Gene expression biomarkers differentiate overall survival of colorectal cancer upon targeted therapies

While monoclonal antibody-based targeted therapies have substantially improved progression-free survival in cancer patients, the variability in individual responses poses a significant challenge in patient care. Therefore, identifying cancer subtypes and their associated biomarkers is required for assigning effective treatment. In this study, we integrated genotype and pre-treatment tissue RNA-seq data and identified biomarkers causally associated with the overall survival (OS) of colorectal cancer (CRC) patients treated with either cetuximab or bevacizumab. We performed enrichment analysis for specific consensus molecular subtypes (CMS) of colorectal cancer and evaluated differential expression of identified genes using paired tumor and normal tissue from an external cohort. In addition, we replicated the causal effect of these genes on OS using validation cohort and assessed their association with the Cancer Genome Atlas Program data as an external cohort. One of the replicated findings was WDR62, whose overexpression shortened OS of patients treated with cetuximab. Enrichment of its over expression in CMS1 and low expression in CMS4 suggests that patients with CMS4 subtype may drive greater benefit from cetuximab. In summary, this study highlights the importance of integrating different omics data for identifying promising biomarkers specific to a treatment or a cancer subtype.

Tumor-Myeloid Cell Crosstalk Pathways Associated with Abscopal Responses in Breast Cancer

PURPOSE/OBJECTIVE(S)

To identify predictive biomarkers for combined radiotherapy (RT) and immune checkpoint inhibitor (ICI) therapy, which can induce systemic anti-tumor immune responses (i.e., "abscopal" responses) in breast cancer.

MATERIALS/METHODS

Four Trp53-/- Balb/c breast cancer syngeneic allograft models with low tumor mutational burden and resistance to dual ICI (anti-PD1 and anti-CTLA4, BioXcell) were used. Two lines exhibited abscopal responses to RT (8Gyx3) plus ICI combination therapy (i.e., "abscopal models"), whereas the other two did not (i.e., "non-abscopal models"). We performed spatial transcriptomics analysis (GeoMx whole transcriptome assay) of pan-cytokeratin positive (panCK+) tumor and tumor-adjacent CD45+ immune cell segments in orthotopically implanted tumors 10 days after initiating treatment with IgG control, RT alone, ICI alone, or RT/ICI combination (N>5 per group). Genes selectively induced by RT/ICI in panCK+ and CD45+ segments in abscopal models and were not induced in non-abscopal models were identified using two-tailed t-tests and FDR correction for multiple testing (FDR<5%). In vitro analyses of RT-induced secreted inflammatory signaling were conducted by exposing supernatant from irradiated (8Gy) tumor cells to a RAW264.7 macrophage cell line expressing an interferon-inducible luciferase reporter construct, followed by luciferase signal quantification (Invivogen).

RESULTS

Hierarchical clustering of immune-related genes in CD45+ immune segments from breast cancer models with abscopal responses to RT/ICI revealed global differences in tumor-adjacent immune profiles at baseline and after RT/ICI treatment. Abscopal responsive breast cancer models at baseline showed an enrichment of myeloid cell subtypes expressing complement protein C1q, Ccl8, and Csf1r in tumor-adjacent CD45+ immune cells, and CD8+ T cells expressing Ccl5. 10 days after treatment with RT/ICI, tumor-adjacent CD45+ immune cells in abscopal models were enriched in Cxcl10, Irf7, and FcgRIV, whereas these genes were not induced in non-abscopal models. CyTOF analyses confirmed the induction of inflamed professional antigen presenting cells (CD11c+CD80+IA/IE+) by RT/ICI in abscopal models. Within panCK+ tumor segments, RT/ICI in abscopal models selectively induced Isg15 and Zbp1, both of which are involved in secreted inflammatory signaling. Consistently, RT-induced secreted inflammatory signaling from tumor cells to macrophages evaluated in vitro was significantly greater in abscopal models compared to non-abscopal models.

CONCLUSION

Breast cancer models with abscopal responses to RT/ICI show increased Isg15- and Zbp1-dependent secreted inflammatory crosstalk that activates Cxcl10 and Irf7 expression in tumor-adjacent myeloid cells, which warrants further investigation as a potential predictive biomarker for combined RT/ICI therapy in breast cancer.

Abstract P2-11-08: Multimodal Prediction of Breast Cancer Recurrence Assays and Risk of Recurrence

Background: Hormone receptor positive breast cancer constitutes about 70% of newly diagnosed early-stage disease in the United States, and gene-expression based recurrence assays such as Oncotype DX (ODX) are strongly recommended by guidelines to aid in treatment decisions. However, recurrence assays are costly, time-consuming, underutilized in low resource settings, and unavailable in developing countries. Deep Learning (DL) using hematoxylin and eosin (H&E) stained digital pathology has been shown to approximate gene expression patterns for multiple cancer types, and may provide a cost-effective, fast, and scalable method to predict risk of recurrence in community settings. Methods: We first developed a model for ODX using only DL on pathology, comprised of two Xception-based modules, trained on 1,039 slides from The Cancer Genome Atlas (TCGA) tessellated into 10x magnification image tiles. The first module predicts tumor likelihood, and was trained using pathologist annotations for tumor regions versus normal stroma. The second module was trained to predict ODX score, estimated from gene expression data within TCGA. Patient-level predictions were calculated by weighting the predicted recurrence score by tumor likelihood for all tiles within a slide. Separately, ODX score was predicted from clinical variables using the University of Tennessee Nomogram, which incorporates grade, progesterone receptor, size, age, and histologic subtype. Finally, we developed a combined model by fitting a logistic regression to the DL pathologic model and the clinical nomogram predictions. Performance of the clinical nomogram, pathologic, and combined models were then compared in a single-institution external validation cohort of patients diagnosed with breast cancer between 2006 and 2020, all of whom had the commercial ODX assay run. Results: We identified 428 cases for our diverse validation cohort (69% White, 24% Black, 6% Asian, and 3% Hispanic) with mean ODX score of 18. Chemotherapy was administered for 104 (24.3%) of patients, the remaining 323 (75.4%) received endocrine therapy alone. Area under the receiver operating characteristic curve (AUROC) for prediction of high ODX score (≥ 26) of the combined model was 0.83 (95% confidence interval [CI] 0.78 – 0.89) in the validation cohort, which was higher than either the DL pathology model (AUROC 0.80, 95% CI 0.75 – 0.85, p = 0.026) or the Tennessee nomogram (AUROC 0.77, 85% CI 0.70 – 0.83, p = 0.003). Performance was similar in Black (AUROC 0.86, 95% CI 0.78 – 0.94) and White (AUROC of 0.81, 95% CI 0.74 – 0.88) subgroups. The combined model was more accurate in prediction of recurrence-free interval in patients receiving endocrine therapy (hazard ratio [HR] 2.02 per standard deviation [SD], 95% CI 1.16 – 3.52, p = 0.013, Concordance [C]-index 0.75) than the clinical nomogram (HR 1.75 per SD, 95% CI 1.09 – 2.81, p = 0.021, C-index 0.68). No model was prognostic in patients receiving chemotherapy. Pathologist review of heatmaps of DL model predictions identified lymphovascular invasion, necrosis, high grade, and infiltrative borders as features contributing to model prediction of high risk. Conclusions: DL can improve on existing clinical prediction of breast cancer with low recurrence risk. This approach could improve the speed at which treatment decisions are made due to the time-consuming nature of genomic testing and simultaneously reduce the cost of care. Given the equal performance in racial subgroups, this approach has promise for application in global health settings where genomic assays are not widely available or are prohibitively expensive.

Citation Format: Frederick M. Howard, James M. Dolezal, Sara Kochanny, Galina Khramtsova, Jasmine Vickery, Andrew Srisuwananukorn, Anna Woodard, Nan Chen, Rita Nanda, Charles Perou, Olufunmilayo I. Olopade, Dezheng Huo, Alexander Pearson. Multimodal Prediction of Breast Cancer Recurrence Assays and Risk of Recurrence [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P2-11-08.

14-gene immunoglobulin (IGG) and proliferation signatures and association with overall survival across cancer-types

2636

Background: The HER2DX prognostic assay in early-stage HER2-positive breast cancer integrates clinical variables and 4 gene expression signatures (GES) tracking IGG, tumor proliferation, luminal cell differentiation, and the expression of the HER2 amplicon. Here, we assessed the prognostic value across cancer-types of each of these four individual GES and a research-based version of the HER2DX risk-score. Methods: RSEM batch normalized RNA-sequencing gene expression data from The Cancer Genome Atlas (TCGA) project were downloaded from cBioPortal. The association between a research-based version of the HER2DX risk-score, and each GES, with overall survival (OS) was assessed as continuous variables using univariate Cox regression model. The research-based version of the HER2DX risk-score tested did not include clinical variables. The Cox model was applied to estimate hazard ratios (HR) with 95% confidence intervals. The threshold for statistical significance was set at p< 0.05. Results: Gene expression data from 9,852 patients representing 30 different cancer-types were evaluated. The proliferation, IGG, luminal and HER2 amplicon GES were significantly associated with OS in 40.0%, 33.3%, 23.0% and 6.7% of the cancer-types tested, respectively. The IGG GES was found significantly associated with a favorable OS in breast cancer (HR: 0.73, p < 0.001), cervical and head & neck squamous cell carcinomas (HR: 0.75, p = 0.021 and HR: 0.78, p < 0.001), lung adenocarcinoma (HR: 0.84, p = 0.020), skin melanoma (HR: 0.73, p < 0.001) and sarcomas (HR: 0.78, p = 0.029). Conversely, association of IGG GES with unfavorable OS was observed in uveal melanoma (HR: 1.74, p = 0.008), kidney clear cell and papillary cell carcinomas (HR: 1.29, p < 0.001 and HR: 1.44, p = 0.017, respectively), as well as brain low-grade gliomas (LGG) (HR: 1.56, p < 0.001). Finally, the research-based HER2DX risk-score was found significantly associated with OS in 11 of 30 (36.7%) cancer-types, including breast (HR: 1.42, p < 0.001) cervical and head & neck squamous cells cancers (HR: 1.36, p = 0.018 and HR: 1.30, p < 0.001), lung adenocarcinoma (HR: 1.34, p < 0.001), skin melanoma (HR: 1.34, p < 0.001), sarcomas (HR: 1.41, p = 0.003), adrenocortical carcinoma (HR: 2.31, p < 0.001), endometrial carcinoma (HR: 1.36, p = 0.005), hepatocarcinoma (HR: 1.26, p = 0.012), LGG (HR: 1.37, p = 0.001) and mesothelioma (HR: 1.48, p = 0.005). Conclusions: The 14-gene IGG and proliferation signatures are strong prognostic biomarkers across cancer-types. The opposite association of IGG with OS depending on the cancer-type warrants further investigation.

Abstract P3-09-08: An automated tool to determine optimal cluster numbers in single-cell RNA sequencing data identifies key prognostic subsets of T cells in breast tumors

Background: Single-cell RNA sequencing (scRNAseq) enables gene expression profiling of individual cells, thus providing new opportunities to identify and characterize cell types and tumor cell states, and which is typically accomplished through some type of clustering analysis. Determination of the optimal number of groups found is a crucial step in data interpretation, and one that is often ignored. Hence, there is still a great need to develop objective tools for estimation of the data driven optimal number of groups/clusters. Methods: We developed “MultiK”, a data-driven tool for objective selection of the optimal number of groups/clusters, which combines multiple resolution solutions together through a consensus clustering approach based upon repeated sub-sampling. MultiK gives multiple diagnostic plots to highlight the number of meaningful groups in the data, and makes objective group number suggestions, which encompasses both high and low-resolution parameters. Results: MultiK successfully identified the ground truth number of groups in a controlled data set of a mixture of 3 breast cell lines, and was sensitive in identification of classes and subclasses in a synthetic “spike in” experiment. We further applied MultiK to identify reproducible groups in complex tissue datasets, including mouse mammary glands and multiple T cell data sets. In both cases, we identified most of the previously known subsets/cell populations and did so without any prior knowledge needed. In the human T cell case, MultiK identified a total of 12 reproducible T cell subsets spanning 6 different data sets that represent multiple cancer types. Moreover, consistent with previous findings, some of these reproducible T cell signatures showed prognostic values in predicting breast cancer patient’s survival including Treg, and multiple CD8 subsets. In particular, we found that the CD4 T naïve signature was significantly associated with overall survival in multiple patient sets, including both HER2+ and TNBCs. We also found that the two CD4 T follicular helper subsets significantly correlated with survival in both HER2+ and TNBC samples. Furthermore, consistent with the previous finding that the CD8 Trm signature associated with good prognosis, we found that our CD8T resident memory signature was prognostic within the HER2+ and TNBC sets. Conclusion: MultiK improves current scRNAseq cluster/group number estimations using an objective data driven approach. This methodology should be important as using our T cell analyses as an example, it shows that many previous published analyses likely overestimated the true number of reproducible T cell subsets in the tumor immune microenvironment, which may lead to irreproducible findings across studies. Additional analyses on tumor cell subsets are also currently underway.

Citation Format: Siyao Liu, Aatish Thennavan, J.S. Marron, Charles Perou. An automated tool to determine optimal cluster numbers in single-cell RNA sequencing data identifies key prognostic subsets of T cells in breast tumors [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-09-08.

Evaluating the efficacy of a priming dose of cyclophosphamide prior to pembrolizumab to treat metastatic triple negative breast cancer

PURPOSE

Triple negative breast cancer (TNBC) is characterized by the presence of immune cells in the tumor microenvironment, however, the response to single-agent immune checkpoint inhibitor (ICI) therapy is modest. Preclinical models have demonstrated that intratumoral regulatory T cells (T_regs) dampen the antitumor response to ICI. We performed a single-arm phase II trial to evaluate the efficacy of a single low dose of cyclophosphamide (Cy) to deplete T_regs administered before initiating pembrolizumab.

PATIENTS AND METHODS

40 patients with pretreated metastatic TNBC were enrolled. The primary endpoints were progression-free survival (PFS) and change in peripheral blood T_regs after Cy. Secondary endpoints included overall response rate (ORR), duration of response, overall survival, treatment-related adverse events (AEs), and correlative evaluations.

RESULTS

Median PFS was 1.8 months, and the ORR was 21%. T_regs were not significantly decreased after Cy prior to ICI (-3.3%, p=0.19), and increased significantly after the first cycle of therapy (+21% between cycles 1 and 2, p=0.005). Immune-related AEs were similar to historical pembrolizumab monotherapy, and were associated with response to therapy (p=0.02). Patients with pretreatment tumors harboring increased expression of B cell metagene signatures and increased circulating B cell receptor repertoire diversity were associated with clinical response and immune-related toxicity (IRT).

CONCLUSIONS

Among patients with heavily pretreated TNBC, Cy prior to pembrolizumab did not significantly deplete T_regs, and in those with decreased numbers there was rapid recovery following therapy. Increased B cell gene expression in baseline samples was associated with clinical response and IRT.

Abstract SP116: B-cells and follicular T cells regulate responses to immune checkpoint inhibitors in breast tumors and melanomas

Triple Negative Breast Cancer (TNBC) makes up 10-15% of breast cancers, disproportionally affects African Americans, and if considered its own disease type, would rank as the #5 cause of cancer deaths in women in the USA each year. Among the most promising new therapeutic approaches for TNBC is immunotherapy using Immune Checkpoint Inhibitors (ICI). Many ICI trials for TNBC patients are underway with early results showing response rates to single agent ICI in the range of 10-20% in the metastatic setting. Thus, there is a good sign of efficacy, but the majority of TNBC patients still fail to respond, even despite the recent improvements made by combination regimens of ICI and chemotherapy. These findings illustrate a clear need for 1) improved biomarker(s) to identify the subset TNBC patients who will respond to current ICI-based regimens, and 2) improvements in ICI regimens that may include additional immune modulating therapeutics, and/or combinations with specific mechanistically chosen chemotherapeutics.

Improving response rates of breast cancer patients to immune checkpoint inhibitors requires an understanding of mechanism of action of ICI, and robust in vivo models to further evaluate mechanisms and to empirically test new regimens in the preclinical setting first. In our prior work, we identified and credentialed mouse models with genomic and transcriptomic similarities to human TNBC, and in a subtype aware manner. In completing these studies, we found that the majority of TNBC mouse models reflect those human tumors with low tumor mutation burden (TMB), and these models had low immune infiltrates. In our efforts to model human TNBCs with elevated tumor mutation burdens we “mutagenized” our low TMB models into the range of high TMB human TNBC. We found that these mouse models with high tumor mutation burden had increased tumor infiltrating lymphocytes and were now responsive to anti-PD1 and anti-CTLA4 immune checkpoint inhibitors. Specifically, we discovered that in ICI sensitive mouse models, CD4+ T follicular helper cells and B cells coactivate one another and both are necessary for the full effects of the anti-tumor response to immune checkpoint therapy. We further used additional genetically engineered mouse models to discover a requirement for active antibody production as being needed for an effective anti-tumor response to immune checkpoint therapy.

The direct and active involvement of B cells in response to ICI has recently been shown to also occur in melanoma patients. This suggests a broader role for B cells as being a key component of a coordinated anti-tumor immune response. In addition, we captured these mechanistic features within a small gene expression signature that we further demonstrated predicted response to ICI treatment in melanoma patients, to trastuzumab in HER2+ breast cancer patients, and to neoadjuvant chemotherapy in TNBC patients. To translate this knowledge to therapeutic advances, we are now using our mouse models to test if immunomodulators of B cells can improve T cell focused immunotherapies, some of which are showing promising results. In summary, the adaptive immune system is a critical component of response to multiple and diverse classes of drugs in breast cancer, including chemotherapy, HER2 targeting therapy, and ICI. Further studies aimed at boosting these responses could have a significant impact for many TNBC patients, and non-TNBC patients, as well.

Citation Format: C Perou. B-cells and follicular T cells regulate responses to immune checkpoint inhibitors in breast tumors and melanomas [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr SP116.

TBCRC 030: a phase II study of preoperative cisplatin versus paclitaxel in triple-negative breast cancer: evaluating the homologous recombination deficiency (HRD) biomarker

BACKGROUND

Cisplatin and paclitaxel are active in triple-negative breast cancer (TNBC). Despite different mechanisms of action, effective predictive biomarkers to preferentially inform drug selection have not been identified. The homologous recombination deficiency (HRD) assay (Myriad Genetics, Inc.) detects impaired double-strand DNA break repair and may identify patients with BRCA1/2-proficient tumors that are sensitive to DNA-targeting therapy. The primary objective of TBCRC 030 was to detect an association of HRD with pathologic response [residual cancer burden (RCB)-0/1] to single-agent cisplatin or paclitaxel.

PATIENTS AND METHODS

This prospective phase II study enrolled patients with germline BRCA1/2 wild-type/unknown stage I-III TNBC in a 12-week randomized study of preoperative cisplatin or paclitaxel. The HRD assay was carried out on baseline tissue; positive HRD was defined as a score ≥33. Crossover to an alternative chemotherapy was offered if there was inadequate response.

RESULTS

One hundred and thirty-nine patients were evaluable for response, including 88 (63.3%) who had surgery at 12 weeks and 51 (36.7%) who crossed over to an alternative provider-selected preoperative chemotherapy regimen due to inadequate clinical response. HRD results were available for 104 tumors (74.8%) and 74 (71.1%) were HRD positive. The RCB-0/1 rate was 26.4% with cisplatin and 22.3% with paclitaxel. No significant association was observed between HRD score and RCB response to either cisplatin [odds ratio (OR) for RCB-0/1 if HRD positive 2.22 (95% CI: 0.39-23.68)] or paclitaxel [OR for RCB-0/1 if HRD positive 0.90 (95% CI: 0.19-4.95)]. There was no evidence of an interaction between HRD and pathologic response to chemotherapy.

CONCLUSIONS

In this prospective preoperative trial in TNBC, HRD was not predictive of pathologic response. Tumors were similarly responsive to preoperative paclitaxel or cisplatin chemotherapy.

Abstract GS2-05: Microscaled proteogenomic methods for precision oncology

Cancer proteogenomics combines genomics, transcriptomics and mass spectrometry-based proteomics to gain insights into cancer biology and treatment responsiveness. While proteogenomics analyses have already shown great potential to deepen our understanding of cancer tissue complexity and signaling, how a patient’s tumor changes upon treatment has largely been the province of genomics. This is due to technical difficulties associated with doing proteogenomic analysis on clinic-derived core-needle biopsies. To address this critical need, we have developed a “microscaled” proteogenomics approach for tumor-rich OCT-embedded core needle biopsies. Tissue-sparing specimen processing (“Biopsy Trifecta EXTraction”, BioTExt) and microscaled proteomics (MiProt) methodologies allowed generation of deep-scale proteogenomics datasets, with copy number and transcript information for &gt;20,000 genes and mass spectrometry-based identification and quantification of nearly all expressed proteins in a tumor (&gt;10,000 proteins) and more than &gt;20,000 phosphosites starting with just 25 micrograms of protein per sample. In order to understand the capabilities and limitations our our approach relative to more conventional deepscale proteomics requiring &gt;10X more starting material, we compared preclinical patient derived xenograft (PDX) models at conventional scale with data obtained by core-needle biopsy of the same tissues. Comparable depth and biological insights were obtained from the cores relative to surgically resected tumors. As a proof-of-concept for implementation in clinical trials, we applied microscaled proteogenomic methods to a small-scale clinical study where biopsies were accrued from patients with ERBB2+ locally advanced breast cancer before and 48 to 72 hours after the first dose of neoadjuvant Trastuzumab-based chemotherapy. Multi-omics comparisons were conducted between samples associated with residual disease versus samples associated with complete pathological response. Integrative, microscaled proteogenomic analyses efficiently diagnosed the molecular bases of diverse candidate treatment resistance mechanisms including: 1) absence of ERBB2 amplification (false-ERBB2+); 2) insufficient ERBB2 activity for therapeutic sensitivity despite ERBB2 amplification (pseudo-ERBB2+); 3) resistance features in true-ERBB2+ cases including androgen receptor signaling, mucin expression and an inactive immune microenvironment; 4) lack of acute phospho-ERBB2 down-regulation in non-pCR cases. In summary, we have developed a robust proteogenomics pipeline well suited for large-scale cancer clinical studies to identify potential resistance mechanism in patients. We conclude that microscaled cancer proteogenomics could improve diagnostic precision in the clinical setting.

Citation Format: Shankha Satpathy, Eric Jaehnig, Krug Karsten, Beom-Jun Kim, Alexander Saltzman, Doug Chan, Kimberly Holloway, Meenakshi Anurag, Chen Huang, Purba Singh, Ari Gao, Noel Namai, Yongchao Dou, Bo Wen, Suhas Vasaikar, David Mutch, Mark Watson, Cynthia Ma, Foluso Ademuyiwa, Mothaffar Rimawi, Jeremy Hoog, Samuel Jacobs, Anna Malovannaya, Terry Hyslop, D.R Mani, Charles Perou, George Miles, Bing Zhang, Michael Gillette, Steven Carr, Matthew Ellis. Microscaled proteogenomic methods for precision oncology [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr GS2-05.

THER-06. GENOMIC AND IMMUNE CHARACTERIZATION OF TRIPLE NEGATIVE BREAST CANCER BRAIN METASTASES

INTRODUCTION: Approximately 50% of patients with metastatic triple negative breast cancer (TNBC) will develop brain metastases (BM). Routinely treated with radiotherapy and/or surgery, survival is generally less than one year. There are no approved systemic therapies to treat TNBC BM. We characterized the genomic and immune landscape of TNBC BM to foster the development of effective brain permeable anti-cancer agents, including immunotherapy. EXPERIMENTAL PROCEDURES: A clinically-annotated BCBM biobank of archival tissues was created under IRB approval. DNA (tumor/normal) and RNA (tumor) were extracted from TNBC primaries and BM; whole exome (WES) and RNA sequencing (RNASeq) was performed. Mutations were determined from WES as those co-identified by two variant callers (Strelka|Cadabra). Immune gene signature expression, molecular subtype identification, and T cell receptor repertoires were inferred from RNAseq. RESULTS: 32 TNBC patient tissues (14 primaries, 18 BCBM, 6 primary-BCBM matched), characterized as basal-like by PAM50, were analyzed. Top exome mutation calls included ten genes in ≥19% of BCBMs including TP53, ATM, and PIK3R1, and four genes in ≥18% of primaries including TP53 and PIK3R1. Many immune gene signatures were lower in BM compared to primaries including B cell, dendritic cell, regulatory T cell, and IgG cluster (p< 0.05). A signature of PD-1 inhibition responsiveness was higher in BM compared with primaries (p< 0.05). BCBM T cell receptor repertoires showed higher evenness and lower read count (both p < 0.01) compared to primaries. CONCLUSIONS: TNBC BM compared to primaries that metastasize to the brain show lower immune gene signature expression, higher PD-1 inhibition response signature expression, and T cell receptor repertoire features less characteristic of an active antigen-specific response. Mutations common to TNBC BM and primaries include TP53 and PIK3R1. Given that non-BCBM (i.e. lung and melanoma) show response to checkpoint inhibitors, these findings collectively support further study of immunotherapy for TNBC BM.

Abstract 4563: Profiling the tumor immune microenvironment of adenocarcinoma and squamous cell NSCLC

The emergence of immune checkpoint inhibitors has changed the treatment paradigm for NSCLC. Recently, immunotherapy in combination with chemotherapy was shown to be a promising first-line treatment strategy for patients with squamous cell lung cancer, where targeted therapy has had little efficacy due to the pathological driver mutations for this disease being poorly understood. However, for both adenocarcinoma and squamous cell NSCLC, use of such immune checkpoint inhibitors has shown to improve the outcome for only a subset of patients with advanced or metastatic disease. Understanding the diversity in the tumor immune microenvironment will be critical to not only advance immune oncology research but to also determine potential clinical distinctions that dictate therapeutic response. Here, we take a multifaceted approach to compare the tumor immune microenvironments of over 50 patients diagnosed with either adenocarcinoma or squamous cell NSCLC. To this end, we performed a comprehensive multicolor flow cytometry analysis to compare tumor immune infiltrates and expression of suppressive markers to adjacent normal lung tissue as well as peripheral blood immune cells from the same patient. This analysis showed differences in the CD4+ and CD8+ T cell, CD19+ B cell, NK cell, and CD14+/CD16+ myeloid cell populations between the tumor and matched tissue and blood. These immune profiles were then compared between adenocarcinoma and squamous cell carcinoma subtypes as well as early versus late stage disease. We then integrated these findings with multiplexed immunofluorescence analysis on a panel of immune cell markers. Finally, 10x genomics single-cell RNA sequencing (scRNAseq) data were used to analyze single-cell transcriptomes from both adenocarcinoma and squamous cell lung tumors. Our result showed great heterogeneity of immune infiltrates across different patient samples, including different subtypes and different disease stages. This detailed comparative analysis will help us further understand the involvement of tumor infiltrating immune cells for different subtypes of lung cancer.

Citation Format: Kayla Harmeyer, Jiehui Deng, Suhagi Shah, Ece Bagdatlioglu, Aarif Ahsan, Christopher Reid, Aatish Thennavan, Charles Perou, Kwok-Kin Wong, Harvey Pass. Profiling the tumor immune microenvironment of adenocarcinoma and squamous cell NSCLC [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 4563.

Abstract P3-07-05: Not presented

This abstract was not presented at the conference.

Citation Format: Hardeman A, Grushko T, Clayton W, Hurley I, Khramtsova G, Parker J, Perou C, Olopade O. Not presented [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-07-05.

Abstract B185: Epithelial cancer cell-expressed genes contribute to clinically relevant immune-based classifications of breast cancer

Introduction: With the development of immunotherapies for breast cancer therapy, reliable methods to evaluate the extent and type of immune involvement present in tumors and investigation of its effect on patient prognosis and treatment are needed. Identifying tumor-specific features that affect immune involvement will be a key to understand tumor-immune involvement. Therefore, we evaluated expression of immune-related mRNA signatures in TCGA breast cancer data to identify distinct immune-related tumor subsets and asociated prognostic values. We also evaluated immune cell features, including B cell and T-cell receptor richness and diversity, as well as epithelial tumor cell-specific features, including somatic mutations, copy number alterations and differential RNA expression between identified groups. Methods: More than 130 published immune related gene signatures were evaluated in 1095 breast tumors and 97 normal mammary samples. Groups were identified by consensus based hierarchical clustering of the immune signatures, using the proportion of ambiguous clustering to select the optimal number of groups. An ElasticNet model trained on TCGA data was applied to two other breast tumor datasets to predict immune group classification. RNA-sequencing (RNAseq) data from 70 breast cancer cell lines and from human tumor xenografts passaged in immune-compromised mice and processed through a human specific sequencing pipeline provided in vitro and in vivo sources of epithelial cancer cell expression with limited stromal content that was used to filter TCGA bulk RNAseq data for epithelial expressed genes. Results: We identified three distinct immune groups present in breast cancer: immune-low, immune-normal, and immune-high. The immune-high group is characterized by high T-cell scores, including both cytotoxic and regulatory T-cell signatures, and increased B cell and macrophage signatures. The immune-normal set shows signs of normal epithelia and low proliferation. The immune-low group has very low immune cell signatures. Intrinsic breast cancer subtypes (Basal, luminal A, Luminal B, Her2 and Normal-like) are present in each of the immune groups; however, enrichment of basal tumors in immune-high, luminal tumors in the immune-low, and normal mammary, normal-like tumors and luminal A tumors in the immune-normal group demonstrate an interaction between intrinsic tumor type and immune involvement. Immune groups are prognostic in TCGA, with the immune-high group having improved recurrence-free survival. Two more breast tumor datasets confirmed improved survival for basal tumors in the immune-high group relative to the immune-low tumors. Total mutation burden, unique somatic mutations, and copy number alterations did not show significant changes between immune-low and –high groups, whereas RNA expression differs between groups. Selecting for genes with evidence of expression by epithelial breast cancer cells identified over 8,000 genes differentially expressed between the immune-low and –high groups, with CCL5, ACAP1, PVRIG, SLA2, LCK and CD8A being among the most significant. Conclusion: Breast cancer can be divided into three clinically relevant immune-related groups. Immune-high has high immune involvement, showing of markers for cytotoxicity and immune suppression and exhaustion. These patients have improved survival, but may still benefit from immune checkpoint inhibition. Immune-normal is reflective of a normal mammary immune state, suggesting a microenvironment that has not been strongly altered by the tumor. Immune-low appears to demonstrate exclusion of immune cells from the tumor, even though tumors contain predicted neoantigens. These patients have poor survival and novel therapeutic strategies to activate immune involvement need to be developed. Epithelial cancer cell expressed many immune-related genes, including CCL5, LCK and CD8A, which may be critical determinants of immune cell attraction.

Citation Format: Jonathan Shepherd, Charles Perou. Epithelial cancer cell-expressed genes contribute to clinically relevant immune-based classifications of breast cancer [abstract]. In: Proceedings of the Fourth CRI-CIMT-EATI-AACR International Cancer Immunotherapy Conference: Translating Science into Survival; Sept 30-Oct 3, 2018; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2019;7(2 Suppl):Abstract nr B185.

Abstract B18: ADSL controls pyrimidine metabolism and triple-negative breast tumorigenesis

Triple-negative breast cancer (TNBC) represents 15% of all breast carcinomas. The unfavorable prognosis and aggressive biology of TNBC highlight the need to find new targeted therapies. Previous study from our lab has shown that the prolyl hydroxylase EglN2 contributes to TNBC progression, with its depletion leading to a decrease in TNBC cell invasion. However, the underlying mechanism remains elusive. In our current study, we developed a novel enzyme-substrate trapping strategy followed by mass spectrometry and identified adenylosuccinate lyase (ADSL) as a potential EglN2 hydroxylase substrate. EglN2 binds directly with ADSL and catalyzes ADSL hydroxylation on proline residues, which might affect ADSL activity. We demonstrated that ADSL is specifically upregulated in TNBC cell lines and patients. Functionally, depletion of ADSL led to decreased anchorage-independent growth, TNBC cell invasion, and orthotopic tumor growth. Mechanistically, integrated analyses of metabolomics and RNA-seq showed that ADSL controls pyrimidine biosynthesis that is essential for TNBC cell proliferation and invasion. Therefore, our study suggests that EglN2-ADSL signaling axis can serve as a novel therapeutic avenue in TNBC by regulating pyrimidine metabolism.

Citation Format: Giada Zurlo, Jeremy Simon, Cheng Fan, Adam Robinson, Javier Rodriguez Martinez, Alex Kriegsheim, Jason Locasale, Charles Perou, Qing Zhang. ADSL controls pyrimidine metabolism and triple-negative breast tumorigenesis [abstract]. In: Proceedings of the AACR Special Conference: Advances in Breast Cancer Research; 2017 Oct 7-10; Hollywood, CA. Philadelphia (PA): AACR; Mol Cancer Res 2018;16(8_Suppl):Abstract nr B18.

Abstract GS2-01: Discovery and characterization of an estrogen bound LncRNA in late-Stage breast cancer

Breast cancer is the second most common newly diagnosed cancer and the second leading cause of cancer death among women in the United States. Despite the proven benefits of adjuvant endocrine therapy in women with hormone receptor positive breast cancer, relapses still occur over 5 years after initial treatment with endocrine therapy, referred to as late-stage relapse. Currently, the mechanisms of driving late-stage relapse are poorly characterized. To date, breast cancer research has primarily focused on protein-coding genes thereby missing the emerging class of long intergenic non-coding RNAs (lncRNAs) that may serve as critical regulators of relapse. Furthermore, the current understanding of lncRNA function is still in its infancy representing a critical gap for translating lncRNA discoveries into real-world applications to benefit patient care. Several well-described examples indicate that lncRNAs may be master epigenetic regulators in cancer biology through their interactions with proteins regulating target gene expression. Therefore, we hypothesize that lncRNAs may interact with estrogen receptor alpha 1 (ESR1) to regulate genes promoting late-stage relapse. To address this, we performed a transcriptome analysis of tumors from a unique cohort of 24 patients that had late-stage relapse to discover a novel set of lncRNAs, most of which have not yet been characterized. Next, we used RNA Immunoprecipitation coupled with transcriptome sequencing (RIP-Seq) to identify transcripts bound to ESR1 in T47D cells. We discovered 217 lncRNAs bound to ESR1 of which 50 were up-regulated in late-stage breast cancer. We chose to focus the most up-regulated lncRNA in late-stage relapse. Since it is an unannotated lncRNA we will refer to it as 'LAte-Stage relapse ESR1-Bound lncRNA 1', or LASER-1. To further understand the interplay between LASER-1 and ESR1, cells endogenously expressing LASER-1 were subjected to partial digestion to preserve lncRNA and protein interactions. Protected RNA fragments were subsequently immunoprecipitated with ESR1 and quantified by qPCR to reveal specific ESR1 interaction sites within LASER-1. Next, we observed increased expression of LASER-1 in ER+ breast cancer cell lines. Notably, LASER-1 expression was elevated in MCF7 long-term estrogen deprived (MCF7 LTED) cells -- that have amplified ESR1 -- relative to parental MCF7 cells. To demonstrate that LASER-1 promotes oncogenic phenotypes we transiently silenced LASER-1 in two cell lines with high endogenous expression of LASER-1 (including MCF LTED) and observed a decrease in cellular proliferation and invasion. Subsequent gene expression analysis after silencing LASER-1 altered mRNA and protein levels of critical cell cycle genes (i.e., p27). Overall, this is the first study to discover ESR1 bound lncRNAs that may be contributing to late-stage relapse in breast cancer. In the short-term, our ongoing research may lead to significant breakthroughs establishing the importance of LASER-1 as a master regulator in late-stage relapse. In the longer-term, we envision this research may lead to the development of novel therapeutics targeting LASER-1 with the potential for rapid clinical translation.

Citation Format: Maher CA, Silva-Fisher J, Eteleeb A, Tang C, Perou C, Reis-Filho JS, Mardis ER, Ellis MJ. Discovery and characterization of an estrogen bound LncRNA in late-Stage breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS2-01.

Intrinsic subtypes and benefit from postmastectomy radiotherapy in node-positive premenopausal breast cancer patients who received adjuvant chemotherapy - results from two independent randomized trials

BACKGROUND

The study of the intrinsic molecular subtypes of breast cancer has revealed differences among them in terms of prognosis and response to chemotherapy and endocrine therapy. However, the ability of intrinsic subtypes to predict benefit from adjuvant radiotherapy has only been examined in few studies.

METHODS

Gene expression-based intrinsic subtyping was performed in 228 breast tumors collected from two independent post-mastectomy clinical trials (British Columbia and the Danish Breast Cancer Cooperative Group 82b trials), where pre-menopausal patients with node-positive disease were randomized to adjuvant radiotherapy or not. All patients received adjuvant chemotherapy and a subgroup of patients underwent ovarian ablation. Tumors were classified into intrinsic subtypes: Luminal A, Luminal B, HER2-enriched, Basal-like and Normal-like using the research-based PAM50 classifier.

RESULTS

In the British Columbia study, patients treated with radiation had an overall significant lower incidence of locoregional recurrence compared to the controls. For Luminal A tumors the risk of loco-regional recurrence was low and was further lowered by adjuvant radiation. These findings were validated in the DBCG 82b study. The individual data from the two cohorts were merged, the hazard ratio (HR) for loco-regional recurrence associated with giving radiation was 0.34 (0.19 to 0.61) overall and 0.12 (0.03 to 0.52) for Luminal A tumors.

CONCLUSIONS

In both postmastectomy trials, patients with Luminal A tumors turned out to have a significant lower incidence of loco-regional recurrence when randomized to adjuvant radiotherapy, leaving no indication to omit postmastectomy adjuvant radiation in pre-menopausal high-risk patients with Luminal A tumors. It was not possible to evaluate the effect of radiotherapy among the other subtypes because of limited sample sizes.

Abstract 1033: Estrogen receptor gene fusions drive endocrine therapy resistance in estrogen receptor positive breast cancer

Dysregulation of estrogen receptor gene (ESR1) is an established mechanism of inducing endocrine therapy resistance. We previously discovered a chromosomal translocation event generating an estrogen receptor gene fused in-frame to C-terminal sequences of YAP1 (ESR1-YAP1) that contributed to endocrine therapy resistance in estrogen receptor positive (ER+) breast cancer models. This current study compares functional and pharmacological properties of additional ESR1 gene fusion events of both early stage (ESR1-NOP2) and advanced endocrine therapy resistant (ESR1-YAP1 and ESR1-PCDH11x) breast cancers. The YAP1 and PCDH11x fusions conferred estrogen-independent and fulvestrant-resistant growth in T47D, an ER+ breast cancer cell line in vitro and in vivo, in contrast to the NOP2 fusion which was sensitive to hormone deprivation. Immunohistochemical (IHC) staining of mouse lungs revealed significantly higher numbers of micrometastatic ER+ cells from the T47D tumors expressing the YAP1 and PCDH11x fusions than YFP control and NOP2 fusion. Estrogen response element (ERE) reporter and pull down assays revealed that although all ESR1 fusions studied bound EREs, only the YAP1 and PCDH11x caused ERE activation. Cell lines containing these “canonical” ESR1 fusions upregulated expression of ER responsive genes such as TFF1 and GREB1 in hormone deprived conditions. In contrast, the NOP2 fusion neither induced ERE activity nor upregulated TFF1 and GREB1 gene expression. The proliferative ability of canonical fusion-containing T47D cells was inhibited by palbociclib, a CDK4/6 inhibitor, in a dose-dependent manner. In vivo growth of patient-derived xenograft tumors naturally harboring the ESR1-YAP1 fusion (WHIM18) was significantly reduced in mice fed palbociclib-containing chow. Mice transplanted with WHIM18 also formed lung micrometastases, with an ER IHC staining pattern similar to lungs from YAP1 and PCDH11x fusion expressing T47D xenografts. In conclusion, in-frame ERE activating canonical fusions occur in end-stage, drug resistant, advanced breast cancer and can be added to ESR1 point mutations as a class of somatic mutation that may cause acquired resistance. Endocrine therapy resistant growth induced by these fusions can be treated with CDK4/6 inhibition, using an FDA approved drug, palbociclib, which could potentially improve outcomes in patients with ESR1 translocated tumors.

Citation Format: Jonathan T. Lei, Jieya Shao, Jin Zhang, Michael Iglesia, Doug W. Chan, Ryoichi Matsunuma, Xiaping He, Purba Singh, Yoshimasa Kosaka, Robert Crowder, Svasti Haricharan, Shyam Kavuri, Jeremy Hoog, Chanpheng Phommaly, Rodrigo Goncalves, Susana Romalho, Wei-Chu Lai, Oliver Hampton, Anna Rogers, Ethan Tobias, Poojan Parikh, Sherri Davies, Cynthia Ma, Vera Suman, Kelly Hunt, Mark Watson, Katherine A. Hoadley, Aubrey Thompson, Charles Perou, Chad J. Creighton, Chris Maher, Matthew J. Ellis. Estrogen receptor gene fusions drive endocrine therapy resistance in estrogen receptor positive breast cancer [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 1033. doi:10.1158/1538-7445.AM2017-1033

Abstract 4970: Oxidative phosphorylation as a target in triple negative breast cancer therapy

Altered cellular metabolism is a hallmark of cancer. It is increasingly recognized that selected tumors are dependent on oxidative phosphorylation (OXPHOS). However, the role of OXPHOS in TNBC is not well understood. We performed RNA sequencing in pre-treatment biopsies from 43 patients with operable triple negative breast cancer (TNBC) who received sequential taxane- and anthracycline-based neoadjuvant chemotherapy. At a median follow-up of 63 months, 14 patients recurred and 12 patients died. At a false discovery rate of 0.05, 33 genes were differentially expressed between the patients who did and did not have a subsequent recurrence. Ingenuity pathway analysis demonstrated that one of the top canonical pathways that differed was higher expression of oxidative phosphorylation signature (p=5.89E-0.7). The patients who recurred had significantly higher levels of mitochondrial genes: MT-ND1 (adjusted p or FDR-BH; q=0.007); MT-ND5 (q=0.03) and MT-ND4 (q=0.04). Further, 21 genes were differentially expressed between patients based on survival, including MT-ND5 (q=0.001); MT-ND4 (q=0.005), MT-ND4L (q=0.015), MT-ND6 (q=0.018), and MT-ATP6 (P=0.03). Top canonical pathway that was differentially expressed based on survival was oxidative phosphorylation (p=9.98E-10). We therefore sought to determine the efficacy of IACS-10759, a novel inhibitor of OXPHOS, in 10 different TNBC patient-derived xenografts representing different gene expression based Lehmann TNBC subtypes. Growth inhibition was observed in multiple subtypes, with regression in one basal-like 1(BL1) 1 model, and stabilization of growth in multiple BL1 and immunomodulatory expression subtypes. Taken together, our data suggests that high OXPHOS is associated with higher recurrence and lower survival. OXPHOS is a promising target in several TNBC subtypes. A Phase I trial of IACS-10759, a potent inhibitor of complex I of oxidative phosphorylation, in leukemia is ongoing and planned in TNBC and other solid tumors.

Citation Format: Funda Meric-Bernstam, Kurt Evans, Xiaofeng Zheng, Xiaoping Su, Erkan Yuca, Stephen Scott, Argun Akcakanat, Naoto Ueno, Bora Lim, Jennifer Litton, Vicente Valero, Fraser Symmans, Gabriel Hortobagyi, Charles Perou, Debu Tripathy, Guilio Draetta, Joe Marszalek, Ana Maria Gonzalez-Angulo, Stacy Moulder. Oxidative phosphorylation as a target in triple negative breast cancer therapy [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 4970. doi:10.1158/1538-7445.AM2017-4970

Pentraxin-3 is a PI3K signaling target that promotes stem cell-like traits in basal-like breast cancers

Basal-like breast cancers (BLBCs) exhibit hyperactivation of the phosphoinositide 3-kinase (PI3K) signaling pathway because of the frequent mutational activation of the PIK3CA catalytic subunit and the genetic loss of its negative regulators PTEN (phosphatase and tensin homolog) and INPP4B (inositol polyphosphate-4-phosphatase type II). However, PI3K inhibitors have had limited clinical efficacy in BLBC management because of compensatory amplification of PI3K downstream signaling loops. Therefore, identification of critical PI3K mediators is paramount to the development of effective BLBC therapeutics. Using transcriptomic analysis of activated PIK3CA-expressing BLBC cells, we identified the gene encoding the humoral pattern recognition molecule pentraxin-3 (PTX3) as a critical target of oncogenic PI3K signaling. We found that PTX3 abundance is stimulated, in part, through AKT- and nuclear factor κB (NF-κB)-dependent pathways and that presence of PTX3 is necessary for PI3K-induced stem cell-like traits. We further showed that PTX3 expression is greater in tumor samples from patients with BLBC and that it is prognostic of poor patient survival. Our results thus reveal PTX3 as a newly identified PI3K-regulated biomarker and a potential therapeutic target in BLBC.

Abstract P6-04-05: Genotype-phenotype classification of triple negative breast cancers (TNBC) in women of African descent using the PAM50 NanoString platform and genomic data

Introduction: TNBC has the highest mortality rate amongst all other breast cancer types due to its complex tumor heterogeneity and lack of well-defined molecular targets. It is known that women of African descent are two to three times more likely to develop TNBC compared to women of European ancestry, yet wide-scale genomic studies of African and African American breast tumors are limited. To elucidate genotypes and molecular subtypes associated with the most aggressive forms of breast cancer, we used the PAM50 NanoString platform to reclassify Nigerian (NG), African American (AA) and Caucasian (CA) tumors previously annotated by Immunohistochemistry (IHC), and correlated our findings to their germline genotype data obtained using high-throughput technologies.

Methods: RNAs were isolated from formalin-fixed, paraffin embedded (FFPE) tumor tissues using the High Pure Paraffin Kit (Roche) following manufacturer's protocol, and assayed on NanoString nCounter Analysis System using a custom Nano110 (PAM50 + claudin-low & VEGF signatures) probe set. Intrinsic subtyping and gene-expression data were evaluated using R statistical software. All study samples were previously annotated and subtyped by the ER/PR/HER2 IHC classifier. Genotypes were obtained from next generation sequencing or Illumina Human2.5M BeadChip platform using germline DNA from more than 2000 breast cancer cases and 2000 controls were studied.

Results: To date, Intrinsic molecular subtyping by Nano110 has been completed on 69 NG, 81 AA and 74 CA tumors. Concordance between IHC and PAM50 was 59%, which is adequate and comparable to previous studies. Basal-like subtype was overrepresented and accounted for nearly 30% of NG and AA cases, compared to 17% in CA cases. HER2-enriched subtype was overrepresented only in NG cases (9%). The proportion with Luminal A tumors were 44% NG, 56% AA and 68% CA, respectively.

Conclusions: PAM50 NanoString assay is reliable and high-throughput for molecular subtyping breast cancer using RNA extracted from FFPE tumors. Ongoing work will correlate PAM50 intrinsic subtypes to genotype data.

Citation Format: Olayiwola OA, Ogundiran TO, Hardeman A, Yoshimatsu TF, Clayton W, Adeoye A, Ademola A, Ajani MA, Khramtsova G, Grushko TA, Huo D, Zheng Y, Parker J, Perou C, Olopade OI. Genotype-phenotype classification of triple negative breast cancers (TNBC) in women of African descent using the PAM50 NanoString platform and genomic data. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-04-05.

Abstract A24: Characterization of a novel p53; BRCA1-deficient claudin-low mammary tumor cell line

Triple-negative breast cancer (TNBC) is a highly diverse and typically aggressive group of cancers. Consequently, elucidating the biology of the various subtypes is necessary for more effective diagnosis and better targeted treatment of this disease. Recently, a novel claudin-low TNBC subtype has been identified. Patients with this subtype have a very poor prognosis, specifically young African-American females. The major focus of our study is to characterize a novel claudin-low cell line cultured from a “humanized” mouse model that is p53 and BRCA1 deficient (K14TRT cell line).

Procedures: This cell line has been maintained for more than 40 months and for over 25 subcultures, and was characterized by microscopy, histopathology, genotyping, cytogenetics, and bioinformatics based techniques.

Results: Histological analysis supports that the primary K14TRT tumors were highly penetrant, with some having spindloid morphology while others were undifferentiated adenocarcinoams, with large areas of necrosis. As compared to normal mammary cells, K14TRT cells possessed larger variable shaped nuclei with a higher nuclear-to-cytoplasmic ratio. The tumor cells also demonstrated a highly aneuploidy karyotype. When isolated via single cell suspension and subsequently passaged, the K14TRT cells were able to form colonies in soft agar. Tumorigenicity was additionally supported by induction of solid tumors following serial injections of the K14TRT cells into the flanks of nude mice. The primary K14TRT tumor overwhelmingly possessed a basal-like breast tumor profile, whereas the transplanted and serially passaged K14TRT cells diverged to a claudin-low profile, as demonstrated by microarray analysis. These serially passaged cells were highly proliferative, with low to absent expression of tight junction proteins (claudin 3, 4, & 7 and E-cadherin), and typically negative for estrogen receptors (ER-), progesterone receptors (PR-), and HER2/neu (HER2-). Since this cell line did demonstrate an ability to form new tumors in recipient mice we speculated they could possess tumor stem cell properties. There competence for self-renewal and capacity to reestablish tumor heterogeneity was supported by fluorescence-activating cell sorting (FACS) analysis of the primary BRCA1/p53 deficient mammary tumor cells in which two populations were identified. Cells that possess low levels of epithelial marker CD24 and high levels of the B1 integrin marker CD29 have been shown to regenerate entire mammary glands indicative of mammary STEM cells. Strikingly, FACS analysis of the K14TRT primary cell line identified two cell populations, one which expressed markers that were CD24high/CD29high (differentiated) and the other population CD24low/CD29high (undifferentiated stem cell-like profile). The sorted undifferentiated mammosphere population when analyzed by microarray analysis clustered with transplanted and serially passaged K14TRT cells identified as being claudin-low, whereas the differentiated sorted population clustered with the primary basal-like cells.

Conclusions: Our current data has demonstrated that the K14TRT cell line is a good candidate for a “humanized” TNBC claudin-low cell line with stem cell properties, and could play a significant role in devising better diagnostic tools and chemotherapeutic strategies for this aggressive tumor subtype.

Citation Format: Stephanie T. Dance-Barnes, Ebone Evans, Shanderys Steward, Xiaping He, Jerry Usary, Charles Perou. Characterization of a novel p53; BRCA1-deficient claudin-low mammary tumor cell line. [abstract]. In: Proceedings of the Fourth AACR International Conference on Frontiers in Basic Cancer Research; 2015 Oct 23-26; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2016;76(3 Suppl):Abstract nr A24.

Abstract IA29: Proteogenomic and phosphoproteomic analysis of breast cancer

The genetic landscape of human breast cancer has been well defined in The Cancer Genome Atlas (TCGA) project. Mass spectrometry (MS)-based global proteome and phosphoproteome analyses provide a complementary, orthogonal approach to genomic studies to further improve the molecular taxonomy and biological understanding of breast cancer. We analyzed human breast cancer samples that had previously undergone comprehensive genomic and reversed phase protein array (RPPA) characterization by TCGA. Tumor samples were analyzed by global shotgun proteomics and phosphoproteomics at an unprecedented coverage of &gt;11,000 quantified proteins and &gt;27,000 phosphorylation sites for each tumor. We verified the translation of hundreds of genomically characterized single nucleotide and splice junction variants at the protein level. The correlation of mRNA to protein abundance was significant for 6,135 out of 9,302 protein/mRNA pairs, but differed amongst protein classes. Genes that did not correlate on the protein/mRNA level included components of basic cellular machineries such as the ribosome, RNA polymerase and spliceosome, as well as those involved in processes regulated by proteolysis. Hierarchical clustering yielded three major clusters in both the proteome and the phosphoproteome data: basal-enriched, luminal-enriched and stroma-enriched groups, the last also enriched for what have been previously designated “reactive-type” tumors by RPPA. Our deep proteome analysis promoted new insights including the consequences of chromosomal loss, such as the 5q deletion characteristic of basal-like breast cancer. The 5q trans effects were interrogated using the Library of Integrated Network-based Cellular Signatures. Theses analyses connected the 5q genes CETN3 and SKP1 to elevated expression of EGFR, and SKP1 also to SRC. Differential phosphopeptide analyses, integrated with activity maps derived from knock-in mutated cell lines, identified multiple novel downstream effects of PIK3CA and TP53 mutation. Besides ERBB2, other amplicon-associated, highly phosphorylated kinases were identified, including CDK12, PAK1, PTK2, RIPK2 and TLK2. These and other examples demonstrate that proteogenomic analysis of breast cancer elucidates functional consequences of somatic mutations, narrows candidate nominations for driver genes within large deletions and amplified regions, and identifies potential therapeutic targets.

Citation Format: Philipp Mertins, DR Mani, Kelly Ruggles, Michael Gillette, Karl Clauser, Pei Wang, Xianlong Wang, Jana Qiao, Song Cao, Francesca Petralia, Filip Mundt, Zhidong Tu, Jonathan Lei, Michael Gatza, Matthew Wilkerson, Charles Perou, Venkata Yellapantula, Kuan-lin Huang, Chenwei Lin, Michael McLellan, Ping Yan, Sherri Davies, Reid Townsend, Steven Skates, Jing Wang, Bing Zhang, Christopher Kinsinger, Mehdi Mesri, Henry Rodriguez, Li Ding, Amanda Paulovich, David Fenyo, Matthew Ellis, Steven Carr, NCI CPTAC. Proteogenomic and phosphoproteomic analysis of breast cancer. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Breast Cancer Research; Oct 17-20, 2015; Bellevue, WA. Philadelphia (PA): AACR; Mol Cancer Res 2016;14(2_Suppl):Abstract nr IA29.

Abstract 2285: Germline genetic variation modulates tumor latency and response to therapy in a mouse model of basal-like breast cancer

An important outcome from gene expression profiling studies of human cancers is the identification of a molecular taxonomy of breast cancer, and in particular, description of the basal-like breast cancer (BBC) subtype. BBC is an aggressive subtype with frequent and early relapse and poor survival. Unfortunately, the mechanisms underlying the poor prognosis associated with BBC are unclear. The over-arching goal of this project was to employ mouse strains generated through the Collaborative Cross (CC), a large panel of recombinant inbred mice strains which can be converted to ‘outbred strains’ with reproducible genomes through the generation of recombinant inbred intercrosses (RIX). Female C3(1)/SV40 T antigen transgenic mice (C3-TAgs; FVB background), a model determined by genomic analysis to reflect the BBC subtype, were bred to recombinant inbred CC males, creating a cohort of 500+ F1 experimental mice in 20 RIX lines to examine various phenotypic and genotypic changes. Mice were weighed weekly and were palpated twice weekly to determine tumor latency. Tumor-bearing mice were observed for tumor progression ([21 day tumor volume - initial volume/initial volume] x 100) and multiplicity (number of tumors at 21 days). Subsequent cohorts of selected RIX lines were treated with trametinib (MEK1/2 inhibitor), BKM120 (panPI3K inhibitor) or everolimus (mTOR inhibitor), and after three weeks of treatment, the percent change in tumor volume was used to assess the objective response rate of the therapies. The average tumor latency of all RIX mice was 22.8 weeks; however 4 RIX lines (1566, 2014, 5155 and 5156) had significantly longer latencies (average 34.1-44.5 wks), while 3 RIX lines (1515, 5066, and 3067) were observed to develop palpable tumors significantly earlier. Line 3067 also had significantly increased body mass relative to parental C3Tag mice. Two lines (559 and 3140) were noted to have increased tumor multiplicity. Efficacy trials showed all of the RIX lines (relative to parental C3TAgs), demonstrated increased resistance to each chemotherapeutic agent studied, with three exceptions: line 559 was sensitive to trametinib; line 1515 was sensitive to BKM120; and line 8005 was sensitive to everolimus. These findings suggest the introduction of germline variation (by breeding C3TAg mice with CC lines) modulates host characteristics as well as BBC development and therapeutic responses. Genetic and genomic analyses (including sequencing of candidate gene regions and microarray analyses) are underway to identify dominant genetic modifiers and associated molecular targets underlying the observed differential phenotypes. We conclude that these RIX strains represent useful model systems for identifying genetic modifiers and host characteristics that influence the development and/or chemotherapeutic response of BBC.

Citation Format: David Darr, Norman Sharpless, Charles Perou, Fernando Pardo-Manuel de Villena, Darla Miller, Katharine Bendt, Stephen D. Hursting. Germline genetic variation modulates tumor latency and response to therapy in a mouse model of basal-like breast cancer. [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 2285. doi:10.1158/1538-7445.AM2015-2285

Abstract 5024: mDX400, the murine analog against the anti-PD1 antibody MK-347 is active in immunocompetent, autochthonous murine models of melanoma and breast cancer

Although effective therapies exist for BRAF-mutant metastatic melanomas (MM) and ER+/PR+/HER2+ breast cancers, fewer options are available for the more aggressive triple negative breast cancers (TNBC) and Ras-mutant MM. Immune infiltration is frequently observed in patient subsets with MM or TNBC. An anti-tumor host immune response may be restrained by the expression of immune checkpoint proteins, such as the programmed death 1 (PD1) protein. We evaluated mDX400, a murine version of the anti-PD-1 antibody MK-3475 that is currently in human clinical trials, in genetically engineered murine models (GEMMs) of melanoma and breast cancer (BC). PD1 antibody was used alone and in combination. Models studied were for basal-like, TNBC (C3TAg), Claudin-Low breast cancer p53null, luminal BC (MMTV-Her2/Neu), RAS-driven melanoma (Tyr-H-Ras and null for Ink4a/Arf (TRIA)) and BRaf-mutant, Pten-deficient melanoma (BRaf/Pten). Mice were housed, treated, and evaluated in the UNC Lineberger Comprehensive Cancer Center Mouse Phase I Unit (MP1U). mDX400 and isotype antibody (10mg/kg IP qw) were supplied by Merck & Co., Inc. Murine cohorts were assessed weekly and therapeutic intervention began once tumors reached 40-64mm3. TNBC models were treated with Carboplatin (C, 50mg/kg IP qwk) in combination with mDX400. mDX400 was administered as a single agent in Melanoma GEMMs. Endpoints were maximal response at ≥21 days and median overall survival (OS). In the C3TAg BC model, mDX400 did not exhibit single-agent activity, but substantial synergy was observed in combination with C [partial response (PR) or CR in 12 of 15 vs. 0 of 19 in non-treated animals, Fisher's exact p=0.006 or vs. C alone 5 of 13 animals, p=0.05] and prolonged OS (41 vs. 28 days in NT, log-rank p=0.006 or C only 28 days, p=0.006) in the C3TAg model. Treatment with single-agent mDX400 induced a marked response in the RAS-driven TRIA melanoma model that is resistant to multiple systemic treatments. mDX400-treated TRIA mice enjoyed an almost tripling of OS [median 56 vs. 21 days, p=0.006]. BRaf/Pten melanomas did not respond to mDX400. Likewise mDX400 plus C provided no benefit to the p53null or MMTV-Her2/Neu BC GEMMs vs. C alone. Our data show differential response of PD-1 antibody therapies to various melanoma and BC models, the latter in combination with C. The responsiveness of a RAS-driven melanoma model to mDX400 is consistent with earlier reports about the clinical benefit of patients with NRAS-mutant melanoma to Ipilimumab, and the lack of PD-L1 expression and immune infiltrates of PTEN-deficient melanomas. At the meeting we will report expression of PD-L1 and other immune checkpoint proteins by cancer cells. The basis whereby carboplatin potentiates the activity of anti-PD1 in TNBC is an area of ongoing study.

Citation Format: David Darr, Kelly S. Clark, Joseph H. Phillips, Elaine Pinheiro, Venkataraman Sriram, Jessie Xiong, Jamie Jordan, Norman E. Sharpless, Charles Perou, Stergios Moschos. mDX400, the murine analog against the anti-PD1 antibody MK-347 is active in immunocompetent, autochthonous murine models of melanoma and breast cancer. [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 5024. doi:10.1158/1538-7445.AM2014-5024

Abstract 360: Thousands of published cancer signatures and pathways can be collapsed into a handful of non-redundant gene programs: a TCGA pan-cancer analysis

Background: The OMIC revolution has produced tens of thousands of published gene expression signatures, pathways, and modules associated with cancer biology or patient outcome. Thanks to the TCGA effort to systematically collect molecular data from multiple cancer types, it is now possible to compare these signatures within and across cancers, and to collapse them into a smaller set of non-redundant features.

Methods: We scored 3602 TCGA samples representing 12 cancer types for expression of a compendium of 6,898 published gene expression prognostic signatures, co-expression modules, amplicons and pathways, using RNAseq expression profiles. 6477 of these features were from the database MsigDB, and the rest from the literature. To define non-redundant gene-programs, we applied a bimodality filter and used Weighted Gene Correlation Network Analysis (WGCNA) to aggregate the bimodal signatures and pathways into coherent clusters. Associations were assessed using standard statistical methods.

Results: From the thousands of cancer signatures in our analysis we identified 22 non-redundant gene-programs, defined as groups of gene signatures with high correlation across the Pan-Cancer dataset. These gene programs represent many of the hallmarks of cancer, including sustained proliferative signaling and DNA repair; immune system signaling; altered glycolytic and fatty acid metabolism yielding resistance to stress; apoptosis evasion; epithelial to mesenchymal transition; co-opted squamous and ‘stemness’ developmental programs; adhesion and cell-cell communication programs, some using plasma membrane and lipid vesicles; and self-sufficiency in growth signals employing estrogen, EGF, and MYC. More than a third of the original signatures clustered together in the proliferation gene-program. In addition to cell cycle and proliferation pathways, this program includes prognostic signatures for lymphoma, bladder, lung, breast, ovarian and other cancers, suggesting that most prognostic signatures are sensing a common proliferative signal associated with cancer aggression. The presence of two dominant immune signals showed a clear divergence in immune signaling between T cell/B cell immune activation and interferon responsive signaling, with the former correlated with immuno-modulatory treatment targets like CTLA4 and PD1. Interestingly, basal, hypoxic, and squamous differentiation programs were associated with poor outcome in many cancer types; in addition, expression of a more novel program dominated by tumor suppressing miRNA associated with good prognosis.

Conclusion: The compendium of gene expression signatures is correlated, and can be reduced to a small set of non-redundant gene-programs that mirror the cancer hallmarks and may provide opportunities for investigating cancer heterogeneity and treatment approaches beyond tissue of origin.

Citation Format: Denise M. Wolf, Cheng Fan, Katherine A. Hoadley, Christina Yau, Artem Sokolov, TCGA Network, Josh Stuart, Charles Perou, Laura van ’t Veer. Thousands of published cancer signatures and pathways can be collapsed into a handful of non-redundant gene programs: a TCGA pan-cancer analysis. [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 360. doi:10.1158/1538-7445.AM2014-360

Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis

BACKGROUND

Pathological complete response has been proposed as a surrogate endpoint for prediction of long-term clinical benefit, such as disease-free survival, event-free survival (EFS), and overall survival (OS). We had four key objectives: to establish the association between pathological complete response and EFS and OS, to establish the definition of pathological complete response that correlates best with long-term outcome, to identify the breast cancer subtypes in which pathological complete response is best correlated with long-term outcome, and to assess whether an increase in frequency of pathological complete response between treatment groups predicts improved EFS and OS.

METHODS

We searched PubMed, Embase, and Medline for clinical trials of neoadjuvant treatment of breast cancer. To be eligible, studies had to meet three inclusion criteria: include at least 200 patients with primary breast cancer treated with preoperative chemotherapy followed by surgery; have available data for pathological complete response, EFS, and OS; and have a median follow-up of at least 3 years. We compared the three most commonly used definitions of pathological complete response--ypT0 ypN0, ypT0/is ypN0, and ypT0/is--for their association with EFS and OS in a responder analysis. We assessed the association between pathological complete response and EFS and OS in various subgroups. Finally, we did a trial-level analysis to assess whether pathological complete response could be used as a surrogate endpoint for EFS or OS.

FINDINGS

We obtained data from 12 identified international trials and 11 955 patients were included in our responder analysis. Eradication of tumour from both breast and lymph nodes (ypT0 ypN0 or ypT0/is ypN0) was better associated with improved EFS (ypT0 ypN0: hazard ratio [HR] 0·44, 95% CI 0·39-0·51; ypT0/is ypN0: 0·48, 0·43-0·54) and OS (0·36, 0·30-0·44; 0·36, 0·31-0·42) than was tumour eradication from the breast alone (ypT0/is; EFS: HR 0·60, 95% CI 0·55-0·66; OS 0·51, 0·45-0·58). We used the ypT0/is ypN0 definition for all subsequent analyses. The association between pathological complete response and long-term outcomes was strongest in patients with triple-negative breast cancer (EFS: HR 0·24, 95% CI 0·18-0·33; OS: 0·16, 0·11-0·25) and in those with HER2-positive, hormone-receptor-negative tumours who received trastuzumab (EFS: 0·15, 0·09-0·27; OS: 0·08, 0·03, 0·22). In the trial-level analysis, we recorded little association between increases in frequency of pathological complete response and EFS (R(2)=0·03, 95% CI 0·00-0·25) and OS (R(2)=0·24, 0·00-0·70).

INTERPRETATION

Patients who attain pathological complete response defined as ypT0 ypN0 or ypT0/is ypN0 have improved survival. The prognostic value is greatest in aggressive tumour subtypes. Our pooled analysis could not validate pathological complete response as a surrogate endpoint for improved EFS and OS.

FUNDING

US Food and Drug Administration.

Abstract A027: Endothelial-like properties of triple-negative claudin-low breast cancer cells promote tumor-vascular interactions and tumor permeability

The different intrinsic subtypes of human breast tumors vary in rate and location of metastatic dissemination. In this study, we aimed to determine if tumor-associated vascular properties could help to explain the differences observed in metastagenicity. We used endothelial and breast cancer cell line models to develop gene expression signatures that measured vascular quantity and vascular proliferation. We tested these signatures, amongst other published endothelial signatures, and found that the genomic programs that measured vascular quantity, vascular proliferation, and a VEGF/Hypoxia-signature were the most highly expressed in claudin-low and basal-like tumors. We then tested the vascular signatures on five human breast cancer datasets (n&gt;3,000) and identified two vascular signatures that added prognostic information to intrinsic subtype classification. Next, microvessel density (MVD) scores were obtained from a set of 70 tumors which were also subjected to gene expression profiling. Interestingly, several of the vascular signatures added metastasis-predicting information in addition to MVD (p&lt;0.05). Similar to the tumor data, claudin-low cell lines also expressed high levels of vascular gene signatures as compared to luminal cell lines. Interestingly, pure claudin-low cell lines, and subsets of claudin-low-like cells within established basal-like cell cancer cell lines, exhibited endothelial-like morphology when culture on Matrigel. In vivo xenografts of claudin-low tumors, but not luminal tumors, extensively perfused injected contrast agent through paracellular spaces and non-vascular tumor-lined channels. Taken together, the endothelial-like characteristics of the cancer cells, combined with both the amount and the physiologic state of the vasculature, contribute to breast cancer metastatic progression. We hypothesize that the genetic signatures we have identified highlight patients that should respond most favorably to anti-vascular agents.

Citation Format: J. Chuck Harrell, Adam Pfefferle, Nicole Zalles, Aleix Prat, Cheng Fan, Andrey Khramtsov, Olufunmilayo Olopade, Melissa Troester, Andrew Dudley, Charles Perou. Endothelial-like properties of triple-negative claudin-low breast cancer cells promote tumor-vascular interactions and tumor permeability. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Breast Cancer Research: Genetics, Biology, and Clinical Applications; Oct 3-6, 2013; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Res 2013;11(10 Suppl):Abstract nr A027.

Abstract LB-265: Patient-derived xenografts from advanced luminal-type breast cancer: insights into endocrine therapy resistance

Deeper understanding the mechanisms by which luminal-type breast cancer develops resistance to endocrine therapy and development of novel strategies to treat these patients requires model systems recapitulate human breast cancer as accurately as possible. An increasing body of work suggests patient derived xenografts (PDX) may represent an informative model for development of novel therapeutics. We therefore established seven xenograft tumor lines from late-stage breast cancer patients with estrogen positive (ER+) disease. To date five ER+ PDX lines have been tested for responses to estradiol treatment in overiectomized NOD/SCID mice. Three showed estradiol independent-growth, one estrogen-stimulated growth and in one estradiol-induced a regression. These patterns mimicked the clinical phenotypes of each patient, tracking survival and responses to serial endocrine treatments. To define new mechanisms for resistance, whole genome DNA sequencing, RNA sequencing and Reverse Phase Protein Assay analysis was conducted. These studies identified an ESR1/YAP1 balanced translocation in a PDX model and tumor of origin showing low levels of ER, paradoxical high level expression form luminal genes and extreme ET resistance. The ESR1 YAP1 fusion maintained the N terminal DNA binding motif of ESR1, but the hormone binding and AF2 motifs were replaced with the C terminal transactivation domain of YAP1. Expression ESR1 YAP1 in ER+ breast cancer models down-regulated ER and induced estrogen independent growth. PDX endocrine phenotypes parallel tumor of origin responses to endocrine therapy and revel novel mechanism for endocrine therapy resistance.

Citation Format: Matthew J. Ellis, Shunqiang Li, Dong Shen, Li Ding, Robert Crowder, Jeiya Shao, Rodrigo Goncalves, Yu Tao, Jingqin Luo, Aleix Prat, Wenbin Liu, Ana Maria Gonzalez-Angulo, Shuying Liu, Joshua F. McMichael, Chris Miller, Dave Larson, Robert S. Fulton, Tom Mooney, Jeremy Hoog, Li Lin, Therese Giuntoli, Caroline Bumb, Crystal Cooper, Rebecca Aft, Robert T. Kitchens, Stephen N. Johnson, Chanpheng Phommaly, Megha Shiyam Kavuri, Katherine DeSchryver, Austin Lin, YiYu Dong, Cynthia X. Ma, Timothy Pluard, Michael Naughton, Ron Bose, Rama Suresh, Reida G. McDowell, Loren Michel, Richard Wilson, Shaomeng Wang, Christopher Maher, Gordon B. Mills, Charles Perou, Elaine R. Mardis. Patient-derived xenografts from advanced luminal-type breast cancer: insights into endocrine therapy resistance. [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 LB-265. doi:10.1158/1538-7445.AM2013-LB-265

Abstract LB-236: Patient derived xenografts as high-fidelity genomic models for advanced breast cancer

A limitation of recent genome studies is that they have almost exclusively focused on early stage untreated samples. Consequently the genomic landscape of advanced and treatment-resistant disease is poorly documented, even though Stage 4 patients stand to benefit most from new therapeutic approaches. To address this issue we developed a panel of patient-derived xenografts (PDX) from Stage 3 and 4 patients with treatment-resistant disease to create a platform for detailed molecular and pharmacological investigation. Recently we reported a whole genome sequencing study of a single example of a breast cancer primary, brain metastasis and PDX “trio” demonstrating that the PDX model efficiently captured almost all of the genome-wide variants observed in the tumor and was enriched for mutations present in the metastatic sample (Ding, Ellis et al., Nature 2010). We now extend these findings by conducting massively parallel sequencing of 13 additional PDX models from a diverse set of locally- advanced or Stage 4 breast cancers, all of whom subsequently died from their disease. Structural variants (deletions, inversions and translocations) were remarkably stable and although new single nucleotide variants were observed post-engraftment, most were either non-coding or not expressed and only one had established functionality (K-RAS). PDX models are therefore high-fidelity genomic models for the biological and pharmacological study of complex, often tumor-unique somatic events responsible for lethal drug-resistant advanced breast cancer.

Citation Format: Shunqiang Li, Dong Shen, Ding Li, Tao Yu, Jingqin Luo, Jeremy Hoog, Joshua McMichael, Chris Miller, Dave Larson, Ron Bose, Bob Fulton, Rick Wilson, Charles Perou, Elaine Mardis, Matthew Ellis. Patient derived xenografts as high-fidelity genomic models for advanced breast cancer. [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 LB-236. doi:10.1158/1538-7445.AM2013-LB-236

Abstract P6-07-10: Luminal A vs. Basal-like Breast Cancer: time dependent changes in the risk of relapse in the absence of treatment

Background: Numerous retrospective analyses of prospective randomized clinical trials of patients treated with adjuvant tamoxifen and chemotherapy have demonstrated that the breast cancer intrinsic subtype Luminal A tumors generally have favorable early initial outcomes, while basal-like tumors are associated with a marked risk of early relapse. To determine the extended natural history of the intrinsic subtypes across two decades of follow up the PAM50 “non-commercial open source bioinformatics” qPCR assay was conducted on node negative tumors accrued through the Cooperative Breast Cancer Tissue Registry (CBCTR) from patients who did not receive systemic therapy.

Methods: Intrinsic subtype calls were obtained from 331 CBCTR cases treated with local interventions only. Tumors were classified into Luminal A (LumA), Luminal B (LumB), HER2-enriched (HER2-E) and Basal-like (BLBC), and correlated relapse-free (RFS). Patient survival and hazard rate were estimated using Kaplan-Meier plots and log-rank test. Multivariable Cox regression analyses were used to determine the significance of the intrinsic subtypes, adjusted with standard clinicopathological variables including tumor size, age at diagnosis, grade, radiation therapy treatment, centralized reviewed estrogen receptor, progesterone receptor and human epidermal growth factor 2 status measured by immunohistochemistry. Patients were diagnosed from 1978 to 1992, with a mean follow-up time of 13 years (range 0.5–31).

Results: Of the 331 tumors tested, 51% of cases were classified as LumA, 18% as LumB, 11% as HER2-E and 20% as BLBC. Although LumA was associated with the best outcome for the first 10-year of follow-up, the final number of RFS events were eventually comparable with those observed for BLBC with prolonged follow up (Table 1). In the multivariable Cox model, only BLBC tumors were associated with worse prognosis than LumA with borderline significance (Hazard ratio: 2.0 (95% CI 0.9–5), p = 0.07). BLBC had the highest hazard rates for the first 5 years (7% at first year to 5% at 5 yr), consistent with previous observations. Interestingly, in the absence of treatment, the slow growing LumA subtype had a gradual increase of hazard for an RFS event from 3% at 5 yrs to 4% at 10 yrs to 7% at 20 yrs. The hazard rates of LumA cross with those of BLBC at 10 years.

Abstract P1-14-20: Meta-analysis Results from the Collaborative Trials in Neoadjuvant Breast Cancer (CTNeoBC)

Withdrawn by Author

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-14-20.

Pathologic complete response predicts recurrence-free survival more effectively by cancer subset: results from the I-SPY 1 TRIAL--CALGB 150007/150012, ACRIN 6657

PURPOSE

Neoadjuvant chemotherapy for breast cancer provides critical information about tumor response; how best to leverage this for predicting recurrence-free survival (RFS) is not established. The I-SPY 1 TRIAL (Investigation of Serial Studies to Predict Your Therapeutic Response With Imaging and Molecular Analysis) was a multicenter breast cancer study integrating clinical, imaging, and genomic data to evaluate pathologic response, RFS, and their relationship and predictability based on tumor biomarkers.

PATIENTS AND METHODS

Eligible patients had tumors ≥ 3 cm and received neoadjuvant chemotherapy. We determined associations between pathologic complete response (pCR; defined as the absence of invasive cancer in breast and nodes) and RFS, overall and within receptor subsets.

RESULTS

In 221 evaluable patients (median tumor size, 6.0 cm; median age, 49 years; 91% classified as poor risk on the basis of the 70-gene prognosis profile), 41% were hormone receptor (HR) negative, and 31% were human epidermal growth factor receptor 2 (HER2) positive. For 190 patients treated without neoadjuvant trastuzumab, pCR was highest for HR-negative/HER2-positive patients (45%) and lowest for HR-positive/HER2-negative patients (9%). Achieving pCR predicted favorable RFS. For 172 patients treated without trastuzumab, the hazard ratio for RFS of pCR versus no pCR was 0.29 (95% CI, 0.07 to 0.82). pCR was more predictive of RFS by multivariate analysis when subtype was taken into account, and point estimates of hazard ratios within the HR-positive/HER2-negative (hazard ratio, 0.00; 95% CI, 0.00 to 0.93), HR-negative/HER2-negative (hazard ratio, 0.25; 95% CI, 0.04 to 0.97), and HER2-positive (hazard ratio, 0.14; 95% CI, 0.01 to 1.0) subtypes are lower. Ki67 further improved the prediction of pCR within subsets.

CONCLUSION

In this biologically high-risk group, pCR differs by receptor subset. pCR is more highly predictive of RFS within every established receptor subset than overall, demonstrating that the extent of outcome advantage conferred by pCR is specific to tumor biology.

Abstract 988: Oncogenic PI3K mutations lead to NF-κB-dependent cytokine expression following growth factor deprivation

While growth factor-independent signaling and proliferation are well-established hallmarks of cancer, little is known regarding growth factor-independent changes in gene expression which occur downstream from oncogenes. The PI3K pathway is one of the most commonly misregulated signaling pathways in human cancers. Here, MCF10A cells expressing the two most common PI3K mutations, PIK3CA E545K and H1047R, were used to identify the repertoire of genes altered by oncogenic PI3K mutations following growth factor deprivation. This gene set most closely correlated with gene signatures from claudin-low and basal-like breast tumors, and categorical enrichment analyses suggested that NF-κB target genes were dramatically upregulated by these mutations. An IKKα inhibitor was used to identify the subset of PI3K-driven genes that is NF-κB dependent. Interestingly, virtually all of these NF-κB dependent genes were secreted proteins, suggesting a paracrine role for this gene set. Among these genes was IL-6, a cytokine frequently expressed in tumors which plays a critical role in generating a tumor-promoting microenvironment. Consistent with this, conditioned media from cells expressing the E545K or H1047R mutations led to increased STAT3 activation in recipient THP-1 monocytes or normal epithelial cells in a NF-κB and IL-6-dependent manner. Together, these data define a PI3K-driven, NF-κB-dependent gene expression profile which may play a critical role in promoting a microenvironment amenable to tumor progression.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 988. doi:1538-7445.AM2012-988

MS1-3: Tumor Heterogeneity and Metastasis

Massively parallel sequencing using Next Gen sequencing approaches have opened a new window into our understanding of the molecular architecture of breast cancer. A significant advantage of these technologies are that they are digital in nature because the measurement technology begins with single DNA molecules attached to a substrate. As a result, not only do these technologies provide an efficient means to identify somatic mutations, information is also provided on the frequency with which any given mutation is observed within the population of DNA molecules in the sample under analysis. The first and most striking feature of Next Gen analysis of breast cancer is the large numbers of somatic mutations that have been identified. Genome wide, thousands of mutations can be present in a single tumor. While this finding generates immense complexity from the perspective of defining biological drivers versus passengers, it does open up a new opportunity to study variations in mutation frequency. In a perfectly monoclonal tumor with a diploid genome and no contaminating normal cell DNA the mutation frequencies generated by digital sequencing will be 50% for a heterozygous mutation or 100% for homozygous mutations. Very few breast cancers subjected to whole genome sequencing exhibit this pattern, rather mutations occur with a wide variation of frequencies from the detection limit, around 5%, through to 100%. Furthermore mutation frequency clustering often occurs, a phenomenon best explained by the presence of a repertoire of mutations present in a founder clone that are present in all tumors, combined with additional less common mutations that represent a subdominant or minority population of cells that have arisen through additional clonal outgrowth. In a multiclonal model one of the most critical questions is “which clone determines the prognosis?”. One way to begin to answer this question is to sequence tumors before and after therapy and to compare primary tumors versus metastases. In our initial experience of a patient with a basal-like breast cancer in which a whole cancer genome was generated from her primary, her subsequent brain metastasis and a mouse xenograft generated from the breast primary, there was clear evidence of a shift in mutation frequency primary to metastasis, suggesting the brain metastasis arose from a minor sub-clone. The mutational profile of the xenograft more closely resembled the metastasis rather than the primary, suggesting the grafting process captured the metastatic clone (1).

In this invited lecture we will present an extension of our sequencing experiences with further human progenitor-mouse xenograft comparisons, comparisons between human primaries and paired metastases and also examples of ER+ breast cancer genomes obtained before and after neoadjuvant endocrine therapy. In all these experiments multiclonality appears to be the rule rather than the exception. The clinical and biological implications of these findings will be discussed.

1. Ding L, Ellis MJ, Li S, Larson DE, Chen K, Wallis JW, et al. Genome remodelling in a basal-like breast cancer metastasis and xenograft. Nature. 2010;464(7291):999–1005. PMCID: 2872544.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr MS1-3.

Phase II study of bortezomib and pegylated liposomal doxorubicin in the treatment of metastatic breast cancer

BACKGROUND

Based on preclinical studies and a phase I trial of the combination of bortezomib and pegylated liposomal doxorubicin (PLD), which both showed activity in breast cancer, we conducted a phase II study of this regimen in patients with metastatic breast cancer.

PATIENTS AND METHODS

Patients received bortezomib 1.3 mg/m2 on days 1, 4, 8, and 11 of an every-21-day cycle, along with PLD 30 mg/m2 on day 4. The primary objective was to evaluate the response rate of this combination, while secondary objectives were to obtain further safety data about this combination, to evaluate the time to disease progression (TTP), and to evaluate response by the breast cancer subtype.

RESULTS

One of 12 evaluable patients had a partial response (8%), while 3 (25%) had stable disease. At 26 months follow-up, the median overall survival was 4.3 months (95% CI, 1.2-26.2) and the median TTP was 1.3 months (95% CI, 0.8-14.0 months). The combination was well tolerated, with the most common events including low-grade nausea and vomiting, neutropenia, and neuropathy, and no cardiac toxicity was seen. Of the 7 tumors subtyped, no association was seen between intrinsic subtype or receptor status and response.

CONCLUSION

The combination of PLD and bortezomib was well tolerated but has minimal activity in heavily pretreated unselected metastatic breast cancer.

Abstract 3004: Specific microRNAs associate with aggressive breast tumor phenotypes in an ethnically diverse patient cohort

Background: Sporadic basal like breast cancer (BLC), a subset of triple negative tumors with bad prognosis appears to be overrepresented in young women and women of African Ancestry. The molecular drivers of basal like and triple negative tumors have not been identified and there is a lack of targeted therapies for women with these tumor phenotypes. MicroRNAs (miRNAs), a new class of small non-coding RNAs, are known to regulate expression of genes important for several biological processes including stem cell division, apoptosis and cell proliferation. Thus, miRNAs have been linked to several diseases including cancer. To examine whether miRNAs play a role in the development and progression of breast cancer, we analyzed 47 primary breast tumors. We hypothesize that specific miRNAs will be associated with particular subtypes of breast cancer especially the biologically aggressive basal like subtype.

Materials and Methods: To test our hypothesis, we used a number of complementary approaches 1.) Using exiqon's miRNA microarray, we profiled 47 breast tumors enriched for African Americans and 4 adjacent normal samples for expression of all known miRNAs. Real time RT-PCR was used to validate the expression data 2.) All tumor samples were classified into the known molecular subtypes using DNA microarray and surrogate immunohistochemistry markers 3.) A subset of the patients with paired normal and tumor samples were analyzed for genomic alterations including deletions and amplifications using 100k Illumina SNP array to corroborate the expression data.

Results: The diverse patient cohort included 36 African Americans, 10 Whites, 1 Hispanic; median age was 51 years. The tumor subtypes included 13 basal like, 11 HER2, 9 Luminal A, 7 Luminal B and 5 normal like. We found 151 miRNA were differentially expressed between primary triple negative and non triple negative tumors. We also found 17 miRNAs that were down regulated in basal tumors. More importantly, some of the 17 microRNAs downregulated in primary BLCs were miRNAs already identified to play a role in de-differentiation or epithelial to mesenchymal transition, all characteristics of BLC. Lastly, some of the miRNAs downregulated in primary BLCs were located at chromosomal regions that are frequently deleted in basal like tumors.

Conclusion: Our data suggests that miRNAs downregulated in BLCs can be lost by several genomic mechanisms and more importantly loss of expression of specific miRNAs may play important roles in development of tumors with this phenotype. Ongoing work is aimed at examining the molecular mechanisms whereby specific miRNAs contribute to the development of basal like tumors.and in understanding the clinical implications of differentially expressed miRNAs in breast tumor progression. These studies underscore the need for an integrated genomic approach to understand breast cancer pathogenesis in diverse populations.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 3004.

Genome-Wide Profiling of Archived Material from CALGB 9840 and 9342 for Paclitaxel (P) and Trastuzumab (T) Response Biomarkers Using Gene Expression and Copy Number Analysis

Background:Emerging data suggest that RNA obtained from formalin-fixed, paraffin-embedded (FFPE) tissue can produce reliable gene expression profiles. Archived material from two taxane monotherapy studies, CALGB 9342 (comparison of 3 doses of P) and 9840 (weekly vs. q3 week P)were profiled for gene expression and DNA copy number.Methods: A total of 238 patients had primary tumor blocks available from a combined sample size of 680. DNA and RNA was extracted from 1.5mm punch cores and the Ambion Recover-All kit™. quality was measured by spectrophotometric analysis, Bioanalyzer RNA Integrity Number (RIN), and housekeeping genes (RPL13A and Actin). A custom DASL™ array containing 779 genes in two-fold redundancy was designed with genes selected to represent the PAM50 intrinsic subtypes, the Oncotype Dx Score, the Netherlands prognostic signature and the genes most frequently found on recurrent breast cancer amplicons. Several methods for identifying outliers were evaluated, including principal components analysis, pairwise correlations as well as the reproducibility of the platform based on replicate samples.Results: Adequate RNA was obtained from 237/238 of these cases which ranged in age from 12-18 years. Of these, 215/237DASL arrays passed further quality control measures. Adequate DNA for CGH was obtained from 227/238 samples. Analysis of PAM50 intrinsic subtypes showed an excess of basal-like tumors (30%) in the primary tumors of this metastatic cohort compared with expected frequency in an early stage population. Luminal A tumors were less frequent than expected (20%). Patients with basal-like tumors did far worse than other tumor types for both PFS on P (p=0.015) and OS (p=2.7X10-6), which persisted in multivariable analysis (p=0.0047), however the interaction term was not significant (Wald p=0.26). While basal-like tumors had similar PFS and OS on both weekly and q3 week P, luminal A tumors appear to achieve more benefit from weekly P (p= 0.0041).The HER2-enriched expression subtype had a similar prognosis to Luminal A and B tumors. This appeared to be due to the presence of T, as the addition of this agent improved PFS (p=0.026) and OS (p=2.0X10-4). Of note, some centrally confirmed HER2 FISH amplified tumors were classified into luminal A, B, and basal-like subtypes. These tumors have similar prognoses to the overall group, for example the basal-like and HER2 tumors had a poor prognosis despite T (p=0.00086). This suggests that HER2 FISH positive tumors may behave based on the underlying tumor subtype. Sawtooth genomes (45% vs 15%) were more frequent than predicted by an early stage tumor dataset as were simplex genomes (3% vs 24%). HER2 by FISH and CGH were highly concordant suggesting data on gene amplification from this platform is robust.Conclusions: Gene expression and copy number profiling of FFPE material from archived tumor blocks (&gt;10 years) produces quality data for biomarker discovery in CALGB clinical trial datasets. These tools allow discovery of novel patterns of gene expression and genomic aberrations that are associated with differential response to P and T. Further studies using these platforms should be performed.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4032.