High content imaging in primary models for near-patient drug repurposing and drug discovery

It is currently difficult to accurately predict efficacy and toxicity of emerging drugs in the clinic using conventional models of disease, particularly at the level of individual patients. This makes it challenging to nominate drug candidates to clinical studies based on accumulated preclinical data, and to match patients to these treatments. In the ideal scenario, cells from multiple patients or donors would be used early in the drug discovery pipeline to understand drug response in models that accurately reflect both disease biology and variability. Furthermore, the same readouts of efficacy and safety could follow each drug into the clinic as a companion precision diagnostic. However, there are currently few methods that enable use of primary material at a scale and cost, such that they can realistically support the full life cycle of drug discovery project at an early stage. In this talk we describe our efforts to bring highly miniaturized primary 3D models to drug discovery. We use microfluidics to generate 3D structures composed of 80-100 cells and high content imaging to characterize drug response. We have preliminary data that miniaturized liver models generated using this approach retain albumin secretion and metabolic activity characteristic of the primary cell phenotype. In addition, we touch upon similar development of primary models and imaging assays for drug profiling cells from ovarian cancer patients. We call this platform Drug Efficacy Testing in Ex Vivo Cultures (DETECt). In this setting we can discriminate between patients with a progression-free interval > 12 months and < 12 months based on the drug response score for the standard of therapy treatment, carboplatin, measured using the DETECt platform.