Heart failure (HF) is characterized by a pathologic process known as remodeling that involves systolic and diastolic impairment, progressive ventricular dilation, and an increase in intracardiac pressures. The process of remodeling is associated with adverse cellular, structural, and functional myocardial changes, that have long been deemed progressive and unidirectional. Clinical experience has shown that the process of remodeling can be delayed or even reversed, either spontaneously in the setting of acute cardiac injury (e.g., acute myocarditis, stress-induced cardiomyopathy etc.), or it can be facilitated through guideline-directed HF therapy including cardiac resynchronization therapy in chronic HF [1,2,3]. Mechanical circulatory support (MCS) with left ventricular assist devices (LVADs) is an established treatment modality for patients with advanced disease and besides its role in supporting systemic circulation by augmenting cardiac output, it provides significant volume and pressure unloading, creating a favorable environment for the reversal of the structural and functional alterations of the failing heart, a process known as reverse remodeling. It has been repeatedly shown that a subset of advanced HF patients can significantly improve their cardiac structure and function while on durable MCS, to the point where withdrawal of the LVAD support can be considered [4,5,6,7,8,9,10,11,12]. In light of these findings, the concept of HF irreversibility has been refuted and the notion that severe HF requiring durable LVAD support indicates irreversible end-stage disease has been revised.

In the recent past, left ventricular remodeling was widely considered irreversible, especially in patients with advanced disease. Besides evidence that the remodeling process could be attenuated with early implementation of angiotensin-converting enzyme inhibitors (ACEi) after a myocardial infarction, the concept that a profoundly dilated ventricle in end-stage HF could revert to a significantly improved phenotype had not been reported [13,14]. The first challenges to this notion were largely driven by the observation that advanced HF patients can improve their cardiac function after MCS. In 1994, Frazier et al. were the first to describe cardiac improvement in a series of 18 patients supported with either a pneumatic or vented electric HeartMate^® (Thoratec Corporation, Pleasanton, CA, USA) device [15,16]. The investigators observed that LVAD support resulted in a significant reduction in the cardiothoracic ratio and left ventricular end-diastolic diameter (LVEDD), and in an improvement in the left ventricular ejection fraction (LVEF) from the baseline. Improved hemodynamics, including a decreased pulmonary capillary wedge pressure (PCWP) and an enhanced cardiac index were also noted in these patients, while histologic examinations demonstrated a reduction in the mean area of myocytes. Close to that time, Levin et al. measured the end-diastolic pressure–volume relationship and examined the cardiac tissue at the time of heart transplantation in patients treated with medical therapy compared to patients bridged with LVAD support [17]. They observed that the LVAD support was associated with a regression of the cellular hypertrophy and a shift of the end-diastolic pressure–volume relationship towards normal values, suggesting a reversal of the remodeling process.

As these findings were often accompanied by a significant recovery of the underlying cardiac function, the first cases where the left ventricular functional improvement was sufficient to allow for LVAD removal (cardiac recovery or a “remission” of HF) were reported [10,11,18]. The first cases of LVAD explantation were performed without using standardized criteria to assess the left ventricular recovery; however, in the following years a more standardized approach was followed and criteria for device weaning were introduced and implemented [4,7,9,19,20,21]. The first anecdotal experiences were followed by reports showing high rates of cardiac recovery and subsequent LVAD explantation, by combining LVAD support with adjuvant reverse remodeling drug therapy, alongside regular testing of the underlying cardiac function [5,6,22,23]. Henceforth, an increasing number of institutions adopted approaches of facilitating and testing for recovery and potential LVAD explantation, leading to the RESTAGE-HF (Remission from Stage D Heart Failure) multicenter study [4]. In this prospective trial, 19 out of the 40 (47.5%) selected chronic advanced HF patients undergoing LVAD support combined with a standardized pharmacologic and cardiac function monitoring protocol, markedly improved their left ventricular structure and function, and had the device explanted. Several cohorts of patients now exist who have had their device removed and have had sustained recovery for many years [4,20,24,25,26]. These patients have been able to return to a normal lifestyle without requiring heart transplantation, thereby enabling the allocation of donor hearts to other individuals in need of this precious resource.

After the initial anecdotal experiences of cardiac recovery in the 1990s, several subsequent reports investigated the phenomenon in a series of LVAD-explanted patients, trying to shed light on the incidence and long-term outcomes.

One of the first reports of marked left ventricular structural and functional improvement on LVAD support and subsequent device explantation was in a small series of five patients with advanced HF and underlying non-ischemic cardiomyopathy (idiopathic dilated cardiomyopathy in three, and postpartum cardiomyopathy in two of the patients) [10]. In three of these patients, the LVAD was removed electively after the recovery of cardiac function, while in the remaining two, it was removed because of device malfunction. While 1 patient died of a non-cardiac cause 10 days after LVAD removal, the other 4 patients remained alive and well 35, 33, 14, and 2 months after LVAD removal, respectively.

The multicenter Thoratec Registry provided one of the largest patient cohorts and included a total of 281 patients with underlying non-ischemic cardiomyopathy [9]. A total of 22 out of 281 (8.1%) patients underwent LVAD explantation, with 17 patients remaining alive, 16 in New York Heart Association (NYHA) class I, and 1 in NYHA class II, after an average follow-up of 3.2 years (1.2–10 years). In another large series, Mancini et al. retrospectively reviewed 111 patients receiving an LVAD as a bridge to transplantation [18]. Only 5 of the 111 patients (4.5% overall, and 9% of patients with non-ischemic HF etiology) had substantial cardiac recovery and were deemed appropriate for device explantation. Notably, just 1 patient remained alive with sustained left ventricular improvement after 15 months of follow-up.

Subsequent studies by the Berlin group reported higher rates of LVAD removal and more sustained left ventricular recovery. Their initial report showed that all 5 patients who underwent explantation exhibited a preserved cardiac function in the following 51 to 592 days [21]. In succeeding reports by 2005, 32 out of 131 (24%) patients with non-ischemic HF underwent device explantation and exhibited a five-year survival rate of 78.3 ± 8.1% [20]. The explanted patients were free from HF symptoms recurrence at a rate of 69.4% and 58.2% at three and five years, respectively. By 2008, 81 patients were weaned from left, right, or biventricular assist devices that had been implanted for end-stage HF in the same center [27]. When only patients with non-ischemic cardiomyopathy were analyzed and after excluding patients with proven myocarditis, it was found that 35 out of 188 (18.5%) patients underwent device explantation. Thirty patients had the device explanted electively, while in another five the decision was precipitated by pump-related complications. In eight of the electively weaned patients, the LVEF had not normalized (30% to 44%) and the LVEDD was between 56 to 60 mm. Nonetheless, the overall 5- and 10-year survival rates following LVAD explantation, including survival after heart transplantation for patients with HF recurrence, were 79.1 ± 7.1% and 75.3 ± 7.7%, respectively. A major finding through these series was that patients with long-term weaning stability had a shorter duration of HF, were younger, and required MCS for a shorter interval [7,28,29].

The LVAD Working group presented a prospective multi-center study of 67 patients who received an LVAD for refractory HF across eight centers in the US [30]. Thirty-seven patients had an underlying non-ischemic, and thirty an ischemic cardiomyopathy, while all patients underwent implantation of the HeartMate^® XVE LVAD (Thoratec Corporation, Pleasanton, CA, USA). On an echocardiographic follow-up, the LVEF increased from 17 ± 7% before LVAD implantation to 34 ± 12% (p < 0.001), the LVEDD decreased from 7.1 ± 1.2 cm to 5.1 ± 1.1 cm (p < 0.001), and the left ventricular mass decreased from 320 ± 113 g to 194 ± 79 g (p < 0.001). Peak oxygen consumption (pVO₂) improved while on LVAD support (13.7 ± 4.2 mL/kg/min at 30 days vs. 18.9 ± 5.5 mL/kg/min at 120 days; p < 0.001). Overall, six (9%) patients were weaned from their LVAD due to cardiac recovery.

The Utah Cardiac Recovery Program reported on 154 consecutive, prospectively enrolled patients with chronic advanced HF receiving a continuous-flow LVAD, after excluding patients with an acute HF etiology [31]. They reported that 21% of patients with an underlying non-ischemic and 5% of those with an ischemic cardiomyopathy achieved an LVEF ≥ 40% after at least six months on LVAD support, while the end-diastolic and end-systolic volumes were significantly and similarly improved in both cohorts.

Recently, the results of the multicenter RESTAGE-HF trial were published [4]. Forty patients with chronic advanced HF receiving the HeartMate II™ LVAD (Thoratec Corporation, Pleasanton, CA, USA) were enrolled across six US centers. An underlying non-ischemic cardiomyopathy, LVEF < 25% with cardiomegaly, age < 60 years, and a duration of HF < 5 years were the inclusion criteria. The LVAD speed was optimized, a HF pharmacological regimen was implemented, and regular echocardiograms were performed at a reduced LVAD speed to test the underlying cardiac function. Prior to LVAD implantation, the LVEF was 14.5 ± 5.3% and the LVEDD was 7.33 ± 0.89 cm. Four enrolled patients did not abide by the protocol due to medical complications unrelated to the study procedures. Overall, 40% of all the enrolled (16/40) patients achieved the primary endpoint of a marked left ventricular structural and functional improvement, a subsequent LVAD explantation within 18 months, and sustained remission from HF at 12 months. Half (18/36) of the patients receiving the protocol were explanted within 18 months (pre-explant LVEF 57 ± 8%; LVEDD 4.81 ± 0.58 cm; LV end-systolic diameter 3.53 ± 0.51 cm; PCWP 8.1 ± 3.1 mmHg; pulmonary artery saturation 63.6 ± 6.8% at 6000 rpm), while overall, 19 patients were explanted (19/36, 52.3% of those receiving the protocol). A post-explantation survival rate, free from reimplantation of an LVAD or heart transplantation, was 90% at one year, and 77% at two and three years. The investigators concluded that a strategy of LVAD support combined with a standardized pharmacologic and cardiac function monitoring protocol resulted in high rates of LVAD explantation and was feasible and reproducible with explantations taking place in all the participating sites.

A recent analysis of 358 consecutive patients with HF with a reduced LVEF (where patients with acute HF etiologies and less than three months of post-LVAD echocardiographic follow-up were excluded by the study design) receiving a continuous-flow LVAD across four US institutions, showed that 34 (10%) patients had an LVEF ≥ 40% and an LVEDD ≤ 6.0 cm at the last available echocardiographic follow-up timepoint within one year of LVAD support [32]. An additional 112 (31%) patients had an absolute LVEF improvement of ≥5% post-MCS (A median LVEF increase of 9% with a range of 6–14%), with the authors suggesting that such patients may benefit from further mechanical unloading or titration of a guideline-directed HF medical therapy to further improve their cardiac structure and function.

The overall rate of cardiac recovery leading to device weaning in the Interagency Registry for Mechanically Assisted Circulatory Support (INERMACS) database has consistently been low, occurring in less than 5% of the implant cases by five years [33]. Topkara et al., however, studied 13,454 patients implanted with a continuous flow LVAD from 2006 to 2015 using the INTERMACS registry [34]. In this line of investigation, cardiac recovery during LVAD support was defined as complete if the device explantation was performed, or as partial if the patient demonstrated a substantial improvement of the left ventricular systolic function (LVEF > 40%) at any follow-up echocardiographic assessment, yet not achieving the device explantation clinical endpoint. Out of 8805 patients with an LVEF < 30% at the time of device implantation, 761 (8.6%) achieved a partial cardiac recovery, with 406 (4.6%) patients reaching an LVEF in the range of 40–50% and 355 (4.0%) patients an LVEF greater than 50%. Again, by using the INTERMACS registry, the Utah group found that out of 15,631 LVAD patients, approximately 13% either underwent device explantation for cardiac recovery or achieved a follow-up LVEF > 40% [35]. This apparently high discrepancy between the patients achieving marked left ventricular structural and functional improvement, and those who eventually had their LVAD explanted, was not unanticipated. This reflects the complexity of decision-making in favor of LVAD weaning, a decision that is affected by various factors pertaining, among others, to the physician’s expertise, institutional experience, and patient and physician perspectives and goals. Probably the most important factor affecting that decision is the fact that the alternative option of heart transplantation is the gold standard therapy for end-stage HF compared to LVAD explantation, which is, in reality, still under clinical investigation/research.

The incidence of LVAD-mediated cardiac recovery is highly variable in the literature, likely representing a variability in study design, patient selection criteria and definition of cardiac recovery, including the acceptable thresholds of left ventricular improvement to allow for device weaning. The results of key clinical outcome studies investigating cardiac functional and structural improvement following long-term MCS therapy in a prospective way are summarized in Table 1 [4,5,6,7,22,23,27,30,31,36,37,38,39,40].

The time course and magnitude of improvement in cardiac structure and function after LVAD placement was prospectively evaluated in a study of 80 consecutive patients with chronic HF, due to both ischemic and non-ischemic cardiomyopathy, who underwent implantation of a continuous-flow device from 2008 to 2011 [8]. The cardiac recovery was assessed on the basis of systolic and diastolic echocardiographic improvement, which was sustained during echocardiograms performed at a reduced pump speed. The serial echocardiographic assessment took place at regular intervals for one year. After six months of LVAD unloading, 34% of the patients had a relative LVEF increase above 50% (compared with the pre-implantation values) and 19% of the patients achieved an LVEF ≥ 40%, irrespective of the underlying HF etiology. This improvement in systolic function was seen as early as 30 days after device implantation with the greatest magnitude of improvement achieved by six months and persisting over the one-year follow-up period. The left ventricular diastolic function parameters improved as early as 30 days and persisted over time. The left ventricular end-diastolic and end-systolic volumes decreased significantly as early as 30 days post-MCS (113 vs. 77 mL/m²; p < 0.01, and 92 vs. 60 mL/m²; p < 0.01, respectively). The left ventricular mass also decreased as early as 30 days after circulatory support initiation (114 vs. 95 g/m²) and continued to decrease over the one-year follow-up. Importantly, it did not reach values below the normal reference range, suggesting there was no atrophic remodeling after prolonged LVAD support.