Addition of propranolol and isosorbide mononitrate to endoscopic variceal ligation does not reduce varicealrebleeding incidence.

1. Addition of Propranolol and Isosorbide Mononitrate to Endoscopic
Variceal Ligation Does Not Reduce Variceal Rebleeding Incidence
ASHISH KUMAR,* SANJEEV KUMAR JHA,†
PRAVEEN SHARMA,* SAROJ DUBEY,†
PANKAJ TYAGI,†
BARJESH CHANDER SHARMA,†
and SHIV KUMAR SARIN*,†
*Department of Medical Hepatology, Institute of Liver and Biliary Sciences, New Delhi, India; and †
Department of Gastroenterology, G B Pant Hospital, University of
Delhi, New Delhi, India
See related article, Rocha ECV et al, on page
988 in CGH.
BACKGROUND & AIMS: Endoscopic variceal ligation
(EVL) and propranolol are standard secondary prophylaxis
therapies for variceal bleeding. Addition of isosorbide mono-
nitrate (ISMN) to propranolol improves its hemodynamic ef-
ficacy; we investigated whether a combination of EVL and
propranolol/ISMN was more effective than EVL alone for
secondary prophylaxis. METHODS: Patients with a prior
variceal bleed were randomly assigned to groups given a com-
bination (n ϭ 88) of EVL, propranolol (dose titrated to reduce
heart rate to 55 beats per minute), and ISMN (40 mg/day) or
EVL alone (n ϭ 89). Primary end points were rebleeding or
death; secondary end points were new complications of portal
hypertension or serious adverse effects. RESULTS: The actu-
arial probabilities of rebleeding 2 years after therapy were 27%
in the combination group and 31% in the EVL alone group (P
ϭ .822). Two patients in the combination group and 3 patients
in the EVL alone group died during the study period (P ϭ
.682); no deaths were caused by variceal hemorrhage. In cir-
rhotic patients, the actuarial probabilities of rebleeding were
24% and 30%, respectively (P ϭ .720). Secondary end points
were comparable between groups. In multivariate analyses,
presence of ascites (P ϭ .003), serum albumin Ͻ 3.3 g/dL (P ϭ
.008), and hepatic venous pressure gradients Ն 18 mm Hg (P
ϭ .009) were independent risk factors for variceal rebleeding.
CONCLUSIONS: EVL alone is sufficient to prevent
variceal rebleeding in cirrhotic and noncirrhotic patients
with history of variceal bleeding. Addition of propranolol
and ISMN to EVL does not reduce the incidence of
variceal rebleeding but increases severe adverse effects.
Risk factors for rebleeding include ascites, low serum al-
bumin, and high hepatic venous pressure gradients.
The lifetime prevalence of varices in cirrhotics has
been reported to be as high as 80%–90%. Variceal
bleeding is a serious complication, and it occurs in 30%–
40% of the patients with varices. The mortality of acute
variceal hemorrhage has decreased significantly but still
ranges between 15% and 20% even with optimal manage-
ment.1–3 Patients with portal hypertension, who have
bled from esophageal varices, have a 60%–80% risk of
variceal rebleeding at 1 year, if no therapy for prevention
of variceal rebleeding is instituted.4,5 Rebleeding from
esophageal varices is associated with a high morbidity
and mortality. Therefore, patients with variceal bleeding
should be given effective long-term therapy to prevent
recurrence of variceal bleeding. The current recommen-
dation, according to the Baveno IV consensus confer-
ence,6 for prevention of rebleeding in patients of cirrho-
sis, is the use of ␤-blockers or endoscopic variceal ligation
(EVL) or both. The combination therapy (␤-blocker plus
EVL) was thought to be better than either therapy alone;
however, definite recommendation for its use could not
be made because of lack of a sufficient number of trials
showing its superiority.6
Conceptually, reduction of portal pressure by combi-
nation pharmacotherapy and local obliteration of varices
should be more effective and synergistic than either ther-
apy administered alone. The hemodynamic response to
␤-blockers is achieved in only approximately one third of
the patients, and addition of nitrates has been found to
work synergistically to reduce further the portal pressure
and improve the overall response rates to pharmacother-
apy.7 We conducted a randomized controlled trial of a
combination of EVL and ␤-blocker plus nitrate vs EVL
alone in the secondary prophylaxis of variceal rebleed
from esophageal varices. The main objectives were to
evaluate, in patients who had bled from varices in the
past, whether endoscopic and combined drug therapy
regimen reduces the incidence of variceal rebleeds and
improves the survival compared with ligation alone in
this group of patients. The adverse effects of the drug
therapy were also carefully monitored.
Patients and Methods
Patients
Patients presenting to the Liver Disease Clinic at
the G B Pant Hospital, New Delhi, with a history of
Abbreviations used in this paper: EVL, endoscopic variceal ligation;
HVPG, hepatic venous pressure gradient; ISMN, isosorbide mononi-
trate.
© 2009 by the AGA Institute
0016-5085/09/$36.00
doi:10.1053/j.gastro.2009.05.049
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
GASTROENTEROLOGY 2009;137:892–901

2. hematemesis and/or melena and proven to have esopha-
geal varices as the bleeding source on upper gastrointes-
tinal (GI) endoscopy were included in the study. The
exclusion criteria were as follows: (1) a history of under-
going endoscopic sclerotherapy, EVL, or cyanoacrylate
injection; (2) a history of surgery for portal hypertension;
(3) coexisting malignancy; (4) severe cardiopulmonary or
renal disease; (5) a history of severe adverse effects or
contraindications to ␤-blockers like bronchial asthma,
uncontrolled diabetes mellitus, heart failure, peripheral
vascular disease, prostatic hypertrophy, arterial hypoten-
sion (systolic blood pressure Ͻ100 mm Hg), bradycardia
(basal heart rate Ͻ55 beats per minute), or complete
heart block; and (6) refusal to give consent to participate
in the trial. A written informed consent was taken from
all the included patients. The Institutional Ethics Com-
mittee approved the study protocol. The trial was regis-
tered with ClincalTrials.gov vide NCT00766805.
Baseline Evaluation
The included patients were broadly categorized
into cirrhotic patients or non-cirrhotic patients. Cirrho-
sis was diagnosed on the basis of clinical, biochemical,
histologic, or ultrasonographic evidence. An effort was
made to determine the etiology of cirrhosis. Hepatitis B
and C virus markers, autoimmune markers, serum ceru-
loplasmin, urinary copper, and serum iron and ferritin
assays were performed wherever needed. History of alco-
hol abuse was sought from all the patients. The severity
of liver disease in cirrhotics was assessed by Child–Tur-
cotte–Pugh (CPT) score. Non-cirrhotic patients were fur-
ther categorized into non-cirrhotic portal fibrosis and
extrahepatic portal vein obstruction. Non-cirrhotic por-
tal fibrosis was diagnosed when varices were present, and
there was no evidence of thrombosis in the splenoportal
axis on ultrasonography and no evidence of cirrhosis on
liver biopsy.8 Extrahepatic portal vein obstruction was
diagnosed when a portal cavernoma was detected by
ultrasonography, and there were no signs of cirrhosis.8
The size of esophageal varices was assessed according
to Conn’s classification9: grade 1, visible only during 1
phase of respiration or on performance of valsalva ma-
neuver; grade 2, visible during both phases of respiration;
grade 3, 3–6 mm; and grade 4, Ͼ6 mm. The size of the
largest varix was assessed by comparison with the shaft of
the biopsy forceps (3 mm) or with the distance between
the open jaws of the biopsy forceps (6 mm) in the lumen
of the lower 2 to 3 cm of the esophagus.
Hepatic Venous Pressure Gradient
Measurement
Hepatic venous pressure gradient (HVPG) was
measured before randomization. HVPG measurement
was done after overnight fast and under antibiotic cover.
Under local anesthesia, a 7-French central venous cathe-
ter (Arrow Medical, Athens, TX) was placed in the right
femoral vein or internal jugular vein under fluoroscopic
guidance, using the Seldinger technique. HVPG was mea-
sured by the standard technique10 in which a balloon
catheter was introduced into the right hepatic vein under
fluoroscopic guidance. The zero reference point was set at
the midaxillary point. The free hepatic venous pressure
was obtained by keeping the catheter free into the lumen
of the hepatic vein. The balloon of the catheter was then
inflated to wedge the lumen of the hepatic vein. Presence
of wedging was confirmed by absence of reflux into the
inferior vena cava, after the injection of 2 mL intravenous
contrast, and appearance of a sinusoidogram. The pres-
sure tracing at this juncture showed absence of wave-
forms, and the pressure was labeled as wedged hepatic
venous pressure. HVPG was determined by subtracting
free from wedged hepatic venous pressures (HVPG ϭ
wedged hepatic venous pressure Ϫ free hepatic venous
pressure). All measurements were performed in triplicate.
If the difference between the 2 readings was more than 1
mm Hg, all the readings were discarded, and a fresh set of
measurements were taken. The normal value of the
HVPG in our hemodynamic laboratory is between 1 and
4 mm Hg.
Randomization
This was an open-label, randomized controlled
trial. Patients were randomized after HVPG, using com-
puter generated random numbers, to receive either of the
2 therapies (EVL or EVL plus drugs). The randomization
sequence remained with the statistician, and the se-
quence remained concealed from the investigators until
the intervention was assigned. Those presenting with an
acute bleed underwent an emergency variceal band liga-
tion and 5-day therapy with terlipressin or somatostatin
prior to HVPG and randomization.
Endoscopic Variceal Ligation
Patients assigned to the EVL group underwent
variceal band ligation at the first endoscopy session
within the next 24 hours. Ligation was done using a
multiband ligator. In each session, as many bands as
possible were placed on the varices in the lower 5–7 cm of
the esophagus, the number varying from 2 to 10. EVL was
done at intervals of 3–4 weeks until the varices were
completely obliterated or reduced to grade 1 size and
could not be banded any further. Once varices were
eradicated, repeat endoscopy was done at monthly inter-
vals for 3 months and then at 3-month intervals to check
for recurrent varices. Further EVL sessions were under-
taken for recurrent varices. Patients were routinely ad-
vised to have liquids on the day of EVL and subsequently
placed on semisolids and solids. They were prescribed 40
mg of pantoprazole twice a day and a suspension of
sucralfate 4 times per day. Patients with ascites were
prescribed oral antibiotics. In general, attempts were
made to correct the coagulation disorders by giving fresh
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
September 2009 EVL VS EVL؉ DRUGS FOR SECONDARY PROPHYLAXIS DRUGS 893

3. frozen plasma. All adverse effects including chest pain,
dysphagia, fever, and GI bleeding were recorded.
EVL Plus Drug Therapy
Patients randomized to the EVL plus drugs ther-
apy received EVL plus ␤-blocker (propranolol) and ni-
trate (isosorbide mononitrate [ISMN]). Within 24 hours,
these patients were started on propranolol after a base-
line electrocardiogram and cardiac evaluation. Treatment
was started with propranolol at a dose of 40 mg twice a
day. The heart rate and blood pressure were checked after
12 to 24 hours. The dose of propranolol was increased at
increments of 20 to 40 mg per day until the patient
achieved a heart rate of 55 beats per minute (bpm) or a
maximum dose of 320 mg/day was achieved. The dose
was reduced if any of the following occurred: systolic
blood pressure (BP) Ͻ90 mm Hg, heart rate Ͻ55 bpm, or
other serious adverse effects. After attaining a stable dose
of propranolol, ISMN was added at a dose of 10 mg twice
a day. The dose was escalated at increments of 10–20
mg/day until a maximum dose of 40 mg/day was reached
or the patient experienced adverse effects such as head-
ache, dizziness, or hypotension. Full dosage of ␤-blocker
and ISMN was achieved within 2–3 weeks of randomiza-
tion, before the second session of EVL. Further EVL
sessions and follow-ups were done as in the EVL arm.
Patients receiving propranolol were monitored daily until
␤-blockade was adequate then monthly for the first 3
months and every 3 months subsequently. Drug compli-
ance was ascertained by interviewing the patient and by
measuring the resting heart rate.
End Points
Primary end points of the study were bleeding and
death. Secondary end points included complications (be-
cause of bleeding, causes related to the underlying liver
disease or unrelated causes), upper GI tract bleeding
because of causes not related to portal hypertension, and
the development of serious adverse effects that required
the discontinuation of therapy.
Bleeding
Patients presenting with active upper GI bleeding
during the study were admitted and subjected to upper
GI endoscopy within 12 hours to determine the source of
bleeding. Upper GI bleeding was diagnosed and classified
using Baveno IV consensus conference criteria.6 Bleeding
from esophageal varices was diagnosed if active bleeding,
a “white nipple,” or a clot was seen at endoscopy or if
there was blood in the stomach in a patient with esoph-
ageal varices and no other potential bleeding source.
Gastric variceal bleed was diagnosed if active bleeding or
a clot was seen on gastric varices on endoscopy or if there
was evidence of recent bleeding in a patient with a gastric
varix and the bleeding had no other possible cause. Bleed-
ing was attributed to portal hypertensive gastropathy
(PHG) if distinct lesions of the gastric mucosa were
present, and there was endoscopic evidence of an active
bleeding lesion, assessed after washing or removal of
clots, and there was no evidence of bleeding from esoph-
ageal, gastric, or ectopic varices. Esophageal ulcer bleed as
a result of band ligation was diagnosed if there was active
bleeding from the ulcer at the site of banding or if there
was an adherent clot on the esophageal ulcer with ab-
sence of any other potentially bleeding lesions in the
upper GI endoscopy. Bleeding was categorized as clini-
cally significant when the heart rate was Ͼ100 bpm, the
systolic blood pressure was Ͻ100 mm Hg, or there was a
postural drop of Ͼ20 mm Hg and transfusion require-
ment of more than 2 U of blood in 24 hours.6 All patients
with bleeding were started on vasoactive drugs (terlipres-
sin or somatostatin) immediately. An emergency session
of EVL was performed in these patients to stop the
bleeding. The vasoactive drugs were continued for 5 days.
Sample Size Calculation and Statistical
Analysis
This study was designed to compare the risks of
variceal rebleed as calculated by Kaplan–Meier curves for
EVL plus drug therapy vs EVL alone. The probability of
rebleed on EVL alone therapy was assumed to be 40% at
1 year. It was hypothesized that EVL plus drug therapy
would halve the risk to 20%. Using a 1-tailed test with an
␣ value of .05 and power (1 Ϫ ␤) of 0.80, the required
sample size would be 148, ie, 74 in each group.
Quantitative data were expressed as mean (ϮSD) or
median (range) and analyzed using Student 2-tailed t test
or Mann–Whitney U test. Qualitative data were analyzed
by Fisher exact test or Pearson ␹2 test. Kaplan–Meier plot
and log-rank test was used to compare cumulative prob-
abilities of rebleeding and death. Subgroup analysis was
done for cirrhotic and non-cirrhotic patients separately.
Univariate and multivariate analyses were performed to
assess the variables predicting rebleed. Statistical analysis
was done using the SPSS 15.0 statistical package (SPSS
Inc, Chicago, IL). The statistical analysis of the entire
data sets pertaining to efficacy (specifically primary and
major secondary efficacy end points) and safety (specifi-
cally, serious adverse events as defined in federal guide-
lines) have been independently confirmed by a biostatis-
tician who is not employed by the corporate entity.
Results
Patients
One thousand nine hundred twenty-one new pa-
tients with portal hypertension were admitted in our
hospital from October 2002 through December 2006
(Supplementary Figure 1). Of these, 653 had history of
variceal bleed in past. Four hundred seventy-six patients
were excluded for the following reasons: history of un-
dergoing endoscopic sclerotherapy, EVL, or cyanoacrylate
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
894 KUMAR ET AL GASTROENTEROLOGY Vol. 137, No. 3

4. injection before presenting to our hospital, 375; history
of surgery for portal hypertension, 12; coexisting malig-
nancy, 27; severe cardiopulmonary or renal disease, 7;
history of severe adverse effects or contraindications to
␤-blockers, 42; and refusal to give consent to participate
in the trial, 13. Hence, a remaining 177 patients were
included in the trial.
One hundred fifty-one patients (85%) were cirrhotic
patients, whereas the remaining 26 (15%) were non-cir-
rhotic patients (extrahepatic portal vein obstruction, 12,
and non-cirrhotic portal fibrosis, 14). Eighteen (10%)
patients had presented with acute variceal bleed (ie,
within 5 days), and the rest had presented with past
history of variceal bleed (Table 1).
Randomization and Follow-Up
Eighty-eight patients were randomized to the
EVLϩdrugs group, and 89 patients were randomized to
the EVL alone group. The baseline characteristics of pa-
tients were similar in the 2 groups (Table 1).
Ten patients were lost to follow-up before the second
session of EVL (4 in EVLϩdrugs group and 6 in EVL
alone group, P ϭ .536). These patients were excluded,
and, thus, subsequent analysis was done on 167 patients.
The mean follow-up was 15 Ϯ 12 months (range, 0.3–52
months) in the EVLϩdrugs group and 15 Ϯ 11 months
(range, 1–55 months) in the EVL group (P ϭ .662). EVL
was done at an average of 4-weeks in both the groups.
The median dose of propranolol used was 120 (range,
40–320) mg/day and of ISMN was 40 (range, 20–40)
mg/day in the EVLϩdrugs group.
Primary End Points
During a mean follow-up of 15 Ϯ 12 months, 30
patients developed rebleed: 14 of 84 (17%) in the
EVLϩdrugs group and 16 of 83 (19%) in the EVL group
Table 1. Baseline Characteristics of Study Groups
Parameters EVLϩdrugs EVL alone P value
All patients (n ϭ 177) (n ϭ 88) (n ϭ 89)
Mean age (ϮSD), y 42 (14) 41 (14) .716
Sex, n (%) .667
Males 75 (85) 78 (88)
Females 13 (15) 11 (12)
Cause of portal hypertension, n (%) .832
Cirrhosis 76 (86) 75 (84)
Non-cirrhotic portal hypertension 12 (14) 14 (16)
EHPVO 6 6
NCPF 6 8
Last bleed, n (%) .590
Յ5 days ago 11 (12) 7 (8)
6Ϫ42 days ago 43 (49) 45 (50)
Ͼ42 days ago 34 (39) 37 (42)
Mean (ϮSD) grade of esophageal varices 3.1 (0.8) 3.2 (0.7) .292
Patients with gastric varices, n (%) 51 (58) 46 (52) .451
Patients with PHG, n (%) 40 (45) 48 (54) .294
Mean (ϮSD) platelet count, 103/mm3 155 (52) 160 (59) .755
Cirrhotic patients (n ϭ 151) (n ϭ 76) (n ϭ 75)
Etiology of cirrhosis, n (%) .354
Alcohol 33 (43) 30 (40)
Viral 13 (17) 20 (27)
Others 30 (40) 25 (33)
CTP class, n (%) .292
A 35 (46) 26 (35)
B 31 (41) 34 (45)
C 10 (13) 15 (20)
Mean (ϮSD) CTP score 7.3 (2.0) 7.8 (2.1) .138
Ascites, n (%) 24 (32) 33 (44) .133
Encephalopathy, n (%) 1 (1) 4 (5) .209
Median (range) serum bilirubin, mg/dL 1.5 (0.5Ϫ16.9) 1.6 (0.2Ϫ14.4) .527
Mean (ϮSD) serum albumin, g/dL 3.2 (0.5) 3.2 (0.6) .707
Prothrombin time prolongation, n (%) .559
0–3 s 42 (56) 35 (47)
4–6 s 17 (22) 21 (28)
Ͼ6 s 17 (22) 19 (25)
Mean (ϮSD) serum creatinine, mg/dL 0.9 (0.5) 0.9 (0.4) .659
Mean (ϮSD) HVPG, mm Hg 18.0 (5.7) 17.7 (5.3) .772
EVL, endoscopic variceal ligation; SD, standard deviation; EHPVO, extrahepatic portal vein obstruction; NCPF, non-cirrhotic portal fibrosis; PHG,
portal hypertensive gastropathy; CTP, Child–Turcotte–Pugh; HVPG, hepatic venous pressure gradient.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
September 2009 EVL VS EVL؉ DRUGS FOR SECONDARY PROPHYLAXIS DRUGS 895

5. (P ϭ .842). The actuarial probability of rebleed at 2 years
was 27% in the EVLϩdrugs group and 31% in the EVL
group (P ϭ .822, log-rank test) (Figure 1).
Of the 14 bleeds in the EVLϩdrugs arm, 6 (43%) were
from esophageal varices prior to eradication (including
post-EVL variceal ulcers), 5 (36%) were from recurrent
esophageal varices, 3 (21%) were from gastric varices, and
none were from portal hypertensive gastropathy. In the
EVL group, of the 16 bleeds, 11 (69%) patients bled from
esophageal varices (including post-EVL variceal ulcers), 4
(25%) from gastric varices, 1 (6%) from portal hyperten-
sive gastropathy, and none from recurrent esophageal
varices.
Five patients died during the study period (2 [2%] in
the EVLϩdrugs group and 3 [4%] in the EVL group, P ϭ
.682) (Figure 2). None of the patients died of variceal
hemorrhage.
In the subgroup analysis of only cirrhotic patients
(Table 2), 11 of 72 (15%) rebled in the EVLϩdrugs group,
and 13 of 69 (19%) rebled in the EVL group (P ϭ .656).
The actuarial probability of rebleed at 2 years was 24% in
the EVLϩdrugs group and 30% in the EVL group (P ϭ
.720, log-rank test) (Figure 3). One (1%) patient died in
the EVLϩdrugs group, whereas 3 (4%) died in the EVL
group (P ϭ .359).
In another subgroup analysis of only those patients
who had their last bleed within 42 days (ie, patients who
had the highest chance of rebleed), no significant differ-
ence was found in the rate of rebleed in the 2 groups: 8
of 52 (15%) rebled in the EVLϩdrugs group, and 10 of 48
(21%) rebled in the EVL group (P ϭ .604).
Secondary End Points
Nine patients developed new ascites (4/84 [5%] in
EVLϩdrugs group vs 5/83 [6%] in EVL group; P ϭ .746),
15 patients had episode(s) of hepatic encephalopathy
(8/84 [10%] in EVLϩdrugs group vs 7/83 [8%] in EVL
group; P ϭ 1), 4 patients had episode(s) of spontaneous
bacterial peritonitis (2/84 [2%] in EVLϩdrugs group vs
2/83 [2%] in EVL group; P ϭ 1), and 6 patients had
episode(s) of hepatorenal syndrome (3/84 [4%] in EVLϩ
drugs group vs 3/83 [4%] in EVL group; P ϭ 1). Fifty-six
patients were hospitalized (26/84 [31%] in EVLϩdrugs
group vs 30/83 [36%] in EVL group; P ϭ .515) for bleed,
ascites, spontaneous bacterial peritonitis, hepatorenal
syndrome, encephalopathy, fever, and others (Table 2).
Adverse Effects of Therapy
Ten patients developed transient chest pain fol-
lowing EVL, which resolved in 1 day (3/84 [4%] in
EVLϩdrugs group and 7/83 [8%] in EVL alone group;
P ϭ .211). Nine (11%) patients developed adverse effects to
drugs in the EVLϩdrugs groups. Adverse effects of pro-
pranolol were as follows: dizziness because of bradycardia
(n ϭ 2, improved with dose reduction), hypotension (n ϭ
2, improved with dose reduction), and dyspnea (n ϭ 1,
improved with stoppage of propranolol). Adverse effect
of ISMN was headache (n ϭ 4, required dose reduction in
2 and stoppage in 2).
Recurrence of Varices
Variceal eradication could be achieved in 77 of
167 patients (44/84 [52%] in the EVLϩdrugs group and
33/83 [40%] in the EVL group, P ϭ .121). The mean
sessions of EVL required to eradicate esophageal varices
were similar in both the groups (4.6 Ϯ 1.9 EVLϩdrugs vs
4.6 Ϯ 1.6 EVL sessions, P ϭ .970). In 28 of 77 (36%)
patients, there was variceal recurrence, defined as appear-
ance or an increase in the grade of varices after achieving
successful eradication. The recurrence rate was similar in
both the groups (16/44 [36%] in the EVLϩdrugs group
and 12 of 33 [36%] in the EVL group, P ϭ 1). Recurrences
of varices were further treated by repeat sessions of EVL.
Effect on Gastric Varices and PHG
At baseline, 97 of 177 (55%) patients had gastric
varices. On follow-up, 91 of 167 (54%) patients had gas-
tric varices that were similar to baseline (P ϭ 1). There
was no change in the frequency of gastric varices in either
of the groups (EVLϩdrugs: baseline 51/88 [58%] vs fol-
Figure 2. Kaplan–Meier graph showing cumulative probability of death
in the 2 groups, in all patients.
Figure 1. Kaplan–Meier graph showing cumulative probability of re-
bleed in the 2 groups, in all patients.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
896 KUMAR ET AL GASTROENTEROLOGY Vol. 137, No. 3

6. low-up 50/84 [60%], P ϭ .878; EVL alone: baseline 46/89
[52%] vs follow-up 41/83 [49%], P ϭ .879).
At baseline, 88 of 177 (50%) patients had PHG. There
was significant increase in the frequency of PHG on
follow-up: 109 of 167 (65%, P ϭ .004). The increase in
frequency of PHG was similar in both the groups
(EVLϩdrugs: baseline 40/88 [45%] vs follow-up 51/84
[61%], P ϭ .049; EVL alone: baseline 48/89 [54%] vs
follow-up 58/83 [70%], P ϭ .041).
Predictors of Rebleed
On univariate analysis, the baseline predictors of
rebleed were as follows: presence of ascites, encephalop-
athy, high bilirubin, low albumin, deranged prothrombin
time, high HVPG, alcohol as etiology, and high CTP score
(Table 3). CTP class of B or C as compared with A was the
strongest risk factor of rebleed (relative risk, 5.04 [95% CI:
1.57–16.11]). All the individual components of CTP score
also were significant risk factors for rebleed. Apart from
these, baseline HVPG also correlated with rebleed risk
(Tables 3 and 4). Alcohol as etiology of cirrhosis as
compared with others also had more chances of rebleed;
however, it did not reach statistical significance (P ϭ
.071). On multivariate analysis (Table 5), the only signif-
icant independent risk factors for rebleed were as follows:
presence of ascites (P ϭ .003), serum albumin Ͻ3.3 g/dL
(P ϭ .008), and baseline HVPG Ն18 mm Hg (P ϭ .009).
Discussion
The results of this large and novel study show that
addition of propranolol and nitrate does not decrease the
probability of variceal rebleed in patients being treated by
EVL. Further addition of these drugs, in fact, leads to
drug-related adverse events and results in some morbid-
Figure 3. Kaplan–Meier graph showing cumulative probability of re-
bleed in the 2 groups, in the subgroup analysis of cirrhotic patients.
Table 2. Primary and Secondary End Points
Parameters EVLϩdrugs EVL alone P value
All patients (n ϭ 167) (n ϭ 84) (n ϭ 83)
Mean (ϮSD) follow-up 15 (12) 15 (11) .662
Patients rebled, n (%) 14 (17) 16 (19) .842
Site of rebleed, n (%)
Esophageal varices before eradication 6 (7) 11 (13) .212
Recurrent esophageal varices 5 (6) 0 (0) .059
Gastric varices 3 (4) 4 (5) .720
PHG 0 (0) 1 (1) .497
Actuarial probability of rebleed at 2 years, % 27 31 .822
Patients died, n (%) 2 (2) 3 (4) .682
Complications, n (%)
New ascites 4 (5) 5 (6) .746
Hepatic encephalopathy 8 (10) 7 (8) 1
SBP 2 (2) 2 (2) 1
HRS 3 (4) 3 (4) 1
Hospitalizations 26 (31) 30 (36) .515
Adverse effects of therapy, n (%)
Chest pain after EVL 3 (4) 7 (8) .211
Adverse effects because of drugs 9 (11) 0 (0) .003
Effect on varices and PHG
Esophageal variceal eradication, n (%) 44 (52) 33 (40) .121
EVL sessions required to eradicate, n (ϮSD) 4.6 (1.9) 4.6 (1.6) .970
Esophageal variceal recurrence, n (%) 16/44 (36) 12/33 (36) 1
Gastric varices, n (%) 50 (60) 41 (49) .215
PHG, n (%) 51 (61) 58 (70) .256
Cirrhotic patients (n ϭ 141) (n ϭ 72) (n ϭ 69)
Patients rebled, n (%) 11 (15) 13 (19) .656
Actuarial probability of rebleed at 2 years, % 24 30 .720
Patients died, n (%) 1 (1) 3 (4) .359
EVL, endoscopic variceal ligation; SD, standard deviation; PHG, portal hypertensive gastropathy; SBP, spontaneous bacterial peritonitis; HRS,
hepatorenal syndrome.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
September 2009 EVL VS EVL؉ DRUGS FOR SECONDARY PROPHYLAXIS DRUGS 897

7. ity. We have also shown that significant independent risk
factors for rebleed are poor liver function (as manifested
by ascites and low albumin) and high portal pressure (as
manifested by high HVPG).
Our study was adequately powered to show non-supe-
riority of EVL plus drugs strategy over EVL alone strat-
egy. We enrolled a sufficient number of patients, 167
bleeders (including 151 cirrhotic patients), which was
sufficient as per the sample size calculations of 148 pa-
tients. The baseline characteristics of the patients were
well matched in both the groups, and the follow-up
duration was adequate to allow meaningful outcome.
The Baveno IV consensus conference recommended
that patients with cirrhosis who had not received primary
prophylaxis could receive nonselective ␤-blocker, EVL, or
both.6 Although it was hypothesized that combination of
nonselective ␤-blocker plus EVL was probably the best
alternative, it concluded that more trials were needed
before a recommendation could be made. Combination
of EVL plus ␤-blocker seemed to be a rational approach
because ␤-blockers would theoretically protect against
rebleeding before variceal obliteration and could prevent
variceal recurrence.
␤-Blockers have been shown to be effective in reducing
the incidence of variceal rebleeding; however, it has been
shown that up to two thirds of the patients may be
nonresponders to ␤-blockers. The addition of ISMN has
been demonstrated to enhance the effect of ␤-blockers in
Table 3. Predictors of Rebleed
Parameters Bleeders Non-bleeders P value
All patients (n ϭ 167) (n ϭ 30) (n ϭ 137)
Mean age (ϮSD), y 44 (16) 40 (14) .187
Sex, n (%) .255
Males 28 (93) 115 (84)
Females 2 (7) 22 (16)
Cause of portal hypertension, n (%) .420
Cirrhotic patients 24 (80) 117 (85)
Non-cirrhotic patients 6 (20) 20 (15)
EHPVO 1 11
NCPF 5 9
Last bleed, n (%) .728
Յ5 days ago 4 (13) 12 (9)
6 to 42 days ago 14 (47) 70 (51)
Ͼ42 days ago 12 (40) 55 (40)
Mean (ϮSD) grade of esophageal varices 3.1 (0.8) 3.2 (0.7) .468
Patients with gastric varices, n (%) 19 (63) 73 (53) .418
Patients with PHG, n (%) 18 (60) 63 (46) .226
Mean (ϮSD) platelet count, 103/mm3 153 (54) 160 (56) .721
Protocol, n (%)
EVLϩdrugs 14 (47) 70 (51)
EVL 16 (53) 67 (49) .691
Cirrhotic patients (n ϭ 141) (n ϭ 24) (n ϭ 117)
Etiology of cirrhosis, n (%) .071
ALD 14 (58) 44 (38)
Other 10 (42) 73 (62)
CTP class, n (%) .006
A 3 (12) 56 (48)
B 15 (63) 45 (38)
C 6 (25) 16 (14)
Mean (ϮSD) CTP score 8.6 (2.1) 7.2 (1.9) .001
Ascites, n (%) 16 (67) 36 (31) .002
Encephalopathy, n (%) 3 (12) 2 (1) .035
Median (range) serum bilirubin, mg/dL 1.9 (0.6Ϫ10.5) 1.5 (0.3Ϫ14.4) .034
Mean (ϮSD) serum albumin, g/dL 2.9 (0.6) 3.3 (0.5) .002
Prothrombin time prolongation, n (%) .032
0–6 seconds 14 (58) 94 (80)
Ͼ6 seconds 10 (42) 23 (20)
Mean (ϮSD) serum creatinine, mg/dL 1.0 (0.5) 0.9 (0.4) .183
Mean (ϮSD) HVPG, mm Hg (n ϭ 115) 20.2 (4.5) 17.3 (5.3) .014
Protocol, n (%) .656
EVLϩdrugs 11 (46) 61 (52)
EVL 13 (54) 56 (48)
SD, standard deviation; EHPVO, extrahepatic portal vein obstruction; NCPF, non-cirrhotic portal fibrosis; PHG, portal hypertensive gastropathy;
EVL, endoscopic variceal ligation; ALD, alcoholic liver disease; CTP, Child–Turcotte–Pugh; HVPG, hepatic venous pressure gradient.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
898 KUMAR ET AL GASTROENTEROLOGY Vol. 137, No. 3

8. reducing portal pressure through the decrease of hepatic
vascular resistance.11 A controlled study showed that
cirrhotic patients receiving propranolol and ISMN had a
lower variceal rebleeding rate compared with patients
who only received propranolol.12 Hence, the combination
of ␤-blockers and ISMN rather than using ␤-blockers
alone has been used in this study as the portal pressure
reducing strategy.
Although the benefit of EVL for secondary prophylaxis
of esophageal variceal bleeding is supported by the med-
ical literature, there has been no consensus regarding the
technical aspects of the procedure. The success of EVL
therapy depends on its technique, and the importance of
technical aspects in influencing the success cannot be
overemphasized. The technique most endoscopists fol-
low13 involves application of bands at the gastroeso-
phageal junction and progressing upwards in a helical
motion for approximately 5–8 cm. There is also no
consensus on when the band ligation sessions should be
repeated. Most endoscopists used to favor an interval of
1–2 weeks,14–18 but, recently, many authors propose that
an interval of 3–4 weeks is more suitable, given that
unhealed ulcers induced by ligation are frequently noted
within 2 weeks of ligation.19–21 One study actually iden-
tified that longer intervals between EVL sessions resulted
in a reduced risk of variceal rebleeding.21 Longer inter-
banding intervals (Ͼ3 weeks) promote optimal results. A
possible explanation for this is that a longer interbanding
interval allows varices to partially recur, allowing them to
be more effectively suctioned and rebanded. Moreover, it
results in a decrease in band-related bleeding, which
occurs because of the ulcers that form at banding sites
contributing to bleeding, especially when banding is per-
formed in rapid succession with short intervals. If EVL is
performed optimally, the addition of drugs to EVL would
not be able to add substantially to the results of EVL
alone, given their equal efficacy in preventing rebleed.22–24
Our results support this hypothesis.
Prior to our study, 2 small randomized controlled
trials had compared the efficacy of combined therapy vs
EVL alone.25,26 In the trial by de la Pena et al,25 efficacy of
EVL combined with nadolol was compared with EVL
alone for the secondary prophylaxis of variceal bleeding.
In a median follow-up of 16 months, nadolol plus EVL
was found to reduce the incidence of variceal rebleeding,
and there was a trend to reduce variceal recurrence com-
pared with EVL alone. Our results differ from the results
of this study because of many reasons. First, in the study
by Pena et al,25 the interval between EVL sessions was
10–12 days. Because of more frequent EVL sessions at
shorter intervals, the chances of rebleeding were higher.
Second, up to one third of the patients had received
endoscopic sclerotherapy as the initial treatment for the
control of active bleeding. Endoscopic sclerotherapy is
inferior to EVL, and, hence, protective effect of ␤-blocker
again comes into play. Third, EVL was performed begin-
ning at the cardia (ie, stomach) or immediately under it.
EVL in the stomach leads to higher chances of bleeding
ulcers.27 In our study, we performed EVL sessions at
intervals of 3–4 weeks that resulted in fewer EVL-induced
ulcer bleeds. We did not perform sclerotherapy in our
patients. Moreover, we performed variceal banding start-
ing approximately a centimeter above the gastroesopha-
geal junction. Hence, the protective effect of ␤-blockers
in our study was minimal, if at all.
In the trial by Lo et al,26 a combination of ␤-blocker,
EVL, and sucralfate was compared with EVL alone in a
group of 122 patients. The combination therapy was
proved to be more effective than EVL alone in terms of
prevention of variceal recurrence and upper GI rebleeding
as well as variceal rebleeding. Survival remained the same
in both the groups. However Lo et al’s study has been
criticized for having observer bias, selection bias, lack of
portal pressure correlations, and difficulty in evaluation
of role of sucralfate without an appropriate control
group.28 We used sucralfate and proton pump inhibitor
in both the groups to prevent EVL-induced ulcer bleeds.
A meta-analysis published recently5 also claims that
combination of endoscopic and drug therapy is superior
to endoscopic therapy alone. However, the major prob-
lem with this meta-analysis is that sclerotherapy and EVL
have been clubbed together as “endoscopic therapy,” and
Table 4. Relative Risks of Factors Found to Be Significant
Predictors of Rebleed on the Univariate Analysis
Factors Relative risk (95% CI) P value
Etiology: ALD vs others 2.00 (0.96Ϫ4.19) .071
Ascites: present vs absent 3.42 (1.57Ϫ7.44) .002
HE: present vs absent 3.88 (1.72Ϫ8.79) .035
Serum bilirubin, mg/dL: Ն2.5 vs
Ͻ2.5
2.77 (1.37Ϫ5.59) .008
Serum albumin, g/dL: Ն3.3 vs
Ͻ3.3
0.34 (0.15Ϫ0.77) .007
PT prolonged by: Ն7 vs Ͻ7
seconds
2.34 (1.15Ϫ4.76) .032
HVPG, mm Hg: Ն18 vs Ͻ18 2.53 (1.12Ϫ5.73) .024
CTP class: B/C vs A 5.04 (1.57Ϫ16.11) .001
ALD, alcoholic liver disease; HE, hepatic encephalopathy; HVPG, he-
patic venous pressure gradient; CTP, Child–Turcotte–Pugh; PT, pro-
thrombin time.
Table 5. Factors Predictive of Rebleed on Multivariate
Analysis
Factors Odds ratio (95% CI) P value
Ascites: present vs absent 4.60 (1.70Ϫ12.48) .003
Serum albumin, g/dL: Ն3.3 vs
Ͻ3.3
0.25 (0.09Ϫ0.69) .008
HVPG, mm Hg: Ն18 vs Ͻ18 4.06 (1.43Ϫ11.55) .009
NOTE. All factors found significant by univariate analysis were entered
into multivariate analysis except CTP score because its individual
components were entered.
HVPG, hepatic venous pressure gradient.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
September 2009 EVL VS EVL؉ DRUGS FOR SECONDARY PROPHYLAXIS DRUGS 899

9. there is no subgroup analysis of EVL plus drugs vs EVL
alone. Clubbing of the 2 endoscopic modalities (sclero-
therapy and EVL) undoubtedly would lead to inferior
results of “endoscopic therapy” because sclerotherapy is
known to be inferior to EVL in the prevention of variceal
rebleeding.
In the present study, there was no difference in the
development of complications like new ascites, hepatic
encephalopathy, spontaneous bacterial peritonitis, he-
patorenal syndrome, or hospitalizations. Adverse ef-
fects because of EVL (eg, transient chest pain) were also
similar in both the groups. However, 9 (11%) patients
developed adverse effects to drugs in the EVL plus
drugs group that included dizziness, bradycardia, hy-
potension, and dyspnea because of propranolol, and
headache because of ISMN. Thus, addition of ␤-blocker
and nitrate adds to morbidity by causing drug-related
adverse effects without leading to any superiority of the
combination over EVL alone. Whereas we had earlier
shown that a combination of ␤-blocker and nitrates
alone is as effective as EVL alone,24 the addition of EVL to
the combination of drugs was not found to be any
superior to EVL alone.
The rate of recurrence of esophageal varices was similar
in both the groups. Moreover, there was no change in the
frequency of gastric varices after EVL in either group.
There was a significant increase in the frequency of PHG,
which was similar in both the groups. Thus, addition of
drugs did not reduce the development of gastric varices
or PHG.
We found that the only significant independent risk
factors for variceal rebleed were as follows: presence of
ascites, serum albumin Ͻ3.3 g/dL, and baseline HVPG
Ն18 mm Hg. In fact, our study confirms the previous
findings that poor liver function and high portal pressure
are significant independent predictors of variceal re-
bleed.19,29
In conclusion, our data show that EVL alone is suffi-
cient to prevent rebleed in patients with portal hyperten-
sion who had bled in the past. Addition of the pharma-
cotherapy regimen of propranolol and ISMN to EVL does
not further reduce the incidence of variceal rebleeding as
achieved by EVL therapy given alone. Moreover, these
drugs add to the morbidity in a proportion of these
patients because of their serious adverse effects. The risk
factors for rebleed are ascites, low serum albumin, and
high baseline HVPG.
Supplementary Data
Note: To access the supplementary material
accompanying this article, visit the online version of
Gastroenterology at www.gastrojournal.org, and at doi:
10.1053/j.gastro.2009.05.049.
References
1. Jensen DM. Endoscopic screening for varices in cirrhosis: find-
ings, implications, and outcomes. Gastroenterology 2002;122:
1620–1630.
2. D’Amico G, De Franchis R, and the Cooperative Study Group.
Upper digestive bleeding in cirrhosis. Post-therapeutic outcome
and prognostic indicators. Hepatology 2003;38:599–612.
3. Carbonell N, Pauwels A, Serfaty L, et al. Improved survival after
variceal bleeding in patients with cirrhosis over the past two
decades. Hepatology 2004;40:652–659.
4. Graham DY, Smith JL. The course of patients after variceal hem-
orrhage. Gastroenterology 1981;80:800–809.
5. Gonzalez R, Zamora J, Gomez-Camarero J, et al. Meta-analysis:
combination endoscopic and drug therapy to prevent variceal
rebleeding in cirrhosis. Ann Intern Med 2008;149:109–122.
6. de Franchis R. Evolving consensus in portal hypertension. Report
of the Baveno IV consensus workshop on methodology of diag-
nosis and therapy in portal hypertension. J Hepatol 2005;43:
167–176.
7. García-Pagán JC, Feu F, Bosch J, et al. Propranolol compared with
propranolol plus isosorbide-5-mononitrate for portal hypertension
in cirrhosis. A randomized controlled study. Ann Intern Med 1991;
114:869–873.
8. Sarin SK, Kumar A. Noncirrhotic portal hypertension. Clin Liver
Dis 2006;10:627–651.
9. Conn HO. Ammonia tolerance in the diagnosis of esophageal
varices. A comparison of endoscopic, radiologic, and biochemical
techniques. J Lab Clin Med 1967;70:442–451.
10. Groszmann RJ, Wongcharatrawee S. The hepatic venous pres-
sure gradient: anything worth doing should be done right. Hepa-
tology 2004;39:280–282.
11. Garcia-Pagán JC, Navasa M, Bosch J, et al. Enhancement of
portal pressure reduction by the association of isosorbide-5-
mononitrate to propranolol administration in patients with cirrho-
sis. Hepatology 1990;11:230–238.
12. Gournay J, Masliah C, Martin T, et al. Isosorbide mononitrate and
propranolol compared with propranolol alone for the prevention of
variceal rebleeding. Hepatology 2000;31:1239–1245.
13. Garcia-Pagán JC, Bosch J. Endoscopic band ligation in the treat-
ment of portal hypertension. Nat Clin Pract Gastroenterol Hepatol
2005;2:526–535.
14. Stiegmann GV, Goff JS, Michaletz-Onody PA, et al. Endoscopic
sclerotherapy as compared with endoscopic ligation for bleeding
esophageal varices. N Engl J Med 1992;326:1527–1532.
15. Laine L, el-Newihi HM, Migikovsky B, et al. Endoscopic ligation
compared with sclerotherapy for the treatment of bleeding esoph-
ageal varices. Ann Intern Med 1993;119:1–7.
16. Gimson AE, Ramage JK, Panos MZ, et al. Randomised trial of
variceal banding ligation versus injection sclerotherapy for bleed-
ing oesophageal varices. Lancet 1993;342:391–394.
17. Lo GH, Lai KH, Cheng JS, et al. A prospective, randomized trial of
sclerotherapy versus ligation in the management of bleeding
esophageal varices. Hepatology 1995;22:466–471.
18. Hou MC, Lin HC, Kuo BI, et al. Comparison of endoscopic variceal
injection sclerotherapy and ligation for the treatment of esopha-
geal variceal hemorrhage: a prospective randomized trial. Hepa-
tology 1995;21:1517–1522.
19. Lo GH, Lai KH. The factors affecting risk of recurrent variceal
bleeding. Gastroenterology 2005;128:244.
20. Lo GH. Prevention of esophageal variceal rebleeding. J Chin Med
Assoc 2006;69:553–560.
21. Harewood GC, Baron TH, Song LM. Factors predicting success of
endoscopic variceal ligation for secondary prophylaxis of esoph-
ageal variceal bleeding. J Gastroenterol Hepatol 2006;21:237–
241.
22. Bosch J, García-Pagán JC. Prevention of variceal rebleeding. Lan-
cet 2003;361:952–954.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
900 KUMAR ET AL GASTROENTEROLOGY Vol. 137, No. 3

10. 23. Sharara AI, Rockey DC. Gastroesophageal variceal hemorrhage.
N Engl J Med 2001;345:669–681.
24. Sarin SK, Wadhawan M, Gupta R, et al. Evaluation of endoscopic
variceal ligation (EVL) versus propanolol plus isosorbide mononi-
trate/nadolol (ISMN) in the prevention of variceal rebleeding:
comparison of cirrhotic and noncirrhotic patients. Dig Dis Sci
2005;50:1538–1547.
25. de la Peña J, Brullet E, Sanchez-Hernández E, et al. Variceal
ligation plus nadolol compared with ligation for prophylaxis of
variceal rebleeding: a multicenter trial. Hepatology 2005;41:
572–578.
26. Lo GH, Lai KH, Cheng JS, et al. Endoscopic variceal ligation plus
nadolol and sucralfate compared with ligation alone for the pre-
vention of variceal rebleeding: a prospective, randomized trial.
Hepatology 2000;32:461–465.
27. Savoye G, Hochain P, Riachi G, et al. Early massive bleeding after
endoscopic ligation for junctional varices. Endoscopy 1998;30:
S101.
28. Bosch J. Prevention of variceal rebleeding: endoscopes, drugs, and
more. Hepatology 2000;32:660–662.
29. Mihas AA, Sanyal AJ. Recurrent variceal bleeding despite endo-
scopic and medical therapy. Gastroenterology 2004;127:621–629.
Received October 6, 2008. Accepted May 20, 2009.
Reprint requests
Address requests for reprints to: S. K. Sarin, MD, DM, Director,
Professor and Head, Department of Gastroenterology, Project
Director, Institute of Liver & Biliary Sciences, Room 201, Academic
Block, G B Pant Hospital, New Delhi 110 002, India.
e-mail: sksarin@nda.vsnl.net.in; fax: (91) 11-23219710.
Conflicts of interest
The authors disclose no conflicts.
CLINICAL–LIVER,
PANCREAS,AND
BILIARYTRACT
September 2009 EVL VS EVL؉ DRUGS FOR SECONDARY PROPHYLAXIS DRUGS 901

11. Supplementary Figure 1. The Consort E-Flowchart Aug. 2005.
901.e1 KUMAR ET AL GASTROENTEROLOGY Vol. 137, No. 3