PLoS By Category | Recent PLoS Articles

Gastroenterology and Hepatology - Pathology - Radiology and Medical Imaging

Prospective Validation of ELF Test in Comparison with Fibroscan and FibroTest to Predict Liver Fibrosis in Asian Subjects with Chronic Hepatitis B
Published: Friday, July 27, 2012
Author: Beom Kyung Kim et al.

by Beom Kyung Kim, Hyon Suk Kim, Jun Yong Park, Do Young Kim, Sang Hoon Ahn, Chae Yoon Chon, Young Nyun Park, Kwang-Hyub Han, Seung Up Kim

Background and Aims

Liver stiffness measurement (LSM) and FibroTest (FT) are frequently used as non-invasive alternatives for fibrosis staging to liver biopsy. However, to date, diagnostic performances of Enhanced Liver Fibrosis (ELF) test, which consists of hyaluronic acid, aminoterminal propeptide of procollagen type-III, and tissue inhibitor of matrix metalloproteinases-1, have not been compared to those of LSM and FT in Asian chronic hepatitis B (CHB) patients.


Between June 2010 and November 2011, we prospectively enrolled 170 CHB patients who underwent liver biopsies along with LSM, FT, and ELF. The Batts system was used to assess fibrosis stages.


Areas under receiver operating characteristic curves (AUROCs) to predict significant fibrosis (F=2), advanced fibrosis (F=3), and cirrhosis (F?=?4) were 0.901, 0.860, and 0.862 for ELF, respectively; 0.937, 0.956, and 0.963 for LSM; and 0.896, 0.921, and 0.881 for FT. AUROCs to predict F=2 were similar between each other, whereas LSM and FT had better AUROCs than ELF for predicting F=3 (both p<0.05), and LSM predicted F4 more accurately than ELF (p<0.05). Optimized cutoffs of ELF to maximize sum of sensitivity and specificity were 8.5, 9.4, and 10.1 for F=2, F=3, and F?=?4, respectively. Using suggested ELF, LSM and FT cutoffs to diagnose F1, F2, F3, and F4, 91 (53.5%), 117 (68.8%), and 110 (64.7%) patients, respectively, were correctly classified according to histological results.


ELF demonstrated considerable diagnostic value in fibrosis staging in Asian CHB patients, especially in predicting F=2. However, LSM consistently provided better performance for predicting F=3 and F4.