PDX model initiatives, collections & standards (first part) 09:00 - 10:15

Acknowledging the genomic heterogeneity between primary and metastatic breast cancer (BC), several studies aiming to investigate the molecular landscape of metastatic BC were launched. Despite these efforts, the molecular characterization of metastatic lesions mostly relies on the analysis of core needle biopsies from a single lesion that might not accurately reflect the morphological, molecular and immune landscape of the various metastases present in the patient. In addition, treatment recommendations are based on the characteristics of the primary and/or a unique metastatic lesion, ignoring disease heterogeneity. Research autopsies or preferably called ‘post-mortem tissue donation’ programs can help to overcome these difficulties by sampling and analyzing multiple lesions. In this context, we have established a unique tissue donation program at our institution called UPTIDER (UZ/KU Leuven Program for Post-mortem Tissue Donation to Enhance Research, NCT04531696). The overarching aim of the program is to unravel metastatic BC evolution, biology, heterogeneity and treatment resistance through multi-level and multiregional analysis, using groundbreaking technologies, of extensive samples rapidly taken in the post-mortem setting. One important aspect of the program focuses on the development of in vivo models, especially for some rare breast cancer subtypes, with the goal to develop innovative therapeutic strategies to optimize BC patients’ care. This is done in collaboration with the TRACE platform located at the KU Leuven and the laboratory of Prof. Brisken located in Lausanne (EPFL, specifically for the development of the lobular BC models) through either implementation of fresh tissues samples into PDX models at the time of the autopsy either storage of frozen tissues samples for future implementation. As of today, we have performed research autopsy on 16 BC patients from whom multiple lesions (range 1-12) were collected for in vivo models from 13 patients. Even though we are at the various stages of implementation and generation, successful uptake was obtained for 6 patients up to now. We believe our models that represent one of the greatest unmet need, metastatic breast cancers, will greatly help to recapitulate the complexity of human tumors and to develop more effective cancer therapies.

Ludovic Bigot¹, Catline Nobre¹, Francesco Facchinetti¹, Jonathan Sabio¹, Loic Poireaudau¹, Floriane Braye¹, Benjamin Besse^1,2, Fabrice Andre^1,2, Luc Friboulet^1,*, Yohann Loriot^1,2,3,*

¹ Université Paris-Saclay, Gustave Roussy Cancer Campus, Inserm U981, Biomarqueurs prédictifs et nouvelles stratégies thérapeutiques en oncologie, 94800, Villejuif, France.
² Department of Medical Oncology, Gustave Roussy Cancer Campus, France
³ Drug Development Department (DITEP), Gustave Roussy Cancer Campus, France

Introduction

In the last 20 years, the advances in molecular oncology and cancer genetics allowed the identification of an increasing number of actionable oncogenic drivers and the development and clinical use of specific inhibitors. Despite these successes, it is now well established that tumour cells adapt and develop acquired resistance to these targeted inhibitors so that disease progresses within 5-12 months. It is crucial to understand these mechanisms of acquired resistance to develop overcoming therapeutic strategies.
At Gustave Roussy, a prospective clinical trial MATCH-R, is conducted to study the acquired resistance mechanisms and to find new therapeutic approaches for patients. A high throughput analysis is performed to compare the genetic of the tumor obtained before starting treatment and the tumor that acquired resistance. At the same time, we develop Patient Derived Xenograft (PDX) from these samples that allow us to validate new therapeutic approaches.

Material and Method

Fresh tumor biopsy specimens were obtained prospectively from patients through a prospective single- institution clinical trial (MATCH-R, NCT02517892). Patient derived xenografts (PDX) in NOD Scid Gamma (NSG) mice as well as patient derived organoids from PDX (PDXO) and Patient derived cell lines were developed and characterized. Extensive molecular profiling including whole exome sequencing (WES), RNA sequencing (RNAseq) and immunohistochemistry were performed on human samples; PDX; Patient derived cells lines and PDXO.

Results and Discussion

As of June 2022, 130 PDX models have been successfully obtained from 330 biopsies (global take rate of 39%). Our focus is the development of models from different cohorts: AR inhibitors in CRPC (18 PDX), ALK inhibitors in NSCLC (15 PDX), EGFR inhibitors in NSCLC (27 PDX), FGFR inhibitors in bladder carcinoma and cholangiocarcinoma (22 PDX) and KRAS inhibitors in NSCLC and pancreatic cancer (9 PDX). Our results validated that our PDX models recapitulate the genetics, the phenotype and the pharmacology of the original biopsies.
Novel mechanisms of resistance to TKI in solid tumors were identified. Adaptive treatment with novel TKI or combinatorial strategies aiming to restore sensitivity in PDX (readout: mean tumor growth). These results confirm that PDX models are useful to study the resistance mechanism and to develop new therapeutic strategies.
We establish a large cohort of models from the most aggressive form of CRPC with a new preclinical approach using both PDX and their matched PDXO. This cohort will constitute a new resource to develop therapeutic approaches in prostate Cancer.

Conclusion

Overall, the MATCH-R project provides a unique preclinical platform to identify resistance mechanisms to innovative therapies and to develop next generation therapeutic strategies.

Petra ter Brugge¹, Ivan Bièche², Sarah Moser¹, Sabrina Ibadioune², Alexandre Eeckhoutte^3,4, Roebi de Bruijn¹, Tatiana Popova^3,4, Marc-Henri Stern^3,4, Jos Jonkers^1* and Elisabetta Marangoni^4*

¹ Division of Molecular Pathology, Oncode Institute, The Netherlands Cancer Institute, Amsterdam, Netherland
² Genetics Department, Institut Curie, PSL Research University, 26 Rue d'Ulm, 75005, Paris, France
³ INSERM U830, Institut Curie, PSL Research University, 75005 Paris, France
⁴ Institut Curie, PSL Research University, 26 Rue d'Ulm, 75005, Paris, France
⁵ Translational Research Department, Institut Curie, PSL Research University, 26 Rue d'Ulm, 75005, Paris, France
*Corresponding authors: elisabetta.marangoni@curie.fr - j.jonkers@nki.nl

The high frequency of homologous recombination deficiency (HRD) is the main rationale of testing platinum-based chemotherapy in triple-negative breast cancer (TNBC). Nevertheless, the existing methods to identify HRD in tumors are controversial and there is a medical need for predictive biomarkers to improve patients selection and avoid toxicity in non-responders. Here, we assessed the in vivo response to platinum agents in 55 molecularly characterized patient-derived xenografts (PDX) of TNBC to identify determinants of response and resistance. The association between BRCA1/2 mutations and response to platinum agents was close to statistical significance, while BRCA1 promoter methylation was not, in part due to residual BRCA1 gene expression and homologous recombination proficiency in different tumours showing mono-allelic promoter methylation. The HRD status, determined from shallow whole genome sequencing (sWGS) based on the number of large-scale genomic alterations, predicted 93% of responses, while homologous recombination proficiency (HRP) predicted resistance in 65% of tumours. Genomic HRD was significantly correlated with a functional defect in homologous recombination, assessed by impaired RAD51 foci formation. Finally, by whole exome sequencing of matched patients’ and PDX tumours, we identified mutations in XRCC3 and ORC1 genes in 2 cisplatin sensitive tumours without any additional alterations the BRCA1/2 locus. Further in vitro validation demonstrated that loss of XRCC3 or mutations in ORC1 sensitize cells to cisplatin. In conclusion, our result demonstrate that the genomic HRD is highly predictive of platinum response in a large cohort of TNBC PDX and identify alterations in XRCC3 and ORC1 genes driving cisplatin response.