Establishment of a new predictive model for esophagogastric variceal rebleeding in liver cirrhosis based on clinical features
- VernacularTitle:基于临床特征构建肝硬化食管胃底静脉曲张再出血的新型预测模型
- Author:
Wen GUO
1
;
Xuyulin YANG
2
;
Run GAO
2
;
Yaxin CHEN
2
;
Kun YIN
2
;
Qian LI
2
;
Manli CUI
2
;
Mingxin ZHANG
2
Author Information
- Publication Type:Journal Article
- Keywords: Liver Cirrhosis; Esophageal and Gastric Varices; Gastrointestinal Hemorrhage; Logistic Models
- From: Journal of Clinical Hepatology 2026;42(1):101-110
- CountryChina
- Language:Chinese
- Abstract: ObjectiveTo establish a new noninvasive， simple， and convenient clinical predictive model by identifying independent predictive factors for rebleeding after endoscopic therapy in cirrhotic patients with esophagogastric variceal bleeding （EGVB）， and to provide a basis for individualized risk assessment and development of clinical intervention strategies. MethodsCirrhotic patients with EGVB who were diagnosed and treated in The First Affiliated Hospital of Xi’an Medical University from September 2018 to October 2023 were enrolled as subjects， and according to whether the patient experienced rebleeding within 1 year after endoscopic therapy， they were divided into rebleeding group with 93 patients and non-rebleeding group with 84 patients. Clinical data were collected and analyzed. The independent samples t-test was used for comparison of normally distributed continuous data between two groups， and the Mann-Whitney U test was used for comparison of non-normally distributed continuous data between two groups； the chi-square test was used for comparison of categorical data between two groups. A Logistic model was established based on the results of the univariate and multivariate analyses， and the receiver operating characteristic （ROC） curve and the area under the ROC curve （AUC） were used to assess the accuracy of the model. R software was used to visualize the model by plotting a nomogram， and the Bootstrap method was used for internal validation of the model. ResultsThe multivariate analysis showed that red blood cell count （RBC）， cholinesterase （ChE）， alkaline phosphatase （ALP）， albumin （Alb）， thrombin time （TT）， portal vein trunk diameter， sequential therapy， and primary prevention were independent predictive factors for rebleeding. Based on the results of the multivariate analysis， a logistic model was established as logit（P）=-0.805-1.978×（RBC）+0.001×（ChE）-0.020×（ALP）-0.314×（Alb）+0.567×（TT）+0.428×（portal vein trunk diameter）-2.303×［sequential therapy （yes=1， no=0）］-2.368×［primary prevention （yes=1， no=0）］. The logistic model （AUC=0.928， 95% confidence interval ［CI］： 0.893—0.964， P<0.001） had a better performance in predicting rebleeding than MELD score （AUC=0.603， 95%CI： 0.520—0.687， P=0.003）， Child-Pugh class （AUC=0.650， 95%CI： 0.578—0.722， P=0.001）， and FIB-4 index （AUC=0.587， 95%CI： 0.503—0.671， P=0.045）. The model had an optimal cut-off value of 0.607， a sensitivity of 0.817， and a specificity of 0.817. Internal validation confirmed that the model had good predictive performance and accuracy. ConclusionSequential therapy， implementation of primary prevention， an increase in RBC， and an increase in Alb are protective factors against rebleeding， while prolonged TT and widened main portal vein diameter are risk factors. The logistic model based on these independent predictive factors can predict rebleeding and thus holds promise for clinical application.
