Author + information
- Received September 21, 2015
- Accepted October 22, 2015
- Published online June 1, 2016.
- Aditya Saini, MDa,
- Munish Kannabhiran, MDa,
- Pratap Reddy, MDa,
- Rakesh Gopinathannair, MD, MAb,
- Brian Olshansky, MDc and
- Paari Dominic, MDa,∗ ()
- aDivision of Cardiology, Department of Medicine, and Center for Cardiovascular Diseases and Sciences, Louisiana State University Health Sciences Center- Shreveport, Shreveport, Louisiana
- bDivision of Cardiovascular Medicine, University of Louisville, Louisville, Kentucky
- cCardiovascular Medicine, University of Iowa, Iowa City, Iowa
- ↵∗Reprint requests and correspondence:
Dr. Paari Dominic, Division of Cardiology, Department of Medicine, LSU Health Sciences Center, 1501 Kings Highway, Shreveport, Louisiana 71103.
Objectives This study sought to study the effect of echocardiographic response to cardiac resynchronization therapy (CRT) on ventricular arrhythmias (VA). The effect of CRT-defibrillator on sustained VA was compared with implantable cardioverter-defibrillator (ICD)-only therapy.
Background CRT is an effective adjunctive therapy in selected patients with advanced congestive heart failure, but its effect on VA remains controversial.
Methods PubMed was searched to identify studies. For primary comparison, studies reporting incidence of VA in patients with congestive heart failure with CRT compared with ICD were included. For secondary comparison, studies reporting incidence of VA in echocardiographic responders compared with nonresponders were included. Studies reporting incidence of VA in CRT nonresponders before and after CRT upgrade from ICD were assessed for the third comparison. Inverse variance method in a random-effects model was used to combine effect sizes.
Results Thirteen studies (4,631 subjects) were included in the primary meta-analysis. Patients with CRT had a significantly lower incidence of VA compared with patients with ICD only (odds ratio: 0.754; confidence interval: 0.594 to 0.959). Thirteen studies (n = 3,667) were included in the meta-analysis of VA in CRT responders versus nonresponders. Responders had a significantly lower risk of VA (odds ratio: 0.436; confidence interval: 0.323 to 0.589). Multivariate meta-regression showed that the percentage beta-blocker use and follow-up duration explained heterogeneity between the studies. Three studies were included in the comparison of VA in CRT nonresponders before and after upgrade from ICD. CRT nonresponders had an elevated risk of VA compared with ICD-only subjects (odds ratio: 1.497; confidence interval: 1.225 to 1.829).
Conclusions CRT may significantly reduce risk of VA compared with ICDs in patients who meet criteria for CRT. CRT responders have significant reduction in VA compared with nonresponders. CRT nonresponse might significantly increase risk of VA.
Cardiac resynchronization therapy (CRT) is an effective adjunctive therapy for patients with advanced systolic heart failure and wide QRS duration (QRSd) who are taking guideline-directed medical therapy. CRT can improve symptoms, exercise capacity, quality-of-life, and long-term survival, and can reduce congestive heart failure (CHF) hospitalizations (1,2). The mechanism of benefit from CRT is related to improved left ventricular (LV) contractile function and reverse remodeling (3,4). Although CRT can reduce mortality and CHF hospitalizations, its effect on ventricular arrhythmias (VA) defined as ventricular tachycardia (VT) or fibrillation episodes that stop only after antitachycardia pacing and/or shocks (or did not otherwise stop spontaneously), as documented by device interrogations, remains controversial.
Various, mostly nonrandomized, prospective and retrospective trials have studied this issue, but results have not been consistent. Although some studies (5,6) have demonstrated significant reduction in risk of VA with active CRT (particularly in subjects with left bundle branch block morphology), others have failed to demonstrate any significant effect of CRT on the incidence of VA when compared with implantable cardioverter-defibrillator (ICD)-only therapy (3,7,8).
Individual studies have suffered the limitation of small sample size, nonrandomized and retrospective study designs, and single center experiences. To complicate the issue further, multiple case reports point toward a proarrhythmic effect of CRT (9,10). Several recent studies have shown that the effect of CRT on VA is determined by the response to CRT (8,11–24). Although multiple studies show a significantly reduced incidence of VA in CRT responders versus nonresponders, 4 published studies (11–14) show no significant difference or only a trend toward reduction in incidence of VA when comparing CRT responders with nonresponders. Three of these studies have relatively large sample sizes.
Hence, we performed a systematic review and combined the data using meta-analytical techniques to strengthen the level of evidence and provide deeper insight into the issue of response-related differences in incidence of VA in CRT patients. In this meta-analysis, we compared: 1) the effect of CRT-defibrillator (CRT-D) versus ICD-only therapy on sustained VA including VT and ventricular fibrillation; 2) the incidence of VA in echocardiographic responders versus nonresponders to CRT; and 3) the incidence of VA among CRT nonresponders before and after upgrade from ICD-only status.
Our meta-analysis is in accordance with recommendations of the Meta-Analysis of Observational studies in the Epidemiology Group (25).
For primary meta-analysis comparing VA in CRT-D with ICD
Studies (retrospective and prospective; randomized and nonrandomized) of CRT-D in patients with CHF with standard indications (left ventricular ejection fraction [LVEF] ≤35%; QRSd >120 ms) were included if they reported the incidence of VA and compared them between patients with CRT-D and ICD or between pre- and post-CRT-D upgrade from ICD. Studies with mean follow-up of ≥3 months post-CRT-D implant were included.
For the secondary meta-analysis comparing VA in CRT responders with nonresponders
Studies comparing the incidence of sustained VA in patients with CHF with echocardiographic response to CRT and those with no response to CRT were included. Echocardiographic response definitions should include 1 of the following: 1) post-implant LVEF of ≥35%; 2) post-implant improvement in LVEF of ≥5%; or 3) post-implant left ventricular end-systolic volume (LVESV) or LVESV index reduction by ≥10%. Studies with mean echocardiographic follow-up duration of ≥6 months post-CRT implant were included.
For the secondary meta-analysis comparing VA in CRT nonresponders with ICD-only patients
Studies comparing the incidence of sustained VA in patients with CHF with echocardiographic nonresponse to CRT before and after CRT upgrade of ICDs were included.
Studies were excluded if they: 1) lacked a control group; 2) were published only in abstract form; or 3) were non-English studies with no English translation.
We searched MEDLINE (1966 to 2014), Cochrane Database of systematic reviews (1999 to 2014), and Google Scholar using keywords: cardiac resynchronization therapy, CRT, biventricular pacing AND ventricular arrhythmia, VT, ventricular fibrillation in various combinations. The “Related Article” feature on PubMed, and a manual search of references was also used to identify additional studies. We reviewed full text of relevant articles. English translations if necessary were obtained. Titles and abstracts were independently reviewed by 2 reviewers (A.S. and M.K.) and cross-verified for inclusion. Details of search strategy are reported in Figure 1.
Data extraction and assessment of study quality
For each included study, all data elements uniformly reported across most studies were extracted by 2 reviewers (A.S. and M.K.) and are shown in Tables 1 and 2. The quality of each study was evaluated in accordance with the guidelines of U.S. Preventive Task Force and the Evidence-Based Management Group (26,27). The following characteristics were assessed: 1) clear inclusion and exclusion criteria; 2) study sample representative of the population; 3) explanation of sample selection; 4) full specification of clinical and demographic variables; 5) reporting loss of follow-up; 6) clear definition of outcomes and outcome assessment; and 7) adjustment of possible confounders in multivariate analysis. Studies were graded as “poor” if they met 3 criteria, “fair” if they met 3 to 5 criteria, and “good” if they met >5 criteria. The quality assessment of individual studies is reported alongside baseline variables in Tables 1 and 2. All disagreements between reviewers were resolved by consensus.
The percentage of patients with VA in patients with CRT-D and ICD reported across studies was used in this meta-analysis to generate degree of association between CRT and VA. Individual studies did not use Cox proportional hazards model or propensity score model to adjust for potential confounders, which is a limitation of our analysis. For all analyses, odds ratio (OR), log OR, and standard errors were calculated from events and sample sizes.
In the secondary meta-analyses, comparing risk of VA in CRT nonresponders before and after upgrade from an ICD, 1 study by Ouellet et al. (5) reported only hazard ratios and we therefore adopted the reported hazard ratio as the best estimate of OR. Also, the hazard ratio reported by this study for cumulative incidence of subsequent VA after the first event (in cases of electrical storm, only the first VA event in any 24-h period follow-up period was included), was taken as the substitute for general incidence of VA for the purpose of the third analysis comparing VA in CRT nonresponders before and after upgrade. We used the inverse variance method to achieve a weighted estimate of the combined overall effect for all the analyses.
We assessed the results for heterogeneity in our analysis by examining the forest plots and then calculating a Q statistic, which we compared with the I2 index. We considered the presence of significant heterogeneity at the 5% level of significance (for the Q test) and values of I2 exceeding 56% as an indicator of significant heterogeneity. Although the Q statistic for the primary analysis showed statistical significance (p = 0.02), I2 index was 49, indicating mild heterogeneity among studies. For the secondary analyses comparing VA in CRT responders with nonresponders, Q statistic (p = 0.003) and I2 index (60) indicated moderate heterogeneity among studies. This prompted us to adopt the random-effects model to pool effect sizes and all analyses were performed using Comprehensive Meta-analysis version 3 (Biostat, Inc., Englewood, New Jersey).
The underlying heterogeneity prompted us to perform meta-regression analysis to investigate factors contributing to heterogeneity and if our study outcome (VA) was affected by factors other than our primary treatment (CRT and response to CRT). For this, we adopted a weighted regression random-effects model and carried out a multivariate regression using up to 3 pre-determined factors using Comprehensive Meta-analysis version 3. These variables were selected on the basis of factors shown to affect VA in individual studies and on availability of data for most of the studies included. A 2-sided p value <0.05 was regarded as significant for all analyses. Data were represented as forest plots for primary analysis. Potential publication bias was assessed with the Egger test and represented graphically with Begg funnel plots of the natural log of the OR versus its standard error.
Meta-analysis of CRT-D versus ICD
Thirteen studies with 4,631 subjects were included in the primary meta-analysis (Table 1). The mean baseline LVEF was 21% to 32%. All studies included patients with QRSd >120 ms in the CRT-D group and had mean follow-up duration of 6 to 41 months. The CRT-D group, across all studies, was comparable and met currently applicable guidelines for CRT-D but the control group with ICDs only varied across studies (Online Table 1).
Only the study by Gopinathannair et al. (28) includes New York Heart Association class IV patients with continuous-flow left ventricular assist device in both study and control population. Because defibrillator indications included primary and secondary prevention of sudden cardiac death, patients on concurrent antiarrhythmic drug therapy included in the individual studies varied (Table 1) but were not statistically different between groups within each study.
VA definitions are included in Online Table 2. Device settings varied across studies (Online Table 2). Only 3 studies (5–7) reported on subjects with left bundle branch block in addition to the QRSd (Table 1) and most of these did not report VA incidence by specific bundle branch block patterns. Hence, to avoid inaccuracies given a small number of studies reporting bundle branch block patterns, a meta-regression analysis using specific bundle branch block pattern as a variable was not performed. Of the studies that showed no statistically significant difference in VA between groups (Figure 2), a significant difference in improvement of 1 or more echocardiographic parameters (LVESV, LVEF, left ventricular end-diastolic volume) between post-CRT and ICD-only group was documented in all but 2 studies where post-CRT remodeling data were either not reported (29) or were not significantly different between groups (28).
Analysis of the funnel plot for the primary analysis showed no significant publication bias (Online Figure 5). In our pooled analysis, we found a significantly lower incidence of VA in CRT-D group in comparison with ICD-only group, hence favoring CRT therapy (OR: 0.754; 95% confidence interval [CI]: 0.594 to 0.959; p = 0.021) (Figure 2). We performed a sensitivity analysis by excluding the MADIT-CRT substudy (5), which had a larger sample size, but a significant difference in outcome persisted, confirming our finding.
Because only the study by Gopinathannair et al. (28) included left ventricular assist device patients in whom the indication and programming of CRT remains controversial, we performed a second sensitivity analysis excluding this study, which showed results similar to our overall analysis (OR: 0.714; 95% CI: 0.554 to 0.922; p = 0.01). Finally, a meta-regression with duration of follow-up as a covariate did not show any correlation to HR for VA, suggesting that the differences in the duration of follow-up between studies did not explain differences in outcome.
Meta-analysis of CRT responders versus nonresponders
Thirteen studies with 3,667 subjects were included in meta-analysis of VA in CRT responders versus nonresponders (Table 2). Nine studies were retrospective and 4 were prospective. All studies fulfilled 1 of the 3 echocardiographic response criteria mentioned in the inclusion criteria. Baseline study characteristics are shown (Table 2). Follow-up durations ranged from 6 to 57 months. VA definitions are included in Online Table 3. Device settings varied across studies (Online Table 3). Analysis of the funnel plot for the secondary analysis showed lack of studies in the right lower corner, suggesting a possible publication bias (Online Figure 6). Our meta-analysis found a significantly lower incidence of VA in echocardiographic responders in comparison with echocardiographic nonresponders, hence favoring CRT responders (OR: 0.436; 95% CI: 0.323 to 0.589; p < 0.05) (Figure 3).
A multivariate meta-regression analysis showed that the percentage beta-blocker use (p < 0.01) and duration of follow up (p < 0.01) together explained all the heterogeneity between the studies with an R2 analog of 1 (Online Figures 1 and 2). Increased beta-blocker use in the sample population increased the positive effect of CRT on incidence of VA in the CRT responders. Additionally, the longer the population was followed the smaller was the difference in outcome of VA between CRT responders and nonresponders. Meta-regression analysis using QRSd and antiarrhythmic drug use were not significant in explaining heterogeneity. Criteria for CRT response did not add to the heterogeneity between the studies in univariate or multivariate analysis (Online Figure 3).
Meta-analysis of CRT nonresponders versus ICD-only patients
Three studies (5,8,24) were included in the pooled analysis of CRT nonresponders versus ICD-only subjects. In the pooled analysis of these 3 studies, we found a significant difference in occurrence of VA among CRT nonresponders compared with ICD-only subjects with higher incidence of VA in CRT nonresponders (OR: 1.497; 95% CI: 1.225 to 1.829; p < 0.05) (Online Figure 4). This is an important finding that may be clinically relevant because very few previous studies have explored this aspect and further randomized studies are required to clarify.
Our meta-analysis, the largest and most complete to date, assessing the impact of CRT therapy on VA shows that CRT decreases risk of VA compared with ICD therapy alone. Within this group of patients, individuals with echocardiographic response to CRT therapy may have increased benefits with lower incidence of sustained VA compared with nonresponders. Finally, patients who do not show evidence of response by echocardiographic parameters to CRT after an upgrade from an ICD-only therapy may have higher incidence of VA compared with their pre-upgrade status.
The impact of CRT on VA has been a matter of debate for more than a decade. Experimental data have demonstrated proarrhythmic effects of CRT. LV epicardial pacing in CRT has been shown to increase the transmural dispersion of refractoriness and QT intervals (30). In addition, there are multiple case reports of increased monomorphic and polymorphic drug-refractory VT that occurred soon after initiation of CRT and resolved only after cessation of biventricular pacing (9,10).
On the other end of the spectrum, there is evidence that acute biventricular pacing decreases the inducibility of sustained monomorphic VT in patients with ischemic cardiomyopathy (31). Major randomized controlled clinical trials (MIRACLE ICD , REVERSE , CONTAK CD ) did not detect a difference in occurrence of VA between CRT-D and ICD. Although a substudy of MADIT-CRT (5) showed that active CRT treatment was associated with significant reduction in risk of life-threatening VA, evidence from other smaller studies has been conflicting.
Given the neutral results obtained in most studies discussed previously, recent studies have focused on assessing whether reverse remodeling or LVEF improvement (echocardiographic response) with CRT therapy is associated with reduction in VA occurrence. Haugaa et al. (32) showed that persistent LV mechanical dyssynchrony by speckle-tracking echocardiography after CRT placement was associated with increased VA. A substudy of the REVERSE trial by Gold et al. (24) revealed that subjects who responded to CRT by reverse remodeling had a reduced incidence of VA compared with those who did not. The results of our meta-analysis help support the findings of these studies that suggest that antiarrhythmic effects of CRT may be a function of echocardiographic response to pacing.
Although no mechanistic conclusions can be made, our meta-analysis suggests that the reduced VA occurrence in CRT patients occurs concurrently with favorable echocardiographic responses to CRT, such as reverse remodeling and improved contractile function. Increased beta-blocker use has been shown to be independently associated with reduced LVESV and cardiac events after receiving CRT (33). Consequently, increased beta-blocker use in the sample population increased the positive effect of CRT on incidence of VA in the CRT responders in our meta-regression analysis.
If antiarrhythmic benefits of CRT are a function of echocardiographic response, the outcome in subjects without a favorable echocardiographic response to CRT becomes important. We address this issue in our study and demonstrate increased risk of VA in CRT nonresponders when compared with ICD-only therapy, underscoring the importance of selection of candidates for CRT. This outcome highlights the need for careful selection of patients with borderline indications for CRT, where there could be a significant risk of increasing VA with inadequate response to CRT. Large randomized controlled clinical trials are needed in the future to confirm the observations of this meta-analysis.
Most of the study designs were observational, retrospective, single center and had inherent limitations. These studies did not report adjusted hazard ratios from Cox proportional hazards or propensity score models and therefore the effect size data could suffer from known and unknown confounders.
Device type, CRT lead location, and device settings, such as detection zone and tachyarrhythmia therapy algorithms, were different across different studies and were not uniformly reported. We recognize this as an important but inherent limitation of this meta-analysis because differences in device programming parameters within a study and across individual studies can potentially affect overall outcomes.
In the primary meta-analysis comparing VA in CRT with ICD-only patients, the patients in the ICD arm varied between studies with some studies including patients with narrow QRS (meeting standard guidelines for ICD) and others including patients with a wide QRS (>120 ms). In the meta-analysis of CRT responders versus nonresponders, we used 3 different response criteria on the basis of previously performed studies; however, a meta-regression analysis by response criteria did not find any significant contribution to heterogeneity by response criteria used. Finally, for the third analysis assessing VA risk in CRT nonresponders compared with ICD-only therapy, we included only 3 studies limiting the power of the analysis.
CRT may significantly reduce risk of sustained VA compared with ICDs in patients who meet criteria for CRT therapy. CRT recipients with echocardiographic evidence of favorable response have a reduced risk of VA versus nonresponders. CRT nonresponders have a higher risk of VA compared with similar patients who receive ICD only. The potential clinical implications of these results need to be further explored with properly powered, prospective randomized trials in a controlled setting where device types and programming parameters can be standardized.
COMPETENCY IN MEDICAL KNOWLEDGE: CRT therapy may reduce the risk of sustained ventricular arrhythmias compared with ICDs in patients who meet criteria for CRT therapy. CRT recipients with echocardiographic evidence of favorable response have significant reduction in the risk of ventricular arrhythmias compared with nonresponders.
TRANSLATIONAL OUTLOOK: Large prospective randomized controlled trials are required to explore our observations that CRT nonresponders may have a significantly higher risk of VA compared with patients who receive only ICD.
For supplemental tables and figures, please see the online version of this article.
Dr. Gopinathannair has been a consultant for St. Jude Medical, Abiomed, HealthTrust PG, and Boston Scientific; has received speaker fees from Pfizer and Bristol-Myers Squibb; and has received honoraria from Boston Scientific. Dr. Olshansky has been a consultant for Boston Scientific, Medtronic, Biotronik, Amgen, Daiichi Sankyo, Boehringer Ingelheim, Amarin, On-X, BioControl, and Lundbeck; and a speaker for Daiichi Sankyo. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- congestive heart failure
- confidence interval
- cardiac resynchronization therapy
- cardiac resynchronization therapy–defibrillator
- implantable cardioverter-defibrillator
- left ventricle
- left ventricular ejection fraction
- left ventricular end-systolic volume
- odds ratio
- QRS duration
- ventricular arrhythmia
- ventricular tachycardia
- Received September 21, 2015.
- Accepted October 22, 2015.
- American College of Cardiology Foundation
- Linde C.,
- Abraham W.T.,
- Gold M.R.,
- et al.,
- REVERSE Study Group
- Ouellet G.,
- Huang D.T.,
- Moss A.J.,
- et al.
- Blaschke F.,
- Knaus T.,
- Celebi O.,
- et al.
- Higgins S.L.,
- Hummel J.D.,
- Niazi I.K.,
- et al.
- Joep Thijssen C.,
- Jan Willem Borleffs,
- Delgado V.,
- et al.
- Medina-Ravell V.A.,
- Lankipalli R.S.,
- Yan G.X.,
- et al.
- Di Biase L.,
- Gasparini M.,
- Lunati M.,
- et al.
- Itoh M.,
- Yoshida A.,
- Fukuzawa K.,
- et al.
- Manfredi J.A.,
- Al-Khatib S.M.,
- Shaw L.K.,
- et al.
- Ruwald M.H.,
- Solomon S.D.,
- Foster E.,
- et al.
- Schaer B.A.,
- Osswald S.,
- Di Valentino M.,
- et al.
- Fish J.M.,
- Brugada J.,
- Antzelevitch C.
- Nabeta T.,
- Inomata T.,
- Iida Y.,
- et al.
- Lin G.,
- Rea R.F.,
- Hammill S.C.,
- et al.
- Ermis C.,
- Seutter R.,
- Zhu A.,
- et al.
- Voigt A.,
- Barrington W.,
- Ngwu O.,
- et al.