Construction and validation of a novel prognostic risk scoring table for patients with acute-on-chronic liver failure
- VernacularTitle:慢加急性肝衰竭患者预后预测评分表的构建及验证
- Author:
Zhanhu BI
1
;
Haifeng HU
1
;
Hong DU
1
;
Linxu WANG
1
;
Xiaofei YANG
1
;
Yidi DING
1
;
Jianqi LIAN
1
Author Information
- Publication Type:Journal Article
- Keywords: Acute-On-Chronic Liver Failure; Proportional Hazards Models; Prognosis; Risk Factors
- From: Journal of Clinical Hepatology 2025;41(10):2102-2109
- CountryChina
- Language:Chinese
- Abstract: ObjectiveTo investigate the clinical features of patients with acute-on-chronic liver failure （ACLF）， and to construct a risk scoring table that can accurately predict the prognosis of patients in the early stage. MethodsA retrospective analysis was performed for the clinical data of 502 patients with ACLF who were admitted to Tangdu Hospital， Air Force Medical University， from January 1， 2010 to December 31， 2020 （training set）， and the influencing factors for 28-day mortality rate were identified. The 69 ACLF patients who were admitted to Tangdu Hospital， Air Force Medical University， from January 1 to December 31， 2021 were enrolled as the validation set. The independent-samples t test or the Mann-Whitney U test was used for comparison of continuous data between two groups， and the chi-square test or the Fisher’s exact test was used for comparison of categorical data between two groups. A univariate Cox regression analysis was used to obtain the early warning indicators associated with the 28-day prognosis of ACLF patients， and variance inflation factors were used to assess multicollinearity among predictors； a multivariate Cox regression analysis was used to construct a risk model for ACLF prognosis （mortality）. A risk scoring table for ACLF prognosis （mortality） was developed based on regression coefficients （β） from the model equation and weight assignments in the nomogram. Internal validation and comparison were performed for the risk model for ACLF prognosis （mortality）， the scoring table for ACLF prognosis （mortality）， and other scoring models （Child-Turcotte-Pugh ［CTP］ score， Model for End-Stage Liver Disease ［MELD］ score， MELD combined with serum sodium concentration ［MELD-Na］ score， and integrated MELD ［iMELD］ score） in the training set， while external validation and comprehensive evaluation of the scoring table and the other scoring models were performed in the validation set. The Nagelkerke’s R2 test and the Hosmer-Lemeshow test were used to assess the degree of fitting of the risk model for ACLF prognosis （mortality）， the scoring table for ACLF prognosis （mortality）， and other scoring models， and fitting curves were plotted. C-index was used to assess the discriminatory ability of the scoring table for ACLF prognosis （mortality） and the other scoring models， and the Z-test was used for comparison of C-index between different models. The decision curve analysis was used to compare the clinical benefits of the scoring table for ACLF prognosis （mortality） and the other scoring models. ResultsThe multivariate Cox regression analysis showed that age （hazard ratio ［HR］=1.027， 95% confidence interval ［CI］： 1.015 — 1.039， P<0.001）， hepatic encephalopathy grade （grade 1： HR=2.928， 95%CI： 1.463 — 5.858， P=0.002； grade 2： HR=3.811， 95%CI： 2.078 — 6.988， P<0.001； grade 3： HR=3.916， 95%CI： 1.917 — 8.001， P<0.001； grade 4： HR=6.966， 95%CI： 4.559 — 10.644， P<0.001）， an increase in total bilirubin （TBil） by ≥17.1 μmol/L per day （HR=1.771， 95%CI： 1.248 — 2.513， P=0.001）， creatinine （HR=1.005， 95%CI： 1.004 — 1.006， P<0.001）， neutrophil count （HR=1.092， 95%CI： 1.060 — 1.126， P<0.001）， and international normalized ratio （HR=1.298， 95%CI： 1.187 — 1.418， P<0.001） were independent risk factors associated with the 28-day mortality rate of ACLF patients， and a risk scoring table was constructed for ACLF prognosis （mortality）. The Nagelkerke’s R2 test showed that the risk scoring table for ACLF prognosis （mortality） had an R2 value of 0.599 in the training set and 0.722 in the validation set， which were higher than the R2 values of CTP， MELD， MELD-Na， and iMELD scores. The Hosmer-Lemeshow test showed that the risk scoring table for ACLF prognosis （mortality） had a P value of 0.280 in the training set and 0.788 in the validation set. The C-index analysis showed that the scoring table had a higher C-index than the other scoring models in the validation set （all P<0.001）， as well as a higher C-index than CTP score in the training set （P<0.001）. The decision curve analysis showed that the risk scoring table for ACLF prognosis （mortality） had higher clinical net benefits than the other scoring models. ConclusionCompared with other scoring models currently used in clinical practice， the novel risk scoring table for ACLF prognosis （mortality） constructed based on the six predictive factors of age， hepatic encephalopathy grade， an increase in TBil by ≥17.1 μmol/L per day， creatinine， neutrophil count， and international normalized ratio has a relatively high value in predicting the 28-day prognosis of ACLF patients.
