The American College of Cardiology/American Heart Association (ACC/AHA) guidelines for the diagnosis and management of heart failure (HF) in adults state that “the addition of an angiotensin receptor blocker [ARB] may be considered in persistently symptomatic patients with reduced left ventricular ejection fraction [LVEF] who are already treated with conventional therapy.”1 This was regarded as a class IIb recommendation with a level of evidence B. In contrast, the 2008 European Society of Cardiology (ESC) guidelines consider the addition of an ARB “in patients with heart failure and an LVEF of equal or less than 40% who remain symptomatic despite optimal treatment with an angiotensin-converting enzyme [ACE] inhibitor and a beta-blocker” as a class I recommendation with a level of evidence A.2 The recently updated 2012 ESC guidelines, however, have revised the above recommendation restricting it to patients who are unable to tolerate a mineralocorticoid receptor antagonist and for the prevention of HF hospitalization.3 Both sets of guidelines are unanimous in that they strongly recommend adding an aldosterone antagonist in symptomatic patients but warn against the triple combination of an ACE inhibitor, ARB, and an aldosterone antagonist. Neither of the guidelines recommend direct renin inhibitors (DRIs). These recommendations leave the practicing clinician in a predicament of how exactly to proceed in a HF patient who remains symptomatic despite optimal conventional therapy. Since a triple blockade of the renin angiotensin aldosterone system (RAAS) is not considered safe, should the next step be an aldosterone antagonist,4,5 an ARB, or a DRI?
In patients on conventional heart failure therapy including angiotensin-converting enzyme (ACE) inhibitors, the addition of angiotensin receptor blockers (ARBs), direct renin inhibitors (DRIs), or aldosterone antagonists are therapeutic options to further reduce the risk of cardiovascular events. However, whether one is preferable over the other is not known. PubMed, EMBASE, and Cochrane Central Register of Controlled Trials databases were searched for randomized clinical trials (RCTs), until March 2011, of trials testing either an ARB, DRI, or an aldosterone antagonist in patients with heart failure who were on conventional heart failure therapy with follow-up of at least 3 months. Efficacy (death, cardiovascular death, nonfatal myocardial infarction, heart failure hospitalization and composite of cardiovascular death or heart failure hospitalization) and safety (hyperkalemia, hypotension, renal failure) outcomes were compared. The authors identified 16 RCTs involving 31,429 participants that satisfied the inclusion criteria. When compared with placebo (reference rate ratio [RR] of 1), aldosterone antagonists reduced the rate of death (RR, 0.79; 95% credibility interval [CrI], 0.66–0.98), cardiovascular death (RR, 0.78; 95% CrI, 0.65–0.93), heart failure hospitalization (RR, 0.74; 95% CrI, 0.55–0.94), and the composite of cardiovascular death or heart failure hospitalization (RR, 0.73; 95% CrI, 0.55–0.90) with no difference for other efficacy outcomes. However, ARBs and DRIs did not result in any significant reduction in the rate of any of the efficacy outcomes when compared with placebo. When compared with placebo (RR=1), ARBs increased the rate of hyperkalemia (138% increase), renal failure (126% increase), and hypotension (63% increase). Similarly, aldosterone antagonists resulted in a 110% increase in hyperkalemia and DRIs with a 98% increase in hypotension. In patients with heart failure and reduced systolic function on conventional heart failure medications, the risk benefit ratio favors the addition of aldosterone antagonists over ARBs or DRIs.
We conducted PubMed, EMBASE, and Cochrane Central Register of Controlled Trials (CENTRAL) searches for randomized clinical trials (RCTs) using the terms: “angiotensin receptor blockers,”“angiotensin receptor antagonists,”“ARBs,”“aldosterone antagonists,”“direct renin inhibitors,”“DRI,” and using the names of individual ARBs, aldosterone antagonists, or DRIs, in patients with HF, until March 2011 (week 2). There was no language restriction for the search. We checked the reference list of published review articles, meta-analyses, and randomized trials identified by the searches to find other eligible trials. We excluded studies that were only in abstract format without a manuscript published. In addition, for studies that did not report the outcomes of interest, we contacted the authors via e-mail. We also searched Food and Drug Administration (FDA) dockets by performing a hand search of all documents submitted for drug approval/labeling change as well as the FDA meeting minutes available on the FDA Web site.
Eligible trials had to fulfill the following criteria: (1) RCTs comparing the above drug classes; (2) cohort with HF and reduced systolic function on conventional HF therapy (including diuretics, β-blockers, and angiotensin-converting enzyme inhibitors); (3) follow-up of at least 3 months; (4) enrolling at least 100 patients; and (5) able to report the outcomes of interest (below).
Selection and Quality Assessment
Two authors (S.B. and S.K.) independently assessed trial eligibility and trial bias risk and extracted data (κ=0.96). Disagreements were resolved by consensus. The bias risk of trials were assessed using the components recommended by the Cochrane Collaboration:6 sequence generation of the allocation; allocation concealment; blinding of participants, personnel, and outcome assessors; incomplete outcome data; selective outcome reporting; and other sources of bias. Trials with high or unclear risk for bias for any one of the first 3 components were considered as trials with a high risk of bias. Otherwise, they were considered as trials with a low risk of bias.
Data Extraction and Synthesis
The following outcomes were evaluated: efficacy outcomes were death, cardiovascular (CV) death, nonfatal myocardial infarction, stroke, HF hospitalization, and the composite of CV death or HF hospitalization. Safety outcomes evaluated were hyperkalemia, renal failure, and hypotension.
Mixed Treatment Comparisons Mixed treatment comparison methods on all available networks of treatment comparisons were used to compare the different treatment regimens using WinBUGS code published by the UK Medical Research Council Health Services Research Collaboration.7 The network meta-analysis allows for comparisons of agents not directly addressed within any of the individual trials such that in addition to analyzing the direct within-trial comparisons between two treatments (such as treatment A vs B), the network framework enabled us to incorporate the indirect comparisons from two trials that have one treatment in common (such as comparison of treatment A vs C using trials comparing A vs B and B vs C). The analysis has the advantage of maintaining the within-trial randomized treatment comparison of each trial while combining all available comparisons between treatments. Moreover, the network meta-analysis produces tighter confidence intervals, implying greater precision of the estimates.
For the purpose of analysis, the trials were grouped into four categories: placebo, ARBs aldosterone antagonists, and DRIs. The analysis compared ARBs, aldosterone antagonists, or DRIs with placebo, which was used as the reference. A random-effects Poisson regression model was fitted after taking into account the correlation structure induced by the multi-arm trials.8 The analysis used the rate of efficacy and safety outcomes per 1000 person-years to obtain the log rate ratios of one treatment relative to another treatment. Rates, rather than number of events, were considered the most appropriate outcome for this analysis because they incorporate the duration of the trials. Patient-years of follow-up were calculated by multiplying the sample size with the mean follow-up duration.
Calculation of the probability that each treatment is best (lowest event proportion) was performed using a Bayesian Markov chain Monte Carlo method, adapted to apply to a connected network set of treatment comparisons. Minimally informative prior distributions were used for comparisons of treatments, so the findings and interpretation are close to those obtained with frequentist methods. All network analyses were conducted using WinBUGS 1.4.3.
Sensitivity analyses were performed after excluding trials that included patients with post–myocardial infarction left ventricular systolic dysfunction.
Role of the Funding Source
This work was not funded and hence there was no role of any funding source in the conception, data synthesis, analysis, data interpretation, or drafting of the manuscript.
We identified 16 RCTs that satisfied our inclusion criteria (Figure 1). The network of treatment comparisons is shown in Figure 2. Trials that compared ACE inhibitor+ARB vs an ACE inhibitor were considered as ARB vs placebo trials for the purpose of this analysis (the Randomized Evaluation of Strategies for Left Ventricular Dysfunction [RESOLVD], Valsartan in Acute Myocardial Infarction [VALIANT], Yashumura and colleagues trials). We excluded the ARBs-only arm of the RESOLVD and VALIANT trials because these patients were not taking ACE inhibitors by design. Of the 16 trials, 6 trials compared ARBs vs placebo, 8 trials compared aldosterone antagonists vs placebo, and 2 trials compared DRIs vs placebo. The network of available treatment comparisons is shown in Figure 1. Of these, 4 trials (the Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study [EPHESUS], VALIANT, Weir and colleagues, Aliskiren Study in Post-MI Patients to Reduce Remodeling [ASPIRE] trials) enrolled patients with post–myocardial infarction left ventricular systolic dysfunction. This later group was excluded in a sensitivity analysis.
Baseline characteristics and quality analysis are summarized in Table I and Table II. The 16 RCTs enrolled 31,429 participants who were followed up for a mean of 16.2 months (range: 3–41 months). The average age was 63±11 years, and an average left ventricular ejection fraction of 31%±7%. Table II shows the conventional HF treatment used in the included trials. While the proportion of patients taking ACE inhibitors was high in most of the trials, the proportion of patients taking β-blockers was variable and dismally low in a few trials. Similarly, the use of implantable cardioverter-defibrillators (ICDs) or cardiac resynchronization therapy (CRT) was low in the few trials that reported these data (Table II).
|Study||Year||Total, No.||Treatment Groups||Follow-Up, mo||Mean Age, y||Men, %||Diabetes, %||Ischemic Etiology, %||NYHA Class III/IV, %||LVEF, %|
|ALOFT14||2008||302||Aliskiren vs placebo||3||68||78||31||55||38||31|
|AREA IN-CHF15||2009||467||Canrenone vs placebo||12||63||84||21||NA||0||40|
|ASPIRE16||2011||820||Aliskiren vs placebo||9||60||83||23||100||NA||38|
|CHARM-added11||2003||2548||Candesartan vs placebo||41||64||79||30||62.5||76||28|
|Cice et al17||2010||332||Telmisartan vs placebo||35.5||63||89||29||NA||66||30|
|Cicoira et al18||2002||106||Spironolactone vs placebo||12||62||87||NA||64||NA||33|
|EMPHASIS-HF12||2010||2737||Eplerenone vs placebo||21||69||78||32||69||0||26|
|EPHESUS19||2003||6632||Eplerenone vs placebo||16||64||71||32||100||NA||33|
|Gao et al20||2007||116||Spironolactone vs placebo||6||55||65||35||51||87||NA|
|RALES21||1999||1663||Spironolactone vs placebo||24||65||73||NA||55||99||25|
|RESOLVD22||1999||441||Enalapril+candesartan vs enalapril||11||63||84||NA||71||36||28|
|Udelson et al23||2010||226||Eplerenone vs Placebo||9||63||84||39||61||26||27|
|Val-HeFT24||2001||5010||Valsartan vs placebo||23||62||80||25.5||57||38||27|
|VALIANT25||2003||9794||Captopril+valsartan vs captopril||24.7||65||69||23||100||NA||35|
|Weir et al26||2009||100||Eplerenone vs placebo||6||59||77||25||100||NA||34|
|Yasumura et al27||2004||135||ACE inhibitor + ARB vs ACE inhibitor or ARB||6||65||80||NA||43||NA||NA|
|Study||Baseline ACE Inhibitor,a%||Baseline BB, %||Baseline Diuretics, %||Baseline Digitalis, %||Baseline ICD/CRT, %||Risk of Biasb|
|Cice et al17||100||61||NA||51.5||NA||+++|
|Cicoira et al18||100||69||NA||NA||NA||+−±|
|Gao et al20||99||56||100||98||2.5||+±+|
|Udelson et al23||97||95||71||NA||NA||+++|
|Weir et al26||12||18||NA||NA||NA||+±+|
|Yasumura et al27||82||48||NA||NA||NA||+−−|
When compared with placebo (reference rate ratio of 1), aldosterone antagonists reduced the rate of death (21% reduction) (Figure 3a), cardiovascular death (22% reduction) (Figure 3b), HF hospitalization (26% reduction) (Figure 3c), and the composite of cardiovascular death or HF hospitalization (27% reduction) (Figure 3d), with no difference for other efficacy outcomes (data not shown). However, ARBs did not result in any significant reduction in the rate of any of the efficacy outcomes (Figure 3a–d). DRIs, when compared with placebo, did not result in any significant reduction in the rate of any of the efficacy outcomes (Figure 3a–d). In the head-to-head comparisons of active comparators, there was no difference for any of the efficacy outcomes for any combination of comparators, although the point estimate favored aldosterone antagonists compared with either ARBs or DRIs (Figure 3a–d).
Table III summarizes the rate per 1000 patient-years of follow-up and the probability that each treatment is the best (lowest event proportion). When compared with placebo, aldosterone antagonists resulted in 9 fewer deaths, 6 fewer cardiovascular deaths, 25 fewer HF hospitalizations, and 44 fewer CV deaths or HF hospitalizations (Table III) per 1000 patient-years of follow-up, and there was >70% to 90% probability that aldosterone antagonists were the best (lowest event proportion) compared with ARBs, DRIs, or placebo (Table III) for these outcomes.
|Treatment||Death Rate |
Rate (95% CrI)
|Probability Best, %||CV Death |
Rate (95% CrI)
|Probability Best||HF Hospitalization |
Rate (95% CrI)
|Probability Best, %||CV death or HF Hospitalization |
Rate (95% CrI)
|Probability Best, %|
|Placebo||40.69 (25.61–66.33)||0.7||33.22 (19.62–57.77)||0.2||95.42 (77.49–113.3)||0.2||159.6 (141.7–178.6)||0.1|
|ARBs||39.06 (23.74–65.24)||4.9||30.94 (17.95–55.20)||6.2||79.82 (60.56–102.9)||12.9||148.8 (119.2–190.1)||3.1|
|AA||32.33 (19.49–54.58)||89.1||25.91 (15.06–46.07)||89.9||69.97 (50.13–92.21)||68.8||116.0 (85.8–145.3)||86.9|
|DRIs||69.53 (25.59–191.6)||5.3||65.24 (20.91–215.1)||3.7||149.6 (21.77–965.5)||18.2||175.0 (94.5–316.0)||9.9|
When compared with placebo (reference rate ratio of 1), ARBs increased the rate of hyperkalemia (138% increase) (Figure 4a), renal failure (126% increase) (Figure 4b), and hypotension (63% increase) (Figure 4c). Similarly, aldosterone antagonists resulted in a 110% increase in the rate of hyperkalemia, with no significant increase in the rate of either renal failure or hypotension when compared with placebo, although the point estimates suggested similar increased risk (Figure 4a–c). In addition, DRIs were associated with a 98% increase in the rate of hypotension, although there was no significant increase in the rate of either hyperkalemia or renal failure when compared with placebo although the point estimates suggested a similar increased risk (Figure 4a–c). In the head-to-head comparisons of active comparators, there was no difference for any of the safety outcomes for any combination of comparators (Figure 4a–c).
Table IV summarizes the event rate per 1000 patient-years of follow-up and the probability that each treatment is the best (lowest event proportion). When compared with placebo, ARBs resulted in 22 excess cases of hyperkalemia, 15 excess cases of renal failure, and 18 excess cases of hypotension (Table IV) for every 1000 patient-years of follow-up. Similarly, when compared with placebo, aldosterone antagonists resulted in 18 excess cases of hyperkalemia, 5 excess cases of renal failure, and 10 excess cases of hypotension (Table IV) for every 1000 patient-years of follow-up. As expected, for all of these adverse safety outcomes, placebo had a greater probability of being the best treatment (lowest event proportion) compared with aldosterone antagonists, ARBs, and DRIs.
Rate (95% CrI)
|Probability Best, %||Renal Failure |
Rate (95% CrI)
|Probability Best, %||Hypotension |
Rate (95% CrI)
|Probability Best, %|
|Placebo||16.56 (11.30–22.10)||94.7||11.8 (7.30–17.24)||73.9||27.14 (21.12–33.03)||84.3|
|ARBs||39.49 (18.59–82.73)||0.8||27.4 (11.78–69.07)||1.3||44.86 (33.39–66.58)||0.1|
|AA||35.08 (20.64–61.85)||0.3||16.57 (7.68–39.61)||16.8||36.62 (20.70–61.21)||13.5|
|DRIs||41.26 (13.08–117.10)||4.2||29.9 (6.83–123.60)||8.0||53.73 (26.59–103.00)||2.1|
In a sensitivity analysis, even after excluding trials of post–myocardial infarction left ventricular systolic dysfunction, the above results were largely unchanged (Table V).
|Treatment||Death||CV Death||HF Hospitalization||CV Death or HF Hospitalization||Hyperkalemia||Renal Failure||Hypotension|
|ARBs||0.94 (0.69–1.42)||0.90 (0.66–1.14)||0.78 (0.63–1.01)||0.89 (0.73–1.25)||3.87 (1.60–8.50)||3.07 (1.00–15.29)||1.89 (1.12–4.36)|
|AA||0.74 (0.46–1.08)||0.73 (0.53–0.97)||0.67 (0.49–0.88)||0.67 (0.46–0.87)||2.27 (1.40–4.36)||1.48 (0.62–5.31)||1.42 (0.59–3.85)|
|DRI||0.36 (0.01–3.92)||NR||1.62 (0.21–10.39)||0.93 (0.21–4.54)||1.40 (0.35–5.16)||1.54 (0.12–25.11)||2.91 (0.41–32.58)|
|AA||0.78 (0.41–1.27)||0.82 (0.55–1.23)||0.86 (0.57–1.20)||0.76 (0.43–1.03)||0.59 (0.24–1.73)||0.47 (0.09–2.62)||0.75 (0.21–2.20)|
|DRI||0.37 (0.01–4.14)||NR||2.06 (0.26–13.47)||1.02 (0.22–4.95)||0.37 (0.07–1.82)||0.48 (0.02–9.14)||1.53 (0.18–17.37)|
|DRI||0.49 (0.01–5.43)||NR||2.42 (0.30–16.25)||1.39 (0.32–7.25)||0.62 (0.13–2.47)||1.02 (0.05–17.15)||2.05 (0.23–26.81)|
We assessed the risk/benefit ratio of addition of an ARB, aldosterone antagonist, or DRI in patients with HF on conventional HF therapy. When compared with placebo, we found a significant benefit of addition of an aldosterone antagonist rather than ARBs or DRIs for the reduction of death, cardiovascular death, HF hospitalization, or the composite of cardiovascular death or HF hospitalization. When compared with placebo, ARBs were associated with a significant increase in the rate of hyperkalemia, renal failure, and hypotension. Similarly, when compared with placebo, aldosterone antagonists were associated with an increase in hyperkalemia and DRI with an increase in hypotension, with the point estimate suggesting a similar increase in other safety outcomes (although statistically nonsignificant). The risk-benefit profile therefore seems to favor the addition of an aldosterone antagonist rather than ARB or DRI in patients with systolic HF who continue to be symptomatic on conventional HF therapy.
Previous randomized trials have shown the beneficial effect of β-blockers and ACE inhibitors in patients with systolic HF.9,10 However, it has also been shown that chronic treatment with ACE inhibitors leads to continued production of angiotensin II (ACE escape) through non-ACE pathways, thus mitigating the beneficial effects of ACE inhibitors to a certain extent. By blocking the AT1 receptors (downstream pathway), ARBs are not affected by ACE escape. Similarly, by blocking the upstream pathway of renin, DRIs are also not influenced by ACE escape. A combination of either ARBs or DRIs with an ACE inhibitor has thus been proposed to provide a more complete RAS blockade. Indeed, some trials have shown the beneficial effects of both an ARB11 and an aldosterone antagonist12 in symptomatic patients taking conventional HF medications including an ACE inhibitor.
A major concern with dual RAAS blockade (either aldosterone antagonists, ARBs, or DRIs) in HF is the safety issue. In the Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity (CHARM)-Added, hyperkalemia was almost 5 times more common, and elevated creatinine occurred twice as much with the addition of candesartan than with placebo in patients taking ACE inhibitors.11 A recent meta-analysis in more than 18,000 patients with left ventricular dysfunction showed a significantly increased risk of adverse events leading to the discontinuation of dual RAAS blockade (ARBs+ACE inhibitor) compared with monotherapy (ACE inhibitor).13 Thus, hypotension, worsening of renal function, and hyperkalemia were more common with combination therapy than with ACE inhibitor therapy alone. Similarly, Kuenzli and colleagues13 found no benefit of dual RAAS blockade (ARBs+ACE inhibitor) compared with ACE inhibitor but more hyperkalemia, renal dysfunction, and hypotension. The analysis found benefit for the outcome of hospitalization due to HF; however, there was significant heterogeneity in this analysis. Our analysis is concordant with this study in that there was no benefit of ARBs when compared with placebo. However, for certain outcomes, such as hospitalization for HF, the point estimate favored ARBs over placebo (although not statistically significant). In addition, the adverse profile of this dual RAAS blockade (ARBs+ACE inhibitor) was not favorable in that it increased the risk of hyperkalemia, renal failure, and hypotension.
In contrast, addition of aldosterone antagonists to patients already taking an ACE inhibitor not only reduced hospitalization for HF but also all-cause mortality and cardiovascular mortality. However, aldosterone antagonists were also associated with hyperkalemia. The pathogenesis of hyperkalemia with aldosterone inhibition is different from the one with the ACE inhibitor/ARB combination. Aldosterone inhibition slows down the Na/K exchange at the distal tubule and has little if any glomerular effects unless there is excessive diuresis. In contrast, the ACE inhibitor/ARB combination only inconsistently affects aldosterone secretion but causes hyperkalemia predominantly by decreasing glomerular filtration rate secondary to dilating the efferent arteriole. Not surprisingly, renal failure was observed significantly more often with the ACE inhibitor/ARB combination than with aldosterone antagonists. Similarly, the risk profile of the DRIs was also not favorable.
Of note, in our efficacy analyses in the indirect head-to-head comparisons of active comparators, there was no difference for any of the outcomes for any combination of comparators, although the point estimate favored aldosterone antagonists compared with either ARBs or DRIs. Not only were aldosterone antagonists beneficial compared with placebo, but the point estimate favored it in comparison with active comparators. This, together with a probability of >70% to 90% to be the best among the four treatment comparators, allows us to reasonably conclude that the aldosterone antagonists are more effective than either ARBs or DRIs. From a safety perspective, however, any combination of dual RAAS blockade is associated with excess adverse events and these patients need to be closely monitored.
As in other meta-analyses, given the lack of data in each trial, we did not adjust our analyses for dosage of medications used or with compliance to assigned treatment. All of the trials did not report each of the outcomes tested. Accordingly, we are not able to exclude outcome measure reporting bias. Of note, the percentage of patients taking β-blockers and those who received ICD/CRT was low in a few trials that reported this. It is unknown whether the mortality benefit seen with aldosterone antagonists would persist in the presence of increased use of therapies known to reduce mortality. In addition, there were only a limited number of trials involving DRIs and it is possible that the results might change when newer trials with larger sample sizes show any beneficial effect of DRIs.
Given the adverse effects and lack of consistent cardiovascular benefits, the routine addition of an ARB or DRI to ACE inhibitor therapy in HF patients should be avoided or used only in select patients who cannot tolerate aldosterone blockade. The data in aggregate seem to favor aldosterone antagonists over ARBs or DRIs as preferred add-on therapy in these patients. Regardless of which drug class is added, dual RAAS blockade will require strict monitoring of potassium and renal function and a careful follow-up for symptoms and signs of hypotension.
Acknowledgments and disclosures: Dr Bangalore had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis and had the final responsibility to submit for publication. Study concept and design: Drs Bangalore and Messerli. Acquisition of data: Drs Bangalore and Kumar. Analysis and interpretation of data: Drs Bangalore and Messerli. Drafting of manuscript: Dr Bangalore and Messerli. Critical revision of the manuscript for important intellectual content: Drs Bangalore, Kumar, and Messerli. Statistical analysis: Dr Bangalore. Study supervision: Drs Bangalore and Messerli. Sripal Bangalore: Advisory boards: Daiichi Sankyo and Boehringer Ingelheim; Sunil Kumar: None; Franz H. Messerli: Ad hoc consultant/speaker for the following organizations: Novartis, Daiichi Sankyo, Abbott. Grant support from Novartis, Forest, and Boehringer Ingelheim.