An increasing number of studies performing correlative drug screens on patient-derived organoids (PDO) are revealing enormous potential for these models in predicting patient response to therapy. Despite this, their future use in a clinical setting is hindered by intrinsic limitations of PDO models, namely low success rates in establishing growing cell cultures from tumor tissue samples and long return times for drug response data that fall outside timescales of clinical actionability. To address these limitations, we leveraged technological advances in emulsion-based microfluidics and droplet generators to develop MicroOrganoSpheres (MOS) - microscale environments containing tumorspheres that retain structural, cellular, and genetic properties of an individual patient’s diseased tissue. Here, we focused on generating MOS from breast cancer tissue biopsies and resections across different subtypes of breast cancer. We established over 30 unique patients’ breast MOS samples at an average of 80% success, with the majority of these samples being hormone receptor positive (ER+/PR+) - the most clinically lethal and challenging subtype to establish. We evaluated MOS dose-responses across several first line standard of care chemotherapies for breast cancer. For this, we created a semi-automated workflow paired with high-content imaging for quantifying dose responses and tracking longitudinal cell dynamics during dosing. We generated dose response data for MOS samples, being able to move from primary tissue samples to MOS to drug dosing data within a clinically relevant timescale of 2-3 weeks. We conclude that our platform demonstrates the feasibility of 1) efficiently establishing MOS from breast cancer patient tumor samples from different subtypes and 2) performing drug dosing studies on these samples with improved turnaround times that enable clinical actionability for the patient. Together, these findings provide a foundation for evaluating this technology as a diagnostic tool in future clinical settings. Citation Format: David M. Graham, Gabrielle Rupprecht, Eric D. Bankaitis, Jeremy M. Force, Wylie Watlington, Steven W. Metzger, Xiling Shen, David Hsu. A novel and rapid patient-derived organoid breast cancer platform for precision medicine [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 192.

3592 Background: Patient-derived models of cancer, such as cell lines, patient-derived organoids, and patient-derived xenografts, are useful models of patient response in the clinic. However, these models are often not clinically applicable within the time periods necessary to inform clinical decision making, as they can take weeks to months to develop. An ideal platform using patient-derived models would be generated from a core biopsy with a subsequent drug screen within 10-14 days to minimize delay in therapy. We recently reported the development of MicroOrganoSpheres (MOS) that can be used in drug screens within 14 days of obtaining a biopsy. In the current study, we use this MOS system as precision oncology platform in colorectal cancer (CRC) to identify new therapies and predict response to therapy. Methods: CRC patient tissue samples were collected under a Duke Institutional Review Board approved protocol (Pro00089222). Resections or biopsies were mechanically and enzymatically digested to obtain a single cell suspension. Cells were then plated in Matrigel at a ratio of 20,000 cells:5 µL Matrigel to establish “mini-bulk” organoid cultures. After establishment for 5-7 days, cultures were harvested with subsequent generation of MOS at a ratio of 50 cells per MOS. After growing for 3-4 days, MOS were used for dose-response curves using oxaliplatin, SN38, and 5-Fluorouracil (5-FU) as well as high-throughput drug screens with the NCI Approved Oncology Drugs Set VI library. Results: We developed and optimized a MOS pipeline on over 50 CRC specimens, including 9 primary rectal, 35 primary CRC, 12 CRC liver metastasis, and 1 CRC lung metastasis lines with a success rate of 80% and an average of 10-21days from biopsy to MOS generation. The high success of generating CRC MOS in a clinically applicable time frame led to the next phase of the project where a total of 10 CRC MOS were tested against standard of care chemotherapy agents used in CRC (oxaliplatin, irinotecan and 5-FU) as well as the NCI Approved Oncology Drugs Set VI within 14-21days of establishment. We noted a range of sensitivity of approximately 100-fold for standard of care agents. The most sensitive drugs found in the high-throughput screen were Bortezomib, Carfilzomib, and Panobinostat and the most resistant were Gefitinib, Chlorambucil, and Procarbazine hydrochloride. Conclusions: These results demonstrate that our MOS pipeline can be used as a precision oncology platform within a clinically applicable time frame to potentially guide therapy. We are now in the process of correlating drug response in MOS to patient outcome data and these findings will be presented at the annual meeting.

1107 Background: ASCO guidelines suggest using single agent chemotherapy for patients with advanced breast cancer (ABC). Single agent chemotherapy provides modest response rates in ABC, causing patients to be exposed to unnecessary toxicity without benefit. Thus, there is an unmet clinical need to develop a clinically applicable assay to guide treatment. We recently reported the use of MicroOrganoSpheres (MOS), which are gel droplets that encapsulate tumor cells creating miniature avatars of a patient’s tumor and are amenable to high throughput dispensing and drug studies. In our current study, we evaluated the feasibility of generating and dosing MOS from ABC samples as a novel drug screen platform that led to the development and enrollment of a precision oncology trial, known as MODEL-ABC. Methods: We first performed a proof-of-concept study on 21 samples from patients with ABC and generated MOS. Drug screens were then performed on these MOS across 7 standard of care (SOC) chemotherapies commonly prescribed for ABC. These data resulted in a platform for the MODEL-ABC study that enrolled patients with ABC of any ER, PR, or HER2 subtype who were eligible for single agent chemotherapy to determine the feasibility of using MOS to predict response to therapy. In this study, biopsies from lesions ≥ 2 cm were obtained as part of SOC to generate MOS and perform drug screens. The patient subsequently received single agent chemotherapy per physicians’ choice with either carboplatin, capecitabine, paclitaxel, eribulin, liposomal doxorubicin, gemcitabine, or vinorelbine. MOS were treated using the same chemotherapy as the patient along with 1-3 alternative single agent chemotherapies. Results: We generated MOS from 19/21 samples (90% success rate) across several breast cancer subtypes. Dose response curves of MOS were successfully generated across the 7 chemotherapies with a mean turnaround time of 21±7 days validating the clinical applicability of MOS and leading to the MODEL-ABC trial. As of February 14, 2023, 4 of 15 patients have enrolled onto MODEL-ABC. All biopsies provided sufficient tissue for MOS generation, and drug screens were performed in 14-28 days across 2-4 chemotherapies. Two patients received capecitabine, one patient received eribulin, and one patient received paclitaxel. Full results of MODEL-ABC including correlation between MOS drug response and patient outcomes will be reported at the meeting. Conclusions: Our platform enables efficient establishment of MOS from ABC patient samples and allows for drug dosing studies to be performed in a clinically meaningful timeframe. Our preliminary data suggests it is feasible to obtain biopsies for MOS development and perform drug screens within 14 days. These findings provided the foundation for evaluating this technology as a potential ABC diagnostic tool and warrants further clinical development in ABC. Clinical trial information: NCT04655573 .