Endoscopic Management of Erosive Causes of UGIB

Endoscopy has only a marginal role in the treatment of hemorrhage caused by erosions. The treatment of choice is acid suppression, leading to healing of erosions and normalization of hemoglobin levels. The outcome is generally excellent. Acute bleeds sometimes require endoscopic treatment such as of a visible vessel, for which the treatment is similar to the approach of gastroduodenal ulcers. A different approach may be the control of bleeding by a topical hemostatic spray. Hemospray (Cook Medical, Winston Salem, NC, USA) is a novel proprietary powder specifically developed for the treatment of UGIB. It is a nonorganic mineral blended powder that is thought to work via absorption of liquids at the bleeding site, forming an adhesive and cohesive mechanical barrier over the bleeding site. In a large European observational study in 63 patients with UGIB due to various causes, the application of Hemospray as primary monotreatment led to immediate hemostasis in 85% of patients with a 7-day rebleeding rate of 15%. When used as salvage therapy, after failure of conventional endoscopic treatment modalities, initial hemostasis was 100% and rebleeding 25%, reflecting the refractory nature of these lesions. Causes of nonvariceal, nonulcer UGIB included erosive esophagitis, gastritis, and/or duodenitis (n = 7), as well as Dieulafoy lesions, gastric antral vascular ectasia (GAVE), and Mallory-Weiss lesions. Concomitant use of antithrombotics, mostly causative for erosive UGIB, do not appear to influence the success rates of Hemospray.

Angiodysplasia

Angiodysplasia is defined as a sharply delineated vascular lesion within the mucosa, with a typical red appearance, with flat or slightly raised surface. Angiodysplasia is generally believed to be acquired, but the exact causes are unknown. These lesions are often multiple and frequently seen in the colon, but may also involve the stomach and small bowel, including the duodenum. Patients with angiodysplasia can present with chronic iron-deficiency anemia, as well as with acute GI bleeding with melena and seldom hematemesis.

Dieulafoy Lesion

Dieulafoy lesions most commonly occur in the proximal stomach along the lesser curve, but they are incidentally seen in the esophagus, small intestine, and colon. A Dieulafoy lesion consists of a dilated tortuous artery with a diameter up to 3 mm that protrudes through the mucosa and may cause massive GI hemorrhage by erosion of the artery. It is thought that this is due to a combination of factors such as straining of the vascular wall during peristalsis and peptic digestion. Bleeding due to Dieulafoy lesions account for 1% to 6% of cases of acute nonvariceal UGIB.

Gastric Antral Vascular Ectasia

GAVE is endoscopically recognized as a pattern of linear red stripes in the antrum separated by normal mucosa. GAVE has been associated with several autoimmune conditions, renal failure, and bone marrow transplantation.

GAVE or so-called watermelon stomach is often confused with portal hypertensive gastropathy. Although GAVE is closely related to portal hypertension, it can also appear in the absence of portal hypertension, and GAVE generally does not respond to treatments that reduce portal pressure.

Portal Hypertensive Gastropathy

Portal hypertensive gastropathy or congestive gastropathy corresponds to dilated gastric mucosal capillaries without inflammation. It is exclusively observed in patients with portal hypertension. When present in the small intestine or colon, it is termed portal hypertensive enteropathy and portal hypertensive colopathy. The degree of portal hypertension needed for development of portal hypertensive gastropathy remains controversial. Bleeding from these conditions usually occurs as slow, diffusely oozing, but it may also be acute (and even) massive.

Endoscopic Management of Vascular Anomalies and Portal Hypertensive Gastropathy

The current standard of endoscopic treatment of bleeding angiodysplasia consists of coagulation therapy. Although high success rates are achieved for treatment of individual lesions, patients may experience recurrent anemia or overt bleeding. Banding of angiodysplasia has also been demonstrated in a small study of 11 subjects. In case of bleeding angiodysplasia in the small intestines, (repeated) argon plasma coagulation (APC) via enteroscopy has been proved feasible and successful.

Endoscopic treatment is also the first choice in bleeding Dieulafoy lesions and is able to achieve hemostasis in more than 90% of cases. Treatment is usually performed with clipping or band ligation of the lesion.

The first-line treatment of actively bleeding GAVE as well as recurrent bleeding from GAVE has for a long time consisted of repetitive sessions of APC. In case of recurrence, patients have been successfully retreated with APC. More recent publications show that band ligation of bleeding GAVE resulted in fewer treatment sessions for control of bleeding and higher rates of cessation of further bleeding than APC treatment. Whether this also pertains to persistently decreased recurrence rates remains to be demonstrated ( Fig. 1 ).

Dieulafoy lesion in the fundus before and after band ligation.

Endoscopic management has no place in treating portal hypertensive gastropathy. Successful treatment should aim at reducing portal venous pressure by means of nonselective β-blockers, vasoactive drugs, and/or transjugular intrahepatic portosystemic shunts.

Endoscopic Management of Neoplastic Bleeding

Conventional endoscopic therapy for bleeding GI tumors is generally not very successful, because most tumors are beyond the reach of thermal or mechanical modalities due to their complex neoangiogenesis. Endoscopic management was assessed in a study of 42 patients with primary upper GI cancer. In approximately half of these patients, severe UGIB was the first presentation of their malignancy, most of which were at an advanced stage with larger ulcerating tumor masses. Endoscopic hemostasis was only achieved in the lesser advanced, nondiffusely bleeding cancers. Severe bleeding correlated with a poor 1-year survival irrespective of initial hemostasis. Against this background, radiotherapy and/or transarterial embolization (TAE) were for a long time the only alternative treatments for management of bleeding. Radiotherapy has a delayed effect, and the effect of both radiotherapy and embolization may be transient. In a retrospective study of 30 patients with UGIB from gastric cancer, almost 75% responded to radiotherapy, but rebled within 3 months. Patients who received concurrent chemoradiotherapy had a significant lower rebleeding rate (18%). In a retrospective study of 23 patients with neoplastic UGIB treated with TAE, the overall treatment success was only 50%.

Hemostatic powders are increasingly advocated as an endoscopic treatment of neoplastic bleeding because of its ability to stop diffuse bleeding. Two case series from Canada and France, including a total of 10 patients, showed promising short-term results in the use of Hemospray in bleeding esophageal, gastric, and pancreatic cancers. In both reports, immediate hemostasis was reached in 100% of the patients and the rebleeding rate after 3 days was 20%. Because powder can be endoscopically delivered during the index endoscopy, it provides a rapid and effective treatment modality in the bleeding oncology patient in the acute setting ( Fig. 2 ).

Hemorrhagic gastric adenocarcinoma of the corpus after hemostasis with Hemospray.

Mallory-Weiss Tear

A Mallory-Weiss tear is a common cause of UGIB and typically presents as hematemesis after an initial episode of vomiting without blood. It is defined as a mucosal laceration at the gastroesophageal junction or gastric cardia, usually caused by retching or forceful vomiting. Many factors have been associated with the development of Mallory-Weiss lesions, including alcohol use, use of aspirin and coumarins, paroxysms of coughing, pregnancy, heavy lifting, straining, seizure, blunt abdominal trauma, colonic lavage, and cardiopulmonary resuscitation. The amount of blood loss is usually mild. Several epidemiologic studies reported these lesions as the bleeding source in 2% to 8% of patients presenting with acute UGIB. In a large UK audit of 5004 patients who underwent endoscopy for UGIB, Mallory-Weiss lesions were the only bleeding source identified in 2.1% of cases and were present in a total 4.3% of cases.

Boerhaave Syndrome and Other Causes of Transmural Perforation

Boerhaave syndrome is a rupture of the esophagus caused by a rapid increase in intraluminal pressure in the distal esophagus combined with negative intrathoracic pressure caused by straining or vomiting. Other causes of benign esophageal perforation include endoscopic and surgical handling and foreign body impaction. In some cases, bleeding may be an accompanying symptom.

Caustic Injury and Foreign Body Ingestion

Ingestion of foreign bodies and toxic substances can be accidental or intentional. The ingestion of strong alkali results in liquefactive necrosis, which is associated with deep tissue penetration and may result in perforation accompanied by bleeding. Acidic agents cause more superficial coagulation necrosis with scarring that may limit the extent of the injury. Bleeding may occur as a result of widespread ulceration. More severe bleeding is however rare.

UGIB is also seen after foreign body ingestion. Most noteworthy in this respect is the ingestion of batteries and magnets leading to leakage and pressure adhesion, which can lead to local ulceration and perforation. Hemorrhage is a frequent complication after foreign body ingestion and results from local pressure or puncture of the mucosa with sometimes fistula formation to an underlying vessel.

Endoscopic Management for Traumatic Bleeding

Most of the Mallory-Weiss lesions stop bleeding spontaneously. A minority of cases require endoscopic treatment. In a French series of 218 patients with Mallory-Weiss bleeding, 56 (26%) had active bleeding on endoscopy. In other series, this prevalence ranged from 5% to 44%. Various studies have looked at optimal endoscopic treatment. In a trial that randomized 63 patients to either supportive treatment or endoscopic therapy, rebleeding rates after only supportive therapy were higher than that after endoscopic epinephrine injection (25% vs 6%). Endoscopic epinephrine injection, hemoclip placement, and band ligation were equivalent for primary hemostasis, all achieving 100% primary hemostasis. Rebleeding rates after band ligation and hemoclip placement were similar in a randomized Korean trial (6%–10%), but significantly lower after band ligation in a French trial (0% vs 18%). It is unknown whether prescription of a proton pump inhibitor accelerates healing.

The “bear-claw” or over-the-scope-clip system (Ovesco Endoscopy, Tübingen, Germany) is a new clipping device developed for closure of large luminal GI defects. These defects may be spontaneous like Boerhaave syndrome or iatrogenic (post-endoscopic submucosal dissection or dehiscence after surgical anastomosis ) and include both perforated and bleeding lesions. Until now, evidence regarding the effectiveness of the device only came from case series, but it is expected that over-the-scope-clip system may become an alternative to conventional surgical therapy for traumatic lesions. Endoscopic placement of a self-expandable metal stent (SEMS) can also be considered for Boerhaave syndrome. The authors reported a series of 33 patients with nonmalignant esophageal perforation treated with SEMS placement. Initial sealing of the perforation was achieved in 97% of patients, long-term success without the need for surgical repair in 88%. Treatment was complicated in one-third of patients by stent migration, managed by SEMS repositioning or restenting. In their center, this has now become the initial endoscopic treatment of choice for esophageal perforations that require intervention. Stents need to be removed within 6 weeks after placement, because a longer time in situ often impairs the ease of removal. When a persistent perforation is observed after stent removal, placement of another stent should be considered.

Endoscopy often has little to offer in terms of treatment of bleeding caused by caustic injury. Early prophylactic esophageal stenting preventing stricture formation is currently under investigation.

Hemobilia

Hemobilia or bleeding from the hepatobiliary tract is a rare cause of UGIB. The causes include liver biopsy, percutaneous transhepatic cholangiography, cholecystectomy, endoscopic biliary biopsies/stenting, and hepatic or bile duct tumors. In a review of 222 cases of hemobilia, 65% were iatrogenic, whereas accidental trauma accounted for only 6%. In rare cases, bleeding may occur due to a fistula to the portal system or the hepatic vein, which can also lead to rapid onset of extreme jaundice. In most cases, these bleeds are self-limiting and do not require intervention other than preventing biliary obstruction due to clots.

Symptoms of hemobilia can be variable. The classic triad of jaundice, biliary colic, and overt UGIB was in the most recent series only observed in 22% of the patients. Endoscopic retrograde cholangiography is helpful to confirm diagnosis and establish the underlying cause. It can also be of help in clearance of the biliary tree, with stenting if necessary to reduce the risk of renewed obstruction. In rare cases, a lesion of the hepatic vein can establish a connection between the hepatic vein and the biliary tract and instead of hemobilia give rise to opposite flow of bile into the vein. Such bilhemia may lead to marked, acute onset of jaundice. In such a case, biliary stenting can reverse the flow, by rapidly reducing jaundice and allow closure of the leak.

Hemosuccus Pancreaticus

Hemosuccus pancreaticus is caused by a bleeding source in the pancreas, pancreatic duct, or adjacent structures, such as the splenic artery connecting to the pancreatic duct. Hemosuccus is most often due to pancreatic pseudoaneurysms resulting from acute or chronic pancreatitis. Usually this involves erosion of the splenic artery leading to hemorrhage into the pancreatic duct. It has been estimated to occur in about 1 in 500 to 1500 cases of UGIB. In 2 retrospective studies (n = 40 total), the most frequent complaints at presentation were melena, hematochezia, hematemesis, and epigastric pain. Patients may also develop symptoms of nausea and vomiting, weight loss, and jaundice (when the bleed also leads to obstruction of the common bile duct). In an observational report including 31 patients with hemosuccus pancreaticus, the presence of duodenal blood was observed during upper GI endoscopy in only half of the patients.

Endoscopic Management of Hemobilia and Hemosuccus Pancreaticus

Because hemobilia is mostly self-limiting, the role of endoscopic treatment is limited to clearance of the biliary system if needed, with occasional stenting to maintain an open biliary tract. Also for hemosuccus pancreaticus, there is no major role for endoscopic treatment. Selective arterial embolization is successful in 50% of the patients, whereas surgery is often necessary in emergency situations or after failure to control the bleeding by TAE.