A Combination Of Ultrasound Elastography And Serum Biomarkers For Non-invasive Evaluation Of Liver Fibrosis In Chronic Hepatitis B

In China, a large number of persons suffer from chronic hepatic B(CHB) which is caused by the hepatic B virus(HBV). Liver fibrosis is a common feature during the course of CHB. Accurate assessment of liver fibrosis in patients with CHB is necessary not only to predict the long-term clinical course but also to determine an appropriate antiviral therapy scheme. Liver biopsy has traditionally been considered the gold standard for staging liver fibrosis. However, it remains an invasive procedure, carrying a risk of rare but potentially life threatening complications. Therefore, clinical practice urgently needs a convenient, accurate and noninvasive diagnosis method for evaluation of liver fibrosis. Non-invasive methods for assessment of liver fibrosis rely on two major techniques: serum biomarkers or imaging techniques. Ultrasound elastography, which can estimate tissue mechanical property, exhibits the advantages of real time, non invasive and repeatable. It has been widely used in clinical researches of liver fibrosis.In this paper, we proposed three new methods to combine the ultrasound elastography and serum biomarkers for the assessment of liver fibrosis. In cooperation with the Shenzhen Third People’s Hospital, we recruited 345 patients with CHB who underwent liver biopsy, biochemical tests, and medical ultrasound examinations prospectively. Using the statistical learning method, three new models combining independent predictors were thus developed for the diagnosis of clinically significant fibrosis(F≥2): the linear model based on ROC criterion(Linear), Logistic regression model(LR) and SVM model(SVM). Overall, the three combined models had a close performance for the diagnosis of F≥2. Compared to a single index ARFI or APRI, the three combined models had better diagnostic performance. The AUROCs of the three combined models were 0.923(Linear), 0.929(LR) and 0.927(SVM) respectively, significantly higher than that of APRI(0.859). Except for the AUROC of Linear, other AUROC values are significantly higher than that of ARFI(0.893). Meanwhile, the three combined models had a similar diagnosis accuracy(87.40%, 87.40% and 88.19%, respectively), higher than the single index ARFI(84.25%) or APRI(66.14%). In addition, the three combined models increased the specificity, positive predictive value and diagnostic odds ratio while compare with single index ARFI or APRI for the diagnosis of F≥2.In conclusion, this paper proposed three new non-invasive models which combines ultrasound elastography and serum biomarkers. The result show the performance improvement for staging liver fibrosis in patients with CHB. Further study is planned to collect more data from multiple clinical centers and constantly optimize the combined model. The proposed models are expected to be applied in clinical practice to replace the liver biopsy in some extent.