Abstract: Objective To evaluate the efficacy of different prognostic scoring models in predicting the postoperative survival rates of patients with acute-on-chronic liver failure (ACLF) after liver transplantation, and to provide a basis for optimizing the decisions of liver transplantation. Methods A retrospective study was conducted on 131 ACLF patients who underwent liver transplantation from June 2018 to October 2023. The clinical data of all these patients were collected, and the following scoring models were calculated: model for end-stage liver disease (MELD) series (MELD, MELD-Na, MELD 3.0), chronic liver failure consortium organ failure (CLIF-C OF), chronic liver failure consortium acute on chronic liver failure (CLIF-C ACLF), transplantation for ACLF grade 3 model (TAM), and Sundaram ACLF-liver transplantation-mortality score (SALT-M) . The predictive efficacy of each scoring model for 90-day and 1-year postoperative survival was analyzed using receiver operating characteristic (ROC) curves method. DeLong tests were used to compare the areas under the ROC curves (AUROC). Kaplan-Meier analysis was performed to assess the prognostic differences stratified by SALT-M scores, and the correlation between SALT-M and CLIF-C ACLF was analyzed. Results Among the seven scoring models, the AUROC values of CLIF-C OF, CLIF-C ACLF, and SALT-M for predicting 1-year postoperative survival are 0.676, 0.695, and 0.680, respectively. However, there was no significant difference in the predictive efficacies among these three models (P>0.05). A weak to moderate positive correlation was found between SALT-M and CLIF-C ACLFs (R=0.319, P<0.001). After stratification by SALT-M tertiles, it was shown that the 1-year survival rate in the T3 group (≥9.34) was significantly lower than that in the T1 group (≤7.43) and T2 group (7.44~9.32) (P=0.015; P=0.028). Conclusion The CLIF-C OF, CLIF-C ACLFs, and SALT-M scores demonstrated a certain predictive value for postoperative survival in ACLF patients after liver transplantation. The SALT-M score is simple and practical but has limited stratification ability for patients with intermediate and low risks. It is recommended to combine with the dynamic changes when scoring to improve their predictive accuracy. Further large-scale studies are needed to validate the discriminative ability for extremely high-risk populations. This study provides theoretical support for individualized decision-making for liver transplantation and postoperative management.

Key words: Acute-on-chronic liver failure, Liver transplantation, SALT-M score, Prognostic scoring models, ROC curve