Several randomised controlled trials (RCTs) have proven the efficacy of the implantable cardioverter-defibrillator (ICD) in preventing sudden cardiac death (SCD) in patients with advanced systolic heart failure (HF) and reduced left ventricular ejection fraction (LVEF).1–3 ICD therapy proved to be efficacious also in other conditions at high risk for arrhythmic death such as familial SCD syndromes.4
However, the choice of implanting an ICD may be challenging in patients with different clinical findings compared to that enrolled in the large randomised trials. An appropriate risk stratification is therefore sometimes difficult to perform, as in the asymptomatic patients with familial syndromes. In these settings, there is no clear evidence so far and an individualised approach seems to be the best way of assessing the risk–benefit of ICD therapy. Therefore, even after the publication of the available 2006 and 2008 practice guidelines,5,6 some of the questions concerning ICD therapy remain unresolved.
Many of the unanswered questions regarding the primary prevention of SCD include the optimal timing of ICD implant after myocardial infarction (MI) or after the first diagnosis of HF and the optimal LVEF cut-off able to identify high-risk patients. RCTs excluded patients in the acute phase of an MI, with a recent coronary revascularisation, New York Heart Association class IV symptoms or a newly diagnosed HF.7
It is well documented that following an MI8 or after initiation of medical therapy for recent-onset congestive HF,9 LVEF may improve over time and increase beyond the guideline range for ICD implantation. While there is convincing data that early implantation of an ICD after an MI does not improve outcomes10–12 and no survival benefit was shown in patients who received an ICD at the time of coronary artery bypass graft,13 data may be less convincing for other conditions. One post-hoc analysis,14 although not definitive, has shown that some patients with non-ischaemic cardiomyopathy may benefit from early ICD implantation. Little is known about the ICD therapy benefit in elderly patients (older than 75 years) with comorbidities such as renal dysfunction, chronic obstructive pulmonary disease and diabetes.
Timing for Implantable Cardioverter-defibrillator Implantation after Myocardial Infarction
Survivors of MI have an increased risk of SCD, particularly those with reduced LVEF (≤35 %).15–17 Randomised trials have investigated the use of an ICD in this setting for both primary and secondary prevention. Results of the secondary prevention trials have led to a general acceptance that ICD therapy improves the survival of most survivors of tachyarrhythmic cardiac arrest.18,19
In contrast to the fairly clear role of ICD therapy for secondary prevention, the optimal approach to the appropriate patient selection and timing of implantation for primary prevention has been more difficult to define. The most important risk parameter for all-cause mortality – as well as sudden arrhythmic death – is reduced LVEF.20 The reported sensitivity, specificity and positive predictive accuracy of a LVEF ≤40 % in predicting major arrhythmic events (fatal or life-threatening arrhythmias) were very variable in different studies, being 45–85, 55–75 and 9–24 %, respectively.21 The relevance of LVEF in patient selection is widely represented in current guidelines and recommendations for ICDs in primary prevention apply only to patients whose LVEF is low (≤30 or ≤35 %) despite receiving optimal medical therapy or <40 % if inducible ventricular sustained tachycardia is documented.5,6
The other variable to consider in obtaining the maximum benefit of ICD therapy is optimal implantation time. Despite the fact that the risk of SCD after an MI is highest during the first weeks after the event (as high as 14–24 % in the first month)22 and tends to decline after 12 months,10 the benefit of ICD therapy increases as the time passes from the myocardial ischaemic event to ICD implantation.23
The Defibrillators in acute myocardial infarction trial (DINAMIT) demonstrated that ICD implantation within 40 days after an acute MI in patients with LVEF ≤35 % does not reduce all-cause mortality, although sudden arrhythmic death was significantly diminished.12 The more recent Immediate risk stratification improves survival trial (IRIS) confirmed these results in patients with LVEF ≤40 % and additional risk factors for sudden death such as a heart rate of 90 or more beats per minute or non-sustained ventricular tachycardia (VT) (≥150 beats per minute) during Holter monitoring.11
In both trials of patients receiving an ICD early after an MI (within the first month), the risk of SCD was reduced by ICD therapy, but this effect was offset by an increase in the risk of non-SCD. Additionally, retrospective studies have suggested that ICD therapy may be of greatest benefit remotely after MI.23 The mechanisms that might have contributed to these apparently paradoxical findings remain unresolved and deserve further study.
Among the reasons to explain this paradox may be that the prognostic significance of EF when measured early after acute MI is strongly reduced in comparison to measurement months or even years after MI.24 EF often increases after initiation of therapy with beta adrenergic blocking agents and after recovery from the acute event,25,26 so that the time-dependent remodelling process modifies the organic and electrical substrate on which ICD therapy acts. Additionally, the benefit to survival obtained with early initiation of medical therapy in the post-MI period, but not with ICD implant, may suggest that treatments affecting remodelling are able to impact on sudden death and survival, whereas treatments focused solely on treating arrhythmias once they occur appear ineffective.
In general, as time passes, the association between baseline clinical characteristics and SCD seems to decrease. If predictors of SCD change with time after MI, a time-updated assessment may probably become more important than the baseline evaluation.27
Another consideration is that despite the fact that sudden death is increased early after MI, it is not synonymous with an arrhythmic event. If the increased risk of sudden death is not due to reversible lethal ventricular arrhythmias, an ICD could not have any impact on this type of death. It is possible that different risk stratifiers are necessary in identifying patients who will experience reversible ventricular tachyarrhythmias and who might benefit from an ICD, and not sudden death in general.10
On the basis of the data collected from randomised trials, current guidelines mandate at least a 40-day period following an MI before an ICD is implanted for a primary prevention indication. Measurement of EF, when used for risk stratification and SCD prediction, should therefore be performed after patients have recovered from the acute event. Unfortunately, the optimal timing in assessing EF to obtain a valuable predictive power remains undefined.
Role of Left Ventricular Ejection Fraction Reduction on Implantable Cardioverter-defibrillator Implant Indications
LVEF occupies a central role in the utilisation of ICDs for primary prevention today. This is because a low LVEF was shown to predict the risk of death after a MI.28 In addition, a qualifying LVEF is the one entry criterion common to all of the ICD studies. Unfortunately, each of the trials used a different single LVEF threshold (most often 35%) with a range of 30 to 40%. Additionally, the differences between the entry criteria and the mean or median values for patients who were actually enrolled in trials were large (7–10 %).7 For example, in the Sudden cardiac death in heart failure trial (SCD-HeFT)3 the LVEF threshold for entry into the trial was 35 %, but the enrolled patients had a median LVEF of 25 %, with an interquartile range of 20–30 %. A subgroup analysis of participants who had LVEF higher than 30 % suggested no benefit from ICD therapy.7 A similar trend was seen in the Multicenter automatic defibrillator implantation trial (MADIT) and MADIT II study populations.1,29
Despite the upper LVEF limit of every trial is the basis for current treatment recommendations, strong evidence supporting the employment of ICD in primary prevention in post-MI patients whose LVEF was near to the upper limit of entry criterion (31–35 %) is currently lacking. In fact, these groups of patients were under-represented, received no benefit or received an uncertain benefit in RCTs performed so far. Impaired LVEF remains a significant contributor to risk for SCD in patients with coronary disease, as well as a variety of non-ischaemic cardiomyopathies. However, when used as the sole criterion to guide prophylactic ICD therapy, it has significant limitations. It seems likely that the relative weight accorded to EF in quantifying risk for SDC will vary depending on the underlying disease substrate, as well as the stage of progression of the disease.
Implantable Cardioverter-defibrillator Therapy in Elderly Patients and the Role of Comorbidities
Patients enrolled in RCTs may significantly differ from the ‘real-life’ community patients in age, gender distribution, disease severity, comorbidity burden and intensity of ancillary therapy.
Starting with age, uncertainty remains whether advantages on survival obtained with ICD therapy are equally applicable in all age groups or if the survival benefit may instead reasonably wane as age increases. This has become a critically important issue in light of the increasing growth in the elderly population in developed countries.
Unfortunately, the average age of the clinical trials’ study populations has tended to be relatively young. For example, the Antiarrhythmics versus implantable defibrillators (AVID),18 Multicenter unsustained tachycardia trial (MUSTT)30 and Defibrillators in non-ischemic cardiomyopathy treatment evaluation (DEFINITE)2 populations, each of which assessed patients having documented ventricular arrhythmias leading to ICD therapy, averaged 65, 66 and 58 years of age, respectively. In the MADIT trial29 the ICD group mean age was 62±9 years, suggesting that many patients were >70 years, but not likely very many >75 years. In the MADIT II trial1 the populations were 64±10 for the ICD group and 65±10 for the conventional treatment arm. In SCD-HeFT3 the mean age was even lower than MADIT or MADIT II, being about 60 years in all groups.
Thus, many patients included in RCTs were close to 70 years, but only a small proportion of them were older than 75 years. It is currently unclear if the observations derived from these important studies can be reliably extended to older populations. Data coming from a MADIT-II substudy31 evaluated the mortality benefit from defibrillator therapy in eligible elderly patients as compared to younger patients. Among patients enrolled in the trial, 204 were older than 75 years and 121 of them underwent defibrillator implant. After a mean follow-up of 17.2 months a 44 % reduction in risk of mortality was shown between patients ≥75 years old with ICD implant compared with those elderly in conventional therapy (hazard ratio [HR] 0.56; 95% confidence interval [CI] 0.29–1.08), but this result was marginally significant (p=0.08). Comparatively, the HR in patients <75 years assigned to defibrillator implant was 0.63 (0.45–0.88; p=0.01) after 20.8 months. Other studies showed no substantial differences in survival benefit between different ages subgroups,32 while a recent meta-analysis performed on data coming from MADIT-II, DINAMIT, DEFINITE, SCD-HeFT and IRIS primary prevention trials showed a major effect of age on reducing prophylactic ICD benefit.33 However, none of these studies, subgroup analyses and meta-analyses could be interpreted as definite proof of presence or absence of ICD benefit because the statistical power of such analyses may be insufficient, involving a small number of patients or being based on partial data deriving from a few studies.
The presence of comorbidities influences the benefits of ICD implantation on survival and a definite role is played by the presence and severity of renal dysfunction. The presence of renal insufficiency adjusted for other covariates, including LVEF, has been shown to be the single strongest predictor of mortality in the HF patient population implanted for primary prevention.34,35 A higher mortality was clearly linked to the degree of renal dysfunction and the impact of primary prevention ICD therapy in patients with a moderate-to-severe renal impairment seems to be limited. Retrospective analysis of the MADIT II trial examining the relationship between renal function, risk of SCD and benefit of the ICD in ischaemic LV dysfunction, found that survival benefit was lost in patients with estimated glomerular filtration <35 ml/min/m2.36 No survival advantages were shown in dialysis-dependent patients in smaller studies.37
The role of other common comorbidities such as chronic obstructive pulmonary disease (COPD) and diabetes mellitus has not been systematically studied, despite the benefits in decreasing total mortality can be affected dramatically by the presence of these conditions. There is almost a complete absence of clinical data confirming the survival benefit conferred by the ICD in patients who have left ventricular dysfunction and severe COPD. Retrospective analyses found a preserved survival benefit of the ICD in the presence of COPD38 but only a small number of patients were considered and no information on COPD severity was available for all of them. In a subgroup analysis of patients enrolled in the MADIT II trial (representing 35% of the overall cohort, of whom 40 % were treated with insulin), diabetic patients derived a similar benefit from ICD therapy despite being sicker and having a higher mortality rate overall compared with non-diabetics.39
In conclusion, at the present time there is no clear evidence to establish definitive limitations to ICD implantation in terms of age or comorbidities. A careful consideration is suggested for patients >75 years old and for those with more advanced stages of renal disease, because in these groups, the benefit of ICD is uncertain. An individualised strategy seems to be the best approach in these circumstances.
Implantable Cardioverter-defibrillator Therapy for Primary Prevention in Familial Sudden Death Syndromes
The relative rarity of the familial sudden death syndromes (ion channelopathies, hypertrophic cardiomyopathy [HCM] and arrhythmogenic right ventricular cardiomyopathy [ARVD]) seriously limits the amount and quality of the data upon which ICD benefit can be evaluated. Therefore, the clinical decision-making process can be very difficult and in many areas uncertainty remains in evaluating the risk–benefit ratio of ICD implant in these often young patients.4
The available guidelines reflect these uncertainties and class I indications (ICDs definitely indicated) are largely restricted to patients who have already experienced malignant ventricular arrhythmias or sudden cardiac arrest.
In the long-QT syndromes (LQTS), no prospective studies are available on the risk of recurrent cardiac arrest in patients with aborted sudden death and not treated with ICDs, because such individuals are thought very likely to experience recurrences of arrhythmic events. However, recent studies40 suggest that the outcomes of these groups of LQTS patients may be good if appropriately treated with beta-blockade. Furthermore, if ICD therapy should be considered in patients experiencing syncope during beta-blocker treatment (with a class IIa indication),41 it is less clear if this approach is equally advantageous in patients experiencing syncope before the start of treatment.
In Brugada syndrome, considerable controversy concerns the management of the asymptomatic patients with a spontaneous type 1 electrocardiogram (ECG) abnormality. While there is a general acceptance that patients with previous syncope should undergo ICD implantation, discordant results were found in risk stratification of asymptomatic patients42,43 and the value of electrophysiological studies to predict sudden death in this setting is unclear and unresolved.
Recommendations for ICD implantation in patients with HCM are based not on randomised trials but on observational data. The American College of Cardiology (ACC)/American Heart Association (AHA)/European Society of Cardiology (ESC) guidelines5 state that ICD implantation can be effective (a class IIa recommendation) in patients with one or more of the recommended major risk factors for SCD (previous cardiac arrest, spontaneous sustained VT, family history of SCD, unexplained syncope, LV thickness ≥30 mm, abnormal exercise BP response, non-sustained spontaneous VT), but the predictive value of every single risk factor is not homogeneous. With the exception of prior cardiac arrest,44 there is very little evidence that any one risk factor is more predictive than another.
While the presence of multiple risk factors appears straightforward and not controversial to consider ICD for primary prevention, the presence of a single risk factor is associated with an SCD annual risk of about 1 % and the life-long morbidity and mortality risks of ICD implantation may outweigh the benefit in this young patient population. Similar considerations can be formulated in ARVD asymptomatic patients with mild disease. This group seems to be at low risk of sudden death45 and therefore the risks of ICD therapy may outweigh the benefits.
A potential aid in the risk stratification of patients with familial SCD syndromes could come from genetic analysis.46,47 The identification of specific mutations linked to a particularly high SCD occurrence may translate into improved prediction of risk facilitating the decision-making process in ICD implantation, especially in asymptomatic patients. Unfortunately, the progress in identifying such genetic markers is very slow and the value of genetic variations important in contributing to SCD susceptibility needs to be validated in large prospective community-based case-control studies. Designing and conducting these studies in these relatively rare conditions seems to be very challenging.
ICD therapy is the most effective therapeutic modality for SCD prevention. Information deriving from large RCTs has led to the identification of patients that may benefit most from this therapy and recommendations given in current guidelines are the results of these investigations. However, an individualised approach remains often the only available option when the patient characteristics differ from that of clinical trials and in a setting in which evidence is not solid enough, such as asymptomatic patients with rare familial SCD diseases. Further research should therefore move towards an improved risk stratification based not only on the role of already known parameters, such as LVEF or electrophysiological studies’ results, that seems to be insufficient in a variety of clinical settings, but also on the identification of individualised, new predictors of SCD, such as genetic profiles or other meaningful biomarkers.