Implantable Cardioverter-Defibrillator Therapy in Patients with Left Ventricular Assist Devices: A Shocking Tale of Survival

This article refers to ‘Cardiac implantable electronic devices with a defibrillator component and all-cause mortality in left ventricular assist device carriers: results from the PCHF-VAD registry’ by M. Cikes et al., published in Eur J Heart Fail 2019;21:1129–1141. ‘I have no desire to suffer twice, in reality and then in retrospect’ Sophocles, Oedipus Rex The mortality benefit of implantable cardioverter-defibrillator (ICD) therapy for primary and secondary prevention in patients with heart failure with reduced ejection fraction (HFrEF) is well-established and recommended by both American and European guidelines.1 Regardless, the rates of ICD implantation in the United States (US) and Europe vary substantially, with far more devices being implanted in the US than in European countries.2, 3 While ICDs have proven benefit in stable chronic HFrEF, they are contraindicated in end-stage HFrEF without a plan for advanced therapies and their role in improving outcomes in patients who have a left ventricular assist device (LVAD) is uncertain. Results of prior retrospective studies of the mortality benefit of ICDs in patients with continuous-flow LVADs are variable.4-7 For the most part, these studies from US patient populations have not shown a survival benefit of ICDs in the LVAD population. The largest studies compared propensity-score matched populations from data collected in the US INTERMACS registry and showed no benefit from ICD therapy.4 However, since practice patterns for ICD implantation in Europe and US differ and there are limited studies using data from European populations, the results of prior studies cannot fully be extrapolated to European patients. In the September issue of the Journal, Cikes et al.8 provide much needed insights into the use and outcomes of ICDs in LVAD patients treated in European centres. Their findings add to a growing collection of retrospective studies that address the important and controversial topic of whether ICDs reduce mortality in patients with durable continuous-flow LVADs. Using the Postgraduate Course in Heart Failure Ventricular Assist Device (PCHF-VAD) registry where data are entered from 12 LVAD centres in Europe, the authors evaluated the impact of ICD therapy on all-cause mortality in 448 LVAD patients. In contrast to most prior studies, Cikes et al. found that LVAD patients with an ICD in place experienced all-cause mortality that was 36% lower than in patients without an ICD. Moreover, the survival advantage associated with an ICD was even higher after adjusting for baseline covariates and with a propensity score adjustment. The striking differences between this and prior studies raises the question of why the results differ so markedly? It is likely that multiple factors contributed to these discrepant results. First, the baseline characteristics of the no-ICD group suggest that these patients are more critically ill at the time of LVAD implant than the patients who had an ICD. Specifically, the no-ICD group was more likely to be INTERMACS class 1 or 2, on extracorporeal membrane oxygenation support, and intubated prior to LVAD. They were also less likely to be receiving guideline-directed medical therapies such as beta-blockers. Differences such as these are common in retrospective analyses and one could easily argue that because the no-ICD population was sicker at baseline, they experienced a higher mortality rate that was driven by issues related to more severe pump failure, as opposed to lack of active ICD therapy. Arguing against this possibility is that when the investigators performed both a covariate-adjusted and propensity-adjusted analysis to control for these differences in illness severity, the results were actually even more in favour of ICD therapy. However, confounding factors that were not addressed by adjustment of the data, and other population characteristics may have played a role in determining these results. Importantly, this is the largest study using data from a European population to date. One of the most notable differences between patients treated at European centres compared to patients included in US-based studies is the significantly lower baseline ICD use in Europe compared to the US, with only 47% baseline ICD use in the present study compared to greater than 80% baseline ICD use in INTERMACS and UNOS registry analyses from the US.4, 7 The lower ICD use in the European population is likely driven by fewer ICDs implanted for primary prevention in Europe as reported previously.2 In the present study, only 34% of patients had an ICD implanted for primary prevention. Thus, patients in the active ICD group were more likely to have ICD therapy for high-risk features such as ventricular arrhythmias or sudden cardiac death and the study population would be less diluted by primary prevention patients who might not benefit from an ICD. This would increase the apparent effectiveness of the ICD in the study population. The corollary is that there may have been many high-risk patients in the no-ICD group who were not protected from death due to ventricular arrhythmias. Both these factors are likely to increase the differences in outcomes associated with ICD therapy between the two groups of patients. The finding in this study and in those previously reported, that pre-LVAD ventricular arrhythmia remained an independent risk factor for mortality in multivariate analysis supports this possibility. When post-LVAD ventricular arrhythmias were analysed as a covariate, they increased the risk of mortality 2.4-fold, while active ICD therapy was associated with a 47% reduction in mortality. These data seem to provide an explanation for the improved survival with ICD therapy found in the present study. While the benefits reported in the present study might be explained by the differences in study populations summarized above, it is also possible that ICDs offer important benefits in the LVAD population.9 It has been postulated that ICDs do not benefit patients with LVADs because the mechanical pump can provide adequate forward blood flow from the heart even in the presence of ventricular arrhythmias5; however, this hypothesis fails to recognize that the right ventricle, which is often dysfunctional in patients with advanced heart failure, may not tolerate episodes of ventricular tachycardia or fibrillation. With ventricular arrhythmias, the right ventricle may lose pulsatility so that it functions like a conduit rather than a pump, a situation analogous to that seen in patients after a Fontan procedure. Whether this Fontan-like circulation is sufficient to provide adequate preload for the LVAD to maintain cardiac output is likely to be variable between patients. Thus, maintenance of cardiac performance in patients who are susceptible to right ventricular dysfunction in the setting of ventricular arrhythmias is a potential mechanism through which the presence of an ICD may improve survival in LVAD patients. An additional possibility for the improved outcomes seen with ICDs is they can provide ongoing protection against bradyarrhythmias as well as monitoring to detect rapid atrial arrhythmias that could compromise cardiac function, even when there is an LVAD in place. In the absence of evidence from prospective randomized clinical trials evaluating active ICD therapy in LVAD patients, Cikes et al. should be congratulated for performing a comprehensive analysis and attempting to adjust for potential confounders in a patient population that is very different from that included in prior studies. While the results of this study suggest that sustained ventricular arrhythmias may be detrimental in the LVAD population and that survival may be improved by ICD implantation, the issue is far from settled. The growth of LVAD therapy and the findings of this study provide a strong argument for a prospective randomized trial of active ICD therapy in LVAD patients. When considering such a study though, care should be taken to address the following: is there a mortality benefit of ICD therapy for both secondary and primary prevention of ventricular arrhythmias after LVAD implantation? Does active ICD therapy benefit all patients equally or are there certain subgroups of patients, such as those with right ventricular dysfunction or ischaemic cardiomyopathy, that are more likely to derive a survival benefit from ICD therapy? Are there other benefits beyond termination of ventricular arrhythmias and defibrillation that an ICD might provide (i.e. arrhythmia detection)? As with any retrospective study, the findings reported by Cikes et al, though thought provoking, may not be ready for widespread application to all patients who are being considered for an LVAD. As we await more definitive evidence, it is best to apply a patient-specific approach such as implanting an ICD for the patient who clearly demonstrates right ventricular failure with ventricular arrhythmias, or not replacing an ICD generator in the patient who has never had ventricular arrhythmias before, particularly if there have been inappropriate shocks. In addition, careful screening of potential LVAD patients using arrhythmia detection devices should be performed as the presence of ventricular arrhythmias is associated with worse outcomes after LVAD implantation. While the question still remains whether or not to routinely place an ICD in an LVAD patient, a personalized approach that is based on careful assessment of arrhythmia risk and arrhythmia surveillance can hopefully prevent our patients from suffering either in retrospect or in the present. Conflict of interest: G.H. has received research grants from the National Institutes of Health (1KL2TR001444) and the American Heart Association (19CDA34760021), grant support from Abbott for work unrelated to this project, and has equity in Vektor Medical Inc., which is unrelated to this topic. The other authors have no disclosures.