Education + Advocacy = Change

Click a topic below for an index of articles:

New Material



Alternative Treatments

Financial or Socio-Economic Issues


Health Insurance



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 at for a review of this paper

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.”

Bleeding esophageal varices

How to treat this dreaded complication of portal hypertension

Ahmed M. Hegab, MD; Velimir A. Luketic, MD


CME learning objectives

  • To understand the conditions under which esophageal bleeding occurs in patients with cirrhosis
  • To learn the best methods of preventing initial bleeding and rebleeding
  • To review the indications for and long-term complications of transjugular intrahepatic portosystemic shunt (TIPS)

The authors disclose no financial interests in this article.

This is the second of three articles on cirrhosis

This page is best viewed with a browser that supports tables.

Preview: Bleeding esophageal varices, one of the most feared complications of portal hypertension, contribute to the estimated 32,000 deaths annually attributed to cirrhosis. Successful control requires knowledge of the pertinent anatomy, underlying pathophysiology of portal hypertension, and natural history of gastroesophageal varices. Drs Hegab and Luketic review these topics and discuss the various prophylactic and therapeutic approaches to management, including pharmacologic agents, endoscopic sclerotherapy, and transjugular intrahepatic portosystemic shunt (TIPS).
Hegab AM, Luketic VA. Bleeding esophageal varices: how to treat this dreaded complication of portal hypertension. Postgrad Med 2001;109(2):75-89

Esophageal varices eventually develop in most patients with cirrhosis, but variceal bleeding occurs in only one third of them. Treatment of patients at highest risk for bleeding is critical because of the high risk of death with each episode of variceal hemorrhage. The goal of treatment of portal hypertension is decreased variceal flow, which is achieved by reducing either portal venous inflow or resistance to portal outflow. This article begins with a review of the pertinent anatomy, which is essential to the understanding of the pathophysiology of portal hypertension and the course of esophageal varices.

Anatomic features

The portal venous system, formed by the confluence of the superior mesenteric vein and the splenic vein (figure 1: not shown), drains the stomach, the large and small intestine, the pancreas, and the spleen. An important feature of this system is that a number of its tributaries also communicate with the systemic circulation. These include the intrinsic and extrinsic veins of the gastroesophageal junction; hemorrhoidal veins of the anal canal; paraumbilical veins and the recanalized falciform ligament; the splenic venous bed and the left renal vein; and the retroperitoneum.

In portal hypertension, these venous collaterals dilate and allow portal venous blood to return to the systemic circulation (figure 2: not shown). Clinically, the most significant collaterals are the intrinsic veins of the gastroesophageal junction, which are located close to the mucosal surface. They are the collaterals most likely to bleed when dilated because of increased blood flow.

The veins of the gastroesophageal junction are classified as intrinsic, extrinsic, and venae comitantes. The intrinsic veins form a subepithelial and submucosal plexus starting at the gastric cardia (upper stomach) and running the length of the esophagus. In healthy persons, these veins drain into the extrinsic plexus through perforating veins 2 to 3 cm above the gastroesophageal junction. Flow through the perforating veins is unidirectional toward the extrinsic plexus and systemic circulation. When portal hypertension develops, however, the valves of the perforating veins become incompetent and allow reversal of flow from the extrinsic to the intrinsic system.

Varices of the gastroesophageal junction usually are classified by location as esophageal or gastric (1). Esophageal varices consist of three or four large trunks that are further characterized by size (table 1). This classification is important because the larger the varix, the more likely it is to bleed. Gastric varices, on the other hand, are by convention classified only by location. Most likely to bleed are the isolated variceal clusters of the fundus, which often are caused by splenic vein thrombosis or spontaneous splenorenal collaterals.

Table 1. Classification of gastroesophageal varices

Small, straight
Enlarged, tortuous; occupy less than one third of the lumen
Large, coil-shaped; occupy more than one third of the lumen

In continuity with esophageal varices

  • Along lesser curve (2 to 5 cm long)
  • Along greater curve extending toward the fundus


  • In the fundus
  • Elsewhere in the stomach


Pathophysiology of portal hypertension

Portal pressure can be measured only angiographically and is expressed in terms of hepatic venous pressure gradient (HVPG) (figure 3: not shown). HVPG is the difference between the wedged hepatic venous pressure and the free hepatic venous pressure. The former is a reflection of sinusoidal pressure, and the latter is a correction for the effects of intra-abdominal pressure (eg, tense ascites). Normally, HVPG is less than 5 mm Hg. Any value greater than 5 mm Hg is by definition portal hypertension. However, several studies have shown that endoscopically identifiable, significant varices form only if HVPG is greater than 12 mm Hg (2). This explains why varices may not be present in every patient with clinical signs of portal hypertension (eg, enlarged spleen, thrombocytopenia, leukopenia). The corollary of this finding is that esophageal hemorrhage occurs only in patients with HVPG greater than 12 mm Hg. Although the absolute value above 12 mm Hg does not correlate with likelihood of bleeding, HVPG of less than 12 mm Hg has become the primary goal of every therapy for portal hypertension.

Portal pressure is directly related to portal venous inflow and the degree of outflow resistance; it can be expressed in terms of Ohm's law as follows:

Portal pressure = portal venous inflow x outflow resistance

The initiating event in the development of portal hypertension is increased resistance to portal outflow. Some causes and sites of the increased resistance are listed in table 2. In cirrhosis, portal hypertension is aggravated by the increase in the portal venous inflow due to splanchnic vasodilation. When portal pressures rise, blood flow is diverted to venous collaterals that dilate to form varices. The likelihood that any one varix will rupture and bleed depends on its wall tension, which can be determined by the application of Poiseuille's and Laplace's laws. In practice, this means that a large, long varix with a high flow rate and a thin wall is most likely to rupture and bleed. Because it is not feasible to shorten a varix or increase its wall thickness, therapies for portal hypertension aim to decrease variceal flow. This decrease is achieved by reducing either portal venous inflow (eg, by splanchnic vasoconstriction) or resistance to portal outflow (eg, by creation of a shunt).

Table 2. Selected causes of portal hypertension

Extrahepatic causes
  • Portal vein thrombosis
  • Extrinsic compression of the portal vein
  • Cavernous transformation of the portal vein

Intrahepatic causes

  • Sarcoidosis
  • Primary biliary cirrhosis
  • Hepatoportal sclerosis
  • Schistosomiasis

Alcoholic hepatitis

Budd-Chiari syndrome (hepatic vein thrombosis)
Veno-occlusive disease
Severe congestive heart failure
Restrictive heart disease


Natural history of gastroesophageal varices

De novo varices develop annually in 5% to 15% of patients with cirrhosis. Once present, varices enlarge by 4% to 10% each year. Varices develop eventually in most patients with cirrhosis, but only one third of them experience variceal bleeding. It is essential to identify and treat those patients at highest risk because each episode of variceal hemorrhage carries a 20% to 30% risk of death, and up to 70% of patients who do not receive treatment die within 1 year of the initial bleeding episode.

Risk factors for variceal bleeding are summarized in table 3. When combined, these factors can predict the risk of bleeding with some precision. Thus, the risk of bleeding within 1 year is less than 10% for a patient with a Child-Pugh class A cirrhotic liver with small varices but is 76% for a patient with a class C cirrhotic liver with large varices and "red signs" (3).

Table 3. Risk factors for variceal bleeding

Portal pressure
  • HVPG >12 mm Hg

Varix size and location

  • Large esophageal varices
  • Isolated cluster of varices in fundus of stomach

Variceal appearance on endoscopy ("red signs")

  • Red wale marks (longitudinal red streaks on varices)
  • Cherry-red spots (red, discrete, flat spots on varices)
  • Hematocystic spots (red, discrete, raised spots)
  • Diffuse erythema

Degree of liver failure

  • Child-Pugh class C cirrhosis*

Presence of ascites

  • Tense ascites

HVPG, hepatic venous pressure gradient.

*The Child-Pugh classification measures liver function on a scale of 5 to 15. A score of 10 or higher indicates class C (advanced disease, severely decreased liver function, increased probability of death).


After an acute variceal hemorrhage, bleeding resolves spontaneously in 50% of patients. Bleeding is least likely to stop in patients with large varices and a Child-Pugh class C cirrhotic liver. Once the bleeding stops, patients remain at increased risk of rebleeding for about 6 weeks; such risk is greatest during the first 48 hours after hemorrhage. Factors associated with early rebleeding include age over 60 years, renal failure, and severe initial bleeding defined by hemoglobin of less than 8 g/dL at presentation. Risk factors for late rebleeding are severe liver failure, continued alcohol abuse, large variceal size, renal failure, and hepatocellular carcinoma.

Management of variceal hemorrhage

Approaches to prevention and treatment of variceal hemorrhage have included pharmacotherapy, endoscopic intervention, surgical therapy and, more recently, radiologic shunting. All of these treatments are limited by their inability to prevent or arrest hemorrhage in a universal manner, extensive side-effect profiles, and failure to improve long-term survival rates. Availability of resources and expertise is an important consideration in determining the best approach.

Primary prophylaxis
Therapy that prevents the initial bleeding episode is considered primary prophylaxis. Surgical portal decompression (diversion of portal flow to the vena cava via an anastomosis between the two veins) was the first such treatment. More than 25 years ago, four seminal studies confirmed the need for randomized clinical trials in gastroenterology. The studies also demonstrated that although surgical shunting is excellent at preventing the first bleeding episode, it also can lead to severe encephalopathy and shorter overall survival. As a result, surgical shunting was abandoned for primary prophylaxis.

Similarly, endoscopic sclerotherapy is effective in preventing the first variceal hemorrhage, but it has been associated with side effects (eg, bleeding, perforation, pleural effusion, strictures) and higher overall mortality rate when compared with sham treatment (4).

Finally, not enough is known about the effectiveness of either radiologically placed shunts or endoscopic variceal ligation (2,5). Currently, only nonselective beta blockers are recommended for primary prevention of variceal hemorrhage.

Nonselective beta blockers: Both propranolol hydrochloride (Inderal) and nadolol (Corgard) reduce portal pressure through beta2 blockade, allowing unopposed alpha-adrenergic activity to constrict mesenteric arterioles and reduce portal venous flow. Higher doses of these drugs than usually used to treat portal hypertension also decrease cardiac output, further limiting flow to the splanchnic circulation.

Ten placebo-controlled studies of more than 1,000 patients with cirrhosis and esophageal varices (6) have evaluated the use of beta blockers to prevent the initial variceal hemorrhage. Differences in size of varices, Child-Pugh class, HVPG, and end points made direct comparison difficult, but several general observations can be made. During the first year of follow-up, those treated with either propranolol or nadolol bled at a lower rate than those who received placebo (0% to 18% versus 12% to 30%). The difference was even more pronounced after 2 years of follow-up. Meta-analysis of the data shows that beta blockade reduced the risk of bleeding by 45% and of bleeding-related death by 50%. However, only one study found an overall survival advantage. Predictors of poor outcome included younger age, large varices, advanced liver disease, and lower propranolol doses.

The beta-blocker dose is adjusted to decrease the resting heart rate by 25% from its baseline, but not to less than 55 to 60 beats/min. In clinical trials, 10 to 480 mg of propranolol daily, in divided doses, or 40 to 320 mg of nadolol daily in a single dose was used.

Nitrates: About 20% of patients do not respond to nonselective beta blockade. In another 20%, beta blockers are contraindicated, and yet another 3% to 27% have side effects that require discontinuation of therapy. All of these patients may benefit from long-acting nitrates. At usual doses, nitrates are vasodilators that decrease cardiac output by reducing venous return. Systemic vasodilation also decreases post-sinusoidal resistance and portal pressure. At high doses, nitrates cause arterial dilation and hypotension, leading to reflex splanchnic vasoconstriction and further decreases in portal pressures.

When compared with propranolol, isosorbide mononitrate was found to be equally effective in preventing bleeding and death after a mean of 2 years (7,8). During long-term follow-up (up to 7 years), however, isosorbide monotherapy was associated with a higher mortality rate, especially in subjects older than 51 years, and thus its prolonged use cannot be recommended. Combination therapy with beta blockers and nitrates has not been validated in controlled trials.

Guidelines: All patients with cirrhosis should undergo endoscopy to screen for varices. The optimum interval for subsequent screening is unknown, but several established centers have adopted a policy of screening every 2 to 3 years. Patients with large varices or endoscopic signs for increased risk of bleeding, or both, should be treated with nonselective beta blockers, which is the only therapy shown to prevent the initial variceal bleeding episode. Whether treating patients with small varices is cost-effective or influences survival has not been determined.

Treatment of acute hemorrhage
Goals in the management of active bleeding are hemodynamic resuscitation, prevention and treatment of complications, and control of bleeding. Hemodynamic resuscitation requires administration of blood products and crystalloid. Care must be taken to avoid overtransfusion, because rebound portal hypertension can lead to early rebleeding. Typically, hematocrit is maintained in the low 30% range. Clotting factors often need to be determined as well. Platelet transfusions are reserved for counts below 50,000/mL in an actively bleeding patient.

Complications related to bleeding or its treatment can substantially increase the risk of death in each episode. Thus, patients with altered mental status and those with massive hematemesis should be intubated for airway protection; prevention of aspiration pneumonia cannot be overemphasized. Because patients with acute variceal hemorrhage are often volume-depleted, nephrotoxins (eg, aminoglycosides, nonsteroidal anti-inflammatory drugs) should be avoided to prevent renal failure. Patients should be monitored for abnormalities such as hypocalcemia and hyperkalemia, which are common during significant blood product transfusions. Because bacteremia often occurs during endoscopic sclerotherapy, patients with valvular heart disease and those with ascites should receive antibiotics at the time of the procedure to prevent subacute bacterial endocarditis and spontaneous bacterial peritonitis, respectively.

Currently, therapeutic endoscopy is the definitive treatment for active variceal hemorrhage. However, when endoscopy is unavailable or delayed, pharmacologic agents play an important role in altering the course of a bleeding episode.

Vasopressin: Vasopressin (Pitressin Synthetic) therapy directly constricts mesenteric arterioles and decreases portal inflow, thus decreasing portal pressure and controlling as many as 60% to 75% of variceal bleeding episodes (6). However, it does not increase the survival rate and may actually increase the mortality rate because of vasoconstrictive effects on other organs (eg, heart, intestine). Nitroglycerin (Nitro-Bid IV, Tridil) given concomitantly alleviates some of the vasoconstriction; a meta-analysis of three controlled trials (1) has shown that this combination is better than vasopressin alone for controlling acute bleeding.

Octreotide acetate: Octreotide acetate (Sandostatin) is a synthetic, long-acting analogue of somatostatin (Zecnil). By inhibiting the release of vasodilatory hormones (eg, glucagon), it indirectly causes splanchnic vasoconstriction and decreased portal flow. Several studies found octreotide to be more effective than either placebo or vasopressin in controlling both initial and sustained bleeding (6). Because it has fewer side effects than vasopressin, octreotide has become the drug of choice in the pharmacologic management of acute variceal bleeding. Unfortunately, like vasopressin, it does not increase the survival rate. Octreotide infusion should be started at initial presentation. In most studies, a loading dose of 50 micrograms was given, followed by an infusion of 50 micrograms/hr. Treatment continued through the fifth hospital day following the initial bleeding episode. Definitive endoscopic therapy usually is performed shortly after hemostasis is achieved.

Endoscopic therapy: Endoscopic sclerotherapy remains first-line therapy for active variceal bleeding. A sclerosant (eg, morrhuate sodium [Scleromate]) is injected into a varix under direct vision during endoscopy. This causes tissue edema and mechanical compression followed by inflammation, variceal thrombosis, fibrosis and, finally, obliteration. Complications include bleeding ulcers, dysphagia due to stricture formation, and pleural effusions. Serious but rare problems are aspiration pneumonia, acute respiratory distress syndrome, and bowel perforation.

Meta-analysis of six randomized studies proved sclerotherapy to be more effective than conservative measures (balloon tamponade or vasopressin therapy) in controlling active bleeding (3). A similar review of five trials comparing sclerotherapy with octreotide or somatostatin showed no clear advantage in either bleeding control or survival rate. The addition of octreotide to sclerotherapy, however, resulted in significant improvement in early bleeding control in four of five trials (1,9).

Endoscopic variceal ligation appears to be emerging as a viable alternative to sclerotherapy, mainly because of fewer complications and similar efficacy in bleeding control (10). Elastic bands are placed around varices using a device attached to the end of the endoscope. Ischemic necrosis, thrombosis, and fibrosis ensue, eradicating the varix. Concomitant use of octreotide may further decrease rebleeding rates, but controlled trials are needed to confirm this.

Transjugular intrahepatic portosystemic shunt: TIPS is an angiographically created shunt between hepatic and portal veins that is kept open by placement of a fenestrated metal stent. It effectively decompresses the portal system, controlling active variceal bleeding over 90% of the time and achieving a mortality rate of less than 10%, even in critically ill patients (2). Immediate complications include secondary bleeding and, in 20% of cases, worsening encephalopathy. The most common long-term complications are stent stenosis or occlusion that requires balloon angioplasty. Causes of bleeding and recurrent portal hypertension after TIPS are summarized in table 4. As a consequence, TIPS is primarily used as rescue therapy when pharmacologic and endoscopic treatment of acute bleeding is unsuccessful (11).

Table 4. Causes of bleeding and recurrent portal hypertension after TIPS

Stent dysfunction
  • Thrombosis
  • Retraction
  • Displacement
  • Stenosis

Severe right-sided heart failure


Persistent gastric varices

  • Associated with spontaneous splenorenal collaterals
  • Associated with massive splenomegaly

TIPS, transjugular intrahepatic portosystemic shunt.


Other considerations: If bleeding persists (or recurs within 48 hours of the initial episode) despite pharmacologic therapy and two endoscopic therapeutic attempts at least 24 hours apart, patients should be considered for salvage therapy. Depending on local expertise and the patient's condition, either TIPS or surgical treatment (transection of esophageal varices and devascularization of the stomach, portacaval shunt, or liver transplantation) should be considered. In general, TIPS is preferred for high-risk patients because of its lower complication rates and ability to act as a bridge to liver transplantation. When bleeding occurs more than 48 hours after the initial episode, a second attempt at endoscopic therapy should be made.

Guidelines: An approach to management of acute variceal hemorrhage is presented in figure 4 (not shown).

Secondary prophylaxis
Without additional therapy, a patient who has survived an episode of variceal hemorrhage has an overall risk of rebleeding that approaches 70% at 1 year. The clinical course often is one of recurrent hemorrhage that leads to hepatic decompensation and death (1). Prevention of recurrent hemorrhage is therefore the key to survival in a patient with varices. All of the treatments previously described also are available for secondary prophylaxis.

Endoscopic therapy: Sclerotherapy is the most extensively studied treatment for prevention of recurrent variceal hemorrhage. When compared with placebo, it has been shown to decrease the risk of rebleeding (50% versus 70%) and death (30% to 60% versus 50% to 75%) (1). None of the nine trials comparing beta blockade plus sclerotherapy with sclerotherapy alone (1) demonstrated a survival advantage for combination therapy, and rebleeding rates were similar. More recently, a series of controlled trials has shown endoscopic variceal ligation to be superior to sclerotherapy in time to variceal obliteration, number of complications, number of rebleeding episodes, and survival rate (10). As a result, band ligation has replaced sclerotherapy as first-line treatment in secondary prevention of variceal hemorrhage.

Banding is carried out every 2 to 3 weeks until obliteration. All the complications of sclerotherapy previously described can occur during these sessions. Most episodes of breakthrough bleeding occur before the varices are completely obliterated. Because varices tend to recur over time, surveillance endoscopy must be performed every 6 to 12 months so that banding can be reinstituted as needed.

Shunts: TIPS has been compared with endoscopic therapy in a number of studies. Patients who have had TIPS are significantly less likely to bleed than those who underwent endoscopic therapy (25% to 59% versus 15% to 25%). Their survival rate, however, is not improved (and may be poorer), and they have to contend with numerous procedure-related complications (eg, hematomas, liver capsule rupture, pulmonary edema) and long-term complications (eg, encephalopathy, liver failure, hemolysis, TIPS stenosis). TIPS therefore is used primarily as salvage therapy in the 40% to 50% of patients who have failed adequate endoscopic and pharmacologic therapy (2). Surgical shunts also are excellent at controlling active bleeding and preventing rebleeding, but at the expense of a significantly increased risk of death (especially in patients with advanced disease) when compared with endoscopic therapy. Liver transplantation is the only effective treatment for both portal hypertension and liver failure and should be considered in any patient who survives a variceal bleeding episode.

Guidelines: An approach to secondary prophylaxis is presented in figure 4 (not shown).


Variceal hemorrhage is one of the most serious complications of portal hypertension. Nonselective beta blockers are the treatment of choice for prevention of the first bleeding episode. Active bleeding is managed with octreotide and endoscopic sclerotherapy. TIPS and shunt surgery are reserved for those in whom octreotide and endoscopic surgery have failed. Endoscopic band ligation should be used for prevention of recurrent bleeding. If endoscopic band ligation fails, patients can be offered TIPS or surgical therapy; they should be evaluated for liver transplantation.


1.      Luketic VA, Sanyal AJ. Esophageal varices. I. Clinical presentation, medical therapy and endoscopic therapy. GI Clin North Am 2000;29(2):337-85

2.      Luketic VA, Sanyal AJ. Esophageal varices. II. Transjugular intrahepatic portosystemic shunt and surgical therapy. GI Clin North Am 2000;29(2):387-421

3.      Sanyal AJ, Purdum PP III, Luketic VA, et al. Bleeding gastro-esophageal varices. Semin Liver Dis 1993;13(4):328-42

4.      The Veterans Affairs Cooperative Variceal Sclerotherapy Group. Prophylactic sclerotherapy for esophageal varices in men with alcoholic liver disease: a randomized, single-blind, multicenter clinical trial. N Engl J Med 1991;324(25):1779-84

5.      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 1999;340(13): 988-93

6.      Sanyal AJ, Shiffman ML. Pharmacologic treatment of portal hypertension. In: Lewis JH, Dubois A, eds. Current clinical topics in gastrointestinal pharmacology. London: Blackwell Scientific, 1997:242-75

7.      Angelico M, Carli L, Piat C, et al. Effects of isosorbide-5-mononitrate compared with propranolol on first bleeding and long-term survival in cirrhosis. Gastroenterology 1997;113(5):1632-9

8.      Merkel C, Marin R, Enzo E, et al. Randomised trial of nadolol alone or with isosorbide mononitrate for primary prophylaxis of variceal bleeding in cirrhosis. Gruppo Triveneto per L'Ipertensione portale. Lancet 1996;348(9043):1677-81

9.      Planas R, Quer JC, Boix J, et al. A prospective randomized trial comparing somatostatin and sclerotherapy in the treatment of acute variceal bleeding. Hepatology 1994;20(2):370-5

10. Laine L, Cook D. Endoscopic ligation compared with sclerotherapy for treatment of esophageal variceal bleeding: a meta-analysis. Ann Intern Med 1995;123(4):280-7

11. Sanyal AJ, Freedman AM, Luketic VA, et al. Transjugular intrahepatic portosystemic shunts for patients with active variceal hemorrhage unresponsive to sclerotherapy. Gastroenterology 1996;111(1):138-46

Dr Hegab is a fellow and Dr Luketic is associate professor of medicine, division of gastoenterology, Virginia Commonwealth University School of Medicine, Richmond. Correspondence: Velimir A. Luketic, MD, Division of Gastroenterology, Medical College of Virginia, Virginia Commonwealth University, PO Box 980341, Richmond, VA 23298-0341.

Symposium Index

  • CIRRHOSIS OF THE LIVER: Introduction to a three-article symposium by Anastasios A. Mihas, MD
  • HEPATIC ENCEPHALOPATHY: Metabolic consequence of cirrhosis often is reversible by Souheil Abou-Assi, MD, Z. Reno Vlahcevic, MD
  • BLEEDING ESOPHAGEAL VARICES: How to treat this dreaded complication of portal hypertension by Ahmed M. Hegab, MD, Velimir A. Luketic, MD
  • MINIMIZING ASCITES: Complication of cirrhosis signals clinical deterioration by Nelson Garcia Jr, MD, Arun J. Sanyal, MBBS, MD