Education + Advocacy = Change

Click a topic below for an index of articles:

New Material



Help us Win the Fight!

Alternative Treatments

Financial or Socio-Economic Issues

Health Insurance

Help us Win the Fight



Institutional Issues

International Reports

Legal Concerns

Math Models or Methods to Predict Trends

Medical Issues

Our Sponsors

Occupational Concerns

Our Board


Religion and infectious diseases

State Governments

Stigma or Discrimination Issues

If you would like to submit an article to this website, email us your paper to




any words all words
Results per page:

“The only thing necessary for these diseases to the triumph is for good people and governments to do nothing.”



Portal Hypertension

Clifford S. Cho, MD

Fellow, Department of Surgical Oncology
Memorial Sloan-Kettering Cancer Center

Layton F. Rikkers, MD, FACS

A.R. Curreri Professor and Chairman, Department of Surgery
University of Wisconsin Medical School

Clinical strategies for managing portal hypertension have undergone significant refinements over the past half-century. This evolution has been driven by advances in our understanding of the physiology of both the disease and the therapies employed against it. Today, clinical management of the portal hypertensive patient is a truly multidisciplinary endeavor, requiring the coordinated efforts of skilled intensivists, gastroenterologists, hepatologists, interventional radiologists, and surgeons. Nevertheless, portal hypertension and its manifold complications remain some of the most vexing problems encountered in modern medicine and surgery.

In this chapter, we briefly review portal venous anatomy and the pathophysiology of portal hypertension [see Sidebar Portal Hypertension: Anatomic and Physiologic Considerations]; however, our main focus is on current practical approaches to managing portal hypertension and its associated sequelae (variceal bleeding, ascites, and hepatic encephalopathy). Of particular relevance to surgeons is that the role of surgical therapy has shifted significantly. Operative treatment now occupies only the final steps in modern treatment protocols for portal hypertension—that is, it serves as a form of salvage for intractable cases that are refractory to other forms of therapy.

Clinical Evaluation


The ultimate aims of diagnostic evaluation in a patient with portal hypertension are (1) to determine the cause of portal hypertension [see Table 1], (2) to estimate hepatic functional reserve, (3) to define the portal venous anatomy and assess hemodynamic status, and (4) to identify the site of GI hemorrhage (if present). Any history of chronic alcohol abuse, hepatitis, or exposure to hepatotoxins raises the suspicion of cirrhotic liver disease. Confirmatory evidence of chronic liver disease on physical examination may be found in the form of jaundice, chest wall spider angiomata, palmar erythema, Dupuytren contractures, testicular atrophy, or gynecomastia. Ascites, splenomegaly, caput medusae, encephalopathic alterations in mental status, and asterixis are all suggestive of portal hypertension.

Investigative Studies


Laboratory studies can also provide indicators of hepatic dysfunction. The hypersplenism that often accompanies cirrhosis can produce mild to moderate pancytopenia. Anemia may also reflect variceal hemorrhage, hemolysis, or simply the chronic malnutrition or bone-marrow suppression associated with chronic alcoholism. Associated hyperaldosteronism, emesis, or diarrhea may give rise to electrolyte derangements, including hyponatremia, hypokalemia, metabolic alkalosis, and prerenal azotemia. Coagulopathy is usually attributable to chronic deficiencies in clotting factors that are normally synthesized by the liver; thus, elevation of the prothrombin time (PT) or the international normalized ratio (INR) often reflects the degree of chronic hepatic impairment. Similarly, the degree of hyperbilirubinemia can be a measure of both acute and chronic hepatic dysfunction. Hepatocellular necrosis results in marked elevations in serum aminotransferases that are readily observed in patients with chronic active viral or alcoholic hepatitis. An alanine aminotransferase (ALT)-aspartate aminotransferase (AST) ratio of 2 or higher is often seen in patients with alcoholic liver disease.

The Child-Pugh scoring system is a useful tool for quantifying hepatic functional reserve [see Table 2].1 Based on total bilirubin and albumin levels, PT (INR), and the clinical severity of ascites and hepatic encephalopathy, the Child-Pugh score predicts both the likelihood of variceal hemorrhage and its anticipated mortality. A newer assessment tool, the Model for End-Stage Liver Disease (MELD) scoring system, which takes the degree of renal impairment and the cause of hepatic dysfunction into account, has also been used to predict outcomes in cirrhotic patients.2

Management of Variceal Bleeding


The prognosis of variceal hemorrhage depends on the presence or absence of underlying cirrhosis. In noncirrhotic patients, the mortality associated with a first episode of variceal hemorrhage ranges from 5% to 10%; in cirrhotic patients, the range is from 40% to 70%. Esophagogastric varices ultimately develop in approximately one half of cirrhotic patients, and bleeding episodes occur in approximately one third of cirrhotic patients with varices. If the initial hemorrhagic episode resolves spontaneously, 30% of patients experience rebleeding within 6 weeks, and 70% experience rebleeding within 1 year. It is noteworthy that overall mortality in patients who survive 6 weeks after an episode of variceal bleeding is statistically indistinguishable from that in persons who have never experienced such an episode.

Further risk stratification is based on the extent of hepatic decompensation. The mortality associated with variceal hemorrhage is 5% for patients with Child class A cirrhosis, 25% for those with Child class B cirrhosis, and over 50% for those with Child class C cirrhosis. The likelihood of recurrent hemorrhage is 28% for patients with Child class A cirrhosis, 48% for those with Child class B cirrhosis, and 68% for those with Child class C cirrhosis.3

Treatment of Acute Variceal Hemorrhage

Section 5 / Chapter 10 - Portal Hypertension

Figure 1. Treatment of acute variceal bleeding

Algorithm outlines treatment of acute variceal bleeding.


Management of acute variceal hemorrhage [see Figure 1] begins with the establishment of adequate airway protection. The risk of aspiration and consequent respiratory deterioration is particularly high among patients with hepatic encephalopathy and those undergoing endoscopic therapy. Accordingly, the threshold for early endotracheal intubation should be low, particularly if endosopic therapy is considered. As with all cases of brisk hemorrhage, adequate venous access is mandatory; placement of a central venous catheter for accurate volume assessment is particularly useful in cases of major bleeding. The presence of chronic liver disease often necessitates vigorous replacement of circulatory volume and coagulation factors, often involving infusion of colloids and transfusion of fresh frozen plasma and packed red blood cells. Antibiotic prophylaxis therapy is recommended because of the propensity of bacterial infections to develop in patients with chronic liver disease after bleeding episodes.

Pharmacologic Therapy

First-line pharmacotherapy for acute variceal bleeding relies on the long-acting somatostatin analogue octreotide, which has been shown to decrease splanchnic blood flow and portal venous pressure. Octreotide is administered in a 250 µg I.V. bolus, followed by infusion of 25 to 50 µg/hr for 2 to 4 days.4 In addition, vasopressin, a strong splanchnic vasoconstrictor, has been shown to control approximately 50% of acute variceal bleeding episodes.4,5 Vasopressin is typically administered in a 20 U I.V. bolus over 20 minutes, followed by infusion of 0.2 to 0.4 U/min. The therapeutic benefits of octreotide and vasopressin appear to be similar, though the side-effect profile of octreotide appears to be much lower than that of vasopressin monotherapy.4 Adjunctive use of nitroglycerin at an initial rate of 50 µg/min (titrated according to blood pressure tolerance) effectively reduces the cardiac complications of vasopressin and thereby facilitates its administration.6 The long-acting vasopressin analogue terlipressin has been shown to be approximately as effective as octreotide.7

Endoscopic Therapy

Endoscopic treatment, in the form of sclerosant injection or band ligation, has become a standard form of therapy for acute variceal hemorrhage. Experienced endoscopists achieve initial control of hemorrhage in 74% to 95% of cases; however, rebleeding rates ranging from 20% to 50% are typically observed.

In endoscopic sclerotherapy, a sclerosant—typically either 5% sodium morrhuate (more common in the United States) or 5% ethanolamine oleate (more common in Europe and Japan)—is injected either intravariceally to obliterate the varix or paravariceally to induce submucosal fibrosis and thereby prevent variceal rupture. Three prospective, randomized, controlled trials demonstrated that endoscopic sclerotherapy, compared with traditional balloon tamponade, achieved better initial hemorrhage control, resulted in fewer episodes of rebleeding, and, in selected cohorts of patients, led to improved long-term survival.8–10 Furthermore, routine use of balloon tamponade after sclerotherapy appeared not to confer any additional therapeutic benefit.8 There are, however, significant risks associated with the use of endoscopic sclerotherapy, including pulmonary complications, transient chest pain, esophageal stricture formation with recurrent sclerotherapy, iatrogenic portal vein thrombosis, hemorrhagic esophageal ulceration, bacteremia, and esophageal perforation.11

Partially in response to the potential complications of endoscopic sclerotherapy, endoscopic variceal band ligation has been advocated as a sclerosant-free therapeutic alternative. The limited data comparing the two approaches suggest a trend toward fewer rebleeding episodes, fewer endoscopic interventions, and significantly lower procedure-related morbidity and overall mortality after variceal ligation.12

Pharmacologic versus Endoscopic Therapy

Two meta-analyses compared medical pharmacotherapy with emergency sclerotherapy as first-line treatment of acute variceal hemorrhage.13,14 No significant differences between the two approaches were demonstrated with respect to initial hemorrhage control or mortality, though treatment-related complications appeared to be significantly more common after sclerotherapy. On the basis of these studies, it has been suggested that endoscopic treatment should be reserved for cases of pharmacotherapeutic failure or that pharmacologic therapy should be initiated in situations where endoscopy is not immediately available. At present, however, it is more common for the two forms of treatment to be employed concurrently. Pharmacotherapy is often initiated in preparation for endoscopy; early mitigation or control of variceal hemorrhage can make endoscopic visualization and intervention easier, safer, and more effective. Indeed, administration of somatostatin before and after endoscopic sclerotherapy has been shown to improve treatment efficacy and decrease transfusion requirements in comparison with endoscopic sclerotherapy alone.15,16

Balloon Tamponade

Section 5 / Chapter 10 - Portal Hypertension

Figure 2. Sengstaken-Blakemore tube

The Sengstaken-Blakemore tube permits tamponade of both the distal esophagus and the gastric fundus. An accessory nasogastric tube permits aspiration of secretions from above the esophageal balloon.

Although the devices used for balloon tamponade have evolved through numerous different forms over the years, all of them rely on the same basic principle—application of direct upward pressure against varices at the esophagogastric junction. Patients for whom balloon tamponade is considered should be intubated endotracheally to prevent airway occlusion and aspiration. The tube is inserted into the stomach, and the gastric balloon is partially inflated with 40 to 50 ml of air [see Figure 2]. An abdominal radiograph is obtained to ensure that the gastric balloon is correctly positioned within the stomach and below the diaphragm. This balloon is then further inflated until it holds 300 ml of air, and the tube is pulled upward with external traction. If hemorrhage is not controlled at this point, the esophageal balloon is inflated to a pressure of 35 to 40 mm Hg. Suction drainage is applied to both the esophageal port and the gastric port to minimize aspiration risk and monitor for recurrent hemorrhage.

When properly applied, direct tamponade therapy is 90% effective in controlling acute hemorrhage. The primary limitation of such therapy is that bleeding resumes in as many as 50% of patients after takedown and removal of the balloon. Furthermore, serious potential complications (e.g., gastric or esophageal perforation, aspiration, and airway obstruction) result in treatment-related mortalities as high as 20%.17,18 Nevertheless, in cases of brisk variceal hemorrhage refractory to pharmacologic and endoscopic therapy, balloon tamponade may have a role to play as a bridge therapy to more definitive forms of treatment, such as transjugular intrahepatic portosystemic shunting (TIPS) (see below) or operative intervention.

Transjugular Intrahepatic Portosystemic Shunting

Section 5 / Chapter 10 - Portal Hypertension

Figure 3. Procedure for performing TIPS

Depicted is the procedure for performing TIPS. (a) A needle is passed under radiologic guidance from a hepatic vein into a major portal venous branch, and a guide wire is advanced through this needle. (b) A balloon is passed over the guide wire, creating a tract in the hepatic parenchyma. (c) An expandable stent is placed though this tract. (d) The effective result is a nonselective portosystemic shunt.


A nonoperative technique for creating an intrahepatic portosystemic fistula for decompression of portal hypertension was proposed in 196919 and first performed in 1982.20 As currently practiced, TIPS is performed by (1) cannulating a hepatic vein (usually the right hepatic vein) via the internal jugular vein, (2) passing a needle from the hepatic vein through the liver parenchyma and into a portal vein branch, (3) passing a guide wire through the needle, (4) dilating the needle tract with a balloon passed over the guide wire, and (5) stenting the tract to a desired diameter, thus effectively constructing a nonselective side-to-side portosystemic shunt [see Figure 3].

Experience with TIPS in the setting of acute variceal hemorrhage is limited. However, one meta-analysis of studies comparing the efficacy of conventional endoscopic therapy (with or without pharmacotherapy) with that of TIPS in treating acute hemorrhagic episodes demonstrated a significant improvement in hemorrhage control with TIPS.21 Unfortunately, this improvement came at the cost of increased rates of hepatic encephalopathy as a consequence of the nonselective shunting of portal venous flow into the systemic venous circulation. Furthermore, the meta-analysis failed to demonstrate a significant improvement in overall mortality with TIPS.21

Given the relative paucity of data on the use of TIPS as first-line therapy for acute variceal hemorrhage, it is logical to recommend that TIPS be employed in cases of pharmacotherapeutic and endoscopic failure; the efficacy of TIPS as salvage therapy in this setting is well documented.22 Contraindications to TIPS include right heart failure and polycystic liver disease. Portal vein thrombosis is a relative contraindication.

Surgical Therapy

The role of surgical management in the treatment of acute variceal bleeding has changed considerably over the past 50 years. At present, operative intervention is reserved for cases that have proved refractory to pharmacotherapy, endoscopy, balloon tamponade, and TIPS. Numerous operations have been developed, each with its own merits and flaws.

Esophageal transection with an end-to-end anastomosis (EEA) stapler has been employed as a means of interrupting blood flow into bleeding esophageal varices. In this technique, the esophagus is mobilized, and the EEA stapler is passed into the distal esophagus through a gastrotomy. With care taken not to injure the vagus nerves and the external periesophageal veins that may be providing collateral venous drainage, a full-thickness segment of the esophagus is transected. When this technique is used on an emergency basis in a patient with acutely bleeding varices, operative mortality is as high as 76%, and the rate of operative complications (e.g., esophageal perforation, stricture, esophagitis, and infection) is approximately 26%.23 Accordingly, esophageal transection is not commonly advocated as a useful form of surgical therapy for acutely bleeding esophageal varices.

In contrast, portosystemic shunting operations have been widely used to treat acute variceal hemorrhage. The largest single body of data on this practice comes from Orloff and associates,24 who reported remarkable outcomes—71% survival at 10 years—in 400 consecutive patients undergoing emergency portacaval shunt operations (mostly side-to-side) over a 28-year period. Unfortunately, these investigators' experience stands in stark contrast to that of most other groups, who uniformly reported operative mortalities of about 40% and 5-year survival rates of about 30%.

Another potential drawback to urgent operative shunting is the manipulation and dissection that are often necessary in the region of the porta hepatis: these measures can result in adhesions and scarring, which can complicate future orthotopic liver transplantation. For this reason, some surgeons have advocated using the mesocaval interposition shunt [see Prevention of Recurrent Variceal Hemorrhage, Surgical Therapy, Portosystemic Shunts, Nonselective Shunts, below] in the emergency setting because of its ability to lower portal pressure without complicating the hilar dissection that will be necessary if transplantation is carried out later.25 In addition, surgeons familiar with the distal splenorenal shunt (DSRS) can employ this selective shunt in some cases of acute variceal hemorrhage unaccompanied by refractory ascites.

Prevention of Recurrent Variceal Hemorrhage



Section 5 / Chapter 10 - Portal Hypertension

Figure 4. Prevention of recurrent variceal bleeding


Algorithm outlines prevention of recurrent variceal bleeding.


Without further treatment, the likelihood that hemorrhage will recur within 1 year after control of an acute episode of variceal bleeding is approximately 70%.26 The pharmacologic maneuver that has been used most extensively to prevent recurrent variceal bleeding [see Figure 4] is nonselective beta-adrenergic blockade, most commonly with propranolol. Although beta blockade has been shown to lower portal pressure and hepatic vein wedge pressure, its ability to induce this effect is variable and unpredictable.27 Nevertheless, a meta-analysis of multiple trials studying the effectiveness of nonselective beta blockade demonstrated a significant decline in recurrent bleeding and a trend toward improved overall survival.4 Patients with decompensated hepatic function appear to derive less benefit from beta blockade, possibly because of the downregulation of beta-adrenergic receptors associated with cirrhosis.28 Adjunctive use of the long-acting vasodilator isosorbide 5-mononitrate (ISMN) appears to potentiate the efficacy of propranolol therapy.29

Endoscopic Therapy

Repeated endoscopic therapy with sclerosant injection or band ligation has been advocated as a means of completely eradicating esophageal varices. Once the varices are eliminated, routine endoscopy is performed at 6- to 12-month intervals to prevent recurrent hemorrhage. Compared with medical treatment, long-term endoscopic therapy results in fewer rebleeding episodes.4 Nevertheless, approximately one half of endoscopically treated patients eventually experience recurrent hemorrhage, usually within the first year. Approximately one third of patients treated with repeated endoscopy ultimately must be converted to another form of therapy because of unrelenting major bleeding.30,31 For this reason, such extended endoscopic surveillance and treatment should be reserved for compliant patients who live in proximity to tertiary medical care and should be administered with the understanding that conversion to a more definitive form of therapy may be necessary if endoscopy fails.

Transjugular Intrahepatic Portosystemic Shunting

TIPS [see   Figure 3 ] has been employed to prevent recurrent episodes of variceal hemorrhage, particularly as a form of bridge therapy for patients awaiting orthotopic liver transplantation. The potential advantage TIPS has over surgical portosystemic shunting is the ability to decompress the portal system without the risks associated with general anesthesia and without postoperative complications. The major limitation of TIPS is the shunt stenosis (caused by neointimal hyperplasia or thrombosis) that occurs in as many as 50% of patients in the first year after the procedure. Fortunately, most such episodes of stenosis are amenable to balloon dilatation or secondary shunt insertion; however, 10% to 15% of TIPS recipients experience total shunt occlusion that cannot be reversed. Furthermore, TIPS functions as a nonselective shunt, leading to hepatic encephalopathy in approximately one third of patients.32

Meta-analytic comparison of TIPS with endoscopic therapy indicates that rebleeding episodes are markedly reduced in patients treated with TIPS, but at the cost of a higher incidence of encephalopathy and a shunt malfunction rate of at least 50%. That the efficacy of TIPS is relatively short-lived makes this modality an ideal form of bridge therapy for patients who are awaiting orthotopic liver transplantation or those who have severe hepatic decompensation and thus are unlikely to live long enough to experience failure of TIPS. TIPS can reduce the number of bleeding episodes for patients on the transplant waiting list. In addition, the significant reduction in portal pressure produced by TIPS technically facilitates future liver transplantation. Finally, unlike surgical shunts, TIPS is completely removed at the time of recipient native hepatectomy.

Surgical Therapy

Surgical therapy is the most effective method of controlling portal hypertension and preventing recurrent variceal hemorrhage. The operative procedures available to the surgeon have undergone numerous modifications and become more effective over the years. Review of the surgical experience reveals that with the onset of alternative modalities (e.g., TIPS and transplantation), the risk status of patients undergoing surgical therapy (as predicted by Child's classification) and the frequency of emergency operations have steadily declined. As a result, the incidence of postoperative hepatic encephalopathy has gradually fallen and overall survival has gradually improved.33 Surgical options for the prevention of recurrent variceal hemorrhage in patients with portal hypertension may be divided into three categories: (1) portosystemic shunt procedures, (2) esophagogastric devascularization, and (3) orthotopic liver transplantation.

Portosystemic shunts Surgical portosystemic shunting provides a means of decompressing the hypertensive portal venous system into the low-pressure systemic venous circulation. Diversion of portal blood flow from the liver also deprives the liver of important hepatotrophic hormones that are present in portal venous blood while routing cerebral toxins normally metabolized by the liver directly into the systemic circulation. As a result, the primary complications of surgical portosystemic shunting are accelerated hepatic dysfunction and hepatic encephalopathy. Primarily in an attempt to minimize these adverse sequelae, various forms of portosystemic shunting operations have evolved, which may be classified as nonselective shunts, selective shunts, or partial shunts.

Nonselective shunts

Section 5 / Chapter 10 - Portal Hypertension

Figure 5. Portosystemic shunts

Nonselective portosystemic shunts either immediately or eventually divert all portal blood flow from the liver into the systemic venous circulation. Shown are the four main variants: (a) end-to-side portacaval shunt, (b) side-to-side portacaval shunt, (c) interposition shunt (portacaval [1], mesocaval [2], and mesorenal [3]), and (d) conventional (proximal) splenorenal shunt.

The classic nonselective portosystemic shunt is the end-to-side portacaval shunt (the so-called Eck fistula) [see Figure 5, part a]. This is the only nonselective shunt that has been rigorously compared with conventional nonoperative therapy. Several randomized, controlled trials demonstrated superior control of bleeding after operative shunting: 9% to 25% of patients experienced rebleeding after portacaval shunting (mostly related to nonvariceal hemorrhage or shunt thrombosis), whereas 65% to 98% of patients experienced rebleeding after medical therapy.34–37 Markedly higher rates of spontaneous posttreatment encephalopathy were reported in the operative shunt groups; however, the overall rates of encephalopathy did not differ between the operative groups and the medical groups, because the encephalopathy seen in the medically treated patients (mainly attributable to hemorrhage and infection) eventually became equivalent to that seen in the surgically treated patients. There were trends toward improved overall survival in the surgical groups, but these trends did not attain statistical significance.

The side-to-side portacaval shunt [see Figure 5, part b] maintains the anatomic continuity of the portal vein as it passes into the liver. However, the high sinusoidal resistance typically present in the setting of cirrhosis effectively renders this shunt a nonselective one, with no measurable antegrade (i.e., hepatopedal) portal blood flow into the liver. Consequently, the encephalopathy rates are no different from those observed after end-to-side portacaval shunting. Side-to-side portacaval shunting does offer the benefit of decompressing the hepatic sinusoidal pressure via reversed (i.e., hepatofugal) flow of blood from the liver into the portal vein. Because transudation of interstitial fluid from both the liver and the intestines is thought to contribute to ascites formation, better control of ascites is achieved with a side-to-side portacaval shunt, which effectively decompresses both the splanchnic veins and the intrahepatic sinusoids, than with an end-to-side portosystemic shunt, which decompresses only the splanchnic veins. The side-to-side portacaval shunt is therefore also recommended for patients with Budd-Chiari syndrome, in whom an end-to-side portacaval shunt would not relieve intrahepatic congestion resulting from hepatic venous outflow occlusion. Otherwise, no significant outcome differences between end-to-side and side-to-side portacaval shunts have been documented. The end-to-side variant is, however, technically easier to construct.

Placement of an interposition mesocaval shunt [see Figure 5, part c] composed of prosthetic or autogenous vein grafts offers the technical advantages of avoiding hilar dissection (thereby making future liver transplantation less complicated) and permitting intentional shunt ligation in the event of refractory postoperative encephalopathy. Like the side-to-side portacaval shunt, the interposition shunt functions physiologically as a nonselective shunt because of the hepatofugal portal venous blood flow. The major drawback to the interposition shunt is shunt thrombosis, which may develop in as many as 35% of cases.

The conventional (proximal) splenorenal shunt [see Figure 5, part d] was initially advocated as a means of decompressing portal venous flow while retaining hepatopedal hepatic portal perfusion. This shunt is constructed by performing a splenectomy and anastomosing the proximal splenic vein to the left renal vein. Physiologic testing of patent conventional splenorenal shunts suggests that they eventually divert all portal flow into the renal vein and therefore effectively function as nonselective shunts. Indeed, long-term rates of hepatic encephalopathy appear to be no lower after conventional splenorenal shunting than after portacaval shunting.38 Shunt occlusion develops in about 18% of cases.39

In short, nonselective portosystemic shunts are an effective means of controlling variceal hemorrhage in cases that are refractory to other therapeutic approaches. Given the absence of any major differences in the rate of encephalopathic complications after the various nonselective shunts, the choice of a nonselective shunting procedure should be based on the surgeon's technical familiarity with the operations and on the patient's candidacy for future transplantation. The end-to-side portacaval shunt can be constructed relatively quickly but should be avoided in patients who have intractable ascites or Budd-Chiari syndrome and those who may subsequently undergo liver transplantation. The side-to-side portacaval shunt may provide better control of ascites but is technically more challenging to construct and should also be avoided if future transplantation is an option. The interposition mesocaval shunt is relatively easy to construct and avoids hepatic hilar dissection but is associated with a relatively high rate of shunt occlusion when a nonautogenous conduit is used. The conventional splenorenal shunt also avoids hilar dissection but is associated with a high shunt occlusion rate and is technically challenging to construct.

Selective shunts

Section 5 / Chapter 10 - Portal Hypertension

Figure 6. Distal splenorenal shunt


The distal splenorenal shunt diverts portal flow from the spleen and short gastric veins into the left renal vein. The DSRS provides selective shunting by preserving portal flow from the mesenteric circulation. Potential sites of collateralization (e.g., the left gastric vein, the gastroepiploic vein, and the umbilical vein) are routinely interrupted to preserve hepatopedal portal flow.

In response to the postoperative complications seen after nonselective portosystemic shunting (hepatic encephalopathy and hepatic failure), Warren and colleagues introduced the distal splenorenal shunt in 1967.40 The DSRS has become the prototypical selective shunt, in that it selectively decompresses the esophagogastric veins while maintaining hepatopedal flow from the mesenteric veins. It is performed by anastomosing the distal splenic vein to the left renal vein and interrupting venous collaterals (e.g., the left gastric and right gastroepiploic veins) [see Figure 6]. As a result, the DSRS effectively separates the portal system into two components: (1) a decompressed esophagogastric venous circuit and (2) a persistently hypertensive mesenteric venous circuit that continues to provide hepatopedal portal flow. Thus, the DSRS does not address the mesenteric and sinusoidal hypertension that is responsible for ascites formation. Indeed, it is believed that the extensive retroperitoneal dissection required to construct this shunt may actually contribute to ascites formation through inadvertent disruption of retroperitoneal lymphatic vessels. The DSRS is contraindicated in patients who have refractory ascites or splenic vein thrombosis, those who have previously undergone splenectomy, and those with an excessively small (< 7 mm) splenic vein diameter.

Unfortunately, perfusion studies indicate that approximately one half of patients lose hepatopedal flow within 1 year after a DSRS procedure. This is a particular problem in patients with alcoholic cirrhosis. The loss of shunt selectivity is believed to result from progressive collateral diversion of portal flow into the splenic vein via a network of pancreatic and peripancreatic veins (the so-called pancreatic siphon effect). Extensive skeletonization of the splenic vein off the pancreas (so-called splenopancreatic disconnection) has been proposed as a means of minimizing this unwanted collateralization,41 but at present, the evidence is insufficient to support routine employment of this measure.

The complications of DSRS procedures are well described. Depending on patient selection, postoperative ascites formation is seen in 7% to 98% of cases; however, in only 0% to 14% of cases is ascites clinically significant and refractory to dietary sodium restriction and diuresis.23 Hepatic encephalopathy is reported in 0% to 32% of cases; several clinical trials comparing DSRS with nonselective shunting demonstrated significantly lower rates of encephalopathy after DSRS, whereas other trials found no statistically significant difference. With respect to overall survival and hemorrhage control, DSRS and nonselective shunts appear to be equivalent.42

Comparison of DSRS construction with endoscopic therapy has yielded interesting results. Two controlled trials comparing endoscopic therapy and salvage DSRS with early DSRS alone demonstrated superior hemorrhage control with early DSRS.30,31 Rates of hepatic encephalopathy did not differ between the two groups. One of the trials, conducted in an urban-suburban area where 85% of sclerotherapy failures could be rescued with salvage DSRS, found survival to be improved in patients treated with endoscopic therapy and salvage DSRS, compared with survival in patients treated with early DSRS alone.30 The other, performed in a less densely populated region where only 31% of sclerotherapy failures could be rescued with salvage DSRS, found survival to be improved in the early DSRS group.31 These data suggest that early definitive surgical intervention may be preferable for patients who are too far from a tertiary medical center to be able to reach one expeditiously in the event of uncontrollable hemorrhage.

Attention is now being turned toward comparisons between the DSRS and TIPS. One uncontrolled comparative study found that with DSRS, hemorrhage control was better, the encephalopathy rate was lower, and shunt occlusion was reduced, but the incidence of postoperative ascites was higher.43 A National Institutes of Health-sponsored randomized comparison between DSRS and TIPS is currently under way at multiple centers.

The other main form of selective portosystemic shunt is the coronary-caval shunt, initially described in Japan in 1984.44 This shunt is constructed by anastomosing an interposition graft to the left gastric (coronary) vein on one end and the inferior vena cava on the other. To date, the applicability of this procedure has been limited, and most surgeons have relatively little experience with it.

Partial shunts Various small-diameter interposition portosystemic shunts have been proposed as partial shunts, designed to achieve partial decompression of the entire portal venous system while maintaining a degree of hepatopedal portal flow to the liver. The most successful of these partial shunts has been the small-diameter portacaval interposition shunt. The use of a 10 mm or smaller interposition shunt, combined with extensive disruption of portosystemic collateral venous circuits, serves to maintain some degree of hepatic portal perfusion. Early experience with the 8 mm ringed polytetrafluoroethylene graft suggests that hepatic encephalopathy rates are lower with this shunt than with nonselective 16 mm grafts and that use of the smaller shunt yields comparable long-term survival.45 An early comparison of the small-diameter portacaval shunt with TIPS demonstrated lower rates of shunt occlusion and treatment failure in the operative therapy group.46


Esophagogastric devascularization

Section 5 / Chapter 10 - Portal Hypertension

Figure 7. Modified Sugiura procedure


Shown is the modified Sugiura procedure. By extensively devascularizing the esophagogastric junction, this procedure may provide a means of interrupting esophagogastric varices without portosystemic shunting.

The most effective nonshunt operation for preventing recurrent variceal hemorrhage is esophagogastric devascularization with esophageal transection and splenectomy, as advocated by Sugiura and associates.47 Unlike simple esophageal transection, which has been used with limited success in the setting of acute hemorrhage, the Sugiura procedure and its subsequent modifications [see Figure 7] involve ligation of venous branches entering the distal esophagus and the proximal stomach from the level of the inferior pulmonary vein, combined with selective vagotomy and pyloroplasty [see 5:20 Gastroduodenal Procedures]. A key point is that the left gastric (coronary) vein and the paraesophageal collateral veins are preserved to permit portoazygous collateralization, which inhibits future varix formation. Initial reports from Japan cited a 5.2% operative mortality and a 6.3% rate of recurrent hemorrhage (most often from nonvariceal causes).47,48 Unfortunately, these successes have not been easily replicated in the United States, where operative mortality with this procedure has exceeded 20%, with bleeding recurring in 35% to 55% of patients.49,50 Nevertheless, modifications of the Sugiura procedure continue to be performed in patients who are unable to undergo shunting procedures because of extensive splanchnic vein thrombosis.

Orthotopic liver transplantation Orthotopic liver transplantation is the most definitive form of therapy for complications of portal hypertension. The cost of cadaveric and living-donor liver transplantation and its attendant immunosuppression, as well as the paucity of available allografts, make liver replacement an option for only a select minority of patients presenting with portal hypertensive sequelae. Accordingly, careful analysis of the outcomes of transplantation procedures in comparison with those of nontransplantation procedures is necessary for optimal allocation of this limited resource.

For patients whose portal hypertension has become refractory to nonoperative management strategies, the decision whether to employ transplantation or nontransplantation operative therapy can be based on the level of hepatic functional reserve. Patients with Child class A or mild class B cirrhosis appear to do well with nontransplantation therapy as first-line operative treatment, with the understanding that liver transplantation may remain an option for salvage therapy in the event of future hepatic functional deterioration. In contrast, patients with more advanced Child class B or Child class C cirrhosis appear to benefit from early transplantation, with nonoperative strategies employed strictly as bridge therapy for maintenance during the time spent on the allograft waiting list.51,52

Prophylaxis of Initial Variceal Hemorrhage

The significant mortality associated with variceal hemorrhage has prompted efforts to devise effective means of preventing the onset of initial variceal bleeding. The difficulty of identifying those 20% to 33% of cirrhotic patients who will experience bleeding episodes remains the primary challenge in the application of prophylaxis for variceal hemorrhage. Patient characteristics that predict an increased likelihood of variceal bleeding include alcoholic cirrhosis, active alcohol consumption, and severe hepatic dysfunction.53 Certain anatomic features of varices seen at the time of endoscopic examination have been shown to predict the likelihood of rupture: evidence of variceal wall thinning (cherry-red spots, red wales), variceal tortuosity, superimposition of varices on other varices, and the presence of gastric varices all appear to be correlated with a higher likelihood of hemorrhage.54

At present, pharmacologic therapy is the only measure that provides effective prophylaxis against variceal hemorrhage. Nonselective beta-adrenergic blockade, either with propranolol or the long-acting agent nadolol, reduces portal venous pressure by decreasing cardiac output and favoring splanchnic vasoconstriction. Clinical trials examining the efficacy of propranolol therapy demonstrated lowered rates of initial variceal bleeding, though the ultimate influence of beta blockade on patient survival was mixed.55–57

Endoscopic sclerotherapy has not been consistently effective in preventing initial variceal bleeding. In fact, several trials found survival to be poorer in patients treated with prophylactic sclerotherapy than in those managed with prophylactic pharmacotherapy.4,58 This difference is probably attributable to the well-documented complications associated with endoscopic sclerotherapy.

The flaws of prophylactic endoscopic sclerotherapy have led some authorities to advocate endoscopic variceal band ligation as a more effective form of prophylaxis. One trial demonstrated that variceal band ligation achieved better prophylaxis of initial variceal bleeding than propranolol therapy did.59 Clearly, this observation warrants further investigation.

Early trials comparing prophylactic portosystemic shunting with medical prophylaxis definitively showed that early operative intervention conferred no significant benefit. In fact, the significant morbidity associated with surgical shunting and the substantial risk of accelerated hepatic dysfunction and encephalopathy led to lower survival rates in patients treated with prophylactic surgical procedures.6,60 At present, the data are insufficient to recommend the use of prophylactic TIPS to prevent acute variceal hemorrhage.

Management of Ascites

The presence of ascites in a patient with portal hypertension is typically an ominous finding that is of significant prognostic importance: 1-year mortalities as high as 50% have been reported in cirrhotic patients with new-onset ascites, whereas baseline 1-year mortalities in cirrhotic patients without ascites are in the range of 10%.61 The pathogenesis of ascites formation appears to be related to the relative hypovolemia and the primary avidity of renal sodium retention that develop in patients with cirrhosis. Hypovolemia induces renin-angiotensin activation and salt and water reabsorption, which, in the setting of chronic liver dysfunction, results in excessive transudation of fluid out of the liver and the intestines and into the peritoneal cavity. The major complications of this process are spontaneous bacterial peritonitis (SBP) and hepatorenal syndrome (HRS) [see Complications, below], which account for the bulk of the morbidity and mortality associated with ascites in patients with portal hypertension.

Nonsurgical Therapy

By addressing the hyperavidity of sodium retention that drives much of ascites formation, restriction of dietary salt intake (to levels as low as 2 g of sodium a day) can resolve ascites in approximately 25% of cases. The hyperaldosteronemic state that exists can be countered by initiating diuresis with spironolactone, which, at dosages ranging from 100 to 400 mg/day, can relieve ascites in an additional 60% to 70% of patients. Although automatic addition of loop diuretics has not been proved to enhance the clinical efficacy of spironolactone, augmentation of spironolactone therapy with furosemide can be helpful for patients whose ascites is refractory to spironolactone monotherapy or who have hyperkalemia as a result of spironolactone treatment. Gradual diuresis is necessary to prevent potential complications (e.g., prerenal azotemia and HRS).62

In cases of ascites that is refractory to medical dietary restriction and diuretic therapy, large-volume paracentesis has been employed with some success. Albumin is typically infused at a dose of 6 to 8 g per liter of ascitic fluid to prevent the hypotension that results from acute volume shifts. Patients in whom ascites recurs after multiple rounds of large-volume paracentesis should be considered for TIPS. TIPS is particularly useful in patients with ascites and a history of bleeding esophageal varices; it corrects as many as 80% of medically refractory cases of ascites.63 However, the efficacy of TIPS is counterbalanced by its attendant risks (i.e., hepatic encephalopathy, shunt occlusion, and accelerated hepatic failure), especially in patients with poor hepatic functional reserve.

Surgical Therapy

Operative intervention plays only a limited role in the management of ascites. Surgically inserted peritoneovenous shunts have been compared with large-volume paracentesis in patients with ascites refractory to medical therapy. No significant differences in early control of ascites have been detected, but patients treated with peritoneovenous shunting appear to benefit from faster ascites resolution, longer palliation, and fewer hospital readmissions.64 Long-term follow-up, however, indicates that shunt occlusion occurs in 47% of patients so treated and disseminated intravascular coagulation in as many as 35%.

The morbidity and mortality associated with operative therapy make routine use of side-to-side portacaval shunts a poor option for managing ascites. The exceptions to this general statement are cases in which ascites proves refractory to medical and TIPS therapy or in which concomitant refractory variceal hemorrhage is present.


SBP is the most common form of ascitic infection. It typically is signaled by fever and abdominal tenderness and often is also accompanied by acute hepatic and renal deterioration. The diagnosis is generally made by analyzing ascitic fluid collected through paracentesis and is defined by the presence of a positive bacterial culture and a neutrophil count higher than 250/mm3 in the absence of an obvious intra-abdominal source of infection. Unlike secondary peritonitis, SBP is typically monomicrobial, and the frequency with which enteric gram-negative rods are found with SBP suggests intestinal bacterial translocation as a potential cause. SBP carries a mortality of 25% and should therefore be treated aggressively with I.V. antibiotic therapy. Given the 70% recurrence rate after an initial episode of SBP, continuation of suppressive antimicrobial therapy until ascites resolves is warranted.65

HRS, a poorly understood state characterized by progressive and refractory renal impairment, typically occurs in the setting of tense ascites and hepatic disease. Management of HRS is strictly supportive, in that the syndrome often responds only to correction of the underlying liver dysfunction. Accordingly, the only proven therapy for HRS is liver transplantation.

Management of Hepatic Encephalopathy

Hepatic encephalopathy is a complex of symptoms characterized by mental status changes ranging from impaired mentation to frank stupor. The classic neurologic finding associated with this symptom complex is asterixis. Typically, hepatic encephalopathy develops in the setting of significant portosystemic shunting or significant hepatic functional impairment. It is most commonly observed after the creation of a therapeutic nonselective portosystemic shunt. Its onset is usually precipitated by dehydration, GI hemorrhage, sepsis, or excessive protein intake; in fact, the spontaneous development of hepatic encephalopathy mandates work-up for these physiologic triggers. It has been speculated that the shunting of intestinally absorbed cerebral toxins (e.g., ammonia, mercaptans, and g-aminobutyric acid) away from hepatic metabolism is what causes hepatic encephalopathy; however, the absolute level of circulating ammonia correlates poorly with the magnitude of encephalopathic symptoms.

Correction of the triggers that cause hepatic encephalopathy often reverses the psychoneurologic disturbances. In severe cases, patients should also receive neomycin (1.5 g every 6 hours), which covers enteric urease-positive bacteria, and lactulose (20 to 30 g two to four times daily), a disaccharide GI cathartic. Both agents are believed to reduce intestinal levels of ammonia and inhibit its enteric absorption. Whereas neomycin has long-term side effects (i.e., nephrotoxicity and ototoxicity), long-term lactulose therapy is generally well tolerated. Dietary protein restriction should also be employed for long-term suppression of hepatic encephalopathy. On occasion, refractory cases of shunt-induced hepatic encephalopathy may be treated by means of intentional ligation or occlusion of the portosystemic shunt.


Figure 2 Carol Donner.

Figures 3 and 5 through 7 Alice Y. Chen.


1. Pugh RN, Murray-Lyon IM, Dawson JL, et al: Transection of the oesophagus for bleeding oesophageal varices. Br J Surg 60:646, 1973

2. Kamath PS, Wiesner RH, Malinchoc M, et al: A model to predict survival in patients with end-stage liver disease. Hepatology 33:464, 2001 

3. Sherlock S: Esophageal varices. Am J Surg 160:9, 1990  

4. D'Amico G, Pagliaro L, Bosch J: The treatment of portal hypertension: a meta-analytic review. Hepatology 22:332, 1995

5. Vlavianos P, Westaby D: Management of acute variceal hemorrhage. Eur J Gastroenterol Hepatol 13:335, 2001  

6. Gimson AE, Westaby D, Hegarty J, et al: A randomized trial of vasopressin and vasopressin plus nitroglycerin in the control of acute variceal hemorrhage. Hepatology 6:410, 1986  

7. Silvain C, Carpentier S, Sautereau D, et al: Terlipressin plus transdermal nitroglycerin vs. octreotide in the control of acute bleeding from esophageal varices: a multicenter randomized trial. Hepatology 18:61, 1993  

8. Barsoum MS, Bolous FI, El-Rooby AA, et al: Tamponade and injection sclerotherapy in the management of bleeding oesophageal varices. Br J Surg 69:76, 1982  

9. Paquet KJ, Feussner H: Endoscopic sclerosis and esophageal balloon tamponade in acute hemorrhage from esophagogastric varices: a prospective controlled randomized trial. Hepatology 5:580, 1985  

10. Larson AW, Cohen H, Zweiban B, et al: Acute esophageal variceal sclerotherapy: results of a prospective randomized controlled trial. JAMA 255:497, 1986 

11. Eckhauser FE, Sarosi G: Endoscopic treatment of esophageal varices and transjugular intrahepatic portal-systemic shunts. Shackleford's Surgery of the Alimentary Tract. Zuidema GD, Yeo CJ, Eds. WB Saunders Co, Philadelphia, 2001

12. Stiegmann GV, Goff JS, Michaletz-Onody PA, et al: Endoscopic sclerotherapy as compared with endoscopic ligation for bleeding esophageal varices. N Engl J Med 326:1527, 1992 

13. Escorsell A, Ruiz del Arbol L, Planas R, et al: Multicenter randomized controlled trial of terlipressin versus sclerotherapy in the treatment of acute variceal bleeding: the TEST study. Hepatology 32:471, 2000  

14. D'Amico G, Pietrosi G, Tarantino I, et al: Emergency sclerotherapy versus vasoactive drugs for variceal bleeding in cirrhosis: a Cochrane meta-analysis. Gastroenterology 124:1277, 2003 

15. Avgerinos A, Nevens F, Raptis S, et al: Early administration of somatostatin and efficacy of sclerotherapy in acute oesophageal variceal bleeds: the ABOVE randomized trial. Lancet 350:1495, 1997  

16. Banares R, Albillos A, Rincon D, et al: Endoscopic treatment versus endoscopic plus pharmacologic treatment for acute variceal bleeding: a meta-analysis. Hepatology 35:609, 2002  

17. Panes J, Teres J, Bosch J, et al: Efficacy of balloon tamponade in treatment of bleeding gastric and esophageal varices: results in 151 consecutive episodes. Dig Dis Sci 33:454, 1988  

18. Avgerinos A, Klonis C, Rekoumis G, et al: A prospective randomized trial comparing somatostatin, balloon tamponade and the combination of both methods in the management of acute variceal haemorrhage. J Hepatol 13:78, 1991  

19. Rosch J, Hanafee WN, Snow H: Transjugular portal venography and radiologic portacaval shunt: an experimental study. Radiology 92:1112, 1969  

20. Colapinto RF, Stronell RD, Birch SJ, et al: Creation of an intrahepatic portosystemic shunt with a Gruntzig balloon catheter. Can Med Assoc J 126:267, 1982  

21. Luca A, D'Amico G, La Galla R, et al: TIPS for the prevention of recurrent bleeding in patients with cirrhosis: meta-analysis of randomized clinical trials. Radiology 212:411, 1999  

22. Sanyal AJ, Freedman AM, Luketic VA, et al: Transjugular intrahepatic portosystemic shunts compared with endoscopic sclerotherapy for the prevention of recurrent variceal hemorrhage: a randomized, controlled trial. Ann Intern Med 126:849, 1997  

23. Maley WR, Klein AS: Portal hypertension. Shackleford's Surgery of the Alimentary Tract. Zuidema GD, Yeo CJ, Eds. WB Saunders Co, Philadelphia, 2001

24. Orloff MJ, Orloff MS, Orloff SL, et al: Three decades of experience with emergency portacaval shunt for acutely bleeding esophageal varices in 400 unselected patients with cirrhosis of the liver. J Am Coll Surg 180:257, 1995  

25. Brems JJ, Hiatt JR, Klein AS, et al: Effect of prior portosystemic shunt on subsequent liver transplantation. Ann Surg 209:51, 1989 

26. Grace ND: A hepatologist's view of variceal bleeding. Am J Surg 160:26, 1990  

27. Garcia-Tsao G, Grace ND, Groszmann RJ, et al: Short-term effect of propranolol on portal venous pressure. Hepatology 6:101, 1986 

28. Gerbes A, Remien J, Jungst D, et al: Evidence for down regulation of beta 2 adrenoceptors in cirrhotic patients with severe ascites. Lancet 21:1409, 1986

29. Garcia-Pagan JC, Fe F, Bosch J, et al: Propranolol compared with propranolol plus isosorbide-mononitrate for portal hypertension in cirrhosis: a randomized controlled study. Ann Intern Med 114:869, 1991  

30. Henderson JM, Kutner MH, Millikan WJ Jr, et al: Endoscopic variceal sclerosis compared with distal splenorenal shunt to prevent recurrent variceal bleeding in cirrhosis: a prospective, randomized trial. Ann Intern Med 112:22, 1990  

31. Rikkers LF, Jin G, Burnett DA, et al: Shunt surgery versus endoscopic sclerotherapy for variceal hemorrhage: late results of a randomized trial. Am J Surg 165:27, 1993  

32. Riggio O, Merlli M, Pedretti G, et al: Hepatic encephalopathy after transjugular intrahepatic portosystemic shunt: incidence and risk factors. Dig Dis Sci 41:578, 1996  

33. Rikkers LF: The changing spectrum of treatment for variceal bleeding. Ann Surg 228:536, 1998  

34. Jackson FC, Perrin EB, Felix W, et al: A clinical investigation of the portacaval shunt: V. Survival analysis of the therapeutic operation. Ann Surg 174:672, 1971 

35. Resnick RH, Iber FL, Ishihara AM, et al: A controlled study of the therapeutic portacaval shunt. Gastroenterology 67:843, 1976

36. Rueff B, Prandi D, Degos F, et al: A controlled study of therapeutic portacaval shunt in alcoholic cirrhosis. Lancet 27:655, 1976

37. Reynolds TB, Donovan AJ, Mikkelsen WP, et al: Results of a 12-year randomized trial of portacaval shunt in patients with alcoholic liver disease and bleeding varices. Gastroenterology 80:1005, 1981 

38. Malt RA, Nabseth DC, Orloff MJ, et al: Occasional notes: portal hypertension, 1979. N Engl J Med 301:617, 1979 

39. Mehigan DG, Zuidema GD, Cameron JL: The incidence of shunt occlusion and portosystemic decompression. Surg Gynecol Obstet 10:661, 1980

40. Warren WD, Zeppa R, Fomon JJ: Selective transsplenic decompression of gastroesophageal varices by distal splenorenal shunt. Ann Surg 166:437, 1967 

41. Inokuchi K, Beppu K, Koyanagi N, et al: Exclusion of nonisolated splenic vein in distal splenorenal shunt for prevention of portal malcirculation. Ann Surg 200:711, 1984 

42. Jin GL, Rikkers LF: Selective variceal decompression: current status. HPB Surg 5:1, 1991  

43. Khaitiyar JS, Luthra SK, Prasad N, et al: Transjugular intrahepatic portosystemic shunt versus distal splenorenal shunt—a comparative study. Hepatogastroenterology 47:492, 2000

44. Inokuchi K, Beppu K, Koyanagi N, et al: Fifteen years' experience with left gastric venous caval shunt for esophageal varices. World J Surg 8:716, 1984 

45. Sarfeh IJ, Rypins EB: Partial versus total portacaval shunt in alcoholic cirrhosis: results of a prospective, randomized clinical trial. Ann Surg 219:353, 1994 

46. Rosemurgy AS, Serafini FM, Zweibel BR, et al: Transjugular intrahepatic portosystemic shunt versus small-diameter prosthetic H-graft portacaval shunt: extended follow-up of an expanded randomized prospective trial. J Gastrointest Surg 4:589, 2000  

47. Sugiura M, Futagawa S: Results of six hundred thirty-six esophageal transactions with paraesophagogastric devascularization in the treatment of esophageal varices. J Vasc Surg 1:254, 1984

48. Idezuki Y, Kokudo N, Sanjo K, et al: Sugiura procedure for management of variceal bleeding in Japan. World J Surg 18:216, 1994 

49. Gouge TH, Ranson JHC: Esophageal resection and paraesophagogastric devascularization for bleeding esophageal varices. Am J Surg 151:47, 1986  

50. Jin G, Rikkers LF: Transabdominal esophagogastric devascularization as treatment for variceal hemorrhage. Surgery 120:641, 1996  

51. Henderson JM: The role of portosystemic shunts for variceal bleeding in the liver transplantation era. Arch Surg 129:886, 1994 

52. Rikkers LF, Jin G, Langnas AN, et al: Shunt surgery during the era of liver transplantation. Ann Surg 228:536, 1997

53. DeFrancis R, Primignani M: Why do varices bleed? Gastroenterol Clin North Am 21:85, 1992 

54. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices: the North Italian endoscopic club for the study and treatment of esophageal varices. N Engl J Med 319:983, 1988

55. Feu F, Bordas JM, Garcia-Pagan JC, et al: Double-blind investigation of the effects of propranolol and placebo in the pressure of esophageal varices in patients with portal hypertension. Hepatology 13:917, 1991

56. LeBrec D: Current status and future goals of the pharmacologic reduction of portal hypertension. Am J Surg 160:19, 1990  

57. Conn HO, Grace ND, Bosch J, et al: Propranolol in the prevention of the first hemorrhage from esophagogastric varices: a multicenter, randomized clinical trial. Hepatology 13:902, 1991  

58. Prophylactic sclerotherapy for esophageal varices in alcoholic liver disease: a randomized, single-blind, multicenter clinical trial. N Engl J Med 324:1779, 1991

59. Sarin SK, Lamba GS, Kumar M, et al: Comparison of endoscopic ligation and propranolol for the primary prevention of variceal bleeding. N Engl J Med 340:988, 1999 

60. Jackson FC, Perrin EB, Smith AG, et al: A clinical investigation of the portacaval shunt: II. Surgical analysis of the prophylactic operation. Am J Surg 115:22, 1968 

61. Gines P, Quintero E, Arroyo V: Compensated cirrhosis: natural history and prognosis. Hepatology 7:122, 1987  

62. Fogel MR, Sawhney VK, Neal EA, et al: Diuresis in the ascitic patient: a randomized controlled trial of three regimens. J Clin Gastroenterol 93:234, 1987

63. Ochs A, Rossle M, Haag K, et al: The transjugular intrahepatic portosystemic stent-shunt procedure for refractory ascites. N Engl J Med 32:1192, 1995

64. Gines P, Arroyo V, Vargas V, et al: Paracentesis with intravenous infusion of albumin as compared with peritoneovenous shunting in cirrhosis with refractory ascites. N Engl J Med 325:829, 1991  

65. Gines P, Rimola A, Planas R, et al: Norfloxacin prevents spontaneous bacterial peritonitis recurrence in cirrhosis: results of a double-blind, placebo-controlled trial. Hepatology 12:716, 1990