ZHANG Zhan-qing,LU Wei,WANG Yan-bing,et al.Logistic regression model for prediction of liver tissue pathological status in patients with hepatitis B[J].同济大学学报（医学版）,2015,36(1):50-57. [点击复制]
(Dept.of Hepatology, Shanghai Public Health Clinical Center of Fudan University, Shanghai 201508, China;Dept.of Clinical Laboratory, Shanghai Public Health Clinical Center of Fudan University, Shanghai 201508, China;Dept.of Pathology, Shanghai Public Health Clinical Center of Fudan University, Shanghai 201508, China)
Abstract:
Objective To construct a Logistic regression model based on age and serum HBsAg, HBV DNA for prediction of liver tissue pathological states in patients with chronic hepatitis B. MethodsTotal 472 consecutive chronic hepatitis B patients with pathological diagnoses of liver tissues, including 279 HBeAg-positive and 193 HBeAg-negative patients, were enrolled in present study. Serum HBsAg and HBeAg were determined by Abbott Architect I2000 and auxiliary reagents, serum HBV DNA was determined by real-time fluorescence quantitative PCR. SPSS 13.0 software was used for statistical analyses. ResultsIn HBeAg-positive patients, serum HBsAg and HBV DNA were negatively correlated with pathological grading and staging(P<0.05). In HBeAg-negative patients, serum HBV DNA was positively correlated with the pathological grading and staging(P<0.01). In both HBeAg-positive and HBeAg-negative patients, the area under ROC curve of the regression model for predicting different pathological states was significantly larger than that of the diagonal reference(P<0.01). In HBeAg-positive patients, the optimal cut-offs of predictive probability in the regression model for predicting advanced fibrosis and serum HBsAg for diagnosis of advanced fibrosis were ≥0.185 and ≤3.797 log10IU/ml respectively, and the corresponding sensitivities, specificities, accuracies were 0.886,0.646,0.706 and 0.800,0.660,0.695 respectively. In HBeAg-negative patients, the optimal cut-offs of predictive probability in the regression model for predicting significant fibrosis and serum HBV DNA for diagnosis of significant fibrosis were ≥0.603 and ≥3.095 log10IU/ml respectively, and the corresponding sensitivities, specificities, accuracies were 0.636,0.720,0.668 and 0.669,0.653,0.663, respectively. ConclusionLogistic regression models based on age and serum HBsAg, HBV DNA can enhance the efficacy of serum HBsAg and HBV DNA for diagnosis of pathological states in patients with chronic hepatitis B.
