Aim/Introduction: In recent years, cancer immunotherapy has become one of the fastest-growing areas in cancer research. Selecting suitable and cost-effective experimental models for developing and validating immunotherapies is one of the major obstacles researchers face today. To overcome this, patient-derived tumor models are of increasing interest because they can better recapitulate many of the properties and the heterogeneity exhibited by the tumor microenvironment at a relatively low cost. Hereby, we propose a high throughput screening platform for an effective and efficient evaluation of cancer immunotherapies in patient-derived tumor models. Methods: Tumor models were established in vitro from patient-derived tumor biopsies. Established tumoroids were engineered using a luciferase-green fluorescent protein (GFP) lentivirus to generate a reporter pool. Transduced pools were enriched for GFP via flow cytometry and characterized using RNA/scRNA-seq and biomarker-based sequencing. Natural Killer (NK) cells were co-cultured with the enriched pool in various effector-to-target ratios and recorded using a live cell imaging and analysis platform. Cytotoxicity and cell health were measured by GFP intensity, luciferase activity, and caspase-based live staining. Results: A patient-derived tumoroid reporter pool was successfully generated through GFP enrichment using a flow cytometer. The killing efficiency of immune cells with various effector(E) to target(T) ratios has been successfully captured in a ratio-dependent manner via the live cell imaging and analysis platform. NK cell-mediated cytotoxicity was successfully measured through GFP intensity, luciferase, and caspase activity. Conclusions: Traditional cell line generation can be used in patient-derived tumoroid models to generate enriched reporter cell pools without selection pressure. Outside of establishing screening platforms, scientists can use this approach to efficiently engineer patient-derived tumoroid models to meet their specific research goals. Here, we used the reporter pool to develop a multiplex-killing assay to measure cell viability and toxicity. This platform can be used in a variety of immune cell workflows, providing a method that can predict tissue-specific responses, and evaluate solid tumor immunotherapies in high throughput cell-based assays. Citation Format: Andrew Tsao, Xiaoyu Yang, Garrett Wong, Vivek Chandra, Jacob Delgadillo, Lindsay Bailey Steinitz, Brittany Balhouse, Colin Paul, Jakhan Nguyen, Sybelle Djikeng, Shyanne Salen, Jason Sharp, Matt Dallas, David Kuninger. Engineering patient-derived tumors to enable high-throughput screening: Immuno-oncology workflows [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 2753.

In vitro cancer research often fails to translate to the clinic, in part due to the use of traditional 2D cancer cell lines as models, which fail to resemble primary cancer cells by a variety of measures, including high mutational burden. An emerging solution to this problem is to replace traditional cell lines with patient tissue-derived cells expanded in 3D, also known as tumoroids or cancer organoids. We developed a defined, serum- and conditioned medium-free system, GibcoTM OncoProTM Tumoroid Culture Medium, that can be used to derive stable tumoroid lines from a variety of tissue sources and maintains the phenotype and genotype of patient-derived tumor cells. By supplementing the base medium with indication-specific growth factors, tumoroid lines were derived from colorectal, lung, and endometrial cancers from both fresh surgical resections and cryopreserved primary cancer cells. To demonstrate the utility of these patient-derived cells as long-term in vitro models, colorectal and lung tumoroid lines were derived from multiple donors and cultured for up to 50 passages. Brightfield microscopy, cell counts, and next-generation sequencing were used to assess maintenance of tumoroid morphology, growth rate, gene expression patterns, and genomic mutations. Patient-derived tumoroid cultures adopted donor-specific morphologies that were maintained during long-term culture. Cell doubling time tended to stabilize within the first few passages as cultures established, was donor-dependent, and averaged around 65 hours for colorectal tumoroids - on par with that of traditional 2D cancer cell lines - and 90-100 hours for lung and endometrial tumoroids, respectively. Importantly, tumoroid lines maintained their gene expression pattern for over 20,000 human RefSeq genes during long-term culture, with correlation between initial tumor material and late-passage samples of R>0.8. Distinct molecular subtypes of colorectal cancer were preserved in cultured tumoroids. The allelic frequency of single nucleotide variations (SNVs) in 161 highly relevant cancer genes was also highly correlated (R>0.9) between uncultured tissue and late-passage tumoroids. Within SNVs, transition/transversion mutation ratios were conserved. Tumoroids were cryopreserved and recovered during this study, demonstrating that biobanking of colorectal and lung tumoroids should not impact their long-term stability. Finally, a subset of the derived colorectal and lung tumoroids were tested and shown to be tumorigenic in mice, where subsequent histology of the tumor was similar to that of in vitro cultures. Altogether, tumoroid derivation and culture in this novel medium enables the long-term preservation of patient-specific cellular genotype and phenotype, which should allow for expansion, biobanking, and performance of experimental repeats within the same patient tissue-derived cultures across labs and over time. Citation Format: Chris Yankaskas, Brittany Balhouse, Colin Paul, Shyanne Salen, Sybelle Djikeng, Pradip Shahi Thakuri, Mark Kennedy, Matt Dallas, David Kuninger. Derivation and long-term maintenance of patient-derived tumoroid lines in a defined, serum-free medium [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 160.

Cancer drug development is an extremely challenging and resource-consuming process with a dismal success rate of 5%. The high failure rate is partly due to the inadequacy of traditional 2D cell culture model to predict drug efficacy and toxicity. A highly accurate and representative system is therefore urgently needed to improve cancer drug discovery and development process. The emerging 3D cancer models can better recapitulate the in vivo tumor microenvironment and have shown closer gene expression profiles to clinical samples compared to 2D models. Another crucial element of in vitro cancer models is the source of cancer cells: patient-derived models have shown advantages to better capture tumor heterogeneity and clinically relevant genetic alterations of the tissue origin compared to traditional cancer cell lines. Several studies showed that response of patient-derived models to treatment reflects the response of patients from which the organoids are derived, suggesting they are great platform for studies of personalized medicine. Hereby, we performed a proof-of-concept high throughput multiplexed plate-reader based drug screening assay on 3D patient-derived tumoroids. Methods: Patient-derived tumoroids were sequenced for cancer relevant mutations, copy number variants, and altered gene expression using multi-biomarker targeted next generation sequencing (NGS) and RNA-seq. 3D tumoroid lines derived from three different colorectal cancer patient samples, together with a colorectal cancer cell line were treated with chemotherapy, and targeted therapy agents. The same tumoroid and cell lines were cultured and treated in 2D as experimental control. Drug response readout was multiplexed using three different plate reader-based assays, measuring the reducing power, ATP, and the release of lactate dehydrogenase from treated cells. Results: Comparable drug responses were observed from three different viability cytotoxicity assays. Different patient-derived tumoroids demonstrated differential drug sensitivity that correlates with their clinical cancer stages. Both patient-derived and cancer cell lines cultured in 3D models showed markedly increased drug resistance compared 2D adherent culture. Differential response was also observed between cancer cell lines vs patient-derived lines. Conclusion: Taken together, these findings suggest the potential advantages of 3D patient-derived tumoroid models over 2D culture and cancer cell lines for accurate prediction of drug response. Through targeted NGS and RNA-seq, patient-specific drug targets could be identified for personalized drug screening. Selected drug candidates could be further subjected to in vivo drug validation using patient-derived xenograft model generated from the same donor. In combination with other derived models or large donor banks, this workflow provides a platform for large-scale drug discovery workflows and precision medicine. Citation Format: Xiaoyu Yang, Andrew Tsao, Garrett Wong, Chris Yankaskas, Colin Paul, Brittany Balhouse, Amber Bullock, Anthony Chatman, Shyanne Salen, Sybelle Djikeng, Matt Dallas, David Kuninger. Multiplexed plate-reader based drug screening of 3D-tumoroid models [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 2750.

Tumoroid technology enables culture of patient tissue-derived cancer cells in 3D, with retention of key characteristics from the original patient tumor. However, current tumoroid culture relies on labor-intensive media formulations and culture workflows. To address these issues, we have developed the serum- and conditioned medium-free Gibco™ OncoPro™ Tumoroid Culture Medium to enable derivation and expansion of patient-derived lines from multiple cancer indications. The system has been optimized to retain key patient genotypic and phenotypic characteristics during in vitro culture in a Wnt agonist-free system that can be easily adopted and transferred between labs. To test the applicability of our system with previously established tumoroid/cancer organoid models, we procured publicly available cancer models from the National Cancer Institute Patient-Derived Models Repository (NCI PDMR). Multiple colorectal, lung, pancreas, and head and neck tumoroid models were tested, including tumoroids originally derived in conditioned medium containing Wnt-3A, R-spondin 3, and Noggin. Cells were assessed for survival, growth, and fidelity to starting material in both OncoPro Tumoroid Culture Medium and in PDMR-recommended homebrew media. Growth and morphology of tumoroid cells was monitored and was comparable between culture conditions. To test for genetic stability of cultures over time, the mutational status of organoids expanded in each condition was characterized using targeted NGS. Both SNV allelic frequency and ploidy values were conserved from initial starting material following expansion to cryopreservation-competent banks (>10e6 cells). Gene expression levels across over >20,000 human RefSeq genes were compared between culture methods and showed high (>0.9) correlation, and Wnt-related signaling pathways were not differentially regulated between media types. In select cases, tumoroid cells expanded in each condition were also analyzed by flow cytometry using optimized antibody panels and protocols. Total cell viability and expression of EpCAM, CD45, CD31, smooth muscle actin, vimentin, CDX2, CEACAM, and cytokeratin 7 was nearly identical across culture conditions. Taken together, our approach represents a simple and effective method to expand and maintain patient-derived human cancer organoids in vitro, with no clear adverse effects from moving to a Wnt-free system. Citation Format: Colin Paul, Brittany Balhouse, Chris Yankaskas, Shyanne Salen, Sybelle Djikeng, Pradip Shahi Thakuri, Matt Dallas, David Kuninger. Expansion of established patient-derived tumoroids in a novel serum-free, Wnt-free media system [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 177.

In this paper we develop convergence and acceleration theory for Anderson acceleration applied to Newton's method for nonlinear systems in which the Jacobian is singular at a solution. For these problems, the standard Newton algorithm converges linearly in a region about the solution; and, it has been previously observed that Anderson acceleration can substantially improve convergence without additional a priori knowledge, and with little additional computation cost. We present an analysis of the Newton-Anderson algorithm in this context, and introduce a novel and theoretically supported safeguarding strategy. The convergence results are demonstrated with the Chandrasekhar H-equation and some standard benchmark examples.