Author + information
- Sean D. Pokorney, MD, MBA∗ ( and )
- Sana M. Al-Khatib, MD, MHS
- Electrophysiology Section, Duke Clinical Research Institute, Duke University Medical Center, Durham, North Carolina
- ↵∗Address for correspondence:
Dr. Sean D. Pokorney, Division of Cardiology, Duke University Medical Center, DUMC 3845, Durham, North Carolina 27710.
- heart failure
- implantable cardioverter-defibrillator
- left ventricular assist device
Implantable cardioverter-defibrillators (ICDs) decrease mortality in patients with heart failure with reduced ejection fraction (1). The outcomes of ICDs in patients with continuous-flow left ventricular assist devices (CF-LVADs) have only been evaluated in small, retrospective studies, despite the fact that implantation rates of CF-LVADs have dramatically increased over time, and greater than one-third of CF-LVAD patients have subsequent ventricular arrhythmias (2,3). The limited availability of data on outcomes of ICDs for patients with CF-LVADs is exemplified by the absence of any ICD recommendations for patients with CF-LVADs in the American College of Cardiology/American Heart Association/Heart Rhythm Society device-based therapy guidelines (1).
The consequences of ventricular arrhythmias in patients with CF-LVADs are not well understood. These patients have left ventricular hemodynamic support, which increases the tolerability of these events; however, ventricular arrhythmias can affect the right ventricle, resulting in lower CF-LVAD flows and hemodynamic compromise. Among patients with CF-LVADs, 16% with ventricular arrhythmias have hypotension or decreased CF-LVAD flows, whereas only 3% have syncope associated with the ventricular arrhythmia (3). However, the effect of ICDs on arrhythmic and all-cause death in patients with CF-LVADs has not been well defined. The risks and drawbacks to ICD therapy in patients with CF-LVADs are important to consider: ICD shocks often take place while the patient is conscious and even asymptomatic, management of anticoagulation and bleeding around the time of ICD implantation or generator replacement can be challenging, and outcomes in patients with infection and concomitant ICDs and CF-LVADs are poor (4).
In this issue of JACC: Heart Failure, Clerkin et al. (5) have presented a retrospective propensity-matched analysis from the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support) registry, comparing outcomes among CF-LVAD patients with and without an ICD in situ. The matched analysis included 4,418 patients with a CF-LVAD, meaning 2,209 patients in each arm (ICD vs. no ICD). The median duration of follow-up was slightly longer than 12 months in both cohorts. The study found that having an ICD was associated with higher all-cause mortality (hazard ratio: 1.20; 95% confidence interval: 1.04 to 1.39; p = 0.013) and a higher rate of unexpected death (hazard ratio: 1.33; 95% confidence interval: 1.03 to 1.71; p = 0.03). Patients in the ICD cohort had statistically significant lower rates of recovery and CF-LVAD explant, relative to patients in the no ICD cohort. The matched ICD patients also had higher rates of rehospitalization and ventricular arrhythmias requiring defibrillation or cardioversion. There were no significant differences in rates of overall infection or rates of bacteremia between the 2 cohorts.
This study by Clerkin et al. (5) had many strengths, including the large number of patients with CF-LVADs and ICDs, as well as the high quality of the data in the INTERMACS registry. The INTERMACS registry also collects data from over 150 sites across the United States (2), so the results are representative of the nationwide CF-LVAD experience. The propensity-matched analysis addressed the measured confounders, which was important based on the fact that nearly all of the baseline variables were statistically different between the unmatched patients with and without ICDs. A falsification-hypothesis analysis did not identify confounding as an explanation for the study results.
The study also had some limitations. As the authors highlighted, propensity matching is unable to account for unmeasured factors; even the most robust statistical methods cannot fully adjust for selection bias and confounding. In this ICD analysis of CF-LVAD patients, several important variables were missing and not included in the matching. First, there were no available data on right ventricular function or pulmonary vascular resistance. This is particularly relevant, as right heart failure was 1 of the 5 highest modes of death seen more frequently in the ICD cohort. Furthermore, better right heart function and lower pulmonary vascular resistance likely resulted in improved tolerability of ventricular arrhythmias, so it would be valuable to understand these characteristics in a CF-LVAD population that was selected in clinical practice to not have an ICD. Second, the initial indication for the ICD was unknown. Recurrent ventricular arrhythmia was included in the matching; however, this variable was limited to patients with frequent defibrillation or cardioversion. Third, the timing of ICD implantation was not known, including whether an ICD had been implanted at the time of CF-LVAD placement or the duration of time the ICD had been in situ prior to CF-LVAD placement. Likewise, data on ICD implantation after CF-LVAD placement were not available. Fourth, beta-blocker use was low (<50%) and was only evaluated at baseline, while beta-blocker use has been shown to increase in the 3 months after LVAD implantation (6). Finally, there were no data on site-level variation in ICD use, which could have affected the primary endpoint based on variation in mortality outcomes and preferential ICD use by site.
Other limitations include the long study period, which spans more than 10 years, as the management of advanced heart failure has changed drastically during this time. Reasons for hospitalization were not provided; these data might have shed some light on the findings. Although mode of death data can be informative, it is not clear whether modes of death in this study were centrally adjudicated using standardized definitions. While the finding of a higher rate of ventricular arrhythmias in the ICD group might be puzzling to some, this finding was likely due to the fact that those patients had devices that detected and treated arrhythmias; patients with no ICDs might have experienced similar rates of arrhythmias that went undetected due to the absence of a device that provides continuous monitoring.
As we evaluate the results of the analysis by Clerkin et al. (5), it would be ideal to identify a physiologic mechanism that would explain the association between ICD use and higher mortality. Patients with an ICD would be at higher risk of infection, which could increase mortality, but there was no difference in rates of infection or bacteremia between the ICD and no ICD cohorts. Right ventricular pacing worsens clinical outcomes in non-LVAD patients. The effects of right ventricular pacing with a CF-LVAD are unknown. Unfortunately, there were no data available on pacing burden or bradycardia programming parameters to explore pacing as a potential factor related to the lower rates of recovery and CF-LVAD explant. A final possibility is that ICD shocks themselves increased mortality. Again, the tachycardia programming parameters were not known, nor have optimal programming parameters been established for patients with CF-LVADs. Withdrawal of support and neurological dysfunction were the 2 modes of death that most accounted for the higher mortality in the ICD cohort, and it would be difficult to identify a physiologic link between an ICD in situ and these modes of death. Regardless, this analysis clearly found that ICD use was not associated with lower mortality in the propensity-matched cohorts.
These limitations aside, the current analysis highlighted an important gap in knowledge with respect to ICD use in patients with CF-LVADs. More importantly, the INTERMACS registry is the most comprehensive and high-quality LVAD database available, yet we have established that multiple limitations remained. Prospective randomized trials are needed in patients with CF-LVADs to guide the use of ICDs in these patients. An optimal approach to ICD programming in patients with CF-LVADs needs to be established. Studies of hemodynamic and physiologic response to ventricular arrhythmias would be informative for understanding the tolerability of ventricular arrhythmias in patients with CF-LVADs. An outcomes-based randomized trial is needed to definitively determine whether ICDs are beneficial in patients with CF-LVADs, as well as to determine the quality of life and healthcare utilization impact of ICDs in this patient population.
The use of CF-LVADs is increasing, and the outcomes of these patients continue to improve with technological and medical interventions. The data from Clerkin et al. (5) call into question the role of ICDs in the CF-LVAD population and present an opportunity for the heart failure and electrophysiology communities to team up in designing and conducting a randomized clinical trial to most definitively define the outcomes of ICDs in patients with a CF-LVAD.
↵∗ Editorials published in JACC: Heart Failure reflect the views of the authors and do not necessarily represent the views of JACC: Heart Failure or the American College of Cardiology.
Dr. Pokorney has received research support from the Food and Drug Administration, Bristol-Myers Squibb, Pfizer, Janssen Pharmaceuticals, Gilead, and Boston Scientific; and has served as a consultant and on the advisory board for Boston Scientific, Medtronic, Bristol-Myers Squibb, and Pfizer. Dr. Al-Khatib has reported that she has no relationships relevant to the contents of this paper to disclose.
- 2017 American College of Cardiology Foundation
- Epstein A.E.,
- DiMarco J.P.,
- Ellenbogen K.A.,
- et al.
- Enriquez A.D.,
- Calenda B.,
- Miller M.A.,
- Anyanwu A.C.,
- Pinney S.P.
- Clerkin K.J.,
- Topkara V.K.,
- Demmer R.T.,
- et al.
- Khazanie P.,
- Hammill B.G.,
- Patel C.B.,
- et al.