MELD score predicts outcomes in patients with advanced heart failure: A longitudinal evaluation.

Abstract

Advanced heart failure (AHF) is characterized by recurrent episodes of haemodynamic instability and frequent hospitalizations, leading to a progressive decline in quality of life and high mortality rates. The objectives of this study were to evaluate the effect of the model for end-stage liver disease (MELD) score and its variations in predicting adverse outcomes [death, urgent heart transplant, and left ventricular assist device (LVAD) implant] among patients with AHF to assess the clinical associations of the MELD score in this population and to compare the efficacy of this tool with other prognostic scores in AHF. In this longitudinal prospective study, 162 patients with advanced heart failure (AHF) were enrolled; all patients included in the study were receiving the maximum tolerated medical therapy according to guidelines. The MELD score was measured at baseline and every 6months during follow-up. All patients underwent echocardiographic assessment and cardiopulmonary testing, which included the evaluation of maximal oxygen uptake (VO2max) and the minute ventilation/carbon dioxide production (VE/VCO2) slope. The mean age of the study group was 57.7±11.6years. There were 26 deaths, 5 urgent transplants, and 1 LVAD implantation during a follow-up period of 31.4±15.6months. The mean New York Heart Association (NYHA) class was 2.8±0.5, ejection fraction (EF) was 26.3±6.5%, the mean VO2max was 11.7±3.5mL/kg/min. Multiple regression analysis revealed a positive correlation between the MELD score and NT-proBNP (β=0.215; P=0.041) and furosemide dosage (β=0.187; P=0.040). Conversely, a negative correlation was observed between the MELD score and TAPSE (β=-0.204; P=0.047). Multivariate Cox regression on combined outcome shows a HR of 1.094 (95% CI 1.003-1.196) for unit increase in MELD considered as a continuous variable. The predictive role is independent by the effect of covariates considered in the analysis such as age, sex, NYHA class, EF, TAPSE, PASP, VO2max, NT-proBNP, MELD score worsening, and NT-proBNP increase. Changes in MELD score percentage, considered as a dichotomous variable (≤100% and >100%), were found to be predictors of mortality, urgent heart transplant and LVAD implant. Receiver operating characteristic (ROC) curves showed an area under the curve (AUC) of 0.887 for MELD score and composite outcome of death, urgent transplant, and need for LVAD. The predictive performance of MELD was even superior compared with MELD-Na, MELD-XI, MAGGIC risk score, and MECKI. The MELD score and its longitudinal changes are effective predictors of adverse outcomes in AHF.