https://orcid.org/0000-0001-7453-3748
Abstract
Heart failure with preserved ejection fraction (HFpEF) is highly prevalent and is associated with relevant morbidity and mortality. However, an evidence-based treatment is still absent. The heterogeneous definitions, differences in aetiology/pathophysiology, and diagnostic challenges of HFpEF made it difficult to define its epidemiological landmarks so far. Several large registries and observational studies have recently disclosed an increasing incidence/prevalence, as well as its prognostic significance. An accurate definition of HFpEF epidemiological boundaries and phenotypes is mandatory to develop novel effective and rational therapeutic approaches.
Introduction
Heart failure with preserved ejection fraction (HFpEF) is highly prevalent and associated with considerable morbidity and mortality. While the prevalence of HFpEF is progressively increasing, few therapeutic successes have been obtained so far. Therefore, after the failure of several trials, the interest of the scientific community has moved backwards to the complex pathophysiology of this condition.1
In fact, the uncertainties related to its definition, its aetiological heterogeneity, and to its epidemiological and clinical relevance should be also dissected to rationally shape future trials on HFpEF patients (i.e. defining entry criteria, making power calculation for sample size based on the expected survival, focus specific treatments on the underlying aetiology and comorbidities).
Therefore, in the current review, we aimed to provide an updated description of HFpEF with a special focus on its definition and epidemiological boundaries, also in light of the different aetiologies and the associated comorbidities. Furthermore, we also summarized the main prognostic studies in this setting and the recent use of predictive prognostic models for HFpEF.
The issue of heart failure with preserved ejection fraction definition
The term HFpEF has replaced the term ‘diastolic HF’, similarly to heart failure with reduced ejection fraction (HFrEF), which is now used in place of ‘systolic HF’. Indeed, while diastolic dysfunction represents a frequent finding in patients with systolic impairment, abnormalities of systolic function can be also demonstrated by myocardial strain analysis in patients with apparently isolated diastolic compromise (Figure 1).2 Therefore, a definition based only on left ventricular ejection fraction (LVEF) has been formulated to avoid pathophysiological ambiguity. However, while patients with reduced LVEF are more easily attributed to the HFrEF phenotype, those with preserved LVEF might have symptoms and signs unrelated to the haemodynamic traits of HF, but due to what is generally considered a comorbidity,3 such as arterial hypertension, atrial fibrillation (AF), obesity, and chronic obstructive pulmonary disease (COPD).
Systolic and diastolic dysfunction across the left ventricular ejection fraction spectrum. Whereas systolic dysfunction represents the landmark sign of heart failure with reduced ejection fraction, diastolic dysfunction may be a common collateral finding in these patients. On the other hand, heart failure with preserved ejection fraction has long been defined ‘diastolic heart failure’, considering impaired myocardial relaxation, rather than contraction, as the pivotal pathophysiological determinant. However, subtle systolic dysfunction could be observed in patients with heart failure with preserved ejection fraction as well, particularly through the use of new diagnostic means (e.g. deformation imaging). DD, diastolic dysfunction; HF, heart failure; HFmrEF, heart failure with mid-range ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LVEF, left ventricular ejection fraction; SD, systolic dysfunction.
In this respect, a more accurate definition of HFpEF was first given in the 2016 European Society of Cardiology (ESC) HF guidelines,4 in which HFpEF was defined by supplementary criteria beyond LVEF >50%, as the presence of:
symptoms ± signs typical of HF (signs may not be present in the early stages of HF, especially in HFpEF, and in patients treated with diuretics).
elevated levels of natriuretic peptides and at least one additional criterion including a relevant structural heart disease (LV hypertrophy—LVH—, and/or left atrium enlargement) or diastolic dysfunction.
Although this definition has improved our capability to properly allocate patients within the HFpEF spectrum, some imprecision is present in real-life populations, and may have partly contributed to the discouraging results of the majority of randomized clinical trials (RCTs).1,5,6 Whereas limited evidence suggests a potential protective role for physical training in this context,1,7 the anti-neurohormonal strategies—considered the standard-of-care therapy for HFrEF—failed to improve outcomes in patients with HFpEF.8 The results of the PARAGON-HF trial, which investigated the possible efficacy of Sacubitril/Valsartan in HFpEF, has further highlighted the limitation of a classification based only on LVEF. Indeed, whereas an overall neutral result relative to the primary endpoint was observed in the whole population, a significant benefit of the treatment has been observed in specific patient subsets, such as in patients with an LVEF <57% and in women.9 In facts, the HFpEF population is a mix of various conditions, more than a single entity. Therefore, whether neurohormonal activation may play some role in the subset with lower LVEF or in specific patient cohorts (e.g. women), other peripheral mechanisms rather than neurohormonal activation could sustain disease progression in patients with higher LVEF. In addition, different cardiovascular substrates, and comorbidity profiles of HFpEF patients could differently impact on pathological mechanisms, symptoms, and responses to therapies.10
Furthermore, thanks to recent therapeutic advances, a growing subset of patients with a prior diagnosis of HFrEF may undergo reverse remodelling during follow-up, achieving an LVEF of 40% and 50%. This condition should be defined HF with recovered LVEF,9 as elegantly reviewed in reference 11. Whereas unified definitions and evidence-based guidelines for this clinical scenario are still missing, it has been recently underlined how these patients should be carefully distinguished from those with HFpEF and still considered and treated as HFrEF, considering the risk of deteriorating to a lower LVEF after treatment discontinuation.12 On the other hand, also patients with an initial diagnosis of HFpEF could sometimes show a progressive reduction in LVEF with time, particularly when considering specific aetiological subsets (e.g. infiltrative diseases or hypertrophic cardiomyopathies).1,13
To overcome the limitations related to an LVEF-centred classification system and to refine the diagnostic definition of HFpEF, thereby meant as an independent clinical condition, two novel multi-parametric scores have been proposed in the last years (Table 1).
Current definitions and diagnostic criteria of HFpEF
| Definition | Diagnostic criteria | Strengths | Weaknesses |
|---|---|---|---|
| ESC 20164 |
| − First standardized definition of HFpEF, replacing the old nomenclature of ‘diastolic HF’ | − Signs may not be present in the early stages of HF − LVEF measurement could be affected by external factors (e.g. different instruments) − Levels of natriuretic peptides may be lower in some categories (e.g. obese) − Not considering the possible improvement or decline of LVEF during follow-up |
| H2FPEF14 | Weighted score derived from 6 variables:
| − Based upon widely available, low-cost, and easy to obtain parameters − Efficient in discriminating the cardiac or non-cardiac origin of exertional dyspnoea | − Young age of patients (mean age 68 ± 11 years) and high prevalence of obesity (mean BMI 33 ± 7.4 kg/m2) in the HFpEF subgroup may limit the performance in different populations |
| HFA–PEFF15 | Stepwise approach:
| − Comprehensive approach to the patient − Not overlooking the role of comorbidities and biomarkers − In deep evaluation of the possible etiologic and/or phenotypic spectra of HFpEF | − The overall complexity may limit applicability in peripheral centres − Predominantly based upon experts’ opinions, still to be tested in validation cohorts |
| Definition | Diagnostic criteria | Strengths | Weaknesses |
|---|---|---|---|
| ESC 20164 |
| − First standardized definition of HFpEF, replacing the old nomenclature of ‘diastolic HF’ | − Signs may not be present in the early stages of HF − LVEF measurement could be affected by external factors (e.g. different instruments) − Levels of natriuretic peptides may be lower in some categories (e.g. obese) − Not considering the possible improvement or decline of LVEF during follow-up |
| H2FPEF14 | Weighted score derived from 6 variables:
| − Based upon widely available, low-cost, and easy to obtain parameters − Efficient in discriminating the cardiac or non-cardiac origin of exertional dyspnoea | − Young age of patients (mean age 68 ± 11 years) and high prevalence of obesity (mean BMI 33 ± 7.4 kg/m2) in the HFpEF subgroup may limit the performance in different populations |
| HFA–PEFF15 | Stepwise approach:
| − Comprehensive approach to the patient − Not overlooking the role of comorbidities and biomarkers − In deep evaluation of the possible etiologic and/or phenotypic spectra of HFpEF | − The overall complexity may limit applicability in peripheral centres − Predominantly based upon experts’ opinions, still to be tested in validation cohorts |
BMI, body mass index; HF, heart failure; HFpEF, heart failure with preserved ejection fraction; LVEF, left ventricular ejection fraction; sPAP, systolic pulmonary arterial pressure; TTE, transthoracic echocardiography.
Current definitions and diagnostic criteria of HFpEF
| Definition | Diagnostic criteria | Strengths | Weaknesses |
|---|---|---|---|
| ESC 20164 |
| − First standardized definition of HFpEF, replacing the old nomenclature of ‘diastolic HF’ | − Signs may not be present in the early stages of HF − LVEF measurement could be affected by external factors (e.g. different instruments) − Levels of natriuretic peptides may be lower in some categories (e.g. obese) − Not considering the possible improvement or decline of LVEF during follow-up |
| H2FPEF14 | Weighted score derived from 6 variables:
| − Based upon widely available, low-cost, and easy to obtain parameters − Efficient in discriminating the cardiac or non-cardiac origin of exertional dyspnoea | − Young age of patients (mean age 68 ± 11 years) and high prevalence of obesity (mean BMI 33 ± 7.4 kg/m2) in the HFpEF subgroup may limit the performance in different populations |
| HFA–PEFF15 | Stepwise approach:
| − Comprehensive approach to the patient − Not overlooking the role of comorbidities and biomarkers − In deep evaluation of the possible etiologic and/or phenotypic spectra of HFpEF | − The overall complexity may limit applicability in peripheral centres − Predominantly based upon experts’ opinions, still to be tested in validation cohorts |
| Definition | Diagnostic criteria | Strengths | Weaknesses |
|---|---|---|---|
| ESC 20164 |
| − First standardized definition of HFpEF, replacing the old nomenclature of ‘diastolic HF’ | − Signs may not be present in the early stages of HF − LVEF measurement could be affected by external factors (e.g. different instruments) − Levels of natriuretic peptides may be lower in some categories (e.g. obese) − Not considering the possible improvement or decline of LVEF during follow-up |
| H2FPEF14 | Weighted score derived from 6 variables:
| − Based upon widely available, low-cost, and easy to obtain parameters − Efficient in discriminating the cardiac or non-cardiac origin of exertional dyspnoea | − Young age of patients (mean age 68 ± 11 years) and high prevalence of obesity (mean BMI 33 ± 7.4 kg/m2) in the HFpEF subgroup may limit the performance in different populations |
| HFA–PEFF15 | Stepwise approach:
| − Comprehensive approach to the patient − Not overlooking the role of comorbidities and biomarkers − In deep evaluation of the possible etiologic and/or phenotypic spectra of HFpEF | − The overall complexity may limit applicability in peripheral centres − Predominantly based upon experts’ opinions, still to be tested in validation cohorts |
BMI, body mass index; HF, heart failure; HFpEF, heart failure with preserved ejection fraction; LVEF, left ventricular ejection fraction; sPAP, systolic pulmonary arterial pressure; TTE, transthoracic echocardiography.
First, the H2FPEF score was presented as a simple method to discriminate between the cardiac or non-cardiac origin of dyspnoea in euvolemic patients with preserved LVEF. In a cohort of consecutive patients (n = 414) undergoing resting and exercise right heart characterization, those with elevated pulmonary capillary wedge pressure were classified as having HFpEF (n = 267).14 At multivariable regression analysis, obesity [i.e. body mass index (BMI) >30 kg/m2], AF, age >60 years, treatment with ≥2 antihypertensives drugs, increased left ventricular filling pressures (i.e. E/e′ >9), and increased systolic pulmonary artery pressure (sPAP) remained as independent predictors of HFpEF and were included in the final score, that was proficiently tested in a validation cohort of 100 patients (area under the curve 0.886; sensitivity 78%; specificity 84%).14 The relatively young age of the patients (mean age 68 ± 11 years) and the high prevalence of obesity (mean BMI 33.0 ± 7.4 kg/m2) within the HFpEF subgroup may have concealed the predictive role of other key variables (such as natriuretic peptides) and thus may show a different performance in different populations. Moreover, as acknowledged by the authors, the fact that natriuretic peptides data were missing in 24% of patients could have further affected their inclusion in the final score.14
Second, a stepwise diagnostic algorithm for HFpEF has been proposed as an expert consensus from the Heart Failure Association of the ESC. According to the consensus, patients should first undergo a pre-test assessment based on their clinical features, laboratory tests, surface electrocardiography, and first-level echocardiography to rule out other causes of dyspnoea. If the clinical suspicious of HFpEF persists, the second step should be represented by the use of natriuretic peptides and a comprehensive echocardiography. A series of parameters are thus derived as either major or minor criteria to obtain a comprehensive score. Finally, stress echocardiography and right-heart catheterization are advised only in case of persisting uncertainty, while advanced imaging and laboratory/genetic testing are indicated to identify specific aetiologies.15 Although apparently more accurate than the H2FPEF, the main limitation of this score remains its complexity and the reliance upon experts’ opinions, without validation studies verifying its accuracy.
Finally, considering the potential combinations of different comorbidity profiles, with distinctive genetic and environmental factors, it should be underlined how the clinical picture of HFpEF appears even more complex and variable. Therefore, a classification of HFpEF based on pathophysiology, disease aetiology, clinical presentation, and ‘phenomapping’ has been suggested to disentangle its complexity.16 Moreover, the distinction of HFpEF patients into ‘phenogroups’, on the basis of different circulating biomarkers, imaging findings, and extracardiac characteristics, may help to identify which subsets are more likely to take advantage from different therapeutic approaches, thus tailoring future clinical trials accordingly.17
Incidence and prevalence trends in heart failure with preserved ejection fraction
The overall prevalence of HFpEF has been reported to be 1.1–5.5% in the general population.18 HFpEF accounts for around 50% of HF cases,19 with a prevalence range spanning from 30% to 70% in different studies. This wide variation is mainly due to the different LVEF cut-off used to define HFpEF previously to the current guideline definition,19 as well as to the diverse populations and clinical settings studied. Although the early detection of individuals at risk of developing HFpEF has been proposed to encourage the adoption of preventive strategies,20 the relative prevalence of HFpEF has been increasing at a rate of 1% per year across the last decades, and it is expected to overcome that of HFrEF in a near future21,22 (Figure 2).
Up-to-date overview of heart failure with preserved ejection fraction epidemiology. In the last decades, the prevalence of heart failure with preserved ejection fraction is increasing and is expected to become the prevalent subtype of heart failure, particularly among women. Nowadays, heart failure with preserved ejection fraction accounts for more than one on three hospitalization for heart failure, characterized by a high risk of re-hospitalization because of cardiac and, more markedly, non-cardiac reasons. In contrast to heart failure with reduced ejection fraction, whose outcomes are improving thanks to therapeutic advances, no specific therapies demonstrated prognostic advantage in heart failure with preserved ejection fraction. Nevertheless, a larger portion of deaths are attributed to non-cardiac causes in this subset, underlining the importance of comorbidities. HF, heart failure; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction.
The proportion of patients hospitalized for HFpEF is also increasing, as evidenced by the Get With The Guidelines–Heart Failure, a cardiovascular quality improvement program including various hospitals in the USA. Among 110 621 patients from 250 centres, 50% (n = 55 083) had HFrEF, 14% (n = 15 184) had HF-mid EF (40–49%), corresponding to heart failure with mid-range ejection fraction (HFmrEF) according to the ESC guidelines, and 36% (n = 40,354) had HFpEF.23 Moreover, in a recent community study considering 40 173 HF-related admissions, the relative prevalence of HFpEF was 39% with a progressive increase in the rate of hospitalization between 2005 and 2014, mostly attributable to an increase in HFpEF in each race-sex group, but more markedly among women.24 Although such findings may have resulted from an improved recognition of HFpEF, a similar trend was reported in other studies.23–25
Cardiovascular substrates of heart failure with preserved ejection fraction
A few clinical features are substantially different between HFpEF and HFrEF, but some overlaps do also exist and are mainly encapsulated in the grey area of HFmrEF. Compared with HFrEF patients, those with HFpEF are usually older, more commonly female, and more likely to have hypertension, AF, and non-cardiovascular comorbidities. On the contrary, male sex, a history of coronary artery disease, higher heart rate on admission, left bundle branch block or ECG abnormalities related to current ischaemia or previous myocardial infarction all significantly increased the odds of having HFrEF in different studies.26,27
Systemic hypertension represents a key determinant of HF and its prevalence is usually higher in HFpEF (in which it ranges from 55% to 90% in various studies) compared to HFrEF populations.28 In particular, whereas LVH, increased ventricular stiffness, and impaired relaxation—secondary to both the increase in overload and neurohormonal activation—may be commonly observed in hypertensive patients, a prior diagnosis of hypertension has been consistently associated with a greater risk of developing HFpEF, independently of possible confounding factors such as aging, sex, or medications.29 On the other hand, although diastolic dysfunction has been historically identified as the main pathophysiological link between systemic hypertension and HFpEF,30 according to more recent studies, also other mechanisms could be involved in this cascade, including subtle systolic dysfunction, reduced atrial function, abnormal ventricular-arterial coupling, and vascular stiffness.28,29
In the last decades, the rising prevalence of HFpEF has been mainly attributed to cardiovascular aging. Indeed, age-related large artery stiffness and the consequent loss of the reservoir function and early wave reflection may be responsible for wide pulse pressure and isolated systolic hypertension, which is hardly reversed by current antihypertensive drugs.31,32 Nevertheless, arterial stiffness, as measured by carotid-femoral pulsed-wave velocity, was not associated with the risk of either HFrEF or HFpEF after adjustment for traditional cardiovascular risk-factors33 and a prospective observational study (IDENTIFY-HF) is ongoing to clarify its role (NCT03186833).
On the other hand, a possible sex interaction has been described as well, as women seem to be at greater risk of developing HFpEF than HFrEF.26,27 A possible explanation has been identified in the different impact of isolated systolic hypertension, that is associated with concentric LVH in women and eccentric LVH in men.34 Furthermore, women show greater pulsatile load than men, which is itself associated with greater subclinical diastolic dysfunction, concentric remodelling, and increased incidence of HFpEF.35,36
The burden of AF is consistently higher in HFpEF than in HFrEF in many studies37–39 and it is associated with older age and worse functional capacity. In a recent observational study from the HF long-term registry of the European Society of Cardiology (n = 14 964), the presence of AF was associated with a greater risk of HF hospitalization across the whole spectrum of LVEF but with all-cause mortality only in HFpEF patients.40 Moreover, the prognostic impact of AF could be greater in women than men in this clinical context.41 However, the pathophysiological role of AF in HFpEF remains poorly understood. A possible hypothesis is that the loss of atrial contribution and rate control may be less tolerated in these patients, due to the increased left atrial stiffness and higher wall stress.
In this regard, the important role of left atrial function in the pathophysiology and progression of HFpEF has been well documented.42 Indeed, left atrial enlargement or dysfunction may be an early marker of HFpEF, and thus the evaluation of these parameters should not be overlooked in subjects at risk [e.g. subjects with hypertension, or obesity, or obstructive sleep apnoeas (OSA)].43 Of note, the evaluation of left atrial deformation by strain analysis allows the evaluation of reservoir, conduit, and contractile functions of the left atrium and may improve the assessment of left ventricular filling pressure and, therefore, of diastolic dysfunction, since its subclinical stages.44 Moreover, abnormal atrial function can contribute to elevated pulmonary vascular resistance, decreased exercise tolerance, and worse outcome in patients with HFpEF.45
The role of ischaemic heart disease in HFpEF has been long debated, and it is usually considered to be less important than in HFrEF. Accordingly, the presence of an ischaemic substrate has been recently associated with a greater decline in systolic function in a cohort of ambulatory patients with HFpEF.46 Again, a possible sex influence has been hypothesized. Indeed, while a history of prior myocardial revascularization represents a possible risk-factor among men hospitalized for HFpEF,47 clinically silent ischaemia or microvascular disease may predispose to HFpEF in women, even though such contribution has been scarcely investigated.48
Beyond systemic hypertension, AF, and ischaemic heart disease, other possible cardiovascular conditions, such as primitive cardiomyopathies and valve diseases, as well as heart and/or vessels involvement in inflammatory, toxic, and metabolic disorders, may predispose to the HFpEF clinical syndrome.1,49 Evidence is increasing on the liaison between HFpEF and cardiac deposition of pathogenic proteins, such as amyloidogenic fibrils. Indeed, following the introduction of non-invasive diagnostic methods (namely diphosphonate scintigraphy), wild-type transthyretin-derived amyloidosis was identified as cause of HFpEF in 13% patients (n = 120), in line with post-mortem studies.50,51 Of note, the prevalence of cardiac amyloidosis could be significantly higher among the elderly. Therefore, such possibility should not be missed among patients with HFpEF of uncertain aetiology, especially considering that different approaches may be advised (also considering the recent therapeutic advances)52 or discouraged in this specific phenotype (i.e. amyloid-related autonomic dysfunction, susceptibility to hypotension in case of use of vasodilators, etc.).
The role of extracardiac comorbidities in heart failure with preserved ejection fraction: much more than innocent bystanders
It is well acknowledged that patients with HFpEF have an increased burden of comorbidities, especially compared with HFrEF1,37–39 (Figure 3). This has been related to a background condition of the systemic pro-inflammatory state, that could further contribute to cardiomyocyte stiffness and interstitial fibrosis.37–39,49 Although the relative prevalence of different comorbidities may be highly variable among different populations, some trends seem to be consistent in most studies.
Pathophysiology of heart failure with preserved ejection fraction. The pathophysiology of heart failure with preserved ejection fraction is far to be clarified. Beyond a cardiovascular substrate of disease, characterized by the complex interaction of ventricular hypertrophy and/or atrial dysfunction, atrial fibrillation, and vascular stiffness, patients with heart failure with preserved ejection fraction often show a variable profile of extracardiac comorbidities that could play a central, albeit not exclusive, role in disease progression. Moreover, many of these comorbidities may underline and sustain a background pro-inflammatory state, furtherly worsening heart function. CA, central apnoeas; CKD, chronic kidney disease; CMs, cardiomyopathies; COPD, chronic obstructive pulmonary disease; HCM, hypertrophic cardiomyopathy; HFpEF, heart failure with preserved ejection fraction; LVH, left ventricular hypertrophy; PH, pulmonary hypertension; RCM, restrictive cardiomyopathy; OSAS, obstructive sleep apnoea syndrome.
The prevalence of obesity is significantly higher as LVEF increases, with over one-third of patients with HFpEF having a BMI >30 kg/m2 in various cohorts.38,40 Although increased body fat is commonly considered to have several detrimental effects on the cardiovascular system,53 mild obesity has been shown to be associated with a better outcome across the whole spectrum of LVEF, including patients with HFpEF, a phenomenon known as ‘obesity paradox’.54 Some specific localization of visceral fat may be notwithstanding more related to a worse outcome than the general increase in body fat.
Indeed, the presence of visceral fat around the upper airways, especially in younger patients, may favour OSA with adrenergic overactivity, oxidative distress, and negative chronobiological consequences.55 An inverse trend has been observed analysing the relative prevalence of either OSA or central apnoeas (CA) in a cohort of HF patients (n = 700) across the spectrum of LVEF. While CA prevailed among HFrEF patients, the prevalence of OSA was higher among HFpEF patients.56 However, while increased neck adiposity could facilitate OSA occurrence,55 a worse diastolic dysfunction may increase the risk of CA also in HFpEF, due to left atrial pressure increase and chemoreflex stimulation.56
In the last decade, several lines of research have also highlighted the role of epicardial fat as a driving force leading to the progression of HFpEF, due to paracrine pro-inflammatory and profibrotic signalling. Conditions such as obesity or other chronic disorders could promote the accumulation and the inflammation of epicardial fat, thus favouring myocardial toxicity, microvascular dysfunction, and pathological remodelling.57 The possible importance of pro-inflammatory and profibrotic mechanisms in this clinical context is further sustained by the evidence that several bio-humoral mediators of active inflammation and fibrosis (e.g. soluble source of tumorogenesis-2 and galectin-3) are commonly elevated and have been associated with the extend of myocardial fibrosis in patients with HFpEF. Although such biomarkers have been shown to be independent predictors of disease severity and poor prognosis in HFpEF—also beyond natriuretic peptides—the possible role of the involved molecular pathways as therapeutic targets is unknown but represents an intriguing question to be addressed in the future.58,59 Similarly, circulating levels of trimethylamine N-oxide—a gut microbiome-mediated metabolite related to the western diet—have been showed to improve risk stratification in HFpEF patients with equivocal levels of natriuretic peptides.60
Interestingly, pro-inflammatory and profibrotic mediators, as well as epicardial and visceral fat, may be further increased in patients with diabetes mellitus.61 Still, epidemiological data suggest that diabetes may be present in up to 45% of patients with HFpEF and is associated with greater clinical severity and poorer outcome.37–39 Therefore, the activation of detrimental bio-humoral cascades, autonomic dysfunction, as well as of altered metabolic pathways may favour cardiac fibrosis and vascular stiffness while increasing at the same time salt-water retention and volume overload.37,61 Interestingly, two inhibitors of the sodium-glucose cotransporter-2 inhibitors, originally approved as antidiabetic drugs, have been recently shown to improve prognosis in patients with HFrEF, regardless of the underlying diabetes.62,63 Whether such findings could be replicated also in the setting of HFpEF is currently under investigation in 2 RCTs.61
Chronic kidney disease (CKD) represents another crucial comorbidity in HFpEF populations: while fluid and sodium retention, uraemic toxins, and neurohormonal activation may precipitate HF syndrome,64 the sustained activation of kidney-related hormonal pathways (e.g. fibroblast-growth-factor-23) may link CKD to HFpEF pathophysiology, promoting vascular stiffness and myocardial fibrosis.65 Moreover, CKD may also contribute to the presence of anaemia, that is more prevalent in HFpEF than in HFrEF, probably also because of the older age of the HFpEF population, and it has been related to worse prognosis in HF.37,39,66
Pulmonary hypertension (PH) constitutes another possible finding in patients with HFpEF. Although Group 2 PH remains the most common in this patient population, some difference has been described compared to HFrEF. In particular, LV stiffness, increased filling pressure, and left atrial dysfunction may be the pivotal determinants of PH in patients with HFpEF, while pro-inflammatory and profibrotic mechanisms could sustain the pathological remodelling of pulmonary vessels.67 Moreover, the overlap with other subtypes of PH may be more common in patients with HFpEF, due to the higher burden of lung disease in these patients.
The prevalence of COPD is consistently greater in HFpEF than in HFrEF patients.37–39 Although bio-humoral (e.g. pro-inflammatory state) and haemodynamic determinants (e.g. abnormal coupling between the pulmonary vein and LV filling due to lung abnormalities) may explain such difference, misdiagnosis of COPD in the presence of dyspnoea and preserved LVEF should also be taken into consideration.68 Furthermore, both natriuretic peptides’ concentrations and echocardiographic markers typical of HFpEF (i.e. increased E/e′ and sPAP) may be abnormal also in patients with COPD.69,70 The research of more specific markers of HFpEF (e.g. left atrial strain, or increased filling pressure) may then prove of value in this context.44,71
Heart failure with preserved ejection fraction hospitalizations and mortality: role of comorbidities and predictive models
It is acknowledged that the rate of re-hospitalization in HFpEF and HFrEF are similar. In the OPTIMIZE-HF registry, that recruited over 40 000 HF patients, the re-hospitalization rate was found to be 29% within 60–90 days of discharge, both in HFrEF and in patients with an LVEF > 40%.72 Similarly, the risk of 30-day re-hospitalization was comparable between HFrEF and HFpEF in a more recent study including 60 640 patients.73 Nevertheless, non-cardiovascular-related diagnosis accounted for the majority of re-hospitalizations among HFpEF patients (59% vs. 47% in HFrEF), highlighting the important role of extracardiac comorbidities in this patient group.73
The mortality rates in HFpEF vary considerably, depending on diagnostic criteria, population analysed (inpatients vs. outpatients) and study design (registries, observational studies, or RCTs). Nonetheless, the prognostic impact of HFpEF seems substantial and has not improved differently from HFrEF, due to the failure of RCTs and absence of guideline-recommended proven therapies.1
Annual mortality in HFpEF ranges from 10% to 30%, being higher in epidemiological studies than in RCTs. In epidemiological studies (observational studies and national registry), the average 1-year mortality of HFpEF is up to 30%.74,75 Due to selection bias (lower age and less comorbidities, strict adherence to therapy) in RCTs the mortality was constantly below 10%.76 In-hospital mortality rates range from 2.5% to 6.5%,27 while long-term 5-year mortality rates vary from 22% to 65%.77
Almost one decade after the first observational studies on outcomes in HFpEF patients, the Meta-analysis Global Group in Chronic Heart Failure (MAGGIC) evaluated 31 observational studies and RCTs and a total of 41 972 patients (n = 10 347 with HFpEF and n = 31 625 with HFrEF), showing that the pooled mortality rate in HFpEF was 121 [95% confidence interval (CI) 117–126] deaths per 1000 patient-years in all the studies, with 146 (95% CI 138–154) deaths per 1000 patient-years in observational studies and 101 (95% CI 96–107) deaths per 1000 patient-years in RCTs. Patients with HFpEF showed 32% lower mortality than those with HFrEF [141(95% CI 138–144) deaths per 1000 patient-years], even after adjustment for age, gender, aetiology, and history of hypertension, diabetes, and AF. Moreover, it was shown that the risk of death did not increase notably until LVEF fell below 40%.78 Remarkably, similar findings have also emerged from a more recent prospective multicentre longitudinal study in an international cohort of 2039 patients. Indeed, after adjusting for age, sex, and other clinical risk factors, patients with HFpEF (n = 574, 28%) showed a lower risk of 2-year all-cause mortality compared to those with HFrEF (n = 1209, 59%) with a HR of 0.62 (95% CI 0.46–0.85). On the other hand, mortality rates were similar between patients with either HFmrEF (n = 256, 13%) or HFpEF.79
According to several observational studies and RCTs, the majority of deaths in HFpEF are due to cardiovascular causes, especially sudden cardiac death, and death related to HF progression. Indeed, cardiovascular death ranges from 50 to 60% of all deaths in observational studies and approaches 70% in RCTs, with sudden death explaining 40–44% and HF 24–30% of cardiovascular deaths.80 However, the clinical definition of HF progression may be different in HFpEF patients compared to HFrEF, while the burden of ventricular arrhythmias is higher as LVEF decreases. Furthermore, as detailed in a recent systematic review, only a minority of the 1608 studies addressing mortality in HFpEF published in a 30-year period reported detailed data about the cause of death and the use of heterogeneous nomenclature could have further affected their external validity.76
On the other hand, it should be highlighted that non-cardiovascular mortality is usually higher in HFpEF than in HFrEF,79 as recently confirmed in a study from our group, enrolling 2791 outpatients with stable HF across the LVEF spectrum, in which non-cardiac mortality was the most prevalent cause of death in HFpEF81 (Figure 4). This suggests that comorbidities should be regarded as a main therapeutic target and object of dedicated quality improvement initiatives in this setting.
Prevalence of cardiac and non-cardiac death in heart failure across left ventricular ejection fraction spectrum. In a prospective cohort of 2791 stable heart failure patients, the prevalence of non-cardiac death was similar between patients with either heart failure with preserved ejection fraction or heart failure with mid-range ejection fraction (62% vs. 54%, respectively) but significantly lower among patients with heart failure with reduced ejection fraction (35%, P < 0.001). HF, heart failure; HFmrEF, heart failure with mid-range ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LVEF, left ventricular ejection fraction. Reproduced with permission from Vergaro et al.81
Among the predictors of poor outcome in HFpEF patients, older age is intuitively associated with higher mortality. Nevertheless, as aging is associated with higher rate of comorbidities, age alone remained as an independent predictor only when considering non-cardiac death, after adjusting for potential confounders.82 With regards to sex-related differences in outcome, some controversy does exist. Indeed, among 4161 patients (2808 women, 67%) admitted to the emergency department for acute HF and LVEF >40%, there were no differences in in-hospital, 30-day, and 180-day all-cause mortality between sexes.83 On the other hand, women may be at lower risk compared with men when considering longer-term outcomes.84 In the last decades, various predictors of poor prognosis have been identified in patients with HFpEF; however, the use of a multiparametric approach (including biomarkers, imaging, and functional tests-derived markers) could be proposed to optimize risk stratification in this context, thus promoting a more personalized treatment.85
The risk to develop HFpEF in the future increases with the number of pre-existing comorbidities.27 Furthermore, the higher the burden of comorbidities the higher the risk of recurrent hospitalizations and overall mortality in HF, independently of LVEF.37,38 The relative contribution of non-cardiac comorbidities on outcomes of either HFrEF or HFpEF patients has been analysed in 31,344 patients in the Swedish Heart Failure Registry from 2000 to 2012. While the impact on all-cause mortality of diabetes (HR 1.57 in HFrEF vs. HR 1.39 HFpEF, P = 0.0002), CKD (HR 1.65 in HFrEF vs. HR 1.44 HFpEF, P = 0.0031), and liver disease (HR 2.13 in HFrEF vs. HR 1.42 HFpEF, P = 0.015) was higher among HFrEF patients, lung diseases had a greater influence in HFpEF patients (HR 1.46 in HFrEF vs. HR 1.66 HFpEF, P = 0.0066).38 Conversely, in a community-based study on 2314 HF outpatients, non-cardiac comorbidities contributed to all-cause mortality, HF-hospitalization, and non-cardiovascular hospitalization, without any significant difference between HFrEF and HFpEF.37 While different populations and methodologies could explain such divergent findings, both these studies underlined the crucial role played by comorbidities, whose tailored targeting should be a fundamental step in patients with HF.
Furthermore, by using the MAGGIC meta-analysis dataset (n = 39 372), a model of mortality prediction in both HFrEF and HFpEF was developed. Among non-cardiac comorbidities, diabetes mellitus, CKD, history of smoking, and COPD emerged as independent predictors of mortality, together with age, male sex, lower blood pressure, lower LVEF, higher New York Class Association class, time since diagnosis, and lack of prescription of anti-neurohormonal therapies. However, the model was not specifically tested in the HFpEF subgroup and the only information provided is that the mortality risk increase due to older age is higher in patients with higher LVEF so that HFrEF and HFpEF survival becomes similar in the elderly.77
Finally, although each comorbidity is associated with a peculiar clinical profile and, potentially, may individually impact on survival, the presence of HFpEF itself is associated with further negative cardiovascular consequences. Indeed, patients with HFpEF when compared with age- and comorbidity-matched patients without HF, actually show worse survival.86,87 These findings thus supported the conclusion that HFpEF is a distinct disease rather than the result of a cluster of comorbidities, and thus needs a dedicated therapeutic effort.
Conclusions
Heart failure with preserved ejection fraction patients are frequently elderly women, with higher BMI, greater prevalence of systolic hypertension and AF and presenting with a high rate of non-cardiac comorbidities, which adversely influence myocardial remodelling and outcome. The use of heterogeneous definitions in various studies makes it difficult to precisely define the epidemiological boundaries for HFpEF. To overcome such a problem, two diagnostic scores have been recently proposed, but some imprecision may still persist. Therefore, a greater effort to improve and validate those scores should be pursued in the next few years. Nonetheless, the relative prevalence of HFpEF has been increasing in the last decades and it is expected to overcome that of HFrEF in the future.
Although the overall mortality remains lower compared with HFrEF, no therapies have been shown to significantly improve outcome in patients with HFpEF, which remain at high risk of re-hospitalization and death for both cardiovascular and non-cardiovascular causes. Therefore, the variable cardiovascular substrate and comorbidity profiles behind HFpEF syndrome could explain—at least partially—the disappointing results obtained in RCTs in this clinical context. We believe that future trials should be tailored on specific aetiologies (i.e. resistant hypertension, cardiac amyloidosis, etc.), specific patient subsets (women, obese patients, elderly), or comorbidities (OSA, renal failure, diabetes) to eventually achieve some benefits at least in some specific patients with HFpEF.
A conceptual shift from a ‘one fits all’ strategy to an individualized approach (precision medicine) based on epidemiological relevance and phenotypic patient characterization may eventually allow the development of drug and/or devices finally able to positively stop or reverse this emerging epidemic condition.
Conflict of interest: none declared.
Comments