Prostate Cancer Program

2018 Annual Research Retreat

Abstracts

Alphabetical by last name of presenting author

Posterboard Assignments by Author Last Name

Last Name⬇ First Name Room Poster-board Title

Balakrishnan Ashwin Cypress 1 An SMS-Based Periprocedural Intervention for Patients Undergoing Prostate Biopsy: Impact on Clinical Utilization, Patient/Provider Communication, and Patient Satisfaction

Belair Cassandra Cypress 2 Improving Prostate Cancer Outcome Prediction Through Noninvasive exRNA Assessment: Year 2 UpdateBhagirath Divya Cypress 3 Role of Exosomes in Inducing Neuroendocrine Differentiation in Advanced Prostate CancerBorno Hala Cypress 4 A Pilot Feasibility Study for Improving Patient Access to Cancer Clinical Trials (Impact Study)Borno Hala Cypress 5 Differential Utilization of Medical versus Surgical Androgen Deprivation therapy for Metastatic Prostate Cancer

Borno Hala Cypress 6 Trial Library – A Pilot Study to Examine the Feasibility, Acceptability, and Preliminary Estimates of Efficacy for a Clinical Trial Online Matching Website on Patient-Prompted Conversations Regarding Prostate Cancer Clinical Trials

Cavazos Taylor Cypress 7 Integrating Genetic information With Machine Learning to Predict Which Prostate Cancer Cases Should Not Be Immediately Treated.

Chen Emmalyn Cypress 8 Quantification of Cell-Free DNA Levels as a Potential Screening and Diagnostic Marker for Localized and Metastatic Prostate Cancer

Chou Jonathan Cypress 9 IMPACT: A Phase II Trial of Combination Immunotherapy in Patients With Metastatic Cancers and CDK12 MutationsDas Rajdep Cypress 10 Investigating the Role of the Long Non-Coding RNA Schlap1 in Prostate Cancer MetastasisDatta Debajyoti Cypress 11 Linking Clinical Molecular Profiles of Tumors to the Electronic Medical RecordFu Qi Cypress 12 Identification of Small Molecule Inhibitors of a Novel Regulatory Pathway of Immune Checkpoint ProteinsGreenland Nancy Cypress 13 Expansile Cribriform Gleason Pattern 4 has Worse Outcomes at Prostatectomy than Glomerulation Gleason Pattern 4Gu Wei Cypress 14 Incidental Detection of Malignancies in Metagenomics Cell-Free DNA TestingHashimoto Yutaka Cypress 15 Up-Regulation of miR-130b Is Involved in Prostate Cancer Racial Disparity Thorough FHIT Pathway inactivationHerlemann Annika Cypress 16 Comparison of TRUS-Targeted vs. MRI-Targeted vs. Systematic Prostate Biopsy in Detecting Prostate CancerJeong Chang Wook Cypress 17 Standard Gamble Utility Measurement from Prostate Cancer Long-Term Survivors: Results from the CaPSURE Registry

Jeong Chang Wook Cypress 18 Validation of the Utility Weighting Function for the Patient-Oriented Prostate Utility Scale (PORPUS-U) Using the Capsure Quality of Life Supplementary Study Data

Kachuri Linda Cypress 19 Disentangling the Relationship Between Genetic Predictors of PSA, Prostate Cancer Susceptibility, and Benign Prostatic Hyperplasia

Kido Tatsuo Cypress 20 The C-Terminal Acidic Domain of TSPX, an X-Linked Tumor Suppressor, Plays Crucial Roles in the Functional Networks of Prostate Cancer

Kith Glenda Hawthorn 30 A Systematic Review of Observational Studies Assessing Outcomes of Prostate Cancer Patients on Active Surveillance: An Examination of Adherence To Protocols

Korenchan David Cypress 21 Hyperpolarized In Vivo pH Imaging Reveals Grade-Dependent Acidification in Prostate Cancer

Korn Natalie Cypress 22 The Rate of Hyperpolarized [1-13C] Pyruvate To [1-13C] Lactate Conversion Distinguishes High-Grade from Low-Grade Prostate Cancer in Patients

Kothari Vishal Cypress 23 Maternal Embryonic Leucine Zipper Kinase (MELK) Is A Novel Therapeutic Target and a Biomarker of Aggressive Prostate Cancer

Langlais Crystal Cypress 24 Obesity and Prostate Cancer Recurrence Following Radical ProstatectomyLi Hui Hawthorn 29 The Long Noncoding RNA PLUTO-201 Is Associated With Prostate Cancer Progression and Metastasis

Lisker Sarah Hawthorn 30 A Systematic Review of Observational Studies Assessing Outcomes of Prostate Cancer Patients on Active Surveillance: An Examination of Adherence To Protocols (With Glenda Kith)

Posterboard Assignments by Author Last Name

Last Name⬇ First Name Room Poster-board Title

Majid Shahana Hawthorn 32 Controversial Role of Micro-RNA-588 in Prostate CancerMeyers Travis Hawthorn 33 inflammatory Bowel Disease and Prostate Cancer Risk: A Mendelian Randomization StudyMoroz Anna Hawthorn 34 Measuring PD-L1 Expression With ImmunoPETNguyen Hao Hawthorn 50 Does PI-RADS V2 Scores Predict Adverse Surgical Pathology At Radical Prostatectomy

Nguyen Hao Hawthorn 51 Evaluating MRI Fusion Biopsy Vs Systematic Ultrasound Guided Biopsy in Predicting High Grade Cancer At Time of Radical Prostatectomy

Nguyen Hao Hawthorn 52 Targeting the Translational oncogenic Program of the Major Cap-Binding Protein eIF4E in Metastatic Castration Resistant Prostate Cancer

Nicholson Lowell Hawthorn 35 A Phase 1/2 Trial of Ibrutinib as Neoadjuvant Therapy in Patients With Localized Prostate CancerNoworolski Susan Hawthorn 36 Characterizing Sparse and Dense Prostate Cancers on Whole-Mount Histopathology and on Multiparametric MRIOh David Cypress 25 Single Cell Discrimination of Immunotherapy-Induced Changes in the Prostate Tumor MicroenvironmentPalmer Nynikka Hawthorn 38 An Ethnographic Study of African American Men's Prostate Cancer Treatment Decision-MakingPalmer Nynikka Hawthorn 39 The Communication Chasm and Prostate Cancer Diagnoses: An Ethnographic Study With African American MenParis Pamela Hawthorn 40 Multiple Biomarkers Independently and Additively Predict Prostate Cancer Pathology OutcomesQuigley David Hawthorn 31 Genomic Hallmarks and Structural Variation in Metastatic Prostate CancerRobinson Troy Hawthorn 41 Targeting Androgen Receptor and the ACK1 Signaling Axis in Castration-Resistant Prostate CancerShiina Marisa Hawthorn 42 High Expression of miR-182 Promotes Prostate Cancer Aggressiveness in African-Americans Compared to CaucasiansShkedi Arielle Hawthorn 37 Identification of Molecular Chaperone Targets for Castration-Resistant Prostate Cancer Treatment

Small Eric Hawthorn 44 Health-Related Quality of Life (HRQoLl) After Disease Progression in SPARTAN: A Phase 3 Trial of Apalutamide (APA) Versus Placebo (PBO) in Men With Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC)

Small Eric Hawthorn 45 Prostate-Specific Antigen (PSA) Outcomes in Patients (Pts) With Nonmetastatic Castration-Resistant Prostate Cancer (Nmcrpc) Treated With Apalutamide (APA): Results From Phase 3 SPARTAN Study

Small Eric Hawthorn 46 SPARTAN, a Phase 3 Double-Blind, Randomized Study of Apalutamide (APA) Versus Placebo (PBO) in Patients (Pts) With Nonmetastatic Castration-Resistant Prostate Cancer (Nmcrpc)

Sun Jinny Hawthorn 43 Resistance to Androgen Deprivation Therapy Leads to Altered Metabolism in Prostate Cancer Cell and Murine ModelsTanaka Yuichiro Hawthorn 47 Functional Significance of Catechol-O-Methyltransferase Gene in Prostate CancerWashington Samuel Cypress 26 MRI-Based Prostate Specific Antigen Density Predicts Gleason Score Upgrade in An Active Surveillance Cohort Xu Melody Hawthorn 48 Genomic Risk Predicts for Molecular Imaging Detected Metastatic Nodal Disease in Prostate CancerXu Melody Hawthorn 49 Single Fraction Brachytherapy as Monotherapy for Early Stage Prostate Cancer: The UCSF ExperienceYang Thao Hawthorn 55 Role of a Novel microRNA Gene at Frequently Deleted Chromosome 8p Region in Prostate Cancer

Zhang Li Cypress 27 Tracking of Long-Term B-Cell Memory Responses Using B-Cell Receptor (BCR) Sequencing in Prostate Cancer (PC) Patients (Pts) Treated With Sipuleucel-T (Sip-T)

Zhao Ning Cypress 28 Exploiting the Labile Fe(II) Pool To Image Prostate Cancer With PET

Zuniga Kyle Hawthorn 53 The Association Between Cribriform Pattern and Adverse Pathology at Radical Prostatectomy in Prostate Cancer Patients Initially on Active Surveillance

Zuniga Kyle Hawthorn 54 Trends in Complementary and Alternative Medicine Use Among Prostate Cancer Patients

Podium Session 2

Clinical and Genomic Hallmarks of Treatment-Associated Small Cell Neuroendocrine Carcinoma: Data from the West Coast Prostate Cancer Dream Team Rahul Aggarwal, Paul Lloyd, Jiaoti Huang, Tomasz Beer, Li Zhang, George Thomas, Lawrence True, Joshi Alumkal, Verena Friedl, Alana Weinstein, Rob Reiter, Matt Rettig, Primo Lara, Martin Gleave, Adam Foye, Denise Playdle, Felix Feng, Kim Chi, Josh Stuart, and Eric Small Background: Genomic alterations in DNA repair genes are present in approximately 20-30% of patients with mCRPC. t-SCNC may be increasing in prevalence and is associated with adverse clinical outcomes. Potential overlap between these two subsets of mCRPC was assessed. Methods: Eligible patients (pts) underwent a metastatic tumor core needle biopsy at one of 5 centers. Tumor tissue was sent for consensus pathology call (JH, GT, LT) and targeted next-generation DNA sequencing. Chi-square test was used to compare the frequency of genomic alterations in select DNA repair genes (BRCA1, BRCA2, ATM, CDK12, RAD51, PALB2, FANCA, CHEK2, MLH1, MSH2, MLH3, and MSH6) between tumors with versus those without small cell histology. Frozen tissue from the same metastatic tumor underwent RNA sequencing (RNA-seq). Pts were prospectively followed for overall survival (OS). Results: 119 consecutive biopsies with sufficient tumor to permit histologic assessment and targeted NGS were included, including 14 biopsies (13%) with t-SCNC. In the overall cohort, DNA repair mutations were present in 41 biopsies (34%). By histologic subtype, mutations were present in 40/105 (38%) of adenocarcinoma tumors vs. only 1 of 14 (7%) biopsies with t-SCNC histology (p = 0.047). In contrast, TP53 and/or RB1 loss were enriched in the t-SCNC cohort (83% vs. 34%, p = 0.0015). Unsupervised hierarchical analysis of the transcriptome identified small cell-enriched cluster of biopsies (N = 12) that were mutually exclusive of DNA repair mutations and enriched for E2F transcriptional targets on RNA-seq. The presence of t-SCNC histologic differentiation was associated with worse OS from date of mCRPC (median OS 44.5 vs. 36.6 months; log-rank p = 0.027); the presence of DNA repair mutations was not (p = 0.734). Conclusions: Tumors harboring DNA repair pathway mutations or t-SCNC differentiation may represent distinct disease subsets of mCRPC with differing clinical outcomes. Added together, the two subsets account for up to 40% of all mCRPC. Independent prospective validation of these findings is warranted.

Podium Session 1 Intraoperative Imaging of Microscopic Margins in Prostate Cancer Using a Microfabricated Fluorescence Imaging “Skin” with Targeted Molecular Agents Efthymios P. Papageorgiou, Hui Zhang, Bernhard E. Boser, Catherine Park, Mekhail Anwar

Background: Patients with high risk prostate cancer undergoing prostatectomy are at risk for microscopic residual disease (MRD) in both the tumor bed (i.e. a positive margin) and lymph nodes. Despite advances in both multiparametric MRI and PSMA-PET, intraoperative visualization of MRD remains elusive: the highly penetrating photons essential for PET imaging contribute significant background for intraoperative hand-held detectors, masking the small signal needed for MRD detection. The self-limiting depth penetration of light reduces background while conveying useful information about the tissue surface and several millimeters below, making fluorescently-tagged molecularly targeted agents ideal for intraoperative margin and node imaging. However, current imagers suffer from lack of mobility and sensitivity due to rigid optics and are (1) limited to millimeter-scale tumor foci numbering in the 100s of millions of cells – far above the few hundred cell sensitivity needed for accurate margin imaging, and (2) fiber optic approaches face a fundamental tradeoff between flexibly and imaging area, precluding comprehensive imaging of a minimally invasive tumor cavity. Here we explore the potential of a targeted anti-PSMA antibody integrated with a custom millimeter-scale imaging “skin” capable of coating surgical instrumentation, to image both tumor margins and occult lymph involvement in prostate cancer in a real-time intraoperative setting.

Methods: Leveraging our custom 4.7mm X 2.25mm, 2880-pixel fluorescence contact imager with 55µm pixel pitch, and scalable architecture, PSMA-overexpressing cells cultured in 3D matrigel, labeled with anti-PSMA antibody, are imaged and quantified to determine imager sensitivity to physiologically relevant levels of PSMA expression. Imaging through blood and tissue was evaluated by imaging 3D clusters of tumor cells through an opaque 250um-layer of blood and over increasing imager-cell separations mimicking the perivascular fat encasing lymph nodes. Imager sensitivity to tumor cells was quantified using a signal to noise ratio (SNR), a logarithmic scale. The minimum number of detectable cells is defined as cell cluster size giving an 𝑆𝑁𝑅𝑑𝐵 = 20𝑙𝑜𝑔10(𝑆/𝑁) ≥ 10𝑑𝐵, where S is the signal from a cell cluster and N is the noise (from both optical and electronic sources). In vivo tumor resection is also performed using a receptor positive mouse model system coupled and systemic labeling with a fluorescently-tagged molecular imaging agent coupled with our custom imager.

Results: Sensitivity (SNR) of our imager is characterized over a broad range of cluster sizes and spacing using prostate cancer cell lines with our custom imager, and compared against the gold-standard, an inverted fluorescence microscope. Dilutions of a relatively thick, highly scattering and nearly opaque 250µm layer of blood reduce intensity (SNR) by a maximum of 10dB, but do not significantly degrade spatial resolution, as scattering, which dominates over absorption, essentially trading signal for background. Proof of concept for surgical resection is demonstrated by imaging a small foci of MRD during mouse resection not visible by eye, but readily detect by our imager in < 0.1s. Conclusions: Given an SNR of 10dB for 100-cells, with a 32dB anticipated improvement with our next generation device, our goal of single cell detection is achievable. We demonstrated in vivo resection identifying MRD. This will help guide precision, individualized surgery and, if necessary, post-operative radiation.

Poster Session An SMS-based Periprocedural Intervention for Patients Undergoing Prostate Biopsy: Impact on Clinical Utilization, Patient/Provider Communication, and Patient Satisfaction Ashwin S. Balakrishnan1, Hao G. Nguyen1, MD, PhD, Katsuto Shinohara1, MD, Peter R. Carroll1, MD, MPH, Anobel Y. Odisho1, MD, MPH 1 Department of Urology, University of California, San Francisco, Helen Diller Family Comprehensive Cancer Center

Introduction/Background: Inadequate patient education and preparation for prostate biopsy leads to care delays, cancelled appointments, decreased patient satisfaction, and costs to the healthcare system. Mobile health (mHealth) approaches can improve outreach and the patient experience. We developed and deployed a text message-based (SMS) reminder and education program for patients scheduled for transrectal prostate biopsy and evaluated the impact on cancellation rates, communication frequency, and patient satisfaction.

Methods/Materials: We developed an SMS-based program with seven reminders containing links to web-based content and surveys sent over an 18-day period (14 days before through 3 days after prostate biopsy). Messages contained educational content and reminders regarding MRI for fusion biopsy, antibiotic adherence, anticoagulation management, and enema use. Demographic information, appointment cancellations/changes statistics, and patient/provider communications were collected between a 12-week pre-launch period (2/1-4/30/18) and a 10-week pilot period (6/1-8/10/18). Patients satisfaction was evaluated in the pilot cohort.

Results: There were 333 scheduled and 224 completed appointments in the pre-launch period, compared to 295 scheduled and 168 completed appointments in the post-launch period. There were no differences in patient age, race/ethnicity, distance traveled to clinic or rural/urban status. There was a higher rate of appointment cancellations or changes in the pilot period compared to pre-launch (42.4 vs 32.1%, p =0.02). However, there were fewer cancellations within 7 days (9.2% vs. 14.7%, p = 0.04) and fewer same-day cancellations (0.7% vs. 5.1%, p<0.01). The median time between cancellation and appointment date improved from 8.8 days (IQR, 1.9-21.9) to 27.9 days (IQR, 8.0-46.9) post-launch (p<0.01) (Figure 1). After launch, there were also fewer secure messages (6.8 vs 3.7, p <0.01) and telephone interactions (1.1 vs 0.7, p <0.01) per patient. Mean patient satisfaction was 4.7 out of 5 (SD 0.6) for the SMS program and 4.8 (SD 0.4) for overall care. 76.5% of patients felt the number of reminders was just right, and 23.5% felt it was too high.

Conclusions: An SMS-based peri-procedural outreach program significantly lowered last-minute appointment cancellations, and was associated with decreased secure messages and phone calls and with high patient satisfaction scores. This leads to fewer under-utilized procedure appointments, more efficient scheduling, decreased inbox and phone call burden, and high patient satisfaction.

Poster Session Improving Prostate Cancer Outcome Prediction Through Noninvasive exRNA Assessment: Year 2 Update Cassandra D. Belair, Brandon Chu, Jacob Freimer, John Neuhaus, Robert Blelloch, Matthew Cooperberg Background: Active surveillance (AS) is arguably the most important advance in localized prostate cancer care in the past decade. In recent analyses, up to 50% of patients with low-risk disease are choosing AS over intervention, figures which represent a major shift in care. While AS is usually limited to low-risk patients, more recent evidence from our group and others suggest that AS could safely be extended to intermediate-risk patients. Our goal is to identify and validate across multiple clinical scenarios a novel circulating biomarker signature that can refine patient stratification in patients eligible for active surveillance, enabling the delivery of aggressive interventional therapy to the subset of patients who are most likely to benefit. We hypothesize that plasma miRNAs can help more definitively stratify the severity of disease, above and beyond current clinicopathological parameters. To test this hypothesis, we will first focus on men whose tumors are low-risk by standard clinical criteria. We will measure pre-surgical plasma miRNA levels and determine whether statistically significant associations exist between these and the post-surgical upgrading/upstaging endpoint. Methods: Plasma samples were collected prospectively at the time of diagnosis by the tissue banks at UCSF and the University of Washington. Radical prostatectomy patients with banked plasma have been identified. Controls were defined as low-risk (pT2, pGS 3+3) at final pathology. Cases were defined as upgrading and/or upstaging (pT3-4 and/or pGS >4+3) at final pathology. RNA was prepared from 200ul archived plasma from each patient using commercially available kits. Sequencing libraries were prepared using methods optimized for low input RNA and multiplexing fifty samples per library. Sequencing was performed by the UCSF CAT core on the Illumina HiSeq4000. Samples were then demultiplexed and adapter trimmed using CutAdapt then mapped using STAR to high confidence mature miRNA sequences obtained from miRBase. Bioinformatic and Statistical analysis was performed using various R packages freely available on CRAN.

Results: We will present our progress at the end of year two of this grant. We have miRNA sequencing data from approximately 400 archived patient samples. Exploratory analyses of the sample and sequencing quality will be presented. Preliminary statistical approach using

the “discovery dataset” to determine whether miRNAs add value to the current diagnostic panel of pathology scores in stratifying low-risk patients will be presented.

Poster Session Role of Exosomes in Inducing Neuroendocrine Differentiation in Advanced Prostate Cancer Divya Bhagirath, Thao Yang, Shahana Majid, Rajvir Dahiya, Yuichiro Tanaka and Sharanjot Saini Department of Urology, Veterans Affairs Medical Center and University of California San Francisco, San Francisco, USA Background: Neuroendocrine prostate cancer (NEPC) is an aggressive variant of advanced prostate cancer (PCa) present in ~30% of metastatic castration-resistant tumors, often emerging as a result of AR-targeted therapies such as enzalutamide, via neuroendocrine differentiation (NED). Owing to NED, tumors show neuroendocrine (NE) features with the expression of neuronal markers such as enolase 2 (ENO2), chromogranin A (CHGA) and synaptophysin (SYP). Clinically, NEPC manifests as the presence of visceral metastatic disease, low serum PSA levels relative to disease burden and limited response to AR signaling inhibitors. The molecular basis of NED/NEPC are poorly understood. We propose that in addition to cell intrinsic genetic determinants of NED, tumor exosomes are important to facilitate neuroendocrine differentiation of prostate tumors via horizontal transfer of functional NE factors and regulatory microRNAs (miRNAs) to recipient cells. Methods: Exosomes were isolated from cell culture models of PCa NED followed by (i) small RNA-next generation sequencing (NGS), (ii) Western blot analyses for oncogenic factors to identify novel regulators that play a role in exosome-mediated intercellular communication underlying NED. Exosome isolation reagent was used for exosome isolation as per manufacturer’s instructions. The integrity of exosomal preparations was confirmed by nanoparticle tracking analysis (NTA). cDNA libraries were generated from purified small RNA (100 ng), index libraries were equally pooled and sequenced. Further, in vitro ‘uptake experiments’ were performed with labelled exosomes isolated from neuroendocrine cell line models and incubation with parental LNCaP cells followed by gene expression profiling by real time PCR and immunoblot analyses. Results: Key oncogenic NE factors and oncogenic miRNAs are released in exosomes by cancer cells induced to undergo neuroendocrine differentiation. These oncogenic factors, via their exosomal uptake, act in a paracrine manner on neighboring non-NE cancer cells leading to initiation of NE-like alterations. Summary/Conclusion: Exosome-mediated intercellular communication is crucial to induction of neuroendocrine differentiation states in PCa.

Poster Session A Pilot Feasibility Study for Improving Patient Access to Cancer Clinical Trials (Impact Study) Hala T Borno, Dana Dornsife, Erin Miller, Paige Wycoff, Eric Small, Li Zhang, Sylvia Zhang, Robert Johnson, Darcy Spicer Background: Disparities in cancer clinical trial participation persist. To date, an effective and scalable model for mitigating contributors to cancer disparities, such as out of pocket costs associated with trial participation, has not been developed. The primary objective of this study is to assess the feasibility of a multi-site study evaluating whether a sliding scale financial reimbursement program (FRP) improves accrual among racial/ethnic minorities to cancer therapeutic clinical trials. Methods: This pilot study will be conducted at UCSF and USC. Over a period of 3 months, each site will accrue 66 participants, for a total of 132 participants overall. Participants will be approached for iMPACT at the time of therapeutic clinical trial (TCT) discussion. Those who consent to iMPACT will be randomized to receive an additional follow-up call about the FRP (intervention/outreach) or no follow-up (usual care). Participants will be surveyed at the time of consent to iMPACT and interviewed 30 days after consent to iMPACT. Participants who choose to enroll in a TCT will be given an additional survey 30 days after consent to the TCT. All participants will be followed for the duration of their enrollment in a TCT with an additional 3-month follow-up after TCT completion. Results: To assess feasibility of a multi-site study, researchers will measure the proportion of participants who sign the iMPACT consent in the intervention arm versus the control arm. To test the efficacy of performing intensive outreach for a sliding scale FRP, researchers will measure the proportion of participants randomized to receive intensive FRP outreach who sign consent for cancer TCTs at day 30. To evaluate if a sliding scale FRP Improves accrual for historically underrepresented populations, researchers will measure the proportion of non-white participants accrued to cancer TCTs per month during the study period. Conclusions Results from this study will characterize the role of FRP and intensive outreach in helping accrue underrepresented patients to TCTs.

Poster Session Differential Utilization of Medical versus Surgical Androgen Deprivation Therapy for Metastatic Prostate Cancer

Hala T. Borno, M.D.1, Daphne Y. Lichtensztajn, M.D., M.P.H.3, Scarlett L. Gomez, Ph.D.,

M.P.H.2,3, Nynikka R. Palmer, Dr.PH., M.P.H.4, Charles J. Ryan, M.D.5

1 University of California at San Francisco, Division of Hematology/Oncology 2 University of California at San Francisco, Department of Epidemiology and Biostatistics 3 Cancer Prevention Institute of California 4 University of California at San Francisco, Division of General Internal Medicine 5 University of Minnesota, Division of Hematology/Oncology Objective: Surgical and medical androgen deprivation therapy (ADT) strategies are comparable in their ability to suppress serum testosterone levels as treatment in metastatic prostate cancer but differ in cost and impact on quality of life. Medical ADT is associated with better long-term quality of life due to the flexibility of possible therapy interruption but comes with a higher cumulative cost. We examined if surgical ADT (i.e., bilateral orchiectomy) was differentially utilized by race/ethnicity and other social factors. Methods: We identified patients with metastatic disease at diagnosis through the California Cancer Registry. The association of race/ethnicity with receipt of surgical ADT was modeled using multivariable Firth logistic regression adjusting for age, Gleason score, prostate specific antigen, clinical tumor and lymph node stage, neighborhood socioeconomic status (nSES), insurance, marital status, comorbidities, initial treatment (radiation, chemotherapy), location of care, rural/urban area, and year of diagnosis. Results: We examined 10,675 patients with metastatic prostate cancer: Non-Hispanic (NH) Black (11.4%), Asian/Pacific Islander (8.4%), Hispanic/Latino (18.5%), and NH White (60.4%). In the multivariable model, patients more likely to receive surgical ADT were Hispanic/Latino (OR=1.32, 95% CI 1.01-1.72), from a low nSES (OR=1.96, 95% CI 1.34-2.89) or rural area (OR=1.49, 95% CI 1.15-1.92), and had Medicaid/public insurance (OR=2.21, 95% CI 1.58-3.10). Patients with Military/Veterans Administration insurance were significantly less likely to receive surgical ADT than patients with private insurance (OR=0.34, 95% CI 0.13-0.88). Conclusion: Race/ethnicity, neighborhood SES, and insurance are significantly associated with receipt of surgical ADT. Future research will need to characterize other differences in initial treatments among men with advanced prostate cancer based on race/ethnicity and aim to better understand what factors drive the association between surgical ADT among men of Hispanic origin or from low nSES. Key words: Metastatic prostate cancer, orchiectomy, androgen deprivation therapy, cancer disparities

Poster Session Trial Library – A Pilot Study to Examine the Feasibility, Acceptability, and Preliminary Estimates of Efficacy for a Clinical Trial Online Matching Website on Patient-Prompted Conversations Regarding Prostate Cancer Clinical Trials Hala T Borno, Brian Bakke, Anke Hebig Prophet, Yoon-Ji Kim, Jan Yeager, Jessica Chao, Pelin Cinar, Christy Boscardin, Ralph Gonzales, Celia P Kaplan, Eric J Small

Background: The ability of clinical trials to effectively translate into therapeutic interventions for the general population can be limited by a lack of representative participants in these trials. There is an acute need to develop informational resources regarding cancer clinical trials that meet the linguistic and literary needs of underrepresented and vulnerable populations. Current practices utilize a one-size fits all approach that may be inaccessible to a diverse audience. Given the increased utilization of online resources for health information among a diverse patient population and across age groups, we developed a website, called Trial Library, which may serve as a prostate cancer clinical trial matching tool. We hypothesize that use of the internet-based clinical trial matching tool in clinic will increase the number of patient-initiated conversations with physicians about clinical trial options, and by extension, improve enrollment to therapeutic cancer clinical trials among a diverse participant pool. Procedures/Design/Aims: This is a non-randomized non-therapeutic interventional pilot study. The study will be performed in genitourinary medical oncology clinic at UCSF Helen Diller Family Comprehensive Cancer Center. We will measure the feasibility and acceptability of a clinical trial matching tool which includes two unique features: user validated Clinical Trial content and navigation schema. We will measure preliminary estimates of efficacy of the online clinical trial matching tool in triggering patient-promoted conversations regarding clinical trials, and ultimately clinical trial enrollment. Discussion: Trial Library is designed to be an accessible informational resource for patients from diverse levels of health literacy and English proficiency. This pilot study will provide important preliminary results to inform the design of the website and subsequent testing outside of clinic. Additionally, this website will ultimately serve as a model for clinical trial information that can be translated across tumor types.

Podium Session 3 Screens to Identify Rationale Drug Combination Strategies for Advanced Prostate Cancer Hui Yu1, Young Sun Lee2, Yupu Liang3, Charles L. Sawyers2,4, Rohit Bose1,5,6* Background: among individual patients with advanced, metastatic prostate cancer, a wide range of responses are observed to second generation anti-androgens. Many features measurable in patients have been demonstrated to correlate with altered clinical responses, including genomic events (AR amplification), alternatively spliced mRNAs, protein subcellular localization (e.g. nuclear vs cytoplasmic AR-V7), as well as the number of prior lines of systemic therapy. To fully leverage this information, we aim to functionally identify modulators of anti-androgen response. Methods: To achieve this, VCaP cells were infected with a whole genome lentiviral library containing 60,000 unique shRNAs to achieve 1 shRNA infected per cell, then passaged in the presence or absence of enzalutamide. Their genomes were subjected to deep sequencing at different time points, to quantify the relative abundance of lentiviral-integrated sequences corresponding to shRNAs, allowing one to infer the effect of each unique shRNA on cellular proliferation under the specified condition. Subtractive analysis between conditions enables housekeeping genes to be eliminated from the analysis, and a focus placed on genes that specifically modulate the proliferation of cells in the presence of anti-androgens. These potential hits were further filtered by genes known to be altered from human tumor profiling data. Results: from this anti-androgen screen, we have identified a handful of exciting epigenetic targets that when inhibited, lead to altered proliferative responses of tumor cells. These are currently undergoing validation. Conclusion: The epigenetic state of prostate tumor cells likely controls their degree of sensitivity to androgen receptor antagonists.

Poster Session Integrating Genetic Information with Machine Learning to Predict Which Prostate Cancer Cases Should Not Be Immediately Treated Taylor B. Cavazos, Nima C. Emami, Janet E. Cowan, Matthew R. Cooperberg, June M. Chan, Peter R. Carroll, John S. Witte Background: Advances in screening and treatment have greatly improved five-year survival for prostate cancer. Approximately 90% of men diagnosed with prostate cancer are characterized as having low-risk disease. Although important strides have been made towards accurate patient stratification, less has been done to help guide treatment decisions for this low risk group. While some patients may experience upgrading or upstaging (UGUS) of their tumor and benefit from more invasive treatments such as radical prostatectomy, others may have slower growing tumors that are unlikely to result in any increased risk of morbidity or mortality due to this disease. In this case, active surveillance may be the best treatment option as surgery or radiotherapy have a number of potential side effects. Methods: To better stratify and guide the medical decisions of low-risk prostate cancer patients, we created a random forest (RF) classifier to predict the likelihood of future UGUS of prostate cancer at diagnosis and prioritize germline variants associated with this phenotype. We analyzed genetic and clinical data from a cohort of 864 low-risk prostate cancer patients whom underwent radical prostatectomy 6 months after initial diagnosis. At radical prostatectomy, 448 patients did not experience a change in tumor stage or grade. The patients were split into a training (n = 690) and validation (n = 174) cohort. Results: With clinical factors alone, including Gleason score and prostate specific antigen levels (PSA), UGUS was predicted with 70% accuracy using a generalized linear model. To identify prognostic signals independent of clinical factors, approximately 700,000 single nucleotide polymorphisms (SNPs) were genotyped in these subjects. Due to the large number of features and small sample size of this study, RF is an appropriate method for analysis given its ensemble learning approach that helps prevent model overfitting. By integrating genetic information and using a RF classifier, we will show the potential increase in prediction accuracy over the clinical model alone. This work was supported in part by the US Department of Defense Prostate Cancer Research Program (W81XWH-13-2-0074).

Poster Session Quantification of Cell-Free DNA Levels as a Potential Screening and Diagnostic Marker for Localized and Metastatic Prostate Cancer Emmalyn Chen1, 2, Clinton L. Cario2, Lancelote Leong2, Karen Lopez3, Patricia S. Li3, Erica Oropeza3, Imelda Tenggara3, Janet Cowan3, Jeffry P. Simko3, 4, June M. Chan2, Terence Friedlander5, Pamela L. Paris3, Peter R. Carroll3, and John S. Witte2, 3 1 PhD Degree Program in Pharmaceutical Sciences and Pharmacogenomics, University of California, San Francisco, California 2 Department of Epidemiology and Biostatistics, University of California, San Francisco, California

3 Department of Urology, Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, California 4 Department of Anatomic Pathology, University of California, San Francisco, California 5 Division of Hematology/Oncology, University of California, San Francisco, California

Early cancer diagnosis, especially while the disease is still localized and before symptoms appear, results in significantly higher survival rates compared to late-stage diagnosis. Cancer is complex, and its underlying mechanisms are still not well-understood, which makes early detection a challenging task. Cell-free DNA (cfDNA) is a promising tool for directly detecting the tumor DNA that is shed into the bloodstream. Measurement of total cfDNA levels is a potentially more affordable and quicker method compared to sequencing. The objective of this study is to determine if cfDNA levels are able to distinguish between healthy individuals from patients with localized or metastatic castrate-resistant prostate cancer (mCRPC).

This study included samples from 146 individuals: 20 healthy donors, 112 patients with localized prostate cancer who underwent radical prostatectomy (RP) at UCSF, and 14 mCRPC patients. Whole peripheral blood was collected in EDTA tubes or PAXgene Blood ccfDNA tubes. Samples collected in EDTA tubes were processed within 2 hrs of collection, and samples collected in PAXgene tubes were processed within 7 days. All samples were centrifuged for 10

mins at 1,900g at 20C. The supernatants were then centrifuged for 10 mins at 16,000g and

stored at 4C prior to extraction within 48 hrs. Extraction was performed on double spun plasma using Qiagen QIAamp Circulating Nucleic Acid Kits. After extraction, the concentration and fragment length distribution were measured with a Bioanalyzer 2100 Instrument (Agilent Technologies). Clinicopathologic characteristics are being collected for all patients. Comparisons of cfDNA levels between two groups were assessed with a Welch’s t-test due to the potential for unequal variances.

The samples used for analysis included 125 individuals to date, with 20 healthy donors, 95 patients with localized disease, and 10 mCRPC patients. The remaining samples are currently being processed. The mean cfDNA level was 1.4 ng/ml (SD, 1.27 ng/ml) for healthy donors, 2.90 ng/ml (SD, 2.66 ng/ml) for patients with localized disease, and 3.6 ng/ml (SD, 3.3 ng/ml) for mCRPC patients. The difference in the mean cfDNA level between the two groups was statistically significant (p < 0.005). Among the 95 patients with localized disease, cfDNA level was not associated with age, PSA, and Gleason score, suggesting the potential for this marker to be independent of factors that are typically used to assess disease burden and risk of progression. Including the remaining mCRPC samples will increase the power to detect a difference between healthy individuals and mCRPC patients. Follow-up samples are being collected for all patients, and future analysis includes investigating the association between cfDNA levels prior to RP surgery and biochemical recurrence with a Cox proportional hazards regression model.

Podium Session 1 Characterizing and Targeting the MYC Surface-ome in Advanced Prostate Cancer Wentao Chen, Hao G. Nguyen, Lingru Xue, Duy Nguyen, Yun Mou, Rahul Aggarwal, Davide Ruggero, James A. Wells Background: Prostate cancer (PC) is the most common malignancy in men and a major cause of death. The 5-year survival rate for distant stage PC falls below 30%. Focal amplification of the region spanning MYC is a genetic defect commonly observed in primary PC, and it happens more frequently in lethal mCRPC. Because of the difficulties in targeting Myc directly with small molecules, therapeutic approaches targeting Myc-driven cancer is limited and underdeveloped. Immunotherapies have come of age as some of the most powerful therapies in cancer. However, there is a dearth of validated membrane protein targets in Myc-driven cancers, especially in prostate cancer, hindering the ability to develop antibody drug conjugates, immunotherapeutics, and other biologic agents. We hypothesize that MYC-induced transcriptional remodeling of the prostate cell surface membrane proteome (surface-ome) will yield therapeutic targets amenable to development of high-affinity antibodies (Abs) with potent anti-tumor activity. Methods: We use isogenic prostate cell lines and PC PDX models to identify the most differentially up-regulated surface targets detected by mass spectroscopy (MS). We also characterize surface-ome remodeling using Fab-Phage display and next-generation sequencing (PhaNGS). We use a high-throughput automated platform developed in our laboratory to isolate phage-derived antibodies from high diversity Fab-phage libraries for MYC-induced surface antigens. We further characterize the specificity of newly derived Abs by immunofluorescence assays. Results: We have observed dramatic surface-ome remodeling by Myc overexpression in PrEC-LHS, RWPE-1 and 22Rv1 prostate cell lines, supporting our hypothesis and the feasibility of the approach. We discovered that the surface-ome remodeling by Myc overexpression is very context-dependent. We find that TNFRSF10B is overexpressed in all 3 Myc overexpressing cell lines, and Myc overexpressing PrEC-LHS and RWPE-1 cells develop strong response to TRAIL treatment. We have established a robust pipeline to characterize the surface-ome of PC PDX models by MS-based membrane proteomics, and we have discovered that two metastatic PDX models share a distinct surface-ome feature that is different from a primary PDX model for PC. Conclusions: Through profiling the surface-ome changes induced by Myc, we confirmed that Myc overexpression is able to dramatically remodel the surface-ome in prostate cells. We have discovered the Myc-induced TNFRSF10B overexpression induced higher sensitivity to TRAIL treatment in prostate cells. We are on the way to characterize the surface-ome of low and high Myc PDX models for PC.

Podium Session 2 Genomic Drivers of Clinical Outcomes in Metastatic Castration-Resistant Prostate Cancer William S. Chen1,2, Shuang G. Zhao3, Rahul Aggarwal2, Li Zhang2, David A. Quigley2, Paul Lloyd2, Adam Foye2, Denise Playdle2, Vishal Kothari2, Christopher A. Maher4, Felix Y. Feng2, Eric J. Small2 1Yale University, 2University of California San Francisco, 3University of Michigan, 4Washington University in St. Louis

Background: Metastatic castration-resistant prostate cancer (mCRPC) is the most lethal form of prostate cancer. There have been a number of recent efforts to identify genomic drivers of the disease, but the clinical implications of these genomic variants remain unknown. We conducted the first prospective cohort study (n=101) to date, using whole genome sequencing (WGS) and whole-transcriptome RNAseq to analyze the association between key driver gene alterations and overall survival. We then performed exploratory expression-based analyses to identify potential mechanisms of enzalutamide resistance in mCRPCs. Methods: We performed whole genome sequencing (WGS) on tissue biopsy samples from 101 mCRPC patients enrolled in the multi-institutional Stand Up 2 Cancer / Prostate Cancer Foundation-funded West Coast Prostate Cancer Dream Team project. The resulting mutation, copy number, and structural variant calls were integrated to determine functional copy number status of candidate genes for downstream clinical analyses. To nominate potential mechanisms of enzalutamide resistance, we performed gene set enrichment analysis. Univariate and multivariable survival analyses were performed to assess the prognostic significance of candidate genomic events. Results: We found that RB1 loss was associated with poor overall survival (median survival 14.1 months vs. 42.0 months, p<0.001). When we compared enzalutamide resistant versus naïve samples using gene set enrichment analysis, we identified the Wnt/beta-catenin pathway as the top differentially modulated pathway in enzalutamide-resistant patients. Furthermore, CTNNB1 (beta-catenin) activating mutations were exclusive to enzalutamide-resistant patients (p=0.013) and predictive of poor overall survival (median survival 13.6 months vs. 41.7 months, p<0.001). Conclusions: Our study is the first to use solid-tumor sequencing results in conjunction with longitudinal clinical outcomes to investigate the genomic underpinnings of mCRPC clinical outcomes. This work sheds new light on clinically important genomic alterations in mCRPC and paves the way for personalized treatment regimens in metastatic prostate cancer.

Poster Session IMPACT: A Phase II Trial of Combination Immunotherapy in Patients with Metastatic Cancers and CDK12 Mutations Melissa Reimers1*, Jonathan Chou2*, Stephanie Daignault-Newton1, Arul Chinnaiyan1, Felix Feng2,3, Eric Small2, and Ajjai Alva1. 1Division of Hematology/Oncology, Department of Medicine, University of Michigan, 2Division of Hematology/Oncology, Department of Medicine, University of California, San Francisco, and 3Department of Radiation Oncology, Helen Diller Comprehensive Cancer Center, University of California, San Francisco Background: Metastatic castrate-resistant prostate cancer (mCRPC) remains a lethal disease. Recent tumor sequencing studies have provided a deeper understanding of the mutational landscape of mCRPC, which in turn has provided novel treatment options. For example, aberrations in DNA repair defect genes, including BRCA2 and ATM, in mCRPC patients has led to several trials evaluating the efficacy of PARP inhibitors. Our recent work has identified a new subtype of prostate cancer, characterized by biallelic, somatic loss-of-function mutations in cyclin-dependent kinase 12 (CDK12). These mutations are found in 5-8% of mCRPC patients, and display widespread focal tandem duplications (i.e. copy-number gains), consistent with global genomic instability. Through computational analysis, we previously showed that these copy-number gains resulted in gene fusions that created chimeric, open reading frames to generate novel fusion neoantigens. In accordance with this hypothesis, CDK12-mutated prostate cancers had higher levels of T-cell infiltrates in the metastatic tumor biopsies. While recent clinical trials have shown that only 5-10% of patients with advanced prostate cancer respond to anti-PD1 therapy, we found that two of four patients (50%) with CDK12 mutations treated on anti-PD1 therapy had a response. This suggests that CDK12-mutated prostate cancer may be a genomic subtype that benefits from immunotherapy. Therefore, we have proposed a single-arm, phase II study of nivolumab (anti-PD1) and ipilimumab (anti-CTLA4) combination therapy followed by nivolumab monotherapy in patients with metastatic CDK12-mutated cancers.

Methods: IMPACT is a phase II, multi-center study of nivolumab and ipilimumab combination therapy in patients with tumors that harbor CDK12 mutations. This study will include a cohort of mCRPC patients who have evidence of disease progression after at least one standard therapy (cohort A) as well as other advanced metastatic cancer patients, including ovarian cancer (cohort B). Patients will receive four cycles of combination nivolumab and ipilimumab every 3 weeks, followed by maintenance nivolumab at a flat dose of 480 mg IV given every 4 weeks, up to 10 doses. The primary objective of the study is to determine the efficacy of checkpoint inhibitor immunotherapy in patients with cancers harboring loss of CDK12. The secondary objectives include safety and tolerability of this combination and quality of life assessment. The exploratory objectives include assessing immune cell infiltrates, as well as determining the genomic and transcriptomic changes with immunotherapy in this genomic subtype of cancer.

Results and Conclusions: This trial is currently under review and is not yet open or accruing patients. We anticipate opening this trial within the next two months at three academic institutions in the US, including the University of Michigan and UCSF.

Podium Session 3 Interactome and Therapeutic Targeting of CDK12 Mutated Prostate Cancer Jonathan Chou1, Troy Robinson2, Rajdeep Das2, Margaret Soucheray3, Alan Ashworth1, Minkyu Kim3*, and Felix Y. Feng2*

1 Department of Medicine, Division of Hematology/Oncology, Helen Diller Family Comprehensive Cancer Center, 2 Department of Radiation Oncology, University of California, San Francisco, and 3 Department of Cellular Molecular Pharmacology, University of California, San Francisco Background: Metastatic disease remains the leading cause of death for patients with castrate-resistant prostate cancer (CRPC). Unfortunately, current treatments for mCRPC extend the lives of patients only by months. Studies by our group and others have shown that genomic analysis can uncover recurring genetic drivers of cancer, which can lead to new precision oncology approaches. For example, we and others have identified a new subtype of prostate cancer (PC) characterized by biallelic, somatic inactivation of cyclin-dependent kinase 12 (CDK12), which is found in 5-10% of mCRPC patients. Most CDK12 mutations (83%) are truncating and result in the loss of the kinase domain. Interestingly, loss of CDK12 associates with a distinct pattern of genomic instability of focal tandem duplications that leads to increased gene fusions and genomic instability. However, whether this genomic instability increases susceptibility to PARP inhibitors or other targeted therapies in PC remains unknown. Furthermore, little is known about the role of CDK12 in maintaining genomic stability, its protein interactions and downstream targets, which will be critical to tackle this aggressive subtype of PC. Methods: Using CRISPR/Cas9, we generated the first CDK12 knockout (KO) and knockdown PC cell line models in LnCAP, C42B, 22RV1 and PC3 cells. We tested the sensitivity of our CDK12-deficient PC cells to multiple PARP inhibitors, including talazoparib, olaparib and rucaparib. In addition, we have generated cell lines containing an inducible FLAG-tagged CDK12 wild-type or kinase dead mutant (KDmut). Using affinity purification and mass spectrometry (AP-MS), we plan to define the protein interactions of CDK12, as well as characterize the phospho-signaling signatures associated with CDK12 using phospho-proteomic analysis. Results: We have successfully generated CDK12 knockdown and KO PC models, which have been extensively validated by sequencing and western blotting. CDK12 knockdown decreases homologous recombination repair (HRR), and increases sensitivity to multiple PARP inhibitors. In addition, CDK12 loss increases cell migration and invasion. Interestingly, CDK12-deficient cells have a prolonged S-phase and form smaller tumors compared to control cells in vivo. Using an inducible, FLAG-tagged CDK12, we have performed AP-MS to identify its protein interaction partners. Conclusion & Future Directions: CDK12 silencing or deletion increases sensitivity to multiple PARP inhibitors in PC cells and decreases HRR. We have identified putative protein interaction partners, which we are in the process of validating. Future work is aimed at defining the phospho-signaling signatures in the setting of CDK12 loss. In addition, we are planning a CRISPR-based screen of the interactome candidates to identify novel synthetic lethal interactions in the context of CDK12-deficiency. Funding: J.C. is supported by an NCI T32 training grant (CA108462), a postdoctoral fellowship from the A.P. Giannini Foundation, as well as the Rosenberg Fellowship in Genitourinary Oncology.

Podium Session 4 When Can Active Surveillance Be Less Active?: Prediction of Long-Term Non-Reclassification for Men with Low-Risk Prostate Cancer Matthew R. Cooperberg, Anna V Faino, Lisa F. Newcomb, Peter R. Carroll, James D. Brooks, Michael Fabrizio, Martin E. Gleave, Todd M. Morgan, Atreya Dash, Peter S. Nelson, Ian M. Thompson, Andrew A. Wagner, Daniel W. Lin, and Yingye Zheng: San Francisco, CA Background: Active surveillance is increasingly consistently endorsed as the preferred management strategy for most men with low-risk prostate cancer. However, nearly all active surveillance protocols entail frequent PSA testing and follow up prostate biopsies every 1 to 2 years. For many men with biologically indolent tumors, this regimen is overly intense, and exposes men to the discomfort, risks, and costs of repeated biopsies. We aimed to determine if some men can be safely selected for a less intense surveillance regimen by predicting the probability of non-reclassification over the next 4 years of surveillance. Methods: Data were collected from men enrolled in the multicenter Canary Prostate Active Surveillance Study (PASS), under which PSAs are collected q3 months and biopsies performed within 12 months of diagnosis and then every 2 years. For inclusion in this study, men had to have undergone at least one follow up biopsy, and those with Gleason grade group >1 at diagnostic biopsy were excluded. Reclassification was defined as increase in Gleason grade group on subsequent biopsy; those without reclassification were censored at last study contact, treatment or 2 years after last biopsy. A dynamic risk prediction model based on a Cox regression with robust variance estimates was used to construct and test a model predicting non-reclassification. Results: Of 1082 men included, 362 (33%) reclassified and the remaining were censored. The final regression model included percent of biopsy cores involved, prior biopsy history, time since diagnosis, BMI, prostate size, diagnostic PSA, and PSAk (a previously described measure of PSA kinetics). This dynamic risk prediction model was assessed at a measurement time of 1 year after diagnosis, predicting out risk of reclassification at 4 years. Men at lowest and highest deciles of this model-based risk faced 6% (95%CI 0-12%) and 73% (55-84%) risks of reclassification within 5 years. For at least 10% of the men in the cohort, the negative predictive value (NPV) for reclassification was 95% or higher. Conclusions: A substantial proportion of men with low-risk prostate cancer can safely be followed with a de-intensified active surveillance protocol, which would improve both the tolerability and cost-effectiveness of this management strategy.

Poster Session Investigating The Role of the Long Non-Coding RNA Schlap1 in Prostate Cancer Metastasis Rajdeep Das, Julia Dancourt, Anirban Sahu, John R. Prensner, Benjamin Chandler, Qi Cao, Nithin Edara, Udit Singhal, Matthew K. Iyer, Rohit Malik, Arul M. Chinnaiyan, Felix Y. Feng

Background: Prostate cancer (PCa) is the second leading cause of cancer death among U.S. men. Clinically, PCa is a heterogeneous disease in which only a small subset of patients will develop aggressive cancer while most cases are indolent and not life threatening. However, the molecular basis for this dichotomy is not fully understood and, furthermore, few biomarkers exist that can predict aggressive versus indolent prostate cancer to guide clinical management of the disease. Recently, we discovered SChLAP-1 (Second Chromosome Locus Associated with Prostate), a seven-exon, prostate-specific, long non-coding RNA (lncRNA) that is overexpressed in a subset of localized and metastatic cancers. Subsequent studies found that SChLAP1 is significantly prognostic for metastatic progression and lethal disease. SChLAP1 mediates cell invasiveness, in part, by interacting with and abrogating genome-wide binding of the tumor suppressive SWI/SNF nucleosome remodeling complex, a multi-subunit epigenetic modifier that regulate gene expression by reorganizing nucleosomes to alter chromatin architecture. However, the nature of this relationship is poorly understood. Methods: To characterize specific regions of SChLAP1 essential for its function, we generated deletion constructs tiling every 250bp and overexpressed in PCa cells. To test whether this region is important for SWI/SNF binding, we performed RNA immunoprecipitation (RIP) for SNF5, a core subunit of the SWI/SNF complex. Next, we performed RIP for BRG1 and BRM, the enzymatic components of SWI/SNF complex in PCa cells to determine whether SChLAP1 binds specifically to either type of complex. Results: We have shown that a 250bp region near the 3’-end of SChLAP1 mediates its invasive phenotype and coordinates its interaction with the SWI/SNF complex. Additionally, we found that SChLAP1 interacts with BRG1-, but not BRM-, containing SWI/SNF complexes and preferentially decreases BRG1 genomic binding. Finally, SChLAP1’s preference for BRG1 may expose BRM as a therapeutic target in PCa. Conclusion: Taken together, our findings have broad implications for cancer biology and provide additional evidence for the development and use of specific BRM inhibitors for therapeutic treatment of cancer.

Poster Session Linking Clinical Molecular Profiles of Tumors to The Electronic Medical Record Debajyoti Datta, M.D. Ph.D.1, Theodore Goldstein, Ph.D.1, Atul Butte, M.D. Ph.D.1 1Bakar Computational Health Sciences Institute, UCSF, CA Background: The past few years have seen the rapid expansion of tumor molecular profiling in the care of cancer patients. The easy availability of clinical tumor sequencing allows for the rapid and efficient identification of genomic alterations in genes and pathways that can be targeted with an ever-widening array of therapeutic agents. As a result, molecular profiling of tumor genomics is considered to be a cornerstone of precision medicine advances in oncology. However, despite their ever-expanding use in the clinical setting, the overall impact and potential benefit of clinical genomics using tumor molecular profiling remains unclear, and evidence demonstrating improvement in patient outcomes is lacking. Methods: The widespread adoption of tumor molecular profiling over the past few years provides a clear opportunity to examine in a retrospective fashion the potential utility and clinical benefit of clinical genomics testing in cancer patients. By linking these results to “real-word” patient data mined from electronic medical records, we can analyze the impact of the rapid expansion in tumor molecular profiling on the management of cancer patients over the past several years. In this study, we present the development of the UCSF Genomic Test Database that combines clinical genomic results from our internal next-generation sequencing panel, as well as commercially available external clinical genomic panels, with a variety of data points from a de-identified representation of our institution’s electronic medical record (EMR). We have obtained the results of clinical genomic testing as XML files from each testing laboratory and assembled these into a relational database that follows a similar format to our de-identified EMR allowing for linking of the two data types for efficient analysis. Results: The linked data set that we have created allows us to measure the utility of genomic tumor testing for the management of cancer patients. We can analyze the magnitude of positive impact for various tumor types, identify barriers to efficient use of genomic tumor testing, and assess for disparities in the utilization of testing. For example, patient survival data relative to the clinical action of obtaining tumor molecular profiling can be measured. We are able to identify a patient death date for a subset of the patients in our cohort, where we find that a small fraction of patients dies before results return and any action can be taken. Conclusions: The results of our survival analysis suggest that molecular profiling should be considered earlier in the course of disease for certain cancer types such as pancreatic cancer. In the future, we expect that combining information on patient outcomes from “real-world” EMR data with clinical genomic testing will provide the additional opportunity of identifying exceptional responders or non-responders with greater efficiency, which can lead to the discovery of new therapeutic targets and treatments as well as the identification of novel biomarkers for response or resistance to therapy.

Podium Session 4 Neighborhoods Archetypes for Understanding Disparities in Prostate Cancer Mortality Mindy DeRouen, Margaret M. Weden, Juan Yang, Jennifer Jain, Scarlett L. Gomez, and Salma Shariff-Marco Background: Neighborhood factors shape and perpetuate disparities in prostate cancer mortality, but existing approaches have made it challenging to assess the synergistic effects of multiple neighborhood factors on health outcomes. An alternative approach is to define neighborhood archetypes that encompass multiple dimensions within a single classification system and capture meaningful distinctions across neighborhoods. The archetype approach has the potential to identify improved opportunities for intervening and reducing health disparities. Methods: Leveraging a comprehensive database of small-area level data on neighborhood social and built environments (the California Neighborhoods Data System), which includes socioeconomic status, racial/ethnic composition, immigration/acculturation factors, urban/rural status, population density, street connectivity, walkable destinations, food environment, recreational opportunities, traffic density, and green space), we applied latent class analysis (LCA) to develop neighborhood archetypes for 2000 and 2010 California block groups and census tracts. Geographic variability in all-case and prostate cancer-specific mortality by neighborhood archetypes was assessed using multivariable Cox proportional hazard models with geocoded cancer registry data. Results: Goodness of fit statistics identified 5-class and 9-class neighborhood archetype models for census tracts that showed significant associations with mortality after prostate cancer. For the 9-class archetype model, compared to those residing in Upper Middle Class suburban neighborhoods, all other neighborhood types, except High Status, had statistically significant higher mortality (for all cause and for prostate cancer-specific mortality), with Hispanic Small Town neighborhoods having the highest morality (HR=1.39, 95% CI=1.35-1.43 for all cause; HR=1.37, 95% CI=1.29-1.45). Heterogeneity was observed for all cause and prostate cancer-specific mortality by race/ethnicity. For all-cause mortality, among NH Whites and AANHPIs, those residing in Inner City neighborhoods had the highest mortality; for NH Blacks and Hispanics, those residing in Hispanic Small Town neighborhoods had the highest mortality. For prostate cancer-specific mortality, among NH Whites and Hispanics, those residing in Hispanic Small Town neighborhoods had the highest mortality; among NH Blacks and AANHPIs, those residing in Rural/Micropolitan neighborhoods had the highest mortality. Variation by nativity (US- or foreign-born) was also observed. Conclusion: The archetype approach yields insights into how neighborhood characteristics work synergistically to influence prostate cancer mortality. This research contributes a more fundamental understanding of how place affects health and can inform multilevel interventions.

Podium Session 1 Boronated PSMA Neutron Capture Therapy for Treatment of Advanced Prostate Cancer Sinan Wang, Charlie Blaha, Andrew Hong, Miko Fogarty, Raquel Santos, Robert Franks, Henry F. VanBrocklin, Tomoko Ozawa, Robert R. Flavell Background: Prostate specific membrane antigen (PSMA) is a cell surface enzyme which is highly expressed on prostate cancer. Various therapeutic and imaging agents have been developed targeting PSMA, with many showing promise in ongoing clinical trials at UCSF and elsewhere. Therapy with radionuclides has been demonstrated using 177Lu and 225Ac, with examples of successful tumor control but also with side effects including xerostomia and bone marrow suppression. Boron neutron capture therapy (BNCT) is based on the selective uptake of a boron containing drugs into cancer over normal tissue, followed by selective neutron beam exposure to the cancerous regions, initiating the release of cytotoxic alpha particles. In this abstract, we demonstrate the synthesis and initial in vitro cell binding evaluation of a series of boron-labeled PSMA derivatives for development for BNCT of advanced prostate cancer. Methods: A series of urea-based PSMA inhibitors were synthesized, which were functionalized with boronic acid or carborane moieties. A competition radioligand binding assay was developed using LNCaP cells treated with 68Ga-PSMA-11 in the presence of varying concentrations of inhibitory PSMA agents. Results: Boron-functionalized PSMA derivatives 1-6 were synthesized in good yields. In a competition radioligand binding assay, compound 4 demonstrated the highest binding affinity of the boronic acids tested to date (IC50 = 130 nM), while the carborane 1 demonstrated an acceptable IC50 of 363 nM. Conclusions: We have demonstrated the synthesis of a series of boron functionalized PSMA inhibitors for the purpose of BNCT of advanced prostate cancer. In initial cell binding assays, compounds 1 and 4 demonstrated high affinity for PSMA. In vivo biodistribution assays are planned for testing of the optimized derivatives in murine xenograft models.

Podium Session 3 Computational Genomic Characterization of Transitional Cancer Subtypes Verena Friedl, Alana S. Weinstein, Artem Sokolov, Christopher K. Wong, Jiaoti Huang, Rahul

Aggarwal, Joshi Alumkal, Adam Foye, Paul Lloyd, Eric J. Small, Joshua M. Stuart

Background: Cancers change over time both as a natural consequence of uncontrolled

proliferation and in response to evolutionary pressures from treatments. During tumor evolution,

cellular differentiation and transdifferentiation may give rise to transitional cancer types that

reside between well-established subtypes. In prostate cancer, some adenocarcinomas develop

neuroendocrine characteristics and eventually transition into small cell neuroendocrine

carcinoma, a rare and highly aggressive subtype. The ability to classify such transitional states

and characterize genomic events driving the transition is crucial for understanding the

progression of advanced prostate cancer.

Methods: We developed a supervised computational approach to assess potential transitional

tumor types using linear discriminant analysis (LDA) on gene expression data. The method

provides a significance estimate on the confidence of a transitional subtype residing between

two established subtypes. Once an intermediate subtype is detected, we perform master

regulator analysis to predict differentially active regulators between subtypes. We applied this

method to 131 metastatic castration-resistant prostate cancer (mCRPC) samples, including

adenocarcinoma, small cell, and Intermediate Atypical Carcinoma (IAC), a possible transitional

subtype.

Results: LDA confirmed IAC as an intermediate subtype between adenocarcinoma and small

cell with a confidence estimate of p<10-14 (Fig. 1A). Master regulator analysis between pairings

of subtypes/subtype groups identified regulators that may drive the transition and intermediate

state seen in IAC (Fig. 1B).

Poster Session Identification of Small Molecule Inhibitors of a Novel Regulatory Pathway of Immune Checkpoint Proteins Qi Fu, Steven Chen, Lin Liu, Mauro Poggio, James Fraser, Robert Blelloch Background: Immune checkpoint proteins aid the growth of cancer cells by suppressing immune cell activity. In the past decade, we have developed antibodies to inhibit the function of these proteins. Using these therapies, patients that were once incurable have remained remission-free for years after treatment. However, one of the major limitations of current immune checkpoint therapies is their low response rates. Therefore, there is a need for better and more effective drugs. We have discovered a novel pathway regulating immune checkpoint proteins. Genetic inhibition of this pathway decreases tumor burden and increases survival in mice. We hypothesize that it is possible to suppress this pathway using small molecules. Methods: We developed and use a protein-based enzymatic screen plan to identify small molecule inhibitors of a target protein in the pathway. Results: We optimized the enzymatic assay by testing a dose curve of different concentrations of enzymes and different incubation times to find the optimal concentration and time that gives the best signal between positive and negative controls. We show that increasing amounts of enzyme results in increased activity. We also show that addition of EDTA, an inhibitor of the enzyme, decreases activity. Z factor scores were greater than 0.5 for the majority of enzyme concentrations we tested, showing the assay is suitable for high-throughput screening. We performed a pilot screen of 2000 bioactive compounds resulting in ~1% rate of active molecules. Conclusions and Future Directions: We have successfully developed an assay that is suitable for high-throughput screening. Going forward, we plan to expand this screen to 50,000 compounds. Active molecules identified from the pilot screen and 50,000 compound screen will be tested further to select a hit series and develop lead compounds. Successful completion of this study could lead to the discovery of novel immune checkpoint therapies that can act as single agents or in combination with current therapies to improve cancer treatment.

Podium Session 1 A Functional Genomics Approach to Identify New Strategies for Imaging and Treating Metastatic Prostate Cancer

Tom O'Loughlin, Rajdeep Das, Avi Baskin, Felix Feng, Luke Gilbert

Prostate cancer is a leading cause of death in men. Most prostate tumors require androgen for survival or cell growth. Chemotherapeutics that chemically castrate the tumor of androgen signaling have been successful. Unfortunately, most prostate cancers patients treated with androgen deprivation therapies (ADT) will eventually become castrate resistant. Nearly all mortality in prostate cancer is due to metastatic disease that is resistant to deprivation of androgens. Imaging and treating metastatic castrate resistant prostate cancer (mCRPC) is a major clinical and scientific challenge, and we need new therapeutic strategies for treating mCRPC. We present new synthetic biology and functional genomics methods that we have developed to systematically unmask the biology of prostate cancer and to define new therapeutic strategies for metastatic prostate cancer.

We have developed an integrated functional genomics platform for mCRPC. We created the first genome-scale CRISPRi/a functional genomics platform for loss- (CRISPRi) and gain- (CRISPRa) of-function screening in human cells. Functional genomics seeks to experimentally define complex genetic interactions that drive cancer progression and anti-cancer drug response. Additionally, we very recently developed a new type of functional genomics approach for large-scale quantitative mapping of human genetic interactions. We have created a panel of CRISPRi/a mCRPC models enabling new functional genomics studies in prostate cancer. We have recently completed CRISPRi functional genomics screens in models of metastatic castration resistant prostate cancer (mCRPC). These experiments are the first CRISPR functional genomics experiments in mCRPC performed in a robust and quantitative manner enabling us to nominate new therapeutic targets in mCRPC. We have also recently used a CRISPRi screen to identify targets that potentiate the anti-cancer activity of a new theranostic targeting PSMA, which is a high value drug and imaging target in mCRPC. Our preliminary data demonstrate we are uniquely positioned to deploy these functional genomics tools in mCRPC to create new therapies that we hope will power the next big advances in anti-cancer medicines for mCRPC.

Podium Session 1 Systematic Identification of Regulatory Pathways that Drive Prostate Cancer Progression Yikai Luo, Matvei Khoroshkin, Hosseinali Asgharian, Hani Goodarzi Background: Prostate cancer progression is fundamentally a disease of disordered gene expression. As the disease progresses, the cancer cells are continuously reprogrammed, starting from a benign neoplasia all the way to aggressive and metastatic tumors. As a result of this complexity, the typical differential gene expression analyses that are often used to describe changes in cellular states is of limited use. This is because the answer to the question “which genes are modulated” is “almost every gene”. We believe a better understanding of prostate cancer progression can be achieved by discovering the underlying regulatory pathways and processes that are hijacked by cancer cells to achieve dysregulated gene expression. Methods: First, we generated a meta-dataset of prostate cancer progression by re-analyzing and integrating data from GTEx (normal prostate), TCGA (normal and primary prostate adenocarcinoma), and WCDT/UMich (metastatic prostate cancer). We then applied our computational platform named iGET, the Integrated Genomics Exploration Tools, which enables the systematic interrogation of whole-transcriptomic measurements in order to pinpoint the regulatory programs that collectively explain the observed modulations in the data. Results: Our initial analyses revealed the existence of a broad gene expression reprogramming as prostate cancer progresses. We successfully recapitulated modulations in previously known gene-sets that associated with prostate cancer. However, iGET has enabled us to go a step further and identify multiple regulatory axes that guide the cells from normal to primary tumor to metastasis. These axes are composed of: (i) HOXB13, (ii) MEF complex, and (iii) Sp1 family of transcription factors. Conclusion: In prostate cancer, hundreds of genes are up and down-regulated. However, the absolute majority of these modulations are phenotypically neutral and do not play a role in the disease. Making sense of these complex gene expression patterns was made possible by identifying the key regulatory programs that collectively explain the transcriptomic modulations observed in clinical samples. This knowledge, in turn, enables us to reorganize our observations into coherent regulatory modules and identify the causal pathways, their upstream regulators, and their downstream effectors.

Poster Session Expansile Cribriform Gleason Pattern 4 has Worse Outcomes at Prostatectomy than Glomerulation Gleason Pattern 4 Nancy Y. Greenland, Li Zhang, Peter Carroll, Bradley A. Stohr, Jeffry P. Simko. Background: In prostate cancer, the expansile cribriform subtype of Gleason pattern 4 has been associated with similar outcomes to Gleason pattern 5 carcinomas. By contrast, the glomerulation subtype has not been consistently associated with worse outcomes. We hypothesized that prostatectomy cases with expansile cribriform pattern 4, but not glomerulation pattern 4, would be associated with increased percentages of pattern 4, rates of seminal vesicle invasion (SVI), and presence and severity of extraprostatic extension (EPE). Methods: We evaluated prostatectomy cases within the past ten years with expansile cribriform pattern 4, glomerulation pattern 4, or both patterns, using these groupings as predictors. Outcomes were the percentage of pattern 4, the presence or absence of SVI, and presence and extent (extensive or limited) of EPE. Associations were determined by chi-squared test and ordered logistic regression models. Results: Cases with expansile cribriform had the worst outcomes at prostatectomy, with the highest rates of >50% pattern 4, SVI, and extensive EPE (Table 1). Cases with glomerulation pattern had the best outcomes at prostatectomy, with the lowest rates of >50% pattern 4, SVI, and extensive EPE. Comparing patients with expansile cribriform and glomerulation pattern, expansile cribriform pattern was associated with a 4.8-fold (95% CI 2.5-9.3) odds of having >50% pattern 4 versus <20% pattern 4, a 3.5-fold (95% CI 1.5-8.4) odds of SVI, and a 3.0-fold (95% CI 1.7-5.4) odds of EPE. The presence of both patterns was associated with an intermediate phenotype.

Glomerulation but no

Expansile Cribriform

Expansile Cribriform but

no Glomerulation

Both Patterns Present

P value1 P value2

Variable # cases 82 137 51

% Gleason pattern 4

>50 %4 28 (34.1%) 89 (65%) 30 (56.9%) <0.001 <0.001

>20-<50 %4

16 (19.5%) 23 (16.8%) 16 (33.3%)

<20 %4 38 (46.3%) 25 (18.2%) 5 (9.8%)

SVI Status SVI 7 (8.5%) 34 (24.8%) 7 (13.7%) 0.007 0.005

No SVI 75 (91.5%) 103 (75.2%) 44 (86.3%)

EPE Status EPE 41 (50%) 103 (75.2%) 31 (56.9%) <0.001 <0.001

No EPE 41 (50%) 34 (24.8%) 20 (39.2%)

Extensive EPE

20 (24.4%) 83 (61.3%) 24 (45.1%) <0.001 <0.001

Limited EPE

21 (25.6%) 20 (14.6%) 7 (13.7%)

1. P value1: P value of the test across all three groups. 2. P value2: P value of the test between glomerulation pattern patients and expansile cribriform pattern

patients.

Conclusions: The expansile cribriform pattern 4 at prostatectomy is associated with more histopathologic features of aggressiveness, while glomerulation pattern 4 is associated with less aggressive features.

Poster Session Incidental Detection of Malignancies in Metagenomics Cell-Free DNA Testing

Wei Gu1, Eric Talevich2,3, Allan Gopez1, Shaun Arevalo1, Scot Federman1, Marc Gottschall1,

Zhongxia Qi1, Jingwei Yu1, Linlin Wang1, Bill Karlon1, Ted Kurtz1, Iwei Yeh2,4, Joseph L.

DeRisi5,6, Steve Miller1, Charles Y. Chiu1,7, Jeff Simko2 1 Department of Laboratory Medicine, University of California San Francisco 2 Department of Pathology, University of California San Francisco 3 DNANexus, Mountain View, California 4 Department of Dermatology, University of California San Francisco 5 Chan-Zuckerberg Biohub, San Francisco, California 6 Department of Biochemistry and Biophysics, University of California San Francisco 7 Department of Medicine, University of California San Francisco Introduction: Metagenomic next-generation sequencing (mNGS) is an unbiased test that can detect virtually all known DNA pathogens through whole genome sequencing of cell-free DNA (cfDNA) in plasma and body fluids. At University of California San Francisco (UCSF) Clinical Laboratories, we launched a clinically validated mNGS test for undiagnosed meningitis and encephalitis using cerebrospinal fluid and are validating a plasma mNGS test for sepsis using cfDNA. Using residual human host data that accounts for >99% of the sequencing data, we found genome-wide copy number variation signatures that are specific for tumor derived cfDNA. Methods: Clinical samples consisting of patient body fluids and plasma were identified due to clinical suspicion or laboratory confirmation of infection or malignancy. In both plasma and body fluids, we reduced the majority of leukocyte human by centrifugation and discarding the vast majority of white blood cells. We then created sequencing libraries with size selection and sequenced on Illumina sequencers. Results: In the initial set of plasma samples from patients newly admitted for suspected sepsis, 4 of 9 plasma from metastatic cancer patients showed at least 5 large (>20 Mb) copy number gains or losses. The positive cases were validated based either on tissue CNV detection or cytogenetics for hematological malignancies. All body fluid samples (e.g. pleural fluid, peritoneal fluid, spinal fluid) that were confirmed positive for malignancy on cytology also carried at least 5 large (>20 Mb) copy number gains or losses. In addition, we found 5 examples of positive tumor signatures in body fluid specimens that had resulted ‘benign’ on cytology. Conclusions: Incidental data from cfDNA metagenomic testing can pick up large CNVs in cfDNA in cases where an infectious and malignant etiology overlaps as possibilities on a clinical differential. The high specificity of requiring multiple large CNVs at a high tumor fraction (>5%) will mitigate the risk of false positives. This test may also be useful for monitoring, working especially well for metastatic disease with high tumor burdens. Larger cohort studies are needed to validate this finding across a broader range of sample types and tumors. Up-regulation of miR-130b is involved in prostate cancer racial disparity thorough FHIT pathway inactivation

Poster Session Up-Regulation of miR-130b is Involved in Prostate Cancer Racial Disparity Thorough

FHIT Pathway Inactivation

Yutaka Hashimoto, Marisa Shiina, Taku Kato1, Soichiro Yamamura, Yuichiro Tanaka, Shahana Majid, Sharanjot Saini, Varahram Shahryari, Priyanka Kulkarni, Pritha Dasgupta, Divya Bhagiratha, Guoren Deng, Laura Tabatabai, Peter R. Carroll, Deepak Kumar and Rajvir Dahiya

Purpose: Prostate cancer (PCa) is a common cancer in men. African-American (AfA) men have higher incidence and significantly higher prostate cancer mortality rates than Caucasian-American (CaA) men. In this study, we investigated the biochemical role of miR-130b in this disparity. Methods: We used meta-analyses and data from Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA). PCa cell lines from CaA (DU145, LNCaP, PC3), AfA (MDA-PCa-2b, E006AA-hT) and a normal epithelial cell line (PWR1E) were used for examining miR-130b expression levels and/or establishing stable miR-130b knock-down clones. Total RNA was extracted from clinical samples and cell lines. miR-130b and mRNA levels of its target genes were determined by quantitative real-time PCR (qPCR). qPCR based gene expression arrays were used to identify miR-130b target genes. Cell proliferation, apoptosis and cell cycle were monitored after stable miR-130b knock down. Western blotting was performed to detect miR-130b target protein expression levels. Results: TCGA data showed up-regulation of miR-130b in AfA PCa tissue samples compared with CaA samples. miR-130b was significantly up-regulated in AfA prostate cancer tissues and cell lines compared to CaA cells and tissues. Utilizing VAMCSF and NDRI patient cohorts, we confirmed that miR-130b expression was linked to a racial difference between AfA/CaA PCa patients. Knock down of miR-130b showed decreased growth and cell cycle arrest, though the effect was less in CaA compared to AfA cells. We found unique changes in biological pathways associated with miR-130b knock down in AfA cells by qPCR array. Evaluation of the altered pathways showed that Fragile Histone Triad (FHIT) axis was markedly changed in the AfA compared with CaA cells. Conclusion: These results demonstrate that miR-130b may be a central regulator of key events that contribute racial differences in prostate cancer. FHIT may also be a new AfA specific tumor-suppressive pathway that may be of therapeutic importance in AfA prostate cancer patients.

Poster Session Comparison of TRUS-targeted vs. MRI-targeted vs. Systematic Prostate Biopsy in Detecting Prostate Cancer Annika Herlemann, MD, Matthew R. Cooperberg, MD, MPH, Maya R. Overland, MD, PhD, Samuel L. Washington III, MD, Peter R. Carroll, MD, MPH, Hao G. Nguyen MD, PhD, Katsuto Shinohara, MD University of California, San Francisco, Helen Diller Family Comprehensive Cancer Center, Dept. of Urology, San Francisco, United States Background: We compared prostate cancer detection rates of transrectal ultrasound (TRUS)-targeted (TBx) vs. magnetic resonance tomography (MRI)-targeted (MBx) vs. systematic (SBx) prostate biopsy. Methods: A prospective, single-center study was conducted on all consecutive patients with prostate cancer suspicion or with low-risk prostate cancer on active surveillance. All patients underwent pre-biopsy multiparametric MRI of the prostate. TBx, MBx and SBx were performed during the same session, respectively. Three experienced surgeons performed the biopsies and were blinded to the MRI results before both SBx and TBx, and MBx was performed at the end. Results: 112 patients were included in our study. Median age and PSA were 67 years (IQR 63-72) and 7.3 ng/mL (IQR 5.03-10.53), respectively. Suspicious lesions were visible on both TRUS (hypoechoic lesion (HEL)) and MRI (≥PIRADS 3) in 77% of patients. More lesions per patient were found on MRI compared to TRUS. PIRADS 3, 4, and 5 MRI lesions were seen as HEL on TRUS in 74%, 89%, and 77% of patients, respectively. There were no significant differences in overall prostate cancer detection when comparing all three techniques. Prostate cancer was detected in 80% of PIRADS 3 lesions, 97% of PIRADS 4 lesions, and 85% of PIRADS 5 lesions. The highest overall Gleason score was detected in 5% by TBx only, and in 8% by MBx only. In 15% of patients no prostate cancer was found by either biopsy method. Conclusions: In the setting of high expertise and experience with TRUS and TBx, MRI and MBx do identify some additional high-grade cases, but this is not common, and the incremental value of MBx over SBx and TBx is relatively modest. SBx should not be omitted routinely from biopsy protocols, and urologists should not abandon TRUS as an important diagnostic imaging modality.

Podium Session 2

Results of the First Phase 3 Trial of 68Ga-PSMA-11 in Biochemical Recurrence

Thomas A. Hope,1 Wolfgang P. Fendler,2 Jeremie Calais,3 Johannes Czernin,3 Rahul Aggarwal,1 Kirsten Greene,1 Matthew Cooperberg,1 Raven Smith,1 Ashley Mishoe,1 Felix Y. Feng,1 Eric J. Small1, Peter R. Carroll1

1) University of California San Francisco, San Francisco, CA 2) Essen University, Essen, Germany 3) University of California, Los Angeles, Los Angeles, CA 4) Technical University of Munich, Munich, Germany

Background: 68Ga-PSMA-11 has been used extensively outside of the United States for the imaging and staging of prostate cancer patients. This prospective Phase III study performed between UCSF and UCLA was conducted in order to obtain approval of 68Ga-PSMA-11 in the United States. Methods: 635 patients with biochemical recurrence (defined as two PSAs greater than 0.2 in the post-radical prostatectomy population, and as a PSA rise of greater than 2.0 over the post-radiation nadir) were included in this study. The imaging studies were distributed among 9 blinded readers so that each imaging study was interpreted by three blinded readers, and the studies were recorded for the presence of prostate cancer on a patient level and region level. Patients were then followed after imaging for up to twelve months, and imaging findings were determined as either true or false positive based on both histopathology as well as a composite imaging follow-up to estimate the positive predictive value of 68Ga-PSMA-11 PET. Results: The average PSA in the imaged population was 2.1, and 136 patients had a PSA less than 0.5, while 173 had a PSA greater than 173. Overall, 75% of patients had a positive PSMA PET with the following distribution: PSA < 0.5 = 38%; 0.5-1.0 = 57%; 1.0-2.0 = 84%; 1.0-5.0 = 86% and >5.0 = 97%. Of the patients imaged, 273 had a lesion with follow-up (43% of patients overall), 37% of which was biopsies of lesions. The positive predictive value in the patients with histopathology correlation was 0.84 on a per patient and per region level, and in patients with the composite endpoint the PPV was 0.92 on a per patient and per region level. Overall there was substantial inter-reader agreement with a Cohen’s Kappa ranging from 0.68-0.78 depending on the region imaged. No patients reported a related adverse event, and no patients reported a Grade 2 CTCAE adverse event that was unrelated. Conclusions: 68Ga-PSMA-11 PET has a high positive predictive value in determining sites of biochemical recurrence in patients with prostate cancer. It is hoped that this data will help support an New Drug Application to the FDA in the coming year. Conflicts of Interest: None Funding Acknowledgements: Partially funding was provided through a Young Investigator Award from the Prostate Cancer Foundation.

Poster Session Standard Gamble Utility Measurement from Prostate Cancer Long-Term Survivors: Results from the CaPSURE Registry Chang Wook Jeong, Janet E. Cowan, Jeanette M. Broering, Leslie S. Wilson, Peter R. Carroll, Matthew R. Cooperberg Background: Health utility represents a quantification of quality of life, with perfect health assigned a value of 1.0 and death equal to 0. It is used to calculate quality-adjusted life years (QALYs). Even it is essential information for most quality of life research, well defined utility for prostate cancer health status are sparse. Thus, we assessed utility values from CaPSURE participants using standard gamble method, which is regarded as the most accurate one. Methods: A total of 3,155 active CaPSURE participants in January of 2012 were eligible to receive the questionnaire as cross-sectional data collection. They were randomly assigned to receive either a version A (utilities ordering high to low) or a version B (opposite ordering) with block design stratified by duration since diagnosis. We separately measured utility values of prostate cancer health, sexual function, urinary function, bowel function and overall health based on current patient status. Probability weighting and mixed model bias correction methods were also applied to individual utility results. We assessed utility values according to prostate cancer status and functional conditions at the time of the survey. Results: A total of 1,884 patients responded the mail survey. After exclusion of incomplete or disqualified data, 1,740 patients were included in final analysis. Version A had slightly higher prostate cancer health utility than version A (0.9213 vs. 0.9460, p<0.001). However, other utilities are not different between groups. Prostate cancer health and overall health utilities were 0.9340 ± 0.1197 and 0.9604 ± 0.1001, respectively in total patients. After bias correction by probability weighting function, they were 0.8656 ± 0.1543 and 0.8965 ± 0.1417. Mixed model correction yielded them as 0.8447 ± 0.1860 and 0.8837 ± 0.1764, respectively. Standard gamble utilities were higher than previously reported values in each disease status and functional condition, even by bias corrections (e.g. prostate cancer health utility for metastatic disease: uncorrected – 0.8940, probability weighting corrected – 0.8259 and mixed model corrected 0.7971). Each dysfunctional status was significantly associated with lower utility value than the others. However, disutilities were smaller than those previously known. Conclusions: Standard gamble utilities which were measured from actual prostate cancer long-term survivors were higher than those of previous studies usually measured using imaginary scenarios or indirect methods. Prostate cancer long-term survivors may well adapt their health conditions even though they experience functional problem.

Poster Session Validation of the Utility Weighting Function for the Patient-Oriented Prostate Utility Scale (PORPUS-U) Using the CaPSURE Quality of Life Supplementary Study Data Chang Wook Jeong, Janet E. Cowan, Jeanette M. Broering, Leslie S. Wilson, Peter R. Carroll, Matthew R. Cooperberg Background: Health utility is essential to calculate quality-adjusted life years (QALYs), with perfect health assigned a value of 1.0 and death equal to 0. The Patient Oriented Prostate Utility Scale (PORPUS) measures quality of life for men with prostate cancer using 10 simple questions and can indirectly generate utility value. However, it was developed and validated in small number of patients. As external validation, we compared utility values converted from PORPUS (PORPUS-U) with those directly measured by standard gamble (SG) method as part of CaPSURE quality of life supplementary study (CaPSURE-QoLSS). Methods: A total of 1,884 patients responded the mail survey of CaPSURE-QoLSS. Of them, 1478 patients completed PORPUS questions and had standard gamble utility values for overall health. PORPUS-U was calculated by proposed utility weighting function for it. PORPUS-U was compared with uncorrected and bias-corrected SG using probability weighting and mixed model. We evaluated correlation coefficient, the root mean square error (RMSE) and the mean absolute error (MAE). Results: Mean age at survey and duration from diagnosis were 72.7 ± 8.2 and 8.8 ± 4.0 years. PORPUS-U was relatively well correlated with uncorrected SG utility with slightly lower value. However, PORPUS-U was significantly higher than bias-corrected utilities (Table 1). These trends were observed in all disease and functional status at survey. PORPUS-Us measured from prostate cancer long-term survivors were generally higher than utilities from previous studies. Conclusions: Utility value converted from PORPUS may be comparable with uncorrected SG utility. However, it may be significantly higher than bias-corrected SG utilities. Thus, further calibration of PORPUS-U could be considered. In addition, Prostate cancer long-term survivors may well adapt and satisfy their health conditions. Table 1. Comparison between PORPUS-U and SG utilities in total patients

Median

Mean ± SD 95%CI r RMSE MAE

PORPUS-U 0.9567 0.9507 ± 0.0520 0.9480 - 0.9533

- - -

SG utility 0.9988 0.9614 ± 0.0982 0.9564 - 0.9664

0.1297 0.1055 0.0626

Corrected SG 1

0.9685 0.8975 ± 0.1400 0.8903 - 0.9046

0.1114 0.1533 0.0944

Corrected SG 2

0.9759 0.8848 ± 0.1742 0.8759 - 0.8937

0.1115 0.1880 0.1103

Poster Session Disentangling the Relationship Between Genetic Predictors of PSA, Prostate Cancer Susceptibility, and Benign Prostatic Hyperplasia Linda Kachuri, Thomas J. Hoffmann, Sara R. Rashkin, Rebecca E. Graff, John S. Witte

Background: The use of prostate-specific antigen (PSA) levels for prostate cancer (PrCa) screening remains controversial due to potential for overdiagnosis and overtreatment of detected disease. A potential avenue for improving PSA performance is to elucidate the relationship between genetic determinants of PSA, benign prostatic hyperplasia (BPH), a condition that also results in elevated PSA levels, and PrCa risk.

Methods: We developed genetic instruments for PSA using summary statistics from a GWAS of 26,307 non-Hispanic whites in Kaiser Permanente. To reduce the possibility of reverse causation, SNP effects were estimated in men without PrCa (at the time of data extraction), only if a PSA measurement 2 years prior to data extraction was available. Associations between PSA instruments and PrCa risk were estimated in the UK Biobank cohort (7441 cases and 169,970 controls), with adjustment for age, genetic ancestry principal components, PSA screening (ever/never), and being diagnosed with BPH.

Results: The main MR analysis used 19 genetic instruments (LD r2<0.05), accounting for 7.14% of variation in log(PSA). Genetically predicted 1-unit increase in log(PSA) was associated with

a significant increase in PrCa risk (OR=2.83, 95% CI: 1.63-4.91, p=2.110-4). These results were not vulnerable to weak instrument bias (mean F = 120; I2=0.99) or directional pleiotropy (MR

Egger 0 p=0.60), however significant heterogeneity between the instruments was observed

(Cochran’s Q p-value: 1.110-74). Most of the variants contributing to this heterogeneity were known PrCa susceptibility loci in 8q24 (MYC), 10q11 (MSMB), 10q26.13 (FGFR2), and 17q12 (HNF1B), however their exclusion did not result in an appreciable change in the MR estimate

(OR for PrCa: 2.86, 95% CI: 2.45-3.33, p=1.110-40). Mediation analyses of these risk loci suggest that their effects on PrCa are mostly operating through PSA-independent mechanisms.

In an attempt to disentangle genetic predictors of PSA and PrCa from loci affecting BPH, we carried out a GWAS of BPH in UK Biobank (3134 cases and 179,754 controls never diagnosed with PrCa or prostatitis). We identified 5 independent loci affecting BPH risk: rs10886890

(P=1.110-12) in 10q26.13 (FGFR2), rs3816659 (P=6.010-11) in 5p15 (CLPTM1L), rs1550348

(P=6.510-10) in 10q26.12 (WDR11), rs9348716 (P=1.710-8) in 6p22.2 (BTN3A2), and

rs919964 (P=3.310-8) in 1p22. Notably, the 19q13.33 (KLK3) region was not associated with BPH, however several BPH loci have been previously linked to PSA and PrCa. Multivariable MR was used to model the effects of these two phenotypes on PrCa simultaneously. After accounting for genetically predicted increase in PSA, there was no association between BPH and PrCa risk

(OR=1.11, 0.88-1.41, p=0.38), while the effect of PSA remained significant (p=1.710-8).

Conclusions: Genetic mechanisms resulting in elevated PSA levels are implicated in PrCa susceptibility, even after the exclusion of variants that are significantly associated with both phenotypes, which may be due to screening bias. In addition, the association between genetically predicted higher PSA levels and PrCa risk persisted after accounting for the influence of BPH susceptibility loci.

Poster Session The C-Terminal Acidic Domain of TSPX, an X-Linked Tumor Suppressor, Plays Crucial Roles in The Functional Networks of Prostate Cancer Tatsuo Kido, Yunmin Li, and Yun-Fai Chris Lau Division of Cell and Developmental Genetics, Department of Medicine, VA Medical Center, University of California, San Francisco Background: TSPX is the tumor suppressor gene located within Xp11.2, a prostate cancer susceptible locus. It is frequently downregulated in multiple cancer types, including lung cancer, glioma, liver cancer, and prostate cancer. Previous studies demonstrated that the C-terminal acidic domain (CAD) of TSPX is crucial for the tumor suppressor functions, such as inhibition of cyclin-B/CDK1 activity and suppression of androgen receptor mediated transactivation. Although TSPX could suppress proliferation of cancer cells, the roles of its CAD in prostate cancer have not been well explored yet, especially at the gene expression level. In the present study, we aimed to elucidate the roles of CAD in the functional networks of prostate cancer. Methods: We generated LNCaP cells overexpressing either TSPX or its CAD-truncated variant under the control of doxycycline (Tet-ON system), and investigated the effect of their overexpression on the gene expression patterns by RNA-seq using the Illumina NextSeq 500. Further, the TSPX-associated genes in the clinical prostate cancer specimens were explored by data-mining of the prostate cancer datasets downloaded from the Cancer Genome Atlas (TCGA). Results: Over than 8,000 genes were affected by TSPX overexpression in LNCaP cells; these genes were involved in various functional networks, such as ‘cell viability of tumor cell lines’, ‘cell survival’, and ‘cell viability’, consistent with the result of cell proliferation assay. Although approximately half of these genes were affected by the CAD-truncated TSPX, only a few interactions were formed in the functional networks, suggesting that CAD could play important roles to effectively suppress tumor growth in prostate cancer. In addition, the data-mining of the TCGA prostate cancer datasets revealed that the most in vivo TSPX-associated genes involved in the functional networks were regulated in a CAD-dependent manner; e.g. 80% of the AR-related network and 79% of the MYC-related network. Conclusions: Our results suggest that the CAD of TSPX plays important roles to efficiently regulate the functional networks for preventing prostate cancer development. Future studies addressing the protein-protein interactions via CAD will provide new insights into the therapeutic targets for prostate cancer.

Poster Session A Systematic Review of Observational Studies Assessing Outcomes of Prostate Cancer Patients on Active Surveillance: An Examination of Adherence to Protocols Authors: Glenda Kith and Sarah Lisker; Urmimala Sarkar; Jill Barr-Walker; Benjamin N. Breyer; Nynikka R. Palmer Background: Active surveillance is an approach to reduce overtreatment of men with low-risk prostate cancer, and some men with intermediate-risk. Active surveillance involves actively monitoring prostate cancer via protocols, avoiding unnecessary morbidity, with the expectation to move to active treatment (surgery or radiation) for curative intent in the event of disease progression. However, the potential for patient safety gaps is concerning if active surveillance protocols are not appropriately followed and if patients are lost to follow-up. We sought to systematically review the extent to which studies evaluating patient outcomes measure and report adherence to active surveillance, and to characterize definitions of adherence and report lost to follow-up.

Methods: We searched PubMed, Embase, and CENTRAL databases for studies published before March 8, 2018. Study inclusion criteria consisted of: (1) observational studies that focused on active surveillance for prostate cancer, (2) authors stated a predetermined active surveillance eligibility and monitoring protocol, and (3) reported outcome data (e.g., disease progression). Two researchers (GK and SL) independently screened and reviewed articles, and consistently met to compare and reach agreement on study inclusion. We collected data on study design, demographics, adherence definition and measures, and outcomes. Adherence definitions were considered precise if they included a pre-specified time frame for protocol activities (e.g. repeat PSA every 6-months). We used the STROBE checklist to assess study quality.

Results: Our search yielded 41 publications that report on real-world active surveillance practice using generally accepted eligibility criteria and monitoring protocols. All studies included a measure of patient outcomes, such as proportion of patients that undergo active treatment, experience disease progression, die, or are lost to follow-up. Only eight studies clearly reported adherence to active surveillance protocols. Five of those studies used precise adherence definitions, while three used adherence definitions without timeframes. Thirty-three publications did not explicitly report adherence; however, twenty three of these studies included measures that could be considered representative of adherence but were not explicitly described as such (e.g., median number of follow-up biopsies). Twenty-three studies reported lost to follow-up rates ranging from 0%-19.5%, and 1 assessed associated factors.

Conclusions: The majority of studies evaluating patient outcomes of men on active surveillance do not clearly measure or report adherence to care; and precise definitions of adherence that consider timeliness were rarely reported. As the number of men managed on active surveillance continues to increase, our findings highlight the need for future research and efforts to improve quality of care, including: (1) establish consensus on a standardize active surveillance protocol, that includes adherence measures; (2) include adherence as a quality improvement measure for active surveillance outcomes; and (3) track and identify multilevel factors associated with adherence and lost to follow-up.

Poster Session

Hyperpolarized In Vivo pH Imaging Reveals Grade-Dependent Acidification in Prostate Cancer

David Korenchan, Robert Bok, Renuka Sriram, Kristina Liu, Romelyn Delos Santos, Hecong Qin, Daniel Vigneron, David Wilson, John Kurhanewicz, and Robert Flavell

Background: Clinical imaging biomarkers distinguishing indolent from aggressive prostate cancer (PCa) are lacking. Changes in metabolism of pyruvate to lactate, perfusion, and lactate export are expected to lead to an extracellular pH (pHe) decrease, which correlates with an aggressive and treatment-resistant phenotype (part a of figure). The goal of this study was to compare between indolent and aggressive disease in a murine model of PCa using a new method of hyperpolarized (HP) 13C pH imaging recently developed in our laboratory.

Methods: Transgenic adenocarcinoma of the mouse prostate (TRAMP) mice were anesthetized and subjected to 1H T2-weighted MRI and an injection of HP [13C]bicarbonate with 13C MR imaging to measure interstitial pH. Within 48 hours of imaging, the mouse was euthanized and the tumor tissue fixed, sectioned, and stained for histology. Stained tumor regions were classified by a trained pathologist as low- or high-grade based upon cell differentiation, glandular pattern, and necrosis. Volume elements (voxels) in images were classified as low-grade or high-grade based upon histology. Two-tailed t-tests with unequal variances were performed between low- and high-grade tumor regions for pHe.

Results: Representative HP 13C overlays for the TRAMP mice studied, along with tumor voxels, are displayed in the figure (part b). The minimum voxel interstitial pH values in low- and high-grade tumors were 7.38 ± 0.13 and 7.04 ± 0.12, respectively (n = 4 low-grade, n = 6 high-grade, p < 0.01, part c of figure).

Conclusions: We demonstrate rapid in vivo imaging of extracellular pH in a murine model of prostate cancer. We observed decreased pHe in high-grade tumor regions, implicating interstitial acidification as a biomarker of PCa indolent-to-aggressive transition. These data suggest potential feasibility of the HP pH imaging method to detect high-grade, clinically significant PCa in men as part of a multiparametric MRI examination.

Poster Session The Rate of Hyperpolarized [1-13C] Pyruvate to [1-13C] Lactate Conversion Distinguishes High-Grade from Low-Grade Prostate Cancer in Patients Natalie Korn1,2, Peder EZ Larson1,2, Hsin-Yu Chen1,2, Jeremy Gordon1, Robert A Bok1, Mark VanCriekinge1, James Slater1, Rahul Aggarwal3, Matthew Cooperberg3, Lynn Delos Santos1, Justin Delos Santos1, Jeffry Simko4, Susan M Noworolski1,2, Daniel Vigneron1,2, John Kurhanewicz1,2 1Department of Radiology and Biomedical Imaging, UCSF 2The Graduate Group in Bioengineering, UCSF 3Department of Urology, UCSF 4Department of Pathology, UCSF Introduction: An accurate method for distinguishing aggressive from indolent prostate cancer at the time of diagnosis is a pressing clinical need. High-spatial and temporal resolution 3D dynamic hyperpolarized (HP) 13C MRSI can provide images of the rate of HP [1-13C]pyruvate to [1-13C]lactate conversion, kPL, that correlate with cancer grade in a murine model of prostate cancer[1]. We present the initial patient data from a phase II clinical trial using a 3D dynamic hyperpolarized 13C MRSI approach focused on investigating the association between kPL and pathologic grade of prostate cancer, using whole-mount step section histopathology after surgery as a reference standard. Methods: 11 prostate cancer patients received a 3T multiparametric 1H/13C pyruvate MRI of the prostate. Dynamic HP 13C spectroscopic imaging data (13C-MRSI) were acquired, and an image map of the apparent rate of [1-13C]pyruvate to [1-13C]lactate conversion (kPL map) [2] was modeled with a total signal-to-noise (tSNR) cutoff of 14. Patients underwent radical prostatectomy with 3mm-slice whole-mount sectioning. Regions of Interest (ROIs) of benign, high-grade primary pattern 4 disease (Gleason score≥4+3) and low-grade primary pattern 3 disease (Gleason≤3+4) were propagated to the kPL maps. Wilcoxon Rank Sum tests with a Holm-Sidak correction were used to determine differences in maximum kPL (kPLmax) between tissue types. Results and Discussion: Seven regions of low-grade prostate cancer, fourteen regions of high-grade prostate cancer, and fourteen regions of benign prostate tissue were included in this analysis. kPLmax was nonsignificantly higher in regions of low-grade cancer (0.0088±0.003sec-1, p=0.218) relative to benign prostate tissue, however high-grade prostate cancer demonstrated significantly higher kPLmax (0.0128±0.003 sec-1), relative to benign (p=0.0003) and low-grade cancer (p=0.034), respectively. Conclusion: This study demonstrated for the first time that kPLmax is significantly elevated in high-grade prostate cancer versus both normal and low-grade disease in the peripheral zone of the human prostate, consistent with the recent pre-clinical study[1]. Future studies will be needed to examine the central gland of the prostate, and to assess the ability of kPLmax to improve diagnostics as part of a multiparametric MRI. [1]Chen, Cancer Res, 2017 [2]Maidens, IEEE Trans Med Im 2016

Rapid-Fire and Poster Sessions Maternal Embryonic Leucine Zipper Kinase (MELK) is A Novel Therapeutic Target and a Biomarker of Aggressive Prostate Cancer Vishal Kothari, Shuang Zhao, Corey Speers, Chao Zhang, Kari-Wilder Romans, Marsha Calvert, Travis Barnard, and Felix Y. Feng. Background: The identification of novel disease drivers and therapeutic targets in aggressive prostate cancer is of paramount importance. In this study, we analyzed prostate cancer samples in order to identify kinases associated with prostate cancer progression and then investigated the top candidate in preclinical models of disease. Methods: We analyzed the expression of all known kinases in 545 prostatectomy samples that were obtained from high-risk patients with long term (>10 years) clinical follow up. We ranked the kinases by their fold change in expression (on a probe-by-probe basis) between patients that progressed to metastasis vs patients that did not. High-throughput proteomic and phospho-proteomic analyses as well as standard molecular assays were used to conduct mechanistic, phenotypic, and in vivo studies. Results: Our analyses revealed Maternal Embryonic Leucine Zipper Kinase (MELK) to be the top differentially expressed kinase associated with metastatic progression of prostate cancer. High expression of MELK is associated with poor clinical outcomes and inhibition of MELK abrogates aggressive cancer phenotypes in prostate cancer cell line models. Immunoprecipitation in combination with mass spectrometry (IP-MS) analysis of prostate cancer cells identified eIF4b as a novel interactor. Mechanistic studies showed that MELK phosphorylates eIF4b to mediate aggressive phenotypes. To better understand the role of MELK in prostate cancer progression, we have generated a knock-in mouse model and are currently evaluating for oncogenic phenotypes. Conclusions: Our results indicate that MELK may serve as a prognostic biomarker and therapeutic target in aggressive PCa. A detailed characterization of the MELK-eIF4b interaction and the mechanisms by which MELK modulates metastasis using in vitro as well as in vivo transgenic models is underway. These initial studies provide rationale for a biomarker-driven clinical trial to assess the efficacy of MELK inhibition in men with advanced prostate cancer.

Poster Session Obesity and Prostate Cancer Recurrence Following Radical Prostatectomy Langlais CSa, Cowan JEb, Broering JMb, Kenfield SAb, Van Blarigan EL,a,b Cooperberg MR, Carroll PRb, Chan JMa,b aDepartment of Epidemiology and Biostatistics, University of California, San Francisco bDepartment of Urology, University of California, San Francisco Purpose: To examine the association between obesity and risk of prostate cancer recurrence in the CaPSURE™ (Cancer of the Prostate Strategic Urologic Research Endeavor) study. Methods: We included 3,491 men with prostate cancer from CaPSURE who underwent radical prostatectomy (RP) between 1998-2017 and had body mass index (BMI) available at baseline. BMI was examined continuously and categorically (<25, 25-29.9, 30-34.9, ≥35 kg/m2). Cancer recurrence was defined as two consecutive prostate-specific antigen (PSA) levels ≥0.2ng/mL after RP or any second treatment. We used Cox proportional hazard models to estimate hazard ratios (HR) and 95% CI for BMI, with adjustment for ethnicity, age, clinical site of recruitment, and diagnostic PSA level, Gleason grade, and T stage. Results: Patients were followed for a median of 4.7 years [IQR: 2.3, 8.1] after RP. Recurrence occurred in 452 (13%) patients a median of 21 months [IQR: 10.4, 41.4] after RP (66% due to PSA rise). There was a positive association between continuous BMI and risk of recurrence (multivariate HR 1.16, CI: 1.04, 1.29, per 5-unit increase in BMI). Men who were very obese had a higher hazard of recurrence (multivariate HR 1.93, CI: 1.27, 2.92) compared to those with normal BMI. Other BMI categories were associated positively, but not statistically significantly, with recurrence (overweight HR: 1.13, CI: 0.87, 1.46; obese HR: 1.26, CI: 0.92, 1.71). Conclusions: Extreme obesity was associated with a higher risk of prostate cancer recurrence, after adjustment for baseline demographic and clinical factors. These results are consistent with prior reports on BMI and post-diagnostic clinical outcomes. Our findings support careful follow-up among obese patients following RP. Additional research is needed to determine if this association is independent of lifestyle behaviors (diet, physical activity) or if losing weight after prostate cancer diagnosis is associated with improved outcomes after RP.

Poster Session The Long Noncoding RNA PLUTO-201 is Associated with Prostate Cancer Progression and Metastasis Hui Li1, Bhavna Malik4, Shuang G. Zhao4, Sethuramasundaram Pitchiaya7, Travis Barnard1, Vishal Kothari1, Rajdeep Das1, Arul M. Chinnaiyan5,6,7,8,9, Felix Y. Feng1,2,3* Departments of Radiation Oncology1, Urology2 and Medicine3, University of California at San Francisco, CA. Departments of Radiation Oncology4, Pathology5 and Urology6, Michigan Center for Translational Pathology7, Howard Hughes Medical Institute8, Comprehensive Cancer Center9, University of Michigan, Ann Arbor, MI. * Correspondence to: Felix Feng, University of California, San Francisco, 1450 3rd St., San Francisco, CA, USA. Email: Felix.Feng@ucsf.edu. Background: Long noncoding RNAs (lncRNAs) have recently been shown to serve as drivers of malignancy and represent novel therapeutic targets. We have recently identified over 46,000 novel lncRNAs through in silico analyses of nearly 8,000 tumor or normal tissue specimens. In this study, we utilize this compendium of lncRNAs to identify those associated with metastatic progression of prostate cancer (PCa), and functionally investigate our top nominated candidate, PLUTO-201, in preclinical models of disease. Methods: Transcriptional profiling of a cohort of prostatectomy patients was used to identify new PCa-associated lncRNAs. The gene most strongly associated with subsequent metastatic progression, PLUTO-201, was further analyzed. Rapid Amplification of cDNA Ends (RACE) was used to define its gene structure, and Fluorescence In-Situ Hybridization (FISH) was used to localize PLUTO-201 transcripts. Knockdown and over-expression studies were performed in vitro and in vivo to elucidate the functions of PLUTO-201. Finally, transcriptional analysis was performed to identify gene sets associated with this candidate. Results: Of all annotated protein-coding genes and lncRNAs, PLUTO-201 was the most strongly associated with metastatic progression, via Area Under the Curve (AUC) assessment on a Receiver Operating Characteristic (ROC) analysis. For multivariable analyses, PLUTO-201 remained prognostic while accounting for standard clinicopathologic variables. Additionally, PLUTO-201 expression is relatively specific for PCa. Knockdown of PLUTO-201 significantly impaired proliferation and invasion of PCa cells in vitro, and tumor growth and metastasis in vivo. Conversely, PLUTO-201 over-expression increases the invasion potential of the prostate epithelial cell line 22Rv1. Conclusions: We have identified a novel, prostate-specific lncRNA (PLUTO-201) that promotes PCa proliferation and metastasis, and is associated with poor clinical prognosis. Future investigations regarding its mechanisms of action are warranted to further elucidate its role in PCa progression.

Poster Session Controversial Role of Micro-RNA-588 in Prostate Cancer Nadeem S Bhat, Varahram Shahryari, Sharanjot Saini, Pritha Dasgupta, Priyanka Kulkarni, Yutaka Hashimoto, Marisa Shiina, Soichiro Yamamura, Yuichiro Tanaka, Rajvir Dahiya and Shahana Majid; Department of Urology, VA Medical Center and UCSF, San Francisco, California

Introduction and Objective: MicroRNAs are small noncoding RNAs that regulate the expression of >60% of all human genes, either inhibiting target mRNA translation or inducing its degradation. In October 2017, a report was published in Journal of Cellular Biochemistry on the role of microRNA-588 (miR-588) in prostate cancer (PCa). The paper reports that miR-588 is oncogenic in PCa. We had performed preliminary experiments to investigate the same miRNA-588 in PCa and our results were indicating a contrasting tumor suppressor role in PCa. We continued our study and performed comprehensive investigation of role of miR-588 in PCa. Methods: We employed a comprehensive approach using a panel of cell lines, in house tissue samples, publicly available large data sets such as TCGA, a series of in vitro assays such as quantitative-real time PCR; western blot; fluorescence-activated cell sorting assays for cell cycle distribution and apoptosis; assays for cell viability, migration and invasion assays. Luciferase reporter assays and in-vivo intra-cardiac and intra-tumoral mouse model studies in nude mice was also performed. Results: The expression of miR-588 was significantly suppressed or silenced in prostate cancer tissue samples and cell lines when compared with normal tissues and a non-malignant cell line. Comparable results were observed by analyzing the publicly available TCGA data sets for prostate adenocarcinoma. Functionally ectopic expression of miR-588 induced G2/M cell cycle arrest and apoptosis and suppressed cell proliferation. miR-588 exerted these functional effects by directly targeting the oncogenic CCNA2 that is involved in cancer cell cycle and proliferation. In silico algorithm showed a complimentary binding sequence in the 3’UTR of CCNA2 for miR-588. Over-expression of miR-588 significantly suppressed the luciferase activity of reporter plasmid containing the wild type 3’UTR sequences of CCNA2 complementary to miR-588, which was abolished by mutations in these 3’UTR regions. miR-588 overexpression also significantly reduced the expression of CCNA2 at both mRNA and protein levels. A significant decrease in the expression of various cell cycle pathway genes such as CCNE1, MCM2, MCM4, CDC7 and CDT1 was also observed. These genes are involved in promoting cell cycle and proliferation and are overexpressed in prostate cancer. In vivo intracardiac implantation of bio-luminescent PC3-MLuc-C6 prostate cancer cells constitutively expressing miR-588 showed a significant inhibition of the metastatic dissemination of these cells compared to the control miR expressing cells. Similarly, intra-tumoral delivery of miR-588 locally in the tumors reduced the tumor burden in nude mice compared to control-miR. To summarize, all the above results confirm that miR-588 has tumor suppressor function in PCa contrary to that what has been published previously. Further, we observed that miR-588 was more downregulated in African American (AfA) PCa cell lines (MDAPCA2b and E00A) compared to Caucasian American (CA) cell lines. In the TCGA data base all the samples that showed some expression of miR-588 belonged to CA whereas AfA tissue samples had no expression. Reconstitution of miR-588 in AfA cell line induced apoptosis and cell cycle arrest and inhibited cell proliferation. These results further indicate that miR-588 not only has overall tumor suppressor effects in PCa but might also be partially responsible for the aggressive disease in African American population. Conclusion: Based on the in-depth and comprehensive investigation of miR-588 in PCa, in contrary to a previous publication, our study demonstrates that miR-588 is significantly silenced and acts as a tumor suppressor in prostate cancer. Moreover, miR-588 might be responsible for aggressive PCa in African American population. Reconstitution of silenced miR-588 may lead to the down-regulation of target oncogenes thereby contributing to novel therapeutic approaches for regulating prostate cancer. This study also highlights the importance of performing multiple and diverse experiments to draw any conclusion about the role of a miRNA in cancer. We hope that our study will encourage other researchers to publish contradicting findings for the overall better understanding of molecular players in human cancers. Source of Funding: This project was funded by NCI-NIH grant number U01CA184966.

Poster Session Inflammatory Bowel Disease and Prostate Cancer Risk: a Mendelian Randomization Study

Travis J. Meyers1, Sara R. Rashkin1, Linda Kachuri1, and John S. Witte1,2 Affiliations: 1. Department of Epidemiology and Biostatistics, UCSF 2. Institute for Human Genetics, UCSF

Introduction: The objective of this study was to investigate the association between inflammatory bowel disease (IBD) with prostate cancer risk using a two-sample Mendelian randomization (MR) design. MR uses genetic variants as surrogate markers for risk factors and can reduce the likelihood of confounding in observational studies. In this study, IBD was hypothesized to be associated with increased risk for prostate cancer due to published links between inflammation with cancer overall and specific to the prostate. Methods: MR analyses were performed in the UK Biobank of 7,441 men with prostate cancer and 169,970 cancer-free controls using single nucleotide polymorphisms (SNPs) associated with overall IBD, ulcerative colitis (UC), and Crohn’s disease (CD). SNP summary statistics were obtained from a large published genome-wide association study (GWAS) of IBD in Europeans. MR associations were measured by the inverse-variance weighted (IVW) meta-analysis method and two sensitivity analysis methods for departure from MR assumptions. In addition, LD score regression was applied to calculate the liability-scale heritability of IBD and prostate cancer and the genetic correlation between both phenotypes. Results: No associations were observed between overall IBD, UC, or CD with prostate cancer in this MR analysis using independent SNP instruments (IVW beta coefficient for IBD = 1.7 x 10-

4, standard error = 2.0 x 10-4). The heritability estimates for prostate cancer and IBD were 0.16 and 0.15, respectively, and the genetic correlation was approximately zero (0.0086, standard error = 0.044). Conclusion: This study does not support a positive association between IBD with prostate cancer risk. Inflammation biomarkers should be further studied for their possible associations with development and progression of prostate cancer.

Poster Session Measuring PD-L1 Expression with ImmunoPET Anna Moroz, Lee, Chia-Yin, Yung-hua Wang, Jeffrey C. Hsiao, Natalia Sevillano, Charles Truillet, Charles S. Craik, Lawrence Fong, Cheng-I Wang, Michael J. Evans Background: The swell of experimental imaging technologies to non-invasively measure immune checkpoint protein expression presents the opportunity for rigorous comparative studies toward identifying a gold standard. 89Zr-atezolizumab is currently in man, and early data show tumor targeting but also abundant uptake in several normal tissues. Therefore, we conducted a reverse translational study, both to understand if tumor to normal tissue ratios for 89Zr-atezolizumab could be improved, and to make direct comparisons to 89Zr-C4, a radiotracer that we showed can detect a large dynamic range of tumor-associated PD-L1 expression. Methods: Atezolizumab was conjugated to the Zr-89 chelator desferrioxamine B (DFO) using a similar approach applied for C4 and DFO-atezo was radiolabeled via incubation with 89Zr-oxalic acid. Its properties were studied in vivo in PD-L1 expressing human and mice cell lines animal models. Mice were imaged (PET CT) at multiple time points followed by biodistrubution study. Results: PET/CT and biodistribution studies in tumor bearing immunocompetent and nu/nu mice revealed that high specific activity 89Zr-atezolizumab (~2 mCi/mg) binds to PD-L1 on tumors, but also results in very high uptake in many normal mouse tissues, as expected. 89Zr-atezolizumab uptake was generally higher in normal mouse tissues compared to 89Zr-C4, and lower in a tumor model with modest PD-L1 expression. Also unexpectedly, reducing the specific activity at least 15 fold suppressed 89Zr-atezo uptake in normal mouse tissues, but increased tumor uptake to levels observed with high specific activity 89Zr-C4. Conclusions: These data reveal that low specific activity 89Zr-atezo may be necessary for accurately measuring PD-L1 in the tumor microenvironment, assuming a threshold can be identified that preferentially suppresses binding in normal tissues without suppressing binding to tumors with abundant expression. Alternatively, high specific activity approaches like 89Zr-C4 PET may be simpler to implement clinically to measure the broad dynamic range of PD-L1 expression known to manifest among tumors.

Poster Session Does PI-RADS v2 Scores Predict Adverse Surgical Pathology at Radical Prostatectomy? Hao G. Nguyen*, Antonio Westphalen*, Niloufar Ameli*, Janet Cowan, Jeff Simko, Katsuto Shinohara and Peter R. Carroll. Background: In recent years, multi-parametric MRI (mpMRI) has gained increased acceptance and utilization as a diagnostic and staging tool for early - stage prostate cancer. Reporting systems, in particular the Prostate Imaging - Reporting and Data System (PI-RADS), now in its second version, has been advanced as means to standardize the grading and reporting of MRI findings. However, it remains to be determined whether PI-RADS scores independently predict the risk of adverse pathology, i.e. high-grade and/or high-stage disease. Objective: To evaluate the association of surgical pathological findings assessed on whole-mount pathology analysis and pre-operative mpMRI suspicion assessed using PI-RADS v2 scores. Methods: We retrospectively analyzed 221 patients who had radical prostatectomy within 12 months of their staging endorectal 3T mpMRI. We examined the association of the PI-RADS v2 scores 1-3, 4 and 5 with adverse surgical pathology, defined as advanced pathologic stage (≥ pT3a) or high-grade disease (primary Gleason pattern ≥ 4) or both, using contingency tables (diagnostic accuracy and chi-square) and logistic regression models. Results: Of 221 patients, 134 (61%) had adverse surgical pathology; 75 men (34%) had high-grade (primary Gleason pattern ≥ 4), 108 (49%) had ≥ pT3 disease, and 38 (17%) had both high-grade and high-stage disease. 89 (40%) had PI-RADS mpMRI score 4 findings, of whom, 53% had adverse pathology. 94 (43%) of cohort had PIRADS 5, of whom 73% had adverse pathology compared to 18 (47%) patients with PI-RADS ≤3. Conversely, 87% (116/134) of patients with adverse pathology had positive MR studies (PI-RADS score 4 or 5). Accordingly, mpMRI PI-RADS 4,5 demonstrated 87% sensitivity (95% CI 81-92), 23% specificity (95% CI 14-32), 64% PPV (95% CI 56-70), 53% NPV (95% CI 37-69), and 62% accuracy (95% CI 55-68) for the detection of adverse surgical pathology. In the multivariable logistic regression analysis, adjusted for PSA density and age, PI-RADS score 5 (odds ratio (OR) 2.4, 95% CI 1.1-5.4, p=0.03) and clinical CAPRA score (OR 1.3, 95% CI 1.1-1.6, p=0.01) were significantly associated with higher risk of adverse pathology. Moreover, PI-RADS score 5 was significantly associated with increased risk of pathologic stage ≥ pT3 (OR 3.0, 95% CI 1.3-6.8, P˂0.01). This study is limited by its retrospective nature. Conclusion: PI-RADS v2 score 5 on mpMRI is highly sensitive for the detection and prediction of adverse pathology. PI-RADS v2 may help improve the detection and staging of prostate cancer and allow for tailored intervention. * equal contributions.

Poster Session Evaluating MRI Fusion Biopsy vs. Systematic Ultrasound Guided Biopsy in Predicting High Grade Cancer at Time of Radical Prostatectomy Hao G. Nguyen*, Katsuto Shinohara*, Janet Cowan, Niloufar Ameli, Antonio Westphalen, Jeff Simko, Matthew Cooperberg and Peter R. Carroll Background: There is much enthusiasm for multi-parametric MRI (mpMRI)-ultrasound fusion biopsy in those with an elevated PSA, a prior negative biopsy or those on active surveillance. However, the predictive value of MRI – targeted biopsy in predicting final cancer grade has not been well addressed. The uncertainties of both over staging and under staging using MRI fusion targeted biopsy have not been well addressed. Objective: We aimed to evaluate the accuracy of cancer risk estimation with MRI fusion biopsy; traditional sextant and anterior (14 cores) ultrasound guided biopsy or the combination, using whole-mount histopathology at time of prostatectomy. Methods: We retrospectively analyzed 510 men with MRI fusion biopsy. 185 patients had radical prostatectomy in 2014-2016. All patients had undergone systematic ultrasound guided biopsy and mpMRI fusion biopsy. We compared Gleason Score (GS) upgrading or downgrading between MRI fusion and systematic ultrasound guided biopsy to that of the final Gleason score evaluated by whole-mount histopathological analysis. Logistic regression was used to evaluate association to adverse pathological outcome for each biopsy approach. Results: Of 185 patients who had RP, significant cancer grade (GS ≥3+4) found on MRI fusion biopsy matched final pathology in 41% of the cases while it was overestimated in 14% of patients and underestimated in 45%. Cancer grade found on traditional systematic biopsy matched final pathology in 51% of patients while it overestimated grade in 18% and underestimated grade in 31% of patients with GS≥7. The combined systematic and MRI fusion matched final pathology 55% of the case while underestimated 18% of patients and overestimated grade in 27% of patients who had GS≥7 on their final pathology. In the logistic regression model, having a GS ≥ 4+3 detected on combination biopsy (MRI +systematic) was associated with higher odds (OR: 14.1 95% CI 5-34, p <0.01) of higher stage cancer (≥pT3a) at RP. The association persisted when the model was adjusted for clinical CAPRA score. The ROC curve (area) of systematic and MRI target is 0.82 while systematic biopsy has 0.78 and MRI target biopsy has 0.76. This study was limited by its retrospective nature. Conclusion: Risk of over - staging using MRI fusion biopsy is low compared to systematic biopsy. However, MRI targeted biopsy alone could significantly underestimate those with clinically significant disease. Using MRI fusion biopsy alone to detect high grade cancer may not be adequate in this contemporary cohort. This data may have important implications for guiding treatment decisions. *Equal contributions

Poster Session Targeting the Translational Oncogenic Program of the Major Cap-Binding Protein eIF4E in Metastatic Castration Resistant Prostate Cancer Hao G. Nguyen, Lingru Xue, Peter Carroll and Davide Ruggero UCSF Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco As a master regulator of translation initiation, the major cap-binding protein eIF4E was thought to be essential for cellular life; thus, it had been widely considered “untouchable.” Surprisingly, we found that a 50% reduction in eIF4E expression in mice is compatible with normal development and global protein synthesis but compromises oncogene-induced cellular transformation. Furthermore, we demonstrated that the ability of these oncogenic pathways to regulate the activity of eIF4E has a direct and causal role in tumorigenesis by regulating the translation of selective mRNAs that are required to sustain cancer cell viability, thereby exposing an “Achilles’ Heel” of this oncogenic pathway. In the current study, our data show that phosphorylation eIF4E is dramatically increased in both primary and metastatic prostate tissue derived from patients. Furthermore, by employing novel Patient-Derived Xenografts (PDX) models of prostate cancer, we demonstrate that inhibition of eIF4E phosphorylation with a novel clinical small molecule, eFT 508, that inactivate the MNKs kinase, results in profound tumor regression and prolongs survival of primary human prostate cancer. These results raise the outstanding question of how specific prostate oncogenic programs are regulated by eIF4E phosphorylation. Strikingly, our preliminary data show that the expression of androgen receptor and its V7 splice variant is dependent on the activity of eIF4E. Therefore, inactivation of MNKs represent a new point of vulnerability of primary and metastatic prostate cancer cells and establishes a promising future clinical intervention for metastatic prostate cancer that remains incurable.

Poster Session A Phase 1/2 Trial of Ibrutinib as Neoadjuvant Therapy in Patients with Localized Prostate Cancer Lowell Nicholson, Won Kim, Matthew Cooperberg, Eric Liu, Whitney Tamaki, Terence Friedlander, Rahul Aggarwal, Eric Small, Peter Carroll, Lawrence Fong Background. Although surgery and radiotherapy are potentially curative treatments for localized prostate cancer, treatment failure occurs in up to 40% of patients. Immunotherapy has tremendous potential as an anti-cancer therapeutic and enhancing the anti-tumor immune response prior to surgery may improve long-term outcomes. While cancer immunotherapy typically focuses on inducing tumor-specific T cell responses, B cells are increasingly recognized as modulators of tumorigenesis and may have an inhibitory effect on the anti-tumor immune response. Ibrutinib, a potent inhibitor of the B cell signaling enzyme Bruton’s tyrosine kinase (BTK), is approved for the treatment of B cell malignancies and may serve as an efficacious adjunct to traditional therapies by depleting inhibitory B cells and thereby increasing the anti-tumor response. The aim of this study is to test the safety and efficacy of neoadjuvant ibrutinib prior to radical prostatectomy (RP) in patients with localized prostate cancer, and to further characterize the immune microenvironment in patients receiving neoadjuvant ibrutinib. Methods. Multicenter phase 1/2 clinical trial of neoadjuvant ibrutinib in men with localized prostate cancer undergoing radical prostatectomy (RP). All patients received 2 weeks of ibrutinib prior to RP and were followed for 4 weeks. Toxicities were assessed according to NCI CTCAE v. 4.03. Serum and tissue samples were collected for immune monitoring and analyzed using cytometry by time of flight (CyTOF) and immunohistochemistry. Results. The study accrued a total of 3 patients and no dose-limiting toxicities were observed. Out of the 3 patients enrolled, 1 patient (33%) experienced a clinical response, defined as a pathologic downstaging of tumor at the time of RP. All 3 patients (100%) are alive with an undetectable prostate specific antigen (PSA) at the time of data collection. CyTOF analysis revealed a significant decrease in the expression of activation markers in B cell subsets and a significant decrease in immune checkpoint expression in T cell, dendritic cell, and NK cell subsets after treatment with ibrutinib. Analysis of tumor samples collected at the time of RP demonstrated a trend towards increased expression of activation markers in tumor infiltrating T cells. Conclusions. Neoadjuvant ibrutinib was safely tolerated in our study population and preliminary findings suggest that it may induce an anti-tumor immune response prior to surgery. Treatment with ibrutinib results in decreased activation of B cell populations with a reciprocal decrease in immune checkpoint expression in multiple cell types that may lead to an increase in the number of active tumor-infiltrating T cells. Our preliminary results suggest that neoadjuvant ibrutinib may be a well-tolerated and efficacious treatment for localized prostate cancer but given our limited sample size, further investigation is needed.

Poster Session Characterizing Sparse and Dense Prostate Cancers on Whole-Mount Histopathology and on Multiparametric MRI Susan M. Noworolski, Olga Starobinets, Kyle Kuchinsky, John Kornak, Peter R. Carroll, Kirsten L. Greene, John Kurhanewicz, Jeffry P. Simko Background: Prostate cancers often present as mixtures of cancerous and benign tissues for which the cancerous tissue may be sparsely distributed, hindering detection. The purpose of the study was 1) to establish the size, incidence and Gleason Score of sparse PCa lesions on whole-mount histopathology in post-prostatectomy samples and 2) to identify the imaging characteristics of sparse cancers on mpMRI modalities. Methods: Seventy-eight men with untreated PCa, scanned with 3T-MR prior to prostatectomy were studied. Post-prostatectomy, all specimens were processed whole-mount. Cancerous regions were outlined and graded, with cancer amounts estimated. Regions with cancer occupying <50% and ≥50% were considered sparse and dense respectively. Histopathology slides were digitized and cancer region sizes estimated. Contiguous 2D cancer regions were grouped into 3D lesions. On T2-weighted MR imaging, regions of interest (ROI) were manually drawn in the peripheral (PZ) and the transition (TZ) zones. Within each patient, ROIs were grouped based on tissue type, Gleason Score, and sparse/dense composition. ROI area-weighted averages for each imaging measure were calculated for each tissue type and sparse/dense composition per patient. Results: Cancer regions were identified on histopathology (n=1193: 939 (PZ) and 254 (TZ)). Sparse lesions were primarily low-grade with the majority of PZ and 100% of TZ sparse lesions ≤GS3+3. Dense lesions were significantly larger than sparse lesions in both PZ and TZ. On imaging, within PZ and TZ, 246/45 and 109/8 dense/sparse 2D cancerous ROIs were drawn. Sparse GS3+3 and ≥GS3+4 cancers had similar imaging characteristics to dense GS3+3 cancers, while sparse PZ GS3+3 cancers differed from benign, P<0.05. Conclusions: Sparse cancers were entirely low grade in the TZ and predominantly low-grade in the PZ and likely pose limited malignant potential for spread and progression. MpMRI may prove valuable for focal treatment planning and establishing lesion margins even in the setting of cancer sparsity.

Rapid-Fire and Poster Sessions Single Cell Discrimination of Immunotherapy-Induced Changes in the Prostate Tumor Microenvironment David Y. Oh1,2, Felix Y. Feng3, Anthony Wong3, Katsuto Shinohara4, Rahul R. Aggarwal1, Terence W. Friedlander1, Eric J. Small1, Lawrence Fong1,2 1. Division of Hematology/Oncology and Genitourinary Medical Oncology, 2. Parker Institute for Cancer Immunotherapy, 3. Department of Radiation Oncology, 4. Department of Urology, University of California, San Francisco Background: Although immunotherapy prolongs survival in metastatic castrate-resistant prostate cancer, the benefit of agents such as sipuleucel-T is modest, and checkpoint inhibitors benefit <10% of patients. That checkpoint inhibitor-responsive patients often have durable responses, however, raises the question of whether specific known or novel T cell populations are mechanistically linked to these deep responses. We hypothesize that immunotherapy combining checkpoint inhibition with immunostimulatory prostate cancer therapies such as radiation and hormonal therapy induces novel populations of effector T cells whose function and antigenic specificity are enriched in prostate tumors and are not present before therapy. Methods: We will test this hypothesis using an unbiased high-resolution approach. Single-cell analysis combining whole-transcriptome RNA sequencing coupled to paired T cell receptor (TCR) analysis will characterize known or unanticipated T cell populations that are generated by combination immunotherapy in an ongoing phase II trial testing novel combinations of immunotherapy (anti-PD-1 immunotherapy +/- the Toll-like receptor 9 agonist SD-101) with radiation and androgen deprivation therapy for newly diagnosed oligometastatic hormone-sensitive prostate cancer. This work will interrogate serial prostate biopsies and blood to determine which functional T cell populations and antigenic specificities are enriched in the tumor over blood, and which are pre-existing versus newly induced by immunotherapy. Results: The first patients have been treated as part of the safety lead-in for this clinical protocol (anti-PD-1 immunotherapy without SD-101), and paired prostate biopsies have been obtained from patients before and after initiation of anti-PD-1 and radiotherapy for single cell sequencing. Ongoing progress in analyzing single-cell RNA sequencing and paired TCR data from these prostate biopsies will be presented. Conclusions: Single-cell transcriptome and TCR analysis in conjunction with a combined modality trial involving established and novel immunotherapies with radiotherapy and hormonal therapy will reveal novel T cell populations and antigenic specificities that can serve not only as biomarkers of response, but also experimental handles for prospective isolation/enhancement of novel T cell populations as well as novel prostate tumor antigens. Conflict of interest: None Funding: D.Y. Oh is supported by NIH 4T32 CA177555, the Harry F. Bisel, MD Endowed Young Investigator Award from the Conquer Cancer Foundation of the American Society of Clinical Oncology, the Bladder Cancer Advocacy Network Palm Beach New Discoveries Young Investigator Award, and the Prostate Cancer Foundation Young Investigator Award. L. Fong is supported by a Challenge Award from the Prostate Cancer Foundation.

Poster Session An Ethnographic Study of African American Men’s Prostate Cancer Treatment Decision-Making Palmer, Nynikka R.; Street, Richard L.; Schillinger, Dean; Shim, Janet K.; Blaschko, Sarah D.; Breyer, Benjamin N.; Pasick, Rena J. Background: Compared to White men, African American men bear a disproportionate burden of prostate cancer (PCa), including higher mortality, widespread under-treatment, and greater dissatisfaction with care and decisional regret. We explored the process of treatment decision-making to inform future interventions to achieve patient-centered communication and guideline-concordant care. Methods: We conducted an ethnographic cohort study at two public hospitals following African American men, from PCa diagnosis through treatment decision. Data sources included audio-recorded observations of urology and radiation oncology appointments, field notes from observations of multiple clinic appointments, and two in-depth in-person interviews with patients (post-diagnosis and post-treatment decision). We explored how patients chose their treatment, who they consulted, and the information they sought. Audio recordings of clinic appointments and in-depth interviews were transcribed verbatim. We used an iterative process of coding and team discussion to conduct a thematic analysis of qualitative data to examine the experiences, refine codes, and identify key themes. Results: Among 16 African American men newly diagnosed with PCa, four key themes emerged regarding treatment decision-making: feelings of anxiety and fear, impact of others’ cancer narratives, interactions with clinical team, and socio-cultural support. Men expressed a desire to quickly get past PCa due to anxiety and fear, and viewed surgery as their best option to achieve this goal. Patients’ sense of urgency varied based on diagnosis (low-risk vs. intermediate- or high-risk PCa). Many men reflected on negative things they heard from other people’s cancer experience, often not related to PCa, which elevated their own apprehension of certain treatment options. Patients expressed confidence in and level of comfort with their urologist and his/her recommendation, despite the absence of a second opinion or consultation with radiation oncology. Notable family/social dynamics emerged, such as men keeping the diagnosis from family members, and family members questioning care plan (e.g., why no active treatment, instead active surveillance). Partners, family members, and friends played an important role, even though they had little or no interface at clinic appointments. Conclusions: Study results thus far reveal a range of influences upon PCa treatment decision-making among African American men, including fear/anxiety and socio-cultural networks that may contribute to rushed and ill-informed decisions. A more nuanced probing of socio-cultural aspects is needed. Future studies should further explore these factors as nodes shaping decision trajectories, and therefore as potential points for intervention. For example, educational efforts could highlight the availability of time to consider treatment options and the value of second opinions. PCa survivors’ stories could also be a powerful resource, although efforts may need to demystify misinformation that incites fear.

Rapid-Fire and Poster Sessions The Communication Chasm and Prostate Cancer Diagnoses: An Ethnographic Study with African American Men Palmer, Nynikka R.; Street, Richard L.; Schillinger, Dean; Shim, Janet K.; Blaschko, Sarah D.; Breyer, Benjamin N.; Pasick, Rena J. Background: Compared to White men, African American men bear a disproportionate burden of PCa, including widespread under-treatment, higher mortality, poorer quality of life, and higher dissatisfaction with care and decisional regret. Among the reasons for suboptimal care is inadequate patient-provider communication, which can be challenging due to low health literacy, physicians’ lack of skills, and time constraints in low resource settings. An in-depth investigation into how PCa diagnoses and treatment options are communicated to African American patients was conducted to inform future interventions to achieve patient-centered communication and shared decision-making.

Methods: We conducted an ethnographic cohort study at two public hospitals of African American men newly diagnosed with PCa. Data sources included (1) audio-recorded observations of urology appointments when men received biopsy results, (2) fieldnotes from clinic observations, and (3) in-depth interviews with patients 1-3 weeks post-diagnosis. These varied forms of data allowed for a multi-faceted description of patient-provider communication at diagnosis. We explored patients’ experiences, understanding of their diagnoses and treatment options, and sources of support in qualitative interviews. Audio recordings of appointments and interviews were transcribed verbatim. We used an iterative process of coding and team discussion to conduct thematic analysis to examine patients’ experiences, refine codes, and identify key themes.

Results: Among 16 African American men newly diagnosed with PCa who completed interviews, three themes emerged: overwhelming amount of information, limited comprehension, and lack of social support. At clinic visits, urologists (N=9) delivered PCa diagnoses using their own styles, including details on patients’ risk category (low, intermediate, high), treatment options, and next steps. Some urologists drew pictures or used diagrams to convey information, yet very few assessed patient comprehension. During clinic visits, many patients appeared distressed with the diagnosis, very few asked questions, and most were unaccompanied. In post-visit interviews, patients expressed being overwhelmed with the diagnosis and related details. Many patients were either unable to or inaccurately described details of their diagnosis. For example, three patients with similar diagnoses had very different perspectives of their disease, ranging from not cancer at all to stage IV disease. Few patients reported seeking informational resources beyond their urologist (e.g., internet, peers). Many patients discussed their diagnosis with partners, family members, and/or friends, noting the value of socio-cultural support despite them being absent at appointments.

Conclusions: While urologists provided detailed information, African American men newly diagnosed with PCa were uniformly overwhelmed, unclear about specifics important to decision-making, and did not retain much information. Future studies should explore the effectiveness of a protocol to: (1) limit and prioritize the most important information to avoid overload; (2) assess patient comprehension; (3) include literacy-appropriate and tailored tools to support comprehension; and (4) promote socio-cultural support.

Poster Session

Multiple Biomarkers Independently and Additively Predict Prostate Cancer Pathology Outcomes Cooperberg MR1, Paris PL1,2, Cowan J1, Lindquist K1, Kobayashi Y1, Simko J1,3, Bengtsson H4, Singh K1, Newcomb L5, Lin D5, Stone S6, Carroll PR1 1Department of Urology, UCSF, San Francisco, CA

2Division of Hematology and Oncology, UCSF, San Francisco, CA 3Department of Pathology, UCSF, San Francisco, CA 4Department of Epidemiology & Biostatistics, UCSF, San Francisco, CA 5Department of Urology, University of Washington, Seattle, WA 6Myriad, Salt Lake City, Utah Background: Distinguishing indolent from aggressive prostate cancer remains a key challenge for prostate cancer management decision-making. A growing number of biomarkers are now on the market to help address this need, but these have rarely been examined together in the same patients to determine their relative value. Methods: We identified men with low-risk cancers defined by biopsy Gleason score ≤3+3, PSA ≤10ng/ml, and clinical stage ≤T2 who underwent immediate prostatectomy. We collected paraffin-fixed prostate tissue and pre-surgical plasma samples from N=381 cases from UCSF

and N=279 cases from UW. Plasma was analyzed for TGF1 and IL6SR, previously validated markers for prostate cancer prognosis, using ELISA. From prostate tissue, DNA was extracted and analyzed for the GEMCaP copy number variation score, and RNA was extracted and analyzed for the cell cycle progression (CCP) score, both well-validated and previously reported scores. Pathologic outcomes were identified as minor upgrading/upstaging (UGUS) (pGS 3+4 or pT3a) or major UGUS (pGS ≥4+3 or ≥pT3b), and multinomial regression was performed to determine putative markers’ ability to predict these outcomes, controlling for PSA, percent of biopsy cores positive, age, and clinical site. Results: Overall, 357 men had no UGUS event at prostatectomy, 236 had a minor event, and

67 had a major event. Neither TGF1 nor IL6SR statistically significantly predicted any UGUS or major UGUS. On the other hand, both CCP and GEMCaP were directly associated with minor UGUS on univariate analysis. On multivariable multinomial analysis including both scores, the CCP score predicted minor UGUS (OR 1.70, 95% CI 1.07-2.72, p=.03) and GEMCaP also predicted minor UGUS (OR 1.04, 95% CI 1.01-1.07, p<.01). Neither CCP nor GEMCaP predicted major UG/US, but the number of events was small. Noteworthy, both GEMCaP and the CCP score were statistically significant predictors of outcomes in the same model, suggesting they can complement each other by offering additional information. The other clinical parameters were not significant in this model. Conclusions: Biomarker signatures based on analysis of DNA and RNA significantly and independently predict adverse pathology among men with clinically low-risk prostate cancer undergoing prostatectomy.

Podium Session 3 Suppression of PD-L1 Secreted in Exosomes Promotes Systemic Anti-Tumor Immunity and Memory

Mauro Poggio1,2, TJ Hu1,2, Chien-Chun Pai3,4, Brandon Chu1,2, Cassandra D. Belair1,2,

Anthony Chang3,4, Elizabeth Montabana5, Qi Fu1,2, Lawrence Fong3,4, and Robert

Blelloch1,2,3*

Affiliations: 1 Department of Urology, University of California, San Francisco, CA, 94146, USA. 2 Eli and Edith Broad Institute for Regeneration Medicine, University of California, San Francisco, CA, 94146, USA. 3Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco CA, 94146, USA. 4 Department of Medicine, University of California, San Francisco, CA, 94146, USA. 5Division of Molecular Biophysics and Integrated Bioimaging, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA.

Antibody blockade of the immune checkpoint protein PD-L1 can lead to durable remissions in a subset of cancer patients. PD-L1 is thought to act in the tumor bed by binding its receptor PD-1 on effector T-cells. However, here we show in some cancer models -including prostate cancer- the vast majority of PD-L1 is secreted within exosomes. Exosomal PD-L1 released from the tumor suppresses T-cell activation in the draining lymph node. Removal of exosomal PD-L1 inhibits tumor growth in an immune dependent fashion, even in models resistant to anti-PD-L1 or anti-PD-1 antibodies. Moreover, exposure to exosomal PD-L1 deficient tumor cells suppresses growth of wild-type tumor cells injected at a distant site, simultaneously or many months later. Together, these findings show that inhibition of exosomal PD-L1 release can result in long-term systemic anti-tumor immunity. Therefore, it represents an unexplored target, which could overcome resistance to current antibody approaches.

Rapid-Fire and Poster Sessions

Genomic Hallmarks and Structural Variation in Metastatic Prostate Cancer

David A. Quigley*1,2, Ha X. Dang*3,4, Shuang G. Zhao*5, Paul Lloyd6, Rahul Aggarwal6, Joshi J. Alumkal7,8, Adam Foye6, Vishal Kothari9, Marc Perry9, Adina M. Bailey6, Denise Playdle6, Travis J. Barnard9, Li Zhang6, Jin Zhang10, Jack F. Youngren6, Marcin P. Cieslik11,12, Abhijit Parolia11,12, Tomasz M. Beer7, George Thomas7,13, Kim N. Chi14,15, Martin Gleave14, Nathan A. Lack14, Amina Zoubeidi14, Robert E. Reiter16, Matthew B. Rettig16, Owen Witte17, Charles J. Ryan18, Lawrence Fong6, Won Kim6, Terence Friedlander6, Jonathan Chou6, Haolong Li9, Rajdeep Das9, Hui Li9, Ruhollah Moussavi-Baygi9, Hani Goodazi19,20, Luke A. Gilbert20, Primo Lara21,22, Christopher P. Evans22,23, Theodore C. Goldstein24,6, Joshua M. Stuart24, Scott A. Tomlins11,12, Daniel E. Spratt5, R. Keira Cheetham25, Donavan T Cheng25, Kyle Farh25, Julian S Gehring25, Jörg Hakenberg25, Arnold Liao25, Phil Febbo25, John Shon25, Brad Sickler25, Serafim Batzoglou25, Karen E. Knudsen26, Housheng H. He27, Jiaoti Huang28, Alexander W. Wyatt14, Scott M. Dehm29,30, Alan Ashworth1,6, Arul M. Chinnaiyan#,11,12 ,31-34, Christopher A. Maher#,3,4, Eric J. Small#,1,6, Felix Y. Feng#,1,6,9,20

Background: While mutations affecting protein-coding regions have been examined across many cancers, structural variants at the genome-wide level are still poorly defined. Genomic structural variants (SVs) present in metastatic castration-resistant prostate cancer (mCRPC) include genomic deletions, insertions, tandem duplications, inversion rearrangements, and inter-chromosomal translocations. Gene fusions involving the E26 transformation-specific (ETS) family of transcription factors have been identified in 40-60% of mCRPC cases, and fusions joining androgen-sensitive genes to oncogenes have been described. However, the majority of SVs involve intergenic or intronic noncoding regions of the genome and are not captured by exome sequencing or transcriptome analysis. Methods: To comprehensively investigate the genomic drivers of mCRPC, we interrogated the whole genomes and transcriptomes of mCRPC samples from over 100 patients at a mean depth of 109X in tumors, a depth 2-3 times greater than that achieved in previous large WGS studies in cancer. Results: Deep sequencing of a large patient cohort permitted us to discover novel recurrent SVs and define the prevalence of these variations in mCRPC. Notably, we observed amplification of an intergenic enhancer region 624 kilobases upstream of the androgen receptor (AR) in 81% of patients, correlating with increased AR expression. Our data support the model that amplification at the putative enhancer locus results in increased AR expression, acting in an additive fashion with, and independently of, amplification of AR itself. In total, 85% of patients had either enhancer amplification or a pathogenic activating AR mutation. Tandem duplication hotspots also occur near MYC, in lncRNAs associated with post-translational MYC regulation. Classes of structural variations were linked to distinct DNA repair deficiencies, suggesting their etiology, including associations of CDK12 mutation with tandem duplications and BRCA2 inactivation with deletions. We observed chromothripsis in 23% of mCRPC patients and demonstrated that chromothripsis was significantly associated with TP53 alterations. This observation supports the proposed but unproven mechanistic association between TP53 alteration and chromothripsis. Conclusions: Our study demonstrates the utility of whole genome analysis across a clinically relevant metastatic tumor cohort, as our analysis led to multiple discoveries that eluded existing exome-centric genomic investigations in the advanced disease setting. We have provided the first landscape of structural variants in mCRPC, a substantial mutational class in this disease that will serve as a repository for other researchers to continue exploring their biological and clinical significance. Together, these observations provide a comprehensive view of how structural variations affect critical regulators in metastatic prostate cancer.

Poster Session Targeting Androgen Receptor and the ACK1 Signaling Axis in Castration-Resistant Prostate Cancer Mithila Sawant1, Jonathan Chou2, Troy Robinson2, Dhivya Sridaran1, Claire Fletcher3, Vishal Kothari2, Felix Feng2* and Nupam Mahajan1* 1Washington University in St. Louis, 2University of California at San Francisco, Helen Diller Comprehensive Cancer Center and 3Imperial College London Background: Over the years, it has emerged that tyrosine kinase and androgen receptor (AR) signaling play important roles in the progression of prostate cancer (PC) from a hormone-sensitive to a castration-resistant state. We previously uncovered a novel signaling pathway whereby AR recruits a non-receptor tyrosine kinase, ACK1, to regulate a distinct transcription program leading to AR upregulation. While enzalutamide and abiraterone are active in the castration-resistant setting, each confers only a modest survival advantage, and resistance to these drugs often develops due to upregulation of AR and/or its splice variants. We have generated a new class of small-molecule inhibitors that target the ACK1 kinase. In preclinical studies, our lead compound, (R)-9bMS, not only suppresses ACK1 activity and mitigates AR and AR-V7 expression, but also overcomes enzalutamide resistance.

Methods: In order to conduct the studies necessary to credential (R)-9bMS as a treatment approach for PC, we first performed large-scale synthesis of (R)-9bMS. We also synthesized various derivatives of (R)-9bMS and assessed their kinase specificity and microsomal stability in mouse, rat and human microsomes. In addition, we performed dose-response experiments with (R)-9bMS to determine the half maximal inhibitor concentration (IC50) in cell line models of hormone-sensitive, castrate-resistant, enzalutamide-resistant as well as taxane-resistant PC cells.

Results: (R)-9bMS is highly stable in human and rat microsomes. In addition, (R)-9bMS inhibits growth of castrate-resistant, enzalutamide-resistant and taxane-resistant PC cells and also suppresses cell migration. Moreover, (R)-9bMS is highly effective in suppressing prostate tumor growth in mice bearing human prostate cancer cell xenografts when (R)-9bMS is given orally at 130 mg/kg/day. Therapeutic experiments with drug combinations and in patient-derived xenograft (PDX) models of PC will be soon underway.

Conclusion: Our data suggest that (R)-9bMS is effective at inhibiting proliferation and migration in multiple models of PC, including castrate-resistant, enzalutamide-resistant, and taxane-resistant settings. In addition, our data suggest that oral administration of (R)-9bMS is effective at suppressing growth in mouse models of PC, and that inhibiting ACK1 using an oral formulation may be a realistic option for future clinical trials.

Funding Acknowledgements: J.C. is supported by an NCI T32 training grant (CA108462) and the A.P. Giannini Foundation. C.F. and V.K. are PCF Young Investigators. N.P.M. is a recipient of grants from the PCF (17CHAL06), NIH/NCI (5R01CA135328), Department of Defense (W81XWH-15-1-0312), and Bankhead-Coley (6BC08). F.F. is the recipient of grants from the PCF (17CHAL06).

Poster Session High Expression of miR-182 Promotes Prostate Cancer Aggressiveness in African-Americans Compared to Caucasians Shiina, Marisa; Hashimoto, Yutaka; Deng, Guoren; Varahram, Shabyari; Kulkarni, Priyanka; Dasgupta, Pritha; Bhagirath, Divya; Yamamura, Soichiro; Majid, Shahana; Saini, Sharanjot; Tanaka, Yuichi and Dahiya, Rajvir Background: African-Americans (AfA) have higher risk for developing prostate cancer and when diagnosed, the cancer is more aggressive with worse survival compared to Caucasians (CA). To clarify the mechanisms involved in this disparity, we analyzed the role of miR-182 in AfA and CA prostate cancer tissues and cells. Materials and Methods: We analyzed miR-182 expression in AfA and CA prostate cancer tissue samples by Real Time PCR. We selected two cell lines, DU-145 (CA) and MDA-PCa-2b (AfA), which express different levels of miR-182, to mimic the tissue samples and help identify the mechanisms related to racial disparity. We generated lentivirus stable miR-182 knockdown cells lines and performed cell viability, migration, invasion, apoptosis assays. Results: We found miR-182 expression to be higher in AfA prostate cancer tissues compared to CA patients with localized disease. Also, expression of miR-182 in AfA cell line, MDA-PCa-2b, was significantly higher compared to CA cell line, DU-145.. miR-182 knockdown in MDA-PCa-2b cells effectively inhibited cell migration, invasion and colony formation compared to control cells, while there was no difference in DU-145 cells. Knockdown of miR-182 caused a decrease in MDA-PCa-2b cell viability compared with negative control. In addition, the cell cycle profile showed an increase in the G1 phase of MDA-PCa-2b cells (control 7.3% compared to miR-182 knockdown cells 18.6%), whereas no change was observed in DU-145 cells. Similar to these results, wound healing was significantly slower in the miR-182 knockdown MDA-PCa-2b cells compared to controls. PCR array analysis was performed to determine the molecular effects of miR-182 knockdown in MDA-PCa-2b and DU-145 cells. We found that Angiopoietin 1 (ANGPT1), which plays a role in the regulation of angiogenesis, endothelial cell survival, proliferation, migration and adhesion was upregulated in knockdown cells. Conclusion: We conclude that high levels of miR-182 is associated with prostate cancer aggressiveness in African-Americans.

Rapid-Fire and Poster Sessions Identification of Molecular Chaperone Targets for Castration-Resistant Prostate Cancer Treatment Arielle Shkedi, Isabelle Taylor, Hao Shao, Martin Kampmann, Jason Gestwicki Background: Castration-resistant prostate cancer (CRPC) is the 3rd leading cause of cancer death among men. In CRPC, disease progression has been attributed to continued activity and signaling of the androgen receptor (AR). Strategies to treat CRPC have therefore involved targeting of the AR pathway; however, resistance typically develops, and the cancer continues to progress. Recently, molecular chaperones have emerged as a promising class of targets for CRPC treatment. Molecular chaperones are responsible for maintaining protein homeostasis within cells, by governing the folding and degradation of “client” proteins. AR is an established client of the molecular chaperone network, and normally depends on heat shock proteins (HSPs), such as Hsp70, Hsp90, and additional co-chaperones for its proper folding and activation. Due to AR dependence on the molecular chaperone network, we hypothesize that targeting chaperones is a viable option for treatment of CRPC. Methods: We aimed to identify molecular chaperones that could be targeted for the treatment of CRPC using a functional screening approach. A custom shRNA library targeting ~140 genes in the protein homeostasis network was used to identify specific chaperones essential for the survival of CRPC cells. Cells were lentivirally transduced with the shRNA library, grown for 10 doublings, and deep sequenced to determine shRNA frequencies. Using this approach, potential molecular chaperone targets for CRPC treatment were identified. Results: Using the functional genomics platform, we identified protein homeostasis network dependencies of CRPC. For example, CRPC cells have been revealed to be particularly sensitive to knockdown of Hsp60, a chaperonin involved in mitochondrial protein folding. Current efforts are now being focused on studying the biological effects of Hsp60 knockdown on CRPC tumorigenesis, understanding the role of Hsp60 on AR signaling, as well as developing inhibitors targeting Hsp60. Conclusions: Molecular chaperones are a new class of targets for the treatment of CRPC. Through functional genomics, we have identified specific molecular chaperones that are now being explored as potential targets for drug therapy. These studies suggest new drug targets for the treatment of CRPC, while also providing a possible template for studying the roles of the protein homeostasis network in other systems. Health-Related Quality of Life (HRQoL) After Disease Progression in SPARTAN: a Phase 3 Trial of Apalutamide (APA) Versus Placebo (PBO) in Men With Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC)

Poster Session

Health-Related Quality of Life (HRQoL) After Disease Progression in SPARTAN: a Phase 3 Trial of Apalutamide (APA) Versus Placebo (PBO) in Men With Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC) Eric J. Small, David Cella, Kelly McQuarrie, Fred Saad, Boris A. Hadaschik, Julie N. Graff, Hiroji Uemura, Stéphane Oudard, Margaret K. Yu, Stacie Hudgens, Angela Lopez-Gitlitz, Brendan Rooney, Mark Morris, and Matthew R. Smith Background: Compared with PBO, APA prolonged the median metastasis-free survival (MFS) by > 2 y (HR = 0.30; 95% CI, 0.24-0.36), and provided a 55% reduction in the risk of symptomatic progression (Sx PD) (HR = 0.45; 95% CI, 0.32-0.63) in patients (pts) with nmCRPC (SPARTAN study, Smith MR, et al. NEJM 2018), with no decline in HRQoL in either treatment group up to the time of developing distant metastases (Mets). Here, we report pt HRQoL following PD. Methods: 1207 pts (median age, both arms: 74 y) with nmCRPC were randomized 2:1 to APA (240 mg QD) or PBO. ADT was continued in all pts. HRQoL was assessed using the pt-reported outcome (PRO) questionnaire Functional Assessment of Cancer Therapy-Prostate (FACT-P). Following development of Mets, pts in the 2 arms received similar treatments, and PROs were collected at 4, 8, and 12 mo. Sx PD was defined as 1) development of a skeletal-related event; 2) initiation of new systemic anticancer treatment due to pain progression or worsening of disease-related symptoms; or 3) development of clinically significant symptoms due to loco-regional tumor progression requiring surgery or radiation. Descriptive statistics were performed for all FACT-P subscales. Results: Group mean PRO scores after PD were available from 341 pts and from 60 pts after Sx PD (Table). These PRO scores were similar for APA vs PBO up to 12 mo after PD. While APA delayed time to Sx PD, once Sx PD was reached there were similar numeric decreases from baseline across FACT-P subdomains up to 12 mo after Sx PD. Conclusions: Relative to PBO, pts treated with APA had a longer MFS, with no decline in HRQoL through the time of Mets, and similar HRQoL after Mets. Sx PD was delayed with APA vs PBO and was associated with a decline in HRQoL in both groups. Thus, HRQoL decline for pts treated with APA was delayed because of a longer time to Sx PD.

Table. PRO group mean scores

APA PBO

Baseline Before Mets After Metsa Baseline Before Mets After Metsa

All pts, n 797 772 157 396 384 184

Group mean (SE) FACT-P FACT-G

117.2 (0.7) 84.1 (0.4)

117.4 (0.7) 83.9 (0.5)

112.5 (1.9) 80.7 (1.3)

116.6 (1.0) 83.4 (0.7)

116.6 (1.0) 83.2 (0.7)

114.5 (1.6) 81.8 (1.1)

Baseline Before Sx

PD After Sx PDa Baseline

Before Sx PD

After Sx PDa

Sx PD subgroup, n 64 64 30 63 63 30

Group mean (SE) FACT-P FACT-G

115.2 (2.5) 84.4 (1.9)

117.0 (2.4) 84.2 (1.7)

108.6 (3.7) 78.6 (2.9)

117.8 (2.0) 85.2 (1.5)

114.5 (2.2) 82.6 (1.7)

105.6 (4.3) 75.3 (3.4)

aIncludes pts with and without subsequent approved treatment for metastatic CRPC. SE, standard error.

Poster Session Prostate-Specific Antigen (PSA) Outcomes in Patients (pts) With Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC) Treated With Apalutamide (APA): Results From Phase 3 SPARTAN Study Eric J. Small,1 Ji Youl Lee,2 Angela Lopez-Gitlitz,3 Fred Saad,4 Brendan Rooney,5 Boris Hadaschik,6 Hiroji Uemura,7 Youyi Shu,3 Margaret Yu,3 Matthew Smith8 Background: SPARTAN is a randomized phase 3, placebo (PBO)-controlled study that evaluates the impact of APA, an orally administered next-generation androgen receptor inhibitor, on metastasis prevention (metastasis-free survival [MFS]) in men with nmCRPC and a rapidly rising PSA. In SPARTAN, APA significantly improved median MFS by 2 years. APA also significantly improved time to metastasis, progression-free survival, and time to symptomatic progression; was associated with a favorable trend of improved overall survival; and significantly improved progression-free survival on second therapy. PSA response and PSA progression were followed to assess treatment responses. In this analysis, we sought to describe PSA outcomes from the SPARTAN study, and correlate PSA parameters with other clinical endpoints. Methods: Pts enrolled in SPARTAN were randomized 2:1 to APA (240 mg QD) or PBO. Androgen deprivation therapy was continued in both groups. PSA was assessed by a central laboratory and blinded to the pts and investigators. PSA response was defined as a > 50% decline from baseline and confirmed on a subsequent measurement at least 4 weeks later. Time to PSA progression was defined as the time from randomization to PSA progression according to Prostate Cancer Working Group 2 criteria. The Cochran-Mantel-Haenszel test was used for the PSA response analysis. Kaplan-Meier methods were used to estimate median time to PSA progression. Cox proportional hazard models were used to estimate the HR and 95% CI. Results: In total, 1207 pts were randomized (806 to APA, 401 to PBO). Baseline median PSA doubling time was 4.4 and 4.5 mos, and median baseline PSA was 7.78 and 7.96 ng/mL in the APA and PBO groups, respectively. We confirmed that shorter baseline PSADT is predictive of an increased risk of metastases or death. In addition, shorter PSADT was also associated with a higher likelihood of both symptomatic progression and of progression following treatment for metastatic disease, PFS2. Confirmed PSA response was observed in 90% and 2% of pts in the APA and PBO groups, respectively (RR = 40.09; 95% CI, 20.99-76.58; p < 0.0001). The median time to PSA response was 29 (range, 8-310) days in the APA group. At 12 weeks after randomization, median PSA decreased by 90% in the APA group and increased by 40% in the PBO group. A > 90% maximum decline in PSA from baseline at any time on study was observed in 66% and 1% of pts in the APA and PBO groups, respectively. Moreover, 40% of patients treated with apalutamide achieved a PSA of less than 0.2. APA significantly decreased the risk of PSA progression by 94% compared with PBO (not reached vs 3.71 mos; HR = 0.064; 95% CI, 0.052-0.080; p < 0.0001). Conclusions: In men with non-metastatic CRPC, a shorter PSADT is associated with an increased risk of metastases or death, an increased risk of symptomatic progression, and a shorter time to progression during treatment for mCRPC. Apalutamide treatment resulted in substantial and rapid declines in PSA, and a greater magnitude of PSA decline with apalutamide was associated with improvements in MFS, Time to symptomatic progression, and PFS2.

Podium Session 2 and Poster Session SPARTAN, a Phase 3 Double-Blind, Randomized Study of Apalutamide (APA) vs Placebo (PBO) in Patients (Pts) with Nonmetastatic Castration-Resistant Prostate Cancer (nmCRPC) Eric J. Small,1 Fred Saad,2 Simon Chowdhury,3 Boris A. Hadaschik,4 Julie N. Graff,5 David Olmos,6 Paul N. Mainwaring,7 Hiroji Uemura,8 Angela Lopez-Gitlitz,9 Géralyn C. Trudel,9 Byron M. Espina,9 Youyi Shu,9 Youn C. Park,9 Wayne R. Rackoff,9 Margaret K. Yu,9 Matthew R. Smith,10 on behalf of the SPARTAN investigators 1Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, CA; 2 Centre Hospitalier de l'Université de Montréal, Université de Montréal, Montreal, Canada; 3Guy’s, King’s and St. Thomas’ Hospitals, Great Maze Pond, London, UK; 4University of Duisburg-Essen, Essen, Germany; 5VA Portland Health Care System, Portland and Knight Cancer Institute, Oregon Health & Science University, Portland, OR; 6Spanish National Cancer Research Centre (CNIO), Madrid and Hospitales Universitarios Virgen de la Victoria y Regional de Málaga, Spain;7Paul Mainwaring Pty Ltd, Hamilton, Queensland, Australia; 8Yokohama City University Medical Center, Yokohama, Japan; 9Janssen Research & Development, Los Angeles, CA; 10Massachusetts General Hospital Cancer Center and Harvard Medical School, Boston, MA Background: Pts with nmCRPC are at risk for developing metastatic disease and cancer-specific mortality. There are no approved treatments for nmCRPC. APA is an orally administered next-generation androgen receptor inhibitor with antitumor activity in CRPC. SPARTAN evaluated the effects of APA on metastasis-free survival (MFS) in men with nmCRPC. Methods: Pts with nmCRPC and prostate-specific antigen doubling time (PSADT) of ≤ 10 mos were randomized 2:1 to APA (240 mg QD) or PBO. The primary end point was MFS, defined as the time from randomization to first radiographic distant metastasis (per blinded central review) or death. Secondary end points included time to metastasis (TTM), progression-free survival (PFS), time to symptomatic progression (SymProg), and overall survival (OS). Pts were eligible to receive study-provided abiraterone acetate plus prednisone after developing distant metastases. Second progression-free survival (PFS2, the time from randomization to disease progression or death after first treatment for metastatic CRPC) was also evaluated. Results: 1207 pts were randomized. Baseline PSADT was < 5 mos in both groups. APA decreased the risk of distant metastasis or death by 72% (HR = 0.28; 95% CI, 0.23-0.35; p < 0.0001), with a median MFS of 40.5 vs 16.2 mos in the PBO group. Secondary end points (TTM, PFS, and SymProg) were all significantly improved. At an interim analysis for OS, there was a trend favoring APA. At a median follow-up of 20.3 mos, 61% of APA and 30% of PBO pts were still on treatment. Rates of discontinuation due to adverse events were low in both groups (10.7% APA, 6.3% PBO). Mean baseline health-related quality of life scores were maintained with treatment, with no difference between groups over time. Of those whose disease progressed, 80% of PBO and 56% of APA pts received therapy for metastatic CRPC. PFS2 was significantly longer for APA vs PBO. Conclusions: APA significantly improved median MFS by 2 years in men with nmCRPC. APA also significantly increased TTM, PFS, SymProg, and PFS2. APA was associated with improved OS. These results support the addition of APA to androgen deprivation therapy in men with nmCRPC.

Rapid-Fire and Poster Sessions Resistance to Androgen Deprivation Therapy Leads to Altered Metabolism in Prostate Cancer Cell and Murine Models Jinny Sun, Justin Delos Santos, Robert Bok, Romelyn Delos Santos, Mark Van Criekinge, Daniel B. Vigneron, Renuka Sriram, John Kurhanewicz

Background: Mainstay treatment for patients with metastatic prostate cancer is androgen deprivation therapy (ADT). However, almost all patients eventually develop androgen-independent prostate cancer, or castration-resistant prostate cancer. Currently no reliable clinical or noninvasive imaging method can predict response to ADT, which is critical in guiding treatment decisions. The goal of this research was to identify metabolic changes associated with resistance to ADT in order to determine the best combination of hyperpolarized 13C MRI probes that can discriminate between androgen-dependent and androgen-independent prostate cancer in future patient studies. Methods: Cell labeling: LnCaP (androgen-dependent) and PC3 cells (androgen-independent) were incubated with either [U-13C]glucose or [U-13C]glutamine. TRAMP treatment and labeling: Adult male TRAMP with a solid tumor mass underwent orchiectomy (equivalent to chemically induced ADT in patients). Mice with <20% or ≥20% increase in tumor volume one-week post-orchiectomy were defined as androgen-dependent or androgen-independent respectively. [U-13C]glucose or [U-13C]glutamine was injected via tail vein. Tissue collected immediately upon sacrifice was flash-frozen in LN2. NMR: Aqueous metabolites were analyzed using 13C-decoupled 1H spectra (Fig 1a) and 2D 1H-13C HSQC to quantify fractional enrichment (FE) = [13C]/[12C+13C]. Results: After [U-13C]glucose labeling, PC3 cells have increased aspartate FE, glutamate FE, and lactate FE compared to LnCaP cells (Fig 1b). PC3 cells also have higher glutamate FE and glutathione FE than LnCaP cells after [U-13C]glutamine labeling (Fig 1c). In [U-13C]glucose infusion studies, androgen-independent TRAMP have increased aspartate FE, glutamate FE, and lactate FE compared to androgen-dependent TRAMP (Fig 1d). Preliminary results on [U-13C]glutamine infusion indicate elevated aspartate FE and glutamate FE in androgen-independent TRAMP (Fig 1e). Together these results indicate that glycolysis, oxidative phosphorylation, and glutaminolysis are significantly upregulated in androgen-independent prostate tumors. Conclusion: This study demonstrates that androgen independence results in elevated flux through glycolysis, oxidative phosphorylation, and glutaminolysis using patient-derived prostate cancer cells and a treatment-driven TRAMP model. These results suggest that a combination of hyperpolarized [1-13C]pyruvate, [2-13C]pyruvate and [5-13C]glutamine can potentially be used to noninvasively discriminate between androgen-dependent and androgen-independent prostate cancer using hyperpolarized 13C MRI.

Poster Session Functional Significance Of Catechol-O-Methyltransferase Gene In Prostate Cancer Shigekatsu Maekawa1,2,3, Taku Kato1,2, Yutaka Hashimoto1,2, Marisa Shiina1,2, Ryan K. Wong1, Varahram Shahryari1, Priyanka Kulkarni1,2, Pritha Dasgupta1,2, Divya Bhagirath1,2, Soichiro Yamamura1,2, Shahana Majid1,2, Sharanjot Saini1,2, Z. Laura Tabatabai4, Peter R. Carroll1,2, Yukio Homma3,5, Rajvir Dahiya1,2, Yuichiro Tanaka1,2 Affiliation: 1) Urology Section, San Francisco Veterans Affairs Medical Center. 2) Department of Urology, University of California, San Francisco. 3) Department of Urology, The University of Tokyo, Tokyo, Japan. 4) Department of Pathology, San Francisco Veterans Affairs Medical Center and University of California, San Francisco. 5) Japanese Red Cross Medical Center, Tokyo, Japan. Abstract: Background: Prostate cancer is one of the most common malignancies and ranks the second most common cause of cancer‐related deaths in men in the United States. Catechol-O-

methyltransferase (COMT) is a phase II enzyme that detoxifies various catechol compounds that are reactive toward DNA and damaging to the cell. Studies have shown COMT to play a protective role against cancers such as renal and breast, but their effect on prostate is not well understood. In this study, the biological properties and function of COMT in prostate cancer were studied. Methods: Expression of COMT was initially measured in normal/benign and cancerous prostate tissues by immunohistochemistry, and cell lines by real‐time PCR and western blotting. Cancerous cells displaying the lowest levels was then transfected with COMT. Gene effect on various cellular properties such as cell proliferation, migration, invasion and apoptosis, as well as growth in athymic nude mice were determined. Results: COMT protein expression was lower in cancer regions compared to benign and normal regions of prostate tissue. Cancerous DU145 and DuPro cells also had reduced mRNA levels of COMT with undetectable protein levels. Interestingly, re‐expressing COMT in DU145 and DuPro cells led to decreased cell proliferation, migration, wound healing ability and invasion, and increased apoptosis compared to vector control. COMT also inhibited cell tumor formation in animal models. As a possible target, the TNFRSF11B gene was upregulated due to COMT. Conclusions: These results demonstrate COMT to protect against prostate cancer progression and to have a functional role by affecting apoptosis. COMT may thus be a potential biomarker or therapeutic target for prostate cancer.

Poster Session MRI-Based Prostate Specific Antigen Density Predicts Gleason Score Upgrade in an Active Surveillance Cohort Samuel L. Washington III MD, Avi S. Baskin MD MPhil, Niloufar Ameli, Hao Nguyen MD PhD, Antonio Westphalen MD, Katsuto Shinohara MD, Peter R. Carroll MD MPH Background: Elevated prostate specific antigen density (PSAD) based on transrectal ultrasound (TRUS) measurements has been shown to be strongly associated with clinically significant disease and to predict progression on active surveillance for men with low stage/grade disease. We hypothesize that elevated MRI PSAD is similarly associated with increased risk of progression on subsequent biopsy. Methods: Patients with Gleason grade 3+3 on diagnostic transrectal ultrasound-guided biopsy who were managed with active surveillance and underwent at least one additional biopsy were included. Patients who underwent MRI greater than 6 months after diagnosis were excluded. Summary statistics were generated for demographic and clinical characteristics. MRI PSAD was calculated using prostate volume on MRI and PSA temporally closest to the MRI. Multivariable logistics regression models were used to evaluate the association between MRI PSAD and predictors of upgrade on serial biopsy. Results: 166 patients were included in the study. Of these patients, 74 of them were upgraded to Gleason grade ≥7 on follow up biopsy. MRI PSAD 0.15-0.225 ng/ml/ml and ≥0.0225 ng/ml/ml were significantly associated with upgrade to Gleason grade 7 compared to MRI PSAD <0.075 ng/ml/ml after controlling for age and time since diagnosis. MRI PSAD less than 0.15 was not associated with upgrade on follow up biopsy. Conclusions: MRI PSAD is significantly associated with Gleason upgrading on follow up biopsy for men initially diagnosed with Gleason grade 3+3 disease. This finding is important because surveillance MRI is increasingly being used to monitor men on active surveillance.

Rapid-Fire and Poster Sessions Genomic Risk Predicts for Molecular Imaging Detected Metastatic Nodal Disease in Prostate Cancer Melody J. Xu, Zachary Kornberg, Adam J. Gadzinski, Dongmei Diao, Janet E. Cowan, Susan Y. Wu, Lauren Boreta, Daniel E. Spratt, Hao G. Nguyen, Matthew R. Cooperberg, Elai Davicioni, Mack Roach III, Thomas A. Hope, Peter R. Carroll, Felix Y. Feng Background: The Decipher genomic classifier (GC) is increasingly used to determine metastasis risk in men with localized prostate cancer (PCa). Whether GCs predict for the presence of occult metastatic disease at presentation or subsequent metastatic progression is unknown. We sought to determine if GC scores predict for extraprostatic Gallium-68 Prostate Specific Membrane Antigen (68Ga-PSMA-11) PET positivity at presentation. Methods: Between December 2015 and December 2017, 82 patients with PCa and both GC scores and pre-treatment 68Ga-PSMA-11 PET scans were identified. Risk stratification by the National Comprehensive Cancer Network (NCCN), CAPRA, and GC scores were performed. Logistic regression was used to identify factors correlated with PSMA-positive disease. Results: 20 (24.4%) and 62 (75.6%) patients were NCCN intermediate- and high-risk, respectively; 24 (29.3%) and 58 (70.7%) were CAPRA low/intermediate- and high-risk. In contrast, only 38 (46.3%) patients had high-risk GC scores. PSMA-avid pelvic nodal involvement was identified in 24 (29.3%) patients. Pelvic nodal involvement was associated with high-risk GC scores (OR 3.3, 95% CI 1.2 – 8.9, p=0.02) and high-risk NCCN (OR 5.0, 95% CI 1.1 – 23.3, p=0.04) and CAPRA (OR 4.0, 95% CI 1.1 – 14.9, p=0.04) grouping. Nodal PSMA-avid disease (pelvic or distant) was also significantly associated with high-risk GC scores (OR 3.6, 95% CI 1.3 – 9.9, p=0.01). Conclusions: High-risk GC scores were associated with a three-fold increase in the likelihood of PSMA-avid lymph node involvement. This data suggests that GC-high patients may benefit from more nodal imaging and treatment intensification, potentially through pelvic nodal dissection, pelvic nodal or irradiation, and/or the addition of chemohormonal agents.

Poster Session Single Fraction Brachytherapy as Monotherapy for Early Stage Prostate Cancer: The UCSF Experience Melody J. Xu MD; Katherine S. Chen MD; Albert J. Chang MD, PhD; Ann Lazar, PhD; Katsuto Shinohara, MD; Adam Cunha, PhD, I-Chow Hsu, MD

Abstract: Background: High-dose rate (HDR) brachytherapy as monotherapy is an effective treatment option for localized prostate cancer, but experience with single-fraction brachytherapy treatment has been limited by studies with small sample size. We report a large single-institution experience with single-fraction HDR brachytherapy as monotherapy for early stage prostate cancer. Methods: Retrospective chart abstraction was performed to identify men treated with HDR brachytherapy as monotherapy for low to intermediate risk prostate cancer. Competing risk analyses were performed to estimate subdistribution hazard ratio (SHR) and cumulative incidence of biochemical recurrence (BCR) and prostate-cancer-specific mortality (PCSM). Results: We identified 124 men with at least 12 months of follow-up. The median follow-up time was 28 months (interquartile range [IQR] 25th to 75th percentile: 21-36 months). Overall, 21.0% of patients (n=26) were low risk, 44.4% (n=55) were favorable intermediate risk, and 34.7% (n=43) were unfavorable intermediate risk. The median PSA nadir was 0.9 ng/mL (IQR 0.3-1.6), with time to nadir of 21 months (IQR 13-30). At 2 years, the cumulative incidence of BCR was 3.5%; 0% for low risk, 4.0% for favorable intermediate risk patients, and 4.2% for unfavorable intermediate risk patients. Compared to low and favorable intermediate risk disease, BCR was significantly associated with unfavorable intermediate risk disease (SHR: 3.6, 95% CI: 1.1 to 11.1, p=0.03). PCSM was 0%. No patients experienced grade 3 or higher acute or late toxicity. Conclusions: Single-fraction brachytherapy for early stage prostate cancer was safe with promising disease control rates, especially for low and favorable intermediate risk patients. Longer term follow-up is needed.

Poster Session Role of a Novel microRNA Gene at Frequently Deleted Chromosome 8p Region in Prostate Cancer Thao Yang, Divya Bhagirath, Thao Yang, Shahana Majid, Varahram Shahryari, Soichiro Yamamura, Z Laura Tabatabai, Yuichiro Tanaka, Rajvir Dahiya, Guoren Deng, Sharanjot Saini

Department of Urology, Veterans Affairs Medical Center, San Francisco and University of California San Francisco, CA A frequent alteration in the prostate oncogenome is the loss of chromosome (chr)8p21-22 region that has been traditionally associated with the loss of homeodomain protein, NKX3.1, that plays important roles in prostate cancer (PCa) initiation. Genomic deletions of this region increase significantly with tumor grade and are associated with poor prognosis in prostate cancer suggesting a link of this region with PCa progression. We proposed and validated a novel, paradigm shifting hypothesis that this frequently deleted locus is associated with a cluster of miRNA genes – miR-3622a/b and miR-383- that are lost in PCa and play an important mechanistic role in PCa progression and metastasis by regulating Epithelial-mesenchymal-transition (EMT) and stemness. Extending our hypothesis, in this study, we evaluated the role of another miRNA gene located within this region- miR-4288- in prostate cancer. To understand the role of miR-4288 in prostate cancer, we performed miRNA expression profiling in laser capture microdissected (LCM) PCa tissues (n=64) and matched adjacent normal regions by real time PCR. miR-4288 expression was down regulated in ~66% of tissue samples. miR-4288 expression was not altered significantly in 14% cases and high miR-4288 expression was observed in 20% cases. We performed ROC (Receiver Operating Characteristic) analyses to test the diagnostic potential of miR-4288 expression. Our analyses showed that miR-4288 expression can be a significant parameter to discriminate between normal and tumor tissues with an area under the ROC curve (AUC) of 0.711 (95% CI: 0.625-0.787, P<0.0001). We found that miR-4288 exhibits 81.54% sensitivity and 51% specificity as a diagnostic parameter. Association with clinicopathological parameters of the disease showed that miR-4288 expression was not significantly correlated with age, tumor stage, Gleason grade, biochemical recurrence or serum PSA levels. However, significant correlation was observed between miR-4288 expression and race. Analyses of miR-4288 expression in prostate cell lines showed that its expression is specifically attenuated in PCa cell lines compared to normal or immortalized prostate epithelial cells. To evaluate the functional role of miR-4288 in prostate cancer, miR-4288 precursor was overexpressed in PCa cell lines followed by functional assays. miR-4288 overexpression in PC3 and LNCaP cell lines led to decreased cellular viabilities as compared to control cells. Our preliminary data suggests that miR-4288 overexpression lead to morphological changes in these cell lines consistent with mesenchymal to epithelial transition (MET). In conclusion, our data suggests that miR-4288 is commonly downregulated in prostate cancer and plays an important regulatory role.

Rapid-Fire and Poster Sessions Tracking of Long-Term B-Cell Memory Responses Using B-Cell Receptor (BCR) Sequencing in Prostate Cancer (PC) Patients (Pts) Treated with Sipuleucel-T (Sip-T) Zhang L1,3, Kandadi H2, Paciorek A3, Chang NN2, Sheikh N2, Fong L1 1Department of Medicine, University of California San Francisco, San Francisco, USA; 2Dendreon Pharmaceuticals, Inc., Seattle, USA; 3UCSF Epidemiology & Biostatistics, University of California San Francisco, San Francisco, USA Background: Sip-T is an autologous cellular immunotherapy for asymptomatic or minimally symptomatic metastatic castration-resistant prostate cancer (mCRPC). Sip-T’s induction of B-cell responses to PAP and other antigens correlates with improved survival (Sheikh 2013, GuhaThakurta 2015). Cancer vaccine-induced changes to BCR repertoire are unknown. To assess changes in BCR repertoire upon booster treatment, we compared patients retreated with sip-T (P10-1) to treatment-naïve patients (STRIDE). Methods: STRIDE pts (N=52) received sip-T with concurrent or sequential enzalutamide for mCRPC (Petrylak 2015). P10-1 pts (N=8) previously treated with sip-T for androgen-dependent PC were retreated with a booster course for mCRPC, after a median of 8.9 years (Beer 2017). Blood samples were collected at baseline (wk 0) and during (wk 2, 4)/post-sip-T (wk 6, 26, 52). Deep sequencing was performed using the ImmunoSEQ assay (Adaptive Biotechnologies). BCR diversity was assessed by clonality, and BCR dynamics by fold-change analysis (Zhang 2017). Results: BCR repertoire had significantly higher clonality in P10-1 vs STRIDE (wk 0: p=0.003, wk 2: p<0.001, wk 4: p<0.001), suggestive of a more focused BCR repertoire. P10-1 also showed increased clonality from wk 0 to 4 (p=0.063), whereas STRIDE showed a significant increase in clonality from wk 0 to 6 (p=0.039), suggesting that BCR repertoire focused earlier in P10-1. Starting at wk 2, more clones remained in the repertoire in P10-1, indicating that sip-T stimulated immunologic memory early, after 1st retreatment (p<0.05). There was less change over time (clonal shuffling) within the 100 most abundant baseline clones in P10-1 (p=0.080), suggesting more relevant clones preexisted at baseline and enriched over time. After the first two sip-T infusions, more clones contracted in P10-1 (p=0.027, p=0.014), whereas more new clones were generated in STRIDE (p=0.083, p=0.003). Conclusions: Sip-T induces long-lasting changes in the BCR repertoire. Sip-T retreatment leads to quicker focusing of BCR repertoire than initial treatment. These results are consistent with sip-T inducing durable immunologic memory.

Poster Session Exploiting the Labile Fe(II) Pool to Image Prostate Cancer with PET Ning Zhao, Yung-Hua Wang, Junnian Wei, Ryan Muir, Adam Renslo, Michael J. Evans Introduction: It has been known for decades that cancer requires a higher level of Fe(III). Once imported into the cancer cell, Fe(III) is rapidly oxidized to Fe(II). Fe(II) is stored inside the cytosol in a partially free population referred to as the “labile iron pool.” The labile iron pool is used for several essential processes like DNA synthesis and repair and energy generation in the mitochondria. We and others have shown that the elevated labile iron pool in cancer can be used to induce a chemical reaction with a trioxolane in the cytosol of the cancer cell. The chemical reaction between Fe(II) and the trioxolane results in radical-based fragmentation of the trioxolane. We and others have shown that by chemically attaching a drug to the trioxolane, improved delivery of the drug within a cancer cell can be achieved by exploiting the unusually high levels of the labile iron pool in a cancer cell. Methods: All 18F Radiolabeling were done in the Sofie Elixys box. Two mice models were used in this study, including b/6 mice and nu/nu mice. Tumor bearing mice received 200 mCi of 18F-TRX by tail vein injections. At 30 minutes, 60minutes, and 90 minutes time point post-injection, mice were euthanized and the animals were dissected for biodistribution. Results: Radiotracer uptake increased in many normal tissues over time including the liver and kidneys, which is suggestive of a chemically catalyzed mechanism of incorporation. Radiotracer activity in the blood pool was very low, even at 30 minutes post-injection, showing that 18F TRX rapidly clears from the bloodstream. Uptake in the bone was low, indicating that 18F TRX is stable to radiodeformation. Liver and Intestine showed the highest uptake in the murine model. Conclusion: High yielding radiosynthesis of a 18F-labeled trioxolane was achieved. Preliminary PET and biodistribution studies show evidence for specific binding of the 18F-TRX compound in normal mouse tissues and one prostate cancer model. We are now developing the method to reduce uptake in the liver and intestine, to improve the imaging efficacy in the prostate tumor bearing mice models.

Poster Session The Association Between Cribriform Pattern and Adverse Pathology at Radical Prostatectomy in Prostate Cancer Patients Initially on Active Surveillance Kyle B Zuniga, Samuel L Washington, Jeffry P Simko, Janet E Cowan, Imelda Tenggara, June M Chan, Matthew R Cooperberg, Peter R Carroll Background: A growing body of literature suggests that the presence of cribriform pattern (CP) in prostate biopsy specimens portends worse outcomes. Furthermore, genomic studies have demonstrated that CP tumors harbor a myriad of molecular alterations associated with more aggressive disease. We sought to examine the relationship between CP and genomic prostate score (GPS) to understand its association with adverse pathology (AP) at radical prostatectomy (RP) in patients managed with active surveillance (AS). Methods: Using the Urologic Outcomes Database (UODB), we identified 180 patients on AS who underwent GPS testing and delayed RP. Biopsy specimens were re-reviewed centrally to confirm Gleason grading. We defined AP as GS > 4+3, pT3/4, or pN1 on surgical pathology. Characteristics of patients with versus without CP were compared using T-tests and chi-square. Cox proportional hazards regression was performed to identify factors associated with AP at RP after adjusting for age, PSA density (log), percentage of positive biopsy cores, biopsy source (UCSF versus outside), and year of diagnosis. Additional models included the base covariates plus GPS, presence of CP, and both GPS and CP. Results: For the cohort, mean age was 60.7 years (SD 7.1) with median PSA of 5.3 ng/ml (interquartile range 4.2, 6.9) at diagnosis. Median PSAD was 0.14 (interquartile range 0.10, 0.20). Among 180 patients, 27 (15%) had CP at biopsy and all but one had Gleason 3+4 on biopsy. Mean GPS differed significantly between CP and non-CP patients 35 (SD 12.7) versus 27 (SD13.7) (p<0.01). On multivariable Cox regression, CP was statistically significantly associated with risk of AP (HR 1.9, 95% CI: 1.1-3.4, p=0.02) but was no longer statistically significantly associated when adjusting for GPS (HR 1.5, 95% CI: 0.8-2.8, p=0.20). Conclusions: The presence of CP at biopsy was significantly associated with higher GPS at RP. While both were associated with adverse outcomes at RP, CP was not associated with AP independent of GPS, suggesting an interaction or commonality between these factors not previously explored.

Poster Session Trends in Complementary and Alternative Medicine Use Among Prostate Cancer Patients Kyle B Zuniga, Shoujun Zhao, Benjamin Cedars, Janet E Cowan, Stacey A Kenfield, Erin L Van Blarigan, Jeanette M Broering, Peter R Carroll, June M Chan Background: Complementary and alternative medicine (CAM) use is common among prostate cancer (PCa) patients. While some studies have clarified the effect of certain CAMs on PCa risk and prognosis, the benefits and harms of the majority of CAMs remain inconclusive. It is important to understand how use of specific CAM types has changed over time to guide patient counseling and prioritize research needs. Therefore, we investigated trends in CAM use using the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE), a large, community-based, national registry of PCa patients. Methods: CaPSURE subjects who completed questionnaires regarding use of over 50 CAM types were identified as ever-users if they indicated CAM use on any questionnaire. We grouped subjects by diagnosis year, created year ranges based on number of subjects, and calculated percent change in ever use between the first period (1996-1999) and the most recent period (2010-2017). Finally, we examined associations of CAM use with sociodemographic, clinical, and treatment characteristics using chi-squared tests. Results: Of the 7,989 respondents, 56.4% reported any CAM use. Commonly used specific CAM types were multivitamins (40.0%), calcium/vitamin D supplements (28.2%), and essential fatty acids and omega-3 (23.7%). Between patients who were diagnosed in 1996-1999 and 2010-2017, overall CAM use had a relative increase of 55% (i.e., from 42% to 66%). For specific CAM, a dramatic relative increase was seen in marijuana (513.9%) and low-meat diets (221.6%), whereas a dramatic relative decrease was seen in selenium (-64.7%) and homeopathic medicine (-33.6%). CAM use was associated with sociodemographic characteristics, including white race, college education, and high income, and clinical characteristics like lower T-stage and lower PSA at diagnosis. Conclusion: PCa patients reported widespread CAM use in 1996-2017. Multivitamins were the most prevalent CAM, and use of marijuana increased dramatically. This information may help to prioritize research needs and reinforces the importance of discussing CAM use with patients and referring to specialists (e.g., nutritionists) when indicated.