Early mortality in implantable cardioverter defibrillator patients: from randomized controlled trials to real life

The implantable cardioverter defibrillator (ICD) has been a story of success over the last three decades. The defibrillator has proven highly effective in termination of ventricular tachyarrhythmias. A plethora of clinical trials have shown survival benefit for patients at risk for life-threatening arrhythmias, resulting in a steady increase in the use of ICDs. 1 , 2 Implantable cardioverter defibrillator guidelines are updated frequently reflecting widening indications. Nevertheless, the reduction in mortality related to ICD depends only on the termination of tachyarrhythmias, regardless of the clinical condition or specific circumstances that may contribute to overall mortality. Most patients at risk for sudden cardiac death suffer from impaired ventricular function and have increased risk of heart failure-related mortality. As resources are limited and cost-effectiveness has to be proven, especially in expensive therapies such as ICD, it is critical to identify patients who will benefit most from the device. In daily routine, some patients could either be managed safely by alternative therapies or might be too sick to benefit from ICD therapy. Prediction of this benefit remains one of the major issues. Presently, only limited data are available reporting predictive variables for either arrhythmic events or mortality from non-arrhythmic events and heart failure in ICD patients. Stein et al.3 reported a large prospective multicenter follow-up of an unselected cohort of new single or dual-chamber ICD recipients. Over 20 parameters in over 1700 patients from 128 US centres were evaluated to establish the survival benefit during the first year after implantation. Such registries provide further information regarding mortality in everyday practice and under routine circumstances. Defining the causes of death usually is difficult in registries and retrospective analyses. Death of unknown cause ranged between 12 and 20% in previous reports 4 , 5 and death circumstances were also unknown in 33% of cases in this paper. 3 Stein et al.3 reported a 1-year mortality of 16%. Does this striking mortality rate in the first year after implantation reflect the real life? Mortality rates in non-randomized secondary prevention cohorts were higher than in randomized controlled trials and ranged between 6.5 and 13%, although the ejection fraction was slightly better and patients seemed to have a better NYHA class ( Table 1 ). Low mortality (6.5%) in the German EURID registry may reflect the inclusion of patients with primary electric disorders. 6 The true mortality in patients with structural heart disease is probably higher. Possible explanations for the relative high mortality rate are comorbidities such as atrial fibrillation (AF) and/or impaired renal function. This is discussed by the authors as the ‘recruitment bias’; in addition, there is probably a ‘participation bias’ in large multicenter trials that may account for a treatment difference. Characteristics of ICD patients assigned to randomized trials and registries/unrandomized cohorts Characteristics of ICD patients assigned to randomized trials and registries/unrandomized cohorts Early mortality was lower in primary prevention trials than in trials that included only survivors of life-threatening arrhythmias. In the large defibrillator trials, early mortality ranged between 2.6 and 9%. Implantable cardioverter defibrillator shocks were more likely in secondary prevention trials. 1 , 2 In contrast, mortality did not differ between the two groups in a recently published study, in which it was very low (4.2%). 7 The study by Stein et al.3 mainly included patients with ICD for secondary prevention (93%). The availability of similar data from MADIT II patients and other primary prevention studies would have even more clinical implication nowadays. We are therefore looking forward to obtaining comparable data on mortality in a cohort of patients with primary prevention and biventricular devices. In the SERF registry, heart failure was one of the main causes of early death in ICD patients. The overall benefit from ICD therapy in secondary prevention trials is the higher the lower the ejection fraction. 2 This is equally true in primary prevention: the MADIT II data revealed that benefit was higher in patients with lower ejection fraction. 8 However, hospitalization for heart failure is associated with a higher mortality rate. Once heart failure needs intervention, mortality rises. The SCD-HeFT study 2 showed no significant risk reduction with the defibrillator in a subgroup of patients in NYHA class III. Data on the impact of the NYHA class on outcome in ICD recipients are not consistent. Moreover, the patient's functional class does not correlate very well with the ejection fraction, although both parameters are independent risk factors for death and arrhythmic events. Hence, in terms of heart failure death, neither ejection fraction nor NYHA class alone seems to be the best predictor of early mortality. Perhaps, a combination of other variables, such as the use of diuretics and digitalis, low mean arterial pressure, etc., best defines the severity of heart failure and identifies patients at high risk of death, despite ICD therapy. Noteworthy, diuretics and digoxin were associated with a 2-fold increase in mortality in other retrospective analysis. 9 The low body mass index and physical inactivity that resisted even multivariate analysis in the present study are also likely to be an expression of advanced heart failure and cardiac cachexia. The ‘obesity paradox’ in terms of mortality may reflect the inability of the NYHA class to describe symptoms exclusively due to heart failure. However, obesity may significantly contribute to long-term cardiovascular mortality and thus may have only an early protective effect. The best parameter for determining the ‘point of no return’ or need for other treatments (e.g. assist devices) is still unclear. Neither NYHA class nor ejection fraction or medications are presently a sufficient basis for withholding ICD therapy. Unfortunately, no information on the QRS width is available in this study which might help to estimate how many patients would have been candidates for biventricular pacing. This could have reduced heart failure-related death, even in the presence of a relatively high rate of non-responders. How about AF? There is a known negative influence on the progression of the left ventricular dysfunction due to AF. Consecutively, mortality attributed to pump-failure was higher in patients with AF, but we should not forget that the presence of AF may not only be a possible cause of poor outcome but also (just) a marker of a more advanced heart disease with a higher risk for decompensation. 10 Noteworthy, the presence of AF was not associated with a higher risk of arrhythmic death in a retrospective analysis of the SOLVD study. 11 It would also be of interest to know if catheter ablation of AF would have influenced outcome. Finally, some other issues concerning prediction of early death after ICD implantation are of interest. Renal function is known to have tremendous impact on cardiovascular outcome. Some studies suggested an almost 4-fold risk of early mortality, if serum creatinine exceeded 1.8 mg/dL. 4 Furthermore, in the SERF registry, one-third of patients were diabetic. This was similar to other multicenter trials. The negative effect of diabetes on cardiovascular outcome is well known. Surprisingly, diabetes was accountable for early mortality in ICD patients, although one may have expected more impact on later survival. Information regarding the quality of medical treatment (e.g. control of HbA1c) or amount of intensified insulin therapy would have been of interest. Identifying patients at risk of dying early after ICD implantation without benefit of device therapy is presently still a challenge. Cooperation between heart failure specialists and electrophysiologists will optimize translation of trial results to everyday routine. The most reasonable approach to optimize ICD therapy is the improvement of risk stratification with studies such as the one presented by Stein et al. 3