Management of gastric varices


  • Conflict of interest
    The authors do not have conflicts of interest to disclose.

Makoto Hashizume, Department of Advanced Medical Initiatives, Faculty of Medical Sciences, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan. Email:


Although the incidence of bleeding from gastric varices is relatively low (10%–36%), the bleeding is massive once it has occurred and it increases the patient's mortality. The management of esophageal variceal bleeding is highly differentiated with several effective treatments available. In contrast, bleeding from gastric varices continues to be a therapeutic challenge. In the last decade, there have been increasing reports regarding the management of gastric varices. In this article we review recent progress in the management of gastric varices and discuss further expected studies.


The classifications of gastric varices commonly used in clinical studies are those of Sarin1 and Hashizume.3 Some modifications have been added and used in various countries.

According to Sarin's classification, diagnosis of gastric varices is based on the presence of anatomical continuation with esophageal varices as well as their location in the stomach (Fig. 1). When the gastroesophageal varices (GOV) are an extension of esophageal varices, they are categorized into two types. The most common are Type 1 gastroesophageal varices (GOV1), which extend along the lesser curvature. They are considered extensions of esophageal varices and the recommended management is the same as that of esophageal varices. Type 2 gastroesophageal varices (GOV2) are those which extend along the fundus; these tend to be longer and more tortuous than Type 1 gastric varices. Isolated gastric varices (IGV) occur in the absence of esophageal varices and are also classified into two types. Type 1 (IGV1) are located in the fundus and tend to be tortuous and complex, and type 2 (IVG2) are located in the body, antrum, or around the pylorus. The IGV1 fundic varices do not include the gastric varices caused by splenic vein thrombosis. The most commonly encountered gastric varices with hemorrhage are GOV1, GOV2 and IGV1. Sarin's classification is useful for considering the management of gastric varices. Bleeding from the GOV1 is relatively straightforward with endoscopic injection sclerotherapy (EIS) or endoscopic variceal ligation (EVL), while it is still hard to control bleeding from fundic gastric varices, such as GOV2 and IGV1.

Figure 1.

Classification of gastric varices (GV) according to Sarin et al.1

Hashizume et al., on the other hand, proposed the classification of the gastric varices based on the clinically significant endoscopic findings, and particularly from the view point of findings associated with the highest risk of these most likely to rupture, as in the classification of esophageal varices (Fig. 2). Thus, endoscopic findings of gastric varices were classified according to their form, location, and color. The form was classified into three types: tortuous (F1), nodular (F2), and tumorous (F3). The location was classified into five types: anterior (La), posterior (Lp), lesser (Ll) and greater curvature (Lg) of the cardia, and fundic area (Lf). The location of the gastric varices depends on hemodynamic factors. The color can be white (Cw) or red (Cr). The glossy, thin-walled focal redness on the varix was defined as red color spot (RC spot). The Hashizume Group reported that the RC spot and larger forms were related to a significantly higher risk of gastric variceal bleeding.

Figure 2.

Schematic diagram of endoscopic findings, classified by Hashizume et al.2

There is a difficulty in comparing the results from clinical studies of gastric varices. This is because each study includes a variety of types of gastric varices according to their own classification. It is clinically important to use the same classification of gastric varices based on the endoscopic findings according to the same rule in each study. Better management of gastric varices would be provided by application of evidence based medicine, in which results have been documented according to the underlying anatomical and endoscopic findings.

Hemodynamics of gastric varices

In patients with portal hypertension, there is a portal and systemic hyperdynamic state, and esophageal or gastric varices develop as one part of the collateral circulation. It is not yet known, when or in whom esophageal or gastric varices will develop. Gastric varices often develop in the submucosal layer at the cardia or the fundus of the stomach, which location is consistent with the boundary line area of porto-systemic shunting. This is mainly because the posterior wall of the cardiac or the fundic area is fixed to the retroperitoneum and is the closest site to the systemic circulation via porto-systemic shunts. The hyperdynamic state of portal hypertension is characterized by the existence of either or both higher arterial and venous inflow, and the higher venous outflow vessels associated with a major decrease in peripheral vascular resistance. The left gastric vein, posterior and short gastric veins are the main supplying vessels to gastric varices,10,11 while the gastro-renal shunt is the main drainage vessel (Fig. 3). It is important to confirm the supplying vessels and the drainage vessels for the management of the gastric varices. To know the local hemodynamics of the gastric varices is the first step to selecting the best choice for the effective treatment of the gastric varices.

Figure 3.

Percutaneous transhepatic portography of gastric varices. Percutaneous transhepatic portography (PTP) showed giant gastro-renal shunt (arrow).

A major porto-systemic shunt, such as a gastrorenal shunt, is present in up to 85% of patients with gastric varices.4,11 The diameter of the huge gastro-renal shunt which is often encountered is about one to three centimeters. The volume of blood flowing through the shunt and the velocity of the porto-systemic shunt are extraordinarily large. This is one reason why conventional endoscopic injection sclerotherapy (EIS) is usually not sufficient. It could also be relative to possible serious complications, such as pulmonary embolism or massive ulcer bleeding. Recently, multidirection-computer tomography (MD-CT) provides the precise information such as the vascular architecture of the gastric varices without angiography.11,12 To know the hemodynamics of the portal circulation, including the supply and the drainage vessels, is very helpful in selecting the best treatment choice for each patient with gastric varices. Balloon-occluded retrograde transvenous obliteration (B-RTO) is the most promising and the most effective treatment in Japan, although it is mostly applied to prophylactic cases when a gastro-renal shunt exists.13–15

Management of gastric varices

A variety of treatments have been developed for esophageal varices since the 1940s including porto-caval shunts, selective shunts, or esophageal transaction, as well as endoscopic treatments. Most approaches have been performed successfully and clinical results have been acceptable when the indications have been appropriately applied. However, the management of gastric varices still remains a therapeutic challenge. Because there are few controlled clinical trials, much less confidence can be placed on guidelines for the management of gastric varices than for their esophageal counterparts.

Endoscopic treatment

Type 1 gastric varices (GOV1) constitute an extension of esophageal varices along the lesser curvature of the stomach. Therefore, the approach to their management should be the same as for esophageal varices. According to the reports about GOV1 gastric variceal bleeding, hemostasis and re-bleeding rates are similar to those in the management of esophageal variceal bleeding.4

On the other hand, the management of bleeding from the cardiac or fundic varices, which are classified into GOV2 or IGV1, is quite different from GOV1. A number of investigators have reported that traditional endoscopic injection sclerotherapy (EIS) is ineffective for the treatment of the isolated gastric varices.16,17 The reason is that gastric varices exist associated with a gastro-renal shunt or a gastro-inferior vena caval shunt, resulting in outflow into the systemic circulation.18 These anatomical characteristics with a major port-systemic shunt create a higher blood flow volume through the shunt, with resultant rapid escape of sclerosant into the systemic circulation during EIS. As a result conventional EIS does not allow the sclerosing agent to initiate thrombosis on the surface endothelium of the gastric varices. Further, there is the risk of such serious complication as pulmonary embolism with the sclerosing agent via the major shunt, or massive ulcer bleeding induced by a puncturing the huge gastric varices.


Compared to endoscopic injection sclerotherapy (EIS) or esophageal variceal ligation (EVL), endoscopic variceal obturation with a tissue adhesive such as N-butyl-cyanoacrylate, or isobutyl-2-cyanoacrylate is more effective for acute fundic gastric variceal bleeding. The results include a better rate of controlling the initial hemorrhage as well as lower re-bleeding rate.19–23 A relatively large prospective randomized trial which compared gastric variceal obturation (GVO) with N-butyl-cyanoacrylate versus EVL in patients with acute gastric variceal hemorrhage demonstrated that the control rate of active bleeding was similar in both groups. However, re-bleeding over a follow-up period of 1.6–1.8 years occurred significantly less frequently in the GVO group (23% versus 47%), with an average of only 1.5 sessions (range 1–3). An international consensus meeting at Baveno IV in 2005 adovocated that a tissue adhesive, such as cyanoacrylate, is the only agent recommendable for control of bleeding from fundic gastric varices.24 Although prospective randomized control studies have recently been reported regarding the management of gastric variceal hemorrhage (Table 1), it still remains difficult to compare each study. The problem is that various types of gastric varices have been included without a definite explanation or classification of the varices. For example, Tan26 and Lo27 's randomized controlled studies including more than 50% of patients, who had GOV1 gastric varices. As reported previously, GOV1 gastric varices are as well controlled by endoscopic ligation or sclerotherapy as esophageal varices. It would be expected that conventional treatments for esophageal varices such as TIPS and EIS would be effective for those patients with GOV1 gastric varices. Therefore, it would be desirable to limit any further studies to isolated cardiac or fundic gastric varices that we classified into GOV2 and IGV1 according to Sarin's classification.

Table 1.  Recent reported randomized controlled trial studies of endoscopic treatment for gastric varices
First author (year)Classification (GOV1/ GOV2 /IGV1)TreatmentHemostasis rateRe-bleeding rateFollow up
Sarin et al.25 (2002)0/8/28GVS (n = 17)62%33%15.4 months
GVO (n = 20))89%27%
Tan et al.26 (2006)53/25/19GVL (n = 48)93% (14/15)44%610 days
GVO (n = 49)93% (14/15)22%680 days
Lo et al.27 (2007)36/33/0TIPS (n = 35GVO (93%)11%32 months
GVO (n = 37) 38%
Mishra et al.28 (2010)0/all GOV2 or IGV1GVO (n = 33)N.D15%26 month
β-blocker (n = 34) 55%

Other new agents


The alternative agent for endoscopic treatment is thrombin. Yang29 evaluated the usefulness of human thrombin in 12 patients with isolated gastric varices. Immediate hemostasis was achieved in all patients, among whom there were six with active bleeding, the remainder with stigmata of recent bleeding. The re-bleeding rate was 27%. Ramesh30 also reported experience with the use of human thrombin in 13 patients. Interestingly, the rates of hemostasis and re-bleeding from gastric varices were 92% and 0%, respectively. The limitation of both studies was small patient number and short duration. It is regrettable that there have been no further studies after these reports. It is also suspicious from the hemodynamic viewpoint as to whether a small volume of thrombin could be truly effective in provoking occlusion of large gastric varices with thrombosis, resulting in control of bleeding from the gastric varices with a major gastro-renal shunt. Thrombin may leak into the systemic circulation in the case of gastric varices with high flow volume and associated with a giant gastro-renal shunt. Intravascular injection of thrombin could then induce disseminated intravascular coagulation (DIC) or pulmonary embolism. Further prospective study is necessary in the future.

Beriplast P

Beriplast P consists of two components, fibrinogen with factor VIII, and human thrombin. Beriplast P has been used with the aim of achieving hemostasis against intra-abdominal oozing during surgery. The procedure requires a double lumen injector to mix the two contents simultaneously on the surface of bleeding tissue. There are two uncontrolled studies which have recently been reported showing the efficacy of Beriplast P in patients with gastric variceal bleeding.31,32 The results were satisfactory, but the number of patients included into the studies was so small that further investigation with significant numbers of patients is needed.


Esophageal variceal ligation (EVL) was introduced by V. Stiegman as a faster and easier treatment against bleeding esophageal varices. It is well indicated for small-sized gastric varices or gastric varices with concurrent esophageal varices. However, gastric varices with active bleeding balloon with tension and are larger in size than esophageal varices, so it is sometimes hard to trap the whole body of the huge varices into a hood. A major problem with use of EVL for management of gastric varices is ulcer formation; this may lead to a severe defect in the gastric wall, including the gastric varix itself. In a randomized controlled trial,20 Lo et al. showed that endoscopic obturation by injection of cyanoacrylate was more effective than EVL. Therefore, EVL is not recommended for large gastric varices. Shiha and Lee reported the usefulness of the detachable snare as an alternative EVL method.33,34 Follow-up data and further results have not yet been reported. Therefore, the efficacy of the detachable snare is to be evaluated in further studies.

Whether snare ligation is successful or not depends on the form of the gastric varices. Because the area with snare ligation is wider than that with a conventional band ligation, ulcer formation following the snare ligation might lead to life-threatening bleeding.

Pharmacologic treatment

There are few reports on pharmacological treatment for gastric varices. Pharmacologic treatment might be effective in control of bleeding from cardiac gastric varices in co-existence with esophageal varices, so called GOV1 according to the Sarin's classification. However, the isolated fundic gastric varices such as GOV2 or IGV1, have not been addressed in previous studies. GOV2 and IGV1 gastric varices are mostly associated with a major port-systemic shunt, so portal vein pressure is lower in patients with those gastric varices than in patients with esophageal varices.4 As a result the efficacy of conventional drugs such as vasodilators or vasoconstrictors is doubtful in the management of gastric varices because of the hyperdynamic state and presence of a major porto-systemic. Only one report has investigated the efficacy of vasoactive agents on bleeding from gastric varices. As shown in Table 2, Mishra et al.28 examined a beta-blocker on secondary prophylaxis of gastric variceal re-bleeding. In this study, a beta-bloker was shown to be inferior to endoscopic cyanoacrylate injection therapy. A beta-blocker with another drug might be effective for prevention of the first gastric variceal bleeding, but a prospective randomized study on the use of vasoactive agents for the purpose of prevention of the first gastric variceal bleed is desirable.

Table 2.  Reported studies of balloon-occluded retrograde transvenous obliteration (B-RTO) for gastric variceal hemorrhage
First author (year)No of patientInitial treatmentSuccess rate5-year rebleeingFollow up (mean)
  • 5 year cumulative gastric varceal re-bleeding rate.

  • including prophylactic treatment.

Ninoi et al.13 (2005)34No data87%3.1%(1/34)700 days (23 month)
Arai et al.45 (2005)11GVL (n = 2)100%0%1136 days (37 month)
Clipping (n = 2)
Spontaneouly Stopped (n = 7)
Hiraga et al.46 (2007)34Spontaneouly Stopped (n = 20)91%0%33 months
Balloon tamponade (n = 6)
Endoscopic treatment (GVL,GVO,EIS,Clipping) (n = 8)
Akahoshi et al.15 (2008) (our data)20GVO (n = 10)94%5.5%(1/18)66 months
Balloon tanmonade (n = 4)
Spontaneously stopped (n = 6)


Transjugular portosystemic shunt (TIPS) is used in cirrhotic patients with liver failure and bleeding esophageal varices as a bridging method until they are able to undergo liver transplantation. It has not been recognized as first line therapy for gastric variceal bleeding. However, when uncontrolled bleeding from gastric varices with endoscopic or pharmacologic treatment had been encountered, TIPS might be a choice for salvage treatment.35,36

If liver function is tolerable for an operation under general anesthesia, surgery might be indicated for uncontrolled bleeding from the gastric varices. If liver function was so poor that surgery would not be tolerated, TIPS might be considered as one of the treatments. It has been reported that the control rate of gastric variceal bleeding with TIPS is over 90%.27,35–37 Although it has been suggested that bleeding from gastric varices can be more difficult to control with TIPS than bleeding from esophageal varices, a prospective study compared salvage TIPS in patients with uncontrolled fundic gastric variceal bleeding (n = 28) versus patients with uncontrolled esophageal variceal bleeding (n = 84) and showed equal efficacy; there was control of hemorrhage in all but one patient in each group.37 When the operator is familiar with TIPS, TIPS might be effective in patients with gastric varices, who had a significantly higher portal venous pressure than the hepatic venous pressure. Thus, the measurement of HVPG would be useful for making a decision to select the TIPS in management of gastric varices.

Since the diagnosis of active bleeding from the gastric varices is endoscopically performed, immediate control of bleeding with an endoscopic procedure is desirable. Whereas TIPS seems to be consuming, the success rate seems to depend on operator skill and the vascular anatomy in each patient. At present, TIPS would be recommended when endoscopic therapy is not successful, or after a single failure of endoscopic treatment. However, the indication should be limited to patients with a higher portal pressure. The operator should keep in mind that β-blocker and TIPS are ineffective in patients with lower portal pressure caused by a major porto-systemic shunt. It should be borne in mind that endoscopic variceal obturation using tissue adhesives such as cyanoacrylate is effective in the management for acute bleeding gastric varices.


Surgical shunting

A recent meta-analysis revealed a significantly better survival and a significantly less frequent shunt failure in patients undergoing surgical shunting compared with TIPS38. However, the problem is that patients included into the studies are limited to those patients with better liver function, classified into Child Pugh Turcotte Classes A or B. Moreover, the majority of patients in those studies had esophageal varices but not gastric varices. The efficacy of surgical shunting for gastric varices has not been evaluated by statistically valid methods. Therefore, the efficacy of surgical shunting for gastric variceal bleeding has not been clearly shown.

Devascularization of the upper stomach with splenectomy

Different from the TIPS or shunt surgery, devascularization of the upper stomach with splenectomy, what is called “Hassab's operation”, has been considered as a feasible procedure for controlling gastric variceal bleeding.39,40 However, Hassab's operation has been shown to confer a higher re-bleeding rate for esophageal varices. In regard to gastric varices, devascularization of the upper stomach with splenectomy has been reported to prevent hemorrhage in a long term follow-up study.40 Even in patients with a poor liver function Child-Pugh-Turcotte score the operation was successfully performed. It is an easier procedure compared to portosystemic shunting surgery, which requires the specific surgical expertise of vascular anastomosis.40 Therefore, it is generally accepted that every surgeon who is an expert in the field of abdominal surgery can perform Hassab's operation, without a need for specific surgical skills in vascular surgery. An additional advantage splenectomy is recovery to the normal range of thrombocyte count from thrombocytopenia, which is caused by hypersplenism following portal hypertension.40,41

However, surgery is limited to patients who can tolerate general anesthesia. A major complication is portal vein thrombosis, but this is easily controlled by postoperative anticoagulation therapy in association with regular ultrasonography to detect the portal thrombosis. As a minimally invasive surgery, a laparoscopic devascularizaion of the upper stomach with splenectomy has been successfully performed.42,43 Splenectomy was not previously recommended in younger patients because of overwhelming postsplenectomy infection (OPSI), a potentially rapidly fatal septicemia. However, surgical technology and vaccination, for example (against pneumococcus), has recently developed to the extent that these problems44 are now largely resolved. The non-re-bleeding rate of 100% over 5-year follow up shows this operation could be the best reliable and promising procedure of a salvage therapy for uncontrolled gastric variceal bleeding.

Balloon-occluded retrograde transvenous obliteration (B-RTO)

Balloon-occluded retrograde transvenous obliteration (B-RTO) have been developed and been established as a superior effective treatment for fundic gastric varices and hepatic encephalopathy18 in Japan. A catheter for B-RTO (6.5 French, Create Medic, Tokyo, Japan) is introduced into the gastro-renal shunt via the right femoral vein. While the gastro-renal shunt of the outflow vessel of the gastric varices is occluded with a balloon, 10 to 20 mL of a 5% solution of ethanolamine olate with iopamidol (EOI) is injected into the gastric varices until their whole length had been visualized (Fig. 4a,b). Gastric varices usually disappear after 2 or 3 months (Fig. 4c).

Figure 4.

(a) Endoscopic image showed tumorous (F3), gastric varices that have a high risk of bleeding. (b) In most cases, after controlling the other out flow routs, a sufficient accumulation of 5% ethanolamine olate with iopamidol (EOI) is obtained in gastric varices. (c) Endoscopic image obtained 3 months after B-RTO revealed eradication of gastric varices.

The long-term effectiveness of B-RTO for the treatment of risky gastric varices has been reported.13–15 In most reports, however, the indication for the B-RTO was prophylactic or elective cases, not acute bleeding. There are few reports about the efficacy of B-RTO for the treatment of patients with gastric variceal bleeding. So far as the authors are aware, there are four reports indicating the effectiveness of B-RTO as a secondary prophylaxis for gastric variceal bleeding (Table 2).15,45,46 According to these reports, the rate of re-bleeding from isolated fundic gastric varices is extremely low by B-RTO compared with that by a previous endoscopic treatment with cyanoacryl, over the longer term. In regard to prophylactic treatment with B-RTO for the large gastric varices without a history of bleeding, most studies show this can be successfully performed, and the results reveal a 100% non-bleeding rate over long term follow up.13–15,47

The psychological advantage of B-RTO to the operator is that it is a non- stressful procedure because no needle puncture to the gastric varices is required. Gastric varices can be treated using B-RTO by an interventional radiology (IVR) technique. When control of gastric variceal bleeding from the puncture site or the ligated site has failed during the endoscopy, the bleeding point may increase in size, resulting in a life threatening hemorrhage.

Another merit of B-RTO is the recovery of liver function, to increased portal flow.48,49 Although renal impairment due to vascular escape of a large amount of sclerosant (ethanolamine olate), and worsening of ascites have been reported, these complications are far less likely with the development of several new techniques.15,50,51 In order to clarify the overall efficacy of B-RTO, a prospective controlled randomized study compared with the other treatments is necessary.


Although there have been an increasing number of reports about gastric variceal bleeding in the last decade, controversy remains about the best approaches to their management and effects over the long term. Thus, it is mainly due to a lack of understanding of anatomical vascular structure and hemodynamics of gastric varices. While a Hassab's operation, B-RTO and obliteration with cyanoacrylate are the most promising among the conventional therapies, combination therapy to completely obliterate the inflow and outflow vessels may lead to better prognosis in the patients with gastric varices. This requires further study.

A better understanding of the hemodynamics and variceal classification based on a statistical evaluation of the risk of bleeding or clinical evidence would be helpful to consider the strategy and to establish the management not only for the gastric varices, but also for the other problems of portal hypertension.