an extra 3.5 months survival compared to surgery alone. Aims: To identify the most activesignalling pathways and compare different therapeutic strategies (upstream and downstreaminhibitors) to induce cytotoxicity. Material and methods: Gene set enrichment analysis(GSEA) using KEGG annotated pathways was utilised to identify the most enriched pathwaysin a gene expression dataset from 75 chemo-naive gastro-oesophageal tumours. For independ-ent validation Erk (total and phosphorylation levels) was used as a read-out for MAPKpathway activation and applied to a tissue microarray containing cores from 434 cases.The identity of the receptor tyrosine kinase(s) (RTK) responsible for MAPK activation wasascertained using a RTK array on 46 snap frozen samples for gastro-oesophageal tumours.The effects of Inhibition of active RTK(s) on proliferation, apoptosis and phosphorylationof downstream MAPK components were tested in 5 cell lines (OE33, OE19, MKN45, HSC39and KATOIII). The most effective RTK inhibitor(s) were compared to downstream inhibitionof Mek. Results: MAPK was the pathway statistically enriched in the greatest proportion ofsamples (32/75 samples, 42.7%) and 148/434 (34.3%) tumour cores displayed immuno-staining for phosphorylated Erk. Based on RTK arrays, the most frequently activated RTKslikely to account for this were members of the EGFR and FGFR family as well as c-met.No RTKs were active in 8/46 cases (17.4%), EGFR alone was active in 8/46 cases (17.4%),2 receptors, usually members of the EGFR family, were active in 13/46 cases (28.3%) andmultiple RTKs (median of 9) were active in 16/46 cases (37.0%). Targeted inhibition ofRTK(s) based on their activation profile led to inhibition of cellular proliferation, increasedapoptosis and abrogation of downstream signalling components In Vitro inall 5 lines. Themost cytotoxic combination of inhibitors was cell line dependent and could not be predictedeasily from the RTK activity profile. Inhibition of the downstream MAPK pathway componentMek induced equivalent or better cytotoxicity in all cell lines. Conclusions: We have shownthat the MAPK pathway is commonly activated and can be effectively silenced either byidentifying and inhibiting key RTKs or via use of a single drug for the downstream hubMek. The efficacy of targeting a downstream fragile point warrants testing in clinical trials.

Mo1877

Plk-1 is Upregulated in Barrett's Adenocarcinoma and is a Surrogate Marker ofAneuploidyJason M. Dunn, Stuart A. McDonald, Dahmane Oukrif, Rehan J. Haidry, Mohammed A.Butt, Laurence Lovat, Gareth H. Williams, Marco Novelli

Introduction DNA ploidy abnormalities, measured by flow or image cytometry, are markersof genomic instability that inform on future cancer risk in patients with Barrett's esophagus(BE). Barriers to routine clinical use include cost, availability and inter laboratory reproducib-ility. DNA replication licensing factors (RLFs) ensure precise duplication of the genome andcontribute to genomic stability, and may therefore act as surrogate markers of DNA ploidyabnormalities. RLFs have great potential for routine clinical use as they are measured byimmunohistochemistry (IHC). Aims To establish a surrogate IHC marker for DNA ploidythat can be utilized as a high throughput screening tool for the assessment of cancer riskin BE. Methods Esophagectomy specimens from 10 patients with Barrett's adenocarcinomawere evaluated. A total of 54 areas were marked representing the metaplasia-dysplasia-carcinoma sequence. These were removed by laser capture microdissection (LCM) andprocessed for image cytometry DNA ploidy analysis. Serial 4 μm sections were cut andstained for 5 RLFs by standard immunohistochemistry protocols, and the labeling index(LI) of percentage positive cells was calculated for each marked area. These scores werethen correlated with histology and DNA ploidy abnormalities. Results There was a significantdifference in the LI of geminin, cdc7 kinase and Polo like kinase 1 (PLK-1) between nondysplastic BE and HGD/Cancer. There was no significant difference between these twogroups for Ki67 or mcm2. PLK-1 had the highest degree of correlation with DNA ploidystatus with a Pearson coefficient r2 = 0.776 (p<0.01). When using a cut off score of 30%,the sensitivity and specificity of PLK-1 for the detection of aneuploidy was 86% and 100%respectively. Conclusion This is the first time geminin, cdc7 kinase and PLK-1 have beenshown to be upregulated in Barrett's dysplasia and adenocarcinoma. PLK-1, a key regulatorof mitosis, correlated most closely with aneuploidy measured on LCM tissue and holds greatpotential as a biomarker in BE.

Mo1878

Aberrant, DNA Damage-Response May Contribute in the CarcinogenesisProcess in Barrett's EpitheliumManisha Bajpai, Kiron M. Das

Background: During the progression of major human cancer types, the early tumorigenicevents may activate the ATR/ATM-regulated checkpoint and the tumour suppressor p53gene. Such early tumorigenic events may be: dysregulated RB, impaired DNA replication,downstream of RB such as dysregulated E2F1 and over expressed cyclin E or Cdc25A, andalterations in DNA repair genes such as BRCA2. Cells defective in their DNA damage responsecomponents compromise cell-cycle arrest, senescence and apoptosis and contribute to carci-nogenesis. We developed a novel in-vitro model of progression of Barrett's metaplasiausing hTERT immortalized benign non-neoplastic Barrett's epithelium (BE) cell line, BAR-T. Chronic intermittent acid and bile (A+B) (5 min/day) exposure caused neoplastic trans-formation of the BAR-T cells evidenced by molecular changes (~20 wks), morphologicchanges (~40 wks) and contact inhibition such as foci formation and soft agar colonyformation (at 60-65 wks) (Int J Cancer 2010). Aim: In this novel BAR-T model we examinedthe ATM/ATR checkpoint molecular events at a series of time points that correlate with theevolutionary process of tumorigenesis from BAR-T cells to neoplasia. Methods: BAR-T cellswere treated with A+B (pH4) for 5 mins everyday for over 65 wks. RT-PCR analyses wasdone for gene expression. Phase contrast microscopy and staining were performed after fociformation and colony formation in soft agar. Untreated cells grown in parallel served ascontrols. Results: Acute induction of the ATM/ATR checkpoint components: RB, E2F, CyclinE, Cdc25A and BRCA1 as well as the tumor suppressor p53 in BAR-T cells was observedat 20 wks of A+B (pH4) treated cells when compared to untreated cells (Table). Activationof the ATM-checkpoint is sustained in later stages at 40, 60 and 80 wks although at lowerlevels. Conclusion: Over expressed cyclin E, Cdc25A and E2F1 can promote unscheduledS-phase entry. Therefore it is possible that the activated DNA damage response machinery

S-667 AGA Abstracts

might play an important role in tumorigenesis in BE. According to our in-vitro model theearly tumorigenic events (20wks) occur even before the appearance of changes in cell shape(~40wks) and loss of contact inhibition (~60wks). As the A+B (pH4) treatment continuesthose cells that could evade the activated ATM-CHK2-P53 barrier imposed by the earlytumorigenic events survived and progressed to neoplasia. Therefore resolving the DNAdamage response may be the predominant mode for prevention of tumorigenic in BE.Table: Expression of selected genes compared to untreated BAR-T cells

Mo1879

Risk Factors for Prevalent Adenocarcinomas in Patients With High-GradeDysplasia in Barrett's Esophagus: A Dutch Population-Based StudyRomy E. Verbeek, Martijn G. van Oijen, Fiebo J. ten Kate, Frank P. Vleggaar, MargueriteE. Schipper, Jantine W. van Baal, Peter D. Siersema

BACKGROUND: The risk of esophageal adenocarcinoma (EAC) is increased in patients withBarrett's esophagus (BE), especially in patients with high-grade dysplasia (HGD). Interest-ingly, there is already a high prevalence of occult EAC at the time of HGD diagnosis. It ishowever unknown which patients are at an increased risk of having prevalent EAC in thepresence of HGD in BE. AIM: To identify independent risk factors for prevalent EAC in alarge nationwide population-based cohort of patients with HGD in BE in the Netherlands.METHODS: All patients with a diagnosis of HGD in BE between 1999 and 2008 in theNetherlands were identified in a nationwide histopathological registry (PALGA). Patientswith HGD and follow-up data after a diagnosis of HGD were included. All pathology reportsrelated to the esophagus were evaluated. An EAC was defined as prevalent when it wasdetected within 6 months following an initial HGD diagnosis. Patients who had at least 2different sets of biopsies from BE in the database prior to a HGD diagnosis were assumedto have participated in a surveillance program. Cox proportional hazards regression analysiswas performed to identify predictors for the presence of prevalent EAC at the moment ofHGD diagnosis in BE. RESULTS: In total, 699 patients with HGD in BE were included.Prevalent EAC was found in 177 (25%) patients with HGD. Multivariate Cox proportionalhazards regression analysis showed that the risk of prevalent EAC was increased whenpatients were between 65 and 75 years compared to a younger age (Hazard ratio (HR) 1.42,95% CI 1.01-1.98), patients were not participating in a surveillance program prior to initialHGD diagnosis compared to patients undergoing surveillance (HR 1.53, 95% CI 1.09-1.14),a diagnosis was made in a general hospital compared to a university hospital (HR 1.85,95% CI 1.14-3.01) and histologic follow-up was performed in case of HGD compared toendoscopic treatment (HR 2.60, 95% CI 1.75-3.85). The presence of unifocal compared tomultifocal HGD (HR 0.33, 95% CI 0.20-0.56) reduced the risk of detecting prevalent EAC.CONCLUSION: In this large nationwide HGD cohort, prevalent EAC is already present inat least a quarter of patients, particular when aged between 65 and 75 years, not participatingin a surveillance program prior to a HGD diagnosis, HGD was diagnosed in a generalhospital, HGD is endoscopically followed-up and not (endoscopically) treated, and HGD inBE is multifocally present. Based on this, we strongly recommend endoscopic treatmentover histologic follow-up in patients with HGD in BE, the former preferably performed incenters with expertise in this field.

Mo1880

Location of Nodal Involvement Influences the Outcome in Patients WithOesophageal AdenocarcinomaChin-Ann Johnny Ong, Susannah Worster, Ayesha Noorani, Richard H. Hardwick,Christopher J. Peters, Rebecca Fitzgerald

INTRODUCTION The incidence of oesophageal adenocarcinoma has quadrupled in the last30 years and outcomes remain poor. Oesophageal adenocarcinoma is currently stagedaccording to the TNM classification. Using reclusive partitioning, we have previously gener-ated and validated a revised lymph node classification to better prognosticate patients. Thissystem highlights the importance of identifying the number of involved lymph nodes andthe relationship to the diaphragm. The importance of nodal burden but not nodal locationwas also embraced in the recent publication of the 7th edition of TNM staging system. AIMS& METHODS The aim of this study is to compare our revised classification with the 7thedition of TNM staging system to evaluate which system is better. A retrospective cohortof 502 patients with oesophageal adenocarcinoma and underwent oesophagectomy in 6 UKtertiary referral centres was used to compare the 2 staging systems. RESULTS Table 1compares the distribution of patients staged by the 2 systems. In our cohort, there wascomplete agreement in both staging systems in identifying 271 patients with N0 and N3disease. The remaining 231 patients were staged differently by the two staging systems andformed the focus of our analysis. 59 patients staged as N2 (worst prognosis group) accordingto our revised lymph node classification were staged differently as N1 (1-2 nodes involved)and N2 (3-6 nodes involved) by the TNM7 staging system (See Table 1). Univariate analysisrevealed that the survival of these 59 patients was no different from the 100 patients stagedas the poorest prognosis group by both staging systems (p =0.542 and 0.342 respectively).Information on the location of nodal involvement was available for 51/59 patients. 88%(46/51) of these patients were classified as N2 by our system due to nodal involvement onboth sides of the diaphragm. This suggests that the TNM7 system would have inaccurately

AG

AA

bst

ract

s