CARC9: Validation of HeartMate 3 Risk Score as a Survival Predictor Calculator

Abstract

Purpose: Recently, a patient-specific HeartMate 3 Survival Risk Score (HM3RS) was developed to predict 1-year and 2-year survival after implantation based on age, prior cardiac surgery, sodium level, blood urea nitrogen (BUN) level, left ventricular end diastolic diameter, and right atrial pressure/pulmonary capillary wedge pressure (RAP/PCWP) ratio. We sought to validate the HM3RS using recent data from our referral center. Methods: Through a retrospective review of 101 HM3 patients implanted between 2017-2021, baseline characteristics, complete hemodynamic profiles, and significant outcomes were collected. Patients were stratified into 3 groups by HM3RS: higher-than-average expected survival (HM3RS <2.41), average expected survival (2.41-2.97), and lower-than-average expected survival (>2.97). Kaplan Meier (KM) curves stratified by HM3RS grouping were plotted and survival was compared with log-rank tests. Time-varying receiver operating characteristic (ROC) curves were also constructed to evaluate survival predictions. Results: Patients with lower-than-average expected survival (n=43) were older (mean age 65±7.6 years), had higher BUN before LVAD (45.3±20.1 mg/dl), were more likely to have had prior coronary artery bypass surgery (41.8%), and RAP/PCWP ratio > 0.6 (56%) (all P < 0.05) (table 1). KM plots showed lower survival in this group at 1-year, 2-years, and during the entire follow-up of 2±1.2 years censored at time of transplant, death or pump exchange (figure 1). However, these results did not reach statistical significance, possibly owing to small sample size. Area under the ROC curve was 0.622 and 0.676 at 1 and 2 year follow-up, respectively (figure 2). Summary: In this study, observed survival was consistent with predictions derived from HM3RS score. The HM3RS appears to be a valid tool to predict survival after LVAD implantation in the real-world and to aid in decision-making. Larger prospective trials are needed to independently validate this tool in clinical practice.Figure 1. Kaplan Meier curves for survival by HM3RS groups through (A) entire follow-up, (B) 1-year of follow-up, (C) 2-years of follow-up.Figure 2. Time dependent receiver operating curve (ROC) to determine survival prediction (A) at 1 year, and (B) at 2 years.