| Objective: A noninvasive diagnostic model consisting of common serological indicators was established to evaluate its diagnostic value for liver fibrosis in chronic hepatitis B and to verify its effectiveness in antiviral therapy.Methods: A group of 326 patients were randomly divided into a modeling group and a verifying group according to 6:4.All the clinical data between the modeling group and the verifying group were analyzed by the rank sum test.In the modeling group,the parameters related to significant liver fibrosis were screened out by Mann-Whitney U rank sum test.Noninvasive diagnostic models were established using binary logisitc regression analysis.The ROC curve is used to judge the diagnostic value of the model in the modeling group and the verification group.MedCalc software is used to compare the diagnostic efficiency of the model with the classic models APRI and FIB-4.Finally,the patients were divided into the HBeAg negative CHB group and the HBeAg positive CHB group,and the ROC curve was used to verify the diagnostic efficiency of the model.Result:1.1.The general clinical data of the modeling group and the validation group were analyzed.The results showed that there was no significant difference in the clinical data between the modeling group and the validation group(P > 0.05).1.2.In the model group,nine parameters of age,PLT,TBIL,ALT,AST,GGT,ALP,ALB and Ln(HBV-DNA)were analyzed by Mann-Whitney U rank sum test in non-significant liver fibrosis And significant difference between the patients with liver fibrosis(P <0.05).1.3.Binary logistic regression analysis was performed on the above nine statistically significant parameters and independent predictors of significant hepatic fibrosis were obtained using the entry method(P <0.05,excluding: P>.1)Binary Logistic Regression Equation is established with parameters and their regression coefficients.The equation is as follows: M = 0.091 * age-0.009 * PLT + 0.239 * TBIL-0.099 * ALB-0.131 * Ln(HBV-DNA)-1.726.1.4.The ROC curve was used to evaluate the diagnostic value of the model in the model group.The AUC was 0.896 [SE 0.023,95% CI0.851-0.941],the optimal diagnostic cutoff was-1.18,SN87.0%,SP83.1%,Youdens index was 0.701,PPV86.36%,NPV85.71%.The AUC of predictive significant liver fibrosis in the validation group was 0.750,SE 0.043,the optimal diagnostic cutoff was-1.54,95% CI(0.665-0.835),SN71.6%,SP71.0% Youdens index was 0.426,PPV was 73.53% and NPV was 69.50%.1.5.The Medcalc software is used to compare the APRI,FIB-4 and the AUC of the model M.The AUC values of the models M,APRI and FIB-4 were 0.896,0.657 and 0.651 respectively.Compared with AUC of APRI,the difference between M model and APRI was statistically significant(Z:5.777,P < 0.0001).Compared with AUC of FIB-4,the difference between M model and FIB-4 was statistically significant(Z:6.215,P < 0.0001).2.1.In the HBeAg positive CHB,the AUC curve of the ROC curve used in the prediction of antiviral therapy is 0.835,the best diagnosis cut-off point is-1.85,95%CI(0.758-0.859),SN80.9%,SP75.6%,Youdens index is 0.565,PPV80.00% and NPV86.36%.2.2.The model has a AUC value of 0.856 for predicting HBeAg in negative CHB,and the best diagnosis cut-off point is-0.87,95%CI(0.767-0.921),SN80.0%,SP80.4%,Youdens index is 0.604,PPV88.00%,and ROC.Conclusion:1.1.The noninvasive diagnostic model we have created has high diagnostic efficiency in both the verification group and the modeling group.1.2.Compared with the classic noninvasive diagnostic model of liver fibrosis,APRI and FIB-4,the model has a better diagnostic efficiency.1.3 Our equation shows a high diagnostic efficacy in the HBeAg-positive CHB and HBeAg-negative CHB groups. |
PDF Full Text Request