Abstract
Gastric antral vascular ectasia (GAVE) is one of the uncommon causes of upper gastrointestinal bleeding. Major treatment of GAVE includes pharmacotherapy, endoscopy, and surgery. The efficacy and safety of pharmacotherapy have not been sufficiently confirmed; and surgery is just considered when conservative treatment is ineffective. By comparison, endoscopy is a common treatment option for GAVE. This paper reviews the currently used endoscopic approaches for GAVE, mainly including argon plasma coagulation (APC), radiofrequency ablation (RFA), and endoscopic band ligation (EBL). It also summarizes their efficacy and procedure-related adverse events. The endoscopic success rate of APC is 40–100%; however, APC needs several treatment sessions, with a high recurrence rate of 10–78.9%. The endoscopic success rates of RFA and EBL are 90–100% and 77.8–100%, respectively; and their recurrence rates are 21.4–33.3% and 8.3–48.1%, respectively. Hyperplastic gastric polyps and sepsis are major adverse events of APC and RFA; and Mallory–Weiss syndrome is occasionally observed after APC. Adverse events of EBL are rare and mild, such as nausea, vomiting, esophageal or abdominal pain, and hyperplastic polyps. APC is often considered as the first-line choice of endoscopic treatment for GAVE. RFA and EBL have been increasingly used as alternatives in patients with refractory GAVE. A high recurrence of GAVE after endoscopic treatment should be fully recognized and cautiously managed by follow-up endoscopy. In future, a head-to-head comparison of different endoscopic approaches for GAVE is warranted.
Keywords
Introduction
Gastric antral vascular ectasia (GAVE), which was first described by Rider et al. 1 in 1953, refers to dilated blood vessels located in the antrum and radiated to the pylorus. It was also named ‘watermelon stomach’ by Jabbari et al. 2 in 1984, as its typical appearance is similar to the stripes seen on watermelons. The most common lesion site of GAVE is the antrum, but it is also observed in other parts of the gastrointestinal tract, including the cardia, duodenum, and jejunum. 3 According to the endoscopic characteristics, GAVE can be divided into two types: striped and diffuse (Figure 1). Histology shows hyaline thrombosis, fibrin hyalinization, and spindle hyperplasia within dilated submucosal capillaries of the gastric antrum. 4 The pathophysiology of GAVE may be attributed to an increased prostaglandin E2 level, 5 a cross-linking of protein antibody in gastric mucosa, 6 an injury of gastric mucosa caused by dyskinesia, 7 or the effect of vasoactive substances, such as gastrin or serotonin. 8

Endoscopic images of striped (a) and diffuse (b) GAVE.
GAVE is often associated with chronic liver disease and autoimmune and connective tissue diseases, such as liver cirrhosis, hepatocellular carcinoma, hypothyroidism, sclerosis, and systemic lupus erythematosus. Most of the patients are elderly, while the ratio of female to male is reported to be 2:1. 9
GAVE accounts for about 4% of non-variceal upper gastrointestinal bleeding. 10 It can be clinically asymptomatic or manifested as occult and dominant upper gastrointestinal bleeding with hematemesis or melena. Many patients present with refractory anemia in spite of blood transfusion and intravenous or oral iron supplementation.
There are several treatment options for GAVE. The efficacy of pharmacologic therapy, such as octreotide, 11 tranexamic acid, 12 bevacizumab, 13 estrogen-progesterone, 14 and thalidomide, 15 has not yet been sufficiently confirmed. Endoscopy has gradually become the first-line treatment option for GAVE in recent years, including cryotherapy, Nd-YAG laser, argon plasma coagulation (APC), radiofrequency ablation (RFA), and endoscopic band ligation (EBL). Surgery can be considered if conservative treatment fails. Antrectomy can effectively control chronic bleeding, 16 but most patients, especially those with comorbid diseases, have a high risk of procedural complications.
To the best of our knowledge, there is lack of consensus statement or guideline recommendation regarding the management of GAVE. This review aims to summarize common endoscopic procedures for the treatment of GAVE and their efficacy and complications. Literature regarding APC, RFA, and EBL for the treatment of GAVE were searched via the PubMed and EMBASE databases from the earliest available publications until 5 June 2021. Search items were [(gastric antral vascular ectasia) OR (watermelon stomach)] AND [(argon plasma coagulation) OR (APC)] OR [(radiofrequency ablation) OR (RFA)] OR [(endoscopic band ligation) OR (EBL)]. Initially, a total of 520 papers were identified. Data on APC, RFA, or EBL alone were collected. The major outcomes of interests included endoscopic success rate, blood transfusion requirement, recurrence rate, and adverse events after endoscopic treatment for GAVE.
Argon plasma coagulation
APC is the most commonly used endoscopic approach for GAVE in recent years. It is a non-contact electrocoagulation device that uses argon plasma to transduce electrical energy to local tissues. The depth of the treated lesions is approximately 1–3 mm, which is sufficient to coagulate the superficial vessels. 17 The electric power output of 40–60 W and the argon flow rate of 0.8–2 l/min are effective to eliminate the lesions 18 (Supplemental Table 1).
Endoscopic success of APC
Endoscopic success, which is often defined as complete hemostasis or elimination of the majority of visible lesions under endoscopy, can reach 40–100% by APC19–28 [Figure 2(a)]. In a study, 21 patients were included, of whom 19 (90.5%) achieved endoscopic success after multiple treatment sessions, but 2 had persistent bleeding and required blood transfusion. It should be noted that 19 patients who achieved endoscopic success presented with iron deficiency anemia alone, but 2 patients who failed were more severe and manifested as acute gastrointestinal bleeding. 21 In another study, the endoscopic success rate was 83.3% (25/30). Only 1 patient received anticoagulant drugs before GAVE was successfully treated by APC, but no patient received antiplatelet drugs. The reasons for endoscopic failure might be that all of the 5 patients had striped lesions, and 3 of them took antiplatelet drugs. 20 By comparison, Garg et al. 27 reported a low success rate of 40% for endoscopic resolution of GAVE. Interestingly, all endoscopic success was just achieved in cirrhotic patients. None of non-cirrhotic patients had endoscopic resolution of GAVE. This may be because diffuse GAVE is more common in liver cirrhosis.16,20 It should be acknowledged that APC is more effective for diffuse GAVE, which is mainly characterized as small lesions with local hemorrhage, but achieves incomplete coagulation for striped GAVE, which is often characterized as large lesions. 29

Outcomes of GAVE patients undergoing APC.
Transfusion after APC
The transfusion-dependent rate after APC is 9.1–50%21–23,25,30–32 [Figure 2(b)]. The causes of transfusion after APC are often varied. In a study by Yussof et al., 23 of the 4 patients who were dependent on blood transfusion before APC, 2 required blood transfusion 8 and 12 months after the last APC due to the recurrence of endoscopically confirmed GAVE. In another study, 2 patients still required continuous blood transfusion after APC, mainly due to bleeding from varices in the esophagus and rectum. 21 In the study by Sebastian et al., 30 two patients required blood transfusion after APC, but did not develop GAVE recurrence. One was due to bleeding from non-steroidal anti- inflammatory drug (NSAID)-related ulcers; another did not have any definite cause of bleeding. As a result, except for GAVE recurrence, other causes for transfusion after APC should be considered. It should be noted that follow-up endoscopy is necessary to clarify the reasons for transfusion requirement.
Recurrence after APC
Dilated capillaries can recur a few months after APC for GAVE. The incidence of recurrence is 10–78.9%21–27,33–38 [Figure 2(c)]. The annual recurrence rate of GAVE is 34–50.3%.24,25 Such a wide difference may be related to the heterogeneity in the definition of recurrence, duration of follow-up, and characteristics of included patients. Herrera et al. 36 reported a low recurrence rate of 12.5%, but the recurrence was evaluated according to the patient’s clinical manifestations without endoscopy follow-up. Similarly, Roman et al. 21 reported a recurrence rate of 15.8% by using the same definition of recurrence without endoscopic follow-up. In addition, it should be noted that 57.9% (11/19) of the patients died without information regarding recurrence, which contributed to a low recurrence rate. By comparison, Nakamura et al. 24 followed 22 patients by blood tests every month and endoscopy every 3 months, and found that the 1-, 2-, and 3-year cumulative recurrence-free rates after APC were 49.7%, 35.5%, and 35.5%, respectively. It should be noted that in the aforementioned study, recurrence was strictly defined as more than 50% of lesions recurred or active bleeding was observed under endoscopy. On the other hand, the studies, in which a majority of patients had chronic liver disease, found a very high recurrence rate of 66.6–78.9%.26,38 Whether chronic liver disease can affect GAVE recurrence after APC is still uncertain. In a study of GAVE patients with liver cirrhosis, some variables, including gender, age, the etiology of cirrhosis, CTP score, medication (blockers, anticoagulant, or antiplatelet drugs), INR, prothrombin rate, and platelet count, were analyzed. However, none of them can significantly predict GAVE recurrence after APC. 25
Adverse events of APC
The incidence of adverse events after APC is 0–20.5%19,21–26,28,30,31,33,35,39–41 [Figure 2(d)]. Procedure-related adverse events are often dependent on the operator’s skills, power output and argon flow rate, distance between equipment and mucosa, and residence time at the same lesion site.
Minor adverse events mainly include epigastric pain, 40 abdominal distention, 40 and mild ulcer bleeding.30,37 They can be alleviated by oral proton pump inhibitors in combination with over-the-counter analgesics in a few days. Hyperplastic gastric polyps25,35,37 and fever 40 after APC have also been reported. Major adverse events are uncommon, including Mallory–Weiss syndrome, 32 scarring of ulcer,22,32 and sepsis.21,35 Probst et al. 22 reported that one patient developed deep antrum ulcers followed by circumferential scarring of the antrum and asymptomatic stenosis 6 months after APC, and planned to undergo Billroth I surgery. Roman et al. 21 found that one patient developed sepsis after APC and then died of infectious peritonitis 4 months after APC.
Hybrid-APC
Hybrid-APC is a modified procedure that creates a ‘safety cushion’ by injecting saline into the submucosa and achieves deeper lesions. In a prospective study, hybrid-APC was used in 9 patients, of whom 8 received 1 session and 1 received 2 sessions. After 6 months, the hemoglobin level reached a normal level in all of the 9 patients without any serious complication. 42 As a result, hybrid-APC may be a safer choice for the treatment of GAVE as compared with traditional APC.
Combination of polidocanol and APC
Polidocanol injection can be used for the management of small-bowel angioectasia, 43 chronic venous disorder, 44 and gastrointestinal ulcer bleeding 45 with excellent results. A combination of polidocanol and APC can treat mucosa and submucosa lesions to radically treat GAVE. 46 In a recent retrospective study, 15 GAVE patients were treated with a combination of polidocanol and APC. Only polidocanol was used as an initial choice of treatment; and repeated APC was employed for the remaining lesions 3–4 days later. All patients achieved a successful hemostasis. However, two (13%) developed treatment-related adverse events, of whom one developed ulceration and another hematoma; and both received conservative treatment. In addition, two (13%) recurred at 36 months and 48 months after hemostasis. 46
Radiofrequency ablation
Since the first report by Ganz et al. 47 in 1999, RFA has been widely used for superficial gastrointestinal diseases, such as angiodysplasia, chronic radiation proctitis, and precancerous lesions. 48 In recent years, RFA has become an alternative technique to APC for the treatment of GAVE (Supplemental Table 2).
The RFA device includes an energy generator and an ablation catheter. 49 The catheters are used to achieve appropriate tissue contact; and they include through-the-scope catheter, HALO 60 , and HALO90, whose ablation area is 1.2 cm2, 1.5 cm2, and 2.6 cm2, respectively. The HALO90 ULTRA catheter can even ablate a larger area of 5.2 cm2 each time and achieve rapid hemostasis. 50
Gross et al. 51 first reported that 6 patients with GAVE were treated with RFA under the HALO90 system in 2008, of whom 4 had failed APC treatment. The mean hemoglobin level increased from 8.6 g/dl to 10.2 g/dl during a mean follow-up period of 2 months after RFA. Of these patients, 5 (83.3%) did not require blood transfusion. All patients tolerated the RFA procedure without complications. However, it should be acknowledged that the sample size is small and follow-up duration is short for the first report. Since then, more studies have explored the efficacy of RFA, especially in patients with refractory GAVE who present with iron deficiency anemia requiring transfusion or gastrointestinal bleeding in spite of several other treatments.
Endoscopic success of RFA
The endoscopic success rate of RFA is up to 90%29,49,52–54 [Figure 3(a)]. However, there is a probability of persistent gastrointestinal bleeding after complete ablation by the HALO90 ablation system.49,52

Outcomes of GAVE patients undergoing RFA.
Transfusion after RFA
Transfusion requirement after RFA remained in 13.3–40% of GAVE patients29,49,51,52,54–58 [Figure 3(b)]. In a large case series by Dray et al., 56 23 GAVE patients required transfusion before RFA; and 65.2% (15/23) of them were weaned off transfusion within 6 months after RFA.
It seems that neither initial endoscopic findings nor previous treatments significantly affect transfusion requirement after RFA. 56 The major causes of transfusion requirement after RFA are persistent hemorrhage49,52 and recurrent bleeding during follow-up. 55 Other sources of bleeding, such as jejunal arteriovenous malformation, 49 necessitate transfusion requirement.
Recurrence after RFA
The recurrence rate after RFA is 21.4–33.3%53,55 [Figure 3(c)]. In a retrospective study, 9 patients with refractory GAVE were included. After the eradication of the lesions, they were followed clinically with serial measurement of their hemoglobin concentration. Three patients (33.3%) recurred and needed repeated treatment. One had the first recurrence 3 months after the first treatment and the second recurrence 3 months after the repeated ablation, while the other 2 had recurrence 11 months after the last treatment. All of the 3 patients had chronic kidney diseases; and 1 of them was also complicated with liver cirrhosis. 53 In another study by Magee et al., 55 14 patients were followed for 12 months after RFA, of whom 3 (21.4%) had recurrences that manifested as recurrent anemia and required blood transfusion or iron supplement. It should be noted that in the two studies mentioned herein, repeated endoscopy after complete ablation was performed only in the case of increased blood transfusion requirement. As a result, more accurate information regarding recurrence of striped or diffuse GAVE lesions after complete ablation needs to be further confirmed by regular follow-up endoscopy.
Adverse events of RFA
The incidence of adverse events after RFA is 0–15%29,49,51–53,56,59 [Figure 3(d)]. Like other endoscopic thermal therapies (ETTs), such as heat probe electrocoagulation and APC, adverse events associated with RFA are primarily attributed to mucosal injury, including hyperplastic polyps, superficial ulcers, hemorrhagic ulcers, and even sepsis. The formation of fragile sessile and pedunculated polyps may be related to both high gastrin status and repeated thermal injury. 60 Superficial and hemorrhagic ulcers may be conservatively controlled without any intervention. 52 Sepsis may be treated by antibiotics. 61 Gastroesophageal junction tear related to RFA with HALO90 has been also reported. 62
RFA versus APC
RFA has some technical advantages, including a contact therapy modality which can counteract severe antral contractions, 51 a wider ablation area which can avoid incomplete treatment, and a reproducible depth of treatment with preset power output and energy density that can achieve a uniform ablation and eliminate the influence of operator.49,63 However, whether RFA is superior to APC for the treatment of GAVE is still controversial. Some studies have found that RFA has higher endoscopic and clinical success rates,64,65 requires less procedural times 66 and treatment sessions, 65 reduces costs, shortens the length of hospital stay,54,59 prolongs the length of retreatment, 66 and improves hemoglobin levels more effectively.64,66 In contrast, another study considered that APC might be more dominant in increasing hemoglobin level and reducing transfusion requirement. 64
In a large cohort study, 24 underwent APC alone and 28 underwent RFA alone. There was no significant difference in the improvement of hemoglobin level, number of treatment sessions, procedural times, and retreatment intervals between the two groups. 63 It should be noted there was a potential bias of patient selection due to the retrospective nature of this study; it was more likely that patients who underwent RFA were more severe. Unfortunately, until now, there is no randomized controlled trial comparing RFA versus APC.
Endoscopic band ligation
EBL is the standard treatment for esophageal varices located in the submucosa. 67 As a result, it is theoretically feasible and effective for GAVE located at the gastric mucosa and submucosa 68 (Supplemental Table 3). EBL is usually initiated from the pylorus until most of the lesions located at the antrum are treated. 69 The number of bands applied for the treatment of GAVE is dependent upon the areas of the lesions and the choices of the endoscopists. After EBL, submucosal capillary thrombosis develops, further leading to various degrees of ischemic necrosis followed by superficial ulcer and finally submucosal fibrosis. 70
Endoscopic success of EBL
The endoscopic success rate of EBL is 77.8–100%38,68 [Figure 4(a)]. In a study, a total of 9 patients were included, of whom 7 and 2 manifested as overt and occult bleeding, respectively. Of these, bleeding events were successfully controlled in 7 patients after an average of 1.9 EBL sessions, while hemostasis failed in the remaining 2 patients. 68

Outcomes of GAVE patients undergoing EBL.
Transfusion after EBL
Transfusion requirement after EBL remained in 15.4%–55.6% of GAVE patients32,69 [Figure 4(b)]. In a retrospective study, the indication of blood transfusion was defined as the amount of blood lost reached beyond 30% of the total blood volume in patients with acute hemorrhage, or if the hemoglobin concentration was less than 7 g/dl in patients with chronic hemorrhage. During a mean follow-up duration of 18 months, 55.6% (10/18) of patients still required blood transfusion after an average of 2.2 treatment sessions. 32 In another study, 2 of 14 patients included did not wean off blood transfusion, probably because both of them had chronic renal failure with diffuse GAVE. 69
Recurrence after EBL
The recurrence rate after EBL is 8.3–48.1%38,71 [Figure 4(c)]. In one study, where 12 patients were treated with EBL, endoscopy revealed that only 1 (8.3%) patient developed the recurrence of GAVE during a mean follow-up duration of 14.8 months. 38 In contrast, in another retrospective study where 27 patients were treated with EBL, 13 (48.1%) recurred during a mean follow-up duration of 18.2 months. 71 A higher transfusion requirement before EBL was the only predictor of recurrence. 71
Adverse events of EBL
The incidence of adverse events after EBL is 8.3–13.6%32,38,40,68,69 [Figure 4(d)]. Common adverse events after EBL include nausea and vomiting 68 and mild esophagus or epigastric pain. 69 Most of them do not require any intervention and disappear within a few days. Other adverse event, such as bleeding from ulcers, can be solved by conservative treatments. 38 Hyperplastic polyps have also been occasionally reported. 32
EBL versus ETT
EBL should be more advantageous on occlusion of mucosal and submucosal vessels as compared with ETT. Some studies have demonstrated that EBL was significantly superior to ETT in hemostasis, 68 eradication of visible lesions on endoscopy, 28 treatment sessions,32,68,72 improvement of hemoglobin, 72 transfusion requirement,68,72 and recurrence. 38 In contrast, other studies found no difference between the two groups in terms of treatment sessions,28,38 blood transfusion requirement, 32 improvement of hemoglobin,28,32,68 and adverse events. 72 In a randomized controlled trial, 88 patients with GAVE were assigned to either the APC or EBL group. They found that the EBL group needed less treatment sessions and a lower amount of blood transfusion than the APC group; however, there was no significant difference in hemoglobin improvement or mild adverse events between the two groups. 40
Conclusion
At the moment, APC is often the first-line choice of endoscopic treatment for GAVE, but it requires more treatment sessions due to its higher risk of recurrence. RFA and EBL have been considered as alternative approaches. Notably, some comparative studies suggest that RFA and EBL may be superior to APC in terms of hemostasis, treatment sessions, and recurrence. However, the optimal endoscopic treatment option for GAVE has not yet been sufficiently confirmed by high-quality randomized controlled trials. In addition, underlying diseases, other sources of bleeding, indications of blood transfusion, and definitions of treatment success vary widely among cohorts, which are also major limitations of the currently published studies.
Supplemental Material
sj-docx-1-taj-10.1177_20406223211039696 – Supplemental material for Endoscopic treatment for gastric antral vascular ectasia
Supplemental material, sj-docx-1-taj-10.1177_20406223211039696 for Endoscopic treatment for gastric antral vascular ectasia by Mengyuan Peng, Xiaozhong Guo, Fangfang Yi, Xiaodong Shao, Le Wang, Yanyan Wu, Chunmei Wang, Menghua Zhu, Ou Bian, Mostafa Ibrahim, Saurabh Chawla and Xingshun Qi in Therapeutic Advances in Chronic Disease
Supplemental Material
sj-docx-2-taj-10.1177_20406223211039696 – Supplemental material for Endoscopic treatment for gastric antral vascular ectasia
Supplemental material, sj-docx-2-taj-10.1177_20406223211039696 for Endoscopic treatment for gastric antral vascular ectasia by Mengyuan Peng, Xiaozhong Guo, Fangfang Yi, Xiaodong Shao, Le Wang, Yanyan Wu, Chunmei Wang, Menghua Zhu, Ou Bian, Mostafa Ibrahim, Saurabh Chawla and Xingshun Qi in Therapeutic Advances in Chronic Disease
Supplemental Material
sj-docx-3-taj-10.1177_20406223211039696 – Supplemental material for Endoscopic treatment for gastric antral vascular ectasia
Supplemental material, sj-docx-3-taj-10.1177_20406223211039696 for Endoscopic treatment for gastric antral vascular ectasia by Mengyuan Peng, Xiaozhong Guo, Fangfang Yi, Xiaodong Shao, Le Wang, Yanyan Wu, Chunmei Wang, Menghua Zhu, Ou Bian, Mostafa Ibrahim, Saurabh Chawla and Xingshun Qi in Therapeutic Advances in Chronic Disease
Footnotes
Authors’ contributions
Conceptualization: Xingshun Qi;
Investigation: Mengyuan Peng, Xiaozhong Guo, Fangfang Yi, Xiaodong Shao, Le Wang, Chunmei Wang, Yanyan Wu, Menghua Zhu, Ou Bian, Mostafa Ibrahim, Saurabh Chawla, and Xingshun Qi;
Writing–original draft: Mengyuan Peng and Xingshun Qi;
Writing–review and editing: Mengyuan Peng, Xiaozhong Guo, Fangfang Yi, Xiaodong Shao, Le Wang, Chunmei Wang, Yanyan Wu, Menghua Zhu, Ou Bian, Mostafa Ibrahim, Saurabh Chawla, and Xingshun Qi;
Supervision: Xiaozhong Guo and Xingshun Qi.
Conflict of interest statement
The authors declare that there is no conflict of interest.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
