Establishment and evaluation of a predictive model for short-time prognosis of patients with hepatitis B virus-related acute-on-chronic liver failure

Objective To investigate the predictive factors for short-term(12-week) survival and prognosis of patients with hepatitis B virus(HBV)-related acute-on-chronic liver failure(HBV-ACLF),and to establish a new predictive model. Methods Related clinical data were collected from 67 patients who were diagnosed with HBV-ACLF in The Affiliated Provincial Hospital of Anhui Medical University from April 2015 to August 2018,and according to their survival at 12-week follow-up after diagnosis,they were divided into survival group with 28 patients and death group with 39 patients. Their clinical data were collected,including sex,age,total bilirubin(TBil),international normalized ratio(INR),creatinine(Cr),serum sodium,platelet count(PLT),alanine aminotransferase(ALT),aspartate aminotransferase(AST),albumin(Alb),serum cystatin C(CysC),and presence or absence of acute kidney injury(AKI). The t-test was used for comparison of normally distributed continuous data between two groups,and the Wilcoxon rank-sum test was used for comparison of continuous data with skewed distribution between two groups; the chi-square test was used for comparison of categorical data between groups; logistic regression was used to perform the factorial analysis and establish a predictive model; the receiver operator characteristic(ROC) curve was used to evaluate the predictive model,and the method by DeLong et al. was used to compare the area under the ROC curve(AUC). Results The death group had significantly higher age,TBil,INR,CysC,and Model for End-Stage Liver Disease(MELD) score than the survival group,and the patients with AKI had a significantly lower short-term survival rate than those without AKI(all P < 0. 05). TBil(odds ratio [OR]= 1. 013,95% confidence interval [CI]: 1. 003-1. 024,P = 0. 014),INR(OR = 6. 857,95%CI: 1. 449-32. 449,P = 0. 015),CysC(OR = 2. 826,95% CI: 1. 001-7. 983,P = 0. 050),and PLT(OR = 0. 982,95% CI: 0. 964-1. 000,P = 0. 048) were independent predictive factors for patient survival. A TICP model was established with the combination of TBil,INR,CysC,and PLT,and there was a significant difference in AUC between the TICP model and MELD score[0. 879(95% CI: 0. 776-0. 946) vs 0. 760(95% CI: 0. 644-0. 859),Z = 2. 708,P = 0. 007]. Compared with MELD score,the TICP model had significantly better accuracy(87. 05% vs 67. 16%),sensitivity(84. 62% vs 56. 41%),and Youden index(0. 70 vs 0. 42). Conclusion TBil,INR,CysC,and PLT are independent influencing factors for short-term prognosis of HBV-ACLF patients,and the TICP predictive model with the combination of these four indices has a good value in predicting the short-term survival and prognosis of HBV-ACLF patients.