5117Lacking treatment effect of ICD in heart failure patients lacking ischemic heart disease and age? A meta-analysis and meta-regression focusing on moderators and the DANISH trial

Abstract

Background: Due to sparse evidence, a Danish recommendation for or against implantable cardioverter defibrillator (ICD) therapy in heart failure (HF) patients without ischemic heart disease (IHD) has awaited the results of the DANISH trial. Published September 2016, these showed that ICD therapy in this patient group did not significantly reduce all-cause mortality, while a subgroup analysis suggested that younger patients may benefit more. Whether HF patients without IHD should be offered an ICD therefore largely remains a patient to patient decision. Purpose(s): To explore effects of age, IHD and other potential moderators on the treatment effect of ICDs in HF patients, allowing for better clinical decision making. Method(s): A literature search for multicenter randomized controlled trials (RCTs) documenting the effect of ICD on mortality in HF patients was performed in the PubMed database. Meta-analysis using the random effects model (DerSimonian- Laird) calculated the relative risk ratio (RR) as the effect size based on the RCTs. Meta-regressions using the mixed-effects model explored whether the following baseline characteristics of the studies were potential moderators of treatment effect: Patient age, proportion with IHD, male gender, diabetes, atrial fibrillation, concurrent medication (beta-blocker, ACE inhibitor/angiotensin II receptor blocker (ARB), aldosterone receptor antagonist, digoxin, diuretic, antiplatelet drug), left ventricular ejection fraction, year of publication, sample size, and follow-up. A sensitivity analysis excluding DANISH was performed. Result(s): 9 multicenter RCTs were identified: MADIT (1996), CABG Patch Trial (1997), MADIT II (2002), DEFINITE (2004), COMPANION (2004), DINAMIT (2004), SCD-HeFT (2005), IRIS (2009), DANISH (2016). The meta-analysis found a significant 15% reduction in the risk of death (RR=0.85, 95% confidence interval (CI)=0.73-0.98). Moderate heterogeneity (I2=62%) allowed for meta-regressions, which found that neither avg. age (mean 62.6 years, range 58.3-66.5) nor proportion with IHD (mean 65.4%, range 0%-100%) were significant moderators, but that concurrent medication with beta-blockers, ACE inhibitors/ARBs, diuretics and antiplatelet drugs were, with diminishing treatment effects with higher proportions of patients on the medications (see figure). The sensitivity analysis found an RR of 0.83 (95% CI=0.71-0.98) and did not meaningfully alter the meta-regression findings. Conclusion(s): We found no evidence suggesting that age or IHD were significant moderators on the treatment effect of ICDs on mortality in HF patients. Instead, we found that concurrent medications incl. beta-blockers and ACE inhibitors were significant moderators. Since the study population in DANISH had a very impressive baseline medication with these drugs (92% and 97% respectively), this begs the question of whether the negative result of DANISH was due to effective guideline-based medical therapy at baseline rather than the absence of IHD. (Figure Presented).