Background

The distal splenorenal shunt (DSRS) was developed by Warren and colleagues in 1967 to achieve selective variceal decompression to prevent recurrent variceal bleeding. Selective variceal decompression combines the benefit of a decompressive shunt to control bleeding with maintenance of portal hypertension and portal flow to the cirrhotic liver to help maintain liver function. The original, classic article on selective variceal decompression described the animal work leading up to the initial clinical data on DSRS (Warren et al, 1967). Proof of concept was shown in the animal studies and in the results of the first six patients in whom DSRS was performed. The unique prior experiences of Warren and Zeppa led to the evolution of this concept: Warren had seen that total shunts control variceal bleeding, but at the cost of liver failure, whereas Zeppa had seen devascularization procedures maintain portal perfusion, but at a cost of significant risk of rebleeding.

Over the next 4 decades, DSRS became the most widely used operation to control variceal bleeding, with a worldwide following (Orozco et al, 2007). The technique continued to evolve, with greater degrees of portal azygos disconnection improving its effectiveness (Henderson et al, 1989). However, in the 1990s, the evolution of improved endoscopic therapy with banding (see Chapter 75B), introduction of transjugular intrahepatic portosystemic shunting (TIPS; see Chapter 76E), and coming of age of liver transplantation (see Chapter 97A, Chapter 97B, Chapter 97C, Chapter 97D, Chapter 97E ) left few indications for shunt surgery for variceal bleeding (see Chapter 76A).

Indications for DSRS

The indications for DSRS have markedly diminished in the last decade, as other options have improved, and the available expertise of surgeons facile at performing DSRS has diminished. That said, DSRS remains a good option for the following patients:

Patient Evaluation

Patients being considered for DSRS require full evaluation to define 1) gastroesophageal varices as the source of bleeding, 2) the cause of the portal hypertension, 3) status of liver function and damage, and 4) vascular anatomy. Evaluation for possible DSRS is only appropriate in patients on propranolol with documented recurrent variceal bleeding or persistent high-risk varices and who have had adequate endoscopic management. Accurate endoscopic evaluation is therefore the first step. In this context, these patients have had upper diagnostic and therapeutic endoscopy at the time of an acute variceal bleed, and follow-up endoscopy is aimed at therapeutic variceal obliteration. Prior to considering patients for DSRS, endoscopy should confirm persistent varices or portal hypertensive gastropathy.

The cause of portal hypertension for most patients in the United States and Europe with variceal bleeding is cirrhosis. In these patients, the first question to be addressed is whether the patient now or in the future is a candidate for liver transplantation. If the patient is likely to need a transplant in the next 1 to 2 years, surgical shunt is not indicated. If the patient has good liver function and is unlikely to need a transplant in the near future, DSRS may be a good option. Other etiologies of portal hypertension—portal vein thrombosis or noncirrhotic portal fibrosis, in which patients have normal liver function—need to be defined early in the evaluation, because the long-term prognosis depends on best prevention of variceal bleeding. Such patients may be candidates for DSRS earlier in their management.

Liver function is assessed from clinical findings and laboratory studies. Jaundice, ascites, and encephalopathy are the three clinical signs and symptoms of advanced liver disease and indicate that patients are not candidates for surgical decompression. Laboratory measurements of serum bilirubin, albumin, serum creatinine, and prothrombin time prolongation are the most useful studies to assess the status of cirrhosis. Combining the clinical and laboratory parameters to discern the Child-Turcotte-Pugh (CTP) class and/or Model for End-Stage Liver Disease (MELD) score gives an objective assessment of risk (Table 76C.1 and Box 76C.1).

Imaging for portal venous anatomy helps with diagnosis and treatment planning and should be performed as part of the evaluation of patients after their initial variceal hemorrhage. Doppler ultrasound can usually visualize the splenic, superior mesenteric, and portal veins as well as the hepatic veins for liver outflow. Early identification of thrombosis of the portal and/or splenic veins may alter the whole approach in the management of the patient. Ultrasound also should examine liver morphology, particularly for evidence of focal lesions suggestive of hepatocellular carcinoma. Computed tomography (CT) or magnetic resonance (MR) vascular imaging and morphologic assessment can augment ultrasound studies; usually these are sufficient to make clinical decisions.

Arteriography may sometimes be required for final definition of the veins before surgical intervention. The components of this evaluation can include the following:

This combination of venous and arterial angiographic study can give information not available with ultrasound, CT, or magnetic resonance imaging (MRI).

Candidates for DSRS are CTP class A or B 7 or 8 patients (i.e., their CTP score is 7 or 8 points), usually without ascites, who are considered to have stable liver disease and are unlikely to need liver transplantation in the next 5 years.

Technique for Distal Splenorenal Shunt

Attention to detail in perioperative and postoperative monitoring and management of a patient undergoing DSRS improves outcome. Monitoring requires an arterial line, a central venous catheter, and a urinary catheter. Good venous access should be obtained in the event that rapid transfusion is required.

Blood and blood products should be available: packed red blood cells, fresh frozen plasma, and platelets may be required if there is major intraoperative blood loss or perioperative hemorrhage. A cell saver can be used in higher risk patients and may reduce blood bank requirements.

The patient is positioned on the operating table with the left arm at the side and the left side slightly elevated. Hyperextending the operating table to open the angle between the left lower ribs and iliac crest aids in exposure and access to the tail of the pancreas. The primary operating surgeon is on the patient’s right. The operation is more easily completed when the first assistant (to the left of the table) also has experience with the procedure.

The steps in DSRS are illustrated. The preferred incision is a long left subcostal incision, extended across the right rectus muscle (Fig. 76C.1). Coagulating diathermy should be used extensively in patients with portal hypertension to achieve hemostasis in dividing tissues. If present, ascites should be aspirated and cultured, and a liver biopsy specimen should be obtained to document the status of the liver at the time of the procedure.

FIGURE 76C.1 The left subcostal incision extends across the midline and right rectus muscle.

Step 1: Exposure

The initial steps of this procedure provide access to the vessels. The pancreas is exposed by opening the lesser sac (Fig. 76C.2) and interrupting the gastroepiploic arcade from the pylorus to the first short gastric vessels. Exposure is greatly enhanced by taking down the splenic flexure of the colon from the lower pole of the spleen. This is done most easily by initially mobilizing the left side of the splenic flexure of the colon, then identifying the plane from there back into the lesser sac at the tail of the pancreas. The splenocolic ligament is divided, giving excellent exposure to the inferior margin of the pancreas from the mesenteric vessels to the splenic hilus. The pancreas is then fully mobilized along its inferior border over its entire length, so that it is turned completely on its side; this is achieved with a combination of Bovie coagulation and ligature of tissues (Fig. 76C.3). The inferior mesenteric vein is the first venous landmark identified: in 50% of patients, this vein enters the splenic vein; in the other 50%, it enters the superior mesenteric vein. The inferior mesenteric vein should be traced up to these vessels and then divided to aid further exposure.

FIGURE 76C.2 Access to the pancreas is through the gastrocolic omentum. Dividing the splenocolic ligament improves exposure and divides collateral varices to the splenic hilus.

FIGURE 76C.3 The pancreas is mobilized from the superior mesenteric vein to the spleen by dividing the posterior parietal peritoneum along its inferior margin.

Step 2: Isolation of the Splenic Vein

The superior mesenteric and splenic vein junction is identified, initially on its posterior surface; this is a safe plane for initial dissection. The splenic vein should be isolated along its inferior and posterior aspect with dissection occuring right on the vessel (Fig. 76C.4). At this point, the splenic, superior mesenteric, and posterior aspects of the portal vein all should be cleared (Fig. 76C.5).

FIGURE 76C.4 The splenic vein is dissected initially along its inferior and posterior edge. This dissection must be directly on the vein.

FIGURE 76C.5 The posterior plane behind the splenic vein should be developed first, and the plane behind the confluence of the splenic vein, superior mesenteric vein, and portal vein is opened.

When the posterior plane is free, attention turns to the anterior and more difficult plane of dissection on the splenic vein. Tributaries rarely enter the anterior surface of the portal vein; so this plane between the neck of the pancreas and the portal vein should be opened first, then the pancreas should be cautiously separated and dissected from the anterior and superior surfaces of the splenic vein. The key is to dissect the pancreas off the splenic vein rather than the other way around. This requires a delicate touch and is best achieved by spreading the tissues gently in the line of the tributaries and at right angles to the splenic vein. This isolates the tributaries, and when identified, a fine right-angle clamp is passed around them, with a 3-0 tie on the vein side and a clip on the pancreatic side (Fig. 76C.6). As much of the splenic vein as possible should be dissected in this manner before dividing the splenic vein at the superior mesenteric vein junction.

FIGURE 76C.6 Isolation of tributaries from the pancreas into the splenic vein requires their dissection at right angles to the splenic vein. These vessels have thin walls and require gentle dissection.

Step 4: Division of the Splenic Vein

This is the step that lines up the splenic and left renal veins for anastomosis. The superior mesenteric end of the splenic vein is ligated with a 2-0 silk, and a large clip is placed flush with the vein behind that tie; this has led to the lowest incidence of thrombus within the portal vein at postoperative angiography. At this point the surgeon must judge whether enough splenic vein has been dissected free of the pancreas to allow it to come down to the left renal vein without kinking or tension. In the event that more dissection of the splenic vein is needed, this can now be performed more easily, because the vein can be manipulated downward, as shown in Figure 76C.7. The disadvantage of this maneuver is that the pressure in the splenic vein has increased with ligation, which leads to greater risk of tearing of the small tributaries.

FIGURE 76C.7 Division of the splenic vein allows the angle between the pancreas and vein to be opened. This improves exposure for dissection of small tributaries to the splenic vein at the cost of increased pressure in the vein.

In patients with alcoholic cirrhosis, data have shown that complete dissection of the splenic vein from the pancreas (splenopancreatic disconnection) is advantageous in leading to improved long-term maintenance of portal perfusion (Henderson et al, 1989). If this modification of DSRS is used, the complete dissection should be done at this time. Data do not support the need for entire splenopancreatic disconnection in patients with nonalcoholic liver disease.

Step 5: The Anastomosis

This is performed without tension, and usually the splenic vein needs to be trimmed so that when the clamps are removed, the vein is not redundant and lies without kinking. This alignment can be difficult to judge, particularly if the two veins are overlying each other. The position of the clamps and trimming of the splenic vein are shown in Figures 76C.8 and 76C.9. The left renal vein is opened over sufficient length without removing an ellipse. The posterior row of the anastomosis is completed with a running suture, with stay sutures placed at either end, and the suture is run on the inside; the anterior row is usually interrupted to avoid risk of a purse-string effect. The completed anastomosis is shown in Figure 76C.10.

FIGURE 76C.8 Sufficient splenic vein must be dissected out of the pancreas for the vein to come down to the left renal vein without tension or kinking. An adequately dissected splenic vein usually needs to be trimmed.

FIGURE 76C.9 The posterior anastomosis is made with a running suture. The clamps must be held for a tension-free anastomosis.

FIGURE 76C.10 The anterior row of the anastomosis is completed with interrupted sutures.

Patient Management

Results

The goals of DSRS are to prevent recurrent variceal bleeding and, by maintaining portal perfusion, not to accelerate the course of the underlying liver disease. In the 1990s and 2000s, several series documented that these goals can be achieved. Most of these series are in good risk patients (CTP class A and B), and results have been excellent.

Bleeding control is equal to or greater than 90% (Henderson et al, 1992; Herman, 1995; Orozco et al, 1997; Rikkers, 1998; Livingstone et al, 2006; Elwood et al, 2006). The highest risk time for rebleeding after DSRS is in the first month (Richards et al, 1987). The technical failure rate should be less than 5%, and it should be defined before hospital discharge with direct shunt catheterization as described earlier. If a technical problem occurs with the anastomosis at that time, surgical reexploration should be undertaken; if it cannot be corrected, splenectomy and devascularization are indicated. Late thrombosis of DSRS is unusual, although patients with portal vein thrombosis undergoing DSRS have a higher rate of shunt stenosis requiring balloon dilation (Warren et al, 1988). Bleeding control is significantly better than that achieved with endoscopic therapy, which is approximately 25% to 30%.

Portal perfusion is maintained at late follow-up in 90% of patients with nonalcoholic liver disease. In patients with alcoholic cirrhosis, loss of portal perfusion occurs in 50% of patients (Henderson et al, 1985). Zeppa and colleagues (1978) documented poorer survival after DSRS in alcoholic patients than in nonalcoholic patients but never proved cause and effect. The mortality rate can be improved in the latter population by complete splenopancreatic disconnection (Henderson et al, 1989). In this modification, the entire splenic vein is dissected free of the pancreas before anastomosis. Splenopancreatic disconnection also has been documented to improve survival in nonalcoholic patients.

Hepatic function is maintained after DSRS in CTP class A and B patients as shown in the series in the 1990s. The rate of transplantation in these patients is currently less than 20%, but longer term follow-up may show an increase in this number at 10 to 20 years. Survival after DSRS is dictated by the severity of the underlying liver disease. The series from the 1990s and 2000s cited above documented 5-year survival of 80% to 90% for CTP class A and B patients.

Two major questions have arisen over the past 5 years: How often is variceal decompression needed, given improved pharmacologic and endoscopic treatment? (D’Amico et al, 2002) And is a surgical shunt ever needed in the era of TIPS? Current data show refractory bleeding through endoscopic and pharmacologic therapy in 15% to 20% of patients; most of these patients have poor liver function, and their only reasonable decompression option is liver transplantation.

The relative efficacy of surgical decompression versus TIPS has been studied in two randomized trials. One study assessed the 8-mm portacaval shunt versus TIPS, and although the surgical limb had better bleeding control, overall outcome was not significantly different (Rosemurgy et al, 2005). In a randomized trial (Henderson et al, 2006) of DSRS versus TIPS in CTP class A and B patients, the rebleeding rates were not significantly different, 6% and 11%, but 83% of TIPS patients required reintervention and dilation to achieve this outcome. Survival and encephalopathy rates were not significantly different in this study. Cost-effectiveness analysis showed a slight advantage to TIPS in this trial (Boyer et al, 2008), and it can be concluded that TIPS is equally effective as DSRS.

In summary, DSRS is an excellent operation for controlling refractory variceal bleeding in patients with portal hypertension and well-preserved liver function. Its current role is, however, minimal given the evolution of other therapeutic options, as succinctly summarized by Hector Orozco (Orozco et al, 2007): “What a pity. There have been many ideas to resolve a problem, many years of work, many patients treated, and one surgical solution that was approaching the ideal: low morbidity, low mortality, low recurrence of the hemorrhagic episodes, and long survival. Despite all of this, we threw it away because we did not know how, when, and to whom we might give it.”