Establish a predictive modeling under antiviral therapy for hepatitis B e antigen seroconversion in chronic hepatitis B
10.3760/cma.j.issn.1007-3418.2018.09.001
- VernacularTitle: 慢性乙型肝炎抗病毒治疗HBeAg血清学转换的预测模型建立
- Author:
Guowang LIU
1
;
Kecheng TANG
1
;
Qian LI
1
;
Wei LU
2
Author Information
1. Department of Critical Liver Disease, Tianjin Second People's Hospital, Tianjin 300192, China
2. Department of Critical Liver Disease, Tianjin Second People's Hospital, Tianjin 300192, China; Tianjin First Central Hospital, Tianjin Institute of Hepatology, Tianjin 300192, China
- Publication Type:Journal Article
- Keywords:
Hepatitis B virus;
Regression analysis;
Hepatitis B e antigen seroconversion;
Prognostic model;
Antiviral therapy
- From:
Chinese Journal of Hepatology
2018;26(9):641-645
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the predictive factors by demonstrating a predictive modeling under antiviral therapy for hepatitis B e antigen seroconversion in HBeAg-positive chronic hepatitis B patients.
Methods:198 cases with HBeAg-positive chronic hepatitis B were enrolled. Fatty liver, family history of hepatitis B, age, sex, drinking history, HBsAg, HBeAg, HBV-DNA levels, total bilirubin (TBil), CD4/CD8, albumin (ALB), alanine amino transferase (ALT) levels were used as a predictor variables of HBeAg seroconversion. Serological seroconversion of HBeAg was observed at 144 weeks of antiviral therapy. Predictive factors of HBeAg seroconversion was analyzed by logistic regression analysis, and the receiver operating characteristic curve was plotted.
Results:HBeAg seroconversion rate was 36.87%. Univariate analysis demonstrated that fatty liver (χ2 = 35.377; P < 0.001), family history of hepatitis B (χ2 = 15.687; P < 0.001), the levels of HBeAg (t = 5.034; P < 0.001), HBsAg (t = 3.454; P < 0.001) and HBV-DNA levels (Z = 4.651; P < 0.001) were predictor variables of HBeAg seroconversion. Multivariate analysis showed that family history of hepatitis B, fatty liver, HBV-DNA levels and HBeAg were independent predictors of HBeAg seroconversion. The established logistic regression model for HBeAg through regression analysis was logit P = 9.623-1.228 × family history of hepatitis B - 1.726 × fatty liver - 0.764 × HBV-DNA levels - 0.146 × HBeAg and area under curve was 0.875. When the cut-off value was -0.9350, the sensitivity and specificity were 92.70%, 75.50%, 83.22%, respectively.
Conclusion:Family history of hepatitis B, fatty liver, HBV-DNA levels and HBeAg may be independent predictors of HBeAg seroconversion at 144 weeks of antiviral therapy in HBeAg-positive chronic hepatitis B patients.
