Management of Lower Gastrointestinal Bleeding

Chapter 54

Management of Lower Gastrointestinal Bleeding

Michael D. Darcy

Clinical Relevance

Lower gastrointestinal bleeding (LGIB) is much less common than upper gastrointestinal (GI) tract bleeding; the incidence of LGIB is only around 20.4 cases per 100,000 adults per year. However, it is associated with a 10% mortality rate,1 and though less common than upper GI bleeding, it more often demands the specialized skills of the interventional radiologist. These cases are not as easily managed by endoscopic techniques, and many gastroenterologists will not attempt endoscopy while the patient is actively bleeding, owing to the difficulty in clearing blood and achieving visualization. Interventional techniques afford the opportunity to localize hemorrhage more precisely and stop bleeding less invasively than surgery.


LGIB is frequently self-limiting, so the first task is to determine whether the patient is bleeding actively enough to make it likely an arteriogram will be positive. Actual output of blood is not always the best criterion. The patient can be grossly bleeding but not have hematochezia because of the large holding capacity of the colon. Alternatively, bloody output from the rectum can continue long after bleeding has stopped as old blood is expelled. Feingold et al.2 found that using pulse and blood pressure assessment alone was very predictive of a positive tagged red blood cell (TRBC) scan. In this study, 62% of unstable patients (defined as a heart rate >100 or systolic blood pressure <100 mmHg) had a positive TRBC, but only 21% of hemodynamically stable patients had positive scans. Similarly, an analysis of 88 patients showed three clinical factors that correlated with positive angiography.3 Angiograms were significantly more positive when (1) blood pressure was less than 90 mmHg (87% positive vs. 12% positive when blood pressure was higher), (2) transfusion requirements were 5 or more units of blood (84% vs. 15% for fewer transfused units), and (3) hemoglobin dropped more than 5 g/dL from prior readings (85% vs. 26% for lesser hemoglobin drops).

On the other hand, angiography should not be delayed until the patient is severely hypotensive. By the time a patient goes into shock, close to 40% of blood volume has been lost.4 The threshold for deciding to do angiography should also be modified by the patient’s cardiovascular and resuscitation history. An elderly patient with moderate cardiovascular disease may need a higher perfusion pressure to avoid end-organ ischemia and thus may poorly tolerate even mild hypotension. End-organ ischemia can also be exacerbated by the decreased oxygen-carrying capacity of blood hemodiluted by resuscitation fluids.

When patients are massively bleeding and hypotensive, the decision to proceed directly to angiography is fairly clear. When uncertain whether there is active bleeding, a TRBC scan may be worthwhile, since it is 5 to 10 times more sensitive than angiography at detecting bleeding. A review of over 600 TRBC scans showed that only 45% were positive.5 However, if a TRBC is negative, its higher sensitivity to bleeding provides evidence that an arteriogram will be very unlikely to demonstrate bleeding. Thus a negative scan is useful in determining the advisability of an arteriogram. Although sensitive to the presence of bleeding, TRBC scans are not as effective at localizing the bleeding site. Zuckerman et al.6 noted that TRBCs were falsely negative or provided incorrect localization in 3% to 59% of cases.

In recent years, computed tomographic angiography (CTA) has been proposed as a minimally invasive way to document and localize active bleeding, with several advantages over TRBC scans. Experimental work in a swine model has shown that CT can detect bleeding at a rate as low as 0.3 mL/min.7 Early studies811 demonstrated the ability of CTA to document active extravasation and localize LGIB. Subsequently, CTA was shown to have 79% to 91% sensitivity, 85% to 99% specificity, and 91% to 98% accuracy for detection of GI bleeding.1214 CTA trumps TRBC scans in that CTA can define the bleeding pathology in 52% to 75% of cases.10,15 This allows formulation of a therapeutic plan better tailored to the patient’s pathology (see Fig. e54-1image). Another advantage of CTA is its ability to depict arterial anatomy (Fig. 54-1). This should facilitate subsequent arterial catheterization as well as increase the ability to localize the bleeding site. In fact, the exquisite depiction of arterial anatomy allowed one sigmoid diverticular bleed to be embolized empirically, which previously would never have been considered in the LGI tract.16 Practical advantages of CTA over TRBC scans are that they tend to be more readily available and more rapidly performed.

Although generally reserved for acute active bleeding, angiography can occasionally be useful in patients with chronic low-grade bleeding in whom multiple endoscopies and other tests have failed to yield the diagnosis. Rollins et al.17 reported that in patients with chronic LGIB, angiography revealed either structural lesions that could account for the bleeding or actual extravasation in 44% of their patients.


Many of the contraindications to elective angiography become relative contraindications when weighed against the need to manage life-threatening bleeding.

Frank discussion about these risks and possible benefits should occur during the informed consent process. History of a life-threatening contrast allergy is a fairly firm contraindication to immediate angiography. Time can be taken to pretreat with steroids if bleeding is not massive, but the magnitude of the reaction must be weighed against the risk of exsanguination in each individual case. Similarly, the risk of renal failure in patients with preexisting renal insufficiency must be balanced against the need for possible life-saving embolization. Vigorous intravenous hydration can be started to reduce the risk of contrast nephropathy, but the patient’s cardiopulmonary status must be monitored to avoid fluid overload and pulmonary edema.

Coagulopathy is a relative contraindication because it increases the risk of access site bleeding. Even with severe coagulopathy, angiography can still be done with a plan to leave a sheath in the arterial access site until the coagulopathy has been corrected. The sheath can also be used to continuously monitor arterial pressure in the postprocedure period. Alternatively, closure devices can be used to manage the arteriotomy. Another reason coagulopathy is a relative contraindication is that it significantly lowers the efficacy of embolization.18 This is because emboli usually reduce blood flow, but complete cessation of flow requires formation of some thrombus around the emboli.


After accessing the femoral artery, the first piece of equipment deployed is a sheath with a hemostatic valve. This will make catheter exchanges easier, but the primary role of the sheath is to insure that arterial access can be maintained throughout the embolization procedure. Emboli can become jammed inside the lumen of the angiographic catheter so tightly that even a guidewire will not pass. If the occluded catheter was introduced through a sheath, it can be safely removed without loss of arterial access.

Angiographic catheters needed to catheterize the mesenteric vessels come in a variety of shapes. Although the catheter used is often a matter of personal preference, certain shapes are more favorable for specific vessels. A recurved catheter like an SOS or Simmons will work well for a caudally directed celiac or superior mesenteric artery (SMA). If the vessel origin comes off at a right angle or is even directed cephalad, then a less curved Cobra or Lev 1 catheter will work better. The inferior mesenteric artery (IMA) generally leaves the aorta in a steep caudal direction. That plus the smaller size of the aorta at this level makes the shape of a Rosch Inferior Mesenteric (RIM) catheter most appropriate for the IMA.

Angiographic catheters are intended to just engage the proximal part of the mesenteric artery, which suffices for diagnostic angiography. If bleeding is identified, these catheters are suitable for therapy only if the plan is to infuse pitressin into the main trunk of the artery. The bleeding focus is usually in the periphery of the vessel distribution, so if embolization is planned, superselective catheterization is required. A 3F microcatheter introduced coaxially through the 5F angiographic catheter is the tool of choice. Modern microcatheters are extremely flexible and often coated to decrease friction. This allows them to be advanced all the way out into the vasa recta in the bowel wall (Fig. e54-2). It is this ability to do superselective catheterization that has revolutionized interventional treatment of LGIB, since superselective catheter placement markedly improves the safety of embolization.

For LGI embolization, embolic agents have evolved along with catheters. When LGI embolization was in its infancy, autologous clot and gelfoam were the available embolic agents. Their use resulted in embolization of larger branch vessels and a high risk of colonic infarction.1921 Currently, microcoils and polyvinyl alcohol (PVA) particles are the most commonly used embolics for LGIB.

Microcoils are small enough to fit through a 3F microcatheter. Their major advantage is that they are very radiopaque and can be easily visualized to allow precise placement. Their main disadvantage is that coils form immediately after exiting the catheter, and there is little option for flow direction of the coil, so the microcatheter has to be able to be maneuvered right to the desired point of embolization. Some physicians will not embolize if the catheter tip cannot be advanced out beyond the marginal artery. In Bandi’s study,22 inability to advance the microcatheter that peripherally was a cause of technical failure in 27% of their cases.

PVA particles are injectable and flow directed. This allows treatment of a lesion even if the catheter cannot be advanced all the way to the bleeding site (Fig. e54-3), but also affords less control over where the emboli go. Thus a wider area of arterial occlusion is likely. Theoretically this might increase the risk of infarction, but this is not proven. To date, there have been no trials randomizing patients to either PVA or microcoils. Larger-diameter (at least 300-500 microns) PVA particles must be used. In an animal model, Kusano et al. showed that using particles smaller than 300 to 500 microns in the bowel led to a significant rate of infarction.23

Ethanol should not be used for LGI embolization. As a liquid, it penetrates to the capillary level and will lead to necrosis. In fact, there have been several case reports of colonic infarction due to ethanol that refluxed into the mesenteric circulation during renal tumor embolization.24,25

An embolic agent that shows some promise is the glue N-butyl cyanoacrylate (NBCA) (Cordis Corp., Miami, Fla.). Although this is a liquid, it can polymerize before reaching the capillary level. In one report,26 no ischemic complications followed NBCA embolization of two LGI bleeds (one middle colic, one right colic). NBCA was successful in cases where embolization with other embolic agents had failed.

Anatomy and Approach

The primary arterial supply to the LGI tract is from the SMA and IMA. The SMA generally supplies the entire small bowel as well as the cecum, ascending colon, and a variable length of the transverse colon. From the splenic flexure down to the rectum is usually IMA territory. Which artery is studied first depends on how well pre-angiographic studies have localized the bleeding. If there is a strong suspicion that the pathology is in a certain bowel segment, the artery that normally supplies that portion should be studied first. If the location of the pathology is uncertain, the IMA can be studied first to avoid overlap between this vascular distribution and excreted contrast that will accumulate in the bladder as the exam progresses.

Clinically significant arterial anatomic variation is more common in the upper GI tract. A rare anatomic variant that must be considered during LGIB cases is a middle colic branch arising from the dorsal pancreatic branch of the celiac artery. With this variant, a celiac arteriogram will be required to evaluate a portion of the transverse colon. This variant can be suspected by comparing the vascular distribution of the IMA and SMA and looking for gaps between them in the region of the distal transverse colon.

In elderly patients with atherosclerotic changes in the aorta, occlusion of the SMA or IMA origin is not uncommon. In cases of ostial occlusion, one should look for collateral supply. The pancreaticoduodenal arcade generally allows free communication between the SMA and celiac artery. Rarely celiac-SMA collateral flow occurs through the arc of Buhler, an embryonic communication between the main trunks. The SMA and IMA can communicate through the marginal artery of Drummond as well as the arc of Riolan (when present). The IMA can also receive collateral supply from middle hemorrhoidal braches off the internal iliacs (Fig. e54-4). Awareness of these collateral pathways is important to knowing which vessel to inject when trying to opacify a target artery with an occluded ostium. When utilizing collateral flow, delayed filming is necessary to allow the contrast to reach the target vessel.

Technical Aspects

Generally, arteriography for LGIB should start with selective catheterizations (see Table 54-1). Aortography is usually not necessary.

It often does not deliver sufficient contrast to the mesenteric vessels to adequately demonstrate extravasation. Even if extravasation is seen on the aortogram, overlap with multiple opacified arteries may make it difficult to determine precisely where the bleeding arises from. However, aortography can be useful to interrogate the mesenteric vessel origins when these vessels are not easily catheterized and an ostial occlusion is suspected.

Without a lateral aortogram, it is usually not possible to ascertain the orientation of the vessel origins prior to engaging them with a catheter, so often physicians will start with their favorite catheter and then switch to another shaped catheter if the first selection fails to adequately engage the target artery. The selective catheter should be advanced far enough into the vessel to prevent recoil out into the aorta, but it should not be advanced so far as to be injecting beyond proximal side branches.

Stay updated, free articles. Join our Telegram channel

Dec 23, 2015 | Posted by in INTERVENTIONAL RADIOLOGY | Comments Off on Management of Lower Gastrointestinal Bleeding

Full access? Get Clinical Tree

Get Clinical Tree app for offline access