1 INTRODUCTION

Patients with chronic kidney disease (CKD), which number more than 20 million in the United States (Pun, 2014), have a 4- to 20-fold higher risk of sudden cardiac death (SCD) than does the general population. Cardiovascular disease (CVD) and CKD are closely related, with the former having a significant impact on CKD mortality (Go et al., 2004; Ronco et al., 2008). A large cohort study has shown that while only 3.1% of stage 2–4 CKD patients progress to renal replacement therapy, 24.9% die (Keith et al., 2004), and that a higher cumulative cardiovascular comorbidity was associated with the risk of death regardless of CKD severity (Jesky et al., 2013). Therefore, in CKD management, clinicians should pay attention not only to renal function deterioration but also to CVD. Identifying CKD patients at high risk, particularly those with structural heart disease (SHD), is of great importance. However, few studies have addressed this issue to date.

It has been reported that noninvasive electrocardiographic markers (NIEMs), including late potentials (LPs) (Gatzoulis et al., 2018), heart rate turbulence (HRT) (Bauer et al., 2008), and nonsustained ventricular tachycardia (NSVT) (Kinoshita et al., 2020), are useful predictors of lethal arrhythmic events and SCD in patients with SHD. We have recently demonstrated that a combined assessment of NIEMs obtained from high-resolution digital electrocardiogram (ECG) systems accurately predicts the occurrence of lethal arrhythmias and SCD in patients with SHD (Hashimoto et al., 2020; Kinoshita et al., 2020). Although LPs have been associated with mortality and SCD in patients with end-stage renal disease (Morales et al., 1998), little is known about the usefulness of other NIEMs in predicting lethal arrhythmias or cardiac mortality in patients with CKD, regardless of disease severity or the presence of concurrent SHD.

This study aimed to investigate whether noninvasive ECG parameters assessed using new ambulatory ECG systems could predict cardiac death, lethal arrhythmic events, and nonfatal cardiovascular events in CKD patients with SHD.

2 METHODS 2.1 Study design and population

This study was a sub-study of the Japanese Noninvasive Electrocardiographic Risk Stratification for Prediction of Sudden Cardiac Death (JANIES) study (Kinoshita et al., 2020). The JANIES study was a multicenter, observational, prospective cohort study. It assessed the role of noninvasive ECG markers, obtained simultaneously using a 24-h high-resolution digital ambulatory ECG system, in predicting severe cardiac events, such as lethal arrhythmias and cardiac mortality in high-risk patients. Recruitment was performed between April 2012 and March 2015. The inclusion and exclusion criteria used have been reported previously (Kinoshita et al., 2020). Briefly, the JANIES study included patients with structural or idiopathic cardiac disorders who underwent 24-h high-resolution digital ambulatory monitoring (Figure 1). The exclusion criteria were as follows: hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, persistent atrial fibrillation or flutter, right or left bundle branch block and intraventricular conduction delay, permanent pacing, second- or third-degree atrioventricular block, nonsimultaneous measurement of ECG markers, and patient dropout (missing follow-up data or unknown cause).

Study population. Patients were enrolled from April 2012 to March 2015. Follow-up data collection was performed every 6 months until September 2015. CKD, chronic kidney disease; JANIES-CKD, The Japanese Noninvasive Electrocardiographic Risk Stratification of Sudden Cardiac Death in Chronic Kidney Disease; SHD, structural heart disease

In this study, we exclusively selected patients with CKD for analysis. CKD was defined according to the Kidney Disease: Improving Global Outcomes (KDIGO) 2012 classification (Levey et al., 2005). Overall, 183 patients (mean age 70 years, 127 men) were eligible for analysis. Patient characteristics are presented in Table 1. The mean left ventricular mass index (LVMI) was 107 ± 37.1 g/m2 (reference values were <90 g/m2 for men and <84 g/m2 for women) (Daimon et al., 2008). Approximately 75% of patients in this study had left ventricular hypertrophy (LVH). Renal replacement therapy (e.g., hemodialysis) was required in 19 patients (10.3%). The major causes of heart disease were ischemic heart disease and chronic heart failure. Other clinical characteristics, including the prevalence of hypertension, dyslipidemia, diabetes mellitus, serum brain natriuretic peptide levels, New York Heart Association (NYHA) functional class, and medication use, are detailed in Table 1.

TABLE 1. Baseline characteristics of the study patients (n = 183) Characteristics Age, years 70 (61.0, 77.0) Male sex 127 (69) Hypertension 133 (73) Dyslipidemia 109 (60) Diabetes mellitus 82 (45) Estimated GFR, ml/min/1.73 m2 Mean 42.4 ± 18.9 45–59 (stage 3a) 94 (51) 30–44 (stage 3b) 44 (24) 15–29 (stage 4) 24 (13) <15 (stage 5) 21 (11) LVEF, % 55.1 (39.8, 68.0) LVDd, mm 51.2 ± 9.6 LVM, g 181.4 (146.2, 224.3) LVMI, g/m2 107.5 ± 37.1 BNP, pg/dl 199.3 (65.9, 397.6) NYHA functional class I 123 (67) II 38 (21) III 15 (8) IV 8 (4) Heart disease Ischemic heart disease 82 (45) Chronic heart failure 71 (39) Dilated cardiomyopathy 9 (5) Hypertensive heart disease 11 (6) Unknown cardiomyopathy 10 (5) ICD implantation 16/183 (9) RRT 19/183 (10) Medications β-blocker 134 (73) RAS inhibitor 124 (68) Ca-channel blocker 56 (31) Diuretic 58 (32) Amiodarone 27 (15) Class Ia or Ic antiarrhythmic 5 (3) Note Values are expressed as n (%), mean ± standard deviation or median (interquartile range). Abbreviations: BNP, brain natriuretic peptide; GFR, glomerular filtration rate; HHD, hypertensive heart disease; ICD, implantable cardioverter defibrillator; LVDd, left ventricular diastolic diameter; LVEF, left ventricular ejection fraction; LVM, left ventricular mass; LVMI, left ventricular mass index; NYHA, New York Heart Association; RAS, renin–angiotensin–aldosterone system; RRT, renal replacement therapy.

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Competent Authorities and Ethics Committees of the participating centers. Written informed consent was obtained from all patients. The JANIES study was approved by the Ethics Committee of Toho University Omori Medical Center (approval number 23-135) and registered in the UMIN Clinical Trials Registry (UMIN000007683).

2.2 Ambulatory ECG recordings and echocardiography assessments

All ECGs were recorded using a FM-180 Digital Holter ECG Recorder (Fukuda Denshi Co., Ltd.) and analyzed using a high-resolution digital ambulatory 24-h ECG system (SCM 8000, Fukuda Denshi Co., Ltd.). Routine ambulatory ECG parameters, including NSVT, ambulatory-based LPs (a-LPs), and HRT, were also analyzed. The presence of NSVT was defined as more than three consecutive ventricular premature contractions (VPCs) at >100 beats/min, as previously reported (Lin et al., 2016). All participants underwent 24-h ECG assessment during their ordinary daily activities. Additionally, ECGs were obtained to assess cardiac structure and function; the left ventricular ejection fraction (LVEF), left ventricular end-diastolic diameter, left ventricular mass (LVM), and LVMI were measured.

2.3 Measurement of ambulatory-based LPs

Electrocardiogram data for measurement of ambulatory-based LPs were obtained at a sampling rate of 1000 Hz and amplitude resolution using a 20-bit analog to digital converter. A three-pole (18 dB/octave) corrected bidirectional filter (infinite pulse response method) was used, with a bandpass ranging from 40 to 200 Hz. ECG data were filtered and ranged from 40 to 250 Hz. Orthogonal X, Y, and Z bipolar leads with silver–silver chloride electrodes (Magnerode®; Fukuda Denshi Co., Ltd.) were used for all recordings. Ambulatory-based LPs were automatically measured every 30 min over 24 h. Values were presented on a trend graph and evaluated according to three parameters: filtered QRS duration (fQRS), duration of low-amplitude signals <40 μV in the terminal filtered QRS complex (LAS40), and root mean square voltage of the terminal 40 ms in the filtered QRS complex (RMS40). Ambulatory-based LPs were considered positive on fulfillment of two of the following three criteria: fQRS > 135 ms, RMS40 < 15 μV, and LAS40 > 39 ms (Abe et al., 2012). The worst values for a-LPs (w-LPs) were calculated and used for analysis in this study according to a previously published method (Hashimoto et al., 2020; Kinoshita et al., 2020). W-LPs were defined as measurements detected when the RMS40 was the smallest over 24 h.

2.4 HRT measurements

Heart rate turbulence was measured according to a previously established protocol (Bauer et al., 2008) and recorded when more than one VPC occurred. HRT is characterized by two parameters, turbulence onset (TO) and turbulence slope (TS). TO captures the early-phase sinus rhythm acceleration after VPC, followed by TS, which captures the compensatory deceleration phase. TO was calculated as the shortening ratio of RR intervals immediately after the compensatory pause of the VPC. TS was calculated as the steepest regression over any five consecutive sinus rhythm RR intervals after VPC and within 15 sinus rhythm beats (Bauer et al., 2008). TO ≥ 0% and TS ≤ 2.5 ms/RR intervals were considered abnormal. HRT was identified when both TO and TS were abnormal. Isolated TO and TS abnormalities were not regarded as HRT, nor was the inability to calculate HRT due to the absence of VPC (Bauer et al., 2008).

2.5 Study endpoints and follow-up

The primary endpoint was the occurrence of lethal ventricular tachyarrhythmias, such as ventricular fibrillation (VF) or sustained ventricular tachycardia (SVT) and cardiac death. Shock delivery using an implantable cardioverter–defibrillator (ICD) and anti-tachycardia pacing for sustained VT were included as lethal ventricular tachyarrhythmias. The secondary endpoint was a composite of hospital admission due to heart failure, percutaneous coronary intervention (PCI), coronary artery bypass surgery (CABG), peripheral artery disease intervention, or aortic dissection. Causes of death were retrieved from medical or autopsy records and testimonies of the primary doctors or witnesses. The occurrence of lethal ventricular tachyarrhythmias was verified through ECG monitoring performed in the hospital or by 24-h Holter ECG recorded during hospitalization or inferred using an ICD. Follow-up data were collected at 6-month intervals until September 2015.

2.6 Statistical analyses

Data are presented as mean ± standard deviation for normally distributed continuous variables and as medians (interquartile range: 25th–75th percentile) for nonnormally distributed variables. The normality of the distribution was tested using the Shapiro–Wilk method. Patient characteristics were compared using the χ2 test for categorical variables, the Student's t-test for continuous and parametric data, and the Mann–Whitney test for nonparametric data. Cox univariate and multivariate regression analyses were performed to investigate associations between the endpoints and clinical parameters. Cardiac event-free survival rates were calculated for each NIEM using the Kaplan–Meier method, and differences in cardiac event-free survival rates were calculated using the log-rank test. Statistical analyses were conducted using SPSS software version 25 (IBM Corp). All tests were two-sided, and a p-value of .05 was considered significant.

3 RESULTS 3.1 Cardiac events during follow-up

During the mean follow-up period of 20.7 ± 11.1 months, 13 patients reached the primary endpoint of the study. Their characteristics are shown in Tables 2 and 3.

TABLE 2. Comparison of risk factors between the primary endpoint group and the event-free group Characteristics Primary endpoint (n = 13) Event free (n = 170) p-value Univariate analysis HR (95% CI) p-value Multivariate analysis HR (95% CI) p-value Age, years 70 (66.5, 81.0) 69.0 (61.0,77.0) .41 1.02 (0.98–1.08) .31 1.02 (0.98–1.08) .27 Male 8 (84.6) 118 (69.4) .25 1.06 (1.03–1.10) .001 1.07 (1.03–1.11) .001 Hypertension 8 (61.5) 125 (73.5) .35 0.56 (0.18–1.72) .31 Dyslipidemia 8 (61.5) 101 (59.4) .88 1.04 (0.34–3.19) .94 Diabetes mellitus 7 (53.8) 75 (44.1) .69 1.43 (0.48–4.26) .52 ICD 4 (30.8) 12 (7.1) .022 6.79 (1.98–23.26) .002 RRT 2 (15.3) 17 (10) .289 2.53 (0.53–11.95) .24 eGFR, (ml/min/1.73 m2) 32.5 ± 24.3 42.8 ± 18 .028 0.97 (0.94–0.99) .014 0.96 (0.93–1.00) .055 LVEF, % 38.0 (29.1, 52.9) 56.9 (41.0, 69.4) .023 0.95 (0.92–0.99) .004 0.98 (0.94–1.03) .44 LVM, g 205.0 (135.5, 236.4) 180.5 (147.8, 224.4) .81 1.00 (0.99–1.02) .98 LVMI, g/m2 108.7 ± 35.9 106.5 ± 37.6 .88 1.003 (0.98–1.029) .79 BNP, pg/dl 393.9 (220.0, 853.1) 174.3 (55.7, 335.1) .005 NYHA Ⅰ, Ⅱ 8 (61.5) 152 (89.4) Ref Ref NYHA Ⅲ, Ⅳ 5 (38.4) 18 (10.5) .003 3.94 (1.29–12.06) .016 β-blocker 11 (92.3) 123 (72.3) .34 RAS inhibitor 8 (61.5) 116 (68.2) .62 Amiodarone 4 (23.4) 23 (13.5) .096 w-LPs 10 (76.9) 69 (40.6) .011 6.04 (1.40–22.3) .007 3.18 (0.72–13.99) .13 HRT 8 (61.5) 68 (40.5) .13 3.01 (0.98–9.25) .54 NSVT 7 (53.8) 20 (11.8) <.001 8.72 (2.80–26.5) <.001 10.41 (2.81–38.51) <.0001 Note Values are expressed as n (%), mean ± standard deviation, median (interquartile range), or hazard ratio (95% confidence interval). Abbreviations: BNP, brain natriuretic peptide; CHF, chronic heart failure; CI, confidence intervals; DCM, dilated cardiomyopathy; GFR, glomerular filtration rate; HHD, hypertensive heart disease; HRT, heart rate turbulence; ICD, implantable cardioverter defibrillator; IHD, ischemic heart disease; LVDd, left ventricular dimension diameter; LVEF, left ventricular ejection fraction; LVM, left ventricular mass; LVMI, left ventricular mass index; NSVT, nonsustained ventricular tachycardia; RAS, renin–angiotensin–aldosterone system; ref, reference; RRT, renal replacement therapy; w-LP, worst value of ambulatory-based late potentials. TABLE 3. The characteristics of patients who developed the primary endpoint No Age/Sex eGFR (ml/min/m2) CKD stage Heart disease NYHA class EF (%) LVDD (mm) NIEP positive ICD RRT Time to event (days) SVT/VF Death 1 76/F 54.3 3a IHD 4 53.0 44.5 HRT 304 SVT/VF 2 49/M 53.9 3a DCM 4 33.7 63.2 w-LPs, NSVT + 126 SVT 3 56/F 48.9 3a DCM 3 27.9 70.1 w-LPs, HRT, NSVT + 495 SVT 4 86/M 44.8 3b HF 2 52.8 55.6 w-LPs, NSVT 978 Death (Heart failure) 5 86/M 34 3b IHD 2 37.0 49.0 w-LPs, HRT + 101 SVT 6 68/M 33 3b DCM 2 38.0 65.0 w-LPs, NSVT 630 Death (Heart failure) 7 68/M 27.2 4 DCM 3 30.3 64.9 w-LPs, NSVT 29 Death (Heart failure) 8 68/M 22 4 IHD 2 22.0 72.0 w-LPs, HRT, NSVT + 33 Death (Heart failure) 9 71/M 20.3 4 IHD 3 48.7 47.9 HRT, NSVT 264 Death (Heart failure) 10 70/M 14.6 5 IHD 2 74.7 37.2 w-LPs, HRT, NSVT 481 Death (Heart failure) 11 89/M 14.1 5 IHD 2 26.0 57.2 w-LPs, HRT 490 Death (Heart failure) 12 74/M 9.8 5 HF 2 47.1 56.8 none + 950 Death (Heart failure) 13 65/M 4.4 5 IHD 1 67.9 54.7 w-LPs, HRT + 384 Death (Heart failure) Abbreviations: CKD, chronic heart failure; DCM, dilated cardiomyopathy; EF, ejection fraction; eGFR, estimated glomerular filtration rate; F, female; HF, heart failure; HRT, heart rate turbulence; ICD, implantable cardioverter defibrillator; IHD, ischemic heart disease; M, male; NIEPs, noninvasive electrocardiographic parameters; NSVT, nonsustained ventricular tachycardia; NYHA, New York Heart Association; RRT, renal replacement therapy; SVT, sustained ventricular tachycardia; VF, ventricular fibrillation; w-LPs, worst value of ambulatory-based late potentials. 3.2 Association between ECG markers and endpoints 3.2.1 Primary endpoint

Univariate analysis (Table 2) revealed that sex, ICD implantation, estimated glomerular filtration rate (eGFR), LVEF, NYHA cardiac functional classification of Ⅲ or Ⅳ, w-LPs, and NSVT were significantly associated with the primary endpoint (p = .001, .002, .014, .004, .016, .007, and <.001, respectively). However, NSVT was the factor that correlated most significantly with the primary endpoint in multivariate analysis (hazard ratio [HR], 10.41; 95% confidence interval [CI], 2.81−38.51). Table 3 demonstrates the individual characteristics of patients who reached the primary endpoint. Six of 13 patients were categorized as having CKD stage 3 (patient No. 1–6); the remaining seven patients were categorized as having CKD stage 4 or 5 (patient No. 7–13). Severe arrhythmic events such as SVT/VF were documented in four of the six patients with CKD stage 3 (No. 1–3, 5). Moreover, two patients of these four were positive for both w-LP and NSVT. These two patients (No. 2 and 3) survived due to anti-tachycardia pacing or appropriate ICD shock delivery. However, all seven patients with CKD stage 4 or 5 died of heart failure. None of the patients developed serious arrhythmic events. The predictive ability of single and combined ECG risk markers is presented in Table 4. The highest positive predictive value (PPV) (47%), positive likelihood ratios (P-LRs) (11.4), predictive accuracy (PA) (92%), and lowest negative likelihood ratios (N-LRs) were obtained when both w-LPs and NSVT were present (HR, 18.22; 95% CI, 5.64−58.81; p < .0001). This combination remained significantly associated with the primary endpoint after adjusting for age, sex, eGFR, and LVEF (HR, 19.18; 95% CI, 4.97−74.08; p < .0001) (Table S1). Kaplan–Meier analysis revealed that the concurrence of w-LPs and NSVT positivity was associated with a significantly lower event-free survival rate than was the presence of either variable alone (log-rank, p < .0001) (Figure 2).