PathAI Reports that Its ML-based NASH Drug Discovery Tool May Identify Clinical Trial Responders Based on Post-Hoc Analysis of Bristol Myers Squibb’s FALCON 1 Study at The Liver Meeting 2021
Post-hoc evaluation of liver biopsies from patients in the Bristol Myers Squibb sponsored FALCON 1 study by AI-based histologic measurement of NASH (nonalcoholic steatohepatitis; AIM-NASH) drug development tool (DDT) suggests that clinical trial endpoints may have been met and shows treatment-associated improvements in key liver tissue features not identified by manual assessment.
Boston, Massachusetts - PathAI, a global provider of AI-powered technology applied to pathology, will announce results from a retrospective analysis of liver biopsy specimens from Bristol Myers Squibb’s FALCON 1 study, a Phase 2b, randomized, multicenter, placebo-controlled study assessing the efficacy and safety of pegbelfermin (PGBF) as a treatment for non-alcoholic steatohepatitis (NASH) at The Liver Meeting, November 12-15, 2021 (NCT03486899).
This exploratory post hoc analysis compared machine learning (ML)-based quantification of histological features with traditional pathology scoring methods, and the results will be presented in the poster Shevell et al., Comparison of manual vs machine learning approaches to liver biopsy scoring for NASH and fibrosis: a post hoc analysis of the FALCON 1 study.
PathAI has developed the AI-based histologic measurement of NASH Drug Development Tool (AIM-NASH DDT) that has been accepted into the FDA Biomarker Qualification Program. The AIM-NASH DDT is intended for use in assessment of endpoints in clinical trials as well as clinical trial enrollment after FDA qualification. AIM-NASH has been trained to detect and quantify the key histological features required to score NASH disease severity using the standard NASH CRN scoring system and generates slide-level scores for those features (lobular inflammation, ballooning, steatosis, and fibrosis) mirroring the standard pathology workflow. In this study, biopsy slides, collected from clinical trial participants within 6 months prior to or during the screening period and after 24 weeks of PGBF treatment, were digitized into whole slide images and evaluated using AIM-NASH. The clinical study central pathologist manually scored these same biopsy samples during the study period.
The FALCON 1 trial had 197 participants randomized to four arms: placebo, plus three treatment arms of PGBF dosed at 10mg, 20mg, and 40mg.
Evaluating the primary clinical trial endpoint of ≥1 stage NASH CRN fibrosis improvement without NASH worsening or NASH improvement without fibrosis worsening at 24 weeks, identified a statistically significant proportion of responders in the treatment arms by the AIM-NASH DDT (p=0.013) that were not reported by manual assessment (p=0.148).
AIM-NASH-based and manual scores for all CRN components showed distinct trends of improvement in all PGBF arms compared to placebo. The AIM-NASH DDT CRN scoring revealed significant improvements in ballooning (p=0.033) and lobular inflammation (p=0.019) in the treatment arms compared with placebo that were not seen by manual scoring (ballooning p=0.274; lobular inflammation p=0.716). Conversely, manual methods showed significant improvements in steatosis for treated patients (p=0.0022) that AIM-NASH did not (p=0.106). Treatment-associated improvements in fibrosis were not seen using either method. Additional assessment by AIM-NASH using a continuous scoring method showed significant differences between placebo and PGBF treated patients for ballooning (p=0.0014), lobular inflammation (p=0.05), and steatosis (p=0.001).
While this study suggests that AIM-NASH-based pathologic assessment of tissue may be more sensitive than manual assessment and may capture changes in histology that could be indicative of drug efficacy, further analyses with larger tissue datasets are required to further support these claims.
PathAI is a leading provider of AI-powered research tools and services for pathology. PathAI’s platform promises substantial improvements to the accuracy of diagnosis and the efficacy of treatment of diseases like cancer, leveraging modern approaches in machine and deep learning. Based in Boston, PathAI works with leading life sciences companies and researchers to advance precision medicine. To learn more, visit pathai.com.