Management of Isolated Gastric Varices by Devascularization and Proximal Gastrectomy in Cirrhotic Patients

Twenty patients with liver cirrhosis were treated by surgery for bleeding from isolated gastric varices. The presence of tortuous and engorged gastric veins connecting with a large splenorenal shunt was demonstrated by transhepatic portography in all patients. The surgical procedures consisted of splenectomy, proximal gastrectomy, paragastric devascularization, and ligation of the splenorenal shunt. Sixteen patients survived the surgery. Four deaths were caused by emergency operation for uncontrollable hemorrhage in extremely poor risk patients. Of the 16 survivors, 15 had been followed wth endoscopy and portography for a mean period of 42 months. The other one died of hepatocellular carcinoma three years after surgery. There was no bleeding episode during the period of follow-up in these patients. Recurrent esophageal varices of mild degree were documented by endoscopy and portography in three patients. Portography demonstrated that several newly formed retroperitoneal veins arising from the junction of the portal and superior mesenteric veins joined to form recurrent varices in these three patients. There was no significant change of the mean portal venous pressure before and after surgery. Our data reveals that elective surgery may provide satisfactory results in patients with isolated gastric varices. Transhepatic portography is the method of choice in radiologic investigation for prominent gastric varices.


INTRODUCTION
In patients with liver cirrhosis the development of esophagogastric varices has been well recognized. However, prominent gastric varices without esophageal varices which are termed isolated gastric varices (IGV) or fundal varices could be found in a small proportion of patients1-3. In such patients with IGV, portographic study usually delineates tortuous and engorged gastric veins with a large spontaneous splenorenal or gastrorenal shunt draining into inferior vena cava4. IGV represent a specific variety of collateral veins in cirrhotic patients. Bleeding from gastric varices is difficult to control by balloon tamponade, especially varices locating in the fundic area5. The experience and success of endoscopic sclerotherapy in achieving hemostasis of IGV is also limited1'3'6. Transhepatic occlusion of IGV is contraindicated because of the potential hazard of inadvertent embolization of adjacent organs via the splenorenal shunt. Surgery becomes necessary in most situations to control hemorrhage from IGV. Herein, we present our experience in dealing with the results of surgical management of IGV with special reference to the study of portography before and after surgery.

Patients
During the period between March 1983 and April 1990, 197 patients with portal hypertension were evaluated by transhepatic portography. Of them, isolated gastric varices were observed in 20(10%), combined esophageal and gastric varices in 76(39%), and esophageal varices alone in 101(51%) based on portographic findings. Of the 20 patients, there were 15 males and 5 females with ages ranging from 37 to 65 years. All the 20 patients had at least one episode of bleeding from gastric varices. All of them had portal hypertension and posthepatitic liver cirrhosis documented by clinical and histological examinations. Preoperative liver function was categorized according to Child's method and, grade A was found in 5, grade B in 11, and grade C in 4 patients. On admission, 6 of the 20 patients with IGV presented clinical features of hepatic encephalopathy and elevated serum ammonia. The degree of encephalopathy became more aggravated when the patient continued to bleed.
Every patient with bleeding gastric varices was given vasopressin soon after admission to our unit. Balloon tamponade was then applied if the patient continued to bleed for one hour after the introduction of vasopressin infusion. Endoscopic sclerotherapy and percutaneous transhepatic embolization were not tried because of the fear of systemic embolization through the large splenorenal shunt. Fifteen patients were able to control bleeding with these treatments. They got recovery from acute bleeding and had an elective surgery later. Unfortunately, five required an emergency procedure due to failure of our medical measures and uncontrollable hemorrhage.

Diagnostic Procedures
Other specific studies included gastroduodenoscopy, transhepatic portography and pressure measurement. All patients had at least one episode of bleeding trom gastric varices. The diagnosis of gastric varices without concomitant esophageal varices was first made with a gastroduodenoscopy (GIF P 10, Olympus Co., Tokyo, Japan). Subsequently the diagnosis was confirmed by detailed angiographic study. The technique for transhepatic portography has been described in more detail in the literature. In all patients a selective splenic venogram was initially obtained to evaluate the patency of the splenic vein, the presence of gastric varices and the flow direction of the splenorenal shunt. Occasionally selective catheterization of the other portal branch and collateral veins was also performed. The portal pressure was measured when the tip of the catheter was in the main portal trunk. The puncture site in the midaxillary line was the reference point and the pressure was measured with a water manometer.

Surgical Therapy
The surgical procedures for IGV included splenectomy, paragastric devascularization, proximal gastrectomy and pyloroplasty. After completion of splenectomy, proximal halves of the lesser and greater curvatures were completely devascularized. The splenorenal shunt was carefully identified, and then ligated and divided.
The stomach was resected 5 cm distal to the esophagogastric junction and the wound was closed with a stapler. A vertical gastrotomy was made to introduce the EEA surgical stapling instrument (Autosuture, USA) for reconstruction of the esophagus and stomach (Figure 1). The same surgical procedures were employed in all patients who were diagnosed to have IGV in this study.

Follow-up Studies
The patients who have survived for more than one year after surgery had regular clinical follow-up including assessment of recurrent hemorrhage, encephalopathy and ascites etc. In addition, to investigate the effect of surgery on variceal recurrence and the change of portal venous pressure, postoperative endoscopy and portography were performed at least one year following surgery. At the follow-up portography, catheterization of the splenic vein was not feasible because it was ligated at the operation. Only superior mesenteric and portal venograms were taken. If there were newly formed collaterals demonstrated, they were selectively cannulated to assess the communication with esophagogastric varices.

Clinical Outcomes
Of the 15 patients who underwent elective surgery, there was no hospital mortality or morbidity related to anastomotic leakage. However, postoperative complications were encountered in four patients. Two patients had left subphrenic abscess due to extravasation of the pancreatic juice which was caused by pancreatic tail injury during splenectomy. They were cured within one month by intraperitoneal drainage. Two patients complained of dysphagia caused by mechanical stapling. The episode improved spontaneously without dilatation 6 months after surgery. Of the five patients who underwent emergency operation, two died 3 and 5 days after surgery, respectively. The causes of death were prolonged shock in one and respiratory failure in the other. Two developed progressive jaundice and died of hepatic failure within one month after surgery. All four patients were of Child's class C. In 4 of the 6 patients with hepatic encephalopathy, the episode did not attack again postoperatively. The other two who underwent emergency surgery died of progressive hepatic insufficiency finally. Of the 16 patients survived surgery, one died of hepatocellular carcinoma 3 years after surgery. The remaining 15 patients had been followed with endoscopy and portography. The mean follow-up period was 42 months ranging from 14 to 72 months, during which no patient had recurrent bleeding (Table I).
Portographic Findings Transhepatic portography was successfully performed in all patients without any serious complications. When the tip of the catheter was positioned in the splenic vein near the hilum, injection of contrast medium demonstrated that the coronary and/or short gastric veins, arising from the splenic vein and forming varices in the proximal stomach, drained into the left renal vein and then into the inferior vena cava ( Figure 2). We believe that such a communicating complex represents a characteristic angiographic appearance. A single and large venous channel, shunting substantial amount of blood from gastric varices into the left renal vein could be demonstrated in every one of our patients. In postoperative portography, neither the coronary vein nor other collaterals leading to the esophagogastric region were visualized in 12 patients (Figure 3). Three patients developed newly formed collaterals which originated from the junction of the portal and superior mesenteric veins. These new collaterals supplied several retroperitoneal veins which joined together and then drained into the esophagus (Figure 4). Portal Pressure Pressure studies confirmed the presence of portal hytal venous pressure before surgery in 20 patients was 34.3 + 5.5 cm H20. In the 15 patients who had been followed up for portography after surgery, the mean portal venous pressure was 33.3 + 4.7 cm H20. There was no significant difference between them.  with a patent splenic vein may occur because of direct anastomoses between the gastric and retroperitoneal veins. Samuel 1 reported that dilatation of the fundal veins may antedate esophageal varices. In our opinion, preexisting venous channels confined to the fundic area may become dilated and lead to the development of gastric varices in response to portal hypertension. Halvorsen and Myking 11 found that the development of portosystemic shunts was not spontaneous, but rather represented dilatation of already existing vascular channels. When dilated venous channels (gastric varices) were eradicated by gastric resection, recurrent varices would not appear in the stomach but in the esophagus as demonstrated by postoperative portography in three of our patients. Our observation is in accord with that of the animal study1. Gastroduodenoscopy has been helpful in detecting gastric vadces in most patients1-4. However, several reports revealed that IGV have been mistaken for neoplasms both radiologically and endoscopically 1'13. We had such experience in three patients. Moreover, one should be careful in taking biopsy from a suspected submucosal lesion in cardiac and fundic regions in cirrhotic patients since it may cause massive and uncontrollable bleeding. In the present study, definite diagnosis was confirmed by transhepatic portography in all patients including three who had been misdiagnosed by endoscopy. Moreover, newly formed collaterals feeding recurrent varices could also be clearly demonstrated by this technique. We recommend that transhepatic portography and endoscopy should be used together for identification of IGV.
In this series the portal pressure is strikingly high in spite of the presence of a large splenorenal shunt. Furthermore, every patient had episodes of variceal bleeding. Our data are in accord with those of Lam and Aseni 4'5 that almost all cirrhotic patients with a large spontaneous portosystemic shunt showed high portal pressure and any protective effect on bleeding by the shunt is unpredictable.
The treatment of choice for the management of hemorrhage from IGV is controversial. Balloon tamponade with the Linton-Nachlas tube could achieve hemostasis with limited success5. The potential risk of emboli migration to the vital organs prevent the widespread use of percutaneous transhepatic embolization for IGV. Previous studies have shown that poor results of endoscopic sclerotherapy in treating gastric varices1'3. However, using improved sclerosing technique and agent, Soehendra et al. 6 have been able to successfully treat bleeding gastric varices and the results are encouraging. Previous reports 3'7-9 revealed that surgical treatments of IGV only limited to sporadic cases and most had inadequate follow-up. Surgical treatments had included ligation of the varices and two-thirds gastrectomy17, splenectomy alone3, esophagogastrectomy and splenectomy13, proximal gastrectomyTM, and selective shunt surgery9.
The rationale of our procedure in treating IGV is to interrupt directly all the varicose network. Resection of a segment of proximal stomach may reduce the immediate recannalization of venous collaterals and the extensive devascularization is to ablate as much as possible the extrinsic collaterals to the stomach. The purpose of splenectomy includes abolishing hypersplenism and eradicating the short gastric veins. Ligation of splenorenal shunt may prevent new development of retroperitoneal collaterals. Pyloroplasty may avoid gastric stasis due to the inevitable vagotomy during the procedure of extensive devascularization.
The late results in this small group presented here seem acceptable as compared to other types of direct surgical intervention2. The mortality was contributed mostly by emergency and poor-risk conditions. The surgical technique should not be recommended in emergency cases, since the management of acute bleeding from IGV is often difficult. Follow-up portography and endoscopy demonstrated that mild recurrent esophageal varices developed in three of 15 patients who survived surgery. These results also suggest that elective surgery may be the most definite approach for the management of IGV. Endoscopy and transhepatic portography provides the most direct methods for access of esophagogastric varices.