S1506

Successful Circumferential Ablation of Barrett’s Esophagus (BE)

with Low Grade Dysplasia (LGD) Using the HALO360 Ablation

System: One-Year Follow-Up of the AIM-LGD Pilot TrialVirender Sharma, H. Jae Kim, Roxane Mclaughlin, Michelle Moirano,Michael Crowell, Patrick J. Dean, David E. FleischerAims: To assess the safety, tolerability and effectiveness of circumferential ablationusing the HALO360 Ablation System (BARRX Medical, Sunnyvale, CA) for theendoscopic ablation of BE-LGD. Methods: To be eligible for enrollment, patientshad up to 6 cm of biopsy-proven BE-LGD demonstrated on at least 2 EGD/biopsysessions in the prior 2 yrs, and independent confirmation by 2 pathologists. TheHALO360 Ablation System consists of a balloon-based ablative catheter which, wheninflated to 7 psi, delivers a pre-set amount of RF energy density (12 J/cm2) to the BEtissue at high power (300 W). In previous reports, these settings result in uniformablation to the level of the muscularis mucosae without subsequent strictureformation. In this study, ablation was delivered twice to the entire length of BE-LGD, followed by lansoprazole 30 mg bid for 1 yr. Post-ablation symptoms (chest,throat and abdominal pain, odynophagia, dysphagia) were quantified for 14 daysusing a diary (visual analog scale, VAS, 0-100 mm). Patients underwent EGD with4Q/1 cm biopsies at 1, 3, 6, 12 mo, with a second ablation at 4 mo if residual BE orLGD was present. Pathology review was blinded. Complete Response (CR) isdefined as all biopsies negative for BE-LGD or BE, analyzed separately. Primaryendpoint: CR rate for BE-LGD at 1 yr (% pts with CR). Secondary endpoint: CR ratefor BE at 1 yr. Results: Ten patients (9 men, mean age 56 years, range 26-79) withBE-LGD (median length 5 cm, range 2-6) were treated. Median primary proceduretime was 26 min (range 19-37). One patient had a brief and mild GI bleed on ASA,which resolved without intervention. There were no other serious adverse eventsand no strictures. Median day #1 chest pain was 8/100 (IQR 0-45), returning to 0/100 on day #9. Median day #1 dysphagia was 8/100 (IQR 5-35), returning to 0/100on day #3. Histological results are available at 6 mo (n Z 10) and 12 mo (n Z 5).At 6 and 12 mo, 100% of patients were CR for the LGD endpoint. At 6 and 12 mo,60% and 80% of patients were CR for BE endpoint, respectively. In those patientswith persistent BE, only very focal (!5 mm) islands remained. There were noburied glands in O450 follow-up biopsies. Conclusion: Circumferential RF ablationof BE-LGD using the HALO360 Ablation System safely and effectively eliminates BE-LGD in 100% of patients at 1 year. Residual BE is uncommon and, when present,occurs as small foci of non-dysplastic BE. This trial has been extended to 2-yearfollow-up in order to collect longer-term data and to enable focal ablation of allpatients with residual BE using a focal ablation device incorporating a similarelectrode technology as used in this study.

S1507

Circumferential RF Ablation for Non-Dysplastic Barrett’s

Esophagus (NDBE) Using the HALO360 Ablation System (AIM

Trial): One-Year Follow-Up of 100 PatientsDavid E. Fleischer, Virender Sharma, Alvaro Reymunde,Michael Kimmey, Ram Chuttani, Bergein Overholt, Kenneth Chang,Charles Lightdale, Nilda Santiago, Douglas Pleskow, M. Brian Fennerty,Patrick J. Dean, Kenneth K. WangAims: To assess the dose-response, safety, tolerability and effectiveness ofcircumferential ablation using the HALO360 Ablation System (BARRX Medical,Sunnyvale, CA) for the endoscopic ablation of NDBE. Methods: This study wasconducted in 2 phases (AIM-I and AIM-II) at 8 U.S. centers from Sep 2003 to Sep2005. The HALO360 Ablation System consists of a balloon-based endoscopiccatheter, which delivers a pre-set amount of RF energy density (J/cm2) to NDBE at300 W. In previous reports, this device ablates to the muscularis mucosae withoutstricture formation. The dose-escalating phase (AIM-I, n Z 32) included pts with2-3 cm of BE and increasing doses (6, 8, 10, 12 J/cm2). The definitive phase (AIM-II,n Z 70) included pts with 2-6 cm of NDBE and treated all with 10 J/cm2 deliveredtwice (2�) per session. Patients received esomeprazole 40 bid for 1 mo post-ablation, and 40 qd thereafter. Post-ablation symptoms (chest, throat, abdominalpain; odynophagia; dysphagia) were quantified with a 14-d diary (visual analogscale, VAS, 0-100). Patients underwent EGD, 4Q/q2 cm biopsies at 1, 3, 6, 12 mo,with a 2nd ablation after 3 mo if residual NDBE. Complete Response (CR): all bxneg for BE. Biopsy Clearance Rate (BCR): % bx per patient per EGD neg for BE.Results: In AIM-I, 32 pts (29 men; mean age 56.8, range 33-75, mean BE 2.3 cm)were enrolled. Median procedure time was 24 min (IQR 20-35). Diary resultsshowed median symptom scores !10/100. At 12 mo (n Z 31; mean 1.8 sessions,range 1-2), CR for BE was achieved in 0%, 10%, 67% and 55% of pts treated at 6, 8,10 and 12 J/cm2, respectively, while the median BCR was 93%, 65%, 100% and 100%.There were no strictures and no buried glands in 1299 bx.In AIM-II, 70 pts (52 men,mean age 55.7 yrs, range 26-79, mean BE 3.2 cm) were enrolled. Median proceduretime was 28 min (IQR 24-33). Diary results showed median symptom scores ! 20/100and complete resolution by 5 d. At 12 mo (n Z 69; mean 1.5 sessions, range 1-2),CR for BE was achieved in 70% of pts (CR in 51% of pts if all cardia bx included inanalysis) and median BCR was 100%. There were no strictures and no buried glands in3007 bx. Conclusion: Circumferential RF ablation of NDBE using the HALO360

Ablation System can be performed with no subsequent strictures or buried glands,and with complete elimination of BE in 70% of patients at 1 yr (AIM-II, excluding

cardia bx). Residual BE, when present, was typically in the form of endoscopicallyvisible, small islands (!5 mm). AIM has been extended to 2-yr follow-up to collectlonger-term data and to enable ablation of residual BE using a novel focal ablationdevice incorporating similar electrode technology as HALO360.

S1508

Computer-Assisted Analysis of Brush Biopsies (EndoCDx)

Increases Detection of Barrett’s Esophagus and Dysplasia:

A Multicenter Prospective Clinical TrialJohn F. JohansonEndocdx Collaborative GroupIntroduction: The current limitations of screening and surveillance of Barrett’sesophagus (BE) include biopsy sampling error and interobserver variation inpathologic interpretation. In this multicenter prospective trial conducted at 8 sites,we evaluated computer assisted analysis of abrasive, transepithelial brush biopsiesas an adjunct to forceps biopsies to increase detection of BE and esophagealdysplasia. Methods: Patients over age 18 with suspected BE or known BEundergoing endoscopic surveillance for dysplasia were enrolled. Each patient had2 brush biopsies (BB) and then random 4 quadrant forceps biopsies (FB) for every1-2 cm of the abnormality. All BB were examined with computer assistance bypathologists at CDx Laboratories (Suffern, NY) and all FB were examined by theinvestigators’ local pathologists. Pathologists analyzing BB and FB samples weremasked from each other’s readings. Results: Of the total 1,183 patients enrolled,363 patients were diagnosed with BE by FB and 340 patients by BB but 144 of the340 patients diagnosed with BE by BB had negative FB results. The addition of BBto the standard FB protocol thus increased the overall detection of BE by 39.6%(95% confidence interval 33% - 47%). Dysplasia was diagnosed in 16 patients by FBand in 19 patients by BB. 14 of the 19 patients diagnosed with dysplasia hadnegative FB results. The addition of BB to the standard FB protocol thus increasedthe detection of esophageal dysplasia by 87.5% (95% confidence interval 44% -156%). Conclusion: These results suggest that adjunctive computer-assisted analysisof an abrasive brush biopsy has the potential to overcome the current limitations ofbiopsy surveillance and substantially improves detection of Barrett’s esophagus anddysplasia.

Increased Detection of Barrett’s Esophagus and Dysplasia with Brush Biopsy(n Z 1,183)

ForcepsBiopsy

Forceps Biopsy CBrush Biopsy

IncreasedDetection

Barrett’s esophagus 363 507 40%)

Dysplasia 16 30 88%))Lower end of 95% CI Z 33%))Lower end of 95% CI Z 44%

S1509

The Use of Thrombin Injections in the Management of Bleeding

Gastric Varices - A Single Centre ExperienceJayapal Ramesh, Jimmy K. Limdi, Vikram Sharma, Ali Aboutwerat,Alistair J. MakinBackground: There is a relative dearth of literature on the definitive endoscopicmanagement of bleeding gastric varices. Variceal ligation with bands and detachablesnares; sclerosants; cyanoacrylate glue, and thrombin injections have been usedwith variable success to control bleeding and reduce the risk of re-bleeding. Aim: Toreport our experience with bovine thrombin injection for the treatment of bleedinggastric varices. Methods: Forty-two cases of gastric varices were identified from ourendoscopy database between July 1998 and July 2003. Thirteen patients hadthrombin injection and data was extracted by retrospective case note analysis.Results: Ten were male and the median age was 55 yrs (range Z 31-74). All 13patients presented with acute bleeding. Ten had gastro-oesophageal varices type 2;two with isolated gastric varices and one gastro-oesophageal varix type 1. Elevenpatients had fundal gastric varices and none were thought to have bled fromoesophageal varices. Twelve of these had moderate to large varices. The underlyingdiagnoses were cirrhosis in nine; hepatocellular carcinoma in two and one each hadportal vein thrombosis and Budd-Chiari syndrome. The cause of cirrhosis wasalcohol in seven, combination of alcohol and hepatitis C in the remaining two. TheChild Pugh classification was grade A in seven; B in four and C in two cases. Patientsreceived one to four sessions (mean Z 2.1) of thrombin with a mean total dose of10.8 mls (range Z 5-42 mls) for variceal eradication. Two patients, in addition tothrombin injection, were also treated with endoscopic variceal ligation. One patientcontinued to bleed needing TIPSS as rescue procedure. Haemostasis in the acutesetting was successful in the rest (92%). The patient with hepatocellular carcinomadied within 30 days and four more patients died after a median follow up of 22months (range 12 - 80), none due to bleeding. There was no re-bleeding in theremainder at follow up (median 25 months; range 8-43). Conclusion: In our series,injection with thrombin proved to be an effective endoscopic treatment in themajority of patients with bleeding gastric varices. The overall mortality, aftercontrolling bleeding, was 38% (5/13) subsequent to a mean follow-up of 28 months.

Abstracts

www.giejournal.org Volume 63, No. 5 : 2006 GASTROINTESTINAL ENDOSCOPY AB127