Angiographic control of nonvariceal gastrointestinal bleeding in adults

INTRODUCTION — In the overwhelming majority of patients with upper or lower gastrointestinal (GI) bleeding, the bleeding either resolves spontaneously or can be controlled endoscopically. However, persistent or recurrent bleeding occurs in 7 to 16 percent of patients with upper GI bleeding [1] and in up to 25 percent of patients with lower GI bleeding [2,3]. Such patients may require angiographic intervention to locate and/or to treat the source of bleeding. Arterial GI bleeding can be controlled by selective embolization of the bleeding artery with coils or particulate matter, selective arterial infusion of vasoconstrictive drugs, or by a combination of these techniques.

This topic will review the angiographic methods available to control upper and lower GI bleeding. The diagnosis of upper and lower GI bleeding, as well as the endoscopic and surgical methods used to control GI bleeding, are discussed separately. (See "Approach to acute upper gastrointestinal bleeding in adults" and "Approach to acute lower gastrointestinal bleeding in adults" and "Overview of the treatment of bleeding peptic ulcers" and "Colonic diverticular bleeding" and "Angiodysplasia of the gastrointestinal tract".)

INDICATIONS — Angiographic control of gastrointestinal (GI) bleeding is indicated for patients with upper or lower GI bleeding who fail to respond to medical and/or endoscopic therapy. In patients with upper GI bleeding, angiographic control of bleeding is generally considered if endoscopic attempts at therapy have failed [4]. In patients with lower GI bleeding, angiographic control of bleeding is used as an alternative to surgery in hemodynamically unstable patients with severe bleeding or for patients with ongoing or recurrent bleeding following attempts to control the bleeding endoscopically [4,5]. (See "Overview of the treatment of bleeding peptic ulcers", section on 'Interventional angiography' and "Colonic diverticular bleeding", section on 'Angiography for unstable patients and endoscopic therapy failure'.)

LESION LOCALIZATION — In most cases, angiographic control of gastrointestinal (GI) bleeding requires that the site of bleeding first be identified. This can be done with endoscopy, nuclear scintigraphy (tagged red blood cell scan), computed tomographic angiography, or standard angiography. These diagnostic tests are discussed in detail elsewhere. (See "Approach to acute lower gastrointestinal bleeding in adults", section on 'Diagnostic studies' and "Approach to acute upper gastrointestinal bleeding in adults", section on 'Upper endoscopy'.)

Prior to attempting angiographic bleeding control, the bleeding vessel(s) are identified angiographically. Optimally, prior to undergoing angiography, patients should have a serum creatinine less than 1.5 mg/dL with an estimated glomerular filtration rate greater than 60 mL/min/1.73 m², an international normalized ratio (INR) less than 1.5, and a platelet count greater than 50,000/mm³ [6]. If needed, fresh frozen plasma or platelets should be transfused prior to or during the procedure.

For lower GI bleeding, the evaluation typically begins with the superior mesenteric artery or the inferior mesenteric artery, based upon the suspected location of the bleeding source, whereas for upper GI bleeding, it begins with the celiac artery or the superior mesenteric artery. Contrast extravasation into the bowel lumen is considered definitive evidence of a bleeding site. Indirect evidence includes visualization of an aneurysm or pseudoaneurysm, filling of spaces outside the bowel lumen (diverticula), early draining vessels (angiodysplasia), neovascularity (tumors), arteriovenous fistulas, and hyperemia (colitis).

Once the site of bleeding is identified, angiographic therapy can be delivered. In addition, in patients with upper GI bleeding, treatment with empiric embolization (ie, embolization of a vessel thought to be supplying the bleeding source without bleeding having been demonstrated on angiography) can be performed, provided the bleeding site was identified endoscopically [7]. This can be aided by placement of clips at the bleeding site during endoscopy.

ANGIOGRAPHIC THERAPIES — Angiographic therapies include the delivery of agents to mechanically occlude the vascular supply of the bleeding lesion (embolization) or the infusion of vasoconstricting medications to decrease the blood flow to the bleeding site.

Embolization — Embolization works by mechanically occluding the arterial blood supply to the bleeding site. The goal of embolization is to decrease the blood flow to the bleeding site enough to achieve hemostasis while maintaining collateral perfusion to adjacent tissues to prevent ischemia or infarction [8]. Embolization should only be performed by interventional radiologists experienced with the technique because of the risk of causing bowel wall ischemia and infarction. In addition, the vessel must be accessible for selective catheterization of the bleeding site.

Indications — The indications for embolization include the following [4,9,10]:

●Massive upper or lower gastrointestinal (GI) bleeding (transfusion requirement of 4 units of blood or more in 24 hours)

●Hemodynamic instability (hypotension with systolic blood pressure lower than 100 mmHg and heart rate of ≥100 beats per minute)

●GI bleeding that fails to respond to conservative medical therapy or endoscopic control

●Active GI bleeding demonstrated by nuclear scintigraphy or computed tomographic angiography examinations

●Bleeding into pancreatic pseudocysts or from visceral artery aneurysms

●Hemobilia

Agents — Materials used for embolization are either temporary (absorbable gelatin sponge) or permanent (coils, particles, glue, ethylene-vinyl alcohol copolymer).

The biodegradable, long-acting gelatin sponge is commonly used for patients with nonvariceal GI bleeding. It causes hemostasis upon contact when injected into a vessel. The gelatin sponge comes as a powder or as small blocks that are cut to the desired size. The latter is preferred since the powdered form is more likely to lead to bowel infarction [11]. The gelatin sponge is usually cut in small pieces and mixed with saline and iodinated contrast to produce a slurry that helps with delivery of the agent [10]. Advantages of using a slurry include widespread availability, cost-effectiveness, and temporary occlusion of the targeted vessels when desired [10].

Microcoils are the preferred embolic agent for treating GI bleeding by most interventionalists. Selective catheterization and embolization of distal arteries in the mesenteric circulation became possible with the introduction of microcatheters. Microcoils can be deployed selectively into the distal bleeding artery and can preserve the collateral blood supply to the area [8].

Particles such as polyvinyl alcohol are another example of permanent embolic agents. Permanent occlusion is achieved through mechanical impaction. Particles should be used with caution in the mesenteric circulation, as they cause distal occlusion at the arteriolar level, increasing the risk of bowel infarction. Particles are used more commonly in combination with coils and gelatin sponge in the treatment of upper GI bleeding, especially when bleeding occurs from a lesion that is not expected to heal spontaneously, such as an ulcer or tumor [10,12].

Liquid agents have also been used for embolization. These agents include N-butyl 2-cyanoacrylate (NBCA) [13-18] and ethylene-vinyl alcohol copolymer [10,19]. Once delivered, these agents solidify, leading to embolization. A potential advantage of NBCA is that the time required for embolization is shorter than that with other agents [16]. However, its use requires significant training and has been associated with bowel ischemia and non-targeted embolization [6,18]. Ethylene-vinyl alcohol is nonadhesive (so there is no risk of it polymerizing to the catheter tip), has high radiopacity, and has long solidification periods that allow for more controllable and predictable embolization [6]. However, the vehicle, dimethylsulfoxide (DMSO), can cause severe vasospasm if injected too rapidly, and it requires special DMSO-compatible catheters. Finally, available preparations of ethylene-vinyl alcohol are significantly more expensive than alternative embolic materials.

Technique — Embolic agents are best delivered via a catheter or microcatheter that is inserted coaxially through a vascular sheath. An initial angiogram must be performed to confirm the bleeding site and to identify anomalous or variant vascular anatomy. The catheter should be inserted into the selected artery in order to avoid deployment of embolic materials into undesired vessels.

Microcoils have become the preferred agent for embolizing bleeding vessels and are simply deployed via a microcatheter to the site of bleeding. Similarly, polyvinyl alcohol particles are delivered via a peripherally placed catheter.

In order to use a gelatin sponge, blocks of gelatin sponge are cut into small 2 to 3 mm strips, inserted into a syringe, mixed with saline and iodinated contrast material, and administered into the percutaneously placed catheter under direct fluoroscopic visualization. The effect upon blood flow is immediately evaluated fluoroscopically; the procedure can be repeated until arterial occlusion occurs and bleeding stops. Care should be used while placing the catheter selectively into peripheral arteries since reflux of the gelatin sponge can occur during diastole or if the material is injected too vigorously, leading to undesired occlusion of additional arteries.

Complications — The complications of embolization include those associated with arteriography itself (eg, hematomas, arterial thrombosis, dissection, embolism, and pseudoaneurysm formation [20]) and bowel infarction. In a systematic review that included 927 patients undergoing embolization for upper GI bleeding, complications occurred in 5 to 9 percent of patients, depending upon the type of bleeding (luminal or transpapillary), with ischemia and infarction accounting for the majority of complications [21].

Intra-arterial vasopressin — Vasopressin causes generalized vasoconstriction via a direct action upon vessel walls, especially the arterioles, capillaries, and venules. Infusion of vasopressin produces a rapid reduction in local blood flow that gradually returns to normal several hours after the infusion has been terminated. The goal is for the decrease in perfusion pressure to permit stable clot formation at the bleeding site. However, rebleeding once the vasopressin has been stopped is common, and its use can be associated with significant ischemic events. As a result, the majority of patients are instead treated with angiographic embolization [8,22]. (See 'Efficacy of intra-arterial vasopressin' below and 'Embolization' above.)

Vasopressin use became much less frequent after the introduction of microcatheters and microcoils, as it increases significantly the complexity and the length of the post-procedure management (see 'Technique' below). However it may be still particularly helpful in patients with [8]:

●Diffuse or multifocal lesions, such as hemorrhagic gastropathy

●Altered or difficult anatomy when superselective catheterization is not technically possible

Technique — Most patients will respond to a vasopressin infusion of 0.2 pressor units/minute. Repeat angiography is performed 30 minutes later, and the rate is increased to 0.3 to 0.4 units/minute if the bleeding persists. If bleeding is controlled, the infusion is continued in an intensive care setting for 12 to 48 hours and then tapered over 24 hours [6,23]. Patients not responding to an infusion rate of 0.4 units/minute are unlikely to respond to higher doses and require alternative therapies, such as embolization or surgery.

Most authors believe that intra-arterial vasopressin is more effective than an intravenous infusion [24,25]. While an intravenous infusion can control upper GI and left-sided colonic bleeding in some cases, it is seldom used clinically [26].

Complications — Vasopressin causes peripheral vasoconstriction that can result in ischemia [27-29]. As a result, it should be used with caution in patients with coronary artery disease, congestive cardiomyopathy, severe hypertension, or severe peripheral vascular disease. It can also cause arrhythmias, especially bradycardia, and water retention leading to hyponatremia.

Fortunately, the luminal GI tract has a rich collateral blood supply, so bowel infarction is unlikely to occur in young patients who have not had previous bowel surgery. Age, vascular disease, vasopressin use, and previous bowel surgery reduce this collateral supply, especially to the colon. Superselective catheter deployment decreases the risk of bowel infarction, but may not always be technically feasible. In addition, despite superselective catheter deployment, bowel wall necrosis may occur if vasopressin therapy is used following embolization.

Choice of therapy — The choice between embolization and vasopressin should be individualized for each patient, taking into account angiographer experience. Although the efficacies of vasopressin and embolization are fairly comparable, embolization should be considered a primary option for treating lower GI bleeding. Vasopressin infusion has significant disadvantages when compared with transcatheter embolization: more complex post-procedure monitoring, longer completion of therapy, and increased likelihood of systemic complications [8]. (See 'Complications' above.)

Of note, patients with bleeding duodenal ulcers are relatively unresponsive to vasopressin infusion [30]. This is due in part to the dual blood supply of the duodenum, the inflammatory reaction produced by penetrating ulcers, and the frequent involvement of large vessels (such as the gastroduodenal artery), which do not constrict in response to vasopressin. Interventions, such as surgery or angiography, are indicated in patients who have active bleeding from a duodenal ulcer that is not stopped or slowed down significantly with endoscopic therapy. Repeat endoscopy and retreatment of the bleeding source can be performed for lesions that are initially controlled by endoscopic therapy. The patient is usually referred for surgery if the bleeding persists or if rebleeding occurs after two therapeutic endoscopies. (See "Overview of the treatment of bleeding peptic ulcers".)

Management of rebleeding — Rebleeding among patients who initially achieve hemostasis following angiographic therapy may exceed 50 percent with some therapies. (See 'Outcomes' below.)

Treatment options in patients with rebleeding include attempts at endoscopic therapy, repeated angiographic therapy, and surgery. The choice of procedure will depend upon the magnitude of the bleed, the likelihood of its being amenable to a specific therapy, and whether the patient is a surgical candidate.

Patients with rebleeding following vasopressin infusion can be treated with a repeat infusion or embolization. If embolic material is to be used, the more distal rather than proximal arteries should be targeted to allow adequate collateral blood supply to develop around the site of occlusion.

Success rates of 85 to 95 percent for managing gastric and duodenal hemorrhage have been reported when failed vasopressin infusion is followed by embolization [24,31]. However, infusion therapy following failed embolization carries a greater risk of infarction and necrosis [31].

OUTCOMES — Angiographic therapy for gastrointestinal (GI) bleeding is initially successful in most patients in whom it is technically feasible, but rebleeding is common.

Efficacy of embolization — When technically feasible, embolization is initially successful in treating approximately 95 percent of patients with GI bleeding. However, rebleeding is common (9 to 56 percent, depending upon the bleeding site) [9,21,32-35].

Systematic reviews have evaluated the efficacy of embolization for upper GI bleeding:

●In a review of 15 studies with 819 patients with upper GI bleeding, endovascular embolization with coils, polyvinyl alcohol particles, blood clots, gelatin sponge, or N-butyl 2-cyanoacrylate (NBCA) was technically successful in 762 patients (93 percent) [9]. Technical failures were due to difficult vascular anatomy, arterial dissection, vasospasm, false-negative read of an angiogram, multiple bleeding sites, and bleeding from a tumor. Of those who underwent technically successful embolization, 549 (67 percent) responded well clinically with cessation of bleeding. Rebleeding occurred in 33 percent, though half responded to repeat embolization.

●A second review included 35 studies with 927 patients [21]. Studies were divided into those that examined primary upper GI tract (luminal) hemorrhage, transpapillary hemorrhage (hemobilia or hemosuccus pancreaticus), or both. Technical success rates ranged from 52 to 100 percent. The pooled mean technical success rates for studies with luminal hemorrhage, transpapillary hemorrhage, or both were 84, 93, and 93 percent, respectively, with clinical success rates of 67, 89, and 64 percent, respectively. Rebleeding rates were 27, 9, and 26 percent, respectively. This review suggests that angiographic embolization may be particularly useful in patients with transpapillary hemorrhage.

Factors associated with failed embolization for upper GI bleeding include multiorgan failure, anticoagulant use, corticosteroid use, pre-embolization vasopressor use, bleeding subsequent to trauma or invasive procedures, longer time to angiography, previous surgery, and embolization using coils alone or coils with another embolic agent [9,21,36]. However, treatment failure does not appear to be associated with empiric embolization of endoscopically identified bleeding sources [36-38]. (See 'Lesion localization' above.)

Initial success rates with embolization for the treatment of lower GI bleeding are also high:

●In a summary of 15 studies with 309 patients, embolization using polyvinyl alcohol, coils, or gelatin sponge was technically feasible in 82 percent of patients [39]. Of those who were embolized, 95 percent of patients had immediate hemostasis and 76 percent were free of bleeding within 30 days. Minor ischemic episodes were noted in 10 percent, with major ischemic episodes being reported in 2 percent.

●A subsequent study examined 78 patients with 83 episodes of lower GI bleeding [40]. Angiography identified active bleeding in 40 episodes (48 percent), and embolization was carried out for 38 episodes (46 percent; 37 episodes with active bleeding on angiography and one without active bleeding).

However, embolization for lower GI bleeding has been associated with rebleeding rates of 22 to 56 percent [40-44].

NBCA embolization has also been used in the treatment of lower GI bleeding. In a study with 27 patients with lower GI bleeding who underwent NBCA embolization, immediate hemostasis was achieved in all 27 (one patient was also treated with a microcoil) [45]. Rebleeding occurred in four patients (15 percent).

Efficacy of intra-arterial vasopressin — The success of intra-arterial vasopressin infusion varies with the site of bleeding. In patients with lower gastrointestinal bleeding, intra-arterial infusion of vasopressin will lead to the cessation of bleeding in approximately 80 to 90 percent of patients [31,46-48]. However, not all patients undergoing angiography will have a bleeding site identified, and even if a bleeding site is identified, intra-arterial vasopressin infusion may not be technically possible, or bleeding may recur once the infusion is stopped (seen in 5 to 50 percent of patients).

Success rates in patients with upper GI bleeding vary depending upon the site of bleeding. Vasopressin appears less likely to control bleeding in patients with bleeding from the pylorus or duodenum. In patients with gastric bleeding, selective angiography identifies a point of extravasation in approximately 60 percent of patients [30]. In those who receive vasopressin, initial success rates are approximately 70 to 80 percent [24,31,49-52].

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Gastrointestinal bleeding in adults".)

SUMMARY AND RECOMMENDATIONS

●Indications – Angiographic control of gastrointestinal (GI) bleeding is indicated for patients with upper or lower GI bleeding who fail to respond to medical and/or endoscopic therapy. In patients with upper GI bleeding, angiographic control of bleeding is generally considered if endoscopic attempts at therapy have failed. In patients with lower GI bleeding, angiographic control of bleeding is used as an alternative to surgery in hemodynamically unstable patients with severe bleeding or for patients with ongoing or recurrent bleeding following attempts to control the bleeding endoscopically. (See 'Indications' above.)

●Angiographic therapies – Angiographic therapies include the infusion of vasoconstricting medications or the delivery of agents to mechanically occlude the vascular supply of the bleeding lesion (embolization). (See 'Angiographic therapies' above.)

•Embolization – Agents used for embolization include biodegradable gelatin sponge, polyvinyl alcohol particles, liquid agents such as glue and ethylene-vinyl alcohol copolymer, and metallic coils. (See 'Embolization' above.)

Superselective transcatheter embolization with microcoils is the primary endovascular treatment option for lower GI bleeding. Endovascular treatment of upper GI bleeding is usually performed with a combination of coils, gelatin sponge, and/or particles.

The complications of embolization include those associated with arteriography itself (eg, hematomas, arterial thrombosis, dissection, embolism, and pseudoaneurysm formation) and bowel infarction. (See 'Complications' above.)

•Vasopressin – Vasopressin causes generalized vasoconstriction via a direct action upon vessel walls, especially the arterioles, capillaries, and venules. The goal is for the decrease in perfusion pressure to permit stable clot formation at the bleeding site. (See 'Intra-arterial vasopressin' above.)

Since vasopressin causes peripheral vasoconstriction, it should be used with caution in patients with coronary artery disease, congestive cardiomyopathy, severe hypertension, or severe peripheral vascular disease. It can also cause arrhythmias and water retention leading to hyponatremia. (See 'Complications' above.)

●Efficacy – Initial hemostasis may be achieved in up to 95 percent of patients in whom angiographic therapy is technically feasible, but rebleeding is a common problem, occurring in up to 56 percent of patients. (See 'Outcomes' above.)