Lower gastrointestinal hemorrhage refers to blood loss originating distal to the ligament of Treitz that is symptomatic and necessitates hospital admission. It is a serious and potentially life-threatening situation. Although rectal hemorrhage can be seen in any age group, most patients requiring admission to the hospital are elderly and have coexistent medical problems. Because of these comorbidities, management is complex. Many patients stop bleeding spontaneously, but up to 30% experience bleeding again during or after their hospitalization. In the adult population, diverticulosis and vascular ectasias cause more than 90% of cases of lower gastrointestinal hemorrhage. Other less common causes are listed in Box 50-1 . Although the focus of this chapter will be on general evaluation and management of lower gastrointestinal hemorrhage, it is worthwhile to briefly examine and contrast the two most common causes.
Inflammatory bowel disease
Solitary rectal ulcer
In Western society, up to 65% of the population will have diverticulosis by age 85 years. Twenty percent of patients with diverticulosis coli will present with bleeding during their lifetime, and 5% will experience a severe hemorrhage. Although almost all of these patients stop bleeding spontaneously, bleeding will recur in 25%. Most diverticula are located in the sigmoid and descending colon, likely as a result of the sigmoid colon’s high intraluminal pressure. Despite this left-sided predilection, diverticular bleeding is distributed fairly equally between the right and left sides of the colon. The pathogenesis is believed to be injury to the submucosal arterial branches of the vasa recta that become stretched over the diverticulum and are then subject to trauma by the passage of stool. It has been speculated that the wider necks of the right-sided diverticula permit a greater length of the artery to be exposed to injury, thus increasing the percentage of bleeding from proximal diverticulosis. Diverticulitis is not usually associated with bleeding vasa recta.
Vascular ectasias of the colon are also believed to be acquired lesions because they are rarely observed in patients younger than 40 years. Also known as angiodysplasia or arteriovenous malformations, these lesions are predominantly located on the right side of the colon. It is likely that the increased wall tension of the cecum accounts for the presence of these lesions. As described by Boley, repeated low-grade obstruction of the submucosal veins over many years leads to the characteristic tortuous, dilated, thin-walled vessels, which can be identified both grossly and histologically. Arteriovenous connections occur relatively late in this process and result from increased pressure, leading to disruption of the precapillary sphincters. A high incidence of cardiac disease, especially aortic stenosis, has been observed in patients with vascular ectasias, and as many as 25% of patients who present with bleeding arteriovenous malformations are noted to have aortic stenosis. First described in 1958, Heyde syndrome describes a triad of aortic stenosis, an acquired coagulopathy, and anemia occurring as a result of intestinal angiodysplasia. Although bleeding has been reported to cease with aortic valve replacement, gastrointestinal hemorrhage is not an indication for open heart surgery. Rather, the decision to perform aortic valve replacement should be made on the basis of traditional indications, with surgery for lower gastrointestinal bleeding proceeding first if critical aortic stenosis is not present. Bleeding from vascular ectasias tends to be venous and, therefore, is not usually as brisk as that seen with diverticulosis. In more than 90% of patients, bleeding will stop spontaneously, but repeated episodes of bleeding are common, with an incidence that approached 85% in one study.
Initial Evaluation and Resuscitation
Because many patients who present with lower gastrointestinal hemorrhage lose a large amount of blood, resuscitation must accompany the initial evaluation. Despite the varied causes for lower gastrointestinal hemorrhage, the initial approach is standard.
In trying to quantify the degree and characteristics of a patient’s bleeding, it is important to realize that even small amounts of blood in the toilet can appear massive to the patient. Because blood is a cathartic, more importance can be attached to the frequency of bloody bowel movements prior to presentation than in trying to quantify the exact amount of blood loss. Significant lower gastrointestinal bleeding often causes hemodynamic instability, and it is essential to treat this condition while evaluating underlying causes. A lack of hemodynamic instability does not necessarily imply a minor bleed, however. A study from our institution revealed that, at presentation, 90% of patients with positive findings of an arteriogram were normotensive or hypotensive, and only 30% were tachycardic.
As the initial evaluation progresses, the basics of cardiopulmonary resuscitation must be followed. Large-bore intravenous catheters should be placed with infusion of a balanced salt solution. Blood samples are drawn for laboratory studies, including hemoglobin, hematocrit, coagulation studies, blood typing, and crossmatching. Attempts should be made to keep the patient normothermic. Placement of a Foley catheter permits accurate assessment of urinary output and assists in fluid replacement. Patients with massive hemorrhage, severe cardiac disease, or multiple comorbidities require intensive monitoring, which may include systemic arterial, pulmonary arterial, electrocardiographic, and oximetric monitors.
Early in the evaluation, a nasogastric tube should be placed. In many studies of lower gastrointestinal bleeding, as many as 10% of patients initially believed to be bleeding from a colonic source were ultimately determined to be bleeding from an upper gastrointestinal lesion. If clear bile is not returned upon nasogastric aspiration, an upper endoscopy should be performed as part of the evaluation. Even if clear bile is noted upon lavage, upper gastrointestinal bleeding can be seen in up to 16% of patients. The nasogastric tube can be left in place to use as access for a rapid mechanical bowel preparation to expedite a colonoscopy.
While resuscitation is proceeding, important information should be obtained from the patient’s history in relation to the risk for continued or recurrent bleeding. Asking about alcohol or aspirin ingestion, a prior history of gastrointestinal bleeding, the presence of any bleeding diathesis, coagulopathy from anticoagulation therapy, and comorbid diseases is extremely important.
Upon physical examination, particular attention should be directed to identifying stigmata of advanced liver disease. In addition, the presence of an abdominal mass may indicate an unsuspected colon carcinoma. Although diverticulitis is not commonly seen with bleeding diverticulosis, a finding of abdominal tenderness may suggest that possibility. More likely, however, would be a diagnosis of ischemic colitis or inflammatory bowel disease when a patient presents with abdominal pain, tenderness, and lower gastrointestinal hemorrhage.
Finally, rigid sigmoidoscopy is essential early in the evaluation of patients with lower gastrointestinal bleeding. It is generally performed in the emergency department to rule out an anorectal source of bleeding. Hemorrhoids associated with portal hypertension can bleed massively, and other low rectal or anal sources of bleeding may be treatable in the acute setting. Additionally, observation of the rectal mucosa may suggest a possible source of bleeding, such as infectious, inflammatory, or ischemic proctocolitis.
The diagnostic phase of lower gastrointestinal bleeding usually proceeds once the patient has been stabilized. However, because of the dynamic nature of colonic hemorrhage, diagnostic testing occasionally must be initiated while the patient is still being stabilized. In fact, aggressive diagnostic maneuvers can have the benefit of localizing bleeding before it has stopped. Identifying the source of bleeding matters, especially for the 25% to 30% of patients who will experience significant recurrent bleeding. Early aggressive diagnostic procedures may achieve this goal and permit nonoperative therapeutic maneuvers as well.
The most commonly used diagnostic studies performed for lower gastrointestinal bleeding are radionuclide scanning, multidetector computed tomography (CT angiography), angiography, and colonoscopy. In many patients, a combination of these tests will be needed, and occasionally, because of recurrent bleeding, it will be necessary to repeat them. Both angiography and colonoscopy can be of potential therapeutic benefit, and this advantage adds to their usefulness as diagnostic modalities.
Two types of radionuclide scans are available to image gastrointestinal bleeding. Initially, sulfur colloid was used as an intravascular marker that could not return to the vascular compartment once bleeding into the intestine had taken place. However, its rapid clearance by the reticuloendothelial system results in two distinct disadvantages. First, the patient must be actively bleeding at the time of the injection because more than 90% of the trace is cleared within 7 minutes. Second, accumulation of activity by the liver and spleen obscures evidence of bleeding from the colonic flexures.
A second technique, injection of technetium-labeled red blood cells, has supplanted sulfur colloid scanning as the nuclear medicine technique of choice. It is equally safe and effective, and its only disadvantage is the 30 to 40 minutes required to label the red blood cells. Technetium-labeled red blood cell scans are reported to be sensitive to bleeding rates as low as 0.05 to 0.1 mL per minute. The tagged red blood cells have an extended half-life, and scanning can take place for 24 hours after injection.
Although tagged red blood cell scanning has gained wide acceptance as a modality to detect gastrointestinal bleeding, it is arguable whether scintigraphy can localize a bleeding site reliably. In a review of 72 technetium-labeled red blood cell scans performed at our institution, 71% of the positive scans accurately localized the site of bleeding as confirmed by surgery, angiography, or endoscopy. This finding is supported by other studies in the literature, which average a false localization rate of 25%. Recently, it has been reported that the accuracy of scanning can be enhanced by performing dynamic scintigraphy, with stratification of results based on early radionuclide blushing. Ideally, scintigraphy should be performed while the patient has active bleeding, yet despite a national trend toward increased 24-hour hospital staff coverage, accommodating on-request scintigraphy continues to be problematic in many hospitals. Even in large centers that employ on-call nuclear medicine technologists, it is difficult to perform scintigraphy in a timely manner during nights and weekends. This suboptimal usage of scintigraphy has resulted in an increased number of false-negative studies. Additionally, despite the theoretical benefit of prolonged bleeding observation (repeat scintigraphy), it has been suggested that scans that are positive within several hours of injection produce more accurate localization, whereas, because of the rapid antegrade and retrograde movement of extravasated blood, the accuracy of delayed-positive scans decreases substantially. Given these factors, the relatively high false localization rate for scintigraphy is not surprising. Therefore, basing a bowel resection on the results of red blood cell scintigraphy alone should be discouraged, and confirmation should be obtained prior to surgery if possible.
Radionuclide scanning has also been suggested as a cost-effective screening tool prior to angiography. Given its increased sensitivity, relative safety, and decreased cost compared with angiography, scintigraphy should be an ideal test to perform to increase the yield of positive angiograms. Unfortunately, it has been difficult to support this supposition in practice, possibly because the delay in obtaining an angiogram while performing scintigraphy allows the “window of opportunity” to find a bleeding vessel to pass. A review of our experience with angiography showed that a prior positive nuclear scan did not increase the percentage of positive angiograms compared with use of angiography as the initial diagnostic procedure, although successful anatomic localization did permit diminished contrast material administration during angiography.
In conclusion, the exact role of radionuclide scanning remains unclear. The examinations are minimally invasive and inexpensive, have low complication rates, and may alert physicians to patients who are likely to require surgery. However, at this point, it is not safe to limit a workup and plan treatment based on the results of a radionuclide scan alone. Furthermore, its value as a screening test for the cost-effective use of angiography remains questionable, and many institutions are now using alternative modalities as first-line diagnostic tools.
Multidetector Computed Tomography
In recent years, contrast-enhanced multidetector computed tomography (MDCT) increasingly has been described as being useful in the diagnosis of active lower gastrointestinal bleeds. Studies have found that MDCT is capable of detecting colonic bleeding at rates between 0.3 and 0.4 mL per minute. This finding is significant because the rate of bleeding detected is lower than the 0.5 mL per minute detected by mesenteric angiography and approaches bleeding rates detected by radionuclide scanning. Additional benefits include the 24-hour availability of MDCT at most hospitals, as well as decreased artifact from bowel gas and peristalsis that can result in false-positive results, specifically during mesenteric angiography. A review of the literature shows that MDCT has a greater than 85% sensitivity and 95% specificity, with 95% overall accuracy for localizing gastrointestinal bleeding. Increased accessibility to MDCT results in a greater likelihood that the study will be performed while the patient has active bleeding, and it theoretically increases the likelihood of having a positive localizing study. After a positive MDCT study, patients should undergo mesenteric angiography and embolization. Concerns of precipitating renal insufficiency as a result of use of this algorithm have been raised because patients receive duplicate intravenous contrast dye loads. In our experience, however, these concerns have not been substantiated. Similar to scintigraphy, we have found that positive localization on MDCT allows for decreased dye administration with the ability to target mesenteric angiography to a specific vascular territory.
Selective mesenteric angiography has become widely used for lower gastrointestinal hemorrhage because it has the benefit not only of being diagnostic but also frequently therapeutic. By localizing bleeding to a specific vessel, angiography tremendously facilitates surgery. Furthermore, transcatheter therapy, either pharmacologic or by embolization, can successfully treat the source of bleeding, thus avoiding surgery.
Minimal preparation is necessary for angiography, but because it must be assumed that the patient is actively bleeding, continuous monitoring is necessary. Resuscitation should be continued while a Foley catheter prevents the bladder from filling with contrast material. Selective injection is performed first through the superior mesenteric artery, because bleeding is most likely in this distribution. Injections of the inferior mesenteric artery and celiac axis follow because the ultimate source is proximal to the ligament of Treitz in up to 10% of patients with presumed lower gastrointestinal bleeding. Bleeding can be detected at rates as low as 0.5 to 1.0 mL per minute. Although extravasation of contrast material is unequivocal evidence for a bleeding source, angiography also can detect other lesions such as a tumor blush or angiodysplasia. Extravasation is seen in fewer than 15% of patients with vascular ectasias; however, angiographic signs of their presence include a prominent early-filling vein, a vascular tuft, or a late-draining vein.
When angiography documents a bleeding site, transcatheter therapy can be instituted in an attempt to stop the bleeding. The two alternatives are intra-arterial infusion of vasopressin or transcatheter embolization of the vessel. Superior mesenteric arterial infusion of vasopressin reduces splanchnic blood flow by up to 65%, thus, it is hoped, allowing a hemostatic plug to form in the bleeding vessel. Vasopressin infusion is initiated at 0.2 units per minute with repeat angiography performed 20 minutes later to document the effectiveness of the infusion. If necessary, the rate can be increased to 0.4 units per minute, after which the marginal benefit is offset by its adverse effects. Because cardiac complications are reportedly as high as 43%, these patients require continuous cardiac monitoring, preferably in an intensive care unit. Although success rates up to 90% have been reported, up to 50% of patients will experience recurrent bleeding upon cessation of therapy. Our experience of a 41% rebleeding rate has led us to conclude that the major benefit of vasopressin is in stabilizing a patient’s clinical situation, thus permitting a semi-elective resection to be performed.
An alternative to vasopressin infusion is transcatheter embolization of the affected vessel. This technique provides permanent control of the bleeding vessel and avoids the troublesome adverse effects of vasopressin. Early techniques used temporary agents placed in a fairly central location so as to spare distal communicating vessels and permit their eventual recanalization. Although theoretically transcatheter embolization was expected to reduce the complication of intestinal ischemia, ischemia still occurred in up to 20% of cases.
The development of small-caliber angiographic catheters has permitted superselective catheterization of peripheral vessels. This advance has permitted a more selective therapeutic intervention and has limited the potential for widespread intestinal ischemia. Platinum-fibered coils or polyvinyl alcohol particles are used as permanent embolic agents. To date, we have used this approach in nearly 200 patients with angiographically proven lower gastrointestinal bleeding. In our experience, recurrent bleeding after seemingly successful selective embolization occurred in approximately 10% of patients. For these patients, repeat embolization remains an option. Similarly, approximately 8% of patients at our institution experienced postembolization ischemia requiring urgent surgical intervention.
Colonoscopy is an extremely valuable diagnostic tool in the evaluation of lower gastrointestinal bleeding. With few exceptions, it should be performed at some time in the evaluation of any patient presenting with acute rectal bleeding. The major issue relates to timing of the colonoscopy during the patient’s hospitalization. There are many proponents of immediate colonoscopy upon admission to the hospital without a colonic preparation. This approach is theoretically possible because blood is an excellent cathartic, and early colonoscopy with “jet” irrigation of the colon has a high likelihood of determining the site of bleeding. A large amount of residual clot is usually present, but the bleeding site can be distinguished either by spurting blood or bright red blood. Blood can travel retrograde up the colon, and thus a clot in the cecum does not necessarily mean that the bleeder is there. However, if the blood stops at a certain level, then the bleeding site is at least distal to that place. Success rates in finding the bleeding site as high as 80% have been reported, but it should be noted that this examination is technically very difficult and has several drawbacks. Even with highly skilled endoscopists performing “emergency” colonoscopy, the cecal intubation rate is less than that for elective procedures. Furthermore, patient instability can severely limit the ability to administer sedatives and analgesics. Finally, residual blood markedly reduces the ability to visualize mucosal detail, which is critical in the diagnosis of angiodysplasia.
A preferable approach is to decide whether a patient has stopped bleeding early in the evaluation period. If the patient is hemodynamically stable and not passing fresh blood per rectum, the patient is prepared for an “urgent” colonoscopy. Polyethylene glycol is administered over 4 to 6 hours, preferably through a nasogastric tube if it has been left in place. A colonoscopy then can be performed in a much more controlled setting. The patient is now hemodynamically stable and can be sedated, permitting a safer procedure. Mucosal detail is usually comparable to that of elective procedures, although high-pressure irrigation to remove an adherent clot must be available.
It is unusual to detect an actively bleeding lesion during colonoscopy performed in this setting. However, suspicious lesions other than diverticula have been reported in up to 50% of patients. Neoplastic lesions can be removed or biopsied at the time of the colonoscopy. Areas of ulceration or active colitis also can be biopsied. Some controversy exists about whether angiodysplastic lesions that are not actively bleeding should be treated prophylactically. We treat them if no other potential bleeding sources are found during the patient’s evaluation and if there is a high degree of suspicion that the vascular ectasia has recently bled. When numerous nonbleeding angiodysplastic lesions are present, their distribution is noted and no active treatment is undertaken. The method of coagulation that is preferred in our gastrointestinal unit is to use the heater probe to cauterize around the lesion and finally to cauterize the central area. Care is taken to use low-power settings, especially in the right colon, which has a relatively thin wall. Other methods of coagulation that can be utilized are bicap electrocautery, needle injection, argon plasma coagulation, and neodymium: yttrium-aluminum-garnet (Nd:YAG) laser therapy. When a diverticulum has stigmata of bleeding, endoscopic clipping has become a favored treatment. Typically, epinephrine is injected into the diverticulum to evert it for better access. This maneuver is then followed by clipping of the vessel or the entire diverticular orifice. Tattooing of suspected bleeding sites is also recommended for future localization should recurrent bleeding occur.
It is important to perform colonoscopy in all patients who have undergone therapeutic angiography. Although the patient may have stopped bleeding after either administration of vasopressin or embolization, there is a 5% to 30% incidence of neoplastic lesions in this setting. Furthermore, the mucosa can be evaluated for evidence of ischemia, especially if abdominal pain or tenderness develops.