Probability-based interpretation of liver stiffness measurement in untreated chronic hepatitis B patients

Abstract

BACKGROUND: Liver stiffness measurement (LSM) by transient elastography is a popular noninvasive test of fibrosis. Traditional LSM cutoffs dichotomize patients and do not clearly indicate the confidence of diagnosis.

AIM: We derived and validated probability functions of fibrosis and cirrhosis based on LSM and determined the effect of alanine aminotransferase (ALT) on the scores.

METHODS: Consecutive chronic hepatitis B patients who underwent liver function tests, LSM, and liver biopsies at six European and Asian centers (2/3 in the training cohort and 1/3 in the validation cohort) were recruited. Binary logistic regression was performed to predict the probabilities of different fibrosis stages based on LSM and/or ALT.

RESULTS: A total of 1,051 patients were included in the final analysis (53 % with ALT ≥ 60 IU/L, 32 % F2, 20 % F3, and 24 % F4). The probability functions (LiFA-HBV score) with and without ALT adjustment closely mirrored the proportion with different fibrosis stages in both the training and validation cohorts. For a range of up to 300 IU/L, ALT maintained a weak linear relationship with LSM for each fibrosis stage (r (2) = 0.018-0.13). Based on relative integrated discrimination improvement, the addition of ALT to the LiFA-HBV score increased the correct reclassification of F3-4 and F4 by 5 and 17 %, respectively.

CONCLUSIONS: ALT increases LSM in a linear fashion in chronic hepatitis B patients at any fibrosis stage. The LiFA-HBV score accurately predicts the probability of fibrosis. ALT adjustment increases the rate of reclassification modestly and is not essential.