Objective To investigate the value of indocyanine green( ICG) clearance test in evaluating the short-term prognosis of patients with hepatitis B virus( HBV)-related acute-on-chronic liver failure( HBV-ACLF),to establish a new model for prognostic evaluation based on indocyanine green retention rate at 15 minutes( ICGR15) combined with Model for End-Stage Liver Disease( MELD) score and Child-Turcotte-Pugh( CTP) score,and to provide a basis for establishing an ideal model for the short-term prognosis of HBV-ACLF.Methods A total of 97 patients with HBV-ACLF who were admitted to Department of Infectious Diseases in The Affiliated Hospital of Southwest Medical University from June 2017 to January 2019 were enrolled and related clinical data were collected. According to the primary liver disease,these patients were divided into chronic hepatitis B( CHB) group and liver cirrhosis( LC) group; according to the outcome at 3-month follow-up,they were divided into death group and survival group. ICGR15,biochemical indices,and coagulation markers were measured at the time of confirmed diagnosis,and CTP and MELD scores were calculated. The t-test was used for comparison of normally distribu-ted continuous data between groups,and the Mann-Whitney U test was used for comparison of non-normally distributed continuous data between groups. The chi-square test was used for comparison of categorical data between groups,and a Spearman rank correlation analysis was used to investigate the correlation of ICGR15 with each variable. A binary logistic regression analysis was used to establish the combined predictive models of MELD-ICGR15 and CTP-ICGR15 for HBV-ACLF. The area under the receiver operating characteristic( ROC) curve( AUC)was used to compare the accuracy of these models in predicting the short-term prognosis of patients with HBV-ACLF,and the Medcalcdcal method was used for comparison of AUC. The maximum of Youden index( sensitivity + specificity-1) was calculated based on the sensitivity and misjudgment rate corresponding to each point on the ROC curve,and the optimal cut-off value for survival or death of patients with HBV-ACLF was determined. Results There were significant differences between the CHB group and the LC group in CTP score( t =-3. 245,P= 0. 002),albumin( Alb)( t = 2. 210,P = 0. 029),alanine aminotransferase( Z =-2. 572,P = 0. 010),and aspartate aminotransferase( Z=-1. 993,P = 0. 046). Compared with the CHB group,the LC group had significantly higher mortality rate( 37. 5% vs 17. 54%,χ2=4. 893,P = 0. 027) and incidence rate of peritonitis( 90% vs 64. 91%,χ2= 7. 945,P = 0. 005). In the CHB group,there were significant differences in ICGR15,MELD score,CTP score,total bilirubin,Alb,prothrombin time activity,prothrombin time,and international normalized ratio between the patients who died and those who survived( t = 3. 036,3. 878,3. 488,2. 756,-3. 049,-3. 954,2. 713,and 2. 717,all P < 0. 05); in the LC group,there were significant differences in ICGR15,CTP score,and Alb between the patients who died and those who survived( t = 3. 476,2. 79,and-2. 274,all P < 0. 05). Regardless of the presence or absence of LC,ICGR15 was positively correlated with MELD score( r = 0. 483 and 0. 35,both P < 0. 05) and CTP score( r = 0. 35 and 0. 411,both P < 0. 05). Based on the binary logistic regression analysis,the combined predictive models of MELD-ICGR15 and CTP-ICGR15 for HBV-ACLF were established,with an AUC of 0. 820 and 0. 847,respectively. The combined predictive models had a larger AUC than each index alone( ICGR15: 0. 787; MELD score: 0. 754;CTP score: 0. 795),and the CTP-ICGR15 model had the largest AUC. ICGR15 had a high sensitivity and a poor specificity,and the combined predictive models improved the specificity of ICGR15 alone; the CTP-ICGR15 model had a higher specificity than the MELD-ICGR15 model.Conclusion ICGR15 has a high sensitivity in predicting the short-term prognosis of HBV-ACLF,and combined predictive models have a higher accuracy than the models of a single index and can improve the specificity of the prediction based on ICGR15 alone. The combined model of CTP-ICGR15 has a higher predictive value than that of MELD-ICGR15.