HeartMate II Risk Score (HMRS) and MELD-Xi Scores Do Not Predict Mortality in HeartMate 3 LVAD Patients

Purpose Heartmate II risk score (HMRS) and MELD-Xi score are validated mortality risk predictors in Heartmate II left ventricular assist device (LVAD) patients. Prognostic utility of these risk scores remains unknown in patients considered for Heartmate 3 (HM3) LVAD therapy. Methods HMRS were calculated for 175 consecutive patients who underwent HM3 LVAD implantation at a single academic institution. Receiver operating characteristic (ROC) curve and logistic- and cox-regression models were developed to assess the impact of HMRS and MELD-Xi scores on short- and long-term outcomes in HM3 patients. Results Mean HMRS and MELD-Xi scores in HM3 population was 1.38±0.57 and 14.4±0.28, respectively. Pre-implant HMRS was not predictive of RVAD use (OR 1.25, p=0.392), CVVHD (OR 1.16, p=0.722), in-hospital mortality (OR 1.52, p=0.409), or 90-day mortality (OR 1.44, p=0.457). Pre-implant MELD-Xi score showed trend towards increased RVAD use (OR 1.09, p=0.056) and need for CVVH (OR 1.14, p=0.064), however was not predictive of in-hospital mortality (OR 1.02, p=0.836), or 90-day post-LVAD mortality (OR 0.958, p=0.636). Area under curve (AUC) for in-hospital mortality was 0.531 for HMRS and 0.525 for MELD-Xi score (Figure) . There was no difference in post-implant survival (88.6% vs. 85.7%, log-rank p=0.569), freedom from GI bleeding (73.5% vs. 65.1%, log-rank p =0.257) or freedom from stroke (93.3% vs. 90.8% log-rank p = 0.478) at 2-year follow-up in patients with below average vs. above average HMRS. Similarly, There was no difference in post-implant survival (91.3% vs. 82.3%, log-rank p=0.121), freedom from gastrointestinal bleeding (72.2% vs. 65.2%, log-rank p =0.257) or freedom from stroke (92.0% vs. 92.4% log-rank p = 0.820) at 2-year follow-up in patients with below average vs. above average MELD-Xi score. Conclusion HMRS and MELD-Xi scores do not predict mortality in patients undergoing HM3 implantation. New prediction models are required to optimize patient selection and outcomes in HM3 patients.