Understanding heart failure with preserved ejection fraction: where are we today?

Under physiological conditions, LV pressure rapidly decreases after systole, allowing fast diastolic LV filling at maintained low filling pressures. Diastolic LV dysfunction in HFpEF is evident from slow LV relaxation and elevated diastolic LV stiffness, which increase diastolic filling pressures and limit cardiac performance at rest, during atrial pacing and exercise [9, 10]. Insight into the pathophysiology of diastolic LV dysfunction in HFpEF has long been missing because of a lack of myocardial tissue obtained from patients with HFpEF. Over the past decade, several groups of investigators were able to obtain myocardial tissue from HFpEF patients revealing specific alterations in myocardial structure, function and intramyocardial signalling, which were relevant to the concentric LV remodelling and diastolic LV dysfunction characteristically observed in patients with HFpEF (Table 1; [11‐18]). Structural alterations consisted of cardiomyocyte hypertrophy [12, 13] and varying degrees of myocardial interstitial fibrosis [11‐13, 17, 18] and capillary rarefaction [18], whereas functional alterations included increased cardiomycyte stiffness [11‐15]. The same studies also demonstrated abnormal intramyocardial signalling, which was evident from endothelial cells expressing adhesion molecules [13, 16], inflammatory cells secreting profibrotic transforming growth factor β (TGF-β) [16] oxidative stress increasing nitrotyrosine content [15, 16] and downregulation of myocardial cyclic guanosine monophosphate (cGMP)-protein kinase G (PKG) signalling [15]. Myocardial cGMP-PKG signalling is crucial for normal cardiovascular physiology, inhibiting maladaptive hypertrophy and enhancing cardiomyocyte compliance through PKG-mediated phosphorylation of the sarcomeric protein titin [19, 20]. Cardiomyocyte stiffness is mainly determined by the elastic sarcomeric protein titin, which functions as a bidirectional spring, responsible for early diastolic recoil and late diastolic distensibility [19]. Titin-based cardiomyocyte stiffness results from dynamic changes in expression of stiff (N2B) and compliant (N2BA) isoforms, from isoform phosphorylation status and from oxidative changes of the N2B segment [19]. Phosphorylation of titin by protein kinase A (PKA) and PKG increase its compliance, thereby acutely lowering cardiomyocyte stiffness (Fig. 1; [11‐15]). Various studies, which procured endomyocardial tissue from patients with HFpEF, HFrEF and aortic stenosis, demonstrated significantly stiffer cardiomyocytes in HFpEF than in HFrEF and aortic stenosis patients [11‐15]. This increased cardiomyocyte stiffness was related to increased titin N2B isoform expression, relative to HFrEF [12], and to reduced phosphorylation of titin [14]. Hypophosphorylation of titin resulted from lower myocardial PKG activity and reduced myocardial cGMP concentration in HFpEF compared with HFrEF and aortic stenosis [15]. The generation of the second messenger molecule cGMP results from activation of soluble guanylate cyclase (sGC) by nitric oxide (NO) and from activation of particulate GC (pGC) by natriuretic peptides (NPs) (Fig. 2; [20]). Once generated, cGMP activates PKG allowing PKG-mediated phosphorylation of a vast number of target proteins, exerting a wide range of downstream effects such as enhanced reuptake of calcium (Ca2+) into the sarcoplasmic reticulum, inhibition of Ca2+ influx, suppression of hypertrophic signalling through inhibition of G-protein coupled receptors and the transient receptor potential canonical channel (TRPC), inhibition of ischaemia-reperfusion injury through phosphorylation of the ATP-sensitive potassium channel and stimulation of LV relaxation and LV distensibility by phosphorylation of troponin I (TnI) and the titin N2B segment (Fig. 2; [19, 20]). Downregulation of myocardial cGMP-PKG signalling in HFpEF is related to reduced myocardial brain-type NP (BNP) expression and increased microvascular inflammation and oxidative stress, which impair both the NP-cGMP and NO-cGMP axes (Fig. 2; [15]). Reduced myocardial BNP expression in HFpEF could have resulted from a number of factors, including concomitant obesity and insulin resistance, which lower myocardial BNP expression [21] and concentric LV remodelling/hypertrophy, which reduces both systolic and diastolic LV wall stress [22]. In addition, low myocardial BNP expression in HFpEF could also have resulted from increased expression of phosphodiesterase (PDE) type 9, which breaks down cGMP specifically generated through the NP-pGC axis [23]. Impaired NO-cGMP signalling could have resulted from the increased inflammation and oxidative stress observed in HFpEF, which was inferred from the high prevalence of comorbidities such as hypertension, obesity and diabetes mellitus type 2 (Fig. 2; [15]).

HFpEF patients are generally older, more often female and have a high prevalence of cardiovascular and non-cardiovascular comorbidities, such as obesity, metabolic syndrome, diabetes mellitus type 2, salt-sensitive hypertension, atrial fibrillation (AF), chronic obstructive pulmonary disease, anaemia and renal dysfunction [24‐26]. Systemic inflammation and endothelial dysfunction are important hallmarks of these comorbidities and are also importantly involved in HFpEF pathophysiology [16, 27]. Myocardial inflammation was shown to contribute to extracellular matrix changes in HFpEF and both myocardial collagen and the amount of inflammatory cells correlated with diastolic LV dysfunction [16]. Endothelial dysfunction is highly prevalent in HFpEF and is related to reduced exercise capacity and worse outcome [27]. Recently, a new paradigm of HFpEF was suggested, which attributes an important role of comorbidities for myocardial dysfunction and remodelling in HFpEF (Fig. 3; [28]).

The new HFpEF paradigm proposes that comorbidities drive structural and functional remodelling in HFpEF through systemic endothelial inflammation [28]. Because of this proinflammatory state, coronary microvascular endothelial cells produce reactive oxygen species, which limits NO bioavailability for adjacent cardiomyocytes. Impaired NO bioavailability results in downregulation of sGC-mediated cGMP-PKG signalling, which augments cardiomyocyte stiffness through hypophosphorylation of titin and increases cardiomyocyte hypertrophy because of impaired PKG-mediated antihypertrophic activity [28]. Furthermore, coronary microvascular endothelial inflammation favours subendothelial migration of leukocytes, which stimulates myofibroblast formation and interstitial collagen deposition. Both increased cardiomyocyte stiffness and interstitial fibrosis induce diastolic LV dysfunction [28].

Microvascular endothelial inflammation is also associated with myocardial capillary rarefaction [29], which was recently demonstrated in HFpEF myocardium [18].

A recent autopsy study demonstrated reduced myocardial capillary density (MCD) in HFpEF patients regardless of the severity of epicardial coronary disease, while the severity of myocardial fibrosis was inversely associated with MCD [18]. Both fibrosis and MCD were similar in those with or without hypertension, suggesting that comorbidities other than hypertension may perpetuate these alterations. Indeed, diabetes mellitus [30] and obesity [30, 31] are known to be associated with MCD. In a recent study performing histological analysis of non-ischaemic myocardium from 57 patients undergoing coronary artery bypass graft surgery, obese patients had lower MCD [31]. Increased body mass index (BMI) was associated with higher pulmonary capillary wedge pressure (PCWP) and lower MCD was associated with both BMI and increased PCWP [31]. Myocardial capillary rarefaction in HFpEF may contribute to decreased maximal myocardial blood flow, impaired oxygen delivery and insufficient metabolic efficiency, which contribute to diastolic LV dysfunction and a higher risk of heart failure in obese individuals [31].

HFpEF is difficult to define as illustrated by various diagnostic classifications and inclusion criteria of clinical trials [8]. These factors contributed to heterogeneity of HFpEF patients recruited into trials and registries. Previously, no consensus was present on the optimal LVEF cut-off value and different cut-offs have been used across classifications and trials ranging from LVEF ≥ 40 % to > 50 % [8]. Although diastolic LV dysfunction represents the dominant abnormality in HFpEF, ancillary mechanisms may also contribute, such as systolic LV dysfunction [32, 33], ventricular-vascular stiffening [34], impaired systemic vasodilatory reserve [35], chronotropic incompetence [33, 35], pulmonary hypertension [36] and right ventricular (RV) dysfunction (Fig. 4; [37, 38]).

The underlying phenotypic heterogeneity is likely far greater in HFpEF than in HFrEF and may be an important reason for the failure of HFpEF clinical trials [42]. For instance, HFpEF patients with pulmonary hypertension and RV systolic dysfunction responded favourably to the phosphodiesterase type 5 inhibitor sildenafil [43], whereas sildenafil exerted no benefit in HFpEF patients without concomitant RV dysfunction [6]. Recently, phenomapping analysis using statistical learning algorithms demonstrated that HFpEF patients recruited according to uniform diagnostic criteria could be divided into three main distinct subgroups, which differed markedly in clinical characteristics, cardiac structure and function, invasive haemodynamics and outcome, despite similar end-systolic and end-diastolic elastances on LV pressure-volume analysis [42]. Therefore, understanding the phenotypic heterogeneity of HFpEF, which includes the aetiological and pathophysiologic heterogeneity of the syndrome, may allow more targeted and successful HFpEF clinical trials.

Recent HFpEF trials demonstrated structural cardiac remodelling in many HFpEF patients including concentric LV remodelling and hypertrophy (59–77 %) and left atrial (LA) dilatation (59–66 %) [5, 7]. Recruitment of LA contractility during stress is impaired in HFpEF and may contribute to the transition from an asymptomatic state to HFpEF, while LA size also predicts clinical outcome [44]. Furthermore, translational studies investigating myocardial tissue from HFpEF patients demonstrated varying degrees of cardiomyoycte hypertrophy, interstitial fibrosis and capillary rarefaction [11, 12, 18], implying distinct and possibly evolutionary stages of myocardial disease progression. Although all large HFpEF trials addressing renin-angiotensin-aldosterone system (RAAS) inhibitors failed to reach statistical significance for the primary outcome, many of them reached borderline significance for a primary endpoint or statistical significance for secondary endpoints, subgroups or post-hoc analyses [3‐5, 7, 45, 46]. Because of these findings, the involvement of RAAS in HFpEF appears more subtle than in HFrEF, probably requiring up-front identification of subgroups of HFpEF patients rather than a ‘one fits all strategy’, which has hitherto been applied in HFpEF in analogy to HFrEF. Such a stratification approach with identification of myocardial structural and functional abnormalities in individual HFpEF patients could be relevant for the determination of potential therapeutic responsiveness and selection of appropriate treatment strategies. For instance, the efficacy of RAAS inhibitors to improve adverse myocardial remodelling is likely very different in HFpEF patients with minor, modest or severe stages of myocardial hypertrophy, fibrosis and capillary rarefaction (Fig. 5; [12]). Indeed, RAAS inhibitors might be effective in HFpEF patients presenting at an early stage, whereas therapeutic efficacy may be lost in HFpEF patients presenting at advanced or final stages. AF is likely to emerge as an indicator of advanced disease in HFpEF. A recent subgroup analysis of HFpEF patients recruited in the RELAX trial indeed showed presence of AF to be indicative of longstanding HFpEF [47]. In this study, HFpEF patients with AF were older than those in sinus rhythm, but had similar symptom severity, comorbidities, and renal function. Despite comparable LV size and mass, AF was associated with worse systolic (lower EF, stroke volume, and cardiac index) and diastolic (shorter deceleration time and larger left atria) function compared with sinus rhythm. Patients with AF had higher PASP and increased NT-proBNP, aldosterone, endothelin-1, troponin I, and C-telopeptide for type I collagen levels, suggesting more severe neurohumoral activation, myocyte necrosis, and fibrosis [47]. Multi-biomarker assessment and cardiac imaging modalities could represent promising tools for diagnostic stratification to identify the underlying stage of myocardial disease in the individual HFpEF patient.