Lower Gastrointestinal Bleeding Imaging 

Lower gastrointestinal (GI) hemorrhage is defined as bleeding from the bowel distal to the ligament of Treitz (see the following image). Acute lower GI bleeding is of recent onset and results in hemodynamic instability and decreasing hemoglobin levels, which need to be treated with transfusions.

Diverticulosis has been implicated as the source of bleeding in as many as 60% of cases of lower GI bleeding. The diverticula are more prevalent in the left or sigmoid colon, but positive arteriographic findings for bleeding localizes the bleeding to the right colon in 60% of cases.

Preferred examination

Colonoscopy after rapid oral purging has emerged as the procedure of choice for the evaluation of acute lower GI bleeding (69-80% yield), and it also provides a means for therapy. This procedure involves the use of thermal contact modalities or epinephrine injections in cases of diverticula and vascular ectasia. Similarly, other causes of lower GI bleeding can be effectively treated during colonoscopy.

However, the 2 widely used diagnostic tests for lower GI bleeding are nuclear scanning (see the image above) during episodes of bleeding or angiography (see the images below). In cases in which colonoscopy is unsuccessful, scanning during episodes of bleeding and angiography are considered to be next imaging tests to determine the cause of the bleeding. Angiography remains the best option in a patient in unstable condition and should be performed in cases of massive bleeding.^[1]

An advantage of angiography over coloscopy or scintigraphy is that no special preparation is needed. Thus, it can be performed in a relatively short period; angiography also provides a means for immediate treatment. However, patients in stable condition can undergo scintigraphy to guide and increase the yield of angiography.

Scintigraphy and/or angiography also play important roles in diagnosis and embolization when colonoscopy reveals negative findings or when it is not feasible. Some investigators also advocate an upper GI endoscopy after colonoscopy yields negative results,^[2] as about 10% of cases are ultimately found to involve an upper GI source of bleeding.

Radiologic intervention

Two transcatheter methods for the treatment of GI bleeding include vasopressin infusion and embolization. Superselective embolization has become more accepted in recent years.^{[3, 4, 5, 6]}

Vasopressin infusion causes vasoconstriction of the small arteries, arterioles, and capillaries, and it has been used for both upper and lower GI bleeds.^[5] The rate of successful control of bleeding has been reported to be 70-90%. The repeat bleeding rate is reported to be 17-27%.

The infusion is started with the catheter in the main trunk of the mesenteric artery that is cause of bleeding. The initial rate is 0.2 U/min. A follow-up angiogram is obtained after about 30 minutes. In cases of active hemorrhage, the rate is increased to 0.4 U/min. Higher rates are not recommended because the potential complications from vasoconstriction can exceed the benefits. The infusion is tapered at 6- to 12-hour intervals and then stopped if no further bleeding ensues.

Complications of vasopressin infusion include myocardial ischemia, arrhythmia, hypertension, bowel ischemia, peripheral vascular ischemia, and antidiuretic effects. Because of significant rebleeding, variable success, the need for ICU monitoring, and the aforementioned adverse effects of the infusion, renewed interest in embolization therapy has developed over the last few years. In fact, embolization has now become the preferred transcatheter therapy. Embolization also achieves immediate control of the bleeding, and ICU monitoring to the degree required with vasopressin infusion is avoided.

The aim of embolization is to decrease the arterial inflow so that the pressure at the bleeding site is decreased and hemostasis occurs. The important issue is to avoid devascularization of the tissues, which leads to ischemia and infarction.^[7]

In 1974, Bookstein first described transcatheter embolization by using an autologous clot.^[8] Since then, various investigators have described small series of transcatheter embolization for lower GI bleeds, with varying success. The potential complication of bowel ischemia and infarction initially limited use of this technique. Earlier groups had described postembolization bowel infarction rates ranging from 0% to 23%. However, some authors claim that the high rate of significant ischemia may have been related to proximal embolization sites in relation to the marginal artery, as lack of large vascular collaterals in large bowel may jeopardize significant areas of colon.

With this issue in mind, further studies with superselective catheterization techniques and embolization were performed using polyvinyl alcohol (PVA, Contour, Medi-tech Inc. Natick, Mass) particles and Gelfoam (Upjohn, Kalamazoo, Mich), although most of the studies have used microcoils (platinum coils), either alone or in conjunction with Gelfoam or PVA.

A study by Evangelista and Hallisey supported the use of superselective embolization as the most effective procedure in reducing complication rates.^[9] The investigators reported an 88% success rate and argued against the use of PVA as the sole embolization agent, as the small particles may reach intramural circulation and thus occlude the submucosal plexus beyond the level of collateralization, leading to significant bowel ischemia. An advantage of coils is that they are visible and, therefore, more controllable.^[9]

Bandi et al also supported superselective embolization as a feasible, safe, and effective technique for treating acute lower GI hemorrhage.^[3] They performed embolization in only those patients in whom they could successfully catheterize the arteria recta leading to the bleeding point. PVA alone (150-500 µm) was used in 28 of these procedures. In 25 patients who underwent objective follow-up with colonoscopy (n = 12), surgery (n = 9), or both (n = 4), mucosal ischemia was demonstrated in 6 (24%) of these patients, but they remained asymptomatic without clinical sequela. No clinically significant bowel ischemia was seen.^[3]

New advances are being made with computed tomographic angiography (CTA) and magnetic resonance angiography (MRA). Junquera et al used CTA to evaluate suspected colonic angiodysplasia.^[10] The sensitivity of CTA in the detection of colonic angiodysplasia was 70%, its specificity was 100%, and it had 100% positive predictive values, compared with findings of angiography or colonoscopy.

In an animal study, Hilfiker et al evaluated use of 3-dimensional (3-D) MRI.^[11] The investigators compared technetium-99m (^99m Tc)-labeled red blood cell (RBC) scintigraphy with 3-D MRI after the intravascular administration of contrast agent. MRI had 100% sensitivity and specificity, compared with 78% sensitivity and 72% specificity for scintigraphy.

Since the late 1970s,^99m Tc sulfur colloid and^99m Tc-labeled RBCs have been used in the evaluation of GI bleeding.^{[12, 13, 14, 15, 16]} In 1977, Alavi et al described use of^99m Tc sulfur colloid tests for diagnosis of bleeding induced in the small bowel and colon of dogs.^[17] Bleeding rates as low as 0.05-0.1 mL/min can be detected. The disadvantage of sulfur colloid scanning is the potential masking of bleeding in the upper abdomen by activity in the liver and spleen.^99m Tc-labeled RBC scanning does not have this problem, and an animal study by Thorne et al in 1987 showed bleeding detection rates as low as 0.04 mL/min, which are comparable to those of sulfur colloid scanning.^[18]

Bunker et al compared the 2 techniques and found that^99m Tc RBC scanning was superior to sulfur colloid scanning.^[19] The former demonstrated a sensitivity of 93%, a specificity of 95%, and an overall accuracy of 94% in detecting and localizing GI hemorrhage. Initial studies showed acceptable sensitivities for GI bleeding, but a few studies in late 1980s raised doubts about the accuracy of scintigraphic studies. One group reported inaccurate localization leading to a surgical error rate of 42%. Some authors claim that most available reports in literature base their conclusions on results obtained with older techniques. The newer techniques involve more dynamic imaging (more frequent acquisition of data), extra-large field-of-view gamma cameras, and cine scintigraphy or movie-mode displays. Later studies have shown that RBC scans have an accuracy of nearly 90% in the localization of the bleeding site.

The following is an example of a ^99m Tc RBC scan.

Degree of confidence

Scintigraphy is reported to increase the diagnostic yield of angiography.^{[12, 20]} Gunderman et al showed that, without scintigraphic screening, angiograms depicted bleeding at a rate of 22%.^[21] After the implementation of scintigraphic screening, 53% of the arteriograms depicted bleeding. In another study, Ng et al showed that, when^99m Tc RBC scans showed immediate blush, patients required urgent angiography.^[22] In contrast, patients whose scans showed delayed blush have low angiographic yields. The investigators recommended that, when scans show delayed blush or negative results, patients may be observed and evaluated with colonoscopy.

The advantages of "bleeding" scanning are that it is noninvasive and that it requires no special patient preparation.^[23] Bleeding is diagnosed when an abnormal focus of activity is seen. This activity increases over time and changes in position due to peristalsis. An advantage of^99m Tc RBC scanning is that repeat imaging can be performed after an interval of as long as 24 hours.

The earliest description of the angiographic demonstration of GI bleeding was in 1963. Since then, angiography has been used in evaluation of GI bleeding, although with varying frequency over time and with the advent of colonoscopy and scintigraphy.^{[24, 25, 26, 27, 28, 29]}

Screen-film arteriography can demonstrate bleeding at rates as low as 0.5 mL/min in dogs, although some authors claim that the actual detectable rate of bleeding in clinical conditions may be in the range of 1.0-1.5 mL/min. Digital subtraction angiography (DSA) has been reported to be more sensitive than conventional screen-film angiography. In 1988, Rees et al showed that DSA tended to be more sensitive than conventional angiography in depicted simulated extravasation in vitro; however, in the clinical study, DSA was severely limited in evaluation of the lower GI tract because of misregistration artifact caused by bowel motion.^[30] In another study, Kruger et al showed similar findings.^[31] The investigators reported that DSA was superior to conventional angiography, provided that it is performed with adequate parasympathicolysis and suspended respiration.

On angiograms, hemorrhage is identified as the extravasation of contrast material into the lumen of the bowel (see the following images). The contrast material extravasation can be free or pooling, and it persists during or even after the injection. The angiodysplasia has been described to have characteristic appearance. It appears as a vascular tuft, along with an early and persistent draining vein.

Degree of confidence

The clinical sensitivity of angiography has been reported variably in different studies.^[32] The typical values are around 60%. Attempts to identify predictors for positive angiographic findings have shown mixed results. In a retrospective study, Pennoyer et al did not identify any single useful predictor to increase the likelihood of obtaining a positive angiographic result.^[24] Evaluated factors included a history of previous GI bleeding, transfusions, orthostatic hypotension, and tachycardia. However, Nicholson et al found a perfect correlation between a systolic blood pressure of less than 100 mm Hg and a positive arteriographic result.^[33] Therefore, some clinicians advocate immediate arteriography rather than nuclear medicine imaging in hemodynamically unstable patients.

Pharmacologic techniques have been used to increase the diagnostic yield of arteriography. These include the use of heparin, vasodilators, and thrombolytics. The reported studies show a 33-65% increase in the yield of angiography. These small studies have not shown any significant complications, although larger studies are needed to prove safety of this method.