Predictive Model Of The Sustainability Of Favorable Response After LVAD Weaning In HF Patients: A VAD Wean Registry Analysis

Abstract

Introduction

During left ventricular assist device (LVAD) support, a subset of heart failure (HF) patients can experience myocardial structural and functional improvement that can lead to device weaning.

Hypothesis

We sought to derive a predictive model for the durability of favorable response after LVAD weaning.

Methods

We studied 400 HF patients enrolled in the international multicenter VAD Wean Registry who received a durable continuous-flow LVAD from 2002 to 2019 and underwent device support weaning (Figure). Indications for LVAD weaning included: (a) structural and functional myocardial improvement meeting institutional criteria for “myocardial recovery” (i.e. responders) or (b) LVAD-related complications accompanied by variable degrees of cardiac improvement (i.e. partial responders). We obtained clinical data before and during LVAD support as well as after LVAD weaning. The primary outcome was 3-year freedom from death, transplant or LVAD re-implantation. Bootstrap imputation and LASSO variable selection techniques were used to derive a predictive model which was then internally validated.

Results

Patients were predominantly White race (71%), Male sex (63%), with an average age of 42±14 years. At 3-years, 78% patients survived without the need for transplant or LVAD re-implantation. Out of 37 variables, we identified four that were included in a multivariate model achieving a C-statistic of 0.70 (95% CI: 0.64-0.76), (Table): indication for LVAD weaning, age, HF duration, and presence of implantable cardioverter-defibrillator (ICD).

Conclusions

Using routine clinical data from a multicenter registry we developed an internally validated risk tool to predict the sustainability of favorable response for 3 years after LVAD weaning. This model could aid clinical discussions related to the long-term management and prognosis of HF patients on LVAD support.