Overview

Variceal bleeding is the most serious complication of cirrhosis, with an up to 50% mortality rate within 30 days of presentation. However, as management has improved, so have patient outcomes (Carbonell et al, 2004; El-Serag & Everhart, 2000; Jamal et al, 2008a; Stokkeland et al, 2006). Specifically, the 30-day mortality rate has decreased from 30% to 43% in earlier eras to only 15% to 21% currently. These lower mortality rates may be attributed to the combined effects of aggressive clinical treatment protocols and the development of more effective therapies such as endoscopic band ligation; intravenous vasoactive medications, such as octreotide and terlipressin; and antibiotic prophylaxis. For example, Carbonell and colleagues (2004) reported that nearly all patients currently admitted for variceal bleeding receive antibiotic prophylaxis (94%) and endoscopic therapy (90%) within 24 hours compared with only 2% and 6%, respectively, 20 years ago. In addition, evidence suggests that prophylactic therapy is reducing the overall rate of bleeding; currently 15% fewer patients are hospitalized than a decade ago (Jamal et al, 2008b). This finding underscores the clinical utility of the early diagnosis of esophageal varices through aggressive screening strategies, careful surveillance for small varices, and prophylatic therapy with either β-blockers or endoscopic band ligation before or after variceal bleeding occurs.

Diagnosis of Esophageal Varices

Screening for esophageal varices is an important component in the long-term management of patients with portal hypertension because effective therapy may be applied once varices are identified. The gold standard test is upper endoscopy (de Franchis, 2005; Garcia-Tsao, 2007, 2008), during which the variceal size and presence of red wale markings should be noted. Varices are classified in three sizes: small varices show minimal elevation of veins above the mucosa; medium describes varices with tortuous veins that occupy less than one third of the lumen; and large varices occupy more than one third of the lumen. However, as recognized by all experienced endoscopists, the interobserver agreement in size estimation of esophageal varices varies (Bendtsen et al, 1990).

The diagnosis of esophageal varices should be considered in any patient with cirrhosis. (Garcia-Tsao et al, 2007; Garcia-Pagan et al, 2008). Because the risk of variceal progression and bleeding depend on both the severity of the underlying liver disease and appearance of the varices at endoscopy, patients with decompensated liver disease and high-risk endoscopic features (large varices, red wale markings) require more aggressive screening protocols. In addition, screening may also be appropriate in selected patients with bridging fibrosis because 16% may have esophageal varices (Sanyal et al, 2006). The likelihood of finding varices in a particular patient largely depends on the severity of the underlying liver disease, with a prevalence of 60% and 30% in decompensated and compensated patients, respectively (D’Amico et al, 1995; Sanyal et al, 2006). Once endoscopy has been performed, the presence or absence of varices has been established, and the varices have been graded according to severity, the requirement for further screening, surveillance, and therapy may be determined. In general, treatment recommendations for medium varices are the same as for large varices. The general protocol for screening and treatment is outlined in Table 75A.1.

Table 75A.1 Screening and Surveillance Recommendation for Esophageal Varices

In cirrhotic patients without varices, the risk of varices subsequently developing is approximately 8% per year (Groszmann et al, 2005; Merli et al, 2003). The recommended screening interval to detect varices in such patients is within 3 years. With a yearly risk of new varices of 8%, the cumulative risk of developing new varices over 3 years would only be about 25%, and during this interval the risk of bleeding is quite small (1% per year) (Groszmann et al, 2005; Merli et al, 2003). Because the risk of developing varices is higher in patients with decompensation, the screening interval may be shortened in this instance (North Italian Endoscopic Club, 1988). Prophylactic therapy with β-blockers to prevent the development of varices or bleeding in cirrhotics without varices is not effective, as was determined by two randomized studies, one with propranolol (Cales et al, 1999) and one with timolol (Groszmann et al, 2005). Because both studies failed to show a benefit with nonselective β-blockers in these populations, the pharmacologic prevention of varices, or “preprimary prophylaxis,” is not possible at present.

In the patient with small varices identified at initial endoscopy, the yearly risk of developing large varices is 7% to 12% (Cales et al 1999; Merkel et al, 2004) and the risk of hemorrhage is only 5%. Repeat endoscopy should be performed at 2 years or yearly in the case of decompensation. Prophylactic treatment with β-blockers in patients with small varices has led to mixed results. In these studies, one trial reported a higher 2-year rate of developing large varices (31%) with treatment compared with placebo (14%), but the dropout rate in the study was very high. The other trial reported a lower rate of progression to large varices with β-blockers (11%) versus placebo (37%) after 3 years. The risk of bleeding was lower (12%) at 5 years with treatment compared with placebo (22%). However, the benefit of β-blockers was mitigated by the fact that the rate of bleeding in the placebo group, which was started on β-blockers once large varices were discovered, was the same as treatment once β-blocker therapy was initiated. In addition, the rate of withdrawal because of side effects was higher with β-blockers (10%) compared with placebo (1%). Because of these mixed results, patients with small varices should receive prophylactic β-blocker therapy only if they have risk factors for bleeding, such as advanced liver disease or red wale markings. In patients not receiving β-blockers, repeat endoscopy is recommended every other year or annually for decompensation, and repeat endoscopy is not required for patients taking β-blockers.

In patients with medium or large varices identified at screening endoscopy, prophylactic therapy is clearly recommended, as described below. This group of patients clearly derives the greatest benefit from therapy. The primary treatment options are obliteration with esophageal variceal band ligation (EVL) or reduction of portal pressure with oral pharmacologic agents. There is currently no role for decompressive therapy, transjugular intrahepatic portosystemic shunt (TIPS), or surgical shunting for prophylaxis of variceal bleeding. In addition, endoscopic sclerotherapy has no role in prophylaxis (see Chapter 70B).

Noninvasive Means of Diagnosing Esophageal Varices

Although upper endoscopy is the most effective screening tool for esophageal varices, the invasive nature of this procedure has led investigators to find alternative noninvasive imaging techniques, as well as methodologies for risk stratifying patients for screening, as shown in Table 75A.2. Noninvasive screening techniques have several potential advantages over endoscopy. They are generally better accepted by patients, which could potentially increase adherence to screening protocols. In addition, these methods are typically less expensive than endoscopy. Finally, varices are absent in most patients (60%) with compensensated cirrhosis. Because the diagnostic yield in this cohort of patients is low, alternative noninvasive screening could risk stratify patients eligible for invasive screening. The four general types of alternative screening modalities are 1) blood chemistries, 2) cross-sectional imaging, 3) transient elastography, and 4) alternative endoscopic techniques, such as ultrathin endoscopy and capsule endoscopy.

Table 75A.2 Diagnosis of Esophageal Varices

The least invasive means of screening for varices is through analysis of blood chemistries. Numerous biochemical parameters have been considered to evaluate the presence of esophageal varices, including platelet count, prothrombin time, serum albumin, and Child-Turcotte-Pugh (CTP) class. However, none of these surrogate markers used alone or in combination is sufficiently reliable for use in clinical practice (D’Amico & Morabito, 2004; de Franchis, 2008; Qamar et al, 2008).

In recent years, cross-sectional imaging, including multidetector computer tomography (CT) and gadolinium-enhanced magnetic resonance (MRI), have been evaluated for esophageal variceal screening. However, compared with CT, there is relatively little experience with MRI (Annet et al, 2006; Goshima et al, 2009; Matsuo et al, 2003). Multidetector CT provides better images than conventional CT, thereby increasing the yield in identifying varices (Kim et al, 2009). In a direct comparison of multidetector CT and conventional endoscopy, both modalities identified varices, and radiologists accurately predicted the presence of large varices 93% of the time (Perri et al, 2008). However, the information from CT imaging was much less detailed compared with endoscopy. The absence of varices was accurately predicted by CT in only about 50% of patients; in more than one third of cases, it inaccurately identified small varices when large varices were seen endoscopically. Aside from limited sensitivity and specificity in variceal screening, CT has several other drawbacks . It cannot provide information about the red wale markings or other stigmata associated with increased risk for bleeding. In addition, exposure to ionizing radiation is significant, especially with repeated CT imaging (de Franchis, 2008; Thabut et al, 2008). However, one potential advantage of CT over endoscopy is its ability to identify other relevant abdominal pathology, namely hepatocellular carcinoma. Currently, cross-sectional imaging is not sufficiently reliable for use in routine clinical screening protocols for esophageal varices.

Transient elastography is an ultrasound technique that noninvasively measures tissue elasticity. This technique is studied primarily for measuring liver stiffness as a means to determine the presence and severity of hepatic fibrosis (Abenavoli et al, 2007; Del Poggio & Columbo, 2009). The test can easily be performed by nonphysicians with a dedicated device by generating a low-amplitude shear wave that propagates through the liver parenchyma. The velocity of propagation is directly proportional to liver stiffness and is automatically calculated by the device and expressed in units of pressure (kPa): the higher the value, the stiffer and more fibrotic the tissue and, by correlation, the more pressure required to deform the tissue. Although universally accepted ranges have not been developed, the general ranges for cutoff values for normal are up to 8 kPa and more than 13 to 18 kPa for cirrhosis.

Aside from being noninvasive and easily performed, the advantages of elastography include assessment of a larger hepatic area than liver biopsy and general correlation with the extent of histologic fibrosis. The disadvantages regarding variceal screening are that it is not a direct assessment of varices, and the test may be inaccurate or unreliable in patients with ascites and in obese patients. In fact, up to 10% of cases may not return any reading (Foucher et al, 2006; Sandrin et al, 2003).

Several studies have evaluated transient elastography in the noninvasive diagnosis and staging of esophageal varices. Theoretically, as liver stiffness and hepatic fibrosis progress, the likelihood of developing esophageal varices should increase. However, this methodology is subject to the same problems as other noninvasive techniques. Using a cutoff value of 12.5 kPa, Castera and colleagues (2009) studied 298 patients with hepatitis C virus (HCV) infection and reported only 76% sensitivity and 78% specificity in the diagnosis of esophageal varices. In another study, investigators compared two cutoff values for liver stiffness in predicting the presence of esophageal varices in 61 patients with advanced fibrosis from HCV (Vizzutti et al, 2007). At a lower cutoff of 17.6 kPa, the sensitivity was 90% and specificity was 43%. At the higher cutoff of 27.4 kPa, more patients with varices were missed (sensitivity of 70%), but specificity improved to 78%. Similar results were reported by Kazemi and colleagues (2006), who examined the correlation between liver stiffness and presence of large esophageal varices.

Although more studies are needed, current data suggest that the diagnostic accuracy of transient elastography is sufficiently limited that it cannot supplant upper endoscopy as a recommended screening technique for esophageal varices (de Franchis, 2008; Del Poggio, 2009). In fact, it seems unlikely that it would ever replace endoscopy as the primary screening tool. However, because it is a less invasive technique, elastography could possibly be used as an adjunct screening methodology. Patients could be monitored for changes in their transient elastography profile between recommended endoscopic screening intervals to potentially identify specific patients at higher risk for variceal progression. However, such screening and surveillance recommendations are only speculative at this point.

During the past decade, the utility of alternative video endoscopic devices has been studied in screening for esophageal varices. One of these modalities is the ultrathin endoscope inserted without sedation via a transnasal and transoral approach. The primary advantage of this procedure compared with conventional endoscopy is the avoidance of conscious sedation with its attendant risks, cost, and inconvenience. Unlike cross-sectional imaging or transient elastography, video endoscopy (ultrathin and capsule endoscopy) provides direct visualization of the esophageal mucosa that may improve estimation of variceal size and red wale markings, which is not possible with cross-sectional imaging. The primary disadvantage of alternative endoscopy is that compared with conventional endoscopy, these devices have a limited field of view and cannot perform interventions such as biopsy or banding.

Limited data are available on ultrathin endoscopy in variceal screening, and most studies, but not all, show that this technique is more easily performed, has greater patient acceptance, and provides an acceptable diagnostic yield compared with conventional endoscopy (Catanzaro et al, 2002; Madhotra et al, 2003; Saeian et al, 2002). However, these results are based on a very small number of patients. Despite some initially promising findings, ultrathin endoscopy has largely fallen out of favor; at most centers, ultrathin unsedated endoscopy is not used for screening or surveillance of esophageal varices.

Esophageal capsule endoscopy is perhaps the most promising option for minimally invasive screening of esophageal varices. This technique involves the ingestion of a capsule videoscope that transmits images as it traverses the esophagus. The patient is supine after ingestion of the device and is gradually placed in the upright position, which increases the transit time in the esophagus to provide better imaging. Compared with conventional endoscopy, capsule endoscopy offers the advantages of direct visualization of the esophagus, the absence of sedation, and less procedural discomfort and time.

Several drawbacks are associated with capsule endoscopy. Interventions obviously cannot be performed, the device cannot be manipulated to change its field of view, and it cannot insufflate the esophagus, which most physicians have found to be a critical step in the proper diagnosis and staging of varices. Finally, capsule endoscopy does not properly visualize gastric varices, which could therefore be missed.

Capsule endoscopy has been studied in the detection and grading of esophageal varices, and de Franchis and colleagues (2008) reported a study comparing capsule endoscopy with conventional endoscopy in 288 patients. The capsule and esophageogastroduodenoscopy (EGD) findings were concordant in 86% of cases. However, capsule endoscopy failed to identify varices in 28 cases, of which 24 (13%) were small and 4 (2.2%) were large varices, although the capsule was significantly more preferred by patients over EGD. Lapalus and colleagues (2009) reported similar results in a cohort of 120 cirrhosis patients, with 77% sensitivity and 88% specificity in the detection of grade 2 and higher esophageal varices and/or red signs compared with upper endoscopy. Smaller studies comparing capsule endoscopy and EGD have yielded similar results (Eisen et al, 2006; Frenette et al, 2008; Lu et al, 2009; Pena et al, 2008). Models comparing cost effectiveness of esophageal capsule endoscopy and EGD imply that they are equivalent strategies (White & Kilgore, 2009).

In summary, upper endoscopy remains the gold standard in screening for the presence and staging of esophageal varices. None of the imaging modalities mentioned above has proven sufficiently reliable to supplant the role of endoscopy in evaluation. However, capsule endoscopy may be used as an alternative, albeit a less effective, screening technique in the small number of patients who either cannot or will not undergo conventional endoscopic screening.

Primary Prophylactic Therapy for Esophageal Varices: Prevention of Initial Bleeding

Once identified, all medium and large varices should be treated. In addition, small varices associated with high-risk features for bleeding, such as decompensated liver disease and red wale markings, should also be treated (Table 75A.3). The primary treatment options are obliteration with EVL or reduction of portal pressure with oral pharmacologic agents. Portal pressure can be pharmacologically reduced in three ways: 1) diminishing the portal inflow with vasoconstrictors, 2) reducing the intrahepatic vascular resistance, or 3) reducing the resistance in the collateral circulation with vasodilators. The most commonly used pharmacologic agents are nonselective β-adrenergic blockers, namely propranolol or nadolol, which act as antagonists for both the β₁ and β₂ receptors; β₁ blockade reduces portal inflow by decreasing cardiac output, and β₂ blockade reduces portal inflow through vasoconstriction of the sphlanchnic circulation. In addition, venodilators such as isosorbide reduce portal pressure by decreasing intrahepatic and/or portocollateral resistance. Isosorbide also has a systemic hypotensive effect and reduces portal pressure by reducing inflow as much as it does by reducing resistance (Blei et al, 1987).

Table 75A.3 Primary Prophylactic Therapy for Esophageal Varices: Prevention of Initial Bleeding

CTP, Child-Turcotte-Pugh classification

The goal of pharmacologic therapy is to reduce portal pressure and thereby reduce variceal bleeding. A decrease in hepatic venous pressure gradient (HVPG) to less than 12 mm Hg essentially eliminates the risk of hemorrhage, and reductions less than 20% from baseline significantly reduce the risk of a first variceal bleed (Casado et al, 1998; Groszmann et al, 1990). Some patients are nonresponders to β-blockers, however, in terms of portal pressure reduction. Bendtsen and colleagues (1991) defined nonresponse as a less than 10% reduction in portal pressure after an oral dose of 80 mg of propranolol. Because their withdrawal returns the risk for bleeding to that of the untreated population, β-blockers should be continued for life (Abraczinskas et al, 2001).

Numerous published primary prophylaxis trials with nonselective β-blockers have been performed that include hundreds of patients. The pooled odds ratio (OR) of bleeding with β-blockers is almost half that of placebo: 0.54 (95% confidence interval [CI], 0.39 to 0.74; P < .001). The number of patients who need to be treated to prevent one adverse bleeding event is 10 (95% CI, 8 to 18). A study of the early trials reported a nonsignificant trend toward a lower mortality rate (OR, 0.75; 95% CI, 0.57 to 1.06) (D’Amico et al, 1999). Another recent meta-analysis similarly found no difference in mortality rate (Tripathi, 2007). Finally, propranolol has been shown to prevent acute and chronic bleeding from portal hypertensive gastropathy in a single-blind, randomized study (Pérez-Ayuso et al, 1991). β-Blockers are typically started at a low dose (propranolol 20 mg bid or nadolol 40 mg qd) and titrated either as maximally tolerated or for a reduction in pulse rate of 25%. However, side effects of β-blockers prevent their use in approximately 15% of patients, and another 15% are intolerant (Garcia-Pagan et al, 2001). In addition, without directly monitoring the effects of β-blockers with HVPG measurements, their clinical effect is difficult to determine because changes in portal pressure correlate poorly with pulse rate (Garcia-Tsao et al, 1986).

In addition to β-blockers, other pharmacologic agents have been evaluated in the prevention of a first variceal bleed, but the results have largely been disappointing. Merkel and colleagues (2000) performed an unblinded study to evaluate the combination of isosorbide mononitrate and nadolol and found the combination to be more effective in reducing bleeding, with only a small increase in side effects. However, a recent larger controlled trial reported no effect with treatment. A double-blind, placebo-controlled trial by Garcia-Pagan and colleagues (2003) composed of 349 patients found no difference in 2-year actuarial probability of variceal bleeding or survival, with adverse effects being more frequent in the combination group. Studies assessing nitrates used alone have failed to demonstrate efficacy (Angelico et al, 1997; Garcia-Pagan et al, 2001). Currently, clinical evidence is insufficient to support the use of nitrates alone or in combination with β-blockers for prophylactic therapy.

Numerous trials have compared endoscopic therapy with β-blockers in primary prophylaxis for variceal hemorrhage (De et al, 1999; Jutabha et al, 2005; Lo et al, 2004; Lui et al, 2002; Psilopoulos et al, 2005; Sarin et al, 1999; Schepke et al, 2004). Meta-analysis showed that variceal ligation significantly reduced the risk of a first bleed compared with propranolol (relative risk [RR], 0.57; 95% CI, 0.38 to 0.85), but there was no difference in terms of mortality rate (RR, 1.03; 95% CI, 0.79 to 1.36) (Khuroo et al, 2005). These data have led to an intense debate as to whether band ligation therapy should replace β-blockers as the first-line therapy for primary prophylaxis.

Proponents of β-blocker therapy point out that band ligation therapy is associated with an inherent risk of bleeding, which can result in death. In addition, there may be a bias toward positive results relative to endoscopic therapy in studies with shorter follow-up and fewer patients. In the analysis of studies that were larger or had longer follow-up—that is, more than 100 patients or follow-up longer than 20 months—no difference in outcomes was reported between the two therapies (Bosch & Garcia-Tsao, 2009). β-Blockers also have a track record of long-term safety, which cannot be said for band ligation therapy. Finally, endoscopic therapy requires ongoing procedures that may be more costly and have lower patient preference compared with pharmacologic therapy (Longacre et al, 2008).

The statements regarding prophylactic therapy from two separate organizations highlight this debate. The recommendations from the 2005 Baveno consensus conference are that β-blockers should be the first-choice treatment to prevent first variceal bleeding in patients with high-risk varices, and EVL should be offered to patients with contraindications or intolerance to β-blockers (de Franchis, 2005). The American Association for the Study of Liver Diseases (AASLD) practice guidelines indicate that nonselective β-blockers and EVL are safe and effective in patients with medium or large varices that have a high risk of hemorrhage (CTP classes B and C or variceal red wale markings on endoscopy). The choice of therapies may be left to physician preference, although β-blockers are the preferred therapy in patients with medium or large varices that have not bled and that are not at the highest risk of hemorrhage (CTP class A patients, no red signs). EVL should be considered in patients with contraindications, intolerance, or noncompliance to β-blockers. Finally, the addition of β-blockers administered with EVL has been evaluated in one study, which failed to demonstrate a reduction in the risk of first bleed or death (Sarin et al, 2005).

Secondary Prophylactic Therapy for Esophageal Varices (prevention of Rebleeding)

Patients who survive a first episode of variceal bleeding are at very high risk of recurrent bleeding (70%) and death (30% to 50%), with the highest risk period occurring within 6 weeks of the index bleed (D’Amico & de Franchis, 2003; Grace et al, 1998). There is consensus that all patients who have previously bled from varices should have secondary therapy to prevent further variceal bleeding (Box 75A.1; Garcia-Tsao et al, 2007). Severity of liver disease (Pagliaro et al, 1994), continued alcohol abuse (Vorobioff et al, 1996), and variceal size have all been associated with variceal rebleeding. In addition, the inability to reduce HVPG to less than 12 mm Hg or achieve a less than 20% reduction from baseline significantly increases the risk of rebleeding. The treatment of acute variceal bleeding is covered in Chapter 75B.

After control of acute bleeding, prevention of recurrent hemorrhage is considered mandatory and should be initiated as soon as the patient is stable. Numerous trials have been performed on β-blockers compared with placebo or no treatment, and nonselective β-blockers have been found to reduce the risk of subsequent bleeding from 63% to 42% and to lower overall mortality rates from 27% to 20% (Bernard et al, 1997; D’Amico et al, 1999; Garcia-Pagan et al, 2008).

The addition of isosorbide to β-blocker therapy has also been evaluated for the prevention of rebleeding. These studies had their impetus from an article that stated approximately one third of hemodynamic nonresponders to β-blockers become responders after the addition of isosorbide (Garcia-Pagan et al, 1991). One of the two trials showed reduced rates of bleeding (Gournay et al, 2000), and one did not show reduced bleeding (Pasta et al, 2001, abstract). Data gathered from other trials compared this regimen with endoscopic therapy and showed that the combination of β-blockers with isosorbide is approximately 33% more effective than β-blockers alone in the prevention of rebleeding (Bosch & Garcia-Pagan, 2003; D’Amico et al, 1999). Therefore the best possible pharmacologic regimen is the combination of β-blockers and nitrates, although in clinical practice this regimen is associated with increased side effects; thus usually only β-blockers are used (Garcia-Tsao et al, 2007). This is especially true for patients with decompensated cirrhosis who are awaiting liver transplantation.

Trials have been conducted comparing pharmacologic therapy with endoscopic treatment in the prevention of rebleeding. The only modern endoscopic therapy for bleeding esophageal varices is band ligation. Sclerotherapy no longer plays a role in treatment because band ligation has clearly been demonstrated as the more efficacious therapy (Laine & Cook, 1995; Garcia-Pagan & Bosch, 2005; Garcia-Pagan et al, 2008) Therefore, although numerous studies have evaluated sclerotherapy in the prevention of rebleeding, these are largely of historic interest only. In general, after an acute variceal bleed, the patient is treated emergently with band ligation therapy and then subsequently receives recurrent EVL until the varices are completely obliterated, which usually requires three to four treatment sessions. Optimal pharmacologic therapy (β-blockers plus nitrates) has been compared with EVL in four published randomized studies with variable results (Lo et al, 2002; Patch et al, 2002; Romero et al, 2006; Villanueva et al, 2001). One study showed a benefit of pharmacologic therapy, another showed a benefit of EVL, and two showed no difference between the two therapies. A meta-analysis (Ding et al, 2009) of these four studies, which included 476 patients, found no significant difference in the rate of rebleeding, overall mortality rate, or complications between β-blocker plus isosorbide and EVL. Therefore both therapies appear to be effective in the prevention of esophageal variceal rebleeding. A more practical approach may be the combination of β-blocker and EVL because endoscopic therapy would obliterate the varices and recurrence would be prevented by β-blockers. Two studies have compared EVL and EVL plus β-blockers, both of which showed that the addition of β-blockers to EVL reduces the risk of rebleeding and variceal recurrence (de la Peña et al, 2005; Lo et al, 2000). Finally, a recent study by Garcia-Pagan and colleagues (2009) compared nadolol plus isosorbide alone or in combination with EVL. These investigators found that both therapies were equally effective in the prevention of rebleeding, but the patients who received EVL and combination drug therapy were more likely to require rehospitalization. They concluded that the addition of EVL to drug therapy was therefore was less efficacious than drug therapy alone.

Four randomized trials have compared TIPS (see Chapter 76E) with EVL, drug therapy, or the combination for the prevention of variceal rebleeding (Escorsell et al, 2002; Gülberg et al, 2002; Pomier-Layrargues & Villenueve, 2001;Sauer et al, 2002). In general, the side-effect profile was worse with TIPS—two studies found significantly more encephalopathy with TIPS, and two found no difference—but generally better control of bleeding was reported with TIPS, with significantly less rebleeding in two studies and no difference in two others. Most importantly, both these studies found no difference in survival rates. Therefore, because TIPS offers no survival advantage and has a worse side-effect profile, it is relegated to salvage therapy for patients with refractory variceal bleeding that is unresponsive to pharmacologic and EVL treatment.

In summary, the current recommendations for the prevention of variceal rebleeding suggest that all patients receive prophylactic therapy to prevent recurrent variceal hemorrhage. The recommended therapy is β-blockers and EVL (Garcia-Tsao et al, 2007; Gonzalez, 2008), although another consensus statement recommends β-blockers plus nitrates or EVL (de Franchis, 2005). TIPS is recommended only in patients with recurrent variceal bleeding refractory to pharmacologic and endoscopic therapy.

Finally, portal hypertensive gastropathy is a recognized cause of upper intestinal bleeding in patients with cirrhosis. The natural history of portal hypertensive gastropathy is that it worsens with increasing severity of liver disease and portal hypertension; acute bleeding is infrequent. Chronic anemia is more common with mild portal hypertensive gastropathy (Merli et al, 2004; Primignani et al, 2000). Propranolol has been shown to prevent acute and chronic bleeding in a single-blind, randomized study (Pérez-Ayuso et al, 1991). The actuarial percentages of patients free of rebleeding from portal hypertensive gastropathy at 12 months were 65% versus 38% (P < .05); at 30 months, the figures were 52% versus 7% (P < .05). In addition, fewer episodes of acute rebleeding were reported (mean episodes per patient per month were 0.01 vs. 0.12).

Hemodynamic Monitoring with Hepatic Venous Pressure Gradient

HVPG is equal to the wedge hepatic venous pressure minus free hepatic venous pressure. Some centers have advocated monitoring of HVPG to assess target reductions of portal pressure because HVPG measurement is the most reproducible and reliable method of assessing the portal pressure in cirrhosis (Groszmann & Wongcharatrawee, 2004). Wedge hepatic venous pressure is equivalent to portal venous pressure in cirrhosis when the major site of resistance is sinusoidal or presinusoidal (alcoholic and viral hepatitis B and C) or both. A decrease in HVPG of 20% or more from baseline or to 12 mm Hg or less has been shown to correlate with a significant reduction in bleeding. Therefore HVPG measurement offers the opportunity to tailor a patient’s reponse to pharmacologic therapy based on changes in portal pressure. Ideally, HVPG could be measured before the initiation of pharmacologic therapy.

After the maximal tolerated dosage is achieved, HVPG could be remeasured. For patients responsive to pharmacologic therapy, no further treatment or adjustment in dosing would be required. However, nonresponders—that is, patients with portal pressure higher than 12 mm Hg or with a less than 20% decrease from baseline—could receive additional pharmacologic therapy. Some data suggest that this approach is efficacious (Bureau et al, 2002; Feu et al, 1995; McCormick et al, 1998; Patch et al, 2002; Villanueva et al, 1996, 2001). However, titration of medication requires multiple invasive procedures to measure HVPG, and some investigators have suggested that patients with early rebleeding are not available for repeated studies; therefore these high-risk patients are excluded from analysis (Thalheimer et al, 2004). Undoubtedly a valuable tool, the applicability of HVPG needs further prospective evaluation in future studies (Bosch et al, 2009; Garcia-Pagan et al, 2008).