Heart failure patients unresponsive to implantable cardioverter-defibrillator therapy: a neglected problem.

Abstract

After publication of the DANISH (Danish Study to Assess the Efficacy of ICDs in Patients with Non-ischemic Systolic Heart Failure on Mortality) trial,1 a debate has arisen in the cardiologists' community on the clinical benefit of implantable cardioverter-defibrillator (ICD) in the modern therapy of non-ischaemic heart failure (HF) patients,2 selected according to the current ICD guideline recommendations (i.e. ejection fraction value and HF functional class).3 The DANISH trial is a recent large randomized controlled trial that compared ICD therapy vs. optimal medical treatment in patients with non-ischaemic cardiomyopathy, and reported no mortality benefit of ICD therapy.1 In support of the current guidelines, recent meta-analyses showed instead a statistically significant clinical efficacy of ICD therapy by pooling DANISH results with those of the earlier studies carried out in the 2000s.4 In this debate, little attention has been paid to the different factors concurring to ICD clinical benefit and to their evolution over time. The clinical benefit of ICD therapy in patients with HF and low ejection fraction is mainly determined by: (i) the baseline risk of total mortality, cardiac death, and sudden cardiac death (SCD); (ii) the percentage of ICD-unresponsive patients (i.e. patients experiencing SCD despite ICD implantation). In particular, the unresponsiveness of patients to ICD therapy and its detrimental effects on ICD clinical benefit are a crucial, but fairly neglected problem.5 Recently, a significant reduction in total mortality, cardiac death, and SCD has been observed in patients with HF and low ejection fraction, thanks to the notable improvement in HF therapy.6 This was obtained by a broad use of beta-blockers, angiotensin-converting enzyme inhibitors, angiotensin-receptor blockers, and mineralocorticoid-receptor antagonists, combined with cardiac resynchronization therapy (CRT) in selected cases. Focusing on the risk of cardiac death, a risk reduction was observed for mortality due to either HF evolution and SCD. The temporal decline in SCD risk was corroborated by a recent study,6 which reported a 44% decrease in SCD risk in the time period from 1995 to 2014 by analysing the results of 12 clinical trials enrolling 40 195 patients with HF and depressed ventricular function. A reduction of SCD baseline risk can determine significant changes in ICD clinical benefit. Considering the simplest scenario (Figure 1), where the number needed to treat (NNT) for total mortality is used as a measure of clinical benefit and a null effect of ICD on death of non-sudden origin is assumed, a decrease in SCD baseline risk determines an hyperbolic increase in NNT values. For instance, considering a low number of ICD-unresponsive patients (25%, red curve), a decrease from 10% to 2% in SCD baseline risk determines a NNT increase from 13 to 67. Although the real clinical scenario may present higher complexity, where both non-sudden cardiac mortality and non-cardiac mortality may vary depending on HF severity and presence of co-morbidities, a considerable effect of SCD baseline risk on ICD clinical benefit is still expected. Although ICDs should theoretically prevent all SCD episodes, a percentage of patients experiencing SCD was observed in the ICD-implanted arm of all trials (Table 1). The occurrence of SCD in ICD patients can be ascribed to two main factors: one is the difficulty to define and assess SCD, and the other is related to the changes induced by novel HF therapies on arrhythmia properties. Considering the first aspect, according to the World Health Organization (WHO) criteria, SCD is defined as: ‘sudden unexpected death either within 1 h of symptom onset (event witnessed), or within 24 h of having been observed alive and symptom free (unwitnessed)’.7 This definition is clearly general and can include different clinical situations. Before 2000, almost 75% of the cases of out-of-hospital cardiac arrest were attributed to ventricular tachycardia/fibrillation (VT/VF). In particular, in a study that revised 90 ICD-unresponsive SCDs (28% of total mortality) in pre-clinical ICD trials, the mechanism of SCD was confirmed in 68 patients.8 The more frequently observed mechanisms of SCD were: VT/VF with post-shock electromechanical dissociation, VT/VF uncorrected by shocks, primary electromechanical dissociation, and incessant VT/VF. More recent studies have depicted a different scenario, where VT/VF are declining as a cause of SCD, being replaced by an increased incidence of pulseless electrical activity. As suggested by a Report from a National Heart, Lung, and Blood Institute Workshop published in 2013,9 the decline in VT/VF incidence may be partially attributed to the extensive use of aggressive pharmacological therapy in HF patients, and specifically to the use of beta-blockers. Indeed, pharmacological therapy may suppress VT/VF at the cost of a relative increase in cardiac arrests related to pulseless electrical activity. On the other hand, instances of sudden vascular death may be wrongly defined as SCD, due to the difficulty in distinguishing the two situations clinically.1 Catastrophic non-cardiac events, such as pulmonary embolus, dissecting aortic aneurysm, and intracerebral haemorrhage, may act as a sudden death inciting event. In a recent large countywide surveillance study, 525 WHO-defined SCDs among 20 440 deaths were reviewed based on autopsy-defined death cause.10 The study showed that nearly half of the presumed SCDs were not of arrhythmic origin. The progressive decline in the number of VT/VF-related SCDs, accompanied by the relative increase of SCDs of non-arrhythmic origin, may explain the increased fraction of ICD-unresponsive patients observed in the recent DANISH trial (53%) (Table 1). In addition to these factors, a small percentage of ICD-unresponsive patients may be ascribed to improper ICD functioning, due to e.g. ventricular under-sensing, unsuccessful final internal shocks, and/or lead failure. An appropriate ICD programming according to recent guideline recommendations11 is thus always advisable, although it is worth to notice that even the best programmed ICD may be ineffective on SCDs of non-arrhythmic origin. The effects of CRT on ICD clinical benefit need also to be clarified. CRT reduces SCD risk, and recent studies have reported very low rates of ventricular arrhythmias in CRT patients who showed a good response to this therapy.12 In the DANISH trial, no difference was found in the relative effect of ICD between patients who received CRT and patients who did not.1 Further studies are thus necessary to substantiate CRT effects on the rate of unresponsive patients. The fraction of ICD-unresponsive patients can have a crucial impact on ICD clinical benefit. As illustrated in Figure 1, an increase of unresponsive fraction (red to blue curve) determines an upward shift of the NNT curve (indicating a decrease in clinical benefit), with more severe effects for lower SCD risk. Indeed, at a baseline risk of SCD of 2%, an increase of ICD-unresponsive patients from 25% to 50% determines a NNT increase from 67 to 100 patients. The clinical scenario of ICD benefit in HF of non-ischaemic origin is exemplified in Table 1, where the mortality outcome data of the DEFINITE (Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation),13 SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial),14 and DANISH1 trials are reported. A progressive increase in the percentage of ICD-unresponsive patients occurred from the DEFINITE to the DANISH trials together with a reduction in the baseline annualized risk of SCD in the trial control arms. The two factors synergistically led to a progressive reduction of ICD clinical benefit, as assessed by the NNT for total mortality at 5 years. However, since non-ischaemic HF patients are expected to live longer thanks to current therapies, a precise evaluation of ICD benefit is conditioned upon the availability of longer follow-up data in future studies for the assessment of the long-term effects of SCD risk exposure. Despite clinical trial results may be not directly transferable to real-world clinical practice, the described clinical scenario depicts a decline in ICD clinical benefit in the modern therapy of HF of non-ischaemic origin. To improve ICD benefit, we believe that the time is ripe to ameliorate risk stratification criteria to identify patients at higher risk of VT/VF who may benefit most from ICD therapy. Accumulating evidence indicates the predictive power for arrhythmic events of ventricular fibrosis assessment, particularly by late gadolinium enhancement-cardiac magnetic resonance (LGE-CMR), in HF of non-ischaemic origin.15 Indeed, patients with ventricular fibrosis have a significantly higher risk for VT/VF compared to those without fibrosis. Ventricular fibrosis assessment by LGE-CMR may be integrated with other risk markers, such as ejection fraction and genetic profiling when available, in a poly-parametric evaluation of arrhythmic risk. This could significantly improve the appropriateness of ICD therapy in non-ischaemic HF patients in the direction of personalized and precision medicine.2 On the other hand, in presence of moderate quality of evidence, individualization of clinical decision-making and truly informed ‘consensus’ should be pursued. The complex decision path of ICD implantation may significantly benefit from detailed discussion with the patient to ensure the ultimate decision is consistent with his ‘preferences and values’. Currently, to make the best use of ICD we need to improve the identification of patient subgroups at higher risk of SCD, but also to reduce the fraction of ICD-unresponsive patients. This may be pursued by using risk markers able to selectively identify patients at higher risk of VT/VF, who can be profitably treated by ICD.2, 15 Conflict of interest: none declared.