Abstracts

with Mantel-Haenszel method (fixed effects model). Heterogeneity was assessed bycalculating the I2 measure of inconsistency. RevMan 5 was utilized for statisticalanalysis. Results: 5 studies (NZ355) met the inclusion criteria with mean ageranging from 67-76 years. The studies were of adequate quality (Jadad scores of 2).No significant differences were noted between early (% 3 hours) and delayed ornext-day feedings for patient complications (OR 0.78, 95% CI: 0.39 - 1.53, pZ0.47),death in %72 hours (OR 0.60, 95% CI: 0.18 - 1.99, pZ0.40), and number ofsignificant gastric residual volume during day 1 (OR 1.46, 95% CI: 0.75 - 2.84,pZ0.27). No publication bias and no significant heterogeneity were noted.Conclusions: Early tube feeding %3 hours after PEG placement has no significantdifferences to delayed or next-day feeding in respect to complications, death in %72 hours, or number of significant gastric residual volumes at day 1. Early feedingwithin 3 hours of PEG placement is a safe alternative to delayed feeding.

S1485

Gastric Antral Pancreatic Rest Resection Using Endoscopic Band

Ligation Snare Polypectomy TechniqueAndrew J. Bain, David J. Owens, Raymond Tang, Thomas SavidesBackground: Although gastric antral pancreatic rests (PR) have characteristicendoscopic and EUS features, confirming a histologic diagnosis may be desirable toexclude other significant pathology. Forceps biopsy and snare polypectomy areoften unsuccessful due to inadequate depth of sampling. Removal with endoscopicband ligation snare polypectomy technique (EBLSP) may be a safe and effective wayof obtaining a tissue diagnosis of PR. Aims: To assess the efficacy and safety ofEBLSP in obtaining a tissue diagnosis of PR compared to sampling with biopsy þ/-snare polypectomy, and to characterize PR endoscopic appearance. Methods:Retrospective case series in which an electronic GI endoscopic report database wassearched for patients who underwent EGD between 1/00 and 10/08 at an academicmedical center by a single endoscopist and were found to have gastric antralsubepithelial masses. The images were reviewed for endoscopic appearance andtissue sampling technique, and final pathologic results were recorded. EBLSP wasperformed with rubber band ligation followed by snare polypectomy and clipplacement. Results: 61 patients with subepithelial antral masses were identified.21(34%) were histology proven PR, of which 16(76%) were obtained with EBLSP,4(19%) with forceps biopsy þ/- snare polypectomy, and 1(5%) with cap EMRtechnique. The median endoscopic size of histology proven PR was 10 mm (range6-15 mm). The median distance from the pylorus was 35 mm (range 10-60 mm).The median location relative to the pylorus was at the 6 o’clock position (range 3-7o’clock). 13(62%) displayed central umbilication. EUS was performed on 19(90%)PR with median dimensions being 8.5 by 4 mm. All PR were hypoechoic orheterogenous lesions located in the submucosal layer. Among patients with antralmasses that endoscopically appeared to be PR, 16 of 18 (89%) who underwentEBLSP had a histologic diagnosis of PR, compared to 4 of 13 (31%) who hadsampling with forceps biopsy þ/- snare polypectomy (pZ0.007). No complicationsoccurred with any tissue sampling method used. Conclusions: 1) Endoscopicsubmucosal resection of gastric antral pancreatic rests can be safely and effectivelyperformed with band ligation snare polypectomy for pathologic diagnosis. 2) EBLSPhas a significantly higher diagnostic yield for suspected pancreatic rests thanforceps biopsy þ/- snare polypectomy (89% vs 31%, pZ0.007). 3) The endoscopicappearance of histology proven pancreatic rests is a raised, 6-15 mm diametersubepithelial mass with central umbilication that is located 1-6 cm from the pylorusin the 3-7 o’clock position.

S1486

Study of Patients with Malignant Lymphoma Diagnosed By

Upper EndoscopyYutaka HirayamaBackground and Aim: The gastrointestinal tract is one of the most commonlyinvolved extranodal sites of malignant lymphoma (ML). These lesions may be seenas primary GI lymphoma or as secondary one. Our interest in lymphomatousneoplasia of the GI tract is due to the large number of patients with ML diagnosedby upper endoscopy. A retrospective study of patients with ML was conducted in anattempt to examine clinical and endoscopic features of ML. Patients and Methods:56 patients with ML (75 lesions) diagnosed by upper endoscopy between April 1,2000 and August 31, 2008 were enrolled. We analyzed clinical features, localization,macroscopic morphology, and differences in the histological diagnosis. And, wealso examined the feature of magnified endoscopic images with the narrow bandimaging (NBI) system in recent 4 patients. Results: 56 patients (age: 62.91 � 14.23,sex: M/F 34/22) with ML were confirmed. 27 lesions (36%) of ML occurred in middlethird stomach, 20 lesions (26.7%) in lower third stomach, 13 lesions (17.3%) induodenum, 8 lesions (10.7%) in upper third stomach, 5 lesions (6.7%) in entirestomach, and 2 lesions (2.7%) in esophagus. In terms of macroscopic morphology,32 lesions (42.7%) were present as superficial depressed type, 19 lesions (25.3%) asulcer type, 13 lesions (17.3%) as fungated type, 6 lesions (8%) as polypoid type,5 lesions (6.7%) as giant fold type. Out of them, 13 lesions (17.3%) were present asa single ulcer. In the histological diagnosis, there were 38 patients (67.9%) with

AB184 GASTROINTESTINAL ENDOSCOPY Volume 69, No. 5 : 2009

diffuse large B-cell lymphoma (DLBCL) (50 lesions, including 5 entire stomachlesions), 8 patients (14.3%) with MALT lymphoma (11 lesions), 4 patients (7.1%)with follicular lymphoma (6 lesions), 4 patients (7.1%) with mantle cell lymphoma(6 lesions), 1 patient (1.8%) with Burkitt’s lymphoma, and 1 patient (1.8%) withadult T-cell lymphoma. Helicobacter pylori was present histologically in 6 patients(75%) of MALT lymphoma group. In a magnified endoscopic imaging with NBI, wecould recognize disappearance of regular gastric pits or appearance of irregularvessel pattern in all of 4 patients with ML (2 with DLBCL, 1 with Mantle celllymphoma, and 1 with MALT lymphoma). Limitation: This was a single-center study.Conclusion: Clinical and endoscopic features of malignant lymphoma were varied.Therefore, we should treat these lesions carefully.

S1487

Matrix Metalloproteinase-3 May Perform An Important Function

in Gastric Ulcer HealingTakafumi Ando, Osamu Watanabe, Kazuhiro Ishiguro,Motofusa Hasegawa, Nobuyuki Miyake, Shinya Kondo, Tsuyoshi Kato,Ryoji Miyahara, Naoki Ohmiya, Yasumasa Niwa, Hidemi GotoBackground/Aims: Matrix metalloproteinases (MMPs) are endopeptidases whichperform important functions in extracellular matrix remodeling, cell proliferation,and inflammatory processes. Here, we compared MMP-3 levels with those of tissueinhibitor of metalloproteinases (TIMP)-1 and several inflammatory cytokines ingastric ulcer (GU) patients. Methods: This study enrolled 50 patients with GU and 6with functional dyspepsia (FD). Samples of gastric mucosa from the antrum andedge of the ulcer were harvested from GU patients and of antral mucosa alone fromFD patients during upper gastrointestinal endoscopy. Mucosal biopsy tissues werecultured for 24 h, and the culture supernatant was measured for levels of MMP-3,TIMP-1, IL-1b, IL-6, and IL-8. Results: All GU patients were positive for Helicobacterpylori (H. pylori) while all FD patients were negative. Antral levels of TIMP-1, IL-1b,IL-6, and IL-8 were significantly higher in GU than FD patients. Further, MMP-3levels were significantly higher in GU patients at the edge of the ulcer than in theantrum, and had a significantly positive correlation with TIMP-1, IL-1b, IL-6, and IL-8. Conclusion: MMP-3 levels were significantly higher at the edge of the ulcer thanin the antrum, suggesting that MMP-3 may perform an important function in gastriculcer healing.

S1488

Risk of Colorectal Polyps in Patients with Gastric Fundic Gland

PolypsPablo Luna, Jose M. Mella, Lisandro Pereyra, Carolina Fischer,Adriana Mohaidle, Adrian Hadad, Beatriz Vizcaino, Mario A. Medrano,Silvia C. Pedreira, Luis A. Boerr, Daniel G. CimminoIntroduction: The risk of gastric and duodenal polyps is higher in severalsyndromes of colonic polyposis. The risk of colonic polyps and adenomas inpatients with gastric polyps, especially the fundic glandular type, is not wellestablished. Aim: To assess the risk of presenting colonic polyps, adenomas andadvanced neoplastic lesions (ANL) in patients with fundic gland gastric polyps.Materials and Methods: The clinical records of patients who had performed a highand a low digestive endoscopy between September 2007 and August 2008 wereretrospectively analyzed. Those with previous digestive endoscopies, an inadequatecolonic cleansing, an incomplete colonoscopy, gastric or colonic surgeries, andintestinal inflammatory disease were excluded. A case-control study was carriedout, defining patients with gastric polyps as ‘‘cases’’, and patients without gastricpolyps as ‘‘controls’’. The risk was assessed, measured in odds ratio (OR) and itscorresponding confidence intervals 95% (CI), of presenting colonic polyps,adenomas and ANL (villous component O25%, size R10 mm, or high gradedysplasia). Results. 247 patients were analyzed: 70 with gastric polyps (cases) and177 without gastric polyps (controls). In the cases, the media age was 60 � 13 yearsold, 71% women; the most frequent types of gastric polyps were fundic glandular(74%, CI 62-84) and hiperplastic (24%, CI 14-36); 72% had inactive or superficialgastritis; and 13%, active or severe gastritis; the Helicobater pylori had a presence of16%; 22% had colonic polyps (25% hyperplastic and 68% adenomas, from which45% were ANL). In the controls, the media age was 61 � 13 years old, 61% women;52% had inactive or superficial gastritis and 30%, active or severe gastritis; theHelicobater pylori had a presence of 27%; 20% had colonic polyps (31%hyperplastic and 63% adenomas, from which 41% were ANL). The presence ofgastric polyps showed an OR of 1.20 (CI 0.62-2.3) for colonic polyps, an OR of 1.65(CI 0.44-5.82) for adenomas, and an OR of 1.44 (CI 0.41-5.20) for ANL. The presenceof fundic gland polyps showed an OR of 0.80 (CI 0.22-2.8) for colonic polyps, an ORof 0.78 (CI 0.08-8.7) for colonic adenomas, and an OR of 0.08 (CI 0.007-1.02) forANL. Conclusion. The results of this work did not show a higher risk of colorectaladenomas or ANL in patients with fundic gland polyps. On the contrary, thereseems to be a tendency to a reduction of such risk before the presence of this typeof gastric polyps.

www.giejournal.org