Gastric antral vascular ectasia (GAVE), first described in 1953, accounts for approximately 5% of nonvariceal upper gastrointestinal bleeding. It can present with frank bleeding or with chronic iron deficiency anemia. In 1984, Jabbari et al. coined the term “watermelon stomach” for the characteristic pattern noted on endoscopy. It was only in 1995 that the distinction was made between portal hypertensive gastropathy (PHG) and GAVE. Despite various attempts at medical, endoscopic, and even surgical treatments over the years, the lack of a complete understanding of the pathophysiology of these diseases has hindered therapy.
It is important to distinguish GAVE from PHG (Table 1). GAVE can be seen in patients with and without cirrhosis or portal hypertension. Patients tend to be predominantly (∼71%) older women with a mean age of 75 years.[4, 5] Several diseases have been associated with GAVE, including connective tissue disorders, bone marrow transplantation, and chronic renal failure. However, there is no clear understanding of the exact link between these disease entities. Endoscopically, GAVE has a characteristic pattern of linear columns of erythematous and at times raised mucosa with underlying tortuous ectatic vessels along the longitudinal folds in the antrum. However, there are various other patterns of GAVE described in the literature (Fig. 1), such as speckled or diffuse patchy erythema, honeycombing, and nodular antral gastropathy, sometimes without a clear distinction from PHG.[5, 6] Histologically, there is an abundance of dilated capillaries in both PHG and GAVE. However, fibrin microthrombi, fibromuscular hyperplasia in the lamina propria (resulting from spindle cell proliferation extending toward antral glands), and increased neuroendocrine cells in the lower lamina propria distinguish GAVE. These findings have a diagnostic accuracy of 85% for GAVE and help distinguish it from PHG[3, 5] (Fig. 2).
|Patient characteristics||No clear association with cirrhosis or portal hypertension; loose association with other autoimmune diseases||Predominantly seen in portal hypertension with/without cirrhosis|
|Endoscopic findingsa||Erythema limited to the antrum||Tends to localize to fundus and corpus|
|Histological findings||Fibrin microthrombi; fibromuscular hyperplasia, increased neuroendocrine cells||Ectatic mucosal capillaries only|
|Pharmacotherapy||Estrogen/progesterone; octreotide; tranexamic acid; thalidomide; serotonin antagonist; corticosteroids||Nonselective beta-blockers; somatostatin/octreotide|
|Invasive therapy||Thermal ablation (argon plasma coagulation/ bipolar electrocoagulation); endoscopic band ligation||TIPS; liver transplantationb|
PHG occurs in approximately 65% of patients with cirrhosis but can also occur in patients with noncirrhotic portal hypertension.[5, 7] PHG may even be a marker of more severe liver disease and possibly a predictor of variceal bleeding.[5, 8] It has been suggested that variceal sclerotherapy exacerbates PHG. An association with variceal banding, however, is more controversial. Endoscopic findings in PHG include patchy erythema with a characteristic “mosaic” pattern (Fig. 3), localizing mostly in the gastric fundus and corpus. Histological findings in PHG are limited to ectatic mucosal capillaries without fibrin microthrombi, spindle cell proliferation, or fibromuscular hyperplasia.
GAVE is an acquired microvascular ectasia thought to result from mechanical stress and/or neuroendocrine abnormalities with hypersecretion of vasoactive substances. Studies regarding pharmacological therapy are mostly in the form of case reports or series. Estrogen and progesterone were reported to decrease bleeding from GAVE but did not eradicate it. Tranexamic acid, an antifibrinolytic agent, has been reported to stop actively bleeding GAVE but carries a risk of thrombotic disease and ischemia. Octreotide has mixed case reports showing efficacy and remains of unclear clinical benefit.[11, 12] Other medications, including thalidomide, serotonin antagonists, and corticosteroids, have a few case reports describing their use with uncertain efficacy.[5, 13] Interestingly, there is a paucity of data in the use of acid suppressive therapy with histamine H2-receptor antagonists or proton pump inhibitors, yet patients are frequently prescribed these agents. In our practice, we also peruse the patient's medications for vasodilating agents, and we consider discontinuing proton pump inhibitors, which theoretically increase antral blood flow. A study of this effect is currently in progress.
Endoscopic thermal ablation therapy has become widely accepted to treat GAVE where it appears to significantly decrease transfusion requirements. Naga et al. demonstrated a decrease of transfusion requirements in 89% of their 29 patients with argon plasma coagulation. Bipolar cauterization is a simpler and less expensive method of endoscopic thermal ablation but requires care due to the potential depth of heat delivery. Endoscopic band ligation is a treatment modality that is less frequently used but is described in the literature. Small studies have shown a reduction of transfusion requirements as well as fewer sessions required for treatment compared with thermal ablation; however, this is significantly more costly and technically more difficult. Other modalities that have fallen out of favor include the Nd:YAG laser, the heater probe, and cryotherapy. At the time of its initial description, GAVE was managed with antrectomy or gastrectomy; however, these methods are used rarely due to the associated morbidity and mortality.
The mechanical effects of portal hypertension in PHG underlie increased gastric blood flow in the submucosal, muscle, and serosal layers, and decreased mucosal blood flow due to congestion and stasis. Nonselective beta-blockers appear to be effective by decreasing portal hypertension; propranolol reduced recurrent bleeding in PHG in a randomized control trial. Somatostatin and its analogue, octreotide, also have been shown to reduce gastric perfusion; several studies have demonstrated that they can temporarily stop active bleeding from PHG. Other agents, such as vasopressin, estrogen, and progesterone, may decrease gastric mucosal blood flow but lack supportive clinical data.
In the past, bleeding PHG was managed with portal decompression with surgical shunts and even gastrectomies for chronic bleeding that was difficult to control. However, these options are becoming obsolete and have now been largely replaced by transjugular intrahepatic portosystemic shunt (TIPS) and liver transplantation when appropriate clinically. Underlining the differences between GAVE and PHG, it was demonstrated in a comparative study that 75% of patients with severe PHG had improvement in endoscopic findings and a decrease in transfusion requirements after TIPS, whereas none of the patients with GAVE demonstrated improvement. Liver transplantation appears to be the definitive therapy for PHG where recovery of liver function reverses portal hypertension.
In conclusion, PHG and GAVE are distinct entities with some overlapping properties. It is critical that each patient with chronic gastrointestinal bleeding and suspected lesions be categorized in the appropriate group, because the management is significantly different. When endoscopic evidence of GAVE or PGH is encountered, it is important to take into consideration the endoscopic, histological, and clinical history and specifically to look for evidence of cirrhosis or other signs of portal hypertension. In our experience, GAVE is often associated with underlying fatty liver disease with or without portal hypertension, but whether or not this is related (perhaps another sequela of hyperinsulinemia) or coincidental remains uncertain. PHG, on the other hand, is more often seen in advanced liver disease with obvious complications of portal hypertension. Future studies are needed to help in the understanding of the pathophysiology of these disease entities so that more tailored therapies can be established.