Elias N. Brountzos, MD

Gastrointestinal bleeding is a frequent cause of hospital admissions. Patients present with melena, hematemesis, hematochezia, and/or shock. GI hemorrhage usually stops spontaneously or responds to conservative management. In approximately 25% of cases, however, the hemorrhage is massive and will be stopped only if the bleeding site is localized.

Endoscopy is the principal method of diagnosing and treating GI hemorrhage. When endoscopy is impossible or ineffective, diagnostic imaging is recommended to localize the site of bleeding and determine its cause.

 GI bleeding is typically intermittent, and imaging must be timed appropriately.

The rate and pattern of the patient’s bleeding will determine which tests to use. Scintigraphic red blood cell scanning can detect bleeding rates of 0.2 to 0.4 mL/min.

 Screen-film angiography can detect bleeds of 0.5 mL/min in animal models. In vitro digital subtraction angiography is five to nine times more sensitive.

 The sensitivity of scintigraphy ranges from 20% to 60%. The accuracy of localization is relatively low, ranging from 41% to 54%.

 Arteriography has reported sensitivities that range between 22% and 87%, but its localization accuracy is excellent (100%).

Multislice CT is emerging as a useful tool for the detection and localization of gastrointestinal bleeding. The principal finding of active bleeding is contrast extravasation in the bowel lumen with attenuation greater than 90 HU. Secondary signs include arterial enhancement or bowel wall thickening and the depiction of tumors or vascular lesions. Reported sensitivity and specificity rates for MSCT are 90.9% and 99%, respectively. Accuracy is 97.6%.

GI bleeding is classified as either upper or lower, depending on the anatomic location of the source. Upper GI bleeding originates proximal to the ligament of Treitz, while lower GI bleeding originates distal to it. These entities differ in etiology, treatment, prognosis, and incidence, with upper GI bleeding six times more common than lower GI bleeding.

Nearly half of all cases of upper GI bleeding are related to peptic ulcers.

 Therapy advances, such as the association of Helicobacter pylori with gastric ulceration, have changed the way that peptic ulcers are managed. Admission rates for upper GI bleeding, approximately 0.1% of the population, have not decreased, however. The incidence of nonvariceal upper GI bleeding is 170 in 100,000 adults per year.

The main causes of bleeding include peptic ulcer, gastritis, Mallory-Weiss tear, marginal ulcer, procedural problems (i.e., following biopsy or gastronomy), arteriovenous malformation or angiodysplasia, Dieulafoy lesion, hemobilia, hemosuccus pancreaticus, and aortoenteric fistula.

Endoscopic therapy is the treatment of choice for upper GI bleeding, but this is ineffective or impossible in 5% to 10% of cases. Bleeding will recur after an initially effective treatment in 15% to 30% of cases.

 Emergency surgery is often required in these patients, who will usually present with massive hemorrhage. The risk of mortality is high, ranging from 20% to 50%.

Selective transcatheter arterial embolization for the control of upper GI bleeding was introduced in 1972.

 An alternative interventional radiological therapy is intra-arterial vasopressin infusion, though the effectiveness of this procedure is reduced by the presence of an abundant arterial supply to the stomach and duodenum.

 Embolotherapy has been shown to be superior to vasopressin infusion in patients with major GI hemorrhage. Investigators have reported an 88% probability of success with embolization compared with 52% with vasopressin treatment.

The goal of embolization is to devascularize the bleeding lesion. This should be done in a superselective way to avoid tissue infarction. Medical or surgical treatment specific to the cause of the bleed is mandatory. The patient should be resuscitated prior to intervention, and fluid and electrolyte deficiencies and coagulation disorders attended to. Coagulation disorders have been implicated in embolization failures and increased mortality rates.

Standard catheters are used to catheterize the celiac artery, superior mesenteric artery, or (if required) gastroduodenal and left gastric arteries. Angiographic findings of GI bleeding include localized hyperemia, a "pseudovein" sign, contrast extravasation, and pseudoaneurysm (Figure 1).

Angiography may be normal in 13% to 63% of the cases, simply because the bleed is intermittent. Many practitioners will then perform a "blind" or "empirical" embolization, that is, embolization without angiographic proof of extravasation. Endoscopy-guided "blind" embolization, in which an endoscopist places radiopaque markers such as clips at the bleeding site to guide the interventional radiologist, has been shown to be effective.

Materials used for embolization include autologous clots, gel foam, polyvinyl alcohol (PVA) particles, coils, and biological glue. The use of microcatheters to gain close access to the bleeding is of paramount importance. Coils should match the diameter of the target artery, and care should be taken to occlude the bleeding site both proximally and distally to avoid retrograde perfusion (Figure 2). Any particles used for embolization should be relatively large (300 to 700 µm) to avoid deep penetration and tissue damage. Liquid embolization materials, such as glue, should be used only as a last resort to stop massive bleeding. The use of glue is technically demanding and may lead to serious complications.

The technical success of embolization is high. Reported occlusion rates range from 89.5% to 98.7%.

The bleeding vessel may, however, be inaccessible in some cases, due to tortuosity, spasm, or atherosclerotic occlusion.

Reported primary clinical success rates range from 58% to 92.3%.

 Factors affecting clinical success include coagulation issues, multiple organ failure, APACHE III score, the sole use of coils for embolization, the time from the start of hemorrhage to arteriography, and the number of red blood cell units infused.

Embolization of upper GI bleeds carries a 4% to 18.1% risk of serious complications.

 Visceral ischemia is the most serious, and this can be lethal. Ischemic complications are thankfully infrequent with meticulous and superselective technique. Caution should also be exercised whenever patients have a history of surgery or radiotherapy because collaterals will be scarce in those patients.

Patients with upper GI bleeding who are treated with embolotherapy are very ill. The reported mortality rates range from 32% to 46%. Patients with multiorgan failure are 17.5 times more likely to die, independent of the outcome of embolization. The risk of death is also 3.4 to 9.6 times higher in cases of coagulopathy. Procedural failure is 2.8 to 2.9 times more common in this group, and the risk of rebleeding is 19.5 times higher.

Patients who undergo successful embolization are 13.3 times more likely to survive than those whose procedure is not successful clinically.

 Investigators have found no difference between embolization and surgery for the rate of rebleeding, need for additional surgery, or death, despite the fact that the embolized patients were older and had a greater incidence of heart disease.

In summary, angiographic evaluation and embolization should be considered the first-line treatment after failure of endoscopic management of upper GI bleeding.

The incidence of lower GI bleeding is 20 to 30 per 100,000 adults. Incidence increases with age. The main causes are diverticular disease, angiodysplasia, malignant tumors, benign tumors, postpolypectomy, coagulopathy, and anticoagulation.

Diverticular disease is frequent after the age of 50 and very common in patients over 80 years old. Right-sided diverticula tend to bleed more frequently than those on the left side. Angiodysplasia is an acquired vascular anomaly that mostly affects the right colon and cecum. Neoplasia may be the cause of lower gastrointestinal bleeding in 5% to 8.3% of patients. This underscores the need for cause-specific treatment after cessation of the hemorrhage.

Patients with lower GI bleeding may have different presentations and can be grouped into four categories. The majority (75% to 90%) will have minor bleeding that stops with conservative therapy. The next group covers patients with chronic intermittent bleeding who are best evaluated with endoscopy. The third category includes patients with life-threatening, intractable bleeding who are hemodynamically stable. The intermittent nature of the bleeds means that these patients may be best evaluated with scintigraphy, followed by angiography if required. The last group covers patients with continuous bleeding who are also hemodynamically unstable. These patients are best managed by emergency angiography.

Embolization for lower GI bleeding was first introduced in the early 1970s. It did not gain widespread acceptance immediately because it was associated with a significant risk of bowel infarction (13% to 33%). This was attributed to the poor anastomotic arterial network of the small and large bowel compared with the upper GI tract.

Interventional radiological management of lower GI bleeding was initially limited to intra-arterial vasopressin infusion. This modality requires long stays in intensive care, fails to control bleeding in more than 20% of patients, and has a 36% to 43% rebleeding rate. The risk of major complications is 20%, and the mortality rate is 9%.

 A comparison of vasopressin infusion with embolization has concluded that the latter technique should be considered the primary treatment option.

Patients should be stabilized prior to embolization, as with upper GI bleeding. All available imaging studies should be used to help plan the angiographic examination. Data suggest that MSCT can localize lower GI bleeds effectively.

 MSCT certainly helps when planning embolization.

A Foley catheter should always be inserted to prevent a full bladder from obscuring a pelvic bleeding site. Selective catheterization of the superior and inferior mesenteric arteries, and occasionally the internal iliac arteries, is performed using standard diagnostic catheters. Coaxial catheterization with a microcatheter is performed after depiction of the bleeding site (Figure 3).

All authors advocate that embolization should be attempted only when the microcatheter can be advanced to the border of the colon (i.e., to the marginal artery or vasa rectae). This minimizes risk of bowel infarction. Most practitioners use microcoils or PVA particles as embolic material, either alone or in combination (Figure 4). Identification of the bleeding branch and subsequent navigation of the microcatheter into the vasa rectae can be a major technical challenge. Coil embolization of the marginal artery is acceptable if there is adequate collateral supply to the adjacent bowel segments.

Studies published after 1990 show that embolization stops lower GI bleeding in 44% to 100% of patients.

 Rebleeding may occur in a significant percentage (14% to 50%), depending on the underlying lesion. Angiodysplasias rebleed most frequently. Endoscopy should be used to establish the cause of the bleed and guide therapy.

Clinically significant bowel ischemia is negligible, according to post-1990 reports, if meticulous technique is applied (i.e., embolization at the utmost distal location with microcoils or large particles). 

 Practitioners are nonetheless urged to follow up patients with endoscopy in case of future bowel strictures.

 Patients with lower GI bleeds who are managed with embolotherapy are seriously sick and face a life-threatening disease. Embolization is life-saving in this setting. The mortality rate for surgical management in this patient group ranges between 10% and 36%.

In summary, although most patients with GI bleeding will respond to conservative or endoscopic treatment, a few require the bleed to be localized on imaging and will need further management. Embolization with superselective techniques will stop hemorrhage in the majority of patients with few major complications. Patients can then undergo treatment targeted at the cause of the bleed.

Articles

Gastrointestinal bleeding is a frequent cause ofhospital admissions. Patients present withmelena, hematemesis, hematochezia, and/orshock. GI hemorrhage usually stops spontaneouslyor responds to conservative management.

Elias N. Brountzos, MD

Articles

Embolotherapy controls largegastrointestinal bleeds safely