Abstract 318: Enhancing drug discovery and development throughput without sacrificing predictivity: ex vivo 3D drug response profiling (DRP) using patient-derived xenografts (PDX)

Abstract

Abstract Background: PDX have become critical elements of preclinical drug development as they better reflect the heterogeneity, molecular and histopathologic signatures of the original tumor than cell lines or genetically engineered mouse models, and their drug response profiles correlate with clinical response. While PDX models have become a powerful tool in drug discovery and development, limitations include low throughput for broad drug screening, lack of dose-response curves, high cost and progressive loss of human-derived stromal elements over serial passages, restricting utility for certain therapeutic classes. A potential mechanism to overcome the low throughput and high cost of PDX models is the incorporation of ex vivo 3D (EV3D) DRP on cells isolated from early passage PDX models. Thus, we correlated DRP results using PDX with genetic mutations and drug response of PDX tested in vivo. Materials & Methods: Cells were isolated from low-passage triple negative breast, invasive bladder, and non-small cell lung PDX tumors propagated in NSG mice and cultured as 3D spheroids. 3D spheroid cultures were exposed to 15 clinically-relevant chemotherapy and targeted agents and assayed for cell viability over a range of concentrations. Non-linear regression curves were generated and relative IC50s estimated. In vivo response with limited numbers of agents at clinically relevant concentrations (3 including controls) was assessed. Results: 3D cultures and testing were successfully established across all PDX and IC50s were successfully generated in 98% of drugs tested. EV3D DRP of PDX tumors differentiated activity of cytotoxic and targeted agents across tumors of similar histologic site of origin. Gemcitabine (IC50 = .007 versus 27 uM) and docetaxel (0.2 versus 40uM) activity was highly correlated with in vivo response in bladder and breast cancers, respectively, whereas cisplatin was equally active across all tumor types (IC50 = 3-8uM). hENT1 mRNA expression was not predictive of gemcitabine activity. EV3D DRP data correlated with PDX and clinical outcome. It identified Erlotinib as being relatively inactive (3 uM) against lung cancer PDX with an EGFR e19del, T790M mutation which correlated with the outcome seen in the PDX mouse and the clinical patient outcome in which the patient became nonresponsive to erlotinib. Trametinib was highly active against lung cancer PDX with a KRAS G12C mutation (IC50 6.7 × 10-6 versus 1.1 × 10-3) and will be used to perform efficacy studies in the KRAS mutant lung PDX model Conclusions: EV3D DRP predicts in vivo response and correlates with pathway activating mutations. EV3D DRP using PDX may represent a novel high throughput and predictive drug response platform that enables compound ranking for preclinical and clinical applications. Citation Format: Tessa M. DesRochers, Christina Mattingly, Stephen Shuford, Matthew Gevaert, David Orr, Carol Bult, Susie Airhart, Mingshan Cheng, Minan Wang, James Keck, Howland Crosswell. Enhancing drug discovery and development throughput without sacrificing predictivity: ex vivo 3D drug response profiling (DRP) using patient-derived xenografts (PDX). [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 318. doi:10.1158/1538-7445.AM2015-318