Comprehensive characterization of granular fibrotic and cellular features in liver tissue enabled by deep learning models

Background & Aims: Histologic staging of metabolic dysfunction-associated steatohepatitis (MASH) requires semiquantitative assessment of hepatocellular ballooning, steatosis, lobular inflammation, and fibrosis. We hypothesize that quantitative histologic analysis will better reflect the continuous distribution of histologic features, and thus the disease biology. Methods: We developed an AI-powered digital pathology tool, Liver Explore, consisting of a suite of machine learning models that detect and classify liver tissue regions, lobular zones, cell types, and fibrosis subtypes from hematoxylin and eosin-stained whole slide images. Human interpretable features (HIFs) were extracted and computed that correspond to predicted substances. The correlation of Liver Explore HIFs with pathologist-provided MASH CRN grades and fibrosis stages, AIM-MASH-generated continuous CRN grades and stages, non-invasive biomarkers, transcriptomics, and outcomes was assessed in participants of the STELLAR-3 and STELLAR-4 trials ( NCT04052516 ). Results: Liver Explore predictions were consistent with manual pathologist annotations. Steatosis, lobular inflammation, hepatocellular ballooning, and fibrosis Liver Explore HIFs were significantly correlated with pathologist CRN grades/stages, while model-derived tissue and cell features revealed quantitative changes in the disease microenvironment as MASH progressed. Pathological and advanced fibrosis HIFs were correlated with non-invasive metrics of fibrosis and a gene signature associated with hepatic stellate cells. HIFs associated with nodular or advanced fibrosis and inflammation were associated with an increased risk of liver-related events in patients from STELLAR-3 and STELLAR-4. Conclusions: The quantitative characterization of the liver disease microenvironment by Liver Explore delivers context relevant to MASH progression beyond the resolution afforded by categorical CRN scoring, highlighting the promise of this tool for broad applications in drug development, from enhancing understanding of mechanisms of action of novel MASH therapeutics to identifying histologic biomarkers for use in clinical trials.