Prediction model of the progression of patients with acute deterioration of hepatitis B virus-related chronic liver disease to acute-on-chronic liver failure

This study aimed to establish a new model for predicting acute-on-chronic liver failure (ACLF) (defined by the Chinese Medical Association), which potentially occurs among patients with acute deterioration (AD) of hepatitis B virus (HBV)-related chronic liver disease (CLD). A total of 754 patients with AD of HBV-related CLD (total bilirubin (TBIL) > 51.3 μmol/L and prothrombin activity (PTA) < 60%, 40% < PTA < 60% when TBIL ≥ 171.1 μmol/L) were retrospectively analyzed and divided into a training cohort (580 patients) and a validation cohort (174 patients). The ACLF occurrence probability of these patients was statistically analyzed within 4 weeks. In the training cohort, multivariate logistic regression analysis was performed to determine the independent predictors of ACLF occurrence and to develop a new prediction model. The validation cohort was utilized to verify and evaluate the value of the new prediction model. Within 4 weeks, 9.9% of the patients progressed to ACLF (12.0 ± 6.7 days). The new prediction model was characterized by R = 3.090 + 0.035 × Age (years) − 0.050 × PTA (%) + 0.005 × TBIL (μmol/L) + 0.044 × D/T (%) − 0.072 × Na (mmol/L) + 0.180 × HBV DNA (log(10)IU/mL). The areas under the receiver operating characteristic curves of the training and validation cohorts in the new model were higher than those in the model for end-stage liver disease. The new prediction model could be used by clinicians to recognize patients with AD of HBV-related CLD with high risks of progressing to ACLF.