Research Article

Efficacy of non-selective b-blockers as adjunct to

endoscopic prophylactic treatment for gastric variceal bleeding:
A randomized controlled trial
Hung-Hsu Hung1,2,3, Chen-Jung Chang1,2,4, Ming-Chih Hou1,2,3,⇑, Wei-Chih Liao1,2,5,
Che-Chang Chan1,2, Hui-Chun Huang1,2, Han-Chieh Lin1,2, Fa-Yauh Lee1,2, Shou-Dong Lee1,6,7
1
Department of Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan; 2Division of Gastroenterology, Department
of Medicine, Taipei Veterans General Hospital, Taipei, Taiwan; 3Endoscopy Center for Diagnosis and Treatment, Taipei Veterans General
Hospital, Taipei, Taiwan; 4Division of Gastroenterology and Hepatology, Department of Internal Medicine, Taipei Medical University Hospital,
Taipei, Taiwan; 5Division of Gastroenterology, Department of Medicine, Taipei Municipal Gan-Dau Hospital, Taipei, Taiwan; 6Deparement of
Medicine, School of Medicine, National Defense Medical Center, Taipei, Taiwan; 7Cheng Hsin General Hospital, Taipei, Taiwan

Background & Aims: Gastric variceal obturation (GVO) therapy
is the current treatment of choice for gastric variceal bleeding
(GVB). However, the efficacy of non-selective b-blockers (NSBB)
in the secondary prevention of GVB is still debatable. This study
aimed at evaluating the efficacy of additional NSBB to repeated
GVO in the secondary prevention of GVB.
Methods: From April 2007 to March 2011, 95 patients with GVB
after primary hemostasis using GVO were enrolled. Repeated
GVO were performed until GV eradication. Forty-eight and 47
patients were randomized into the GVO alone group (Group A)
and the GVO + NSBB group (Group B), respectively. Primary out-
comes in terms of re-bleeding and overall survival were analyzed
by multivariate analysis.
Results: After a mean follow-up of 18.10 months in group A, 26
patients bled and 20 died. In group B, 22 patients bled and 22
died after a mean follow-up of 20.29 months. The overall re-
bleeding and survival rates analyzed by the Kaplan–Meier
method were not different between the two groups (p = 0.336
and 0.936, respectively). The model of end-stage liver disease
(MELD) score and main portal vein thrombosis (MPT) were inde-
pendent determinants of re-bleeding while MPT and re-bleeding
Keywords: Beta-blocker; Cyanoacrylate injection; Gastric variceal bleeding;
Re-bleeding; Secondary prevention.
Received 18 August 2011; received in revised form 2 December 2011; accepted 3
December 2011; available online 17 January 2012
⇑ Corresponding author. Address: Department of Medicine, School of Medicine,
National Yang-Ming University, Taipei Veterans General Hospital, No. 201, Sec. 2,
Shih-Pai Rd., Taipei, Taiwan. Tel.: +886 2 2871 2121x3218; fax: +886 2 2874
5074.
E-mail address: mchou@vghtpe.gov.tw (M.-C. Hou).
Abbreviations: GV, gastric varices; EV, esophageal varices; GVB, gastric variceal
bleeding; EVB, esophageal variceal bleeding; TIPS, trans-jugular intra-hepatic
porto-systemic shunt; GOV, gastro-esophageal varices; IGV, isolated gastric var-
ices; GVO, gastric variceal obturation; NSBB, non-selective b-blockers; RCT, ran-
domized controlled trial; OPD, out-patient department; HCC, hepatocellular
carcinoma; MPT, main portal vein thrombosis; PHG, portal hypertensive gas-
tropathy; MELD, model of end-stage liver disease; EVL, esophageal variceal lig-
ation; HR, hazard ratio; CI, confidence interval; INR, international normalized
ratio; IRB, Institutional Review Board; SD, standard deviation; SPSS, Statistical
Package for Social Sciences.
were independent factors of mortality by time-dependent Cox-
regression model. Asthenia was the most common adverse event
and was higher in group B (p <0.001).
Conclusions: Adding NSBB therapy to repeated GVO provides no
benefit for the secondary prevention of bleeding and mortality in
patients with GVB.
Ó 2012 European Association for the Study of the Liver. Published
by Elsevier B.V. All rights reserved.
Introduction
Gastric varices (GVs) are less common than esophageal varices
(EVs) and occur in around 20% of patients with portal hyperten-
sion [1]. Although the incidence of gastric variceal bleeding (GVB)
is lower than esophageal variceal bleeding (EVB), GVB has a
higher re-bleeding rate, requires more blood transfusion, and
has a higher mortality rate [1–3]. The importance of preventing
GV re-bleeding cannot be over-emphasized.
Secondary prevention of GVB includes gastric variceal obtura-
tion (GVO), non-selective b-blockers (NSBB), trans-jugular intra-
hepatic porto-systemic shunt (TIPS), surgical shunts, and liver
transplant [4]. Among these therapies, GVO is shown to be very
effective with a consistent success rate of 90–100% in controlling
acute GVB [5–8]. However, cyanoacrylate injection is complicated
and can lead to bacteremia, sepsis, embolism, and endoscopic
injury [9,10].
NSBB is effective in reducing re-bleeding and death from EVs
by 40% and 20%, respectively [11]. However, its efficacy as sec-
ondary prevention of GVB has limited evidence [12]. In a previous
study, repeated GVO appeared to be more effective than NSBB as
secondary prevention and in improving survival of GVB patients
[12].
Although the recent meta-analysis and Baveno V consensus
suggests a combination of endoscopic and drug therapy is more
effective than either therapy alone in the secondary prevention
of EV bleeding [13], the evidence is limited for the combination
therapy in the secondary prevention of GV bleeding. As a whole,

Journal of Hepatology 2012 vol. 56 j 1025–1032

Research Article
Endoscopic-proved acute GVB
met the inclusion criteria (GOV2 or IGV1)
n = 110
From April 2007 to March 2010
Randomization
n = 95
Allocated to repeated GVO
(Group A)
n = 48
Lost to follow-up
n=2
Analyzed patients
n = 48
Fig. 1. The flow chart of the study.
the secondary prevention of GV bleeding remains sub-optimal in
comparison to the secondary prevention of EV bleeding [12,14–
16]. The current prospective randomized controlled trail (RCT)
aimed at evaluating whether additional NSBB can improve re-
bleeding and mortality rates in patients who undergo repeated
GVO after primary hemostasis.
Materials and methods
Patients
Cirrhotic patients with acute GVB were enrolled consecutively at the Taipei Veter-
ans General Hospital from April 2007 to March 2011 and the last randomized sub-
ject received the treatment protocol for at least 3 months. The inclusion criteria
were (1) age of 18–80 years; (2) type 2 gastro-esophageal varices (GOV2) or type
1 isolated gastric varices (IGV1); (3) GVB proven by endoscopy and GVO within
24 h of bleeding; and (4) stable hemodynamic condition for at least 3 days after
GVO.
The exclusion criteria were (1) previous treatment of GV, including endoscopic
therapy, NSBB, TIPS, or surgery; (2) contraindications to NSBB or cyanoacrylate injec-
tion; (3) serum total bilirubin >10 mg/dl; (4) hepatic encephalopathy grade III/IV; (5)
hepato-renal syndrome; (6) severe heart failure (NYHA Fc III/IV); (7) end-stage renal
disease under renal replacement therapy; (8) severe chronic obstructive pulmonary
disease; (9) proven malignancy other than hepatocellular carcinoma (HCC); (10)
pregnancy; (11) pacemaker use; and (12) refusal to participate.
The contraindications of NSBB were (1) bronchial asthma; (2) chronic obstruc-
tive pulmonary disease (COPD); (3) uncontrolled heart failure; (4) sinus bradycar-
dia <60/min; (5) heart block greater than first degree; and (6) cardiogenic shock.
The patients were randomized consecutively to two groups using numbered
envelopes that contained the treatment regimens, which were generated by a
system using computer-allocated random digit numbers. Randomization was
performed as long as hemodynamic conditions were stable, usually 3–5 days after
initial GVO for acute bleeding (Fig. 1). In group A, the patients underwent
repeated endoscopic cyanoacrylate injection every 3–4 weeks until GV eradica-
tion. Patients in group B also underwent the repeated GVO protocol as group A
but had additional NSBB treatment. The hospital’s Institutional Review Board
(IRB) approved the study and written informed consent was obtained from each
patient before enrollment.
1026 Journal of Hepatology 2012 vol. 56 j 1025–1032
• Previous treatment of GV
(n = 8, GVO = 4, NSBB = 3, TIPS = 1)
• Serum total bilirubin >10 mg/dl (n = 1)
• Hepatic encephalopathy grade III/IV
(n = 1)
• Severe heart failure (NYHA-Fc III/IV)
and end-stage renal disease under
renal replacement therapy (n = 1)
• Proven malignancy other than hepato-
cellular carcinoma (n = 1)
• Refusal (n = 3)
Allocated to repeated GVO
combined with NSBB (Group B)
n = 47
Lost to follow-up
n=0
Analyzed patients
n = 47
Gastric variceal obturation
Endoscopic intervention was performed by using an Olympus XQ-240 or XQ-260
endoscope (Olympus Optical Co. Ltd., Tokyo, Japan) and a 22-gauge disposable
injection needle (EIS 01943, Top Co., Tokyo, Japan). Each shot contained 0.5 ml
n-butyl-2-cyanoacrylate (Histoacryl blue, Braun, Melsungen, Germany) and
0.5 ml lipiodol (Guerbet Laboratory, Aulnay-Sous-Bris, France), with no more than
six shots per session. After GVO, omeprazole 40 mg intravenous infusion twice
daily was prescribed for 2 days, and then shifted to oral esomeprazole 40 mg
per day for another 12 days.
Beta-blockers therapy
Propranolol was started at a dose of 10 mg twice daily as soon as hemodynamic
stability was achieved after GVO. The dose was doubled every 3 days in the
hospital or every 7 days in the out-patient department (OPD) if the target
dose was not reached. The target dose was achieved when one of the following
criteria was satisfied: (1) >25% reduction of baseline heart rate; (2) a target
heart rate of approximately 55/min; and (3) a maximum dose of 320 mg per
day if well tolerated and with systolic blood pressure >90 mm Hg [4,17].
Compliance was carefully assessed by interview with patients or relatives on
every OPD follow-up.
Clinical assessment and follow-up
Information regarding presentation of upper gastrointestinal bleeding was care-
fully recorded from the patients and their relatives. Vital signs, biochemistry data,
amount of blood transfusion, infection status, medication, and endoscopic find-
ings were also recorded as detailed as possible.
If there were no symptoms or signs of bleeding, follow-up endoscopy was
regularly performed every 3–4 weeks and repeated GVO was performed until
the GVs were eradicated (Fig. 1). After eradication, surveillance endoscopy was
performed every 3 months. If bleeding symptoms or signs were noted during fol-
low-up, emergency endoscopy was performed to localize and treat the bleeding
source as soon as possible. Vasoactive agents, proton-pump inhibitor, antibiotics,
or balloon tamponade were allowed as treatment of acute upper gastrointestinal
bleeding according to the bleeding focus and indications. If re-bleeding occurred,
conservative treatment, endoscopic treatment, TIPS, or surgery were offered
based on the preference of the patients or their families.
 JOURNAL OF HEPATOLOGY
Table 1. Clinical characteristics of patients with gastric variceal bleeding.

Clinical characteristics GVO group (Group A) GVO + BB group (Group B) p

(n = 48) (n = 47)
Age (yr) 59.13 ± 13.45 59.19 ± 12.21 0.980
Sex (M/F) 35/13 34/13 0.950
Etiology of portal hypertension (viral/alcoholic/others) 28/11/9 32/11/4 0.336
Functional status (ECOG) (0/1/2/3) 14/23/11/0 13/17/16/1 0.297
Ascites (-/+) 24/24 20/27 0.242
Hepatic encephalopathy (-/+) 38/10 32/15 0.200
Albumin (g/dl) 2.94 ± 0.61 2.88 ± 0.69 0.624
Total bilirubin (mg/dl) 2.44 ± 3.00 2.01 ± 2.38 0.442
Creatinine (mg/dl) 1.29 ± 0.84 1.27 ± 1.05 0.939
Prothrombin time (INR) 1.29 ± 0.29 1.26 ± 0.28 0.525
Platelet (/mm3) 97.67 ± 62.32 107.23 ± 53.40 0.424
Child-Pugh score 7.96 ± 2.02 8.28 ± 2.37 0.482
Child-Pugh class (A/B/C) 14/23/11 12/22/13 0.847
MELD score 13.55 ± 5.85 12.48 ± 4.50 0.323
Primary/secondary GV 47/1 43/4 0.161
GOV2/IGV1 39/9 38/9 0.960
Form of GV (F1/F2/F3) 9/19/20 12/18/17 0.709
Size of GV (mm) 9.96 ± 2.93 10.17 ± 3.59 0.753
Active bleeding before randomization (-/+) 32/16 37/10 0.188
Infection (yes/no) 17/31 12/35 0.460
Blood transfusion (unit) 3.13 ± 3.04 3.60 ± 4.01 0.526
Hepatocellular carcinoma (-/+) 38/10 39/8 0.635
Main portal vein thrombosis (-/+) 41/7 41/6 0.797
Amount of cyanoacrylate mixture before randomization (ml) 3.44 ± 2.65 3.06 ± 1.62 0.328
Follow-up (month) 18.10 ± 15.31 20.29 ± 16.02 0.498
SD, standard deviation; INR, international normalized ratio; MELD, model of end-stage liver disease; GOV2, gastro-esophageal varix, type 2; IGV1, isolated gastric varix,
type 1, by Sarin’s classification.
All continuous variables were expressed as mean ± SD.

The diagnosis of cirrhosis was based on history, physical finding, biochemical
data, and imaging studies, including abdominal ultrasonography, computed
tomography (CT), or magnetic resonance imaging (MRI). All patients were fol-
lowed-up until their last hospital visit or death. Serum liver biochemistries, a-
fetoprotein levels, and abdominal ultrasonography were arranged every 3–
4 months because of national health insurance policy and the high prevalence
of HCC in Taiwan.
Definitions
The classification of GV was suggested by Sarin et al.: GOV1, varices continuous
with EV and extending along the lesser curve for approximately 2–5 cm below
the gastro-esophageal junction; GOV2, varices extending from the esophagus to
below the gastro-esophageal junction toward the fundus; IGV1, varices located
in the fundus; and IGV2, ectopic varices in the antrum, corpus, and around the
pylorus [1].
The size of GV was measured in comparison to the open diameter of the
biopsy forceps (FB-24K-1, Olympus medical systems Corp., Tokyo, Japan), which
was 8 mm between the tips.
The severity of cirrhosis was classified according to Pugh’s modification of
Child’s classification and models for end-stage liver disease (MELD) score
[18,19]. GVB was defined as either active GVB, including active spurting or oozing,
or stigmata of recent hemorrhage, including blood clot or ulcer, on the GV [18,19].
Clinically significant re-bleeding was defined as new onset hematemesis, cof-
fee-ground vomitus, hematochezia, or melena resulting in hospital admission,
blood transfusion, 3-g drop in hemoglobin, or death within 6 weeks after endo-
scopic treatment [4]. A single episode of clinically significant re-bleeding from
portal hypertensive sources after randomization was considered failure of sec-
ondary prevention and primary end point [4]. The secondary end point was
mortality.
Eradication was defined as non-visualization of patent GV. If mucosal prom-
inence could not be differentiated from scar tissue or patent GV, endoscopic ultra-
sound was performed to assess the obliteration [8].
Statistical analysis
Chi-square analysis or Fisher’s exact test was used to compare categorical vari-
ables while the Mann–Whitney U-test was used to compare continuous variables.
Cumulative bleeding-free rates or overall survival rates were estimated by the
Kaplan–Meier method and compared by Cox’s proportional hazards model.
Covariate of re-bleeding was considered a time-dependent event in the multivar-
iate Cox-regression.
The re-bleeding rate was assumed to be 40% for repeated GVO [8,20]. Addi-
tional NSBB could reduce re-bleeding rates by approximately 25% in cirrhotic
patients with EV who underwent repeated sclerotherapy [21], thus the re-bleed-
ing rate after repeated GVO with additional NSBB was set at 15%. The type I (a)
error and type II (b) error were set to 0.05 and 0.2, respectively. The proposed
sample size was 47 cases in each group as calculated by MedCalc (MedCalc for
Windows, version 4.20.011, Mariakerke, Belgium).
Variables with statistical significance (p <0.05) or proximate to it (p <0.1) by
univariate analysis underwent multivariate analysis using Cox-regression model.
A two-tailed p <0.05 was considered statistically significant. All statistical analy-
ses were performed using the Statistical Package for Social Sciences (SPSS 17.0 for
Windows, SPSS, Inc., Chicago, IL, USA).

Journal of Hepatology 2012 vol. 56 j 1025–1032 1027

Research Article
Table 2. Re-bleeding and mortality in patients with gastric bleeding (GVO vs. GVO + NSBB).

GVO group (Group A) GVO + BB group (Group B) p

(n = 48) (n = 47)
Hemostatic outcome
Number of re-bleeding patients 26 22§ 0.609
Amount of cyanoacrylate mixture after randomization (ml) 0.625 ± 1.44 0.47 ± 1.40 0.591
Sources of re-bleeding†
GV or GV ulcer 17 15 0.886
EV 3 1 0.617
PHG 3 2 0.663
Undetermined 2 1 0.583
Mortality 20 22 0.766
Cause of death
Bleeding 6 6
Hepatic failure 10 9
Sepsis 4 5
Others* 0 2
All continuous variables were expressed as mean ± SD.
Bleeding death was defined as death within 6 weeks of re-bleeding.
 
In group A, one patient bled due to gastric ulcer. In group B, one and 2 patients bled due to reflux esophagitis and duodenal ulcer, respectively.
⁄
In group B, one patient died of hyperkalemia-induced arrhythmia and another patient died of hepatocellular carcinoma rupture.
§
Four patients experienced re-bleeding before the stable dose was reached. If the 4 patients experienced re-bleeding before reaching the stable dose was excluded for
analysis, the re-bleeding rate (GVO vs. GVO + NSBB = 26/48 vs. 18/43, p = 0.128) and overall survival (GVO vs. GVO + NSBB = 20/48 vs. 20/43, p = 0.953) were still not
statistical significant.

Results common bleeding source was GV or GV ulcers, followed by portal

Baseline clinical characteristics
A total of 110 cirrhotic patients with GOV2 or IGV1 were enrolled
and 15 patients were excluded (Fig. 1). The 95 patients who ful-
filled the inclusion criteria were randomized as soon as they were
hemodynamically stable, and 48 and 47 patients were random-
ized to groups A and B, respectively. The days between index
bleeding and randomization were 4.06 ± 0.84 and 4.26 ± 0.97,
respectively (p = 0.301). In terms of baseline demographic data,
there was no significant difference in clinical data, follow-up
duration, etiology of portal hypertension, liver function reserve,
classification of GV, extent and form of GV, infection status,
severity of bleeding, and association with HCC between the two
groups (Table 1).
Hemodynamic outcomes at randomization and follow-up
Blood pressure and heart rate were not different at randomiza-
tion (Supplementary Table 1). After NSBB treatment, the heart
rate in group B was lower than that in group A (66.3 ± 12.5 vs.
83.5 ± 12.5 beats per min, p <0.001). During follow-up, the heart
rate was reduced in both group A (6.13 ± 14.08 bpm, p = 0.004)
and group B (23.83 ± 14.73 bpm, p = 0.001) but was greater in
group B (p <0.001).
Re-bleeding and mortality
During the follow-up period, 26 and 22 patients developed gas-
trointestinal bleeding in group A and group B, respectively
(Table 2). The amount of cyanoacrylate mixture used after ran-
domization was not different between the two groups. The most
hypertensive gastropathy (PHG) and EV.
After a mean follow-up of 18.10 months, 20 patients in group
A died. In group B, 22 patients died after a mean follow-up of
20.29 months. The overall mortality during follow-up period
was similar (p = 0.766). Hepatic failure was the most common
cause of death in both groups, followed by bleeding and sepsis.
Factors associated with re-bleeding
The overall cumulative re-bleeding free rates at 0.5, 1, 2, and
3 years were 93.8%, 77.1%, 58.1%, and 47.0%, respectively, in
group A and 89.4%, 76.6%, 62.1%, and 52.4%, respectively, in group
B (p = 0.336) (Fig. 2A). In patients without concomitant HCC, the
cumulative re-bleeding free rates at 0.5, 1, 2, and 3 years were
86.8%, 71.1%, 56.9%, and 41.5%, respectively, in group A and
87.2%, 79.2%, 64.2%, and 50.0%, respectively, in group B
(p = 0.416) (Fig. 2B). There was no difference in re-bleeding-free
rates between the two groups. The re-bleeding rates were not dif-
ferent on per-protocol analysis (data not shown) either.
Univariate analysis showed that MELD score, infection, HCC,
and main portal vein thrombosis (MPT) were associated with
higher incidence of re-bleeding (Table 3). On multivariate analy-
sis, MELD score and MPT were independent determinants of re-
bleeding.
Factors associated with overall survival
The overall survival rates at 0.5, 1, 2, and 3 years were 89.6%,
83.3%, 72.7%, and 52.5%, respectively, in group A and 89.4%,
78.7%, 67.4%, and 51.8%, respectively, in group B (p = 0.936)
(Fig. 3A). In patients without concomitant HCC, the overall sur-
vival rates at 0.5, 1, 2, and 3 years were 92.1%, 86.1%, 63.5%,

1028 Journal of Hepatology 2012 vol. 56 j 1025–1032

 JOURNAL OF HEPATOLOGY
A B
GVO GVO
Cumulative re-bleeding-free rate (%)

Cumulative re-bleeding-free rate (%)

100 GVO + NSBB 100 GVO + NSBB

80 80

60 60

40 40

20 20
p = 0.336 p = 0.416
0 0
0 6 12 18 24 30 36 42 48 54 0 6 12 18 24 30 36 42 48 54
Follow-up (months) Follow-up (months)
Patients at risk Patients at risk
GVO 48 22 17 15 12 8 6 3 1 GVO 38 21 17 15 12 8 6 3 1
GVO + NSBB 47 24 22 17 12 10 8 2 1 GVO + NSBB 39 22 20 16 11 9 7 2 1

Fig. 2. The cumulative re-bleeding free rates. (A) The cumulative re-bleeding free rates in patients with GVB undergoing secondary prevention. There was no difference
between the two groups (p = 0.336). (B) The cumulative re-bleeding free rates in patients without concomitant hepatocellular carcinoma undergoing secondary prevention.
There was no difference between the two groups (p = 0.416).

Table 3. Univariate and multivariate analysis of re-bleeding in patients with gastric variceal bleeding undergoing secondary prevention (n = 95).

Univariate analysis Multivariate analysis

Variable Hazard ratio (95% CI) p Hazard ratio (95% CI) p
GVO/GVO + NSBB 0.758 (0.429-1.338) 0.339
Age (yr) 0.997 (0.975-1.019) 0.768
Albumin (g/dl) 0.761 (0.465-1.246) 0.278
Bilirubin (mg/dl) 1.106 (0.984-1.242) 0.091
Prothrombin time (INR) 1.096 (0.359-3.343) 0.873
Hepatic encephalopathy (+/-) 1.226 (0.746-2.146) 0.382
Child-Pugh score 1.021 (0.896-1.164) 0.753
MELD score 1.068 (1.006-1.134) 0.032 1.008 (1.022-1.158) 0.008
Active bleeding before randomization (+/-) 1.513 (0.820-2.793) 0.186
Infection (+/-) 2.070 (1.157-3.703) 0.014
HCC (+/-) 2.577 (1.340-4.950) 0.005
MPT (+/-) 3.344 (1.613-6.897) 0.001 4.049 (1.908-8.621) <0.001
GOV2/IGV1 1.233 (0.613-2.480) 0.557
INR, international ratio; MELD, model of end-stage liver disease; HCC, hepatocellular carcinoma; MPT, main portal vein thrombosis; GOV2, gastro-esophageal varix, type 2;
IGV1, isolated gastric varix, type 1.

and 46.2%, respectively, in group A and 87.2%, 76.5%, 62.4%, and
52.1%, respectively, in group B (p = 0.912) (Fig. 3B). There was
not difference in overall survival between the two groups regard-
less of concomitant HCC. The mortality rates were still not differ-
ent on per-protocol analysis (data not shown).
Univariate analysis showed that MELD score, HCC, MPT, active
bleeding before randomization, and re-bleeding were associated
with higher mortality (Table 4). While covariate of re-bleeding
was considered a time-dependent event in the multivariate
Cox-regression model, MPT and re-bleeding remained as inde-
pendent determinants of GV mortality.
Adverse events
Adverse events were recorded within one month after randomi-
zation when titration of NSBB was achieved in most cases. There
was no difference between the two groups (p = 0.062) (Supple-
mentary Table 2). Asthenia was the most common adverse event
and was higher in group B (p <0.001). There was no difference in
gastro-intestinal discomfort and infection between the two
groups. Eight and three patients in group B discontinued and
reduced NSBB, respectively due to side effects.
Discussion
GVB is characterized by high re-bleeding rate and mortality rates,
particularly in patients with GOV2 or IGV1 [1,17]. Pharmacologic
and endoscopic treatments, TIPS, and surgery have been
performed to prevent re-bleeding in GVB patients [4,6,8,12,22].
Randomized controlled trials (RCT) to test the best methods for
secondary prevention of GVB are scarce. Previous RCTs have

Journal of Hepatology 2012 vol. 56 j 1025–1032 1029

Research Article
A B
100
Cumulative survival rate (%)
80
60
40
20
p = 0.936
0 0
0 6 12 18 24 30 36
Follow-up (months)
Patients at risk
GVO 48 35 25 16
GVO + NSBB 47 35 30 22
Fig. 3. The cumulative overall survival rates. (A) The cumulative overall survival rates in GVB patients undergoing secondary prevention. There was no difference between
the two groups (p = 0.936). (B) The cumulative overall survival rates in GVB patients without concomitant HCC undergoing secondary prevention (p = 0.912).
shown that GVO is superior to endoscopic ligation in preventing
re-bleeding [8,23]. Thus, it is recommended as the current first-
line treatment for acute GVB, particularly for those with bleeding
from GOV2 and IGV1 [17]. However, the re-bleeding rate of GV
remains high after GVO and the best therapy for the secondary
prevention of GVB is still not satisfactory. This is fast emerging
as one of the most important issues to be investigated [4,12,17].
Sarin et al. have found that NSBB alone is not as effective as
repeated GVO in preventing GV re-bleeding [12]. Moreover,
patients treated with NSBB are associated with higher mortality
compared to repeated GVO [12]. Repeated GVO is superior to
NSBB alone in term of re-bleeding prevention and of survival in
patients with GVB [12]. In patients with EVs, the addition of NSBB
to endoscopic ligation is the best method for preventing EV re-
bleeding [4,17]. However, the role of NSBB in addition to repeated
GVO is uncertain. Therefore, the current study is the first one
designed to test if the addition of NSBB to repeated GVO can pro-
vide better benefit in preventing re-bleeding and in reducing
mortality in GVB patients, with special emphasis on GOV2 or
IGV1 because they have the highest re-bleeding risk and benefit
most from GVO when compared to those of GOV1 [4,10].
In the current study, NSBB in addition to repeated GVO is not
better than repeated GVO alone in preventing re-bleeding in GVB
patients. Possible reasons are: (1) most patients with GV already
have large gastro-renal porto-systemic shunting and/or segmental
portal hypertension such that TIPS, and therefore NSBB may be not
as effective in this category [24,25], or (2) portal pressure is lower
in patients with GVB than in patients with EVB [26] and thus, portal
pressure may be not as critical in GVB patients or a 20% reduction in
hepatic venous pressure gradient (HVPG) may be inadequate to
prevent re-bleeding [12,26]. Although the target dose in the cur-
rent study is lower than in other NSBB studies in Western coun-
tries, there may be higher sensitivity to beta-blocking and
hypotensive effects among Chinese than in Caucasians [27,28].
The ineffectiveness of NSBB cannot be attributed to an inade-
quate dose because the target dose of NSBB is higher than in a
previous study that shows effective beta-blockade and reduction
of re-bleeding in patients with EVB [29]. Actually, a small-sized
retrospective study and a recent RCT have also shown that NSBB
cannot decrease the risk of GV re-bleeding and does not improve
1030
GVO GVO
GVO + NSBB 100 GVO + NSBB
Cumulative survival rate (%)
80
60
40
20
p = 0.912
42 48 54 0 6 12 18 24 30 36 42 48 54
Follow-up (months)
Patients at risk
14 11 9 5 3 GVO 38 32 24 15 13 10 8 4 2
17 14 11 5 2 GVO + NSBB 39 32 27 21 16 13 10 4 2
the overall survival [12,28]. Moreover, ineffectiveness of addi-
tional NSBB in the current study persists regardless of inten-
tion-to-treat or per-protocol analysis.
A small-scale RCT shows that TIPS is better than GVO in pre-
venting GV re-bleeding [22]. However, that study included 50% of
patients with GOV1, with clinical characteristics that resemble EV
and are associated with lower re-bleeding risk than GOV2 and
IGV1 [1,10]. TIPS is better in reducing portal pressure in patients
with EV than in those with GV [26].
In univariate analysis, MELD score, infection, HCC, and MPT
were associated with higher incidence of re-bleeding. Although
antibiotics are routinely used in patients with acute variceal
bleeding in the study hospital, the infection rate associated with
acute gastric variceal bleeding remains at 30.5% and is a risk fac-
tor of re-bleeding. In multivariate analysis, MELD score and MPT
are the independent factors of re-bleeding, which are consistent
with the current and previous studies [8,28,30–33].
Aside from re-bleeding, mortality is not different between
repeated GVO alone or GVO + NSBB either. Univariate analysis
showed that MELD score, HCC, MPT, active bleeding before ran-
domization and re-bleeding were associated with higher mortal-
ity. MELD score is superior to Child–Pugh classification in
predicting short- and long-term mortality after acute variceal
bleeding [31,34,35]. It can be used to stratify the risk of mortality
and re-bleeding instead of Child–Pugh classes [31]. While covari-
ate of re-bleeding was considered a time-dependent event, multi-
variate analysis showed re-bleeding and MPT are independent risk
factors of mortality in the present study. Therefore, the importance
of secondary prevention of GVB and further investigation of a bet-
ter strategy to manage GVB cannot be over-emphasized enough.
Although embolic complications of GVO have been found in a
previous study [36], a serious complication has not occurred in
the current study. The overall side effects or complications are
not significantly different between the two groups, but asthenia
is higher in patients treated with additional NSBB.
The limitations of the study were no HVPG measurement and
the lack of a double-blinded comparison. The study was designed
in order to assess the usefulness of the addition of a treatment
decreasing the absolute risk of bleeding of 25% or more compared
with a baseline treatment with a 40% absolute risk. Any smaller
Journal of Hepatology 2012 vol. 56 j 1025–1032
 JOURNAL OF HEPATOLOGY
Table 4. Univariate and multivariate analysis of overall survival in patients with gastric variceal bleeding undergoing secondary prevention (n = 95).

Univariate analysis Multivariate analysis

Variable Hazard ratio (95% CI) p Hazard ratio (95% CI) p
GVO/GVO + NSBB 1.025 (0.559-1.881) 0.936
Age (yr) 1.015 (0.992-1.038) 0.203
Albumin (g/dl) 0.948 (0.587-1.531) 0.827
Bilirubin (mg/dl) 1.056 (0.931-1.197) 0.395
Prothrombin time (INR) 1.100 (0.329-3.676) 0.877
Hepatic encephalopathy (+/-) 1.233 (0.715-2.123) 0.452
Child-Pugh score 1.075 (0.944-1.224) 0.273
MELD score 1.063 (1.002-1.128) 0.044
Active bleeding before randomization (+/-) 1.946 (1.032-3.663) 0.040
Infection (+/-) 1.623 (0.862-3.049) 0.134
HCC (+/-) 3.546 (1.815-6.944) <0.001
MPT (+/-) 6.024 (2.770-13.158) <0.001 3.390 (1.499-7.692) 0.003
GOV2/IGV1 1.194 (0.370-1.898) 0.671
Re-bleeding (+/-)* 7.839 (3.976-15.457) <0.001 6.719 (3.378-13.364) <0.001
INR, international ratio; MELD, model of end-stage liver disease; HCC, hepatocellular carcinoma; MPT, main portal vein thrombosis; GOV2, gastro-esophageal varix, type 2;
IGV1, isolated gastric varix, type 1.
⁄
Covariate of re-bleeding was considered a time-dependent event in multivariate analysis.

decrease in absolute risk of bleeding could not be excluded Hospital Institutional Review Board (96-04-01), and Clinical trial
according to the present trial design. Further studies will be nec- number (NCT00567216).
essary to elucidate this point.
Acknowledgements
To date, the current study is the only RCT with a large case
number to provide evidence on additional NSBB in patients
The authors thank Mrs. Pui-Ching Lee for her assistance with the
who undergo repeated GVO. Adding NSBB does not prevent re-
statistical analysis and Hai-Shin Lee and Yi-Chian Chiang for the
bleeding or reduce mortality in GVB patients who undergo
technical assistance with GVO and patient follow-up.
repeated GVO but increases adverse effects. Therefore, repeated
GVO is the first choice for secondary prevention but additional
NSBB is not suggested. Active bleeding and re-bleeding represent Supplementary data
risks of mortality, therefore better methods of bleeding preven-
tion should be explored and investigated. Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jhep.2011.12.021.
Authors’ contribution

Hung-Hsu Hung, study concept and design, data analysis and
interpretation, and manuscript drafting; Ming-Chih Hou, study
concept and design, GVO and NSBB dosing, data analysis and
interpretation, and manuscript revision; Chen-Jung Chang and
Wei-Chih Liao, patient enrollment and follow-up, and data collec-
tions; Che-Chang Chan and Hui-Chun Huang, patient enrollment;
Han-Chieh Lin, Fa-Yauh Lee, and Shou-Dong Lee, study concept
and design.
Conflict of interest
The authors who have taken part in this study declared that they
do not have anything to disclose regarding funding or conflict of
interest with respect to this manuscript.
Financial support
This study was supported by grants from the National Science
Council (NSC 96-2314-B-075-037-MY3), Taipei Veterans General
References
[1] Sarin SK, Lahoti D, Saxena SP, Murthy NS, Makwana UK. Prevalence,
classification and natural history of gastric varices: a long-term follow-up
study in 568 portal hypertension patients. Hepatology 1992;16:1343–1349.
[2] Korula J, Chin K, Ko Y, Yamada S. Demonstration of two distinct subsets of
gastric varices. Observations during a seven-year study of endoscopic
sclerotherapy. Dig Dis Sci 1991;36:303–309.
[3] Mathur SK, Dalvi AN, Someshwar V, Supe AN, Ramakantan R. Endoscopic and
radiological appraisal of gastric varices. Br J Surg 1990;77:432–435.
[4] De Franchis R. Portal hypertension V. In: Proceedings of the fifth Baveno inter
-national consensus workshop. Chichester, West Sussex: Wiley & Sons; 2011.
[5] Huang YH, Yeh HZ, Chen GH, Chang CS, Wu CY, Poon SK, et al. Endoscopic
treatment of bleeding gastric varices by N-butyl-2-cyanoacrylate (Histo-
acryl) injection: long-term efficacy and safety. Gastrointest Endosc
2000;52:160–167.
[6] Sarin SK, Jain AK, Jain M, Gupta R. A randomized controlled trial of
cyanoacrylate versus alcohol injection in patients with isolated fundic
varices. Am J Gastroenterol 2002;97:1010–1015.
[7] Akahoshi T, Hashizume M, Shimabukuro R, Tanoue K, Tomikawa M, Okita K,
et al. Long-term results of endoscopic Histoacryl injection sclerotherapy for
gastric variceal bleeding: a 10-year experience. Surgery 2002;131:
S176–S181.
[8] Tan PC, Hou MC, Lin HC, Liu TT, Lee FY, Chang FY, et al. A randomized trial of
endoscopic treatment of acute gastric variceal hemorrhage: N-butyl-2-
cyanoacrylate injection versus band ligation. Hepatology 2006;43:690–697.

Journal of Hepatology 2012 vol. 56 j 1025–1032 1031

Research Article
[9] Chen WC, Hou MC, Lin HC, Yu KW, Lee FY, Chang FY, et al. Bacteremia after
endoscopic injection of N-butyl-2-cyanoacrylate for gastric variceal bleed-
ing. Gastrointest Endosc 2001;54:214–218.
[10] Ryan BM, Stockbrugger RW, Ryan JM. A pathophysiologic, gastroenterologic,
and radiologic approach to the management of gastric varices. Gastroenter-
ology 2004;126:1175–1189.
[11] 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. An analysis of data and
prognostic factors in 589 patients from four randomized clinical trials.
Franco-Italian Multicenter Study Group. N Engl J Med 1991;324:1532–1538.
[12] Mishra SR, Chander Sharma B, Kumar A, Sarin SK. Endoscopic cyanoacrylate
injection versus beta-blocker for secondary prophylaxis of gastric variceal
bleed: a randomised controlled trial. Gut 2010;59:729–735.
[13] Gonzalez R, Zamora J, Gomez–Camarero J, Molinero LM, Banares R, Albillos
A. Meta-analysis: combination endoscopic and drug therapy to prevent
variceal rebleeding in cirrhosis. Ann Intern Med 2008;149:109–122.
[14] Sarin SK. Long-term follow-up of gastric variceal sclerotherapy: an eleven-
year experience. Gastrointest Endosc 1997;46:8–14.
[15] Trudeau W, Prindiville T. Endoscopic injection sclerosis in bleeding gastric
varices. Gastrointest Endosc 1986;32:264–268.
[16] Soehendra N, Nam VC, Grimm H, Kempeneers I. Endoscopic obliteration
of large esophagogastric varices with bucrylate. Endoscopy 1986;18:
25–26.
[17] Garcia-Tsao G, Bosch J. Management of varices and variceal hemorrhage in
cirrhosis. N Engl J Med 2010;362:823–832.
[18] Pugh RN, Murray-Lyon IM, Dawson JL, Pietroni MC, Williams R. Transection
of the oesophagus for bleeding oesophageal varices. Br J Surg
1973;60:646–649.
[19] Hsu CY, Lin HC, Huang YH, Su CW, Lee FY, Huo TI, et al. Comparison of the
model for end-stage liver disease (MELD), MELD-Na and MELDNa for
outcome prediction in patients with acute decompensated hepatitis. Dig
Liver Dis 2010;42:137–142.
[20] De Franchis R. Portal hypertension III. In: Proceedings of the third Baverno
international consensus workshop on definitions, methodology, and thera-
peutic strategies. Oxford, Malden, MA: Blackwell Science; 2001.
[21] Elsayed SS, Shiha G, Hamid M, Farag FM, Azzam F, Awad M. Sclerotherapy
versus sclerotherapy and propranolol in the prevention of rebleeding from
oesophageal varices: a randomised study. Gut 1996;38:770–774.
[22] Lo GH, Liang HL, Chen WC, Chen MH, Lai KH, Hsu PI, et al. A prospective,
randomized controlled trial of transjugular intrahepatic portosystemic
shunt versus cyanoacrylate injection in the prevention of gastric variceal
rebleeding. Endoscopy 2007;39:679–685.
[23] Lo GH, Lai KH, Cheng JS, Chen MH, Chiang HT. A prospective, randomized
trial of butyl cyanoacrylate injection versus band ligation in the manage-
ment of bleeding gastric varices. Hepatology 2001;33:1060–1064.
1032 Journal of Hepatology 2012 vol. 56 j 1025–1032
[24] Sanyal AJ, Freedman AM, Luketic VA, Purdum 3rd PP, Shiffman ML, DeMeo J,
et al. The natural history of portal hypertension after transjugular intrahe-
patic portosystemic shunts. Gastroenterology 1997;112:889–898.
[25] Watanabe K, Kimura K, Matsutani S, Ohto M, Okuda K. Portal hemodynamics
in patients with gastric varices. A study in 230 patients with esophageal and/
or gastric varices using portal vein catheterization. Gastroenterology
1988;95:434–440.
[26] Tripathi D, Therapondos G, Jackson E, Redhead DN, Hayes PC. The role of the
transjugular intrahepatic portosystemic stent shunt (TIPSS) in the manage-
ment of bleeding gastric varices: clinical and haemodynamic correlations.
Gut 2002;51:270–274.
[27] Hayes PC, Davis JM, Lewis JA, Bouchier IA. Meta-analysis of value of
propranolol in prevention of variceal haemorrhage. Lancet 1990;336:
153–156.
[28] Wu CY, Yeh HZ, Chen GH. Pharmacologic efficacy in gastric variceal
rebleeding and survival: including multivariate analysis. J Clin Gastroenterol
2002;35:127–132.
[29] Lo GH, Lai KH, Cheng JS, Chen MH, Huang HC, Hsu PI, et al. Endoscopic
variceal ligation plus nadolol and sucralfate compared with ligation alone
for the prevention of variceal rebleeding: a prospective, randomized trial.
Hepatology 2000;32:461–465.
[30] Berzigotti A, Gilabert R, Abraldes JG, Nicolau C, Bru C, Bosch J, et al.
Noninvasive prediction of clinically significant portal hypertension and
esophageal varices in patients with compensated liver cirrhosis. Am J
Gastroenterol 2008;103:1159–1167.
[31] Bambha K, Kim WR, Pedersen R, Bida JP, Kremers WK, Kamath PS. Predictors
of early re-bleeding and mortality after acute variceal haemorrhage in
patients with cirrhosis. Gut 2008;57:814–820.
[32] Abraldes JG, Villanueva C, Banares R, Aracil C, Catalina MV, Garci APJC, et al.
Hepatic venous pressure gradient and prognosis in patients with acute
variceal bleeding treated with pharmacologic and endoscopic therapy. J
Hepatol 2008;48:229–236.
[33] Hou MC, Lin HC, Liu TT, Kuo BI, Lee FY, Chang FY, et al. Antibiotic prophylaxis
after endoscopic therapy prevents rebleeding in acute variceal hemorrhage:
a randomized trial. Hepatology 2004;39:746–753.
[34] Chalasani N, Kahi C, Francois F, Pinto A, Marathe A, Bini EJ, et al. Model for
end-stage liver disease (MELD) for predicting mortality in patients with
acute variceal bleeding. Hepatology 2002;35:1282–1284.
[35] Amitrano L, Guardascione MA, Bennato R, Manguso F, Balzano A. MELD score
and hepatocellular carcinoma identify patients at different risk of short-term
mortality among cirrhotics bleeding from esophageal varices. J Hepatol
2005;42:820–825.
[36] Hou MC, Lin HC, Lee HS, Liao WC, Lee FY, Lee SD. A randomized trial of
endoscopic cyanoacrylate injection for acute gastric variceal bleeding:
0.5 mL versus 1.0 mL. Gastrointest Endosc 2009;70:668–675.