A New Model Score To Predict Survival In Non-malignant Complications Of Chronic Hepatitis B Virus-related Cirrhosis Of Liver

Objective: To establish a new model score, the Model for Non-malignantComplications of Liver Cirrhosis (MNC) score, by logistic regression analysis, and tocompare the predictive accuracy of MNC model score with the Model for End-stageLiver Disease (MELD) score and the Model for End-stage Liver Disease-Change(MELD-Chg) score for predicting the prognosis of patients with non-malignantcomplications of chronic hepatitis B virus-related cirrhosis of liver.Methods: A total of160patients, consisting of118males and42females, with anaverage age of62.5years (62.5±13.1years) with chronic hepatitis B virus-related livercirrhosis were retrospectively evaluated at the time of their last admission and at theirprevious admission3months earlier. Laboratory data including Alanineaminotransferase (ALT), Aspartate aminotransferase (AST), γ-glutamyl transferase(GGT), Alkaline phosphatase (ALP), total serum bilirubin, serum albumin, Red bloodcell count (RBC), White blood cell count (WBC), platelet count, Internationalnormalized ratio for prothrombin time (INR), serum urea and serum creatinine wereretrieved from each folder. Logistic regression analysis was used to construct the MNCmodel score equation. The predictive accuracy of the MNC model score, MELD-Chgscore and MELD score were compared by the area under the receiver-operatorcharacteristic curve (AUC).Results: Using3-month survival as end-point, the AUC of MNC model score,MELD-Chg score and MELD score were0.765,0.719and0.644respectively. Both theMNC model score and the MELD-Chg score had higher AUC compared to the MELDscore (X2=6.82; p=0.0331, p <0.05), the three scores showing statistical significance.The AUC of the MNC model score was18.8%higher than that of the MELD score(AUC of MNC=0.765, AUC of MELD=0.644) and the AUC of the MELD-Chg score was7.5%higher than that of the MELD score (AUC of MELD-Chg=0.719, AUC ofMELD=0.644). The AUC of MNC model score was higher than the AUC of theMELD-Chg score by4.6%(AUC of MNC=0.765, AUC of MELD-Chg=0.719).Comparison of the MNC model score with the MELD-Chg score showed the X2valuewas2.05and the two scores were not statistically significant (p=0.1523, p>0.05).Comparison between MNC model score and MELD score had an X2value of6.81, thetwo scores showed statistical significance (p=0.009, p <0.05). In the comparisonbetween the MELD score and the MELD-Chg score, the X2value was3.61and the twoscores were not statistically significant (p=0.058, p>0.05). The cut-off value with thebest predictive accuracy for the MNC model score was-0.6589, with sensitivity andspecificity of65.5%and77.1%respectively. The positive predictive value was60.0%and the negative predictive value was80.0%for the MNC model score, with accuracyof73.1%. The cut-off value with the best predictive accuracy for the MELD-Chg scorewas3.9348, with sensitivity and specificity of65.5%and74.3%respectively. Thepositive predictive value was57.1%and the negative predictive value was80.4%forthe MELD-Chg score, with accuracy of71.3%.Conclusions:1. A new prognostic score derived by logistic regression analysis incorporatingchange of serum levels of albumin, bilirubin, creatinine, sodium and INR, the Model forNon-malignant Complications of Liver Cirrhosis (MNC) score, improves the prognosticaccuracy of3-month survival in chronic hepatitis B virus infection-related livercirrhosis.2. The MNC model score shows better accuracy than the MELD score and theMELD-Chg score. The MELD-Chg score is better than the MELD score.3. With an MNC model score target of less than-0.6589, a patient with non-malignant complications of liver cirrhosis may have up to80%chance of survival.