AI portal tract detection and characterisation for a regional analysis of steatosis and inflammation in MASLD, MASH and AIH.

AIMS: Annotation of liver biopsies for disease staging is increasingly aided by digital pathology; however, existing systems do not quantify inflammation and steatosis within an anatomical framework. We hypothesise that an artificial intelligence (AI) system that quantifies portal tracts (PT) and the anatomical distribution of steatotic vesicles and inflammatory cells will align with manual pathologist scoring and stratify liver diseases.

METHODS

In this observational, cross-sectional study, digitised images of haematoxylin and eosin-stained specimens were pooled from four independent cohorts of metabolic dysfunction-associated steatotic liver disease (MASLD) or steatohepatitis (MASH) or autoimmune hepatitis (AIH) (n=390: 89 MASLD, 238 MASH, 63 AIH). PT, steatosis, and inflammation were quantified using a proprietary AI system and scored by expert pathologists.

RESULTS

The percentage of steatosis was higher in MASH (7.5%) than in MASLD (3.2%). Lobular regions had larger steatotic vesicles (260 vs 190 μm2). AI-derived steatosis quantification correlated with manual grading (rs=0.72). The inflammatory cell number (ICN) was twofold higher in AIH than MASLD/MASH in interface (390 vs 140), portal (4600 vs 1500) and lobular (1500 vs 650) regions. Portal inflammation from manual grading correlated with ICN count at PT (rs=0.71) but not lobular regions (rs≤0.29). For equivalent grades of portal inflammation, the ICN was up to threefold higher in AIH than in MASLD/MASH (rs=0.71).

CONCLUSION

A new AI system for anatomical quantification of liver biopsy features measured variation in fat and inflammation across the lobule. It showed that inflammation burden was higher in AIH than MASLD/MASH, despite equivalent portal grades, providing objective support for histological scoring.