Increased Frequency of Clonal Hematopoiesis of Indeterminate Potential in Bloom Syndrome Probands and Carriers

Background

Bloom Syndrome (BSyn) is an autosomal recessive disorder caused by biallelic germline variants in BLM, which functions to maintain genomic stability. BSyn patients have poor growth, immune defects, insulin resistance, and a significantly increased risk of malignancies, most commonly hematologic. The malignancy risk in carriers of pathogenic variants in BLM (BLM variant carriers) remains understudied. Clonal hematopoiesis of indeterminate potential (CHIP) is defined by presence of somatic mutations in leukemia-related genes in blood of individuals without leukemia and is associated with increased risk of leukemia. We hypothesize that somatic mutations driving clonal expansion may be an underlying mechanism leading to increased cancer risk in BSyn patients and BLM variant carriers.

Methods

To determine whether de novo or somatic variation is increased in BSyn patients or carriers, we performed and analyzed exome sequencing on BSyn and control trios.

Results

We discovered that both BSyn patients and carriers had increased numbers of low-frequency, putative somatic variants in CHIP genes compared to controls. Furthermore, BLM variant carriers had increased numbers of somatic variants in DNA methylation genes compared to controls. There was no statistical difference in the numbers of de novo variants in BSyn probands compared to control probands.

Conclusion

Our findings of increased CHIP in BSyn probands and carriers suggest that one or two germline pathogenic variants in BLM could be sufficient to increase the risk of clonal hematopoiesis. These findings warrant further studies in larger cohorts to determine the significance of CHIP as a potential biomarker of aging, cancer, cardiovascular disease, morbidity and mortality.

Multimodal Biomarkers That Predict the Presence of Gleason Pattern 4: Potential Impact for Active Surveillance

PURPOSE

Latent grade group ≥2 prostate cancer can impact the performance of active surveillance protocols. To date, molecular biomarkers for active surveillance have relied solely on RNA or protein. We trained and independently validated multimodal (mRNA abundance, DNA methylation, and/or DNA copy number) biomarkers that more accurately separate grade group 1 from grade group ≥2 cancers.

MATERIALS AND METHODS

Low- and intermediate-risk prostate cancer patients were assigned to training (n=333) and validation (n=202) cohorts. We profiled the abundance of 342 mRNAs, 100 DNA copy number alteration loci, and 14 hypermethylation sites at 2 locations per tumor. Using the training cohort with cross-validation, we evaluated methods for training classifiers of pathological grade group ≥2 in centrally reviewed radical prostatectomies. We trained 2 distinct classifiers, PRONTO-e and PRONTO-m, and validated them in an independent radical prostatectomy cohort.

RESULTS

PRONTO-e comprises 353 mRNA and copy number alteration features. PRONTO-m includes 94 clinical, mRNAs, copy number alterations, and methylation features at 14 and 12 loci, respectively. In independent validation, PRONTO-e and PRONTO-m predicted grade group ≥2 with respective true-positive rates of 0.81 and 0.76, and false-positive rates of 0.43 and 0.26. Both classifiers were resistant to sampling error and identified more upgrading cases than a well-validated presurgical risk calculator, CAPRA (Cancer of the Prostate Risk Assessment; P < .001).

CONCLUSIONS

Two grade group classifiers with superior accuracy were developed by incorporating RNA and DNA features and validated in an independent cohort. Upon further validation in biopsy samples, classifiers with these performance characteristics could refine selection of men for active surveillance, extending their treatment-free survival and intervals between surveillance.

Radiographic-pathologic concordance in the workup of locally radiorecurrent prostate cancer

313

Background: Advanced molecular PET/CT (mPET) studies are increasingly being utilized in conjunction with multiparametric MRI (mpMRI) to evaluate the burden of radiorecurrent disease in men who develop a biochemical recurrence following definitive radiotherapy (RT) for prostate cancer (PCa). However, radiographic concordance with pathologic confirmation of radiorecurrent disease in this setting is poorly described. We sought to conduct a patient-level analysis comparing concordance of radiographic and pathologic findings between mpMRI and mPET. Methods: Men who had previously undergone definitive RT for PCa and subsequently experienced treatment failure defined by the Phoenix definition were enrolled in a prospective registry study wherein radiographically identified local PCa recurrences were biopsied using mpMRI or mPET fusion (Artemis) with real-time ultrasound. Prior to biopsy, men underwent diagnostic imaging with mpMRI, advanced mPET (68Ga-PSMA-11 or 18F-FACBC), or both in order to identify a biopsy target. At least one imaging modality had to reveal a recurrent lesion based on PIRADS or PROMISE imaging classifications in order to prompt biopsy. Radiographic and pathologic findings were classified as either “treatment effect” or “recurrent disease”. Using biopsy as the reference standard, positive predictive value (PPV) was evaluated for mpMRI and mPET modalities separately. Results: Of 28 patients with radiographic recurrence on mpMRI or mPET, 10/28 (35.7%) exhibited treatment effect without evidence of active cancer on biopsy confirmation. Prostate adenocarcinoma was identified in 17/28 patient biopsies, whereas small cell prostate cancer was present in 1 patient. All 28 men underwent mpMRI prior to biopsy and 23/28 (82.1%) additionally underwent mPET; 19/23 (82.6%) underwent 68Ga-PSMA-11 and 4/23 (17.4%) were imaged with 18F-FACBC. Concordance in the assessment of recurrent disease between mpMRI and mPET was achieved in 12/23 (52.2%) men who underwent both imaging modalities. Among the 28 men who underwent mpMRI, PPV was 0.84, whereas PPV for the 23 men who underwent mPET was 0.70. Conclusions: In patients for whom clinical suspicion of radiorecurrence was high enough to warrant a biopsy, pre-biopsy mpMRI outperforms mPET in terms of PPV for detecting pathologically confirmed locally radiorecurrent PCa. Used in tandem, mpMRI and mPET might better select appropriate candidates for biopsy than either radiographic modality alone. While advanced mPET remains promising for detecting distant recurrences at the time of RT failure, biopsy confirmation following radiographic detection of local radiorecurrence remains essential for evaluating the true burden of local recurrence.

Multimodal biomarkers overcome sampling bias to predict presence of aggressive localized prostate cancer

209

Background: Histopathologic investigation of diagnostic prostate biopsies both confirms the presence of disease and estimates its potential for distal spread via tumour grade. The accuracy of biopsy grading is limited by intra-tumoral heterogeneity, inter-observer variability, and other factors. To improve risk stratification at the time of diagnosis, we sought to create objective molecular biomarkers of radical prostatectomy grade that are resistant to sampling error and should be useful when applied to biopsy tissue. Methods: We developed and validated a robust objective biomarker of prostate cancer grade using pathologic grading of prostatectomy tissues as the gold standard. We created training (333 patients) and validation (202 patients) cohorts of Cancer of the Prostate Risk Assessment (CAPRA) low- and intermediate-risk prostate cancer patients. To address intra-tumoral heterogeneity, each tumor was sampled at two locations. We profiled the abundance of 342 mRNAs complemented by 100 canonical DNA copy number aberration loci (CNAs) and 14 hypermethylation events. Using the training cohort with cross-validation, we evaluated models for training classifiers of pathologic Grade Group ≥2, Restricting to strategies resulting in true negative rates ≥0.5, true positive (TP) rates ≥0.8, we selected two strategies to train classifiers, PRONTO-e and PRONTO-m. Results: The PRONTO-e classifier comprises 353 mRNA and CNA features, while the PRONTO-m classifier comprises 94 mRNA, CNA, methylation and clinical features. Both classifiers (PRONTO-e, PRONTO-m) validated in the independent cohort, with respective TP rates of 0.809 and 0.760, false positive rates of 0.429 and 0.262, F1 scores of 0.709 and 0.724, and AUCs of 0.792 and 0.818. Conclusions: Two classifiers were developed and validated in separate cohorts, each achieved excellent performance by integrating different types of molecular data. Implementation of classifiers with these performance characteristics could markedly improve current active surveillance approaches without increasing patient morbidity and may help better inform patients on their individual need for definitive therapy versus active surveillance.

Multiparametric magnetic resonance imaging of multifocal prostate cancer to reveal intra-prostatic genomic heterogeneity and novel radio-genomic correlates: Results of the Smarter Prostate Interventions and Therapeutics (SPIRIT) study

20

Background: Multi-focality and heterogeneity in prostate cancer can confound the selection of appropriate clinical management. Our study aimed to explore radio-genomic correlations using multiparametric magnetic resonance imaging (mpMRI) against a histopathologic reference standard. Methods: Eight men with prostate cancer who underwent mpMRI followed by prostatectomy were selected for this pilot. Whole-mount histopathology was digitized and co-registered to corresponding MRI slices using a validated high-fidelity methodology.(1) Foci, including central/transitional and peripheral zone lesions were identified by a pathologist, and contoured on digitized histopathology specimens and these digitized maps were used to guide macrodissection of the individual foci for genomic copy-number aberration (CNA) analysis. Correlation of radiomics signatures with the histologic findings and genomic analysis was performed. Results: We found a broad range of CNAs revealing inter-patient and intra-prostatic heterogeneity. Recurrently-altered loci ( e.g., 8p21) containing genes of known significance ( e.g., NKX3-1) were observed. Only radiomic features derived from apparent diffusion coefficient (ADC) independently correlated with both Gleason grade (Rho=-0.62, p=0.003) and median CNA burden (Rho=-0.68, p<0.001). While greater CNA burden expectedly correlated with higher grade, intermediate-grade (Gleason score 3+4 or 4+3) lesions appeared more like either high-grade (Gleason scores ≥4+4) or low-grade (Gleason score 3+3) disease when clustered based on CNA and ADC metrics. Conclusions: These findings suggest ADC derived radiomic metrics may be a useful imaging biomarker across both central and peripheral zone lesion and could aid in further characterization of intra-prostatic biologic heterogeneity. These proof-of-principle data reveal novel radio-genomic correlations that could supplement histologic grading and conventional imaging, thus warranting expanded study and validation. 1) Int J Rad Oncol Biol Phys. 2016; 96(1):188-96.

Association of black race with improved outcomes following definitive radiotherapy with androgen deprivation therapy for high-risk prostate cancer: A meta-analysis of eight randomized trials

327

Background: Though Black men with prostate cancer are more likely to have aggressive disease features than White men, race-specific differences in initial treatment responses in localized disease remains unknown. Methods: Individual patient data were obtained for 9259 patients (including 1674 [18.1%] Black men and 7585 [81.9%] White men) enrolled on eight randomized controlled trials evaluating definitive radiotherapy (RT) ± short-term or long-term androgen deprivation therapy (STADT and LTADT). The primary endpoints were biochemical recurrence (BCR), distant metastasis (DM), and prostate cancer-specific mortality (PCSM). Fine-Gray subdistribution HR (sHR) models were developed to evaluate the cumulative incidences of all endpoints after stratification by National Comprehensive Cancer Network risk grouping. A meta-analysis was done to estimate pair-wise comparisons of treatments within and between Black and White men, after adjusting for age, Gleason score, clinical T stage, and initial PSA. Results: Black men were more likely to have NCCN high-risk disease at enrollment (656/1674 [39.2%] vs 2506/7585 [33%], p<0.001). However, within the high-risk stratum Black men had lower 10-year rates of BCR (46.1% vs. 50.4%, p=0.02), DM (14% vs. 21.6%, p<0.001), and PCSM (4.9% vs. 9.8%, p<0.001). After adjusting for age and disease characteristics, Black men with high-risk prostate receiving RT+STADT had lower rates of BCR (sHR 0.73, 95% CI 0.62-0.86, p<0.001), DM (sHR 0.64, 95% CI 0.49-0.84, p=0.001) and PCSM (sHR 0.49, 95% CI 0.25-0.95, p=0.04). There were no differences in BCR, DM, or PCSM among men receiving RT+LTADT. The interaction between race and the impact of adding STADT to RT alone on BCR was statistically significant (p=0.003). Conclusions: Black men enrolled on randomized trials with long-term follow-up have higher risk disease at enrollment, but have better BCR, DM, and PCSM outcomes with RT-based therapy compared with White men, particularly with the addition of STADT.

The impact of intratumoral heterogeneity on prognostic biomarkers in localized prostate cancer

46

Background: Genomic biomarkers can identify patients that harbour aggressive disease. The utility of these biomarkers is uncertain due to genomic variation between prostate biopsy specimens. To quantify the robustness of genomic biomarkers, we performed spatio-genomic characterization of distinct tumor foci. We scored three validated DNA-based biomarkers of early biochemical recurrence: percentage of genome with a copy number aberration (PGA), a 100-loci biomarker, and an optimized 31- loci biomarker derived from the previous. For each biomarker, we determined their robustness to intratumoral heterogeneity in association with predicting early biochemical recurrence (eBCR; ≤18 months) and long term control (LTC; ≥48 months). Methods: We queried a registry of 1054 patients with high-risk prostate cancer who underwent a radical prostatectomy (RP). We developed a cohort (n = 42) risk matched by clinicopathologic prognostic indices. Half of the patients had eBCR, while the other half had LTC. We profiled multiple tumor foci per patient, analyzing 119 tumor foci. For each focus, CNA profiles were generated, and three biomarker scores were calculated. For each patient and biomarker, we calculated the score of the lowest-score region, the highest-score region, or sampling of all foci and use the mean score. Results: All three biomarkers distinguished LTC from eBCR. PGA scores separated the two groups with an area under the receiver operator curves (AUC) ranging from 0.75-0.80. The 100- and 31-loci signatures, had AUCs ranging from 0.76-0.85 and 0.76-0.80 respectively. Using Cox proportional hazards modeling, we found that PGA was associated with LTC (Hazard ratio (HR) range: 2.56-6.22; p < 0.05. This was replicated for the 100-loci signature (HR range: 3.55-5.23; p < 0.05). The 31-loci detected associations with eBCR independent of how different foci were summarized (log-rank p-value range: 5.1 x 10-4- 5.9 x 10-3). Conclusions: Despite divergence in biomarker scores, all three predicted eBCR. Our study suggests that genomic biomarkers can overcome intratumoral heterogeneity, making discrete samples potentially adequate in patients with high-risk disease to determine the risk of eBCR after radical treatment.

The molecular hallmarks and clinical consequences of tumor hypoxia in prostate cancer

81

Background: Localised prostate cancers are classified into risk-groups using clinical measurements like grade and stage to inform treatment decisions. However, these groupings are imprecise: ~30% of intermediate-risk patients suffer relapse of their disease despite precision image-guided radiotherapy or radical prostatectomy. One reason for this variability in response to treatment is the underlying cellular and molecular heterogeneity of tumours. Prostate tumour cells exist within a microenvironment characterized by gradients of oxygen levels and prostate tumours with low levels of oxygen (hypoxia) have poor clinical outcomes. Methods: Hypoxia was measured using multiple mRNA-based signatures. We examined 548 patients with localised prostate cancer and statistically assessed the association of hypoxia with copy-number alterations (CNAs), single-nucleotide variants (SNVs), genomic rearrangements, focal genomic events ( i.e. kataegis, chromothripsis), telomere length, clinical indices ( i.e. grade, stage) and subclonal architecture. Results: Elevated hypoxia was associated with allelic loss of PTEN, higher rates of chromothripsis and intraductal and cribriform carcinoma (IDC-CA). To translate these findings into a biomarker for prostate cancer precision medicine, we integrated tumour microenvironmental data with genomic and pathological information to stratify patients into distinct prognostic groups. Patients with localized prostate cancer that have polyclonal tumours with elevated hypoxia, allelic loss of PTEN and IDC-CA were at the highest risk of rapid biochemical failure (P = 3.48 x10-3, Logrank test) and metastasis (P = 4.61 x 10-3, Logrank test), even after controlling for T-category, Gleason score and pre-treatment PSA. Conclusions: These data suggest that the aggressiveness of prostate cancers is driven by the interplay of the tumour microenvironment, tumour evolutionary trajectories and its genomic mutational profile.

Abstract P6-19-01: Evaluation of multiple transcriptomic gene risk signatures in male breast cancer

Introduction: Male breast cancer (MBC) is a rare disease accounting for less than 1% of all breast cancers (BC) and 1% of all cancers in males. The clinical management is largely extrapolated from female BC. Several multigene assays are increasingly used to guide clinical treatment decisions in female BC, however there is little data on the utility of these tests in MBC.

Methods: Here we present the gene expression results of 380 M0, ER+ve, HER2-ve MBCs enrolled in the Part 1 (retrospective joint analysis) International Male Breast Cancer Program of 1483 patients diagnosed between 1990-2010 (Cardoso et al. Annals of Oncology, 2018). Using a custom Nanostring™ panel comprised of the genes from the commercial risk tests Prosigna®, OncotypeDx® and Mammaprint®, risk scores and intrinsic subtyping data were generated to recapitulate the commercial tests as described by Bayani and Yao et al (npjBreast Cancer, 2017). Survival outcomes by risk classification were analyzed using Cox models with time-dependent covariates when the proportional hazard assumption was not met and adjusted for clinical and treatment variables.

Results: Prosigna-like risk scores identified 99 (26.1%) as low-risk, 159 (41.8%) as intermediate-risk, and 122 (32.1%) as high-risk. Using the TAILORx cut-off (25) for OncotypeDx-like risk of recurrence scoring, 158 (41.6%) were identified as low-risk, while 222 (58.4%) were identified as high-risk. MammaPrint-like results identified 175 (46.1%) as low-risk and 205 (53.9%) as high-risk. Overall, patients classified as high-risk had higher grade, more nodal involvement, larger tumors, and more frequently treated with chemotherapy than low-risk patients. Survival analyses demonstrated clear clinical utility for each test, showing patients at high-risk with poor relapse-free survival (RFS) as compared to patients classified as low-risk: Prosigna-like RFS at 3-years (HR=2.20, 95% CI, 1.28-3.80); Oncotype-like RFS at 3-years (HR=1.92, 95% CI, 1.17-3.17); MammaPrint-like RFS (HR=1.51, 95% CI, 1.00-2.27); with similar findings for distant relapse-free survival (DRFS) and overall survival (OS). Across outcomes and all gene signatures, patients with concordant Low/Low risk classification had better prognosis than those with concordant High/High risk classification. PAM50 intrinsic subtyping identified 147 (38.7%) as Luminal A, 57 (15.0%) as Luminal B, 80 (21.1%) as Her2-enriched and 96 (25.3%) as Basal-like; showing overall 34.5% concordance to clinic-pathological subtyping by central pathology (95% CI, 29.7%-39.5%). Comparison between the tests in the MBC cohort and a comparable cohort of female BC from the Tamoxifen Exemestane Adjuvant Multinational (TEAM) trial processed in the same way will be presented.

Conclusion: Common transcriptomic assays designed to assess residual risk, validated in female BC, provide similar information in male BC patients. Not surprisingly, disagreement between test results at the individual patient level was observed. To our knowledge, this is the largest study of MBC assayed to generate risk scores of the current commercial BC tests to demonstrate their clinical utility and their differences and similarity to female BC.

This work has been funded by the Breast Cancer Research Foundation (BCRF).

Citation Format: Bayani J, Poncet C, Yao CQ, Crozier C, Anouk N, Piper T, Cunningham C, Sobol M, Aebi S, Benstead K, Bogler O, Dal Lago L, Fraser J, Hilbers FH, Hedenfalk I, Korde L, Linderholm B, Martens J, Middleton L, Murray M, Kelly C, Nilsson C, Nowaczyk M, Peeters S, Peric A, Porter P, Schröder C, Rubio IT, Ruddy KJ, van Asperen C, Van Den Weyngaert D, van Deurzen C, van Leeuwen-Stok E, Vermeij J, Winer E, Boutros PC, Giordano SH, Cardoso F, Bartlett JM. Evaluation of multiple transcriptomic gene risk signatures in male breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P6-19-01.

Personalized risk stratification for patients with early prostate cancer (PRONTO): A Canadian team biomarker project

109

Background: Current practice stratifies men with prostate cancer into risk groups based primarily on Gleason grade. When applied to biopsy samples, the Gleason grading is inaccurate due to sampling error and inter-observer variation. The result is that men either receive unnecessary surgical treatment, or they don’t receive adequate treatment, leading to worse outcomes. Previously published genomic tests have not successfully distinguished indolent low grade (G6 or GG1) cancers from their more aggressive intermediate grade (G7 or GG2 and 3) counterparts. PRONTO is specifically aimed at creating a multi-modal risk stratification tool to improve treatment stratification following a core biopsy diagnosis. Methods: PRONTO links 7 projects, each with novel diagnostic assays for risk stratification that focus on analysis of copy number variations (CNV), DNA hypermethylation, trans-differentiation, cancer metabolism, or the tumor microenvironment. We merged the best transcripts from each project into a single NanoString gene expression assay, measuring 393 transcripts, in a cohort of 365 cases of radical prostatectomy from low-to-intermediaterisk patients. To minimize sampling error, we took multiple samples, and obtained high grade, low grade and benign areas for each radical prostatectomy case. Results: Our primary goal was to develop a multivariate molecular classifier of grade that distinguished G6 from G7 (3+4 or 4+3). Cases were randomly partitioned into five equally sized groups. A supervised machine learning algorithm (random forests) was trained on samples from four of the groups, and then evaluated by testing on the fifth group. This process was repeated for each of the five groups, yielding a combined clinical and molecular classifier. DNA methylation profiles and CNV profiles are currently being integrated into our classifier Conclusions: We have developed a multivariate classifier that distinguishes low grade from intermediate grade prostate cancer. It will be clinically validated in biopsy samples from large cohorts of early prostate cancer patients.

Microrna-198: A novel tumor suppressor in prostate cancer

92

Background: microRNAs (miRNAs) are small non-coding RNA molecules which act as repressors of gene function, and have been identified as playing substantial roles in cancer as both tumor suppressors and oncogenes. miR-198 has been reported to be down-regulated in several cancers, with low expression associated with worse overall survival. In addition, miR-198 has demonstrated tumor suppressor effects by altering several hallmarks of cancer. Despite compelling evidence in other cancers, miR-198's role in prostate cancer aggression has not yet been evaluated. Methods: Experiments were conducted by overexpressing miR-198 in three prostate cancer cell lines: DU145, LNCaP and 22RV1. To examine miR-198's effect on hallmarks of cancer in vitro, we used standard protocols to assay proliferation, colony formation, cell cycle profile, migration, and invasive potential. Gene array, qPCR, western blotting, and siRNA knockdown experiments were used to establish miR-198 downstream targets. Results: Overexpression of the candidate tumor suppressor miR-198 diminished proliferation in all prostate cancer cell lines and significantly impaired colony formation in soft agar. Subsequently, miR-198 expression was also demonstrated to reduce growth and tumor formation in vivo using LNCaP xenografts. Gene arrays and in silico target prediction identified several candidate targets, none of which have been previously linked to miR-198. MIB1, an E3 ubiquitin ligase, was the only target we have since identified to be reduced with miR-198 overexpression at both the RNA and protein levels. Knockdown of MIB1 recapitulated miR-198's effects on proliferation and colony formation, and further experiments are underway to demonstrate a direct binding relationship. An additional gene array with MIB1 siRNA will be performed to highlight pathways through which MIB1/miR-198 suppress tumorigenesis. Conclusions: Our evidence supports miR-198 as an important tumor suppressor in prostate cancer, with elevated expression impairing proliferation, colony formation, and in vivo tumor formation. This investigation into miR-198 highlights a potential role as a prostate cancer biomarker, or as a potential target of therapeutics to restore miRNA activity.

Overcoming miR-106a induced radioresistance in prostate cancer: Targeting senescence with KU-55933

77

Background: Prostate cancer is a leading cause of cancer related death in men worldwide, with recurrence being a major clinical problem after radiotherapy. There is an unmet need to better characterize radioresistant tumors and identify biomarkers to improve patient outcomes. Methods: We identified that miR-106a was overexpressed in radiation resistant cell lines compared to parental cells. We analyzed The Cancer Genome Atlas dataset to assess miR-106a expression in normal prostate, and low- to high-grade prostate tumors. To assess the functional role of miR-106a, we performed in vitro and in vivo assays for radiation response, including clonogenic survival, proliferation, senescence, and tumor xenograft growth after radiation. We performed gene array and pathway analyses to identify downstream effectors of miR-106a. Results: MiR-106a expression was significantly higher in prostate tumors with Gleason score > 7 compared to Gleason ≤ 7, suggesting miR-106a is involved in high grade disease. MiR-106a overexpression confers radioresistance in vitro and in vivo, by targeting LITAF. We now extend miR-106a’s effects to upregulation of ATM at the promoter level, thereby increasing ATM transcript and protein in the cell. Unexpectedly, we found that miR-106a’s mechanism of radioresistance through ATM upregulation does not alter DNA damage repair. ATM upregulation affects clonogenic survival through reduced senescence. KU-55933, a specific ATM kinase inhibitor, resensitizes miR-106a overexpressing cells to radiation by inducing senescence, a predominant mode of cell death in prostate cancer. Conclusions: Our research challenges the current paradigm of ATM and DNA damage repair by outlining another mechanism of radioresistance through alteration of senescence. Our findings suggest that miR-106a may be a promising biomarker for high-grade disease and radioresistant prostate cancer. In addition, we have identified a therapeutic intervention for miR-106a induced radioresistance. Improvements in bioavailability of KU-55933 may lead to its clinical use in combination with radiation therapy to radiosensitize miR-106a induced radioresistant prostate cancer.

Hypoxia related mRNA biomarker to predict biochemical failure and metastasis for prostate cancer

5

Background: Hypoxia is an important regulatory factor in tumorigenesis and is associated with a poor prognosis. Patients with high risk locally advanced disease account for 13-21% of prostate cancer cases and the ten year cancer specific survival rate for these patients is 62%. Patients with hypoxic tumours could benefit from hypoxia modifying therapeutics in addition to radiotherapy. Clinical companion biomarkers are needed to stratify patients who would benefit from hypoxia modifying therapy in addition to radiotherapy. Methods: RNA-seq analysis was performed on prostate cell lines (PNT2-C2, PC-3, LNCaP and DU145) exposed to 1% hypoxia for 24 hrs. A prostate cancer hypoxia gene signature was derived in silico using publicly available prostate gene expression data sets and the RNA-seq data. The biomarker was then independently validated in multiple cohorts of prostate cancer patients with localized diseases receiving prostatectomy alone, prostatectomy plus adjuvant radiotherapy, prostatectomy plus salvage radiotherapy, or definitive radiotherapy alone. Results: In vitro the hypoxia inducible expression of the hypoxia gene signature was tested at 1% and 0.1% oxygen of which 21 of the 28 genes were regulated by hypoxia. Patients stratified as high hypoxia were associated with significantly poorer 5-year biochemical recurrence free survival in patients undergoing prostatectomy alone, prostatectomy plus adjuvant radiotherapy and definitive radiotherapy alone. In multivariable analysis, the biomarker retained significance after correcting for confounding factors including Gleason group, PSA, a molecular classifier, etc. In another cohort of prostatectomy and salvage radiotherapy treated patients, the mRNA signature predicts metastasis free survival in both univariable and multivariable analyses. Conclusions: We derived a de novo mRNA signature based on hypoxia-regulated genes. The biomarker consistently predicts biochemical failure and metastasis for prostate cancer patients with localized disease.

A biopsy-based genomic classifier to predict biochemical failure after definitive radiation without hormone therapy in a prospective cohort of intermediate risk prostate cancer

68

Background: Recently, NCCN adopted the Zumsteg-Spratt subclassification to define NCCN favorable and unfavorable intermediate-risk prostate cancer (IR-PCa). NCCN unfavorable disease is recommended to receive combination androgen deprivation therapy (ADT) and radiotherapy. To determine if genomics could help identify a subset who may safely avoid ADT, we evaluated the Decipher genomic classifier (GC) in IR-PCa treated with dose-escalated image-guided radiotherapy (DE-IGRT) alone. Methods: Our cohort comprised of 121 patients with NCCN favorable (N = 49, 40%) and unfavorable (N = 74, 60%) IR-PCa, who received 78 Gy without ADT. Diagnostic needle biopsies with the highest Gleason score (GS) and %tumor involvement were macrodissected for RNA extraction. GC scores were determined from the Decipher prostate cancer classifier assay (GenomeDx Biosciences, San Diego, CA). Primary clinical endpoint was biochemical relapse ([BCR], PSA nadir + 2ng/ml) post-DE-IGRT. We compared association with BCR against known clinicopathologic prognostic indices and the NCCN risk strata. Results: With a median follow up of 7.5y, 24 (19%) patients experienced BCR. Individual clinical indices did not predict BCR-free survival rate (BFS). NCCN risk strata was however associated with a small but significant difference in BFS (5-y 93%, favorable vs 88%, unfavourable, P = 0.046). GC scores stratified 85 (70%), 19 (16%), and 17 (14%) men into low, intermediate, and high risk of recurrence; 5-y BFS were 95%, 89%, and 59%, respectively (P < 0.001). On multivariable analysis, a hazard ratio of 4.71 (95% CI 1.81-12.28, P = 0.0015) for BCR was observed for the GC high risk group compared to low/intermediate; NCCN risk strata and intraductal variant did not achieve significance. Conclusions: In IR-PCa men treated with DE-IGRT monotherapy, Decipher GC was an independent predictor of BCR. While most men in this our cohort were stratified as NCCN unfavorable IR-PCa, the majority were GC low risk with excellent outcomes from DE-IGRT alone. In contrast, a minority with GC high risk had suboptimal outcomes, and may benefit from ADT intensification.

Abstract P1-06-02: Comparative survival analysis of multiparametric tests in the TEAM pathology study: What to do when molecular tests disagree?

Multiparametric assays for risk are increasingly used in the management of node-negative and node-positive hormone receptor-positive invasive breast cancer. Data from multiple sources suggests different tests may provide different risk estimates at the individual patient level1. Analysis from the TEAM pathology study (Bayani and Yao et al npjBreast Cancer, 2017) allows direct comparison of prognostic information from gene signatures in a clinical trial cohort of postmenopausal patients. Risk classifications using genes comprising the following multi-parametric tests: OncotypeDx® (Genomic Health Inc.)2,3, Prosigna™(NanoString Technologies, Inc.)4-6, Mammaprint® (Agendia Inc.)7,8 were performed. For the OncotypeDX-Like Recurrence Score (RS), RNA abundance was processed to fit the measurement range as described2,3, with classification into high, intermediate or low risk groups based the derived RS and modeled for DRFS. For the Prosigna-Like Risk of Recurrence Score (ROR), samples were processed as previously outlined9, then modelled against DRFS. For the MammaPrint-Like Risk Score, samples were processed by published methods8 and modelled for DRFS. Comparing OncotypeDx-Like with Prosigna-Like showed that 45% of cases were classified identically by both (3.3% low risk, 20.9% intermediate, 20.7% high). Of 3370 cases, 353 (10.5%) had scores differing by more than 1 classification (i.e. hi/low or low/high). Almost all (343) of these were cases classified high risk by OncotypeDX-Like RS/low risk by Prosigna-Like ROR (Table 1). Univariate Cox regression analysis, using low/low cases as a reference (relative risk of distant metastasis =1.0), suggested that cases called low risk by Prosigna-Like ROR/High risk by OncotypeDx-Like RS did not perform differently from cases called low risk by both tests (Table 2). However, all cases called intermediate by one test and high risk by another appeared to be high risk (Table 2). Comparisons between Prosigna-Like ROR and MammaPrint-Like scores showed similar concordance between low/low and high/high (52.5% of cases with concordant results). In Prosigna-Like ROR intermediate risk cases, MammaPrint-Like results divided cases between low and high risk, as predicted. Comparisons between these tests is challenging, and evidence on their discordance in risk stratification presents further dilemmas. Preliminary analysis of TEAM suggests a complex inter-relationship between test results in the same patient cohorts requiring careful evaluation.

Table 1OncotypeDX-Like RSLowInt.HighTotalLow1126163431071Prosigna-Like RORInt.1677046151486High10106697813Total289142616553370

Table 2OncotypeDX-Like RSLowInt.HighLowRef1.26 (0.57-2.79)1.13 (0.49-2.62)Prosigna-Like RORInt.1.2 (0.47-3.05)2.22 (1.03-4.78)4.27 (2.01-9.08)High6.10 (1.58-23.6)4.15 (1.79-9.59)4.92 (2.32-10.42)

Citation Format: Bartlett JMS, Bayani J, Kornaga E, Piper T, Mallon E, Yao CQ, Boutros PC, Hasenburg A, Kieback DG, Markopoulos C, Dirix L, Seynaeve C, Can de Velde CJH, Rea DW. Comparative survival analysis of multiparametric tests in the TEAM pathology study: What to do when molecular tests disagree? [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P1-06-02.

Abstract P2-10-01: Validation that a histone gene signature predicts anthracycline response in early breast cancer

Background: The use of anthracycline-based chemotherapies has improved overall and disease free survival in breast cancer. However, anthracyclines can have significant toxicities including cardiotoxicity and leukemia. It is, therefore, imperative to identify those patients who will benefit from adjuvant anthracycline treatment and patients who could be spared unnecessary toxicities and be considered for alternative adjuvant therapy. Previous work performed by our laboratory identified a histone gene expression signature as a predictive marker of anthracycline benefit in the BR9601 clinical trial. In this study we validate the 18 histone gene signature in the MA.5 clinical trial and examine the role of the signature in individual intrinsic subtypes of breast cancer.

Methods We analysed the CCTG MA.5 clinical trial in a prospectively planned retrospective biomarker approach to validate this signature and tested the role of intrinsic subtyping as predictive markers of anthracycline benefit. RNA was extracted from patients in the MA.5 adjuvant trial evaluating the addition of epirubicin (E) to CMF and analysed using NanoString technology. Log-rank analyses validated the predictive values of the signature on distant relapse-free survival (DRFS). Cox-regression models tested independent predictive value on DRFS in the presence of treatment, age, tumour size, nodal status, HER2, ER status and grade, and treatment by marker interactions.

Results Analysis of the MA.5 clinical cohort revealed that patients whose tumour had low histone gene signature expression experienced increased DRFS (HR: 0.54, 95% CI 0.38-0.76, p=0.001) when treated with CEF compared with patients treated with CMF alone. Conversely, there was no apparent benefit of CEF vs CMF in patients with high histone gene expression signature (HR: 1.01, 95%CI 0.66-1.55, p=0.963). After multivariate analysis and adjustment for HER2, nodal status, age, grade and ER, the treatment by marker interaction for the gene signature was 0.54 (95%CI 0.31-0.94, p=0.030) for DRFS.

The predictive impact of the histone signature was independent of intrinsic subtype.

Conclusion The histone gene expression signature is an independent predictor of anthracycline benefit and could be a potential candidate diagnostic assay for patients with early breast cancer.

Citation Format: Spears M, Kornaga E, Lyttle N, Liao L, Bayani J, Quintayo M-A, Yao CQ, D'Costa A, Boutros PC, Twelves CJ, Pritchard KI, Levine MN, Nielsen TO, Shepherd L, Bartlett JMS. Validation that a histone gene signature predicts anthracycline response in early breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-10-01.

Targeted sequencing in a phase III trial of luminal breast cancer: Identification of novel targets

505

Background: The International Cancer Genome Consortium and The Cancer Genome Atlas have had a global transformative impact on our understanding of cancer. These programs have mapped the genomic landscape of common and rare tumors setting the scene for a comprehensive change in the approach to cancer diagnosis and treatment. However, the task remains incomplete until these mutational events are linked to clinical outcomes in the context of current therapeutic intervention to drive future stratified medicine approaches. Methods: We performed targeted sequencing in patients from the Tamoxifen Exemestane Adjuvant Multicentre trial. DNA was extracted and a 101 gene panel analysed using a novel mutation calling pipeline. Both a priori and machine learning analyses were performed using distant recurrence free survival as the primary endpoint. Results: In 1,491 successfully analyzed samples 1,070 (71.76%) samples exhibited at least one single nucleotide mutation (range 0-94, 1.828+/-0.133, mean+/-s.e.). 98/101 genes were mutated in at least one patient. Only variants in PIK3CA, TP53, MLL3, CDH1 were detected in 5% or more of samples. Twenty genes were associated with increased risk of recurrence in multivariate analyses corrected for clinic-pathological variables, 50% of these genes were involved in transcriptional regulation or RNA/protein processing. In a multivariate analysis, two combined signalling modules were independently prognostic for residual risk following hormone therapy (HRvalidation 3.10 95%CI 1.78-5.40 and HRvalidation 2.70 95%CI 1.57-4.64). Conclusions: We successfully performed a signalling pathway-based targeted sequencing analysis within predefined signalling modules. In supervised and unsupervised analyses we identified multiple signalling cassettes linked to poor outcome in patients with ER+ve breast cancers treated with modern endocrine therapy in the context of a phase III clinical trial. These results identify novel candidates as targets to treat endocrine refractory breast cancers.

Copy-number and targeted sequencing analyses to identify distinct prognostic groups: Implications for patient selection to targeted therapies amongst anti-endocrine therapy resistant early breast cancers

524

Background: Hormone receptor positive breast cancer is a therapeutic challenge. Despite optimal anti-endocrine therapies, most breast cancer deaths follow a diagnosis of early luminal cancer. To understand the impact of multiple aberrations in the context of current therapy, we assessed the prognostic ability of genomic signatures as a putative stratification tool to targeted therapies. Methods: This a priori study is based on molecular pathways which might predict response to targeted therapies. DNA from 420 patients from the phase III TEAM pathology cohort were used. Patients with a distant recurrence within 5 years were matched by clinical variables to those disease-free at follow up. Copy number analysis was performed using the Affymetrix Oncoscan Assay. Targeted sequencing was performed in a subset of samples for genes based on signaling cassettes mined from the ICGC. Pathways were identified as aberrant if there were copy number variations (CNVs) and/or mutations in any of the pre-determined pathway genes: 1) CCND1/CCND2/CCND3/CDK4/CDK6; 2) FGFR1/FGFR2/FGFR2/FGFR4; and 3) AKT1/AKT2/PIK3CA/PTEN. Kaplan-Meier and log-rank analyses were used for DFS between groups. Hazard ratios were calculated using the Cox proportional hazard models adjusted for age, tumour size, grade, lymph node and HER2 status. Results: 390/420 samples passed informatics QC filters. For the CCND/CDK pathway, patients with no CNV changes experienced a better DFS (HR = 1.7, 95% CI 1.3-2.3, p < 0.001). For the FGFR/FGF pathway, a similar outcome is seen among patients without CNVs (HR = 1.5, 95% CI 1.1-2.0; p = 0.005). For AKT/PIK3CA, a decrease in DFS was seen in those with aberrations (HR = 1.4, 95% CI 1.0-1.8, p = 0.03). Conclusions: We demonstrated that CNVs of genes within CDK4/CCND, PIK3CA/AKT and FGFR pathways are independently linked to high risk of relapse following endocrine treatment, with implications for identifying those patients who are at high-risk for recurrence despite optimal anti-endocrine therapy and linking molecular features driving these cancers to targeted therapies.

Abiraterone +/- cabazitaxel in defining complete response in prostatectomy (ACDC-RP) trial

TPS5095

Background: Given recent advances in the management of de novo metastatic hormone-sensitive prostate cancer with both docetaxel and abiraterone, as well as evidence of significant activity of cabazitaxel in the post-abiraterone castrate-resistant setting, we hypothesized that the addition of cabazitaxel to neoadjuvant abiraterone will improve pathological complete response rates by overcoming mechanisms of resistance in localized high-risk prostate cancer. Aim: To determine the relative efficacy of the addition of cabazitaxel to abiraterone in the neoadjuvant treatment of prostate cancer to achieve a complete response. Methods: Open label, randomized, 2-arm multi-centre, phase 2 clinical trial. Primary endpoint: Pathological complete response rate (pCR). Secondary endpoints: surgical outcomes (positive margins, extracapsular extension, seminal vesicle or nodal involvement), pharmacodynamic markers in residual tumour (apoptosis, androgen receptor expression, localization, and signaling), biomarkers (intra-prostatic androgen levels), and safety. Design: Study participants will be randomized in a 1:1 ratio to receive either: Arm A: Abiraterone (1000 mg/day), prednisone (5 mg b.i.d.), leuprolide (22.5 mg s.c. every 3 months), and cabazitaxel (25 mg/m2 starting at week 2, with 6 mg pegfilgrastim 24 h following cabazitaxel) or Arm B: Abiraterone (1000 mg/day), prednisone (5 mg b.i.d.) and leuprolide (22.5 mg s.c. every 3 months). Assessments will take place biweekly for the first 12 weeks, then monthly until the prostatectomy (scheduled for 24 weeks following start of treatment). Target accrual is 88 participants within 36 months. Study is powered to detect a 15% difference with 85% power, assuming a one-sided type 1 error rate of 20%. A 6 patient safety run-in is included. As of Jan 2017, 1 site is open in Canada, with 4 additional Canadian sites and 1 site in Australia pending. To date, 4 participants are randomized and undergoing treatment. ACDC-RP is an investigator-initiated trial led by the Princess Margaret Urology Trials Group with funding from Ontario Institute for Cancer Research (OICR) and in-kind contributions from Janssen and Sanofi. Clinical trial information: NCT02543255.

Pdxdc1 modulates prepulse inhibition of acoustic startle in the mouse

Current antipsychotic medications used to treat schizophrenia all target the dopamine D2 receptor. Although these drugs have serious side effects and limited efficacy, no novel molecular targets for schizophrenia treatment have been successfully translated into new medications. To identify novel potential treatment targets for schizophrenia, we searched for previously unknown molecular modulators of acoustic prepulse inhibition (PPI), a schizophrenia endophenotype, in the mouse. We examined six inbred mouse strains that have a range of PPI, and used microarrays to determine which mRNA levels correlated with PPI across these mouse strains. We examined several brain regions involved in PPI and schizophrenia: hippocampus, striatum, and brainstem, found a number of transcripts that showed good correlation with PPI level, and confirmed this with real-time quantitative PCR. We then selected one candidate gene for further study, Pdxdc1 (pyridoxal-dependent decarboxylase domain containing 1), because it is a putative enzyme that could metabolize catecholamine neurotransmitters, and thus might be a feasible target for new medications. We determined that Pdxdc1 mRNA and protein are both strongly expressed in the hippocampus and levels of Pdxdc1 are inversely correlated with PPI across the six mouse strains. Using shRNA packaged in a lentiviral vector, we suppressed Pdxdc1 protein levels in the hippocampus and increased PPI by 70%. Our results suggest that Pdxdc1 may regulate PPI and could be a good target for further investigation as a potential treatment for schizophrenia.

Genomic architecture of radioresistant prostate cancer

26

Background: Spatial intra-tumoral heterogeneity of prostate cancer is secondary to genomic diversity and multi-clonality. These unique features potentially promote resistance to treatment. Here, we investigated if clonal selection or adaptation of new clones dominates in prostate cancer at the time of recurrence following radiotherapy. Methods: We identified 11 patients with biopsy-proven multifocal recurrent prostate cancer following radiotherapy. Copy number aberration (CNA) profiling was performed on 33 anatomically distinct tumor foci with 11 matched-normals. 4 cases had matched pre-radiotherapy tumors for CNA profiling to assess for clonality. We evaluated for recurrent gene amplifications and deletions, and determined genomic instability by percent genome aberration (PGA). We also compared these genomic indices against 373 sporadic prostate cancers from the Canadian Prostate Cancer Gene Network. Results: We observed large intra- and inter-patient variation (p <0.001, one-way ANOVA) in PGA scores among the radioresistant tumors. Interestingly, although total CNA counts did not differ between the radioresistant and sporadic cohorts (median = 40, radioresistant vs 33, sporadic, p = 0.20], there was a trend for increased genomic instability in the radioresistant cohort (median PGA = 8.8 vs 4.9, p = 0.059). Spatial resolution of gene-level CNAs revealed the acquisition of CNAs that were both common and non-recurrent in the multi-focal radioresistant tumors, thus suggesting a common clonal origin, with subsequent divergent evolution. Importantly, we observed a mixture of CNAs, including known prognostic genes in prostate cancer, namely NKX3-1, PTEN, TP53, CDKN1B, and CDH1,that was shared between pre-treatment and radioresistant tumors, favoring clonal selection. We also discovered a novel deleted region on Chr3p, consisting of RAD18 and FANCD2, which was uniquely present in the radioresistant tumors. Conclusions: Our novel observations in a small cohort of radioresistant prostate cancers favor the model of clonal selection, as opposed to new-onset tumors. These results support the discovery of biomarkers a priori, and targeted treatment of these radioresistant clones to improve the therapeutic ratio of precision radiotherapy.

Integrated somatic subtypes of localized intermediate-risk prostate cancer

e560

Background: Approximately two thirds of intermediate risk prostate cancer patients are over- or under- treated because we cannot correctly prognose this risk group; therefore we require novel biomarkers to better direct patient therapies and avoid subjecting patients to side effects without benefit. One reason genomic biomarkers are not currently used in clinical settings is because they are notoriously difficult to validate in follow-up studies. Furthermore, the lack of clear prostate cancer subtypes prevents the development of subtype specific biomarkers as is standard practice in breast cancer. We aim to improve biomarker validation rates by defining prostate cancer subtypes that can be used to create subtype specific biomarkers. Methods: First, we assess large scale genomic patterns using whole genome sequencing and methylation data and create integrative subtypes for intermediate risk prostate cancer. Second, we assess associations between specific aberrations and subtypes, and determine whether certain types of molecular aberrations are more important background aberrations for subtype specific biomarker development. Finally, we assess biases in prognostic performance of the published Lalonde biomarker between groups associated with patient subtypes to show that subtype aware biomarkers are necessary. Results: We demonstrate that the Lalonde biomarker is biased by the cohorts’ proportion of TMPRSS2-ERG (T2E) aberrations illustrating the need to develop different biomarkers for patients with T2E and patients without T2E. Further, we suggest integrative subtypes can be used to select patients with similar genomic profiles for biomarker analysis to improve biomarker validation rates. Conclusions: This analysis provides direct guidance for future biomarker development and addresses an important barrier to clinical use of genomic biomarkers for prostate cancer.

Abstract P3-07-32: Tumour infiltrating lymphocyte (TIL) and chemokine gene signature predicts for benefit of anthracycline-containing chemotherapy in breast cancer patients

Background: The contribution of immune cells has long been appreciated in tumour development and disease progression; however, their translational potential as cancer-associated prognostic and predictive markers was only recently recognized. High densities of tumour-infiltrating lymphocytes (TILs) correlate with improved clinical outcome in breast cancer; whether TILs also predict anthracycline benefit in all, or only a particular subgroup, of breast cancer patients remains largely unknown. Furthermore, since identification of TILs is generally based on H&E staining, it has not previously been possible to evaluate relative contribution of distinct T-cell types, and B cells, to patient outcome.

Methods: We assessed 290 patient samples from the BR9601 adjuvant breast cancer trial for the capacity of TIL contexture to predict for anthracycline (E-CMF) benefit over CMF. We immunoprofiled patient samples on the Nanostring platform to gain insight into the impact of lymphocyte populations predicting for anthracycline benefit. Our immunoprofiling panel included 38 genes representing TIL-gene signatures and chemokines that may be responsible for recruiting TILs to the tumour site.

Results: The analyses revealed two important findings. First, refinement of the 38-gene panel resulted in the generation of a novel 9-gene signature that includes cytotoxic T lymphocytes (CTL) and chemokine genes. Low CTL gene expression correlated with ER+ expression while high expression correlated with ER- expression (p&lt;0.0001), consistent with the notion that high TIL densities are predominantly observed in non-luminal breast cancers. Second, in an univariate Cox regression analysis, this 9-gene signature was a predictive biomarker of anthracycline benefit with respect to breast-cancer specific OS (HR: 0.371, 95%CI 0.158-0.868, p=0.022) and DRFS (HR: 0.395, 95%CI 0.172-0.907, p=0.028); this effect was no longer significant after adjustment for other prognostic factors (OS HR: 0.437, 95%CI 0.166-1.150, p=0.094; DRFS HR: 0.488, 95%CI 0.185-1.287, p=0.147).

Conclusion: This study highlights the significance of assessing the entire tumour since TILs, tumour and stromal cells collectively engage in a complex interplay that contributes to disease development and progression. Importantly, it reveals that not only CTLs but also chemokines may be clinically relevant and should be validated as potential biomarkers of anthracycline benefit and as therapeutic targets.

Citation Format: Braunstein M, Yao C, Lyttle N, Liao L, Boutros PC, Twelves CJ, Bartlett JMS, Spears M. Tumour infiltrating lymphocyte (TIL) and chemokine gene signature predicts for benefit of anthracycline-containing chemotherapy in breast cancer patients. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-07-32.

Abstract P1-05-01: The epithelial to mesenchymal transition: Identifying a signature of recurrence in ductal carcinoma in situ

Background: The epithelial to mesenchymal transition (EMT) plays a critical role in the progression from non-invasive to invasive breast carcinomas (IBC). It is characterized by alterations in gene expression, changes in cellular polarity, the disruption of tight junctions; production of metalloproteinases, transforming growth factor-β (TGFβ) induction, expression of cancer stem cell markers, hypoxia, decrease in e-cadherin expression, along with other molecular biological events. Several transcription factors including ZEB1/2, TWIST1, SNAIL1/2, FOX family, GATA4/6 are involved in the process. There is a need to identify the molecular events driving the progression of ductal carcinoma in situ (DCIS); and to derive a signature that differentiates DCIS lesions that have the potential to recur as a subsequent DCIS, an IBC, or to not recur. To catalog the changes associated with EMT that may reveal a clinically relevant signature of progression from DCIS to DCIS or IBC recurrences using a panel of 200 genes related to EMT.

Methodology: RNA was extracted from formalin-fixed paraffin embedded (FFPE) sections of pure primary DCIS lesions representing three categories of outcome: those that did not recur; those that recurred with a subsequent DCIS; and those that recurred with invasive cancer. RNA abundance profiling was performed using Nanostring platform and data processing using an R statistical environment. Levels of mRNA abundance were modelled as a function of recurrence status. Coefficients were fit to terms representing the effect and the standard errors of the coefficient were adjusted with an empirical Bayes moderation. Model-based t-tests were then used to test if the coefficients were significantly different from zero.

Results: Using a technical control sample, pairwise comparisons across three replicates showed high correlation (ρ=0.99, Pρ&lt;2.2x10-16 for all 3 comparisons), suggesting the robustness of the assay. In our preliminary survey of 45 patients across the three groups, we have identified a number of genes that showed differential mRNA abundance levels between patients who recurred (either DCIS or invasive recurrence) vs. those who did not recur. Using Random Forest analysis in a leave-one-out cross-validation approach, we were able to obtain a classifier with a sensitivity of 82% and specificity of 58%. Based on these initial findings, an additional 200 samples have been processed to support these initial findings.

Conclusion: The current literature provided increasing evidence that transcriptomic patterns reflecting the EMT may reveal novel biomarkers and elucidate molecular mechanisms leading to improved prognosis. Among breast carcinomas, differential expression of the EMT genes has been associated with a worse outcome, among estrogen receptor-negative and basal-like carcinomas. However, the understanding of the role of EMT genes in DCIS is limited; therefore, to elucidate whether the EMT plays a role in the progression of DCIS, we have designed an EMT gene panel that also includes genes that are significant prognosticators for IBC, including ER, PgR, Ki67 and HER2. In an exploratory analysis of cases trained based on clinical outcome, the sensitivity for predicting recurrence (whether DCIS or invasive) was 82%.

Citation Format: Felipe Lima J, Yao CQ, Yan F, Dion D, Quintayo MA, Lungu I, Nofech-Mozes S, Pruneri G, Viale G, Boutros PC, Bartlett JMS, Bayani J. The epithelial to mesenchymal transition: Identifying a signature of recurrence in ductal carcinoma in situ. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P1-05-01.

Abstract P2-08-29: Defining a signature of residual risk following endocrine treatment in the tamoxifen and exemestane adjuvant multinational (TEAM) trial

Introduction: There are a number of commercially-available tests to stratify risk for women diagnosed with early breast cancer. While such "Generation I" tests are increasingly being used, a consensus is growing that these tests are moderately accurate in assessing risk. Moreover, Generation I tests fail to direct more personalized treatment. Therefore, there is a clear need for more informative "Generation II" tests that better assess risk, also on the long term, and provide theranostic targets. To this end, we have performed an mRNA abundance-based analysis trained in the 790 patients of the UK TEAM cohort to identify a signature of residual risk , to be validated in the remaining 3000 patients from the TEAM pathology study.

Methods: RNA extracted from the tumors of respective TEAM pathology study patients were profiled using a 165-gene NanoString code set. The gene list was compiled from targets that comprise many of the existing risk assessment tests, in addition to genes known to be of importance for breast cancer pathogenesis. Signal intensities were normalized using the R statistical environment; 336 different combinations of preprocessing methods were assessed and the most optimal method selected using unbiased criteria. A10-fold cross-validation approach, in combination with a network-based patient risk score calculation formula, was used to derive a 95-gene signature. Briefly, genes were first filtered based on a Cox regression p-value threshold of 0.25; the sum of the weighted mRNA abundance levels of the result genes was calculated for each patient as the risk score. Patient-wise risk scores were then used in a multivariate Cox proportional hazards model along with clinical covariates such as age, grade, HER2 status and nodal status, using DRFS truncated to 10 years as an end-point.

Results: Univariate survival analysis revealed a number of significantly prognostic candidates. The resulting 95-gene signature identified in the training set, stratified patients into high and low risk with an HRhigh of 2.74 (p&lt;2.06 x10-4) when adjusted for age, grade, HER2 status and nodal status; resulting in an AUC of 0.73. Modular analyses of the genes comprising the 95-gene signature identified pathways associated with receptor tyrosine kinase signalling, regulation of cell cycle, and the spindle assembly checkpoint. Additionally, the composition of the gene-list made it possible to characterize the patients into their intrinsic subtypes and to determine their relative risk for recurrence relative to assessment tools available today . The validation of the 95-gene signature will be conducted in the remaining samples in the TEAM pathology study using the bioinformatics strategy described above.

Conclusions: The impact of test-guided therapy using multi-parametric tests is increasingly being felt in the clinic, and is reshaping modern health-care economics. A successful Generation II multi-parametric test will better discriminate those that are truly at high risk for recurrence following endocrine therapy and indicate potential therapeutic options for intervention for those who would not benefit from current modalities.

Citation Format: Bayani J, Yao CQ, Quintayo MA, Haider S, Brookes CL, Yan F, van de Velde CJH, Hasenburg A, Kieback DG, Markopoulos C, Dirix L, Seynaeve C, Boutros PC, Rea DW, Bartlett JMS. Defining a signature of residual risk following endocrine treatment in the tamoxifen and exemestane adjuvant multinational (TEAM) trial. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P2-08-29.

Role of Nek2 on centrosome duplication and aneuploidy in breast cancer cells

Breast cancer is the most common solid tumor and the second most common cause of death in women. Despite a large body of literature and progress in breast cancer research, many molecular aspects of this complex disease are still poorly understood, hindering the design of specific and effective therapeutic strategies. To identify the molecules important in breast cancer progression and metastasis, we tested the in vivo effects of inhibiting the functions of various kinases and genes involved in the regulation/modulation of the cytoskeleton by downregulating them in mouse PyMT mammary tumor cells and human breast cancer cell lines. These kinases and cytoskeletal regulators were selected based on their prognostic values for breast cancer patient survival. PyMT tumor cells, in which a selected gene was stably knocked down were injected into the tail veins of mice, and the formation of tumors in the lungs was monitored. One of the several genes found to be important for tumor growth in the lungs was NIMA-related kinases 2 (Nek2), a cell cycle-related protein kinase. Furthermore, Nek2 was also important for tumor growth in the mammary fat pad. In various human breast cancer cell lines, Nek2 knockdown induced aneuploidy and cell cycle arrest that led to cell death. Significantly, the breast cancer cell line most sensitive to Nek2 depletion was of the triple negative breast cancer subtype. Our data indicate that Nek2 has a pivotal role in breast cancer growth at primary and secondary sites, and thus may be an attractive and novel therapeutic target for this disease.

Risk factors for voriconazole hepatotoxicity at 12 weeks in lung transplant recipients

Voriconazole is commonly used for prophylaxis and treatment of invasive aspergillosis in lung transplant recipients. However, the use of voriconazole may at times be limited by the development of hepatotoxicity. Our goal is to determine predictors of voriconazole-associated hepatotoxicity in lung transplant recipients. We conducted a single center retrospective cohort study of lung transplant recipients from 2006 to 2010 who received voriconazole therapy. We compared characteristics of patients who developed hepatotoxicity and those who did not. One hundred five lung transplant recipients received voriconazole. Hepatotoxicity occurred in 51% (54/105) of patients and lead to discontinuation in 34% (36/105). In univariate analysis, age less than 40 years, cystic fibrosis, use of azathioprine, history of liver disease and early initiation of voriconazole were associated with hepatotoxicity. In multivariable logistic regression analysis, perioperative initiation of voriconazole (within 30 days of transplantation) was independently associated with hepatotoxicity (OR 4.37, 95% CI: 1.53-12.43, p = 0.006). The five risk factors identified in the univariate analysis were used to build a K-nearest neighbor algorithm predictive model for hepatotoxicity. This model predicted hepatotoxicity with an accuracy of 70%. Voriconazole therapy initiated within the first 30 days of transplantation is associated with a greater risk of developing hepatotoxicity.

C-Terminal region of teneurin-1 co-localizes with dystroglycan and modulates cytoskeletal organization through an extracellular signal-regulated kinase-dependent stathmin- and filamin A-mediated mechanism in hippocampal cells

The pyramidal neurons in the hippocampus are extremely neuroplastic, and the complexity of dendritic branches can be dynamically altered in response to a variety of stimuli, including learning and stress. Recently, the teneurin family of proteins has emerged as an interneuronal and extracellular matrix signaling system that plays a significant role in brain development and neuronal communication. Encoded on the last exon of the teneurin genes is a new family of bioactive peptides termed the teneurin C-terminal-associated peptides (TCAPs). Previous studies indicate that TCAP-1 regulates axon fasciculation and dendritic morphology in the hippocampus. This study was aimed at understanding the molecular mechanisms by which TCAP-1 regulates these changes in the mouse hippocampus. Fluoresceinisothiocyanate (FITC)-labeled TCAP-1 binds to the pyramidal neurons of the CA2 and CA3, and dentate gyrus in the hippocampus of the mouse brain. Moreover, FITC-TCAP-1 co-localizes with β-dystroglycan upon binding to the plasma membrane of cultured immortalized mouse E14 hippocampal cells. In culture, TCAP-1 stimulates ERK1/2-dependent phosphorylation of the cytoskeletal regulatory proteins, stathmin at serine-25 and filamin A at serine-2152. In addition, TCAP-1 induces actin polymerization, increases immunoreactivity of tubulin-based cytoskeletal elements and causes a corresponding increase in filopodia formation and mean filopodia length in cultured hippocampal cells. We postulate that the TCAP-1 region of teneurin-1 has a direct action on the cytoskeletal reorganization that precedes neurite and process development in hippocampal neurons. Our data provides novel evidence that functionally links the teneurin and dystroglycan systems and provides new insight into the molecular mechanisms by which TCAP-1 regulates cytoskeletal dynamics in hippocampal neurons. The TCAP-dystroglycan system may represent a novel mechanism associated with the regulation of hippocampal-function.