Peer reviewed article
66
Review article
SWISS MED WKLY 2010 ; 139 ( 5 –6 ) : 66–72 · www.smw.ch
Heart failure with normal ejection fraction
(HFNEF): is it worth considering?
Coralie Blanche, Thierry Fumeaux, Ralf Polikar
Internal Medicine and Intensive Care Unit, GHOL, Regional Hospital Nyon, Switzerland
Summary
A significant proportion of patients with heart
failure happen to have a normal ventricular ejec-
tion fraction at echocardiography during exami-
nation. Previously called diastolic heart failure, it
is nowadays referred to as heart failure with nor-
mal ejection fraction (HFNEF) or HF with pre-
served ejection fraction. The European Society of
Cardiology, recognizing the importance of this
type of heart failure, recently issued new definition
criteria for it.This review will discuss the different
steps that lead to such a diagnosis, as well as some
new aspects of its pathophysiology. Finally, the
management of this form of HF, that is not as
straightforward as HF due to systolic dysfunction,
will be discussed.
Key words: heart failure; HFPEF; HFNEF;
Diastolic dysfunction; BNP
No financial
support.
Introduction
Heart failure (HF) is a complex syndrome, re-
sulting from structural or functional cardiac disor-
ders that impair the ability of the cardiac pump to
support a physiological circulation. HF is a fre-
quent pathology in Switzerland, with a significant
proportion of patients who need to be admitted to
the hospital, and a high case fatality rate [1]. Ad-
vances in the management of patients with left
ventricular (LV) systolic dysfunction have resulted
in a significant extension of life expectancy, but
this is not the case for patients with HF in the ab-
sence of echocardiographic LV systolic dysfunc-
tion [2–5]. This type of HF is now being called
Heart Failure with Normal Ejection Fraction
(HFNEF) although HF with Preserved EF (HF-
PEF), which probably better delineates the fact
that systolic function is not completely normal in
such patients but only apparently preserved, and
diastolic heart failure (DHF) have also been used
[6]. The European Society of Cardiology (ESC)
recently underlined the clinical importance and
complexity of HFNEF, and issued new definition
criteria, based on clinical signs and/or symptoms
of HF, echocardiographic and biological parame-
ters [7]. This review will focus on the most recent
data gained on the mechanisms of HFNEF, as well
as on the results of the latest clinical studies on
HFNEF. It is intended to help the general practi-
tioner understand and apply the new ESC defini-
tion of HFNEF into their daily practice.
Clinical recognition and epidemiology
Numerous publications report that many pa-
tients who present with symptoms or signs of HF
do not have decreased LV ejection fraction
(LVEF). Data from recent studies show a preva-
lence rate of HFNEF ranging from 30 to 50% in
patients with HF, at least in patients with hospital
admission [8–11].HFNEF patients are likely to be
older women, in whom systolic hypertension and
myocardial hypertrophy with fibrosis are the main
contributors to cardiac dysfunction.
The general prevalence of HFNEF seems to
be on the rise, as a possible consequence of grow-
ing recognition, population ageing and increases
in hypertension and obesity prevalence [8]. Even
though HFNEF mortality may be lower than in
HFREF, high NYHA functional capacity carry a
poor prognosis, whatever the LV systolic function.
Finally, HFNEF patients have a reduced life
expectancy, due to an increased risk of sudden
cardiac death and acute HF, implying that these
individuals have a significant functional cardiac
disease [8, 10, 12]. For these reasons, HFNEF has
become a public health problem and deserves
more attention [8–10].
67
SWISS MED WKLY 2010 ; 140 ( 5 –6 ) : 66–72 · www.smw.ch
HFNEF implies some degree of LV filling al-
teration despite an apparently normal LV systolic
function: this so-called LV diastolic dysfunction
(DD) results from various mechanisms,mainly ab-
normal relaxation and decreased compliance, with
a rise in LV end-diastolic pressure (LVEDP) [13].
Systolic function seems preserved, at least as as-
sessed by 2D echocardiography, but this may be
more complex, as this technique does not take into
account all aspects of LV contractility.
Diastolic function
Diastolic function of the LV depends on active
relaxation and on the stiffness characteristics of the
ventricular wall [13].Diastolic dysfunction is not only
the consequence of altered geometric and elastic
properties of themyocardium, but is also influenced
by pericardial distensibility, interventricular depen-
dency, left atrial pressure, and electrical atrio-ven-
tricular coupling. In addition, the vascular bed’s
compliance and resistance may play a significant
role in the development of diastolic dysfunction.
Zile et al. invasively recorded ventricular pres-
sures in patients with HFNEF, and confirmed the
increase in LVEDP, together with smaller end-dia-
stolic volumes, compared to controls [13]. They
found evidence of a slower and incomplete relax-
ation associated with an increased passive ventricu-
lar stiffness, which could explain the high LVEDP.
Westermann et al. confirmed these observations,
with an invasive measure of ventricular pressure-
volume loops in HFNEF patients: they showed an
elevated LVEDP that might be due to higher LV
stiffness and increased relaxation time [14]. These
alterations were more marked during handgrip or
tachycardia, but with a different pattern of altera-
tions in these two situations, suggesting a complex
mechanism of diastolic dysfunction during exercise,
not only due to heart rate increase [15].
Tachycardia may reveal diastolic dysfunction
and is often not well tolerated by patients with HF-
NEF: Wachter et al. demonstrated that relaxation
timewas prolonged inHFNEF, and that the normal
frequency-dependent acceleration of relaxation was
blunted, resulting in significant decrease of end-di-
astolic volume and stroke volume with higher heart
rates [16].
Eventually, recent and precise descriptions of
diastolic function, by the association of echocardio-
graphicmeasures of ventricular dimensions coupled
with invasive measures of ventricular pressures us-
ing an implantable device called e-PAD,addedmore
evidence that diastolic distensibility is decreased in
HFNEF [17, 18].
In the daily practice, the non-invasive evalua-
tion of LV diastolic function is performed with
transthoracic echocardiography (TTE). Early mi-
tral diastolic flow (called Ewave), due to the ventric-
ular relaxation, and late mitral diastolic flow (called
A wave), secondary to the atrial kick, can be mea-
sured with Doppler-imaging. The velocity ratio of
these waves, called E/A ratio, has been used as an in-
direct evaluation of ventricular filling pressure, and
as a surrogate marker of diastolic function. Three
grades of diastolic dysfunction have been defined.
Grade I (E/A <1) is referred to as relaxation abnor-
mality and is often encountered in elderly patients.
Grade II (E/A >1) is called pseudo-normalization,
since it is also recorded in young healthy individuals.
As LA pressure increases further, relaxation time
decreases and the contribution of the LA kick to LV
filling diminishes, with a “restrictive” pattern (E/A
>1,5 and DT <140 msec), referred to as Grade III.
This range of severity has been shown to represent
a negative prognostic marker, especially in patients
with systolic dysfunction [19, 20].
However, the E/A ratio may not precisely re-
flect diastolic function, partly because it is load de-
pendant and it has variable performances [21, 22].
OtherTTE parameters give some indication on di-
astolic function, such as the deceleration time of the
Ewave, the isovolumic relaxation time, the left atrial
size, and the LV wall thickness, and, as elegantly
stated by L.Hatle, diastolic dysfunction is better as-
sessed when all TTE and Doppler parameters are
put together [23].
Tissue Doppler imaging (TDI) is a novel tech-
nique, which allows analysis of global diastolic and
systolic velocities of the myocardial wall. The E’
wave is the peak velocity of early diastolic myocar-
dial lengthening, recorded at the mitral annulus.
When combined with the E wave, measured with
conventional Doppler, the E/E’ ratio is well corre-
lated to LVEDP and is less load-dependent than
E/A ratio [24–26]. It also correlates very well with
HF symptoms, both in HFREF and HFNEF [27–
29]. An E/E’ ratio >15 has been chosen as a cut-off
value for diastolic dysfunction, and a value <8 is sen-
sitive enough to exclude it.
When E/E’ ratio is between 8 and 15, other ap-
proaches, such as pulmonary vein Doppler analysis,
M-mode colorDoppler of LVfilling, the “time con-
stant” of LV relaxation (t), the ‘stiffness’ constant of
LV (b), the left atricular volume index (LAVI), and
the left ventricular mass index (LVMI) can be used
in conjunction with LA size [30]. However, in clini-
cal practice, these novel approaches are not yet
widely available, and cardiac catheterization remains
the gold standard for filling pressure assessment.
The E’/E ratio is of limited value in patients
with abnormal relaxation but normal LVEDP at
rest.Marked diastolic dysfunction can develop with
exercise, and diastolic dysfunction can be missed,
unless there is an LA enlargement. Likewise, some
patients with a decrease in compliance without sig-
nificant relaxation abnormality may have a normal
E/E’, and once again LA size and blood flow Dop-
pler analysis might then be the only clues of a dia-
stolic dysfunction [23, 31].
Pathophysiology of HFNEF
68
Heart failure with normal ejection fraction (HFNEF): is it worth considering?
Systolic function
By definition, systolic function should be nor-
mal or preserved in HFNEF, at least when mea-
sured with echocardiography.However,TTEmay
be misleading, because it only evaluates radial
shortening and doesn’t take into account other as-
pects of systolic function, such as longitudinal
shortening, circumferential twist, and regional ab-
normalities. This is important, since loss of longi-
tudinal motion of the LV is the most obvious
change associated with ageing, and because radial
shortening is often increased in the early stage of
diabetic and hypertensive heart disease [32–34].As
elderly, diabetic and hypertensive subjects repre-
sent a significant proportion of HFNEF patients,
the relative preservation of LVEF may lead to a
wrong impression.
LV systolic function can nowadays be evalu-
ated more precisely with TDI. LV longitudinal
shortening is assessed by the amplitude of mitral
annular shortening, and when associated with the
evaluation of systolic myocardial velocities, this
approach yields a much better detection of global
and regional systolic dysfunction thanTTE.With
this tool, it has been shown that systolic function
may be altered not only in HFREF, but also in pa-
tients with HFNEF [35–37]. Other recently de-
veloped techniques, such as strain rate imaging
(SRI) which uses the time integral of the velocity
gradient between two adjacent myocardial seg-
ments, may also be better at detecting regional
and global alterations of systolic function [38].
The application of these new techniques
shows that LV systolic function is, in fact, probably
slightly altered in patients with HFNEF:Tan et al.
have studied HFNEF patients by various echocar-
diographic techniques, both at rest and at exercise,
and have shown that HFNEF patients have sys-
tolic segmental alterations, such as lower apical
rotation, decreased ventricular suction and re-
duced untwisting, and that these alterations are
more pronounced at exercise [15]. One can con-
clude that standard TTE doesn’t seem to have
perfect diagnostic performance for systolic func-
tion, and that EF doesn’t reflect systolic function
in its entirety, and may seem normal because of a
change in ventricular geometry. Based on a mathe-
matical model, Maclver et al have shown that EF
preservation in HFNEF is probably related to in-
creased muscle mass and increased radial thicken-
ing of the ventricular wall, compensating for the
reduced long-axis shortening [39].
All these data suggest that indeed HFNEF
and HFREF may represent a continuum, and not
separate entities.
Cellular alterations associated with HFNEF
Van Heerbeck et al. demonstrated that the
myocardiumof patients withHFNEF andHFREF
present the same degree of histological fibrosis,
but that cardiomyocytes express different isoforms
of the constitutive cytoskeletal protein titin, which
increases diastolic compliance upon phosphoryla-
tion [40]. The N2B isoform expressed in HFNEF
is stiffer and its increased expression may contrib-
ute to the increased ventricular stiffness, and par-
ticipate in the development of diastolic dysfunc-
tion.
Diagnosis of HFNEF
HF is a clinical syndrome and may present in
either acute or chronic form. Acute onset of dys-
pnea and/or congestion will often lead to hospital-
ization, and the clinical suspicion of HF is high in
most cases. More chronic symptoms, such as dys-
pnea on exertion, nocturnal paroxysmal dyspnea,
orthopnea, or fatigue, may be the only features of
HF, particularly in the outpatient setting and
aren’t specific of any type of HF.
Signs and symptoms of HF result from ele-
vated LV filling pressure and/or low cardiac out-
put, which may be the consequence of either in-
complete emptying, as in systolic dysfunction, or
impaired ventricular filling as in diastolic dysfunc-
tion. Therefore, the clinical presentation of HF-
PEF and HFNEF share similarities such as a third
heart sound, elevated jugular venous pressure and
signs of pulmonary congestion, which, in the past,
had been considered to be the hallmark of LV sys-
tolic dysfunction but are also found in patients
with diastolic dysfunction. As their performance
characteristics may vary, classical symptoms and
signs are not sufficient for the diagnosis of HF
(fig. 1) [41].The Framingham criteria are sensitive
but moderately specific, and apply poorly to non
congestive or moderate HF (table 1) [42–45].
In the daily practice, the physician firstly
needs to confirm that HF is the cause of the pa-
Figure 1
Diagnostic value of symptoms and signs of HF.
Adapted with permission from Brunner-La Rocca [41] and Schweiz Med Forum
2007;7(Suppl. 39):3–14 S, http://www.medicalforum.ch HF: Heart failure
MI: myocardial infarction COPD: chronic obstructive pulmonary disease
69
Heart failure with normal ejection fraction (HFNEF): is it worth considering?
tient’s presenting symptoms and signs, and then to
distinguish between HFREF and HFNEF. The
most recent guidelines, including the ESC algo-
rithm, recommend that the diagnosis of HF be
based upon ECG, chest X-ray and measurement
of natriuretic peptides, and confirmed by echocar-
diographic evaluation, which in addition will help
differentiate HFREF from HFNEF (fig. 2 and 3)
[6, 41].
Brain natriuretic peptide (BNP) and its pre-
cursor (NT-proBNP) are produced by cardiomy-
ocytes in response to volume or pressure overload,
and are very sensitive markers, so that HF is
unlikely in presence of normal levels [46–48].
Specificity is not as good, because blood levels are
influenced by many factors, such as age, BMI and
kidney function, and may increase in various clini-
cal situations. This is why the ‘rule-in’ cut-off val-
ues may differ in various guidelines, and the inter-
pretation of these values always has to be cautious
and linked to the clinical situation [46, 49]. It has
to be emphasized that, in recent years, natriuretic
peptides have been made readily available in each
emergency room and private practice, and have
Major criteria
– Paroxysmal nocturnal dyspnea
– Neck vein distention
– Rales
– Radiographic cardiomegaly
– Acute pulmonary edema
– S3 gallop
– Increased central venous pressure (>16 cm H
2
O at right atrium)
– Hepatojugular reflux
–Weight loss >4.5 kg in five days in response to treatment
Minor criteria
– Bilateral ankle edema
– Nocturnal cough
– Dyspnea on ordinary exertion
– Hepatomegaly
– Pleural effusion
– Decrease in vital capacity by one third from maximum
recorded
– Tachycardia (heart rate >120 beats/min.)
Table 1
Framingham criteria
for the diagnosis
of heart failure (HF):
HF is diagnosed
when 2 major criteria
OR 1 major criteria
plus 2 minor criteria
are present.
Figure 2
ESC diagnosis
algorithm.
How to diagnose HFNEF
Symptoms or signs of heart failure
Normal or mildly reduced left ventricular systolic function
LVEF >50%
and
LVEDVI <97 mL/m
2
Evidence of abnormal LV relaxation, filling, diastolic
distensibility, and diastolic stiffness
Invasive Haemodynamic measurements
mPCW >12 mm Hg
or
LVEDP >16 mm Hg
or
r >48 ms
or
b >0.27
Biomarkers
NT-proBNP >220 pg/mL
or
BNP >200 pg/mL
TD
EIE’ >8
TD
EIE’ >15 15 >EIE’ >8
Biomarkers
NT-proBNP >220 pg/mL
or
BNP >200 pg/mL
Echo – blodflow Doppler
EIA
–50yt
<0.5 and DT
–50yt
>280 ms
or
Ard-Ad >30 ms
or
LAVI >40 mL/m
2
LVMI >122 g/m
2
(/); >149 g/m
2
(?)
or
Atrial fibrillation
HFNEF
Reproduced with permission from Paulus et al. [7]. HFNEF: heart failure with normal ejection fraction LVEF: left ventricular
ejection fraction LVEDVI: left ventricular end-diastolic volume indexed to body surface LV: left ventricular mPCW: mean
pulmonary capillary wedge pressure LVEDP: left ventricular end-diastolic pressureTD: tissue Doppler LAVI: left atrial volume
indexed to body surface area LVMI: left ventricular mass indexed to body surface area. See text for other abbreviations.
70
Heart failure with normal ejection fraction (HFNEF): is it worth considering?
been better validated than the various echo-Dop-
pler variables for the diagnosis and prognosis of
HF.Therefore, they are often put forward by most
experts for the diagnosis of HFNEF (fig. 3) and
used as inclusion criteria in most ongoing studies
of HFNEF [46].
After the diagnosis of HF has been made, dia-
stolic dysfunction must be proved to confirm the
presence of HFNEF. The hallmark of diastolic
dysfunction is elevated filling pressures, such as
LVEDP and pulmonary capillary wedge pressure
(mPCWP), which can be measured invasively [7].
However, invasive haemodynamic studies are
rarely undertaken and non invasive parameters are
most often used to confirm diastolic dysfunction.
TTE is the most widely available confirma-
tory tool for HF, and the ESC and Swiss guide-
lines both recommend its use. As already dis-
cussed, it can distinguish between HFNEF and
HFREF, but also gives clues for the etiology of
HF. In addition,TTE has a prognostic value, since
low EF, LV remodeling and LV restrictive filling
patterns are associated with a high mortality.TTE
has some limitations, and cardiac catheterization
should be considered for more accurate measure-
ment of LVEDP and PCWP.
Figure 3
Use of BNP for the diagnosis of HF.
Adapted and reproduced with permission from Maisel et al. [46]
HF: heart failure
Clinical suspicion of HF
BNP levels
BNP <100 pg/ml
HF very improbable
(<2%)
BNP 100–400 pg/ml
Past history of HF
HF probable (75%)
BNP >400 pg/ml
HF very probable
(>98%)
Management of HFNEF
As HFNEF is associated with significant
morbidity and mortality [1], patients should have
regular medical visits to evaluate worsening of
signs and symptoms, associated with regular echo-
cardiographic evaluations, in order to initiate ap-
propriate therapy in case of a worsening of systolic
function.Cardiovascular risk factors should be de-
tected and treated. Rapid or unexpected deteriora-
tion should be quickly assessed by cardiologists.
Medi