Abstract 3176: Biomarker identification using xenograft mouse model based clinical trial simulation and artificial intelligence data analytics

Abstract

Abstract The growing number of anti-cancer drugs available at different stages of clinical development combined with the broadening potential use of combination therapy further complexifies the early identification of companion markers, markers of synergy as well as novel indications for existing and new drug combinations. Well characterized patient derived xenograft mouse models (PDX), combined with Artificial Intelligence tools that can integrate and analyze the broad range of generated data can help address this challenge. PDX experiments can providing an opportunity to simulate a clinical assessment using multiple mice models.In this study, we developed a PDX platform combined with the KEM® Artificial Intelligence data analytics, that is based on Formal Concept Analysis, to simulate a clinical trial and identify biomarkers of response. The platform was tested on colon cancer patient derived PDX. Respectively mRECIST response and survival of respectively 21 and 26 PDXs against Oxalipaltin combined with 5-Fluorouracil and folinic acid (Folfox) was experimentally assessed against a placebo, simulating a clinical trial-like setting with 2 arms. Biomarkers of response (mRECIST) and survival were identified using KEM®, combined with statistical modelling (Cox survival-modelling). 24 candidate biomarker genes were identified including PGAP3, ERBB2, NOTCH2, WDR70, and ZNF227. Alone or combined, these biomarkers are significantly linked to an increase or decrease of the survival PDX (p ranging from 2.2e-10 to 0.048, odd-ratio ranging from 0.12 to 10.00), with the potential to be used as inclusion or exclusion biomarkers. This work demonstrates the ability of a combined PDX / Artificial Intelligence platform to simulate clinical trials and identify biomarkers of drug efficacy and synergy, thus fostering the design of precision medicine clinical trials. Citation Format: M Afshar, Francis Bichat, Olivier Duchamp, A Etcheto, Damien France, M Kindermans, Caroline Mignard, F Parmentier, Hery Ratsima. Biomarker identification using xenograft mouse model based clinical trial simulation and artificial intelligence data analytics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3176.