Gastric antral vascular ectasia (GAVE) is an infrequently encountered vascular lesion located primarily in the gastric antrum. It accounts for approximately 4% of cases of non-variceal upper gastrointestinal bleeding. It is characterised by a distinctive endoscopic appearance consisting of visible convoluted vessels which may present in a diffuse disorganised and nodular pattern located in the antrum, or in prominent organised streaks and stripes emanating from the pylorus resembling the stripes of a watermelon (Fig. 1). Endoscopic description of extent and severity is largely subjective, with the decision to treat and the selection of treatment modality generally based on endoscopists’ preference. While histological diagnosis is rarely required, biopsies show dilated, ectatic capillaries in the mucosa and submucosa with microthrombi and minimal inflammation [1,2].

GAVE was first described as an arteriovenous malformation of the stomach and treated by antrectomy with Billroth-I anastomosis formation, which while resulting in an improvement in hemoglobin levels, resulted in significant morbidity [3,4]. Early endoscopic therapies included heat probe [5], bipolar electrocoagulation [6], sclerotherapy [7], and laser therapy [8]. These were initially effective endoscopic modalities but were limited by cost, availability, and the risk of deep thermal injury. Argon plasma coagulation (APC), which induces coagulation of the tissue surface using a high-frequency electrical current conducted via ionised argon gas, has become the treatment modality of choice for most endoscopists as it is easy to use, relatively safe, and familiar to endoscopists (Fig. 2). The depth of penetration usually ranges from 1–3 mm, significantly reducing the risk of perforation. Initial reports demonstrated an 80% success rate in reducing transfusion requirements [9,10], for both the diffuse-type and the stripe-type of GAVE [11]; however, it often required multiple treatment sessions [12]. More recent studies have highlighted the limitations of APC on follow-up, with higher treatment failures, and only a small number of patients remaining transfusion free post treatment [13-16]. Radiofrequency ablation (RFA), which transmits high-power energy in short bursts to ablate superficial mucosal lesions, has been proposed as an alternative to APC, particularly for treatment-refractory cases [17,18]. However, there are few prospective trials investigating RFA in GAVE with sparse data on long-term outcomes.

Endoscopic band ligation (EBL) has become an increasingly used alternative treatment modality [19], demonstrating a good safety profile and favourable effectiveness [20]. A clinical response of >90% has been reported on short term follow-up [21]. Given that the vascular lesions observed in GAVE often involve the submucosa, the deeper penetration offered by band ligation should result in more effective hemostasis (Fig. 3). Recent studies demonstrating lower re-bleeding rates following EBL as compared to APC, and higher treatment success rates would support this [19,22]. However, a recent retrospective study demonstrated a high initial rate of clinical response with falling recurrence-free survival over time, suggesting the longterm outcomes may not be as successful as proposed [23].

While an increasing number of treatment options are being explored for GAVE, currently no evidence-based treatment algorithm exists to direct the appropriate treatment approach for GAVE. It remains unknown whether certain sub-types of GAVE respond better to particular treatment modalities, or whether treatment could be tailored or indeed, individualised.

The aim of this study was to retrospectively evaluate the clinical course of patients treated with either APC or EBL for GAVE to determine whether either one conferred better longterm outcomes.

This study evaluated the comparative effectiveness of APC and EBL as treatment modalities for GAVE. Using a retrospective approach, we tracked outcomes using the index treatment modality as a starting point. The decision to treat and the choice of therapy were based on the endoscopists’ preference in real time, with experience the only influencing factor. The median follow-up period for the cohort was 19.9 months (range, 2–55).

The precise goal of treatment in GAVE is an interesting subject and one that lacks consensus—whether total eradication of GAVE is necessary, or whether control of bleeding by way of stable hemoglobin/iron levels is sufficient. As a retrospective study, we aimed to evaluate clinical outcomes based on the need for repeat procedure/intervention, hemoglobin levels post procedure, and transfusion requirements. In that regard, treatment outcomes were based on the more conservative measure of evidence of bleeding. Subsequent treatment sessions were booked based on the endoscopists’ subjective assessment of necessity.

In our centre, as in most, APC remains the default treatment of choice in the treatment in GAVE. This is largely due endoscopists’ experience with this modality in other settings such as in the treatment of radiation proctitis and angiodysplasia. It is familiar to both endoscopists and endoscopy nurses, relatively easy to use, and provides immediate results. The applied mucosal injury gives a satisfactory visual result; however, the superficial nature of the therapy leads to significant rates of recurrence. As a result, alternative therapeutic modalities have been explored.

As a relatively novel treatment modality, the incidence of EBL use is comparatively low in our cohort and its use was limited in the early stages of the study period. However, we note that its use increased during the study period, with patients initially treated with APC receiving crossover treatment with EBL. The ability of this modality to penetrate deeper into the submucosa, with resulting and subsequent thrombosis, underlies its potential as a more effective therapeutic option [24]. As endoscopists become more familiar with this treatment option, the frequency of its use will likely rise further.

Our data suggests a more effective endoscopic response with EBL compared to APC, with fewer subsequent treatment sessions required and a lower transfusion requirement post treatment. This analysis includes those treated with primary EBL and those treated with ‘rescue’ EBL, or EBL after treatment failure with APC. The analysis in the rescue group failed to reach statistical significance likely due to the small number of patients involved in this subgroup. Blood transfusion requirements were noted to be lower 3 months post procedure in the EBL group when compared to the APC group. This crude data looked at overall transfusion requirements for the EBL and APC groups and did not correct for potential bias. As we posit that more severe cases were selected for treatment with EBL (this is supported by the lower mean hemoglobin level observed in the EBL group pre-treatment), any bias would likely inflate the pre-procedure transfusion requirement in this group. The overall reduction (reduction of 55% in EBL compared to 15% in APC), while crude, demonstrates an important difference between the groups.

In our cohort, patients with cirrhosis and GAVE were less represented than in some earlier studies. Nonetheless, the proportion of associated conditions reported are similar to those in recent studies assessing therapeutic intervention in GAVE [18]. The lower proportion of cirrhosis in our study may be explained by: (1) over-diagnosis of GAVE in cirrhotic patients with portal hypertensive gastropathy, and (2) cirrhotic patients with GAVE may represent a milder form and were not considered for therapy.

The main limitation of this study is its retrospective nature. Patients were not randomised to a particular treatment modality. In spite of this, the data we present does demonstrate a clear difference in outcomes between the two groups and presents EBL as a more effective treatment modality.

There is currently no consensus or definition of severity of disease in GAVE, or indeed, evidence to suggest whether different subtypes of disease (i.e., diffuse vs. striped) would benefit from particular treatment modalities. We suggest that cases receiving endoscopic therapy by definition represent moderate to severe GAVE; however, this is based on a subjective assessment by the endoscopist. In the absence of evidence-based treatment algorithms, the decision to treat endoscopically as well as the choice of therapy is based on endoscopist’s experience and preference. This may result in inconsistent approaches to the management of GAVE and/or inappropriate or ineffective therapeutic interventions.

In our cohort, with a mean age of 74 years and a female preponderance, APC was the commonest treatment modality employed; however, we demonstrated an increasing incidence of EBL. As a relatively new treatment modality, the incidence of EBL as an index treatment was perhaps disappointingly low in our cohort. However, we do note an increase in its use as a rescue therapy for patients who are refractory to APC. Patients treated with EBL at their index treatment required fewer subsequent treatment sessions, had a greater mean rise in hemoglobin, and had a lower transfusion requirement post treatment. This data suggests a more effective endoscopic response when EBL is employed in the treatment of GAVE, either as a primary treatment modality or after APC failure. Whether certain GAVE subtypes may benefit from different treatment modalities is beyond the scope of this study but represents an appealing future research question. We suggest in our study that more severe cases of GAVE may be preferentially selected for treatment with EBL as more diffuse and severe presentations may respond better to the deeper penetration of EBL compared to APC. Differentiating between GAVE subtypes may facilitate the development of an evidence-based treatment algorithm which could be employed to select which patients require treatment as well as the most appropriate treatment modality.