PathAI Publications

Simulated machine learning-based fibrosis assessment reveals that adjusting for steatosis content in baseline and follow-up MASH biopsies can overestimate fibrosis reduction

Written by Admin | Nov 15, 2024 5:00:00 AM

Study Background

In MASH clinical trials, the accurate and sensitive measurement of the change in fibrosis from baseline remains a challenge, exacerbated by concurrent changes in liver volume and steatotic content.

Reduction of liver volume and fat content have been hypothesized to result in apparent increase of quantitative fibrosis measurements (e.g., collagen proportionate area, CPA) due to apparent tissue compression [1].

Adjustment by subtracting steatosis content from tissue has been proposed to account for the effects of fat reduction [2].

Here, we (i) test the assumptions of steatosis adjustment using simulated follow-up biopsies, (ii) investigate changes in the relative abundance of hepatocyte subtypes using clinical trial data, and (iii) demonstrate an AI measure of fibrosis change that shows little dependence on steatosis.

 

Conference

AASLD 2024

View Poster