Angiodysplasia is an acquired submucosal arteriovenous malformation which may cause lower gastrointestinal bleeding in elderly patients. It is often difficult to diagnose, and many aspects of the disease are unclear; reported statistics about incidence and results of treatment therefore differ widely. Angiodysplasia has been reported to account for between 2 and 60 per cent of adult patients with lower gastrointestinal bleeding; it probably causes the majority of incidents of massive lower gastrointestinal bleeding in older patients, with diverticulosis accounting for most of the remainder. There is also little agreement about the prevalence of angiodysplasia in adults; estimates range from 1 to 30 per cent, but the true figure is probably 1 to 5 per cent.

Typical angiodysplastic lesions are 0.5 to 1.0 cm in diameter, bright red, flat or slightly raised, and covered by very thin epithelium. Between 70 and 90 per cent of the lesions appear in the right colon, although they may also be found in the distal colon, and can also occur, less often, anywhere else in the gastrointestinal tract. Most patients have two or three lesions. Angiodysplasia probably accounted for many episodes of what were in the past thought to be diverticular bleeds. Thus, there is only partial truth to the maxim that diverticulitis is predominantly a left-sided disease, and bleeding diverticulosis a right-sided disease.

Theories of aetiology abound, but it is probably an acquired degenerative lesion resulting from chronic, intermittent, partial low grade obstruction to submucosal venous outflow within the muscular bowel wall, due to contraction and distension of the colon. This in turn causes dilatation first of submucosal veins, and then capillaries and precapillary sphincters, leading to a small arteriovenous malformation. Since the right colon has the widest diameter in the large bowel, the law of LaPlace predicts that wall tension is greatest there. Thus, the intramural venous obstruction would also be maximal in that location, which could account for the right-sided predilection of angiodysplasia. Alternatively, but less convincingly, the lesions might appear because of low mucosal blood flow leading to chronic submucosal shunting. The relatively low mucosal perfusion could be accounted for by high right colon wall tension with or without chronic constipation, as well as by cardiac valvular disease. The least likely explanation is that it may simply arise because the thin wall of the right colon provides minimal support for the microscopic vasculature, leading to foci of dilatation which degenerate into arteriovenous malformations. It has been suggested that an abnormal pulse wave pattern in the ileocaecal artery due to cardiac valve disease contributes to this degeneration. Numerous accounts have documented an association with ‘cardiovascular disease’ (50–100 per cent), and especially aortic valve disease (20–30 per cent). Not all studies have revealed a correlation with valve lesions, however, and assertions of cause and effect are conjectural. There are major methodological deficiencies in all reports of correlation between angiodysplasia and heart valve disease.

Inherited gastrointestinal arteriovenous malformations such as hereditary haemorrhagic telangiectasia or Osler–Weber–Rendu syndrome are similar but unrelated. Unlike angiodysplasia, hereditary haemorrhagic telangiectasia is familial, tends to affect younger patients, and produces many more lesions which are much more widely dispersed throughout the gastrointestinal tract as well as in the skin and mucous membranes. Unlike haemangiomas, angiodysplasia is not neoplastic.

Gastrointestinal bleeding is the only clinical manifestation of angiodysplasia, and patients are usually over 60 years old. The most common result of bleeding lesions is gross haemochezia or maroon stools, and sometimes melaena, although some reports claim that 30 to 50 per cent of patients show only anaemia and guaiac-positive faeces. Less than 10 per cent is a more likely proportion. The natural history of angiodysplasia is uncertain. Because the bleeding occurs from tiny capillaries and venules covered by very thin epithelium, episodes tend to be self limiting, but chronic and recurrent. Massive haemorrhage with hypotension is unusual. In contrast, diverticular bleeding, which comes from one or two larger eroded arterioles, is more likely to be acute and massive. After clot has formed around the responsible diverticular vessel, recurrent bleeding is less likely than with angiodysplasia. Patients with angiodysplasia may have a long history of recurrent idiopathic gastrointestinal bleeding, with multiple unrevealing prior investigations, possibly including laparotomy. The arteriovenous malformations are too small to cause haemodynamic changes (in the absence of bleeding), pain, or perforation.

Angiodysplasia has largely been recognized and characterized only since the advent of colonoscopy and mesenteric angiography. Because the lesions are small, flat, and not transmural, they are overlooked by barium enema radiography or inspection of the serosal surface at laparotomy. Even pathological examination will usually miss angiodysplasia unless special injection techniques are used. The mesenteric vessels of the specimen may be injected with barium followed by radiography, or the vessels may be perfused with silicone rubber followed by tissue clearing with methyl salicylate (Fig. 2) 1115, so that only the vascular network remains opaque. Specialized pathological examinations reveal tortuous, enlarged, thin walled submucosal veins communicating with clusters of dilated mucosal capillaries (Fig. 3) 1116. Endoscopic biopsies are rarely helpful, however, because the mucosal lesion is variable, and the biopsy forceps do not usually reach as deep as the abnormal submucosal veins.

The appearances at angiography reflect the theory of pathogenesis. In 90 per cent of cases a large, tortuous, slowly draining submucosal vein is found; a mucosal vascular tuft is seen in 70 per cent; 60 per cent of cases show an early draining dilated vein, which correlates with the final stage of formation of the arteriovenous malformation. Small angiodysplasias can be quite inconspicuous on angiography, unless active haemorrhage is taking place and extravasation of contrast is demonstrated. Unfortunately, because bleeding from angiodysplasia is typically episodic, extravasation is not often demonstrable at angiography.

The colonoscopic appearance is pathognomonic. The lesions are flat or slightly raised, bright red, and 0.5 to 1.0 cm in diameter. Because of the magnification afforded by the colonoscope, the tiny dilated mucosal vessels in and around the lesion can frequently be seen on close inspection, although these small malformations can be overlooked on endoscopy unless the colonic mucosa is quite clean. Polyethylene glycol gut lavage solutions clean out the colon more effectively than older laxative and enema preparations, and make discovery of angiodysplasia more likely. Care must be taken not to confuse angiodysplasia with traumatic ecchymoses (which are seen on withdrawal but not insertion of the colonoscope), and prominent veins (which are bluish, not bright red). Angioplastic lesions may be almost hidden behind a mucosal fold in the right colon, or just proximal to the ileocaecal valve. The diagnosis can be made 80 to 90 per cent of the time using either colonoscopy or angiography. Endoscopy is probably the more sensitive and specific technique, but opinions differ, depending upon local expertise, and upon which method is accepted as the standard by which to judge the other. Both angiography and colonoscopy may produce false positive results. Colonoscopy is less invasive than angiography, and also has the advantage of providing biopsies if desired. Furthermore, the entire colon and sometimes the terminal ileum can be meticulously assessed to exclude other diseases or detect synchronous angiodysplasia. A growing array of endoscopic therapeutic manoeuvres are also possible, including fulguration and polypectomy.

Treatment is indicated for an angiodysplasia that has bled, because of its tendency to cause chronic recurrent haemorrhage. The choice is between surgical resection and endoscopic coagulation. Both treatments are fairly uncomplicated; the most difficult challenge is to make an accurate diagnosis, using the methods discussed above, in order to direct therapy appropriately. Other possible sources of bleeding must be ruled out, and care must be taken not to overlook synchronous angiodysplasia spots. Since angiodysplasia is usually found in the right colon, right hemicolectomy is often curative (or removal of whatever bowel segment is responsible). Rebleeding occurs in 15 to 25 per cent of surgically treated patients, usually because of overlooked arteriovenous malformations in other areas of the bowel, or an error in diagnosis. If preoperative evaluation has reasonably convincingly demonstrated right-sided angiodysplasia, and left-sided diverticulosis is found at laparotomy, it is still acceptable to limit the resection to the right colon. If no specific treatment is undertaken at least half of the patients with angiodysplasia will rebleed.

Despite satisfactory results with surgery, the modern treatment of angiodysplasia is endoscopic coagulation, unless the lesions are large or quite numerous. Either laser photocoagulation (Fig. 4) 1117 or electrocoagulation may be used. Photocoagulation is preferable: the use of ‘hot biopsy’ forceps allows the target to be grasped, the mucosa drawn up off the underlying bowel wall, coagulating current applied, and the lesion released (Fig. 5) 1118. The technique minimizes the risk of perforation, which is an important consideration in the thin walled caecum (in contrast to coagulation of bleeding ulcers in a duodenum or stomach thickened by chronic inflammation). No actual biopsy of the arteriovenous malformation is required, since the diagnosis may be made by inspection alone. If a laser is to be used, argon is preferable to neodymium-yttrium aluminium garnet (Nd:YAG). The light wavelength from an argon unit is preferentially absorbed by red pigment and also penetrates more superficially than does light from a Nd:YAG machine. Both characteristics are nicely suited to treatment of superficial blood vessel malformations. Whether using laser photocoagulation or hot biopsy electrocoagulation, it is best to start at the periphery of the lesion and progress toward the centre. The mucosa should be cauterized until it is white, not black. Coagulation of large angiodysplasias should be carried out in stages, several weeks apart, to minimize the risk of perforation. The colon and terminal ileum should be searched carefully for synchronous lesions. Rebleeding occurs in only 10 to 30 per cent of patients, and this is usually from lesions overlooked or incompletely coagulated, making early follow-up colonoscopy is prudent. Nevertheless, even in patients who rebleed after endoscopic treatment of arteriovenous malformation of the gastrointestinal tract the frequency of haemorrhage episodes and number of transfusions declines. Perforation follows in up to 7 per cent of treatments, usually when a Nd:YAG laser is used. The best treatment for an asymptomatic angiodysplasia lesion that has been found incidentally during colonoscopy is unclear. The risk of later haemorrhage is unknown but is probably low; it is therefore reasonable to leave quiescent lesions alone. Experienced endoscopists, however, may justifiably prefer to eliminate the uncertainty.

Baum S, et al. Angiodysplasia of the right colon: a cause of gastrointestinal bleeding. Am J Roentgenol 1977; 129: 789–94.

Boley SJ, et al. On the nature and etiology of vascular ectasias of the colon. Gastroenterology 1977; 72: 650–60.

Boley SJ, et al. Vascular ectasias of the colon. Surg Gynecol Obstet 1979; 149: 353–9.

Danesh BJZ, et al. Angiodysplasia, an uncommon cause of colonic bleeding. Int J Colorect Dis 1987; 2: 218–22.

Emanuel RB, et al. Arteriovenous malformations as a cause of gastrointestinal bleeding. J Clin Gastroenterol 1985; 7: 237–46.

Gostout CJ, et al. Mucosal vascular malformations of the gastrointestinal tract: clinical observations and results of endoscopic Nd-YAG laser therapy. Mayo Clin Proc 1988; 63: 993–1003.

Greenstein RJ, et al. Colonic vascular ectasias and aortic stenosis: coincidence or casual relationship? Am J Surg 1986; 151: 347–51.

Imperiale TF, Ransohoff DF. Aortic stenosis, idiopathic gastrointestinal bleeding and angiodysplasia: is there an association? Gastroenterology 1988; 95: 1670–6.

Price AB. Angiodysplasia of the colon (review). Int J Colorect Dis 1986; 1: 121–8.

Richter JM, et al. Angiodysplasia: natural history and efficacy of therapeutic interventions. Dig Dis Sci 1989; 34: 1542–6.

Richter JM, et al. Angiodysplasia: clinical presentation and colonoscopic diagnosis. Dig Dis Sci 1984; 29: 481–5.

Talman EA, Dixon DS, Gutierrez FE. Role of arteriography in rectal hemorrhage due to arteriovenous malformations and diverticulosis. Ann Surg 1979; 190: 203–13.

Tedesco FJ, Griffin JW, Khan AQ. Vascular ectasia of the colon: clinical, colonoscopic, and radiographic features. J Clin Gastroenterol 1980; 2: 233–8.

Trudel JL, Fazio VW, Sivak MR. Colonoscopic diagnosis and treatment of arteriovenous malformations in chronic liver gastrointestinal bleeding. Dis Colon Rect 1988; 31: 107–10.