Automated fibrosis phenotyping of liver tissue from non-tumor lesions of patients with and without hepatocellular carcinoma after liver transplantation for non-alcoholic fatty liver disease

Background Fibrosis is the most important pathological feature in predicting development of Hepatocellular carcinoma (HCC). However, the incidence rate of HCC in patients with non-alcoholic fatty liver disease (NAFLD) is relatively low. We evaluated phenotypic histological features to differentiate HCC from non-HCC in patients with non-tumor lesions of cirrhotic livers. Methods Seventeen patients with NAFLD who underwent liver transplantation were enrolled. FibroNest was used to quantify histological phenotypes of non-tumor fibrosis lesions. Quantification included collagen content and structure traits, fiber morphometric traits, and fibrosis architecture traits. Each trait was described by up to seven quantitative fibrosis traits (qFTs). Among the qFTs measured in each specimen, those that described most of the variability between consecutive groups were automatically detected and combined into a normalized Phenotypic Composite Fibrosis Score (Ph-CFS). We trained FibroNest to identify the principal traits that differentiate HCC from non-HCC. Results HCC was found in 8 cases and non-HCC in 9 cases. The Ph-CFS significantly differentiated HCC from non-HCC (4.6 vs. 5.9, p < 0.05). Individual qFTs for morphometric features including collagen fiber length, width, perimeter, and area denoted significant differences between HCC and non-HCC. The Ph-CFS could be used to distinguish HCC (Ph-FCS < 5.0) from non-HCC (Ph-FCS ≥ 5.0) with 75% sensitivity and 100% specificity. Conclusion In patients who underwent liver transplantation, fibrotic histological phenotypes in non-tumor lesions appeared to be different between HCC and non-HCC. Phenotypic analysis of collagen in non-tumor lesions might be an effective and automated method to distinguish HCC from non-HCC on histopathology imaging.