Ligation Versus Propranolol for the Primary

Prophylaxis of Variceal Bleeding in Cirrhosis
Michael Schepke,
Gerhard Kleber,
Dieter N¨ urnberg,
J ¨ org Willert,
Lydia Koch,
Wilfried Veltzke-Schlieker,
Claus Hellerbrand,
Johannes Kuth,
Stefan Schanz,
Stefan Kahl,
Wolfgang E. Fleig,
and Tilman Sauerbruch
for the
German Study Group for the Primary Prophylaxis of Variceal Bleeding
In this randomized controlled multicenter trial, we compared endoscopic variceal band-
ing ligation (VBL) with propranolol (PPL) for primary prophylaxis of variceal bleeding.
One hundred fifty-two cirrhotic patients with 2 or more esophageal varices (diameter >5
mm) without prior bleeding were randomized to VBL (n ؍ 75) or PPL (n ؍ 77). The
groups were well matched with respect to baseline characteristics (age 56 ؎ 10 years,
alcoholic etiology 51%, Child-Pugh score 7.2 ؎ 1.8). The mean follow-up was 34 ؎ 19
months. Data were analyzed on an intention-to-treat basis. Neither bleeding incidence
nor mortality differed significantly between the 2 groups. Variceal bleeding occurred in
25% of the VBL group and in 29% of the PPL group. The actuarial risks of bleeding after
2 years were 20% (VBL) and 18% (PPL). Fatal bleeding was observed in 12% (VBL) and
10% (PPL). It was associated with the ligation procedure in 2 patients (2.6%). Overall
mortality was 45% (VBL) and 43% (PPL) with the 2-year actuarial risks being 28%
(VBL) and 22% (PPL). 25% of patients withdrew from PPL treatment, 16% due to side
effects. In conclusion, VBL and PPL were similarly effective for primary prophylaxis of
variceal bleeding. VBL should be offered to patients who are not candidates for long-
term PPL treatment. (HEPATOLOGY 2004;40:65–72.)
pper intestinal hemorrhage is a common and
often fatal complication of portal hypertension.
It occurs in 30% of patients with cirrhosis, with
each bleeding episode bearing a mortality risk of 30% to
Thus, prophylactic treatment prior to the first
bleeding (i.e., primary prophylaxis) is mandatory in high-
risk patients.
Nonselective ␤-blockers (i.e., proprano-
lol, nadolol)— the current standard prophylaxis—reduce
bleeding incidence and bleeding-related mortality.
However, pharmacotherapy with ␤-blockers is not opti-
mal: 30% to 40% of patients will not achieve a sufficient
reduction of portal pressure to prevent bleeding.
thermore, contraindications and side effects are com-
and may require withdrawal, which reincreases
the risk of bleeding.
Therefore, therapeutic alternatives
to ␤-blockers are warranted. Due to heterogeneous re-
sults, prophylactic endoscopic sclerotherapy is not recom-
mended for the primary prevention of variceal
Compared to injection sclerotherapy,
variceal banding ligation (VBL) allows a more rapid and
more effective eradication of varices with fewer side
To date, only 2 trials comparing VBL with stan-
dard treatment (␤-blockers) have been fully pub-
The results of these studies are controversial
and partly inconclusive. In the present article, we report
on the results of a prospective randomized multicenter
trial comparing propranolol (PPL) and banding ligation
for the primary prophylaxis of variceal bleeding in pa-
tients with cirrhosis.
Abbreviations: VBL, variceal banding ligation; PPL, propranolol; NIEC, North
Italian Endoscopic Club.
From the
Department of Internal Medicine I, University of Bonn, Bonn, Ger-
Department of Internal Medicine I, University of Halle-Wittenberg, Halle-
Wittenberg, Germany;
Medical Department B, Ruppiner Kliniken, Neuruppin,
Department of Medicine, Knappschaftskrankenhaus, University of Bo-
chum, Bochum, Germany;
Department of Internal Medicine and Gastroenterol-
ogy, University Clinic Charite ´, Campus Virchow, Berlin, Germany;
of Internal Medicine I, University of Regensburg, Regensburg, Germany;
Department, Malteserkrankenhaus Ju¨lich, Germany;
Medical Department, Evan-
gelisches Krankenhaus, Ko¨ln-Kalk, Germany, and
Department of Gastroenterol-
ogy, Hepatology, and Infectious Diseases, University of Magdeburg, Magdeburg,
Received January 5, 2004; accepted March 23, 2004.
Supported by the German Association for the Study of the Liver (GASL) and the
Ernst und Berta Grimmke Stiftung, Dusseldorf, Germany.
Presented, in part, at the 54th Annual Meeting of the American Association for
the Study of Liver Diseases, Boston, MA, October 24-28, 2003.
Address reprint requests to: Tilman Sauerbruch, MD, Department of Internal
Medicine I, University of Bonn, Sigmund-Freud-Str.25, D-53127 Bonn, Ger-
many. E-mail:; fax: 49-228-2874322.
Copyright © 2004 by the American Association for the Study of Liver Diseases.
Published online in Wiley InterScience (
DOI 10.1002/hep.20284
Patients and Methods
Patients and Inclusion Criteria. Patients fulfilling
the following inclusion criteria participated in the present
study after giving written informed consent: 2 or more
esophageal varices with a diameter greater than 5 mm;
proven liver cirrhosis (diagnosis based on histology, n ϭ
54, or unequivocal clinical, sonographic, and laboratory
findings, n ϭ98); a Child-Pugh score
below12; and age
18 to75 years. Exclusion criteria were previous variceal
bleeding or any other reported episode of hematemesis or
melena unless clearly proven to be independent frompor-
tal hypertension; prehepatic portal hypertension; brady-
cardia less than 64 per minute; systolic blood pressure less
than 100 mm Hg; contraindications to PPL (obstructive
airway disease, congestive heart failure, severe peripheral
vascular disease, diabetes mellitus type 1); severe comor-
bidity substantially reducing life expectancy (advanced
cancer including hepatocellular carcinoma, renal failure, ad-
vanced HIVinfection); being listed for liver transplantation;
long-term anticoagulant treatment; inability to give in-
formed consent; treatment with ␤-blockers or nitrates
within 30 days before randomization; existing transjugular
intrahepatic portosystemic shunt or surgical shunt; and as-
sumed incompliance with the study protocol.
The study was performed according to the 1975 Dec-
laration of Helsinki and European Community Directive
91/507 (”good clinical practice”). The study protocol was
approved by the ethics committee of the University of
Bonn, Germany and by the ethics committees of all par-
ticipating centers.
Two hundred three eligible patients fulfilling the in-
clusion criteria were screened between October 1, 1996
and March 31, 2001. Of these, 46 had to be excluded due
to existing exclusion criteria (contraindications to PPL,
10; patients listed for liver transplantation, 13; refusal to
participate, 9; severe comorbidity, 8; assumed incompli-
ance, 4; inability to give informed consent, 2). Five addi-
tional patients had been included in the study despite
having existing exclusion criteria or lacking inclusion cri-
teria (prior bleeding episode, 1; age greater than 75, 1;
advanced oropharyngeal cancer diagnosed prior to ran-
domization, 1; Child-Pugh score greater than 11, 1; long-
term anticoagulant treatment, 1) and were therefore
excluded from the analysis.
Randomization. Patients were centrally assigned to
the 2 treatment arms at the Institute of Medical Biometry,
University of Bonn, Germany, by a block randomization
with blocks of 6 patients for each center. Immediately
after inclusion of a particular patient, the randomization
site informed the participating center about the individ-
ual outcome of randomization by telefax. The hospitals
that recruited patients for this study are listed in the Ap-
pendix. Of the 27 centers, 10 enrolled 6 or more patients
(total, 100 patients).
Study Design and Follow-up. In the VBL arm, li-
gations were performed at weekly intervals under mild
sedation using midazolam. During each session, up to
10 bands were placed beginning in the distal esophagus
using a multiband ligation device (Sixshooter; Wilson-
Cook Inc., Winston-Salem, NC or Speedband; Boston
Scientific, Inc., Natick, MA) until eradication of vari-
ces was achieved. Religation was performed when at
least 1 varix with a diameter greater than 5 mm reoc-
In the PPL arm, treatment was started at a dosage of 40
mg twice daily. This dosage was increased by 10 mg twice
daily either until a reduction of the resting heart rate of
20%—compared to the pretreatment heart rate—was
achieved, or up to the maximal dose without side effects.
In case of side effects already induced by the initial daily
dose of 80 mg, propranolol dosage was reduced by 10 mg
twice daily, and the maximumtolerated dosage was given.
A crossover of patients due to treatment intolerance was
not allowed. After dose-finding for PPL or successful li-
gation, clinical follow-up visits, which included surveil-
lance endoscopies, were performed at 6-month intervals
in both groups. In the VBL group, 1 additional endos-
copy was performed 3 months after randomization. At
each follow-up visit, the following sequelae of treatment
had to be assessed according to the study protocol. For the
VBL group, these were bleeding, complicated ulcer (de-
fined as bleeding or penetrating ulcer), stenosis, pleural
effusions, mediastinitis, aspiration, perforation, or newly
developed gastric varices; for the PPL group, these were
dyspnea, arrhythmia, edema, arterial hypotension, impo-
tence, or Raynaud phenomenon or deterioration of pe-
ripheral vascular disease. In the PPL group, compliance
was assessed by pill count. All patients were asked to grade
the tolerability of treatment on a standard questionnaire
with a semiquantitative scale (1, not harmful at all; 2,
slightly harmful; 3, moderately harmful; 4, harmful) at
each follow-up visit. The grading of treatment tolerability
during the first 6 months was averaged, and during the
further follow-up mean values at every visit are given.
Each patient was followed until death or for at least 2
years, the minimum follow-up period defined by the
study protocol. Patients surviving a variceal bleeding as a
defined end point were kept in the study, but only survival
was documented. Patients who underwent liver trans-
plantation during follow-up were censored. Recruitment
of patients was stopped on March 31, 2001, and the study
was terminated on March 31, 2003.
End Points, Definitions. The primary study end
points were defined as gastrointestinal bleeding due to
portal hypertension and death from any cause. Portal hy-
pertensive bleeding was defined (according to Baveno
)as hematemesis and/or melena together with di-
rect endoscopic signs of bleeding (i.e., visualization of
blood emanating from a varix or thrombus or fibrin clot
on a varix). In cases of hematemesis and/or a melena with-
out direct endoscopic signs of variceal bleeding and the
absence of other potential bleeding sources, a fall of he-
moglobin by 2 g/dL or more was required to fulfill the
end-point definition. This definition of bleeding was also
applied to patients bleeding from ligation ulcers. The
bleeding episodes observed in the present study were ex-
clusively due to variceal bleeding. Management of acute
bleeding was standardized at each participating center as
follows: emergency endoscopy within 6 hours (ligation or
sclerotherapy); adjuvant vasoactive treatment (octreotide
or terlipressin); and supportive therapy, including blood
transfusions. Any death within 6 weeks after a bleeding
episode was considered to be bleeding-induced (fatal
Successful variceal eradication was defined as
the visual impression of absence of esophageal varices
large enough for ligation. Recurrent varices were defined
as the presence of varices with a diameter of at least 5 mm
after initial successful eradication.
Sample Size Calculation and Interim Analysis.
When we planned the study, no data were available on the
incidence of variceal bleeding following banding ligation
for primary prophylaxis. The study committee decided
that the trial should be designed to reveal a 10% differ-
ence in bleeding rates in favor of VBL. However, due to
the uncertainty of the ligation effect in primary prophy-
laxis, the protocol included an interim analysis to be per-
formed 6 months after the inclusion of 100 patients. We
presumed 2- year-bleeding risks of 20% in the PPL group
and 10% in the VBL group. Under these assumptions,
and with a type I error of 0.05, a type II error of 0.1, and
1 interim analysis, the sample size calculation demon-
strated that 200 patients had to be recruited for each
treatment arm. The interim analysis, however, demon-
strated virtually no differences between the treatment
arms with respect to bleeding. Thus, according to the
results of the interimanalysis, even the projected 2 ϫ200
patients would have been far from sufficient to demon-
strate a difference in bleeding rates. Therefore, after con-
sultation with two external advisories, the study
committee decided to stop randomization on March 31,
Statistical Analysis. Data were analyzed on an inten-
tion-to-treat basis. Data are expressed as mean (Ϯ SD).
Differences between groups were analyzed by Chi-square
test, Fisher exact test, and the unpaired Student t test. The
actuarial probabilities of bleeding or death were calculated
by the Kaplan-Meier method, and comparisons were
made using the log-rank test. Observed bleeding episodes
within 2 years were further compared by chi-square test.
The effect of confounding risk factors was assessed by Cox
proportional hazards regression analysis. Scoring models
(Child-Pugh, North Italian Endoscopic Club [NIEC] in-
dex) were not included in the multivariate regression anal-
ysis. A two-tailed P value of less than .05 was considered
to demonstrate statistical significance.
Seventy-seven patients were randomized to receive
PPL, and 75 patients were randomized to the VBL arm
(Table 1). Significant differences between the two study
groups were not observed in any of the baseline parame-
ters. Patients were followed for 34.4 Ϯ 18.9 months
(range, 0.1-73.2). The mean time period between ran-
domization and termination of the study was 51.8 Ϯ15.0
months. Except for 5 patients who had to be censored due
to liver transplantation during the first 2 years after ran-
Table 1. Patient Characteristics
(n ؍ 75)
(n ؍ 77)
Age (y) 54.3 Ϯ 10.5 57.3 Ϯ 9.7
Sex, n (%)
Male 50 (66.7) 54 (70.1)
Female 25 (33.3) 23 (29.9)
Etiology of cirrhosis, n (%)
Alcoholic 40 (53.3) 38 (49.4)
Viral 22 (29.3) 25 (32.5)
Other* 13 (17.3) 14 (18.2)
Child-Pugh class, n (%)
A 34 (45.3) 37 (48.1)
B 31 (41.3) 31 (40.3)
C 10 (13.3) 9 (11.7)
Child-Pugh score 7.3 Ϯ 1.8 7.0 Ϯ 1.9
Ascites, n (%) 32 (42.7) 29 (37.7)
Hepatic encephalopathy, n (%) 10 (13.3) 9 (11.7)
Bilirubin (mg/dL) 2.5 Ϯ 2.5 2.2 Ϯ 2.0
Albumin (g/L) 37 Ϯ 6.0 37 Ϯ 5.6
Prothrombin index (%) 72 Ϯ 16 72 Ϯ 14
Creatinine (mg/dL) 0.9 Ϯ 0.3 0.8 Ϯ 0.3
Endoscopic findings, n (%)
Esophageal varices grade II 32 (42.7) 35 (45.5)
Esophageal varices grade III 43 (57.3) 42 (54.5)
Red markings 29 (38.7) 30 (39.0)
Gastric fundal varices 10 (13.3) 10 (13.2)
NIEC score 32.5 Ϯ 6.2 31.9 Ϯ 6.3
NOTE. Baseline clinical, biochemical, and endoscopic characteristics of pa-
tients enrolled in the study. None of the baseline characteristics differed signifi-
cantly between the treatment groups.
*Primary biliary cirrhosis (VBL, n ϭ 1; PPL, n ϭ 4); primary sclerosing
cholangitis (PPL, n ϭ 1); autoimmune hepatitis (VBL, n ϭ 1; PPL, n ϭ 1);
drug-induced cirrhosis (VBL, n ϭ 1); alpha 1-antitrypsin deficiency (PPL, n ϭ 1);
hemochromatosis (PPL, n ϭ 1); cryptogenic (VBL, n ϭ 10; PPL, n ϭ 6).
HEPATOLOGY, Vol. 40, No. 1, 2004 SCHEPKE ET AL. 67
domization, the entire cohort could be followed for at
least 2 years. No patient was lost to follow-up. During the
entire follow-up period, a total of 10 patients (VBL, 6;
PPL, 4) had to be censored due to liver transplantation.
The mean daily dosage of PPL was 77.3 Ϯ 39.5 mg
after titration, according to the protocol. A20%and 25%
reduction of resting heart rate was achieved in 79.2% and
52.8% of our patients, respectively. Side effects were re-
ported in 53 patients (69%). In 12 patients (16%), these
PPL side effects did not resolve after dosage reduction,
and PPL had to be withdrawn. The adverse events requir-
ing withdrawal of treatment were symptomatic arterial
hypotension (n ϭ 10), deterioration of peripheral vascu-
lar disease (n ϭ 1), and exanthema suspected to be in-
duced by PPL (n ϭ1; Table 2). An additional 7 patients
stopped PPL intake (despite absence of side effects) due to
incompliance. Thus, a total of 19 patients (25%) with-
drew from PPL treatment. Of these 19 patients, 6
(31.6%) bled thereafter (bleeding was fatal in 2 patients,
0.6 and 11.9 months after PPL withdrawal). On average,
variceal bleeding occurred 13.4 Ϯ 9.8 months after PPL
In the VBL group, successful eradication of varices was
achieved in 69 patients (92%). Initial variceal eradication
required 10.3 Ϯ 6.2 rubber bands placed during 2.0 Ϯ
1.2 endoscopy sessions. Recurrent varices occurred in 42
patients (60%) after a mean of 11.2 Ϯ11.3 months after
eradication. Eradication of recurrent varices required
1.2 Ϯ 0.9 sessions and 5.5 Ϯ 3.3 rings. After the second
eradication period, large varices (i.e., diameter Ն 5 mm)
recurred in 11 patients (15%). Adverse events were re-
ported in 47% of patients in the VBL group (Table 2). In
the majority, these adverse events were mild or moderate.
The only severe side effect in the VBL group was bleeding
from ligation ulcers; this occurred in 5 patients (6.7%). It
was unexpectedly serious in 3 patients: in 1 patient, a
life-threatening hemorrhage from a ligation ulcer could
only be stopped endoscopically by cyanobucrylate (His-
toacryl) injection. An additional patient, who had been
discharged from the hospital immediately after the first
ligation session, was readmitted 12 days later with a severe
hemorrhage from a bleeding ligation ulcer and conse-
quently died from uncontrolled bleeding despite emer-
gent endoscopy. One other patient was found dead with
signs of severe hematemesis 3 days after the first ligation
session. Although no endoscopy was performed, and thus
the bleeding source remains uncertain, it was counted as
fatal hemorrhage related to banding ligation. Thus, we
observed a treatment-related mortality of 2.6% in the
VBL group. Patients’ feeling with respect to tolerability
showed a tendency in favor of VBL, a difference that was
not statistically significant (Table 3).
The actuarial risks of variceal bleeding at 2 years were
20.0%Ϯ4.8%for VBLand 17.6%Ϯ4.6%for PPL. The
figures for actuarial bleeding risk at 3 years were 24.8%Ϯ
5.6% (VBL) and 30.0%Ϯ5.9% (PPL). Figure 1 depicts
the Kaplan-Meier plot of variceal bleeding for the two
study groups. There were no statistically significant dif-
ferences between the groups with respect to bleeding in-
cidence (P ϭ .87, log-rank test). The actual (observed)
bleeding rates at 2 years were 18.7% (VBL) and 15.6%
(PPL); P ϭ .41). During the entire follow-up period, 19
patients (25.3%) bled in the VBL group and 22 (28.6%)
bled in the PPL group (P ϭ.62). Bleeding-related deaths
occurred in 9 patients (12%) treated with banding liga-
tion compared to 8 patients (10.4%) in the PPL group
(P ϭ.58; Table 3). One patient in each treatment group
died after a recurrent bleeding episode; the other bleed-
ing-related deaths occurred after the initial bleeding.
Overall mortality, the second primary end point of this
study, did not differ significantly between the 2 study
Table 2. Adverse Events
VBL n (%) PPL* n (%)
Bleeding from ligation ulcer† 5 (7) Symptomatic hypotension 30 (39)
Peripheral edema 2 (3) Peripheral edema 19 (25)
Gastrointestinal discomfort 12 (16) Gastrointestinal discomfort 10 (13)
Symptomatic hypotension 1 (1) Arrhythmia including symptomatic
bradycardia 15 (19)
Dizziness 3 (4) Dizziness 8 (10)
Dysphagia 15 (20) Impotence 3 (4)
Raynaud symptoms and
deterioration of peripheral
vascular disease 3 (4)
Exanthema 1 (1)
NOTE. Some patients reported more than 1 side effect.
*PPL had to be withdrawn in 12 patients due to side effects (symptomatic
hypotension, n ϭ 10; exanthema, n ϭ 1; peripheral vascular disease, n ϭ 1).
†Treatment-related mortality 2.6% (2 patients).
Table 3. Outcomes and Treatment Tolerability
VBL n (%) PPL n (%)
Variceal bleeding 19 (25.3) 22 (28.6)
Overall mortality 34 (45.3) 33 (42.9)
Bleeding related deaths 9 (12) 8 (10.4)
Treatment tolerability*
6 mo† 1.4 Ϯ 0.41 1.66 Ϯ 0.67
12 mo 1.24 Ϯ 0.43 1.43 Ϯ 0.58
18 mo 1.24 Ϯ 0.48 1.69 Ϯ 0.9
24 mo 1.15 Ϯ 0.35 1.6 Ϯ 0.67
36 mo 1.3 Ϯ 0.56 1.44 Ϯ 0.62
NOTE. End points observed during the entire follow-up of 34.2 Ϯ19.1 months.
None of the figures differed significantly between the 2 groups.
*Patients were asked to grade treatment tolerability on a semiquantitative
scale; (1, not harmful at all; 2, slightly harmful; 3, moderately harmful; 4,
†Treatment tolerability values during the first 6 months of follow-up were
averaged, during further follow-up mean values at each visit are given.
groups: 34 (45.3%) and 33 patients (42.9%) died in the
VBL and PPL group, respectively (P ϭ 0.59; Table 3).
Actuarial mortality risks at 2 years were 28.3% Ϯ 5.2%
(VBL) and 22.4% Ϯ 4.8% (PPL); at 3 years, the risks
were 38.5% Ϯ 5.9% (VBL) and 32.7% Ϯ 5.5% (PPL);
P ϭ .37, log-rank test (Fig. 2).
To identify prognostic variables associated with the
risk of first bleeding and death, we included the baseline
parameters of age, sex, treatment arm, etiology (alcoholic/
viral/other), variceal grading, ascites (ϩ/Ϫ), hepatic en-
cephalopathy (ϩ/Ϫ), prothrombin index, creatinine, and
bilirubin in a Cox proportional hazards regression analy-
sis. With respect to first variceal bleed, the parameters of
ascites, bilirubin, prothrombin index, and creatinine were
identified as significant (P Ͻ0.1) risk factors on univari-
ate analysis (Table 4). On multivariate analysis, bilirubin
(hazard ratio [HR]1.27; 95% CI 1.13–1.43 P ϭ .0001)
and creatinine (HR 3.32; 95% CI 1.14–9.96 P ϭ .003)
turned out as independently predictive for the risk of first
bleeding. The NIEC score, designed to predict bleeding
in untreated patients, was not predictive for first variceal
bleeding in the cohort receiving a prophylactic regimen.
Analysis of prognostic variables with respect to survival
generated the parameters of age, ascites, prothrombin in-
dex, creatinine, bilirubin, and number of recruited pa-
tients per center as significant (P Ͻ.1) predictive variables
in the univariate model (Table 5). In the multivariate Cox
model, the parameters of age (HR 1.029; 95% CI 1.002–
1.06 P ϭ .034) and prothrombin index (HR 0.97 95%
CI 0.96, 0.99 P ϭ.002) were identified as independently
predictive of survival.
In the present randomized controlled multicenter trial,
we compared VBL with PPL as treatment for the preven-
tion of first variceal bleeding in patients with high-risk
varices. We did not find significant differences between
these two regimens with respect to our end points: variceal
bleeding and overall mortality.
According to results from the trials comparing VBL
with nonactive treatment
and the 2001 meta-analy-
it can be concluded that VBL is effective in patients
with high-risk varices. It reduces not only the risk of first
Fig. 1. Kaplan-Meier plot comparing bleeding incidence in the endo-
scopic ligation group (VBL) and the propranolol group (PPL). P ϭ .87,
log-rank test. Fig. 2. Kaplan-Meier plot comparing survival in the endoscopic liga-
tion group (VBL) and the propranolol group (PPL). P ϭ.37, log-rank test.
Table 4. Univariate Cox Regression Analysis (First Bleeding)
Variable Hazard Ratio 90% CI P
Age 0.979 0.954 1.005 .180
Sex 1.055 0.613 1.818 .871
Etiology (alcoholic/viral/other) 0.938 0.852 1.033 .277
Variceal grading (II/III) 0.715 0.427 1.198 .285
Bilirubin* 1.241 1.129 1.365 .0002
Ascites 1.329 1.070 1.651 .031
Creatinine* 2.545 1.019 6.357 .093
Prothrombin index 0.974 0.957 0.991 .013
Hepatic encephalopathy 1.004 0.467 2.156 .994
No. of patients per center 0.984 0.951 1.018 .438
Treatment arm 0.950 0.567 1.593 .871
*Significant (P Ͻ .05) in the multivariate analysis.
HEPATOLOGY, Vol. 40, No. 1, 2004 SCHEPKE ET AL. 69
bleed but also,according to the meta-analysis, mortality.
However, in these trials, patients did not receive the stan-
dard treatment for primary prophylaxis, namely, PPL.
Since VBL is still regarded as an experimental treatment
in this setting, its superiority over PPL has to be proven
before it can be recommended as a first-line option. To
date, only two randomized studies investigating VBL for
the primary prophylaxis of variceal bleeding in compari-
son to PPL have been finalized and fully published.
addition, preliminary results of a third small study have
been reported.
The present trial has the highest number
of patients in the VBL arm (n ϭ75)compared to studies
by Lui et al. (n ϭ44),
Sarin et al. (n ϭ45),
and De et
al. (n ϭ 15).
Notably, we did not lose a single patient
from follow-up during the 24 months after randomiza-
tion, and the average follow-up time of almost 3 years is
the longest of all trials on VBL for primary bleeding pro-
phylaxis published to date.
The present study is a mul-
ticenter trial with a rather high number of centers that
included fewer than 6 patients (1 randomization block).
Results between patients from large centers enrolling 6 or
more patients (n ϭ 100) and small centers (n ϭ 52) did
not differ either with respect to primary end points or
with respect to baseline parameters; this suggests that our
findings are robust.
Because monotherapy
and combination treatment
with nitrates
have have failed to prove superiority, PPL is
still the adequate standard treatment for any trial on pri-
mary prophylaxis of variceal bleeding. The results ob-
tained in our standard treatment arm (PPL) are well in
line with the literature: the actuarial bleeding risk after 2
years is 17.6% in the present study and 22% in the large
meta-analysis of Poynard et al.
Recent trials by Lui et al.
Garcia-Pagan et al. (17% in patients with vari-
ces Ͼ5 mm),
and several others
also reported similar
bleeding rates. In the first trial comparing VBL and PPL
for primary prophylaxis,
the actuarial bleeding risk was
very high in the PPL group (40% at 18 months); this
prompted criticism concerning compliance and PPL dos-
In the present trial, 9.1% of patients stopped PPL
due to noncompliance compared to 7.6% in the Scottish
trial by Lui at al.
The mean dose of PPL given in the
present study was lower than that in Lui et al.
—77 mg
versus 113 mg—but comparable to the mean dosages
reported in the meta-analysis by Imperiale et al.
70-73 mg). To minimize both the number of patients that
had to be excluded from the study and the incidence of
adverse events, the target heart rate reduction in the PPL
group was 20%, lower than in the majority of trials
(25%). PPL dosage has been shown to be an important
predictor of variceal rebleeding.
It is therefore possible
that a more aggressive PPL treatment would have resulted
in a lower bleeding rate. However, given the rather high
incidence of adverse events attributable to PPL treatment
in the present study (symptomatic hypotension 39%, pe-
ripheral edema 25%, bradycardia 19%), we believe that a
target heart-rate reduction of 25% instead of 20% would
have resulted in a higher PPL dosage in only very few
patients. This assumption is supported by the fact that in
one of the largest studies evaluating drug treatment for the
primary prophylaxis of variceal bleeding,
the mean daily
PPL dosage was 67 mg (vs. 77 mg in the present study),
although a heart-rate reduction of 25% was targeted.
Regarding the prophylactic efficacy of VBL compared
to PPL, the present study differs from the two previously
published studies
in several important aspects. In the
first study, a single center trial,
the bleeding risk at 18
months was 15%(24 months not reported); this is similar
to our results (18.5%at 18 months). Due to a 40%bleed-
ing incidence in the PPL group, that study found a signif-
icant benefit in favor of VBL. By contrast, both the
present study and the Scottish multicenter trial by Lui et
failed to demonstrate a significant difference between
VBL and PPL. Remarkably, the bleeding rate in the VBL
group of the Scottish study was much lower than in ours
(6.2%vs. 20.0%at 2 years). However, the authors did not
prove superiority of VBL statistically, probably due to a
type II error (only 44 patients in the VBL arm). Our
actuarial 2-year bleeding risk of 20% is similar to many
other evaluations of VBL for primary bleeding prophyl-
Compared to the Scottish trial, the present
study included more patients with grade III varices
(55.9% vs. 16.3%) and with red color signs (38.8% vs.
5.1%) but fewer Child-Pugh class C patients (12.5% vs.
32.6%). The baseline bleeding risk, as assessed by the
NIEC index,
was slightly higher in the present trial
(32.2 vs. 30.4). Furthermore, one might speculate that
ligation was less intensive in the present study. Indeed, the
number of ligation sessions required to achieve variceal
eradication was lower in our study (2.0 vs. 3.2). However,
Table 5. Univariate Cox Regression Analysis (Mortality)
Variable Hazard Ratio 90% CI P
Age* 1.026 1.003 1.049 .062
Sex 0.852 0.536 1.353 .569
Etiology (alcoholic/viral/other) 1.017 0.954 1.083 .672
Variceal grading (II/III) 0.720 0.471 1.100 .202
Bilirubin 1.123 1.033 1.221 .022
Ascites 1.284 1.077 1.531 .019
Creatinine 2.548 1.172 5.538 .047
Prothrombin index* 0.974 0.960 0.989 .003
Hepatic encephalopathy 1.685 0.997 2.846 .102
No. of patients per center 0.966 0.938 0.995 .053
Treatment arm 0.802 0.528 1.220 .387
*Significant (P Ͻ .05) in the multivariate analysis.
the intensity of endoscopic treatment is difficult to com-
pare because the total number of rubber bands placed is
not mentioned by Lui et al.
Furthermore, 60% of our
patients received follow-up ligation of recurrent varices (6
rubber bands on average).
The total incidence of side effects tended to be lower in
the VBL arm: However, fatal (n ϭ2) and life-threatening
(n ϭ1) complications were only encountered in the VBL
group; PPL side effects were usually mild and resolved
after dosage reduction or withdrawal. PPL did not cause
serious adverse events, however, 1 of the 2 fatal hemor-
rhages observed after drug withdrawal in the PPL arm
occurred shortly after PPL discontinuation (18 days); this
may suggest a causal relationship. The tolerability of treat-
ment tended to be better in the VBL group throughout
the follow-up.
Given the very similar rates of bleeding in both groups,
PPL should be more cost-effective. Based on current Ger-
man prices, present data suggest that, excluding prices for
the first diagnostic endoscopy, the cost of ligation devices
and endoscopies until eradication (not including cost of
ligation of recurrent varices) amounts to 570 Euros on
average in the VBL arm. This is similar to the cost of 6
years of PPL treatment (540 Euros). In addition, the high
incidence of recurrent varices requiring religation has to
be weighed in favor of PPL treatment with respect to cost.
One might question whether VBL and PPL have been
effective at all in our study. If we compare the present
results with our previous trial,
in which we studied pro-
phylactic sclerotherapy—in a very similar cohort and with
similar end point definitions—versus no prophylactic
regimen (bleeding rate 37%during a median follow-up of
22 months), it seems that both regimens in the present
study had a beneficial effect on bleeding, although their
effect on mortality is much more questionable.
Where is the place for VBL in primary prophylaxis of
variceal bleeding? Assuming an equivalent efficacy, VBL
can be offered to patients who are noncompliant with
PPL (9.1% in our study) as well as to patients in whom
severe side effects (16%in our study) or contraindications
(5%of patients screened for our study) preclude the use of
PPL. Furthermore, prophylactic VBL might be consid-
ered for patients who show no adequate hemodynamic
response to PPL (40% according to the literature).
In summary, the present randomized controlled mul-
ticenter trial comparing endoscopic variceal ligation and
propranolol for the primary prevention of variceal hem-
orrhage did not show significant differences between reg-
imens with respect to bleeding incidence and bleeding-
induced and overall mortality. Although endoscopic
ligation has been shown to be effective in such setting,
propranolol should be considered first; ligation should be
offered to patients with side effects due to propranolol, in
whom propranolol is contraindicated, who are noncom-
pliant, or who do not respond hemodynamically,
amounting to 50%of patients for whomprimary prophy-
laxis of variceal hemorrhage is indicated. After acceptance
of this manuscript, a further study in line with our results
has been published.
Acknowledgment: The authors thank Ms. H.
Diedenhofen for documentary assistance and Dr. R. Fim-
mers, Institute for Medical Biometry, Informatics, and
Epidemiology, University of Bonn, Germany, for statis-
tical advice.
The following hospitals recruited patients for this study:Mediz-
inische Universita¨tsklinik und Poliklinik I, Bonn (M. Schepke, L.
Koch, T. Sauerbruch), Klinik und Poliklinik fu¨r Innere Medizin I,
Universita¨t Halle-Wittenberg (G. Kleber, C. Go¨bel, W.E. Fleig),
Medizinische Klinik B, Ruppiner Kliniken GmbH, Neuruppin (T.
Liebig, D. Nu¨rnberg), Medizinische Universita¨tsklinik, Knappschafts-
Krankenhaus, Bochum-Langendreer (J. Willert), Klinik und Po-
liklinik fu¨r Innere Medizin I, Universita¨tsklinikum Regensburg (C.
Hellerbrand), Klinik fu¨r innere Medizin mit Schwerpunkt Gastroen-
terologie, Universita¨tsklinikum Charite´, Campus Virchow-Klinikum,
Berlin (W. Veltzke-Schlieker), Malteserkrankenhaus St. Elisabeth,
Medizinische Abteilung, Ju¨lich (J. Kuth), Evangelisches Krankenhaus
Ko¨ln-Kalk, Innere Abteilung (S. Schanz), Klinik fu¨r Gastroenterolo-
gie, Hepatologie und Infektiologie, Universita¨tsklinikum Magdeburg
(S. Kahl), Abteilung fu¨r Gastroenterologie, Klinikum Ernst von Berg-
mann, Potsdam (F. Jopke), Innere Klinik II, Kreiskrankenhaus As-
chersleben (S. Deist), Innere Abteilung, Malteserkrankenhaus Bonn
(B. Wo¨stmann), Medizinische Klinik II, Klinikum St. Marien, Am-
berg (M. Ko¨llinger), Medizinische Klinik, Allgemeines Krankenhaus
Hagen (R. Hanrath), Medizinische Universita¨tsklinik Kiel (H. Hin-
richsen), Medizinische Klinik und Poliklinik, Abteilung fu¨r Gastroen-
terologie und Hepatologie, Berufsgenossenschaftliche Kliniken
Bergmannsheil, Bochum (T. Griga), Medizinische Klinik, St. Elisa-
beth Krankenhaus Ko¨ln (C. Pohl), Medizinische Klinik I, Klinikum
Landshut (R. Schaller), Medizinische Klinik, Gastroenterologie und
Hepatologie, Krankenhaus Siegburg GmbH (K. Simon), Mediz-
inische Universita¨tsklinik III, Rheinisch-Westfa¨lische Technische
Hochschule Aachen (F. Lammert), I. Medizinische Klinik und Po-
liklinik, Johannes-Gutenberg-Universita¨tsklinikumMainz (K. Merge-
ner), Medizinische Klinik I, Klinikum Lippe-Detmold, Detmold (H.
Waltke), Innere Abteilung, Elisabeth-Krankenhaus, Essen (B. Tillen-
burg), Medizinische Universita¨tsklinik II, Klinikum Großhadern,
Ludwig-Maximilians-Universita¨t Mu¨nchen (A. Gerbes), Klinik fu¨r In-
nere Medizin II, Dr. Horst-Schmidt-Kliniken GmbH, Wiesbaden (R.
Henrich), Klinik fu¨r Innere Medizin und Gastroenterologie, Florence-
Nightingale-Krankenhaus Du¨sseldorf (J. Erckenbrecht), Abteilung In-
nere Medizin I, Medizinische Klinik und Poliklinik, Universita¨t
Tu¨bingen (S. Dette).
1. Smith JL, Graham DY. Variceal hemorrhage: a critical evaluation of sur-
vival analysis. Gastroenterology 1982;82:968–973.
2. Christensen E, Fauerholdt L, Schlichting P, Juhl E, Poulsen H, Tygstrup
N. Aspects of the natural history of gastrointestinal bleeding in cirrhosis
and the effect of prednisone. Gastroenterology 1981;81:944–952.
HEPATOLOGY, Vol. 40, No. 1, 2004 SCHEPKE ET AL. 71
3. McCormick PA, OЈKeefe C. Improving prognosis following a first variceal
haemorrhage over four decades. Gut 2001;49:682–685.
4. Garcia-Tsao G. Current management of the complications of cirrhosis and
portal hypertension: variceal hemorrhage, ascites and spontaneous bacte-
rial peritonitis. Gastroenterology 2001;120:726–748.
5. De Franchis R, ed. Portal Hypertension III. Proceedings of the third
Baveno international consensus workshop on definitions, methodology
and therapeutic strategies. Oxford, England: Blackwell Science; 2001.
6. Lebrec D, Poynard T, Hillon P, Benhamou JP. Propranolol for prevention
of gastrointestinal bleeding in patients with cirrhosis. N Engl J Med 1981;
7. D‘Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a
meta-analytic review. HEPATOLOGY 1995;22:332–354.
8. Poynard T, Cales P, Pasta L, Ideo G, Pascal JP, Pagliaro L, et al. Beta-
adrenergic-antagonist drugs in the prevention of gastrointestinal bleeding
in patients with cirrhosis and esophageal varices. N Engl J Med 1991;324:
9. Garcia-Pagan JC, Morillas R, Banares R, Albillos A, Villanueva C, Vila C,
et al. Propranolol plus placebo versus propranolol plus isosorbide-5-mono-
nitrate in the prevention of a first variceal bleed: a double-blind RCT.
HEPATOLOGY 2003;37:1260–1266.
10. Bosch J, Mastai R, Kravetz D, Bruix J, Gaya J, Rigau J, et al. Effects of
propranolol on azygos blood flowand hepatic and systemic hemodynamics
in cirrhosis. HEPATOLOGY 1984;4:1200–1205.
11. Garcia-Tsao G, Grace ND, Groszmann RJ, Conn HO, Bermann MM,
Patrick MJ, et al. Short-term effects of propranolol on portal venous pres-
sure. HEPATOLOGY 1986;6:101–106.
12. Vorobioff J, Picabea E, Villavicencio R, Puccini V, Rossi O, Bordato J, et
al. Acute and chronic hemodynamic effects of propranolol in unselected
cirrhotic patients. HEPATOLOGY 1987;7:648–653.
13. Lui HF, Stanley AJ, Forrest EW, Jalan R, Hislop WS, Mills PR, et al.
Primary prophylaxis of variceal hemorrhage: a randomized controlled trial
comparing band ligation, propranolol, and isosorbide mononitrate. Gas-
troenterology 2002;123:735–44.
14. D‘Amico G, Pagliaro L, Bosch J. Pharmacological treatment of portal
hypertension: an evidence-based approach. Semin Liver Dis 1999;19:475–
15. Abraczinskas D, Ookubo R, Grace N, Groszmann R, Bosch J, Garcia-Tsao
G, et al. Propranolol for the prevention of first variceal bleeding: a lifetime
commitment? HEPATOLOGY 2001;34:1096–1102.
16. Sauerbruch T, Wotzka R, Kopcke W, Harlin M, Heldwein W, Bayer-
dorffer E, et al. Prophylactic sclerotherapy before the first episode of
variceal hemorrhage in patients with cirrhosis. N Engl J Med 1988;319:
17. Laine L, Cook D. Endoscopic ligation compared with sclerotherapy for the
treatment of esophageal variceal bleeding. A meta-analysis. Ann Int Med
18. Sarin SK, Lamba G, Kumar M, Misra A, Murthy N. Comparison of
endoscopic ligation and propranolol for the primary prevention of variceal
bleeding. N Engl J Med 1999;340:988–993.
19. Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Tran-
section of the esophagus for bleeding esophageal varices. Br J Surg 1973;
20. Sarin SK, Guptan R, Jain A, SundaramK. Arandomized controlled trial of
endoscopic variceal band ligation for primary prophylaxis of variceal bleed-
ing. Eur J Gastroenterol Hepatol 1996;4:337–342.
21. Lay CS, Tsai Y, Teg C, Shyu WS, Guo W, Wu K, et al. Endoscopic variceal
ligation in prophylaxis of first variceal bleeding in cirrhotic patients with
high-risk esophageal varices. HEPATOLOGY 1997;25:1346–1350.
22. Lo GH, Lai KH, Cheng JS, Lin CK, Hsu PI, Chiang HAT. Prophylactic
banding ligation of high risk esophageal varices in patients with cirrhosis: a
prospective, randomized trial. J Hepatol 1999;31:451–456.
23. Imperiale T, Chalasani N. A meta-analysis of endoscopic variceal ligation
for primary prophylaxis of esophageal variceal bleeding. HEPATOLOGY
24. De BK, Goshal U, Das T, Santra A, Biswas P. Endoscopic variceal ligation
for primary prophylaxis of oesophageal variceal bleeding: preliminary re-
port of a randomized controlled trial. J Gastroenterol Hepatol 1999;14:
25. Angelico M, Carli L, Piat C, Gentile S, Capocaccia L. Effects of isosorbide-
5-mononitrate compared with propranolol on first bleeding and long-term
survival in cirrhosis. Gastroenterology 1997;113:1632–1639.
26. Pascal JP, Cales P, and the Multicentre Study Group. Propranolol in the
prevention of first upper gastrointestinal tract haemorrhage in patients
with cirrhosis of the liver and oesophageal varices. N Engl J Med 1987;
27. Italian Multicenter Project for Propranolol in Prevention of Bleeding.
Propranolol prevents first gastrointestinal bleeding in non-ascitic cirrhotic
patients. Final report of a multicenter randomized trial. J Hepatol 1989;
28. Burroughs AK, Patch D. Primary prevention of bleeding from esophageal
varices. N Engl J Med 1999;340:1033–1035.
29. Abraldes JG, Tarantino I, Turnes J, Garcia-Pagan JC, Rodes J, Bosch
J. Hemodynamic response to pharmacological treatment of portal hyper-
tension and long-term prognosis of cirrhosis. HEPATOLOGY 2003;37:902–
30. North Italian Endoscopic Club (NIEC) for the study and treatment of
esophageal varices. Prediction of first variceal hemorrhage in patients with
cirrhosis of the liver and esophageal varices. Aprospective multicenter trial.
N Engl J Med 1988;319:983–989.
31. Lo GH, Chen WC, Chen MH, Lin CP, Lo CC, Hsu PI, et al. Endoscopic
ligation vs. nadolol in the prevention of first variceal bleeding in patients
with cirrhosis. Gastrointest Endosc 2004;59:333–338.