Occult and Unexplained Chronic Gastrointestinal Bleeding

Repeat EGD and Colonoscopy

Second-look upper endoscopy may be helpful in identifying bleeding lesions potentially overlooked or unrecognized at the time of the initial endoscopic evaluation. Data suggest that 2% to 25% of lesions causing obscure gastrointestinal bleeding may be overlooked and are within reach of a standard upper endoscope. More recent studies using double-balloon enteroscopy (DBE) and CE have also confirmed these findings. Commonly overlooked lesions in the upper GI tract include Cameron's erosions or ulcerations in large hiatal hernias, isolated gastric fundal varices, peptic ulcers, AVMs including gastric antral vascular ectasia (watermelon stomach), and Dieulafoy's lesion. An aorto-enteric fistula should always be considered in patients with prior abdominal aortic aneurysm repair. If second-look upper endoscopy is selected, push enteroscopy may be the chosen method to more thoroughly examine the distal duodenum and proximal jejunum in addition to the more proximal upper GI tract.

Although the reported yield of repeat colonoscopy is low (6%–23%), it may be helpful in selected patients when the original colonoscopy examination was documented to have had a mediocre or poor preparation, the extent of the examination was not to the cecum, the patient is older and risk of a previously missed neoplastic lesion is considered, or the terminal ileum was not previously evaluated. Lesions missed or unrecognized during colonoscopy may include AVMs, polyps, solitary rectal ulcers, rectal varices, and neoplasms.

End-Stage Renal Disease

AD is the second most common cause of GI bleeding in patients with end-stage renal disease. These lesions account for approximately 20% of upper GI bleeds and 30% of lower GI bleeds, and approximately 50% of recurrent upper GI bleeds. In a prospective study of upper GI hemorrhage over a 50-month period, vascular ectasia was the etiology of upper GI hemorrhage in 13% of patients with renal insufficiency and was the etiology of bleeding more often in patients with renal insufficiency than in patients with normal renal function. The prevalence of vascular ectasia as a cause of upper GI bleeding was related to the duration of renal failure and the requirement of hemodialysis. Although the reason for the increased prevalence among patients with end-stage renal disease is unknown, a possible explanation is that there is an increased risk of bleeding associated with uremia-induced platelet dysfunction (aggregation and adhesion).

von Willebrand's Disease

von Willebrand's disease is a bleeding disorder that results from a qualitative or quantitative defect in von Willebrand factor (vWF). vWF is a complex multimeric glycoprotein present in platelets, plasma, and subendothelium, and is essential to platelet adhesion and aggregation at the site of vascular injury. In a study of patients with both bleeding and nonbleeding angioectasias of the GI tract and control patients with colonic diverticular hemorrhage, Veyradier et al (2001) showed that most patients with bleeding angioectasias of the GI tract lack the largest multimers of vWF induced by a latent acquired form of von Willebrand's disease. Because these specific multimers are the most effective in inducing platelet aggregation in high shear stress that is commonly present in the microcirculation of angioectasias, it was concluded their deficiency contributes to active bleeding.

Aortic Stenosis

Approximately 50% of patients with bleeding vascular ectasias have evidence of cardiac disease, and 25% have been reported to have aortic stenosis (AS). Bleeding from angiodysplasias in patients with AS (Heyde's syndrome) has been repeatedly reported but is highly controversial. Support for this hypothesis comes from evidence of improvement or cessation of chronic GI bleeding in the vast majority of patients with AS after aortic valve replacement. This effect was sustained for up to 12 years after surgery in the largest case series of 91 patients.

Two possible explanations have been proposed to explain this observation. Patients with AS may develop an acquired form of von Willebrand's disease, which can be reversed after aortic valve replacement. Patients with AS may be more likely to bleed from existing AD. Another explanation is that existing AD may bleed as a result of ischemic necrosis in patients who have a low cardiac output. However, this explanation is inconsistent with the observations that bleeding AD have not been observed with other forms of heart disease associated with a low cardiac output and that a low cardiac output is a late complication of AS.

Several retrospective, uncontrolled studies and a prospective, controlled investigation do not substantiate a causative role or association of aortic valve diseases with colonic angioectasias. Replacement of the aortic valve for control of bleeding secondary to these vascular lesions is not universally accepted. A logical approach to patients with both lesions is to treat the colonic lesion first endoscopically, regardless of whether the patient's cardiac status warrants surgery. If valve replacement is necessary and endoscopic therapy is unsuccessful, further endoscopic or surgical treatment of the colonic AD should be delayed until after cardiac surgery. Further attempts at endoscopic treatment or surgical resection are indicated if bleeding recurs.

Progressive Systemic Sclerosis

Vascular lesions are a prominent feature of progressive systemic sclerosis, especially in the CREST variant. In patients with progressive systemic sclerosis, sites most frequently affected by telangioectasias are the hands, lips, tongue, and face, but gastric, intestinal, and colorectal lesions have been reported. These lesions may be the source of occult or clinically significant bleeding and are best treated by endoscopic coagulation.

Left Ventricular Assist Devices

Patients with left ventricular assist devices have higher rates of GI bleeding, most commonly from angioectasias in the upper digestive tract and small bowel.

Endoscopic Imaging: Standard Upper Endoscopy and Colonoscopy

Angioectasias have a characteristic appearance of a cherry-red, fern-like pattern of arborizing, ectatic blood vessels radiating from a central vessel ( Fig. 18.1 ). This pattern should be specifically looked for because angioectasias may be confused with other erythematous mucosal lesions or with normal vessels ( Table 18.2 ). Because traumatic and endoscopic suction artifacts may resemble vascular lesions, all lesions must be evaluated immediately on insertion, rather than during withdrawal. “Anemic halos” are often seen surrounding angioectasias of the bowel. Although these halos do not differentiate the various types of vascular lesions, they distinguish true vascular lesions from artifacts. Newer alternative imaging options, such as narrow-band imaging, allow precise discrimination of vascular structures from artifactual mucosal hemorrhage. Punch biopsy samples of vascular lesions obtained during endoscopy are usually nonspecific; the bleeding induced by performing biopsies of these abnormalities is not justified.

Angioectasias may be difficult to visualize during colonoscopy in patients who do not have an optimal bowel cleaning. Because the appearance of angioectasias is influenced by blood pressure, blood volume, and state of hydration, these lesions may not be evident in patients with severely reduced blood volumes or who are in shock until red blood cell and volume deficits are corrected. Cold water lavage of the colon, as is sometimes done to clean the luminal surface from debris during colonoscopy, reportedly may mask these lesions. Meperidine also has been implicated in masking lesions because of a transient decrease in mucosal blood flow. Minimizing use of meperidine and reversal with naloxone to increase the yield of detection has been advocated by some clinicians. Naloxone has been reported to enhance the appearance of normal vasculature in approximately 10% of patients and to cause angioectasias to appear (2.7%) or increase in size (5.4%). Reversal of narcotic analgesia may affect the comfort of an examination, particularly if therapeutics are performed.

Assessment of the Small Bowel

Some patients presenting with GI bleeding have no source found by upper endoscopy and colonoscopy, even after second-look endoscopy. In these cases of unexplained chronic GI bleeding, endoscopic examination of the small bowel has been limited by several factors. The length of the small intestine, in addition to its free intraperitoneal location, vigorous contractility, and overlying loops, confounds the usual diagnostic techniques, including barium studies, endoscopic intubation, and identification of specific sites by special imaging techniques of nuclear medicine scans and angiography. The bleeding rate may be slow or intermittent, not allowing identification by either angiography or radionuclide bleeding scan. Because of the inability to localize a bleeding site in the small bowel, patients with obscure GI bleeding from a small bowel source typically presented with prolonged occult blood loss or recurrent episodes of melena or maroon stool without a specific diagnosis. In this group of patients, previous noninvasive tests, such as small bowel follow-through, radioisotope-labeled red blood cell scan, and push enteroscopy, have had suboptimal diagnostic yields of 20% to 40%. Invasive methods, such as laparotomy or intraoperative enteroscopy, may improve the yield up to 70%.

An early diagnosis of the bleeding site has become possible recently with the development of several endoscopic modalities for assessment of the small bowel that include wireless CE and deep small bowel enteroscopy (DBE, single-balloon enteroscopy, and spiral enteroscopy [SE]).

When performing a second-look endoscopy, useful adjunctive diagnostic maneuvers include use of a cap-fitted endoscope to examine blind areas, such as the high lesser curve, under the incisura angularis, and the posterior wall of the duodenal bulb; use of a side-viewing endoscope to examine the ampulla in patients with suspected pancreaticobiliary pathology; and use of a push enteroscope to examine the C-loop of duodenum carefully after injection of glucagon. Although the yield is low (6%), repeat colonoscopy may be useful in the setting of prior poor bowel preparation. Use of naloxone may improve the detection of colonic angioectasias that were not obvious at the index examination. When all the findings on standard examinations (upper endoscopy and colonoscopy) are negative, the small bowel may be presumed to be the source of blood loss. CE should be the third test in the evaluation of patients with GI bleeding ( Fig. 18.2 ). CE, balloon, and spiral enteroscopy, as well as other diagnostic modalities for suspected small bowel bleeding, are discussed in detail in Chapter 17 : Middle Gastrointestinal Bleeding.

Diagnostic Algorithm for Endoscopy in Suspected Middle Gastrointestinal Bleeding

Similar to other authors, we recommend CE as a first-line investigation over balloon-assisted enteroscopy in view of its convenience, higher chance to visualize the entire small intestine, and similar diagnostic yield. A recent updated and revised meta-analysis of 10 studies involving 642 patients demonstrated that the pooled overall diagnostic yield for video CE (VCE) and DBE was 62% and 56%, respectively. The diagnostic yield of DBE was significantly higher when performed after a positive VCE compared with negative VCE (75% vs. 27.5%). No test substitutes for good clinical judgment, however, and all small bowel diagnostic studies must be considered in difficult cases of obscure GI bleeding, particularly in a young patient.

In the setting of unexplained overt GI bleeding, CE should be performed close to the bleeding episode. In case of negative CE, we recommend that a second endoscopist reread the CE study. We also recommend a second-look EGD with special attention to areas less optimally examined by CE, especially the duodenum, before concluding with a final diagnosis of obscure GI bleeding. Second-look CE has also been suggested for patients with a prior nondiagnostic CE. Bar-Meir et al (2004) reported that 7 out of 20 patients (35%) who underwent second-look CE had positive or suspicious findings. Another study showed that patients with a nondiagnostic CE test would benefit from a second-look CE if the bleeding presentation changed from occult to overt, or if the hemoglobin value decreased 4 g/dL or more.

Pharmacologic Treatment

Pharmacotherapy should be considered whenever endoscopic therapy, surgical intervention, or angiographic therapy is impractical or ineffective, such as in patients in whom the source and the etiology of bleeding are unknown or the pathology is too diffuse to be amenable to ablative therapies.

Nonbleeding AD detected during evaluation of occult bleeding or iron deficiency anemia should be considered to be causative, if no other cause is found. In patients with occult bleeding, bleeding from AD may be more likely in patients who have multiple lesions and a bleeding diathesis (e.g., anticoagulation). As a result, a graduated approach with primary or adjunctive iron-replacement therapy may be initiated, with pursuit of more aggressive therapeutic options guided by the clinical circumstances. The aims of iron-replacement therapy should be to restore hemoglobin levels and mean corpuscular volume to normal, and to replenish body stores.

Numerous oral iron preparations are available, although ferrous sulfate and ferrous gluconate are the preferred forms because of low cost and high bioavailability. A liquid preparation is often tolerated better. Ascorbic acid enhances iron absorption. Parenteral iron should be used when there is intolerance to at least two oral iron preparations or noncompliance, when there is a suboptimal clinical response secondary to suspected iron malabsorption, and when blood transfusion becomes difficult to achieve because of excessive antibodies that challenge crossmatching.

Estrogen-progesterone combination hormonal therapy has been used to treat patients with angioectasias of the GI tract. The effect, which is not immediate, seems to be estrogen dose–dependent. Hormonal therapy acts by enhancing microvascular circulation, coagulation, and vascular endothelial integrity. The most common combination schedule has been ethinyl estradiol 0.01 to 0.05 mg and norethisterone 1 to 3 mg. This therapy should be used in 6-month courses with pauses to reduce the incidence of adverse effects, mostly secondary to the estrogen component. The results of several prospective, controlled trials examining hormonal therapy have been divergent. In a long-term observational study, combination hormonal therapy was shown to stop bleeding in patients with occult GI bleeding of obscure origin, likely to have resulted from small bowel AD. Although uncontrolled studies suggest that combination estrogen-progesterone therapy prevents bleeding episodes secondary to AD, the evidence from the largest placebo-controlled trial to date suggests that this therapy is ineffective. These authors considered that efficacy of hormonal therapy in these patients remains to be proven by a large, randomized, placebo-controlled trial with long-term follow-up.

Reports of efficacy of octreotide in the treatment of AD have been limited to case reports and small series in which a response has been reported in some patients. Octreotide produces vasoconstriction secondary to inhibitory effects on growth hormone and multiple GI vasodilator hormones, and markedly reduces splanchnic blood flow. Its antiangiogenic properties have been shown in different tissues (eye, placenta, liver, and GI neuroendocrine tumors); however, applicability and utility in obscure GI bleeding remain unknown.

The dosage of octreotide can be tapered to the lowest quantity that prevents rebleeding. Response is immediate, and the drug can be administered intravenously (50 µg/hr) or subcutaneously (50–100 µg two or three times a day). Its subcutaneous administration and its longer half-life (90–100 minutes) make octreotide superior to somatostatin and allow use in the outpatient setting. A 6-month course of therapy has been used to treat most patients. A meta-analysis reported by Jackson and Gerson in 2014 including four studies with a total sample size of 77 patients concluded that there is a significant effect on bleeding cessation rates for the somatostatin analogs with a pooled odds ratio of 14.52 (95% confidence interval [CI]: 5.9–36). The standard mean reduction in number of transfusions after 1 year of therapy was 0.55 (95% CI: 0.29–0.82). However, it is difficult to ascertain whether this represents the true effect of octreotide, as this meta-analysis had some inherent weaknesses.

Sandostatin LAR Depot (Novartis Pharmaceuticals Corporation, East Hanover, NJ) is a depot formulation of octreotide for long-term maintenance therapy currently approved for acromegaly and GI and pancreatic neuroendocrine tumors. Compared with conventional octreotide, Sandostatin LAR Depot is administered at a dose of 20 mg intramuscularly once a month with a similar efficacy and safety profile. Sandostatin LAR Depot does not require hospital admission, which makes it an attractive outpatient option for long-term therapy in patients with chronic GI bleeding, but is more expensive than conventional octreotide.

Thalidomide is a drug with powerful immunomodulatory, antiinflammatory, and antiangiogenic effects that was withdrawn from the market in the 1960s because of its teratogenicity. It has been reintroduced more recently and has been demonstrated to inhibit VEGF-dependent angiogenesis. Thalidomide administered orally at a variable dose of 100 to 300 mg/day is an innovative and promising therapeutic option for GI bleeding associated with AD and can be used in refractory cases or when other drugs or therapies are contraindicated. Thalidomide is contraindicated in patients with peripheral neuropathy and pregnant women and women with childbearing potential because of its teratogenic effects, and it should be used cautiously in patients with cardiovascular or neurologic disorders and hepatic or renal impairment. Owing to its immunosuppressant activity by blocking tumor necrosis factor, use of thalidomide may also be discouraged in patients at risk for infection or chronic infectious disease, especially patients with HIV infection. In all these clinical settings, Sandostatin LAR Depot may be safer than thalidomide.

Two recent studies explore the use of thalidomide for gastrointestinal angiodysplastic lesions (GIADs) in a series of patients. The first study from China in 2011 randomized patients to 100 mg of thalidomide versus iron therapy for 4 months and demonstrated efficacy when rebleeding rates and transfusional requirements were analyzed. The second study from Spain treated 12 refractory patients and demonstrated an increase in hemoglobin values after 4 months.

Other agents that have been used for management of angioectasias with limited available data include tranexamic acid, danazol, desmopressin, and recombinant activated factor VII.

Endoscopic Treatment

The goal of endoscopic therapy is thrombosis of the bleeding vessel. Studies directly comparing the effectiveness of the different approaches have not been performed. The approach depends on the location of the lesion, the experience of the endoscopist, and the availability of equipment.