Patient-derived organoids are an emerging 3D model system that more closely recapitulate the organ functionality of the tissue of origin compared to traditional 2D cell lines. Further advantages to the organoid system include tunability i.e. adding additional cell types, adjusting matrix stiffness, and establishing nutrient gradients, in a physiologically relevant setting, as well as the ability to quickly scale up from a small initial sample. We have established an actively expanding biobank of primary and metastatic colorectal cancer (CRC) tumor organoids under normoxic (21% O2) and physioxic (5% O2) conditions, from a diverse patient population. Concurrently, we isolated and cultured matched cancer-associated fibroblasts (CAFs). With these samples we are able to generate data from a cohort that mimics the CRC population at large, including different ethnic groups, and mutational status. Here we detail a quantitative imaging platform that captures 3D morphometric information in addition to traditional live/dead readouts.
Organoids from each patient are heterogeneous in size, shape, symmetry, and various other phenotypic features. Furthermore, the perturbation of microenviromental factors i.e. drugs, CAFs, etc. cause phenotypic changes over time. Common population-based viability assays used in drug screens, such as ATP or MTT, may be inappropriate to capture the complexities of drug response since they are often single end point measurements of an entire population of cells, and do not account for phenotypic variations. Short term quantitative live cell imaging incorporates phenotypic information, and can extend the lifetime of samples, yet still require manipulating samples by using dyes. However, these dyes are phototoxic, making them less than ideal for long term live cell imaging. By restricting image acquisition to brightfield we are able to minimize manipulation of patient samples, leaving them intact and viable. The additional benefit of imaging organoids in 3D enables us to get spatial information not available from assays with a single readout.
Using our non-destructive imaging technique we can track 3D morphometric changes in the same organoid population over time. By using a flexible analysis method, and unbiased machine learning algorithms to determine relevant features we can account for these differences yet still get comparable data. Our organoid repository combined with long term image analysis and machine learning techniques provide a robust platform that is flexible enough to handle the heterogeneity seen across the patient population. This platform could be used to advance personalized medicine allowing clinicians to quickly and more accurately determine the appropriate treatment for a patient by screening their tumor prior to determining a course of treatment.
Citation Format: Erin Spiller, Roy Lau, Colin Flinders, Shannon Mumenthaler. A robust, non-destructive image analysis method for the quantitation and characterization of patient derived organoids [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4835. doi:10.1158/1538-7445.AM2017-4835