Article Text

Download PDFPDF

Heart failure with a normal ejection fraction: new developments
  1. Gabriel Wai-Kwok Yip1,
  2. Michael Frenneaux2,
  3. John E Sanderson2
  1. 1
    Division of Cardiology, Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong, Peoples’ Republic of China
  2. 2
    Department of Cardiovascular Medicine, University of Birmingham, Birmingham, UK
  1. Correspondence to Professor J E Sanderson, Department of Cardiovascular Medicine, University of Birmingham, Birmingham B15 2TT, UK; j.e.sanderson{at}

Statistics from

About half of patients with symptoms and signs of heart failure will be found to have a normal or near-normal left ventricular ejection fraction (LVEF). These patients are mostly elderly women, and the majority have a history of hypertension in contrast to a predominantly ischaemic aetiology in those with heart failure and a reduced ejection fraction (HFREF). Heart failure with a normal ejection fraction (HFNEF) is proving to be intriguing with only a few established facts but many myths. First, the precise underlying pathophysiology is still debated. Recent work has demonstrated that abnormalities exist in LV systolic properties, ventricular–arterial coupling, LV diastolic function, torsion or twist, ventricular–ventricular interaction, pericardial constraint, with impaired chronotropic, vasodilator reserves and pulmonary hypertension. Thus, the term HFNEF is preferred rather than diastolic heart failure (DHF) as the term DHF implies that the primary or dominant abnormality is in diastole alone to which treatment should be targeted, which may be misguided. Second, the prognosis appears to be similar in both HFNEF and HFREF. Third, dichotomising heart failure into systolic and diastolic clinical entities based on the LVEF has led to a paucity of clinical trials of treatments for HFNEF. Over the past 20 years, significant advances in drug and device therapy have improved survival in patients with HFREF, yet evidence-based treatment for reducing cardiovascular mortality and morbidity in HFNEF is lacking. Finally, on present trends, HFNEF may become the most common form of heart failure and thus a major health problem for which there is little of proven therapeutic value, although some doubt how common true HFNEF is. There is, therefore, an urgent need for coordinated efforts to examine this apparently common condition.

Definitions and diagnosis

Several criteria for defining HFNEF have been proposed. In the recently updated European Society of Cardiology guidelines, diagnosis of HFNEF depends on clinical symptoms and signs of heart failure in the presence of an EF >50% in a non-dilated left ventricle (LV end-diastolic volume <97 ml/m2) and abnormalities in LV diastolic function/filling.1 Although invasive assessment using LV pressure–volume relations are considered to be the “gold standard” for objective evidence of diastolic dysfunction, the results are variable and do not precisely measure either relaxation or “stiffness” as commonly supposed.2 The time constant of relaxation (τ) is determined by fitting a monoexponential curve to the isovolumic fall in LV pressure and calculating the relaxation half-time (t1/2). However, the curve is not always exponential, especially if left atrial pressure is high, and τ cannot distinguish between reduced myocyte relaxation and incoordinate relaxation. The end-diastolic pressure–volume relationship is not a measure of stiffness but the extent to which stiffness depends on volume. A normal left ventricle will become “stiff” if volume is increased abnormally, as for example in renal failure. Echocardiography remains the most widely accessible and versatile clinical tool for assessment of ventricular function. Recent tissue Doppler and strain parameters have given a better insight into the mechanics of HFNEF, and a non-invasive method for estimating LV filling pressures (the ratio E/e′) can be derived,3 although recently doubt has arisen about the accuracy of this index, at least in HFREF.4 However, the simplest measure is the left atrial volume index (LAVI) as an enlarged LAVI suggests a chronically raised LV end-diastolic pressure.5 The presence of left ventricular hypertrophy (LVH) and an increased LAVI in a breathless patient strongly suggests the likelihood of HFNEF. B-type natriuretic peptide (BNP) and N-terminal pro-brain natriuretic peptide (NT-proBNP) <100 and <120 pg/ml, respectively, are helpful to exclude the diagnosis, especially in presence of indeterminate diastolic function based on tissue Doppler and strain echocardiography.1

HFNEF and HFREF phenotypes: from microscopy to the whole person

Cardiomyocytes of patients with HFNEF and HFREF appear to be structurally and functionally different. HFNEF is characterised by cardiomyocyte hypertrophy and preserved myofilamentary density, whereas HFREF has minimal cardiomyocyte hypertrophy and loss of myofilaments.6 The resting tension and calcium sensitivity of isolated cardiomyocytes are also higher in HFNEF.7 However, even in studies showing increased passive tension in myocytes from patients with HFNEF, myocyte total force is reduced similar to HFREF myocytes. The problem with isolated myocyte studies is that the sarcomeres shorten by 15%, whereas the whole ventricle ejects greater than 50% of its volume.8 This can only be explained by taking account of the architecture of the heart and fibre orientation. The main pathological differences are in the extracellular matrix with an increase in interstitial collagen in HFNEF in contrast to a degradation of endo- and perimysial components of the collagen scaffolding in HFREF.9 The difference may be related to balance or imbalance between collagen production and turnover, as reflected by the ratio of matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs)—for example, MMP-1/TIMP-1 ratio.10 These extracellular matrix changes may have an important effect on myocardial fibre architecture, function and remodelling. In addition, the higher ratio of the shorter and stiffer N2B isoform to the longer but more compliant N2BA titin isoform, a giant sarcomeric cytoskeletal protein extending from the Z lines to the centre of the thick filament that serves as a molecular spring, may contribute to impaired passive recoil and delayed early filling in both conditions.11

Despite these pathological differences, the separation of patients with heart failure into two phenotypes is contentious.12 Patients with HFNEF have similar though not as severe pathophysiological characteristics as patients with typical HFREF. Neuroendocrine activations (norepinephrine and BNP levels) are similar in both groups. There are subtle abnormalities of systolic LV function in HFNEF (see below) further worsened during exercise which are not appreciated from an LV ejection fraction measurement at a single time point. Diastolic LV dysfunction, impaired chronotropic, systemic vasodilator and contractile reserves on exercise are seen in both entities. Thus, these apparently separate phenotypes differ only in the degree of ventricular remodelling and volume changes paralleling the extent of ventricular fibrosis and scarring. It appears more useful to classify heart failure according to the aetiology and the mechanisms involved in the individual patient, which may be different.12 Table 1 summarises the similarities and differences between HFNEF and HFREF and illustrates that increased ventricular volume, reflecting remodelling, is the main difference between the two. In addition, MacIver and Townsend have demonstrated that in the presence of LVH and a normal LV volume LVEF is mathematically bound to increase, falsely creating the impression of normal systolic function.13 Essentially, there is heart failure with large or small hearts.

Table 1

Morphological, microscopic, resting and dynamic functional changes in heart failure with normal ejection fraction (HFNEF) and heart failure with reduced ejection fraction (HFREF)

Postulated mechanisms of HFNEF

Mechanisms of HFNEF were postulated based on measurements acquired at rest,

  1. An increase in intrinsic myocardial stiffness and impaired relaxation in diastole result in higher left atrial pressures to fill adequately, predisposing these patients to pulmonary venous congestion and dyspnoea, especially on exertion.14

  2. Increased systolic ventricular and arterial stiffening—that is, deranged ventriculoarterial coupling, may contribute to the pathophysiology of HFNEF by exaggerating hypertensive response with increased systolic pressure load, and inducing load-dependent diastolic dysfunction, especially during exercise or other stresses, in addition to sensitivity of blood pressure to circulating volume and diuretics.15

  3. Enhanced sensitivity to volume overload from increased LV remodelling and dilatation with volume-dependent elevation of filling pressures was seen in a small subgroup of hypertensive patients with HFNEF who had renal impairment and larger LV volumes but normal systolic ventricular and vascular stiffness.16 Thus, impaired renal function and renal arterial atherosclerosis in the elderly may also help to cause rapid rises in blood pressure and excessive fluid retention.

These concepts underestimate the impact of the previous systole on early diastolic filling particularly incoordination, dyssynchrony, reduced longitudinal function and torsion. Furthermore, the orthodox view that systolic function is entirely normal has been challenged in studies using newer echocardiographic techniques which have shown that systolic function does not appear to be entirely normal in all subjects with HFNEF, or those with LVH and diabetes both common aetiological factors for HFNEF.12 New insights and a more holistic explanation have come from studying ventricular function not only at rest but on exercise. After all, heart failure is a disease of exercise.

New insights from exercise echocardiography

Systolic and diastolic function are intimately linked17 and the architecture of the heart is engineered to provide both rapid ejection and filling of the heart. This is achieved through the helical structure of the myocardium fibres in opposing layers, which allows the heart to twist.8 In the normal heart, left ventricular twist during systole stores energy, and pulls the mitral annulus towards the apex (which also helps suck blood into the atrium), and the corresponding untwisting process and recoil in early diastole when that energy is released generates the negative intraventricular pressure gradient or suction in early diastole which is vital for rapid filling. This process is followed by recoil of the mitral annulus back towards the base of the heart, which also aids ventricular filling by moving the mitral annulus around the column of the incoming blood. All these aspects of ventricular function increase on exercise, not only to accelerate ventricular ejection but also, more importantly, to enable rapid filling of the ventricle during a shortened diastole while maintaining a low filling pressure.18

In HFNEF, this close relation between systole and diastole is disrupted, and our recent studies have shown a variety of abnormalities of systolic and diastolic function on exercise: reduced myocardial systolic strain, reduced ventricular systolic rotation at rest (which fails to increase normally on exercise), reduced mitral annular motion in systole and diastole and delayed ventricular untwisting associated with reduced left ventricular suction. Mean mitral annular systolic and diastolic velocities, systolic LV rotation and early diastolic untwist on exercise correlated with peak Vo2 max.19 Impaired atrial function on exercise may also contribute to breathlessness.20 Thus, HFNEF is not an isolated disorder of diastole. It is not clear how frequently HFNEF evolves into HFREF but hypertension is often the only cause of heart failure in many areas of the world, including Africa and Asia, suggesting that hypertensive heart failure passes from a phase of LVH alone to HFNEF and then HFREF.

Other factors

Recent community-based data have shown that pulmonary hypertension is common and often severe in HFNEF and it cannot be fully accounted for by passive contribution of pulmonary venous hypertension,21 and exercise-induced pulmonary hypertension is common even in those with a normal LV ejection fraction.22

Ageing is associated with decreases in the elastic properties of the heart and great vessels, which leads to increased systolic blood pressure and myocardial stiffness. Arterial stiffness is an independent predictor of diastolic dysfunction and linked to impaired coronary flow velocity reserve, which may cause subendocardial ischaemia and reduced longitudinal ventricular function.23 Patients with HFNEF have attenuated heart rate responses to exercise, similar to patients with HFREF.24 Patients with chronotropic incompetence have more severe exercise intolerance than those without it. The slower heart rate rise, lower peak heart rate and impaired recovery suggest autonomic (parasympathetic) dysfunction.

Metabolic abnormalities in HFNEF

Cardiac contraction and active relaxation are both highly energy requiring processes (the heart generates several times its own weight in ATP every day). Impairment of cardiac energetic status (assessed by cardiac MR spectroscopy) is well described in HFREF and carries adverse prognostic significance in these patients.25 One study reported reduced cardiac energetics in hypertensive patients with “diastolic dysfunction”,26 and in another study the cardiac phosphocreatine (PCr)/ATP ratio was found to be reduced in hypertensive subjects with LVH compared with healthy controls, and this reduction was more marked in those patients undergoing transition to “heart failure” (whether “systolic” or “diastolic”).26 More recently, in a carefully characterised population of elderly patients with HFNEF we have shown that the cardiac PCr/ATP ratio was markedly reduced in patients compared with age-matched controls and that this was associated with both an abnormal slowing of the rate of LV active relaxation on exercise and a failure of contractility to increase appropriately on exercise.27 It seems likely that increased arterial stiffness (with ventriculoarterial mismatch) combined with cardiac energetic impairment may underlie these dynamic changes in cardiac function on exercise. This opens the possibility of “metabolic modulators” (such as trimetazidine and perhexiline) as therapeutic agents.


Because patients with HFNEF were excluded from all major therapeutic trials based on a normal LVEF, there is very little evidence on which to base treatment. One trial found that the angiotensin receptor antagonist candesartan modestly reduced hospital admissions for heart failure but did not significantly affect mortality in patients with HFNEF.27 Another recent study also evaluated the effect of the angiotensin receptor blocker irbesartan on mortality and cardiovascular morbidity in 4128 patients with HFNEF. It found no benefit of irbesartan over placebo in reducing mortality or morbidity from cardiovascular disease.28 However, a small randomised controlled trial found that diuretics alone reduced symptoms and improved quality of life significantly, but that adding ramipril or irbesartan was not more efficacious.29 These negative results are surprising. Fibrosis of the left ventricle is increased with LVH and hypertension. Angiotensin converting enzyme inhibitors and angiotensin receptor blockers can block the fibrogenic action of angiotensin experimentally and have been shown to reduce fibrosis in patients with hypertension.30 Fibrosis and altered collagen in LVH may have a deleterious effect on overall myocardial architecture, particularly ventricular twist and torsion. Nevertheless, the reduction of fibrosis may be an important therapeutic target, and the current studies of spironolactone in HFNEF will be interesting.


HFNEF is not a disease of diastole alone and there are complex mechanical and metabolic abnormalities of ventricular systolic and diastolic function that lead to impaired early diastolic filling especially on exercise. The aetiology is probably multifactorial, although hypertension with LVH, diabetes, ischaemia and ageing may all conspire to alter myocardial architecture and ultimately, function (fig 1). Treatment is entirely empirical and although diuretics do help the breathlessness no treatment at present seems to have any significant impact on mortality.

Figure 1

Schema of pathophysiology of breathlessness on exertion in heart failure with normal ejection fraction. LVH, left ventricular hypertrophy. Reproduced, with permission, from Tan et al.19


View Abstract


  • Competing interests GW-KY has no competing interests. MF is a consultant to Medtronic and St Jude and has a patent for perhexiline in heart failure (no current financial value). JES has received travel grants from Sanofi-Aventis, Boehringer-Ingelheim and Pfizer and lecture fees from Pfizer.

  • Provenance and peer review Not commissioned; not externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.