Pharmacological Treatment of Heart Failure with Preserved Ejection Fraction: A Glimpse of Light at the End of the Tunnel?

This editorial refers to ‘Effects of carvedilol on heart failure with preserved ejection fraction: the Japanese Diastolic Heart Failure Study (J-DHF)’, by K. Yamamoto et al., published in this issue on pages 110–118. Heart failure with preserved ejection fraction (HFPEF) is an increasingly large medical problem.1 This is due not only to the fact that its incidence and prevalence are increasing, but also because its diagnosis is difficult, and—lastly—because no treatment is currently available. Traditionally, HF was seen as a disorder in which there was a ‘pump problem’ which would lead to signs and symptoms, and the pump problem was generally considered a contractile problem, which could be quantitated as a reduced LVEF. In recent years, it is increasingly being recognized, however, that HF can also occur in the setting of a normal or near-normal (‘preserved’) EF, and this is often associated with an impaired filling, causing impaired pumping capacity of the heart in a different way.2–4 As far back as 2006, Owan and co-workers5 showed that the incidence of HFPEF was increasing, while the incidence of HF with reduced EF (HFREF) was decreasing. The decreased incidence of HFREF appears to be due, at least in part, to improved (invasive) treatment of acute myocardial infarction and the fact that fewer large infarcts (leading to HF) are encountered. The increased incidence of HFPEF may be related to the ageing of the population, particularly in the Western world,6 as stiffness of the heart increases with age. More recent studies have confirmed the findings of Owan et al.,5 including a large study from 275 hospitals in the USA which showed that the number of HF hospitalizations gradually increased for patients with HFPEF, while it decreased in patients with HFREF in the same time period.7 A second problem associated with HFPEF is its definition and diagnosis.4 Older trials in patients with HFPEF, such as CHARM-Preserved,8 basically enrolled patients who had a normal (or near normal) EF (usually with a cut-off of 40%) in combination with signs and symptoms of HF, with or without a previous (recent) HF hospitalization. In the subsequent PEP-CHF9,10 trial, additional echocardiographic criteria were required for enrolment, which included evidence of left atrium enlargement and LV hypertrophy, while in I-PRESERVE they could be used, but were not mandatory.11,12 There is a lot of debate, however, regarding the value and clinical applicability of echocardiographic criteria for the diagnosis of HFPEF, while the cut-off value of the EF (i.e. to determine what is preserved EF) is another issue.13 Recently, it has been suggested that a more prominent place should be given to the use of biomarkers in the diagnosis of HFPEF,14 although most, if not all, physicians who treat patients with HFPEF would agree that some evidence of structural (or functional) heart disease must be documented. The third problem in the management of patients with HFPEF is that so far no single drug has been shown to improve outcome significantly in this population.12 Potentially interesting findings have been observed with the ACE inhibitor perindopril in the PEP-CHF study,10 with the ARB candesartan in CHARM-Preserved,8 and with the beta-blocker nebivolol in the SENIORS Study,15 but none of these findings was robust, and, as a consequence, no drug has received a recommendation in patients with HFPEF in the recently published European Society of Cardiology (ESC) guidelines.13 In the present issue of the journal, the findings of the Japanese Diastolic Heart Failure (J-DHF) study are reported.16 In this study, Yamamoto et al. studied 245 patients with HF and LVEF > 40% who had a clinical diagnosis of HF based on clinical criteria,17 and no additional echo criteria were required. After enrolment, patients were randomized to carvedilol at a target dose of 10 mg twice daily, the primary endpoint was a composite of cardiovascular (CV) death and (unplanned) hospitalization for HF, and—assuming a risk reduction of 20% on beta-blockade after 2 years—it was calculated that a total of 800 patients (2 × 400) would be necessary to achieve a power of 80%. Because of slow enrolment, however, only 245 patients were randomized between 2004 and 2009.16 Patients were not required to have sinus rhythm at baseline, and 34% of all patients had AF. During a mean follow-up of 3.2 years, the primary endpoint occurred in 29 patients in the carvedilol group, compared with 34 patients in the placebo group [hazard ratio (HR) 0.90, P = 0.69]. All-cause mortality alone was also not different in the two groups (18 vs. 21 deaths, respectively). The effect of carvedilol on all-cause mortality or CV hospitalizations (the primary endpoint in the SENIORS Study15) was also reported in the J-DHF study, and this showed that there were 45 patients on carvedilol who experienced this compared with 56 patients on placebo [HR 0.85, 95% confidence interval (CI) 0.57–1.26, P = 0.41). Although the target dose was 20 mg of carvedilol, the median prescribed dose was only 7.5 mg, and, in the patients who tolerated higher doses (>7.5 g/day, n = 58), the effect of carvedilol appeared to be better overall. However, the study was probably too small to show significant differences between groups, and P-values for interaction were not given. Although the study of Yamamoto et al. was greatly underpowered,16 the results are nevertheless interesting, particularly as they provide new data on beta-blockers in patients with HFPEF. There are two other randomized studies available which provide data in this patient population.15,18 The first set of data comes from a substudy from the SENIORS trial, in which the primary endpoint was a composite of all-cause mortality or CV hospitalizations.15 In that substudy, the 643 patients with an EF ≥40% included had been randomized to nebivolol (n = 320, target dose 10 mg, achieved dose 7.6 mg) or placebo (n = 323). The effect of nebivolol on the primary outcome was non-significant; the nebivolol:placebo HR was 0.82 (95% CI 0.62–1.08). In the third study by Aronow et al.,18 158 patients with mild to moderate HF and an LVEF ≥40%, who were treated with an ACE inhibitor and diuretics, were randomized to a target dose of propranolol 30 mg, three times daily, or no propranolol (control group). During a follow-up of 32 months, 44 of 79 patients (56%) randomized to propranolol, and 60 of 79 patients (76%) randomized to no propranolol died (P = 0.007 between groups), but, regarding other endpoints, only non-fatal myocardial infarction was reported. What do these findings mean? If the present data from the J-DHF and the data from the SENIORS substudy are pooled19 for the composite endpoint all-cause mortality or CV hospitalizations, the observed odds ratio (OR) is 0.78 (95% CI 0.59–1.04, P = 0.09). If we add the Aronow data18 for all-cause mortality alone, and pool the three studies described above,19 the observed OR is 0.75 (95% CI 0.54–1.03, P = 0.08) (Figure 1). Still, the numbers of patients remain small and no firm conclusions can be drawn from these findings. Nevertheless, some remarks can be made. Beta-blockers primarily act by decreasing oxygen consumption and by reducing myocardial ischaemia. In patients with HF, there is a mismatch between oxygen demand (increased in HF) and supply (decreased in HF). This mismatch increases during exercise, leading to myocardial ischaemia.20,21 Animal experiments have shown that myocardial blood flow reserve (i.e. the response to coronary vasodilation, such as that caused by adenosine) is markedly attenuated in LV hypertrophy, but even more so when LV hypertrophy is associated with HF.22 Because of their antiadrenergic and anti-ischaemic effects, beta-blockers reduce atrial and ventricular arrhythmias (and sudden death) in HFREF,13,23 and it can be speculated that this effect could be similar in patients with HFPEF. Another issue is the presence of AF in HFPEF. In J-DHF, 34% of patients had AF, and in SENIORS AF was present in 36% of those with HFPEF.15 There are data that show that beta-blockers may be less effective in HFREF patients with AF than in those with sinus rhythm,23,24 and it can be speculated that if beta-blockers would be studied only in (HFPEF) patients with sinus rhythm, the treatment effect would be more favourable. Indeed, in J-DHF, the effect of carvedilol in patients with sinus rhythm (HR 0.68) also appeared different from that in patients with AF (HR 1.12; P-value for interaction 0.33), but of course the numbers here were small. Hopefully, in the future, trials with beta-blockers in patients with HFPEF (and sinus rhythm), such as the β-PRESERVE Trial, that aims to enrol 1200 patients who will be randomized to metoprolol or placebo, and will be conducted in China,25 will provide more data in this population. Neurohormonal activation is one of the hallmarks of HF, and neurohormonal inhibitors have been shown to be the most effective drugs in patients with HF with systolic dysfunction.26 However, as described above, so far, none of the studies in HFPEF investigating the effect of increased neurohormonal blockade showed a significantly improved outcome.4,13 In recent years, the new first in class angiotensin receptor neprilysin inhibitor (ARNI) compound LCZ696 was developed. This drug combines the effect of the ARB valsartan and a neprilysin (neutral endopeptidase) inhibitor prodrug AHU377 in one compound.27 Neprilysin degrades biologically active natriuretic peptides such as BNP (and ANP), and, by inhibiting this enzyme, LCZ696 increases natriuretic peptides, which may have favourable cardiovascular effects, such as vasodilation and natriuresis. A recent study showed that LCZ causes a greater blood pressure reduction in patients with hypertension than valsartan, with similar tolerability.28 At the 2012 ESC Congress in Munich, the results of the PARAMOUNT Trial were presented, and they have now also appeared in the Lancet.29 In that study, Solomon et al. studied 301 patients with HF and an LVEF ≥45%, who had signs and symptoms of HF, and they were required to have an NT-proBNP >400 pg/mL. Patients were randomized to a target dose of 200 mg LCZ696 twice daily or valsartan 160 mg twice daily. The primary endpoint of the study was the change from baseline in NT-proBNP after 12 weeks, and total treatment duration was 36 weeks. (Note: NT-proBNP is not a substrate for neprilysin, so NT-proBNP levels in patients taking LCZ696 reflect LV wall stress, or the patient's haemodynamic condition more broadly.) After 12 weeks, LCZ696 reduced the mean NT-proBNP from 783 to 605 pg/mL, while no significant effect was observed with valsartan (NT-proBNP from 862 pg/ml to 835 pg/mL; P = 0.005 between groups). The difference between the two groups was no longer statistically significant after 36 weeks as NT-proBNP levels in the valsartan group decreased somewhat between 12 and 36 weeks. The drug was generally well tolerated, and left atrial dimensions decreased more on LCZ696 than on valsartan. The number of clinical events was unaffected, and too small to allow any meaningful comparison. What do the data of the PARAMOUNT study mean? NT-proBNP has become one of the most widely used (surrogate) markers of efficacy in HF studies,14 and the data from the PARAMOUNT study are intriguing. By increasing natriuretic peptides, LCZ696 may lead to vasodilation and natriuresis, and possibly, further suppression of the renin-angiotensin system, which could contribute to the potentially beneficial morphological changes, as observed with echocardiography in PARAMOUNT. This may well lead to a lower incidence of AF, which is beneficial in these patients, since AF is more troublesome in HFPEF patients than in patients with HFREF.30,31 The incidence of AF in PARAMOUNT was low, and the follow-up was too short to provide useful data on this. The comparator to LCZ696 used in PARAMOUNT was valsartan given at a dose equivalent to this antagonist provided by LCZ696. It should be noted that valsartan was given not because ARBs are of benefit in HFPEF, but because most patients with HFPEF have an indication for a renin–angiotensin–aldosterone system (RAAS) blocker related to co-morbidities such as hypertension and diabetes mellitus, and in PARAMOUNT 80% of patients were treated with a beta-blocker, and 93% with an ACE inhibitor and/or an ARB (before randomization). An alternative comparator could have been an ACE inhibitor, but then the study design would not have been as clear a test of the effect of neprilysin inhibition as the equivalence of the level of concomitant RAAS blockade would have been less certain.32 Lastly, the total number of deaths in PARAMOUNT was three, and therefore, although the findings of this study are promising, an outcome trial will be necessary before any firm statements regarding the potential place of this drug in the treatment of HFPEF can be made. HFPEF has for a long time been a very difficult field for many reasons, as discussed above, but recent data from different studies suggest that we may see a first glimpse of light at the end of the tunnel. In addition to the studies discussed above, aldosterone blockade has been suggested to be potentially beneficial in the treatment of patients with HFPEF.33 In the recently presented ALDO-DHF study, 422 patients with HFPEF (EF ≥50%) and mild symptoms who received 25 mg spironolactone or placebo were studied. Although the results have not been published, spironolactone was found not to increase exercise capacity after 12 weeks, but it did improve echocardiographic parameters of diastolic function.34 In the much larger TOPCAT trial, which investigates the effect of spironolactone on outcome, 3500 patients with HFPEF and EF ≥45% are being investigated,35 and the results of this trial will be available in the next 12 months. Hopefully, all these new studies will lead to a new perspective for patients with HFPEF. We thank Kevin Damman for carrying out the statistical analysis of beta-blockers in HF-PEF. Conflict of interest: D.J.v.V. has received Board Membership Fees from Novartis. J.J.V.M's employer, Glasgow University, has been paid by sponsors for his participation in a number of trials in HF-PEF including CHARM-Preserved (Astra Zeneca), I-Preserve (BMS-Sanofi) and PARAMOUNT (Novartis).