High In Vitro and In Vivo Activity of BI-847325, a Dual MEK/Aurora Kinase Inhibitor, in Human Solid and Hematologic Cancer Models

BI-847325 is an ATP-competitive inhibitor of MEK/Aurora kinases with the potential to treat a wide range of cancers. In a panel of 294 human tumor cell lines in vitro, BI-847325 was found to be a highly selective inhibitor that was active in the submicromolar range. The most sensitive cancer types were acute lymphocytic and myelocytic leukemia, melanomas, bladder, colorectal, and mammary cancers. BI-847325 showed a broader range of activity than the MEK inhibitor GDC-0623. The high efficacy of BI-847325 was associated with but not limited to cell lines with oncogenic mutations in NRAS, BRAF, and MAP2K1.The high antiproliferative activity of BI-847325 was validated in vivo using subcutaneous xenograft models. After oral administration of 80 and 40 mg/kg once weekly for 3 or 4 weeks, BI-847325 was highly active in four of five colorectal, two of two gastric, two of two mammary, and one of one pancreatic cancer models (test/control < 25%), and tumor regressions were observed in five of 11 cancer models. The treatment was well tolerated with no relevant lethality or body weight changes. In combination with capecitabine, BI-847325 displayed synergism over single-agent therapies, leading to complete remission in the triple-negative mammary model MAXFTN 401, partial regression in the colon model CXF 1103, and stasis in the gastric models GXA 3011 and GXA 3023. In conclusion, dual MEK/Aurora kinase inhibition shows remarkable potential for treating multiple types of hematologic and solid tumors. The combination with capecitabine was synergistic in colorectal, gastric, and mammary cancer.

SIGNIFICANCE

We report the preclinical evaluation of BI-847325, a MEK/Aurora kinase inhibitor. Our data demonstrate that BI-847325 has potent antitumor activity in a broad range of human solid and hematologic cancer models in vitro and in vivo and is well tolerated in animal models. It also shows synergistic effect when combined with capecitabine. These findings provide a strong rationale for further development of BI-847325 as a potential therapeutic for patients with cancer.

Abstract 1132: A systems biology approach combining ProLiFiler and Cancer Data Miner for an enhanced preclinical characterization of the WEE-1 inhibitor Adavosertib

Adavosertib (AZD1775, MK-1775), a clinical stage inhibitor of the tyrosine kinase WEE1, was investigated in a cell proliferation and survival assay with 140 human cancer cell lines (CLs) representing all major tumor types (ProLiFiler platform of Reaction Biology) followed by mechanism of action (MoA) and biomarker analyses using 4HF Biotec’s Cancer Data Miner in silico platform. Adavosertib exhibited a broad anti-cancer activity across all hematological and solid tumor types with IC50 values ranging from 0.06 to 10 µM (median: 0.38 µM), matching the consistently high expression of the WEE1 gene. Among 900 reference compounds, the activity profile of adavosertib correlated best with the profiles of compounds targeting the replication stress response including other WEE1 inhibitors but also inhibitors of checkpoint kinase 1 and 2 (CHK1/2) or ataxia telangiectasia-mutated (ATM). Significant correlations were also seen with compounds blocking mitosis, DNA replication and DNA repair. Interestingly, we observed a subset of cell lines that were resistant to both DNA synthesis and PARP inhibitors but were sensitive to WEE1 inhibition. Moreover, by using multiple datasets of WEE1 inhibitors connected to the molecular annotations of CLs for a data driven biomarker screening, we identified MYC mutations as a predictive marker of sensitivity and PIK3CA or ERBB2 gene amplifications as predictors of resistance. Transcriptome analysis identified up to 900 genes for which higher expression in CLs was associated with CL sensitivity to the compound. Preliminary pathway analysis indicated that these genes are well represented among nuclear factor and Myc-regulated genes. In conclusion, our studies demonstrate broad anticancer activity of adavosertib and confirm its proposed MoA. The biomarkers we identified will facilitate the selection of pre-clinical in vivo tumor models and, if confirmed, even patient selection for clinical trials. The combined use of the ProLiFiler and Cancer Data Miner Platforms has the potential to accelerate and de-risk the development of anti-cancer agents.

Citation Format: Vincent Vuaroqueaux, Daniel Feger, Anne-Lise Peille, Oliver Siedentopf, Sadhana Panzade, Sarah Ulrich, Sebastian Dempe, Heinz-Herbert Fiebig, Jan Erik Ehlert. A systems biology approach combining ProLiFiler and Cancer Data Miner for an enhanced preclinical characterization of the WEE-1 inhibitor Adavosertib [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1132.

Pharmacogenomics characterization of the MDM2 inhibitor MI-773 reveals candidate tumours and predictive biomarkers

MI-773 is a recently developed small-molecule inhibitor of the mouse double minute 2 (MDM2) proto-oncogene. Preclinical data on the anti-tumour activity of MI-773 are limited and indicate that tumour cell lines (CLs) with mutated TP53 are more resistant to MI-773 than wild type TP53. Here, we explored the compound's therapeutic potential in vitro using a panel of 274 annotated CLs derived from a diversity of tumours. MI-773 exhibited a pronounced selectivity and moderate potency, with anti-tumour activity in the sub-micromolar range in about 15% of the CLs. The most sensitive tumour types were melanoma, sarcoma, renal and gastric cancers, leukaemia, and lymphoma. A COMPARE analysis showed that the profile of MI-773 was similar to that of Nutlin-3a, the first potent inhibitor of p53-MDM2 interactions, and, in addition, had a superior potency. In contrast, it poorly correlates with profiles of compounds targeting the p53 pathway with another mechanism of action. OMICS analyses confirmed that MI-773 was primarily active in CLs with wild type TP53. In silico biomarker investigations revealed that the TP53 mutation status plus the aggregated expression levels of 11 genes involved in the p53 signalling pathway predicted sensitivity or resistance of CLs to inhibitors of p53-MDM2 interactions reliably. The results obtained for MI-773 could help to refine the selection of cancer patients for therapy.

Abstract 1173: Identification of novel targets for the treatment of pancreatic cancer patients

Pancreatic cancer (PC) is the 12th most common cancer worldwide. Despite a large panel of chemo- and targeted therapeutics options, patient prognosis remains poor with a 5-years overall survival below 10%. Thus, there is still a critical need to develop more efficient therapeutic alternatives. Antibody drug conjugates (ADC) and small molecule drug conjugates (SMDC) combine the oncolytic activity of highly potent chemotherapies with the target specificity of an antibody or a small molecule. Both ADC and SMDC are of increasing interest for cancer treatment, as they allow more specific delivery of chemotherapies to the tumor site. Facing the clinical needs for PC treatment, here we present an in-silico analysis to reveal specific targets for further ADC/SMDC development. 4HF Biotec has developed a proprietary platform connecting large clinical, OMICS and drug sensitivity data from various sources. It includes annotation for more than 1,800 preclinical models (cell lines, cell line-derived xenografts, and patient-derived xenografts), up to 11,000 patient tumors and 22,000 normal tissues (TCGA, GTEx and various GEO datasets). For tumor target discovery purposes, we designed and implemented the platform with specific analytics tools. To identify specific targets for PC, we first decided to analyze preclinical models, to focus on genes expressed by tumor cells and not by stroma cells. This aspect is particularly important in the context of PC which often have a high stroma content. Differential gene expression analysis of 113 PC preclinical models versus 1,737 tumor models from up to 30 tumor entities revealed 327 PC specific genes potentially targetable. Then, a similar analysis was performed by testing TCGA patient tumors (179 pancreatic tumors vs 9,521 patient tumors from other entities) and revealed 1,292 pancreatic specific genes. Finally, PC patient tumors were compared to 709 samples from various normal organs allowing to identify 1,156 tumor specific genes. At the intersection of these three analyses, we identified 56 PC-specific target candidates for ADC/SMDC development. Among the top candidates, MUCL3 (mucin like 3) was one of the most promising genes. Its mRNA expression is almost exclusively restricted to pancreatic and stomach samples in both preclinical models and TCGA patient tumors. It is overall not frequently expressed by normal tissues, and restricted to subsets of stomach, esophagus, and lung samples. The gene encodes for a transmembrane protein with a long weakly glycosylated extracellular part. A detailed analysis of the protein characteristics and expression modalities will be shown. The present work demonstrates that our in silico platform helps to identify promising targets for PC treatment using ADC/SMDC approaches. Our analyses revealed MUCL3 as one of the top candidates, further analyses will be needed to determine its druggability using small molecules or antibodies.

Citation Format: Anne-Lise Peille, Alexandra Musch, Hoor Al-Hasani, Heinz-Herbert Fiebig, Vincent Vuaroqueaux. Identification of novel targets for the treatment of pancreatic cancer patients [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 1173.

Abstract 1027: ProLiFiler and Cancer Data Miner, combined platforms for preclinical investigation to scrutinize impact of inhibitors on the KRAS, PI3K and MDM2 signaling pathways

Testing novel anti-cancer agents across large panel of tumor models covering genetic diversity of cancers is increasingly considered as a cornerstone of preclinical development. For this purpose, Reaction Biology developed “ProLiFiler” a standard panel of 140 cell lines (CLs) covering most common cancer types to evaluate anti-proliferative activity of novel drugs. Partnering has been made with 4HF Biotec and their in-silico platform, named “Cancer Data Miner”, to investigate and to understand molecular basis of drug sensitivity. Here we report the use of our platforms to realize integrative pharmacogenomic studies for three recent small molecules targeting major altered pathways in cancers. It includes SOS1::KRAS interaction inhibitor BI-3406, MDM2 inhibitor Nutlin-3a, and PI3K inhibitor Taselisib. Main goal of the study is to provide meaningful information for these three drugs regarding their efficacy and potency, the validation of their mechanism of actions (MOA), the suitable clinical indications, possible drug combinations and the predictive biomarkers of sensitivity or resistance. The three compounds are tested for anti-proliferative activity in vitro in a 2D monolayer assay using the “ProLiFiler”CLs panel. For data analytics, the resulting in vitro data are loaded on the “Cancer Data Miner” platform and connected to CL annotations including whole exome mutations, chromosomal aberrations, gene expression profiles or drug sensitivity profiles. The drug response profiles will be reported for the three compounds individually and compare between them, showing respective efficacy, potency, and CL/cancer entity selectivity. Using the MOA Finder tool, we will correlate BI-3406, Nutlin-3a, and Taselisib individual IC50 profiles to those of more than 800 compounds with known MOA that are integrated on the platform. The analyses will show the drugs most closely related to the 3 compounds and that are expected to have similar MOA. With the biomarker discovery tools, we will run high throughput statistical analyses to reveal whole exome mutations, copy number variations and expression significantly associated with drug sensitivity/resistance. For interpretation, pathway and enrichment analysis will be presented. A focus will be made on key alterations like KRAS and MAPK related genes, TP53-MDM2 and PIK3CA-PTEN to evaluate their predictivity. The work will be also completed by functional analysis, for instance, by assessing the effect of BI-3406 on ERK-MEK activation, and the impact of MDM2 inhibition on apoptotic markers. The present work will show the whole panel of analyses proposed by 140 CL-ProLiFiler and Cancer Data Miner complementary platforms, allowing to acquire key information at an early stage of drug development and helping to setup next steps such as selection of models for in vivo testing.

Citation Format: Vincent Vuaroqueaux, Daniel Feger, Hoor Al-Hasani, Oliver Siedentopf, Anne-Lise Peille, Sarah Ulrich, Sebastian Dempe, Heinz-Herbert Fiebig, Jan Erik Ehlert. ProLiFiler and Cancer Data Miner, combined platforms for preclinical investigation to scrutinize impact of inhibitors on the KRAS, PI3K and MDM2 signaling pathways [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 1027.

Abstract LB107: Identification of a promising renal cell carcinoma target using a unique in silico approach

In 2020, kidney cancers accounted for more than 4.1% of new cancer cases worldwide. Among these, more than 90% were renal cell carcinoma (RCC), a very heterogenous disease consisting of 3 major subtypes: chromophobe , papillary and clear cell. These subtypes differ in histology, genetic alterations, clinical outcome and therapeutic response. Over the last decades, great advances in the treatment of RCC have been made with additional targeting agents inhibiting RTKs, VEGFR and mTORC1. However, for local RCC surgery remains the main treatment modality whereas metastatic RCC is still difficult to treat due to chemoresistance. Antibody drug conjugates (ADC) or small molecule drug conjugates (SMDC) are emerging approaches for the treatment of cancer patients. By coupling highly potent chemotherapeutic agents to a specific antibody/small molecule, tumor cells can be selectively targeted allowing treatment with higher drug doses leading to an enhanced efficacy, while maintaining low toxicity in healthy tissues. 4HF Biotec has developed a proprietary database containing, among others, curated molecular data from tumor patient samples, preclinical models (cell lines, cell line-derived xenografts, and patient-derived xenografts) and normal tissues. Using this platform, we established a unique methodology to identify and prioritize genes that could be targeted by ADCs/SMDCs. By performing bioinformatics analyses of this molecular data, we identified genes that are differentially expressed across cancer (sub)types. We further selected those that are overexpressed in the tumor entity compared to matching normal tissue and in the tumor entity compared to all normal tissues. The overlapping genes were assessed in preclinical models to confirm their tumor-specificity. For the final selection, we investigated target druggability by curating data regarding protein function, targetability and localization. With this approach, we identified a total of 374 potential drug targets across 15 cancer (sub)types for the development of ADCs or SMDCs. Analyses of kidney cancer identified Semaphorin 5B (SEMA5B) among the top candidates. SEMA5B was significantly upregulated in patient samples of the kidney compared to other tumor entities. Comparison with normal kidney samples revealed a significant higher expression in tumor samples. In addition to the expression profiles, association between SEMA5B expression and clinical and molecular features will be presented. Furthermore, we will present curated information about protein function and structure that will facilitate antibody generation. Our findings indicate that targeting SEMA5B on tumor cells could represent a new therapeutic option in the treatment of renal cancer.

Citation Format: Alexandra Musch, Anne-Lise Peille, Hoor Al Hasani, Heinz-Herbert Fiebig, Vincent Vuaroqueaux. Identification of a promising renal cell carcinoma target using a unique in silico approach [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 LB107.

Abstract 2774: Longitudinal characterization of patient-derived orthotopic xenograft brain tumor models

Gliomas are the most common primary brain tumors and currently the prognosis is still poor. Due to this, it is one of the main areas in oncological research and drug development programs. Innovative therapies are vital to improve treatment outcomes but must be developed to enable trafficking across the blood brain barrier (BBB). For this advent, animal models provide important information prior to clinical studies. Among the different in vivo models, orthotopic PDX models represent best the tumor microenvironment and natural variability of tumors, hence providing the most reliable results over species.

In the brain tumor field, imaging has a central role in clinical diagnosis and as a prognostic factor to monitor therapy response in patients. Magnetic resonance imaging (MRI) is widely used for clinical diagnosis and disease follow up as it offers unprecedented soft tissue contrast and high spatial resolution non-invasive way. In addition to imaging, molecular profiling, e.g. circulating immune biomarkers and local oncometabolites in the tumor environment may facilitate as important translational biomarkers in development of immunotherapy for gliomas in addition to traditional histopathological readouts.

The purpose of this work was to analyze possible heterogeneity of tumors in vivo, growth rate and volume in stereotactically implanted orthotopic PDX brain tumor models using MRI/MRS imaging. In addition to imaging, neurological scoring was performed to monitor general animal welfare, cytokine profiles from plasma to observe immunological responses over time and determination of oncometabolites in plasma and brain tissues combined with traditional histopathological changes were performed. The data from orthotopic models was also compared to readouts in subcutaneous models.

As a conclusion, translational in vivo imaging techniques were applied to study orthotopic tumor progression combined with circulating immune biomarkers, and general welfare scoring. These readouts provide a powerful and translational research tool together with oncological disease animal models allowing comprehensive evaluation of disease progression and treatment interventions for in vivo studies.

Citation Format: Julia Schueler, Mariette Heins, Artem Shatillo, Kimmo Lehtimäki, Anne-Lise Peille, Taina-Kaisa Stenius, Timo Bragge, Jussi Rytkönen, Diana Miszczuk, Tuulia Huhtala. Longitudinal characterization of patient-derived orthotopic xenograft brain tumor models [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 2774.

Abstract 2813: Tumor models driven by EGFR: Optimizing the preclinical profiling of EGFR-targeting agents

The epidermal growth factor receptor (EGFR, HER1, ERBB1) is a driver of many human cancers. Standard of care treatment for colon, head and neck and non-small cell lung cancer includes drugs targeting EGFR. Numerous molecular alterations activating the oncogenic potential of the EGFR gene have been described including activating point mutations, point mutations causing resistance against EGFR-targeting drugs, activating deletions and truncations as well as overexpression of EGFR and its ligands, occasionally induced by gene amplification. In other situations, EGFR has been identified as a driver in the absence of any obvious molecular alterations of the EGFR gene. The discovery and development of EGFR-targeting agents depends on the availability of relevant tumor models. We provide an overview of our collection of EGFR-driven tumor models, including PDXs along with PDX-derived cell lines and human tumor cell lines. We compare sensitivity profiles of tumor models for EGFR-targeting agents obtained in vivo and in vitro and demonstrate that for EGFR-targeting agents, data obtained in 2D and 3D assays are predictive for the in vivo situation. We propose an optimized strategy for the preclinical profiling of EGFR-targeting anti-cancer agents.

Citation Format: Markus Posch, Hagen Klett, Anne-Lise Peille, Gerhard Kelter, Armin Maier, Julia Schüler, Thomas M. Metz. Tumor models driven by EGFR: Optimizing the preclinical profiling of EGFR-targeting agents [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 2813.

Evaluation of molecular subtypes and clonal selection during establishment of patient-derived tumor xenografts from gastric adenocarcinoma

Patient-derived xenografts (PDX) have emerged as an important translational research tool for understanding tumor biology and enabling drug efficacy testing. They are established by transfer of patient tumor into immune compromised mice with the intent of using them as Avatars; operating under the assumption that they closely resemble patient tumors. In this study, we established 27 PDX from 100 resected gastric cancers and studied their fidelity in histological and molecular subtypes. We show that the established PDX preserved histology and molecular subtypes of parental tumors. However, in depth investigation of the entire cohort revealed that not all histological and molecular subtypes are established. Also, for the established PDX models, genetic changes are selected at early passages and rare subclones can emerge in PDX. This study highlights the importance of considering the molecular and evolutionary characteristics of PDX for a proper use of such models, particularly for Avatar trials.

Abstract LB-B05: Characterization of a panel of 79 PDX-derived cell lines with a focus on the EGFR exon 20 insertion mutation-driven NSCLC model LXFE 2478

Patient-derived tumor xenograft (PDX) models have become indispensable for the preclinical profiling of novel anti-cancer agents as they retain histological, molecular and pharmacological characteristics of the parental patient tumors and for many tumor types collectively replicate the diversity of patient tumors. To make PDX models available for in vitro assays, we have established a panel of 79 low passage cell lines derived from PDXs representing 15 different tumor histologies. Using, among others, the non-small cell lung cancer model LXFE_2478 which is driven by the exon 20 insertion EGFR mutation M766_A767insASV as an example, we demonstrate the following similarities of PDX models and the corresponding PDX-derived cell lines. (i) Subcutaneous xenografts established from PDX-derived cell lines mirror the histology of the corresponding PDX models and usually also patient tumors. (ii) For LXFE_2478 both the PDX model and the corresponding PDX-derived cell line express high levels of EGFR, suggesting that the oncogenic driver is still present. (iii) For LXFE_2478 the PDX and the PDX-derived cell line subcutaneously implanted in mice display comparable sensitivity to a variety of EGFR inhibitors and cytotoxics. (iv) For LXFE_2478 the sensitivity of the PDX-derived cell line to a variety of EGFR inhibitors in an in vitro 2D cell survival and proliferation assay in general corresponds to the EGFRi sensitivity of the corresponding PDX model in vivo. (v) For several additional PDX models representing various histotypes the sensitivity to first generation EGFR inhibitors of PDX-derived cell lines in the 2D assay is in line with the in vivo EGFR inhibitor sensitivity of the corresponding PDXs. In conclusion, at least for some mechanisms of action in vitro data obtained with PDX-derived cell lines can predict the in vivo behavior of the corresponding PDX model and can thereby accelerate, and reduce the costs of, the discovery of novel anticancer agents.

Citation Format: Gerhard Kelter, Anne-Lise Peille, Jutta Fehr, Hagen Klett, Armin Maier, Markus Posch, Thomas M Metz. Characterization of a panel of 79 PDX-derived cell lines with a focus on the EGFR exon 20 insertion mutation-driven NSCLC model LXFE 2478 [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr LB-B05. doi:10.1158/1535-7163.TARG-19-LB-B05

Abstract LB-C16: The percentage of tumor infiltrating lymphocytes in a panel of solid cancer PDX models correlates with a specific human cytokine pattern in peripheral blood of the tumor bearing animal

In the advent of immune modulatory compounds being part of most of the oncology drug development pipelines, the available preclinical models have to be characterized with respect to their interaction with the tumor microenvironment in general and with the immune cells of the host specifically. In the current study we analyzed 31 PDX models covering a broad range of solid cancer entities with regard to the phenotypic characteristics in a humanized environment. Well established PDX models from six different entities (breast, colon, renal, pancreatic, NSCL cancer and glioblastoma) were implanted subcutaneously into immune-compromised mice substituted with CD34+ hematopoietic stem cells (hu-mice). Tumor characterization included: tumor growth curves, take rates, human immune cell infiltrates of tumor (=TILs), peripheral blood and bone marrow (determined by flow cytometry, FC, and IHC) and secretion of human as well as murine cytokines in mouse serum (bead array: human 40plex (Biorad, #171AK99MR2; murine 23plex (Biorad, #M60009RDPD). The panel of 31 different PDX models of solid cancer displayed a distinct TIL and cytokine profile per model. The analysis of TILs revealed a large bandwidth of immune cell infiltration in the respective model ranging from 94% huCD45+ cells (renal cancer RXF 2773) to 3.54% huCD45+ (NSCLC LXFA 1041). Of note, the percentage and phenotype of the TILs was not related to the percentage and phenotype of the human immune cells in the BM or peripheral blood. In a next step, we categorized the PDX models into hot (&gt;5% TILs =huCD45+) and cold (&lt;5% TILs=huCD45+) tumors and compared the two groups with regard to the different read-outs. Differences between the two groups were found with regard to their sensitivity towards checkpoint inhibitor treatment (&gt;5% TILS were more sensitive), the human cytokine profile and their CD4/CD8. A trend towards a higher mutational burden was observe in the group with &gt;5%TILs. No difference between the two categories was determined in TIL composition (despite CD4/CD8 ratio), tumor growth behavior and secretion of mouse cytokines. Taken together, the characterization of PDX models with respect to their TIL composition facilitates model selection for innovative drugs in the immune-oncology field. The possibility to discriminate hot vs cold tumors by a minimal-invasive method like serum analysis of cytokine expression profile as potential biomarker will help to characterize the complete PDX collection of more than 500 models but will as well serve as additional read-out in preclinical I-O studies.

Citation Format: Eva Oswald, Anna Edinger, Anne-Lise Peille, Hagen Klett, Julia Schueler. The percentage of tumor infiltrating lymphocytes in a panel of solid cancer PDX models correlates with a specific human cytokine pattern in peripheral blood of the tumor bearing animal [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference on Molecular Targets and Cancer Therapeutics; 2019 Oct 26-30; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2019;18(12 Suppl):Abstract nr LB-C16. doi:10.1158/1535-7163.TARG-19-LB-C16

Abstract 2709: Modulation of the tumor-infiltrating lymphocyte population by PARP inhibitor talazoparib in combination with anti-PD1 treatment significantly enhances overall survival in a murine BRCA1-/- breast cancer model

Targeted therapy of BRCA-deficient cancers has been achieved using poly(ADP-ribose) polymerase (PARP) inhibitors, which block BRCA-independent DNA repair. With first approval in 2014 of Olaparib the concept of tumor-specific synthetic lethality was added to the treatment portfolio of cancer patients. Although the effects of PARPi have shown promising results in multiple cancer types, how and whether patients might benefit from combination with compounds modulating the immune landscape of a tumor is largely unknown. In the current study, we investigate the cross-talk between PARPi and immune checkpoint inhibition, in particular, anti PD-1 and anti CTLA-4, as the most advanced targets in the field. PARP inhibitors Niraparib, Rucaparib, Talazoparib and Olaparib were investigated in vivo using the murine EMT6/BRCA1-/- model in monotherapy as well as in combination with anti-CTLA-4 or anti-PD1 treatment. The four PARPi showed distinct activity profiles in the two breast cancer models. Talazoparib was the most active compound in the BRCA1-/- model (optimal T/C (test/control) of 60%), followed by Niraparib and Rucaparib (65% and 67%, respectively). Olaparib was considered inactive with a T/C value of 80% in monotherapy. The EMT6/BRCA1-/- turned out to be sensitive towards anti CTLA-4 treatment (optimal T/C of 35% - 38% in three independent experiments). Anti PD-1 treatment in monotherapy induced no significant reduction in tumor growth (optimal T/C of 80% - 82%, in two independent experiments). The combination of PARPi and anti CTLA-4 induced a transient but significant reduction of tumor load early in the treatment phase (p&lt; 0.002, one-way ANOVA, day 8). However, anti PD-1 treatment significantly prolonged overall survival in combination with Talazoparib (p&lt; 0.011 log-rank test). The analysis of tumor infiltrating lymphocytes (TILs) by flow cytometry revealed that Talazoparib as monotherapy and in combination with checkpoint (CP) inhibitors enhanced the number of gMDSC as well as the number of CD3/CD11b double positive T cells. Weekly cytokine analysis in the serum of tumor bearing mice will elucidate possible interactions between PARP and CP inhibitors and give guidance to optimal combination and schedules. First results indicate that PARPi as well as CPi induce unique cytokine profiles correlating well with the modified TIL composition of the respective treatment groups. Further mechanistic studies as well as comparative studies with the EMT6/BRCAwt model will elucidate the tumor biology behind these observations and might lead to beneficial combination strategies in patients suffering from triple negative and BRCA1-/- breast cancer.

Citation Format: Anya Avrutskaya, Cordula Tschuch, Astrid Jensen, William Durham, Maycee Robinson, Charles Krause, Emily O’Koren, Gerhard Kelter, Anne-Lise Peille, Armin Maier, Anne-Marie Zuurmond, Julia Schüler. Modulation of the tumor-infiltrating lymphocyte population by PARP inhibitor talazoparib in combination with anti-PD1 treatment significantly enhances overall survival in a murine BRCA1-/- breast cancer model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2709.

Patient derived renal cell carcinoma xenografts exhibit distinct sensitivity patterns in response to antiangiogenic therapy and constitute a suitable tool for biomarker development

Systemic treatment is necessary for one third of patients with renal cell carcinoma. No valid biomarker is currently available to tailor personalized therapy. In this study we established a representative panel of patient derived xenograft (PDX) mouse models from patients with renal cell carcinomas and determined serum levels of high mobility group B1 (HMGB1) protein under treatment with sunitinib, pazopanib, sorafenib, axitinib, temsirolimus and bevacizumab. Serum HMGB1 levels were significantly higher in a subset of the PDX collection, which exhibited slower tumor growth during subsequent passages than tumors with low HMGB1 serum levels. Pre-treatment PDX serum HMGB1 levels also correlated with response to systemic treatment: PDX models with high HMGB1 levels predicted response to bevacizumab. Taken together, we provide for the first time evidence that the damage associated molecular pattern biomarker HMGB1 can predict response to systemic treatment with bevacizumab. Our data support the future evaluation of HMGB1 as a predictive biomarker for bevacizumab sensitivity in patients with renal cell carcinoma.

Abstract 2169: Metastases of a temozolimide-sensitive patient-derived melanoma xenograft model show distinct biologic features and developed resistance against temozolomide

Patient-derived tumor xenografts (PDX) play a major role in the development of new cancer therapies and their strengths and weaknesses have gradually been elucidated. In the current study we established a melanoma PDX from donor patient tissue. In addition, we were able to create two sublines from spontaneous metastases occurring in the murine host during the establishment phase of the original model. All three lines were characterized by tumor growth kinetics, antitumoral activity against standard of care Temozolomide and patho-histological examination. Furthermore, whole exome sequencing and RNAseq data of primary PDX and its metastases are available. Two out of the three sublines have corresponding cell lines for 2D and 3D testing. The PDX model was developed from a biopsy of a 68 year old woman undergoing surgery due to a non-pretreated melanoma. After seven subcutaneous passages in immune-compromised mice, individual animals showed tumor growth in the liver as well as the spleen. We were able to passage and characterize those metastases in parallel to the original model. The human origin of the lines as well as cell lines established in 2D was confirmed by str-analysis. All three in vivo lines depicted distinct growth kinetics: The doubling times varied significantly (Kruskal-Wallis, p&lt; 0.0018) between 12.34 days (primary tumor, MEXF 2090P) and 30.78 day (spleen metastasis, MEXF 2090S) and 19.01 days doubling time for the liver metastasis model (MEXF 2090L). MEXF 2090S thereby depicted the slowest growth rate in our melanoma PDX panel (24 models, 10.09 days mean doubling time). The patho-histological examination revealed a well differentiated melanoma with low stroma content (3-5%) in all three lines. The molecular analysis (whole exome sequencing) identified distinct differences. Nevertheless, the potential driver mutations, Tp53 (R213Q) and Nras (Q61K), were recognized in all investigated samples. Temozolomide was applied to all three lines in vivo. 6 mice per group and line were treated either with Temozolomide (40 mg/kg/d, iv, twice a week for three weeks) or the control vehicle (10% DMSO, 90% NaCl). The primary tumor depicted statistically significant antitumoral activity with a T/C (test vs control) value of 36% (p&lt; 0.05, t test, two-tailed) two weeks after the last treatment (experiment day 32). The two lines derived of metastases were resistant against the alkylating agent depicting T/C values of 76% (2090L) and 79% (2090S), respectively. Spontaneous metastases are a rare event in PDX models growing subcutaneously in NMRI nude mice. Thus, shedding some light into the biology of the metastatic event in the current model will help to understand and influence the metastatic process in general. In any case, those lines can serve as indispensable tools in the oncology drug development pipeline.

Citation Format: Kerstin Klingner, Dorothee Lenhard, Elke Simon, Anne-Lise Peille, Julia Schüler. Metastases of a temozolimide-sensitive patient-derived melanoma xenograft model show distinct biologic features and developed resistance against temozolomide [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2169.

Abstract A019: Establishment and characterization of a melanoma patient-derived xenograft model comprising three different sublines with distinct biologic features

Patient-derived tumor xenografts (PDX) play a major role in the development of new cancer therapies and their strengths and weaknesses have gradually been elucidated. Large panels of solid cancer PDX are available to screen innovative compounds, identify new targets, and study tumor biology. In the current study we established a melanoma PDX from donor patient tissue. In addition, we were able to create two sublines from spontaneous metastases occurring in the murine host during the establishment phase of the original model. All three lines were characterized by tumor growth kinetics, antitumoral activity against standard-of-care temozolomide, and pathohistologic examination. Furthermore, molecular examination of primary PDX and its metastases is under way as well as cell line establishment. The PDX model was developed from a biopsy of a 68-year-old woman undergoing surgery due to a non-pretreated melanoma. After seven subcutaneous passages in immune-compromised mice, individual animals showed tumor growth in the liver as well as the spleen. We were able to passage and characterize those metastases in parallel to the original model. The human origin of the lines as well as one cell line established in 2D from the primary PDX was confirmed by str-analysis. All three in vivo lines depicted distinct growth kinetics: The doubling times varied significantly (Kruskal-Wallis, p&lt; 0.0018) between 12.34 days (primary tumor, MEXF 2090P) and 30.78 days (spleen metastasis, MEXF 2090S) and 19.01 days doubling time for the liver metastasis model (MEXF 2090L). The pathohistologic examination revealed a well-differentiated melanoma with low stroma content (3-5%) in all three lines. The molecular analysis (whole exome sequencing) is ongoing. So far the primary tumor has been characterized and two potential driver mutations, Tp53 (R213Q) and Nras (Q61K), were recognized. To characterize the models in more detail, temozolomide was applied to all three lines. 6 mice per group and line were treated either with temozolomide (40 mg/kg/d, iv, twice a week for three weeks) or the control vehicle (10% DMSO, 90% NaCl). The primary tumor depicted statistically significant antitumoral activity with a T/C (test vs control) value of 36% (p&lt; 0.05, t test, two-tailed) two weeks after the last treatment (experiment day 32). The two lines derived of metastases were resistant against the alkylating agent, depicting T/C values of 76% (2090L) and 79% (2090S), respectively. Further characterization of the PDX model and its different sublines will help to elucidate the resistance mechanism behind these data. Spontaneous metastases are a rare event in PDX models growing subcutaneously in NMRI nude mice. Thus, shedding some light into the biology of the metastatic event in the current model will help to understand and influence the metastatic process in general. In any case, those lines can serve as indispensable tools in the oncology drug development pipeline.

Citation Format: Kerstin Klingner, Dorothee Lenhard, Elke Simon, Anne-Lise Peille, Julia Schüler. Establishment and characterization of a melanoma patient-derived xenograft model comprising three different sublines with distinct biologic features [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A019.

Abstract 1840: Whole-exome somatic mutation analysis of mouse cancer models and implications for preclinical immunomodulatory drug development

Experimental tumors raised in rodents represent an important preclinical tool to develop innovative anticancer compounds before clinical testing. Amongst others such models include solid tumors raised in syngeneic fully immunocompetent hosts and tumors spontaneously growing in genetically engineered mice (GEM) and derivate thereof. These model platforms have gained additional value since the manipulation of the immune system to fight cancer has led to tangible benefits for cancer patients. In the current study, we analyzed somatic mutation profiles from whole-exome sequencing (WES) data in a panel of 14 different mouse models covering 6 major cancer types. 4 models were GEM-derived, all other lines were developed by injection of established cell lines into the corresponding mouse strain. In parallel, these models were evaluated for their sensitivity towards checkpoint inhibitors (α-CTLA-4, α-PD-1 or α-PDL-1) in mono- or combined therapy with cytostatic and/or targeted agents.WES achieved an average-of-coverage of 165X in tumor models and normal DNA. A median mutation rate of 34 somatic mutations (m)/MB was detected, ranging from 7 m/MB (GEM derived NSCLC model KP) to 328 m/MB (syngeneic NSCLC line Lewis Lung) in exons. Mutation rates were markedly lower in GEM-derived models as in syngeneic lines (median of 9 vs 43 m/MB). This reflects very well the different underlying carcinogenic mechanism of these two types of models. The cross-comparison of tissue-transplants vs cell lines from GEM-derived model KP revealed that 75% of the mutations found in the primary KP could also be detected in the corresponding cell lines KP1 and KP4. Of note, the mutation count increased 1.3- (KP4) and 2.9-fold (KP1) during cell line establishment. Every model depicted a distinct profile against modulators of the immune system dividing the panel in responders and non-responders. In our hands no significant correlation could be determined between mutational load and sensitivity towards checkpoint inhibition in vivo. This might be related to the fact that the dataset was not broad enough and the number of models per entity was too small, rendering the subtype analysis within the panel not feasible. However, a strong tendency was observed when investigating the colon lines Colon26, CT26 and MC38 showing best response to the combination of PD-1+CTLA-4 inhibitors and in parallel the highest mutation rates (52, 64 and 59 m/MB, respectively) compared to non-responders B16-F10, CloudmanS91, 4T1 and KP1 (23 m/MB on average). Mouse models of cancer are a relevant tool for preclinical studies specifically for immuno-oncology. The molecular characterization of these models will help to optimize their use in drug discovery. They will support the development of innovative drugs and indentification of biomarkers to classify the patient cohort profiting the most from these new compounds.

Citation Format: Bruno Zeitouni, Cordula Tschuch, Jason M. Davis, Anne-Lise Peille, Yana Raeva, Manuel Landesfeind, Sheri Barnes, Julia B. Schüler. Whole-exome somatic mutation analysis of mouse cancer models and implications for preclinical immunomodulatory drug development [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1840. doi:10.1158/1538-7445.AM2017-1840

Abstract 4815: Humanized single mouse trial: A preclinical platform feasible for immune-oncology drug screening and translational biomarker development

The field of cancer immunology is rapidly moving towards innovative therapeutic strategies. As a consequence the need for robust and predictive preclinical platforms arises just as well. The current project aims to establish a drug screening workflow bridging between innovative mouse models and clinical biomarker development. A total of 69 NOG (NOD/Shi-scid/IL-2Rγnull) mice were engrafted with CD34+ hematopoietic stem cells. Thereafter, tumor material from 11 different lung cancer patient derived xenograft models (NSCLC PDX) was implanted subcutaneously. Individual mice were treated with α-CTLA-4, α-PD-1 or the combination thereof. With n=1 per treatment arm and model the study design followed the screening approach of the single mouse trial (SMT). Infiltration of human immune cells was detected by flow cytometry (FC) and immunohistochemistry (IHC) in hematopoietic organs and tumor tissue. A computerized analysis for digitized whole-slide images of the samples was used to quantify the lymphocyte infiltration using color classification and morphological image processing techniques. All 3 treatment arms displayed a discrete activity pattern throughout the PDX panel. Tumor models with high tumor infiltrating lymphocyte (TIL) rates in the donor patient material tended to be more sensitive towards checkpoint inhibitor treatment as models with low rates. Numbers of TILs in the PDX detected by FC and IHC were significantly increased in the treatment groups as compared to control vehicle. In parallel, hematopoietic organs showed high (&gt;25%) amounts of huCD45 cells in all groups and models. PDX models being sensitive towards checkpoint inhibitor treatment (responders) displayed a higher percentage of DAB+ nuclei in huCD45 IHC stains than non-responder models as determined by image analysis. Irrespective thereof, in responders as well as non-responders the treatment with checkpoint inhibitors enhanced the percentage of DAB+ nuclei. Whole-slide image analysis of the H&E stains revealed an increase of the stromal compartment proportion in the tumor tissue under treatment with checkpoint inhibitors in responder models. In non-responder models the ratio between tumor and stroma was not influenced by drug treatment. The use of PDX based humanized mouse models in a SMT format allows screening approaches in complex mouse models. The combination with a comprehensive image analysis tool enables additional read-outs to quantify antitumoral activity of immune modulatory compounds. The latter can be used to identify possible biomarkers in the preclinical setting. Moreover, the translation and validation of these biomarker candidates in a clinical setting is self-evident as primary material needed for these types of analyses is easily accessible.

Citation Format: Daniel Bug, Eva Oswald, Anne Grote, Anne-Lise Peille, Gabriele Niedermann, Dorit Merhof, Friedrich Feuerhake, Julia B. Schüler. Humanized single mouse trial: A preclinical platform feasible for immune-oncology drug screening and translational biomarker development [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4815. doi:10.1158/1538-7445.AM2017-4815

Abstract 639: Subtyping of pancreatic cancer patient-derived xenograft tumors and implications for anticancer agent testing

Despite improvements in treatment, pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal cancers, with a continuous increase in incidence emphasizing the need for further research and therapeutic development. In recent years, we have developed a collection of &gt;40 patient-derived xenograft (PDX) models from PDAC. In order to determine their relevance for anticancer agent testing, we extensively characterized our models for histology features, whole exome mutations (Hiseq 2000), chromosome rearrangements, gene copy number variations (Affymetrix SNP6) and gene expression (Affymetrix U133 Plus2.0).

PDAC from 65 patients were implanted into immuno-compromised mice, resulting in the development of 42 PDX models (success rate 65%). The PDAC from which models were established included moderate and poorly differentiated tumors and were heterogeneous for stroma content. In patient tumors, we showed fibroblast activation but not stroma content predicted poor patient outcome (p&lt;0.0001) and success of PDX establishment (p&lt;0.01). Correlating with those of the parental tumors, the resulting PDAC_PDX also showed heterogeneity for stroma content and for murine fibroblast activation. As seen in patient tumors, the PDAC_PDX were characterized by frequent chromosomal instability, with 38% of the models presenting a moderate hypoploidy while the 62% remaining ones showed a hyperploidy. Two models showed hypermutation due to mismatch repair deficiency while the others had on average a lower mutation load when compared to other histotypes such as colon, lung or melanoma tumors. In total, we identified more than 9000 genes altered by homozygous deletions, high gene amplifications and/or mutations, most of the models showed alterations in KRAS (81%), TP53 (67%), CDKN2A (64%) and TGFBR2/SMAD4 (64%) genes. Next, PDAC_PDX profiles were merged with those of patient tumors to determine PDAC_PDX subtypes and were found to be of the classical (61%), quasi mesenchymal (QM) (34%) and exocrine like subtype (5%) according to the 62 gene expression signature established by Collisson et al. 2011. Preliminary biomarker analysis revealed certain associations between genomic alterations and transcriptome subtypes, such as more frequent alterations in the TGFB pathway in classical PDX (80%) compared to QM (47%). Further detailed biomarkers analyses also addressing questions of molecular determinants of stroma content, of fibroblast activation and of sensitivity toward anticancer agents will be presented.

Extensive characterization of our PDAC_PDX collection revealed similarities with patient tumors with regards of histology features including stroma content and fibroblast activation. At molecular level, the models showed similar genomic and transcriptomic patterns as those reported for patient PDAC, altogether, proving the value of this collection for drug development investigations.

Citation Format: Peter Bronsert, Tim Kees, Bruno Zeitouni, Anne-Lise Peille, Manuel Landesfeind, Heinz-Herbert Fiebig, Simon Kuesters, Vincent Vuaroqueaux. Subtyping of pancreatic cancer patient-derived xenograft tumors and implications for anticancer agent testing. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 639.

Abstract 2701: Combining whole-exome and RNA-Seq data improves the quality of PDX mutation profiles

Patient-derived xenograft tumor models (PDX) are of increasing interest for anti-cancer agent testing due to their close resemblance to patient tumors. An accurate molecular characterization of the models is essential 1) to select the PDX that best fit the genetic requirements for a successful cancer therapy investigation and 2) to identify potential predictive biomarkers of response. In this study, we evaluated the quality of mutation profiles from whole-exome sequencing (WES) in terms of concordance with previously acquired mutation data in a large collection of PDX. Further, we analyzed the persistence of disclosed mutations at the transcript level with RNA-Seq.

From 339 PDX, DNA was extracted and enriched in exonic regions with Agilent SureSelect kits before Illumina HiSeq 2000 sequencing with a minimum expected average-of-coverage of 100X. Raw paired-end reads were analyzed by a PDX-specific bioinformatics pipeline to identify the human mutation profile. Sequenom Oncocarta and Sanger sequencing data acquired for 29 cancer genes in 272 PDX was used to evaluate the WES mutation profiles. In parallel, 92 PDX were profiled with RNA-Seq (100M sequencing reads required) and we investigated the expressed mutation profiles by comparing with mutations from WES data.

Among 502 point mutations found with classical methods, 95% were retrieved by WES analyses, revealing the very high sensitivity of the PDX-specific bioinformatics pipeline. 5% of mutations were missed because of a low coverage, particularly in the STK11 gene and in the KRAS gene of pancreatic models, possibly due to poor gene enrichment and high mouse stroma content, respectively. Deeper sequencing could potentially overcome this lack of coverage. Additionally, the WES analysis pipeline displays a high specificity, reporting only 1 additional mutation at gene positions covered with the classical methods. Finally, 507 mutations were detected by WES at positions not interrogated by classical methods emphasizing the necessity for next-generation sequencing (NGS) to obtain a comprehensive mutational spectrum. The number of mutations found using RNA-Seq data was on average two times lower and covered 15% of the mutations detected in WES. This was mainly due to the non-expression of genes or isoforms (40%), the mono-allelic expression of genes (30%), and low coverage data (15%). RNA-Seq analysis restricted to expressed genes represents a substantial complement to WES mutation data and enhances understanding of actual gene alterations in cancer cells.

This study demonstrated the high quality of mutation profiles obtained by WES and highlights the importance of integrating expression data to accurately predict the impact of a mutation at the protein level. An accurate molecular characterization of models is crucial for the selection of PDX with a specific genetic background for the evaluation of anticancer agents.

Citation Format: Manuel Landesfeind, Bruno Zeitouni, Anne-Lise Peille, Vincent Vuaroqueaux. Combining whole-exome and RNA-Seq data improves the quality of PDX mutation profiles. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2701.

Abstract A101: Comprehensive genomic profile analyses of small cell lung cancer patient-derived xenografts for pharmacogenomics

Small Cell Lung Cancers (SCLC) are fast growing tumors with frequent neuroendocrine differentiation. They represent 10-15% of lung cancers and arise in heavy smokers. Most SCLC cancer patients respond well to chemotherapy initially but rapidly develop resistance and die within a few months. Thus, the identification of alternative therapeutic approaches is urgently needed to improve patient's prognosis. Since SCLC is rarely treated by surgery, the number of SCLC models developed for pharmacogenomic research is very limited, however we have established 6 patient-derived xenograft (PDX) models from primary and metastasis SCLC. Complementary to histology, growth characteristic and response to SOCs, here we present comprehensive genomic analyses of these models, revealing similarity with patient tumors and druggable genes and pathways.

First we confirmed that the SCLC PDX models retained the morphology characteristics of SCLC tumors and all were heterogeneous in stroma and vasculature content. The median volume doubling time of SCLC-PDX was 7.4 days, which was faster than adenoma (9.6 days) or squamous (9.5 days) NSCLC and slower than large cell NSCLC-PDX (5.7 days). In vivo analyses demonstrated that the SCLC-PDX responded to the SOCs irinotecan and cyclophosphamide but were relatively resistant to cisplatin.

Whole exome sequencing, RNAseq and SNP6.0 microarray data revealed mutations in 1542 genes (bi-allelic mutations occurring in 328 of these) as well as larger chromosomal alterations including high copy number gains in 33 genes (of the 33 gene copy gains found, 28 occurred in a model with amplification of chromosome 1), 3 homozygous gene deletions and &gt;20 gene fusions. Genes involved in cell cycle/apoptosis, ECM and focal adhesion via PI3K/AKT were frequently altered, as identified using KEGG database resources and as in patient tumors, recurrent mutations were seen in TP53, RB1 and SPDYE5. Furthermore, similar to SCLC data reported by the TCGA, we saw alterations of the NOTCH signaling pathway in some models, with mutations in NOTCH 1, 2 or 4 genes as well as downstream signaling molecules (CTBP2, KAT2A, DTX2). A search for drug targetable gene alterations identified mutations in BRAF, ERBB2, ALK, mTOR and JAK3 genes while EGFR, KRAS, PI3KCA and PTEN genes were not affected in these models. We also discovered gene fusions such as IP6K1-TRAIP or TM9SF4 fused with the oncogene SRC. At the gene expression level (RNAseq and Affymetrix U133 2Plus), 5 out of 6 SCLC-PDX expressed high levels of neuroendocrine markers characteristic of SCLC such as ASCL1, DLK1, GRP or CHGA, and had gene signatures characteristic of high proliferation and expression of stem cell markers such as LGR5. Analysis of the model with amplification of chromosome 1 confirmed that gene copy number gains were frequently associated with increased expression. This included the MYC family member MCL1, suggesting MYC-driven tumors and sensitivity toward compounds such as aurora kinase inhibitors.

In conclusion, we show here SCLC-PDX closely resemble to patient tumors for histology and genomic profiles resulting in fast growing models of various sensitivity toward chemotherapeutics. Numerous drug targetable alterations have been revealed by comprehensive genomic profile analysis. These models therefore represent useful tools for pharmacogenomics investigations.

Citation Format: Vincent Vuaroqueaux, Anne-Lise Peille, Bruno Zeitouni, Vanina Fiebig, Heinz-Herbert Fiebig. Comprehensive genomic profile analyses of small cell lung cancer patient-derived xenografts for pharmacogenomics. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A101.

Abstract A20: A systematic patient-derived xenograft based solution for pre-clinical biomarker discovery

There is an acute need for biomarkers at every phases of drug development from selecting preclinical models in pharmacogenomic studies to enrollment of patients in clinical trials. However, their identification remains extremely challenging due to the limited availability of clinical samples. In contrast, standard tumor models such as cell lines are available but are genetically relatively far from patient tumors.

Use patient-derived xenografts (PDX) for anticancer agent-testing is of increasing interest due to their closer similarity to patient tumors compared to cell lines. Over the last 30 years, we have established a collection of 400 PDX covering more than 30 different cancer types. PDX models have been extensively characterized using the microarray or next-generation sequencing technologies for gene expression, copy number variations and whole-exome mutations. Biomarker research is now possible using these data in combination with drug response data from in vivo or in vitro 2D or 3D assays routinely performed on-site with large panels of 100-200 PDX.

We present here a fully integrated bioinformatics pipeline dedicated to biomarker discovery in which the complete molecular profiles of our PDX have been systematically tested for association with drug sensitivity. To identify the biomarkers associated with drug response, several statistical tests have been performed. Drug response data were treated either as continuous variables using the Spearman or Wilcoxon tests, or as categorical variables (with two groups of responders and non-responders) using the LIMMA, t-test or Fisher exact test. Given that high throughput data frequently leads to large biomarker lists, we used specific filters to narrow down the list of candidates by defining thresholds based on corrected p-values, by intersecting results from different tests, or by integrating the tumor type into the statistical tests. Since sensitivity to anticancer agents is often multi-factorial, we also used integrative approaches that combined gene mutations, copy number loss and lack of gene expression for association with drug response. Finally, significant biomarkers were visualized using clustering heatmaps and enrichment GO/pathway approaches to get more insight in their biological function.

Using a selection of several datasets of PDX drug responses to chemotherapeutics and targeted therapies (targeting RTK/RAS/RAF and PI3K/MTOR pathways and using specific compounds such as Vemurafenib, Erlotinib or Cetuximab), we demonstrate the efficacy of our approach to retrieve biomarkers of known clinical utility. Using these datasets we also could address the questions of model panel sizes, molecular data type and tumor subtype representation, and show how more accurate biomarkers can be validated using an independent dataset of samples.

The development of strategies for testing anticancer agents using PDX in mouse clinical trials, or high throughput in vitro 2D, 3D screening approaches coupled to a more systematic biomarker research should significantly contribute to early biomarker identification and facilitate drug development.

Citation Format: Bruno Zeitouni, Anne-Lise Peille, Zakia Amalou, Thomas Metz, Heinz-Herbert Fiebig, Vincent Vuaroqueaux. A systematic patient-derived xenograft based solution for pre-clinical biomarker discovery. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A20.

Abstract A102: Molecular profiling of a non-small cell lung PDX collection by whole exome sequencing and RNAseq revealed subtype specificities with therapeutic implications

Non-small cell lung cancer (NSCLC) is a heterogeneous disease comprising different histological subtypes with therapeutic implications. We have established a collection of 76 NSCLC patient-derived xenografts (PDX) comprising adenocarcinoma (39), squamous cell carcinoma (24) and large-cell carcinoma (11) models (2 mixed) for use in anti-cancer drug development and pharmacogenomics research. We analyzed the molecular diversity of our PDX collection using whole-exome sequencing, Affymetrix SNP6.0 and RNAseq, assessed their similarity to patient tumors and the impact on drug testing.

Similar to patient lung cancers, genomic analyses revealed that the NSCLC PDX were heavily altered tumors with an overall mutations load ranking from 317 to 1549 per model (mean = 682 mutations). We retrieved signatures of mutational processes known to be associated with ageing, smoking and APOBEC alterations. In addition to the mutations, chromosomal instability characterized by polyploidy including focal gene amplifications (most frequently 2p11.2, 14q32.33) and deletions (e.g. recurrent deletions of 3p14.2, 9p21.3) and gene fusions were detected. While no significant differences in average mutation loads or levels of polyploidy between the histotypes were found, histotype-specific gene alterations and gene expression profiles were identified.

In line with in vivo resistance to anti-EGFR treatment, alterations of the RAS/RAF pathway, PIK3CA and PTEN, as well as the MET amplification occurred more frequently in adenocarcinoma and large cell PDX explaining the resistance of 90% of models to anti-EGFR. Among adenocarcinoma and large cell PDX, we identified three models with EGFR mutations, one of which contained the codon insertion EGFRM766X associated with a strong sensitivity towards cetuximab but not erlotinib.

In addition, we identified adenocarcinoma PDX models in which KIT was amplified and large cell PDX with deletions in CDKN2A/B, alterations in the RAS/RAF pathway (73%) and epigenetic regulation pathways (64%). Squamous cell PDX were less frequently altered in the RAS/RAF pathway, but frequently had altered TP53 (90%), PIK3CA and PTEN (30%) and high expression of TP63 and FGFR3. Furthermore, a highly expressed FGFR3-TACC3 fusion gene was identified in one squamous cell PDX, which is under evaluation for sensitivity towards anti-FGFR3. Regarding PDX response to standard-of-care agents, squamous cell PDX were the most sensitive to cisplatin whereas adenocarcinoma PDX were the most resistant (p = 0.003). In contrast, both groups were more sensitive to paclitaxel than large cell PDX (p = 0.03). These findings suggest distinct molecular profiles influence the response to cytotoxic agents.

Molecular profiling of our NSCLC PDX collection confirmed its similarity to patient tumors and highlighted the distinct molecular specificities of each subtype, suggesting different therapeutic approaches. This extensive characterization will help improve the selection of PDX models in future drug development.

Citation Format: Anne-Lise Peille, Bruno Zeitouni, Vanina Fiebig, Heinz-Herbert Fiebig, Vincent Vuaroqueaux. Molecular profiling of a non-small cell lung PDX collection by whole exome sequencing and RNAseq revealed subtype specificities with therapeutic implications. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A102.

Abstract B70: High-throughput analysis of 3D tumor colony formation of primary cell suspensions derived from xenografts to identify efficacy of anti-tumor agents in single agent or combination therapy

Recent experience in anti-cancer research has demonstrated that three-dimensional (3D) tumor cell culture models represent a biological assay system, being more relevant than two-dimensional (2D) monolayer cell cultures. Anchorage-independent 3D growth is one of the hallmarks of cell transformation and performance of 3D colony formation assays in semi-solid media is considered as an accurate and stringent measure for detecting growth of malignant, transformed cells.

Here, we describe the high-throughput application of a 3D soft-agar clonogenic assay using primary cell suspensions of patient-derived (PDX) and cell line-derived (CDX) xenografts in a 96 well microplate format, using targeted anti-tumor agents, such as Braf, Wnt pathway, and telomerase inhibitors. Colony formation was determined by automated image analysis. Our data show that this is a more appropriate read-out compared to indirect measurements, such as determination of metabolic activity by measuring intracellular ATP.

Specificity of the assay was demonstrated by testing Braf inhibitors vemurafenib and dabrafenib, which potently inhibited colony formation of V600E mutated tumors of melanoma and colon carcinoma (MEXF 1732, MEXF 672, MEXF 989, HT-144, and Colo 205), whereas Braf wildtype tumors of the same histotypes were less sensitive (MEXF 1792, MEXF 1870, MEXF 535, MEXF 622, and HCT-116). Moreover, anti-tumor activity was determined for several standard of care agents. Erlotinib, for example, exhibited differential in vitro activity, which closely matched results obtained for xenografts when tested in vivo, and EGFR was overexpressed in sensitive tumor models.

The 3D soft-agar clonogenic assay is customizable for different treatment layouts, which makes it a valuable tool for efficacy determinations of single drug treatments in broad screens or drug combinations. By using a 5×5 matrix combination approach, we identified synergistic interactions between Wnt pathway (XAV-939) and telomerase (RHPS4) inhibitors using Bliss independence analysis. These results are in line with a postulated molecular link between Wnt/β-catenin signalling and telomerase expression (Hoffmeyer et al. Science 336, 1549, 2012).

In conclusion, the 3D assay platform provides a fast and robust method for the assessment of anti-tumor agents using primary cell suspensions in vitro without needing to establish permanent cell lines. Screening approaches with single agents and combinations in broad tumor panels are of high value for investigating cytotoxic and new targeted anti-tumor agents, as well as for hypothesis generation and selection of development or treatment strategies.

Citation Format: Armin Maier, Sumeer Dhar, Andreas Maunz, Bruno Zeitouni, Anne-Lise Peille, Torsten Giesemann, Heinz-Herbert Fiebig. High-throughput analysis of 3D tumor colony formation of primary cell suspensions derived from xenografts to identify efficacy of anti-tumor agents in single agent or combination therapy. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr B70.

Abstract A10: Establishment and characterization of a patient-derived non-small cell lung cancer mouse model of acquired resistance towards anti-EGFR treatment

Non-small cell lung cancer (NSCLC) is the largest subgroup of lung cancer, occurring at a frequency of over 80% of lung cancer cases. In up to 30% of NSCLC patients the oncogenic driver of tumor growth is a constitutively activated EGF receptor (EGFR), which plays a critical role in regulating multiple cellular processes, including proliferation, survival and apoptosis. Although these patients gain great benefit from treatment with EGFR tyrosine kinase inhibitors (TKI, e.g. erlotinib or gefitinib), development of resistance is inevitable.

To model the emergence of drug resistance, an EGFR driven, gefitinib sensitive, patient-derived xenograft (PDX) NSCLC model was treated continuously with gefitinib in immunocompromised mice. The dose of daily treatment was adjusted according to tumor growth over a period of up to 91 days. In a first phase, dosing was high (40-50 mg/kg) to eradicate EGFR TKI sensitive cells. At the time point of maximal antitumoral activity dosing was reduced to 20-30 mg/kg ( = low dose) to preserve selection pressure. Between 69 and 91 days after dosing was initiated, drug-resistant tumors emerged in 4 out of 10 mice under high dose treatment. Resistant tumor fragments, which were re-implanted into a new cohort of mice and continuously treated with gefitinib, kept resistance also under high dose treatment. A comprehensive analysis using Western blot (WB), qPCR and sequencing was performed to identify the reason for resistance.

In WB analysis we could show that signalling through EGFR was completely abrogated in all four resistant tumor sublines. Neither secondary mutations in EGFR (ex19-21) or KRAS (ex 1+2) could be detected, nor was the expression of cMET, AXL, HGF, PTEN or HER3 significantly increased in resistant tumors as shown by sequencing and qPCR respectively. However a more comprehensive WB analysis revealed several genes being activated in resistant compared to primary tumors. Depending on the subline, phospho-(p)-cMET, p-AKT, AXL & p-AXL, p-cRAF, p-MEK, p-ERK as well as HER3, IGF-R and ALK were up-regulated in resistant tumors. Based on these data we determined that signalling pathways such as the (RAS)-cRAF-MEK-ERK signalling cascade or enhanced cMET/AKT signalling were activated in drug-resistant tumors. These signalling pathways are known to be alternative pathways for EGFR signalling also in patients with acquired resistance indicating the clinical relevance of these models. To shed more light into the mechanism of resistance, whole exome sequencing analyses of the four resistant sublines as well as the original tumor model are underway.

In summary, we have developed four NSCLC tumor sublines each harbouring a different mechanism of resistance to EGFR TKI treatment, modelling the emergence of drug-resistant NSCLC in patients. Herewith a strong preclinical tool for the development of innovative compounds targeting acquired EGFR resistance is now available.

Citation Format: Cordula Tschuch, Kerstin Klingner, Dorothee Lehnhard, Anne Löhr, Yana Raeva, Anne-Lise Peille, Eva Oswald, Julia B. Schüler. Establishment and characterization of a patient-derived non-small cell lung cancer mouse model of acquired resistance towards anti-EGFR treatment. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A10.

Abstract 1705: Classification of colorectal PDX into transcriptomic subtypes associated with distinct genomic alteration profiles and in vivo response patterns to therapies

Recently, unsupervised gene expression-based signatures with prognostic and potential predictive implications were proposed for colorectal cancer (CRC) classification. The challenge is now to characterize the genomic alterations of CRC subtypes and their sensitivity to therapies. Patient derived xenograft models (PDX) may be a valuable tool for this purpose, as they retain the molecular features and drug response patterns of their parental patient tumors. In this study, we classified our collection of colon PDX into transcriptomic subtypes and investigated the associations with genomic alterations and in vivo responses to cetuximab (CTX), oxaliplatin (OXT) and irinotecan (IR). We determined expression profiles of 67 PDX models using Affymetrix HGU133 Plus2.0 arrays and applied the CRC assigner-786 gene expression signature reported by Sadanandam et al. to classify our PDX into 5 subtypes. We identified 34 (51%) transit-amplifying (TA), 15 (22%) goblet-like (GL), 10 (15%) inflammatory (IF), 7 (10%) enterocyte (ET) and 1 (1%) stem-like (SL) PDX models. TA showed a gene signature of WNT pathway activation whereas GL, ET and IF PDX subtypes harbored gene signatures of KRAS pathway activation. As analyzed by whole exome sequencing and Affymetrix SNP6.0 array, the GL, ET and IF subtypes were found to display different signatures of mutational processes, mutations and chromosomal rearrangement patterns than TA and SL. Of particular note, mutations in BRAF, PIK3CA/PTEN, TGFBR2/SMAD4 or NOTCH1 were mainly found in GL, ET or IF while mutations in APC, TP53 and KRAS were not associated with a given subtype. Regarding PDX drug sensitivity, 10/11 TA/SL PDX with unaltered KRAS/PIK3CA/PTEN/HER2 status were sensitive to CTX whereas 9/12 models with alterations were resistant. TA and SL PDX were also frequently sensitive to IR and OXT (6/10 and 4/10, respectively). In the IF subtype, 6/6 PDX showing KRAS, PIK3CA or PTEN alterations were resistant to CTX. Of interest, 5/5 IF PDX were sensitive to IR treatment. As in IF PDX, GL and ET often had alterations in KRAS/PIK3CA/PTEN/HER2 and were resistant to CTX. Moreover, these subtypes were resistant to both IR and OXT. Furthermore, the frequent KRAS pathway activation and the presence of actionable mutations in GL, ET and IF would advocate the testing of inhibitors (as monotherapy or in combination) of BRAF, PI3K or MEK in tumors of these subtypes while WNT inhibitors should be tested in TA models.Molecular classification of our colon PDX collection confirmed their similarities with patient tumors. Investigation of drug sensitivity resulted in the identification of PDX subtypes which respond poorly to standard therapies and which require different treatment options. Testing alternative and combined therapies in PDX may be a valuable approach to optimize personalized medicine.

Citation Format: Anne-Lise Peille, Christina Gredy, Bruno Zeitouni, Armin Maier, Kerstin Klingner, Tim Kees, Julia Schüler, Thomas Metz, Heinz. Herbert Fiebig, Vincent Vuaroqueaux. Classification of colorectal PDX into transcriptomic subtypes associated with distinct genomic alteration profiles and in vivo response patterns to therapies. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1705. doi:10.1158/1538-7445.AM2015-1705

Anti-tumor activity of the TGF-β receptor kinase inhibitor galunisertib (LY2157299 monohydrate) in patient-derived tumor xenografts

Purpose

The transforming growth factor-beta (TGF-β) signaling pathway is known to play a critical role in promoting tumor growth. Consequently, blocking this pathway has been found to inhibit tumor growth. In order to achieve an optimal anti-tumor effect, however, it remains to be established whether blocking the TGF-β signaling pathway alone is sufficient, or whether the tumor microenvironment plays an additional, possibly synergistic, role.

Methods

To investigate the relevance of blocking TGF-β signaling in tumor cells within the context of their respective tissue microenvironments, we treated a panel of patient-derived xenografts (PDX) with the selective TGF-β receptor kinase inhibitor LY2157299 monohydrate (galunisertib) and assessed both the in vitro and in vivo effects.

Results

Galunisertib was found to inhibit the growth in an in vitro clonogenic assay in 6.3 % (5/79) of the examined PDX. Evaluation of the expression profiles of a number of genes, representing both canonical and non-canonical TGF-β signaling pathways, revealed that most PDX exhibited expression changes affecting TGF-β downstream signaling. Next, we subjected 13 of the PDX to an in vivo assessment and, by doing so, observed distinct response patterns. These results suggest that, next to intrinsic, also extrinsic or microenvironmental factors can affect galunisertib response. pSMAD2 protein expression and TGF-βRI mRNA expression levels were found to correlate with the in vivo galunisertib effects.

Conclusions

From our data we conclude that intrinsic, tumor-dependent TGF-β signaling does not fully explain the anti-tumor effect of galunisertib. Hence, in vivo xenograft models may be more appropriate than in vitro clonogenic assays to assess the anti-tumor activity of TGF-β inhibitors such as galunisertib.

Electronic supplementary material

The online version of this article (doi:10.1007/s13402-014-0210-8) contains supplementary material, which is available to authorized users.

Abstract 3699: Molecular profiling of BRAFi-resistance in melanoma cancer models using high-throughput sequencing in patient-derived xenografts

Patients with metastatic V600E mutant melanomas treated with BRAF inhibitors (BRAFi) frequently develop resistant tumors with aggressive phenotypes. Furthermore, approximately 20-40% of V600E mutant tumors are intrinsically resistant to BRAFi. Suitable models for studying mechanisms of acquired and intrinsic resistance to BRAFi are therefore necessary. In the present study, we applied deep sequencing techniques to identify possible mechanisms of intrinsic and acquired resistance in our collection of melanoma patient-derived xenograft models (MEXFs) treated with BRAFi.

Mutational and expression profiles in MEXFs were characterized by whole-exome sequencing (WES) and HG-U133 Plus 2.0 Affymetrix chips and correlated with BRAFi efficacy data from 3D Tumor Clonogenic Assays (TCA). In addition, four resistant cell lines were created by continuously treating 2D monolayer cultures of tumor cells with Vemurafenib, all of which were initially sensitive to BRAFi and carry the BRAF V600E mutation. Expression profiles and mutations of these cell lines were analyzed from RNA-seq data. Potential gene candidates responsible for conferring resistance were further investigated by Q-PCR and Western-blot experiments.

WES data revealed that the number of mutations per model was highly variable. Some MEXFs showed a hyper-mutated profile (&gt;2000 mutations) and were characterized by specific mutational signatures in agreement with those found in melanoma (Alexandrov et al, Nature, 2013). Mutation frequencies of genes typically mutated in melanoma are very similar to those found in The Cancer Genome Atlas. We observed a high correlation of mutations between our MEXFs and their respective cell lines. Among the models with a V600E mutation, only one (MEXF 462) showed resistance to BRAFi-treatment as identified by 3D TCA analyses. In this model, we identified gene point mutations and a high over-expression of EGFR, MEK1, PDGFRA and NF1 in contrast to the other models, suggesting a specific regulation of the RTK signaling pathways. In 2D assays, synergistic interaction of Vemurafenib and Erlotinib was shown. RNA-seq analysis of the cell line established from MEXF 276, in which resistance to Vemurafenib was induced, revealed around 20% of genes being differentially expressed (FC&gt;2) between sensitive and resistant cell lines. An up-regulation of EGFR was also found in this resistant cell line and was confirmed by Western-blot. In addition, an overexpression of PLAU, a biomarker of invasiveness, was identified. We currently perform expression profiling of three other models resistant to BRAFi. Preliminary results suggest different patterns of gene regulation involved in acquisition of resistance. A precise map of transcriptomic and mutational profiles of the four cell lines will be generated and we are investigating if invasiveness is increased upon acquisition of resistance.

Citation Format: Bruno Zeitouni, Gerhard Kelter, Armin Maier, Florian Kiefer, Frederic Foucault, Anne-Lise Peille, Tim Kees, Torsten Giesemann, Vincent Vuaroqueaux, Thomas Metcalfe, Heinz-Herbert Fiebig. Molecular profiling of BRAFi-resistance in melanoma cancer models using high-throughput sequencing in patient-derived xenografts. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 3699. doi:10.1158/1538-7445.AM2014-3699

Abstract LB-314: Whole exome sequencing analyses of gastric cancers reveal two distinct genomic alteration patterns with implications in drug sensitivity

Gastric cancer is the fourth most common cancer diagnosed and the second most frequent cause of cancer-related death worldwide. Multiple factors can contribute to the development of gastric cancer, including H. pylori infection, dietary behaviour and life style, possibly resulting in distinct cancer subtypes with different drug sensitivity profiles. In the present study we searched for gastric cancer mutation patterns in the dataset of the “The Cancer Genome Atlas” (TCGA) and in our collection of patient derived xenografts (PDX). In a second part, we evaluated gene alteration patterns for their implications for drug sensitivity.

In both TCGA and our PDX datasets, Whole Exome Sequencing analyses revealed two subsets of gastric tumors characterized by specific mutation signatures, with different types and numbers of genomic alterations. The first subset (60% and 75% of samples) contained lower levels of mutations and was characterized by increased numbers of large chromosomal rearrangements resulting in gene loss or amplifications. The second subset of tumors (25%-40% of samples) revealed higher levels of mutations that were predominantly nucleic acid substitutions and small indels linked to mismatch repair genes including MLH1 or MSH3 and to high microsatellite instability. In both subsets, the mutation spectrum was dominated by C&gt;T transitions with an increase of small indels in the subset of highly-mutated tumors.

At the gene level, the genes which were mutated in our gastric PDX collection overlapped to great extent with the mutations found in TCGA tumors, especially regarding the most frequently mutated genes. In the first subset, high levels of gene amplifications and deletions were found, including growth factor receptor amplifications in EGFR and HER2. Furthermore, the mutation frequency in genes associated with drug resistance such as KRAS was decreased. The tumors with growth factor receptor amplification responded consistently to therapies such as Cetuximab or Trastuzumab. In contrast, an increased frequency of mutations in oncogenes and tumor suppressors, including KRAS (n=5/10), PIK3CA (n=5/10) and PTEN (n=7/10), was found in the second subset. The mutational profile of these tumors suggest the use of compounds targeting downstream molecules, such as PIK3CA, or targeting effectors of DNA repair, such as PARP, for anti-cancer therapy. Of note, no association was found between the mutation groups and sensitivity to chemotherapeutic agents such as 5FU, Cisplatin or Paclitaxel.

In conclusion, we identified two subsets of gastric tumors both in the TCGA dataset and in our collection of PDX models, characterized by distinct genomic alteration profiles suggesting different therapeutic approaches. Currently, we are assessing drug sensitivity profiles within the two subsets in our PDX models.

Citation Format: Anne-Lise Peille, Swee-Seong Wong, Florian Kiefer, Bruno Zeitouni, Armin Maier, Frederic Foucault, Tim Kees, Vincent Vuaroqueaux, Amit Aggarwal, Christoph Reinhard, Heinz Herbert Fiebig. Whole exome sequencing analyses of gastric cancers reveal two distinct genomic alteration patterns with implications in drug sensitivity. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr LB-314. doi:10.1158/1538-7445.AM2014-LB-314

Abstract 100: Establishment and characterization of allografts derived from a genetically engineered mouse model of non-small cell lung cancer (NSCLC)

Introduction: Genetically engineered mouse models (GEMM) represent an attractive system for preclinical research since GEMM tumors develop in the presence of a competent immune system, thereby closely resembling the tumor microenvironment in patients. One example is the KP model (DuPage et al., 2009), which reflects the tumorigenesis of NSCLC in humans. KP tumors carry Kras and Trp53 mutations, comparable to subtypes of human NSCLC. Typical drawbacks of GEMM tumors include differences in their genetic make-up to human tumors and the very slow and heterogeneous tumor development, making the use of these models challenging for routine in vivo efficacy studies. The latter can be improved by grafting primary tumors on a genetically dissimilar member of the same strain, so-called allografting. Thus, the advantages of xenograft and GEMM models can be combined. Here, we described the establishment and characterization of allografts from the KP GEMM.

Material & Methods: KP tumors were excised from the lungs of KP mice and implanted subcutaneously (s.c.) into C57BL/6N female mice. When tumor growth was detected, passaging steps from animal to animal were performed. Furthermore, two cell lines derived from KP tumors, KP1 and KP4 (provided by Müller et al.), were injected s.c., the developing tumors excised and directly re-implanted as described above. The passages were analyzed molecularly, histologically, and via IHC (e.g. Ki67 proliferation marker) , and compared to the original in-situ tumors. A therapeutic study with Erlotinib and BEZ235 was performed on KP1 allografted mice.

Results: All investigated allografts showed similar histological patterns. Ki67 scores reflected the average time from implantation to passaging. The average passage times ranged from 19±3 days (KP4 cell line) to 26±6 days (KP1 cell line). Allografts from the original KP model showed a time to passaging of 23±5 days.

The mRNA expression of EpCAM (CD326) was on average 10-fold higher in KP and KP1 allografted tumors compared to normal C57BL/6 lung tissue, which could be confirmed by flow cytometry and cell-binding assay, determining &gt;99% EPCAM positive cells in KP1 tumors. All allografts carried the Kras mutation and loss of function point mutations in Trp53.

The therapy study revealed no effect of Erlotinib on tumor growth, but a significant (p=0.0003) tumor load reduction in mice treated with BEZ235 (519 ± 244 mm³, n=5) compared to the control group (1678 ± 275 mm³, n=4) on day 28, respectively.

Conclusion: The generation of s.c. transplantable allografts from the KP GEMM was successfully conducted. The allograft models enable the use of immunocompetent mice in a feasible time frame and still resemble key aspects of human disease. Further studies will elucidate the interaction of the allografts with the host and compare sensitivity characteristics between GEMM and allograft of the same model.

Citation Format: Damaris Kukuk, Philipp Müller, Kerstin Klingner, Anne-Lise Peille, Alfred Zipelius, Julia B. Schüler. Establishment and characterization of allografts derived from a genetically engineered mouse model of non-small cell lung cancer (NSCLC). [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 100. doi:10.1158/1538-7445.AM2014-100

Abstract C25: The use of a patient derived tumor xenograft collection to assess different Met and HGF detection methods and their predictive values for therapy response

Aims: Met inhibitors are targeted drugs which hold promise for the treatment of tumors with HGF/Met pathway activation. To accurately identify such tumors, relevant and well validated predictive biomarkers are needed. In the present study, we used our large collection of tumor models to investigate the value of Met and HGF as determined by multiple methods for predicting sensitivity to Met-inhibitors.

Materials and Methods: HGF and Met were assessed at DNA, RNA and protein levels in a panel of 292 tumor models covering 15 different histotypes. The tumors were analyzed for Met amplification and protein expression in a tissue microarray analysis using standard FISH and IHC assays. Met and HGF were also assessed for gene copy number variations, mutations, mRNA and protein expression levels and phosphorylation. For relevant tumors, sensitivity to the Met inhibitors PF-04217903 and/or JNJ-38877605 was evaluated by Tumor Clonogeneic Assays (TCA) and/or in vivo testing.

Results: Regarding Met and HGF expression, results obtained at the DNA, RNA and protein level were highly consistent. Affymetrix analyses showed expression of Met mRNA in a large proportion of the models, with a good correlation to protein levels detected by IHC or Western blot. Western blot analyses revealed that most tumors which overexpressed Met showed a high phosphorylation level of the receptor. The FISH and SNP6 analyses allowed identification of a subset of tumors with Met gene amplification, associated with overexpression of Met mRNA, protein and receptor phosphorylation. These models were sensitive to Met-inhibitors in both TCA and in vivo assays. In addition, a subset of tumor xenografts displayed very high Met expression levels and sensitivity to Met-inhibitors, although they did not carry an amplified Met gene. IHC analyses revealed that Met was diversely expressed in some tumors, indicative of intra tumoral heterogeneity that could impact on response to Met inhibitors. HGF, both at mRNA and protein level, was detectable in a subset of tumors, including tumors which did not overexpress Met. Some of these tumors were strongly sensitive to Met inhibitors, suggesting a predictive value of HGF expression. Of interest, these tumors had different topography, such as the liver, the kidney and the lung as well as sarcomas.

Conclusion: This study reveals Met overexpression as one of the major determinants of tumor sensitivity to Met-inhibitors. Accurate assessment of Met levels by mRNA array, western blot and IHC analyses showed that Met overexpression was not only due to gene amplification, demonstrating the importance of gene expression assessment complementary to gene copy number evaluation. The in situ analysis also demonstrated that intra-tumoral heterogeneity in Met expression levels can be of importance. Furthermore, HGF levels may be a good predictor for response to Met inhibitors in tumors. Taken together, out data show that in addition to the well-established ex vivo 3D assay with PDX, Met and HGF expression emerges as a powerful parameter to predict tumor response to cMet-directed therapies in vivo.

Citation Information: Mol Cancer Ther 2013;12(11 Suppl):C25.

Citation Format: Vincent Vuaroqueaux, Torsten Giesemann, Luigi Tornillo, Armin Maier, Rebekka Krumbach, Anne-Lise Peille, Tim Kees, Jianing Guo, Frederic Foucault, Zakia Amalou, Serenella Eppenberger, Luigi Terracciano, Heinz-Herbert Fiebig. The use of a patient derived tumor xenograft collection to assess different Met and HGF detection methods and their predictive values for therapy response. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr C25.

Abstract B111: Molecular and chemosensitivity profiling of melanoma cell lines with acquired resistance to BRAF inhibitor Vemurafenib

Introduction: One of the main limitations of patient benefit from targeted therapies is the rapid development of resistance. The BRAF inhibitor Vemurafenib is approved for BRAF mutant melanoma. Here we generated four Vemurafenib resistant melanoma cell lines and found that each had a unique profile of receptor tyrosine kinase over or underexpression, compared to the parent cell lines.

Methods: Cell lines used were the commercially available HT-144 and three Oncotest patient-derived xenograft-derived cell lines MEXF 276L, MEXF 394L and MEXF 520L, all sensitive to Vemurafenib and carrying the BRAF V600E or V600K mutation. Each was treated continuously with increasing Vemurafenib concentration up to 6 µM.

Results: Upon continuous Vemurafenib treatment all four melanoma cell lines developed Vemurafenib resistance. The resistant cell lines had between 19 and 168 fold higher IC50 values compared to the mock-treated lines. These IC50 values are comparable with IC50 values for cell lines without BRAF mutations, meaning that these cell lines are truly resistant to Vemurafenib.

Chemosensitivity profiling revealed cross-resistance to other BRAF inhibitors (n=3) as well as to MEK inhibitors (n=6), while for AKT, c-MET, Eg5, EGFR or PI3K inhibitors sensitivities were mostly similar to those of the parental cell lines. The cross-resistance with BRAF and MEK inhibitors suggests a mechanism of resistance either downstream of MEK or independent of the RAS/RAF/MEK pathway.

Excess receptor tyrosine kinase (RTK) expression has been associated with resistance of BRAF mutant melanomas to Vemurafenib. Molecular analyses showed that each of the four resistant cell lines displayed an altered expression of RTKs with each showing a different combination of over- and under-expressed receptors. Downstream phosphorylation of Erk1/2, Akt and S6RP were also found to be altered in the resistant cell lines compared to their sensitive counterparts.

Conclusions: Four BRAF-mutant melanoma cell lines acquired resistance to BRAF inhibitor Vemurafenib with cross- resistance to other BRAF and MEK inhibitors. However, the cross resistance seems to be the only similarity between the resistant cell lines, as changes in RTK expression and downstream Erk1/2, Akt and S6RP phosphorylation were unique to each cell line.

This leads to the conclusion that each of the four resistant cell lines may represent a resistance mechanism present in a sub-group of the patient population.

Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B111.

Citation Format: Rebekka Krumbach, Anne-Lise Peille, Torsten Giesemann, Vincent Vuaroqueaux, Heiner Fiebig, Thomas Metcalfe, Gerhard Kelter. Molecular and chemosensitivity profiling of melanoma cell lines with acquired resistance to BRAF inhibitor Vemurafenib. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B111.

Abstract 2774: The molecular determinants of sensitivity to HER2 targeted therapy in Patient Derived Xenograft gastric tumor models from Caucasian and Eastern Asian patients

Introduction: Gastric cancer is common in Asia and Eastern Asia where more than half of the world's cases arise. As is the case for Caucasian patients, amplification and overexpression of the HER2 gene is a critical event in the tumorigenesis of gastric cancer and is present in 10 to 20% of Asian patients. Trastuzumab has been approved for treatment of HER2-positive gastric cancer. However, primary and secondary resistance to Trastuzumab is a significant problem and new strategies to overcome resistance are needed. Models which recapitulate this differential response aid the development of new therapies targeting HER2. We recently enlarged our collection of gastric cancer Patient Derived Xenografts (PDX), which in the past focused on Caucasian patients, by developing PDX from Eastern Asian patients. In this study we investigated whether the newly developed PDX collection is reflective of the clinical situation and contains HER2 amplified/overexpressing tumors. We evaluated the sensitivity to Trastuzumab treatment of gastric cancer PDX which are HER2 amplified/overexpressing and searched for possible additional molecular determinants of sensitivity.

Material and Methods: Gastric tumors were xenografted in nude mice and were characterized by Affymetrix SNP V6.0 array and qPCR for gene copy number variation, Sanger sequencing and Sequenom MassARRAY OncoCarta panels 1, 2 and 3 for mutations, Affymetrix HGU133 plus 2.0 arrays for gene expression and using immunohistochemistry (IHC) for protein expression. Response to HER2-targeted therapy was assessed in vivo by treating gastric PDX with Trastuzumab 10 mg/kg/day at days 7, 14, and 21.

Results: A total of 29 gastric cancer PDX were established (7 PDX of Caucasian and 22 of Eastern Asian origin). Among these 29 PDX, 1 PDX from a Caucasian patient and 4 from Eastern Asian patients expressed high amounts of HER2 mRNA due to gene amplification that ranged from 10 to &gt;50 copies. IHC in situ analyses revealed that the HER2 amplification/mRNA overexpression correlated with strong HER2 protein expression (score 3+). Ongoing analyses investigating these HER2-amplified/overexpressing PDX for sensitivity to Trastuzumab revealed one PDX sensitive and one PDX resistant to treatment. We will present analyses of key biomarkers such as NRAS/KRAS/BRAF mutations, PIK3CA/PTEN status, HER2 integrity, gene expression profiles and their correlation with sensitivity/resistance to trastuzumab.

Conclusion: In agreement with what is observed in clinical practice, we identified Caucasian and Eastern Asian gastric tumors amplified and overexpressing HER2. The PDX from these tumors allowed the investigation of response to therapy targeting HER2. These PDX will aid in testing new HER2 targeted treatments and in identifying potential molecular determinants of resistance to Trastuzumab and other HER2 targeting agents.

Citation Format: Vincent Vuaroqueaux, Andreas Ackermann, Jianing Guo, Anne-Lise Peille, Rebekka Krumbach, Frederic Foucault, Thomas Metz, Heinz-Herbert Fiebig. The molecular determinants of sensitivity to HER2 targeted therapy in Patient Derived Xenograft gastric tumor models from Caucasian and Eastern Asian patients. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2774. doi:10.1158/1538-7445.AM2013-2774

Correlating global gene regulation to angiogenesis in the developing chick extra-embryonic vascular system

Background

Formation of blood vessels requires the concerted regulation of an unknown number of genes in a spatial-, time- and dosage-dependent manner. Determining genes, which drive vascular maturation is crucial for the identification of new therapeutic targets against pathological angiogenesis.

Methology/Principal Findings

We accessed global gene regulation throughout maturation of the chick chorio-allantoic membrane (CAM), a highly vascularized tissue, using pan genomic microarrays. Seven percent of analyzed genes showed a significant change in expression (>2-fold, FDR<5%) with a peak occurring from E7 to E10, when key morphogenetic and angiogenic genes such as BMP4, SMO, HOXA3, EPAS1 and FGFR2 were upregulated, reflecting the state of an activated endothelium. At later stages, a general decrease in gene expression occurs, including genes encoding mitotic factors or angiogenic mediators such as CYR61, EPAS1, MDK and MYC. We identified putative human orthologs for 77% of significantly regulated genes and determined endothelial cell enrichment for 20% of the orthologs in silico. Vascular expression of several genes including ENC1, FSTL1, JAM2, LDB2, LIMS1, PARVB, PDE3A, PRCP, PTRF and ST6GAL1 was demonstrated by in situ hybridization. Up to 9% of the CAM genes were also overexpressed in human organs with related functions, such as placenta and lung or the thyroid. 21–66% of CAM genes enriched in endothelial cells were deregulated in several human cancer types (P<.0001). Interfering with PARVB (encoding parvin, beta) function profoundly changed human endothelial cell shape, motility and tubulogenesis, suggesting an important role of this gene in the angiogenic process.

Conclusions/Significance

Our study underlines the complexity of gene regulation in a highly vascularized organ during development. We identified a restricted number of novel genes enriched in the endothelium of different species and tissues, which may play crucial roles in normal and pathological angiogenesis.