Introducing an optimal liver allocation system for liver cirrhosis patients

Abstract

Background: Liver transplantation (LT) is the only treatment option for patients with advanced liver disease. Currently, liver donation to these patients, considering priorities, is based on the Model for End-Stage Liver Disease (MELD). MELD score is a tool for predicting the risk of mortality in patients with advanced liver disease. However, few studies have so far been conducted in Iran on the efficacy of MELD score of these patients.

Objectives: This study reviews the present status of the MELD score and introduces a new model for optimal prediction of the risk of mortality in Iranian patients with advanced liver disease.

Patients and methods: Data required were collected from 305 patients with advanced liver disease who enrolled in a waiting list (WL) in Imam Khomeini Hospital from May 2008 to May 2009. All of the patients were followed up for at least 3 years until they died or underwent LT. Cox regression analysis was applied to select the factors affecting their mortality. Survival curves were plotted. Wilcoxson test and receiver operating characteristics curves for survival predictive model were used to compare the scores. All calculations were performed with the SPSS (version 13.0) and R softwares.

Results: During the study, 71 (23.3%) patients died due to liver cirrhosis and 43 (14.1%) underwent LT. Viral Hepatitis (43.7%) is the most common cause of end-stage liver disease among Iranian patients. A new model (NMELD) was proposed with the use of the natural logarithms of two blood serum variables (total bilirubin and albumin) and the patients' age (year) by applying the Cox model: NMELD = 10 × (0.736 × ln (bilirubin) - 1.312 × ln (albumin) + 0.025 × age + 1.776).

Conclusions: The results of the Wilcoxon test showed that there is a significant difference between the usual MELD and our proposed NMELD scores (P < 0.001). Receiver operating characteristics curve for survival predictive model indicated that the NMELD score is more efficient compared with the MELD score in predicting the risk of mortality. Since serum creatinine was not significant in NMELD score, further studies to clarify this issue are suggested.

Keywords: Allocation; End-Stage Liver Disease; Liver Cirrhosis; Liver Transplantation.

Figure 1.. Kaplan–Meier Survival Curves for CP and MELD Scores

a) Categories of CP scores: 1) CP < 6 as class A, 2) CP 7-9 as class B and 3) CP > 10 as class C b) Categories of MELD score: 1) MELD < 10, 2) MELD 10-19 and 3) MELD > 20

Figure 2.. Accuracy of the NMELD Score (Dash Line) Using the Covariates of ln (Bilirubin), ln (Albumin) and age vs. MELD (Solid Line) Score Using the Covariates of ln (Bilirubin), ln (INR) and ln (Creatinine). Lines Plot the Estimates of Incident/Dynamic AUC (t) Versus Time Under the Assumption of Proportional Hazards.

Figure 3.. Comparison of the Area Under ROC Curves (AUC) for Predicting the Risk of Mortality at 3 (AUC = 0.916), 6 (AUC = 0.811), 9 (AUC = 0.788) and 12 (AUC = 0.780) Months for NMELD Scores.

Figure 4.. Comparison of the Area Under the ROC Curves (AUC) for Predicting the Risk of Mortality at 3 (A), 6 (B), 9 (C) and 12 (D) Months Between NMELD (Dash Line) and MELD (Solid Line). AUC for Prediction of the Risk of Mortality at 3 (AUC = 0.916), 6 (AUC = 0.811), 9 (AUC = 0.788) and 12 (AUC = 0.780) Months for the NMELD Score and AUC for Prediction of the Risk of Mortality at 3 (AUC = 0.636), 6 (AUC = 0.697), 9 (AUC = 0.672) and 12 (AUC = 0.640) Months for MELD Score.