Long-term Outcome of Patients With Obscure Gastrointestinal Bleeding Investigated by Double-Balloon Endoscopy

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2010;8: Long-term Outcome of Patients With Obscure Gastrointestinal Bleeding Investigated by Double-Balloon Endoscopy SATOSHI SHINOZAKI, HIRONORI YAMAMOTO, TOMONORI YANO, KEIJIRO SUNADA, TOMOHIKO MIYATA, YOSHIKAZU HAYASHI, MASAYUKI ARASHIRO, and KENTARO SUGANO Department of Medicine, Division of Gastroenterology, Jichi Medical University, Tochigi, Japan BACKGROUND & AIMS: It is often difficult to determine the cause of obscure gastrointestinal bleeding (OGIB). We evaluated the diagnostic yield and long-term outcome of patients with OGIB by using double-balloon endoscopy (DBE). METHODS: In this large, retrospective cohort study, DBE was performed in 200 consecutive patients with OGIB. Follow-up data were available from 151 patients for 29.7 months (range, 6 78 months), and clinical outcome was assessed. RESULTS: DBE detected bleeding sources in 155 of 200 patients (77.5%). The most frequent source detected was small intestine ulcers/erosions (64 patients). Patients who underwent DBE within 1 month after the last episode of overt bleeding had a better yield of positive findings than those who did not (84%, 107/128 patients vs 57%, 24/42; P.002). The overall rate of control of OGIB was 64% (97/151 patients). Patients with vascular lesions of the small intestine had a significantly lower rate of control of OGIB than those with other small intestine lesions (40%, 12/30 patients vs 74%, 52/70; P.001). A requirement for a large transfusion before DBE (P.012), multiple lesions (P.010), and suspicious (not definite) lesions (P.038) each significantly increased the likelihood of overt rebleeding in patients with vascular lesions of the small intestine. CONCLUSIONS: DBE is useful for the diagnosis of patients with OGIB and should be performed as soon as possible after overt OGIB. Patients with vascular lesions of the small intestine should be followed with particular care. View this article s video abstract at Obscure gastrointestinal bleeding (OGIB) is defined as bleeding of unknown origin that persists or recurs after a negative esophagogastroduodenoscopy (EGD) and colonoscopy. Although OGIB accounts for 5% of all cases of gastrointestinal bleeding, 1 it is often difficult to locate the source of bleeding in patients with it; frequent blood transfusions are thus often required, and OGIB might prove life-threatening. Capsule endoscopy (CE) was recently developed for small intestine exploration 2 and has markedly increased the diagnostic yield in patients with OGIB to around 63% (35% 76%). 1 CE can provide an endoscopic view of most of the small intestine mucosa, although diagnosis with it depends on analysis of recorded images, and neither biopsy nor endoscopic treatment can be performed. Yamamoto et al 3 were the first to report total enteroscopy with double-balloon endoscopy (DBE), and the usefulness of this method in the diagnosis and treatment of small bowel lesions has been documented. 4 7 The long-term outcome and diagnostic yield of DBE in patients with OGIB must be determined in a large cohort study to establish diagnostic and therapeutic guidelines. The aim of the present study was to assess the long-term outcome of patients who underwent DBE for OGIB. Patients and Methods Inclusion of Patients in the Study We prospectively registered the medical history, endoscopic findings, and clinical findings of all patients who had undergone DBE at Jichi Medical University Hospital in a database. Consecutive patients whose indication for DBE was OGIB were selected for this study from a total of 552 patients (1042 procedures) who underwent DBE from September 2000 to March Of the 552 subjects, 200 patients underwent DBE for investigation of OGIB. Methods of Obtaining Clinical Information In addition to the information in the database, medical records were retrospectively reviewed in this cohort study. Follow-up data were obtained from medical records at our hospital and from questionnaires collected by mail, , or fax from the doctors of other hospitals/clinics. We did not interview patients by phone. The questionnaires included date of final examination, changes in hemoglobin level, existence of overt bleeding, additional examinations, amounts of transfusions, amounts of iron replacements, date of transfusion, and last date of iron replacement. Of the 200 patients, 83 and 117 patients were followed in our hospital and in other hospitals/ clinics, respectively. The questionnaires for the 117 patients were sent to their doctors, and we received completed responses for 92 patients (79%). Categorization of Bleeding Pattern We divided the 200 patients into 3 groups by pattern of bleeding as follows. Ongoing overt bleeding was defined as melena or hematochezia within 24 hours before DBE. Previous overt bleeding was defined as at least 1 episode of melena or hematochezia before DBE, with a longer than 24-hour interval between the last episode of overt bleeding and DBE. Occult Abbreviations used in this paper: APC, argon plasma coagulation; CE, capsule endoscopy; DBE, double-balloon endoscopy; EGD, esophagogastroduodenoscopy; OGIB, obscure gastrointestinal bleeding; SD, standard deviation; TAE, transarterial embolization by the AGA Institute /10/$36.00 doi: /j.cgh

2 152 SHINOZAKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 2 bleeding was defined as persistent iron deficiency anemia with positive fecal occult blood test. Definition of Outcome To investigate outcome, we defined controlled OGIB as follows: (1) no further overt bleeding, (2) hemoglobin level not below 10 g/dl at final examination, and (3) no transfusion and no iron replacement within the 6-month period preceding the final examination. If anemia could be clearly explained by conditions other than gastrointestinal bleeding, OGIB was considered controlled if the other criteria were met, even if hemoglobin level was below 10 g/dl. We defined uncontrolled OGIB as OGIB for which the above criteria were not met. Additional Double-Balloon Endoscopies During Follow-up In cases of overt rebleeding or persistent iron deficiency anemia with positive fecal occult blood test, another DBE was performed during the follow-up period on the basis of the patient s condition. The decision was made by the doctors in charge. Follow-up Period The follow-up period was defined as the time between the first DBE and the last date of medical examination. Double-Balloon Endoscopy: Procedure and Findings The endoscope and technique used for DBE have been described elsewhere. 3,4,8 The route of insertion was chosen on the basis of the location of the suspected lesion as estimated from clinical findings and results of prior examinations. Antegrade insertion was chosen when jejunal lesions were suspected, whereas retrograde insertion was chosen when ileal lesions were suspected. In the present study, DBE was performed by the antegrade route alone in 47 patients (24%), by the retrograde route alone in 49 patients (25%), and by both routes in 104 patients (52%). When a definite bleeding source was not observed with one route, the other route was used as well. When a bleeding source was clearly observed during the procedures, we did not always attempt total enteroscopy. Of the latter 104 patients, we attempted total enteroscopy in 84 and were successful in 64 (76%). The main cause of failure in the 20 patients without success (24%) was intestinal adhesions. No major complications of DBE procedures such as perforation or pancreatitis were noted in this study. Other Examinations In addition, 20 (10%), 87 (44%), and 104 (52%) of 200 patients underwent CE, red blood cell scintigraphy, and abdominal computed tomography, respectively, before DBE. Endoscopic Treatments For ulcerous lesions, we did not perform endoscopic treatment unless the ulcer was accompanied by visible bleeding vessels. For vascular lesions, we generally performed endoscopic hemostasis by using argon plasma coagulation (APC), electrocoagulation, or clip placement according to the type of lesion on the basis of the Yano Yamamoto classification. 5 The Yano Yamamoto classification is used for endoscopic findings of small intestine vascular lesions. In this classification, type 1, type 2, and type 3 lesions are considered to represent angioectasia, Dieulafoy s lesion, and arteriovenous malformation, respectively. In addition, type 1 and type 2 are subdivided into types 1a and 1b and types 2a and 2b according to lesion size (1a and 2a are punctulate lesions ( 1 mm), whereas types 1b and 2b are larger lesions (a few mm in size). 5 We used this endoscopic classification because histologic confirmation of angioectasia or Dieulafoy s lesion was not obtained for most of the vascular lesions. We basically attempted to treat all vascular lesions, but sometimes we left nonbleeding small vascular lesions (type 1a without oozing) untreated when a patient had too many lesions to treat in one session or had other lesions more likely to be responsible for bleeding. Categorization of Endoscopic Findings We categorized small bowel bleeding sources into 3 groups as follows: ulcers/erosions, vascular lesions, and tumors/polyps. Furthermore, we divided DBE findings into definite or suspicious, considering the likelihood of their being a bleeding source. In the case of ulcers/erosions, we defined ulcers more than 10 mm in diameter or bleeding lesion as definite and all other ulcers/erosions as suspicious. In the case of vascular lesions, we defined type 1a without bleeding as suspicious and all other vascular lesions as definite according to the Yano Yamamoto classification. 5 In the case of tumors/polyps, we defined lesions with an ulcer, lesions larger than 2 cm in diameter, and bleeding lesions as definite and all other lesions as suspicious. When 2 or more possible bleeding lesions were observed, we documented the lesions most likely to be responsible for bleeding in this study. Ultimately, no patients were determined to have 2 or more definite bleeding sources. Regarding small intestine ulcers/erosions, we grouped ulcers classified into known disease categories into known categories (ie, NSAID ulcer/erosion, Meckel s diverticular ulcer/erosion, Crohn s disease ulcer, and so forth) and ulcers of unknown disease category into the unknown category (ie, nonspecific ulcer/erosion). Furthermore, we did not include small findings that could not be considered bleeding sources, such as small lipomas, tiny nonbleeding polyps, Meckel s diverticulum without ulcers and lymphangiectasia, and other nonsignificant findings. Statistical Analysis Data with a normal distribution were compared by using Student paired t test, whereas categorical data were compared by using the 2 test. For correlation analysis, Pearson correlation coefficients were calculated. Follow-up data regarding rebleeding-free intervals were analyzed by means of Kaplan Meier survival curves and the log-rank test. Differences were considered significant when P.05. Results Investigation of Obscure Gastrointestinal Bleeding by Using Double-Balloon Endoscopy Demographic data for the patients undergoing DBE for OGIB are shown in Table 1. Despite the large range in age of patients, those in the fifth through seventh decades of age were predominant, and only 45 patients (23%) were younger than 40 years of age. Before DBE examination, hemoglobin level was increased by transfusion or iron replacement to stabilize vital signs. In total, 123 patients (62%) required

3 February 2010 OUTCOME OF PATIENTS WITH OGIB AFTER DBE 153 Table 1. Demographic Data of Patients Undergoing DBE for OGIB Overall Ongoing overt bleeding Previous overt bleeding Occult bleeding No. of patients Sex (male/female) 122/78 17/13 88/52 17/13 Median age (range), y 60 (11 88) 67 (12 88) 60 (11 88) 58 (13 72) Median length of bleeding history (range), mo 2.5 (1 382) 1 (1 137) 2.5 (1 157) 13 (1 382) Median no. of overt bleeding episodes (range) 2 (0 20) 3 (1 20) 2 (1 12) 0 No. of patients with blood transfusion Mean transfusion requirements (range), units 4.5 (0 40) 8.1 (0 30) 4.2 (0 40) 2.8 (0 20) No. of patients observed for 6 months transfusions, whereas 116 patients (58%) required iron replacement. Diagnostic Usefulness of Double-Balloon Endoscopy in Patients With Obscure Gastrointestinal Bleeding Details of DBE diagnosis are summarized in Table 2. DBE revealed positive findings in 155 (78%) of 200 patients. When positive findings were divided into definite and suspicious findings, they were detected by DBE in 124 (62%) and 31 (16%) patients, respectively, in total. The yields of definite findings in the 3 groups of bleeding pattern (ongoing overt, previous overt, and occult) were 83% (25 of 30 patients), 58% (81 of 140), and 60% (18 of 30), respectively. The yield of definite findings in the ongoing overt bleeding group was significantly higher than those in the other groups (ongoing overt vs previous overt, P.008; ongoing overt vs occult, P.045; and previous overt vs occult, P.829). Relationship Between Diagnostic Yield and Timing of Double-Balloon Endoscopy in Patients With Obscure Gastrointestinal Bleeding We also evaluated the interval between last bleeding and first DBE in patients with previous overt bleeding (Table 3), as described by Pennazio et al 9 in their CE study of OGIB. The yield of positive findings on DBE for previous overt bleeding significantly decreased with the length of this interval (correlation analysis: P.001) (Table 3). Patients who underwent DBE within 1 month after the last episode of overt bleeding had a significantly better yield than the other patients (84%, 107 of 128 patients vs 57%, 24 of 42 patients; P.002). Bleeding Pattern and Type of Lesions in the Small Intestine In 126 patients with small intestine lesions, the proportions of types of lesions were examined in relation to bleeding Table 2. DBE Diagnosis of Patients With OGIB Type of bleeding DBE diagnosis Overall Ongoing overt bleeding Previous overt bleeding Occult bleeding Small intestine lesion 126 (63%) 23 (77%) 83 (59%) 20 (67%) Ulcer/erosion NSAID ulcer/erosion Meckel s diverticular ulcer Crohn s disease 6 6 Blind loop ulcer Behcet disease Diverticulum-related duodenal ulcer in the third portion 2 2 Steroid-induced duodenal ulcer in the second portion 1 1 Anastomotic ulcer Schonlein Henoch ulcer 1 1 Pouch ulcer 1 1 Radiation enteritis 1 1 Nonspecific jejunal ulcer 1 1 Nonspecific ileal ulcer Vascular lesion Tumor Polyp Colonic lesion 12 (6%) 1 (3%) 9 (6%) 2 (7%) Esophageal or gastric lesion 16 (8%) 2 (7%) 12 (9%) 2 (7%) Biliary bleeding 1 (0.5%) 1 (3%) Negative 45 (22.5%) 3 (10%) 36 (26%) 6 (20%) Total

4 154 SHINOZAKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 2 Table 3. Diagnostic Yield of DBE Type of bleeding Definite Positive Suspicious Negative Ongoing overt bleeding 25 (83%) 2 (7%) 3 (10%) (n 30) Previous overt bleeding, 81 (58%) 23 (16%) 36 (26%) overall (n 140) 2 weeks (n 54) 36 (67%) 8 (15%) 10 (19%) 3 4 weeks (n 44) 28 (64%) 8 (18%) 8 (18%) 2 3 months (n 32) 13 (41%) 7 (22%) 12 (38%) 4 6 months (n 4) 2 (50%) 0 (0%) 2 (50%) 7 months (n 6) 2 (33%) 0 (0%) 4 (67%) Occult bleeding (n 30) 18 (60%) 6 (20%) 6 (20%) Total (n 200) 124 (62%) 31 (16%) 45 (23%) pattern (overt OGIB, 106 patients vs occult OGIB, 20 patients). Of small intestine lesions presenting with overt OGIB, ulcers/ erosions, vascular lesions, and tumors/polyps accounted for 55% (58 of 106 patients), 30% (32 of 106), and 15% (16 of 106) of lesions, respectively. In patients with occult OGIB, small intestine ulcers/erosions, vascular lesions, and tumors/polyps accounted for 30% (6 of 20 patients), 25% (5 of 20), and 45% (9 of 20) of lesions, respectively. Taken together, small intestine tumors/polyps were the most frequent cause of bleeding in the occult OGIB group (45%), although they were responsible for bleeding in only 15% of patients in the overt OGIB group (P.005). Bleeding Sources Outside the Small Intestine The 155 positive findings in this study included those in 29 patients (19%) with non small intestine lesions, including 16 esophagogastric lesions (9 ulcers, 5 vascular lesions, 1 polyp, 1 bleeding erosion), 12 colonic lesions (5 vascular lesions, 3 cases of colitis, 3 cancers, 1 case of bleeding diverticula), and 1 case of biliary bleeding. Retrospectively, it was found that 28 patients other than the single one with biliary bleeding had a bleeding source in a region detectable with conventional EGD or colonoscopy. However, 6 of 16 esophagogastric lesions and 3 of 12 colonic lesions had already been detected before DBE. Because the lesions detected had not been conclusively shown to be bleeding sources, DBE was performed as well on these 9 patients. These non small intestine lesions were considered definite bleeding sources on the basis of exclusion of small intestine lesions on DBE. Long-term Outcome of Patients With Obscure Gastrointestinal Bleeding Follow-up data during 6 months after DBE were obtained for 151 of 200 patients (76%); 75 of these 151 patients were followed at our hospital, whereas 76 patients were followed at other hospitals/clinics. In total, 49 of 200 patients were excluded from outcome analysis for the following reasons: (1) patient did not visit the hospital/clinic, and/or (2) physicians at other hospitals/clinics did not respond to the questionnaire. The frequencies of medical examination and blood tests were determined by the physicians in charge. The mean follow-up period for the 151 patients followed for more than 6 months was 29.7 months (range, 6 78 months). Of these 151 patients, 100 patients had small intestine lesions, 19 patients had non small intestine lesions, and 32 patients had negative findings for bleeding sources. We focused evaluation on the outcome of the 100 patients with small intestine lesions. Overall Outcome of 100 Patients With Small Intestine Lesions Overall, the proportion of patients with controlled OGIB in the small intestine was 61 of 100 (61%) (Table 4). On the basis of DBE diagnosis, the rate of control of patients with small intestine vascular lesions (40%, 12 of 30 patients) was significantly lower than that for other small intestine lesions (vascular lesions vs ulcers/erosions, 40% vs 65%, P.031; vascular lesions vs tumors/polyps, 40% vs 84%, P.002). Furthermore, the rate of control of patients with small intestine ulcers/erosions (65%, 33 of 51 patients) was lower than that of those with small intestine tumors/polyps (84%, 16 of 19), although not to a statistically significant extent (P.113). Outcomes for specific small bowel lesions were as follows. Small Intestine Ulcers/Erosions Follow-up data were available for 51 of 64 patients (80%) with small intestine ulcers/erosions (Table 5). Eighteen of the 51 patients (35%) were categorized as uncontrolled. First, we examined the impact of known categories diagnosed by DBE. Thirty-six patients (26 definite and 10 suspicious) were diagnosed with ulcers/erosions of known categories and 15 patients (11 definite and 4 suspicious) as having lesions of unknown category. We compared the clinical outcome of these 2 groups. Patients with ulcers/erosions of known categories exhibited a significantly higher rate of control of OGIB than those with ulcers/erosions of unknown category (75%, 27 of 36 patients vs 40%, 6 of 15 patients, P.017). Of the 18 patients with uncontrolled OGIB, follow-up DBE was performed in 8 for investigation of recurrent overt bleeding, and DBE findings suggested bleeding sources in the same location in 7 of these 8 patients (88%). To elucidate the clinical impact of DBE on the 51 patients, we examined the changes in transfusion requirements, iron replacement requirements, and hemoglobin level before to after DBE. The number of patients requiring transfusion significantly decreased from 29 to 1 (P.001) and the number requiring iron replacement from 26 to 4 (P.001), whereas hemoglobin level significantly increased from g/dl (mean standard deviation [SD]) (range, g/dl) to g/dl (range, g/dl) (P.001). The overall rate of control was 65% (33 of 51 patients). Small Intestine Vascular Lesions Follow-up data were available for 30 of 37 patients (81%) with small intestine vascular lesions (Table 5). Of the 30 patients, 18 patients (60%) were categorized as uncontrolled, although treatment was performed in most of the patients (26 Table 4. Overall Outcome of 100 Patients With Small Intestine Lesions Type of lesions Control rate of OGIB Ulcers/erosions 65% (33/51) P.031 Vascular lesions 40% (12/30) Tumors/polyps 84% (16/19) P.002 Overall 61% (61/100)

5 February 2010 OUTCOME OF PATIENTS WITH OGIB AFTER DBE 155 Table 5. Long-term Outcome of 100 Patients With Small Intestine Lesions DBE findings Treatment Outcome Controlled Uncontrolled Ulcers/erosions Definite: 37 Known categories: 26 NSAID ulcer: 5 Long-term discontinuation of NSAIDs: No long-term discontinuation of NSAIDs: Meckel s diverticular ulcer: 4 Resection: Observation: Crohn s disease ulcer: 3 Steroid and mesalamine: Blind loop ulcer: 5 Resection: Metronidazole: Behcet ulcer: 3 Steroid and mesalamine: Diverticulum-related duodenal ulcer: 1 Proton pump inhibitor: Steroid-induced duodenal ulcer: 1 Clipping: Anastomotic ulcer: 2 Resection: 1, proton pump inhibitor: Schonlein Henoch ulcer: 1 Steroid: Ileal pouch ulcer: 1 Surgical suture: Unknown category: 11 Multiple jejunal ulcers: 1 Resection: Multiple ileal ulcers: 2 Mesalamine and levofloxacin: 1, anti-tuberculosis therapy: 1 a 2 0 Multiple and annular ileal ulcers: 5 Resection: 2, mesalamine: Observation: Solitary ileal ulcers: 3 Resection: 1, TAE: 1, b anti-tuberculosis therapy: 1 a 3 0 Suspicious: 14 Known categories: 10 NSAID erosion and small ulcer: 7 NSAID discontinuation: Meckel s diverticular ulcer: 1 Resection: Diverticulum-related duodenal ulcer: 1 Observation: Radiation enteritis: 1 Observation: c Unknown category: 4 Unclassified ileal ulcer: 4 Observation: 1, clipping: 1, metronidazole: 1, mesalamine: Total: 51 (1) Further overt rebleeding (yes/no), 17 (33%)/34 (67%) 33 (65%) 18 (35%) (2) Requiring iron/transfusions, 4 (8%)/1 (2%) Vascular lesions Definite: 20 Type 1a with oozing: 2 Clipping: 1, electrocoagulation: Type 1b with oozing: 3 APC: 1, electrocoagulation: Type 1b without oozing: 11 APC: 8, electrocoagulation: 2, clipping: Type 2a: 3 Clipping: Varix: 1 Resection: d Suspicious: 10 Type 1a without oozing: 10 APC: 5, electrocoagulation: 1, observation: Total: 30 (1) Further overt rebleeding (yes/no), 16 (53%)/14 (47%) 12 (40%) 18 (60%) (2) Requiring iron/transfusions, 9 (30%)/5 (17%) Tumors/polyps Definite: 16 Gastrointestinal stromal tumor: 7 Resection: Adenocarcinoma: 2 Resection: Hemangioma: 2 Resection: Lymphangioma with bleeding: 1 Lipid restriction: Large cell carcinoma: 1 Resection: Diffuse large B-cell lymphoma: 2 Chemotherapy: 1, chemoradiotherapy: Ileal submucosal tumor with ulcer: 1 Observation: Suspicious: 3 Non-neoplastic polyp: 2 Snare polypectomy: e Lymphangioma: 1 Endoscopic mucosal resection: Total: 19 (1) Further overt rebleeding (yes/no), 3 (16%)/16 (84%) (2) Requiring iron/transfusions, 2 (11%)/1 (5%) 16 (84%) 3 (16%) a Empiric therapy. No evidence of intestinal tuberculosis. b Subsequent angiography showed massive bleeding from the jejunal branch of superior mesenteric artery treated by TAE. c Rebled from the small intestine perforation, resulting in emergency operation. d Rebled from the postoperative anastomotic ulcer. e Rebled from the type 2a vascular lesion.

6 156 SHINOZAKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 2 of 30). All definite lesions were treated endoscopically or surgically. The patients with definite small intestine vascular lesions exhibited a significantly better outcome than those with suspicious lesions (55%, 11 of 20 patients vs 10%, 1 of 10; P.017). Furthermore, endoscopically treated definite lesions exhibited a better outcome than those with endoscopically treated suspicious lesions, although not to a statistically significant extent (58%, 11 of 19 patients vs 17%, 1 of 6; P.078). The bleeding in all 4 patients observed was not controlled. Of the 18 patients with uncontrolled OGIB, another DBE was performed during the follow-up period in 11 for investigation of recurrent overt bleeding. Follow-up DBE revealed definite vascular lesions in 10 of the 11 patients (91%) that were considered different from the primary lesions (type 1a with oozing in 2 patients, type 1b without oozing in 1, type 2a in 5, type 2b in 1, and postoperative anastomotic ulcer in 1). In particular, 5 of 9 patients with uncontrolled suspicious lesions underwent DBE, and 4 of the 5 patients (80%) exhibited definite vascular lesions (type 1a with oozing in 2 patients and type 2a in 2). To assess the clinical impact of DBE on the 30 patients, we examined changes in transfusion requirements, iron replacement requirements, and hemoglobin level before to after DBE. The number of patients requiring transfusion significantly decreased from 23 to 5 (P.001) and the number requiring iron replacement from 19 to 9 (P.009), whereas hemoglobin level significantly increased from g/dl (mean SD) (range, g/dl) to g/dl (range, g/dl) (P.001). The overall rate of control was 40% (12 of 30 patients). As indicated above, patients with small intestine vascular lesions as a source of OGIB had a significantly lower rate of control of OGIB than those with other types of small intestine lesions. We therefore attempted to determine predictors of overt rebleeding in patients with small intestine vascular lesions among age, sex, type of OGIB, amount of transfusion, length of history of bleeding, anticoagulant therapy (warfarin), platelet count, hemodialysis, number of vascular lesions, and type of vascular lesions. Three significant predictors were found (Table 6). One was the amount of transfusion before DBE. Patients with no overt rebleeding after DBE required units (mean SD) (range, 0 10 units) before DBE, whereas those with overt rebleeding required units (range, 0 30 units) (P.012). The second predictor was recognition of Table 6. Factors Assessed for Predicting Overt Rebleeding in Patients With Small Intestine Vascular Lesion Factor Overt rebleeding, none (n 14) vs 1 or more (n 16) P value Age.25 Sex.07 Type of OGIB.24 Amount of transfusion before DBE.012 Length of bleeding history.51 Anticoagulant drug (warfarin).78 Platelet count.90 Hemodialysis.11 Number of lesions (solitary vs multiple).010 Type of lesions (definite vs suspicious).038 multiple lesions on DBE, compared with solitary lesions. Six of 14 patients with no rebleeding and 13 of 15 with overt rebleeding had multiple lesions (P.010). The third predictor was suspicious vascular lesions on DBE. Eight of 20 patients with definite lesions and 8 of 10 with suspicious lesions experienced overt rebleeding (P.038). These findings indicate that large amount of transfusion before DBE, multiple vascular lesions, and suspicious lesions are significant predictors of overt rebleeding after DBE in patients with small intestine vascular lesions. Small Intestine Tumors/Polyps Follow-up data were available for 19 of 25 patients (76%) with small intestine tumors/polyps (Table 5). Of the 19 patients, 3 patients were categorized as uncontrolled. The OGIB in 2 patients with diffuse large B-cell lymphoma was uncontrolled despite chemoradiotherapy. A patient with non-neoplastic polyp treated by snare polypectomy exhibited overt rebleeding 122 days after DBE, and subsequent DBE revealed a type 2a vascular lesion in the jejunum, which was treated by clipping. To examine the clinical impact of DBE on the 19 patients, we examined the changes in transfusion requirements, iron replacement requirements, and hemoglobin level before to after DBE. The number of patients requiring transfusion significantly decreased from 10 to 1 (P.001) and the number requiring iron replacement from 15 to 2 (P.001), whereas hemoglobin level significantly increased from g/dl (mean SD) (range, g/dl) to g/dl (range, g/dl) (P.001). The overall rate of control was 84% (16 of 19 patients). Patients With Negative Double-Balloon Endoscopy Follow-up data were available for 32 of 45 patients (71%) with negative DBE. In 12 of the 32 patients (37%), bleeding was uncontrolled. Although 4 of the 12 patients underwent another DBE for recurrent bleeding, DBE revealed no bleeding sources in them. In 1 of these 12 patients, angiography after DBE revealed pseudoaneurysm of the right hepatic artery, which was treated by transarterial embolization (TAE). Because massive overt bleeding was stopped with this intervention, the bleeding was considered to have resulted from aneurysmal penetration into the bile duct. The overall rate of control was 63% (20 of 32 patients). Pattern of Rebleeding After Treatment of Small Intestine Lesions To clarify the pattern of rebleeding after treatment of small intestine lesions, we evaluated the rebleeding-free intervals in 4 groups (negative, treated definite tumors/polyps, treated definite ulcers/erosions, and treated definite vascular lesions). The rebleeding-free interval was defined as the interval between first DBE and first overt bleeding after DBE (Figure 1). We focused on rebleeding-free intervals for treated definite lesions by excluding suspicious and non-treated lesions, because suspicious and non-treated lesions included both lesions that did not require treatment and lesions that could not be treated. As a control, we also analyzed findings for patients with negative findings on DBE. As indicated in Figure 1, patients with treated definite vascular lesions had a shorter rebleedingfree interval than those with other lesions (vascular lesions vs

7 February 2010 OUTCOME OF PATIENTS WITH OGIB AFTER DBE 157 Figure 1. Rebleeding-free interval after DBE. tumors/polyps, P.019; vascular lesions vs ulcers/erosions, P.075; vascular lesions vs negative, P.074, log-rank test). One of 15 patients with treated definite tumors/polyps rebled; this patient had diffuse large B-cell lymphoma treated by chemotherapy. None of the patients with definite treated tumors/ polyps that were resected exhibited rebleeding. In the treated definite vascular lesion group, approximately 40% of patients experienced rebleeding within 1 year after DBE. These findings suggest that patients with definite vascular lesions of the small intestine were prone to rebleed within 1 year after DBE even after treatment. Discussion We have reported here the findings of a large cohort study of DBE diagnosis and long-term outcome of patients with OGIB after DBE. The DBE diagnosis of 200 patients and long-term clinical outcomes of 151 patients who had been followed for 29.7 months on average were determined. To the best of our knowledge, both this length of follow-up and the number of patients followed are higher than in any previous reports on the clinical outcome of patients with OGIB after DBE. 7,10,11 We found that the high diagnostic yield of and possibility of endoscopic intervention with DBE contributed to determination of therapeutic strategy and improved the longterm outcome of patients. DBE yielded positive findings suggestive of bleeding sources in 155 of 200 (78%) patients. This yield of positive findings was not less than in other studies (41% 80%, mean 65%) of OGIB with DBE. 1 Overall, on DBE diagnosis of small intestine lesions in this study, ulcers/erosions were found to comprise 51% of small intestine bleeding sources. Although European DBE studies of OGIB found that vascular lesions were the most common cause of OGIB, 12 ulcerous lesions were the most frequent bleeding source in OGIB in the present study and other Japanese DBE studies as well. 10,11 Ethnic differences might explain this difference in findings. In the present study, small intestine tumors/polyps were the most frequent cause of bleeding in patients with occult OGIB. Raju et al 1 also reported that younger patients (younger than 40 years of age) with OGIB are prone to have tumor lesions. Our findings and the report by Raju et al suggest that small intestine tumors/polyps should be considered in patients with occult OGIB and/or in younger patients, and that contrast-enhanced computed tomography scan might be useful in searching for mass lesions and determining the route of insertion of DBE in such cases. To determine the appropriate timing of DBE in patients with overt bleeding, we evaluated the interval between last overt bleeding and first DBE in patients with overt OGIB. The yield of positive findings of DBE for previous overt OGIB was higher in those with a shorter interval than in those with a longer interval (Table 3). The yield of definite findings of DBE for ongoing overt OGIB was significantly higher than for previous overt OGIB (P.008). Furthermore, patients who underwent DBE within 1 month after the last episode of overt bleeding had a significantly better yield of positive findings than the other patients (P.002). DBE should thus be performed within 1 month after last overt bleeding and preferably within 24 hours. We examined the clinical impact of DBE diagnosis of small intestine ulcers/erosions. Patients with small intestine ulcers/ erosions of known categories had a significantly better outcome than patients with ulcers/erosions of unknown category (P.017). Diagnosis of ulcers/erosions of known categories by DBE might thus enable specific treatment resulting in improving outcomes. It is also necessary to elucidate the pathogenesis of small intestine ulcers of unknown category and to develop the specific treatments for them. In the present study, small intestine vascular lesions had a lower rate of control of OGIB than other small intestine lesions, despite endoscopic treatment. In the present study, suspicious lesions had a lower rate of control than definite lesions. We considered some possible reasons for the lower rate of control in the patients with suspicious lesions as follows. (1) Patients with suspicious lesions had higher tendency of multiple lesions. Nine of the ten (90%) patients suspicious lesions were multiple. (2) Although all the definite lesions were treated, some of the suspicious lesions were not treated: four of the ten (40%) patients with suspicious lesions were not endoscopically treated. (3) Definite lesions could be overlooked by the first DBE in some of the patients with suspicious lesions. Actually, 5 of 10 patients with suspicious lesions underwent another DBE dur-

8 158 SHINOZAKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 8, No. 2 ing follow-up because of the overt rebleedings, which revealed definite vascular lesions in 4 of the 5 patients. Careful follow-up is thus required when only suspicious vascular lesions are detected as bleeding sources. In some patients, even if lesions responsible for bleeding are treated, other vascular lesions can bleed later. In fact, 2 of 3 type 2a lesions that were thought to represent Dieulafoy s lesions rebled even after successful endoscopic hemostasis with clip placement. Follow-up DBE at rebleeding suggested bleeding from a different site. Because small intestine vascular lesions are frequently multiple and we were unable to perform systemic treatment for multiple lesions of this type, the usefulness of local endoscopic treatment in a single session for them might be limited. However, frequency of overt bleeding and amount of transfusion could have decreased after endoscopic hemostasis with DBE. Even if rebleeding occurs after hemostasis, repeated hemostasis with DBE might improve management of refractory OGIB. To confirm this, a large prospective study of endoscopic intervention for small intestine vascular lesions is needed. A US cohort study of OGIB after DBE whose length of follow-up was similar to the present study was reported recently. 13 Similar to our study, it also showed that patients with small intestine vascular lesions are more likely to report recurrent hemorrhage than other lesions, despite endoscopic interventions. However, the rates of patients with small intestine ulcers/erosions and tumors/polyps were higher in the present study than in the US cohort. The present study has some limitations. First, we did not assess the outcome of patients who did not undergo DBE for OGIB as a control. Second, because we treated all of the patients with definite vascular lesions, we did not assess the outcome of patients with non-treated definite vascular lesions. However, for ethical reasons it is difficult to leave definite vascular lesions untreated. In conclusion, DBE is useful for diagnosis of OGIB. Furthermore, DBE diagnosis and subsequent intervention yield good long-term outcomes. To maximize the diagnostic yield of DBE for overt bleeding, DBE should be performed as soon as possible, preferably within 1 month after bleeding. Careful follow-up is needed for patients with small intestine vascular lesions as sources of bleeding, and large amount of transfusion, multiple lesions, and suspicious lesions can be considered strong risk factors for rebleeding. Supplementary data To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi: / j.cgh References 1. Raju GS, Gerson L, Das A, et al. American Gastroenterological Association (AGA) Institute medical position statement on obscure gastrointestinal bleeding. Gastroenterology 2007;133: Iddan G, Meron G, Glukhovsky A, et al. Wireless capsule endoscopy. Nature 2000;405: Yamamoto H, Sekine Y, Sato Y, et al. Total enteroscopy with a nonsurgical steerable double-balloon method. Gastrointest Endosc 2001;53: Yamamoto H, Kita H, Sunada K, et al. Clinical outcomes of double-balloon endoscopy for the diagnosis and treatment of small-intestinal diseases. Clin Gastroenterol Hepatol 2004;2: Yano T, Yamamoto H, Sunada K, et al. Endoscopic classification of vascular lesions of the small intestine (with videos). Gastrointest Endosc 2008;67: Shinozaki S, Yamamoto H, Ohnishi H, et al. Endoscopic observation of Meckel s diverticulum by double balloon endoscopy: report of five cases. J Gastroenterol Hepatol 2008;23:e308 e Sun B, Rajan E, Cheng S, et al. Diagnostic yield and therapeutic impact of double-balloon enteroscopy in a large cohort of patients with obscure gastrointestinal bleeding. Am J Gastroenterol 2006; 101: Yamamoto H, Sugano K. A new method of enteroscopy: the double-balloon method. Can J Gastroenterol 2003;17: Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology 2004;126: Ohmiya N, Yano T, Yamamoto H, et al. Diagnosis and treatment of obscure GI bleeding at double balloon endoscopy. Gastrointest Endosc 2007;66:S72 S Manabe N, Tanaka S, Fukumoto A, et al. Double-balloon enteroscopy in patients with GI bleeding of obscure origin. Gastrointest Endosc 2006;64: Heine GD, Hadithi M, Groenen MJ, et al. Double-balloon enteroscopy: indications, diagnostic yield, and complications in a series of 275 patients with suspected small-bowel disease. Endoscopy 2006;38: Gerson LB, Batenic MA, Newsom SL, et al. Long-term outcomes after double-balloon enteroscopy for obscure gastrointestinal bleeding. Clin Gastroenterol Hepatol 2009;7: Reprint requests Address requests for reprints to: Hironori Yamamoto, MD, PhD, Department of Medicine, Division of Gastroenterology, Jichi Medical University, Yakushiji, Shimotsuke, Tochigi , Japan. yamamoto@jichi.ac.jp; fax: (81) Conflicts of interest The authors disclose the following: Dr Yamamoto has applied for a patent in Japan for the double-balloon endoscopy technique described in this article. The remaining authors disclose no conflicts.