Exploration and verification of liver fibrosis non-invasive diagnostic model in patients with chronic hepatitis B

Objective: This study aimed to establish a new non-invasive diagnostic model for liver fibrosis in Chinese patients with chronic hepatitis B (CHB) and to evaluate its diagnostic accuracy. Methods: This was a prospective study on the liver histology of 240 CHB patients. Patients were divided into 2 groups: 120 patients were included for the establishment of diagnostic model (modeling group) and remaining patients for the validation. Results: A new non-invasive diagnostic model for liver fibrosis contained two parameters: P = -0.32+0.568 * LnFS (kPa)-0.035 * LnPLT (109/L), and liver fibrosis index was calculated as eP/(1+eP). The area under the receiver operating characteristic curve (AUROC) for significant fibrosis was 0.934 (95% CI = 0.865, 1.0) in modeling group (n = 120) and 0.891 ((95% CI = 0.811, 0.971) in validation group (n = 120). When the model value was <0.58, the negative predictive value was 76%, the sensitivity was 100% for significant fibrosis, the specificity was 95%, the positive predictive value was 100% and the negative predictive value was 76% with the model index >0.88. When the data of all patients were analyzed with Forns Index, Fibro index and Fibrotest model, the AUROC was 0.756, 0.705 and 0.742, respectively, which were lower than in the new model. Conclusion: The newly established non-invasive diagnostic model is good for the diagnosis of liver fibrosis in CHB patients, especially in those with significant liver fibrosis. Fibroscan has a higher diagnostic accuracy in liver fibrosis caused by CHB and can be used as an independent predictor.