Pre-clinical cancer studies have been limited by the availability of patient cell models that accurately represent the diversity of patient populations. The utility of these studies is further impacted by the lack of normal tissue from the same patient. Furthermore, clinical evaluation of targeted therapeutics, like tarceva (erlotinib), would benefit from patient models expressing the target for robust characterization of drug response. Moreover, individual patient responses to the same anti-cancer compound cannot be assessed with high precision unless healthy controls are used from the same tumor donor. Here, we present six breast cancer patient-derived tumor models, accompanied by cancer (tumor)-associated fibroblasts (CAFs) and mesenchymal stem cells (MSCs). These paired and matched sample cohorts were studied in the context of spheroid models of the metastatic breast cancer niche. We show that these models allow an increased understanding of chemotherapy modulation by non-cancerous cells, which are present within the tumor but not targeted by the current chemotherapies or other anti-tumor approaches. We used a three-drug set to create a viability profile for each patient-associated tumor sample, which in turn can be used to aid clinical decision-making and increase the personalization of each treatment for an individual. Further, our approach shows different genomic and proteomic profiles for the co-culture models when compared between the tumor 3D and healthy controls. Citation Format: Liam Deems, Dmitry Shvartsman, Maria Ivanova, Cheryl Murphy, David Deems. Enabling therapeutic decisions for a breast cancer patient cohort using matched diseased and normal tissue in tumor organoids [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-06-15.

Abstract Cancer research relies on a plethora of disease-relevant tumor models, but patient specific models are still being developed. In this study, we show the utility of patient-specific pancreatic, ovarian and breast cancer models in the context of the metastatic niche, composed of co-cultured mesenchymal stem cells (MSC) in three-dimensional tumor spheroids. Additionally, different ratios of MSC's were seeded with the cancer spheroids to interrogate the effect of MSC's on chemotherapeutic drug response in defined media conditions. The patient-specific and treatment-naive cancer models represent a diverse range of disease type, progression grade, and genetic profiles. These models were co-cultured with MSC's sourced from both bone marrow and adipose tissue. These co-culture models were tested with a panel of common, broad-range chemotherapeutic agents as well as several disease-specific drugs. The resulting drug responses show that the presence of MSC's within the context of the metastatic niche causes attenuation towards a higher chemotherapeutic drug resistance. The tumor models exhibit a varied chemotherapeutic response when MSC's are seeded in 50:1, 5:1, or 1:1 ratio with cancer cells, suggesting that even a small presence of MSC's may have a significant effect on the drug resistance of metastatic tumor spheroids. While this MSC-induced drug response attenuation is observed across the broad spectrum of patient profiles and disease types, the strength of the effect is not homogenous. The observed variability in the strength of the effect was impacted by the disease type of the cancer cells, the source of the MSC's, and the specific drug used in the metastatic tumor model. We propose that using these reproducible and easily scaled models of metastatic tumors to find effective chemotherapeutic drugs could increase our understanding of the tumor microenvironment, the onset of metastasis, and the process of cancer grafting into other tissues. Discovering effective and targeted therapies utilizing a more advanced model of the metastatic tumor niche could increase the success rate of an eventual clinical trial. Citation Format: Liam Deems, Amit Shahar, Amy Pepicelli, Cheryl Murphy, Daniel Solomon, David Deems, Dmitry Shvartsman. Modeling the metastatic niche interactions between patient tumor and mesenchymal cells to identify drivers of chemotherapy drug resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 3164.

Abstract Creating patient-specific models with a high level of characterization and functionality in tumoroid modeling enables a better understanding of the disease drivers in metastatic cancers. It presents the opportunity for personalized treatment with novel therapeutic reagents in underrepresented cancers such as pancreatic ductal adenocarcinoma. Using a defined and treatment-naive patient-specific model cohort of nine patients diagnosed with pancreatic adenocarcinoma, we aimed to represent the diversity of progression grade and assay profiles. These models were tested with a panel of common, broad-range chemotherapeutic agents and characterized for genetic, proteomic, and tumoroid drug response profiles. A detailed analysis of gene expression values from the original tissue specimen, compared with cancer cultures in hypoxic, normoxic planar, and spheroid (3D) culture conditions, enabled the identification of C-MET as a critical gene that shows upregulation closely resembling the expression profile observed in the original patient biopsy and was modulated by tissue culture format. Once this potential target was identified, the pancreatic models were treated with five C-MET-targeting chemotherapeutic agents with different modes of action. Proto-oncogene receptor C-MET (HGF receptor) is a potential therapeutic target in many metastatic cancers due to its integral role in angiogenesis, proliferation, and cancer cell survival¹. We propose that these reproducible and easily scaled tumoroid models can be used to find new therapeutic targets in-vitro and provide the example of C-MET in our pancreatic cancer patient cohort. Moreover, only 3D hypoxic models closely resembled original patient tissues in their genomic expression profiles. Using these model formats, we distinguished differences between various C-MET targeting compounds and their ability to induce cell death, which correlated with their mode of action on C-MET neutralization. Identifying these targets in 3D produces new and potentially more effective drug targets where they may not have been observed solely in planar conditions with legacy cell lines. Our models of metastatic pancreatic cancer tumoroids can be readily utilized to benefit researchers and clinicians seeking new targets for patient treatment. Citation Format: Liam Deems, Maria Ivanova, Cheryl Murphy, Amit Shahar, David Deems, Dmitry Shvartsman. Identification of C-MET receptor as a therapeutic target in patient-specific tumoroid models of metastatic pancreatic adenocarcinoma allows identification of a new mode of action for its inhibitors [abstract]. In: Proceedings of the AACR Virtual Special Conference on Pancreatic Cancer; 2021 Sep 29-30. Philadelphia (PA): AACR; Cancer Res 2021;81(22 Suppl):Abstract nr PO-074.

Abstract Reproducible, scalable, and patient-specific 3D models of the tumor microenvironment are required to study tumor growth, metastasis and dormancy. In this study, patient-derived cancer cells, mesenchymal stem cells (MSCs), and optimization of the oxygen level and extracellular matrix (ECM) were incorporated in the development of a spheroid system using 5 cancer cell models (pancreatic [Panc], lung adenocarcinoma [LA], colon adenocarcinoma [CA], endometrioid ovarian [EO] and high-grade serous carcinoma [HGSC]). The cell models were isolated and scaled up using a feeder-free culture medium in either ambient oxygen (approximately 21% O2) or hypoxic (5% O2) conditions, depending on the individual model requirements. Cell migration was assessed via monolayer scratch assays, and the relative rates were compared to the T stage of the tumor from which the models were derived. To assess spheroid formation, the 5 models were cultured in ECM-containing medium. The Panc and HGSC spheroids were also generated with the addition of MSCs at ratios of 1-100 MSCs per 100 cancer cells. The relative rates of migration of the cell models in ambient oxygen (CA > EO ≈ Panc > HGSC ≈ LA) differed significantly from that in 5% O2 (CA = Panc = HGSC > EO > LA). Interestingly, the T stage of the original tumors (CA = Panc = HGSC > LA > EO) correlated more closely to the relative rate of cell migration when the cell models were cultured in 5% compared to 21% O2. When cultured in a 3D (spheroid) format containing ECM, the Panc and EO models demonstrated increased invasion into the surrounding ECM compared to the LA, HGSC, and CA models. The addition of MSCs into the LA model dramatically increased the invasive phenotype of the LA model. A seeding ratio of 1:100 MSC:LA cells enabled spheroid formation and invasion into the ECM, previously unobserved for the LA model in this format. The Panc model became extremely aggressive with respect to invasion into the ECM, at a ratio as low as 1:100 MSCs to cancer cells. In pilot studies with the Panc and HGSC models, spheroid size and invasion into the ECM increased upon adding the MSCs. Spheroid diameter increased by approximately 30%, and there was a notable increase in invasiveness of the already mobile and invasive Panc cancer cells into the surrounding ECM. We present here an efficient, patient-specific, spheroid model system representing dormant and metastatic tumor states that is suitable for studying antitumor drug response, personalized therapy, and disease mechanisms. The relative rates of cell migration positively correlated with the T stage of the original tumors in the more physiologically-relevant oxygen condition (5%), underscoring the need to optimize this parameter during model development. Continued optimization will include incorporating tissue-specific ECMs from lung, liver and pancreas, and assessing the robustness and scalability of these models for high-throughput drug screens. Citation Format: Dmitry Shvartsman, Amy Pepicelli, Liam Deems, David Deems, Elin S. Agoston, Amit Shahar. A metastatic phenotype is reproduced in spheroids containing patient-specific cancer cells and mesenchymal stem cells [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 337.