Vascular Lesions of the Gastrointestinal Tract

Colonic Angioectasia

AE of the colon is a distinct clinical and pathologic entity. It is the most common vascular abnormality of the GI tract and probably the most frequent cause of recurrent or chronic lower intestinal bleeding in persons older than 60 years of age. AEs are acquired with aging, and there does not appear to be a gender predominance. In contrast to congenital or neoplastic vascular lesions of the GI tract, acquired AEs are not associated with lesions of the skin or other viscera, although some 10% of patients with colonic AE have similar lesions in the small intestine when evaluated by angiography or enteroscopy. AEs almost always are confined to the cecum or ascending colon, often are multiple rather than single, and usually are smaller than 10 mm in diameter. They are seldom identified by the surgeon at operation or by the pathologist using standard histologic techniques, but usually they can be diagnosed by angiography (discussed later); colonoscopy ( Figs. 38.1 and 38.2 ); or helical CTA.

The roles of CT and MRI for vascular lesions of all types are evolving but are certain to increase as these sophisticated modes of diagnosis become more widely available; it is also clear that conventional angiography now is more important for therapy than for diagnosis. To determine the precise nature of a vascular lesion, histologic examination, with or without injection studies of the vasculature, is necessary. For example, in one publication in which histologic confirmation of vascular lesions was not performed, AEs reportedly occurred distal to the hepatic flexure in 46% of patients ; subsequent review of tissue sections from the supposed AEs in the small bowel or left colon revealed histologic changes different from those of AEs in the right colon (personal review by S.J. Boley and L.J. Brandt).

Pathology

Histologic identification of AEs is difficult unless special techniques are used. Although usually less than one third of lesions are found by routine pathologic examination, almost all can be identified by injecting the colonic vasculature with silicone rubber, dehydrating the cells with increasing concentrations of ethyl alcohol, clearing the specimen by immersing it for 24 hours in a bath of methylsalicylate, and then viewing the specimen by dissecting stereomicroscopy ( Fig. 38.3 ). In a study using these methods, surgically resected colons were found to have 1 or more mucosal AEs that measured 1 to 10 mm in diameter. AEs were usually multiple, and all were located in the cecum and ascending colon.

Microscopically, mucosal AEs consist of ectatic, distorted, thin-walled venules, capillaries, and arterioles. The earliest abnormality is the presence of dilated, tortuous, submucosal veins ( Fig. 38.4A ), often in areas where overlying mucosal vessels appear normal. More advanced lesions show increasing numbers of dilated and deformed vessels traversing the muscularis mucosa and involving the mucosa (see Fig. 38.4B and C ) until, in the most severe lesions, the mucosa is replaced by a maze of distorted, dilated vascular channels (see Fig. 38.4D ). Enlarged arteries and thick-walled veins occasionally are seen in advanced lesions, in which the dilated arteriolar-capillary-venular unit has become a small arteriovenous (AV) fistula because of loss of prearteriolar sphincter function. Large thick-walled arteries, however, are more typical of congenital AVMs.

Pathogenesis

The previously described studies using injection and clearing techniques indicated that AEs are acquired with aging and that they represent a unique clinical and pathologic entity. Clinically, AEs are frequently identified at colonoscopy in older adults and in injected colons resected from older patients with no history of bleeding. Boley postulated that the likely cause of AEs is partial, intermittent, low-grade obstruction of submucosal veins at the site where these vessels pierce the muscular layers of the colon ( Figs. 38.5 and 38.6 ). He then went on to propose a schema for their development, suggesting that repeated episodes of transiently elevated pressure during muscular contraction and distention of the cecum over many years result in dilatation and tortuosity of the submucosal vein and, later, of the venules and capillaries of the mucosal units that drain into it. He further suggested that, ultimately, the capillary rings dilate, the precapillary sphincters lose their competency, and a small AV fistula is produced; the latter is responsible for the “early-filling vein,” which was the original angiographic hallmark of this lesion ( Fig. 38.7 ). Prolonged increased flow through the AV fistula can then produce alterations in the arteries supplying the area and in the extramural veins that drain it. This developmental concept of the cause of AEs was based on the finding of (1) a prominent submucosal vein, either in the absence of any mucosal lesion, or underlying only a minute mucosal AE supplied by a normal artery; (2) dilatation of the veins, starting where they traverse the muscularis propria (see Fig. 38.5 ); and (3) previous studies showing that venous flow in the bowel may be diminished by increases in colon motility, intramural tension, and intraluminal pressure. Following this logic, the prevalence of AEs in the right colon can be attributed to the greater tension in the cecal wall compared with that in other parts of the colon, according to Laplace’s principle: T ∝ PR (where T is tension, P is intraluminal pressure, and R is radius).

An alternative concept for the development of AEs is based on the demonstration that AEs have been shown to express vascular endothelial growth factor (VEGF) and its receptors along their endothelial lining in surgical specimens from patients who have undergone colectomy for recurrent bleeding ; this indicates a proliferative phase of angiogenesis. VEGF and VEGF receptor 1 have been shown to be upregulated by hypoxia, and therefore a role also has been suggested for hypoxia in the pathogenesis of AEs. It further has been proposed that von Willebrand factor (vWf) regulates angiogenesis through multiple “cross-talking” pathways that involve VEGF receptor 2 signaling, angiopoietins, and integrin ανβ3. In a mouse model, inhibition of vWf in endothelial cells results in increased in vitro angiogenesis and increased VEGF receptor proliferation and migration, coupled to decreased integrin ανβ3 levels and increased angiopoietin release. Further research still is needed to clarify the pathophysiology of AEs.

Clinical Features and Associated Conditions

In 1961, Baum and colleagues used intraoperative angiography to show that cecal AEs may bleed. Today such an observation is well documented in daily practice by colonoscopy. Recent publications have cited AEs and diverticulosis to be responsible for 3% to 15% and 20% to 65%, respectively, of acute LGIB episodes (see Chapters 20 and 121 ). The problem of attributing bleeding to one or the other cause, when bleeding from the lesion is not demonstrated by colonoscopy or by extravasation of contrast material on vascular imaging studies, is compounded by the frequency and coexistence of these disorders without bleeding in people older than 60 years of age. The prevalence of diverticulosis is estimated to be as high as 50% in the population older than age 60. Mucosal and submucosal AEs of the right colon can be found by injection studies of colons removed at surgery in more than 25% and 50%, respectively, of patients in this age range without evidence of bleeding. In large series of colonoscopic examinations, AEs have been seen in 0.2% to 2.9% of nonbleeding persons and 2.6% to 6.2% of patients evaluated specifically for occult blood in the stool, anemia, or hemorrhage. In a patient being studied for GI bleeding, in whom the site of active bleeding is unproven, the only basis for determining that an identified AE or diverticulum is responsible is the indirect evidence provided by the patient’s course after ablation or resection of the suspected lesion. It is unusual for incidentally found AEs to bleed, and an AE, even in a patient with a history of bleeding, cannot be assumed to be the cause.

Bleeding from AEs typically is recurrent and low grade, although patients can present with massive hemorrhage. The nature and degree of bleeding frequently vary in the same patient with different episodes: Patients may have bright red blood, maroon stools, or melena on separate occasions. This spectrum reflects the varied rates of bleeding from the ectatic capillaries, venules, and AV communications, which depends on the developmental stage of the lesions. In one study, bleeding from AEs was characterized by tarry stools in 20% to 25% of cases, whereas the minority (10% to 15%) of patients exhibited solely iron deficiency anemia, with stools that were intermittently positive for occult blood. Another study reported that AEs resulted in hemodynamically significant LGIB in 21% of cases, although 42% exhibited chronic LGIB or anemia without evidence of acute hemorrhage. Today, AEs are thought to be asymptomatic or to result in occult obscure GI bleeding in most patients. Bleeding from AEs stop spontaneously in more than 90% of cases.

In 1958, Heyde described what is still a controversial association: AE, GI bleeding, and aortic stenosis (AS); aortic valve replacement (AVR) had even been recommended for “Heyde syndrome” when bleeding could not be managed by medical means. Numerous reports of Heyde syndrome appear in the literature, although some analyses and studies have failed to support the association. The existence of Heyde syndrome has been suggested again in a retrospective study in which the frequency of AS was 31.7% in patients with “AVMs” compared with 14% in the general population. It has been postulated that deficiencies of the largest forms of vWf multimers (von Willebrand disease, type 2A) result in hemostatic abnormalities that may predispose preexisting AEs to bleed. It is now believed that increased shear stress results in unfolding of the globular von Willebrand polymer into an elongated highly asymmetric protein, which exposes the A2 domain. ADAMTS13 then binds to the A2 domain, which results in cleavage of this high molecular weight multimer into smaller polymers, which are less hemostatic than their parent molecules. Preoperative deficiency of these multimers reverses after AVR, with resolution of bleeding in most patients with Heyde syndrome who underwent AVR; recommendation to replace the aortic valve to control GI bleeding from AEs is controversial. Currently AVR is only recommended for patients with severe AS and not for those with GI bleeding or iron deficiency anemia in the setting of asymptomatic AS because most bleeding from AEs can be controlled by any one or more of a variety of endoscopic techniques (see Chapter 20 ).

GI bleeding is a major occurrence in those with a left ventricular assist device (LVAD), and now recognized to result most commonly from UGI tract angiodysplasia (discussed later). A recent meta-analysis of 17 case-control and cohort studies showed a pooled prevalence of GI bleeding in continuous-flow LVAD patients to be 23% with potential risk factors of older age and elevated creatinine. The mechanism of LVAD-associated GI bleeding is still not well understood but has been attributed to impaired vWf-dependent primary hemostasis. Such impairment may be a result of decreased specific activity of vWf, shear stress that results in release of preformed vWf from endothelial cells, and decreased high molecular weight multimers in nonpulsatile flow regimens leading to acquired von Willebrand syndrome. It has been noted that wide pulse pressures are associated with increased von Willebrand multimers, but that patients with LVADs and AS have narrow or zero pulse pressures. Studies are underway to determine whether decreasing the speed of these devices and hence inducing more “pulsatile” flow will result in a reduction in GI bleeding, although preliminary results do not show improvement in acquired von Willebrand syndrome with these maneuvers. Patel et al. published a novel approach using nasal endoscopy to determine the risk of GI bleeding for patients with LVADs. The presence of nasal hypervascularity as a potential surrogate for GI vascular lesions helped determine the risk of GI bleeding in patients who were to undergo LVAD. They found that nasal hypervascularity was equally common in patients with heart failure regardless of whether or not they received an LVAD (63% in the LVAD group and 57% in the heart failure group versus 20% in the control group), but there was a statistically significant association between GI bleeding in patients with LVAD and nasal hypervascularity with an incidence of 32%. There was no statistically significant association between GI bleeding and heart failure patients with hypervascularity but without an LVAD. Thus, nasal endoscopy may be a potential surrogate marker for GI mucosal vascular alterations. Further studies are being done to compare nasal endoscopy findings in LVAD patients with and without small bowel vascular lesions diagnosed by VCE to confirm this hypothesis.

Diagnosis and Management

Management of colonic AEs begins with suspecting the lesion in an older person who has acute or chronic LGIB (see Chapter 20 ). Colonoscopy is the primary means of both diagnosis and treatment. If the suspected lesion cannot be found, or if bleeding is massive and colonoscopy cannot be performed, radionuclide scintigraphy followed by CTA should be performed. One retrospective study compared CTA to ⁹⁹ mTc-labeled red blood cell scintigraphy (RBCS) for the overall evaluation and management of acute LGIB and found that both CTA and RBCS could be used to identify active bleeding (38% of cases), but the site of bleeding was localized with CTA in a significantly higher proportion of studies. Of 24 patients in whom the site of LGIB was accurately localized by CTA, 2 patients were diagnosed with “AVMs.” RBCS did not establish causation of bleeding in any patient.

The endoscopist’s ability to diagnose the specific nature of a vascular lesion is limited by the similar appearance of different types of lesions. AEs, spider angiomas, telangiectasias, angiomas, the focal hypervascularity of radiation colitis, UC, Crohn disease, ischemic colitis, certain infections, hyperplastic and adenomatous polyps, and malignancies, including lymphoma and leukemic infiltrations, can all, on occasion, resemble each other ( Box 38.1 ). Because traumatic and endoscopic suction artifacts may resemble vascular lesions, all lesions must be evaluated on insertion of the colonoscope, rather than during withdrawal. Pinch biopsy samples obtained from small, nonelevated vascular lesions during endoscopy usually are nonspecific; therefore, the risk of performing biopsies of these abnormalities is not justified. Sometimes the prominent feeding vessel of an AE might be appreciated at the time of endoscopy, and the mucosa at the AE margin may be more pale than distant mucosa, although such a “pale halo” also may be seen in other vascular lesions.

Because the appearance of vascular lesions is influenced by a patient’s blood pressure and blood volume, such lesions may not be evident in those with significant anemia or hypotension; thus, accurate evaluation may not be possible until red cell and volume deficits are corrected. Meperidine also may diminish the prominence of finer vascular abnormalities (e.g., AEs, the telangiectasias of HHT); use of meperidine, therefore, should be avoided and, if used, its effects reversed by naloxone so that vascular lesions can be detected accurately; such a masking effect does not occur with fentanyl. In patients who have received meperidine, naloxone has been shown to enhance the appearance of normal colonic vasculature in approximately 10% of patients and to cause existing AEs to appear (2.7%) or increase in size (5.4%) ( Fig. 38.8 ). For these reasons, naloxone is an important adjunctive medication for patients undergoing endoscopic evaluation for lower intestinal bleeding and who have received meperidine. Cool water lavage, to cleanse the mucosal surface during colonoscopy, also may cause underlying AEs to vasoconstrict and disappear transiently.

Angiography is used to determine the site and nature of vascular lesions during active bleeding and can identify some vascular lesions even after bleeding has ceased. The 3 reliable angiographic signs of AEs are a densely opacified, slowly emptying, dilated, tortuous vein; a vascular tuft; and an early-filling vein (see Fig. 38.7 ). A fourth sign, extravasation of contrast material, identifies the site of bleeding when the rate of bleeding is at least 0.5 mL/min but is not specific for AE. The slowly emptying vein (see Fig. 38.7 A ) persists late into the venous phase, after the other mesenteric veins have emptied. Vascular tufts (see Fig. 38.7 B ) are created by the ectatic venules that join the mucosal component of the AE and its submucosal vein. They are seen best in the arterial phase; are usually located at the termination of a branch of the ileocolic artery; appear as small candelabra-like or oval clusters of vessels; and still are seen in the venous phase communicating with a dilated, tortuous, intramural vein. The early-filling vein is seen in the arterial phase within 4 or 5 seconds of injection (see Fig. 38.7 B ) but is not a valid sign of AE if vasodilators such as papaverine or tolazoline (Priscoline) have been used to enhance the study. When the lesion is bleeding, intraluminal extravasation of contrast material usually appears during the arterial phase and persists throughout the study. Extravasation identifies a site of active bleeding, but in the absence of other signs of AEs, it suggests another cause for the bleeding.

Management of nonbleeding AEs incidentally found at colonoscopy is expectant. In such cases, endoscopic therapy is not indicated because the risk of bleeding in asymptomatic patients with AEs has been shown in a prospective study to be low (0% in 3 years), which clearly does not warrant the potential risks of bleeding and perforation with colonoscopic ablation.

Bleeding from AEs can be controlled endoscopically or angiographically in most patients, thereby avoiding the morbidity and mortality of emergency operation. Today, super-selective microcoil embolization has replaced intra-arterial vasopressin infusion for the treatment of LGIB. Such embolization is highly effective and safe although complications occur in 5% to 9% of cases; serious complications (e.g., gangrene, hematoma formation, arterial dissection, thrombosis, pseudoaneurysm formation) are reported in less than 2% of cases. Vasopressin still is recommended, however, when intestinal vascular lesions are diffuse throughout the bowel or when super-selective catheterization is not possible.

Hormonal therapy, using estrogens in combination with progestins, had been used to treat patients with a variety of bleeding vascular lesions of the GI tract. The mechanisms by which such agents work are not known, although procoagulant effects and enhanced endothelial integrity are popular theories. A long-term observational study showed that combination hormonal therapy stopped bleeding in patients with occult GI bleeding of obscure origin (likely to have resulted from angiodysplasias in the small intestine), although a recent meta-analysis detailed 2 case-control studies in which hormonal therapy was ineffective for bleeding cessation. It is conceivable that vascular lesions in the small intestine may respond differently to such treatment than similar appearing lesions in the colon; no study of hormonal therapy has been done for proven colonic AEs.

Somatostatin analogs are another option for the treatment of bleeding from GI vascular lesions. These agents work by inhibiting angiogenesis, decreasing splanchnic blood flow, increasing vascular resistance, and improving platelet aggregation. In the recent meta-analysis mentioned previously, 4 cohort studies were found assessing the efficacy of either daily or monthly octreotide. The pooled odds ratio for bleeding cessation was 14.5 (95% confidence interval [CI]: 5.9 to 36), and there was a decrease in transfusion requirements seen after 1 year of therapy with an odds ratio of 0.55 (95% CI: 0.29 to 0.82).

A novel therapy for AEs, and perhaps other vascular lesions in the GI tract, is the use of anti-angiogenic factors, including thalidomide, bevacizumab, and lenalidomide. Thalidomide was developed in the 1950s as a sedative, sleeping pill, and antiemetic for pregnant women, but it soon became notorious for causing phocomelia and other malformations in the newborn. In 1994, D’Amato and colleagues reported that thalidomide inhibited VEGF and basic fibroblast growth factor-mediated angiogenesis. Data suggest the mechanism for its antiangiogenic effect is related to reduced expression of integrin genes with resultant decreased cell-cell surface interactions and response to angiogenic cytokines. Several case reports and case series have described the successful use of thalidomide to treat life-threatening or refractory bleeding from intestinal AEs and Crohn disease with refractory bleeding. After treatment with thalidomide for 3 months, substantial reductions in the number, size, and color intensity of AEs were documented by VCE. A controlled trial of patients with GI angiodysplasias and GAVE (see later) randomized patients to thalidomide or iron supplementation and reported a 50% or greater decrease in bleeding episodes at the 1-year follow-up in 71.4% patients in the thalidomide group compared with 3.7% in the iron supplementation group; these patients had lesions that were predominantly confined to the stomach and small intestine and there still is no experience with these treatments for colonic AEs.

Bevacizumab (Avastatin) is a humanized monoclonal antibody against VEGF that is effective against colon and renal cancers and also has strong antiangiogenic activity. Curiously, dose-dependent nasal and GI bleeding is observed in up to 59% of patients during treatment, possibly caused by a loss of vascular integrity as a result of bevacizumab-induced endothelial-cell shedding in highly regenerative mucosal tissues with active angiogenesis. It is unclear why some antiangiogenic substances like bevacizumab cause mucosal bleeding and others like thalidomide do not; this disparity effect may be related to the phase of angiogenesis that is antagonized or might reflect a particularly strong antiangiogenic activity.

Lenalidomide is a newer angiogenesis inhibitor that is an analog of thalidomide but with fewer adverse effects. Lenolidamide could present a new therapeutic asset for AEs, although its role remains to be evaluated in controlled studies. VEGF-based antiangiogenic therapy is a promising therapy, but a more detailed understanding of the angiogenic cascade and how antiangiogenic substances act within it will be needed to resolve this issue.

In the past, neodymium:yttrium-aluminum-garnet laser, endoscopic sclerosis, monopolar and bipolar electrocoagulation, and heater probe had been used to ablate a variety of vascular lesions throughout the GI tract and to control active bleeding. More recently, however, hemoclips in combination with cautery, endoscopic band ligation, and argon plasma coagulation (APC) have been used for this purpose ( Fig. 38.9 ). For AEs, heater probe, and APC are most commonly used. Control of bleeding has been obtained with a variety of endoscopic thermal means in 47% to 88% of cases, and no technique has been established as superior. Severe delayed bleeding occurs in 5% of patients with colonic AEs after thermal therapy. Recurrent bleeding from colonic AEs appears to be reduced after endoscopic therapies, but more than 1 treatment session may be necessary. Rebleeding can be expected to increase with time and has been seen in 28% to 52% of patients over a follow-up period ranging from 15 to 36 months.

In preparation for endoscopic ablation of vascular lesions, aspirin and aspirin-containing drugs, other NSAIDs, anticoagulants, and antiplatelet agents should be withdrawn, if possible, at least several days before the procedure and depending on the agent. Aspiration of some intraluminal gas just before thermal therapy is applied adds a measure of safety as the colon wall is not so thinned with a smaller-diameter lumen. Right hemicolectomy is indicated when AE has been identified by either colonoscopy or angiography and when therapy by either or both of these modalities is unsuccessful, cannot be performed, or is unavailable and the patient has continuous or recurrent LGI bleeding. The presence or absence of diverticulosis in the left colon does not alter the extent of colonic resection in this circumstance; only the right half of the colon is removed, but it is important that the entire right half of the colon be removed to ensure that no AEs are left behind. If the site of bleeding (and its cause) is not identified, and bleeding recurs or is continuous, a subtotal colectomy (STC) is appropriate. Morbidity and mortality rates of an STC are not statistically different from those accompanying a “blind” hemicolectomy, i.e., when the bleeding site is not identified. In one surgical review, mortality and rebleeding rates for STC, directed limited colectomy, and blind limited colectomy were 0% to 40% and 0% to 8%; 2% to 22% and 0% to 15%; and 20% to 57% and 35% to 75%, respectively.