A Single-Center Experience Evaluating Variable Prognostic Indicators of 30- and 90-Day Mortality in Patients with End-Stage Liver Disease After Creation of Transjugular Intrahepatic Portosystemic Shunt

Abstract

Introduction: Transjugular intrahepatic portosystemic shunt (TIPS) procedures are used to treat complications of portal hypertension such as refractory ascites and variceal hemorrhage. Previous studies have shown that the model for end-stage liver disease (MELD) score is superior to other liver disease scoring systems to establish optimal candidates for TIPS procedure and liver transplantation. The purpose of our study was to analyze various prognostic indicators of 30- and 90-day mortality after TIPS creation. Methods: We conducted a retrospective chart review on cirrhotic patients who underwent TIPS procedure from June 2009 to September 2013. All TIPS revisions were excluded. Data on age, gender, underlying liver disease, indication for TIPS, blood pressure, hemoglobin, platelets, INR, serum albumin, ALT, AST, bilirubin, creatinine, glucose, and sodium upon admission prior to TIPS placement was collected. Pre- and post-TIPS, and absolute portosystemic reduction gradients were recorded. Physical examination ascertained the presence and severity of ascites and hepatic encephalopathy. Electronic medical records were reviewed to determine 30- and 90-day mortality rates. Continuous variables were compared by Student’s t-test and categorical variables by chi-square tests. In some cases, non-parametric tests were used. A logistic regression was performed to determine the effects of age, gender, presence of ascites or encephalopathy, BMI, and MELD-Na score on the likelihood of death within 90 days of TIPS procedure. Results: From June 2009 to September 2013, 69 patients underwent TIPS placement. We found MELD (p=0.026) and MELD-Na scores (p=0.014) to be significantly different between those who died within 30 days and those who did not following TIPS placement. MELD-Na score (p=0.01) was significantly different between the deceased and persons who survived the 90-day period. Our logistic regression model itself was statistically significant (p<0.05), explaining 29.6% (Nagelkerke R2) of the variance in 90-day mortality. It correctly classified 81.6% of cases. Sensitivity was 33.3%, specificity was 94.5%, positive predictive value was 62.5%, and negative predictive value was 83.6%. Of the six predictor variables, only MELD-Na score was statistically significant (p=0.028). For each point increase in MELD-Na score, chance of death increased by 1.15 times. Conclusion: Our single-center experience suggests that the MELD-Na score is the most effective predictor of survival after TIPS creation. In our study, an individual’s likelihood of death was 1.15 times greater for every 1-point increase in the MELD-Na score. The use of this data will help classify patients with endstage liver disease, establishing the best candidates for TIPS creation and liver transplantation.