Workshop Opening

Uri Ben-David, Ben Siranosian, Gavin Ha, Helen Tang, Rameen Beroukhim and Todd R. Golub

The Broad Institute of Harvard and MIT, Cambridge, MA, USA

Introduction

Patient-derived models are invaluable for cancer research and cancer precision medicine. While tumor heterogeneity and genomic instability have received much attention lately, the heterogeneity and instability of cancer models – and how cancer model evolution affects cancer research – remain under-explored.

Material and Methods

We studied the genomic evolution of human cancer cell lines (CLs) and patient-derived xenografts (PDXs), using various genetic, transcriptional and phenotypic assays.

Results and Discussion

CLs
Genomic analyses of 106 CLs grown in two laboratories revealed extensive genetic diversity. Follow-up comprehensive genomic characterization of 27 strains of the common breast cancer CL MCF7 uncovered rapid genetic diversification. Similar results were obtained with multiple strains of 13 additional CLs. Genetic changes were associated with differential activation of gene expression programs and marked differences in cell morphology and proliferation. Barcoding experiments showed that CL evolution occurs as a result of positive clonal selection that is highly sensitive to culture conditions. Analyses of single cell-derived clones showed that ongoing instability quickly translates into CL heterogeneity. Testing of the 27 MCF7 strains against 321 anti-cancer compounds uncovered strikingly disparate drug response: at least 75% of compounds that strongly inhibited some strains were completely inactive in others.

PDXs
We followed the genomic evolution of PDXs throughout their derivation and in vivo propagation. We observed extensive evolution of the PDX copy-number landscapes, mostly driven by strong clonal dynamics leading to the expansion of pre-existing minor subclones. Clonal dynamics were strongest during derivation and early propagation of the models, and attenuated at later passages. Reproducible changes were observed across independent PDXs generated from the same primary tumor, indicating the involvement of selection, rather than mere genetic drift. Importantly, the rate of genomic evolution in PDXs was similar to that observed in patient-derived CLs. Tumor evolution in PDXs and in patients followed distinct trajectories. Specifically, aneuploidy events recurrently observed in primary tumors gradually disappeared in PDXs. The genomic stability of PDXs was associated with their response to chemotherapy and targeted drugs.

Conclusion

We found evidence for extensive genomic evolution in cancer models, and explored its biological origins and functional consequences. Our findings have broad implications for basic cancer research and cancer precision medicine, especially in the context of tumor “avatars”. We suggest practical ways to mitigate the risks posed by genomic evolution, and propose how to constructively build upon this phenomenon in future studies.

James H. Doroshow¹, Yvonne A. Evrard², Melinda G. Hollingshead¹, Alice Chen¹, Biswajit Das², Vivekanand Datta², Michelle M. Gottholm-Ahalt¹, Chris Karlovich², Dianne L. Newton², P. Mickey Williams²

¹ Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD, USA

² Frederick National Laboratory for Cancer Research of the National Cancer Institute, and Leidos Biomedical Research, Inc., Frederick, MD, USA

The National Cancer Institute (NCI) has developed a national repository of patient-derived models (PDMs) comprised of patient-derived tumor xenografts (PDXs), in vitro patient-derived tumor cell cultures (PDCs), cancer associated fibroblasts (CAFs), and patient-derived organoids (PDOrg). NCI has focused on generating models to complement existing PDX collections and address unmet needs in the preclinical model space.

These models serve as a resource for public-private partnerships and for academic drug discovery efforts. The PDMs are clinically-annotated with molecular information available in an easily accessible database available to the extramural community (pdmr.cancer.gov). The NCI is producing PDMs from primary and metastatic tumor tissues and blood specimens supplied by NCI-supported clinical trials and NCI-designated Cancer Centers; the models include patient data such as previous clinical therapies, confirmed histologic diagnosis, age, stage, smoking history, and self-reported race/ethnicity, and provide representative PDX histology and NextGen sequencing data for each model. In addition, wherever available germline sequence and somatic variant calls are made available.

The PDMR also accepts previously derived PDX models developed at external sites that have followed NCI’s standard operating procedures.

Currently, over 170 PDX models are being distributed (including tumor fragments appropriate for implantation or protein extraction, DNA, and RNA) to the research community; by 9/1/2018 over 40 PDC lines (often with matching PDX) and over 100 CAF lines will also become available to the public. Additional PDXs, PDCs, CAFs, and tumor organoids will be released continuously as they pass final quality control standards. Over the past year, the NCI PDMR has distributed more than 300 PDX models to the extramural research community in the United States.

The goal of the PDMR is to develop 50 or more models per common tumor type such that the size of each molecularly-characterized subgroup is sufficient to power subsequent validation and/or efficacy studies. We are also attempting to generate models from pre- and post-treatment specimens from the same patients. Finally, NCI is targeting collection of infrequently-observed histologies to advance both biological investigations and drug development efforts for under-studied malignancies. Funded by NCI Contract No. HHSN261200800001E

Botteron Catherine¹, Werno C.¹, Treitschke S.¹, Weidele K.¹, Scheitler S.¹, Polzer B.¹, Werner-Klein M.², Klein C.A.^{1, 3}

¹ Fraunhofer Institute for Toxicology and Experimental Medicine, Project Group Personalized Tumor Therapy, 93053 Regensburg, Germany

² Institute for Immunology, University of Regensburg, 93053 Regensburg, Germany

³ Experimental Medicine and Therapy Research, University of Regensburg, 93053 Regensburg, Germany

Introduction

We recently showed that dissemination of few melanoma cells to the lymph nodes is a quantitative risk factor for death from melanoma. However, the analysis of disseminated tumor cells (DTC) is limited by their extremely low frequency, resulting in a lack of in vitro/in vivo models for mechanistic studies of early systemic disease. Therefore, the objective of this study was to develop protocols to expand single DTC from lymph nodes of melanoma patients to establish better preclinical models for adjuvant cancer therapies

Material and Methods

Sentinel lymph nodes from melanoma patients were cut in halves, disaggregated and analyzed for the presence of gp100-positive DTC. The other half was used for diagnostic assessment by histopathology. Single DTC within the lymph node-derived single cell suspension were then propagated under specific sphere forming conditions and expanded by transplantation in immunodeficient mice. DTC based pre-clinical in vitro/in vivo models were then characterized on molecular level and were used for functional analyses. To test drugs on DTC in presence of a human immune system we additionally generated humanized xenograft models

Results and Discussion

We successfully established cell lines and xenografts from DTCs of 17 melanoma patients. We confirmed the origin of the cells by genomic fingerprint and analyzed the genomic profile of the expanded cells over various culture and animal passages. DTC cell lines were used for functional analysis and for testing the efficacy of targeted therapies. In addition, we set up a novel preclinical mouse model based on expanded melanoma DTC. For this, we established a human immune system in immunodeficient mice, which were then transplanted with melanoma DTC-derived cells. Interestingly, the presence of a human immune system significantly induced tumor formation, as well as dissemination of melanoma cells and improved the response to a targeted therapy compared to mice without a humanized immune system.

Conclusion

The development of DTC based pre-clinical in vitro/in vivo models enables mechanistic studies on DTCs and drug tests against the target cells of adjuvant therapies in the presence of human immune cells. These models may help to identify candidate adjuvant therapies targeting DTCs and to understand mechanisms of drug resistance of currently applied targeted therapies

Filomena Di Giacomo¹, Jude M. Phillip², Rohan Bareja³, Danilo Fiore¹, Xujun Wang¹, Balvir Kunar⁴, Michael Ginsberg⁴, Lorena Consolino⁵, Zhiwei Yang¹, Joseph Scandura⁶, Wayne Tam¹, Sabina Chiaretti⁷, Anna Guarini⁷, Robin Foà⁷, Olivier Elemento³, Leandro Cerchietti², Shahin Rafii⁸, Giorgio Inghirami^1,5

¹ Department of Pathology and Laboratory Medicine, 10065, New York, NY, USA

² Department of Medicine and Meyer Cancer Center, 10065, New York, NY, USA

³ Institute for Computational Biomedicine & Caryl and Israel Englander Institute for Precision Medicine, Weill Cornell Medicine 10021, New York, NY, USA

⁴ Angiocrine Bioscience, 92122, San Diego CA, USA

⁵ Department of Molecular Biotechnology and Health Sciences, University of Turin, 10126, Turin, Italy

⁶ Department of Medicine, Hematology-Oncology, Weill Cornell Medicine and the New York Presbyterian Hospital 10065, New York, NY, USA

⁷ Department of Cellular Biotechnologies and Hematology, Sapienza University of Rome, 00161 Rome, Italy

⁸ Department of Medicine, Ansary Stem Cell Institute, Division of Regenerative Medicine, Weill Cornell Medicine 10065, New York, NY, USA

Introduction

T-cell acute lymphoblastic leukemia (T-ALL) remain frequently incurable. Poor clinical responses in T-ALL patients are often due to unchecked growth of relapsing/refractory leukemia cells protected in the safe haven of niches.

Material and Methods

Conventional in vitro screens overestimate drug sensitivities, lacking protective components of T-ALL niches. Here, we developed a platform of PDTX-derived leukemic cells to discover patient-specific drug responses in the context of a physiologic niche.

Results and Discussion

PDTX and engineered ECs were used to dissect disease evolution demonstrating that tumor-educated aberrant ECs counteract individual leukemia addictions and overcome stress conditions via chemotherapy. Using engineered endothelial cells (ECs) and patient-derived-tumor-xenografts (PDTX) we discovered that T-ALL instruct tumor ECs to acquire an aberrant pre-natal signature. These subverted tumor ECs by supplying angiocrine factors counteract and overcome chemotherapy sensitivity. Employing cocultured EC/ PDTX T-ALL cells, we carried out drug discovery unraveling patient-specific drug-response fingerprints and predicting responses, which effectively induced in vivo regression of T-ALLs. This underscores the ability to prospectively design personalized therapies integrating genomic and functional readouts within the umbrella of aberrant tumor niche models, marking the pivotal contribution of the host-mediated protumorigenic role of ECs.

Conclusion

Our platform identified functionally distinct subgroups of T-ALL, which specifically responded to unique agents, allowing the construction of patient specific drug biograms. We anticipate that host cancer models and clinical application of drug-library screening will foster implementation of patient-specific drug combinations and curative tailored protocols of naïve and refractory T-ALL patients.