2016 UCSF Breast Oncology Program Scientific Retreatcancer.ucsf.edu/research/BOP2016/sm_files/BOP...

This retreat is supported in part by Nektar Therapeutics, Puma Biotechnology, Color Genomics, Inc., Celgene Corporation, and the UCSF Helen Diller Family Comprehensive Cancer Center Support Grant (P30 CA 082103).

Page 1

2016 UCSF Breast Oncology Program Scientific Retreat

Poster Session Abstracts

Table of Contents Poster Category Detection and Biomarkers 2-27 Epidemiology and Population Sciences 28-33 Molecular Biology Poster 34-44 Physical Sciences Poster 45-48 Therapeutics and Clinical Trials Poster 49-61 The Breast Oncology Program Scientifc Retreat is supported in part by:

• Nektar Therapeutics • Color Genomics, Inc. • Celgene Corporation • UCSF Helen Diller Family Comprehensive Cancer Center Support Grant (P30

CA 082103)


Page 2


Detection and Biomarkers Poster #1 A Quadrant-Based Quantitative Analysis of Background Parenchymal Enhancement in Breast MRI Ella F. Jones, Natalie Hartman, Helen Park, Wen Li, Ania Azziz, David C. Newitt, John Kornak, Bonnie N. Joe, and Nola M. Hylton

Presenting Author: Ella Jones Radiology Breast Imaging [email protected] Abstract PURPOSE: Background parenchymal enhancement (BPE) is the contrast enhancement of normal fibroglandular breast tissues measured by dynamic contrast-enhanced (DCE) MRI. High level of BPE has been associated with breast cancer risk. Despite the potential importance of BPE, there is little effort to systematically assess BPE quantitatively and compare it to the qualitative Breast Imaging-Reporting and Data System (BI-RADS) classification. Herein, we present a quadrant-based analysis of breast parenchyma, in an attempt to search for regional variations in BPE that may influence corresponding qualitative assessment. METHODS: With institutional review board approval, a retrospective analysis of breast DCE-MRI was performed on 172 patients with BI-RADS 1 (normal) or 2 (benign) final assessments from 2007 to 2010. Qualitative BPE was classified into categories of minimal (n=58), mild (n=47), moderate (n=40), and marked (n=27). Fibroglandular tissue was segmented from pre-contrast T1 weighted images using a fuzzy C-means clustering method 4. One pixel erosion at the periphery of each breast tissue mask was applied to compensate for motion artifacts between pre- and post-contrast scans. The quadrant analysis of the left breast was performed using the nipple as an anatomical reference on the sagittal image. Automatic segmentation yielded four approximately equal quadrants: lower inner (LIQ), lower outer (LOQ), upper inner (UIQ) and upper outer (UOQ) which were confirmed on coronal and axial images. A percent enhancement map (percent BPE=100*(S1-S0)/S0 where S0/S1=pre/post-contrast signal intensity) was generated. The mean percent BPE value, termed PBPE, was calculated over the breast tissue mask for each quadrant. The Wilcoxon signed-rank test was used to compare the quadrant PBPEs. Linear models were used to determine the effect of age on quadrant PBPE. Spearman’s correlation was used to assess the correlation between quantitative PBPE and qualitative BI-RADS classification. RESULTS: PBPE values within the four separate quadrants exhibited distinct contrast enhancement behaviors. The upper quadrants showed significantly higher PBPE values than the lower quadrants with median PBPE at 16.91% for UIQ and 16.87% for UOQ vs. 14.40% for LIQ and 14.49% for LOQ. Estimated (pseudo-) median difference = 2.02%, 95% CI (1.19, 2.87), p < 0.0001 for UIQ vs. LIQ and 2.42%, 95% CI (1.57, 3.26), p < 0.0001 for UOQ vs. LOQ. The differences between estimated inner and outer quadrant medians for upper and lower sections were small and did not achieve statistically significance (p = 0.11 and 0.53 for UIQ/UOQ and LIQ/LOQ respectively). The effect of age had a statistically significant association with 3 out of 4 quadrant PBPE values. In a single predictor/unadjusted model, LIQ showed an estimated change of -18.4% per 10 years increase in age,


Page 3


95% CI (-34.5, -2.20), p = 0.03. Similar relationship was found in UIQ and UOQ: -17.6% for UIQ (95% CI (-35.3, 0.224), p = 0.05) and -23.2% for UOQ (95% CI (-41.3, -5.07), p = 0.01) per 10 years. Although LOQ also exhibited a similar trend with estimated change of -12.4% and 95% CI (-27.1, 2.42), it did not achieve statistical significance (p = 0.10). Statistically significant but mild Spearman correlations were detected between all quadrant PBPEs and qualitative BI-RADS with UOQ being the highest: estimated correlation rho = 0.33, 95% CI (0.19, 0.46), p < 0.0001 and rho = 0.29, 95% CI (0.06, 0.50), p = 0.017 in the patient subset of moderate and marked BI-RADS classifications. The distribution of quantitative UOQ PBPE values in each qualitatively assessed BI-RADS BPE shows higher categories of BI-RADS corresponded to an overall increase in PBPE (this association was more pronounced in younger patients: data not shown). It is worth noting that there was considerable overlap in PBPE values across BI-RADS BPE classifications. DISCUSSIONS: The quadrant analysis reveals a distinct regional enhancement behavior within the normal breast. Upper quadrants exhibit higher enhancement with the UOQ PBPE values estimated to have the highest (positive) correlation with BI-RADS classification. The covariate effect of age indicates that younger patients having higher regional PBPE may result in higher qualitative interpretations. The high overlap of PBPE values across BI-RADS categories indicates a lack of agreement between quantitative and qualitative assessments. Possible contributing factors include image artifact effects on quantitative PBPE and variability of observer’s subjective BI-RADS interpretations. In conclusion, this study provides the first quantitative assessment of BPE by quadrants in a non-cancer population and indicates that regional enhancement in the breast has an influence on the BI-RADS interpretation, particularly in the moderate and marked categories.


Page 4


Detection and Biomarkers Poster #2 Breast Cancer Risk Prediction using a Combined Clinical Risk Calculator and Polygenic Risk Score Yiwey Shieh - Division of General Internal Medicine, UCSF; Donglei Hu - Division of General Internal Medicine, UCSF; Lin Ma – UCSF; Charlotte C. Gard - Department of Economics, Applied Statistics, and International Business; New Mexico State University; Jessica Leung - Department of Diagnostic Radiology, MD Anderson Cancer Center; Jeffrey A. Tice - Division of General Internal Medicine, UCSF; Celine M. Vachon - Department of Health Sciences Research, Division of Epidemiology; Mayo Clinic; Steven R. Cummings - San Francisco Coordinating Center; Karla Kerlikowske - Division of General Internal Medicine, UCSF; Elad Ziv - Division of General Internal Medicine, UCSF

Presenting Author: Yiwey Shieh Division of General Internal Medicine [email protected] Abstract Background: Personalized assessment of breast cancer risk can inform the use of screening and prevention modalities. The Breast Cancer Surveillance Consortium (BCSC) risk model, which includes breast density, has been validated in multiple racial/ethnic groups. Adding the cumulative effects of multiple single nucleotide polymorphisms (SNPs) as a polygenic risk score was recently shown to improve the performance of the BCSC model. We sought to replicate these findings in an independent population. Methods: We conducted a nested case-control study within a cohort of women receiving mammographic screening at the California Pacific Medical Center. Women with a diagnosis of breast cancer were matched 1:1 with controls based on age and race/ethnicity. The polygenic risk score was calculated as the product of the likelihood ratios for breast cancer associated with 91 SNPs, with the assumption that the SNPs are independent. Multivariate logistic regression was used to explore the contributions of the polygenic risk score and variables in the BCSC model to breast cancer risk. Discriminatory accuracy of models was quantified using the area under the receiver operator characteristic curve (AUROC). Results: We included 486 cases and 495 matched controls. Eighty percent were Caucasian and 11% were Asian. Increasing quartiles of polygenic risk score (PRS) were associated with higher risk of breast cancer, with the highest quartile having a 2.71-fold odds (95% CI 1.87-3.93) compared with the lowest quartile. In a multivariate model, the PRS, family history, and breast density remained strong risk factors. The combined model (PRS + BCSC) had better discriminatory ability compared with the BCSC model (AUROC 0.65 vs 0.62, p < 0.001). The polygenic risk score had an AUROC of 0.61 (95% CI 0.56 to 0.64) across all subjects, and its performance was preserved in the subset of Asian participants, with an AUROC 0.64 (95% CI 0.53 to 0.75). The combined model classified 20% of cases as high-risk (5-year risk greater than 3%), compared with 10% using the BCSC model. Conclusions: Adding the polygenic risk score to the BCSC model improved discrimination and identified additional high-risk women who met the 5-year risk threshold for chemoprevention. Precision screening strategies are evaluating the clinical role of measuring SNPs.


Page 5


Detection and Biomarkers Poster #3 Non-invasive measurement of mTORC1 signaling with 89Zr-transferrin Charles Truillet,1 John T. Cunningham,2 Davide Ruggero,2 Jason S. Lewis,3 Michael J. Evans1 1 Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA 94143, USA 2 School of Medicine and Department of Urology, Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, San Francisco, CA 94158, USA 3 Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York, New York, NY 10065, USA

Presenting Author: Charles Truillet Radiology and and Biomedical Imaging Michael J. Evans Laboratory [email protected] Abstract Glioblastoma multiforme (GBM) and prostate cancer (PCa) generally harbor hyperactive PI3K/Akt/mTOR signaling. However, clinical responses to small molecule therapies targeting kinases in this pathway have been modest, and to date, there are no drugs cleared for these indications. While feedback mechanisms capable of overcoming single agent therapy have been defined, the robust antitumor effects observed in preclinical models suggest that they may not solely account for the weak clinical responses. We hypothesized that incomplete target inhibition may also account for poor clinical responses. To test this hypothesis in man, we show that 89Zr-transferrin, a molecule that we are actively translating into the clinic for positron emission tomography, can measure mTORC1 activity. Genetic manipulations that activate mTORC1 signaling elevated transferrin uptake into GBM and PCa cells in vitro. Moreover, 89Zr-transferrin uptake was higher in prostate adenocarcinoma due to homozygous loss of PTEN compare to wild type mouse prostates. Conversely, several mechanistically distinct drugs that suppress mTORC1 activity inhibited transferrin uptake in vitro. Finally, we show that TFRC expression does not change after chronic mTORC1 inhibition, suggesting a post-translational mechanism of TFCR regulation. Collectively, these data support the use of 89Zr-Tf to interpret the pharmacology of targeted therapies that inhibit mTORC1, a finding that could have immediate implications for drug development.


Page 6


Detection and Biomarkers Poster #4 Development of radiolabeled artificial receptors to target important cancer signaling pathways Matthew F. L. Parker, Department of Radiology and Biomedical Imaging, UCSF Charlie Truillet, Department of Radiology and Biomedical Imaging, UCSF Michael Evans, Department of Radiology and Biomedical Imaging, UCSF

Presenting Author: Matthew Parker Radiology and Biomedical Imaging Michael Evans Laboratory [email protected] Abstract A key feature found akin in many malignant cell lines is an increase in the presence of transmembrane cell signaling proteins (a class of proteins called receptor tyrosine kinases or RTK) on the surfaces of these cells. The function of RTKs is involving with signaling and communication. In addition to an increase of volume, increases in signaling are also prevalent in tumors. On the biochemical level, the signaling is achieved through a phosphorylation event on select residues of the amino acid tyrosine present in the intercellular tail of RTKs. Due to the increase of volume and frequency of phosphorylation occurrences in tumor cells versus normal tissues, we are interested in evaluating phosphorylated tyrosine (p-tyr) as a useful biomarker for imagining and detection of cancer. Through the use of the literature and computational chemistry, we have identified and are developing β-cyclodextrin (a cyclic compound possessing 7 glucose molecules connected through 1,4-glycocidic linkages) based receptors to bind p-tyr. A key feature of our proposed receptors is the use of multiple hydrophillic guanidinium groups that will act as hydrogen-bonding pairing to the phosphate group of p-tyr while the hydrophobic interior of the cyclodextrin can accommodate the tyrosine moiety. Our proposed cyclodextrin receptors offer multiple opportunities for the introduction of 11C and 18F radionuclides, in order to develop them for nuclear imaging tools. Here we show our progress toward the synthesis of these receptors and their evaluation in simple, controlled systems and more complex biological systems. If successful, our hope is that this work will lead to new strategies for rapid detection and monitoring of cancer using positron emission tomography (PET) imaging.


Page 7


Detection and Biomarkers Poster #5 Development of radiolabeled artificial substrates to target important cancer metabolic pathways Matthew F. L. Parker, Department of Radiology and Biomedical Imaging, UCSF Charlie Truillet, Department of Radiology and Biomedical Imaging, UCSF Michael Evans, Department of Radiology and Biomedical Imaging, UCSF

Presenting Author: Matthew Parker Radiology and Biomedical Imaging Michael Evans Laboratory [email protected] Abstract A known consequence to the abnormal and rapid cell division observed in many malignant cell lines is the increased need for many of the essential raw materials used to construct important biological macromolecules like DNA, RNA, and proteins. To accommodate this demand, many tumors overexpress much of the biochemical machinery. One specific enzyme of interest to us is phosphoribosyl pyrophosphate synthetase 2 (PRPS2) because it has been found to be a rate-limiting enzyme on the purine biosynthesis pathway (an important anabolic energy supply for the cancer cells). Another key feature of this enzyme is that the chemical transformation it is responsible for converts a relative hydrophobic compound into a significantly hydrophilic compound, which prevents it from leaving the cells. Due to the increased presence and capability to metabolically trap substrates within the cell, we are interested in evaluating PRPS2 as a useful biomarker for imaging and detection of cancer. Through the use of the literature and computational chemistry, we have identified and are developing artificial substrates for PRPS2 that bear similarity to its natural substrate but small modification have been made to address cell permeability and stability. Due to the very small size of the natural substrate for PRPS2, we have developed some rather clever chemical strategies to introduce 18F radionuclides, in order to develop new nuclear imaging tools. Here we show our progress toward the synthesis of these substrates and their evaluation in simple, controlled systems and more complex biological systems. If successful, our hope is that this work will lead to new strategies for rapid detection and monitoring of cancer using positron emission tomography (PET) imaging.


Page 8


Detection and Biomarkers Poster #6 Node-pore sensing: a novel platform to characterize and differentiate cellular mechanical phenotypes Junghyun Kim1, Sewoon Han1, Andy Lei2, Masaru Miyano3, Mark A. LaBarge3, and Lydia L. Sohn1 Author Affiliations 1 Department of Mechanical Engineering, UC Berkeley, Berkeley, California, USA. 2 Department of Bioengineering, UC Berkeley, Berkeley, California, USA. 3 Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, USA.

Presenting Author: Junghyun Kim Department of Mechanical Engineering, UC Berkeley Sohn Laboratory [email protected] Abstract Mechanical properties of cells could potentially be used as label-free biomarkers for the metastatic potential of cancer cells, cell-cycle stages, and degree of stem-cell differentiation. However, current techniques for measuring cellular mechanical properties have had limited adoption for biological application because of their low throughput or restricted number of measured parameters. Here, we demonstrate a microfluidic platform that measures multiple biophysical parameters of cells, simultaneously. Our method employs node-pore sensing (NPS): a method by which cells produce a uniquely shaped current pulse as they transit a microfluidic channel that has been segmented by a series of nodes. For our application, one of the channel segments (which hereafter we refer to as the “contraction channel”) is narrower than the diameter of the cells, leading to their deformation (under constant force) as they transit this portion of our platform. An analysis of the current pulses produced leads to specific quantitative information about the individual cells, including the free cell diameter, the deformed cellular diameter, and their transit time through the contraction channel. Using these parameters and the known fluid velocity within the contraction channel, we can define a dimensionless parameter, the cell deformability index (CDI), for each cell. As we show in a series of proof-of-principle experiments, we can successfully use the CDI to distinguish normal breast (MCF-10A) and lung (BEAS-2B) epithelial cells from cancer cells (MCF-7 an A549) with extremely high accuracy. Furthermore, we can detect and quantify cytoskeleton network perturbations within cells. Finally, we show that we can use the CDI to distinguish luminal epithelial cells from myoepithelial cells and determine their ratios in primary samples. Overall, our NPS platform is a novel, high-throughput method for evaluating the mechanical phenotype of cells label-free. By combining the traditional immune-based assays with cellular mechanical phenotyping, we envision numerous applications in clinical diagnostics.


Page 9


Detection and Biomarkers Poster #7 Centromere and kinetochore gene misexpression predicts breast cancer patient outcome after radiotherapy and chemotherapy Weiguo Zhang[1,2], Jian-Hua Mao[1,3], Ke Liu[1,2], Ben Brown[1,2], Gary Karpen[1,2] 1: Lawrence Berkeley National Laboratory; 2: University of California, Berkeley; 3: University of California, San Francisco

Presenting Author: Weiguo Zhang Lawrence Berkeley National Laboratory [email protected] Abstract Chromosomal instability (CIN) is a cancer hallmark that contributes to tumor heterogeneity and other malignant properties. Conversely, the genotoxicity and proteotoxicity that accompany high levels of CIN can inhibit cancer cell growth. Aberrant centromere and kinetochore function causes CIN through chromosome missegregation and fragmentation, resulting in aneuploidy, rearrangements and micronucleus formation. We developed a CES system (for Centromere and kinetochore gene Expression Score) that quantitates the misexpression of genes required for centromere and kinetochore function in a wide spectrum of human cancers. In breast cancers, high CES values strongly correlate with increased copy number alterations and mutation frequencies, and effectively predict poor patient survival and disease progression in breast cancer subtypes. Importantly, the CES system also forecasts patient outcomes after adjuvant chemo- and radio- therapies. These results suggest that the CES system serves as an effective biomarker for patient prognosis and choosing therapeutic options. They also validate the utility of incorporating basic knowledge of chromosome segregation pathways into breast cancer research and clinical applications.


Page 10


Detection and Biomarkers Poster #8 The I-SPY 2 TRIAL: Serial sample collection for biomarker studies and early identification of non-responders. Lamorna Brown Swigart, Pat Glenn, Aye Aye Ma, Licia Dean, Laura van 't Veer, Laura Esserman and the I-SPY 2 Trial consortium

Presenting Author: Lamorna Brown Swigart Laboratory Medicine van 't Veer Laboratory [email protected] Abstract The I-SPY 2 TRIAL is an adaptive neoadjuvant trial for women with newly diagnosed, locally advanced breast cancer. The trail is enrolling patients at approximately 16 centers across North America. The trial design includes a master Investigational New Drug FDA approval that allows multiple novel drugs to be tested in the trial. Patients who consent and meet inclusion criteria receive standard of care treatment with or without one of the novel agents. The goal of the trial is to identify drug regimens that have ≥ 85% predictive probability of success in a neoadjuvant 300-patient phase 3 trial of patients in any of 10 possible signatures defined by HR, HER2, and MammaPrint. The adaptive aspect of the trial is achieved through a randomization engine that learns, from patients that have completed therapy, which signatures respond best to a given drug. The randomization engine then alters the odds of assignment to that arm accordingly. A secondary aim of the trial is to identify companion biomarkers that can be used to select patients who would benefit from the novel drug. Women who consent to screening undergo biopsies to provide material for MammaPrint and provide further biopsies three weeks after the start of treatment and at the time of surgery. These, and concurrent blood samples, are sent to the central I-SPY lab at UCSF. The longitudinal collection of samples allows evaluation of changes within the tumor under the influence of the drug combinations. Such changes may be indicative of a pathological response that will result in reduction of cancer burden. Conversely change, or the absence of change, may be an early indication of failure to respond. This knowledge may be used to adapt treatment for a non-responding patient earlier in her treatment. Here we describe the samples collected and the methods used to ensure pathological quality in material used for biomarker assays.


Page 11


Detection and Biomarkers Poster #9 Developing a Breast Fine Needle Aspiration Biopsy Service in Peru Dianna Ng1, Britt-Marie Ljung1, Richard Bardales2, Milagros Abad3, Manuel Cedano3, Tara Hayes Constant4, Jennifer L. Winkler4, Ronald Balassanian1 1 Department of Pathology, UCSF Medical Center, San Francisco, CA 2 Outpatient Pathology Associates, Sacramento, CA 3 Northern Regional Cancer Institute, Trujillo, Peru 4 PATH, Seattle Washington

Presenting Author: Dianna Ng Department of Pathology [email protected] Abstract Program/Project Purpose: Breast cancer incidence is increasing in low- and middle-income countries and may be related to changes in lifestyle, reproductive practices and life expectancy. In 2010, 57.8% of breast cancers in Peru were stage 3 or higher at diagnosis. As part of a multi-modal approach, federal and regional Peruvian institutions partnered with several organizations, including UCSF, to improve access to, and quality of breast cancer care in Peru. Phase 1 involved screening with clinical breast exam, creating a community-based referral system for women with masses, and promoting evaluation of lesions using fine needle aspiration biopsy (FNAB) by trained physicians, with treatment at the regional cancer institute. Curricula for each activity were developed and validated. During phase 1, it was recognized in-country FNAB expertise was limited. For phase 2, the critical objective was to solidify local capacity for high-quality FNAB and integrate FNAB into standard of care for breast cancer. Structure/Method/Design: Three main activities were completed during phase 2. Activity 1: Develop a Training of Trainers (ToT) curriculum for FNAB, and identify master trainers and trainees. Activity 2: Facilitate endorsement of a national approach to training FNAB and undertake preliminary rollout in La Libertad region. Activity 3: Strengthen interpretation and reporting of FNA results. Outcome & Evaluation: A 5-day ToT pilot course occurred during October 19-23, 2014. Prior to the course, women with palpable masses were identified during a breast screening campaign and scheduled for FNAB. Local, regional, national and international clinical teams observed, procured and interpreted FNAB. Four new cases of cancer and 8 cases of granulomatous mastitis clinically mimicking tumor were diagnosed. For evaluation, slide quality before and after training intervention will be compared. Following the course, a one-day validation meeting including physicians and national leaders took place at the national cancer center in Lima, Peru. FNAB was endorsed for early diagnosis and triage. Subsequently a senior leader/cancer surgeon and cytopathologist studied at UCSF for 2 days and 2 weeks respectively. Going Forward: To ensure sustainability and expansion of FNAB, cytology fellowships and training centers are needed. Strengthening partnerships will be critical for long-term capacity. An electronic reporting system will be developed to replace paper-based methods. Funding: Norwegian Cancer Society/Norwegian Breast Cancer Society (NCS/NBCS), Susan G. Komen for the Cure, The PATH Catalyst Fund.


Page 12


Detection and Biomarkers Poster #10 Data Integration to Identify Markers of Drug Response in Breast Cancer James T Webber, Nevan J Krogan, Sourav Bandyopadhyay UCSF

Presenting Author: James Webber Bioengineering and Therapeutic Sciences Bandyopadhyay Laboratory [email protected] Abstract Currently, treatment for breast cancer is determined by the use of clinical subtypes and a relatively small number of molecular markers. While these subtypes are an effective way to classify patients, a significant number of patients do not respond to the indicated treatment as expected. Moreover, many patients do not have any clear oncogenic target and cannot be effectively treated. Thus there is a need for more precise subtyping and for new biomarkers of drug sensitivity and response in breast cancer. In this project, we are integrating multiple data types from two different breast cancer datasets−a collection breast cancer tumors, and a panel of breast-cancer-derived cell lines. By finding gene clusters that are differentially active across both tumors and cell lines, and then correlating these clusters with drug response data, we will identify sets of genes that are both predictive of drug sensitivity and also commonly activated in clinical samples.


Page 13


Detection and Biomarkers Poster #11 Drug-induced kinome dynamics identifies new combination therapies Hayley Donnella (UCSF), James Webber (UCSF), John Gordan (UCSF), Rebecca Levin (UCSF), Nora Bayani (Lawrence Berkeley National Laboratories), James Korkola (OHSU), Kevan Shokat (UCSF), Sourav Bandyopadhyay (UCSF)

Presenting Author: Hayley Donnella Bioengineering and Therapeutic Sciences Bandyopadhyay Laboratory [email protected] Abstract In recent years, the advent of targeted therapy has revolutionized the field of cancer therapeutics. Protein kinases are arguably the most tractable candidates for therapeutic targeting in breast cancer, and many are well established in the clinic including HER2 (Herceptin™, Tykerb™, or pertuzumab) and EGFR (Erbitux™, Iressa™, or Tarceva™). However, molecularly targeted cancer therapies fail when tumor cells circumvent the action of these small-molecule inhibitors, facilitating the development of resistance despite the target kinase being present and active in the cell. Tumor cells have an extraordinarily complex adaptive rewiring response to drug treatment. This dynamic reprogramming can serve as a path to drug resistance by allowing the cell to modulate kinase activation, thus compensating for the loss of activity from the drug target. Here we have taken a systems-kinome approach to uncover these mechanisms of rewiring during innate resistance. Using multiplexed inhibitor beads coupled with mass spectrometry (MIB/MS), we enriched for and quantified the activated kinome across a panel of breast cancer cell lines in response to different kinase inhibitors targeting key pathways in breast cancer. Through label-free quantitation we can compare the kinase activity profiles of drug-sensitive and drug-resistant cell lines both before and after treatment. Analysis of responses reveals common protein changes associated with drug resistance and sensitivity. These kinome-response signatures can then be used in the rational design of combination therapies that block signaling pathways mediating drug resistance.


Page 14


Detection and Biomarkers Poster #12 Functional detection of phospho-circuits identifies the kinome vulnerabilities of BRAFV600E colorectal cancer Jean-Philippe Coppé, Changjun Wang, Diede Brunen, Anirudh Prahallad, Ana Ruiz-Saenz, Danislav Spassov, Bo Pan, Christina Yau, Denise Wolf, Christian Posch, Pei Rong Evelyn Lee, Miki Mori, Julia Malato, Donghui Wang, Paul Phojanakong, Zhongzhong Chen, Byron Hann, Lamorna Brown-Swigart, Mark Moasser, René Bernards, and Laura J. van ’t Veer

Presenting Author: Jean-Philippe Coppe Laboratory Medicine [email protected] Abstract BACKGROUND: Kinase inhibitors are used widely to treat various cancers, but intrinsic feedback loop mechanisms or adaptive reprogramming of kinase circuits often contribute to therapeutic resistance. Measuring the activity of kinase enzymes, and potentially monitoring the functionality of the entire human phospho-reactome at once, remains largely unexplored. To address this challenge, we developed a high throughput kinase-activity mapping (HT-KAM) assay that monitors the phospho-catalytic activity of multitude kinase enzymes. The assay successfully identified the phospho-signatures of BRAFV600E colorectal and thyroid cancer cells intrinsically resistant to BRAFV600E-targeted therapy (Vemurafenib), which represents a very-poor prognosis condition that affects an estimated 15,000 patients per year across America. METHODS: 228 phospho-sensing probes derived from biological target sites of kinase enzymes were developed into an ATP-consumption screen to identify the activity signatures of >50 kinases, including EGFR, MAPK, AKT, ABL and SRC family kinases. Functional phospho-signatures were derived from HT-KAM measurements testing cell extracts from cells treated or not with inhibitors. For in vitro cell survival screens, in vivo xenograft tumor modeling, and exogenous genetic alteration of SRC-regulated mechanisms, we used a series of colorectal and thyroid cancer cell lines treated with combinations of kinase-targeting agents (representing 29 kinase inhibitors, and 26 cell lines total). RESULTS: The phospho-catalytic fingerprint of Vemurafenib-resistant BRAFV600E colorectal cancer identified actionable kinases that regulate mechanisms of CRC resistance to Vemurafenib alone, and Vemurafenib + Cetuximab/Gefitinib combination therapy. SRC kinases emerged from phospho-circuits as conserved, central mediators of resistance to BRAFV600E-targeted therapy. SRC functioned in parallel to (and independently of) EGFR (or other RTKs, such as HER3). SRC kinases-targeting therapies Dasatinib or Saracatinib combined with Vemurafenib, and with or without Gefinitib or Lapatinib, had strong synergistic effects, and efficiently restored therapeutic response of BRAFV600E colorectal cancer and thyroid cancer cells, in vitro and in vivo. CONCLUSION: We successfully identified new feedback mechanisms regulating drug-sensitivity to single or combined targeted therapies. We defined a new, pivotal role for SRC kinases in regulating innate resistance to Vemurafenib treatment. Our high throughput kinase-activity mapping (HT-KAM) platform innovates the exploration of strategic, actionable kinase dependencies of tumors, and may provide insights to guide precision medicine.


Page 15


Detection and Biomarkers Poster #13 Investigation into the clinical significance and utility of common genetic variants in predicting the occurrence of bilateral breast cancer Karolina Plonowska BA1, Sarah Dilys Sawyer1 PhD, Carlie Thompson MD1, Beth Crawford MS2, and Laura Esserman MD MBA1,3,4 1 School of Medicine, University of California, San Francisco; 2 UCSF Cancer Risk Program; 3 UCSF Carol Franc Buck Breast Care Center; 4 Athena Breast Health Network.

Presenting Author: Karolina Plonowska Surgery [email protected] Abstract Women diagnosed with breast cancer commonly ask if they would benefit from a risk-reducing contralateral prophylactic mastectomy (CPM). In the US there has been a rapid increase in this surgery, particularly amongst young women with an early onset- and family history of breast cancer [1]. Currently, CPM is considered to only benefit a small proportion of women, like BRCA1 and BRCA2 mutation carriers. However, common genomic variants (SNPs) that are associated with breast cancer risk may identify an additional group of women that may benefit from CPM. We previously showed 22 SNPs could identify women with breast cancer at risk of a second primary breast cancer [2]. We hypothesize that common variant genetic testing could identify an additional group of women at high- or low-risk of developing bilateral breast cancer. Hence, we examined 75 SNPs in population controls, and women diagnosed with unilateral- and bilateral breast cancer. Methods: 99 unilateral breast cancer (UBC) and 97 bilateral breast cancer (BBC) cases were recruited from the Cancer Risk Program at UCSF. 427 controls that had no breast cancer were recruited from Athena Breast Health Network. Participants underwent genetic testing for 75 SNPs, as well as 19 genes that have been associated with breast cancer risk. Results: 7.1% (7/99) UBC and 7.2% (7/97) BBC had pathogenic mutations detected in high-to-moderate risk breast cancer genes and were excluded from further analysis. For 90 UBC and 92 BBC, a polygenic risk score (PRS) based on 75 SNPs was calculated as described by Sawyer et al [2]. UBC and BBC had a significantly higher mean PRS than controls (UBC PRS = 0.22, p=0.0014; BBC PRS =0.36, p<1x10-7). The mean PRS was higher in BBC than UBC but, likely due to low power, the difference is not statistically significance (0.36 vs 0.22, P=0.125). Conclusions: This study confirms that SNPs are associated with higher risk of breast cancer. To assess whether SNPs predict bilateral breast cancer risk, a larger sample set (n=500 each). If there is a clinically meaningful difference in bilateral risk, these results could be used in a shared decision-making tool by women to make more informed CPM decisions.


Page 16


Detection and Biomarkers Poster #14 Common genetic variants associated with breast cancer risk used in the Athena study to enhance models identifying women for breast cancer chemoprevention Sarah Theiner(1), Sarah D. Sawyer(1), Paige Kendall(1), Alexandra S. Perry(1), Denise Wolf(1), Scott Huntsman(1), Bo Pan(2), Jeffery A. Tice(1), David A. Pearce(3), Thomas Cink(3), Laura Esserman(1), Elad Ziv(1), Laura van ‘t Veer(1) (1)University of California, San Francisco, San Francisco, CA; (2)Peking Union Medical College Hospital, Beijing, China; (3)Sanford Health, Sioux Falls, SD

Presenting Author: Sarah Theiner Laboratory Medicine van 't Veer Laboratory [email protected] Abstract Background: The U.S. Preventive Services Task Force recommends that women with a >3% five-year risk of developing breast cancer consider taking selective estrogen receptor modifiers (SERMs) or aromatase inhibitors (AIs) to reduce their risk. Polygenic risk score (PRS), calculated by adding the individual breast cancer risk association for each common genetic variant (SNP), has been found to predict women at low- to high-risk of breast cancer. We analyze associations between SNP risk alleles and known breast cancer risk factors (ethnicity, family history of breast cancer and number of biopsies); furthermore, we quantify the likely impact on chemoprevention recommendations by adding the PRS to known risk models in a subset of women participating in the University of California 100,000 women Athena Breast Health Network. Methods: Our research cohort included 838 women with no previous diagnosis of breast cancer from the University of California, San Francisco, and was enriched for women determined to be at elevated risk for developing breast cancer by the Gail model. A panel of 75 breast cancer risk SNPs were evaluated on saliva and blood samples (Akesogen Inc; COGS oncochip array). The PRS for each patient was calculated by converting the odds ratio for each SNP into a likelihood ratio (LR) and combining LR’s across SNPs. Breast Cancer Surveillance Consortium (BCSC), Gail, BCSC-PRS and Gail-PRS scores (risk models incorporating PRS within a Bayesian framework), were evaluated for each patient. Associations between variables were assessed using t-test or ANOVA. A threshold of p<0.05 was used to assess significance. Results: Women in this study carry an average of 65 risk allele SNPs (of 150, 2 per locus). By ANOVA, there is a statistically significant association between the SNPs risk allele count and ethnicity (p=0.014), with a trend towards association with a family history of a first-degree relative with breast cancer (p=0.053). PRS is significantly associated with a family history breast cancer (p=0.031); neither SNP allele count nor PRS associates with previous biopsy status. We found by adding PRS that 12% (86/707) and 13% (104/776) of patients with a prior BCSC or Gail score <3% five-year risk, respectively, changed classifications and would be eligible for chemoprevention. Conversely, 37% (36/98) and 36% (22/62) of patients with a BCSC or Gail score of


Page 17


>3% five-year risk, respectively, changed classifications by adding PRS and would no longer be eligible for chemoprevention. Conclusion: The addition of SNP based PRS to BCSC and Gail models significantly changes how women are classified and as a result changes whether risk reducing agents are recommended. PRS will be combined with BCSC and genetic test results for 9 breast cancer genes to calculate a women’s breast cancer risk in the PCORI-funded Athena WISDOM study of 100,000 women, comparing risk-based vs. annual mammography screening.


Page 18


Detection and Biomarkers Poster #15 Receptor activator of nuclear factor kappa B (RANK) expression in primary breast cancer correlates with recurrence free survival and development of bone metastases in the I-SPY 1 trial (CALGB 150007/150012; ACRIN 6657) Neelima Vidula, Christina Yau, Jiali Li, and Hope S. Rugo University of California, San Francisco

Presenting Author: Neelima Vidula Hematology/Oncology Breast Oncology [email protected] Abstract Background: The Receptor Activator of Nuclear Factor Kappa B (RANK)/RANK ligand (RANKL)/Osteoprotegerin (OPG) axis may be involved in the development of bone metastases. We studied gene expression in this pathway in primary breast cancer (BC) to determine correlations with clinical characteristics and outcomes in the neoadjuvant I-SPY 1 study. Methods: We evaluated RANK/RANKL/OPG expression with gene microarray studies in I-SPY 1 (n=149), using the external GSE25066 (n=425) dataset for validation. Associations with clinical features were determined with the t-test and ANOVA. An optimal biomarker cut point to dichotomize recurrence free survival (RFS) curves was determined using I-SPY 1, and then the association between dichotomized biomarkers and RFS was evaluated with the multivariate Cox proportional hazard model. Associations with site specific relapse were determined with the t-test and multivariate logistic regression. Pearson correlations with a significance threshold of p <0.05 with a filter to identify genes with similar concordance in both datasets were used to identify external genes associated with RANK. Results: In I-SPY 1, at 3.5 years, 41 patients developed recurrent disease. 12 patients had any bone metastases (BM) and 22 had non bone metastases (NBM). RANK was significantly higher in hormone receptor (HR) negative vs. HR positive (p=0.027), basal vs. non basal (p=0.004), and pathologic complete response (p=0.038) tumors; the associations with HR negative and basal BC were also significant in GSE25066. In both datasets, higher RANK associated with significantly worse RFS (I-SPY 1: p=0.045, GSE25066: p=0.044). In I-SPY 1, higher RANK expression significantly correlated with BM vs. NBM (p=0.045), even on HR status adjustment (p=0.035). 9 external genes had positive concordance with RANK in both datasets. Conclusion: RANK is higher in HR negative and basal BC, and correlates with worse RFS and BM. Targeting this pathway may improve outcomes.


Page 19


Detection and Biomarkers Poster #16 Analysis of primary breast cancer (BC) expression of programmed cell death 1 (PD-1) receptor and programmed death ligand 1 (PD-L1) to determine associations with clinical characteristics and outcomes Neelima Vidula, Christina Yau, Andrei Goga, Hope S. Rugo University of California, San Francisco

Presenting Author: Neelima Vidula Hematology/Oncology Breast Oncology [email protected] Abstract Background: Antitumor immunity may be dampened by the interaction of the PD-1 receptor on tumor infiltrating lymphocytes and PD-L1 on tumor cells. Two recent phase I trials have suggested efficacy of anti-PD-1/PD-L1 antibodies in triple negative (TN) BC. In this study, we investigated associations between primary BC PD-1 and PD-L1 expression, clinical characteristics, and patient outcomes in publically available databases. Methods: We evaluated PD-1 and PD-L1 expression using microarray data from the neoadjuvant I-SPY 1 study (n = 149). Associations with clinical features and chemotherapy response were assessed by Kruskal-Wallis and Wilcoxon rank sum tests, respectively. Recurrence free survival (RFS) associations were assessed by the Cox proportional hazard model. Pearson correlations between PD-1 and expression of PD-L1, HAVCR2, STAT5A, FOXP3, MYC, and ESR1 were determined in I-SPY 1 and 2 other datasets: METABRIC (n = 1992) and TCGA (n = 817). Results: In I-SPY 1, PD-1 expression was significantly higher in HER2+ and TNBC (p = 0.003), and in grade 2/3 tumors (p = 0.043); this association was also seen in METABRIC. PD-1 expression was associated with pathologic complete response (p = 0.006), but this correlation did not remain significant on subtype correction. PD-1 was not associated with tumor stage, nodal status, lymphovascular invasion or RFS. While PD-L1 did not correlate with tumor features, patients with PD-L1 expression in the lowest quintile had worse RFS, even after subtype adjustment (HR 2.33, p = 0.01). In all 3 datasets, PD-1 significantly correlated with PD-L1, HAVCR2, and STAT5A, and inversely with ESR1. In the TN subset of TCGA and METABRIC, PD-1 significantly correlated with PD-L1, HAVCR2, and STAT5A. In TCGA and METABRIC, PD-L1 significantly correlated with HAVCR2 and STAT5A, and this was also seen in the TN subset. In TCGA alone, PD-1 and PD-L1 significantly correlated with FOXP3, and PD-1 with MYC. Conclusions: PD-1 expression is higher in TN and other aggressive BC subtypes. PD-1 and PD-L1 correlate with immune related genes HAVCR2 and STAT5A. Low PD-L1 expression may be an adverse prognostic factor. Trials are underway to investigate the activity of anti-PD-1/PD-L1 antibodies in TNBC and to elucidate markers of response.


Page 20


Detection and Biomarkers Poster #17 Single-Cell Analysis Characterizes Cellular Heterogeneity in Mammary Cancer Development Elizabeth Willey (UCSF), Kai Kessenbrock (UC Irvine), Devon Lawson (UC Irvine), Zena Werb (UCSF)

Presenting Author: Elizabeth Willey Anatomy Werb Laboratory [email protected] Abstract Breast cancer is the most commonly diagnosed form of cancer and remains the leading cause of cancer deaths in women worldwide. Despite recent major advances in understanding molecular and genetic factors driving the initiation and progression of breast cancer, prognosis still remains poor. Normal breast tissue is comprised of ducts and lobules of epithelial cells (basal and luminal) nestled among adipose-rich stromal tissue. Breast cancer develops when these epithelial cells acquire pro-tumorigenic mutations that lead to uncontrolled cellular proliferation and subsequent loss of normal tissue structure. Over the past few decades, we have significantly advanced our understanding of the genetic mutations that drive tumorigenesis and how signaling circuits within the cell are consequently altered. However, it remains unclear how these mutations affect the behavior of breast epithelial cell populations during the initial steps of tumor formation. By investigating the heterogeneity of gene expression in normal breast tissue, we can compare it to early neoplastic and cancerous specimens, allowing us to identify those cells that are potentially preferentially tumorigenic. To show the spectrum of cellular heterogeneity at the single-cell level, we isolated both basal and luminal mammary epithelial cells (MECs) from reduction mammoplasty samples by Fluorescence Activated Cell Sorting (FACS) and analyzed them on the Fluidigm® C1 platform. After sequencing 396 cDNA libraries from single MECs and analyzing the datasets using several statistical tools, we were able to identify 5-7 distinct subpopulations within the basal cell cluster and between 2 and 5 subpopulations within the luminal cell cluster. Our analysis revealed hundreds of specific cell surface markers for these newly defined mammary subpopulations, which were further substantiated using immunofluorescent staining, flow cytometry, and single-cell western blot. Our work provides novel insights into the heterogeneity of the human mammary epithelial system under normal tissue homeostasis, which will form the basis for ongoing experiments. Understanding the early stages of breast tumorigenesis in high single-cell resolution will lead to the identification of novel biomarkers for early cancer detection and may lay the groundwork for developing new therapeutic approaches to prevent breast cancer from progressing into a life threatening condition for patients around the world. Funding: This study was supported by the National Institute of Health (NIH/NCI K99/R00-CA181490 to K.K. and 5R01CA129523-05 to Z.W.).


Page 21


Detection and Biomarkers Poster #18 A comprehensive approach to molecular profiling of circulating tumor cells from metastatic breast cancer patients Mark Jesus M. Magbanua, Louai Hauranieh, Ritu Roy, Praveen Pendyala, Eduardo V. Sosa, Janet H. Scott, Jin Sun Lee, Andrea Ordonez Coronel, Bao Ho, Tulasi Solanki, Hope S. Rugo and John W. Park Affiliation of authors: Division of Hematology/Oncology, Helen Diller Family Comprehensive Cancer Center, UCSF; Helen Diller Family Comprehensive Cancer Center Computational Biology Core (RR), University of California San Francisco

Presenting Author: Mark Jesus Magbanua Hematology Oncology John Park Laboratory [email protected] Abstract Background: Circulating tumor cells (CTCs) in blood offer a relatively non-invasive source of metastatic tumor material for molecular characterization. We isolated and profiled CTCs from metastatic breast cancer patients to elucidate underlying biology of blood-borne metastasis. Methods: CTCs were isolated via immunomagnetic enrichment followed by fluorescence-activated cell sorting (IE/FACS). Expression of 64 cancer-related genes was analyzed in 105 patients using microfluidic qRT-PCR. In 49 of the 105 patients, parallel genome-wide copy number analysis was performed on CTCs isolated from the same enriched blood samples. Whole genome and whole exome next generation sequencing was performed on CTCs from an index patient. Results: Transcriptional profiling of CTCs revealed up-regulation of several genes including EPCAM and CCND1 and down-regulation of hematopoietic-related genes including PTPRC (CD45) and CD68. Cluster analysis of expression data revealed three groups corresponding to basal-like, normal-like, and luminal subtypes. Cell-to-cell heterogeneity was observed when CTCs were analyzed at the single cell level. Serial analysis displayed changes in gene expression over time. CTCs exhibited numerous genomic aberrations including 1q/8q gains and 8p/16q losses, consistent with breast cancer origin. Copy number profiles grouped into three major clusters: CTCs exhibiting low genomic instability, 8q gain, and 1q gain/11q loss. Overexpression of CCND1 and ERBB2 was associated with CTC clusters exhibiting high genomic instability. Changes in ER (n=102) and HER2 (n=130) status between CTCs and matched primary tumors were observed in 27% and 23% of the patients, respectively. Next-generation whole genome (38x) and exome (140x) sequencing analysis of CTCs from an index patient diagnosed with invasive lobular carcinoma discovered a mutation in E-cadherin (CDH1), a gene commonly altered in this histological subtype. Conclusions: Comprehensive molecular characterization of CTCs provided novel insights into the biology of these cells. CTC profiling may open avenues for discovery of biomarkers for personalized treatment.


Page 22


Detection and Biomarkers Poster #19 Science Advocacy Exchange as a Cancer Center Priority: From Rhetoric to Reality Susan Samson1 Alicia Y. Zhou1*, Maria Martins 1*, Alexandra Corella1, Dai Horiuchi1, Christina Yau1, Taha Rakshandehroo1, John Gordan1, Rebecca Levin1, Jeff Johnson1, John Jascur1, Mike Shales1, Antonio Sorrentino1, Jaime Cheah2, Paul Clemens2,Alykhan Shamji2, Stuart Schreiber2,3, Nevan Krogan1, Kevan Shokat1, Frank McCormick1, Sourav Bandyopadhyay2 Andrei Goga1 1 University of California San Francisco; 2 The Broad Institute of Harvard and MIT; 3 Howard Hughes Medical Institute

Presenting Author: Susan Samson UCSF BOP, Breast Science Advocacy Core [email protected] Abstract BACKGROUND Collaborative team science provides a starting point for comprehensive change, and advocates have a unique and important role developing and engaging in transdisciplinary collaboratives that focus on new questions and new possibilities to advance the science of ethnic and medically underserved health care disparities. Participating in four areas: 1) research and programmatic support, 2) education and outreach, 3) policy and strategy, and 4) representation and advisory, the UCSF Breast Science Advocacy Core (BSAC) Program, a volunteer affiliate of the Breast Oncology Program (BOP), one of ten multidisciplinary research programs under the umbrella of the UCSF Helen Diller Comprehensive Cancer Center promotes a transformative, transdisciplinary, integrated environment to study the biological basis of the diseases that comprise breast cancer; to define the risk of developing or progressing with specific types of breast cancer; to develop novel interventions that work locally and globally to reduce morbidity and mortality from breast cancer and its treatment; and to leverage new collaborative research, education, and mentoring/training opportunities that address cancer outcome disparities. STUDY OBJECTIVES Genomic analyses of patient tumors have unearthed an overwhelming number of recurrent somatic alterations in genes that have dramatic effects on tumor biology, patient drug responses, and clinical outcomes. In one study, high grade triple negative breast cancer (TNBC) accounts for 34% of breast cancers in African American women versus 21% in white women. A growing body of evidence has shown that African American women have biologically more aggressive disease, independent of social determinants, and suffer the highest mortality rates. While biological breakthroughs of the last decade have greatly advanced our understanding of cancer, in advanced TNBC, a poor prognosis subtype, there is an urgent need to translate this evolving patient genomic data into new therapeutic paradigms. Our study focuses on the intersection of synthetic lethal approaches, MYC driven human cancers, and immunotherapy as an “innovation agenda”. A distinct MYC vision highlights how overexpression is associated with aggressive outcomes and poor patient outcomes, and synthetic lethal strategies to


Page 23


target MYC (CDK inhibitors, PIM2, as well as the PDI immune pathways) have potential for addressing outcome disparities In African American Women with Triple Negative Breast Cancer (TNBC). KEY FINDINGS We have developed a screening technique that can be used to rapidly and accurately identify potential synthetic lethal interactions in TNBC. This platform utilizes an isogenic cell line system that we have developed to model oncogene activation in TNBC. A growing body of evidence has shown that • Quantitative approach maps genotype-specific drug responses in isogenic cells • Systematic discovery of biomarkers for cancer drugs under clinical investigation • Clinically actionable synthetic lethal interaction between MYC and dasatinib is discovered • Mechanism of dasatinib action through inhibition of LYN kinase is described KEY TAKE-AWAY MESSAGE The inclusion of advocates in convergent science settings remind academic stakeholders that research is there to benefit the patient as they attempt to spark innovation, democratize science, and support smarter interventions that expedite the incredible potential of future investments in bioscience within disparities arenas.


Page 24


Detection and Biomarkers Poster #20 Co-Expression Modules Identified From Published Immune Signatures Reveals Five Distinct Immune Subtypes in Breast Cancer Dominic Amara BA, Denise Wolf PhD, Laura van’t Veer PhD, Laura Esserman MD MBA, Michael Campbell PhD, Christina Yau PhD UCSF Breast Care Center

Presenting Author: Dominic Amara Surgery [email protected] Abstract Introduction Immune modulating therapies offer an attractive novel approach in the treatment of breast cancer. There is a growing body of literature demonstrating that immune-related expression signatures predict breast cancer prognosis and chemo- and/or targeted therapy responsiveness. However, it is unclear how these signatures relate to one another. Here we evaluated 57 immune signatures in breast cancer and generated co-expression modules to classify patients into immune subtypes. Methods We evaluated 57 published gene expression signatures related to immune function in 5 public breast cancer gene expression datasets (TCGA (n=817), METABRIC (n=1992), EMC344 (n=344), pooled triple negative set: GSE31519 (n=579), pooled neoadjuvant chemotherapy-treated set: GSE25066 (n=508)). For each dataset, consensus clustering was used to subset the signatures based on their co-expression pattern. Signatures that were in the same consensus cluster across all 5 datasets were used to define immune modules; and module scores were computed as the average across their constituent signatures. Patients were then classified into immune subtypes based on their module scores using consensus clustering. Overall survival (OS) differences between immune subtypes were assessed using Cox proportional hazard modeling in basal breast cancers from the METABRIC dataset (n=329). Results Consensus clustering of the 57 expression signatures consistently yields four distinct co-expression modules across the five datasets. These modules appear to represent distinct immune components and signals, with constituent signatures relating to: 1) T-cells and/or B-cells (T/B-cell), 2) interferon (IFN), 3) transforming growth factor beta (TGFB), 4) core serum response, dendritic cells and/or macrophages (CSR). Of note, the T/B-cell module contains 20 of the 57 signatures evaluated; and the CSR module is highly correlated to proliferation (r=0.81). Subtyping of patients based on these co-expression modules consistently yields subsets that fall into five major immune subtypes. These immune subtypes are associated with differences in overall survival in the METABRIC basal breast cancer cases, where the CSR High subtype has the worst outcome (10-year OS: 23%). In comparison, the subsets corresponding to the T/B-cell/IFN High subtype have better outcome (Hazard ratio: 0.43, p = 0.018). In contrast, no significant outcome differences were observed between the poor outcome CSR-High subtype and the remaining three immune subtypes (p>0.05).


Page 25


Conclusion Our exploratory study identified four distinct immune co-expression modules (T/B-cell, IFN, TGFB, or CSR) from a collection of published immune signatures. Using these modules, we identified 5 immune subtypes with prognostic significance in basal breast cancers. We propose to test representative signatures from the 4 modules and the combined immune subtypes as predictive biomarkers of response to immunotherapies.


Page 26


Detection and Biomarkers Poster #21 Breast Cancer Screening in the Precision Medicine Era 1Carlie Thompson, 1Allison Stover Fiscalini, 1Patricia Donnellan, 1Celia Kaplan, 2Lisa Madlensky, 3Martin Eklund, 1Elad Ziv, 1Laura van’ t Veer, 1Jeffrey Tice, Athena Breast Health Network investigators, 1Laura J Esserman 1University of California, San Francisco, 2University of California San Diego, 3Karolinska Institutet

Presenting Author: Allison Stover Fiscalini Surgery [email protected] Abstract We are entering the era of precision medicine in which cancer screening, prevention and treatment will be tailored to each individual. The progress made in this field is due, in part, to advances in our understanding of cancer risk and tumor biology. The challenge before us is to harness this knowledge and apply it in the clinical setting. Breast cancer screening provides an excellent opportunity to test the value of precision medicine in the real world. In this report we describe the process of designing a model of personalized breast cancer screening. Methods Risk factors were selected that have the greatest impact, have been validated and can be measured across a population. A risk model was selected that is highly calibrated, has been validated in a large screening cohort and is easy to apply in a large population of women. An expert committee was convened that set risk thresholds for stratifying women into groups that will be recommended to undergo biennial, annual or every six month screening. Risk thresholds and screening schedules are in accordance with the United States Preventive Services Task Force breast cancer screening recommendations. Results Risk factors: Age, race/ethnicity, personal history of breast biopsies and benign breast disease, family history, breast density and breast cancer-associated genetic mutations and single nucleotide polymorphisms (SNPs) were chosen as the risk factors that will be used to determine breast cancer risk. Risk model: The Breast Cancer Surveillance Consortium risk model will be used to calculate a woman's 5-year risk and will be modified by a polygenic risk score based on 81 SNPs. Risk thresholds: Women will be recommended to undergo biennial screening mammography when they reach the age of 50 or have the risk of an average 50 year-old woman (1.3% 5-year risk). Women will be advised to undergo annual screening if they are at increased risk of developing an interval cancer (women in their forties with extremely dense breasts and women at increased risk of developing estrogen receptor negative breast cancer based on their SNPs). Women will be recommended to undergo annual mammography and annual MRI if they are found to be gene mutation positive, have the risk of a BRCA1 mutation carrier (6% 5-year risk) or have a history of mantle radiation.


Page 27


Discussion Selecting the appropriate risk factors and risk model and determining risk thresholds are key components of designing a personalized breast cancer screening model. Personalized screening may be the way forward, but this can only be determined within the setting of a randomized controlled trial. We will conduct such a trial to determine if personalized screening is as safe as, less morbid than, more preferred by women than and enables prevention when compared to annual screening. The WISDOM (Women Informed to Screen Depending on Measures of risk) study will compare risk-based screening to annual screening within the Athena Breast Health Network with support from the Patient-Centered Outcomes Research Institute. Our intent is that this trial will provide us with the data that we need to determine the safest and most effective way to screen women for breast cancer in the era of precision medicine.


Page 28


Epidemiology and Population Sciences Poster #1 Lack of Payer Coverage for Multi-Gene Panels that include BRCA1/2 Tests Elizabeth Clain Albert Einstein School of Medicine Michael Douglas, Julia Trosman, and Kathryn Phillips UCSF Christine Weldon Center for Business Models in Healthcare

Presenting Author: Michael Douglas, co-author, UCSF Clin Pharm HDFCC, IHPS [email protected] Abstract In recent years, genetic testing for heritable cancer syndromes - such as BRCA1/2 testing for hereditary breast and ovarian cancer - has been shifting from single-gene analysis to multi-gene panels, typically using next-generation sequencing technologies. UCSF Center for Translational and Policy Research on Personalized Medicine (TRANSPERS) collaborators conducted the first review of the BRCA1/2 testing landscape since the historic 2013 Supreme Court decision that allowed the entry of new testing providers We found that the number of BRCA1/2-only tests and panels including BRCA1/2 has increased since June 2013, and average prices have decreased. Although payers cover single gene BRCA1/2 single-gene testing for high risk individuals, many payers consider panels investigational or experimental and thus do not cover them. Although 76% of payers have coverage policies about panels, none of these policies provides positive coverage. Of payers with policies on panels, most (77%) consider panels investigational or experimental, and the remainder limits coverage to those panels on which all the genes are considered medically necessary. The experience with panels including BRCA1/2 may be instructive in understanding the evolution of payer coverage policies as other multi-gene panels become more commonly used. TRANSPERS is currently developing a Payer Coverage Policy Registry that systematically synthesizes payer coverage policies on multi-gene panels in order to be able to assess when payers cover panels, how panels are included in payer policies, and how this is changing over time. Our Registry is unique in its focus on coverage policies, inclusion of a wide range of key factors that influence coverage, and objective analyses by a non-profit academic institution


Page 29


Epidemiology and Population Sciences Poster #2 Longitudinal analysis of breast cancer recurrence and mortality among U.S. Latina women with breast cancer in the Pathways Study. Natalie J. Engmann, MSc (UCSF), Isaac J. Ergas, MPH (Kaiser Permanente), Song Yao, PhD (Roswell Park Cancer Institute), Marilyn L. Kwan, PhD (Kaiser Permanente), Janise M. Roh, MPH, MSW (Kaiser Permanente), Christine B. Ambrosone, PhD (Roswell Park Cancer Institute), Lawrence H. Kushi, ScD (Kaiser Permanente), Laura Fejerman, PhD (UCSF)

Presenting Author: Natalie Engmann Department of Epidemiology & Biostatistics [email protected]; [email protected] Abstract Although Latina women in the United States have a lower incidence of breast cancer than non-Latina White women, they have poorer survival. The U.S. Latina population is heterogeneous, and research has suggested that genetic admixture may in part account for poor breast cancer survival among Latina women independent of tumor characteristics and socioeconomic variables. We evaluated the association of genetic ancestry with breast cancer recurrence, and breast cancer-specific and all-cause mortality among 506 Latina women diagnosed with invasive breast cancer participating in the Pathways Study. The Pathways Study is a prospective cohort following women diagnosed with breast cancer in Kaiser Permanente Northern California. European, African and Indigenous American ancestry proportions were estimated with a panel of 118 ancestry informative markers and the program ADMIXTURE. Proportional hazards models were used to assess the effect of genetic ancestry on breast cancer recurrence (53 events), breast cancer-specific mortality (31 events) and all-cause mortality (54 events), with a mean follow-up time of 6 years. Indigenous American ancestry was not associated with breast cancer recurrence (HR=0.99 per 10% increase in Indigenous American ancestry, 95% CI: 0.86, 1.16, p=0.99), breast cancer mortality (HR=0.95, 95% CI: 0.77, 1.17, p=0.65), or all-cause mortality (HR=0.91, 95% CI: 0.77, 1.07, p=0.25). Adjustment for markers of socioeconomic status, clinical characteristics and treatment modalities did not alter the associations. Although this cohort of Latina women with breast cancer is not large, results suggest that previously observed differences in survival by genetic ancestry may be explained by disparities in healthcare access or quality of service in the general population of Latina women from Northern California, whereas the present analysis was restricted to women enrolled in Kaiser Permanente in which access to care is largely uniform.


Page 30


Epidemiology and Population Sciences Poster #3 Developing a Breast Fine Needle Aspiration Biopsy Service in Peru Dianna Ng, Britt-Marie Ljung, Richard Bardales, Milagros Abad, Manuel Cedano, Tara Hayes Constant, Jennifer L. Winkler, Ronald Balassanian

Presenting Author: Ronald Balassanian Pathology Cytopathology [email protected] Abstract Program/Project Purpose: Breast cancer incidence is increasing in developing countries. In 2010, 57.8% of breast cancers in Peru were at least stage 3 at diagnosis. Federal and regional Peruvian institutions partnered with several organizations, including UCSF, to improve access and quality of breast cancer care in Peru. Phase 1 involved screening with clinical breast exam, creating a community-based referral system for women with masses, and promoting evaluation of lesions using fine needle aspiration biopsy (FNAB) by trained physicians, with treatment at the regional cancer institute. Curricula for each activity were developed and validated. During phase 1, in-country FNAB expertise was recognized as limited. For phase 2, the critical objective was to solidify local capacity for high-quality FNAB and integrate FNAB into standard of care for breast cancer. Structure/Method/Design: Three main activities were completed during phase 2. 1) Develop a Training of Trainers (ToT) curriculum for FNAB, and identify master trainers and trainees. 2) Facilitate endorsement of a national approach to training FNAB and undertake preliminary rollout in La Libertad region. 3) Strengthen interpretation and reporting of FNA results. Outcome & Evaluation: A 5-day ToT pilot course occurred during October 19-23, 2014. Prior to the course, women with palpable masses were identified during a breast screening campaign and scheduled for FNAB. Regional, national and international clinical teams observed, procured and interpreted FNAB. Four new cases of cancer and 8 cases of granulomatous mastitis were diagnosed. Slide quality before and after training intervention will be compared. Following the course, a one-day validation meeting including physicians and national leaders took place at the national cancer center in Lima, Peru. FNAB was endorsed for early diagnosis and triage. Going Forward: To ensure sustainability of FNAB, cytology fellowships and training centers are needed. Strengthening partnerships will be critical for long-term capacity. An electronic reporting system will be developed to replace paper-based methods.


Page 31


Epidemiology and Population Sciences Poster #4 Evaluating the frequency of supportive care referrals generated using patient reported data from the Athena Health Questionnaire System Emily C. Wong, BA1; Celia P. Kaplan, PhD1; Dianne M. Shumay, PhD1; Yan Leykin, PhD1; Kim A. Etzel, BS1; Allison Stover Fiscalini, MPH1; Laura J. van ‘t Veer, PhD1; Laura J. Esserman, MD, MBA1; Michelle E. Melisko, MD1 1 = University of California, San Francisco

Presenting Author: Emily Wong Surgery van 't Veer Laboratory/Breast Cancer Survivorship [email protected] Abstract Background Patients at risk for or diagnosed with breast cancer have many symptoms and need for supportive care services. As part of the Athena Breast Health Network (a University of California-wide collaboration), the UCSF Breast Care Center (BCC) has incorporated an electronic health questionnaire system (HQS) prior to new patient and follow-up clinic visits, allowing patients to provide information on their personal health and family history, physical and psychological symptoms, and lifestyle. Based on these patient-reported outcomes (PRO), automated referrals for services including genetic counseling, psycho-oncology, social work, fertility preservation, and smoking cessation are generated. Algorithms defining thresholds to trigger these referrals were developed by clinicians and supportive care providers to proactively meet patients’ needs. Objectives To evaluate the incidence and outcomes of supportive care referrals based on existing algorithms, and identify reasons for non-utilization of the services offered. The ultimate goal for this evaluation is to modify the existing algorithms to better meet patients’ needs. Methods Patients initiating care at the UCSF BCC are invited by email to complete an HQS that provides information relevant to their clinical care. Patients sign an electronic consent, agreeing to have their PRO stored and accessed for research purposes. Family history, health behaviors, desired services, and responses to National Cancer Institute Patient Reported Outcomes Measurement Information System (PROMIS) items are processed through algorithms, generating referrals based on defined thresholds. A clinician summary report is generated and scanned into the electronic medical record (EMR), identifying services for which the patient has met thresholds. Referrals are sent to the clinician as pended orders through the EMR. Once signed by the care provider (physician or nurse practitioner), the order is routed through the EMR to the appropriate service and the patient is offered a visit or phone consultation when appropriate.


Page 32


Results Between 1/1/14 and 12/31/14, 1297 patients initiating care at the UCSF BCC completed an HQS prior to their clinic visit. 1108 patients (85.4%) agreed to have their data used for research. 623 patients (56.2%) were referred to at least one supportive care service. Ongoing analyses are underway to determine the percentage of patients who received services, explore barriers to accessing these services, and evaluate patients’ preferences regarding provision of services in alternate formats, including webinars, online content, and group sessions. Conclusions Effective use of PRO identifies a high percentage of patients in need of supportive care services. Through analysis of utilization of services based on our existing thresholds, we hope to optimize our algorithms to better serve our patients’ needs throughout the continuum of cancer care.


Page 33


Epidemiology and Population Sciences Poster #5 Patient utilization of supportive care services identified through the Athena Breast Health Questionnaire System Emily C. Wong, Dianne M. Shumay, Celia P. Kaplan, Yan Leykin, Laura J. van ‘t Veer, Allison Stover Fiscalini, Laura J. Esserman, Michelle E. Melisko, University of California, San Francisco

Presenting Author: Emily Wong Surgery van 't Veer Laboratory/ Breast Cancer Survivorship [email protected] Abstract Background Psychological wellbeing and lifestyle changes are important factors in long-term health of cancer survivors. As part of the Athena Breast Health Network, the UCSF Breast Care Center (BCC) incorporated an electronic health questionnaire system (HQS) that collects patient-reported data on physical and psychological symptoms, medical conditions, family history, and lifestyle to identify patients’ supportive care needs. Methods Women with appointments at the UCSF Breast Care Center (BCC) completed an electronic HQS. Self-reported family history, personal health, and health behaviors were used to generate referrals to supportive care services based on defined thresholds. Women consenting to be contacted for future research studies were invited to complete an additional survey assessing (1) Utilization of supportive care referrals made for them, (2) usefulness of these supportive care services, and (3) interest in other modalities for providing supportive care resources. Results From 11/1/14-5/31/15, 425 women seen at the BCC completed an HQS and agreed to have their data used for research. 141 women (33.2%) completed the additional survey and were included in this analysis, 101 of whom had confirmed breast cancer diagnoses. 28% of patients referred for psychological services completed appointments with psychological services, and those who did not complete appointments explained that they were already seeing another mental health provider. At the time of responding to the survey, participants were most interested in learning about nutrition, integrative approaches to healthcare, exercise, and hormonal therapy, and preferred to receive information about these additional services via in-person individual sessions. Conclusions HQS can assist in identifying patient supportive care needs, but despite automation of referrals, many patients do not follow through to use supportive care and clinical resources. Timing and resource accessibility may be key. Through analysis of reasons for non-utilization, we hope to better tailor the delivery of supportive care services, when indicated, to better serve patient needs.


Page 34


Molecular Biology Poster #1 High dimensional analysis of age-related phenotypic diversity in human mammary epithelial cells Fanny A. Pelissier1,2, Denis Schapiro3, Hang Chang2, Bernd Bodenmiller3, Mark A. LaBarge2 and James B. Lorens1

1 Department of Biomedicine, University of Bergen, Bergen N-5009, Norway 2 Life Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA 3 Institute of Molecular Life Sciences, University of Zürich, CH-8057 Zürich

Presenting Author: Fanny Pelissier University of Bergen LBNL [email protected] Abstract The accumulation of defective stem or progenitor cells may be common in aging tissues, and a skewed differentiation potential may increase susceptibility to oncogenic events. The incidence of luminal breast tumors increases with age, whereas the incidence of basal-like tumors decreases and there does not appear a strong age-dependent association with particular mutations does not seem to exist. However, the epigenetic landscape of human mammary epithelial cells changes with age. The functional heterogeneity within a tissue reflects the phenotypic divergence between different cell states. Hence quantifying age-dependent phenotypic heterogeneity can reveal subpopulations of cells that accumulate with age. We applied high-dimensional (30-parameter) mass cytometry single cell analysis to profile the phenotype of human mammary epithelial cells from forty-four women from 16 to 91 years old. Computational analysis using unsupervised population partitioning identified clusters of a specific subset of luminal cells that acquired a more basal phenotype and accumulate with age. A regularized classification model robustly predicted the age-group of each sample based on alterations in phenotypic diversity. Finally, age-related signatures were present detectable at the progenitor cell level. This study presents a unique in-depth perspective of age-related phenotypic diversity of human mammary epithelial cells and pinpoints a specific cell luminal subpopulation that may be an origin of age-related breast cancer.


Page 35


Molecular Biology Poster #2 Age-dependent gene expression patterns in human luminal epithelial cells are established through microenvironment-imposed epigenetic modifications. Masaru Miyano, Marcus Stoiber, Martha R Stampfer, James B Brown and Mark A LaBarge Lawrence Berkeley National Laboratory

Presenting Author: Masaru Miyano Lawrence Berkeley National Laboratory [email protected] Abstract Changing patterns of gene expression is a normal consequence of aging in healthy tissues, but there is a functional cost to these changes that can be relevant to disease pathology. The most obvious age-related disease in breast is cancer, with a majority of newly diagnosed breast cancer occurs in women over 50. The gene expression changes that occur with age in breast have functional consequences in the epithelial progenitor and somatic cells. The result is a decline of the myoepithelial lineage, loss of luminal cell specificity, and accumulation of differentiation defective multipotent progenitor cells. We have hypothesized that these tissue-level changes make older epithelia more susceptible to transformation. Because gene expression patterns reflect the wiring and response diagrams of cells, it is of central importance to understand the origins of age-related transcriptomes. Additionally, early passage normal human mammary epithelial cells (HMEC) show age-dependent functional and molecular hallmarks consistent with in vivo, suggesting that the underlying gene expression patterns are metastable. DNA methylation is a stable, but malleable, form of epigenetic regulation that may underpin these biologically metastable states. Genome-wide analysis of primary epithelia was used to identify a set of genes that exhibit age- and lineage-specific expression that was inversely correlated with promoter CpG methylation. Chemical perturbation of DNA methylation in pre-menopausal HMEC resulted in a biochemical phenocopy of more advanced age. Tissue-mimetic cultures were used to demonstrate that lineage-specific gene expression and methylation in luminal cells were imposed by distinct microenvironments. Optimal maintenance of the luminal phenotype required direct contact with the apical surface of myoepithelial cells. Mimetic tissues built with HMEC that differed in chronological age of their donors revealed that age-dependent gene expression and methylation patterns are communicated between the two different lineages, as exposure of pre-menopausal luminal cells to a post-menopausal microenvironment imposed transcriptional patterns in luminal cells consistent with post-menopause. These data demonstrate that lineage- and age-dependent phenotypes in HMEC are maintained by microenvironment-imposed metastable epigenetic states.


Page 36


Molecular Biology Poster #3 Unmasking latent heterogeneity in transformed epithelial cells through single cell isolation drives epithelial-mesenchymal transition and creates phenotypic diversity. Amanda K. Paulson, Chithra Krishnamurthy, and Zev J. Gartner (University of California, San Francisco)

Presenting Author: Chithra Krishnamurthy Pharmaceutical Chemistry [email protected] Abstract Tumor heterogeneity, arising from local differences in the microenvironment, noisy gene expression, or acquired genetic and epigenetic changes, poses serious challenges in the diagnosis and treatment of cancer. Ensemble measurements can mask the presence of rare subpopulations, such as drug resistant clones, or those that enhance the malignant properties of neighboring cells. The transformed epithelial cell line MCF10AT is an example of a population that displays deceptive phenotypes as an ensemble despite heterogeneity at a single cell level. We found that individual cells in the bulk population displayed considerable heterogeneity with respect to their epithelial-mesenchymal transition (EMT) status that was masked when the cells were cultured in aggregate in synthetic organoids. To investigate the origin and consequences of this heterogeneity, we derived single cell clones from the parental population. Clone-to-clone variability spanned a range of phenotypes; many clones displayed a more purified epithelial or mesenchymal phenotype than the parental population, showcasing differentiated states that were suppressed in the bulk population. In order to investigate how this variability might affect the behavior at an ensemble level, we created mosaic aggregates of the bulk population combined with a single clone. Depending on the clone involved, these organoids exhibited a wide range of collective behaviors, suggesting plasticity within the parental population. Behaviors spanned from normal morphogenesis to tissue scattering and hypermotility. Analysis of mRNA transcripts among phenotypically divergent clones showed elevated levels of fibronectin among clones that induced tissue scattering, while clones that allowed tissues to condense displayed higher levels of E-cadherin and laminin V. We hypothesize that cell-cell and cell-ECM signaling among a heterogeneous parental population maintains the MCF10AT line in a metastable state by averaging the microenvironment generated from numerous subclones. However, when isolated from the bulk population, clones differentiate in response to purely autocrine feedback, driving them into a specific state. This dynamic reciprocity between different cell states in a well-mixed population can mask underlying phenotypic heterogeneity that only emerges when clones are isolated from the bulk population, as may happen during tumor cell dissemination or therapeutic intervention.


Page 37


Molecular Biology Poster #4 Processing by MRE11 is involved in the sensitivity of subtelomeric regions to DNA double-strand breaks Keiko Muraki, Limei Han, Douglas Miller and John P. Murnane UCSF Department of Radiation Oncology

Presenting Author: Keiko Muraki Radiation Oncology Mt. Zion [email protected] Abstract Double-strand breaks (DSBs) are very hazardous to cells, because they promote chromosome rearrangements leading to cell death or cancer if improperly repaired. We have investigated the consequences of DSBs near telomeres using integrated plasmids that contain an I-SceI endonuclease recognition sequence to selectively introduce DSBs, and transgenes to analyze the consequence of DSBs. We have previously shown that DSBs generated near telomeres in a human cancer cell line show a much higher frequency of large deletions and gross chromosome rearrangements (GCRs), and a lower frequency of nonhomologous end joining (NHEJ). This repair deficiency at telomeres has been proposes as a mechanism by which oncogene-induced replication stress promotes telomere dysfunction, leading to chromosome instability in human cancer cells (Cancer Res. 70:4255, 2010), or senescence in normal human cells (EMBO J, 31:2839, 2012). In addition, the repair deficiency in repair of DSBs has been proposes as a mechanism for ionizing radiation-induced senescence in normal mammalian cells (Nat Cell Biol. 14:355, 2012; Nat Commun. 3:708, 2012). Our results show that inhibition of MRE11 3′-5′ exonuclease activity with Mirin reduces the frequency of large deletions and GCRs at both interstitial and subtelomeric DSBs, indicating that inappropriate processing of DSBs by MRE11 is involved in the formation of large deletions and GCRs at both interstitial and telomeric DSBs, although the frequency and extent of this inappropriate processing is greatly increased at DSBs near telomeres. In contrast, MRE11 3′-5′ exonuclease activity is not involved in the formation of small deletions, which occur at the same frequency at interstitial and telomeric DSBs, leading us to conclude that small deletions occur through C-NHEJ, and that C-NHEJ functions normally at telomeric DSBs. Our results also suggest that the ATM inhibits large deletions at interstitial sites by protecting DSBs, while ATM inhibits large deletions at telomeric sites due to its requirement for Classical nonhomologous end joining (C-NHEJ). Combined, our results demonstrate that telomeric regions are proficient in C-NHEJ, but that telomeric DSBs are highly prone to inappropriate processing, which promotes large deletions and GCRs, involving Alternative NHEJ (Nucleic Acids Res. 43:7911, 2015).


Page 38


Molecular Biology Poster #5 Cytometric Atlas of the Human Breast: Comprehensive Characterization Reveals 12 Distinct Cell Types William C. Hines†, Ambrose Carr‡, Kate Thi†, Zhenmao Wan‡, Dana Pe’er‡ and Mina J. Bissell† †Life Sciences Division, Lawrence Berkeley National Laboratory; Mailstop 977R225A, 1 Cyclotron Road; Berkeley, CA 94720, USA. ‡Department of Biological Sciences, Columbia University; 1212 Amsterdam Avenue, New York, NY 10027, USA

Presenting Author: Curt Hines LBNL-Bissell Lab [email protected] Abstract Since formulation of the cellular theory in the early 19th century—relatively not so long ago—it has become universally recognized that tissues and organisms are formed and shaped by cells of many different types, all operating in beautiful harmony. Many mechanisms of this orchestration remain largely a mystery to this day, yet are important for us to understand if we are to fully comprehend fundamental processes in disease, especially one as formidable and perplexing as cancer. In the current era of progressive technological advances, many areas of biology are converging, and we now recognize and expect differences between cells, even among those of the same type, but whose origins rest in different tissues. Yet, many core tenets are being universally applied. The detailed cellular composition of tissues is clouded with uncertainty for many tissue types, and this includes the breast. To clarify the cell types in the breast and build a solid foundation for future work, we have meticulously examined a large collection of normal breast tissues, microscopically and by flow cytometry. We developed a complex antibody panel and rigorous gating strategy capable of objectively resolving and sorting each major cell type. We have sequenced mRNAs expressed by even the rarest of cell subpopulations and refined or developed methods for culturing each. This includes: several different luminal and epithelial fractions, myoepithelial cells, adipocytes, leukocytes, pericytes, erythrocytes, adipose-derived mesenchymal stem cells, and both lymphatic and vascular endothelial cells –12 different types in all. Future studies are aimed at developing heterotypic culture models using these cells, but also to use our new knowledge and experience to similarly explore heterogeneity of ductal carcinoma in situ (DCIS) and invasive breast cancers. We will present the development of this method and characterization to date of the sorted cell populations.


Page 39


Molecular Biology Poster #6 Radiation-induced TGFβ promotes malignant progression on human epithelial cells by triggering an EMT program that selects for TGFβ-resistant cells Gonzalez-Junca, A.; Barcellos-Hoff, MH.

Presenting Author: Alba Gonzalez Junca Radiation Oncology [email protected] Abstract Ionizing radiation is considered a complete carcinogen, with the capacity to both initiate and promote cancer by inducing the mutation and altering signals that can affect tumor progression. Recent evidence has demonstrated that radiation-induced changes in the microenvironment are essential for the promotion of more aggressive epithelial tumors. One of the earliest events in the epithelial carcinogenesis program is dysplasia, which is characterized by a loss of correct tissue organization and expansion of cells. Our hypothesis is that the execution of an early epithelial to mesenchymal transition (EMT) program is an essential step that drives dysplasia. We have observed that ionizing radiation is able to elicit at TGFβ-mediated EMT program in non-malignant epithelial cells MCF10A when they are irradiated as single-cells. In contrast, the epithelial phenotype is stabilized by cell-to-cell contacts, which prevent the cells to undergo EMT. So, in tissues, EMT and therefore dysplasia would only occur when cell-to-cell junctions are dismantled. One of the main components of the microenvironment are immune system cells; among them, macrophages have a prominent role in inflammation as well as tumor promotion. Radiation can alter macrophage polarization and we have observed a radiation-induced macrophage gene-expression signature in vivo. We speculated that macrophages may collaborate with radiation-induced TGFβ and contribute to the execution of an EMT program by disrupting epithelial cell-to-cell junctions. To further study the contribution of macrophages and radiation-induced TGFβ in the carcinogenesis process, we established an in vitro model consisting in the non-malignant human breast cells, MCF10A co-cultured with distinct subtypes of human macrophages differentiated from circulating human peripheral blood mononuclear cells (PBMC). We found that M1 and M2 subtype macrophages, as well as a macrophage-secreted cytokine IFNɣ, cooperate with radiation-induced TGFβ to induce EMT in non-malignant epithelial cells, which is evidenced by morphological changes, E-cadherin loss, β-catenin nuclear translocation, and increased expression of EMT transcription factors. Importantly, macrophages and IFNɣ cooperate with radiation-induced TGFβ to select for cells that have lost the TGFβ anti-proliferative response, which we define as cells that are double positive staining for pSmad2 and the proliferation marker Ki67. Furthermore, we found this population of TGFβ-resistant cells is also present in human early breast cancer lesions. These studies shed light to the initial events of that establish a tumor microenvironment, which promotes epithelial carcinogenesis process and selection of clones resistant to the TGFβ anti-proliferative effect.


Page 40


Molecular Biology Poster #7 HZE RADIATION PROMOTES MALIGNANT PROGRESSION OF HUMAN EPITHELIAL CELLS BY TRIGGERING AN EMT PROGRAM AND SELECTING FOR TGFΒ-RESISTANT CELLS Alba Gonzalez-Junca1, Lin Ma1, Mary Helen Barcellos-Hoff1,2 1 NYU School of Medicine, Department of Radiation Oncology, New York, NY; 2 UCSF Department of Radiation Oncology, San Francisco, CA

Presenting Author: Lin Ma Radiation Oncology Barcellos-Hoff Laboratory [email protected] Abstract Exposure to HZE radiation in deep space is expected to increase cancer risk. Our laboratory has reported that mice irradiated with HZE and subsequently transplanted with non-irradiated oncogenically-primed mammary epithelial cells develop more aggressive tumors compared to sham or γ-irradiated mice, underscoring a role for radiation non-targeted effects (NTE) in carcinogenesis (Illa-Bochaca, et al. 2014) One of the earliest events in the epithelial carcinogenesis program is the dysplasia, which is characterized by a loss of correct tissue organization and aberrant expansion of cells. Our hypothesis is that the execution of an early EMT program is an essential step that drives dysplasia. Our previous studies of epithelial carcinogenesis have shown that both high and low LET radiation primes cells to undergo a TGFβ-mediated epithelial to mesenchymal transition (EMT) when they are irradiated as single cells (Andarawewa et al. 2011). However, the epithelial phenotype is stabilized by cell-to-cell contacts that prevent the epithelial cells from undergoing EMT. Thus in intact tissues, it is unlikely epithelial cells execute EMT in response to radiation unless cell-to-cell junctions are dismantled. Macrophages dismantle junctions as they move about a tissue and we have identified a macrophage gene-expression signature in HZE irradiated tissue (Tang et al. 2014). Here we tested whether primary human monocyte populations could release cell-contact inhibition of EMT when co-cultured with irradiated, confluent human epithelial cells. Macrophages were differentiated in vitro from CD14+ peripheral blood mononuclear cells (PBMC) from healthy donors using published protocols. MCF10A human mammary epithelial cells were irradiated with low fluence 600 MeV/amu Fe56, 350 MeV/amu Si36 and H+ 1000MeV/amu particles. Co-cultures were established by addition of macrophage subpopulations to established MCF10A monolayers shortly after irradiation and phenotype was analyzed 3 days later. We observed that activated M1 and M2, but not un-differentiated M0, macrophages, cooperate with TGFβ to induce an EMT program by disrupting cell-to-cell contacts in HZE irradiated cultures but not in γ-irradiated cultures. EMT was evidenced by changes in cell morphology, loss of epithelial markers E-cadherin and β-catenin, and increased mRNA of EMT transcription factors, SNAIL and TWIST. Notably, exogenous IFNɣ, a cytokine released by activated macrophages, is sufficient to elicit EMT under these conditions, regardless of radiation quality. Activated macrophages are also a significant source of TGFβ activity. As normal epithelial cells are very sensitive to TGFβ-mediated cell cycle arrest, yet all breast cancers escape this growth control, we


Page 41


postulated that abnormally high TGFβ levels in the microenvironment could act as a positive pressure to select for populations that have escaped from the TGFβ cell cycle control. To assess the proportion of epithelial cells that are resistant to TGFβ's anti-proliferative effect we used double staining for pSmad2 (readout of TGFβ pathway activation) and Ki67 (marker of proliferation), which are mutually exclusive in non-malignant epithelia. Activated M2 macrophages and TGFβ co-cultured with Si-irradiated MCF10A cells, increased the proportion of pSmad2 + Ki67 double positive cells by 1.8 fold (p=0.004). Notably, cooperation between activated M1 and M2 macrophages was only observed in HZE irradiated cells, suggesting that aspects of this mechanism depends on radiation quality. In contrast, IFNɣ increased the population of TGFβ resistant cells in irradiated cultures independent of radiation quality. Our findings using this in vitro human model link innate immune cells to the TGFβ-mediated EMT program and selection of clones resistant to TGFβ anti-proliferative effects during early stages of radiation carcinogenesis process. Understanding the mechanisms by which HZE radiation increases the risk of aggressive cancers may help estimate risk, identify co-factors, like microgravity, that might exacerbate this mechanism and offer the potential for designing rational countermeasures that are needed for human deep space exploration.


Page 42


Molecular Biology Poster #8 Tumor microenvironments alter myoepithelial function by modulating its tumor-suppressive properties and the myoepithelial secretome Rosalyn W. Sayaman (1,2), Masaru Miyano (2), Wen-Rong Lie (3), Mark A. LaBarge (2) and Mina J. Bissell (2) 1. Interdisciplinary Graduate Group in Comparative Biochemistry, University of California, Berkeley, Berkeley, CA, USA; 2. Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA; 3. EMD Millipore Corporation, St Charles, MO, USA

Presenting Author: Rosalyn Sayaman Lawrence Berkeley National Lab/UC Berkeley Mina Bissell [email protected] Abstract Loss of the myoepithelial layer in the mammary gland is a hallmark of breast cancer progression from benign to invasive. Myoepthelial cells have previously described tumor-suppressive properties and its loss is agreed to be a permissive factor in breast cancer. However, why this loss occurs remains unclear. We hypothesized that during tumor progression, dynamic and reciprocal signaling between the altered myoepithelial microenvironment and the myoepithelial cells lead to disruption of proper function of the myoepithelial lineage. To address this question, we cultured myoepithelial cells isolated from normal human mammary epithelial cells in either ‘normal-like’ extracellular matrices (ECM) – recapitulating the major protein components of epithelial basement membrane (BM) in vivo, or ‘tumor-like’ ECM – proteins found in the stromal ECM to which the epithelial layer becomes more and more exposed as the BM breaks down during cancer progression. To explore the effects of ECM on myoepithelial function, we assayed for cellular changes in myoepithelial lineage specific markers and known tumor-suppressive factors, and secretory levels of angiogenic and cancer-associated factors in these two types of microenvironments. We found higher expression of known myoepithelial tumor-suppressive factors in cells grown on laminin-111-rich ‘normal-like’ ECM. On the other hand, myoepithelial cells grown in ‘tumor-like’ ECM exhibited much higher expression of an array of secreted factors previously mapped to known cancer pathways. Our findings demonstrate a potential role of laminin-111 in regulating proper myoepithelial function in the normal context, and link stromal ECM to disruption of myoepithelial tumor-suppressive functions and the up-regulation of tumor-promoting factors in the tumor context. These results highlight the dynamic interactions between the cell and the ECM, and the feedback mechanism that leads to the synthesis and secretion of key factors into the microenvironment, which in turn changes myoepithelial properties. An understanding of how the ECM regulates proper maintenance and function of myoepithelial cells in normal vs. tumor-context is important in elucidating the mechanism behind the loss of myoepithelial cells, and more significantly how this loss contributes to the progression of breast cancer.


Page 43


Molecular Biology Poster #9 Investigating the mechanisms of resistance to neratinib in HER2-positive breast cancer Pei Rong Evelyn Lee, Denise Wolf, Jean-Philippe Coppe, Laura van't Veer Department of Lab Medicine, UCSF

Presenting Author: Pei Rong Evelyn Lee Laboratory Medicine van't Veer Laboratory [email protected] Abstract Approximately 20% of human breast cancers overexpress the HER2 proto-oncogene, which confers a more aggressive tumor phenotype and is associated with a poor prognosis. Despite the initial clinical efficacy of tyrosine kinase inhibitors in HER2-positive breast cancer, resistance invariably develops, and overcoming or preventing resistance is key to unleashing their full therapeutic potential. We therefore sought to study the mechanisms of resistance to neratinib, an irreversible EGFR, HER2 and HER4 tyrosine kinase inhibitor which has recently graduated from the I-SPY2 randomized phase II clinical trial. We developed neratinib-resistant cell lines by exposing three HER2-overexpressing breast cancer cell lines to increasing concentrations of neratinib over several months. Cell viability assays and immunoblots were used to establish neratinib IC50 in both parental and resistant cell lines and examine the changes in signaling dynamics, respectively. We observed a 20-60-fold increase in neratinib IC50 in resistant cell lines. Compared to the sensitive parental strains, resistant cells exhibited activated MAPK signaling. Ongoing studies aim to further elucidate the role of MAPK activation and other molecular mechanisms driving neratinib resistance.


Page 44


Molecular Biology Poster #10 Studying the Interactome of Breast Cancer: the Cancer Cell Map Initiative Kyumin Kim, Margaret Soucheray, Jeremy Chang, Nevan Krogan, Danielle Swaney, and Minkyu Kim Department of Cellular and Molecular Pharmacology, and California Institute for Quantitative Biosciences, University of California, San Francisco

Presenting Author: Minkyu Kim Cellular and Molecular Pharmacology [email protected] Abstract Our genetic information is stored in the form of DNA, and proteins – the actual working part of a cell – are made based upon the instruction on DNA sequence. Thus, changes to DNA sequence lead to changes in protein and ultimately how protein works. Each protein has a unique job and is involved in various cellular processes and pathways by interacting and working together with other companion proteins in teams, which enable cells to properly function in the body. Recent technical progress revealed numerous DNA sequence mutations in cancer cells. But how many of these mutations consequently result in changes in normal cellular processes is poorly understood. Through years of study, it is becoming clear that cancer is a disease that arises not only because of defects in individual proteins, but also because of the action of hallmark cellular processes and biological pathways in which they involve. Therefore, what is urgently needed is to understand the ways these mutated proteins interact with other proteins and comprise an altered working team (i.e. complex) or pathway, leading to a devastating cellular outcome – cancer. The goal of this project is to comprehensively uncover protein-protein (physical) and genetic (functional) interactions occurring in various breast cancer subtypes to better recognize how cells with varying mutations are capable of hijacking and modifying regular pathways and complexes toward breast tumorigenesis. Our approach will be powerful to define and characterize the hallmark pathways operating in breast cancer and to pinpoint key proteins that may be specifically targeted to eradicate breast tumors, while minimizing harm to healthy tissues in patients.


Page 45


Physical Sciences Poster #1 A continuum of therapeutic responses to a single pathway-targeted compound composed by combinatorial microenvironments Chun-Han Lin (a, b) , Fanny A Pelissier (a, c), Catherine Park (d), and Mark A LaBarge (a) a. BSE Division, Lawrence Berkeley National Laboratory, Berkeley, CA b. Program in Comparative Biochemistry, University of California Berkeley, Berkeley, CA c. Department of Biomedicine, University of Bergen, Bergen, Norway d. Department of Radiation Oncology, University of California San Francisco, San Francisco, CA e. Center for Bioengineering, Tissue Regeneration, Department of Surgery, UCSF, San Francisco, USA

Presenting Author: Chun-Han Lin Lawrence Berkeley National Laboratory Mark LaBarge Laboratory [email protected] Abstract The tumor microenvironment is emerging as a likely determinant of drug response efficacy, but the underlying mechanisms are poorly understood. Physical properties, such as matrix rigidity, and the chemical properties, such as the extracellular matrix (ECM) and soluble factor composition, of the microenvironment are known to modulate cellular functions. Here we demonstrate that the anti-‐proliferative effect of the HER2-‐targeted drug lapatinib in HER2-‐amplified breast cancer cells is modulated by the rigidity of the adhesive substrata. Resistance to lapatinib and expression of the HER2 receptor were proportional to elastic modulus, whereas the proportion of phosphorylated HER2 receptors was inversely correlated. Modulus-‐ dependent lapatinib resistance required the mechanosensitive transcription co-‐ activators YAP and TAZ. We used MicroEnvironment microarray (MEArray) to further study the effect of distinct molecular microenvironments, in addition to matrix rigidity, inlapatinib responses. Using this combinatorial microenvironment approach revealed that, within an isogenic population of cells, resistance or sensitivity fell along a continuum that was determined by the sum of molecular and physical properties of any given microenvironment. Thus both the mechanical and chemical properties of tumor microenvironments are plausible determinants of resistance to and efficacy of HER2 pathway-‐targeted therapeutics.


Page 46


Physical Sciences Poster #2 Interplay between ECM stiffness, miR-203 and mammographic density Ivory Dean1, Irene Acerbi1, Janna Mouw1, Alfred Au2, Yunn-yi Chen3, Shelley Hwang4, Valerie M. Weaver1,5,6 1 Department of Surgery and Center for Bioengineering and Tissue Regeneration, UCSF (San Francisco, CA); 2 UCSF Carol Buck Breast Cancer Center (San Francisco, CA); 3 Department of Pathology, UCSF (San Francisco, CA); 4 Department of Surgery, Duke University (Durham, NC); 5 Department of Bioengineering and Therapeutic Sciences, University of California at San Francisco; 6 Department of Anatomy, University of California at San Francisco, 7UCSF Helen Diller Family Comprehensive Cancer Center

Presenting Author: Ivory Dean Surgery Valerie Weaver Laboratory [email protected] Abstract High mammographic density (MD) is associated with an overall greater lifetime risk of breast cancer and is characterized by higher tissue levels of extracellular matrix (ECM) collagen. Tissues with elevated levels of type I collagen are stiffer and ECM stiffness can promote the malignant transformation of oncogenically-primed mammary epithelial cells in culture and in vivo. Whether or not high MD reflects a breast stroma that is stiff, and if this stiffer ECM could account for the elevated breast cancer risk observed in women with high MD, has yet to be resolved. To address this question we have been conducting a comprehensive biophysical and molecular analysis of the breast parenchyma in breast obtained by prophylactic mastectomy in women with BIRADS status classified as low (BIRADS 1) versus high MD (BIRADS 4). Thus far, our data has revealed that the intra-lobular ECM associated with the terminal end-buds contains anisotropic compliant and relaxed collagen fibrils. By contrast, the inter-lobular ECM contains stiff, oriented collagen fibrils. Preliminary data suggest that the ECM associated with the terminal end-buds in the upper outer quadrant may be stiffer in women with high MD as compared to low MD. We also found that ECM stiffness modulates microRNAs (miRs), which are a class of short non-coding RNAs that act post-transcriptionally to regulate gene expression. Interestingly, the mammary tissue obtained from women with high MD had lower levels of miR-203, which has been implicated in epithelial-to-mesenchymal transition (EMT) inhibition. Consistently, in vitro studies revealed that miR-203 is repressed in human mammary epithelial cells cultured on a stiff ECM. We also found that a stiffened mammary ECM decreases miR-203 expression and promotes an EMT. Importantly, inhibiting ECM stiffening in mouse tissue in vivo restored miR-203 levels and reduced EMT induction and metastasis. Moreover and critically, we observed that triple negative breast tumors that often express EMT markers develop the stiffest ECM and have reduced miR-203 levels. Thus, in high MD tissue, ECM tension may influence cancer risk by modulating miRs. Studies are underway to elaborate these preliminary findings and to assess their clinical relevance. (This work is supported by the: NIH NSRA T32 CA 108462-11 to Ivory Dean, Susan G Komen PDF12230246 to Irene Acerbi, W81XWH-13-1-0216, USMRAA DOD-BCRP BC122990, NIH R01 CA138818-01A1 and NIH R01 CA192914-01 to Valerie Weaver.)


Page 47


Physical Sciences Poster #3 High spatial resolution DWI for evaluation of breast tumor early treatment response: Association of apparent diffusion coefficient changes with pathologic complete response Lisa J Wilmes1, Wei-Ching Lo1, Wen Li1, David C Newitt1, Suchandrima Banerjee2, Evelyn Proctor1, Emine U Saritas3, Ajit Shankaranarayanan2, Nola M Hylton1 1 Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA; 2 Applied Science Laboratory, GE Healthcare, Menlo Park, CA; 3 Bilkent University, Ankara, Turkey

Presenting Author: Lisa Wilmes Radiology and Biomedical Imaigng Hylton Laboratory [email protected] Abstract Background: Diffusion-weighted imaging (DWI) provides information about tissue microstructure and has shown promise as a potential biomarker of early treatment response in breast cancer. One limitation of standard echo-planar imaging (EPI) DWI is that the spatial resolution is typically much lower than the T1-weighted acquisition used for dynamic contrast enhanced (DCE)-MRI. Our group has optimized a high spatial-resolution single-shot EPI reduced-FOV DWI (HR-DWI) acquisition for breast imaging. The sequence differs from standard DWI by utilizing a 2D spatially-selective echo-planar RF excitation pulse. This results in higher spatial resolution and improved image quality compared to standard DWI. This work investigated the association between tumor pathologic complete response (pCR), a clinical measure of tumor response, and early changes in tumor apparent diffusion coefficient (ADC) metrics measured with HR-DWI. Methods: Twenty patients with invasive breast cancer were scanned with HR-DWI and DCE-MRI before (pre-treatment) and after one cycle (early-treatment) of neoadjuvant taxane based treatment as part of an ongoing IRB approved study at our institution. All patients gave informed consent. Imaging was performed on a 1.5T GE scanner. Early changes in mean tumor ADC, as well as 5th, 15th, 25th, 50th, 75th, and 95th percentile ADCs, and DCE-MRI tumor volume were calculated and evaluated as predictors of pCR. The association between early change predictors (HR-DWI ADC metrics and tumor volume) and pCR was evaluated using receiver operating characteristic analysis to obtain the area under the curve (AUC) for the full cohort of patients, and also for the subset of 14 patients that were both hormone receptor positive (HR+) and human epidermal growth factor receptor 2 negative (HER2-). Results: For early percent change in tumor ADC a trend of increasing AUC with decreasing percentile ADC was observed. Additionally, the AUCs for the lower percentile tumor ADC were higher than for the early tumor volume change. These effects were stronger when only the HR+/HER2- subset was considered. Conclusions: These results suggest that HR-DWI may be of value in evaluating early breast tumor response to neoadjuvant chemotherapy and that characterization of early change in breast tumor ADC metrics may be affected by the tumor genetic subtype.


Page 48


Physical Sciences Poster #4 Breaking the tension: investigating a link between tissue mechanics and tumor immunity in breast cancer Ori Maller, Luke Cassereau, Alex Barrett, Brian Ruffell, Irene Acerbi, Miranda L. Broz, Jennifer M. Munson, Melody A. Swartz, Kirk Hansen, Matthew F. Krummel, Lisa M. Coussens, and Valerie M. Weaver

Presenting Author: Ori Maller Surgery Valerie M. Weaver Laboratory [email protected] Abstract Human breast tumors are highly fibrotic and their extracellular matrices (ECMs) are stiffer relative to benign lesions. We previously demonstrated stiffer ECM and elevated mechanosignaling promoted mammary tumorigenesis, whereas reducing ECM stiffening impeded tumor formation. More recently, we established a positive correlation between the number and location of infiltrating immune cells and ECM stiffness in human breast tumors. In addition, infiltrating immune cells have been implicated in ECM remodeling associated with mammary gland development and tumorigenesis in vivo. This has led us to hypothesize that tumor-associated macrophages (TAMs) drive tissue fibrosis and subsequently may stimulate inflammatory signaling. Using the MMTV-PyMT model, early macrophage depletion not only ablated lung metastases, but demonstrated an anti-fibrotic role for TAMs depicted by a decrease in fibrillar collagen deposition and reduced ECM stiffness. Interestingly, CSF1 antagonist reduced phospho-STAT3 levels and FAK signaling in mammary tumors. To further substantiate the effect of tissue fibrosis on inflammatory signaling in mammary tumor cells, we inhibited lysyl oxidase (LOX), a collagen crosslinking enzyme, in MMTV-PyMT mice to attenuate ECM stiffness — we determined phospho-STAT3 levels decreased in mammary tumor cells in the treated mice. Lastly, we demonstrated ECM stiffness directly caused STAT3 phosphorylation in tumor cells in vitro. Moreover, we found that LOX inhibitor treatment caused a shift in the cytokine milieu consistent with an anti-tumor immune response. The link between tissue fibrosis and STAT3 activity has been further established when we observed decreases in ECM deposition and stiffness levels in mammary tumor cells lacked STAT3 (STAT3 KO). Similar to the LOX inhibition study, STAT3 KO group had a cytokine profile consistent with an anti-tumor immune response, suggesting a causal relationship between tissue mechanics, STAT3 activity, and changes in cytokine milieu. Finally, we found that primary tumor organoids isolated from MMTV-PyMT mice embedded in stiff collagen gels induced macrophage invasion and TGFβ-mediated polarization, but not when these tumor organoids embedded in soft collagen gels. Collectively, our data provide multiple lines of evidence suggesting TAMs promote fibrosis that stimulates inflammatory signaling in tumor cells in early mammary tumorigenesis — and this feed-forward loop induces a pro-tumor immune response.


Page 49


Therapeutics and Clinical Trials Poster #1 HDAC inhibitors modulation of immune-check point pathway in breast cancer Manuela Terranova Barberio1,2, Scott Thomas2, Niwa Ali3, Robert Sanchez3, Michael D. Rosenblum3, Alfredo Budillon2 and Pamela N. Munster1 1 Department of Hematology and Oncology, University of California, San Francisco, CA 2 Experimental Pharmacology Unit - National Cancer Institute G. Pascale - Naples, Italy 3 Department of Dermatology, University of California, San Francisco, CA

Presenting Author: Manuela Terranova Barberio Department of Medicine, Division of Hematology and Oncology Munster Laboratory [email protected] Abstract Background: Breast cancer is one of the most common diseases as leading cause of cancer death in women. Despite aggressive therapeutic options for triple-negative (TNBC) and hormone-sensitive metastatic breast cancer (MBC), survival is poor due to the development of resistance. This prompts the need for novel approaches. Histone deacetylase inhibitors (HDACi) represent a new class of anticancer agents that can reverse hormone therapy resistance. In addition to their effects on ER signaling, HDACi have immune modulatory functions, including modulation of regulatory T-cells, Foxp3 expression, changes in tumor-infiltrating lymphocyte (TILs) composition, induction of PD-L1/PD-1 expression. Under normal conditions the PD-1/PD-L1 pathway down-regulates cytotoxic T-cell activity to maintain immune homeostasis. Cancer cells take advantage of this pathway to suppress and inhibit the anti-tumor immune response. In breast cancer, PD-L1 expression is less frequent and mainly found in TNBC, HER2+, ER- and PR- tumors. Increased PD-L1 expression correlates with higher TILs and those data together correlate with better response in breast cancer patients. PD-1/PD-L1 pathway represents one of the primary immunosuppressive drivers in multiple types of cancer. Thus, inhibiting PD-1/PD-L1 interactions may prevent T-cell suppression and reactivate immune-surveillance. Results: we evaluated PD-L1 basal expression in different breast cancer cell lines, with TNBC and HER2+ cells expressing the higher PD-L1 levels. We found that HDACi up-regulated PD-L1 in a time-dependent manner with a direct transcriptional effect. This was confirmed even in tamoxifen-resistant breast cancer cells, characterized by higher PD-L1 basal expression compared to the parental cells. Co-culturing tumor cells with human peripheral blood mononuclear cell (PBMCs) we performed comprehensive flow cytometric immunophenotyping (FCI) to define the role of epigenetic priming in promoting immune cell activation. We found that HDACi up-regulate PD-L1 on tumor cells independently from PBMCs presence and downregulate CD4 FoxP3+ cells. We performed FCI in


Page 50


tumors and PBMCs from hormone therapy resistant MBC patients enrolled in a phase II clinical trial receiving combination of the HDACi vorinostat and the PD-1 inhibitor pembrolizumab. Conclusion: Overall, this study suggests that the combination of HDACi with immune checkpoint inhibitors identify a novel therapeutic strategy and warrants further clinical evaluation for the treatment of breast cancer.


Page 51


Therapeutics and Clinical Trials Poster #2 Diagnosis and Therapy of Aggressive Breast Cancers by Targeting Urokinase Receptor Efrat Harel - Department of Pharmaceutical Chemistry, University of California, San Francisco Samantha Liang- Department of Pharmaceutical Chemistry, University of California, San Francisco Zev Gartner- Department of Pharmaceutical Chemistry, University of California, San Francisco Rabuka, David- Catalent Pharma Solutions, Emeryville, CA 94608, USA. Susie Brain – Breast Science Advocacy Core, University of California, San Francisco Laura van 't Veer- Laboratory Medicine, Helen Diller Family Comprehensive Cancer Center Charles Craik- Department of Pharmaceutical Chemistry, University of California, San Francisco

Presenting Author: Efrat Harel Pharmaceutical Chemistry Craik Laboratory [email protected] Abstract Triple negative Breast Cancer (TNBC) is a highly aggressive BC subtype, with an increased likelihood of distant recurrence and of death compared with other types of BC. Patients diagnosed with TNBC lack the estrogen and progesterone receptor, the human epidermal growth factor 2 and do not respond well to current therapies. Targeted therapies such as epidermal growth factor receptor inhibitors, showed variable response rates but no survival benefit. One hallmark of aggressive cancers is increased urokinase receptor (uPAR) expression. The over-expression of urokinase plasminogen activator (uPA) and its receptor (uPAR) have been found to contribute to the aggressive phenotype in a number of cancers and is found uniformly in many metastases. uPAR ubiquitousness in TNBC makes it an attractive target for uPAR-directed therapies. It participates in many protein/protein interactions, which leads to pericellular proteolysis and signaling that is mediated by specific integrins. Therefore, inhibition of one or more of these interactions would progressively contribute to a reduction in aggressive behavior. Six inhibitory anti-uPAR Antibodies (Abs) were identified. Two of those Abs, referred to as 2G10 and 3C6, obstruct protein-protein interactions between uPAR and uPA or beta 1 integrin, respectively. The Abs were tested in in vitro and in vivo TNBC models and showed diagnostic and therapeutic potential. In a mouse model of TNBC the Abs targeting two distinct subdomains of slowed or blocked tumor growth. Moreover, treatment of TNBC cell lines in vitro with a combination of 2G10 and 3C6 demonstrated synergy, suggesting that blocking multiple uPAR effector functions simultaneously may provide a dramatically enhanced response. The anti uPAR Abs, have been assembled using a modular platform to give bi-specific Abs based on DNA linkers. The platform enables a library of heterofunctional molecules to be made with precise geometries, valencies, and rigidities. Basic linear construct combining 3C6 and 2G10, have been tested for uPAR recognition ability and therapeutic effect in vitro in TNBC cell lines. Our results indicate that at fixed total protein concentration, the scaffolded Abs outperform the unscaffolded Abs in blocking MDA-MD-231 cell invasion.


Page 52


Therapeutics and Clinical Trials Poster #3 Carboplatin and talazoparib combination therapy results in severe hematologic toxicity, specifically in BRCA 1/ 2 mutated patients: dose optimization may improve outcomes Imke H. Bartelink, Jim Leng, Mallika Dhawan, Scott Thomas, Rahul Aggarwal, Rada Savic, Leslie Floren, Denise Wolf, Laura van ‘t Veer, Pamela Munster

Presenting Author: Imke Bartelink Department of Medicine Munster Laboratory [email protected] Abstract Background The Poly(ADP-ribose) polymerase inhibitor (PARPi) talazoparib is currently in clinical development. Synergistic antitumor effects of talazoparib and platinum drugs have been observed in preclinical models. However DNA damage repair-deficient cancer cells may slow down repair of platinum intercalation. Therefore, we hypothesize that patients with deficient DNA repair mechanisms show improved response, but at the cost of higher risk of toxicity. Methods In a Phase 1 study, talazoparib was dosed 750 or 1000 µg daily and carboplatin at AUC 1 or 1.5 mg*min/mL weekly or every 2 out of 3 weeks in metastatic patients. Patients were grouped in either deleterious BRCA 1/2 mutations or non BRCA 1/2 carriers based on germline or somatic screening tests. Plasma pharmacokinetics of carboplatin and toxic effects of carboplatin and talazoparib on platelets and neutrophils were described using non-linear mixed effect modeling. Results At the time of evaluation 17 patients (14F/3M) were evaluable for analysis. Of these patients two patients had a deleterious mutation in BRCA1, two in BRCA 2, ten had no deleterious mutations in BRCA genes and in two patients no genetic screening was performed. In the four BRCA 1/2 mutation and six non-BRCA mutation carriers, grade 3/4 hematologic toxicity was anemia (0 versus 5) neutropenia (2 versus 3) thrombocytopenia (0 versus 2). All patients required dose reductions until hematologic toxicities were resolved. Previously published parameters of carboplatin single agent on neutrophils and platelets under-predicted the severity of neutrophil and platelet suppression in all patients. We observed a linear correlation between talazoparib dose and toxicity (platelets: ΔOFV=-10, neutrophils ΔOFV=-61, P<0.0001). Talazoparib/carboplatin affected platelet counts (-11.4% versus -1.1%, ΔOFV=-28, P<0.001) and neutrophil counts (-9.1%, versus -3.8%, ΔOFV= -17, P<0.001) significantly larger in BRCA 1/2 mutation carriers compared to non-carriers. The neutropenia model simulations suggests that monthly carboplatin dosing in BRCA 1/2 mutated patients and every other week in non-carriers, would significantly reduce neutropenic severity. Conclusion and future research Talazoparib in combination with carboplatin shows large hematologic toxicity specifically in BRCA 1/2 mutated patients. These results suggest that reduction of dosing frequency may decrease the severity of hematologic toxicity, especially in BRCA mutated patients.


Page 53


Therapeutics and Clinical Trials Poster #4 Inhibition of fatty acid oxidation as a therapy for MYC-overexpressing triple-negative breast cancer Roman Camarda (UCSF), Alicia Y Zhou (UCSF), Rebecca A Kohnz (UC Berkeley), Sanjeev Balakrishnan (UCSF), Brittany Anderton (UCSF), Celine Mahieu (UCSF), Henok Eyob (UCSF), Aaron Tward (UCSF), Gregor Krings (UCSF), Daniel K Nomura (UC Berkeley), and Andrei Goga (UCSF)

Presenting Author: Roman Camarda Cell and Tissue Biology Andrei Goga Laboratory [email protected] Abstract Expression of the oncogenic transcription factor MYC is disproportionately elevated in triple-negative breast cancer (TNBC) compared to estrogen, progesterone and/or human epidermal growth factor 2 receptor-positive (RP) breast tumors. We and others have shown that MYC alters metabolism during tumorigenesis. However, the role of MYC in TNBC metabolism remains largely unexplored. We hypothesized that pharmacologic inhibition of MYC-driven metabolic pathways may serve as a therapeutic strategy for this clinically challenging subtype of breast cancer. Using a targeted metabolomics approach, we identified fatty acid oxidation intermediates as dramatically up-regulated in a MYC-driven model of TNBC. A lipid metabolism gene signature was identified in patients with TNBC in the TCGA and multiple other clinical datasets, implicating fatty acid oxidation as a deregulated pathway critical for TNBC metabolism. We find that MYC-overexpressing TNBC, including a transgenic model and patient-derived xenograft (PDX), display increased bioenergetic reliance upon fatty-acid oxidation (FAO). Pharmacologic inhibition of FAO catastrophically decreases energy metabolism of MYC-overexpressing breast cancer, blocks growth of a MYC-driven transgenic TNBC model and MYC-overexpressing PDX. Our results demonstrate that inhibition of FAO is a novel therapeutic strategy against TNBCs that overexpress MYC.


Page 54


Therapeutics and Clinical Trials Poster #5 Advancing Research on Cancer Therapeutic Resistance: A request for UCSF community planning involvement Lara Cobler, PhD1, Luika Timmerman, PhD*1, Catherine Smith, MD*1,2, Cancer Center Therapeutic Resistance Symposium Participants and Organizers, Pamela Munster, MD3, Alan Ashworth, PhD, FRS1,4 1 Helen Diller Family Comprehensive Cancer Center, UCSF; 2 Assistant Adjunct Professor Division of Hematology/Oncology, Department of Medicine; 3 Professor in Residence, Division of Hematology/Oncology, Department of Medicine; 4 President, UCSF Helen Diller Family Comprehensive Cancer Center and Senior Vice President for Cancer Services, UCSF Health, Professor of Medicine, Division of Hematology/Oncology, Department of Medicine *Equal contributors

Presenting Author: Luika Timmerman and Lara Cobler Helen Diller Family Comprehensive Cancer Center Timmerman Laboratory [email protected]; [email protected] Abstract Hundreds of cancer therapeutics have been developed and are used to treat a wide spectrum of tumor types. However the pervasive ability of tumors to evolve resistance to essentially all of these drugs results in more than half of a million US cancer deaths each year, including more than 40,000 women with metastatic breast cancer. Patients succumb when their tumors become resistant to the available arsenal of therapeutics appropriate for their tumor type. Despite these statistics there is a paucity of research focused on elucidating the mechanisms that specific tumor types commonly use to evade particular therapeutics and no longitudinal study of the stepwise acquisition of drug resistance in clinical populations as they are moved through various therapeutic options. There is even less genome-era interest in classical multi-drug resistance to mainstay chemotherapeutics, although the majority of cancer patients will receive some form of chemotherapy together with or subsequent to modern targeted therapeutics. Recent analyses of resistance to targeted tyrosine kinases-specific therapeutics are the exception to this rule. But for the most part these involve simply hunting for mutations in the target protein by gene sequencing methods. A wealth of preclinical studies suggest that resistance can be much more complex, involving for example activation of alternate proliferative signaling pathways, differentiation via activation of epithelial-mesenchymal transition, differentiation to other tissue types, or outgrowth of pre-existing insensitive clonal populations. In order to examine current challenges and to promote discussion, we organized a fact-finding symposium at the Mission Bay campus of UCSF, supported by the Developmental Therapeutics Program of the UCSF Helen Diller Family Comprehensive Cancer Center (HDFCCC). Participants were asked to discuss barriers that prevented or vastly slowed their ability to gain traction on drug resistance, either in their clinical experiences or in laboratory-based preclinical studies. They were also asked to suggest actions that the Cancer Center could take to lower these barriers. This poster summarizes the outcome of these conversations, and asks for your input to influence the time and resources at the UCSF Cancer Center that will be dedicated to this important topic.


Page 55


Therapeutics and Clinical Trials Poster #6 Toward Developing a Metastatic Breast Cancer Treatment Strategy that Incorporates History of Response to Previous Treatments Aleksandra Olow, Denise Wolf, Laura van ‘t Veer

Presenting Author: Aleksandra Olow Laboratory Medicine [email protected] Abstract Introduction: Typically, managing metastatic breast cancer care involves selecting multiple drugs regimens over patient’s lifetime, with new therapeutics being introduced once treatment resistance develops. Though information regarding response to past treatments may provide clues regarding the classes of drugs most and least likely to work for a particular patient, currently there is no systematized knowledge base that would support clinical treatment decision-making that takes past treatments and responses into account. As a first step in this direction, we calculate the conditional probability of response to a variety of compounds given information on sensitivity to other agents using an in vitro cell line drug screen panel. Methods: To calculate the conditional response probability matrix we utilized cell line viability data from 84 metastatic breast cancer (MBC) cell lines treated with 86 compounds (Daemen et al., 2013). GI50 thresholds were used as a cut-off for treatment response and encoded in a binary. A selected subset of non-responding cell lines exposed to a given agent had an average response value calculated for all the remaining 85 compounds. Next, drugs with largest dynamic range of responses were determined by IQR > 0.13, which yielded 30 compounds. The conditional probability data was clustered using Euclidean distance matrix and ward clustering algorithm, and represented on a heatmap plot; additionally annotated with a color bar legend of generalized primary mode of action categories. Results: The emerging in vitro response patterns grouped compounds into three categories: 1) cell cycle inhibitors, 2) topoisomerase inhibitors 3) PI3K/EGFR inhibitors. Given insensitivity to compounds in the cell cycle inhibitor group, we observe a dramatic increase in sensitivity to both topoisomerase inhibitors, and PI3K/EGFR and HDAC inhibitors. Moreover, reduced sensitivity to topoisomerase inhibitors associates with an increased responsiveness to cell cycle and PI3K/EGFR/HDAC inhibitors. Interestingly, given insensitivity to PI3K/EGFR/HDAC compounds, there is a slight increase in sensitivity to topoisomerase targeting compounds group, and a substantial increase in response to cell cycle inhibitors. Conclusions: Investigating probabilities of drug sensitivity conditioned on history of drug resistance could be pivotal in informing clinicians deciding on the next line of breast cancer treatments for treatment-resistant patients.


Page 56


Therapeutics and Clinical Trials Poster #7 A Randomized Phase II Study of Pembrolizumab, an anti-PD (programmed cell death)-1 Antibody, in Combination with Carboplatin Compared to Carboplatin Alone in Breast Cancer (BC) Patients with Chest Wall Disease, with Immunologic and Genomic Correlative Stu Neelima Vidula, Andrei Goga, Matthew Krummel, Hope S. Rugo University of California, San Francisco

Presenting Author: Neelima Vidula Hematology/Oncology Breast Oncology [email protected] Abstract Background: Patients with BC and chest wall disease have a poor prognosis, necessitating better therapies. The interaction of the PD-1 receptor on tumor infiltrating lymphocytes and programmed death ligand 1 (PD-L1) on tumor cells dampens antitumor immunity. Pembrolizumab, a monoclonal antibody, binds PD-1, promoting tumor immune rejection. Recent phase I studies of anti-PD-1/PD-L1 antibodies have shown efficacy in BC. We hypothesize that pembrolizumab may be effective in treating chest wall disease in combination with carboplatin. Methods:This is a phase II multicenter study including BC patients with chest wall disease that is hormone resistant or triple negative. Eighty-four patients at Translational Breast Cancer Research Consortium sites will be randomized 2:1 to receive treatment with pembrolizumab and carboplatin (n=56, Arm A) or carboplatin alone (n=28, Arm B) until disease progression. Patients randomized to Arm B may cross-over following progression to pembrolizumab alone (Arm Bx). Patients in Arm A will be treated with pembrolizumab 200 mg IV and carboplatin AUC 5 IV every 3 weeks for at least 6 cycles followed by maintenance pembrolizumab 200 mg IV every 3 weeks if stable or responding disease. Patients in Arm B will be treated with carboplatin AUC 5 IV every 3 weeks until progression, whereupon they may cross-over to pembrolizumab 200 mg IV every 3 weeks alone (Arm Bx). An interim analysis for futility will be performed after 18 patients are enrolled into Arm B to allow early stopping of that arm for lack of efficacy. The primary endpoint is to compare disease control rates at 18 weeks of treatment. This study is powered to detect a 20% difference in disease control rates between arms (hazard ratio 0.52, alpha= 0.10, beta= 0.20). Secondary endpoints include progression free survival, toxicity, and overall response rate. Exploratory endpoints include assessing changes in tumor PD-L1 gene expression, tumor and peripheral blood immune composition and cytokine expression, plasma tumor DNA, circulating tumor cells, and tumor MYC genomic expression using tumor biopsy and peripheral blood testing before and after treatment. This study promises to improve our understanding of pembrolizumab for the treatment of BC with chest wall disease.


Page 57


Therapeutics and Clinical Trials Poster #8 Developing silastic tubing for local delivery of hormonal therapy: a novel approach to breast cancer prevention Jeenah Park, UCSF Hem/Onc; Scott Thomas, UCSF Hem/Onc; Allison Zhong, UC Berkeley Molecular & Cell Biology; Alan Wolfe, UCSF Bioengineering and Therapeutic Sciences; Merisa Piper, UCSF Surgery; Manuela Terranova Barberio, UCSF Hem/Onc; Les Benet, UCSF Bioengineering and Therapeutic Sciences; Pamela Munster, UCSF Hem/Onc

Presenting Author: Jeenah Park Hem/Onc Munster Laboratory [email protected] Abstract For women at high risk for breast cancer, preventative interventions are limited to bilateral mastectomy with or without oophorectomy or prolonged anti-estrogen therapy. For many, these two options are unacceptable due to drug related toxicity or the irreversible consequences of prophylactic surgery. As such, many women will choose neither preventive measures. This highlights the need for prevention alternatives that are less invasive, less toxic, and less permanent. Ideally, only the tissue at risk should be treated. Therefore, we have developed a novel approach to release an anti-estrogen to the breast tissue only, with the goal to reach high breast tissue concentrations with minimal systemic exposure. Cultured ER-positive breast cancer cells were used to validate the activity of fulvestrant, 4-hydroxytamoxifen, and raloxifene released from the silastic tubing. CD-1 mice were used to demonstrate accumulation of fulvestrant released from the silastic tubing in the mammary fat pad. LC-MS/MS was used to quantify released fulvestrant in both in vitro and in vivo experiments. NSG mice were used to determine efficacy of fulvestrant released from the silastic tubing in preventing tumor formation. After 10 cm of silastic tubing was loaded with fulvestrant, it was placed in tissue culture media. Media was collected and replaced with fresh media every 3.5 days for one year. It was then used to treat MCF7 and T47D cancer cells for 3 days. Within 7 days, fulvestrant released from the tubing was sufficient to modulate the ER signaling pathway of both cell lines and inhibit cell growth comparable to cells directly treated with a clinically feasible concentration of fulvestrant (100 nM). LC-MS/MS analysis demonstrated a fulvestrant release rate of 179 ± 4 nM fulvestrant every 3.5 day in culture media. To ascertain differential uptake of fulvestrant in the mammary tissue, we implanted 2 cm fulvestrant-loaded tubing proximal to the inguinal mammary fat pad of CD-1 female mice and characterized biodistribution of fulvestrant. Various organs (blood, heart, lung, liver, kidney, and mammary fat pad) were harvested over time post-implantation. Using LC-MS/MS, we determined that fulvestrant preferentially accumulates in the mammary fat pad with minimal to no detection in other organs. To determine whether this local delivery of fulvestrant is sufficient to prevent tumorigenesis in vivo, we implanted 1 cm ethanol or fulvestrant loaded tubing across the lateral mammary fat pad in NSG female mice. Our work has revealed that fulvestrant released from silastic tubing can significantly reduce the size of the orthotopic tumors.


Page 58


Our data from in vitro and in vivo breast cancer models suggest that implantable silastic tubing has the capacity for long-term release of the anti-estrogen fulvestrant at high local concentrations that are sufficient to inhibit ER signaling and tumor cell proliferation with minimal systemic exposure. Further work is underway to design the optimal design for delivery. If successful, this option will provide a more acceptable alternative for breast cancer prevention and allow women at high risk to delay or forgo bilateral mastectomies.


Page 59


Therapeutics and Clinical Trials Poster #9 A pilot feasibility study of the WISDOM (Women Informed to Screen Depending on Measures of Risk) Study, a preference-tolerant randomized control trial evaluating a risk-based breast cancer screening strategy 5Sarah Rosenberg-Wohl, 5Prihatha Narasimmaraj, 5Allison Stover Fiscalini, 5Celia Kaplan, 5Laura van’t Veer, Anne Marie Hallada, 5Carlie Thompson, 5Sarah Theiner, 1Alexander Borowsky, 3Arash Naeim, 2Hoda Anton-Culver, 4Andrea LaCroix, Athena Breast Health Network, 5Laura J Esserman 1 University of California Davis, 2 University of California Irvine, 3 University of California Los Angeles, 4 University of California San Diego, 5 University of California, San Francisco

Presenting Author: Sarah Rosenberg-Wohl Surgery [email protected] Abstract Background: For almost 30 years, annual mammograms for women over 40 have been a cornerstone of the US strategy to reduce breast cancer mortality. Introduction of the 2009 USPSTF screening guidelines, though based on a thorough review of the scientific literature, has triggered scientific debate and a stalemate. The solution is not to prolong the controversy with repetitious reviews of past studies, but rather to test and implement a personalized model that leverages advances in breast cancer biology, risk assessment, and imaging to provide screening recommendations based upon well-characterized measures of risk. Our WISDOM (Women Informed to Screen Depending On Measures of risk) study, a preference-tolerant randomized controlled trial funded through PCORI, will evaluate whether such a risk-based screening strategy, compared to annual screening, is as safe, is less morbid, enables prevention and is preferred by women. This upcoming pilot study will test the feasibility and technical implementation of the WISDOM trial, focusing on recruitment, enrollment, and randomization processes, a coverage with evidence development approach to enable rapid adoption, and patient experience and satisfaction. Trial Design: 225 participants will be recruited from Athena patients receiving care at UCSF. Participants must be female; between age 40 and 75; have had a normal mammogram at UCSF in the past 6 months; and in the Athena research cohort. Exclusion criteria are a breast cancer or DCIS diagnosis; inability to provide consent; or inability to speak English. After education about the trial, patients will be asked if they are willing to be randomized to either the risk-based or annual screening schedule; if not, they can self-assign to their preferred schedule. The risk-based screening strategy will incorporate risk assessment based on the latest Breast Cancer Surveillance Consortium model along with established and recently validated genetic risk factors, co-morbidities, and breast density, and will be used to tailor individual recommendations for starting and stopping age, frequency, and screening modality. A saliva assay will be administered to participants in the risk-based arm to screen for genetic breast cancer risk factors (BRCA1 & BRCA2 + 9 additional genes + 157 single nucleotide polymorphisms).


Page 60


Therapeutics and Clinical Trials Poster #10 Characterization of the oral, gut, and breast tissue microbiomes in women with invasive breast cancer, ductal carcinoma in situ (DCIS), or no history of breast cancer Breanna Johnson, Tess O’Meara, Sarah Theiner, Maribel Campos, Laura Esserman, Michael Campbell

Presenting Author: Breanna Johnson Department of Surgery Campbell Laboratory [email protected] Abstract Background: We now know that there is an interplay between our bodies and the bacteria that we carry, both in health and in disease. Inflammatory diseases associated with disruptions in microbial equilibria, such as periodontal disease and ulcerative colitis, as well as chronic antibiotic use, have been associated with increased risk for cancer development. In a previous small pilot study, we observed differences in the oral microbiomes of women with breast cancer and healthy women. This presentation describes an ongoing study in which we aim to validate our previous findings in a larger cohort of women, as well as extend the microbiomes studied to include the gut and local breast tissue microbiota. Trial design and eligibility criteria: Women with invasive breast cancer or with ductal carcinoma in situ (DCIS) who have not received prior therapy for their disease are eligible for enrollment. A cohort of healthy women will also be enrolled. Sample collection kits containing cheek and stool swab materials are distributed to patients in the clinic, and breast tumor tissue is collected at the time of surgery. Specific aims: 1. To characterize and compare oral and gut microbial diversity from women with early stage invasive breast cancer, women with DCIS, and healthy women; and 2. To characterize the microbiota associated with breast tumor tissue and compare this to healthy breast tissue. Methods: DNA from the three sites is isolated, the bacterial 16S rRNA gene is PCR amplified and sequenced, and the bacterial species present are enumerated. Oral, gut, and breast tissue microbial diversity at the genus and species levels will be assessed using the Shannon diversity index. Student's t-test will be used to compare the mean diversity index values. Evaluation of microbial diversity and clinical variables will be determined using Spearman or Pearson correlations depending on the distribution of the data. We will also compare oral microbial diversity and clinical variables across racial and ethnic groups to determine if differences in the microbiomes may explain disparities in incidence across these groups. Classification and regression trees (CART) will be employed to develop rule sets based on the abundance of oral bacteria that discriminate breast cancer vs. DCIS vs. healthy controls. Present accrual: As of January 2016, 64 women with early-stage invasive cancer, 23 women with DCIS, and 82 healthy women received sample collection kits. Breast tumor tissue has been collected from 42 invasive breast cancer cases and 11 DCIS cases. Target accrual: For the oral and gut microbiome analyses: 100 women with early-stage invasive breast cancer, 50 women with ductal carcinoma in situ (DCIS), and 100 healthy women with no history of breast cancer. For the breast tissue microbiome analyses: 15 women with invasive breast cancer, 10 patients with DCIS, and 10 reduction mammoplasty cases from healthy women.


Page 61


Therapeutics and Clinical Trials Poster #11 Testing the ability of pembrolizumab to alter the tumor immune microenvironment of high risk ductal carcinoma in situ Jasmine Wong, Michael J. Campbell, Jo Chien, Breanna Johnson, Laura J. Esserman

Presenting Author: Breanna Johnson Department of Surgery Campbell Laboratory [email protected] Abstract Background: Ductal carcinoma in situ (DCIS) of the breast is a pre-malignant condition that represents a spectrum of disease that ranges from indolent lesions to true precursors of triple negative or Her2 positive cancers. DCIS that is associated with a high risk of recurrence have features that include high grade (II or III), hormone receptor negative, Her2 positive, presentation as a palpable mass or large size (greater than 5 cm), and in patients younger than 45 years old. Changing the immune microenvironment of high-risk DCIS lesions represents a potential opportunity to prevent cancer. The focus of this study is to see if the immune microenvironment of DCIS can be changed with short-term exposure to an immunomodulating agent, pembrolizumab. Pembrolizumab is a humanized monoclonal antibody that blocks the interaction between programmed death 1 (PD-1), a receptor expressed on several immune cells and its ligands programmed death ligands 1 and 2 (PD-L1 and PD-L2). Trial Design: This study will include 3 dose cohorts using a 3+3 cohort dose escalation design followed by a dose expansion cohort at the maximum tolerated dose. Eligible subjects in the dose escalation cohorts will be offered 2 doses of pembrolizumab injected intralesionally (IL) followed by surgery as determined by the surgeon and the subject (partial mastectomy or mastectomy). The maximum tolerated dose will be used in the expansion cohort, with a target enrollment of another 30 subjects randomized to either the control group (10 patients) or the treatment group (20 patients). The control group will proceed to surgery alone. The treatment group will receive 2 doses of intralesional pembrolizumab prior to surgery. All subjects in the expansion cohort will also undergo a baseline MRI at diagnosis and undergo a 2nd MRI prior to surgery. Specific aims: 1. To show that intralesional injection of pembrolizumab into high risk DCIS is safe and feasible 2. To show a change in the tumor immune microenvironment 3. To determine whether short-term exposure to pembrolizumab decreases tumor volume on MRI imaging.

Date post:	02-Jun-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

2016 UCSF Breast Oncology Program Scientific Retreatcancer.ucsf.edu/research/BOP2016/sm_files/BOP...

Documents