Search

Purpose Conventional staging methods are inadequate to identify patients with stage II colon cancer (CC) who are at high risk of recurrence after surgery with curative intent. ColDx is a gene expression, microarray-based assay shown to be independently prognostic for recurrence-free interval (RFI) and overall survival in CC. The objective of this study was to further validate ColDx using formalin-fixed, paraffin-embedded specimens collected as part of the Alliance phase III trial, C9581. Patients and Methods C9581 evaluated edrecolomab versus observation in patients with stage II CC and reported no survival benefit. Under an initial case-cohort sampling design, a randomly selected subcohort (RS) comprised 514 patients from 901 eligible patients with available tissue. Forty-nine additional patients with recurrence events were included in the analysis. Final analysis comprised 393 patients: 360 RS (58 events) and 33 non-RS events. Risk status was determined for each patient by ColDx. The Self-Prentice method was used to test the association between the resulting ColDx risk score and RFI adjusting for standard prognostic variables. Results Fifty-five percent of patients (216 of 393) were classified as high risk. After adjustment for prognostic variables that included mismatch repair (MMR) deficiency, ColDx high-risk patients exhibited significantly worse RFI (multivariable hazard ratio, 2.13; 95% CI, 1.3 to 3.5; P < .01). Age and MMR status were marginally significant. RFI at 5 years for patients classified as high risk was 82% (95% CI, 79% to 85%), compared with 91% (95% CI, 89% to 93%) for patients classified as low risk. Conclusion ColDx is associated with RFI in the C9581 subsample in the presence of other prognostic factors, including MMR deficiency. ColDx could be incorporated with the traditional clinical markers of risk to refine patient prognosis.

1074 Background: In the Triple Negative Trial we observed no improved response rate (RR) to C over D in aTNBC [Tutt et al, Nat Med 2018], but we did in BRCA1/2 mutated (mut) patients (pts). We hypothesise tumors with other aberrant DNA damage response (DDR) characteristics having higher RR to DNA damage inducing C than D. Methods: We tested the predictive value of DDR process related gene expression signatures (PARPi7, chromosomal instability CIN70, TP53 & DDR Deficiency (DDRD)) on 192 treatment naïve primary tumours (PT) by total RNA-sequencing. Odds ratio (OR) for RR are reported. Paired PT & recurrent (REC) signature scores were compared. Results: Unexpectedly, high DDRD and PARPi7 were associated with higher RR to D than C ( p =0.01 & 0.06). No effect was observed for CIN70 or TP53 signature. To assess whether the unexpected results were due to biological changes 12 PT-REC pairs were available from pts who received chemotherapy (CT) between PT & REC. CIN70 increased from PT to REC, DDRD (non-significantly) & PARPi7 decreased. 4/5 TP53 wildtype classified PT samples classified as mut in REC. The BRCA1/2 & DDRD-treatment interactions only held in pts who received CT before trial entry (table). The PARPi7-treatment interaction only held in CT naïve pts. In CT naïve pts, high CIN70 tumors suggested higher C RR as hypothesized. Restricted to the 149 PAM50 basal-like pts, results were non-significant but similar trends seen. Conclusions: In this trial of aTNBC, DDRD high pts with prior CT had better RR to D than C. A possible explanation for this unexpected result is selective pressure of adjuvant DNA damaging CT and selection for relative taxane sensitivity in those who recur despite a high DDRD score. The hypothesised CIN70 treatment interaction was observed in CT naïve pts. Our results suggest care is required in application of signatures to initial diagnostic material when predicting response to DNA damaging agents at REC particularly in pts with prior CT. [Table: see text]

439 Background: The benefit of checkpoint blockade in advanced gastric cancer is limited and biomarkers related to response are needed. Novel gene expression analysis software was used to identify Hallmarks of Cancer associated with clinical benefit following nivolumab treatment in >2nd line advanced gastroesophageal adenocarcinoma (GEA). Methods: RNA-sequencing data from baseline GEA patient diagnostic tumor samples (103 from NCT02862535; 5 from NCT02862535) were analyzed using the claraT platform (V2.0.0, Almac Diagnostic Services). 62 gene signatures were quantified representing 6 key Hallmarks of Cancer (Avoiding Immune Destruction, Activating Invasion and Metastases, Sustaining Proliferative Signaling, Inducing Angiogenesis, Resisting Cell Death and Genome Instability and Mutation). Clinical benefit (CB) was defined as tumor response or overall survival (OS) > 1 year. HER2 status was from medical records. Survival analyses used cox proportional hazards models. Results: Gene expression signatures (GES) identified 5 molecular subgroups (C1-C5). Rate of CB in each molecular subtype are outlined in Table. C3 and C4 had significantly improved OS compared to C2, (HR = 0.45; p= 0.02 and HR = 0.42; p= 0.02). Greater proportions of HER2+ subjects were present in C4 and C3 vs. C2, with C3 statistically significant (60% vs. 14%; p= 0.012). Gene expression characterized by chromosomal instability (CIN) and homologous recombination repair deficiency (HRD) were associated with HER2(+) (wilcox p= < 0.05). Patients selected by only using CIN & HRD had significant improvement in OS (HR = 0.63; p= 0.03). Conclusions: Interferon-based GES did not predict benefit from immune checkpoint blockade. GES representing HRD and activation of HER2, EGFR and MAPK (each enriched in CIN) were associated with improved survival upon checkpoint blockade in advanced GEA patients. Clinical trial information: NCT02862535 . [Table: see text]

e21045 Background: Traditionally gene expression signatures (GES) are used individually to classify patients into subgroups. Signatures targeting the same biology are often developed independently and may not classify identically. We developed the claraT software tool that uses consensus between multiple published GES categorised by the Hallmarks of Cancer (Hanahan & Weinberg, 2011) to classify cancers. As metastatic melanoma represents poor prognostic disease (5-yr survival 15-20%), we applied claraT to the TCGA melanoma dataset to identify targetable biologies, validated in a cohort of melanoma patients treated with Ipilimumab. Methods: TCGA RNA-seq data ( n= 472) was analysed using the claraT platform including GES for immune ( n= 14), angiogenesis ( n= 9) and epithelial-mesenchymal transition (EMT) ( n= 12) Hallmarks. Samples were clustered for the combined and individual Hallmarks. Median progression-free (PFS) and overall-survival (OS) differences were analysed across identified subgroups. Analysis was validated in an Ipilimumab treated melanoma dataset ( n= 42) (Van Allen, 2015). Results: Clustering the combined Hallmarks identified 4 subgroups in the TCGA cohort: 1) Immune active, 2) Immune-EMT active, 3) EMT-Angiogenesis active, 4) All inactive. Groups 1&2 had significantly improved OS compared to Groups 3&4 (HR = 0.50, p< 0.0001). Clustering using single Hallmarks revealed that immune-positive tumours had significantly improved OS (HR = 0.53, p< 0.0001) compared to immune-negative tumours. Angiogenesis-negative tumours displayed improved PFS (HR = 0.73, p= 0.03) and OS (HR = 0.53, p

e15535 Background: The incidence of Esophageal Adenocarcinoma (EAC) has risen in the western world but response rates to chemotherapy are low and survival is poor. Increased molecular understanding is needed to develop novel treatments. Methods: Transcriptional profiling of 274 treatment naïve EAC biopsies was performed using the Almac Diagnostics Xcel array™. All patients received platinum-based neo-adjuvant chemotherapy prior to surgical resection at four United Kingdom centers between 2004-2012. Semi-supervised clustering was performed followed by functional enrichment using DAVID. Cluster membership was assessed for independence of prognostic factors using Cox proportional hazards. Candidate targets were identified by siRNA screening in OE33 cells. Treatment with the survivin inhibitor, YM155 in EAC cell lines was also assessed. Results: Semi-supervised hierarchical clustering identified two groups with significant differences in RFS [HR = 0.54 (0.29-0.99), p = 0.05] and OS [HR = 0.52 (0.28-0.96), p = 0.04]. There were significant associations between the clusters and both nodal and TNM downstaging but not with pathological response. The PI3K-AKT, p53, tight junction and HIF-1 signaling pathways were upregulated in the poor prognostic group. Eighty-four genes were selected and taken forward in a functional genomic siRNA screen. Twenty-seven genes showed a significant reduction in viability following siRNA-mediated knockdown and further verification resulted in twelve candidate genes. Finally, target knockdown in seven EAC cell lines resulted in four interrelated hits- BIRC5, JAK1, OSMR and SLC2A1. Knock down of BIRC5 (Survivin) induced apoptosis, as evidenced by PARP cleavage, in both the parental OE33 and cisplatin-resistant OE33CDDPR cell lines. YM155, a survivin inhibitor, is shown to induce apoptosis at nanomolar concentrations across a panel of parental and cisplatin-resistant EAC cell lines and further mechanistic work is ongoing. Conclusions: We have performed clustering of a large transcriptomic dataset and defined a poor prognostic group of EAC patients. We identified Survivin (BIRC5) as a mediator of cisplatin resistance in EAC and a potential novel drug target. Further pre-clinical and clinical work to assess the benefit of survivin inhibition in EAC is warranted.

e15142 Background: Ulcerative colitis (UC), is a known risk-factor for colorectal cancer (CRC). 10% of UC-patients develop UC-specific CRC yet there are currently no biomarkers for early detection or understanding of drivers of disease. Unresolved inflammation is key in UC-CRC development, yet the aetiology is unclear. It is unknown if UC-CRC falls into a specific consensus molecular subgroup in the CMS classification (Guinney et al 2015). This bioinformatics approach aimed to identify patients at-risk of developing UC-CRC, whilst informing biological and clinical understanding of the disease. Methods: We used publicly available UC data and in-silico, cross platform, class label transfer (GECA) to determine mRNA similarity between UC/UC-CRC and sporadic CRC (sCRC) and to identify a UC-predominant CMS. Pathogenic mutations were identified for each CMS as well as gene set enrichment and survival analysis to determine UC-CRC behaviour comparative to sCRC. Transcriptional relationships were assessed via unsupervised and semi-supervised clustering by immune-signatures and scores to determine molecular patterns between phenotypes. We derived a gene list from UC data informed by clustering to stratify patients based on risk to UC-CRC progression. Results: UC-CRC featured predominantly in the NA group, with 3 attributed pathogenic mutations, an oxidative stress environment and neuroendocrine dysregulation. The subgroup had UC-like clinical features; younger age, distal tumour location and TP53 mutations along with the highest rate of cancer recurrence (~20%) and a unique cellular composition in CMS being highly immune and stromal. Importantly this immune signal was distinct from the interferon type signalling observed in CMS1 and was not associated with PD-L1 activation. Clustering revealed a loss of IFN-type gene expression in colon tissue as UC progressed to UC-CRC. Conclusions: The NA group was thought to be unclassifiable samples between CMS however, analysis shows UC-CRC enrichment with defined biological characteristics. We have identified novel biology that may inform the risk of developing UC-CRC and guide potential preventative and therapeutic approaches.

Purpose Current prognostic factors are poor at identifying patients at risk of disease recurrence after surgery for stage II colon cancer. Here we describe a DNA microarray–based prognostic assay using clinically relevant formalin-fixed paraffin-embedded (FFPE) samples. Patients and Methods A gene signature was developed from a balanced set of 73 patients with recurrent disease (high risk) and 142 patients with no recurrence (low risk) within 5 years of surgery. Results The 634–probe set signature identified high-risk patients with a hazard ratio (HR) of 2.62 (P < .001) during cross validation of the training set. In an independent validation set of 144 samples, the signature identified high-risk patients with an HR of 2.53 (P < .001) for recurrence and an HR of 2.21 (P = .0084) for cancer-related death. Additionally, the signature was shown to perform independently from known prognostic factors (P < .001). Conclusion This gene signature represents a novel prognostic biomarker for patients with stage II colon cancer that can be applied to FFPE tumor samples.

5578Background: Patients with High Grade Serous Ovarian Cancer (HGSOC) initially respond to SOC platinum based treatment but most will eventually relapse with platinum resistant disease. Angiogenes...

12111Background: Unsupervised hierarchical clustering of gene expression data from 265 high grade serous ovarian cancer (HGSOC) patients identified 3 major molecular subgroups. One subgroup is driv...

103 Background: Recent studies have demonstrated limited success of immune checkpoint therapies in unselected prostate cancer. We therefore assessed an immune based DNA Damage Repair Deficiency (DDRD) assay, that we have previously reported represents activation of the cGAS STING pathway, in the TCGA prostate cancer dataset to investigate the presence of targetable immune biology in prostate cancer. In addition we applied a second assay (the prostate cancer metastatic signature-PCM) that predicts the risk of metastatic recurrence for early prostate cancer, in order to assess if immune therapy could have a role in treating high risk disease. Methods: 498 samples with RNA sequencing data were scored with the PCM and DDRD assays. Integrative analysis was performed on 488 samples with RNA sequencing, promoter site methylation, somatic mutation and somatic copy number variation. Gene expression of n = 6 immune checkpoint targets was investigated with the subgroups identified using T-tests. The prevalence of immune infiltration in each subgroup was tested by applying a cut off to the leukocyte fraction. Cox proportional hazards regression analysis of 441 patients was assessed for biochemical recurrence. Results: Integrative analysis identified four patient subgroups characterised primarily by variances in copy number and genomic mutation. One of these subgroups ‘Metastatic-like DDRD’ had significantly higher PCM scores and DDRD immune scores compared to the other subgroups (p < 2E-12). This subgroup of patients showed elevated leukocyte fraction and expression of immune checkpoint genes: CD274 (PDL1), CTLA4, ICOS, IDO1, HAVCR2 (TIM3) & LAG3 (p < 2E-6). Genomic instability with amplification of 8q and a larger prevalence of somatic mutations including that of TP53 was also detected in this subgroup. These patients had an increased risk of biochemical relapse in both univariate (p < 2E-5) and multivariate (p < 0.008) analysis. Conclusions: We identified a poor prognostic subgroup, representing 17% of early prostate cancer patients that are at increased risk of developing metastatic disease and present with targetable immune biology. These patients may represent a viable target population for immune checkpoint and DNA damaging therapies in prostate cancer.

Human cancers exhibit genomic instability and an increased mutation rate due to underlying defects in DNA repair. Cancer cells are often defective in one of six major DNA repair pathways, namely: mismatch repair, base excision repair, nucleotide excision repair, homologous recombination, nonhomologous endjoining and translesion synthesis. The specific DNA repair pathway affected is predictive of the kinds of mutations, the tumor drug sensitivity, and the treatment outcome. The study of rare inherited DNA repair disorders, such as Fanconi anemia, has yielded new insights to drug sensitivity and treatment of sporadic cancers, such as breast or ovarian epithelial tumors, in the general population. The Fanconi anemia pathway is an example of how DNA repair pathways can be deregulated in cancer cells and how biomarkers of the integrity of these pathways could be useful as a guide to cancer management and may be used in the development of novel therapeutic agents.

529 Background: Biomarkers of response and resistance to adjuvant chemotherapy for TNBC are needed. Deficiency in DNA damage response (DDR) and repair pathways have been reported in TNBC and may impact response to chemotherapy. Aims: To investigate DNA damage response deficiency (DDRD) molecular signature, BRCA1mRNA expression and Tumor Infiltrating Lymphocytes (TILs) as prognostic markers in TNBC patients treated with adjuvant AC on S9313. Methods: S9313 accrued 3125 early stage BC patients to two alternative schedules of AC with no difference in outcomes between the two arms. We identified 425 (14%) patients with centrally determined TNBC with tissue availability. DDRD signature (44 gene signature, Almac Inc.) and BRCA1expression (NanoString nCounter) were performed on RNA isolated from pre-treatment FFPE tumor tissue. DDRD score was classified in quartiles. TILs evaluation was performed using previously described criteria. Markers were tested for prognostic effect on DFS and OS using Cox regression model with adjustment for randomized treatment assignment. Results: For 425 TNBC patients median age: 45 yrs, and 5 year DFS and OS = 74% and 82%, respectively. DDRD signature was successfully evaluated in 89.6% (381/425) but only 267 (62.8%) met 60% tumor content criterion for inclusion. DDRD score quartiles were associated with DFS (5 year DFS 59% & 82% in the lowest & highest quartiles respectively, p = 0.0005) and OS (5 year OS 74% and 86% in lowest and highest quartiles respectively, p = 0.008). BRCA1 expression and TILs were successfully determined in 78% and 99% samples, respectively. BRCA1expression was not associated with DFS. TILs were associated with DFS (10% increase HR = 0.88; 95% CI 0.79-0.97; p = 0.016) and OS (HR = 0.84; 95% CI 0.74-0.94; p = 0.0005). DDRD score and TILs were highly correlated (Pearson = 0.62). In multivariate model of DFS including TILs and DDRD quartiles, only DDRD remains significant (p = 0.018). Conclusions: DDRD signature was prognostic in TNBC patients treated with AC chemotherapy and has the potential to be used as a selection criterion to identify TNBC patients whose prognosis is sufficiently poor to justify evaluation of alternative treatment.

4026 Background: Current strategies to guide the selection of neo-adjuvant or adjuvant therapy in esophageal and gastric adenocarcinomas (EAC/GAC) are inadequate. We assessed a clinically validated 44 gene DNA Damage Response Deficiency (DDRD) assay to predict prognosis following neo-adjuvant DNA damaging chemotherapy (CT) in EAC and adjuvant CT or chemo-radiotherapy (CRT) in GAC. Methods: Transcriptional profiling of 273 formalin fixed paraffin embedded pre-treatment endoscopic EAC biopsies was performed using the Almac Diagnostics Xcel array. All EAC patients were treated with cisplatin-based neo-adjuvant chemotherapy followed by surgical resection between 2003 and 2014 at four UK centers in the OCCAMS consortium. Further validation was performed using a publically available dataset of 270 resected gastric cancers treated with adjuvant platinum-based CT, CRT or surgery alone at the Samsung Medical Centre, Seoul, Korea. The association between the DDRD score and prognosis was assessed by Kaplan-Meier analysis and Cox Proportional Hazards regression. Results: A total of 66 EAC samples (24%) were characterized as DDRD positive with the remaining 207 samples (76%) being DDRD negative. DDRD assay positivity was associated with improved DFS (HR 0.58; 95% CI 0.36-0.93; p = 0.024) and OS (HR 0.56; 95% CI 0.34-0.92; p = 0.023) following multivariate analysis. DDRD positive patients had a higher pathological response rate (p = 0.033) and a higher rate of loco-regional versus distant relapse (30% vs 20%; p = 0.013). For GAC, 132 samples (49%) were characterized as DDRD positive with the remaining 138 (51%) being DDRD negative. DDRD positivity was associated with improved DFS (HR 0.48; 95% CI 0.25-0.96; p = 0.037) following D2 gastrectomy and adjuvant CT or CRT. DDRD status was not associated with DFS in the surgery alone cohort (HR 0.87; 95% CI 0.55-1.38; p = 0.562). Conclusions: The DDRD assay is strongly predictive of benefit from DNA damaging neo-adjuvant CT and esophagectomy in EAC and gastrectomy and CT/CRT in GAC and can be applied to routine diagnostic material.

11616 Background: Epithelial–mesenchymal transition (EMT) is the conversion of epithelial cells to mesenchymal cells and involves loss of cell–cell adhesion and cell polarity and increased motility, invasiveness and metastases. The EMT process has therefore been under investigation as a new target for anticancer drug discovery. The aim of this work was to develop an EMT biomarker suitable for formalin-fixed paraffin embedded (FFPE) tissue that could be used for patient treatment selection. Methods: Unsupervised hierarchical clustering of ovarian cancer gene expression data (TCGA. 2011) previously identified an EMT subgroup. We confirmed this EMT subgroup in FFPE tissue using 265 high grade serous ovarian cancer (HGSOC) FFPE samples. The analysis was extended to show existence of an EMT subgroup across a range of solid tumours including colon, lung, melanoma and prostate cancer. Further to this, a common gene list was generated to include only transcripts with high variability and expression across diseases, and used as a starting list for development of a 15 transcript assay which can be used to prospectively identify the EMT subgroup from archived tissue. Results: The 15 gene expression assay was a poor prognostic marker in Colorectal, (Relapse free survival: HR = 1.46 [95% CI:1.07-1.98]); Lung, (Relapse free survival: HR = 2.18 [95% CI:1.33-3.56]); Prostate cancer, (Biochemical recurrence: HR 2.49 CI: 1.43-4.34) and was associated with activated MAPK (phospho-MAPK) in preclinical models and clinical samples (p < 0.05). The assay score was reduced by MEK inhibition (p < 0.05) and elevated by KRAS, NRAS and MEK1 overexpression (p < 0.05). The assay predicted response to the MEK inhibitors Trametinib and Selumetinib in cell line models (p < 0.001). Conclusions: A 15 gene expression assay has been developed from FFPE samples across multiple diseases to detect an EMT molecular subgroup associated with MAPK signalling. The assay predicted sensitivity to MEK inhibitors in pre-clinical model systems. Further work aims to validate the assay as a predictive biomarker in clinical samples from patients treated with EMT or MEK targeted therapies.

4023 Background: Programmed Death-1 Receptor (PD-1) and its ligand (PD-L1) downregulate T cell activation and suppress tumor killing. This study investigated the role of PD-L1 and tumor infiltrating lymphocytes (TILs) in response to neo-adjuvant therapy and prognosis in esophageal adenocarcinoma (EAC). Methods: Transcriptional profiling of 273 formalin fixed paraffin embedded pre-treatment endoscopic EAC biopsies was carried out using the Almac Diagnostics Xcel array and the expression levels of PD-L1 probesets corresponding to protein encoding extracted. Response was assessed by tumor regression grade (TRG; score ≤ 2 = response). Immunohistochemistry (IHC) for PD-L1 and CD8 was performed in matched resection specimens from 135 patients. All EAC patients were treated with cisplatin-based neo-adjuvant chemotherapy followed by surgical resection between 2003 and 2014 at four UK centers as part of the OCCAMS consortium. Associations between expression, protein levels and TRG were assessed by Kruskal-Wallis, Mann-Whitney Unpaired, Spearman rank correlation or chi-squared tests. Survival analysis was performed using Cox Proportional Hazards regression. Results: High PD-L1 gene expression in the pre-chemotherapy biopsies was associated with pathological response (TRG ≤ 2; p = 0.02) following neo-adjuvant chemotherapy. PD-L1 ( > 5%) was expressed in the tumor or stromal cells in 4% and 15% of resection specimens respectively. PD-L1 gene and IHC expression ( > 5%) were closely associated between the biopsies and both the tumor (p = 0.032) and stroma (p = 0.019) of the matched resection specimens. Patients with PD-L1 IHC positivity in tumor cells demonstrated improved relapse-free survival (HR 0.314; 95% CI 0.099-0.997; p = 0.049) and positive stromal PD-L1 IHC staining correlated with pathological response (p = 0.05). Biopsy gene expression of PD-L1 and CD8 was closely associated (p = 0.024) and the presence of CD8+ TILs in the microenvironment strongly correlated with tumor (p < 0.001) and stromal (p < 0.001) PD-L1 positivity. Conclusions: High PD-L1 expression in the pre-treatment biopsies in EAC is predictive of response to neo-adjuvant chemotherapy and may aid selection of conventional and immune-targeted agents.

4030 Background: Prognostication for cancer patients is based upon factors determined at baseline and becomes less relevant over time. Conditional survival (CS) estimates future prognosis based upon survival to a specific time point after treatment. We analyzed CS for patients in the United Kingdom (UK) undergoing surgery and neoadjuvant chemotherapy (NAC) for gastro-esophageal junction (GEJ) or esophageal adenocarcinoma (EAC). Methods: 1409 patients with GEJ/EAC treated with NAC and surgical resection at 7 centers across the UK from 2002-2014 were identified. Clinicopathological and survival data was collected as part of the Oesophageal Cancer Clinical and Molecular Stratification (OCCAMS) consortium. A multivariable Cox survival model was used to analyze the association of factors such as node positivity (N+), lymphovascular invasion (LVI+), tumor differentiation, circumferential resection margin involvement (CRM+) and pathological response by tumor regression grade (TRG ≤2) with risk of relapse (RR) or death from time of surgery. Results: Of 1409 patients, 726 (51.5%) were aged

11 Background: Approximately 20% of patients with organ confined prostate cancer (PCa) will develop disease recurrence following radical treatment (surgery or external beam radiotherapy (EBRT)). We hypothesized that a molecular subgroup of early PCa may have metastatic potential at presentation, resulting in disease recurrence. Methods: Using unsupervised hierarchical clustering of gene expression from a PCa dataset we identified a novel molecular subgroup with a transcriptional profile similar to metastatic disease. We developed a 70 gene expression assay to prospectively identify patients within the subgroup from formalin fixed and paraffin embedded tissue (FFPE). Initial assessment found the assay to be prognostic in three independent publicly available prostatectomy datasets. We therefore assessed the prognostic value of the assay in FFPE clinical samples collected from multiple international sites. FFPE tumor resections and tumor biopsy specimens were obtained from 322 surgical patients and 248 patients treated with EBRT. Regions of highest Gleason grade were identified for macrodissection, RNA extraction and gene expression analysis. Samples were dichotomized as metastatic biology assay positive or negative using a pre-specified cut-off. The association of assay results with biochemical failure (BF) and distant metastases (DM) was tested on multivariate (MVA). Results: The assay was significantly associated with BF on MVA (HR 1.67 [1.16-2.38]; p = 0.0059), (HR 2.26 [1.26-4.04]; p = 0.0062) and DM on MVA (HR 3.39 [1.88-6.12]; p = 0.0001), (HR 3.26 [1.27-8.30]; p = 0.0137) for surgery and EBRT cohorts respectively. Importantly, in a combined model, the assay demonstrated additional information to the commonly used CAPRA clinical tool for prediction of DM, HR 2.72 [2.10-3.51]; p < 0.0001, and HR 2.72 [1.42-5.20]; p = 0.0026 (Prostate Metastatic Assay combined with CAPRA-S and CAPRA respectively). Conclusions: The metastatic biology assay predicts BF and DM in PCa patients treated with surgery or EBRT. The assay may help to select patients at risk of metastatic disease for additional treatment aimed at preventing disease recurrence.

3035Background: We previously described a molecular subtype of tumors that presents an innate immune response to intrinsic DNA damage. The associated gene expression pattern is mediated by the STIN...

5502 Background: HGSOC is a histopathological diagnosis and may represent multiple diseases at a molecular level. We investigated whether distinct molecular subgroups may influence treatment choice...

4015 Background: The incidence of esophageal adenocarcinoma (EAC) has increased 6-fold in the last 30 years and in early stage EAC 5-year survival remains poor at 25-35%. Neoadjuvant therapy confer...

63 Background: Accurate prediction of outcome for neo-adjuvant therapy in early stage esophageal adenocarcinoma (EAC) is essential for clinical trial design, patient counseling and treatment decisions. The use of 18F-fluorodeoxyglucose positron emission tomography (PET) has improved the staging of EAC and can predict pathological response and survival. Our aim was to develop a pre-operative prognostic model based on metabolic imaging and clinical parameters. Methods: Between April 2004 and December 2010, 118 patients with early stage adenocarcinoma of the distal esophagus or gastro-esophageal junction were treated with platinum and fluorouracil-based combination chemotherapy prior to surgical resection. Patients were staged by endoscopic ultrasound and PET/CT was performed prior to, and upon completion of, chemotherapy. A metabolic response was defined as a >35% reduction in the maximum standard uptake variable (SUVmax). The final prognostic model was selected by multivariate Cox regression analyses and the concordance-index (c-index) was used to assess the fit of the final multivariate model. Results: In both the univariate and multivariate analysis of patient prognosis post-surgery, five pre-surgery factors were included: clinical T and N staging, PET response, neutrophil-lymphocyte ratio and albumin levels. The majority of the patients were T3 (82.1%) and N1 (73.7%) and 69.8% of patients demonstrated a PET response. In the univariate analysis PET response was the only significant prognostic factor with a median overall survival (OS) of 60.4 months for responders compared to 23.3 months for non-responders (p=0.0241). The final multivariate model of T staging and PET response had a c-index of 0.624 with T3/PET non-response resulting in a poorer prognosis compared to T3/PET response and T2/PET responder or non-responder (median OS 11.5 vs 44.5 vs undefined months; p=0.0284). We seek to validate our model in an independent patient dataset to assess its predictive capability. Conclusions: Metabolic imaging and clinical staging can be combined in a prognostic model in early stage EAC. Our model can inform patient counseling and the development of clinical trial strategies.

455 Background: Only 15-25% of pts with stage II colon cancer (CC) experience recurrence and conventional staging methods neither allow accurate identification of low (L) and high-risk (H) subgroups nor predict benefit of adjuvant chemotherapy. The ColDx assay (Almac Diagnostics) is a 634-probeset gene expression signature shown to be independently prognostic for recurrence-free interval (RFI). The objective of this study was to assess the ability of ColDx to classify stage II CC pts at L- and H-risk of relapse. Methods: This validation study was conducted using formalin fixed paraffin embedded biospecimens and clinical data from CALGB 9581, a phase III trial of edrecolomab v. observation in pts with normal risk, stage II CC. 1,454 CALGB 9,581 pts met eligibility criteria. A case-cohort sampling design was used to randomly select (RS) 514 pts from 901 eligible pts with available tissue; supplemented by 49 non-RS recurrent pts (total 563). Risk status for each pt was based on a positive or negative ColDx score using a pre-specified cutpoint, 0.4377. The Self Prentice method was used to test the association between ColDx categories and RFI (distant recurrence or death due to primary disease). Results: Initial results in 563 pts were erroneous due to a quality failure in a batch of reagent. 524 samples were re-labeled, re-ordered, and re-assayed using reagents that passed quality control (36 samples had insufficient material; 95 failed ColDx QC). Final analysis comprised 393 pts, 360 RS (58 events; 16%); 33 non-RS events. 216 pts (55%) were predicted H (62 events); 177 (45%) pts were predicted L (29 events). H pts exhibited significantly worse RFI (univariable hazard ratio (HR), 2.0; 95% CI, 1.3-3.3; p < 0.01). ColDx remained significant after adjustment for prognostic factors; HR, 2.1 (95% CI, 1.3-3.4; p < 0.01). Conclusions: The ColDx assay result is associated with RFI in the CALGB 9,581 sub-sample and is independent from other prognostic factors, including MSI. Further investigation is needed to establish the role of this classifier in guiding treatment decisions in this patient population.

45 Background: Patients with operable esophageal adenocarcinoma have a poor prognosis (median survival

TPS11120 Background: Currently there is no biomarker to predict specific benefit from DNA-damaging anthracycline and cyclophosphamide-based chemotherapy in the clinic. Loss of the Fanconi anemia/BRCA (FA/BRCA) DNA-damage response pathway occurs in approximately 25% of breast cancer and results in sensitivity to DNA-damaging agents. We therefore developed an assay to detect loss of the FA/BRCA pathway, for the purpose of predicting benefit from chemotherapy. Methods: 21 FA patient samples were analyzed to identify genetic processes associated with loss of the FA/BRCA pathway. Unsupervised hierarchical clustering was then performed using 60 BRCA1/2 mutant and 47 sporadic tumor samples and a molecular subgroup was identified that was defined by the molecular processes representing loss of the FA/BRCA pathway. A 44-gene DNA Damage response deficient (DDRD) assay was developed that could identify this subgroup from formalin fixed, paraffin embedded (FFPE) samples in the clinic. Results: In a publicly available independent cohort of 204 patients, the assay predicted response to neoadjuvant DNA-damaging chemotherapy (5-fluorouracil, anthracycline and cyclophosphamide) with an odds ratio of 4.01, (95% Cl:1.69-9.54). We also analysed samples from an independent cohort of 114 node-negative breast cancer patients treated with adjuvant 5-fluorouracil, epirubicin and cyclophosphamide treatment at the Northern Ireland Cancer Centre. The DDRD assay significantly predicted 5-year relapse free survival with a hazard ratio of 0.27 (95% Cl:0.10-0.83). The assay was not predictive of survival in patients who did not receive chemotherapy. Conclusions: An FFPE tissue-based assay that detects loss of the FA/BRCA pathway has been developed and independently validated as a predictor of response and prognosis following DNA damaging anthracycline/cyclophosphamide-based chemotherapy in the neoadjuvant and adjuvant settings.

10553 Background: Loss of PTEN function has been widely reported to cause up-regulation of the PI3K/AKT signalling pathway resulting in increased cell growth, proliferation and survival. More recently it has been reported that PTEN null cells demonstrate genomic instability and have increased production of reactive oxygen species (ROS) and oxidative stress induced DNA damage. Ataxia Telangiectasia Mutated (ATM) is the primary response kinase, which responds to stalled DNA replication and DNA double strand breaks due to oxidative DNA damage. Methods: A metagene representing ATM activation was generated from cell line data and used to perform hierarchical clustering analysis of public DNA microarray profiling datasets of breast cancer, ovarian cancer and glioblastoma with known PTEN IHC/mutation status. Furthermore, we ask if ATM activation may be therapeutically exploited in PTEN null tumours using ATM specific siRNA and compounds in 2 PTEN isogenic cell line model systems. Results: We show that PTEN null cells have elevated levels of ROS, DNA damage and have endogenous activation of ATM, an enzyme important in responding to DNA damage resulting from oxidative stress. We hypothesised that PTEN deficient tumours may rely on ATM enzyme for survival. To investigate this we generated a 189-gene list representing ATM activation and used this to perform hierarchical clustering analysis of a breast cancer DNA microarray dataset. This list was able to significantly cluster tumours with known loss of PTEN expression (p=0.004). Furthermore, this gene list was able to segregate PTEN null/mutant tumours from PTEN wild-type tumours in 2 independent datasets of glioblastoma and ovarian cancer (p=0.015 and p=0.012). In addition, we found that inhibition of ATM using the selective inhibitor KU-55933 caused DNA damage, cell cycle arrest and apoptosis specifically in PTEN deficient cells when compared to PTEN wild-type cells. Conclusions: These observations suggest that ATM may represent a therapeutic target in PTEN deficient tumours and furthermore ATM activation may be the basis of a biomarker of PTEN status in human cancers.

10511 Background: Loss of a functional DNA-damage response (DDR) sensitizes tumors to DNA-damaging as well as targeted therapeutics such as PARP-1 inhibitors. However, there is no assay to detect D...

5000 Background: Despite histological heterogeneity and evidence of some molecular subtypes (e.g. BRCA1/2-deficient) EOC continues to be treated as a single disease. Using unsupervised hierarchical...

e21024 Background: There is considerable interest in the use of gene expression data to generate prognostic/predictive biomarkers from NSCLC tissue. Since individual tumors exhibit histological div...

4041 Background: The discovery and widespread acceptance of molecular subtypes in breast cancer has changed the way that this disease is viewed. Similarly, we have focussed on stage II colorectal cancer where 20–25% of patients are at risk of recurrent disease following surgery. There are currently several types of colorectal cancer recognised by histopathological methods, however, little has been published regarding disease classification at a molecular level. Methods: We have taken an unbiased approach to the molecular classification of colorectal cancer. Using a colorectal cancer disease specific microarray platform we have performed expression profiling and bioinformatic analysis of 500 formalin fixed and paraffin embedded (FFPE) stage II colorectal cancer samples, in order to identify and characterise molecular subtypes. Results: We have identified molecular subtypes for colorectal cancer. One of these subtypes is enriched with tumours with microsatellite instability (MSI) that are predominantly proximal, mucinous and poorly differentiated. It is notable that while MSI is generally accepted to be associated with a good prognosis, we have found both good and poor prognosis patients within this subtype. A second subtype contains tumors with a low CpG island methylator phenotype. Another subtype is defined by an expression pattern consistent with activation of tumor suppressors. The remaining subtypes are newly identified. Conclusions: We have identified molecular subtypes in colorectal cancer. We will present how knowledge about these subtypes may affect patient management and may be used to guide the development of novel therapeutic strategies. [Table: see text]

22086 Background: Non-small cell lung cancer (NSCLC) is the leading cause of cancer mortality worldwide with low response rates to standard chemotherapy treatment. Identification of reliable genomi...

10509 Loss of the fanconi anemia (FA) pathway function has been described in a number of sporadic tumor types including breast, ovarian, pancreatic, head and neck and hematological malignancies. Functionally, the FA pathway responds to stalled DNA replication following DNA damage. Given the importance of the FA pathway in the response to DNA damage, we hypothesized that cells deficient in this pathway may become hyper-dependent on alternative DNA damage response pathways in order to respond to endogenous genotoxic stress such as occurs during metabolism. Therefore, targeting these alternative pathways could offer therapeutic strategies in FA pathway deficient tumors. To identify new therapeutic targets we treated FA pathway competent and deficient cells with a DNA damage response siRNA library, that individually knocked out 230 genes. We identified a number of gene targets that were specifically toxic to FA pathway deficient cells, amongst which was the DNA damage response kinase Ataxia Telangiectasia Mutated (ATM). To test the requirement for ATM in FA pathway deficient cells, we interbred Fancg ± Atm± mice. Consistent with the siRNA screen result, Fancg-/- Atm-/- mice were non viable and Fancg± Atm-/- and Fancg-/- Atm ± progeny were less frequent that would have been expected. Several human cell lines with FA gene mutations were observed to have constitutive activation of ATM which was markedly reduced on correction with the appropriate wild-type FA gene. Interestingly, FA pathway deficient cells, including the FANCC mutant and FANCG mutant pancreatic cancer cell lines, were selectively sensitive to monotherapy with the ATM inhibitor KU55933, as measured by dose inhibition and colony count assays. FA pathway deficient cells also demonstrated an increased level of chromosomal breakage, cell cycle arrest and apoptosis following KU55933 treatment when compared to FA pathway corrected cells. We conclude that FA pathway deficient cells have an increased requirement for ATM activation in order to respond to sporadic DNA damage. This offers the possibility that monotherapy with ATM inhibitors could be a therapeutic strategy for tumors that are deficient for the FA pathway. No significant financial relationships to disclose.