Author + information
- Received December 29, 2014
- Revision received January 26, 2015
- Accepted January 29, 2015
- Published online March 1, 2015.
- Saadia Sherazi, MD, MS∗∗ (, )
- Valentina Kutyifa, MD, PhD∗,
- Scott McNitt, MS∗,
- Mehmet K. Aktas, MD∗,
- Jean-Philippe Couderc, PhD∗,
- Benjamin Peterson, MD∗,
- Poul Erik Bloch Thomsen, MD†,
- Joseph Kautzner, MD, PhD‡,
- Arthur J. Moss, MD∗ and
- Wojciech Zareba, MD, PhD∗
- ∗Cardiology Division, Heart Research Follow-up Program, University of Rochester Medical Center, Rochester, New York
- †Department of Cardiology, Aalborg University Hospital, Aalborg, Denmark
- ‡Department of Cardiology, Institute for Clinical and Experimental Medicine, Prague, Czech Republic
- ↵∗Reprint requests and correspondence:
Dr. Saadia Sherazi, Heart Research Follow-up Program, Cardiology Division, University of Rochester Medical Center, 265 Crittenden Boulevard, Box 653, Rochester, New York 14642.
Objectives This study sought to evaluate the prognostic value of heart rate variability (HRV) for death or heart failure in patients with mildly symptomatic heart failure undergoing cardiac resynchronization therapy with a defibrillator (CRT-D).
Background There are limited data regarding the prognostic value of HRV as a means of identifying high-risk patients treated with CRT-D.
Methods We analyzed the relationship between pre-implant time-domain (SD of all normal-to-normal RR intervals [SDNN], SDs of averaged 5-min normal-to-normal RR intervals, root mean square of successive differences, and mean of the SDs of all normal-to-normal RR intervals for all 5-min segments of the entire recording), and frequency-domain (low-frequency power, very-low-frequency power [VLF], high-frequency power, low-frequency power/low-frequency power ratio) HRV parameters, and the end point of death or heart failure and death alone. Study subjects include 719 patients in normal sinus rhythm enrolled in MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial–Cardiac Resynchronization Therapy); outcomes of CRT-D patients with low HRV (lower tertile) were compared with CRT-D patients with preserved HRV (2 upper tertiles) and with patients receiving implantable cardioverter-defibrillators only.
Results During a mean 3.4 ± 0.9 years of follow-up, 124 patients reached the primary end point of death or heart failure, and 47 died. In multivariate analysis, low SDNN (≤93 ms) was associated with significantly higher risk of death or heart failure (hazard ratio [HR] 1.63 [95% confidence interval (CI): 1.12 to 2.36]; p = 0.010) and mortality (HR 2.10 [95% CI: 1.14 to 3.87]; p = 0.017) compared with higher SDNN (>93 ms). Similarly, low VLF (≤179 ms2) was associated with an increased risk of death or heart failure (HR 2.14 [95% CI: 1.46 to 3.13]; p < 0.001) and death alone (HR 2.49 [95% CI: 1.35 to 4.57]; p = 0.003). There was no significant difference in outcome between low HRV patients treated with CRT-D and patients receiving an implantable cardioverter-defibrillator only.
Conclusions Our findings indicate that autonomic dysfunction (quantified by low SDNN and low VLF) identified patients with no benefit or limited benefit from cardiac resynchronization therapy. Pre-implant HRV analysis might help in optimizing qualifications for this treatment.
Analysis of heart rate variability (HRV) is a reliable and reproducible technique for assessing autonomic balance in patients with cardiovascular disease. Autonomic dysfunction plays a significant role in the pathophysiology of heart failure; patients with heart failure have increased sympathetic activity, decreased vagal tone, and depressed baroreceptor responsiveness (1,2). Previous studies have suggested that HRV provides important prognostic information in patients with heart failure (3–5). However, its use as a means of identifying high-risk patients treated with cardiac resynchronization therapy with a defibrillator (CRT-D) is not well established.
The objective of the present study, conducted in patients enrolled in the CRT-D arm of MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial–Cardiac Resynchronization Therapy) (6), was 2-fold: 1) to evaluate the association between autonomic activity, assessed by using baseline HRV, before CRT-D implantation on the risk of death or heart failure; and 2) to determine whether HRV parameters could be used as prognostic markers before CRT-D implantation.
The design of the MADIT-CRT study has been reported previously (6). Briefly, 1,820 patients with ischemic cardiomyopathy (New York Heart Association functional class I or II) or nonischemic cardiomyopathy (New York Heart Association functional class II), left ventricular ejection fraction (LVEF) ≤0.30, and a wide QRS complex ≥130 ms were randomized to receive either CRT-D or an implantable cardioverter-defibrillator (ICD) alone in a 3:2 ratio. Screened patients were excluded from enrollment if they had an existing indication for CRT, New York Heart Association functional class III or IV in the 90 days before enrollment, an implanted pacemaker, coronary artery bypass graft surgery, percutaneous coronary intervention, or myocardial infarction within the 90 days before enrollment.
Patients in the CRT-D arm of the trial were requested to undergo 24-h Holter monitoring before device implantation. HRV analyses were performed in 719 patients in normal sinus rhythm. Death or heart failure was selected as the primary end point for the present study, consistent with the primary end point for the overall MADIT-CRT trial. The diagnosis of heart failure required signs and symptoms consistent with congestive heart failure that were responsive to intravenous decongestive therapy on an outpatient basis or an augmented decongestive regimen with oral or parenteral medications during an in-hospital stay. Using pre-specified criteria, adjudication of the end points was conducted by an independent mortality committee and by a heart failure committee that was unaware of study group assignments.
HRV study and analysis
Analysis of the 24-h Holter electrocardiogram recordings was performed by using the HScribe Mortara System (Mortara Instruments, Milwaukee, Minnesota). After automatic beat annotation (verified by a technician), the following time-domain HRV parameters were calculated on the entire 24-h recording: SD of all normal-to-normal RR intervals (SDNN), SD of averaged 5-min normal-to-normal RR intervals (SDANN), mean of the SDs of all normal-to-normal RR intervals for all 5-min segments of the entire recording (SDNNIX), and root mean square of successive differences (RMSSD) (7). At the beginning of the Holter recording, there was a 20-min supine resting period, from which a 5-min segment (middle of the 20-min period) was taken for analysis and computation of frequency-domain HRV parameters, as follows: low-frequency (LF) (0.04 to 0.15 Hz), very-low-frequency (VLF) (0.01 to 0.04 Hz), high-frequency (HF) (0.15 to 0.40), and the vagosympathetic balance (LF/HF) ratio according to the Mortara Super electrocardiogram program (8).
Baseline clinical characteristics and HRV parameters among CRT-D patients categorized according to the primary end point of death or heart failure were compared by using Kruskal-Wallis tests for continuous variables and the chi-square test for categorical variables. Patients were divided a priori into tertile groups according to the baseline HRV parameters: lowest tertile (T1), representing patients with depressed HRV, and 2 upper tertiles (T2 and T3), representing patients with preserved HRV. The probability of events was compared between these groups. The prognostic significance of HRV parameters was evaluated by comparing the log-likelihoods of the various multivariate Cox proportional hazards regression models each fitted with a different HRV parameter but all adjusted by using the same set of 6 covariates. The clinical covariates in the multivariate model were: left bundle branch block (LBBB), LVEF, female sex, glomerular filtration rate ≥60 ml/min/1.73 m2, history of heart failure hospitalization, and duration of QRS. Clinical covariate selection was conducted by using the best subset selection method. Possible interaction between HRV parameter and LBBB electrocardiogram pattern was checked with interaction terms in the multivariate models. Separate analyses were conducted comparing outcome of ICD patients with outcome of CRT-D patients stratified according to HRV parameters. A p value <0.05 was considered significant, and all tests were 2-sided. Analyses were performed by using SAS version 9.3 (SAS Institute, Inc., Cary, North Carolina).
Baseline clinical and HRV characteristics by the primary end point
A total of 719 patients with normal sinus rhythm were included in this study. There were 124 patients who had the primary end point of death or heart failure, and 47 patients died during a mean follow-up period of 3.4 ± 0.9 years. There were significant differences in baseline clinical covariates among those with or without death or heart failure (Table 1). Patients with death or heart failure were significantly older, less often female, and were less likely to have LBBB. They had more comorbid conditions such as diabetes, ischemic cardiomyopathy, and renal dysfunction. Patients with death or heart failure had a significantly lower baseline LVEF; however, baseline left ventricular end-systolic volume and left ventricular end-diastolic volume were similar. The use of medications (including beta-blockers and angiotensin-converting enzyme inhibitors/angiotensin receptor blockers) was similar between the groups. The mean values for the following HRV parameters were significantly lower in patients who had the primary end point: SDNNIX, VLF, LF, HF, and LF/HF ratio. Mean SDNN, SDANN, and root mean square of successive differences did not differ between patients with or without the primary end point.
Time-domain HRV parameters
CRT-D patients in the lowest tertile of SDNN (≤93 ms) at baseline had a significantly greater cumulative probability of death or heart failure compared with patients with SDNN in T2 and T3 (>93 ms). At 4 years, the cumulative probability of death or heart failure was 17% for patients with preserved SDNN versus 24% in CRT-D patients with low SDNN and 23% in patients who had an ICD only (p = 0.004) (Figure 1A). CRT-D patients with preserved SDNN (T2 and T3) had 5% mortality, whereas CRT-D patients with lower SDNN had a higher probability of mortality (13%), similar to patients with an ICD only (10%) (p = 0.011) (Figure 1B).
These findings were confirmed in the multivariate analysis. CRT-D patients with lower baseline SDNN (T1) had an increased risk of death or heart failure (hazard ratio [HR] 1.63 [95% confidence interval (CI): 1.12 to 2.36]; p = 0.010) and an increased risk of all-cause mortality (HR 2.10 [95% CI: 1.14 to 3.87]; p = 0.017) (Table 2) compared with CRT-D patients with higher SDNN. Among the time-domain HRV measurements, SDNNIX ≤37 ms (T1) was the best predictor of heart failure or death (HR: 1.78 [95% CI: 1.23 to 2.58]; p = 0.002), with the best fit in the multivariate model using the –2 Log L test value. However, the end point for all-cause mortality, SDNN (T1), was the best predictor among all time-domain HRV parameters. There was no interaction between LBBB and SDNN T1 for primary end point (p = 0.810) and death (p = 0.325).
When comparing low SDNN patients versus ICD patients, there was no significant difference in the risk of death or heart failure (HR: 1.11 [95% CI: 0.79 to 1.56]; p = 0.525) and no significant difference in death alone (HR: 1.17 [95% CI: 0.69 to 1.96]; p = 0.549). After adjustment for clinical covariates, patients with preserved HRV measured according to SDNN in T2 and T3 had a significantly lower risk of death or heart failure (HR: 0.66 [95% CI: 0.48 to 0.90]; p = 0.008) and death alone (HR: 0.55 [95% CI: 0.32 to 0.94]; p = 0.029) compared with ICD-only patients.
CRT-D patients with low VLF (T1) ≤179 ms2 at baseline had a significantly higher risk of death or heart failure compared with preserved VLF >179 ms2 (T2 and T3). At 4 years, the cumulative probability of death or heart failure was 14% in patients with preserved VLF and 28% in patients with low VLF, compared with 23% in patients with ICD only (p < 0.001) (Figure 2A). CRT-D patients with preserved VLF had a 5% low risk of death, compared with 14% in CRT-D patients with low VLF and 10% in patients with an ICD only (p = 0.003) (Figure 2B). Multivariate analysis found that patients with low baseline VLF were at a significantly increased risk of death or heart failure (HR: 2.14 [95% CI 1.46 to 3.13]; p < 0.001) and an increased risk of death (HR: 2.49 [95% CI: 1.35 to 4.57]; p = 0.003) (Table 2). Among all the frequency-domain parameters, low VLF (T1) was the predictor in the models that was the best fit for both the death or heart failure and death alone end points. Models including VLF T1 also had superior log-likelihood values compared with the time-domain parameters (SDNNIX or SDNN). There was no significant interaction between the presence of LBBB and VLF T1 with the primary end point (p = 0.257) and death only (p = 0.204).
When comparing low VLF patients with ICD patients, there was no significant difference in the risk of death or heart failure (HR: 1.18 [95% CI: 0.85 to 1.63]; p = 0.316) and no significant difference in death alone (HR: 1.24 [95% CI: 0.75 to 2.06]; p = 0.393). Patients with preserved HRV measured according to VLF in T2 and T3 had a significantly lower risk of death or heart failure (HR: 0.58 [95% CI: 0.42 to 0.80]; p = 0.001) and death alone (HR: 0.53 [95% CI: 0.30 to 0.92]; p = 0.024) compared with ICD-only patients.
Beyond the significant difference in outcome according to baseline VLF, CRT-D patients with a lower (T1 vs. T2 and T3) baseline LF, HF, and LF/HF ratio also had a significantly higher risk of death or heart failure. Lower baseline LF was also significantly related to an increased risk of death after adjustment for clinical covariates.
The primary findings of this MADIT-CRT substudy was that HRV analysis might help in identifying patients who would benefit from CRT. Preserved baseline HRV, as assessed by using time- and frequency-domain analyses, was associated with a lower risk of death or heart failure among patients with mild to moderate heart failure who were treated with CRT-D. A number of the HRV measures, especially SDNN and VLF, were found to be predictive of death or heart failure. Importantly, CRT-D patients with low HRV did not have a better outcome than ICD-only patients, indicating that HRV analysis performed before implantation might optimize qualifications for CRT-D therapy.
The SDNN taken from routine 24-h Holter monitoring, as well as the VLF computed from supine 5-min recordings, performed similarly in identifying CRT-D patients with improved outcomes. Therefore, a 24-h Holter recording should be performed before a decision is made regarding CRT to determine whether the given patients have depressed or preserved HRV and whether patients with preserved SDNN or VLF values should be considered as high priority for CRT.
Prognostic value of HRV in CRT-D patients
We found that both time- and frequency-domain measures of HRV were predictive of the end point of death or heart failure and death alone. We found that the risk prediction for the end point of death or heart failure was strongest in the presence of low SDNNIX, whereas low SDNN was the strongest predictor of death. Similarly, patients with low VLF were at particularly high risk of adverse cardiac events. The autonomic variables we analyzed provided independent prognostic information beyond clinical variables, including LBBB, female sex, LVEF, glomerular filtration rate ≥60 ml/min/1.73 m2, and QRS duration, which are known to influence prognosis and identification of responders to CRT (9–12).
SDNN is 1 of the most widely studied and easy to measure parameters of autonomic function (3,13–15). Similar to our findings, the UK-HEART (United Kingdom Heart Failure Evaluation and Assessment of Risk) trial found that low SDNN was associated with a high risk of death in patients with chronic heart failure (3). The annual mortality rate ranged from 17.9% in the lower tertile of SDNN (<93 ms) to 6.2% in the middle tertile (93 to 130 ms) and 5.5% in the upper tertile (>130 ms). Coincidently, the lower tertile for SDNN was <93 ms, which was similar to the lower tertile cutoff in our study.
As shown in Figure 2B, the mortality of patients with low VLF was higher than the annual mortality of patients who were implanted with an ICD alone. The exact underlying mechanism for the association between depressed HRV and poor outcome is not well understood. The VLF power spectral component of HRV has been shown to be an independent predictor of all-cause mortality and sudden death in patients with acute myocardial infarction (16,17). Similarly, VLF was identified as an independent risk predictor in a small cohort of patients with acute decompensated heart failure (18). La Rovere et al. (19) have shown that autonomic markers (including VLF, LF, and turbulence slope) provide incremental value over traditional clinical risk predictors in patients with symptomatic heart failure. VLF power accounts for a significant portion of the total power in the 24-h heart rate spectrum, and it is believed to be a gauge of several important mechanisms, including thermoregulation, fluctuation in the activity of renin-angiotensin systems, and the function of peripheral chemoreceptors (19–22). Other physiological mechanisms that have been proposed for VLF oscillations include slow respiratory patterns, physical activity, and parasympathetic mechanisms (21). Low VLF may reflect a reduction in the cardiac defense response toward external stress, which in turn may be associated with increased risk for adverse cardiac events in patients with heart failure (18).
HRV and CRT
To our knowledge, the present study is the first to describe the relationship between baseline HRV measurements and subsequent cardiac events in a large population of patients with heart failure treated with CRT. This therapy promotes cardiac reverse remodeling, improving clinical outcomes and reducing the risk of adverse cardiac events (23,24).
HRV analysis helps to identify patients who benefit from CRT-induced changes in reverse remodeling. There is emerging evidence that CRT may be associated with acute improvements in the measure of HRV. Fantoni et al. (25) have shown that in patients with advanced heart failure, CRT could significantly modify the heart rate profile and SDANN. Device-based HRV was analyzed in this study, and the authors found that the 2-year event-free survival rate was significantly lower in patients who had an increase in HRV 4 weeks after CRT initiation. In the study by Landolina et al. (26), which included 509 patients, a reduced HRV monitored by the implanted device at the first week after CRT implantation was the sole predictor of major cardiovascular events among several baseline variables.
The present study was a subanalysis of the main trial, with HRV measurements pre-specified in the MADIT-CRT protocol. MADIT-CRT was not designed to select patients according to HRV results. Patients in the ICD arm did not undergo Holter monitoring. The Holter analyses were performed only in patients receiving CRT-D.
In patients with heart failure treated with CRT, a variety of autonomic function measures can provide independent prognostic information. HRV highlights the presence of pathological mechanisms (autonomic dysfunction) that are associated with adverse clinical outcomes in patients with mildly symptomatic heart failure that are not explained by traditional risk factors. The use of baseline HRV measures may be considered for purposes of risk-stratifying patients before consideration of CRT; patients with preserved HRV might benefit more from CRT than patients with low HRV.
COMPETENCY IN MEDICAL KNOWLEDGE: Preserved HRV, as assessed by using time- and frequency-domain analyses, is associated with a lower risk of death or heart failure among patients with mild to moderate heart failure who are treated with CRT-D.
COMPETENCY IN PATIENT CARE: Among patients with mild to moderate symptoms of heart failure, LBBB, and LVEF <30%, low HRV (particularly SDNN and VLF before CRT implantation) predicts higher risk of future death or heart failure events. This simple, readily available tool (SDNN is automatically computed in each Holter system) can be used in patient care to identify patients at higher risk of future events after optimization of both medical and device therapy for mildly symptomatic heart failure.
TRANSLATIONAL OUTLOOK: Although depressed HRV is well recognized in patients with heart failure, the clinical usefulness of this testing is limited. Our results indicate that this test is linked to optimized therapy and should become part of the routine Holter-based assessment of patients considered for CRT.
MADIT-CRT was supported by a research grant from Boston Scientific to the University of Rochester School of Medicine and Dentistry. Dr. Kautzner is on the advisory board for Boston Scientific, Biosense Webster, Medtronic, and St. Jude Medical. All other authors have reported that they have no other relationships relevant to the contents of this paper to disclose. Drs. Sherazi and Kutyifa contributed equally to this work.
- Abbreviations and Acronyms
- confidence interval
- cardiac resynchronization therapy
- cardiac resynchronization therapy with a defibrillator
- high-frequency power
- hazard ratio
- implantable cardioverter-defibrillator
- left bundle branch block
- low-frequency power
- left ventricular ejection fraction
- SD of averaged 5-minute normal-to-normal RR intervals
- SD of all normal-to-normal RR intervals
- mean of the SDs of all normal-to-normal RR intervals for all 5-min segments of the entire recording
- lowest tertile
- T2 and T3
- 2 upper tertiles
- very-low-frequency power
- Received December 29, 2014.
- Revision received January 26, 2015.
- Accepted January 29, 2015.
- American College of Cardiology Foundation
- Francis G.S.,
- Benedict C.,
- Johnstone D.E.,
- et al.
- Nolan J.,
- Batin P.D.,
- Andrews R.,
- et al.
- Woo M.A.,
- Stevenson W.G.,
- Moser D.K.,
- Middlekauff H.R.
- ↵(1996) Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation 93:1043–1065.
- Goldenberg I.,
- Moss A.J.,
- Hall W.J.,
- et al.
- Arshad A.,
- Moss A.J.,
- Foster E.,
- et al.
- Zareba W.,
- Klein H.,
- Cygankiewicz I.,
- et al.
- La Rovere M.T.,
- Bigger J.T. Jr..,
- Marcus F.I.,
- Mortara A.,
- Schwartz P.J.
- Bigger J.T. Jr..,
- Fleiss J.L.,
- Rolnitzky L.M.,
- Steinman R.C.
- Bigger J.T. Jr..,
- Fleiss J.L.,
- Steinman R.C.,
- Rolnitzky L.M.,
- Kleiger R.E.,
- Rottman J.N.
- Hadase M.,
- Azuma A.,
- Zen K.,
- et al.
- Malliani A.,
- Pagani M.,
- Lombardi F.,
- Cerutti S.
- Taylor J.A.,
- Carr D.L.,
- Myers C.W.,
- Eckberg D.L.
- Solomon S.D.,
- Foster E.,
- Bourgoun M.,
- et al.
- Barsheshet A.,
- Wang P.J.,
- Moss A.J.,
- et al.
- Fantoni C.,
- Raffa S.,
- Regoli F.,
- et al.