AIM: As liver biopsy has considerable limitations in the assessment of liver fibrosis, non-invasive models have achieved great progress in the past. However, many tests consist of variables that are not readily available, and there are few data about patients with hepatitis B e-antigen (HBeAg) negative chronic hepatitis B (CHB). The aim of this study was to develop a model using routine data to predict liver fibrosis in HBeAg negative CHB patients. METHODS: We randomly divided 349 patients who underwent liver biopsy into training (n = 200) and validation (n = 149) sets. Multivariable logistic regression and receiver-operator curve (ROC) analyses were used to develop a model for predicting both significant fibrosis (stages 2-4) and cirrhosis (stage 4) in the training set. The model was validated in 149 patients in comparison to FIB-4, Forn's, S and aspartate aminotransferase-to-platelet ratio index indices using ROC. RESULTS: Multivariable logistic regression analysis showed that the parameters of the model for predicting both significant fibrosis and cirrhosis included sex, age, prothrombin time, platelet count, cholesterol and γ-glutamyltransferase. In the training set, the areas under the ROC (AUC) for predicting significant fibrosis and cirrhosis were 0.856 and 0.956, respectively. In the validation group, the AUC for predicting significant fibrosis and cirrhosis were 0.889 and 0.937, respectively. Using the best cut-off values, significant fibrosis and cirrhosis can be accurately predicted in 40.9% and 91.3% of patients, respectively. CONCLUSION: Our model can accurately predict both significant fibrosis and cirrhosis and may decrease the need of liver biopsy in a considerable proportion of patients with HBeAg negative CHB.
AIM: As liver biopsy has considerable limitations in the assessment of liver fibrosis, non-invasive models have achieved great progress in the past. However, many tests consist of variables that are not readily available, and there are few data about patients with hepatitis B e-antigen (HBeAg) negative chronic hepatitis B (CHB). The aim of this study was to develop a model using routine data to predict liver fibrosis in HBeAg negative CHB patients. METHODS: We randomly divided 349 patients who underwent liver biopsy into training (n = 200) and validation (n = 149) sets. Multivariable logistic regression and receiver-operator curve (ROC) analyses were used to develop a model for predicting both significant fibrosis (stages 2-4) and cirrhosis (stage 4) in the training set. The model was validated in 149 patients in comparison to FIB-4, Forn's, S and aspartate aminotransferase-to-platelet ratio index indices using ROC. RESULTS: Multivariable logistic regression analysis showed that the parameters of the model for predicting both significant fibrosis and cirrhosis included sex, age, prothrombin time, platelet count, cholesterol and γ-glutamyltransferase. In the training set, the areas under the ROC (AUC) for predicting significant fibrosis and cirrhosis were 0.856 and 0.956, respectively. In the validation group, the AUC for predicting significant fibrosis and cirrhosis were 0.889 and 0.937, respectively. Using the best cut-off values, significant fibrosis and cirrhosis can be accurately predicted in 40.9% and 91.3% of patients, respectively. CONCLUSION: Our model can accurately predict both significant fibrosis and cirrhosis and may decrease the need of liver biopsy in a considerable proportion of patients with HBeAg negative CHB.