disease. In addition, in paired biopsies within one patient no significant differences in expression were found between affected and unaffected parts of the colon. Conclusion: This study demonstrates for the first time that Claudin 18 is expressed in the human colon. The expression is largely increased in patients with ulcerative colitis. The observation that this appears unrelated to the severity of the inflammation might point to a primary defect in regulation in patients with ulcerative colitis and warrants further examination. M1694 Ampullary Lesions: Incidental Adenomas Vs FAP Associated Adenomas. Results of a Single Referral Center Shayan Irani, Andrew D. Arai, Kamran Ayub, John J. Brandabur, Russell Dorer, Margaret Farrell-Ross, Michael Gluck, Geoffrey Jiranek, David Patterson, Drew Schembre, Richard A. Kozarek BACKGROUND & PURPOSE: FAP is associated with a high risk of duodenal and ampullary neoplasms. Tumors of the ampulla occur rarely in non-FAP patients (prevalence of .04 -.12%). We compared these 2 groups of patients with ampullary adenomas. METHODS: All patients with ampullary adenomas referred for endoscopic papillectomy between 9/1997 and 9/2007 were reviewed. Data was gathered on clinical presentation, endscopic details (tumor location/lateral extension/associated duodenal adenomas etc.), complications, histo- logy, follow up and need for surgery. Recurrence was defined as the reappearance of gross or microscopic adenoma with a negative endoscopy and biopsy between the ampullectomy and the endoscopy demonstrating a recurrence. Endoscopic success was defined as no recurrence after ampullectomy and no invasive cancer on ampullectomy. RESULTS: From 9/97 to 9/07, 193 patients (including 7 patients with minor and major) underwent endoscopy for ampullary lesions. Of these, 25 were either juxta-ampullary lesions(15) or normal variants of the ampulla(10). There were 18 non-adenomatous lesions of the ampulla, 112 ampullary adenomas (Incidental 89, FAP 23) and 38 ampullary adenocarcinomas (all in the incidental group). The patients in the incidental group tended to be older and had a larger mean tumor size. 6 of the 7 patients with minor and major ampullary adenomas, were in the FAP group. As expected, associated c-loop adenomas were also more common in the FAP group vs the incidental group (11 vs 1). Initial endoscopic unresectability was more common in the incidental group vs FAP group (22% vs 4%). There were 8 patients with invasive cancer found after ampullectomy in the incidental group vs 0 in the FAP group. There were a total of 2 severe complications in the incidental group (1 retroperitoneal and 1 intraperitoneal perforation). The other complications in both groups were mild and included mild pancreat- itis or endoscopically controlled GI bleed requiring a maximum of 3 days of admission. CONCLUSIONS: 1) The majority of ampullary adenomas are endoscopically amenable in both incidental and FAP groups of patients. 2) The mean size of adenomas and the number of adenocarcinomas tends to be smaller in the FAP group. 3) The recurrence rates and the success rates tend to be comparable in the two groups. 4) The latter two findings can be attributed to surveillance programs. Incidental vs FAP associated ampullary adenomatous lesions M1695 Cyclical Rifaximin for Symptomatic, Uncomplicated Diverticular Disease: A Meta-Analysis Krishna S. Kasturi, Rajasekhara R. Mummadi, Sathya Jaganmohan Background & Aim: About 20% of patients with colonic diverticulosis manifest symptoms, and some individuals experience persistent symptoms despite adequate dietary fiber use. Intestinal bacterial overgrowth causes mucosal inflammation and contributes to symptoms. The efficacy of long-term use of poorly absorbed antibiotics, such as rifaximin in such patients is unclear. Our aim was to evaluate this effect size. Methods: Electronic databases MEDLINE, EMBASE, and Cochrane Controlled Trials Register were searched for published randomized controlled trials that compared cyclical rifaximin to a placebo in addition to dietary fiber for patients with symptomatic uncomplicated diverticular disease (UDD). Data was extracted on the mean symptom scores (tenesmus, abdominal pain/ discomfort, bloating, diarrhea) in the control and treatment groups along with the reported incidence of complica- tions. Weighted mean difference and pooled proportion improvement in symptom scores were computed (Review Manager 5.0). Results: A total of 5 studies (Table) met the inclusion criteria. Total 1782 (808M,974F). Baseline symptom score ranged from 6.2 - 6.8 in the placebo group and 6.4 - 7.1 in the rifaximin group (p = NS). Studies were heterogeneous. The weighted mean difference for improvement in baseline symptom score was 1.71 [95% CI = 1.28 - 2.14] favoring the Rifaximin group (Inverse Variance, Random effects). Pooled percentage improvement in symptom scores was 57.75% (placebo group) and 80.51% (rifaximin group). The occurrence of complications (diverticulitis, bleeding) was also less in the rifaximin group (Mantel-Haenszel, Fixed effect, OR 0.40 [95% CI 0.21 - 0.75]). Conclusion: Cyclical Rifaximin appears to be an effective and safe adjunct to dietary fiber in reducing symptoms and complications in UDD. Study characteristics A-399 AGA Abstracts M1696 The Quality of On-Line He@lth Information in Inflammatory Bowel Disease Sander van der Marel, Herma Fidder, Daniel W. Hommes Background: The internet is now the largest source of health information and is widely used by both patients and health professionals. The growing amount of uncontrolled information available on inflammatory bowel disease (IBD) makes an evaluation necessary. Aims and Methods: To assess the quality of IBD information on the internet we performed a search using the phrases ‘inflammatory bowel disease', ‘Crohn's disease' and ‘ulcerative colitis'. The search terms were entered separately into the six most commonly used search engines1: Google, Yahoo! Search, AOL, MSN search, AlltheWeb, and Altavista. The first 30 results were used for initial evaluation. Web sites were categorized as institutional, pharmaceutical, non-pharmaceutical commercial sites, charitable, support or alternative medicine. Web sites were evaluated for content quality using the validated DISCERN2 rating instrument. Based on the total DISCERN score the web sites were grouped into categories: excellent (63-75), good (51-62), fair (39-50), poor (27-38), and very poor (15-26). In order to assess readability we used Word processor Microsoft Word 2003 in which free text was scored by using the Flesch Reading Ease and the Flesch-Kincaid Grade Level score. The Flesch Reading Ease score rates text on a 100-point scale. A score between 60-70 has been recommended for health information3. The Flesch-Kincaid Grade Level score rates text on a U.S. school-grade level (recommended level 6 to 8) 3. Results: A total of 86 websites were evaluated: 33% were institutional, 6% pharmaceutical, 29% non-pharmaceutical commercial sites, 5% charit- able, 21% support, and 7% alternative medicine. The DISCERN tool rated 16% of the sites as excellent, 26% good, 38% fair, 12% poor, and 8% very poor. The average rating for institutional websites was 48 (fair), pharmaceutical 38 (poor), non-pharmaceutical commer- cial sites 51 (good), charitable 57 (good), support 50 (fair) and alternative medicine 35 (poor). Interobserver agreement was good (weighted kappa statistics 0.6 (95% CI 0.41- 0.79). Most sites scored low on questions related to sources of information, no treatment and shared decision making. The median Flesch Reading Ease score for the sites was 42 (range 0-78), whereas only 3 sites scored a Flesch-Kincaid Grade Level below 8. Conclusions: The quality of most websites containing information is substandard, representing an unmet need for optimization. Charitable websites appear to have the highest quality of written information and alternative medicine websites the lowest. Most sites have high reading levels, which may compromise the accessibility of the information to large groups of the population. AGA Abstracts