Prevalence, risk factors and causes of discordance in fibrosis staging by transient elastography and liver biopsy

Abstract

Liver stiffness measurement (LSM) by transient elastography (TE) is widely used for the noninvasive assessment of fibrosis. Our objectives were to examine the prevalence, risk factors and causes of discordance between fibrosis estimated by TE and liver biopsy. Two hundred and fifty-one patients with hepatitis B, C and nonalcoholic fatty liver disease underwent LSM by TE and liver biopsy. Predictors of discordance (≥2 fibrosis stages) between measures, which occurred in 14% of patients (n=35), were identified by comparing patient, TE and biopsy characteristics of discordant and nondiscordant cases. According to predefined criteria, 40% of discordances were attributed to TE error and 23% to biopsy error; 37% were indeterminate. In multivariate analysis, mild fibrosis (F0-2 vs. F3-4), and higher body mass index (BMI), ALT and LSM variability [assessed by the ratio of the interquartile range to median LSM (IQR/M)] were independently associated with discordance. Discordance was three-fold more common in patients with obesity (28 vs. 9%), ALT ≥60 U/L (20 vs. 7%) and IQR/M ≥0.17 (22 vs. 7%; all P<0.005). Based on these variables, a discordance risk score assigning 1 point to each factor was developed. The prevalence of discordance in patients with 0, 1, 2 and 3 factors were 2, 7, 20, and 55% respectively (P<0.0005). Discordance between liver fibrosis estimated by TE and biopsy occurs in one in seven patients. In assessing the validity of TE results, clinicians must recognize risk factors for discordance and in at-risk patients, consider alternative measures including biomarkers and possibly biopsy.