Location of portosystemic shunting

Published on 10/04/2015 by admin

Filed under Surgery

Last modified 10/04/2015

Print this page

rate 1 star rate 2 star rate 3 star rate 4 star rate 5 star
Your rating: none, Average: 0 (0 votes)

This article have been viewed 2541 times

Chapter 76A Location of portosystemic shunting


Esophageal varices develop in patients with portal hypertension, most commonly secondary to hepatic cirrhosis (see Chapter 70A, Chapter 70B ). They occur most frequently in the distal esophagus, although they may be accompanied by gastric varices. Rupture of varices is associated with massive upper gastrointestinal (GI) bleeding with an attendant high mortality rate. Therapy aimed at the prevention and treatment of bleeding varices has included pharmacologic, endoscopic, radiologic, and surgical strategies. All these therapies have evolved technically, and increasing clinical experience has resulted in more accurate definition of the role of each treatment modality. This chapter discusses the appropriate role of surgical shunts for the management of bleeding esophageal varices. An understanding of the role of surgical therapy also requires an understanding of the context in which it is applied, however. The natural history of bleeding esophageal varices is discussed first, followed by a description of the roles of alternative therapies. In current medical practice, it is most appropriate to apply surgical shunts within the context of medical (see Chapter 75A) and endoscopic management (see Chapter 75B), transjugular intrahepatic portosystemic shunts (TIPS; see Chapter 76E), and liver transplantation (see Chapter 97A, Chapter 97B, Chapter 97C, Chapter 97D, Chapter 97E ). Many patients are treated sequentially with more than one modality, and algorithms are presented to help establish the appropriate clinical context for surgical shunt therapy.

Natural History of Esophageal Varices

Esophageal varices may produce massive upper GI bleeding that is difficult to control. Not all varices bleed, and not all patients with cirrhosis or portal hypertension will have esophageal varices develop. Clinical studies have sometimes included control groups without medical intervention, and analysis of these trials has helped define the natural history of esophageal varices. In one series, 46% of 819 patients with biopsy or clinical evidence of cirrhosis and no history of bleeding had esophageal varices by endoscopy (PROVA Study Group, 1991).

Over time, varices may appear, disappear, or change in size depending on alterations in patient physiology. A study of 84 patients with cirrhosis without previous bleeding who were monitored by serial endoscopy over 2 years showed that 31% of patients without varices progressed to large varices over 2 years, whereas in 70% of patients with small varices, the varices enlarged after 2 years (Cales et al, 1990). Dagradi (1972) studied the influence of alcohol on varices in patients with cirrhosis and found that variceal length increased in 65% of patients with cirrhosis who continued to consume alcohol, but it decreased in 80% of patients with cirrhosis who abstained from alcohol. Baker and colleagues (1959) reported that varices regress in 25%, disappear in 32%, and progress in 21% of patients with cirrhosis whose varices are monitored by endoscopy.

Most bleeding episodes in long-term studies occur during the first 1 to 2 years after identification of varices (Baker et al, 1959; Groszmann et al, 1990; Siringo et al, 1994; Triger et al, 1991). Average mortality rates after bleeding from esophageal varices are 23% at 1 year, 34% at 2 years, and 58% at 3 years. Approximately one third of deaths in patients with known esophageal varices are attributable to upper GI bleeding; a larger proportion die as a result of liver failure. The mortality rate directly attributable to variceal hemorrhage is 10% to 17% for cirrhotic patients (Baker et al, 1959; Sauerbruch et al, 1988; Triger et al, 1991). In patients with varices, upper GI bleeding is attributable to variceal hemorrhage in roughly two thirds of patients (Gebhard, 1998). Clinical parameters associated with increased risk of hemorrhage and death from esophageal varices include large varices, those with cherry-red spots (Dagradi, 1972), concurrent gastric varices (Kleber et al, 1991), Child-Turcotte-Pugh (CTP) classification, continued alcohol use (Dagradi, 1972), and infection (Goulis et al, 1999). Death correlates more closely with CTP classification (Merkel et al, 1989) than with any other parameter studied.

Rebleeding and mortality rates markedly increase after varices bleed. Studies have reported rebleeding rates to be 30% within 6 weeks of an initial variceal hemorrhage (Copenhagen Esophageal Varices Sclerotherapy Project [CVESP], 1984; Graham & Smith, 1981) and 60% to 75% within 1 year (Baker et al, 1959; Graham & Smith, 1981). Esophageal varices are the cause of bleeding in approximately 16% of hospital admissions for upper GI bleeding (de Franchis et al, 1991). Mortality rates from all causes within 1 year of initial hemorrhage have been estimated at 40% to 66% (Burroughs et al, 1989; CEVSP, 1984; Graham & Smith, 1981; Le Moine et al, 1992). The risk of dying increases as the interval between initial and second hemorrhage decreases (Gebhard, 1998). If patients survive for more than 12 weeks after a variceal hemorrhage, the risk of rebleeding or dying returns to that of patients who have never bled (Gebhard, 1998).

Pharmacologic Management of Portal Hypertension (See Chapter 75A, Chapter 75B, Chapter 75C )


β-Blocker therapy has been studied to test its efficacy in preventing primary variceal hemorrhage in patients with known varices. Nadolol is a nonselective β-blocker, meaning it blocks both β1 and β2 receptors; patients given nadolol were compared with untreated control subjects. Nadolol reduced the incidence of bleeding from 35% ± 3% to 12% ± 3%, and the incidence of fatal bleeds was reduced from 18% ± 3% to 10% ± 2%. There was no difference in overall mortality rate (Poynard et al, 1991). This study is used to support the use of prophylactic β-blockade to prevent a first variceal hemorrhage.

Nitrates are vasodilators whose action is mediated by nitric oxide on vascular smooth muscle. Nitroglycerin decreases portal pressure in patients with cirrhosis when high doses are used (Moreau et al, 1989). In animal studies, nitroglycerin lowered portal pressure 13%, and systemic blood pressure decreased 25%. This drug lowers portal pressure less than systemic pressure. Nitrates in combination with β-blockade may offer prophylaxis against an initial variceal bleed.

Randomized, controlled clinical trials comparing nonselective β-blockers (propranolol or nadolol) with no therapy in cirrhotic patients showed that drug treatment effectively reduced the risk of a first variceal hemorrhage (Poynard et al, 1991). The combination of isosorbide mononitrate and β-blockade further reduces portal pressure and has been shown in three studies to effectively reduce the risk of a first variceal bleed compared with β-blockade alone (Garcia-Pagan et al, 1990; Villanueva et al, 1996; Vorobioff et al, 1993). These investigations have noted, however, the difficult problem of compliance, particularly in patients with alcoholism. In addition, fatigue may be a side effect of therapy with β-blockade, and even more seriously, if patients do bleed, their ability to compensate for blood loss by tachycardia is compromised.

Acute Variceal Hemorrhage

The posterior pituitary hormone vasopressin causes splanchnic arteriolar vasoconstriction, reducing portal blood pressure by approximately 15% when given intra-arterially or intravenously (Chojkier et al, 1979; Huet et al, 1987). Intravenous use is preferred for safety and convenience, and the optimal dose of the drug is 0.3 to 0.4 U/min intravenously. As a result of simultaneous vasoconstrictive effects on the cardiac, mesenteric, and cerebral circulations, the complications increase when doses of 0.5 to 0.7 U/min are administered. It is not necessary to taper the dose; the infusion can be stopped when the therapeutic end point is reached. In a controlled study comparing vasopressin with no therapy, approximately half of the patients on vasopressin stopped bleeding, but this result did not differ from control subjects (Chojkier et al, 1979; Fogel et al, 1982).

Nitrogen is often administered concurrently with vasopressin to reduce the systemic vasoconstrictive effects of vasopressin, and it may further reduce portal pressure. Nitroglycerin infusion begins at 40 µg/min and is titrated to a mean arterial blood pressure of 65 to 75 mm Hg (Gimson et al, 1986).

Octreotide reduces bleeding (D’Amico et al, 1995) and enhances the results of sclerotherapy (Besson et al, 1995). Somatostatin and octreotide are endogenous peptides that act by reducing splanchnic, hepatic, and azygos blood flow (Bosch et al, 1981). Their principal advantage over vasopressin is that they do not cause vasoconstriction of the myocardial and cerebral circulations. Somatostatin and octreotide should be administered continuously at 250 µg/h and increased to 500 µg/h if bleeding continues. Preliminary studies showed that octreotide helped arrest acute variceal bleeding in six of six patients (Thulin et al, 1979; Tyden et al, 1978). Randomized, controlled trials comparing somatostatin or octreotide with vasopressin versus no infusion have shown equivocal results, which suggests that vasopressin and somatostatin have similar efficacy (Burroughs, 1996; Burroughs et al, 1990; Imperiale et al, 1995). Neither vasopressin nor somatostatin has been approved by the U.S. Food and Drug Administration (FDA) for treatment of variceal bleeding, although both agents are commonly used for this purpose (Korula, 1998). A prospective, randomized trial showed equivalence of somatostatin and sclerotherapy in the treatment of acute variceal bleeding (Planas et al, 1994).

Prevention of Rebleeding After Initial Control

Propranolol was shown by Lebrec and colleagues (1980, 1981) to reduce rebleeding significantly after acute variceal hemorrhage. This effect may be mediated by a decrease in cardiac output (β1-blockade), increased splanchnic arteriolar resistance (β2-blockade), and consequent decrease in portal blood flow (Lebrec et al, 1982) and collateral blood flow via the azygos venous system (Feu et al, 1993). β-Blockade is not widely used in the United States to prevent rebleeding after an episode of variceal hemorrhage because endoscopic sclerotherapy and ligation are preferred, and β-blockade after acute bleeding has not been shown to reduce mortality rate (Pagliaro et al, 1989). Meta-analysis comparing β-blockade with endoscopic sclerotherapy showed a non–statistically significant decrease in pooled relative risk for bleeding in the sclerotherapy group and no difference in mortality rate between the two groups (D’Amico et al, 1995). A randomized, controlled study showed, however, that isosorbide mononitrate (80 mg/day) in combination with nadolol (80 mg/day) was more effective than sclerotherapy in reducing rebleeding (Villanueva et al, 1996), and complications were less frequent in the group treated with drugs (16% vs. 37%). If this study is confirmed by other investigators, pharmacologic therapy may play a larger role than it currently does in the United States in the prevention of rebleeding.

Endoscopic Therapy of Variceal Hemorrhage


The use of prophylactic sclerotherapy to prevent a first hemorrhage was studied in three meta-analyses (see Chapter 75A, Chapter 75B, Chapter 75C ; Fardy & Laupacis, 1994; Pagliaro et al, 1989; Van Ruiswyk & Byrd, 1992). One study concluded that paravariceal injection with polidocanol decreased mortality rates (Fardy & Laupacis, 1994). The other two reports found that prophylactic sclerotherapy did not reduce bleeding or mortality rate and concluded that sclerotherapy was not indicated in this setting (D’Amico et al, 1995; Pagliaro et al, 1989; Van Ruiswyk & Byrd, 1992). The largest trial of prophylactic sclerotherapy was the Veterans Affairs (VA) cooperative trial. This trial included 281 patients but was prematurely closed because of excess mortality rate in the sclerotherapy group (VA Cooperative Variceal Sclerotherapy Group [CVSG], 1991). Sclerotherapy prevented variceal hemorrhage but substituted bleeding from sclerotherapy-induced ulceration. This study effectively ended the use of prophylactic sclerotherapy in the United States.

Acute Variceal Hemorrhage

When it became apparent that the once predominant therapy for variceal hemorrhage, emergency surgical shunts, were not improving survival but rather substituting death from liver failure for death from bleeding, endoscopic variceal injection was evaluated as a less invasive therapy. In 1980, a prospective randomized trial with 107 patients from King’s College Hospital showed control of bleeding by sclerotherapy in 57% of 51 treated patients compared with 25% of 56 patients treated medically (MacDougall et al, 1982). Two years later, a follow-up study showed improved patient survival with sclerotherapy compared with controls who received blood transfusions, vasopressin, and a Sengstaken-Blakemore tube when necessary (Fig. 76A.1).

When interpreting this and subsequent trials, it is important to understand that the King’s College trial had more nonalcoholic patients than alcoholic patients (60 vs. 47) and had patients with relatively mild liver failure (74 were CTP class A or B, 33 were class C). The more patients in any study of variceal hemorrhage who are alcoholic or who have CTP class C liver disease, the more difficult it is to show a survival advantage of therapy. Death from bleeding in such patients tends to be replaced by death from liver failure (Block & Reichelderfer, 1998). The VA cooperative study showed no reduction of long-term survival when acute hemorrhage was treated with sclerotherapy (CVSG, 1994).

Sclerotherapy has been shown to stop acute variceal hemorrhage effectively (Gregory, 1990; Westaby et al, 1989). Meta-analysis of 20 trials of emergency sclerotherapy versus a variety of alternative therapies supported the superiority of sclerotherapy with its success rate of 71% to 100%; however, the complication rate was high (18%), and 2.7% patients died as a direct result of sclerotherapy (D’Amico et al, 1995).

Endoscopic variceal ligation (EVL) has been developed as an endoscopic alternative to sclerotherapy, potentially lowering the risk of ulceration and perforation of the esophagus. Seven prospective, randomized, controlled trials compared EVL with endoscopic sclerotherapy (Gimson et al, 1993; Hashizume et al, 1993; Hou et al, 1995; Laine et al, 1993; Lo et al, 1995, 1997; Stiegmann et al, 1992). In all studies, EVL and sclerotherapy were equally effective in controlling active bleeding. Complications were significantly lower with EVL in all studies. No esophageal strictures were seen in patients treated with EVL compared with 5% to 33% of patients treated with sclerotherapy. The development of the multiple-band ligating device, which allows banding without repeatedly reinserting the endoscope, has made this modality of endoscopic control of varices much more attractive, such that it has now become the endoscopic therapy of choice (Laine, 1997).

Prevention of Rebleeding

Although sclerotherapy effectively stops acute variceal bleeding, rebleeding remains a problem, and intermediate (2- to 5-year) survival is not improved in many trials. A confounding variable confusing interpretation of the results in many of these trials is continued alcoholism. Alcohol abstinence for 6 months, CTP class, and aspartate aminotransferase level all were independent predictors of survival in the VA trial (CVSG, 1994). Meta-analyses of trials comparing sclerotherapy with pharmacologic management have shown sclerotherapy to prevent rebleeding more effectively and sometimes improve survival (D’Amico et al, 1995; Infante-Rivard et al, 1989). When EVL was compared with sclerotherapy, rebleeding rates were significantly decreased with EVL in three studies (Gimson et al, 1993; Hou et al, 1995; Lo et al, 1995), and mortality rates were significantly lower in three studies (Hou et al, 1995; Lo et al, 1995; Stiegmann et al, 1992). EVL seems to be at least as effective as sclerotherapy in preventing rebleeding.

Transjugular Intrahepatic Portosystemic Shunt (See Chapter 76E)

The development and clinical use of TIPS in the treatment of portal hypertension first occurred in the 1990s, and its use for the treatment of variceal hemorrhage has expanded (LaBerge et al, 1992; McCormick et al, 1994). TIPS is used electively far more often than in the emergency setting. Despite the effectiveness of TIPS in abruptly stopping variceal hemorrhage, overall patient mortality rates remain high (Smith & Graham, 1982). Death often is related to multisystem organ failure, progressive liver failure or sepsis, or disseminated intravascular coagulation, reflecting the use of TIPS in patients with end-stage disease. Complications are usually related to the underlying cirrhosis and associated comorbidities. In addition to relieving variceal hemorrhage, TIPS effectively relieves ascites in these patients (Crenshaw et al, 1996; Martin et al, 1993b) because TIPS is functionally a nonselective, side-to-side portacaval shunt. In contrast to all surgical shunts, TIPS creates a shunt to the suprahepatic inferior vena cava (Fig. 76A.2).

TIPS was developed as a minimally invasive procedure performed by radiologists using fluoroscopic imaging to place a noncompressible stent between the portal vein and hepatic vein. Successful TIPS lowers portal pressure, and the procedure is typically well tolerated even in very ill patients. Complications include encephalopathy secondary to portosystemic shunting, shunt stenosis and occlusion, inability to place a TIPS, and intraperitoneal bleeding if the liver capsule is punctured. Mortality rate at 30 days has been reported at 20%, but half of these deaths were unrelated to the procedure itself (Darcy et al, 1993). In patients with CTP class C cirrhosis, a 30-day mortality rate of 67% has been reported (Martin et al, 1993a, 1993b); in a lower risk population of 100 patients, a 30-day mortality rate of 5.3% was reported (Richter et al, 1994).

Primary patency—that is, patency without radiologic intervention to revise the TIPS—has been reported to be 46% to 85% in the first 3 to 6 months (Richter et al, 1994; Saxon et al, 1993; Sterling & Darcy, 1995) and 27% to 57% at 1 year (Haskal et al, 1994; Malisch et al, 1993; Saxon et al, 1993; Sterling & Darcy, 1995). Primary assisted patency, meaning patency after revision, has been reported to be 85% (Haskal et al, 1994). LaBerge and colleagues (1995) reported shunt stenosis or occlusion in 47% of 90 TIPS patients over a 2-year period.

The newer polytetrafluoroethylene (PTFE)-covered stents are associated with less need for intervention, less shunt dysfunction, and better outcomes (Bureau et al, 2007). If post-TIPS ultrasound shows narrowing or thrombosis of the shunt, patency can be restored by repeat balloon dilation and stenting or by thrombectomy. Color Doppler ultrasound of TIPS is routinely performed at 1- and 6-month intervals after the procedure to evaluate luminal narrowing or increased flow velocity, which would suggest impending thrombosis of the TIPS. The rate of TIPS restenosis or occlusion is higher than the rate of recurrent symptoms because in some patients, occlusion does not produce symptoms. Nevertheless, recurrent variceal hemorrhage occurs in approximately 50% of patients with TIPS stenosis or occlusion. A multicenter trial of TIPS involving 100 patients reported that 16% of patients had rebleeding by 6 months, and five of these were from nonvariceal sources (Coldwell et al, 1995). Similarly, LaBerge and colleagues (1995) reported variceal rebleeding in 32% of 90 patients at 2 years.

Acute Variceal Bleeding and Transjugular Intrahepatic Portosystemic Shunting

TIPS may be used effectively in the control of acute variceal hemorrhage when medical management or endoscopic variceal ligation or both are ineffective. Barton and colleagues (1995) found that TIPS controlled acute variceal bleeding in 91% of patients, whereas Helton and associates (1993) reported control in 17 (74%) of 23 patients. A report by Encarnacíon and colleagues (1995) reported on 65 patients with acute variceal bleeding unresponsive to sclerotherapy or not treated with sclerotherapy because of recurrent massive hemorrhage. Acute bleeding stopped before the TIPS procedure in 26 patients, but not in the other 39 patients. Of the 65 patients with acute bleeding, 64 had successful placement of TIPS, and all these patients stopped bleeding within 3 days. The 30-day survival rate of patients who stopped bleeding before TIPS was 96%, but it was only 69% for patients actively bleeding at the time of TIPS. Survival was also linked to CTP class, with a 30-day survival rate of 91% for class A (n = 2) and class B (n = 32) patients, but survival rate was 71% for class C (n = 31) patients.

When used as primary therapy for acute variceal bleeding, TIPS may reduce treatment failure and mortality rate in high-risk patients. Monescillo and colleagues (2004) reported that in patients defined as high risk by a hepatic venous pressure gradient (HVPG) greater than 20 mm Hg randomized to treatment with TIPS (n = 26) versus no TIPS (n = 26), the no-TIPS group required more transfusions (P = .002), needed more intensive care unit care, had more treatment failures, and had poorer survival (P < .05). The no-TIPS group was treated with β-blockers, variceal banding, or sclerotherapy.

Prevention of Rebleeding by Transjugular Intrahepatic Portosystemic Shunting

TIPS has been used most frequently to prevent recurrent variceal hemorrhage. The results of four large series are shown in Table 76A.1 (Henderson et al, 1998). Rebleeding rates are similar in these series and are approximately 25% at 1 year. Thirty-day mortality rates were 14% to 16% except for the Rössle series (Rössle et al, 1994). Most deaths within 30 days were due to multisystem organ failure, whereas most later deaths were attributable to progressive liver failure. Because TIPS is a nonselective shunt, encephalopathy rates were relatively high at 25%, although in most patients this was not debilitating because it is usually controllable with lactulose, neomycin or rifaximin, and a low-protein, low-ammonia diet.

Most episodes of rebleeding after TIPS were related to stenosis or thrombosis of the shunt. In addition, many asymptomatic patients had shunt stenosis or thrombosis detected by ultrasound. Primary patency was 40% to 67% at 1 year, which improved to 79% to 88% with revision of stenotic stents (assisted primary patency). Secondary patency, which is patency after TIPS thrombectomy or revision, was 95% to 100% at 1 year (Coldwell et al, 1995; Fillmore et al, 1996; LaBerge et al, 1995; Rössle et al, 1994).

In the short time since its introduction, TIPS has had a dramatic impact on the treatment of variceal hemorrhage. In addition to its use in preventing variceal hemorrhage, TIPS has the added benefit of often improving overall liver function, as measured by CTP status, and of effectively bridging patients to liver transplantation (Abouljoud et al, 1995; Menegaux et al, 1994; Millis et al, 1995; Odorico, 1998; Suc et al, 1995). Despite the suggestion that TIPS may reduce operative time and blood loss during liver transplantation, data are not yet available to support this contention. Nevertheless, it has been shown that the TIPS procedure effectively prevents rebleeding (D’Amico et al, 1995; Ring et al, 1992).

TIPS has been compared with endoscopic therapy for the long-term prevention of recurrent bleeding. In a meta-analysis of 11 randomized trials, fewer patients rebled after TIPS (19%) than after endoscopic therapy (47%), encephalopathy was more common after TIPS (34%), and TIPS dysfunction developed in 50% of patients overall (Papatheodoridis et al, 1999).

A major recent change in therapy recommendations is that in creating a TIPS, use of expanded PTFE (ePTFE)-covered stents is now preferred. The basis for this recommendation is the decreased need for shunt intervention and a suggestion of better outcomes with covered stents rather than bare ones (Angermayr, et al, 2003). Although TIPS increases the risk of hepatic encephalopathy, prophylactic use of nonabsorbable disaccharides or antibiotics does not reduce this risk and is not recommended (Riggio et al, 2005).

TIPS also has been compared with shunt surgery. The distal splenorenal shunt showed lower rates of rebleeding, encephalopathy, and shunt thrombosis than TIPS, but ascites was less common after TIPS (Khaitiyar et al, 2000). A multi-institutional, randomized trial compared TIPS with the distal splenorenal shunt in CTP class A and B cirrhotic patients (Henderson et al, 2004). Initial analysis of the results showed no significant differences between TIPS and the distal splenorenal shunt in variceal rebleeding, shunt occlusion, and survival. However, of the TIPS patients, 80% required reintervention to maintain shunt patency, and close surveillance was required.

TIPS also has been compared with the small-diameter interposition shunt. In a controlled trial, shunt occlusion, death from hepatic failure, and the need for liver transplantation all were significantly more common after TIPS (Rosemurgy et al, 2000). At the 10-year follow-up report, the small-diameter interposition shunts continued to perform better in terms of shunt occlusion, and survival was also superior in CTP class A and B patients and in those with Model for End-Stage Liver Disease (MELD) scores less than 13 (Rosemurgy et al, 2000). Indications for TIPS supported by current data include 1) continued variceal hemorrhage after sclerotherapy or banding, 2) prevention of rebleeding or treatment of ascites in patients awaiting liver transplantation, and 3) prevention of rebleeding in patients who are not candidates for a surgical shunt or liver transplantation because of expected short survival.

Treatment of Ascites with Transjugular Intrahepatic Portosystemic Shunting

TIPS has been used effectively to relieve ascites in patients refractory to pharmacologic therapy with diuretics. Salerno and associates (2004) reported a multicenter, randomized, controlled trial comparing TIPS (n = 33) with paracentesis plus albumin (n = 33) in patients with CTP class B and C cirrhosis. Survival without liver transplantation was superior in patients treated with TIPS (P = .021; Fig. 76A.3). By multivariate analysis, a higher MELD score and paracentesis independently predicted death. Treatment failure was more common in patients treated with paracentesis, although encephalopathy occurred more commonly in patients receiving TIPS (Salerno et al, 2004).

Surgical Shunts for Bleeding Esophageal Varices (See Chapters 76C and 76D)

Prophylactic Surgery

Early trials of prophylaxis for variceal bleeding compared portacaval shunts with medical therapy. Although bleeding was effectively prevented, survival was not significantly enhanced with surgery because of a marked increase in deaths from accelerated hepatic failure (Grace, 1992). Because only one third of patients with varices eventually bleed, surgery cannot be justified as prophylaxis and is not recommended in this setting.

In a prospective, controlled study to evaluate prophylactic surgery in 112 patients with portal hypertension and esophageal varices, Inokuchi (1984) found the bleeding rates were 19.2% in the medical group and 0% in the surgical group. No difference was reported in the survival rate between the two groups at 2-year follow-up, and prophylactic surgery led to a prevention of esophageal bleeding without any increase in the mortality rates. This is the only study to support a role for prophylactic surgery, but it should be noted that the majority of these patients had posthepatic cirrhosis with reasonably well-preserved liver function.