JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 76, NO.

6, 2020

ª 2020 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

PUBLISHED BY ELSEVIER

THE PRESENT AND FUTURE

JACC SCIENTIFIC EXPERT PANEL

Heart Failure With Recovered

Left Ventricular Ejection Fraction
JACC Scientific Expert Panel

Jane E. Wilcox, MD,a James C. Fang, MD,b Kenneth B. Margulies, MD,c Douglas L. Mann, MDd

ABSTRACT

Reverse left ventricular (LV) remodeling and recovery of LV function are associated with improved clinical outcomes in
patients with heart failure with reduced ejection fraction. A growing body of evidence suggests that even among patients
who experience a complete normalization of LV ejection fraction, a significant proportion will develop recurrent LV
dysfunction accompanied by recurrent heart failure events. This has led to intense interest in understanding how to
manage patients with heart failure with recovered ejection fraction (HFrecEF). Because of the lack of a standard definition
for HFrecEF, and the paucity of clinical data with respect to the natural history of HFrecEF patients, there are no current
guidelines on how these patients should be followed up and managed. Accordingly, this JACC Scientific Expert Panel
reviews the biology of reverse LV remodeling and the clinical course of patients with HFrecEF, as well as provides
guidelines for defining, diagnosing, and managing patients with HFrecEF. (J Am Coll Cardiol 2020;76:719–34) © 2020 by
the American College of Cardiology Foundation.

OVERVIEW OF HEART FAILURE WITH A their outcomes and clinical management, as well as
RECOVERED LEFT VENTRICULAR raised interest in understanding how these patients
EJECTION FRACTION differ from those with more modest to little positive
change in LVEF (nonresponders). Improvements in
The recognition that left ventricular ejection fraction LVEF with guideline-directed medical therapy
(LVEF) improves substantially in a subset of heart (GDMT) can lead to a complete normalization of LVEF
failure (HF) patients with reduced ejection fraction (i.e., >50%) or a partial normalization of LVEF (40%
(HFrEF) who are treated with evidenced-based med- to 50%) (Figure 1). Estimates of the proportion of pa-
ical and device therapies has led to intense interest in tients with improved LVEF range widely (e.g., 10% to

From the aDivision of Cardiovascular Medicine, Department of Medicine, Northwestern University Feinberg School of Medicine,
Chicago, Illinois; bDivision of Cardiology, Department of Medicine, University of Utah, Salt Lake City, Utah; cTranslational
Research Center, Department of Medicine, University of Pennsylvania Pearlman School of Medicine, Philadelphia, Pennsylvania;
and the dCardiovascular Division, Department of Medicine, Washington University School of Medicine, St. Louis, Missouri. Dr.
Wilcox has received funding from the National Institutes of Health (NIH) and the American Heart Association; has received
consulting/speaking honoraria from Abbott and Medtronic; and serves as a scientific consultant/advisory board member for
Cytokinetics. Dr. Fang has received funding from the National Institutes of Health (NIH) and the American Heart Association; is on
Listen to this manuscript’s the steering committee for clinical trials sponsored by Novartis, Amgen, and AstraZeneca; and is on the Data Safety and Moni-
audio summary by toring Board for a clinical trial sponsored by AstraZeneca. Dr. Margulies holds research grants from Thoratec Corporation (Abbott)
Editor-in-Chief and Sanofi; and serves as a scientific consultant/advisory board member for Pfizer, MyoKardia, Inc., and American Regent. Dr.
Dr. Valentin Fuster on Mann has received funding from the NIH; serves on the Scientific Advisory Board for MyoKardia, Inc.; and served on the steering
JACC.org. committee for a clinical trial sponsored by Novartis. David Burkhoff, MD, PhD, served as Guest Associate Editor for this paper. P.K.
Shah, MD, served as Guest Editor-in-Chief for this paper.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the JACC author instructions page.

Manuscript received January 7, 2020; revised manuscript received May 7, 2020, accepted May 14, 2020.

ISSN 0735-1097/$36.00 https://doi.org/10.1016/j.jacc.2020.05.075

720 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

ABBREVIATIONS 40%) due to variable definitions and the use

AND ACRONYMS HIGHLIGHTS
of both observational and clinical trial data-
sets. There is no consensus definition
This consensus document was created
CMR = cardiac magnetic
resonance
currently available for identifying HFrEF pa- because there are no guidelines for the
tients with a recovered LVEF. management of patients with HFrecEF.
CRT = cardiac
resynchronization therapy Although increases in LVEF may occur

A working definition of HFrecEF that is
DCM = dilated cardiomyopathy “spontaneously” in some forms of dilated
consistent with the majority of studies in
ECG = electrocardiogram
cardiomyopathy (DCM), the changes gener-
the literature includes the following: 1)
ally occur in the setting of the use of
GDMT = guideline-directed
documentation of a decreased
medical therapy guideline-directed medical and device ther-
LVEF <40% at baseline; 2) $10% abso-
HF = heart failure apy (1). Moreover, it is usually not possible
lute improvement in LVEF; and 3) a sec-
HFmrEF = heart failure mid-
to clearly discern the spontaneous compo-
ond measurement of LVEF >40%.
range ejection fraction nent of the improvement in myocardial
HFpEF = heart failure with function, as most patients are treated with
Guideline-directed medical and device
preserved ejection fraction GDMT. It is important to recognize that the therapy for patients with HFrecEF should
HFrecEF = heart failure with subgroup of HFrEF patients with a recov- be continued indefinitely until the
recovered ejection fraction
ered LVEF are clinically distinct from pa- biology and clinical epidemiology of
HFrEF = heart failure with tients with heart failure with a preserved EF HFrecEF is better understood.
reduced ejection fraction
(HFpEF), who also have an LVEF >50%
ICD = implantable-cardioverter
HFrecEF patients should have close clin-
along with the presence of HF signs and
defibrillator
ical follow-up due to the high risk of
symptoms (2,3). The issue of modest recov-
LVEF = left ventricular ejection heart failure relapse.
fraction ery of LVEF with a resultant LVEF that falls
NT-proBNP = N-terminal pro–
between 40% and 50% has resulted in
B-type natriuretic peptide the proliferation of new nomenclatures,
SCD = sudden cardiac death including HF improved EF (4), HFpEF,
borderline HFpEF (1), HF recovered EF
BIOLOGY OF REVERSE REMODELING AND
(HFrecEF) (3,5), and HF mid-range EF (HFmrEF) (6).
RECOVERY OF VENTRICULAR FUNCTION
In the absence of prior documentation of LVEF, one
can only speculate as to whether LVEF is either ris-
Insofar as the calculation of LVEF incorporates LV
ing or falling in this group of patients, which em-
end-diastolic volume in the denominator of the
phasizes the need to follow the LVEF trajectory in
equation, improvements in LVEF are associated with
this group of patients by performing serial assess-
a reciprocal decrease in LV end-diastolic volume,
ments of LVEF over time. Indeed, previous studies
which has been referred to as reverse LV remodeling.
have shown that patients with mid-range LVEF be-
In the context of the present discussion, reverse LV
tween 40% and 50% (HFmrEF) are heterogeneous in
remodeling refers to the restoration of more normal
nature and represent an admixture of HFrEF pa-
cardiac myocyte size and LV chamber geometry,
tients with improved LVEF and patients with HFpEF
resulting in a leftward shift of the end-diastolic
whose LVEF has declined (7,8). Although HFmrEF is
pressure volume relationship toward normal values
endorsed by the European Society of Cardiology (6)
(9,10). Importantly, reverse LV remodeling is associ-
as a new category of HF, we suggest that, absent
ated with improved myocyte contractility and
knowledge of the LVEF trajectory, HFmrEF patients
improved LV chamber contractility (11–13). It is
should be considered neither biologically nor clini-
important to recognize that reverse LV remodeling is
cally synonymous with HFrEF patients with a
associated with fewer HF hospitalizations and
recovered LVEF.
decreased cardiovascular mortality, and that there is
Given the complexity and heterogeneity of pa-
a direct correlation between the extent of reverse LV
tients with heart failure with a recovered EF
remodeling and the improvement in cardiac survival
(HFrecEF), it is not surprising that there is little or
(14).
no consensus with respect to how to define, di-
agnose, and manage this growing population of HF MECHANISMS OF REVERSE LV REMODELING.
patients. Accordingly, the current JACC Scientific Although the biological basis for reverse LV remod-
Expert Panel reviews the biology and clinical course eling and recovery of LV function is incompletely
of HFrecEF patients, as well as provides guidelines understood, several general concepts have emerged.
for defining, diagnosing, and managing this group The most prominent theme is that cardiac remodeling
of patients. is a dynamic process that occurs in a bidirectional
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 721
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

F I G U R E 1 Changes in LVEF With GDMT in Patients With Heart Failure With a Reduced EF

Full Partial
Recovery (EF >50%) Recovery (EF 40%-50%)

LV Reverse LV
Remodeling Remodeling
Reverse LV
Remodeling

Medical
and Device
Therapy

No Functional
Heart Failure Recovery (EF <40%)
(EF <40%)

Patients with heart failure with recovered ejection fraction (HFrecEF) treated with guideline-directed medical and device therapies (GDMT)
may have a complete recovery of left ventricular ejection fraction (LVEF) >50%, partial recovery of LVEF (EF 40% to 50%), or no functional
recovery of LVEF (EF <40%).

manner (i.e., forward and reverse) and that cardiac that occur within the myocardial extracellular matrix
remodeling involves the coordinated regulation of (15). A second important theme, which has direct
multiple molecular and cellular changes that bearing on the concept of myocardial remission (dis-
contribute to phenotypic changes in the size, shape, cussed later), is that many of the multilevel molecular
and function of the heart. Basic and clinical studies changes that occur during forward LV remodeling
have consistently shown that many of the cellular and remain dysregulated in reverse remodeled hearts,
anatomic changes that occur during forward LV despite improvements in structural and functional
remodeling revert toward the normal less pathologic abnormalities. Transcriptional profiling of reverse
phenotype during reverse LV remodeling. A complete remodeled hearts reveals the emergence of new sets
description of the molecular changes that occur dur- of genes that belong to ontogenies that are not
ing reverse LV remodeling has been the subject of expressed in nonfailing hearts (19). Viewed together,
several reviews (15–17). these findings suggest that reverse LV remodeling is
Table 1 shows that pharmacological and device not simply a mirror image of the molecular and
therapies that lead to reverse LV remodeling are cellular pathways that become dysregulated during
accompanied by salutary changes in the biology of the forward LV remodeling but rather that reverse LV
cardiac myocyte, the composition of the extracellular remodeling represents a coordinated multilevel pro-
matrix, and the chamber properties of the left cess that allows the heart to adopt a new, less path-
ventricle. Moreover, studies that have examined se- ological steady state that is associated with enhanced
rial changes in gene expression during reverse LV pump function and improved clinical prognosis.
remodeling have shown that the normalization of Given the multilevel adaptations associated with
gene transcription related to myocyte contractility reverse remodeling and the diversity of clinical con-
occurs before changes in genes related to the extra- texts in which it occurs, it is reasonable to speculate
cellular matrix, suggesting that return of myocyte about whether there are primary drivers of reverse
function is required for reversal of the changes in LV remodeling that enable the activation of other sec-
geometry in the failing heart (18). In addition to ondary downstream processes. In this regard, there is
changes in adult cardiac myocytes during reverse LV strong evidence supporting the concept of biome-
remodeling, there are a number of important changes chanical load as one such primary driver (20). The
722 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

T A B L E 1 Cellular and Molecular Determinants of Recovery of LV Function

Aldosterone
Beta-Blocker ACE Inhibitor ARB Antagonists LVAD CRT CSD

Myocyte defects
Hypertrophy Decreased Decreased Decreased Decreased Decreased Decreased Decreased
Fetal gene expression Decreased Decreased Decreased ND Decreased Decreased Decreased
Myocytolysis Decreased ND ND ND Decreased ND ND
Beta-adrenergic desensitization Decreased Decreased Decreased ND Decreased Decreased Decreased
EC coupling Increased Increased Increased ND Increased Increased Increased
Cytoskeletal proteins ND ND ND Increased Increased ND Increased
Myocardial defects
Myocyte apoptosis Decreased Decreased Decreased ND Decreased Decreased Decreased
MMP activation Decreased Decreased Decreased Decreased Decreased Decreased Decreased
Fibrosis Decreased Decreased Decreased Decreased Increased* Decreased Decreased
Angiogenesis Increased Increased Increased Increased Decreased Increased Increased
LV dilation Decreased Stabilized Stabilized Stabilized Decreased Decreased Decreased

Reproduced with permission from Mann et al. (17)

ACE ¼ angiotensin-converting enzyme; ARB ¼ angiotensin receptor blocker; CRT ¼ cardiac resynchronization therapy; CSD ¼ cardiac support device; EC ¼ excitation-
contraction; LV ¼ left ventricular; LVAD ¼ left ventricular assist device; MMP ¼ matrix metalloproteinase; ND ¼ not done.

pathological hypertrophy resulting from a discrete highest rates of recovery of LV function have been
process such as aortic stenosis ultimately entrains a associated with amelioration of adverse metabolic or
wide variety of pathological signaling processes that energetic circumstances known to compromise car-
contribute to the resultant structural and functional diac function such as chronic tachycardia, hyperthy-
abnormalities; myocardial unloading likewise invokes roidism, and hypothyroidism (16,23). The second
varied and potent reverse LV remodeling signaling highest rates of recovery of LV function have been
pathways. This is particularly evident in studies associated with dilated cardiomyopathies that are
of hearts supported by a LV assist device (19,21) associated with immune responses, such as peri-
but also apparent after cardiac resynchronization partum cardiomyopathy, acute lymphocytic myocar-
therapy (CRT) (22), in which decreased cellular and ditis, and the systemic inflammatory response
chamber hypertrophy are accompanied by changes in syndrome. Recovery of LV function has also been
gene expression that regulate different functional associated with the discontinuation of cardiotoxins,
domains of the cell; these include the sarcomere, most commonly ethanol and cancer chemotherapies,
ß-adrenergic signaling, excitation contraction including anthracyclines, tyrosine kinase inhibitors,
coupling, metabolism, and the cytoskeleton. and monoclonal antibodies (16). There also seem to be
differences in recovery of LV function in DCM among
EPIDEMIOLOGY OF REVERSE women and men, as well as White and Black patients,
LV REMODELING AND RECOVERY with greater recovery of LV function and event-free
OF LV FUNCTION survival in women and White patients (24).
A large body of evidence shows reverse LV
Reverse LV remodeling with recovery of LV function remodeling and recovery of LV function after imple-
can occur spontaneously in a variety of different mentation of evidence-based medical, device-based,
clinical settings (Figure 2). These clinical observations and surgical interventions in patients with chronic
provide clues to the underlying biology of reverse LV HFrEF. Because this topic has been reviewed exten-
remodeling. Remarkably, recovery of LV function sively elsewhere (16,25), it is discussed here only
occurs in a significant proportion of individuals even briefly. Numerous studies in patients with ischemic
when the severity of HF or cardiac dysfunction is heart disease have shown that there is substantial
severe. Spontaneous recovery of LV function most potential for reverse LV remodeling after coronary
commonly occurs after resolution of the inciting artery revascularization (26). Although the role of
stress that compromised myocardial function. As myocardial viability testing in this setting remains
shown in Figure 2, 3 major etiologies of myocardial uncertain, multiple studies have reported that there
injury are associated with spontaneous recovery of is a greater likelihood of improved LV function,
LV function and reverse LV remodeling: abnormal functional capacity, and survival after revasculariza-
energetics, toxic insults, and inflammation. The tion when viable myocardial segments are present.
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 723
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

F I G U R E 2 Recovery of LVEF in Clinical Settings

Toxic Pathological
Insults Immune
Responses

Alcohol Peri-
partum
Chemo- ALM
therapy
Beta-
Tachycardia Blockade Excess
Neurohormonal
Thyroid Activation
Reverse CPAP
Abnormal Disease
Remodeling
Energetics RAAS
Revascularization Blockade

CRT LVAD

CSD MVR
AVR

Excess
Hemodynamic
Load

The segments of the outmost ring highlight pathophysiological processes implicated by reverse left ventricular remodeling, in particular
clinical settings that comprise the middle ring. Reproduced with permission from Hellawell et al. (16). ALM ¼ acute lymphocytic myocarditis;
AVR ¼ aortic valve replacement; CPAP ¼ continuous positive airway pressure; CRT ¼ cardiac resynchronization therapy; CSD ¼ cardiac support
device; LVAD ¼ left ventricular assist device; LVEF ¼ left ventricular ejection fraction; MVR ¼ mitral valve repair/replacement; RAAS ¼ renin-
angiotensin-aldosterone system.

Pharmacological inhibition of the sympathetic ner- remodeling and improve LV function in patients with
vous system and the renin-angiotensin-aldosterone HFrEF (27). Viewed together, these studies implicate
system have also been associated with reverse LV adrenergic and renin-angiotensin-aldosterone system
remodeling, improved LV function, and improved signaling in the pathogenesis of forward LV remod-
clinical outcomes. Among the different evidence- eling and indicate that mitigating these mechanisms
based neurohormonal antagonists, the use of beta- favors reverse LV remodeling. Table 2 summarizes
adrenergic blockers are most strongly associated established clinical predictors of reverse remodeling.
with reverse LV remodeling. Although there is sub- The prevalence of HFrecEF has been gleaned pri-
stantial evidence that antagonism of the renin- marily from retrospective single-center reports or
angiotensin-aldosterone system prevents forward LV aggregated from a few centers with research data-
remodeling, the evidence for regression of established bases. In one of the first reports to carefully charac-
LV remodeling is less definitive for angiotensin- terize such patients, Punnoose et al. (28) noted that
converting enzyme inhibitors and aldosterone antag- the prevalence of heart failure with a recovered LVEF
onists (25). However, treatment with angiotensin re- was 34% in their center; notably, 70% of patients
ceptor blockers is associated with significant labeled as HFpEF had documentation of prior HFrEF.
reductions in LV internal diastolic diameter and in- Patients with HFrecEF with a baseline LVEF <40%
creases in LVEF (4). More recently, treatment with that improved to an LVEF >40% were younger, less
sacubitril/valsartan was shown to induce reverse LV likely to have coronary artery disease, and had fewer
724 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

Larger, more contemporary series have generally

T A B L E 2 Predicting Reverse LV Remodeling Among Patients With HFrEF
confirmed these early observations and provided
Predictors of Reverse LV Remodeling additional insights (30). In a prospective consecutive
Clinical parameters Nonischemic etiology series of 1,057 patients from a university program in
Lower duration of HF
Spain with baseline and follow-up echocardiograms
Female
at 1 year, one-quarter of patients with a baseline
No LBBB
LBBB in CRT
LVEF <45% experienced an average increase of 21%
Genetic factors Pathogenic gene variants not involving structural in their LVEF. In addition to female sex, nonischemic
cytoskeletal proteins or Z-disk proteins etiology, and younger age, the investigators found
Echocardiography/CMR Lower LVEF, greater contractility on strain imaging that shorter duration of HF and absence of left bundle
imaging Greater LV diameters
LGE absence branch block predicted reverse LV remodeling and
Biomarkers Lower NT-proBNP recovery of LVEF (31). In a cohort of ambulatory pa-
Lower troponin
Lower sST2
tients studied at Emory University, w16% of patients
Galectin-3, emerging biomarkers (mimecan, microRNAs, orexin) with a baseline LVEF <50% recovered their LVEF
>50% on GDMT. Patients with a recovered LVEF
Modified with permission from Aimo et al. (58).
CRT ¼ cardiac resynchronization therapy; HF ¼ heart failure; HFrEF ¼ heart failure with a reduced ejection
>50% had a decreased risk of HF hospitalizations, as
fraction; LBBB ¼ left bundle branch block; LGE ¼ late gadolinium enhancement; LVEF ¼ left ventricular ejection well as all-cause and cardiovascular mortality
fraction; NT-proBNP ¼ N-terminal pro–B-type natriuretic peptide; sST2 ¼ soluble ST (suppression of tumori-
genicity) 2. compared with HFpEF and HFrEF cohorts (5). In the
Heart Muscle Disease Registry of Trieste (32), Merlo
et al. reported that 38 (9%) of 408 DCM patients
comorbidities. Basuray et al. (3) extended these pre- recovered their LVEF >50% and normalized their LV
liminary findings in the 3-center collaborative Penn end-diastolic dimension on GDMT. Importantly,
Heart Study, which followed up HF patients for w40% of this subgroup experienced a subsequent
almost 9 years. These authors defined HFrecEF as an decline in LVEF, and 5% required heart trans-
LVEF >50% with prior documentation of an plantation or died after 15  4.7 years of follow-up.
LVEF <50%. Of the entire cohort, only 10% (n ¼ 176 of Clinical trial data provide important insights into
1,821) were classified as HFrecEF, with an average recovery of LVEF on GDMT insofar as the data are
improvement in LVEF of 28% over a mean of drawn from multiple centers in a prospective manner
29 months. Similar to the report by Punnoose et al. and are therefore less subject to survival bias than
(28), the HFrecEF patients were younger and had less single-center studies. In the Val-HeFT (Valsartan
coronary disease with fewer comorbidities and Heart Failure Trial), 9% of patients (n ¼ 3,517) with an
symptoms. Importantly, HFrecEF patients were on LVEF <35% had an improved LVEF >40% at 1 year (4).
comparable GDMT compared with HFrEF patients. Seven variables (male sex [odds ratio (OR): 0.69],
The hazard ratio for all-cause death, cardiac trans- ischemic origin [OR: 0.41], body mass index [OR:
plantation, or left ventricular assist device placement 0.96], diastolic blood pressure [OR: 1.01], left ven-
in the HFrEF group versus the HFrecEF group was 4.1 tricular internal diameter/body surface area [OR: 0.5],
(95% confidence interval: 2.4 to 6.8; p < 0.001), baseline beta-blocker therapy [OR: 1.9], and valsartan
whereas the unadjusted hazard ratio for HFpEF treatment [OR: 1.5]) were independently associated
versus HFrecEF was 2.3 (95% confidence interval: 1.2 with HFrecEF. However, even in the presence of all 7
to 4.5; p ¼ 0.013). Intriguingly, the patients with factors, only a few of the patients had substantial
HFrecEF had a risk of HF hospitalizations that was recovery in LV function (median probability of
similar to that of patients with HFpEF (1.3; 95% con- HFrecEF: 0.15 [95% CI: 0.08 to 0.22]; area under the
fidence interval: 0.9 to 2.0; p ¼ 0.15), as well as curve: 0.76 [95% confidence interval: 0.72 to 0.79]).
persistent biomarker evidence of inflammatory, Dramatic responses among some HFrEF patients to
neurohormonal, and myocardial injury. These find- cardiac resynchronization therapy (CRT), so-called
ings were also confirmed in the prospective registry super-responders, can also provide insight into
to Improve the Use of Evidence-Based Medical Ther- HFrecEF (33). Patients with nonischemic HF, very
apies in the Outpatient Setting (IMPROVE-HF), which wide QRS with left bundle branch block morphology,
enrolled 3,994 patients (29), in which almost one- female sex, and echocardiographic evidence of dys-
third (29%) of patients experienced a 10% absolute synchrony responded favorably to CRT. Notably, a
increase in LVEF (24.5% to 46.2%) over a 24-month small prospective randomized experience showed
follow-up. In this series, female sex, nonischemic that 78% of super-responders experienced a deterio-
etiology, and lack of myocardial infarction were ration in clinical and echocardiographic parameters
associated with improvement in LVEF. within 12 months after CRT deactivation (34).
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 725
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

C ENTR AL I LL U STRA T I O N Clinical Assessment of Patients With Recovered Left Ventricular Ejection Fraction

Wilcox, J.E. et al. J Am Coll Cardiol. 2020;76(6):719–34.

Clinical assessment of symptoms and electrocardiogram (ECG) can identify patients with heart failure with recovered ejection fraction (HFrecEF) at higher risk of
relapse. The need for persistent diuretics and left bundle branch block (LBBB) represents higher risk subgroups. Absence of late gadolinium enhancement (LGE) on
cardiac magnetic resonance imaging is a strong predictor of recovery or remodeling and is associated with improved prognosis. Elevated extracellular volume (ECV)
values (suggested of edema or fibrosis) can also improve diagnostic understanding of specific cardiomyopathies. Higher absolute global longitudinal strain (GLS) (e.g.,
>16%) is associated with stability of left ventricular ejection fraction over short-term follow-up and higher GLS even among dilated hearts is associated with
HFrecEF status. Prognosis in genetic dilated cardiomyopathy varies, with truncating variants of the titin gene (TTNtv) more likely to respond favorably to guideline-
directed medical therapy and achieve HFrecEF status, and LMNA mutations less likely to respond and confer high risk for sudden cardiac death despite HFrecEF status.
Greater reductions in N-terminal pro–B-type natriuretic peptide (NT-proBNP) with neurohormonal heart failure (HF) therapy is associated with greater improvements
in left ventricular structure and function, as well as improved clinical outcomes. A rise in NT-proBNP in HFrecEF patients may precede HF relapse. Gaps remain with
regard to development of inception cohorts to better understand the natural history of HFrecEF, and additional clinical trials are needed to define which elements of
clinical care are important for maintaining clinical remission, as well as basic studies to better define the biology of HFrecEF in order to develop new therapeutic targets.
Echo ¼ echocardiography; HFrEF ¼ heart failure with reduced ejection fraction; LVEDD ¼ left ventricular end diastolic dimension; MRI ¼ magnetic resonance imaging.
726 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

T A B L E 3 Recommended Interval Follow-Up for HFrecEF

Clinical Holter Echocardiography

Examination Monitoring NT-pro With Mechanics
Interval Follow-Up Time Period (After Meeting the HFrecEF Definition) and ECG (24 h) BNP (Strain) CMR

Every 6 months (until 12–18 months of HFrecEF). X X X

After w1 yr of “clinically stable” HFrecEF X*
Every 6–12 months (at minimum). X X
Optimal interval of echocardiography/imaging is unknown. It is X
reasonable clinical practice to assess durability every 1–3 yrs
after stable recovery depending on etiology.
Every 1–2 yrs for certain genetic cardiomyopathies at risk of atrial X
dysrhythmias (e.g., TTN).

*Consider CMR if one was not performed at de novo diagnosis of HFrEF.

CMR ¼ cardiac magnetic resonance; ECG ¼ electrocardiogram; HFrecEF ¼ heart failure with recovered ejection fraction; other abbreviations as in Table 2.

In summary, regardless of the definition of heart to maintain LV pump function under normal
HFrecEF, the preponderance of data suggests that conditions; however, this adaptation has less biolog-
younger age, female sex, nonischemic etiology, ical and contractile reserve capacity, and is therefore
shorter duration of disease, and fewer comorbidities more prone to redevelop LV dysfunction in response
are associated with higher likelihood of recovery of to hemodynamic, neurohormonal, or environmental
LVEF. Moreover, the clinical outcomes of HFrecEF stress. Although the precise biological motifs that are
patients are also improved compared with patients responsible for this loss of reserve capacity are not
with HFrEF and HFpEF with respect to death. How- known, it is likely that progressive loss of cardiac
ever, HF events still occur and symptoms remain myocytes, persistent dysregulation of the tran-
present in HFrecEF patients, although they may be scriptome, metabolome, and proteome of cardiac
less frequent than in patients with HFrEF (31). A myocytes, and the progressive erosion of the native
number of limitations of the available data should be 3-dimensional organization of the extracellular ma-
recognized, including survival bias, the imprecise trix surrounding the cardiac myocytes contribute to
nature of quantitative data in the clinical setting, the stability of the reverse LV remodeled heart (17).
missing data, treatment disparities, and variable This point of view is supported by the observation
clinical protocols with respect to surveillance testing that the great majority of clinical examples of spon-
and therapeutic management. taneous recovery of LV function associated with du-
rable clinical stability occur after transient injury
NATURAL HISTORY OF RECOVERED LVEF
(e.g., energetic defects or myocardial toxins), rather
than more long-standing and/or permanent injury
Although reverse LV remodeling and recovery of LV
(e.g., myocardial infarction, genetic abnormalities).
function are associated with improved clinical out-
comes, there is a growing body of evidence suggest- ETIOLOGY OF THE
ing that, even among patients who experience a CARDIOMYOPATHY MATTERS
complete normalization of LV structure and function
after implementation of GDMT, a significant propor- Understanding the pathophysiological basis for HF is
tion will develop recurrent LV dysfunction accom- essential for understanding the prognosis and man-
panied by recurrent HF events (3,32). The biological agement of patients with HFrecEF. Indeed, manage-
explanation for why some patients who have ment of specific types of cardiomyopathies is a
improved LV structure and function remain free from burgeoning field, especially in the cardio-oncology
HF events (“myocardial remission”) and why other arena, such as with trastuzumab-related ventricular
patients who have a similar improvement in LV dysfunction, and inflammatory responses to immune-
structure and function continue to have recurrent HF checkpoint inhibitors (35). Monoclonal antibodies
events is not known. One plausible explanation for that block ErbB2 (HER2/neu) signaling (e.g., trastu-
this phenomenon, which is based on the consistent zumab) disrupt cardiac homeostasis and myocardial
finding that the reverse remodeled heart retains repair and cause systolic and diastolic dysfunction. In
many of the molecular features of the failing heart, is a seminal paper, Narayan et al. (36) showed in a
that reverse LV remodeling represents a transition to prospective cohort of 277 patients with breast cancer
a new less pathological “steady state” that allows the that trastuzumab therapy resulted in early LVEF
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 727
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

F I G U R E 3 Sample Follow-Up and Clinical Testing Schedule for an HFrecEF Patient Deemed High-Risk for Recurrence of HF

Echo with mechanics

Minimum interval 3-6 months,

Echo with
recovery can occur up to >12
CMR* or mechanics every
months Echo 1-3 years

Variable 3 6 9 12 18 24 30
MONTHLY FOLLOW-UP SCHEDULE

Time Point 1) Clinical Exam 1) Clinical Exam

Zero: 2) ECG 2) ECG
HFrecEF 3) NT-proBNP 3) NT-proBNP
Diagnosis

Patients with heart failure with recovered ejection fraction (HFrecEF) at high risk of relapse (persistent left bundle branch block, genetic
dilated cardiomyopathy, higher biomarker profiles, or more comorbidities) require close clinical follow-up, with biomarker and imaging, with
shorter intervals immediately following HFrecEF diagnosis out to 12 months, and longer intervals thereafter. Abnormal or worsening global
longitudinal strain (GLS) may identify patients at higher risk of HF relapse. *Consider CMR if was not performed at de novo time of HFrEF
diagnosis. CMR ¼ cardiac magnetic resonance; ECG ¼ electrocardiogram; Echo ¼ echocardiography; HF ¼ heart failure; HFrEF ¼ heart
failure with a reduced ejection fraction; NT-proBNP ¼ N-terminal pro–B-type natriuretic peptide.

declines and incomplete recovery (e.g., persistent time of manuscript writing, the proportion of patients
subclinical dysfunction) at 3 years. The echocardio- with Takotsubo cardiomyopathy who will develop an
graphic parameters most consistently associated with HFrEF phenotype is unknown, nor is it clear whether
LVEF decline were LV volumes, longitudinal and these patients will benefit from continued GDMT.
circumferential strain, arterial load, and the ventric- Patients with alcohol-induced cardiomyopathy
ular arterial coupling ratio. often have significant LV reverse remodeling
Takotsubo cardiomyopathy has previously been following cessation of alcohol consumption (40). One
considered a transient period of profound LV observational study suggests improved durability of
dysfunction brought on by emotional distress that recovery over 15 years of follow-up among subjects
resolves with absolute recovery of LVEF. However, with alcohol-induced cardiomyopathy compared with
Scally et al. (37) recently highlighted that in fact, other HF etiologies; however, survival bias limits
despite a return to normal LVEF and normalization of interpretation. Ware et al. (41) have reported a shared
serum biomarkers, patients with previous Takotsubo genetic predisposition in alcohol-induced cardiomy-
cardiomyopathy experienced persistent reduced api- opathy and DCM, mostly due to truncating variants in
cal circumferential strain and reduced global longi- titin (TTN). Given the cohort data showing similar
tudinal strain on speckle-tracking echocardiography, prognosis compared with DCM (42), and other studies
as well as increased native T 1 mapping values on limited by survival bias, we recommend that medical
cardiac magnetic resonance (CMR). This persistent therapy for HF generally be continued, even in
subclinical cardiac dysfunction is also characterized HFrecEF patients with alcohol-induced
by a low-grade chronic inflammatory state with a cardiomyopathy.
myocardial macrophage inflammatory infiltrate and Fulminant myocarditis is a relatively uncommon
an increase in systemic proinflammatory cytokines syndrome that is associated with excellent 11-year
(38). Moreover, recent data from the InterTAK Reg- transplant-free survival of >90%, if patients survive
istry highlight the morbidity of Takotsubo cardiomy- the initial episode, which often requires acute me-
opathy, with long-term outcomes similar to those chanical support for cardiogenic shock (43). In a
with acute coronary syndromes (39). However, at the clinicopathological study from Johns Hopkins, a more
728 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

catastrophic clinical presentation was associated with includes: 1) documentation of a decreased

higher likelihood of complete myocardial recovery in LVEF <40% at baseline; 2) $10% absolute
survivors of the acute episode (44). The biology of the improvement in LVEF; and 3) a second measure-
fulminant presentation of myocarditis is likely ment of LVEF >40%. These improvements in LVEF
different from the biology of acute myocarditis are typically accompanied by a reduction in LV
without fulminant presentation, given that these 2 volumes.
conditions have dramatically different survival
Measurement of the changes in LVEF to ascertain
curves. In one study, nonfulminant acute myocarditis HFrecEF should be obtained at least 3 to 6 months
was associated with a 45% 11-year transplant free after the baseline LVEF, when the patient is stable
survival rate. In contrast, data from an international hemodynamically, to avoid acute changes in LVEF
registry of fulminant and nonfulminant myocarditis that are secondary to changes in heart rate or
revealed that cardiac death or transplantation loading conditions.
occurred more frequently among adults with fulmi-
nant lymphocytic myocarditis compared with those
MANAGEMENT OF PATIENTS WITH
with nonfulminant lymphocytic myocarditis at
RECOVERED CARDIOMYOPATHY
60 days (19.5% vs. 0%; p ¼ 0.005) and at 7 years of
follow-up (41.4% vs. 3.1%; p ¼ 0.0004) (45). Because
Despite an improvement in outcomes, patients with
of the conflicting nature of the long-term clinical
HFrecEF are still at risk for subsequent HF hospitali-
outcomes for fulminant and nonfulminant myocar-
zation and death relative to patients without HF and
ditis patients, we believe that it is prudent to
hence are not “normal” or truly cured from their HF
continue GDMT therapy for both of these sub-
(3,5,28). As discussed earlier, despite substantial
populations of HFrecEF patients, pending additional
reverse LV remodeling in certain patients, even with
long-term outcome data.
normalizing their LVEF and LV size, these improve-
WORKING DEFINITION OF HFrecEF ments most often represent myocardial “remission”
rather than a true cure of HF. Optimal clinical man-
Patients with HFrecEF represent a distinctly respon- agement of this significant population, albeit with a
sive subset of patients with HFrEF whose biological less severe phenotype, remains unclear due to lack of
substrate for HF and clinical course are distinct from robust prospective data. In fact, there has only been
patients with HFrEF and HFpEF. We propose that one randomized-controlled clinical trial (discussed
these patients be referred to as HFrecEF, to denote later) in 50 patients with nonischemic HFrecEF to
that they were initially HF patients with a remodeled assess the safety of weaning GDMT in this patient
(e.g., dilated) left ventricle. This terminology also population (46). Using best available evidence, puta-
avoids confusing these patients with patients with tive mechanistic insights, and clinical practice expe-
HFpEF who have an LVEF >50%, as well as with rience, we propose the following framework for
patients with an intermediate LVEF (40% to 50%), assessment, surveillance, and treatment that can help
which may represent HFpEF patients with deterio- inform the answers to these questions (Central
rating LVEF. As noted earlier, one of the major hur- Illustration).
dles toward our understanding of this unique group
CLINICAL EXAMINATION, SYMPTOMATOLOGY, AND
of patients is the lack of standardization with regard
ELECTROCARDIOGRAM. Jugular venous distension
to the definition of HFrecEF. A second limitation is
and signs of volume overload are particularly con-
that there are no current guidelines with respect to
cerning in HFrecEF. Patients with HFrecEF who still
how these patients should be followed up. Accord-
require loop diuretics for symptom relief may repre-
ingly, we propose the following general recommen-
sent a higher risk population, suggesting that they are
dations to guide clinical care:
at higher risk of recurrent HF events (i.e., relapse).

When identifying patients with HFrecEF, it is Moreover, persistent exertional dyspnea is common,
important to focus on the LVEF “trajectory” of the although usually less severe, and may reflect resting
patient, recognizing that changes in LVEF integrate or exertional increases in pulmonary capillary wedge
a number of different variables, including the na- pressure, chronotropic incompetence, pulmonary
ture and extent of myocardial injury, the degree hypertension, lack of systemic vasodilator reserve,
and duration of LV remodeling, and the type of and/or microvascular dysfunction.
therapy that is being initiated. The electrocardiogram (ECG) is also a cost-effective

A working definition of HFrecEF that is consistent modality to risk-stratify patients with HFrecEF.
with the majority of studies in the literature Although left bundle branch block is predictive of
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 729
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

CRT response, it is associated with lower likelihood of generally distinct mechanistic pathways of myocar-
improvement with GDMT alone (47). In addition, pa- dial injury and repair, biomarkers such as natriuretic
rameters of repolarization heterogeneity on surface peptides (ventricular remodeling), troponin (myocar-
ECG (e.g., QRST angle, QT dispersion) are associated dial injury), ST2 (inflammation), and galectin-3
with myocardial recovery among patients with acute (fibrosis) can provide independent additive
HF and nonischemic cardiomyopathy (48). In the prognostic information in HFrEF (58). Relevant to
absence of a complete normalization of the ECG, it HFrecEF, a greater reduction in N-terminal pro–B-
should be assumed that myocardial disease remains type natriuretic peptide (NT-proBNP) with GDMT is
present. associated with greater improvements in LVEF and
FAMILY HISTORY OF DCM AND EVALUATION OF greater reduction in LV volumes as well as improved
UNDERLYING GENETIC RISK. A thorough 3-generation clinical outcomes (59). A subsequent study showed
family history is always recommended among pa- that multiple biomarkers are associated with im-
tients with nonischemic DCM, regardless of the LVEF provements in LVEF (60). Because there is a risk of
or clinical trajectory, insofar as there are implications recurrent LV dysfunction in HFrecEF (61,62), consid-
for the patient’s prognosis and recovery as well as eration should also be given to performing serial
offspring and other first-degree relatives. All measurements of specific biomarkers over time, in
first-degree relatives of a DCM proband are recom- conjunction with regular surveillance provider visits
mended to undergo clinical screening with echocar- and selected imaging modalities.
diography, ECG, and clinical examination (49). In 2-DIMENSIONAL ECHOCARDIOGRAPHY. In addition
addition, we recommend that clinicians should have to clinical status, responder status to GDMT is
low threshold to perform genetic testing even among ultimately determined by 2-dimensional echocardi-
those with recovered DCM, because the results may ography. Specific echocardiographic features are
inform prognosis about durability of recovery, risk for characteristic of HFrecEF, including decreases in LV
HF relapse or sudden death, and risk for atrial ar- end-systolic and end-diastolic volumes, improve-
rhythmias, in addition to informing HF risk for family ments in functional mitral regurgitation, and the lack
members. Herman et al. (50) published seminal find- of right ventricular dysfunction (27,63). Despite im-
ings in 2012 that 15% to 25% of cases of DCM were provements in gross myocardial functioning, global
associated with rare truncating TTN variants. In longitudinal strain and diastolic function rarely
addition, Ware et al. (51) have shown that 15% of normalize in HFrecEF (62). However, among HFrecEF
women with peripartum cardiomyopathy carried patients, higher global longitudinal strain (e.g., >16%
truncations in TTN. Titin is the largest protein in the absolute global longitudinal strain) is associated with
heart, spanning the length of the sarcomere, from the stability of LVEF over short-term follow-up (w2
Z disc to the M band, and it functions as a molecular years). Swat et al. (64) further showed that higher
spring that regulates contraction. In contrast to other baseline absolute longitudinal strain (e.g., >8%) is
known genetic etiologies of DCM, truncating variants associated with HFrecEF status in a retrospective
in TTN are compatible with recovery after exposure to cohort of DCM patients with acute decompensated
GDMT (52,53) and left ventricular assist device HF, even among patients with larger left ventricu-
unloading; however, it should be emphasized that the lar dimensions.
sustainability of these improvements are not CMR IMAGING. CMR is best utilized to characterize
well characterized. the myocardial substrate around the time of de novo
Another important consideration in the care of the diagnosis of HFrEF to provide insights into etiology.
HFrecEF patient is determining the risk of sudden For example, the pattern of late gadolinium
cardiac death (SCD), a risk that still persists in certain enhancement can be very suggestive of specific car-
underlying genetic cardiomyopathies, despite recov- diomyopathies, such as sarcoidosis and certain
ery of LVEF. For example, pathogenic mutations in muscular dystrophies (65,66). The presence and
DSP, SCN5A, LMNA, and FLNC carry a significantly extent of late gadolinium enhancement are also pre-
higher risk of malignant arrhythmias compared with dictors of response to neurohormonal HF therapy and
other genetic and nongenetic etiologies of DCM risk prediction in both ischemic (67) and nonischemic
(54–57) despite therapy with GDMT and improve- (68,69) conditions. In nonischemic cardiomyopathy,
ments in LVEF. the absence of late gadolinium enhancement is a
BIOMARKERS. Circulating biomarkers are influenced strong predictor of recovery or remodeling and is
by intrinsic myocardial properties as well as periph- associated with improved prognosis (70). The use of
eral and local metabolic factors. Because they reflect T 1 mapping to measure extracellular volume and
730 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

interstitial fibrosis (71) shows promise as another and Morbidity in Heart Failure) and PARAGON-HF
predictor of response to therapy and improved prog- (Prospective Comparison of ARNI with ARB Global
nostication in HF (72). However, the utility of CMR Outcomes in HF With Preserved Ejection Fraction)
after some degree of LV remodeling or recovery has trials revealed that the therapeutic effects of
occurred is largely unknown. sacubitril/valsartan, compared with a RAS inhibitor
CLINICAL SCENARIOS. Within the framework of the alone, seem to extend to patients with HF and
information and the limitations discussed here, we mildly reduced EF (74). It is likely that most
propose the following answers to frequently asked patients with HFmrEF in current clinical practice
clinical questions in managing patients with HFrecEF. have HFrecEF, which likely explains why there is a
1. Can any or all of the HF medications be therapeutic benefit for continued neurohormonal
stopped in HFrecEF? Is there a “signature” for blockade in HFmrEF (74,75). In our experience, if
c u r e ? In the open-label randomized pilot TRED-HF there is uncertainty around whether to continue
(Withdrawal of Pharmacological Treatment for GDMT in a particular patient, medications should be
Heart Failure in Patients With Recovered Dilated continued because of the clinical observation that,
Cardiomyopathy) trial, the investigators tested the among patients who experience a relapse and
hypothesis that GDMT could be withdrawn in recurrent decline in LVEF, there is a higher
asymptomatic HFrEF patients if the LVEF rose likelihood of recurring myocyte injury and a
from <40% to >50%, the left ventricular end-diastolic diminished ability to recover LVEF the second time
volume normalized, and NT-proBNP was <250 ng/l around.
after treatment (46). After screening 936 patients, 51
patients were randomized to either a phased with- 2. How should these patients be followed up?
drawal protocol of the HF GDMT or continued therapy What should be the frequency of follow-up?
with GDMT; the participants initially randomized to Table 3 summarizes our consensus recommendation
continued therapy also had their medications for interval testing to guide management in HFrecEF
weaned. Within 6 months, 11 (44%) of 25 from the first patients. Once patients with HFrecEF are deemed
withdrawal group and 9 (36%) of 25 from the second “stable” for at least 1 year by the treating provider, we
group experienced a recurrence of HF, defined by a recommend that they should be seen every 6 months
fall in LVEF >10% to <50%, an increase in left ven- for at least 3 years, then every year at minimum due
tricular end-diastolic volume >10% to greater than to the risk of relapse and HF hospitalization
the normal range, a doubling of the NT-proBNP to (3,5,28,76). Such a practice ensures regular laboratory
>400 ng/l, or clinical evidence of HF. Importantly, surveillance (e.g., biomarker data), review of symp-
there were no deaths. Based on this single random- toms and signs of HF, and to encourage compliance
ized trial and clinical reports reviewed herein, we with polypharmacy, a well-known limitation in
recommend not stopping GDMT in patients with chronic conditions, particularly when such conditions
HFrecEF unless there are mitigating circumstances. In are asymptomatic.
this regard it is notable that in TRED-HF, the recur- Follow-up imaging is clearly part of surveillance to
rence of HF occurred over months, rather than days or assess for durability of recovery, and the interval of
weeks, after weaning GDMT. Accordingly, if patients imaging is patient and risk dependent. The current
with HFrecEF stop neurohormonal antagonists for echocardiographic-appropriate use criteria do not
several days due an intercurrent illnesses or for some discuss this patient population (77). Based on available
other clinical reason, it is unlikely that they will data (acknowledging its limitations) and our clinical
redevelop recurrent HF in the immediate short term. experience, we suggest that echocardiography be
Cessation of diuretic agents is encouraged among performed annually for the first 2 years at a minimum
recovered-EF patients; indeed, the ability to tolerate to assess durability of recovery (62), or earlier if signs
the lack of diuretics may be indicative of a lower risk and/or symptoms of HF develop. Imaging intervals can
of recrudescent HF in HFrecEF (73). We recommend be lengthened after a period of stability; however, the
that if a patient with HFrecEF continues to require optimal frequency is unclear at the time of this writing.
diuretics, then further titration of GDMT (to target Given that the durability of LV function in HFrecEF is
doses) should be considered. In addition, consider- uncertain, we recommend that patients should be
ation should be given to substituting an angiotensin imaged at least every 3 to 5 years until further longi-
receptor neprilysin inhibitor for an angiotensin- tudinal data can be obtained on this patient popula-
converting enzyme inhibitor. Recent pooled ana- tion. Patients at higher risk (e.g., those with persistent
lyses of the PARADIGM-HF (Prospective Comparison left bundle branch block, genetic DCM, higher
of ARNI with ACEI to Determine Global Mortality biomarker profiles, more comorbidities) may be
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 731
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

selected for a shorter interval of imaging follow-up. In the best evidence to date, the data would sup-
addition, patients with long-term HFrecEF may port ICD generator change for most patients with
relapse due to a newly acquired condition, such as HFrecEF (85), especially if a deleterious genetic
coronary disease, or new onset of atrial fibrillation that mutation associated with high arrhythmia risk is
was not present when diagnosed at an earlier time present, history of appropriate shocks is docu-
point or the appearance of a new arrhythmia. HFrecEF mented, or the ECG remains abnormal. As a rule,
patients should be vigilantly followed up as described CRT should be maintained because electrical
earlier, and clinicians should recognize that not all dyssynchrony and forward LV remodeling are
recrudescent HF is “failure of GDMT,” and new con- known to recur with loss of resynchronization
current cardiac conditions may be an underlying (33,86).
explanation. Figure 3 shows an example of a follow-up
GAPS IN KNOWLEDGE/FUTURE PERSPECTIVES
schedule for an HFrecEF patient deemed to be higher
risk for HF relapse.
We have learned a great deal about the biology,
3 . Is implantable-cardioverter defibrillator generator
epidemiology, clinical predictors, and outcomes of
change indicated in HFrecEF patients? It is unclear if
reverse LV remodeling and recovery of LV function
the subset of HFrecEF patients who had an
over the past 2 decades. Unfortunately, far less is
implantable-cardioverter defibrillator (ICD) placed for
known about the biology, natural history, and the
primary prevention of SCD when the LVEF was #35%
long-term clinical outcomes for HFrecEF patients.
continue to benefit from ICD therapy after the LVEF
This knowledge gap represents a significant unmet
has improved. However, the greater duration of
clinical need, insofar as it is directly relevant to the
documented LV dysfunction may place the HFrecEF
reemergence of clinical HF in this seemingly stable
subgroup with an ICD in an inherently higher risk SCD
patient population. Until recently, these HF read-
category than an HFrecEF patient who recovered to an
missions have been ascribed to inadequate clinical
LVEF >35% to 40% within a relatively short time
care. As discussed in the current review, the rede-
frame, and therefore did not meet appropriate use
velopment of HF in patients with HFrecEF likely has
criteria for ICD implantation (78). A recent meta-
more to do with our lack of understanding with
analysis supports the notion that there is persistent
respect to how to manage HFrecEF than to inade-
arrhythmic risk among recovered EF patients, with a
quate clinical care.
3.3% per year rate of appropriate ICD therapy among
We believe that future research in this area will
those with LVEF $45% (79). An analysis of SCD HeFT
benefit from improved phenotyping of HFrEF pa-
(Sudden Cardiac Death in Heart Failure Trial) showed
tients to guide care, developing inception cohorts of
that patients who had an improvement in EF to >35%
HFrecEF patients to better understand the natural
during follow-up accrued a similar mortality benefit
history of HFrecEF and additional clinical trials to
with an ICD as those whose EF remained at #35% (80).
define which elements of clinical care are important
Device therapy in HFrecEF patients is not specif-
for maintaining clinical remission, as well as basic
ically addressed in current practice guidelines (81).
studies to better define the biology of HFrecEF. The
However, guidelines do mention that ICD therapy
goal is to develop new therapeutic targets that will
may be appropriate for patients who carry certain
enable patients with HFrecEF to experience a durable
pathogenic genetic mutations associated with high
remission from HF.
arrhythmia risk regardless of LVEF if clinical HF is
present (49). Because fatal arrhythmias may occur
despite normalization of LVEF, considerations for ADDRESS FOR CORRESPONDENCE: Dr. Jane E. Wil-
prophylactic placement of an ICD have been recom- cox, 676 North Saint Clair, Suite 600, Chicago,
mended independent of EF for specific mutations in Illinois 60611. E-mail: [email protected].
genes such as LMNA, SCN5A, and FLNC (82–84). Twitter: @WilcoxHeart. OR Dr. Douglas L. Mann,
There are no prospective trials of ICD therapy Campus Box 8086, 660 S. Euclid Avenue, St. Louis,
among HFrecEF populations; however, based on Missouri 63110. E-mail: [email protected].
732 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

REFERENCES

1. Yancy CW, Jessup M, Bozkurt B, et al. 2013 14. Kramer DG, Trikalinos TA, Kent DM, with recovered ejection fraction: a distinct clinical
ACCF/AHA guideline for the management of heart Antonopoulos GV, Konstam MA, Udelson JE. entity. J Card Fail 2011;17:527–32.
failure: a report of the American College of Car- Quantitative evaluation of drug or device effects
29. Wilcox JE, Fonarow GC, Yancy CW, et al.
diology Foundation/American Heart Association on ventricular remodeling as predictors of thera-
Factors associated with improvement in ejection
Task Force on Practice Guidelines. J Am Coll Car- peutic effects on mortality in patients with heart
fraction in clinical practice among patients with
diol 2013;62:e147–239. failure and reduced ejection fraction: a meta-
heart failure: findings from IMPROVE HF. Am
analytic approach. J Am Coll Cardiol 2010;56:
2. Wilcox JE, Yancy CW. Heart failure—a new Heart J 2012;163:49–56.e2.
392–406.
phenotype emerges. JAMA Cardiol 2016;1:507–9.
30. Ghimire A, Fine N, Ezekowitz JA, Howlett J,
15. Kim GH, Uriel N, Burkhoff D. Reverse remod-
3. Basuray A, French B, Ky B, et al. Heart failure Youngson E, McAlister FA. Frequency, predictors,
elling and myocardial recovery in heart failure. Nat
with recovered ejection fraction: clinical descrip- and prognosis of ejection fraction improvement in
Rev Cardiol 2018;15:83–96.
tion, biomarkers, and outcomes. Circulation 2014; heart failure: an echocardiogram-based registry
129:2380–7. 16. Hellawell JL, Margulies KB. Myocardial reverse study. Eur Heart J 2019;40:2110–7.
remodeling. Cardiovasc Ther 2012;20:172–81. 31. Lupon J, Diez-Lopez C, de Antonio M, et al.
4. Florea VG, Rector TS, Anand IS, Cohn JN. Heart
failure with improved ejection fraction: clinical 17. Mann DL, Barger PM, Burkhoff D. Myocardial Recovered heart failure with reduced ejection
characteristics, correlates of recovery, and sur- recovery: myth, magic or molecular target? J Am fraction and outcomes: a prospective study. Eur J
vival: results from the Valsartan Heart Failure Coll Cardiol 2012;60:2465–72. Heart Fail 2017;19:1615–23.
Trial. Circ Heart Fail 2016;9.
18. Weinheimer CJ, Kovacs A, Evans S, 32. Merlo M, Stolfo D, Anzini M, et al. Persistent
5. Kalogeropoulos AP, Fonarow GC, Matkovich SJ, Barger PM, Mann DL. Load-depen- recovery of normal left ventricular function and
Georgiopoulou V, et al. Characteristics and out- dent changes in left ventricular structure and dimension in idiopathic dilated cardiomyopathy
comes of adult outpatients with heart failure and function in a pathophysiologically relevant murine during long-term follow-up: does real healing
improved or recovered ejection fraction. JAMA model of reversible heart failure. Circ Heart Fail exist? J Am Heart Assoc 2015;4:e001504.
Cardiol 2016;1:510–8. 2018;11:e004351. 33. Proclemer A, Muser D, Facchin D. What we can
6. Ponikowski P, Voors AA, Anker SD, et al. 2016 19. Margulies KB, Matiwala S, Cornejo C, Olsen H, learn from “super-responders”. Heart Fail Clin
ESC guidelines for the diagnosis and treatment of Craven WA, Bednarik D. Mixed messages: tran- 2017;13:225–32.
acute and chronic heart failure: the Task Force for scription patterns in failing and recovering human 34. Cay S, Ozeke O, Ozcan F, Aras D, Topaloglu S.
the diagnosis and treatment of acute and chronic myocardium. Circ Res 2005;96:592–9. Mid-term clinical and echocardiographic evalua-
heart failure of the European Society of Cardiology tion of super responders with and without pacing:
20. Mann DL, Bristow MR. Mechanisms and
(ESC). Developed with the special contribution of the preliminary results of a prospective, random-
models in heart failure: the biomechanical model
the Heart Failure Association (HFA) of the ESC. Eur ized, single-centre study. Europace 2016;18:
and beyond. Circulation 2005;111:2837–49.
Heart J 2016;18:991–1075. 842–50.
21. Chen Y, Park S, Li Y, et al. Alterations of gene
7. Dunlay SM, Roger VL, Weston SA, Jiang R, 35. Salem JE, Manouchehri A, Bretagne M, et al.
expression in failing myocardium following left
Redfield MM. Longitudinal changes in ejection Cardiovascular toxicities associated with ibrutinib.
ventricular assist device support. Physiol Geno-
fraction in heart failure patients with preserved J Am Coll Cardiol 2019;74:1667–78.
mics 2003;14:251–60.
and reduced ejection fraction. Circ Heart Fail 2012;
5:720–6. 22. Barth AS, Chakir K, Kass DA, Tomaselli GF. 36. Narayan HK, Finkelman B, French B, et al.
Transcriptome, proteome, and metabolome in Detailed echocardiographic phenotyping in breast
8. Wilcox JE, Mann DL. Beta-blockers for the cancer patients: associations with ejection fraction
dyssynchronous heart failure and CRT.
treatment of heart failure with a mid-range ejec- decline, recovery, and heart failure symptoms over
J Cardiovasc Transl Res 2012;5:180–7.
tion fraction: deja-vu all over again? Eur Heart J 3 years of follow-up. Circulation 2017;135:
2018;39:36–8. 23. Givertz MM, Mann DL. Epidemiology and nat-
1397–412.
ural history of recovery of left ventricular function
9. Kass DA, Baughman KL, Pak PH, et al. Reverse 37. Scally C, Rudd A, Mezincescu A, et al. Persis-
in recent onset dilated cardiomyopathies. Curr
remodeling from cardiomyoplasty in human heart tent long-term structural, functional, and meta-
Heart Fail Rep 2013;10:321–30.
failure. External constraint versus active assist. bolic changes after stress-induced (Takotsubo)
Circulation 1995;91:2314–8. 24. McNamara DM, Starling RC, Cooper LT, et al.
cardiomyopathy. Circulation 2018;137:1039–48.
Clinical and demographic predictors of outcomes
10. Levin HR, Oz MC, Chen JM, Packer M, Rose EA, 38. Scally C, Abbas H, Ahearn T, et al. Myocardial
in recent onset dilated cardiomyopathy. Results of
Burkhoff D. Reversal of chronic ventricular dilation and systemic inflammation in acute stress-induced
the IMAC (Intervention in Myocarditis and Acute
in patients with end-stage cardiomyopathy by (Takotsubo) cardiomyopathy. Circulation 2019;
Cardiomyopathy)-2 Study. J Am Coll Cardiol 2011;
prolonged mechanical unloading. Circulation 139:1581–92.
58:1112–8.
1995;91:2717–20.
25. Saraon T, Katz SD. Reverse remodeling in 39. Ghadri JR, Kato K, Cammann VL, et al. Long-
11. Dipla K, Mattiello JA, Jeevanandam V, term prognosis of patients with Takotsubo syn-
systolic heart failure. Cardiol Rev 2015;23:173–81.
Houser SR, Margulies KB. Myocyte recovery after drome. J Am Coll Cardiol 2018;72:874–82.
mechanical circulatory support in humans with 26. Mrosek M, Labeit D, Witt S, et al. Molecular
determinants for the recruitment of the ubiquitin- 40. Djousse L, Gaziano JM. Alcohol consumption
end-stage heart failure. Circulation 1998;97:
ligase MuRF-1 onto M-line titin. FASEB J 2007;21: and heart failure: a systematic review. Curr Athe-
2316–22.
1383–92. roscler Rep 2008;10:117–20.
12. Hutchinson KR, Guggilam A, Cismowski MJ,
41. Ware JS, Amor-Salamanca A, Tayal U, et al.
et al. Temporal pattern of left ventricular struc- 27. Januzzi JL Jr., Prescott MF, Butler J, et al.
Genetic etiology for alcohol-induced cardiac
tural and functional remodeling following reversal Association of change in N-terminal pro-B-type
toxicity. J Am Coll Cardiol 2018;71:2293–302.
of volume overload heart failure. J Appl Physiol natriuretic peptide following initiation of
(1985) 2011;111:1778–88. sacubitril-valsartan treatment with cardiac struc- 42. Fauchier L, Babuty D, Poret P, et al. Compar-
ture and function in patients with heart failure ison of long-term outcome of alcoholic and idio-
13. Topkara VK, Chambers KT, Yang KC, et al.
with reduced ejection fraction. JAMA 2019;322: pathic dilated cardiomyopathy. Eur Heart J 2000;
Functional significance of the discordance be-
1–11. 21:306–14.
tween transcriptional profile and left ventricular
structure/function during reverse remodeling. JCI 28. Punnoose LR, Givertz MM, Lewis EF, 43. McCarthy RE 3rd., Boehmer JP, Hruban RH,
Insight 2016;1:e86038. Pratibhu P, Stevenson LW, Desai AS. Heart failure et al. Long-term outcome of fulminant
JACC VOL. 76, NO. 6, 2020 Wilcox et al. 733
AUGUST 11, 2020:719–34 HFrecEF Consensus Recommendations

myocarditis as compared with acute (non- 60. Ky B, French B, Levy WC, et al. Multiple bio- 74. Solomon SD, Vaduganathan M, Claggett BL,
fulminant) myocarditis. N Engl J Med 2000;342: markers for risk prediction in chronic heart failure. et al. Sacubitril/valsartan across the spectrum of
690–5. Circ Heart Fail 2012;5:183–90. ejection fraction in heart failure. Circulation 2020;
141:352–61.
44. Lieberman EB, Herskowitz A, Rose NR, 61. Lupon J, Gavidia-Bovadilla G, Ferrer E, et al.
Baughman KL. A clinicopathologic description of Dynamic trajectories of left ventricular ejection 75. Cleland JGF, Bunting KV, Flather MD, et al.
myocarditis. Clin Immunol Immunopathol 1993; fraction in heart failure. J Am Coll Cardiol 2018;72: Beta-blockers for heart failure with reduced, mid-
68:191–6. 591–601. range, and preserved ejection fraction: an
individual patient-level analysis of double-blind
45. Ammirati E, Veronese G, Brambatti M, et al. 62. Adamo L, Perry A, Novak E, Makan M,
randomized trials. Eur Heart J 2018;39:26–35.
Fulminant versus acute nonfulminant myocarditis Lindman BR, Mann DL. Abnormal global longitu-
in patients with left ventricular systolic dysfunc- dinal strain predicts future deterioration of left 76. de Groote P, Fertin M, Pentiah AD,
tion. J Am Coll Cardiol 2019;74:299–311. ventricular function in heart failure patients with a Goéminne C, Lamblin N, Bauters C. Long-term
recovered left ventricular ejection fraction. Circ functional and clinical follow-up of heart failure
46. Halliday BP, Wassall R, Lota AS, et al. With-
Heart Fail 2017;10. patients with recovered left ventricular ejection
drawal of pharmacological treatment for heart
fraction after b-blocker therapy. Circ Heart Fail
failure in patients with recovered dilated cardio- 63. Tayal U, Prasad SK. Myocardial remodelling
2014;7:434–9.
myopathy (TRED-HF): an open-label, pilot, rand- and recovery in dilated cardiomyopathy. JRSM
omised trial. Lancet 2019;393:61–73. Cardiovasc Dis 2017;6:2048004017734476. 77. American College of Cardiology Foundation
Appropriate Use Criteria Task Force, American
47. Sze E, Samad Z, Dunning A, et al. Impaired 64. Swat SA, Cohen D, Shah SJ, et al. Baseline
Society of Echocardiography, American Heart As-
recovery of left ventricular function in patients longitudinal strain predicts recovery of left ven-
sociation, et al. ACCF/ASE/AHA/ASNC/HFSA/HRS/
with cardiomyopathy and left bundle branch tricular ejection fraction in hospitalized patients
SCAI/SCCM/SCCT/SCMR 2011 appropriate use
block. J Am Coll Cardiol 2018;71:306–17. with nonischemic cardiomyopathy. J Am Heart
criteria for echocardiography. A Report of the
Assoc 2018;7:e09841.
48. Prenner SB, Swat SA, Ng J, Baldridge A, American College of Cardiology Foundation
Wilcox JE. Parameters of repolarization hetero- 65. Ganigara M, Sharma B, Komalla RB, Vyas SY, Appropriate Use Criteria Task Force, American
geneity are associated with myocardial recovery in Mannam G, Rao NK. Unique pattern of late gado- Society of Echocardiography, American Heart As-
acute heart failure. Int J Cardiol 2020;301:147–51. linium enhancement on cardiac magnetic reso- sociation, American Society of Nuclear Cardiology,
49. Hershberger RE, Givertz MM, Ho CY, et al. nance imaging in Duchenne muscular dystrophy. Heart Failure Society of America, Heart Rhythm
Genetic evaluation of cardiomyopathy—a Ann Pediatr Cardiol 2016;9:190–1. Society, Society for Cardiovascular Angiography
Heart Failure Society of America Practice Guide- 66. Florian A, Ludwig A, Engelen M, et al. Left and Interventions, Society of Critical Care Medi-
line. J Card Fail 2018;24:281–302. ventricular systolic function and the pattern of cine, Society of Cardiovascular Computed To-
late-gadolinium-enhancement independently and mography, and Society for Cardiovascular
50. Herman DS, Lam L, Taylor MR, et al. Trunca-
additively predict adverse cardiac events in Magnetic Resonance Endorsed by the American
tions of titin causing dilated cardiomyopathy.
muscular dystrophy patients. J Cardiovasc Magn College of Chest Physicians. J Am Coll Cardiol
N Engl J Med 2012;366:619–28.
Reson 2014;16:81. 2011;57:1126–66.
51. Ware JS, Seidman JG, Arany Z. Shared genetic
67. Felker GM, Anstrom KJ, Adams KF, et al. Ef- 78. Russo AM, Stainback RF, Bailey SR, et al.
predisposition in peripartum and dilated cardio-
fect of natriuretic peptide-guided therapy on ACCF/HRS/AHA/ASE/HFSA/SCAI/SCCT/SCMR 2013
myopathies. N Engl J Med 2016;374:2601–2.
hospitalization or cardiovascular mortality in high- appropriate use criteria for implantable
52. Jansweijer JA, Nieuwhof K, Russo F, et al. risk patients with heart failure and reduced ejec- cardioverter-defibrillators and cardiac resynchro-
Truncating titin mutations are associated with a tion fraction: a randomized clinical trial. JAMA nization therapy: a report of the American College
mild and treatable form of dilated cardiomyopa- 2017;318:713–20. of Cardiology Foundation appropriate use criteria
thy. Eur J Heart Fail 2017;19:512–21. task force, Heart Rhythm Society, American Heart
68. Barison A, Aimo A, Ortalda A, et al. Late
53. Lim SY, Yuzhalin AE, Gordon-Weeks AN, Association, American Society of Echocardiogra-
gadolinium enhancement as a predictor of func-
Muschel RJ. Targeting the CCL2-CCR2 signaling phy, Heart Failure Society of America, Society for
tional recovery, need for defibrillator implantation
axis in cancer metastasis. Oncotarget 2016;7: Cardiovascular Angiography and Interventions,
and prognosis in non-ischemic dilated cardiomy-
28697–710. Society of Cardiovascular Computed Tomography,
opathy. Int J Cardiol 2018;250:195–200.
and Society for Cardiovascular Magnetic Reso-
54. Wilcox JE, Hershberger RE. Genetic cardio-
69. Gulati A, Jabbour A, Ismail TF, et al. Associa- nance. J Am Coll Cardiol 2013;61:1318–68.
myopathies. Curr Opin Cardiol 2018;33:354–62.
tion of fibrosis with mortality and sudden cardiac
79. Smer A, Saurav A, Azzouz MS, et al. Meta-
55. Gigli M, Merlo M, Graw SL, et al. Genetic risk of death in patients with nonischemic dilated car-
analysis of risk of ventricular arrhythmias after
arrhythmic phenotypes in patients with dilated diomyopathy. JAMA 2013;309:896–908.
improvement in left ventricular ejection fraction
cardiomyopathy. J Am Coll Cardiol 2019;74:
70. Masci PG, Schuurman R, Andrea B, et al. during follow-up in patients with primary pre-
1480–90.
Myocardial fibrosis as a key determinant of left vention implantable cardioverter defibrillators.
56. McNally EM, Mestroni L. Dilated cardiomyop- ventricular remodeling in idiopathic dilated car- Am J Cardiol 2017;120:279–86.
athy: genetic determinants and mechanisms. Circ diomyopathy: a contrast-enhanced cardiovascular
80. Adabag S, Patton KK, Buxton AE, et al. Asso-
Res 2017;121:731–48. magnetic study. Circ Cardiovasc Imaging 2013;6:
ciation of implantable cardioverter defibrillators
790–9.
57. Burke MA, Cook SA, Seidman JG, Seidman CE. with survival in patients with and without
Clinical and mechanistic insights into the genetics 71. Youn JC, Hong YJ, Lee HJ, et al. Contrast- improved ejection fraction: secondary analysis of
of cardiomyopathy. J Am Coll Cardiol 2016;68: enhanced T1 mapping-based extracellular volume the Sudden Cardiac Death in Heart Failure Trial.
2871–86. fraction independently predicts clinical outcome in JAMA Cardiol 2017;2:767–74.
patients with non-ischemic dilated cardiomyopa-
58. Aimo A, Gaggin HK, Barison A, Emdin M, 81. Al-Khatib SM, Stevenson WG, Ackerman MJ,
thy: a prospective cohort study. Eur Radiol 2017;
Januzzi JL Jr. Imaging, biomarker, and clinical et al. 2017 AHA/ACC/HRS guideline for manage-
27:3924–33.
predictors of cardiac remodeling in heart failure ment of patients with ventricular arrhythmias and
with reduced ejection fraction. J Am Coll Cardiol 72. Puntmann VO, Carr-White G, Jabbour A, et al. the prevention of sudden cardiac death: a report
HF 2019;7:782–94. T1-Mapping and outcome in nonischemic cardio- of the American College of Cardiology/American
myopathy: all-cause mortality and heart failure. Heart Association Task Force on Clinical Practice
59. Daubert MA, Adams K, Yow E, et al. NT-
J Am Coll Cardiol Img 2016;9:40–50. Guidelines and the Heart Rhythm Society. J Am
proBNP goal achievement is associated with sig-
Coll Cardiol 2018;72:e91–220.
nificant reverse remodeling and improved clinical 73. Wilcox J, Yancy CW. Stopping medication for
outcomes in HFrEF. J Am Coll Cardiol HF 2019;7: heart failure with improved ejection fraction. 82. Ackerman MJ, Priori SG, Willems S, et al. HRS/
158–68. Lancet 2019;393:8–10. EHRA expert consensus statement on the state of
734 Wilcox et al. JACC VOL. 76, NO. 6, 2020

HFrecEF Consensus Recommendations AUGUST 11, 2020:719–34

genetic testing for the channelopathies and car- 84. Ortiz-Genga MF, Cuenca S, Dal Ferro M, et al. 86. Yu CM, Chau E, Sanderson JE, et al. Tissue
diomyopathies this document was developed as a Truncating FLNC mutations are associated with Doppler echocardiographic evidence of reverse
partnership between the Heart Rhythm Society high-risk dilated and arrhythmogenic cardiomy- remodeling and improved synchronicity by simul-
(HRS) and the European Heart Rhythm Association opathies. J Am Coll Cardiol 2016;68:2440–51. taneously delaying regional contraction after
(EHRA). Heart Rhythm 2011;8:1308–39. biventricular pacing therapy in heart failure. Cir-
85. Thomas IC, Wang Y, See VY, Minges KE,
culation 2002;105:438–45.
83. Halliday BP, Cleland JGF, Goldberger JJ, Curtis JP, Hsu JC. Outcomes following implantable
Prasad SK. Personalizing risk stratification for cardioverter-defibrillator generator replacement
sudden death in dilated cardiomyopathy: the in patients with recovered left ventricular systolic
past, present, and future. Circulation 2017;136: function: the National Cardiovascular Data Regis- KEY WORDS heart failure with recovered
215–31. try. Heart Rhythm 2019;16:733–40. ejection fraction, myocardial recovery