Fig. 6

Identification of highly predictive fibrosis markers in patients with chronic liver diseases. A Differential expression of switching genes derived from HSC differentiation trajectory analysis in human biopsies from NAFLD patients with mild (F0–F1, n = 40) or advanced (F3–F4, n = 32) fibrosis. Average expression, standard error, and FDR-values from differential expression analysis are shown. ROC curves show the discriminatory power of each transcript and of the combination of AEBP1, ITGAV, LOXL2, and TAGLN (A-I-L-T). For validation of this four-gene combination, ROC curves were drawn using the microarray data from another NAFLD cohort with mild (F0–F2, n = 31) or significant (F3–F4, n = 8) fibrosis (B), and from HBV patients with mild (F0–F2, n = 96) or severe (F3–F4, n = 28) fibrosis (C). SE sensitivity, SP specificity, AEBP1 adipocyte enhancer binding protein 1, ITGAV integrin subunit alpha V, LOXL2 lysyl oxidase like 2, TAGLN transgelin