REVIEWS

Evaluation and management of heart

failure with preserved ejection fraction
Barry A. Borlaug
Abstract | Heart failure with preserved ejection fraction (HFpEF) has grown to become the
dominant form of heart failure worldwide, in tandem with ageing of the general population and
the increasing prevalences of obesity, diabetes mellitus and hypertension. The clinical syndrome
of HFpEF is heterogeneous and must be distinguished from heart failure with reduced ejection
fraction as well as other aetiologies that have different treatment strategies. The diagnosis of
HFpEF is challenging and ultimately relates to the conceptual definition of heart failure as a
clinical syndrome characterized by symptoms that are associated with a reduced capacity of
the heart to pump blood adequately at normal filling pressures during diastole. Clinical trials
to date have been largely unsuccessful in identifying effective treatments for HFpEF but
evidence supports the use of diuretics, mineralocorticoid antagonists and lifestyle interventions.
Pathophysiological heterogeneity in the presentation of HFpEF is substantial, and ongoing
studies are underway to evaluate the optimal methods to classify patients into phenotypically
homogeneous subpopulations to facilitate better individualization of treatment.

At present, >70% of patients with heart failure (HF) aged accordingly, they might respond in a less uniform fash-
>65 years have a preserved ejection fraction1–3. The inci- ion to treatments. This pathophysiological heterogeneity
dence and prevalence of HF with preserved ejection frac- has served as the impetus for new efforts to more care-
tion (HFpEF) has been growing by 10% every 10 years fully phenotype individual patients13. In this Review, I
relative to HF with reduced ejection fraction (HFrEF). discuss the definitions, diagnosis and treatment strate-
This gap is expected to widen in the coming years owing gies for HFpEF, and highlight the importance of disease
to the ageing of the general population and the increas- phenotyping for this increasingly prevalent syndrome.
ing prevalence of conditions associated with the devel-
opment of HFpEF, particularly obesity, metabolic Definitions of heart failure
syndrome and diabetes mellitus3–5. Although HFpEF and HF was historically defined with the use of clinical
HFrEF have been proposed to be part of the same spec- impressions alone14. Although useful, the assessment of
trum, with HFpEF originally thought to be a precursor to clinical features is highly subjective and less reproduci-
HFrEF, the totality of evidence supports the concept that ble than other diagnostic methods. In 1971, McKee et al.
they are two different disorders6. When evaluated longi- proposed new criteria for diagnosis15 (Table 1). Although
tudinally over many years, the transition from HFpEF to these Framingham criteria have served as a reference
HFrEF is rare7, and risk factors such as obesity, metabolic standard for decades, they are largely predicated on the
syndrome and sedentary lifestyle are known to be asso- presence of congestion at rest. Importantly, this clinical
ciated with HFpEF but not with HFrEF8. In contrast to feature is often absent in ambulatory patients who have
HFrEF, the diagnosis of HFpEF is challenging because well-compensated HF, or in patients with HF who
ejection fraction is normal, cardiac congestion is diffi- develop abnormal haemodynamics exclusively during
cult to evaluate non-invasively and many patients have exercise9,10. Therefore, despite being highly specific,
haemodynamic abnormalities only during exercise9–12. the Framingham criteria have a poor sensitivity for the
Few treatment options have been shown in clinical trials diagnosis of HF. Other diagnostic criteria from different
to be effective in reducing the morbidity and mortality professional societies, on the basis of clinical trial entry
Department of Cardiovascular associated with HFpEF1. One potential reason for the criteria, and from individual investigators have been
Medicine, Mayo Clinic,
Rochester, MN, USA.
failure of clinical trials to identify effective treatment applied12,16. Disease prevalence and outcome rates vary
strategies thus far is thought to relate to the pathophysi- markedly depending on the different measures used to
e-mail: borlaug.barry@
mayo.edu ological heterogeneity within the broader clinical spec- define HFpEF16.
https://doi.org/10.1038/ trum of what is considered HFpEF. Not all patients with Ideally, HF should not be defined subjectively but
s41569-020-0363-2 HFpEF have uniform pathophysiological features and, rather conceptually and with the use of objective measures

Nature Reviews | Cardiology

Reviews

Key points The definition of a ‘preserved’ ejection fraction

in the literature has consisted of an ejection fraction in
• Heart failure with preserved ejection fraction (HFpEF) has become the most common the range of 35–50%1. The 2016 ESC/Heart Failure
form of heart failure, associated with substantial morbidity and mortality. Association (HFA) guidelines included a new category,
• HFpEF is defined haemodynamically as a clinical syndrome associated with a lack of termed HF with mid-range ejection fraction (HFmrEF),
capacity of the heart to pump blood adequately without the requirement for elevated that was intended to stimulate greater research into the
cardiac filling pressures. cohort of patients with HF and an ejection fraction in
• Typical HFpEF must be distinguished from other causes of the clinical syndrome the range of 40–49%19. Most studies to date, includ-
of heart failure, which are treated differently. ing ancillary analyses from clinical trials that involved
• Diagnosis is challenging and requires the demonstration of objective evidence patients with HF and a mildly depressed ejection frac-
of congestion or poor cardiac output using assessment of clinical history, physical tion, have shown that patients with HFmrEF respond to
examination, natriuretic peptide testing, echocardiography data and invasive
conventional therapies proven to be beneficial in patients
exercise testing.
with HFrEF, suggesting that HFmrEF might simply be
• To date, most clinical trials on the efficacy of treatments for HFpEF have produced
a mild form of HFrEF12,20,21. As a result, some clinicians
neutral results, but strong evidence supports the use of diuretics, mineralocorticoid
receptor antagonists and exercise training as effective therapies.
have proposed the classification of all patients with HF
and an ejection fraction <50% under the syndrome of
• Ongoing studies are evaluating the utility of more rigorous pathophysiological
characterization of HFpEF into distinct phenotypes to improve the matching of
HFrEF22. A case can also be made for an even higher
individualized treatments to patients who are most likely to respond favourably. ejection fraction threshold. The few patients who pro-
gress from HFpEF to HFrEF are more likely to have a
borderline reduction in ejection fraction (50–55%)7, and
that reflect this conceptual definition. HF is well accepted evidence from a 2019 clinical trial suggests that patients
as a clinical syndrome characterized by symptoms such as with an ejection fraction of up to 57% might respond
dyspnoea and fatigue caused by cardiac dysfunction. differently to treatment than patients with an ejection
Therefore, HF can be defined haemodynamically in a fraction >57%23. However, consensus statements from
patient with a preserved ejection fraction as a reduced both the ACC/AHA/Heart Failure Society of America
capacity of the heart to pump blood to the body at a rate (HFSA) and the ESC continue to use an ejection fraction
commensurate with the needs of the body at rest and ≥50% as the threshold to identify HFpEF19,24, and this
with exertion at normal diastolic filling pressures9,10,17,18 cut-off value is used in this Review.
(Table 1). This definition requires invasive measure- Patients with HFpEF with the most severe abnor-
ments, limiting its applicability in everyday practice. malities in cardiac function, excessive sodium and water
However, an important strength is that this definition retention, and high plasma levels of natriuretic peptides
provides an objective gold standard against which other have a greater risk of death or hospitalization for HF
less invasive approaches can be measured1. This concep- than patients with HFpEF with normal resting hae-
tual, haemodynamics-related definition forms the basis modynamics but abnormal responses upon exercise16.
for diagnostic strategies, as discussed below10–12. However, this finding does not refute the observation
that the clinical syndrome of HFpEF (as defined concep-
tually in Table 1) is also present in the group of patients
Table 1 | definitions of heart failure with abnormalities that only become apparent during
Scheme definition/criteria application exercise. Despite normal resting haemodynamics, these
patients with less advanced HFpEF also develop cardiac
Framingham Major criteria: paroxysmal Requires two major criteria injury25 and an impairment in myocardial function26,27
nocturnal dyspnoea; jugular vein or two minor criteria plus one
distention; rales; radiographic major criterion simultaneously; that worsen with exercise, leading to marked elevation
cardiomegaly; acute pulmonary minor criteria are acceptable in pulmonary vascular pressures that causes lung con-
oedema; S3 gallop; increased only if not ascribable to other gestion28, altered breathing mechanics29 and dyspnoea29.
central venous pressure (>16 cm); conditions Over sustained periods of time, these haemodynamic
hepatojugular reflux; weight loss
>4.5 kg in 5 days in response to perturbations that arise during exercise lead to reduced
treatment aerobic capacity30 and increased risk of hospitalization
Minor criteria: bilateral ankle
for HF16,31, and are associated with increased mortality32.
oedema; nocturnal cough; Typical ‘garden variety’ HFpEF, as seen among most
dyspnoea on ordinary exertion; elderly patients, is associated with comorbidities such
hepatomegaly; pleural effusion; as obesity, hypertension, kidney disease, diabetes and
decrease in vital capacity metabolic syndrome1,13. The underlying mechanisms
by one-third from maximum
recorded; tachycardia (heart by which these conditions cause HF are not yet fully
rate >120 bpm) understood, but are clearly related to stresses associated
with these comorbidities, including intermittent pres-
Haemodynamic Lack of capacity of the heart to ↑ Left ventricular filling pressure
pump blood adequately to meet at rest or exercisea; cardiac sure overload, metabolic duress, endothelial and coro-
metabolic demand at normal output reserve <80%, predicted nary microvascular dysfunction, tissue ischaemia and
cardiac filling pressures based on metabolic demandb fibrosis, and low-grade systemic inflammation1,13,33.
a
Defined by left ventricular end diastolic pressure or pulmonary capillary wedge pressure A number of other cardiovascular disorders that
≥15 mmHg at rest and/or ≥25 mmHg with exercise. bPredicted increase in cardiac output with differ from this typical form of HFpEF can also result
exercise is the product of an increase in oxygen consumption multiplied by six. Cardiac output
reserve is defined as the quotient of measured exercise minus rest output and predicted minus in the same haemodynamic profile of elevated filling
rest output. pressures with an ejection fraction ≥50% but should

www.nature.com/nrcardio
 Reviews

Table 2 | Causes of the clinical syndrome of HFpEF

aetiology diagnostic tools Treatment considerations distinct from primary HFpEF
HFpEF See section on diagnosis NA
Cardiac amyloidosis Monoclonal proteins, radionuclide Tafamidis (for transthyretin amyloidosis) or chemotherapy
scintigraphy, biopsy (for light-chain amyloidosis); avoid neurohormonal
antagonists
Hypertrophic Echocardiography, cardiac MRI β-Blockers, calcium-channel blockers or septal-reduction
cardiomyopathy therapies (for obstructive cardiomyopathy); avoid
vasodilators
Cardiac sarcoidosis Cardiac MRI, FDG-PET, biopsy Immunosuppressive agents
Constrictive Echocardiography, cardiac MRI or Pericardiectomy
pericarditis CT imaging, invasive haemodynamic
measurements
Valvular heart Echocardiography, invasive Surgical or percutaneous valve interventions
diseasea haemodynamic measurements with
ventriculography
Coronary artery Invasive coronary angiography, Revascularization, aspirin, statins, β-blockers and nitrates
diseasea stress imagingb or CT imaging
High-output heart Evaluate for arteriovenous shunts Treatments directed at the cause of high cardiac output
failure and liver disease (such as fistula ligation for shunts, liver transplantation
for cirrhosis)
Myocarditis Cardiac MRI, endomyocardial biopsy Immunosuppressive agents for some types (such as giant
cell myocarditis or eosinophilic myocarditis)
Toxinsa Assessment of clinical history, blood Removal of offending toxin (such as alcohol, cocaine,
testing, endomyocardial biopsy chemotherapy or radiation therapy, or heavy metals)
FDG-PET, fluorodeoxyglucose-positron emission tomography; HFpEF, heart failure with preserved ejection fraction; NA , not
applicable. aThese aetiologies can coexist with ‘primary’ HFpEF. Contribution to the clinical syndrome needs to be judged on a
case-by-case basis. bSensitivity and specificity of stress imaging is poorer in HFpEF than in other conditions35.

not be regarded as true HFpEF (Table 2), and include (LV) ejection fraction. In patients with HFpEF, diag-
aetiologies with distinct mechanisms, separate natural nostic evaluation is much more challenging than in
histories and unique treatments. Some clinicians prefer patients with HFrEF given that the ejection fraction is
to merge all these specific aetiologies together under normal and, therefore, similar to that of patients with
the umbrella of HFpEF, mirroring the practice used for non-cardiac dyspnoea. The diagnosis of HFpEF fun-
HFrEF19,24. However, this practice of excessive merging damentally relies upon demonstration of evidence of
of patient subgroups only serves to further increase het- congestion or high filling pressures, tying back into the
erogeneity34. Given the availability of specific treatments a priori haemodynamics-related definition of HFpEF
for these alternative causes, the inappropriate grouping (Table 1). Clinical history and physical examination alone
of disease types could deprive patients of potential should be sufficient in patients with frank congestion at
life-saving therapies (Table 2). rest, for example, in patients who fulfil the Framingham
One unresolved problem in field of HFpEF is in how criteria (Table 1). However, in patients with a lesser
causality is ascribed. Whether a patient with apparent degree of congestion or with abnormal haemodynamics
‘primary’ HFpEF with evidence of coexisting coronary that develop only during exercise, evidence of high filling
artery disease or heart valve disease on echocardiogra- pressures might only be obtained with advanced imaging
phy should be diagnosed with HFpEF or with the latter or invasive cardiopulmonary exercise testing9–12,38.
disorders is unclear35–37. Furthermore, the degree of cor-
onary (or valve) disease severity necessary and sufficient Clinical history and physical examination. A detailed
to cause symptoms of cardiac failure, as opposed to being clinical history and physical examination is the first step
a secondary condition that coexists with HFpEF, is also in the evaluation of potential HFpEF. Symptom severity
unclear. The distinction might be very difficult to make, and provocative factors should be determined. Certain
particularly because patients with HFpEF show evidence symptoms, such as orthopnoea and paroxysmal noc-
of cardiac injury (elevated troponin levels) and a myocar- turnal dyspnoea, are highly specific for HF and provide
dial supply–demand mismatch that worsens during exer- robust evidence in support of the diagnosis (Table 1) but
cise25. These questions will only be answerable through have low sensitivity39. Other symptoms such as exer-
data obtained from longitudinal studies and randomized tional dyspnoea and fatigue are much less specific for
clinical trials but, at present, this ascription is performed HF but also much more common and thus more sen-
based on the judgement of the caregiving physician. sitive. Some patients with HFpEF complain of chest
discomfort, which might resemble angina or have more
Diagnosis of HFpEF atypical characteristics40. Chest pain in HFpEF might
The diagnosis of HFrEF is straightforward and made be related to epicardial coronary disease35, microvascu-
upon symptomatic presentation of dyspnoea and an lar dysfunction41–43, myocardial supply–demand mis-
echocardiogram that reveals a depressed left ventricular match that worsens with activity25 or any combination

Nature Reviews | Cardiology

Reviews

of these factors. Peripheral oedema, abdominal fullness, A physical examination is generally insensitive for
anorexia and early satiety are symptoms that gener- HFpEF diagnosis but might provide important clues.
ally indicate advanced HFpEF with right-sided heart The vast majority of patients with HFpEF (80–90%)
failure44. Dizziness and orthostatic intolerance suggest have a history of hypertension1,13 but blood pressure
the possibility of cardiac amyloidosis. In addition to the might be well controlled by medication at the time of
elicitation of symptoms, the medical history should be evaluation. Overall, 60–75% of patients with HFpEF
reviewed for comorbidities that are strongly linked to have a high BMI (>30 kg/m2)46. Careful evaluation for
HFpEF. A history of atrial fibrillation (AF), obesity, dia- evidence of congestion should be performed, with par-
betes or metabolic syndrome all substantially increase ticular attention to the assessment of jugular distention,
the likelihood of HFpEF11,45 (Table 3). gallop sounds and peripheral oedema47,48. The presence
of an irregular heart rhythm suggests AF, whereas pul-
monary rales are generally absent in chronic HFpEF
Table 3 | Clinical, laboratory and imaging predictors of HFpEF11 unless the patient has acutely decompensated HF at the
time of evaluation49.
Parameters odds C-statistic Sensitivity Specificity P value
ratio (%) (%)
Imaging and electrocardiography. A 12-lead electro-
Clinical
cardiogram should be routinely obtained to evaluate
Age >60 years 6.20 0.704 80 60 <0.0001 for AF11,45 as well as other evidence that might suggest
Obesity 3.46 0.651 65 65 <0.0001 coronary disease or underlying cardiomyopathy. Other
Grade II obesity or highera 4.02 0.615 35 88 <0.0001
findings from the electrocardiogram that support the
diagnosis of HFpEF include a prolonged PR interval
Chronic kidney disease 3.38 0.584 26 90 <0.0001 or paced rhythm11 (Table 3). In particular, chest radiog-
stage 3 or higher
raphy should be performed to evaluate for pulmonary
Hypertension 5.33 0.664 86 47 <0.0001 aetiologies. Pulmonary congestion is usually absent in
Atrial fibrillation 12.35 0.652 35 96 <0.0001 patients with compensated HF, but cardiomegaly or
Diabetes mellitus 2.80 0.579 28 88 0.0003 pleural effusions, if present, support the diagnosis of
HFpEF (Table 3).
Prediabetes or diabetes 2.82 0.624 55 70 <0.0001
Transthoracic echocardiography has a central role in
Laboratory the evaluation of HFpEF and has been reviewed in detail
NT-proBNP >125 pg/ml 3.74 0.649 77 53 <0.0001 elsewhere38. Echocardiography allows for the exclusion
of patients with HFrEF or other causes of clinical HF
NT-proBNP >275 pg/ml 4.82 0.680 59 77 <0.0001
with a normal ejection fraction such as valvular or peri-
NT-proBNP >450 pg/ml 4.93 0.657 46 85 <0.0001 cardial disease (Table 2). The most robust indicators that
Electrocardiography support the diagnosis of HFpEF include an elevation in
First degree AV block 1.92 0.532 15 92 0.006 the E/e′ ratio, which correlates with LV filling pressures,
and an increase in the estimated pulmonary artery sys-
PR interval per millisecond 1.02 0.644 69 56 <0.0001
tolic pressure (PASP), which develops secondary to left
Pacemaker 21.30 0.560 13 99 <0.0001 atrial hypertension11,50 (Table 3). Other findings that
Chest radiography increase the probability of HFpEF include left atrial
Cardiomegaly 7.56 0.601 24 96 <0.0001
enlargement, reduced left atrial strain, impaired LV early
diastolic tissue velocity (e′), reduced global longitudinal
Pleural effusion 4.96 0.537 9 98 0.002 strain, and right ventricular (RV) dilatation or dysfunc-
Echocardiography tion11 (Table 3). Notably, all these indices are either surro-
Global longitudinal 2.10 0.591 62 56 0.0004 gates for elevated filling pressure (E/e′ ratio and PASP),
strain <16% a sequelae of chronically elevated filling pressures (left
LV hypertrophy 2.55 0.570 26 88 0.0006 atrial remodelling and RV dilatation or dysfunction), or
causes of high filling pressure (reduced e′ velocities and
LA volume index >30 ml/m 2
5.65 0.704 70 71 <0.0001
strain)38. Most of these indices are specific but have low
E/e′ ratio >9 5.23 0.687 78 59 <0.0001 sensitivity for HFpEF (Table 3). As such, these parame-
E/e′ ratio >13 5.20 0.661 46 86 <0.0001 ters are most useful for diagnosis when abnormalities
are present but do not exclude disease when absent10,11.
Septal e′ velocity <7 cm/s 2.90 0.619 48 76 <0.0001
Stress imaging for myocardial ischaemia in patients with
PASP >35 mmHg 5.05 0.657 46 86 <0.0001 HFpEF is often performed but is of uncertain utility
RV fractional area 4.88 0.637 39 88 <0.0001 given the high false-positive and false-negative results
change <48% in this patient population35.
TAPSE <21 mm 3.69 0.637 46 81 <0.0001
Qualitative RV dysfunction 4.26 0.578 22 94 <0.0001 Role of natriuretic peptides. Plasma natriuretic peptide
levels increase in response to elevations in cardiac wall
Qualitative RV dilatation 3.45 0.598 32 88 <0.0001 stress, which is related to chamber dimension and dis-
AV, atrioventricular; HFpEF, heart failure with preserved ejection fraction; LA , left atrial; LV, left tending pressure according to the law of Laplace51,52.
ventricular; NT-proBNP, N-terminal pro-B-type natriuretic peptide; PASP, pulmonary artery
systolic pressure; RV, right ventricular; TAPSE, tricuspid annular plane systolic excursion. Similar to HFrEF, patients with HFpEF and elevated
a
Defined as a BMI ≥35 kg/m2. natriuretic peptide levels have an increased risk of HF

www.nature.com/nrcardio
 Reviews

hospitalization and death compared with patients with is obtained in the radial artery for measurement of blood
HFpEF and normal natriuretic peptide levels52–54. Given pressure and sampling of arterial blood. Exercise can
that natriuretic peptide levels are easy to measure and be performed in either the supine or upright position
can help to identify patients with HFpEF (Table 3) using an ergometer. The protocols employed depend on
and stratify those at higher risk of events, the presence the abilities and fitness level of the patient but generally
of elevated natriuretic peptide levels in plasma has been start at a low workload of 20 W and increase in 2–3 min
required for enrolment eligibility in a large number of intervals by 10–20 W increments to exhaustion. Oxygen
clinical trials for HFpEF in the past 10 years1, and test- consumption is measured through a mouthpiece or
ing of natriuretic peptide levels is recommended in tight-fitting mask, and cardiac output is determined
contemporary guidelines12,19,24. with the use of the direct Fick method9–12. Symptoms of
However, compared with patients with HFrEF, fatigue and dyspnoea are recorded at each stage.
patients with HFpEF have lower wall stress owing to a Because invasive exercise testing directly assesses
smaller cavity size and thicker ventricular walls6. Many the parameters that define HFpEF (Table 1), this meas-
patients with HFpEF have normal distending cardiac urement serves as the gold standard test to prove
pressures at rest and increased distending pressures dur- (or refute) that HFpEF is the cause of symptoms9–12.
ing exercise when the patient becomes symptomatic9,10. HFpEF is defined invasively by an elevated pulmo-
Therefore, natriuretic peptide levels are substantially nary capillary wedge pressure ≥15 mmHg at rest and
lower in patients with HFpEF than in patients with ≥25 mmHg during exercise9–12. Invasive cardiopulmo-
HFrEF, and are indeed often normal1,9,10,52,55. nary exercise testing also enables the assessment of
The most conspicuous difference between patients cardiac output reserve, which requires measurement
with HFpEF with and without elevated natriuretic pep- of cardiac output and oxygen consumption in tan-
tide levels is in the prevalence of obesity46,55. Obesity dem. Under normal conditions, cardiac output should
is associated with reduced natriuretic peptide levels increase by 6 ml for every 1 ml increase in whole-body
in plasma, related in part to increased clearance via oxygen consumption17. Values <80% of this level are con-
natriuretic peptide receptors in adipocytes56 as well as sidered to indicate impairment in cardiac output reserve,
with a reduction in cardiac chamber distention owing to which is commonly seen in HFpEF9,17,27,61.
heightened pericardial restraint46. A 2017 study showed Invasive exercise testing is associated with high
that 18%, 30% and 40% of patients with invasively costs and the need for operator expertise. Alternative
proven HFpEF have N-terminal pro-B-type natriu- approaches that have been evaluated include acute
retic peptide levels in plasma below the thresholds of saline loading in the catheterization laboratory, which
≤125 pg/ml, <225 pg/ml and <300 pg/ml, respectively, can improve the discrimination of patients with HFpEF
used in clinical practice and trials10. An elevated circu- from those without HF compared with the use of
lating natriuretic peptide level is now understood to not resting measurements alone62. Although helpful, this
be necessary for a diagnosis of HFpEF11,12, although the method is not as robust as exercise testing59 and still
presence of elevated natriuretic peptide concentrations requires cardiac catheterization. Non-invasive cardio­
does increase the likelihood of HFpEF (Table 3). pulmonary exercise testing has been recommended
for the evaluation of unexplained dyspnoea, and might
Exercise testing. A substantial proportion of patients provide important data that can facilitate the identifi-
with HFpEF develop abnormalities in cardiac filling cation of pulmonary abnormalities63. However, a 2018
pressures exclusively during exercise9–12,27,57,58. Given study found that non-invasive cardiopulmonary exer-
that invasively measured haemodynamics are normal cise testing did not robustly discriminate patients with
at rest in these patients, they would not be expected to HFpEF from those with non-cardiac causes of dyspnoea,
show evidence of elevated cardiac filling pressures at particularly among patients with mild-to-moderately
rest on echocardiography, by physical examination or depressed functional capacity30.
by measuring natriuretic peptide levels10. Therefore, each Exercise echocardiography has increasingly been
of the aforementioned tests has a poor sensitivity for the advocated for as an alternative to invasive stress testing64.
identification of HFpEF10,11 (Table 3). Given that variables on Doppler echocardiography, such
The diagnostic conundrum in these patients has led as the E/e′ ratio, correlate moderately well with invasively
to the introduction of invasive haemodynamic exercise measured pressures, a number of investigators have eval-
testing to evaluate for HFpEF9–12,31,57,59–61. This testing uated the utility of exercise echocardiography stress test-
involves right heart catheterization for the direct meas- ing for the diagnosis of HFpEF. Some studies have shown
urement of central haemodynamics at rest and then dur- robust correlations of variables measured using exercise
ing graded increases in exercise workload to volitional echocardiography with invasively obtained data10,65–67,
fatigue, ideally with simultaneous expired gas analysis to but other studies have not reproduced these favourable
evaluate aerobic capacity (peak oxygen consumption). results61,68–71.
Venous access is generally obtained through the jugu-
lar vein, with a balloon-tipped catheter advanced to the Integrated diagnostic approaches. In the past 2 years, an
pulmonary artery for measurement of pulmonary artery evidence-based study11 and an expert consensus guide-
pressure and pulmonary capillary wedge pressure, and line12 have reported on potential approaches for the diag-
sampling of venous blood9–12. Central venous pressure nostic work-up of HFpEF. In the first study, Reddy et al.
is measured throughout the study with the use of the evaluated 414 consecutive patients undergoing eval-
sidearm of the central venous sheath, and arterial access uation for unexplained dyspnoea (267 patients with

Nature Reviews | Cardiology

Reviews

Patient with unexplained dyspnoea

Assessment of pre-test probability:

clinical evaluation plus echocardiography

H2FPEF score HFA-PEFF score

Score Clinical Characteristics Points Domains Major criteria Minor criteria
label variable (2 points) (1 point)
H2 Heavy BMI >30 kg/m2 2 Functional • Septal e′ <7 cm/s • Average E/e′ ratio 9–14
Hypertension ≥2 antihypertensive 1 • Lateral e′ <10 cm/s • Global longitudinal
medications • Average E/e′ ratio ≥15 strain <16%
• Velocity of the tricuspid
F Atrial Persistent or 3 regurgitation >2.8 m/s
fibrillation paroxysmal (pulmonary artery systolic
pressure >35 mmHg)
P Pulmonary Pulmonary artery 1
hypertension systolic pressure Morphological • LAVI >34 ml/m2 • LAVI 29–34 ml/m2
>35 mmHg • LVMI ≥149/122 g/m2 • LVMI >115/95 g/m2 (m/w)
(measured on Doppler (m/w) and relative • Relative wall thickness >0.42
echocardiography) wall thickness >0.42 • Left ventricular wall thickness ≥12 mm
E Old age Age >60 years 1 Biomarker • NT-proBNP >220 pg/ml • NT-proBNP 125–220 pg/ml
(sinus rhythm) • BNP >80 pg/ml • BNP 35–80 pg/ml
F Filling E/e′ ratio >9 1
pressure (measured on Doppler Biomarker • NT-proBNP >660 pg/ml • NT-proBNP 365–660 pg/ml
echocardiography) (atrial fibrillation) • BNP >240 pg/ml • BNP 105–240 pg/ml

Low probability, unlikely HFpEF Intermediate probability High probability, likely HFpEF
(H2FPEF score of 0–1, HFA-PEFF score of 0–1) (H2FPEF score of 2–5, HFA-PEFF score of 2–4) (H2FPEF score of 6–9, HFA-PEFF score of 5–6)

Haemodynamic exercise test

Fig. 1 | diagnostic approach for HFpEF. In the patient with unexplained dyspnoea, assessment of the pre-test probability
that heart failure with preserved ejection fraction (HFpEF) is present is first performed via clinical, echocardiographic and
laboratory testing. Two scoring systems have been developed for this purpose: the H2FPEF score11 (left) and the HFA–PEFF
score12 (right). With either system, patients deemed at high and low probability are generally diagnosed as having or not
having HFpEF, respectively, without further testing. In patients with intermediate probability of HFpEF, haemodynamic
exercise testing is needed, which can be performed using cardiac catheterization or stress echocardiography. BNP,
B-type natriuretic peptide; LAVI, left atrial volume index; LVMI, left ventricular mass index; m/w, men/women; NT-proBNP,
N-terminal pro-B-type natriuretic peptide.

HFpEF and 147 control patients with non-cardiac causes validation cohort of 100 patients (61 with HFpEF and
of dyspnoea)11. All patients underwent definitive assess- 39 controls), in which the H2FPEF score showed an
ment of case or control status by the gold standard of area under the curve of 0.886 (P < 0.0001)11. The num-
invasive haemodynamic exercise testing. The discrimi- ber of points in the score provides an estimate of the
natory capacity of different clinical findings from med- probability that HFpEF is the cause of dyspnoea (Fig. 1).
ical history, laboratory testing, electrocardiography and A model incorporating the same values as continuous
imaging were evaluated to determine diagnostic utility variables was found to have even better discriminatory
(Table 3). The authors then derived a weighted, com- capacity11,72. This method enables a Bayesian approach
posite score based on six variables (termed the H2FPEF to HFpEF diagnosis, whereby patients with a low or high
score), ranging from 0 to 9, incorporating the strong- likelihood of disease are diagnosed, and the more costly
est independent parameters that discriminated cases and more definitive test (that is, invasive exercise test-
from controls and using the variables that remained ing) is reserved for patients with an intermediate pre-test
independent predictors of HFpEF in a multivariable probability of disease. Importantly, the H2FPEF score
logistic model (Fig. 1). Some variables, such as left atrial can be calculated using simple clinical characteristics
volume, showed good discriminatory capacity in a uni- and echocardiographic data that are universally available
variate analysis (Table 3) but did not remain significant in clinical practice11,72.
in the multivariable model11. Plasma natriuretic pep- Since the original publication of the H2FPEF score by
tide levels were not independently predictive of HFpEF Reddy et al.11, the generalizability and prognostic value
in this analysis and were therefore not included in the of the score have been demonstrated in two independ-
final score11. ent ancillary analyses73,74. A further study has evaluated
In the derivation cohort, the H2FPEF score displayed a series of patients who were thought to have HFpEF
robust discrimination of cases from controls, with on the basis of physician opinion together with patients
an area under the curve of 0.841 (P < 0.0001)11. This deemed to be at risk of HFpEF but assumed not to have
discriminatory capacity was confirmed in a separate the syndrome75. The overall discriminatory capacity of

www.nature.com/nrcardio
 Reviews

the H2FPEF score was favourable in patients with a nor- ESC/HFA algorithm, known as the HFA-PEFF score
mal ejection fraction but not in those presenting with (Fig. 1), involves pre-test assessment, diagnostic work-up
dyspnoea, leading the authors to question the veracity with echocardiography imaging data and natriuretic
of the H2FPEF model. However, fundamental flaws in peptide score, advanced work-up with functional test-
the design of this study raise questions about the con- ing in case of uncertainty, and final aetiological work-up.
clusions75. The true cause of dyspnoea was not defini- In patients with a high score, the diagnosis is deemed
tively ascertained in any of the patients. Patients with secure, whereas in patients with very low scores, HFpEF
dyspnoea who were assumed not to have HFpEF were is deemed unlikely to be present. In patients with inter-
analysed as a disease-free control group, but definitive mediate HFA-PEFF scores, the algorithm calls for
assessment was not performed to prove that HFpEF was exercise haemodynamics assessment through either
not present75. This design flaw is relevant given that pre- invasive exercise testing or exercise echocardiography,
vious studies have shown that 50–60% of patients with though the literature is more equivocal for the latter12.
indeterminate dyspnoea and a normal ejection fraction At present, no direct comparisons of the approaches to
are ultimately diagnosed with HFpEF based on invasive diagnosis based on the H2FPEF score or the HFA-PEFF
testing9–11. Similarly, HFpEF case status was also not score have been made, but they are likely to perform
confirmed in this cohort and, notably, patients deemed similarly well given the central role of exercise testing
to have HFpEF had only borderline echocardiographic in both algorithms among patients with diagnostic
abnormalities associated with HFpEF, whereby only uncertainty (Fig.1).
38% had an E/e′ ratio >9 and only 43% had an estimated
PASP >35 mmHg (ref.75). Therefore, although invasive Treatment of HFpEF
testing is clearly not necessary in all patients undergoing To date, all large-scale clinical trials to assess the effi-
an evaluation for HFpEF, when performing studies to cacy of medical therapies for HFpEF have had neutral
assess the discriminatory capacity of different diagnostic results1,23,76–91 (Fig. 2; Table 4). However, the strategy of
approaches, the definitive evaluation of case or control decongestion has been shown to reduce hospitalization
status is essential, rather than relying on assumptions rates92,93 and lifestyle interventions, such as exercise
and physician opinion alone. training, can improve functional capacity and quality of
A consensus statement designed to help guide the diag- life94,95. As such, current management of HFpEF is pri-
nostic evaluation of HFpEF was published by the ESC/HFA marily directed towards control of hypervolaemia with
in 2019 (ref.12). Although the recommended diagnos- diuretics, the use of mineralocorticoid antagonists in
tic algorithm is not evidence-based like the H2FPEF selected patients, and exercise training and management
score11, the approach is very similar. The proposed of associated comorbidities1,19,24.

a b
Aldo-DHF83
CHARM-Preserved77
RELAX84

INDIE-HFpEF86
I-PRESERVE78
SECRET exercise95

PEP-CHF76 SECRET diet95

–1.0 0 1.0 2.0
Treatment effect on peak VO2
TOPCAT82 (ml/min/kg)

PARAGON23 NEAT85

EDIFY90

CHAMPION92,93 SECRET exercise95

SECRET diet95
0.4 0.6 0.8 1.0 1.2 –20 0 20 40
Hazard ratio or incidence rate ratio Treatment effect on 6-MWD (m)
for HF hospitalization or deatha
Fig. 2 | Evidence base for the treatment of HFpEF. a | Clinical trials of pharmacological therapies for heart failure with
preserved ejection fraction (HFpEF) have not yet identified an effective medical therapy, but the strategy of decongestion
to treat elevated pulmonary artery pressures, tested in the CHAMPION trial92,93, has been shown to reduce the risk of
hospitalization for heart failure (HF). b | Clinical trials of medical therapies to improve peak exercise capacity (as assessed by
peak oxygen consumption (VO2)) and submaximal exercise capacity (assessed by 6-min walking distance (6-MWD)) have also
not demonstrated a favourable effect of these approaches. By contrast, lifestyle interventions, including exercise training
and weight loss induced by caloric restriction, improved functional capacity in a single-centre study. aThe CHAMPION
trial and CHAMPION-PEF ancillary study evaluated the effects on the single end point of hospitalization for HF, and not the
combined end point of hospitalization for HF and mortality as in the other trials of pharmacological therapies for HFpEF.

Nature Reviews | Cardiology

Reviews

Table 4 | Major randomized clinical trials of pharmacological treatments for HFpEF

Study (publication year) drug or Number Main entry Primary end point Finding Notes refs
strategy of patients criteria
Angiotensin-converting enzyme inhibitor
PEP-CHF (2006) Perindopril 850 EF >40% plus All-cause death or Neutral Improved 6-MWD, but high 76

SHD HF hospitalization treatment discontinuation

rates
Angiotensin receptor blocker
CHARM-Preserved (2003) Candesartan 3,023 EF >40% CV-related death or Neutral Greater benefit with lower EF, 77

HF hospitalization ↓ HF hospitalization
I-PRESERVE (2008) Irbesartan 4,128 EF ≥45% All-cause death or Neutral Potential benefit in patients 78

CV hospitalization with low BNP levels

Angiotensin receptor–neprilysin inhibitor
PARAMOUNT (2012) Sacubitril– 301 EF ≥45%, ↑ BNP Reduction in Positive ↓ LA volume 79

valsartan NT-proBNP levels

PARAGON (2019) Sacubitril– 4,822 EF ≥45%, ↑ BNP CV-related death or Neutral Possible benefit in women 23

valsartan plus SHD HF hospitalization and patients with EF <57%

β-Adrenergic receptor antagonist
J-DHF (2013) Carvedilol 245 EF >40% CV-related death or Neutral Open-label design, low 80

HF hospitalization event rate

ELANDD (2012) Nebivolol 116 EF >45% plus 6-MWD Neutral ↓ HR correlated with 81

LVDD ↓ exercise capacity

Mineralocorticoid receptor antagonist
TOPCAT (2014) Spironolactone 3,445 EF ≥45% and HF CV-related death, Neutral Geographical variation 82

hospitalization aborted SCD or HF

or ↑ BNP levels hospitalization
Aldo-DHF (2013) Spironolactone 422 EF ≥50% plus Peak VO2 and E/e′ Neutral Favorable reduction in E/e′, 83

LVDD ↓ NT-proBNP levels,

↓ LV mass, no effect on QOL
Nitric oxide and cGMP-stimulating therapies
RELAX (2013) Sildenafil 216 EF ≥50% plus Peak VO2 Neutral Trend for ↓ GFR, no change 84

objective in QOL or 6-MWD

diagnosis of HFa
NEAT (2015) Isosorbide 110 EF ≥50% plus Daily physical Neutral Trend for ↓ physical activity, 85

mononitrate objective activity level no change in QOL or 6-MWD

diagnosis of HFa
INDIE-HFpEF (2018) Inorganic 105 EF ≥50% plus Peak VO2 Neutral No effect on physical activity 86

nitrite objective level, QOL or E/e′ ratio

diagnosis of HFa
SOCRATES-PRESERVED Vericiguat 477 EF ≥45% and NT-proBNP levels Neutral ↑ QOL 87,88

(2017) ↑ BNP levels and LA volume

Others
DIG-PEF (2006) Digoxin 988 EF >45% HF-related death or Neutral Trend for 89

HF hospitalization ↓ HF hospitalizations but

↑ unstable angina incidence
EDIFY (2017) Ivabradine 179 EF ≥45% plus E/e′ ratio, 6-MWD Neutral No benefit despite 90

↑ BNP levels plus and NT-proBNP ↓ HR by 7.7 bpm

↑ HR levels
PANACHE (2019) Neladenoson 305 EF ≥45% plus 6-MWD Neutral No effect on activity, QOL 91

↑ BNP levels plus or NT-proBNP levels

SHD
CHAMPION (2014) PA pressure- 119 EF ≥40% plus HF HF hospitalization Positive Similar benefit in HFpEF 92,93

guided therapy hospitalization and HFrEF

6-MWD, 6-min walking distance; BNP, B-type natriuretic peptide; CV, cardiovascular; EF, ejection fraction; GFR , glomerular filtration rate; HF, heart failure;
HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; HR , heart rate; LA , left atrial; LV, left ventricular; LVDD,
left ventricular diastolic dysfunction; NT-proBNP, N-terminal pro B-type natriuretic peptide; PA , pulmonary artery; QOL , quality of life; SCD, sudden cardiac
death; SHD, structural heart disease; VO2, oxygen consumption rate. aDefined by any of the following: hospitalization for decompensated HF in the past year
but not within the past 30 days, invasively documented elevated pulmonary capillary wedge pressure at rest (>15 mmHg) or with exercise (≥25 mmHg) at
catheterization performed in the evaluation of dyspnoea, elevated NT-proBNP (>400 pg/ml) or BNP (>200 pg/ml) levels at enrolment, and echocardiographic
evidence of HFpEF, defined as ≥2 of the following: E/e′ ratio ≥15, moderate LA enlargement and pulmonary artery systolic pressure >35 mmHg.

www.nature.com/nrcardio
 Reviews

Pharmacological therapy. Diuretics are the first-line the metabolite was undetectable in a higher percentage
treatment for patients with HFpEF and congestion. of Russian patients who were assigned to and reported
Although diuretics have not been directly studied in taking spironolactone than in those from the USA and
a controlled trial in patients with HFpEF, indirect evi- Canada (30% versus 3%)98. These analyses suggest that
dence supports the efficacy of diuretics in this patient the patients enrolled in Russia and Georgia did not have
cohort. In the CHAMPION trial92, patients with HF typical HFpEF and were also less likely to adhere to the
were randomly assigned to treatment prescribed with assigned medication.
or without knowledge of pulmonary artery pressures When restricting the TOPCAT analysis to patients
measured with an implantable haemodynamic moni- enrolled in the Americas, in whom the confidence in
tor. Higher pulmonary artery pressures over short peri- the diagnosis and compliance with study procedures was
ods of time reflect increases in downstream left heart greater than with the Russian and Georgian cohort, the
filling pressures, signalling the need for a reduction in risks of the composite primary outcome (HR 0.82, 95%
these pressures. Over a mean follow-up of 15 months, CI 0.69–0.98), cardiovascular death (HR 0.74, 95% CI
patients assigned to pressure-informed therapy had a 0.57–0.97) and hospitalization for HF (HR 0.82, 95%
39% reduction in hospitalizations for HF92, a reduction CI 0.67–0.99) were each reduced significantly with
that was also observed when the analysis was restricted spironolactone treatment compared with placebo97.
to patients with HFpEF93 (Fig. 2). Diuretics were the most A subsequent analysis also restricted to patients enrolled
commonly titrated medication in the group allocated to in the Americas showed that spironolactone treatment
pulmonary artery pressure-guided therapy, providing significantly reduced the clinical signs of congestion
evidence to support the efficacy of diuretics in reducing compared with placebo and decongestion was inde-
the risk of hospitalization for HF. pendently associated with improved quality of life and
To date, the most substantial evidence supporting clinical outcomes99. The benefits from spironolactone
pharmacological therapies in HFpEF is derived from treatment have also been shown to be greater in patients
trials evaluating the mineralocorticoid antagonist with a lower ejection fraction, particularly <50%21.
spironolactone and the dual angiotensin–neprilysin Guidelines from the ACC, AHA and HFSA now rec-
inhibitor sacubitril–valsartan23,82. Investigators of the ommend that patients with HF and an ejection fraction
TOPCAT trial82 randomly assigned 3,445 patients with ≥45% (who meet the TOPCAT entry criteria) be treated
symptomatic HFpEF (ejection fraction ≥45%) to receive with a mineralocorticoid receptor antagonist (class IIb
either spironolactone or placebo. The composite pri- indication)24.
mary outcome was death from cardiovascular causes, The 2019 PARAGON trial23 compared the clinical
aborted cardiac arrest or hospitalization for HF. Over a outcomes of treatment with sacubitril–valsartan or
mean follow-up of 3.3 years, spironolactone treatment with valsartan only in 4,822 patients with symptomatic
did not reduce the rate of the primary outcome com- HFpEF (ejection fraction ≥45%), structural heart dis-
pared with placebo (18.6% versus 20.4%, HR 0.89, 95% ease and elevated natriuretic peptide levels. At a median
CI 0.77–1.04)82. Hospitalization for HF occurred less fre- follow-up of 35 months, sacubitril–valsartan treatment
quently in the spironolactone group than in the placebo did not significantly reduce the frequency of the primary
group (12.0% versus 14.2%, HR 0.83, 95% CI 0.69–0.99), composite outcome of hospitalizations for HF and death
but the other components of the primary outcomes as from cardiovascular causes compared with treatment
well as all cause death and hospitalizations for any cause with valsartan only (894 primary events in 526 patients
occurred at similar rates in both groups. A subgroup versus 1,009 primary events in 557 patients; rate ratio
analysis showed a significant reduction in the rate of the (RR) 0.87, 95% CI 0.75–1.01, P = 0.06). A small, bor-
primary outcome with spironolactone compared with derline significant reduction with sacubitril–valsartan
placebo among patients who were enrolled according to treatment was observed in the rates of hospitalization
elevated natriuretic peptide levels (15.9% versus 23.6%, for HF (RR 0.85, 95% CI 0.72–1.00). All-cause mortality
HR 0.65, 95% CI 0.49–0.87)82. However, a subsequent was similar in the two groups23.
post hoc analysis reported that patients with lower An improvement in NYHA class was slightly more
natriuretic peptide levels paradoxically responded more frequent in the sacubitril–valsartan group than the
favourably to spironolactone than patients with high valsartan-only group (15.0% versus 12.6%; odds ratio
natriuretic peptide levels96. 1.45, 95% CI 1.13–1.86)23. Sacubitril–valsartan treat-
Approximately half of the TOPCAT cohort was ment improved quality of life but the effect size was
enrolled in North America and South America, and small (between-group difference of 1.0 point on the
the other half in the Republic of Georgia and Russia. Kansas City Cardiomyopathy Questionnaire, with a
In a post hoc analysis, Pfeffer et al. reported several geo- minimal clinically significant difference considered to
graphical differences that confounded the interpretation be 5 points). Patients in the sacubitril–valsartan group
of the TOPCAT results97. Patients enrolled in Georgia had higher rates of hypotension and angioedema, and
and Russia had four-fold lower event rates than their lower rates of hyperkalaemia and adverse renal outcomes
American counterparts and were less likely to display compared with the valsartan-only group. Furthermore,
typical effects of spironolactone treatment, including systolic blood pressure was 4.5 mmHg lower (95% CI
hyperkalaemia, increased creatinine levels in blood or 3.6–5.4 mmHg) in the sacubitril–valsartan group than in
reduced blood pressure levels. A sub-study measur- the valsartan-only group23. Prespecified subgroup anal-
ing plasma levels of canrenone (an active metabolite yses suggested a possible benefit of sacubitril–valsartan
of spironolactone) at the 12-month visit reported that treatment among patients with an ejection fraction below

Nature Reviews | Cardiology

Reviews

the median (≤57%), similar to the ejection fraction-based revascularized35. However, this observation has not
analyses from TOPCAT21, as well as in women. Given been confirmed in a prospective study. AF is also very
that the primary end point was not met, each of the sec- common in patients with HFpEF and is associated with
ondary end points and subgroup analyses must be con- a depressed exercise capacity, greater right-sided HF and
sidered as exploratory23. Finally, a subsequent ancillary increased mortality105–107. Catheter ablation has emerged
analysis revealed that the risk of total HF hospitalizations as an effective means to restore sinus rhythm in patients
and cardiovascular death was inversely and non-linearly with AF108 but has not been well-studied in patients with
associated with the timing from previous HF hospitaliza- HFpEF and AF. Prospective trials testing the efficacy of
tions100. Compared with valsartan-only treatment, abso- ablation are urgently required in this patient cohort, par-
lute risk reductions with sacubitril–valsartan treatment ticularly given the known causal relationship between
were more pronounced in patients enrolled early after ablation and left atrial dysfunction109, which is known
hospitalization (≤30 days) than in patients who were to already be present in HFpEF37,110–113.
enrolled later after hospitalization or never hospitalized.
This finding emphasizes how patient characteristics, such Therapies targeting cardiometabolic risk. Exercise train-
as the degree of congestion, might influence responses ing has been shown to improve exercise capacity and
to different treatments and that important differences quality of life in patients with HFpEF8,94,95,114,115. The
might exist between congested or hospitalized patients benefits of exercise training seem to be mediated by
and patients with compensated HF who only develop salutary effects outside of the heart94,116, although one
abnormal haemodynamics during exercise. Regulatory study did observe improvements in cardiac function
bodies have not yet made decisions on the labelling of after training115. Sodium restriction is generally advised
sacubitril–valsartan treatment in HFpEF. and has been associated with improvements in ventricu-
Given the current evidence base, pharmacological lar and vascular stiffening in patients with HFpEF117.
treatment of HFpEF includes a diuretic for patients Obesity is highly prevalent among patients with HFpEF
with congestion and addition of a mineralocorticoid (60–75%) and is associated with more adverse haemo-
receptor antagonist for patients who are acceptable can- dynamic measures, poorer right ventricular function,
didates19,24. Comorbid conditions are very common in increased volume retention, greater ventricular inter-
patients with HFpEF and require careful treatment101. action, reduced exercise capacity and efficiency, greater
For example, angiotensin-converting enzyme inhibitors symptom severity, increased vulnerability to kidney dys-
and angiotensin-receptor blockers are frequently pre- function, and poorer quality of life46,118–121. In patients
scribed for hypertension, particularly when kidney dis- without HF, weight loss has been reported to reduce car-
ease is present. The ACC/AHA guidelines recommend diac filling pressures as well as blood pressure levels and
the treatment of hypertension in HFpEF to a target sys- heart rate122. In a single-centre trial enrolling patients
tolic blood pressure of <130 mmHg (ref.24) but, impor- with obesity and HFpEF (mean BMI 39.3 kg/m2, among
tantly, the observation that patients with HFpEF show an whom 95% had hypertension and 35% had diabetes),
exaggerated hypotensive response to vasodilators com- Kitzman et al. found that a modest weight loss induced
pared with patients with HFrEF must be considered102. by caloric restriction reduced body mass by 6.6%, which
In addition, observational data suggest that patients with resulted in significant improvements in exercise capacity
HFpEF and systolic blood pressure <120 mmHg have and quality of life95 (Fig. 2).
an increased risk of death and hospitalization for HF103. In addition to weight loss and exercise training,
Furthermore, observational studies have suggested a research has also focused on targeting cardiometabolic
possible benefit of statin use in HFpEF104, and patients risk with the use of pharmacotherapy in patients with
with standard indications for statin therapy, such as cor- HFpEF. Inhibitors of the sodium–glucose transport
onary artery disease, should be treated with this class protein 2 (SGLT2) have been shown to reduce the
of drug19,24. The management of AF in HFpEF involves rate of incident hospitalization for HF by 31% among
the use of anticoagulants and consideration of rate ver- patients with diabetes123. In the DAPA-HF trial124, treat-
sus rhythm control strategies, similar to that of non-HF ment with the SGLT2 inhibitor dapagliflozin was shown
populations19,24. Organic nitrates, inorganic nitrite and to reduce the primary outcome of hospitalization for HF
phosphodiesterase 5 inhibitors have not been shown to or cardiovascular death in patients with HFrEF by 26%
be beneficial for patients with HFpEF in randomized compared with placebo, a benefit that was similar in
clinical trials84–86 (Table 4), and as such are considered patients with or without diabetes. A number of ongo-
to be contraindicated in these patients according to the ing trials are evaluating whether SGLT2 inhibitors can
ACC/AHA/HFSA guidelines24. improve clinical outcomes in patients with HFpEF125.

Non-pharmacological therapy. Coronary artery disease Device-based therapies. In contrast to HFrEF, no device-
is common among patients with HFpEF, observed in based therapy has been approved for the treatment of
approximately two-thirds of this cohort35,43. Myocardial HFpEF. On the basis of data from TOPCAT, sudden car-
ischaemia and injury are related to the severity of ven- diac death has been estimated to account for up to 20%
tricular dysfunction and haemodynamic derangements of deaths in HFpEF126, but no trial has yet evaluated the
that are observed in patients with HFpEF25. One ret- potential role of implantable defibrillators for patients
rospective observational study reported that survival with HFpEF. Chronotropic incompetence is common in
was greater in patients with HFpEF undergoing com- HFpEF and the strategy of rate-adaptive atrial pacing is
plete revascularization than in those who were not currently being tested in the RAPID-HF trial127,128.

www.nature.com/nrcardio
 Reviews

In addition to modulation of heart rate and rhythm, pulmonary and muscle structure or function155–157,
other trials are currently evaluating the capacity of comorbid conditions such as obesity46, presence of cor-
device-based approaches to reduce cardiac filling onary disease35 and diabetes158, or unbiased clustering
pressures. A percutaneously implanted intra-atrial analyses incorporating data from multiple sources159–161.
septostomy device has been shown, in uncontrolled Some of the apparent differences in phenotypes might
and sham-controlled trials, to improve symptoms, be related to the relative timing in the natural history
functional capacity and exercise haemodynamics in of disease when patients are observed162. For example,
patients with HFpEF129–132. The device works by creat- patients with advanced HFpEF tend to have more RV
ing a small left-to-right shunt (shunt fraction of 25%) remodelling and dysfunction than patients with less
that results in a reduction in left atrial pressure. The advanced disease105. Many of these potential phenotyp-
shunt also increases pulmonary blood flow, which ing approaches overlap considerably and some might
might enhance pulmonary vascular recruitment and also even conflict with one another (Fig. 3). For example,
distention, and increases the oxygen content of venous two different machine learning-based approaches using
blood to reduce pulmonary hypoxic vasoconstriction133. the same technology have revealed different clustering
These changes have been associated with salutary effects of HFpEF phenotypes159,161. How can one approach be
on pulmonary vascular resistance and compliance deemed superior to the other? The ultimate value of any
in patients with HFpEF after septostomy but, impor- of the potential schemes for phenotyping will reside it
tantly, patients with marked pre-existing pulmonary its simplicity, feasibility and capacity to predict distinct
vascular disease were excluded from these studies133. responses to treatment.
Although RV dysfunction has not been observed in the Phenotypes in HFpEF that are based on pathophys-
patients with HFpEF studied to date, an increase in RV iology might hold the most promise as they are easy to
volume has been observed after septostomy132, which conceptualize, and treatments will target the abnor-
might augment pericardial constraint in some patients malities on which the phenotype is based. Examples
with HFpEF134. Atrial septostomy is being tested in two of these phenotypes include patients with HFpEF and
ongoing clinical trials in HFpEF135,136. pulmonary vascular disease147,148, systemic inflamma-
A major component of the increase in LV filling pres- tion33 or coronary microvascular dysfunction25,41–43.
sure in normal and diseased hearts is caused by extrin- Alternative approaches include phenotyping on the
sic restraint mediated by the interventricular septum basis of comorbid conditions such as obesity, coro-
and the pericardium134. A novel device-based therapy nary disease or metabolic syndrome35,46,163. These phe-
has been developed to open the anterior pericardium notypes are easy to identify and have established and
through a minimally invasive subxiphoid approach137. novel treatments that can be tested. One important
In animal and human experiments, this device substan- difference in phenotypes might exist between patients
tially mitigated the rise in LV filling pressures during with congestion that is apparent at rest and those with
volume loading137,138, and is currently being tested in a exercise-induced haemodynamic abnormalities9. The
single-centre study involving patients with HFpEF139. latter group might require specific treatments targeting
abnormalities that develop preferentially during stress,
Emerging role for phenotyping. To date, the failure either pharmacologically164–166 or through device-based
of clinical trials to identify an effective treatment for approaches129–134,137,138. Other therapies, including those
HFpEF (Fig. 2; Table 4) has been attributed, at least targeting neurohormonal activation, might be more
in part, to heterogeneity in the clinical syndrome of effective for patients with congestion that is apparent
HFpEF. Some patients might respond favourably to at rest100. Phenotypes that are based on sophisticated
certain therapies whereas others might be harmed by machine-learning techniques also hold promise159–161,
the same intervention, nullifying the ability to identify but might be more difficult to understand and apply,
a treatment effect when all patients are combined and and do not necessarily include common pathophysio-
treated together (Fig. 3). Therefore, greater efforts have logical features; nevertheless, new mechanisms might be
been made in trials to avoid enrolling patients with cer- discovered with the use of these approaches.
tain specific aetiologies that cause the clinical syndrome At present, only two examples of successful pheno-
of HFpEF but carry their own distinct pathophysiology, typing in HFpEF according to this approach have been
natural history and treatment (Table 2). reported. One approach is to target patients with HFpEF
Even after the aetiologies listed in Table 2 are and obesity given that caloric restriction or aerobic
excluded, substantial heterogeneity among the broader exercise training was shown to increase peak oxygen
clinical syndrome of HFpEF remains (Fig. 3), which consumption in these patients95. The second example
has led to calls for increased phenotyping of patients is in patients with transthyretin cardiac amyloidosis167.
into more mechanistically homogeneous groups13,38. Although amyloidosis is a separate entity that should
However, at present, no consensus exists on how not be considered as HFpEF (Table 2), these two diseases
this phenotyping should be conducted. Proposed were formerly grouped together. Myocardial dysfunction
approaches include profiling on the basis of plasma bio- in transthyretin amyloidosis is caused by deposition
markers25,140–145, rest or exercise haemodynamic signa- of unstable, misfolded amyloid fibrils. Maurer et al.
tures28,146–148, the degree of LV or RV dysfunction105,149–151, demonstrated, in a multicentre, placebo-controlled trial,
left atrial abnormalities37,110–113, coexisting valvular heart that treatment with tafamidis, a drug that binds to tran-
disease36,37, coronary microvascular dysfunction41–43, sthyretin to prevent tetramer dissociation, in patients
systemic inflammation33,152,153, exercise responses27,154, with transthyretin cardiac amyloidosis was associated

Nature Reviews | Cardiology

Reviews

a HFpEF

b
Comorbidities
Hypertension
Obesity
Coronary microvascular and macrovascular disease
Diabetes mellitus and metabolic syndrome

Cardiac and pulmonary vascular function

LV dysfunction only
LV and LA dysfunction and/or atrial fibrillation
Pulmonary vascular dysfunction
RV dysfunction

Haemodynamics
↑ LVFP with exercise only
↓ Pulmonary vasodilatation with exercise
↑ LVFP at rest with pulmonary hypertension
↑ RVFP and LVFP at rest

Extra-cardiac structure and function

↑ Arterial stiffness
Endothelial and coronary microvascular dysfunction
Sarcopenia and mitochondrial dysfunction
Tissue fibrosis

Biomarkers
Normal natriuretic peptide levels
Pro-inflammatory markers
Cardiac injury markers
Fibrotic markers

Increasing disease severity and risk of adverse outcomes

Fig. 3 | Phenotyping in patients with HFpEF. a | Most clinical trials published to date have broadly enrolled all patients with
heart failure with preserved ejection fraction (HFpEF), without regard to specific phenotypes, and have generated largely
neutral results. b | Classifying patients with HFpEF into subgroups of pathophysiologically homogeneous phenotypes might
allow for more precise and targeted therapies. Several strategies have been proposed to classify patients with different
characteristics, including the presence of comorbid conditions, cardiac functional status, invasive haemodynamics
data, presence of extra-cardiac abnormalities and plasma biomarker levels. The size of each bar is intended to roughly
approximate the relative prevalence of each phenotype in the broad population of ‘garden variety’ HFpEF. Disease severity
and risk of adverse outcomes increase on moving from left to right. As illustrated, each of these domains overlaps with one
another, both within a specific grouping scheme and between different potential phenotyping schemes. LA, left atrial; LV, left
ventricular; LVFP, left ventricular filling pressure; RV, right ventricular; RVFP, right ventricular filling pressure.

with a reduction in all-cause mortality (HR 0.70, 95% treatment is available, the approach to separate patients
CI 0.51–0.96) and a lower rate of cardiovascular-related with cardiac amyloidosis from those with HFpEF is well
hospitalizations (RR 0.68, 95% CI 0.56–0.81) compared supported. Through more rigorous characterization of
with placebo167. This finding clearly establishes tran- specific HFpEF phenotypes, and together with clinical
sthyretin cardiac amyloidosis as a separate disorder trials testing therapeutic approaches targeted to the
that is distinct from HFpEF (Table 2), an important specific pathophysiology identified, we will hopefully
distinction because 5–13% of patients originally con- be able to define other groups of patients who should
sidered to have typical HFpEF had evidence of cardiac also be separated from typical HFpEF, recapitulat-
amyloidosis when evaluated with nuclear scintigraphy ing the successes that have been observed for cardiac
or at post-mortem examination168,169. Now that a specific amyloidosis167.

www.nature.com/nrcardio
 Reviews

Conclusions mineralocorticoid receptor antagonists, management

HFpEF has grown to become the dominant form of HF of comorbidities and promotion of a healthy active
worldwide and continues to present a diagnostic and lifestyle, with prescription of exercise training where
therapeutic challenge for clinicians. Diagnosis relies feasible. Although most clinical trials of therapy for
on the demonstration of objective evidence of cardiac HFpEF have had neutral results to date, future efforts
dysfunction, on the basis of either physical examina- to more rigorously characterize and group patients into
tion, imaging data, blood testing results or echocardi- discrete phenotypes hold great promise to allow the
ography data. When these tests are inconclusive, invasive individualization of therapy to improve outcomes.
exercise testing is necessary. Treatment of HFpEF is
Published online xx xx xxxx
aimed at control of congestion and involves the use of

1. Pfeffer, M. A., Shah, A. M. & Borlaug, B. A. Heart 20. Lund, L. H. et al. Heart failure with mid-range ejection 38. Obokata, M., Reddy, Y. N. V. & Borlaug, B. A.
failure with preserved ejection fraction in perspective. fraction in CHARM: characteristics, outcomes and Diastolic dysfunction and heart failure with preserved
Circ. Res. 124, 1598–1617 (2019). effect of candesartan across the entire ejection ejection fraction: understanding mechanisms by using
2. Shah, A. M. et al. Heart failure stages among older fraction spectrum. Eur. J. Heart Fail. 20, 1230–1239 noninvasive methods. JACC Cardiovasc. Imaging 13,
adults in the community: the Atherosclerosis Risk (2018). 245–257 (2019).
in Communities Study. Circulation 135, 224–240 21. Solomon, S. D. et al. Influence of ejection fraction on 39. Davie, A. P., Francis, C. M., Caruana, L.,
(2017). outcomes and efficacy of spironolactone in patients Sutherland, G. R. & McMurray, J. J. Assessing
3. Dunlay, S. M., Roger, V. L. & Redfield, M. M. with heart failure and preserved ejection fraction. diagnosis in heart failure: which features are any use?
Epidemiology of heart failure with preserved ejection Eur. Heart J. 37, 455–462 (2016). QJM 90, 335–339 (1997).
fraction. Nat. Rev. Cardiol. 14, 591–602 (2017). 22. Butler, J., Anker, S. D. & Packer, M. Redefining heart 40. Mentz, R. J., Broderick, S., Shaw, L. K., Fiuzat, M. &
4. Sidney, S. et al. Association between aging of the failure with a reduced ejection fraction. JAMA 322, O’Connor, C. M. Heart failure with preserved ejection
US population and heart disease mortality from 2011 1761–1762 (2019). fraction: comparison of patients with and without
to 2017. JAMA Cardiol. 4, 1280–1286 (2019). 23. Solomon, S. D. et al. Angiotensin-neprilysin inhibition angina pectoris (from the Duke Databank for
5. Flegal, K. M., Kruszon-Moran, D., Carroll, M. D., in heart failure with preserved ejection fraction. Cardiovascular Disease). J. Am. Coll. Cardiol. 63,
Fryar, C. D. & Ogden, C. L. Trends in obesity among N. Engl. J. Med. 381, 1609–1620 (2019). 251–258 (2014).
adults in the United States, 2005 to 2014. JAMA 24. Yancy, C. W. et al. 2017 ACC/AHA/HFSA focused 41. Shah, S. J. et al. Prevalence and correlates of
315, 2284–2291 (2016). update of the 2013 ACCF/AHA guideline for the coronary microvascular dysfunction in heart failure
6. Borlaug, B. A. & Redfield, M. M. Diastolic and systolic management of heart failure: a report of the American with preserved ejection fraction: PROMIS-HFpEF.
heart failure are distinct phenotypes within the heart College of Cardiology/American Heart Association Eur. Heart J. 39, 3439–3450 (2018).
failure spectrum. Circulation 123, 2006–2013 task force on clinical practice guidelines and the 42. Yang, J. H. et al. Endothelium dependent and
(2011). Heart Failure Society of America. Circulation 136, independent coronary microvascular dysfunction in
7. Lupon, J. et al. Heart failure with preserved ejection e137–e161 (2017). patients with heart failure with preserved ejection
fraction infrequently evolves toward a reduced 25. Obokata, M. et al. Myocardial injury and cardiac fraction. Eur. J. Heart Fail. https://doi.org/10.1002/
phenotype in long-term survivors. Circ. Heart Fail. 12, reserve in patients with heart failure and preserved ejhf.1671 (2020).
e005652 (2019). ejection fraction. J. Am. Coll. Cardiol. 72, 29–40 43. Mohammed, S. F. et al. Coronary microvascular
8. Pandey, A. et al. Physical activity, fitness, and (2018). rarefaction and myocardial fibrosis in heart failure
obesity in heart failure with preserved ejection 26. Borlaug, B. A. et al. Global cardiovascular reserve with preserved ejection fraction. Circulation 131,
fraction. JACC Heart Fail. 6, 975–982 (2018). dysfunction in heart failure with preserved ejection 550–559 (2015).
9. Borlaug, B. A., Nishimura, R. A., Sorajja, P., Lam, C. S. fraction. J. Am. Coll. Cardiol. 56, 845–854 (2010). 44. Gorter, T. M. et al. Right heart dysfunction and failure
& Redfield, M. M. Exercise hemodynamics enhance 27. Borlaug, B. A., Kane, G. C., Melenovsky, V. & in heart failure with preserved ejection fraction:
diagnosis of early heart failure with preserved ejection Olson, T. P. Abnormal right ventricular-pulmonary mechanisms and management. Position statement
fraction. Circ. Heart Fail. 3, 588–595 (2010). artery coupling with exercise in heart failure with on behalf of the Heart Failure Association of the
10. Obokata, M. et al. Role of diastolic stress testing in preserved ejection fraction. Eur. Heart J. 37, European Society of Cardiology. Eur. J. Heart Fail. 20,
the evaluation for heart failure with preserved ejection 3293–3302 (2016). 16–37 (2018).
fraction: a simultaneous invasive-echocardiographic 28. Reddy, Y. N. V. et al. The haemodynamic basis 45. Reddy, Y. N. V., Obokata, M., Gersh, B. J.
study. Circulation 135, 825–838 (2017). of lung congestion during exercise in heart failure & Borlaug, B. A. High prevalence of occult heart
11. Reddy, Y. N. V., Carter, R. E., Obokata, M., with preserved ejection fraction. Eur Heart J. 40, failure with preserved ejection fraction among patients
Redfield, M. M. & Borlaug, B. A. A simple, 3721–3730 (2019). with atrial fibrillation and dyspnea. Circulation 137,
evidence-based approach to help guide diagnosis of 29. Obokata, M. et al. Hemodynamics, dyspnea, and 534–535 (2018).
heart failure with preserved ejection fraction. pulmonary reserve in heart failure with preserved 46. Obokata, M., Reddy, Y. N., Pislaru, S. V., Melenovsky, V.
Circulation 138, 861–870 (2018). ejection fraction. Eur. Heart J. 39, 2810–2821 & Borlaug, B. A. Evidence supporting the existence of
12. Pieske, B. et al. How to diagnose heart failure with (2018). a distinct obese phenotype of heart failure with
preserved ejection fraction: the HFA-PEFF diagnostic 30. Reddy, Y. N. V., Olson, T. P., Obokata, M., Melenovsky, V. preserved ejection fraction. Circulation 136, 6–19
algorithm: a consensus recommendation from the & Borlaug, B. A. Hemodynamic correlates and (2017).
Heart Failure Association (HFA) of the European diagnostic role of cardiopulmonary exercise testing 47. Drazner, M. H. et al. The relationship of right- and
Society of Cardiology (ESC). Eur. Heart J. 40, in heart failure with preserved ejection fraction. left-sided filling pressures in patients with heart failure
3297–3317 (2019). JACC Heart Fail. 6, 665–675 (2018). and a preserved ejection fraction. Circ. Heart Fail. 3,
13. Shah, S. J. et al. Phenotype-specific treatment of heart 31. Eisman, A. S. et al. Pulmonary capillary wedge 202–206 (2010).
failure with preserved ejection fraction: a multiorgan pressure patterns during exercise predict exercise 48. From, A. M. et al. Bedside assessment of cardiac
roadmap. Circulation 134, 73–90 (2016). capacity and incident heart failure. Circ. Heart Fail. hemodynamics: the impact of noninvasive testing and
14. Klainer, L. M., Gibson, T. C. & White, K. L. The 11, e004750 (2018). examiner experience. Am. J. Med. 124, 1051–1057
epidemiology of cardiac failure. J. Chronic Dis. 18, 32. Dorfs, S. et al. Pulmonary capillary wedge pressure (2011).
797–814 (1965). during exercise and long-term mortality in patients 49. Stevenson, L. W. & Perloff, J. K. The limited reliability
15. McKee, P. A., Castelli, W. P., McNamara, P. M. & with suspected heart failure with preserved ejection of physical signs for estimating hemodynamics in
Kannel, W. B. The natural history of congestive heart fraction. Eur. Heart J. 35, 3103–3112 (2014). chronic heart failure. JAMA 261, 884–888 (1989).
failure: the Framingham study. N. Engl. J. Med. 285, 33. Paulus, W. J. & Tschope, C. A novel paradigm for heart 50. Lam, C. S. et al. Pulmonary hypertension in
1441–1446 (1971). failure with preserved ejection fraction: comorbidities heart failure with preserved ejection fraction:
16. Ho, J. E. et al. Differential clinical profiles, exercise drive myocardial dysfunction and remodeling through a community-based study. J. Am. Coll. Cardiol. 53,
responses and outcomes associated with existing coronary microvascular endothelial inflammation. 1119–1126 (2009).
HFpEF definitions. Circulation 140, 353–365 (2019). J. Am. Coll. Cardiol. 62, 263–271 (2013). 51. Iwanaga, Y. et al. B-type natriuretic peptide strongly
17. Abudiab, M. M. et al. Cardiac output response to 34. Luscher, T. F. Lumpers and splitters: the bumpy road reflects diastolic wall stress in patients with chronic
exercise in relation to metabolic demand in heart failure to precision medicine. Eur. Heart J. 40, 3292–3296 heart failure: comparison between systolic and
with preserved ejection fraction. Eur. J. Heart Fail. 15, (2019). diastolic heart failure. J. Am. Coll. Cardiol. 47,
776–785 (2013). 35. Hwang, S. J., Melenovsky, V. & Borlaug, B. A. 742–748 (2006).
18. Denolin, H., Kuhn, H., Krayenbuehl, H. P., Loogen, F. Implications of coronary artery disease in heart failure 52. Mueller, C. et al. Heart failure association of the
& Reale, A. The definition of heart failure. Eur. Heart J. with preserved ejection fraction. J. Am. Coll. Cardiol. European Society of Cardiology practical guidance
4, 445–448 (1983). 63, 2817–2827 (2014). on the use of natriuretic peptide concentrations.
19. Ponikowski, P. et al. 2016 ESC Guidelines for the 36. Eleid, M. F., Nishimura, R. A., Sorajja, P. & Eur. J. Heart Fail. 21, 715–731 (2019).
diagnosis and treatment of acute and chronic heart Borlaug, B. A. Systemic hypertension in low-gradient 53. Cleland, J. G., Taylor, J. & Tendera, M. Prognosis in
failure: the task force for the diagnosis and treatment severe aortic stenosis with preserved ejection fraction. heart failure with a normal ejection fraction. N. Engl.
of acute and chronic heart failure of the European Circulation 128, 1349–1353 (2013). J. Med. 357, 829–830 (2007).
Society of Cardiology (ESC). Developed with the 37. Tamargo, M. et al. Functional mitral regurgitation and 54. van Veldhuisen, D. J. et al. B-type natriuretic peptide
special contribution of the Heart Failure Association left atrial myopathy in heart failure with preserved and prognosis in heart failure patients with preserved
(HFA) of the ESC. Eur. Heart J. 37, 2129–2200 ejection fraction. Eur. J. Heart Fail. https://doi.org/ and reduced ejection fraction. J. Am. Coll. Cardiol. 61,
(2016). 10.1002/ejhf.1699 (2020). 1498–1506 (2013).

Nature Reviews | Cardiology

Reviews

55. Anjan, V. Y. et al. Prevalence, clinical phenotype, and ejection fraction: the CHARM-preserved trial. Lancet 99. Selvaraj, S. et al. Utility of the cardiovascular
outcomes associated with normal B-type natriuretic 362, 777–781 (2003). physical examination and impact of spironolactone
peptide levels in heart failure with preserved ejection 78. Massie, B. M. et al. Irbesartan in patients with in heart failure with preserved ejection fraction.
fraction. Am. J. Cardiol. 110, 870–876 (2012). heart failure and preserved ejection fraction. N. Engl. Circ. Heart Fail. 12, e006125 (2019).
56. Wang, T. J. et al. Impact of obesity on plasma natriuretic J. Med. 359, 2456–2467 (2008). 100. Vaduganathan, M. et al. Prior heart failure
peptide levels. Circulation 109, 594–600 (2004). 79. Solomon, S. D. et al. The angiotensin receptor hospitalization, clinical outcomes, and response to
57. Andersen, M. J. & Borlaug, B. A. Invasive hemodynamic neprilysin inhibitor LCZ696 in heart failure with sacubitril/valsartan compared with valsartan in HFpEF.
characterization of heart failure with preserved ejection preserved ejection fraction: a phase 2 double-blind J. Am. Coll. Cardiol. 75, 245–254 (2020).
fraction. Heart Fail. Clin. 10, 435–444 (2014). randomised controlled trial. Lancet 380, 1387–1395 101. Mentz, R. J. et al. Noncardiac comorbidities in heart
58. Borlaug, B. A. et al. Diastolic relaxation and compliance (2012). failure with reduced versus preserved ejection fraction.
reserve during dynamic exercise in heart failure with 80. Yamamoto, K., Origasa, H. & Hori, M. Effects of J. Am. Coll. Cardiol. 64, 2281–2293 (2014).
preserved ejection fraction. Heart 97, 964–969 (2011). carvedilol on heart failure with preserved ejection 102. Schwartzenberg, S. et al. Effects of vasodilation
59. Andersen, M. J., Olson, T. P., Melenovsky, V., Kane, G. C. fraction: the Japanese Diastolic Heart Failure Study in heart failure with preserved or reduced ejection
& Borlaug, B. A. Differential hemodynamic effects (J-DHF). Eur. J. Heart Fail. 15, 110–118 (2013). fraction implications of distinct pathophysiologies
of exercise and volume expansion in people with 81. Conraads, V. M. et al. Effects of the long-term on response to therapy. J. Am. Coll. Cardiol. 59,
and without heart failure. Circ. Heart Fail. 8, 41–48 administration of nebivolol on the clinical symptoms, 442–451 (2012).
(2015). exercise capacity, and left ventricular function of 103. Tsimploulis, A. et al. Systolic blood pressure and
60. Maron, B. A., Cockrill, B. A., Waxman, A. B. & patients with diastolic dysfunction: results of the outcomes in patients with heart failure with preserved
Systrom, D. M. The invasive cardiopulmonary ELANDD study. Eur. J. Heart Fail. 14, 219–225 ejection fraction. JAMA Cardiol. 3, 288–297 (2018).
exercise test. Circulation 127, 1157–1164 (2013). (2012). 104. Alehagen, U., Benson, L., Edner, M., Dahlstrom, U.
61. Maeder, M. T., Thompson, B. R., Brunner-La Rocca, H. P. 82. Pitt, B. et al. Spironolactone for heart failure with & Lund, L. H. Association between use of statins and
& Kaye, D. M. Hemodynamic basis of exercise limitation preserved ejection fraction. N. Engl. J. Med. 370, mortality in patients with heart failure and ejection
in patients with heart failure and normal ejection 1383–1392 (2014). fraction of ≥50. Circ. Heart Fail. 8, 862–870 (2015).
fraction. J. Am. Coll. Cardiol. 56, 855–863 (2010). 83. Edelmann, F. et al. Effect of spironolactone on diastolic 105. Obokata, M., Reddy, Y. N. V., Melenovsky, V., Pislaru, S.
62. Fujimoto, N. et al. Hemodynamic responses to rapid function and exercise capacity in patients with heart & Borlaug, B. A. Deterioration in right ventricular
saline loading: the impact of age, sex, and heart failure with preserved ejection fraction: the Aldo-DHF structure and function over time in patients with heart
failure. Circulation 127, 55–62 (2013). randomized controlled trial. JAMA 309, 781–791 failure and preserved ejection fraction. Eur. Heart J. 40,
63. Guazzi, M. et al. 2016 Focused update: clinical (2013). 689–697 (2019).
recommendations for cardiopulmonary exercise 84. Redfield, M. M. et al. Effect of phosphodiesterase-5 106. Zakeri, R. et al. Impact of atrial fibrillation on exercise
testing data assessment in specific patient populations. inhibition on exercise capacity and clinical status capacity in heart failure with preserved ejection
Circulation 133, e694–e711 (2016). in heart failure with preserved ejection fraction: fraction: a RELAX trial ancillary study. Circ. Heart Fail.
64. Lancellotti, P. et al. The clinical use of stress a randomized clinical trial. JAMA 309, 1268–1277 7, 123–130 (2014).
echocardiography in non-ischaemic heart disease: (2013). 107. Zakeri, R., Chamberlain, A. M., Roger, V. L. &
recommendations from the European association 85. Redfield, M. M. et al. Isosorbide mononitrate in Redfield, M. M. Temporal relationship and prognostic
of cardiovascular imaging and the american society of heart failure with preserved ejection fraction. N. Engl. significance of atrial fibrillation in heart failure patients
echocardiography. Eur. Heart J. Cardiovasc. Imaging J. Med. 373, 2314–2324 (2015). with preserved ejection fraction: a community-based
17, 1191–1229 (2016). 86. Borlaug, B. A. et al. Effect of inorganic nitrite vs study. Circulation 128, 1085–1093 (2013).
65. Holland, D. J., Prasad, S. B. & Marwick, T. H. placebo on exercise capacity among patients with 108. Packer, D. L. et al. Effect of catheter ablation vs
Contribution of exercise echocardiography to the heart failure with preserved ejection fraction: the antiarrhythmic drug therapy on mortality, stroke,
diagnosis of heart failure with preserved ejection INDIE-HFpEF randomized clinical trial. JAMA 320, bleeding, and cardiac arrest among patients with
fraction (HFpEF). Heart 96, 1024–1028 (2010). 1764–1773 (2018). atrial fibrillation: the CABANA randomized clinical
66. Burgess, M. I., Jenkins, C., Sharman, J. E. & 87. Pieske, B. et al. Vericiguat in patients with worsening trial. JAMA 321, 1261–1274 (2019).
Marwick, T. H. Diastolic stress echocardiography: chronic heart failure and preserved ejection fraction: 109. Packer, M. Effect of catheter ablation on pre-existing
hemodynamic validation and clinical significance of results of the SOluble guanylate Cyclase stimulatoR abnormalities of left atrial systolic, diastolic, and
estimation of ventricular filling pressure with exercise. in heArT failurE patientS with PRESERVED EF neurohormonal functions in patients with chronic
J. Am. Coll. Cardiol. 47, 1891–1900 (2006). (SOCRATES-PRESERVED) study. Eur. Heart J. 38, heart failure and atrial fibrillation. Eur. Heart J. 40,
67. Talreja, D. R., Nishimura, R. A. & Oh, J. K. Estimation 1119–1127 (2017). 1873–1879 (2019).
of left ventricular filling pressure with exercise by 88. Filippatos, G. et al. Patient-reported outcomes in 110. Melenovsky, V. et al. Left atrial remodeling and function
Doppler echocardiography in patients with normal the SOluble guanylate Cyclase stimulatoR in heArT in advanced heart failure with preserved or reduced
systolic function: a simultaneous echocardiographic- failurE patientS with PRESERVED ejection fraction ejection fraction. Circ. Heart Fail. 8, 295–303 (2015).
cardiac catheterization study. J. Am. Soc. Echocardiogr. (SOCRATES-PRESERVED) study. Eur. J. Heart Fail. 19, 111. Freed, B. H. et al. Prognostic utility and clinical
20, 477–479 (2007). 782–791 (2017). significance of cardiac mechanics in heart failure with
68. Obokata, M. & Borlaug, B. A. The strengths and 89. Ahmed, A. et al. Effects of digoxin on morbidity preserved ejection fraction: importance of left atrial
limitations of E/e’ in heart failure with preserved and mortality in diastolic heart failure: the ancillary strain. Circ. Cardiovasc. Imaging 9, e003754 (2016).
ejection fraction. Eur. J. Heart Fail. 20, 1312–1314 digitalis investigation group trial. Circulation 114, 112. Reddy, Y. N. V. et al. Left atrial strain and compliance
(2018). 397–403 (2006). in the diagnostic evaluation of heart failure with
69. Nauta, J. F. et al. Correlation with invasive left 90. Komajda, M. et al. Effect of ivabradine in patients preserved ejection fraction. Eur. J. Heart Fail. 21,
ventricular filling pressures and prognostic relevance with heart failure with preserved ejection fraction: 891–900 (2019).
of the echocardiographic diastolic parameters used in the EDIFY randomized placebo-controlled trial. 113. Telles, F. et al. Impaired left atrial strain predicts
the 2016 ESC heart failure guidelines and in the 2016 Eur. J. Heart Fail. 19, 1495–1503 (2017). abnormal exercise haemodynamics in heart failure
ASE/EACVI recommendations: a systematic review in 91. Shah, S. J. et al. Effect of neladenoson bialanate on with preserved ejection fraction. Eur. J. Heart Fail. 21,
patients with heart failure with preserved ejection exercise capacity among patients with heart failure 495–505 (2019).
fraction. Eur. J. Heart Fail. 20, 1303–1311 (2018). with preserved ejection fraction: a randomized clinical 114. Kitzman, D. W., Brubaker, P. H., Morgan, T. M.,
70. Santos, M. et al. E/e’ ratio in patients with unexplained trial. JAMA 321, 2101–2112 (2019). Stewart, K. P. & Little, W. C. Exercise training in older
dyspnea: lack of accuracy in estimating left ventricular 92. Abraham, W. T. et al. Wireless pulmonary artery patients with heart failure and preserved ejection
filling pressure. Circ. Heart Fail. 8, 749–756 (2015). haemodynamic monitoring in chronic heart failure: a fraction: a randomized, controlled, single-blind trial.
71. Sharifov, O. F. & Gupta, H. What is the evidence randomised controlled trial. Lancet 377, 658–666 Circ. Heart Fail. 3, 659–667 (2010).
that the tissue doppler index E/e’ reflects left (2011). 115. Edelmann, F. et al. Exercise training improves exercise
ventricular filling pressure changes after exercise or 93. Adamson, P. B. et al. Wireless pulmonary artery capacity and diastolic function in patients with heart
pharmacological intervention for evaluating diastolic pressure monitoring guides management to reduce failure with preserved ejection fraction: results of the
function? A systematic review. J. Am. Heart Assoc. 6, decompensation in heart failure with preserved Ex-DHF (Exercise training in Diastolic Heart Failure) pilot
e004766 (2017). ejection fraction. Circ. Heart Fail. 7, 935–944 (2014). study. J. Am. Coll. Cardiol. 58, 1780–1791 (2011).
72. Paulus, W. J. H2FPEF score. Circulation 138, 871–873 94. Pandey, A. et al. Exercise training in patients 116. Haykowsky, M. J. et al. Effect of endurance training on
(2018). with heart failure and preserved ejection fraction: the determinants of peak exercise oxygen consumption
73. Segar, M. W., Patel, K. V., Berry, J. D., Grodin, J. L. meta-analysis of randomized control trials. in elderly patients with stable compensated heart
& Pandey, A. Generalizability and implications of the Circ. Heart Fail. 8, 33–40 (2015). failure and preserved ejection fraction. J. Am. Coll.
H2FPEF score in a cohort of patients with heart failure 95. Kitzman, D. W. et al. Effect of caloric restriction or Cardiol. 60, 120–128 (2012).
with preserved ejection fraction. Circulation 139, aerobic exercise training on peak oxygen consumption 117. Hummel, S. L. et al. Low-sodium DASH diet improves
1851–1853 (2019). and quality of life in obese older patients with heart diastolic function and ventricular-arterial coupling in
74. Myhre, P. L. et al. Application of the H2 FPEF score to failure with preserved ejection fraction: a randomized hypertensive heart failure with preserved ejection
a global clinical trial of patients with heart failure clinical trial. JAMA 315, 36–46 (2016). fraction. Circ. Heart Fail. 6, 1165–1171 (2013).
with preserved ejection fraction: the TOPCAT trial. 96. Anand, I. S. et al. Interaction between spironolactone 118. Reddy, Y. N. V. et al. Characterization of the obese
Eur. J. Heart Fail. 21, 1288–1291 (2019). and natriuretic peptides in patients with heart phenotype of heart failure with preserved ejection
75. Sepehrvand, N. et al. External validation of the failure and preserved ejection fraction: from the fraction: a RELAX trial ancillary study. Mayo Clin.
H2F-PEF model in diagnosing patients with heart TOPCAT trial. JACC Heart Fail. 5, 241–252 (2017). Proc. 94, 1199–1209 (2019).
failure and preserved ejection fraction. Circulation 97. Pfeffer, M. A. et al. Regional variation in patients 119. Reddy, Y. N. V. et al. Adverse renal response to
139, 2377–2379 (2019). and outcomes in the Treatment of Preserved Cardiac decongestion in the obese phenotype of heart failure
76. Cleland, J. G. et al. The perindopril in elderly people Function Heart Failure with an Aldosterone Antagonist with preserved ejection fraction. J. Card. Fail. 26,
with chronic heart failure (PEP-CHF) study. Eur. Heart (TOPCAT) trial. Circulation 131, 34–42 (2015). 101–107 (2020).
J. 27, 2338–2345 (2006). 98. de Denus, S. et al. Spironolactone metabolites 120. Miller, W. L. & Borlaug, B. A. Impact of obesity on
77. Yusuf, S. et al. Effects of candesartan in patients with in TOPCAT - new insights into regional variation. volume status in patients with ambulatory chronic
chronic heart failure and preserved left-ventricular N. Engl. J. Med. 376, 1690–1692 (2017). heart failure. J. Card. Fail. 26, 112–117 (2020).

www.nature.com/nrcardio
 Reviews

121. Reddy, Y. N. V. et al. Quality of life in heart failure with 138. Borlaug, B. A. et al. Pericardiotomy enhances left are reduced in older patients with heart failure and
preserved ejection fraction: importance of obesity, ventricular diastolic reserve with volume loading in preserved ejection fraction and are related to exercise
functional capacity, and physical inactivity. Eur. J. humans. Circulation 138, 2295–2297 (2018). intolerance. JACC Heart Fail. 4, 636–645 (2016).
Heart Fail. https://doi.org/10.1002/ejhf.1788 (2020). 139. US National Library of Medicine. ClinicalTrials.gov 157. Rommel, K. P. et al. Extracellular volume fraction for
122. Reddy, Y. N. V. et al. Hemodynamic effects of weight https://clinicaltrials.gov/ct2/show/NCT03923673 characterization of patients with heart failure and
loss in obesity: a systematic review and meta-analysis. (2020). preserved ejection fraction. J. Am. Coll. Cardiol. 67,
JACC Heart Fail. 7, 678–687 (2019). 140. Tromp, J. et al. Identifying pathophysiological 1815–1825 (2016).
123. Zelniker, T. A. et al. SGLT2 inhibitors for primary and mechanisms in heart failure with reduced versus 158. Yap, J. et al. Association of diabetes mellitus on cardiac
secondary prevention of cardiovascular and renal preserved ejection fraction. J. Am. Coll. Cardiol. 72, remodeling, quality of life, and clinical outcomes in
outcomes in type 2 diabetes: a systematic review and 1081–1090 (2018). heart failure with reduced and preserved ejection
meta-analysis of cardiovascular outcome trials. Lancet 141. Obokata, M. et al. The neurohormonal basis of fraction. J. Am. Heart Assoc. 8, e013114 (2019).
393, 31–39 (2019). pulmonary hypertension in heart failure with preserved 159. Shah, S. J. et al. Phenomapping for novel classification
124. McMurray, J. J. V. et al. Dapagliflozin in patients with ejection fraction. Eur. Heart J. 40, 3707–3717 (2019). of heart failure with preserved ejection fraction.
heart failure and reduced ejection fraction. N. Engl. 142. Tromp, J. et al. Novel endotypes in heart failure: Circulation 131, 269–279 (2015).
J. Med. 381, 1995–2008 (2019). effects on guideline-directed medical therapy. 160. Kao, D. P. et al. Characterization of subgroups of heart
125. Lam, C. S. P., Chandramouli, C., Ahooja, V. & Eur. Heart J. 39, 4269–4276 (2018). failure patients with preserved ejection fraction with
Verma, S. SGLT-2 inhibitors in heart failure: current 143. Tromp, J. et al. Biomarker profiles in heart failure possible implications for prognosis and treatment
management, unmet needs, and therapeutic patients with preserved and reduced ejection fraction. response. Eur. J. Heart Fail. 17, 925–935 (2015).
prospects. J. Am. Heart Assoc. 8, e013389 (2019). J. Am. Heart Assoc. 6, e003989 (2017). 161. Segar, M. W. et al. Phenomapping of patients with
126. Vaduganathan, M. et al. Sudden death in heart failure 144. Sanders-van Wijk, S. et al. Circulating biomarkers of heart failure with preserved ejection fraction using
with preserved ejection fraction: a competing risks distinct pathophysiological pathways in heart failure machine learning-based unsupervised cluster analysis.
analysis from the TOPCAT trial. JACC Heart Fail. 6, with preserved vs. reduced left ventricular ejection Eur. J. Heart Fail. 22, 148–158 (2020).
653–661 (2018). fraction. Eur. J. Heart Fail. 17, 1006–1014 (2015). 162. Senni, M., Caravita, S. & Paulus, W. J. Do existing
127. Borlaug, B. A. et al. Impaired chronotropic and 145. Tromp, J. et al. Biomarker correlates of coronary definitions identify subgroup phenotypes or reflect the
vasodilator reserves limit exercise capacity in patients microvascular dysfunction in heart failure with natural history of heart failure with preserved ejection
with heart failure and a preserved ejection fraction. preserved ejection fraction. Circulation 140, fraction? Circulation 140, 366–369 (2019).
Circulation 114, 2138–2147 (2006). 1359–1361 (2019). 163. Andersson, C. et al. Risk factor-based subphenotyping
128. US National Library of Medicine. ClinicalTrials.gov 146. Hoeper, M. M. et al. Pulmonary hypertension in heart of heart failure in the community. PLoS One 14,
https://clinicaltrials.gov/ct2/show/NCT02145351 failure with preserved ejection fraction: a plea for e0222886 (2019).
(2020). proper phenotyping and further research. Eur. Heart J. 164. Borlaug, B. A., Melenovsky, V. & Koepp, K. E.
129. Hasenfuss, G. et al. A transcatheter intracardiac 38, 2869–2873 (2017). Inhaled sodium nitrite improves rest and exercise
shunt device for heart failure with preserved ejection 147. Borlaug, B. A. & Obokata, M. Is it time to recognize a hemodynamics in heart failure with preserved ejection
fraction (REDUCE LAP-HF): a multicentre, open-label, new phenotype? Heart failure with preserved ejection fraction. Circ. Res. 119, 880–886 (2016).
single-arm, phase 1 trial. Lancet 387, 1298–1304 fraction with pulmonary vascular disease. Eur. Heart J. 165. Borlaug, B. A., Koepp, K. E. & Melenovsky, V.
(2016). 38, 2874–2878 (2017). Sodium nitrite improves exercise hemodynamics and
130. Kaye, D. M. et al. One-year outcomes after 148. Gorter, T. M., Obokata, M., Reddy, Y. N. V., ventricular performance in heart failure with preserved
transcatheter insertion of an interatrial shunt device Melenovsky, V. & Borlaug, B. A. Exercise unmasks ejection fraction. J. Am. Coll. Cardiol. 66, 1672–1682
for the management of heart failure with preserved distinct pathophysiologic features in heart failure with (2015).
ejection fraction. Circ. Heart Fail. 9, e003662 (2016). preserved ejection fraction and pulmonary vascular 166. Reddy, Y. N. V. et al. The β-adrenergic agonist
131. Feldman, T. et al. Transcatheter interatrial shunt disease. Eur. Heart J. 39, 2825–2835 (2018). albuterol improves pulmonary vascular reserve
device for the treatment of heart failure with preserved 149. Borlaug, B. A., Lam, C. S., Roger, V. L., Rodeheffer, R. J. in heart failure with preserved ejection fraction.
ejection fraction (REDUCE LAP-HF I [Reduce elevated & Redfield, M. M. Contractility and ventricular systolic Circ. Res. 124, 306–314 (2019).
left atrial pressure in patients with heart failure]): stiffening in hypertensive heart disease insights into 167. Maurer, M. S. et al. Tafamidis treatment for patients
a phase 2, randomized, sham-controlled trial. the pathogenesis of heart failure with preserved with transthyretin amyloid cardiomyopathy. N. Engl.
Circulation 137, 364–375 (2018). ejection fraction. J. Am. Coll. Cardiol. 54, 410–418 J. Med. 379, 1007–1016 (2018).
132. Shah, S. J. et al. One-year safety and clinical outcomes (2009). 168. Gonzalez-Lopez, E. et al. Wild-type transthyretin
of a transcatheter interatrial shunt device for the 150. Shah, A. M. et al. Prognostic importance of impaired amyloidosis as a cause of heart failure with preserved
treatment of heart failure with preserved ejection systolic function in heart failure with preserved ejection fraction. Eur. Heart J. 36, 2585–2594
fraction in the reduce elevated left atrial pressure in ejection fraction and the impact of spironolactone. (2015).
patients with heart failure (REDUCE LAP-HF I) trial: Circulation 132, 402–414 (2015). 169. Mohammed, S. F. et al. Left ventricular amyloid
a randomized clinical trial. JAMA Cardiol. 3, 968–977 151. Melenovsky, V., Hwang, S. J., Lin, G., Redfield, M. M. deposition in patients with heart failure and preserved
(2018). & Borlaug, B. A. Right heart dysfunction in heart ejection fraction. JACC Heart Fail. 2, 113–122 (2014).
133. Obokata, M. et al. Effects of interatrial shunt on failure with preserved ejection fraction. Eur. Heart J.
pulmonary vascular function in heart failure with 35, 3452–3462 (2014). Acknowledgements
preserved ejection fraction. J. Am. Coll. Cardiol. 74, 152. Sabbah, M. S. et al. Obese-inflammatory phenotypes The author is supported by grants RO1 HL128526 and UO1
2539–2550 (2019). in heart failure with preserved ejection fraction. HL125205.
134. Borlaug, B. A. & Reddy, Y. N. V. The role of the J. Am. Coll. Cardiol. 73 (Suppl. 1), 661 (2019).
pericardium in heart failure: implications for 153. Van Tassell, B. W. et al. IL-1 blockade in patients Competing interests
pathophysiology and treatment. JACC Heart Fail. 7, with heart failure with preserved ejection fraction. The author declares no competing interests.
574–585 (2019). Circ. Heart Fail. 11, e005036 (2018).
135. US National Library of Medicine. ClinicalTrials.gov 154. Houstis, N. E. et al. Exercise intolerance in heart Peer review information
https://clinicaltrials.gov/ct2/show/NCT03499236 failure with preserved ejection fraction: diagnosing Nature Reviews Cardiology thanks A. Desai, M. Metra
(2020). and ranking its causes using personalized O2 pathway and W. Paulus for their contribution to the peer review of
136. US National Library of Medicine. ClinicalTrials.gov analysis. Circulation 137, 148–161 (2018). this work.
https://clinicaltrials.gov/ct2/show/NCT03088033 155. Fayyaz, A. U. et al. Global pulmonary vascular
(2020). remodeling in pulmonary hypertension associated Publisher’s note
137. Borlaug, B. A. et al. Percutaneous pericardial resection: with heart failure and preserved or reduced ejection Springer Nature remains neutral with regard to jurisdictional
a novel potential treatment for heart failure with fraction. Circulation 137, 1796–1810 (2018). claims in published maps and institutional affiliations.
preserved ejection fraction. Circ. Heart Fail. 10, 156. Molina, A. J. et al. Skeletal muscle mitochondrial
e003612 (2017). content, oxidative capacity, and Mfn2 expression © Springer Nature Limited 2020

Nature Reviews | Cardiology