Introduction and Objective: In a Phase III, randomized, placebo-controlled, multi-center study in which T2DM patients were treated with oral insulin for 26 weeks, the primary endpoint defined as a decrement of 0.6% HbA1c by end of treatment was not met. A post-hoc analysis aimed to identify subpopulations with substantial clinical responses to treatment.Methods: Explainable AI (XAI) and Phase II data (n=206) collected with the same oral insulin formulation were used to train a model to estimate the expected treatment effect in the Phase III trial (n=489). The analysis also sought to identify causal interrelations among markers that could explain treatment-disease mechanisms and recommend responder profiles. Profiles were validated using real-world data (RWD) of ~2,200 T2DM patients to guarantee sufficient representation in the T2DM patient population.Results: XAI-identified demographic (e.g., age, sex), clinical (e.g., BMI), and biological (e.g., baseline HbA1c) markers provided optimal responder profiles that enabled ~1% HbA1c reduction over placebo. Profiles derived based on Phase II data were similarly accurate on Phase III data (ρ=0.96).Conclusion: Variability in response to oral insulin can be minimized by segmenting the population based on BMI and other clinical and biological markers. These findings will inform eligibility criteria for a newly planned Phase III study.
R. Eldor: Consultant; Oramed Pharmaceuticals. Speaker’s Bureau; Novo Nordisk, Lilly Diabetes, Sanofi, Medtronic. Stock/Shareholder; Oramed Pharmaceuticals. Speaker’s Bureau; Novartis Pharmaceuticals Corporation, Boehringer-Ingelheim. M. Kidron: Board Member; Oramed Pharmaceuticals. Employee; Oramed Pharmaceuticals. Stock/Shareholder; Oramed Pharmaceuticals. R.H. Wolf: Other Relationship; Oramed Pharmaceuticals. S. Chaim: None. R. Giami: Other Relationship; Oramed Pharmaceuticals. B. Lerner: None.
Source link
Leave a Reply