5,6,8,9,13,14 Statistical analysis was carried out using the selleck compound SPSS V15.0 (SAS Institute Inc., Cary, NC, USA) software. The demographic characteristics and histological data of the training and validation cohorts of 532 chronic HBV carriers are summarized in Table 1. The patients in the validation cohort were younger than those in the training cohort and had milder fibrosis. There was no significant difference between the training cohort and the validation cohort in the degree of necro-inflammation and fibrosis. Correlations of fibrosis staging from biopsy and routine demographic or laboratory markers were evaluated by Spearman rank correlation coefficient

in the training cohort. GGT (r = 0.299), globulin (GLO) (r = 0.178), serum total bilirubin (STB) (r = 0.154), Age (r = 0.150), indirect bilirubin (IB) (r = 0.146), prothrombin time (PT) (r = 0.123), Selleckchem Palbociclib alkaline phosphatase (ALP) (r = 0.122) and direct bilirubin (DB) (r = 0.117) were positively correlated with fibrosis staging, while PLT (r = −0.221), ALB (r = −0.167), red blood cell count (RBC) (r = −0.131) and white blood cell count (WBC) (r = −0.120) were

negatively correlated with fibrosis staging. Correlation coefficient r of each marker was significant (P < 0.05). In the training cohort, routine markers associated with the presence of significant fibrosis (S2-4) were assessed by univariate analysis. The diagnostic value of each single marker was assessed by calculating the area under the ROC (AUROC) (Table 2). Univariate analysis showed that Age, PLT, GGT, ALP, STB, IB and ALB were able to predict significant fibrosis in the training

cohort (P < 0.05). But all of them failed to achieve an AUROC better than 0.6, except GGT (AUROC = 0.660), indicating that significant fibrosis was difficult to be distinguished effectively by single routine markers. Accordingly, ALB, selleck products PLT, GLO, GGT, WBC, ALP, RBC, HB, DB and AST were identified as predictors of cirrhosis by univariate analysis (P < 0.05). The patients in the training cohort were divided into different groups according to three study endpoints: significant fibrosis, advanced fibrosis and cirrhosis (S0-1 vs S2-4, S0-2 vs S3-4 and S0-3 vs S4). Logistic regression was carried out to identify independent factors associated with each endpoint. The regression models and their diagnostic performance are shown in Table 3. All the three marker panels could predict the degree of fibrosis with a higher degree of accuracy than individual markers. Regression function 1 performed best in predicting significant fibrosis (S0-1 vs S2-4, AUROC = 0.694), function 2 performed best in predicting advanced fibrosis (S0-2 vs S3-4, AUROC = 0.744), function 3 performed best in predicting cirrhosis (S0-4 vs S4, AUROC = 0.838).