCD was more frequently observed in higher ADMIRE-HF risk score group (high:50[5/10]% vs intermediate:33[10/30]% vs low:14[7/50]%, p = 0.0090; hazard ratio, high vs low 4.3 [95% CI 1.1 to 17.2], intermediate vs low 3.3 [95% CI 1.3 to 8.7]) (Figure 2). Moreover, patients with ERP had a higher risk of SCD than those without ERP (67[8/12] % vs 18[14/78] %, p = 0.0001, hazard ratio 4.9 [95% CI 1.3 to 19.5]) (Figure 3).

Figure 2
figure 2

Sudden cardiac death (SCD)-free rate and severe arrhythmic event-free rate curves by Kaplan–Meier analysis in HFrEF patients with high-, intermediate- and low-ADMIRE-HF risk scores

Full size image
Figure 3
figure 3

Sudden cardiac death (SCD)-free rate and severe arrhythmic event-free rate curves by Kaplan-–Meier analysis in HFrEF patients with or without early repolarization pattern (ERP)

Full size image

At univariate analysis, serum sodium and uric acid levels, plasma norepinephrine concentration, ADMIRE-HF risk score and ERP showed a significant association with SCD. Multivariate Cox analysis revealed that both ADMIRE-HF risk score and ERP were independently and significantly associated with SCD (Table 3). SCD was most frequently observed in patients with intermediate/high ADMIRE-HF score and ERP (80[4/5] %) (Figure 4). Moreover, patients with intermediate/high ADMIRE-HF risk score without ERP (31[11/35] %) and those with low ADMIRE-HF risk score and ERP (57[4/7] %) experienced SCD significantly more frequently than in those with low ADMIRE-HF risk score without ERP (7[3/43] %). The adjusted hazard ratio for SCD prediction in patients with intermediate/high ADMIRE-HF score and ERP was 27.0 (95%CI 2.0 to 361.4), which was about 4-fold higher that the hazard ratio in those with either intermediate/high ADMIRE-HF score or ERP (6.5 [95%CI 2.8 to 15.3]).

Table 3 Univariate and multivariate Cox proportional hazards analysis to identify patients with HFrEF at risk of sudden cardiac death
Full size table
Figure 4
figure 4

Sudden cardiac death (SCD)-free rate and severe arrhythmic event-free rate curves in HFrEF patients stratified by the ADMIRE-HF risk score and early repolarization pattern (ERP)

Full size image

During the follow-up period, three patients had sustained VT followed by ICD implantation, and one of two patients with the implantation of CRT-D had appropriate ICD therapy, so that 26 patients had SAE as secondary endpoint. ADMIRE-HF score (12 [2-15] vs 3 [0-8], p = 0.0041) and ERP-positive rate (31% vs 6%, p = 0.0041) were significantly higher in patients with than without SAE. Kaplan–Meier analysis revealed that SAE was more frequently observed in higher ADMIRE-HF risk score group (high:60[6/10]% vs intermediate:40[12/30]% vs low:16[8/50]%, p=0.0008; hazard ratio, high vs low 5.2 [95% CI 1.4 to 20.2], intermediate vs low 3.7 [95% CI 1.5 to 9.0]) (Figure 2). Moreover, patients with ERP had a higher risk of SAE than those without ERP (67[8/12] % vs 23[18/78] %, p = 0.0002, hazard ratio 4.2 [95% CI 1.1 to 15.4]) (Figure 3). At univariate analysis, LVEF, serum sodium and plasma norepinephrine concentration, ADMIRE-HF risk score and ERP showed a significant association with SCD. Multivariate Cox analysis revealed that ADMIRE-HF risk score and ERP were independently and significantly associated with SAE. Patients with intermediate/high ADMIRE-HF score and ERP (80[4/5]%) had a significantly higher risk of SAE than those with either (43[18/42]%) or none of them (9[4/43]%) (Figure 4).

Discussion

The major finding of the present study was that we verified the prognostic value of ADMIRE-HF risk score for the prediction of SCD and the combination of ADMIRE-HF risk score and ERP identified a subset of HFrEF patients at high risk of SCD.

In HFrEF patients, sympathetic overactivity and parasympathetic withdrawal are associated with poor outcomes.23 It is well known that increased sympathetic activity is associated with lethal arrhythmias and following SCD, and that increased parasympathetic activity can exert a protective effect.24 The status of the cardiac autonomic nervous system can play a role in all three major pathways (trigger, substrate, and modulator) that are believed to contribute to the initiation and perpetuation of lethal arrhythmias.25 ADMIRE-HF risk score includes a cardiac MIBG variable, which reflects cardiac sympathetic nerve activity. On the other hand, although the exact mechanism for the generation of ERP in patients with structural heart disease remains unknown, it has been reported that ERP in patients with coronary artery disease may be related to the presence of myocardial scar,26 which creates an arrhythmic substrate. This might account for our results that ADMIRE-HF risk score and ERP were independently associated with SCD in HFrEF patients.

ADMIRE-HF risk score has been shown to be a novel tool that effectively stratifies HFrEF patients according to their risk of serious arrhythmic events, and the risk is independent of other conventional risk factors.14 In the sub-analysis of ADMIRE-HF study, which showed the utility of ADMIRE-HF risk score, the primary endpoint was a composite of SCD, resuscitated cardiac arrest, appropriate device therapy, and sustained ventricular tachycardia. Furthermore, no information is available on the external validation of the ADMIRE-HF score for predicting SCD in HFrEF patients. In the present study, ADMIRE-HF risk score was associated with SCD in HFrEF patients. Although there were differences in patients’ characteristics between ADMIRE-HF study and the present study (e.g., less ischemic origin and greater LVEF, compared with ADMIRE-HF study), the present study confirmed that ADMIRE-HF risk score is useful for predicting SCD in patients with stable HFrEF.

The prevalence of ERP has been previously reported to be 1% to 13%.16,17,18,27 ERP has been generally considered to be an innocuous finding in healthy subjects. Previous reports showed the association between ERP and fatal ventricular tachyarrhythmia in patients without structural heart disease.16,17 An experimental study showed that the presence of a prominent action potential notch in the epicardium but not in the endocardium provided a voltage gradient that manifests as a J-wave or elevated J-point in the ECG.15 This voltage gradients can initiate arrhythmogenesis to increase the vulnerability to lethal ventricular tachyarrhythmias in healthy subjects. Although the precise mechanism for generation of ERP remains unknown in patients with HFrEF, ERP in HFrEF patients might differ from that in subjects without structural heart disease, as we previously reported.21

Recent clinical trials have shown that ICDs can reduce the risk of SCD in symptomatic HFrEF patients who have low LVEF.28,29 Therefore, many guidelines recommend ICD implantation for the primary prevention of SCD based on LVEF.30 However, the actual rate of appropriate shocks remains low in HFrEF patients with ICD implantation, according to the current guidelines.31,32 Thus, alternative means are needed to refine the criteria for ICD implantation. In the present study, the combination of ADMIRE-HF risk score and ERP provided incremental prognostic information for the prediction of SCD in HFrEF patients. This result suggests that the combination of ADMIRE-HF risk score and ERP can identify a subset of HFrEF patients at high risk for SCD that could benefit more from ICD implantation. Furthermore, in the present study, it is notable that SCD occurred in only one patient in the group with low ADMIRE-HF risk score and without ERP during the initial 10 years of follow-up. This result also suggests that the combination of ADMIRE-HF risk score and ERP can identify patients with HFrEF who do not need ICD implantation.

Our study has several limitations. First, the small sample size is a major limitation. The results might have been influenced by fortuitous circumstances in this small number of patients. Second, no study patients received the current standard of care (i.e., there was insufficient use of beta-blockers, ICD, biventricular pacemakers, and biventricular defibrillators [CRT-D]) and the measurement of plasma brain natriuretic peptide level at entry because this study was started in the last century. The medications used during follow-up may have affected cardiac MIBG uptake, ERP, and the clinical outcomes. During the follow-up period, five patients received ICD/CRT-D treatment. Patients without SCD had a higher incidence of ICD/CRT-D therapy than those with SCD, although the difference was not statistically significant (7% vs 0%, p = 0.32). Even if the patients with ICD implantation during the follow-up period were excluded from analysis, the similar result was obtained. Third, there was a considerable overlap in ADMIRE-HF risk score between patients with and without SCD, which might reduce the usefulness of ADMIRE-HF risk score. Although the sensitivity (68[15/22]%) and positive predictive value (38[15/40]%) of high/intermediate ADMIRE-HF risk score for SCD prediction was not adequate, the negative predictive value was relatively high (86[43/50]%). This finding suggests that we might be able to identify patients at low risk of SCD if they have low ADMIRE-HF risk score. Fourth, we had no data from T-wave alternans testing, although we previously showed the comparison the prognostic value of heart rate variability with cardiac MIBG imaging.7 Lastly, because we included only stable outpatients with mild-to-moderate HFrEF, patients in NYHA functional class IV were not included in this study. Therefore, our results should not be generalized to patients with severe HFrEF.

New Knowledge Gained

The prognostic value of ADMIRE-HF risk score for the prediction of SCD was verified in HFrEF patients, and the combination of ADMIRE-HF risk score and ERP could identify the HFrEF patients with a higher risk of SCD.

Conclusion

The combination of ADMIRE-HF risk score and ERP provides incremental prognostic value for predicting SCD in patients with HFrEF.