Author + information
- Received April 12, 2013
- Accepted April 29, 2013
- Published online June 25, 2013.
- Arthur R. Garan, MD∗,
- Melana Yuzefpolskaya, MD∗,
- Paolo C. Colombo, MD∗,
- John P. Morrow, MD∗,
- Rosie Te-Frey, NP†,
- Drew Dano, MS∗,
- Hiroo Takayama, MD†,
- Yoshifumi Naka, MD, PhD†,
- Hasan Garan, MD∗,
- Ulrich P. Jorde, MD∗ and
- Nir Uriel, MD∗∗ ( )()
- ∗Department of Medicine, Division of Cardiology, Columbia University Medical Center, New York, New York
- †Department of Surgery, Division of Cardiothoracic Surgery, Columbia University Medical Center, New York, New York
- ↵∗Reprint requests and correspondence:
Dr. Nir Uriel, Mechanical Circulatory Support Program, Division of Cardiology, Columbia University Medical Center/New York Presbyterian Hospital, 622 West 168th Street, PH-12, Room 134, New York, New York 10032.
Objectives This study sought to evaluate the prevalence and significance of ventricular arrhythmia (VA) and the role of an implantable cardioverter-defibrillator (ICD) in patients supported by a continuous-flow left ventricular assist device (CF-LVAD).
Background VAs are common in patients supported by CF-LVADs but prospective data to support the routine use of ICDs in these patients are lacking.
Methods All patients supported by long-term CF-LVAD receiving care at our institution were enrolled. The ICDs were interrogated at baseline and throughout prospective follow-up. The VA was defined as ventricular tachycardia/fibrillation lasting >30 s or effectively terminated by appropriate ICD tachytherapy. The primary outcome was the occurrence of VA >30 days after CF-LVAD implantation.
Results Ninety-four patients were enrolled; 77 had an ICD and 17 did not. Five patients with an ICD had it deactivated or a depleted battery not replaced during the study. Twenty-two patients had a VA >30 days after LVAD implantation. Pre-operative VA was the major predictor of post-operative arrhythmia. Absence of pre-operative VA conferred a low risk of post-operative VA (4.0% vs. 45.5%; p < 0.001). No patients discharged from the hospital without an ICD after CF-LVAD implantation died during 276.2 months of follow-up (mean time without ICD, 12.7 ± 12.3 months).
Conclusions Patients with pre-operative VA are at risk of recurrent VA while on CF-LVAD support and should have active ICD therapy to minimize sustained VA. Patients without pre-operative VA are at low risk and may not need active ICD therapy.
- heart failure
- implantable cardioverter-defibrillator
- left ventricular assist device
- ventricular arrhythmia
- ventricular tachycardia
The role of the implantable cardioverter-defibrillator (ICD) for primary prevention of sudden cardiac death in patients with heart failure has been well established in multiple large randomized trials (1,2). Prophylactic ICDs have been shown to confer a mortality benefit in patients with heart failure and reduced ejection fraction with New York Heart Association functional class II and III symptoms, but not in patients with class IV symptoms. More recently, mechanical circulatory support devices have assumed an important role in the treatment of patients with class IV symptoms either as a bridge to cardiac transplantation (BTT) or as destination therapy (DT).
Although ventricular arrhythmia (VA) is often a lethal event in patients with heart failure, the presence of continuous-flow left ventricular assist devices (CF-LVAD) may alter the impact of VAs for these patients. For patients supported by LVADs, VAs are common, and numerous studies have demonstrated that appropriate tachytherapies are delivered by ICDs in 22% to 52% of patients with LVADs (3–10). Some investigators have suggested that LVADs are arrhythmogenic, either by introducing new areas of scar or by altering gene expression of ion channels possibly involved in arrythmogenesis (3,4). Although VAs may be common, there have been reports of patients surviving after hours—and in extreme cases, months—of ventricular fibrillation (VF) (11–17).
Prospective data defining the role of ICDs in patients with continuous-flow LVADs are lacking. Accordingly, we sought to characterize the prevalence, risk factors, and significance of VAs in patients with long-term CF-LVADs to determine whether concomitant ICD therapy is necessary for all patients receiving long-term mechanical support.
This study was approved by the institutional review board of the Columbia University Medical Center. All patients 18 years of age or older with a long-term CF-LVAD presenting to our institution between January 1, 2012, and October 1, 2012, were enrolled and followed up prospectively. The primary outcome was the occurrence of “late” VAs (>30 days post-operatively) after LVAD implantation. Any VAs occurring within 30 days of LVAD implantation were excluded because of the likelihood that they would occur in a hospitalized setting and often in the presence of intravenous vasoactive therapies. Secondary outcomes were all-cause mortality and catheter ablation of VA. Deaths were classified as “cardiac” when a definitive cause of death related to a cardiovascular event could be identified, “noncardiac” when the cause of death identified did not relate to the cardiovascular system, or “indeterminate” when the cause of death was unknown. Whenever possible, ICDs were interrogated at the time of death to ascertain whether VA might have been the cause of death.
A VA was defined as ventricular tachyarrhythmia (ventricular tachycardia [VT] or ventricular fibrillation [VF]) that received appropriate therapy (antitachycardia pacing or shock) from an ICD or was sustained for >30 s in the absence of effective treatment. Intracardiac electrograms, telemetry tracings, or electrocardiograms were obtained for each episode and reviewed by an electrophysiologist. Device detection thresholds and treatment algorithms were individualized for each patient. The VAs were divided into those occurring early (within 30 days after LVAD implant surgery) or late (occurring >30 days after surgery).
Enrollment and follow-up
Patients with an ICD and patients without an ICD were enrolled. At our institution, patients who undergo CF-LVAD implantation without an ICD do not routinely have an ICD inserted before discharge. After the informed consent process, a list of active medications, current LVAD settings, and demographic information were recorded. Next, patients with ICDs underwent device interrogation. Information collected included the tachyarrhythmia settings and details for each VA event stored within the device that resulted in therapy. For patients who already had an LVAD at the time of enrollment, VAs that had occurred before the enrollment period were identified by the initial ICD interrogation and were included in the analysis with those occurring during study follow-up. After enrollment, patients were followed up prospectively for the occurrence of VAs.
Routine evaluation of patients presenting with VAs included a transthoracic echocardiogram to exclude a “suction” event, defined as obstruction of the inflow cannula by myocardium, which frequently results in a tachyarrhythmia. All patients without ICDs underwent electrocardiography or telemetry monitoring whenever they reported symptoms suggestive of VAs (e.g., dizziness, palpitations) or manifested signs of right ventricular failure.
Data were collected using Excel software (Microsoft Corp., Redmond, Washington). All data were analyzed using R version 2.15.2. Categorical values are presented as percentages and continuous variable as mean ± SD and medians with interquartile ranges. Fisher's exact test was used to examine the association between binary variables. For comparison of continuous variables between groups, the Wilcoxon rank-sum test was used. For the survival analysis, the Mantel-Haenszel test was used to compare survival curves between groups. For multivariate analysis, Cox regression models were used. A p value <0.05 was considered statistically significant.
Ninety-four patients were enrolled; 77 had an ICD and 17 did not. One patient without an ICD at the time of LVAD implantation underwent ICD insertion after experiencing sustained VF after hospital discharge (Fig. 1A). None of the remaining 16 patients without an ICD underwent “primary prevention” insertion of an ICD. In addition, 5 patients had their ICD either deactivated or did not have a depleted generator changed after the device had reached end of life. Total time with a deactivated ICD was 40.1 patient-months.
The baseline characteristics of patients with and patients without an ICD are presented in Tables 1 and 2. Overall, 85.1% (n = 80) were men, 52.1% (n = 49) had an ischemic cardiomyopathy, and 48.9% (n = 46) were implanted as BTT. The average age was 62.2 ± 12.0 years. Seven (7.4%) had undergone catheter ablation for VT before surgery, and 44 (46.8%) had had at least 1 VA before surgery. Forty patients (42.6%) were receiving an antiarrhythmic drug at the time of LVAD implantation. Eighty-nine (94.7%) patients were taking a beta-blocker. The mean left ventricular end-diastolic dimension was 6.8 ± 1.0 cm and the mean left ventricular ejection fraction was 17.2 ± 5.3%. Patients without an ICD were more often women (41.2% vs. 9.1%; p = 0.003), were more likely to have had <3 months of disease duration (64.7% vs. 2.6%; p < 0.001), and were more likely to have a smaller left ventricular end-diastolic dimension (5.8 ± 1.0 cm vs. 7.0 ± 0.9 cm; p < 0.001). There were no differences in the number of patients taking beta-blockers or antiarrhythmic medications in the 2 groups during the study period.
Sixty-two patients (66.0%) had undergone CF-LVAD implantation before the enrollment period. Total time supported by CF-LVAD in patients with an active ICD was 969.1 patient-months, and prospective follow-up time totaled 281.0 patient-months. Total time supported by CF-LVAD for patients without an active ICD was 276.2 patient-months (mean 12.7 ± 12.3 months), and prospective follow-up time totaled 139.6 patient-months.
Prevalence of VA
The prevalence of VAs occurring within the first 30 days post-operatively was 20.2%; the prevalence occurring >30 days after LVAD implantation was 23.4%. Excluding VAs that occurred within the first 30 days, 22 (23.4%) patients had 98 VAs for an event rate of 1.2 VAs per patient-year. Four patients had 5 or more late VAs. Among patients without an active ICD, the prevalence of VAs occurring >30 days post-operatively was 4.0%. The distribution of post-operative VAs is shown in Figure 2.
Predictors of late VA
Predictors of late VAs among patients with an ICD are shown in Tables 3 and 4. In a multivariate analysis, the only independent predictor of the occurrence of late post-operative VAs was the presence of VA before LVAD implantation. Age, sex, and the INTERMACS (Interagency Registry for Mechanical Assisted Circulatory Support) profile were not statistically different between patients with and patients without late VAs. Similarly, the etiology of the patient's cardiomyopathy, baseline left ventricular function, and left ventricular end-diastolic dimension were not independent predictors of post-operative VAs. Lastly, the occurrence of right ventricular failure in the post-operative period was not predictive of the occurrence of late VAs. The positive predictive value of pre-operative VAs was 45.5%, and the negative predictive value was 96.0% (Table 5). A time-to-event analysis for freedom from post-operative VAs is shown in Figure 3.
VA untreated by ICD
Six patients had sustained VAs that were not immediately terminated (Table 6); of those, 5 had ICDs and 1 did not. In 3 patients, attempts by the ICD to cardiovert VT failed; 2 survived despite transformation to sustained VF; and 1 died as a result of several weeks of recurrent, intractable VT that ultimately led to right ventricular failure and multisystem organ failure. In the 2 other instances where the arrhythmias were sustained despite the presence of an ICD, the patients had VT in the monitor-only zone: 1 at a rate of 160 beats/min and the other at a rate of 170 beats/min (Fig. 1B). On device interrogation, both of these arrhythmias had lasted for 2 weeks before clinical detection. The patient without an ICD presented with dizziness for 4 h and was found to be in VF. He was externally cardioverted and a secondary prevention ICD was implanted (Fig. 1A).
All 6 patients who had a prolonged VA presented with right ventricular failure and worsening renal function; 4 also had worsening liver function test abnormalities. Except for the single patient in this group who had <3 months of disease duration pre-operatively, all had experienced 2 or more VAs pre-operatively.
VA requiring ablation
Three (3.2%) patients underwent catheter ablation for VT after LVAD implantation. Of these, 2 procedures were performed endocardially and 1 was performed epicardially. A total of 8 VTs were inducible, and 5 were successfully ablated; 2 patients had 4 inducible VTs each, and 1 patient had none. Of the 2 patients who underwent successful ablation, both had subsequent VAs. One ultimately died of intractable VT with subsequent right ventricular failure (detailed in preceding text), and the other had a reduction in the frequency of VAs.
During the follow-up period, there were 7 deaths (Table 7). One death was cardiac, 4 were noncardiac, and 2 were indeterminate. Two deaths occurred in patients with ICDs that were no longer active; 1 was secondary to persistent Enterococcus faecium infection, and the other was indeterminate. The remaining 5 deaths were of patients with an active ICD; causes of death were right ventricular failure secondary to weeks of intractable VA (detailed in preceding text), intracranial hemorrhage, Clostridium perfringens infection, hypoxemic respiratory failure, and indeterminate. There was no difference in survival when patients were stratified by the presence or absence of active ICD therapy (Fig. 4). There were no deaths among patients who were discharged after LVAD implantation without an ICD. The overall mortality was 6.7% per year, and the overall mortality from cardiac or indeterminate causes was 2.9% per year.
Of 35 patients with an LVAD as DT and an active ICD, 91.4% (n = 32) remain on support, 5.7% (n = 2) died, and 2.9% (n = 1) had the LVAD explanted during the study period. Of 37 patients with an LVAD as BTT and an active ICD, 54.1% (n = 20) remain on support, 37.8% (n = 14) underwent orthotopic heart transplant (OHT), and 8.1% (n = 3) died during the study period.
Of 13 patients with an LVAD as DT and without an active ICD, 84.6% (n = 11) remain on support and 15.4% (n = 2) died. Finally, of 9 patients with an LVAD as BTT and without an active ICD, 55.6% (n = 5) remain on support, 33.3% (n = 3) have undergone OHT, and 11.1% (n = 1) had the LVAD explanted.
In the current study, we aimed to determine the prevalence and clinical significance of ventricular arrhythmias after LVAD implantation. Furthermore, we sought to determine the role of the ICD in this patient population. Our principal findings are as follows: 1) VAs were common among patients with long-term CF-LVADs. 2) VAs are usually not life-threatening during CF-LVAD support, although a subset of patients has significant arrhythmic morbidity and represents those who may benefit from aggressive antiarrhythmic therapies including pharmacotherapy, ICD therapies, and catheter ablation. 3) Pre-operative VAs were the major predictor of late post-operative VAs in this population, and the risk of developing de novo VAs after LVAD implantation is low. 4) There was no difference in mortality between patients with and patients without active ICDs, as CF-LVADs appear to provide enough hemodynamic support to prevent sudden arrhythmic death in patients without an ICD.
To our knowledge, this is the first prospective analysis of the clinical impact of VAs in patients supported by long-term CF-LVADs. Increasing numbers of patients are undergoing implantation of long-term mechanical circulatory support devices as DT and BTT. As patients with CF-LVADs are supported for increasing duration, the following questions have arisen: 1) Should ICDs be implanted in patients who undergo CF-LVAD implantation without a pre-existing device; and 2) should generators be changed when depleted in patients with CF-LVADs and an ICD?
In the largest retrospective study to date, Cantillon et al. (5) demonstrated that the presence of an ICD was associated with improved survival in patients with LVADs. However, this analysis included mostly pulsatile devices. Pulsatile flow pumps are filled by contribution from the native heart, relying on systolic filling, and in this way are pre-load sensitive. In contrast, continuous-flow pumps fill actively in both systole and diastole and are less sensitive to pre-load and contribution from the native heart. Therefore, results from the pulsatile pump era may not apply to today's patients with contemporary continuous-flow pumps.
Because of a lack of prospective data supporting their use, ICDs are not routinely inserted in patients undergoing CF-LVAD implantation at our center. Hence, our analysis includes 17 patients supported by CF-LVAD who were discharged without an ICD. Not surprisingly, these patients had shorter disease duration—often undergoing CF-LVAD implantation during their index admission for heart failure—and less ventricular dilation than patients who had an ICD at the time of CF-LVAD implantation. Importantly, none of the patients discharged from the hospital without an ICD after their index admission for CF-LVAD implantation died during the study period.
During >400 patient-months of prospective follow-up time, we identified pre-operative VA as the lone independent predictor of post-operative VAs. Neither type of cardiomyopathy nor degree of ventricular dilation was predictive of late VAs. Finally, right ventricular failure was common among our cohort but it did not predict the occurrence of post-operative VAs, although patients with prolonged ventricular tachyarrhythmias uniformly presented with right ventricular failure.
It is difficult to assess the impact of VAs when they are treated effectively—and often without the patient's awareness—by an ICD. In our study, a number of these events were prolonged and allowed us to observe the consequences of untreated VTs in the presence of a CF-LVAD. Although ICDs may prevent adverse sequelae of sustained VAs in a subset of patients (e.g., right ventricular failure, renal failure), they do not appear to confer survival benefit through prevention of sudden cardiac death for this population as they do for patients with advanced heart failure not supported by LVADs, for CF-LVADs seem to prevent complete hemodynamic collapse in the setting of VA.
Implementation of ICD therapy is not without risk. The incidence of ICD infection appears to be rising and may be as high as 7% in some reports (18). Furthermore, ICD shocks have been associated with decreased quality of life (19). With these potential risks in mind, we believe our data support a patient-specific approach to arrhythmia management in the LVAD population. Therefore, our center has adopted the following strategy (Fig. 5): 1) Patients without pre-operative VAs do not need a primary prevention ICD placed if they do not have a VA at the time of CF-LVAD implantation; devices may be placed in the event of VAs occurring de novo after CF-LVAD implantation. 2) Patients with pre-operative VAs should have active ICD therapy maintained to prevent the adverse sequelae of sustained arrhythmias. 3) Patients without pre- or post-operative VAs and an ICD with a depleted generator do not need it replaced. Furthermore, deactivation of the ICD should be performed for these patients when the device reaches end of life to avoid the risk of device malfunction.
We believe such an approach to the question of ICD therapy for patients with CF-LVADs warrants further evaluation in a larger, multicenter prospective study. In addition, it must be noted that this strategy does not take into account the need for cardiac pacing or the potential effects of cardiac resynchronization therapy. Whether cardiac resynchronization therapy conveys a survival benefit or improved functional capacity in patients supported by long-term CF-LVAD is entirely unknown and warrants prospective evaluation. Until we are able to answer this question, however, we will continue to recommend reactivation of whatever pre-operative pacing settings the patient had after CF-LVAD implantation.
Lastly, current practice is to implant ICDs in patients with heart failure and refractory VAs. Our data indicate that patients with a higher burden of pre-operative VAs were at risk for having more post-operative VAs. Such arrhythmias may put them at risk for the development or worsening of right ventricular failure as well as recurrent shocks. Therefore, for patients with ICDs at the time of CF-LVAD implantation, we believe arrhythmia burden should be carefully evaluated by device interrogation before the implementation of mechanical circulatory support. That will allow patients at high risk of post-operative VAs to be identified so that interventions aimed at minimizing the risk of recurrent VAs, including antiarrhythmic medication use, catheter ablation, and even biventricular support, may be considered.
Our study represents a single-center experience. The ICD therapy was not randomized, and there was no standardization of ICD programming. Patients with more aggressive tachytherapy settings might have had higher rates of appropriately delivered tachytherapies that would inflate their risk of VA. In addition, as patients without ICDs were monitored clinically for evidence of VAs, their risk might have been underestimated; VAs lasting >30 s, then self-terminating, might have been entirely undetected without an ICD to store or treat the arrhythmia. It is important to note, however, that if they occurred, such arrhythmias were of no clinical consequence. As Oswald et al. (7) have suggested, less aggressive ICD tachytherapy settings are appropriate for LVAD supported patients given that the risk of sudden hemodynamic collapse is obviated by mechanical support. Finally, while every effort was made to identify all VAs, some might not have been identified if they were erased from the ICD at the time of interrogation by another healthcare provider without documentation in the medical record.
Late ventricular arrhythmias are common after implantation of a CF-LVAD. Pre-operative VAs predict post-operative VAs, and patients without a history of arrhythmias are at low risk of experiencing them de novo post-operatively. Routine primary prevention implantation of ICDs or generator changes for pre-existing devices are unnecessary for patients without a history of VAs because CF-LVADs provide a level of hemodynamic support that allows such interventions to be performed according to a secondary prevention strategy.
Dr. Naka has received honorarium from Thoratec Corporation. Dr. Jorde has received honorarium from Thoratec Corporation; and is supported by a National Institutes of Health grant (R01-HL 868845-04). Dr. Uriel has received honorarium from HeartWare; and is supported by a National Institutes of Health grant (1 K23 HL 103795-01A1). All other authors have reported they have no relationships relevant to the contents of this paper to disclose. Dr. Jorde and Dr. Uriel are joint senior authors.
- Abbreviations and Acronyms
- bridge to transplantation
- continuous-flow left ventricular assist device
- destination therapy
- implantable cardioverter-defibrillator
- left ventricular assist device
- ventricular arrhythmia
- ventricular fibrillation
- ventricular tachycardia
- Received April 12, 2013.
- Accepted April 29, 2013.
- American College of Cardiology Foundation
- Ziv O.,
- Dizon J.,
- Thosani A.,
- Naka Y.,
- Magnano A.R.,
- Garan H.
- Oz M.C.,
- Rose E.A.,
- Slater J.,
- Kuiper J.J.,
- Catanese K.A.,
- Levin H.R.
- Salzberg S.P.,
- Lachat M.L.,
- Zünd G.,
- Turina M.I.