NADPH Oxidase NOX5-S Mediates Acid-Induced mPGES1 Expression in Barrett\'s Esophageal Adenocarcinoma Cells

should undergo BE screening can decrease over-testing and significantly improve the costeffectiveness of endoscopic screening. The aim of this study is to develop a prediction rule to identify patients at high risk for BE. Methods: We are in the midst of performing a retrospective case-control logistic regression analysis. The cases are a cohort of 380 adults who were found to have histologically confirmed BE at the first (index) endoscopy at a single institution. Each case will be matched with two controls (no BE diagnosis at endoscopy) on the basis of the following: indication for endoscopy, year of endoscopy and endoscopist who performed the procedure. As we continue to accrue patient data toward our final goal of 380 BE cases and 760 controls (1:2), this is an interim analysis of 195 patients and 195 controls. Risk factors analyzed (age, gender, BMI, medical and social history and concurrent medications) were extracted from the patients' medical records. In Analysis I, we performed univariate analysis and multivariate logistic regression modeling to identify pivotal risk factors for BE. In Analysis 2, we used similar methods but in a subset of 97 BE cases and 97 controls who underwent an endoscopy for the clinical indication of GERD. Results: Univariate risk factors significantly associated with BE at endoscopy included younger age, male gender, absence of any history of cancer, aspirin use and statin use. BMI > 25 approached statistical significance (OR= 1.517; 95% CI 0.95-2.43). Three of these predictors, male gender (OR= 8.34; 95% CI 3.17-21.9), having a history of cancer (OR= 0.37; 95% CI 0.160.88) and increasing age but with a marginal effect size (OR= 0.97; 95% CI 0.94-0.99), remained significant in the multivariate regression model. In the subset analysis of patients with GERD symptoms, only male gender was identified as a predictor for Barrett's esophagus in both univariate and multivariate (OR= 6.5; 95% CI 2.27-18.7) regression models. Conclusion: Our multivariate regression model showed that male gender is a highly significant predictor for BE with an adjusted OR >8, a ratio that is higher than previously published results. Furthermore, we found that a history of cancer was associated with a significantly lower risk of BE. We hypothesize that this finding is a result of surveillance bias, as cancer patients are more likely to undergo unrelated screening studies and are therefore lower risk individuals. These interim findings suggest that effective risk stratification may be possible to determine patients most likely to benefit from endoscopic screening for BE.

Sa1078

Background: Barrett's esophagus (BE) is a risk factor for esophageal carcinoma. Therefore, regression or prevention of progression of BE has raised much interests. PPIs are commonly used for BE but the effectiveness on regression of BE is controversial. In a Cochrane metaanalysis of RCTs in 2007, PPIs up to 2 years were not significantly different from H2RAs or surgical therapy in the regression of BE. We aimed to investigate the effect of PPIs on regression of BE in non-randomized studies. Methods: A comprehensive literature search was conducted using MEDLINE and EMBASE (up to Sep. 2010) for fully published, Englishlanguage cohort studies that reported the length or area of BE at endoscopy before and after PPI treatment. Post-surgery PPI arms were excluded. StatsDirect® 2.7.7 was used to calculate pooled effect size with a random effect model. If a study had more than 1 time points, we took data from the last observed endpoint (longest duration) in the total analysis. Subgroup analysis was performed for different time points. I2 >25% was considered significant for heterogeneity between studies. Results: Literature search yielded 439 abstracts. Among 15 included cohort studies, 6 studies were subsequently excluded due to insufficient data. New squamous islands were observed in 55% (163/ 295) patients (7 studies, pooled proportion= 72%, 95% CI 54-84%). Seven studies with 24 arms (n=286) reported data (at 1 to 6 time periods) for the length of BE. Duration of PPIs ranged from 6 months to 13 years. In comparison with baseline, length of BE was significantly reduced in the total analysis of 7 studies (n=83) with overall pooled mean change as -0.52 (-0.99, -0.07, p=0.025); significant heterogeneity was detected (I2 = 50%). In subgroup analyses, the pooled mean change in length of BE at 6 months, 12 months, 1-3 years and 4-5 years were -0.38 (-0.87, 0.12; I2 = 34.4%; 3 studies, n=53); -0.54 (-1.05, -0.03; I2 = 48%; 4 studies, n=64); -0.33 (-0.81, 0.15; I2 = 56.2% ; 5 studies, n=121) and -0.45 (-1.05, 0.15; I2 = 45%; 3 studies, n=53), respectively. Data for duration longer than 5 years was all reported from the same, largest study (n=188) with duration up to 13 years (n=7) and no significant change in length was seen. Conclusion: Only a few small size cohort studies looked at the effect of long-term PPIs on regression of BE but with significant heterogeneity between studies. Present evidence indicates that PPIs lead to development of squamous islands in more than half of the patients; the length of BE was significantly reduced by ~0.5cm. However, the clinical significance of this is unclear, especially for any long term cancer prevention effect.

Sa1076 No Benefit of Autoflorescence (AFI) in a Diagnosis of Low Grade Intraepitelial Neoplasia (Lgin) in Patients With Barrett's Esophagus (BE) Magdalena Stefanova, Miroslav Zavoral, Filip Zavada, Stepan Suchanek, Inna Tuckova, Jan Martinek INTRODUCTION: Barrett esophagus(BE) is a risk factor for esophageal adenocarcinoma. Better imaging modalities such as chromoendoscopy, narrow band imaging (NBI), and autofluorescence may have the potential to improve an early detection of Barrett esophagus related neoplasia (BORN). BACKGROUND AND AIMS: The aim of this study was to evaluate the diagnostic accuracy of AFI in the detection of LGIN in patients with BE. PATIENTS AND METHOD: A prospective cohort study. In 2010 a total of 39 patients with a suspected Barrett esophagus related neoplasia were investigated using high resolution endoscopy, methylene-blue chromoendoscopy and NBI. The suspicion for BORN was based on the results of previous endoscopies with biopsies evaluated by the referring physician. All patients were also examined (during the same session) with an AFI. In each patient, suspected regions were visualised and only targeted biopsies were taken. The biopsies were also taken from all AFI positive regions. In patients without any suspected regions standard random biopsies were taken. The slides were assessed by two experienced pathologists. RESULTS: Out of 39 patients with a suspected BORN one patient with an early mucosal adenocarcinoma (EAC), 2 patients with HGIN and 8 patients with LGIN were diagnosed. In the remaining 28 patients no BORN was confirmed. In all three patients with EAC or HGIN, AFI examination was positive. Among 8 patients with LGIN, AFI was positive in 6 patients and negative in 2 patients. In the remaining patients without BORN, AFI was falsely positive in 5 patients and negative in 23 patients. The sensitivity, specificity, PPV and NPV of AFI for detecting of LGIN were 75%; 82%; 54% and 92%. CONCLUSION: AFI, in addition to high resolution endoscopy with NBI, has no clear benefit in the diagnosis of LGIN. Its positive predicted value is low. Our results further suggest the over-diagnosing of LGIN in patients with Barrett esophagus. Supported by a grant from Ministry of Defence, MO 1610

Sa1079 Medical and Surgical Therapy in Barrett's Esophagus (BE): Meta-Analyses Using Various Outcomes Mohammad Yaghoobi, Raheleh Bijarchi, Richard H. Hunt Background: Treatment of Barrett's esophagus (BE) and assessment of efficacy are challenging. Different types of outcome have been used in studies to measure the efficacy of treatment with no consensus on any one of them. We aimed to perform meta-analyses to investigate the efficacy of medical and surgical modalities in treatment of BE by using the most common outcomes applied in the literature. Methods: A comprehensive electronic literature search from 1966 to Nov 2010 was done by two reviewers for randomised controlled trials (RCTs) or observational cohorts comparing medical [H2-receptor antagonists (H2RAs) or proton pump inhibitors (PPIs)] and surgical (anti-reflux surgery) treatment of BE. There was no restriction in terms of language, location or quality score of the studies. Meta-analyses were done individually by using regression of BE, length of BE, rate of development of high grade dysplasia or adenocarcinoma, and % time with pH