Prevalence and prognostic association of ventricular arrhythmia in non-ischaemic heart failure patients: results from the DANISH trial

Abstract

Aims

Improved risk stratification to identify non-ischaemic heart failure patients who will benefit from primary prophylactic implantable cardioverter-defibrillator (ICD) is needed. We examined the potential of ventricular arrhythmia to identify patients who could benefit from an ICD.

Methods and results

A total of 850 non-ischaemic systolic heart failure patients with left ventricle ≤35% and elevated N-terminal pro-brain natriuretic peptides had a 24-h Holter monitor recording performed. We examined present non-sustained ventricular tachycardia (NSVT), defined as ≥3 consecutive premature ventricular contractions (PVCs) with a rate of ≥100/min, and number of PVCs per hour stratified into low (<30) and high burden (≥30) groups. Outcome measures were overall mortality, sudden cardiac death (SCD), and cardiovascular death (CVD). In total, 193 patients died, 49 from SCD and 125 from CVD. Non-sustained ventricular tachycardia (365 patients) was significantly associated with increased all-cause mortality [hazard ratio (HR) 1.47; 95% confidence interval (CI) 1.07–2.03; P = 0.02] and to CVD (HR 1.89; CI 1.25–2.87; P = 0.003). High burden PVC (352 patients) was associated with increased all-cause mortality (HR1.38; CI 1.00–1.90; P = 0.046) and with CVD (HR 1.78; CI 1.19–2.66; P = 0.005). There was no statistically significant association with SCD for neither NSVT nor PVC. In interaction analyses, neither NSVT (P = 0.56) nor high burden of PVC (P = 0.97) was associated with survival benefit from ICD implantation.

Conclusion

Ventricular arrhythmia in non-ischaemic heart failure patients was associated with a worse prognosis but could not be used to stratify patients to ICD implantation.

Introduction

The role of primary prophylactic implantation of implantable cardioverter-defibrillator (ICD) in patients with non-ischaemic heart failure (HF) remain less clear than for ischaemic HF. Several studies have investigated the impact of treatment with ICD on mortality and prognosis in these patients, but no single trial has demonstrated a clear survival benefit from ICD implantation.^1–4

Ventricular ectopic activity may indicate an increased risk of malignant arrhythmia,⁵ and consequently a potentially protective effect from ICD implantation. The aims of this study were to describe the amount of non-sustained ventricular tachycardia (NSVT) and premature ventricular contractions (PVCs) in the setting of non-ischaemic systolic HF and to examine whether either arrhythmia could identify high-risk patients who might benefit from a prophylactic ICD implantation.

Methods

Trial overview

This is a Holter substudy to the Danish Study to Assess the Efficacy of ICD’s in Patients with Non-Ischaemic Systolic Heart Failure on Mortality (DANISH), a randomized, multicentre study carried out at five centres in Denmark from February 2008 to June 2016. The DANISH study enrolled a total of 1116 patients with non-ischaemic systolic HF, randomized to either standard medical care or standard medical care with the addition of a primary prophylactic ICD.

The DANISH study protocol, available at NEJM.org, was approved by the regional scientific ethics committee for the Capital Region of Denmark (H-D-2007-0101) and the Danish Data Protection Agency. The trial was also registered at http://www.clinicaltrials.gov. Unique identifier: NCT00542945. The trial was performed in accordance with the principles of the Declaration of Helsinki. All patients were enrolled only after providing written informed consent. The data underlying this article will be shared on reasonable request to the corresponding author.

Study population

The DANISH study’s inclusion and exclusion criteria of have been described previously.⁶ In brief, included patients had to be New York Heart Association (NYHA) class II–IV (NYHA II–IV) patients could be included if cardiac resynchronization therapy (CRT) was planned), had to have a documented left ventricular ejection fraction (LVEF) ≤35% and an elevated N-terminal pro B-type natriuretic peptide (NT-proBNP) (>200 pg/mL), and they had to be on optimal medical therapy [maximal tolerable angiotensin-converting enzyme inhibitors (ACE-I)/angiotensin II receptor blocker (ARB), a beta-blocker and a mineralocorticoid receptor antagonist (MRA)].

Holter recordings

All Holter recordings were digitally transferred to the Holter Laboratory, University of Copenhagen, Amager Hvidovre Hospital. For this study, the 7-lead Rozinn RZ153+ recorder (ScottCare) was used. All recordings were scheduled to last 24-h, but periods with excessive noise were removed. The analysis of the Holter recordings was performed by two experienced technicians and was supervised by cardiologist with arrhythmia expertise. The technicians and the arrhythmia expert were blinded to ICD randomization.

Holter recordings were performed in a total of 931 patients at baseline, immediately after randomization before ICD implantation. For logistical reasons, Holter recordings were not performed on the remaining 185 patients enrolled in the main study. Another 81 patients were excluded from this study because of poor recording quality, leaving a total of 850 eligible patients (see Supplementary material online, Figure S1). There was no difference in outcomes between included and excluded patients (see Supplementary material online, Table S1).

Endpoints and variables of interest

The primary endpoint was death from any cause, while secondary endpoints were sudden cardiac death (SCD) and cardiovascular death (CVD). All endpoints were adjudicated by an endpoint classification committee (see the DANISH Supplementary material online, Appendix for further information).⁷

Non-sustained ventricular tachycardia was defined as a minimum of three consecutive PVC with a rate of ≥100/min and a duration shorter than 30 s.

We considered PVC burden as a categorical variable: we divided the total number of PVCs during the recording by the number of analysis hours, and then separated patients into a low-burden group (<30 PVCs per hour) and a high-burden group (≥30 PVCs per hour). This clustering was based upon the European Society of Cardiology guidelines and the Copenhagen Holter Study.^8,9

Statistical analyses

Descriptive statistics of continuous data are presented with a median and interquartile range (IQR), while categorical data are presented with total number and percentage. Baseline characteristics were compared by using Wilcoxon’s test for continuous variables and by χ² test for categorical variables.

Proportional-hazard assumption was tested by Schoenfeld residuals. We used Cox Regression models to analyse associations with outcomes. In multivariable analyses, we adjusted for age, sex, diabetes, CRT, estimated glomerular filtration rate (eGFR), syncope history, family history of sudden death, LVEF, NT-proBNP, and NYHA class. The Cox regression model was also used to test for interactions between NSVT or PVC and ICD implantation.

Results are graphically presented with cumulative incidence plots and forest plots, and with a hazard ratio (HR) and 95% confidence interval (CI).

All statistical work was performed using the SAS Statistical Software Program Version 9.4. Two-sided P-values of 0.05 or less were considered statistically significant.

Results

Patients were included from 7 February 2008 to 30 June 2014. The median follow-up time was 4 years and 11 months (IQR 3 years and 4 months), the median age was 63 years (IQR 56–70), and 27% were women. The median Holter recording analysis time was 23 h and 59 min (IQR 12 min).

In the study population, 94% of the patients received beta blockers, while 97% received either ACE-I or ARB and 58% received an MRA. Furthermore, 51% received an ICD (see Table 1).

The majority of patients had ventricular ectopic activity during the recording; we observed PVC in 825 (97%) of the patients.

Non-sustained ventricular tachycardia

Of the 850 patients, 365 (43%) had at least one run of NSVT. Patients with NSVT were more often males, had lower systolic blood pressure, lower LVEF, shorter QRS duration, higher levels of NT-proBNP, longer HF duration, more often had diabetes mellitus and were less likely to receive a CRT (Table 1).

In the NSVT group, 111 patients died (30%), vs. 82 of patients (17%) without NSVT. When analysed in a univariable model, NSVT was a strong predictor of death from any cause (HR 1.86; CI 1.40–2.47; P < 0.0001), which remained significant after multivariable adjustment (HR 1.47; CI 1.07–2.03; P = 0.02) (see Table 2, Figures 1 and 2).

A total of 80 patients with NSVT (22%) and 45 from the group without (9%) died of CVD. NSVT was strongly associated with CVD in both univariable (HR 2.43; CI 1.69–3.50; P < 0.0001) and multivariable analyses (HR 1.89; CI 1.25–2.87; P = 0.003) (see Table 2 and Figure 2).

Sudden cardiac death occurred in 49 patients of the entire study population (6%), 29 patients in the NSVT group (8%) and 20 patients in the non-NSVT group (4%). In univariable analysis, NSVT was a significant predictor of SCD (HR 1.97; CI 1.12–3.47; P = 0.02), but not in the adjusted multivariate analysis (HR 1.78; CI 0.96–3.27; P = 0.07) (see Table 2 and Figure 2).

In the interaction analyses, we found no association between NSVT and ICD effect on death from any cause (P-value for interaction = 0.56). Additionally, we found no significant interaction between NSVT and the effect of ICD on CVD (P-value for interaction = 0.07) nor with SCD (P-value for interaction =0.09) (see Table 3 and Figure 4). However, in subgroups, we did see a reduction of CVD (HR 0.50; CI 0.27–0.93; P = 0.03) and SCD (HR 0.23; CI 0.08–0.69; P = 0.009) in patients without NSVT (see Table 3).

Premature ventricular contraction burden

The low-burden group consisted of 498 patients (59%), while the high-burden group consisted of 352 patients (41%). At baseline the high-burden group was older, consisted more of men, had shorter QRS duration, higher levels of NT-proBNP, had longer duration of HF, had more frequently diabetes, less frequent CRT and a higher mean heart rate (see Supplementary material online, Table S2).

In the low-burden group, 88 patients (18%) died, while 105 patients (30%) died in the high-burden group. In univariable analysis, high-burden PVC was associated with increased all-cause mortality (HR 1.78; CI 1.34–2.36; P < 0.0001), and remained associated with increased all-cause mortality after adjustment (HR 1.38; CI 1.00–1.90; P = 0.046) (see Table 2, Figures 1 and 3).

A total of 50 patients (10%) died from CVD in the low-burden group and 75 patients (21%) in the high-burden group. In univariable analyses, the high-burden group was associated with a worse prognosis (HR 2.24; CI 1.57–3.20; P < 0.0001). After adjustment, the association was still significant (HR 1.78; CI 1.19–2.66; P = 0.005) (see Table 2 and Figure 3).

The endpoint SCD occurred in a total of 24 low-burden patients (5%), and in 25 high-burden patients (7%). High-burden PVC was not statistically associated with SCD in neither univariate (HR 1.56; CI 0.89–2.72; P = 0.12) nor multivariable analyses (HR 1.51; CI 0.80–2.86; P = 21) (see Table 2 and Figure 3).

In the interaction analyses with ICD implantation, we found no overall association of effect on neither all-cause mortality (P-value for interaction = 0.47), CVD (P-value for interaction = 0.66) or SCD (P-value for interaction = 0.84) (see Table 4 and Figure 4). When looking at the subgroups individually, ICD implantation was associated with a mortality reduction of SCD in the high-burden group (HR 0.44; CI 0.12–0.98; P = 0.04), but not in any other group (see Table 4).

Discussion

In patients with non-ischaemic systolic HF, ventricular ectopic activity was substantial and associated with a poor prognosis. However, we did not find that the degree of ventricular ectopy was associated with a benefit of ICD implantation.

The prevalence of ventricular ectopy has been studied in other populations. In the Prophylactic Defibrillator Implantation in Patients with Nonischemic Dilated Cardiomyopathy trial (DEFINITE), 91% of the patients had NSVT, while 78% had >10 PVC/h.³ In the Amiodarone or an ICD for congestive heart failure study (SCD-HeFT), 23% of the patients had NSVT.⁴

Sajadieh et al.⁹ studied ventricular arrhythmia in apparently healthy subjects aged 55 years or older. Looking at the PVC burden, they found that 8% of their population belonged in the group corresponding to our high-burden group, and 92% belonged in the group corresponding to our low-burden group. Similar to our results, Sajadieh et al. found that high ectopic activity (>30 PVCs/h) was associated with a worse prognosis, which was similar to our results. In addition, the presence of more than five NSVT episodes within the first 6 months after ICD implantation has been found to be an independent predictor of cardiac mortality.¹⁰ However, it is not completely understood why ventricular ectopy is associated with a worse prognosis, but recent studies have found that frequent PVCs are associated with incident HF, chronic HF, worsening of HF symptoms and increased mortality.^11,12 One theory that could explain this is that frequent ventricular ectopy leads to increased left ventricle dysfunction and reduction in LVEF.^11,13,14

In our study, the high-burden group made up 41% of the entire study population, and 43% of the patients had NSVT. Looking at baseline, both the NSVT population and the high-burden PVC population did show a higher a priori risk, with both groups being more frequently males, having higher levels of NT-proBNP and had longer duration of HF. Thus, their progression of HF could be more advanced, leading to increased mortality. However, after adjustment for age, sex, diabetes, CRT, eGFR, syncope history, family history of sudden death, LVEF, NT-proBNP, and NYHA class, both NSVT and high-burden PVC remained significantly associated with all-cause mortality and CVD. The role of ventricular arrhythmias in HF patients is therefore not completely clear. However, both NSVT and high-burden PVC seem to be indicators of a worsened prognosis in HF patients. In our patients, there was an association between increasing NT-proBNP levels and both NSVT and PVCs (see Supplementary material online, Tables S3 and S4). Although these results do not describe the nature of the relationship, these findings underline a potential relationship between worsened HF and ventricular ectopic activity.

Our interaction analyses with ICD implantation did not reveal any association with prognosis based on the presence of neither NSVT nor PVC burden. Looking at the individual subgroups, ICD implantation was associated with a significantly lower risk of CVD and SCD in the non-NSVT patients. In the high-burden PVC group, we also saw a significant reduction in SCD. The results from the DANISH study showed that ICD implantation did reduce the risk of SCD, but not all-cause mortality or CVD. The reasons behind this are not clear, but the DANISH study indicated that younger patients might have a survival benefit from ICD implantation, and that the entire study population therefore might be too old to show an effect from ICD implantation. Another theory is, that the population simply has HF that is too advanced, and the patients are therefore more prone to die from other causes than ICD treatable malignant arrhythmias. Our baseline data indicated that the non-NSVT patients had less severe HF, which might in part explain the positive effect of a prophylactic ICD implantation in this subgroup.

Additionally, we carried out a univariable analysis of respectively NSVT and PVC the endpoints ATP and appropriate shocks. The association was highly significant, which indicates that both NSVT and PVC are valid risk markers of electrical instability. However, in the light of overall neutral results from the DANISH study, the ICD unit does not alter the prognosis, appropriate therapies given or not. For additional data, please see Supplementary material online, Table S5. However, these are subgroup analyses and the results should be interpreted with caution.

Goldberger et al.¹⁵ evaluated the 12 most commonly reported risk factors of arrhythmic events in non-ischaemic dilated cardiomyopathy. These risk factors included baroreflex sensitivity, heart rate turbulence, heart rate variability, left ventricular end-diastolic dimension, electrophysiological study outcome, NSVT, left bundle branch block, signal-averaged electrocardiogram parameters, fragmented QRS, QRS-T angle, and T-wave alternans. The study demonstrated that several of these established risk factors had high sensitivity, but low specificity. Since non-ischaemic HF include various geneses the cause of the arrhythmic event might be multifactorial. Therefore, it appears to be unlikely that any established risk factor alone could predict beneficial outcomes of prophylactic ICD implantation, as shown in the study by Goldberger et al.

Instead, these risk factors might be used as an a priori stratification for further examination. In a study by Gatzoulis et al.,¹⁶ post-myocardial infarction patients with preserved LVEF were grouped based on the presence of non-invasive risk-factors (PVCs, NSVT, late potentials, prolonged QTc, increased T-wave alternans, reduced heart rate variability, and abnormal deceleration capacity with abnormal turbulence). If at least one of these were present, the patient underwent programmed ventricular stimulation. If sustained monomorphic ventricular tachycardia or ventricular flutter or polymorphic ventricular tachycardia were induced, an ICD was offered. A total of nine patients with at least one non-invasive risk-factor and inducible ventricular arrhythmia had a major arrhythmic event during follow-up, compared to none in the group without non-invasive risk-factors and the group with at least one non-invasive risk factor but without inducible ventricular arrhythmia. This resulted in 100% sensitivity, and a specificity of 93.8%. As mentioned, the study was performed on patients with preserved LVEF and after a myocardial infarction, and hence the results cannot be transferred to non-ischaemic HF patients. However, this does suggest a potential future use of non-invasive risk-factors, including NSVT and PVCs.

Limitations

Our study has several limitations. Firstly, this was a substudy to the DANISH trial which presented overall neutral results and was consequently not specifically powered to investigate the effect of arrhythmia load. Any conclusions should therefore be drawn with caution, and our results should be considered as hypothesis generating.

Secondly, event rates were low, especially the risk of SCD where ICD implants are considered efficient in preventing death. Thirdly, baseline comparison revealed residual confounding, and despite multivariable analyses, this might still affect the results.

A technical limitation was the challenge for the Holter device to detect ventricular runs when a bundle branch block was present. This was also the case in subjects with high levels of pacing. Therefore, some runs of NSVT may have been missed. However, all Holter recordings have been reviewed by an experienced cardiologist to keep this to a minimal level.

It could also be debated if a single 24-h Holter recording is sufficient to categorize a patient’s arrhythmia burden. With circadian changes and most likely also arrhythmic differences over longer periods of time, several recordings would make a more solid foundation for stating a patient’s arrhythmic burden. Nevertheless, a 1-day recording is a very common clinical practice, and our method therefore mirrors daily procedure, strengthening the clinical use of our results.

Conclusion

In conclusion, a high ventricular arrhythmia burden in non-ischaemic HF patients indicate a worsened prognosis; however, these arrhythmia analyses cannot be used to stratify patients to ICD implantation.

Supplementary material

Supplementary material is available at Europace online.

Funding

This study was supported by the Danish Heart Foundation [17-R116-A7647-22064] and by the Hospital’s Independent Research Funds, Copenhagen University Hospital Amager and Hvidovre, Denmark. Neither funds have had any influence on the design, conduct, analyses or publications of the study.

Conflict of interest: J.J.T. discloses lecture fees from Novartis and is a member of the Pfizer advisory board. L.K. is speaker honoraria from Novartis, AstraZeneca and Boehringer. J.C.N. discloses a grant from the Novo Nordisk Foundation. E.K. discloses consulting fees from Edwards. A.B. discloses consulting fees from Bayer, Boehringer-Ingelheim, Bristol-Myers Squibb and MDS, research grant s from RSD/Region Sjælland and Theravance and travel grants from Biotronik. J.H.S. is a member of Medtronic advisory boards and discloses speaker honoraria and research grants from Medtronic, in addition to research grants from Gilead. U.D. discloses Advisory Board membership and a research grant from Boehringer-Ingelheim, Advisory Board Merck, and speaker fees Pfizer and BAYER. All reported conflicts of interests are unrelated to the work in this paper.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author

References