Edward W. Lee, MD, PhD
Sodium does not improve prognostic value of model for end stage of liver disease (MELD) in TIPS patients with refractory ascites
Transjugular intrahepatic portosystemic shunt (TIPS) may be used to treat complications of end stage liver disease, such as refractory ascites and variceal hemorrhage. Several studies have validated both the Model for End-Stage Liver Disease (MELD) and MELD-Sodium (MELD-Na) scores for predicting mortality following TIPS.1-3 However, prior research directly comparing prognostic value of MELD-Na versus MELD are limited and have revealed conflicting results.1,2,4-7 This single center retrospective study aims to compare the prognostic utility of MELD and MELD-Na scores for post-TIPS survival, with the hypothesis that inclusion of sodium does not improve model accuracy within the refractory ascites population.
Retrospective electronic medical records at a single liver transplant center were reviewed during the years 2007 to 2019. A total of 203 patients were included in this study for analysis. The standard TIPS technique was used in all cases.8 Correlation between MELD or MELD-Na with post-TIPS survival was performed using logistic regression and cox proportional hazards (CPH) models. To directly compare model fit, Akaike Information Criteria (AIC) and area under the curve (AUC) of receiver operator characteristic (ROC) plots were calculated for 30 day, 90 day, 1 year, and all time survival. Statistical tests were performed with R (RStudio Team v2.1, 2015).
Higher scores in both MELD and MELD-Na significantly correlated with increased risk of post-TIPS mortality at 30 days, 90 days, and 1 year in logistic regression models and all time survival in CPH models. The effects of MELD and MELD-Na were greatest on 90 day mortality with associated odds ratios of 1.15 (1.08 – 1.25) and 1.12 (1.05 – 1.21), respectively. ROC AUC was larger for MELD relative to MELD-Na at all logistic regression model endpoints, and was statistically significantly greater in 30 and 90 day mortality models. Relative to MELD-Na, AIC comparisons demonstrated significant improvement in fit for modeling 90 day and all time mortality using MELD.
MELD score is superior to MELD-Na in predicting 30 and 90 day mortality and all time survival in patients undergoing TIPS for refractory ascites.