Is the glass half full or half empty after PARAGON-HF?

Heart failure with preserved ejection fraction (HFpEF) is an increasingly common condition, particularly with the aging of the population, and it remains a public health problem because of its poor prognosis and high rate of rehospitalization.1,2 Various drugs effective in heart failure with reduced ejection fraction (HFrEF) have been evaluated to improve the natural history and prognosis in the disease with disappointing results. Thus, beta-blockers, ACE inhibitors (ACE-I), angiotensin receptors blockers (ARB) and mineralocorticoid receptor antagonists (MRA) have not been shown to be convincingly effective. Sacubitril/valsartan, a leader in a new therapeutic class, angiotensin receptor and neprilysin inhibitors (ARNi), has been shown to be particularly effective in improving the prognosis and quality of life of patients with low LVEF (HFrEF). In the PARADIGM trial, sacubitril/valsartan improved symptoms and reduced total patient mortality by 16% and HF rehospitalization by 21%.3 It was therefore logical to think that this drug could also be effective in patients with LVEF greater than 40% (HFpEF). Especially as a preliminary study, PARAMOUNT,4 conducted in a population comparable to PARADIGM, but with LVEF > 45%, had shown a beneficial effect on plasma natriuretic peptides which decreased compared to the comparator, valsartan. The PARAGON trial5 was thus performed on nearly 5000 patients who had stable chronic heart failure (CHF) (NYHA II–III) with a LVEF greater than 45% (this cut-off was decided prior to the ESC 2016 Guidelines that defined HFpEF when LVEF is >50%). Sacubitril/valsartan (97/103 mg) bid was tested vs. valsartan (160 mg bid). The primary endpoint was the association of cardiovascular mortality and total hospitalizations for heart failure. The mean age of participants in the study was 72 years, about 7 years more than in HFrEF trials, with 51% being women. The result appears mixed and can be seen optimistically or pessimistically, as with the expression of a glass being half empty or half full. Indeed, the study was negative with regards to the main combined primary endpoint, with a P-value of 0.059 (HR 0.87). Logically, our analysis should stop there because this difference did not meet the predetermined level of statistical significance, and subsequent analyses should be considered to be only exploratory. One cannot help however to look at the various components of the primary criterion: cardiovascular mortality and total mortality were not significantly affected by sacubitril/valsartan and there was only a trend towards a reduction in hospitalizations for heart failure (HR 0.85, P = 0.056). Sacubitril/valsartan also improved quality of life but resulted in more hypotensions. Is the molecule ineffective? Has the study been poorly designed or conducted? Was the study population not the right one? It is now known that HFpEF is an extremely heterogeneous condition and that heart failure is less often than in HFrEF, the leading cause of cardiovascular and total mortality in these patients. Even if one found a decrease in HF-related deaths, it may be difficult to show a reduction in all-cause mortality in a generally very old population in whom death may also occur because of cancer, cerebrovascular disease, etc. This may partly explain these disappointing results on mortality. With regards to hospitalizations for heart failure, the result is also somewhat mixed because in this study, all hospitalizations for heart failure were evaluated, and not the time between randomization and the first hospitalization for heart failure as classically, the criterion chosen being a priori more favourable for a drug improving cardiac function. After PARADIGM, it has been stated that sacubitril/valsartan would have been much more effective if compared to a placebo and not to enalapril, given the known efficacy of enalapril in HFrEF.6 However, this argument cannot be used in PARAGON as the comparator, valsartan at recommended target doses, has not been shown to be more effective than a placebo to improve outcome in I-PRESERVE.7 In an exploratory manner, one can also look at the 12 pre-specified subgroups. In a multivariable model that accounted for all potential interactions, there was suggestion of heterogeneity of treatment effect with possible benefit in patients with lower ejection fraction and in women. Indeed, sacubitril/valsartan was more effective regarding the primary endpoint in women, which is original even though women are more represented in this study population; in most studies conducted in CHF, the effects on women—the proportion of which at baseline rarely exceeds 30%—are equal or lesser than in men, but with non-significant interaction between gender and outcome. More interesting is the interaction between outcome and LVEF: in PARAGON, sacubitril/valsartan was more effective than valsartan in patients with LVEF < 57%, the median value in the study. Is this surprising? Probably not as no-one knows what is the correct upper limit of LVEF for HFrEF. Most of the patients with LVEF < 57% probably have mild LV systolic dysfunction as an important cause—not necessarily unique—of their HF and given the results of PARADIGM, one can imagine that sacubitril/valsartan may also be more effective than an ARB in this population. Similar findings have been made in TOPCAT8 and in CHARM.9 Evaluated initially in HFpEF, sacubitril/valsartan may be added to the list of neurohormonal antagonists effective in HFmrEF, although not specifically evaluated in this population!10 Finally, there is also a trend in difference of effect according to whether patients were included in Western Europe or North America or in the rest of the world, possibly because of different epidemiology or ways of care in different countries as found previously11 but the interaction was not significant. In fact, the most probable explanation is related to the fact that the population suffering from HFpEF is much more heterogeneous than in HFrEF. In HFrEF, whatever the mode of entry into, the phenotype is very generally relatively uniform with dilated, hypokinetic hearts where fibrosis gradually replaces dead myocytes; ventricular hyperexcitability; significant activation of neurohormonal systems (sympathetic, renin–angiotensin–aldosterone, and endothelin). And a neurohormonal antagonist (beta-blocker, ACE-I, ARB, and MRA) may logically be effective in many patients. HFpEF is much more heterogeneous and several studies have shown that there are very different phenotypes.12 What is there in common between (i) an obese diabetic patient with nephroangiosclerosis-related renal failure and a LVEF of 50%, (ii) a 90-year-old woman with systolic hypertension related to a rigid arterial system and who decompensates during infections or runs of atrial fibrillation? or (iii) a patient with a slightly dilated ventricle with LVEF 42% who is likely to have minimal systolic dysfunction. We could multiply the examples. It is also possible that there were patients included in all HFpEF studies with cardiac amyloidosis13 and conventional drugs of heart failure and perhaps more so sacubitril/valsartan are known to be ineffective or perhaps deleterious in this disease. In total, sacubitril/valsartan in PARAGON is equal to or slightly better than valsartan. All of this confirms that it is unlikely that an effective drug will ever be found in all forms of HFpEF if one does not change the way of running studies in this population. It would probably be necessary to target patients a little more and administer the treatment according to the phenotype. It can be imagined that RAS and neprilysin inhibitors may be relatively effective in subjects with LVEF below 57%, in diabetic patients or in case of significant cardiac remodelling; or that calcium antagonists or diuretics may be more effective when systolic hypertension is the primary cause of HFpEF. Treating decompensating factors may also be as effective as the treatment regimens evaluated to date: influenza vaccination may be one of the most effective treatment in HFpEF patients.14 Conflict of interest: A.C.S. received fees from Novartis, Astra Zeneca, Servier, Vifor, Abbott, Roche. Biography: Alain Cohen-Solal is Professor of Cardiology at the Faculty of Medicine (Paris University) and Head of the Department of Cardiology and Cardiac Rehabilitation at the Hospital Lariboisière in Paris, France. Professor Cohen-Solal’s research is focused on several aspects of acute and chronic heart failure (HF), mainly: the abnormalities of the cardiac and peripheral responses to exercise in HF, the evaluation of HF by cardiopulmonary exercise testing, the diagnostic and prognostic role of neuro-hormones and biomarkers, the pathophysiology of HFpEF and CardioOncology. He headed (2009–19) a National Institute of Health and Medical Research Unit (U942) ‘Biomarkers in Heart Failure’. Alain Cohen-Solal is a member of the European Society of Cardiology. He has been member of the Working Group on ‘Heart Failure’ (ESC) and of the Board of the European Association of Cardiac Prevention and Rehabilitation. He has also been member of the Heart Failure Guidelines Writing Committee of the ESC in 2008. He currently is also on the board of the GERS-P, the WG on Exercise and Rehabilitation and Prevention of the French Society of Cardiology; member of the CardioOncology WG of the ESC and leads the CardioOncology Council of the French Society of Cardiology. He has been the national coordinator in France for several trials in heart failure, including DIAMOND-HF, VICTORIA, ATOMIC-HF, RELAX-AHF2, TRUE-HF, EFFECT-HF, REACH 1, CHARM, SENIORS, SURVIVE, and RED-HF. Biography: Damien Logeart is Professor of Cardiology at Paris University (Paris Diderot medical school). He is head of the echocardiography lab and works in the cardiology department as a heart failure specialist, as well as an interventional cardiologist. Damien Logeart is a member of French Society of Cardiology, the ESC and the HFA. He was president of the French Working Group on ‘Heart failure and cardiomyopathy' (SFC, 2009–2010). His research is focused mainly on the care of patients with heart failure and on biomarkers. He has published as an author and co-author of over 150 articles in peer-reviewed medical journals. He obtained several grants from the government and is the coordinator of several academic multicenter trials and registries (ECAD-HF, OFICA, FRESH, PREGICA).