Durable left ventricular assist devices (LVADs) have classically been understood as a bridge to transplant or as destination therapy. However, sustained mechanical unloading of the left ventricle can induce, in a selected subgroup, structural and functional improvement sufficient to consider a bridge-to-recovery strategy.
The problem is that this possibility remains underused. In many centers recovery is detected incidentally, without systematic protocols, and the decision to explant depends heavily on local experience. On top of this, previous models have often used device explantation as a synonym for recovery, when that decision is conditioned by clinical, surgical, institutional and even psychological factors. Having a simple tool that estimates the probability of recovery before implantation could help select patients better, set expectations and design dedicated programs.
The study develops and externally validates the RecoverHeart calculator from a multicenter collaboration with a prominent role for the Utah Cardiac Recovery program. The derivation cohort comprised 509 patients from five institutions and the external validation cohort 375 from four additional centers. The primary response criterion was reaching, within the first year after implantation, an LVEF ≥ 40% together with a left ventricular end-diastolic diameter ≤ 6 cm, regardless of whether the device was ultimately explanted. Four preoperative variables were selected using penalized regression (LASSO).
Significant reverse remodeling was reached by 76 patients (14.9%) in the derivation cohort and 42 (11.2%) in the validation cohort. The predictors were female sex, non-ischemic cardiomyopathy, shorter heart failure duration and a smaller pre-implant end-diastolic diameter. The calculator showed acceptable discrimination, with a C statistic of 0.72 in the derivation cohort and 0.75 in validation, generated individual probabilities ranging from 0.2% to 67.6%, allowed patients to be stratified into low, intermediate and high probability, and outperformed both a previous score (I-CARS) and isolated predictors. The authors conclude that RecoverHeart is a practical tool, based on variables available before implantation, which (by defining recovery through objective echocardiographic criteria rather than explantation) could improve the selection of candidates for a bridge-to-recovery strategy and refine the complementary use of assist devices and transplantation.
COMMENTARY:
The value of this work is not mathematical but clinical: it forces the question to be asked before implantation. Which patient with advanced heart failure about to receive an LVAD has a reasonable chance of recovering ventricular function? In a setting of organ scarcity, complex waiting lists and increasingly heterogeneous patients, identifying those who might benefit from a genuine bridge-to-recovery strategy can change the conversation before the pump goes in.
It is worth gauging the magnitude without overstatement. Recovery is not anecdotal, but neither is it frequent: roughly one in seven patients in the derivation cohort and one in nine in validation. This fits accumulated experience. Complete recovery should not be sold as a general expectation, but it should not be ignored either.
The four predictors are clinically plausible: female sex, non-ischemic etiology, shorter time of evolution and a smaller end-diastolic diameter. Non-ischemic cardiomyopathy carries the most weight among the positive predictors, while longer-standing disease and a more dilated ventricle progressively lower the probability. One caveat worth keeping in mind: female sex did not reach conventional statistical significance in the multivariable model (p = .131) and is retained because the penalized method keeps it, not because it is a robust predictor on its own. Together they outline a recognizable phenotype: less chronically remodeled myocardium, less scar, more room for recovery after unloading.
From the standpoint of an assist-device program, the work shifts the focus from “implant and wait” to “implant with a defined plan.” A non-ischemic patient, of short evolution and with a not-excessively-dilated ventricle who scores high should perhaps enter a specific pathway from the outset: aggressive optimization of medical therapy, serial echocardiograms, speed-reduction tests, right ventricular assessment and periodic discussion within a team that truly knows recovery. The tool does not replace clinical judgment; it helps recovery stop being a matter of chance and become an explicit goal in selected patients.
A real methodological strength is that recovery is not defined as explantation but echocardiographically (LVEF ≥ 40% and LVEDD ≤ 6 cm). This matters, because explantation depends on many factors unrelated to the myocardium: surgical experience, center culture, fear of relapse, patient preference, device complications or transplant availability. A patient may have recovered relevant ventricular parameters and still not be explanted. That said, the opposite should be stated just as clearly: predicting reverse remodeling is not predicting safe explantation, cure or freedom from recurrence. Withdrawing support requires more elements: clinical stability, functional reserve, response to partial-unloading tests, right ventricular function, mitral or tricuspid regurgitation, pulmonary pressure, tolerance of medical therapy and team experience. RecoverHeart is a good entry point to the recovery protocol, not the complete protocol.
The limitations matter and deserve a practical reading. The multicenter design strengthens external validity but introduces variability in selection, follow-up, protocols and echocardiographic measurement, and there was no central imaging core laboratory—something that affects precisely the key variables (LVEF and diameters). Some centers enriched their sampling with recovered patients, so the absolute rates are not population incidence: the authors themselves clarify that the aim was relative individual probability. And much of the cohort predates four-pillar therapy; it is reasonable to think that sacubitril/valsartan, SGLT2 inhibitors, optimized beta-blockade and mineralocorticoid antagonists could modify the probability of recovery, so performance in contemporary programs—with more HeartMate 3 and intensive pharmacological optimization—remains to be confirmed. Discrimination, moreover, is acceptable but not spectacular (C 0.72–0.75): useful for stratifying, not for deciding on its own.
In daily practice, this calculator should not be used to deny an LVAD to a patient with a low probability of recovery. A low score does not mean the device will not help; it simply indicates that the goal will remain bridge to transplant or destination therapy. A high score, by contrast, does change the conversation: it allows the patient to be told that, beyond stabilizing advanced heart failure, there is a reasonable chance of significant ventricular improvement and, in very selected cases, of future weaning from support.
The deeper contribution is that it helps mature the idea of the LVAD as a dynamic therapeutic tool, not a static destination. Not all patients with advanced heart failure share the same biological fate: some have a ventricle too dilated, fibrotic or ischemic to expect much; others retain a margin of reversibility we may not be exploiting enough. RecoverHeart does not resolve every doubt, but it offers a simple way to begin that stratification and, above all, it forces us to think about recovery before implanting, rather than stumbling upon it during follow-up.
REFERENCE:
Taleb I, Kyriakopoulos CP, Wever-Pinzon O, Maneta E, Selzman CH, Dranow E, et al. Determining the individualized probability of myocardial recovery: The multicenter RecoverHeart calculator. J Heart Lung Transplant. 2026. doi:10.1016/j.healun.2026.04.008.
