gastric epithelial cells with wild-type H. pylori strain 7.13 or an isogenic 7.13 0310- mutantand quantified CagA translocation by Western blot using anti-phosphotyrosine and anti-CagA antibodies. Significantly higher levels of CagA were translocated by the 7.13 0310-

mutant compared to wild-type H. pylori strain 7.13. By utilizing proteomics technology tocontrast two closely related H. pylori strains that induce distinct phenotypes, we have shownthat HP0310 may function to alter levels of peptidoglycan, thereby affecting the compositionof substrates translocated by the cag secretion system. These studies also provide an importantfoundation for investigating the role of bacterial virulence factors in H. pylori-inducedpathologic sequelae, as such results would ultimately allow for generation of a profile of H.pylori proteins to use for screening persons infected with strains most likely to inducesevere disease.

225

Intravenous Esomeprazole for Prevention of Peptic Ulcer Re-Bleeding in aPredominantly Caucasian Population: Results On Clinical Benefits andHospital Resource UseAlan N. Barkun, Joseph J. Sung, Ernst J. Kuipers, Joachim Mössner, Dennis M. Jensen,Robert Stuart, James Y. Lau, Henrik Ahlbom, Jan Kilhamn, Peter Wahlqvist, Bengt Liljas,Tore Lind

Background: The efficacy of proton pump inhibitors in peptic ulcer bleeding (PUB) haspreviously been demonstrated in Asian, but not Caucasian populations. From a health policyperspective, it is important to assess if clinical benefits and the impact on hospital resourceuse are generalizable across populations. Methods: In this multinational, double-blind,randomized clinical trial (NCT00251979), the efficacy of intravenous (i.v.) esomeprazole(80mg infusion over 30min, then 8mg/h for 71.5h) was compared with matching placeboafter successful endoscopic hemostasis of PUB. After this initial therapy, all patients receivedoral esomeprazole 40mg qd for a further 27d. The primary outcome was clinically significantre-bleeding within 72h. Outcome variables were analyzed for the Caucasian sub-populationin a post-hoc analysis. Results: Of 767 patients randomized, 764 provided data for anITT analysis (375 esomeprazole, 389 placebo). The numbers needed to treat (NNT) withesomeprazole to avoid re-bleeding, surgery, and endoscopic re-treatment (all within 30d)were 17, 37 and 19, respectively. Clinical benefits for the Caucasian sub-population (87%of patients, n=667) were similar to the total study population for all variables investigated(table). Conclusions: Esomeprazole, given as a continuous high-dose i.v. infusion followedby an oral regimen after successful endoscopic therapy for PUB, showed low NNTs to avoidmajor clinical events with a corresponding reduction in use of hospital resources, andindicated clinical benefits in Caucasian PUB patients. Supported by AstraZeneca R&D,Mölndal, Sweden

226

Scheduled 2nd Endoscopy or High Dose Omeprazole Infusion in PreventingPeptic Ulcer Rebleeding - A Prospective Randomized TrialPhilip W. Chiu, Henry Joeng, Catherine Choi, Kelvin K. Tsoi, Kwok hung Kwong, Siu HoLam

Background We previously reported an interim analysis of this study comparing secondendoscopy against omeprazole infusion in preventing peptic ulcer rebleeding. We now reportthe final analysis of this randomized study. Method Patients who had bleeding peptic ulcerwith endoscopic stigmata of acute bleeding, visible vessel or adherent clot and hemostasisachieved on endoscopy were randomly assigned to two treatment groups. One (2nd OGD)received scheduled second endoscopy 16-24 hours after initial haemostasis, and furthertherapy applied if endoscopic stigmata persisted. Another group (PPI) received high doseadjunctive omeprazole infusion. Those who developed rebleeding in either group wouldreceive operation if further endoscopic therapy failed. Results From Nov 2003 to May 2008,556 patients presented to United Christian Hospital with bleeding peptic ulcers. Afterendoscopic haemostasis, 305 eligible patients were randomized, 153 in PPI group and 152in 2nd OGD group. 10 (6.5%) in PPI group developed rebleeding, while 12 (7.9%) in 2ndOGD group sustained rebleeding (p = 0.646; OR = 1.226, 95% CI 0.51 - 2.93). There isno difference in the probably of rebleeding within 30 days after initial endoscopy(Log ranktest p = 0.64) (Fig 1). The number of patients that required surgery for rebleeding was 6in the PPI group and 3 in 2nd OGD group (p = 0.51; OR = 0.49, 95%CI 0.12-2.0). Therewas no difference in the hospital stay, ICU stay, transfusion or mortality between the twogroups. The total hospital cost for PPI group was US$541,656.4, while that for 2nd OGDgroup was US$594654.8. Conclusion For patients with bleeding peptic ulcers that requiredendoscopic therapy, omeprazole infusion achieved similar clinical outcomes as compared

A-43 AGA Abstracts

to second endoscopy. PPI infusion incurred a lower hospital cost than scheduled second endo-scopy.

Fig 1

227

Prolonged Low-Dose Omeprazole Infusion for 7 Days Can Reduce Peptic UlcerRe-Bleeding for High-Risk Co-Morbid PatientsHsiu-Chi Cheng, Hsiao-Bai Yang, Wei-Lun Chang, Bor-Shyang Sheu

Background: Patients with co-morbidities have an increased risk of peptic ulcer re-bleedingafter gastroscopic hemostasis. The duration of re-bleeding is elongated up to 14th days afterthe first bleeding episode in these patients. Omeprazole infusion can effectively prevent pepticulcer rebleeding after gastroscopic therapy in patients without co-morbidities. However,rebleeding risk is high in co-morbid patients receiving either high-dose or low-dose omepra-zole infusion for 3 days. The optimal duration is uncertain for those with co-morbidities.Aims: To determine if prolonged low-dose intravenous omeprazole infusion up to 7 dayscould reduce peptic ulcer re-bleeding for patients with co-morbidities. Patients and methods:One-hundred and forty-seven co-morbid patients with peptic ulcer bleeding were enrolledprospectively. Five were excluded because of cancer bleeding. The majority of patientsenrolled in both groups were Class III and IV by the American Society of Anesthesiologistclassification and had similarly high proportions of Rockall risk score ≥6 (low-dose groupvs. high-dose group: 90.9% vs. 85.5%, p>0.05 and 89.4% vs. 89.5%, p>0.05, respectively).After establishing hemostasis by gastroscopic therapy, these patients received omeprazole80 mg bolus infusion soon, and were randomized into 7-day low-dose group (n=66) or 3-day high-dose group (n=76), receiving 3.3 mg/hr or 8mg/hr continuous omeprazole infusion,respectively. After omeprazole infusion, oral esomeprazole 40 mg qd was given. The primaryend-point was defined as the rebleeding within 28 days after gastroscopy. Results: The 7-day cumulative re-bleeding rate was similar between the two groups (9.5% vs. 9.7%, p>0.05)but the 7-day low-dose group had a lower risk of re-bleeding between the 8th to the 28thday, as compared to the 3-day high-dose group (0% vs. 10.7%, p=0.03; relative risk [95%]:0.52 [0.43~0.63]). The Kaplan-Meier curves confirmed that the 7-day low-dose group hadsignificantly higher cumulative re-bleeding-free proportion between the 8th to the 28th daythan the 3-day high-dose group (Log-rank test, p= 0.02). Multiple logistic regressionsconfirmed that patients with cancer and serum albumin < 3 g/dL were two independentrisk factors related to re-bleeding within the first 7 days. There was no significant independentrisk factor to determine re-bleeding in high-dose group during the 8th to the 28th day onfollow-up. Conclusions: For peptic ulcer patients with co-morbidities, delayed re-bleedingrisk during the 8th to the 28th day after gastroscopic hemostasis can be significantly reducedby a prolonged course of low-dose omeprazole infusion up to 7 days.

228

Clinical Judgement Versus Risk Scores for Predicting Outcome in PatientsBleeding from High Risk Peptic Ulcer. Preliminary Results of a ProspectiveMulticenter StudyEnric Brullet, Xavier Calvet, Rafel Campo, Michel Papo, Montserrat Planella, FélixJunquera, Pilar Garcia-Iglesias, Silvia Montoliu, Albert Pardo, Raquel Ballester-Clau,Mireia Miquel, Valentí Puig Diví, Eva Martinez-Bauer, Albert Villoria, Marta Gallach,Meritxell Casas Rodrigo, Mercedes Vergara

No studies comparing the different risk scores for predicting outcome in peptic ulcer bleeding(PUB) patients have previously been performed. In addition, it is totally unknown whetherrisk scores are superior to the clinical judgement. AIM: To compare the accuracy of Rockall,Baylor, Cedars-Sinai and Blatchford scores with the endoscopist's judgement for the predic-tion of outcome in PUB patients. PATIENTS AND METHODS: During a two years period,238 PUB patients were included at the 3 participant centers. Patients received intravenousPPI and emergency endoscopy was performed within 12 hours of admission. Combinedendoscopic therapy (epinephrine injection plus polidocanol or clip) was performed in 136patients with high risk ulcers. The endoscopist classified the patients in high-medium-lowrisk for both rebleeding and mortality and calculated the different risk scores. ROC curves,sensitivity, specificity, positive (PPV) and negative predictive values (NPV) and its 95%confidence intervals for predicting both rebleeding and mortality for each risk score andclinical judgment were calculated. RESULTS: Rebleeding and mortality rates were 14% (19/136) and 5.1% (7/136) respectively. Comparisons between clinical judgement and thedifferent risk scores are shown in the Tables. CONCLUSION: Findings of the study suggeststhat available risk scores did not add prognostic value to the clinical judgement of theendoscopist. In addition, either clinical judgement or risk scores miss 4% to 10% of patients

AG

AA

bst

ract

s

AG

AA

bst

ract

swho will rebleed following endoscopic therapy. More accurate risk scores for the predictionof rebleeding should be developed.Rebleeding

Mortality

229

Does Proton Pump Inhibitor (PPI) Treatment Initiated Before Endoscopy Workin Unselected Upper Gastrointestinal Bleeding? A Cochrane SystematicReview UpdateAravamuthan Sreedharan, Janet Martin, Grigorios I. Leontiadis, David Forman, Colin W.Howden, Paul Moayyedi

Background The clinical efficacy of PPIs initiated before index endoscopy in patients withupper gastrointestinal (UGI) bleeding is not proven. A previous Cochrane systematic review(SR) published in 2006 did not show any benefit for clinical outcomes. Objectives Toupdate the previous Cochrane SR, by systematically reviewing the evidence from randomizedcontrolled trials (RCTs) in UGI bleeding where PPI treatment was initiated before endoscopy.Methods We performed a search of CENTRAL, MEDLINE, EMBASE, CINAHL databases andmajor conference proceedings up to October 2008. Eligible RCTs compared pre-endoscopicintravenous or oral PPIs with either placebo, H2-receptor antagonists or no acid suppressivetreatment in patients with unselected acute UGI bleeding. Two reviewers independentlyassessed the eligibility and extracted data regarding outcomes and methodological quality.The primary outcome was 30-day mortality; secondary outcomes were 30-day re-bleedingand surgical rates, stigmata of recent hemorrhage (SRH; active bleeding or adherent clot)and requirement for endoscopic therapy at index endoscopy, length of hospital stay andblood transfusion requirement. Sub-group analyses were performed for the outcomes inpatients with peptic ulcer bleeding. Results Six RCTs comprising 2223 patients were included.5 were published in full and 1 as an abstract. 3 RCTs reported adequate concealment ofallocation. Meta-analyses were performed for all dichotomous outcomes. Meta-analysis forthe continuous outcome measure for length of hospital stay was not performed due toinadequate data. Sensitivity analyses were performed as appropriate and sources of heterogen-eity were explored. The main results are presented in the Table, and did not change materiallyfor sub-analysis of patients with peptic ulcer bleeding. Conclusions PPI treatment initiatedprior to endoscopy in unselected patients with UGI bleeding significantly reduced theproportion with SRH and requirement for therapy at index endoscopy. However, there wasno evidence that PPI treatment improved mortality, re-bleeding, or need for surgery.Results

*unweighted event rates

A-44AGA Abstracts

230

Active Gastrointestinal Bleeding: Detection and Localization with Contrast-Enhanced UltrasonographyNoriaki Manabe, Jiro Hata, Ken Haruma, Hiroshi Imamura, Tomoari Kamada, HiroakiKusunoki

Background and Aims: Although early detection of the bleeding site and treatment of activegastrointestinal (GI) bleeding is essential in its management, it may sometimes be difficultto establish diagnoses. Advances in contrast-enhanced ultrasonography (CEUS) could makeit possible to detect contrast leakage from GI tracts. The aim of this study was to prospectivelyevaluate the accuracy of CEUS for the depiction and localization of active GI bleeding.Subjects and Methods: Thirty-two patients (18 men 14 women; mean age, 69.0 years) withevidence of active GI bleeding were enrolled. After injection of Sonazoid, pathologic GIlesions were imaged by CEUS for 2 min. The sign of active GI bleeding was defined aspositive when pooling or leakage of contrast materials from the GI wall was observed.Calculation of sensitivity, specificity, accuracy, and positive and negative predictive valuesfor the detection of active GI bleeding by CEUS was performed on the basis of a per-patientand per-location analysis in relation to GI endoscopy. Results: CEUS depicted extravasationin 6 of 32 patients (18.8%). Patient-based sensitivity, specificity, accuracy, and positive andnegative predictive CEUS values for the detection of active GI bleeding were 71.4% (5 of7), 96.0% (24 of 25), 90.6% (29 of 32), 83.3% (5 of 6) and 92.3% (24 of 26), respectively.The location-based indexes were 71.4% (5 of 7), 98.5% (64 of 65), 95.8% (69 of 72),83.3% (5 of 6) and 97.0% (64 of 66), respectively. Conclusions: CEUS is accurate fordetection and localization of bleeding sites in patients with active GI bleeding.

231

Usefulness of Endoscopic Ultrasound (EUS) Elastography for the Detection ofMalignant Infiltration of Mediastinal and Abdominal Lymph NodesJose Larino, Julio Iglesias-Garcia, Ana Alvarez-Castro, Jose Mera, Marta Iglesias-Rivas,Ihab Abdulkader, Jeronimo Forteza, Enrique Dominguez-Munoz

Background: EUS-elastography allows analyzing tissue stiffness during a standard EUS exam-ination, which may be of help in the differential diagnosis of solid lesions. Detection ofmalignant infiltration of LN is highly relevant to define the optimal therapeutic strategy fordifferent tumors. We hypothesized that EUS-elastography may provide with additionalinformation to the conventional EUS B-mode for the detection of malignancy in mediastinaland abdominal lymph nodes (LN). The aim of the study was to evaluate the usefulness ofEUS-elastography in this setting. METHODS: 57 consecutive patients (mean age 64 years,19-82, 43 males) who underwent EUS for the evaluation of LN were prospectively includedin the study. EUS and elastography were performed under conscious sedation by using thelinear Pentax EUS (EG 3830 UT) together with the Hitachi EUB 8500 and 900. EUS-guidedfine needle aspiration was performed in all cases. Histology of surgical specimens wasconsidered as the reference method in operated cases. Positive cytology for malignancytogether with compatible EUS, PET and CT imaging were considered as the reference methodto define malignancy in non-operated cases. EUS, PET, CT imaging, clinical presentationand a minimum follow-up of six months were required for final diagnosis of benign diseasein cases of benign cytology. Elastographic pattern of the different LN are described. Probabilityof malignancy according to the EUS-elastographic pattern was calculated. Results: A totalof 63 LN were evaluated. Size of LN was 17.4mm as a mean (range 4-59mm). 54 LN werelocated in the mediastinum, and 9 in the abdomen. Malignancy was confirmed by referencemethods in 31 cases, whereas the remaining 32 LN were finally considered as benign. Threedifferent elastographic patterns were identified: 1) a heterogeneous blue-predominant pattern(n=26 LN), 2) a heterogeneous green-predominant pattern (n=23 LN), and 3) a heterogeneousmixed green-blue pattern with geographical appearance and no color predominance (n=14LN). Malignant LN showed either a blue-predominant pattern (n=24 LN) or a mixed green-blue pattern (n=7 LN). On the contrary, most benign LN (72%) showed a green-predominantpattern. The probability of malignancy in a LN showing a green-predominant pattern wasof 0%, and of 92% in case of a blue-predominant pattern. Finally, the probability ofmalignancy in a LN showing a mixed green-blue pattern was of 50%. Conclusions: EUS-elastography is a very useful tool for the differential diagnosis of mediastinal and abdominalLN. It provides with specific colour patterns supporting the malignant or benign nature ofthe LN.

232

High-Resolution Magnetic Resonance Colonography and Contrast-EnhancedMagnetic Resonance Imaging Can Distinguish Early Murine Colorectal Tumorsfrom Colitis In VivoDevkumar Mustafi, Urszula Dougherty, Xiaobing Fan, Gregory S. Karczmar, EricaMarkiewicz, Marta Zamora, Marc Bissonnette

PURPOSE: Ulcerative colitis is characterized by persistent or recurrent inflammation andcan progress to colon cancer. Currently, there are no sensitive and specific non-invasivemethods to identify early tumors. The aim of this study was to develop a high-resolutionmagnetic resonance imaging (MRI) method for early detection of colitis-associated colorectaltumors. METHODS AND MATERIALS: C57Bl6 mice (n=10) were treated with 2.5% dextransulfate sodium (DSS) in the drinking water for five days to induce colitis. In a second group(n=10), tumors were induced with azoxymethane weekly for two weeks (10 mg/kg bodywt), followed by two cycles of DSS, one at week 2 and one at week 6. These models mimicmany clinical and pathological features of colitis and colorectal cancer. In the group receivingDSS alone, mice were imaged weekly from day 5 to day 25 after DSS. Mice bearing colitis-associated early tumors were imaged weekly between 8-11 weeks after AOM/DSS treatments.We used a 9.4 Tesla Bruker scanner for In Vivo MR colonography. High-resolution T1/T2-weighted MR images were acquired from multiple slices with thickness≤1 mm. For dynamiccontrast-enhanced MRI (DCEMRI), we injected a gadolinium-based contrast agent (Gd-DTPA, ~0.13 mmol/kg) and acquired images for ~12 min. Rate constants for Gd uptake(Ktrans) and its distribution (Ve) in muscle, normal and inflamed colons, and tumors were