ICDs were initially studied in patients who survived SCD or in those patients with sustained ventricular arrhythmias. From such secondary prevention trials, several findings suggested that patients with reduced systolic function had the greatest benefit from ICD implantation. In the AVID study, patients who survived SCD or had sustained ventricular arrhythmias were randomized to ICD or Class III anti-arrhythmic drug therapy [5]. Left Ventricular Ejection Fraction (LVEF) was 0.31 and 0.32 in the ICD and medical treatment arms, respectively. Fifty-five percent of the ICD group and 60 % of the medical therapy group had clinical heart failure symptoms. The primary outcome was a significant reduction in mortality for patients in the ICD arm. The benefit of ICD therapy was more pronounced among patients with LVEF <0.35. CIDS (Canadian Implantable Defibrillator Study) randomized patients with ventricular arrhythmias to ICD implantation versus amiodarone therapy [6]. The mean LVEF was 0.33 and 0.34 in the ICD and amiodarone groups, respectively, with half of both groups diagnosed with clinical heart failure. Subanalysis of the CIDS population revealed that depressed LVEF (<0.35) and advanced NYHA (New York Heart Association) heart failure class were independent predictors of benefit from ICD therapy [7]. CASH evaluated the effect of metoprolol, propafenone, amiodarone or ICD therapy in survivors of SCD [8]. All patients had clinical heart failure with mean LVEF ranging from 0.44 to 0.47 in the treatment arms. Again, ICD implantation provided a significant reduction in mortality compared with AADs. Furthermore, propafenone increased mortality leading to early cessation of the treatment arm. A meta-analysis of the AVID, CIDS and CASH trials yielded a 28 % relative risk reduction in mortality with ICD therapy compared with amiodarone therapy [9]. Furthermore, the difference in mortality was attributed to a 50 % reduction to arrhythmic deaths in the ICD group. There was no statistical difference between mortality for patients with LVEF >0.35; however among patient(s) with LVEF <0.35, there was a 34 % reduction in mortality with ICD therapy. Secondary prevention of SCD trials provided the basis for future studies investigating ICD implantation as a primary prevention therapy. Patients with reduced LVEF were identified from these studies as a sub-population with the potential for significant benefit from ICD implantation. Additionally, ICD therapy reduced mortality compared with AAD therapy supporting ICD implantation as first-line therapy for patients with history of SCD or ventricular arrhythmias.

Due to the results of secondary prevention trials, the focus of ICD investigation switched to primary prevention of sudden death. Summaries of primary prevention ICD trials are included in Table 16.2. The MADIT and MUSTT trials were the first large primary prevention trials evaluating the impact ICD therapy on mortality. In both studies, the population included those with coronary artery disease, LV systolic dysfunction, non-sustained ventricular tachycardia and inducible sustained ventricular tachycardia. In the MADIT trial, ICD therapy resulted in a 54 % reduction in mortality compared with AAD therapy. The mean LVEF was ~0.25 in the study population and two-thirds of the patients had clinical heart failure. Amiodarone consisted of 74 % of the AAD group and was associated with increased mortality compared with beta blocker therapy. Although ICD was not a treatment arm of the MUSTT trial, the results suggested that patients with inducible ventricular tachycardia at the time of electrophysiology study were at high risk for SCD (MUSTT). Patients with coronary artery disease, LVEF <0.40, and non-sustained ventricular tachycardia underwent electrophysiology study. Patients with inducible sustained ventricular tachycardia were randomized to medical therapy or a strategy of antiarrhythmia guided therapy. In the latter arm, they received antiarrhythmic drugs and if unsuccessful for rendering VT/VF non-inducible then an ICD was implanted. Patients who were non-inducible at baseline were included in a registry. Fourteen percent (49/353) of patients randomized to no AAD therapy after inducible ventricular tachycardia had syncope, sustained ventricular tachycardia or SCD leading to ICD implantation. This was markedly greater than the 3 % (49/1397) of patients that were non-inducible and had similar events leading to ICD implantation. These results suggested that depressed LVEF and inducible ventricular tachycardia were associated with a high risk of SCD. Furthermore, ICD therapy significantly reduced mortality in this population compared with conventional AAD therapy. However, total mortality was less strongly affected by inducibility of VT [10].

Several trials explored the effect of primary prevention ICD implantation at the time of coronary artery revascularization or shortly after myocardial infarction. CABG-PATCH assessed the impact of ICD implantation at the time of coronary artery bypass graft (CABG) surgery for patients at high risk for SCD [11]. There were 900 patients with reduced LVEF and abnormal signal averaged ECG randomized to CABG with or without ICD implantation. The mean age was 62 years old, LVEF was 0.27, and greater than 70 % had NYHA Class II/III heart failure. At the end of 4 years there was no statistical difference in mortality among the two groups. Subsequent analysis demonstrated that ICD implantation significantly reduced arrhythmic deaths between groups by 45 % [12]. However, due to high percentages of non-arrhythmic death in the population the protective effect of ICD therapy was not effective in reducing all-cause mortality. Furthermore, revascularization may have lead to improvements in LVEF that protected from arrhythmic deaths. The DINAMIT trial evaluated ICD implantation in a high risk population at the time of acute myocardial infarction [13]. Patients with recent (6–40 days) myocardial infarction who had reduced LVEF and autonomic dysfunction (reduced heart rate variability or elevated 24 h average heart rate) were randomized to optimal medical therapy with and without ICD implantation. In 676 patients, the mean LVEF was 0.28, >85 % had NYHA Class II/III heart failure and >70 % of patients suffered a new Q-wave myocardial infarction. All-cause mortality was not statistically different among the two groups; however, there was a statistically significant reduction in arrhythmic mortality in the ICD (HR 0.42). This was offset by an increase in non-arrhythmic deaths in the ICD group compared with control. The findings do not support implantation of primary prevention ICDs at the time of myocardial infarction. Furthermore this trial was the basis for the 40-day post-MI requirement for ICD implantation in the current guidelines. In a secondary analysis of DINAMIT, mortality in patients with ICD implantation differed greatly by whether they had received appropriate ICD shocks [14]. Patients in the ICD arm with appropriate ICD shocks had 15 % annual mortality compared with 6 % annual mortality in ICD patients with appropriate shocks. Effectively, the reduction in arrhythmic mortality observed in the ICD arm was negated by non-arrhythmic death in those ICD patients receiving appropriate ICD shocks. The DINAMIT results were supported by the IRIS trial where 898 patients with recent (5–31 days) myocardial infarction were randomized to conventional medical therapy with or without ICD implantation [15]. In the IRIS trial many more patients underwent PTCA (72 %) compared with the DINAMIT trial (27 %). There were similar rates of beta blocker and ACE/ARB use in both studies. All-cause mortality at an average of 37 months was not different in the ICD arm (HR 1.04) compared with control. Subgroup analysis demonstrated that patients with left main disease benefitted from ICD therapy whereas those with thrombolytic therapy fared better in the control group. There was a non-significant trend toward NYHA Class III/IV and smoking as predictors of ICD benefit. Similar to the DINAMIT trial, reduction in arrhythmic death in the ICD arm (HR 0.55) was offset by increase in non-arrhythmic death (HR 1.92). The results of the CABG-Patch, DINAMIT and IRIS trials found no mortality benefit of primary prevention ICD implantation in patients with reduced LVEF at the time of CABG or acute myocardial infarction.

MADIT II randomized 1232 patients with prior myocardial infarction and LVEF <0.30 to conventional medical therapy with and without ICD implantation [16]. Compared with MADIT, sustained VT or electrophysiology studies were not required for enrollment. The population was 85 % male with mean LVEF 0.23 in each group. Roughly two-thirds of the patient had at least NYHA Class II heart failure symptoms and 15 % were on AADs. There were 105/742 (14.2 %) and 97/490 (17.9 %) deaths in the ICD and conventional groups, respectively (p = 0.016), with a hazard ratio for ICD implantation of 0.69. The trial was stopped prematurely once superiority of ICD therapy was detected. The effect of ICD on survival was not affected by age, sex, LVEF or NYHA Class. These results supported ICD implantation for primary prevention of SCD in patients with reduced LVEF and ischemic cardiomyopathy, a majority of whom have symptomatic heart failure.

The next generation of primary prevention trials included patients with non-ischemic cardiomyopathy as well as ischemic cardiomyopathy. The CAT trial randomized 104 patients with reduced LVEF <0.30 and recent onset (<9 months) of dilated cardiomyopathy to conventional medical therapy with or without ICD implantation [17]. At a mean follow up of 5.5 years there was no statistical difference in mortality with ICD implantation. Of note, the trial was stopped prematurely due to lower than expected SCD event rates in the control arm at 1 year. AMIOVERT randomized 103 patients with non-ischemic cardiomyopathy, LVEF >0.35 and asymptomatic non-sustained ventricular tachycardia to amiodarone versus ICD implantation [18]. There was no statistical difference in mortality at 1 and 3 years with a non-statistical trend toward reduced arrhythmic burden in the amiodarone arm. The DEFINITE trial [20] enrolled 458 patients with non-ischemic cardiomyopathy and ventricular ectopy to optimal medical therapy versus optimal medical therapy plus ICD. The mean LVEF was 0.21, roughly 80 % had NYHA Class II/III symptoms and clinical heart failure duration was 2.8 years. Beta blocker and ACE/ARB use was ~85 % in the population. ICD therapy reduced all-cause mortality by 35 %, although this was not statistically significant (HR 0.65, CI 0.4–1.06). ICD did significantly reduce arrhythmic deaths (HR 0.2) without changing deaths related to heart failure. Subgroup analysis demonstrated that men and patients with NYHA Class III symptoms benefitted the most from ICD implantation. Age, LVEF and QRS duration had no significant impact on the outcome. All of these studies were underpowered so it is difficult to make conclusions regarding the role of ICD therapy in this population based on these trials.