S3-E14.1 - How Ballooned Hepatocyte Analytics Weaken NASH Clinical Trials

Send us Fan MailOne major discussion at NASH-TAG this year was about the inconsistency in ballooned hepatocyte identification and how this inconsistency inflates screen fail rates and possibly placebo response across studies.This conversation explores that issue in detail. It begins with Quentin Anstee eloquently laying out the challenge that led to this study and the approach it took.The challenge: recent studies have produce Kappa values for coding NAS scores in the 0.55 - 0.60 range. These scores suggests a level of error-related noise that might swamp a signal that a drug actually works.The solution: nine of the world's leading hepatopathologists (who, Quentin notes, were "amazingly generous with their time") scored ten different liver biopsies by drawing circles around the slides they determined were ballooned cells and then repeated the process 90 days later in a different order, with some mirrored or rotated.First finding: Concordance was "relatively modest." There was only one slide in the entire exercise where all nine pathologists found a ballooned hepatocyte. This clearly suggested that relying on a measure of the number of ballooned cells on a slide as a measure of disease was likely to produce inconsistent results due to what Quentin describes as "the lack of a common vocabulary" or vision between the pathologists.After noting how small a sample a single cell was, Stephen Harrison asks whether we would be served better if we evaluated the volume of ballooned cells on a slide, rather than the absolute number.As the conversation ends, Quentin notes analyzing a slide looking for ballooned cell is a much more complex procedure than nonhepatologists previously appreciated. He suggests that if we are looking for an analysis that is quantitative and consistent, volume may provide more robust results than looking at an actual number.The episode and this conversation are sponsored by HistoIndex. Conversation 14.5 is a discussion of how artificial intelligence driven assistive technology can improve the consistency of ballooned hepatocyte scoring in advanced fibrosis and support development of robust outcomes for fibrosis studies.