Search

Metabolic dysfunction-associated steatohepatitis (MASH) is a major cause of liver-related morbidity and mortality, yet treatment options are limited. Manual scoring of liver biopsies, currently the gold standard for clinical trial enrollment and endpoint assessment, suffers from high reader variability. This study represents the most comprehensive multisite analytical and clinical validation of an artificial intelligence (AI)-based pathology system, AI-based measurement of metabolic dysfunction-associated steatohepatitis (AIM-MASH), to assist pathologists in MASH trial histology scoring. AIM-MASH demonstrated high repeatability and reproducibility compared to manual scoring. AIM-MASH-assisted reads by expert MASH pathologists were superior to unassisted reads in accurately assessing inflammation, ballooning, MAS ≥ 4 with ≥1 in each score category and MASH resolution, while maintaining non-inferiority in steatosis and fibrosis assessment. These findings suggest that AIM-MASH could mitigate reader variability, providing a more reliable assessment of therapeutics in MASH clinical trials.

The gold standard for enrollment and endpoint assessment in metabolic dysfunction‐associated steatosis clinical trials is histologic assessment of a liver biopsy performed on glass slides. However, obtaining the evaluations from several expert pathologists on glass is challenging, as shipping the slides around the country or around the world is time‐consuming and comes with the hazards of slide breakage. This study demonstrated that pathologic assessment of disease activity in steatohepatitis, performed using digital images on the AISight whole slide image management system, yields results that are comparable to those obtained using glass slides. The accuracy of scoring for steatohepatitis (nonalcoholic fatty liver disease activity score ≥4 with ≥1 for each feature and absence of atypical features suggestive of other liver disease) performed on the system was evaluated against scoring conducted on glass slides. Both methods were assessed for overall percent agreement with a consensus "ground truth" score (defined as the median score of a panel of three pathologists' glass slides). Each case was also read by three different pathologists, once on glass and once digitally with a minimum 2‐week washout period between the modalities. It was demonstrated that the average agreement across three pathologists of digital scoring with ground truth was noninferior to the average agreement of glass scoring with ground truth [noninferiority margin: −0.05; difference: −0.001; 95% CI: (−0.027, 0.026); and p < 0.0001]. For each pathologist, there was a similar average agreement of digital and glass reads with glass ground truth (pathologist A, 0.843 and 0.849; pathologist B, 0.633 and 0.605; and pathologist C, 0.755 and 0.780). Here, we demonstrate that the accuracy of digital reads for steatohepatitis using digital images is equivalent to glass reads in the context of a clinical trial for scoring using the Clinical Research Network scoring system. [ABSTRACT FROM AUTHOR]