Saperas E, Dot J, Videla S, et al. (Barcelona, Spain). Capsule endoscopy versus computed tomographic or standard angiography for the diagnosis of obscure gastrointestinal bleeding. Am J Gastroenterol 2007;102:731–737.Obscure gastrointestinal (GI) hemorrhage is defined as the presence of hematochezia, melena, or iron-deficiency anemia with or without guaiac-positive stool testing and no detected source of blood loss on conventional upper endoscopy, colonoscopy, or small bowel series. The most common etiology of obscure blood loss in patients >40 years old is arteriovenous malformations of the small intestine, vascular lesions that can be causal in up to 30%–40% of patients, but are not detected on standard imaging studies, including small bowel series and computed tomographic scans (Br Med J 1984;288:1663–1665). Less common etiologies for blood loss include ulcerations, neoplasia diverticulae, and other vascular disorders, including Dieulafoy’s lesions and hemorrhage from manifestations of portal hypertension in the small intestine.Endoscopic imaging tests available for small bowel imaging include push enteroscopy, which can visualize mucosa 50–100 cm distal to the ligament of Treitz (Endoscopy 1995;27:164–170). Video capsule endoscopy (CE), developed by Given Imaging (Given Imaging Limited, Yoqneam, Israel) entails swallowing a capsule measuring 11 × 26 mm that subsequently obtains images throughout the GI tract and transmits the data during an 8-hour recording period via radio frequency to a recording device. Multiple studies have demonstrated the superiority of CE compared to push enteroscopy for patients with obscure GI bleeding. In patients presenting with obscure bleeding, the yield of CE has ranged from 55%–75% with a 25%–50% higher yield compared with push enteroscopy (Endoscopy 2002;34:685–689; Aliment Pharmacol Ther 2004;20:189–194). Double balloon enteroscopy (DBE) allows complete visualization of the small intestine using a 200-cm enteroscope with an outer diameter of 8.5 mm equipped with a 140-cm overtube with an outer diameter of 12 mm (Fujinon Inc, Saitama, Japan; Gastrointest Endosc 2001;53:216–220). With the advent of these new technologies, intraoperative enteroscopy has been reserved for patients who fail these examinations or have adhesions requiring lysis, potentially via a laparoscopic approach, so that DBE can be performed (Gastrointest Endosc 2006;64:984–988).Imaging modalities for GI hemorrhage include technetium-99m–labeled red blood cell scans and angiography. The utility of these modalities relies on the frequency of bleeding, with highest rates found if bleeding rates approximate 1.0 mL/min. Although tagged red blood cell scans can detect sources bleeding at a rate of 0.5 mL/min, their ability to locate lesions remains limited (Nucl Med Commun 2002;23:591–594). Angiography offers the potential for rapid treatment of identified bleeding lesions. In most patients, the current technique is to perform selective mesenteric angiographic examination with selective embolotherapy with either microcoils, gelatin sponge pledglets, or polyvinyl alcohol embolospheres (Cardiovasc Intervent Radiol 2006;29:49–58).In this prospective study performed by Saperas et al in Spain, the goal was to compare the yield of CE to that of angiography in patients presenting with obscure GI bleeding. Enrolled patients with obscure GI bleeding underwent initial computed tomographic angiography (CTA) followed by selective mesenteric angiography and CE performed independently by blinded investigators within 7 days. Twenty-eight patients with obscure bleeding (20 with overt bleeding and 8 with occult bleeding) were enrolled.CTA was performed initially when the patients were hemodynamically stable after either fluid or blood resuscitation. Selective angiography of the visceral trunks (celiac axis, superior mesenteric, and inferior mesenteric arteries) was performed subsequently. Patients then underwent preparation for video CE using 2 liters of polyethylene glycol preparation.The mean age of the enrolled patients was 74 years; 57% were men, and 82% with at least one comorbid condition. Of the cohort, 89% had required blood transfusions, 36% were using anticoagulant therapy, and 18% antiplatelet therapy. CTA and angiography were performed in 86% owing to contrast allergy in 1 patient and renal failure in 2 patients. CTA identified a bleeding source in 6 of 25 patients (24%; 95% confidence interval [CI], 9.4%–45.1%), including angiodysplasia (n = 2), jejunal tumor (n = 1), aortoenteric fistula (n = 1), ischemic enteritis (n = 1), and jejunal diverticula (n = 1). Mesenteric angiography identified an actively bleeding source or an obvious vascular abnormality suggestive of a bleeding site in 14 of 25 patients (56%; 95% CI, 34.9%–75.6%). Angiodysplasia was the bleeding source identified in 13 of 14 (95%) patients. Angiography also established the diagnosis in one additional patient diagnosed with small bowel stromal tumor. CTA identified 2 of 13 (15%) angiodysplasias depicted on standard angiograms. Among patients with normal CTA examinations, location of bleeding was revealed by conventional angiography in 11 of 19 (60%) patients, whereas conventional angiography was negative in 3 of the 6 (50%) patients in whom CTA was positive. No complications occurred after CTA or conventional visceral angiography.Wireless CE was able to be performed in all patients (applicability 100%) without any complications. Average gastric and small bowel transit times for the capsule were 19 (range, 2–102) minutes and 262 (range, 63–411) minutes, respectively. Failure to visualize the cecum occurred in 3 (11%) patients and the quality of the images was considered to be good or very good in all cases.CE identified a bleeding source in 19 of the 28 patients (68%; 95% CI, 47.7%–84.1%) with angiodysplasia being identified in 43% of patients. There was no difference between patients with obscure overt bleeding (70%) or patients with obscure-occult bleeding (63%; P = 1.0). In the 25 patients actually tested, CE identified a bleeding source in 72% of patients (95% CI, 50.6%–87.9%) compared with 24% for CTA (95% CI, 9.4%–45.1%; P = .005 vs CE), 56% for angiography (95% CI, 34.9%–75.6%; P = .29), and 68% for CTA-angiography (95% CI, 46.5%–85.1%; P = 1.00). Similarly, CE was able to diagnose 100% (6 of 6) of patients diagnosed by CTA and 86% (12 of 14) of patients diagnosed by angiography. Moreover, CE was positive in 63% (12 of 19) of negative cases on CTA, in 55% (6 of 11) of negative cases on angiography, and in 36% (3 of 7) of those negative on both tests. Angiography was diagnostic in 2 cases (22%) where CE was normal. The agreement between CE and the 2 other techniques was 52% (P = .03).As a result of the CE findings, therapeutic intervention was undertaken in 9 of 19 (47%) patients with positive findings. In 1 patient with jejunoileal ulcers diagnosed with Crohn’s disease, specific medical therapy was started; 4 patients with angiodysplasia underwent argon beam coagulation, which stopped hemorrhage, and 4 patients underwent surgery (1 patient with jejunal tumor was diagnosed with a gastrointestinal stromal tumor, 1 with jejunal diverticular bleeding, 1 with segmental ischemic enteritis, and 1 with an aortoenteric fistula).CommentApproximately 5% of patients with GI hemorrhage will have no source found on conventional upper and lower endoscopy and will be classified with obscure bleeding (Am J Surg 1992;163:90–92). Although obscure bleeding is uncommon, diagnostic evaluation has traditionally been prolonged and associated with considerable expense (Gastrointest Endosc 1990;36:337–341). The introduction of CE and DBE have dramatically enhanced the clinician’s ability to diagnose and treat patients with small bowel lesions. Novel radiographic imaging techniques include helical CT-enteroclysis and helical CT angiography. Helical CT–enteroclysis combines standard enteroclysis to distend the small bowel followed by helical CT (Clin Radiol 2006;61:31–39). Case reports have shown the ability of helical CT to identify small and large intestinal bleeding angioectasias (Clin Imaging 2000;24:61–63). In a report of 22 patients with obscure GI bleeding, CT enteroclysis was found to be inferior to CE in the detection of potential bleeding lesions such as angioectasia in the small bowel (Endoscopy 2003;35:1009–1014).CTA involves catheterization of the abdominal aorta followed by helical CTA before and after intra-arterial injections of a contrast medium. In a prospective study of 18 patients with bleeding colonic angioectasias, the sensitivity, specificity, and positive predictive value of helical CT angiography were 70%, 100%, and 100%, respectively, when compared with a gold standard of colonoscopy and mesenteric angiography (Gastroenterology 2000;119:293–299).The data on the clinical utility of angiography in the specific setting of obscure GI bleeding have been limited. Although prior studies have clearly shown an improved diagnostic accuracy for CE compared with push enteroscopy (Gut 2003;52:1122–1126), there has been limited information to date regarding yield of angiography and comparison to CE. The current study by Saperas et al demonstrates that CE is indeed superior to initial angiographic examination for patients with obscure bleeding. Although the authors state that the angiography occurred once patients were stabilized, we are not provided with information regarding the timing of the angiography in relation to the last bleeding episode. It could be that angiography would have a higher yield if performed only in patients within a few hours of overt bleeding. Nonetheless, initial CE appears to be associated with a higher diagnostic yield for the majority of patients.Interim results from a randomized controlled trial being performed in Hong Kong comparing CE with mesenteric angiography were presented at Digestive Disease Week 2007 (Gastrointest Endosc 2007;AB). Results were presented from 41 patients who were randomized, 20 to the capsule and 21 to the angiographic examination. Nineteen patients in each arm completed the study. Overall the diagnostic yield was 55% for CE compared with 9.5% for angiography (P = .003), supporting the findings from the Saperas study.In summary, initial video capsule endoscopy appears to be superior to angiography for the immediate investigation of patients with active obscure overt or occult bleeding. CE has been demonstrated to be superior to push enteroscopy and other radiographic imaging techniques for small bowel lesions. Whether DBE is associated with equivalent or superior diagnostic yield compared with CE is an area of active investigation. Both CE and DBE have been shown to have miss rates between 20% and 30% (Gastrointest Endosc 2006;64:740–750) and may therefore be complementary in the diagnosis and management of small bowel lesions. Given recent evidence, angiography may be relegated to a back seat for patients with obscure GI hemorrhage, but should still be considered in patients with ongoing bleeding who are unable to be diagnosed and treated after initial CE or DBE examinations. Saperas E, Dot J, Videla S, et al. (Barcelona, Spain). Capsule endoscopy versus computed tomographic or standard angiography for the diagnosis of obscure gastrointestinal bleeding. Am J Gastroenterol 2007;102:731–737. Obscure gastrointestinal (GI) hemorrhage is defined as the presence of hematochezia, melena, or iron-deficiency anemia with or without guaiac-positive stool testing and no detected source of blood loss on conventional upper endoscopy, colonoscopy, or small bowel series. The most common etiology of obscure blood loss in patients >40 years old is arteriovenous malformations of the small intestine, vascular lesions that can be causal in up to 30%–40% of patients, but are not detected on standard imaging studies, including small bowel series and computed tomographic scans (Br Med J 1984;288:1663–1665). Less common etiologies for blood loss include ulcerations, neoplasia diverticulae, and other vascular disorders, including Dieulafoy’s lesions and hemorrhage from manifestations of portal hypertension in the small intestine. Endoscopic imaging tests available for small bowel imaging include push enteroscopy, which can visualize mucosa 50–100 cm distal to the ligament of Treitz (Endoscopy 1995;27:164–170). Video capsule endoscopy (CE), developed by Given Imaging (Given Imaging Limited, Yoqneam, Israel) entails swallowing a capsule measuring 11 × 26 mm that subsequently obtains images throughout the GI tract and transmits the data during an 8-hour recording period via radio frequency to a recording device. Multiple studies have demonstrated the superiority of CE compared to push enteroscopy for patients with obscure GI bleeding. In patients presenting with obscure bleeding, the yield of CE has ranged from 55%–75% with a 25%–50% higher yield compared with push enteroscopy (Endoscopy 2002;34:685–689; Aliment Pharmacol Ther 2004;20:189–194). Double balloon enteroscopy (DBE) allows complete visualization of the small intestine using a 200-cm enteroscope with an outer diameter of 8.5 mm equipped with a 140-cm overtube with an outer diameter of 12 mm (Fujinon Inc, Saitama, Japan; Gastrointest Endosc 2001;53:216–220). With the advent of these new technologies, intraoperative enteroscopy has been reserved for patients who fail these examinations or have adhesions requiring lysis, potentially via a laparoscopic approach, so that DBE can be performed (Gastrointest Endosc 2006;64:984–988). Imaging modalities for GI hemorrhage include technetium-99m–labeled red blood cell scans and angiography. The utility of these modalities relies on the frequency of bleeding, with highest rates found if bleeding rates approximate 1.0 mL/min. Although tagged red blood cell scans can detect sources bleeding at a rate of 0.5 mL/min, their ability to locate lesions remains limited (Nucl Med Commun 2002;23:591–594). Angiography offers the potential for rapid treatment of identified bleeding lesions. In most patients, the current technique is to perform selective mesenteric angiographic examination with selective embolotherapy with either microcoils, gelatin sponge pledglets, or polyvinyl alcohol embolospheres (Cardiovasc Intervent Radiol 2006;29:49–58). In this prospective study performed by Saperas et al in Spain, the goal was to compare the yield of CE to that of angiography in patients presenting with obscure GI bleeding. Enrolled patients with obscure GI bleeding underwent initial computed tomographic angiography (CTA) followed by selective mesenteric angiography and CE performed independently by blinded investigators within 7 days. Twenty-eight patients with obscure bleeding (20 with overt bleeding and 8 with occult bleeding) were enrolled. CTA was performed initially when the patients were hemodynamically stable after either fluid or blood resuscitation. Selective angiography of the visceral trunks (celiac axis, superior mesenteric, and inferior mesenteric arteries) was performed subsequently. Patients then underwent preparation for video CE using 2 liters of polyethylene glycol preparation. The mean age of the enrolled patients was 74 years; 57% were men, and 82% with at least one comorbid condition. Of the cohort, 89% had required blood transfusions, 36% were using anticoagulant therapy, and 18% antiplatelet therapy. CTA and angiography were performed in 86% owing to contrast allergy in 1 patient and renal failure in 2 patients. CTA identified a bleeding source in 6 of 25 patients (24%; 95% confidence interval [CI], 9.4%–45.1%), including angiodysplasia (n = 2), jejunal tumor (n = 1), aortoenteric fistula (n = 1), ischemic enteritis (n = 1), and jejunal diverticula (n = 1). Mesenteric angiography identified an actively bleeding source or an obvious vascular abnormality suggestive of a bleeding site in 14 of 25 patients (56%; 95% CI, 34.9%–75.6%). Angiodysplasia was the bleeding source identified in 13 of 14 (95%) patients. Angiography also established the diagnosis in one additional patient diagnosed with small bowel stromal tumor. CTA identified 2 of 13 (15%) angiodysplasias depicted on standard angiograms. Among patients with normal CTA examinations, location of bleeding was revealed by conventional angiography in 11 of 19 (60%) patients, whereas conventional angiography was negative in 3 of the 6 (50%) patients in whom CTA was positive. No complications occurred after CTA or conventional visceral angiography. Wireless CE was able to be performed in all patients (applicability 100%) without any complications. Average gastric and small bowel transit times for the capsule were 19 (range, 2–102) minutes and 262 (range, 63–411) minutes, respectively. Failure to visualize the cecum occurred in 3 (11%) patients and the quality of the images was considered to be good or very good in all cases. CE identified a bleeding source in 19 of the 28 patients (68%; 95% CI, 47.7%–84.1%) with angiodysplasia being identified in 43% of patients. There was no difference between patients with obscure overt bleeding (70%) or patients with obscure-occult bleeding (63%; P = 1.0). In the 25 patients actually tested, CE identified a bleeding source in 72% of patients (95% CI, 50.6%–87.9%) compared with 24% for CTA (95% CI, 9.4%–45.1%; P = .005 vs CE), 56% for angiography (95% CI, 34.9%–75.6%; P = .29), and 68% for CTA-angiography (95% CI, 46.5%–85.1%; P = 1.00). Similarly, CE was able to diagnose 100% (6 of 6) of patients diagnosed by CTA and 86% (12 of 14) of patients diagnosed by angiography. Moreover, CE was positive in 63% (12 of 19) of negative cases on CTA, in 55% (6 of 11) of negative cases on angiography, and in 36% (3 of 7) of those negative on both tests. Angiography was diagnostic in 2 cases (22%) where CE was normal. The agreement between CE and the 2 other techniques was 52% (P = .03). As a result of the CE findings, therapeutic intervention was undertaken in 9 of 19 (47%) patients with positive findings. In 1 patient with jejunoileal ulcers diagnosed with Crohn’s disease, specific medical therapy was started; 4 patients with angiodysplasia underwent argon beam coagulation, which stopped hemorrhage, and 4 patients underwent surgery (1 patient with jejunal tumor was diagnosed with a gastrointestinal stromal tumor, 1 with jejunal diverticular bleeding, 1 with segmental ischemic enteritis, and 1 with an aortoenteric fistula). CommentApproximately 5% of patients with GI hemorrhage will have no source found on conventional upper and lower endoscopy and will be classified with obscure bleeding (Am J Surg 1992;163:90–92). Although obscure bleeding is uncommon, diagnostic evaluation has traditionally been prolonged and associated with considerable expense (Gastrointest Endosc 1990;36:337–341). The introduction of CE and DBE have dramatically enhanced the clinician’s ability to diagnose and treat patients with small bowel lesions. Novel radiographic imaging techniques include helical CT-enteroclysis and helical CT angiography. Helical CT–enteroclysis combines standard enteroclysis to distend the small bowel followed by helical CT (Clin Radiol 2006;61:31–39). Case reports have shown the ability of helical CT to identify small and large intestinal bleeding angioectasias (Clin Imaging 2000;24:61–63). In a report of 22 patients with obscure GI bleeding, CT enteroclysis was found to be inferior to CE in the detection of potential bleeding lesions such as angioectasia in the small bowel (Endoscopy 2003;35:1009–1014).CTA involves catheterization of the abdominal aorta followed by helical CTA before and after intra-arterial injections of a contrast medium. In a prospective study of 18 patients with bleeding colonic angioectasias, the sensitivity, specificity, and positive predictive value of helical CT angiography were 70%, 100%, and 100%, respectively, when compared with a gold standard of colonoscopy and mesenteric angiography (Gastroenterology 2000;119:293–299).The data on the clinical utility of angiography in the specific setting of obscure GI bleeding have been limited. Although prior studies have clearly shown an improved diagnostic accuracy for CE compared with push enteroscopy (Gut 2003;52:1122–1126), there has been limited information to date regarding yield of angiography and comparison to CE. The current study by Saperas et al demonstrates that CE is indeed superior to initial angiographic examination for patients with obscure bleeding. Although the authors state that the angiography occurred once patients were stabilized, we are not provided with information regarding the timing of the angiography in relation to the last bleeding episode. It could be that angiography would have a higher yield if performed only in patients within a few hours of overt bleeding. Nonetheless, initial CE appears to be associated with a higher diagnostic yield for the majority of patients.Interim results from a randomized controlled trial being performed in Hong Kong comparing CE with mesenteric angiography were presented at Digestive Disease Week 2007 (Gastrointest Endosc 2007;AB). Results were presented from 41 patients who were randomized, 20 to the capsule and 21 to the angiographic examination. Nineteen patients in each arm completed the study. Overall the diagnostic yield was 55% for CE compared with 9.5% for angiography (P = .003), supporting the findings from the Saperas study.In summary, initial video capsule endoscopy appears to be superior to angiography for the immediate investigation of patients with active obscure overt or occult bleeding. CE has been demonstrated to be superior to push enteroscopy and other radiographic imaging techniques for small bowel lesions. Whether DBE is associated with equivalent or superior diagnostic yield compared with CE is an area of active investigation. Both CE and DBE have been shown to have miss rates between 20% and 30% (Gastrointest Endosc 2006;64:740–750) and may therefore be complementary in the diagnosis and management of small bowel lesions. Given recent evidence, angiography may be relegated to a back seat for patients with obscure GI hemorrhage, but should still be considered in patients with ongoing bleeding who are unable to be diagnosed and treated after initial CE or DBE examinations. Approximately 5% of patients with GI hemorrhage will have no source found on conventional upper and lower endoscopy and will be classified with obscure bleeding (Am J Surg 1992;163:90–92). Although obscure bleeding is uncommon, diagnostic evaluation has traditionally been prolonged and associated with considerable expense (Gastrointest Endosc 1990;36:337–341). The introduction of CE and DBE have dramatically enhanced the clinician’s ability to diagnose and treat patients with small bowel lesions. Novel radiographic imaging techniques include helical CT-enteroclysis and helical CT angiography. Helical CT–enteroclysis combines standard enteroclysis to distend the small bowel followed by helical CT (Clin Radiol 2006;61:31–39). Case reports have shown the ability of helical CT to identify small and large intestinal bleeding angioectasias (Clin Imaging 2000;24:61–63). In a report of 22 patients with obscure GI bleeding, CT enteroclysis was found to be inferior to CE in the detection of potential bleeding lesions such as angioectasia in the small bowel (Endoscopy 2003;35:1009–1014). CTA involves catheterization of the abdominal aorta followed by helical CTA before and after intra-arterial injections of a contrast medium. In a prospective study of 18 patients with bleeding colonic angioectasias, the sensitivity, specificity, and positive predictive value of helical CT angiography were 70%, 100%, and 100%, respectively, when compared with a gold standard of colonoscopy and mesenteric angiography (Gastroenterology 2000;119:293–299). The data on the clinical utility of angiography in the specific setting of obscure GI bleeding have been limited. Although prior studies have clearly shown an improved diagnostic accuracy for CE compared with push enteroscopy (Gut 2003;52:1122–1126), there has been limited information to date regarding yield of angiography and comparison to CE. The current study by Saperas et al demonstrates that CE is indeed superior to initial angiographic examination for patients with obscure bleeding. Although the authors state that the angiography occurred once patients were stabilized, we are not provided with information regarding the timing of the angiography in relation to the last bleeding episode. It could be that angiography would have a higher yield if performed only in patients within a few hours of overt bleeding. Nonetheless, initial CE appears to be associated with a higher diagnostic yield for the majority of patients. Interim results from a randomized controlled trial being performed in Hong Kong comparing CE with mesenteric angiography were presented at Digestive Disease Week 2007 (Gastrointest Endosc 2007;AB). Results were presented from 41 patients who were randomized, 20 to the capsule and 21 to the angiographic examination. Nineteen patients in each arm completed the study. Overall the diagnostic yield was 55% for CE compared with 9.5% for angiography (P = .003), supporting the findings from the Saperas study. In summary, initial video capsule endoscopy appears to be superior to angiography for the immediate investigation of patients with active obscure overt or occult bleeding. CE has been demonstrated to be superior to push enteroscopy and other radiographic imaging techniques for small bowel lesions. Whether DBE is associated with equivalent or superior diagnostic yield compared with CE is an area of active investigation. Both CE and DBE have been shown to have miss rates between 20% and 30% (Gastrointest Endosc 2006;64:740–750) and may therefore be complementary in the diagnosis and management of small bowel lesions. Given recent evidence, angiography may be relegated to a back seat for patients with obscure GI hemorrhage, but should still be considered in patients with ongoing bleeding who are unable to be diagnosed and treated after initial CE or DBE examinations.