The generation of clinical practice guidelines accomplishes several aims. First, it reviews available evidence that when placed in a graded hierarchal framework informs clinical practice statements. Second, it identifies gaps in evidence intended to spur future research. Third, it introduces new concepts that orient the discipline toward new models of understanding. Therefore, with regard to the latter aim, the American College of Cardiology/American Heart Association 2013 heart failure (HF) guidelines for the first time articulated a theoretical categorization of HF based on ejection fraction (EF) measurements that differed from prior schemes.1 Rather than wrestle with the demarcation of HF with reduced EF (HFrEF) vs HF with preserved EF (HFpEF) at a single cut point, the writing group recognized that no evidence base existed to accommodate such precision and that, indeed, the experiential observations captured the theme of a “gray zone” with borderline EF (ie, left ventricular ejection fraction [LVEF] >0.40 and <0.50). Moreover, clinical empiricism led the committee to opine that the pathway to this borderline zone was either de novo discovery or recovery from prior documented HFrEF. Such a categorization raised several important questions. Are those patients still candidates for class of recommendation I evidence-based medical and device therapy as indicated by the guidelines for HFrEF? What is the natural history of HF with improved EF? How large is this cohort? What are the patient experiences of those who have experienced recovery? Most important, is there a science to recovery that creates insight regarding new mechanisms and treatments of left ventricular (LV) dysfunction?
In this issue of JAMA Cardiology, Kalogeropoulos et al2 add nicely to this emerging phenotype with a careful single-center analysis of all patients with HF, including detailed phenotyping and longitudinal follow-up with the best evidence to date to endorse the original effort of the American College of Cardiology/American Heart Association HF guideline writing committee to identify this new and potentially important category of HF.1 The work by Kalogeropoulos et al2 establishes several new insights: (1) 16.2% (350 of 2166) of those with an LVEF exceeding 0.40 potentially represented improved or “recovered” HF (ie, HFrecEF), (2) 3-year mortality was lowest for HFrecEF compared with HFrEF and HFpEF, and (3) hospitalization burden was significantly lower for HFrecEF compared with the other phenotypes. These data seemingly confirmatory of this new HF phenotype are supportive but not definitive. As pointed out by the authors, the limitations are important and merit emphasis, including it being a single-center study, academic referral bias, the absence of antecedent ventricular function measurements in some patients with LVEF exceeding 0.40, and a probable survivor bias that may overstate the true incidence of HFrecEF. Most important, the absence of a concurrent comparator group or inception cohort3 could very well represent a survival bias alone, which leads us to interpret the data with interest but not yet with conviction. Added to those important limitations are the uncertainty of reversible disease causality (eg, inflammatory myocarditis, peripartum cardiomyopathy, toxin-induced LV dysfunction, or the variability of LVEF assessment). Yet, the conclusions by Kalogeropoulos et al2 are inescapable: a patient population exists that has experienced recovery of ventricular function from frank HFrEF. The foregoing questions raised regarding therapeutic implications cannot be answered with the data from the study by Kalogeropoulos et al,2 but it is important to address what we understand regarding mechanisms of recovery.
Despite reports of myocardial recovery,4 we have a limited understanding of the determinants of recovery and their associated underlying mechanisms because most published data are limited to the small number of patients who have undergone LV assist device (LVAD) implantation, with a lack of systematic data from clinical cohorts.5- 7 In large-scale clinical trials for HFrEF, there are many patients who derive benefit, leading to approval of medical and device therapies. However, even in the most convincing trials, there are also many patients who derive no benefit and may even experience harm.8 The ability of the myocardium to recover (ie, cardiac reserve) is fundamental for the capability to respond to therapies (ie, reverse remodel). In fact, reverse LV remodeling is the only surrogate marker shown to be predictive of improved outcome in HFrEF.9,10 Therefore, better understanding of the ability or inability to improve LVEF (or reverse remodel) may help to (1) determine mechanisms of recovery, (2) assist in risk discrimination (eg, identify patients with little potential for recovery and therefore in need of intensive disease management or referral for LVAD and heart transplantation or palliative care), and (3) enhance clinical trial design (eg, identify patients with the potential for myocardial recovery and subsequent testing of new strategies to promote or even accelerate recovery).
The growing field of precision medicine may provide a framework to find a common biological basis of different subclasses of HFrEF beyond ischemic and nonischemic or dilated cardiomyopathy and to provide information about trajectories in HF (including cardiac recovery) after initial diagnosis. The ability to deeply phenotype HF populations with advanced imaging techniques (eg, strain echocardiography and cardiac magnetic resonance imaging), biomarker information, and hemodynamic data, in addition to other clinical factors (eg, medication and clinical laboratory data), allows newer bioinformatics approaches to determine those characteristics that best predict the outcome of improved ventricular function.
Given the sophisticated nature of many electronic health records and the application of novel data mining techniques, it is now possible to build robust, phenotypically enriched data sets to study clinical predictors of myocardial recovery in HFrEF. Thus far, most studies examining outcomes in HF (including myocardial recovery) use basic statistical techniques. However, biological systems are much more complex, and a systems biology view may be appropriate. Machine learning can account for these higher-order interactions because it is a scientific discipline that focuses on how computers learn from data, arising from the intersection of statistics and computer science.11 Of note, machine learning algorithms have already been applied to well-phenotyped HFpEF clinical cohorts, with the identification of unique subgroups with differing natural histories and likely varied responses to candidate medical therapies.12 A similar approach may provide the same precision in the description of HFrecEF.
We have some scientific insight regarding LV recovery from the biological response to mechanical circulatory support. Bruckner et al13 have reported that patients with a higher degree of fibrosis at the time of LVAD implantation had less recovery of LV systolic function during LVAD unloading. It is well described that HF is associated with defects in calcium homeostasis.14 Notably, among patients with myocardial recovery and subsequent LVAD removal, favorable changes in myocardial excitation-contraction coupling were seen, including shorter action potential duration and increased sarcoplasmic reticulum calcium content.15 Together, these important observations suggest that there are specific molecular pathways involved in the myocardial recovery process.
Exploiting the potential for cardiac recovery via mechanical unloading is under active investigation in the multicenter prospective Remission From Stage D Heart Failure study,16 using LVAD as a bridge to recovery in patients with nonischemic cardiomyopathy. Promising interim results have been presented: 5 out of 20 patients (25%) have been successfully explanted after a mean of 265 days (range, 197-417 days) of support.17
Pharmacologic studies have also provided understanding into myocardial recovery. The work done coincident with the introduction of β-blocker therapy for HFrEF demonstrated a restoration of β-adrenergic density and a return to predominant β1 receptors on “restored” myocytes.18 Additional insights from the use of aldosterone receptor antagonists demonstrated enhanced benefit aligned with active collagen turnover.19
Myocardial recovery is not the opposite of disease progression and likely represents more than reversal of already described pathophysiological models of LV dysfunction. It is a distinct response to therapy or withdrawal of negative stimulus. This process may be quite intricate or may represent the benefit of timely interventions before an adverse burden of remodeling gone too far.20 Where this “point of no return” lies in the natural history of HF is unknown and likely varies based on HF etiology. A major limitation of the study by Kalogeropoulos et al2 is that the duration of HF is not included (or was unknown). Indeed, HF duration may be viewed as a surrogate of chronic adverse myocardial structural and molecular remodeling beyond which there is no potential for repair.
Going forward, it is critically important that clinical trials must harness the opportunity to study in parallel the cellular and molecular processes associated with the observed varying degrees of clinical myocardial recovery or reverse remodeling. Prospective studies should be commenced that include well-phenotyped inception cohorts to accurately identify the incidence of recovery and to discriminate those likely to recover and those likely to fail evidence-based medical and device therapy. We agree wholeheartedly with a newly announced clinical and translational research effort championed by the National Heart, Lung, and Blood Institute and a newly formed working group to address myocardial recovery (Utah Cardiac Recovery Symposium, oral communication, January 2016). We would implore the panel to consider myocardial recovery not only in the context of mechanical circulatory support but also as the results of best application of evidence-based medical therapy. Indeed, work our group has previously published demonstrated the potential for substantial recovery with process improvement and quality-driven optimization of evidence-based medical therapy. In this ambulatory population of patients with HFrEF (Registry to Improve the Use of Evidence-Based Heart Failure Therapies in the Outpatient Setting [IMPROVE HF] cohort), almost one-third of patients experienced a doubling of LVEF (25% to 46%).21 We would further submit that the future of successful cardiac recovery programs includes all stages of HF, especially stage B or asymptomatic LV dysfunction, when certain pathophysiological mechanisms may still be reversible.
It is our opinion that myocardial recovery exists, as evidenced by clinical trials, observational data, and recent integration into current guidelines. Now is the time to recognize recovery as a clinical reality for patients with HFrEF and to begin a deliberate pursuit of the underlying mechanisms and future clinical considerations. Indeed, a new phenotype of HF has emerged.
Corresponding Author: Jane E. Wilcox, MD, MSc, Division of Cardiology, Department of Medicine, Northwestern University Feinberg School of Medicine, 676 N St Clair, Ste 600, Chicago, IL 60611 (firstname.lastname@example.org).
Published Online: July 6, 2016. doi:10.1001/jamacardio.2016.1356.
Conflict of Interest Disclosures: Both authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and none were reported.
Wilcox JE, Yancy CW. Heart Failure—A New Phenotype Emerges. JAMA Cardiol. 2016;1(5):507-509. doi:10.1001/jamacardio.2016.1356