Variceal ligation plus nadolol compared with ligation for prophylaxis of variceal rebleeding: A multicenter trial


  • Conflict of interest: Nothing to report.

  • Preliminary results of the study were presented at the Annual Meeting of the Digestive Disease Week, May 17-22, 2003, Orlando, FL, and published as an abstract.38


β-Blockers and endoscopic variceal ligation (EVL) have proven to be valuable methods in the prevention of variceal rebleeding. The aim of this study was to compare the efficacy of EVL combined with nadolol versus EVL alone as secondary prophylaxis for variceal bleeding. Patients admitted for acute variceal bleeding were treated during emergency endoscopy with EVL or sclerotherapy and received somatostatin for 5 days. At that point, patients were randomized to receive EVL plus nadolol or EVL alone. EVL sessions were repeated every 10 to 12 days until the varices were eradicated. Eighty patients with cirrhosis (alcoholic origin in 66%) were included (Child-Turcotte-Pugh A, 15%; B, 56%; C, 29%). The median follow-up period was 16 months (range, 1–24 months). The variceal bleeding recurrence rate was 14% in the EVL plus nadolol group and 38% in the EVL group (P = .006). Mortality was similar in both groups: five patients (11.6%) died in the combined therapy group and four patients (10.8%) died in the EVL group. There were no significant differences in the number of EVL sessions to eradicate varices: 3.2 ± 1.3 in the combined therapy group versus 3.5 ± 1.3 in the EVL alone group. The actuarial probability of variceal recurrence at 1 year was lower in the EVL plus nadolol group (54%) than in the EVL group (77%; P = .06). Adverse effects resulting from nadolol were observed in 11% of the patients. In conclusion, nadolol plus EVL reduces the incidence of variceal rebleeding compared with EVL alone. A combined treatment could lower the probability of variceal recurrence after eradication. (HEPATOLOGY 2005;41:572–578.)

Variceal bleeding is a serious complication of portal hypertension and a leading cause of death in patients with liver cirrhosis. The patients who survive an initial episode have a risk of rebleeding approaching 80% at 2 years.1 After the first episode of hemorrhage from esophageal varices, β-blockers and endoscopic treatment are effective therapies in the secondary prevention of variceal bleeding.2 Unfortunately, nonselective β-blockers achieve target reductions in hepatic venous pressure gradient in only approximately one third of patients.3–5 Moreover, endoscopic injection sclerotherapy (EIS) has been replaced almost universally by endoscopic variceal ligation (EVL), because variceal eradication is faster and provides a lower variceal rebleeding rate with fewer secondary effects than in EIS.6–13 In addition, a meta-analysis has shown a lower mortality in EVL than in EIS.14

Some studies comparing band ligation with β-blockers plus isosorbide mononitrate show contradictory results regarding the effectiveness of both treatments in the prevention of variceal rebleeding.15–17

The rationale behind the use of a combination therapy is that agents acting through different mechanisms may have additive or even synergistic benefits.5 There are few studies analyzing the efficacy of associating both therapeutic methods, β-blockers and EVL, in the prevention of recurrent hemorrhage after variceal bleeding.18 The goals are first, to reduce the risk of rebleeding during the course of initial endoscopic therapy, because this is the stage with the highest rate of rebleeding1; and second, to decrease the risk of recurrent varices and rebleeding after initial endoscopic therapy is completed.

The aim of this study was to assess the efficacy of a combination therapy between a portal hypotensive agent such as nadolol and EVL versus EVL alone for the secondary prophylaxis of variceal bleeding.


EIS, endoscopic injection sclerotherapy; EVL, endoscopic variceal ligation.

Patients and Methods


This prospective, randomized, controlled study was conducted in four hospitals in Spain. These centers are the reference hospitals for four different regions: 535,000 (Santander), 390,000 (Sabadell), 340,000 (Ourense), and 160,000 (Burgos) inhabitants.


From June 1999 through October 2003, all emergency ward patients with upper gastrointestinal bleeding (hematemesis, melena, and fall of hemoglobin values) were included in a standard care protocol. This protocol included the following measures: confirmation of upper gastrointestinal bleeding, review of the complete medical history, intravenous access, determination of hemodynamic status (pulse and blood pressure), placement of gastric aspiration, and basic analytical workup. Volume replacement was initiated immediately in all patients with hemodynamically significant upper gastrointestinal bleeding (fall of systolic arterial pressure to less than 100 mm Hg and cardiac rate of more than 100 beats/min). Associated disorders were treated when necessary. All patients underwent upper gastrointestinal endoscopy within 12 hours of admission. Endoscopy was performed under continuous monitoring of pulse rate, oxygen saturation, and blood pressure.

Inclusion Criteria.

Patients were considered for enrollment in the study after fulfilling the following criteria: (1) age between 18 and 75 years; (2) diagnosis of cirrhosis on the basis of previous liver biopsy or clinical, biochemical, and ultrasound findings; and (3) bleeding from esophageal varices or esophagogastric type gastroesophageal varices 1 (gastric varices 2 to 5 cm below gastroesophageal junction as continuation of esophageal varices). Bleeding was considered of variceal origin when the emergency endoscopy showed any of the following: active bleeding from an esophageal varix, stigmata of recent hemorrhage over varix (clot adherent), or when there was no other cause of upper gastrointestinal bleeding and fresh blood was found in the stomach.

Patients fulfilling the inclusion criteria and presenting active variceal bleeding or stigmata of a recent hemorrhage were treated with EVL or EIS during emergency endoscopy.

Exclusion Criteria.

Exclusion criteria were: (1) declining to participate in the study; (2) bleeding from fundal varices (gastroesophageal varices 2); (3) portal vein thrombosis; (4) known hepatocellular carcinoma; (5) contraindications to β-blocker therapy (chronic obstructive pulmonary disease, asthma, aortic stenosis, atrioventricular block, intermittent claudication, insulin-dependent diabetes mellitus); (6) previous endoscopic scheduled treatment (EVL or EIS) in the previous 12 months; (7) previous portosystemic shunt; (8) pregnancy; (9) refractory ascites19; or (10) American Society of Anesthesia classification stages IV or V.20


On the fifth day, the patients who met the inclusion criteria were invited to participate in the study. After obtaining informed written consent, they were stratified according to the Child-Turcotte-Pugh score (A and B versus C class) and randomized to receive EVL plus nadolol or EVL alone by opening an opaque, sealed envelope containing the allocation scheme. Implementation of random allocation was performed by sequences of 10 numbers obtained from a computer. Each hospital consecutively included patients according to the list of numbers attributed to the center, being impossible for two patients to have the same number allocation.


After endoscopic emergency treatment, all patients received somatostatin (250 μg/h) for 5 days. During this period, a complete clinical evaluation was carried out, as well as laboratory tests, an abdominal ultrasound, and an electrocardiogram.

The first EVL session was carried out in the initial emergency endoscopy in those patients with active bleeding and in those with a clot adherent on varix, in the emergency endoscopy or during the 24 hours after index bleeding. In both groups, EVL sessions were scheduled every 10 to 12 days until variceal eradication was achieved. EVL sessions were performed under conscious sedation with midazolam (0.035–0.05 mg/kg) and continuous monitoring of oxygen saturation and pulse rate. EVL was performed using multiband devices: Six Shooter-Saeed Multi-Band Ligator (Wilson-Cook Medical GI Endoscopy, Winston-Salem, NC) and speed band (Microvasive Endoscopy; Boston Scientific Corp., Natick, MA). EVL was performed beginning at the cardia or immediately under it, and each varix was ligated as many times as necessary to make it no longer visible.

Nadolol (Sanofi-Synthelabo, Barcelona, Spain) was given continuously, with an initial dose of 40 mg once daily and then in increasing doses until a 25% decrease in heart rate was achieved.


After variceal eradication, endoscopic controls were performed in both groups at 3-, 6-, and 12-month intervals and then annually, with further treatment if new varices appeared. Clinical follow-up was carried out in the outpatient clinic at 1-, 3-, 6-, 12-, 18-, and 24-month intervals. An abdominal ultrasound study was carried out twice yearly.


The primary endpoint of the study was the rate of variceal rebleeding after intervention. Recurrent variceal bleeding was considered if the patient had hematemesis, melena, and fall of hemoglobin values and active bleeding or stigmata of recent hemorrhage was seen from a varix or from a scar of previous ligation. Blood transfusion was not necessary to define recurrent variceal bleeding. Secondary endpoints included the following:

  • 1Mortality rate
  • 2Eradication of varices, defined as nonendoscopic visualization of veins in the lower third of the esophagus.
  • 3Recurrence of esophageal varices, defined as the observation of new varices after eradication had been achieved. Variceal size was classified in four grades (I to IV). Gastric varices were classified according to the Sarin's system of gastroesophageal varices 1 and gastroesophageal varices 2.21
  • 4Complications of pharmacological and endoscopic treatments were defined according to the Baveno III consensus workshop.22
  • 5Treatment failure, defined as the inability to control active bleeding after two attempts with the same method, recurrence of bleeding twice in a 3-month period, death related to bleeding or complications, or the need to place a transjugular intrahepatic portosystemic shunt to control acute or recurrent bleeding.

Sample Size Determination.

The study was designed to compare the risk for variceal recurrent bleeding. The actuarial risk for recurrent bleeding at 2 years is approximately 31% for EVL.13 No figures were available for EVL plus nadolol at the time this study was designed. This study hypothesized a reduction of the risk for recurrent variceal bleeding by EVL plus nadolol at 16%. According to the sample size calculation, the study would require 85 patients in each group. An interim analysis was planned when half of the sample was included without a predetermined stopping rule. In this interim analysis, we found relevant statistical differences between the two treatment groups (P = .006) in the primary endpoint, very close to the .005 required if the O'Brien-Fleming method had been used at the study planning stage.

Statistical Methods.

Student t test or the Mann-Whitney U test were used for quantitative variables, and the chi-square test was used for qualitative ones. The Kaplan-Meier method was used to estimate the time to event for mortality, first episode of recurrent bleeding, and variceal recurrence. The log-rank test was used to compare resulting curves for the two treatment groups. Analysis was carried out using the intention-to-treat method. Differences were considered significant when the P value was less than .05. Calculations were performed using the SPSS statistical software package (SPSS Inc., Chicago, IL).


The study was approved by the research ethics committee of each participant center.


The results shown below correspond to those found at interim analysis (after half of the calculated sample was included). Although a post hoc allocation of α is not an appropriate method to conclude the study, the finding of relevant statistical differences between the two treatment groups obliged us to finalize the study at this point for ethical reasons.

During the study period, 240 patients with liver cirrhosis with acute variceal bleeding were encountered. A total of 131 patients were excluded (causes of exclusion are shown in Fig. 1). Thus, 84 patients were eligible for inclusion in our study. Nevertheless, one patient was diagnosed with lymphoma 3 months after inclusion and three patients abandoned the study by the third month, so 80 patients were finally evaluated: 43 patients in the EVL plus nadolol group and 37 patients in the EVL group. Figure 1 shows the flow of participants.

Figure 1.

Outcome of patients. ASA, American Society of Anesthesia classification; TIPS, transjugular intrahepatic portosystemic shunt; EVL, endoscopic variceal ligation; Nad, nadolol.

There were no differences between both groups regarding age, sex, cause of cirrhosis, Child-Turcotte-Pugh class, severity of index bleeding, or size of varices (Table 1). The initial bleeding episode was treated by EVL in 73% of cases and by EIS in 34% of the patients without differences between groups.

Table 1. Patient Characteristics
 EVL (n = 37)EVL + Nadolol (n = 43)P Value
Sex (male/female)27/1033/10.7
Age (mean ± SD)60 (18–75)60 (36–75).4
Cause of cirrhosis  .3
Child-Turcotte-Pugh score  .9
Variceal size (II/III/IV)7/17/1310/26/7.2
Endoscopic findings  .1
 Active variceal bleeding2425 
 Recent clot1318 
Initial hemostatic treatment   
 Endoscopic band ligation2926.3
Previous history of variceal bleeding92.1
Previous prophylaxis with β-blockers93.1
Red packed blood cells units, median (range)2 (0–9)2 (0–9).7

The median number of endoscopic sessions (3; range, 1–7) and the duration of the program to achieve eradication was similar in both groups (Table 2). The mean dose of nadolol used was 58 mg (range, 10–120 mg). This resulted in a reduction in the pulse rate from 79 ± 11 beats/min at baseline to 63 ± 7 beats/min during follow-up evaluation. The pulse rate in the patients of EVL group did not change (81 beats/min to 84 beats/min).

Table 2. Results
 EVL (n = 37)EVL + Nadolol (n = 43)P Value
  1. Abbreviation: PHG, portal hypertensive gastropathy.

Number of EVL sessions to eradicate, median (range)3 (1–7)3 (1–7) 
Recurrent variceal bleeding, n (%)14 (38)6 (14).006
 Source of bleeding   
  Esophageal varices104 
  Esophageal ulcer31 
  Gastric varices11 
Bleeding from PHG1 
Probability of variceal recurrence at years 1 and 277%, 97%54%, 68%.06
Mortality, n (%)4 (10.8)5 (11.6) 
 Cause of death   
  Hepatic failure22 
  Variceal bleeding1 
  Hepatocellular carcinoma11 
  Hepatorenal syndrome2 
Follow-up (mo)   
 Mean ± SD15 ± 817.5 ± 7.8 

Recurrent bleeding from esophagogastric varices occurred in six patients (14%) in the EVL plus nadolol group, versus in 14 patients (38%) in the EVL group (P = .006, log-rank test). Another patient in the EVL group bled from portal hypertensive gastropathy. The difference was maintained adjusted to the Child-Turcotte-Pugh class and cause, being more evident in the first 10 months after the intervention. Both curves became parallel after the 10-month follow-up (Fig. 2).

Figure 2.

Kaplan-Meier probability of being free of variceal rebleeding. VL, variceal ligation; NAD, nadolol.

The median follow-up time was similar in both groups: 17.5 ± 7.8 months in the EVL plus nadolol versus 15 ± 8 months in the EVL group. Mortality also was similar in both groups. Four patients (10.8%) died in the EVL group (two of hepatic failure, one of variceal rebleeding, and one of hepatocellular carcinoma). In the EVL plus nadolol group, five patients (11.6%) died (two of hepatic failure, two of hepatorenal syndrome, and one of hepatocellular carcinoma). The most important factor influencing the mortality rate was the Child-Turcotte-Pugh class. In the Child-Turcotte-Pugh class A-B group, there were three deaths (5.3%), as opposed to six deaths (25%) in the class C group.

There were seven treatment failures in the EVL group and two in the EVL plus nadolol group (P < .05): three patients died of bleeding or complications related to the bleeding episode, two patients were given β-blocker treatment, three patients underwent transjugular intrahepatic portosystemic shunt placement, and one patient underwent a splenorenal shunt.

The probability of variceal recurrence during the first year of follow-up after eradication was 54% in the EVL plus nadolol group, compared with 77% for EVL patients (P = .06, log-rank test). Moreover, there was a tendency for fewer EVL sessions to treat new varices in the combined therapy group (two versus one in EVL alone group) without achieving statistical significance (P ± .056; Table 3).

Table 3. Results
 EVL (n = 37)EVL + Nadolol (n = 43)P Value
  1. Abbreviations: GOV, gastroesophageal varices; ns, not significant.

Months of follow-up until new varices apparition, median (range)8 (1–24)9 (2–24)ns
Number of sessions to treat new varices after eradication, median (range)2 (0–5)1 (0–6).056
Number of new varices appearing during follow-up, median (range)1 (0–3)1 (0–5)ns
Fundal varices at the end of follow-up   
 No fundal varices2738 

Fourteen patients in the EVL plus nadolol group had adverse effects. Three of these patients abandoned the protocol study because of nadolol's adverse effects. One patient in the combined therapy group experienced esophageal stenosis that resolved after endoscopic dilation (Table 4). In the EVL group, only one patient had transient dysphagia, which did not require treatment.

Table 4. Adverse Effects
 EVL (n = 37)EVL + Nadolol (n = 43)
Atrioventricular blocking02
Thoracic pain01
Ear drum bleed01
Esophageal stenosis01


β-Blockers and EVL have proven to be valuable methods for secondary prophylaxis of variceal bleeding. β-Blocker drugs decrease portal pressure by reducing portal and azygos blood flow. Several studies have demonstrated its efficacy in reducing the risk of variceal bleeding and mortality in primary and secondary prevention.23 EVL diminishes variceal rebleeding and the mortality rate by local mechanisms,14 but after eradication there is a high tendency for variceal recurrence—between 29% and 50% by the first year—which demands frequent endoscopic follow-up.9, 13 To improve the prophylactic efficacy of β-blockers, addition of 5-isosorbide mononitrate achieves lower portal pressure.24 However, comparison between β-blockers plus 5-isosorbide mononitrate and EVL shows contradictory results. Patch et al.17 found β-blockers plus nitrates to be effective as EVL, with rebleeding rates of 43.7% and 53.8% in a series of patients predominantly of Child-Turcotte-Pugh class C (53%). Villanueva et al.15 showed that nadolol plus isosorbide mononitrate was more effective than EVL, but Lo et al.16 found the contrary. The rate of recurrent bleeding in the EVL group was very high (49%) in the second study, which may be the result of the long interval between endoscopic sessions (2 to 3 weeks). Previous studies with EVL showed a rate of recurrent bleeding of 16% to 36%, with endoscopic sessions every 7 to 14 days.6–9, 11, 12 In the Lo et al. study,16 the rate of recurrent bleeding in the nadolol plus isosorbide mononitrate group also was very high (42%). This increase may be the result of the low dose of nadolol administered. A previous study comparing EVL with propranolol in primary prophylaxis found the same high rate of variceal bleeding.25 It was interpreted as being the result of a low dose of propranolol.26

There are few studies analyzing the efficacy of combined therapeutic methods in the prevention of recurrent hemorrhages after variceal bleeding.18 There is evidence of the advantages of associating endoscopic and pharmacological treatments in first and recurrent variceal bleeding. The association of β-blockers with EIS reduces rebleeding and the risk of death compared with β-blockers alone.27, 28 The association of β-blockers plus EIS shows a reduction of bleeding risk compared with only EIS in individual studies, although a meta-analysis did not show any significant differences.23, 29, 30 The initial hypothesis of the present study was that the association of β-blockers and EVL would increase the therapeutic effects of both methods. The improvement of results combining these therapies could be greater, because EVL has been shown to result in less recurrent bleeding, fewer complications, and lower mortality rates than EIS. The studies comparing EIS with EIS plus propranolol show a rebleeding rate between 18% and 63% in the combined treatment group,29–34 whereas the Lo et al.18 study showed a 23% rebleeding rate combining EVL and nadolol, in accordance with less recurrent bleeding with EVL than with EIS.

In the present study, we found a significant reduction in the rebleeding rate associating EVL and nadolol, compared with EVL alone. This finding was neither the result of reducing the number of ligation sessions nor of shortening the duration of the ligation program until variceal eradication, because rebleeding rates were similar in both therapeutic groups. Obviously, the reduction of rebleeding in the group of combined treatment was the result of the beneficial effects of nadolol. Similar findings have been reported by Lo et al.,18 associating EVL with nadolol and sucralfate compared with EVL alone, although it was not possible to determine which part of this reduction was the result of nadolol or of sucralfate. Our rebleeding rate for EVL (38%) is within the usual range. There have been eight randomized trials comparing EVL with EIS; in two of them, the rebleeding rate in the EVL group was low (16% and 18%), but the other studies reported rebleeding rates between 26% and 36%.6–13 There have been three randomized controlled trials comparing β-blockers plus isosorbide mononitrate with EVL, and ligation rebleeding rates were between 20% and 54%.15–17 Moreover, the two multicenter trials showed EVL rebleeding rates within our rate: 36% and 54%.6, 17 No study showed a rebleeding rate as low as our combined EVL plus nadolol in any therapeutic group, which is promising for the treatment of patients with cirrhosis.

It should be noted that the interim analysis was planned without a predetermined stopping rule other than the usual (and arbitrary) P = .05 for the entire study. Although a post hoc allocation of α is not appropriate, we wish to remark that the study would have been stopped with P = .005 if the O'Brien–Fleming method35 had been used at the study planning stage. As a consequence of this weakness, our conclusions may require slight modification, and the independent replication of our results would be helpful.

An important point is the reappearance of new varices after initial eradication by EVL. In the present study, the association of nadolol also was beneficial in slowing down the speed of variceal reappearance and in reducing the number of endoscopic sessions. Differences were close to statistical significance, suggesting that this beneficial effect could reduce the risk of late rebleeding.

Nevertheless, the side effects and contraindications of β-blockers are a limitation for their use. It is estimated that approximately 15% to 20% of patients who take these medications will have contraindications.36 In our study, only 9 (3.75%) of 240 patients had a β-blocker contraindication. The rate of side effects is more than 30%, resulting in withdrawal from treatment in 20% to 30% of the patients.37 We observed adverse effects in 5 (11%) of 43 patients, whereas three of them had to be withdrawn from the treatment. Moreover, considering the difference in recurrent bleeding between both groups observed in the first 10 to 12 months of treatment, the period of combined treatment could be restricted to this period if secondary effects limit their quality of life.

We conclude that the association of nadolol to EVL reduces the rate of recurrent bleeding from esophageal varices compared with EVL alone and achieves a low rate of treatment failure, although it does not reduce mortality. The combination of nadolol with variceal ligation could delay the appearance of new varices after eradication. Nevertheless, the combined therapy increases the incidence of adverse effects in 11% of patients. The results of this prematurely terminated study would benefit from confirmation in further studies.


Members of the EVL Study Group: M. Hernandez-Guerra, E. Fabrega, J. Crespo, F. Pons-Romero (Hospital Marques de Valdecilla); R. Campo, M. Gil, B. Dalmau (Hospital de Sabadell). The authors thank Javier Llorca (Public Health, Medicine School, Santander) and Jordi Real (CS Parc Taulí) for statistical consultation and Carolyn Oldham and Cristina Fuentes for their comments and for reviewing the manuscript.