Ejection Fraction in Heart Failure Revisited-Where Does the Evidence Start?

Michael Böhm, Yvonne Bewarder, Ingrid Kindermann


Eur Heart J. 2020;41(25):2363-2365. 

Randomized placebo controlled trials (RCTs) are the mainstay of evidence-based medicine and particularly important in chronic diseases such as heart failure with a high prevalence, incidence, morbidity, and mortality. Guideline-directed therapies consist of renin–angiotensin–aldosterone system inhibitors, beta-blockers, and mineralocorticoid receptor antagonists (MRAs),[1] the evidence for which was generated in large landmark trials.[2–6] All these studies recruited early (1991–2011) and have selected appropriately matched placebo-treated control groups. As the majority of trials usually are designed to test new treatments on top of guideline-recommended therapies, the problem of finding appropriate control groups arose with the testing of angiotensin receptor blockers (ARBs). In CHARM-Alternative, valsartan was tested against a placebo group, however in patients who could not tolerate an angiotensin-converting enzyme (ACE) inhibitor.[7] This control group might not be applicable to the overall population of heart failure patients as coughing is associated with different genetic backgrounds such as polymorphisms of the bradykinin receptor gene[8] and is differently prevalent in certain populations.[9] Therefore, a large body of data on ARBs in heart failure are generated as add-on therapy to ACE inhibitors (CHARM-Added trial).[10] In particular, novel studies such as PARADIGM[11] tested sacubitril/valsartan against guideline-recommended therapy, which was enalapril according to the SOLVD trial.[2] It appears difficult to find appropriate placebo controls for novel treatments, because for ethical reasons evidence-proven therapies cannot be withheld and long-term pre-treatment with related drugs (the ARB component in sacubitril/valsartan) might produce carry-over effects, limiting scientific reliability.[1]

In heart failure with preserved ejection fraction (HFpEF), similar problems arose. While there is no direct placebo-controlled evidence for any treatment, these patients are usually intensively treated with drugs which are recommended for heart failure with reduced ejection fraction (HFrEF) such as beta-blockers and ACE inhibitors/ARBs (40–75%) for an ejection fraction of >50% and close to 90% treatment rates in patients with an EF of 40–50%, which is similar to that in patients with HFrEF (EF <40%).[12] This might be due to the perception of physicians that any patient with heart failure needs neuroendocrine antagonist treatment or to the fact that these agents are applied for the treatment of hypertension, which is highly prevalent in this population.[12] This was one of the problems with the PARAGON study,[13] which compared sacubitril/valsartan with valsartan. In a previous study, the ARB drug class member candesartan provided a nominal albeit non-significant reduction in cardiovascular death and heart failure hospitalization in the CHARM-Preserved trial of 11%.[14] Therefore, the question remains of whether PARAGON was neutral due to a small but difficult to detect effect of valsartan. Notably, valsartan improved diastolic function in patients with hypertension and HFpEF in the VALIDD trial, in patients in whom blood pressure was reduced by >10 mmHg.15 In PARAGON, 95.5% of patients had hypertension and the median systolic blood pressure (SBP) of 131 ± 16 mmHg indicates that many patients with uncontrolled hypertension and significant decreases of on-treatment SBP might have been enrolled.

In their study reported in this issue of the European Heart Journal, Vaduganathan et al.[16] took a reasonable approach by applying statistical methods to impute a historical placebo group[14] as comparator for the pooled treatment arms of PARADIGM[11] and PARAGON.[13] This has been done with the PARADIGM study before[17] using data from SOLVD.[2] The authors observed a similar effect on treatment with sacubitril/valsartan as in placebo-controlled trials in the low EF group. Interestingly, at an EF of from 40% to <55%, a clear sign for risk reduction was also observed. Therefore, these data are reassuring despite a lack of a placebo group but extending clinical information to a less well studied but large[12] group of heart failure patients with intermediate EF.[16,18] These data are in line with a secondary analysis from the TOPCAT study19 and individualized patient data from beta-blocker trials,[20] also providing evidence that the cut-off of EF for beneficial treatment effects is ~55%.

These analyses using imputed control groups might have some limitations. As the authors correctly pointed out, these studies were generated at different time points with different concomitant treatments. Treatment intensity with MRAs was lower and ivabradine for patients with higher heart rates was not available.[21] Furthermore, the mortality rates for cardiovascular causes as one crucial endpoint in heart failure trials declined over time, while the prevalence and mortality of non-cardiovascular comorbidities are steadily increasing.[22] Therefore, the generally used primary endpoint of cardiovascular death and heart failure hospitalization could have been more prevalent in the older studies from where placebo groups were imputed than they would have been in contemporary control groups.

Taken together, the authors have to be congratulated for providing important data and reassuring physicians to use novel heart failure drugs such as sacubitril/valsartan by imputing a placebo control group as comparator for PARADIGM and PARAGON treatments. Importantly, this analysis, in addition to previous studies,[18–20] shows that in the sparsely studied population of patients with an EF of 40–55%, several heart failure treatments including sacubitril/valsartan might provide benefit (see Take home figure). These and previous data should be the subject for intensive discussions among scientists, physicians, and healthcare providers. Future guidelines need to find a way to implement these data into recommendations for daily clinical practice.


Treatment effects over a broad range of ejection fractions from PARAGON/PARADIGM, TOPCAT, Beta Blocker Trialists investigation, and CHARM trials. Please note that the cut-off of ejection fractions appears to be around 55% from where neuroendocrine targets provide risk reduction (green), while non-modifiable targets and non-cardiovascular comorbidities are associated with outcomes not modifiable by currently available treatments at an ejection fraction >55% (red area, lower panel). Dark red areas depict ejection fractions which are not covered by dedicated placebo-controlled trials, but where evidence exists from exploratory analyses.