Portal hypertension and variceal bleeding—Unresolved issues. Summary of an American Association for the study of liver diseases and European Association for the study of the liver single-topic conference

Many randomized controlled trials (RCTs) have advanced the knowledge of the complications of portal hypertension, specifically in the management of varices and variceal hemorrhage. The endpoints in most of these trials have used definitions attained at international consensus workshops that have been published in medical journals since the first Baveno conference in 1992.1 Three more international consensus conferences have helped further define clinical endpoints and practice recommendations.2-5 This article summarizes the results of an Endpoints Single Topic Conference on “Portal Hypertension and Variceal Bleeding—Unresolved Issues” that took place in Atlanta, GA, in June 4-6, 2007 and that was sponsored jointly by the American Association for the Study of Liver Diseases (AASLD) and the European Association for the Study of the Liver and constitutes the 6th international consensus conference in the area of varices and variceal hemorrhage. As outcomes have improved and knowledge has been gained, it is necessary to prioritize areas that require further research, to define surrogate markers of outcome and to stratify patients in different risk groups. These were the objectives of this single-topic conference. For areas in the management of varices and variceal hemorrhage in which it was decided that no further trials were necessary or plausible, practice recommendations were put forward. In some instances, these differed from recommendations put forward at the most recent consensus conference in Baveno 20045 and were incorporated into the recently published AASLD/American College of Gastroenterology (ACG)-sponsored recommendations.6, 7 Recommendations were obtained by majority agreement defined as agreement by greater than 70% of 23 expert participants (listed at the end of the article). There are two main stages in the natural history of cirrhosis: compensated and decompensated cirrhosis, defined by the absence or presence of ascites, variceal hemorrhage, encephalopathy, or jaundice.8, 9 Both entities differ in clinical presentation, outcome, mortality, and predictors of death. Compensated cirrhosis has a median survival of more than 12 years (while remaining in the compensated stage), which is significantly longer than that of decompensated patients (approximately 2 years). Compensated patients die mostly after decompensation or of causes unrelated to liver disease, whereas in patients with decompensated cirrhosis mortality is mostly liver-related. Transition from a compensated to a decompensated stage is the most common outcome in patients with compensated cirrhosis and occurs at a rate of 5%-7% per year.8, 10 Although the Child-Turcotte-Pugh (CTP) score (or its components) is the most robust predictor of death in cirrhosis (compensated or decompensated), other prognostic markers differ, depending on the stage of cirrhosis.9 In compensated cirrhosis, markers of portal hypertension (varices, platelet count, spleen size, gamma-globulins) are predictive of death, whereas in decompensated cirrhosis, markers of circulatory dysfunction (such as renal dysfunction) and hepatocellular carcinoma are more predictive. More recently, four different stages have been proposed based on 1-year mortality data in a large cohort of untreated patients.11 The first two stages occur in compensated cirrhosis and are defined by the absence (stage 1) or presence of varices (stage 2). The other two stages occur in decompensated cirrhosis and are defined by the presence of ascites (with or without varices, stage 3) and by variceal hemorrhage (with or without ascites; stage 4). The 1-year mortality in these four stages is 1%, 3%, 20%, and 57%, respectively. Assessment of portal pressure by the hepatic venous pressure gradient (HVPG) has been a useful predictor of outcomes in both stages. In patients with compensated cirrhosis, an HVPG greater or equal to 10 mmHg is the most important predictor of the development of varices12 and clinical decompensation.13 This HVPG level has been designated “clinically significant portal hypertension.”14 Furthermore, in posttransplantation recurrent hepatitis C, HVPG has a higher diagnostic accuracy than liver biopsy in predicting decompensation.15 In decompensated cirrhosis, HVPG obtained at the time of variceal hemorrhage predicts outcome16 and is an independent predictor of death in a model adjusted by MELD (model for end-stage liver disease), ascites, and age.17 An HVPG of 20 mmHg appears to be the best cutoff in this patient population. Compensated and decompensated cirrhosis should be considered two separate entities both in clinical practice and in clinical research Studies in cirrhosis should consider and analyze these entities separately HVPG >10 mmHg is the best predictor of variceal development and decompensation and should be used to stratify patients with compensated cirrhosis in clinical trials Substaging of patients with cirrhosis requires validation in prospective cohort studies ACG, American College of Gastroenterology; AASLD, American Association for the Study of Liver Diseases; CE, capsule endoscopy; CTP, Child-Turcotte-Pugh; EGD, esophagogastroduodenoscopy; EVO, endoscopic variceal obturation; HVPG, hepatic venous pressure gradient; LS, liver stiffness; NSBB, nonselective beta-blockers; RCT, randomized controlled trials; TIPS, Transjugular intrahepatic portosystemic shunt. Current guidelines recommend screening esophagogastroduodenoscopy (EGD) when the diagnosis of cirrhosis is made so that effective prophylaxis of variceal hemorrhage can be applied.5-7 An alternative to EGD is capsule endoscopy (CE), which, in pilot studies, had a negative predictive value of 57%-100%18-20 and may have an acceptable budget impact.21 Preliminary results of a multicenter study comparing CE versus EGD in 288 patients (180 with varices by EGD), showed a good agreement in detecting both the presence (86%) and size (77%) of esophageal varices.22 However, 23% of patients with negative CE had varices on EGD and, of 79 patients with medium/large varices on EGD, 17 (22%) were diagnosed as having no or small varices on CE. There was greater preprocedure acceptance and postprocedure satisfaction for CE. Capsule endoscopy requires further study to better define its reproducibility, reliability, and accuracy, as well as patient preference and cost-effectiveness, but may be a reasonable alternative to EGD in patients unable or unwilling to undergo EGD A large multicenter placebo-controlled study showed that nonselective beta-blockers (NSBBs) are not effective in preventing the development of varices in patients with portal hypertension (HVPG > 6 mmHg), and its use is associated with a higher rate of side effects.12 Most agreed that it is currently unnecessary to perform similar trials. RCTs aimed at preventing first variceal hemorrhage in patients with small varices will require a huge sample size and therefore should be discouraged. Because NSBBs may have other beneficial effects via reduction in portal pressure, RCTs combining endpoints such as an increase in HVPG, variceal growth, cirrhosis decompensation, and death may be more appropriate. RCTs of endoscopic variceal ligation (EVL) in this setting also should be explored. Given that high-risk small varices bleed at a similar rate to large varices and that growth of non–high-risk small varices has been shown in an RCT to be slower in patients treated with nadolol,23 there was majority agreement in recommending that patients with small varices and a high risk of bleeding (CTP class C or red signs) should receive NSBB and that in other patients with small varices NSBB are optional; if NSBB are given, repeat EGD is not necessary, whereas if NSBB are not given, EGD should be repeated in 2 years. These recommendations are different from those proposed in the last Baveno conference5 and were incorporated in the AASLD/ACG guidelines.6, 7 Two therapies are effective in these patients: NSBB and EVL. Two published meta-analyses of 12 trials comparing them show that EVL is associated with a significantly lower incidence of first variceal hemorrhage but without differences in mortality.24, 25 The results of a third Cochrane meta-analysis are awaited. The most recent Baveno conference considered that, given a lack of survival benefit and uncertainty regarding the long-term benefits of EVL, NSBB should be first-line therapy and that EVL should be offered to patients who have contraindications or intolerance to NSBB.5 Data beyond these meta-analyses were reviewed at the single-topic conference, showing that analysis of six high-quality RCTs show no difference in first hemorrhage between therapies26; that side effects are more severe with EVL, as there have been three reported deaths secondary to bleeding from post-EVL ulcers27, 28; that EVL was cost-effective when cost per quality-adjusted life-year is considered29; and that predicted preferences are higher for EVL in both patients and physicians.30 Advantages of NSBB such as prevention of bleeding from other portal hypertension sources (portal hypertensive gastropathy and gastric varices) and possible reduction in the incidence of spontaneous bacterial peritonitis,31 and other complications of portal hypertension, were also discussed. Nevertheless, most agreed that both NSBB and EVL are effective therapies and that the clinical decision depends on patient characteristics and preferences, local resources, and expertise. These recommendations differ from those put forward in the Baveno conference and were incorporated in the AASLD/ACG guidelines.6, 7 Further studies comparing NSBB versus EVL are not justified. If a new treatment becomes available, it should be compared with either NSBB or EVL and not with a placebo or an untreated group. Studies of NSBB plus EVL have been inconclusive,32, 33 and, given a perceived unacceptable risk–benefit ratio with combined therapy, most agreed that combination EVL + NSBB is not recommended in this setting.6, 7 In patients with no varices and portal hypertension (HVPG ≥6 mmHg), the best predictor of the development of varices is a baseline HVPG greater than 10 mmHg.12) Furthermore, an HVPG reduction greater than 10% at 1 year was associated with a significantly lower incidence of varices.12 Interestingly, even though there was no overall difference in the development of varices between patients treated with NSBB and with placebo, a greater proportion of patients randomized to NSBB achieved a reduction in HVPG greater than 10% (53% on NSBB vs. 38% on placebo), indicating that a drug that could increase the proportion of patients that achieve this HVPG reduction would probably be effective in preventing varices. In patients with varices (mostly medium/large) that have never bled, reductions greater than 10%-20% reduce the incidence of first variceal hemorrhage, and patients in whom the HVPG decreases below 12 mmHg are essentially protected from bleeding.34-36 In this patient population, a reduction in HVPG of more than 15% has been associated with a reduction in spontaneous bacterial peritonitis.31 Unless a new and effective therapy becomes available, further trials of preprimary or primary prophylaxis with existing therapies are unnecessary In studies of prevention of varices (preprimary prophylaxis), only patients with an HVPG greater than 10 mmHg should be included In studies of prevention of variceal hemorrhage (primary prophylaxis), new pharmacological treatments should be compared with NSBB in a double-blind design and should include HVPG measurements. In patients with small varices, combining endpoints will be necessary In RCTs using EVL, because blinding is not applicable, treatment outcome and adverse events should be predefined and assessed by physicians blinded to trial treatment Currently, it is recommended that short-term antibiotic prophylaxis, a measure that reduces bacterial infections,37 variceal rebleeding,38 and death37 be used in every patient with cirrhosis admitted with gastrointestinal hemorrhage.5-7 Different antibiotics have been used in different trials and, given different local antibiotic susceptibility patterns and different availability it is unlikely that a definitive trial in this area will be performed. Specific therapy is based on the combination of pharmacological and endoscopic therapy, which is better than either treatment alone,39-41 particularly with early administration of pharmacological therapy.42, 43 No RCTs compare different combinations of endoscopic/pharmacological therapy. RCTs comparing different pharmacological agents (vasopressin, somatostatin, terlipressin, octreotide) demonstrate no differences among them regarding control of hemorrhage and early rebleeding, although vasopressin is associated with more adverse events.40 In practice, the choice of pharmacological agent is usually based on availability and cost. The optimal duration of pharmacological therapy has not been well established. In RCTs, the duration of vasoactive treatment has varied between 8 hours42 and 6 days. Trials aimed at determining the best duration of therapy are impractical and costly. There was majority agreement that an appropriate length of therapy would be anywhere between 2 and 5 days,6, 7 depending on control of hemorrhage and the presence or absence of predictors of rebleeding (for example, CTP class, HVPG). EVL is more effective than endoscopic variceal sclerotherapy (EVS) with greater control of hemorrhage, lower rebleeding, and lower adverse events but without differences in mortality.44, 45 No further trials are necessary to determine the best endoscopic therapy. Prospective cohort studies in which HVPG has been measured within 48 hours of admission for hemorrhage show that levels greater than 20 mmHg are associated with increased rebleeding and mortality.16, 44, 46-48 A more recent study performed in the era of combined vasoactive drug plus endoscopic therapy confirms this HVPG cutoff and shows that an index including CTP score and blood pressure at admission has similar prognostic value.49 Furthermore, a drug-induced HVPG reduction of less than 10% predicts 5-day failure. This response may improve by doubling the dose of somatostatin or switching to another agent (such as terlipressin).50 Transjugular intrahepatic portosystemic shunt (TIPS) is a reasonable alternative in the face of failure of combined pharmacologic plus endoscopic therapy. In the Baveno conference, it was considered that a second attempt at endoscopic therapy was one possibility but that one could perform TIPS after failure of the first endoscopic therapy.5 A small study suggests that early TIPS placement (within 24 hours of hemorrhage) is associated with a significant improvement in survival in patients with an HVPG greater than 20 mmHg.48 Therefore, HVPG can provide useful information that allows for risk stratification and more aggressive treatment in high-risk patients. Risk-stratification of patients presenting with acute variceal hemorrhage is necessary to better assess duration of pharmacological therapy and to determine whether different treatment strategies (such as early TIPS in high-risk patients) are HVPG greater than 20 mmHg measured within 48 hours of is the best predictor of outcome markers of a outcome have been and require validation TIPS therapy in high-risk patients should be further The most recent Baveno consensus conference recommended that patients with cirrhosis who had not primary prophylaxis could receive NSBB, EVL, or Although it was that combination of NSBB plus EVL was probably the best it that more trials were a could be EVL plus NSBB is a because NSBB will rebleeding variceal and can variceal In two small RCTs demonstrate the of combined therapy versus EVL on this there was majority agreement that, as incorporated in AASLD/ACG the best in the prevention of recurrent esophageal variceal hemorrhage is the combination of NSBB plus 7 Trials of combination pharmacological therapy plus versus EVL have shown no differences in rebleeding or death. A recent RCT comparing EVL + combination drug therapy + versus combination drug therapy showed no differences in overall rebleeding or survival but a lower incidence of variceal rebleeding in the EVL indicating that the benefit of EVL in variceal hemorrhage was by an increase in bleeding from who a reduction of HVPG below 12 mmHg are protected from whereas those that reduce it by more than from baseline have a risk of rebleeding compared with that repeat are performed within from within 1 to of patients treated with NSBB are HVPG Therefore, and a therapy becomes that will significantly increase the proportion of HVPG it would that HVPG should this In a study the of three different strategies in the setting of secondary combination pharmacological therapy with HVPG was more cost-effective than EVL or combination drug therapy without HVPG, and its improved with of a there have been no RCTs comparing an therapy versus therapy, and such a trial is The between EVL and the EGD is and it is considered unnecessary to perform RCTs to these Given data from RCTs of and EVL in RCTs of versus EVL, it appears that the the between EVL the the of varices. EGD, were used in RCTs two of them EGD was also performed 1 after and in a of it was to at 6 without differences in recurrent bleeding compared with trials that Most will the first after the second and at to depending on variceal This majority was incorporated into the AASLD/ACG guidelines.6, 7 TIPS has shunt in most patients and is the main in patients in whom first-line therapy has in most However, a large multicenter trial of TIPS versus shunt shows similar of rebleeding, encephalopathy, and mortality in patients with CTP class cirrhosis who had therapy, with a higher rate of shunt dysfunction in the TIPS because TIPS were Because both have is an in when is Given this and the that there are or there is to TIPS and lower availability of most agreed that in shunt is but that further trials are not therapy should be further its on and on outcomes pharmacological treatments should be compared with NSBB in a double-blind design and should include HVPG A surrogate outcome measure is a a or another that is to an on a outcome occur or more are and are achieved less than the clinical outcomes be The first in validation is to demonstrate a between the surrogate and the clinical The second is to whether the on the surrogate outcome predicts the on the clinical As HVPG predicts and in HVPG that occur time are negative predictors of development of the risk of variceal the development of complications of portal and pressure measured by the HVPG may be as as can to a surrogate outcome measure in In HVPG could be a surrogate not only in clinical trials for portal hypertension but also in trials of liver and liver disease), as in a recent study performed in posttransplantation recurrent hepatitis C in which HVPG was more in predicting the development of decompensated cirrhosis than was liver The cutoff levels that would define in RCTs on prevention of varices would be to achieve a in HVPG of greater than 10% from to less than 10 In trials of prevention of variceal hemorrhage, the outcome would be to the HVPG to less than 12 mmHg or to achieve a of greater than from although it would that for studies of primary prophylaxis of variceal bleeding a of greater than 10% from baseline would be In secondary prophylaxis, the second should be performed as as possible after an optimal dose and no than 1 after the first because one third of rebleeding will occur within this time studies have that an acute of HVPG response to at the time of the first HVPG is predictive of long-term response and but this will require further The is and guidelines are has been assessed the analysis of studies in which baseline and repeat HVPG were performed between 20 and from showing a of only between measurements. In a recent study showed an agreement in The overall rate in performed in a between and was only with large or the most There appears to be a were and after No deaths have been differences are between in the and those in the is used and is performed by In the HVPG are performed by mostly for clinical with a and a and a is used by only and use and not Most agreed that there is a for of the the of guidelines as well as and of the of HVPG measurements. is a and for liver Prospective studies show that are useful in the of portal hypertension (HVPG ≥6 has been shown to with HVPG in hepatitis and in However, although this is for HVPG, less than 10 or less than 12 mmHg and for mmHg or mmHg significant portal the is not optimal and the between HVPG and to differ depending on the of and the cutoff value that best significant portal hypertension is not well established. blinded prospective studies in patients with cirrhosis of causes are HVPG is the best surrogate in portal hypertension trials and should be measured in every trial pharmacological therapy HVPG and of liver and should be measured in trials therapies in which of is an Although its prognostic value in cirrhosis has been prognostic studies of HVPG in other causes of cirrhosis are is a to significant portal hypertension to assess response are necessary varices occur in of patients with portal varices are the of gastric varices with bleeding and rebleeding large varices may bleed HVPG less than 12 NSBB are recommended as primary prophylaxis gastric variceal to control hemorrhage and recurrent hemorrhage include endoscopic and data comparing these therapies in bleeding varices show that the best control of hemorrhage is achieved with TIPS, or Three small RCTs compare endoscopic variceal with versus or EVL in bleeding gastric 79 three RCTs are for regarding control of acute or less than of the patients included in these studies had varices, and a separate analysis was not performed. is recommended that TIPS be used in acute bleeding from varices when is or if rebleeding occurs after this has not been A small study comparing versus TIPS in the prevention of recurrent hemorrhage in patients in whom acute gastric variceal hemorrhage was controlled with showed similar rebleeding but than of the patients were bleeding from Studies on gastric varices should on patients with varices Trials of TIPS versus are The most common causes of portal hypertension in are and portal Recommendations from with cirrhosis be to patients. may require different depending on the of the portal characteristics of such as differences in rate and for for such as EGD and HVPG the of RCTs in There are data from studies using NSBB in or endoscopic therapy versus in portal that included patient EVL appears to be a reasonable in acute variceal hemorrhage and for preventing hemorrhage but it has not been compared with However, EVL is not in than 1 year In with portal would to be the best for secondary Although RCTs are in prospective cohort studies in defined risk (compensated or with or without variceal hemorrhage) are therapies such as NSBB require further This of portal hypertension is mostly to portal and occurs mostly in patients with an There are RCTs on therapies and primary and secondary prophylaxis as well as and Given the large of patients for an RCT of this most agreed that useful therapies to be those in with cirrhosis, specifically and endoscopic with shunt or and TIPS as on cohort studies, the use of in patients with a history of history of more than one history of or has been associated with a reduction in complications and and improved Prospective cohort studies in this area are The were the of this In the expert that, with the in the and the recommendations of of and of of of of of