Angiotensin receptor–neprilysin inhibition in heart failure with preserved ejection fraction: lessons from PARAGON‐HF

Identifying effective therapies for patients living with heart failure (HF) with preserved ejection fraction (HFpEF) has remained challenging. The recently completed PARAGON-HF (Prospective Comparison of ARNI with ARB Global Outcomes in HF With Preserved Ejection Fraction),1 the largest HFpEF trial conducted to date, examined a first-in-class angiotensin receptor–neprilysin inhibitor (ARNI), sacubitril/valsartan, compared with valsartan among 4822 patients with signs and symptoms of HF, left ventricular ejection fraction (LVEF) ≥45%, elevated natriuretic peptides, and evidence of structural heart disease. Patients allocated to sacubitril/valsartan experienced modestly lower rates of composite cardiovascular death and total HF hospitalizations compared with patients randomized to valsartan after a median follow-up of 35 months, but this treatment effect did not meet pre-specified statistical significance thresholds (Figure 1). We contextualize these trial findings, including learnings from subsequent secondary analyses, and reflect on key trial elements that may inform the next phase of therapeutic development in HFpEF. Neprilysin, the primary target of sacubitril, is a membrane-bound and circulating enzyme that is broadly expressed across tissue beds. It is responsible for cleaving several potentially biologically important molecules, many of which, including certain natriuretic peptides, are thought to mediate its therapeutic benefits. While a diversity of biological processes may contribute to the development and progression of HFpEF, perturbed natriuretic peptide signalling appears to be important in its pathogenesis. There is down-regulation of natriuretic peptide–cyclic guanosine monophosphate (cGMP)–protein kinase G (PKG) signalling in the cardiomyocyte in HFpEF2 (Figure 2). This is partially mediated by up-regulated phosphodiesterase type 9 activity, which breaks down cGMP stimulated by non-nitric oxide mechanisms (such as via natriuretic peptide signalling). The cGMP–PKG pathway is key in modulating myocardial and vascular stiffness.2 Indeed, patients with early HFpEF have an impaired atrial natriuretic peptide, renal cGMP, and natriuretic responses to acute volume expansion, which is corrected by subcutaneous B-type natriuretic peptide pre-treatment.3 Natriuretic peptides are also anti-proliferative (i.e. inhibit hypertrophy and fibrosis), and neprilysin inhibition has been shown to lower markers of profibrotic signalling in HF.4 Moreover, neprilysin inhibition may possess favourable properties in slowing kidney progression in HF, especially among patients with diabetes mellitus, a common comorbidity in HFpEF.5 PARAGON-HF was preceded by a phase 2 study of ∼300 patients with HFpEF, which demonstrated a 23% reduction in 12-week changes in N-terminal prohormonal of B-type natriuretic peptide (NT-proBNP).6 As such, there was strong rationale that the beneficial effects of neprilysin inhibition may extend to HFpEF. On the other hand, some reports have suggested that circulating neprilysin levels may be lower in HFpEF,7, 8 and thus the biological effects of neprilysin inhibition may be more complex in this population. Were the intended biological pathways affected and anticipated haemodynamic responses achieved in PARAGON-HF? Irrespective of starting levels, average systolic blood pressure was reduced by ∼5 mmHg compared with valsartan at 4 weeks in PARAGON-HF9; these haemodynamic responses are even greater than the observed blood pressure lowering in HF with reduced ejection fraction (HFrEF).10 In PARAGON-HF, NT-proBNP was rapidly reduced by 19% with sacubitril/valsartan compared with valsartan by 16 weeks.11 Taken together, blood pressure and natriuretic peptides were indeed successfully reduced with sacubitril/valsartan compared with valsartan, suggesting that drug dosing and adherence to the study drug were reasonable in the trial. Yet these changes in natriuretic peptide levels were lower in magnitude compared with changes observed in controlled trials of HFrEF (28% at 8–10 weeks12 and 33% at 12 weeks13). While natriuretic peptides are lower on average in HFpEF compared with HFrEF and subject to variability across key subgroups, short-term changes in natriuretic peptides in response to investigational therapies appear to still be interpretable and are least directionally linked with treatment effects on HF hospitalization when examining a range of HFpEF programmes.14 Based on the NT-proBNP response observed in PARAGON-HF, a ∼10–20% relative risk reduction in composite cardiovascular death or HF hospitalization would have been anticipated based on data from prior HFpEF trial programmes.15 Over the last three decades of HF clinical trials, the choice of the primary endpoint has evolved from all-cause mortality to more disease-specific composites including non-fatal as well as fatal events, such as time to first occurrence of hospitalization for HF or cardiovascular death. However, as overall mortality and the relative proportion of cardiovascular deaths in HFpEF are lower compared with HFrEF,16 new endpoints and analytic approaches have been considered. As a result, the primary efficacy endpoint chosen PARAGON-HF reflected a further evolution of thinking about the impact of HFpEF and what the key goal of treatment is in this condition. Recognizing the chronicity of HFpEF, and the potential occurrence of multiple non-fatal events before death, the primary efficacy endpoint chosen PARAGON-HF was a composite of total (first and recurrent) HF hospitalizations and cardiovascular death. Quantifying total HF hospitalizations may more completely capture patient experience, is linked closely with disease progression, and is meaningful from medical, economic, and societal perspectives. While several analytic approaches have been proposed in recurrent event analyses, each, including the semi-parametric proportional rates method employed in PARAGON-HF, appears to demonstrate stronger treatment effects and generally augments power (and reduces necessary sample size).17 Indeed, in PARAGON-HF, there were 1083 patients experiencing at least one primary event (inclusive of only first HF hospitalizations) and 1903 total events (inclusive of first and recurrent HF hospitalizations). Sacubitril/valsartan showed more robust effects on recurrent HF hospitalizations compared with first HF events in PARAGON-HF.1 Beneficial effects of sacubitril/valsartan were also observed with respect to metrics of patient-reported quality of life and physician-assessed functional status, which were key secondary endpoints. While these results must be considered exploratory given hierarchical testing structures, consideration of patient-centred endpoints are in line with recent regulatory guidance put forth by the US Food and Drug Administration.18 As adverse quality of life is similar or worse among HFpEF compared with HFrEF (especially in women19, 20), ongoing HFpEF trial programmes should continue to target measures of health status. Modest average differences in the Kansas City Cardiomyopathy Questionnaire clinical summary score (1.0, 95% confidence interval 0–2.1) may be reflective of high individual patient-level responses. In a pre-specified responder analysis, a greater proportion of patients randomized to sacubitril/valsartan garnered clinically meaningful improvements in health-related quality of life measures.1 Kidney dysfunction is an integral part of the pathogenesis of HFpEF and slowing or even reversing progressive worsening of kidney function over time is another important treatment goal. A composite kidney outcome, a secondary endpoint in PARAGON-HF, was halved with sacubitril/valsartan compared with valsartan,1 suggesting incremental kidney benefits of neprilysin inhibition to renin–angiotensin blockade in this high-risk population. Given the large number of patients with HFpEF who are already on renin–angiotensin system blockers, this incremental renal benefit due to neprilysin may be meaningful. While there is no clearly demonstrated standard of care for patients with HFpEF, valsartan, an angiotensin receptor blocker (ARB), was selected as a comparator instead of placebo in PARAGON-HF for several reasons.21 First, nearly 90% of patients in PARAGON-HF were treated with an angiotensin-converting enzyme inhibitor (ACEi) or ARB before they were screened for inclusion in the trial. Unlike typical therapeutic scenarios when investigational therapies can be added on, the risk of angioedema would be excessive with concurrent use of a neprilysin inhibitor and ACEi. Second, discontinuation of ACEi/ARB prior to randomization to ARNI or placebo would risk renin–angiotensin system withdrawal and resultant inadequate blood pressure control (a key guideline-recommended component of HFpEF management) in the placebo arm, and would likely have raised ethical concerns. Nevertheless, while the selected design isolated the study of the incremental effects of the novel compound, sacubitril, the use of an active comparator may have diminished the observed treatment effect since ARBs may themselves be beneficial with respect to HF hospitalization in HFpEF.22 To investigate this further, we used data from PARADIGM-HF (Prospective comparison of ARNI with ACEI to Determine Impact on Global Mortality and morbidity in Heart Failure) and PARAGON-HF (n = 13 194) and the CHARM programme (Candesartan in Heart failure–Assessment of Reduction in Mortality and morbidity) (n = 5050) to make an indirect comparison between sacubitril/valsartan and a putative placebo. In this analysis, there was a substantial and nominally significant benefit of sacubitril/valsartan on cardiovascular death and total HF hospitalizations up to a LVEF of approximately 60%.21 Researchers and clinician experts have made a call to move beyond LVEF in defining HF groups.23 Indeed there is a growing perspective that HFpEF is both aetiologically and phenotypically heterogeneous.24, 25 While more rigorous phenotyping with imaging and biomarkers has been suggested, this would add cost and complexity to clinical care and trial designs. In the near-term, LVEF will remain a cheap, prognostically meaningful, and widely available tool in characterizing HF. In PARAGON-HF, the benefits of sacubitril/valsartan appeared to be more apparent among patients at the lower range of ejection fraction-based eligibility,26 one of 12 pre-specified subgroups in the trial, that held up to multivariable adjustment. This finding based on the pre-specified subgroup of LVEF at or below the median (≤57%) was consistent with similar findings in the CHARM-Preserved and TOPCAT (Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist) trials27-29 in which renin–angiotensin–aldosterone system inhibitors had shown directionally favourable benefits in reducing HF hospitalization among patients at the lower end of the ejection fraction spectrum in HFpEF trials.27 In a pre-specified pooled patient-level analysis of data from both PARADIGM-HF and PARAGON-HF encompassing over 13 000 patients with HF, sacubitril/valsartan was superior overall to renin–angiotensin system inhibitors across a broad range of endpoints, including all-cause mortality. Nevertheless, treatment effects were modified by LVEF such that more prominent effects were observed among participants with ejection fraction below the normal range.26 These observations lend external validity to these findings and are supported by two mechanistic trials demonstrating that the beneficial effects of ARNIs in HFrEF may be mediated through reverse remodelling.13, 30 Taken together, these data suggest that patients with HF and LVEF below normal may indeed benefit from therapies known to benefit patients with frankly reduced LVEF. While cautious interpretation of subgroup analysis is always appropriate, ARNI effects in patients with LVEF below the normal range are substantiated by pre-specified analyses (with significant interaction testing), biological plausibility, and external consistency. The exact upper and lower bounds of this LVEF range of benefit remain poorly defined. Like other continuous markers of risk (low-density lipoprotein, blood pressure, fasting plasma glucose) with attendant measurement error and inter-test variability, there appears to be a gradient of risk and ARNI benefit at lower LVEF ranges.26 Other factors may also explain the attenuated benefits of ARNI at higher LVEF ranges. Higher competing risks of non-cardiovascular death among patients with HFpEF may be less modifiable with ARNI. Amyloid deposition and sarcomeric myocardial abnormalities (including hypertrophic cardiomyopathy) may be important to disease progression in some patients with HFpEF (especially with normal or supra-normal LVEF),31 which would not be expected to be modified with ARNI. More women than men are affected by HFpEF. PARAGON-HF enrolled 52% women, one of the largest cohorts of women studied in a HF trial. Ensuring adequate representation and rigorous study of women in HF trial programmes has been an important goal of regulatory agencies and federal funding bodies. Few HF therapies are known to differ significantly in efficacy between men and women.32 In a pre-specific subgroup analysis of PARAGON-HF, women were observed to garner significantly more benefit from sacubitril/valsartan compared with men, an effect driven by reductions in HF hospitalizations.33 While both men and women had declining benefit with increasing LVEF, women derived benefit to a substantially higher LVEF than men. Similar observations with more favourable treatment responses among women compared with men have been seen with trials of other renin–angiotensin–aldosterone system inhibitors in HFpEF.27 This gender-based treatment differential requires further study and may elucidate new biological understanding of this disease entity. In PARAGON-HF, the overall treatment effect, together with observed effects on secondary endpoints and in large subgroups, suggest potentially clinically relevant benefits of sacubitril/valsartan that extend to patients with LVEF higher than the range that is currently considered frankly ‘reduced,’ findings that may be common to therapies that benefit patients with systolic dysfunction. Indeed, growing evidence supports the concept that evidence-based therapies that are disease-modifying in HFrEF may also have beneficial properties among patients with HF with LVEF below normal, or ‘mildly reduced’ ejection fraction.26, 34 Recognizing the degree of heterogeneity in this population, future trials may consider employing various approaches to improve precision of estimates of treatment effects including adaptive designs (allowing for continual enrolment beyond initial targets if point estimates are in a promising zone). As clinicians, guideline writers, and global regulatory bodies interpret the totality of evidence from PARAGON-HF to determine its potential use in clinical care of HFpEF patients, these data advance current understanding of HF across the spectrum of LVEF and inform next directions in therapeutic development for this important syndrome. Conflict of interest: M.V. is supported by the KL2/Catalyst Medical Research Investigator Training award from Harvard Catalyst (NIH/NCATS Award UL 1TR002541), serves on advisory boards for Amgen, AstraZeneca, Baxter Healthcare, Bayer AG, Boehringer Ingelheim, Cytokinetics, and Relypsa, and served on the clinical endpoints committee of PARAGON-HF. J.J.V.McM.'s employer, Glasgow University, has been paid by Novartis (who manufacture sacubitril/valsartan) for his time spent as committee member for the trials listed (using sacubitril/valsartan), meetings related to these trials and other activities related to sacubitril/valsartan, e.g. lectures, advisory boards and other meetings. Novartis has also paid his travel and accommodation for some of these meetings. These payments were made through a Consultancy with Glasgow University and he has not received personal payments in relation to these trials/this drug. The trials include – PARADIGM-HF: co-PI; PARAGON-HF: co-PI; PERSPECTIVE, PARADISE-MI and UK HARP III trial: executive/steering committees. S.D.S. has received research grants from Alnylam, Amgen, AstraZeneca, Bellerophon, Celladon, Gilead, GlaxoSmithKline, Ionis Pharmaceutics, Lone Star Heart, Mesoblast, MyoKardia, NIH/NHLBI, Novartis, Sanofi Pasteur, Theracos, and has consulted for Alnylam, Amgen, AstraZeneca, Bayer, Bristol-Myers Squibb, Corvia, Gilead, GlaxoSmithKline, Ironwood, Merck, Novartis, Pfizer, Takeda, and Theracos.