AGA Abstracts in EGD, and greater primary SUV in PET scan were identified as clinical predictors of locally advanced EAC by pathological staging. Locally Advanced Factor Early EAC N=69 P-value EAC N=29 Age (years) 67.2±9.3 64.0±10.0 0.145 Male gender 58 (84.0%) 26 (89.6%) 0.546 Smoking status: 21 (30.4%) 21 (30.4%) 0.842 Never Smoking: Pack/year 18 [2;57] 20 [5;55] 0.478 Alcoholism history 2 (4.8%) 3 (15.7%) 0.314 BMI (kg/m2) 27.2 [24.5;32.9] 22.4 [19.9;25.1] <0.001 Hemoglobin (mg/ 13.6±1.9 14.2±1.7 0.224 dl) Dysphagia 21 (30.4%) 17 (58.6%) 0.017 Heartburn/Regurgi- 24 (34.7%) 12 (41.3%) 0.697 tation Chest pain 5 (7.2%) 2 (6.8%) 0.99 Post prandial dis- 7 (10.1%) 3 (10.3%) 0.99 tress Unintentional 12 (17.3%) 11 (37.9%) 0.054 weight loss GI bleeding at diag- 15 (21.7%) 4 (13.7%) 0.53 nosis Prior diagnosis of 52 (75.3%) 17 (58.6%) 0.157 Barrett's esophagus Barrett's length in 5.6±3.3 6.0±4.1 0.39 EGD (cm) Tumor size in EGD 2.6±1.8 3.5±1.9 0.014 (cm) Endoscopic appear- 0.001 ance -Mass 27 (39.1%) 20 (68.9%) -Nodule 37 (53.3%) 5 (17.2%) -Submucosal infil- 1 (1.9%) 3 (10.5%) tration -Ulcerated 4 (5.7%) 1 (3.8%) Tumor Circumfer- 0.002 ence: - 0% - 33% 54 (84.5%) 14 (53.6%) -34% - 66% 9 (14.062%) 7 (26.3%) -67% - 100% 1 (1.562%) 5 (19.1%) Tumor SUV on PET 0.022 Scan: 15 (21.7%) / 33 2 (6.8%) / 10 -less than 2.5/ 2.6 (47.8%) / 21 (34.4%) / 17 to 5/ larger than 5 (30.4%) (58.6%) SUV of lymph 0.237 nodes on PET scan -less than 2.5/ 2.5 59 (85.5%) / 5 25 (86.2%) / 4 to 5/ larger than 5 (7.2%) / 5 (7.2%) (13.7%) / 0 (0.0%) Statistics presented as Mean ± SD or N (column %). BMI=Body mass index; GEJ= gastroesophageal junction; SUV=standardized uptake value Su1127 INCREASED SOX2 EXPRESSION IN THE EARLY STAGE OF ESOPHAGEAL SQUAMOUS CELL CARCINOGENESIS INDUCE CELL PROLIFERATION AND TUMOR ANGIOGENSIS Kiichi Takahashi, Naoki Asano, Akira Imatani, Xiaoyi Jin, Masahiro Saito, Waku Hatta, Kiyotaka Asanuma, Kaname Uno, Tomoyuki Koike, Atsushi Masamune Background and Aim: Recent whole-genome and exome sequencing analyses have revealed the genetic alterations in the advanced esophageal squamous cell carcinoma (ESCC). These analyses showed that Sox2, a transcription factor essential for the maintenance of stem cell, is overexpressed in the advanced ESCC. On the other hand, we previously reported that deviation from the regulation by Sox2 causes gastric carcinogenesis upon Helicobacter pyloriin- fection (DDW2012, 2014 and 2018). Hence, we hypothesized that Sox2 overexpression may also contribute to the initiation of esophageal carcinogenesis and aimed to investigate whether Sox2 overexpression is involved in the carcinogenesis and tumor angiogenesis in the early stage of ESCC. Methods and Results: Immunohistochemistry was performed to determine the expression of Sox2 in human intraepithelial ESCC obtained by endoscopic resection (n=47). Quantitative image analysis using TissueFAXS ® revealed that the expression of Sox2 was significantly increased to 1.37±0.16 folds in the cancer tissues compared to the adjacent normal tissues, suggesting that Sox2 expression is amplified even in the initiation of esophageal carcinogenesis. Next, after confirming that several human ESCC cell lines highly expressed Sox2 compared to the normal human esophageal epithelial cell line Het- 1A, we established Het-1A cells that overexpress Sox2 by the addition of Doxycycline (Dox) (HetSox2) as an in vitromodel of esophageal carcinogenesis. In MTS assay, Sox2 overexpressing HetSox2 cells displayed increased cell proliferation (1.26±0.18 folds, p<0.05). In endothelial cell (HUVEC) transwell migration assay, HUVECs with Dox-treated HetSox2 S-512 AGA Abstracts migrated through the membrane to a greater degree than untreated control cells (2.26±0.82- fold, p<0.05). In sprouting angiogenesis assay, sprout length of HUVECs with Dox-treated HetSox2 tended to be longer compared to that of Dox-untreated HetSox2 (1.72±0.46 folds).These results suggested that Sox2 induced cell proliferation and angiogenesis in normal human esophageal epithelial cells. We then performed gene expression microarray analysis of Dox-treated and -untreated HetSox2 to investigate genes variated by Sox2 overexpression. Sox2 overexpression up-regulated 95 genes (IGFBP3, LOXL2, CA9)(>2-folds), and down- regulated 75 genes (EGFL6, FGF9, LCN2) (<0.5-folds). Among them, 15 genes were related to angiogenesis, 11 were related to carcinogenesis, and 38 were related to cell proliferation. These genes could be acting downstream of Sox2 to induce cell proliferation and tumor angiogenesis. Conclusions. Our study suggested that overexpression of Sox2 contributed to initial carcinogenesis and tumor neovascularization in the early stage of ESCC. This may give us a clue to understand the mechanism of esophageal carcinogenesis, which could lead us to a new endoscopic diagnosis and therapy for ESCC Su1128 CLINICAL EPIDEMIOLOGY OF ESOPHAGEAL CANCER: NATIONWIDE MULTICENTER HISTORICAL COHORT OVER 10 YEARS Hye-Kyung Jung, Chung Hyun Tae, Hyuk Lee, Kee Don Choi, Jun Chul Park, Joong Goo Kwon, Yoon Jin Choi, Su Jin Hong, Jaekyu Sung, Woo Chul Chung, Kibae Kim, Seung Young Kim, Kyung Ho Song, Kyung Sik Park, Seong Woo Jeon, Byung-Wook Kim, Han Seung Ryu, Ok-Jae Lee, Gwang Ho Baik, Yong Sung Kim, Hwoon-Yong Jung Aims: There are few esophageal cancer studies that reflect real world data including actual clinical situations or advances in treatment modality. This study was aimed to analyze the treatment and prognosis according to the epidemiological change of esophageal cancer based on the hospital cohort over a 10-year period in Korea. Methods: This historical cohort was recruited by 20 referral hospitals nationwide for newly diagnosed esophageal cancer patients from January 1, 2005 to December 31, 2017 (N=6354). The cancer stage at the initial diagnosis was defined according to the 7th edition of American Joint Committee on Cancer (AJCC) on TNM staging system. Results: The mean age of patients with 6354 incident cases was 64.9 ± 9.0 years with male predominance (92.9%). Squamous cell cancer was 96.9% and early esophageal cancer (stage I) has been increased around 2010 (24.7% vs. 37.2%, p < 0.001). Lower located esophageal cancer was most frequent in both squamous cell cancer and adenocarcinoma (43.2% vs. 43.1%). Endoscopic resection (ER) was performed in 5.3%, surgery in 52.3% of esophageal cancer, and definitive concurrent chemoradiotherapy (CCRT) 27.0%. Risk adjusted overall survival (OR) of stage I was 9.6 years, stage II, 4.3 years, stage III, 1.8 years, stage IV 0.9 years (p value for all pairwise log-rank test < 0.001). In stage Ia, ER showed similar OR compared to patients who underwent surgery. In locally advanced patients, OR in patients with surgery with/without chemotherapy or radiotherapy was better than that in patients with definitive CCRT (6.9 ± 0.3 years vs. 1.5 ± 0.1 years, Log Rank p value < 0.001). Conclusions: ER showed an equivalent effect to surgery in esophageal mucosal cancer. In patients who had locally advanced esophageal cancer, surgery with/without chemotherapy or radiotherapy revealed better survival than definitive CCRT, therefore, surgery should be actively considered in these groups. Su1129 PREDICTIVE SIGNIFICANCE OF ENDOSONOGRAPHIC (EUS) STAGING IN LOCALLY ADVANCED GASTRIC (GC) OR GASTROESOPHAGEAL- JUNCTION (AEG) ADENOCARCINOMA: ANALYSIS OF DATA FROM A PERIOPERATIVE AIO-CAO PHASE II STUDY Visvakanth Sivanathan, Thomas Thomaidis, Michael Stahl, Christoph Utz, Annett Maderer, Florian Lordick, André Mihaljevic, Stephan Kanzler, Thomas Höhler, Peter Thuss-Patience, Stefan Moenig, Volker Kunzmann, Sebastian Schroll, Stefan Ibach, Andrea Tannapfel, Hans-Joachim Meyer, Christoph Schuhmacher, Hans Jochen Wilke, Katrin Krause, Aysun Karatas, Peter R. Galle, Markus Moehler Introduction: This AIO-CAO Phase II Study investigated the additional administration of Panitumumab to a basic ECX therapy (Epirubicin, Cisplatin and Capecitabine) in GC or AEG. This evaluation correlated the preoperative EUS, after neoadjuvant chemotherapy, with the therapy response in addition to the initial examination (baseline EUS). Methods: